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

BooksAudiobooksComicsSheet Music### Categories

### Categories

### Categories

Editors' Picks Books

Hand-picked favorites from

our editors

our editors

Editors' Picks Audiobooks

Hand-picked favorites from

our editors

our editors

Editors' Picks Comics

Hand-picked favorites from

our editors

our editors

Editors' Picks 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

Welcome to Scribd! Start your free trial and access books, documents and more.Find out more

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

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

· ) is a function diﬀerentiable in its three arguments. equating these terms with δS[q(t). t0 (1. q(t). q(t) + δq(t) is a ‘slightly’ diﬀerent trajectory than q(t). For this purpose we consider M ˙ ˙ ˙ L(q(t) + δq(t). (1. t1 ] where the integrand is a function L(q(t).g.e.. t) (1. · ] in conventional diﬀerential calculus does not correspond to a diﬀerentiated function.. ∂f in case of a function of M variables.6) immediately.6) For a proof we can use conventional diﬀerential calculus since the functional (1. dt q(t). q(t). i. · .5) where L( · .10) From this follows (1. q(t) + δ q(t). but rather to a diﬀerential of the function which is simply the df diﬀerentiated function multiplied by the diﬀerential increment of the variable. ˙ In the following we will often use the notation for velocities and other time derivatives d q(t) = q(t) and dtj = xj . df = dx dx or. q(t) + δ q(t). the velocity vector dt q(t) and time t. It is also important to appreciate that δS[ · . ˙ Theorem: Let S[q(t)] = t0 dx dt t1 ˙ dt L(q(t). t) (1. δq(t)] = dt − δqj (t) + δqj (t) ∂qj dt ∂ qj ˙ ∂ qj ˙ t0 j=1 j=1 t1 . t) of the conﬁguration d vector q(t). It holds M M ∂L t1 d ∂L ∂L δS[q(t). We attempt to evaluate t1 S[q(t) + δq(t)] = t0 ˙ ˙ dt L(q(t) + δq(t). df = M ∂xj dxj . t) = L(q(t).e. e. at the ends of the time interval of the trajectories the variations vanish. δq(t)].2 Lagrangian Mechanics Note: δq(t) describes small variations around the trajectory 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 . 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. .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 . i.6) is expressed in terms of ‘normal’ functions. t) + j=1 ∂L ∂L δqj + δ qj ˙ ∂qj ∂ qj ˙ + O( 2 ) (1. We focus on such functionals because they play a central role in the so-called action integrals of Classical Mechanics.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 ) (1.7) through Taylor expansion and identiﬁcation of terms linear in δqj (t).

11) The extremals qe (t) can be identiﬁed through a condition which provides a suitable diﬀerential equation for this purpose. t1 ] ⊂ R → R holds t1 (1. if and only if it satisﬁes the conditions (j = 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 . · ] ≡ 0 and. . The length of such path can be determined starting from the fact that the incremental length ds in going from point (x.12) for j = 1.5) assume extreme values.13) dt f (t)h(t) = 0 t0 (1.16) the property δS[qe (t).14) for any continuous function h(t) ∈ F with h(t0 ) = h(t1 ) = 0. y(x + dx)) is ds = (dx)2 + ( dy dx)2 = dx dx 1+( dy 2 ) . (1.15) We will not provide a proof for this Lemma. (1. 2. dx (1. δq(t)] = dt − δqj (t) . δq(t)] = 0 for all δq(t) ∈ F . .12) for j = 1. . M . the above theorem. . . .12) The proof of this theorem is based on the property Lemma: If for a continuous function f(t) f : [t0 . 2. For this purpose we deﬁne Deﬁnition: An extremal of a diﬀerentiable functional S[ ] is a function qe (t) with the property δS[qe (t). from (1. y(x2 ) = y2 . t1 ]. y(x1 ) = y1 . Let us assume that the two points are connected by the path y(x). Theorem: Euler–Lagrange Condition For the functional deﬁned through (1. y(x)) to (x + dx. then f (t) ≡ 0 on [t0 . This condition is stated in the following theorem.16) This property holds for any δqj with δq(t) ∈ F . . . hence.6) which reads in the present case M ∂L t1 d ∂L δS[q(t). 2. ∂qj dt ∂ qj ˙ t0 j=1 (1. y2 ).1. The proof of the above theorem starts from (1. it holds in case δq(t0 ) = δq(t1 ) = 0 that qe (t) is an extremal.1: Variational Calculus 3 We now consider the question for which functions the functionals of the type (1. On the other side. y1 ) and (x2 . M ) d dt ∂L ∂ qj ˙ − ∂L = 0 ∂qj (1.17) . . According to the Lemma above follows then (1. . M and δqj (t0 ) = δqj (t1 ) = 0 follows according to (1.5).

The shortest path is an extremal of s[y(x)] which must. . 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 . M (1. .1. . Velocity–dependent potentials which describe particles moving in electromagnetic ﬁelds will be considered below. This in turn yields y(x) = ax + b.23) Presently we consider only velocity–independent potentials.22) ˙ where L(q(t). j = 1. dx ) = 1 + (dy/dx)2 . q(t). hence.4 The total path length is then given by the integral x1 Lagrangian Mechanics s = x0 dx 1+( dy 2 ) . t1 S[q(t)] = t0 ˙ dt L(q(t). (1. dx (1. q(t).5) with L(y(x). . q(t).1: Show that the shortest path between two points on a sphere are great circles. .21) are extremals of the functional. One obtains y 1 − y2 y(x) = (x − x2 ) + y2 . . . 1. y = const. i. Using y = dx the condition reads d dx ∂L ∂y = d dx y 1 + (y )2 = 0. obey the Euler–Lagrange dy condition. the so-called action integral. circles whose centers lie at the center of the sphere. (1. t) = j=1 1 mj qj − U (q1 . . qM ) . ˙2 2 (1. t) (1. 2. q2 .18) dy s is a functional of y(x) of the type (1. t) is the so-called Lagrangian M ˙ L(q(t)..e.20) x1 − x2 Exercise 1.19) From this follows y / 1 + (y )2 = const and.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. according to the theorems above. The constants a and b are readily identiﬁed through the conditons y(x1 ) = y1 and y(x2 ) = y2 .

respectively. t) 1∂ K(r. t) = A(r. t). The equations of motion for such a particle are d q ˙ (mr) = F (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.1. t) . t) as follows B E = ×A 1 ∂A . These conditions read in the present case ∂L d ∂L ∂U d − = 0 → − − (mj qj ) = 0 ˙ (1. dt The ﬁelds E(r.28) (1.23).27) and (1. t) c ∂t (1. t) obey the Maxwell equations ×E + ·B ×B − ·E 1 ∂ c ∂t E 1 ∂ c ∂t B (1. t) and B(r.33) . t) = V (r.30) (1. The transformation which leaves the ﬁelds invariant is A (r. t) describes the charge current density. x3 (t)) moving in an electromagnetic ﬁeld described through the electrical and magnetic ﬁeld components E(r. t) describes the charge density present in the ﬁeld and J(r. t) + v × B(r. Equations (1.24) ∂qj dt ∂ qj ˙ ∂qj dt which are obviously equivalent to the Newtonian equations of motion.25) = 0 = 0 4π J c = 4πρ = (1. t) dt c where dr = v and where F (r. t) = q E(r. c ∂t (1.27) (1.26) (1. x2 (t).30.28) can be satisﬁed implicitly if one represents the ﬁelds through a scalar potential V (r.31) = − V − Gauge Symmetry of the Electromagnetic Field It is well known that the relationship between ﬁelds and potentials (1. t) and B(r. t) and a vector potential A(r.22.2: 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 (1.25) of a particle moving in the ﬁeld.32) (1. t) + V (r. t) − K(r. 1.29) where ρ(r. 1. F (r.31) allows one to transform the potentials without aﬀecting the ﬁelds and without aﬀecting the equations of motion (1.

35) − ∂A1 ∂x2 [v × B]1 = x2 B3 − x3 B2 = x2 ˙ ˙ ˙ ∂A2 ∂A1 − ∂x1 ∂x2 − x3 ˙ ∂A1 ∂A3 − ∂x3 ∂x1 . Comparing then (1.36) This expression allows us to show that (1.41) with (1. B3 = ∂A2 ∂x1 (1. in fact. 2 c (1.41) which is identical to the term (1. if one assumes for a particle the Lagrangian ˙ L(r..35) shows that the Newtonian equations of motion and the Euler–Lagrange conditions are. t) = 1 q mv 2 − q V (r.38) ∂L ∂ x1 ˙ − ∂L = 0.36) in the Newtonian equation of motion. namely.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. 1.25). equivalent. . 1.40. dt ∂x1 c We notice using (1. r.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 . t) · v . (1.25) follows from the Hamiltonian Principle of Least Action.40) (1. t) + A(r.39) ∂A1 ∂A2 ∂A3 x1 + ˙ x2 + ˙ x3 ˙ ∂x1 ∂x1 ∂x1 . (1.g. ∂x1 (1. e.38.37) is ∂L ∂V q = −q + ∂x1 ∂x1 c The ﬁrst term in (1. d ∂V q (mv1 ) = F1 = −q + [v × B]1 .30).35) is equivalent to the Euler–Lagrange condition d dt The second term in (1.37) The results (1. (1.34) For this purpose we consider only one component of the equation of motion (1.6 Lagrangian of Particle Moving in Electromagnetic Field Lagrangian Mechanics We want to show now that the equation of motion (1.

33) ˙ L (r. 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.e. q. Eq. 1. (1.42) is. For this purpose we consider the transformation of the single particle Lagrangian (1. t) + q K(q.43) To prove this theorem we determine the action integral corresponding to the transformed Lagrangian t1 S [q(t)] = t0 ˙ dtL (q. t) + K ·v (1. t) the Lagrangian. encountered above in connection with the transformations (1.45) . r. t) + dt c (1. q. They are the subject of the following theorem.32.42) This transformation is termed gauge transformation. 1. t) 1 1 ∂K q mv 2 − q V (r.33) of electromagnetic potentials. t) + A(r.32.45) and reordering terms yields using (1. t) · v . i. t) − + 2 c ∂t c 1 q = mv 2 − q V (r. Note that one adds the total time derivative of a function K(r. q. t) + A (r. t) c t1 t0 = S[q(t)] + (1. a term not aﬀected by the variations allowed. t) L (q. r.46) (1.34) q 1 q d ˙ L (r. q. in fact. t) . The gauge symmetry.1.43) into (1. 2 c c dt Inserting (1. 2 c = A(r.44) implies that the gauge transformation amounts to adding a constant term to the action integral. t) = q K(q. appear in a diﬀerent.32. t) · v + K(r. We will consider in the following two symmetries. We want to demonstrate now that the transformation (1. 1. t) = x1 + ˙ x2 + ˙ x3 + ˙ = ( K) · v + .33) of electromagnetic potentials. equivalent to the gauge transformation (1. t) c t1 t0 t1 t0 ˙ dtL(q.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. The factor q has been introduced to make this c transformation equivalent to the gauge transformation (1. t) = mv 2 − q V (r.32.3: Symmetry Properties 7 1.. gauge symmetry and symmetries with respect to spatial transformations.44) Since the condition δq(t0 ) = δq(t1 ) = 0 holds for the variational functions of Lagrangian Mechanics. This term is d ∂K ∂K ∂K ∂K ∂K K(r.33) of electyromagnetic potentials. dt ∂x1 ∂x2 ∂x3 ∂t ∂t (1. t) = L(q. 1. One can conclude then immediately that any extremal of S [q(t)] is also an extremal of S[q(t)]. and since these properties allow one often to simplify mathematical descriptions. surprisingly simple fashion in Lagrangian Mechanics.

We consider now invariance properties connected with coordinate transformations. t). The transformations considered are speciﬁed in the following deﬁnition. r ∈ R . the transformation (1.50) A non-trivial example is furnished by the inﬁnitesimal rotation around axis e ˆ r = r + e×r . for the connection between invariance properties and constants of motion. 1.. t).49) =0 ∈ R (1. The following description of spatial symmetry is important in two respects. for example. The corresponding inﬁnitesimal transformation is deﬁned for small r (r. t). where the origin of is chosen such that r (r. Deﬁnition: Inﬁnitesimal One-Parameter Coordinate Transformations A one-parameter coordinate transformation is decribed through r = r (r.32. The transformations we consider are the most simple kind.52) x3 0 0 1 x3 . t) by gauge transformed potentials V (r.47) (1. ˆ (1.42) corresponds to replacing in the Lagrangian potentials V (r. the equivalence of (1. but nevertheless instructive way considering rotations around the x3 –axis. ) .e.33). A (r. R(r) = through ∂r ∂ (1. A(r. In case of a translation transformation in the direction e nothing new is gained. Such invariance properties are very familiar. we like to ˆ provide the transformation here anyway for later reference r = r + e. ˆ (1.8 Lagrangian Mechanics Obviously. therefore. 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.51) We would like to derive this transformation in a somewhat complicated. ˆ ˆ ˆ Examples of Inﬁnitesimal Transformations The beauty of inﬁnitesimal transformations is that they can be stated in a very simple manner. and for the introduction of inﬁnitesimal transformations which will provide a crucial method for the study of symmetry in Quantum Mechanics. in the case of central force ﬁelds which are invariant with respect to rotations of coordinates around the center. In this case the transformation can be written in matrix form cos −sin 0 x1 x1 x2 = sin cos 0 x2 (1. We have proven. ) = r + R(r) + O( 2 ) . r. for such vectors holds a · a = 1. which has an important analogy in Quantum Mechanics.42) and (1. However. 0) = r . i.48) In the following we will denote unit vectors as a.

3: Symmetry Properties 9 In case of small this transformation can be written neglecting terms O( 2 ) using cos = 1 + O( 2 ). 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.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. Using the fact..k=1 ∂L ∂Qj qk ˙ ∂ qj ∂qk ˙ (1. In the context of Lagrangian Mechanics this implies that such transformation leaves the Lagrangian of the system unchanged. namely for ﬁelds rather than only for particle motion. q.58) From this follows the statement of the theorem. t) yields after Taylor ˙ 2 ). We have used here the notation corresponding to single particle motion. Anytime.57) ∂ d where we used dt Qj = M ( ∂qk Qj )qk . a classical mechanical system is invariant with respect to a coordinate transformation ˙ a constant of motion exists. The property has been shown to hold in a more general context. ˙ ∂qk ˙ Inserting these inﬁnitesimal changes of qj and qj into the Lagrangian L(q. then M ∂L j=1 Qj ∂ xj is a constant of motion. We consider here only the ‘particle version’ of the theorem.e. q. q. however.53). sin = + O( 2 ) 0 − 0 x1 x1 x1 x2 = x2 + 0 0 x2 + O( 2 ) .e. 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. a quantity C(r. r) which is constant along the classical path of the system. In order to prove Noether’s theorem we note qj qj ˙ = qj + = qj + ˙ k=1 Qj (q) M (1.57) must cancel each other or both vanish. q.55) (1. Invariance implies L = L. ˙ We have generalized in this theorem the deﬁnition of inﬁnitesimal coordinate transformation to M –dimensional vectors q.. t) is invariant with respect to an inﬁnitesimal transformation q = q + Q(q).54) which is equivalent to (1. the second and third term ˙ k=1 in (1. expansion.51) reads r = r − x2 e1 + x1 e2 ˆ ˆ (1. by Noether. i. (1. i. the property holds for any system. neglecting terms of order O( M ˙ ˙ L (q. t) = L(q. t) + j=1 ∂L Qj + ∂qj M j.1. .56) ∂Qj qk . Theorem: Noether’s Theorem ˙ If L(q.57) M Qj j=1 d dt ∂L ∂ qj ˙ + ∂L d Qj ∂ qj dt ˙ = d dt M Qj j=1 ∂L = 0 ∂ qj ˙ (1.

hence. ˆ ˙ ˆ j=1 (1. It is not necessary to investigate the consequences of global. j = 1. i. 2. 2.51). We will now investigate the consequence of rotational invariance as described according to the inﬁnitesimal transformation (1. j = 1.59) We obtain the well known result that in this case the momentum in the direction. 2 We ﬁrst determine the constant of motion in case of invariance with respect to translations as deﬁned in (1. for which translational invariance holds. except using now Qj = ej · (ˆ × r). In this case we will use the same notation as in (1. 3 and. of course. 3) 3 Qj j=1 ∂L = e· ˆ ∂ xj ˙ 3 ej mxj = e · mv . In this case we have Qj = ej · e. not inﬁnitesimal transformations.59). to be expected that this is the constant of motion. . 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. is conserved. v) = 1 mv 2 − U (r). It was.50). 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. Noether’s theorem yields ˆ ˆ the constant of motion (qj = xj . A calculation similar to that in (1.59) yields the constant of motion ˆ e (ˆ × r) · mv.e.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.

We provide now a set of formal rules which state how the probability to observe such a particle at some space–time point r. t1 ] .1) (2. The probability to detect the particle with a detector of sensitivity f (r) is d3 r f (r) |ψ(r. 1. t)|2 = ψ(r. r.Chapter 2 Quantum Mechanical Path Integral 2.1 The Double Slit Experiment Will be supplied at later date 2. The particle is described by a wave function ψ(r. is unity. 4.2) (2. (2. 11 . The wave function ψ(r. 2. t)|2 = 1 Ω ∀t. t) ψ : R3 ⊗ R → C. placed into the space at some particular time and with a detector behind each slit. t is described in Quantum Mechanics. At present one may think of f (r) as a sum over δ–functions which represent a multi–slit screen.4) a condition which enforces that the probability of ﬁnding the particle somewhere in Ω at any particular time t in an interval [t0 . The probability that the particle is detected at space–time point r. t ∈ [t0 . t). t)|2 Ω (2. t1 ] in which the particle is known to exist.3) where Ω is the space volume in which the particle can exist. t)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) is normalized d3 r |ψ(r. 3. t) where z is the conjugate complex of z. t is |ψ(r.

t1 ]) Ωd Ωd 3r Ωd 3r Ωd 3r 3 r |ψ(r. t|r .. t ) φ(r .8) 7. the path integral r(tN )=rN d[r(t)] · · · r(t0 )=r0 (2. The following so-called completeness relationship holds for the propagator (t > t [t0 . t1 |r0 .10) We have introduced here a new symbol.12 Quantum Mechanical Path Integral 5. the propagator is unitary. t|r0 . t )φ(r. The generalization of the completeness property to N − 1 intermediate points t > tN −1 > tN −2 > . t ) φ(r.11) which denotes an integral over all paths r(t) with end points r(t0 ) = r0 and r(tN ) = rN .e. t|r0 . This is diﬀerent from the classical situation where the particle follows a discrete path and. namely the point on the classical path at time t .9) Employing a continuum of intermediate times t ∈ [t0 . t ) ψ(r . t0 ) = δ(r − r0 ). t0 ) = φ(r. i. t)|2 = (2.. t. t|r . tN −1 ) φ(rN −1 . Since (2. (2. t ) ψ(r .9) for N → ∞. . ψ(r0 . t| rN −1 . t). t0 ) = Ωd 3r N −1 Ωd 3r N −2 ··· Ωd 3r 1 φ(r. 8. t|r . 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. t0 ) = r(t0 )=r0 d[r(t)] Φ[r(t)] . i. The deﬁnition will actually assume an inﬁnite number of intermediate times and express the path integral through integrals of the type (2. . This symbol will be deﬁned further below.8) a knowledge of the state at all space points r ∈ Ω at some intermediate time t . t. (2.5) 6.7) t. . hence. t ∈ (2. t ∈ [t0 . The state of a system at time t. at any intermediate time the particle needs only be known at one space point. tN −2 ) · · · φ(r1 . t0 ) . requires then according to (2. t ) Ωd 3 r |ψ(r. (t > t .4) holds for all times. t) is described by a linear map of the type ψ(r. t ) which requires d3 r φ(r. t|r0 . t = )|2 = 1 . t1 ] (2. > t1 > t0 is φ(r. t) = Ω d3 r φ(r. t ) t > t . t |r0 . t ) = δ(r − r ) Ω (2. tN −1 |rN −2 . t|r . t1 ]) d3 r φ(r. described by ψ(r. This must hold for any ψ(r .e. t ) ψ(r . t|r . t ∈ [t0 . t1 ] yields an expression of the form r(tN )=rN φ(r.6) φ(r. The time evolution of ψ(r. t|r .

. Since this number is multiplied by ‘i’. Its value is extremely small in comparision with typical values for action integrals of macroscopic particles.12..5 · 1027 . a very large number. This implies that small variations of the path near the classical path alter the value of the action integral by very little.15) The exponent of (2.2: Axioms 9. for microscopic particles like the electron this is by no means the case. However. (2. The constant given in (2. However.12.14) has the same dimension as the action integral S[r(t)].12) is then Scl / ≈ 0. accordingly.13) L(r. expressions (2. namely due to the property of the classical path to be an extremal of the action integral. i.11) is Φ[r(t)] = exp where S[r(t)] is the classical action integral tN 13 i S[r(t)] (2. The situation is very diﬀerent for microscopic particles. 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.2. The functional Φ[r(t)] in (2. 2. 2. it is comparable to action integrals as they arise for microscopic particles under typical circumstances. i. actually often essentially exclusive.e. r. such that destructive interference of the contributions of such paths does not occur. the exponent is a very large imaginary number. However.0545 · 10−27 erg s . Scl / ≈ 8. not only for the classical path. 2 2 (2. 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. t) is the Lagrangian of the classical particle. One would still expect signiﬁcant descructive interference between contributions of diﬀerent paths since the value calculated is comparable to 2π. The value of S[r(t)] is then Scl = 1 1 m v 2 t = erg s .13). . However. in complete distinction from Classical Mechanics. interferences should be much less dramatic than in case of the macroscopic particle.14) ˙ In (2. contribution to the path integral (2. r. 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.e. Situations which are well described classically will be distinguished through the property that the classical path gives the dominant. t) (2. Only for paths close to the classical path is such interference ruled out. for the electron many paths contribute and the action integrals for non-classical paths need to be known.13) and = 1.6725 · 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.12) S[r(t)] = t0 ˙ dt L(r.13) are built on action integrals for all possible paths. In case of a proton with mass m = 1.

provides an avenue to evaluate path integrals. . namely 1 m(xj+1 − xj )2 / 2 and U (xj ).21) N 2. however. . This allows us to write the evolution operator using (2. −∞ dxN −1 (xj+1 −xj )2 1 − U (xj ) . (2. (x1 . x2 . ˙ 2 (2. (xN . (2. .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. namely tj+1 − tj = (tN − t0 )/N = N . (2.20) can be properly taken. .12) φ(xN . t) = ˙ 1 mx2 − U (x) . t1 ). For the sake of simplicity we will consider the case of particles moving in one dimension labelled by the position coordinate x. . at the same time. They can be anywhere in the allowed volume which we will choose to be the interval ] − ∞. the xj values are not. > t1 > t0 letting N go to inﬁnity later.19) The main idea is that one can replace the path integral now by a multiple integral over x1 . tN |x0 . 2.3 How to Evaluate the Path Integral In this section we will provide an explicit algorithm which deﬁnes the path integral (2.16) In order to deﬁne the path integral we assume. ∞[. respectively.10) and (2. tN )} . .13) and. as in (2. . Its value is CN = m 2πi N 2 (2. tN −1 ). x. 2 2m N (2. . t0 ).20) Here.9). The particles have associated with them a Lagrangian L(x. 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. 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 .18) The time instances are ﬁxed. tj+1 ) we assume that kinetic energy and potential energy are constant. In passing from one space–time instance (xj . tj ) to the next (xj+1 .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 .12. t0 ) = limN →∞ CN exp i N N −1 j=0 +∞ +∞ +∞ −∞ dx1 −∞ dx2 .14 Quantum Mechanical Path Integral 2. The spacings between the times tj+1 and tj will all be identical. a series of times tN > tN −1 > tN −2 > . etc. (xN −1 .

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.26) The ﬁrst term on the r.2. t0 ) = exp S[x(t)] (2. t0 ) = exp 1 im (xN − x0 )2 2 tN − t0 φ(0.10.s. of (2.13). (2. the origin of which coincides with the classical path of the particle.e.. it is more suitable to deﬁne new integration variables yj . To see the beniﬁt of such approach we deﬁne a path y(t) as follows x(t) = xcl (t) + y(t) (2. t0 ) and (xN .29) This expression corresponds to the action integral in (2. ˙ t0 (2. The resulting expression for S[x(t)|x(t0 ) = x0 .31) We have denoted explicitly that the action integral for a path connecting the space–time points (x0 . x(tN ) = 0. namely. xN − x0 xcl (t) = x0 + ( t − t0 ) . (2.23).h. it holds y(t0 ) = y(tN ) = 0 . ˙ ˙ ˙ 2 + y2) = ˙ (2. 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 . using (2. tN |x0 . since x(t0 ) = xcl (t0 ) = x0 and x(tN ) = xcl (tN ) = xN . tN ) is to be evaluated. = (2. can be expressed through a path integral with endpoints x(t0 ) = 0. Inserting the result into (2.27) The third term can be written in the notation introduced tN 1 dt my 2 = S[x(t)|x(t0 ) = 0. One obtains S[x(t)|x(t0 ) = x0 . t0 ) and (xN .22) where xcl (t) is the classical path which connects the space–time points (x0 .25) + The condition (2. The ˙ 1 S[x(t)|x(t ) = x .4: Propagator for a Free Particle 15 Rather then using the integration variables xj . 2.30) d[x(t)] exp 2 tN − t0 x(t0 )=0 a result. tN ). tN |0. t0 ) (2. x(tN ) = 0] . due to (2.s. x(tN ) = xN ] 1 (xN − x0 )2 m + 0 + 2 tN − t 0 + S[x(t)|x(t0 ) = 0.23) tN − t0 It is essential for the following to note that. ˙ 2 t0 (2.25) is.24) implies for the second term on the r. 2 2 tN − t0 (2. x(tN ) = xN ] is S[x(t)|x(t0 ) = x0 . tN dt y = y(tN ) − y(t0 ) = 0 .24) Also we use the fact that the velocity of the classical path xcl = (xN −x0 )/(tn −t0 ) is constant.24). tN dt t0 1 1 (xN − x0 )2 m x2 = m ˙ cl . x(tN ) = 0] .h. .12) yields x(tN )=0 im (xN − x0 )2 i φ(xN . which can also be written φ(xN . tN |x0 .28) i.

t0 ) = limN →∞ × +∞ +∞ −∞ dy1 · · · −∞ dyN −1 exp i N m 2πi N 2 N × (2. noting y0 = yN = 0. 2 −1 −1 2 (2. . 0 0 .k=1 yj ajk yk − 1) × (N − 1) matrix . (2. . . .33) where ajk are the elements of the following symmetric (N 2 −1 0 −1 2 −1 2 m 0 −1 ajk = . In the appendix we prove +∞ +∞ d j.32) we split oﬀ the factor 2 mN in the deﬁnition (2. .34) of (ajk ) deﬁning a new matrix (Ajk ) through ajk = Using det(ajk ) = m 2 N N −1 m 2 N Ajk .34) dy1 · · · −∞ −∞ dyN −1 exp i N −1 j. .. 0 0 0 0 0 0 The following integral +∞ +∞ (2.36) which holds for a d-dimensional.38) . ....37) det(Ajk ) . 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.. . tN |0. . . 0 0 .16 Evaluation of the necessary path integral Quantum Mechanical Path Integral To determine the propagator (2. (2. . . .32) (yj+1 − yj )2 N −1 1 2 j=0 2 m N The exponent E can be written. .k=1 dy1 · · · −∞ −∞ dyN −1 exp i yj bjk yk = (iπ)d det(bjk ) 1 2 .k=1 yj ajk yk (2. (2.31) for a free particle one needs to evaluate the following path integral φ(0.. 0 0 .. In order to complete the evaluation of (2. .35) must be determined... 2 N . symmetric matrix (bjk ) and det(bjk ) = 0. real.

43) We like to note here for further use below that one might as well employ the ‘artiﬁcial’ starting values D0 = 1.4: Propagator for a Free Particle a property which follows from det(cB) = cn detB for any n × n matrix B. .46) . (2. .45) Expressions for Free Particle Propagator We have now collected all pieces for the ﬁnal expression of the propagator (2. . (2. 0 0 Dn = . (2.. x = xN φ(x.2. tN |0. which follows from (2.41) D2 = 2 −1 −1 2 = 3 (2..39) yields φ(0.40) .31) and obtain. D3 . deﬁning t = tN . t0 ) = limN →∞ and with NN (2. . 0 0 0 −1 2 . n = 1. we obtain φ(0. let say n.. . One can readily verify that this procedure leads to the following recursion equation for the determinants Dn = 2 Dn−1 − Dn−2 . 2 −1 0 0 0 −1 2 For this purpose we expand (2. 2. . .42) m 2πi NN 1 2 (2. .44) = tN − t0 . t0 ) = m 2πi (tN − t0 ) 1 2 ..41) the same result for D2 . . t0 ) = limN →∞ m 2πi N 2 17 N 2πi m N N −1 2 1 . one can readily verify Dn = n + 1 . Using D1 = |(2)| = 2 . .18) we obtain φ(0. . . (2.. . 0 0 0 . ..39) In order to determine det(Ajk ) we consider the dimension n of (Ajk ).40) in terms of subdeterminants along the last column. .. presently N − 1. We seek then to evaluate the determinant of the n × n matrix 2 −1 0 . t0 ) = m 2πi (t − t0 ) 1 2 exp im (x − x0 )2 2 t − t0 . . To solve this three term recursion relationship one needs two starting values. . tN |0. D1 = 2 and obtain from (2... det(Ajk ) (2. t|x0 . Our derivation has provided us with the value det(Ajk ) = N . . . variable. Inserting this into (2. . . 0 0 −1 2 −1 . . tN |0. . .

We want to apply axiom (2. x) + E(x) (2.47) 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 .48) as the initial state using the propagator (2.5) allows us to predict the time evolution of any state function ψ(x.5) to (2. The result (2. t) of a free particle. centered around x0 = 0. t0 ) = m 2πi (t − t0 ) 3 2 exp im (r − r0 )2 2 t − t0 .51) as follows Eo (xo . t|r0 . (2.10) φ(r. (2. t0 )| = 2 1 πδ 2 1 2 exp − x2 0 δ2 (2.48) Obviously. x) = im 2 t x2 1 + i o t mδ 2 − 2xo x − po t m + f (x) (2. We need to evaluate for this purpose the integral ψ(x. 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 . the wave function of the particle at later times.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. t0 ) = 1 πδ 2 1 4 exp − x2 po 0 +i x 2 2δ (2.53) . and describes a single particle since 1 πδ 2 1 2 +∞ dx0 exp −∞ − x2 0 δ2 = 1.52) and applying (2. according to (2.50) One refers to such states as wave packets. We will obtain.46).18 Quantum Mechanical Path Integral This propagator. One obtains then for the propagator (2. We devide the contributions to the exponent Eo (xo . x) + E(x) in (2. the associated probability distribution |ψ(x0 . Below we will apply this to a particle at rest and a particle forming a so-called wave packet. thereby.46) can be generalized to three dimensions in a rather obvious way.247).49) is Gaussian of width δ.

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 . according to (2. z2 = +∞ × i 1−i t mδ 2 (2.53) E(x) = x− po mt 19 (2. We proceed as follows.54) f (x) = This yields Eo (xo . however.x) (2.58) The integrand is an analytical function everywhere in the complex plane and we can alter the integration path. making certain. (2.59) An integration path in the complex x0 –plane along the direction i 1−i t mδ 2 (2. that the new path does not lead to additional contributions to the integral. 2 t One chooses then f (x) to complete.56) m 2πi t 1 2 +∞ eE(x) −∞ dx0 eEo (xo . 2 (2. The beginning and the end of such path are the points z1 = −∞ × i 1−i t mδ 2 .61) . 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 .4: Propagator for a Free Particle im x2 − f (x) .2. x) = One can write then (2.55) im xo 2 t 1+i t − mδ 2 x− 1+ po m t t i mδ2 . the square in (2.51) ψ(x.52).57) and needs to determine the integral +∞ I = −∞ +∞ dx0 eEo (xo . t) = 1 πδ 2 1 4 2 t 1 + i mδ2 . (2.60) is then represented by real ρ values. (2.

(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) ] . z2 = +∞ .54) using (2.64) ψ(x.67) We need to determine ﬁnally (2.58) has a leading contribution in x0 of −x2 /δ 2 the integrand of (2. (2.63) holds and. Hence.63) which can be readily evaluated. accordingly. one can show that z1 lies at −∞ − i × ∞ and z2 at +∞ + i × ∞. Quantum Mechanical Path Integral (2. using a ab = a − .58) by t mδ 2 +∞ I = i 1−i dρ exp − −∞ m 2 t 1+ t mδ 2 2 ρ2 (2.69) t)2 t 2δ 2 (1 + i mδ2 ) + i po x − i p2 o t 2m (2.20 whereas the original path in (2. 1+b 1+b ﬁnally E(x) = − (x − po m (2.70) .57) reads then 1 πδ 2 1 4 2πi t . We can conclude then that (2.62) If one can show that an integration of (2. Since the exponent in (2. In fact. I = Equation (2. t m(1 + i mδ2 ) (2.58) has the end points z1 = −∞ .68) and.55).58) vanishes 0 for Re x0 → ±∞. t) = Seperating the phase factor 1 t 1 + i mδ2 1 4 1 2 exp [ E(x) ] . 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. the paths between z1 → z1 and z2 → z2 have a real part of x0 of ±∞.66) yields ψ(x.65) t 1 − i mδ2 t 1 + i mδ2 .58) along the path z1 → z1 and along the path z2 → z2 gives only vanishing contributions one can replace (2. (2.

74) remains invariant in time.4: Propagator for a Free Particle which inserted in (2.72). t) = exp i i p2 po o x − (t − to ) m 2m .71. However.49)..72) 2 t2 δ 1+ 2 4 (2. the spatial dependence of the initial state (2. 2. (2.48) and for the ﬁnal state (2.74) which corresponds to (2.77) . (2. This yields.67) provides the complete expression of the wave function at all times t ψ(x. The conservation of this factor is particularly striking for the (unnormalized) initial wave function ψ(x0 . instead of (2.75) ψ(x.e.72).76) From this we conclude that an initial wave function ψ(xo . In this case holds ψ(x.72) Comparision of Moving Wave Packet with Classical Motion It is revealing to compare the probability distributions (2. 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 .72) for the initial state (2. 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 . The width of the distribution (2.75) i. t0 ) = exp i po xo m . The center of the distribution (2. the case of arbitrary to is recovered iby replacing t → to in (2. (2. t0 ) = exp i po i p2 o xo − to m 2m .48) for δ → ∞. Another interesting observation is that the wave function (2.71). (2. We had assumed above [c. The corresponding probability distribution is |ψ(x.73) m δ increases with time. 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 (2.48)] to = 0.f. (2. (2. respectively.72) moves in the direction of the positive x-axis with velocity vo = po /m which identiﬁes po as the momentum of the particle. This ‘spreading’ of the wave function is a genuine quantum phenomenon.71) × exp − (x − po m t)2 ) 2δ 2 (1 + 2 t2 m2 δ 4 (1 − i .2. (2. coinciding at t = 0 with the width of the initial distribution (2.71) conserves the phase factor exp[i(po / )x] of the original wave function (2. t) = exp i po i p2 o x − t m 2m .49).48) and that the respective phase factor is related with the velocity of the classical particle and of the center of the distribution (2.

Obviously. y(t).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 . xcl + y(t). 2 . For this purpose we use d xcl y = ˙ ˙ (xcl y) − xcl y ˙ ¨ (2. t) ˙ L (y.83) (2. One refers to the respective states as stationary states.82) which describes the deviation from the classical path xcl (t) with end points xcl (t0 ) = x0 and xcl (tN ) = xN . t) ˙ (2.10. 2. tN |x0 . t) = L(xcl . 1. y(tN ) = 0 . Such states play a cardinal role in quantum mechanics.22 becomes at t > to ψ(x.80) into (2.13) φ(xN .79) for a quadratic Lagrangian L(x. t) + L (y. t) = ˙ 1 1 mx2 − c(t)x2 − e(t)x .80) For this purpose we need to determine the action integral S[x(t)] = t0 dt L(x..e.85) We want to show now that the contribution of δL to the action integral (2. xcl . (2.81) vanishes2 . 2. x.78) i. the end points of y(t) are y(t0 ) = 0 Inserting (2.81) for an arbitrary path x(t) with end points x(t0 ) = x0 and x(tN ) = xN . 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 2 tN (2. y(t). y(t)] which vanishes according to the Hamiltonian principle as discussed in Sect. 2.12.82) one obtains L(xcl + y. t) = exp i Quantum Mechanical Path Integral po i p2 o x − t m 2m .5 Propagator for a Quadratic Lagrangian x(tN )=xN We will now determine the propagator (2. (2. t) + δL ˙ ˙ ˙ ˙ where L(xcl . 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 . xcl . t0 ) = x(t0 )=x0 d[x(t)] exp i S[x(t)] (2.84) . ˙ ˙ = (2.

83) the ﬁrst term on the r.88) vanishes. of (2. . 0 0 0 c2 0 . For the evaluation of φ(0. . . .88) and. hence. tN |0. 0 0 0 −1 2 . .80) yields for the classical path m¨cl + c(t) xcl + e(t) = 0 x (2. . . t0 ) we will adopt a strategy which is similar to that used for the evaluation of (2. tN |0. .87) According to (2..h. . 0 0 0 c3 .24) to the Lagrangian (2.93) . t0 ) = y(tN )=0 d[y(t)] exp y(t0 )=0 i tN dt L (y. .32). . . . y. . . 0 0 m ajk = .s. cN −2 0 0 0 0 0 cN −1 (2. The discretization scheme adopted above yields in the present case ˜ φ(0. .h. (2. also the second contribution on the r. . tN |0.2. .91) (yj+1 − yj )2 1 2 2m N where cj = c(tj ).91) through the quadratic N −1 E = i j.. 2 −1 0 0 0 −1 2 c1 0 0 . . . 2 N .31) for the free particle propagator. t0 ) (2.79) i ˜ φ(xN . One can then express the propagator (2. x t0 (2. . . . . . t0 ) = limN →∞ × +∞ +∞ −∞ dy1 · · · −∞ dyN −1 exp N i N N −1 j=0 m 2πi N 2 N × − 1 2 cj 2 yj (2.. . . vanishes. .. 0 0 0 .89) where ˜ φ(0... 2 . 0 0 −1 2 −1 .k=1 yj ajk yk (2. . .92) where ajk are the elements of the following (N − 1) × (N − 1) matrix 2 −1 0 . . tj = t0 + form j. .. Applying the Euler–Lagrange conditions (1. tN |x0 . . t) ˙ t0 . . . 0 0 N 0 − . . One can express the exponent E in (2. t0 ) = exp S[xcl (t)] φ(0. 0 0 0 . . .. tN |0. .s.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) .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. .

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

In contrast to W (x.147) in (2. 2 (2. t − to ) 1 ˜ exp[−iω(m + 1 ) to ] φm (xo ) 2 m! (2. (2. xo . 2.186) Fourier transform.182) Comparision with (2.. t) = exp[−iω(n + 1 ) t] φn (x) as invariants under the 2 action of the propagator φ(x. . xo .e. one obtains for the r. of (2.172).184) 1 ˜ exp[−iω(n + 1 ) t] φn (x) 2 n! ∞ +∞ = m=0 −∞ dxo Φ(x.183) where the expansion coeﬃcients are functions of x. Noting that the propagator (2. to ). xo . . results in 1 ˜ exp[−iω(n + 1 ) t] φn (x) 2 n! +∞ = −∞ dxo Φ(x. but not of t. to ) exp[−iω(n + 1 ) to ] φn (xo ) .188) ˜ ˜ Equation (2.185) Replacing t → t + to yields ∞ n=0 1 ˜ exp[−iω(n + 1 ) (t + to )] φn (x) 2 n! ∞ +∞ = m=0 −∞ dxo Φ(x. We note that the factor exp(2 mω/ xe−iωt − e−2iωt ) in (2. t − to ) = φ(x.172) is a function of t − to and deﬁning accordingly Φ(x.172) I = exp 2 mω x e−iωt − e−2iωt − Quantum Mechanical Path Integral mω 2 1 x − iωt 2 2 . i.172).h. . t) +∞ −∞ dto 1 ˜ exp[−iω(m + 1 ) to ] φm (xo ) 2 m! (2. 2. which also exhibits such invariance.187) or 1 ˜ exp[−iω(n + 2 ) t] φn (x) +∞ = −∞ ˜ dxo φ(x. t − to ) 1 ˜ φn (xo ) n! (2. We want to inspect the consequences of the invariance property (2. to ) we express (2.. t|xo . t|xo . t|xo . xo .172) in the form ∞ n=0 (2.171) concludes the proof of (2.36 Altogether.171) ∞ W (x. n = 1.171. Accoordingly. one can expand (2.171) can be expanded in terms of e−inωt . .s.188) identiﬁes the functions ψn (x. t) = n=0 1 ˜ exp[ − iω(n + 1 ) t ] φn (x) 2 n! (2. t). 1 exp[iω(n + 2 ) to ] · · · .

194) plays a central role for the Hermite polynomials since it contains. t) provide all stationary states of the quantum mechanical harmonic oscillator.171). (2. characterizes the wave functions φn (x). t)|2 = ˜n (x)|2 . .5: Propagator for a Quadratic Lagrangian 37 ˜ ˜ the functions ψn (x.183). t)|2 = Nn +∞ −∞ n = 0. t) are associated with a time-independent probablity density |ψn (x. ˜ |φ stationary wave functions ψn (x. z) = n=0 1 2 /2 1 2 /2 = = mω e−iωt x (2.190) ˜ is obeyed. t) of the quantum mechanical harmonic oscillator ˜ ψn (x.193) (2.183).183) reads then w(y.195) zn Hn (y) n! ˜ φn (y) .183). according to (2. Expansion (2.189) ˜ dx φ2 (x) = 1 n (2.197) are called Hermite polynomials which are polynomials of degree n which will be evaluated below. The Hermite Polynomials ˜ The function (2. . through expansion (2. We will also argue that the functions ψn (x.192) w(y. z) (2. z) = exp(2yz − z 2 ) . In the following we will characterize the functions φn (x) and determine the normalization constants Nn . t) such that +∞ −∞ 2 dx |ψ(x. To obtain ˜n (x) we simplify the expansion (2.196) where Hn (y) = ey 2 /2 (2. we have identiﬁed then.191) (2. (2. . through the expansion coeﬃcients φn (x) in (2. Expression (2. z) = Hn (y) (2. This follows from ∂n w(y.171) W (x. 1. 2.198) ∂z n z=0 .194) = z2 n=0 ∞ 1 ∞ zn ˜ φn (y) n! (2.2.197) The expansion coeﬃcients Hn (y) in (2. 2 Here Nn are constants which normalize ψn (x.194). For this purpose we introduce ﬁrst closed expressions for φ the new variables y z and write (2. z) z 2 e−y or w(y. t) = exp[−iω(n + 1 ) t] Nn φn (x) . Actually. in a ‘nutshell’ all information on the Hermite polynomials. t) = z 2 e−y where w(y.

closed expressions for Hn (y).200) yields for the ﬁrst Hermite polynomials H0 (y) = 1. 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. ν H2 (y) = 4y 2 − 2. and recursion equations for the eﬃcient evaluation of Hn (y).1: Schematic representation of change of summation variables ν and µ to n = ν + µ and m = ν −µ.196).e. H1 (y) = 2y. Boston.Graham. ∂y n (2. We want to derive now explicit expressions for the Hermite polynomials.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). m lattice (right diagram).Wilf (Academic Press. .E. Massachusetts. For this purpose we expand the generating function (2. The diagrams illustrate that a summation over all points in a ν. i. . Inc. As will become evident in the present case generating functions provide an extremely elegant access to the special functions of Mathematical Physics3 .Patashnik (AddisonWesley. ν+µ (2.L. 2. ν+µ ν H3 (y) = 8y 3 − 12y.194.. normalization factors for φ(y). One calls w(y. among other properties.201) ν−µ µ µ ν−µ Figure 2. Reading.Knuth.196) to ˜ derive. 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 (2.200) One can deduce from this expression the polynomial character of Hn (y). and O. . 1990) is a useful introduction to this tool as is a chapter in the eminently useful Concrete Mathematics by R. that Hn (y) is a polynomial of degree n..196) results in the so-called Rodrigues formula for the Hermite polynomials Hn (y) = (−1)n ey 2 ∂ n −y2 e . .S. The diagrams also identify the areas over which the summation is to be carried out. z) the generating function for the Hermite polynomials.38 Quantum Mechanical Path Integral which is a direct consequence of (2. D. 1989).196) ∂n w(y.198) for the Hermite polynomials can be expressed in a more convenient form employing deﬁnition (2. We will employ (2. µ lattice (left diagram) corresponds to a summation over only every other point in an n.199) Comparision with (2.202) generatingfunctionology by H. (2. The identity (2.

201).1.5: Propagator for a Quadratic Lagrangian and introduce now new summation variables n = ν + µ.207) one can deduce that Hn (y). µ expressend in terms of n. The lattices representing the summation terms are shown in Fig.206) From this expansion we can identify Hn (y) [n/2] Hn (y) = k=0 (−1)k n! (2y)n−2k . according to 2. m = ν − µ 0 ≤ n < ∞. p ≤ x. m must be restricted such that either both n and m are even or both n and m are odd. According to (2. 2 µ = n−m ..194) holds w(−y.208) which agrees with the expressions in (2. H3 (y) = 8y 3 − 12y. Hence. the sum in (2.e. 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 . 39 (2. (2.207) contains only even powers. in fact. From (2. 0 ≤ m ≤ n .205) Since (n − m)/2 is an integer we can introduce now the summation variable k = (n − m)/2 . 2. 2 (2. is a polynomial of degree n.2. hence. it contains only odd powers.204) Since ν. .203) The old summation variables ν. −z) and. it holds Hn (−y) = (−1)n Hn (y) .207) This expression yields for the ﬁrst four Hermite polynomials H0 (y) = 1. k! (n − 2k)! (2.210) from which one can conclude the property (2. . . for odd n. H2 (y) = 4y 2 − 2. otherwise. i. H1 (y) = 2y. m are ν = n+m .209). .197) ∞ n=0 zn Hn (−y) = n! ∞ n=0 (−z)n Hn (y) = n! ∞ n=0 zn (−1)n Hn (y) n! (2. With this restriction in mind one can express (2. (2. In case of even n . (2. z) = w(y.209) This property follows also from the generating function. 0 ≤ k ≤ [n/2] where [x] denotes the largest integer p. k! (n − 2k)! = Hn (y) (2.202) e 2yz − z 2 ∞ = n=0 zn n! ≤n m≥0 n! (−1) n−m 2 n−m 2 ! m! (2y)m . µ are integers the summation over n.

Hn+1 (y) − 2y Hn (y) + 2n Hn−1 (y) = 0 . (2.211) m! n! m=0 n=0 Comparing terms on both sides of the equation yields (2n)! .209) since odd functions have a node at the origin. z). H2n+1 (0) = 0 .194).215) ∂z which can be readily veriﬁed using (2. z) should be equivalent to the relationship (??). 2. z) = 0 (2. H2n (0) = (−1)n Recursion Relationships A useful set of properties for special functions are the so-called recursion relationships. dy n = 1.197) above and given by (2.233) below. Starting point of the derivation is the property of w(y. For Hermite polynomials holds.214) We want to derive (??) using the generating function. 2.196) into the diﬀerential equation (2. . . (2.213) which allow one to evaluate Hn (y) from H0 (y) and H1 (y) given by (2. . . . n = 1. . for example.212) n! This implies that stationary states of the harmonic oscillator φ2n+1 (x). n = 0. One can then readily state which diﬀerential equation of 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 . . Substituting expansion (2. n! (2. dw/dy − 2zw = 0. 2. 2. .194). z) − (2y − 2z) w(y. have a node at y = 0.218) 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. as deﬁned through (2. a factor z reduces the order of Hn by one and introduces also a factor n.40 Quantum Mechanical Path Integral The generating function allows one to determine the values of Hn (y) at y = 0. z) ∂ w(y. Another relationship is d Hn (y) = 2n Hn−1 (y). as given in (2. The reader may verify that w(y. (2. . .208). (2. 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) . namely.217) gives H1 (y) − 2y H0 (y) = 0. For this purpose one considers w(0. indeed satisﬁes the latter relationship. . Hn+1 (y) − 2y Hn (y) + 2n Hn−1 (y) = 0.188. n = 1. 1. a property which is consistent with (2. . 2. .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. (2.

225) We want to derive from the generating function (2. dy n (2. .219) which can be written I(y) = e−y 2 +∞ −∞ dte−(t−iy) = e−y √ 2 2 +∞ −∞ dze−z . z ) e−y 2 +∞ = e2 z z = √ ∞ dy e−(y−z−z ) = .196) the orthogonality properties of the Hermite polynomials.223) dn 2iy t − t2 2 e = (2 i t)n e2iy t − t n dy results. 2 (2. of the Rodrigues formula (2.194.s. ﬁnally. 2 √ π e2 z z (2.194. z) w(y.h.226a).224) +∞ −∞ dt tn e2iy t − t .h.196) yields ∞ n.s. in the integral representation of the Hermite polynomials 2n (−i)n ey √ Hn (y) = π Orthonormality Properties 2 (2.222) Employing this expression now on the r. (2.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. 2. 2 n = 0.228) .2. 2 (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 . (2. 2 (2. .226) π −∞ 2n z n z n n=0 n! Expressing the l. acording to the deﬁnition (2.200) yields Hn (y) = The identity (−1)n y2 √ e π +∞ dt −∞ dn 2iyt − t2 e . 2. through a double series over Hermite polynomials using (2.220) Using (2. 2. 1.247) for a = i one obtains I(y) = and.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 .221) 1 = √ π +∞ −∞ dt e2iyt − t . For this purpose we consider the integral +∞ −∞ dy w(y. e−y 2 π e−y 2 (2. .

2: Stationary states φn (y) of the harmonic oscillator for n = 0. 3. the value of the wave function at y = 0 is positive for n = 0.230) (2. that the wave functions are even for n = 0. 2.231) √ π (2. in harmony with (??). 3. (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 2.189. The normalized states are [c. 2. 4. 3.228) allow us to construct normalized stationary states of the harmonic oscillator. 4 and odd for n = 1. 1. ¡ ¢ ¡ -4 -2 0 2 4 y (2.190)] φn (y) = Nn e−y 2 /2 and for the normalization constants Nn follows from (2. 4. in agreement with our above discussions.2.232) Hn (y) .229) Hn (y) . One can recognize. Normalized Stationary States The orthonormality conditions (2. According to (2. 2. (2. 2. 4 in Fig. One can also recognize that n is equal to the number of nodes of the wave function.233) are presented for n = 0.f. 3. 2. (2.197) holds 2 ˜ φn (y) = e−y /2 Hn (y) .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. Furthermore. negative for n = 2 and vanishes for n = 1. 1.233) .

e. denoted by P (x) and introduced in Sect. w(x) ≡ 1.189) φn (x) = √ 1 2n n! mω π 1 4 e− mωx2 2 Hn ( mω x) . of functions which obey dx f 2 (x) w(x) = < ∞ .228) shows that the orthonogality condition of the Hermite polynomials is in complience with (2.228).5: Propagator for a Quadratic Lagrangian 43 The normalization condition (2.189) by the Jacobian dx/dy. w(x) = xα e−x . 5. g ∈ F . (10.10 and eq. 5.. ??? and eq. The latter are written for polynomials pn (x) in the general form dx pn (x) pm (x) w(x) = In δnm Ω (2.151. 2. Other examples of orthogonal polynomials are the Legendre and Jacobi polynomials which arise in solving three-dimensional stationary Schr¨dinger equations. In = Γ(n+α+1)/n! where Γ(z) is the so-called Gamma function. Comparision with (2. Ω (2.150 . i.236) where w(x) is a so-called weight function with the property w(x) ≥ 0. i. 5 below [c. The Hermite polynomials are so-called orthonogal polynomials since they obey the conditions (2.235) 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.236 . The orthogonal polynomials pn mentioned above have the important property that they form a complete basis in the space F of normalizable functions. 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. using (2. w(x) = 0 only at a discrete set of points xk ∈ Ω (2. and I = 2/(2 + 1).2. (5. In case of the associated Laguerre (α) polynomials.156. +∞[.237) and where In denotes some constants.179] holds Ω = [−1. holds Ω = [0. In case of the Legendre polynomials.233) and (2. ???] and relativistic [see Sect. w(x) = √ exp(−x2 ). by dx = dy mω 1 4 . denoted by Ln (x) and encountered in case of the stationary states of the nonrelativistic [see Sect. 10.237) for Ω = R..239) .234) The explicit form of the stationary states of the harmonic oscillator in terms of the position variable x is then.e. (2. (2. 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.f.459] hydrogen atom. 1]. and In = 2n n! π. 5. (2.238) where the space is endowed with the scalar product (f |g) = Ω dx f (x) g(x) w(x) = < ∞ .231) of the wave functions diﬀers from that postulated in (2. f.

236).. ψ(y.. t) is ∞ ψ(y. t) = n=0 cn (t) e−y 2 /2 Hn (y) . symmetric a.44 As a result holds for any f ∈ F f (x) = n Quantum Mechanical Path Integral cn pn (x) (2.e. t) must be identical to one of the stationary states (2. If one replaces for all f ∈ F: f (x) → w(x) f (x) and. Appendix: Exponential Integral We want to prove +∞ +∞ I = dy1 .246) for det(a) = 0 and real.197) it must hold cn (t) = dn exp[−iω(n + 1 )t] and.242) Let us assume now the case of a function space governed by the norm (2. 2 (2.243) To be consistent with(2.240) where cn = 1 In dx w(x) f (x) pn (x) . pn (x) → w(x) pn (x) the scalar product (2. the states (2. 2. t) which is stationary under the action of the harmonic oscillator propagator (2.147). .247) . In case of n = 1 this reads +∞ −∞ dx ei a x = 2 iπ . except for a single n = no . i. . −∞ dyn ei −∞ n j. a (2.172) holds. Therefore.188.242) and the existence of a normalizable state ψ(y.245) must be time-independent. aT = a. (2. ∞ n. hence.244) For the state to be stationary |ψ(x. i.233). t) = n=0 1 2 dn exp[−iω(n + )t] e−y /2 Hn (y) . Ω (2. (2.241) The latter identity follows from (2.239) becomes the conventional scalar product of quantum mechanics f |g = Ω dx f (x) g(x) . det(a) (2.233) exhaust all stationary states of the harmonic oscillator. the 2 stationary state ψ(y.e. Since the Hermite polynomials form a complete basis for such states we can expand ∞ ψ(y. i. a state for which (2.k yj ajk yk = (iπ)n . The only possibility for this to be true is dn = 0. in particular.e.e. t)|2 .m=0 d∗ dm exp[iω(m − n)t] e−y Hn (y)Hm (y) . n 2 (2.. i.

(2. . .254) ajj . 2. according to (2. .252) ˜ For the determinant of a holds det(˜) = a n ajj ˜ j=1 (2. .k n n yj = ˜ k (S −1 )jk yk .253) as well as det(˜) = det(S−1 aS) = det(S−1 ) det(a) .246) exploits that for any real. The proof of (2. This property allows one to simplify the bilinear ˜ form n yj ajk yk by introducing new integration variables j. . ST S = 1 1 or S = S−1 .. 0 ˜ ˜ S−1 a S = a = . 0 0 .m (S T )jl alm Smk .248) . ann ˜ where S can be chosen as an orthonormal transformation..250) The bilinear form in (2. . det(S) a = (det(S))−1 det(a) det(S) = det(a) .k yj ajk yk n = j.248. ˜ (2. symmetric matrix exists a similarity transformation such that a11 0 .249) n ajk = ˜ l. . .k n m = = j.k yj ajk yk ˜˜ ˜ (2.k n m n y Sj ajk Skm ym ˜ ˜ y (S T ) j ajk Skm ym ˜ ˜ j. . (2.255) . . . yk = k Skj yj . ˜ (2. . .2. One can conclude det(a) = j=1 n (2.e.249) The akk are the eigenvalues of a and are real. 0 ˜ 0 a22 . i.251) where. . .7: Appendix / Exponential Integral 45 which holds for a ∈ C as long as a = 0.246) reads then in terms of yj ˜ n n j. . (2.

Accordingly.261) ∂(y1 . (2. .46 We have assumed det(a) = 0. . yn ) y ˜ ∂yj . i. . yn ) y ˜ = ( det S )2 (2. y −∞ d˜n det y −∞ ∂(y1 . .260) (2. yn ) ∂(˜1 .257) Substitution of the integration variables (2. . . yn ) ∂(˜1 . . holds n Quantum Mechanical Path Integral ajj = 0 ˜ j=1 (2. ajj = 0 . d˜n e y −∞ = (2.264) I = d˜1 . . . det( From (2. n (2. . . . ∂ ys ˜ (2.249) follows 1 = det ST S such that one can conclude det S = ±1 One can right then (2.259) (2.250) +∞ +∞ I = d˜1 .262) (2. . . ˜ for j = 1.. 2. . hence. . .265) k=1 −∞ which leaves us to determine integrals of the type +∞ dx eicx −∞ 2 (2. . .250) allows one to express (2.. .266) . . . .263) (2.258) where we introduced the Jacobian matrix J = with elements Jjs = According to (2.256) such that none of the eigenvalues of a vanishes.258) +∞ +∞ ∂(y1 . . . . ∂(˜1 . 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 −∞ . .e. . .. yn ) y ˜ ei n k akk yk ˜ ˜2 . . yn ) ) = det(S) . . .250) holds J = S and.

268) can be written as the sum of the following path integrals J = J1 + J2 + J3 + J4 .3. the integrand does not exhibit any singularities anywhere in C.268) along the path γ = γ1 + γ2 + γ3 + γ4 displayed in Figure 2. for x. (2.257) holds c = 0. p ∈ IR. We consider ﬁrst the case c > 0 and discuss the case c < 0 further below.c > 0 .266) to the well-known Gaussian integral +∞ dx e−cx −∞ 2 = π . 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 . i.271) .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 √ (2.268) vanishes. according to (2.e.. c (2.270) dx e−cx 2 2 − 2p √ γ4 : z = ix − p J4 = −p i dx eic(ix−p) . Jk = γk dz eicz 2 (2. The contour intergral (2. The contour integral (2.7: Appendix / Exponential Integral 47 where.267) by considering the contour integral J = γ dz eicz = 0 2 (2. One can relate integral (2. 2 since eicz is a holomorphic function.

(2.48 Quantum Mechanical Path Integral Im(z) ip γ2 ¡ γ4 ¡ γ3 ¡ -i p Figure 2.267) one has shown then ∞ dx eicx −∞ 2 = iπ . 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.3: Contour path γ in the complex plain.268) for p = +∞. (2. (2. c ¡ -p γ1 p Re(z) 2 | |e−2cxp | . Hence.272) = 0 . p p→+∞ lim |J2 or 4 | ≤ = p→+∞ 0 lim dx |e−2cxp | 1 − e−2cp 2cp = 0.275) .274) (2. One can state accordingly ∞ J = dx e −∞ icx2 − √ ∞ i dx e−cx −∞ 2 Using 2.273) p→+∞ lim J2 and J4 do not contribute then to integral (2. We will now show that the two integrals J2 and J4 vanish for p → +∞.

2.277) We apply the above results (2. This leads to ∞ icx2 J = dx e −∞ + √ −∞ −i ∞ dx ecx = 0 2 (2.275. c (2.2.276) and (c < 0) ∞ dx eicx −∞ 2 = −iπ = −|c| iπ .277) to (2.279) . if one chooses the same contour integral as (2.265). (iπ)n det(a) (2.255) this result can be expressed in terms of the matrix a I = which concludes our proof.268). but with a path γ that is reﬂected at the real axis.7: Appendix / Exponential Integral 49 One can derive the same result for c < 0. It holds n I = k=1 iπ = akk ˜ (iπ)n .278) Noting (2. n ˜ j=1 ajj (2.

50 Quantum Mechanical Path Integral .

t) satisﬁes a partial diﬀerential equation. x3 51 . t) ψ(r + s. t + ) = Ω m 2πi 3 2 exp i m (r − r0 )2 − 2 U (r. Generalizing (2. even in x1 . of this equation in terms of powers of and we will demonstrate that the terms of order require that ψ(r. .2) d 3 r0 m 2πi 3 2 exp i m (r − r0 )2 − 2 U (r.h. It holds then according to (2. 2.21).3) In order to carry out the integration we set r0 = r + s and use s as the new integration variable.4) − U (r. t + ) = × exp i 2 +∞ +∞ +∞ −∞ dx1 −∞ dx2 −∞ dx3 m x2 + x2 + x2 1 2 3 m 2πi 3 2 × (3.Chapter 3 The Schr¨dinger Equation o 3. This yields ψ(r. x3 )T .s. t) . namely the Schr¨dinger equation. and the r. (3.1 Derivation of the Schr¨dinger Equation o We will consider now the propagation of a wave function ψ(r. t) .h. t) . t + ) = Ω d3 r0 φ(r.20. t + |r0 . t) by an inﬁnitesimal time step .5) ψ(r. t + |r0 . t) = From this follows ψ(r. Notable exceptions are non-stationary systems involving time-dependent linear and quadratic Lagrangians. x2 . t) ψ(r0 . The propagator in (3. but by no means all. For many situations. the Schr¨dinger o o equation provides the simpler avenue towards describing quantum systems than the path ingral formulation of Section 2. (3.20) to evaluate the propagator. (3. We will denote the components of s by (x1 . t) . x2 .s. In the limit of very small it is suﬃcient to employ a single discretization interval in (2.20) to R one obtains then for small φ(r.1) We will expand the l.1) can be expressed through the discretization scheme (2. t) ψ(r0 .

52 Schr¨dinger Equation o It is important to note that the integration is not over r. t). U (r.5) assuming that only the leading terms contribute. 1 2 3 2 ∂2 j=1 xj ∂x2 ψ(r. O( 5 2 ) .6) make contributions of the order O( 3 2 ) . 1. t) + j=1 xj ∂ 1 ψ(r.7) According to (2. the terms collected in (3. . t) = ψ(r. t) j 1 12 .10) It is now important to note that in case of integral (3.12) Here one needs to note that we are actually dealing with a three-fold integral. a I2 (a) = − 3 4a2 iπ . 1 4 3 2 2 ∂4 j. (3. x2 . a (3.6) 3 4 ∂4 j=1 xj ∂x4 ψ(r. t) j . t) and the kinetic energy term.11) holds 3 m 3 iπ 2 2 × = 1 (3.. of the expansion of ψ(r + s.g.4) one identiﬁes 2 1 = a m = O( ) (3. a supposition which will be examined below.11) and. x3 )T . consequently..k=1 ∂2 ψ(r.. i ∂a Starting from (3.k=1 xj xk ∂x2 ∂x2 ψ(r. only terms of the expansion (3. x2 . t) + . It is then suﬃcient to consider the terms ψ(r. ∂xj ∂xk (3.. x3 . t) j k ..13) 2πi a . e. we need then to evaluate integrals of the type +∞ In (a) = −∞ dx x2n exp i a x2 . t) is a constant with respect to this integration.9) iπ a (3. Obviously. Since the kinetic energy contribution in (3.8) one can evaluate recursively all integrals In (a). Since the latter contributes to (3.36) holds I0 (a) = Inspection of (3. O( 7 2 ) .4) only for small x2 + x2 + x2 values 1 2 3 we expand 3 ψ(r + s. It holds In+1 (a) = I1 (a) = i 2a iπ . t) ..5) which are even separately in all three coordinates yield non-vanishing contributions. n = 0. 2 (3.4) is even in all three coordinates x1 .7) shows 1 ∂ In (a). O( 7 2 ). According to (3. t) + ∂xj 2 3 xj xk j. (3. but over s = (x1 .8) (3. . The integration involves only the wave function ψ(r +s.

19) ∂t where 2 2 ˆ H = − + U (r.18) This is the celebrated time-dependent Schr¨dinger equation.2 Boundary Conditions The time-dependent Schr¨dinger equation is a partial diﬀerential equation. i. Usually. t)ψ(r.17) 2m ψ(r. t). (3. t) + and exp − i U (r. t) . t) + O( 2 ) . t) ψ(r.14) This expansion in terms of powers of suggests that we also expand ∂ ψ(r. one speciﬁes the so-called initial condition. t) (3. ψ(r. t) = ∂t 2 − 2 2m + U (r. t) = H ψ(r. t) = ψ(r.15) ψ(r. 2nd o order in the spatial variables and linear in the solution ψ(r. t) . We will derive brieﬂy the type of boundary conditions encountered.16) Inserting this into (3.10).20) 2m 3. The following general remarks can be made about the solution.g. this equation is trivially satisﬁed to order O( 0 ). t) + ∂ ∂t ψ(r. using (3. t) + O( 2 ) ∂t i U (r. t) − + i 2 2 i U (r. t) + 1 2i 4 m 2 53 ψ(r. Obviously. t1 ) = f (r). t) + O( 2 ) .. a solution is thought for t ≥ t0 and the solution is speciﬁed at the intial time t0 . t) + O( 2 ) . e. 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. This equation is often written in the o form ∂ ˆ i ψ(r. (3. (3. Due to its linear character any linear combination of solutions of the time-dependent Schr¨dinger o equation is also a solution. The 1st order time derivative requires that for any solution a single temporal condition needs to be speciﬁed.. 1st order in time. In order O( ) the equation reads i ∂ ψ(r. t) = 1 − (3.2: Boundary Conditions and one can conclude. t + ) = exp − i U (r. t) ψ(r.3. t) (3. 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 .14) results in ψ(r. ψ(r. (3. t + ) = ψ(r.e.

f ∗ (r) d3 rg(r) Ω 2 = − 2m d3 rf ∗ (r) Ω 2 g(r) − f ∗ (r) (3.26) where ∂Ω da · A(r) denotes an integral over the surface ∂Ω of the volume Ω.27) ˆ We will postulate below that H is an operator in H which represents energy. D. (John Wiley.. t). 1975). 2nd Ed. hence. . that H is 2 . The hermitian property. g|r) = − 2 2m ( f ∗ (r) g(r) − g(r) f ∗ (r) ) (3. Calssical Electrodynamics. Jackson. Chapter 1. The reader is advised to consult a reference text on ‘Linear Algebra’ to follow this argument. t) g(r) − Ω d3 rg(r) U (r.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. J.22) ˆ where H is deﬁned in (3. for example. In (3.54 Schr¨dinger Equation o linear vector spaces of functions f (r) in which a scalar product between two elements f. Interchanging f ∗ (r) and g(r) results in g| H |f Ω = Ω ˆ d3 r g(r) Hf ∗ (r) .28) See.20). in principle. 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. implies hermitian ˆ f | H |g 1 2 Ω ˆ = g| H |f Ω (3. g|r) (3.21) Ω This leads one to consider the integral f | H |g Ω = Ω ˆ d3 rf ∗ (r) H g(r) (3. g ∈ H is deﬁned as follows f |g Ω = d3 rf ∗ (r)g(r) (3. Since energy is a real ˆ ˆ quantity one needs to require that the eigenvalues of the operator H are real and. (3.23) ˆ Since H is a diﬀerential operator the expressions (3.23).25) ˆ f | H |g Ω ˆ = g| H |f Ω + ∂Ω da · P (f ∗ . however.26) the vector–valued function P (f ∗ . 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. diﬀer from each other. New York. g|r) is called the ˆ concomitant of H and is P (f ∗ .22) and (3.

Since the total probability of ﬁnding the particle anywhere in space is d3 r|f (r)|2 = 1 (3. t) on a single connected surface ∂Ωj only either one of the conditions (3. obey (3.30) Note that these boundary conditions are linear in f . In this case one can postulate both conditions (3. κ > 0 or like r−α . t) = ω d3 r |ψ(r.e. ∂Ω = ∂Ω1 ∩ ∂Ω2 ∩ ∂Ω3 ∩ . t)|2 . 3. i. (3. . The probability to observe the particle anywhere in the subvolume ω ⊂ Ω is p(ω.30) can be postulated.g. t) + ψ(r. t) is ∂t p(ω.29) In fact. namely. i. t) which describes the state of o a particle moving in the potential U (r. like exp(−κr).31) |r|→∞ (3.3 Particle Flux and Schr¨dinger Equation o The solution of the Schr¨dinger equation is the wave function ψ(r.. t)∂t ψ(r. if f and g satisfy these conditions than also does any linear combination αf + βg. .. t). t) ] . and that the density decays rapidly when one moves away from that area. in which case ∂Ω1 is the inner sphere and ∂Ω2 is the outer sphere. i. r ∈ ∂Ω or da · f (r) = 0 ∀r. |ψ(r. 3.3.e. in this case also all derivatives of f (r) vanish at inﬁnity.. 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. e. t)|2 . (3. r ∈ ∂Ω (3. Often the closed surface of a volume ∂Ω is the union of disconnected surfaces3 ∂Ωj .29. to avoid discontinuities in ψ(r.34) An example is a volume between two concentric spheres. 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. (3. The observable directly linked to the wave function is the probability to ﬁnd the particle at position r at time t. therefore. we can only allow functions which make the diﬀerential da · P (f ∗ .31) usually arises when a particle existing in a bound state is described.32). In either case does f (r) and all of its derivatives vanish asymptotically. However.30) each condition holding on an entire surface ∂Ωj .e. It must hold then for all f ∈ H f (r) = 0 ∀r. t)∂t ψ ∗ (r.33) The time derivative of p(ω.32) the wave function must decay for |r| → ∞ rapidly enough to be square integrable.29.3: Particle Flux 55 and. The latter property stems from the fact that the boundary condition (3. t) = ω 3 d3 r [ ψ ∗ (r. g|r) vanish on ∂Ω. α > 2.

Note that for a normalized wave function the quantity ρ(r. One refers to such solution as normalized. t)r. ψ|r. t) = 0 . (3. We will assume in the remainder of this Section that the solutions discussed are normalized. 3.19) and its conjugate complex4 −i yields i ∂t p(ω. t) − ψ ∗ (r. Accordingly the probability to observe the particle anywhere in the total volume Ω is constant. t).37) can be expressed through a volume integral according to da · A(r) = ∂ω ω (3.37) If one applies this identity to ω = Ω one obtains according to (3. t) ψ ∗ (r. t) H ψ ∗ (r. t) − ψ(r.27) one can express this j(r. (3. t) such that d3 r|ψ(r. t) = ω Schr¨dinger Equation o ∂ ∗ ˆ ψ (r. (3.36) According to (3. t) ψ(r. t) + ω · j(r. The surface integral (3. t) . t) = 0 (3.43) (3. Using (3.27) this can be written i ∂t p(ω.41) holds for any volume ω ⊂ Ω one can conclude ∂t ρ(r.26.39) d3 r · A(r) (3.37) d3 r ∂t ρ(r.41) where j(r.18) by any complex number and accordingly one can deﬁne ψ(r. t) H ψ(r. t) = 0. t) ∂t ˆ ˆ ψ ∗ (r. t) ] .44) ˆ Note that the Hamiltonian H involves only real terms.40) One can rewrite then (3. t)|2 is a probability density with units 1/volume. t)|2 = 1 Ω (3. t) + 4 · j(r.35) d3 r . t) = [ ψ(r.30) ∂t p(Ω. One can multiply the solution of (3. t) (3.56 Using (3. (3.29. ψ(r. 3. t) = ∂ω da · P (ψ ∗ (r. A natural choice for this constant is 1.42) 2mi Since (3. t) . t) = P (ψ ∗ .38) holds. t) = |ψ(r. . t) = H ψ ∗ (r.

(2.48) that the general solution is of the form ψ(r. t) has vanishing ﬂux anywhere. t0 ) = [2π]− 2 Ω∞ 3 ˜ d3 k φ(k) exp i(k · r − ωt0 ) . Equation (3.43) that any real wave function ψ(r.50) 3 ˜ d3 k φ(k) exp i(k · r − ωt) Ω∞ (3.18) in Ω∞ = R in the case o U (r.125)] and in a quadratic [c. 2.51) . 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.49) ˜ Obviously.4: Free Particle 57 The interpretation of j(r. One can readily show by insertion into (3. t0 ) determines φ(k). 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. t) (3. t) . t) = − ψ(r. (3. t) = 0 2 ∂ 2 i ψ(r.e..71)]. Note that j(r. (3. for f (r) ∈ R.47) . (3.f. (2. equal to the velocity of the particle. This ﬁnding is consistent with the present evaluation of the particle ﬂux: a factor exp(ipo xo / ) gives rise to a ﬂux po /m. i.41) written in the form ∂t ω d3 r ρ(r. This follows directly from an inspection of Eq. 3.48. 2m (3. t) = [2π]− 2 where the dispersion relationship holds ω = k2 . t) is that of density ﬂux. t) = − ∂ω da · j(r.f.105.45) Obviously.148. m (3.49) reads at t = t0 ψ(r. (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) . (2.4 Solution of the Free Particle Schr¨dinger Equation o We want to consider now solutions of the Schr¨dinger equation (3.f.46) i..3. t) gives rise to a descrease of the total probability in volume ω due to the disappearence of probability density at the surface ∂ω. and for particles moving in a linear [c. arises solely from the complex factor exp(i k · r).e.167)] potential. 2. It is of interest to note from (3. 2. Such case arose in Sect. In Sect. t) points in the direction to the outside of volume ω. the initial condition at ψ(r. 2 for a free particle [c. j(r.48) ∂t 2m which describes the motion of free particles.

t0 ) (3.31) applies. Ω∞ (3. (3.52) above ψ(r.54) to the case of non-vanishing potentials U (r). t|r0 . t|r0 . t0 ) ψ(r0 . evaluating the integral in (3. (t − t0 ) (3.e. hence. In case of the harmonoc oscillator the expansion can be stated in a closed form given in (4. i.54) an alternative representation of the propagator (2.54) valid for this case.5). i. 3. In fact. To show this one needs to note 1 2π = +∞ dk1 exp −∞ ik1 (x − x0 ) − 1 2 i 2 k2 1 2m .51) to be square integrable it follows according to the properties of the Fourier–transform that ψ(r.e.49).247).47).49.81) . The general form for this propagator involves an expansion in terms of a complete set of eigenfunctions as in (3.70) derived below for a particle in a box and the harmonic oscillator.114) and (4. The solution as stated is deﬁned in the inﬁnte space Ω∞ = R .58 The inverse Fourier transform yields 3 ˜ φ(k) = [2π]− 2 Schr¨dinger Equation o d3 r0 exp(−ik · r0 ) ψ(r0 . The ensuing solutions are called wave packets.56) and using (2. derive an expression similar to (3. If one chooses the initial state f (r) deﬁned in (3. respectively. t) as given by (3.51. 3. that the “natural boundary condition” (3.54) yields (2.53) where φ(r. in (2.49) is square integrable at all subsequent times and.. t0 ) = 1 2π 3 d3 k exp Ω∞ ik · (r − r0 ) − i 2 k2 2m (t − t0 ) .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. t) = Ω∞ d3 r0 φ(r.47). We have identiﬁed then with (3.52) We have not speciﬁed the spatial boundary condition in case of (3.54) This expression obviously has the same form as postulated in the path integral formulation of Quantum Mechanics introduced above. Comparision with Path Integral Formulation One can write solution (3.. Below we will generalize the propagator (3. t0 ) .

