Notes on Quantum Mechanics

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

Chapter 1 Lagrangian Mechanics Our introduction to Quantum Mechanics will be based on its correspondence to Classical Mechanics. i. . ∀t. Definition: A functional S[ ] is differentiable.e. often in the Cartesian coordinate system..3) (ii) δS[q(t). 1. qM (t)). δq1 (t)] + α2 δS[q(t). · ] exists with the properties (i) S[q(t) + δq(t)] = S[q(t)] + δS[q(t). t1 ] ⊂ R} dt (1.1 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. t1 ] ⊂ R → RM . · ] is called the differential of S[ ]. t ∈ [t0 . there are 3N configurational coordinates.1) from a space F of vector-valued functions q(t) onto the real numbers. namely. namely. 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). . In case of N classical particles holds M = 3N . δq(t)] + O( 2 ) (1.2) . q(t) is called the trajectory of a system of M degrees of freedom described by the configurational coordinates q(t) = (q1 (t). if for any q(t) ∈ F and δq(t) ∈ F where F = {δq(t). 1 (1. An important concept is that the equations of motion of Classical Mechanics can be based on a variational principle. F = {q(t). | a functional δS[ · . The latter is the velocity vector of the system. |δq(t)| < . δq2 (t)] .4) d δq(t)| < . δq(t) ∈ F. Definition: A functional S[ ] is a map S[ ] : F → R . . q(t) differentiable} (1. the position coordinates of the particles in any kind of coordianate system. q2 (t). For this purpose we will review the relevant concepts of Classical Mechanics. q : [t0 . δS[ · . The linearity property above implies δS[q(t). δq(t)] is linear in δq(t). 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).

i.6) is expressed in terms of ‘normal’ functions. t) of the configuration d vector q(t). q(t).2 Lagrangian Mechanics Note: δq(t) describes small variations around the trajectory q(t). We attempt to evaluate t1 S[q(t) + δq(t)] = t0 ˙ ˙ dt L(q(t) + δq(t).e. equating these terms with δS[q(t). ∂f in case of a function of M variables.e. · ) is a function differentiable in its three arguments. t) (1. the velocity vector dt q(t) and time t. For this purpose we consider M ˙ ˙ ˙ L(q(t) + δq(t). We focus on such functionals because they play a central role in the so-called action integrals of Classical Mechanics.6) For a proof we can use conventional differential calculus since the functional (1. df = M ∂xj dxj . . t1 ] where the integrand is a function L(q(t). q(t) + δq(t) is a ‘slightly’ different trajectory than q(t). ˙ In the following we will often use the notation for velocities and other time derivatives d q(t) = q(t) and dtj = xj . t) = L(q(t).g. t0 (1. · . q(t) + δ q(t). δq(t)] = dt − δqj (t) + δqj (t)   ∂qj dt ∂ qj ˙ ∂ qj ˙ t0 j=1 j=1 t1 . · ] in conventional differential calculus does not correspond to a differentiated function. dt q(t). t) (1.8) 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 .5) where L( · . q(t).. δq(t)]. (1. t) + j=1 ∂L ∂L δqj + δ qj ˙ ∂qj ∂ qj ˙ + O( 2 ) (1. but rather to a differential of the function which is simply the df differentiated function multiplied by the differential increment of the variable. e.6) immediately. i. df = dx dx or.7) through Taylor expansion and identification of terms linear in δqj (t). ˙ Theorem: Let S[q(t)] = t0 dx dt t1 ˙ dt L(q(t). q(t) + δ q(t).10) From this follows (1. 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. It is also important to appreciate that δS[ · .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. j=1 We will now consider a particular class of functionals S[ ] which are expressed through an integral d over the the interval [t0 . at the ends of the time interval of the trajectories the variations vanish. It holds   M  M ∂L  t1 d ∂L ∂L δS[q(t)..

. 2. it holds in case δq(t0 ) = δq(t1 ) = 0 that qe (t) is an extremal.   ∂qj dt ∂ qj ˙ t0 j=1 (1. The length of such path can be determined starting from the fact that the incremental length ds in going from point (x. For this purpose we define Definition: An extremal of a differentiable functional S[ ] is a function qe (t) with the property δS[qe (t).6) which reads in the present case     M ∂L t1 d ∂L δS[q(t). t1 ] ⊂ R → R holds t1 (1. On the other side. . t1 ].12) for j = 1. .12) for j = 1. Let us assume that the two points are connected by the path y(x). if and only if it satisfies the conditions (j = 1.16) the property δS[qe (t).5). δq(t)] = 0 for all δq(t) ∈ F . 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 . from (1. Theorem: Euler–Lagrange Condition For the functional defined through (1. y(x)) to (x + dx. hence. then f (t) ≡ 0 on [t0 . 2. · ] ≡ 0 and. y(x1 ) = y1 .12) The proof of this theorem is based on the property Lemma: If for a continuous function f(t) f : [t0 .13) dt f (t)h(t) = 0 t0 (1. 2. This condition is stated in the following theorem.15) We will not provide a proof for this Lemma.5) assume extreme values.1. (1. . . (1. y(x2 ) = y2 .14) for any continuous function h(t) ∈ F with h(t0 ) = h(t1 ) = 0.17) . y1 ) and (x2 . . δq(t)] = dt − δqj (t) . .11) The extremals qe (t) can be identified through a condition which provides a suitable differential equation for this purpose. . y(x + dx)) is ds = (dx)2 + ( dy dx)2 = dx dx 1+( dy 2 ) . . .1: Variational Calculus 3 We now consider the question for which functions the functionals of the type (1. the above theorem. M ) d dt ∂L ∂ qj ˙ − ∂L = 0 ∂qj (1. dx (1. According to the Lemma above follows then (1. y2 ).16) This property holds for any δqj with δq(t) ∈ F . M . M and δqj (t0 ) = δqj (t1 ) = 0 follows according to (1. The proof of the above theorem starts from (1.

t) (1.20) x1 − x2 Exercise 1. The shortest path is an extremal of s[y(x)] which must..5) with L(y(x). (1. y = const. M (1. This in turn yields y(x) = ax + b. One obtains y 1 − y2 y(x) = (x − x2 ) + y2 . 1. i.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.e. circles whose centers lie at the center of the sphere. q(t). t1 S[q(t)] = t0 ˙ dt L(q(t). hence. j = 1. Threorem: Hamiltonian Principle of Least Action The trajectories q(t) of systems of particles described through the Newtonian equations of motion d ∂U (mj qj ) + ˙ = 0 dt ∂qj . Using y = dx the condition reads d dx ∂L ∂y = d dx y 1 + (y )2 = 0.19) From this follows y / 1 + (y )2 = const and. qM ) . q(t).4 The total path length is then given by the integral x1 Lagrangian Mechanics s = x0 dx 1+( dy 2 ) . . . t) = j=1 1 mj qj − U (q1 . 2. (1. .23) Presently we consider only velocity–independent potentials.18) dy s is a functional of y(x) of the type (1. the so-called action integral. dx ) = 1 + (dy/dx)2 . .21) are extremals of the functional.1. The constants a and b are readily identified through the conditons y(x1 ) = y1 and y(x2 ) = y2 . ˙2 2 (1. dx (1. . q(t). according to the theorems above.22) ˙ where L(q(t).1: Show that the shortest path between two points on a sphere are great circles. . . q2 . t) is the so-called Lagrangian M ˙ L(q(t). . obey the Euler–Lagrange dy condition. Velocity–dependent potentials which describe particles moving in electromagnetic fields will be considered below.

t) + V (r.22.26) (1.24) ∂qj dt ∂ qj ˙ ∂qj dt which are obviously equivalent to the Newtonian equations of motion. t) and B(r. t) as follows B E = ×A 1 ∂A .31) allows one to transform the potentials without affecting the fields and without affecting the equations of motion (1.1. t).2: Lagrangian 5 For a proof of the Hamiltonian Principle of Least Action we inspect the Euler–Lagrange conditions associated with the action integral defined through (1. 1.27) and (1. x3 (t)) moving in an electromagnetic field described through the electrical and magnetic field components E(r.30) (1. t) . t) obey the Maxwell equations ×E + ·B ×B − ·E 1 ∂ c ∂t E 1 ∂ c ∂t B (1. Equations (1. respectively.30. The transformation which leaves the fields invariant is A (r. t) + v × B(r.32) (1. t) describes the charge current density. t) and B(r. c ∂t (1. x2 (t).31) = − V − Gauge Symmetry of the Electromagnetic Field It is well known that the relationship between fields and potentials (1. t) = V (r. t) describes the charge density present in the field and J(r. These conditions read in the present case ∂L d ∂L ∂U d − = 0 → − − (mj qj ) = 0 ˙ (1. t) 1∂ K(r.25) = 0 = 0 4π J c = 4πρ = (1. 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). The equations of motion for such a particle are d q ˙ (mr) = F (r. t) c ∂t (1.28) (1. F (r.33) . t) and a vector potential A(r. t) − K(r. dt The fields E(r. t) = q E(r. 1.23).27) (1.29) where ρ(r. t) is the Lorentz force. t) = A(r.25) of a particle moving in the field.28) can be satisfied implicitly if one represents the fields through a scalar potential V (r. t) dt c where dr = v and where F (r.

30). dt ∂x1 c We notice using (1. 1. (1. e. (1.40.35) − ∂A1 ∂x2 [v × B]1 = x2 B3 − x3 B2 = x2 ˙ ˙ ˙ ∂A2 ∂A1 − ∂x1 ∂x2 − x3 ˙ ∂A1 ∂A3 − ∂x3 ∂x1 .34) For this purpose we consider only one component of the equation of motion (1. t) · v .41) with (1. if one assumes for a particle the Lagrangian ˙ L(r.36) in the Newtonian equation of motion.37) The results (1.35) is equivalent to the Euler–Lagrange condition d dt The second term in (1. d ∂V q (mv1 ) = F1 = −q + [v × B]1 . 2 c (1. (1.36) This expression allows us to show that (1.25) follows from the Hamiltonian Principle of Least Action.38) ∂L ∂ x1 ˙ − ∂L = 0.39) ∂A1 ∂A2 ∂A3 x1 + ˙ x2 + ˙ x3 ˙ ∂x1 ∂x1 ∂x1 .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 .g. .37) is ∂L ∂V q = −q + ∂x1 ∂x1 c The first term in (1.35) shows that the Newtonian equations of motion and the Euler–Lagrange conditions are. 1. ∂x1 (1.6 Lagrangian of Particle Moving in Electromagnetic Field Lagrangian Mechanics We want to show now that the equation of motion (1.25).38. r.40) (1. equivalent.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. B3 = ∂A2 ∂x1 (1.. in fact.41) which is identical to the term (1. Comparing then (1. t) = 1 q mv 2 − q V (r. namely. t) + A(r.

q.33) of electromagnetic potentials. encountered above in connection with the transformations (1. equivalent to the gauge transformation (1. Note that one adds the total time derivative of a function K(r. t) · v + K(r.45) .33) of electromagnetic potentials. t) − + 2 c ∂t c 1 q = mv 2 − q V (r. 1. 1. t) c t1 t0 = S[q(t)] + (1.45) and reordering terms yields using (1. We will consider in the following two symmetries. a term not affected by the variations allowed. t) = mv 2 − q V (r. dt ∂x1 ∂x2 ∂x3 ∂t ∂t (1. t) the Lagrangian. t) = x1 + ˙ x2 + ˙ x3 + ˙ = ( K) · v + . t) · v . q.32. i. 2 c c dt Inserting (1.32. t) = q K(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. t) + A (r.46) (1.43) To prove this theorem we determine the action integral corresponding to the transformed Lagrangian t1 S [q(t)] = t0 ˙ dtL (q. q. r. t) + A(r. 1.1.3: Symmetry Properties 7 1. appear in a different.43) into (1.34) q 1 q d ˙ L (r.33) of electyromagnetic potentials. The factor q has been introduced to make this c transformation equivalent to the gauge transformation (1. t) + dt c (1. The gauge symmetry. r. They are the subject of the following theorem. t) + q K(q. 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. t) = L(q. This term is d ∂K ∂K ∂K ∂K ∂K K(r. t) c t1 t0 t1 t0 ˙ dtL(q.e. surprisingly simple fashion in Lagrangian Mechanics. t) 1 1 ∂K q mv 2 − q V (r. 2 c = A(r. 1. t) + K ·v (1. For this purpose we consider the transformation of the single particle Lagrangian (1. t) L (q. (1.33) ˙ L (r.. and since these properties allow one often to simplify mathematical descriptions.42) is. One can conclude then immediately that any extremal of S [q(t)] is also an extremal of S[q(t)].44) implies that the gauge transformation amounts to adding a constant term to the action integral.44) Since the condition δq(t0 ) = δq(t1 ) = 0 holds for the variational functions of Lagrangian Mechanics. gauge symmetry and symmetries with respect to spatial transformations. Eq. in fact. t) . q.32.32.42) This transformation is termed gauge transformation. We want to demonstrate now that the transformation (1.

8 Lagrangian Mechanics Obviously. which has an important analogy in Quantum Mechanics. t).50) A non-trivial example is furnished by the infinitesimal rotation around axis e ˆ r = r + e×r .48) In the following we will denote unit vectors as a. In this case the transformation can be written in matrix form      cos −sin 0 x1 x1  x2  =  sin cos 0   x2  (1..33). 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. 1. The corresponding infinitesimal transformation is defined for small r (r. t). the reason being that our interest lies in achieving familiarity with the principles (just mentioned above ) of symmetry properties rather than in providing a general tool in the context of Classical Mechanics. i. ˆ (1. Definition: Infinitesimal One-Parameter Coordinate Transformations A one-parameter coordinate transformation is decribed through r = r (r. the equivalence of (1. r ∈ R . ) . A(r. 0) = r . for such vectors holds a · a = 1. for example. ˆ (1. ) = r + R(r) + O( 2 ) . The transformations considered are specified in the following definition.e. the transformation (1. therefore.47) (1. ˆ ˆ ˆ Examples of Infinitesimal Transformations The beauty of infinitesimal transformations is that they can be stated in a very simple manner. for the connection between invariance properties and constants of motion. However. r. A (r.51) We would like to derive this transformation in a somewhat complicated.49) =0 ∈ R (1. In case of a translation transformation in the direction e nothing new is gained. R(r) = through ∂r ∂ (1. Such invariance properties are very familiar. t). The transformations we consider are the most simple kind.32.42) and (1. We consider now invariance properties connected with coordinate transformations. where the origin of is chosen such that r (r. t) by gauge transformed potentials V (r.52) x3 0 0 1 x3 . We have proven. The following description of spatial symmetry is important in two respects.42) corresponds to replacing in the Lagrangian potentials V (r. we like to ˆ provide the transformation here anyway for later reference r = r + e. but nevertheless instructive way considering rotations around the x3 –axis.

sin = + O( 2 )        0 − 0 x1 x1 x1  x2  =  x2  +  0 0   x2  + O( 2 ) . then M ∂L j=1 Qj ∂ xj is a constant of motion. neglecting terms of order O( M ˙ ˙ L (q. We have used here the notation corresponding to single particle motion. The property has been shown to hold in a more general context.58) From this follows the statement of the theorem. Using the fact.3: Symmetry Properties 9 In case of small this transformation can be written neglecting terms O( 2 ) using cos = 1 + O( 2 ). expansion. i.54) which is equivalent to (1. t) is invariant with respect to an infinitesimal transformation q = q + Q(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. We consider here only the ‘particle version’ of the theorem. Anytime. (1. by Noether.51) reads r = r − x2 e1 + x1 e2 ˆ ˆ (1. In order to prove Noether’s theorem we note qj qj ˙ = qj + = qj + ˙ k=1 Qj (q) M (1. Theorem: Noether’s Theorem ˙ If L(q.e.. i.k=1 ∂L ∂Qj qk ˙ ∂ qj ∂qk ˙ (1. the second and third term ˙ k=1 in (1.53) 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. t) + j=1 ∂L Qj + ∂qj M j. In the context of Lagrangian Mechanics this implies that such transformation leaves the Lagrangian of the system unchanged. q..56) ∂Qj qk . Invariance implies L = L. q. however. 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. namely for fields rather than only for particle motion.55) (1. q. 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. t) = L(q. a classical mechanical system is invariant with respect to a coordinate transformation ˙ a constant of motion exists.1. t) yields after Taylor ˙ 2 ).e. .57) must cancel each other or both vanish. ˙ We have generalized in this theorem the definition of infinitesimal coordinate transformation to M –dimensional vectors q. a quantity C(r. r) which is constant along the classical path of the system.57) ∂ d where we used dt Qj = M ( ∂qk Qj )qk . ˙ ∂qk ˙ Inserting these infinitesimal changes of qj and qj into the Lagrangian L(q.53). q. the property holds for any system.

2. In this case we have Qj = ej · e.e. for which translational invariance holds. hence. not infinitesimal transformations. v) = 1 mv 2 − U (r). It is not necessary to investigate the consequences of global.51).59).50).59) yields the constant of motion ˆ e (ˆ × r) · mv. 2. is conserved. . to be expected that this is the constant of motion. It was. 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. We will now investigate the consequence of rotational invariance as described according to the infinitesimal transformation (1. A calculation similar to that in (1. Noether’s theorem yields ˆ ˆ the constant of motion (qj = xj . 3 and. 2 We first determine the constant of motion in case of invariance with respect to translations as defined in (1. j = 1. j = 1. of course. 3) 3 Qj j=1 ∂L = e· ˆ ∂ xj ˙ 3 ej mxj = e · mv . In this case we will use the same notation as in (1. except using now Qj = ej · (ˆ × r).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. ˆ ˙ ˆ j=1 (1. i.59) We obtain the well known result that in this case the momentum in the direction. 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.

4) a condition which enforces that the probability of finding the particle somewhere in Ω at any particular time t in an interval [t0 . t ∈ [t0 . t) ψ : R3 ⊗ R → C. We provide now a set of formal rules which state how the probability to observe such a particle at some space–time point r. 11 . The particle is described by a wave function ψ(r. t) is normalized d3 r |ψ(r. r. 2. At present one may think of f (r) as a sum over δ–functions which represent a multi–slit screen.2) (2. (2.2 Axioms for Quantum Mechanical Description of Single Particle ˙ Let us consider a particle which is described by a Lagrangian L(r. The probability that the particle is detected at space–time point r. 4. t) where z is the conjugate complex of z. t1 ] . t is |ψ(r. t)|2 = ψ(r. 3. The wave function ψ(r. placed into the space at some particular time and with a detector behind each slit. The probability to detect the particle with a detector of sensitivity f (r) is d3 r f (r) |ψ(r. t is described in Quantum Mechanics.Chapter 2 Quantum Mechanical Path Integral 2. t1 ] in which the particle is known to exist. t). 1.1 The Double Slit Experiment Will be supplied at later date 2. t)|2 = 1 Ω ∀t. t)ψ(r.1) (2. t)|2 Ω (2.3) where Ω is the space volume in which the particle can exist. is unity.

12 Quantum Mechanical Path Integral 5. t|r . The definition will actually assume an infinite number of intermediate times and express the path integral through integrals of the type (2. t1 ] yields an expression of the form r(tN )=rN φ(r. the propagator is unitary. t0 ) = φ(r. t|r . The following so-called completeness relationship holds for the propagator (t > t [t0 . t0 ) = δ(r − r0 ). t1 ]) Ωd Ωd 3r Ωd 3r Ωd 3r 3 r |ψ(r. . t| rN −1 . t1 ] (2. t1 ]) d3 r φ(r. t|r . (t > t . t0 ) = Ωd 3r N −1 Ωd 3r N −2 ··· Ωd 3r 1 φ(r. namely the point on the classical path at time t . This must hold for any ψ(r .9) Employing a continuum of intermediate times t ∈ [t0 . The state of a system at time t. t. t ) Ωd 3 r |ψ(r. i. t) is described by a linear map of the type ψ(r. t ) φ(r. t ) which requires d3 r φ(r.10) We have introduced here a new symbol. t|r0 . t0 ) . t ) ψ(r . (2. t|r .5) 6. t ) ψ(r . tN −1 |rN −2 . t)|2 = (2. t1 |r0 . tN −2 ) · · · φ(r1 . t ∈ [t0 . 8.6) φ(r.11) which denotes an integral over all paths r(t) with end points r(t0 ) = r0 and r(tN ) = rN . t|r . This symbol will be defined further below. t. t ) φ(r . t ) ψ(r . t |r0 . 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.8) 7. Since (2. t). t ∈ (2. t ) = δ(r − r ) Ω (2. hence. t ∈ [t0 .9) for N → ∞. t )φ(r. The generalization of the completeness property to N − 1 intermediate points t > tN −1 > tN −2 > . This is different from the classical situation where the particle follows a discrete path and.7) t. t ) t > t . t0 ) = r(t0 )=r0 d[r(t)] Φ[r(t)] .8) a knowledge of the state at all space points r ∈ Ω at some intermediate time t . i. t|r0 . at any intermediate time the particle needs only be known at one space point. The time evolution of ψ(r. .. ψ(r0 ..e. t = )|2 = 1 . . described by ψ(r. > t1 > t0 is φ(r.e. requires then according to (2.4) holds for all times. the path integral r(tN )=rN d[r(t)] · · · r(t0 )=r0 (2. t|r0 . t|r . (2. tN −1 ) φ(rN −1 . t) = Ω d3 r φ(r.

This implies that small variations of the path near the classical path alter the value of the action integral by very little.0545 · 10−27 erg s .e. 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.11) is Φ[r(t)] = exp where S[r(t)] is the classical action integral tN 13 i S[r(t)] (2. a very large number. However. However. contribution to the path integral (2. The constant given in (2. Its value is extremely small in comparision with typical values for action integrals of macroscopic particles.13) and = 1. interferences should be much less dramatic than in case of the macroscopic particle. Scl / ≈ 8. t) is the Lagrangian of the classical particle. in complete distinction from Classical Mechanics.14) ˙ In (2.2: Axioms 9. expressions (2.13) are built on action integrals for all possible paths. not only for the classical path. for microscopic particles like the electron this is by no means the case.14) has the same dimension as the action integral S[r(t)].12) is then Scl / ≈ 0.12. Situations which are well described classically will be distinguished through the property that the classical path gives the dominant.15) The exponent of (2. However.12) S[r(t)] = t0 ˙ dt L(r. the exponent is a very large imaginary number.2. t) (2. 2. i. In case of a proton with mass m = 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.5 · 1027 . 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. for the electron many paths contribute and the action integrals for non-classical paths need to be known. Only for paths close to the classical path is such interference ruled out. The situation is very different for microscopic particles. 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.13) L(r. namely due to the property of the classical path to be an extremal of the action integral.. r. The value of S[r(t)] is then Scl = 1 1 m v 2 t = erg s . The functional Φ[r(t)] in (2. One would still expect significant descructive interference between contributions of different paths since the value calculated is comparable to 2π. . 2 2 (2.e. However. (2.12. such that destructive interference of the contributions of such paths does not occur.13). 2. r. accordingly. it is comparable to action integrals as they arise for microscopic particles under typical circumstances. i. actually often essentially exclusive.. Since this number is multiplied by ‘i’.

˙ 2 (2. For the sake of simplicity we will consider the case of particles moving in one dimension labelled by the position coordinate x. t1 ). In passing from one space–time instance (xj . (xN −1 . This allows us to write the evolution operator using (2. Its value is CN = m 2πi N 2 (2.9). 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 . x. 2 2m N (2. t) = ˙ 1 mx2 − U (x) . t0 ). tj+1 ) we assume that kinetic energy and potential energy are constant.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 . > t1 > t0 letting N go to infinity later. tN |x0 . t0 ) = limN →∞ CN exp i N N −1 j=0 +∞ +∞ +∞ −∞ dx1 −∞ dx2 . .18) The time instances are fixed. x2 .14 Quantum Mechanical Path Integral 2. namely 1 m(xj+1 − xj )2 / 2 and U (xj ). a series of times tN > tN −1 > tN −2 > .20) can be properly taken.21) N 2. the xj values are not. (2. The particles have associated with them a Lagrangian L(x. however. . at the same time.13) and. tj ) to the next (xj+1 . (2.12) φ(xN . ∞[. tN )} .20) Here. .16) In order to define the path integral we assume. as in (2. The spacings between the times tj+1 and tj will all be identical. etc. 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. tN −1 ).3 How to Evaluate the Path Integral In this section we will provide an explicit algorithm which defines the path integral (2. namely tj+1 − tj = (tN − t0 )/N = N . They can be anywhere in the allowed volume which we will choose to be the interval ] − ∞. (x1 . provides an avenue to evaluate path integrals. (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.10) and (2. respectively.12. . 2. . . (xN . . .19) The main idea is that one can replace the path integral now by a multiple integral over x1 . −∞ dxN −1 (xj+1 −xj )2 1 − U (xj ) .

(2. can be expressed through a path integral with endpoints x(t0 ) = 0.13). the origin of which coincides with the classical path of the particle.23). t0 ) and (xN .31) We have denoted explicitly that the action integral for a path connecting the space–time points (x0 . since x(t0 ) = xcl (t0 ) = x0 and x(tN ) = xcl (tN ) = xN .h.23) tN − t0 It is essential for the following to note that. tN ). To see the benifit of such approach we define a path y(t) as follows x(t) = xcl (t) + y(t) (2. x(tN ) = 0] .24). ˙ t0 (2.12) yields x(tN )=0 im (xN − x0 )2 i φ(xN .27) The third term can be written in the notation introduced tN 1 dt my 2 = S[x(t)|x(t0 ) = 0. tN |0. x(tN ) = xN ] is S[x(t)|x(t0 ) = x0 . t0 ) (2.25) + The condition (2. x(tN ) = 0] . . which can also be written φ(xN . Inserting the result into (2.e.30) d[x(t)] exp 2 tN − t0 x(t0 )=0 a result. 2. The resulting expression for S[x(t)|x(t0 ) = x0 .2. using (2. x(t ) = x ] for any path x(t) can then be expressed through action integral 0 0 N N an action integral over the path y(t) relative to the classical path.24) implies for the second term on the r.22) where xcl (t) is the classical path which connects the space–time points (x0 .10. = (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 . ˙ 2 t0 (2.h.s. x(tN ) = 0.s. t0 ) = exp S[x(t)] (2. tN dt y = y(tN ) − y(t0 ) = 0 . t0 ) and (xN . tN |x0 . ˙ ˙ ˙ 2 + y2) = ˙ (2.26) The first term on the r. it is more suitable to define new integration variables yj .29) This expression corresponds to the action integral in (2. xN − x0 xcl (t) = x0 + ( t − t0 ) .25) is. The ˙ 1 S[x(t)|x(t ) = x . due to (2.24) Also we use the fact that the velocity of the classical path xcl = (xN −x0 )/(tn −t0 ) is constant. of (2. tN |x0 . x(tN ) = xN ] 1 (xN − x0 )2 m + 0 + 2 tN − t 0 + S[x(t)|x(t0 ) = 0. it holds y(t0 ) = y(tN ) = 0 . (2. t0 ) = exp 1 im (xN − x0 )2 2 tN − t0 φ(0.. namely. 2 2 tN − t0 (2. tN ) is to be evaluated.28) i.4: Propagator for a Free Particle 15 Rather then using the integration variables xj . One obtains S[x(t)|x(t0 ) = x0 . tN dt t0 1 1 (xN − x0 )2 m x2 = m ˙ cl .

.  0 0 0 0 0 0 The following integral +∞ +∞ (2.31) for a free particle one needs to evaluate the following path integral φ(0. .  . 2 N  .16 Evaluation of the necessary path integral Quantum Mechanical Path Integral To determine the propagator (2.36) which holds for a d-dimensional. . noting y0 = yN = 0..  .k=1 yj ajk yk   (2. . ..32) (yj+1 − yj )2 N −1 1 2 j=0 2 m N The exponent E can be written. 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 |0.33) where ajk are the elements of the following symmetric (N  2 −1 0  −1 2 −1  2 m  0 −1  ajk =  . .38) .34) of (ajk ) defining a new matrix (Ajk ) through ajk = Using det(ajk ) = m 2 N N −1 m 2 N Ajk ..35) must be determined.. t0 ) = limN →∞ × +∞ +∞ −∞ dy1 · · · −∞ dyN −1 exp i N m 2πi N 2 N × (2. .. In the appendix we prove  +∞ +∞ d j.k=1 dy1 · · · −∞ −∞ dyN −1 exp i yj bjk yk  =  (iπ)d det(bjk ) 1 2 . . (2. 0 0   ..k=1 yj ajk yk − 1) × (N − 1) matrix  . . (2. In order to complete the evaluation of (2. .  . symmetric matrix (bjk ) and det(bjk ) = 0.37) det(Ajk ) .. . real. 2 −1  −1 2 (2.. (2. 0 0 .34) dy1 · · · −∞ −∞ dyN −1 exp i  N −1 j. 0 0   .32) we split off the factor 2 mN in the definition (2.  .

42) m 2πi NN 1 2 (2.. tN |0.   . . Using D1 = |(2)| = 2 .18) we obtain φ(0. . .46) . .. 0 0     0 −1 2 . variable. .45) Expressions for Free Particle Propagator We have now collected all pieces for the final expression of the propagator (2. .41) the same result for D2 .41) D2 = 2 −1 −1 2 = 3 (2. det(Ajk ) (2.. (2. . .31) and obtain. defining t = tN .  . 2.40) . (2. t0 ) = m 2πi (tN − t0 ) 1 2 . t0 ) = limN →∞ and with NN (2. n = 1. . Our derivation has provided us with the value det(Ajk ) = N . t0 ) = limN →∞ m 2πi N 2 17 N 2πi m N N −1 2 1 .39) yields φ(0. (2.2. . x = xN φ(x. . t0 ) = m 2πi (t − t0 ) 1 2 exp im (x − x0 )2 2 t − t0 . . presently N − 1.43) We like to note here for further use below that one might as well employ the ‘artificial’ starting values D0 = 1.40) in terms of subdeterminants along the last column. . . Inserting this into (2. t|x0 ... ..    0 0 0 ..39) In order to determine det(Ajk ) we consider the dimension n of (Ajk ). . D1 = 2 and obtain from (2. 2 −1  0 0 0 −1 2 For this purpose we expand (2.. We seek then to evaluate the determinant of the n × n matrix   2 −1 0 . . To solve this three term recursion relationship one needs two starting values. One can readily verify that this procedure leads to the following recursion equation for the determinants Dn = 2 Dn−1 − Dn−2 . .. tN |0.4: Propagator for a Free Particle a property which follows from det(cB) = cn detB for any n × n matrix B. D3 . (2. 0 0    Dn =  .44) = tN − t0 . . let say n. tN |0. we obtain φ(0. one can readily verify Dn = n + 1 . which follows from (2. . 0 0  −1 2 −1 . . .

We need to evaluate for this purpose the integral ψ(x.52) and applying (2.46). x) + E(x) in (2.18 Quantum Mechanical Path Integral This propagator. the associated probability distribution |ψ(x0 . We will obtain. (2. One obtains then for the propagator (2.10) φ(r. x) = im 2 t x2 1 + i o t mδ 2 − 2xo x − po t m + f (x) (2. The result (2. the wave function of the particle at later times. We devide the contributions to the exponent Eo (xo . t|r0 . t0 )| = 2 1 πδ 2 1 2 exp − x2 0 δ2 (2.46) can be generalized to three dimensions in a rather obvious way.49) is Gaussian of width δ.5) allows us to predict the time evolution of any state function ψ(x.53) .50) One refers to such states as wave packets. and describes a single particle since 1 πδ 2 1 2 +∞ dx0 exp −∞ − x2 0 δ2 = 1.48) as the initial state using the propagator (2. t0 ) = 1 πδ 2 1 4 exp − x2 po 0 +i x 2 2δ (2. t0 ) = m 2πi (t − t0 ) 3 2 exp im (r − r0 )2 2 t − t0 . 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 . thereby. centered around x0 = 0. x) + E(x) (2.48) Obviously.51) as follows Eo (xo .5) to (2. (2. We want to apply axiom (2.51) 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. Below we will apply this to a particle at rest and a particle forming a so-called wave packet.247). according to (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 . t) of a free particle.

that the new path does not lead to additional contributions to the integral. (2.53) E(x) =  x− po mt 19 (2.2.59) An integration path in the complex x0 –plane along the direction i 1−i t mδ 2 (2.57) and needs to determine the integral +∞ I = −∞ +∞ dx0 eEo (xo .61) .54) f (x) =  This yields Eo (xo . however. t) = 1 πδ 2 1 4 2 t 1 + i mδ2  .4: Propagator for a Free Particle im x2 − f (x) . the square in (2.x)   im  xo 2 t  t − mδ 2 x− 1+ x− 1+ po m = −∞ +∞ dx0 exp  1+i t 2    2 t i mδ2 = −∞ im dx0 exp  2 t t 1+i 2 mδ xo − po m t t i mδ2  . x) = One can write then (2. 2 (2. z2 = +∞ × i 1−i t mδ 2 (2.51) ψ(x.55) im  xo 2 t  1+i t − mδ 2 x− 1+ po m t t i mδ2  . (2. We proceed as follows. (2. 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 .60) is then represented by real ρ values.52).58) The integrand is an analytical function everywhere in the complex plane and we can alter the integration path. making certain. 2 t One chooses then f (x) to complete. The beginning and the end of such path are the points z1 = −∞ × i 1−i t mδ 2 .x) (2.56) m 2πi t 1 2 +∞ eE(x) −∞ dx0 eEo (xo . according to (2.

63) which can be readily evaluated. z2 = +∞ .58) along the path z1 → z1 and along the path z2 → z2 gives only vanishing contributions one can replace (2. (2. 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 (2.55).70) . t m(1 + i mδ2 ) (2.67) We need to determine finally (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) ] . the paths between z1 → z1 and z2 → z2 have a real part of x0 of ±∞.68) and. (2.63) holds and. using a ab = a − .62) If one can show that an integration of (2.58) has a leading contribution in x0 of −x2 /δ 2 the integrand of (2.54) using (2. I = Equation (2.20 whereas the original path in (2.58) by t mδ 2 +∞ I = i 1−i dρ exp − −∞ m 2 t 1+ t mδ 2 2 ρ2 (2.58) vanishes 0 for Re x0 → ±∞. In fact.65) t 1 − i mδ2 t 1 + i mδ2 . Quantum Mechanical Path Integral (2.66) yields ψ(x. Hence.58) has the end points z1 = −∞ . 1+b 1+b finally E(x) = − (x − po m (2. accordingly.69) t)2 t 2δ 2 (1 + i mδ2 ) + i po x − i p2 o t 2m (2. (2. Since the exponent in (2.57) reads then 1 πδ 2 1 4 2πi t .64) ψ(x. one can show that z1 lies at −∞ − i × ∞ and z2 at +∞ + i × ∞. t) = Seperating the phase factor 1 t 1 + i mδ2 1 4 1 2 exp [ E(x) ] . We can conclude then that (2.

This ‘spreading’ of the wave function is a genuine quantum phenomenon. Another interesting observation is that the wave function (2. (2.48) and for the final state (2. The corresponding probability distribution is |ψ(x. t0 ) = exp i po i p2 o xo − to m 2m . t) = exp i po i p2 o x − t m 2m .71). We had assumed above [c. (2.72).76) From this we conclude that an initial wave function ψ(xo .48) and that the respective phase factor is related with the velocity of the classical particle and of the center of the distribution (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 .f.74) which corresponds to (2.72).. The conservation of this factor is particularly striking for the (unnormalized) initial wave function ψ(x0 . This yields. coinciding at t = 0 with the width of the initial distribution (2. the case of arbitrary to is recovered iby replacing t → to in (2.48) for δ → ∞.e. (2.71. (2. instead of (2.67) provides the complete expression of the wave function at all times t ψ(x. t) = exp i i p2 po o x − (t − to ) m 2m . In this case holds ψ(x.75) ψ(x.4: Propagator for a Free Particle which inserted in (2.75) i.73) m δ increases with time. (2. (2.71) conserves the phase factor exp[i(po / )x] of the original wave function (2.71) × exp − (x − po m t)2 ) 2δ 2 (1 + 2 t2 m2 δ 4 (1 − i . However. 2. t0 ) = exp i po xo m .72) for the initial state (2. the spatial dependence of the initial state (2.72) Comparision of Moving Wave Packet with Classical Motion It is revealing to compare the probability distributions (2. (2.48)] to = 0. The center of the distribution (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. 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 . 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 (2.49). The width of the distribution (2.2.72) 2 t2 δ 1+ 2 4 (2.77) .49).74) remains invariant in time.

t) = L(xcl .12. the end points of y(t) are y(t0 ) = 0 Inserting (2. 2. y(t). One refers to the respective states as stationary states.82) one obtains L(xcl + y. t) = exp i Quantum Mechanical Path Integral po i p2 o x − t m 2m .. t) + L (y. t) = ˙ 1 1 mx2 − c(t)x2 − e(t)x . For this purpose we use d xcl y = ˙ ˙ (xcl y) − xcl y ˙ ¨ (2.79) for a quadratic Lagrangian L(x.78) i. ˙ 2 2 tN (2. t) ˙ δL 1 1 mx2 − c(t)x2 − e(t)xcl ˙ cl cl 2 2 1 1 = my 2 − c(t)y 2 ˙ 2 2 = mxcl y(t) − c(t)xcl y − e(t)y . ˙ ˙ = (2. Such states play a cardinal role in quantum mechanics. y(tN ) = 0 . y(t)] which vanishes according to the Hamiltonian principle as discussed in Sect.81) for an arbitrary path x(t) with end points x(t0 ) = x0 and x(tN ) = xN .13) φ(xN .80) into (2. In order to simplify this task we define again a new path y(t) x(t) = xcl (t) + y(t) (2.85) We want to show now that the contribution of δL to the action integral (2.81) vanishes2 . (2. 2. the spatial as well as the temporal dependence of the wave function remains invariant in this case.82) which describes the deviation from the classical path xcl (t) with end points xcl (t0 ) = x0 and xcl (tN ) = xN .22 becomes at t > to ψ(x.83) (2.10. xcl . 2. xcl .80) For this purpose we need to determine the action integral S[x(t)] = t0 dt L(x. 2 . Obviously. x. 1. y(t). t) ˙ L (y.e.5 Propagator for a Quadratic Lagrangian x(tN )=xN We will now determine the propagator (2. t0 ) = x(t0 )=x0 d[x(t)] exp i S[x(t)] (2. (2.84) .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 . x. tN |x0 . xcl + y(t). t) + δL ˙ ˙ ˙ ˙ where L(xcl . t) ˙ (2.

87) According to (2. .89) where ˜ φ(0.32). 0 0  0 c2 0 .    0 0 0 . 0 0     0 −1 2 . . . One can express the exponent E in (2. tj = t0 + form j. t0 ) (2. . . t0 ) we will adopt a strategy which is similar to that used for the evaluation of (2. .31) for the free particle propagator.h. vanishes. . y. tN |0.91) (yj+1 − yj )2 1 2 2m N where cj = c(tj ).s. .  .2.91) through the quadratic N −1 E = i j.  ... . cN −2 0  0 0 0 0 cN −1 (2. of (2. 2 −1  0 0 0 −1 2   c1 0 0 . 0 0  −1 2 −1 . . t0 ) = exp S[xcl (t)] φ(0.93) . . . .   . t0 ) = limN →∞ × +∞ +∞ −∞ dy1 · · · −∞ dyN −1 exp N i N N −1 j=0 m 2πi N 2 N × − 1 2 cj 2 yj (2. . Applying the Euler–Lagrange conditions (1. For the evaluation of φ(0. (2... . . t0 ) = y(tN )=0 d[y(t)] exp y(t0 )=0 i tN dt L (y. 2 . The discretization scheme adopted above yields in the present case ˜ φ(0.88) and.. . . . . .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. . tN |x0 . ..24) to the Lagrangian (2.83) the first term on the r.h.88) vanishes. also the second contribution on the r. . . hence.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) . t) ˙ t0 . . 0 0  m   ajk =  ..k=1 yj ajk yk (2. . tN |0.. .79) i ˜ φ(xN .  2 N  . tN |0. tN |0. . . . 0 0     0 c3 . One can then express the propagator (2. . 0 0   N  0 −  .80) yields for the classical path m¨cl + c(t) xcl + e(t) = 0 x (2. x t0 (2.  0 0 0 .s.   . . . . .92) where ajk are the elements of the following (N − 1) × (N − 1) matrix   2 −1 0 .

24

Quantum Mechanical Path Integral

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

N

(iπ)N −1 det(a) 1

1 2

= limN →∞

1
2

N

2

N

N −1

m

det(a)

 

.

(2.94)

˜ In order to determine φ(0, tN |0, t0 ) we need to evaluate the function f (t0 , tN ) = limN →∞ According to (2.93) holds DN −1 =      =        2−
m c1
2 N 2 N

2
N

N

N −1

m

det(a)

.

(2.95)

def

2

N

N −1

m

det(a) 0 0 0 . . .
m cN −2
2 N

(2.96) 0 

−1 −1 2 − . . . 0 0
2 N

0 ... −1 . . . ... . .. . . .
m c3

−1 2 − 0 . . . 0 0

m c2

0 ... 2 − 0

−1 2 −

m cN −1

2 N

 0    0   .  .  .  −1  

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

m

cn

Dn−1 − Dn−2

(2.97)

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

m

c1

(2.98)

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

= −

cn+1 Dn . m

(2.99)

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

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

25

= =

N N

D0 = 0 ; D1 − D0
N

= 2−

2 N

m

c1 − 1 = 1 .

(2.101)

We have then finally for the propagator (2.79) φ(x, t|x0 , t0 ) = m 2πi f (to , t)
1 2

exp

i

S[xcl (t)]

(2.102)

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

2.6

Wave Packet Moving in Homogeneous Force Field

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

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

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

exp

i

S[xcl (τ )] .

(2.106)

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

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

Quantum Mechanical Path Integral

(2.108)

x − xo 1 f − (t − to ) t − to 2m

τ +

1 f 2 τ 2m

(2.109)

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

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

=

+ 2f

x − xo 1 f − (t − to ) t − to 2m

τ (2.111)

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

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

=

1 m 2 +f +

(t − to ) (t − to )2

1 f2 (t − to )3 + xo f (t − to ) 3 m

=

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

and, finally, for the propagator φ(x, t|xo , to ) = × exp m 2πi (t − to )
1 2

×

(2.113)

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

2.5: Propagator for a Quadratic Lagrangian

27

The propagator (2.113) allows one to determine the time-evolution of the initial state (2.48) using (2.5). Since the propagator depends only on the time-difference t − to we can assume, withoult loss of generality, to = 0 and are lead to the integral
+∞ m 1 1 4 2 ψ(x, t) = dx0 πδ 2 2πi t −∞ im (x − x0 )2 x2 po i i f2 3 exp − 02 + i xo + (x + xo ) f t − t 2 t 2δ 2 24 m
1

(2.114)

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

E(x) =

f t2 x − f (x) m

1 f 2 t3 . 24 m

One chooses then f (x) to complete, according to (2.52), the square in (2.115)  2 2 x − po t − f t m 2m  f (x) =  . t 1 + i mδ2 This yields Eo (xo , x) = im  xo 2 t  1+i x− t− t − 2 mδ t 1 + i mδ2
1 4 2 t2 ) m2 δ 4

(2.117)

po m

f t2 2m

2

 .

(2.118)

Following in the footsteps of the calculation on page 18 ff. one can state again ψ(x, t) =
t 1 − i mδ2 t 1 + i mδ2
1 4

1 πδ 2 (1 +

exp [ E(x) ]

(2.119)

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

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

(2.120)

1+i t mδ 2 −

− x−

x−

x− x

f t2 po t− m 2m

po f t2 t− m 2m
2

f t2 + 2x m

po f t2 t+ m 2m

f t2 po t+ m 2m

1+i

t mδ 2 (2.121)

28 Inserting this into (2.120) yields x− i
po m

Quantum Mechanical Path Integral

E(x)

=

− +

t−

2δ 2 1 + i

f t2 2m t mδ 2

2

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

(po + f t) x −

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

dτ (po + f τ )2 .
0

(2.123)

Altogether, (2.119, 2.122, 2.123) provide the state of the particle at time t > 0
t 1 − i mδ2 t 1 + i mδ2 
1 4

ψ(x, t)

=

1 πδ 2 (1 + x− 2δ 2
po m
2 t2 ) m2 δ 4

1 4

× 

 × exp  − × exp i

t−

f t2 2m

2

1+

2 t2 m2 δ 4

1−i
t

t mδ 2

  × . (2.124)

(po + f t) x −

i
0

(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

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


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 −

[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

e

−iωto

.

(2.177)

=

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

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)

=

x2 − 4

mω mω

x e−iωt +

2 e−2iωt mω (2.181)

=

mω 2 x + 2 2

x e−iωt − e−2iωt

. 2.s.172) is a function of t − to and defining accordingly Φ(x.e.182) Comparision with (2.186) Fourier transform.188) ˜ ˜ Equation (2. . t − to ) 1 ˜ φn (xo ) n! (2. t|xo .171) concludes the proof of (2.147) in (2.172) I = exp 2 mω x e−iωt − e−2iωt − Quantum Mechanical Path Integral mω 2 1 x − iωt 2 2 . i. 1 exp[iω(n + 2 ) to ] · · · . Accoordingly.172).183) where the expansion coefficients are functions of x.172). In contrast to W (x.h.188) identifies the functions ψn (x. xo . t) = exp[−iω(n + 1 ) t] φn (x) as invariants under the 2 action of the propagator φ(x. 2. We note that the factor exp(2 mω/ xe−iωt − e−2iωt ) in (2. t − to ) 1 ˜ exp[−iω(m + 1 ) to ] φm (xo ) 2 m! (2. t − to ) = φ(x. (2. We want to inspect the consequences of the invariance property (2. but not of t.185) Replacing t → t + to yields ∞ n=0 1 ˜ exp[−iω(n + 1 ) (t + to )] φn (x) 2 n! ∞ +∞ = m=0 −∞ dxo Φ(x. .172) in the form ∞ n=0 (2. of (2. t).187) or 1 ˜ exp[−iω(n + 2 ) t] φn (x) +∞ = −∞ ˜ dxo φ(x. xo . 2 (2. t|xo . one obtains for the r. . n = 1.36 Altogether. to ) we express (2. which also exhibits such invariance.171.184) 1 ˜ exp[−iω(n + 1 ) t] φn (x) 2 n! ∞ +∞ = m=0 −∞ dxo Φ(x. to ).171) can be expanded in terms of e−inωt . Noting that the propagator (2.. t|xo . one can expand (2. results in 1 ˜ exp[−iω(n + 1 ) t] φn (x) 2 n! +∞ = −∞ dxo Φ(x.171) ∞ W (x. xo . to ) exp[−iω(n + 1 ) to ] φn (xo ) . t) = n=0 1 ˜ exp[ − iω(n + 1 ) t ] φn (x) 2 n! (2. xo . t) +∞ −∞ dto 1 ˜ exp[−iω(m + 1 ) to ] φm (xo ) 2 m! (2. .

Expression (2. The Hermite Polynomials ˜ The function (2. t)|2 = ˜n (x)|2 .183). through expansion (2. .196) where Hn (y) = ey 2 /2 (2. 2. Actually. according to (2.183).194) plays a central role for the Hermite polynomials since it contains. t) of the quantum mechanical harmonic oscillator ˜ ψn (x. (2. z) z 2 e−y or w(y.2. t) = exp[−iω(n + 1 ) t] Nn φn (x) . in a ‘nutshell’ all information on the Hermite polynomials.190) ˜ is obeyed.192) w(y. z) = n=0 1 2 /2 1 2 /2 = = mω e−iωt x (2. t) = z 2 e−y where w(y.5: Propagator for a Quadratic Lagrangian 37 ˜ ˜ the functions ψn (x. t) such that +∞ −∞ 2 dx |ψ(x. .183) reads then w(y.197) The expansion coefficients Hn (y) in (2.183).189) ˜ dx φ2 (x) = 1 n (2. This follows from ∂n w(y. For this purpose we introduce first closed expressions for φ the new variables y z and write (2.194). t) provide all stationary states of the quantum mechanical harmonic oscillator. . t)|2 = Nn +∞ −∞ n = 0. 2 Here Nn are constants which normalize ψn (x. ˜ |φ stationary wave functions ψn (x.198) ∂z n z=0 .171) W (x. In the following we will characterize the functions φn (x) and determine the normalization constants Nn . z) (2. t) are associated with a time-independent probablity density |ψn (x. We will also argue that the functions ψn (x. 1.194) = z2 n=0 ∞ 1 ∞ zn ˜ φn (y) n! (2. z) = Hn (y) (2.197) are called Hermite polynomials which are polynomials of degree n which will be evaluated below. z) = exp(2yz − z 2 ) . Expansion (2.171). To obtain ˜n (x) we simplify the expansion (2. we have identified then. characterizes the wave functions φn (x). through the expansion coefficients φn (x) in (2.193) (2.191) (2. (2.195) zn Hn (y) n! ˜ φn (y) .

∂y n (2..202) generatingfunctionology by H.Wilf (Academic Press.198) for the Hermite polynomials can be expressed in a more convenient form employing definition (2. Reading.194. . closed expressions for Hn (y).200) yields for the first Hermite polynomials H0 (y) = 1.E. z) the generating function for the Hermite polynomials. 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. D.L. and O. The identity (2. . normalization factors for φ(y).201) ν−µ     µ µ ν−µ Figure 2.200) One can deduce from this expression the polynomial character of Hn (y). We want to derive now explicit expressions for the Hermite polynomials. m lattice (right diagram). For this purpose we expand the generating function (2. Massachusetts. As will become evident in the present case generating functions provide an extremely elegant access to the special functions of Mathematical Physics3 .196) ∂n w(y. .Knuth.196) to ˜ derive. ν+µ   (2. ν+µ ν   H3 (y) = 8y 3 − 12y. Inc. 1990) is a useful introduction to this tool as is a chapter in the eminently useful Concrete Mathematics by R.196).Graham. µ lattice (left diagram) corresponds to a summation over only every other point in an n.199) Comparision with (2. that Hn (y) is a polynomial of degree n.S.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).e. 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. Boston. 2. among other properties. ν     H2 (y) = 4y 2 − 2..196) results in the so-called Rodrigues formula for the Hermite polynomials Hn (y) = (−1)n ey 2 ∂ n −y2 e . 1989). H1 (y) = 2y. and recursion equations for the efficient evaluation of Hn (y). One calls w(y.1: Schematic representation of change of summation variables ν and µ to n = ν + µ and m = ν −µ.Patashnik (AddisonWesley. (2. We will employ (2. The diagrams illustrate that a summation over all points in a ν.38 Quantum Mechanical Path Integral which is a direct consequence of (2. i. The diagrams also identify the areas over which the summation is to be carried out. .

z) = w(y.. −z) and.204) Since ν. according to 2.197) ∞ n=0 zn Hn (−y) = n! ∞ n=0 (−z)n Hn (y) = n! ∞ n=0 zn (−1)n Hn (y) n! (2.207) This expression yields for the first four Hermite polynomials H0 (y) = 1. H1 (y) = 2y. k! (n − 2k)! = Hn (y) (2. 2 µ = n−m .1. in fact.e.2. With this restriction in mind one can express (2.201).209) This property follows also from the generating function. .206) From this expansion we can identify Hn (y) [n/2] Hn (y) = k=0 (−1)k n! (2y)n−2k .209).203) The old summation variables ν. According to (2.5: Propagator for a Quadratic Lagrangian and introduce now new summation variables n = ν + µ.207) one can deduce that Hn (y). m must be restricted such that either both n and m are even or both n and m are odd. The lattices representing the summation terms are shown in Fig. 0 ≤ k ≤ [n/2] where [x] denotes the largest integer p.194) holds w(−y. otherwise. hence. m = ν − µ 0 ≤ n < ∞. .210) from which one can conclude the property (2. H3 (y) = 8y 3 − 12y. H2 (y) = 4y 2 − 2. i. 2 (2.207) contains only even powers. 0 ≤ m ≤ n . . From (2. Hence.208) which agrees with the expressions in (2. is a polynomial of degree n. 39 (2. k! (n − 2k)! (2. One can write then using m = n − 2k e 2yz − z 2 ∞ = n=0 zn n! [n/2] k=0 n! (−1)k (2y)n−2k . (2. it contains only odd powers. µ expressend in terms of n. . (2. it holds Hn (−y) = (−1)n Hn (y) . In case of even n . the sum in (2. p ≤ x.205) Since (n − m)/2 is an integer we can introduce now the summation variable k = (n − m)/2 . 2. for odd n. m are ν = n+m . (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.

196) into the differential equation (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) . Starting point of the derivation is the property of w(y. (2. (2. 2.188. z) should be equivalent to the relationship (??).215) ∂z which can be readily verified using (2. . n! (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. z) − (2y − 2z) w(y. as defined through (2.209) since odd functions have a node at the origin. 2.233) below. z) ∂ w(y.40 Quantum Mechanical Path Integral The generating function allows one to determine the values of Hn (y) at y = 0. dy n = 1. dw/dy − 2zw = 0.194). 2.213) which allow one to evaluate Hn (y) from H0 (y) and H1 (y) given by (2. 2. One can then readily state which differential equation of w(y. . indeed satisfies the latter relationship. a property which is consistent with (2. have a node at y = 0. n = 1. Another relationship is d Hn (y) = 2n Hn−1 (y). For this purpose one considers w(0. . as given in (2.217) gives H1 (y) − 2y H0 (y) = 0. Hn+1 (y) − 2y Hn (y) + 2n Hn−1 (y) = 0 . n = 1. z). 1.211) m! n! m=0 n=0 Comparing terms on both sides of the equation yields (2n)! . a factor z reduces the order of Hn by one and introduces also a factor n. . . .214) We want to derive (??) using the generating function.208).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. Substituting expansion (2. .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 . Hn+1 (y) − 2y Hn (y) + 2n Hn−1 (y) = 0. namely.197) above and given by (2. . for example. . z) = 0 (2. H2n+1 (0) = 0 . n = 0. . 2.212) n! This implies that stationary states of the harmonic oscillator φ2n+1 (x). (2. The reader may verify that w(y. (2. (2. .194). For Hermite polynomials holds. . H2n (0) = (−1)n Recursion Relationships A useful set of properties for special functions are the so-called recursion relationships.

acording to the definition (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. finally. 2.226a).s. 2 n = 0.2. 2 (2.247) for a = i one obtains I(y) = and. 2 (2. 2.196) yields ∞ n. 2 (2. z) w(y.228) . For this purpose we consider the integral +∞ −∞ dy w(y. 1. . (2.h.223) dn 2iy t − t2 2 e = (2 i t)n e2iy t − t n dy results. of the Rodrigues formula (2.222) Employing this expression now on the r.219) which can be written I(y) = e−y 2 +∞ −∞ dte−(t−iy) = e−y √ 2 2 +∞ −∞ dze−z .200) yields Hn (y) = The identity (−1)n y2 √ e π +∞ dt −∞ dn 2iyt − t2 e .220) Using (2. 2.226) π −∞ 2n z n z n n=0 n! Expressing the l. in the integral representation of the Hermite polynomials 2n (−i)n ey √ Hn (y) = π Orthonormality Properties 2 (2.224) +∞ −∞ dt tn e2iy t − t . 2 √ π e2 z z (2.194. .221) 1 = √ π +∞ −∞ dt e2iyt − t . dy n (2.194.s. z ) e−y 2 +∞ = e2 z z = √ ∞ dy e−(y−z−z ) = . e−y 2 π e−y 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 . through a double series over Hermite polynomials using (2.225) We want to derive from the generating function (2.196) the orthogonality properties of the Hermite polynomials.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 .h.

According to (2. negative for n = 2 and vanishes for n = 1. that the wave functions are even for n = 0. (2. in harmony with (??). 2. One can also recognize that n is equal to the number of nodes of the wave function.233) . 1. 4.189. 4 and odd for n = 1. One can recognize. 3.190)] φn (y) = Nn e−y 2 /2 and for the normalization constants Nn follows from (2. 1.233) are presented for n = 0. 4 in Fig. ¡ ¢ ¡ -4   -2 0 2 4 y (2. 2. (2.197) holds 2 ˜ φn (y) = e−y /2 Hn (y) .232) Hn (y) .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. 4. (2.f.231) √ π (2.2: Stationary states φn (y) of the harmonic oscillator for n = 0. 2. the value of the wave function at y = 0 is positive for n = 0. 3.2.229) Hn (y) . Furthermore. 2. Normalized Stationary States The orthonormality conditions (2.228) +∞ 2 Nn −∞ √ 2 2 2 dy e−y Hn (y) = Nn 2n n! π = 1 1 2n n! 1 2n n! We conclude Nn = and can finally state the explicit form of the normalized stationary states φn (y) = −y √ e π 2 /2 The stationary states (2. 3.230) (2. 2. 3.228) allow us to construct normalized stationary states of the harmonic oscillator. in agreement with our above discussions. The normalized states are [c.

denoted by P (x) and introduced in Sect.e. and In = 2n n! π. The Hermite polynomials are so-called orthonogal polynomials since they obey the conditions (2. denoted by Ln (x) and encountered in case of the stationary states of the nonrelativistic [see Sect. i.238) where the space is endowed with the scalar product (f |g) = Ω dx f (x) g(x) w(x) = < ∞ . ??? and eq.189) by the Jacobian dx/dy.231) of the wave functions differs from that postulated in (2. In case of the Legendre polynomials. 5 below [c. In case of the associated Laguerre (α) polynomials. Comparision with (2.236) where w(x) is a so-called weight function with the property w(x) ≥ 0.239) . of functions which obey dx f 2 (x) w(x) = < ∞ . (2.233) and (2. Other examples of orthogonal polynomials are the Legendre and Jacobi polynomials which arise in solving three-dimensional stationary Schr¨dinger equations. The latter are written for polynomials pn (x) in the general form dx pn (x) pm (x) w(x) = In δnm Ω (2. (2.10 and eq. g ∈ F . 5.150 . 10. w(x) ≡ 1. holds Ω = [0.237) and where In denotes some constants. w(x) = √ exp(−x2 ). by dx = dy mω 1 4 . 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.2. (2. ???] and relativistic [see Sect. w(x) = xα e−x .156.228).179] holds Ω = [−1. using (2. f.f. In = Γ(n+α+1)/n! where Γ(z) is the so-called Gamma function. (10.e. 5.237) for Ω = R. w(x) = 0 only at a discrete set of points xk ∈ Ω (2.189) φn (x) = √ 1 2n n! mω π 1 4 e− mωx2 2 Hn ( mω x) .235) Completeness of the Hermite Polynomials The Hermite polynomials are the first members of a large class of special functions which one encounters in the course of describing stationary quantum states for various potentials and in spaces of different dimensions. +∞[.5: Propagator for a Quadratic Lagrangian 43 The normalization condition (2. 2.236 . i. 1]. 5.234) The explicit form of the stationary states of the harmonic oscillator in terms of the position variable x is then..151. 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. The orthogonal polynomials pn mentioned above have the important property that they form a complete basis in the space F of normalizable functions.459] hydrogen atom. (5..228) shows that the orthonogality condition of the Hermite polynomials is in complience with (2. and I = 2/(2 + 1). Ω (2.

i.44 As a result holds for any f ∈ F f (x) = n Quantum Mechanical Path Integral cn pn (x) (2. i. Since the Hermite polynomials form a complete basis for such states we can expand ∞ ψ(y.233) exhaust all stationary states of the harmonic oscillator. Therefore. ∞ n. except for a single n = no .244) For the state to be stationary |ψ(x. t) which is stationary under the action of the harmonic oscillator propagator (2. .172) holds.. the states (2.188.197) it must hold cn (t) = dn exp[−iω(n + 1 )t] and. hence.147). The only possibility for this to be true is dn = 0.246) for det(a) = 0 and real. If one replaces for all f ∈ F: f (x) → w(x) f (x) and. (2.245) must be time-independent.e. in particular.e.236). ψ(y. a (2. a state for which (2.243) To be consistent with(2. the 2 stationary state ψ(y.242) Let us assume now the case of a function space governed by the norm (2. −∞ dyn ei −∞ n j. n 2 (2.m=0 d∗ dm exp[iω(m − n)t] e−y Hn (y)Hm (y) . . pn (x) → w(x) pn (x) the scalar product (2.242) and the existence of a normalizable state ψ(y. Ω (2.247) . 2. Appendix: Exponential Integral We want to prove +∞ +∞ I = dy1 .241) The latter identity follows from (2.239) becomes the conventional scalar product of quantum mechanics f |g = Ω dx f (x) g(x) . (2. i. symmetric a. t)|2 .e. aT = a. 2 (2. t) must be identical to one of the stationary states (2.e.. det(a) (2.240) where cn = 1 In dx w(x) f (x) pn (x) . i. t) is ∞ ψ(y.. t) = n=0 cn (t) e−y 2 /2 Hn (y) . t) = n=0 1 2 dn exp[−iω(n + )t] e−y /2 Hn (y) .k yj ajk yk = (iπ)n . In case of n = 1 this reads +∞ −∞ dx ei a x = 2 iπ .233).

246) exploits that for any real.253) as well as det(˜) = det(S−1 aS) = det(S−1 ) det(a) . (2. . . 0 0 . . .249) n ajk = ˜ l.251) where. .248.255) .k yj ajk yk ˜˜ ˜ (2.252) ˜ For the determinant of a holds det(˜) = a n ajj ˜ j=1 (2.   .246) reads then in terms of yj ˜ n n j.250) The bilinear form in (2. ˜ (2. 0 ˜  0 a22 .248) .m (S T )jl alm Smk . The proof of (2.k n m n y Sj ajk Skm ym ˜ ˜ y (S T ) j ajk Skm ym ˜ ˜ j. .  . symmetric matrix exists a similarity transformation such that   a11 0 . .k n n yj = ˜ k (S −1 )jk yk .254) ajj . . One can conclude det(a) = j=1 n (2.. .. yk = k Skj yj .k yj ajk yk n = j. . .249) The akk are the eigenvalues of a and are real. according to (2. . i. 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 .e.7: Appendix / Exponential Integral 45 which holds for a ∈ C as long as a = 0. ann ˜ where S can be chosen as an orthonormal transformation. ˜ (2. (2. 2. (2. .2. det(S) a = (det(S))−1 det(a) det(S) = det(a) .k n m = = j. 0  ˜   ˜ S−1 a S = a =  . .

. . .263) (2. hence.257) Substitution of the integration variables (2.. . Accordingly. yn ) y ˜ = ( det S )2 (2. .250) holds J = S and.e. holds n Quantum Mechanical Path Integral ajj = 0 ˜ j=1 (2. .46 We have assumed det(a) = 0. ..262) (2. . . .258) +∞ +∞ ∂(y1 . .256) such that none of the eigenvalues of a vanishes. ∂(˜1 . . yn ) y ˜ ei n k akk yk ˜ ˜2 . . 2.250) allows one to express (2.261) ∂(y1 . .266) . . . det( From (2. ..260) (2. yn ) y ˜ ∂yj . .250) +∞ +∞ I = d˜1 . . . yn ) ∂(˜1 . .264) I = d˜1 . . d˜n e y −∞ = (2.249) follows 1 = det ST S such that one can conclude det S = ±1 One can right then (2. . ∂ ys ˜ (2. n (2.259) (2. . . ajj = 0 . 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 −∞ . . . (2. ˜ for j = 1.265) k=1 −∞ which leaves us to determine integrals of the type +∞ dx eicx −∞ 2 (2. yn ) ) = det(S) . yn ) ∂(˜1 . . . . .258) where we introduced the Jacobian matrix J = with elements Jjs = According to (2. i. y −∞ d˜n det y −∞ ∂(y1 . .

One can relate integral (2.268) can be written as the sum of the following path integrals J = J1 + J2 + J3 + J4 .2.. We consider first the case c > 0 and discuss the case c < 0 further below.271) . Jk = γk dz eicz 2 (2.e.267) by considering the contour integral J = γ dz eicz = 0 2 (2.3. p ∈ IR.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. The contour integral (2.268) along the path γ = γ1 + γ2 + γ3 + γ4 displayed in Figure 2.7: Appendix / Exponential Integral 47 where. 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 .270) dx e−cx 2 2 − 2p √ γ4 : z = ix − p J4 = −p i dx eic(ix−p) . the integrand does not exhibit any singularities anywhere in C. The contour intergral (2. c (2. (2. 2 since eicz is a holomorphic function. for x. according to (2.257) holds c = 0.266) to the well-known Gaussian integral +∞ dx e−cx −∞ 2 = π .c > 0 .268) vanishes. i.

48 Quantum Mechanical Path Integral Im(z) ip γ2 ¡ γ4 ¡ γ3 ¡ -i p Figure 2. c   ¡ -p γ1 p Re(z) 2 | |e−2cxp | . p p→+∞ lim |J2 or 4 | ≤ = p→+∞ 0 lim dx |e−2cxp | 1 − e−2cp 2cp = 0.3: Contour path γ in the complex plain. Hence.273) p→+∞ lim J2 and J4 do not contribute then to integral (2. (2. 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.268) for p = +∞.275) .274) (2. We will now show that the two integrals J2 and J4 vanish for p → +∞.272) = 0 . One can state accordingly ∞ J = dx e −∞ icx2 − √ ∞ i dx e−cx −∞ 2 Using 2.267) one has shown then ∞ dx eicx −∞ 2 = iπ . (2. (2.

if one chooses the same contour integral as (2. It holds n I = k=1 iπ = akk ˜ (iπ)n .7: Appendix / Exponential Integral 49 One can derive the same result for c < 0. (iπ)n det(a) (2.276) and (c < 0) ∞ dx eicx −∞ 2 = −iπ = −|c| iπ .277) We apply the above results (2. This leads to ∞ icx2 J = dx e −∞ + √ −∞ −i ∞ dx ecx = 0 2 (2.268).277) to (2. c (2. 2.255) this result can be expressed in terms of the matrix a I = which concludes our proof. but with a path γ that is reflected at the real axis.278) Noting (2.265). n ˜ j=1 ajj (2.2.279) .275.

50 Quantum Mechanical Path Integral .

(3. t) ψ(r + s.1 Derivation of the Schr¨dinger Equation o We will consider now the propagation of a wave function ψ(r. of this equation in terms of powers of and we will demonstrate that the terms of order require that ψ(r. t) .4) − U (r. x2 . the Schr¨dinger o o equation provides the simpler avenue towards describing quantum systems than the path ingral formulation of Section 2. t) satisfies a partial differential equation.21).5) ψ(r. (3.1) can be expressed through the discretization scheme (2. t) ψ(r0 .20) to R one obtains then for small φ(r.s.h. 2. t) . x3 51 . t) .20) to evaluate the propagator. For many situations. t + |r0 . t + ) = × exp i 2 +∞ +∞ +∞ −∞ dx1 −∞ dx2 −∞ dx3 m x2 + x2 + x2 1 2 3 m 2πi 3 2 × (3.s. x2 . but by no means all.1) We will expand the l. . In the limit of very small it is sufficient to employ a single discretization interval in (2. x3 )T . namely the Schr¨dinger equation. It holds then according to (2. t + ) = Ω d3 r0 φ(r.2) d 3 r0 m 2πi 3 2 exp i m (r − r0 )2 − 2 U (r. and the r. t) ψ(r0 .h. t + |r0 .20.3) In order to carry out the integration we set r0 = r + s and use s as the new integration variable. t + ) = Ω m 2πi 3 2 exp i m (r − r0 )2 − 2 U (r. (3. t) by an infinitesimal time step . even in x1 . Generalizing (2. This yields ψ(r. t) = From this follows ψ(r.Chapter 3 The Schr¨dinger Equation o 3. The propagator in (3. Notable exceptions are non-stationary systems involving time-dependent linear and quadratic Lagrangians. We will denote the components of s by (x1 . t) .

It holds In+1 (a) = I1 (a) = i 2a iπ ..k=1 ∂2 ψ(r.4) one identifies 2 1 = a m = O( ) (3.g. 2 (3.5) assuming that only the leading terms contribute. a I2 (a) = − 3 4a2 iπ .4) is even in all three coordinates x1 . we need then to evaluate integrals of the type +∞ In (a) = −∞ dx x2n exp i a x2 . According to (3.. x2 . e. of the expansion of ψ(r + s. x3 )T ..36) holds I0 (a) = Inspection of (3. only terms of the expansion (3.. It is then sufficient to consider the terms ψ(r. . t) j 1 12 . 1.9) iπ a (3. t) j k . 1 2 3 2 ∂2 j=1 xj ∂x2 ψ(r. t) .8) one can evaluate recursively all integrals In (a). t) + j=1 xj ∂ 1 ψ(r.12) Here one needs to note that we are actually dealing with a three-fold integral. t). The integration involves only the wave function ψ(r +s.52 Schr¨dinger Equation o It is important to note that the integration is not over r. O( 5 2 ) .6) 3 4 ∂4 j=1 xj ∂x4 ψ(r. .k=1 xj xk ∂x2 ∂x2 ψ(r.7) shows 1 ∂ In (a). t) is a constant with respect to this integration..11) holds 3 m 3 iπ 2 2 × = 1 (3. n = 0. Obviously.13) 2πi a . Since the kinetic energy contribution in (3.8) (3. t) + ∂xj 2 3 xj xk j. ∂xj ∂xk (3. t) = ψ(r.. 1 4 3 2 2 ∂4 j.6) make contributions of the order O( 3 2 ) . U (r.11) and. a (3.5) which are even separately in all three coordinates yield non-vanishing contributions. Since the latter contributes to (3. t) and the kinetic energy term. consequently. (3.7) According to (2.4) only for small x2 + x2 + x2 values 1 2 3 we expand 3 ψ(r + s. but over s = (x1 .10) It is now important to note that in case of integral (3. the terms collected in (3. t) j . a supposition which will be examined below. x2 . (3. x3 . i ∂a Starting from (3. O( 7 2 ). O( 7 2 ) . t) + .

17) 2m ψ(r. t + ) = exp − i U (r.19) ∂t where 2 2 ˆ H = − + U (r. t) (3.10). one specifies the so-called initial condition. t) + ∂ ∂t ψ(r.2 Boundary Conditions The time-dependent Schr¨dinger equation is a partial differential equation.g. The 1st order time derivative requires that for any solution a single temporal condition needs to be specified. ψ(r. t) . ψ(r. t1 ) = f (r). The 2nd order spatial derivatives require that one specifies also properties of the solution on a closed boundary ∂Ω surrounding the volume Ω in which a solution is to be determined. Obviously.14) results in ψ(r. t) = ψ(r.. t). t) + and exp − i U (r. As we will discuss in Chapter 5 below the solutions of the Schr¨dinger equation are restricted to particular Hilbert spaces H which are o . t) = ∂t 2 − 2 2m + U (r. We will derive briefly the type of boundary conditions encountered.15) ψ(r. using (3. t) ψ(r. t) + 1 2i 4 m 2 53 ψ(r. t) + O( 2 ) . In order O( ) the equation reads i ∂ ψ(r.14) This expansion in terms of powers of suggests that we also expand ∂ ψ(r. (3. (3. t) = 1 − (3. t) (3. The following general remarks can be made about the solution. e.. a solution is thought for t ≥ t0 and the solution is specified at the intial time t0 . t)ψ(r.20) 2m 3. t) + O( 2 ) .16) Inserting this into (3.3. t) + O( 2 ) ∂t i U (r. 1st order in time.2: Boundary Conditions and one can conclude. Usually. (3. 2nd o order in the spatial variables and linear in the solution ψ(r.e. t + ) = ψ(r. Due to its linear character any linear combination of solutions of the time-dependent Schr¨dinger o equation is also a solution. This equation is often written in the o form ∂ ˆ i ψ(r.18) This is the celebrated time-dependent Schr¨dinger equation. t) . (3. i. t) − + i 2 2 i U (r. this equation is trivially satisfied to order O( 0 ). t) + O( 2 ) . t) = H ψ(r. t) ψ(r.

The hermitian property. 2nd Ed. Calssical Electrodynamics. D.22) and (3. (John Wiley.27) ˆ We will postulate below that H is an operator in H which represents energy. in principle. differ from each other. (3.22) ˆ where H is defined in (3.23) ˆ Since H is a differential operator the expressions (3. hence. 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.54 Schr¨dinger Equation o linear vector spaces of functions f (r) in which a scalar product between two elements f. .21) Ω This leads one to consider the integral f | H |g Ω = Ω ˆ d3 rf ∗ (r) H g(r) (3. Interchanging f ∗ (r) and g(r) results in g| H |f Ω = Ω ˆ d3 r g(r) Hf ∗ (r) . In (3. 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.28) See.20). that H is 2 . g|r) = − 2 2m ( f ∗ (r) g(r) − g(r) f ∗ (r) ) (3. New York.23). J. t). 1975). however.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. t) g(r) − Ω d3 rg(r) U (r. Since energy is a real ˆ ˆ quantity one needs to require that the eigenvalues of the operator H are real and. f ∗ (r) d3 rg(r) Ω 2 = − 2m d3 rf ∗ (r) Ω 2 g(r) − f ∗ (r) (3. Chapter 1. Jackson. g|r) is called the ˆ concomitant of H and is P (f ∗ . The reader is advised to consult a reference text on ‘Linear Algebra’ to follow this argument. g|r) (3.26) where ∂Ω da · A(r) denotes an integral over the surface ∂Ω of the volume Ω. g ∈ H is defined as follows f |g Ω = d3 rf ∗ (r)g(r) (3. implies hermitian ˆ f | H |g 1 2 Ω ˆ = g| H |f Ω (3.25) ˆ f | H |g Ω ˆ = g| H |f Ω + ∂Ω da · P (f ∗ . for example..26) the vector–valued function P (f ∗ .

i. In either case does f (r) and all of its derivatives vanish asymptotically. in which case ∂Ω1 is the inner sphere and ∂Ω2 is the outer sphere. t) which describes the state of o a particle moving in the potential U (r. t). However.e. and that the density decays rapidly when one moves away from that area. |ψ(r. namely.e.31) usually arises when a particle existing in a bound state is described. like exp(−κr). . t)|2 . Often the closed surface of a volume ∂Ω is the union of disconnected surfaces3 ∂Ωj .29. 3. In this case one can postulate both conditions (3.. obey (3. to avoid discontinuities in ψ(r. Since the total probability of finding the particle anywhere in space is d3 r|f (r)|2 = 1 (3. (3. It must hold then for all f ∈ H f (r) = 0 ∀r. therefore. α > 2. t) + ψ(r. t) is ∂t p(ω. t)|2 . 3. (3.29) In fact. .32). 3.31) |r|→∞ (3.32) the wave function must decay for |r| → ∞ rapidly enough to be square integrable.g. ∂Ω = ∂Ω1 ∩ ∂Ω2 ∩ ∂Ω3 ∩ .3: Particle Flux 55 and. t) ] . The observable directly linked to the wave function is the probability to find the particle at position r at time t. κ > 0 or like r−α . if f and g satisfy these conditions than also does any linear combination αf + βg. t) = ω d3 r |ψ(r.29. The latter property stems from the fact that the boundary condition (3. t) on a single connected surface ∂Ωj only either one of the conditions (3.3. r ∈ ∂Ω or da · f (r) = 0 ∀r. .3 Particle Flux and Schr¨dinger Equation o The solution of the Schr¨dinger equation is the wave function ψ(r. In this case one can expect that the particle density is localized in the area where the energy of the particle exceeds the potential eenrgy.30) each condition holding on an entire surface ∂Ωj .30) Note that these boundary conditions are linear in f .e. t) = ω 3 d3 r [ ψ ∗ (r.. g|r) vanish on ∂Ω. A most common boundary condition is encountered for the volume Ω = R3 in which case one postulates lim f (r) = 0 “natural boundary condition” . i. i.. in this case also all derivatives of f (r) vanish at infinity. (3. we can only allow functions which make the differential da · P (f ∗ ..33) The time derivative of p(ω. t)∂t ψ ∗ (r.30) can be postulated. r ∈ ∂Ω (3. e. The probability to observe the particle anywhere in the subvolume ω ⊂ Ω is p(ω. t)∂t ψ(r.34) An example is a volume between two concentric spheres.

t) ψ ∗ (r. t). t) = 0. Note that for a normalized wave function the quantity ρ(r.27) one can express this j(r. t) + ω · j(r.42) 2mi Since (3. t)r. (3. t) ] . One can multiply the solution of (3. t) − ψ ∗ (r.19) and its conjugate complex4 −i yields i ∂t p(ω.41) holds for any volume ω ⊂ Ω one can conclude ∂t ρ(r.27) this can be written i ∂t p(ω. Using (3. t) ∂t ˆ ˆ ψ ∗ (r. t) − ψ(r. t) = H ψ ∗ (r. t)|2 is a probability density with units 1/volume. (3.43) (3. ψ|r. t) + 4 · j(r.29. (3.41) where j(r.37) If one applies this identity to ω = Ω one obtains according to (3. ψ(r.56 Using (3.35) d3 r . t) = 0 . 3. t) H ψ ∗ (r. t) = ∂ω da · P (ψ ∗ (r.18) by any complex number and accordingly one can define ψ(r. The surface integral (3.36) According to (3. t) = P (ψ ∗ . We will assume in the remainder of this Section that the solutions discussed are normalized. t) . A natural choice for this constant is 1. t) = |ψ(r.30) ∂t p(Ω. t) ψ(r. One refers to such solution as normalized. t) (3. .37) can be expressed through a volume integral according to da · A(r) = ∂ω ω (3. t) = 0 (3. 3.38) holds. t) = [ ψ(r.26. t) . t) such that d3 r|ψ(r. (3. t)|2 = 1 Ω (3. t) = ω Schr¨dinger Equation o ∂ ∗ ˆ ψ (r.44) ˆ Note that the Hamiltonian H involves only real terms.39) d3 r · A(r) (3.40) One can rewrite then (3. t) H ψ(r. Accordingly the probability to observe the particle anywhere in the total volume Ω is constant.37) d3 r ∂t ρ(r.

t) is that of density flux.f. In Sect. (2.. m (3. Equation (3.105. This follows directly from an inspection of Eq. t) = 0 2 ∂ 2 i ψ(r. (3. arises solely from the complex factor exp(i k · r).45) Obviously. t) = − ψ(r.167)] potential.41) written in the form ∂t ω d3 r ρ(r. 3. j(r. This finding is consistent with the present evaluation of the particle flux: a factor exp(ipo xo / ) gives rise to a flux po /m. 2. the initial condition at ψ(r. Such case arose in Sect.71)].. 2 for a free particle [c.48) ∂t 2m which describes the motion of free particles. It is of interest to note from (3.e.43) that any real wave function ψ(r.48) that the general solution is of the form ψ(r. (3. 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.f. t) has vanishing flux anywhere.125)] and in a quadratic [c. t0 ) = [2π]− 2 Ω∞ 3 ˜ d3 k φ(k) exp i(k · r − ωt0 ) . equal to the velocity of the particle. Note that j(r.49) reads at t = t0 ψ(r.4: Free Particle 57 The interpretation of j(r. i.f. t) = [2π]− 2 where the dispersion relationship holds ω = k2 . t) (3. 2. (2.47) .18) in Ω∞ = R in the case o U (r.3. t0 ) determines φ(k).148. (3. One can readily show by insertion into (3. for f (r) ∈ R. 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. 2. and for particles moving in a linear [c.4 Solution of the Free Particle Schr¨dinger Equation o We want to consider now solutions of the Schr¨dinger equation (3. t) = − ∂ω da · j(r.48.e. 2m (3.46) i. t) gives rise to a descrease of the total probability in volume ω due to the disappearence of probability density at the surface ∂ω.49) ˜ Obviously. t) . t) points in the direction to the outside of volume ω.51) . (2. 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) .50) 3 ˜ d3 k φ(k) exp i(k · r − ωt) Ω∞ (3. (3.

e. t) = Ω∞ d3 r0 φ(r.247). t0 ) = 1 2π 3 d3 k exp Ω∞ ik · (r − r0 ) − i 2 k2 2m (t − t0 ) . The solution as stated is defined in the infinte space Ω∞ = R . In fact. hence. t|r0 .58 The inverse Fourier transform yields 3 ˜ φ(k) = [2π]− 2 Schr¨dinger Equation o d3 r0 exp(−ik · r0 ) ψ(r0 . Comparision with Path Integral Formulation One can write solution (3. that the “natural boundary condition” (3.114) and (4. The ensuing solutions are called wave packets. (3. The general form for this propagator involves an expansion in terms of a complete set of eigenfunctions as in (3.5).54) This expression obviously has the same form as postulated in the path integral formulation of Quantum Mechanics introduced above.49) is square integrable at all subsequent times and.e. In case of the harmonoc oscillator the expansion can be stated in a closed form given in (4. 3.54) valid for this case.81) . t|r0 .52) above ψ(r.49). We have identified then with (3. t0 ) (3.49. t) as given by (3.52) We have not specified the spatial boundary condition in case of (3.70) derived below for a particle in a box and the harmonic oscillator.47). t0 ) . To show this one needs to note 1 2π = +∞ dk1 exp −∞ ik1 (x − x0 ) − 1 2 i 2 k2 1 2m . t0 ) ψ(r0 . (t − t0 ) (3. Ω∞ (3. evaluating the integral in (3.53) where φ(r.31) applies. respectively.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. Below we will generalize the propagator (3.51) to be square integrable it follows according to the properties of the Fourier–transform that ψ(r.54) yields (2.47). 3.56) and using (2.54) an alternative representation of the propagator (2. derive an expression similar to (3. i. If one chooses the initial state f (r) defined in (3.54) to the case of non-vanishing potentials U (r).51... i. in (2.

(3. t) = −∞ dx0 φ(x. We have demonstrated then in this case that the Schr¨dinger formulation of Quantum o Mechanics is equivalent to the Feynman path integral formulation. a result which is identical to the expression (??) obtained by means of the path integral propagator (2. t) = f (t) φ(r) .59) which is solved through completion of the square in the exponent [c. r ∈ ∂Ω or da · φ(r) = 0 ∀r. 3. 3. 3.56)].49. i. r ∈ ∂Ω (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. Stationary States We consider now solutions of the time-dependent Schr¨dinger equation (3. (3.54) is +∞ ψ(x. t|x0 . the functions obey boundary conditions of the type (3.f. t0 ) = 1 2π +∞ dk exp −∞ ik (x − x0 ) − i 2m (t − t0 ) .20) which are of o the form ψ(r.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..60) yields with t0 = 0 ψ(x.62) We will restrict the space of allowed solutions to a volume Ω such that the functions also make the concomitant (3. Accordingly. t|x0 .55.46).61) × exp − (x − po m t)2 2δ 2 (1 + 2 t2 ) m2 δ 4 (1 − i .4: Free Particle Free Particle at Rest 59 We want to apply solution (3.3.57) where φ(x. the boundary conditions are φ(r) = 0 ∀r. t0 ) ψ(x0 .29.27) vanish on the surface ∂Ω of Ω.53.63) . t) is given by (??).19.64) (3. (3. 3. t) 2 k2 (3.57–3. t) = t 1 − i mδ2 t 1 + i mδ2 1 4 1 πδ 2 (1 + 2 t2 m2 δ 4 1 4 ) × t po i p2 o ) + i x − t mδ 2 2m (3.60) =???? Combining (3. (3.30).e.52) to the case that the initial state of a 1-dimensional free particle ψ(x0 . The 1-dimensional version of (3.

68) ˆ where g1 (t) = i ∂t f (t). according to (3. One calls such states stationary states.e. n = 1. It is important to realize that the separable solutions (3.63. It follows then from (3. 2. t) = Ω Ω d3 r |ψ(r. for the eigenvalue problems in the confined function space. One can conclude that the probability density for the state (3.e.. and h2 (r) = H φ(r). the solutions o (3.71) .27) holds and. In order to further characterize the solution (3. solutions exist only for a set of discrete E ˆ values.62) exists. We will demonstrate that solutions of the type (3. We have then shown that ψ(r.e. g2 (t) = f (t).62) we insert it into (3. It turns out that a solution for f (t) can be found for any E.62) which consist of two factors. 3.62) do exist and we will characterize the two factors of the solution f (t) and φ(r).65) is constant. often only for a discrete set of values En .64). As pointed out above. We will encounter many such problems in the subsequent Sections. t)|2 = |φ(r)|2 . t)|2 = |f (t)|2 Ω d3 r |φ(r)|2 (3.66) i. . The identity (3.70) The task of finding solutions φ(r) which solve (3. h1 (r) = φ(r).62) are special solutions of the timedependent Schr¨dinger equation. We denote the corresponding solution by φE (r). At this point we state without proof that.. We must postulate therefore ∂t f (t) ˆ H φ(r) = E f (t) = E φ(r) .62) is |ψ(r. one factor depending only on the time variable and the other only on the space variables are called separable in space and time. the eigenvalues of the operator H. hence. We may note in passing that solutions of the type (3. It follows from the observation that for the solution space considered (3.69) If these two equations can be solved simultaneously a solution of the type (3. in general. t) vanishes on the surface of ∂Ω. is time-independent. This yields an expression g1 (t) h1 (r) = g2 (t) h2 (r) (3. t) = f (0) exp − i E φE (r) where ˆ H φE (r) = E φE (r) . a common case is Ω = Ω∞ and the ‘natural boundary condition’ (3.62) have the particular property that the associated probability distributions are independent of time.60 Schr¨dinger Equation o and affect only the spatial wave function φ(r).69) is called an eigenvalue problem. In fact. i. if f (t) = eiα .42) the flux j(r.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.45) that the total probability d3 r ρ(r. α ∈ R . (3. namely f (t) = f (0) exp − i Et . (3. by no means all solutions are of this type. . .19). We want to demonstrate this property now. i. At this point we will accept that solutions φ(r) of the type (3. (3. for functions required to obey boundary conditions (3. however.31). This can hold only if |f (t)| is time-independent.67) (3..69) exist. (3.

74) The classical free particle with constant energy E > 0 moves without bounds in the space Ω∞ . a property which is to be expected since the energy is purely kinetic energy which. We start our proof using the property (3. k ∈ R 2 k2 φk (r) = N exp i k · r .20).3. Obviously. should be positive.75) 2m One can ascertain this statement by inserting the expression for φk (r) into the eigenvalue problem posed in (3.g.74) using exp i k · r = ik exp i k · r . t) = exp − i 2 k2 2m t exp i k · r (3. m (3.73) from which follows E = E ∗ and.71) are.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 . The solution φE (r) corresponding to the eigenvalue problem posed by (3. particles scattered of a potential.28) for the special case that f and g in (3. − 2 2m φ(r) = E φ(r) .e. The resulting total energy values E are positive. As a result we cannot postulate in the present case that wave functions are localized and normalizable. E ∈ R.28) both represent the state φE (r). in fact. i.43) is j(r. t) = exp − E t 2 φE (r) .. (3. We will wave this assumption as we always need to do later whenever we deal with unbound particles.74) is actually best labelled by an index k. real. 3.76) has kinetic energy 2 k 2 /2m.76) according to (3. ˆ d3 r φ∗ (r) H φE (r) = E Ω Ω ˆ d3 r φE (r) H φ∗ (r) . We will show in Section 5 that E can be interprerted as the total energy of a stationary state. one can interpret then k as the magnitude of the momentum of the particle. . E (3. (3. o According to (3.4: Free Particle 61 is a solution of the time-dependent Schr¨dinger equation (3. The corresponding stationary solution ψ(r. hence. The flux corresponding to (3. We want to prove now that the eigenvalues E which arise in the eigenvalue problem (3. Stationary State of a Free Particle We consider now the stationary state of a free particle described by i ψ(r.66) E must be real. e. E (3.72) According to (3.71) this yields E Ω d3 r φ∗ (r) φE (r) = E ∗ E Ω d3 r φE (r) φ∗ (r) . t) = |N |2 k . of course.19. = E.

82) have the property that either both boundary conditions are satisfied or none. (o) 2 k2 (e) 2 k2 2m = E (3. . therefore.62 Schr¨dinger Equation o 3. The boundary conditions which according to (3.83) The solutions (3. It turns out that this boundary condition can only be satisfied for a discrete set of k–values kn . we have to consider only one boundary condition.o) (a) = 0 and φE (e.80). n ∈ N. . We can expect.81. t) = exp(−iEt/ )φE (x) where φE (x) is determined by 2 d2 − φE (x) = E φE (x) . a] ⊂ R which vanish on the surface ∂Ω1 = {−a. 2m dx2 2 (± ) = } (3. (3.81) 2m = E. f continuous. 2a n = 1. 5 . .80) need to be satisfied are φE (e.82) satisfy the differential equation in (3.78) (3.82) One can readily verify that (3. so-called even solutions obeying φ(x) = φ(−x) φE (x) = A coskx .5 Particle in One-Dimensional Box As an example of a situation in which only bound states exist in a quantum system we consider the stationary states of a particle confined to a one-dimensional interval [−a. a] ⊂ R → R.79) The stationary solutions have the form ψ(x. i. hence. Hence. f ∈ F1 = {f : [−a.80) 2m dx2 We note that the box is symmetric with respect to the origin. and so-called odd solutions obeying φ(x) = −φ(−x) φE (x) = A sinkx. that the solutions obey this symmetry as well. We assume.o) (−a) = 0 (3. 3..81) they are kn = nπ .84) . In case of the even solutions (3. t) . a}. t) = − d2 ψ(x. a] ⊂ R assuming that the potential outside of this interval is infinite. (3. 3. two types of solutions.e. Solutions of the Schr¨dinger Equation in F1 o The time–dependent solutions satisfy i ∂t ψ(x.81. 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. let say the one at x = a. φ(±a) = 0 . (3. We will refer to this as a particle in a one-dimensional ‘box’.

94) . respectively. By counting the number of their nodes one can determine the energy ordering of the wave functions. according to the dispersion relationships given in (3.o) (a.5: Particle in One-Dimensional Box since for such kn 63 nπa nπ = cos 2a 2 In case of the odd solutions (3. 6 . n is assumed to be even. are En = 2π2 8ma2 n2 . (3. = 0. 5 . 4. (3.91) holds.87) (Note that. (3. 3 . (3.90) 2a The wave functions represent stationary states of the particle in a one-dimensional box. x)|2 = 1 n (3. . 4.82). 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 .3. x) = An cos n and φ(o) (a. 6 . 5 . .86). 2a n = 1. .89) nπx . 3. 6 . 2a 1 . n = 2. 4. i. . This condition implies for the even states +a |An |2 −a dx cos2 nπx = |An |2 a = 1 . (3.93) The normalization constants are then An = (3.86) satisfy the boundary condition since for such kn nπa sin(kn a) = sin 2a = sin nπ 2 = 0.e.81) and (3.85) (3. (3. n = 1. (3. 2. 3. . The wave functions for the five lowest energies En are presented in Fig. . . .81.86). 2a n = 1. . according to (3. .84. 3.92) and for the odd states +a |An |2 −a dx sin2 nπx = |An |2 a = 1 .82) only the kn –values cos(kn a) = cos kn = nπ . (3. . a n = 2. 2a n = 2.82).. . x) = An sin n nπx . 3.1). .) The Energy Spectrum and Stationary State Wave Functions The energy values corresponding to the kn –values in (3.88) where the energies for odd (even) n–values correspond to the even (odd) solutions given in (3. φ(e) (a. It is desirable to normalize the wave functions such that +a −a dx |φ(e.

64 Schr¨dinger Equation o E h2 ] [ 8 m a2 25 ¢ 16 9 4 Figure 3. x) for n = 1.1: Eigenvalues En and eigenfunctions φn (e.o) (a. ¡   1 -a a x . 4. 5 of particle in a box. 3. 2.

} where φn (a. . x) .95) 1 a cos nπx 2a sin nπx 2a for n = 1. this integral vanishes. it holds: f |f Ω1 = 0 → f (x) ≡ 0.. . In case of n = m we have to consider three cases. m both odd.e. 5 . . g ∈ F1 (3. the integral arises φn |φm 1 a Ω1 = 1 a +a nπx −a dx cos 2a cos mπx = 2a (3. (3. Because of the property (3. 2. (i) n. m odd. Similarly.90. i.98) ∞ f |f 5 Ω1 = n=1 d2 . 4. N ≥ 1.103) We will show in Section 5 that the property of a scalar product do indeed apply. one obtains for n. the integrals vanish.102) holds according to (3. 3.97) form an orthonormal basis set. . . n (3. and (iii) the mixed case.101) +a −a dx cos (n−m)πx − cos (n+m)πx 2a 2a and.5: Particle in One-Dimensional Box The Stationary States form a Complete Orthonormal Basis of F1 We want to demonstrate now that the set of solutions (3. . m even φn |φm 1 a Ω1 = 1 a +a nπx −a dx sin 2a sin mπx = 2a (3.99) Since in this case the integrand is a product of an even and of an odd function. m both even.e.89. In particular.98) The latter property is obviously true for n = m. hence. it holds φn |φm Ω1 = δn m .96) f |g Ω1 = −a dxf (x) g(x) . (ii) n. .3. n = 1.100) +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. the integral vanishes. 3. for ∞ f (x) = n=1 dn φn (a.98) the elements of B1 must be linearly independent. (3. In case of n.94) B1 = {φn (a. n = m. for n = 2. f. too. . a] are N. In fact. 3. Obviously. The latter case leads to integrals φn |φm Ω1 = 1 a +a dx cos −a nπx mπx sin . Hence we need to consider only the first two cases. 6 .. 3. the integrand is odd. 2a 2a (3. x) (3. x) = together with the scalar product5 +a 65 (3.

.108) where the functions cn (t) are to be determined from the Schr¨dinger equation (3. any element of the function space F1 defined in (3. they can be expanded in terms of a Fourier series. . there exist real constants {an .107) Inserting this into (3. a] yields then also an expansion for any element in F1 and B1 is a complete basis for F1 . x) cn (t) −a dx0 φn (a. t0 ) in terms of eigenfunctions φn (a. n = 1. such demonstration can be based on the theory of Fourier series. Hence. 3.105) must vanish. (3.104) ˜ where [y]a = y mod2a. . . Demonstration of completeness is a formidable task. x) . x) to obtain an expression for ψ(x. x0 ) ψ(x0 . i.79) requiring the o initial condition cn (t0 ) = 1 . . This implies that only the trigonometric functions which are elements of B1 enter into the Fourier ˜ series. 2.66 Schr¨dinger Equation o f (x) ≡ 0 implies f |f Ω1 = 0 which in turn implies dn = 0 since d2 ≥ 0. 5 . t0 ) . In the present case. For this purpose we extend the definition of the elements of F1 to the whole real axis through ˜ f : R → R.110) . t) = n=1 φn (a. and bn . . x) ∂t cn (t) −a dx0 φn (a. Restricting the expansion (3. the coefficients an .105) to the interval [−a. t0 ) = n=1 dn φn (a. 1. m = 1. It follows that B1 n defined in (3. . x) cn (t) −a dx0 φn (a. (3. x0 ) ψ(x0 . . x0 ) ψ(x0 . t) at times t > t0 . 3. .106) and generalizing to t ≥ t0 one can write ∞ +a ψ(x. .} such that ∞ ∞ nπx nπx ˜ = + an sin (3.106) Using the orthonormality property (3.} and {bn . 3.105) f an cos 2a 2a n=0 n=1 ˜ The functions f corresponding to the functions in the space F1 have zeros at x = ±m a. t0 ) = dm . n = 2.78) can be expressed as a linear combination of the elements of B1 defined in (3.109) Insertion of (3. 4.. . n = 1.98) one obtains +a dx0 φm (a. ( ) = ([ + ] − ) (3. Accordingly. however. i. in (3.96).108) into the Schr¨dinger equation yields o +a ∞ n=1 φn (a. .95) is an orthonormal basis. We like to show finally that the basis (3.. We have then shown that any f corresponding to elements of F1 can be expanded in terms of elements in B1 . . 2. . For this purpose we expand ∞ ψ(x. −a (3. (3.e.95. The functions f are periodic with period 2a. Evaluating the Propagator We can now use the expansion of any initial wave function ψ(x. t0 ) (3. t0 ) = +a ∞ i n=1 − En φn (a.95) is also complete. x0 ) ψ(x0 .e. n = 0.

the elements of which satisfy the appropriate boundary conditions. Example of a Non-Stationary State As an illustration of a non-stationary state we consider a particle in an initial state ψ(x0 . rather than by an integral (3. x) exp n=1 i − En (t − t0 ) φn (a. t) for any initial condition ψ(x. (3.113). a] yields. t0 ) itself is a solution of the time-dependent Schr¨dinger o equation (3. according to (3. but the collision with the left wall leads to new interference effects. The interference pattern begins to ‘smear out’ first. t0 ) = δ(x − x0 ) as can be readily verified using (3.112) Equations (3. k0 = . t|x0 . being reflected at the right wall. (3.98).113) where φ(x. t0 ). σ = a 15 . t0 |x0 . The last frame shows the wave front reaching again the right wall and the onset of new interference. i ∂t cm (t) = − Em cm (t) . Note that the propagator has been evaluated in terms of elements of a particular function space F1 . . t0 ) is also referred to as a Greens function.111) The solutions of these equations which satisfy (3. 4 a (3. x0 ) .108. One can recognize that the particle moves first to the left and that near the right wall of the box interference effects develop. t0 ) (3. Often the propagator φ(x. t|x0 .109) are cm (t) = exp i − Em (t − t0 ) .114) This expression has the same form as postulated in the path integral formulation of Quantum Mechanics introduced above. Figure 3. The respective initial condition is φ(x.112) determine now ψ(x..e. In case that different boundary conditions hold the propagator will be different as well. t0 ) = 1 2πσ 2 1 4 exp − x2 0 + i k0 x0 2σ 2 . t0 ) ψ(x0 . We have identified then with (3. This solution can be written +a ψ(x.2 presents the probability distribution of the particle at subsequent times.78). i. The particle moves then to the left. t0 ) = ∞ φn (a.5).3. It is of interest to note that φ(x. cm (t0 ) = 1 . t|x0 .5: Particle in One-Dimensional Box Multiplying both sides by φm (a. x) and integrating over [−a. t|x0 . 3.115) 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.79) which lies in the proper function space (3.114) the representation of the propagator for a particle in a box with infinite walls. 67 (3. In the present system which is composed solely of bound states the propagator is given by a sum. t) = −a dx0 φ(x.54) as in the case of the free particle system which does not exhibit any bound states. in (2.

48 m a h2 |Ψ|2 2 t = 0.68 Schr¨dinger Equation o |Ψ|2 2 t = 0.2: Stroboscopic views of the probability distribution |ψ(x.20 m a h2 1 πσ 1 πσ -a a -a a |Ψ|2 2 t = 1.44 m a h2 |Ψ|2 2 t = 1.96 m a h2   |Ψ|2 2 t = 1.16 m a h2   1 πσ 1 πσ -a a -a a Figure 3.68 m a h2 1 πσ 1 πσ -a a -a a |Ψ|2 2 t = 1. t)|2 for a particle in a box starting in a Gaussian distribution with momentum 15 /a.00 m a h2 |Ψ|2 2 t = 0.92 m a h2 |Ψ|2 2 t = 2.24 m a h2 1 πσ k = 15 a 1 πσ -a a -a a |Ψ|2 2 t = 0. .72 m a h2 1 πσ 1 πσ -a a -a a |Ψ|2 2 t = 0.

x3 . can be counted. 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. 3 (3. .121) where − 2 2m 2 ∂j φ(j) (xj ) = Ej φ(j) (xj ) . 3.122) . 2.116) where Ω is the interior of the box and ∂Ω its surface Ω ∂Ω = [−a1 .119) where φE (x1 . ) = ∀( . i. one can express H 3 φ(x1 . The corresponding solutions have the form i ψ(x1 . f continuous. hence. ˆ H = − 2 (3. x3 )T ∈ Ω.95).e. This property allowed us to evaluate the propagator (3. (3. )T ∈ ∂ } . x3 ) = E φE (x1 . x2 ..78).116) and obeys the partial differential equation ˆ H φE (x1 . The description employed a space of functions F1 defined in (3. x2 .6 Particle in Three-Dimensional Box We consider now a particle moving in a three-dimensional rectangular box with side lengths 2a1 . a3 ] ⊂ R = {(x1 .3. Placing the origin at the center and aligning the x1 . x2 . x2 .. x3 )T ∈ Ω.117) 2m 2 2 2 ∂1 + ∂2 + ∂3 (3. x2 . . x2 . ˆ = O(x1 ) + O(x2 ) + O(x3 ). a1 ] ⊗ [−a2 . x3 ) .118) which are stationary states. j = 1. x2 . ( . of the space F1 is that the elements of the basis set can be enumerated by integer numbers. x2 . t) = H ψ(x1 . x3 .e.120) ˆ Since the Hamiltonian H is a sum of operators O(xj ) each dependent only on a single variable. We have defined in the space F1 a scalar product (3. t) = exp − E t φE (x1 . x3 ) is an element of the function space F3 defined in (3. x2 . x3 ) = j=1 φ(j) (xj ) (3. x3 ) (3. x3 = ±a3 } . A complete basis of F1 is given by the infinite set B1 (3. x2 . 2a3 . (3. x1 = ±a1 } ∪ {(x1 . φ(j) (±aj ) = 0 . We seek then solutions of the time-dependent Schr¨dinger equation o ˆ i ∂t ψ(x. x2 = ±a2 } ∪ {(x1 . t) . 2a2 .114) in terms of which the solutions for all initial conditions can be expressed. x2 .97) with respect to which the eigenfunctions are orthonormal.6: Particle in Three-Dimensional Box Summary: Particle in One-Dimensional Box 69 We like to summarize our description of the particle in the one-dimensional box from a point of view which will be elaborated further in Section 5. a2 ] ⊗ [−a3 . An important property of this basis and. i. x3 )T ∈ Ω.

n2 . . if all three lengths a1 .n3 ) (a1 . t0 ) = n1 . .} ∞ (3. a2 . . x3 –axes.126) Symmetries The three-dimensional box confining a particle allows three symmetry operations that leave the box unchanged. the solutions of (3. 3.n2 . n 3 and E(n1 . x2 .n2 .n2 . a3 . (3. n1 . n 2 .123) 8m n2 n2 n2 1 2 3 + 2 + 2 a2 a2 a3 1 .122) with (3. a2 .124) The same considerations as in the one-dimensional case allow one to show that B3 = {φ(n1 .n3 ) (a1 . 2.125) is a complete orthonormal basis of F3 and that the propagator for the three-dimensional box is φ(r.n2 .n3 ) = 2π2 = φn1 (a1 . a3 . a3 .96) and.n3 ) (a1 .n2 . 2. a2 . r0 ) . For example. a3 are identical. (3. i. n3 = 1.122) are given by (3. x3 ) . x3 ) n1 . x2 .e.. a2 . in the present case . hence. .70 Schr¨dinger Equation o and E1 + E2 + E3 = E. If two or all three orthogonal sides of the box have the same length further symmetry operations leave the system unaltered. . x3 ) = 1. 3. r) exp i − E(n1 n2 n3 ) (t − t0 ) φ(n1 . t|r0 .n2 . x2 . This additional symmetry is also reflected by degeneracies in the energy levels. . This symmetry has been exploited in deriving the stationary states. n3 = 1. . . Such degeneracies are always a signature of an underlying symmetry. 2. n2 . x1 . Actually. a2 .n3 =1 φ(n1 . x1 . x1 ) φn2 (a2 .n3 ) (a1 . . namely rotation by π around the x1 . Comparing (3. (3.120) can be written φ(n1 . a1 = a2 = a3 = a then rotation around the x1 . x2 ) φn3 (a3 . x3 –axes by π/2 also leaves the system unaltered. a3 .80) shows that the solutions of (3. 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. n1 . x2 . 3.

3. due to the hermitian character of ˆ H. e.1) (a.6: Particle in Three-Dimensional Box 71 ‘accidental’ degeneracies also occur. 6 This is a result of linear algebra which the reader may find in a respective textbook. However.3. a. n2 .2) (a. a. One particular aspect of the degeneracies illustrated in the Table above is worth mentioning. a. a. in case of degenerate states one cannot necessarily expect that stationary states are orthogonal. a. a. 5) as shown in the Table above. for (n1 . a.2. . for example. the identity 32 + 32 + 32 = 12 + 12 + 52 . φ(1. Any linear combination of wave functions ˜ φ(r) = α φ(1.2) (a.g. n3 ) = (3. a.2.1. to construct n orthogonal stationary states6 . 1. We consider the degeneracy of the energy E122 which is due to the identity E122 = E212 = E221 . 3) (1. The origin of this degeneracy is.. Hence.127) ˜ ˆ ˜ obeys the stationary Schr¨dinger equation H φ(r) = E122 φ(r). r). However. this linear combination o is not necessarily orthogonal to other degeneraste states. r) + β φ(2. r) (3.2. in case of an n–fold degeneracy it is always possible.2) (a. however. r) + γ φ(2.

72 Schr¨dinger Equation o .

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

i dx x = x ˆ (4.2) can be expressed in terms of the two operators p = ˆ d . The stationary states of the harmonic oscillator have been considered already in Chapter 2 where the corresponding wave functions (2.235) had been determined. act.1 Creation and Annihilation Operators The Hamiltonian operator (4. t) = exp(−iEt/ )φE (x.3) would be continuous and the eigenfunctions φE (x) would not be normalizable. t) where ˆ H φE (x) = E φE (x) . however.5). the natural boundary conditions apply x→±∞ lim φE (x) = 0 . 4. We will point out explicitly where assumptions are made which built on this restriction. is the commutation property 1 1 (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. therefore.4) be satisfied. The operators act ˆ on the space of functions N1 defined in (4.7) of the Hamiltonian. The Hamiltonian H can be expressed in terms of the operators acting on the space (4.1) of the form ψ(x. lim ( ) = } (4.3) can be solved for any E ∈ R. however. It is important to keep in mind this restriction of the space. This property states that p and x obey an ˆ ˆ algebra in which the two do not commute. in which the operators used below.7) which is why these operators are of interest to us.3) Due to the nature of the harmonic potential which does not allow a particle with finite energy to move to arbitrary large distances. only for a discrete set of E values can the boundary conditions (4. the commutator has a simple form. x] = p ˆ . The cardinal property exploited below.5) ˆ H = 1 1 2 p + m ω 2 x2 ˆ ˆ 2m 2 (4. (4. If this restriction would not apply and all functions f : R → R would be admitted. In the following algebraic solution of (4.4) Equation (4.8) i which holds for any position and momentum operator.5) →±∞ where C∞ denotes the set of functions which together with all of their derivatives are continuous. ∈ C∞ . (4.6) the first being a differential operator and the second a multiplicative operator. beside the representation (4. the spectrum of ˆ H in (4. all stationary states of the harmonic oscillator must be bound states and.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. In the framework of the Schr¨dinger equation the stationary o states are solutions of (4. In order [ˆ.

8) a ˆ ˆ ˆ ˆ ˆ implies [ˆ− .15) in a form which will be useful below ˆ H = ˆ H = ω a− a+ − ˆ ˆ ω a+ a− + ˆ ˆ ω 1 1 2 ω 1 . Before we continue we like to write (4.12) To prove this commutation property we determine using (4. the action of [ˆ. x] p ˆ on such functions must be considered.7) assuming that the factors are easier to handle than the Hamiltonian in yielding spectrum and eigenstates. x] f (x) = p ˆ d d x f (x) − x f (x) = f (x) = 1 f (x) . 2mω dx a− = ˆ mω x + ˆ 2 d .15) together lead to the commutation property (4.16) (4. 1 i dx i dx i i (4. p ˆ 2 2m ω 2 (4. 1 ω 2 (4.8) yield a− a+ = ˆ ˆ Similarly one can show a+ a− = ˆ ˆ 1 ˆ 1 H − 1 . ˆ 2 2m ω (4.9) From this follows (4. x] .1: Creation and Annihilation Operators 75 to derive (4.10) we define a+ = ˆ mω i x − √ ˆ p.15) 1 ˆ 1 H + 1 .11) ˆ The reader may note that we have attempted.14) and (4.13) (4. to factor H/ ω.14. 4. One obtains [ˆ. 2mω dx (4.4. 1 ω 2 (4.12). hence. Our next step is an attempt to factorize the Hamiltonian (4. Since a+ and a− are ˆ ˆ + and a− since operators and not scalars we cannot simply expect that the identy (4.7) and (4.14) 1 i mω 2 x + ˆ p2 + ˆ [ˆ.18) It is of interest to note that the operators a+ and a− are real differential operators. a ˆ 1 (4. a+ ] = a− a+ − a+ a− does not necessarily vanish. ˆ 2 2m ω a− = ˆ mω i x + √ ˆ p. the commutator property (4. ˆ ˆ .17) d dx We also express a+ and a− directly in terms of the coordinate x and the differential operator ˆ ˆ a+ = ˆ mω x − 2 d .8) we recall that all operators act on functions in N1 and.8).11) a− a+ = ˆ ˆ (4. in fact.10) holds for a ˆ ˆ [ˆ− . In fact. 1 2 (4. Being guided by the identity for scalar numbers (b − ic)(b + ic) = b2 − i(cb − bc) + c2 = b2 + c2 (4. a+ ] = 1 .

24) − m This property states that the operators in the function space N1 are the hermitian conjugate of each other.19) in the form1 f |ˆ+ g a Ω∞ = g|ˆ− f a Ω∞ .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) . (4.76 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. (4. 4. Using (??) we like to state (4. 4. 4. one obtains using (4. . 4.22) Similarly one can show using again (4.21) a− φE (x) is ˆ a stationary state of energy E − ω and a+ φE (x) is a stationary state of energy E + ω. 4.16. In fact.21) is available. The property ˆ ˆ follows directly from (??). 4. 4. ˆ (4. ˆ The results (4.17. a+ and a− as Ladder Operators ˆ ˆ The operators a+ and a− allow one to generate all stationary states of a harmonic oscillator once ˆ ˆ one such state φE (x) ˆ H φE (x) = E φE (x) (4. ˆ (4.23) 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) .19) This property states that the operators a+ and a− are the adjoints of each other. The derivation will be based solely on the properties (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) . (4.20) ˆ In the following we will determine the spectrum of H and its eigenstates.17. This property of operators is investigated more systematically in Section 5.16. g ∈ N1 +∞ +∞ dx f (x) a+ g(x) = −∞ −∞ dx g(x) a− f (x) .17.22.12.23) Together.19). 4.16. it holds that for a stationary state φE (x) of energy E defined through (4.

(4. (4. a state φ0 (x) which obeys the property a− φ0 (x) = 0 ˆ (4. 4. m ∈ Z when m is decreased. φ0 (y) dy dy the solution of which is 1 lnφ0 (y) = − y 2 + c0 2 φ0 (y) = c0 exp 1 − y2 2 . Another name is creation (ˆ+ ) and annihilation (ˆ− ) operators since these operators a a create and annihilate vibrational quanta ω.6. two difficulties: (i) one needs to know at least one stationary state to start the construction. E .4. in fact.. There arise. (4. be found.11) one can rewrite (4.2 Ground State of the Harmonic Oscillator d + y dy A suitable solution of (4.5).) ˆ ˆ 0 = φE |HφE Ω∞ − φE |HφE Ω∞ = ( E − E ) φE |φE Ω∞ . (ii) the construction appears to yield energy eigenvalues without lower bounds when.25) needs to be normalizable in order to represent a bound state of the harmonic oscillator. however.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.e.29) d 1 d φ0 (y) = = lnφ0 (y) = − y . on ˆ the other side such a state is itself an eigenstate of H as can be shown using (4.25) φ0 (y) = 0 (4.17) ˆ H φ0 (x) = ( ω a+ a− + ˆ ˆ 1 ω 1 ) φ0 (x) = 1 ω φ0 (x) .25) can. the solution φ0 (x) of (4. 2 2 (4.28) 4. one expects that E = 0 should be a lower bound.25) on one side would lead to termination of the sequence E0 + m.29) where y = mω x.32) for some constant c0 or (4. φ0 (x) should be an element of the function space N1 defined in (4.27) Since E = E one can conclude φE |φE Ω∞ = 0. It turns out that both difficulties can be resolved together. For f (x) = φE (x) and g(x) = φE (x) in (??) follows (Note that according to (??) the eigenvalue E is real. E = E . (4. Using (4. in fact.26) Of course.30) Assuming that φ0 (y) does not vanish anywhere in its domain ] − ∞. i. The property (??) has an important consequence for the stationary states φE (x). In fact.31) (4.33) . +∞[ one can write (4. For obvious reasons one refers to a+ and a− as ladder ˆ ˆ operators. Let φE (x) and φE (x) be two normalized stationary states corresponding to two different energies E.

it holds ˆ H φn (x) = ω(n + 1 ) φn (x) . We recall that according to (4. .34) dx |φ0 (x)|2 = |c0 |2 −∞ −∞ dx exp − mωx2 2 = |c0 |2 π ..e.24) can then be written as follows En = ω(n + 1 ). which for a narrowly localized state yields a large positive value.39) We notice that the ground state wave function φ0 (x) as well as the operators (ˆ+ )n are real. . i.25) admits only one solution there is only one set of states with energy E + m ω. a state confined to the minimum corresponding to the classical state of lowest energy. m ∈ Z which properly terminate at some minimum value E + m0 ω ≥ 0. Such states need to be suitably normalized for which purpose we introduce a normalization constant cn φn (x) = cn a+ ˆ n φ0 (x) .35) The appropriate ground state is φ0 (x) = mω π 1 4 exp − mωx2 2 . 2 This state of lowest energy is called the ground state of the oscillator.78 Linear Harmonic Oscillator This solution is obviously normalizable. . (4.26) the energy value associated with this state is 1 ω. we construct the states for n = 1. the so-called excited states. . would yield a vanishing contribution. 1. We a can.. i. 2 n = 0. 2.e. The reason is that in the Hamiltonian (4. The set of allowed energies of the oscillator according to (4. . 2.36) assumes a functional form such that both terms together assume a minimum value. . We will consider this point more systematically in Section 5. .. n = 0.39). . 2. For this purpose we apply the operator a+ to the ˆ ground state (4. therefore. i. choose the normalization constants cn and the functions φn (x) real as well. (4. . The conventional normalization condition φ0 |φ0 reads +∞ +∞ Ω∞ = 1 (4. (4. 2. the potential energy contribution which for a state δ(x).38) These states correspond to the energy eigenvalues (4.e.36) Since the first order differential equation (4. mω (4. we can now construct the stationary states corresponding to energies (4.37) above the ground state energy. (4. . 4. The ground state (4. 1.2) there are two competing contributions to the energy. . .36) n times. 1. and the kinetic energy contribution.37) It is most remarkable that the energy 1 ω of the ground state is larger than the lowest classically 2 allowed energy E = 0.3 Excited States of the Harmonic Oscillator Having obtained a suitable stationary state with lowest energy. 2 n = 0.. .

42) ˆ Using (4. √ a+ φn (x) = n + 1 φn+1 (x) .15 .46) According to (4.39) yield √ From this follows βn = 1/ n and. n = 0. For n = 0 holds c0 = 1. βn must be real as well. 4.45) We like to note that these functions according to (4. (4. i. they obey φn |φn = δn n . .50) φn (x) .40).e.24) holds in analogy to (4. To determine βn we note using (4. Ω∞ 2 = βn a− φn |ˆ− φn ˆ a Ω∞ 2 = βn φn |ˆ+ a− φn a ˆ Ω∞ .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. 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. 2. To determine these constants we adopt a recursion scheme. .47).51) .48) (4. n = 0. We consider then the situation that we have obtained a properly normalized state φn (x).41) Employing the adjoint property (4. (4. . Since a− is a real differential operator [see (4. 1.45) form an orthonormal set. 1. . according to (4. (4.18)] and since the ˆ φn (x) are real functions. ˆ (4.43) (4. (4. .47) for some suitable constants βn .3: Excited States 79 We need to determine now the normalization constants cn .20) yields 2 αn φn |ˆ− a+ φn a ˆ Ω∞ = 1. 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 .40–4.49) (4.. 2. n. according to (4. (4.4.44) One can conclude then that the stationary states of the oscillator are described by the functions 1 φn (x) = √ a+ ˆ n! n φ0 (x) .28) and the construction (4. (4.20) 1 = φn−1 |φn−1 Equations (4. . √ a− φn (x) = n φn−1 (x) .40) φn−1 (x) = βn a− φn (x) ˆ (4.40) = 1. .

For this purpose we start from expression (4. 2.54) 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 . We will demonstrate below the identity Hn (y) = e y2 2 y − d dy n e− y2 2 (4.57) which includes the factor 1/ n! according to Eqs. by dx = dy mω 1 4 .233) derived in Sect. (4.46) yields a set of normalized states.30) and employing the normalization +∞ −∞ dy φ2 (y) = 1 n (4.200).52). i. for example.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) .7. by the Rodrigues formula (2. however.39).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 (4.35) rather than (4. .58) This result agrees with the expression (2.53) We will later re-introduce this factor to account for the proper normalization (4.55) 1 y2 2 (4. by introducing the variable y defined in (4. (4.40–4. 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 .28) of the ground state and the definition (4.45). (4. (4. n–independent factor..7 and given. 2. It should be noted that √ the normalization (4.35) by a constant.e.52) This normalization of the wave functions differs from that postulated in (4.56) can be expressed in terms of Hermite polynomials Hn (y) introduced in Sect. (4. namely the square root of the Jacobian dx/dy. simplifying the calculation.56) Relationship to Hermite Polynomials We want to demonstrate now that the expression (4.

200) we need to verify y − which implies Hn (y) = (−1)n ey 2 d dy n e− y2 2 = (−1)n e y2 2 dn −y2 e . For the sake of notational simplicity we employ initially the coordinate y as defined in (4.59) 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 .64) which implies that (4.30) and return later to the coordinate x.59) by induction noting first that (4. g(y) = (−1)n+1 e y2 2 dn+1 −y2 e = dy n+1 y − d dy n+1 e− y2 2 . Starting point of our derivation is the assumption that the initial condition can be expanded in terms of the eigenstates φn (y) (4. We prove (4.106–3. hence. 4. (3.59) hold.200).65) . t0 ).59) holds for n = 0.e. the Rodrigues formula (2. therefore. (4. and showing then that the property also holds for n + 1 in case it holds for n. (4. 1. t0 ) = n=0 dn φn (y0 ) .39) ∞ ψ(y0 .60) and.4: Propagator 81 To verify the realtionship between the definition (4. (4. dy dy (4.63) one obtains g(y) = y − d dy f (y) (4.4.57) of the hermite polynomials and the definition given by (2.61) and.2) for an arbitrary initial wave function ψ(x0 . dy n (4. Our derivation will follow closely the procedure adopted for the case of a ‘particle in a box’ [see Eqs.4 Propagator for the Harmonic Oscillator We consider now the solution of the time-dependent Schr¨dinger equation of the harmonic oscillator o (4.. i.1.62) Denoting f (y) = (y − d/dy)exp(−y 2 /2) and employing y2 y2 d d − y2 e 2 f (y) = −y e− 2 f (y) + e− 2 f (y) dy dy (4.59) dn −y2 e dy n (4.61) One can factor g(y) and employ (4.114)]. 4.

5). sec:harm.46) the expansion coefficients dn are +∞ dn = −∞ dy0 ψ(y0 . 1965) Sect. (4.67) for which according to (3. 2 A demonstration of this property can be found in Special Functions and their Applications by N. t0 ) ψ(y0 .Lebedev (Prentice Hall.s.71) Applying (??) to express Hn (y) and Hn (y0 ) yields exp(−2tuv) w(y. In order to prove the equivalence of (4. Inc..147) for the harmonic iscillator determined in Sect. this is an excellent textbook from which we have borrowed heavily in this Section.72) × exp( −u − v + 2iyu + 2iy0 v ) . 2.225) which allows one to derive a generating function for products of Hermite polynomials which can be applied to the r.65) to times t ≥ t0 through insertion of time-dependent coefficients cn (t1 ) ∞ ψ(y. (4. t) (4.N.112) and (4.70).147) we employ the technique of generating functions as in Sect. pp.15. t0 ) φn (y0 ) . a supposition which is not proven here2 .39) holds cn (t1 ) = exp −i ω ( n + 1 ) ( t1 − t0 ) 2 . t1 |y0 . 4. t) = π −3/2 exp 2 2 y 2 y0 + 2 2 2 +∞ +∞ ∞ du −∞ −∞ dv n=0 1 ( −2tuv )n × n! (4. t|y0 .66) One can extend expansion (4. 68.69) where ∞ φ(y. N. For this purpose we start from the integral representation (2. Englewood Cliffs.J.82 Linear Harmonic Oscillator Such expansion is possible for any f (y0 ) = ψ(y0 . t1 ) = n=0 dn φn (y) cn (t) (4.108–3. t) = n=0 Hn (y)e−y /2 Hn (y0 )e−y0 /2 n √ t . t1 ) = −∞ dy0 φ(y. 2n n! π 2 2 (4..68) Altogether one can express then the solution +∞ ψ(y. Employing orthogonality condition (4. t0 ) which is an element of N1 defined in (4. We want to demonstrate now that this propagator is identical to the propagator (2.70) and (2.70) is the propagator of the linear harmonic oscillator. t0 ) = n=0 φn (y) φn (y0 ) tn+ 2 . y0 .h. .7. y0 . We consider for this purpose the following expression for |t| < 1 ∞ w(y. 1 t = e−iω(t1 −t0 ) (4. of (4.

147) derived by means of the path integral description. t1 |y0 . identical to the generating function (4.e. 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 = . i. −∞ (4. y0 . a Re a2 > 0 (4. i. (4.70) is. t0 ) = where F (t1 . (4. t0 ) exp i 2 1 2 (y + y0 ) G(t1 . it holds φ(y.4. t0 ) = i φ(x. √ πexp( t2 v 2 − 2iytv − y 2 ) (4.78) 1 − e−2iω(t1 −t0 ) = sinω(t1 − t0 ) 2 i e−iω(t1 −t0 ) (4.74) applied once results in w(y.s.78) in terms of the coordinates x and x0 .71) of w(y. by a factor 4 mω/ for both φn (y) and φn (y0 ). t0 ) (4..79) 1 + e−2iω(t1 −t0 ) cosω(t1 − t0 ) = . t0 ) − i y y0 2 F (t1 .h.4: Propagator Carrying out the sum on the r.77) The sum in (4. The resulting propagator is G(t1 . one obtains w(y. indeed.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 . This requires that we employ (4. . (4. y0 .30) to replace y and y0 and that we multiply the propagator by mω/ . (4. t0 ) = exp mω 2iπ sinω(t1 − t0 ) 1 2 × .80) −2iω(t1 −t0 ) sinω(t1 − t0 ) 1−e We can finally express the propagator (4.e..81) imω 2 sinω(t1 − t0 ) (x2 + x2 ) cosω(t1 − t0 ) − 2 x x0 0 This result agrees with the propagator (2.77).74) a second time yields finally together with the definition (4. t) = π −3/2 exp +∞ 83 y2 y2 + 0 2 2 +∞ dv exp( −v 2 + 2iy0 v ) × −∞ × −∞ du exp( −u2 − 2u(tv − iy) ) .76) One can express this in terms of the stationary states (4.73) The incomplete Gaussian integral +∞ −∞ dx e−a 2 x2 √ − 2bx = π b2 /a2 e . t|x0 . t) = π −1 exp − 2 y 2 y0 + 2 2 +∞ dv exp(−(1 − t2 ) v 2 − 2i (yt − y0 ) v) . y0 .75) Applying (4. t0 ) = and 1 2 i π F (t1 .

(4.e. a (4. 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. ˆ ˆ Calculus of Creation and Annihilation Operators We summarize first the key properties of the operators a+ and a− ˆ ˆ [ˆ− . hence..5 Working with Ladder Operators In the last section we have demonstrated the use of differential equation techniques. We want to introduce now techniques based on the ladder operators a+ and ˆ a− . phonons. The reason for introducing the calculus of the operators a+ and a− is that ˆ ˆ this calculus proves to be useful for the description of vibrations in crystals. Here f (ν) (y0 ) denotes the ν-th derivative of f (y) taken at y = y0 . 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. In this case we define ∞ f (ν) (0) + ν (4. a− = √ ˆ y + .. e. 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! .g. one can express y = 2(ˆ− + a+ ) and d/dy = 2(ˆ− − a+ ) and.84) = 1 1 = 0 √ = n + 1 φn+1 (y) √ = n φn−1 (y) = f |ˆ+ g a Ω∞ . both quantum systems are endowed with a large number of modes. 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 − . i. a− must be equivalent.83) We will encounter below functions of a+ and a− .85) is convergent everywhere in R. (4. and of the modes of the quantized electromagnetic field. the approaches a ˆ a ˆ using y. the use of generating functions. 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! (4.82) a+ = √ ˆ dy dy 2 2 √ √ Obviously.86) f (ˆ+ ) = a a ˆ ν! ν=0 .84 Linear Harmonic Oscillator 4. d/dy and a+ . each corresponding to a single harmonic oscillator of the type studied presently by us. f (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 .88) (4. f (ˆ+ )] + f (ˆ+ ) a− ˆ a a a a ˆ which implies that in order to prove (4.90) we show that (4. (ˆ+ )n a+ ] = (ˆ+ )n [ˆ− . property (4. i.96) .88) we need to show actually [ˆ− . a− f (ˆ+ ) φ0 (y) = f (1) (ˆ+ ) φ0 (y) ˆ a a which follows from a− φo (y) = 0. fn+1 (ˆ+ )] a a = [ˆ− . An example is the so-called shift operator exp(uˆ− ). a a (4.90) To prove (4.4. (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. Note that ˆ a an immediate consequence of (4. a+ ] + [ˆ− . The following important property holds a a− f (ˆ+ ) = f (1) (ˆ+ ) + f (ˆ+ ) a− . ˆ a We proceed by induction noticing first that (4.93) one can reduce all operators acting on some proper state function by operators acting on the state φ0 (y). a a − (4. The first case is trivial.87. 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+ ).91) (4.94) This identity follows from (4. ˆ a a a ˆ In particular. a a a ˆ 1 a Let us assume that (4. For fn+1 follows then [ˆ− .88) is (ˆ− )n f (ˆ+ ) φo (y) = f (n) (ˆ+ ) φ0 (y) .90) holds for f (ˆ+ ). It holds a a euˆ f (ˆ+ ) φ0 (y) = f (ˆ+ + u) φ0 (y) .89) (4. a+ ] = 1 = f1 (ˆ+ ) .90) holds for any function fn (ˆ+ ) = (ˆ+ )n which is a power of a a a+ . the second case follows from [ˆ− . 4. Hence. f1 (ˆ+ )] = [ˆ− .83) and can be proven for all powers fn by induction and then inferred for all proper functions f (ˆ± ). a An important operator function is the exponential function.90) holds for fn .95) (4.88) is a fundamental one and will be used in the following. a a a 85 (4. f (ˆ+ )] = f (1) (ˆ+ ) .5: Working with Creation and Annihilation Operators and similarly for f (ˆ− ). The convergence of the Taylor expansion ascertains then that (4.e. we will only assume operator functions acting on φ0 (y).87) (4.90) holds for f0 and for f1 .92) f (y) = n=0 dn φn (y) = n=0 (ˆ+ )n a dn √ φ0 (y) n! (4. We note ˆ a− f (ˆ+ ) = [ˆ− ..

106) yields a eαˆ + ∗ a− ˆ a a eαˆ φ0 (y) = e−αα eαˆ φ0 (y) .105) a eαˆ + − α∗ a− ˆ φ0 (y) = e αα∗ 2 e−α ∗ a− ˆ a eαˆ φ0 (y) . + (4.108) .99) ˆ C(u) = zˆ a zˆ+ + z ∗ a− a ˆ + +z a ˆ ∗ − a euzˆ euz ˆ − u C(u) . one obtains a evˆ + − v ∗ a− ˆ a φ0 (y) = e− 2 vv evˆ e− v 1 ∗ + ∗ a− ˆ φ0 (y) (4. hence. the solution of (4. a a ν! (4.101) 1. 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.98) for u = −α∗ and v = α yields e−α This together with (4. that the operators considered act on φ0 (y). + (4. without explicitly stating this. euzˆ ˆ euz ∗ a− ˆ (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) . 1 + + uz ∗ a− ). + ∗ a− ˆ a − u euzˆ euz + ∗ a− ˆ = (4. One can conclude then using the definition of C(u) and defining ˆ ˆ is D(u) = exp(uzˆ a ˆ v = uz 1 ∗ ˆ− ∗ ∗ ˆ− a+ a+ evˆ + v a φ0 (y) = e 2 vv evˆ ev a φ0 (y) .96) is applied is a a a euˆ evˆ φ0 (y) = ev (ˆ − + + + u) a φ0 (y) = euv evˆ φ0 (y) .107) − α∗ a− ˆ φ0 (y) = e− αα∗ 2 a eαˆ φ0 (y) .103) (4.98) The related operators exp[vˆ+ ± v ∗ a− ] play also an important role. + ∗ + (4.90) and zz ∗ = 1 we can write this d du e uzˆ+ a e uz ∗ a− ˆ a − z ∗ a− .106) Applying (4. Below we will use the operator identity which follows for the choice of v = α in (4.101) ˆ ˆ ˆ C(u) obviously obeys C(0) = 1 This results in D(0) = 1 and. + (4. ∗ = exp(uzˆ a ˆ and determine its derivative d du ˆ C(u) a = zˆ+ euzˆ euz a + ∗ a− ˆ a + euzˆ z ∗ a− euz ˆ + ∗ a− ˆ = + zˆ+ + z ∗ a− a ˆ Using (4. (4. z ∈ C. We assume in the following a ˆ derivation.97) An example in which (4.86 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) .104) (4.100) 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) (4.102) Similarly.

e.109) e = π4 n! n=0 √ Using (4. We want to derive now the values Hn (0).s. (??). Similarly.95) one obtains a a a a euˆ φ0 |evˆ φ0 = φ0 |euv evˆ φ0 = euv evˆ φ0 |φ0 = euv − + + + − (4.116) = δµ ν . in the ladder operator calculus. Expanding r.113) Comparision of all terms on the l.s.s. We start from the generating function (??) and replace according to (4.h.84) yields ∞ ν.111) the orthonormality properties of the wave functions φn (y).4. the expected from which follows by comparision of each term on both sides φµ |φν orthonormality property.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! (4.115) a where the latter step follows after expansion of evˆ and using a− φ0 (y) = 0. and ˆ l.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 will derive the equivalent of a generating function and use it to rederive the values Hn (0) and the orthonormality properties of φn (y).112) Expanding both sides of this equation and using (4. of (4..58) the Hermite polynomials Hn (y) by the eigenstates φn (y) √ ∞ 1 ( 2t)n y2 /2 2yt − t2 √ e φn (y) (4. i. and on the r.98) and again (4. + (4. 4.h.h.111) This expression.95).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. provide the same values for Hn (0) as provided in Eq.h. is the equivalent of the generating function (??).111) yields π − 4 exp − 1 u2 2 a = euˆ φ0 (0) .110) √ y2 u2 + 2uy − 2 2 a = euˆ φ0 (y) . Setting y = 0 in (4.115) and using (4. .µ=0 uµ v ν √ φµ |φν µ!ν! ∞ = µ=0 uµ v µ µ! (4.114) where we have employed (4. (4. Using (4.s. we can reproduce by means of the generating function (4.84. For this purpose we consider a a a a euˆ φ0 |evˆ φ0 = φ0 |euˆ evˆ φ0 + + − + (4.

(4. (−i)n ˜ φn (p) = √ 2n n! m˜ ω π 1 4 exp − m˜ p2 ω 2 Hn ( m˜ ω p) . In the momentum representation the wave function is a function of p. 1. according to (4. For example. the momentum and position operators are p = p. n = 0.7) is given by (4.118) d2 1 2 + p dp2 2m ˜ ˜ φE (p) = E φE (p) .124) ). using (4..125) . The eigenvalues..123) (4. The momentum representation eigenfunctions are. ˜ (4. ω = 1 / m2 ω . (4. (4.88 Linear Harmonic Oscillator 4. i. is given by the ˜ function ψ(p). i.121). .37).3)..235). using (2.119) 2m 2 dp2 Accordingly. 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. dp (4.6) and the Hamiltonian (4. the probability density is P (x) = |φn (x)|2 . . In the position representation the wave function is a function of x. 1.235). 4.120) Multiplying this equation by 1/m2 ω 2 yields − where ˜ E = E / m2 ω 2 .e. n = 0. for an oscillator in a stationary state φn (x) as given by (2. (4.2.122) For a solution of the Schr¨dinger equation in the momentum representation. For this purpose we employ the Schr¨dinger equation in the momentum o 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.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. As to be expected.7) is 1 2 m 2 ω 2 d2 p − . the eigenvalues are identical to those determined in the position representation.e. ˆ and the Hamiltonian (4. En = ω (n + 1 ) 2 d2 1 + m˜ 2 p2 ω 2 2m dp 2 2 ˜ ˜˜ φE (p) = E φE (p) (4.121) ω (n + ˜ 1 ) 2 . (4.122). The solutions can be stated readily exploiting the earlier results. . i.2). the momentum and position operators are as stated in (4. . 2. the time-independent Schr¨dinger equation for the oscillator can be written o ˆ H = − mω 2 2 2 x = i ˆ d . 2. is given by a function ψ(x).e. . of (4. are ˜ En = or.

129) m ω one obtains. i. through direct integration. m ω (4. m ω (4.235)] absorbs the replacement dx → dy.e.127) exp − p2 m ω 2 Hn ( 1 p) . 1 ˜ φn (k) = √ 2π +∞ (4.58).132) dy exp(−iky) φn (y) −∞ (4. defined only up to an arbitrary phase factor eiα . ˜ We can now state the probability density Pn (p) for an oscillator with energy En to assume momentum p.134) Exercise 4.e.133) where we note that a change of the normalization of φn (y) [c. hence.129) follows ky = px/ and. 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. 4.126).1: Demonstrate the validity of (4.. Defining k = m˜ / p or ω 1 k = p (4.126) Normalized eigenfunctions are.30) to obtain (4. This implies the property of Hermite polynomials (−i)n e− k2 2 +∞ Hn (k) = −∞ dy exp(−iky) e− y2 2 Hn (y) (4.131) in the present case employing dimensionless units.125.122) (−i)n ˜ φn (p) = √ 2n n! 1 πm ω 1 4 89 exp − p2 2m ω Hn ( 1 p) .4. (4. 2n n! π (4. φn (k) = We want to finally apply the relationship 1 ˜ φn (p) = √ 2π +∞ 2 (−i)n −k √ e 2 Hn (k) .f. the correctness of the phase factors (−i)n .30) and (4. instead of (4.126) a phase factor (−i)n .6. this allowed us to introduce in (4.128) We may also express the eigenstates (4. α ∈ R.130) dx exp(−ipx/ ) φn (x) −∞ (4.126) in terms of dimensionless units following the procedure adopted for the position representation where we employed the substitution (4. From (4. of course. . i..30) and (2.6: Momentum Representation or with (4. Proceed as follows. the Jacobian.134).

135) g(k.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. for example. a pendulum swinging at small amplitudes.139) where we have added the condition that the states be normalized. namely through the eigenvalue problem for normalized states a− φ(α) (y) = α φ(α) (y) . y) −∞ (4. and G. for example.138) We will show that a harmonic oscillator prepared in such a state will remain forever in such a state. 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.90 (a) Prove first the property 1 f (k.1 showing in its top row the attributes of such state just discussed. We want to construct and characterize such states. coherent states. 2. (4. The answer is ‘yes’. 1989) which discusses quasi-classical states of the free electromagnetic field in Sect.Chen-Tannoudji.196) of Hermite polynomials and equate coefficients of equal powers of z to prove (4. z) = exp −2ikz + z 2 − k 2 /2 . New York. The quasi-classical states can be obtained by generalizing the construction of the ground state of the oscillator.— III. J. +∞ Linear Harmonic Oscillator dy exp(−iky) g(z.C. Inserting this expansion into √ (4. The question arises if similar states exist also for the quantum oscillator. z) = exp 2yz − z 2 − y 2 /2 (4. pp.194. For this purpose we assume such state exists and expand ∞ ∞ φ(α) (y) = n=0 d(α) φn (y) . We first construct the solution of (4. 4. carries out a periodic motion described by y(t) = Re y0 eiωt (4. φ(α) | φ(α) = .7 Quasi-Classical States of the Harmonic Oscillator A classical harmonic oscillator.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. Inc. ˆ α ∈ R.136) (b) Employ the generating function (2. or Glauber states of the harmonic oscillator and they play.Dupont-Roc. The motion of the probability distribution of such state is presented in Fig.. 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 . Such states are referred to as quasi-classical states.140) An excellent textbook is Photons and Atoms: Introduction to Quantum Electrodynamics by C.138). except that α changes periodically in time. 4.134). 192 3 .4. n n=0 d(α) n 2 = 1 (4.Grynberg (John Wiley & Sons. z) = √ 2π where f (k.

138). This interpretation justifies the choice of α ∈ R in (4.4.146) √ 1 − (y − 2 α)2 2 (4. [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) . i.142) The solution is αn d(α) = √ c n n! ∞ ∞ (4.e. hence. n! n (4.148) The identity (4.144) Normalization requires c = exp(−α2 /2) and.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) .147) allows one to interprete T (α) as an operator √ which shifts the ground state by 2α.149) with (4.7: Quasi-Classical States or ∞ n=0 91 √ n + 1 dn+1 − α d(α) n (α) φn (y) = 0 . n n+1 (4. Comparing (4. However.150) for α presently real..149) − α∗ a− ˆ (4.145) Using the generating function in the form (4.145) using (4.139)] 1 = n=0 d(α) n 2 = |c|2 n=0 α2 2 = |c|2 eα . n! (4.108) 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. the quasi-classical states are φ (α) (y) = exp α2 − 2 ∞ n=0 αn √ φn (y) .84) φ(α) (y) = exp − α2 2 ∞ n=0 αα∗ (αˆ+ ) a a+ φ0 (y) = e− 2 eαˆ φ0 (y) . n! (4. One can also write (4.143) for a constant c which is determined through the normalization condition [see (4. n! (4.147) √ which identifies this state as a ground state of the oscillator displaced by 2α. . we will see below that α ∈ C are also admissible and will provide a corresponding interpretation. (4.109).

150) for complex α obeys the unitary property T + (α) T (α) = 1 1 (4. t1 ) = exp or φ (α) − α2 2 ∞ exp n=0 −i ω ( n + ∞ n=0 1 ) ( t1 − t0 ) 2 αn √ φn (y) n! (4.151) which follows from a generalization of (4. n = 0.150) 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. is the adjoint of the operator function on the l. .154) .67. the shift operator leaves the ground state normalized.h. Relationship (4.158) Comparision of this expression with (4. .159) Noticing the identity which holds for |u|2 = 1 u2 + 1 = (Re u)2 − (Im u)2 + 1 + 2 i Re u Im u = 2 (Re u)2 + 2 i Re u Im u (4. t1 ) = u exp 1 2 α2 − 2 (α u)n √ φn (y) .156) (4. n! u = e−iω(t1 − t0 ) .95) to linear combinations αˆ+ + βˆ− and reads then a a f (αˆ+ + βˆ− ) φ|ψ = φ|f (α∗ a− + β ∗ a+ ) ψ .160) .s. α ∈ C (4.68)] one can write the time-dependent solution with (4.h. Following the procedure adopted in determining the propagator of the harmonic oscillator [see (4. In particular.155) (4.145) shows that α in (4. .109) serves again to simplify this sum   √  y2 1 1 α2 (u2 + 1)   φ(α) (y.92 Linear Harmonic Oscillator The shift operator as defined in (4. (4.e. i.145) as the initial state φ(α) (y. it holds T + (α) φ0 |T + (α) φ0 = f |T (α) T + (α) g = φ0 |φ0 .s. a a ˆ ˆ (4. becomes real. t1 ) = π − 4 u 2 exp  − + 2yαu −  2  2 E (4.145) is replaced by a complex number αu which at times t1 = t0 + 2πn/ω.. 4.157) (y. one can write in case of T + (α) T + (α) + = T (α) = eα ∗ a− ˆ − αˆ+ a a = e − (αˆ + − α∗ a− ) ˆ .152) T + (α) f |T + (α) g = f |T (α) T + (α) g = f |g . 1.153) This is obviously the inverse of T + (α) as defined in (4.152) Since the operator function on the r. (4.

obviously.150) with (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 (4.161) and is found to describe a displaced Gaussian. t1 ) = u 2 T + (α u) φ0 (y) (4.4.7: 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. 4.158) allows one to write 1 φ(α) (y.138. 4. a momentum factor and a phase factor. Comparing (4. except that it is not pointlike. this characterization corresponds closely to that of the possible initial states of a classical oscillator.163) This solution corresponds to the initial state given by (4.1 (top row) we present the probability distribution of the Glauber state for various instances in time.148.165) .162) 2 with a periodic change. 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. experiences a change of its width (relative to . moving solely its center of mass in an oscillator fashion around the minimum of the harmonic potential. Our interpretation 2 explains also the meaning of a complex α in (4. The oscillations of the mean position y0 (t1 ) and of the mean momentum k0 (t1 ) are out of phase by π just as in the case of the classical oscillator. a complex α corresponds to a displaced oscillator which has initially a non-vanishing velocity. (4.138): a real α corresponds to a displacement such that initially the oscillator is at rest. (4.164) where according to (4.159) is then 1 1 i φ(α) (y. One can express (4. in general.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. t1 ) = π − 4 exp − [y − y0 (t1 )]2 + iy k0 (t1 ) − iφ0 (t1 ) − ω(t1 − t0 ) 2 2 . being displaced around the center with period 2π/ω and. The diagram illustrates that the wave function retains its Gaussian shape with constant width for all times.163) also through the shift operator. 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. This shows clearly that the Glauber state is a close analogue to the classical oscillator.147). The time-dependent wave function (4.

σ|y.1: Time dependence of Gaussian wave packets in harmonic oscillator potential.168) dy e−p(y−q) Hn (y) where p = 1 + σ2 . 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. t0 ) = [πσ 2 ]− 4 exp we expand ∞ 1 − (y − ao )2 2σ 2     (4.169) . (4. To describe such state which satisfies the initial condition φ(ao . t0 ) = n=0 dn φn (y) .166) φ(ao . This behaviour is illustrated in Fig. that of the Glauber states) with a period π/ω.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. 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. 1 + σ2 (4.167) One can derive for the expansion coefficients +∞ dn = −∞ dy φn (y) φ(ao . σ|y. σ|y. 2σ 2 q = a .1 (second and third row). 4.

P. therefore.7: Quasi-Classical States The solution of integral (4. p−1 1 − σ4 t = 1 − σ 2 −iω(t1 −to ) e 1 + σ2 (4.172) One can then conclude e−a /4 n lim dn = √ a σ→1 2n n! (4.170) In order to check the resulting coefficients we consider the limit σ → 1.166) in analogy to (4. New York. Yu. Prudnikov.175) ∞ tn n=0 2n n! Hn (yo ) e−yo /2 Hn (y) e−y 2 2 /2 The generating function (4. Brychkov. 1990). σ|y. i.. t1 ) = √ 2 × exp − 1 (y 2 + yo ) 1+t2 − 2 1−t 2 1 1 √ √ pσ π 1 − t2 × t 1 − t2 (4.157) as follows φ(ao . in fact. a worthy successor of the famous Gradshteyn. σ→1 O(y n−2 ) for n ≥ 2 1 for n = 0.174) Hn (q p p−1 ) φn (y) . 1 a2 2 1 + σ2 + 1 2 2 y o + 2 y yo This integral can be found in Integrals and Series. a ground state shifted by ao = 2α. unfortunately very expensive.4.177) − 1 ω(t1 − to ) 2 . namely $750 4 . t1 ) = √ × where yo = q p a = √ .143). vol. Following the strategy adopted previously one can express the time-dependent state corresponding to the initial condition (4. 2 by A. In this limit the coefficients dn should√ become identical to the coefficients (4.176) 1 √ pσ π a 2 exp − 1 1+σ2 + 1 yo − 1 ω(t1 − to ) 2 2 2 2 × (4. For this purpose we note that according to the explicit expresssion (??) for Hn (y) the leading power of the Hermite polynomial is Hn (y) = (2 y)n + and. σ→1 p−1 2 (4.e. 1 (4.I.168) is4 e−aq/2 (p − 1)n/2 dn = √ Hn (q 2n n!σ p(n+1)/2 p ). σ|y.76) permits us to write this φ(ao . this 3 volume integral table is likely the most complete today.171) lim p−1 p n 2 Hn (q p ) = lim (2q)n = an .173) which is. in agreement with the expected result (4. This expansion can be written φ(ao . t1 ) = e−aq/2 √ pσ ∞ n=0 e −iω(n+ 1 )(t1 −to ) √ 2 1 2n n! p−1 p n/2 (4.143) of the Glauber state. Marichev (Wiley. σ|y. p−1 95 (4. and O.

96 Linear Harmonic Oscillator .

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

(5. We assume the convention that rotations are right-handed. Let us define the rotated vector as r = R(ϑ) r .98 Properties of Rotations in R Theory of Angular Momentum and Spin Rotational transformations of vectors r ∈ R . i. (5. (5. i.e. (5. ϕ)T ) is negative.7) Let us consider briefly an example to illustrate rotational transformations and to interpret the sign of detR(ϑ).6) (5. In the latter case the determinant of R(ϑ = (0. 2. From (5. x3 )T . .6) follows immediatety for the determinant of R(ϑ) using detA B = detA detB and detAT = detA detR(ϑ) = ±1 . 0. ϑ parametrizes the rotations uniquely as long as |ϑ| < 2π. x2 . the matrix represents a proper rotation.k. i.e. ϕ)T ) = ±  sinϕ cosϕ 0  (5. In the following we want to exclude inversions and consider only rotation matrices with detR = 1. they conserve a · b = 3 aj bj . in Cartesian coordinates r = (x1 . A rotation around the x3 -axis by an angle ϕ is described by the matrix   cosϕ − sinϕ 0 R(ϑ = (0. =1 [R(ϑ)]jk [R(ϑ)]j ak b = j=1 aj bj .1) In Cartesian coordinates this reads 3 xk = j=1 [R(ϑ)]kj xj k = 1. a minus sign in Eq. are linear and.8) 0 0 1 In case of a prefactor +1. it follows 3 [R(ϑ)]jk [R(ϑ)]j = δk j=1 . 3 . (5. in case of a prefactor −1 a rotation and an inversion at the origin. (5...(5.2) Rotations conserve the scalar product between any pair of vectors a and b.e. then the fingers of your fist point in the direction of the rotation. can be represented by 3 × 3 matrices R(ϑ). if the thumb in your right hand fist points in the direction of ϑ. where ϑ denotes the rotation.5) This equation states the key characteristic of rotation matrices.7) implies that a rotation is associated with an inversion.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 . namely around an axis given by the direction of ϑ and by an angle |ϑ|. therefore. 0. It holds then j=1 3 3 a ·b = j.3) Since this holds for any a and b.

1 T Similarly. forms a group G satisfying the axioms (i) For any pair a. We would like to point out the obvious property that for all elements R of SO(3) holds R−1 = RT (5. Since (R1 )T = (R1 )−1 it follows −1 −1 −1 T (R1 )T R1 = (R1 )−1 R1 = −1 which implies R1 ∈ SO(3). one can show R3 R3 = 1 Furthermore. R3 = R1 R2 is a real. it holds 1. (5.11) (iii) From detR1 = 0 follows that R1 is non-singular and. We have shown altogether that the elements of SO(3) form a group.13) . there exists a real 3 × 3–matrix −1 R1 which is the inverse of R1 . (iii) ∀a. R RT = RT R = 1 detR = 1 } 1. We need to demonstrate that this inverse belongs also to −1 T SO(3).12) (iv) Since the associative law holds for multiplication of any square matrices this property holds for the elements of SO(3).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. in fact. 99 (5. T R1 R1 −1 = 1 −1 = 1 1 1 (5. It holds T T T T R3 R3 = ( R1 R2 )T R1 R2 = R2 R1 R1 R2 = R2 R2 = 1 . (5. b ∈ G the product c = a ◦ b is also in G. a ∈ G → ∃a−1 ∈ G with a ◦ a−1 = a−1 ◦ a = e. We want to prove now that SO(3) as defined in (5.9) This set of matrices is closed under matrix multiplication and. (i) Obviously. In the following we will assume R1 . (ii) The group element ‘e’ is played by the 3 × 3 identity matrix 1 1. (iv) For products of three elements holds the associative law a ◦ (b ◦ c) = (a ◦ b) ◦ c. (ii) There exists an element e in G with the property ∀a ∈ G → e ◦ a = a ◦ e = a. hence. R2 ∈ SO(3) and do not write out “◦” explicitly since it represents the matrix product. It follows that R3 is also an element of SO(3).5. 3 × 3–matrix.10) detR3 = det ( R1 R2 ) = detR1 detR2 = 1 .9) forms a group. For this purpose we must demonstrate that the group properties (i–iv) hold.

J3 as follows i R(ϑ) = exp − ϑ · J (5. This representation expresses rotation matrices in terms of three 3 × 3– matrices J1 . Lie Group.13) R(ϑk ek )−1 = R(ϑk ek )T = [exp ( ϑk Lk )]T = exp ϑk LT ˆ ˆ k and.15) Below we will construct appropriate matrices Jk . Lie Algebra We want to consider now a most convenient. 0 1 −1 0 . −i} together with multiplication as the binary operation ‘◦’.18) (5. we want to assume that they exist. 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. i.1: Test if the following sets together with the stated binary operation ‘◦’ form groups.20) (5. will play a central role in the representation of many other symmetry transformations in Physics. due to the uniqueness of the inverse. k = exp ( −ϑk Lk ) (5. establish if 1-1 homomorphic mappings exist.19) (5. presently. x2 –plane {rotation by Oo . −1. M3 . −1 0 0 −1 . J2 . (b) the set of matrices {M1 .100 Theory of Angular Momentum and Spin Exercise 5. We want to show that for the property R RT = 1 to hold. M4 } = 1 0 0 1 . Generators. (c) the set of four rotations in the x1 .17) .14) together with matrix multiplication as the binary operation ‘◦’.16) must be antisymmetric. M2 . rotation by 90o . compact representation of R(ϑ) which will be of general use and. the matrices i Lk = − Jk (5.e. mappings between the sets which conserve the group properties: (a) the set of real and imaginary numbers {1. 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. rotation by 180o . in fact. identify subgroups. 0 −1 1 0 (5. we arrive at the property exp ϑk LT k from which follows LT = −Lk .1. i. rotation by 270o } and consecutive execution of rotation as the binary operation ‘◦’.

. 0)T ) − 1 1 1 ϑ1 →0 (5.e. 0. One can determine then the matrices Lk defined in (5.15) as the derivatives of R(ϑ) with respect to the Cartesian components ϑk taken at ϑ = (0. Exercise 5. This property gives A then only three independent matrix elements. ϑ3 and three matrices L1 .24) (5.22). the real matrix A is anti-symmetric.        (5. lim R[(0. 0. 0)T . 0)T the rotation is the identity.2: Determine the generator L1 according to Eq. ϑ2 .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 . 0. i.23) (5.27) . A must have the form   0 a b A =  −a 0 c  . (5. ϑ3 )T ] = lim  sinϑ3 cosϑ3 0  =  ϑ3 ϑ3 →0 ϑ3 →0 0 0 1 0 0 1 Insertion into (5.1: Matrix Representation of SO(3) 101 We can conclude then that if elements of SO(3) can be expressed as R = eA .22) and similar for L2 and L3 .(5. 0.8) we can state    cosϑ3 −sinϑ3 0 1 −ϑ3 0 1 0  .25)     0 −ϑ3 0 0 −1 0  ϑ3 0 0  =  1 0 0  . L3 which are independent of the rotation angles. L2 .21) −b −c 0 It is this the reason why one can expect that three-dimensional rotation matrices can be expressed through (5. Using the definition of partial derivatives we can state L1 = lim ϑ−1 R(ϑ = (ϑ1 . ϑ3 )T such that for ϑ = (0. Using (5.1.26) (5. ϑ2 .15) with three real parameters ϑ1 . Let us use this definition to evaluate L3 . 0 0 0 0 0 0  (5.