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. The remaining integration over k leads to the integral +∞ −∞ dk exp ikx − 1 2 (k − po 2 2 ) δ − i k2 m (t − t0 ) (3.30). Accordingly.54) is +∞ ψ(x. (3.63) . We have demonstrated then in this case that the Schr¨dinger formulation of Quantum o Mechanics is equivalent to the Feynman path integral formulation. (3. (3.49.64) (3.56)].29. t) is given by (??). t) = f (t) φ(r) .57–3. r ∈ ∂Ω (3.27) vanish on the surface ∂Ω of Ω. 3.f. t) = −∞ dx0 φ(x.61) × exp − (x − po m t)2 2δ 2 (1 + 2 t2 ) m2 δ 4 (1 − i . a result which is identical to the expression (??) obtained by means of the path integral propagator (2. 3. 3.59) which is solved through completion of the square in the exponent [c. t|x0 .20) which are of o the form ψ(r.55.60) yields with t0 = 0 ψ(x. t0 ) = 1 2π +∞ dk exp −∞ ik (x − x0 ) − i 2m (t − t0 ) .4: Free Particle Free Particle at Rest 59 We want to apply solution (3.46).60) =???? Combining (3. Stationary States We consider now solutions of the time-dependent Schr¨dinger equation (3.62) We will restrict the space of allowed solutions to a volume Ω such that the functions also make the concomitant (3. t) 2 k2 (3.3.e.52) to the case that the initial state of a 1-dimensional free particle ψ(x0 .57) where φ(x. (3. r ∈ ∂Ω or da · φ(r) = 0 ∀r. the boundary conditions are φ(r) = 0 ∀r.. i. 3. The 1-dimensional version of (3. t|x0 .53. t0 ) ψ(x0 . 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. the functions obey boundary conditions of the type (3.19.

t) vanishes on the surface of ∂Ω. In fact. (3. At this point we state without proof that.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.62) exists.e.69) If these two equations can be solved simultaneously a solution of the type (3. (3. h1 (r) = φ(r).71) . One can conclude that the probability density for the state (3.62) we insert it into (3. (3. g2 (t) = f (t).e.. the eigenvalues of the operator H. We will demonstrate that solutions of the type (3. (3.45) that the total probability d3 r ρ(r.42) the ﬂux j(r. 2. in general. t)|2 = |φ(r)|2 . This yields an expression g1 (t) h1 (r) = g2 (t) h2 (r) (3. We want to demonstrate this property now.66) i. t) = Ω Ω d3 r |ψ(r. for functions required to obey boundary conditions (3.67) (3. solutions exist only for a set of discrete E ˆ values. the solutions o (3.e.64).69) is called an eigenvalue problem.19). i. We may note in passing that solutions of the type (3.62) have the particular property that the associated probability distributions are independent of time. i. This can hold only if |f (t)| is time-independent.31). namely f (t) = f (0) exp − i Et .60 Schr¨dinger Equation o and aﬀect only the spatial wave function φ(r). according to (3.63. It is important to realize that the separable solutions (3.62) are special solutions of the timedependent Schr¨dinger equation. t)|2 = |f (t)|2 Ω d3 r |φ(r)|2 (3. We denote the corresponding solution by φE (r). .65) is constant. 3. At this point we will accept that solutions φ(r) of the type (3. In order to further characterize the solution (3. We will encounter many such problems in the subsequent Sections. .68) ˆ where g1 (t) = i ∂t f (t).69) exist. It turns out that a solution for f (t) can be found for any E. We must postulate therefore ∂t f (t) ˆ H φ(r) = E f (t) = E φ(r) . It follows from the observation that for the solution space considered (3.. hence. a common case is Ω = Ω∞ and the ‘natural boundary condition’ (3.. however. α ∈ R .27) holds and.62) do exist and we will characterize the two factors of the solution f (t) and φ(r).62) which consist of two factors. is time-independent. It follows then from (3. As pointed out above. One calls such states stationary states. by no means all solutions are of this type. n = 1. We have then shown that ψ(r.70) The task of ﬁnding solutions φ(r) which solve (3. and h2 (r) = H φ(r). often only for a discrete set of values En . The identity (3. if f (t) = eiα . . one factor depending only on the time variable and the other only on the space variables are called separable in space and time. t) = f (0) exp − i E φE (r) where ˆ H φE (r) = E φE (r) .62) is |ψ(r. for the eigenvalue problems in the conﬁned function space.

The resulting total energy values E are positive. E ∈ R. t) = exp − E t 2 φE (r) . We will show in Section 5 that E can be interprerted as the total energy of a stationary state. We will wave this assumption as we always need to do later whenever we deal with unbound particles. a property which is to be expected since the energy is purely kinetic energy which. particles scattered of a potential.43) is j(r.74) is actually best labelled by an index k.71) this yields E Ω d3 r φ∗ (r) φE (r) = E ∗ E Ω d3 r φE (r) φ∗ (r) . (3.74) using exp i k · r = ik exp i k · r .4: Free Particle 61 is a solution of the time-dependent Schr¨dinger equation (3. We want to prove now that the eigenvalues E which arise in the eigenvalue problem (3. t) = exp − i 2 k2 2m t exp i k · r (3.g. t) = |N |2 k . Obviously. . should be positive..71) are. As a result we cannot postulate in the present case that wave functions are localized and normalizable.76) according to (3. real. The corresponding stationary solution ψ(r.74) The classical free particle with constant energy E > 0 moves without bounds in the space Ω∞ . ˆ d3 r φ∗ (r) H φE (r) = E Ω Ω ˆ d3 r φE (r) H φ∗ (r) . (3. E (3.75) 2m One can ascertain this statement by inserting the expression for φk (r) into the eigenvalue problem posed in (3. in fact. m (3. one can interpret then k as the magnitude of the momentum of the particle.19.76) has kinetic energy 2 k 2 /2m.66) E must be real. i.3. hence. of course. 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 . e. Stationary State of a Free Particle We consider now the stationary state of a free particle described by i ψ(r. The ﬂux corresponding to (3.28) for the special case that f and g in (3. E (3.28) both represent the state φE (r). − 2 2m φ(r) = E φ(r) .20).72) According to (3. We start our proof using the property (3. = E.e. k ∈ R 2 k2 φk (r) = N exp i k · r .73) from which follows E = E ∗ and. o According to (3. The solution φE (r) corresponding to the eigenvalue problem posed by (3.

We can expect. 5 . n ∈ N.e. a] ⊂ R which vanish on the surface ∂Ω1 = {−a. 2a n = 1. therefore. t) = − d2 ψ(x.82) satisfy the diﬀerential equation in (3. .80) need to be satisﬁed are φE (e. (3. φ(±a) = 0 .81.82) have the property that either both boundary conditions are satisﬁed or none. t) . so-called even solutions obeying φ(x) = φ(−x) φE (x) = A coskx . we have to consider only one boundary condition. let say the one at x = a.79) The stationary solutions have the form ψ(x.o) (−a) = 0 (3. two types of solutions. 3.62 Schr¨dinger Equation o 3.. .80).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. The boundary conditions which according to (3. Solutions of the Schr¨dinger Equation in F1 o The time–dependent solutions satisfy i ∂t ψ(x. t) = exp(−iEt/ )φE (x) where φE (x) is determined by 2 d2 − φE (x) = E φE (x) .84) . f ∈ F1 = {f : [−a. . 3.81) they are kn = nπ . (o) 2 k2 (e) 2 k2 2m = E (3. (3. We will refer to this as a particle in a one-dimensional ‘box’. In case of the even solutions (3. 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. It turns out that this boundary condition can only be satisﬁed for a discrete set of k–values kn . a] ⊂ R assuming that the potential outside of this interval is inﬁnite.78) (3.o) (a) = 0 and φE (e. i. and so-called odd solutions obeying φ(x) = −φ(−x) φE (x) = A sinkx. hence. We assume.80) 2m dx2 We note that the box is symmetric with respect to the origin.83) The solutions (3. f continuous.82) One can readily verify that (3. that the solutions obey this symmetry as well. Hence.81. a}. (3. a] ⊂ R → R.81) 2m = E. 2m dx2 2 (± ) = } (3.

This condition implies for the even states +a |An |2 −a dx cos2 nπx = |An |2 a = 1 . n = 2.81.82). (3.86). . 3.85) (3. 2a n = 1. . respectively. x) = An cos n and φ(o) (a. i. . (3. . . 2a n = 1. .84.86). 6 . (3. .82). (3. 2.. It is desirable to normalize the wave functions such that +a −a dx |φ(e.86) satisfy the boundary condition since for such kn nπa sin(kn a) = sin 2a = sin nπ 2 = 0. 4.90) 2a The wave functions represent stationary states of the particle in a one-dimensional box.93) The normalization constants are then An = (3. 3. 4. (3. . 3 .) The Energy Spectrum and Stationary State Wave Functions The energy values corresponding to the kn –values in (3.1). x) = An sin n nπx .94) .91) holds. (3. n is assumed to be even.92) and for the odd states +a |An |2 −a dx sin2 nπx = |An |2 a = 1 . a n = 2. are En = 2π2 8ma2 n2 . 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 .88) where the energies for odd (even) n–values correspond to the even (odd) solutions given in (3. . . 4. = 0. . By counting the number of their nodes one can determine the energy ordering of the wave functions.87) (Note that.o) (a. .81) and (3. 6 . The wave functions for the ﬁve lowest energies En are presented in Fig. n = 1. φ(e) (a.82) only the kn –values cos(kn a) = cos kn = nπ .5: Particle in One-Dimensional Box since for such kn 63 nπa nπ = cos 2a 2 In case of the odd solutions (3. 5 . 5 . 3. according to (3. 2a n = 2. 2a 1 . (3. (3.3. . 6 .89) nπx . according to the dispersion relationships given in (3. 3.e. x)|2 = 1 n (3.

1: Eigenvalues En and eigenfunctions φn (e.o) (a. 2. 3.64 Schr¨dinger Equation o E h2 ] [ 8 m a2 25 ¢ 16 9 4 Figure 3. x) for n = 1. 5 of particle in a box. 4. ¡ 1 -a a x .

Similarly.98) The latter property is obviously true for n = m.. too. for ∞ f (x) = n=1 dn φn (a. (3. .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. (i) n. it holds: f |f Ω1 = 0 → f (x) ≡ 0. x) = together with the scalar product5 +a 65 (3.3.e. i.98) ∞ f |f 5 Ω1 = n=1 d2 . it holds φn |φm Ω1 = δn m . The latter case leads to integrals φn |φm Ω1 = 1 a +a dx cos −a nπx mπx sin .101) +a −a dx cos (n−m)πx − cos (n+m)πx 2a 2a and. 2a 2a (3.90. m even φn |φm 1 a Ω1 = 1 a +a nπx −a dx sin 2a sin mπx = 2a (3. x) (3. n = m. one obtains for n. 3. . 2. Hence we need to consider only the ﬁrst two cases. In fact. n = 1. In case of n = m we have to consider three cases.. i. . x) . 3. m odd. 3.97) form an orthonormal basis set.} where φn (a.99) Since in this case the integrand is a product of an even and of an odd function. a] are N. 4. In case of n.103) We will show in Section 5 that the property of a scalar product do indeed apply. . f.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. In particular. g ∈ F1 (3. n (3. (3. . 3.95) 1 a cos nπx 2a sin nπx 2a for n = 1. . the integrand is odd. m both odd. (ii) n. Because of the property (3.89.98) the elements of B1 must be linearly independent. 5 .94) B1 = {φn (a. 6 . the integral vanishes. this integral vanishes. . N ≥ 1. for n = 2.102) holds according to (3. . Obviously. hence. and (iii) the mixed case. m both even.e. the integral arises φn |φm 1 a Ω1 = 1 a +a nπx −a dx cos 2a cos mπx = 2a (3.96) f |g Ω1 = −a dxf (x) g(x) . the integrals vanish.

105) to the interval [−a. x0 ) ψ(x0 .98) one obtains +a dx0 φm (a. t0 ) = n=1 dn φn (a. t0 ) .79) requiring the o initial condition cn (t0 ) = 1 . Evaluating the Propagator We can now use the expansion of any initial wave function ψ(x. The functions f are periodic with period 2a. x) cn (t) −a dx0 φn (a. 4. .} and {bn . t0 ) in terms of eigenfunctions φn (a. −a (3.109) Insertion of (3. there exist real constants {an .78) can be expressed as a linear combination of the elements of B1 deﬁned in (3. x) ∂t cn (t) −a dx0 φn (a. x0 ) ψ(x0 . . . 1. Restricting the expansion (3.. however. . t0 ) (3. 5 . in (3.95. 2. (3. such demonstration can be based on the theory of Fourier series. n = 2.107) Inserting this into (3. Hence. . t) = n=1 φn (a.e. . any element of the function space F1 deﬁned in (3.66 Schr¨dinger Equation o f (x) ≡ 0 implies f |f Ω1 = 0 which in turn implies dn = 0 since d2 ≥ 0.95) is also complete.106) and generalizing to t ≥ t0 one can write ∞ +a ψ(x. This implies that only the trigonometric functions which are elements of B1 enter into the Fourier ˜ series. We like to show ﬁnally that the basis (3.105) must vanish. 2. . (3. 3. i. n = 1. n = 0. the coeﬃcients an . We have then shown that any f corresponding to elements of F1 can be expanded in terms of elements in B1 .e. they can be expanded in terms of a Fourier series. (3. 3. 3.108) into the Schr¨dinger equation yields o +a ∞ n=1 φn (a. . ( ) = ([ + ] − ) (3. . For this purpose we expand ∞ ψ(x. It follows that B1 n deﬁned in (3. . t0 ) = +a ∞ i n=1 − En φn (a. and bn . .. a] yields then also an expansion for any element in F1 and B1 is a complete basis for F1 .106) Using the orthonormality property (3. .104) ˜ where [y]a = y mod2a.108) where the functions cn (t) are to be determined from the Schr¨dinger equation (3. Demonstration of completeness is a formidable task.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. In the present case.} such that ∞ ∞ nπx nπx ˜ = + an sin (3. i.110) . n = 1. x0 ) ψ(x0 . t) at times t > t0 . . x) cn (t) −a dx0 φn (a. For this purpose we extend the deﬁnition of the elements of F1 to the whole real axis through ˜ f : R → R. x0 ) ψ(x0 . x) to obtain an expression for ψ(x. x) .96). . m = 1. t0 ) = dm . Accordingly.95) is an orthonormal basis.

t0 ) = δ(x − x0 ) as can be readily veriﬁed using (3. t) = −a dx0 φ(x.78). rather than by an integral (3. We have identiﬁed then with (3. This solution can be written +a ψ(x. x0 ) . according to (3. t|x0 .54) as in the case of the free particle system which does not exhibit any bound states. k0 = .109) are cm (t) = exp i − Em (t − t0 ) . t0 ). t) for any initial condition ψ(x. 3.79) which lies in the proper function space (3. 4 a (3. t|x0 . a] yields.5: Particle in One-Dimensional Box Multiplying both sides by φm (a. t|x0 . The respective initial condition is φ(x.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. .2 presents the probability distribution of the particle at subsequent times.108.e. In the present system which is composed solely of bound states the propagator is given by a sum. It is of interest to note that φ(x. One can recognize that the particle moves ﬁrst to the left and that near the right wall of the box interference eﬀects develop. The particle moves then to the left.112) Equations (3.112) determine now ψ(x. t0 ) (3. i ∂t cm (t) = − Em cm (t) . in (2.113) where φ(x. t|x0 . 67 (3. t0 |x0 . t0 ) = 1 2πσ 2 1 4 exp − x2 0 + i k0 x0 2σ 2 . t0 ) itself is a solution of the time-dependent Schr¨dinger o equation (3. x) and integrating over [−a. t0 ) = ∞ φn (a..5). Often the propagator φ(x. Example of a Non-Stationary State As an illustration of a non-stationary state we consider a particle in an initial state ψ(x0 . σ = a 15 . (3. The interference pattern begins to ‘smear out’ ﬁrst. cm (t0 ) = 1 .114) the representation of the propagator for a particle in a box with inﬁnite walls. but the collision with the left wall leads to new interference eﬀects.114) This expression has the same form as postulated in the path integral formulation of Quantum Mechanics introduced above. The last frame shows the wave front reaching again the right wall and the onset of new interference. (3.111) The solutions of these equations which satisfy (3. In case that diﬀerent boundary conditions hold the propagator will be diﬀerent as well. x) exp n=1 i − En (t − t0 ) φn (a. t0 ) is also referred to as a Greens function. Figure 3. being reﬂected at the right wall.98). t0 ) ψ(x0 . i.113). Note that the propagator has been evaluated in terms of elements of a particular function space F1 .3. the elements of which satisfy the appropriate boundary conditions.

96 m a h2 |Ψ|2 2 t = 1.20 m a h2 1 πσ 1 πσ -a a -a a |Ψ|2 2 t = 1.72 m a h2 1 πσ 1 πσ -a a -a a |Ψ|2 2 t = 0.00 m a h2 |Ψ|2 2 t = 0.16 m a h2 1 πσ 1 πσ -a a -a a Figure 3.24 m a h2 1 πσ k = 15 a 1 πσ -a a -a a |Ψ|2 2 t = 0.68 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.92 m a h2 |Ψ|2 2 t = 2.48 m a h2 |Ψ|2 2 t = 0.68 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. .2: Stroboscopic views of the probability distribution |ψ(x.

x3 ) is an element of the function space F3 deﬁned in (3. x1 = ±a1 } ∪ {(x1 . ( . x3 )T ∈ Ω. hence. x2 . a1 ] ⊗ [−a2 . a3 ] ⊂ R = {(x1 . The description employed a space of functions F1 deﬁned in (3. (3. An important property of this basis and. a2 ] ⊗ [−a3 .6 Particle in Three-Dimensional Box We consider now a particle moving in a three-dimensional rectangular box with side lengths 2a1 .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. of the space F1 is that the elements of the basis set can be enumerated by integer numbers. x2 . i.118) which are stationary states.117) 2m 2 2 2 ∂1 + ∂2 + ∂3 (3. x2 . f continuous. (3. i. Placing the origin at the center and aligning the x1 ..120) ˆ Since the Hamiltonian H is a sum of operators O(xj ) each dependent only on a single variable. j = 1. ˆ = O(x1 ) + O(x2 ) + O(x3 ). . ) = ∀( . x2 . 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. . x2 = ±a2 } ∪ {(x1 . We have deﬁned in the space F1 a scalar product (3. 2. x3 ) = j=1 φ(j) (xj ) (3. 3 (3. φ(j) (±aj ) = 0 . x3 )T ∈ Ω. can be counted. x3 = ±a3 } . x2 . x3 ) . This property allowed us to evaluate the propagator (3. x2 .114) in terms of which the solutions for all initial conditions can be expressed.116) and obeys the partial diﬀerential equation ˆ H φE (x1 . x2 . A complete basis of F1 is given by the inﬁnite set B1 (3. x3 . one can express H 3 φ(x1 . x3 . ˆ H = − 2 (3. x2 .119) where φE (x1 . 3.121) where − 2 2m 2 ∂j φ(j) (xj ) = Ej φ(j) (xj ) . 2a2 . t) .95).e. We seek then solutions of the time-dependent Schr¨dinger equation o ˆ i ∂t ψ(x. t) = H ψ(x1 . x2 .e.3. t) = exp − E t φE (x1 .78). x3 ) (3. The corresponding solutions have the form i ψ(x1 .122) .116) where Ω is the interior of the box and ∂Ω its surface Ω ∂Ω = [−a1 . )T ∈ ∂ } . x2 . x2 . x3 ) = E φE (x1 .97) with respect to which the eigenfunctions are orthonormal. x3 )T ∈ Ω.. 2a3 .

a2 . if all three lengths a1 . a2 . x3 –axes. 3. t|r0 .n2 . x1 ) φn2 (a2 . x3 ) . 3.80) shows that the solutions of (3.120) can be written φ(n1 . (3. x2 . Actually. a1 = a2 = a3 = a then rotation around the x1 .123) 8m n2 n2 n2 1 2 3 + 2 + 2 a2 a2 a3 1 .n3 ) (a1 . . n1 . Comparing (3. n 3 and E(n1 . . . r) exp i − E(n1 n2 n3 ) (t − t0 ) φ(n1 . namely rotation by π around the x1 . in the present case .} ∞ (3.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. x3 –axes by π/2 also leaves the system unaltered. . x3 ) n1 . n2 . a3 . x1 .n2 . For example. (3. n3 = 1.n2 . This additional symmetry is also reﬂected by degeneracies in the energy levels.126) Symmetries The three-dimensional box conﬁning a particle allows three symmetry operations that leave the box unchanged. 2. . a3 .124) The same considerations as in the one-dimensional case allow one to show that B3 = {φ(n1 . n1 . a2 . . x2 .n2 .122) are given by (3. . a2 . 3. hence. Such degeneracies are always a signature of an underlying symmetry. This symmetry has been exploited in deriving the stationary states.n3 ) (a1 .n3 ) (a1 . a3 . a3 . 2. a3 are identical. x2 .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.e. the solutions of (3.122) with (3.. 2.n3 =1 φ(n1 .70 Schr¨dinger Equation o and E1 + E2 + E3 = E. . x1 . i. t0 ) = n1 . n 2 .n3 ) (a1 . (3. x2 ) φn3 (a3 . n3 = 1. r0 ) . n2 . x3 ) = 1.125) is a complete orthonormal basis of F3 and that the propagator for the three-dimensional box is φ(r.96) and. x2 . . a2 .n2 .

1) (a.2. n2 . r) + γ φ(2.1. Hence. a. Any linear combination of wave functions ˜ φ(r) = α φ(1. a.2) (a. to construct n orthogonal stationary states6 . . for example. The origin of this degeneracy is. 1. in case of an n–fold degeneracy it is always possible.2) (a.g. 6 This is a result of linear algebra which the reader may ﬁnd in a respective textbook.127) ˜ ˆ ˜ obeys the stationary Schr¨dinger equation H φ(r) = E122 φ(r). r) + β φ(2. φ(1.2) (a. 3. However.3.6: Particle in Three-Dimensional Box 71 ‘accidental’ degeneracies also occur. due to the hermitian character of ˆ H. a. However. n3 ) = (3. a. the identity 32 + 32 + 32 = 12 + 12 + 52 . One particular aspect of the degeneracies illustrated in the Table above is worth mentioning. r) (3. We consider the degeneracy of the energy E122 which is due to the identity E122 = E212 = E221 . e.. 3) (1.2. for (n1 . a. 5) as shown in the Table above.2. a. a. a. however. r). in case of degenerate states one cannot necessarily expect that stationary states are orthogonal. this linear combination o is not necessarily orthogonal to other degeneraste states.

72 Schr¨dinger Equation o .

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

4) Equation (4.5). the commutator has a simple form.5) ˆ H = 1 1 2 p + m ω 2 x2 ˆ ˆ 2m 2 (4. x] = p ˆ .3) Due to the nature of the harmonic potential which does not allow a particle with ﬁnite energy to move to arbitrary large distances. 4. In order [ˆ. (4.1) of the form ψ(x. The cardinal property exploited below.8) i which holds for any position and momentum operator.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.2) can be expressed in terms of the two operators p = ˆ d .7) which is why these operators are of interest to us. is the commutation property 1 1 (4.3) would be continuous and the eigenfunctions φE (x) would not be normalizable. In the following algebraic solution of (4. however.4) be satisﬁed. t) where ˆ H φE (x) = E φE (x) . beside the representation (4. all stationary states of the harmonic oscillator must be bound states and.235) had been determined. The Hamiltonian H can be expressed in terms of the operators acting on the space (4. t) = exp(−iEt/ )φE (x. In the framework of the Schr¨dinger equation the stationary o states are solutions of (4. the spectrum of ˆ H in (4. therefore. This property states that p and x obey an ˆ ˆ algebra in which the two do not commute.1 Creation and Annihilation Operators The Hamiltonian operator (4. i dx x = x ˆ (4. The stationary states of the harmonic oscillator have been considered already in Chapter 2 where the corresponding wave functions (2.5) →±∞ where C∞ denotes the set of functions which together with all of their derivatives are continuous. We will point out explicitly where assumptions are made which built on this restriction. the natural boundary conditions apply x→±∞ lim φE (x) = 0 . It is important to keep in mind this restriction of the space. in which the operators used below. only for a discrete set of E values can the boundary conditions (4.6) the ﬁrst being a diﬀerential operator and the second a multiplicative operator. If this restriction would not apply and all functions f : R → R would be admitted. (4. ∈ C∞ . act.3) can be solved for any E ∈ R.7) of the Hamiltonian. lim ( ) = } (4. however. The operators act ˆ on the space of functions N1 deﬁned in (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.

8) a ˆ ˆ ˆ ˆ ˆ implies [ˆ− .15) together lead to the commutation property (4.18) It is of interest to note that the operators a+ and a− are real diﬀerential operators. 1 ω 2 (4.8) we recall that all operators act on functions in N1 and. the action of [ˆ. ˆ ˆ .1: Creation and Annihilation Operators 75 to derive (4. a+ ] = a− a+ − a+ a− does not necessarily vanish.8).14) and (4.10) we deﬁne a+ = ˆ mω i x − √ ˆ p.11) a− a+ = ˆ ˆ (4. 1 i dx i dx i i (4. hence. a+ ] = 1 . 2mω dx (4. x] .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. 2mω dx a− = ˆ mω x + ˆ 2 d . x] p ˆ on such functions must be considered.7) and (4.12). p ˆ 2 2m ω 2 (4.14.9) From this follows (4.7) assuming that the factors are easier to handle than the Hamiltonian in yielding spectrum and eigenstates.11) ˆ The reader may note that we have attempted.12) To prove this commutation property we determine using (4. 1 2 (4. a ˆ 1 (4. the commutator property (4.4. One obtains [ˆ.15) 1 ˆ 1 H + 1 . Being guided by the identity for scalar numbers (b − ic)(b + ic) = b2 − i(cb − bc) + c2 = b2 + c2 (4.16) (4. 4. 1 ω 2 (4.17) 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 . In fact.13) (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 . in fact. Before we continue we like to write (4. Our next step is an attempt to factorize the Hamiltonian (4. ˆ 2 2m ω (4.10) holds for a ˆ ˆ [ˆ− . ˆ 2 2m ω a− = ˆ mω i x + √ ˆ p. to factor H/ ω. x] f (x) = p ˆ d d x f (x) − x f (x) = f (x) = 1 f (x) .8) yield a− a+ = ˆ ˆ Similarly one can show a+ a− = ˆ ˆ 1 ˆ 1 H − 1 .

The derivation will be based solely on the properties (4. 4.24) − m This property states that the operators in the function space N1 are the hermitian conjugate of each other. 4.21) is available.19) in the form1 f |ˆ+ g a Ω∞ = g|ˆ− f a Ω∞ .20) ˆ In the following we will determine the spectrum of H and its eigenstates. (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. The property ˆ ˆ follows directly from (??). ˆ (4.17. (4.22. 4.16. . 4.19).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) . 4.23) Together.16. In fact. 4.17. 4. 4. (4. Using (??) we like to state (4. it holds that for a stationary state φE (x) of energy E deﬁned through (4.22) Similarly one can show using again (4.12.16. ˆ (4. ˆ The results (4. This property of operators is investigated more systematically in Section 5. one obtains using (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) . 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) (4.19) This property states that the operators a+ and a− are the adjoints of each other. g ∈ N1 +∞ +∞ dx f (x) a+ g(x) = −∞ −∞ dx g(x) a− f (x) .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) .17.21) a− φE (x) is ˆ a stationary state of energy E − ω and a+ φE (x) is a stationary state of energy E + ω.

2 2 (4. For f (x) = φE (x) and g(x) = φE (x) in (??) follows (Note that according to (??) the eigenvalue E is real.25) φ0 (y) = 0 (4.26) Of course.32) for some constant c0 or (4.30) Assuming that φ0 (y) does not vanish anywhere in its domain ] − ∞.4. (4.25) on one side would lead to termination of the sequence E0 + m. m ∈ Z when m is decreased.25) can.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. Another name is creation (ˆ+ ) and annihilation (ˆ− ) operators since these operators a a create and annihilate vibrational quanta ω.29) where y = mω x.2 Ground State of the Harmonic Oscillator d + y dy A suitable solution of (4. φ0 (y) dy dy the solution of which is 1 lnφ0 (y) = − y 2 + c0 2 φ0 (y) = c0 exp 1 − y2 2 . Using (4. however. There arise. (4.e. +∞[ one can write (4. E = E .17) ˆ H φ0 (x) = ( ω a+ a− + ˆ ˆ 1 ω 1 ) φ0 (x) = 1 ω φ0 (x) . in fact. a state φ0 (x) which obeys the property a− φ0 (x) = 0 ˆ (4.) ˆ ˆ 0 = φE |HφE Ω∞ − φE |HφE Ω∞ = ( E − E ) φE |φE Ω∞ .6. The property (??) has an important consequence for the stationary states φE (x). (ii) the construction appears to yield energy eigenvalues without lower bounds when. In fact.27) Since E = E one can conclude φE |φE Ω∞ = 0. in fact. the solution φ0 (x) of (4. on ˆ the other side such a state is itself an eigenstate of H as can be shown using (4.28) 4. (4.11) one can rewrite (4..29) d 1 d φ0 (y) = = lnφ0 (y) = − y . It turns out that both diﬃculties can be resolved together. E . 4. one expects that E = 0 should be a lower bound.25) needs to be normalizable in order to represent a bound state of the harmonic oscillator.33) . For obvious reasons one refers to a+ and a− as ladder ˆ ˆ operators. i. be found. φ0 (x) should be an element of the function space N1 deﬁned in (4. two diﬃculties: (i) one needs to know at least one stationary state to start the construction.31) (4. Let φE (x) and φE (x) be two normalized stationary states corresponding to two diﬀerent energies E.5). (4.

Such states need to be suitably normalized for which purpose we introduce a normalization constant cn φn (x) = cn a+ ˆ n φ0 (x) . . choose the normalization constants cn and the functions φn (x) real as well. m ∈ Z which properly terminate at some minimum value E + m0 ω ≥ 0. . 2 n = 0.3 Excited States of the Harmonic Oscillator Having obtained a suitable stationary state with lowest energy. 2. (4.39) We notice that the ground state wave function φ0 (x) as well as the operators (ˆ+ )n are real.2) there are two competing contributions to the energy.36) assumes a functional form such that both terms together assume a minimum value. i. The conventional normalization condition φ0 |φ0 reads +∞ +∞ Ω∞ = 1 (4. a state conﬁned to the minimum corresponding to the classical state of lowest energy.39). n = 0. 2. 1. The ground state (4. The set of allowed energies of the oscillator according to (4.24) can then be written as follows En = ω(n + 1 ). it holds ˆ H φn (x) = ω(n + 1 ) φn (x) . . we construct the states for n = 1. the potential energy contribution which for a state δ(x). 2 n = 0.. (4. . .35) The appropriate ground state is φ0 (x) = mω π 1 4 exp − mωx2 2 .78 Linear Harmonic Oscillator This solution is obviously normalizable.36) n times. . 2. We a can. (4. .34) dx |φ0 (x)|2 = |c0 |2 −∞ −∞ dx exp − mωx2 2 = |c0 |2 π .e. We recall that according to (4. .25) admits only one solution there is only one set of states with energy E + m ω. The reason is that in the Hamiltonian (4. (4.36) Since the ﬁrst order diﬀerential equation (4. .37) It is most remarkable that the energy 1 ω of the ground state is larger than the lowest classically 2 allowed energy E = 0.26) the energy value associated with this state is 1 ω. 1. . we can now construct the stationary states corresponding to energies (4. therefore. 4. For this purpose we apply the operator a+ to the ˆ ground state (4. i. 2 This state of lowest energy is called the ground state of the oscillator. We will consider this point more systematically in Section 5.37) above the ground state energy. ... . would yield a vanishing contribution. and the kinetic energy contribution. .e. 2. 1. which for a narrowly localized state yields a large positive value..38) These states correspond to the energy eigenvalues (4. mω (4. i. the so-called excited states.e.

15 .18)] and since the ˆ φn (x) are real functions. 1. 4.40) = 1.48) (4. n = 0. 2. . . 2. (4. n = 0.45) form an orthonormal set.45) We like to note that these functions according to (4.47) for some suitable constants βn . Ω∞ 2 = βn a− φn |ˆ− φn ˆ a Ω∞ 2 = βn φn |ˆ+ a− φn a ˆ Ω∞ . . .43) (4.3: Excited States 79 We need to determine now the normalization constants cn .24) holds in analogy to (4.14) 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. (4.40) φn−1 (x) = βn a− φn (x) ˆ (4. To determine these constants we adopt a recursion scheme. Since a− is a real diﬀerential operator [see (4. i.20) 1 = φn−1 |φn−1 Equations (4.42) ˆ Using (4. √ a+ φn (x) = n + 1 φn+1 (x) . . .e. √ a− φn (x) = n φn−1 (x) . For n = 0 holds c0 = 1.44) 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) . 1. (4.39) yield √ From this follows βn = 1/ n and.40). 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 . βn must be real as well.49) (4. they obey φn |φn = δn n .50) φn (x) . We consider then the situation that we have obtained a properly normalized state φn (x).20) yields 2 αn φn |ˆ− a+ φn a ˆ Ω∞ = 1. To determine βn we note using (4. (4. n. according to (4. (4.28) and the construction (4.46) According to (4.40–4. ˆ (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. according to (4. (4..4.51) . .41) Employing the adjoint property (4.47).

It should be noted that √ the normalization (4.53) We will later re-introduce this factor to account for the proper normalization (4. (4.57) 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) . . simplifying the calculation.. 2. namely the square root of the Jacobian dx/dy.58) This result agrees with the expression (2.7 and given. however. by introducing the variable y deﬁned in (4.56) can be expressed in terms of Hermite polynomials Hn (y) introduced in Sect. For this purpose we start from expression (4.52).57) which includes the factor 1/ n! according to Eqs.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 (4.7.52) This normalization of the wave functions diﬀers from that postulated in (4. by dx = dy mω 1 4 .233) derived in Sect.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 .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 . i. n–independent factor. 2. for example.28) of the ground state and the deﬁnition (4.46) yields a set of normalized states.39).55) 1 y2 2 (4. We will demonstrate below the identity Hn (y) = e y2 2 y − d dy n e− y2 2 (4.200).56) Relationship to Hermite Polynomials We want to demonstrate now that the expression (4.45). (4. (4.40–4. (4. by the Rodrigues formula (2. (4.30) and employing the normalization +∞ −∞ dy φ2 (y) = 1 n (4.35) rather than (4.e.

(3. dy n (4.4. therefore. 1..2) for an arbitrary initial wave function ψ(x0 .65) .e.200). 4.30) and return later to the coordinate x. g(y) = (−1)n+1 e y2 2 dn+1 −y2 e = dy n+1 y − d dy n+1 e− y2 2 .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 (4.4: Propagator 81 To verify the realtionship between the deﬁnition (4.1.200) 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 . i. (4. t0 ). (4.59) by induction noting ﬁrst that (4.59) holds for n = 0. 4. t0 ) = n=0 dn φn (y0 ) .114)]. We prove (4.59) dn −y2 e dy n (4. For the sake of notational simplicity we employ initially the coordinate y as deﬁned in (4.61) and.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 .39) ∞ ψ(y0 . hence.61) One can factor g(y) and employ (4.4 Propagator for the Harmonic Oscillator We consider now the solution of the time-dependent Schr¨dinger equation of the harmonic oscillator o (4.64) which implies that (4.63) one obtains g(y) = y − d dy f (y) (4.57) of the hermite polynomials and the deﬁnition given by (2. 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. dy dy (4. (4. the Rodrigues formula (2.106–3.60) and.59) hold. Our derivation will follow closely the procedure adopted for the case of a ‘particle in a box’ [see Eqs.

1 t = e−iω(t1 −t0 ) (4. a supposition which is not proven here2 .108–3. N. . t0 ) which is an element of N1 deﬁned in (4.. t0 ) ψ(y0 .69) where ∞ φ(y. Employing orthogonality condition (4. t) = π −3/2 exp 2 2 y 2 y0 + 2 2 2 +∞ +∞ ∞ du −∞ −∞ dv n=0 1 ( −2tuv )n × n! (4.h.46) the expansion coeﬃcients dn are +∞ dn = −∞ dy0 ψ(y0 . t) (4. 2 A demonstration of this property can be found in Special Functions and their Applications by N. t0 ) φn (y0 ) .72) × exp( −u − v + 2iyu + 2iy0 v ) . For this purpose we start from the integral representation (2.147) we employ the technique of generating functions as in Sect. Englewood Cliﬀs.70). 1965) Sect. pp. 68. We want to demonstrate now that this propagator is identical to the propagator (2. t) = n=0 Hn (y)e−y /2 Hn (y0 )e−y0 /2 n √ t .N.s. y0 . 2. t0 ) = n=0 φn (y) φn (y0 ) tn+ 2 .Lebedev (Prentice Hall.112) and (4. y0 . Inc.71) Applying (??) to express Hn (y) and Hn (y0 ) yields exp(−2tuv) w(y. t1 ) = −∞ dy0 φ(y. t|y0 . t1 ) = n=0 dn φn (y) cn (t) (4.39) holds cn (t1 ) = exp −i ω ( n + 1 ) ( t1 − t0 ) 2 .147) for the harmonic iscillator determined in Sect. In order to prove the equivalence of (4. this is an excellent textbook from which we have borrowed heavily in this Section.J. sec:harm.225) which allows one to derive a generating function for products of Hermite polynomials which can be applied to the r. We consider for this purpose the following expression for |t| < 1 ∞ w(y. 4.70) is the propagator of the linear harmonic oscillator.68) Altogether one can express then the solution +∞ ψ(y.. (4.70) and (2.15. (4.67) for which according to (3.82 Linear Harmonic Oscillator Such expansion is possible for any f (y0 ) = ψ(y0 . of (4. t1 |y0 .65) to times t ≥ t0 through insertion of time-dependent coeﬃcients cn (t1 ) ∞ ψ(y.7.5).66) One can extend expansion (4. 2n n! π 2 2 (4.

y0 .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 . t|x0 . t0 ) = i φ(x.80) −2iω(t1 −t0 ) sinω(t1 − t0 ) 1−e We can ﬁnally express the propagator (4. it holds φ(y.74) a second time yields ﬁnally together with the deﬁnition (4. a Re a2 > 0 (4. (4. t) = π −1 exp − 2 y 2 y0 + 2 2 +∞ dv exp(−(1 − t2 ) v 2 − 2i (yt − y0 ) v) .e. t0 ) (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 = . (4. t0 ) = and 1 2 i π F (t1 .78) in terms of the coordinates x and x0 .. y0 .s.4: Propagator Carrying out the sum on the r.e. t) = π −3/2 exp +∞ 83 y2 y2 + 0 2 2 +∞ dv exp( −v 2 + 2iy0 v ) × −∞ × −∞ du exp( −u2 − 2u(tv − iy) ) .77) The sum in (4. (4.. t0 ) = exp mω 2iπ sinω(t1 − t0 ) 1 2 × . . (4.4. √ πexp( t2 v 2 − 2iytv − y 2 ) (4. t0 ) = where F (t1 .h. by a factor 4 mω/ for both φn (y) and φn (y0 ).71) of w(y.76) One can express this in terms of the stationary states (4. i. −∞ (4.81) imω 2 sinω(t1 − t0 ) (x2 + x2 ) cosω(t1 − t0 ) − 2 x x0 0 This result agrees with the propagator (2. t1 |y0 . identical to the generating function (4. indeed.74) applied once results in w(y.147) derived by means of the path integral description.73) The incomplete Gaussian integral +∞ −∞ dx e−a 2 x2 √ − 2bx = π b2 /a2 e . This requires that we employ (4.75) Applying (4.70) is.79) 1 + e−2iω(t1 −t0 ) cosω(t1 − t0 ) = . The resulting propagator is G(t1 .78) 1 − e−2iω(t1 −t0 ) = sinω(t1 − t0 ) 2 i e−iω(t1 −t0 ) (4. t0 ) − i y y0 2 F (t1 . i. y0 .77). t0 ) exp i 2 1 2 (y + y0 ) G(t1 . one obtains w(y.

Here f (ν) (y0 ) denotes the ν-th derivative of f (y) taken at y = y0 .86) f (ˆ+ ) = a a ˆ ν! ν=0 . f (ˆ+ ). 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. 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.. phonons. the use of generating functions. 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 − . each corresponding to a single harmonic oscillator of the type studied presently by us.85) is convergent everywhere in R. (4. 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! . 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.82) a+ = √ ˆ dy dy 2 2 √ √ Obviously. We want to introduce now techniques based on the ladder operators a+ and ˆ a− . ˆ ˆ Calculus of Creation and Annihilation Operators We summarize ﬁrst the key properties of the operators a+ and a− ˆ ˆ [ˆ− .83) We will encounter below functions of a+ and a− . a− = √ ˆ y + . and of the modes of the quantized electromagnetic ﬁeld. a (4. a− must be equivalent.5 Working with Ladder Operators In the last section we have demonstrated the use of diﬀerential equation techniques. both quantum systems are endowed with a large number of modes. one can express y = 2(ˆ− + a+ ) and d/dy = 2(ˆ− − a+ ) and. In this case we deﬁne ∞ f (ν) (0) + ν (4. 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.. d/dy and a+ . hence.84) = 1 1 = 0 √ = n + 1 φn+1 (y) √ = n φn−1 (y) = f |ˆ+ g a Ω∞ .g.e. e.84 Linear Harmonic Oscillator 4. (4. the approaches a ˆ a ˆ using y. i.

90) holds for any function fn (ˆ+ ) = (ˆ+ )n which is a power of a a a+ .93) one can reduce all operators acting on some proper state function by operators acting on the state φ0 (y).90) we show that (4.87) (4.91) (4. ˆ a We proceed by induction noticing ﬁrst that (4. We note ˆ a− f (ˆ+ ) = [ˆ− .90) To prove (4. 4. property (4. a+ ] + [ˆ− . The ﬁrst case is trivial. Hence. The convergence of the Taylor expansion ascertains then that (4. a a a 85 (4..87. Note that ˆ a an immediate consequence of (4. f (ˆ+ )] = f (1) (ˆ+ ) . 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 second case follows from [ˆ− . a An important operator function is the exponential function. a+ ] = 1 = f1 (ˆ+ ) .88) is a fundamental one and will be used in the following. a a (4.88) (4. For fn+1 follows then [ˆ− . a a − (4. (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) (4.92) f (y) = n=0 dn φn (y) = n=0 (ˆ+ )n a dn √ φ0 (y) n! (4. i.95) (4. a a a ˆ 1 a Let us assume that (4.89) (4.83) and can be proven for all powers fn by induction and then inferred for all proper functions f (ˆ± ). f (ˆ+ )] + f (ˆ+ ) a− ˆ a a a a ˆ which implies that in order to prove (4. (ˆ+ )n a+ ] = (ˆ+ )n [ˆ− . ˆ a a a ˆ In particular.4. f1 (ˆ+ )] = [ˆ− . The following important property holds a a− f (ˆ+ ) = f (1) (ˆ+ ) + f (ˆ+ ) a− . An example is the so-called shift operator exp(uˆ− ).88) we need to show actually [ˆ− .88) is (ˆ− )n f (ˆ+ ) φo (y) = f (n) (ˆ+ ) φ0 (y) .94) This identity follows from (4. It holds a a euˆ f (ˆ+ ) φ0 (y) = f (ˆ+ + u) φ0 (y) . we will only assume operator functions acting on φ0 (y).96) . a− f (ˆ+ ) φ0 (y) = f (1) (ˆ+ ) φ0 (y) ˆ a a which follows from a− φo (y) = 0. 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+ ).5: Working with Creation and Annihilation Operators and similarly for f (ˆ− ).90) holds for f0 and for f1 .90) holds for fn .90) holds for f (ˆ+ ).e. fn+1 (ˆ+ )] a a = [ˆ− .

108) .90) and zz ∗ = 1 we can write this d du e uzˆ+ a e uz ∗ a− ˆ a − z ∗ a− . + (4.96) is applied is a a a euˆ evˆ φ0 (y) = ev (ˆ − + + + u) a φ0 (y) = euv evˆ 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) (4.103) (4.105) 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) . + ∗ a− ˆ a − u euzˆ euz + ∗ a− ˆ = (4. one obtains a evˆ + − v ∗ a− ˆ a φ0 (y) = e− 2 vv evˆ e− v 1 ∗ + ∗ a− ˆ φ0 (y) (4. We assume in the following a ˆ derivation. z ∈ C.106) Applying (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. Below we will use the operator identity which follows for the choice of v = α in (4.97) An example in which (4. + ∗ + (4. without explicitly stating this. + (4. 1 + + uz ∗ a− ).99) ˆ C(u) = zˆ a zˆ+ + z ∗ a− a ˆ + +z a ˆ ∗ − a euzˆ euz ˆ − u C(u) .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) .105) a eαˆ + − α∗ a− ˆ φ0 (y) = e αα∗ 2 e−α ∗ a− ˆ a eαˆ φ0 (y) .107) − α∗ a− ˆ φ0 (y) = e− αα∗ 2 a eαˆ φ0 (y) . ∗ = 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 (4.98) The related operators exp[vˆ+ ± v ∗ a− ] play also an important role.104) (4. a a ν! (4. euzˆ ˆ euz ∗ a− ˆ (4. (4. that the operators considered act on φ0 (y).101) ˆ ˆ ˆ C(u) obviously obeys C(0) = 1 This results in D(0) = 1 and. 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) . the solution of (4.106) yields a eαˆ + ∗ a− ˆ a a eαˆ φ0 (y) = e−αα eαˆ φ0 (y) .102) Similarly.101) 1.98) for u = −α∗ and v = α yields e−α This together with (4. hence.

116) = δµ ν .h.112) Expanding both sides of this equation and using (4. Setting y = 0 in (4. Using (4. 4. and on the r.µ=0 uµ v ν √ φµ |φν µ!ν! ∞ = µ=0 uµ v µ µ! (4. the expected from which follows by comparision of each term on both sides φµ |φν orthonormality property.95). provide the same values for Hn (0) as provided in Eq. and ˆ l.h. (??). i.110) √ y2 u2 + 2uy − 2 2 a = euˆ φ0 (y) . of (4.115) and using (4.98) and again (4.h.111) This expression..58) the Hermite polynomials Hn (y) by the eigenstates φn (y) √ ∞ 1 ( 2t)n y2 /2 2yt − t2 √ e φn (y) (4.h.s. .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! (4.109) e = π4 n! n=0 √ Using (4.84) 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. Expanding r. in the ladder operator calculus.111) the orthonormality properties of the wave functions φn (y). we can reproduce by means of the generating function (4. (4.95) one obtains a a a a euˆ φ0 |evˆ φ0 = φ0 |euv evˆ φ0 = euv evˆ φ0 |φ0 = euv − + + + − (4. + (4.4.114) where we have employed (4. We want to derive now the values Hn (0).e.113) Comparision of all terms on the l.84.115) a where the latter step follows after expansion of evˆ and using a− φ0 (y) = 0. For this purpose we consider a a a a euˆ φ0 |evˆ φ0 = φ0 |euˆ evˆ φ0 + + − + (4.s. is the equivalent of the generating function (??).111) yields π − 4 exp − 1 u2 2 a = euˆ φ0 (0) .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. We start from the generating function (??) and replace according to (4. Similarly. 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.s.84) yields ∞ ν.

is given by a function ψ(x). dp (4.125) . In the momentum representation the wave function is a function of p. 1. (4. ˆ and the Hamiltonian (4.7) is given by (4. the momentum and position operators are as stated in (4.7) is 1 2 m 2 ω 2 d2 p − . (−i)n ˜ φn (p) = √ 2n n! m˜ ω π 1 4 exp − m˜ p2 ω 2 Hn ( m˜ ω 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. i.120) Multiplying this equation by 1/m2 ω 2 yields − where ˜ E = E / m2 ω 2 . .2). the time-independent Schr¨dinger equation for the oscillator can be written o ˆ H = − mω 2 2 2 x = i ˆ d . ...121).122).2. As to be expected.123) (4.e. 2.118) d2 1 2 + p dp2 2m ˜ ˜ φE (p) = E φE (p) . (4. . the probability density is P (x) = |φn (x)|2 . n = 0. are ˜ En = or. For example.88 Linear Harmonic Oscillator 4. . ˜ (4. i. In the position representation the wave function is a function of x. is given by the ˜ function ψ(p). . 1.235). The momentum representation eigenfunctions are.124) ). the momentum and position operators are p = p. (4.122) For a solution of the Schr¨dinger equation in the momentum representation.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.e. The solutions can be stated readily exploiting the earlier results. of (4. using (2. (4.235). . n = 0. The eigenvalues.37). according to (4. for an oscillator in a stationary state φn (x) as given by (2. 4. we o note that the posed eigenvalue problem is formally identical to the Schr¨dinger equation in the o position representation as given by (4.3).. ω = 1 / m2 ω .6) and the Hamiltonian (4. i. For this purpose we employ the Schr¨dinger equation in the momentum o representation. En = ω (n + 1 ) 2 d2 1 + m˜ 2 p2 ω 2 2m dp 2 2 ˜ ˜˜ φE (p) = E φE (p) (4. using (4. (4.119) 2m 2 dp2 Accordingly. 2. the eigenvalues are identical to those determined in the position representation.121) ω (n + ˜ 1 ) 2 .e.

through direct integration.122) (−i)n ˜ φn (p) = √ 2n n! 1 πm ω 1 4 89 exp − p2 2m ω Hn ( 1 p) . the correctness of the phase factors (−i)n .130) dx exp(−ipx/ ) φn (x) −∞ (4.6. Deﬁning k = m˜ / p or ω 1 k = p (4. of course.133) where we note that a change of the normalization of φn (y) [c.. deﬁned only up to an arbitrary phase factor eiα . 4. From (4. .30) and (2.127) exp − p2 m ω 2 Hn ( 1 p) . Proceed as follows.f.129) m ω one obtains. α ∈ R. ˜ We can now state the probability density Pn (p) for an oscillator with energy En to assume momentum p.126) in terms of dimensionless units following the procedure adopted for the position representation where we employed the substitution (4. This implies the property of Hermite polynomials (−i)n e− k2 2 +∞ Hn (k) = −∞ dy exp(−iky) e− y2 2 Hn (y) (4. φn (k) = We want to ﬁnally apply the relationship 1 ˜ φn (p) = √ 2π +∞ 2 (−i)n −k √ e 2 Hn (k) .126) Normalized eigenfunctions are. m ω (4.235)] absorbs the replacement dx → dy.4. the Jacobian.30) and (4. i. instead of (4. 2n n! π (4. this allowed us to introduce in (4.126). m ω (4.131) in the present case employing dimensionless units..128) We may also express the eigenstates (4.126) a phase factor (−i)n .e.125.6: Momentum Representation or with (4.e.1: Demonstrate the validity of (4. (4.129) follows ky = px/ and.134) Exercise 4. i. hence.132) dy exp(−iky) φn (y) −∞ (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 (4.134).30) to obtain (4. 1 ˜ φn (k) = √ 2π +∞ (4.58).

except that α changes periodically in time. carries out a periodic motion described by y(t) = Re y0 eiωt (4.138).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. 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. We want to construct and characterize such states. For this purpose we assume such state exists and expand ∞ ∞ φ(α) (y) = n=0 d(α) φn (y) . coherent states. 4.136) (b) Employ the generating function (2.134). (4. pp. z) = exp −2ikz + z 2 − k 2 /2 . 4.196) of Hermite polynomials and equate coeﬃcients of equal powers of z to prove (4.Dupont-Roc.139) where we have added the condition that the states be normalized. for example. The quasi-classical states can be obtained by generalizing the construction of the ground state of the oscillator. The answer is ‘yes’. namely through the eigenvalue problem for normalized states a− φ(α) (y) = α φ(α) (y) .90 (a) Prove ﬁrst the property 1 f (k.C. φ(α) | φ(α) = . Such states are referred to as quasi-classical states. +∞ Linear Harmonic Oscillator dy exp(−iky) g(z. The question arises if similar states exist also for the quantum oscillator.7 Quasi-Classical States of the Harmonic Oscillator A classical harmonic oscillator. Inc. or Glauber states of the harmonic oscillator and they play. The motion of the probability distribution of such state is presented in Fig. 192 3 . Inserting this expansion into √ (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.1 showing in its top row the attributes of such state just discussed.. J. and G. y) −∞ (4. 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 . 2. We ﬁrst construct the solution of (4.4.Grynberg (John Wiley & Sons.Chen-Tannoudji.194. z) = exp 2yz − z 2 − y 2 /2 (4.135) g(k. ˆ α ∈ R. for example.138) We will show that a harmonic oscillator prepared in such a state will remain forever in such a state. 1989) which discusses quasi-classical states of the free electromagnetic ﬁeld in Sect.— III. a pendulum swinging at small amplitudes. New York. n n=0 d(α) n 2 = 1 (4.140) An excellent textbook is Photons and Atoms: Introduction to Quantum Electrodynamics by C. z) = √ 2π where f (k.

n! n (4. n n+1 (4. i.109).147) allows one to interprete T (α) as an operator √ which shifts the ground state by 2α. hence.147) √ which identiﬁes this state as a ground state of the oscillator displaced by 2α. (4.145) Using the generating function in the form (4.144) Normalization requires c = exp(−α2 /2) and.146) √ 1 − (y − 2 α)2 2 (4. the quasi-classical states are φ (α) (y) = exp α2 − 2 ∞ n=0 αn √ φn (y) . However.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.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) . [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) .150) for α presently real. n! (4. Comparing (4.4.142) The solution is αn d(α) = √ c n n! ∞ ∞ (4.7: Quasi-Classical States or ∞ n=0 91 √ n + 1 dn+1 − α d(α) n (α) φn (y) = 0 .149) − α∗ a− ˆ (4. n! (4.138).e.148) The identity (4.145) using (4.143) for a constant c which is determined through the normalization condition [see (4.149) with (4.84) φ(α) (y) = exp − α2 2 ∞ n=0 αα∗ (αˆ+ ) a a+ φ0 (y) = e− 2 eαˆ φ0 (y) .. n! (4. This interpretation justiﬁes the choice of α ∈ R in (4. One can also write (4.139)] 1 = n=0 d(α) n 2 = |c|2 n=0 α2 2 = |c|2 eα .

67. i.154) .145) is replaced by a complex number αu which at times t1 = t0 + 2πn/ω.152) T + (α) f |T + (α) g = f |T (α) T + (α) g = f |g . n = 0. (4.158) Comparision of this expression with (4. In particular.s. t1 ) = exp or φ (α) − α2 2 ∞ exp n=0 −i ω ( n + ∞ n=0 1 ) ( t1 − t0 ) 2 αn √ φn (y) n! (4.s.109) serves again to simplify this sum √ y2 1 1 α2 (u2 + 1) φ(α) (y.160) .h.150) for complex α obeys the unitary property T + (α) T (α) = 1 1 (4. it holds T + (α) φ0 |T + (α) φ0 = f |T (α) T + (α) g = φ0 |φ0 .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 (4.150) 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 (4.145) as the initial state φ(α) (y. is the adjoint of the operator function on the l. 4.157) (y.155) (4. (4. the shift operator leaves the ground state normalized. one can write in case of T + (α) T + (α) + = T (α) = eα ∗ a− ˆ − αˆ+ a a = e − (αˆ + − α∗ a− ) ˆ .145) shows that α in (4. . n! u = e−iω(t1 − t0 ) . t1 ) = π − 4 u 2 exp − + 2yαu − 2 2 E (4.152) Since the operator function on the r. becomes real. 1. Following the procedure adopted in determining the propagator of the harmonic oscillator [see (4.h. Relationship (4.92 Linear Harmonic Oscillator The shift operator as deﬁned in (4. . α ∈ C (4.e.156) (4..95) to linear combinations αˆ+ + βˆ− and reads then a a f (αˆ+ + βˆ− ) φ|ψ = φ|f (α∗ a− + β ∗ a+ ) ψ . a a ˆ ˆ (4.68)] one can write the time-dependent solution with (4. t1 ) = u exp 1 2 α2 − 2 (α u)n √ φn (y) .153) This is obviously the inverse of T + (α) as deﬁned in (4. .151) which follows from a generalization of (4.

Our interpretation 2 explains also the meaning of a complex α in (4.150) with (4.138. except that it is not pointlike.158) allows one to write 1 φ(α) (y. (4. t1 ) = u 2 T + (α u) φ0 (y) (4. (4. t1 ) = π − 4 exp − [y − y0 (t1 )]2 + iy k0 (t1 ) − iφ0 (t1 ) − ω(t1 − t0 ) 2 2 .162) 2 with a periodic change. 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 (4.159) is then 1 1 i φ(α) (y. 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.4.161) 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. this characterization corresponds closely to that of the possible initial states of a classical oscillator. a momentum factor and a phase factor. moving solely its center of mass in an oscillator fashion around the minimum of the harmonic potential.147). being displaced around the center with period 2π/ω and. a complex α corresponds to a displaced oscillator which has initially a non-vanishing velocity.1 (top row) we present the probability distribution of the Glauber state for various instances in time.138): a real α corresponds to a displacement such that initially the oscillator is at rest. Comparing (4.161) and is found to describe a displaced Gaussian.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 (4.163) This solution corresponds to the initial state given by (4. 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. in general. This shows clearly that the Glauber state is a close analogue to the classical oscillator.164) where according to (4. The diagram illustrates that the wave function retains its Gaussian shape with constant width for all times. One can express (4. experiences a change of its width (relative to .165) . The time-dependent wave function (4. In Fig.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. 4.148. 4.163) also through the shift operator. obviously.

166) φ(ao .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. This behaviour is illustrated in Fig.169) . To describe such state which satisﬁes the initial condition φ(ao . σ|y. t0 ) = [πσ 2 ]− 4 exp we expand ∞ 1 − (y − ao )2 2σ 2 (4. 4.1: Time dependence of Gaussian wave packets in harmonic oscillator potential. 2σ 2 q = a . 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. σ|y. 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). σ|y. t0 ) = n=0 dn φn (y) .168) dy e−p(y−q) Hn (y) where p = 1 + σ2 . (4.167) One can derive for the expansion coeﬃcients +∞ dn = −∞ dy φn (y) φ(ao . 1 + σ2 (4. that of the Glauber states) with a period π/ω.

a worthy successor of the famous Gradshteyn.143) of the Glauber state. σ→1 p−1 2 (4.157) as follows φ(ao . therefore. 1990).171) lim p−1 p n 2 Hn (q p ) = lim (2q)n = an . namely $750 4 . 1 (4. This expansion can be written φ(ao .76) permits us to write this φ(ao .170) In order to check the resulting coeﬃcients we consider the limit σ → 1. Brychkov. vol. 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. 2 by A.173) which is.4.e.I. i.172) One can then conclude e−a /4 n lim dn = √ a σ→1 2n n! (4. σ|y.168) is4 e−aq/2 (p − 1)n/2 dn = √ Hn (q 2n n!σ p(n+1)/2 p ).P. this 3 volume integral table is likely the most complete today.175) ∞ tn n=0 2n n! Hn (yo ) e−yo /2 Hn (y) e−y 2 2 /2 The generating function (4. p−1 1 − σ4 t = 1 − σ 2 −iω(t1 −to ) e 1 + σ2 (4. σ|y. t1 ) = √ × where yo = q p a = √ . p−1 95 (4.. New York.7: Quasi-Classical States The solution of integral (4. and O.174) Hn (q p p−1 ) φn (y) . Marichev (Wiley. Prudnikov. In this limit the coeﬃcients dn should√ become identical to the coeﬃcients (4. t1 ) = e−aq/2 √ pσ ∞ n=0 e −iω(n+ 1 )(t1 −to ) √ 2 1 2n n! p−1 p n/2 (4.177) − 1 ω(t1 − to ) 2 . a ground state shifted by ao = 2α. σ→1 O(y n−2 ) for n ≥ 2 1 for n = 0.176) 1 √ pσ π a 2 exp − 1 1+σ2 + 1 yo − 1 ω(t1 − to ) 2 2 2 2 × (4. in fact. t1 ) = √ 2 × exp − 1 (y 2 + yo ) 1+t2 − 2 1−t 2 1 1 √ √ pσ π 1 − t2 × t 1 − t2 (4. σ|y. unfortunately very expensive.143). in agreement with the expected result (4. Yu. Following the strategy adopted previously one can express the time-dependent state corresponding to the initial condition (4. 1 a2 2 1 + σ2 + 1 2 2 y o + 2 y yo This integral can be found in Integrals and Series.166) in analogy to (4.

96 Linear Harmonic Oscillator .

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

ϑ parametrizes the rotations uniquely as long as |ϑ| < 2π. it follows 3 [R(ϑ)]jk [R(ϑ)]j = δk j=1 . namely around an axis given by the direction of ϑ and by an angle |ϑ|.6) follows immediatety for the determinant of R(ϑ) using detA B = detA detB and detAT = detA detR(ϑ) = ±1 .2) Rotations conserve the scalar product between any pair of vectors a and b. x2 . in Cartesian coordinates r = (x1 . they conserve a · b = 3 aj bj . i.7) implies that a rotation is associated with an inversion. It holds then j=1 3 3 a ·b = j.8) 0 0 1 In case of a prefactor +1. (5. In the following we want to exclude inversions and consider only rotation matrices with detR = 1. (5.k.e.6) (5. a minus sign in Eq.98 Properties of Rotations in R Theory of Angular Momentum and Spin Rotational transformations of vectors r ∈ R . A rotation around the x3 -axis by an angle ϕ is described by the matrix cosϕ − sinϕ 0 R(ϑ = (0. (5. if the thumb in your right hand ﬁst points in the direction of ϑ.7) Let us consider brieﬂy an example to illustrate rotational transformations and to interpret the sign of detR(ϑ). in case of a prefactor −1 a rotation and an inversion at the origin. 0. therefore. (5. From (5. can be represented by 3 × 3 matrices R(ϑ). x3 )T . i.1) In Cartesian coordinates this reads 3 xk = j=1 [R(ϑ)]kj xj k = 1.5) This equation states the key characteristic of rotation matrices. are linear and...e.e. 0. (5. =1 [R(ϑ)]jk [R(ϑ)]j ak b = j=1 aj bj .(5. In the latter case the determinant of R(ϑ = (0. 2. then the ﬁngers of your ﬁst point in the direction of the rotation. . ϕ)T ) = ± sinϕ cosϕ 0 (5. i. 3 .3) Since this holds for any a and b. ϕ)T ) is negative. where ϑ denotes the rotation. the matrix represents a proper rotation. (5.4) 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 . We assume the convention that rotations are right-handed. Let us deﬁne the rotated vector as r = R(ϑ) r .

We have shown altogether that the elements of SO(3) form a group.9) This set of matrices is closed under matrix multiplication and. (5. In the following we will assume R1 . (5. in fact. T R1 R1 −1 = 1 −1 = 1 1 1 (5. R RT = RT R = 1 detR = 1 } 1. b ∈ G the product c = a ◦ b is also in G. there exists a real 3 × 3–matrix −1 R1 which is the inverse of R1 .13) . (ii) There exists an element e in G with the property ∀a ∈ G → e ◦ a = a ◦ e = a. hence.10) detR3 = det ( R1 R2 ) = detR1 detR2 = 1 . Since (R1 )T = (R1 )−1 it follows −1 −1 −1 T (R1 )T R1 = (R1 )−1 R1 = −1 which implies R1 ∈ SO(3). a ∈ G → ∃a−1 ∈ G with a ◦ a−1 = a−1 ◦ a = e. forms a group G satisfying the axioms (i) For any pair a. 99 (5. It follows that R3 is also an element of SO(3). (i) Obviously. 1 T Similarly. For this purpose we must demonstrate that the group properties (i–iv) hold.5. one can show R3 R3 = 1 Furthermore.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.9) forms a group. R2 ∈ SO(3) and do not write out “◦” explicitly since it represents the matrix product. We need to demonstrate that this inverse belongs also to −1 T SO(3). R3 = R1 R2 is a real. (ii) The group element ‘e’ is played by the 3 × 3 identity matrix 1 1.11) (iii) From detR1 = 0 follows that R1 is non-singular and. it holds 1. We want to prove now that SO(3) as deﬁned in (5. (iv) For products of three elements holds the associative law a ◦ (b ◦ c) = (a ◦ b) ◦ c. We would like to point out the obvious property that for all elements R of SO(3) holds R−1 = RT (5. 3 × 3–matrix.12) (iv) Since the associative law holds for multiplication of any square matrices this property holds for the elements of SO(3). It holds T T T T R3 R3 = ( R1 R2 )T R1 R2 = R2 R1 R1 R2 = R2 R2 = 1 . (iii) ∀a.

the matrices i Lk = − Jk (5. in fact. Lie Algebra We want to consider now a most convenient. presently.13) R(ϑk ek )−1 = R(ϑk ek )T = [exp ( ϑk Lk )]T = exp ϑk LT ˆ ˆ k and. establish if 1-1 homomorphic mappings exist. compact representation of R(ϑ) which will be of general use and. (b) the set of matrices {M1 . 0 1 −1 0 . M2 . −1 0 0 −1 . J2 . Generators. −i} together with multiplication as the binary operation ‘◦’. M4 } = 1 0 0 1 . mappings between the sets which conserve the group properties: (a) the set of real and imaginary numbers {1.e. identify subgroups.16) must be antisymmetric. M3 . we arrive at the property exp ϑk LT k from which follows LT = −Lk . rotation by 270o } and consecutive execution of rotation as the binary operation ‘◦’.1.20) (5.18) (5. 0 −1 1 0 (5. we want to assume that they exist. (c) the set of four rotations in the x1 .19) (5. will play a central role in the representation of many other symmetry transformations in Physics. rotation by 180o . −1.100 Theory of Angular Momentum and Spin Exercise 5. J3 as follows i R(ϑ) = exp − ϑ · J (5. due to the uniqueness of the inverse. i. k = exp ( −ϑk Lk ) (5. 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 (5.15) Below we will construct appropriate matrices Jk . This representation expresses rotation matrices in terms of three 3 × 3– matrices J1 .14) together with matrix multiplication as the binary operation ‘◦’. i.1: Test if the following sets together with the stated binary operation ‘◦’ form groups. rotation by 90o . x2 –plane {rotation by Oo .17) . 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. We want to show that for the property R RT = 1 to hold. Lie Group.

Let us use this deﬁnition to evaluate L3 . 0)T . 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 = (0.23) (5.8) we can state cosϑ3 −sinϑ3 0 1 −ϑ3 0 1 0 . ϑ3 )T such that for ϑ = (0. ϑ3 )T ] = lim sinϑ3 cosϑ3 0 = ϑ3 ϑ3 →0 ϑ3 →0 0 0 1 0 0 1 Insertion into (5. 0.24) (5.2: Determine the generator L1 according to Eq. ϑ3 and three matrices L1 .1.27) . Using (5.22).25) 0 −ϑ3 0 0 −1 0 ϑ3 0 0 = 1 0 0 . ϑ2 . This property gives A then only three independent matrix elements. ϑ2 .26) (5.(5.. 0. (5. (5. 0 0 0 0 0 0 (5. 0)T ) − 1 1 1 ϑ1 →0 (5.15) with three real parameters ϑ1 . One can determine then the matrices Lk deﬁned in (5.5. i.21) −b −c 0 It is this the reason why one can expect that three-dimensional rotation matrices can be expressed through (5.22) and similar for L2 and L3 . the real matrix A is anti-symmetric.e. 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). We assume now that we parameterize rotations through a three-dimensional rotation vector ϑ = (ϑ1 . L2 . L3 which are independent of the rotation angles. 0.1: Matrix Representation of SO(3) 101 We can conclude then that if elements of SO(3) can be expressed as R = eA . lim R[(0. A must have the form 0 a b A = −a 0 c . 0. 0)T the rotation is the identity.

B]] + [[A. 0. m = 3 for any even permutation of k = 1. Finite rotations can be obtained by applications of many inﬁnitesimal rotations of the type R(ϑ = ( 1 . x2 -.15) and their generators obey the property (5. L ] = k m Lm (5. only the latter example and .g. 2 .29) where Z1 = A + B and where the remaining operators Zn are commutators of A and B. From this result one can conclude that expressions of the type (5. e. An answer can be found considering the Baker-Campbell-Hausdorf expansion ∞ exp( λA ) exp( λB ) = exp( n=1 λn Zn ) (5. 0)T ) = exp( 1 L1 ).30) is called the Lie algebra. We want to consider.33) We want to demonstrate now that exp(ϑ · L) yields the known rotational transformations. as we see later. 0. (5. that the generators represent inﬁnitesimal rotations with 1 >> |ϑ| around the x1 -. B] ). holds the algebra [Jk . however. etc. it must hold 3 [Lk . (5. L ] = m=1 fk m Lm . if the resulting products of exponential operators can be expressed in the form (5. m = 3 for any odd permutation of k = 1..15). 1 1 e. the property (5. are related to angular momentum operators. (5.31) where k m are the elements which are 0 1 k m = 1 −1 of the totally antisymmetric 3-dimensional tensor. R(ϑ = (0. In fact. B]. ϕ)T . i. Of course. i.28) The construction of the generators through the limit taken in Eq. and x3 -axes.. in fact.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. m = 3 .g..15) obey the group property. The expression for Zn are actually rather complicated. Z2 = 2 [A.32) For the matrices Jk = i Lk which. three being a special case. (5. Z3 = 12 ( [A.e. 0)T ) = exp( 2 L2 ) and R(ϑ = (0. 3 )T ) = exp( 3 L3 ).e. however. and the constants fk m are called the structure constants. the Lie algebra of SO(3) can be written [Lk .8) in case of ϑ = (0. J ] = i k m Jm .30) are called Lie groups. The question is then. the elements of if any two indices are identical for k = 1.22) implies. 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. 0. [A.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. Lie groups can have any number of generators. B].30) Groups with the property that their elements can be expressed like (5. = 2. = 2. In case of the group SO(3) the structure constants are particularly simple. commutators of commutators of A and B. that the operators (5. = 2. the matrix (5.

(5.37) = (−1)n 1 0 0 1 . (5.39) Recognizing the Taylor expansions of cos ϕ and sin ϕ one obtains eA = and. that the matrix ϕL3 .27).1.1: Matrix Representation of SO(3) determine exp(ϕL3 ). eϕL3 which agrees with (5. cos ϕ − sin ϕ 0 = sin ϕ cos ϕ 0 0 0 1 (5.38) eA = n=0 (−1)n 2n ϕ (2n)! 1 0 0 1 ∞ + n=0 (−1)n ϕ2n+1 (2n + 1)! 0 −1 1 0 . accordingly.35). We note. (2n + 1)! (5. 0 −1 1 0 = (−1)n 0 −1 1 0 (5.8). A = ϕ 0 −1 1 0 . ∞ 2n = − 1 0 0 1 2n+1 (5. (5.31).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) . 5.3: Show that the generators Lk of SO(3) obey the commutation relationship (5. using (5.41) cos ϕ − sin ϕ sin ϕ cos ϕ (5.34. according to (5.36) The property 0 −1 1 0 allows one to write 0 −1 1 0 and. Exercise 5.34) 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 . can be written ϕL3 = 0 103 A 0 0 0 0 .

(5.48) .1: Assume the deﬁnition ρ ρ (A B) = ρ (B) ρ (A).2. a relationship which we like to express as follows R(ϑ) = ρ R(ϑ) .44) holds.e.104 Theory of Angular Momentum and Spin 5. (5.44) 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) . the transformation R(ϑ): C∞ ( ) → C∞ ( ) is related to the transformation R(ϑ): R → R . hence. one isomorphic to SO(3) and. the R(ϑ) ψ(r) = ψ(R(ϑ) r ).42) Obviously. we deﬁne rotations R(ϑ) as linear maps C∞ ( ) → C∞ ( ). Since this holds for any ψ(r ) one can conclude ρ R(ϑ1 ) ρ R(ϑ2 ) group property (5. For this purpose we require stringent continuity properties of the wavefunction: the wavefunctions under consideration must be elements of the set C∞ ( ). (5. Exercise 5.43) ρ( ) conserves the group property of SO(3). In analogy to R(ϑ) being a linear map R → R . In particular.e. i. in fact. Show that in this case holds Generators One can assume then that R(ϑ) should also form a Lie group.46) (5. for A.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). i.. we will learn that transformation patterns and invariances are connected with the angular momentum states of quantum mechanics.45) (5. its elements can be represented through an exponential R(ϑ) = exp i − ϑ·J (5. Deﬁnition We ﬁrst deﬁne how a rotational transformation acts on a wavefunction ψ(r ). namely R(ϑ)ψ(r ) = ψ(R−1 (ϑ) r ) .47) = ρ R(ϑ1 ) R(ϑ2 ) . B ∈ SO(3) holds ρ(A B) = ρ(A) ρ(B) . i.e. complex-valued functions over the 3-dimensional space R which can be diﬀerentiated inﬁnitely often. = ψ([ R(ϑ1 ) R(ϑ2 ) ]−1 r) = ρ R(ϑ2 ) ψ([R(ϑ1 )]−1 r) (5.

in case of J3 .56) We like to verify this property for [J1 . it holds [Jk . 5. (5. 0.51) Inserting this into (5. show that (5.54) (5.49) to a function f (r).53–5.5.e. 0.22) according to i − J1 = lim ϑ−1 R(ϑ = (−ϑ1 . k = 1. − 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) . 0. −ϑ3 ) r ) − f (r) ) .2: (a) Derive the generators Jk .49).53) (5. but also the Lie algebra of the generators. (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 (5.52) ( x2 ∂1 − x1 ∂2 ) f (r) .23. J2 ]. The generators Jk can be determined by applying (5.2. 0)T ) − 1 1 ϑ1 →0 105 (5..50) lim ϑ−1 ( f ( R(0. J1 ].24) one can expand ϑ3 →0 ϑ3 →0 lim ϑ−1 R(ϑ = (0. J3 ] and [J3 . 0. J2 ] f (r) = − = − 2[ 2 [− i J .56) holds for [J2 .55) Not only is the group property of SO(3) conserved by ρ( ). i − J3 f (r) = = Using (5. J ] = i k m Jm .57) Exercise 5. (5. One obtains using (5. (5.(5. ϑ3 )T ) f (r) − f (r) 3 (5.55) and f (r) = f [J1 .42)) and similarly for J2 and J3 . 0 0 1 x3 x3 (5. −ϑ3 ) r ≈ −ϑ3 1 0 x2 = −ϑ3 x1 + x2 .2: Function Space Representation of SO(3) where the generators can be determined in analogy to Eq. i. 2 by means of limits as suggested in Eq. . 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 . In fact.49) (note the minus sign of −ϑ1 which originates from the inverse of the rotation in Eq. 3 1 ϑ3 0 x1 x1 + ϑ3 x2 R(0. (5.

Using [AB.53–5.63) (5.106 Theory of Angular Momentum and Spin Exercise 5. the operator 2 2 2 J 2 = J1 + J2 + J3 (5.2. J3 ] + [J2 . i. 3 . identiﬁes J as the quantum mechanical angular momentum operator.65) . φ). J3 ] + [J1 .56) are the famous commutation relationships of the quantium mechanical angular momentum operators. according to the correspondence principle between classical and quantum mechanics.3: Consider a wave function ψ(φ) which depends only on the azimutal angle φ of the spherical coordinate system (r. (5.e.55) can be written as a vector product of r and i − J = −r × From this follows. . However.58) i = r ×p ˆ (5.64) mY m (θ. To show this we ﬁrst note that (5. 5. φ) as well as the eﬀect of the so-called raising and lowering operators J± = J1 ± i J2 (5. These eigenstates are the well-known spherical harmonics Y m (θ. φ) m (θ. i.61) We demonstrate this property for k = 3. 3. C] = A[B. J3 ] = = = 2 2 [J1 + J2 . φ).. φ) (5. using the momentum operator p = ˆ J = r × i .61) simultaneous eigenstates of J 2 and of one of the Jk . usually chosen as J3 . J3 ] (5.e. Jk ] = 0 . and x3 = r cosθ. We will show now that the properties of spherical harmonics J2Y J3 Y m (θ.59) which.3 Angular Momentum Operators The generators Jk can be readily recognized as the angular momentum operators of quantum mechanics. The commutation relationships imply that no simultaneous eigenstates of all three Jk exist. x2 = r sinθ sinφ. can be found. (5. Show that for J3 as deﬁned above holds exp( iα J3 ) ψ(φ) = ψ(φ + α). k = 1.62) J1 [J1 . 2. k = 1.60) commutes with all three Jk . φ) = = 2 ( + 1) Y m (θ. 2.56) one obtains [J 2 . J3 ] J2 −i J1 J2 − i J2 J1 + i J2 J1 + i J1 J2 = 0 According to (5. Equation (5. the system related to the Cartesian coordinates through x1 = r sinθ cosφ. C]B and (5. θ. J3 ] J1 + J2 [J2 . [J 2 . C] + [A.

(∗) (5.1 Let Lk . i. It holds L+ L− | = (L+ L− − L− L+ + L− L+ )| = ([L+ . φ) − m (θ. φ) (θ.e.56). L− ] + L− L3 ) | . Continuing our proof through induction we assume now that property (i) holds for m. α −1 β −1 = 2 .71) To prove this theorem we ﬁrst show by induction that (i) holds. φ) are a consequence of the Lie algebra (5. φ) 107 = = 0 = = 0 ( + m + 1)( − m) Y ( + m + 1)( − m) Y m+1 (θ.3: Angular Momentum Operators on the spherical harmonics. and since we have already shown | − 1 = (iβ −1 )−1 L− | we can state L3 | − 1 = −i( − 1)| − 1 . L ] = k m Lm and let L± = L1 ± i L2 . We note L3 L− | = ( [L3 . An Important Theorem Theorem 1.72) = 0 = −i | (5.70) (5. i. L+ raises the m-value of | − 1 from m = − 1 to m = .73) (5.76) 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 . 2. For this purpose we prove the following theorem. The deﬁnitions of the coeﬃcients α −1 and β −1 yield L+ L− | = L+ ( iβ −1 | − 1 = − α −1 β −1 | .66) (5. L2 = L2 + L2 + L2 . in fact. Noting [L+ .68) (5. We will show that this property holds then also for m − 1. φ) (5. L− ] + L− L+ )| . L− ] + L− L+ )| m = ( −2iL3 + . 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. namely.5.75) (5. k = 1. L− ] = [L1 + iL2 . J+ Y J− Y J− Y m (θ. From this follows that L− | is an eigenstate of L3 .e. L1 − iL2 ] = L2 + iL1 = i (L1 − iL2 ) = iL− and L3 | = −i | we obtain L3 L− | = ( iL− + L− L3 )| = i(− + 1) L− | . We consider L+ L− | m = ( [L+ . L1 − iL2 ] = −2iL3 and L+ | = 0 one obtains L+ L− | = −2iL3 | = −2 | . L− ] = [L3 . Using [L3 . For this purpose we ﬁrst demonstrate that the property holds for m = − 1 by considering (i) for the state L− | ∼ | − 1 .69) J+ Y (θ..67) (5.74) (5. φ) m+1 (θ. The arguments are very similar to the ones used above and we can be brief.