From this result one can conclude that expressions of the type (5. B]. are related to angular momentum operators. the property (5.8) in case of ϑ = (0. An answer can be found considering the Baker-Campbell-Hausdorf expansion ∞ exp( λA ) exp( λB ) = exp( n=1 λn Zn ) (5. J ] = i k m Jm .28) The construction of the generators through the limit taken in Eq. In fact.22) implies. the elements of if any two indices are identical for k = 1. i. The question is then. L ] = k m Lm (5. In case of the group SO(3) the structure constants are particularly simple. Of course. however.e. 0.. B]. B] ).30) Groups with the property that their elements can be expressed like (5.15) and their generators obey the property (5. that the generators represent infinitesimal rotations with 1 >> |ϑ| around the x1 -.g. and the constants fk m are called the structure constants. (5. L ] = m=1 fk m Lm . We want to consider. = 2. and x3 -axes. 0.30) is called the Lie algebra. it must hold 3 [Lk . x2 -.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. Z3 = 12 ( [A. B]] + [[A.29) where Z1 = A + B and where the remaining operators Zn are commutators of A and B. (5. 2 . The expression for Zn are actually rather complicated. as we see later. however.30) are called Lie groups. 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.e. 0)T ) = exp( 1 L1 ). e. if the resulting products of exponential operators can be expressed in the form (5. (5. holds the algebra [Jk . the Lie algebra of SO(3) can be written [Lk . the matrix (5.g.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. m = 3 for any even permutation of k = 1. (5. 0)T ) = exp( 2 L2 ) and R(ϑ = (0. ϕ)T . [A. m = 3 .31) where k m are the elements which are   0   1 k m =  1   −1 of the totally antisymmetric 3-dimensional tensor.32) For the matrices Jk = i Lk which. that the operators (5. m = 3 for any odd permutation of k = 1. 3 )T ) = exp( 3 L3 ).15) obey the group property. commutators of commutators of A and B. Finite rotations can be obtained by applications of many infinitesimal rotations of the type R(ϑ = ( 1 . etc. Z2 = 2 [A. = 2. three being a special case. only the latter example and . 1 1 e. R(ϑ = (0. = 2. Lie groups can have any number of generators. i...15).33) We want to demonstrate now that exp(ϑ · L) yields the known rotational transformations. 0. in fact.

1. using (5. (5.37) = (−1)n 1 0 0 1 . eϕL3 which agrees with (5. 5. 0 −1 1 0 = (−1)n 0 −1 1 0 (5. (5.5.39) Recognizing the Taylor expansions of cos ϕ and sin ϕ one obtains eA = and.41) cos ϕ − sin ϕ sin ϕ cos ϕ (5. can be written ϕL3 = 0 103 A 0 0 0 0 .35).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 . ∞ 2n = − 1 0 0 1 2n+1 (5. (5.36) The property 0 −1 1 0 allows one to write 0 −1 1 0 and.31).  cos ϕ − sin ϕ 0 =  sin ϕ cos ϕ 0  0 0 1  (5. A = ϕ 0 −1 1 0 . Exercise 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 . according to (5. that the matrix ϕL3 . accordingly.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 . We note. (2n + 1)! (5.3: Show that the generators Lk of SO(3) obey the commutation relationship (5.40) .1: Matrix Representation of SO(3) determine exp(ϕL3 ).8).27).34.

e. Definition We first define how a rotational transformation acts on a wavefunction ψ(r ). the R(ϑ) ψ(r) = ψ(R(ϑ) r ). (5. i. complex-valued functions over the 3-dimensional space R which can be differentiated infinitely often. for A. In particular.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). In analogy to R(ϑ) being a linear map R → R ..46) (5.42) Obviously.e. hence. Since this holds for any ψ(r ) one can conclude ρ R(ϑ1 ) ρ R(ϑ2 ) group property (5. we will learn that transformation patterns and invariances are connected with the angular momentum states of quantum mechanics.e. = ψ([ R(ϑ1 ) R(ϑ2 ) ]−1 r) = ρ R(ϑ2 ) ψ([R(ϑ1 )]−1 r) (5. the transformation R(ϑ): C∞ ( ) → C∞ ( ) is related to the transformation R(ϑ): R → R . we define rotations R(ϑ) as linear maps C∞ ( ) → C∞ ( ). in fact. namely R(ϑ)ψ(r ) = ψ(R−1 (ϑ) r ) .2. i. Exercise 5. B ∈ SO(3) holds ρ(A B) = ρ(A) ρ(B) . Show that in this case holds Generators One can assume then that R(ϑ) should also form a Lie group. (5.104 Theory of Angular Momentum and Spin 5.45) (5. i.44) holds. one isomorphic to SO(3) and.48) . a relationship which we like to express as follows R(ϑ) = ρ R(ϑ) .43) ρ( ) conserves the group property of SO(3).47) = ρ R(ϑ1 ) R(ϑ2 ) .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) . (5.1: Assume the definition ρ ρ (A B) = ρ (B) ρ (A). its elements can be represented through an exponential R(ϑ) = exp i − ϑ·J (5. For this purpose we require stringent continuity properties of the wavefunction: the wavefunctions under consideration must be elements of the set C∞ ( ).

55) and f (r) = f [J1 . k = 1. 5.56) We like to verify this property for [J1 . (5. J3 ] and [J3 . J2 ].57) Exercise 5.53–5. 0. Carrying out similar calculations for J1 and J2 one can derive i − J1 i − J2 i − J3 = = = x3 ∂2 − x2 ∂3 x1 ∂3 − x3 ∂1 x2 ∂1 − x1 ∂2 .53) (5. The generators Jk can be determined by applying (5. In fact.2: (a) Derive the generators Jk . 2 by means of limits as suggested in Eq.24) one can expand  ϑ3 →0 ϑ3 →0 lim ϑ−1 R(ϑ = (0.49).49) (note the minus sign of −ϑ1 which originates from the inverse of the rotation in Eq. . show that (5. ϑ3 )T ) f (r) − f (r) 3 (5.  0 0 1 x3 x3 (5. (5. in case of J3 . 0. One obtains using (5. J2 ] f (r) = − = − 2[ 2 [− i J .2. J ] = i k m Jm . i.5.e. − 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) .50) lim ϑ−1 ( f ( R(0. (5. 3     1 ϑ3 0 x1 x1 + ϑ3 x2 R(0.23.50) 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.(5.2: Function Space Representation of SO(3) where the generators can be determined in analogy to Eq.42)) and similarly for J2 and J3 .54) (5.55) Not only is the group property of SO(3) conserved by ρ( ). −ϑ3 ) r ) − f (r) ) . −ϑ3 ) r ≈  −ϑ3 1 0   x2  =  −ϑ3 x1 + x2 .49) to a function f (r).. (5. J1 ].51) Inserting this into (5.22) according to i − J1 = lim ϑ−1 R(ϑ = (−ϑ1 . 0.52) ( x2 ∂1 − x1 ∂2 ) f (r) . it holds [Jk . 0. (5.56) holds for [J2 . but also the Lie algebra of the generators. 0)T ) − 1 1 ϑ1 →0 105 (5. i − J3 f (r) = = Using (5.

φ) (5.e.65) .61) We demonstrate this property for k = 3. according to the correspondence principle between classical and quantum mechanics. identifies J as the quantum mechanical angular momentum operator. (5. J3 ] + [J2 . Jk ] = 0 .e. Using [AB. the operator 2 2 2 J 2 = J1 + J2 + J3 (5. using the momentum operator p = ˆ J = r × i . These eigenstates are the well-known spherical harmonics Y m (θ.3: Consider a wave function ψ(φ) which depends only on the azimutal angle φ of the spherical coordinate system (r.55) can be written as a vector product of r and i − J = −r × From this follows.63) (5. J3 ] + [J1 .64) mY m (θ. To show this we first note that (5. [J 2 . J3 ] J1 + J2 [J2 .61) simultaneous eigenstates of J 2 and of one of the Jk . θ. C] + [A. usually chosen as J3 . The commutation relationships imply that no simultaneous eigenstates of all three Jk exist. φ).. can be found. φ) = = 2 ( + 1) Y m (θ. 2.62) J1 [J1 . 2.53–5. .56) are the famous commutation relationships of the quantium mechanical angular momentum operators. C]B and (5.58) i = r ×p ˆ (5.2. and x3 = r cosθ. φ) as well as the effect of the so-called raising and lowering operators J± = J1 ± i J2 (5. i. φ) m (θ. 5. Show that for J3 as defined above holds exp( iα J3 ) ψ(φ) = ψ(φ + α).59) which. k = 1. J3 ] (5. However.60) commutes with all three Jk . x2 = r sinθ sinφ. Equation (5. 3 . φ).56) one obtains [J 2 .3 Angular Momentum Operators The generators Jk can be readily recognized as the angular momentum operators of quantum mechanics. J3 ] = = = 2 2 [J1 + J2 . 3. We will show now that the properties of spherical harmonics J2Y J3 Y m (θ. (5. the system related to the Cartesian coordinates through x1 = r sinθ cosφ.106 Theory of Angular Momentum and Spin Exercise 5. k = 1. C] = A[B. J3 ] J2 −i J1 J2 − i J2 J1 + i J2 J1 + i J1 J2 = 0 According to (5. i.

k = 1.71) To prove this theorem we first show by induction that (i) holds. and since we have already shown | − 1 = (iβ −1 )−1 L− | we can state L3 | − 1 = −i( − 1)| − 1 . i. We will show that this property holds then also for m − 1. The definitions of the coefficients α −1 and β −1 yield L+ L− | = L+ ( iβ −1 | − 1 = − α −1 β −1 | . (∗) (5. 3 be operators acting on a Hilbert space H which obey the algebra [Lk . 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. The arguments are very similar to the ones used above and we can be brief.75) (5. L− ] + L− L3 ) | . Noting [L+ .76) We now show that property (ii) also holds for m = − 1 proceeding in a similar way. J+ Y J− Y J− Y m (θ. L1 − iL2 ] = L2 + iL1 = i (L1 − iL2 ) = iL− and L3 | = −i | we obtain L3 L− | = ( iL− + L− L3 )| = i(− + 1) L− | .56). We note L3 L− | = ( [L3 . φ) 107 = = 0 = = 0 ( + m + 1)( − m) Y ( + m + 1)( − m) Y m+1 (θ. L+ raises the m-value of | − 1 from m = − 1 to m = .74) (5. L1 − iL2 ] = −2iL3 and L+ | = 0 one obtains L+ L− | = −2iL3 | = −2 | .72) = 0 = −i | (5. namely. φ) − m (θ. L− ] + L− L+ )| . An Important Theorem Theorem 1.73) (5.. L2 = L2 + L2 + L2 . φ) (5. α −1 β −1 = 2 . L− ] = [L3 . For this purpose we prove the following theorem. L− ] = [L1 + iL2 . For this purpose we first demonstrate that the property holds for m = − 1 by considering (i) for the state L− | ∼ | − 1 . L− ] + L− L+ )| m = ( −2iL3 + .67) (5.66) (5. Using [L3 . 2.68) (5. in fact.e. φ) are a consequence of the Lie algebra (5. Continuing our proof through induction we assume now that property (i) holds for m.1 Let Lk .69) J+ Y (θ. It holds L+ L− | = (L+ L− − L− L+ + L− L+ )| = ([L+ . i.e. We consider L+ L− | m = ( [L+ . φ) (θ.5. L ] = k m Lm and let L± = L1 ± i L2 .70) (5. φ) m+1 (θ. From this follows that L− | is an eigenstate of L3 .3: Angular Momentum Operators on the spherical harmonics.

(5. 5. For this purpose we need to express the angular momentum operators in terms of spherical coordinates (r.77) We will show now that if (ii) holds for m. We note L3 L− | m = ( [L3 . 5. 5.e. We note that we can write L2 = 1 1 1 1 2 2 2 2 L+ L− + 2 L− L+ + L3 .79) We can now show that (iii) holds for all proper m–values.84) . A first application will involve the construction of the eigenstates of the angular momentum operators as defined in (5. i. 2 We note that Eqs.e. This implies that (i) holds for m − 1. in terms of spherical coordinates defined through x1 x2 x3 = = = r sin θ cos φ r sin θ sin φ r cos θ (5.53–5. it also holds for m − 1. 5. 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 .64).69). 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 .55).. θ. (5. φ. it holds αm βm = 2 one obtains αm βm = = ( + 1) k=m+1 k . Before we can finally show that property (iii) holds as well for all − ≤ m ≤ we need to determine the coefficients αm and βm .66.80) (5.78) ( + m + 1)( − m) ( + m + 1)( − m) .83) (5.63.79) read L+ | m L− | m + 1 = −i = −i ( + m + 1)( − m) | m + 1 ( + m + 1)( − m) | m .73. Using the familiar formula + m − n k=0 k = n (n + 1)/2 − (m + 1) m = ( + 1) m − (m + 1) m (5. Again the argumemts are similar to the ones used above. yield properties (5. We will have various opportunities to employ the Theorem just derived. 5. We can deduce from (∗) and (∗∗) α α α −1 β −1 −2 β −2 −3 β −3 = 2 = 2( − 1 + ) = 2( − 2 + · · −1 + ) Obviously.82) (5. Angular Momentum Operators in Spherical Coordinates We want to express now the generators Jk . One can normalize the states | m such that αm = βm . finally αm = βm = (5.81) i. from L+ L− | m = − αm−1 βm−1 | m follows the recursion relationship αm−1 βm−1 = 2m + αm βm . in particular.70..108 Theory of Angular Momentum and Spin L− L+ )| m = (−2m − αm βm )| m .72. given in (5. This completes the proof of the theorem. (∗∗) (5.

85). using ∂/∂tanφ = cos2 φ ∂/∂φ and ∂/∂cosθ = −(1/sinθ) ∂/∂θ.87) To derive these properties we consider.86). φ) ∂x ∂x3 ∂ tan φ =0 (5. φ) ∂x2 /x1 ∂ ∂x3 /r ∂ = x3 + x3 ∂x2 ∂ tan φ ∂x2 ∂ cos θ ∂x2 /x1 /r ∂ ∂ − x2 − x2 ∂x3 3 ∂ cos θ f (θ.85). the latter defined in (5. φ) (5. using (5.82–5.3: Angular Momentum Operators 109 We first like to demonstrate that the generators Jk .60).93) and similarly. We like to express now the operators J3 and J 2 . In fact. ∂φ = sin φ (5. for example. L1 f (θ. φ) = (x3 ∂2 − x2 ∂3 )f (θ. (5.90) This yields. (L2 )2 = − cos φ ∂ ∂ + cot θ sin φ ∂θ ∂φ 2 .89) = ∂ x2 x2 ∂ x2 + x2 ∂ x3 − 33 − x2 1 3 2 x1 ∂ tan φ r ∂ cos θ r ∂ cos θ f (θ. Using (5.85) (5. φ) sin θ ∂θ sin θ ∂θ L1 f (θ.91) which agrees with expression (5. in terms of spherical coordinates. actually. φ) = ( x3 ∂2 − x2 ∂3 ) f (θ. φ) = (5. applying repeatedly the chain rule. cos θ ∂ cos2 φ + sin θ cos φ ∂φ cos2 θ sin θ sin φ ∂ sin θ sin φ ∂ + + sin2 θ f (θ.84 ) one obtains. 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 ∂ = − . involve only the angular variables θ and φ and not the radius r. According to (5.92) (5. using (5.87) holds J3 = To determine J 2 we note. L1 f (θ. φ).86) (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 .5.88) (5.

98) where Jclass = r × mv is the angular momentum and pr = mr is the radial momentum. 5.85. 2m r ∂r2 2m r2 2 (5.96) and the relationship between Lk and Jk as given in (5.101) 1 ∂2 r r ∂r2 and comparision with (5.94) It follows (L1 )2 + (L2 )2 + (L3 )2 = With cot2 θ + 1 = 1/sin2 θ.95) sinθ ∂ ∂θ = ∂2 ∂ + cotθ .87).4 Angular Momentum Eigenstates According to the theorem on page 107 above the eigenfunctions (5. 5.99) = = ∂2 2 ∂ 1 + + 2 2 2 ∂r r ∂r r sin θ ∂2 2 ∂ + 2 ∂r r ∂r sinθ ∂ ∂θ 2 + ∂2 ∂φ2 (5. The operators in the present case are Lk = − i Jk where the Jk are differential operators in θ and φ given in (5. sphere S2 . These generators act on a subspace of C∞ ( ). One first identifies the Hilbert space H on which the generators Lk operate. 5.63. The functions in C∞ (S ) have real variables . 5.86. ∂θ ∂φ ∂2 ∂ ∂2 + cotθ + ( cot2 θ + 1 ) 2 .87).97). one can conclude J2 = − 2 sin θ 2 sinθ ∂ ∂θ 2 + ∂2 ∂φ2 . 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 + .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. 5.100) (5.85.e. namely on the space of complex-valued functions on the unit 3-dim.64) can be constructed as follows.86. 2 ∂θ ∂θ (5. ∂θ2 ∂θ ∂φ (5. i.97) Kinetic Energy Operator The kinetic energy of a classical particle can be expressed J2 p2 p2 = r + class 2m 2m 2m r2 (5. 5. C∞ (S ). 1 sin2 θ sinθ ∂ ∂θ (5.

103) iteratively for m = − 1. } applying L− Y m+1 (θ. φ) in H which satisfy L+ Y (θ. φ) = ∆( . can be expressed L± = Accordingly.87). .104) (5. φ) = 0 as well as L3 Y (θ. 0 ≤ φ < 2π. (5. φ) L3 Y (θ. φ). The eigenfunctions (5.63. φ) . (5. 5. The raising and lowering operators L± = L1 ± i L2 . is indeed a Hilbert space.105) explicitly.111) . (5. The function space endowed with this norm and including only functions for which the integral (5. − + 1. φ) = = 0 −i Y (θ.64) can be constructed then by seeking first functions Y (θ. φ) = −i Y (θ. . m = − . the so-called spherical harmonics. . and to be admissable for description of quantum states. φ)f (θ. φ). normalizing these functions. . φ) = −i ( + m + 1)( − m)Y m (θ. φ) (5. − 2.107) reads ∂ ∂ + i cotθ ∂θ ∂φ P (cosθ) ei φ i e±iφ ∂ ∂ ± i cotθ ∂θ ∂φ .102) where dΩ = sinθ dθ dφ is the volume element of S2 .109) using (5. . . (5. by carrying out the construction according to (5.104.107) (5. φ).106) The latter property is obviously satisfied for the chosen φ-dependence.85–5.105) ∆( . φ) = 2 + 1 ( − m)! P 4π ( + m)! m (cosθ) eimφ . 5. φ. m) = 1 −m J− 1 2 Y (θ.4: Angular Momentum Eigenstates 111 θ.5. − .110) = 0 (5. 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. 0 ≤ θ < π. φ) (5. and then determining the family {Y m (θ. .108) m (θ. must be cyclic in φ with period 2π. The norm in H is defined through |f |2 = S2 dΩ f ∗ (θ. m) Constructing Y → Y →→ ··· Y ( + m)! (2 )! ( − m)! −1 − We like to characterize the resulting eigenfunctions Y m (θ. φ) (5. One obtains in this way Y m (θ. (5. .102) exists.

New York.120) See. one for which π dθ sinθ |P (cosθ)|2 (5. 5..119) provide the following expression for Y Y (θ. The normalization factor N is chosen such that π (5. We note that (5.e. for example. φ) = (−1) 1 1 = 1 [1 · 3 · 5 · · · (2 − 1)]2 (5.119) where the factor (−1) has been included to agree with convention1 . (5. 5. using the new integration variable x = cos θ. (5.117) Accordingly.106.113) 0 is finite. i. . ”Classical Electrodynamics.D.114) dθ sinθ |Y (cosθ)|2 = 1 0 (5. π +1 dθ sin2 0 +1 θ = −1 +1 −1 dx (1 − x2 ) +1 = = .112 or Theory of Angular Momentum and Spin ∂ − ∂θ cotθ P (cosθ) = 0 (5. Jackson (John Wiley.118) 2 +1 1 1 sin θ ei 4π (2 )! 2 ! φ .114. . (5.115) holds.106. 5.107).116) The integral appearing in the last expression can be evaluated by repeated integration by parts. 2nd Ed.114) yield |N |2 2 +1 1 2π 4π (2 )! π dθ sin2 0 +1 θ = 1. One obtains.” by J. (5. = = x (1 − x2 ) 2 x3 3 + 2 −1 +1 −1 −1 dx x2 (1 − x2 ) + 2 · (2 − 2) 1·3 −1 +1 (1 − x2 ) dx x4 (1 − x2 ) −1 −2 +1 2 · (2 − 2) · · · 2 dx x2 1 · 3 · 5 · · · (2 − 1) −1 2 · (2 − 2) · · · 2 2 (2 )! 2 = 2 2 +1 1 · 3 · 5 · · · (2 − 1) 2 + 1 [1 · 3 · 5 · · · (2 − 1)] (5. 1975) .116) implies |N |2 from which we conclude N = (−1) 1 · 3 · 5 · · · (2 − 1) (5. A proper solution of this equation.112) where we employed the definition of P (cosθ) in (5. is P (cos θ) = N sin θ as one can readily verify.

. we seek to determine the functions P m (cos θ) and. and can obtain now.119). φ).125) which agrees with (5. m (cos θ) e (5. namely Y (θ. according to (5. the eigenfunctions Y m (θ. describes the φ-dependence of Y m (θ.122) The latter identity can be written.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 .64). φ). P m (x) = −1 ( + m)! 2 ! ( − m)! × (1 − x2 )− m+1 2 −(1 − x2 ) 2 1 ∂ x + (m + 1) 1 ∂x (1 − x2 ) 2 × (5.124) called the associated Legendre polynomials. 5.124) holds for m + 1. For m = (5. − defined in (5. . We want to demonstrate now that the recursion equation (5.e. The reader should note that the associated Legendre polynomials. P m (x) = −1 × ( + m + 1) ( − m) × −(1 − x2 ) 2 1 (5.123).106). as specified in (5.103.123) P m+1 (x) ∂ x + (m + 1) 1 ∂x (1 − x2 ) 2 . .110).123) leads to the expression P m (x) = −m m ∂ 1 ( + m)! (1 − x2 )− 2 (x2 − 1) 2 ! ( − m)! ∂x −m (5. Actually.63. 5. use (5. therefore. − 2. (5.4: Angular Momentum Eigenstates 113 We have constructed a normalized solution of (5.124) we proceed by induction.110) together with (5. 5.5. Let us then assume that (5. P m+1 (x) = −m−1 m+1 ∂ 1 ( + m + 1)! (1 − x2 )− 2 (x2 − 1) .121) This expressions shows that the factor eimφ .127) ∂ −m−1 (x2 − 1) .124). indeed. employing again the variable x = cos θ.124) reads P (x) = = (2 )! (1 − x2 )− 2 (x2 − 1) 2 ! (−1) 1 · 3 · 5 · · · (2 − 1) sin θ (5.126) Then holds.. (5. 5.114. m = − 1. φ). every application of L− reduces the power of eiφ by one. To prove (5.103. i. 2 ! ( − m − 1)! ∂x −m−1 (5. are real.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) . ∂x −m−1 . through repeated application of (5.

φ) = Y − (θ. 5.128) one can show that (5. therefore..132). appears not very interesting. which is equivalent to recursive application of L− . φ) = (−1)m Y m −m (θ. φ) (5. However.127) reproduces (5. φ) as given in (5. Y − +2 .124).124) holds for m if it holds for m + 1. it holds ∂2 (x2 − 1) = (2 )! ∂x2 and.124).129) holds.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. φ) .123) terminates for m = − .131) identifies f (θ. if L− Y (θ. In that case (1 − x2 ) is a polynomial of degree 2 and. 2 ! (2 )! ∂x2 1 (5.134) (x) = 1 2 ! (1 − x2 ) 2 . φ) An alternative route to construct the eigenfunctions Y of L− f (θ. Constructing Y − → Y − +1 → ···Y m (θ. We want to test if the recursion (5. therefore. We first determine Y − (θ. φ) L3 f (θ. φ) using (5.132) which reduces the number of spherical harmonics which need to be evaluated independently roughly by half . i. the derivative ∂ 2 +1 /∂x2 +1 of this expression vanishes. we embark on this construction in order to prove (5. that expression (5. (5. 5. etc. therefore.135) .124) and. φ) choosing the proper sign. using x = cos θ.124) holds for m = . therefore. It holds. Y ∗ (θ. Since (5. yields a vanishing expression for m = − − 1 as long as is a non-negative integer.133) Since the term with the highest power of (x2 − 1) is x2 . namely.e. P − (x) = 1 ∂2 (1 − x2 ) 2 (x2 − 1) .106. hence. φ). from such construction emerges an important symmetry property of Y m (θ. expression (5.106. and constructs then the eigenfunctions Y − +1 .130) (5. it holds then for all m. by repeated application of the operator L+ . φ) = = determines first a normalized solution 0 i f (θ. (5. φ) = 0 (5. Such construction reproduces the eigenfunctions Y m (θ. P − (5.124) and. In fact.

110) applies the = ( + m + 1)( − m) Y m+1 (θ. φ). (5.141) We want to demonstrate now that this recursion equation leads to the expression P m (x) = +m m ∂ (−1)m (1 − x2 ) 2 (x2 − 1) 2 ! ∂x +m (5.136) is normalized and has the proper sign.129).132).122. 5. One can readily verify that this expression provides a solution of (5. This follows from the identity ∂ ∂ − i cotθ ∂θ ∂φ One can use (5.f. one can conclude for the associated Legendre polynomials ( − m)! ( + m)! = or P m+1 (cos θ) ∂ − m cotθ ∂θ P m (cos θ) ( + m + 1)( − m) ( − m − 1)! P ( + m + 1)! m+1 (cos θ) (5. According to (5. (5.e.143) . (5. We first note that for m = − (5.4: Angular Momentum Eigenstates Due to (5. φ) = 2 +1 1 1 sin θ e−i 4π (2 )! 2 ! φ 115 − . (5.124) one arrives at Y (θ. (5.136) to construct all other Y recursion eiφ ∂ ∂ + i cotθ ∂θ ∂φ Y m (θ. φ) sin θ e−i φ = 0. the expression (5. 5.142) yields P − (x) = (−1) 1 (1 − x2 )− 2 (x2 − 1) = (1 − x2 ) 2 2 ! 2 ! (5.139) = ∂ − m cotθ ∂θ P m (cos θ) .138) Employing (5. φ) . i. Obviously. a sign consistent with the family of functions Y m constructed above.136) We note in passing that this expression and the expression (5. (5.80) and (5.124).140) Introducing again the variable x = cos θ leads to the recursion equation [c.137) m (θ.106).123)] P m+1 (x) = −(1 − x2 ) 2 1 ∂ x − m 1 ∂x (1 − x2 ) 2 P m (x) . following closely the proof of eq.106.. (5.120) for Y obey the postulated relationship (5.5.142) For this purpose we proceed by induction.

148) According to (5. P (x) = P 0 (x) are called Legendre polynomials. We then assume that (5. 2 ! ∂x +m+1 (5. given by (5.h..145) Hence.142) and by (5. of which agrees with the r.s.142) one can state P (x) = ∂ (x2 − 1) . one notes that application of (5.147) the r. Hence.142). However.142) holds for m + 1 if it holds for m.132). equivalently.141) reads then P m+1 (x) = 1 (5. Y −2 .144) reads P m+1 (x) = +m+1 m+1 ∂ (−1)m+1 (1 − x2 ) 2 (x2 − 1) .124) and (5. 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 . (5.146) i.135).124). (5.150) . 1 (1 − x ) ∂x +m (1 − x2 ) 2 1 (5. etc.106) this implies for Y The Legendre Polynomials The functions the identity (5. (5.s..144) ∂ x − m 1 ∂x (1 − x2 ) 2 +m m ∂ (−1)m (1 − x2 ) 2 (x2 − 1) . 2 ! ∂x 1 (5. etc. i.116 Theory of Angular Momentum and Spin wich agrees with (5.e. 2 ! ∂x +m (5. According to both (5. are identical. one can conclude P m (x) = (−1)m m (θ.149) (5. Y − +2 . Since (5.142) holds for m. of (5.124) yields (−1)m ( − m)! P ( + m)! −m (x) = +m m ∂ (−1)m (1 − x2 ) 2 (x2 − 1) . (5.h.142) holds for m = − we verified that it holds for all m. Accordingly.128) m + 1 → −m or. also the associated Legendre polynomials determined this way. yields the same eigenfunctions as the previous construction beginning with Y and stepping down the series of functions Y −1 . φ) ( − m)! P ( + m)! −m (x) .e. The construction beginning with Y − and continuing with Y − +1 . 2 ! ∂x +m −(1 − x2 ) 2 Replacing in (5.

156) yields P (1) = 1 .158) see. r1 is the point where the potential is measured. (5. =0 (5. and r2 denotes the location of the charge.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) .132). In case q = 1 and |r2 | < |r1 | holds the identity2 ∞ 1 r2 = P (cos γ) (5.150) and (5.5.156) w(x.155) P (x) t . 2nd Ed. 2 3 1 1 = = 2 1−t 1 − 2t + t ∞ t . (5. 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..153) P (cos θ) eimφ .” by J.154) can be written 1 = w(x. The generating function allows one to derive useful properties of the Legendre polynomials. P3 (x) = 1 2 117 P2 (x) = − 3x) 1 2 2 (3x − 1) . for example. t) = √ 1 − 2xt + t2 ∞ 1 − 2 x t + t2 . 1975).. For example. New York. 92 To prove this property see. φ) = 2 + 1 ( − m)! (−1)m sinm θ × 4π ( + m)! × ∂ ∂ cos θ m (5.152) and. accordingly. Jackson (John Wiley. t) is called a generating function of the Legendre polynomials3 . Lebedev (Prentice-Hall. . Using x = cos γ. 45 of “Special Functions and their Applications” by N. Englewood Cliffs. ”Classical Electrodynamics. m ≥ 0 ∂xm (5. 1965) which is an excellent compendium on special functions employed in physics. t = r2 /r1 . .154) |r1 − r2 | r +1 =0 1 where γ is the angle between r1 and r2 .N. P1 (x) = x . the spherical harmonics for m ≥ 0 Y m (θ. 2. t) = √ Comparision with (5.157) = 0. 1. N.151) (5x3 Comparision of (5. . in case of x = 1 holds w(1. The spherical harmonics for m < 0 can be obtained using (5.g. and |r1 − r2 | = r1 (5. pp. e.J.4: Angular Momentum Eigenstates The first few polynomials are P0 (x) = 1 . . (5. pp. =0 (5.D.

159) by w(1 − 2xt + t2 ) leads to the differential equation obeyed by w(x. φ) . θ. φ) goes over to Y m (π − θ. φ) m (θ.f. under inversion Y m (θ.166) We want to summarize the properties of the spherical harmonics Y derived above. 2. π + φ) = (−1) Y m (θ.161) Collecting coefficients with equal powers t yields 0 = P1 (x) − x P0 (x) ∞ (5. ∂t 1 − 2xt + t2 (5. Accordingly. (5. (5. . holds P1 (x) ( + 1) P +1 (x) = = x P0 (x) (2 + 1) x P (x) − P −1 (x) (5. φ) (5. ∂t P (x)t −1 −1 (5. using P0 (x) = 1 [c. t) holds Theory of Angular Momentum and Spin ∂lnw(x. .118 For w(x.164) which. Due to cos(π − θ) = − cos θ the replacement r → −r alters P m (x) into P m (−x). Inspection of (5. Accordingly. then the coordinates of − r are (r. φ) m (θ.151)] allows one to determine P (x) for = 1. . φ) = = 2 ( + 1) Y m (θ.. π + φ). φ). π + φ).163) (5.142) allows one to conclude P m (−x) = (−1) +m P m (x) (5. φ) . π − θ.162) − (2 + 1)x P (x) + P −1 (x) ] t + =1 [( + 1) P +1 (x) The coefficients for all powers t must vanish individually.168) mY m (θ. Noting exp[im(π + φ)] = (−1)m exp(imφ) we determine Y Properties of Y m (θ.167) (5.160) this differential equation is equivalent ∞ (1 − 2xt + t2 ) =0 P (x) t + (t − x) =0 P (x) t = 0 (5. t) x−t = .159) Using ∂lnw/∂t = (1/w) ∂w/∂t and multiplying (5. φ) Under inversion at the origin vectors r are replaced by − r. 1. Inversion Symmetry of Y m (θ. φ) m (π − θ.156) and (∂/∂t) to P (x)t = ∞ ∂w + (t − x) w = 0 . . The spherical harmonics are eigenfunctions of the angular momentum operators J2Y J3 Y m (θ. t) (1 − 2xt + t2 ) Employing (5. If the spherical coordinates of r are (r.165) since ∂ n /∂(−x)n = (−1)n ∂ n /∂xn .

P −1 (x) ] (5. The spherical harmonics form. 4π (5. m 4. 2.173) m+1 (θ.172) (5.) P0 (x) = 1 . φ) 2π 0 (5. The spherical harmonics are given by the formula Y m (θ. φ) = = ( + m + 1)( − m) Y ( + m + 1)( − m) Y m+1 (θ. φ) = δ δm m . 5.e.4: Angular Momentum Eigenstates 119 2. φ) ∈ C∞ (S2 ) holds ∞ f (θ.5. where ( = 1. P +1 (x) P1 (x) = x . φ) C = =0 m=− π C sin θdθ 0 m Y m (θ. . for any f (θ. (5. φ) = (−1)m Y ∗ (θ. φ)f (θ. φ) (5.. φ) . The spherical harmonics obey the recursion relationships J+ Y J− Y m (θ. a complete basis of C∞ (S2 ). The spherical harmonics for m < 0 are given by Y −m (θ. The spherical harmonics Y 0 are Y 0 (θ. . i. uniquely defined functions over the unit sphere S2 which are infinitely often differentiable π 2π sin θdθ 0 0 dφY ∗m (θ. φ) . φ) where J± = J1 ± iJ2 . φ) = 2 +1 P (cos θ) .179) . φ) = 2 + 1 ( − m)! (−1)m sinm θ × 4π ( + m)! 2 ! × ∂ ∂ cos θ m (5.177) Note that the spherical harmonics are real.170) (5. (5. 6.175) (5. except for the factor exp(imφ). .171) m = dφ Y ∗ (θ. For the Legendre polynomials holds the orthogonality property +1 dx P (x) P ( (x) = −1 2 δ 2 +1 . The spherical harmonics form an orthonormal basis of the space C∞ (S2 ) of normalizable. φ) Y m (θ. m (5. φ) m (θ.176) = 1 +1 [ (2 + 1) x P (x) − are the Legendre polynomials.178) 7.169) 3.174) m ≥ 0 P (cos θ) eimφ . in fact.

φ) = δm0 2 +1 . .183) (5. t) stated in (5.4. For the Laplacian holds 2 = = = (5.h.2: Derive the orthogonality property for the Legendre polynomials (5. expand the result in powers of t and equate the resulting powers to those arising on the r. 4π (5.181) = 0.120 8. The spherical harmonics for m (θ.182) Y11 Y10 = = − (5. For this purpose start from the identity +1 ∞ +1 dx w(x. φ) . The spherical harmonics for θ = 0 are Y m (θ Theory of Angular Momentum and Spin = 0. φ) = (−1) Y m (π − θ.87).h.86).185) (5.189) evaluate the integral on the l. (5. The spherical harmonics obey the inversion symmetry Y 10.180) 9.4.179) using the generating function w(x. φ). (5.184) Y22 Y21 Y20 together with (5.188) Exercise 5. =0 t + −1 dxP (x) P (x) .s.. (5. (5. Exercise 5.s. 2 are given by Y00 = 1 √ 4π 3 sin θ eiφ 8π 3 cos θ 4π 1 4 − 15 sin2 θ e2iφ 2π 15 sin θ cos θ eiφ 8π 5 4π 3 1 cos2 θ − 2 2 (5.1: Derive expressions (5.4.177). π + φ) .187) h(r) Y m (θ.186) (5.189).156). Exercise 5. 1. t)2 = −1 . of (5.3: Construct all spherical harmonics Y3m (θ. 11. φ) = 1 ∂2 r − r ∂r2 ( + 1) r2 h(r) Y m (θ.

and (ii) they form the lowest dimensional sets X obeying C∞ (S ) = X which have this invariance property..e. if one specifies the functional form of the coefficients [D(ϑ)] m . ∞. ˜ f (θ. exp − i ϑ · J . (5. (5.− + .190) provide a complete. . }] . = − . . 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 .. ..196) C∞ (S ) = = X . = .195) i. .194) Y m (θ.193) where [ { } ] denotes closure of a set by taking all possible linear combinations of the elements of the set. m = S2 i dΩ Y ∗m (θ. . We denote the image of a function f (θ.193) shows ∞ = − .. called the irreducible representation.. i. .. exp − i ϑ · J (X ) = X . φ). with elements given by (5.m m m (θ. i. (5. .. φ) i − ϑ·J f (θ. m Y m (θ. m . For this purpose we consider the subspaces of C∞ (S ) X = [ {Y . φ) D(ϑ) = . φ) c = . on functions f (θ.. are characterized completely if one specifies the transformation of any Y m (θ. φ) exp − ϑ · J .. φ) ˜ under such rotational transformation by f (θ. }] (5.195)..e.197) The subspaces X have the important property that (i) they are invariant under rotations exp − i ϑ · J . φ) ˜ Y m (θ. φ ) (5. . (5. − + 1. We like to argue that the matrix representing the rotational transformations of the function space C∞ (S ). . . . therefore. orthonormal basis for the function c m Y m (θ. .192) = S2 dΩ Y ∗ (θ .5. φ ) exp m One can also represent C∞ (S ) C∞ (S ) = [ {Y . .42). φ) ∈ C∞ (S ). φ) m. Comparision with (5. φ) = exp Since the spherical harmonics Y space C∞ (S ) one can expand ˜ f (θ. m = − .e. ∞. .m D(ϑ) m.. The transformations exp − i ϑ · J . = 0. .191) i − ϑ·J f (θ . φ) (5.5: Irreducible Representations 121 5. φ) (5. assumes a particularly simple form.− + .. φ) . 1. = . }. .5 Irreducible Representations We will consider now the effect of the rotational transformations introduced in (5.

e. with blocks of dimension 1. If one orders the basis according to the partitoning (5.e. produces 0.   .201) From expression (5.202) D(ϑ) =           (2 + 1) × (2 + 1)               . i.199) [J+ . equivalently.200) leaves X invariant. leaves X invariant.194) only spherical harmonics with or.. One = contribute can conclude then that in the expansion (5. the transformation exp − i ϑ · J leaves X invariant as well.   1×1         3×3           .. φ) .. one concludes for the matrix elements (5. Similarly. 5.197) of C∞ (S ) the matrix representation of D(ϑ) assumes a block-diagonal form. One proceeds by expanding first ϑ · J in terms of J+ .. one can conclude that J+ leaves X invariant. one needs to consider the action of J+ 1 J− 2 J3 3 on the subspaces X . can be expanded in powers One employs then the commutation properties [J3 .195) [D(ϑ)] m. The exponential operator exp − i ϑ · J + ϑ3 J3 )n .199) allows one to express exp i − ϑ·J = n1 .200) in (5. i. J− 2 also leaves n1 X invariant.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 . . in case m = − . (5. m = δ S2 dΩ Y ∗ (θ.. Since any additive term n n n cn1 . .n2 . 5.     (5. which follow readily from (5.n3 n n n Accordingly.n2 . .. J− ] = 2 J3 . Since the Y m are n3 eigenfunctions of J3 the action of J3 reproduces the basis functions of X . J− . .n3 J+ 1 J− 2 J3 3 .198) 2 2 where ϑ± = ϑ1 ( 1 ϑ− J+ 2 + 1 2 ϑ+ J− ϑ2 .e. (5. 3. n The action of J− on Y m produces Y m−1 or. J± ] = ± J± (5. and J3 1 1 ϑ · J = ϑ− J+ + ϑ+ J− + ϑ3 J3 (5. Application of (5. i.56.200) 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. .65).

3 certainly allow one to describe any rotation around the origin. f (α + 2π) = f (α).6: Wigner Rotation Matrices 123 This representation of D(ϑ) is the one wich has blocks of the lowest dimensions possible. (ii) a second rotation around the new x2 –axis by an angle β. other representations are termed reducible representations. 5. (d) Show that the similarity transformation T (ϕ) defined in the space of non-singular. (b) Prove that the elements of SO(2) can be completely characterized through a single parameter.e. 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. However. (e) In the space C∞ ( ) of infinitely often differentiable. and for which detR = 1. (f) The transformation ρ(ϕ) as defined in (e) leaves the following subspace of functions considered in (e) Xm = { f (α) = A e−imα . 2.1: Representations of the group SO(2) SO(2) is the set of all 2×2 matrices R which are orthogonal.5.e. Exercise 5. (iii) a third rotation around the new x3 –axis by an angle γ. Determine the generator of ρ(ϕ) in analogy to (5. Give the corresponding expression of ρ(ϕ).5. This step reconsiders the parametrization of rotations by the vector ϑ assumed sofar. 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). this parametrization. (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). 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). ∈ N} invariant. does not provide the simplest mathematical description of rotations.6 Wigner Rotation Matrices We will now take an important first step to determine the functional form of the matrix elements of D(ϑ). and periodic functions f (α). though seemingly natural. 5. (a) Show that SO(2) with the group operation ◦ defined as matrix multiplication. The representation is referred to as the irreducible representation. for which holds RT R = R RT = 1 1.49). A ∈ C. k = 1. i. The three components ϑk . .22. is a group. i.

211) Using (5. m i Y m (θ.208) αJ3 βJ2 = e− αJ3 e− i βJ2 e i (5. the transformation from the rotated frame to the original frame.195) by ˜ Y m (θ. in terms of rotations defined with respect to the original frame.124 Theory of Angular Momentum and Spin The angles α. followed by J2 in the original frame. the transformation from the original frame to the rotated frame. φ) = S2 dΩ Y ∗m (θ. followed by transformation (i).209) The first two rotations in (5. φ) [D(α. The axis x2 is defined in the coordinate frame which is related to the original frame by rotation (i).h.203) can be expressed. can then be written e− i βJ2 − i αJ3 e = e− i αJ3 e− i βJ2 (5. Accordingly. and the r.194. Obviously.e.s. i. i. We will demonstrate that (5. of (5. For this purpose we notice that J2 can be expressed through the similarity transformation J2 = e− i αJ3 J2 e i αJ3 (5.206) The expression (5.203) has the disadvantage that it employs rotations defined with respect to three different frames of reference.m m.210) The third rotation in (5.203) For any ϑ ∈ R one can find Euler angles α.204) is satisfied. 5. in analogy to (5.h. the l. γ)] = . β.203).212) . The Euler rotation replaces exp − i ϑ · J by e− i γJ3 − i βJ2 − i αJ3 e e . i (5. is e− i γJ3 = e− i βJ2 e− i αJ3 e− i γJ3 e i αJ3 e i βJ2 (5. m [D(α. γ will be referred to as Euler angles. one can replace (5.e. For any similarity transformation involving operators A and S holds eS A S Accordingly.e. however.207) which replaces J2 by the inverse transformation (i). φ) e− γJ3 − i βJ2 − i αJ3 e e .207) are equivalent. the axis x3 is defined in the coordinate frame which is related to the original frame by the consecutive rotations (i) and (ii).209).. β.. (5. γ ∈ R such that exp i − ϑ·J = e− i γJ3 − i βJ2 − i αJ3 e e (5.205) Y m (θ. β.. β. we can write e− i −1 = S eA S −1 .203). γ)] m.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.s. i. (5. (i) and (ii). φ) (5.

γ)] m.218) Defining the so-called Wigner rotation matrix dmm (β) = one can express the rotation matrices [D(α. φ) e− m i βJ2 Y m (θ. the muon µ and its neutrino νµ of the second . (5.m m. Examples are the other five members of the lepton family to which the electron belongs. m ( ) S2 dΩ Y ∗ (θ. φ) (5. φ) = e−im α S2 dΩ Y ∗ (θ.206) by ˜ Y m (θ.220) We will derive below [see Eqs. (5.309.215) benefits from the choice of Euler angles for the parametrization of rotational transformations.217) Accordingly. This allows one to express the rotational trasnformation (5. φ) Y m (θ. γ)] = . 5.205. β. to redefine the three rotations in (5. φ) e− αJ3 e− βJ2 Y m (θ. one can write [D(α. m i Y m (θ. φ) . φ) f (θ. 5. β.203) with respect to the original (unprimed) frame one simply needs to reverse the order of the rotations. (5.e.. φ) e−iγm δ (5. (5. 5. but many other elementary components of matter are endowed with spin-like properties. φ) = S2 dΩ f (θ.214) e− γJ3 = S2 dΩ Y ∗m (θ.213) i. β. (5. γ)] m. φ) e− S2 i γJ3 Y m (θ.215) The evaluation of the matrix elements (5.210) yields the simple result e− i γJ3 − i βJ2 − i αJ3 e e = e− i αJ3 e− i βJ2 e− i γJ3 . m = e−iαm S2 dΩ Y ∗ (θ. φ) i i (5. the self-adjointness of the operator J3 one can state dΩ Y ∗ (θ.64) yields dΩ f (θ. γ)] m.219) = e−iαm dm m (β) e−iγm δ ( ) .7: Spin 1 2 and the group SU(2) 125 Multiplication from the left with (5. The eigenvalue property (5.216) Using. the electron e and the electron neutrino νe of the first generation.7 Spin 1 2 and the group SU(2) The spin describes a basic and fascinating property of matter.5. φ) e− m S2 i αJ3 f (θ. β. φ) e−imγ .219).310)] an explicit expression for the Wigner rotation matrix (5. m (5. φ) [D(α. in addition. φ) e− m i βJ2 Y m (θ. m [D(α. φ) . Best known is the spin of the electron.

and three of which make up the baryons which carry spin. Such systems can also assume any linear combination c+ χ+ + c− χ− .1 systems 2 2 can be related to two states which we denote by χ+ and χ− . e. We will consider first only the so-called spin– 1 . The transformation matrices are then elements of the set SU (2) = { complex 2 × 2 matrices U .1 and spin. leaving its imprint on the properties of nuclei. the tau τ and its neutrino ντ of the third generation and their antiparticles carry a 1 spin. hence. eight real numbers to specify the matrix elements. which might be the reason why consideration of rotational symmetry is so often fruitful in the study of matter. j. 1 We will specify for the transformations considered detU = 1. two of which (in certain linear combinations) make up the vector mesons which carry spin 1. As complex 2 × 2 matrices one needs. as long as |c+ |2 + |c− |2 = 1. generalizing then further below. there is nothing more elementary to matter than the spin property. particles which mediate the strong and the electro-weak interactions. the degrees of freedom of the matrices U are three real quantities. The presence of this property permeates matter also at larger scales than those of the elementary particles mentioned.2 . [Sk ]mn = ([Sk ]nm )∗ . c± ∈ C. and because of detU = 1. in principle.e. (5.222) where the vector S has three components. there are together five conditions in terms of real quantities to be met by the matrix elements and. have properties beyond those of single spins. k = 1. Spin. It appears to be rather impossible for a Physicist not to be enamored with the spin property. This specification implies that we consider transformations save for overall factors eiφ since such factors are known not to affect any observable properties. and the condition detU = 1 requires the Sk to have vanishing trace. each component Sk . One can show that the unitarity condition requires these matrices to be hermitian. atoms and molecules. a relationship. in fact. the photon. however. the two W± and the 2 2 Zo . U U † = U † U = 1 detU = 1 } 1. 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 . The important feature is that all U ∈ SU(2) can be parametrized by an exponential operator U = exp − i ϑ · S (5. like superconductivity and magnetism. k = 1. four real and four imaginary parts of Ujk . one for each matrix element of U † U .126 Theory of Angular Momentum and Spin generation.3 . 3 representing a 2 × 2 matrix. There are also the mediators. We will find that the spin in its transformation behaviour is closely related to angular momentum states.g. The particles mentioned. carry other spin-like properties which together. the spin of the electron is likely the most important property in Chemistry.221) One can show readily that this set forms a group with the groups binary operation being matrix multiplication. So do the three generations of six quarks. the quarks. the gluon. In any case. We may finally mention that the spin is at the heart of many properties of condensed matter systems. then allowed + − symmetry transformations of spin states are described by 2 × 2-matrices U with complex-valued matrix elements Ujk . 2. Because of the unitarity condition U † U = 1 which are really four equations in terms of real 1 quantities. How can the elements of SU(2) be parametrized. 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− . i. 2. and which carry spin 1. If we identify the states χ+ and χ− with the basis of a Hilbert space. There .

∈ R .229) and avoids commuting the components aj and bk . but their comonents must not necessarily commute with each other.e. Any such A can be expressed in terms of three real parameters x. 2 S2 = 1 σ2 .. (5. (5. In fact. Clifford algebras play an important role in the mathematical structure of physics. 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.229) = 1 . 3 are the Pauli matrices σ1 = 0 1 1 0 . z A = In fact. (5. b are vectors commuting with σ. According to this property the Pauli matrices generate a 3dimensional Clifford algebra C3 . y. σk ]− = 2i jk z x − iy x + iy −z .227) which is essentially identical to the Lie algebra (5.k=1 σ σ aj bk + j>k aj bk = 3 j. x. y. 1 σj σk = − σk σj for j = k which can also be readily verified. k = 1. The proof of this relation rests on the commutation relationship and the anti-commutation relationship (5. 2 S3 = 1 σ3 2 (5. e. they are associated with the important fermion property of matter. i.e. z.225) σ (5.224) Algebraic Properties of the Pauli Matrices The Pauli matrices (5. (5.231) .7: Spin 1 2 and the group SU(2) 127 exist three such linear independent matrices.g. σj 2 (5. 2. σ3 = 1 0 0 −1 . 1 i.227. namely.223) where σk . one obtains (σ · a) (σ · b) = 3 j 2 j=1 (σ ) aj bj 3 j k j. A = x σ 1 + y σ 2 + z σ3 .k=1 j<k + σ j σ k aj bk . (σ · a) (σ · b) = a · b 1 + iσ · (a × b) 1 (5. The Pauli matrices obey the following anti-commutation properties σj σk + σk σj = [σj .224) provide a basis in terms of which all traceless.k=1 σ σ 3 j k j. hermitian 2 × 2–matrices A can be expandend.5.230) where a. σ2 = 0 −i i 0 .228) (5. 5. it might hold aj bk − bk aj = 0.31) of the group SO(3). the simplest choice being S1 = 1 σ1 . σk ]+ = 2δjk 1 .226) The Pauli matrices satisfy very special commutation and anti-commutation relationships. At this point we will state a useful property. One can readily verify the commutation property σj σk − σk σj = [σj . it holds.

h. 1 1 1 ( + 1) χ± 2 2 . Well almost like it.k=1 (σ σ j>k σ k σ j (aj bk − ak bj ) = (5.e.231) proves (5. since there is a slight difference: when you interprete the 3-component ϑ in (5.227) leads to 3 3 jk j. to rotations in space. however.222) as a rotation vector and you rotate once.232) − σ k σ j ) (aj bk − ak bj ) .222) by the Euler form exp(−iγS3 ) exp(−iβS2 ) exp(−iαS3 ) .222). S3 χ± = ± 1 χ± 2 . 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. (5. S ] = i k m Sm (5.233) This result together with (5.234) 5. 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 .k=1 j>k σ j σ k (aj bk − ak bj ) − 1 2 3 j k j.236) The matrix elements of this 2 × 2 operator can be labelled by the basis states χ+ and χ− . For this purpose we choose to replace (5. The two remaining terms yield using σ j σ k = −σ k σ j for j = k and altering ‘dummy’ summation indices 3 j. through the algebra obeyed by the operators Sk [Sk .237) + 1)1 and.238) S 2 χ± = . We like to derive this result now by evaluating the transformations of SU(2) explicitly. i. yields a · b.235) which is identical to that of the generators of SO(3). The commutation property (5.s. (5. only a 720o rotation leaves the spin invariant. hence.128 Theory of Angular Momentum and Spin Using (σ j )2 = 1 the first term on the r.k=1 j>k σ j σ k (aj bk − ak bj ) = 1 2 3 j. spin changes sign. In the special case a = b ∈ R holds (σ · a)2 = a 2 1 . however. let say around the x2 –axis by 360o .8 Generators and Rotation Matrices of SU(2) Sofar there is nothing which relates the transformations U . (5. 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. 1 (5.k=1 j>k i σl (aj bk − ak bj ) = i =1 σ a×b . This relationship emerges. (5.k=1 j>k 1 2 3 j.230).

. We may.5. no two such particles can exist in the same state.. . therefore.236) 1 2m | exp(−iγS3 ) exp(−iβS2 ) exp(−iαS3 ) | 1 m 2 1 2m = e−iγm | exp(−iβS2 ) | 1 m e−iαm 2 (1) 2 = e−iγm dmm (β)e−iαm (5.e. .242) The expressions in brackets [.] can be identified with the Taylor expansion of the cos.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. . (5. For this purpose one needs to determine the powers of −iβS2 . 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 .244) i.e. (5. following Jourdan and Schwinger. that states | m for higher quantum numbers = 0. The complete matrix elements (5. 3 .240) . 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.239) in the notation (5. namely. − + 1. .9 Constructing Spin States with Larger Quantum Numbers Through Spinor Operators In this section we like to demonstrate.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). m = − .219. rotation by 360o changes the sign of the spin state. . i. 2 2 We obtain with this notation for the transformations (5.245) 5. identical particles any number of which can exist in the same state χ+ and χ− . The idempotence property of Sk yields particularly simple expressions for these powers. (5. .e.239 ) one expands the exponential operator exp(−iβS2 ). 2. .219).220) are [D(α. β.243) We note the property of this rotation matrix 2 dmm (2π) (1) = −1 1 . 1. i. ( − 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. . (5.239) where we made use of the notation (5. γ)] m. use the label 2 2 | 1 ± 1 for the states χ± . 1 . In order to determine the Wigner matrix in (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 . 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.5.

253) = δζζ . −) bζ . . b− † = − sin β † β b + cos b† . b † ζ ζ =0 (5.243) 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 − . b− )T is expressed through the commutation relationships + − (ζ. using (5. The operators b† allow one to construct a set of states which represent j + m–fold and j − m–fold ± χ+ and χ− states as follows b† + j+m |Ψ (j.248) that bζ relationships † † b ζ. 2 2 . b ζ = b† . m) = b† − j−m (j + m)! (j − m)! |Ψ0 . j. These operators are associated with a given spatial reference system. 2 + 2 − † (5. ζ = +. b † ζ For the vacuum state |Ψ0 holds = δζζ .254) We will show below that these states are orthonormal and form spin states with higher quantum numbers. also operators of the type x b† = x+ b† + x− b† + − . . j = 0. . b ζ = b ζ. b ζ = 0 b ζ. i. . (5. . 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± . m = −j. The boson character of the operators b† = (b† . b† |Ψ0 = χ+ + .247.e. b† )T and b = (b+ .247) (5. b ζ The States |Ψ(j.248) bζ . −j + 1. + − For example. x∗ b = x∗ b+ + x∗ b− + − (5. We consider.5. therefore. 1.250) which for x∗ x = x∗ x+ + x∗ x− = 1 represent spinor operators in an arbitrary reference system.246) The corresponding adjoint operators are denoted by b+ and b− . i. . . 3 . b† |Ψ0 = χ− − . b† )T and b = (b+ . b± |Ψ0 = 0 (5.e. b− )T as spinor creation and annihilation + − operators.249) In the following will refer to b† = (b† . (5.252) (5. m) † and bζ obey the commutation (5.251) One can show using this property and (5.

5. that the annihilation operators play a role similar to the differential operator in calculus. (5.256) In the special case |state = |Ψ0 this reads using (5. possibly infinite power series.255) is b+ .258) Equations (5. The Spinor Derivative Operator A most important operation is the action of the annihilation operators b+ and b− on operators which can be expressed as polynomials. For this purpose we consider the following polynomial of creation operators N j+m j−m f (b† ) = + n=0 cn b† + n . b† + n b† + b† + b† + + n n n b+ . f (b† ) + (5. b† + b+ . b† + − well-known for the quantum mechanical harmonic oscillator. b† + n .260) . An example + − is the monomial b† which generates the states (5. We will show that for this commutator holds n b+ . b† + n = n cn b+ .247. (5. If it is true for n then using (5.248) and from (5. f b† Obviously.10 Algebraic Properties of Spinor Operators We want to establish first a few important and useful algebraic properties which result from the commutation relationships (5.10: Algebraic Properties of Spinor Operators 131 5. + + (5.257) which for (5. (5. b † + = n b† + n−1 b† + b † + + n = (n + 1) b† + .259) The property is obviously true for n = 0. in b† and b† . f (b† ) |state + f (b† ) b+ |state .256) and (5.248) we obtain b+ . 5. We will derive the result. we need to evaluate b+ . b† + = n b† + n−1 .249). .257) motivate us to determine the commutator (5.254). + b+ .255) In order to determine how this operator is modified by multiplication from the left by b+ we note b+ f (b† ) |state = + b+ . f (b† ) |Ψ0 .249) b+ f (b† ) |Ψ0 = + b+ . b† + n+1 = = b+ .

f (b† ) − = f (b† ) − (5. m) m=−j (5. (5.257) we can conclude in particular bζ f (b† .262) Because of the commutation relationships (5. b† ) . the property holds also for n + 1. = f (b† ) + (5. + − (5. one can prove b− . m) .132 Theory of Angular Momentum and Spin i. By induction we can conclude that (5. + − (5.261) Similarly. f (x)] g(x) = (∂k f (x))g(x) or [∂k .267) exp xb† |Ψ0 is called a generating function of |Ψ(j.250) and where φjm (x) represents the function of the two variables x+ and x− closely related to |Ψ(j.259) holds for all n ∈ N. −) + − + − bζ . Equation (5.266) we compare the terms x+ b† + j+m φjm (x) Ψ (j. b† ) + − = ∂ ∂b† ζ f (b† .264) This demonstrates the equivalence of the spinor operators bζ and the derivative operation. m) = x− b† − j−m (j + m)! (j − m)! |Ψ0 . (5. m) φjm (x) = j−m xj+m x− + (j + m)! (j − m)! .1. In order to derive (5.263) 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 . f (x)] = ∂k f (x) . m) We want to prove now the property ∞ j exp xb † |Ψ0 = 1 j=0. (5.265) Generating Function of the States |Ψ(j.263) According to (5..247) the more general property for polynomials f (b† . b† ) |Ψ0 . f (b† ) + where f (x) = df dx .255) b+ .e.266) where xb† has been defined in (5. φjm (x) Ψ (j.268) .. b† ) in b† and b† holds (ζ = +. b† ) |Ψ0 = + − ∂ ∂b† ζ f (b† . One can finally conclude for polynomials (5. f (b† . 2 ..

. ... 2 2 leads to ∞ j ∞ φjm (x) |Ψ (j.269) (5..1. (2j)! 2j |Ψ0 (5. 2 . . (5.. 1 xb† u! |Ψ0 (5. s! (n − s)! Defining j+m = N −s j−m = s.1. . ..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..1. m=−j = j=0.273) Change of the summation indices allows one to conclude j φjm (x) Ψ (j. m) m=−j = = 1 x+ b† + x− b† + − (2j)! 2j 1 xb† |Ψ0 .274) Summing this expression over 2j = 0.275) = exp xb† |Ψ0 . chosing the summation index j = 0... 2. one obtains φjm (x) Ψ (j. 1 . 2. 1.. 2 ∞ 1 xb† 2j! u 2j |Ψ0 = u=0. 1 ..272) The summation over s can be written in terms of j and m using (5.270) s x+ b† + N −s x− b† − s |Ψ0 . (5. . m) = = |Ψ0 (N − s)! s! N −s 1 N! x+ b† x− b† + − N ! (N − s)!s! N −s 133 (5. 1.5. . 3 . m) 1 j=0. i.10: Algebraic Properties of Spinor Operators with the s–th term in the binomial expansion of (a + b)n n! an−s bs .270) N 2j j → s=0 j−m=0 → m=−j .e.

b† ] = 0 allow one to state that for + − + any polynomial f (b+ . b− )e(yb ) |Ψ0 = f (y+ . m) |Ψ j .284) .277) by Ψ0 |e(x b) e(yb ) Ψ0 .j . (5. of (5. Orthonormality of the States |Ψ (j.278) † ∗ which allows us to replace the l. we can conclude † † ∗ e(xb ) Ψ0 |e(yb ) Ψ0 = e(x y) (5. hence. b− ] = 0. (5. b− ) holds † † f (b+ . that Ψ (j.h.m φjm (x∗ ) φj m (y) Ψ (j.254) are orthonormal.134 Theory of Angular Momentum and Spin which concludes our derivation. The generating function allows one to derive various properties of the states |Ψ (j.264) implies † † bζ e(yb ) |Ψ0 = yζ e(yb ) |Ψ0 . i. m) We want to show now that the states (5.s. m .282) where we defined x∗ y = x∗ y+ + x∗ y− . b† ] = 0 and [b− .249) for any non-vanishing integer s holds + − Ψ0 |bs Ψ0 = 0 and.e.280) The commutation properties [b+ . (5.276) = j.283) (5. m) |Ψ j . 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. For this purpose we consider the inner product † † e(xb ) Ψ0 |e(yb ) Ψ0 = δjj δmm (5. In order to evaluate the ∗ b) (yb† ) operator e(x e we notice that the derivative property (5. m) and will be used for this purpose below. m holds.281) and.m. According to (5. therefore. [b† .279) One can generalize this to † † s (bζ )s e(yb ) |Ψ0 = yζ e(yb ) |Ψ0 .277) To evaluate this expression we first notice † e(xb ) † = e(x ∗ b) . y− )e(yb ) |Ψ0 (5. (5.

.m 135 = e(x ∗ y) . m) |Ψ j . 2.289) < n |σi | m >< n |σj | m > {b† bm .m.. k = 1.276).277) φjm (x∗ ) φj m (y) Ψ (j. States 2 We want to demonstrate now that the states |Ψ (j.m ijk Jk .285) Following steps similar to those in Eqs.n . 3 denote the Pauli spin matrices (5.5. 1 New Representation of Spin 0. (5. (5. 1. (5. Jj ] = = 1 4 1 4 < n |σi | m >< n |σj | m > b† bm . (5. − 2 + (5. using < + |σ1 | + > = < − |σ1 | − > = 0.m φjm (x∗ ) φjm (y) . bm bm } n n = 1 4 < n |σi | m >< n |σj | m > {b† bm δmn − b† bm δm n } n n n.268–5. Jj ] = i This property is derived as follows [Ji . b† bm n n n. < + |σ1 | − > = < − |σ1 | + > = 1.m +b† b† . One defines corresponding operators Jk through 1 Jk = b† < ζ |σk | ζ > bζ .m. 2 .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). m j. .288) The three operators obey the Lie algebra of SU(2) [Ji .ζ where σk . etc.10: Algebraic Properties of Spinor Operators and comparing (5. 5.223. 3 .287) ζ 2 ζ.n .n .m. m) as given in (5. The 2 operators are explicitly.282) and (5.275) one can show e(x ∗ y) = j. . (5. J1 = J2 = J3 = 1 † b b− + b† b+ − 2 + 1 † b b− − b† b+ − 2i + 1 † b b+ − b† b− .m .286) from which follows immediately the orthonormality property (5. b† bm n n n.224). In our present notation the matrix elements < ζ |σk | ζ > for ζ = ± correspond to the matrix elements < ζ |σk | ζ > for σ = ± 1 .

k ijk ijk σk | m > b † bm n = i = i 1 2 < n |σk | m > b† bm n n. To this end we note 2 2 2 J 2 = J1 + J2 + J3 = 1 1 2 J+ J− + J− J+ + J3 2 2 . σj ]| m > b† bm n n. J+ ] = [J1 − iJ2 .294) . (5. (5. J1 ] + i [J1 .136 = 1 4 − = = 1 4 1 4 Theory of Angular Momentum and Spin < n |σi | m >< m |σj | m > b† bm n n.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).290) |Ψ (j.m < n |2i n. m) are eigenstates of J 2 and J3 . J2 ] = −2 J3 from which we can conclude the property 1 1 J− J+ = J+ J− − J3 2 2 Hence. (5. − (5.295) The last result together with (5. m) as Eigenstates of J 2 and J3 We wish to show now that the states |Ψ (j.297) .293) . (5.m 1 4 < n |σj | m >< m |σi | m > b† bm n n .291) Here we have defined J+ J− = J1 + iJ2 = J1 − iJ2 = b† b− + = b† b+ .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 .292) The commutation relationships (5.m < n |[σi .m ijk Jk . (5.289) yield [J− .m. 2 J 2 = J+ J− + J3 − J3 = J3 (J3 − 1) + J+ J− (5. J1 + iJ2 ] = −i [J2 . (5.m.m.

. State the corresponding eigenvalues and the degree of degeneracy. . (5.299) (5.298) (5. 2 2 137 (5. 1 2. .302) (5.e. ˆ respectively. using operators I± = I1 ± iI2 2 2 2 and the subspace {|n1 . . m = λ||λ. m) One can furthermore derive readily J+ |Ψ (j. One can then conclude J 2 |Ψ (j. . b† . [bj .s. m) J3 |Ψ (j. 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† . −λ + 1. Ik ] = i jk I . . 1 2 k (a) Show that the eigenstates are given by 1 |n1 . m = m||λ. Construct.h.5. b+ ] = δjk .10: Algebraic Properties of Spinor Operators The last term on the r. 1.1: The system investigated in this section is formally identical to a 2-dimensional isotropic harmonic oscillator governed by the Hamiltonian H = ω(b+ b1 + b+ b2 + 1) . . m − 1) . .303) Exercise 5. 2. m + 1) (j + m) (j − m + 1)|Ψ (j. 2 . n1 = 0. b † + b + + b † + b+ b † − b − 1 1 b† + b + + b † + b + b † − b− + b † − b − b † + b + . n2 . m . . [Ij . m) = = (j + m + 1) (j − m) |Ψ (j. Show λ = n1 +n2 n1 +n2 2 ( 2 + 1) and m = 1 2 (n1 − n2 ). j = 1. m = −λ. I3 = (b+ b1 + b+ b2 ) 2 2 2 2 2i 2 1 satisfy the Lie algebra of SU(2).305) = = j ( j + 1 ) |Ψ (j. (b) Show that the three operators 1 + 1 1 + (b1 b2 + b+ b1 ) . I2 = (b1 b2 − b+ b1 ) .300) (5. i.10.301) Obviously.. I3 ||λ. . . and eigenstates of k with eigenvalues j. n = n1 + n2 fixed. m) J− |Ψ (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. i.304) (5. m I1 = where λ = 0.254) are eigenstates of + b+ and − b− with eigenvalues j + m and j − m. m) m |Ψ (j. the number of states to the possible energy eigenvalues. . λ. the states (5.e. . m) . n} eigenstates of I 2 = I1 + I2 + I3 and of I3 I 2 ||λ. (5.

m ) are related to the + − states |Ψ (j.5. For this purpose we note that the states |Ψ(jm) in a rotated coordinate system according m to (5. m) . The combination of σ.306) and (5. + − One may expect that these two conditions restrict both σ and σ.306) where b † and b † are given by (5.138 Theory of Angular Momentum and Spin 5.308) The latter sum involves terms (b† )j+m (b† )j−m for σ + σ = j + m and 2j − σ − σ = j − m.309) . σ values which yields (b† )j+m (b† )j−m is + − then σ = j + m − σ. m = b† + j+m b† − j−m (j + m )! (j − m )! |Ψ0 (5. m = m=−j dm m (β) |Ψ (j.306. However. m) in the original coordinate system by j |Ψ j.251). 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. On the other side the states |Ψ (j.254) are |Ψ j.307) Comparision of (5.310) it reduces to (5.307) shows that the elements of the rotation matrix can be obtained by binomial expansion of b † and b † in terms of b† and b† . (j) (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). In case j = 1 this expression yields.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 × (5.243). . both conditions are satisfied for σ = j + m − σ. for example. The prefactor of (b† )j+m (b† )j−m which according to (5.  1 1 √ sinβ 2 (1 + cosβ) 2  (1) 1 cosβ (dm m ) =  − √2 sinβ 1 1 √ 2 (1 − cosβ) − 2 sinβ and in case j = 1 2 1 2 (1 − cosβ) 1 √ sinβ 2 1 (1 + cosβ) 2    (5.

This factor implies.12 Mapping of SU(2) onto SO(3) The representation (5.e. 2 2 − (5. that the the representation (5.. therefore. Replacing the latter transformation by its negative form. b+ † = −cos β † β b+ − sin b† 2 2 − . is the expected element of SO(3).251). Hence.314) leaves (5. conclude that the mapping from SU(2) onto SO(3) is a 2–1 mapping. however. U =  0 1 1 i √ ( x1 − ix2 ) √ x3 − √2 0 2 2 (5. the orthogonal 3 × 3 matrix which describes a rotation around the x2 –axis.251) and (5..s.313) × 0 0 2  =  0 sinβ 0 cosβ 1 −i 0 which.309) and the particular matrix (5.12: Mapping of SU(2) onto SO(3) 139 5. x2 .311) (5.310) establishes a mapping of SU(2) onto SO(3).310) to the SU(2) transformation assumed in deriving the general result (5.309). . of (5. One can. b− † = sin β † β b+ − cos b† . This can be shown by (1) applying to the matrix A = (dm m ) in (5. in fact.308) and had the form (5.   = U  x2  .309) does not distinguish between SU(2) transformations (5.312) transforms like an angular momentum state ˜ |1 m . namely that the component of the vector on the l. 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.312) The choice of this transformation derives from a property shown further below. the matrix A should represent a rotation around the x2 –axis in the space of vectors ˜ (x1 . x3 ) ∈ R .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  .308) unaltered except for a factor (−1)2j which multiplies then also the final result (5.g.e. in case of j = 1.5. i.310). i. e. It is of interest to trace the mapping from SU(2) onto SO(3) as represented by (5. The SU(2) transformation entered in (5.h.

140 Theory of Angular Momentum and Spin .

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

In fact. . . 14 641} . i. = B such.. mB . composite systems behaving like elementary objects are common. 1. 2.max d(B) = 1 =0 (2 1 + 1) 2 =0 (2 2 + 1) = ( 1.max denote the largest orbital and rotational angular momentum values.142 Addition of Angular Momenta and Spin ρC . ρ 1 = 0. to separate the 14 641 dimensional space of rotational states Y 1 m1 (ˆAB ) Y 2 m2 (ˆC ) into r ρ subspaces Bk . .max = ∆ 2 mA mB mC E . . Such extremely useful transformation of the problem can be achieved through the choice of a new basis set B = { 1 . 2. the dimensions d(Bk ) of these subspaces does not exceed 100. will make their appearence a natural consequence of the symmetry of the building blocks of matter. and the following mathematical description will shed light on their ubiquitous appearence in physics.max = 1 2 IA−B E . . the values of which are determined by the size of the interaction domain ∆V .1) where 1. The latter often arrives at the physical properties of composite systems by adding the . .m1 . rather than by Classical Mechanics.m1 2 . n = 1. − 2 ≤ m1 ≤ (6. .max 2. vibrations and rearrangement of the particles in reactions like AB + C → A + BC. mA mB + mB mC + mA mC 2. = 0.max + 1)2 ( 2. (6. by the masses mA . . 2 . (6. .max .m2 Y 1 m1 (ˆAB ) Y 2 m2 (ˆC ). 2} 1.max = 2.max and 2. by the total energy E. albeit puzzling. have not even be considered. 1.2) The dimension d(B) of B is 1. in fact.e. . The rotational symmetry of the interaction between the particles allows one. since further important degrees of freedom. mC . 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 ). and by the moment of inertia IA−B of the diatomic molecule A–B approximately as follows 1. each of which stands for two angles.m2 c (n) r ρ 1.max .3) For rather moderate values 1.max + 1)2 (6.4) 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. a very large number. − 1 ≤ m1 ≤ 1. . . as we will demonstrate below. Such large number of dynamically coupled states would constitute a serious problem in any detailed description of the scattering process. . 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. however. Obviously. in particular. 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. that only states within the subspaces Bk are coupled in the scattering process.max = 10 one obtains d(B) = 14 641. B1 ⊕ B2 ⊕ .

the existence of a basis (6.4) which decouples rotational states is connected with the rotational symmetry of the Hamiltonian of the three particle system considered. ∂yAB ∂zAB i (2) ∂ ∂ − J3 = ρy − ρx . 2.6.8) Obviously. Following our description of rotations of single particle wave functions we express (6. the commutation relationships (1) Jp . To specify this property mathematically let us denote by H the Hamiltonian in the rotated frame.48) i i R(ϑ) = exp − ϑ · J (1) exp − ϑ · J (2) (6.55) holds i (1) − J1 = zAB ∂ ∂ − yAB .5) according to (5. be different from H. One can equivalently express therefore (6. For ˆ ˆ example. q = 1.e. 3 (6. ρC ) = ψ(R−1 (ϑ) rAB . according to (5. presumably a facility the reader would like to acquire with great eagerness. It holds then H R(ϑ) ψ = R(ϑ) H ψ. 5. from which follows in turn the well-known result that H is related to H through the similarity transformation H = R(ϑ) H R−1 (ϑ). Since this is true for any ψ(rAB . ρC ) defined through R(ϑ) ψ(rAB . connected with the fact that the Hamiltonian H does not depend on the overall orientation of the three interacting particles.5) do not affect the Hamiltonian.0: Addition of Angular Momenta and Spin 143 corresponding physical properties of the elementary components.7) where the generators J (k) are differential operators acting on rAB (k = 1) and on ρC (k = 2). Jq(2) = 0 for p. (6.. in fact. i. 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. In this sense. Rotational Symmetry of the Hamiltonian As pointed out already. ∂ρx ∂ρy (6. rotations R(ϑ) of the wave functions ψ(rAB . The invariance of the Hamiltonian under overall rotations of the three particle system implies then H = R(ϑ) H R−1 (ϑ) .53. assuming presently that H might. 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. Hence.9) hold since the components of J (k) are differential operators with respect to different variables. R−1 (ϑ) ρC ) (6. the reader will learn in the following section how to add and subtract in the microscopic world of Quantum Physics.6) For the following it is essential to note that H is not invariant under rotations of only rAB or ρC .7) i R(ϑ) = exp − ϑ · J (6.10) . but solely under simultaneous and identical rotations of rAB or ρC . ρC ) it follows H R(ϑ) = R(ϑ) H.

3 do not commute.e. J ] = i k m Jm (6. k = 1. 2. One can show readily that the commutation relationships [Jk . however. k = 1.14) implies that the total angular momentum is conserved during the scattering process. (6.e. 2 together with the property [A.. componentwise.f. B + C] = [A. Before we apply the procedure (5. k = 1. i. The property (6. J ] = i k m Jm for n = 1.17) .15) are satisfied. i. 2. C]. of course. the operators Jk . For a proof one uses (6. (5. r ρ Definition of Total Angular Momentum States The commutation property (6. k = 1. According to the theorem above this property implies that one can construct eigenstates YJM of J3 and of J2 = J2 + J2 + J2 1 2 3 (6. [H.71–5.6) for rotational invariance of the Hamiltonian in the form i i H = exp − ϑ · J H exp + ϑ · J . the (n) (n) (n) properties [Jk . To order O(|ϑ|) one obtains i ϑ · JH . 3 . a supposition which can be tested through the commutation properties of these operators. 2. Eqs. In the following we will use the notation r ρ Ω1 = rAB .12) each individually can have simultaneous eigenstates with the Hamiltonian. We suspect.144 where J = J (1) + J (2) Addition of Angular Momenta and Spin . In fact.14) We will refer in the following to Jk .e.12) We consider this equation for infinitesimal rotations. that the components Jk .71–5. ˆ Ω 2 = ρC . (6.. i. (6. 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 ).9). and the eigenvalues of J2 and J3 are good quantum numbers. ˆ (6. that energy.16) following the procedure stated in the theorem above [c.11) By means of (6. 3 as the three components of the total angular momentum operator. 3 cannot have simultaneous eigenstates among each other.81) we want to consider the relationship between YJM and Y 1 m1 (ˆAB ) Y 2 m2 (ˆC ).14) implies that the components of the total angular momentum operator (6. We recognize. the important fact that the Jk obey the Lie algebra of SO(3).81)]. B] + [A. (6. Jk ] = 0 . for |ϑ| H ≈ 1 − 1 i ϑ·J H 1 + 1 i ϑ·J ≈ H + i 1.13) Hϑ · J − Since this holds for any ϑ it follows HJ − J H = 0 or.11) we can write the condition (6. we will find that the states yield the basis B with the desired property of a maximal uncoupling of rotational states. 2.