5. 5.79) read L+ | m L− | m + 1 = −i = −i ( + m + 1)( − m) | m + 1 ( + m + 1)( − m) | m .80) (5.64). We can deduce from (∗) and (∗∗) α α α −1 β −1 −2 β −2 −3 β −3 = 2 = 2( − 1 + ) = 2( − 2 + · · −1 + ) Obviously. in terms of spherical coordinates deﬁned through x1 x2 x3 = = = r sin θ cos φ r sin θ sin φ r cos θ (5.55). We note that we can write L2 = 1 1 1 1 2 2 2 2 L+ L− + 2 L− L+ + L3 . given in (5. Again the argumemts are similar to the ones used above. in particular. One can normalize the states | m such that αm = βm . 5. A ﬁrst application will involve the construction of the eigenstates of the angular momentum operators as deﬁned in (5. yield properties (5.82) (5.77) We will show now that if (ii) holds for m.72. (∗∗) (5.79) We can now show that (iii) holds for all proper m–values. This completes the proof of the theorem. Angular Momentum Operators in Spherical Coordinates We want to express now the generators Jk .108 Theory of Angular Momentum and Spin L− L+ )| m = (−2m − αm βm )| m .81) i. (5. 5. For this purpose we need to express the angular momentum operators in terms of spherical coordinates (r. from L+ L− | m = − αm−1 βm−1 | m follows the recursion relationship αm−1 βm−1 = 2m + αm βm . φ. θ. it also holds for m − 1.84) . ﬁnally αm = βm = (5. (5.66.78) ( + m + 1)( − m) ( + m + 1)( − m) ..63. 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 . We will have various opportunities to employ the Theorem just derived. This implies that (i) holds for m − 1. We note L3 L− | m = ( [L3 . 5..73. Using the familiar formula + m − n k=0 k = n (n + 1)/2 − (m + 1) m = ( + 1) m − (m + 1) m (5.69). 2 We note that Eqs. it holds αm βm = 2 one obtains αm βm = = ( + 1) k=m+1 k .e.53–5.83) (5.e.70. i. 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 . Before we can ﬁnally show that property (iii) holds as well for all − ≤ m ≤ we need to determine the coeﬃcients αm and βm .

for example. 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 fact. involve only the angular variables θ and φ and not the radius r. φ) = ( x3 ∂2 − x2 ∂3 ) f (θ. φ) (5. actually. using (5.82–5.92) (5.93) and similarly. L1 f (θ. φ) = (5. (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 . ∂φ = sin φ (5. cos θ ∂ cos2 φ + sin θ cos φ ∂φ cos2 θ sin θ sin φ ∂ sin θ sin φ ∂ + + sin2 θ f (θ.85) (5. using ∂/∂tanφ = cos2 φ ∂/∂φ and ∂/∂cosθ = −(1/sinθ) ∂/∂θ. (5. L1 f (θ. According to (5.84 ) one obtains.85). in terms of spherical coordinates. the latter deﬁned in (5. Using (5. φ).89) = ∂ x2 x2 ∂ x2 + x2 ∂ x3 − 33 − x2 1 3 2 x1 ∂ tan φ r ∂ cos θ r ∂ cos θ f (θ. applying repeatedly the chain rule.86) (5.87) To derive these properties we consider.91) which agrees with expression (5.3: Angular Momentum Operators 109 We ﬁrst like to demonstrate that the generators Jk .86). using (5. (L2 )2 = − cos φ ∂ ∂ + cot θ sin φ ∂θ ∂φ 2 .85). φ) = (x3 ∂2 − x2 ∂3 )f (θ. 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 ∂ = − .5.60). φ) sin θ ∂θ sin θ ∂θ L1 f (θ.87) holds J3 = To determine J 2 we note.88) (5. φ) ∂x ∂x3 ∂ tan φ =0 (5.90) This yields. We like to express now the operators J3 and J 2 .

∂θ ∂φ ∂2 ∂ ∂2 + cotθ + ( cot2 θ + 1 ) 2 .4 Angular Momentum Eigenstates According to the theorem on page 107 above the eigenfunctions (5.95) sinθ ∂ ∂θ = ∂2 ∂ + cotθ .97) Kinetic Energy Operator The kinetic energy of a classical particle can be expressed J2 p2 p2 = r + class 2m 2m 2m r2 (5. 2m r ∂r2 2m r2 2 (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 + . 2 ∂θ ∂θ (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.e. One ﬁrst identiﬁes the Hilbert space H on which the generators Lk operate.86. These generators act on a subspace of C∞ ( ). 5. 5.98) where Jclass = r × mv is the angular momentum and pr = mr is the radial momentum. 1 sin2 θ sinθ ∂ ∂θ (5.64) can be constructed as follows.94) It follows (L1 )2 + (L2 )2 + (L3 )2 = With cot2 θ + 1 = 1/sin2 θ.87). 5.86.99) = = ∂2 2 ∂ 1 + + 2 2 2 ∂r r ∂r r sin θ ∂2 2 ∂ + 2 ∂r r ∂r sinθ ∂ ∂θ 2 + ∂2 ∂φ2 (5.85. one can conclude J2 = − 2 sin θ 2 sinθ ∂ ∂θ 2 + ∂2 ∂φ2 . The operators in the present case are Lk = − i Jk where the Jk are diﬀerential operators in θ and φ given in (5.96) and the relationship between Lk and Jk as given in (5. (5.100) (5. C∞ (S ). 5.101) 1 ∂2 r r ∂r2 and comparision with (5. ∂θ2 ∂θ ∂φ (5. The functions in C∞ (S ) have real variables . namely on the space of complex-valued functions on the unit 3-dim.97).85. sphere S2 . i.63.87). 5.

104. 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.111) .85–5. − + 1. 0 ≤ φ < 2π. φ).5. (5. . φ)f (θ. 5.109) using (5. One obtains in this way Y m (θ. φ) (5. φ) L3 Y (θ.87). The raising and lowering operators L± = L1 ± i L2 .107) reads ∂ ∂ + i cotθ ∂θ ∂φ P (cosθ) ei φ i e±iφ ∂ ∂ ± i cotθ ∂θ ∂φ . φ) in H which satisfy L+ Y (θ. . φ) = ∆( . must be cyclic in φ with period 2π. m) Constructing Y → Y →→ ··· Y ( + m)! (2 )! ( − m)! −1 − We like to characterize the resulting eigenfunctions Y m (θ. (5. − . normalizing these functions. and to be admissable for description of quantum states. can be expressed L± = Accordingly. . . . (5. φ).103) iteratively for m = − 1.106) The latter property is obviously satisﬁed for the chosen φ-dependence. the so-called spherical harmonics.107) (5. φ) = = 0 −i Y (θ.105) ∆( . φ) = −i ( + m + 1)( − m)Y m (θ. φ) = 2 + 1 ( − m)! P 4π ( + m)! m (cosθ) eimφ . and then determining the family {Y m (θ. m = − . φ) = −i Y (θ.4: Angular Momentum Eigenstates 111 θ.104) (5.102) exists.64) can be constructed then by seeking ﬁrst functions Y (θ. } applying L− Y m+1 (θ. . . φ) (5. (5. by carrying out the construction according to (5. 5.105) explicitly.108) m (θ. (5.102) where dΩ = sinθ dθ dφ is the volume element of S2 .110) = 0 (5. is indeed a Hilbert space. 0 ≤ θ < π.63. The norm in H is deﬁned through |f |2 = S2 dΩ f ∗ (θ. − 2. The function space endowed with this norm and including only functions for which the integral (5. m) = 1 −m J− 1 2 Y (θ. φ). φ) (5. The eigenfunctions (5. φ) = 0 as well as L3 Y (θ. . φ. φ) .

5.116) implies |N |2 from which we conclude N = (−1) 1 · 3 · 5 · · · (2 − 1) (5.D.119) where the factor (−1) has been included to agree with convention1 . is P (cos θ) = N sin θ as one can readily verify. i.118) 2 +1 1 1 sin θ ei 4π (2 )! 2 ! φ . Jackson (John Wiley. for example.106. (5.120) See. (5. (5.113) 0 is ﬁnite. . The normalization factor N is chosen such that π (5.107). using the new integration variable x = cos θ.119) provide the following expression for Y Y (θ. 1975) ..114) yield |N |2 2 +1 1 2π 4π (2 )! π dθ sin2 0 +1 θ = 1.e. = = 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. ”Classical Electrodynamics.116) The integral appearing in the last expression can be evaluated by repeated integration by parts.112 or Theory of Angular Momentum and Spin ∂ − ∂θ cotθ P (cosθ) = 0 (5. A proper solution of this equation.115) holds. φ) = (−1) 1 1 = 1 [1 · 3 · 5 · · · (2 − 1)]2 (5.112) where we employed the deﬁnition of P (cosθ) in (5. One obtains.114. New York.106. . 5. 2nd Ed. We note that (5.” by J. 5.114) dθ sinθ |Y (cosθ)|2 = 1 0 (5. π +1 dθ sin2 0 +1 θ = −1 +1 −1 dx (1 − x2 ) +1 = = .117) Accordingly. one for which π dθ sinθ |P (cosθ)|2 (5. (5.

therefore..124) called the associated Legendre polynomials. (5. i.124) holds for m + 1. . .124). m (cos θ) e (5. P m (x) = −1 ( + m)! 2 ! ( − m)! × (1 − x2 )− m+1 2 −(1 − x2 ) 2 1 ∂ x + (m + 1) 1 ∂x (1 − x2 ) 2 × (5.114. ∂x −m−1 . We want to demonstrate now that the recursion equation (5. 5. every application of L− reduces the power of eiφ by one. and can obtain now.121) This expressions shows that the factor eimφ . we seek to determine the functions P m (cos θ) and. For m = (5.106).124) we proceed by induction. according to (5.5. 2 ! ( − m − 1)! ∂x −m−1 (5.125) which agrees with (5. indeed.e. namely Y (θ. P m+1 (x) = −m−1 m+1 ∂ 1 ( + m + 1)! (1 − x2 )− 2 (x2 − 1) .123). use (5.124) reads P (x) = = (2 )! (1 − x2 )− 2 (x2 − 1) 2 ! (−1) 1 · 3 · 5 · · · (2 − 1) sin θ (5. . through repeated application of (5. − deﬁned in (5. the eigenfunctions Y m (θ. describes the φ-dependence of Y m (θ. employing again the variable x = cos θ. as speciﬁed in (5.122) The latter identity can be written. Let us then assume that (5.119).63.103. are real.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 . The reader should note that the associated Legendre polynomials. φ).64).103.126) Then holds.123) P m+1 (x) ∂ x + (m + 1) 1 ∂x (1 − x2 ) 2 . m = − 1. Actually. P m (x) = −1 × ( + m + 1) ( − m) × −(1 − x2 ) 2 1 (5. − 2.110).4: Angular Momentum Eigenstates 113 We have constructed a normalized solution of (5. 5.121) 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) . To prove (5.110) together with (5. φ). 5.127) ∂ −m−1 (x2 − 1) .123) leads to the expression P m (x) = −m m ∂ 1 ( + m)! (1 − x2 )− 2 (x2 − 1) 2 ! ( − m)! ∂x −m (5. 5. φ). (5.

Y ∗ (θ. using x = cos θ.132). φ) using (5. appears not very interesting.128) one can show that (5. that expression (5. from such construction emerges an important symmetry property of Y m (θ.124). φ) An alternative route to construct the eigenfunctions Y of L− f (θ..106. 5. therefore. We ﬁrst determine Y − (θ. yields a vanishing expression for m = − − 1 as long as is a non-negative integer. which is equivalent to recursive application of L− .124).124) and. φ) L3 f (θ. etc.e. i. (5. φ) = Y − (θ. Constructing Y − → Y − +1 → ···Y m (θ. P − (x) = 1 ∂2 (1 − x2 ) 2 (x2 − 1) .132) which reduces the number of spherical harmonics which need to be evaluated independently roughly by half .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. 2 ! (2 )! ∂x2 1 (5.133) Since the term with the highest power of (x2 − 1) is x2 . It holds. In that case (1 − x2 ) is a polynomial of degree 2 and. φ).129) holds. φ) = = determines ﬁrst a normalized solution 0 i f (θ. φ) = 0 (5. φ) = (−1)m Y m −m (θ.124) and. the derivative ∂ 2 +1 /∂x2 +1 of this expression vanishes. hence.135) . We want to test if the recursion (5. Since (5. it holds then for all m.127) reproduces (5.130) (5. P − (5. therefore. However. φ) as given in (5.106. φ) . if L− Y (θ. φ) (5. we embark on this construction in order to prove (5.124) holds for m = . namely. Such construction reproduces the eigenfunctions Y m (θ. and constructs then the eigenfunctions Y − +1 .134) (x) = 1 2 ! (1 − x2 ) 2 . it holds ∂2 (x2 − 1) = (2 )! ∂x2 and.124) holds for m if it holds for m + 1. therefore. φ) choosing the proper sign. expression (5. therefore. 5. In fact. Y − +2 . (5. by repeated application of the operator L+ .123) terminates for m = − .131) identiﬁes f (θ.

. According to (5.143) .e.139) = ∂ − m cotθ ∂θ P m (cos θ) . the expression (5. (5.129).106.141) 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 (5.124) one arrives at Y (θ. (5. (5.110) applies the = ( + m + 1)( − m) Y m+1 (θ.136) We note in passing that this expression and the expression (5. Obviously. a sign consistent with the family of functions Y m constructed above. We ﬁrst note that for m = − (5.f. i.138) Employing (5.5. 5.137) m (θ. (5. φ) = 2 +1 1 1 sin θ e−i 4π (2 )! 2 ! φ 115 − . 5. φ) sin θ e−i φ = 0.106).140) Introducing again the variable x = cos θ leads to the recursion equation [c. (5. (5.120) for Y obey the postulated relationship (5.136) to construct all other Y recursion eiφ ∂ ∂ + i cotθ ∂θ ∂φ Y m (θ.132).142) yields P − (x) = (−1) 1 (1 − x2 )− 2 (x2 − 1) = (1 − x2 ) 2 2 ! 2 ! (5.80) and (5. 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. following closely the proof of eq. One can readily verify that this expression provides a solution of (5. φ) . This follows from the identity ∂ ∂ − i cotθ ∂θ ∂φ One can use (5.123)] P m+1 (x) = −(1 − x2 ) 2 1 ∂ x − m 1 ∂x (1 − x2 ) 2 P m (x) . (5.124).142) For this purpose we proceed by induction.136) is normalized and has the proper sign. φ).4: Angular Momentum Eigenstates Due to (5.122.

However..124) and (5. 1 (1 − x ) ∂x +m (1 − x2 ) 2 1 (5. one can conclude P m (x) = (−1)m m (θ. The construction beginning with Y − and continuing with Y − +1 . i. (5. 2 ! ∂x +m+1 (5. given by (5. Hence.148) According to (5.142) holds for m + 1 if it holds for m. We then assume that (5. 2 ! ∂x 1 (5.h. Accordingly.e.135).106) this implies for Y The Legendre Polynomials The functions the identity (5..141) reads then P m+1 (x) = 1 (5. (5.150) . also the associated Legendre polynomials determined this way. φ) ( − m)! P ( + m)! −m (x) .149) (5. yields the same eigenfunctions as the previous construction beginning with Y and stepping down the series of functions Y −1 .147) the r. Y −2 .144) reads P m+1 (x) = +m+1 m+1 ∂ (−1)m+1 (1 − x2 ) 2 (x2 − 1) . (5.116 Theory of Angular Momentum and Spin wich agrees with (5.146) i.132).e. (5.s.144) ∂ x − m 1 ∂x (1 − x2 ) 2 +m m ∂ (−1)m (1 − x2 ) 2 (x2 − 1) .142) and by (5. 2 ! ∂x +m −(1 − x2 ) 2 Replacing in (5. Y − +2 . etc. of which agrees with the r.142) holds for m.142). Since (5.s. According to both (5.124) yields (−1)m ( − m)! P ( + m)! −m (x) = +m m ∂ (−1)m (1 − x2 ) 2 (x2 − 1) .142) holds for m = − we veriﬁed that it holds for all m.142) one can state P (x) = ∂ (x2 − 1) . equivalently. of (5. one notes that application of (5. etc. P (x) = P 0 (x) are called Legendre polynomials.128) m + 1 → −m or.145) Hence. 2 ! ∂x +m (5. are identical.124). 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 .h.

(5. m ≥ 0 ∂xm (5. The spherical harmonics for m < 0 can be obtained using (5. P3 (x) = 1 2 117 P2 (x) = − 3x) 1 2 2 (3x − 1) .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) . and r2 denotes the location of the charge. 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..D. (5. P1 (x) = x . in case of x = 1 holds w(1. . accordingly.150) and (5. In case q = 1 and |r2 | < |r1 | holds the identity2 ∞ 1 r2 = P (cos γ) (5. pp. ”Classical Electrodynamics.156) yields P (1) = 1 .155) P (x) t .153) P (cos θ) eimφ . For example.158) see.J.151) (5x3 Comparision of (5. r1 is the point where the potential is measured. New York. φ) = 2 + 1 ( − m)! (−1)m sinm θ × 4π ( + m)! × ∂ ∂ cos θ m (5. and |r1 − r2 | = r1 (5.” by J.154) can be written 1 = w(x. . Englewood Cliﬀs. The generating function allows one to derive useful properties of the Legendre polynomials. .g. Using x = cos γ. t) = √ Comparision with (5.N. 2. t = r2 /r1 . pp.157) = 0. N. 1975). 2 3 1 1 = = 2 1−t 1 − 2t + t ∞ t . for example. t) is called a generating function of the Legendre polynomials3 . the spherical harmonics for m ≥ 0 Y m (θ.5. Lebedev (Prentice-Hall. =0 (5.. 45 of “Special Functions and their Applications” by N. =0 (5.132).4: Angular Momentum Eigenstates The ﬁrst few polynomials are P0 (x) = 1 . (5. Jackson (John Wiley. e. 1. 1965) which is an excellent compendium on special functions employed in physics.154) |r1 − r2 | r +1 =0 1 where γ is the angle between r1 and r2 . 2nd Ed. t) = √ 1 − 2xt + t2 ∞ 1 − 2 x t + t2 . . 92 To prove this property see.152) and.156) w(x.

Accordingly.151)] allows one to determine P (x) for = 1. Accordingly. π + φ).156) and (∂/∂t) to P (x)t = ∞ ∂w + (t − x) w = 0 . . π + φ).164) which.118 For w(x. Inspection of (5. ∂t P (x)t −1 −1 (5.160) this diﬀerential equation is equivalent ∞ (1 − 2xt + t2 ) =0 P (x) t + (t − x) =0 P (x) t = 0 (5.159) Using ∂lnw/∂t = (1/w) ∂w/∂t and multiplying (5. .162) − (2 + 1)x P (x) + P −1 (x) ] t + =1 [( + 1) P +1 (x) The coeﬃcients for all powers t must vanish individually.159) by w(1 − 2xt + t2 ) leads to the diﬀerential equation obeyed by w(x. φ) m (θ. If the spherical coordinates of r are (r. φ) . .161) Collecting coeﬃcients with equal powers t yields 0 = P1 (x) − x P0 (x) ∞ (5. φ) Under inversion at the origin vectors r are replaced by − r.168) mY m (θ. φ) . (5.167) (5. Due to cos(π − θ) = − cos θ the replacement r → −r alters P m (x) into P m (−x). t) (1 − 2xt + t2 ) Employing (5. φ) = = 2 ( + 1) Y m (θ. θ. ∂t 1 − 2xt + t2 (5. 1. Inversion Symmetry of Y m (θ.f.165) since ∂ n /∂(−x)n = (−1)n ∂ n /∂xn . φ) (5. 2. holds P1 (x) ( + 1) P +1 (x) = = x P0 (x) (2 + 1) x P (x) − P −1 (x) (5. .163) (5. under inversion Y m (θ. π + φ) = (−1) Y m (θ. using P0 (x) = 1 [c. φ) m (π − θ. Noting exp[im(π + φ)] = (−1)m exp(imφ) we determine Y Properties of Y m (θ. t) holds Theory of Angular Momentum and Spin ∂lnw(x. φ) m (θ. (5. π − θ. then the coordinates of − r are (r. φ) goes over to Y m (π − θ. t) x−t = .166) We want to summarize the properties of the spherical harmonics Y derived above. φ).142) allows one to conclude P m (−x) = (−1) +m P m (x) (5. The spherical harmonics are eigenfunctions of the angular momentum operators J2Y J3 Y m (θ..

where ( = 1. m 4. For the Legendre polynomials holds the orthogonality property +1 dx P (x) P ( (x) = −1 2 δ 2 +1 . P −1 (x) ] (5. for any f (θ.4: Angular Momentum Eigenstates 119 2.172) (5. φ) m (θ. a complete basis of C∞ (S2 ).. i.169) 3. φ)f (θ. The spherical harmonics are given by the formula Y m (θ. . φ) C = =0 m=− π C sin θdθ 0 m Y m (θ. . The spherical harmonics form an orthonormal basis of the space C∞ (S2 ) of normalizable. φ) = (−1)m Y ∗ (θ. P +1 (x) P1 (x) = x . φ) . The spherical harmonics obey the recursion relationships J+ Y J− Y m (θ. 5. φ) 2π 0 (5.5. φ) = 2 + 1 ( − m)! (−1)m sinm θ × 4π ( + m)! 2 ! × ∂ ∂ cos θ m (5. The spherical harmonics for m < 0 are given by Y −m (θ.174) m ≥ 0 P (cos θ) eimφ . uniquely deﬁned functions over the unit sphere S2 which are inﬁnitely often diﬀerentiable π 2π sin θdθ 0 0 dφY ∗m (θ.176) = 1 +1 [ (2 + 1) x P (x) − are the Legendre polynomials. φ) = 2 +1 P (cos θ) .171) m = dφ Y ∗ (θ. (5. . 6. 4π (5.178) 7.177) Note that the spherical harmonics are real.179) .175) (5.e. The spherical harmonics form. m (5. φ) where J± = J1 ± iJ2 . φ) Y m (θ. 2. The spherical harmonics Y 0 are Y 0 (θ.173) m+1 (θ. φ) ∈ C∞ (S2 ) holds ∞ f (θ.170) (5. φ) . in fact. φ) = = ( + m + 1)( − m) Y ( + m + 1)( − m) Y m+1 (θ. except for the factor exp(imφ). (5. φ) (5.) P0 (x) = 1 . φ) = δ δm m .

156). =0 t + −1 dxP (x) P (x) .87). φ) .s. The spherical harmonics for m (θ.181) = 0. t) stated in (5. The spherical harmonics obey the inversion symmetry Y 10.180) 9.4. φ) = (−1) Y m (π − θ. 4π (5.184) Y22 Y21 Y20 together with (5. (5.185) (5. For this purpose start from the identity +1 ∞ +1 dx w(x. 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 (5.177).189).h. φ). (5.1: Derive expressions (5.187) h(r) Y m (θ. φ) = 1 ∂2 r − r ∂r2 ( + 1) r2 h(r) Y m (θ.h. .186) (5. (5.86).183) (5.182) Y11 Y10 = = − (5.188) Exercise 5. 1. Exercise 5. 11. (5.189) evaluate the integral on the l.s.120 8.3: Construct all spherical harmonics Y3m (θ. t)2 = −1 .4.2: Derive the orthogonality property for the Legendre polynomials (5.4. For the Laplacian holds 2 = = = (5. expand the result in powers of t and equate the resulting powers to those arising on the r. π + φ) . Exercise 5. φ) = δm0 2 +1 . of (5. The spherical harmonics for θ = 0 are Y m (θ Theory of Angular Momentum and Spin = 0.179) using the generating function w(x..

called the irreducible representation. . φ) c = . (5.m m m (θ. and (ii) they form the lowest dimensional sets X obeying C∞ (S ) = X which have this invariance property. i. . .193) shows ∞ = − ..5 Irreducible Representations We will consider now the eﬀect of the rotational transformations introduced in (5.42). exp − i ϑ · J (X ) = X . .195) i..5. φ).. . φ) (5. m = S2 i dΩ Y ∗m (θ.. 1. For this purpose we consider the subspaces of C∞ (S ) X = [ {Y .. φ) = exp Since the spherical harmonics Y space C∞ (S ) one can expand ˜ f (θ. ˜ f (θ. . . .e. φ) D(ϑ) = . i. assumes a particularly simple form. therefore. (5.196) C∞ (S ) = = X ... φ) . with elements given by (5. if one speciﬁes the functional form of the coeﬃcients [D(ϑ)] m . φ) (5. on functions f (θ. φ) i − ϑ·J f (θ. The transformations exp − i ϑ · J . φ) exp − ϑ · J . . m .194) Y m (θ.192) = S2 dΩ Y ∗ (θ . φ) ˜ under such rotational transformation by f (θ.190) provide a complete. φ ) exp m One can also represent C∞ (S ) C∞ (S ) = [ {Y .191) i − ϑ·J f (θ .193) where [ { } ] denotes closure of a set by taking all possible linear combinations of the elements of the set. (5. .. = . m Y m (θ.. ∞. }. − + 1. = .− + .. . .5: Irreducible Representations 121 5. We denote the image of a function f (θ. }] . . .e. = − .195). We like to argue that the matrix representing the rotational transformations of the function space C∞ (S ). }] (5. . φ) m. φ ) (5.− + . = 0.197) The subspaces X have the important property that (i) they are invariant under rotations exp − i ϑ · J . ∞.e. φ) ∈ C∞ (S ). Comparision with (5.. 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 .m D(ϑ) m. . (5. are characterized completely if one speciﬁes the transformation of any Y m (θ. orthonormal basis for the function c m Y m (θ. φ) ˜ Y m (θ. m = − .. exp − i ϑ · J .

e. n The action of J− on Y m produces Y m−1 or.. J− . φ) .. (5. 5.200) leaves X invariant.e.201) From expression (5.200) of the rotational transformations one can also conclude that the X are the smallest invariant subspaces of C∞ (S ). φ) exp m i − ϑ·J Y m (θ. produces 0. in case m = − .65). J± ] = ± J± (5.194) only spherical harmonics with or. The exponential operator exp − i ϑ · J + ϑ3 J3 )n . Since any additive term n n n cn1 .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 .56. 3. one concludes for the matrix elements (5.200) in (5.. i. If one orders the basis according to the partitoning (5. i.n2 . one needs to consider the action of J+ 1 J− 2 J3 3 on the subspaces X .n3 n n n Accordingly.. (5. 5.197) of C∞ (S ) the matrix representation of D(ϑ) assumes a block-diagonal form. (5.199) allows one to express exp i − ϑ·J = n1 . with blocks of dimension 1. can be expanded in powers One employs then the commutation properties [J3 .n2 .n3 J+ 1 J− 2 J3 3 . m = δ S2 dΩ Y ∗ (θ. i. Since the Y m are n3 eigenfunctions of J3 the action of J3 reproduces the basis functions of X . . . equivalently. . Application of (5. leaves X invariant. One proceeds by expanding ﬁrst ϑ · J in terms of J+ . . which follow readily from (5.198) 2 2 where ϑ± = ϑ1 ( 1 ϑ− J+ 2 + 1 2 ϑ+ J− ϑ2 .. one can conclude that J+ leaves X invariant. . the transformation exp − i ϑ · J leaves X invariant as well.. J− 2 also leaves n1 X invariant. One = contribute can conclude then that in the expansion (5.202) D(ϑ) = (2 + 1) × (2 + 1) . J− ] = 2 J3 .199) [J+ . . Similarly. 1×1 3×3 .195) [D(ϑ)] m. and J3 1 1 ϑ · J = ϑ− J+ + ϑ+ J− + ϑ3 J3 (5.e.

5. (f) The transformation ρ(ϕ) as deﬁned in (e) leaves the following subspace of functions considered in (e) Xm = { f (α) = A e−imα . Give the corresponding expression of ρ(ϕ). f (α + 2π) = f (α). However. (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). for which holds RT R = R RT = 1 1. The representation is referred to as the irreducible representation. though seemingly natural. The three components ϑk .6: Wigner Rotation Matrices 123 This representation of D(ϑ) is the one wich has blocks of the lowest dimensions possible.e. (e) In the space C∞ ( ) of inﬁnitely often diﬀerentiable. Exercise 5.49). 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. (ii) a second rotation around the new x2 –axis by an angle β. (b) Prove that the elements of SO(2) can be completely characterized through a single parameter. (a) Show that SO(2) with the group operation ◦ deﬁned as matrix multiplication. and periodic functions f (α). This step reconsiders the parametrization of rotations by the vector ϑ assumed sofar. Determine the generator of ρ(ϕ) in analogy to (5. A ∈ C. this parametrization. ∈ N} invariant. and for which detR = 1. (d) Show that the similarity transformation T (ϕ) deﬁned in the space of non-singular. i. 2. 5. is a group. (iii) a third rotation around the new x3 –axis by an angle γ. 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). k = 1.1: Representations of the group SO(2) SO(2) is the set of all 2×2 matrices R which are orthogonal.e.22. i. 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). . other representations are termed reducible representations.6 Wigner Rotation Matrices We will now take an important ﬁrst step to determine the functional form of the matrix elements of D(ϑ).5. 3 certainly allow one to describe any rotation around the origin. does not provide the simplest mathematical description of rotations.

205) Y m (θ. i.e. followed by transformation (i).207) are equivalent.s. in terms of rotations deﬁned with respect to the original frame. m i Y m (θ. m [D(α. φ) (5. β. For this purpose we notice that J2 can be expressed through the similarity transformation J2 = e− i αJ3 J2 e i αJ3 (5. γ)] m. the transformation from the rotated frame to the original frame.206) The expression (5. We will demonstrate that (5. the l. 5. φ) = S2 dΩ Y ∗m (θ.. (5.e. The axis x2 is deﬁned in the coordinate frame which is related to the original frame by rotation (i).203).207) which replaces J2 by the inverse transformation (i).203). For any similarity transformation involving operators A and S holds eS A S Accordingly.s. γ will be referred to as Euler angles. γ)] = . β. and the r.. the axis x3 is deﬁned in the coordinate frame which is related to the original frame by the consecutive rotations (i) and (ii). one can replace (5.212) . can then be written e− i βJ2 − i αJ3 e = e− i αJ3 e− i βJ2 (5. is e− i γJ3 = e− i βJ2 e− i αJ3 e− i γJ3 e i αJ3 e i βJ2 (5.h.. β.203) has the disadvantage that it employs rotations deﬁned with respect to three diﬀerent frames of reference. i. we can write e− i −1 = S eA S −1 .211) Using (5. (i) and (ii). i. Accordingly. i (5.m m. β. φ) [D(α. of (5. The Euler rotation replaces exp − i ϑ · J by e− i γJ3 − i βJ2 − i αJ3 e e .203) can be expressed. in analogy to (5.210) The third rotation in (5.194.195) by ˜ 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 (5.203) For any ϑ ∈ R one can ﬁnd Euler angles α. γ ∈ R such that exp i − ϑ·J = e− i γJ3 − i βJ2 − i αJ3 e e (5.209). however.e.204) is satisﬁed.h.124 Theory of Angular Momentum and Spin The angles α.209) The ﬁrst two rotations in (5.208) αJ3 βJ2 = e− αJ3 e− i βJ2 e i (5. the transformation from the original frame to the rotated frame. Obviously. φ) e− γJ3 − i βJ2 − i αJ3 e e . (5. followed by J2 in the original frame.

(5. φ) f (θ.210) yields the simple result e− i γJ3 − i βJ2 − i αJ3 e e = e− i αJ3 e− i βJ2 e− i γJ3 . (5. φ) e− αJ3 e− βJ2 Y m (θ. one can write [D(α. m [D(α. φ) = e−im α S2 dΩ Y ∗ (θ. 5. m i Y m (θ. φ) (5.310)] an explicit expression for the Wigner rotation matrix (5.214) e− γJ3 = S2 dΩ Y ∗m (θ. φ) . φ) e− m i βJ2 Y m (θ.309. β.213) i. the muon µ and its neutrino νµ of the second . β.203) with respect to the original (unprimed) frame one simply needs to reverse the order of the rotations. φ) e− S2 i γJ3 Y m (θ. φ) i i (5. φ) e−iγm δ (5. φ) . Best known is the spin of the electron. φ) Y m (θ. The eigenvalue property (5. 5.216) Using. 5.218) Deﬁning the so-called Wigner rotation matrix dmm (β) = one can express the rotation matrices [D(α. φ) e−imγ . m ( ) S2 dΩ Y ∗ (θ. but many other elementary components of matter are endowed with spin-like properties. β. m (5. (5.64) yields dΩ f (θ.217) Accordingly. φ) [D(α. γ)] = . φ) e− m S2 i αJ3 f (θ. φ) e− m i βJ2 Y m (θ.e. the self-adjointness of the operator J3 one can state dΩ Y ∗ (θ. β. in addition.220) We will derive below [see Eqs.206) by ˜ Y m (θ.m m.7 Spin 1 2 and the group SU(2) The spin describes a basic and fascinating property of matter. φ) = S2 dΩ f (θ.219) = e−iαm dm m (β) e−iγm δ ( ) . to redeﬁne the three rotations in (5. γ)] m. m = e−iαm S2 dΩ Y ∗ (θ. This allows one to express the rotational trasnformation (5.215) beneﬁts from the choice of Euler angles for the parametrization of rotational transformations. (5. γ)] m. (5.219).215) The evaluation of the matrix elements (5. Examples are the other ﬁve members of the lepton family to which the electron belongs.205. the electron e and the electron neutrino νe of the ﬁrst generation. γ)] m..5.7: Spin 1 2 and the group SU(2) 125 Multiplication from the left with (5.

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

At this point we will state a useful property. x. it might hold aj bk − bk aj = 0. one obtains (σ · a) (σ · b) = 3 j 2 j=1 (σ ) aj bj 3 j k j. σ2 = 0 −i i 0 .228) (5.5. ∈ R . y.227.231) . the simplest choice being S1 = 1 σ1 . 2 S3 = 1 σ3 2 (5.224) provide a basis in terms of which all traceless.229) and avoids commuting the components aj and bk .e. but their comonents must not necessarily commute with each other. The proof of this relation rests on the commutation relationship and the anti-commutation relationship (5.7: Spin 1 2 and the group SU(2) 127 exist three such linear independent matrices. they are associated with the important fermion property of matter. A = x σ 1 + y σ 2 + z σ3 . 1 i. b are vectors commuting with σ. y.k=1 σ σ 3 j k j. e. 2 S2 = 1 σ2 . According to this property the Pauli matrices generate a 3dimensional Cliﬀord algebra C3 ..226) The Pauli matrices satisfy very special commutation and anti-commutation relationships. it holds. (5. z A = In fact.k=1 j<k + σ j σ k aj bk .k=1 σ σ aj bk + j>k aj bk = 3 j. (5.223) where σk . 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.31) of the group SO(3).229) = 1 .227) which is essentially identical to the Lie algebra (5. i. 2.225) σ (5. (5. Cliﬀord algebras play an important role in the mathematical structure of physics.224) Algebraic Properties of the Pauli Matrices The Pauli matrices (5. 1 σj σk = − σk σj for j = k which can also be readily veriﬁed. hermitian 2 × 2–matrices A can be expandend. (σ · a) (σ · b) = a · b 1 + iσ · (a × b) 1 (5.230) where a. (5.g. σ3 = 1 0 0 −1 .e. One can readily verify the commutation property σj σk − σk σj = [σj . namely. In fact. σk ]+ = 2δjk 1 . σj 2 (5. 5. σk ]− = 2i jk z x − iy x + iy −z . k = 1. 3 are the Pauli matrices σ1 = 0 1 1 0 . z. Any such A can be expressed in terms of three real parameters x. The Pauli matrices obey the following anti-commutation properties σj σk + σk σj = [σj .

S ] = i k m Sm (5. In the special case a = b ∈ R holds (σ · a)2 = a 2 1 .234) 5.k=1 j>k 1 2 3 j.235) which is identical to that of the generators of SO(3). hence. For this purpose we choose to replace (5.236) The matrix elements of this 2 × 2 operator can be labelled by the basis states χ+ and χ− .222). however. The commutation property (5.h.e. S3 χ± = ± 1 χ± 2 . i.128 Theory of Angular Momentum and Spin Using (σ j )2 = 1 the ﬁrst term on the r. 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 . through the algebra obeyed by the operators Sk [Sk . spin changes sign.232) − σ k σ j ) (aj bk − ak bj ) . This relationship emerges. We like to derive this result now by evaluating the transformations of SU(2) explicitly. The two remaining terms yield using σ j σ k = −σ k σ j for j = k and altering ‘dummy’ summation indices 3 j. 1 1 1 ( + 1) χ± 2 2 .8 Generators and Rotation Matrices of SU(2) Sofar there is nothing which relates the transformations U . only a 720o rotation leaves the spin invariant. to rotations in space.222) as a rotation vector and you rotate once. (5. 1 (5. however. (5.k=1 j>k σ j σ k (aj bk − ak bj ) = 1 2 3 j. since there is a slight diﬀerence: when you interprete the 3-component ϑ in (5.k=1 (σ σ j>k σ k σ j (aj bk − ak bj ) = (5. (5.238) S 2 χ± = . Well almost like it.s. (5. 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. yields a · b. 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 (5.231) proves (5.227) leads to 3 3 jk j.222) by the Euler form exp(−iγS3 ) exp(−iβS2 ) exp(−iαS3 ) .237) + 1)1 and.233) This result together with (5.k=1 j>k σ j σ k (aj bk − ak bj ) − 1 2 3 j k j.230).k=1 j>k i σl (aj bk − ak bj ) = i =1 σ a×b . let say around the x2 –axis by 360o .

9 Constructing Spin States with Larger Quantum Numbers Through Spinor Operators In this section we like to demonstrate. .220) are [D(α. m = − . 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. 3 . i.244) i.. i.e. − + 1. γ)] m. 1 .239 ) one expands the exponential operator exp(−iβS2 ).219).242) The expressions in brackets [.5. ( − β S2 )2n = (−1)n ( − β S2 )2n+1 = (−1)n β 2 β 2 2n 1 1 0 1 −1 0 2n+1 (5.239) in the notation (5. . The complete matrix elements (5.240) .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). The idempotence property of Sk yields particularly simple expressions for these powers. following Jourdan and Schwinger. (5.241) 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. . no two such particles can exist in the same state. that states | m for higher quantum numbers = 0. rotation by 360o changes the sign of the spin state. For this purpose one needs to determine the powers of −iβS2 . namely.e.243) We note the property of this rotation matrix 2 dmm (2π) (1) = −1 1 . use the label 2 2 | 1 ± 1 for the states χ± . β. therefore. .] can be identiﬁed with the Taylor expansion of the cos. 2. . 1.245) 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. . .5. . (5. In order to determine the Wigner matrix in (5. (5. 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.e.239) where we made use of the notation (5. We may. 2 2 We obtain with this notation for the transformations (5.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 . identical particles any number of which can exist in the same state χ+ and χ− .219. . 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 .

. ζ = +. . i. + − For example. b± |Ψ0 = 0 (5. The boson character of the operators b† = (b† . 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. x∗ b = x∗ b+ + x∗ b− + − (5. b† )T and b = (b+ . b† |Ψ0 = χ+ + . b ζ = 0 b ζ.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 − . 1 .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± . 1.248) that bζ relationships † † b ζ. .247.253) = δζζ .246) The corresponding adjoint operators are denoted by b+ and b− . b− )T as spinor creation and annihilation + − operators.e. m) = b† − j−m (j + m)! (j − m)! |Ψ0 . b † ζ ζ =0 (5. using (5. −) bζ .248) bζ .252) (5. We consider. m) † and bζ obey the commutation (5. j = 0.254) We will show below that these states are orthonormal and form spin states with higher quantum numbers. m = −j. .249) In the following will refer to b† = (b† . b† )T and b = (b+ . −j + 1. .5. i.e. . These operators are associated with a given spatial reference system. b ζ = b ζ.247) (5.250) which for x∗ x = x∗ x+ + x∗ x− = 1 represent spinor operators in an arbitrary reference system. therefore. b− )T is expressed through the commutation relationships + − (ζ. (5. b † ζ For the vacuum state |Ψ0 holds = δζζ . b ζ = b† . (5.251) One can show using this property and (5. 2 + 2 − † (5. 3 . also operators of the type x b† = x+ b† + x− b† + − . . b† |Ψ0 = χ− − . j. 2 2 . b ζ The States |Ψ(j. b− † = − sin β † β b + cos b† .

+ + (5. An example + − is the monomial b† which generates the states (5. (5.256) In the special case |state = |Ψ0 this reads using (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+ .248) we obtain b+ . If it is true for n then using (5.255) is b+ . (5. f (b† ) |state + f (b† ) b+ |state . b† + n+1 = = b+ . 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. we need to evaluate b+ .257) motivate us to determine the commutator (5.5.256) and (5. We will derive the result. b† + n = n cn b+ . f (b† ) |Ψ0 . that the annihilation operators play a role similar to the diﬀerential operator in calculus.259) The property is obviously true for n = 0. b† + n b† + b† + b† + + n n n b+ .254).249) b+ f (b† ) |Ψ0 = + b+ .247. For this purpose we consider the following polynomial of creation operators N j+m j−m f (b† ) = + n=0 cn b† + n . f b† Obviously.260) . + b+ . b† + n . b † + = n b† + n−1 b† + b † + + n = (n + 1) b† + . possibly inﬁnite power series. b† + = n b† + n−1 . (5. 5.10: Algebraic Properties of Spinor Operators 131 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. b† + − well-known for the quantum mechanical harmonic oscillator.257) which for (5.258) Equations (5. . in b† and b† .249). f (b† ) + (5. We will show that for this commutator holds n b+ . b† + b+ .248) and from (5.

257) we can conclude in particular bζ f (b† . f (b† ) − = f (b† ) − (5.261) Similarly.262) Because of the commutation relationships (5. (5. By induction we can conclude that (5. + − (5. b† ) . 2 . + − (5.265) Generating Function of the States |Ψ(j. m) .263) According to (5.132 Theory of Angular Momentum and Spin i. f (x)] g(x) = (∂k f (x))g(x) or [∂k . 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. φjm (x) Ψ (j.266) we compare the terms x+ b† + j+m φjm (x) Ψ (j.267) exp xb† |Ψ0 is called a generating function of |Ψ(j. In order to derive (5. one can prove b− . b† ) |Ψ0 = + − ∂ ∂b† ζ f (b† .268) . m) m=−j (5. b† ) |Ψ0 . One can ﬁnally conclude for polynomials (5.. f (x)] = ∂k f (x) .264) This demonstrates the equivalence of the spinor operators bζ and the derivative operation.. b† ) in b† and b† holds (ζ = +. = f (b† ) + (5.1. f (b† .e. f (b† ) + where f (x) = df dx . the property holds also for n + 1.250) and where φjm (x) represents the function of the two variables x+ and x− closely related to |Ψ(j.. (5.266) where xb† has been deﬁned in (5. b† ) + − = ∂ ∂b† ζ f (b† . (5. −) + − + − bζ .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 .247) the more general property for polynomials f (b† .255) b+ . Equation (5.259) holds for all n ∈ N. m) φjm (x) = j−m xj+m x− + (j + m)! (j − m)! .

1.270) N 2j j → s=0 j−m=0 → m=−j . . chosing the summation index j = 0.275) = exp xb† |Ψ0 .1... . 2. (5.. s! (n − s)! Deﬁning j+m = N −s j−m = s. i.270) s x+ b† + N −s x− b† − s |Ψ0 . 1 xb† u! |Ψ0 (5... m=−j = j=0. 3 ..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. one obtains φjm (x) Ψ (j. 1 . 2. (2j)! 2j |Ψ0 (5. m) 1 j=0.1. 2 . m) m=−j = = 1 x+ b† + x− b† + − (2j)! 2j 1 xb† |Ψ0 . m) = = |Ψ0 (N − s)! s! N −s 1 N! x+ b† x− b† + − N ! (N − s)!s! N −s 133 (5.1.. 1.. 2 2 leads to ∞ j ∞ φjm (x) |Ψ (j. . ..e.5. (5..272) The summation over s can be written in terms of j and m using (5.274) Summing this expression over 2j = 0. .273) Change of the summation indices allows one to conclude j φjm (x) Ψ (j. 2 ∞ 1 xb† 2j! u 2j |Ψ0 = u=0.10: Algebraic Properties of Spinor Operators with the s–th term in the binomial expansion of (a + b)n n! an−s bs .. 1 . .269) (5.

we can conclude † † ∗ e(xb ) Ψ0 |e(yb ) Ψ0 = e(x y) (5. The generating function allows one to derive various properties of the states |Ψ (j. For this purpose we consider the inner product † † e(xb ) Ψ0 |e(yb ) Ψ0 = δjj δmm (5.134 Theory of Angular Momentum and Spin which concludes our derivation.m φjm (x∗ ) φj m (y) Ψ (j.j . m) and will be used for this purpose below. m holds. [b† . (5. b− ] = 0. b† ] = 0 and [b− .e. of (5.264) implies † † bζ e(yb ) |Ψ0 = yζ e(yb ) |Ψ0 .280) The commutation properties [b+ . therefore.277) To evaluate this expression we ﬁrst notice † e(xb ) † = e(x ∗ b) . According to (5. b− )e(yb ) |Ψ0 = f (y+ . Orthonormality of the States |Ψ (j.281) and.h. m) We want to show now that the states (5. i.279) One can generalize this to † † s (bζ )s e(yb ) |Ψ0 = yζ e(yb ) |Ψ0 .282) where we deﬁned x∗ y = x∗ y+ + x∗ y− .s.254) are orthonormal. hence. that Ψ (j.277) by Ψ0 |e(x b) e(yb ) Ψ0 .276) = j. m) |Ψ j . (5. (5. b† ] = 0 allow one to state that for + − + any polynomial f (b+ . In order to evaluate the ∗ b) (yb† ) operator e(x e we notice that the derivative property (5. m) |Ψ j . b− ) holds † † f (b+ . 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. (5.249) for any non-vanishing integer s holds + − Ψ0 |bs Ψ0 = 0 and.284) .278) † ∗ which allows us to replace the l. y− )e(yb ) |Ψ0 (5. m .m.283) (5.

m 135 = e(x ∗ y) .224). b† bm n n n.289) < n |σi | m >< n |σj | m > {b† bm . 1.m +b† b† . . (5. etc.. Jj ] = i This property is derived as follows [Ji . . m) |Ψ j .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.m.224).277) φjm (x∗ ) φj m (y) Ψ (j.282) and (5.285) Following steps similar to those in Eqs.m ijk Jk .288) The three operators obey the Lie algebra of SU(2) [Ji .275) one can show e(x ∗ y) = j. bm bm } n n = 1 4 < n |σi | m >< n |σj | m > {b† bm δmn − b† bm δm n } n n n. In our present notation the matrix elements < ζ |σk | ζ > for ζ = ± correspond to the matrix elements < ζ |σk | ζ > for σ = ± 1 . < + |σ1 | − > = < − |σ1 | + > = 1. Jj ] = = 1 4 1 4 < n |σi | m >< n |σj | m > b† bm . J1 = J2 = J3 = 1 † b b− + b† b+ − 2 + 1 † b b− − b† b+ − 2i + 1 † b b+ − b† b− . using < + |σ1 | + > = < − |σ1 | − > = 0. b† bm n n n. . (5. 5. States 2 We want to demonstrate now that the states |Ψ (j.n . 2. (5. m j.m.ζ where σk .m. 1 New Representation of Spin 0. − 2 + (5.n .m φjm (x∗ ) φjm (y) . 3 . 2 .n . (5.j .10: Algebraic Properties of Spinor Operators and comparing (5. 3 denote the Pauli spin matrices (5.286) from which follows immediately the orthonormality property (5. (5.287) ζ 2 ζ. The 2 operators are explicitly. k = 1.276).5.m. m) as given in (5.223.m .268–5. One deﬁnes corresponding operators Jk through 1 Jk = b† < ζ |σk | ζ > bζ .

k ijk ijk σk | m > b † bm n = i = i 1 2 < n |σk | 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 . (5. J1 + iJ2 ] = −i [J2 . σj ]| m > b† bm n n. m) are eigenstates of J 2 and J3 .m < n |2i n. (5.m ijk Jk . − (5.290) |Ψ (j. (5.295) The last result together with (5.296) ( 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).292) The commutation relationships (5.136 = 1 4 − = = 1 4 1 4 Theory of Angular Momentum and Spin < n |σi | m >< m |σj | m > b† bm n n. J1 ] + i [J1 .m. m) as Eigenstates of J 2 and J3 We wish to show now that the states |Ψ (j.m.291) Here we have deﬁned J+ J− = J1 + iJ2 = J1 − iJ2 = b† b− + = b† b+ .297) . To this end we note 2 2 2 J 2 = J1 + J2 + J3 = 1 1 2 J+ J− + J− J+ + J3 2 2 .m. 2 J 2 = J+ J− + J3 − J3 = J3 (J3 − 1) + J+ J− (5. J+ ] = [J1 − iJ2 .294) .m < n |[σi .293) .m 1 4 < n |σj | m >< m |σi | m > b† bm n n .289) yield [J− . (5. J2 ] = −2 J3 from which we can conclude the property 1 1 J− J+ = J+ J− − J3 2 2 Hence. (5. (5.

m = λ||λ. m . . m) J3 |Ψ (j. . m + 1) (j + m) (j − m + 1)|Ψ (j. n1 = 0. b† . j = 1.e.301) Obviously.304) (5. and eigenstates of k with eigenvalues j. 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. One can then conclude J 2 |Ψ (j. the number of states to the possible energy eigenvalues.303) Exercise 5. m) One can furthermore derive readily J+ |Ψ (j. m = m||λ. . b+ ] = δjk . .h. State the corresponding eigenvalues and the degree of degeneracy. m I1 = where λ = 0. n2 . 1 2. n = n1 + n2 ﬁxed. . λ. I3 ||λ. . . m) m |Ψ (j. i.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) .10: Algebraic Properties of Spinor Operators The last term on the r. I3 = (b+ b1 + b+ b2 ) 2 2 2 2 2i 2 1 satisfy the Lie algebra of SU(2). Ik ] = i jk I . 1 2 k (a) Show that the eigenstates are given by 1 |n1 . m) J− |Ψ (j. 2.5. i.. 2 2 137 (5. using operators I± = I1 ± iI2 2 2 2 and the subspace {|n1 . 1.10.298) (5. .299) (5. ˆ respectively. −λ + 1.302) (5. Show λ = n1 +n2 n1 +n2 2 ( 2 + 1) and m = 1 2 (n1 − n2 ).305) = = j ( j + 1 ) |Ψ (j. (b) Show that the three operators 1 + 1 1 + (b1 b2 + b+ b1 ) . . . b † + b + + b † + b+ b † − b − 1 1 b† + b + + b † + b + b † − b− + b † − b − b † + b + .254) are eigenstates of + b+ and − b− with eigenvalues j + m and j − m. . the states (5. .s. n} eigenstates of I 2 = I1 + I2 + I3 and of I3 I 2 ||λ.e. I2 = (b1 b2 − b+ b1 ) . m) = = (j + m + 1) (j − m) |Ψ (j.300) (5. 2 . Construct. m = −λ. 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† . [bj . m − 1) . . (5. [Ij . (5. m) .

On the other side the states |Ψ (j. m) .307) Comparision of (5. The prefactor of (b† )j+m (b† )j−m which according to (5.138 Theory of Angular Momentum and Spin 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).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 (5.310) it reduces to (5. + − One may expect that these two conditions restrict both σ and σ.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 × (5.306) where b † and b † are given by (5.309) .251).306) and (5. (j) (5. The combination of σ.308) The latter sum involves terms (b† )j+m (b† )j−m for σ + σ = j + m and 2j − σ − σ = j − m. In case j = 1 this expression yields.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† . For this purpose we note that the states |Ψ(jm) in a rotated coordinate system according m to (5. m ) are related to the + − states |Ψ (j. However. 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. both conditions are satisﬁed for σ = j + m − σ.306. m = b† + j+m b† − j−m (j + m )! (j − m )! |Ψ0 (5. for example.254) are |Ψ j. m) in the original coordinate system by j |Ψ j. σ values which yields (b† )j+m (b† )j−m is + − then σ = j + m − σ. .243). m = m=−j dm m (β) |Ψ (j.

of (5. = U x2 .5. x3 ) ∈ R .251).313) × 0 0 2 = 0 sinβ 0 cosβ 1 −i 0 which. namely that the component of the vector on the l.s.e.309) and the particular matrix (5.12: Mapping of SU(2) onto SO(3) 139 5. therefore. The SU(2) transformation entered in (5. This factor implies.314) leaves (5. in case of j = 1.g. conclude that the mapping from SU(2) onto SO(3) is a 2–1 mapping. is the expected element of SO(3). in fact. i.e. that the the representation (5. Hence. x2 .h.310). b+ † = −cos β † β b+ − sin b† 2 2 − . the orthogonal 3 × 3 matrix which describes a rotation around the x2 –axis. the matrix A should represent a rotation around the x2 –axis in the space of vectors ˜ (x1 .. 2 2 − (5.310) to the SU(2) transformation assumed in deriving the general result (5.251) and (5. Replacing the latter transformation by its negative form.308) unaltered except for a factor (−1)2j which multiplies then also the ﬁnal result (5.12 Mapping of SU(2) onto SO(3) The representation (5. however.308) and had the form (5. 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 (5.314) in case of integer j–values. This can be shown by (1) applying to the matrix A = (dm m ) in (5.309) does not distinguish between SU(2) transformations (5.312) transforms like an angular momentum state ˜ |1 m . e.309).311) (5. . i. b− † = sin β † β b+ − cos b† . It is of interest to trace the mapping from SU(2) onto SO(3) as represented by (5.312) The choice of this transformation derives from a property shown further below.310) establishes a mapping of SU(2) onto SO(3).. One can. U = 0 1 1 i √ ( x1 − ix2 ) √ x3 − √2 0 2 2 (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 .

140 Theory of Angular Momentum and Spin .

γ are needed to specify a general rotational transformation 1 141 .. each carrying orbital angular momentum decribed by spherical harmonics Y m (θ. rB .g. 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 . e. Further below we will consider composite systems involving spin states. The overall orientation of any three particle conﬁguration can be speciﬁed by three parameters1 . How should they be chosen? Actually there is no unique choice. and (rAB . φ) as eigenfunctions and/or spin. and particle C impinges on the compound AB coming from a large distance. 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 . β. 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. Often the socalled total angular momentum. give rise to good quantum numbers. Since the Hamiltonian does not depend on the position of the center of mass R = (mA rA + mB rB + mC rC )/(mA +mB +mC ). since related operators. B. The rotational part of the scattering motion is described then in terms of the unit vectors rAB and ˆ We remind the reader that. 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. and to express the Hamiltonian in terms of |rAB |. on the distances between the three particles. hence.. classically speaking the sum of all angular momenta and spins of the composite system. rC of the particles.g. C governed by a Hamiltonian H which depends only on the internal coordinates of the system. ρC ). This eleminates three further parameters from the dependency of the Hamiltonian on the three particle conﬁguration and one is left with three parameters. e. Example: Three Particle Scattering Consider the scattering of three particles A. |ρC |. is the quantity of interest. for example. sums of orbital angular momentum and of spin operators of the particles. three Eulerian angles α. by a rotational vector ϑ. commute with the Hamiltonian of the composite system and. six parameters must suﬃce to describe the interaction of the system. We like to illustrate this for an example involving particle motion.Chapter 6 Quantum Mechanical Addition of Angular Momenta and Spin In this section we consider composite systems made up of several particles.

. each of which stands for two angles. in particular. composite systems behaving like elementary objects are common. − 2 ≤ m1 ≤ (6. rather than by Classical Mechanics.. . n = 1. 2 .max + 1)2 ( 2. mA mB + mB mC + mA mC 2.max . . 1. . = B such. ρ 1 = 0. the values of which are determined by the size of the interaction domain ∆V . albeit puzzling. vibrations and rearrangement of the particles in reactions like AB + C → A + BC. (6.max = 10 one obtains d(B) = 14 641.max = ∆ 2 mA mB mC E . mC . 2. Obviously. (6. Such large number of dynamically coupled states would constitute a serious problem in any detailed description of the scattering process.max 2.max and 2.max + 1)2 (6. − 1 ≤ m1 ≤ 1. . .1) where 1. . as we will demonstrate below. i.3) For rather moderate values 1. 1. .max = 1 2 IA−B E . The latter often arrives at the physical properties of composite systems by adding the . however. mB . by the total energy E. that only states within the subspaces Bk are coupled in the scattering process. have not even be considered. . . the dimensions d(Bk ) of these subspaces does not exceed 100. to separate the 14 641 dimensional space of rotational states Y 1 m1 (ˆAB ) Y 2 m2 (ˆC ) into r ρ subspaces Bk .max . . 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 ). 2} 1. and the following mathematical description will shed light on their ubiquitous appearence in physics. a very large number.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. by the masses mA .max d(B) = 1 =0 (2 1 + 1) 2 =0 (2 2 + 1) = ( 1.m1 .142 Addition of Angular Momenta and Spin ρC . in fact. and by the moment of inertia IA−B of the diatomic molecule A–B approximately as follows 1. . The rotational symmetry of the interaction between the particles allows one.m2 Y 1 m1 (ˆAB ) Y 2 m2 (ˆC ). 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. Such extremely useful transformation of the problem can be achieved through the choice of a new basis set B = { 1 .m2 c (n) r ρ 1. . = 0. since further important degrees of freedom.2) The dimension d(B) of B is 1. will make their appearence a natural consequence of the symmetry of the building blocks of matter. 2.max = 2. In fact.max denote the largest orbital and rotational angular momentum values. B1 ⊕ B2 ⊕ . 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. 14 641} .m1 2 .

ρC ) deﬁned through R(ϑ) ψ(rAB . 2. 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. Jq(2) = 0 for p. in fact. the reader will learn in the following section how to add and subtract in the microscopic world of Quantum Physics. ρC ) = ψ(R−1 (ϑ) rAB .9) hold since the components of J (k) are diﬀerential operators with respect to diﬀerent variables.5) do not aﬀect the Hamiltonian. To specify this property mathematically let us denote by H the Hamiltonian in the rotated frame. i. 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).. ∂ρx ∂ρy (6.0: Addition of Angular Momenta and Spin 143 corresponding physical properties of the elementary components. ∂yAB ∂zAB i (2) ∂ ∂ − J3 = ρy − ρx . the commutation relationships (1) Jp .6) For the following it is essential to note that H is not invariant under rotations of only rAB or ρC . The invariance of the Hamiltonian under overall rotations of the three particle system implies then H = R(ϑ) H R−1 (ϑ) .8) Obviously. For ˆ ˆ example. (6.53. the existence of a basis (6. presumably a facility the reader would like to acquire with great eagerness. Rotational Symmetry of the Hamiltonian As pointed out already. Following our description of rotations of single particle wave functions we express (6. be diﬀerent from H.48) i i R(ϑ) = exp − ϑ · J (1) exp − ϑ · J (2) (6. It holds then H R(ϑ) ψ = R(ϑ) H ψ. rotations R(ϑ) of the wave functions ψ(rAB . 5. q = 1. ρC ) it follows H R(ϑ) = R(ϑ) H.10) .4) which decouples rotational states is connected with the rotational symmetry of the Hamiltonian of the three particle system considered.7) i R(ϑ) = exp − ϑ · J (6. according to (5. Since this is true for any ψ(rAB . Hence. assuming presently that H might.e. connected with the fact that the Hamiltonian H does not depend on the overall orientation of the three interacting particles. but solely under simultaneous and identical rotations of rAB or ρC . One can equivalently express therefore (6. 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. 3 (6.55) holds i (1) − J1 = zAB ∂ ∂ − yAB . R−1 (ϑ) ρC ) (6. In this sense.5) according to (5.6.7) where the generators J (k) are diﬀerential operators acting on rAB (k = 1) and on ρC (k = 2).

J ] = i k m Jm for n = 1. ˆ (6.12) We consider this equation for inﬁnitesimal rotations.17) . Before we apply the procedure (5. ˆ Ω 2 = ρC . of course.. For a proof one uses (6. the important fact that the Jk obey the Lie algebra of SO(3). i. k = 1. One can show readily that the commutation relationships [Jk . [H.e. 3 . 3 do not commute. i.144 where J = J (1) + J (2) Addition of Angular Momenta and Spin .f. B] + [A. k = 1. 3 as the three components of the total angular momentum operator. To order O(|ϑ|) one obtains i ϑ · JH .71–5. that the components Jk . the (n) (n) (n) properties [Jk . 2. 2 together with the property [A. The property (6. (6.81) we want to consider the relationship between YJM and Y 1 m1 (ˆAB ) Y 2 m2 (ˆC ). 2.. the operators Jk . 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 ). Eqs. In fact. r ρ Deﬁnition of Total Angular Momentum States The commutation property (6. we will ﬁnd that the states yield the basis B with the desired property of a maximal uncoupling of rotational states.6) for rotational invariance of the Hamiltonian in the form i i H = exp − ϑ · J H exp + ϑ · J . that energy. i. however. and the eigenvalues of J2 and J3 are good quantum numbers.14) We will refer in the following to Jk .16) following the procedure stated in the theorem above [c.11) By means of (6. for |ϑ| H ≈ 1 − 1 i ϑ·J H 1 + 1 i ϑ·J ≈ H + i 1.14) implies that the components of the total angular momentum operator (6. a supposition which can be tested through the commutation properties of these operators.14) implies that the total angular momentum is conserved during the scattering process.9). In the following we will use the notation r ρ Ω1 = rAB . (5.e. (6. (6.15) are satisﬁed. We suspect. componentwise. 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. 2.81)]. 3 cannot have simultaneous eigenstates among each other. B + C] = [A. (6. k = 1.e. k = 1.71–5.12) each individually can have simultaneous eigenstates with the Hamiltonian. We recognize. J ] = i k m Jm (6. 2.11) we can write the condition (6.13) Hϑ · J − Since this holds for any ϑ it follows HJ − J H = 0 or. Jk ] = 0 . C].

assume the expansion YJM ( 1 . restrict the sum in (6. the reader should always keep in mind that conditions equivalent to (6.18) yields using the property (k) J3 Y k mk (Ωk ) = mk Y k mk (Ωk ) . (6. These coeﬃcients. .e. J (2) . i. Ω2 ) = m1 (J M | 1 m1 2 M − m1 ) Y 1 m1 (Ω1 ) Y 2 m2 (Ω2 ) .m2 (J M | 1 m1 2 m2 ) Y 1 m1 (Ω1 ) Y 2 m2 (Ω2 ) (6. hence. 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.21) One can. 2 |Ω1 .22) We will not adopt such explicit summation since it leads to cumbersum notation.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) . an important example being the symmetry groups SU(N) governing elementary particles made up of two quarks. m1 . (6. of (6.1: Clebsch-Gordan Coeﬃcients 145 6. J (1) . J (2) .18) to avoid summation of vanishing terms YJM ( 1 .e.1: Show that J2 ... We ﬁrst notice that YJM in (6. Since YJM sofar speciﬁes solely two quantum numbers. i. therefore. J3 YJM = M YJM .20) This equation can be satisﬁed only if the Clebsch-Gordan coeﬃcients satisfy (J M | 1 m1 2 m2 ) = 0 for m1 + m2 = M . (6. k = 1. ρC ) are normalized. Equivalent coeﬃcients exist for other symmetry properties of multi–component systems.h. 2 |Ω1 . two further quantum numbers need to be speciﬁed for a complete characterization of the total angular momentum states. baryons. 2 M YJM ( 1 .21) hold.6. Exercise 6. play a cardinal role in the mathematical description of microscopic physical systems.19) Noting J3 = J3 + J3 and applying this to the l. The two missing quantum 2 2 numbers are 1 and 2 corresponding to the eigenvalues of J (1) and J (2) . or the closely related Wigner 3j-coeﬃcients introduced further below. i. the eigenvalue being speciﬁed by the quantum number M .s. (6. mesons. Ω2 ) = m1 .18) is an eigenfunction of J3 . The expansion coeﬃcients (J M | 1 m1 2 m2 ) ˆ are called the Clebsch-Gordan coeﬃcients which we seek to determine now. and three quarks. However. 2 |Ω1 .18) where the states YJM ( 1 .e. and J3 . and J (1) · J (2) commute. Ω2 ) (1) (2) = 1 m1 2 m2 ) Y 1 m1 (Ω1 ) Y 2 m2 (Ω2 ) (m1 + m2 ) (J M | .m2 2 |Ω1 . J3 .1. J3 . We.

m2 = − 2 . . Ω2 ) = δJJ δM M (6. . In fact. 2 ) and B ( 1 . . . .e. (6. . m1 = − 1 . 2) = {Y 1 m1 (Ω1 ) Y 2 2 m2 (Ω2 ). J = | 1 − 2 |. to a new basis B ( 1 . .18) constitutes a change of an orthonormal basis B( 1 . (6. .h.s.. the range of values assumed for J is correct. . . . it holds 1+ 2 ( 2J + 1 ) = (2 J=| 1 − 2 | 1 + 1) (2 2 + 1) .27) together with (6. Our derivation below will also yield real values for the Clebsch-Gordan coeﬃcients. it is not immediately obvious what the J– values are. J. −J + 1.146 Theory of Angular Momentum and Spin The expansion (6. (6. . . . 2 |Ω1 .25) We want to state now two summation conditions which follow from the orthonormality of the two basis sets B( 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.28) This property follows from the fact that the basis elements are eigenstates of hermitian operators with diﬀerent eigenvalues. 1.24) The basis B( 1 . 1 + 2 . .m2 2 M and with (6. The property dΩ1 dΩ2 Y∗ ( 1 .h. | 1 − 2 | + 1. (6. the basis B ( 1 .26) We will show below in an explicit construction of the Clebsch-Gordan coeﬃcients that. This obviously corresponds quantum mechanically to a range of J–values J = | 1 − 2 |.24) yields 1 m1 2 m2 ) ∗ 1 m1 2 m2 ) (J M| = δJJ δM M . The basis B ( 1 . Ω2 ) YJ M ( 1. property implies dΩ1 dΩ2 Y 1 m1 corresponding to the l. B2 2) should be 2 |Ω1 . 2 ). and that the states can be normalized. M = −J. − + 1.23) corresponding to the r.18) applied to Y∗ and to YJ JM (J M | m1 . . .. .1. 2) . However.2: Prove Eq. −J + 1. − 2} 1 + 1. respectively. . In this case the range of |J|–values would be the interval [ |J (1) | − |J (2) | .s. Exercise 6. JM 2 |Ω1 . J} . in fact. . |J (1) | + |J (2) |]. 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 . 1 + 2. Ω2 ). 2 ) is also orthonormal2 and must have the same number of elements. = {YJM ( 1 . For each quantum number J there should be 2J + 1 elements YJM with M = −J. | 1 − 2| + 1. 2 ) has (2 1 + 1)(2 2 + 1) elements.25) i. (6. (6. J (1) and J (2) .

(6. Ω2 ) and integrating.81) stated above. Ω2 ) . One can show readily using the same reasoning as applied in the derivation of (6.2: Construction of Clebsch-Gordan Coeﬃcients 147 The second summation condition starts from the fact that the basis sets B( 1 .e.e.6. Hence. The result of this construction will include all the properties previewed above. At this point we like to stress that the construction will be based on the theorems (5. 2 |Ω1 . 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. (6. rather the convention (J M | has been assumed.33) JM The latter summation has not been restricted explicitly to allowed J–values. i.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 . 2 |Ω1 .32) which complements (6.g.27) yields δJJ δM M = m1 . .18).30) The latter property follows from multiplying (6. to composite systems involving spin. also apply the results.. 2 |Ω1 .1 states...28) allows one to conclude that the coeﬃcients cJ (J M | 1 m1 2 m2 )∗ . Ω2 ) . i.71–5. (6. it is possible to expand Y 1 m1 (Ω1 )Y 2 m2 (Ω2 ) in terms of YJM ( 1 . and actually also the properties of Clebsch-Gordan coeﬃcients stated above. 2 |Ω1 . or J > 1 + 2 (6. Ω2 ). or baryons multiplets involving three quarks. e.2 Construction of Clebsch-Gordan Coeﬃcients We will now construct the Clebsch-Gordan coeﬃcients. (6. will be based solely on the commutation properties of the operators J and J (k) . Ω2 )Y 1 m1 (Ω1 )Y 2 m2 (Ω2 ) . 2 ) span the same function space. We can. 2 |Ω1 . the group SU(3) in case of meson multiplets involving two quarks. 1 m1 2 m2 ) = 0 if J < | 1 − 2| .31) are identical to Comparision with (6.. A similar 2 construction will also be applied to composite systems governed by other symmetry groups.e. (6. i. 2 ) and B ( 1 . 1+ 2 J Y 1 m1 (Ω1 )Y 2 m2 (Ω2 ) = J =| 1 − 2 | M =−J cJ M YJ M ( 1.28) from (6.18) by Y∗ J orthogonality property (6.34) 6. therefore. The (J M | 1 m1 2 m2 ) cJ M .29) where the expansion coeﬃcients are given by the respective scalar products in function space cJ M = dΩ1 dΩ2 Y∗ J M ( 1.

therefore. Ω2 ) = Y 1 1 (Ω1 ) Y 2 2 (Ω2 ) . 1+ 2 ( 1. Similarly. J − 2.38) for M = J − 1.68) J+ = (1) + J+ (1) (2) Y 1 1 (Ω1 ) Y 2 2 2 2 (Ω2 ) 1 1 (6. . (6. 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 ) 2 |Ω1 . we can also demonstrate using (??) that G dΩ1 = We.35) 1 − 2 |. . Ω2 ) . 1 + 2} and for (6. 2 |Ω1 .39) which we will try for a solution of (6. .25) using (6. (6. Ω2 ). Ω2 ) (Ω1 ) dΩ2 Y 2 2 (Ω2 ) = 1.36) = 0 = J YJJ ( 1 . Quantum mechanically this corresponds to a state G 1+ 2. 1+ 2 ( 1. .18) which satisﬁes J+ YJJ ( 1 . 2 |Ω1 .66. 1+ 2. 2 |Ω1 . 2 |Ω1 . For this purpose we insert (6. We ﬁrst seek an unnormalized solution GJJ and later normalize. J3 YJJ ( 1 . Ω2 ) . 2 |Ω1 . have shown Y 1+ 2. We embark on the suggested construction for the choice J = 1 + 2 .11) J+ by J+ + J+ . The construction assumes a particular choice of J ∈ {| such J–value seeks an expansion (6. . . 2 |Ω1 .37) and replace (1) (2) according to (6.64) 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. 2 |Ω1 .37). Ω2 ) Addition of Angular Momenta and Spin (6.39) into (6.37) The solution needs to be normalized. 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. we can demonstrate condition (6. . . M = −J. Ω2 ) = (J + M + 1)(J − M )YJM ( 1 .42) ( 1. (6.148 For the construction of YJM we will need the operators J± = J1 + iJ2 . −J. 2 |Ω1 . We obtain using (5. Ω2 ) is normalized dΩ2 G dΩ1 Y 1 1 1+ 2. Having determined such YJJ we then construct the whole family of functions XJ = {YJM ( 1 . (6. J} by applying repeatedly J− YJM +1 ( 1 . −J + 1. 2 |Ω1 .11) and (5. . Ω2 ) = Y 1 1 (Ω1 ) Y 2 2 (Ω2 ) (6.41) 2) Y (Ω1 ) Y (Ω2 ) . 1+ 2 ( 1. 1+ 2 In fact.40) (Ω1 ) J+ Y (2) 2 2 J+ Y 1 1 (Ω1 ) Y (Ω2 ) + Y (Ω2 ) = 0 .5.43) . | 1 − 2| + 1.

h. According to the condition (6. We like to construct now the family of total angular momentum functions B + − = {Y + − M ( .2: Construction of Clebsch-Gordan Coeﬃcients 149 We now employ property (6. M = −( + − ). . The reader should familiarize himself with the entries of the Table. Having constructed this family we have determined the Clebsch-Gordan coeﬃcients ( 1 + 2 − 1M | 1 m1 2 m2 ). Ω2) for M = 1 + 2 − 2 and J = 1 + 2 .38) yields pression 2( 1 + 2 )Y 1 + 2 1 + 2 −1 ( 1 .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. 2 = 1 in Table 1.1: Construct following the procedure above the three functions YJM ( 1 .37).2. . ). To demonstrate (6. . The l. G 1 + 2 −1 1 + 2 −1 in (6. The r.45) for some unknown constant c. . M = −( + ). 1 + 2 − 1. 1 + 2 − 2. ( + )}. 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. 2 |Ω1 . . One obtains then Y 1 1+ 2 1 + 2 1 + 2 −1 2 2 ( 1.s. . 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.38) to construct the family of functions B + = {Y + M ( . 2 |Ω1 . in particular. We demonstrate the procedure ex(1) (1) plicitly only for M = 1 + 2 − 1.27). of (6.37) is satisﬁed. and all total angular momentum functions for a given choice of 1 and 2 . . −( 1 + 2 − 1). ( + − )}. One can obviously continue the construction outlined to determine Y 1 + 2 −2 1 + 2 −2 .h. . Exercise 6.47) can serve now to obtain recursively the elements of the family B + − for M = 1 + 2 − 2.36. 6.s. The result is illustrated for the case 1 = 2. ).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. Ω2 ) 2 1+ 2 = (Ω1 )Y 2 2 −1 (6. . etc. This expression is orthogonal to (6. with the symmetry pattern and with the terms Y 1 m1 Y 2 m2 contributing to each YJ M . We seek for this purpose ﬁrst an unnormalized solution G 1 + 2 −1 1 + 2 −1 of (6. of (6.45). Normalization yields furthermore Y = 1 1 + 2 −1 1 + 2 −1 We have thereby determined ( 1 2 |Ω1 Ω2 ) (6.39) as required by (6. | .46) 2 1 +c Y (Ω1 ) Y 2 2 (Ω2 ) = 0 . . 2 |Ω1 . . Expression (6. To satisfy this equation one needs to choose c = − 1 / 2 .44) (Ω2 ) . .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 ). Ω2 ). . | . One can readily show that (6. Show that the resulting functions are orthonormal. all the Clebsch-Gordan coeﬃcients ( 1 + 2 M | 1 m1 2 m2 ). 1 + 2 − 3.6.47) (Ω2 ) − 1 1 2 + Y 2 1 1 −1 (Ω1 )Y 2 2 + Y 2 1 1 (Ω1 )Y 2 2 −1 (Ω2 ) . thereby.

632456 0. Ω2) Y1−1 (2. 1|Ω1 . Ω2) Y30 (2. Ω2) Y10 (2.774597 0 2 5 8 15 1 6 3 10 1 3 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 0.816497 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. Ω2) Y3−1 (2.258199 −0.816497 −0. 1|Ω1 .316228 0. Ω2) Y20 (2. 1|Ω1 .774597 − − − − 0. 1|Ω1 .408248 −0.707107 0.730297 −0. 1|Ω1 .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 ) 0.632456 0. .547723 0.150 Addition of Angular Momenta and Spin Y33 (2. Ω2) Y31 (2. Ω2) Y11 (2.57735 0.57735 0.730297 0.447214 −0.1: 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. Ω2) Y2−1 (2.57735 3 10 3 5 Y2−1 (Ω1 )Y11 (Ω2 ) Y20 (Ω1 )Y10 (Ω2 ) Y21 (Ω1 )Y1−1 (Ω2 ) −0.547723 0.316228 0. Ω2) Y3−3 (2. 1|Ω1 . Ω2) Y21 (2. 1|Ω1 .816497 0.57735 −0.547723 0. 1|Ω1 .57735 0.447214 0. 1|Ω1 .707107 0. Ω2) Y3−2 (2.774597 0. 1|Ω1 .57735 0. Ω2) Y32 (2.707107 0. 1|Ω1 . 1|Ω1 .547723 0. Ω2) Y22 (2. 1|Ω1 .632456 −0.707107 0.408248 −0. 1|Ω1 . 1|Ω1 .258199 0. Ω2) Y2−2 (2.

m − 2 . m + 1 .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 ). These expressions will be derived now. (6. Deﬁnition of Spinor Operators for Two Particles In contrast to Sections 5. commute with each other ζ ζ aζ . 2 . Accordingly. 5. aζ . b † ζ = δζζ = δζζ . bζ = a† .10 [cf. J.2. we extent deﬁnition (5. b† ζ ζ = aζ .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 .ζ a† < ζ |σk | ζ > aζ ζ b† < ζ |σk | ζ > bζ ζ (6. a† and bζ . (5.54) 2 + = √ b† ( 2 +m2 )! √− ( 2 −m2 )! .49) where ζ. M | .ζ ζ.3: Explicit Expressions of CG–Coeﬃcients 151 Exercise 6. − 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† ζ ζ = 0 = 0 .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. M | . b† refer to diﬀerent particles and. (6. hence. ζ = ± and the matrix elements < ζ |σk | ζ > are as deﬁned in Section 5. It is actually possible to state explicit expressions for any single coeﬃcient (JM | 1 m1 2 m2 ). + 2 − J−M +1 for = J + 1 2J+2 2 J−M for = J − 1 2J 2 = .10 where we studied single particle angular momentum and spin.10. 6.6.53) (6. b† ζ = 0. The creation and annihilation operators are again of the boson type with commutation properties aζ .9. we are dealing now with two particles carrying angular momentum or spin.48) (6.10. a† ζ . 1 .51) The operators aζ .50) (6. bζ . For this purpsose we will employ the spinor operators introduced in Sections 5. a† ζ ζ b† .9.52) and | 2 m2 2 According to Section 5. 5. bζ = = a† . 1 (6.287) to two particles (1) J k k = = (2) J 1 2 1 2 ζ. aζ bζ .

m2 1 2 ( 1 . 2) . M ( 1 . + 1) |J..68) The states |J. as usual are denoted by their respective quantum numbers J. M ( 1 . ˆ kj |J. (6. hence.65) (6. 1.152 It holds. 2. (6.69) − 2| ≤ J ≤ + 2 . for such states should hold J2 |J. 2 . m1 . m1 1 | 2 . in analogy to Eqs. M ( 1 . M ( 1 . m1 1 | 2 . 5. k = 1. . −J ≤ M ≤ J . 2) + 1) |J. M. (1) J (2) J 3 | 1 m1 1 3 | 2 m2 2 = m1 | 1 m1 = m2 | 2 m2 1 2 (6.m2 | 1 .62) . mj j = j | j .61) (6.59) J± = J1 ± i J2 .53. (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 . (1) J 2 . M ( 1 . (6. The aim of the present Section is to determine closed expressions for the Clebsch–Gordan coeﬃcents ( 1 . m2 |J.54). 1 .64) (6. m2 2 . mj j and.60) (6. J3 .300). = M |J. k2 = b b+ + b† b− k1 = + − − 2 2 + namely. m2 a† + ( 1 2 = a† − ( 1 1 −m1 (6. M ( (2) J 2 |J. M ( 1 .67) = j |J. can be expressed in terms of the number operators 1 † 1 ˆ ˆ a† a+ + a† a− . (1) J 2 |J. |J. ˆ For the operators kj holds ˆ kj | j .303). m1 1 | 2 .18) as follows = | 1 m1 . 2) 2) (6.e. which describe a two particle system according to (6.302. 2 )). 2 . M ( 1 . 2 )) . J3 |J. 2 . j = 1.57) b† + ( (2) 2 2 +m2 1 +m1 b† − ( 2 2 −m2 + m1 )! − m1 )! (1) + m2 )! − m2 )! |Ψo . M ( 1 . 2) 1. M ( 1 . M ( 1 . (2) J which. 6. (6. 2) We will also require below the raising and lowering operators associated with the total angular momentum operator (6. m2 |J. m1 .58) Jk + (2) Jk . are | 1 . = = 1( 1 2( 2 2) 2) (6. 2) 2) can be expressed in terms of Clebsch-Gordan coeﬃcients (6. (6. M ( 2) 2) = J(J + 1) |J. we like to recall for future reference that the operators (5.55) . M ( 1 .63) according to At this point. . 2 ) which are simultaneous eigenstates of the operators 2 2 J2 . i. M ( 1 . (5.59) We seek to determine states |J. M ( 1 . 3 .56) The states | 1 . (j) 2 ˆ ˆ J = kj ( kj + 1 ) . 2 . (1) J . (2) J 2 . The operator of the total angular momentum/spin of the two particle system is J = with Cartesian components Jk = (1) J + J (6.66) (6.