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. Ω2 ) (1) (2) = 1 m1 2 m2 ) Y 1 m1 (Ω1 ) Y 2 m2 (Ω2 ) (m1 + m2 ) (J M | .21) One can. We first notice that YJM in (6. of (6.20) This equation can be satisfied only if the Clebsch-Gordan coefficients satisfy (J M | 1 m1 2 m2 ) = 0 for m1 + m2 = M . i. hence.m2 (J M | 1 m1 2 m2 ) Y 1 m1 (Ω1 ) Y 2 m2 (Ω2 ) (6. restrict the sum in (6. mesons.e.6. However.s. and three quarks. therefore. 2 |Ω1 . J (2) . These coefficients. 2 |Ω1 .18) to avoid summation of vanishing terms YJM ( 1 . The expansion coefficients (J M | 1 m1 2 m2 ) ˆ are called the Clebsch-Gordan coefficients which we seek to determine now.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) . Ω2 ) = m1 (J M | 1 m1 2 M − m1 ) Y 1 m1 (Ω1 ) Y 2 m2 (Ω2 ) . and J (1) · J (2) commute. J (1) . Ω2 ) = m1 . an important example being the symmetry groups SU(N) governing elementary particles made up of two quarks.1. J (2) . baryons. i.e. J3 .22) We will not adopt such explicit summation since it leads to cumbersum notation.18) where the states YJM ( 1 . two further quantum numbers need to be specified for a complete characterization of the total angular momentum states. ρC ) are normalized.18) yields using the property (k) J3 Y k mk (Ωk ) = mk Y k mk (Ωk ) . We. 2 M YJM ( 1 .h. k = 1. (6.e. Since YJM sofar specifies solely two quantum numbers.1: Show that J2 .m2 2 |Ω1 . i. m1 . assume the expansion YJM ( 1 . J3 YJM = M YJM . and J3 . 2 |Ω1 . Equivalent coefficients exist for other symmetry properties of multi–component systems. (6. play a cardinal role in the mathematical description of microscopic physical systems. the eigenvalue being specified by the quantum number M . (6..19) Noting J3 = J3 + J3 and applying this to the l. or the closely related Wigner 3j-coefficients introduced further below. J3 .21) hold.1: Clebsch-Gordan Coefficients 145 6. The two missing quantum 2 2 numbers are 1 and 2 corresponding to the eigenvalues of J (1) and J (2) . Exercise 6..18) is an eigenfunction of J3 . . (6. the reader should always keep in mind that conditions equivalent to (6.

−J + 1. (6. In fact. Our derivation below will also yield real values for the Clebsch-Gordan coefficients.s. J} . .. and that the states can be normalized. J (1) and J (2) . m1 = − 1 . 1.. . In this case the range of |J|–values would be the interval [ |J (1) | − |J (2) | . = {YJM ( 1 . .28) This property follows from the fact that the basis elements are eigenstates of hermitian operators with different eigenvalues. (6. . Ω2 ) = δJJ δM M (6. 2 ) has (2 1 + 1)(2 2 + 1) elements. . JM 2 |Ω1 . (6. . M = −J. 2) = {Y 1 m1 (Ω1 ) Y 2 2 m2 (Ω2 ). .18) constitutes a change of an orthonormal basis B( 1 . property implies dΩ1 dΩ2 Y 1 m1 corresponding to the l. 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. (6. B2 2) should be 2 |Ω1 . the basis B ( 1 . 2) .25) i. 2 ) is also orthonormal2 and must have the same number of elements. 1 + 2.18) applied to Y∗ and to YJ JM (J M | m1 . (6. 2 |Ω1 .m2 2 M and with (6. . respectively. . to a new basis B ( 1 . . 2 ). . (6. This obviously corresponds quantum mechanically to a range of J–values J = | 1 − 2 |.27) together with (6. . . J = | 1 − 2 |. . . However.25) We want to state now two summation conditions which follow from the orthonormality of the two basis sets B( 1 . it holds 1+ 2 ( 2J + 1 ) = (2 J=| 1 − 2 | 1 + 1) (2 2 + 1) . . m2 = − 2 . J. The property dΩ1 dΩ2 Y∗ ( 1 . 2 ) and B ( 1 . the range of values assumed for J is correct. For each quantum number J there should be 2J + 1 elements YJM with M = −J. .24) The basis B( 1 .2: Prove Eq.1. 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 | + 1. . Ω2 ) YJ M ( 1. it is not immediately obvious what the J– values are. in fact.146 Theory of Angular Momentum and Spin The expansion (6. −J + 1. .23) corresponding to the r. − + 1. .h.e. Exercise 6.26) We will show below in an explicit construction of the Clebsch-Gordan coefficients that.h. The basis B ( 1 . 1 + 2 . Ω2 ).s. − 2} 1 + 1. | 1 − 2| + 1.24) yields 1 m1 2 m2 ) ∗ 1 m1 2 m2 ) (J M| = δJJ δM M . |J (1) | + |J (2) |].

1 states. the group SU(3) in case of meson multiplets involving two quarks.g. The (J M | 1 m1 2 m2 ) cJ M . or J > 1 + 2 (6. (6. i. .18). (6. to composite systems involving spin.28) allows one to conclude that the coefficients cJ (J M | 1 m1 2 m2 )∗ .34) 6. The result of this construction will include all the properties previewed above.6. A similar 2 construction will also be applied to composite systems governed by other symmetry groups.. (6. We can. 2 ) and B ( 1 .2: Construction of Clebsch-Gordan Coefficients 147 The second summation condition starts from the fact that the basis sets B( 1 . i.81) stated above.e.33) JM The latter summation has not been restricted explicitly to allowed J–values. 2 |Ω1 .32) which complements (6. will be based solely on the commutation properties of the operators J and J (k) . Ω2 ) . 1 m1 2 m2 ) = 0 if J < | 1 − 2| ..e.29) where the expansion coefficients are given by the respective scalar products in function space cJ M = dΩ1 dΩ2 Y∗ J M ( 1. 2 |Ω1 .71–5. Ω2 )Y 1 m1 (Ω1 )Y 2 m2 (Ω2 ) . 2 ) span the same function space. Ω2 ). 1+ 2 J Y 1 m1 (Ω1 )Y 2 m2 (Ω2 ) = J =| 1 − 2 | M =−J cJ M YJ M ( 1.e. therefore.2 Construction of Clebsch-Gordan Coefficients We will now construct the Clebsch-Gordan coefficients. 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. and actually also the properties of Clebsch-Gordan coefficients stated above.m2 M ( 1. At this point we like to stress that the construction will be based on the theorems (5. Hence. (6.18) 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 . i. Ω2 ) .30) The latter property follows from multiplying (6. One can show readily using the same reasoning as applied in the derivation of (6. e.. 2 |Ω1 .31) are identical to Comparision with (6. or baryons multiplets involving three quarks.28) from (6.18) by Y∗ J orthogonality property (6.27) yields δJJ δM M = m1 . 2 |Ω1 . it is possible to expand Y 1 m1 (Ω1 )Y 2 m2 (Ω2 ) in terms of YJM ( 1 . Ω2 ) and integrating. rather the convention (J M | has been assumed. (6. 2 |Ω1 .. also apply the results.

1+ 2 ( 1. . Having determined such YJJ we then construct the whole family of functions XJ = {YJM ( 1 .35) 1 − 2 |. 2 |Ω1 . 2 |Ω1 .64) J3 = = (1) + J3 (1) 1 (2) Y 1 1 (Ω1 ) Y 2 2 2 2 (Ω2 ) 1 1 J3 Y ( + 1 1 (Ω1 ) Y 1 1 (Ω2 ) + Y 2 2 (Ω1 ) J3 Y (2) 2 2 (Ω2 ) (6. For this purpose we insert (6. . −J. Ω2 ) is normalized dΩ2 G dΩ1 Y 1 1 1+ 2.148 For the construction of YJM we will need the operators J± = J1 + iJ2 . .43) . we can demonstrate condition (6.11) and (5. Ω2 ) .66. We first seek an unnormalized solution GJJ and later normalize. (6. M = −J. To find GJJ we start from the observation that GJJ represents the state with the largest possible quantum number J = 1 + 2 with the largest possible component M = 1 + 2 along the z–axis.37). 1+ 2 ( 1. Ω2 ) = (J + M + 1)(J − M )YJM ( 1 . 2 |Ω1 . J − 2.39) into (6.36) = 0 = J YJJ ( 1 . 2 |Ω1 .40) (Ω1 ) J+ Y (2) 2 2 J+ Y 1 1 (Ω1 ) Y (Ω2 ) + Y (Ω2 ) = 0 .41) 2) Y (Ω1 ) Y (Ω2 ) .37) and replace (1) (2) according to (6. 1+ 2 ( 1.25) using (6. therefore. (6.18) which satisfies J+ YJJ ( 1 . 2 |Ω1 .68) J+ = (1) + J+ (1) (2) Y 1 1 (Ω1 ) Y 2 2 2 2 (Ω2 ) 1 1 (6. Ω2 ) = Y 1 1 (Ω1 ) Y 2 2 (Ω2 ) . The construction assumes a particular choice of J ∈ {| such J–value seeks an expansion (6.37) The solution needs to be normalized. Ω2 ) 2 |Ω1 .11) J+ by J+ + J+ . Ω2 ) (Ω1 ) dΩ2 Y 2 2 (Ω2 ) = 1. . . Similarly. 2 |Ω1 . . 1 + 2} and for (6. J} by applying repeatedly J− YJM +1 ( 1 . . we can also demonstrate using (??) that G dΩ1 = We.5.38) for M = J − 1. Ω2 ). 1+ 2. Ω2 ) . −J + 1. Ω2 ) Addition of Angular Momenta and Spin (6. 1+ 2 In fact. 2 |Ω1 . J3 YJJ ( 1 . | 1 − 2| + 1. Quantum mechanically this corresponds to a state G 1+ 2.39) which we will try for a solution of (6. (6. 2 |Ω1 . . We embark on the suggested construction for the choice J = 1 + 2 .42) ( 1. (6. have shown Y 1+ 2. Ω2 ) = Y 1 1 (Ω1 ) Y 2 2 (Ω2 ) (6. . 2 |Ω1 . 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. . We obtain using (5.

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.h. 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.2: Construction of Clebsch-Gordan Coefficients 149 We now employ property (6. M = −( + − ). This expression is orthogonal to (6. We demonstrate the procedure ex(1) (1) plicitly only for M = 1 + 2 − 1.44) (Ω2 ) . Ω2 ) 2 1+ 2 = (Ω1 )Y 2 2 −1 (6. with the symmetry pattern and with the terms Y 1 m1 Y 2 m2 contributing to each YJ M . According to the condition (6.46) 2 1 +c Y (Ω1 ) Y 2 2 (Ω2 ) = 0 . . thereby. To satisfy this equation one needs to choose c = − 1 / 2 . .45). . The r. ( + )}.47) (Ω2 ) − 1 1 2 + Y 2 1 1 −1 (Ω1 )Y 2 2 + Y 2 1 1 (Ω1 )Y 2 2 −1 (Ω2 ) . of (6.38) yields pression 2( 1 + 2 )Y 1 + 2 1 + 2 −1 ( 1 . | . | . G 1 + 2 −1 1 + 2 −1 in (6.s. One can readily show that (6. 1 + 2 − 2. 2 = 1 in Table 1.6. M = −( + ). all the Clebsch-Gordan coefficients ( 1 + 2 M | 1 m1 2 m2 ). . . 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.27). To demonstrate (6. Normalization yields furthermore Y = 1 1 + 2 −1 1 + 2 −1 We have thereby determined ( 1 2 |Ω1 Ω2 ) (6. . ). We seek for this purpose first an unnormalized solution G 1 + 2 −1 1 + 2 −1 of (6. Exercise 6. . Having constructed this family we have determined the Clebsch-Gordan coefficients ( 1 + 2 − 1M | 1 m1 2 m2 ). 1 + 2 − 1. . Ω2 ). 2 |Ω1 . and all total angular momentum functions for a given choice of 1 and 2 .38) yields with J− = J− + J− the ex√ √ 2 1 Y 1 1 −1 (Ω1 )Y 2 2 (Ω2 ) + 2 2 Y 1 1 −1 (Ω1 )Y 2 2 −1 (Ω2 ). The reader should familiarize himself with the entries of the Table.38) to construct the family of functions B + = {Y + M ( . The l. . . ).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 ) (6.2.s. 2 |Ω1 . 1 + 2 − 3.1: Construct following the procedure above the three functions YJM ( 1 . .37) is satisfied. The result is illustrated for the case 1 = 2. . Ω2) for M = 1 + 2 − 2 and J = 1 + 2 . One can obviously continue the construction outlined to determine Y 1 + 2 −2 1 + 2 −2 . 6. Show that the resulting functions are orthonormal. ( + − )}. etc.37). in particular. Expression (6. 2 |Ω1 .39) as required by (6.36.47) can serve now to obtain recursively the elements of the family B + − for M = 1 + 2 − 2. . We like to construct now the family of total angular momentum functions B + − = {Y + − M ( .45) for some unknown constant c.h. One obtains then Y 1 1+ 2 1 + 2 1 + 2 −1 2 2 ( 1. of (6. −( 1 + 2 − 1).

632456 −0. Ω2) Y32 (2. 1|Ω1 .816497 −0.57735 3 10 3 5 Y2−1 (Ω1 )Y11 (Ω2 ) Y20 (Ω1 )Y10 (Ω2 ) Y21 (Ω1 )Y1−1 (Ω2 ) −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.707107 0. Ω2) Y20 (2.408248 −0.632456 0.547723 0. 1|Ω1 . 1|Ω1 .707107 0.57735 0. 1|Ω1 .774597 0 2 5 8 15 1 6 3 10 1 3 0. Ω2) Y3−1 (2. 1|Ω1 .707107 0.816497 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. 1|Ω1 .447214 0.57735 0.774597 − − − − 0.316228 0. Ω2) Y21 (2.730297 0.774597 0.57735 0.730297 −0. Ω2) Y2−2 (2.632456 0. . Ω2) Y31 (2. 1|Ω1 .547723 0. Ω2) Y1−1 (2.57735 −0. Ω2) Y3−2 (2.547723 0. 1|Ω1 .816497 0. 1|Ω1 . Ω2) Y2−1 (2. 1|Ω1 . Ω2) Y3−3 (2. Ω2) Y22 (2. Ω2) Y30 (2.816497 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 6.57735 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 0.258199 −0. Ω2) Y11 (2.447214 −0. 1|Ω1 .707107 0.408248 −0. 1|Ω1 .316228 0. Ω2) Y10 (2.547723 0. 1|Ω1 . 1|Ω1 .150 Addition of Angular Momenta and Spin Y33 (2. 1|Ω1 .258199 0.

we are dealing now with two particles carrying angular momentum or spin. bζ .6. Accordingly. b† ζ ζ = 0 = 0 .53) (6.ζ a† < ζ |σk | ζ > aζ ζ b† < ζ |σk | ζ > bζ ζ (6. 6.3: Explicit Expressions of CG–Coefficients 151 Exercise 6. hence.10 [cf. bζ = a† . For this purpsose we will employ the spinor operators introduced in Sections 5. aζ bζ . (6.48) (6. 1 .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 . m + 1 . a† ζ ζ b† . b† ζ ζ = aζ .50) (6. M | .10. 1 (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. a† ζ . a† and bζ . (6.52) and | 2 m2 2 According to Section 5. The creation and annihilation operators are again of the boson type with commutation properties aζ . bζ = = a† . b† refer to different particles and. ζ = ± and the matrix elements < ζ |σk | ζ > are as defined in Section 5. M | . (5. b † ζ = δζζ = δζζ . aζ .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 ).2: Use the construction for Clebsch-Gordan coefficients above to prove the following formulas  J+M 1  for = J − 2 2J 1 1 1 = J.9. J. we extent definition (5. b† ζ = 0.54) 2 + = √ b† ( 2 +m2 )! √− ( 2 −m2 )! . m − 2 . commute with each other ζ ζ aζ .10.2. 5.9. 5. 2 .51) The operators aζ .287) to two particles (1) J k k = = (2) J 1 2 1 2 ζ. These expressions will be derived now.10 where we studied single particle angular momentum and spin. + 2  − J−M +1 for = J + 1 2J+2 2  J−M  for = J − 1 2J 2 = .ζ ζ. Definition of Spinor Operators for Two Particles In contrast to Sections 5.49) where ζ. It is actually possible to state explicit expressions for any single coefficient (JM | 1 m1 2 m2 ).

300). mj j and. k = 1. 2 )).18) as follows = | 1 m1 .66) (6. M. 5. M ( 1 . 1. we like to recall for future reference that the operators (5. (5. M ( 1 .60) (6. ˆ kj |J. M ( 2) 2) = J(J + 1) |J. 2 . 2 .303). as usual are denoted by their respective quantum numbers J. which describe a two particle system according to (6. The operator of the total angular momentum/spin of the two particle system is J = with Cartesian components Jk = (1) J + J (6. 2) 2) (6. 2) . 2. M ( 1 . M ( 1 .56) The states | 1 . can be expressed in terms of the number operators 1 † 1 ˆ ˆ a† a+ + a† a− .58) Jk + (2) Jk .302. J3 . M ( 1 .61) (6. 6. m1 1 | 2 . (1) J 2 |J. (6. m1 . 2 . j = 1. −J ≤ M ≤ J . The aim of the present Section is to determine closed expressions for the Clebsch–Gordan coefficents ( 1 . M ( (2) J 2 |J. 2) + 1) |J.69) − 2| ≤ J ≤ + 2 . for such states should hold J2 |J. M ( 1 . 2) 2) can be expressed in terms of Clebsch-Gordan coefficients (6.152 It holds. (6. M ( 1 . = M |J. (1) J (2) J 3 | 1 m1 1 3 | 2 m2 2 = m1 | 1 m1 = m2 | 2 m2 1 2 (6.59) We seek to determine states |J. + 1) |J. 2 )) . 2 .63) according to At this point. M ( 1 . (j) 2 ˆ ˆ J = kj ( kj + 1 ) .64) (6. m2 2 .65) (6.. = = 1( 1 2( 2 2) 2) (6. m1 1 | 2 . M ( 1 . (6. hence.54). m1 1 | 2 .53. k2 = b b+ + b† b− k1 = + − − 2 2 + namely. in analogy to Eqs. (2) J which. m2 |J. i.m2 | 1 . M ( 1 . are | 1 . 2 ) which are simultaneous eigenstates of the operators 2 2 J2 .67) = j |J. m2 1 2 ( 1 . mj j = j | j . 3 .e. M ( 1 . (1) J 2 .62) . (1) J .55) . (6. J3 |J. m2 a† + ( 1 2 = a† − ( 1 1 −m1 (6. |J. ˆ For the operators kj holds ˆ kj | j .57) b† + ( (2) 2 2 +m2 1 +m1 b† − ( 2 2 −m2 + m1 )! − m1 )! (1) + m2 )! − m2 )! |Ψo . m2 |J. 2) 1.59) J± = J1 ± i J2 . m1 . (6. (2) J 2 . (1) J 2 | (2) J 2 | 1 m1 1 2 m2 2 Addition of Angular Momentum and Spin = = 1( 1 2( 2 + 1) | 1 m1 + 1) | 2 m2 1 2 . 2) We will also require below the raising and lowering operators associated with the total angular momentum operator (6. M ( 1 . M ( 1 .68) The states |J. . 1 . .

a† b† + a† a− . 4 Using J3 = (1) J 3 + (2) J 3 . K † (kj + 1) 2 2 1ˆ † 1 ˆ kj K + K † (kj + 1) 2 2 1 ˆ 1 ˆ K † kj + kj . This operator obeys the following commutation relationships with the other pertinent angular momentum / spin operators ˆ kj . K † 2 1ˆ ˆ 1 ˆ ˆ kj kj + 1. b† = 0 . ˆ A similar calculation yields [k2 .73.6.3: Explicit Expressions of CG–Coefficients The Operator K † The following operator K † = a† b† − a† b† . 2 4 = 0 = = 0.75) The relationships (6. K † + K † 2 2 3 † †ˆ K kj + K . 2 2 3 ˆ = K † kj + K † . K † 1 a† a+ − a† a− + b† b+ − b† b− . K† J3 . We note that.71–6. and applying the relationships (6. 6. using (6. a† b† − a† a− . K † (j) J 2 . (6.73) (6. K † J± . j = 1. b† − b† a† b− . expressing (k) J 3 through the creation and annihilation operators according to (5.71–6. + − + − 2 2 − + = . (6.64. due to J2 = K† 1 2 J+ J− 1 † K .70) will play a crucial role in the evaluation of the Clebsch-Gordan-Coefficients. j = 1. 6. a† b† − a† b† + − + − + − − + 2 1 1 = a† a+ . a† b† + + − − + 2 2 − 1 1 = a† b† − a† b† = K† . a† b† + − − + 2 + 2 − 1 1 − b† a† b+ .72) (6.50.74) + 1 J− J+ + J2 . a† b† − a† b† − + − − + 2 + 1 1 = a† a+ . K † 1 † a a+ + a† a− .71) (6. For example. K † = 0. the relationships (6.73) yields J3 .73) can be readily proven.65) and (6. Employing (6.71) one can show K† = = = = = 1 ˆ ˆ kj (kj + 1). K † + kj . + − − + 2 2 = 1 † 2K .288).51) one obtains ˆ k1 . K † ] = (j) J 2 .74) imply 3 2 J2 . + − − + 153 (6. 6.

2) 2) ˆ K † kj + = K† = j 3 † K + K † j ( j + 1) |J.77) where N is another normalization constant. 2 ) j + 4 1 3 + K † |J. 2 ) produces again total angular momentum eigenstates to the same J and M quantum numbers of J2 and J3 .78) a state which has been used already in the construction of Clebsch-Gordan coefficients in Section 6. j + 2 2 2) However.72) and (6. M ( 1 . j1 + j2 (j1 . To demonstrate that the resulting states are eigenstates of (1) J 2 and (2) J 2 we exploit (6. Application of (K † )n to this state yields. 6. Action of K † on the states |J. 2 2 + n ) 2 (6. M ( 1 + n . j1 + j2 (j1 . (6. 2 ) . = − a† a− .76) and state K† n |J. M ( 1 . M ( 1 . M ( 1 . j1 1 + . M ( 1 .. it holds K † |J. 1 2 + 2 . 6. M ( 1 .75) ascertain that under the action of K † the states |J.77). One can generalize property (6.73. The commutation properties (6. M ( 1 . j2 ) = |j1 . 6. M ( 1. j1 1 |j2 . j2 + ) 2 2 n n = N (n. 2) is a state with quantum numbers 1 1 + 1 2 and = N |J. |j1 + j2 .e.62. K† + K † (j) J 2 |J. We consider now the case that (K † )n acts on the simplest total angular momentum / spin state. M ( 1 .79) n |j1 + j2 . a† b− − a† b† + + + − + = 0. a† b† + b† a† b− . j1 + ) K † 2 2 (6. but for different 1 and 2 quantum numbers of the operators (1) J 2 and (2) J 2 . M ( 1 . j2 ) . namely. b† + − + + + − The property [J− .67) (j) J 2 K † |J. according to (6. 2) = = (j) 2 J . M ( 1 + .154 Starting from (5. 2) = N |J. on the state |j1 + j2 .292) one can derive similarly J+ . 2) We want to demonstrate now that the action of K † on the states |J. j1 + j2 (j1 + n n . 2 ) remain eigenstates of J2 and J3 with the same quantum numbers. K † ] = 0 is demonstrated in an analoguous way.2. 2 2 1 + ) . K † = Addition of Angular Momentum and Spin a† a− + b† b− . 2 (6. j2 2 . 4 3 + j ( j + 1) |J.76) Here N is an unknown normalization constant. this result implies that K † |J. i. M ( 1 .63.

86) b† + J+ 2 − 1 |Ψo . Strategy for Generating the States |J. (6. 2) = ∆(J. M ) (J− )J−M |J. in analogy to the construction (5.83) The derivation of this expression will be provided further below (see page 158 ff). holds |J. j1 . M ( 1 . j1 + n .57) and exploiting the fact that J− and K † commute [c. 2 ) can be expressed through the r. (6. J( 1 . − 1. (6. = (J + M )! (2J)! (J − M )! 1 2 2) (6.79) n. 2) 2) . 2 = j2 + n 2 (6. M ( 1 .f. M ( 1 . 2 2 It is now important to notice that any state of the type |J. M ( 1 .82) 1 × | (J + 2 1 × | (J + 2 1 (J + 2 1 − 1 ) . (6.6.h. J( 1 . J( 1 . M ( 1 .80) = = = 1 1 2 2 − J 1 n − = (J + 2 2 n 1 − = (J + 2 2 + 1 1 − − 2) 2) . M ) (J + × K† 1 − 2 )! (J + 2 − 1 )! J+ 1 − 1 + 2 −J (J− )J−M a† + × 2 (6. 1.105) of the spherical harmonics. Combining (6. 6. j2 as follows J = j1 + j2 which is equivalent to n j1 j2 Accodingly. 2) Our construction of the states |J. (J + 2 − − 2) 1 1) 2 The normalization constant appearing here is actually N( + − J.81) = N( 1 + 2 1 2 − J. 2 ) for −J ≤ M ≤ J as follows |J. 5.85) ∆(J.3: Explicit Expressions of CG–Coefficients 155 where we denoted the associated normalization constant by N (n.79). j2 + n ). 2) 1 2 = ( 1 + 2 (2J + 1)! − J)! ( 1 + 2 + J + 1)! 1 2 . 2) = N( 1 + 2 − J. 2 ) . For this purpose one needs to choose in (6.84) with (6. 2) .82) for |J. 2 ) exploits the expression (6. 1 .82. allow one to obtain the states |J. The latter states.104.74)] yields |J. 1 = j1 + n 2 . K† 1 2 1 + 2 −J × × . of (6. M ) .84) (6. 2 ) ∆(J.s. J( 1 .

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. in analogy to (5.87) + − + − i.86) in terms of monomials 1 +m1 1 −m1 2 +m2 2 −m2 a† a† b† b† |Ψo . r!s!t! r.t (6. x.50–6.e.86) |Gr = Jr a† st − + s b† + t |Ψo (6. y) is a generating function for the states |Gr defined in (6. (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 .93) = f (a† + λ a† ) g(b† + λ b† ) |Ψo . x.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. 2 [cf.s. m1 1 | 2 .92) − − and noting [a+ . + + Taylor expansion of the two exponential operators yields immediately I(λ. y) = λr xs y t r |Gst .90) (6. b† ] = 0 [cf. a† ] = [b+ .89) i.. follows from the commutation properties (6.87). (6. m2 Gordan-coefficients.57)].88). st The desired expansion of |Gr can be obtained from an alternate evaluation of I(λ. x.264).e. y) = exp (λJ− ) exp x a† exp x b† |Ψo .52). I(λ. (6.156 Addition of Angular Momentum and Spin Our strategy for the evaluation of the Clebsch-Gordan-coefficients is to expand (6. One obtains then using J− = a† a+ + b† b+ (6. x. For this purpose we introduce the generating function I(λ.91) which.88) in terms of monomials (6.. Comparision with (6.

2) = s. y) = s.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.s. M ( 1 .t = q r! × a† + s q s−q t r−q a† − q × b† + t−r+q b† − r−q |Ψo (6. x. Expanding the exponentials in st (6. using the commutation property (6. 2 ) in terms of states | 1 . Operation of this operator on (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.t × a† + s−q s q b† + t r−q t−r+q a† − q b† − |Ψo (6. one can write the right factor in (6.98) a† − s a† + b† + s . x.3: Explicit Expressions of CG–Coefficients We conclude I(λ.96) and.90) allows one to infer |Gr s.86) (J− )J−M a† + = q J+ 1 − 2 b† + J+ 2 − 1 |Ψo 1 (6. With K † given by (6.q (−1)s (6. 157 (6.94) yields I(λ.6.94) One can infer from this result the desired expressions for |Gr .50).95) Comparision with (6. using the definition (6. m1 1 | 2 .97). |J.70) holds K† 1 + 2 −J = s 1 + 2 −J b† − s 1 + 2 −J−s (−1)s 1 + 2 −J−s (6. to obtain the desired expansion of |J. M ( 1 . m2 2 .97) yields. y) = exp a† + λ a† + − exp b† + λ b† + − |Ψo .t × λ v b† − × r−q v × = t s−u v λu a† − r! q |Ψo λr xs y t r! s! t! r.99) .88).

j2 . 2 ) we consider the scalar product J. m2 = M − 1 + q + s. The summation over q corresponds then to the summation over m1 . 2 ) − 2) .82) and (6. M − +q+s One can conclude that this expression reproduces (6. 1 .83) of the normalization constant N ( 1 + defined through (6.102) We want to determine now the expression (6.69) since.6. M ( 1 .54) between creation operator monomials and angular momentum states allow one to write this |J.103) = 1. √ 2J + 1 1 2 . 2 ) (6. − 1 . n)|ψ(j1 . The Clebsch-Gordan coefficents are then finally ( 1 . j2 = 1 (J + 2 2 − 1) . according to (6.158 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 (6.104) .103) this can be written 1 = N 2 ψ(j1 . n = 1 + 2 −J. J( 1 . (6.101). 2 ) ∆(J. 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. Using (6. (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 .69) if one identifies m1 = 1 −q − s . q = 1 − m1 − s and m2 = M − m1 . M ) (J + 2 )! (J + 2 − 1 )! s.82). To determine N = N ( 1 + 2 −J. j2 . m2 |J. m2 in (6. J( 1 .q (−1)s × (6. n) 2 )|J. m1 . 1 . × 2) = N( 1 + 2 1 − J. For this purpose we introduce j1 = 1 (J + 2 1 2 − J.53.101) Note that m1 + m2 = M holds. − q − s)!(q + s)! 2 × (M + 1 − 1 + q + s)!( 1 1 + 2 − M − q − s)! 2 − q − s 1 | 2.

108). m1 1 |j2 .108) = λ=0 (n1 + ν)! n1 ! (6. n) = 1 (2j1 )!(2j2 )! a† + n! (2j1 )!(2j2 )! a† + 2j1 −n−s s 2j1 s n s 2j2 a† b† + − |Ψo = n−s (−1)s a† b† − + s b† + (−1)s a† − s (2j1 + n − s)!s!(2j2 + s)!(n − s)! s!(n − s)! 2j2 +s b† + b† − n−s (2j1 + n − s)!s!(2j2 + s)!(n − s)! |Ψo .106) The orthonormality of the states occurring in the last expression allows one to write (6. One obtains then. j2 .112) s .3: Explicit Expressions of CG–Coefficients where |ψ(j1 .6. 1 1−λ n1 +1 1 1−λ n2 +1 = m1 . j2 . (6.57) for these states and the expression (6. j1 1 |j2 .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.m2 n1 + m1 n1 n2 + m2 n2 λm1 +m2 (6.109) and Taylor expansion of the left hand side of (6. j2 2 . n) in terms of states |j1 . applying (6. (6. Accordingly.104) 1 = N2 (n!)2 (2j1 )!(2j2 )! (2j1 + n − s)!(2j2 + s)! s!(n − s)! 2j2 + s 2j2 (6.110) which can be written 1 1−λ n1 +n2 +2 = r s n1 + r − s n1 n2 + s n2 λr (6.108) twice. n) = K† n 159 |j1 . we obtain |ψ(j1 .70) for the operator K † .108) yields the identity n1 + r − s n1 n2 + s n2 = n1 + n2 + r + 1 n1 + n2 + 1 . j2 . (6. We employ the expression (6.111) Comparision with (6.105) The first step of our calculation is the expansion of ψ(j1 . m2 2 .

and J (2) are vectors in R . 1 .116) one expresses the Clebsch-Gordan coefficient on the r.103) one obtains the desired result (6.83). −M ) . vice versa. J.115) This relationship is a trivial consequence of (6. M ) = (−1) 1 + 2 −J ( 2 . m2 |J. m2 |J.113) Using the identities (6. (6. m1 . (6. of (6.120) The symmetry properties (6. −M ) = (−1) 1 + 2 −J ( 1 . (6. 2 . of (6. 2 .s. m2 |J. To derive relationship (6.118). (6. m2 ) by ( 1 .s. M | 1 .118) Finally.120) can be readily derived from the expression (6. 6.h. m1 . and (6. m1 ) by ( 2 . −m1 . J (1) on the r.114) one obtains − J = − J (1) − J (2) The respective Clebsch-Gordan coefficients ( 1 . ( 2 . J (1) . M ) corresponding to (6. m1 |J. m2 |J. −m1 2|J. 2 .114) For example. 2 . interchanging the operators J (1) and J (2) results in J = J (2) + J (1) . M ) corresponding to (6.g. −m1 . We will demonstrate this now.119) The corresponding symmetry property of the Clebsch-Gordan coefficients is ( 1 . M ) .116) If one takes the negatives of the operators in (6.102) of the Clebsch-Gordan coefficients. m2 |J. m2 ) and.4 Symmetries of the Clebsch-Gordan Coefficients The Clebsch-Gordan coefficients obey symmetry properties which reflect geometrical aspects of the operator relationship (6.102) by replacing ( 1 .116). one can interchange also the operator J on the l. M ) = (−1) 2 +m2 2J + 1 ( 2 . m1 .115). M ) ( 1 . m1 ) . For the quantum mechanical addition of angular momenta the Clebsch Gordan coefficients ( 1 .h.h. 2 . (6. m2 . (6. −m2 . m1 . (2j1 + 2j2 + 1)! (6. m1 ).114) show a simple relationship to the Clebsch Gordan coefficients ( 2 . of this equation J (1) = J (2) − J . M ) . m1 |J. m2 .114) as long as J.. (6. 2 .s. −m2 |J.117) are again related in a simple 2 .160 Applying this to (6.107) 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 . namely. and .114) by.11) J = J (1) + J (2) . 1 . m1 . manner to the coefficients ( 1 .116) through formula (6. 2 1+1 (6. e. ( 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

180) where the sign could depend on j.186) Here (( · · · )) denotes again confinement of the diffusion operator ∂r to within the double bracket. have proven (6. φ) and (5. ˆ 2 2 2 2 2 2 for θ = 0. for θ = 0 (6. 1 |ˆ) r r r 2 2 2 2 2 2 (6. identified the sign of (6. 1 |θ. ˆ (6. given by ˆ σ · r = σ1 sin θ cos φ + σ2 sin θ sin φ + σ3 cos θ .181) 0 = j+ 1 2 4π Yj 1 (j + 1 . 1 |ˆ) r 2 (6.148) the particular ˆ 1 1 states Yj 1 (j − 2 . 6. 1 |θ = 0. φ) 2 2 2 = − j+ 1 2 4π (6. 1 |θ = 0.147.187) 1 2 . ˆ (6. 1 |ˆ) + f (r) σ · p Yjm (j ± 1 . φ) 2 2 2 .174–5. 1 |ˆ) at θ = 0 are r r 2 2 2 2 Yj 1 (j − 1 . therefore. 1 |θ = 0. φ) = − Yj 1 (j + 1 . We want to demonstrate finally that the “−”-sign holds in (6.1 states χ 1 ± 1 2 2 2 [c. 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 .180). 1 |θ = 0.183) in case θ = 0 becomes in the standard representation with respect to the spin.224)] 1 0 σ·r = ˆ .180) and (6. According to (5.182) σ · r Yj 1 (j − 1 . Since f (r) is independent of θ and φ follows ˆ ((σ · p f (r))) = −i (( ∂r f (r) )) σ · r . 1 |ˆ) = ((σ · p f (r)))Yjm (j ± 1 .181. 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) . (6.180) and. (5.168 Addition of Angular Momentum and Spin 1 Yjm (j ± 2 .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. (6.160).182) 0 Since σ · r. One can conclude then ˆ σ·r = ± ˆ 0 1 1 0 (6.f.185) We have. 1 |ˆ).188) . For this purpose we consider the application of σ · r in the case of θ = 0.184) 0 1 space χ 1±1 2 2 one can conclude from (6. 6. 2 |ˆ) and Yj 1 (j + 1 . hence. We want to determine ˆ its action on the wave functions f (r) Yjm (j ± 1 . The result can also be stated in the compact form σ · r Yjm (j ± 1 .

1 |ˆ) r 2 = i ∂r + −j r 1 g(r) Yjm (j + 2 . 1 |ˆ) r 2 (6.155. 1 |ˆ) .156) this yields finally 1 ˆ σ · p f (r) Yjm (j + 2 . To demonstrate this we consider one of its cartision components.195) .196) = i j+ ∂r + r 1 2 3 2 1 f (r) Yjm (j − 2 . 1 |ˆ) = r 2 2 ˆ i ∂r f (r) − f (r) σ · p σ · r Yjm (j ˆ 1 2 169 . 1 |ˆ) = i [ ∂r + r + ] f (r) Yjm (j 2 r r Using (6.197) 1 ˆ σ · p g(r) Yjm (j − 2 . we conclude the intuitively expected identity 1 ˆ ˆ p×r = − J .192) ˆ ˆ The operator p × r in (6. 1 |ˆ) = 0 and σ = 2S/ .230) allows one to express ˆ ˆ ˆ ˆ ˆ σ · p σ · r = p · r + i σ · (p × r) .53). r 2 2 1 J ·S 2 1 ˆ σ · p f (r) Yjm (j ± 2 .190) can be related to the angular momentum operator. ∂3 (1/r) = −x3 /r3 and ∂2 x3 = x3 ∂2 . h r (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 (6.5: Spin–Orbital Angular Momentum States Using (6.193) r r 1 1 1 = −i (∂2 x3 − ∂3 x2 ) h − i h (x3 ∂2 − x2 ∂3 ) . Corresponding results are obtained for the other components of p × r and.g.6. r r r Using ∂2 (1/r) = −x2 /r3 .187) one obtains ˆ σ · p f (r) Yjm (j ± 1 .194) ˆ ˆ where J1 is defined in (5. 6. r Altogether we obtain.191) (6. e. using ∂r Yjm (j ± 1 . 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.186) for both terms in (6.189) The celebrated property of the Pauli matrices (5. r 2 (6.190) 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. hence. x3 x2 ˆ ˆ ( p × r )1 h = −i (∂2 − ∂3 ) h (6..198) . 1 |ˆ) r 2 1 2 (6. 1 |ˆ) r 2 (6. 1 |ˆ) r 2 (6.

For this purpose we use the construction method introduced in Sec.146). 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. 6.143–6. in fact. 1 |ˆ) r 2 2 r r2 and.205) which agrees with the expressions (6.206) . are consistent with this identity. 6. according to (6. − |j.e. 5. 2 2 (6.170 Addition of Angular Momentum and Spin To demonstrate the validity of this key result we note that according to (5.2.144) for m = j.199) We want to show that eqs.203) (6. 6. 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 .199). j + . 1 |ˆ) r r 2 2 2 2 2 2 2 yields ˆ (σ · p)2 f (r) Yjm (j + 1 .. as well as (6. r (6. 1 |ˆ) r 2 2 i 3 ˆ = σ · p [ i∂r + (j + ) ] f (r) Yjm (j − 1 .143. 1 |ˆ) = Yj− 1 j− 1 (ˆ) χ 1 1 . adopting the method in Sec. 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 .204) (6. 1 |ˆ) = J 2 Yjm (j + 1 . j 2 + 2j + 2 3 4 (6. We note ˆ (σ · p)2 f (r) Yjm (j + 1 . 1 |ˆ) r 2 2 The Clebsch-Gordan coefficients are then (j − 1 .198). according to (6.200) = (j + 1 )(j + 3 ).101).197. 1 |j. consider first the case of the largest m-value m = j. We begin with the coefficients (6. r 2 2 2 2 2 2 − r 2 ∂r r + J2 r2 f (r) Yjm (j + 1 . . In this case holds. Yjj−1 (j − 1 . j − 1 .43). 6. using (5. i. Evaluation of Relevant Clebsch-Gordan Coefficients We want to determine now the Clebsch-Gordan coefficients (6.2. 6. 1 .201) 1 3 (j + )(j + ) Yjm (j + 1 .144) and.143. For m = j − 1 one can state.202) Yjj (j − 1 . r ∂r2 r 2 (6.45). 1 |ˆ) r 2 2 r r2 r2 j 2 + 2j + 3 2 2 2 4 = [ −∂r − ∂r + ] f (r) Yjm (j + 1 . (6. 1 |ˆ) = r 2 2 which agrees with (6. j) 2 2 2 2 (j − . j) 1 2 1 2 1 2 1 2 = = 1 0 (6.151).230.99) holds ˆ (σ · p)2 = − 2 2 = ∂2 J2 r + 2 .

[c.144) for the Clebsch-Gordan coefficients by induction.143. 1 .202).s. and J− + S− to the r. m − 1) 2 2 2 2 = j +m−1 2j (6. Expression (6. The further Clebsch-Gordan coefficients (· · · |jj − 2). 1 .f.206). (6.6. (· · · |jj − 3). (6. 6. 2 2 2 2 2j .209) (j − 1 . 6. j − 1) 2 2 2 171 = = 2j − 1 2j 1 2j (6. j − 1 . j − 1) 2 2 2 2 which again agrees with the expressions (6. j − 3 . (6. Let us verify then the expression (6.210) to (6. 1 |ˆ) = r 2 2 Applying J− (tot) j+m Yj− 1 m− 1 χ 1 1 + 2 2 2 2 2j j−m Yj− 1 m+ 1 χ 1 − 1 .2.144) implies for j = m Yjm (j − 1 . 2 2 2 2 2j (6. is obtained by applying the operator [c. 1 |j.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 .143. m − 3 .f.5: Spin–Orbital Angular Momentum States The corresponding Clebsch-Gordan coefficients are then 1 (j − 1 .137)] J− (tot) = J− + S− (6.211) to the l.143. are obtained by iterating the application of (6.. 1 |ˆ) = r 2 2 j+m−1 Yj− 1 m− 3 χ 1 1 + 2 2 22 2j (j − m) = This implies (j − 1 . 2 |j. etc.210).144) for m = j − 1. 1 |ˆ) = r 2 2 j+m 2j + (j + m − 1)(j − m + 1) Yj− 1 m− 3 χ 1 1 2 2 2 2 j+m Yj− 1 m− 1 χ 1 − 1 2 2 2 2 2j j−m 2j (j + m)(j − m) Yj− 1 m− 1 χ 1 − 1 2 2 2 2 + or Yjm−1 (j − 1 . 6.h.210)] yields (j + m)(j − m + 1) Yjm−1 (j − 1 . 1 .208) (6. − 1 |j. as described in Sec.207) (6.s.h. 6.

m) = 2 2 2 2 Similarly.144). 1 |j + 1 . j2 m2 ) 2 . (1) (2) (−tot) Obviously. m3 .215) which follows from (6. m + 1 . all three vectors Jclass . m. − 1 |j.218) where we have replaced the quantum numbers J. 2 . one obtains (j + 1 . 1 . 1 . 1 . 6. 2 . The latter relationships applied together read ( 1 . 1 . m + 1 ) = 2 2 2 j+m+1 . M. 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. j2 . −m2 | 1 . m) 2 2 2 2 2j + 1 (6. The Clebsch-Gordan coefficients (6.116. 6. − 1 |j. one can obtain expression (6.143) by applying the symmetry relationships (6. to reflect in the notation the symmetry of these quantities. m1 . In this context Jclass and Jclass play the same role. m2 1 (6.144) for j = m − 1.143.120).144) by induction. m2 . − 1 |j. m + 1 . m. m1 . m − 1) 2 2 2 2 = (6. Jclass and Jclass play equivalent roles.143). m1 . m − 1 . m1 ) (6.146) can be obtained from (6. 2j + 1 (6.145) from (6.217) 2j + 2 (j + 1 .215).214) yields (j. m3 . m + 1 ) = 2 2 2 2 and.172 Addition of Angular Momentum and Spin j −m+1 2j (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 . 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 . 2j + 2 (6. 1 |j + 1 .216) 6. 1 . hence.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 . m3 ) = (−1) × 2 + 3 − 1 + 2 +m2 × 2 .213) which is in agreement with (6.214) 2 2 +1 ( 3 . m2 | 3 . (1) (2) (−tot) The coefficients which are the quantum mechanical equivalent to Jclass + Jclass + Jclass = 0 are the 3j–coefficients introduced by Wigner. 1+1 3 For 1 (j. 6. using (6. the relationship (6. proven (6.143. j3 . by the set j1 . We have. j+m+1 . However.

as follows   −j1 + j2 + j3 j1 − j2 + j3 j1 + j2 − j3 j1 j2 j3 j1 − m1 j2 − m2 j3 − m3  . m3 . e. m2 . −j1 + j2 + j3 . However.6..e. In this way the values of 12 3j-coefficients are related. The two integer entries J − 1 − m2 and J − 2 + m1 in (6.21. m1 . The reader may note that the entries of the Regge-symbol.102) are obtained each through the difference of two entries of the Regge-symbol.222) m1 m2 m3 j1 + m1 j2 + m2 j3 + m3 The Regge symbol vanishes. For this purpose one can use the symmetry transformations to place the smallest element into the upper left corner of the Regge symbol.6: The 3j–Coefficients We first like to point out that conditions (6. except when all elements are non-negative integers and each row and column has the same integer value Σ = j1 + j2 + j3 . 6. The Regge symbol reflects a remarkable degree of symmetry of the related 3j–coefficients: One can exchange rows. m2 and j3 . the Regge-symbol. one can exchange columns and one can reflect at the diagonal (transposition). Because of its high degree of symmetry the Regge symbol is very suited for numerical evaluations of the 3j–coefficents. These symmetry operations relate altogether 72 3j–coefficients. The Regge symbol also vanishes in case that two rows or columns are identical and Σ is an odd integer. 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. the so-called triangle condition. with respect to alltogether 12 symmetry operations. 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. =  (6.102) of the Clebsch-Gordan coefficients.223) (Σ + 1)!n11 !n22 !n33 !n23 !n32 !)) n=0 n31 n32 n33 .g.221) j1 j2 j3 −m1 −m2 −m3 where the the results of a cyclic. j2 . According to the definition of the 3j–coefficients one would expect symmetry properties with respect to exchange of j1 .219) (6. for which no known classical analogue exists. states that j1 . j2 . 173 (6.218) relating 3jcoefficients and Clebsch-Gordan coefficients. i. √ safe for the prefactor 2j3 + 1 which is cancelled according to the definition (6. discovered by Regge. are identical to the integer arguments which enter the analytical expression (6. To represent the full symmetry one expresses the 3j–coefficients through a 3 × 3–matrix. m3 and with respect to sign reversals of all three values m1 .220) The latter condition |j1 − j2 | ≤ j3 ≤ j1 + j2 . j3 form the sides of a triangle and the condition is symmetric in the three quantum numbers. there exist even further symmetry properties. anti-cyclic exchange and of a sign reversal are stated.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 . In case of a non-cyclic exchange of rows and columns the 3j–coefficent assumes a prefactor (−1)Σ .

m1 + m2 ) (2 − m1 )! (2 + m1 )! √ 10 (1 − m)! (1 + m)! (1 − m2 )! (1 + m2 )! 1 ≥ |m| ∧ 1 ≥ |m2 | ∧ 2 ≥ |m1 | (6. m2 ) √ 2 1 ≥ |m1 |) 2 (6.223-6.229) 1 1 (0m| m1 m2 ) = 2 2 i(−1)1−m2 δ(0. 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 (6.228) √ (−1)2m1 −2m2 7 δ(m.1.232) .224) (6.231) Here is an explicit value of a Clebsch-Gordan coefficient: 1 1 (70 2 −15 1 |120 −10 60 2 −5 1 ) = 2 2 √ √ √ 4793185293503147294940209340 127 142 35834261990081573635135027068718971996984731222241334046198355 0.10752786393409395427444450130056540562826159542886 We also illustrate the numerical values of a sequence of 3j-coefficients in Figure 6. m1 + m2 ) (2 − m1 )! (2 + m1 )! √ 6 (2 − m)! (2 + m)! (1 − m2 )! (1 + m2 )! 1 ≥ |m2 | ∧ 2 ≥ |m| ∧ 2 ≥ |m1 | (6. (6.225) sn−1 .226) We like to state finally a few explicit analytical expressions for Clebsch-Gordan coefficients which were actually obtained using (6.226) by means of a symbolic manipulation package (Mathematica) (1m|2m1 1m2 ) = (−1)1+m+m1 δ(m.230) 1 1 (1m| m1 m2 ) = 2 2 √ (−1)2m1 −2m2 3 δ(m. m1 + m2 ) (3 − m)! (3 + m)! (3m|2m1 1m2 ) = √ 105 (2 − m1 )! (2 + m1 )! (1 − m2 )! (1 + m2 )! 1 ≥ |m2 |) ∧ 2 ≥ |m1 | ∧ 3 ≥ |m| (6. (6.227) (2m|2m1 1m2 ) = (−1)m+m1 (m + 2m2 ) δ(m. m) δ(−m1 .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.

. .+ 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 . (2 (2 1 2 + 1) (2 + 1) × (2 1 2 + 1) + 1) .5). Ω2 ) provide the irreducible representation for the rotation R(ϑ) defined in (6.233) . m2 = − 2 .e. Similarly.. m1 = − 1 . rotations in 2–particle function space. . Irreducible Representation We had stated above that the spherical harmonics Y m (Ω)). 2 = 1. i. .e. . . If we define the matrix representation of R(ϑ) by D(ϑ). 1 . provide the irreducible representation for D(ϑ).1: Oscillatory behavior of 3j-coefficients.             (6.6: 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. . .              . the rotations in single particle function space. . . eigenfunctions of the single particle angular momentum operators J 2 and J3 . 2 |Ω1 .6. . 2. then for a basis {Y 1 m1 (Ω1 )Y 2 m2 (Ω2 ). i. the 2–particle total angular momentum wave functions YJM ( 1 . .+ 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. .3: Two triplet states |1m1 (1) |1m2 (2) are coupled to an overall singlet state Y00 (1. 1 .6. each of the blocks in (6. (6. .6. 2.1: Prove Eqs. Ω2 ). 2 |Ω1 . .233) is further block-diagonalized as follows + 2. 1). Exercise 6. J} (2 (2 1 2 + 1) (2 + 1) × (2 1 2 + 1) + 1) = (2| 1 − 2 | + 1) ×(2| 1 − 2 | + 1) (2| 1 − 2 | + 3) ×(2| 1 − 2 | + 3) (6. M = −J. .234) . . . . Show that the probability of detecting a triplet substate |1m1 (1) for arbitrary polarization (m2 – 1 value) of the other triplet is 3 .6.234) Exercise 6. J = | 1 − 2 |.233. . . . . .6. (2( 1 + 2 ) + 1) ×(2( 1 + 2 ) + 1) Partitioning in smaller blocks is not possible. . 2 = 1..176 Addition of Angular Momentum and Spin 1 For the basis {YJM ( 1 . Exercise 6.

.222) in case (ii) of single particle spin operators. S 2 = S1 + S2 + S3 or any other polynomial of Sk .238) .g. e. 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).e. e. ∂x2 3 1 2 2 2 2 case (ii) the operator T could be a spin operator Sk defined in (5. The latter implies that T transforms like an angular momentum or spin state | m . In fact.7: Tensor Operators 177 6. that T belongs to a family of operators {Tkq . Note that in the examples mentioned the operator T would be defined within the same representation as R(ϑ). 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.237) are called tensor operators of rank k. k} such that k Tkq = q =−k (k) Dq q (ϑ) Tkq . .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.223). for example.236) In this equation Dq q (ϑ) denotes the rotation matrix Dq q (ϑ) = kq |R(ϑ)|kq . . (6. The operator T may also act on multi-particle states like Y 1 m1 (ˆ1 )Y 2 m2 (ˆ2 ). that in case (i) T is an operator C∞ ( ) → C∞ ( ) acting on single particle wave functions. the operator (5.6. (k) (6.g. i. q = −k. Examples of Tensor Operators The multiplicative operators C∞ ( ) → C∞ ( ) Y ( ) = def Y ( ) (6. −k + 1.42) in case (i) of the position representation of single particle wave functions or the operator (5.236. . −k + 1. (k) (6. . 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 . q = −k. e.g. k} with the transformation property (6. Rotations transform |ψ as |ψ = R(ϑ)|ψ and T |ψ as R(ϑ)T |ψ .237) The operators T ∈ {Tkq . The latter can be written T |ψ where T denotes T in the rotated frame given by T = R(ϑ) T R−1 (ϑ) . some of the r r examples considered below involve tensor operators T of this type. . This implies. 6.

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. In fact.235 . x3 .178 are tensor operators of rank k. 1. but rather any rotation and any parametrization can be assumed.7.7. Exercise 6.2: 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. 2.240) 1 where we have defined ϑ± = 2 (ϑ1 iϑ2 ) and L± = L1 ± iL2 .239) These operators can be expressed in terms of the coordinates x1 . Exercise 6. The fact that these operators form tensor operators of rank 1.6.203). This choice of parametrization allows us to derive conditions which are equivalent to the property (6.3.237). but are far easier to ascertain for any particular operator.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. For the following it is important to note that the rotation matrix elements (6. 3 follows from the transformation properties of the spherical harmonics derived in Section 1. 2. we will assume presently that the rotation is chosen as follows R(ϑ) = exp ( ϑ+ L+ + ϑ− L− + ϑ3 L3 ) (6. x2 . .237) do not require that the rotation R−1 (ϑ) is expressed in terms of Euler angles according to (5.

243) From (5.237) be assumed then.246.7. ϑ− . J− .7.4: Is the 1-particle Hamiltonian 2 H = − 2 2m + V (|r|) (6.7.6. Show that S (1) · S (2) is a tensor operator of rank 0 in the space of the products of the corresponding spin states | 1 m1 (1) | 1 m2 (2) . Expressing the results in terms of the angular momentum operators J+ . (k + q + 1)(k − q) (k + q)(k − q + 1) (6. (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 . ϑ3 . 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) . 1 (6. Tkq ] = −iTk q+1 (k + q + 1)(k − q) . Tkq ] = = δqq and by subtracting Tkq on both sides of the k ϑ+ kq |L+ |kq Tkq .237) yields first k R(ϑ) Tkq R −1 (ϑ) = q =−k kq |R(ϑ)|kq Tkq . ϑ− . (6. q =−k q+1 (6. The property (6.242) from which follows by means of kq |1 1|kq equation ϑ+ [L+ . 6. 0)T in case of small ϑ+ .235 . 0.6. (0.235 . 2 Exercise 6.7. Exercise 6. ϑ3 )T .80) follows kq |L+ |kq = −i (k + q + 1)(k − q)δq and. (6.6: Form tensor operators of rank 1 in terms of the three components of acting on the space of 1-particle wave functions.244) [L+ .6. Tkq ] = = = Tkq+1 Tkq−1 q Tkq . Tkq ] [J− . J3 yields [J+ . We first consider a rotation with ϑ = (ϑ+ .7: Tensor Operators 179 The conditions can be derived if we consider the property (6. Exercise 6.246) (6. Tkq ] [J3 .235 . hence. 0. Similar equations can be derived for infinitesimal rotations of the form ϑ = (0.6.1 particles for which the first spin is described by 2 the operator S (1) and the second spin by the operator S (2) .247).5: Consider a system of two spin.237) for transformations characterized by infinitesimal values of ϑ+ .247) These properties often can be readily demonstrated for operators and the transformation properties (6. 0)T .3: Derive Eqs.245) (6.248) a tensor operator? Exercise 6.

J3 . i. 2 |γ2 ) = ( 1 m1 |kq 2 m2 ) Tkq | 2 m2 γ2 . 2 |γ2 ) = q.250) q. possibly the total angular momentum– spin state of a compositie system. 2 |γ2 ) = Cδ 1 1 δm1 m1 2 1( 1 (6. 2 |γ2 ) is an eigenstate of J 2 with eigenvalue + 1).249) which demonstrates. 2 |γ2 ) is an eigenstate of J 2 with eigenvalue 2 1 ( 1 + 1). behave like |kq | 1 m1 .8 Wigner-Eckart Theorem A second important property of tensor operators Tkq beside (6. J2 . 1 m1 |Φ 1 m1 (k. (6. To prove this we consider the transformation of Tkq | 2 m2 γ2 R(ϑ) Tkq | 2 m2 γ2 = R(ϑ) Tkq R−1 (ϑ) R(ϑ) | 2 m2 γ2 = q m2 Dq q Tkq Dm2 m2 | 2 m2 γ2 2 (k) ( ) (6.m2 2 m2 ) (q + m2 ) Tkq | 2 m2 γ2 | 2 m2 γ2 (6.252) ∼δm1 q+m2 = ( 1 m1 |kq 2 m2 ) Tkq Similarly. hence. 2 |γ2 ) can be shown readily as follows using (6.1: Show that Φ 1 m1 (k.247). (6.m2 ( 1 m1 |kq m1 q.180 Theory of Angular Momentum and Spin 6.e. J3 | 2 m2 γ2 = m2 γ2 | 2 m2 and the property (6.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 . . m1 . one can show that Φ 1 m1 (k. J 2 | 2 m2 γ2 = 2 2( 2 + 1) | 2 m2 γ2 . Before we proceed we like to point out that the states | 2 m2 γ2 are also eigenstates of J 2 . in fact.237) is that their matrix elements 1 m1 γ1 |Tkq | 2 m2 γ2 obey simple relationships expressed in terms of Clebsch-Gordan coefficients. The hermitian property of J3 and J 2 implies that states Φ 1 m1 (k.253) Exercise 6.235 . 2 |γ2 ) which correspond to total angular momentum states.e. i. construct states Φ 1 m1 (k.18) are defined through Φ 1 m1 (k. | 1 m1 γ1 denotes an angular momentum (spin) state. 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. These states according to (6.21) of Clebsch-Gordan coefficients J3 Φ 1 m1 (k.6. i. the stated property. which is characterized also by a set of other quantum numbers γ1 which are not affected by the rotational transformation R(ϑ). One can. J3 | 2 m2 γ2 = m2 | 2 m2 γ2 .m2 ( 1 m1 |kq 2 m2 ) (J3 Tkq − Tkq J3 + Tkq J3 ) | 2 m2 γ2 = qTkq = q. 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. 2 |γ2 ) are orthogonal to | 1 m1 in case of different quantum numbers 1 .251) The corresponding property for Φ 1 m1 (k.8.e. J3 are the generators of the rotation R(ϑ).

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

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

4) is by far the most relevant of the four choices.3) describes an isotropic harmonic oscillator .. i. The fourth potential V (r) = − Z e2 r (7. Poststrasse 22.Chapter 7 Motion in Spherically Symmetric Potentials We describe in this section the stationary bound states of quantum mechanical particles in spherically symmetric potentials V (r). The potentials (7. for example. 7000 Chur. Potential (7.1) confines a freely moving particle to a spherical box of radius R. The corresponding wave functions serve basis functions for multi-electron systems in atoms. The potential (7. in case of the so-called bag model of hadronic matter. 7. molecules.e. Four examples will be studied.3) describes the motion of a charge in a uniformely charged sphere and can be employed to describe the motion of protons and neutrons in atomic nuclei1 . The second potential is of the square well type −Vo 0 ≤ r ≤ R . See.4) governs the motion of electrons in hydrogen-type atoms. It leads to the stationary electronic states of the hydrogen atom (Z = 1).2) serve as schematic descriptions of quantum particles. Switzerland. and crystals. φ. Simple Models of Complex Nuclei / The Shell Model and the Interacting Boson Model by I. Talmi (Harwood Academic Publishers. The first potential V (r) = 0 0 ≤ r ≤ R ∞ r > R (7. for example.1. (7. in potentials which are solely a function of r and are independent of the angles θ. 1993) 1 183 .2) V (r) = 0 r > R The third potential V (r) = 1 mω 2 r2 2 (7.

m (r) . i. ˆ dt (7. stationary states ψE. 2 . 3 .8) This property can also be proven readily employing the expression of the kinetic energy operator [c. . the time variation of J can be written. one can state ˙ J = m r2 θ . (5. 2 2mr2 (7. k = 1.14) (7. Accordingly. J3 ψE.87) for Jk as well as the commutation property (5. k = 1.m (r) can be chosen as simultaneous eigenstates of the Hamiltonian ˆ operator H as well as of J 2 and J3 . (7. J ψE.7) ˆ The correspondence principle dictates then for the quantum mechanical Hamiltonian operator H and angular momentum operators Jk ˆ [H. 2 .1 Radial Schr¨dinger Equation o ˙ mv = − er ∂r V (r) . ˆ A classical particle moving in a potential V (r) is governed by the Newtonian equation of motion (7.f.m (r) . 2.11) (7. one can conclude ˙ Jk = {H. d ˙ J = v × mv + r × mv = 0 + r × er (−∂r V (r)) = 0 .e. using (7. .99)] 2 1 2 J2 2 2 = − ∂r r + .184 Spherically Symmetric Potentials 7.15) (7.85– 5. 3..10) (7.12) . ˆ H ψE.6) ˙ Since Jk is also equal to the poisson bracket {H.5) In the case of an angular independent potential angular momentum J = m r × v is a constant of ˙ motion.13) . Employing in this plane the coordinates r.61) of J 2 and Jk . (7.9) − 2m 2m r 2mr2 the expressions (5. θ for the distance from the origin and for the angular position. In fact. 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 . Jk ] = 0 .m (r) m ψE. 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.m (r) . For this purpose one concludes first that the conservation of angular momentum J implies a motion of the particle confined to a plane. Jk } where H is the Hamiltonian. Jk } = 0 .m (r) (7.m (r) = = = E ψE. (7.5) and r = v. 2. ( + 1) ψE.

− Adopting for ψE. 2 (7.m vE.m (r) Y m (θ. (7. φ) are the angular momentum eigenstates defined in Section 5.16) m 2mr2 Once.m (r) = 0. This barrier. (7.m (r) 2 .4.10). (7. . .23) This demonstrates that the function r vE.e. but can also trap particles in the latter space giving rise to strong scattering resonances (see Section ??).18) = vE. φ) (7.m (r) = 0. for the wave function of a quantum mechanical particle a differential equation which governs solely the r-dependence.20) Since this equation is independent of the quantum number m we drop the index m on the radial wave function vE. .m (r) and E .m (r) describes the radial motion as a one-dimensional motion in the interval [0. Employing the kinetic energy operator in the form (7. .24) .12) are obeyed and one obtains for (7.17) In analogy to the classical description one can derivem. ( + 1) 2mr2 . (r) = 0 .20) in the form of the one-dimensional Schr¨dinger equation o 2 − where 2m 2 ∂r + Veff (r) − E φE (r) = 0 . r(t) is determined the angular motion follows from (7.20) by −2mr/ 2 yields the so-called radial Schr¨dinger equation o 2 ∂r − ( + 1) − U (r) − κ2 r2 r vκ.21) Veff (r) φE (r) = = V (r) + r vE.7. ∞[ governed by the effective potential (7. It can be solved by integration of 1 2 2 J2 r = ± ˙ E − V (r) − . equations (7. 7. can exclude particles from the space with small r values.10) − 1 2 ∂ r + 2m r r 2 2 ( + 1) + V (r) − E 2mr2 .19) where Y m (θ. together with the original potential.m (r) 1 2 J2 ∂r r + + V (r) − E 2m r 2mr2 the functional form ψE. Multiplying (7.11.11). i.13). using (7. in the present case. (7. mr2 (t) (7. One can write (7.m ψE.m (r) .m . . by integration of ˙ θ = J . (7.9) one can write the stationary Schr¨dinger o equation (7.. .1: Radial Schr¨dinger Equation o 185 This is a differential equation which governs solely the radial coordinate. .22) which is the original potential 2 ( +1) V (r) with an added rotational barrier potential 2mr2 .22) (7.

(r) ∼ A r + B r− −1 +1 + B r− (7.186 where we defined U (r) κ2 = = − − 2m 2 Spherically Symmetric Potentials V (r) E (7. II by A. 1975).” by J. This follows for > 0 from consideration of the integral which measures the total particle density.26) 2m 2 In case E < 0.m (r) ∼ √ B .20) the index E by the equivalent index κ. r → 0 (7. For r → 0 one may assume that the term ( + 1)/r2 becomes larger than the potential U (r). .33) A detailed discussion of the proper boundary conditions.m (r) ∼ √ 4π B δ(x1 )δ(x2 )δ(x3 ) . for > 0 is not integrable. .D.29) Only the first term is admissable. ψE. 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). accordingly. 2 2 2m 2 ψE. in particular. (7. see. . at r = 0 is found in the excellent monographs Quantum Mechanics I. (r) ∼ A r or vκ. (r) = 0 . New York. In this case the solution is governed my ( + 1) 2 r vκ. 4π|r| (7. for example. Berlin.25) (7. For = 0 the contribution of Br−( function is. the term Br−( +1) would contribute |B|2 0 dr r−2 +1) (7. assumes the general form r vκ. Galindo and P. The radial part of this integral is ∞ 0 2 dr r2 vκ.28) (7. Jackson (John Wiley. according to a well-known result in Classical Electromagnetism3 . hence. We replaced in (7. Boundary Conditions In order to solve (7.182) for Y00 .27) ∂r − r2 and. 2nd Ed. κ assumes real values. (r) (7. ”Classical Electrodynamics. Pascual (Springer.20) one needs to specify proper boundary conditions2 .30) and. using the expression (5.31) to the complete wave which.32) The total kinetic energy resulting from this contribution is.

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).

m a mj (kr) Y m (θ.194 Generating Function of Spherical Bessel Functions Spherically Symmetric Potentials The stationary Schr¨dinger equation of free particles (7. can be written exp(ikr cos θ). φ) .106).178) are given in terms of Legendre polynomials P (cos θ) one can replace the expansion in (7. of (7. One can show d −1 2 (x − 1) ∼ (x2 − 1) × polynomial in x dx −1 (7. according to (5. in case of a free particle moving along the x3 -axis one expands eik3 x3 = . The orthogonality properties (5. −1 .179) yield from (7. One can expand the former solution in terms of solutions (7.99) +1 d cos θ eikr cos θ P (cos θ) = b j (kr) −1 2 .102) −1 d −1 +1 −1 dx −1 (x2 − 1) .98) The l.98) by ∞ eik3 x3 = =0 b j (kr) P (cos θ) . (7. φ).s.. In this case the expansion on the r. 2 +1 (7.100) Defining x = cos θ. one given o by (7. namely. z = kr.s. (7.e.99) We want to determine the expansion coefficients b . φ) for non-vanishing m has a non-trivial φ-dependence as described by (5. the wave function does not depend on φ. and using the Rodrigues formula for Legendre polynomials (5. For example.50).44. 7.h.h. Since the spherical harmonics Y 0 (θ. (7.103) and. hence.45) and one given by (7.101) 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. This holds for −1 and one can conclude +1 consecutive integrations by part d izx e dx (7.50).43) has two solutions.104) dx eizx P (x) −1 = (−1) 2 ! (iz) = 2 ! +1 dx (x2 − 1) −1 +1 dx (1 − x2 ) eizx . i. the surface term ∼ [· · ·]+1 vanishes.150) one obtains +1 +1 1 ∂ dx eizx P (x) = dx eizx (x2 − 1) .98) does not involve any non-vanishing m-values since Y m (θ.

1 (7.99). after insertion into (7.106) yields finally b = i (2 + 1) or. Integral Representation of Bessel Functions Combining (7.f. 2 +1 −1 dx (1 − x2 )ν− 2 eizx .7.108) where the last factor is equal to unity.106) Employing (5.107) 1 2 (2 )! 1 · 3 · 5 · · · (2 + 1) 2 + 1 1 · 2 · 3 · 4 · · · (2 − 1) · 2 (7.109) eikr cos θ = =0 i (2 + 1) j (kr) P (cos θ) . of (7.105) and (7.114) . −1 (7.105) must hold for all powers of z.109) results in the integral representation of j (z) j (z) = (z) 2 +1 ! +1 dx (1 − x2 ) eizx . as the generating function of the spherical Bessel functions. −1 (7. The identity (7. ∞ (7. Comparision with the l.h.s. (7.117) one can write the r.96) one can express this. (7.h. using ν = Jν (z) = √ 1 πΓ(ν + 1 ) 2 z 2 ν + 1. 1 (7.113) dt (1 − t2 )ν− 2 eizt . for the leading power x [c. −1 (7.s.s.105) This expression allows one to determine the expansion coefficients b . z i or i (2 + 1) z (2 )! 2 2 2 +1 2 ! [1 · 3 · 5 · · · (2 − 1)] 1 (7.111) Employing (7.112) We want to consider the expression Gν (z) = aν z ν fν (z) where we define fν (z) = C (7. in particular.2: Free Particle Comparision with (7.h.110) One refers to the l.69)] b z 2 (iz) = 1 · 3 · 5 · · · (2 + 1) 2 + 1 2 ! +1 dx (1 − x2 ) .100) gives b j (kr) 2 (iz) = 2 +1 2 ! +1 195 dx (1 − x2 ) eizx .

119) We note that equations (7. To prove this we note fν (z) = i C dt (1 − t2 )ν− 2 t eizt . In fact.113) obeys the Bessel equation (7.117) dt (1 − t2 )ν− 2 t2 eizt 1 (7.123) Insertion of these identities into Bessel’s equation leads to a differential equation for fν (z) which is identical to (7. t2 of the integration paths C. 7.196 Spherically Symmetric Potentials Here C is an integration path in the complex plane with endpoints t1 . We can now demonstrate that Gν (z) defined in (7.93).93) as long as the integration path in (7.116) fν (z) = − or fν (z) = − From this we can conclude fν (z) + 2ν + 1 fν (z) + fν (z) = 0 . We consider now the functions u(j) (z) = √ 1 πΓ(ν + 1 ) 2 z 2 ν Cj dx (1 − x2 )ν− 2 eizx . (7. z (7.122) (7. 1 (7. for properly chosen endpoints t1 . 2.121).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 . 4.4: Prove (7.116) In case that the endpoints of the integration path C satisfy (1 − t2 )ν+ 2 eizt one can write (7.113) satisfies (7. C 1 (7.124) . 1 j = 1.118) z 2ν + 1 fν (z) + fν (z) . it holds for the derivatives of Gν (z) Gν (z) G (z) = = ν aν z ν fν (z) + aν z ν fν (z) z 2ν ν(ν − 1) aν z ν fν (z) + aν z ν fν (z) + aν z ν fν (z) 2 z z (7.120) z 2ν + 1 dt (1 − t2 )ν− 2 eizt + C C 1 1 t2 t1 = 0 (7. obeys Bessel’s equation (7.114. 3. t2 . Exercise 7.117).93) for arbitrary ν.120) such that we can conclude that Gν (z) for proper integration paths is a solution of (7.2. (7.121) (7. Gν (z).120) imply also the property (2ν + 1) fν (z) + z fν+1 (z) = 0 .

(2) Hν (z) = −2 u(4) (z) .117) and.125) 0 ≤ s < ∞ (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 . Following convention. one can state Jν (z) = − u(2) (z) + u(4) (z) .132) For Hν (z) one derives.130) holds Jν (z) = (1) 1 2 1 (1) (2) Hν (z) + Hν (z) .128) C = C1 ∪ C2 ∪ C3 ∪ C4 is a closed path.134) Similarly. one can derive (2) Hν (z) = e −i(z−πν/2−π/4) ∞ 0 √ ds [s(1 − is/2)]ν− 2 e−zs . 1 (7. (7. therefore. hence. dt = i ds.124) vanishes along the whole path C3 and. . Since the integrand in (7. we introduce the so-called Hankel functions (1) Hν (z) = −2 u(2) (z) . the integral expression (1) Hν (z) 1 1 1 1 (7.2: Free Particle 197 for the four integration paths in the complex plane parametrized as follows through a real path length s C1 (t1 = −1 → t2 = +1) : C2 (t1 = 1 → t2 = 1 + i ∞) : C3 (t1 = 1 + i ∞ → t2 = −1 + i ∞) : C4 (t1 = −1 + i ∞ → t1 = −1) : t = s t = 1 + is t = 1 + is t = −1 + is −1 ≤ s ≤ 1 (7.135) (7. u(2) (z) and u(2) (z) are both solutions of Bessel’s equation (7.7. 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 . 1 (7. (7.127) 0 ≤ s < ∞ (7. Accordingly.93). u(3) (z) ≡ 0.129) The integrand in (7. using t = 1 + is.131) According to (7.124) is analytical in the part of the complex plane surrounded by the path C we conclude 4 u(j) (z) ≡ 0 j=1 (7.126) −1 ≤ s ≤ 1 (7. Comparision with (7.134) and (7.135) are known as the Poisson integrals of the Bessel functions. for Rez > 0 and ν ∈ R obey (7.112) shows u(1) (z) = Jν (z). (7.130) We note that the endpoints of the integration paths C2 and C4 .

143) . z (7. We employ for this purpose the Poisson integrals (7. the Wronskian W (g.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.142) and h(z) = H (2) (z) +1 2 one can identify the constant c by using the leading terms in the expansions (7. converge fast for |z| → ∞. Hν (z) as well as H−ν (z) are solutions of this equation.141) is o W (z) = − In case of g(z) = H (1) (z) +1 2 c .93) only as ν 2 .139) (z) = 2 ±i[z − ( e πz +1) π ] 2 r=0 ( + r)! r!( − r)! i ± 2z r (7. hence.140) Bessel Functions with Negative Index Since ν enters the Bessel equation (7.140). (7.e. For such solutions g(z). h(z) to be linearly independent. πz (7. i.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. we obtain H (1) 2 1 +2 (1) 2 (z) = r=0 ( + r)! ! r!( − r)! ± i 2 r 1 z +r+1 (7.134) and (7. One obtains W (z) = − 4i . (7. For the Wronskian connected with the Bessel equation (7. h) must be a non-vanishing function..93) holds the identity 1 W = − W z (7. As a second order differential equation the Bessel equation has two linearly independent solutions.135) which read for ν = + 1 2 H where f The binomial formula yields f (1) 2 (1) 2 1 +2 (1. The general solution of (7.136) (1) 2 (z) = 0 ds [s(s ± is/2)] e−zs .137) (z) = r=0 r ± i 2 r 0 ∞ ds s +r e−zs .141) (1) (1) the derivation of which follows the derivation on page 192 for the Wronskian of the radial Schr¨dinger equation.138) The formula for the Laplace transform of sn leads to f and.

.151) are equivalent to h h (1) (2) (z) (z) = = j (z) + i n (z) j (z) − i n (z) .2: Free Particle i. 7. H (1. B can be obtained from the asymptotic expansion (7.146) (z) = −i (−1) H (2) 1 (z) +2 .53).2) (7.144) The expansion coefficients A. (7.152) (7. 7.146. (7. one can show H−( Spherical Hankel Functions In analogy to equations (7.97).132) follows n (z) = 1 2 π (−1) 2z +1 H−( (1) 1 +2) (z) + H−( (2) +1) 2 (z) . (7.97) one defines the spherical Hankel functions h (1. (7.90. We conclude 2 H−( Simarly.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. (7. in fact.147) (z) = π (1.151) Equations (7. 2z 2 (1.140).149. From (7.150) According to (7.2) H + 1 (z) .153) .2) (7.2) (2) +1) 2 (1) +1) 2 (z) = i (−1) H (1) (z) +1 2 . H−ν (z) = A H (1) (1) (z) +1 2 One can then expand + BH (2) (z) +1 2 .132. (7.149) We want to establish also the relationship between h (z) and the irregular spherical Bessel function n (z) defined in (7. linearly independent. According to (7. 7.97.147) this can be written n (z) = 1 2i h (1) (z) − h (2) (z) .2) (z) +1 2 199 are. 2 (1.148) holds for the o regular spherical Bessel function j (z) = 1 (1) (2) [ h (z) + h (z) ] .e. (7.96.145) This equation can hold only for B = 0 and A = exp[i( + 1 )π]. 7. 7. 7. For |z| → ∞ the leading terms yield π π π π π 1 1 1 √ ei(z+ 2 ) = √ A ei(z− 2 − 2 ) + √ B e−i(z− 2 − 2 ) z z z |z| → ∞ .7.

are j (z) n (z) = = (7.164) .159) (7. j (z) (z) = ( i) z +1 e±iz r=0 ( + r)! r!( − r)! ± i 2z r (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.155) (z) and To determine j (z) and n (z) we note that for z ∈ R the spherical Hankel functions h (2) h (z). 7.2) (z). 2z 2 z sin(z − π/2) z cos(z − π/2) − .161) (7. as given by (7. at large |z|. = sin(z − π/2) and sin[z − ( + 1) π ] .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 (7.140.162) The leading terms in these expansions.163) (7. at large |z|. 2 2z z (7. From (7.148). Hence. it follows z ∈ R : Using p (z) = Re r=0 j (z) = Re[h (1) (z)] .160) Employing cos[z − ( + 1) π ] .154) The leading term in this expansion. (7.148) one obtains readily h (1) 2 (1.156) ( + r)! r!( − r)! ( + r)! r!( − r)! i 2z r [ /2] = r=0 ( + 2r)! (2r)!( − 2r)! −1 4z 2 r (7.158) one can derive then from (7. z (7. (1) (7. ne ll(z) = Im[h (1) (z)] . = −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) . is h (1) 2 (z) = ( i) z +1 e±iz .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) .140) and (7.

(1.166) (7.169) (7. defining a = −1 and employing fν (z) as defined in (7.148) and express g (z) = h (1.2: Free Particle Expressions for the Spherical bessel Functions j (z) and n (z) 201 The identities (7.167) (7.171. 7. j (z) and n (z). From the linearity of the relationships (7.174) and from (7. 7.2) g (z) − g (z) z 2 +1 g (z) − g z (7.2) To demonstrate that (7.4 dx (1 − x2 )ν− 2 eizx .170) Recursion Formulas of Spherical Bessel Functions The spherical Bessel functions obey the recursion relationships g g +1 (z) +1 (z) = = (1.173) A (z) =  z ( +1) z2 +2 z (7. 7.174) We want to prove these relationships.2) relationships hold for g (z) = h (z).149.114) we can write g (z) = a z f 2 ! + 1 (z) 2 . (7.165) (7. 1 (7.2) (z) = −2 π 1 √ 2z πΓ(ν + 1 ) 2 z 2 ν C2.168) (7. 7.162) allow one to provide explicit expressions for j (z) and n (z).172) to +1 (z) +1 (z) = A (z)  1− g (z) g (z) −1   (7. 7.171–7.175) Using Γ( + 1) = !.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. 1.124.171) holds for g (z) = h (z) we employ (7.171) −1 (z) (7.131. One obtains for = 0.176) .161.172) where g (z) is either of the functions h obtain the recursion relationship g g (z). For this purpose we need to demonstrate only that the (1.151) follows then that the relatiosnhips hold also for g (z) = j (z) and g (z) = n (z). One can combine (7.

z2 (7.58) it holds o 2 g (z) = − g (z) + z ( + 1) g (z) − g (z) .165).174) we start from (7.5: Provide a detailed derivation of (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) .202 The derivative of this expression is g (z) = Employing (7.2. Exercise 7.2.171). To prove (7. Exercise 7.121).179) from which follows (7..166). n2 (z) from n0 (z).171).174) to determine (a) j1 (z). (7.173) is equivalent to (7.182). 7. 7. z (7.e.172). n0 (z) using (7.173.179) leads to (7.173. is equivalent to (7. f yields. In order to prove (7. The second component of (7.172). j0 (z) using (7.180) yields g +1 (z) = g (z) − ( + 2) g (z) + z2 +2 g (z) . (7. (7.171).180) Since g (z) is a solution of the radial Schr¨dinger equation (7. The first component of (7.177) (z) = − z f 2 +2 + 3 (z) 2 .178) (7.179) g (z) = − z2 g (z) + g (z) − g +1 (z) z . j2 (z) from j0 (z). in fact.182) Replacing all first derivatives employing (7.6: Employ the recursion relationship (7.172) we differentiate (7. and (b) n1 (z).181) Using this identity to replace the second derivative in (7. .173). i.

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

(8. The following substitutions. . t). t)] and.9) (8. From this follows E(r. The reader is referred to the well-known textbook ”Classical Electrodynamics”. t) through the scalar and vector potentials V (r.3) (8.10) (8. t) + ∂t A(r. In the non-relativistic limit holds p ≈ mr and (8. t) − ∂t χ(r. t) + χ(r. t). D. called the scalar potential. t] + v × B[ro (t).8) = 0 (8. t) in (8. t) in (8.2) In addition. t] (8. t) + v × B(r.1) (8. t) is a scalar function. in turn. t) + ∂t B(r. t) × B(r. t) . the two homogeneous Maxwell equations hold × E(r. obeys the equation of motion d p = q dt E[ro (t). t) −→ 1 A(r. t) . The particle. by J.1) the term qδ(r − ro (t)) and to the current density ˙ J(r. called the vector potential.2) the term q r˙o δ(r − ro (t)). t) −→ V (r. t) = 0 = 0.4). Equation (8.7) Gauge Transformations We have expressed now the electric and magnetic fields E(r. t) and B(r. As is well known. accordingly. t) Interaction of Radiation with Matter = 4 π ρ(r. 1975) for a discussion of these and other conventional units. contributes to the charge density ρ(r.11) V (r. t) which is solved by E(r. called gauge transformations. t) (8. t) and A(r. t) + ∂t A(r. t) = × A(r. 2nd Edition.5) where p is the momentum of the particle and ro (t) it’s position at time t.3) reads then × E(r.6) for some vector-valued function A(r. (8. t) − ∂t E(r. t) = − V (r. the relationship between fields and potentials is not unique.25). 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. solves implicitly (8. alter the potentials. Jackson (John Wiley & Sons.5) above agrees with the equation of motion as given in (1. New York. Scalar and Vector Potential Setting B(r. t) where V (r. t) − ∂t A(r. t) = − V (r. t) = 4 π J(r. t) · B(r. t) . We assume so-called Gaussian units. but leave the fields unaltered: A(r. (8.204 Maxwell equations1 · E(r.

t) = 1 ∂t V (r. t) = − 4 π J(r. 4π (8. One would prefer a description in which the Coulomb potential (8. t) = · A(r. be achieved. Such description can.20) . t) (8.12) yields then the Poisson equation 2 V (r. t) + ∂t V (r. t) . t) − ∂t A(r.1) · − V (r. t = 4 π J(r. t) = 0 the solution is given by the Coulomb integral V (r.19) and define J (r. For this purpose we examine the offending term ∂t V (r. t) Using · ∂t A(r. t) + ∂t A(r. t) = ∂t = 4 π ρ(r. and the former the Coulomb interactions in the unperturbed bound particle system. t) = 0 . t) the condition · A(r. the second inhomogeneous Maxwell equation.19) couples the vector potential A(r.16) This is the potential commonly employed in quantum mechanical calculations for the description of Coulomb interactions between charged particles.1: Electromagnetic Field 205 This gauge freedom will be exploited now to introduce potentials which are most suitable for the purpose of separating the electromagnetic field into a component arising from the Coulomb potential connected with the charge distribution ρ(r. t) can be obtained employing (8.6) and (8.13) · A(r. (8. t). (8. (8. (8.12) leads us to 2 2 A(r. t) . (8. Using the expressions (8. t) .15) d3 r ρ(r . this potential results from inserting (8. t). The vector potential A(r.17) A(r. t) |r − r | (8.19) Unfortunately.16) and the vector potential are uncoupled.8. t) = Ω∞ for r ∈ ∂Ω∞ (8.14) In case of the boundary condition V (r. t) − 2 × A(r. and a component due to the remaining currents. In fact. In fact. such that the latter describes the electromagnetic radiation. the gauge freedom allows us to impose on the vector potential A(r.9) for the fields results in × The identity × × A(r. t) .12) The corresponding gauge is referred to as the Coulomb gauge. t) and V (r. equation (8. t) = − 4 π ρ(r.18) together with condition (8. in fact. t) . t) and the current due to moving net charges. t) together with (8. a name which is due to the form of the resulting scalar potential V (r.9) into (8. t) in (8. t) − ∂t V (r.2). t) − ∂t A(r.

. t) = − V (r. t) contributes an electrical field component as well as the total magnetic field Et (r. t) = 0 . t) These fields are obviously divergence -free (e. for example. hence. t) = − ∂t ρ(r. t) . t) be the total current of the system under investigation and let Jt = J − J .21) and (8. t) − ∂t A(r. Let J(r.28) (8.g. t) = 0 (8. (8. t) = 0).206 For the curl of J holds × J (r. t) = × A(r. hence. The vector potential determined through (8. V (r. t) as the current due to the time-dependence of the charge distribution ρ(r. 8. the name longitudinal electric field. (8.24) (8. For J also holds the continuity equation · J(r. respectively.2 Planar Electromagnetic Waves The current density Jt describes ring-type currents in the space under consideration. t) = 0 . t) = or = 2 Interaction of Radiation with Matter (8. t) = − 4 π Jt (r.26) turns into the well-known wave equation 2 2 A(r.23) This continuity equation identifies J (r. t) holds.21) ∂t and the Poisson equation (8. t) Bt (r. yet a current.22) 1 ∂t 4π · J (r. no divergence-free currents.12) and the Coulomb potential (8. the name transverse fields. t) (8. in a ring-shaped antenna which exhibits no net charge.26) This equation does not couple anymore scalar and vector potentials. such current densities exist. Presently. t).14). = − ∂t A(r. t) . we want to assume that no ring-type currents. For the divergence of J (r. The definitions of J and Jt applied to (8.29) · Et (r. (8. (8. V (r.30) . t) (8.26) and (8. t) = 0 and from this follows · Jt (r.16) yield finally the electric and magnetic fields. t) = 0). t) − ∂t A(r.e.19) yield 2 2 A(r. A(r. t) + ∂t ρ(r. using ∂t · J (r.25) one refers to J and Jt as the longitudinal and the transverse currents.. t) = 0 . exist in the space considered. In this case (8.27) which is obviously curl-free ( × E (r. t) + ∂t ρ(r.25) Because of properties (8. i. t) contributes solely an electric field component E (r.

A complete set of solutions is given by the so-called plane waves A(r. The corresponding electric and magnetic fields. The latter vector describes the direction of propagation of the energy flux connected with the plane wave electromagnetic radiation.8.32. 8. t) . is orthogonal to k. Note that in the units chosen the velocity of light is c = 1. (8. Obviously. accordingly.34.34.2: Planar Electromagnetic Waves 207 which describes electromagnetic fields in vacuum.36) 4π Using the identity a × (b × c) = b (a · c) − c (a · b) and (8. the constant k is referred to as the wave vector. 8. 8.38. The Coulomb gauge condition (8. t) in (8. at each point r and moment t. t) = i k × A(r. Time average over one period 2π/ω yields S(r.31. t) = ±i ω A(r. t) = Re Et (r.35) are complex-valued quantities.35) one obtains S(r.31) u is a unit vector (|ˆ| = 1) which. ω2 = |Ao |2 (8. In applying the potential and fields to physical observables and processes we will only employ the real parts. (8. there exist two linearly ˆ u independent orientations for u corresponding to two independent planes of polarization.158). This flux is given by 1 S(r.34.31) and the resulting fields (8. t) and Bt (r.35). t) = ± ω2 ˆ |Ao |2 k . 8.31).38) ˆ In this expression for the energy flux one can interprete k as the propagation velocity (note c = 1) and. (8. (8.29). t) × Re B(r. see also the comment below Eqs.35) The vector potential in (8.37) ˆ ˆ where k is the unit vector k = k/|k|. are orthogonal to each other and are both orthogonal to the wave vector k.32) holds. t) = ± ω2 ˆ |Ao |2 k sin2 ( k · r − ωt ) 4π (8.28. t) Bt (r. Et (r. according to (8.34) (8.39). 8. obviously. 8π (8. 2 . ˆ We want to characterize now the radiation field connected with the plane wave solutions (8. ˆ (8.39) 8π The definition incoming waves and outgoing waves is rationalized below in the discussion following Eq. 8. are Et (r. hence. Here the “-” sign corresponds to so-called incoming waves and the “+” sign to outgoing waves2 . 8.12) yields u·k = 0. t) = Ao u exp i(k · r ˆ where the dispersion relationship |k| = ω (8.33) ωt) (8. t) .

31.32).40) = V It should be pointed out that Nω represents the number of photons for a specific frequency ω. t).42) 8.29) together with the potentials (8. ˆ (8. a ˆ specific k and a specific u.27. hence. 8. (8. Comparision of (8.31). These photons each carry the energy ω. defined below Eqs. respectively.28.31) allows one finally to state for the planar wave vector potential A(r.45) .40) allows one to express then the field ˆ amplitudes 8πNω Ao = .39) through the density of photons connected with the planar waves (8. (8. the energy of the plane wave is transported in the direction of −k. t) = 8πNω u exp i(k · r − ωt) ˆ ωV . 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. (8. 8.8. t) 2m 2 + qV (r) .16. i (8.43). t). A correct description of the electromagnetic field requires that the field be quantized. t) . whereas in the case of outgoing waves the energy is transported in the direction of k. t) are complex leading to the difference of a factor 2 in the energy densities of the lontitudinal and transverse components of the fields.44) .41) ωV Inserting this into (8. (8.208 Interaction of Radiation with Matter as the energy density. u · k = 0 .43) Here the fields are defined through Eqs. The integrals express the integration over the energy density of the fields.38) implies that for incoming waves.3 Hamilton Operator The classical Hamiltonian for a particle of charge q in a scalar and vector potential V (r) and A(r. is p − q A(r. Note that E (r. We are left with a classical Hamiltonian function which has an obvious quantum mechanical analogue ˆ H = ˆ p − q A(r. We assume that the energy content of the fields is not altered significantly in the processes described and. The sign in (8. Bt (r. t) = H Ψ(r. If we consider a volume V with a number of photons Nω the energy density is obviously Nω ω . (8. The wave function Ψ(r. t) is real 1 and that Et (r. A ‘poor man’s’ quantization of the field is possible at this point by expressing the energy density (8. we will neglect the respective terms in the Hamiltonian (8. (8. t) 2m + 1 8π Ω∞ 2 H = + qV (r) 1 16π d3 r |Et |2 + |Bt |2 Ω∞ d3 r E 2 + . |k| = ω . 8.31).39) and (8.

hence.50) The equivalence of (8. (8.t)/ should not affect a system and that. (8.t)/ p + q(( χ)) Ψ(r. not to phenomenological interactions like the molecular van der Waals interaction. t) .47) is equivalent to   2 ˆ p − qA   i ∂t eiqχ(r. t) by a local and time-dependent phase factor eiqχ(r. One can show that (8. t)|2 and.t)/ constitutes a unitary transformation of a scalar quantity.51) 2m the potentials A(r.44.46) the fields enter. t) =  2m where ((· · ·)) denotes derivatives in (( χ)) and ((∂t χ)) which are confined to the function χ(r. t) . t) ˆ = eiqχ(r. t) . It should be noted.t)/ Ψ(r. Obviously.47..10. by the group 3 The effect on other expectation values is not discussed here. t) enter whereas in the classical equations of motion ˙ m¨ = q E(r.48) 2m For this purpose one notes i ∂t eiqχ(r. does not change expectation values which contain the probability densities3 . Elementary constituents of matter which are governed by other symmetry groups.8. 8. This leads to the question in how far the gauge transformations (8. such phase factor does not change the probability density |Ψ(r. in the Schr¨dinger equation o   2 ˆ p − qA   i ∂t Ψ(r.48) implies that the gauge transformation (8. e. (8.t)/ Ψ(r.t)/ Ψ(r. . An important conceptual step of modern physics has been to turn the derivation given around and to state that introduction of a local phase factor eiqχ(r.t)/ .t)/ [ i ∂t − q((∂t χ)) ] Ψ(r.46). t) ˆ p eiqχ(r. t) =  + qV  Ψ(r. t) (8.3: Hamilton Operator 209 Gauge Transformations It is interesting to note that in the quantum mechanical description of a charged particle the potentials V (r.e.45) leads to the Schr¨dinger equation o   2 ˆ p − q A − q(( χ))   + qV − q((∂t χ))  Ψ(r. that this principle applies only to fundamental interactions. 8. t) are necessary to compensate terms which arise through the phase factor. t) r (8. Applying the gauge transformations (8. 8.g..47) i ∂t Ψ(r.10.11) affect the quantum mechanical description. t) =  + qV  eiqχ(r. t) and A(r. i. t) inside the double brackets. t) and V (r.49) (8.10. The idea just stated can be generalized by noting that multiplication by a phase factor eiqχ(r. t) = eiqχ(r. an element of the group U(1). have no effect on the motion of the particle described by (8. accordingly. In the classical case such question is mute since the gauge transformations do not alter the fields and.11) to (8.t)/ Ψ(r. 8.11) of the potentials is equivalent to multiplying the wave function Ψ(r. hence. t) + q r × B(r. 8. however.

therefore. e..54) ˆ2 p q ˆ q2 2 − p·A + A + qV .57) satisfies · A = 0 and. The stationary homogeneous magnetic field B(r. the so-called Landau gauge ˆ associates the vector potential AL (r) = Bo x1 e2 ˆ (8.g. one can employ the Hamiltonian (8. the generators of SU(2). consequently.58).53) ˆ ˆ p·A + A·p + q2 2 A + qV 2m (8.57) with a homogeneous magnetic field Bo . (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. (8. k3 . t) ≡ 0.55). · A = 0 [cf.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. The resulting fields are called Yang-Mills fields. holds (8. The Hamiltonian (8. x2 .12)].56) due to the gauge freedom.210 Interaction of Radiation with Matter SU(2).45) with Hamiltonian (8. In this o case holds V (r.55).56). affects the complexity of the mathematical derivation of the wave functions.t) where σ is the vector of Pauli matrices.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. x1 . We want to describe the stationary states corresponding to the Hamiltonian (8. likewise can demand the existence of fields which compensate local transformations described by eiσ·χ(r. can be described by various vector potentials. j = 1. (8. t) = Bo .52) This expression vanishes in the present case since since ˆ ˆ p · Af = A · pf and. t). thereby. For this purpose we use the wave function in the form Ψ(E. k2 . In case of a homogeneous potential pointing in the x3 -direction. (8. The vector potential (8. H = 2m m 2m (8.55) 8. 2. Accordingly. for Bo = Bo e3 in (8. 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. The choice of a vector potential affects the form of the wave functions describing the eigenstates and.59) . x3 ) = exp(ik2 x2 + ik3 x3 ) φE (x1 ) . 3.58) where ∂j = (∂/∂xj ).

The stationary Schr¨dinger equation (8.67) .64) is the classical Larmor frequency (c = 1). the familiar harmonic oscillator quantum number. We want to employ .66) Obviously. Unfortunately. therefore.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. x1 .62) x1o = and where ω = k2 eBo eBo me (8. (8. Without affecting the energy one can form wave packets in terms of the solutions (8. However.63) this induces a spread of the wave function in the x1 direction. 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. like the corresponding gauge (8.56) through the vector potential A(r) = 1 Bo × r .4: Electron in Homogenous Magnetic Field This results in a stationary Schr¨dinger equation o 2 211 − 2me 2 ∂1 + 2 k2 2 2me + 2 k2 3 2me + 2 eBo k2 e2 Bo 2 x1 + x φE (x1 ) me 2me 1 = E φE (x1 ) . the so-called symmetric gauge which expresses the homogeneous potential (8. 2 2me (8.63) (8. 2 (8. the states are degenerate in the quantum number k2 describing displacement along the x2 coordinate. x2 . is not symmetric in the x1 . according to (8. is Ψ(n.59).8. Solution for Symmetric Gauge The solution obtained above has the advantage that the derivation is comparatively simple. From this observation one can immediately conclude that the wave shifted (by e 3 function of the system. 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. 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 ) . the wave function (8.57). k2 .59). The energies corresponding to these states are E(n.and x2 -coordinates. k2 . k3 ) = ω(n + 2 k2 1 3 ) + .65). (8.65) where we replaced the parameter E by the integer n.65) which localize the electrons.62) is that of a displaced (by x1o ) harmonic oscillator with o 2 k 2 /2m ) energies. k3 .

78) . For the vector potential (8.73) where µclass is the magnetic moment connected with a current density j µclass = 1 2 r × j(r) dr (8.73. (8.76) e 2me class class · Bo .12) for the Coulomb gauge. (8.52) becomes ˆ2 e e2 p + Bo · L + H = 2me 2me 8me Of particular interest is the contribution Vmag = e L · Bo 2me Bo × r 2 . 8. The theory of classical electromagnetism predicts an analogue energy contribution .72) and (8. Accordingly. ˆ (8. namely. y-plane with constant velocity v on a ring of radius r. the Hamiltonian (8. (8.77) Comparision with (8.68) × u yields. namely. Vmag = −µclass · Bo (8.74) is in the present case 1 µclass = − e r v e3 .212 Interaction of Radiation with Matter One can readily verify that this vector potential satisfies the condition (8. using × (uf ) = −u × f +f ˆ ˆ Bo · p × r f = Bo · r × p f .72) to Hamiltonian (8.70) ˆ ˆ p · A f = − Bo · p × r f . for any (8. accordingly. in the present case of constant Bo .74) We consider a simple case to relate (8. (8.75) 2 The latter can be related to the angular momentum µclass = − and.69) × u and × r = 0.72) allows one to interpret µ = − e L 2me (8. The latter can be rewritten.71). Vmag = e 2me class = r me v e3 of the electron ˆ (8. an electron moving in the x. ˆ Identifying r × p with the angular momentum operator L.67) one can write ˆ p·A = Using · (u × v) = − u · function f (r) ×v + v· 2i · Bo × r .74).71) (8. In this case the current density measures −e v oriented tangentially to the ring. the magnetic moment (8.

j = 1. x2 . we separate the variable x1 . 10 that the spin of the electron. (8. Accordingly.71) is ˆ H = 2 ˆ2 p e2 Bo + 2me 8me x2 + x2 + 1 2 eBo L3 . A derivation of his property and the value of g.84) . 2me (8.88) me ω xj . the Hamiltonian (8. x3 ) = E ΨE (x1 . requires a Lorentz-invariant quantum mechanical description as provided in Sect. x2 . For a magnetic field (8. 2me We have used here the expression for the angular momentum operator ˆ L3 = ( /i)(x1 ∂2 − x2 ∂1 ) . (8.87) . x2 .and the ±x2 -direction. x2 ) where ˆ Ho = − 2 2 2 ∂1 + ∂2 (8.8.e.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 . x3 ) = exp(ik3 x3 ) ψ(x1 .80) (8.56) pointing in the x3 -direction the symmetric gauge (8. In this case. x2 ) . 2 . i.83) describes two identical oscillators along the x1 -and x2 -directions which ˆ are coupled through the angular momentum operator L3 . 10.83) (8. for Bo = Bo e3 . ˆ i. e gives rise to an energy contribution (8.e. We will demonstrate in Sect. the so-called gyromagnetic ratio of the electron. x2 ) = E ψ(x1 . 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 (8.86) 1 2 2me 1 2 ˆ as well as of the angular momentum operator L3 .82) 1 ω (x1 ∂2 − x2 ∂1 ) i 2me + 1 me ω 2 x2 + x2 1 2 2 2 k2 3 + (8. x2 ) obey then ˆ Ho ψ(x1 . The functions ψ(x1 . x3 ) . x2 from x3 setting ΨE (x1 ..79) To obtain the stationary states.85) The Hamiltonian (8. To obtain these eigenstates we introduce the customary dimensionless variables of the harmonic oscillator Xj = (8. E = E − (8.72) with an associated magnetic moment − g 2me S where g ≈ 2. the solutions of ˆ H ΨE (x1 .4: Electron in Homogenous Magnetic Field 213 as the quantum mechanical magnetic moment operator for the electron (charge −e).67) yields a more symmetric solution which decays to zero along both the ±x1 . likewise.81) (8. described by the operator S.

94) is an eigenstate of the vibrational Hamiltonian. x2 ) (8. attempt to express eigenstates in terms of vibrational wave functions a† 1 j+m Ψ(j. x2 ) is the wave function for the state with zero vibrational quanta for the x1 .93) ˆ The states (8.as well as for the x2 -oscillator. obeyed. 1 2 since one phonon is annihilated and one created. 2 (8.92) are not eigenstates of L3 . m . in fact. x1 .214 Employing the creation and annihilation operators 1 a† = √ j 2 and the identity ˆ ω L3 = which can readily be proven. . . m . (8. (8. 0. 1 3 .90) a† a2 − a† a1 1 2 . through a combination of states j Ψ (j.. x1 . −j + 1. x1 .92) represents a state with j + m quanta in the x1 -oscillator and j − m quanta in the x2 -oscillator. m . x2 ) . Such eigenstates can be expressed. m = −j. (j) (8. We.e. the total vibrational energy being ω(2j + 1). 1. m . m. therefore.89) .. j = 1. (8. such that 1 i a† a2 − a† a1 1 2 Ψ (j.e. m . x2 ) = m=−j αmm Ψ(j. If this property is. it holds a† a1 + a† a2 + 1 1 1 2 Ψ (j. 2 2 . x2 ) . (8.94) is also an eigenstate of L3 . . . x2 ) = 2m Ψ (j. m . x1 . x2 ) = a† 1 j−m (j + m)! (j − m)! Ψ(0. x1 .92) where Ψ(0. (8. i. however. x2 ) = ω 2j + 2m + 1 Ψ (j. aj = √ 2 Interaction of Radiation with Matter Xj + ∂ ∂Xj . +j .97) . . x2 ) = ( 2j + 1 ) Ψ (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 .94) is an eigenstate of Ho ˆ Ho Ψ (j. x1 . . x2 ) . x1 . m as follows: j = 0. (8. x1 .95) (j) ˆ We want to choose the coefficients αmm such that (8.96) ˆ holds. x2 ) (8. i. x1 . (8. . 0. In order to cover all posible vibrational quantum numbers one needs to choose j. . x1 . m .91) We note that the operator a† a2 − a† a1 leaves the total number of vibrational quanta invariant. x1 .94) Since this state is a linear combination of states which all have vibrational energy (2j + 1) ω. (8. m.

a transformation moving the x3 –axis into the x2 – 2 axis. the eigenstates of (8. 5. thus. the second rotation a factor dmm ( π ). According to Sect.100) J2 = 1 2 2i 1 ˆ i. x2 ) (8.97). x1 . x1 . m ..5. x1 . π . x2 ) . . 5. i. The required rotation must transform the x3 –axis into the x2 –axis. a1 . We assume that the scalar potential V in (8.10.e. 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 .5. except for a factor 2 . a2 1 2 present notation ←→ b† . cf. If we identify 2 a† .(5. a† . The first rotation contributes a factor exp(−im π ).5. which corresponds there to the angular momentum in the x2 –direction.31). 5.11.5: 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. 0) Ψ(j.5.309) yields finally Ψ (j. Eq. m.10. m) in Sect.83) and confirmed the eigenvalues stated in (8.10. t) = 1/4πr confining an electron with energy E < 0 to move in the well known orbitals of the hydrogen atom.99) with eigenvalue m. (8. x2 ). m. is identical to the operator L3 introduced in (8. 5. x1 . b+ . In the simplest case the radiation field consists of a single planar electromagnetic wave described through the potential (8.101) 2 2 where Dmm ( π . 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.84) above.9. the states are eigenstates of the operator [we use for the operator the notation of Sect. The connection with the present problem arises due to the fact that the operator J2 in Sect. m. 5. 5. the 2 2 latter representing the Wigner rotation matrix of Sect. According to the derivation given there. This implies that ˆ we can obtain eigenstates of L3 by rotation of the states Ψ(j. b− + − notation in Sects. The external radiation field is accounted for by the vector potential A(r.55) confines the particle to stationary bound states.9. Using the explicit form of the Wigner rotation matrix as given in (5. x1 . x2 ) defined in (8.11.92) correspond to the eigenstates |Ψ(j. b† ..e. 5. x1 .11 such transformation is provided through π π (j) Ψ (j.8.5 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. t) introduced above.11 (8. m .10. 8.9.98) then the states Ψ(j.55).102) We have identified. Other radiation fields can . an example is the Coulomb potential V (r. (8. is in the notation of the present section 1 ˆ a† a2 − a† a1 .288)] 1 ˆ J3 = 2 a† a1 − a† a2 1 2 (8. m. x2 ) = Dmm ( .

and that the first term in (8. is much 2 (r. using (8. 1. this assumption can be waved as our results below will not depend on it. In fact. t). (8. However.e. .105) as follows4 . i. The states |n are thought to form a complete..106) where we adopted the Dirac notation for the states of the quantum system. the term ∼ p · A(r. 2 . i. Here the so-called unperturbed system is governed by the Hamiltonian Ho with stationary states defined through the eigenvalue problem Ho |n = n |n .e.41) e ˆ e p2 p·A ∼ 2me me me 2me 4 1 2 8πNω .104) (8.109) The reader should note that the estimates are very crude since we are establishing an order of magnitude estimate only. (8..42).. orthonormal basis. t) + A (r. can be considered a weak perturbation.105). the term ∼ A second term in (8. in this case. For an electron charge q = −e and an electron mass m = me one can provide the estimate for the first term of (8.e. This result will allow us to neglect the larger than the second term. one can estimate that the perturbation. 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.107) 1 = 1 n=0 |n n| . we discount the possibility of a continuum of eigenstates. t) is given by (8. 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. t).108) We assume for the sake of simplicity that the eigenstates of Ho can be labeled through integers.103) (8. ωV (8. We first note. n = 0.105) where A(r.216 Interaction of Radiation with Matter be expanded through Fourier analysis in terms of such plane waves. i. i. as given in (8.105) for the case of radiationinduced transitions in atomic systems. we assume n|m = δnm and for the identity ∞ (8. is much smaller than the eigenvalue differences near ˆ typical atomic bound states.105) in further calculations and to expand the wave function in terms of powers of VS (t) in a perturbation calculation. t) − m 2m (8.. Estimate of the Magnitude of VS We want to demonstrate now that the interaction VS (t). . .e.

5 ˚.42) it holds VS ≈ e m k.117) In case that such perturbation acts on an electron and is due to superpositions of planar waves described through the vector potential (8.105) for radiation fields can be considered a small perturbation. Using again (8.e. i.112) yields e2 A2 ∼ 10 eV · 10−8 = 10−7 eV ..113) This magnitude is much less than the differences of the typical eigenvalues of the lowest states of the hydrogen atom which are of the order of 1 eV. u) p · u exp i(k · r − ωt) ˆ ˆ ˆ kV . t) ..118) . altogether. the first term in (8. i.105).112) π λ me c2 and with e2 /ao ≈ 27 eV. We can. with ao ≈ 0. Assuming a single photon. one can neglect this term as long as the first term gives non-vanishing contributions. e ˆ p · A ∼ 10 eV · 10−4 = 10−3 eV .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 .116) This term is obviously much smaller than the first term. Consequently. We want to estimate now the second term in (8. me (8. a volume V = λ3 where λ is the wave length corresponding to a plane wave with frequency ω. Nω = 1.105) due to a radiation field by VS = − q ˆ p · A(r. Hence. 2me (8. and me c2 ≈ A A 500 keV 2 ao ω ≈ 10−8 (8.ˆ u 4πNk α(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. λ = 2πc/λ.115) (8. and as long as the photon densities Nω /V are small.111) For ω = 3 eV and a corresponding λ = 4000 ˚ one obtains. m (8.109) 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. one obtains for (8. replace the perturbation (8.114) Employing for the second factor the estimate as stated in (8.8. (8. (8. hence.e.110) where ao is the Bohr radius.

127) i ∂t |ΨD (t) = VD (t) |ΨD (t) .123) i = Ho exp − Ho (t − to ) (8. n = 1. describes the effect of a radiation field on an electron system and which will be assumed below to describe radiative transitions. The sum runs over all possible k vectors and might actually be an integral. Multiplying the latter equation by the operator exp i (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) . The probability to find the particle in the state |n at time t is then p0→n (t) = | n|ΨS (t) |2 . One seeks to predict the probability to observe the particle in one of the states |n . .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. For this purpose one needs to determine the state |ΨS (t) of the particle. under ordinary circumstances..122) where |ΨD (to ) = |0 .124) (8. Using i i ∂t exp − Ho (t − to ) and (8. (8. Equation (8.106–8. |n are defined in (8. u) has been added to describe eliptically or circularly polarized ˆ waves.42) through k = |k| = ω. the sum over u involves the two possible polarizations of planar ˆ electromagnetic waves.118) is the form of the perturbation which. |Ψ(to ) = |0 (8.218 Interaction of Radiation with Matter where we have replaced ω in (8. .125) (8. A factor α(k. The states |0 . 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 . 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 .126) Ho (t − to ) yields finally .121) (8. .120) (8. 2.108) as the eigenstates of the unperturbed Hamiltonian Ho . n = 0 at some later time t ≥ to .

0 VD (t) |ΨD (t) (1) VD (t) |ΨD (t) (0) (8. . We will consider here only the two leading contributions to |ΨD (t) . we define |ΨD (t) through the evolution equations i ∂t |ΨD (t) i ∂t |ΨD (t) i (2) ∂t |ΨD (t) (1) (0) = = = .137) One can readily verify that (8.137) |ΨD (t) = (1) (0) (8.121) expressed in terms of ΨD (t) is i p0→n (t) = | n|exp − Ho (t − to ) |ΨD (t) |2 .136) together with the initial conditions |ψD (to ) = |0 0 for n = 0 for n = 1. 8.5: Time-Dependent Perturbation Theory where VD (t) = exp 219 i i Ho (t − to ) VS (t) exp − Ho (t − to ) . (8.137) are consistent with (8.127. = . to (8.134) (8.131) p0→n (t) = | n |ΨD (t) |2 .128) We note that the transition probability (8. . .135) i ∂t |ΨD (t) (n) VD (t) |ΨD (n−1) (t) (8. Equations (8. Employing this result one obtains for (8. In order to determine |ΨD (t) described through (8. . using |exp[− i n (t n (8.128).130) = exp − i n (t − to ) − to )]| = 1. .8. . (8. From (8. 8. Due to the Hermitean property of the Hamiltonian Ho holds n|Ho = i n|exp − Ho (t − to ) from which we conclude. (8. n| (8.132) where |ΨD (t) accounts for the contribution due to n-fold products of VD (t) to |ΨD (t) .133.127) we assume the expansion ∞ |ΨD (t) = n=0 (n) |ΨD (t) (n) (8.133) (8.132–8. . 8. consequently.138) 1 i t dt VD (t ) |0 .139) .133–8. 2.134.129) n| and.137) can be solved recursively.137) follows |ΨD (t) = |0 . Accord(n) ingly.

according to (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).142). 8. We want to apply now the perturbation expansion derived to such perturbations. . 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. (8.135. (8.e. The coefficients A1 . + m=0 dt to to dt n|VD (t )|m m|VD (t ) |0 8.137) |ΨD (t) = (2) Interaction of Radiation with Matter 1 i 2 t t dt to to dt VD (t ) VD (t ) |0 .118). that two planar waves simulataneously interact with an electronic system. according to (8. i.6 Perturbations due to Electromagnetic Radiation We had identified in Eq. λ → 0+ has been introduced which describes that at time to → −∞ the perturbation is turned on gradually. The factor III in (8..220 This result.142) combining an incoming and an incoming or outgoing wave. λ → 0+ . the direction of ˆ the propagation being determined by k = k/|k|. For the sake of including the effect of superpositions of plane waves we will assume. In (8. 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. in turn. 8. II III (8. note that uj . A2 are defined through (8. non-singular.141) + .. defines the direction of the E-field of the ˆ radiation.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. to → −∞ (8. such that the combined radiation field is decribed by the vector potential A(r.41). The resulting perturbation on an electron system. however. is VS = ˆ ˆ V1 exp(−iω1 t) + V2 exp( iω2 t) eλt . This factor will serve mainly the purpose of keeping in the following derivation all mathematical quantities properly behaved.144) r is the position of the electron and p = ( /i) is the momentum operator of the electron.144) describes the propagation of the planar wave.34. A factor exp(λt).. yields for (8.143) 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) 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.

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. (8.6: Perturbations due to Electromagnetic Radiation 1st Order Processes 221 We employ now the perturbation (8. (8.149) Using the definition of VD stated in (8.128). 8.8.141).130) and (for m = 0) i i exp − Ho (t − to ) |m = exp − for (8.145) n|ΨD (t) (1) m (t − to ) |m . According to (8. (8.141) this is n|ΨD (t) = − (2) 1 2 ∞ m=0 t t dt to to dt n|VD (t ) |m m|VD (t ) |0 .140. For the 1st order contribution to the transition amplitude n|ΨD (t) = (1) 1 i t dt n|VD (t ) |0 to (8.145) we obtain then. employing the perturbation (8. (8.143) to the expansion (8.147) ˆ ˆ n|V1 |0 e−iω1 t + n|V2 |0 e Carrying out the time integration and taking the limit limt→−∞ yields (1) n|ΨD (t) = λ→0+ lim e λt exp ˆ n|V1 |0 o − o − ω1 ) t + + ω1 − n + iλ ( n i i ˆ + n|V2 |0 2nd Order Processes exp o − o ω2 ) t ± ω2 − n + iλ ( n .128) one obtains k|VD (t) | = k|VS (t)| exp i ( k − ) (8.150) and. (8.146) = 1 λ→0+ t→−∞ i lim lim × t dt exp to i ( n − o − i λ) t iω2 t × . using (8.143).148) We consider now the 2nd order contribution to the transition amplitude.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 .

e. (8..152).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 (8.148) and (8. We assume first that the first order transi(1) tion amplitude n|ΨD (t) is non-zero.131).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. We also assume for the final state n = 0 such that n|0 = 0 and p0→n (t) = | n|ΨD (t) |2 holds.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. respectively. i. in which case one can expect that it is larger than the 2nd (2) order contribution n|ΨD (t) which we will neglect. provide now the transition probability p0→n (t) according to Eq. For this rate holds .153) (8. Using (8.155) λ→0+ ( o We are actually interested in the transition rate. the time derivative of p0→n (t).148) and ∗ |z1 + z2 |2 = |z1 |2 + |z2 |2 + 2Re(z1 z2 ) (1) (8.

156) such that the 1st order contributions of the two planar waves of the perturbation simply add.156) ( o The period of electromagnetic radiation absorbed by electronic systems in atoms is of the order 10−17 s. The result supports our definition of incoming and outgoing waves in (8. the two terms apearing on the rhs.160) leads to final states |n with energy n = o ± ω2 .142–8. 8.142) ˆ ˆ The matrix elements n|V1 |0 and n|V2 |0 in (8. For the resulting expression the limit limλ→0+ can be taken. due to the uncertainty relationship would introduce a considerable perturbation to the system. In case the matrix elements are non-zero.143.158). however.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. one should average the rate over many periods of the radiation. to cancel the third term in (8.159) leads to final states |n with energy n = o + ω1 . The time average will be denoted by · · · t . 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. due to the vector potential A1 u1 exp i (k1 · r − ω1 t) . (8.158) play an essential role for the transition rates of radiative transitions.e. a property. Hence. The result of such average is. any attempt to do so.144). First.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. The second contribution to (8.158) ˆ + | n|V2 |0 |2 δ( − ω2 ) Obviously. is much shorter than could be resolved in any observation. which is observed through the so-called spectral intensities of transitions |0 → |n ..31) and (8. of this expression describe the individual effects of the two planar wave contributions of the perturbation (8. in fact. describing either an incoming or an outgoing plane wave due to the vector potential A2 u2 exp i (k1 · r ˆ ω2 t) . the matrix elements can vary strongly for different states |n of the system. The δ-functions appearing in this expression reflect energy conservation: the incoming plane wave contribution of (8. .8.144). Using lim x2 + 2 δ→0+ = π δ(x) (8. ˆ (8. i.

158).167) Taking now the limit limλ→0+ and using (8. 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.g. exp[ i (±ω2 − ω1 )]. as we did in the the case of 1st order transitions.e.161) To determine the transition rate we proceed again..162) × and.e.. (8.158). ∞ − 2 ω1 − 2i λ)t n − o − 2 ω1 − 2i λ ( n o − z2 = m=0 ˆ ˆ n|V2 |m m|V2 |0 ω2 − i λ m− o exp i × (8. in Eqs.168) absorption of 2 photons ω1 . We define ∞ z1 = m=0 ˆ ˆ n|V1 |m m|V1 |0 m − o − ω1 − i λ exp i × (8. Time average · · · t of expression (8.153) vanishes such that the leading contribution to the transition probability p0→n (t) arises from the 2nd order amplitude (8. |z1 + z2 + z3 |2 t = |z1 |2 + |z2 |2 + |z3 |2 (8. e.157) yields. (2) (8. hence.224 2nd Order Radiative Transitions Interaction of Radiation with Matter We now consider situations where the first order transition amplitude in (8. i. it holds p0→n (t) = | n|ΨD (t) |2 .166) In this expression the terms |zj |2 exhibit only a time dependence through a factor e2λt whereas the terms zj z ∗ k for j = k have also time-dependent phase factors. i..153–8. in analogy to (8.165) × ∞ 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=1 j=k 2 2 2 2 ∗ zj zk (8.152).164) (8.163) × (8.166) over many periods of the radiation yields exp[ i (±ω2 − ω1 )] t = 0 and. similarly.

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

m denote the relevant quantum numbers) ψn=1. t) = 2πN ωV u exp ˆ ı (k · r − ωt) + 2πN ωV u exp ˆ ı (−k · r + ωt) . 8.174) We seek to evaluate the matrix element ˆ M = n| p eik·r |0 .179) . according to (8.e. φ) a3 ao o 2p (8. (8.173) V1. the second term can be discounted in case of absorption. For example.175) The matrix element involves a spatial integral over the electronic wave functions associated with states |n and |0 .2 = m ωV Only the first term of (8. φ) (8. However. φ) = 6 r −r/2ao e Y1m (θ.143) will contribute to the absorption process.178) These wave functions make significant contributions to this integral only for r-values in the range r < 10 ao .171) The perturbation on an atomic electron system is then according to (8. to → −∞ (8. in this range one can expand eik·r ≈ 1 + ik · r + .226 Absorption of a Plane Polarized Wave Interaction of Radiation with Matter We consider first the case of absorption of a monochromatic. and the integral is =1. (8. i.m=0 (r.158). (8.m (r. φ) = 2 − 1 2 1 −r/ao e Y00 (θ. φ) × × eik·r e−r/ao Y00 (θ.144) VS = where e 2πN ˆ ˆ p · u e±ik·r .172) δ( n − o − ω) (8. φ) a3 o 1s (8. . The absorption rate.. θ. λ → 0+ .177) √ M = 6 0 ∞ 1 2π ia4 o r2 dr −1 dcosθ 0 ∗ dφ r e−r/2ao Y1m (θ. (8. θ.170) We will employ only the real part of this potential. in case of a radiative transition from the 1s state of hydrogen to one of its three 2p states. . .143. 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 . ˆ (8. t) = 8πN ωV u exp ˆ ı (k · r − ωt) .176) ψn=2. =0. the wave functions are (n. plane polarized wave described through the complex vector potential A(r. the vector potential actually assumed is A(r.

Ho ] = ˆ2 p [ r.178) holds eik·r ≈ A 1 1 + O( 50 ) such that one can approximate eik·r ≈ 1 . in case of the hydrogen atom. (8. pj ] = i δkj one obtains ˆ [ r.181) and (8. pk ] pk ˆ k=1 (8.186) δ( n − o − ω) (8. Ho ] |0 = i i n) n|r |0 .180) ∆E2p−1s and ao = 0.104) and.182) one obtains M ≈ Insertion into (8.182). One refers to this approximation as the dipole approximation. Ho ] i where Ho is the Hamiltonian given by (8. Using (8. Using |k| = 2π/λ.187) . V (r) = − .185) from which follows (8. pk ] + ˆ ˆ k=1 1 2me 3 [ r.179) and other terms are never larger than 20π ao /λ. the value of the wave length for the 1s → 2p transition 2π c λ = = 1216 ˚ A (8. Transition Dipole Moment A further simplification of the matrix element (8.175). We are now in a position to obtain an alternative expression for the matrix element (8.183) (8.174) yields kabs = 4π 2 e2 N ω V u · n| ˆ |0 ˆ r 2 m( o − m n| [r.529 ˚ one concludes that in the significant integration range in (8. ] + [ r. This simplification results from the identity m ˆ p = [ r.182) one finds [ r.184) and the commutation property [xk . is Ho = ˆ (p)2 e2 + V (r) .182) (8. 2me r (8.7: 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.k=1 i m 3 pk ek = ˆ j.8.k=1 i ˆ p m (8.175) can then be ˆ achieved and the differential operator p = i replaced by by the simpler multiplicative operator r. V (r) ] 2me =0 = Using r = 3 ˆ j=1 xj ej 1 2me 3 pk [ r. Ho ] = i m 3 pk ej δjk = ˆ j.

i. Absorption of Thermal Radiation We want to assume now that the hydrogen atom is placed in an evironment which is sufficiently hot. u2 and k ˆ ˆ coordinate system. and all polarizations of the radiation. directions.33) are orthogonal to each other. |0 . planar waves as defined in (8. ρ) holds then u cosα = cosϑ1 cosϑ + sinϑ1 sinϑ cos(ϕ1 − ϕ) = sinϑcosϕ and cosβ = cosϑ2 cosϑ + sinϑ2 sinϑ cos(ϕ2 − ϕ) = sinϑsinϕ . such that the thermal radiation present supplies a continuum of frequencies. ϕ. according to (8. We have demonstrated in our derivation of the rate of one-photon processes (8. u2 . the direction of u1 is described by the ˆ ˆ ˆ ˆ angles ϑ1 = π/2.190) (8. x2 .193) . 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. ϕ1 = 0 and of u2 by ϑ2 = π/2. will actually be replaced by a distribution function which reflects (1) the finite life time of the states |n .174) one can replace n − o by ω. The δ-function appearing in this expression. 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. one can express | u · n| r |0 |2 = |ρ|2 ( cos2 α + cos2 β ) = sin2 θ . and (2) the fact that strictly monochromatic radiation cannot be prepared such that any radiation source provides radiation with a frequency distribution. in practical situations. say u1 and u2 . The direction of this vector in the u1 . x3 -axes of a right-handed cartesian vectors u1 . ϕ2 = π/2. ρ) and ˆ u β = ∠(ˆ2 . obtain the transition rate in the present case by adding the individual transition rates of all planar waves present in the oven.189) where the factor 1/ arose from the integral over the δ-function. hence.228 Interaction of Radiation with Matter where we used the fact that due to the δ-function factor in (8. three-dimensional vector.e.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 (8.31–8. The unit ˆ ˆ ˆ ˆ can be chosen to point along the x1 . a very hot oven. ˆ u ˆ (8.35).191) (8. there are ony two linearly independent directions of u possible. ˆ In order to carry out the integral dk we note that u describes the possible polarizations of the ˆ ˆ and u. Accordingly.158) above that in first order the contributions of all components of the radiation field add. Integrating and summing accordingly over all contributions as given by (8. k frame is described by the angles ϑ.. For the two angles α = ∠(ˆ1 . k ˆ As a result. We can.188) ˆ Here dk is the integral over all orientations of k.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.

However. For this purpose we use ω/c = 2π/λ and obtain 4 e2 ω 3 32π 3 e2 ao 1 ao = = 1. Nω = 0. as for the absorption process.. 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. can be inserted into (8.194) combined are typically somewhat smaller than (1 ˚/1000 ˚)3 = A A 10−9 .171) and perturbation (8. The corresponding process is termed spontaneous emission.37 × 1019 × 3 3 c3 3 ao λ3 s λ and kabs where λ = 2πc n − (8. 8.. . for Nω = 0. for atoms or symmetric molecules. The value of | n| r |0 | determines the strength of an optical transition.. However. Transition Dipole Moment The expression (8.189) to yield the total absorption rate kabs (tot) = Nω 4 e2 ω 3 | n|r|0 |2 3 c3 . in particular.8. . However.37 × 1019 1 ao | n|r|0 |2 × s λ λ2 . ˆ In case of emission only the second term V2 exp(+iωt) in (8. The transition dipole moment can vanish for many transitions between stationary states of a quantum system. in agreement with observations.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 |. that emission occurs even if no photon is present in the environment. In case of a classical radiation field one would expect that emission cannot occur.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 . i.e. however.e. The radiation field is described.173). (8. the calculation of the emission rate proceeds as in the case of absorption. a problem in case of emission by quantum systems in complete darkness.197) o (tot) (8.173) contributes. Nω photons before absorption. This difference between emission and absorption is due to the quantum nature of the radiation field. there is also a contribution to the emission rate which is proportional to Nω . the resulting total rate of emission bears a different dependence on the number of photons present in the environment. linear molecules. This dependence predicts.196) The last two factors in (8.172. . (8. the absorption rate is of the order of 109 s−1 or 1/nanosecond. Emission of Radiation We now consider the rate of emission of a photon. 1. i.7: One-Photon Emission and Absorption 229 This geometrical average. The most intensely absorbing molecules are long.195) = Nω 1. finally. Accordingly. For practical evaluations we provide an expression which eliminates the physical constants and allows one to determine numerical values readily. Otherwise. This poses. 2 . The quantum nature of radiation manifests itself in that the number of photons Nω msut be an integer. by planar waves with vector potential (8.

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

ˆ (8. (8. The resulting rate of the transition depends on 2 Nω . (8. even if n| r |0 vanishes. If the sum in (8. under ordinary circumstances. as described through the vector potential ˆ A(r.168) where the first term describes the relevant contribution. 8. finally.173). Employing the dipole approximation (8.187). We wish to describe an observation in which a detector is placed at a solid angle element dΩ2 = sinθ2 dθ2 dφ2 with respect to the origin of the coordinate system in which the electron is described.207) one is. In applying (8. a transition between the states |0 and |n may be possible. t) = Ao1 u1 cos(k1 · r − ω1 t) .8.208) has been prepared. dropping the index 1 characterizing the radiation.205) The respective radiative transition is of 2nd order as described by the transition rate (8. Scattering of Photons at Electrons – Kramers-Heisenberg Cross Section We consider in the following the scattering of a photon at an electron governed by the Hamiltonian Ho as given in (8.172.207) 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.207) does not converge rapidly.207) does not provide a suitable avenue of computing the rates of 2-photon transitions. One obtains. in case ˆ of absorption only V1 contributes.207) Expression (8. The electron absorbs the radiation and emits immediately a second photon. 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. The perturbation which accounts for the coupling of the electronic system and the radiation field is the same as in case of 1st order absorption processes and given by (8. however. (8.182) yields. The transition requires that the transition dipole moment n| r |0 does not vanish for two states |0 and |n . . however. We assume that a planar wave with wave vector k1 and polarization u1 . induce single photon absorption processes as described by the transition rate Eq. which is not necessarily the case. then expression (8.8: Two-Photon Processes Two-Photon Absorption 231 The interaction of electrons with radiation.181) and using (8. The intense radiation fields of lasers allow one to increase transition rates to levels which can readily be observed in the laboratory. but then requires the absorption of two photons. In this case one needs to choose the energy of the photons to obey n = o + 2 ω. 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) . However. faced with the dilemma of having to sum over all intermediate states |m of the system.206) − o − 2 ω) .104) with stationary states |n defined through (8.106).

according to the ˆ description of emission processes on page 229.141).210).211) is stated in Eq. 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  (8.208). t) = A+ u1 exp[ i (k1 · r − ω1 t) ] + A− u2 exp[ i (k2 · r − ω2 t) ] . t) = Ao2 u2 cos(k2 · r − ω2 t) . The second term in (8.105) using the vector potantial (8.. This expansion yields. ˆ The perturbation which arises due to the vector potential (8.210) is accounted for by the following contributions of (8. 8.211) is o2 A− = o2 8π (N2 + 1) ω1 V (8.213) where N2 /V is the density of photons characterized through k2 .211) accounts for the emitted wave and. ˆ (8. the wave characterized through k1 .105). later we will allow the quantum system to select appropriate values. The vector potential describing the emitted plane wave is then A(r. (8.210) The vector potential which describes both incoming wave and outgoing wave is a superposition of the potentials in (8.e. 8. the amplitude A+ is given by o1 A+ = o1 8πN1 ω1 V (8.212) where N1 /V is the density of photons for the wave described by (8.208. i. u2 . In the present case we consider only scattering processes which absorb radiation corresponding to the vector potential (8. u1 .208) and the emission of the radiation in (8.208) and emit radiation corresponding to the vector potential (8. hence. the amplititude A− defined in (8.211) The first term describes the absorption of a photon and.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 (8.232 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 .209) cosθ2 where the value of k2 has been fixed. however. We know already from our description in Section 8.210). for the components of the wave (8.6 above that the absorption of the radiation in (8. The relevant terms of the perturbation (8.208. in the present case.210) A(r. o1 ˆ o2 ˆ (8.214) .

e. Accordingly. one needs to integrate the rate as given by (8.181) in stating this result. Inserting also the values (8. Only the second (1st order) and the third (2nd order) terms in (8. Hence.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. 8.216) over all available modes of the field.217) 2 Here ro denotes the classical electron radius ro = e2 = 2.215) correspond to scattering processes in which the radiation field ‘looses’ a photon ω1 and ‘gains’ a photon ω2 . evaluating the time integrals in (8.218) .8.212.216) 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 . i..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. me c2 (8. Following closely the procedures adopted in evaluating the rates of 1st order and 2nd order radiative transitions on page 222–225. only these two terms contribute to the scattering amplitude.8 · 10−15 m .e.. i.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 (8.215) 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. one needs to carry out the integration 2 V(2π)−3 k2 dk2 · · ·.

. one can relate (8.228) S (0) = m S (0) = 2 m . i.234 Interaction of Radiation with Matter The factor N1 c/V can be interpreted as the flux of incoming photons. ω1 = ω2 . in fact. for all states |m of the electronic system (8.. 2 . one assumes the even stronger condition ω1 << | o − m| . − o )2 ( m − o )3 (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.221) suggests to expand S( ω) S( ω) = S(0) + S (0) ω + Using 1 − o ± ω one can readily determine m (8.227) 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.e. (8.225) S(0) = m u ˆ 0|ˆ1 · p |m m|ˆ2 · p |0 u ˆ u ˆ 0|ˆ2 · p |m m|ˆ1 · p |0 u ˆ + me ( m − o ) me ( m − o ) (8. a process in which the electronic state remains unaltered after the scattering.219) (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. Accordingly.224) = m 1 − o ( m ω ( ω)2 + + .223) m Condition (8. consequently.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 + .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..221) Using |n = |0 and. it follows 2 dσ = ro (N2 + 1) dΩ2 |ˆ1 · u2 − S( ω) |2 u ˆ (8. 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. Rayleigh Scattering We turn first to an example of so-called elastic scattering.

k=1 (ˆ1 )j (ˆ2 )k δjk = u1 · u2 u u ˆ ˆ (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. For this purpose we replace p using (8.236) is of great practical importance. It explains. as a compass.226). Using again (8.108) for the identity operator. .e.182) for p and the expression (8.8: Two-Photon Processes 235 ˆ These three expressions can be simplified using the expression (8.8.229) both for the u1 · p and the to u2 · p terms in (8.232) Obviously.108) this is 1 0|ˆ1 ·r u2 · p − u2 · p u1 ·r |0 . Expression (8. u1 ·r ] |0 = 0 u ˆ (8. ˆ We want to simplify first (8.232.228) provides then the first non-vanishing contribution to the scattering cross section (8. 8.108) yields S (0) = me 2 0| [ˆ2 ·r. For this purpose we apply (8. i.229) (8.235) and obtain the leading contribution to the expression (8. m|ˆ1 · p |0 u ˆ 1 = − m|ˆ1 ·r |0 u me ( m − o ) i (8. 8.228) we obtain ˆ ˆ ˆ ˆ S (0) = 2me 2 m 0|ˆ2 ·r |m m|ˆ1 ·r |0 u u 0|ˆ1 ·r |m m|ˆ2 ·r |0 u u + m − o m − o (8.222) of the cross section for Rayleigh scattering dσ = 2 ro m2 ω 4 (N2 + 1) dΩ2 × e (8.224.234.182) 0|ˆ1 · p |m u ˆ 1 = 0|ˆ1 ·r |m u me ( m − o ) i This transforms (8.235) We can now combine eqs.227) vanishes.233) m Employing again (8.222)..234) where we used for the second identity the fact that u1 ·r and u2 ·r commute. pk ]|0 = u u ˆ j. (8. for example. ˆ ˆ S (0) given in (8. the blue color of the sky and the polarization pattern in the sky which serves many animals.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 ) . u ˆ ˆ ˆ ˆˆ i The commutator property [xj . this term cancels the u1 · u2 term in (8.231) (ˆ1 )j (ˆ2 )k 0|[xj . 8.222). ˆ ˆ We want to prove now that expression (8. honey bees.230) According to (8.k=1 j. pk ] = i δjk yields finally ˆ S(0) = 1 S(0) = i 3 3 (8. u u u u (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 .

237) The resulting scattering is called Thomson scattering.244) .238) (8. |n = |0 and ω1 = ω2 = ω in (8.220) 2 dσ = ro (N2 + 1) dΩ2 |ˆ1 · u2 |2 . ao Bohr radius .220).e.. (8.and the x2 -axes     1 0 (1) (2) u1 =  0  . however. for all states |m of the electronic system . u2 = u2 ˆ ˆ ˆ ˆ (8. the emitted u ˆ radiation is decribed by the wave vector   sinθ2 cosφ2 k2 = k1  sinθ2 sinφ2  (8. We want to assume. u1 =  1  ˆ ˆ (8. hence.241) 0 0 Similarly. u2 = u2 ˆ ˆ ˆ ˆ (1) (2) (2) (1) (1) (2) (1) (2) (2) (1) needs to be (8.240) cosθ2 We choose for the polarization of the incoming radiation the directions along the x1 . 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 .236 Thomson Scattering Interaction of Radiation with Matter We consider again elastic scattering.239) We will show below that this expression decribes the non-relativistic limit of Compton scattering. assume now that the scattered radiation has very short wave length such that ω >> | o − m| . i.240). u2 = u2 ˆ ˆ ˆ ˆ cos2 φ2 cos2 θ2 sin2 φ2 for u1 = u1 . u ˆ 1 ao . though. u2 = u2 ˆ ˆ ˆ ˆ for u1 = u1 . 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 . that the dipole approximation is still valid which restricts the applicability of the following derivation to k1 >> One obtains immediately from (8. (8..243) cos2 θ2 cos2 φ2 for u1 = u1 . To evaluate |ˆ1 · u2 |2 we assume that k1 is oriented along the x3 -axis and. The first choice is ˆ   sinφ2 k2 × k1 (1) u2 = ˆ =  −cosφ2  (8.242) |k2 × k1 | 0 where the second identity follows readily from k1 = e3 and from (8.

If energy is absorbed by the system.e. (8. defined in (8. In case that the nuclear degrees of freedom excited absorb very little energy. i. Since the scattering is inelastic.. in case of a diatomic molecule |n = |φ(elect. The energy deficit is used to excite the system.. the scattered radiation is polarized to some degree. In the case that the scattering excites other than electronic degrees of freedom. φ(vibr.220) represent actually electronic as well as nuclear motions. or in case of translational motion of molecules in liquids. 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 first term in (8. i.247) becomes identical with the Thomson scattering cross section (8.246). Such scattering is called Raman 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. For electronic systems in molecules or crystals the degrees of freedom excited are nuclear motions. The excitation can be electronic.)n . the states |n etc. one speaks of anti-Stokes scattering. but most often involves other degrees of freedom.220) vanishes and one obtains in case of Raman scattering ω2 2 dΩ2 | u2 · R · u1 |2 ˆ ˆ (8.247) ( 1 − cosθ2 ) (8. Raman Scattering and Brillouin Scattering We now consider ineleastic scattering described by the Kramers-Heisenberg formula. as in the case of excitations of accustical modes of crystals. if energy is released.g. one speaks of Stokes scattering.. the resulting scattering cross section is 2 ro (N2 + 1) ( 1 + cos2 θ2 ) dΩ2 .8.245)  1 cos2 θ for u = u(2) ˆ ˆ 2 2 2 2 The result implies that even though the incident radiation is unpolarized. The radiation scattered at right angles is even completely polarized (1) along the u2 -direction.247. 8. the scattering is termed Brillouin scattering.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 .)n . molecular vibrations or crystal vibrational modes. ˆ In case that one measures the scattered radiation irrespective of its polarization.248) me c2 One can readily show that in the non-relativistic limit.246) 2 This expression is the non-relativistic limit of the cross section of Compton scattering.249) dσ = ro (N2 + 1) ω1 −1 −1 ω2 − ω1 = . for c → ∞ the Compton scattering cross section (8. In the case of such scattering an electron system absorbs and re-emits radiation without ending up in the initial state. e.e.

250) (8. Of course. e. a case called resonant Raman scattering. the cross section for resonant Raman scattering can be several orders of magnitude larger than that of ordinary Raman scattering.g. the Raman scattering cross section will be much enhanced. a property which can be exploited to selectively probe suitable molecules of low concentration in bulk matter. Nevertheless. ω1 ≈ m − o for a particular state |m .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. In case that the incoming photon energy ω1 is chosen to match one of the electronic excitations.251) ω1 − ( We define x · R · y = j..k xj Rjk yk . . no singlularity developes in such case due to the finite life time of the state |m .

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

. one can state P Ψ(x1 . For the permutations P1 and P2 given above holds P1 = T (3. x2 . . 5). the top row representing the numbers 1 to N labelling the N particles under consideration. i..e. . Denoting by P (j) the image of j. We can then state P2 = T (4. xj . 4) T (4. k) of the two particles transposed. xP (2) . . . 6) T (6. the set of all permutations of N objects. in particular. k) = (4. . in notation (9. (9. . xN ) = Ψ(xP (1) . a given P can either only be presented by even numbers of transpositions or by odd numbers of transpositions.240 Many–Particle Systems repeatedly like degenerate states.3) the j-th index in the second row. 8) .e. The product P1 P2 of even and odd permutations P1 and P2 obey the following multiplication table: P2 \ P1 even odd even even odd odd odd even We finally determine the action of permutations P on the wave functions (9. . We will not discuss here at any depth the properties of the permutation groups SN . A reader interested in the quantum mechanical description of N particle systems is strongly encouraged to study these representations. (9. namely implicitly the group property.. . . P2 = T (4.3) P2 which affects only two particles. For the following we will require only two properties of SN .e. 5) T (5. . i. 5) in case of P2 . (j. . i. One calls permutations of the first type even permutations and permutations of the second type odd permutations. and the second row showing the numbers 1 to N again. rather only states which obey a certain transformation property under permutations are acceptable. . i. The elements P ∈ SN can be represented conveniently as 2 × N matrices. . . However.. k). particles 4 and 5. 5) (9. This property will be established now.2). The group SN is then composed of two disjunct classes of even and of odd permutations. 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. k) are characterized by the two indices (j. xP (N ) ) . number k under the entry j of the first row indicating that particle j is switched with particle k. Transpositions denoted by T (j. The permutations we need to consider for a system of N particles are the elements of the group SN . is called a transposition.e. . but in a different order. . The latter factorization has the essential property that the number of factors is either even or odd. . the representations in the space of N –particle wave functions are extremely useful in dealing with N –particle systems..4) both permutations being obviously odd. 3) T (7.5) . even though these properties. and explicitly the fact that any P ∈ SN can be given as a product of transpositions T (j. 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 = . . not any state can be chosen as to represent the many– particle system. xP (j) . leaving all other particles unchanged. P2 = . .

In fact. for the scalar product holds j|k = δj k . xk ) is symmetric in its two arguments.e. Presently. Obviously. implies that the permutation operators can be chosen in a diagonal representation. i.-1}. In the latter case all even permutations are represented by ‘1’ and all odd permutations by ‘-1’.9) = 1 for P even −1 for P odd (9. −1. . .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). particles with half–integer spin values. i.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. In fact. i. particles with integer spin values. hence.8) |Ψ (9. the Hamiltonian must be of the type N H = j=1 1 F (xj ) + 2 N G(xj . S where S denotes the number of available single particle states. the Hamiltonian for permuted indices P HP −1 is identical to H. . 1 together with multiplication. provide a representation of the group structure represented by the multiplication table of even and odd permutations given above. The first case applies to bosons. all particles being governed by the same interactions.11) .e. both groups. a diagonal representation can only berealized in a simpler group. xk ) j. in turn. it is a postulate of quantum mechanics that for any description of many–particle systems the permutation operators must be diagonal.5) implies that a permutation effectively changes the indices of the particles) and.e.. or the corresponding group of two elements 1. ∀P ∈ SN . 2.9. From this property follows that H is independent of any specific assignment of indices to the particles (note that (9.. {1} (trivial) and {1.k=1 (9. i.10) We will construct now wave functions which exhibit the appropriate properties. P ] = 0 . In fact. i. Since permutations do not necessarily commute.e. (9. it is essential that the functions F ( ) and G( . (9. ) are the same for all particles and for all pairs of particles.. (9. The terms of the Hamiltonian will be discussed further below. From the latter results [H. the representation is either isomorphic to the group composed of only the ‘1’.. The single particle states are assumed to be orthonormal. usually a very large number. G(xj . k = 1. the second group to fermions.6) which describes one–particle interactions and two–particle interactions of the system. 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.

. n2 . ..9). The numbers nj . S} of the type (9. λN ) = j=1 φλj (xj ) . . . . . i. .15) with λk = j. .. . . 9. . . .. the reason is that such transposition duplicates the state which would.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. . is a particular choice of placing N particles consistent with the occupation numbers (n1 . 9.8) is x1 . λ2 . λ2 . . . . . x2 . λ2 . . . . . λj ∈ {1. . λN ) ⊂ SN is the set of all permutations involving only particles j with different λj . .8. . . It holds N ΨFock (λ1 . . . . . This detail is connected with the fact that a two–particle state of the type ΨFock (λ.12). with λj = λk = . are essential. . λN )|ΨFock (λ1 . . referred to as the occupation numbers. .g. (9. k. The Fock states represent N particles which are placed into S states |λ1 . We first consider orthonormal many–particle wave functions. T (j. i. therefore. 9.15) (nj !) N! 1 2 P ∈Λ(λ1 . The integers nj are equal to the number of λk in (9. |λN .λN ) P x1 .13) These states form an orthonormal basis of N -particle states. The reader is well advised to pause and grasp the definition of the nj . each particle j into a specific state |λj . . . . A wave function which obeys the boson symmetry (9. .11. 2. . λ2 . . . λN ) do not exhibit the symmetries dictated by (9. . n2 . xN |ΨFock (λ1 .14) Obviously. In a second step we form linear combinations of Fock states with the desired symmetries. hence.. characterize the wave function (9. the so-called Fock-states. the states |ΨFock (λ1 . which do not yet obey the symmetries (9. An important detail of the definition (9.. . . N x1 .. e. . . 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.15) completely and.12) where the second factor represents the scalar product between spin states. λN ) = j=1 δλj λj . . the nj specify how often a single particle state |j is occupied. .15) is that the sum over permutations does not involve permutations among indices j. xN |ΨB (n1 . k) ∈ Λ(λ1 . xN |Ψ Fock (λ1 . . . . λ2 . . x2 . . |λ2 . . Λ(λ1 . . x2 .8. . nS ). . (9.e. .15). nS ) = S j=1 (9. λN ) only if λj = λk .λN ) on the rhs.e. λ) in which two particles occupy the same single– particle state |λ does not allow transposition of particles. . λ2 . . . . .9).λN ) Here the states (λ1 . appear severalfold in (9.

.9). . . the photons of the electromagnetic field or the phonons in a crystal. . (b) Show that for the states (9. . . . (9. . electrons in an atom. . . . .λN ) (9.16). . . φλN (xN ) Here. . n2 . For any value nj > 1 two or more of the columns of the Slater matrix are identical and the wave function vanishes. n2 . . . nS ) holds S Ψ (N )|Ψ (N ) = j=1 B B δnj nj . .. for two sets of occupation numbers N = (n1 . . . ... nS ) = φλ1 (x1 ) φλ1 (x2 ) .1. x2 .e. Wave Functions with Fermion Symmetry in the Occupation Number Representation One can construct similarly a wave function which obeys the fermion symmetry property (9. √1 N! (9..1: Bosons and Fermions 243 The wave functions (9. . x2 .15) holds (9. xN |ΨF (n1 . . . nS ) holds S ΨF (N )|ΨF (N ) = j=1 δnj nj . . correspond to a particular choice of placing N particles consistent with the occupation numbers (n1 . . . . the integers nj denote the occupancy of the single–particle state |j . . .1: (a) Show that the states (9. . .17) describe. . . .16) Exercise 9. λ2 . . a molecule or a crystal. n2 . . . i.. . for example.18) this wave function can also be expressed through the so-called Slater determinant x1 . Such wave function is x1 . Using the identity governing the determinant of N × N –matrices A N det(A) = P ∈SN P j=1 Aj P (j) (9.19) φλ1 (xN ) φλ2 (xN ) . nS ) = √1 N! P ∈SN P P x1 . x2 . . . . . .15) obey the boson symmetry (9. .20) . The fermion states (9. . . . . xN |ΨFock (λ1 . nS ) and N = (n1 .e. . nS ). . n2 . n2 . . . xN |ΨF (n1 . φλN (x1 ) φλN (x2 ) . φλ2 (x1 ) φλ2 (x2 ) . These states form an orthonormal basis. nS ) and N = (n1 .17) Here the states (λ1 .λN ) on the rhs. .8). The important property holds that nj can only assume the two values nj = 0 or nj = 1.9. These states form an orthonormal basis. n2 . i. for two sets of occupation numbers N = (n1 . . . .15) describe. (9. for example. λ2 . .. n2 .

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

9. a† ] = 1 For j = k follows similarly 1. . (9.21. . . (9. obviously. . (9. n2 .27. The related operator S ˆ N = j=1 ˆ Nj (9. . . i. . j (9.33) is called the particle number operator since. . .22) ˆ Nj |ΨB (N ) = nj |ΨB (N ) . nS ) = j=1 a† j nj nj ! |0 . nS ) N |ΨB (n1 . nS ) = j=1 nj |ΨB (n1 .. . . S ˆ N |ΨB (n1 . .34) = .21) and vice versa..23) follow in an analogous way. . . .24). The creation operators a† allow one to construct boson states ΨB (N ) from the vacuum state j |0 = |ΨB (n1 = 0. the eigenvalues of Nj are the occupation numbers nj . (9. . n2 .28) (9. the properties (9. We will prove the commutation properties (9.29) |ΨB (n1 . .27) Equations (9. (9. .32) ˆ ˆ i. 9.e. . n2 = 0.e. n2 .26) From this result one can immediately conclude that (9. . nS ) = 0 . nS ) .30) Of particular interest is the operator ˆ Nj = a† aj . It holds for j = k (aj a† − a† aj )|Ψ(N ) = (nj + 1 − nj )|Ψ(N ) = |Ψ(N ) . . nS ) = |ΨB (N ) . . . . nk + 1 .22) follows from (9. . . . .29) yield the commutation relationship (9.31) This operator is diagonal in the occupation number representation. j j Since this holds for any |Ψ(N ) it follows [aj . n2 . .24). 9. nj − 1 . 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 . n2 . nS = 0) as follows S (9. One refers to Nj as the occupation number operator. j (aj a† − a† aj )|Ψ(N ) = k k ( nj (nk + 1) − (nk + 1)nj )|Ψ(n1 .2: Operators of 2nd Quantization Let A† be the matrix adjoint to Aj . One can readily show using (9. . . for basis states |ΨB (n1 .

24) and the property aj |0 = 0. ...36) Creation and Annihilation Operators for Fermions We want to consider now creation and annihilation operators for fermions.38) corresponds to x1 . .23. . . As long as one starts from a wave function with the proper boson symmetry. i. .. . Prove this by induction. . .37) (9.2.e.35) and where the derivative operation is defined through ∂f (A) = ∂A ∞ n=1 1 f (n) (0) An−1 . .39) . since they are induced through the algebra of a† and aj . φj (x1 ) φj (x2 ) .. Before proceeding in this respect we need to account for the following property of the fermion wave function which applies in the case nj . . it also holds for nj + 1.2: Show that the boson operators a† and aj satisfy j [aj . . . To prove this property we notice that the l. .s. nj+1 . in general. j Exercise 9. . nj+1 . . 9.23. x2 . nj .. from the vacuum state |0 or states |ΨB (N ) . . . Exercise 9. .17. 9. nj . φj (xN ) φj+1 (xN ) . . nj+1 = 1. the fermion wave function changes sign when one exchanges the order of the occupancy.. . . φj+1 (x1 ) φj+1 (x2 ) . nS ) = . in case it holds for nj . .. . .9). 9.19) without affecting the fermion symmetry (9. . Obviously. (9. 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. .1: The commutation relationships (9. .2.. nS ) = − x1 . e. 9. operators which can alter the occupancy of the wave function (9. (n − 1)! (9.29. . showing that the property holds for nj = 0 and. 9. . nj (9.21.. . . one does not need to worry ever about proper symmetries of wave functions. i. x2 . . . . x2 . . nS ) . . . xN |ΨF (n 1 . . in case that the single particle states |j and |j + 1 are both occupied.30). nj+1 .23) hold for the state defined through (9. of (9.e. √1 N! .246 Many–Particle Systems In using the boson creation and annihilation operators. one needs to apply only the commutation propertes (9. f (a† )] j [a† . . .24) and aj |0 = 0 imply that the the properties (9.. xN |ΨF (n1 . . x1 . xN |ΨF (n1 ..22. . 9.38) .g.h. .

. We are now ready to define the annihilation operator cj for fermions in the single–particle state |j as follows cj |ΨF (n1 . i. nj → 0.e. .. n2 = 1.19). xN |ΨF (n1 . We will be adopting below the notational convention that qj corresponds to occupancies N = (n1 . however. n6 = n7 = . Because of the property of the determinant to change sign when two columns are interchanged. . . . x2 . x2 |ΨF (n1 = 0. A proper example is the two particle fermion wave function x1 ... φj+1 (xN ) φj (xN ) . . .s. n2 . . 2. that occupancies are always ordered according to a monotonous increase of the single–particle state index. i. it is necessary to specify for a fermion wave function the order in which the single– particle states |λj are occupied. nS = 0) φ2 (x1 ) φ5 (x1 ) 1 1 = √2 = √2 [φ2 (x1 )φ5 (x2 ) − φ5 (x1 )φ2 (x2 )] . nj . .. nS ) as follows: j−1 qj = k=1 nk .39) and (9. . . . This change of position is connected with a possible sign change (−1)qj where qj is defined for a given N = (n1 . . For this purpose one must adhere to a strict convention: the labelling of single–particle states by indices j = 1. One adopts the convention that particles are created and annihilated by these operators always at the first position of the wave function. . . . .43) . . . n2 . . . . particles are being created or annihilated. n5 = 1. i.. . φ2 (x2 ) φ5 (x2 ) (9. ..40) . ..40) are identical. nS ) = nj (−1)qj |ΨF ( n1 .42) i. nj+1 . vanishes. This requires one.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. at which column of the Slater determinant (9. .. nj+1 . nj−1 . n3 . at the first column of the Slater determinant (9. φj (x1 ) φj (x2 ) . the expressions (9. qj is the number of states |k with k < j which are occupied in a fermion wave function.e. n3 = 0.e.. (9.9.h. nj = 0. . . . . . n4 = 0.19). . 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). nS ) qj states occupied In case that the single particle state |j is not occupied. . n2 . . . . nS ) = . of (9. (9. . n2 . .. . . the rhs.38) reads − x1 .. . Obviously. . 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. nj . in order to be consistent with this convention. . .e. φj+1 (x1 ) φj+1 (x2 ) . .. (9. . 247 − √1 ! N .2: Operators of 2nd Quantization The r. nS ).

. .44) (9.19). .48) vanishes.44).17. nS ) = j Many–Particle Systems (−1)qj (1 − nj ) |ΨF ( n1 . . otherwise. We consider (9. B]+ = AB + BA. nS ) = j j (−1)qj (1 − nj + 1)(−1)qj nj |ΨF (N ) = nj ΨF (N ) (9. nS ) qj states occupied The operators thus defined obey the commutation properties [cj . nj+1 . c† ]+ = 0 j k (9. c† ]+ = δj k k where we have used the definition of the so-called anti-commutators [A.46) [cj . . nj−1 . . n2 .48) From this follows (9. (9. The anti-commutation properties (9. nS ) = j (−1)qj (nj + 1))(−1)qj (1 − nj ) |ΨF (N ) = (1 − nj ) |ΨF (N ) (9. . 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 . We will prove now the anti-commutation property (9. (9. . . n3 .43). the last factor on the rhs. . of (9.50) .45) follow in an analogous way and will not be derived here. 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 . . n2 . . It holds c† cj |ΨF (N ) = c† (−1)qj nj |ΨF (n1 .44) follows from the definition (9. . Let Cj be the matrix corresponding to the operator cj in the basis of fermion states (9. nj . ck ]+ = 0 . .47) Let C† be the matrix adjoint to Cj . [c† . nj → 1. Similarly one obtains cj c† |ΨF (N ) = cj (−1)qj (1 − nj ) |ΨF (n1 . 9. nj → 1 .46) first for the case j = k.46). We first show that (9. . nj → 0 .45) (9. . . We have used here the definition qj = k<j nk along with qj = qj .49) The derivation involves realization of the fact that a non-zero result is obtained only in case nj = 1.248 The operator c† adjoint to cj exhibits the following property j c† |ΨF (n1 .