+ − + − 2 2 − + = .3: Explicit Expressions of CG–Coeﬃcients The Operator K † The following operator K † = a† b† − a† b† . a† b† + a† a− . K † + K † 2 2 3 † †ˆ K kj + K . ˆ A similar calculation yields [k2 . K † ] = (j) J 2 . We note that. a† b† − a† a− . a† b† − a† b† + − + − + − − + 2 1 1 = a† a+ .71) one can show K† = = = = = 1 ˆ ˆ kj (kj + 1).288).73) can be readily proven. 2 2 3 ˆ = K † kj + K † .71) (6. 6. using (6. and applying the relationships (6. b† = 0 .75) The relationships (6. a† b† + − − + 2 + 2 − 1 1 − b† a† b+ . 6. a† b† − a† b† − + − − + 2 + 1 1 = a† a+ .71–6. Employing (6. j = 1. (6. K† J3 .74) imply 3 2 J2 .73) (6. K † = 0.72) (6.70) will play a crucial role in the evaluation of the Clebsch-Gordan-Coeﬃcients. 2 4 = 0 = = 0. expressing (k) J 3 through the creation and annihilation operators according to (5. K † J± . (6. due to J2 = K† 1 2 J+ J− 1 † K .65) and (6.73) yields J3 . a† b† + + − − + 2 2 − 1 1 = a† b† − a† b† = K† . K † 1 † a a+ + a† a− .50. b† − b† a† b− . K † 2 1ˆ ˆ 1 ˆ ˆ kj kj + 1. + − − + 2 2 = 1 † 2K .6. For example.71–6. 4 Using J3 = (1) J 3 + (2) J 3 .74) + 1 J− J+ + J2 .73. K † (kj + 1) 2 2 1ˆ † 1 ˆ kj K + K † (kj + 1) 2 2 1 ˆ 1 ˆ K † kj + kj . j = 1.64.51) one obtains ˆ k1 . K † + kj . K † 1 a† a+ − a† a− + b† b+ − b† b− . 6. + − − + 153 (6. the relationships (6. K † (j) J 2 . This operator obeys the following commutation relationships with the other pertinent angular momentum / spin operators ˆ kj .

K† + K † (j) J 2 |J. a† b− − a† b† + + + − + = 0. 2 2 1 + ) .76) and state K† n |J. 2 ) .78) a state which has been used already in the construction of Clebsch-Gordan coeﬃcients in Section 6. K † = Addition of Angular Momentum and Spin a† a− + b† b− .292) one can derive similarly J+ . 2) We want to demonstrate now that the action of K † on the states |J. Action of K † on the states |J. One can generalize property (6. j2 ) .75) ascertain that under the action of K † the states |J. on the state |j1 + j2 .154 Starting from (5. namely. The commutation properties (6. M ( 1 + . |j1 + j2 . M ( 1 . To demonstrate that the resulting states are eigenstates of (1) J 2 and (2) J 2 we exploit (6. = − a† a− .77) where N is another normalization constant. 2) is a state with quantum numbers 1 1 + 1 2 and = N |J. according to (6. 2 2 + n ) 2 (6. 2 ) produces again total angular momentum eigenstates to the same J and M quantum numbers of J2 and J3 . j2 + ) 2 2 n n = N (n. j1 + j2 (j1 . 2) = N |J..63.79) n |j1 + j2 .76) Here N is an unknown normalization constant.77). K † ] = 0 is demonstrated in an analoguous way. 2 (6. M ( 1 . We consider now the case that (K † )n acts on the simplest total angular momentum / spin state. 1 2 + 2 . M ( 1 . j1 + ) K † 2 2 (6. M ( 1 . j + 2 2 2) However. j2 ) = |j1 . 6. j2 2 . 6.2. 4 3 + j ( j + 1) |J. Application of (K † )n to this state yields. but for diﬀerent 1 and 2 quantum numbers of the operators (1) J 2 and (2) J 2 . M ( 1 . 2 ) j + 4 1 3 + K † |J. M ( 1 . M ( 1.67) (j) J 2 K † |J. 6.72) and (6. j1 + j2 (j1 .73.e. i.62. j1 1 |j2 . M ( 1 . b† + − + + + − The property [J− . 2) 2) ˆ K † kj + = K† = j 3 † K + K † j ( j + 1) |J. a† b† + b† a† b− . M ( 1 . (6. M ( 1 + n . M ( 1 . 2 ) remain eigenstates of J2 and J3 with the same quantum numbers. j1 1 + . M ( 1 . j1 + j2 (j1 + n n . it holds K † |J. this result implies that K † |J. 2) = = (j) 2 J .

2) = N( 1 + 2 − J. holds |J. J( 1 . (6.80) = = = 1 1 2 2 − J 1 n − = (J + 2 2 n 1 − = (J + 2 2 + 1 1 − − 2) 2) . M ( 1 . (J + 2 − − 2) 1 1) 2 The normalization constant appearing here is actually N( + − J.79). of (6. M ( 1 . J( 1 . 2) 2) .81) = N( 1 + 2 1 2 − J.86) b† + J+ 2 − 1 |Ψo . 2 ) ∆(J. For this purpose one needs to choose in (6. 2) = ∆(J. 2 = j2 + n 2 (6.57) and exploiting the fact that J− and K † commute [c. (6. j2 + n ).3: Explicit Expressions of CG–Coeﬃcients 155 where we denoted the associated normalization constant by N (n. 1. allow one to obtain the states |J. M ( 1 .82) for |J. M ) (J + × K† 1 − 2 )! (J + 2 − 1 )! J+ 1 − 1 + 2 −J (J− )J−M a† + × 2 (6. = (J + M )! (2J)! (J − M )! 1 2 2) (6. 1 . 2) 1 2 = ( 1 + 2 (2J + 1)! − J)! ( 1 + 2 + J + 1)! 1 2 .84) with (6. M ( 1 . j1 .82) 1 × | (J + 2 1 × | (J + 2 1 (J + 2 1 − 1 ) . J( 1 . 1 = j1 + n 2 . M ) (J− )J−M |J.104. 2 ) .6.s.105) of the spherical harmonics. in analogy to the construction (5. M ( 1 . Strategy for Generating the States |J. − 1.h.83) The derivation of this expression will be provided further below (see page 158 ﬀ). J( 1 .79) n. j2 as follows J = j1 + j2 which is equivalent to n j1 j2 Accodingly. 2 ) for −J ≤ M ≤ J as follows |J. 2 ) can be expressed through the r.f. 2 2 It is now important to notice that any state of the type |J. (6.74)] yields |J. 2) . The latter states. j1 + n . 2) Our construction of the states |J. M ) . 5.85) ∆(J. 2 ) exploits the expression (6. (6.82.84) (6. K† 1 2 1 + 2 −J × × . 6. Combining (6.

69) yields then the Clebsch- Expansion of an Intermediate State We ﬁrst consider the expansion of the following factor appearing in (6. in terms of | 1 . x. (6.91) which.88).57)].. follows from the commutation properties (6. b† ] = 0 [cf.e.87).88) in terms of monomials (6. x.50–6. y) = λr xs y t r |Gst . m2 Gordan-coeﬃcients. x.93) = f (a† + λ a† ) g(b† + λ b† ) |Ψo .87) + − + − i. st The desired expansion of |Gr can be obtained from an alternate evaluation of I(λ.e. (6. m1 1 | 2 . in analogy to (5. For this purpose we introduce the generating function I(λ. One obtains then using J− = a† a+ + b† b+ (6..92) − − and noting [a+ . y) is a generating function for the states |Gr deﬁned in (6.86) in terms of monomials 1 +m1 1 −m1 2 +m2 2 −m2 a† a† b† b† |Ψo . I(λ.t (6.264). Comparision with (6. 2 [cf.52).156 Addition of Angular Momentum and Spin Our strategy for the evaluation of the Clebsch-Gordan-coeﬃcients is to expand (6. + − + − .89) i. y) = exp (λJ− ) exp x a† exp x b† |Ψo .90) (6.50)] − − 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. a† ] = [b+ . + + Taylor expansion of the two exponential operators yields immediately I(λ.s. r!s!t! r. (6.86) |Gr = Jr a† st − + s b† + t |Ψo (6. x. 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.

95) Comparision with (6.t × a† + s−q s q b† + t r−q t−r+q a† − q b† − |Ψo (6.97) + 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. 2) = s.s. x.3: Explicit Expressions of CG–Coeﬃcients We conclude I(λ.94) One can infer from this result the desired expressions for |Gr . using the commutation property (6. one can write the right factor in (6.94) yields I(λ.97) yields. M ( 1 .6. Expanding the exponentials in st (6. to obtain the desired expansion of |J.88).90) allows one to infer |Gr s.70) holds K† 1 + 2 −J = s 1 + 2 −J b† − s 1 + 2 −J−s (−1)s 1 + 2 −J−s (6.t × λ v b† − × r−q v × = t s−u v λu a† − r! q |Ψo λr xs y t r! s! t! r. using the deﬁnition (6.98) a† − s a† + b† + s . m2 2 .q (−1)s (6. x.t = q r! × a† + s q s−q t r−q a† − q × b† + t−r+q b† − r−q |Ψo (6.96) and. 2 ) in terms of states | 1 .99) .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.50).97). M ( 1 .86) (J− )J−M a† + = q J+ 1 − 2 b† + J+ 2 − 1 |Ψo 1 (6. With K † given by (6. y) = exp a† + λ a† + − exp b† + λ b† + − |Ψo . y) = s. 157 (6. |J. m1 1 | 2 . Operation of this operator on (6.

q (−1)s × (6.100) ( 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 . 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. M ) (J + 2 )! (J + 2 − 1 )! s. m1 . n) 2 )|J. 2 ) ∆(J.83) of the normalization constant N ( 1 + deﬁned through (6. 1 . M ( 1 . (6. n)|ψ(j1 . j2 . − q − s)!(q + s)! 2 × (M + 1 − 1 + q + s)!( 1 1 + 2 − M − q − s)! 2 − q − s 1 | 2. 2 ) we consider the scalar product J. M − +q+s One can conclude that this expression reproduces (6.101) Note that m1 + m2 = M holds.102) We want to determine now the expression (6. 2 ) − 2) .82) and (6.103) this can be written 1 = N 2 ψ(j1 .54) between creation operator monomials and angular momentum states allow one to write this |J. m2 = M − 1 + q + s. 1 .6.82). − 1 . Using (6. m2 in (6. j2 . j2 = 1 (J + 2 2 − 1) . 2 ) (6. J( 1 .104) .101). m2 |J. according to (6.103) = 1. × 2) = N( 1 + 2 1 − J.69) since. To determine N = N ( 1 + 2 −J. √ 2J + 1 1 2 . J( 1 . The Clebsch-Gordan coeﬃcents are then ﬁnally ( 1 . n = 1 + 2 −J.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 (6.69) if one identiﬁes m1 = 1 −q − s . The summation over q corresponds then to the summation over m1 . (6. For this purpose we introduce j1 = 1 (J + 2 1 2 − J. q = 1 − m1 − s and m2 = M − m1 .53.

j2 . We employ the expression (6. n) in terms of states |j1 .3: Explicit Expressions of CG–Coeﬃcients where |ψ(j1 .57) for these states and the expression (6. One obtains then. 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 .70) for the operator K † .104) 1 = N2 (n!)2 (2j1 )!(2j2 )! (2j1 + n − s)!(2j2 + s)! s!(n − s)! 2j2 + s 2j2 (6.109) and Taylor expansion of the left hand side of (6.108) twice. j1 1 |j2 . j2 2 . m2 2 .105) The ﬁrst step of our calculation is the expansion of ψ(j1 . applying (6. (6. (6. n) = K† n 159 |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. (6.108) = λ=0 (n1 + ν)! n1 ! (6. j2 .m2 n1 + m1 n1 n2 + m2 n2 λm1 +m2 (6. 1 1−λ n1 +1 1 1−λ n2 +1 = m1 .110) which can be written 1 1−λ n1 +n2 +2 = r s n1 + r − s n1 n2 + s n2 λr (6.108) yields the identity n1 + r − s n1 n2 + s n2 = n1 + n2 + r + 1 n1 + n2 + 1 .111) Comparision with (6. we obtain |ψ(j1 .108).106) The orthonormality of the states occurring in the last expression allows one to write (6.6. m1 1 |j2 . j2 . Accordingly.112) s .

m1 ) by ( 2 . For the quantum mechanical addition of angular momenta the Clebsch Gordan coeﬃcients ( 1 .117) are again related in a simple 2 . m2 . 2 . −M ) ..114) one obtains − J = − J (1) − J (2) The respective Clebsch-Gordan coeﬃcients ( 1 . 2 .114) show a simple relationship to the Clebsch Gordan coeﬃcients ( 2 .g. 6. −m2 . −M ) = (−1) 1 + 2 −J ( 1 . M | 1 .114) by.160 Applying this to (6. m2 . M ) = (−1) 1 + 2 −J ( 2 . one can interchange also the operator J on the l. m1 . m1 .120) The symmetry properties (6. 2 . (6. (6.h.114) For example. 2 . m1 |J. 1 . of (6.115). M ) corresponding to (6.102) by replacing ( 1 . (2j1 + 2j2 + 1)! (6. J (1) on the r. (6. J.11) J = J (1) + J (2) .102) of the Clebsch-Gordan coeﬃcients. m2 |J. manner to the coeﬃcients ( 1 . ( 2 . (6. and J (2) are vectors in R . m2 |J. −m1 . M ) = (−1) 2 +m2 2J + 1 ( 2 .120) can be readily derived from the expression (6. −m2 |J.116) through formula (6. of this equation J (1) = J (2) − J . To derive relationship (6. M ) . m2 ) and. 2 1+1 (6. 1 . e. m1 . m1 |J. m1 .103) one obtains the desired result (6.116) one expresses the Clebsch-Gordan coeﬃcient on the r. m1 ) . of (6. (6.119) The corresponding symmetry property of the Clebsch-Gordan coeﬃcients is ( 1 .116) If one takes the negatives of the operators in (6. We will demonstrate this now. M ) corresponding to (6. m2 ) by ( 1 .s. m1 ). (6. M ) .118) Finally. m1 .115) This relationship is a trivial consequence of (6.s.4 Symmetries of the Clebsch-Gordan Coeﬃcients The Clebsch-Gordan coeﬃcients obey symmetry properties which reﬂect geometrical aspects of the operator relationship (6. interchanging the operators J (1) and J (2) results in J = J (2) + J (1) . m2 |J. −m1 .113) Using the identities (6.s.118).h. and . −m1 2|J. 2 .114) as long as J.83).h.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 . (6. m2 |J. and (6. 2 . m2 |J. namely. M ) ( 1 . J (1) .116). vice versa. ( 1 .

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

1 |ˆ) at θ = 0 are r r 2 2 2 2 Yj 1 (j − 1 . For this purpose we consider the application of σ · r in the case of θ = 0. φ) = − Yj 1 (j + 1 . (6. 1 |ˆ) + f (r) σ · p Yjm (j ± 1 . φ) 2 2 2 .147.187) 1 2 . (6.148) the particular ˆ 1 1 states Yj 1 (j − 2 . One can conclude then ˆ σ·r = ± ˆ 0 1 1 0 (6. According to (5. 1 |ˆ) r r r 2 2 2 2 2 2 (6.177) one notes that for φ = 0 the spin-angular momentum states 2 are entirely real such that σ · r must be real as well.181. φ) 2 2 2 = − j+ 1 2 4π (6.f. 1 |θ = 0.184) 0 1 space χ 1±1 2 2 one can conclude from (6. 1 |θ = 0. therefore. 1 |ˆ). 1 |θ.185) We have.183) in case θ = 0 becomes in the standard representation with respect to the spin.174–5. 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. φ) and (5. The result can also be stated in the compact form σ · r Yjm (j ± 1 . identiﬁed the sign of (6.1 states χ 1 ± 1 2 2 2 [c. φ) .182) σ · r Yj 1 (j − 1 .182) 0 Since σ · r. We want to demonstrate ﬁnally that the “−”-sign holds in (6. (5. 1 |ˆ) = ((σ · p f (r)))Yjm (j ± 1 . 1 |ˆ) r 2 (6.160).180) and.180) and (6.180) where the sign could depend on j. for θ = 0 (6.186) Here (( · · · )) denotes again conﬁnement of the diﬀusion operator ∂r to within the double bracket. ˆ (6. hence. 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 . ˆ 2 2 2 2 2 2 for θ = 0.188) . have proven (6. Since f (r) is independent of θ and φ follows ˆ ((σ · p f (r))) = −i (( ∂r f (r) )) σ · r . ˆ (6.181) 0 = j+ 1 2 4π Yj 1 (j + 1 . 1 |θ = 0.224)] 1 0 σ·r = ˆ . 6.180). 6. given by ˆ σ · r = σ1 sin θ cos φ + σ2 sin θ sin φ + σ3 cos θ .168 Addition of Angular Momentum and Spin 1 Yjm (j ± 2 . 2 |ˆ) and Yj 1 (j + 1 . 1 |θ = 0. We want to determine ˆ its action on the wave functions f (r) Yjm (j ± 1 .

1 |ˆ) r 2 1 2 (6.. Corresponding results are obtained for the other components of p × r and. x3 x2 ˆ ˆ ( p × r )1 h = −i (∂2 − ∂3 ) h (6.187) one obtains ˆ σ · p f (r) Yjm (j ± 1 .191) (6. using ∂r Yjm (j ± 1 . 1 |ˆ) = i [ ∂r + r + ] f (r) Yjm (j 2 r r Using (6. 1 |ˆ) r 2 = i ∂r + −j r 1 g(r) Yjm (j + 2 .193) r r 1 1 1 = −i (∂2 x3 − ∂3 x2 ) h − i h (x3 ∂2 − x2 ∂3 ) .155. r 2 2 1 J ·S 2 1 ˆ σ · p f (r) Yjm (j ± 2 . ∂3 (1/r) = −x3 /r3 and ∂2 x3 = x3 ∂2 .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. 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. r r r Using ∂2 (1/r) = −x2 /r3 . 1 |ˆ) .189) The celebrated property of the Pauli matrices (5. 1 |ˆ) r 2 (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. To demonstrate this we consider one of its cartision components.190) can be related to the angular momentum operator. we conclude the intuitively expected identity 1 ˆ ˆ p×r = − J . e.197) 1 ˆ σ · p g(r) Yjm (j − 2 .194) ˆ ˆ where J1 is deﬁned in (5. 1 |ˆ) r 2 (6.g.156) this yields ﬁnally 1 ˆ σ · p f (r) Yjm (j + 2 .192) ˆ ˆ The operator p × r in (6. 6. 1 |ˆ) = r 2 2 ˆ i ∂r f (r) − f (r) σ · p σ · r Yjm (j ˆ 1 2 169 . 1 |ˆ) r 2 (6.5: Spin–Orbital Angular Momentum States Using (6. hence. r Altogether we obtain. r 2 (6.196) = i j+ ∂r + r 1 2 3 2 1 f (r) Yjm (j − 2 .195) .198) .230) allows one to express ˆ ˆ ˆ ˆ ˆ σ · p σ · r = p · r + i σ · (p × r) .53).6. 1 |ˆ) = 0 and σ = 2S/ . h r (6.186) for both terms in (6.

r ∂r2 r 2 (6.203) (6. are consistent with this identity.198). 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 .204) (6. j) 1 2 1 2 1 2 1 2 = = 1 0 (6.199) We want to show that eqs.101).197. (6. 6.170 Addition of Angular Momentum and Spin To demonstrate the validity of this key result we note that according to (5.201) 1 3 (j + )(j + ) Yjm (j + 1 . 6. r (6. 1 |ˆ) r r 2 2 2 2 2 2 2 yields ˆ (σ · p)2 f (r) Yjm (j + 1 . 1 |j.43). 2 2 (6. We note ˆ (σ · p)2 f (r) Yjm (j + 1 . 1 |ˆ) = J 2 Yjm (j + 1 . using (5.144) for m = j. 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 .143–6.199). according to (6.144) and. j 2 + 2j + 2 3 4 (6. 5. in fact. For m = j − 1 one can state.99) holds ˆ (σ · p)2 = − 2 2 = ∂2 J2 r + 2 . according to (6.202) Yjj (j − 1 .. j) 2 2 2 2 (j − . j + . Yjj−1 (j − 1 . 6. 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 (6.2.2.205) which agrees with the expressions (6.151). For this purpose we use the construction method introduced in Sec. .206) . consider ﬁrst the case of the largest m-value m = j. as well as (6. 1 |ˆ) = Yj− 1 j− 1 (ˆ) χ 1 1 . 1 .45).143. − |j.230. We begin with the coeﬃcients (6.e. r 2 2 2 2 2 2 − r 2 ∂r r + J2 r2 f (r) Yjm (j + 1 . 1 |ˆ) r 2 2 i 3 ˆ = σ · p [ i∂r + (j + ) ] f (r) Yjm (j − 1 . j − 1 .143. In this case holds. adopting the method in Sec. i. Evaluation of Relevant Clebsch-Gordan Coeﬃcients We want to determine now the Clebsch-Gordan coeﬃcients (6. 1 |ˆ) r 2 2 r r2 and. 1 |ˆ) r 2 2 r r2 r2 j 2 + 2j + 3 2 2 2 4 = [ −∂r − ∂r + ] f (r) Yjm (j + 1 .146). 1 |ˆ) r 2 2 The Clebsch-Gordan coeﬃcients are then (j − 1 .200) = (j + 1 )(j + 3 ). 6. 6. 1 |ˆ) = r 2 2 which agrees with (6.

6. The further Clebsch-Gordan coeﬃcients (· · · |jj − 2). j − 3 . and J− + S− to the r.f.h. j − 1 .137)] J− (tot) = J− + S− (6.144) for m = j − 1. are obtained by iterating the application of (6.2. 6.143.143.s.144) for the Clebsch-Gordan coeﬃcients by induction.h. 1 . as described in Sec..207) (6. 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 . 6.s.202).209) (j − 1 . 1 .f. j − 1) 2 2 2 171 = = 2j − 1 2j 1 2j (6. 6. (6. 6. − 1 |j.5: Spin–Orbital Angular Momentum States The corresponding Clebsch-Gordan coeﬃcients are then 1 (j − 1 . m − 3 . (6. 1 |j. 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 . 2 2 2 2 2j . etc.208) (6.210)] yields (j + m)(j − m + 1) Yjm−1 (j − 1 . Expression (6. [c. (6.212) 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 . 2 2 2 2 2j (6.210). 2 |j.206). m − 1) 2 2 2 2 = j +m−1 2j (6. Let us verify then the expression (6. is obtained by applying the operator [c. j − 1) 2 2 2 2 which again agrees with the expressions (6. 1 .144) implies for j = m Yjm (j − 1 .210) to (6.211) to the l. (· · · |jj − 3).143. 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 .

Jclass and Jclass play equivalent roles.213) which is in agreement with (6.144). 2j + 1 (6.214) 2 2 +1 ( 3 .218) where we have replaced the quantum numbers J. proven (6. M.146) can be obtained from (6. 1 . m + 1 ) = 2 2 2 2 and.215) which follows from (6. the relationship (6. m. m2 1 (6. m3 .120). 1 . all three vectors Jclass . m1 ) (6.143. m2 | 3 . − 1 |j. m − 1 . 6. m) = 2 2 2 2 Similarly.217) 2j + 2 (j + 1 . 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 .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 . 6. − 1 |j.143).145) from (6. m3 . one can obtain expression (6. (1) (2) (−tot) Obviously.143. m + 1 ) = 2 2 2 j+m+1 . m1 . 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 . In this context Jclass and Jclass play the same role. However.172 Addition of Angular Momentum and Spin j −m+1 2j (j − 1 . 1 . − 1 |j.116. hence.143) by applying the symmetry relationships (6. j2 . m − 1) 2 2 2 2 = (6. one obtains (j + 1 .214) yields (j. m. m2 .216) 6. m) 2 2 2 2 2j + 1 (6. m1 . The Clebsch-Gordan coeﬃcients (6. 6. using (6. 2 . j2 m2 ) 2 . 2 . We have.215). m + 1 .144) by induction. The latter relationships applied together read ( 1 . 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. m1 . 1 . 1 |j + 1 . 1 . j+m+1 . by the set j1 . to reﬂect in the notation the symmetry of these quantities. 2j + 2 (6. m + 1 .144) for j = m − 1. 1 |j + 1 . −m2 | 1 . m3 ) = (−1) × 2 + 3 − 1 + 2 +m2 × 2 . 1+1 3 For 1 (j. j3 . (1) (2) (−tot) The coeﬃcients which are the quantum mechanical equivalent to Jclass + Jclass + Jclass = 0 are the 3j–coeﬃcients introduced by Wigner.

173 (6. one can exchange columns and one can reﬂect at the diagonal (transposition). for which no known classical analogue exists.220) The latter condition |j1 − j2 | ≤ j3 ≤ j1 + j2 . −j1 + j2 + j3 . The reader may note that the entries of the Regge-symbol. The Regge symbol also vanishes in case that two rows or columns are identical and Σ is an odd integer. the so-called triangle condition.102) are obtained each through the diﬀerence of two entries of the Regge-symbol.e.21. In this way the values of 12 3j-coeﬃcients are related. m3 . with respect to alltogether 12 symmetry operations. = (6. as follows −j1 + j2 + j3 j1 − j2 + j3 j1 + j2 − j3 j1 j2 j3 j1 − m1 j2 − m2 j3 − m3 . For this purpose one can use the symmetry transformations to place the smallest element into the upper left corner of the Regge symbol.6. In case of a non-cyclic exchange of rows and columns the 3j–coeﬃcent assumes a prefactor (−1)Σ . states that j1 . 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. j3 form the sides of a triangle and the condition is symmetric in the three quantum numbers. These symmetry operations relate altogether 72 3j–coeﬃcients. are identical to the integer arguments which enter the analytical expression (6. m1 . j2 . anti-cyclic exchange and of a sign reversal are stated. To represent the full symmetry one expresses the 3j–coeﬃcients through a 3 × 3–matrix.102) of the Clebsch-Gordan coeﬃcients.6: The 3j–Coeﬃcients We ﬁrst like to point out that conditions (6. there exist even further symmetry properties. i. The two integer entries J − 1 − m2 and J − 2 + m1 in (6. √ safe for the prefactor 2j3 + 1 which is cancelled according to the deﬁnition (6. According to the deﬁnition of the 3j–coeﬃcients one would expect symmetry properties with respect to exchange of j1 . Because of its high degree of symmetry the Regge symbol is very suited for numerical evaluations of the 3j–coeﬃcents.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 . discovered by Regge.218) relating 3jcoeﬃcients and Clebsch-Gordan coeﬃcients.219) (6. m2 and j3 .223) (Σ + 1)!n11 !n22 !n33 !n23 !n32 !)) n=0 n31 n32 n33 . However. The Regge symbol reﬂects a remarkable degree of symmetry of the related 3j–coeﬃcients: One can exchange rows.g..221) j1 j2 j3 −m1 −m2 −m3 where the the results of a cyclic.222) m1 m2 m3 j1 + m1 j2 + m2 j3 + m3 The Regge symbol vanishes. except when all elements are non-negative integers and each row and column has the same integer value Σ = j1 + j2 + j3 . 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. 6. m3 and with respect to sign reversals of all three values m1 . j2 . m2 . the Regge-symbol. e.

1. m1 + m2 ) (2 − m1 )! (2 + m1 )! √ 10 (1 − m)! (1 + m)! (1 − m2 )! (1 + m2 )! 1 ≥ |m| ∧ 1 ≥ |m2 | ∧ 2 ≥ |m1 | (6.226) We like to state ﬁnally a few explicit analytical expressions for Clebsch-Gordan coeﬃcients which were actually obtained using (6.227) (2m|2m1 1m2 ) = (−1)m+m1 (m + 2m2 ) δ(m.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. (6.174 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) (6.10752786393409395427444450130056540562826159542886 We also illustrate the numerical values of a sequence of 3j-coeﬃcients in Figure 6.223-6. m2 ) √ 2 1 ≥ |m1 |) 2 (6. m1 + m2 ) (2 − m1 )! (2 + m1 )! √ 6 (2 − m)! (2 + m)! (1 − m2 )! (1 + m2 )! 1 ≥ |m2 | ∧ 2 ≥ |m| ∧ 2 ≥ |m1 | (6.226) by means of a symbolic manipulation package (Mathematica) (1m|2m1 1m2 ) = (−1)1+m+m1 δ(m. 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. (6.232) . m) δ(−m1 .224) (6. m1 + m2 ) (3 − m)! (3 + m)! (3m|2m1 1m2 ) = √ 105 (2 − m1 )! (2 + m1 )! (1 − m2 )! (1 + m2 )! 1 ≥ |m2 |) ∧ 2 ≥ |m1 | ∧ 3 ≥ |m| (6.228) √ (−1)2m1 −2m2 7 δ(m.225) sn−1 .230) 1 1 (1m| m1 m2 ) = 2 2 √ (−1)2m1 −2m2 3 δ(m.229) 1 1 (0m| m1 m2 ) = 2 2 i(−1)1−m2 δ(0.

m2 = − 2 . Irreducible Representation We had stated above that the spherical harmonics Y m (Ω)). .. the rotations in single particle function space.e. i. . then for a basis {Y 1 m1 (Ω1 )Y 2 m2 (Ω2 ). . .e. Ω2 ) provide the irreducible representation for the rotation R(ϑ) deﬁned in (6. i.5).6. eigenfunctions of the single particle angular momentum operators J 2 and J3 . m1 = − 1 . provide the irreducible representation for D(ϑ). 2. .+ 1 . . . 1 . 2 = 1. the 2–particle total angular momentum wave functions YJM ( 1 . .233) . . . .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. rotations in 2–particle function space. . Similarly. 2 |Ω1 . (2 (2 1 2 + 1) (2 + 1) × (2 1 2 + 1) + 1) . If we deﬁne the matrix representation of R(ϑ) by D(ϑ).+ 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. ..1: Oscillatory behavior of 3j-coeﬃcients.

6. Exercise 6. 1).6. (6. .6. . M = −J. . J = | 1 − 2 |.1: Prove Eqs. Ω2 ).234) .3: Two triplet states |1m1 (1) |1m2 (2) are coupled to an overall singlet state Y00 (1.. 2 |Ω1 . 1 . . . . 2. each of the blocks in (6. Exercise 6. Show that the probability of detecting a triplet substate |1m1 (1) for arbitrary polarization (m2 – 1 value) of the other triplet is 3 . .176 Addition of Angular Momentum and Spin 1 For the basis {YJM ( 1 . 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. .6. .234) Exercise 6. .233) is further block-diagonalized as follows + 2. 2 = 1. .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( 1 + 2 ) + 1) ×(2( 1 + 2 ) + 1) Partitioning in smaller blocks is not possible.233. .

(6.237) are called tensor operators of rank k. The operator T may also act on multi-particle states like Y 1 m1 (ˆ1 )Y 2 m2 (ˆ2 ). −k + 1. ∂x2 3 1 2 2 2 2 case (ii) the operator T could be a spin operator Sk deﬁned in (5. Note that in the examples mentioned the operator T would be deﬁned within the same representation as R(ϑ).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.237) The operators T ∈ {Tkq .6. q = −k. 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. This implies. The latter can be written T |ψ where T denotes T in the rotated frame given by T = R(ϑ) T R−1 (ϑ) .236) In this equation Dq q (ϑ) denotes the rotation matrix Dq q (ϑ) = kq |R(ϑ)|kq . e. Rotations transform |ψ as |ψ = R(ϑ)|ψ and T |ψ as R(ϑ)T |ψ . −k + 1.7: Tensor Operators 177 6. k} such that k Tkq = q =−k (k) Dq q (ϑ) Tkq . Examples of Tensor Operators The multiplicative operators C∞ ( ) → C∞ ( ) Y ( ) = def Y ( ) (6.236. e. . i. In fact.g. that in case (i) T is an operator C∞ ( ) → C∞ ( ) acting on single particle wave functions. 6. some of the r r examples considered below involve tensor operators T of this type.238) .e. k} with the transformation property (6. for example.223). Such operators T can be characterized through their behaviour under rotational transformations.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 . . (k) (6.222) in case (ii) of single particle spin operators. e. The latter implies that T transforms like an angular momentum or spin state | m . q = −k. the operator (5. . that T belongs to a family of operators {Tkq . . (k) (6. S 2 = S1 + S2 + S3 or any other polynomial of Sk . 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). . .g.g.42) in case (i) of the position representation of single particle wave functions or the operator (5.

x2 . This choice of parametrization allows us to derive conditions which are equivalent to the property (6.6. Exercise 6.7.240) 1 where we have deﬁned ϑ± = 2 (ϑ1 iϑ2 ) and L± = L1 ± iL2 . For the following it is important to note that the rotation matrix elements (6.237).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 0.239) These operators can be expressed in terms of the coordinates x1 . In fact.178 are tensor operators of rank k. . Exercise 6.235 .237) do not require that the rotation R−1 (ϑ) is expressed in terms of Euler angles according to (5. 2. but are far easier to ascertain for any particular operator. 1. 3 follows from the transformation properties of the spherical harmonics derived in Section 1. The fact that these operators form tensor operators of rank 1.3.2: 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. 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. x3 . we will assume presently that the rotation is chosen as follows R(ϑ) = exp ( ϑ+ L+ + ϑ− L− + ϑ3 L3 ) (6. 2.203). but rather any rotation and any parametrization can be assumed.7.

Exercise 6.237) for transformations characterized by inﬁnitesimal values of ϑ+ . Tkq ] = = δqq and by subtracting Tkq on both sides of the k ϑ+ kq |L+ |kq Tkq . J3 yields [J+ . 0)T in case of small ϑ+ . Tkq ] = = = Tkq+1 Tkq−1 q Tkq .7. 0)T .237) yields ﬁrst k R(ϑ) Tkq R −1 (ϑ) = q =−k kq |R(ϑ)|kq Tkq .7. q =−k q+1 (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 . Tkq ] [J− .6.245) (6. (6. (6.244) [L+ . We ﬁrst consider a rotation with ϑ = (ϑ+ . (0. hence.248) a tensor operator? Exercise 6.5: Consider a system of two spin.237) be assumed then. ϑ3 .235 . 0. Exercise 6. 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) . Tkq ] = −iTk q+1 (k + q + 1)(k − q) .6.6. (6. Tkq ] [J3 . ϑ3 )T .7. ϑ− .80) follows kq |L+ |kq = −i (k + q + 1)(k − q)δq and. Similar equations can be derived for inﬁnitesimal rotations of the form ϑ = (0.247) 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) . 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) .246.243) From (5. (k + q + 1)(k − q) (k + q)(k − q + 1) (6. ϑ− . 0.4: Is the 1-particle Hamiltonian 2 H = − 2 2m + V (|r|) (6.242) from which follows by means of kq |1 1|kq equation ϑ+ [L+ .7.7: Tensor Operators 179 The conditions can be derived if we consider the property (6. 2 Exercise 6.3: Derive Eqs.247).6: Form tensor operators of rank 1 in terms of the three components of acting on the space of 1-particle wave functions.246) (6. Expressing the results in terms of the angular momentum operators J+ . 1 (6.235 . The property (6.235 . .6. 6. J− .

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 .251) The corresponding property for Φ 1 m1 (k.253) Exercise 6.m2 ( 1 m1 |kq m1 q. Before we proceed we like to point out that the states | 2 m2 γ2 are also eigenstates of J 2 .e. 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. 2 |γ2 ) = Cδ 1 1 δm1 m1 2 1( 1 (6. one can show that Φ 1 m1 (k. | 1 m1 γ1 denotes an angular momentum (spin) state.m2 ( 1 m1 |kq 2 m2 ) (J3 Tkq − Tkq J3 + Tkq J3 ) | 2 m2 γ2 = qTkq = q.m2 2 m2 ) (q + m2 ) Tkq | 2 m2 γ2 | 2 m2 γ2 (6. i.235 . 2 |γ2 ) is an eigenstate of J 2 with eigenvalue 2 1 ( 1 + 1). hence. One can.1: Show that Φ 1 m1 (k. J3 | 2 m2 γ2 = m2 γ2 | 2 m2 and the property (6.e. 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. behave like |kq | 1 m1 . J 2 | 2 m2 γ2 = 2 2( 2 + 1) | 2 m2 γ2 . J3 are the generators of the rotation R(ϑ).6.237) is that their matrix elements 1 m1 γ1 |Tkq | 2 m2 γ2 obey simple relationships expressed in terms of Clebsch-Gordan coeﬃcients. 2 |γ2 ) is an eigenstate of J 2 with eigenvalue + 1). The hermitian property of J3 and J 2 implies that states Φ 1 m1 (k. which is characterized also by a set of other quantum numbers γ1 which are not aﬀected by the rotational transformation R(ϑ).180 Theory of Angular Momentum and Spin 6. 2 |γ2 ) = ( 1 m1 |kq 2 m2 ) Tkq | 2 m2 γ2 . (6. possibly the total angular momentum– spin state of a compositie system. 2 |γ2 ) = q. i. These states according to (6. 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.e. .250) q. J3 | 2 m2 γ2 = m2 | 2 m2 γ2 . i.21) of Clebsch-Gordan coeﬃcients J3 Φ 1 m1 (k. construct states Φ 1 m1 (k. J2 . in fact.18) are deﬁned through Φ 1 m1 (k. 1 m1 |Φ 1 m1 (k. (6.247).8. J3 . the stated property.252) ∼δm1 q+m2 = ( 1 m1 |kq 2 m2 ) Tkq Similarly.8 Wigner-Eckart Theorem A second important property of tensor operators Tkq beside (6. 2 |γ2 ) can be shown readily as follows using (6. 2 |γ2 ) are orthogonal to | 1 m1 in case of diﬀerent quantum numbers 1 . m1 . 2 |γ2 ) which correspond to total angular momentum states.249) which demonstrates.

e.2: Determine the matrix elements of the gradient operator d3 rF (r) G(r) of the type (6. γ1 ||Tk || 2 .255) At this point the important property can be proven that 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) is independent of m1 . γ2 ( 1 m1 |kq (6. γ1 ||Tk || 2 .g. can be evaluated as easily as possible. its m1 –dependence being expressed solely through a Clebsch-Gordan coeﬃcient.8: Wigner-Eckart Theorem 181 In order to evaluate the matrix elements (6.s. m1 = q = m2 = 0. γ2 k− 2 + 1 √ 1 2 m2 ) (−1) 2 1 +1 = 1 . for which the l. One can then evaluate also the corresponding Clebsch-Gordan coeﬃcient ( 1 m1 |kq 2 m2 ) and determine 1 . say m1 + 1. 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 . (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. m2 .259) The socalled reduced matrix element 1 . γ2 (−1)k− 2 + 1 ( 1 m1 |kq 2 m2 ) (6.260) Exercise 6. (6.256) and noting that the operator adjoint to J+ is J− . the matrix elements 1 m1 γ1 |Tkq | 2 m2 γ2 can be reduced to an m1 –independent factor. i. γ1 |Tkq | 2 m2 .32) Tkq | 2 m2 γ2 = and orthogonality property (6. Using | 1 m1 + 1γ1 = 1 ( 1 + m1 + 1)( 1 − m1 ) J+ | 1 m1 γ1 (6. 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 .259) to a combination of magnetic quantum numbers m1 . γ2 is determined by applying (6.261) . q .258) We can then ﬁnally express the matrix elements of the tensor operators Tkq as follows 1 m1 . To prove this property we consider 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) for a diﬀerent m1 value.h. γ1 ||Tk || 2 .8. γ2 = 2 1 +1 1 m1 .257) which establishes the m1 –independence of 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) .254) = 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) ( 1 m1 |kq 2 m2 ) . γ1 ||Tk || 2 .e. γ1 |Tkq | 2 m2 .6.

φ) +1m (θ. φ) = Y −1m (θ. evaluate the matrix for m1 = q = m2 = 0 using cosθ Y ( +1−m)( +1+m) (2 +1)(2 +3) m (θ. φ) = ( −m)( +m) (2 −1)(2 +1) Y +1m (θ.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). φ) sinθ Y √ ( +1) (2 +1)(2 +3) Y m (θ. but a very useful exercise!) . (The necessary evaluations are cumbersome. For this purpose relate r operator T1q . φ) + Y −1m (θ. φ) − √ ( −1) 2 −1)(2 +1) and express the remaining matrix elements using the Wigner–Eckart theorem.

The corresponding wave functions serve basis functions for multi-electron systems in atoms. Poststrasse 22. and crystals.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 . for example. It leads to the stationary electronic states of the hydrogen atom (Z = 1). for example. molecules. Potential (7. See. 7. Four examples will be studied. The second potential is of the square well type −Vo 0 ≤ r ≤ R . Simple Models of Complex Nuclei / The Shell Model and the Interacting Boson Model by I. 1993) 1 183 .2) serve as schematic descriptions of quantum particles. The potential (7. φ. The ﬁrst potential V (r) = 0 0 ≤ r ≤ R ∞ r > R (7. (7.1) conﬁnes a freely moving particle to a spherical box of radius R.3) describes an isotropic harmonic oscillator .4) is by far the most relevant of the four choices.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). Switzerland. i.. The fourth potential V (r) = − Z e2 r (7.2) V (r) = 0 r > R The third potential V (r) = 1 mω 2 r2 2 (7. The potentials (7.4) governs the motion of electrons in hydrogen-type atoms. in case of the so-called bag model of hadronic matter. in potentials which are solely a function of r and are independent of the angles θ. 7000 Chur.1. Talmi (Harwood Academic Publishers.e.

m (r) . Jk } where H is the Hamiltonian.5) In the case of an angular independent potential angular momentum J = m r × v is a constant of ˙ motion.6) ˙ Since Jk is also equal to the poisson bracket {H. For this purpose one concludes ﬁrst that the conservation of angular momentum J implies a motion of the particle conﬁned to a plane.61) of J 2 and Jk . ˆ A classical particle moving in a potential V (r) is governed by the Newtonian equation of motion (7.7) ˆ The correspondence principle dictates then for the quantum mechanical Hamiltonian operator H and angular momentum operators Jk ˆ [H.12) .87) for Jk as well as the commutation property (5. one can state ˙ J = m r2 θ . Employing in this plane the coordinates r. k = 1. .m (r) . 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.8) This property can also be proven readily employing the expression of the kinetic energy operator [c..85– 5. d ˙ J = v × mv + r × mv = 0 + r × er (−∂r V (r)) = 0 . k = 1. 3. i.15) (7.5) and r = v. (7. Accordingly. 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 . 2 2mr2 (7. J3 ψE.13) .184 Spherically Symmetric Potentials 7. (7. ˆ H ψE.m (r) = = = E ψE.99)] 2 1 2 J2 2 2 = − ∂r r + .9) − 2m 2m r 2mr2 the expressions (5. 2 . 3 . Jk ] = 0 .m (r) can be chosen as simultaneous eigenstates of the Hamiltonian ˆ operator H as well as of J 2 and J3 .m (r) (7. stationary states ψE.e. using (7.f. ( + 1) ψE. 2. the time variation of J can be written.1 Radial Schr¨dinger Equation o ˙ mv = − er ∂r V (r) . Jk } = 0 .m (r) . In fact. θ for the distance from the origin and for the angular position. 2 . (7. .10) (7. J ψE. (5.14) (7.11) (7. 2. ˆ dt (7.m (r) m ψE. one can conclude ˙ Jk = {H.

. One can write (7. . using (7.17) In analogy to the classical description one can derivem. .24) . (7.19) where Y m (θ.11).21) Veﬀ (r) φE (r) = = V (r) + r vE. φ) (7. for the wave function of a quantum mechanical particle a diﬀerential equation which governs solely the r-dependence. . (7. mr2 (t) (7.10) − 1 2 ∂ r + 2m r r 2 2 ( + 1) + V (r) − E 2mr2 .m (r) 2 .m (r) Y m (θ.m (r) . (7. in the present case.12) are obeyed and one obtains for (7. .m (r) 1 2 J2 ∂r r + + V (r) − E 2m r 2mr2 the functional form ψE.m (r) = 0.m vE.22) which is the original potential 2 ( +1) V (r) with an added rotational barrier potential 2mr2 .1: Radial Schr¨dinger Equation o 185 This is a diﬀerential equation which governs solely the radial coordinate. (7.22) (7. It can be solved by integration of 1 2 2 J2 r = ± ˙ E − V (r) − ..11.18) = vE. . − Adopting for ψE.20) in the form of the one-dimensional Schr¨dinger equation o 2 − where 2m 2 ∂r + Veﬀ (r) − E φE (r) = 0 .13). .e.20) by −2mr/ 2 yields the so-called radial Schr¨dinger equation o 2 ∂r − ( + 1) − U (r) − κ2 r2 r vκ.m ψE. ∞[ governed by the eﬀective potential (7.m . Multiplying (7. together with the original potential.10). 7. 2 (7. (7.7.20) Since this equation is independent of the quantum number m we drop the index m on the radial wave function vE. φ) are the angular momentum eigenstates deﬁned in Section 5. can exclude particles from the space with small r values. but can also trap particles in the latter space giving rise to strong scattering resonances (see Section ??).23) This demonstrates that the function r vE.9) one can write the stationary Schr¨dinger o equation (7. ( + 1) 2mr2 . (r) = 0 .4. by integration of ˙ θ = J . This barrier.m (r) = 0.m (r) describes the radial motion as a one-dimensional motion in the interval [0.m (r) and E . equations (7.16) m 2mr2 Once. i. r(t) is determined the angular motion follows from (7. Employing the kinetic energy operator in the form (7.

at r = 0 is found in the excellent monographs Quantum Mechanics I. 2nd Ed. using the expression (5.20) the index E by the equivalent index κ. (r) ∼ A r or vκ. see. Pascual (Springer. For r → 0 one may assume that the term ( + 1)/r2 becomes larger than the potential U (r). 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). r → 0 (7. We replaced in (7.186 where we deﬁned U (r) κ2 = = − − 2m 2 Spherically Symmetric Potentials V (r) E (7.m (r) ∼ √ 4π B δ(x1 )δ(x2 )δ(x3 ) . Berlin. (7.26) 2m 2 In case E < 0. 4π|r| (7. Jackson (John Wiley.28) (7. (r) (7.182) for Y00 .27) ∂r − r2 and. . ”Classical Electrodynamics. In this case the solution is governed my ( + 1) 2 r vκ.31) to the complete wave which. accordingly. ψE.D. (r) = 0 .32) The total kinetic energy resulting from this contribution is. . II by A. .30) and. the term Br−( +1) would contribute |B|2 0 dr r−2 +1) (7.m (r) ∼ √ B . for > 0 is not integrable.” by J. for example. 1975).20) one needs to specify proper boundary conditions2 . assumes the general form r vκ. 2 2 2m 2 ψE. according to a well-known result in Classical Electromagnetism3 . For = 0 the contribution of Br−( function is.33) A detailed discussion of the proper boundary conditions. (r) ∼ A r + B r− −1 +1 + B r− (7. New York. in particular. This follows for > 0 from consideration of the integral which measures the total particle density.29) Only the ﬁrst term is admissable. κ assumes real values.25) (7. Galindo and P. Boundary Conditions In order to solve (7. hence. The radial part of this integral is ∞ 0 2 dr r2 vκ.

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).

99) +1 d cos θ eikr cos θ P (cos θ) = b j (kr) −1 2 .100) Deﬁning x = cos θ.s.. (7. φ) . φ) for non-vanishing m has a non-trivial φ-dependence as described by (5.e.99) We want to determine the expansion coeﬃcients b . 2 +1 (7. For example. 7. can be written exp(ikr cos θ). one given o by (7. i. One can expand the former solution in terms of solutions (7.98) by ∞ eik3 x3 = =0 b j (kr) P (cos θ) . In this case the expansion on the r.h.98) does not involve any non-vanishing m-values since Y m (θ. of (7.45) and one given by (7. hence.50). (7. φ). in case of a free particle moving along the x3 -axis one expands eik3 x3 = .106). The orthogonality properties (5. namely.43) has two solutions. This holds for −1 and one can conclude +1 consecutive integrations by part d izx e dx (7. and using the Rodrigues formula for Legendre polynomials (5. One can show d −1 2 (x − 1) ∼ (x2 − 1) × polynomial in x dx −1 (7.h.104) dx eizx P (x) −1 = (−1) 2 ! (iz) = 2 ! +1 dx (x2 − 1) −1 +1 dx (1 − x2 ) eizx .150) one obtains +1 +1 1 ∂ dx eizx P (x) = dx eizx (x2 − 1) . z = kr.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 (7.102) −1 d −1 +1 −1 dx −1 (x2 − 1) .m a mj (kr) Y m (θ.98) The l. the surface term ∼ [· · ·]+1 vanishes.44.103) and.50).179) yield from (7. the wave function does not depend on φ. −1 . (7. according to (5. Since the spherical harmonics Y 0 (θ.s.194 Generating Function of Spherical Bessel Functions Spherically Symmetric Potentials The stationary Schr¨dinger equation of free particles (7.178) are given in terms of Legendre polynomials P (cos θ) one can replace the expansion in (7.

1 (7.7.96) one can express this.108) where the last factor is equal to unity. for the leading power x [c.f.114) .112) We want to consider the expression Gν (z) = aν z ν fν (z) where we deﬁne fν (z) = C (7.110) One refers to the l. ∞ (7. (7. −1 (7.s. in particular. as the generating function of the spherical Bessel functions. 2 +1 −1 dx (1 − x2 )ν− 2 eizx .h.105) This expression allows one to determine the expansion coeﬃcients b .105) and (7. after insertion into (7.109) results in the integral representation of j (z) j (z) = (z) 2 +1 ! +1 dx (1 − x2 ) eizx . −1 (7. Integral Representation of Bessel Functions Combining (7.106) Employing (5.h.105) must hold for all powers of z.107) 1 2 (2 )! 1 · 3 · 5 · · · (2 + 1) 2 + 1 1 · 2 · 3 · 4 · · · (2 − 1) · 2 (7.69)] b z 2 (iz) = 1 · 3 · 5 · · · (2 + 1) 2 + 1 2 ! +1 dx (1 − x2 ) . 1 (7.109) eikr cos θ = =0 i (2 + 1) j (kr) P (cos θ) .117) one can write the r.2: Free Particle Comparision with (7.111) Employing (7. z i or i (2 + 1) z (2 )! 2 2 2 +1 2 ! [1 · 3 · 5 · · · (2 − 1)] 1 (7. of (7. (7. Comparision with the l.100) gives b j (kr) 2 (iz) = 2 +1 2 ! +1 195 dx (1 − x2 ) eizx .106) yields ﬁnally b = i (2 + 1) or. using ν = Jν (z) = √ 1 πΓ(ν + 1 ) 2 z 2 ν + 1.113) dt (1 − t2 )ν− 2 eizt . The identity (7.s.h.99). −1 (7.s.

196 Spherically Symmetric Potentials Here C is an integration path in the complex plane with endpoints t1 . We consider now the functions u(j) (z) = √ 1 πΓ(ν + 1 ) 2 z 2 ν Cj dx (1 − x2 )ν− 2 eizx .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 .4: Prove (7. We can now demonstrate that Gν (z) deﬁned in (7. Exercise 7. Gν (z). z (7.118) z 2ν + 1 fν (z) + fν (z) .124) . In fact.93) as long as the integration path in (7.120) such that we can conclude that Gν (z) for proper integration paths is a solution of (7.120) imply also the property (2ν + 1) fν (z) + z fν+1 (z) = 0 .113) satisﬁes (7.2. for properly chosen endpoints t1 .121) (7. t2 .122) (7. 1 (7.93).93) for arbitrary ν.116) In case that the endpoints of the integration path C satisfy (1 − t2 )ν+ 2 eizt one can write (7. 1 j = 1. 3.123) Insertion of these identities into Bessel’s equation leads to a diﬀerential equation for fν (z) which is identical to (7. obeys Bessel’s equation (7. (7.119) We note that equations (7.117).121). To prove this we note fν (z) = i C dt (1 − t2 )ν− 2 t eizt .114.117) dt (1 − t2 )ν− 2 t2 eizt 1 (7.113) obeys the Bessel equation (7. 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. 7. (7. t2 of the integration paths C. C 1 (7. 4.116) fν (z) = − or fν (z) = − From this we can conclude fν (z) + 2ν + 1 fν (z) + fν (z) = 0 . 2.120) z 2ν + 1 dt (1 − t2 )ν− 2 eizt + C C 1 1 t2 t1 = 0 (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 .130) We note that the endpoints of the integration paths C2 and C4 . one can derive (2) Hν (z) = e −i(z−πν/2−π/4) ∞ 0 √ ds [s(1 − is/2)]ν− 2 e−zs .131) According to (7.124) vanishes along the whole path C3 and. Since the integrand in (7.7.117) and.124) is analytical in the part of the complex plane surrounded by the path C we conclude 4 u(j) (z) ≡ 0 j=1 (7.125) 0 ≤ s < ∞ (7. 1 (7. 1 (7. for Rez > 0 and ν ∈ R obey (7. therefore.112) shows u(1) (z) = Jν (z). the integral expression (1) Hν (z) 1 1 1 1 (7.130) holds Jν (z) = (1) 1 2 1 (1) (2) Hν (z) + Hν (z) . Comparision with (7. dt = i ds. u(2) (z) and u(2) (z) are both solutions of Bessel’s equation (7.128) C = C1 ∪ C2 ∪ C3 ∪ C4 is a closed path.129) The integrand in (7. u(3) (z) ≡ 0.135) are known as the Poisson integrals of the Bessel functions.93).134) Similarly. one can state Jν (z) = − u(2) (z) + u(4) (z) .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. (7.134) and (7.127) 0 ≤ s < ∞ (7. we introduce the so-called Hankel functions (1) Hν (z) = −2 u(2) (z) . (7. using t = 1 + is. (2) Hν (z) = −2 u(4) (z) .132) For Hν (z) one derives. Accordingly.135) (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 . (7.126) −1 ≤ s ≤ 1 (7. Following convention. . hence.

πz (7.e.135) which read for ν = + 1 2 H where f The binomial formula yields f (1) 2 (1) 2 1 +2 (1.134) and (7.140). One obtains W (z) = − 4i .136) (1) 2 (z) = 0 ds [s(s ± is/2)] e−zs . Hν (z) as well as H−ν (z) are solutions of this equation..93) holds the identity 1 W = − W z (7.139) (z) = 2 ±i[z − ( e πz +1) π ] 2 r=0 ( + r)! r!( − r)! i ± 2z r (7. For such solutions g(z). converge fast for |z| → ∞. (7.142) and h(z) = H (2) (z) +1 2 one can identify the constant c by using the leading terms in the expansions (7.2) (z) +1 2 in terms of z −1 such that (z) = e±i[z − ( ∞ +1) π ] 2 2z z ( 1 ) f 2 (z) π ! (7.141) (1) (1) the derivation of which follows the derivation on page 192 for the Wronskian of the radial Schr¨dinger equation. For the Wronskian connected with the Bessel equation (7.140) Bessel Functions with Negative Index Since ν enters the Bessel equation (7. As a second order diﬀerential equation the Bessel equation has two linearly independent solutions.138) The formula for the Laplace transform of sn leads to f and. the Wronskian W (g.141) is o W (z) = − In case of g(z) = H (1) (z) +1 2 c . hence. h(z) to be linearly independent. We employ for this purpose the Poisson integrals (7. The general solution of (7. z (7.93) only as ν 2 .143) .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. h) must be a non-vanishing function.137) (z) = r=0 r ± i 2 r 0 ∞ ds s +r e−zs . (7. i. we obtain H (1) 2 1 +2 (1) 2 (z) = r=0 ( + r)! ! r!( − r)! ± i 2 r 1 z +r+1 (7.

2) (z) +1 2 199 are.96.2) (2) +1) 2 (1) +1) 2 (z) = i (−1) H (1) (z) +1 2 .149.2) H + 1 (z) . 7.147) (z) = π (1. linearly independent.97). (7.150) According to (7. (7.2: Free Particle i. in fact. 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| → ∞ .148) holds for the o regular spherical Bessel function j (z) = 1 (1) (2) [ h (z) + h (z) ] .145) This equation can hold only for B = 0 and A = exp[i( + 1 )π].151) are equivalent to h h (1) (2) (z) (z) = = j (z) + i n (z) j (z) − i n (z) . 2z 2 (1. (7.53).147) this can be written n (z) = 1 2i h (1) (z) − h (2) (z) .148) Following the arguments provided above (see page 193) the functions h (z) are solutions of the radial Schr¨dinger equation of free particles (7.2) (7. B can be obtained from the asymptotic expansion (7.e. (7.151) Equations (7.146.146) (z) = −i (−1) H (2) 1 (z) +2 . H (1. 7. H−ν (z) = A H (1) (1) (z) +1 2 One can then expand + BH (2) (z) +1 2 . one can show H−( Spherical Hankel Functions In analogy to equations (7.140).132) follows n (z) = 1 2 π (−1) 2z +1 H−( (1) 1 +2) (z) + H−( (2) +1) 2 (z) .97) one deﬁnes the spherical Hankel functions h (1.97..153) . 7. We conclude 2 H−( Simarly.149) We want to establish also the relationship between h (z) and the irregular spherical Bessel function n (z) deﬁned in (7. 2 (1.7. 7. From (7.90. (7. (7.132.144) The expansion coeﬃcients A. 7. (7. According to (7. 7.2) (7.152) (7.

are j (z) n (z) = = (7. = −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) . z (7.148). 2z 2 z sin(z − π/2) z cos(z − π/2) − . j (z) (z) = ( i) z +1 e±iz r=0 ( + r)! r!( − r)! ± i 2z r (7. = sin(z − π/2) and sin[z − ( + 1) π ] . (7. ne ll(z) = Im[h (1) (z)] .161) (7. From (7. 2 2z z (7.163) (7. (1) (7. 7. Hence.159) (7.158) one can derive then from (7.156) ( + r)! r!( − r)! ( + r)! r!( − r)! i 2z r [ /2] = r=0 ( + 2r)! (2r)!( − 2r)! −1 4z 2 r (7.2) (z).160) Employing cos[z − ( + 1) π ] .154) The leading term in this expansion.155) (z) and To determine j (z) and n (z) we note that for z ∈ R the spherical Hankel functions h (2) h (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 (7.140. is h (1) 2 (z) = ( i) z +1 e±iz .148) one obtains readily h (1) 2 (1.162) The leading terms in these expansions. as given by (7.200 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). are complex conjugates. at large |z|.164) .140) and (7. it follows z ∈ R : Using p (z) = Re r=0 j (z) = Re[h (1) (z)] . at large |z|.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) .

151) follows then that the relatiosnhips hold also for g (z) = j (z) and g (z) = n (z).2) relationships hold for g (z) = h (z). 1.4 dx (1 − x2 )ν− 2 eizx .171) −1 (z) (7.131.171–7.7.149.148) and express g (z) = h (1. (1. One obtains for = 0. 7. (7.174) We want to prove these relationships.2) To demonstrate that (7.176) . 7. 7.2) g (z) − g (z) z 2 +1 g (z) − g z (7. deﬁning a = −1 and employing fν (z) as deﬁned in (7.173) A (z) = z ( +1) z2 +2 z (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. j (z) and n (z).171.161.2: Free Particle Expressions for the Spherical bessel Functions j (z) and n (z) 201 The identities (7. 7.114) we can write g (z) = a z f 2 ! + 1 (z) 2 .124. 7.170) Recursion Formulas of Spherical Bessel Functions The spherical Bessel functions obey the recursion relationships g g +1 (z) +1 (z) = = (1.162) allow one to provide explicit expressions for j (z) and n (z).166) (7.172) to +1 (z) +1 (z) = A (z) 1− g (z) g (z) −1 (7. 1 (7.175) Using Γ( + 1) = !.171) holds for g (z) = h (z) we employ (7.168) (7.167) (7. From the linearity of the relationships (7.172) where g (z) is either of the functions h obtain the recursion relationship g g (z).165) (7.174) and from (7. For this purpose we need to demonstrate only that the (1.169) (7.2) (z) = −2 π 1 √ 2z πΓ(ν + 1 ) 2 z 2 ν C2. One can combine (7.

f yields. (7.171). j2 (z) from j0 (z). i.178) (7.2.182) Replacing all ﬁrst derivatives employing (7.6: Employ the recursion relationship (7. The second component of (7.172).171). n2 (z) from n0 (z).121).173. In order to prove (7. z2 (7.171). The ﬁrst component of (7.174) we start from (7. Exercise 7.182).172) we diﬀerentiate (7.177) (z) = − z f 2 +2 + 3 (z) 2 . (7.58) it holds o 2 g (z) = − g (z) + z ( + 1) g (z) − g (z) .173) is equivalent to (7.166). 7. .179) leads to (7. To prove (7. is equivalent to (7.202 The derivative of this expression is g (z) = Employing (7. 7. j0 (z) using (7. in fact. and (b) n1 (z).2. z (7. together with (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) . (7.181) Using this identity to replace the second derivative in (7.180) Since g (z) is a solution of the radial Schr¨dinger equation (7.5: Provide a detailed derivation of (7. n0 (z) using (7.173..173).179) g (z) = − z2 g (z) + g (z) − g +1 (z) z .174) to determine (a) j1 (z). Exercise 7.172).179) from which follows (7.e.180) yields g +1 (z) = g (z) − ( + 2) g (z) + z2 +2 g (z) .165).

The classical electromagnetic ﬁeld is governed by the Maxwell equations stated already in (1... 8.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. These sources enter the two inhomogeneous 203 . the attractive Coulomb force between proton and electron in case of the hydrogen atom. and the external electromagnetic ﬁeld are of electromagnetic origin and. We assume that the system considered is in vacuum in which charge and current sources described by the densities ρ(r.29). e. for situations in which the photons absorbed or emitted by the quantum system do not alter the energy content of the ﬁeld. Such description should be suﬃcient for high quantum numbers..e. However.g.27– 1. It turns out that the proper description of the electromagnetic ﬁeld requires a little bit of eﬀort. and are aﬀected by the external electromagnetic ﬁeld. then state the Hamiltonian which describes the resulting interaction. for the existence of discrete photons. i.g. The problem is that the latter electron. This is achieved in the following derivation. t) are present. t) and J(r. e. both the Coulomb interaction due to charges contributing to binding the particle. on an electron in a hydrogen atom. hence.. We will describe the electromagnetic ﬁeld classically. are aﬀected by the Coulomb interaction V (r) which is part of the forces which produce the bound state. e.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. i.e. For this purpose we need to establish a suitable description of this ﬁeld. is aﬀected by electromagnetic ﬁelds.. must be described consistently. an electron in a hydrogen atom. or other charged particles.g.

1) (8. accordingly. t) is a scalar function. t) −→ 1 A(r.204 Maxwell equations1 · E(r. the two homogeneous Maxwell equations hold × E(r. in turn. 2nd Edition. contributes to the charge density ρ(r. t) . The reader is referred to the well-known textbook ”Classical Electrodynamics”. obeys the equation of motion d p = q dt E[ro (t). called the scalar potential. . called gauge transformations.7) Gauge Transformations We have expressed now the electric and magnetic ﬁelds E(r. t] + v × B[ro (t). t) and B(r. 1975) for a discussion of these and other conventional units. t)] and. t) · B(r. the relationship between ﬁelds and potentials is not unique. t) . t) + χ(r. t) + ∂t B(r. called the vector potential. t) . (8. Scalar and Vector Potential Setting B(r.4). t) and A(r. D. Jackson (John Wiley & Sons. t) = 4 π J(r.9) (8. t) × B(r. t) in (8. From this follows E(r.2) the term q r˙o δ(r − ro (t)).5) above agrees with the equation of motion as given in (1. The following substitutions. (8. t] (8. t) (8.4) Lorentz Force A classical particle with charge q moving in the electromagnetic ﬁeld experiences the so-called Lorentz force q[E(r. t).3) reads then × E(r.2) In addition. t) (8. t) − ∂t A(r. Equation (8. t) = 0 = 0. t) which is solved by E(r.11) V (r. t) = − V (r. t) where V (r. In the non-relativistic limit holds p ≈ mr and (8. New York. The particle. t) −→ V (r.3) (8. solves implicitly (8. t) = × A(r. t) + ∂t A(r.5) where p is the momentum of the particle and ro (t) it’s position at time t. t) = − V (r. t). t) + v × B(r. As is well known. alter the potentials. t) through the scalar and vector potentials V (r. by J. We assume so-called Gaussian units. t) Interaction of Radiation with Matter = 4 π ρ(r.25).10) (8.6) for some vector-valued function A(r.8) = 0 (8. t) in (8. t) − ∂t E(r. t) − ∂t χ(r. t) + ∂t A(r. (8. but leave the ﬁelds unaltered: A(r.1) the term qδ(r − ro (t)) and to the current density ˙ J(r.

t) − ∂t A(r. t) = − 4 π J(r. 4π (8. (8. 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. Using the expressions (8. t) = Ω∞ for r ∈ ∂Ω∞ (8. t) + ∂t A(r.2).6) and (8.17) A(r. a name which is due to the form of the resulting scalar potential V (r. t) the condition · A(r. Such description can. t) = ∂t = 4 π ρ(r.18) together with condition (8. t) and the current due to moving net charges.19) Unfortunately.19) and deﬁne J (r. this potential results from inserting (8.14) In case of the boundary condition V (r. in fact. and a component due to the remaining currents. t) − 2 × A(r.15) d3 r ρ(r . In fact. t = 4 π J(r.19) couples the vector potential A(r.12) yields then the Poisson equation 2 V (r. t) − ∂t V (r. t) can be obtained employing (8. t) together with (8. t) = − 4 π ρ(r. the gauge freedom allows us to impose on the vector potential A(r. (8. t) = 0 the solution is given by the Coulomb integral V (r. t) . t) . t) = · A(r. t). t) (8. t) + ∂t V (r. equation (8.9) into (8. The vector potential A(r. be achieved.9) for the ﬁelds results in × The identity × × A(r. t) and V (r. t) in (8.12) leads us to 2 2 A(r. One would prefer a description in which the Coulomb potential (8. t). t) − ∂t A(r. t) = 1 ∂t V (r. and the former the Coulomb interactions in the unperturbed bound particle system. t) |r − r | (8. In fact. t) . t) .1) · − V (r. such that the latter describes the electromagnetic radiation.8.16) and the vector potential are uncoupled.16) This is the potential commonly employed in quantum mechanical calculations for the description of Coulomb interactions between charged particles. the second inhomogeneous Maxwell equation.13) · A(r. (8. t) . t) = 0 . (8.20) . For this purpose we examine the oﬀending term ∂t V (r.12) The corresponding gauge is referred to as the Coulomb gauge.

2 Planar Electromagnetic Waves The current density Jt describes ring-type currents in the space under consideration. t) = 0 . t) .. t) . In this case (8. Let J(r.26) This equation does not couple anymore scalar and vector potentials. (8. the name longitudinal electric ﬁeld. t) = 0 (8.25) one refers to J and Jt as the longitudinal and the transverse currents. no divergence-free currents. t) = 0 . t) (8. yet a current. V (r. For J also holds the continuity equation · J(r. hence. t) contributes solely an electric ﬁeld component E (r.12) and the Coulomb potential (8.27) which is obviously curl-free ( × E (r.30) . exist in the space considered. t). t) = − ∂t ρ(r.16) yield ﬁnally the electric and magnetic ﬁelds. Presently. 8. t) = − 4 π Jt (r. respectively.14). using ∂t · J (r. t) = − V (r. For the divergence of J (r. t) Bt (r. t) be the total current of the system under investigation and let Jt = J − J .26) turns into the well-known wave equation 2 2 A(r. (8. t) = 0 . V (r.23) This continuity equation identiﬁes J (r..28) (8. in a ring-shaped antenna which exhibits no net charge. t) = 0). such current densities exist. t) holds. t) contributes an electrical ﬁeld component as well as the total magnetic ﬁeld Et (r. t) = 0). t) + ∂t ρ(r.25) Because of properties (8. t) − ∂t A(r. we want to assume that no ring-type currents. = − ∂t A(r.26) and (8.29) · Et (r. i. for example.19) yield 2 2 A(r.21) ∂t and the Poisson equation (8. t) + ∂t ρ(r.24) (8. t) These ﬁelds are obviously divergence -free (e.22) 1 ∂t 4π · J (r. (8. t) = × A(r.e. the name transverse ﬁelds. hence.21) and (8. t) = or = 2 Interaction of Radiation with Matter (8.206 For the curl of J holds × J (r. t) as the current due to the time-dependence of the charge distribution ρ(r. (8. t) = 0 and from this follows · Jt (r. The vector potential determined through (8.g. t) − ∂t A(r. t) (8. A(r. The deﬁnitions of J and Jt applied to (8.

39).38) ˆ In this expression for the energy ﬂux one can interprete k as the propagation velocity (note c = 1) and.8. (8.37) ˆ ˆ where k is the unit vector k = k/|k|.2: Planar Electromagnetic Waves 207 which describes electromagnetic ﬁelds in vacuum. obviously. t) × Re B(r. Time average over one period 2π/ω yields S(r.158).34. A complete set of solutions is given by the so-called plane waves A(r. (8. (8. is orthogonal to k.32) holds. ˆ (8. 8.35) are complex-valued quantities. t) and Bt (r. t) = Ao u exp i(k · r ˆ where the dispersion relationship |k| = ω (8.31.29). are orthogonal to each other and are both orthogonal to the wave vector k. (8. hence. 2 . see also the comment below Eqs.34. there exist two linearly ˆ u independent orientations for u corresponding to two independent planes of polarization.31) u is a unit vector (|ˆ| = 1) which. 8. Et (r. t) .35). In applying the potential and ﬁelds to physical observables and processes we will only employ the real parts. the constant k is referred to as the wave vector. t) = Re Et (r. Obviously. ω2 = |Ao |2 (8. t) in (8. The Coulomb gauge condition (8. 8. at each point r and moment t.31). Here the “-” sign corresponds to so-called incoming waves and the “+” sign to outgoing waves2 . t) . are Et (r.34. 8.28. t) = ±i ω A(r.31) and the resulting ﬁelds (8.32. The corresponding electric and magnetic ﬁelds. 8. t) Bt (r. according to (8. accordingly. Note that in the units chosen the velocity of light is c = 1.33) ωt) (8.12) yields u·k = 0.38. This ﬂux is given by 1 S(r. t) = i k × A(r. The latter vector describes the direction of propagation of the energy ﬂux connected with the plane wave electromagnetic radiation. t) = ± ω2 ˆ |Ao |2 k . 8π (8. 8.36) 4π Using the identity a × (b × c) = b (a · c) − c (a · b) and (8. t) = ± ω2 ˆ |Ao |2 k sin2 ( k · r − ωt ) 4π (8.35) The vector potential in (8.34) (8.39) 8π The deﬁnition incoming waves and outgoing waves is rationalized below in the discussion following Eq. 8.35) one obtains S(r. ˆ We want to characterize now the radiation ﬁeld connected with the plane wave solutions (8.