Equation (9. Correspondingly.52) One can obtain similarly the same relationship in the case j > k.57) On the l.h. hence. . S (9.) = c† 1 c† 2 · · · c† N |0 .50) yield ( cj c† + c† cj ) |ΨF (N ) = |ΨF (N ) . the creation operators must operate in the proper order. nλ2 . 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 . hence. . We j will often separate the states |j . On the r. nk = 1 . nλ1 . . of this equation we meant to indicate only those N occupation numbers nλj which are non-vanishing..46). i. with eigenvalues equal to the occupation numbers nj ˆ Nj |ΨF (N ) = nj |ΨF (N ) ˆ Nj ..51. . coordinates x and index j into a space part and a spin part. . .49) shows that the operator ˆ Nj = c† cj j (9. i.29)) |ΨF (. nS ) k and.h. .53) is diagonal in the occupation number representation. .s. j j For j < k one obtains cj c† |ΨF (N ) = (−1)qj +qk nj (1 − nk )|ΨF (n1 . 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 . nj = 0 . . nj = 0 . k k 249 (9. . is referred to as the occupation number operator. j → j. 9.e.e.58) . σ 1 (σ = ± 2 ) .e. 2 (9. .54) ˆ N = j=1 ˆ Nj (9. .9. 1 In the following we will apply fermion operators c† and cj only to electrons which carry spin 2 . 1}. in the basis |ΨF (N ) . i.51) (9. .2: Operators of 2nd Quantization (9.46). x|j → φj (r) | 1 σ . 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 (9.49) and (9.55) is called the particle number operator. . nλN . Since (9.56) Here we have made use of cj cj = 0 and c† c† = 0 which follows from nj ∈ {0. for which purpose one needs to envoke only the algebraic (anti-commutation) properties (9. . nk = 1 . . . . . an operator cj must be on the left of an operator ck for j < k.s.52) hold for any |ΨF (N ) one can conclude (9.. j j j j j j j (9. ( cj c† + c† cj ) |ΨF (N ) = 0 . . λ λ λ (9.

59) ˆ ˆ are called single–particle operators.e. These operators had been introduced already in Eq. . 2.F (N ) .3 One– and Two–Particle Operators Definition Operators acting on N particle systems of the type N ˆ F = j=1 ˆ f (xj ) (9. 9. S of single–particle spinorbital states (see page 241) and the operators are specified through the associated generic operators ˆ f and g .62) These matrix elements can be obtained by chosing the operators and many-particle states in the position representation. (9. ˆ .F (N )|G|ΨB. 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. 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 (9.61) is not a one particle operator as long as the N operators rj (x). Operators of the type 1 ˆ G = 2 N g (xj . The operators g (x.59) and (9. 9. ˆ ˆ ˆ We seek to determine now how one–particle operators F and two–particle operators G act on many– B (N ) and |ΨF (N ) .6) above. .. respectively – which are all identical. x2 . The essential aspect of definition (9. The action of the operators boson and many–fermion states |Ψ is obviously described through the many–particle state matrix elements ˆ ΨB. .17) and integrating over all particle space-spin coordinates x1 . .60) involves generic operators – acting on xj and on xj . .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. An operator N ˆ R = j=1 rj (xj ) ˆ (9.250 Many–Particle Systems 9.59) and over pairs of particles in (9.15. σr ) f (x) ψs (r. N are not all identical. ˆ r|f |s = 1 ∗ ˆ d3 r ψr (r. r = 1. xN . .59.F (N )|F |ΨB.60) is that the sum over particles in (9. .60) ˆ correspondingly are called two–particle operators. The operators f (x) which constitute F are called the associated generic single–particle operators1 . xk . x ) which constitute G are ˆ called the associated generic two–particle operators. j = 1.17). ˆ ΨB.59. σs ) (9.F (N ) . . (9. We adopt a similar expression to distinguish two–particle operators G and g . 9. xk ) ˆ j. Since the many–particle states are built-up from a basis |r . . i. namely. (9. 2. respectively. .k=1 j=k (9. 9.15.

64). σr )ψs (r . that the symmetry implies that one can switch simultaneously the pairs of indices r. r|ˆ|u. An example for a two–particle operator is the Coulomb repulsion operator 1 ˆ V = 2 N j. is t(x) = (− 2 /2m)(∂ 2 /∂r 2 ). ˜ (9.71) .59). one cannot switch the indices of one pair individually. according to (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 . i. u which follows directly g from the definition (9..67) ˆ In this case the generic single–particle operator. u. σu ) .3: One– and Two–Particle Operators r. σt )ψu (r . s and t.e. x ) ψt (r. s|ˆ|t.70) Both operators.63. σs ) g (x. u = s. Notice. are spin-independent as is evident from the factor δσσ . (9.68) Another single–particle operator is the one–particle density operator N ρ(r ) = ˆ˜ j=1 δ(r − rj ) . We will show that the boson and fermion creation and annihilation operators serve this purpose. t . ˆ 251 (9.62) in terms of the matrix elements of single–particle states (9. 9.9. g g (9.65) This symmetry will be exploited repeatedly below.69). 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 (9.64) We like to note an important symmetry of the matrix element r.69) ˆ In this case the generic operator is f (x) = δ(r − r). 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.k=1 j=k q2 |rj − rk | (9. ∂r 2 (9. 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) . s|ˆ|t.66) Here we adopt the notation for the Laplacian ∂2 ∂2 ∂2 ∂2 = + + . ∂r 2 ∂x2 ∂x2 ∂x2 1 2 3 (9.64) r. u = g d3 r ∗ ∗ d3 r ψr (r.66) and (9. ∂rj2 (9. s|ˆ|t.

. The ˆ matrix elements of the generic operator in terms of single–particle states are ˆ r. The generic operator is given through the Pauli matrices.3 2 2 (9. 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 + . α|ˆ3 |s.+ S2. The non-vanishing matrix elements of the spherical components of s are (α = + 1 .252 Many–Particle Systems which is also spin-independent.77) δσ σ ( δσσ − δ−σσ ) The operators (9. u. σ. In this case the generic operator is v (x. i. σ | s12 |t. 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 (9.3 S2. β = ˆ 2 2 1 −2) r.− . u.− S2.− + S1. . σ | t |t. not in the strict sense of our definition above since the restriction 2 j = k does not apply in the summation. Sj is the spin operator for particle j with components (in the spherical representation) Sj. However. Here. however. s. (9. β|ˆ− |s. Sj. σ . α|ˆ+ |s.74) and (9.74) =   N j=1 Sj  = 2 N Sj · Sk j. σ .e.73) is a one-particle operator.76) the single–particle state matrix elements of which are r. one can obviously extract the term j Sj to be left with a two–particle operator in the strict sense of our definition. x ) = q 2 /|r1 − r2 |. σ. Sj. β|ˆ3 |s. s = 1 σ. β s r. α s r.+ .+ + S1.77) are orbital-independent as evidenced by the factors δrs and δrt δsu in the respective formulas.3 . The operator N S = j=1 Sj (9. s.72) As a final example of one– and two–particle operators we consider operators of total spin.75) is a two–particle operator.k=1 (9. α s r. The generic operator is s12 = S1 · S2 = ˆ 1 1 S1. β s The operator S 2 = δrs = δrs = = + − 1 2 1 2 δrs δrs (9.

46). In order to determine these matrix elements one needs to evaluate first the matrix elements ˆ of the generic operators r|f |s r. r ΨF (N )|c† cs |ΨF (N ) .s=1 S r.s=1 ˆ r|f |s a† as bosons r ˆ r|f |s c† cs fermions r r. The matrix elements are then actually those of the number operator c† cr which is diagonal in the occupation number representation.t. u a† a† at au bosons g r s † † r.80) (9.u=1 S r.82) ΨF (N )|c† cs |ΨF (N ) = r cr 2 (9.u=1 (9.63. 9.9. r ΨB (N )|a† a† at au |ΨB (N ) r s ΨF (N )|c† c† ct cu |ΨF (N ) r s bosons fermions (9.62) are correct one needs to compare the result derived with conventional derivations of the matrix elements2 . 9.e. u as defined in (9. 9.62) through single–particle state matrix ˆ ˆ elements (9. The Matrix Elements ΨF (N )|c† cs |ΨF (N ) r We consider first the case r = s. (9.79) is opposite to that in the matrix element r.81) For the latter matrix elements simple rules can be derived from the algebraic properties of the boson and fermion operators (9. To show that the resulting values of the matrix elements (9.171 and pp.79) in the basis of many–particle states |ΨB.62).24) and (9.60) as follows N ˆ f (xj ) → j=1 N 1 2 1 2 S r.t. xk ) → ˆ j.3: One– and Two–Particle Operators Definition in Terms of Creation and Annihilation Operators 253 In order to express many–particle state matrix elements (9.176 where the matrix elements of fermion states can be found. respectively.63) and (9.. u .F (N ) have the same values as the respective matrix elements (9.s. . i.64) one replaces the operators F in (9.s.79) It is crucial to notice in the expression for the fermion two–particle operator that the order of the annihilation operators in (9. s|ˆ|t. and then in g a second step the matrix elements ΨB (N )|a† as |ΨB (N ) . namely cu ct .78) 1 2 g (xj . g and not ct cu .78) and (9. u cr cs cu ct g fermions (9. s|ˆ|t. 9. r S Ψ (N In case of r = s one obtains F )|c† cr |ΨF (N ) r = nr r=1 δnr nr .78.45.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’.83) t=1 t=r.79) have the following implication: The operators (9. s|ˆ|t. Expressions (9. s|ˆ|t. pp. These rules will be provided below only for the case of fermions.64).23.59) and G in (9. respectively.k=1 j=k S r.

t. Starting from the definition (9. 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.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) are included in the summation for which all indices are different. t. c2 = 0 r (9. s. Case 1 three of the four indices are different. Case 2 all four indices are different.d. g r s r. s.85) All other matrix elements vanish. (9.254 From (9.e.t. (r. The anti-commutation property (9.86) into three contributions.84) The off-diagonal elements with nr = nr except for r = s. in which case ns = ns + 1 and nt = nt − 1 holds.u r=s.’ (all indices different) that only tuples (r.i. 9.s.d.88) . r = t etc. Comparision with the results in Condon&Shortley (pp.s=1 a. u need to be considered. where each contribution corresponds to one of the three cases mentioned. ˆ The first contribution G0 is 1 ˆ G0 = 2 S r s|ˆ|r s c† c† cs cr + g r s r. 171) shows that the operator defined in (9. i. ˆ ˆ Nr Ns .. t.e. u in this sum three possibilities exist Case 0 two of the four indices are different.i. for which holds r = s.53) 1 ˆ G0 = 2 S r s|ˆ|r s − r s|ˆ|s r g g r. we split the sum in (9. those for which N and N differ in more than two occupation numbers. ˆ ˆ ˆ ˆ Accordingly. s.i. s.d.45) yields together with the definition of the occupation number operator (9.s=1 a.i. t). s).s=1 a.78) does yield the same results as the operator defined in (9.. i. 1 2 S r s|ˆ|s r c† c† cr cs .59) when the matrix elements are determined between Slater determinant wave functions. (9.87) Here and in the following we denote by ‘a. t=u 2 = 0. are ˆ ˆ ΨF (N )|F |ΨF (N ) = (1 − ns )(−1)qs nt (−1)qt s|f |t (9.84) we can conclude that for diagonal elements holds S Many–Particle Systems ˆ ΨF (N )|F |ΨF (N ) = r=1 ˆ nr r|f |r (9. namely. G = G0 + G1 + G2 . (r. t.d.86) For the combination of indices r.82.

s.90) ˆ Since Nr is diagonal. As we will find.t>u S r.s. One obtains 1 ˆ ΨF (N )|G|ΨF (N ) = 2 S nr ns r. 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. this contribution obviously behaves similarly to the off-diagonal part (9. 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 .88)..93) . s − r.91) ˆ G2 = r.s.89) S r.t. given in (9.u r<s.d. N = N or nr = nr for all r.t<u S r.s.e. 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.t.i.u r>s. i.d.t=1 a.t>u r s|ˆ|t u − r s|ˆ|u t g g c† c† cu ct . ˆ The second contribution G1 is ˆ G1 = + + + 1 2 1 2 1 2 1 2 S r.s.45).u r>s.t. Commutation of ct cu according to (9. S r. this contribution to the two–particle operator is diagonal in the occupation number ˆ representation.92) ˆ This contribution to G has non-vanishing matrix elements only when N differs from N in four ˆ occupation numbers nr .t=1 a. (9. ns .s. s|ˆ|s.t.t=1 a. nu . ˆ Case 0 This case covers diagonal matrix elements.86).s.d.i. s (9. this part accounts for all diagonal contributions to G.84) ˆ of the one–particle operator F . r g g .i.9. s|ˆ|r.i. r s (9.t>u S r. ˆ A similar series of transformations allows one to express G2 which can be written ˆ G2 = + + + as follows S 1 2 1 2 1 2 1 2 r.d. nt .u r<s.s.s r.s.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. g r s (9.d.t=1 a.t=1 a. contributes in this case. employing the symmetry property (9.t. S r. (9. A similar contribution does not arise for F . F (N )|G|ΨF (N ) .3: One– and Two–Particle Operators 255 Obviously.65). Only G0 . r s|ˆ|r t − r s|ˆ|t r g g ˆ c† ct Nr .i.u r>s.

94) Case 2 This case covers off-diagonal elements with nr = nr except for r = s.3.92) from (9. u g g (9.5: Derive (9. 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. respectively.4: 9.92). Only G2 . s|ˆ|u.90). u < v. All matrix elements which are not covered by the three cases above vanish.90) from (9. the ‘−’-sign applies for s < r.1: 9.. t for which holds ˆ ns = ns + 1 and nt = nt − 1. contributes in this case.256 Many–Particle Systems Case 1 This case covers off-diagonal elements with nr = nr except for r = s. ˆ ˆ Show that the matrix elements of F and of G conserve particle number. v − r. i. particle number is conserved.2: 9. u > v or s > r. One obtains ˆ ΨF (N )|G|ΨF (N ) = (−1)qs +qt (1 − ns )nt r r=s. nt = nt + 1. s|ˆ|r. it holds S nr = S nr . nu = nu − 1. Only G1 . u > v. nv = nv − 1.96) . contributes in this case.3.75) which are.92). Derive the non-vanishing matrix elements (9. v for which holds ˆ ns = ns + 1.3: 9. r g g (9. given in (9. In particular.95) In the latter formula the ‘+’-sign applies for s < r. t − r.89). for ˆ non-vanishing matrix elements ΨF (N )|G|ΨF (N ) at most four occupation numbers nr and nr can differ. Spin Operators ˆ ˆ In this section we will briefly consider the spin operators S given in (9.62) for bosons.3. One obtains ˆ ΨF (N )|G|ΨF (N ) = ± (−1)qs +qr +qu +qv (1 − ns )(1 − nr )nu nv r.t nr r. given in (9. Derive (9. We like to express these operators through ˆ fermion creation and annihilation operators.3.3. Derive (9. one–particle and two–particle operators. Furthermore. t. u < v or s > r. s|ˆ|t.74) and (9.73) and S 2 given in (9.76). u. The matrix elements of the corresponding generic operators are (9. r=1 r=1 Exercise Exercise Exercise Exercise Exercise 9.e. s|ˆ|v.91).95) from (9.

superconductivity. v c† c† cu ct . the systems investigated involve a macroscopic number of components such that statistical mechanical approaches are envoked.g. In the latter case the indices r.g. 9. the study of problems governed by Hamiltonians of the type (9.97). 9. t. v r s 2 r.98) In many cases the Hamiltonian is spin-independent and the equivalent notation S H = r.97) is over the orbital states.t.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. e.99) has preoccupied an important part of all intellectual efforts in Theoretical Physics during the past fifty years.s=1 r. The great generality of the concepts developed in the context of many–particle systems can be judged from the fact .s.6: Derive (9. e. 9. the two–particle operator makes the description of many–particle systems a very hard problem. that the two–particle operator representing interactions between particles induces a variety of interesting phenomena. 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.96) and (9.t.9. The outcome of these studies is that even when interactions between components (particles) are simple.97) The summation in (9.98. but also ordinary phenomena like freezing and evaporation.u=1 σ.96. the concerted behaviour of interacting systems can deviate qualitatively from that of systems made up of independent components. which represents the interactions between particles. s. u refer only to the orbital part of the single–particle basis. Examples is the laser action. 2 2 Exercise 9. evaluation of their stationary states and excitation energies. molecules.σ (9. Unfortunately. Actually. If it were not for the two–particle contribution to the Hamiltonian. In fact. the spin part of the single–particle states is represented by α = 1 and β = − 1 . Often. the description of many–particle systems..99) will be used. 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. The fortunate side of this is. u c† c† cuσ ctσ v rσ sσ r. s|ˆ|u.u S (9.s.s=1 σ 1 ˆ rσ r|t|s c† csσ + 2 S r. would be a simple exercise in linear algebra. in crystals. however.3. atoms and nuclei. s|ˆ|t.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..s=1 c† c† crβ csα . rα sβ (9.

100) ˆ ˆ We will denote the matrix elements of t as r|t|s = trs . (9.e. We will restrict our description in the following to systems composed of an even number (2N ) of particles and to spin-independent interactions. {|m). What might happen? Independent-Particle Hamiltonian A many–fermion system governed by an independent-particle Hamiltonian.1: Imagine that in a closed. a Hamiltonian without a two–particle operator contribution accounting for interactions among the particles. i. electrons and nucleons..s=1 σ ˆ rσ r|t|s c† csσ (9. The Aufbau principle. 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). water-filled jar all electrons of water turn their fermion nature into a boson nature. If it were’nt for the fermion nature. according to which nuclei and atoms change their qualitative properties when they grow larger. namely of neurons3 . Amit (Cambridge University Press. m = 1. . The Hamiltonian of such system is S Ho = r. In such a representation Ho is diagonal and the stationary states can be stated readily.. Before continuing one may finally point out that the material world as we know it and as it establishes the varieties of natural substances. i. . Exercise 9.100) in a form in which the factors c† csσ become rσ occupation number operators. 1989) . ultimately depend in a crucial way on the fermion character of its constituent building blocks. Transformation to a New Set of Creation and Annihilation Operators Our aim is to represent the Hamiltonian (9.J. 2.103) An account of some of these efforts can be found in Modelling Brain Function – The World of Attractor Neural Networks by D. ranging from nuclei to the molecules of living systems . another prototypical systems of many strongly interacting components.102) U ∗rm Urn = δmn . The electronic properties of atoms with different numbers of electrons would differ little. S}. can be described in a rather straightforward way. where S |m) = r=1 Urm |r . would not exist. all material systems would condensate into states which would depend little on particle number.e. . Chemistry essentially would know only one element and Life would not exist. New York.4.101) The unitary property of Urn U† U = U U† = 1 1 or S S (9. r=1 3 Urm U ∗sm = δrs m=1 (9.258 Many–Particle Systems that today these concepts are providing insight into the functioning of biological brains.

4: Independent-Particle Models allows one to express conversely |r in terms of |m) S 259 |r = m=1 ∗ Urm |m) . of N -Fermion states in which single particle states |nj ) as defined in (9.110) mσ nσ [dmσ . i. .107) Urn c† rσ . nσ We demonstrate here the anti-commutation property (9.104) One can.e. of (9.46) as c† and crσ .108) obey the same anti-commutation properties (9.s.111). (9. the operators d† and dnσ .108) and (9.111) can be written. (9.s r The unitarity property r ∗ Urm Urn = δmn yields immediately identity (??). [d† .46). 9.111) and.103) allows one to express [c.109) The operators (9. the anti-commutation properties (9.106) which together with (9. d† ]+ = δmn δσσ nσ (9.45.105) where ˜ ˆ tmn = (m|t|n) = S ∗ trs Urm Usn = r. c† ]+ = δσσ [dmσ . accordingly.f. The unitary property (9. using (9.110) can be demonstrated similarly. The l.and two-particle operators hold in an analogous way for d† and dnσ . hence. csσ = n=1 Usn dnσ .104)] S S ∗ Urm d† mσ m=1 c† rσ = . they are Fermion creation and annihilation operators.h.113) Urm Usn [crσ . (9. [dmσ .101) of states |r . In a basis of states N d† j σj |0 n j=1 (9. the expressions drived above for the matrix elements of one.s=1 ˆ U † tU mn (9.112) . d† ]+ = 0 (9. rσ namely. ∗ ∗ Urm Urn .101) are occupied. expand S trs = mn ∗ ˜ tmn Urm Usn (9. for nσ rσ example.9.108) These operators describe the creation and annihilation operators of fermions in states |n) which are linear combinations (9.. (9. behave exactly like the operators c† and crσ behave for states |ΨF (N ) .100) yields ˜ Ho = tmn d† dnσ mσ where d† nσ = r=1 S S (9. dnσ ]+ = 0 . d† ]+ = sσ nσ r. dnσ = r=1 ∗ Urn crσ (9.

n = 1. the associated ( properly normalized) eigenvectors Vn are the columns of (Urn ). 2 2 2 2 i.106) diagonal. σj ) = (nk .. 2.114) Ho = n=1 † n dnσ dnσ (9. The condition (9. the so-called ground state.. . .106) and the unitary property of Urn .115) ˜ and involves solely occupation number operators N nσ = d† dnσ . i. It is. n2N . j j (9. (9.57) to the case of 2N particles.116) which follows from (9. We apply for this purpose (9. 2.106) is equivalent to S trs Usn = s=1 n Urn ∀r. σ2N . USn ). . i. σ1 . S are T real.118) Ground State In case of an ordering m < n for m < n the state of lowest energy. ˜ tmn = In this case Ho has the desired form S n δmn .114) together with (9. n2 . .e. transformation matrices Urn and energy values n . U2n . (9. ) = j=1 nj .115) with eigenvalues 2N E(n1 . Before proceeding we like to address the issue how the new representation.e. is obtained. . . .. . σ2 . r = 1. i. Vn = (U1n . 1 as well as a fermion in a spin state | 1 . The 2N fermion state 2N d† j σj |0 n j=1 where (nj . a fermion in a spin state | 1 . The corresponding eigenvalues n . . . .. A non-degenerate orbital state which is occupied by two fermions.260 Diagonal Representation Many–Particle Systems The transformation (9. 1 and |β = | 1 . S (9. .e. hence.117) are eigenstates of (9. is N |Φo = j=1 d† α d† β |0 .119) In this state the N lowest orbital eigenstates of trs are occupied each by an electron with spin |α = | 1 . a simple matter to nσ state many–particle states in the new representation. − 1 is called a closed 2 2 2 2 .e. σk ) for j = k (9. This equation poses the eigenvalue problem for the hermitian matrix (trs ).101) gives us the freedom to choose the matrix (9. Eigenstates of (9. − 1 .115) can be stated immediately since any many–particle wave function in the occupation number representation is suitable.

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

one with eigenvalue zero (singlet) and one with eigenvalue ‘2’ (triplet). . α and |β. α ± |β. . |β. β are not eigenstates of S 2 . β qualify as eigenstates. c† +1 σ = c† σ 1 2N (9. We identify the states |2N + 1 = |1 to avoid cumbersome summation limits.σ = (9.122) is correct. 2N which are located on a ring. the linear combinations 1/2(|α. The system is assumed to posses a 2N –fold symmetry axis and interactions connect only orbital states |r → |r ± 1 .126) which follows from the occurence of squares of fermion operators which. 2. mα nβ (9. do not need to be considered since they ˆ give vanishing contributions.262 and ˆ Σ3 = − m. . also eigenstates of Σ2 . β . 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 . . β and |β.4. for |α.e. ˆ An eigenvalue ‘2’ of the spin operator S 2 identifies the states |α. ˆ 3 on the two states produces . The action of Σ N +1 ˆ Σ3 |α. β =− m.n=N Many–Particle Systems N +1 d† d† dmβ dnα . . ˆ The remaining states |α. (9. . 2N ) drσ = √ 1 2N 2N e−irnπ/N cnσ n=1 (9. r = 1. for example.n=N d† d† dmβ dnα d† +1α dN β |Φo = 0 mα nβ N (9. . however.σ d† +1. of course.2: Construct four operators O ˆ O ˆ such that O ˆ S3 .129) The symmetry of the Hamiltonian suggests to choose a new representation defined through the operators (r = 1. i. conclude that (9. m |Φo m m m γσ. The two states |α. One can.128) The cyclic property of the system is expressed through c2N +1 σ = c1 σ .127) ˆ are triplet and singlet states. etc. α as triplet states. . hence. of course.100) outlined above we consider a system of 2N electrons which move in a set of atomic orbitals |r . vanish. . β both with eigenvalues 1. |N + 3 . α . α are.σ dN σ N σ. The same result holds for |β.125) ˆ Contributions to S 2 acting on states |N + 2 .. 2. ˆ Exercise 9.130) . 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.

One can then express[.109)] ¸ cnσ = √ 1 2N 2N eirnπ/N drσ r=1 2N (9.s. eisπ/N 1 − e2N isπ/N 1 − eisπ/N (9. . 4N.9.130.4: Independent-Particle Models d† = √ rσ 1 2N 2N 263 eirnπ/N c† nσ n=1 √ 1 eirnπ/N . (9. rσ (9. thereby.135) c† nσ 1 = √ 2N e−irnπ/N d† rσ r=1 (9. .133) Application to the l. . indeed.139) −2t cos r=1 σ rπ N d† drσ .132) an identity which can be derived employing the well-known expression for a finite geometric series m as = a s=1 1 − am . 9.h. .f.138) ˜ trs = from which follows ˆ Ho = eirπ/N + e−irπ/N 2N δrs (9. n=1 (9.132) one can conclude 2N ei(s−r)nπ/N + eisπ/N n=1 1 2N 2N ei(r−s)nπ/N . 1−a (9.140) The Hamiltonian is now diagonal and the construction outlined above can be applied.134) which. vanishes for integer s. (9. of (9. one obtains the new representation of Ho 2N ˆ Ho = −t r.137) where 1 ˜ trs = e−irπ/N 2N Using (9.131) The (9.131) 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 = 0.136) ˆ and.s=1 σ ˜ rσ trs d† dsσ (9. . 2N (9. 2N.

does not spoil it. those with energy above that of the (energetically) highest occupied single-particle state. i. 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 . Of course. There are two reasons why this is so: first.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. 9. and that independent-particle behaviour surfaces mainly because it applies well to the ground state. 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. the everpresent Coulomb repulsion between electrons. if one places fermions into an independent-particle ground state of the type (9. rather they assume that each particle experiences the average interaction due to the fermions frozen into the ground state. i. These states do not altogether neglect the pair interactions. A similar principle accounts approximately for properties of nuclei as described by the nuclear shell model. i. assuming that electrons move independently from each other in so-called bands.4.5 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.e. 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. 2. i. . Hence. .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 . We will present in this and the following section a method which constructs such optimal single-fermion states. in states which are constructed according to the ‘Aufbauprinzip’ in a manner which neglects the mutual repulsion between electrons. .. ... n = 1. valence bands or conduction bands.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. Many electronic properties of solids can be understood in a similar way.264 Many–Particle Systems Exercise 9.g. the fermion nature restricts the possibility for perturbations on the system. 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. Since independent-particle behaviour is so universal one may wonder why the presence of particleparticle interactions.. 2N with interactions between neighbouring (n. e.e.. one expects that mean-field descriptions should be rather sufficient. The slow decay of the interaction makes the motion of any one electron rather independent of the exact position of most other electrons.e. The regularities of the elements can be rationalized in terms of electrons moving in hydrogen atom-type orbitals. in particular. if one can choose the singleparticle states such that the residual perturbations are small. (9.e..e. i.

. .t. Mean-Field Potential The mean field for the reference state |Φo (U) is defined in a straightforward way by averaging the two-particle contribution to (9. U = ( Urm ) as defined in (9. . is not unitary. n.146) . . The states |m are connected with the states |r through (m = 1. S on which (9.143) with r. .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. p. .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.9. s|ˆ|t. ˜ (9. We will adopt a second set of single-particle states {|m .s.144) defined as in (9. . S) ˜ S S |m = ˜ r=1 Urm |r . |r = m=1 ∗ Urm |m . If one restricts m = 1.142) r|t|s c† csσ + rσ sσ 2 r. q. r = 1. S is a unitary matrix. s. u refer to the initial basis of single-particle states |r .142) is based.u=1 σ. naturally. u c† c† cuσ ctσ v (9. 2. m = 1. . This is done as follows: ˆ Vmf (|Φo (U) ) = 1 2 S r. S} the elements of which will be labelled by indices ˜ m. m = 1. s|ˆ|t.145). . .s=1 σ r. 2. N . . the corresponding U = ( Urm ) forms an S × N matrix and. t.108). 2. . . 2. u v r. In our formulas below we will adopt strictly the convention that indices r.142) over this reference state to turn the contribution into an effective one-particle operator. .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.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. . j j (9.s. . 2.u=1 σ. as in in (9.σ and derive a single-particle operator which approximates this Hamiltonian.t. The states |m serve to define a reference state of the system ˜ N |Φo (U) = j=1 d† α d† β |0 .

s|ˆ|m.σ S ctσ + m. d† d† mσ nσ dmσ dnσ d† dnσ mσ = = = = and.m. u ˜ ˜v m. 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.n. n = ˜ ˆ ˜ s. u c† ˜ vn sσ m. n c† c† v ˜ ˜ rσ sσ r.148) one obtains. s| v t.u ∗ r.146) accordingly.t.u=1 σ.m. m.σ S d† d† mσ nσ cuσ ctσ + r.149) The remaining matrix elements appearing in (9.n=1 σ.148) r.n. m| v |t. by means of the rules (9.σ S dmσ dnσ + r.150) r. Many–Particle Systems (9. hence.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.σ S 1 − 2 where.266 Here ··· denotes the average O = Φo (U)| O |Φo (U) .t=1 σ. ˆ (9.t=1 σ. m|ˆ|t. u Usm Uun . n|ˆ|t.σ S d† dnσ cuσ mσ − r. m|ˆ|˜ . m|ˆ|t.s.s.n=1 σ. for example. s|ˆ|t.σ S d† dnσ mσ ctσ (9.s.u=1 σ.n.148) are obtained in a similar way by transforming only two of the four single-particle states into the new representation.n. s|ˆ|˜ .84).t. u c† d† dnσ cuσ ˜ vn rσ mσ r.σ . 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.u=1 σ. n c† ctσ δσ σ ˜ v ˜ rσ m≤N r. n c† d† dnσ ˜ v ˜ rσ mσ r. n c† ˜ v ˜ sσ m. For the averages ··· in (9.m.

152) can then be written in the form ˆ Vmf (|Φo (U) ) = r| vmf (|Φo (U) ) |s c† csσ ˆ (9.u=1 σ S − m.m.σ S r. The mean field approximation. as introduced here.. n c† ctσ .σ 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.S.t=1 σ.154) ˆ The mean field approximation replaces then the Hamiltonian (9. At this point this state is completely arbitrary.5: Self-Consistent Field Theory 1 2 1 2 1 2 S 267 m..t=1 σ (9.149) for the matrix elements in the mixed representation one can state the right hand side explicitly in terms of Urm .u=1 σ. m=1. s|ˆ|t.. m|ˆ|˜ .151) Carrying out the sum σ δσσ = 2 leads to a factor 2 for the first two terms.9.s σ where r| vmf (|Φo (U) ) |s = ˆ t. We will address now a proper choice of the reference state.2.. r| v |u.142) by Hmf (U) defined through S ˆ Hmf (U) = r. s| v |t.152) By means of expression (9. m|ˆ|t.156) which defines the reference state |Φo (U) .2.N (9. t| v |u. n c† ctσ δσ σ ˜ v ˜ rσ − r. The one-particle operator (9.u=1 σ.155) which is a function of the S × N –matrix U = ( Urm )r=1.153) rσ r. Exploiting the symmetry r. leaves open the question how the reference state should be chosen. u − r. s|ˆ|˜ . u c† cuσ δσσ ˜ vn sσ m≤N s. s )  ˆ ˆ  m≤N ∗ Utm Uum   (9.. ˜ v ˜ sσ m≤N s.s=1 σ ˆ r|t|s + r| vmf (|Φo (U) ) |s ˆ c† csσ rσ (9.m. .σ S + − r. u c† cuσ ˜ vn rσ (9. m|ˆ|˜ .. u = s.u ( 2 r. u c† cuσ ˜ vn rσ m≤N r. t| v |s.

. Let us state now the construction of the self-consistent field ground state in more detail. 2. can achieve this goal only iteratively. We will argue below that the self-consistent field ground state. .158) where the labels m are ordered to obey the condition (1) m < (1) n for m < n . Step 2M. steps (2) and (3) are being repeated M times until the procedure converges. (9. . the reference state resulting from step 2M + 1 (within numerical error) is equal to the reference state resulting from step 2M − 1. SCF-Algorithm. 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. M = 1. . S . SCF-Algorithm. 2. 2. (9. . The procedure determines the reference state as the ground state of the independent-particle Hamiltonian (9. m = 1. the independent-particle nature stemming from the fact that the functional form of the ground state.145).s=1 σ ˆ r|hmf |s c† csσ rσ (M ) (9. often also referred to as the Hartree-Fock approximation. S . .s=1 σ S ˆ r|t|s + r| vmf |Φo (U(M ) ) ˆ |s c† csσ rσ = r.6 Self-Consistent Field Algorithm The Self-Consistent Field (SCF) approximation. . i.. r = 1. is exact only for an independent-particle Hamiltonian. and obtaining in a step (3) the ground state of this one-particle Hamiltonian according to the construction in Section 9. is the lowest energy independent-particle ground state one can construct.155) following the method outlined in Section 9.4 and defining this as the new reference state.4. r = 1.119. m = 1. This procedure.157) and determines the associated diagonal representation defined through S (1) ˆ r|t|s Usm = s=1 (1) (1) m Urm . .160) .142). (9.s=1 σ S ˆ rσ r|t|s c† csσ (9. S.268 Many–Particle Systems 9.145). however. Step 1: Choosing an Initial Reference State One defines the Hamiltonian ˆ (1) Hmf = r. i. . 9.e. 2. 2.159) One defines the reference state |Φo (U(1) ) using the definition (9. . .e. .155). . The state thus determined is referred to as the self-consistent independent-particle ground state. under conditions which often are realized. assuming in an intial step (1) a properly chosen reference state. .. . i. S .: Determine Mean Field Hamiltonian In this step the mean field Hamiltonian S ˆ (M ) Hmf = r. U(1) is the S ×N –matrix (1) of the first N column vectors of Urm .e. determining in a step (2) the corresponding mean field Hamiltonian (9.

S. ˆ (b) How can the expectation values of the operator Prr be interpreted.145). . . S . (c) Show that any spin-independent one-particle operator S ˆ F = r. S . 2. m = 1. 2.e. . 2 .6: Self-Consistent Field Algorithm is evaluated. SCF-Algorithm.s=1 (9. . ˆ Exercise 9.s=1 σ ˆ r|f |s c† csσ rσ (9. 2.145). 2. Step 2M + 1. 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. M = 1. . t| v |s.: 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 . t| v |u.163) The result yields the reference state |Φo (U(M +1) ) using the definition (9.167) . . (9. u − r. U(M +1) is the (M +1) S × N –matrix of the first N column vectors of Urm . . s ) ˆ ˆ m=1 Utm ∗(M ) (M Uum ) (9. r = 1. indicating that the state converged.165) can be written ˆ F = S ˆ ˆ r|f |s Prs r. . the S × S–matrix S (M ) r|hmf |s N 269 ˆ = r|t|s + t. i.166) ˆ (d) Define a similar two-particle operator Prstu . m = 1. (e) Consider the creation operators N † gqσ = m=1 Vmq d† mσ (9. step 2M + 2 is carried out.9. ˆ t. etc. SCF-Algorithm: Continuation and Convergence Condition When step 2M + 1 is completed.u=1 ( 2 r. S do not exceed a preset threshold. . σ p = δtr δsu δσσ . 2. (9. .1: Let Prs be the one-particle operator with the generic operator prs = |r s|.164) ˆ (a) Determine the expectation value of Prs for the state as defined in (9. . σ|ˆrs |u. . r = 1. i. s = 1. .. . .162) is solved adopting an ordering of the labels m which obeys the condition (M +1) m < (M +1) n for m < n .161) is calculated. .e. i. S (9.6. .e..

Note that this is not an mσ S × S–matrix! Show that the reference state defined through N † † gqα gqβ |0 q=1 (9.163).171) where the convention (9.162.. For this representation holds N m| t |˜ + ˜ ˆn m 2 mm | v |˜ m ˜ ˜ ˆ n˜ − mm | v |m n ˜˜ ˆ ˜ ˜ = (SCF ) δmn m (9. (9. .e.172) is obeyed.170) (SCF ) Here |r . Equation (9.270 Many–Particle Systems which are connected with d† through a unitary N × N –matrix (Vqm ). We denote the representation which results from the SCF algorithm after its convergence as follows: S |m = ˜ r=1 (SCF Urm ) |r . . (SCF ) m < (SCF ) n for m < n (9.σ + 2v ( c† c1α c† c1β + c† c2α c† c2β ) + v 1α 2α 1β 2β + c† c2σ c† 2σ 1σ c1σ ) . . i.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.163). (f) Can one distinguish the states (9.171) implies that for the mean field Hamiltonian holds S ˆ Hmf (U(SCF ) ) = m=1 σ (SCF ) † dmσ dmσ m .168) ˆ ˆ has the same expectation values for Prs and Prstu as the reference state (9. S denotes the initial single-particle states and Urm is the unitary S × S– transformation matrix obtained in (9. r = 1.6. We begin by summarizing the result of the SCF algorithm.168) through a physical observation? Exercise 9. (9.9. (9.145) and (9.169) 9.145).173) . 2.7 Properties of the SCF Ground State We want to investigate now the properties of the SCF ground state.

2 m=1 m|t|m + ˜ ˆ˜ m. r.162) is non-linear in Urm . Within the mean field approximation as described by (9. mα mβ (9. We like to determine next the energy expectation value of SCF . ˆ SCF H SCF . the mean-field Hamiltonian is diagonal in the SCF representation.176) 2 mm | v |mm ˜˜ ˆ ˜˜ − mm | v |m m ˜˜ ˆ ˜ ˜ .7: Properties of the SCF Ground State 271 i. exploiting the spinindependence of (9. Finally. the mean field potential ˆ Vmf (|Φo (U) ) defined in (9. and once as a probe particle being subject to the mean pair interaction.174) The first property we consider is the total spin character of SCF . Since this state is composed only of closed shells one can conclude following Section 9. Exercise 9.142). Employing expressions (9. .171) yields ˆ SCF H SCF N = N (9. v (SCF ) Show that the resulting eigenvalue problem of the type (9. This over-counting leads to the correction term in (9.93) for the expectation values (diagonal elements) of the one-particle and two-particle part of the Hamiltonian (9.177) 2 mm | v |mm ˜˜ ˆ ˜˜ − mm | v |m m ˜˜ ˆ ˜ ˜ .142).177). (9. 2 m=1 (SCF ) m − m.173) holds N ˆ SCF Hmf (U(SCF ) ) SCF = 2 r=1 (SCF ) m . once as a member of the ground state. one can also determine the expectation value of SCF for the Hamiltonian (9.e.2: Derive (9.m =1 Comparision with (9.1: Reformulate (9.177).84) and (9.4 that SCF is a total singlet state.. the SCF ground state is N SCF = m=1 d† d† |0 .171) in terms of matrix elements r|t|s . s|ˆ|t. (SCF ) ˆ Exercise 9. u and Urm .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.175) However.e.7. one obtains ˆ SCF H SCF N = N (9.. Since the system under consideration has only 2N particles altogether.7.146) counts a particle twice. i.9.142).

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. t describes the coupling of state |r to states |r ± 1 at the two neighboring lattice sites. as stated through Hamiltonian (9. . of (9.h.t.272 Many–Particle Systems 9. u c† c† cuσ ctσ . reads 1 ˆ r.178).σ rσ The potential energy term. s|ˆ|u. We consider as an example the Hubbard model of an infinite linear lattice and apply the model to describe magnetic instabilities in metals. In the present Section we want to adapt the algorithm to systems containing a macroscopically large number of fermions.h. is characterized through the parameters t.s.178) and through the so-called filling factor n n= 2N N = .s. i. for example. . Both N0 and N are macroscopically large numbers. referring to the fact that in this case the band of single particle energies (9. cannot be solved exactly. rσ rσ r σσ (9. t = v δru δst δrs .178) v contributes only in case that two electrons occupy the same lattice site. r ∈ Z. s|ˆ|t..σ crσ + crσ cr+1. β for ‘up’ and ‘down’ rσ spins. except in case of one-dimensional systems. The first term on the r.4–9.181) The Hubbard model.7 for finite systems.180) V = rσ sσ 2 r. is identical to that of the independent–particle Hamiltonian H0 (9. t is assumed to be a positive.182) n can assume values 0 ≤ n ≤ 2. in the usual two–particle operator form. i. In spite of the simplicity of its Hamiltonian the Hubbard model.179) r+1.u σσ where r. The case n = 1 is termed the half filled band case.e.σ + v 2 c† c† crσ crσ . v represents the ‘on-site’ Coulomb repulsion.178). Due to the lack of an exact solution in dimensions higher than one.. . † ˆ Ho = −t c† .e.128). The operator c† (crσ ) creates (destroys) an electron with spin σ (σ = α. v of the Hubbard Hamiltonian (9. describes the sites of a linear lattice.s. labeled r = −N0 + 1. v (9. S N0 (9.8 Mean Field Theory for Macroscopic Systems The SCF algorithm has been developed in Sections 9.σ crσ + crσ cr+1. i. (9. in the two-dimensional copper oxide lattices of high temperature superconductors. N0 which are populated by 2N electrons (we choose an even number of electrons for convenience). .. of (9. .e. real number. respectively) in state |r at lattice site r.178) Here the index r.178). The one-dimensional Hubbard model is described by the Hamiltonian ˆ H = −t rσ † c† r+1. According to (9.192) derived below is filled half. We assume that there are altogether S = 2N0 lattice sites. approximation schemes like the self-consistent field approximation play an important role. v (9. the second term on ther.

183) where the operators d† are related to the original creation operators c† through the unitary mσ rσ ˆ transformation U defined in (9.184) and using (9. Inserting (9. boundary effects are assumed to be negligible. 2N0 (9.186) where m = −2t cos mπ . by N −1 ||SCF = m=0 d† d† |0 . The unitary transformation which diagonalizes H0 .188) . < ±(N0 −1) < N0 holds as can be readily verified. Accordingly. Using (9.6.185) into (9. rσ rσ (9. This task has ˆ been solved already on page 262. Accordingly.8: Macroscopic Systems 273 Since we are dealing presently with a macroscopic system.e.181. 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..155.186) one obtains for the mean field Hamiltonian N0 ˆ (2) Hmf = m=−N0 +1 σ m d† dmσ + v mσ N 2N0 c† crσ . SCF Ground State for the Hubbard Model We seek to determine the ground state of the Hubbard model stated by (9. one obtains for the mean field Hamiltonian N −1 ˆ ˆ Hmf (U) = H0 + v r.178) can be determined by applying (9. .144). rσ (9. the operator H0 can be re-written in the diagonal form N0 ˆ H0 = m=−N0 +1 σ m d† dmσ mσ (9.140).154). The self-consistent field Hamiltonian Hmf which corresponds to the Hubbard Hamiltonian (9.174). i. 9. N0 (9. we identify |r + N0 = |r and later let N0 go to infinity.179). .185) ˆ According to (9. we assume that the SCF ground state is given by (9.σ m=0 ∗ Urm Urm c† crσ .187) The energy levels are labeled such that 0 < ±1 < . We can now embark on step 2 of the SCF algorithm. mα mβ (9.184) ˆ We are now ready to apply step 1 of the SCF algorithm and diagonalize term H0 .9. For this purpose we apply the SCF theory presented in Section 9. is Urm = √ 1 exp (irmπ/N0 ) .178) within the framework of the self-consistent field theory. 9. in fact.

π]. 4 So far we have not exploited the fact that N0 and N are macroscopically large quantities. The energy Emf of this state can be calculated readily by using (9. The ground state of the system. When N0 becomes macroscopically large the single particle discrete energy levels (9. 9. the self–consistency condition is exactly met and the SCF algorithm converged after two steps.190) We have used in the latter expression the definition (9. called the Fermi energy. hence. as given in (9. i..178).1).187). according to the Pauli principle.182)].191) where m is given in (9. the results of different formulations agree qualitatively.187) form a quasi1 2 continuous energy band. N0 ˆ (2) ˆ Hmf = Hmf = m=−N0 +1 σ m + vn † dmσ dmσ 2 (9. Indeed. there exist several ways of formulating mean field approximations.190). Indeed. denoted by F (see Fig. Usually. Alternative Description of the Mean Field Approximation The state (9.182) of the filling factor n [c. the ground state of the Hubbard model in the self-consistent field approximation can be determined exactly. ˆ The Hamiltonian Hmf . but may differ quantitatively. In other words. 9. can be obtained. Apparently. (9.f. in the limit N0 → ∞. crσ through d† . dnσ according to (9.185) one can prove rσ mσ N0 c† crσ = rσ rσ m=−N0 +1 σ d† dmσ . π].183) is not the only candidate for the mean field ground state of the Hubbard model.177) and mm |ˆ|mm = mm |ˆ|m m = ˜˜ v ˜˜ ˜˜ v ˜ ˜ which holds in the present case.e. −π ≤k ≤π . namely. To demonstrate this we consider an alternative formulation of the mean field approximation for Hamiltonian (9.1.. .109) and using (9.274 Many–Particle Systems Expressing c† . holds k ≡ km ∈] − π. the electrons in the present description behave like independent particles with energies ¯m = m + vn . the state with the lowest possible energy. (9. for km = mπ/N0 holds km ∈ −π + N0 .189) and. even for one and the same Hamiltonian.e. by filling up this energy band with electrons from the bottom of the band (which corresponds to k = 0) to a maximum energy.192) This dispersion law is presented in Fig. i. . One can see that −2t ≤ (k) ≤ 2t holds. mσ (9. is already diagonal. . 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. through the so-called dispersion law (k) = −2t cos k . π and. The energy m can be expressed as a function of a continuous variable k ∈] − π. The bottom of the energy band corresponds to (0) = −2t and the width of the energy band is 4t. . −π + N0 .

45. i.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 . First.9. the deviation of Nrσ from its mean value Nrσ for the corresponding reference state..178) and performing the summation over the spin indices results in ˆ ˆ H = H0 + v r ˆ ˆ Nrα Nrβ . where Nrσ is the mean occupation number of the one particle state |rσ . The anti-commutation properties (9.193) where. in the last term on the right hand side.8: Macroscopic Systems 275 Figure 9. The mean field approximation neglects the . 9. the operator Nrσ can be written ˆ Nrσ = Nrσ + δ (Nrσ ) .e.195) ˆ ˆ ˆ and where δ (Nrσ ) = Nrσ − Nrσ describes the fluctuations of Nrσ .183). Nrσ = SCF ||Nrσ ||SCF .. say the one given in (9.178) in terms of the occupation number operaˆ tors Nrσ = c† crσ .e.1: The dispersion law (k) for independent electrons on an infinite one dimensional lattice.193) into (9. we have employed (9. rσ rσ rσ (9.194) ˆ For a given reference state. let us express the interaction term in (9. (9.196) (9.56). F = (±kF ) denotes the Fermi energy. (9. i. Inserting (9.

consequently. 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 . yields essentially the same mean field Hamiltonian (9.198) Let us assume that the ground state of the Hubbard model.m ∗ Urm Urm d† dm σ mσ = m=0 ∗ Urm Urm = n N = . i. Spin-Polarized Mean Field Ground State ˆ We want to consider now a ground state for Hmf defined through Nrσ = nσ = const (9. Therefore. In case nα = nβ the following construction will lead to a ground state which coincides with the non-magnetic ground state (9.183). m+ 2 2 c† crσ − 2N0 rσ (9. in the mean field approximation. .200) allowing. is given by (9. 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 (9.146). (9. however.198) as the one obtained by using the mean field potential (9. Note that the procedure employed in (9. 2N0 2 holds ˆ Hmf = = mσ mv ˆ H0 + 2 vn2 2 rσ nv † vn dmσ dmσ − N . For such state the mean occupation number nrσ is uniform at all lattice sites.183) . that nα and nβ assume different values. but the mean number of particles with spin α can be different from the mean number of particles with spin β.194) by dropping all the terms which contain the fluctuations to second order.197) and the corresponding mean field Hamiltonian becomes ˆ ˆ Hmf = H0 + v r ˆ ˆ Nrα Nrβ + Nrβ Nrα − v r Nrα Nrβ . the ground state of the system can have a non-zero local spin and. Because of Nrσ = c† crσ rσ N −1 = m. separating the occupation number operator into a mean value plus fluctuation and neglecting the fluctuations in 2nd order. non-vanishing magnetization.e.191) for the ground state energy.276 Many–Particle Systems fluctuations to second order and the mean field Hamiltonian is obtained from (9.199) which is identical to expression (9.195)..

In the limit N0 → ∞ this sum can be expressed as an integral. In fact.9. According to this method one minimizes ˜ Emf = Emf + µ (n − nα − nβ ) .206) ˜mf is minimized where µ is the Lagrangian multiplier considered an independent variable. 2π (9. (9..206).192). as is shown schematically in Figure 9. At this point we exploit the fact that our system is macroscopically large.202) This Hamiltonian describes a system of non-interacting electrons with a spin-dependent dispersion law (9. For this purpose the sum in (9.205) is met. N0 ∼ 1023 . one can expect the system to lower its energy by assuming nα = nβ and. (9. (9. Since β (k) − α (k) = v(nα − nβ ). 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.8: Macroscopic Systems from which we obtain nσ = The mean field Hamiltonian is then ˆ Hmf = m 277 1 2N0 Nmσ . a spin–polarized ground state. subject to the constraint n = nα + nβ . Therefore. The actual values of nσ (σ = α. nβ and µ. In this limit the discrete energy spectrum (9. The values of nα and nβ are determined by minimizing the energy density (9. such that the larger nσ .201) ( m ˆ + vnβ ) Nmα + ( m ˆ + vnα ) Nmβ − 2N0 vnα nβ . i.206) needs to be evaluated. β) are determined by minimizing the energy density of the ground state Hmf 1 = [ mσ Nmσ ] − vnα nβ (9. The additional term in (9.2a. E with respect to nα .203) is provided by the continuous function σ (k) = (k) + vn−σ . 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) . if nα > nβ then β (k) > α (k). 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 ).205) Such minimization is realized through the method of Lagrangian multipliers.207) where (k) is given by (9.204) Emf ≡ 2N0 2N0 m. the smaller is σ (k) for a given k..203) mσ = m + vn−σ . (9.g. 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.e..σ with respect to nα and nβ .208) . hence.206) ascertains that condition (9. e. i. m (9. the so-called Fermi energy F = µ (see below).e.

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. therefore. 2π (9. 2π (9. and noting |f (k)| = |2t sin k| = 2t one obtains δ( − (k)) = 1 − cos2 k = (2t)2 − 2 . (b) Graphical definition of the limiting energies α and β . Since the equation f (k) ≡ 2t cos k + = 0 has two simple roots. holds π ρ( ) = −π dk δ( + 2t cos k) . In our case (k) is given by (9. If the function f depends explicitly only on (k) one can state π −π dk f ( (k)) = 2π π ∞ ρ( )d f ( ) −∞ (9. −π .212) where xi are the simple roots of the function f (x).209) where ρ( ) = −π dk δ ( − (k)) . Inserting this last result into (9.210). |f (xi )| (9.278 Many–Particle Systems (a) (b) Figure 9. gives the the number of available one particle states per site.211) and using π dkδ(k ± (π − arccos( /2t))) = θ(2t − | |) .2 = ±(π − arccos( /2t)). namely k1.210) Here δ(x) is the Dirac–delta function and ρ( ). The shaded areas represent the filled portions of these bands by electrons. δ(k + π − arccos( /2t)) + δ(k − π + arccos( /2t)) (2t)2 − 2 . which is usually called the density of states.192) and.211) In order to calculate the integral we recall the following property of the Dirac–delta function δ [f (x)] = i δ(x − xi ) . unit energy and a given spin orientation of the particle.

(9.e. i. One obtains ˜ Emf = α β ρ( )d + −2t −2t ρ( )d + vnα nβ + µ(n − nα − nβ ) . if we assume that is restricted to the interval ] − 2t.221) ∂µ Equations (9.222) ..214) are given by the sum (9. (9. Replacing again the sum by an integral over energy one can write σ nσ = −2t ρ( )d (9. (9. (9.9.204) through an integral expression.216) Using (9. 9. µ.216) allow one.219. (9. β (9. v and n. 9. of t.e.201).. + vnα − µ = 0 .2]. on conditions for a ground state of minimum energy are ˜ ∂ Emf =0. one arrives at the following expression of the density of states ρ( ) = θ(2t − | |) π (2t)2 − 2 . From (9. in principle.221) and (9.f. ∂ α ˜ ∂ Emf =0.213) The presence of the θ function in the above formula guaranties that the density of states vanishes for | | > 2t.215) nα and nβ in (9.217) and µ. Fig.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..220) β ˜ ∂ Emf = n − nα − nβ = 0 . the necessary ˜ Since Emf [c. namely.219–9. to determine the unknown quantities α .8: Macroscopic Systems 279 where θ(x) is the step function. We can employ the results obtained to express the ground state energy density (9. namely. ∂ β ˜ ∂ Emf =0. ∂µ (9.206)] depends only on continuous quantities. (9.f. 2 (9.218) read ˜ ∂ Emf = ∂ α ˜ ∂ Emf = ∂ β and α + vnβ − µ = 0 . i.e. β .219) (9. nα and nβ as a function of the parameters of the Hubbard model. θ(x) = 1 if x > 0 and θ(x) = 0 if x < 0.218) The derivatives can be readily determined and the conditions (9.220) µ can be readily obtained µ= 1 ( 2 α + β) + vn . outside the energy band (k). 2t[ then the θ function in (9.213) can simply be replaced by 1. i. F = α + vnβ = β + vnα .213) and carrying out the resulting integral yields nσ = 1 arcsin( σ /2t) . π α.

229) has a wider range of validity than one can infer from the present analysis of the Hubbard model. (9.228) Since the total number of states per site is finite.227) will have a second.222) and (9. Accordingly. (9. f (∆) converges to a maximum value as ∆ → ∆max = n.280 Many–Particle Systems Comparision of (9. From the definition (9. One can determine vc by equating the slope of f (∆) at the origin to 1.2b).. (9. if.227) has only the trivial solution ∆ = 0.3].223) ∆ = 0 corresponds to the ground state (9. and only if. (9.3. a magnetic ground state. The slope of f (∆) at the origin depends on the Hubbard model parameter v. If the slope of f (∆) at the origin is less than 1. so called ferromagnetic solution.f. condition (9.225) Defining the function v∆ f (∆) = 0 ρ( β + )d . non trivial.226) follows that f (∆) is a monotonically increasing function of ∆. Equation (9. For this purpose one plots f (∆) versus ∆ as is shown schematically in Figure 9.219–9.229) 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.227) Since f (0) = 0 one can infer that ∆ = 0 is always a solution (the so called paramagnetic solution) of condition (9. i. For v above a critical value vc . i. from (9. the slope of f (∆) is positive for all ∆ ≥ 0.220) and (9.217) and (9. The Stoner criterion (9.227).226) ∆ = f (∆) . First.183) without magnetization. (9. . We assume in the following discussion that the implicit ∆ dependence of β in f (∆) can be neglected.. Figure 9.227) can be solved graphically. One still needs to determine α and β . For this purpose we consider the magnetization of the mean field ground state ∆ ≡ nα − nβ . as reflected by the expression d[f (∆)] = vρ( d∆ β + ∆) = π (2t)2 v − (v∆ + 2 β) >0.226) one can obtain ∆ by solving (cf. (9.223) follows α α − v β . from (9.228)] vc ρ ( β ) = 1 . the slope of f (∆) at the origin is larger than 1 [c. condition (9.f. (9. Indeed.223 one obtains ∆= and second. (9. One can express ∆ as a function of α and β in two different ways.215) reveals that µ is indeed equal to the Fermi energy F (see also Figure 9.e.224) v∆ ∆= β ρ( )d = 0 ρ( β + )d .e. the Hubbard model has a ground state with ∆ = 0. (9. through the condition [c.

3: Graphical solution of the mean field equation ∆ = f (∆). t. n).230) one obtains from (9.220.205) the following set of equations determining x. (9. 9. any . (9. 2 2 2 (9.219. Since ∆ vanishes near vc we set x = y + δ where δ is a small quantity. Accordingly..232) states that x − y is proportional to ∆. α.9. We like to determine finally the dependence of vc on the filling factor n.234) 2t 2 The resulting phase diagram of the ground state.8: Macroscopic Systems 281 Figure 9.233) yield the desired expression vc π = π cos (n − 1) . (9.231) (9. β x+y x−y = = π(n − 1) 2t π (sin x − sin y) = π∆ .4. v (9. The corresponding 2t analytical expression provides the phase diagram of the ground state: when v approaches vc from above. β = 2t sin y . From (9. the plot of vc vs. i. the magnetization ∆ of the ground state vanishes and remains zero for values v < vc . the ground state is characterized by the quantities (v. y and.231) follows y= δ π π (n − 1) − ≈ (n − 1) .e.232) together with (9.232) For a given 2t the values of x and y can be obtained numerically and other relevant quantities v mentioned above can be calculated.233) Employing the approximation sin(y + δ) − sin y ≈ δ cos y for small δ. hence. 9. n. is presented in Figure 9. Equation (9. 2t As already mentioned. By parameterizing α and β α = 2t sin x .

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

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

284 Many–Particle Systems .

(refeq:ham2. t) c 2 i + qV (r. x2 . any of these equations being specific for certain classes of particles. spin–0 particles. The representation 285 .1) which connect space time coordinates (x1 .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.f. some of the equations describe a particle together with its anti-particle.g. t ) in another frame. It is not possible to uncouple the equations to describe only a single type particle without affecting negatively the Lorentz invariance of the equations. e.45)] described by an electrical potential V (r. 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. spin– 1 particles. 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. t) and a vector potential A(r. x3 = x3 (10. t) ∂ i ψ(r. 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. Here c denotes the velocity of light.. t) ψ(r. v = v1 x1 . 8.e. x2 . 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. etc. We will find that actually several Lorentz–invariant equations which replace (10.2) will result. t) (10. Furthermore. Obviously.e. t) are vectors of dimension larger one. 2 As mentioned..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. these ˆ transformations are x1 = x1 − v1 t 1 − v1 2 c .2) The replacement of (10. i. t) in one frame with space time coordinates (x1 . x3 . The transformations beween such frames according to the Theory of Special Relativity are described by Lorentz transformations. x3 . i. In case that v is oriented along the x1 –axis. t = t− 1 − v1 x c2 1 v1 2 c . t) = ∂t 1 2m q − A(r. x2 = x2 .. the components representing the spin attribute of particles and also representing together with a particle its anti-particle.

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

9) and (10. This will be explained further below. ν=0 (10. i. We will exploit this property below to determine the general form of the Lorentz transformations. xµ .9)  1 0 0  0 −1 0 0   ( gµν ) =   0 0 −1 0  = g .7) The reason for the notation is two-fold. 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 .1: Natural Representation of the Lorentz Group 287 coordinates in one reference frame and by x µ the coordinates in the other reference frame.8) new old old The summation convention allows us to write (10. the Lorentz transformations constitute a linear transformation which we denote by 3 x µ = ν=0 Lµ ν xν . Since this holds for any xµ it must be true Lµ ρ gµν Lν σ = gρσ . Aµ ν xν = µ=0 new old Aµ ν xν . Aµν . the notation in (10. The Lorentz transformations form the subgroup of all matrices which leave the expression (10. Aµ ν .e. [ A more conventional notation would be Lµν . Aµν .e.5) invariant. the .6) allows us to introduce the so-called summation conventon: any time the same index appears in an upper and a lower position. 2. 3: 4-vector: xµ . L ∈ GL(4.11) This condition specifies the key property of Lorentz transformations.5) into scalar products. (10. R). summation over that index is assumed without explicitly noting it. 3 3 3 yµ xµ = new yµ xµ . The second reason is that upper and lower positions allow us to accomodate the expression (10. ν = 0. however.. Aµ ν B ν ρ = ν=0 Aµ ν B ν ρ . 1.] The following possibilities exist for the positioning of the indices µ. 4 × 4 tensor: Aµ ν .. (10.12) (10. (10. R). The subset of GL(4. 0 0 0 −1 (10. i. the latter notation will be used for different quantities further below. This condition can be written xµ gµν xν = x gµν x where  µ ν (10. First. the latter set constituting a group.6) x µ = Lµ ν xν .6) Here Lµ ν are the elements of a 4 × 4–matrix representing the Lorentz transformation. The Lorentz transformations are non-singular 4 × 4–matrices with real coefficients.10.10) Combining condition (10.

is called O(3.288 Relativistic Quantum Mechanics elements of which satisfy this condition. hence. LT gL = g } . i.22) (10. a group.e. R) is a group.10) of g one can rewrite the invariance property (10. 10.14). (10. L−1 ∈ O(3. . R). L2 ∈ O(3.e.14) from the right by gLT and using (10.14) holds for the inverse of L. To demonstrate the group property of O(3.14)  L0 0 −L1 0 −L2 0 −L3 0  −L0 1 L1 1 L2 1 L3 1   . L3 0 L3 1 L3 2 L3 3 Using the definition (10.1) ⊂ GL(4. 1) = { L.16. 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   (10.21) i. i.e. This set is identical with the set of all Lorentz transformations L. the inverse of L 1  L−1 (10.17) one obtains (L−1 )T g L−1 = gLgggLT g = gLgLT g .15) (10.13) L = ( Lµ ν ) =  2 0  L 0 L2 1 L2 2 L3 3  . employing expressions (10. L ∈ GL(4.20) results in the desired property (L−1 )T g L−1 = g . 3 2 1 2 1 2 (10.19) i.1).. property (10.1). of (10.16) The corresponding expression for (LT )−1 is obviously (LT )−1 = (L−1 )T = g L g .1) since it satisfies (10..1). = 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. This stipulates that O(3.15) one can derive LT gLgLT = LT and multiplying this from the left by by g(LT )−1 yields L g LT = g Using this result to simplify the r. In fact.20) Multiplying (10. For L3 = L1 L2 holds LT g L3 = LT LT g L1 L2 = LT (LT gL1 ) L2 = LT g L2 = g .18) (10.12) LT gL = g .1). (10.1). the associated inverse obeys (10. (10. of O(3.. We consider 1 then L1 .e. One can also show that if L ∈ O(3.s. From this one can obtain using g2 = 1 1 (gLT g)L = 1 and.14).17) we note first that the identity matrix 1 is an element of O(3.1).. L3 ∈ O(3. in fact.h.1) is.

It holds g ∈ L and −1 ∈ L as one can readily verify testing for property (10. det L = −1.28) (10.23) One can classify Lorentz transformations according to the value of the determinant into two distinct classes. L ∈ O(3.33) where we used the definition aM = {M1 . L↓ . 1). there exist two other distinct classes. M2 ∈ M.. (10. The above shows that the set of proper Lorentz transformations L↑ allows one to generate all Lorentz transformations. 1). σ = 0 yields L0 0 2 − L1 0 2 − L2 0 2 − L3 0 2 = 1. = { L.12).31) (10. = { L. Considering in (10. (10. L0 0 ≥ 1} . L0 0 ≤ −1} . 1).1: Natural Representation of the Lorentz Group Classification of Lorentz Transformations 289 We like to classify now the elements of L = O(3. + − gL↑ + = L↑ g . L↑ contains 1 1. hence. L ∈ O(3. One can also 1 verify that one can write L↑ − L↓ + L↓ − = gL↑ + = = −L↑ . B ∈ L+ holds C = AB ∈ L+ . (10. Obviously. L0 0 ≥ 1} . Properties (10. + except for the trivial factors g and −1 It is. M1 = a M2 }. L0 0 (10.34) j=1 j=1 .12) the case ρ = 0.24) or since (L1 0 )2 + (L2 0 )2 + (L3 0 )2 ≥ 0 it holds (L0 0 )2 ≥ 1. det L = −1. = { L. L↓ are disjunct sets defined as follows + + − − L↑ + L↓ + L↑ − L↓ − = { L. det L = 1. Using det AB = det A det B and det AT = det A yields (det L)2 = 1 or det L = ±1 . entirely suitable to investigate first only 1.1).14).29) (10. ∃M2 . (10.30) ≤ −1} .14).23) and (10. Lorentz transformations in L↑ .26) (10. 1).27) (10.25) i. From this we can conclude L0 0 ≥ 1 or L0 0 ≤ −1 . For this purpose we consider first the value of det L. 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 .10. A statement on this value can be made on account of property (10. + = − L↑ g + (10. L ∈ O(3. A second class property follows from (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 .e. L ∈ O(3. L↑ .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↑ . det L = 1. + + ↑ ↑ We can also demonstrate that for A.32) (10. + We start our investigation by demonstrating that L↑ forms a group.14) which we employ in the formulation (10.

therefore.13). (10. + i.37) (10.e.39) where we have employed the matrix form T 1 defined as in (10.1) ascertains CT gC = g it must hold C 0 0 ≥ 1. L↑ is called the subgroup of proper. 1 We will then employ the Lie group properties to generate all transformations in L↑ . (10. 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. µ ν small . hence. if we enforce (10.34) provides then the estimate conclude from (10.14) implies 1 + T g (1 + ) = g 1 1 (10. Similarly.16). 1 0 0 = ≥ 1 and.1). In fact. obviously j=1 A0 0 B 0 0 ≥ 1.35) One can conclude.1).290 Relativistic Quantum Mechanics From (10.12) follows (B 0 0 )2 − 3 (B j 0 )2 = 1 or 3 (B j 0 )2 = (B 0 0 )2 − 1. from which one can conclude L−1 ∈ L↑ . Infinitesimal Lorentz transformations The transformations in L↑ have the property that they are continously connected to the identity 1 1. i. accordingly. δ00 = 1 and It should be noted that according to our present definition holds δµν = gµρ δ δ11 = δ22 = δ33 = −1. (10. + transformations in a small neighborhood of 1 which we parametrize by infinitesimal parameters. since the group property of O(3. Since the associative property holds for matrix multipli1 1 + ↑ cation we have verified that L+ is indeed a subgroup of SO(3. | 3 A0 j B j 0 | < A0 0 B 0 0 .36) It holds. In the following we + will consider solely this subgroup of SO(3. Since A0 0 ≥ 1 and B 0 0 ≥ 1. ρ ν (10.14) actually L0 0 ≥ 1 and det L = 1 must hold. . To order O( 2 ) holds g + g = 0. holds L0 0 > 0 and the value of the determinant is close to unity. This relationship shows (L−1 )0 0 = L0 0 . 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. we consider transformations Lµ ν = δ µ ν + µ ν . one can j=1 j=1 3 0 )2 = (A0 )2 − 1. For the inverse of + ↑ any L ∈ L+ holds (10.40) and. 1 ∈ L↑ ..38) For these transformations. Property (10. + Accordingly.e.. orthochronous Lorentz transformations. Using the above expression for C 0 0 one can state C 0 0 > 0.21) j=1 (A j 0   3 j=1 A0 j B j 0  ≤ 2 (A0 0 )2 − 1 (B 0 0 )2 − 1 < (A0 0 )2 (B 0 0 )2 . The next group property of L↑ to be demonstrated is the existence of the inverse. obviously.

 (10.1: Natural Representation of the Lorentz Group Using (10. ϑ3 . 2.42) This relationship implies µ 0 j µ j = 0 = j 0 . other five parameters zero) . K2 =  . (ϑ1 .44) This result allows us now to define six generators for the Lorentz transformations(k = 1.46)  0 0   0 0  . 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 . 2.10. w3 ) =   −w2 ϑ3 0 −ϑ1  −w3 −ϑ2 ϑ1 0 (10. j. J =  0  3  0 1 0 0 0 0 0 0 0 0 0 0 0     (10. J ] [ Kk .48) (10. K ] [ Jk . 3) Jk = (ϑk = 1. 3 (10.15) one can conclude T 291 = −g g (10.45) (10. w2 . The generators are explicitly  0 0 0 0  0 0 0 0 J1 =   0 0 0 −1 0 0 1 0  (10. 2. ϑ2 . other five parameters zero) Kk = (wk = 1.49) . k j j = 1. w1 . k = 1.50) . = − k m Km (10.41) 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 ] The operators also obey J · K = J1 J1 + J2 J2 + J3 J3 = 0 = = k m Jm k m Jm  0 −1 0 0   0 0  0 0 (10.43) k = − Inspection shows that the matrix has 6 independent elements and can be written   0 −w1 −w2 −w3  −w1 0 −ϑ3 ϑ2   .47)     0 −1 0 0 0 0 −1 0 0  −1   0 0   0 0 0 0  0 0  K1 =  . J2 =    0 0 0 −1  0 0 0 0   0 0 0 0 0  0 0 −1 0 1  . 3 .

Exercise 7.e. that the transformations (10. For the parameters ϑk of the Lorentz transformations holds obviously ϑk ∈ [0. To determine the explicit form of this transformation we evaluate the exponential operator by Taylor expansion.1: Demonstrate the commutation relationships (10.49.292 as can be readily verified. Kk . 10. (10.56) = 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. 2π[ . 3 is closed.54) which. 3 where we have defined ϑ · J = k=1 ϑk Jk and w · K = following the algebra in Chapter 5. 0  1 n (10. k = 1. Relativistic Quantum Mechanics The commutation relationships (10.. constitutes an overcomplete parametrization of the rotations (see Chapter 5). A boost in the x1 –direction is L = exp(w1 K1 ). i.50). 2. In analogy to equation (5. and using the relationship Jk = 0 0 0 Lk (10. however.53) Here R(ϑ) are the 3 × 3–rotation matrices constructed in Chapter 5. w ∈ R3 3 k=1 wk Kk .49) define the Lie algebra associated with the Lie group L↑ . 3 (10. w) = exp ϑ · J + w · K .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   . 2.57) . k = 1.51) One can readily show.51) for w = 0 correspond to rotations of the spatial coordinates. L(ϑ. w = 0) = 0 0 0 R(ϑ) . (10.52) where the 3 × 3–matrices Lk are the generators of SO(3) defined in Chapter 5.55) L 0 0 0 0  exp (w1 K1 ) = exp   Using the idempotence property 0 −1 −1 0 2 0 0 1 0 (10. + 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’. ϑ.

t = t − v1 x1 2 1 − v1 . in fact.59).60) sinh w1 = v1 / 2 1 − v1 . We expect that vk can only assume values less than the velocity of light c which in the present units is c = 1. we can state vk ∈ ] − 1.51) the explicit transformation for space–time–coordinates is then x1 = x1 − v1 t 1 − 2 v1 .59) for the case k = 1 corresponds to the relative velocity of two frames of reference. consistent with (10. one obtains from (10.61) √ √−v1 √   exp (w1 K1 ) =    2 1 − v1 1 2 1 − v1 0 0 0 0 0 0  0 0    1 0  0 1  (10. for wk wk ∈ ] − ∞. The range of the parameters wk can now be specified.1) of the Lorentz transformations is obtained through the parameter change sinh w1 v1 = = tanh w1 (10. x2 = x2 .1). From (10.59) follows. ∞[ . (10.56.58) = cosh w1 1 + sinh w1 1 0 −1 −1 0 The conventional form (10.62) According to (10. 1[. x3 = x3 (10. 10. Accordingly.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 .59) can be written  1 2 1 − v1 √−v1 2 1 − v1 2 1 − v1 . 10. This property is. . (10. (10.64) We note that the range of wk -values is not a compact set even though the range of vk -values is compact.59) cosh w1 Using cosh2 w1 − sinh2 w1 = 1 one can identify sinhw1 = sinh2 w1 + 1.59) v1 = These two equations yield cosh w1 = 1/ and (10. (10.63) which agrees with (10. 10. cosh2 w1 − 1 and coshw1 = cosh2 w1 − 1 = cosh w1 sinh w1 sinh2 w1 + 1 .3.10. This property of the wk -values contrasts with the property of the parameters ϑk specifying rotational angles which assume only values in a compact range.6. Correspondingly. however. vk defined in (10.

67) the transformation behaviour of which we will demonstrate further below. t) are the charge density and the current density.66) Examples of 4-vectors beside xµ are the momentum 4-vector pµ = (E. f = f. a1 . We have encountered examples in connection with rotational transformations. not any physical property f ∈ R is a scalar. We will see below how true scalars under Lorentz transformations can be constructed. a3 )T and transform according to a µ = Lµ ν aν . another example is the rest mass m of a particle. respectively. Such quantities appear in the description of physical systems and statements about transformation properties are often extremely helpful and usually provide important physical insight.e.3) are the so-called 4–vectors aµ . the z–component x3 of a particle. Hence. namely. Presently. However.4). These quantites always come as four components (a0 . vectors like r = (x1 .2 Scalars. (10. finally. the square of the electric field |E(r. E = √ m mv .294 Relativistic Quantum Mechanics 10. A) (10.69) where V (r. Nevertheless.. 4–Vectors and Tensors In this Section we define quantities according to their behaviour under Lorentz transformations. x3 )T . t) are the electrical and the vector potential of an electromagnetic field. . p = √ 2 1−v 1 − v2 (10. x2 . A third 4-vector is the so-called current vector J µ = (ρ. of a system of charges. 4-Vectors The quantities with the transformation behaviour like that of the position–time vector xµ defined in (10. tensor operators Tkm . p) . Another example is the potential 4-vector Aµ = (V. (10. Counterexamples are the energy. r2 ) and. our definition of quantities with special properties under Lorentz transformations presently is confined to the natural representation. we have not yet defined representations of Lorentz transformations beyond the ‘natural’ representation acting in the 4–dimensional space of position– and time–coordinates. a2 . total angular momentum r 1 2 3 states of composite systems like Y m ( 1 .65) An example is s2 defined in (10. we will encounter an impressive example of physical properties. scalars like r = x2 + x2 + x2 . i. t)|2 or the scalar product r1 · r2 of two particle positions. Some of these r ˆ quantities were actually defined with respect to representations of the rotation group in function spaces. The 4-vector character of J µ and of Aµ will be demonstrated further below. not in the so-called natural representation associated with the 3–dimensional Euclidean space E3 . spherical harmonics Y m (ˆ). t) and A(r. Scalars The quantities with the simplest transformation behaviour are so-called scalars f ∈ R which are invariant under transformations. the charge density. J) (10.68) where ρ(r. 2 |ˆ1 . t) and J(r.

66) together with (10. accordingly. The 4-Vector ∂µ (10.76) An important example of a covariant 4-vector is the differential operator ∂µ = ∂ = ∂xµ ∂ . The respective vectors aµ are associated with the 4-vectors aµ .71) Contravariant and Covariant 4-Vectors It is convenient to define a second class of 4-vectors. −a1 .17) can be written (L−1 )ν µ = Lµ ν . We start from x µ = Lµ ν xν . i. We do not fact. i. g µν ∂ν = Lµ ρ g ρσ ∂σ . ∂x µ (10. Covariant 4-vectors transform like a µ = gµν Lν ρ g ρσ aσ where we defined g µν = gµν . Multiplication (and summation) of x µ = Lµ ν xν by Lρ σ gρµ yields. lower.73) Note that according to (10.66). We have duplicated the expression for the inverse of Lµ ν to obtain the correct notation in terms of covariant. This property follows from (10.12). one can employ the tensors g µν to raise and lower indices of L ν establish here the consistency of the ensuing notation..16). We like to point out that from definition (10. 4–Vectors and Tensors Scalar Product then 295 4-vectors allow one to construct scalar quantities.70) is a scalar. In fact. using (10.79) This is the inverse Lorentz transformation consistent with (10. aµ (10.e.75) (10. one can express [(L−1 )T ]µ ν = (L−1 )ν µ and. In any case one can express (10.17) Lµ σ is the transformation inverse to Lσ µ .10. the relationship being aµ = gµν aν = (a0 .74) g µν aν . If aµ and bµ are 4-vectors aµ gµν bν (10.77) The transformed differential operator will be denoted by ∂µ = def ∂ .12) a gµν b µ ν = Lµ ρ gµν Lν σ aρ bσ = aρ gρσ bσ (10. gσν xν = Lρ σ gρµ x µ and g µσ gσν = δ µ ν . and . xν = g νσ Lρ σ gρµ x µ . (10. One calls 4-vectors aµ covariant and 4-vectors aµ contravariant.78) To prove the 4-vector property of ∂µ we will show that g µν ∂ν transforms like a contravariant 4vector. (10.73) aµ = Lµ σ aσ . −a2 . −a3 ) (10.e. ∂t (10.72) of the covariant 4-vector follows = In µν and g µ as well.2: Scalars. (10..72) where aν is a vector with transformation behaviour as stated in (10.