(8.42) 8. we will neglect the respective terms in the Hamiltonian (8. 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.43). (8. t).31) allows one ﬁnally to state for the planar wave vector potential A(r. (8.40) = V It should be pointed out that Nω represents the number of photons for a speciﬁc frequency ω.39) through the density of photons connected with the planar waves (8.8.38) implies that for incoming waves. A correct description of the electromagnetic ﬁeld requires that the ﬁeld be quantized.39) and (8. the energy of the plane wave is transported in the direction of −k.27.31). i (8. 8. (8. hence. The integrals express the integration over the energy density of the ﬁelds.31. deﬁned below Eqs. We are left with a classical Hamiltonian function which has an obvious quantum mechanical analogue ˆ H = ˆ p − q A(r. u · k = 0 . t) = H Ψ(r. t) 2m + 1 8π Ω∞ 2 H = + qV (r) 1 16π d3 r |Et |2 + |Bt |2 Ω∞ d3 r E 2 + . These photons each carry the energy ω.208 Interaction of Radiation with Matter as the energy density. t) is real 1 and that Et (r. We assume that the energy content of the ﬁelds is not altered signiﬁcantly in the processes described and.32).31). The wave function Ψ(r. The sign in (8. Bt (r.43) Here the ﬁelds are deﬁned through Eqs.45) .44) . a ˆ speciﬁc k and a speciﬁc u.28.41) ωV Inserting this into (8. 8. respectively. Comparision of (8. t) 2m 2 + qV (r) . |k| = ω . If we consider a volume V with a number of photons Nω the energy density is obviously Nω ω .3 Hamilton Operator The classical Hamiltonian for a particle of charge q in a scalar and vector potential V (r) and A(r. 8.40) allows one to express then the ﬁeld ˆ amplitudes 8πNω Ao = . ˆ (8. (8. whereas in the case of outgoing waves the energy is transported in the direction of k. t) . A ‘poor man’s’ quantization of the ﬁeld is possible at this point by expressing the energy density (8. (8. t) = 8πNω u exp i(k · r − ωt) ˆ ωV . is p − q A(r. t). Note that E (r.29) together with the potentials (8.16. t) are complex leading to the diﬀerence of a factor 2 in the energy densities of the lontitudinal and transverse components of the ﬁelds.

hence. t) are necessary to compensate terms which arise through the phase factor. The idea just stated can be generalized by noting that multiplication by a phase factor eiqχ(r.44.11) aﬀect the quantum mechanical description. t) . i. t) inside the double brackets. t) ˆ = eiqχ(r. t) .t)/ Ψ(r. not to phenomenological interactions like the molecular van der Waals interaction.51) 2m the potentials A(r.g.10. It should be noted. t) ˆ p eiqχ(r.e. t) = + qV Ψ(r. t)|2 and. t) enter whereas in the classical equations of motion ˙ m¨ = q E(r. t) (8. 8. (8. hence.49) (8.10.3: Hamilton Operator 209 Gauge Transformations It is interesting to note that in the quantum mechanical description of a charged particle the potentials V (r. Obviously. such phase factor does not change the probability density |Ψ(r.10. t) and A(r. 8.t)/ p + q(( χ)) Ψ(r. have no eﬀect on the motion of the particle described by (8. In the classical case such question is mute since the gauge transformations do not alter the ﬁelds and.11) to (8. 8. . Applying the gauge transformations (8.47) i ∂t Ψ(r. accordingly.t)/ Ψ(r. One can show that (8. t) + q r × B(r. however.47) is equivalent to 2 ˆ p − qA i ∂t eiqχ(r.11) of the potentials is equivalent to multiplying the wave function Ψ(r. 8. e.45) leads to the Schr¨dinger equation o 2 ˆ p − q A − q(( χ)) + qV − q((∂t χ)) Ψ(r.t)/ [ i ∂t − q((∂t χ)) ] Ψ(r.47.48) implies that the gauge transformation (8. t) and V (r. t) . 8. t) = + qV eiqχ(r. an element of the group U(1).t)/ Ψ(r.t)/ should not aﬀect a system and that.46) the ﬁelds enter.. 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.t)/ Ψ(r. t) r (8. (8..8. by the group 3 The eﬀect on other expectation values is not discussed here. in the Schr¨dinger equation o 2 ˆ p − qA i ∂t Ψ(r.46). that this principle applies only to fundamental interactions. t) = 2m where ((· · ·)) denotes derivatives in (( χ)) and ((∂t χ)) which are conﬁned to the function χ(r. Elementary constituents of matter which are governed by other symmetry groups. t) by a local and time-dependent phase factor eiqχ(r.48) 2m For this purpose one notes i ∂t eiqχ(r.50) The equivalence of (8.t)/ . does not change expectation values which contain the probability densities3 . t) = eiqχ(r.t)/ constitutes a unitary transformation of a scalar quantity. (8. This leads to the question in how far the gauge transformations (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. likewise can demand the existence of ﬁelds which compensate local transformations described by eiσ·χ(r. · A = 0 [cf.12)]. t). x1 . k3 . for Bo = Bo e3 in (8.57) satisﬁes · A = 0 and.52) This expression vanishes in the present case since since ˆ ˆ p · Af = A · pf and. the so-called Landau gauge ˆ associates the vector potential AL (r) = Bo x1 e2 ˆ (8. (8. The vector potential (8. k2 . For this purpose we use the wave function in the form Ψ(E. e.54) ˆ2 p q ˆ q2 2 − p·A + A + qV .59) . 2. Accordingly.56) due to the gauge freedom. the generators of SU(2). (8.g. 3. j = 1.45) with Hamiltonian (8. The Hamiltonian (8. The resulting ﬁelds are called Yang-Mills ﬁelds. The choice of a vector potential aﬀects the form of the wave functions describing the eigenstates and. (8. (8. consequently. one can employ the Hamiltonian (8. 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. The stationary homogeneous magnetic ﬁeld B(r. can be described by various vector potentials.. 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. therefore. aﬀects the complexity of the mathematical derivation of the wave functions. t) = Bo . H = 2m m 2m (8.56). x2 .55).58) where ∂j = (∂/∂xj ).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. thereby. We want to describe the stationary states corresponding to the Hamiltonian (8. x3 ) = exp(ik2 x2 + ik3 x3 ) φE (x1 ) .55). In case of a homogeneous potential pointing in the x3 -direction. t) ≡ 0.55) 8.58).57) with a homogeneous magnetic ﬁeld Bo . holds (8.210 Interaction of Radiation with Matter SU(2).53) ˆ ˆ p·A + A·p + q2 2 A + qV 2m (8. In this o case holds V (r.t) where σ is the vector of Pauli matrices.

k3 .57). according to (8. The stationary Schr¨dinger equation (8. 2 (8. (8.65).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 (8.8. x2 .67) .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 ) .65) where we replaced the parameter E by the integer n. The energies corresponding to these states are E(n.63) (8. We want to employ . the wave function (8. 2 2me (8. (8. k3 ) = ω(n + 2 k2 1 3 ) + . 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 ) (8.66) Obviously.62) x1o = and where ω = k2 eBo eBo me (8. therefore. according to (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 ) .64) is the classical Larmor frequency (c = 1). like the corresponding gauge (8.and x2 -coordinates. k2 . From this observation one can immediately conclude that the wave shifted (by e 3 function of the system. Solution for Symmetric Gauge The solution obtained above has the advantage that the derivation is comparatively simple. the so-called symmetric gauge which expresses the homogeneous potential (8. is not symmetric in the x1 . Without aﬀecting the energy one can form wave packets in terms of the solutions (8.59). is Ψ(n. Unfortunately. k2 . the states are degenerate in the quantum number k2 describing displacement along the x2 coordinate.56) through the vector potential A(r) = 1 Bo × r .65) which localize the electrons.63) this induces a spread of the wave function in the x1 direction.59). the familiar harmonic oscillator quantum number. x1 . However. 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.

ˆ Identifying r × p with the angular momentum operator L.72) allows one to interpret µ = − e L 2me (8. the Hamiltonian (8.71). namely.74) is in the present case 1 µclass = − e r v e3 .69) × u and × r = 0. y-plane with constant velocity v on a ring of radius r. Accordingly.72) and (8. 8. (8.73) where µclass is the magnetic moment connected with a current density j µclass = 1 2 r × j(r) dr (8.71) (8. an electron moving in the x.68) × u yields.75) 2 The latter can be related to the angular momentum µclass = − and. The theory of classical electromagnetism predicts an analogue energy contribution . In this case the current density measures −e v oriented tangentially to the ring. for any (8. accordingly.77) Comparision with (8. using × (uf ) = −u × f +f ˆ ˆ Bo · p × r f = Bo · r × p f .74) We consider a simple case to relate (8.12) for the Coulomb gauge. (8.74). the magnetic moment (8.67) one can write ˆ p·A = Using · (u × v) = − u · function f (r) ×v + v· 2i · Bo × r . namely. For the vector potential (8.70) ˆ ˆ p · A f = − Bo · p × r f . Vmag = −µclass · Bo (8.78) .212 Interaction of Radiation with Matter One can readily verify that this vector potential satisﬁes the condition (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 .72) to Hamiltonian (8. The latter can be rewritten. (8.73. (8. in the present case of constant Bo . Vmag = e 2me class = r me v e3 of the electron ˆ (8.76) e 2me class class · Bo . ˆ (8.

For a magnetic ﬁeld (8. 2 .84) .71) is ˆ H = 2 ˆ2 p e2 Bo + 2me 8me x2 + x2 + 1 2 eBo L3 .82) 1 ω (x1 ∂2 − x2 ∂1 ) i 2me + 1 me ω 2 x2 + x2 1 2 2 2 k2 3 + (8.8.67) yields a more symmetric solution which decays to zero along both the ±x1 . the solutions of ˆ H ΨE (x1 . We will demonstrate in Sect. for Bo = Bo e3 .83) (8.56) pointing in the x3 -direction the symmetric gauge (8. x2 ) where ˆ Ho = − 2 2 2 ∂1 + ∂2 (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. we separate the variable x1 . (8. x3 ) . likewise.e. i.80) (8.88) me ω xj .83) describes two identical oscillators along the x1 -and x2 -directions which ˆ are coupled through the angular momentum operator L3 . E = E − (8.81) (8. e gives rise to an energy contribution (8. To obtain these eigenstates we introduce the customary dimensionless variables of the harmonic oscillator Xj = (8.87) .e. j = 1. 2me (8. x2 .85) The Hamiltonian (8.4: Electron in Homogenous Magnetic Field 213 as the quantum mechanical magnetic moment operator for the electron (charge −e). x2 . Accordingly. x3 ) = exp(ik3 x3 ) ψ(x1 . 10. In this case. described by the operator S.and the ±x2 -direction. 2me We have used here the expression for the angular momentum operator ˆ L3 = ( /i)(x1 ∂2 − x2 ∂1 ) . the so-called gyromagnetic ratio of the electron. (8. A derivation of his property and the value of g. x2 ) = E ψ(x1 .86) 1 2 2me 1 2 ˆ as well as of the angular momentum operator L3 .72) with an associated magnetic moment − g 2me S where g ≈ 2.79) To obtain the stationary states. x3 ) = E ΨE (x1 . requires a Lorentz-invariant quantum mechanical description as provided in Sect. 10 that the spin of the electron. The functions ψ(x1 .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 . x2 . the Hamiltonian (8. x2 from x3 setting ΨE (x1 . x2 ) . x2 ) obey then ˆ Ho ψ(x1 . ˆ i.

through a combination of states j Ψ (j. x1 .92) are not eigenstates of L3 .89) . m . x1 . .92) where Ψ(0. x2 ) . m . 0. 1 2 since one phonon is annihilated and one created.94) Since this state is a linear combination of states which all have vibrational energy (2j + 1) ω. .as well as for the x2 -oscillator. . x2 ) = ω 2j + 2m + 1 Ψ (j. aj = √ 2 Interaction of Radiation with Matter Xj + ∂ ∂Xj . m. We. x2 ) = 2m Ψ (j. m . obeyed. x2 ) = m=−j αmm Ψ(j. it holds a† a1 + a† a2 + 1 1 1 2 Ψ (j. x1 . . x2 ) . Such eigenstates can be expressed.94) is also an eigenstate of L3 . 2 (8. (8.214 Employing the creation and annihilation operators 1 a† = √ j 2 and the identity ˆ ω L3 = which can readily be proven. the total vibrational energy being ω(2j + 1). j = 1. x1 . i.90) a† a2 − a† a1 1 2 ..97) . attempt to express eigenstates in terms of vibrational wave functions a† 1 j+m Ψ(j.91) We note that the operator a† a2 − a† a1 leaves the total number of vibrational quanta invariant. m . x1 . x2 ) (8. such that 1 i a† a2 − a† a1 1 2 Ψ (j. x1 . 1 3 .94) is an eigenstate of the vibrational Hamiltonian. x2 ) . in fact. 1. (8. . (8. (8. m . m = −j. 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 . x1 . i.93) ˆ The states (8.e. 0. . therefore.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 . −j + 1. In order to cover all posible vibrational quantum numbers one needs to choose j. x2 ) = a† 1 j−m (j + m)! (j − m)! Ψ(0.e. x1 . m as follows: j = 0.94) is an eigenstate of Ho ˆ Ho Ψ (j. If this property is. (8. . (j) (8. x1 .96) ˆ holds. x1 .95) (j) ˆ We want to choose the coeﬃcients αmm such that (8. m . . (8. x2 ) is the wave function for the state with zero vibrational quanta for the x1 . x2 ) (8. however. x1 . (8. m . 2 2 .. (8. m.

5. The external radiation ﬁeld is accounted for by the vector potential A(r.9. x1 . 5. Other radiation ﬁelds can . 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. 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. x2 ) (8.10. π . x1 . 5. x1 . m.11. The required rotation must transform the x3 –axis into the x2 –axis.11 such transformation is provided through π π (j) Ψ (j. x1 .31). .10. Using the explicit form of the Wigner rotation matrix as given in (5. b† . The connection with the present problem arises due to the fact that the operator J2 in Sect.5. We assume that the scalar potential V in (8. i.102) We have identiﬁed. x2 ) deﬁned in (8. m. 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 . If we identify 2 a† . According to the derivation given there.. a transformation moving the x3 –axis into the x2 – 2 axis. x2 ).99) with eigenvalue m.10. (8.84) above..5.98) then the states Ψ(j. The ﬁrst rotation contributes a factor exp(−im π ). a† .5. the second rotation a factor dmm ( π ).11.288)] 1 ˆ J3 = 2 a† a1 − a† a2 1 2 (8.e. x2 ) = Dmm ( . 5. 0) Ψ(j. m. an example is the Coulomb potential V (r.9.9.92) correspond to the eigenstates |Ψ(j. (8. except for a factor 2 .10. t) = 1/4πr conﬁning an electron with energy E < 0 to move in the well known orbitals of the hydrogen atom.e.100) J2 = 1 2 2i 1 ˆ i. is identical to the operator L3 introduced in (8. b− + − notation in Sects. Eq. which corresponds there to the angular momentum in the x2 –direction.101) 2 2 where Dmm ( π . 5. 5.5 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. This implies that ˆ we can obtain eigenstates of L3 by rotation of the states Ψ(j. x2 ) . a2 1 2 present notation ←→ b† .11 (8. cf. m. the 2 2 latter representing the Wigner rotation matrix of Sect. In the simplest case the radiation ﬁeld consists of a single planar electromagnetic wave described through the potential (8. m . x1 .5. x1 .97).(5. a1 .55). the states are eigenstates of the operator [we use for the operator the notation of Sect. According to Sect. the eigenstates of (8.8. m) in Sect. b+ .55) conﬁnes the particle to stationary bound states.309) yields ﬁnally Ψ (j. t) introduced above. m . is in the notation of the present section 1 ˆ a† a2 − a† a1 .83) and conﬁrmed the eigenvalues stated in (8. thus. 5.

as given in (8.105). one can estimate that the perturbation. i.108) We assume for the sake of simplicity that the eigenstates of Ho can be labeled through integers. t). i. i.105) as follows4 . 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. 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.105) for the case of radiationinduced transitions in atomic systems. In fact. For an electron charge q = −e and an electron mass m = me one can provide the estimate for the ﬁrst term of (8.105) in further calculations and to expand the wave function in terms of powers of VS (t) in a perturbation calculation.104) (8. can be considered a weak perturbation. is much smaller than the eigenvalue diﬀerences near ˆ typical atomic bound states.. 1. i.42)..216 Interaction of Radiation with Matter be expanded through Fourier analysis in terms of such plane waves. Estimate of the Magnitude of VS We want to demonstrate now that the interaction VS (t). However. in this case.. we discount the possibility of a continuum of eigenstates.e. . is much 2 (r. the term ∼ A second term in (8. we assume n|m = δnm and for the identity ∞ (8. the term ∼ p · A(r. . (8. t) + A (r.103) (8. this assumption can be waved as our results below will not depend on it. t) − m 2m (8. using (8.105) where A(r.e..41) e ˆ e p2 p·A ∼ 2me me me 2me 4 1 2 8πNω . (8.e.e. orthonormal basis. This result will allow us to neglect the larger than the second term. The states |n are thought to form a complete. We ﬁrst note. . 2 . n = 0.106) where we adopted the Dirac notation for the states of the quantum system.109) The reader should note that the estimates are very crude since we are establishing an order of magnitude estimate only. Here the so-called unperturbed system is governed by the Hamiltonian Ho with stationary states deﬁned through the eigenvalue problem Ho |n = n |n .107) 1 = 1 n=0 |n n| . t). ωV (8. and that the ﬁrst term in (8. t) is given by (8.

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. a volume V = λ3 where λ is the wave length corresponding to a plane wave with frequency ω.8.e. altogether.116) This term is obviously much smaller than the ﬁrst term. one obtains for (8. 2me (8. i. and me c2 ≈ A A 500 keV 2 ao ω ≈ 10−8 (8. replace the perturbation (8. hence. Assuming a single photon. Using again (8.115) (8.e.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. with ao ≈ 0.ˆ u 4πNk α(k. (8.105) for radiation ﬁelds can be considered a small perturbation.117) In case that such perturbation acts on an electron and is due to superpositions of planar waves described through the vector potential (8. and as long as the photon densities Nω /V are small. the ﬁrst term in (8. We can. λ = 2πc/λ. e ˆ p · A ∼ 10 eV · 10−4 = 10−3 eV .118) . i. one can neglect this term as long as the ﬁrst term gives non-vanishing contributions. (8.41) 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 .105) due to a radiation ﬁeld by VS = − q ˆ p · A(r..105).111) For ω = 3 eV and a corresponding λ = 4000 ˚ one obtains.112) yields e2 A2 ∼ 10 eV · 10−8 = 10−7 eV .110) where ao is the Bohr radius. We want to estimate now the second term in (8. t) . Hence.114) Employing for the second factor the estimate as stated in (8. me (8. m (8.112) π λ me c2 and with e2 /ao ≈ 27 eV. Consequently. u) p · u exp i(k · r − ωt) ˆ ˆ ˆ kV ..5 ˚. Nω = 1.42) it holds VS ≈ e m k.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.

. |n are deﬁned in (8.122) where |ΨD (to ) = |0 . u) has been added to describe eliptically or circularly polarized ˆ waves.42) through k = |k| = ω.121) (8. n = 0 at some later time t ≥ to .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. Equation (8.108) as the eigenstates of the unperturbed Hamiltonian Ho . 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 . |Ψ(to ) = |0 (8. Using i i ∂t exp − Ho (t − to ) and (8..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. Multiplying the latter equation by the operator exp i (8.127) i ∂t |ΨD (t) = VD (t) |ΨD (t) . (8.125) (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) . A factor α(k.120) (8. 2.124) (8.118) is the form of the perturbation which. The probability to ﬁnd the particle in the state |n at time t is then p0→n (t) = | n|ΨS (t) |2 . For this purpose one needs to determine the state |ΨS (t) of the particle. One seeks to predict the probability to observe the particle in one of the states |n . the sum over u involves the two possible polarizations of planar ˆ electromagnetic waves.123) i = Ho exp − Ho (t − to ) (8. 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 . The states |0 . n = 1. under ordinary circumstances.106–8. . 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 .

Employing this result one obtains for (8. = .137) can be solved recursively.128).139) . .134. Equations (8. n| (8.137) are consistent with (8.128) We note that the transition probability (8.137) One can readily verify that (8.127.5: Time-Dependent Perturbation Theory where VD (t) = exp 219 i i Ho (t − to ) VS (t) exp − Ho (t − to ) . . From (8. Due to the Hermitean property of the Hamiltonian Ho holds n|Ho = i n|exp − Ho (t − to ) from which we conclude. (8. 0 VD (t) |ΨD (t) (1) VD (t) |ΨD (t) (0) (8.130) = exp − i n (t − to ) − to )]| = 1. (8. .138) 1 i t dt VD (t ) |0 .129) n| and.135) i ∂t |ΨD (t) (n) VD (t) |ΨD (n−1) (t) (8.136) together with the initial conditions |ψD (to ) = |0 0 for n = 0 for n = 1.137) follows |ΨD (t) = |0 . 2. We will consider here only the two leading contributions to |ΨD (t) .133. 8.131) p0→n (t) = | n |ΨD (t) |2 .133) (8.121) expressed in terms of ΨD (t) is i p0→n (t) = | n|exp − Ho (t − to ) |ΨD (t) |2 .127) we assume the expansion ∞ |ΨD (t) = n=0 (n) |ΨD (t) (n) (8.134) (8. 8. (8. .133–8. In order to determine |ΨD (t) described through (8. 8. we deﬁne |ΨD (t) through the evolution equations i ∂t |ΨD (t) i ∂t |ΨD (t) i (2) ∂t |ΨD (t) (1) (0) = = = . using |exp[− i n (t n (8. . to (8. .132) where |ΨD (t) accounts for the contribution due to n-fold products of VD (t) to |ΨD (t) .8. .137) |ΨD (t) = (1) (0) (8.132–8. consequently. Accord(n) ingly.

137) |ΨD (t) = (2) Interaction of Radiation with Matter 1 i 2 t t dt to to dt VD (t ) VD (t ) |0 . + m=0 dt to to dt n|VD (t )|m m|VD (t ) |0 8. however.6 Perturbations due to Electromagnetic Radiation We had identiﬁed in Eq.142) combining an incoming and an incoming or outgoing wave.144) describes the propagation of the planar wave.141) + . 8.41). (8. that two planar waves simulataneously interact with an electronic system.34.118). according to (8. The coeﬃcients A1 . . the direction of ˆ the propagation being determined by k = k/|k|.e. In (8. A factor exp(λt). i. λ → 0+ .. in turn. non-singular. 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. The factor III in (8. such that the combined radiation ﬁeld is decribed by the vector potential A(r. We want to apply now the perturbation expansion derived to such perturbations. is VS = ˆ ˆ V1 exp(−iω1 t) + V2 exp( iω2 t) eλt . 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. 8. according to (8.220 This result. λ → 0+ has been introduced which describes that at time to → −∞ the perturbation is turned on gradually.135.143) 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 .144) r is the position of the electron and p = ( /i) is the momentum operator of the electron.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. The resulting perturbation on an electron system. to → −∞ (8. yields for (8.. (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). For the sake of including the eﬀect of superpositions of plane waves we will assume. II III (8.142).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. A2 are deﬁned through (8. This factor will serve mainly the purpose of keeping in the following derivation all mathematical quantities properly behaved.. note that uj . deﬁnes the direction of the E-ﬁeld of the ˆ radiation.

6: Perturbations due to Electromagnetic Radiation 1st Order Processes 221 We employ now the perturbation (8. 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. employing the perturbation (8.128).8.150) and. For the 1st order contribution to the transition amplitude n|ΨD (t) = (1) 1 i t dt n|VD (t ) |0 to (8. (8. using (8.148) We consider now the 2nd order contribution to the transition amplitude.141).130) and (for m = 0) i i exp − Ho (t − to ) |m = exp − for (8.143). 8. According to (8. (8.145) n|ΨD (t) (1) m (t − to ) |m .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 . (8.128) one obtains k|VD (t) | = k|VS (t)| exp i ( k − ) (8. (8.149) Using the deﬁnition of VD stated in (8.145) we obtain then.143) to the expansion (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 .140. (8.146) = 1 λ→0+ t→−∞ i lim lim × t dt exp to i ( n − o − i λ) t iω2 t × .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 .

.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 ) (8.222 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 (8.148) and ∗ |z1 + z2 |2 = |z1 |2 + |z2 |2 + 2Re(z1 z2 ) (1) (8. the time derivative of p0→n (t). Using (8.153) (8. respectively. (8.152).155) λ→0+ ( o We are actually interested in the transition rate.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. We also assume for the ﬁnal state n = 0 such that n|0 = 0 and p0→n (t) = | n|ΨD (t) |2 holds. i.148) and (8.e. provide now the transition probability p0→n (t) according to Eq.131). We assume ﬁrst that the ﬁrst order transi(1) tion amplitude n|ΨD (t) is non-zero. For this rate holds . in which case one can expect that it is larger than the 2nd (2) order contribution n|ΨD (t) which we will neglect.

i. First. due to the uncertainty relationship would introduce a considerable perturbation to the system.142–8. to cancel the third term in (8.144).e. one should average the rate over many periods of the radiation.31) and (8. in fact.8.158) play an essential role for the transition rates of radiative transitions. The result of such average is. The result supports our deﬁnition of incoming and outgoing waves in (8. The second contribution to (8. In case the matrix elements are non-zero.158) ˆ + | n|V2 |0 |2 δ( − ω2 ) Obviously.160) leads to ﬁnal states |n with energy n = o ± ω2 . ˆ (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. of this expression describe the individual eﬀects of the two planar wave contributions of the perturbation (8..144). (8. describing either an incoming or an outgoing plane wave due to the vector potential A2 u2 exp i (k1 · r ˆ ω2 t) . . which is observed through the so-called spectral intensities of transitions |0 → |n . the matrix elements can vary strongly for diﬀerent states |n of the system. any attempt to do so.156) such that the 1st order contributions of the two planar waves of the perturbation simply add.143. For the resulting expression the limit limλ→0+ can be taken.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. due to the vector potential A1 u1 exp i (k1 · r − ω1 t) .142) ˆ ˆ The matrix elements n|V1 |0 and n|V2 |0 in (8.159) leads to ﬁnal states |n with energy n = o + ω1 . a property. 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. The δ-functions appearing in this expression reﬂect energy conservation: the incoming plane wave contribution of (8. Using lim x2 + 2 δ→0+ = π δ(x) (8.156) ( o The period of electromagnetic radiation absorbed by electronic systems in atoms is of the order 10−17 s. The time average will be denoted by · · · t . 8.158). Hence. the two terms apearing on the rhs. however.

168) absorption of 2 photons ω1 .166) over many periods of the radiation yields exp[ i (±ω2 − ω1 )] t = 0 and.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. 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. i.153–8.164) (8. exp[ i (±ω2 − ω1 )]. ∞ − 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. similarly.g.158). Time average · · · t of expression (8.166) 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.153) vanishes such that the leading contribution to the transition probability p0→n (t) arises from the 2nd order amplitude (8.161) To determine the transition rate we proceed again.k=1 j=k 2 2 2 2 ∗ zj zk (8. in Eqs. (2) (8. it holds p0→n (t) = | n|ΨD (t) |2 .157) yields. as we did in the the case of 1st order transitions.163) × (8.158). We deﬁne ∞ z1 = m=0 ˆ ˆ n|V1 |m m|V1 |0 m − o − ω1 − i λ exp i × (8. hence..167) Taking now the limit limλ→0+ and using (8. i.162) × and.152).e. in analogy to (8.224 2nd Order Radiative Transitions Interaction of Radiation with Matter We now consider situations where the ﬁrst order transition amplitude in (8..e. |z1 + z2 + z3 |2 t = |z1 |2 + |z2 |2 + |z3 |2 (8. (8.. e.

describe the absorption/emission of two photons. this state is only virtually excited. accordingly. The factors | · · · |2 in (8. m = o + ω1 ) and the evolution of state |m is described by a factor 1/( m − o − ω1 −i λ). there is no energy conservation necessary (in general. interference between the contributions from all intermediate states |m can arise. In case of the ﬁrst term in (8.168) describe the time sequence of the two photon absorption/ emission processes. namely absorption/ emission ﬁrst of photon ω2 and then absorption of photon ω1 and. each of energy ω2 . describe the absorption of two photons. the second term describes the processes leading to ﬁnal states |n with energy n = o ± 2 ω2 and. i..e. For this purpose we will employ the transition rate as given in Eq. Most remarkably. (8.168) describe similar histories of the excitation process.. each of energy ω1 . describes processes which lead to ﬁnal states |n with energy n = o + 2 ω1 and. accordingly. 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 .e. i. Similarly. The remaining two contributions in (8.e..8.7: 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. promotes the system then to the state |n .e. i. which is stationary and energy is conserved. |n ← |m . the third term in (8. The expression sums over all possible virtually occupied states |m and takes the absolute value of this sum.158) which accounts for such transitions.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. i. through absorption of a photon.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. Similarly.. The ﬁrst term. ﬁrst absorption of photon ω1 and then absorption/ emission of photon ω2 . .168) describes two intermediate histories. the third term describes processes in which a photon of energy ω1 is absorbed and a second photon of energy ω2 is absorbed/emitted. a second perturbation. according to its δ-function factor. |m ← |0 (8.169) pert. 8.

i. m denote the relevant quantum numbers) ψn=1. in this range one can expand eik·r ≈ 1 + ik · r + . and the integral is =1. the vector potential actually assumed is A(r. θ.170) We will employ only the real part of this potential. λ → 0+ .144) VS = where e 2πN ˆ ˆ p · u e±ik·r .176) ψn=2. 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 .2 = m ωV Only the ﬁrst term of (8. to → −∞ (8. . ˆ (8. the wave functions are (n. t) = 8πN ωV u exp ˆ ı (k · r − ωt) . (8. φ) a3 ao o 2p (8. . plane polarized wave described through the complex vector potential A(r. t) = 2πN ωV u exp ˆ ı (k · r − ωt) + 2πN ωV u exp ˆ ı (−k · r + ωt) . For example.. . according to (8.178) These wave functions make signiﬁcant contributions to this integral only for r-values in the range r < 10 ao .143. φ) a3 o 1s (8. 8.177) √ M = 6 0 ∞ 1 2π ia4 o r2 dr −1 dcosθ 0 ∗ dφ r e−r/2ao Y1m (θ.173) V1.175) The matrix element involves a spatial integral over the electronic wave functions associated with states |n and |0 .e.m=0 (r. (8.158). (8.143) will contribute to the absorption process.174) We seek to evaluate the matrix element ˆ M = n| p eik·r |0 .172) δ( n − o − ω) (8. φ) × × eik·r e−r/ao Y00 (θ.226 Absorption of a Plane Polarized Wave Interaction of Radiation with Matter We consider ﬁrst the case of absorption of a monochromatic. θ.179) .m (r. (8.171) The perturbation on an atomic electron system is then according to (8. φ) = 6 r −r/2ao e Y1m (θ. The absorption rate. the second term can be discounted in case of absorption. φ) (8. However. =0. φ) = 2 − 1 2 1 −r/ao e Y00 (θ. in case of a radiative transition from the 1s state of hydrogen to one of its three 2p states.

104) and.184) and the commutation property [xk .186) δ( n − o − ω) (8.174) yields kabs = 4π 2 e2 N ω V u · n| ˆ |0 ˆ r 2 m( o − m n| [r. This simpliﬁcation results from the identity m ˆ p = [ r.182) one ﬁnds [ r. pk ] pk ˆ k=1 (8. We are now in a position to obtain an alternative expression for the matrix element (8. Using (8.8. Ho ] i where Ho is the Hamiltonian given by (8. ] + [ r. is Ho = ˆ (p)2 e2 + V (r) . V (r) ] 2me =0 = Using r = 3 ˆ j=1 xj ej 1 2me 3 pk [ r. V (r) = − . pk ] + ˆ ˆ k=1 1 2me 3 [ r.182) one obtains M ≈ Insertion into (8. Ho ] = i m 3 pk ej δjk = ˆ j.529 ˚ one concludes that in the signiﬁcant integration range in (8.k=1 i m 3 pk ek = ˆ j.187) . One refers to this approximation as the dipole approximation. 2me r (8.185) from which follows (8. Transition Dipole Moment A further simpliﬁcation of the matrix element (8. the value of the wave length for the 1s → 2p transition 2π c λ = = 1216 ˚ A (8.183) (8.178) holds eik·r ≈ A 1 1 + O( 50 ) such that one can approximate eik·r ≈ 1 .175) can then be ˆ achieved and the diﬀerential operator p = i replaced by by the simpler multiplicative operator r.175).180) ∆E2p−1s and ao = 0. Using |k| = 2π/λ.181) and (8. in case of the hydrogen atom.7: One-Photon Emission and Absorption 227 One can estimate that the absolute magnitude of the second term in (8.181) For the commutator in (8. pj ] = i δkj one obtains ˆ [ r.182) (8. (8.k=1 i ˆ p m (8.179) and other terms are never larger than 20π ao /λ.182). Ho ] |0 = i i n) n|r |0 . Ho ] = ˆ2 p [ r.

i. directions.35). Integrating and summing accordingly over all contributions as given by (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. according to (8. u2 . there are ony two linearly independent directions of u possible. planar waves as deﬁned in (8..193) . |0 . and (2) the fact that strictly monochromatic radiation cannot be prepared such that any radiation source provides radiation with a frequency distribution. will actually be replaced by a distribution function which reﬂects (1) the ﬁnite life time of the states |n . such that the thermal radiation present supplies a continuum of frequencies. ϕ1 = 0 and of u2 by ϑ2 = π/2. k ˆ As a result. k frame is described by the angles ϑ. 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.188) ˆ Here dk is the integral over all orientations of k.189) where the factor 1/ arose from the integral over the δ-function.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. say u1 and u2 .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. ϕ.e.33) are orthogonal to each other. three-dimensional vector. x2 . one can express | u · n| r |0 |2 = |ρ|2 ( cos2 α + cos2 β ) = sin2 θ .31–8. We can. The δ-function appearing in this expression. We have demonstrated in our derivation of the rate of one-photon processes (8. x3 -axes of a right-handed cartesian vectors u1 . ρ) and ˆ u β = ∠(ˆ2 . The direction of this vector in the u1 . hence.190) (8.174) one can replace n − o by ω. The unit ˆ ˆ ˆ ˆ can be chosen to point along the x1 .228 Interaction of Radiation with Matter where we used the fact that due to the δ-function factor in (8. u2 and k ˆ ˆ coordinate system. and all polarizations of the radiation. obtain the transition rate in the present case by adding the individual transition rates of all planar waves present in the oven. in practical situations. ˆ In order to carry out the integral dk we note that u describes the possible polarizations of the ˆ ˆ and u. ˆ u ˆ (8.158) above that in ﬁrst order the contributions of all components of the radiation ﬁeld add. For the two angles α = ∠(ˆ1 . the direction of u1 is described by the ˆ ˆ ˆ ˆ angles ϑ1 = π/2. Absorption of Thermal Radiation We want to assume now that the hydrogen atom is placed in an evironment which is suﬃciently hot. ϕ2 = π/2. ρ) holds then u cosα = cosϑ1 cosϑ + sinϑ1 sinϑ cos(ϕ1 − ϕ) = sinϑcosϕ and cosβ = cosϑ2 cosϑ + sinϑ2 sinϑ cos(ϕ2 − ϕ) = sinϑsinϕ . Accordingly. a very hot oven.191) (8.

173). 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. as for the absorption process.. The quantum nature of radiation manifests itself in that the number of photons Nω msut be an integer. 2 . Nω photons before absorption. Transition Dipole Moment The expression (8. However.172. For practical evaluations we provide an expression which eliminates the physical constants and allows one to determine numerical values readily. . the calculation of the emission rate proceeds as in the case of absorption. This dependence predicts. The value of | n| r |0 | determines the strength of an optical transition. the absorption rate is of the order of 109 s−1 or 1/nanosecond. ˆ In case of emission only the second term V2 exp(+iωt) in (8.. for atoms or symmetric molecules. However. This diﬀerence between emission and absorption is due to the quantum nature of the radiation ﬁeld.e.37 × 1019 1 ao | n|r|0 |2 × s λ λ2 . i.e. Emission of Radiation We now consider the rate of emission of a photon.194) combined are typically somewhat smaller than (1 ˚/1000 ˚)3 = A A 10−9 .171) and perturbation (8.197) o (tot) (8. a problem in case of emission by quantum systems in complete darkness. there is also a contribution to the emission rate which is proportional to Nω .37 × 1019 × 3 3 c3 3 ao λ3 s λ and kabs where λ = 2πc n − (8. The corresponding process is termed spontaneous emission. 1. .189) to yield the total absorption rate kabs (tot) = Nω 4 e2 ω 3 | n|r|0 |2 3 c3 .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 . by planar waves with vector potential (8. The most intensely absorbing molecules are long. The transition dipole moment can vanish for many transitions between stationary states of a quantum system. For this purpose we use ω/c = 2π/λ and obtain 4 e2 ω 3 32π 3 e2 ao 1 ao = = 1. .196) The last two factors in (8.195) = Nω 1. however. Accordingly. i. for Nω = 0. However. Otherwise. The radiation ﬁeld is described. 8. (8. (8. that emission occurs even if no photon is present in the environment.173) contributes. ﬁnally. Nω = 0. This poses. linear molecules.8. In case of a classical radiation ﬁeld one would expect that emission cannot occur. can be inserted into (8.194) 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 |. the resulting total rate of emission bears a diﬀerent dependence on the number of photons present in the environment. in agreement with observations..7: One-Photon Emission and Absorption 229 This geometrical average. in particular.

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. exp[− ω/kB T ] = This equation yields Nω = 1 . 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.200) where kB is the Boltzmann constant. 8..8 Two-Photon Processes In many important processes induced by interactions between radiation and matter two or more photons participate. exp[ ω/kB T ] − 1 (8.204) Nω . using (8. Deﬁning the rate of spontaneous emission 4 e2 ω 3 | n|r|0 |2 (8. the well-known Planck radiation formula.203) i.g. Hence. = (8.200). For these numbers holds Nn / No = exp[−( n − o )/kB T ] (8. Nω + 1 (8.230 Interaction of Radiation with Matter which is termed induced emission since it can be induced through radiation provided. accordingly. respectively.202) It follows.198) to determine the stationary distribution of photons ω in an oven of temperature T . .195) and (8.198) (8.. The total rate of emission.e.201) ksp = 3 c3 the rates of absorption and emission are Nω ksp and (Nω + 1)ksp . in lasers. 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 . it must hold Nω ksp N0 = (Nω + 1) ksp Nn (8. The number of atoms undergoing absorption in unit time are Nω ksp No and undergoing emission are (Nω +1)ksp Nn . To demonstrate this property we apply the transition rates (8.198) is consistent with Planck’s radiation law which reﬂects the (boson) quantum nature of the radiation ﬁeld. We assume n − o = ω. respectively.199) Planck’s Radiation Law The postulate of the Nω + 1 dependence of the rate of emission as given in (8. e. In the following we discuss several examples. Let No and Nn denote the number of atoms in state |0 and |n .

however. however. which is not necessarily the case. 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 × (8.106). a transition between the states |0 and |n may be possible. If the sum in (8. 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) . In applying (8. The electron absorbs the radiation and emits immediately a second photon.208) has been prepared. 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. The transition requires that the transition dipole moment n| r |0 does not vanish for two states |0 and |n .205) The respective radiative transition is of 2nd order as described by the transition rate (8. 8. then expression (8. The intense radiation ﬁelds of lasers allow one to increase transition rates to levels which can readily be observed in the laboratory. as described through the vector potential ˆ A(r.207) one is. ﬁnally. The resulting rate of the transition depends on 2 Nω . (8. t) = Ao1 u1 cos(k1 · r − ω1 t) . but then requires the absorption of two photons.8.207) Expression (8.187). .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.172. Employing the dipole approximation (8. (8. under ordinary circumstances.8: Two-Photon Processes Two-Photon Absorption 231 The interaction of electrons with radiation.173). induce single photon absorption processes as described by the transition rate Eq.104) with stationary states |n deﬁned through (8.206) − o − 2 ω) . In this case one needs to choose the energy of the photons to obey n = o + 2 ω. even if n| r |0 vanishes. 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. dropping the index 1 characterizing the radiation. in case ˆ of absorption only V1 contributes.182) yields. faced with the dilemma of having to sum over all intermediate states |m of the system.168) where the ﬁrst term describes the relevant contribution. One obtains. 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 (8.207) does not provide a suitable avenue of computing the rates of 2-photon transitions.207) does not converge rapidly. However. We assume that a planar wave with wave vector k1 and polarization u1 . ˆ (8. (8.181) and using (8.

6 above that the absorption of the radiation in (8. ˆ The perturbation which arises due to the vector potential (8. (8. u1 .208).208) and the emission of the radiation in (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 . t) = Ao2 u2 cos(k2 · r − ω2 t) .e.141). 8. in the present case.210) is accounted for by the following contributions of (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 (8.211) 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. t) = A+ u1 exp[ i (k1 · r − ω1 t) ] + A− u2 exp[ i (k2 · r − ω2 t) ] .210). for the components of the wave (8. In the present case we consider only scattering processes which absorb radiation corresponding to the vector potential (8.212) where N1 /V is the density of photons for the wave described by (8. i.208) and emit radiation corresponding to the vector potential (8.105).210).209) cosθ2 where the value of k2 has been ﬁxed. the wave characterized through k1 . o1 ˆ o2 ˆ (8.214) . The vector potential describing the emitted plane wave is then A(r. The second term in (8.210) A(r. The relevant terms of the perturbation (8. later we will allow the quantum system to select appropriate values.211) is stated in Eq. the amplititude A− deﬁned in (8.208.210) The vector potential which describes both incoming wave and outgoing wave is a superposition of the potentials in (8. the amplitude A+ is given by o1 A+ = o1 8πN1 ω1 V (8. according to the ˆ description of emission processes on page 229.208.211) is o2 A− = o2 8π (N2 + 1) ω1 V (8. hence. u2 .105) using the vector potantial (8. 8.213) where N2 /V is the density of photons characterized through k2 .211) The ﬁrst term describes the absorption of a photon and. This expansion yields.. ˆ (8. We know already from our description in Section 8. however.211) accounts for the emitted wave and.

215) correspond to scattering processes in which the radiation ﬁeld ‘looses’ a photon ω1 and ‘gains’ a photon ω2 .181) in stating this result..218) . one needs to carry out the integration 2 V(2π)−3 k2 dk2 · · ·. i.e.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 . only these two terms contribute to the scattering amplitude. 8. Inserting also the values (8. Hence. me c2 (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..8 · 10−15 m . Only the second (1st order) and the third (2nd order) terms in (8.212. evaluating the time integrals in (8.e.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. one needs to integrate the rate as given by (8.8: 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 ∞ (8.215) 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. Accordingly.217) 2 Here ro denotes the classical electron radius ro = e2 = 2.216) over all available modes of the ﬁeld.8. Following closely the procedures adopted in evaluating the rates of 1st order and 2nd order radiative transitions on page 222–225. i.

a process in which the electronic state remains unaltered after the scattering. one can relate (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 + . in fact.. ω1 = ω2 . − o )2 ( m − o )3 (8. it follows 2 dσ = ro (N2 + 1) dΩ2 |ˆ1 · u2 − S( ω) |2 u ˆ (8.e. 2 . Rayleigh Scattering We turn ﬁrst to an example of so-called elastic scattering. Accordingly. (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. .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 (8.228) S (0) = m S (0) = 2 m ..223) m Condition (8. for all states |m of the electronic system (8.217) 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 (8.219) (8. i. one assumes the even stronger condition ω1 << | o − m| .221) Using |n = |0 and.. consequently. 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 (8.234 Interaction of Radiation with Matter The factor N1 c/V can be interpreted as the ﬂux of incoming photons. .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 (8.224) = m 1 − o ( m ω ( ω)2 + + .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.

228) provides then the ﬁrst non-vanishing contribution to the scattering cross section (8.233) m Employing again (8.231) (ˆ1 )j (ˆ2 )k 0|[xj .222) of the cross section for Rayleigh scattering dσ = 2 ro m2 ω 4 (N2 + 1) dΩ2 × e (8. ˆ ˆ S (0) given in (8. i.182) 0|ˆ1 · p |m u ˆ 1 = 0|ˆ1 ·r |m u me ( m − o ) i This transforms (8.235) and obtain the leading contribution to the expression (8. for example.234) where we used for the second identity the fact that u1 ·r and u2 ·r commute.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.k=1 j.222). u ˆ ˆ ˆ ˆˆ i The commutator property [xj . as a compass.. For this purpose we apply (8.236) is of great practical importance. Expression (8. (8. honey bees.222). u u u u (8. It explains.108) yields S (0) = me 2 0| [ˆ2 ·r.227) vanishes.235) We can now combine eqs. pk ]|0 = u u ˆ j.e. 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 .8.232) Obviously.229) (8. . ˆ ˆ We want to prove now that expression (8. the blue color of the sky and the polarization pattern in the sky which serves many animals. ˆ We want to simplify ﬁrst (8. Using again (8. m|ˆ1 · p |0 u ˆ 1 = − m|ˆ1 ·r |0 u me ( m − o ) i (8.230) According to (8.k=1 (ˆ1 )j (ˆ2 )k δjk = u1 · u2 u u ˆ ˆ (8.232.108) this is 1 0|ˆ1 ·r u2 · p − u2 · p u1 ·r |0 .224.234.229) 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 ) .182) for p and the expression (8.8: Two-Photon Processes 235 ˆ These three expressions can be simpliﬁed using the expression (8.226). For this purpose we replace p using (8. 8. u1 ·r ] |0 = 0 u ˆ (8.108) for the identity operator. this term cancels the u1 · u2 term in (8. 8.236) 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. pk ] = i δjk yields ﬁnally ˆ S(0) = 1 S(0) = i 3 3 (8.229) both for the u1 · p and the to u2 · p terms in (8.

243) cos2 θ2 cos2 φ2 for u1 = u1 . (8. 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 .236 Thomson Scattering Interaction of Radiation with Matter We consider again elastic scattering. however. u2 = u2 ˆ ˆ ˆ ˆ (8.238) (8.240) cosθ2 We choose for the polarization of the incoming radiation the directions along the x1 .220). 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 .239) We will show below that this expression decribes the non-relativistic limit of Compton scattering.237) The resulting scattering is called Thomson scattering. though. i. the emitted u ˆ radiation is decribed by the wave vector sinθ2 cosφ2 k2 = k1 sinθ2 sinφ2 (8. The ﬁrst choice is ˆ sinφ2 k2 × k1 (1) u2 = ˆ = −cosφ2 (8. ao Bohr radius . |n = |0 and ω1 = ω2 = ω in (8.e. We want to assume. u ˆ 1 ao .242) |k2 × k1 | 0 where the second identity follows readily from k1 = e3 and from (8.. for all states |m of the electronic system .. hence.220) 2 dσ = ro (N2 + 1) dΩ2 |ˆ1 · u2 |2 . u2 = u2 ˆ ˆ ˆ ˆ for u1 = u1 . that the dipole approximation is still valid which restricts the applicability of the following derivation to k1 >> One obtains immediately from (8. u1 = 1 ˆ ˆ (8. u2 = u2 ˆ ˆ ˆ ˆ cos2 φ2 cos2 θ2 sin2 φ2 for u1 = u1 . u2 = u2 ˆ ˆ ˆ ˆ (1) (2) (2) (1) (1) (2) (1) (2) (2) (1) needs to be (8.241) 0 0 Similarly. assume now that the scattered radiation has very short wave length such that ω >> | o − m| .and the x2 -axes 1 0 (1) (2) u1 = 0 . To evaluate |ˆ1 · u2 |2 we assume that k1 is oriented along the x3 -axis and.240). (8.244) .

φ(vibr.246) 2 This expression is the non-relativistic limit of the cross section of Compton scattering. 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. or in case of translational motion of molecules in liquids. molecular vibrations or crystal vibrational modes.e. The energy deﬁcit is used to excite the system. the scattered radiation is polarized to some degree. as in the case of excitations of accustical modes of crystals. In the case that the scattering excites other than electronic degrees of freedom.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. one speaks of anti-Stokes scattering. the scattering is termed Brillouin scattering. but most often involves other degrees of freedom. one speaks of Stokes scattering. the ﬁrst term in (8. For electronic systems in molecules or crystals the degrees of freedom excited are nuclear motions. 8. ˆ In case that one measures the scattered radiation irrespective of its polarization.)n . in case of a diatomic molecule |n = |φ(elect.. if energy is released..e.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 . In case that the nuclear degrees of freedom excited absorb very little energy. The radiation scattered at right angles is even completely polarized (1) along the u2 -direction. i. for c → ∞ the Compton scattering cross section (8. e. Raman Scattering and Brillouin Scattering We now consider ineleastic scattering described by the Kramers-Heisenberg formula.248) me c2 One can readily show that in the non-relativistic limit.220) represent actually electronic as well as nuclear motions. the states |n etc.246).g. If energy is absorbed by the system.220) vanishes and one obtains in case of Raman scattering ω2 2 dΩ2 | u2 · R · u1 |2 ˆ ˆ (8. Such scattering is called Raman scattering.)n . (8. the resulting scattering cross section is 2 ro (N2 + 1) ( 1 + cos2 θ2 ) dΩ2 . In the case of such scattering an electron system absorbs and re-emits radiation without ending up in the initial state.247) becomes identical with the Thomson scattering cross section (8..247.247) ( 1 − cosθ2 ) (8. i. 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. The excitation can be electronic.8. Since the scattering is inelastic.249) dσ = ro (N2 + 1) ω1 −1 −1 ω2 − ω1 = . deﬁned in (8.

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 )/ (8.251) ω1 − ( We deﬁne x · R · y = j. a case called resonant Raman scattering. 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 .k xj Rjk yk .250) (8. In case that the incoming photon energy ω1 is chosen to match one of the electronic excitations. e. the cross section for resonant Raman scattering can be several orders of magnitude larger than that of ordinary Raman scattering. Of course.g. . ω1 ≈ m − o for a particular state |m . a property which can be exploited to selectively probe suitable molecules of low concentration in bulk matter.

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

and explicitly the fact that any P ∈ SN can be given as a product of transpositions T (j. leaving all other particles unchanged. . even though these properties. not any state can be chosen as to represent the many– particle system. . . 5) (9. . A reader interested in the quantum mechanical description of N particle systems is strongly encouraged to study these representations. We will not discuss here at any depth the properties of the permutation groups SN . rather only states which obey a certain transformation property under permutations are acceptable.4) both permutations being obviously odd. xj . . k). i. k) = (4. Transpositions denoted by T (j. xP (2) . Denoting by P (j) the image of j. is called a transposition. (j.. 5) in case of P2 . and the second row showing the numbers 1 to N again. one can state P Ψ(x1 . (9. . xP (N ) ) . . 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 = . For the following we will require only two properties of SN . P2 = T (4. The permutations we need to consider for a system of N particles are the elements of the group SN . . One calls permutations of the ﬁrst type even permutations and permutations of the second type odd permutations. . the representations in the space of N –particle wave functions are extremely useful in dealing with N –particle systems. xP (j) . . However. xN ) = Ψ(xP (1) . . ..e. .240 Many–Particle Systems repeatedly like degenerate states. . 6) T (6. the set of all permutations of N objects. .3) P2 which aﬀects only two particles. i. . P2 = . This property will be established now. 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. but in a diﬀerent order. . 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. For the permutations P1 and P2 given above holds P1 = T (3. namely implicitly the group property. i. the top row representing the numbers 1 to N labelling the N particles under consideration. The latter factorization has the essential property that the number of factors is either even or odd. particles 4 and 5. in particular. in notation (9.3) the j-th index in the second row. .e. 4) T (4. k) are characterized by the two indices (j. We can then state P2 = T (4.e. 5). i. 3) T (7. k) of the two particles transposed.5) . x2 .. (9. The group SN is then composed of two disjunct classes of even and of odd permutations.e. a given P can either only be presented by even numbers of transpositions or by odd numbers of transpositions. number k under the entry j of the ﬁrst row indicating that particle j is switched with particle k. 5) T (5.. The elements P ∈ SN can be represented conveniently as 2 × N matrices.2). . . 8) .

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

. . .. . . . .15) completely and. . (9. λ2 . . .. x2 . N x1 . is a particular choice of placing N particles consistent with the occupation numbers (n1 . n2 .8. . λj ∈ {1.g. k) ∈ Λ(λ1 . The reader is well advised to pause and grasp the deﬁnition of the nj . This detail is connected with the fact that a two–particle state of the type ΨFock (λ. . . . . . . xN |Ψ Fock (λ1 . each particle j into a speciﬁc state |λj . λ2 . λ2 . The integers nj are equal to the number of λk in (9. . λN )|ΨFock (λ1 . λN ) do not exhibit the symmetries dictated by (9. . S} of the type (9. therefore. The numbers nj . i. nS ) = S j=1 (9. .. .13) These states form an orthonormal basis of N -particle states. 9. . λN ) only if λj = λk . |λN . the so-called Fock-states. . . An important detail of the deﬁnition (9. which do not yet obey the symmetries (9. the nj specify how often a single particle state |j is occupied. the reason is that such transposition duplicates the state which would. 9. λN ) ⊂ SN is the set of all permutations involving only particles j with diﬀerent λj .15) (nj !) N! 1 2 P ∈Λ(λ1 . xN |ΨFock (λ1 .12). . k.. λ2 . . 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. . λN ) = j=1 δλj λj . .9). . x2 . In a second step we form linear combinations of Fock states with the desired symmetries. xN |ΨB (n1 . . . . . are essential. with λj = λk = .11. . . . 2. characterize the wave function (9.9). |λ2 . . T (j. 9. . (9. . .14) Obviously. . the states |ΨFock (λ1 .15). . . referred to as the occupation numbers. A wave function which obeys the boson symmetry (9. . . . e.e. Λ(λ1 . . .e. The Fock states represent N particles which are placed into S states |λ1 . .λN ) Here the states (λ1 . . .λN ) P x1 .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 (9.15) with λk = j. appear severalfold in (9. . .. λ) in which two particles occupy the same single– particle state |λ does not allow transposition of particles. . nS ). . x2 . .12) where the second factor represents the scalar product between spin states. We ﬁrst consider orthonormal many–particle wave functions. i. .λN ) on the rhs. .8) is x1 . . hence. . . λ2 .8. λ2 . n2 . It holds N ΨFock (λ1 . . .15) is that the sum over permutations does not involve permutations among indices j. λN ) = j=1 φλj (xj ) .

for two sets of occupation numbers N = (n1 . correspond to a particular choice of placing N particles consistent with the occupation numbers (n1 . . √1 N! (9. a molecule or a crystal. . The fermion states (9.. . . (9. n2 . n2 .e. .16) Exercise 9. x2 . . n2 . φλN (xN ) Here. . i.1. . . n2 . . These states form an orthonormal basis. nS ) = φλ1 (x1 ) φλ1 (x2 ) .18) this wave function can also be expressed through the so-called Slater determinant x1 . xN |ΨF (n1 . . the photons of the electromagnetic ﬁeld or the phonons in a crystal. .λN ) on the rhs.15) obey the boson symmetry (9.8). for two sets of occupation numbers N = (n1 . for example. xN |ΨF (n1 . .e. . . . . . . . (9. . . .9).. for example. . nS ) holds S Ψ (N )|Ψ (N ) = j=1 B B δnj nj .15) describe. . . .. . .15) holds (9. .17) describe. nS ) = √1 N! P ∈SN P P x1 . φλN (x1 ) φλN (x2 ) . φλ2 (x1 ) φλ2 (x2 ) .19) φλ1 (xN ) φλ2 (xN ) . nS ) and N = (n1 . .λN ) (9. Wave Functions with Fermion Symmetry in the Occupation Number Representation One can construct similarly a wave function which obeys the fermion symmetry property (9..9. . . n2 . nS ) holds S ΨF (N )|ΨF (N ) = j=1 δnj nj . . Using the identity governing the determinant of N × N –matrices A N det(A) = P ∈SN P j=1 Aj P (j) (9. . nS ).20) .. . Such wave function is x1 . . The important property holds that nj can only assume the two values nj = 0 or nj = 1. i. . nS ) and N = (n1 . . . λ2 . .1: Bosons and Fermions 243 The wave functions (9. electrons in an atom. These states form an orthonormal basis. n2 . . x2 . . the integers nj denote the occupancy of the single–particle state |j . . .1: (a) Show that the states (9. (b) Show that for the states (9. . .17) Here the states (λ1 . n2 . xN |ΨFock (λ1 . . x2 . . . . . . λ2 . For any value nj > 1 two or more of the columns of the Slater matrix are identical and the wave function vanishes. .16). . . .. .

. . a† ] = 0 j k (9. . . n2 . . . nS ). . nj + 1. (b) Show that the states (9. Employing the superindices N and N introduced above and using the orthonormality property (9. . n ) are normalΨ 1 j 1 j S S ized according to (9. . . n ) and the N − 1-particle wave function ΨB (n .17) obey the orthonormality property (9. nj .16) one obtains (Aj )N = ΨB (N )| aj |ΨB (N ) = √ S N nj δnj nj −1 k=1 k=j δnk nk . . These operators allow one to construct and manipulate many–particle wave functions while preserving permutation symmetry. .15). .22) The operators aj and a† obey the commutation properties j [aj . . (9. . . . .17) obey the fermion symmetry (9. . . n − 1.23) (9. [a† . . . 9.21) √ The factor nj appears here on the rhs. since both the N -particle wave function B (n .9). . . To prove that (9. . Exercise 9. . . . . . a† ] = δj k . . nj . nS ) .24) [aj . .22) follows from (9. .21) we consider the matrix Aj corresponding to aj in the basis of many–particle states (9. (9. . . . 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 .2 Operators of 2nd Quantization An important tool to describe mathematically systems of many bosons or fermions. . . . nS ) = j nj + 1 ΨB (n1 .15) and (9. ak ] = 0 . are the so-called operators of 2nd quantization. . . n . . .2: (a) Show that the states (9. .25) . Creation and Annihilation Operators for Bosons We consider the operator aj deﬁned through aj ΨB (n1 . nj − 1. . nS ) = nj ΨB (n1 . . .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 . . (9.20). guaranteeing the correct permutation properties of the many–particle states. nS ) nj ≥ 1 0 nj = 0 √ . . .15).244 Many–Particle Systems The representation of wave functions (9.1. For the adjoint operator a† holds j a† ΨB (n1 . . .

We will prove the commutation properties (9.23) follow in an analogous way.29) |ΨB (n1 . . nS ) = j=1 a† j nj nj ! |0 . nS ) N |ΨB (n1 . . the eigenvalues of Nj are the occupation numbers nj . .24). . .24). The related operator S ˆ N = j=1 ˆ Nj (9.33) is called the particle number operator since. . obviously. . One can readily show using (9. nk + 1 . 9. The creation operators a† allow one to construct boson states ΨB (N ) from the vacuum state j |0 = |ΨB (n1 = 0. . .32) ˆ ˆ i.27) Equations (9. . n2 .28) (9.21. i. . for basis states |ΨB (n1 . . n2 . . .30) Of particular interest is the operator ˆ Nj = a† aj .34) = . nS = 0) as follows S (9. nj − 1 . (9. (9. 9. .2: Operators of 2nd Quantization Let A† be the matrix adjoint to Aj . . . (9. . nS ) = |ΨB (N ) . 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 . . .22) ˆ Nj |ΨB (N ) = nj |ΨB (N ) . (9. It holds for j = k (aj a† − a† aj )|Ψ(N ) = (nj + 1 − nj )|Ψ(N ) = |Ψ(N ) .21) and vice versa. nS ) = j=1 nj |ΨB (n1 .22) follows from (9. .29) yield the commutation relationship (9. j (aj a† − a† aj )|Ψ(N ) = k k ( nj (nk + 1) − (nk + 1)nj )|Ψ(n1 . nS ) . . n2 . nS ) = 0 . j j Since this holds for any |Ψ(N ) it follows [aj ..9.31) This operator is diagonal in the occupation number representation. . . (9.27. . .26) From this result one can immediately conclude that (9. One refers to Nj as the occupation number operator. j (9. n2 . S ˆ N |ΨB (n1 . n2 .. a† ] = 1 For j = k follows similarly 1. . . the properties (9. . .e. . n2 = 0.e.

. in general.24) and aj |0 = 0 imply that the the properties (9. nj+1 . . . . . operators which can alter the occupancy of the wave function (9. showing that the property holds for nj = 0 and. . one does not need to worry ever about proper symmetries of wave functions.35) and where the derivative operation is deﬁned through ∂f (A) = ∂A ∞ n=1 1 f (n) (0) An−1 .. . .9). 9. f (a† )] j [a† . nj (9.21. Exercise 9. .2.. . nS ) . xN |ΨF (n1 .e. . .39) .23. j Exercise 9. x2 . xN |ΨF (n1 .17.h.1: The commutation relationships (9.29. 9. . . it also holds for nj + 1. . nj .37) (9. x2 . the fermion wave function changes sign when one exchanges the order of the occupancy.23) hold for the state deﬁned through (9. (n − 1)! (9.24) and the property aj |0 = 0. from the vacuum state |0 or states |ΨB (N ) . Before proceeding in this respect we need to account for the following property of the fermion wave function which applies in the case nj .. i. Prove this by induction.38) . nj+1 = 1. . . . xN |ΨF (n 1 .. nj+1 . nj . 9.19) without aﬀecting the fermion symmetry (9. .. . . φj+1 (x1 ) φj+1 (x2 ) . . . nS ) = .30). .g.23. . .2. . nS ) = − x1 . To prove this property we notice that the l. 9.36) Creation and Annihilation Operators for Fermions We want to consider now creation and annihilation operators for fermions. φj (xN ) φj+1 (xN ) .. φj (x1 ) φj (x2 ) . . of (9.246 Many–Particle Systems In using the boson creation and annihilation operators. nj+1 . one needs to apply only the commutation propertes (9.. Obviously.22. . x1 . . As long as one starts from a wave function with the proper boson symmetry. . . since they are induced through the algebra of a† and aj . in case that the single particle states |j and |j + 1 are both occupied. 9. √1 N! . i. . e.. . .38) corresponds to x1 . 9. . . x2 . .. . . in case it holds for nj .s. . (9. . . .2: Show that the boson operators a† and aj satisfy j [aj .e.. 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.

. φj (x1 ) φj (x2 ) .. at the ﬁrst column of the Slater determinant (9. however.9..e. nj+1 . xN |ΨF (n1 . . . . n2 . n5 = 1. . n2 = 1. in order to be consistent with this convention. nj = 0. . .s. One adopts the convention that particles are created and annihilated by these operators always at the ﬁrst position of the wave function. nj . .. . n3 .2: Operators of 2nd Quantization The r. (9.19). nS ) qj states occupied In case that the single particle state |j is not occupied. . n4 = 0. (9. n2 . nj .. 247 − √1 ! N . . . . n2 . This change of position is connected with a possible sign change (−1)qj where qj is deﬁned for a given N = (n1 . i. We are now ready to deﬁne the annihilation operator cj for fermions in the single–particle state |j as follows cj |ΨF (n1 . nS ) = nj (−1)qj |ΨF ( n1 . . 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. . (9. nS ) as follows: j−1 qj = k=1 nk . . Obviously. . it is necessary to specify for a fermion wave function the order in which the single– particle states |λj are occupied. vanishes. We will be adopting below the notational convention that qj corresponds to occupancies N = (n1 . . nS ) = . . . . . φj+1 (x1 ) φj+1 (x2 ) . . A proper example is the two particle fermion wave function x1 . of (9. the rhs. at which column of the Slater determinant (9. . ... . nj → 0... i. x2 |ΨF (n1 = 0.39) and (9. . particles are being created or annihilated. .. For this purpose one must adhere to a strict convention: the labelling of single–particle states by indices j = 1. 2. .40) are identical. the expressions (9. x2 . .19). . . nS = 0) φ2 (x1 ) φ5 (x1 ) 1 1 = √2 = √2 [φ2 (x1 )φ5 (x2 ) − φ5 (x1 )φ2 (x2 )] .e.h. . .e.. that occupancies are always ordered according to a monotonous increase of the single–particle state index. . . nj+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). n3 = 0. φj+1 (xN ) φj (xN ) . . . .40) . i.e. nj−1 . n2 . .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. qj is the number of states |k with k < j which are occupied in a fermion wave function. . This requires one. . . n6 = n7 = . nS ). Because of the property of the determinant to change sign when two columns are interchanged.. . .38) reads − x1 . φ2 (x2 ) φ5 (x2 ) (9. .43) .42) i.

ck ]+ = 0 . nS ) qj states occupied The operators thus deﬁned obey the commutation properties [cj . . of (9.50) . the last factor on the rhs.45) (9. .44). . .17. . nj → 1 . n3 .44) (9. We consider (9.44) follows from the deﬁnition (9. .48) From this follows (9. otherwise.46) [cj . n2 .46). nj−1 . nS ) = j Many–Particle Systems (−1)qj (1 − nj ) |ΨF ( n1 . 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 . . .43). We have used here the deﬁnition qj = k<j nk along with qj = qj . . c† ]+ = 0 j k (9. 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 . 9. The anti-commutation properties (9.19). n2 . . . nj → 0 . It holds c† cj |ΨF (N ) = c† (−1)qj nj |ΨF (n1 . . nS ) = j (−1)qj (nj + 1))(−1)qj (1 − nj ) |ΨF (N ) = (1 − nj ) |ΨF (N ) (9. Let Cj be the matrix corresponding to the operator cj in the basis of fermion states (9. We ﬁrst show that (9. Similarly one obtains cj c† |ΨF (N ) = cj (−1)qj (1 − nj ) |ΨF (n1 . nj → 1.45) follow in an analogous way and will not be derived here.48) vanishes. . . nj+1 . . c† ]+ = δj k k where we have used the deﬁnition of the so-called anti-commutators [A. (9. (9. [c† .49) The derivation involves realization of the fact that a non-zero result is obtained only in case nj = 1. We will prove now the anti-commutation property (9. nj .46) ﬁrst for the case j = k. B]+ = AB + BA.47) Let C† be the matrix adjoint to Cj .248 The operator c† adjoint to cj exhibits the following property j c† |ΨF (n1 . . nS ) = j j (−1)qj (1 − nj + 1)(−1)qj nj |ΨF (N ) = nj ΨF (N ) (9. .

j j For j < k one obtains cj c† |ΨF (N ) = (−1)qj +qk nj (1 − nk )|ΨF (n1 . .h.2: Operators of 2nd Quantization (9. in the basis |ΨF (N ) . i.) = c† 1 c† 2 · · · c† N |0 . nλ2 . for which purpose one needs to envoke only the algebraic (anti-commutation) properties (9.52) One can obtain similarly the same relationship in the case j > k. i..57) On the l.s. nj = 0 ..s. .46). j j j j j j j (9. 2 (9.53) is diagonal in the occupation number representation. . ˆ It is an interesting exercise to demonstrate that Nj only has eigenvalues 0 or 1. . nS ) k c† cj |ΨF (N ) = (−1)qj +qk −1 nj (1 − nk )|ΨF (n1 . .9. nk = 1 . nS ) k and. of this equation we meant to indicate only those N occupation numbers nλj which are non-vanishing. nλN .46).56) Here we have made use of cj cj = 0 and c† c† = 0 which follows from nj ∈ {0. 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 .49) shows that the operator ˆ Nj = c† cj j (9. Since (9. .e. .49) and (9. .e. . i. λ λ λ (9.29)) |ΨF (.58) . σ 1 (σ = ± 2 ) . hence. . ( cj c† + c† cj ) |ΨF (N ) = 0 . . 1}.e.50) yield ( cj c† + c† cj ) |ΨF (N ) = |ΨF (N ) . j → j. 1 In the following we will apply fermion operators c† and cj only to electrons which carry spin 2 .h. . .54) ˆ N = j=1 ˆ Nj (9. nk = 1 . x|j → φj (r) | 1 σ . with eigenvalues equal to the occupation numbers nj ˆ Nj |ΨF (N ) = nj |ΨF (N ) ˆ Nj . We j will often separate the states |j . . S (9. is referred to as the occupation number operator. .51. the creation operators must operate in the proper order. nλ1 .52) hold for any |ΨF (N ) one can conclude (9. k k 249 (9. coordinates x and index j into a space part and a spin part. . Equation (9. . an operator cj must be on the left of an operator ck for j < k.55) is called the particle number operator. .. hence. .51) (9. 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. . nj = 0 . 9. On the r.

9. Operators of the type 1 ˆ G = 2 N g (xj . . 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. ˆ ΨB.250 Many–Particle Systems 9.60) ˆ correspondingly are called two–particle operators. respectively. ˆ .. r = 1.F (N ) .61) is not a one particle operator as long as the N operators rj (x). 9. Since the many–particle states are built-up from a basis |r .k=1 j=k (9. i. The operators f (x) which constitute F are called the associated generic single–particle operators1 .e. (9. The essential aspect of deﬁnition (9. . We adopt a similar expression to distinguish two–particle operators G and g .F (N )|F |ΨB. j = 1. These operators had been introduced already in Eq. 2.59) and over pairs of particles in (9.59) ˆ ˆ are called single–particle operators. (9. .59. . σs ) (9. x ) which constitute G are ˆ called the associated generic two–particle operators.59) and (9. The operators g (x. 9. .17). N are not all identical. x2 . S of single–particle spinorbital states (see page 241) and the operators are speciﬁed through the associated generic operators ˆ f and g . xk .60) is that the sum over particles in (9.15. σr ) f (x) ψs (r. namely. (9. ˆ r|f |s = 1 ∗ ˆ d3 r ψr (r. An operator N ˆ R = j=1 rj (xj ) ˆ (9.59.63) ˆ ˆ This expression has been speciﬁed for the purpose of these notes to distinguish F and f and is not common ˆ terminology. xN . respectively – which are all identical. .17) and integrating over all particle space-spin coordinates x1 . 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. xk ) ˆ j. . .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. .60) involves generic operators – acting on xj and on xj . The action of the operators boson and many–fermion states |Ψ is obviously described through the many–particle state matrix elements ˆ ΨB.6) above.F (N )|G|ΨB.F (N ) . .62) These matrix elements can be obtained by chosing the operators and many-particle states in the position representation. 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 ) .15. 9. 2.

ˆ 251 (9. according to (9. that the symmetry implies that one can switch simultaneously the pairs of indices r.71) .64) We like to note an important symmetry of the matrix element r.65) This symmetry will be exploited repeatedly below. σu ) .3: One– and Two–Particle Operators r. 9. 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 .66) and (9. r|ˆ|u.70) Both operators. are spin-independent as is evident from the factor δσσ .e.59).64). ∂rj2 (9.69). (9. σs ) g (x. ∂r 2 ∂x2 ∂x2 ∂x2 1 2 3 (9. σr )ψs (r . i.. x ) ψt (r.62) in terms of the matrix elements of single–particle states (9.9.67) ˆ In this case the generic single–particle operator. Notice. σt )ψu (r . 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. one cannot switch the indices of one pair individually. u = g d3 r ∗ ∗ d3 r ψr (r. 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) . ˜ (9.66) Here we adopt the notation for the Laplacian ∂2 ∂2 ∂2 ∂2 = + + . An example for a two–particle operator is the Coulomb repulsion operator 1 ˆ V = 2 N j.64) r. u. s|ˆ|t.k=1 j=k q2 |rj − rk | (9. s|ˆ|t. u = s.69) ˆ In this case the generic operator is f (x) = δ(r − r). t . ∂r 2 (9.68) Another single–particle operator is the one–particle density operator N ρ(r ) = ˆ˜ j=1 δ(r − rj ) . s|ˆ|t. is t(x) = (− 2 /2m)(∂ 2 /∂r 2 ). g g (9. u which follows directly g from the deﬁnition (9. We will show that the boson and fermion creation and annihilation operators serve this purpose. 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. s and t.

x ) = q 2 /|r1 − r2 |.+ .3 2 2 (9. one can obviously extract the term j Sj to be left with a two–particle operator in the strict sense of our deﬁnition.− + S1.− . The generic operator is s12 = S1 · S2 = ˆ 1 1 S1. The operator N S = j=1 Sj (9. The non-vanishing matrix elements of the spherical components of s are (α = + 1 .77) δσ σ ( δσσ − δ−σσ ) The operators (9.. β|ˆ− |s. u. β s The operator S 2 = δrs = δrs = = + − 1 2 1 2 δrs δrs (9. i. The generic operator is given through the Pauli matrices. α|ˆ+ |s. α s r. Sj.74) and (9. σ. However. σ | t |t. not in the strict sense of our deﬁnition above since the restriction 2 j = k does not apply in the summation. σ. s. (9.k=1 (9.77) are orbital-independent as evidenced by the factors δrs and δrt δsu in the respective formulas. 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 + .3 S2.252 Many–Particle Systems which is also spin-independent. β|ˆ3 |s.− S2.73) is a one-particle operator. In this case the generic operator is v (x. u. β = ˆ 2 2 1 −2) r.3 . Sj.e.74) = N j=1 Sj = 2 N Sj · Sk j. α s r. σ | s12 |t. . however. α|ˆ3 |s.+ S2. 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.+ + S1. The ˆ matrix elements of the generic operator in terms of single–particle states are ˆ r. s.75) is a two–particle operator. σ . β s r. Here.72) As a ﬁnal example of one– and two–particle operators we consider operators of total spin. Sj is the spin operator for particle j with components (in the spherical representation) Sj.76) the single–particle state matrix elements of which are r. s = 1 σ. σ .

79) have the following implication: The operators (9. (9. namely cu ct . u . respectively. s|ˆ|t. r ΨF (N )|c† cs |ΨF (N ) .79) in the basis of many–particle states |ΨB.62) are correct one needs to compare the result derived with conventional derivations of the matrix elements2 .e. u cr cs cu ct g fermions (9. i.s.59) and G in (9.24) and (9. 9. u a† a† at au bosons g r s † † r.81) For the latter matrix elements simple rules can be derived from the algebraic properties of the boson and fermion operators (9. s|ˆ|t.60) as follows N ˆ f (xj ) → j=1 N 1 2 1 2 S r. and then in g a second step the matrix elements ΨB (N )|a† as |ΨB (N ) . 9.79) is opposite to that in the matrix element r.F (N ) have the same values as the respective matrix elements (9. The matrix elements are then actually those of the number operator c† cr which is diagonal in the occupation number representation.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’. The Matrix Elements ΨF (N )|c† cs |ΨF (N ) r We consider ﬁrst the case r = s. These rules will be provided below only for the case of fermions. xk ) → ˆ j. . g and not ct cu .78) and (9.78. pp.45.3: One– and Two–Particle Operators Deﬁnition in Terms of Creation and Annihilation Operators 253 In order to express many–particle state matrix elements (9.64).u=1 S r. u as deﬁned in (9.t. s|ˆ|t. In order to determine these matrix elements one needs to evaluate ﬁrst the matrix elements ˆ of the generic operators r|f |s r.u=1 (9.s=1 ˆ r|f |s a† as bosons r ˆ r|f |s c† cs fermions r r.78) 1 2 g (xj . To show that the resulting values of the matrix elements (9.9.176 where the matrix elements of fermion states can be found.. 9.23.46).s=1 S r.62). respectively.k=1 j=k S r.80) (9.64) one replaces the operators F in (9.82) ΨF (N )|c† cs |ΨF (N ) = r cr 2 (9.79) It is crucial to notice in the expression for the fermion two–particle operator that the order of the annihilation operators in (9.83) t=1 t=r.63) and (9.171 and pp.t. r ΨB (N )|a† a† at au |ΨB (N ) r s ΨF (N )|c† c† ct cu |ΨF (N ) r s bosons fermions (9.s. r S Ψ (N In case of r = s one obtains F )|c† cr |ΨF (N ) r = nr r=1 δnr nr .63. 9.62) through single–particle state matrix ˆ ˆ elements (9. Expressions (9. s|ˆ|t.

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.s.i. ˆ ˆ Nr Ns .59) when the matrix elements are determined between Slater determinant wave functions. t.86) For the combination of indices r. for which holds r = s.e.86) into three contributions.78) does yield the same results as the operator deﬁned in (9. are ˆ ˆ ΨF (N )|F |ΨF (N ) = (1 − ns )(−1)qs nt (−1)qt s|f |t (9.d. u) are included in the summation for which all indices are diﬀerent. u need to be considered.i. t. s. c2 = 0 r (9. (r. s... we split the sum in (9.s=1 a.84) The oﬀ-diagonal elements with nr = nr except for r = s. t.i. s). 9.u r=s. Comparision with the results in Condon&Shortley (pp. Case 2 all four indices are diﬀerent. The anti-commutation property (9.i.45) yields together with the deﬁnition of the occupation number operator (9.79) and using c† r we can write S 1 ˆ G = rs|ˆ|tu c† c† cu ct g r s 2 r. u in this sum three possibilities exist Case 0 two of the four indices are diﬀerent.d. in which case ns = ns + 1 and nt = nt − 1 holds.s=1 a.254 From (9.e. where each contribution corresponds to one of the three cases mentioned.85) All other matrix elements vanish. G = G0 + G1 + G2 . (9.88) . s. 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.’ (all indices diﬀerent) that only tuples (r.82. namely.53) 1 ˆ G0 = 2 S r s|ˆ|r s − r s|ˆ|s r g g r. 1 2 S r s|ˆ|s r c† c† cr cs . 171) shows that the operator deﬁned in (9. r = t etc. t). t=u 2 = 0.t. those for which N and N diﬀer in more than two occupation numbers. i. Case 1 three of the four indices are diﬀerent. i.d.87) Here and in the following we denote by ‘a. t. Starting from the deﬁnition (9. (r. ˆ ˆ ˆ ˆ Accordingly. g r s r. (9.d.s=1 a.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.

t=1 a. 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.t>u r s|ˆ|t u − r s|ˆ|u t g g c† c† cu ct . 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.i.t<u S r.s.90) ˆ Since Nr is diagonal.3: One– and Two–Particle Operators 255 Obviously. this part accounts for all diagonal contributions to G. ˆ The second contribution G1 is ˆ G1 = + + + 1 2 1 2 1 2 1 2 S r. this contribution to the two–particle operator is diagonal in the occupation number ˆ representation.s r.9.u r>s. S r.91) ˆ G2 = r.e.u r<s.93) . s|ˆ|r.u r<s. (9.t. given in (9.u r>s. s|ˆ|s. (9. r s|ˆ|r t − r s|ˆ|t r g g ˆ c† ct Nr .d. nt .65).d. r g g . Commutation of ct cu according to (9.i.i. Only G0 . S r. employing the symmetry property (9.86).i. A similar contribution does not arise for F . F (N )|G|ΨF (N ) .d. s − r.t.t=1 a.t>u S r. g r s (9.t.t. nu . As we will ﬁnd. s (9.d.s.s. i. 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.s.s. contributes in this case.i.92) ˆ This contribution to G has non-vanishing matrix elements only when N diﬀers from N in four ˆ occupation numbers nr . r s (9. this contribution obviously behaves similarly to the oﬀ-diagonal part (9.s.89) S r.84) ˆ of the one–particle operator F .t. One obtains 1 ˆ ΨF (N )|G|ΨF (N ) = 2 S nr ns r. N = N or nr = nr for all r.u r>s.t=1 a.s. ˆ Case 0 This case covers diagonal matrix elements.t=1 a.45).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.t>u S r.88).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 .s.t=1 a.d. ns ..s.

95) In the latter formula the ‘+’-sign applies for s < r. it holds S nr = S nr .92). All matrix elements which are not covered by the three cases above vanish.73) and S 2 given in (9.256 Many–Particle Systems Case 1 This case covers oﬀ-diagonal elements with nr = nr except for r = s.5: Derive (9.89). 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. v − r. We like to express these operators through ˆ fermion creation and annihilation operators.3. r=1 r=1 Exercise Exercise Exercise Exercise Exercise 9.90) from (9. s|ˆ|r. s|ˆ|v. the ‘−’-sign applies for s < r. nt = nt + 1.4: 9.75) which are. In particular. nv = nv − 1.76). u < v. t − r. Only G1 .e. u g g (9. nu = nu − 1. The matrix elements of the corresponding generic operators are (9.1: 9. s|ˆ|t. s|ˆ|u. Furthermore. given in (9.92). u < v or s > r.t nr r.3. particle number is conserved.3. given in (9. Derive the non-vanishing matrix elements (9. r g g (9. One obtains ˆ ΨF (N )|G|ΨF (N ) = ± (−1)qs +qr +qu +qv (1 − ns )(1 − nr )nu nv r.95) from (9.2: 9.62) for bosons. Derive (9. respectively.94) Case 2 This case covers oﬀ-diagonal elements with nr = nr except for r = s. contributes in this case. contributes in this case.91).3. u. i. u > v.3: 9.74) and (9. Only G2 . t. for ˆ non-vanishing matrix elements ΨF (N )|G|ΨF (N ) at most four occupation numbers nr and nr can diﬀer.90). One obtains ˆ ΨF (N )|G|ΨF (N ) = (−1)qs +qt (1 − ns )nt r r=s..3. Derive (9. ˆ ˆ Show that the matrix elements of F and of G conserve particle number. v for which holds ˆ ns = ns + 1.96) . one–particle and two–particle operators. Spin Operators ˆ ˆ In this section we will brieﬂy consider the spin operators S given in (9.92) from (9. u > v or s > r. t for which holds ˆ ns = ns + 1 and nt = nt − 1.