296 Relativistic Quantum Mechanics contravariant.81) ∂ . i. ∂µ does indeed transform like a covariant vector.67) transforms like (10.84) = 1 − (v)2 (10. Using the chain rule of differential calculus we obtain 3 ∂µ = ν=0 ∂xν ∂ = g νσ Lρ σ gρµ ∂ν = Lµ ν ∂ν ∂x µ ∂xν (10..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 .79) allows us to determine the connection between ∂µ and ∂µ . d’Alembert Operator We want to construct now a scalar differential operator. (10. 2 dt dτ 1 − v (10.83) is a scalar under Lorentz transformations.85) 1 d d = √ .88) One can express then the momentum vector pµ = √ d m dxµ = m xµ . dτ 1 − v 2 dt p0 = E = √ m m dt = √ . 1 − v2 1 − v 2 dt (10. this operator is equal to the d’Alembert operator which is known to be Lorentz-invariant.89) . upper.e. indices.− ∂t . Proof that pµ is a 4-vector We will demonstrate now that the momentum 4-vector pµ defined in (10. (10. i. p1 = √ m v1 m dx1 = √ . 2 2 dt 1 − v 1 − v (10. e.g. For this purpose we define first the contravariant differential operator ∂ µ = g µν ∂ν = Then the operator 2 ∂µ ∂ µ = ∂t − 2 (10.87) The remaining components of pµ can be written.82) (10.86) One can write (10..66). 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. In fact..e.

90) (10. . can be demonstrated readily using the 4-vector notation (10.92) (10.93) (10.s.96) in the form ∂µ Aµ = 0 . namely pµ pµ = pµ gµν pν = E 2 − p 2 Evaluation of the r. (10. We will demonstrate now the 4-vector property of Aµ .77) which allow one to express (10. a contravariant 4-vector then the l. t). in fact.h. of (10. Expansion in should then be rapidly convergent. (10. t) and A(r.89) must be a 4-vector. in (10. is a scalar.h.10. i.97) and.97) We have proven already that ∂µ is a contravariant 4-vector. for these potentials.h.12).3: Relativistic Electrodynamics 297 d The operator m dτ transforms like a scalar.s. in fact.69) for the electrodynamic potential and the 4-vector derivative (10. ∂t V (r. Since xµ transforms like a contravariant 4-vector.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. respectively.95) 2 p This obviously describes the energy of a free particle with rest energy ±m.s.91) (10.89) alltogether transforms like a contravariant 4-vector. We like to rewrite the last result E 2 = p 2 + m2 or E = ± p 2 + m2 . the r. of (10.s. hence. This gauge is not Lorentz-invariant and we will adopt here another gauge. pµ on the l. the so-called Lorentz gauge. equivalently.67) E2 − p 2 = or pµ pµ = m2 which. and had chosen the so-called Coulomb gauge (8. in (10. yields according to (10. namely.h. hence. t) + · A(r.96) The Lorentz-invariance of this gauge.e. t) = 0 . (10. kinetic energy ± 2m and relativistic corrections. Lorentz-invariant.94) 1 m In the non-relativistic limit the rest energy m is the dominant contribution to E. One obtains E = ±m ± p2 2m (p 2 )2 + O 4m3 (p 2 )3 4m5 .69) is. The momentum 4-vector allows us to construct a scalar quantity. 8. 10. If we can show that Aµ defined in (10. m2 m2 v 2 − = m2 1 − v2 1 − v2 (10. and.96) is a scalar and. Lorentz Gauge In our previous description of the electrodynamic field we had introduced the scalar and vector potential V (r. · A = 0..

The respective equation for A0 = V can be obtained from Eq.104) where mn = mn is the totally anti-symmetric three-dimensional tensor defined in (5. 10. one obtains 2 ∂t V (r. must be a scalar.77) and (10. this can be stated F k0 = −F 0k = E k . it follows that J µ . transforms like a 4-vector. t) together with (10.s. The l..e.100) Similarly. 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   .69) and ∂ µ in (10.99) Since this equation must be true in any frame of reference the right hand side must vanish in all frames.99) must be a scalar. t) 2 ∂t A(r. according to (10. i. (10. ∂µ J µ (xµ ) = 0 . F mn = − mn (10. a 4-vector. in fact. This principle should hold in any frame of reference. . t) = 4 π J(r. i.18) and. of (10. 2. that F µν can be expressed through the potential Aµ in (10. one obtains for A(r. Since ∂µ ∂ µ transforms like a scalar one can conclude that Aν (xσ ) must transform like a 4-vector.s. 3 (10. t) − 2 V (r. using (10. t) = 0 (10. Consequently. t) .e.6) and (8.298 Transformation Properties of J µ and Aµ Relativistic Quantum Mechanics The charge density ρ(r. t) from (8. (8. t) and current density J(r. Equation (10.100.96). prove that Aµ is.97) and. t) = −∂t V (r. t). · A(r.h. m.101) Combining equations (10.s.69).105) . thereby. (10. t) − 2 A(r.h.13). t) = −∂t V (r.96). (10. F µν =  x  Ey B z 0 −Bx  Ez −By Bx 0 Alternatively. One can readily verify.102) In this equation the r. also the l. using (8. t) = 4πρ(r. (10.98) can be written. Using · ∂t A(r. in fact.103) B .17) using the identity (8. must transform likewise.h. n = 1.32).82) as follows F µν = ∂ µ Aν − ∂ ν Aµ . We want to derive now the differential equations which determine the 4-potential Aµ in the Lorentz gauge (10.68).83) and (10. · A(r. t) + · J(r. (10. yields ∂µ ∂ µ Aν (xσ ) = 4 π J ν (xσ ) . t) are known to obey the continuity property ∂t ρ(r. has to transform like a contravariant 4-vector..101). k.98) which reflects the principle of charge conservation.9). t) . using (10. Since ∂µ transforms like a covariant 4-vector. t) = ∂t · A(r.

by (10.2).63).3: Relativistic Electrodynamics 299 The relationships (10.102) ∂µ F µν = 4π J ν . one can conclude from (10.104) establishe the transformation behaviour of E(r. Noting ∂µ F µν = ∂µ ∂ µ Aν − ∂µ ∂ ν Aµ = ∂µ ∂ µ Aν − ∂ ν ∂µ Aµ = ∂µ ∂ µ Aν . . where we used (10. Maxwell Equations in Lorentz-Invariant Form One can express the Maxwell equations in terms of the tensor F µν in Lorentz-invariant form.4). respectively. equivalently.103.10. B and E⊥ . 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. These equations show that under Lorentz transformations electric and magnetic fields convert into one another. The results can be put into the more general form E B = = E .113) which can be shown to be equivalent to the two homogeneous Maxwell equations (8.109) (10.106) In case that the Lorentz transformation Lµ ν is given by (10. E⊥ = B⊥ E⊥ + v × B √ 1 − v2 B⊥ − v × E = √ 1 − v2 (10. 8.112) (10. B .105) and (10. In a new frame of reference holds F µν = Lµ α Lν β F αβ (10. B⊥ are.62) or.108) yields then the expressions for the transformed fields E and B .97).110) where E .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  (10.3.111) One can readily prove that this equation is equivalent to the two inhomogeneous Maxwell equations (8.1.105) of the tensor F µν one can conclude the property ∂ σ F µν + ∂ µ F νσ + ∂ ν F σµ = 0 (10. 10. From the definition (10. the components of the fields parallel and perpendicular to the velocity v which determines the Lorentz transformation. 8. t) and B(r. t). (10.

yields dpx = q dt Ex + (v × B)x (10. (10.. hence.86) dE q = p·E .5). We want to √ demonstrate now that this equation is equivalent to the equation of motion (8.116) (10.117) (10. demonstrated that (10. It holds for a particle with 4-momentum pµ as given by (10.87)] and retain the definition E for the electric field.5) where p = mv/ 1 − v 2 .120) is referred to as the Lorentz force.104). To avoid √ confusion we will employ in the following for the energy of the particle the notation E = m/ 1 − v 2 [see (10. e.120) (10. dτ m √ Employing again (10. equivalent to (8. For the spatial components. We want to express this force through the tensor F µν .e. using (10.115) This equation follows. dt E From this one can conclude.114). provides an alternative description of the action of the Lorentz force.67). also from the equation of motion (8. employing (10.114) where d/dτ is given by (10.114) reads using (10.119) which is the x-component of the equation of motion (8.g.67) and charge q dpµ q = pν F µν dτ m (10. We have. hence.103) dpx q = ( EEx + py Bz − pz By ) . however.86). for µ = 1. dE q = p·E dτ m or with (10. The µ = 0 component of (10.118) p· dt √ where we exploited the fact that according to p = mv/ 1 − v 2 holds p v. of (10.h.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.93). 1 dE 2 1 dp 2 = = qp·E 2 dt 2 dt (10. Equation (10. E = m/ 1 − v 2 . .114) reads then.5) taking the scalar product with p dp = qp · E (10.5).114) is. The term on the r.86) and (10. i. in fact.s..

obey a different transformation law. We need to emphasize that (10.. We will denote the intended representation by L(ϑ. w) = ρ(Lµ ν (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 .g.48) and (10. the charge-current density in case of Sect. i.e.125)]. ψ ∈ C∞ (4).123) This definition corresponds closely to the function space representation (5. w.3 or the bi-spinor wave function of electron-positron pairs in Sect.48) of SO(3). i.47). w) which correspond to the generators Jk and Kk in (10.122) 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 ) . e.126) ρ ewk Kk − 1 1 . 10.127) .121) by (L−1 )µ ν xν and obtains ψ (xµ ) = ψ((L−1 )µ ν xν ) . In this section we will investigate the transformation properties of scalar functions ψ(xµ ). J2 . (10.125) def (10.. The resulting expression should be a generalization of the function space representation (5. For such functions holds in the transformed frame ψ (Lµ ν xν ) = ψ(xµ ) (10. (10.42) of SO(3).124) In this expression J = (J1 . (5. w)) = ρ eϑ·J + w·K which we present in the form L(ϑ.e. K3 ) are the generators of L(ϑ. 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.121) covers solely the transformation behaviour of scalar functions. xµ ). in the present case we exclude the factor ‘−i’ [cf. However.1) is a generalization (rotation + boosts) of the group SO(3). we seek an expression which corresponds to (10.51) for the natural representation of the Lorentz group. Functions which represent 4-vectorsor other non-scalar entities.10. def (10. Accordingly. w) = exp ϑ · J + w · K . (10. K2 .4 Function Space Representation of Lorentz Group In the following it will be required to decribe the transformation of wave functions under Lorentz transformations. For this purpose one replaces xµ in (10. and which can be constructed following the procedure adopted for the function space representation of SO(3).7.. 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. taken at the pairs of points (x µ = Lµ ν xν . J3 ) and K = (K1 . We like to express now ψ (x µ ) in terms of the old coordinates xµ .4: Function Space Representation of Lorentz Group 301 10. in as far as SO(3.

sinhw1 x0 + coshw1 x1 .131) This yields for small w1 ρ ew1 K1 ψ(xµ ) = ψ(coshw1 x0 + sinhw1 x1 . K2 ] = K3 : [ J1 . One can determine similarly K1 starting from coshw1 sinhw1  sinhw1 coshw1 =   0 0 0 0  0 0 1 0  0 0   . [K1 .128). K2 ] = −J3 .126). 1 This result. J ] [ Kk . x1 ∂3 − x3 ∂1 ] = [x3 ∂2 . i. J2 ] = J3 . J3 = x2 ∂1 − x1 ∂2 . reproduces the expression for J1 in (10. x1 . x1 ∂3 ] − [x2 ∂3 . J2 = x1 ∂3 − x3 ∂1 .49) as the generators of the natural representation. −sinϑ1 x2 + cosϑ1 x3 ) = ψ(xµ ) + ϑ1 (x3 ∂2 − x2 ∂3 ) ψ(xµ ) + O(ϑ2 ) . x3 ) 2 = ψ(xµ ) + w1 (x1 ∂0 + x0 ∂1 ) ψ(xµ ) + O(w1 ) (10.134) .302 One obtains J1 = x3 ∂2 − x2 ∂3 . (10.e.129) which yields for small ϑ1 ρ eϑ1 J1 ψ(xµ ) = ψ(x0 . [ Jk . The generators J . In order to evaluate (10.130) ew1 K1 −1 = e−w1 K1 (10. the expression for K1 in (10.126) 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. x3 ∂1 ] = −x1 ∂2 + x2 ∂1 = J3 . K obey the same Lie algebra (10.132) and.128) eϑ1 J1 = e−ϑ1 J1 1  0 =   0 0   0 0 0 1 0 0   0 cosϑ1 sinϑ1  0 −sinϑ1 cosϑ1 (10. namely [J1 . obviously.133) We demonstrate this for three cases. x2 . J2 ] = [x3 ∂2 − x2 ∂3 . K ] [ Jk . and [J1 . K ] = = k m Jm k m Jm = − k m Km . cosϑ1 x2 + sinϑ1 x3 . obviously. which we like to demonstrate for J1 and K1 . 0  1 (10. (10.

[ J1 .142) (10. Ω which are connected with x0 . x2 ∂0 ] − [x1 ∂0 .125) in the case of a one-dimensional transformation of the type L(w3 ) = exp w3 K3 . ∂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 . ∂Ω .136) (10. x0 ∂2 ] = x3 ∂0 + x0 ∂3 = K3 . x2 + x2 + x2 1 2 3 in the case of SO(3) and (x0 )2 − (x3 )2 in case of transformation (10. x2 ∂0 ] − [x2 ∂3 .137) where K3 is given in (10. 10.135) (10. ∂3 in terms of ∂R . x3 as follows x0 = R coshΩ . cothΩ = 3 .141) (10. 303 (10. We note tanhΩ = ∂R x0 = . For this purpose one expresses K3 in terms of hyperbolic coordinates R.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. x0 ∂2 + x2 ∂0 ] = [x0 ∂1 .4: Function Space Representation of Lorentz Group [ K1 . Eqs.138) x3 x0 .e. a relationship which can also be stated R = and + − (x0 )2 − (x3 )2 (x0 )2 − (x3 )2 if x0 ≥ 0 if x0 < 0 (10.139) x3 = R sinhΩ (10. K2 ] = [x0 ∂1 + x1 ∂0 .139. The relationships (10. x0 ∂2 + x2 ∂0 ] = [x3 ∂2 ..137). i. K2 ] = [x3 ∂2 − x2 ∂3 . (5.87)]. x0 ∂2 ] = −x2 ∂1 + x1 ∂2 = − J3 . In the following we consider solely the case x0 ≥ 0. (10. 0 ∂cothΩ ∂x x = ∂R x0 = − ∂x3 R (10. In both cases the radial coordinate is the quantity conserved under the transformations.10.85-5. 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 .140) allow one to ∂ ∂ express the derivatives ∂0 .143) .128). can be simplified considerably. One-Dimensional Function Space Representation The exponential operator (10.

122). and (time)−1 all have the same dimension. In the position representation the correspondence principle states ˆ E =⇒ E = − ∂t i ˆ p =⇒ p = i which. In these units the quantities energy. − ) = i ∂ µ . ˆ (10.92) one obtains the wave equation − or 2 2 µ ∂ ∂µ ψ(xν ) = m2 ψ(xν ) ψ(xν ) = 0 . The latter property implies that upon change of reference frame ψ(xµ ) transforms according to (10. We will provide a ‘derivation’ of these two equations which stem from the historical development of relativistic quantum mechanics.147) (10.121. 10. mass.144) The action of the exponential operator (10.5 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. (length)−1 .304 Inserting these results into the definition of K3 in (10.146) Applying the correspondence principle to (10. which allows one to replace classical observables by quantum mechanical operators acting on wave functions. however. Further below we will provide a more systematic. ∂Ω Relativistic Quantum Mechanics (10. (10. in 4-vector notation reads pµ =⇒ pµ = i (∂t . The partial differential equation (10.150) . The historic route to these two equations.145) ∂Ω 10. Free Particle Case A quantum mechanical description of a relativistic free particle results from applying the correspondence principle. momentum.128) yields K3 = x0 ∂3 + x3 ∂0 = ∂ .151) is called the Klein-Gordon equation.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 ) . but certainly is short and. In natural units the Klein–Gordon equation (10. representation theoretic treatment. complex-valued function. ˆ pµ =⇒ pµ = i(∂t . namely the Klein–Gordon and the Dirac equations. therefore. is not very insightful.149) where ψ(xµ ) is a scalar.151) reads ∂µ ∂ µ + m2 ψ(xµ ) = 0 (10. In the following we will employ so-called natural units = c = 1. (10.148) ∂ µ ∂µ + m2 (10. extremely useful. ) = i ∂µ .

155) describes the positive definite probability to detect a particle at position r at time t and where jS (r. t)∂t ψ(r. t) ψ(r. t) − ψ(r. 2mi (10.160) ∂t (ψ ∗ ∂t ψ which corresponds to − ψ∂t ψ∗) + · (ψ ψ∗ ∂t ρKG (r. t) ) . t) ] 2mi (10. Under Lorentz transformations the free particle Klein–Gordon equation (10. t) ψ ∗ (r. In order to obtain the conservation law connected with the Klein–Gordon equation (10.153) is associated with a conservation law for particle probability ∂t ρS (r.157) 2m which is equivalent to (10. t) ψ(r.154). (10. t)∂t ψ ∗ (r. t) = ψ ∗ (r.158) which yields 2 2 ψ ∗ ∂t ψ − ψ∂t ψ ∗ − ψ ∗ 2ψ +ψ ψ∗ − 2ψ∗ = ψ) = 0 (10. t) 1 [ ψ ∗ (r. t) + where ρKG (r.156) · jS (r. t) jKG (r.10.5: Klein–Gordon Equation or 2 ∂t − 2 305 + m2 ψ(xµ ) = 0 . t) − ψ(r. t) + where ρS (r. t) = 0 i ( ψ ∗ (r.161) . 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) ψ ∗ (r.159) (10.151) One can notice immeadiately that (10. t) − ψ(r. t) = · jKG (r.150) is invariant under Lorentz transformations.152) which has the same form as the Klein–Gordon equation in the original frame. t) ) 2m 1 = ( ψ ∗ (r. t) = (10. t) (10.150) one considers Im ψ ∗ ∂µ ∂ µ + m2 ψ = 0 (10. and that (as postulated) ψ(xµ ) is a scalar. t) = − 1 2m 2 ψ(r.154) describes the current density connected with motion of the particle probability distribution. 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) = 0 (10. This follows from the fact that ∂µ ∂ µ and m2 are scalars. t) ψ(r.150) becomes ∂µ ∂ µ + m2 ψ (x µ ) = 0 (10.

e.168) .166) ψo (p. m 2 2 E0 = m2 + po .93] dispersion relationship connecting E0 .306 Relativistic Quantum Mechanics This conservation law differs in one important aspect from that of the Schr¨dinger equation (10. and of negative energies E ≤ −m. λ) actually describes charge density rather than probability. for example pions. i. ±) = ± 2 m2 + po (10.167) m which is positive for λ = + and negative for λ = −. λ) = λ ψo (p. The density ρKG (p. 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.164) The corresponding energy is Eo (po . in that the expression for ρKG is not positive definite.151) yields 2 2 Eo − p0 − m2 ψ(r. Today. We denote this by summarizing the solutions as follows ∂µ ∂ µ + m2 ψo (p. λ) associated with the corresponding wave functions ψo (p. (10. λ|xµ ) according to (10. the anti-particles carry a charge opposite to that of the associated particles. not of particle probability. λ|xµ ) = N e−iλmt .150) are actually determined by po and by the choice of sign ±. The proper interpretation of the two cases is that the Klein–Gordon equation describes particles as well as anti-particles. corresponding to λ = −. λ|xµ ) = 0 2 m2 + po . corresponding to λ = +.166) is Eo (p) ∗ ρKG (p. with p = 0.162) shows that the solutions of the free particle Klein-Gordon equation (10. t).150) is a continuum of positive energies E ≥ m. the Klein-Gordon equation is considered as a suitable equation to describe spin–0 particles. λ|xµ ) (10.154). p0 . t) = 0 (10.p ei(p·r − λEo (p)t) Eo (p) = The spectrum of the Klein–Gordon equation (10. according to (??) is decribed by the r–independent wave function ψo (p = 0. λ = ± .162) Inserting this into the Klein–Gordon equation (10. λ|xµ )ψo (p. Generating a Solution Through Lorentz Transformation A particle at rest. is actually that of a charge density.163) which results in the expected [see (10.165) This result together with (10. The proper interpretation of ρKG (r. 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) . (10. λ = ± (10. (10..161) and (10. λ|xµ ) = Nλ. it had been realized later. o namely. and the density ρKG (p.

166)] of the particle. (10.173) (10.175) as the momentum of the particle and one obtains L(w3 )ψo (p = 0. t)) . t)) (10.172) as the energy [c.169) −iλ ( m coshw3 ) ( R coshΩ ) − iλ ( m sinhw3 ) ( R sinhΩ ) .h.171) One can interpret then for λ = +. of (10.166).s. A(r.170) The coordinate transformation (10. (10.174) which agrees with the expression given in (10.. for the wave function (10. L(w3 )ψo (p = 0. λ = +|xµ ) = N ei(px 3 − Eo (p)x0 (10. Ω. 2 2 1−v 1−v (10.138) to replace t = x0 by hyperbolic coordinates R. one has to interprete p = mv/ 1 − v 2 (10. for positive energy solutions.61) yield for this expression −iλ √ m mv x0 − iλ √ x3 . λ|xµ ) = = exp w3 ∂ ∂Ω N e−iλmRcoshΩ 3) N e−iλmRcosh(Ω+w . t). Such wave function should be generated by applying the Lorentz transformation in the function space representation p (10.168).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. Aµ (xµ ) = (V (r. λ = −|xµ ) = N ei(px 3 + Eo (p)x0 .145) choosing m = sinhw3 . In case of λ = −.e. using (10.. For this purpose we consider the wave function for a particle moving with momentum velocity v in the direction of the x3 –axis. −A(r.177) . (10. i.10.e. t). for negative energy solutions. This yields. In case of λ = + one obtains finally L(w3 )ψo (p = 0. (10.176) 10. p = − mv/ 1 − v 2 as the momentum of the particle relative to the moving frame and √ m = 1 − v2 m2 = 1 − v2 m2 + m2 v 2 = 1 − v2 m2 + p2 = Eo (p) (10.138) and the relationships (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.f. i.168). in fact.169) The addition theorem of hyperbolic functions cosh(Ω + w3 ) = coshΩ coshw3 + sinhΩ sinhw3 allows us to rewrite the exponent on the r.

in keeping withnthe non-relativistic limit.150) −g µν (i∂µ )(i∂ν ) + m2 ψ(xµ ) = 0 . According to the principle of minimal coupling [see (10.180) + m2 ψ(r. albeit in a form. For this purpose one writes the Klein-Gordon equation (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 − .178) According to the replacements in Table 10.. t) = −i − q A(r. we define ψ(r. According to the so-called minimum coupling principle the presence of the field is accounted for by altering energy.e.147). t) 2 (10. t).e.150). (10. t) . For this purpose we split-off a factor exp(−imt) which describes the oscillations of the wave function due to the rest energy. that the mass m of the particle. and focus on the remaining part of the wave function. m2 ) ψ(xµ ) = 0 . i∂t and −i in (10. 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. A) free quantum particle quantum particle in field (V. t))2 ψ(r. i. A) or Aµ = (V. in . which couples a particle of charge q to an electromagnetic field described through the potential Aµ (xν ).93).182) and seek an equation for Ψ(r. In terms of space-time derivatives this reads (i∂t − qV (r. (10. momenta for classical particles and the respective operators for quantum mechanical particles in the manner shown. t) = e−imt Ψ(r. See also Eq. is much larger than all other energy terms.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.. t) . (10. i.1.. i.e. (10. it’s rest energy. We will also assume.1: Coupling of a particle of charge q to an electromagnetic field described by the 4-vector potential Aµ = (V. according to the rules shown in Table 10. 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 (10.308 free classical particle classical particle in field (V.179) (10. −A).69)] one replaces the quantum mechanical operators.

(10. | i∂t Ψ | << m . π − mesons are generated through inelastic proton–proton scattering p + p −→ p + p + π − + π + . 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− .181) can then be approximated as follows: (i∂t − qV )2 e−imt Ψ = = (i∂t − qV ) (me−imt Ψ + e−imt i∂t Ψ − qV e−imt Ψ) m2 e−imt Ψ + me−imt i∂t Ψ − qV e−imt Ψ +me−imt i∂t Ψ − e−imt ∂ 2 Ψ − qV e−imt i∂t Ψ −me−imt qV Ψ − e−imt i∂t qV Ψ + q 2 V 2 e−imt Ψ ≈ m2 e−imt Ψ − 2mqV e−imt Ψ − 2me−imt i∂t Ψ (10. (10. This application demonstrates that the Klein–Gordon equation describes spin zero particles. of (10. e. Pionic Atoms To apply the Klein–Gordon equation (10.185) This is.e.183) The term on the l. t) = ˆ [p − q A(r.183)..h.10.181) reads then i ∂t Ψ(r. i.. t) ≡ 0 r + Ze2 r 2 + 2 − m2 π ϕ(r ) = 0 .6: Klein–Gordon Equation with Electromagnetic Field particular. identical to the Schr¨dinger equation (10.181) to a physical system we consider pionic atoms. t) 2m Ψ(r.184) where we neglected all terms which did not contain factors m.188) Inserting this into (10. i. t) (10.. atoms in which one or more electrons are replaced by π − mesons. (10.g. however.186) then are slowed down.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.e.2) of a non-relativistic spin-0 particle o moving in an electromagnetic field. To ‘manufacture’ pionic atoms. The Klein-Gordon equation (10. larger than |i∂t Ψ/Ψ| and alrger than qV . t) = − Ze and A(r. A stationary state of the Klein–Gordon equation is described by a wave function ψ(xµ ) = ϕ(r ) e−i t . Ψ |q V | << m . spin-0 mesons.181) yields (we assume now that the Klein–Gordon equation describes a 2 particle with mass mπ and charge −e) for qV (r. filtered out of the beam and finally fall as slow pions onto elements for which a pionic variant is to be studied. 309 (10. t)]2 + qV (r. (10.189) .s. The approximation is justified on the ground of the inequalities (10. The process of π − meson capture involves the so-called Auger effect.

φ) ..194) and (10. φ) (10.195) The latter problem leads to the well-known spectrum En = − mπ (Ze2 )2 .192) leads then to the ordinary differential equation d2 − dr2 ( + 1) − Z 2 e4 2 Ze2 + + r2 r 2 − m2 π R (r) = 0 .196) In this expression the number n defined through n = n− −1 (10. φ) = ( + 1) Y m (θ. .e.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 . n = 1. The Laplacian is 2 = ˆ 1 2 1 1 1 2 L2 2 ∂r r + 2 2 ∂θ sinθ∂θ + 2 2 ∂φ = ∂r r − 2 . . . . this quantity definitely must be an integer. 2. (10.197) counts the number of nodes of the wave function. 1. (10.198) . φ) R (r) Y r m (θ. the eigenfunctions of the operator L2 ˆ L2 Y m (θ. θ.191) m (θ. i.e. . (10. i.192) ˆ are spherical harmonics. (10. (10. .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 ) (10.194) 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 (10.310 Relativistic Quantum Mechanics Because of the radial symmetry of the Coulomb potential we express this equation in terms of spherical coordinates r. .191) ˆ The operator L2 in this equation suggests to choose a solution of the type ϕ(r ) = where the functions Y in (10. φ. The similarity of (10.199) ( + 1) − Z 2 e4 .193) (10. n − 1 . 2n2 = 0.

EKG = Using (10. 2 e2 −→ e2 mπ . n − 1 m = − . . (10.204) reads 1 (10..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 ). . .10. . (10. . 10. results in the (10. i. . . 1. . .6: Klein–Gordon Equation with Electromagnetic Field and one can write (10.199) and defining EKG (n.202) Solving this for state) yields (choosing the root which renders = 1 + mπ Z 2 e4 ( n + λ( )+1 )2 ≤ mπ . which corresponds to a bound = 0. . 2mπ 2 (n + λ( ) + 1)2 (10.194) d2 λ( ) ( λ( ) + 1 ) 2 Ze2 − + + dr2 r2 r 2 311 − m2 π R (r) = 0 . . Introducing α = Z 2 e4 and 1 β = + 2 (10.203) spectrum . 2mπ 2 −→ λ( ) . 1. . 2.205) 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 ) . = 0.e. 1. . . . m) = mπ 1 + Z 2 e4 (n− − 1 2 + ( + 1 )2 − Z 2 e4 )2 2 n = 1. . − + 1. .196) if one makes the replacement 2 values which can be obtained E −→ One obtains − m2 π . n = 0. .201) mπ Z 2 e4 m2 − m2 π π = − . + (10. .197.200) The bound state solutions of this equation should correspond to from (10. .

e. arises because the Klein–Gordon equation. Many attempts were made by theoretical physicists to ‘linearize’ the square root operator in (10. the Klein–Gordon equation had been rejected since it did not yield a positive-definite probability density. is related to the equation E 2 = m2 + p 2 of the classical theory which involves a term E 2 . Dirac succeeded. a feature which is connected with the 2nd order time derivative in this equation. a splitting of the six n = 2.209). i. 2n2 2βn3 8n4 4n4 (10. 10. His solution . Also. the second term the non-relativistic energy. Finally. one with a positive-definite density. Equations (10. in fact.208. It must be pointed out here that does not denote energy. but not an operator iγ · as suggested by the principle of relativity (equivalence of space and time). t) (10. but for a long time to no avail.e. The latter term agrees with observations of pionic atoms. However. The latter spectrum shows. however. would have only a first order time derivative.312 ≈ From this results for (10.208. through the correspondence principle.209) which.208).2 particles. t) = ± m2 − 2 Ψ(r.146) leads to the wave equation i∂t Ψ(r. however. (10. is identical to the two equations (10. are unsatisfactory since expansion of the square root operator involves all powers of the Laplace operator. It should feature then a differential operator of the type D = iγ µ ∂µ . 4n (10. In order to describe electron spectra one must employ the Lorentz-invariant 1 wave equation for spin. the wave function changes sign under reflection. 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 . i.209).. and the third term gives the leading relativistic correction. . This derivative. 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.. 10.e. i. it does not agree with observations of the hydrogen spectrum. m) ≈ m − mZ 2 e4 mZ 4 e8 − 2n2 2n3 1 + − 1− Relativistic Quantum Mechanics α α2 α2 α2 − + + + O(α3 ) . the Dirac equation introduced below. Application of the correspondence principle (10. for example. A more satisfactory Lorentz–invariant wave equation. but in the present case rather the negative of the energy.207) Here the first term represents the rest energy..7 The Dirac Equation Historically. = 1 states into groups of two and four degenerate states. 10.204) EKG (n. in turn. the π − meson is a pseudoscalar particle.208) These two equation can be combined i∂t + m2 − 2 i∂t − m2 − 2 Ψ(r.206) 1 2 3 + O(Z 6 e12 ) . t) .

ψd (xµ ).208). Instead. In fact.211) into (10. 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. Properties of the Dirac Matrices Let us now trace Dirac’s steps in achieving the linearization of the ‘square root operator’ in (10.210) 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. (10. 2. the operator of the Klein–Gordon equation ∂µ ∂ µ + m2 = −( P + m ) ( P − m ) (10. However. one route to important discoveries in physics is the creation of new mathematical descriptions of nature. 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 ). exploited ∂µ ∂ν = ∂ν ∂µ . therefore. γ 1 . therefore. 3 are anti-hermitian . The discovery by Dirac.1 2 and anti-particles. . achieved through a beautiful mathematical theory. spin. (10.e. i.214) From this follows that γ 0 has real eigenvalues ±1 and γ j . but did not commute the. the quantities γ 0 . 3 has imaginary eigenvalues ±i.7: Dirac Equation 313 to the problem involved an ingenious step.212) where we have changed ‘dummy’ summation indices. according to (10.10. ψ2 (xµ ). γ 2 .213) reads (γ µ )2 = 1 µ = 0 −1 µ = 1.210) where P = iγ µ ∂µ . 2.211) Obviously. the realization that the linearization can be carried out only if one assumes a 4-dimensional representation of the coefficients γ µ . are associated with a positive-definite particle density. For µ = ν condition (10. j = 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 . so far. γ ν ]+ = 2 g µν (10. γ j . It turns out that no 4-vector of real or complex coefficients can satisfy these conditions. unspecified algebraic objects γ µ and γ ν .215) . 3 . these descriptions ultimately merging with the true theory of matter. (10. Starting point is to boldly factorize. this was not so. beautiful mathematical theory and that. the 4-dimensionsional representation discovered by Dirac involved new physical properties. Initially. it was assumed that the 4-dimensional space introduced by Dirac could be linked to 4vectors. j = 1. namely. Inserting (10.209). quantities with the transformation law (10. 2. Accordingly.213) We want to determine now the simplest algebraic realization of γ µ . yet to be discovered.. . strengthens the believe of many theoretical physicists today that the properties of physical matter ultimately derive from a.66). We seek to identify the coefficients γ µ . one can impose the condition γ0 is hermitian . .

222) and multiplication from the right by γ 0 yields the adjoint Dirac equation Ψ† (xµ ) γ 0 iγ µ ∂µ + m ← =0.. ψ3 . σ 2 .215).213) is satisfied.223) Inserting this into (10.e. 1 (10. namely the Pauli matrices σ 1 . however. This property can also be written (γ µ )† = γ 0 γ µ γ 0 . rather than to the right side. A proper choice is γ0 = 1 1 0 0 −1 1 . σ j σ k = − σ k σ j for j = k .224) . σ 3 for which. rather than a scalar wave function.218) of the Pauli matrices one can readily prove that condition (10. One can readily show using the hermitian property of the Pauli matrices (γ 0 )† = γ 0 and (γ j )† = −γ j for j = 1. (10.216) i. Relativistic Quantum Mechanics (10. The Adjoint Dirac Equation The adjoint equation is Ψ† (xµ ) i(γ µ )† ∂µ + m ← ← (10. ψ2 .219) Using property (10. is implied by (10. except for similarity trasnformations. is unique. ψ4 ) and where ∂µ denotes the differential operator ∂ µ operating to the left side. (10.218) The Pauli matrices allow one.213) reads γ µ γ ν = −γ ν γ µ . the γ µ are anti-commuting. as long as det(γ µ ) = 0 the dimension d of the square matrices γ µ must be even so that (−1)d = 1.220) where Ψ(xµ ) represents a 4-dimensional wave function. We will argue further below that the choice f γ µ . 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 ) .314 For µ = ν (10. in fact. in fact. (10. 3 which. 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 (10.217) Obviously. 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. For d = 2 there exist only three anti-commuting matrices. to construct four matrices γ µ for the next possible dimension d = 4. γj = 0 −σ j σj 0 .222) ∗ ∗ ∗ ∗ where we have defined Ψ† = (ψ1 . holds 2 σj = 1 . 2.221) =0 (10. (10.

2.232) This form of the Dirac equation is called the Schr¨dinger form since it can be written in analogy o to the time-dependent Schr¨dinger equation o i∂t Ψ(xµ ) = Ho Ψ(xµ ) .7: Dirac Equation Similarity Transformations of the Dirac Equation .219) is the socalled chiral representation defined through ˜ Ψ(xµ ) = S Ψ(xµ ) . ˆ ˆ ˆ Ho = α · p + β m . 3 and ˆ β = 1 1 0 0 −1 1 .233) The eigenstates and eigenvalues of H correspond to the stationary states and energies of the particles described by the Dirac equation.e.228) (10.231) ˆ whereα has the three components αj .10.230) Schr¨dinger Form of the Dirac Equation o Another form in which the Dirac equation is used often results from multiplying (10.229) The similarity transformation (10.Chiral Representation 315 The Dirac equation can be subject to any similarity transformation defined through a non-singular ˜ 4 × 4–matrix S. j = 1.226) (10. t) = 0 (10. 10. (10. j = 1.227) A representation often adopted beside the one given by (10. 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 ˜ (10.227) leaves the algebra of the Dirac matrices unaffected and commutation property (10. γj = ˜ 0 σj 1 S = √ 2 −σ j 0 1 1 1 1 1 −1 1 1 (10. γ ˜ Exercise 10.225) . 3 . 3 .221) from the left by γ 0 ˆ ˆ i ∂t + i α · − β m Ψ(r. 2. i.1: Derive (refeq:Dirac-intro20a) from (10.7. [˜ µ .. j = 1. (10.227).213) still holds. (10. (10.213).222. (10. γ ν ]+ = 2 g µν . . 2. αj = ˆ 0 σj σj 0 . and γ0 = ˜ 0 1 1 1 0 1 .

any product 1 of the dj ’s can be brought to the form (10.e. 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 . Γ±14 = ±d1 d3 d4 .236) where ‘⊗’ denotes the Kronecker product between matrices.235) and. d2 = . Γ±4 = ±d3 .Good. There are (including the different signs) 32 elements in G4 which we define as follows Γ±1 = ±1 1 Γ±2 = ±d1 .. i. Rev. the matrix elements of C = A ⊗ B are Cjk. To complete the proof in this section the reader may consult Miller ‘Symmetry Groups and their Application’ Chapter 9.237) which are the ordered products of the operators ±1 and d1 . The associative algebra generated by d1 . Γ±9 = ±d2 d3 Γ±10 = ±d2 d4 . This property extends then to 4 × 4–matrices which obey (10. d4 = 1 0 0 −1 1 0 0 −1 0 1 1 0 0 i −i 0 (10. The representations of Clifford algebras Cm are well established.6 and R. form a Clifford algebra C3 .235). d2 .235) as can be readily verified from (10.e. to Dirac matrices. (1955). The representations of this group are given by a set of 32 4 × 4–matrices which are equivalent with respect to similarity transformations. satisfy the anti-commutation property dj dk + dk dj = 2 for j = k 0 for j = k Relativistic Quantum Mechanics d4 = γ 0 (10. The three Pauli matrices also obey the property (10. Γ±13 = ±d1 d2 d4 .Phys. m = Aj Bkm . Γ±8 = ±d1 d4 . in case of C4 .235) can differ only with respect to unitary similarity transformations. j = 1.. 3 . Since the Γj are hermitian the similarity transformations are actually given in terms of unitary transformations. For example.316 Clifford Algebra and Dirac Matrices The matrices defined through dj = iγ j .213). Obviously. .234) (10. The Clifford algebra C4 entails a subgroup G4 of elements ± dj1 dj2 · · · djs . Γ±3 = ±d2 .Mod. 27. hence. Γ±5 = ±d4 Γ±6 = ±d1 d2 .238) These elements form a group since obviously any product of two Γr ’s can be expressed in terms of a third Γr .213). j1 < j2 < · · · < js s ≤ 4 (10.237) by means of the property (10. Γ±15 = ±d2 d3 d4 Γ±16 = ±d1 d2 d3 d4 (10. page 187. Γ±7 = ±d1 d3 . d4 is called a Clifford algebra C4 . d4 . d3 . i. . 2. Γ±11 = ±d3 d4 Γ±12 = ±d1 d2 d3 . The reader may . One can conclude then that also any set of 4 × 4–matrices obeying (10.H.

and derivatives according to (10.. Exercise 7. The transformation is assumed to be linear and of the form Ψ (x ) = S(Lµ ν ) Ψ(xµ ) µ µ (10. in a moving frame should hold iγ ∂µ − m Ψ (x ) = 0 ..236) with the Dirac representation (10. The latter transformations imply that coordinates transform according to (10.80). However. in frames connected by Lorentz transformations.8 Lorentz Invariance of the Dirac Equation We want to show now that the Dirac equation is invariant under Lorentz transformations. Multiplication and summation of (10. transform coordinates and derivatives of the Dirac equation.6). the form of the Dirac equation is identical in equivalent frames of reference.215). x = Lµ ν xν . in fact.214) follows that γ 0 has real eigenvalues ±1 and can be represented by a hermitian matrix.213.e.e. exist.e.6).4: Show that from (10.218) of the Dirac matrices γ µ . Exercise 7. 10.3: Demonstrate the anti-commutation relationships (10.76) yields ∂ρ = Lν ρ ∂ ν . i. 10. in a moving frame holds iγ µ ∂µ − m Ψ (x ) = 0 . 2.80) by Lµ ρ and using (10. µ µ (10. therefore.. Exercise 7. these properties determine the matrices γ µ uniquely.218) of the Pauli matrices σ j .e.. (10.240) .241) . and γ j .10.219). i. Infinitesimal Bispinor State Transformation We want to show now that a suitable transformation of Ψ(xµ ) does. i. i. We can.e. Lorentz Transformation of the Bispinor State Actually.. we do not know yet how to transform the 4-dimensional wave function Ψ and the Dirac matrices γ µ . Hence. a result one could have also obtained by applying the chain rule to (10.2: Demonstrate the anti-commutation relationships (10. 3 has imaginary eigenvalues ±i and can be represented by an anti-hermitian matrix. it must hold γ µ = γ µ . x µ = Lµ ν xν (10.239) Form invariance implies that the matrices γ µ should have the same properties as γ µ . 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. Except for a similarity transformation. i.8: Lorentz Invariance of the Dirac Equation 317 also want to establish the unitary transformation which connects the Dirac matrices in the form (10. namely. j = 1.

s.249) from which we can conclude σµν = −σµν νµ = −σνµ i. νµ (10. Multiplication of this property by g from the right yields ( g)T = − g.243) in a form in which the Lorentz transformation in the form Lµν is on the r. (10.250) σµν = −σνµ . of the equation. it must hold (10.243) We want to determine now the 4 × 4–matrix S(Lη ξ ) which satisfies this condition.245) The construction of S(Lη ξ ) will proceed using the avenue of infinitesimal transformations. however. Multiplication of (10. follows S(Lη ξ )γρ S −1 (Lη ξ ) = Lσ ρ γσ . µ g ρν = µν and. We had introduced in (10. i.241) implies a similarity transformation Sγ µ S −1 .. using gρµ γ µ = γρ . It holds then S( µν i ) = 1 − σµν 1 4 µν µν = S(− νµ i ) = 1 + σµν 1 4 µν . For this purpose we exploit (10.h. The infinitesimal transformation S(Lρ σ ) which corresponds to (10.242) 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 .243) from the left by Lσ ρ gσν yields S(Lη ξ )γ µ S −1 (Lη ξ ) gρµ = Lσ ρ gσν γσ . hence.318 Relativistic Quantum Mechanics where S(Lµ ν ) is a non-singular 4 × 4–matrix. state that the Dirac equation (10.244) from which.248) Here σµν denote 4 × 4–matrices operating in the 4-dimensional space of the wave functions Ψ. hence.e. The elements of g are. µ (10. One can finally conclude multiplying both sides by g ρµ S(Lη ξ )γ µ S −1 (Lη ξ ) = Lν µ γν .247) the infinitesimal matrix µν is anti-symmetric. g is an anti-symmetric matrix.. in the expression of the infinitesimal transformation ρ Lµν = g µν + µν (10.12) in the form Lν µ gνσ Lσ ρ = gµρ = gρµ .e. S( µν ) should not change its value if one replaces in its argument µν by − νµ . The proper starting point for a constructiuon of S(Lη ξ ) is actually (10. (10. the transformation (10.221) 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. (10. One can. 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. Ob1 viously. .246) (10.38) the infinitesimal Lorentz transformations in the form Lµ ν = δ µ ν + µ ν where the infinitesimal operator µ ν obeyed T = − g g.

this condition can actually be expressed through six independent conditions. namely.257) .251) S −1 ( µν ) = 1 + σµν µν 1 4 Inserting (10. a solution of condition (10. 10. For this purpose one needs to express formally the r.254) is σαβ = i [ γα .251) into (10. like the expression on the l.h.e.228.8: Lorentz Invariance of the Dirac Equation 319 One can readily show expanding SS −1 = 1 to first order in µν that for the inverse infinitesimal 1 transformation holds i (10. the chiral representation introduced above in Eqs. γj = ˜ 0 −σ j σj 0 . (10. (10.254).s. j = 1.h. γβ ]− 2 (10.s. γ µ ]− = 2i ( δ µ β γα − δ µ α γβ ) .s.252) Since six of the coefficients αβ can be chosen independently.254) The proper σαβ must be anti-symmetric in the indices α.255) satisfy condition (10..h. 2. Exercise 7. β and operate in the same space as the Dirac matrices.253) 1 2 αβ ( δ µ β γα − δ µ α γβ ) . In this representation the Dirac matrices γµ = (˜ 0 .229). i. For this purpose we need to verify that the algebra of the generators involving addition and multiplication is closed. β [ σαβ .248. of (10.255) which can be demonstrated using the properties (10. the expression on the r.10.256) One can readily verify that the non-vanishing generators σµν are given by (note σµν = −˜νµ . 10.h.252) also as a sum over both indices of αβ . σjk = [˜j ..s. of (10. (10.252) results in the condition for each α.246) results then in a condition for the generators σµν − i ( σαβ γ µ − γ µ σαβ ) 4 αβ = νµ γν .216) of the Dirac matrices. For this purpose we inspect the properties of the generators in a particular representation. Algebra of Generators of Bispinor Transformation We want to construct the bispinor Lorentz transformation by exponentiating the generators σµν . −γ ) are ˜ γ ˜ γ0 = ˜ 0 1 1 1 0 1 . must be symmetric with respect to interchange of the indices α and β. (10. (10. ˜ ˜ σ only six generators need to be determined) σ0j = ˜ i [˜0 . Furthermore. γj ] = γ ˜ 2 −iσ j 0 0 iσ j .213. 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. In fact. 10. Comparing this with the l. γk ] = ˜ γ ˜ jk σ 0 0 σ .5: Show that the σαβ defined through (10. 3 .

248) allows us to write the transformation S for any. i. One should. In fact. therefore.262) This expression allows one to transform according to (10. be able to state S in terms of the same parameters w and ϑ as the Lorentz transformation in (10.51). any representation of the generators.255) yields a closed algebra. ϑ) = e 2 (w − iϑ)·σ 0 − 1 (w + iϑ)·σ 0 e 2 1 (10. therefore. Finite Bispinor Transformation The closedness of the algebra of the generators σµν defined through (10. 2 0 −(w + iϑ) · σ 13 + σ23 ˜ 23 ) (10. and since the algebra of the Pauli matrices is closed.320 Relativistic Quantum Mechanics Obviously.259) We had stated before that the transformation S is actually determined through the Lorentz transformation Lµ ν . one can express the tensor µν through w and ϑ using µν = µ ρ g ρν and the expression (10.259) becomes in the chiral representation ˜ S(w.44)  0 −w1 −w2 −w3  w 0 ϑ3 −ϑ2   =  1  w2 −ϑ3 0 ϑ1  w3 ϑ2 −ϑ1 0  µν (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 = . We can finally note that the closedness of the algebra of the generators σµν is not affected by similarity transformations and that. (10. in particular.259) yields the desired connection between the Lorentz transformation (10. since both operations convert block-diagonal operators A 0 0 B (10. the representation (10. not necessarily infinitesimal.e. In order to construct an explicit expression of S in terms of w and ϑ we employ again the chiral representation.261) We note that this operator is block-diagonal. µν in the exponential form i S = exp − σµν 4 µν .241) bispinor wave functions from one frame of reference into another frame of reference. the algebra of these generators is closed under addition and multiplication. Since such operator does not change its block-diagonal form upon exponentiation the bispinor transformation (10.258) again into block-diagonal operators.51) and S. .260) Inserting this into (10..

Multiplying (10. The reason is that the r.270) We conclude that (10. (10. Since ∂µ transforms like a covariant 4-vector.268) We will prove this property in the chiral representation. j) = Ψ† (xµ )γ 0 γ µ Ψ(xµ ) . (10.266) ρ(xµ ) = Ψ† (xµ )Ψ(xµ ) = s=1 |ψs (xµ )|2 (10.263) obviously is a scalar under Lorentz transformations and that the left hand side must then also transform like a scalar.221) and the adjoint Dirac equation (10. For this purpose we prove first the relationship S −1 = γ 0 S † γ 0 . the conservation law (10.h.10. j µ must transform like a contravariant 4-vector.221) from the left by Ψ† (xµ )γ 0 . (10. and addition yields Ψ† (xµ )γ 0 The last result can be written ∂µ Ψ† (xν )γ 0 γ µ Ψ(xν ) = 0 .s. Obviously. The time-like component ρ of j µ 4 iγ µ ∂µ + iγ µ ∂µ ← Ψ(xµ ) = 0 .e. of (10.224) from the right by Ψ(xµ ). This transformation behaviour can also be deduced from the transformation properties of the bispinor wave function Ψ(xµ ). i.224). −ϑ) = e− 2 (w − iϑ)·σ 0 1 (w + iϑ)·σ 0 e2 1 (10. in fact. (10.268) holds for the bispinor Lorentz transformation. 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 .267) has the desired property of being positive definite. We will now determine the relationship between the flux j µ = Ψ (x )γ 0 γ µ Ψ (x ) † µ (10.. ϑ) = S(−w. the property applies then in any representation of S.269) One can readily show that the same operator is obtained evaluating γ S (w.263) allows one to conclude that j µ must transform like a contravariant 4-vector as the notation implies. (10.271) .263) does hold. The conservation law (10. For our proof we note first ˜ ˜ S −1 (w.265) (10.8: 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 .264) (10.263) Starting point are the Dirac equation in the form (10. for j µ (xµ ) = (ρ.

As pointed out.274) 10. The operators commute also with Ho in (10.268) µ j = Ψ† (xµ )S † γ 0 γ µ SΨ(xµ ) = Ψ† (xµ )γ 0 S −1 γ µ SΨ(xµ ) .232. µ (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. One obtains using (10. In the following we want to provide this characterization.246) S −1 (Lη ξ )γ µ S ( Lη ξ ) = (L−1 )νµ γν = (L−1 )ν γ ν = Lµ ν γν . i. of which the wave functions are eigenfunctions as well.272) With S −1 (Lη ξ ) = S((L−1 )η ξ ) one can restate (10.273) where we have employed (10. (10. not only as the most simple demonstration of the theory. The additional degrees of freedom described by the four components of the bispinor wave function require. 10. Like for the free particle wave functions of the non-relativistic Schr¨dinger and the Klein–Gordon o equations one expects that the space–time dependence is governed by a factor exp[i(p · r − t)]. (10. Note that we have assumed in (10.275) The free particle wave function is an eigenfunction of Ho . as is required for the mentioned property. 10.76). In case of non-relativistic quantum mechanics the free particle wave function has a single component ψ(r.271) that the Dirac matrices are independent of the frame of reference.272) results in the expected transformation behaviour µ j = Lµ ν j ν . a property which leads to the energy– momentum (dispersion) relationship of the Dirac particle.282. only two degrees of freedom of the bispinor four degrees of freedom are independent [c. (10. 10..283)]. additional characterizations. orthonormal basis and allows one to quantize the Dirac field Ψ(xµ ) much like the classical electromagnetic field is quantized through creation and annihilation operators representing free electromagnetic waves of fixed momentum and frequency.233). bispinor wave functions and we need to determine corresponding components of the wave function. the wave function has four components which invite further characterization of the free particle state.f. specific for the Dirac free particle. Combining this with (10. The solutions provide also a complete.9 Solutions of the Free Particle Dirac Equation We want to determine now the wave functions of free particles described by the Dirac equation. These operaˆ tors. however. commute with each other. but also as providing a basis in which the wave functions of interacting particle systems can be expanded and characterized. We will start from the Dirac equation in the Schr¨dinger form (10. (10.e.322 Relativistic Quantum Mechanics in a moving frame of reference and the flux j µ in a frame at rest. As it turns out. the Dirac particles are described by 4-dimensional. t) and is determined through the momentum p ∈ R3 . In relativistic quantum mechanics a Dirac particle can also be characterized through a momentum.231. For this purpose we consider . the identification of observables and their quantum mechanical operators.233) o i∂t Ψ(xµ ) = Ho Ψ(xµ ) . as just mentioned. Like in non-relativistic quantum mechanics the free particle wave function plays a central role.

234) and (10. (10.280) which has a positive and a negative solution = ± E(p) . In fact. like the Klein–Gordon equation.10. and φo .279) According to the property (5.284) The relationships (10. (10.281) Obviously. 10.277) explicitly ( − m) 1 φo − σ · p χo 1 − σ · p φo + ( + m) 1 χo 1 = 0 = 0.278) 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 . +m (10. namely φo and χo are related as follows φ0 χo = = σ·p χ0 −m σ·p φ0 . E(p) = m2 + p 2 . the socalled helicity of the particle.283) where is defined in (10. reproduce the classical relativistic energy–momentum relationships (10. 10. from (10.282) (10.283) imply that the bispinor part of the wave function allows only two degrees of freedom to be chosen independently. χo each represent a constant.281). two-dimensional spinor state. later to be identified with momentum and energy.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. hence.278) provides us with information about the components of the bispinor wave function (10. We want to show now that these degrees of freedom correspond to a spin-like property. 10. one finds using (5.279) the well-known relativistic dispersion relationship 2 = m2 + p 2 (10. These two relationships are consistent with each other.276). 2 − m2 0 (10.93. 1 (10.234) of Pauli matrices holds (σ · p)2 = p 2 1 One can. .282. Inserting (10.276) into (10.276) where p and together represent four real constants.94) Equation (10. the Dirac equation.280) σ·p (σ · p)2 φ0 = χ0 = +m ( + m) ( + m) p2 χ = χ0 . (10.277) To solve this problem we write (10. conclude 1.231.232) leads to the 4-dimensional eigenvalue problem m σ·p σ · p −m φo χo = φo χo .

287. ±) denote the solution of a free particle moving with momentum p in the laboratory frame.290) should differ. 10. +) = −1 . through χo given by χo = u1 u2 = u ∈ C2 .290).h. (10. u† u = |u1 |2 + |u2 |2 = 1 .268). = +E(p) . +|xµ ) = N+ (p) u σ·p E(p) + m u ei(p·r − t) .. S † = γ 0 S −1 γ 0 .287. (10.259). (10. ±) γ 0 S −1 γ 0 γ 0 S Ψ(0.285)]. Employing (10. (10. (10.286) Here N+ (p) is a constant which will be chosen to satisfy the normalization condition Ψ† (p. For the positive energy solution. 10. i. is Ψ(p. according to (10.e. ±) = = = Ψ† (0.285) The corresponding free Dirac particle is then described through the wave function Ψ(p. Hence. the solution for = +E(p).286) for p = 0 yields.287) the form of which will be justified further below. The form of this condition and of (10.287) will be justified now. is invariant under Lorentz transformations.292) .e.f. (10. ±). +) = |N+ (0)|2 (u† .s. ±)γ 0 Ψ(p. One can see this as follows: Let Ψ(p. 0) γ 0 u 0 = |N+ (0)|2 . we present φo through the normalized vector φo = u1 u2 = u ∈ C2 ..219) and u† u = 1 [c.241). we demonstrate that the product Ψ† (p. ±) γ 0 Ψ(0. u† u = |u1 |2 + |u2 |2 = 1 . The connection between the solutions. S −1 = γ 0 S † γ 0 and. +) = 1 . +) γ 0 Ψ† (p. the solution for = −E(p).290) which differs from the normalization condition (10.h. (10. ±) S † γ 0 S Ψ(0.281)] separately. ±) = S Ψ(0.289) Here N− (p) is a constant which will be chosen to satisfy the normalization condition Ψ† (p. Similarly. ±) γ 0 Ψ(p. according to (10.324 Relativistic Quantum Mechanics For our further characterization we will deal with the positive and negative energy solutions [cf.s. (10. and let Ψ(0. ±) denote the corresponding solution of a particle in its rest frame.e. +) γ 0 Ψ† (p. i. where S is given by (10.s. = −E(p) . the l. of (10. of (10. ±) . ±) Ψ† (0. (10. i. we present the negative energy solution.287) in the minus sign on the r.288) corresponding to the wave function Ψ(p.. Ψ† (0. Ψ† (p. (10. −|xµ ) = N− (p) −σ·p E(p) + m u u ei(p·r − t) . We want to demonstrate now that the signs on the r. ±) Ψ† (0. For this purpose we consider first the positive energy solution. ±). hence.291) Note that we have used that. +) γ 0 Ψ(0.h. using γ 0 as given in (10. First.

σ·p u∗ E(p) + m T γo u σ·p E(p) + m u = 1 (10.297) 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.h.294) We want to determine now N± (p) for arbitrary p.286).9: Solutions of the Free Particle Dirac Equation The same calculation for the negative energy wave function as given in (10.299) ( m + E(p) )2 .286) is 2 N+ (p) (u∗ )T .295) Evaluating the l. (10. 325 (10.300) (10.234)] and using the normalization of u in (10. Exercise 7. −) = |N− (0)|2 (0. 2m (10.298) becomes N+ (p) = m + E(p) . u† ) γ 0 0 u = − |N− (0)|2 . Condition (10.302) .301) yields the normalization coefficient N+ (p) = m + E(p) .285) results in 2 N+ (p) 1 − p2 (E(p) + m)2 = 1 (10.219) yields 2 N+ (p) u† u − u† (σ · p)2 u (E(p) + m)2 = 1. (10. 2 E(p) (10. (5.6: Show that the normalization condition 2 N+ (p) (u∗ )T . We consider first the positive energy solution.296) Replacing (σ · p)2 by p 2 [c.h.287) written explicitly using (10.293) Obviously. this requires the choice of a negative side on the r. We can also conclude from our derivation N± (0) = 1 .10. of (10. ( m + E(p) )2 − p 2 (10.298) This result completes the expression for the wave function (10. using γ 0 as given in (10.s. −) γ 0 Ψ(0.289) yields Ψ† (0. σ·p u∗ E(p) + m T u σ·p E(p) + m u = 1 (10.f.290) to assign a positive value to |N− (0)|2 .s.

j = 1. λ|xµ ) = λ E(p) Ψ(p.306) i. i.s. exp[i(p · r − t)] = exp(ipµ xµ ). 3 is fairly obvious.e. conclude N− (p) = m + E(p) 2m (10.286. λ|xµ ) (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. 10. (10.289).300) have been constructed to satisfy the free particle Dirac equation (10. Inserting (10.. have completed the determination for the wave function (10.289) is T −σ·p −σ · p 2 E(p) + m u N− (p) u∗ . (u∗ )T γ o = −1 (10. We can. 10. defined as the component of the particle spin along the direction of motion. identical to the condition (10.307) where pµ = ( . i∂µ Ψ(p. This can be verified expressing the space–time factor of Ψ(p. Helicity The free Dirac particle wave functions (10. Rather than considering the ˆ observable (10. (10.309) . the two components of u indicate another degree of freedom which needs to be defined. the wave functions constructed represent eigenstates of Ho . thereby.303) E(p) + m u Evaluating the l. −p). This degree of freedom 1 describes a spin– 2 attribute. hence. The commutation property [σ · p. This attribute is the so-called helicity.e. λ|xµ ) in 4-vector notation. (10. The corresponding operator which measures this observable is 1 p ˆ Λ = σ· .275). however. pj ] = 0 . λ|xµ ) .304) This condition is. not a constant vector. i.286. The property [σ · p.289. The wave functions (10. The wave functions are also eigenstates of the momentum operator i∂µ .. Condition (10. 2.275) yields Ho Ψ(p.289) are not completely specified. 10.308) 2 |p| ˆ Note that p represents here an operator.286) into (10.296) for the normalization constant N+ (p) of the positive energy solution.h.290) written explicitly using (10.305) and.307) we investigate first the observable due to the simpler operator σ · p.e. yields 2 N− (p) u† (σ · p)2 u − u† u (E(p) + m)2 = −1 .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. λ|xµ ) = pµ Ψ(p. Ho ] = 0 follows from (10..326 Relativistic Quantum Mechanics We consider now the negative energy solution.

311) 2  −p  0 m + Ep 1 where e3 denotes the unit vector in the x3 -direction and where ˆ Ep = m2 + p2 .. According to the definition (5. t) = Np  e 2   1 p m + Ep 0   0   i(px3 − E t) 1 p e Ψ(pˆ3 . We consider first the simplest case that particles move along the x3 –direction. Ho and σ·p commute with each other and. 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 . States which are simultaneously eigenvectors of these three operators are also simulteneously eigenvectors of the three operators p.312). In this case Λ = 2 σ 3 .10. We assume first particles with positive energy. . p3 ). = −E(p). 327 0 (σ · p)2 ˆ 2 (σ · p) ˆ 0 0 (σ · p)2 ˆ 2 (σ · p) ˆ 0 (10.. Np = m + Ep . Since helicity is defined relative to the direction of motion of a particle the characterization of u as an eigenvector of the helicity operator. − 1 |r. i. can ˆ ˆ be simultaneously diagonal.308) above. 2 the wave functions which are eigenstates of the helicity operator. Therefore. i. − 1 |r. −. Ho and Λ defined in (10. +. in principle. t) = Np  e   1 0   0 p  m + Ep  i(px3 + E t) 1 p e Ψ(pˆ3 . +.e. 2m (10. 0. + 2 |r.e.312) We assume now particles with negative energy.224) 1 of σ3 the two u vectors (1. i. = +E(p).286) are eigenfunctions of Λ as well will specify now the vectors u. 0)T and (0. t) = Np  e (10. are   1   i(px3 − E t) 0 p e Ψ(pˆ3 . hence.310) We have shown altogether that the operators p. 1)T are eigenstates of 2 σ 3 with eigenvalues ± 1 ..313) 2   0 1 where Ep and Np are defined in (10. ˆ The condition that the wave functions (10.9: 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. t) = Np  e (10.e. is independent of the direction of motion of the particle. 1 p = (0. + 1 |r. −.

(10. The corresponding wave functions are determined through ˜ i˜ 0 ∂t − m Ψ(t) = 0 . t) = Np ei(p·r + Ep t) (10.312) can be obtained also by means of the Lorentz transformation (10. Generating Solutions Through Lorentz Transformation The solutions (10. t) 2 = Np 1 u(p.123).. 10. 2  1  0  (10. +. denoted by ˜.328 Relativistic Quantum Mechanics To obtain free particle wave functions for arbitrary directions of p one can employ the wave functions (10.314) (10. + 2 ) u(p.262) for the bispinor wave function and the transformation (10. For this purpose one starts from the solutions of the Dirac equation in the chiral representation (10. 10. [One can also express the rotation through Euler angles α.229). − ) 2 where ϑ(p) = = = exp exp i − ϑ(p) · σ 2 i − ϑ(p) · σ 2 1 0 0 1 .312). + 2 |r.317) Ψ(p.319) 1 Ψ(p. These eigenstates are obtained through a rotational transformation (5. −. − 2 ) u(p. − 2 ) −p 1 m + Ep u(p. γ and are  1  0  1 ˜ Ψ(p = 0.220).316) describes a rotation which aligns the x3 –axis with the direction of p. 0)T and (0. − 2 |r. + ) 2 1 u(p. 1)T have to be replaced by eigenstates u± (p) of the spin operator along the direction of p. 5. β. a wave function which represents free particles at rest.226.313) except that the states (1.e. − 1 ) 2 ei(p·r − Ep t) (10. for an r–independent wave function. +. p) e |p| (10. γ . + 2 ) ei(p·r − Ep t) (10. +.321) . 10. + 1 |r.311. 1 |t) = √2   e−imt .318) 1 Ψ(p. − 1 ) 2 −p 1 m + Ep u(p. − 1 |r. −.222.311.] The corresponding free particle wave functions are then Ψ(p. i.223) as follows 1 u(p. + 2 ) p 1 m + Ep u(p. in which case the transformation is given by (5. t) = N√ ei(p·r + Ep t) (10.315) e3 × p ˆ ∠(ˆ3 .320) where Ep and N are again given by (10. t) 2 = Np u(p. + 1 ) 2 p 1 m + Ep u(p.

10.311.9: Solutions of the Free Particle Dirac Equation  0  1  1 ˜ Ψ(p = 0.  −1  e 0   0  1  +imt 1 ˜  Ψ(p = 0. yields for the exponential space–time dependence according to (10.311)  e 1 w 2 3 ˜ Ψ(p(w3 )ˆ3 . 2  0 e −1  329 (10.325) One should note that φo .322).323) should yield the solutions for non-vanishing momentum p in the x3 –direction. (10. 2 The solutions (10. −.313) in the p → 0 limit. − 1 |r.325) to (10.10.174. − 1 |t) = √2   e−imt .322) can be written in spinor form 1 √ 2 φo χo e imt . φo . Applying (10.324) 0 e − 1 w3 σ 3 2 φo χo = e 2 w3 σ φo 1 3 e− 2 w3 σ χo 1 3 . 0 1 (10. This correspondence justifies the characterization ±. i. w3 ).323) Transformation (10.324. For the resulting wave functions in the chiral representation one can use then a notation corresponding to that adopted in (10. t) e 2 e− 2 w3 1  = √  1 2  e 2 w3  i(px3 − E t) p e  .. +.262) for a boost in the x3 –direction. 0. 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 . χo ∈ 1 0 . + 1 |r. χo are eigenstates of σ 3 with eigenvalues ±1. − 1 |t) = √2  . −. 10. 2 |t) = √2  .322) The reader can readily verify that transformation of these solutions to the Dirac representationsas defined in (10. +.176) imt → i ( p3 x3 and for the bispinor part according to (10. +.228) yields the corresponding solutions (10. ± 1 of the wave functions stated in (10. 2  0  1   1  0  +imt 1 1 ˜  Ψ(p = 0.262) φo χo → e 2 w3 σ 0 1 3 Et ) (10.e. for w = (0. 10.

+ 2 |r. −. t) e  1 0 1 0 0 1 0 1 1 0 1 0 0 1 0 1   i(px3 − E t) p e   1 Ψ(p(w3 )ˆ3 . t) e 1  = √  2   −e− 2 w3 − 1 w3 2 1  i(px3 + E t) p e   ˜ Ψ(p(w3 )ˆ3 . 2 (10.61) p(w3 ) = m sinhw3 . t) e 2 e 1   = √  1 2 − e 2 w3  i(px3 + E t) p e  (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. Transformation to the Dirac representation by means of (10. −. 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 . + 2 |r.327) Employing the hyperbolic function properties cosh x = 2 coshx + 1 . −. +. and the expression (10.228) yields w3  cosh 2 =   sinh w3 2 1 Ψ(p(w3 )ˆ3 .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 (10.61) between the parameter w3 and boost velocity v3 .329) . +.328) the relationship (10. − 1 |r.326) where according to (10. − 2 |r. − 2 |r. t) e 2  =     i(px3 + E t) p e   1 Ψ(p(w3 )ˆ3 . t) e  =    i(px3 + E t) p e  (10. 2 sinh x = 2 coshx − 1 . −.

243) and. The change of sign for the latter solutions had to be expected as it was already noted for the negative energy solutions of the Klein–Gordon equation (10. λ. λ. refered to as the active transformation. ρ(Lη ξ ) has been defined in (10. This transformation.176). = γν . of course. 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 .243) considered solely the limit of infinitesimal transformations anyway.331) i S(Lη ξ )γ µ S −1 (Lη ξ ) ρ(Lη ξ )∂µ ρ−1 (Lη ξ ) − m Ψ (xµ ) = 0 (10.330) Inserting expressions (10.311) as well as the negative energy solutions (10. expresses the system in the old coordinates.334) (10. we will derive this equation only for infinitesimal transformations. when we generated the solutions Ψ(p.327). The reason why we provide another derivation of (10.331) is a solution as well. and since (2) the calculations following (10. indeed.330) into (10.243).221) using (10. Such description will be essential for the formal description of Lorentz invariant wave equations for arbitray spin further below. Λ|xµ ) from the solutions describing particles at rest Ψ(p = 0.168–10.333) . Actually.332) Here we have made use of the fact that S(Lη ξ ) commutes with ∂µ and ρ(Lη ξ ) commutes with γ µ .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. (10. To show this we rewrite the Dirac equation (10. thereby.123) above and characterized there. Such transformation had been applied by us.9: Invariance of Dirac Equation Revisited and similarly sinh w3 = 2 Ep − m = 2m p 2m (Ep + m) 331 (10. We expect. In the new derivation we consider the particle described by the wave function transformed. Making this expectation a postulate allows one to derive the condition (10. Invariance of Dirac Equation Revisited At this point we like to provide a variation of the derivation of (10. in general.10. 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|t).329. the essential property stating the Lorentz–invariance of the Dirac equation. is sufficient since (1) it must hold then for any finite transformation. that if Ψ(xµ ) is a solution of the Dirac equation that Ψ (xµ ) as given in (10. which however. but not the observer. reproduces the positive energy wave functions (10. The transformation is Ψ (xµ ) = S(Lη ξ ) ρ(Lη ξ ) Ψ(xµ ) (10. the proper transformation S(Lη ξ ).313) for −p.

∂µ ] = [x2 ∂1 − x1 ∂2 . .342) One can readily convince oneself that these results are consistent with (10. Inspection of (10. ∂µ ] = (10.339) [K1 .335) shows that we need to demonstrate [J .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 ] = [x1 ∂3 − x3 ∂1 .337) [J2 . 1) . it is met by S(Lη ξ ) as given in the chiral representation by (10. ∂µ ] = (10.243) and. . i. ∂µ ] = [x3 ∂2 − x2 ∂3 .336). ∂µ ] = (10. of course. We have demonstrated. [K . ∂µ ] = [x0 ∂1 + x1 ∂0 . 0. ∂µ ] = (10. ∂µ ] = (K )ν µ ∂ν . 1. ∂µ ] = (10.332 Relativistic Quantum Mechanics We will demonstrate now that the first condition is satisfied by ρ(Lη ξ ). We will proceed by employing the generators (10. ϑ = (0. w = (0. 0).331) is again a solution of the Dirac equation.. . w = (0. 0). 0. ϑ = (0. 0. the Dirac equation is invariant under active Lorentz transformations. ∂µ ] = (10.     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. 0). The second condition is identical to (10. w = (0.340) [K2 . As mentioned already we will show condition (10. 0.e.334) for infinitesimal Lorentz transformations Lη ξ .341) [K3 . 0). 0.335) = ∂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. ∂µ ] = [x0 ∂2 + x2 ∂0 . ∂µ ] = [x0 ∂3 + x3 ∂0 .338) [J3 . ∂µ ] = (J )ν µ ∂ν .336) We will proceed with this task considering all six cases: [J1 .262). . 0). therefore. that any solution Ψ(xµ ) transformed according to (10. .

345) in the so-called non-relativistic limit in which all energies are much smaller than m. we replace the operator ∂µ by ∂µ + iqAµ .345) (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. Following the respective procedure developed for the Klein-Gordon equation in Sect. (10. accordingly.10: Dirac Particles in Electromagnetic Field 333 10.349) We want to focus on the stationary positive energy solution. we define φ(xµ ) = e−imt Φ(xµ ) χ(x ) = e µ −imt (10. σ3 )T one obtains then i∂t φ i∂t χ = = ˆ σ · π χ + qV φ + mφ ˆ σ · π φ + qV χ − mχ .344) Using the notation σ = (σ1 . (10. This solution exhibits a timedependence exp[−i(m + )t] where for holds in the non-relativistic limit | | << m.346) ˆ ˆ ˆ α · π + qV + β m Ψ(xµ ) (10.233) by replacing i∂t by (see Table 10.1 in Sect 10.343) One may also include the electromagnetic field in the Dirac equation given in the Schr¨dinger form o ˆ (10.g. σ2 .221) reads then [ iγ µ (∂µ + iqAµ ) − m ] Ψ(xν ) = 0 (10. The Dirac equation (10. e.. 10.351) X (x ) µ .350) (10. Equivalently. For this purpose we choose the decomposition Ψ(xµ ) = φ(xµ ) χ(xµ ) .347) (10.6 we assume that the field is described through the 4-vector potential Aµ and. Accordingly. 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 .345) holds |qV | << m . for the scalar field V in (10.232). Non-Relativistic Limit We want to consider now the Dirac equation (10.348) (10.10 Dirac Particles in Electromagnetic Field We like to provide now a description for particles governed by the Dirac equation which includes the coupling to an electromagnetic field in the minimum coupling description.10.6 above).

For this purpose we employ the identity (5.356) i∂t Φ ≈ Φ + qV Φ . One obtains [πj .334 and assume that for the time-derivative of Φ and X holds ∂t Φ << m . ∂k ] + q [ ∂j . does not need to be considered anymore.355) ˆ σ·π Φ 2m (10.359) We want to evaluate the latter commutator. ∂k ] + q [Aj . Ak ] . 5. (10. due to the m−1 factor.7. which in the present case states ˆ ˆ2 ˆ ˆ (σ · π)2 = π + i σ · (π × π) .352) allow one to approximate (10. πk ] = = 1 1 [ ∂j + qAj .360) For an arbitrary function f (r) holds ( [Aj .351) in (10. 10. accordingly. Ak ] + q 2 [Aj .348. Ak ] i i i i =0 =0 = q q [Aj . (10. Ak ] ) f = ( ∂j Ak − Ak ∂j + Aj ∂k − ∂k Aj ) f .230). and.356).354) The properties (10. ∂k + qAk ] i i 1 1 1 1 [ ∂j . (10.350.353) by X ≈ to obtain a closed equation for Φ ˆ σ·π 2m 2 (10.361) . i i (10.354) ˆ 0 ≈ σ · π Φ − 2m X . ˆ Equation (10. ∂k ] + [∂j . (10. ˆ ˆ For the components of π × π holds ˆ ˆ π×π = jk (10. ∂k ] + [∂j .357) Equation (10. henceforth. identifies X as the small component of the bi-spinor wave function which.353) (10.349) yields i∂t Φ i∂t X = = ∂t X X Relativistic Quantum Mechanics << m . (10.352) ˆ σ · π X + qV Φ ˆ σ · π Φ + qV X − 2mX . one can replace X in (10.346.357) for Φ can be reformulated by expansion of (σ · π)2 . 10. 10. Φ Using (10.358) ( πj πk − πk πj ) = jk [πj . derived in Sect. π ] .

2) it introduces the o extra term qσ · B Φ which describes the well-known interaction of a spin. r (10. t) [see (8.e. −A).2 around the magnetic field.365) Comparision of this expression with (5.367) . Ak ] ) f = [ ((∂j Ak )) − ((∂k Aj )) ] f (10. 0.360) and Aµ = (V. 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) . (10.. 0. t) 2m 2m Φ(r. Dirac Particle in Coulomb Field . t)]2 q − σ · B(r. Equations (10. t) ≈ ˆ [p − q A(r.Spectrum We want to describe now the spectrum of a relativistic electron (q = −e) in the Coulomb field of a nucleus with charge Ze.366) can be interpreted as a rotation around the field B by an angle qtB. one obtains ( [Aj .363) allow us to write (10. t) + q V (r. 0) .344. Let us consider briefly the consequences of the interaction of a spin. o reveals a stunning feature: the Pauli equation does justice to its two-dimensional character.1 particle with a magnetic 2 field B. The formal solution of this equation is Φ(t) = e−iqtB·σ Φ(0) . using (10. Comparision of (10.1 with the magnetic field. but rather remains as a steady “guest” of non-relativistic physics with the proper interaction term.358.363) where we employed B(r.343) for the vector potential Aµ = (− Ze2 .2) governing a one-dimensional wave function ψ ∈ C. i. 10.222.362) or. The respective bispinor wave function Ψ(xµ ) ∈ C4 is described as the stationary solution of the Dirac equation (10. t) (10. while agreeing in all other respects with the non-relativistic Schr¨dinger equation (10. [πj . In other words. 10.366) (10.223) shows that the propagator in (10.6)].364) governing a two-dimensional wave function Φ ∈ C2 with the corresponding non-relativistic Schr¨dinger equation (10. t) = × A(r. does not leave the theory again when one takes the non-relativistic limit. 10.364) which is referred to as the Pauli equation.1 which emerged in the Lorentz-invariant theory as an algebraic 2 necessity.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 ((· · ·)). πk ] = q q (( ∂j Ak − ∂k Aj )) = − i i ×A jk q = − B i jk (10. ∂k ] + [∂j .357) in the final form i∂t Φ(r.10. the spin. 5. the interaction qσ · B induces a 1 precession of the spin.359.