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

(9..J. If it were’nt for the fermion nature.101) The unitary property of Urn U† U = U U† = 1 1 or S S (9.s=1 σ ˆ rσ r|t|s c† csσ (9. 1989) .1: Imagine that in a closed. 2. i. r=1 3 Urm U ∗sm = δrs m=1 (9. The Aufbau principle. The Hamiltonian of such system is S Ho = r. ranging from nuclei to the molecules of living systems . where S |m) = r=1 Urm |r . .4. all material systems would condensate into states which would depend little on particle number. another prototypical systems of many strongly interacting components. ultimately depend in a crucial way on the fermion character of its constituent building blocks. electrons and nucleons. a Hamiltonian without a two–particle operator contribution accounting for interactions among the particles. namely of neurons3 . In such a representation Ho is diagonal and the stationary states can be stated readily. What might happen? Independent-Particle Hamiltonian A many–fermion system governed by an independent-particle Hamiltonian. New York. Before continuing one may ﬁnally point out that the material world as we know it and as it establishes the varieties of natural substances. . m = 1.e.258 Many–Particle Systems that today these concepts are providing insight into the functioning of biological brains.100) in a form in which the factors c† csσ become rσ occupation number operators.103) An account of some of these eﬀorts can be found in Modelling Brain Function – The World of Attractor Neural Networks by D. S}.e. would not exist. Chemistry essentially would know only one element and Life would not exist. The electronic properties of atoms with diﬀerent numbers of electrons would diﬀer little. We will restrict our description in the following to systems composed of an even number (2N ) of particles and to spin-independent interactions. . Exercise 9.100) ˆ ˆ We will denote the matrix elements of t as r|t|s = trs . {|m). according to which nuclei and atoms change their qualitative properties when they grow larger.. water-ﬁlled jar all electrons of water turn their fermion nature into a boson nature. i.102) U ∗rm Urn = δmn . can be described in a rather straightforward way. Transformation to a New Set of Creation and Annihilation Operators Our aim is to represent the Hamiltonian (9. Amit (Cambridge University Press. 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).

110) can be demonstrated similarly. nσ We demonstrate here the anti-commutation property (9.108) obey the same anti-commutation properties (9. they are Fermion creation and annihilation operators. the expressions drived above for the matrix elements of one..s=1 ˆ U † tU mn (9.4: Independent-Particle Models allows one to express conversely |r in terms of |m) S 259 |r = m=1 ∗ Urm |m) . [d† .107) Urn c† rσ . ∗ ∗ Urm Urn . of N -Fermion states in which single particle states |nj ) as deﬁned in (9.s r The unitarity property r ∗ Urm Urn = δmn yields immediately identity (??). using (9. hence. (9.f.111). In a basis of states N d† j σj |0 n j=1 (9. The l.46). (9.s. rσ namely. (9.105) where ˜ ˆ tmn = (m|t|n) = S ∗ trs Urm Usn = r.100) yields ˜ Ho = tmn d† dnσ mσ where d† nσ = r=1 S S (9. [dmσ . d† ]+ = 0 (9.46) as c† and crσ .104) One can.104)] S S ∗ Urm d† mσ m=1 c† rσ = .e. d† ]+ = sσ nσ r.and two-particle operators hold in an analogous way for d† and dnσ .106) which together with (9. the anti-commutation properties (9. accordingly. . csσ = n=1 Usn dnσ . the operators d† and dnσ .108) and (9.103) allows one to express [c. The unitary property (9. dnσ ]+ = 0 .9. i.h. 9.111) can be written.101) of states |r . behave exactly like the operators c† and crσ behave for states |ΨF (N ) .111) and.45. d† ]+ = δmn δσσ nσ (9. for nσ rσ example. dnσ = r=1 ∗ Urn crσ (9.101) are occupied.109) The operators (9.112) . (9. expand S trs = mn ∗ ˜ tmn Urm Usn (9.110) mσ nσ [dmσ .108) These operators describe the creation and annihilation operators of fermions in states |n) which are linear combinations (9.113) Urm Usn [crσ . c† ]+ = δσσ [dmσ . of (9.

σk ) for j = k (9. This equation poses the eigenvalue problem for the hermitian matrix (trs ). It is. We apply for this purpose (9.115) can be stated immediately since any many–particle wave function in the occupation number representation is suitable.. n2N . The corresponding eigenvalues n . (9.116) which follows from (9. 2 2 2 2 i. i. a fermion in a spin state | 1 . . is obtained. The condition (9. . USn ).e. The 2N fermion state 2N d† j σj |0 n j=1 where (nj . ˜ tmn = In this case Ho has the desired form S n δmn . transformation matrices Urn and energy values n . Vn = (U1n . 2. ) = j=1 nj . .119) In this state the N lowest orbital eigenstates of trs are occupied each by an electron with spin |α = | 1 . σ2 .101) gives us the freedom to choose the matrix (9.117) are eigenstates of (9. S are T real. − 1 . σj ) = (nk . j j (9. a simple matter to nσ state many–particle states in the new representation. . σ2N . r = 1. the so-called ground state..106) diagonal.e.. . σ1 . .114) Ho = n=1 † n dnσ dnσ (9.57) to the case of 2N particles. . is N |Φo = j=1 d† α d† β |0 . 1 as well as a fermion in a spin state | 1 . . A non-degenerate orbital state which is occupied by two fermions.106) and the unitary property of Urn .e. i. .115) with eigenvalues 2N E(n1 . S (9.260 Diagonal Representation Many–Particle Systems The transformation (9. n = 1. Before proceeding we like to address the issue how the new representation. 2.115) ˜ and involves solely occupation number operators N nσ = d† dnσ . 1 and |β = | 1 . Eigenstates of (9. (9.114) together with (9. U2n . . n2 . the associated ( properly normalized) eigenvectors Vn are the columns of (Urn ).e. − 1 is called a closed 2 2 2 2 .106) is equivalent to S trs Usn = s=1 n Urn ∀r..118) Ground State In case of an ordering m < n for m < n the state of lowest energy. . i. . hence.

α = 2 |β. (9.9. the third term − S m. The ground state has only closed shells. 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 = 2 |α. i. β .97). Alltogether. of ∆E = and |β.e.124) . the states diﬀer in their spin character. is a singlet state.e. The two states |α. namely. β ˆ are eigenstates of S 2 given in (9.n=1 † † if either |m) or |n) are closed shells.100)..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. One can conclude immediately that the ˜ ˜ ˜ ˜ (N mα − N mβ )(N nα − N nβ ) does not give any contribution second term in (9. energy above the ground state. α . Obviously. In case of m = n it gives a contribution ‘1’ for each closed shell which cancels the contribution of the ﬁrst term. β |β. 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. |α.n=1 cmα cnβ cmβ cnα does not give a contribution if m = n and either |m) or |n) are closed shells. α N +1 − N .n=1 d† d† dmβ dnα mα nβ (9.n=1 m. both with eigenvalues 2 ˆ S 2 |α.4: Independent-Particle Models 261 shell..n=1 d† d† dmβ dnα mα nβ (9.n=N (9. Similarly. However. ˆ S 2 ||β. We will demonstrate now that this property endows the ground state with total spin zero. namely. ˆ one can conclude that |Φo is an eigenstate of S 2 with eigenvalue zero. There are four such states.120) where the occupation number operators refer to the single-particle states |m).120). Excited States We want to construct now excited states for the system governed by the independent-particle Hamiltonian (9. α |β. α |α. 1 S 4 m.121) All four states have the same excitation energy. i. 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.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. The action of the operator (9.n=1 m.120) is particularly simple for closed shells. 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).97). 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.

do not need to be considered since they ˆ give vanishing contributions. . mα nβ (9.129) The symmetry of the Hamiltonian suggests to choose a new representation deﬁned through the operators (r = 1. of course. ˆ 3 on the two states produces . The system is assumed to posses a 2N –fold symmetry axis and interactions connect only orbital states |r → |r ± 1 . We identify the states |2N + 1 = |1 to avoid cumbersome summation limits.126) which follows from the occurence of squares of fermion operators which. |N + 3 . . β both with eigenvalues 1. conclude that (9. . i. 2. β and |β. . one with eigenvalue zero (singlet) and one with eigenvalue ‘2’ (triplet). One can. r = 1. α and |β. vanish. α are. for example.127) ˆ are triplet and singlet states. 2. β qualify as eigenstates.122) is correct.σ = (9. hence.100) outlined above we consider a system of 2N electrons which move in a set of atomic orbitals |r . α as triplet states. .262 and ˆ Σ3 = − m. 2N ) drσ = √ 1 2N 2N e−irnπ/N cnσ n=1 (9. also eigenstates of Σ2 .130) . β are not eigenstates of S 2 .4. for |α.128) The cyclic property of the system is expressed through c2N +1 σ = c1 σ . |β. . ˆ An eigenvalue ‘2’ of the spin operator S 2 identiﬁes the states |α..2: Construct four operators O ˆ O ˆ such that O ˆ S3 . α . The action of Σ N +1 ˆ Σ3 |α. of course. c† +1 σ = c† σ 1 2N (9. 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 . (9.n=N d† d† dmβ dnα d† +1α dN β |Φo = 0 mα nβ N (9. 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. however.σ dN σ N σ. . α ± |β.e.σ d† +1. β . The two states |α. 2N which are located on a ring. β =− m. The same result holds for |β. .125) ˆ Contributions to S 2 acting on states |N + 2 . the linear combinations 1/2(|α. ˆ The remaining states |α. etc. m |Φo m m m γσ. .n=N Many–Particle Systems N +1 d† d† dmβ dnα . ˆ Exercise 9.

9.s=1 σ ˜ rσ trs d† dsσ (9.f. 2N.137) where 1 ˜ trs = e−irπ/N 2N Using (9. n=1 (9.131) The (9. .130.h. (9.109)] ¸ cnσ = √ 1 2N 2N eirnπ/N drσ r=1 2N (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 . rσ (9. 2N (9. .139) −2t cos r=1 σ rπ N d† drσ . one obtains the new representation of Ho 2N ˆ Ho = −t r.140) The Hamiltonian is now diagonal and the construction outlined above can be applied.134) which. 1−a (9. (9.4: Independent-Particle Models d† = √ rσ 1 2N 2N 263 eirnπ/N c† nσ n=1 √ 1 eirnπ/N .135) c† nσ 1 = √ 2N e−irnπ/N d† rσ r=1 (9. . of (9. thereby. eisπ/N 1 − e2N isπ/N 1 − eisπ/N (9. One can then express[. vanishes for integer s.131) constitute a unitary transformation U deﬁned through Urn = unitarity property follows from 2N eirsπ/N = 0 r=1 for 0 < s < 2N .132) gives for r = 0. . .9.133) Application to the l. 4N.s.136) ˆ and.138) ˜ trs = from which follows ˆ Ho = eirπ/N + e−irπ/N 2N δrs (9. indeed.132) one can conclude 2N ei(s−r)nπ/N + eisπ/N n=1 1 2N 2N ei(r−s)nπ/N .

. 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 . i.. those with energy above that of the (energetically) highest occupied single-particle state. in states which are constructed according to the ‘Aufbauprinzip’ in a manner which neglects the mutual repulsion between electrons. i..g. n = 1..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 . rather they assume that each particle experiences the average interaction due to the fermions frozen into the ground state. (9.4.e. 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. 2. i. the fermion nature restricts the possibility for perturbations on the system. We will present in this and the following section a method which constructs such optimal single-fermion states. . Hence. i. in particular. The regularities of the elements can be rationalized in terms of electrons moving in hydrogen atom-type orbitals.e. There are two reasons why this is so: ﬁrst.e.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. if one can choose the singleparticle states such that the residual perturbations are small. valence bands or conduction bands. 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.141) State the excitation energy ∆E. properties of many-fermion properties deviant from such simple description is then of as much interest as independent fermion descriptions are useful. A second reason why independent-particle models can be successful is that the pair interaction for electrons is the slowly decaying Coulomb interaction. The most famous example is the periodic table of the elements.e.. the everpresent Coulomb repulsion between electrons. Many electronic properties of solids can be understood in a similar way. if one places fermions into an independent-particle ground state of the type (9.. .. assuming that electrons move independently from each other in so-called bands. e.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. Of course. i. Since independent-particle behaviour is so universal one may wonder why the presence of particleparticle interactions. does not spoil it.264 Many–Particle Systems Exercise 9. 9. . The slow decay of the interaction makes the motion of any one electron rather independent of the exact position of most other electrons. A similar principle accounts approximately for properties of nuclei as described by the nuclear shell model. 2N with interactions between neighbouring (n. These states do not altogether neglect the pair interactions. and that independent-particle behaviour surfaces mainly because it applies well to the ground state.

N . as in in (9. the corresponding U = ( Urm ) forms an S × N matrix and. . .s.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. . 2. S is a unitary matrix.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. 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. 2. m = 1.145). r = 1. s|ˆ|t. .142) r|t|s c† csσ + rσ sσ 2 r.9. m = 1.142) is based.108). . |r = m=1 ∗ Urm |m .u=1 σ. . t. u refer to the initial basis of single-particle states |r . S} the elements of which will be labelled by indices ˜ m. u c† c† cuσ ctσ v (9. 2. u v r. . p. 2. . This is done as follows: ˆ Vmf (|Φo (U) ) = 1 2 S r. . q. ˜ (9.t.t. . . s. s|ˆ|t.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. U = ( Urm ) as deﬁned in (9. j j (9. . If one restricts m = 1. The states |m are connected with the states |r through (m = 1. is not unitary.143) with r. The states |m serve to deﬁne a reference state of the system ˜ N |Φo (U) = j=1 d† α d† β |0 . We will adopt a second set of single-particle states {|m . S) ˜ S S |m = ˜ r=1 Urm |r . In our formulas below we will adopt strictly the convention that indices r.146) .σ and derive a single-particle operator which approximates this Hamiltonian. naturally.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. .142) over this reference state to turn the contribution into an eﬀective one-particle operator.144) deﬁned as in (9. . . S on which (9.u=1 σ. 2.s.s=1 σ r. .

s. d† d† mσ nσ dmσ dnσ d† dnσ mσ = = = = and.m. 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 = ˜ ˆ ˜ s.149) The remaining matrix elements appearing in (9.n=1 σ.146) accordingly.u ∗ r.t=1 σ.σ S d† d† mσ nσ cuσ ctσ + r.t.148) are obtained in a similar way by transforming only two of the four single-particle states into the new representation.t=1 σ.84).n.150) r. ˆ (9. hence.σ S ctσ + m. u ˜ ˜v m. u Usm Uun . m|ˆ|t.σ S d† dnσ cuσ mσ − r. u c† d† dnσ cuσ ˜ vn rσ mσ r.u=1 σ.σ S d† dnσ mσ ctσ (9. For the averages ··· in (9.s.m. s|ˆ|t.σ S 1 − 2 where. n c† ˜ v ˜ sσ m.n=1 σ.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 (9.148) r.n.u=1 σ.148) one obtains. 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. u c† ˜ vn sσ m. n c† ctσ δσ σ ˜ v ˜ rσ m≤N r. s| v t.u=1 σ.σ . m. s|ˆ|˜ . m| v |t.s.m. n c† d† dnσ ˜ v ˜ rσ mσ r. n|ˆ|t.n. Many–Particle Systems (9.266 Here ··· denotes the average O = Φo (U)| O |Φo (U) . n c† c† v ˜ ˜ rσ sσ r.σ S dmσ dnσ + r. by means of the rules (9.t.n. for example. m|ˆ|t. s|ˆ|m.

153) rσ r.N (9.m.5: Self-Consistent Field Theory 1 2 1 2 1 2 S 267 m. m|ˆ|˜ .s σ where r| vmf (|Φo (U) ) |s = ˆ t.154) ˆ The mean ﬁeld approximation replaces then the Hamiltonian (9. m=1..152) can then be written in the form ˆ Vmf (|Φo (U) ) = r| vmf (|Φo (U) ) |s c† csσ ˆ (9.9. m|ˆ|t. n c† ctσ .2. s|ˆ|t.S. u c† cuσ δσσ ˜ vn sσ m≤N s.m.u=1 σ.σ r. t| v |u...155) which is a function of the S × N –matrix U = ( Urm )r=1. t| v |s.t=1 σ (9. r| v |u.u ( 2 r. u c† cuσ ˜ vn rσ (9.2.σ S r. n c† ctσ δσ σ ˜ v ˜ rσ − r. leaves open the question how the reference state should be chosen. s| v |t. .. u = s. s ) ˆ ˆ m≤N ∗ Utm Uum (9.u=1 σ.s=1 σ ˆ r|t|s + r| vmf (|Φo (U) ) |s ˆ c† csσ rσ (9. The one-particle operator (9. s|ˆ|˜ .142) by Hmf (U) deﬁned through S ˆ Hmf (U) = r. We will address now a proper choice of the reference state..σ S + − r. 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. as introduced here. At this point this state is completely arbitrary. m|ˆ|˜ . ˜ v ˜ sσ m≤N s.u=1 σ S − m. u c† cuσ ˜ vn rσ m≤N r..t=1 σ.152) By means of expression (9. u − r.156) which deﬁnes the reference state |Φo (U) .151) Carrying out the sum σ δσσ = 2 leads to a factor 2 for the ﬁrst two terms. The mean ﬁeld approximation.149) for the matrix elements in the mixed representation one can state the right hand side explicitly in terms of Urm . Exploiting the symmetry r.

r = 1. (9.s=1 σ S ˆ rσ r|t|s c† csσ (9. 2. 2. S . i. . (9.158) where the labels m are ordered to obey the condition (1) m < (1) n for m < n . steps (2) and (3) are being repeated M times until the procedure converges. S . S . and obtaining in a step (3) the ground state of this one-particle Hamiltonian according to the construction in Section 9.142). S. assuming in an intial step (1) a properly chosen reference state.159) One deﬁnes the reference state |Φo (U(1) ) using the deﬁnition (9. 2. We will argue below that the self-consistent ﬁeld ground state. M = 1. . . .e. i. is exact only for an independent-particle Hamiltonian. . (9.. Let us state now the construction of the self-consistent ﬁeld ground state in more detail. often also referred to as the Hartree-Fock approximation.4 and deﬁning this as the new reference state.145). . Step 2M. .e. is the lowest energy independent-particle ground state one can construct. however.: Determine Mean Field Hamiltonian In this step the mean ﬁeld Hamiltonian S ˆ (M ) Hmf = r.s=1 σ S ˆ r|t|s + r| vmf |Φo (U(M ) ) ˆ |s c† csσ rσ = r. the independent-particle nature stemming from the fact that the functional form of the ground state. . i. . SCF-Algorithm.6 Self-Consistent Field Algorithm The Self-Consistent Field (SCF) approximation. m = 1.268 Many–Particle Systems 9.155) following the method outlined in Section 9. U(1) is the S ×N –matrix (1) of the ﬁrst N column vectors of Urm .s=1 σ ˆ r|hmf |s c† csσ rσ (M ) (9. . the reference state resulting from step 2M + 1 (within numerical error) is equal to the reference state resulting from step 2M − 1.160) . determining in a step (2) the corresponding mean ﬁeld Hamiltonian (9.. . can achieve this goal only iteratively. 2. SCF-Algorithm. . Step 1: Choosing an Initial Reference State One deﬁnes the Hamiltonian ˆ (1) Hmf = r. r = 1. . The state thus determined is referred to as the self-consistent independent-particle ground state. 9.4. .119. 2. 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 . This procedure.e. .145).155). The procedure determines the reference state as the ground state of the independent-particle Hamiltonian (9. under conditions which often are realized. m = 1.

: 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 .s=1 (9. t| v |u. 2. S . .145). i. m = 1. (9. S. . 2.6. i.1: Let Prs be the one-particle operator with the generic operator prs = |r s|. ˆ (b) How can the expectation values of the operator Prr be interpreted. 2. i.e. indicating that the state converged.9.162) is solved adopting an ordering of the labels m which obeys the condition (M +1) m < (M +1) n for m < n . (c) Show that any spin-independent one-particle operator S ˆ F = r.167) . (9. . r = 1.u=1 ( 2 r.. m = 1. . 2. 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. σ p = δtr δsu δσσ . 2 . ˆ Exercise 9. SCF-Algorithm: Continuation and Convergence Condition When step 2M + 1 is completed. s ) ˆ ˆ m=1 Utm ∗(M ) (M Uum ) (9.161) is calculated..e. S (9. . . . . σ|ˆrs |u.s=1 σ ˆ r|f |s c† csσ rσ (9. etc. .163) The result yields the reference state |Φo (U(M +1) ) using the deﬁnition (9. t| v |s.e. M = 1. . . s = 1. .6: Self-Consistent Field Algorithm is evaluated. . U(M +1) is the (M +1) S × N –matrix of the ﬁrst N column vectors of Urm . u − r. S do not exceed a preset threshold. the S × S–matrix S (M ) r|hmf |s N 269 ˆ = r|t|s + t. step 2M + 2 is carried out. .145). . ˆ t. Step 2M + 1. . SCF-Algorithm.166) ˆ (d) Deﬁne a similar two-particle operator Prstu . 2. (e) Consider the creation operators N † gqσ = m=1 Vmq d† mσ (9. .164) ˆ (a) Determine the expectation value of Prs for the state as deﬁned in (9.165) can be written ˆ F = S ˆ ˆ r|f |s Prs r. r = 1.

168) through a physical observation? Exercise 9.170) (SCF ) Here |r .. We denote the representation which results from the SCF algorithm after its convergence as follows: S |m = ˜ r=1 (SCF Urm ) |r . 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 (9.145). i.171) implies that for the mean ﬁeld Hamiltonian holds S ˆ Hmf (U(SCF ) ) = m=1 σ (SCF ) † dmσ dmσ m .7 Properties of the SCF Ground State We want to investigate now the properties of the SCF ground state. (SCF ) m < (SCF ) n for m < n (9.163).173) .162. .e. Equation (9.σ + 2v ( c† c1α c† c1β + c† c2α c† c2β ) + v 1α 2α 1β 2β + c† c2σ c† 2σ 1σ c1σ ) .163).172) is obeyed. .169) 9. 2. (9. S denotes the initial single-particle states and Urm is the unitary S × S– transformation matrix obtained in (9.6.9. Note that this is not an mσ S × S–matrix! Show that the reference state deﬁned through N † † gqα gqβ |0 q=1 (9.171) where the convention (9. .270 Many–Particle Systems which are connected with d† through a unitary N × N –matrix (Vqm ). We begin by summarizing the result of the SCF algorithm.168) ˆ ˆ has the same expectation values for Prs and Prstu as the reference state (9.2: 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. r = 1. (9. (f) Can one distinguish the states (9. (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. (9.1: Reformulate (9. Employing expressions (9. the mean-ﬁeld Hamiltonian is diagonal in the SCF representation. once as a member of the ground state.142).177).93) for the expectation values (diagonal elements) of the one-particle and two-particle part of the Hamiltonian (9.84) and (9. (SCF ) ˆ Exercise 9.175) However. Since the system under consideration has only 2N particles altogether. exploiting the spinindependence of (9.7: Properties of the SCF Ground State 271 i. i. This over-counting leads to the correction term in (9. mα mβ (9.142).176) 2 mm | v |mm ˜˜ ˆ ˜˜ − mm | v |m m ˜˜ ˆ ˜ ˜ . v (SCF ) Show that the resulting eigenvalue problem of the type (9.146) counts a particle twice. Finally.. Since this state is composed only of closed shells one can conclude following Section 9.e. the mean ﬁeld potential ˆ Vmf (|Φo (U) ) deﬁned in (9.171) in terms of matrix elements r|t|s .7. We like to determine next the energy expectation value of SCF .142).7. 2 m=1 m|t|m + ˜ ˆ˜ m. r.2: Derive (9. u and Urm . Within the mean ﬁeld approximation as described by (9. .4 that SCF is a total singlet state.e.9.162) is non-linear in Urm . and once as a probe particle being subject to the mean pair interaction.171) yields ˆ SCF H SCF N = N (9. ˆ SCF H SCF .177).. s|ˆ|t.173) holds N ˆ SCF Hmf (U(SCF ) ) SCF = 2 r=1 (SCF ) m . one obtains ˆ SCF H SCF N = N (9.177) 2 mm | v |mm ˜˜ ˆ ˜˜ − mm | v |m m ˜˜ ˆ ˜ ˜ . Exercise 9.174) The ﬁrst property we consider is the total spin character of SCF .m =1 Comparision with (9. 2 m=1 (SCF ) m − m. the SCF ground state is N SCF = m=1 d† d† |0 . one can also determine the expectation value of SCF for the Hamiltonian (9.

u c† c† cuσ ctσ . i. the second term on ther. . labeled r = −N0 + 1. s|ˆ|t. .. (9. real number. S N0 (9. The ﬁrst term on the r. describes the sites of a linear lattice.h.178). approximation schemes like the self-consistent ﬁeld approximation play an important role. s|ˆ|u.178) and through the so-called ﬁlling factor n n= 2N N = .s. 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.u σσ where r.s. . v represents the ‘on-site’ Coulomb repulsion.7 for ﬁnite systems.. In spite of the simplicity of its Hamiltonian the Hubbard model.180) V = rσ sσ 2 r. v (9. † ˆ Ho = −t c† . is identical to that of the independent–particle Hamiltonian H0 (9. of (9. i.s.128).272 Many–Particle Systems 9. v (9. We consider as an example the Hubbard model of an inﬁnite linear lattice and apply the model to describe magnetic instabilities in metals.178). The one-dimensional Hubbard model is described by the Hamiltonian ˆ H = −t rσ † c† r+1.8 Mean Field Theory for Macroscopic Systems The SCF algorithm has been developed in Sections 9.σ rσ The potential energy term.182) n can assume values 0 ≤ n ≤ 2.192) derived below is ﬁlled half. t describes the coupling of state |r to states |r ± 1 at the two neighboring lattice sites.. in the usual two–particle operator form. t = v δru δst δrs . i.e. r ∈ Z. rσ rσ r σσ (9. Both N0 and N are macroscopically large numbers. We assume that there are altogether S = 2N0 lattice sites.179) r+1. Due to the lack of an exact solution in dimensions higher than one. According to (9. of (9. t is assumed to be a positive.σ + v 2 c† c† crσ crσ . In the present Section we want to adapt the algorithm to systems containing a macroscopically large number of fermions.178) v contributes only in case that two electrons occupy the same lattice site.181) The Hubbard model. N0 which are populated by 2N electrons (we choose an even number of electrons for convenience). respectively) in state |r at lattice site r.h. is characterized through the parameters t. in the two-dimensional copper oxide lattices of high temperature superconductors. β for ‘up’ and ‘down’ rσ spins.178) Here the index r.178).e.t. referring to the fact that in this case the band of single particle energies (9.e.σ crσ + crσ cr+1. except in case of one-dimensional systems. cannot be solved exactly. v of the Hubbard Hamiltonian (9. . for example. as stated through Hamiltonian (9. . The operator c† (crσ ) creates (destroys) an electron with spin σ (σ = α. reads 1 ˆ r.4–9.σ crσ + crσ cr+1. The case n = 1 is termed the half ﬁlled band case.

e.9. We can now embark on step 2 of the SCF algorithm.179). Accordingly. 2N0 (9. we identify |r + N0 = |r and later let N0 go to inﬁnity. is Urm = √ 1 exp (irmπ/N0 ) .186) where m = −2t cos mπ .144). boundary eﬀects are assumed to be negligible. SCF Ground State for the Hubbard Model We seek to determine the ground state of the Hubbard model stated by (9. mα mβ (9. 9.181..178) can be determined by applying (9.185) into (9.6. .187) The energy levels are labeled such that 0 < ±1 < . one obtains for the mean ﬁeld Hamiltonian N −1 ˆ ˆ Hmf (U) = H0 + v r.140).185) ˆ According to (9. N0 (9. . < ±(N0 −1) < N0 holds as can be readily veriﬁed.183) where the operators d† are related to the original creation operators c† through the unitary mσ rσ ˆ transformation U deﬁned in (9. rσ (9. Accordingly. i. in fact.8: Macroscopic Systems 273 Since we are dealing presently with a macroscopic system.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 .174).188) . The unitary transformation which diagonalizes H0 . by N −1 ||SCF = m=0 d† d† |0 .178) within the framework of the self-consistent ﬁeld theory. This task has ˆ been solved already on page 262. 9. The self-consistent ﬁeld Hamiltonian Hmf which corresponds to the Hubbard Hamiltonian (9. the operator H0 can be re-written in the diagonal form N0 ˆ H0 = m=−N0 +1 σ m d† dmσ mσ (9.184) ˆ We are now ready to apply step 1 of the SCF algorithm and diagonalize term H0 . rσ rσ (9. For this purpose we apply the SCF theory presented in Section 9. we assume that the SCF ground state is given by (9.155.σ m=0 ∗ Urm Urm c† crσ . Inserting (9. Using (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 262.154).184) and using (9.

183) is not the only candidate for the mean ﬁeld ground state of the Hubbard model. −π + N0 . To demonstrate this we consider an alternative formulation of the mean ﬁeld approximation for Hamiltonian (9. the results of diﬀerent formulations agree qualitatively. Indeed. the electrons in the present description behave like independent particles with energies ¯m = m + vn . for km = mπ/N0 holds km ∈ −π + N0 . 4 So far we have not exploited the fact that N0 and N are macroscopically large quantities. the state with the lowest possible energy. Apparently. i. (9. (9. π].178). The ground state of the system. holds k ≡ km ∈] − π. N0 ˆ (2) ˆ Hmf = Hmf = m=−N0 +1 σ m + vn † dmσ dmσ 2 (9. through the so-called dispersion law (k) = −2t cos k . 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. One can see that −2t ≤ (k) ≤ 2t holds. can be obtained. 9. The bottom of the energy band corresponds to (0) = −2t and the width of the energy band is 4t. Usually. . as given in (9.e. even for one and the same Hamiltonian. ˆ The Hamiltonian Hmf .182) of the ﬁlling factor n [c.1). In other words. The energy m can be expressed as a function of a continuous variable k ∈] − π. When N0 becomes macroscopically large the single particle discrete energy levels (9. .. i. . Alternative Description of the Mean Field Approximation The state (9. crσ through d† .187). dnσ according to (9. denoted by F (see Fig.1. by ﬁlling up this energy band with electrons from the bottom of the band (which corresponds to k = 0) to a maximum energy.109) and using (9.. but may diﬀer quantitatively. the ground state of the Hubbard model in the self-consistent ﬁeld approximation can be determined exactly. π]. Indeed. .e. is already diagonal.192) This dispersion law is presented in Fig. in the limit N0 → ∞.190) We have used in the latter expression the deﬁnition (9. −π ≤k ≤π . The energy Emf of this state can be calculated readily by using (9.191) where m is given in (9. called the Fermi energy.f.177) and mm |ˆ|mm = mm |ˆ|m m = ˜˜ v ˜˜ ˜˜ v ˜ ˜ which holds in the present case. hence. according to the Pauli principle. the self–consistency condition is exactly met and the SCF algorithm converged after two steps. namely. mσ (9.182)].190).274 Many–Particle Systems Expressing c† .189) and.187) form a quasi1 2 continuous energy band. there exist several ways of formulating mean ﬁeld approximations.185) one can prove rσ mσ N0 c† crσ = rσ rσ m=−N0 +1 σ d† dmσ . π and. 9.

the deviation of Nrσ from its mean value Nrσ for the corresponding reference state. let us express the interaction term in (9. Inserting (9. we have employed (9..194) ˆ For a given reference state. F = (±kF ) denotes the Fermi energy. i.193) where. Nrσ = SCF ||Nrσ ||SCF .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 .e. First. The anti-commutation properties (9. i. 9.196) (9.1: The dispersion law (k) for independent electrons on an inﬁnite one dimensional lattice.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 . (9. The mean ﬁeld approximation neglects the .193) into (9.9. say the one given in (9.183).195) ˆ ˆ ˆ and where δ (Nrσ ) = Nrσ − Nrσ describes the ﬂuctuations of Nrσ . the operator Nrσ can be written ˆ Nrσ = Nrσ + δ (Nrσ ) .178) and performing the summation over the spin indices results in ˆ ˆ H = H0 + v r ˆ ˆ Nrα Nrβ .8: Macroscopic Systems 275 Figure 9.45. rσ rσ rσ (9. (9.. in the last term on the right hand side.56).e.

however.200) allowing. Therefore. 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 (9.199) which is identical to expression (9. 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.. that nα and nβ assume diﬀerent values. 2N0 2 holds ˆ Hmf = = mσ mv ˆ H0 + 2 vn2 2 rσ nv † vn dmσ dmσ − N . Spin-Polarized Mean Field Ground State ˆ We want to consider now a ground state for Hmf deﬁned through Nrσ = nσ = const (9. Because of Nrσ = c† crσ rσ N −1 = m. in the mean ﬁeld approximation.198) Let us assume that the ground state of the Hubbard model.191) for the ground state energy.198) as the one obtained by using the mean ﬁeld potential (9. but the mean number of particles with spin α can be diﬀerent from the mean number of particles with spin β. . (9.183) . Note that the procedure employed in (9. non-vanishing magnetization. m+ 2 2 c† crσ − 2N0 rσ (9.183). the ground state of the system can have a non-zero local spin and. 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 . consequently. is given by (9. For such state the mean occupation number nrσ is uniform at all lattice sites.197) and the corresponding mean ﬁeld Hamiltonian becomes ˆ ˆ Hmf = H0 + v r ˆ ˆ Nrα Nrβ + Nrβ Nrα − v r Nrα Nrβ . In case nα = nβ the following construction will lead to a ground state which coincides with the non-magnetic ground state (9.195).146). yields essentially the same mean ﬁeld Hamiltonian (9.m ∗ Urm Urm d† dm σ mσ = m=0 ∗ Urm Urm = n N = .276 Many–Particle Systems ﬂuctuations to second order and the mean ﬁeld Hamiltonian is obtained from (9. i.e.

. nβ and µ. In fact. 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 ).e. m (9. subject to the constraint n = nα + nβ . Therefore. For this purpose the sum in (9.9. the so-called Fermi energy F = µ (see below). 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. Since β (k) − α (k) = v(nα − nβ ).σ with respect to nα and nβ .e.204) Emf ≡ 2N0 2N0 m. e. if nα > nβ then β (k) > α (k). (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. (9.201) ( m ˆ + vnβ ) Nmα + ( m ˆ + vnα ) Nmβ − 2N0 vnα nβ . The additional term in (9. one can expect the system to lower its energy by assuming nα = nβ and. N0 ∼ 1023 . At this point we exploit the fact that our system is macroscopically large.8: Macroscopic Systems from which we obtain nσ = The mean ﬁeld Hamiltonian is then ˆ Hmf = m 277 1 2N0 Nmσ .. such that the larger nσ .206). 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) .208) . 2π (9.. β) are determined by minimizing the energy density of the ground state Hmf 1 = [ mσ Nmσ ] − vnα nβ (9. the smaller is σ (k) for a given k. a spin–polarized ground state. The values of nα and nβ are determined by minimizing the energy density (9.g.206) ascertains that condition (9.207) where (k) is given by (9.2a. as is shown schematically in Figure 9. (9. hence.192). In the limit N0 → ∞ this sum can be expressed as an integral.206) ˜mf is minimized where µ is the Lagrangian multiplier considered an independent variable. i.203) is provided by the continuous function σ (k) = (k) + vn−σ . E with respect to nα . (9.205) is met.202) This Hamiltonian describes a system of non-interacting electrons with a spin-dependent dispersion law (9.203) mσ = m + vn−σ .205) Such minimization is realized through the method of Lagrangian multipliers. i. In this limit the discrete energy spectrum (9. According to this method one minimizes ˜ Emf = Emf + µ (n − nα − nβ ) .206) needs to be evaluated.

If the function f depends explicitly only on (k) one can state π −π dk f ( (k)) = 2π π ∞ ρ( )d f ( ) −∞ (9. |f (xi )| (9.211) and using π dkδ(k ± (π − arccos( /2t))) = θ(2t − | |) .278 Many–Particle Systems (a) (b) Figure 9. 2π (9.210) Here δ(x) is the Dirac–delta function and ρ( ). holds π ρ( ) = −π dk δ( + 2t cos k) .211) In order to calculate the integral we recall the following property of the Dirac–delta function δ [f (x)] = i δ(x − xi ) . gives the the number of available one particle states per site.212) where xi are the simple roots of the function f (x). −π . The shaded areas represent the ﬁlled portions of these bands by electrons.2 = ±(π − arccos( /2t)). δ(k + π − arccos( /2t)) + δ(k − π + arccos( /2t)) (2t)2 − 2 . Since the equation f (k) ≡ 2t cos k + = 0 has two simple roots.209) where ρ( ) = −π dk δ ( − (k)) .210). therefore.192) and. For a given dispersion law (k) the density of states ρ( ) can be calculated by employing equation (9.2: (a) Relative position of the two energy sub–bands α (k) and β (k) for ∆ = nα − nβ > 0. Inserting this last result into (9. 2π (9. In our case (k) is given by (9. which is usually called the density of states. (b) Graphical deﬁnition of the limiting energies α and β . unit energy and a given spin orientation of the particle. namely k1. and noting |f (k)| = |2t sin k| = 2t one obtains δ( − (k)) = 1 − cos2 k = (2t)2 − 2 .

2].221) ∂µ Equations (9. 2t[ then the θ function in (9.. (9. the necessary ˜ Since Emf [c. (9.213) The presence of the θ function in the above formula guaranties that the density of states vanishes for | | > 2t. µ. π α. ∂µ (9.214) where σ denotes the value of (k) which corresponds to the top of the ﬁlled portion of the energy sub–band σ (k) [c.219) (9. (9. ∂ α ˜ ∂ Emf =0. β (9.216) allow one.204) through an integral expression.8: Macroscopic Systems 279 where θ(x) is the step function. β . From (9. nα and nβ as a function of the parameters of the Hubbard model. + vnα − µ = 0 .219. if we assume that is restricted to the interval ] − 2t. in principle.f. i.213) can simply be replaced by 1. We can employ the results obtained to express the ground state energy density (9. v and n.221) and (9. θ(x) = 1 if x > 0 and θ(x) = 0 if x < 0. (9.220) β ˜ ∂ Emf = n − nα − nβ = 0 . i. on conditions for a ground state of minimum energy are ˜ ∂ Emf =0. outside the energy band (k). one arrives at the following expression of the density of states ρ( ) = θ(2t − | |) π (2t)2 − 2 .216) Using (9. i.206)] depends only on continuous quantities. Replacing again the sum by an integral over energy one can write σ nσ = −2t ρ( )d (9. One obtains ˜ Emf = α β ρ( )d + −2t −2t ρ( )d + vnα nβ + µ(n − nα − nβ ) . ∂ β ˜ ∂ Emf =0.215) nα and nβ in (9.218) The derivatives can be readily determined and the conditions (9.220) µ can be readily obtained µ= 1 ( 2 α + β) + vn . to determine the unknown quantities α . (9.f. of t.213) and carrying out the resulting integral yields nσ = 1 arcsin( σ /2t) . namely. 2 (9. 9.9. 9.217) and µ.. F = α + vnβ = β + vnα .222) . Fig. namely..e.e.201).219–9.e.218) read ˜ ∂ Emf = ∂ α ˜ ∂ Emf = ∂ β and α + vnβ − µ = 0 . (9.214) are given by the sum (9.

(9. (9.e.220) and (9.f. The Stoner criterion (9. condition (9.3]. (9. The slope of f (∆) at the origin depends on the Hubbard model parameter v.f.227) can be solved graphically. a magnetic ground state. From the deﬁnition (9.223 one obtains ∆= and second.217) and (9.229) 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. (9. from (9. so called ferromagnetic solution.228) Since the total number of states per site is ﬁnite.280 Many–Particle Systems Comparision of (9. condition (9. i.226) ∆ = f (∆) .183) without magnetization. the slope of f (∆) is positive for all ∆ ≥ 0. If the slope of f (∆) at the origin is less than 1. For this purpose we consider the magnetization of the mean ﬁeld ground state ∆ ≡ nα − nβ . i.226) follows that f (∆) is a monotonically increasing function of ∆. as reﬂected by the expression d[f (∆)] = vρ( d∆ β + ∆) = π (2t)2 v − (v∆ + 2 β) >0. non trivial. Equation (9.227) has only the trivial solution ∆ = 0.215) reveals that µ is indeed equal to the Fermi energy F (see also Figure 9.225) Deﬁning the function v∆ f (∆) = 0 ρ( β + )d . (9.e. the Hubbard model has a ground state with ∆ = 0. through the condition [c. Figure 9. the slope of f (∆) at the origin is larger than 1 [c. from (9. (9.227) will have a second..223) ∆ = 0 corresponds to the ground state (9. We assume in the following discussion that the implicit ∆ dependence of β in f (∆) can be neglected. First.2b).. Indeed.227).223) follows α α − v β . if.222) and (9. One can express ∆ as a function of α and β in two diﬀerent ways. . (9.228)] vc ρ ( β ) = 1 .227) Since f (0) = 0 one can infer that ∆ = 0 is always a solution (the so called paramagnetic solution) of condition (9.229) has a wider range of validity than one can infer from the present analysis of the Hubbard model.3. One still needs to determine α and β . (9.219–9. (9. Accordingly. One can determine vc by equating the slope of f (∆) at the origin to 1. For v above a critical value vc . and only if. f (∆) converges to a maximum value as ∆ → ∆max = n.224) v∆ ∆= β ρ( )d = 0 ρ( β + )d . For this purpose one plots f (∆) versus ∆ as is shown schematically in Figure 9.226) one can obtain ∆ by solving (cf.

n).9. the magnetization ∆ of the ground state vanishes and remains zero for values v < vc .8: Macroscopic Systems 281 Figure 9.232) together with (9. (9. β = 2t sin y . From (9. Accordingly.233) yield the desired expression vc π = π cos (n − 1) .4..234) 2t 2 The resulting phase diagram of the ground state. (9. Since ∆ vanishes near vc we set x = y + δ where δ is a small quantity. The corresponding 2t analytical expression provides the phase diagram of the ground state: when v approaches vc from above. Equation (9.230) one obtains from (9.205) the following set of equations determining x. β x+y x−y = = π(n − 1) 2t π (sin x − sin y) = π∆ . 2 2 2 (9.e. (9. any .231) follows y= δ π π (n − 1) − ≈ (n − 1) . 2t As already mentioned.232) states that x − y is proportional to ∆.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. the plot of vc vs. y and. n. the ground state is characterized by the quantities (v. α. t.3: Graphical solution of the mean ﬁeld equation ∆ = f (∆).220. is presented in Figure 9. 9.231) (9.219. By parameterizing α and β α = 2t sin x . 9. v (9. We like to determine ﬁnally the dependence of vc on the ﬁlling factor n. hence.233) Employing the approximation sin(y + δ) − sin y ≈ δ cos y for small δ. i.

Second. First. can lead to serious modiﬁcations of the mean ﬁeld ground state. strong electron correlation eﬀects . even in these materials. However. 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. Third. the analysis presented above is not quite useless. in one spatial dimension is very strong. Well.282 Many–Particle Systems mean ﬁeld ground state is represented by a point in the phase diagram. which are neglected in the mean ﬁeld theory. In general. the answer is no. the properties of the real ground state of the Hubbard model. Concluding Remarks At the end of our analysis of the mean ﬁeld ground state of the one-dimensional Hubbard model. in three spatial dimensions the theory presented above works ﬁne in the case of the transitional–metal oxides such as NiO and CoO. 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 reason is that the eﬀect of quantum ﬂuctuations.4: Ground state phase diagram of the mean ﬁeld Hamiltonian (9. the eﬀect of quantum ﬂuctuations is getting less important as the dimensionality of the system is increased. 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.202). For example. at least qualitatively. The real ground state of the one-dimensional Hubbard model is quite diﬀerent in nature from the mean ﬁeld ground state discussed here. Nevertheless. since in one spatial dimension there are many exact results available. . once we specify a class of possible states to which the real ground state might belong to. 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. Figure 9.

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

284 Many–Particle Systems .

45)] described by an electrical potential V (r. It is not possible to uncouple the equations to describe only a single type particle without aﬀecting negatively the Lorentz invariance of the equations. Here c denotes the velocity of light. We will ﬁnd that actually several Lorentz–invariant equations which replace (10. t) and a vector potential A(r. x3 . 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. t) are vectors of dimension larger one. x2 .. t ) in another frame. e.f. x3 . etc. t) in one frame with space time coordinates (x1 . The transformations beween such frames according to the Theory of Special Relativity are described by Lorentz transformations.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. i.2) The replacement of (10. v = v1 x1 . t = t− 1 − v1 x c2 1 v1 2 c . x2 . Obviously.e. these ˆ transformations are x1 = x1 − v1 t 1 − v1 2 c .e. 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.1) which connect space time coordinates (x1 . spin–0 particles. t) = ∂t 1 2m q − A(r. any of these equations being speciﬁc for certain classes of particles. In case that v is oriented along the x1 –axis. (refeq:ham2. 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. x2 = x2 . 2 As mentioned. the components representing the spin attribute of particles and also representing together with a particle its anti-particle. t) (10. spin– 1 particles. some of the equations describe a particle together with its anti-particle.g. t) ∂ i ψ(r.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.2) will result. Furthermore. t) ψ(r.. i. t) c 2 i + qV (r. x3 = x3 (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. The representation 285 .. 8.

In such a space Lorentz transformations correspond to 4-dimensional rotations.7 a heuristic derivation of the two most widely used and best known Lorentz–invariant ﬁeld equations. x3 ) (10.e. Note that the components of xµ all have the same dimension. 10.7) equation.1).1). i. The reason for this choice is that the transformations (10. x3 ) = r describes the space coordinates. x 2 . however. namely that of length. x2 . This will provide us with a general set of Lorentz–invariant equations which for various particles take the place of the Schr¨dinger equation.4) invariant.1) leave the quantity s2 = (x0 )2 − (x1 )2 − (x2 )2 − (x3 )2 (10.e. wave functions ψ(r. 10.4 to cover ﬁelds. 5 to the groups SO(3) and SU(2) of rotation transformations of space coordinates and of spin.1 will be extended in Sect. i. We will ﬁnd that this deﬁnition will lead us back to the transformation law (10.1) 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. x2 . Minkowski Space Historically.e. We will denote these vectors as follows def xµ = (x0 . Before introduco ing these general Lorentz–invariant ﬁeld equations we will provide in Sects. This space is called the Minkowski space. Rather than following this avenue we will introduce Lorentz transformations within a setting which does not require real and imaginary coordinates.286 Relativistic Quantum Mechanics in Sect. Rather than starting from (10. 10.. t) and vectors with functions ψ(r. Denoting by xµ the .1 Natural Representation of the Lorentz Group In this Section we consider the natural representation of the Lorentz group L. The Group of Lorentz Transformations L = O(3. x1 . but in a setting of representation theory methods as applied in Secti. 10. t) as components. assume new units for time such that the velocity of light c becomes c = 1. We will. henceforth. namely the Klein–Gordon (Sect. i. the group of Lorentz transformations (10. x 1 . This choice implies dim(time) = dim(length).5) √ One can interprete the quantity −s2 as a distance in a 4–dimensional Euclidean space if one chooses the time component purely imaginary. 0 1 2 3 (10.. for the transformed space-time–cordinates x µ = (x 0 .5) and the Dirac (Sect. The elements L ∈ L act on 4–dimensional vectors of position– and time–coordinates.1). 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.3) where x0 = ct describes the time coordinate and (x1 . x 3 ) holds (x0 )2 − (x1 )2 − (x2 )2 − (x3 )2 = (x )2 − (x )2 − (x )2 − (x )2 .5. 10. 10. 10. we will provide a more basic deﬁnition of the transformations.

Aµν . L ∈ GL(4. 0 0 0 −1 (10. Aµ ν B ν ρ = ν=0 Aµ ν B ν ρ .5) invariant.12) (10. (10. Aµ ν xν = µ=0 new old Aµ ν xν . The subset of GL(4. the notation in (10.] The following possibilities exist for the positioning of the indices µ. i. ν = 0. Since this holds for any xµ it must be true Lµ ρ gµν Lν σ = gρσ . xµ . (10.. i. The Lorentz transformations form the subgroup of all matrices which leave the expression (10. however.1: Natural Representation of the Lorentz Group 287 coordinates in one reference frame and by x µ the coordinates in the other reference frame. Aµ ν . [ A more conventional notation would be Lµν . R). 3 3 3 yµ xµ = new yµ xµ .. 1.e. 2. the Lorentz transformations constitute a linear transformation which we denote by 3 x µ = ν=0 Lµ ν xν . (10.6) Here Lµ ν are the elements of a 4 × 4–matrix representing the Lorentz transformation.10) Combining condition (10. the latter notation will be used for diﬀerent quantities further below.10.9) and (10. the latter set constituting a group. This will be explained further below.6) x µ = Lµ ν xν . 4 × 4 tensor: Aµ ν .11) This condition speciﬁes the key property of Lorentz transformations.6) allows us to introduce the so-called summation conventon: any time the same index appears in an upper and a lower position.e. 3: 4-vector: xµ . This condition can be written xµ gµν xν = x gµν x where µ ν (10. 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.6) yields Lµ ρ gµν Lν σ xρ xσ = gρσ xρ xσ . We will exploit this property below to determine the general form of the Lorentz transformations. the .8) new old old The summation convention allows us to write (10. summation over that index is assumed without explicitly noting it. First. ν=0 (10. R).7) The reason for the notation is two-fold.5) into scalar products. Aµν .9) 1 0 0 0 −1 0 0 ( gµν ) = 0 0 −1 0 = g . The second reason is that upper and lower positions allow us to accomodate the expression (10. The Lorentz transformations are non-singular 4 × 4–matrices with real coeﬃcients.

1).1). i. This stipulates that O(3. hence.13) L = ( Lµ ν ) = 2 0 L 0 L2 1 L2 2 L3 3 . One can also show that if L ∈ O(3.17) one obtains (L−1 )T g L−1 = gLgggLT g = gLgLT g . For L3 = L1 L2 holds LT g L3 = LT LT g L1 L2 = LT (LT gL1 ) L2 = LT g L2 = g .. a group.1) since it satisﬁes (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. To demonstrate the group property of O(3. the associated inverse obeys (10.18) (10.1) is. From this one can obtain using g2 = 1 1 (gLT g)L = 1 and.1).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. .h. 3 2 1 2 1 2 (10.14).17) we note ﬁrst that the identity matrix 1 is an element of O(3.19) i. L3 ∈ O(3. R). 10. L2 ∈ O(3.14) holds for the inverse of L..14) L0 0 −L1 0 −L2 0 −L3 0 −L0 1 L1 1 L2 1 L3 1 .1).22) (10.12) LT gL = g .s. (10. L−1 ∈ O(3. i. L ∈ GL(4. L3 0 L3 1 L3 2 L3 3 Using the deﬁnition (10.16) The corresponding expression for (LT )−1 is obviously (LT )−1 = (L−1 )T = g L g .1). of O(3. LT gL = g } . is called O(3... employing expressions (10. in fact.15) (10. (10. 1) = { L.16.e. In fact.1) ⊂ GL(4. We consider 1 then L1 . This set is identical with the set of all Lorentz transformations L.e.e.20) results in the desired property (L−1 )T g L−1 = g . (10.10) of g one can rewrite the invariance property (10.14). property (10.20) Multiplying (10. R) is a group.21) i.14) from the right by gLT and using (10.288 Relativistic Quantum Mechanics elements of which satisfy this condition.e.1). of (10. = g LT g = 0 1 2 −L 2 L 2 L 2 L3 2 −L0 3 L1 3 L2 3 L3 3 (10. We want to show now L = O(3. the inverse of L 1 L−1 (10.

M2 ∈ M. L↓ .30) ≤ −1} . L0 0 (10.24) or since (L1 0 )2 + (L2 0 )2 + (L3 0 )2 ≥ 0 it holds (L0 0 )2 ≥ 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 . = { L. + + ↑ ↑ We can also demonstrate that for A.10.25) i. = { L.12).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↑ . The above shows that the set of proper Lorentz transformations L↑ allows one to generate all Lorentz transformations.23) and (10.. One can also 1 verify that one can write L↑ − L↓ + L↓ − = gL↑ + = = −L↑ . + We start our investigation by demonstrating that L↑ forms a group. A statement on this value can be made on account of property (10.28) (10. (10. L ∈ O(3. 1). + except for the trivial factors g and −1 It is. L0 0 ≤ −1} . det L = −1.14).34) j=1 j=1 . (10. Considering in (10. From this we can conclude L0 0 ≥ 1 or L0 0 ≤ −1 . L↑ contains 1 1. + − gL↑ + = L↑ g .26) (10. For this purpose we consider ﬁrst the value of det L. 1). M1 = a M2 }. there exist two other distinct classes. Using det AB = det A det B and det AT = det A yields (det L)2 = 1 or det L = ±1 .33) where we used the deﬁnition aM = {M1 . Lorentz transformations in L↑ .31) (10.23) One can classify Lorentz transformations according to the value of the determinant into two distinct classes.32) (10. det L = 1. ∃M2 . 1). (10. Schwartz’s inequality yields j=1 2 3 3 3 j=1 A jB 0 ≤ 0 j A 0 2 j Bj 0 2 . L↓ are disjunct sets deﬁned as follows + + − − L↑ + L↓ + L↑ − L↓ − = { L.14).1: Natural Representation of the Lorentz Group Classiﬁcation of Lorentz Transformations 289 We like to classify now the elements of L = O(3. L ∈ O(3. A second class property follows from (10. det L = 1. Obviously. L0 0 ≥ 1} . (10.29) (10. = { L. det L = −1. L0 0 ≥ 1} . entirely suitable to investigate ﬁrst only 1. L ∈ O(3. L↑ . + = − L↑ g + (10. It holds g ∈ L and −1 ∈ L as one can readily verify testing for property (10.14) which we employ in the formulation (10. B ∈ L+ holds C = AB ∈ L+ .1). 1). L ∈ O(3. Properties (10. σ = 0 yields L0 0 2 − L1 0 2 − L2 0 2 − L3 0 2 = 1.27) (10. hence.e.12) the case ρ = 0.

39) where we have employed the matrix form T 1 deﬁned as in (10. The next group property of L↑ to be demonstrated is the existence of the inverse. (10. hence.14) actually L0 0 ≥ 1 and det L = 1 must hold.34) provides then the estimate conclude from (10.. obviously j=1 A0 0 B 0 0 ≥ 1.37) (10.1). orthochronous Lorentz transformations.1) ascertains CT gC = g it must hold C 0 0 ≥ 1. 1 ∈ L↑ . i. Property (10. In fact.13)..36) It holds. + i. 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. since the group property of O(3. µ ν small . | 3 A0 j B j 0 | < A0 0 B 0 0 .e. Similarly.16).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 . L↑ is called the subgroup of proper. + transformations in a small neighborhood of 1 which we parametrize by inﬁnitesimal parameters. For the inverse of + ↑ any L ∈ L+ holds (10. we consider transformations Lµ ν = δ µ ν + µ ν . δ00 = 1 and It should be noted that according to our present deﬁnition holds δµν = gµρ δ δ11 = δ22 = δ33 = −1. ρ ν (10. Since A0 0 ≥ 1 and B 0 0 ≥ 1. To order O( 2 ) holds g + g = 0.1). (10. one can j=1 j=1 3 0 )2 = (A0 )2 − 1. . if we enforce (10. This relationship shows (L−1 )0 0 = L0 0 .290 Relativistic Quantum Mechanics From (10. from which one can conclude L−1 ∈ L↑ .14) implies 1 + T g (1 + ) = g 1 1 (10.40) and. In the following we + will consider solely this subgroup of SO(3.e. Since the associative property holds for matrix multipli1 1 + ↑ cation we have veriﬁed that L+ is indeed a subgroup of SO(3.12) follows (B 0 0 )2 − 3 (B j 0 )2 = 1 or 3 (B j 0 )2 = (B 0 0 )2 − 1. 1 We will then employ the Lie group properties to generate all transformations in L↑ .35) One can conclude.38) For these transformations. 1 0 0 = ≥ 1 and. holds L0 0 > 0 and the value of the determinant is close to unity. (10. therefore. 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. obviously. Inﬁnitesimal Lorentz transformations The transformations in L↑ have the property that they are continously connected to the identity 1 1. accordingly. + Accordingly. Using the above expression for C 0 0 one can state C 0 0 > 0.

2.44) This result allows us now to deﬁne six generators for the Lorentz transformations(k = 1.48) (10.50) .46) 0 0 0 0 . ϑ2 . 3 (10. 2.47) 0 −1 0 0 0 0 −1 0 0 −1 0 0 0 0 0 0 0 0 K1 = . other ﬁve parameters zero) Kk = (wk = 1.43) k = − Inspection shows that the matrix has 6 independent elements and can be written 0 −w1 −w2 −w3 −w1 0 −ϑ3 ϑ2 . = − k m Km (10. k j j = 1. 3) Jk = (ϑk = 1. 2.49) . The generators are explicitly 0 0 0 0 0 0 0 0 J1 = 0 0 0 −1 0 0 1 0 (10. w3 ) = −w2 ϑ3 0 −ϑ1 −w3 −ϑ2 ϑ1 0 (10. (ϑ1 . j. J ] [ Kk .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 . K ] [ Jk . 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 . J = 0 3 0 1 0 0 0 0 0 0 0 0 0 0 0 (10. k = 1. J2 = 0 0 0 −1 0 0 0 0 0 0 0 0 0 0 0 −1 0 1 .42) This relationship implies µ 0 j µ j = 0 = j 0 . K2 = . 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 (10.45) (10. w2 . w1 . other ﬁve parameters zero) . 3 . ϑ3 . (10.10.1: Natural Representation of the Lorentz Group Using (10.15) one can conclude T 291 = −g g (10.

51) One can readily show. w ∈ R3 3 k=1 wk Kk .49. that the transformations (10.52) where the 3 × 3–matrices Lk are the generators of SO(3) deﬁned in Chapter 5. Kk .. In analogy to equation (5. constitutes an overcomplete parametrization of the rotations (see Chapter 5). w = 0) = 0 0 0 R(ϑ) . 10. however.49) deﬁne the Lie algebra associated with the Lie group L↑ . i.56) = 1 0 0 1 = 1 1 (10. Following the treatment of the rotation group SO(3) one can express the elements of L↑ through + the exponential operators L(ϑ. + The commutation relationships imply that the algebra of the generators Jk . We consider now the Lorentz transformations for ϑ = 0 which are referred to as ‘boosts’. w) = exp ϑ · J + w · K .292 as can be readily veriﬁed.50).1: Demonstrate the commutation relationships (10. Relativistic Quantum Mechanics The commutation relationships (10. 2. 3 where we have deﬁned ϑ · J = k=1 ϑk Jk and w · K = following the algebra in Chapter 5. 3 (10. k = 1.e. Exercise 7.57) .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 . 2. 2π[ .51) for w = 0 correspond to rotations of the spatial coordinates. and using the relationship Jk = 0 0 0 Lk (10. k = 1.54) which.55) L 0 0 0 0 exp (w1 K1 ) = exp Using the idempotence property 0 −1 −1 0 2 0 0 1 0 (10. L(ϑ. (10. (10. For the parameters ϑk of the Lorentz transformations holds obviously ϑk ∈ [0. ϑ. To determine the explicit form of this transformation we evaluate the exponential operator by Taylor expansion.53) Here R(ϑ) are the 3 × 3–rotation matrices constructed in Chapter 5. A boost in the x1 –direction is L = exp(w1 K1 ). 3 is closed. 0 1 n (10.

59) v1 = These two equations yield cosh w1 = 1/ and (10.3.59) cosh w1 Using cosh2 w1 − sinh2 w1 = 1 one can identify sinhw1 = sinh2 w1 + 1. however.60) sinh w1 = v1 / 2 1 − v1 . cosh2 w1 − 1 and coshw1 = cosh2 w1 − 1 = cosh w1 sinh w1 sinh2 w1 + 1 .59) for the case k = 1 corresponds to the relative velocity of two frames of reference. 10. (10.59) can be written 1 2 1 − v1 √−v1 2 1 − v1 2 1 − v1 .1). 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.6. 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. we can state vk ∈ ] − 1. x2 = x2 .59) follows.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. Correspondingly. This property is. (10. one obtains from (10. t = t − v1 x1 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 = 1.63) which agrees with (10. consistent with (10.51) the explicit transformation for space–time–coordinates is then x1 = x1 − v1 t 1 − 2 v1 .58) = cosh w1 1 + sinh w1 1 0 −1 −1 0 The conventional form (10. 10. Accordingly.62) According to (10. in fact. x3 = x3 (10.59). ∞[ . . (10.1) of the Lorentz transformations is obtained through the parameter change sinh w1 v1 = = tanh w1 (10. The range of the parameters wk can now be speciﬁed.10.56. From (10. vk deﬁned in (10.64) We note that the range of wk -values is not a compact set even though the range of vk -values is compact. for wk wk ∈ ] − ∞.

(10. our deﬁnition of quantities with special properties under Lorentz transformations presently is conﬁned to the natural representation. A third 4-vector is the so-called current vector J µ = (ρ. r2 ) and. These quantites always come as four components (a0 . namely.294 Relativistic Quantum Mechanics 10. ﬁnally. t) are the charge density and the current density. 4–Vectors and Tensors In this Section we deﬁne quantities according to their behaviour under Lorentz transformations.4).3) are the so-called 4–vectors aµ . a1 . tensor operators Tkm .67) the transformation behaviour of which we will demonstrate further below. not in the so-called natural representation associated with the 3–dimensional Euclidean space E3 . A) (10. not any physical property f ∈ R is a scalar. t)|2 or the scalar product r1 · r2 of two particle positions. . a2 . Such quantities appear in the description of physical systems and statements about transformation properties are often extremely helpful and usually provide important physical insight. Presently. the z–component x3 of a particle. a3 )T and transform according to a µ = Lµ ν aν . the square of the electric ﬁeld |E(r. Scalars The quantities with the simplest transformation behaviour are so-called scalars f ∈ R which are invariant under transformations. We will see below how true scalars under Lorentz transformations can be constructed. of a system of charges. Counterexamples are the energy. Hence. total angular momentum r 1 2 3 states of composite systems like Y m ( 1 . However.66) Examples of 4-vectors beside xµ are the momentum 4-vector pµ = (E. E = √ m mv . x2 . f = f. we will encounter an impressive example of physical properties.65) An example is s2 deﬁned in (10..69) where V (r.68) where ρ(r. 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. the charge density. p = √ 2 1−v 1 − v2 (10. Another example is the potential 4-vector Aµ = (V. t) and A(r. i.2 Scalars. We have encountered examples in connection with rotational transformations. t) are the electrical and the vector potential of an electromagnetic ﬁeld. The 4-vector character of J µ and of Aµ will be demonstrated further below. 2 |ˆ1 . J) (10. scalars like r = x2 + x2 + x2 .e. (10. we have not yet deﬁned representations of Lorentz transformations beyond the ‘natural’ representation acting in the 4–dimensional space of position– and time–coordinates. vectors like r = (x1 . Nevertheless. x3 )T . respectively. Some of these r ˆ quantities were actually deﬁned with respect to representations of the rotation group in function spaces. spherical harmonics Y m (ˆ). t) and J(r. p) .

e.78) To prove the 4-vector property of ∂µ we will show that g µν ∂ν transforms like a contravariant 4vector. −a1 .71) Contravariant and Covariant 4-Vectors It is convenient to deﬁne a second class of 4-vectors.77) The transformed diﬀerential operator will be denoted by ∂µ = def ∂ .16).2: Scalars.75) (10.12).10. g µν ∂ν = Lµ ρ g ρσ ∂σ . aµ (10. −a2 . This property follows from (10. one can express [(L−1 )T ]µ ν = (L−1 )ν µ and.66) together with (10. The 4-Vector ∂µ (10. If aµ and bµ are 4-vectors aµ gµν bν (10.72) where aν is a vector with transformation behaviour as stated in (10. accordingly. We have duplicated the expression for the inverse of Lµ ν to obtain the correct notation in terms of covariant. −a3 ) (10. and . lower. (10.e.74) g µν aν .76) An important example of a covariant 4-vector is the diﬀerential operator ∂µ = ∂ = ∂xµ ∂ .12) a gµν b µ ν = Lµ ρ gµν Lν σ aρ bσ = aρ gρσ bσ (10. one can employ the tensors g µν to raise and lower indices of L ν establish here the consistency of the ensuing notation.17) can be written (L−1 )ν µ = Lµ ν ..72) of the covariant 4-vector follows = In µν and g µ as well.. the relationship being aµ = gµν aν = (a0 .70) is a scalar.79) This is the inverse Lorentz transformation consistent with (10. The respective vectors aµ are associated with the 4-vectors aµ . We do not fact. In fact. i. gσν xν = Lρ σ gρµ x µ and g µσ gσν = δ µ ν . (10. Multiplication (and summation) of x µ = Lµ ν xν by Lρ σ gρµ yields.73) aµ = Lµ σ aσ . 4–Vectors and Tensors Scalar Product then 295 4-vectors allow one to construct scalar quantities.17) Lµ σ is the transformation inverse to Lσ µ . We start from x µ = Lµ ν xν . (10. xν = g νσ Lρ σ gρµ x µ . ∂t (10. ∂x µ (10. using (10. One calls 4-vectors aµ covariant and 4-vectors aµ contravariant. We like to point out that from deﬁnition (10. In any case one can express (10.66). Covariant 4-vectors transform like a µ = gµν Lν ρ g ρσ aσ where we deﬁned g µν = gµν . i.73) Note that according to (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. e.82) (10.83) is a scalar under Lorentz transformations. 2 2 dt 1 − v 1 − v (10.− ∂t . 2 dt dτ 1 − v (10.e. 1 − v2 1 − v 2 dt (10.85) 1 d d = √ .. (10.88) One can express then the momentum vector pµ = √ d m dxµ = m xµ .296 Relativistic Quantum Mechanics contravariant. For this purpose we deﬁne ﬁrst the contravariant diﬀerential operator ∂ µ = g µν ∂ν = Then the operator 2 ∂µ ∂ µ = ∂t − 2 (10.86) One can write (10. (10.. d’Alembert Operator We want to construct now a scalar diﬀerential operator. upper. i.79) allows us to determine the connection between ∂µ and ∂µ .66). dτ 1 − v 2 dt p0 = E = √ m m dt = √ .87) The remaining components of pµ can be written.81) ∂ .. In fact. ∂µ does indeed transform like a covariant vector.e. 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. Proof that pµ is a 4-vector We will demonstrate now that the momentum 4-vector pµ deﬁned in (10.89) . p1 = √ m v1 m dx1 = √ . indices.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 . this operator is equal to the d’Alembert operator which is known to be Lorentz-invariant.g.84) = 1 − (v)2 (10.67) transforms like (10. i.

92) (10.12). 10. t) = 0 . Lorentz-invariant.96) in the form ∂µ Aµ = 0 . t) and A(r. pµ on the l. namely pµ pµ = pµ gµν pν = E 2 − p 2 Evaluation of the r. is a scalar. in (10. (10.96) is a scalar and. hence.96) The Lorentz-invariance of this gauge. can be demonstrated readily using the 4-vector notation (10.h.3: Relativistic Electrodynamics 297 d The operator m dτ transforms like a scalar. i. ∂t V (r. namely.h. of (10.91) (10.69) is. in fact. Expansion in should then be rapidly convergent. t). m2 m2 v 2 − = m2 1 − v2 1 − v2 (10. t) + · A(r.97) and.95) 2 p This obviously describes the energy of a free particle with rest energy ±m.s.94) 1 m In the non-relativistic limit the rest energy m is the dominant contribution to E. for these potentials. the r. This gauge is not Lorentz-invariant and we will adopt here another gauge. and.89) alltogether transforms like a contravariant 4-vector. of (10. · A = 0. One obtains E = ±m ± p2 2m (p 2 )2 + O 4m3 (p 2 )3 4m5 .77) which allow one to express (10.s. We will demonstrate now the 4-vector property of Aµ . in (10. in fact. 8. and had chosen the so-called Coulomb gauge (8.69) for the electrodynamic potential and the 4-vector derivative (10. hence. respectively. If we can show that Aµ deﬁned in (10.89) must be a 4-vector. (10. the so-called Lorentz gauge. We like to rewrite the last result E 2 = p 2 + m2 or E = ± p 2 + m2 . (10. Since xµ transforms like a contravariant 4-vector. The momentum 4-vector allows us to construct a scalar quantity. Lorentz Gauge In our previous description of the electrodynamic ﬁeld we had introduced the scalar and vector potential V (r.h. a contravariant 4-vector then the l.h. equivalently.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.93) (10.10.s.e.97) We have proven already that ∂µ is a contravariant 4-vector.67) E2 − p 2 = or pµ pµ = m2 which.. kinetic energy ± 2m and relativistic corrections. .s. yields according to (10.90) (10.

18) and.298 Transformation Properties of J µ and Aµ Relativistic Quantum Mechanics The charge density ρ(r.e. t) . t) = 4 π J(r.h. · A(r.13).83) and (10. this can be stated F k0 = −F 0k = E k . t) are known to obey the continuity property ∂t ρ(r..68). (8. 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 .82) as follows F µν = ∂ µ Aν − ∂ ν Aµ . t) − 2 A(r. using (10. i. thereby. one obtains 2 ∂t V (r. This principle should hold in any frame of reference.100) Similarly. using (10. also the l. Using · ∂t A(r. must transform likewise. t). t) 2 ∂t A(r.102) In this equation the r.100. Equation (10. ∂µ J µ (xµ ) = 0 .98) can be written. F µν = x Ey B z 0 −Bx Ez −By Bx 0 Alternatively.105) .s. 3 (10. according to (10. t) .99) Since this equation must be true in any frame of reference the right hand side must vanish in all frames. One can readily verify. Consequently.104) where mn = mn is the totally anti-symmetric three-dimensional tensor deﬁned in (5. a 4-vector. transforms like a 4-vector. has to transform like a contravariant 4-vector. must be a scalar. of (10.101). (10.s. (10.69).103) B . yields ∂µ ∂ µ Aν (xσ ) = 4 π J ν (xσ ) .97) and. F mn = − mn (10.s. t) from (8.101) Combining equations (10. k.6) and (8. 2.17) using the identity (8. m. Since ∂µ ∂ µ transforms like a scalar one can conclude that Aν (xσ ) must transform like a 4-vector.e. .96). t) = −∂t V (r. We want to derive now the diﬀerential equations which determine the 4-potential Aµ in the Lorentz gauge (10.98) which reﬂects the principle of charge conservation. · A(r.99) must be a scalar.. t) and current density J(r. n = 1. t) + · J(r. prove that Aµ is. t) − 2 V (r.77) and (10. in fact. t) = 4πρ(r.69) and ∂ µ in (10. (10. 10. one obtains for A(r.h. (10. it follows that J µ . (10. in fact.9). t) = ∂t · A(r.96). t) = −∂t V (r.h. t) = 0 (10.32). using (8. t) together with (10. that F µν can be expressed through the potential Aµ in (10. The respective equation for A0 = V can be obtained from Eq. The l. Since ∂µ transforms like a covariant 4-vector. i.

8.3: Relativistic Electrodynamics 299 The relationships (10. t) and B(r. one can conclude from (10.105) of the tensor F µν one can conclude the property ∂ σ F µν + ∂ µ F νσ + ∂ ν F σµ = 0 (10.109) (10. E⊥ = B⊥ E⊥ + v × B √ 1 − v2 B⊥ − v × E = √ 1 − v2 (10.108) yields then the expressions for the transformed ﬁelds E and B . From the deﬁnition (10.113) which can be shown to be equivalent to the two homogeneous Maxwell equations (8. B .104) establishe the transformation behaviour of E(r.111) One can readily prove that this equation is equivalent to the two inhomogeneous Maxwell equations (8. 10.4).105) and (10. equivalently. respectively. B and E⊥ . by (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 (10.106) In case that the Lorentz transformation Lµ ν is given by (10. the components of the ﬁelds parallel and perpendicular to the velocity v which determines the Lorentz transformation.103. .112) (10. t).102) ∂µ F µν = 4π J ν .10. B⊥ are. The results can be put into the more general form E B = = E . 8. 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. In a new frame of reference holds F µν = Lµ α Lν β F αβ (10. where we used (10.62) or. These equations show that under Lorentz transformations electric and magnetic ﬁelds convert into one another.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 .3. (10.2). Noting ∂µ F µν = ∂µ ∂ µ Aν − ∂µ ∂ ν Aµ = ∂µ ∂ µ Aν − ∂ ν ∂µ Aµ = ∂µ ∂ µ Aν .1.63).

67). of (10.87)] and retain the deﬁnition E for the electric ﬁeld. demonstrated that (10.5).5). . Equation (10. for µ = 1.114). hence. e.s. dt E From this one can conclude. however.120) (10. 1 dE 2 1 dp 2 = = qp·E 2 dt 2 dt (10.67) and charge q dpµ q = pν F µν dτ m (10.h.86). in fact. provides an alternative description of the action of the Lorentz force.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. E = m/ 1 − v 2 . The term on the r.g..5) where p = mv/ 1 − v 2 . hence. It holds for a particle with 4-momentum pµ as given by (10. To avoid √ confusion we will employ in the following for the energy of the particle the notation E = m/ 1 − v 2 [see (10.118) p· dt √ where we exploited the fact that according to p = mv/ 1 − v 2 holds p v. also from the equation of motion (8. For the spatial components. using (10.120) is referred to as the Lorentz force. employing (10.104). dE q = p·E dτ m or with (10. dτ m √ Employing again (10.93).e. i.114) where d/dτ is given by (10. equivalent to (8.116) (10.119) which is the x-component of the equation of motion (8.114) is.86) and (10.103) dpx q = ( EEx + py Bz − pz By ) . The µ = 0 component of (10.5) taking the scalar product with p dp = qp · E (10.. (10.115) This equation follows.117) (10.86) dE q = p·E . yields dpx = q dt Ex + (v × B)x (10. 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 (8.114) reads using (10. We have.114) reads then.

ψ ∈ C∞ (4).47). obey a diﬀerent transformation law. i.e. w. in as far as SO(3. In this section we will investigate the transformation properties of scalar functions ψ(xµ ). (10. def (10.126) ρ ewk Kk − 1 1 .123) This deﬁnition corresponds closely to the function space representation (5. (10. 10. The resulting expression should be a generalization of the function space representation (5. e. J2 . and which can be constructed following the procedure adopted for the function space representation of SO(3).g.3 or the bi-spinor wave function of electron-positron pairs in Sect.. w) which correspond to the generators Jk and Kk in (10. Accordingly.48) of SO(3). i.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 ) . the charge-current density in case of Sect.48) and (10. 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. (10.4 Function Space Representation of Lorentz Group In the following it will be required to decribe the transformation of wave functions under Lorentz transformations.121) covers solely the transformation behaviour of scalar functions.125) def (10. w) = ρ(Lµ ν (ϑ. w)) = ρ eϑ·J + w·K which we present in the form L(ϑ.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 . We need to emphasize that (10.125)]. we seek an expression which corresponds to (10.51) for the natural representation of the Lorentz group. For such functions holds in the transformed frame ψ (Lµ ν xν ) = ψ(xµ ) (10.. w) = exp ϑ · J + w · K . taken at the pairs of points (x µ = Lµ ν xν . Functions which represent 4-vectorsor other non-scalar entities. We like to express now ψ (x µ ) in terms of the old coordinates xµ .. We will denote the intended representation by L(ϑ.42) of SO(3). 10. (5.7.121) by (L−1 )µ ν xν and obtains ψ (xµ ) = ψ((L−1 )µ ν xν ) .1) is a generalization (rotation + boosts) of the group SO(3). J3 ) and K = (K1 .127) .e.124) In this expression J = (J1 . 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. xµ ). K3 ) are the generators of L(ϑ. However.4: Function Space Representation of Lorentz Group 301 10. in the present case we exclude the factor ‘−i’ [cf. For this purpose one replaces xµ in (10.10. K2 .

K2 ] = −J3 .131) This yields for small w1 ρ ew1 K1 ψ(xµ ) = ψ(coshw1 x0 + sinhw1 x1 . −sinϑ1 x2 + cosϑ1 x3 ) = ψ(xµ ) + ϑ1 (x3 ∂2 − x2 ∂3 ) ψ(xµ ) + O(ϑ2 ) . x1 . x3 ∂1 ] = −x1 ∂2 + x2 ∂1 = J3 .129) which yields for small ϑ1 ρ eϑ1 J1 ψ(xµ ) = ψ(x0 . x1 ∂3 ] − [x2 ∂3 . namely [J1 . the expression for K1 in (10. which we like to demonstrate for J1 and K1 . J ] [ Kk . 1 This result. K ] [ Jk . K ] = = k m Jm k m Jm = − k m Km . [ Jk .128).49) as the generators of the natural representation. cosϑ1 x2 + sinϑ1 x3 . and [J1 .e. x1 ∂3 − x3 ∂1 ] = [x3 ∂2 .130) ew1 K1 −1 = e−w1 K1 (10. (10.302 One obtains J1 = x3 ∂2 − x2 ∂3 .126). 0 1 (10. J2 = x1 ∂3 − x3 ∂1 . J3 = x2 ∂1 − x1 ∂2 . reproduces the expression for J1 in (10. x2 . One can determine similarly K1 starting from coshw1 sinhw1 sinhw1 coshw1 = 0 0 0 0 0 0 1 0 0 0 . K obey the same Lie algebra (10.133) We demonstrate this for three cases. [K1 .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. K2 ] = K3 : [ J1 . obviously. J2 ] = [x3 ∂2 − x2 ∂3 . obviously. (10. sinhw1 x0 + coshw1 x1 . The generators J . x3 ) 2 = ψ(xµ ) + w1 (x1 ∂0 + x0 ∂1 ) ψ(xµ ) + O(w1 ) (10.132) and. J2 ] = J3 .128) eϑ1 J1 = e−ϑ1 J1 1 0 = 0 0 0 0 0 1 0 0 0 cosϑ1 sinϑ1 0 −sinϑ1 cosϑ1 (10. i.134) . In order to evaluate (10.

141) (10. 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 .85-5.136) (10.. In both cases the radial coordinate is the quantity conserved under the transformations. x2 ∂0 ] − [x2 ∂3 . x0 ∂2 + x2 ∂0 ] = [x3 ∂2 . 303 (10. [ J1 . For this purpose one expresses K3 in terms of hyperbolic coordinates R. x0 ∂2 ] = x3 ∂0 + x0 ∂3 = K3 . x0 ∂2 + x2 ∂0 ] = [x0 ∂1 . ∂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 . (10.137) where K3 is given in (10.10. The relationships (10.135) (10.139) x3 = R sinhΩ (10. (5. K2 ] = [x0 ∂1 + x1 ∂0 . In the following we consider solely the case x0 ≥ 0.e. x2 + x2 + x2 1 2 3 in the case of SO(3) and (x0 )2 − (x3 )2 in case of transformation (10.138) x3 x0 .87)]. x2 ∂0 ] − [x1 ∂0 .140) allow one to ∂ ∂ express the derivatives ∂0 . i.137).4: Function Space Representation of Lorentz Group [ K1 . cothΩ = 3 . x3 as follows x0 = R coshΩ . 10. a relationship which can also be stated R = and + − (x0 )2 − (x3 )2 (x0 )2 − (x3 )2 if x0 ≥ 0 if x0 < 0 (10.125) in the case of a one-dimensional transformation of the type L(w3 ) = exp w3 K3 .128). K2 ] = [x3 ∂2 − x2 ∂3 . ∂Ω . One-Dimensional Function Space Representation The exponential operator (10. We note tanhΩ = ∂R x0 = . Eqs.139. Ω which are connected with x0 . can be simpliﬁed considerably. x0 ∂2 ] = −x2 ∂1 + x1 ∂2 = − J3 .142) (10.140) 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.143) . 0 ∂cothΩ ∂x x = ∂R x0 = − ∂x3 R (10. ∂3 in terms of ∂R .

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. ˆ pµ =⇒ pµ = i(∂t . however. ) = i ∂µ . The latter property implies that upon change of reference frame ψ(xµ ) transforms according to (10.304 Inserting these results into the deﬁnition of K3 in (10.148) ∂ µ ∂µ + m2 (10. The historic route to these two equations. Free Particle Case A quantum mechanical description of a relativistic free particle results from applying the correspondence principle. mass. which allows one to replace classical observables by quantum mechanical operators acting on wave functions.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 ) . Further below we will provide a more systematic.122). but certainly is short and. In the following we will employ so-called natural units = c = 1. in 4-vector notation reads pµ =⇒ pµ = i (∂t .121. (10. extremely useful. and (time)−1 all have the same dimension.92) one obtains the wave equation − or 2 2 µ ∂ ∂µ ψ(xν ) = m2 ψ(xν ) ψ(xν ) = 0 . In natural units the Klein–Gordon equation (10. In these units the quantities energy. 10.151) reads ∂µ ∂ µ + m2 ψ(xµ ) = 0 (10.147) (10. The partial diﬀerential equation (10.128) yields K3 = x0 ∂3 + x3 ∂0 = ∂ . We will provide a ‘derivation’ of these two equations which stem from the historical development of relativistic quantum mechanics. complex-valued function. (10. ∂Ω Relativistic Quantum Mechanics (10. (length)−1 .151) is called the Klein-Gordon equation. − ) = i ∂ µ . In the position representation the correspondence principle states ˆ E =⇒ E = − ∂t i ˆ p =⇒ p = i which.146) Applying the correspondence principle to (10. momentum. representation theoretic treatment.145) ∂Ω 10.150) . ˆ (10.144) The action of the exponential operator (10. namely the Klein–Gordon and the Dirac equations. therefore.149) where ψ(xµ ) is a scalar. is not very insightful.

t) ψ ∗ (r. t) ] 2mi (10. t) = · jKG (r. t) ψ(r. This follows from the fact that ∂µ ∂ µ and m2 are scalars. t) = (10. t) + where ρKG (r.154).153) is associated with a conservation law for particle probability ∂t ρS (r. (10. t) ) 2m 1 = ( ψ ∗ (r.154) describes the current density connected with motion of the particle probability distribution. t) = − 1 2m 2 ψ(r.10. t)∂t ψ ∗ (r. t) = ψ ∗ (r.161) . 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) + where ρS (r. t) − ψ(r.150) is invariant under Lorentz transformations. t) 1 [ ψ ∗ (r.152) which has the same form as the Klein–Gordon equation in the original frame.160) ∂t (ψ ∗ ∂t ψ which corresponds to − ψ∂t ψ∗) + · (ψ ψ∗ ∂t ρKG (r. t) − ψ(r. t)∂t ψ(r. t) jKG (r.155) describes the positive deﬁnite probability to detect a particle at position r at time t and where jS (r. and that (as postulated) ψ(xµ ) is a scalar. Under Lorentz transformations the free particle Klein–Gordon equation (10. t) ψ(r. t) = 0 (10. t) − ψ(r. 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. t) (10.5: Klein–Gordon Equation or 2 ∂t − 2 305 + m2 ψ(xµ ) = 0 .159) (10.156) · jS (r.150) becomes ∂µ ∂ µ + m2 ψ (x µ ) = 0 (10. t) = 0 i ( ψ ∗ (r. 2mi (10.151) One can notice immeadiately that (10. t) ψ ∗ (r. t) ) .158) which yields 2 2 ψ ∗ ∂t ψ − ψ∂t ψ ∗ − ψ ∗ 2ψ +ψ ψ∗ − 2ψ∗ = ψ) = 0 (10.157) 2m which is equivalent to (10. In order to obtain the conservation law connected with the Klein–Gordon equation (10.150) one considers Im ψ ∗ ∂µ ∂ µ + m2 ψ = 0 (10. t) ψ(r.

the Klein-Gordon equation is considered as a suitable equation to describe spin–0 particles. m 2 2 E0 = m2 + po . is actually that of a charge density. the anti-particles carry a charge opposite to that of the associated particles. p0 . i. (10. o namely. λ|xµ ) = 0 2 m2 + po . λ) associated with the corresponding wave functions ψo (p. We denote this by summarizing the solutions as follows ∂µ ∂ µ + m2 ψo (p. (10.161) and (10. λ|xµ ) (10. λ|xµ )ψo (p. and the density ρKG (p. λ) = λ ψo (p.164) The corresponding energy is Eo (po . for example pions. The proper interpretation of the two cases is that the Klein–Gordon equation describes particles as well as anti-particles.166) ψo (p. and of negative energies E ≤ −m.151) yields 2 2 Eo − p0 − m2 ψ(r.162) Inserting this into the Klein–Gordon equation (10. λ = ± . The density ρKG (p.162) shows that the solutions of the free particle Klein-Gordon equation (10.166) is Eo (p) ∗ ρKG (p. The proper interpretation of ρKG (r.163) which results in the expected [see (10. λ) actually describes charge density rather than probability. it had been realized later. (10. λ = ± (10. λ|xµ ) = Nλ. t). Today.93] dispersion relationship connecting E0 .165) This result together with (10. t) = 0 (10..e. not of particle probability.154). ±) = ± 2 m2 + po (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.150) are actually determined by po and by the choice of sign ±. corresponding to λ = +. λ|xµ ) = N e−iλmt .150) is a continuum of positive energies E ≥ m.167) m which is positive for λ = + and negative for λ = −. λ|xµ ) according to (10. according to (??) is decribed by the r–independent wave function ψo (p = 0. corresponding to λ = −. with p = 0.p ei(p·r − λEo (p)t) Eo (p) = The spectrum of the Klein–Gordon equation (10. in that the expression for ρKG is not positive deﬁnite. 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) . Generating a Solution Through Lorentz Transformation A particle at rest.168) .306 Relativistic Quantum Mechanics This conservation law diﬀers in one important aspect from that of the Schr¨dinger equation (10.

138) and the relationships (10. L(w3 )ψo (p = 0. A(r. one has to interprete p = mv/ 1 − v 2 (10.f.. For this purpose we consider the wave function for a particle moving with momentum velocity v in the direction of the x3 –axis. In case of λ = + one obtains ﬁnally L(w3 )ψo (p = 0. of (10..138) to replace t = x0 by hyperbolic coordinates R.173) (10.175) as the momentum of the particle and one obtains L(w3 )ψo (p = 0. In case of λ = −. λ|xµ ) = = exp w3 ∂ ∂Ω N e−iλmRcoshΩ 3) N e−iλmRcosh(Ω+w . for positive energy solutions. 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. Such wave function should be generated by applying the Lorentz transformation in the function space representation p (10.61) yield for this expression −iλ √ m mv x0 − iλ √ x3 .169) The addition theorem of hyperbolic functions cosh(Ω + w3 ) = coshΩ coshw3 + sinhΩ sinhw3 allows us to rewrite the exponent on the r. for the wave function (10. −A(r. Aµ (xµ ) = (V (r.168).166). t)) (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. t). for negative energy solutions. using (10. λ = +|xµ ) = N ei(px 3 − Eo (p)x0 (10. (10.e.177) . This yields.h. (10. Ω.174) which agrees with the expression given in (10. (10. i. 2 2 1−v 1−v (10. i.e.10.6 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.166)] of the particle. in fact. λ = −|xµ ) = N ei(px 3 + Eo (p)x0 . t)) . t).171) One can interpret then for λ = +.169) −iλ ( m coshw3 ) ( R coshΩ ) − iλ ( m sinhw3 ) ( R sinhΩ ) .145) choosing m = sinhw3 .170) The coordinate transformation (10.s.172) as the energy [c.176) 10.168). (10.

93).e. according to the rules shown in Table 10. t))2 ψ(r.. i. t) .308 free classical particle classical particle in ﬁeld (V. t) = −i − q A(r. m2 ) ψ(xµ ) = 0 . See also Eq. −A).179) (10.1: Coupling of a particle of charge q to an electromagnetic ﬁeld described by the 4-vector potential Aµ = (V. i. A) free quantum particle quantum particle in ﬁeld (V.150).147). According to the principle of minimal coupling [see (10. For this purpose we split-oﬀ a factor exp(−imt) which describes the oscillations of the wave function due to the rest energy.1. (10. we deﬁne ψ(r. and focus on the remaining part of the wave function. t) = e−imt Ψ(r. In terms of space-time derivatives this reads (i∂t − qV (r. (10.1 this becomes g µν (i∂µ − qAµ )(i∂ν − Aν ) ψ(xµ ) = m2 ψ(xµ ) which can also be written g µν πµ πν − . t) 2 (10.. (10. According to the so-called minimum coupling principle the presence of the ﬁeld is accounted for by altering energy. i. We will also assume. in . For this purpose one writes the Klein-Gordon equation (10. which couples a particle of charge q to an electromagnetic ﬁeld described through the potential Aµ (xν ).e.e.. that the mass m of the particle.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.182) and seek an equation for Ψ(r. momenta for classical particles and the respective operators for quantum mechanical particles in the manner shown. is much larger than all other energy terms. i∂t and −i in (10.150) −g µν (i∂µ )(i∂ν ) + m2 ψ(xµ ) = 0 .69)] one replaces the quantum mechanical operators. albeit in a form. t). in keeping withnthe non-relativistic limit.180) + m2 ψ(r.178) According to the replacements in Table 10. it’s rest energy. 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. t) . (10. A) or Aµ = (V. 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.

309 (10. Ψ |q V | << m .g. (10. ﬁltered out of the beam and ﬁnally fall as slow pions onto elements for which a pionic variant is to be studied. t) (10.. t) = − Ze and A(r. π − mesons are generated through inelastic proton–proton scattering p + p −→ p + p + π − + π + . (10.189) . i.183) The term on the l. identical to the Schr¨dinger equation (10. This application demonstrates that the Klein–Gordon equation describes spin zero particles. however. The process of π − meson capture involves the so-called Auger eﬀect.188) Inserting this into (10.185) This is.186) then are slowed down.e. t) ≡ 0 r + Ze2 r 2 + 2 − m2 π ϕ(r ) = 0 .183). larger than |i∂t Ψ/Ψ| and alrger than qV . t) 2m Ψ(r..184) where we neglected all terms which did not contain factors m. e.6: Klein–Gordon Equation with Electromagnetic Field particular. The Klein-Gordon equation (10.h. | i∂t Ψ | << m .181) reads then i ∂t Ψ(r.. t) = ˆ [p − q A(r. of (10.2) of a non-relativistic spin-0 particle o moving in an electromagnetic ﬁeld.e.s. A stationary state of the Klein–Gordon equation is described by a wave function ψ(xµ ) = ϕ(r ) e−i t . Pionic Atoms To apply the Klein–Gordon equation (10. (10. atoms in which one or more electrons are replaced by π − mesons. 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− .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.10. spin-0 mesons. The approximation is justiﬁed on the ground of the inequalities (10. (10. i. t)]2 + qV (r.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. To ‘manufacture’ pionic atoms.181) yields (we assume now that the Klein–Gordon equation describes a 2 particle with mass mπ and charge −e) for qV (r.181) to a physical system we consider pionic atoms.

195) The latter problem leads to the well-known spectrum En = − mπ (Ze2 )2 . (10. φ) R (r) Y r m (θ.. i. 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 . n − 1 .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.. this quantity deﬁnitely must be an integer.199) ( + 1) − Z 2 e4 . . n = 1. . . φ) (10. θ.e. . .e.191) m (θ. (10.193) (10.192) leads then to the ordinary diﬀerential equation d2 − dr2 ( + 1) − Z 2 e4 2 Ze2 + + r2 r 2 − m2 π R (r) = 0 . i. φ) . φ) = ( + 1) Y m (θ. .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. r r r r sin θ r sin θ Ze2 2 r ) (10. φ. the eigenfunctions of the operator L2 ˆ L2 Y m (θ. 1.194) and (10.190) 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 . .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 (10.191) ˆ The operator L2 in this equation suggests to choose a solution of the type ϕ(r ) = where the functions Y in (10. (10.197) counts the number of nodes of the wave function. (10. 2. (10.196) In this expression the number n deﬁned through n = n− −1 (10.192) ˆ are spherical harmonics.198) . The similarity of (10.

1.199) and deﬁning EKG (n. results in the (10.201) mπ Z 2 e4 m2 − m2 π π = − .197. . . − + 1. 2mπ 2 −→ λ( ) . . 2. 2 e2 −→ e2 mπ . Introducing α = Z 2 e4 and 1 β = + 2 (10. .10.e. . i. . 10. + (10. .200) The bound state solutions of this equation should correspond to from (10.203) spectrum . . .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 ) . .196) if one makes the replacement 2 values which can be obtained E −→ One obtains − m2 π . n = 0. (10. n − 1 m = − . m) = mπ 1 + Z 2 e4 (n− − 1 2 + ( + 1 )2 − Z 2 e4 )2 2 n = 1. .. . (10. . .6: Klein–Gordon Equation with Electromagnetic Field and one can write (10. 1. . = 0. 1.204) 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 ). EKG = Using (10. . . 2mπ 2 (n + λ( ) + 1)2 (10.202) Solving this for state) yields (choosing the root which renders = 1 + mπ Z 2 e4 ( n + λ( )+1 )2 ≤ mπ .194) d2 λ( ) ( λ( ) + 1 ) 2 Ze2 − + + dr2 r2 r 2 311 − m2 π R (r) = 0 . which corresponds to a bound = 0. . . .204) reads 1 (10.

one with a positive-deﬁnite density.208). i. in turn. is related to the equation E 2 = m2 + p 2 of the classical theory which involves a term E 2 . but not an operator iγ · as suggested by the principle of relativity (equivalence of space and time). a splitting of the six n = 2.312 ≈ From this results for (10.206) 1 2 3 + O(Z 6 e12 ) . but in the present case rather the negative of the energy. in fact. for example.. Finally. 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.209).204) EKG (n. would have only a ﬁrst order time derivative. it does not agree with observations of the hydrogen spectrum.208. However. the π − meson is a pseudoscalar particle. It must be pointed out here that does not denote energy. Dirac succeeded. the wave function changes sign under reﬂection.209) which. 4n (10. It should feature then a diﬀerential operator of the type D = iγ µ ∂µ . i. the Klein–Gordon equation had been rejected since it did not yield a positive-deﬁnite probability density. Many attempts were made by theoretical physicists to ‘linearize’ the square root operator in (10. a feature which is connected with the 2nd order time derivative in this equation. t) . i. His solution . 10. but for a long time to no avail. (10. 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 . arises because the Klein–Gordon equation. and the third term gives the leading relativistic correction. . Also. m) ≈ m − mZ 2 e4 mZ 4 e8 − 2n2 2n3 1 + − 1− Relativistic Quantum Mechanics α α2 α2 α2 − + + + O(α3 ) . In order to describe electron spectra one must employ the Lorentz-invariant 1 wave equation for spin. the Dirac equation introduced below..e. the second term the non-relativistic energy. = 1 states into groups of two and four degenerate states. 10. however. t) (10. through the correspondence principle. 10.2 particles. A more satisfactory Lorentz–invariant wave equation. The latter spectrum shows.e. Application of the correspondence principle (10.146) leads to the wave equation i∂t Ψ(r. t) = ± m2 − 2 Ψ(r. This derivative. 2n2 2βn3 8n4 4n4 (10.209). is identical to the two equations (10.7 The Dirac Equation Historically. however. The latter term agrees with observations of pionic atoms. Equations (10.208. are unsatisfactory since expansion of the square root operator involves all powers of the Laplace operator.e..208) These two equation can be combined i∂t + m2 − 2 i∂t − m2 − 2 Ψ(r.207) Here the ﬁrst term represents the rest energy.

(10. 2. spin. according to (10. one can impose the condition γ0 is hermitian . .. 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 ). γ j . For µ = ν condition (10. 3 . the realization that the linearization can be carried out only if one assumes a 4-dimensional representation of the coeﬃcients γ µ .212) where we have changed ‘dummy’ summation indices. . γ 1 . 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 . The discovery by Dirac. 2. γ ν ]+ = 2 g µν (10. one route to important discoveries in physics is the creation of new mathematical descriptions of nature. j = 1. unspeciﬁed algebraic objects γ µ and γ ν . We seek to identify the coeﬃcients γ µ . Instead. these descriptions ultimately merging with the true theory of matter. (10. therefore. It turns out that no 4-vector of real or complex coeﬃcients can satisfy these conditions.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 (10.7: Dirac Equation 313 to the problem involved an ingenious step.210) where P = iγ µ ∂µ . Inserting (10.211) Obviously.213) We want to determine now the simplest algebraic realization of γ µ . this was not so. j = 1.211) into (10. yet to be discovered. the quantities γ 0 .e. it was assumed that the 4-dimensional space introduced by Dirac could be linked to 4vectors.10. . exploited ∂µ ∂ν = ∂ν ∂µ . Initially. 3 has imaginary eigenvalues ±i. namely. but did not commute the. are associated with a positive-deﬁnite particle density. γ 2 . strengthens the believe of many theoretical physicists today that the properties of physical matter ultimately derive from a. the operator of the Klein–Gordon equation ∂µ ∂ µ + m2 = −( P + m ) ( P − m ) (10. Properties of the Dirac Matrices Let us now trace Dirac’s steps in achieving the linearization of the ‘square root operator’ in (10.1 2 and anti-particles. so far. Accordingly. beautiful mathematical theory and that. 2. i. ψd (xµ ). 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. achieved through a beautiful mathematical theory.66). However. Starting point is to boldly factorize. ψ2 (xµ ). (10. In fact. the 4-dimensionsional representation discovered by Dirac involved new physical properties.215) .209). quantities with the transformation law (10.208).213) reads (γ µ )2 = 1 µ = 0 −1 µ = 1. therefore. 3 are anti-hermitian .214) From this follows that γ 0 has real eigenvalues ±1 and γ j .

222) ∗ ∗ ∗ ∗ where we have deﬁned Ψ† = (ψ1 . A proper choice is γ0 = 1 1 0 0 −1 1 .218) of the Pauli matrices one can readily prove that condition (10. rather than to the right side.216) i.314 For µ = ν (10. in fact. 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. however.213) is satisﬁed. For d = 2 there exist only three anti-commuting matrices. as long as det(γ µ ) = 0 the dimension d of the square matrices γ µ must be even so that (−1)d = 1. (10. to construct four matrices γ µ for the next possible dimension d = 4. 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 ) . γj = 0 −σ j σj 0 . 1 (10. (10.. Relativistic Quantum Mechanics (10. namely the Pauli matrices σ 1 . (10.223) Inserting this into (10. σ j σ k = − σ k σ j for j = k .218) The Pauli matrices allow one.224) . 2. is unique. This property can also be written (γ µ )† = γ 0 γ µ γ 0 . holds 2 σj = 1 .217) Obviously.e. σ 3 for which.220) where Ψ(xµ ) represents a 4-dimensional wave function. σ 2 . ψ4 ) and where ∂µ denotes the diﬀerential operator ∂ µ operating to the left side. the γ µ are anti-commuting. The Adjoint Dirac Equation The adjoint equation is Ψ† (xµ ) i(γ µ )† ∂µ + m ← ← (10. (10. is implied by (10. ψ2 . rather than a scalar wave function.222) and multiplication from the right by γ 0 yields the adjoint Dirac equation Ψ† (xµ ) γ 0 iγ µ ∂µ + m ← =0.213) reads γ µ γ ν = −γ ν γ µ .221) =0 (10. in fact. ψ3 . except for similarity trasnformations.215). 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. We will argue further below that the choice f γ µ . One can readily show using the hermitian property of the Pauli matrices (γ 0 )† = γ 0 and (γ j )† = −γ j for j = 1. 3 which.219) Using property (10.

3 .229) The similarity transformation (10. αj = ˆ 0 σj σj 0 . j = 1. 2. j = 1. (10. 3 and ˆ β = 1 1 0 0 −1 1 . (10.7.Chiral Representation 315 The Dirac equation can be subject to any similarity transformation deﬁned through a non-singular ˜ 4 × 4–matrix S. 3 . (10. (10.227) leaves the algebra of the Dirac matrices unaﬀected and commutation property (10.226) (10.227) A representation often adopted beside the one given by (10.227).233) The eigenstates and eigenvalues of H correspond to the stationary states and energies of the particles described by the Dirac equation.230) Schr¨dinger Form of the Dirac Equation o Another form in which the Dirac equation is used often results from multiplying (10. 2. γ ˜ Exercise 10.10.7: Dirac Equation Similarity Transformations of the Dirac Equation .219) is the socalled chiral representation deﬁned through ˜ Ψ(xµ ) = S Ψ(xµ ) .231) ˆ whereα has the three components αj .1: Derive (refeq:Dirac-intro20a) from (10. j = 1. i. (10.228) (10. [˜ µ . ˆ ˆ ˆ Ho = α · p + β m .213). t) = 0 (10.. 2. and γ0 = ˜ 0 1 1 1 0 1 .e.225) .222. .213) still holds.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 ) . γj = ˜ 0 σj 1 S = √ 2 −σ j 0 1 1 1 1 1 −1 1 1 (10. γ ν ]+ = 2 g µν . 10.221) from the left by γ 0 ˆ ˆ i ∂t + i α · − β m Ψ(r. 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.

The Cliﬀord algebra C4 entails a subgroup G4 of elements ± dj1 dj2 · · · djs . Γ±8 = ±d1 d4 .235) can diﬀer only with respect to unitary similarity transformations. Γ±9 = ±d2 d3 Γ±10 = ±d2 d4 . 2. . any product 1 of the dj ’s can be brought to the form (10. .235) as can be readily veriﬁed from (10. Γ±7 = ±d1 d3 . Γ±11 = ±d3 d4 Γ±12 = ±d1 d2 d3 .e. One can conclude then that also any set of 4 × 4–matrices obeying (10. i. For example. i. (1955). Rev. This property extends then to 4 × 4–matrices which obey (10. d2 . d4 is called a Cliﬀord algebra C4 . Γ±14 = ±d1 d3 d4 . There are (including the diﬀerent signs) 32 elements in G4 which we deﬁne as follows Γ±1 = ±1 1 Γ±2 = ±d1 . Γ±5 = ±d4 Γ±6 = ±d1 d2 .. page 187. satisfy the anti-commutation property dj dk + dk dj = 2 for j = k 0 for j = k Relativistic Quantum Mechanics d4 = γ 0 (10. 3 .213). Γ±15 = ±d2 d3 d4 Γ±16 = ±d1 d2 d3 d4 (10. The reader may . j = 1. The representations of Cliﬀord algebras Cm are well established. d2 = . Since the Γj are hermitian the similarity transformations are actually given in terms of unitary transformations. form a Cliﬀord algebra C3 . Γ±4 = ±d3 . To complete the proof in this section the reader may consult Miller ‘Symmetry Groups and their Application’ Chapter 9.238) These elements form a group since obviously any product of two Γr ’s can be expressed in terms of a third Γr . d3 . 27.Good.236) where ‘⊗’ denotes the Kronecker product between matrices. hence. 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 .. The associative algebra generated by d1 .Mod.316 Cliﬀord Algebra and Dirac Matrices The matrices deﬁned through dj = iγ j . Γ±3 = ±d2 . to Dirac matrices.6 and R. d4 = 1 0 0 −1 1 0 0 −1 0 1 1 0 0 i −i 0 (10.237) by means of the property (10.237) which are the ordered products of the operators ±1 and d1 . d4 . The three Pauli matrices also obey the property (10. m = Aj Bkm .213). j1 < j2 < · · · < js s ≤ 4 (10.e.234) (10.235) and. Obviously. in case of C4 .Phys.235). the matrix elements of C = A ⊗ B are Cjk. Γ±13 = ±d1 d2 d4 . The representations of this group are given by a set of 32 4 × 4–matrices which are equivalent with respect to similarity transformations.H.

. and derivatives according to (10. namely. Exercise 7. these properties determine the matrices γ µ uniquely. (10. Hence. in frames connected by Lorentz transformations. Exercise 7.6).218) of the Dirac matrices γ µ .e.241) . 10.. i. 10.219). transform coordinates and derivatives of the Dirac equation. i. j = 1.2: Demonstrate the anti-commutation relationships (10. exist. i.213. Exercise 7. Except for a similarity transformation. we do not know yet how to transform the 4-dimensional wave function Ψ and the Dirac matrices γ µ ..239) Form invariance implies that the matrices γ µ should have the same properties as γ µ .80).240) . 3 has imaginary eigenvalues ±i and can be represented by an anti-hermitian matrix.4: Show that from (10.80) by Lµ ρ and using (10. 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. and γ j . Lorentz Transformation of the Bispinor State Actually. x µ = Lµ ν xν (10. in fact. However.e. µ µ (10. in a moving frame should hold iγ ∂µ − m Ψ (x ) = 0 .3: Demonstrate the anti-commutation relationships (10.76) yields ∂ρ = Lν ρ ∂ ν . The latter transformations imply that coordinates transform according to (10.215). x = Lµ ν xν . i.214) follows that γ 0 has real eigenvalues ±1 and can be represented by a hermitian matrix.218) of the Pauli matrices σ j . a result one could have also obtained by applying the chain rule to (10.. The transformation is assumed to be linear and of the form Ψ (x ) = S(Lµ ν ) Ψ(xµ ) µ µ (10.8: Lorentz Invariance of the Dirac Equation 317 also want to establish the unitary transformation which connects the Dirac matrices in the form (10. it must hold γ µ = γ µ . the form of the Dirac equation is identical in equivalent frames of reference..e. Multiplication and summation of (10. therefore. i.e. Inﬁnitesimal Bispinor State Transformation We want to show now that a suitable transformation of Ψ(xµ ) does. 2.e.6).8 Lorentz Invariance of the Dirac Equation We want to show now that the Dirac equation is invariant under Lorentz transformations.236) with the Dirac representation (10. We can. in a moving frame holds iγ µ ∂µ − m Ψ (x ) = 0 .10.

244) from which. The inﬁnitesimal transformation S(Lρ σ ) which corresponds to (10. using gρµ γ µ = γρ . of the equation.243) in a form in which the Lorentz transformation in the form Lµν is on the r. g is an anti-symmetric matrix. The elements of g are. i. hence. For this purpose we exploit (10. (10. however. state that the Dirac equation (10.. Multiplication of this property by g from the right yields ( g)T = − g.247) the inﬁnitesimal matrix µν is anti-symmetric.248) Here σµν denote 4 × 4–matrices operating in the 4-dimensional space of the wave functions Ψ. . (10. Ob1 viously.249) from which we can conclude σµν = −σµν νµ = −σνµ i. We had introduced in (10. 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. (10. One can ﬁnally conclude multiplying both sides by g ρµ S(Lη ξ )γ µ S −1 (Lη ξ ) = Lν µ γν . it must hold (10.241) implies a similarity transformation Sγ µ S −1 .221) upon Lorentz transformation yields iS(Lη ξ )γ µ S −1 (Lη ξ ) Lν µ ∂ν − m Ψ (x ) = 0 .247) can be expanded S( µν i ) = 1 − σµν 1 4 µν (10.e.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.e. µ (10.s.h.318 Relativistic Quantum Mechanics where S(Lµ ν ) is a non-singular 4 × 4–matrix.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 . νµ (10.12) in the form Lν µ gνσ Lσ ρ = gµρ = gρµ . the transformation (10. in the expression of the inﬁnitesimal transformation ρ Lµν = g µν + µν (10. It holds then S( µν i ) = 1 − σµν 1 4 µν µν = S(− νµ i ) = 1 + σµν 1 4 µν . S( µν ) should not change its value if one replaces in its argument µν by − νµ .246) (10.245) The construction of S(Lη ξ ) will proceed using the avenue of inﬁnitesimal transformations. µ g ρν = µν and. One can. follows S(Lη ξ )γρ S −1 (Lη ξ ) = Lσ ρ γσ ..243) from the left by Lσ ρ gσν yields S(Lη ξ )γ µ S −1 (Lη ξ ) gρµ = Lσ ρ gσν γσ .243) We want to determine now the 4 × 4–matrix S(Lη ξ ) which satisﬁes this condition. The proper starting point for a constructiuon of S(Lη ξ ) is actually (10. hence. Multiplication of (10.250) σµν = −σνµ .

254) The proper σαβ must be anti-symmetric in the indices α. (10. 10. the chiral representation introduced above in Eqs. a solution of condition (10.8: 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. For this purpose one needs to express formally the r.s.256) One can readily verify that the non-vanishing generators σµν are given by (note σµν = −˜νµ .255) satisfy condition (10. −γ ) are ˜ γ ˜ γ0 = ˜ 0 1 1 1 0 1 .257) .e.10. 10. (10. (10. 3 . of (10.254).248.251) into (10.h.s. ˜ ˜ σ only six generators need to be determined) σ0j = ˜ i [˜0 .h.254) is σαβ = i [ γα .252) also as a sum over both indices of αβ .252) Since six of the coeﬃcients αβ can be chosen independently. Furthermore.s. must be symmetric with respect to interchange of the indices α and β.251) S −1 ( µν ) = 1 + σµν µν 1 4 Inserting (10..s. like the expression on the l. γj ] = γ ˜ 2 −iσ j 0 0 iσ j . In fact. γβ ]− 2 (10. 2.246) results then in a condition for the generators σµν − i ( σαβ γ µ − γ µ σαβ ) 4 αβ = νµ γν .h. (10.213.216) of the Dirac matrices. of (10.253) 1 2 αβ ( δ µ β γα − δ µ α γβ ) . γk ] = ˜ γ ˜ jk σ 0 0 σ . Algebra of Generators of Bispinor Transformation We want to construct the bispinor Lorentz transformation by exponentiating the generators σµν . β [ σαβ . For this purpose we inspect the properties of the generators in a particular representation. γj = ˜ 0 −σ j σj 0 .252) results in the condition for each α. 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. this condition can actually be expressed through six independent conditions.h. γ µ ]− = 2i ( δ µ β γα − δ µ α γβ ) . β and operate in the same space as the Dirac matrices. For this purpose we need to verify that the algebra of the generators involving addition and multiplication is closed. i.5: Show that the σαβ deﬁned through (10. σjk = [˜j .229). the expression on the r.255) which can be demonstrated using the properties (10. (10.. namely. Comparing this with the l.228. Exercise 7. 10. j = 1. In this representation the Dirac matrices γµ = (˜ 0 .

241) bispinor wave functions from one frame of reference into another frame of reference. We can ﬁnally note that the closedness of the algebra of the generators σµν is not aﬀected by similarity transformations and that.320 Relativistic Quantum Mechanics Obviously. the algebra of these generators is closed under addition and multiplication.261) We note that this operator is block-diagonal. One should. ϑ) = e 2 (w − iϑ)·σ 0 − 1 (w + iϑ)·σ 0 e 2 1 (10. 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 = . therefore. In fact.259) We had stated before that the transformation S is actually determined through the Lorentz transformation Lµ ν . (10. .262) This expression allows one to transform according to (10..51). therefore. Finite Bispinor Transformation The closedness of the algebra of the generators σµν deﬁned through (10.259) becomes in the chiral representation ˜ S(w. not necessarily inﬁnitesimal. since both operations convert block-diagonal operators A 0 0 B (10.44) 0 −w1 −w2 −w3 w 0 ϑ3 −ϑ2 = 1 w2 −ϑ3 0 ϑ1 w3 ϑ2 −ϑ1 0 µν (10. be able to state S in terms of the same parameters w and ϑ as the Lorentz transformation in (10. i. 2 0 −(w + iϑ) · σ 13 + σ23 ˜ 23 ) (10. any representation of the generators. in particular.51) and S. one can express the tensor µν through w and ϑ using µν = µ ρ g ρν and the expression (10. µν in the exponential form i S = exp − σµν 4 µν . 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.e.248) allows us to write the transformation S for any.255) yields a closed algebra.260) Inserting this into (10.258) again into block-diagonal operators. and since the algebra of the Pauli matrices is closed. In order to construct an explicit expression of S in terms of w and ϑ we employ again the chiral representation.

266) ρ(xµ ) = Ψ† (xµ )Ψ(xµ ) = s=1 |ψs (xµ )|2 (10. For our proof we note ﬁrst ˜ ˜ S −1 (w. This transformation behaviour can also be deduced from the transformation properties of the bispinor wave function Ψ(xµ ).221) from the left by Ψ† (xµ )γ 0 . Since ∂µ transforms like a covariant 4-vector. (10. (10.263) allows one to conclude that j µ must transform like a contravariant 4-vector as the notation implies.269) One can readily show that the same operator is obtained evaluating γ S (w. The reason is that the r. j) = Ψ† (xµ )γ 0 γ µ Ψ(xµ ) . −ϑ) = e− 2 (w − iϑ)·σ 0 1 (w + iϑ)·σ 0 e2 1 (10. j µ must transform like a contravariant 4-vector.268) holds for the bispinor Lorentz transformation.e.h. for j µ (xµ ) = (ρ. (10. (10.263) obviously is a scalar under Lorentz transformations and that the left hand side must then also transform like a scalar. We will now determine the relationship between the ﬂux j µ = Ψ (x )γ 0 γ µ Ψ (x ) † µ (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 .268) We will prove this property in the chiral representation. in fact.265) (10. For this purpose we prove ﬁrst the relationship S −1 = γ 0 S † γ 0 .264) (10. of (10.224).224) from the right by Ψ(xµ ). the property applies then in any representation of S. The conservation law (10. and addition yields Ψ† (xµ )γ 0 The last result can be written ∂µ Ψ† (xν )γ 0 γ µ Ψ(xν ) = 0 .267) has the desired property of being positive deﬁnite. ϑ) = S(−w.263) Starting point are the Dirac equation in the form (10.s..263) does hold.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 .10. the conservation law (10. Obviously. The time-like component ρ of j µ 4 iγ µ ∂µ + iγ µ ∂µ ← Ψ(xµ ) = 0 . Multiplying (10.221) and the adjoint Dirac equation (10.271) . i. (10.270) We conclude that (10.

275) The free particle wave function is an eigenfunction of Ho . 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. As it turns out. t) and is determined through the momentum p ∈ R3 . a property which leads to the energy– momentum (dispersion) relationship of the Dirac particle. In relativistic quantum mechanics a Dirac particle can also be characterized through a momentum. (10.283)].231.271) that the Dirac matrices are independent of the frame of reference. The operators commute also with Ho in (10.274) 10.232. bispinor wave functions and we need to determine corresponding components of the wave function.272) results in the expected transformation behaviour µ j = Lµ ν j ν . Combining this with (10.282.246) S −1 (Lη ξ )γ µ S ( Lη ξ ) = (L−1 )νµ γν = (L−1 )ν γ ν = Lµ ν γν . The solutions provide also a complete. the wave function has four components which invite further characterization of the free particle state.233). the identiﬁcation of observables and their quantum mechanical operators. as is required for the mentioned property. These operaˆ tors. (10. (10. additional characterizations.233) o i∂t Ψ(xµ ) = Ho Ψ(xµ ) . commute with each other.322 Relativistic Quantum Mechanics in a moving frame of reference and the ﬂux j µ in a frame at rest. speciﬁc for the Dirac free particle.9 Solutions of the Free Particle Dirac Equation We want to determine now the wave functions of free particles described by the Dirac equation. 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)].273) where we have employed (10. not only as the most simple demonstration of the theory.76). 10. Like in non-relativistic quantum mechanics the free particle wave function plays a central role. of which the wave functions are eigenfunctions as well. Note that we have assumed in (10. 10. (10. In the following we want to provide this characterization.268) µ j = Ψ† (xµ )S † γ 0 γ µ SΨ(xµ ) = Ψ† (xµ )γ 0 S −1 γ µ SΨ(xµ ) .e. One obtains using (10. only two degrees of freedom of the bispinor four degrees of freedom are independent [c. the Dirac particles are described by 4-dimensional. For this purpose we consider ..f. µ (10. In case of non-relativistic quantum mechanics the free particle wave function has a single component ψ(r. i. but also as providing a basis in which the wave functions of interacting particle systems can be expanded and characterized. as just mentioned. We will start from the Dirac equation in the Schr¨dinger form (10. The additional degrees of freedom described by the four components of the bispinor wave function require. As pointed out.272) With S −1 (Lη ξ ) = S((L−1 )η ξ ) one can restate (10. 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. however.

282.94) Equation (10.10.279) the well-known relativistic dispersion relationship 2 = m2 + p 2 (10. Inserting (10. (10. We want to show now that these degrees of freedom correspond to a spin-like property. and φo . +m (10.276) into (10. the Dirac equation.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 .280) which has a positive and a negative solution = ± E(p) . hence. two-dimensional spinor state. 10. .277) explicitly ( − m) 1 φo − σ · p χo 1 − σ · p φo + ( + m) 1 χo 1 = 0 = 0.276).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 (10.276) where p and together represent four real constants.281) Obviously. like the Klein–Gordon equation.281).93. E(p) = m2 + p 2 . 2 − m2 0 (10.282) (10. 10.284) The relationships (10. from (10.283) where is deﬁned in (10.234) and (10. the socalled helicity of the particle. conclude 1.280) σ·p (σ · p)2 φ0 = χ0 = +m ( + m) ( + m) p2 χ = χ0 . χo each represent a constant.277) To solve this problem we write (10. namely φo and χo are related as follows φ0 χo = = σ·p χ0 −m σ·p φ0 . one ﬁnds using (5.231.279) According to the property (5. In fact.234) of Pauli matrices holds (σ · p)2 = p 2 1 One can.278) provides us with information about the components of the bispinor wave function (10.283) imply that the bispinor part of the wave function allows only two degrees of freedom to be chosen independently. These two relationships are consistent with each other. later to be identiﬁed with momentum and energy. reproduce the classical relativistic energy–momentum relationships (10. 10. 1 (10.232) leads to the 4-dimensional eigenvalue problem m σ·p σ · p −m φo χo = φo χo . (10. (10.

±) γ 0 Ψ(0..s. The form of this condition and of (10. = −E(p) . 10.287) will be justiﬁed now. according to (10. the solution for = +E(p). is invariant under Lorentz transformations.290) which diﬀers from the normalization condition (10. according to (10. (10. Hence.287) the form of which will be justiﬁed further below.288) corresponding to the wave function Ψ(p.290).e.s. (10. We want to demonstrate now that the signs on the r. ±). of (10. the l. i. i.259). of (10. (10.285)].287.h. For the positive energy solution. S −1 = γ 0 S † γ 0 and.281)] separately. +) = −1 . using γ 0 as given in (10. 10. (10. ±) = S Ψ(0. The connection between the solutions. First. +) γ 0 Ψ† (p.286) Here N+ (p) is a constant which will be chosen to satisfy the normalization condition Ψ† (p.. (10. (10. and let Ψ(0. ±) γ 0 Ψ(p.324 Relativistic Quantum Mechanics For our further characterization we will deal with the positive and negative energy solutions [cf.291) Note that we have used that. (10. 0) γ 0 u 0 = |N+ (0)|2 .e. ±) S † γ 0 S Ψ(0. (10. ±). i. ±) denote the corresponding solution of a particle in its rest frame. u† u = |u1 |2 + |u2 |2 = 1 .. Ψ† (0. is Ψ(p. +|xµ ) = N+ (p) u σ·p E(p) + m u ei(p·r − t) . (10. S † = γ 0 S −1 γ 0 .s. through χo given by χo = u1 u2 = u ∈ C2 . ±) denote the solution of a free particle moving with momentum p in the laboratory frame. +) = |N+ (0)|2 (u† .268). (10. One can see this as follows: Let Ψ(p.h. ±)γ 0 Ψ(p. ±) γ 0 S −1 γ 0 γ 0 S Ψ(0.289) Here N− (p) is a constant which will be chosen to satisfy the normalization condition Ψ† (p. we present φo through the normalized vector φo = u1 u2 = u ∈ C2 . +) γ 0 Ψ(0. where S is given by (10.219) and u† u = 1 [c. u† u = |u1 |2 + |u2 |2 = 1 . ±) . we demonstrate that the product Ψ† (p. Ψ† (p. ±) = = = Ψ† (0. we present the negative energy solution.h.287) in the minus sign on the r. +) = 1 . For this purpose we consider ﬁrst the positive energy solution. hence. ±) Ψ† (0. Similarly. the solution for = −E(p).f. ±) Ψ† (0.241). +) γ 0 Ψ† (p.292) .285) The corresponding free Dirac particle is then described through the wave function Ψ(p.e. Employing (10.290) should diﬀer.286) for p = 0 yields.287. = +E(p) . −|xµ ) = N− (p) −σ·p E(p) + m u u ei(p·r − t) .

u† ) γ 0 0 u = − |N− (0)|2 . 2 E(p) (10.298) This result completes the expression for the wave function (10. (10.300) (10.285) results in 2 N+ (p) 1 − p2 (E(p) + m)2 = 1 (10.h.289) yields Ψ† (0.f.6: Show that the normalization condition 2 N+ (p) (u∗ )T . −) γ 0 Ψ(0. σ·p u∗ E(p) + m T γo u σ·p E(p) + m u = 1 (10.9: Solutions of the Free Particle Dirac Equation The same calculation for the negative energy wave function as given in (10.286) is 2 N+ (p) (u∗ )T .10.302) .296) Replacing (σ · p)2 by p 2 [c.298) becomes N+ (p) = m + E(p) .s. σ·p u∗ E(p) + m T u σ·p E(p) + m u = 1 (10.s. −) = |N− (0)|2 (0. this requires the choice of a negative side on the r. Condition (10. We consider ﬁrst the positive energy solution.219) yields 2 N+ (p) u† u − u† (σ · p)2 u (E(p) + m)2 = 1.301) yields the normalization coeﬃcient N+ (p) = m + E(p) .286).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. (5.295) Evaluating the l.h. 325 (10. Exercise 7.287) written explicitly using (10. ( m + E(p) )2 − p 2 (10.293) Obviously.290) to assign a positive value to |N− (0)|2 .299) ( m + E(p) )2 . of (10. We can also conclude from our derivation N± (0) = 1 . using γ 0 as given in (10. 2m (10.294) We want to determine now N± (p) for arbitrary p. (10.234)] and using the normalization of u in (10.

λ|xµ ) = λ E(p) Ψ(p. deﬁned as the component of the particle spin along the direction of motion.289) are not completely speciﬁed.286) into (10.e.275).e. The corresponding operator which measures this observable is 1 p ˆ Λ = σ· . hence..289). The wave functions (10. j = 1. The wave functions are also eigenstates of the momentum operator i∂µ . 10. The commutation property [σ · p.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. Inserting (10.275) yields Ho Ψ(p. Rather than considering the ˆ observable (10.s.305) and. This attribute is the so-called helicity. 3 is fairly obvious. This degree of freedom 1 describes a spin– 2 attribute.289) is T −σ·p −σ · p 2 E(p) + m u N− (p) u∗ . λ|xµ ) . The property [σ · p. Helicity The free Dirac particle wave functions (10. not a constant vector.. identical to the condition (10.300) have been constructed to satisfy the free particle Dirac equation (10.286. pj ] = 0 . i. however.303) E(p) + m u Evaluating the l.290) written explicitly using (10.h. λ|xµ ) = pµ Ψ(p. i∂µ Ψ(p. thereby.296) for the normalization constant N+ (p) of the positive energy solution.326 Relativistic Quantum Mechanics We consider now the negative energy solution. the wave functions constructed represent eigenstates of Ho .309) . the two components of u indicate another degree of freedom which needs to be deﬁned. yields 2 N− (p) u† (σ · p)2 u − u† u (E(p) + m)2 = −1 .307) we investigate ﬁrst the observable due to the simpler operator σ · p. have completed the determination for the wave function (10. 2. We can. (10. λ|xµ ) (10.308) 2 |p| ˆ Note that p represents here an operator. i. 10.304) This condition is. (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.286. (u∗ )T γ o = −1 (10. conclude N− (p) = m + E(p) 2m (10. (10. 10.. −p). Ho ] = 0 follows from (10.306) i. λ|xµ ) in 4-vector notation.289. exp[i(p · r − t)] = exp(ipµ xµ ).e. Condition (10.307) where pµ = ( . This can be veriﬁed expressing the space–time factor of Ψ(p.

. . 2m (10.312). 0)T and (0. − 1 |r.310) We have shown altogether that the operators p.308) above. t) = Np e (10.312) We assume now particles with negative energy. can ˆ ˆ be simultaneously diagonal. +. Therefore. t) = Np e 1 0 0 p m + Ep i(px3 + E t) 1 p e Ψ(pˆ3 . According to the deﬁnition (5.311) 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. −. 327 0 (σ · p)2 ˆ 2 (σ · p) ˆ 0 0 (σ · p)2 ˆ 2 (σ · p) ˆ 0 (10. = +E(p).e. −. i. hence. 1)T are eigenstates of 2 σ 3 with eigenvalues ± 1 . + 2 |r.224) 1 of σ3 the two u vectors (1. +. States which are simultaneously eigenvectors of these three operators are also simulteneously eigenvectors of the three operators p. − 1 |r.e. In this case Λ = 2 σ 3 . p3 ). We consider ﬁrst the simplest case that particles move along the x3 –direction. Ho and σ·p commute with each other and... are 1 i(px3 − E t) 0 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 .286) are eigenfunctions of Λ as well will specify now the vectors u. 2 the wave functions which are eigenstates of the helicity operator. t) = Np e 2 1 p m + Ep 0 0 i(px3 − E t) 1 p e Ψ(pˆ3 . Np = m + Ep . is independent of the direction of motion of the particle. + 1 |r. = −E(p).10. ˆ The condition that the wave functions (10. i. 0. Ho and Λ deﬁned in (10. t) = Np e (10. 1 p = (0.9: 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 = 0.313) 2 0 1 where Ep and Np are deﬁned in (10. We assume ﬁrst particles with positive energy.e. in principle. i.

e. + 2 |r. p) e |p| (10. The corresponding wave functions are determined through ˜ i˜ 0 ∂t − m Ψ(t) = 0 . 10. + 1 |r. −. 5.229). + 1 ) 2 p 1 m + Ep u(p. + 2 ) p 1 m + Ep u(p.312) can be obtained also by means of the Lorentz transformation (10. +.262) for the bispinor wave function and the transformation (10. 1)T have to be replaced by eigenstates u± (p) of the spin operator along the direction of p. 10. t) = Np ei(p·r + Ep t) (10.222.312).315) e3 × p ˆ ∠(ˆ3 .328 Relativistic Quantum Mechanics To obtain free particle wave functions for arbitrary directions of p one can employ the wave functions (10. Generating Solutions Through Lorentz Transformation The solutions (10.223) as follows 1 u(p. 10. + ) 2 1 u(p. These eigenstates are obtained through a rotational transformation (5. (10. + 2 ) u(p.316) describes a rotation which aligns the x3 –axis with the direction of p. t) = N√ ei(p·r + Ep t) (10. t) 2 = Np 1 u(p.318) 1 Ψ(p. i. +. − 2 ) −p 1 m + Ep u(p.317) Ψ(p. 1 |t) = √2 e−imt . − ) 2 where ϑ(p) = = = exp exp i − ϑ(p) · σ 2 i − ϑ(p) · σ 2 1 0 0 1 . denoted by ˜. − 1 ) 2 −p 1 m + Ep u(p.220).311. γ . for an r–independent wave function. For this purpose one starts from the solutions of the Dirac equation in the chiral representation (10.319) 1 Ψ(p. 2 1 0 (10. [One can also express the rotation through Euler angles α. a wave function which represents free particles at rest. − 2 |r.321) .226.320) where Ep and N are again given by (10.] The corresponding free particle wave functions are then Ψ(p. − 1 ) 2 ei(p·r − Ep t) (10. +. − 1 |r. β. − 2 ) u(p. −.311..123). t) 2 = Np u(p.313) except that the states (1. in which case the transformation is given by (5. 0)T and (0. + 2 ) ei(p·r − Ep t) (10. γ and are 1 0 1 ˜ Ψ(p = 0.314) (10.

322) The reader can readily verify that transformation of these solutions to the Dirac representationsas deﬁned in (10.262) φo χo → e 2 w3 σ 0 1 3 Et ) (10..311. −. 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 e −1 329 (10. − 1 |t) = √2 . + 1 |r. +. − 1 |t) = √2 e−imt .324) 0 e − 1 w3 σ 3 2 φo χo = e 2 w3 σ φo 1 3 e− 2 w3 σ χo 1 3 . 10. χo are eigenstates of σ 3 with eigenvalues ±1. ± 1 of the wave functions stated in (10.325) to (10. w3 ). 2 The solutions (10. 0. χo ∈ 1 0 .324.e. −1 e 0 0 1 +imt 1 ˜ Ψ(p = 0. 2 |t) = √2 . +. yields for the exponential space–time dependence according to (10.322) can be written in spinor form 1 √ 2 φo χo e imt .10. 2 0 1 1 0 +imt 1 1 ˜ Ψ(p = 0. φo .322).176) imt → i ( p3 x3 and for the bispinor part according to (10. Applying (10. −.262) for a boost in the x3 –direction.323) should yield the solutions for non-vanishing momentum p in the x3 –direction. 10. For the resulting wave functions in the chiral representation one can use then a notation corresponding to that adopted in (10. 10. +.228) yields the corresponding solutions (10. t) e 2 e− 2 w3 1 = √ 1 2 e 2 w3 i(px3 − E t) p e . − 1 |r. This correspondence justiﬁes the characterization ±.323) Transformation (10. i.311) e 1 w 2 3 ˜ Ψ(p(w3 )ˆ3 . (10. 0 1 (10.313) in the p → 0 limit.174.325) One should note that φo .9: Solutions of the Free Particle Dirac Equation 0 1 1 ˜ Ψ(p = 0. for w = (0.

326) where according to (10. − 2 |r. + 2 |r. 2 (10. −.61) between the parameter w3 and boost velocity v3 . 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 .61) p(w3 ) = m sinhw3 . 2 sinh x = 2 coshx − 1 . and the expression (10. − 1 |r. +.328) the relationship (10. −. −. Transformation to the Dirac representation by means of (10. +. t) e 2 = i(px3 + E t) p e 1 Ψ(p(w3 )ˆ3 . 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 .327) Employing the hyperbolic function properties cosh x = 2 coshx + 1 . t) e = i(px3 + E t) p e (10. − 2 |r.329) .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 (10.330 e 1 w 2 3 Relativistic Quantum Mechanics 1 0 1 0 0 1 0 1 1 ˜ Ψ(p(w3 )ˆ3 . + 1 |r. + 2 |r. −.228) yields w3 cosh 2 = sinh w3 2 1 Ψ(p(w3 )ˆ3 . t) e 2 e 1 = √ 1 2 − e 2 w3 i(px3 + E t) p e (10. t) e 1 = √ 2 −e− 2 w3 − 1 w3 2 1 i(px3 + E t) p e ˜ Ψ(p(w3 )ˆ3 .

Λ|xµ ) from the solutions describing particles at rest Ψ(p = 0. To show this we rewrite the Dirac equation (10. refered to as the active transformation.329.9: Invariance of Dirac Equation Revisited and similarly sinh w3 = 2 Ep − m = 2m p 2m (Ep + m) 331 (10. 10. 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 (10.243) and. Such description will be essential for the formal description of Lorentz invariant wave equations for arbitray spin further below. the essential property stating the Lorentz–invariance of the Dirac equation.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.334) (10.330) into (10.331) 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 ). λ.10.327).221) using (10. indeed. (10. of course.332) Here we have made use of the fact that S(Lη ξ ) commutes with ∂µ and ρ(Lη ξ ) commutes with γ µ . Λ|t). Such transformation had been applied by us. reproduces the positive energy wave functions (10. and since (2) the calculations following (10. that if Ψ(xµ ) is a solution of the Dirac equation that Ψ (xµ ) as given in (10. the proper transformation S(Lη ξ ).311) as well as the negative energy solutions (10.176).168–10.330) Inserting expressions (10.123) above and characterized there.331) is a solution as well. = γν . In the new derivation we consider the particle described by the wave function transformed. thereby. in general. The reason why we provide another derivation of (10. ρ(Lη ξ ) has been deﬁned in (10. Making this expectation a postulate allows one to derive the condition (10.243). expresses the system in the old coordinates.313) for −p. but not the observer. which however.243) considered solely the limit of inﬁnitesimal transformations anyway. λ. 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 . when we generated the solutions Ψ(p. we will derive this equation only for inﬁnitesimal transformations. This transformation.333) . Actually. The transformation is Ψ (xµ ) = S(Lη ξ ) ρ(Lη ξ ) Ψ(xµ ) (10. Invariance of Dirac Equation Revisited At this point we like to provide a variation of the derivation of (10.331) i S(Lη ξ )γ µ S −1 (Lη ξ ) ρ(Lη ξ )∂µ ρ−1 (Lη ξ ) − m Ψ (xµ ) = 0 (10. is suﬃcient since (1) it must hold then for any ﬁnite transformation. We expect.

∂µ ] = (10. ∂µ ] = (10. 0). that any solution Ψ(xµ ) transformed according to (10. it is met by S(Lη ξ ) as given in the chiral representation by (10. 0).334) for inﬁnitesimal Lorentz transformations Lη ξ . 1. ∂µ ] = [x0 ∂1 + x1 ∂0 .332 Relativistic Quantum Mechanics We will demonstrate now that the ﬁrst condition is satisﬁed by ρ(Lη ξ ).331) is again a solution of the Dirac equation. 0.337) [J2 .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. . ∂µ ] = [x3 ∂2 − x2 ∂3 .338) [J3 . ∂µ ] = (10. 0. .340) [K2 .262).336) We will proceed with this task considering all six cases: [J1 . ∂µ ] = [x0 ∂2 + x2 ∂0 . ∂µ ] = [x0 ∂3 + x3 ∂0 .341) [K3 . 0. We will proceed by employing the generators (10. 0. w = (0. . As mentioned already we will show condition (10. 1) . ϑ = (0.339) [K1 . w = (0.243) and. We have demonstrated.. 0). w = (0. therefore. ∂µ ] = (10. 0). ∂µ ] = [x1 ∂3 − x3 ∂1 . [K . ϑ = (0. 0). The second condition is identical to (10. i. ∂µ ] = [x2 ∂1 − x1 ∂2 .335) shows that we need to demonstrate [J . ∂µ ] = (10. . ∂µ ] = (J )ν µ ∂ν . 0.342) One can readily convince oneself that these results are consistent with (10. the Dirac equation is invariant under active Lorentz transformations. Inspection of (10.e. ∂µ ] = (K )ν µ ∂ν . 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 (10.336). ∂µ ] = (10.335) = ∂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. of course. .

345) holds |qV | << m . 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 .10.345) (10.350) (10. for the scalar ﬁeld V in (10.347) (10. The Dirac equation in the Schr¨dinger form reads then o i∂t Ψ(xµ ) = ˆ ˆ where α and β are deﬁned in (10.1) i∂t − qV and p by ˆ ˆ π = p − qA .348) (10.10: Dirac Particles in Electromagnetic Field 333 10. σ2 . Accordingly. 10.351) X (x ) µ .6 we assume that the ﬁeld is described through the 4-vector potential Aµ and. (10. Equivalently. e.g.344) Using the notation σ = (σ1 .343) One may also include the electromagnetic ﬁeld in the Dirac equation given in the Schr¨dinger form o ˆ (10.345) in the so-called non-relativistic limit in which all energies are much smaller than m.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. For this purpose we choose the decomposition Ψ(xµ ) = φ(xµ ) χ(xµ ) .1 in Sect 10. 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 10.221) reads then [ iγ µ (∂µ + iqAµ ) − m ] Ψ(xν ) = 0 (10. The Dirac equation (10.. Non-Relativistic Limit We want to consider now the Dirac equation (10. we deﬁne φ(xµ ) = e−imt Φ(xµ ) χ(x ) = e µ −imt (10.349) We want to focus on the stationary positive energy solution.233) by replacing i∂t by (see Table 10. (10. This solution exhibits a timedependence exp[−i(m + )t] where for holds in the non-relativistic limit | | << m. we replace the operator ∂µ by ∂µ + iqAµ .232). Following the respective procedure developed for the Klein-Gordon equation in Sect.6 above).346) ˆ ˆ ˆ α · π + qV + β m Ψ(xµ ) (10. accordingly.

357) for Φ can be reformulated by expansion of (σ · π)2 . π ] .356) i∂t Φ ≈ Φ + qV Φ . i i (10. For this purpose we employ the identity (5. ∂k + qAk ] i i 1 1 1 1 [ ∂j . 10.349) yields i∂t Φ i∂t X = = ∂t X X Relativistic Quantum Mechanics << m . πk ] = = 1 1 [ ∂j + qAj .358) ( πj πk − πk πj ) = jk [πj . identiﬁes X as the small component of the bi-spinor wave function which.346. which in the present case states ˆ ˆ2 ˆ ˆ (σ · π)2 = π + i σ · (π × π) .348. (10. Φ Using (10. (10. 10. does not need to be considered anymore.354) ˆ 0 ≈ σ · π Φ − 2m X . derived in Sect.356).353) by X ≈ to obtain a closed equation for Φ ˆ σ·π 2m 2 (10. Ak ] + q 2 [Aj .353) (10. due to the m−1 factor.359) We want to evaluate the latter commutator.351) in (10.230). ∂k ] + q [Aj . accordingly.354) The properties (10. ∂k ] + [∂j . (10. ∂k ] + [∂j .360) For an arbitrary function f (r) holds ( [Aj .334 and assume that for the time-derivative of Φ and X holds ∂t Φ << m . one can replace X in (10.7. Ak ] ) f = ( ∂j Ak − Ak ∂j + Aj ∂k − ∂k Aj ) f . ˆ ˆ For the components of π × π holds ˆ ˆ π×π = jk (10. (10.352) ˆ σ · π X + qV Φ ˆ σ · π Φ + qV X − 2mX . ˆ Equation (10. Ak ] i i i i =0 =0 = q q [Aj .352) allow one to approximate (10. (10. One obtains [πj .350. 10.357) Equation (10. henceforth. and. ∂k ] + q [ ∂j .355) ˆ σ·π Φ 2m (10.361) . 5. Ak ] .

10.366) can be interpreted as a rotation around the ﬁeld B by an angle qtB. o reveals a stunning feature: the Pauli equation does justice to its two-dimensional character. 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) .363) allow us to write (10.e. t) (10.10. πk ] = q q (( ∂j Ak − ∂k Aj )) = − i i ×A jk q = − B i jk (10. 10. −A). 0. In other words. Dirac Particle in Coulomb Field .2) governing a one-dimensional wave function ψ ∈ C. Ak ] ) f = [ ((∂j Ak )) − ((∂k Aj )) ] f (10. t) [see (8.343) for the vector potential Aµ = (− Ze2 . t) 2m 2m Φ(r. Comparision of (10. Let us consider brieﬂy the consequences of the interaction of a spin. 0.1 which emerged in the Lorentz-invariant theory as an algebraic 2 necessity.. ∂k ] + [∂j .222.344. t)]2 q − σ · B(r.2) it introduces the o extra term qσ · B Φ which describes the well-known interaction of a spin. Equations (10.2 around the magnetic ﬁeld.Spectrum We want to describe now the spectrum of a relativistic electron (q = −e) in the Coulomb ﬁeld of a nucleus with charge Ze. but rather remains as a steady “guest” of non-relativistic physics with the proper interaction term. (10. 10. i.358.364) which is referred to as the Pauli equation. r (10.360) and Aµ = (V. one obtains ( [Aj .357) in the ﬁnal form i∂t Φ(r. The respective bispinor wave function Ψ(xµ ) ∈ C4 is described as the stationary solution of the Dirac equation (10.365) Comparision of this expression with (5.363) where we employed B(r.364) governing a two-dimensional wave function Φ ∈ C2 with the corresponding non-relativistic Schr¨dinger equation (10.362) or.223) shows that the propagator in (10.367) . the spin. 0) . while agreeing in all other respects with the non-relativistic Schr¨dinger equation (10.359.1 particle with a magnetic 2 ﬁeld B. The formal solution of this equation is Φ(t) = e−iqtB·σ Φ(0) .366) (10. t) = × A(r.6)]. [πj .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 ((· · ·)). using (10. does not leave the theory again when one takes the non-relativistic limit. t) + q V (r. t) ≈ ˆ [p − q A(r. the interaction qσ · B induces a 1 precession of the spin. 5.1 with the magnetic ﬁeld.

according to (10.374). ˜ (10.231).228) and in (10.370) Any solution of (10.s.370) resembles closely the Klein-Gordon equation (10. 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.h. 1. (5.180). but the converse is not necessarily true. 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 .373) (10. Employing πµ as deﬁned ˜ in Table 10.ν=1 µ =ν γ µ γ ν πµ πν .369) from the left by γ ν πν + m.370) ˜ [ i˜ µ (∂µ + iqAµ ) − m ] [ i˜ µ (∂µ + iqAµ ) + m ] Ψ(xµ ) = 0 γ γ (10.369) For our solution we will adopt presently a strategy which follows closely that for the spectrum of pionic atoms in Sect.370). (5. µ = ν and altering ‘dummy’ summation ˜ ˜ ˜ ˜ . 3 µ=0 ˆ2 (˜ µ )2 πµ = π0 − π . Equation (10. multiplying (10. 10. γ ˆ2 ˆ2 (10.369).7 one can write the second term in (10.372) (10.370) is obtained then ˜ [ i˜ µ (∂µ + iqAµ ) + m ] Ψ(xµ ) γ is a solution of (10. indeed. 3.6.230) γ µ γ ν = − γ ν γ µ . respectively. is a solution of (10. (˜ 0 )2 = 1 and (˜ j )2 = γ 1 γ −1 j = 1. The diﬀerence arises from the term γ µ πµ γ ν πν in (10.370) for which holds ˜ ˆ ˜ ˆ 3 γ πµ γ πν = ˜ ˆ ˜ ˆ µ=0 µ ν (˜ µ )2 πµ + γ ˆ2 µ. but diﬀers from it in an essential way.369). noting from (10.368) ˜ ( γ µ πµ − m ) Ψ(xµ ) = 0 .374) The ﬁrst term on the r. 5. we solve ˜ [ i˜ µ (∂µ + iqAµ ) − m ] Ψ(xµ ) = 0 γ (10.368) is also a solution of (10.e.368) ˜ where Ψ(xµ ) and γ µ are deﬁned in (10. ˜ However.371) such that we can conclude that (10.232) in Sect. γ ˆ ˜ ˆ (10. once a solution Ψ(xµ ) of (10.1 one can write (10.336 Relativistic Quantum Mechanics For the purpose of the solution we assume the chiral representation. i.230).375) Following the algebra that connected Eqs. For this purpose we ‘square’ the Dirac equation.371). can be rewritten using. ˜ ˜ ˆ ˆ (10. 2.229).

229). r2 (10.381) According to (10. i.376) This expression can be simpliﬁed due to the special form (10.344) π ˆ [ˆ0 . the commutators [˜ 0 . it holds 1 4 [˜ µ . γ j ] [ˆ0 .378) According to the deﬁnition (10. γ ˜ π ˆ j=1 (10. γ ν ] [ˆµ . π0 ] γ ˜ π ˆ j=1 = [˜ 0 .ν=1 µ =ν 337 1 2 ( γ µ γ ν πµ πν + γ ν γ µ πν πµ ) ˜ ˜ ˆ ˆ ˜ ˜ ˆ ˆ µ. 3 (10. γ ] [ˆ0 .382) .ν=1 µ =ν σ 0 0 −σ · (( qA0 )) (10..ν=1 µ =ν = [˜ µ . πν ] γ ˜ π ˆ µ. γ µ γ ν πµ πν = ˜ ˜ 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.367) and the deﬁnition (10.378) can be evaluated using (10. 2. πj ] in (10. due to A = 0. −i∂j ] = − [ (∂t + iqA0 ) ∂j − ∂j (∂t + iqA0 ) ] f i ((∂j qA0 )) f (10.ν=1 µ =ν 3 j=1 3 3 = 1 4 1 2 1 [˜ .10: Dirac Particles in Electromagnetic Field indices.344).377) which follows readily from the deﬁnition (10. πν ] = 0 π ˆ for µ.380) γ µ γ ν πµ πν = i ˜ ˜ ˆ ˆ µ.e. t) is a suitable test function and where ((· · ·)) denotes the range to which the derivative is limited. Since [ˆµ . γ 0 ] [ˆj .367) holds qA0 = r ˆ Ze2 . ν = 1. γ j ] are γ ˜ [˜ 0 .367) of Aµ .379) The commutators [ˆ0 . πν ] γ ˜ π ˆ µ.ν=1 µ =ν (10. one can summarize (10. πj ] + γ ˜ π ˆ 4 0 j [˜ j . πj ] . Altogether.10.380) where f = f (r. 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 (10.376–10. γ ν ] [ˆµ . πj ] π ˆ = = (−i∂t + qA0 .

375) the ˆ ‘squared’ Dirac equation (10.148). 6.383) We seek stationary solutions of this equation.338 Relativistic Quantum Mechanics where r = r/|r| is a unit vector.5 the operator iσ · r is block-diagonal in this basis such that only ˆ 1 the states for identical j. 2 3 2 . (10.147.384) into (10. Accordingly.191) solved in Sect. couples the orbital angular momentum of the electron to its spin.389) According to the results in Sect. however. −j + 1. 6. (10. . Insertion of (10. 1 |ˆ). . it is this quantity that we want to determine.2 components ˜ Ψ(r) = and obtain for the separate components φ± (r) Ze2 + r 2 ˜ φ+ (r) ˜ φ− (r) (10.385) 1 We split the wave function into two spin. 10. . + j } (10. (10.383) 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 . (10..381).e. . In the latter case. is genuinely two-dimensional and. . r r 2 2 2 j = 1.387) The expression (10. (10. 1 |ˆ) ) . m = −j.374). Combining this result with (10. only the two states {Yjm (j − 2 . We note that these states are also eigenstates of the angular r 2 .384) can be interpreted as the energy of the stationary state and.388) in terms of states introduced in Sect. Yjm (j + r 2 1 1 . 6.189) 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 . we express 2 the solution of (10. Yjm (j + 1 .1 . (10.6. 1 |ˆ). 2 |ˆ)} as given in (6.368) reads − Ze2 ∂t − i r 2 + 2 + i σ·r ˆ 0 0 −σ · r ˆ Ze2 − m2 r2 Ψ(xµ ) = 0 (10.388) Except for the term iσ · r this equation is identical to that posed by the one-dimensional Kleinˆ Gordon equation for pionic atoms (10. hence. Such solutions are of the form ˜ ˜ Ψ(xµ ) = Φ(r) e−i t . m values are coupled. i. . in r ˆ fact. a solution of the form ∼ Y m (ˆ) can be obtained.386) + 2 ± iσ · r ˆ Ze2 − m2 r2 φ± (r) = 0 .5 which describe the coupling of orbital angular momentum and spin 1 {( Yjm (j − 2 . The term iσ · r.

396) and λ2 (j) [λ2 (j) + 1] (10.392) into diagonal form. − 1 r2 (j − 1 )(j + 1 ) − Z 2 e4 2 2 ±i Ze2 (j + )(j + ) − Z 2 e4 We seek to bring (10. (6. in the present treatment. . not the wave functions. λ1 (j) and λ2 (j). 1 |ˆ) = − Yjm (j ˆ r 2 2 1 2 h± (r) g± (r) Yjm (j − 1 .391) derived in Sect. (6. 6. .395) (10.393) but do not explicitly consider further the wavefunctions.2 (j) as given by (10.390) (10.397) This equation is identical to the Klein-Gordon equation for pionic atoms written in the form (10. . 10. except for the slight diﬀerence in the expression of λ1.199). 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 .f.5 [c. unaltered. 1. 1 |ˆ) r 2 (10. we want to determine solely the spectrum.392).395.10. However.10: Dirac Particles in Electromagnetic Field 339 ˆ momentum operator L2 [cf. 3 .151). rather than = 0.2 (j) being 2 j = 1 . . such transformation can be chosen as to diagonalize the second term.2 (j)[λ1.2 (j) + 1) 2 Ze2 + + r2 r 2 − m2 f1.2 (r) = 0 (10. a speciﬁc pair of total spin-orbital angular momentum quantum numbers j. 1 |ˆ) r 2 2 2 2 r r . conclude that the . 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. 1 |ˆ) are r 2 2 ˆ eigenfunctions of L2 . 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. . involving the 2 × 2 unit matrix. 1 |ˆ) + r Yjm (j + 1 . Obviously. We can. Any similarity transformation leaves the ﬁrst term in (10.186)]. namely. m and expand φ± (r) = Using σ · r Yjm (j ± 1 . namely. therefore.2 (j).392) h± (r) g± (r) = 0. . (10. the values of the argument of λ1. We select.392) reads then in the diagonal representation 2 ∂r − λ1.151]. as in the case of pionic atoms.200). which states that the states Yjm (j ± 1 . property (6. Since.394) Equation (10. the eigenvalues are independent of m. and except for the fact that 2 2 we have two sets of values for λ1. the missing additive term − 1 . hence.396) and (10..

. 2 2 m = −j. j = 1.340 Relativistic Quantum Mechanics spectrum of (10. . of these equations describe the binding energy. ( +1) − Z 2 e4 )2 (10.397) is again given by eq. 2. One obtains in case of (10. . .2 . = 1 + m Z 2 e4 √ ( n − − 1 + 2 − Z 2 e4 )2 . accordingly. m) 2 n = 1. (10. In case of non-relativistic hydrogen-type 1 atoms. −j + 1. . . .h. . .s. . . 2. j . Using (10. 1 ms = ± 2 (10. .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 = 0.398. . 10.395. . the second term on the r.403) √ Z 2 e4 2 ED (n. . . m = − . 10. Obviously. 2. The magnitude of the relativistic eﬀect 2 2 e4 . (10. j = 1 .396) we obtain. One can equate (10. including spin.398) 2 = 1 + m . . 2n2 = 0. m) = 1 + m (n− + . 1. . It is given by n = n + + 1 where is the orbital angular momentum quantum number and n = 0. j . 1.401) if one attributes to the respective 1 states the angular momentum quantum numbers = j + 2 and = j − 1 . . . . and spin-orbital angular momentum j = − 1 .395) and 2 corresponds to λ2 (j) as given in (10. (10. . 1. m = −j. . . . .396).400) and (10.401) corresponds to a non-relativistic state 2 with quantum numbers n. . . 1. 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. 3 . . . . .402) with (10. .400) corresponds to a non-relativistic state with quantum numbers n. −j + 1. . . 2 2 1 ≈ m − (10. n − 1 . .401) These expressions can be equated with the non-relativistic spectrum. . −j + 1. 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 .404) = 0. n − 1 . j = + 1 . .399) n = 0. . 2. − + 1. . . counts the nodes of the wave function. . . (10. . These considerations 2 are summarized in the following equations 1 ED (n. (10. 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. j where 1 corresponds to λ1 (j) as given in (10. 1. . . 3.400) (10. .203). One may also state 2 this in the following way: (10.402) In this expression n is the so-called main quantum number. . . and spin-orbital angular momentum j = + 1 . j = − 2 . albeit with some modiﬁcations. . . m = −j. . .

40151 ⇑ . relativistic. The energies were evaluated with m = 511. binding energy / eV Eq..10.404) ﬁnally into the single formula ED (n. However. 1. relate closely to the corresponding nonrelativistic states. notation 1s 1 2 non-rel. binding energy / eV Eq. ± 3 .551177 ⇑ . .405). (10. . 2 |ˆ) for m = ± 1 . j (10. is in the range of 10 eV. mom. n − 1 j = 1 2 1 2 ± m = −j.60583 -3. the non-relativistic and relativistic energies are hardly discernible. 10. One can combine the expressions (10.405) In order to demonstrate relativistic eﬀects in the spectrum of the hydrogen atom we compare in Table 10. j.40146 ⇑ ⇑ -1.3. . This is clearly demonstrated by the comparision of non-relativistic and relativistic spectra of a hydrogen-type atom with Z = 100 in Table 10. . 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. . . −j + 1. Of particular interest is the eﬀect of spin-orbit coupling which removes. .40147 -1. 0 0 1 1 0 1 1 2 2 spinorbital ang. the 2p 3 states with r 2 2 2 2 1 involve four degenerate states corresponding to Y 3 m (1.2: Binding energies for the hydrogen (Z = 1) atom. . 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 (1.405) in terms of the non-relativistic quantum numbers n.402) -13.10: Dirac Particles in Electromagnetic Field main quantum number n 1 2 2 2 3 3 3 3 3 orbital angular mom. i.. (10.405)] spectrum of the hydrogen atom. = 0. 1 |ˆ) for m = ± 1 .551176 spectr.1. Degeneracies are denoted by ⇑.402)] and the relativistic [cf. much less than the rest mass of the electron (511 keV). (10.036 by means of Eqs.405 ) -13. 2.0041 keV and e2 = 1/137. . j 1 2 1 2 1 2 3 2 1 2 1 2 3 2 3 2 5 2 341 rel.403. r 2 2 2 . . (10. case these six states are split into energetically diﬀerent 2p 1 and 2p 3 states.3.51176 ⇑ ⇑ ⇑ ⇑ 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. in fact.402. . 10.e. i. The reason is that the mean kinetic energy of the electron in the hydrogen atom. the nonrelativistic degeneracy for the six 2p states of the hydrogen atom: in the present. (10.60601 -3.e.1. for example. . m) = m 1 + Z 2 e4 (n−j −1 + 2 (j+ 1 )2 − Z 2 e4 )2 2 n = 1. The table entries demonstrate that the energies as given by the expression (10.551178 ⇑ .2 the non-relativistic [cf.

0.409) where V (r) is spherically symmetric.406) α √ 1+ (n − β + β 2 −α)2 The expansion (10. 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 10.6 -42. (10. 10.206) provides in the present case ED (n. The energies were evaluated with m = 511. In order to investigate further the deviation between relativistic and non-relativistic spectra of hydrogen-type atoms we expand the expression (10. Degeneracies are denoted by ⇑.1 -34. An example for such potential is the Coulomb potential V (r) = −Ze2 /r considered further below.2.2 -17. binding energy / keV Eq. binding energy / keV Eq. Tables 10. 0) (10. (10.15. Introducing α = Z 2 e4 1 and β = j + 2 (10.35. We assume for the wave function the stationary state .036 by means of Eqs. to estimate the diﬀerence between the energies of the states 2p 3 and 2p 1 (cf. (10.405 ) -161.1 ⇑ ⇑ ⇑ ⇑ rel.8 ⇑ . 32 (10.3).407) 1 2 This expression allows one.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).402. j. 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 .10.9 ⇑ . It holds for n = 2 and j = 3 .402) -136. .15. for example.3: Binding energies for the hydrogen-type (Z = 100) atom.342 main quantum number n 1 2 2 2 3 3 3 3 3 orbital angular mom.405).1 ⇑ . j 1 2 1 2 1 2 3 2 1 2 1 2 3 2 3 2 5 2 Relativistic Quantum Mechanics non-rel.405) to order O(Z 4 e8 ). mom.405) reads 1 (10. m) ≈ m − mZ 2 e4 mZ 4 e8 − 2n2 2n3 1 j+ − 3 4n + O(Z 6 e12 ) .0 ⇑ ⇑ -15. 0 0 1 1 0 1 1 2 2 spinorbital ang. (10.3 spectr.0041 keV and e2 = 1/137.

414) ˆ σ · p rf (r) Yjm (j + 1 .5. ˆ σ · p X + m Φ + V (r) Φ ˆ σ · p Φ − mX + V (r) X = = Φ X (10. 2 |ˆ) + r Yjm (j − 2 . pp. 2 |ˆ) r Φ(r) r .413. This term has been discussed in detail in Sect.415) 1 ˆ σ · p rg(r) Yjm (j − 2 . 1 |ˆ) r 2 2 = i ∂r + j+ r 1 2 3 2 f (r) Yjm (j − 1 .. 10. according to (10. r 2 (10. 1 |ˆ).10. The Dirac equation reads then. 10. 1 |ˆ) r 2 = i ∂r − 1 2 r g(r) Yjm (j + 1 . 1 |ˆ) r 2 2 (10. The arguments above allow one to eliminate the angular dependence by expanding Φ(r) and X (r) 1 in terms of Yjm (j + 2 .10: Dirac Particles in Electromagnetic Field form Ψ(xµ ) = e−i t 343 Φ(r) X (r) . 1 |ˆ) . 168]: the term is a scalar (rank zero tensor) in the space of the spin-angular momentum states Yjm (j ± 1 . 1 |ˆ) and Yjm (j − 1 . 6.416) The diﬀerential equations (10.411.198)] not couple states with diﬀerent j. (10. 1 |ˆ) .232. m-values. r 2 2 (10.411) (10.2 degrees of freedom.197. but are timeindependent. = r (10. i.345). 1 |ˆ) r 2 2 = i j+ ∂r + r j+ r 1 2 1 r f (r) Yjm (j − 2 .414) These equations can be brought into a more symmetric form using ∂r + which allows one to write (10.e. in particular.e. 1 |ˆ) r 2 2 = i ∂r + −j r 1 g(r) Yjm (j + 2 . (6.417) c(r) d(r) X (r) 1 1 1 1 Yjm (j + 2 .413) ˆ σ · p g(r) Yjm (j − 1 .. 6.5 [see. i.f.412) are four-dimensional with r-dependent wave functions.410) 1 where Φ(r). σ · p ˆ σ · p f (r) Yjm (j + 1 . 2 |ˆ) r r r . 1 |ˆ) r 2 1 1 = ∂r r r r (10. 6.412) ˆ In this equation a coupling between the wave functions Φ(r) and X (r) arises due to the term σ · p. 1 |ˆ) introduced in r 2 2 1 Sect. 10. X (r) ∈ C2 describe the spatial and spin. the term is block-diagonal in the space spanned by the states Yjm (j ± 1 . 2 |ˆ) and does r 2 ˆ has odd parity and it holds [c. r r 2 2 2 a(r) b(r) 1 1 1 1 Yjm (j + 2 . 2 |ˆ) + r Yjm (j − 2 .

417) of the form f1 (r) i Yjm (j + 1 .423) .415. there exist two independent solutions (10.411. 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. Presently. we consider the case that only states for one speciﬁc j.417) into (10. c(r) are coupled. the orthonormality property (6. 10. only a(r).421) g2 (r) X (r) 1 1 − Yjm (j + 2 .416). According to (10. g2 (r) ∂r + 1 2 f1 (r) − [ + m − V (r) ] g1 (r) = 0 (10.344 Relativistic Quantum Mechanics In general. 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. such expansion must include states with all possible j. (10. m values.412).418) holds for f1 (r).420) g1 (r) X (r) 1 1 − Yjm (j − 2 . g1 (r) ∂r − j+ r j+ r 1 2 g1 (r) + [ − m − V (r) ] f1 (r) = 0 ∂r + and for f2 (r). 10. d(r) are coupled and b(r). 2 |ˆ) r r where the factors i and −1 have been introduced for convenience. m pair contribute. using (10. 2 |ˆ) r r f2 (r) 1 1 r Φ(r) i r Yjm (j − 2 .157).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. Accordingly.419) Obviously. Inserting (10. 1 |ˆ) r Φ(r) 2 2 r = (10.

enforcing the polynomial to be of ﬁnite order leads to discrete eigenvalues .422) determines solutions of the form (10. .422. (10.425) r 0 covers states with angular momentum 1s 1 . 2 2 2 2 2 2 = j − 1 2 and. one demonstrates then that the wave functions for r → ∞ behaves as exp(−µr) for some suitable µ. together with the appropriate boundary conditions at r = 0 and r → ∞. . .420). 3d 3 .. except for the opposite sign of the term (j + 1 ).409). = 0.421). . 2. The solution of (10.422). in the non-relativistic limit wave functions f2 (r) 1 1 i Yjm (j − 2 . 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. 3d 5 .424) Ψ(xµ ) ≈ r 0 1 i. i.367) which is spherically symmetric such that equations (10. . etc. Similarly.423) are identical.405) of the relativistic hydrogen atom the corresponding states have quantum numbers n = 1.423). Dirac Particle in Coulomb Field . According to the discussion of the spectrum (10. In the non-relativistic limit. .10.422) corresponds to states f1 (r) Yjm (j + 1 .422) describes the states 2p 1 . According to this procedure.e. determining wave functions of the type (10.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. Behaviour at r → 0 We consider ﬁrst the behaviour of the solutions f1 (r) and g1 (r) of (10. . 2 |ˆ) r Ψ(xµ ) ≈ . Equation (10. 10. the 2 equations determine. (10. n − 1. We assume the 4-vector potential of pure Coulomb type (10.405).422) and (10. 3p 3 . and obtains ﬁnally a polynomial function p(r) such that rγ exp(−µr)p(r) solves (10. correspodingly the states We consider ﬁrst the solution of (10. 2 2 2 (10.10: Dirac Particles in Electromagnetic Field 345 Equations (10. 1. (10.422). the radial wave functions for Dirac particles in the potential (10. one demonstrates ﬁrst that the wave function at r → 0 behaves as rγ for some suitable γ.420) is the large component and X is the small component. Hence.e.. 2s 1 .422).426) . 2p 3 . . 3p 1 . the ones given in (10.405). 1 |ˆ) r i 2 2 . (10. We note that (10. namely. 3s 1 . (10.423) apply for V (r) = −Ze2 /r.422) near r = 0. to states with angular momentum = j + 2 . Hence. Φ in (10. etc.422) follows in this case from the same procedure as that adopted for the radial wave function of the non-relativistic hydrogen-type atom. for small r.

for small r. 10.434) √ The solutions of these equations are f1 . 10.427) with 1 γ = (j + )2 − Z 2 e4 . g1 (r) →∞ ∼ e−µr . becomes imaginary. hence. we conclude f1 (r) →∞ ∼ e−µr .430) 2 Note that the exponent in (10. Such r-dependence 2 would make the expectation value of the potential r2 dr ρ(r) 1 r (10.267. (10. in case j + 1 )2 < Ze2 .266. Only the exponentially decaying solution is admissable and. according to (10. One obtains γ = ± (j + 1 )2 − Z 2 e4 . hence.429) This equation poses an eigenvalue problem (eigenvalue −γ) for proper γ values.432) = = − ( − m.422) becomes ∂r g1 (r) ∂r f1 (r) Iterating this equation once yields 2 ∂r g1 (r) 2 ∂r f1 (r) 1 .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 . for the particle density holds then ρ(r) ∼ |rγ−1 |2 = Behaviour at r → ∞ For very large r values (10.428) = 0.433) = = ( m2 − 2 ) g1 (r) ( m2 − 2 ) f1 (r) (10. (10.431) inﬁnite since. determined that the solutions f1 (r) and g1 (r). The assumed r-dependence in (10. r2 (10. g1 ∼ exp(± m2 − 2 r). µ = m2 − 2 (10. We have.435) . ) f1 (r) ( + m ) g1 (r) (10.427) = = 0 0. g1 (r) →0 ∼ b rγ (10.427).420). assume the r-dependence in (10.427) makes only the positive solution 2 possible. (10.

h.441) correspond to (10.435 ). Equation (10.422) for the Coulomb potential V (r) = −Ze2 /r We assume a form for the solution which is adopted to the asymptotic solution (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. Accordingly.440) (10.442) which leads to a partial cancellation of the asymptotically dominant terms.10: Dirac Particles in Electromagnetic Field For bound states holds 347 < m and. Let us consider then for the solution of (10.440) and (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 ˜ (10.437) where they ˜ ˜ cancelled in case f1 = g1 = a. Solution of the Radial Dirac Equation for a Coulomb Potential To solve (10. We also introduce the new variable ρ = 2µ r .434) results in √ √ (m − ) m + a − (m − ) m + a √ √ (m + ) m − a − (m + ) m − a which is obviously correct. g1 (r) = φ1 (r) − φ2 (r) ˜ (10. (10. hence. √ g1 (r) = − m − a e−µr .438) (10. µ is real.436). we set f1 (r) g1 (r) = = √ √ ˜ m + e−µr f1 (r) −µr = = 0 0 (10.445) . of both (10. 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.441) The last two terms on the l.10.437) (10.434) √ m + a e−µ r . (10.444) (10.s.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 . However.436) f1 (r) = Insertion into (10. ρ m+ ρ [∂ρ − (10.

454) .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 (10.446.452) From (10. (10.446 . 10.447) We seek solutions of (10.430) one can write this βs = s+γ − 2 √ Ze m2 − 2 s (s + 2γ) βs−1 .447) leads to (s + γ) αs + (j + 1 ) βs − √ 2 s Ze2 m2 − 2 αs (10.446) (10.449) for γ given in (10. (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 (10.453) (s + γ)2 + Z 2 e4 − (j + 1 )2 2 2 √ Ze m2 − 2 βs . 10. Inserting (10.450) −√ mZe2 m2 − 2 βs ρs+γ−1 = 0 mZe2 m2 − αs (s + γ) β2 + (j + 1 ) αs + √ 2 s 2 +√ From (10.450) follows Ze2 m2 − 2 βs − βs−1 =0 (10. 10.426–10.448) φ2 (ρ) = ρ (10. Using (10.448.f.430)].451) 1 √mZe − (j + 2 ) 2 2 αs .447) of the form n φ1 (ρ) = ργ s=0 n γ s=0 αs ρ2 βs ρ2 (10.449) into (10. = m− Ze2 βs s + γ − √m2 − 2 2 (10.430) which conform to the proper r → 0 behaviour determined above [c.

.448. n It holds φ1 (ρ) φ2 (ρ) = = ργ F (−so . . conclude 2 2 2 that the polynomials in (10.. namely. 10. i. . (s − so ) β0 s! (2γ + 1)(2γ + 2) · · · (2γ + s) (10.405). . n = 0..456) − so 2 √mZe m2 − 2 (−so )(1 − so )(2 − so ) . ρ) .449) must be of ﬁnite order.405) allows one to identify n = n − j + 2 which. in fact.457) with the conﬂuent hypergeometric functions F (a. for the states 2p 1 .461 for values given by (10. c. (s − so ) β0 s! (2γ + 1)(2γ + 2) · · · (2γ + s) (10. 2. . 1. We can..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. x) .457) One can relate the polynomials φ1 (ρ) and φ2 (ρ) deﬁned through (10. . α + 1. m (n ) = . = From (10.e. Comparision with (10.455) are ﬁnite. ρ) so β0 ργ F (1 − so .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 . with the associated Laguerre polynomials L(α) = F (−n.430. The expressions (10. β0 j+ 1 2 (10. 2.459) − 2 √mZe m2 − 2 (10.452) follows j+ αs = 1 2 349 Ze2 m2 − 2 −γ (10. so = n . 2γ + 1.456) and (10.448.10: Dirac Particles in Electromagnetic Field Deﬁning so = √ one obtains βs = = . 3d 3 holds n = 1. For example. . (10. c c(c + 1) 2! (10. 1.461) In order that the wave functions remain normalizable the power series (10.456.458) or. is an integer. 10. 10. 10. This requires that all coeﬃcients αs and βs must vanish for s ≥ n for some n ∈ N. equivalently. 2γ + 1.405) and the ensuing so values ( 10.10. . x) = 1 + a a(a + 1) x2 x + + . According to the deﬁnitions (10.457) for βs and αs imply that so must then be an integer.460) (10. hence.449) and (10. .455) this conﬁnement of so implies discrete values for .

460.439).158) of the latter states absorb the angular integral in (10.410) of the stationary state wave function the density ρ(xµ ) of the states under consideration. which involve the conﬂuent hypergeometric functions in (10.421) with radial wave functions f1 (r). 2γ + 1. (10. Greiner. can follow the procedure adopted for the wave functions of the non-relativistic hydrogen atom and will not be carried out here. 1 |ˆ) 2 2 r κ = j+ 1 2 (10.464) 0 The evaluation of the integrals. 2µr) (2µ) 2 Γ(2γ + 1) 3 N = and m 4m )Γ(2γ + n + 1) (n +γ)m (n +γ)m (10.421)] ∞ dr (|f1 (r)|2 + |g1 (r)|2 ) = 1 (10. g1 (r) determined by (10.m (j − 1 . Sect.463) where Φ and X .465) (10. 1990). (10. m. is time-independent.267). 10. The coeﬃcients β0 in (10.442). Due to the form (10. j.463) and yield [note the 1/r factor in (10.421). The wave functions (10.467) ± n F (1 − n . 1 |ˆ) in (6.460. 10.438.469) This formula has been adapted from ”Relativistic Quantum Mechanics” by W. The normalization integral is then ∞ 0 π 2π r2 dr 0 sin θdθ 0 dφ (|Φ(r)|2 + |X (r)|2 ) = 1 (10. m|xµ ) = e−i t iF1 (r) Yj. 2γ + 1. Berlin. The complete wave function is given by the following set of formulas 1 Ψ(n. where2 F± (κ|r) = G1 (r) = F+ (κ|r) .148). = j + 2 . and (10. are two-dimensional vectors determined through the explicit form of the spin-orbital angular momentum states Yjm (j ± 1 . 10.466) F1 (r) = F− (κ|r) .157. 9.147. (n + γ)m N (2µr)γ−1 e−µr − κ F (−n . 10. = j + 2 .m (j + 1 .350 Relativistic Quantum Mechanics Altogether we have determined the stationary states of the type (10. 2µr) (10.460. The orthonormality 2 2 r properties (6.461). They are described through quantum numbers n. 6. 1 |ˆ) 2 2 r G1 (r) Yj. as in (10.461) are to be chosen to satisfy a normalization condition and to assign an overall phase. (Springer. 6. 2 .6.468) −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 .461).421) correspond to non-relativistic states with orbital angular momentum 1 1 = j + 2 . j. given by expression (10.

describes the complementary set of states 1s 1 .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. = j − µ 1 2 .10. G2 (r) = F+ (κ|r) .472) F2 (r) = F− (κ|r) . this wavefunction has an orbital angular momentum quantum number = j − 1 and. 3p 3 .m (j + 1 . .469) that the following wave functions result Ψ(n. 1 |ˆ) 2 2 r κ = −j − 1 2 (10. 2s 1 . hence.471) (10.469).465–10. 3d 5 .10: Dirac Particles in Electromagnetic Field 351 We want to consider now the stationary states of the type (10. obtained closed expressions for the wave functions of all the stationary bound states of relativistic hydrogen-type atoms. We have. The radial wave functions f2 (r) and g2 (r) in (10. One can verify. 2 accordingly. tracing all steps which lead from (10. (10. j.422) to (10.421) which. 3s 1 . where F± (κ|r) are as given in (10. 2p 3 .470) r 0 Obviously.m (j − 1 . not not 2 2 2 2 2 2 covered by the wave functions given by (10.469). etc. 2 |ˆ) r Ψ(xµ ) ≈ e−1 t . 1 |ˆ) 2 2 r G2 (r) Yj.421) are governed by the radial Dirac equation (10. in the non-relativistic limit. m|x ) = e −i t iF2 (r) Yj. become f2 (r) 1 1 i Yjm (j − 2 .467–10.

352 Relativistic Quantum Mechanics .

in fact.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. ϑ). isomorphic to the natural representation. i.. in a sense. ϑ through 3. We will proceed by building as much as possible on the results obtained sofar in the chiral representation of the Dirac equation.2) (11. in particular.. indeed. must be two-dimensional irreducible representations of the Lorentz group.1) (11. diﬀerent L(w. a complex variable z. The lower dimensionality of a(z) and b(z) implies. z ∈ C ˜ Because of its block-diagonal form each of the diagonal components of S(z).3) (11. 1 353 . ϑ)µ ν correspond to diﬀerent a(z) and b(z). 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. ϑ)µ ν . i.229).e. a(z) and b(z).Chapter 11 Spinor Formulation of Relativistic Quantum Mechanics 11. may be exploited to express the Lorentz-invariant equations of relativistic quantum mechanics. This is.262). 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. expressing its dependence on w. 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. what will be achieved in the following.225–10. ˜ We have altered here slightly our notation of S(w.4) = exp = w − iϑ . Starting point is the Lorentz transformation S(w. ϑ) for the ˜ in the chiral representation as given by (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.e.

β. ϑ ∈ R3 . however. ϑ ∈ R3 (11. ϑ ∈ R3 . θ ∈ R3 . A First Characterization of the Bispinor States We note that in case w = 0 the Dirac transformations are pure rotations. ﬁnally. In this case a(z) and b(z) are identical and read a(i ϑ) = b(i ϑ) = exp 1 − ϑ·σ 2 . neutrino equation. (11. For ϑ = (0. γ (see Chapter 5). ± 1 represents the familiar spin– 1 states. −1 2 2 z = iϑ. + 1 2 2 state 2 ∼ | 1 .e.9) a∗ (z) = exp 2 . (11. i. +1 2 2 |1. are elements of SU(2) and correspond.e. usually expressed 2 as product of rotations and of functions of Euler angles α. (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. will establish the map between L(w.6) as given by (5. − 1 2 2 z = iϑ.243). In this case holds a(z) = b(z) ˜ 1 (xµ ). express 4-vectors Aµ .5) The transformations in this case. in 2 fact. Ψ2 (xµ ) and Ψ3 (xµ ). +1 2 |1. to 1 ∗ ∗ z ·σ (11. i. a distinct diﬀerence in the transformation behaviour of ˜ ˜ ˜ ˜ Ψ1 (xµ ). Aν . to the rotational transformations of spin. Ψ4 (xµ ) in case z = w + iϑ for w = 0.7) where “∼” stands for “transforms like”. This characterization allows one to draw conclusions regarding the state space in which a(z) and b(z) operate.1 states as described by Dm m (ϑ).354 Spinor Formulation and b(z) act. Ψ4 (xµ ) transform according to ˜ ˜ ˜ and Ψ b(z).8) We like to stress that there exists. and the Klein–Gordon equation in the spinor representation. a space of vectors state1 for which holds state2 state 1 ∼ | 1 . 2 2 2 ˜ Since a(z) acts on the ﬁrst two components of the solution Ψ of the Dirac equation. Here | 1 . ϑ)µ ν and a(z). the operator ∂µ and the Pauli matrices σ in the new representation and. for z = iϑ.. 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 . Relationship Between a(z) and b(z) The transformation b(z) can be related to the conjugate complex of the transformation a(z). namely. formulate the Dirac equation. −1 2 2 |1. the so-called spinor space.. β. Ψ2 (xµ ) transform according to a(z) whereas Ψ3 (xµ ). b(z).

however.10) From this one can derive b(z) = where = 0 1 −1 0 .10. one can express a∗ (z) −1 = exp 1 − z ∗· σ 2 = b(z) .14) .2. −1 a∗ (z) −1 (11.13) Explicit matrix multiplication using (5. µ ). 1. (11. therefore. 11. but leave rotation angles ϑ unaltered.f. The eﬀect of inversion on Lorentz transformations is.12) does. −1 given by (11. 11. ∗ σ 3 = σ3 . in fact. (5.224.224)] ∗ σ1 = σ1 . σ2 . This demonstrates that a(z) is the ˜ transformation which characterizes both components of S(z).12) yields σ1 σ3 or in short σ Similarly. Ψ (xµ ) as well as Ψ (xµ ). Ψ (xµ ) even though these pairs of ˜ ˜3 ˜4 featuring the components Ψ1 (x ˜ 2 components transform independently of each other. . that the transformation (11. σ3 ). (11. (11.12). 355 (11. (11. one can show −1 −1 −1 = −σ1 = σ2 = −σ3 ∗ = −σ1 ∗ = −σ2 σ2 −1 (11. using f (a) −1 = f ( a −1 ). One can readily verify [c. −1 = ∗ −σ3 = −σ ∗ . Hence. a result to be used further below.15) (11. ∗ σ 2 = − σ2 .17) We conclude.11. 11. hold −1 = 1 One notices then.1: Lorentz Transformation of Dirac Bispinor ∗ ∗ ∗ where σ ∗ = (σ1 . that they alter w into −w.18) with a(z) given by (11.12) To prove (11. Spatial Inversion ˜ One may question from the form of S(z) why the Dirac equation needs to be four-dimensional.16) σ = −σ ∗ . 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.11) 0 −1 1 0 = .11) one ﬁrst demonstrates that for and −1 as given in (11.1) can be written in the form ˜ S(z) = a(z) 0 0 a∗ (z) −1 (11. a∗ (z) −1 = exp 1 ∗ z σ∗ 2 −1 .4) and .

Amsterdam. for particles like the neutrinos which do not obey inversion symmetry two components of the wave function are suﬃcient. (11. P −1 = P.224)] tr( σj ) = 0 . the Lorentz invariant equation for neutrinos is only 2–dimensional.3.1: Show that SL(2..e. ˜ ˜ P S(w + iϑ) P = S(−w + iϑ) = b(z) 0 0 a(z) .e. 2 = 1 i. de Kerf Lie a Algebras. One can verify this by evaluating the determinant of a(z) det( a(z) ) = det e 2 z·σ 1 (11. C) deﬁned in (11. ϑ ∈ R3 .20) ˜ Obviously. 1990). 2 . does not suﬃce for particles like the electron which obey inversion symmetry. a(w + iϑ) for w = 0 is an element of SL(2. However.10.21) = etr( 2 z·σ) = 1 1 (11.2.A. The transformation S(z) in the transformed space is then ˜ P 1. For the general case the proof. The proof of this important property is straightforward in case of hermitian M (see Chapter 5).G.24) = etr(M ) (11.. (11. j = 1.21) together with matrix multiplication as the binary operation forms a group.23) Exercise 11. can be found in G. based on the Jordan–Chevalley theorem.C) and SO(3. In fact.A. C) = { M. This implies ˜ Ψ1 ˜ Ψ2 P ˜ Ψ3 ˜ Ψ4 ˜ Ψ3 ˜ Ψ4 = Ψ1 ˜ ˜ Ψ2 . (5. the transformations a(z) and b(z) become interchanged. det(M ) = 1 } . which describes pure rotations. 2. Part (Elsevier. non-singular matrix M holds2 det eM and from [c. M is a complex 2 × 2–matrix. 11.22) which follows from the fact that for any complex. is an element of SU(2).2 Relationship Between the Lie Groups SL(2.f.1) We have pointed out that a(iϑ). the space spanned by only two of the components of Ψ is not invariant under spatial inversion and. (11. Exercise 1. B¨uerle and E. However. hence. Obviously.356 Spinor Formulation ˜ Let P denote the representation of spatial inversion in the space of the wave functions Ψ. 3 .19) i.

We will prove below [cf. Argue why M (Aµ ) = σµ Aµ provides a bijective map. (11.26) (11.2: Show that σ0 . σ1 . the matrix M (Aµ ) is hermitian as can be seen from inspection of (11.75) holds aν = Lν µ aµ one realizes that σµ transforms inversely to covariant 4–vectors.2. The function M (Aµ ) is bijective.28) σµ = 1 0 0 1 .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 (11.30) Since the components of Aµ are real.29) where we used (10. σ2 . σ2 . Noting that for covariant vectors according to (10.31) Exercise 11. (11. 1 0 0 −1 . or+ thochronous Lorentz transformations. one wishes that the deﬁnition (11.11.135)] this transformation behaviour. Demonstrate that (11. M (Aµ ) according to (11. i.25) σ0 σ1 σ2 σ3 The quantity σµ thus deﬁned does not transform like a covariant 4–vector.27) requires then σ µ Aµ Lν µ σµ = σν (11. in a transformed frame should hold Lρ ν σ µ Aµ σµ A µ . C) and the group L↑ of proper. (11. 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 . −→ µ Consistency of (11. (11.25) can also be written M (Aµ ) = A0 + A3 A1 − iA2 A1 + iA2 A0 − A3 . σ3 provide a linear–independent basis for the space of hermitian 2 × 2–matrices.28) and (11. σ3 are hermitian.C) and SO(3. one can provide a simple expression for the inverse of M (Aµ ) M = M (Aµ ) ↔ Aµ = 1 tr M σµ 2 .26) or from the fact that the matrices σ0 .25) of the matrix M (Aµ ) is independent of the frame of reference.. in fact.2: Lie Groups SL(2. (11.31) holds. 0 −i i 0 . σ1 . 0 1 1 0 .27) −→ Straightforward transformation into another frame of reference would replace σµ by σµ . (11. In fact.30). The following important property holds for M (Aµ ) det ( M (Aµ ) ) = Aµ Aµ which follows directly from (11.e.76).32) .

Due to det(a) = 1 the transformation (11. + .37) using (11. (11.31)] 1 a Aµ −→ A µ = tr a M (Aµ ) a† σµ 2 Because of (11. In fact.36) The suitable A µ can readily be constructed using (11. The linear character of the transformation becomes apparent expressing A µ as given in (11. (11.35) holds for this transformation A µ Aµ = Aµ Aµ (11. In fact. any a ∈ SL(2.34) where we used the properties M † = M and (a† )† = a. (11. L(a)µ ν = 1 tr a σν a† σµ 2 . 11.3: Show that L(a)µ ν deﬁned in (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† .31).33) to M (Aµ ) describing the action of the transformation in terms of transformations of Aµ . it holds for the matrix M deﬁned through (11.40) is an element of L↑ . (11.37) deﬁnes actually a Lorentz transformation.25) Aµ = which allows us to express ﬁnally A µ = L(a)µ ν Aν .37) which implies that (11.38) .f. Accordingly. (11.32. C) .2.40) 1 tr a σν a† σµ 2 Aν (11.33) det(M ) = = det( a M a† ) = det(a) det(a† ) det(M ) [det(a)]2 det(M ) = det(M ) .39) Exercise 11. for any a ∈ SL(2. (11. C) and for any Aµ there exists an A µ such that M (A µ ) = a M (Aµ ) a† . C) deﬁnes the transformation [c.33) This transformation conserves the hermitian property of M since (M )† = ( a M a† )† = (a† )† M † a† = a M a† = M (11.35) We now apply the transformation (11. a ∈ SL(2.33) conserves the determinant of M .

(11. 2 (11. ¯ Lµ ρ = L(a1 )µ ν L(a2 )µ ρ = L( product in SO(3.C) )µ ρ (11.e. 1 ¯ and using the deﬁnition of Lµ ρ in (11.42) = ν Deﬁning Γ = a† σµ a1 .δ (11.α.1) deﬁned through L(a)µ ν [cf.47) This completes the proof of the homomorphic property of L(a)µ ν .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 .40)] respects the group property of SL(2.44) (σν )αβ (σν )δγ .δ 1 4 Aαβγδ Γβα Γγδ α. (11.1). (11.C) and 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 δ = δ = = δ = δ 1 2 = 2 = 1 Aαβγδ (11.11.β γ.45) One can demonstrate through direct evaluation 2 α = 2 α = 2 α = = 2 α = 0 else β β γ γ = = = = γ = γ = 1.1) L(a)µ ν provides a homomorphism 359 We want to demonstrate now that the map between SL(2. C) and of SO(3.β γ.2: Lie Groups SL(2.1) 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. β 2. C) and SO(3. i. .43) (σν )αβ Γβα Γγδ (σν )δγ = ν..41) results in 1 ¯ Lµ ρ = 4 where Aαβγδ = ν Γ = a2 σρ a† 2 (11.

53) Equating (11. correspondingly.49).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 ) .e. J3 of the Lorentz transformations Lµ ν in the natural representations. i.50) (11.48) yields for the l. 1) − σ1 σ0 0 0 One can verify readily that 1 κ1 = − σ 1 2 (11.52) and (11. 10. we write (11. 0.52) where we employed (11. correspond to the generators given by (10.25) in the last step. To this end we consider inﬁnitesimal transformations and keep only terms of zero order and ﬁrst order in the small variables.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. (11. of (11. M (Aµ + (K1 )µ ν Aν ) = = M (Aµ ) + M (Aµ ) − M ( (−A1 . (11.53) results in the condition σ 0 A1 − σ 1 A0 = ( κ 1 σ µ + σ µ κ † ) Aµ . 0. denoted below as κ1 . J Spinor Formulation The transformations a ∈ SL(2. 1 (11. 0). K3 .49) we obtain using (11.57) .h.2) which correspond to the generators K1 . 0) ) σ0 A 1 − σ1 A 0 (11. J1 . 0. 0.58) .s.48) We want to determine now the generators of the transformation a(z) as deﬁned in (11.55) (11. noting the linearity of M (Aµ ).49) assuming (note that g µ ν is just the familiar Kronecker δµν ) Lµ ν a = = gµν + 1 + 1 (K1 )µ ν κ1 . J2 . Aµ = (0. For the r. 1.s. Aµ = (0. K2 .. 0). eight real numbers. −A0 . C) which correspond to K. 0. Aµ = (0.54) This reads for the four cases Aµ = (1. 0).360 Generators for SL(2. 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 . C) as complex 2 × 2–matrices are deﬁned through four complex or. (11.51) Insertion of (K1 )µ ν as given in (10.47. 0.48). 1 (11.56) (11. Because of the condition det(a) = 1 only six independent real numbers actually suﬃce for the deﬁnition of a.36) M (Lµ ν Aν ) = a M (Aµ ) a† (11. 1.h. of (11. To obtain the generator of a(z) corresponding to the generator K1 .

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 . (11.3: Spinors 361 obeys these conditions. Here a(z) = ( aα β ) = describes the matrix a(z). λ2 .4: Show that the generators (11. a1 1 a1 2 a2 1 a2 2 (11. J3 are given by 1 κ = − σ.1) acts on 4–vectors Aµ .62) We denote the general (z = w + iϑ) transformation by φ α = aα β φβ = aα 1 φ1 + aα 2 φ2 .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.61) This identiﬁes the transformations a(z = w − iϑ) as representations of Lorentz transformations. state that the following two transformations are equivalent L(w. κ3 of a(z) corresponding to K2 . C) acts on spinors φα ∈ C2 (characterized below) 1 (11. 2 (11. 2 λ = i σ. + 1 2 2 1 transforms under rotations (z = iϑ) like a spin– 2 state | 1 .11. hence. ϑ) = ew·K + ϑ·J ∈ SO(3.60) of a ∈ SL(2. 2 2 φ1 φ2 ∈ C2 .60) We can. def α = 1.63) where we extended the summation convention of 4-vectors to spinors. − 1 .64) . Similarly. J2 . the so-called contravariant spinors. C) correspond to the generators K and J of Lµ ν . Exercise 11. a(w − iϑ) = e− 2 (w − iϑ)·σ ∈ SL(2. λ3 corresponding to J1 .2. K3 and λ1 . w describing boosts and ϑ describing rotations. 11. one can show that the generators κ2 . 2 (11.

as we will demonstrate below such states are invariant under general Lorentz transformations a(z). this scalar product is anti-symmetric. Such a scalar product does. indeed. i. ± 1 . . ± 1 ..e. 0| 1 . The existence of a scalar product gives rise to the deﬁnition of a dual representation of the states φα denoted by φα . etc. (11. those of “particle 2” with the spinor χβ one can state that the quantity 1 Σ = √ 2 φ1 χ2 − φ2 χ1 (11. In fact. as covariant spinors.67) We will refer to φα . In this case spinors φα transform like spin– 1 states and an invariant. The relationship between the two can be expressed φ1 φ2 = = 3 φ1 φ2 0 1 −1 0 (11. χβ . exchange of φα and χβ alters the sign of the expression.70) ‘Scalar’ implies invariance under rotations and is conventionally generalized to invariance under other symmetry trasnformations.e. should remain invariant under transformations a(iϑ).e. m 1 | . 0 = 2 2 m=± 1 2 1 1 (0. To arrive at a suitable scalar product we consider ﬁrst only rotational transformations a(iϑ). The corresponding states are deﬁned through φ1 χ2 − φ2 χ1 = φ1 χ1 + φ2 χ2 It obviously holds χα = χ1 χ2 = χ2 −χ1 . m. In the notation developed in Chapter 5 holds for the singlet state | . ..66) should constitute a singlet spin state. i. . Deﬁnition of covariant spinors Expression (11. 0. 0| 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.68) (11. . . −m) | . 1 ) stands for the Clebsch–Gordon coeﬃcient. . is the singlet state. χβ .66) is a bilinear form. . invariant under transformations a(z). . However. 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. . 0| . Using (0. invariant and as such has the necessary properties of a scalar3 product. as contravariant spinors and to φα . which 2 can be constructed from products φ1 χ2 . 1 .65) where | · · · 1 describes the spin state of “particle 1” and | · · · 2 describes the spin state of “particle 2” and (0. −m 2 2 2 2 1 1 1 1 2 (11. 1 ) = 2 2 2 2 2 2 2 2 √ ±1/ 2 and equating the spin states of “particle 1” with the spinor φα . 1 ..69) (11.. i.

72) 0 −1 1 0 Exercise 11. Proof that φα χα is Lorentz invariant We want to verify now that the scalar product (11.69.70) is φ1 φ2 −1 363 = = −1 φ1 φ2 . The transformation behaviour of φα according to (11. (10.74)].72)] φα φ α = = = = αβ αβ φβ φβ 11 21 11 21 12 22 12 22 (11. 11. (11.78) aβ γ γδ φδ (11.3. In the transformed frame holds φ α χα = aα β αγ aγ δ δκ φβ χκ . The inverse of (11.11.3: Spinors as can be veriﬁed from (11.73) (11.79) = a −1 T a κβ .71.70) and (11. 11. −1 connecting contravariant and covariant spinors play the role of the metric tensors gµν . (11. Accordingly.81) .63. 11.77) (11. (10. For this scalar product holds φα χα = −χα φα . 11.10.71) (11.68).69.1: Show that for any non-singular complex 2 × 2–matrix M holds M −1 = det(M ) M −1 T The matrices .74) = = 0 1 −1 0 0 −1 1 0 (11.73.75) is given by φα = as can be readily veriﬁed.80) One can write in matrix notation aα β γ δκ αγ a δ αβ (11. we will express (11. g µν of the Minkowski space [cf. 11.76) αβ αβ The scalar product (11.72) in a notation analogous to that chosen for contravariant and covariant 4–vectors [cf. (11.67) will be expressed as φα χα = φ1 χ1 + φ2 χ2 = φ1 χ2 − φ2 χ1 .77) is Lorentz invariant. 10.75) (11.

the ˙ conjugate complex spinors φα need to be considered separately from the spinors φα ..85) φ1 φ2 ∗ ∗ .83) into (11. We denote ˙ (aα β )∗ = aα β ˙ (11. σ † = (σ ∗ )T = σ. ˙ We also deﬁne covariant versions of φα φα = (φα )∗ .e. (11. k = 1. Hence. Insertion of (11.14) one can write a −1 Spinor Formulation = e 2 z·σ 1 −1 = e 2 z· 1 σ −1 = e− 2 z·σ 1 ∗ (11.84) (11.63). i.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. As discussed above.87) which we will employ from now on.88) which one veriﬁes taking the conjugate complex of (11.88) reads ˙ ˙ φ α = aα β φβ ˙ ˙ (11. a∗ (z) provides a representation of the Lorentz group which is distinct from that provided by a(z). hence.89) such that (11.83) Here we have employed the hermitian property of σ. Obviously. ˙ (11.81) yields aα β and. 2.2) and (11. The transformation ˙ behaviour of φα is ˙ φ α = (aα β )∗ φβ ˙ (11. it holds φk = (φk )∗ .364 Using (11. from (11.80) φ α χα = φβ χβ .86) A concise notation of the conjugate complex spinors is provided by ˙ (φα )∗ = φα = φ1 ˙ φ2 ˙ ˙ (11.91) . 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 (11.90) extending the summation convention to ‘dotted’ indices.

in matrix notation.93) (11. For the spinors φα and χα thus ˙ deﬁned holds that the scalar product ˙ φα χα = φ1 χ1 + φ2 χ2 ˙ ˙ ˙ ˙ ˙ (11. (11.97) The transformation. is governed by the operator a∗ (z) −1 which arises in ˜ the Lorentz transformation (11.100) .96) is Lorentz invariant. one can state ˜ and (11. Ψ4 .92) (11.75). i. (11.11. 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 = (11.95) ˙ where αβ and αβ are the real matrices deﬁned in (11. The transformation behaviour of φα is ˙ φα = ˙ ˙ β γδ ˙˙ φδ ˙ ˙ αβ a γ ˙ ˙ . 11..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 ) = = . − 1 .94) (11. A comparision of (11.73.97) implies then that φα transforms like Ψ ˙ φ1 ˙ φ2 ˙ transforms under rotations (z = iϑ) like a spin– 1 state | 1 .18) ˜ 3 . a property which is rather evident.75.4 Spinor Tensors ˙ ˙ ˙ We generalize now our deﬁnition of spinors φα to tensors.18) of the bispinor wave function Ψ.4: Spinor Tensors 365 The relationship between contravariant and covariant conjugate complex spinors can be expressed in analogy to (11. A tensor tα1 α2 ···αk β1 β2 ···β (11. + 1 2 2 2 1 transforms under rotations (z = iϑ) like a spin– 2 state | 1 .e.76) φα ˙ ˙ φα αβ ˙ ˙ αβ ˙ ˙ = = = = αβ ˙ ˙ αβ ˙ ˙ αβ αβ φβ φβ ˙ ˙ (11. 11.99) 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.

102.103) A αβ = aα γ aβ δ Aγ δ .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. k. αβ = αβ (11.105) This tensor is actually invariant under Lorentz transforma= αβ .102) = .. m. i. i. using conventional matrix indices j.106) Exercise 11. An obvious candidate is [cf. ˙ ˙ ˙ ˙ (11.25) the tensor is explicitly Aαβ = ˙ A1 1 A1 2 ˙ ˙ A2 1 A2 2 ˙ ˙ = A0 + A3 A1 − iA2 A1 + iA2 A0 − A3 .25] M (Aµ ) = σµ Aµ . .m aj akm t m (11.e. (11.e. tjk = a t a† jk ˙ ˙ ˙ ˙ (11. ˙ ˙ ˙ ˙ (11. i.106).1: Prove equation (11. the transformation bevaviour of a tensor tαβ reads in spinor and matrix notation t αβ = aαγ aβδ tγδ . tjk = a t aT jk = .109) . This tensor transforms according to t αβ = aα γ aβ δ tγ δ ˙ This reads in matrix notation.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 . the elements of which are linear functions of Aµ .101) An example is the tensor tαβ which will play an important role in the spinor presentation of the Dirac equation. This task implies that we seek a tensor. it holds αβ . αβ ˙ αγ t ˙ γβ (11. with ˙ contravariant indices αβ.103) Similarly. We had demonstrated that M (Aµ ) transforms according to M = M (Lµ ν Aν ) = a M (Aµ ) a† (11. The 4–vector Aµ in spinor form We want to provide now the spinor form of the 4-vector Aµ . (11.4.e. An example of a tensor is αβ and tions..108) Obviously.m aj a∗ t km m .. (11. 11.107) which reads in spinor notation [cf. we want to express Aµ through a spinor tensor. (11. According to (11.

Using (11.117) ˙ ˙ ∂0 − ∂3 −∂1 + i∂2 −∂1 − i∂2 ∂0 + ∂3 . employing (11. Aµ and tensors tαβ can be applied to the partial diﬀerential operator ∂µ . (11.115) is correct.113) Aα β = ˙ (11. ˙ 2 (11.110) The 4-vectors Aµ .110) one can state ∂ αβ = Similarly.115) Exercise 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. ∂µ in spinor notation The relationship between 4–vectors Aµ .4. (11.2: Prove that (11.111) A11 A12 ˙ ˙ A21 A22 ˙ ˙ ˙ 0 −1 1 0 (11.4: Spinor Tensors One can express Aαβ also through Aµ Aα β = ˙ ˙ 367 A0 − A3 −A1 + iA2 −A1 − iA2 A0 + A3 . (11.112) A22 −A21 ˙ ˙ −A12 A1 1 .11. 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 . Hence. 11. (11.109. (11.110) one obtains Aα β = ˙ A0 − A3 −A1 − iA2 −A1 + iA2 A0 + A3 A0 + A3 A1 + iA2 A1 − iA2 A0 − A3 .116) .114) yields ∂αβ = ˙ ∂0 + ∂3 ∂1 + i∂2 ∂1 − i∂2 ∂0 − ∂3 .

243) in the chiral representation expressing S(Lη ξ ) by (11.229) and (11.18) and γ µ by (11.124).126) which is the complex conjugate of (11.. 1 0 0 −1 .120) σµ 0 .h. (11.124).119) (11. (11. 0 −i i 0 .121).e. hence.368 σµ in Tensor Notation Spinor Formulation We want to develop now the tensor notation for σµ (11. one can conclude aσ µ ∗ −1 Lν µ σν 0 = . (11.243)]. −1 0 0 1 (11. (11.e.121) One expects then that the transformation properties of σ µ should follow from the transformation properties established already for γ µ [c. To obtain the transformation properties of σ µ we employ then (10.15) one can write σµ = and.125) a (σ ) Equation (11.124) constitutes the essential transformation property of σ µ and will be considered further.123) (a∗ )−1 Lν µ a L ν µ = = σν (σ ) ν ∗ −1 (11. i. σµ building on the known transformation behaviour of γ µ . Equation (10.122) aσ µ −1 a−1 −1 (a∗ )−1 0 Lν µ (11.. holds also in the chiral representation.26) and its contravariant analogue σ µ σµ = σµ = 1 0 0 1 1 0 0 1 .118) 0 −1 −1 0 0 i −i 0 For this purpose we consider ﬁrst the transformation behaviour of σ µ and σµ .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. . Hence.125) is equivalent to µ ∗ −1 −1 a∗ (σ µ )∗ −1 −1 a Lν µ = (σ ν )∗ (11. 0 1 1 0 . i. .243) holds independently of the representation chosen. of this equation is 0 ∗ (σ µ )∗ a and. We will obtain the transformation behaviour of σ µ .124) . (11. Comparision of (10. Note that (10.125) is equivalent to (11. γµ = ˜ 0 µ )∗ (σ −1 0 σµ σµ 0 (σ µ )∗ −1 . (10. hence.f.118) yields ˜ γµ = ˜ Using (11. This is ˜ possible since γ µ can be expressed through σ µ .s. . σµ . (11.

107) to (11. 11.124) a σ µ [a∗ ]T Lν µ = σ ν . −1 0 0 1 µ=3 . (σ ∗ )T = σ yields using (11.g. In fact..134) . 3) σj −→ a(iϑ) σj a(iϑ) † = a(iϑ) σj a∗ (iϑ) T .2) −1 369 (a∗ )−1 = −1 − 1 z ∗ ·σ ∗ 2 e = e− 2 z 1 ∗ · −1 σ ∗ = e2z 1 ∗ ·σ . 0 1 1 0 µ=1 .2. (11. 1 0 0 −1 µ=3 . .4: Spinor Tensors One can rewrite (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. (11. It holds according to (11. 11.108) that the same transformation behaviour applies then for general Lorentz transformations. 0 i −i 0 µ=2 .133) Combining (11. e. equation (11. hence. 11.e.103) ˙ identiﬁes σ µ as a tensor of type tαβ . 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 .4). (11.2.130) We argue in analogy to the logic applied in going from (11. under rotations the Pauli matrices transform like (j = 1.131) where a is given by (11. state that σ µ in a new reference frame is σ µ = a σ µ a† (11. transformations (11. therefore.132) 1 0 0 1 µ=0 .. This transformation behaviour.129.2) −1 (a∗ )−1 = e2z 1 ∗ ·(σ ∗ )T = e2z 1 ∗ ·σ ∗ T = [a∗ ]T . 11. according to (11.16.11. (11.127) Exploiting the hermitian property of σ.124) using (11. = . 11. 2. One can. (11.102.4) with w = 0. 0 −1 −1 0 µ=1 .128) One can express. (11. i.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.

˙ ˙ In (11. This yields 1 0 0 1 0 0 0 0 σ0 + σ3 σ1 + iσ2 (11. Each of the 4 × 4 = 16 matrix elements in (11. (11..140) where we employed the expressions (11.114).370 Inverting contravariant and covariant indices one can also state Lν µ σµ = σν .29)]. whereas ˙ account for the elements of the individual Pauli the ‘outer’ contravariant spinor indices. The ˙ desired change of representation(σµ )αβ −→ (σµ )αβ corresponds to a transformation of the basis of ˙ spin states f g f −→ = (11. ˙˙ (11. i.e.s. δ.139) g −f g and.135) This is the property surmised already above [cf. account for the 4-vector µ. 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 .132) and (11.138) the ‘inner’ covariant spinor indices.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 ˙ ˙ ˙ ˙ . We will now consider the representation of σ µ . in fact.136) results in the succinct expression 1 2 ˙ γδ ˙ (11.138) Note that all elements of σαβ are real and that there are only four non-vanishing elements.224) ˙ Using (11. account for the 4–vector index µ. however.h.12) for of σµ yields together with (5.133) is characterized through a 4-vector ˙ index µ. σµ in which the contravariant indices are moved ‘inside’.136) σαβ = = 2 ˙ σ1 − iσ2 σ0 − σ3 0 0 0 0 1 0 0 1 where on the rhs. the submatrices correspond to tαβ spinors. i. We can. We want to express now σ µ and σµ also with respect to the 4–vector index µ in spinor form employing (11. µ = 0. i. and the covariant indices are moved outside. α. = σαβ ˙ ˙ ˙ ˙ ˙ (σ21 )11 (σ21 )12 (σ22 )11 (σ22 )12 ˙ ˙ ˙ ˙ ˙ ˙ ˙ ˙ 21 (σ )22 21 (σ )22 (σ21 ) (σ22 ) ˙ ˙ ˙ ˙ 21 22 Equating this with the r.137) σαβ ˙ = δαγ δβ δ .e. Spinor Formulation (11. the transformation is inverse to that of ordinary 4–vectors. 1. 3 as well as through two spinor indices αβ. matrices. of (11. β. hence.e.. We can summarize that σ µ and σµ transform like a 4–vector.. 2. ˙ ˙ 0 −1 1 0 ˙ (11.110) to express σ αβ in terms 0 −1 1 0 0 0 0 1 (11. γ.

˙˙ (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 ˙ .142) = This can be expressed 1 2 σ αβ ˙ ˙ γδ = δαγ δβ δ . This expression implies that the element of γ µ given by σ µ −1 is in the expression (11.121) of γ ˜ ˜ ˙ basis |αβ whereas the element of γ µ given by σ µ is in the basis |αβ . we write ˜ ˙ γµ = ˜ 0 (σ µ )αβ µ ) )∗ ((σ αβ 0 ˙ ˙ . (11. One can write then the scalar product using (11. (11.145) Let Aµ be a covariant 4–vector. which in ˙ (11. ˙˙ (11.143) Combined with (11.140) to transform each of the four submatrices.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 ˙ ˙ .11.144) The Dirac Matrices γ µ in spinor notation We want to express now the Dirac matrices γ µ in spinor form.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 .146) Exploiting the property (11. employing then transformation (11. Accordingly. ˙ ˙ ˙ ˙ (11.4: Spinor Tensors 371 and. For this purpose we start from the ˜ µ . (11.147) .

99) yields the Dirac equation in spinor form i ∂ αβ χβ ˙ i ∂ αβ φ ˙ The simplicity of this equation is striking. (11.149) (11.147) allows us to rewrite the Dirac equation in the chiral representation (10.226) ˜ i γ µ ∂µ Ψ(xµ ) = 0 i ∂ αβ ˙ i ∂ αβ 0 ˙ ˜ ˜ Ψ(xµ ) = m Ψ(xµ ) .148) ˜ Employing Ψ(xµ ) in the form (11.5 Lorentz Invariant Field Equations in Spinor Form Dirac Equation (11.372 Spinor Formulation 11. α ˙ = = m φα m χβ .150) . ˙ (11.

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

For example. . generates a state which has the same energy as the original one. Lie groups play an essential role in the discussion of mass degeneracies in particle physics. The energy levels are mk 2 En = − 2 2 . . can be identiﬁed with the angular momentum operators.... (12. . as represented by an element of SO(3). 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 . (12. k = 1. Therefore. Jk . In order to illustrate this... · · · . n − 1. as studied in chapter 7. The spherical symmetry implies the commutation of the Hamiltonian with angular momentum operators (7. on an energy eigenstate. .18). Presence of additional symmetries may further increase the degeneracy of the system. m r (12. 3 . the states 3s. l. Classically. V (r). . 2m r (12. . . l and the corresponding energy levels are independent of m.λn . we ﬁrst consider a particular example of the implications of symmetry.. φ).. 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. Sk . (12. ψE.8) ˆ [H.λ1 . is allowed to take values in 0. (7. The energy levels are totally independent of l.λn = E exp(iαk Sk ) ψE. As an example of this we will consider the Coulomb problem with the Hamiltonian H = p2 k − . (7.λn .12).λ1 .m (r) = vE.m (r) Y m (θ. Sk . (12. leaves the energy of the state invariant H exp(iαk Sk ) ψE. In chapter 7 the dynamics of a particle moving in three dimensions under the inﬂuence of a spherically symmetric potential.39).5) with m = −l. where the symmetry generators.λ1 . each energy level is (2l + 1)-fold degenerate..4) The stationary Schr¨dinger equation. has been discussed..7) 2 n where the orbital angular momentum. 3p and 3d all have the same energy. can then be reduced to a one-dimensional radial o equation.374 Spinor Formulation The action of any symmetry generator...3) If the newly obtained state is linearly independent of the original one. This degeneracy follows from the fact that any rotation. namely motion in a spherically symmetric potential described by the group SO(3) (or its double covering SU (2) as discussed in section 5. We will investigate this shortly in more detail in the case of systems with spherical symmetry.6) From elementary quantum mechanics we know the spectrum of the hydrogen atom.. this implies a degeneracy in the spectrum.24). 2. which yields a set of eigenstates of the form ψE. Jk ] = 0 . (7.8) .

In the following we consider the bound states.11. 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 . 2 1 B = (J − K).13) It can be veriﬁed explicitly that its components commute with the Hamiltonian. Jj ] = i [Ki .10) (12.21) .12) (12. 2E (12.8) is not a hermitian operator.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.9) (12. Kj ] = i ijk Kk . we will compute the hydrogen spectrum using the additional symmetry. 2 which can be shown to satisfy [Ai . which have a negative energy E. We will also need the following identity [4] 2 = 2H J2 + m 2 + k2 . (12.19) (12.20) (12. For this purpose we ﬁrst note that J · = 0.11) (12. Therefore. Now we scale the eccentricity vector as follows K= − m . ijk Jk . in the subspace of the Hilbert space corresponding to a certain energy we can replace H by E. (12. 2m r 1 x1 (J3 p2 − J2 p3 + i p1 ) − k .18) which complement the familiar angular momentum algebra of section 5. We introduce the following new operators 1 A = (J + K). In order to understand the aforementioned extra degeneracy. Aj ] = i ijk Ak . The vector in (12.15) which can be proved after very considerable algebra. (12. 2m r (12.16) Through some algebra [4] the following commutation relations can be veriﬁed [Ki .17) (12. 2m r 1 x2 (J1 p3 − J3 p1 + i p2 ) − k .14) This follows from a · (a × b) = (a × b) · a = 0. 2m r 1 x3 (J2 p1 − J1 p2 + i p3 ) − k . (12. which is valid even when a and b do not commute.3.

. .27) (12. Chaos is described classically as exponential . which is identical to the the direct sum of the Lie algebra of of two SO(3) (or SU (2)) algebras. . . . 1. . Furthermore the bound on the orbital angular momentum. . 1. In contrast to the discussion above about extra symmetries. Using this equation the energy eigenvalues can be written in terms of the eigenvalues of A2 and B 2 operators.7) tells us that (2a + 1) = n.21) and (12. = 0. By comparing to the rotation algebra introduced in chapter 5. a lack of symmetry implies a lack of degeneracy in the energy levels of a quantum mechanical system.31) where we have used (12. 1. namely that J J > < A − B =0 A + B =2 A (12. a = 0.22) (12.25) So far we have shown that the symmetry generators form an algebra. . Bj ] = 0. . b = 0. . can be seen to follow from the triangle inequality as applied to J = A + B. A.34) It follows that l has to have values in {0 = |a − b|.33) (12. 2 (12. 2 (2a + 1)2 2 2 (12. Bj ] = i [Ai . .32) A comparison with (12. we can read oﬀ the eigenvalues of A2 and B 2 from (12. (12. Noticing that A2 and B 2 have the same eigenvalues because of (12. the energy eigenvalues are found to be E=− mk 2 1 . .15). . a + b = n − 1}. ijk Bk . . 2 1 . Spinor Formulation (12. a = 0. .22): A2 = a(a + 1) B 2 = b(b + 1) Following (12. l. 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 (12.28) . H = 0. This illustrates the eﬀect of additional symmetries to the degeneracy structure of a quantum mechanical system. . The most extreme case of this is the quantum analogue of a classically chaotic system. H B.26) (12.24) (12.14) we note that A2 − B 2 = J · = 0.28). This implies that a = b. 1.30) 2 2 2 1 .23) (12. . .376 [Bi . .29) (12.

A typical example of this so called quantum chaos is the quantum billiard problem. that the proton and the neutron behave identically under the so-called strong interactions and that their diﬀerence is solely in their charge content. (12. This was motivated by the observation that their masses are approximately equal: mp = 938. It is important to place the isospin concept in its proper historical context. the isospin symmetry is also governed by an SU (2) group ‘rotating’ components in (12. 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. Following the mass-energy equivalence of special relativity E = mc2 . which was discovered by observing degeneracies in the particle spectrum. it is natural to represent them in terms of a two component vector. This is known as level repulsion. For example. for example. mp = 939. This enables us to utilize what we already know about the SU (2) symmetry group from the study of angular momentum.) If the proton and the neutron are to be viewed as two linearly independent states of the same particle. analogous to the 1 spin-up and spin-down states of a spin.57M eV /c2 .2 system p= 1 0 . This (approximate) degeneracy led into the idea of the existence of an (approximate) symmetry obeyed by the underlying nuclear interactions. 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. 12. (Strong interactions bind the atomic nucleus together. in the sense that nearby trajectories in the phase space diverge from each other over time.2 Isospin and the SU (2) ﬂavor symmetry The concept of isospin goes back to Heisenberg.36) In analogy to the concept of spin regarding the rotations in 3-space as discussed in chapter 5.36) into each other in abstract isospin space. after the discovery of the neutron in 1932. n= 0 1 . for the . who. another manifestation of chaos is the lack of independent operators commuting with the Hamiltonian. (12. If the chosen boundary is ‘irregular’ in a suitably deﬁned sense. the classical trajectories will diverge from each other after successive bounces from the boundary.5: Lorentz Invariant Field Equations 377 sensitivity to initial conditions. which is a particle in box problem in two dimensions with a boundary which can be chosen arbitrarily. suggested that the proton and the neutron can be regarded as two states of a single particle. (Weak interactions are responsible.11.35) this mass equivalence can be viewed as an energy degeneracy of the underlying interactions. However. Originally it was believed that isospin was an exact symmetry of strong interactions and that it was violated by electromagnetic and weak interactions. namely. In the next section we will proceed with the discussion of a symmetry.28M eV /c2 . For a more detailed discussion of quantum chaos in billiard systems we refer the reader to [7] and the references therein.

(mπ± = 139.378 Spinor Formulation beta decay). 2. sec. π 0 = |1. The mass diﬀerence between the neutron and the proton could then be attributed to the charge content of the latter. If the mass diﬀerence (or the energy diﬀerence) were to be to be purely electrostatic in nature. Denoting the total isospin. Shortly we 1 will see how to describe both the pion and nucleon states as composites of more fundamental I = 2 states. .− 2 2 . n = I = . 1 π 0 = |1. u¯ and d¯ states.39) All other isomultiplets.37) As an example of a multiplet with I = 1 we have the three pions or π-mesons π + = |1. 1 we can re-write (12. including the proton and the neutron. 2 2 1 2 2 2 1 1 1 1 1 √ . −1 .− 2 2 1 2 2 2 1 1 1 1 . 2 2 1 1 1 . 1 . the proton is the lighter of the two. I3 = 2 2 1 1 . including the spin and color quantum numbers of their constituent quarks. 0 π − = |1. as we shall see below. as good quantum numbers. π − = |1. the proton had to be heavier.42) (12. Isospin symmetry is not an exact symmetry of strong interactions. For example. and its third component. . Therefore it remains a useful concept. They can be constructed with the same rules that have been used for angular momentum addition ¯ u in chapter 6.6M eV /c2 . I3 = − 2 2 . we refer the reader to [1]. If it were otherwise the proton would be unstable by decaying into the neutron.11. 2 2 1 2 2 2 (12. (12. 0 .38) are ud. For a precise description of the two nucleons as composite states.43) Similarly.− 2 2 .38) which have all nearly identical masses. the three pions in (12. −1 = = = 1 1 1 1 . albeit it is a good approximate one. In the framework of the quark of model. . the fundamental representation of the isospin symmetry corresponds to the doublet that contains the so-called up and down quarks u= 1 1 . 2 (12. the proton and the neutron can be written as totally symmetric uud and udd states. I. (12. I3 . are made up of these two quarks. spelling disaster for the stability of matter.0M eV /c2 ). 2 2 .40) + 2 1 1 .− . However. respectively.36) as a multiplet with I = 2 1 1 p = I = . . it can be seen as part of a larger (and more approximate) symmetry which is of great utility to classify observed particle families. They u form an isotriplet: π + = |1.− . Further than that. We can describe isospin multiplets the same way we have described the angular momentum and spin multiplets.41) (12. mπ0 = 135. (12. d= 1 1 .

1 |1. mesons have baryon number 0.11. baryon and meson. −1 = |p > |p >. I3 = 0. The reader may know that the deuteron. It shall be noted here that the quark model was not invented until 1960’s. All other composite states have their strangeness given by the sum of the strangeness content of their constituents.44) 2 where B is the baryon number and S is the strangeness. The up and down quarks have strangeness zero. Therefore the deuteron has to be an isosinglet state (|0. by accidental convention. the up and down quarks are not the only quark ‘species’ . Now let us consider another example of combining the isospins of two particles. The possibilities are that of an isosinglet 1 |0. (12. but at the time the empirical concepts like isospin and strangeness quantum numbers were in use. as some mesons discovered later are heavier than some baryons and so on. we shall setup some terminology: baryons are qqq states. Therefore it has to have isospin. In the late 1940’s and early 1950’s. strongly interacting particles) which is about a GeV /c2 . originally refer to the relative weight of particles. (12. Naturally. . 1 = √ (|p > |n > +|n > |p >).43). but such states do not exist.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 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. We will now try to describe its wave function in terms of its constituent nucleons. baryons generally are heavy. All antiparticles have their 3 quantum numbers reversed.46) (12.) are light. 2 = |n > |n > . 0 = √ (|p > |n > −|n > |p >) 2 and that of an isotriplet |1. consists of a proton and a neutron.37) and in analogy to (12. muon. The names.e. Before proceeding further. and quarks have baryon number 1 . the pions being examples thereof. This is why isospin is such a good symmetry and why isomultiplets have nearly identical masses. By convention ¯ baryons have baryon number 1. Following (12. to be −1 for the strange quark.47) (12. 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). 0 |1. where leptons (electron. The value of the strangeness quantum number is taken. a hydrogen isotope. As it later turned out.48) (12. the neutrinos etc. mesons have intermediate mass ranges.45) Is the deuteron an isosinglet state or an isotriplet? If it were an isotriplet (|1. this will be mathematically identical to adding two spins. whereas mesons are q q states. 0 ) we should have seen nn and pp bound states of comparable energy in nature (because of isospin symmetry). such as proton and the neutron. a number strange particles have been found which presumably contained a third quark species: the strange quark. 0 ). this remains only an inaccurate naming convention today. If taken literally.or ﬂavors as they are commonly called.

7 to the case of isospin. γ (f ) ||T0 ||I (i) . γ (f ) ||T ||I (i) . 0 + |0. I3 . is proportional to |M|2 . I3 (f ) = 1.52) where γ (i) and γ (f ) denote degrees of freedom other than isospin. such as the spatial dependence of the wave function and spin. which is an isoscalar. γ (f ) |T00 |I (i) I3 . Show that the expectation of I(1) · I(2) is state. 1 and (1/ 2)(|1. respectively. The initial and ﬁnal states can be denoted in more detail as | initial | ﬁnal = = I (i) . 1 and |1. γ (i) I (f ) . 0 . (12.45) and (12. 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. We will now employ the (isospin analogue) of the Wigner-Eckart theorem (6. γ (i) (i) (12. 1 4 in an isotriplet state and − 3 in an isosinglet 4 Using (12. We will eventually be able to compute ratios of scattering cross-sections between diﬀerent processes. γ (f ) T00 I (i) = 1.7 and 6.7. I3 . (f ) (i) (I (f ) I3 |00I (i) I3 ) is a Clebsch-Gordon coeﬃcient and I (f ) .48) the initial states in (I) and (II) have isospin values |1. For completeness let us start by restating the Wigner-Eckart theorem (6.55) . . Exercise.38) and the fact that the deuteron is an isosinglet we know that the isospins of the ﬁnal states in (I) and (II) are |1.5 for the spin-analogue of the same problem. where M is the scattering amplitude given by M = ﬁnal | α I(1) · I(2) | initial .259) in the present context: I (f ) I3 .49) The only assumption that we put in will be that the interaction is of the form V = α I(1) · I(2) . 0 ). Consider the generalization of tensor operators discussed in section 6. consider (I) p + p → d + π + (II) p + n → d + π 0 (12. γ (i) is a reduced matrix element deﬁned in the same sense as in section 6. Exercise. The cross-section. Show that I(1) ·I(2) is an ‘isotensor’ of rank zero. The dot product here refers to the abstract isospin space.259) discussed in detail in sections 6.380 Spinor Formulation As an exercise on the implications of isospin symmetry we will consider nucleon-nucleon scattering. γ (i) .54) Here T00 ≡ V = α I(1) · I(2) . I3 = 1.50) . √ According to (12. as discussed in the exercise above.8. For example. (12. σ. γ (i) = (f ) (i) (I (f ) I3 |00I (i) I3 ) (f ) (i) (−1)I −I (f ) (i) (12.8 to compute the ratio of the scattering amplitudes MI and MII . Refer to exercise 6. γ (f ) (f ) (i) (12.51) (12.53) 0 √ 1 2I (i) +1 I (f ) .

[2] As a further example.66) = 2 3 1 3 2. γ (f ) T00 I (i) = 0. particles with strangeness. γ (i) = √ 2 1 (f ) (i) +√ I (f ) = 1. I3 = 0.59) (12. involving. γ (f ) T00 I (i) = 1. = 381 (12. −1 π0 + n : |1. which is also manifested by a vanishing Clebsch-Gordon prefactor. −1 2 3 2 − + .63) where common dynamical factors (which would not be as easy to compute) have dropped out thanks to the Wigner-Eckart theorem. We want to compute the ratio of total cross-sections assuming a similar interaction as in the previous example σ( π + + p → anything ) .64) There are more exotic possibilities. 0 1 1 3 3 2. I3 = 0. −2 2 1 1 3 2.65) (12. we will consider pion-nucleon scattering. −2 1 1 2. γ (i) 3 1 (f ) (i) I (f ) = 1. .58) (12. 2 . 2 = 3 2.60) (12. σ( π − + p → anything ) The possibilities are (a) π + + p → π + + p . γ (i) 3 +0. −2 .61) MII Note that the second term in MII vanishes due to the isospin conservation. I3 = 0. σII which is in approximate agreement with the observed ratio.5: Lorentz Invariant Field Equations 1 (11|0011) √ I (f ) = 1. (c) π − + p → π 0 + n . (12. The relevant Clebsch-Gordon coeﬃcients are easily evaluated: (11|0011) = (10|0010) = 1. 1 1 1 1 3 √ 2. 2 = 2. for example. Once again we need the isospins for the initial and ﬁnal states.56) (12.57) (12. = MII (1/ 2) (12. γ (f ) ||T00 || I (i) = 1. γ (f ) ||T00 || I (i) = 1. (b) π − + p → π − + p . (12.11. I3 = 0. 1 π − + p : |1. γ (i) 2 1 = (10|0010) √ I (f ) = 1. (12. but these are not dominant at relatively low energies. It follows σI = 2.62) We can now write the ratio of the scattering amplitudes: MI 1 √ . −2 1 1 √ . which are obtained by a standard Clebsch-Gordan expansion π + + p : |1.

For instance. which can be produced by π − + p → K + + Σ− . It was assumed that strong interactions conserved S (at least approximately). . Hence the strangeness changing strong decays of K + (or Σ− ) were forbidden.382 Spinor Formulation As in the example of nucleon-meson scattering we deﬁne the relevant matrix elements M3 = 2 M1 = 2 3 2. giving it a higher than usual lifetime. are up to 40% heavier. 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. mentioned in the previous section.s = 0 . K + . m 1 2. Hence Gell-Mann and independently Nishijima postulated the existence of a separate quantum number. S. m .d = 1 . m 1 2. m V V 3 2. such that S(K + ) = 1. was motivated by the existence of particles that are produced strongly but decay only weakly. for example. called strangeness.71) u = 0 .3 The Eightfold Way and the ﬂavor SU (3) symmetry The discovery of the concept of strangeness. however. which leads to the following 2 2 ratios for the cross-sections σa : σb : σc = 9 : 1 : 2. making an identiﬁcation of the underlying symmetry and the multiplet structure less straightforward.69) 12. has a lifetime which is comparable to that of π + albeit being more than three times heavier. S(Σ− ) = −1 and S(π) = S(N ) = 0. while weak interactions did not. it appears an obvious generalization to add another component for an extra quark to the isospin vector space 1 0 0 (12. The classiﬁcation of the newly found particles as members of some higher multiplet structure was less obvious then the case of isospin. etc. [2] (12.70) (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. In the light of the quark model. (12.65) Ma = M 3 2 Mb = 1 M 3 + 2 M 1 3 3 √ 2 Mc = 2 3 M3 2 √ 2 (12.67) which are independent of m. Strange partners of the familiar nucleons. 0 0 1 .68) − 2 3 M1 2 Guided by empirical data we will further assume that M 3 >> M 1 .

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

As in the case of SU (2) higher dimensional representations obeying the same structure can be found. U . .78) where fjkl are the antisymmetric structure constants similar to the jkl of SU (2) given in (5.75) αk λk (12. (12. δjkl . −2 λ4 λ6 λ8 The generators. can be written in the form U = eiH where H is a hermitian matrix. . As discussed in section 5.79). obey the following relation tr(λj λk ) = 2δjk . λk ]+ = δjk + 2δjkl λl . (We shall at times refer to the Lie algebra with the name of the group. 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 . (12. λ3 = 0 −1 0 . λk ] = 2ifjkl λl . regardless of the dimension of the representation. Therefore we will express elements of SU (3) as U = ei k (12. λ7 = 0 0 . 3 (12.77) λ1 . λk .76) where λk are 8 linearly independent matrices forming the basis of the Lie algebra of SU (3).7. λ5 = . λ2 = . since det(eA ) = etrA . The Lie algebra structure is given by the commutators of λk [λj .32). 0 0 0 .80) This is the fundamental or deﬁning representation of SU (3). The reader is invited to compare the structure of SU (3) to that of SU (2) discussed in section 5.384 Spinor Formulation SU (3) and not U (3) that is of main interest. any unitary matrix. which can be veriﬁed explicitly as matrix identities. We can also introduce the constants.1. via the anticommutator relation 4 [λj .79) (12.) The unit determinant condition requires that all λk are traceless. The fundamental relation to be preserved is (12. the meaning being apparent from the context.

In the basis (12. I3 . [Y.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 . T0 } and [Y. (Recall that the strange quark has S = −1 by convention). Derive the Gell-Mann–Nishijima relation (12.81) In chapter 5. (12. starting with the observation that Y = B + S.44). Exercise. T± ] = ±T± .84) (12.89) (12.88) which deﬁnes (not uniquely) a maximal set of mutually commuting operators {Y.86) (12.87) the commutation relations between the generators can be expressed in a succinct manner. these relations have been used to construct the angular momentum spectrum as well as the function space representation of the rotation algebra. which satisﬁed an ‘eigenvalue equation’ [J0 . V± ] = ±V± . V± = F4 ± iF5 .83) (12. (12. We ﬁrst introduce the F-spin operators 1 Fi = λ i .90) (12. In the case of SU (2) the identiﬁcation of ‘ladder’ operators {J+ .92) (12.85) (12.5: Lorentz Invariant Field Equations 385 As it turns out the set of generators in (12.77) is not the most useful basis in the study of SU (3). 2 1 [T0 . 3 (12. The generators of SU (3) can be arranged into a very similar form to that of SU (2). 2 (12. 2 Y = √ F8 .94) . J± ] = ± J± .87) Exercise. J− } proved useful. namely spherical harmonics.93) (12. Y is called the hypercharge.71) show that T0 is the isospin operator. T± ] = 0. U± ] = ±U± . First. 2 (12. we have [Y. [Y. (12. U± ] = U± . 1 [T0 . V± ] = ± V± . T0 ] = 0.11. U± = F6 ± iF7 . Using the convention in (12. T 0 = F3 .

we have [T+ . We can construct two such independent Casimir operators for the group SU (3).97) which relate commutators of generators with opposite eigenvalues to the maximal set {Y. while SU (3) has rank 2. SU (2) has rank 1. the representations will now lie in a T0 − Y -plane.5. U− ] = 0. However. jkl (12.98) (12. This form is essential for easy labeling of the representations of the group. U− ] = Y − T0 = 2U0 . [U+ . Finally. V− ] = T− . [T+ .104) (12.386 Spinor Formulation which relate the remaining generators {T± .88).101) (12. When the basis of a Lie algebra is expressed in such a way to satisfy the form of the eigenvalue relations as given above.105) C2 = jkl djkl λj λk λl In general the number of independent Casimir operators of a Lie group is equal to its rank. The interested reader is instead referred to a very readable account given in chapter 12 of [4]. U+ ] = V+ . as the relation between the states in a given representation can be conveniently expressed in terms of ladder operators. (12. 2 3 Y + T0 = 2V0 . V+ ] = 0. Casimir operators can be used to label irreducible representations of the Lie algebra. V+ ] = 0. we will use these commutation relations to construct representations of SU (3). Another important property of SU (2) is the existence of an operator (namely the total angular momentum. [U+ . T0 }. T− ] = 2T0 . V± . Note that. U± } to this maximal set by ‘eigenvalue equations’ and [T+ . In the case of SU (2) the representations lay on a line on the J0 axis. U0 and V0 are linear combinations of T0 and Y .99) (12. [T+ . 3 [U+ . A formal deﬁnition and a detailed discussion of the Cartan-Weyl form is beyond the scope of this chapter.95) (12. [T+ . [V+ . As in the case of J 2 and SU (2). The same way the angular momentum ladder operators have been used to construct the representations of SU (2). similar to the way it was done in section 5. C1 = k λ2 = − k 2i 3 fjkl λj λk λl . An operator which commutes with all generators of a Lie group is called a Casimir operator. V− ] = 2 (12. J 2 ) which commutes with all of the generators. it is said to be in Cartan-Weyl form.100) (12. Thus.102) (12.96) (12. The maximum number of mutually commuting generators of a Lie algebra is called its rank.103) Any remaining commutators follow from hermiticity. since there are two mutually commuting generators in SU (3) as given in (12. V− ] = −U− . .

As another example for the representations of SU (3). |t3 .88) we can label states by the eigenvalues of T0 and Y operators.111) (12. 2 1 = β t3 ± .108) (12.) All other representations can be constructed by combining these two conjugate representations.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 .106) (12. The details of this procedure is beyond the scope of this chapter.109) The same way that J± |m is proportional to |m ± 1 in the case of the angular momentum algebra. (12. y . The representations D(1. the pion family forms part of an octet corresponding to the D(1.y ± 1 .2) shows the representation D(2.1). 2 (12. 1) representation. (12. U± . Because of (12. 2 1 = γ t3 . 2 (12. respectively. All other states of the representation are then constructed by successive application of ladder operators T± . This property can be used to label representations in terms of the values of the Casimir operators. The interested reader is referred to [4].11.5 how to label the irreducible representations of the angular momentum algebra SU (2) in terms of the value of the total angular momentum. = y |t3 . A notation suggestive of the dimensionality of the representation can be used to identify representations. The representations of SU (3) are constructed analogous to those of SU (2) by identifying the ‘boundary’ states annihilated by raising (or lowering) operators. y 3 = ( y− 4 3 = ( y+ 4 1 t3 ) |t3 . Such a representation is labeled as 1 D(p. The representations for hexagons with sides of length p and q in the T0 − Y -plane. Figure (12. For example.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. For example. . Now we will construct explicit representations of SU (3). q) and it has a dimensionality of 2 (p + 1)(q + 1)(p + q + 2). we have U0 |t3 .3).antiquark pair. V± . y . y U± |t3 . 2 1 t3 ) |t3 . For example the pion octet (or any other meson octet) is therefore realized as states of a quark . 1) as an example. y V± |t3 . y . y V0 |t3 .110) (12. y = t3 |t3 . (See Fig. y . y 1 = α t3 ± . it was shown in section 5. 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. y . we have T± |t3 . y : T0 |t3 . y . y Y |t3 . 0) and D(0. 1) correspond to the triplets of quarks and antiquarks. y ± 1 . (12.112) The eﬀect of these operators to the states in the y − t3 plane have been outlined in Fig. especially chapters 7 and 8. For example.

1) representation of SU (3). ¡ ¢ ¡ ¢ 8 6 2 '745301) ¥¦¤ ¤ #" ! (& % '$ © ¨ ¦ §¥ ©§ ¨ . £ Figure 12. The states in the inner triangle are doubly degenerate.1: The eﬀect of SU (3) ladder operators on the y − t3 plane.2: D(2.388 £ Spinor Formulation Figure 12.

. − . where we would write [3] ⊗ [3] = [1] ⊕ [3] ⊕ [5]. 2 (12. for example. This way the quark-antiquark states can be represented as follows [3] ⊗ [¯ = [8] ⊕ [1] 3] (12. 1) is written as [8] etc.116) (12. 0) or the fundamental representation of ﬂavor SU (3) symmetry.113) The additional singlet state corresponds to the η meson.115) (12. 2 3 2 0. 0) and D(0.114) Figure 12.117) The Gell-Mann–Nishijima relation can then be succinctly expressed as 1 Q = y + t3 .3: D(1. The three quarks in the fundamental representation can now be written as u = d = s = 1 1 .5: Lorentz Invariant Field Equations 389 D(1. to the case of adding two spin triplet states. 3 (12.118) ¡ ¢ ¨ ©§ ¨¥¦¤ " ! . The octet D(1. 1) are denoted by [3] and [¯ respectively. in the case of SU (2). . 2 3 1 1 − . This expansion can be compared. 3]. £ (12.11.

119) (12. T. p. Wiley. 1984. Phys. Eisenberg. Wiley. 1989. Martin. W. D. Wiley. D. Oxford. [7] Expansion method for stationary states of quantum billiards. Halzen. Am. G. M¨ller. Schulten. (1999). 3 1 Qd = − . J. 1987. B. F. J.121) L 0 X - X 0 Figure 12. Cheng. Springer-Verlag u [5] Principles of symmetry.4: The baryon octet as a D(1. Harter. dynamics. K.120) (12. L. 1993. D. I. Kosztin. Kaufman. M. [4] Quantum mechanics: symmetries.-F. References [1] Quarks and leptons: an introductory course in particle physics. 3 £ Spinor Formulation (12.-P. [6] Gauge theory of elementary particle physics. ¡ ¢ S - S 0 ¤ ¥ S + . 1) representation of SU (3). North holland. [2] Introduction to elementary particles. 3 1 Qs = − . Li. and spectroscopy. W. 1984. L. W. 67 (2). 1972. Greiner. [3] Microscopic theory of the nucleus.390 from which the quark charges follow 2 Qu = . Griﬃths. A. Greiner. 133.

Are you sure?

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

We've moved you to where you read on your other device.

Get the full title to continue

Get the full title to continue listening from where you left off, or restart the preview.

scribd