1 one can write (10.369).228) and in (10. but differs from it in an essential way. 3 µ=0 ˆ2 (˜ µ )2 πµ = π0 − π .s.369) For our solution we will adopt presently a strategy which follows closely that for the spectrum of pionic atoms in Sect. respectively. Employing πµ as defined ˜ in Table 10. For this purpose we ‘square’ the Dirac equation.375) Following the algebra that connected Eqs.229). indeed.371). ˜ However. according to (10.7 one can write the second term in (10.232) in Sect.368) ˜ ( γ µ πµ − m ) Ψ(xµ ) = 0 .374).230) γ µ γ ν = − γ ν γ µ .6.370) resembles closely the Klein-Gordon equation (10.368) is also a solution of (10.180). ˜ (10. (˜ 0 )2 = 1 and (˜ j )2 = γ 1 γ −1 j = 1. γ ˆ2 ˆ2 (10.373) (10. can be rewritten using. multiplying (10. 5.h.371) such that we can conclude that (10. 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 .374) The first term on the r. 1.231).368) ˜ where Ψ(xµ ) and γ µ are defined in (10.e.370) for which holds ˜ ˆ ˜ ˆ 3 γ πµ γ πν = ˜ ˆ ˜ ˆ µ=0 µ ν (˜ µ )2 πµ + γ ˆ2 µ. 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. i. (5. once a solution Ψ(xµ ) of (10.230).369). The difference arises from the term γ µ πµ γ ν πν in (10.372) (10. we solve ˜ [ i˜ µ (∂µ + iqAµ ) − m ] Ψ(xµ ) = 0 γ (10.370) is obtained then ˜ [ i˜ µ (∂µ + iqAµ ) + m ] Ψ(xµ ) γ is a solution of (10. 3. but the converse is not necessarily true. noting from (10. ˜ ˜ ˆ ˆ (10.ν=1 µ =ν γ µ γ ν πµ πν . 2.336 Relativistic Quantum Mechanics For the purpose of the solution we assume the chiral representation.370) ˜ [ i˜ µ (∂µ + iqAµ ) − m ] [ i˜ µ (∂µ + iqAµ ) + m ] Ψ(xµ ) = 0 γ γ (10. µ = ν and altering ‘dummy’ summation ˜ ˜ ˜ ˜ . γ ˆ ˜ ˆ (10. is a solution of (10. Equation (10.370) Any solution of (10.369) from the left by γ ν πν + m. (5.370).

r2 (10.ν=1 µ =ν 3 j=1 3 3 = 1 4 1 2 1 [˜ . due to A = 0.10. γ ν ] [ˆµ .344). Since [ˆµ .ν=1 µ =ν = [˜ µ . 2. πj ] in (10. πν ] γ ˜ π ˆ µ. γ µ γ ν πµ πν = ˜ ˜ ˆ ˆ µ. γ j ] are γ ˜ [˜ 0 . πj ] .382) . one can summarize (10.ν=1 µ =ν (10.376–10.380) γ µ γ ν πµ πν = i ˜ ˜ ˆ ˆ µ. πj ] + γ ˜ π ˆ 4 0 j [˜ j .378) According to the definition (10. Altogether.378) can be evaluated using (10. i. ν = 1.380) where f = f (r.367) and the definition (10. πν ] = 0 π ˆ for µ. πj ] π ˆ = = (−i∂t + qA0 .e.229). γ ν ] [ˆµ . 3 (10.367) of Aµ . π0 ] γ ˜ π ˆ j=1 = [˜ 0 . t) is a suitable test function and where ((· · ·)) denotes the range to which the derivative is limited.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. the commutators [˜ 0 .381) According to (10. −i∂j ] = − [ (∂t + iqA0 ) ∂j − ∂j (∂t + iqA0 ) ] f i ((∂j qA0 )) f (10.. γ ] [ˆ0 . it holds 1 4 [˜ µ .379) The commutators [ˆ0 . 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 (10.367) holds qA0 = r ˆ Ze2 . γ ˜ π ˆ j=1 (10. γ j ] [ˆ0 .10: Dirac Particles in Electromagnetic Field indices. γ 0 ] [ˆj .376) This expression can be simplified due to the special form (10.ν=1 µ =ν σ 0 0 −σ · (( qA0 )) (10. πν ] γ ˜ π ˆ µ.344) π ˆ [ˆ0 .377) which follows readily from the definition (10.ν=1 µ =ν 337 1 2 ( γ µ γ ν πµ πν + γ ν γ µ πν πµ ) ˜ ˜ ˆ ˆ ˜ ˜ ˆ ˆ µ.

1 |ˆ) ) . 1 |ˆ).. We note that these states are also eigenstates of the angular r 2 . a solution of the form ∼ Y m (ˆ) can be obtained. couples the orbital angular momentum of the electron to its spin. 2 |ˆ)} as given in (6.381). + j } (10.6. 2 3 2 . only the two states {Yjm (j − 2 .374). hence.384) into (10.384) can be interpreted as the energy of the stationary state and.389) According to the results in Sect. (10. 6.388) in terms of states introduced in Sect.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. . (10. (10.148). i. 1 |ˆ). in r ˆ fact.147.2 components ˜ Ψ(r) = and obtain for the separate components φ± (r) Ze2 + r 2 ˜ φ+ (r) ˜ φ− (r) (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 . is genuinely two-dimensional and. 6. −j + 1. it is this quantity that we want to determine. Yjm (j + 1 . . In the latter case.383) We seek stationary solutions of this equation. . 6. Insertion of (10.e.387) The expression (10.5 the operator iσ · r is block-diagonal in this basis such that only ˆ 1 the states for identical j. r r 2 2 2 j = 1. Combining this result with (10. m = −j. (10. Such solutions are of the form ˜ ˜ Ψ(xµ ) = Φ(r) e−i t . The term iσ · r.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 . .375) the ˆ ‘squared’ Dirac equation (10. . however. (10. Yjm (j + r 2 1 1 . .5 which describe the coupling of orbital angular momentum and spin 1 {( Yjm (j − 2 . Accordingly.385) 1 We split the wave function into two spin. 10.191) solved in Sect. m values are coupled.386) + 2 ± iσ · r ˆ Ze2 − m2 r2 φ± (r) = 0 .368) reads − Ze2 ∂t − i r 2 + 2 + i σ·r ˆ 0 0 −σ · r ˆ Ze2 − m2 r2 Ψ(xµ ) = 0 (10.338 Relativistic Quantum Mechanics where r = r/|r| is a unit vector.1 . we express 2 the solution of (10. (10.

(6.2 (j)[λ1. 1. . namely.10. 1 |ˆ) = − Yjm (j ˆ r 2 2 1 2 h± (r) g± (r) Yjm (j − 1 . in the present treatment. .397) This equation is identical to the Klein-Gordon equation for pionic atoms written in the form (10.151]. and except for the fact that 2 2 we have two sets of values for λ1.396) and (10. except for the slight difference in the expression of λ1. Obviously. a specific pair of total spin-orbital angular momentum quantum numbers j.. the eigenvalues are independent of m. 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 (10.393) but do not explicitly consider further the wavefunctions.2 (j) + 1) 2 Ze2 + + r2 r 2 − m2 f1.f. − 1 r2 (j − 1 )(j + 1 ) − Z 2 e4 2 2 ±i Ze2 (j + )(j + ) − Z 2 e4 We seek to bring (10. 1 |ˆ) + r Yjm (j + 1 . not the wave functions. which states that the states Yjm (j ± 1 . therefore.2 (j) being 2 j = 1 .396) and λ2 (j) [λ2 (j) + 1] (10. 3 . 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 (10.2 (j).395.390) (10.10: Dirac Particles in Electromagnetic Field 339 ˆ momentum operator L2 [cf.199).391) derived in Sect. the missing additive term − 1 . λ1 (j) and λ2 (j). we want to determine solely the spectrum. However.394) Equation (10. (6. 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 . Since. rather than = 0.392). .392) into diagonal form.151).2 (r) = 0 (10. the values of the argument of λ1.395) (10. Any similarity transformation leaves the first term in (10. 1 |ˆ) r 2 2 2 2 r r .200). as in the case of pionic atoms. namely. hence. 6.5 [c. (10. 1 |ˆ) are r 2 2 ˆ eigenfunctions of L2 . 1 |ˆ) r 2 (10. We can.392) reads then in the diagonal representation 2 ∂r − λ1. m and expand φ± (r) = Using σ · r Yjm (j ± 1 . unaltered. .2 (j) as given by (10.186)]. We select. .392) h± (r) g± (r) = 0. conclude that the . 10. property (6. . such transformation can be chosen as to diagonalize the second term. involving the 2 × 2 unit matrix.

. 2n2 = 0. . .395. including spin.397) is again given by eq.395) and 2 corresponds to λ2 (j) as given in (10. . . j . . . 2. 2. 3. . and spin-orbital angular momentum j = + 1 . Obviously. −j + 1. .s. . 1. (10. 2. 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 .396). . . One may also state 2 this in the following way: (10. . It is given by n = n + + 1 where is the orbital angular momentum quantum number and n = 0. . j = + 1 . m = −j. (10. . . These considerations 2 are summarized in the following equations 1 ED (n. . j . One obtains in case of (10. j = − 2 . . of these equations describe the binding energy. . One can equate (10.403) √ Z 2 e4 2 ED (n. . Using (10. −j + 1. . . m) = 1 + m (n− + . the second term on the r. 10. . .399) n = 0. . .401) These expressions can be equated with the non-relativistic spectrum. . 1 ms = ± 2 (10. .402) with (10. −j + 1. . . n − 1 . − + 1.401) corresponds to a non-relativistic state 2 with quantum numbers n. m = −j.398. .400) (10. The magnitude of the relativistic effect 2 2 e4 . . . counts the nodes of the wave function. .203). the stationary states have binding energies E = − mZ 2 e4 n = 1. j where 1 corresponds to λ1 (j) as given in (10.2 . m) 2 n = 1. 2 2 1 ≈ m − (10. . . = 1 + m Z 2 e4 √ ( n − − 1 + 2 − Z 2 e4 )2 . (10. . In case of non-relativistic hydrogen-type 1 atoms. m = − . (10. 10. accordingly. n − 1 . . . .398) 2 = 1 + m . . 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 . .400) and (10.340 Relativistic Quantum Mechanics spectrum of (10. 1. . ( +1) − Z 2 e4 )2 (10. .400) corresponds to a non-relativistic state with quantum numbers n. . 1.396) we obtain.402) In this expression n is the so-called main quantum number. . 3 . 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. albeit with some modifications.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. j = 1 . .404) = 0. . .401) if one attributes to the respective 1 states the angular momentum quantum numbers = j + 2 and = j − 1 .h. . 2 2 m = −j. 1. j = 1. and spin-orbital angular momentum j = − 1 . 1. . . . 2. (10.

10. This is clearly demonstrated by the comparision of non-relativistic and relativistic spectra of a hydrogen-type atom with Z = 100 in Table 10.405) In order to demonstrate relativistic effects in the spectrum of the hydrogen atom we compare in Table 10.51176 ⇑ ⇑ ⇑ ⇑ 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. . the non-relativistic and relativistic energies are hardly discernible.551177 ⇑ . case these six states are split into energetically different 2p 1 and 2p 3 states. Of particular interest is the effect of spin-orbit coupling which removes. . j.10: Dirac Particles in Electromagnetic Field main quantum number n 1 2 2 2 3 3 3 3 3 orbital angular mom. r 2 2 2 . j 1 2 1 2 1 2 3 2 1 2 1 2 3 2 3 2 5 2 341 rel. the nonrelativistic degeneracy for the six 2p states of the hydrogen atom: in the present.3.60601 -3. .405)] spectrum of the hydrogen atom.e. notation 1s 1 2 non-rel. mom. (10. One can combine the expressions (10.402. . ± 3 . .2: Binding energies for the hydrogen (Z = 1) atom. (10. 1 |ˆ) for m = ± 1 . The reason is that the mean kinetic energy of the electron in the hydrogen atom.. 10. j (10.1. = 0.405 ) -13. (10. . 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. is in the range of 10 eV.10. 1.2 the non-relativistic [cf. i.405). in fact.3. binding energy / eV Eq.40147 -1.1. . i. . However.e.551176 spectr. .60583 -3. much less than the rest mass of the electron (511 keV). . relativistic.403. . n − 1 j = 1 2 1 2 ± m = −j. Degeneracies are denoted by ⇑.402) -13..0041 keV and e2 = 1/137. −j + 1. (10.40146 ⇑ ⇑ -1. 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. for example. (10. 0 0 1 1 0 1 1 2 2 spinorbital ang.036 by means of Eqs.405) in terms of the non-relativistic quantum numbers n.40151 ⇑ . m) = m 1 + Z 2 e4 (n−j −1 + 2 (j+ 1 )2 − Z 2 e4 )2 2 n = 1.551178 ⇑ . relate closely to the corresponding nonrelativistic states. .404) finally into the single formula ED (n. binding energy / eV Eq.402)] and the relativistic [cf. The energies were evaluated with m = 511. 2. The table entries demonstrate that the energies as given by the expression (10. the 2p 3 states with r 2 2 2 2 1 involve four degenerate states corresponding to Y 3 m (1. 2 |ˆ) for m = ± 1 .

15. 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. (10.036 by means of Eqs. binding energy / keV Eq. We assume for the wave function the stationary state .3: Binding energies for the hydrogen-type (Z = 100) atom. j. In order to investigate further the deviation between relativistic and non-relativistic spectra of hydrogen-type atoms we expand the expression (10. An example for such potential is the Coulomb potential V (r) = −Ze2 /r considered further below.3).35.405). 0. 10. m) ≈ m − mZ 2 e4 mZ 4 e8 − 2n2 2n3 1 j+ − 3 4n + O(Z 6 e12 ) . The energies were evaluated with m = 511. It holds for n = 2 and j = 3 . mom.402.15.0 ⇑ ⇑ -15.1 -34. Introducing α = Z 2 e4 1 and β = j + 2 (10.409) where V (r) is spherically symmetric.9 ⇑ .8 ⇑ .2. 0.6 -42. for example.10.1 ⇑ . 32 (10. to estimate the difference between the energies of the states 2p 3 and 2p 1 (cf. (10. 1 2 2 2 2 E (2p 3 ) − E (2p 1 ) ≈ − 2 2 mZ 4 e8 2 · 23 1 −1 2 = mZ 4 e8 . Tables 10. j 1 2 1 2 1 2 3 2 1 2 1 2 3 2 3 2 5 2 Relativistic Quantum Mechanics non-rel. 0) (10.402) -136.405) to order O(Z 4 e8 ).3 spectr. binding energy / keV Eq.0041 keV and e2 = 1/137. 0 0 1 1 0 1 1 2 2 spinorbital ang. .406) α √ 1+ (n − β + β 2 −α)2 The expansion (10.405) reads 1 (10.2 -17.408) 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).1 ⇑ ⇑ ⇑ ⇑ rel. (10. (10.405 ) -161.342 main quantum number n 1 2 2 2 3 3 3 3 3 orbital angular mom. Degeneracies are denoted by ⇑.407) 1 2 This expression allows one.206) provides in the present case ED (n.

168]: the term is a scalar (rank zero tensor) in the space of the spin-angular momentum states Yjm (j ± 1 . (6. 1 |ˆ) r 2 = i ∂r − 1 2 r g(r) Yjm (j + 1 .2 degrees of freedom. (10. 10. 2 |ˆ)  r Φ(r)  r  . r 2 (10. 6. 6.414) These equations can be brought into a more symmetric form using ∂r + which allows one to write (10. m-values.412) are four-dimensional with r-dependent wave functions.10: Dirac Particles in Electromagnetic Field form Ψ(xµ ) = e−i t 343 Φ(r) X (r) .410) 1 where Φ(r). 2 |ˆ) + r Yjm (j − 2 . in particular.197. but are timeindependent.. 1 |ˆ) and Yjm (j − 1 .411) (10.411.232. The Dirac equation reads then. 1 |ˆ) r 2 1 1 = ∂r r r r (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 . 6.e. X (r) ∈ C2 describe the spatial and spin.. ˆ σ · p X + m Φ + V (r) Φ ˆ σ · p Φ − mX + V (r) X = = Φ X (10.5.198)] not couple states with different j. r 2 2 (10. 2 |ˆ) and does r 2 ˆ has odd parity and it holds [c.f. i.412) ˆ In this equation a coupling between the wave functions Φ(r) and X (r) arises due to the term σ · p. 2 |ˆ) + r Yjm (j − 2 . σ · p ˆ σ · p f (r) Yjm (j + 1 .416) The differential equations (10.e. 1 |ˆ) r 2 2 = i ∂r + j+ r 1 2 3 2 f (r) Yjm (j − 1 .   =  r (10. 10.414) ˆ σ · p rf (r) Yjm (j + 1 . 1 |ˆ) r 2 2 = i ∂r + −j r 1 g(r) Yjm (j + 2 .415) 1 ˆ σ · p rg(r) Yjm (j − 2 .345). pp. r r 2 2 2     a(r) b(r) 1 1 1 1 Yjm (j + 2 . 1 |ˆ) introduced in r 2 2 1 Sect.5 [see. i. 1 |ˆ). 1 |ˆ) . the term is block-diagonal in the space spanned by the states Yjm (j ± 1 . 1 |ˆ) r 2 2 = i j+ ∂r + r j+ r 1 2 1 r f (r) Yjm (j − 2 .413. 1 |ˆ) r 2 2 (10.417)  c(r)  d(r) X (r) 1 1 1 1 Yjm (j + 2 . This term has been discussed in detail in Sect. according to (10. 1 |ˆ) .413) ˆ σ · p g(r) Yjm (j − 1 . 2 |ˆ) r r r .10. 10.

According to (10.418) holds for f1 (r). 2 |ˆ) r r   f2 (r) 1 1 r Φ(r)  i r Yjm (j − 2 . 10. only a(r). Accordingly. m values. Inserting (10.421)   g2 (r) X (r) 1 1 − Yjm (j + 2 .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 (10.157).411. 1 |ˆ)  r Φ(r) 2 2  r    =  (10.419) Obviously. we consider the case that only states for one specific j. 2 |ˆ)   =  (10. the orthonormality property (6. g1 (r) ∂r − j+ r j+ r 1 2 g1 (r) + [ − m − V (r) ] f1 (r) = 0 ∂r + and for f2 (r).417) into (10.412). such expansion must include states with all possible j. 10. Presently. d(r) are coupled and b(r).420)   g1 (r) X (r) 1 1 − Yjm (j − 2 . using (10.423) . (10. 2 |ˆ) r r where the factors i and −1 have been introduced for convenience. 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. m pair contribute. g2 (r) ∂r + 1 2 f1 (r) − [ + m − V (r) ] g1 (r) = 0 (10. c(r) are coupled.416).344 Relativistic Quantum Mechanics In general.417) of the form     f1 (r) i Yjm (j + 1 .415. there exist two independent solutions (10.418) ∂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.

e. (10. for small r... (10. In the non-relativistic limit. 2 2 2 (10. 3p 1 . etc. one demonstrates first that the wave function at r → 0 behaves as rγ for some suitable γ. Equation (10.424) Ψ(xµ ) ≈  r 0 1 i.422. = 0.405). . determining wave functions of the type (10. one demonstrates then that the wave functions for r → ∞ behaves as exp(−µr) for some suitable µ. the 2 equations determine.421).405). (10. 2 2 2 2 2 2 = j − 1 2 and. Dirac Particle in Coulomb Field . n − 1. . to states with angular momentum = j + 2 .422) near r = 0. Hence.422).422) describes the states 2p 1 . 1 |ˆ)  r i 2 2 .422). According to the discussion of the spectrum (10. 10. 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. According to this procedure.425) r 0 covers states with angular momentum 1s 1 . .367) which is spherically symmetric such that equations (10.423) apply for V (r) = −Ze2 /r.409). 3s 1 . 2s 1 .422). (10. except for the opposite sign of the term (j + 1 ). i. . 2 |ˆ)  r Ψ(xµ ) ≈  . the radial wave functions for Dirac particles in the potential (10. 3d 5 . Behaviour at r → 0 We consider first the behaviour of the solutions f1 (r) and g1 (r) of (10. together with the appropriate boundary conditions at r = 0 and r → ∞. 1. the ones given in (10. Φ in (10.422) determines solutions of the form (10. The solution of (10.426) . .422) corresponds to states   f1 (r) Yjm (j + 1 . in the non-relativistic limit wave functions   f2 (r) 1 1 i Yjm (j − 2 . 2p 3 . correspodingly the states We consider first the solution of (10. 2. Similarly.420) is the large component and X is the small component. and obtains finally a polynomial function p(r) such that rγ exp(−µr)p(r) solves (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. (10.10: Dirac Particles in Electromagnetic Field 345 Equations (10.e.420). We note that (10. . 3d 3 . . etc.422) and (10. Hence.422) follows in this case from the same procedure as that adopted for the radial wave function of the non-relativistic hydrogen-type atom.10.423) are identical. namely. We assume the 4-vector potential of pure Coulomb type (10.423). enforcing the polynomial to be of finite order leads to discrete eigenvalues .405) of the relativistic hydrogen atom the corresponding states have quantum numbers n = 1. 3p 3 .

r2 (10. Such r-dependence 2 would make the expectation value of the potential r2 dr ρ(r) 1 r (10. (10. determined that the solutions f1 (r) and g1 (r). One obtains γ = ± (j + 1 )2 − Z 2 e4 . 10. hence. (10. µ = m2 − 2 (10.434) √ The solutions of these equations are f1 . for small r. g1 (r) →0 ∼ b rγ (10.429) This equation poses an eigenvalue problem (eigenvalue −γ) for proper γ values. ) f1 (r) ( + m ) g1 (r) (10.267.432) = = − ( − m. in case j + 1 )2 < Ze2 . assume the r-dependence in (10. g1 (r) →∞ ∼ e−µr . We have. Only the exponentially decaying solution is admissable and.427).427) = = 0 0.422) becomes ∂r g1 (r) ∂r f1 (r) Iterating this equation once yields 2 ∂r g1 (r) 2 ∂r f1 (r) 1 . according to (10. becomes imaginary.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. g1 ∼ exp(± m2 − 2 r). 10.430) 2 Note that the exponent in (10. The assumed r-dependence in (10.427) with 1 γ = (j + )2 − Z 2 e4 .420).427) makes only the positive solution 2 possible.431) infinite since. (10.435) . hence.433) = = ( m2 − 2 ) g1 (r) ( m2 − 2 ) f1 (r) (10.266. we conclude f1 (r) →∞ ∼ e−µr .

445) .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 .434) √ m + a e−µ r .10: Dirac Particles in Electromagnetic Field For bound states holds 347 < m and.440) and (10.438) (10.441) The last two terms on the l. However.440) (10. we set f1 (r) g1 (r) = = √ √ ˜ m + e−µr f1 (r) −µr = = 0 0 (10.422) for the Coulomb potential V (r) = −Ze2 /r We assume a form for the solution which is adopted to the asymptotic solution (10.422 ) 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 ˜ (10.439) − m− e g1 (r) ˜ where µ is given in (10.10.436). of both (10.436) f1 (r) = Insertion into (10. µ is real.441) correspond to (10. without loss of generality we can choose ˜ f1 (r) = φ1 (r) + φ2 (r) .444) (10.h.435 ). hence.437) where they ˜ ˜ cancelled in case f1 = g1 = a. g1 (r) = φ1 (r) − φ2 (r) ˜ (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. Let us consider then for the solution of (10. ρ m+ ρ [∂ρ − (10. (10. Accordingly.434) results in √ √ (m − ) m + a − (m − ) m + a √ √ (m + ) m − a − (m + ) m − a which is obviously correct. √ g1 (r) = − m − a e−µr .s. Equation (10.442) which leads to a partial cancellation of the asymptotically dominant terms. (10.437) (10. We also introduce the new variable ρ = 2µ r . Solution of the Radial Dirac Equation for a Coulomb Potential To solve (10.

450) follows Ze2 m2 − 2 βs − βs−1 =0 (10.448) φ2 (ρ) = ρ (10. 10.426–10.451) follows βs−1 = = Ze2 s+γ + √ m2 − s+γ − βs + m2 Z 2 e4 m2 − 2 − (j + 1 )2 2 2 √ Ze m2 − 2 2 s+γ − βs (10.430) one can write this βs = s+γ − 2 √ Ze m2 − 2 s (s + 2γ) βs−1 .449) into (10.446) (10.446 .446.f.447) of the form n φ1 (ρ) = ργ s=0 n γ s=0 αs ρ2 βs ρ2 (10.451) 1 √mZe − (j + 2 ) 2 2 αs . (10.448.449) for γ given in (10.447) We seek solutions of (10. 10.452) From (10.450) −√ mZe2 m2 − 2 βs ρs+γ−1 = 0 mZe2 m2 − αs (s + γ) β2 + (j + 1 ) αs + √ 2 s 2 +√ From (10. Inserting (10. Using (10. 10. = m− Ze2 βs s + γ − √m2 − 2 2 (10.430)].430) which conform to the proper r → 0 behaviour determined above [c.454) .447) leads to (s + γ) αs + (j + 1 ) βs − √ 2 s Ze2 m2 − 2 αs (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 .

2. x) = 1 + a a(a + 1) x2 x + + .458) or.459) − 2 √mZe m2 − 2 (10.e.461 for values given by (10. with the associated Laguerre polynomials L(α) = F (−n. . c c(c + 1) 2! (10. (s − so ) β0 s! (2γ + 1)(2γ + 2) · · · (2γ + s) (10. in fact. ρ) .405). n It holds φ1 (ρ) φ2 (ρ) = = ργ F (−so . . i. . 10. for the states 2p 1 .457) with the confluent hypergeometric functions F (a.461) In order that the wave functions remain normalizable the power series (10. α + 1.455) 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 ) . m (n ) = .452) follows j+ αs = 1 2 349 Ze2 m2 − 2 −γ (10. We can. conclude 2 2 2 that the polynomials in (10. . .456) and (10. (10.449) and (10. 2γ + 1. Comparision with (10. .462) Z 2 e4 1 + 1 (n + (j+ 2 )2 − Z 2 e4 )2 This expression agrees with the spectrum of relativistic hydrogen-type atoms derived above and 1 given by (10. . For example. ρ) so β0 ργ F (1 − so . 10. 3d 3 holds n = 1. . so = n .457) One can relate the polynomials φ1 (ρ) and φ2 (ρ) defined through (10. (s − so ) β0 s! (2γ + 1)(2γ + 2) · · · (2γ + s) (10. This requires that all coefficients αs and βs must vanish for s ≥ n for some n ∈ N. According to the definitions (10.457) for βs and αs imply that so must then be an integer. = From (10.448. hence. β0 j+ 1 2 (10.405) allows one to identify n = n − j + 2 which.460) (10. is an integer. x) .. . The expressions (10. . 2. 10. 1. 2γ + 1. namely.455) are finite.456. c. equivalently.10: Dirac Particles in Electromagnetic Field Defining so = √ one obtains βs = = .449) must be of finite order. 1.448. 10.430.. n = 0.405) and the ensuing so values ( 10.455) this confinement of so implies discrete values for ..10.456) − so 2 √mZe m2 − 2 (−so )(1 − so )(2 − so ) . 3p 1 .

147.465) (10. can follow the procedure adopted for the wave functions of the non-relativistic hydrogen atom and will not be carried out here. Berlin.421)] ∞ dr (|f1 (r)|2 + |g1 (r)|2 ) = 1 (10.461). given by expression (10. The complete wave function is given by the following set of formulas 1 Ψ(n.421). They are described through quantum numbers n.460. Sect. (Springer. 1 |ˆ) 2 2 r G1 (r) Yj.464) 0 The evaluation of the integrals. The orthonormality 2 2 r properties (6. The wave functions (10. j. 1990).421) with radial wave functions f1 (r).267).148).350 Relativistic Quantum Mechanics Altogether we have determined the stationary states of the type (10. 1 |ˆ) in (6. The coefficients β0 in (10.438. as in (10. 10. Due to the form (10.158) of the latter states absorb the angular integral in (10. (10. which involve the confluent hypergeometric functions in (10. is time-independent. (10. g1 (r) determined by (10. j. where2 F± (κ|r) = G1 (r) = F+ (κ|r) . = j + 2 . 2 . 9.460. 2γ + 1. 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 .m (j − 1 .460. 10. 2µr) (10.461) are to be chosen to satisfy a normalization condition and to assign an overall phase.157. m|xµ ) = e−i t iF1 (r) Yj. 6.461).466) F1 (r) = F− (κ|r) . = j + 2 . and (10. are two-dimensional vectors determined through the explicit form of the spin-orbital angular momentum states Yjm (j ± 1 .m (j + 1 . 10.6. 10. Greiner.469) This formula has been adapted from ”Relativistic Quantum Mechanics” by W.421) correspond to non-relativistic states with orbital angular momentum 1 1 = j + 2 .410) of the stationary state wave function the density ρ(xµ ) of the states under consideration.463) where Φ and X .439).442).463) and yield [note the 1/r factor in (10. m. 2µr) (2µ) 2 Γ(2γ + 1) 3 N = and m 4m )Γ(2γ + n + 1) (n +γ)m (n +γ)m (10. 1 |ˆ) 2 2 r κ = j+ 1 2 (10. The normalization integral is then ∞ 0 π 2π r2 dr 0 sin θdθ 0 dφ (|Φ(r)|2 + |X (r)|2 ) = 1 (10.467) ± n F (1 − n . (n + γ)m N (2µr)γ−1 e−µr − κ F (−n . 2γ + 1.

j.m (j − 1 .10. One can verify.421) are governed by the radial Dirac equation (10.470) r 0 Obviously. become   f2 (r) 1 1 i Yjm (j − 2 . m|x ) = e −i t iF2 (r) Yj. 1 |ˆ) 2 2 r κ = −j − 1 2 (10.10: Dirac Particles in Electromagnetic Field 351 We want to consider now the stationary states of the type (10. . 3d 5 .421) which.m (j + 1 . etc. = j − µ 1 2 . 2 accordingly. 3p 3 . 1 |ˆ) 2 2 r G2 (r) Yj. in the non-relativistic limit. tracing all steps which lead from (10. describes the complementary set of states 1s 1 .465–10. We have. this wavefunction has an orbital angular momentum quantum number = j − 1 and. 2p 3 .471) (10.423) which differs from the radial Dirac 1 equation for f1 (r) and g1 (r) solely by the sign of the terms (j + 2 )/r. 2s 1 .469) that the following wave functions result Ψ(n. obtained closed expressions for the wave functions of all the stationary bound states of relativistic hydrogen-type atoms. 2 |ˆ)  r Ψ(xµ ) ≈ e−1 t  . G2 (r) = F+ (κ|r) . (10. not not 2 2 2 2 2 2 covered by the wave functions given by (10. 3s 1 .467–10. hence.422) to (10. where F± (κ|r) are as given in (10.472) F2 (r) = F− (κ|r) .469). The radial wave functions f2 (r) and g2 (r) in (10.469).

352 Relativistic Quantum Mechanics .

what will be achieved in the following.. must be two-dimensional irreducible representations of the Lorentz group.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 (10.2) (11. ϑ). ϑ through 3.229). different L(w.262). i.Chapter 11 Spinor Formulation of Relativistic Quantum Mechanics 11. This is. We will characterize the space on which the transformations a(z) We will see below that the representations a(z) and b(z) are. indeed. in particular. We will proceed by building as much as possible on the results obtained sofar in the chiral representation of the Dirac equation.1) (11. a(z) and b(z). ϑ) for the ˜ in the chiral representation as given by (10. 1 353 . may be exploited to express the Lorentz-invariant equations of relativistic quantum mechanics.e.3) (11. in fact.4) = exp = w − iϑ . 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. 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. isomorphic to the natural representation. a complex variable z. The lower dimensionality of a(z) and b(z) implies.. in a sense.225–10. ϑ)µ ν . 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.e. Starting point is the Lorentz transformation S(w. ˜ We have altered here slightly our notation of S(w. ϑ)µ ν correspond to different a(z) and b(z). expressing its dependence on w. i. z ∈ C ˜ Because of its block-diagonal form each of the diagonal components of S(z).

neutrino equation. in 2 fact. the operator ∂µ and the Pauli matrices σ in the new representation and. ϑ ∈ R3 .. Ψ2 (xµ ) transform according to a(z) whereas Ψ3 (xµ ).6) as given by (5.e. will establish the map between L(w. 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 .. Aν .e.5) The transformations in this case.1 states as described by Dm m (ϑ). and the Klein–Gordon equation in the spinor representation. β. to 1 ∗ ∗ z ·σ (11. ± 1 represents the familiar spin– 1 states. Ψ2 (xµ ) and Ψ3 (xµ ). (11. Ψ4 (xµ ) transform according to ˜ ˜ ˜ and Ψ b(z). β. For ϑ = (0. namely. In this case a(z) and b(z) are identical and read a(i ϑ) = b(i ϑ) = exp 1 − ϑ·σ 2 . to the rotational transformations of spin. formulate the Dirac equation.354 Spinor Formulation and b(z) act.9) a∗ (z) = exp 2 . finally. a space of vectors state1 for which holds state2 state 1 ∼ | 1 . This characterization allows one to draw conclusions regarding the state space in which a(z) and b(z) operate. i. ϑ ∈ R3 . i. A First Characterization of the Bispinor States We note that in case w = 0 the Dirac transformations are pure rotations. +1 2 |1.243).7) where “∼” stands for “transforms like”. Relationship Between a(z) and b(z) The transformation b(z) can be related to the conjugate complex of the transformation a(z). (11. Here | 1 . ϑ)µ ν and a(z). a distinct difference in the transformation behaviour of ˜ ˜ ˜ ˜ Ψ1 (xµ ). express 4-vectors Aµ . − 1 2 2 z = iϑ. ϑ ∈ R3 (11. −1  2 2  |1. 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. however. b(z). γ (see Chapter 5). θ ∈ R3 . Ψ4 (xµ ) in case z = w + iϑ for w = 0. for z = iϑ. usually expressed 2 as product of rotations and of functions of Euler angles α. 2 2 2 ˜ Since a(z) acts on the first two components of the solution Ψ of the Dirac equation. + 1 2 2 state 2 ∼ | 1 . are elements of SU(2) and correspond. −1 2 2 z = iϑ. the so-called spinor space. In this case holds a(z) = b(z) ˜ 1 (xµ ). (11.8) We like to stress that there exists. +1  2 2 |1.

355 (11. −1 a∗ (z) −1 (11. (11.224. 11. 11.10) From this one can derive b(z) = where = 0 1 −1 0 . Spatial Inversion ˜ One may question from the form of S(z) why the Dirac equation needs to be four-dimensional. This demonstrates that a(z) is the ˜ transformation which characterizes both components of S(z).1) can be written in the form ˜ S(z) = a(z) 0 0 a∗ (z) −1 (11. one can express a∗ (z) −1 = exp 1 − z ∗· σ 2 = b(z) . using f (a) −1 = f ( a −1 ).14) .11) one first demonstrates that for and −1 as given in (11. 11. in fact.17) We conclude.11) 0 −1 1 0 = . ∗ σ 2 = − σ2 . one can show −1 −1 −1 = −σ1 = σ2 = −σ3 ∗ = −σ1 ∗ = −σ2 σ2 −1 (11. σ2 . −1 = ∗ −σ3 = −σ ∗ .f. therefore.12) yields σ1 σ3 or in short σ Similarly. (11.12) does.1: Lorentz Transformation of Dirac Bispinor ∗ ∗ ∗ where σ ∗ = (σ1 .12).15) (11.18) with a(z) given by (11. a result to be used further below.4) and . −1 given by (11. µ ). Ψ (xµ ) as well as Ψ (xµ ).224)] ∗ σ1 = σ1 . that they alter w into −w. Hence. The effect of inversion on Lorentz transformations is. (5. (11. 1. One can readily verify [c. however. but leave rotation angles ϑ unaltered. Ψ (xµ ) even though these pairs of ˜ ˜3 ˜4 featuring the components Ψ1 (x ˜ 2 components transform independently of each other.10. hold −1 = 1 One notices then. σ3 ).2. 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. a∗ (z) −1 = exp 1 ∗ z σ∗ 2 −1 . . ∗ σ 3 = σ3 . that the transformation (11.11. (11.16) σ = −σ ∗ .13) Explicit matrix multiplication using (5.12) To prove (11.

e.20) ˜ Obviously. Exercise 1. does not suffice for particles like the electron which obey inversion symmetry.2 Relationship Between the Lie Groups SL(2. This implies  ˜ Ψ1 ˜  Ψ2 P  ˜  Ψ3 ˜ Ψ4  ˜ Ψ3 ˜   Ψ4  =    Ψ1 ˜ ˜ Ψ2     . Part (Elsevier. C) = { M.C) and SO(3.356 Spinor Formulation ˜ Let P denote the representation of spatial inversion in the space of the wave functions Ψ.24) = etr(M ) (11.21) = etr( 2 z·σ) = 1 1 (11. hence.2. Amsterdam. 2 . which describes pure rotations. ϑ ∈ R3 . The transformation S(z) in the transformed space is then ˜ P 1. j = 1.A. 2. (11. based on the Jordan–Chevalley theorem. the space spanned by only two of the components of Ψ is not invariant under spatial inversion and. non-singular matrix M holds2 det eM and from [c.21) together with matrix multiplication as the binary operation forms a group.1: Show that SL(2. P −1 = P. M is a complex 2 × 2–matrix.19) i.f.10.  (11.e. the transformations a(z) and b(z) become interchanged. det(M ) = 1 } . C) defined in (11.23) Exercise 11. (5. In fact.224)] tr( σj ) = 0 . can be found in G.3. However. For the general case the proof. (11. Obviously. de Kerf Lie a Algebras. However.A.1) We have pointed out that a(iϑ).. for particles like the neutrinos which do not obey inversion symmetry two components of the wave function are sufficient. a(w + iϑ) for w = 0 is an element of SL(2. 3 . The proof of this important property is straightforward in case of hermitian M (see Chapter 5). is an element of SU(2). 2 = 1 i.G. ˜ ˜ P S(w + iϑ) P = S(−w + iϑ) = b(z) 0 0 a(z) . the Lorentz invariant equation for neutrinos is only 2–dimensional. One can verify this by evaluating the determinant of a(z) det( a(z) ) = det e 2 z·σ 1 (11. 11. B¨uerle and E.. 1990).22) which follows from the fact that for any complex.

Argue why M (Aµ ) = σµ Aµ provides a bijective map.26) or from the fact that the matrices σ0 . (11. −→ µ Consistency of (11.e.135)] this transformation behaviour. σ2 . We will prove below [cf. σ2 . Noting that for covariant vectors according to (10. 0 −i i 0 .32) .31) holds.27) −→ Straightforward transformation into another frame of reference would replace σµ by σµ . in a transformed frame should hold Lρ ν σ µ Aµ σµ A µ .. The following important property holds for M (Aµ ) det ( M (Aµ ) ) = Aµ Aµ which follows directly from (11. σ3 provide a linear–independent basis for the space of hermitian 2 × 2–matrices. σ1 . σ3 are hermitian.30) Since the components of Aµ are real. σ1 .28)   σµ =   1 0 0 1 . 1 0 0 −1    . in fact. 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 .2. (11.29) where we used (10. (11. Starting point is a bijective map between R4 and the set of two-dimensional hermitian matrices defined through M (Aµ ) = σµ Aµ where   (11.1) Mapping Aµ onto matrices M (Aµ ) 357 We want to establish now the relationship between SL(2. one can provide a simple expression for the inverse of M (Aµ ) M = M (Aµ ) ↔ Aµ = 1 tr M σµ 2 . M (Aµ ) according to (11. Demonstrate that (11. 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.25) σ0 σ1 σ2 σ3 The quantity σµ thus defined does not transform like a covariant 4–vector.11.28) and (11.2: Show that σ0 .25) of the matrix M (Aµ ) is independent of the frame of reference.31) Exercise 11. C) and the group L↑ of proper.26) (11. i.2: Lie Groups SL(2.76).C) and SO(3. The function M (Aµ ) is bijective.27) requires then σ µ Aµ Lν µ σµ = σν (11. 0 1 1 0 . (11.30). In fact.25) can also be written M (Aµ ) = A0 + A3 A1 − iA2 A1 + iA2 A0 − A3 . (11. or+ thochronous Lorentz transformations.  (11. one wishes that the definition (11.

(11. a ∈ SL(2. (11. it holds for the matrix M defined through (11.25) Aµ = which allows us to express finally A µ = L(a)µ ν Aν .31)] 1 a Aµ −→ A µ = tr a M (Aµ ) a† σµ 2 Because of (11. C) and for any Aµ there exists an A µ such that M (A µ ) = a M (Aµ ) a† .33) This transformation conserves the hermitian property of M since (M )† = ( a M a† )† = (a† )† M † a† = a M a† = M (11.36) The suitable A µ can readily be constructed using (11. any a ∈ SL(2. Due to det(a) = 1 the transformation (11.35) We now apply the transformation (11.39) Exercise 11.37) defines actually a Lorentz transformation. Accordingly. 11.2.40) is an element of L↑ .33) to M (Aµ ) describing the action of the transformation in terms of transformations of Aµ .37) which implies that (11.33) det(M ) = = det( a M a† ) = det(a) det(a† ) det(M ) [det(a)]2 det(M ) = det(M ) .3: Show that L(a)µ ν defined in (11.40) 1 tr a σν a† σµ 2 Aν (11.34) where we used the properties M † = M and (a† )† = a. for any a ∈ SL(2. + . The linear character of the transformation becomes apparent expressing A µ as given in (11.32.37) using (11. In fact. (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) conserves the determinant of M . C) defines the transformation [c. (11. (11. (11. C) .f. In fact. L(a)µ ν = 1 tr a σν a† σµ 2 .38) .35) holds for this transformation A µ Aµ = Aµ Aµ (11.

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 . (11.δ 1 4 Aαβγδ Γβα Γγδ α.β γ.43) (σν )αβ Γβα Γγδ (σν )δγ = ν.1).45) One can demonstrate through direct evaluation   2 α =    2 α =  2 α = =   2 α =    0 else β β γ γ = = = = γ = γ = 1..C) )µ ρ (11.2: Lie Groups SL(2.e.47) This completes the proof of the homomorphic property of L(a)µ ν .δ (11.44) (σν )αβ (σν )δγ .1) L(a)µ ν provides a homomorphism 359 We want to demonstrate now that the map between SL(2.40)] respects the group property of SL(2. (11. ¯ Lµ ρ = L(a1 )µ ν L(a2 )µ ρ = L( product in SO(3.1) 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 2.46) 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 .α. C) and of SO(3. (11.β γ. 2 (11. 1 ¯ and using the definition of Lµ ρ in (11.42) = ν Defining Γ = a† σµ a1 .41) results in 1 ¯ Lµ ρ = 4 where Aαβγδ = ν Γ = a2 σρ a† 2 (11.C) and SO(3.11. i. β δ = δ = = δ = δ 1 2 = 2 = 1 Aαβγδ (11. C) and SO(3.1) defined through L(a)µ ν [cf. .

0. noting the linearity of M (Aµ ). 0. Aµ = (0.52) where we employed (11. 0). M (Aµ + (K1 )µ ν Aν ) = = M (Aµ ) + M (Aµ ) − M ( (−A1 . Aµ = (0. (11. denoted below as κ1 . we write (11. 0). 10. Because of the condition det(a) = 1 only six independent real numbers actually suffice for the definition of a.58) . 1.59) = = = = 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. K2 . 1.2) which correspond to the generators K1 .. K3 .h. J2 . eight real numbers.36) M (Lµ ν Aν ) = a M (Aµ ) a† (11. C) as complex 2 × 2–matrices are defined through four complex or.55) (11.50) (11.56) (11. i. 0. 0. −A0 . To obtain the generator of a(z) corresponding to the generator K1 .s.57) .s.48). J1 .48) We want to determine now the generators of the transformation a(z) as defined in (11. For the r. (11. of (11.h.53) results in the condition σ 0 A1 − σ 1 A0 = ( κ 1 σ µ + σ µ κ † ) Aµ . Aµ = (0. 0.49) we obtain using (11.51) ( 1 + κ1 ) M (Aµ ) ( 1 + κ† ) 1 1 1 = = M (Aµ ) + M (Aµ ) + ( κ1 M (Aµ ) + M (Aµ )κ† ) + O( 2 ) 1 ( κ1 σµ + σµ κ† ) Aµ + O( 2 ) .49) assuming (note that g µ ν is just the familiar Kronecker δµν ) Lµ ν a = = gµν + 1 + 1 (K1 )µ ν κ1 . correspond to the generators given by (10. correspondingly.51) Insertion of (K1 )µ ν as given in (10.25) in the last step. 0. 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  . of (11. J3 of the Lorentz transformations Lµ ν in the natural representations.e.360 Generators for SL(2. To this end we consider infinitesimal transformations and keep only terms of zero order and first order in the small variables. 1 (11.53) Equating (11.48) yields for the l. 0). 0) ) σ0 A 1 − σ1 A 0 (11. 1) − σ1 σ0 0 0 One can verify readily that 1 κ1 = − σ 1 2 (11. C) which correspond to K. (11.52) and (11.49). 1 (11.54) This reads for the four cases Aµ = (1. 0. J Spinor Formulation The transformations a ∈ SL(2.47.

C) acts on spinors φα ∈ C2 (characterized below) 1 (11. w describing boosts and ϑ describing rotations. a1 1 a1 2 a2 1 a2 2 (11.60) of a ∈ SL(2. + 1 2 2 1 transforms under rotations (z = iϑ) like a spin– 2 state | 1 . 2 λ = i σ. the so-called contravariant spinors.4: Show that the generators (11.3: Spinors 361 obeys these conditions.64) .60) We can. state that the following two transformations are equivalent L(w. J2 .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. Exercise 11. C) correspond to the generators K and J of Lµ ν . 2 2 φ1 φ2 ∈ C2 .11. λ3 corresponding to J1 . hence. def α = 1. 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 .62) We denote the general (z = w + iϑ) transformation by φ α = aα β φβ = aα 1 φ1 + aα 2 φ2 . 2 (11. Similarly. Here a(z) = ( aα β ) = describes the matrix a(z).61) This identifies the transformations a(z = w − iϑ) as representations of Lorentz transformations. − 1 . K3 and λ1 . 2 (11. (11.63) where we extended the summation convention of 4-vectors to spinors. ϑ) = ew·K + ϑ·J ∈ SO(3. one can show that the generators κ2 . J3 are given by 1 κ = − σ.1) acts on 4–vectors Aµ . κ3 of a(z) corresponding to K2 . 11. λ2 .2. a(w − iϑ) = e− 2 (w − iϑ)·σ ∈ SL(2.

1 ) = 2 2 2 2 2 2 2 2 √ ±1/ 2 and equating the spin states of “particle 1” with the spinor φα .. .67) We will refer to φα . should remain invariant under transformations a(iϑ).66) should constitute a singlet spin state. . . 0| . 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. m 1 | . .e.68) (11. . 1 . Such a scalar product does.e. In fact. m. indeed. 0. The relationship between the two can be expressed φ1 φ2 = = 3 φ1 φ2 0 1 −1 0 (11. invariant and as such has the necessary properties of a scalar3 product. as we will demonstrate below such states are invariant under general Lorentz transformations a(z). which 2 can be constructed from products φ1 χ2 . exchange of φα and χβ alters the sign of the expression. . i. . . −m 2 2 2 2 1 1 1 1 2 (11. those of “particle 2” with the spinor χβ one can state that the quantity 1 Σ = √ 2 φ1 χ2 − φ2 χ1 (11. 0| 1 . as covariant spinors. χβ . invariant under transformations a(z).66) is a bilinear form. The corresponding states are defined through φ1 χ2 − φ2 χ1 = φ1 χ1 + φ2 χ2 It obviously holds χα = χ1 χ2 = χ2 −χ1 .. However. χβ . Definition of covariant spinors Expression (11. 1 . To arrive at a suitable scalar product we consider first only rotational transformations a(iϑ). ± 1 . The existence of a scalar product gives rise to the definition of a dual representation of the states φα denoted by φα .e.. In this case spinors φα transform like spin– 1 states and an invariant. . Using (0. i. −m) | .. i. 0 = 2 2 m=± 1 2 1 1 (0.362 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. is the singlet state. etc. ± 1 . 1 ) stands for the Clebsch–Gordon coefficient.70) ‘Scalar’ implies invariance under rotations and is conventionally generalized to invariance under other symmetry trasnformations. In the notation developed in Chapter 5 holds for the singlet state | . (11. as contravariant spinors and to φα .65) where | · · · 1 describes the spin state of “particle 1” and | · · · 2 describes the spin state of “particle 2” and (0. 0| 1 .69) (11. this scalar product is anti-symmetric.

72)] φα φ α = = = = αβ αβ φβ φβ 11 21 11 21 12 22 12 22 (11.81) .63.80) One can write in matrix notation aα β γ δκ αγ a δ αβ (11.74) = = 0 1 −1 0 0 −1 1 0 (11.67) will be expressed as φα χα = φ1 χ1 + φ2 χ2 = φ1 χ2 − φ2 χ1 . we will express (11.72) 0 −1 1 0 Exercise 11.70) is φ1 φ2 −1 363 = = −1 φ1 φ2 . (10.78) aβ γ γδ φδ (11.3. 10.3: Spinors as can be verified from (11. 11.73.75) is given by φα = as can be readily verified.74)]. (11.71) (11. g µν of the Minkowski space [cf. The transformation behaviour of φα according to (11.69. In the transformed frame holds φ α χα = aα β αγ aγ δ δκ φβ χκ . (11. The inverse of (11. 11.77) (11.68). 11. Accordingly.73) (11. (10.69.76) αβ αβ The scalar product (11.11.70) and (11. 11. For this scalar product holds φα χα = −χα φα . (11. 11.79) = a −1 T a κβ .72) in a notation analogous to that chosen for contravariant and covariant 4–vectors [cf.71.1: Show that for any non-singular complex 2 × 2–matrix M holds M −1 = det(M ) M −1 T The matrices . Proof that φα χα is Lorentz invariant We want to verify now that the scalar product (11.77) is Lorentz invariant. −1 connecting contravariant and covariant spinors play the role of the metric tensors gµν .75) (11.10.

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. Hence. k = 1. a∗ (z) provides a representation of the Lorentz group which is distinct from that provided by a(z). the ˙ conjugate complex spinors φα need to be considered separately from the spinors φα .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. σ † = (σ ∗ )T = σ.364 Using (11.84) (11.85) φ1 φ2 ∗ ∗ .80) φ α χα = φβ χβ . i.90) extending the summation convention to ‘dotted’ indices.63). ˙ (11. 2. We denote ˙ (aα β )∗ = aα β ˙ (11. ˙ We also define covariant versions of φα φα = (φα )∗ .2) and (11. from (11.86) A concise notation of the conjugate complex spinors is provided by ˙ (φα )∗ = φα = φ1 ˙ φ2 ˙ ˙ (11.89) such that (11.e. it holds φk = (φk )∗ ..91) .87) which we will employ from now on.88) which one verifies taking the conjugate complex of (11.81) yields aα β and. hence.83) Here we have employed the hermitian property of σ. 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. (11. Insertion of (11.88) reads ˙ ˙ φ α = aα β φβ ˙ ˙ (11. Obviously. The transformation ˙ behaviour of φα is ˙ φ α = (aα β )∗ φβ ˙ (11.83) into (11. As discussed above.

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 =  (11.97) The transformation. The transformation behaviour of φα is ˙ φα = ˙ ˙ β γδ ˙˙ φδ ˙ ˙ αβ a γ ˙ ˙ . in matrix notation.96) is Lorentz invariant. a property which is rather evident. one can state ˜ and (11.94) (11. A tensor tα1 α2 ···αk β1 β2 ···β (11.93) (11.75..e. A comparision of (11. Ψ4 . For the spinors φα and χα thus ˙ defined holds that the scalar product ˙ φα χα = φ1 χ1 + φ2 χ2 ˙ ˙ ˙ ˙ ˙ (11.11.75).100) . + 1 2 2 2 1 transforms under rotations (z = iϑ) like a spin– 2 state | 1 . i.95) ˙ where αβ and αβ are the real matrices defined in (11.18) ˜ 3 . is governed by the operator a∗ (z) −1 which arises in ˜ the Lorentz transformation (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 )   =   = . a∗ (z) −1 accounting for the ˜ transformation behaviour of the third and fourth spinor component of Ψ.73. (11. − 1 .18) of the bispinor wave function Ψ.99)  Ψ3 (xµ )   χ ˙ (xµ )  ˜ χβ (xµ ) ˙ 1 ˜ χ2 (xµ ) Ψ4 (xµ ) ˙ 11.97) implies then that φα transforms like Ψ ˙ φ1 ˙ φ2 ˙ transforms under rotations (z = iϑ) like a spin– 1 state | 1 . 11. 11.4 Spinor Tensors ˙ ˙ ˙ We generalize now our definition of spinors φα to tensors.92) (11.4: Spinor Tensors 365 The relationship between contravariant and covariant conjugate complex spinors can be expressed in analogy to (11. (11.76) φα ˙ ˙ φα αβ ˙ ˙ αβ ˙ ˙ = = = = αβ ˙ ˙ αβ ˙ ˙ αβ αβ φβ φβ ˙ ˙ (11.

This tensor transforms according to t αβ = aα γ aβ δ tγ δ ˙ This reads in matrix notation.e.101) An example is the tensor tαβ which will play an important role in the spinor presentation of the Dirac equation..4. this transformation behaviour is in harmony with the tensor notation adopted.m aj akm t m (11. An obvious candidate is [cf. αβ = αβ (11.e. 11.102.102) = .25) the tensor is explicitly Aαβ = ˙ A1 1 A1 2 ˙ ˙ A2 1 A2 2 ˙ ˙ = A0 + A3 A1 − iA2 A1 + iA2 A0 − A3 . αβ ˙ αγ t ˙ γβ (11.106) Exercise 11. i.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 .e. with ˙ contravariant indices αβ. We had demonstrated that M (Aµ ) transforms according to M = M (Lµ ν Aν ) = a M (Aµ ) a† (11.103) A αβ = aα γ aβ δ Aγ δ . i. An example of a tensor is αβ and tions. This task implies that we seek a tensor.108) Obviously.. using conventional matrix indices j. ˙ ˙ ˙ ˙ (11. tjk = a t a† jk ˙ ˙ ˙ ˙ (11. k. the transformation bevaviour of a tensor tαβ reads in spinor and matrix notation t αβ = aαγ aβδ tγδ . m.105) This tensor is actually invariant under Lorentz transforma= αβ .104) Indices on tensors can also be lowered employing the formula tα β = and generalizations thereof.25] M (Aµ ) = σµ Aµ . (11.107) which reads in spinor notation [cf. the elements of which are linear functions of Aµ .1: Prove equation (11. According to (11. tjk = a t aT jk = . .106). The 4–vector Aµ in spinor form We want to provide now the spinor form of the 4-vector Aµ .109) . (11. ˙ ˙ ˙ ˙ (11. (11. it holds αβ . (11..m aj a∗ t km m .103) Similarly. i. we want to express Aµ through a spinor tensor.

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. Hence.116) . (11.113) Aα β = ˙ (11.110) one obtains Aα β = ˙ A0 − A3 −A1 − iA2 −A1 + iA2 A0 + A3 A0 + A3 A1 + iA2 A1 − iA2 A0 − A3 . Aµ and tensors tαβ can be applied to the partial differential operator ∂µ .4: Spinor Tensors One can express Aαβ also through Aµ Aα β = ˙ ˙ 367 A0 − A3 −A1 + iA2 −A1 − iA2 A0 + A3 .117) ˙ ˙ ∂0 − ∂3 −∂1 + i∂2 −∂1 − i∂2 ∂0 + ∂3 .110) one can state ∂ αβ = Similarly.111) A11 A12 ˙ ˙ A21 A22 ˙ ˙ ˙ 0 −1 1 0 (11.115) is correct.112) A22 −A21 ˙ ˙ −A12 A1 1 .110) The 4-vectors Aµ . 11. (11.11. ∂µ in spinor notation The relationship between 4–vectors Aµ .2: Prove that (11. (11. ˙ 2 (11. Using (11. (11.4.109. 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 . employing (11. (11.114) yields ∂αβ = ˙ ∂0 + ∂3 ∂1 + i∂2 ∂1 − i∂2 ∂0 − ∂3 .115) Exercise 11.

s. (11. Comparision of (10. γµ = ˜ 0 µ )∗ (σ −1 0 σµ σµ 0 (σ µ )∗ −1 . (11. .243) in the chiral representation expressing S(Lη ξ ) by (11. We will obtain the transformation behaviour of σ µ . To obtain the transformation properties of σ µ we employ then (10.26) and its contravariant analogue σ µ σµ = σµ = 1 0 0 1 1 0 0 1 .123) (a∗ )−1 Lν µ a L ν µ = = σν (σ ) ν ∗ −1 (11. (11. i.e.124) constitutes the essential transformation property of σ µ and will be considered further.229) and (11. Note that (10.368 σµ in Tensor Notation Spinor Formulation We want to develop now the tensor notation for σµ (11.f. −1 0 0 1 (11.124) . (10.243) 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.. .124).126) which is the complex conjugate of (11.h. one can conclude aσ µ ∗ −1 Lν µ σν 0 = . (11. 0 −i i 0 . 0 1 1 0 . σµ .15) one can write σµ = and. Equation (10.243) holds independently of the representation chosen. (11.125) is equivalent to (11. σµ building on the known transformation behaviour of γ µ . of this equation is 0 ∗ (σ µ )∗ a and. hence.120) σµ 0 . holds also in the chiral representation. 1 0 0 −1 . .121).124).118) 0 −1 −1 0 0 i −i 0 For this purpose we consider first the transformation behaviour of σ µ and σµ . (11.119) (11.121) One expects then that the transformation properties of σ µ should follow from the transformation properties established already for γ µ [c.e. i.122) aσ µ −1 a−1 −1 (a∗ )−1 0 Lν µ (11. This is ˜ possible since γ µ can be expressed through σ µ . hence.243)]. Hence.125) is equivalent to µ ∗ −1 −1 a∗ (σ µ )∗ −1 −1 a Lν µ = (σ ν )∗ (11.118) yields ˜ γµ = ˜ Using (11..18) and γ µ by (11.125) a (σ ) Equation (11.

11.124) a σ µ [a∗ ]T Lν µ = σ ν .2) −1 (a∗ )−1 = e2z 1 ∗ ·(σ ∗ )T = e2z 1 ∗ ·σ ∗ T = [a∗ ]T . 11. under rotations the Pauli matrices transform like (j = 1.132) 1 0 0 1 µ=0 .131) where a is given by (11. (11.2. −1 0 0 1 µ=3   .e.107) to (11. e. 3) σj −→ a(iϑ) σj a(iϑ) † = a(iϑ) σj a∗ (iϑ) T . In fact. (11.4).2.16. (11. state that σ µ in a new reference frame is σ µ = a σ µ a† (11. (σ ∗ )T = σ yields using (11. i.11.103) ˙ identifies σ µ as a tensor of type tαβ .   =  .130) We argue in analogy to the logic applied in going from (11. 0 −1 −1 0 µ=1 . (11. 11. 2.4: Spinor Tensors One can rewrite (11.2) −1 369 (a∗ )−1 = −1 − 1 z ∗ ·σ ∗ 2 e = e− 2 z 1 ∗ · −1 σ ∗ = e2z 1 ∗ ·σ .  (11.102.g. therefore.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 (11.124) using (11.133) Combining (11. This transformation behaviour. 1 0 0 −1 µ=3   .134) .. (11.108) that the same transformation behaviour applies then for general Lorentz transformations. One can.128) One can express..129. equation (11.127) Exploiting the hermitian property of σ. according to (11. hence.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. 11.4) with w = 0. 0 −i i 0 µ=2 . 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 . 0 1 1 0 µ=1 . transformations (11.

This yields   1 0 0 1   0 0 0 0 σ0 + σ3 σ1 + iσ2  (11.136) σαβ = = 2 ˙   σ1 − iσ2 σ0 − σ3 0 0 0 0 1 0 0 1 where on the rhs. the transformation is inverse to that of ordinary 4–vectors.133) is characterized through a 4-vector ˙ index µ.114). We can.h.136) results in the succinct expression 1 2 ˙ γδ ˙ (11. The ˙ desired change of representation(σµ )αβ −→ (σµ )αβ corresponds to a transformation of the basis of ˙ spin states f g f −→ = (11. Spinor Formulation (11.. of (11. in fact.137) σαβ ˙ = δαγ δβ δ .224) ˙ Using (11. ˙ ˙ 0 −1 1 0 ˙ (11. We will now consider the representation of σ µ .e.. 1. the submatrices correspond to tαβ spinors.29)].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 want to express now σ µ and σµ also with respect to the 4–vector index µ in spinor form employing (11. ˙˙ (11. (11.s. µ = 0. whereas ˙ account for the elements of the individual Pauli the ‘outer’ contravariant spinor indices.e. γ. 2. ˙ ˙ In (11. account for the 4–vector index µ. Each of the 4 × 4 = 16 matrix elements in (11.e.135) This is the property surmised already above [cf. i. and the covariant indices are moved outside.132) and (11. σµ in which the contravariant indices are moved ‘inside’.139) g −f g and.370 Inverting contravariant and covariant indices one can also state Lν µ σµ = σν . matrices. account for the 4-vector µ.12) for of σµ yields together with (5. hence. δ. β.140) where we employed the expressions (11. We can summarize that σ µ and σµ transform like a 4–vector.138) the ‘inner’ covariant spinor indices.. =  σαβ ˙  ˙ ˙ ˙ ˙ (σ21 )11 (σ21 )12 (σ22 )11 (σ22 )12   ˙ ˙ ˙ ˙ ˙ ˙ ˙ ˙ 21 (σ )22 21 (σ )22 (σ21 ) (σ22 ) ˙ ˙ ˙ ˙ 21 22 Equating this with the r. i.110) to express σ αβ in terms 0 −1 1 0 0 0 0 1     (11. 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 . i. α. however. 3 as well as through two spinor indices αβ.138) Note that all elements of σαβ are real and that there are only four non-vanishing elements.

121) of γ ˜ ˜ ˙ basis |αβ whereas the element of γ µ given by σ µ is in the basis |αβ .146) Exploiting the property (11.140) to transform each of the four submatrices.142) = This can be expressed 1 2 σ αβ ˙ ˙ γδ = δαγ δβ δ . which in ˙ (11. ˙˙ (11.11. For this purpose we start from the ˜ µ .4: Spinor Tensors 371 and. we write ˜ ˙ γµ = ˜ 0 (σ µ )αβ µ ) )∗ ((σ αβ 0 ˙ ˙ . employing then transformation (11. (11. (11. Accordingly. One can write then the scalar product using (11.115) 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 .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.143) Combined with (11.147) .  ˙ ˙ ˙ ˙       (11. This expression implies that the element of γ µ given by σ µ −1 is in the expression (11.144) results in the simple relationship γ µ Aµ = ˜ 0 Aα β ˙ Aα β 0 ˙ .138) one can conclude that the following property holds 1 2 ˙ γδ σ αβ ˙ = 1 2 σ αβ ˙ ˙ γδ = δαγ δβ δ .145) Let Aµ be a covariant 4–vector. ˙˙ (11.

5 Lorentz Invariant Field Equations in Spinor Form Dirac Equation (11.149) (11.147) allows us to rewrite the Dirac equation in the chiral representation (10.148) ˜ Employing Ψ(xµ ) in the form (11.150) . (11.226) ˜ i γ µ ∂µ Ψ(xµ ) = 0 i ∂ αβ ˙ i ∂ αβ 0 ˙ ˜ ˜ Ψ(xµ ) = m Ψ(xµ ) . α ˙ = = m φα m χβ .372 Spinor Formulation 11.99) yields the Dirac equation in spinor form i ∂ αβ χβ ˙ i ∂ αβ φ ˙ The simplicity of this equation is striking. ˙ (11.

1 Symmetry and Degeneracies The degeneracies of energy levels of a quantum mechanical system are related to its symmetries. (12.e.Chapter 12 Symmetries in Physics: Isospin and the Eightfold Way by Melih Sener and Klaus Schulten Symmetries and their consequences are central to physics. 373 (12. In the following section we will introduce the concept of isospin as an approximate SU (2) symmetry. will commute with the Hamiltonian of the system. 12.1) Then the generators. 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. we will discuss the SU (3) symmetry of three quark flavors. Let us assume a continuous symmetry obeyed by a quantum mechanical system. the so-called Eightfold Way. These are the isospin symmetry of nuclear interactions and its natural extension. Sk ] = 0. We will apply these concepts to an analysis of nucleon-nucleon and nucleon-meson scattering..2) . which identifies the proton and the neutron as different states of the same particle. [H. O = exp k αk Sk . The action of the symmetry operations on quantum mechanical states are given by elements of a corresponding Lie group. i. In the final section. 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. We will particularly use the example of spherically symmetric potentials. 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. Sk .

. n − 1. can be identified with the angular momentum operators.12). 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 . · · · .m (r) Y m (θ.λn . Jk .. Classically.. Presence of additional symmetries may further increase the degeneracy of the system.. this implies a degeneracy in the spectrum.λn .8) ˆ [H. 2. is allowed to take values in 0. namely motion in a spherically symmetric potential described by the group SO(3) (or its double covering SU (2) as discussed in section 5. 3 . leaves the energy of the state invariant H exp(iαk Sk ) ψE. has been discussed. This degeneracy follows from the fact that any rotation.18).6) From elementary quantum mechanics we know the spectrum of the hydrogen atom.3) If the newly obtained state is linearly independent of the original one. (12. . We will investigate this shortly in more detail in the case of systems with spherical symmetry.5) with m = −l.λ1 . k = 1. ψE. Sk .7) 2 n where the orbital angular momentum... on an energy eigenstate. (12.λ1 .m (r) = vE. .. φ). (12.λ1 . 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. the states 3s. Jk ] = 0 . as represented by an element of SO(3). The spherical symmetry implies the commutation of the Hamiltonian with angular momentum operators (7.. which yields a set of eigenstates of the form ψE.. As an example of this we will consider the Coulomb problem with the Hamiltonian H = p2 k − . The energy levels are mk 2 En = − 2 2 . (7. V (r).. 2m r (12. Lie groups play an essential role in the discussion of mass degeneracies in particle physics. where the symmetry generators. we first consider a particular example of the implications of symmetry. (12. For example. Sk . In chapter 7 the dynamics of a particle moving in three dimensions under the influence of a spherically symmetric potential. can then be reduced to a one-dimensional radial o equation. .. In order to illustrate this..4) The stationary Schr¨dinger equation. (7.374 Spinor Formulation The action of any symmetry generator. Therefore.39). . each energy level is (2l + 1)-fold degenerate. generates a state which has the same energy as the original one. l and the corresponding energy levels are independent of m. 3p and 3d all have the same energy.8) .λn = E exp(iαk Sk ) ψE. The energy levels are totally independent of l. (7.. m r (12. . l. as studied in chapter 7. .24).

10) (12.8) is not a hermitian operator. ijk Jk . 2 which can be shown to satisfy [Ai .13) It can be verified explicitly that its components commute with the Hamiltonian.17) (12. The vector in (12. 2E (12. In the following we consider the bound states.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. We will also need the following identity [4] 2 = 2H J2 + m 2 + k2 . (12.21) .12) (12. 2m r 1 x3 (J2 p1 − J1 p2 + i p3 ) − k . 2m r (12.19) (12. 2 1 B = (J − K).18) which complement the familiar angular momentum algebra of section 5. Therefore. which have a negative energy E. In order to understand the aforementioned extra degeneracy. 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 .15) which can be proved after very considerable algebra. Jj ] = i [Ki .14) This follows from a · (a × b) = (a × b) · a = 0. 2m r 1 x1 (J3 p2 − J2 p3 + i p1 ) − k . (12.3.9) (12.11) (12. Now we scale the eccentricity vector as follows K= − m .20) (12.11. 2m r 1 x2 (J1 p3 − J3 p1 + i p2 ) − k . which is valid even when a and b do not commute. in the subspace of the Hilbert space corresponding to a certain energy we can replace H by E. (12. Aj ] = i ijk Ak . Kj ] = i ijk Kk . (12. For this purpose we first note that J · = 0. We introduce the following new operators 1 A = (J + K). we will compute the hydrogen spectrum using the additional symmetry.16) Through some algebra [4] the following commutation relations can be verified [Ki .

29) (12. . H B. Using this equation the energy eigenvalues can be written in terms of the eigenvalues of A2 and B 2 operators. . ijk Bk . This implies that a = b. . .15). . l. In contrast to the discussion above about extra symmetries. 2 (12. H = 0. a + b = n − 1}.7) tells us that (2a + 1) = n. 1. Spinor Formulation (12. . Bj ] = 0. Bj ] = i [Ai . b = 0. a lack of symmetry implies a lack of degeneracy in the energy levels of a quantum mechanical system. Chaos is described classically as exponential .28) . . a = 0.23) (12. Furthermore the bound on the orbital angular momentum. .34) It follows that l has to have values in {0 = |a − b|.24) (12. the energy eigenvalues are found to be E=− mk 2 1 .33) (12.376 [Bi .26) (12. (12.27) (12. . 1.31) where we have used (12. By comparing to the rotation algebra introduced in chapter 5. 1.21) and (12. Noticing that A2 and B 2 have the same eigenvalues because of (12.22) (12. . a = 0. This illustrates the effect of additional symmetries to the degeneracy structure of a quantum mechanical system.28). we can read off the eigenvalues of A2 and B 2 from (12.22): A2 = a(a + 1) B 2 = b(b + 1) Following (12. .14) we note that A2 − B 2 = J · = 0. = 0. namely that J J > < A − B =0 A + B =2 A (12. 1. can be seen to follow from the triangle inequality as applied to J = A + B. . . 2 1 .32) A comparison with (12. In order to arrive at the spectrum a final bit algebra is needed A2 + B 2 = J 2 + K 2 m 2 = J2 − 2E mk 2 1 = − − 4E 2 (12. .25) So far we have shown that the symmetry generators form an algebra. 2 (2a + 1)2 2 2 (12.30) 2 2 2 1 . . which is identical to the the direct sum of the Lie algebra of of two SO(3) (or SU (2)) algebras. . . . A. . The most extreme case of this is the quantum analogue of a classically chaotic system.

In the next section we will proceed with the discussion of a symmetry. For a more detailed discussion of quantum chaos in billiard systems we refer the reader to [7] and the references therein. This was motivated by the observation that their masses are approximately equal: mp = 938. 12. which was discovered by observing degeneracies in the particle spectrum. (12.36) In analogy to the concept of spin regarding the rotations in 3-space as discussed in chapter 5. This enables us to utilize what we already know about the SU (2) symmetry group from the study of angular momentum. another manifestation of chaos is the lack of independent operators commuting with the Hamiltonian. However.) If the proton and the neutron are to be viewed as two linearly independent states of the same particle.35) this mass equivalence can be viewed as an energy degeneracy of the underlying interactions.2 system p= 1 0 . Originally it was believed that isospin was an exact symmetry of strong interactions and that it was violated by electromagnetic and weak interactions.5: Lorentz Invariant Field Equations 377 sensitivity to initial conditions. the classical trajectories will diverge from each other after successive bounces from the boundary.57M eV /c2 . 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. (Strong interactions bind the atomic nucleus together. A typical example of this so called quantum chaos is the quantum billiard problem. the isospin symmetry is also governed by an SU (2) group ‘rotating’ components in (12. Following the mass-energy equivalence of special relativity E = mc2 .36) into each other in abstract isospin space. mp = 939. in the sense that nearby trajectories in the phase space diverge from each other over time. namely. analogous to the 1 spin-up and spin-down states of a spin. n= 0 1 . for the . This is known as level repulsion.2 Isospin and the SU (2) flavor symmetry The concept of isospin goes back to Heisenberg. after the discovery of the neutron in 1932.28M eV /c2 . It is important to place the isospin concept in its proper historical context. for example. it is natural to represent them in terms of a two component vector. (Weak interactions are responsible. that the proton and the neutron behave identically under the so-called strong interactions and that their difference is solely in their charge content.11. For example. 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. This (approximate) degeneracy led into the idea of the existence of an (approximate) symmetry obeyed by the underlying nuclear interactions. who. which is a particle in box problem in two dimensions with a boundary which can be chosen arbitrarily. If the chosen boundary is ‘irregular’ in a suitably defined sense. suggested that the proton and the neutron can be regarded as two states of a single particle. (12.

(mπ± = 139. The mass difference between the neutron and the proton could then be attributed to the charge content of the latter. Further than that. 1 π 0 = |1.39) All other isomultiplets. as good quantum numbers. π 0 = |1. the proton is the lighter of the two. If it were otherwise the proton would be unstable by decaying into the neutron. . In the framework of the quark of model. 2 2 1 1 1 . 0 π − = |1.− . it can be seen as part of a larger (and more approximate) symmetry which is of great utility to classify observed particle families.43) Similarly.36) as a multiplet with I = 2 1 1 p = I = . If the mass difference (or the energy difference) were to be to be purely electrostatic in nature. including the spin and color quantum numbers of their constituent quarks. 2 2 1 2 2 2 1 1 1 1 1 √ . d= 1 1 . 0 .− 2 2 . albeit it is a good approximate one. .38) which have all nearly identical masses. .11.− 2 2 . including the proton and the neutron.0M eV /c2 ).40) + 2 1 1 . mπ0 = 135. −1 . −1 = = = 1 1 1 1 . They u form an isotriplet: π + = |1. . the proton and the neutron can be written as totally symmetric uud and udd states. π − = |1. We can describe isospin multiplets the same way we have described the angular momentum and spin multiplets. sec.6M eV /c2 . are made up of these two quarks. the fundamental representation of the isospin symmetry corresponds to the doublet that contains the so-called up and down quarks u= 1 1 . I3 = 2 2 1 1 . Shortly we 1 will see how to describe both the pion and nucleon states as composites of more fundamental I = 2 states. Isospin symmetry is not an exact symmetry of strong interactions.− 2 2 1 2 2 2 1 1 1 1 . (12. we refer the reader to [1].− . respectively. For example. 2 2 1 2 2 2 (12.41) (12. 2 2 . Therefore it remains a useful concept. as we shall see below. spelling disaster for the stability of matter. Denoting the total isospin. However. the three pions in (12. I3 . n = I = . I. 1 . (12. the proton had to be heavier. 1 we can re-write (12.37) As an example of a multiplet with I = 1 we have the three pions or π-mesons π + = |1. (12. 2 (12. They can be constructed with the same rules that have been used for angular momentum addition ¯ u in chapter 6. 2. and its third component. u¯ and d¯ states.42) (12.38) are ud. For a precise description of the two nucleons as composite states.378 Spinor Formulation beta decay). I3 = − 2 2 .

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).46) (12. I3 = 0. Now let us consider another example of combining the isospins of two particles.47) (12. the neutrinos etc. 0 = √ (|p > |n > −|n > |p >) 2 and that of an isotriplet |1.43). and quarks have baryon number 1 . This is why isospin is such a good symmetry and why isomultiplets have nearly identical masses. a hydrogen isotope. but at the time the empirical concepts like isospin and strangeness quantum numbers were in use.44) 2 where B is the baryon number and S is the strangeness. . All other composite states have their strangeness given by the sum of the strangeness content of their constituents. By convention ¯ baryons have baryon number 1.37) and in analogy to (12. Naturally.) are light. the up and down quarks are not the only quark ‘species’ . a number strange particles have been found which presumably contained a third quark species: the strange quark.45) Is the deuteron an isosinglet state or an isotriplet? If it were an isotriplet (|1. Before proceeding further.11. 2 = |n > |n > . this remains only an inaccurate naming convention today. where leptons (electron. The up and down quarks have strangeness zero. Following (12. such as proton and the neutron. It shall be noted here that the quark model was not invented until 1960’s. −1 = |p > |p >. originally refer to the relative weight of particles. to be −1 for the strange quark. 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. by accidental convention. In the late 1940’s and early 1950’s. whereas mesons are q q states. The names. mesons have baryon number 0. We will now try to describe its wave function in terms of its constituent nucleons. we shall setup some terminology: baryons are qqq states. (12. 0 |1. (12.e. 1 |1. 0 ). In the next section we will be able to view the Gell-Mann–Nishijima relation in the light of the representation theory for the flavor SU (3) symmetry. If taken literally. Therefore it has to have isospin. mesons have intermediate mass ranges. The value of the strangeness quantum number is taken. 1 = √ (|p > |n > +|n > |p >). baryons generally are heavy. baryon and meson. All antiparticles have their 3 quantum numbers reversed. The reader may know that the deuteron. this will be mathematically identical to adding two spins.or flavors as they are commonly called. The possibilities are that of an isosinglet 1 |0. strongly interacting particles) which is about a GeV /c2 .48) (12. as some mesons discovered later are heavier than some baryons and so on. the pions being examples thereof. 0 ) we should have seen nn and pp bound states of comparable energy in nature (because of isospin symmetry). consists of a proton and a neutron. Therefore the deuteron has to be an isosinglet state (|0. muon. As it later turned out.

1 and (1/ 2)(|1.7.45) and (12.8 to compute the ratio of the scattering amplitudes MI and MII .55) . Refer to exercise 6. . is proportional to |M|2 .7 and 6. σ. (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. γ (i) I (f ) . γ (f ) T00 I (i) = 1. γ (f ) (f ) (i) (12.7 to the case of isospin. I3 . 1 4 in an isotriplet state and − 3 in an isosinglet 4 Using (12. γ (i) is a reduced matrix element defined in the same sense as in section 6. γ (i) (i) (12. We will eventually be able to compute ratios of scattering cross-sections between different processes. Show that I(1) ·I(2) is an ‘isotensor’ of rank zero.259) discussed in detail in sections 6. 1 and |1. γ (f ) |T00 |I (i) I3 . Exercise.8. For completeness let us start by restating the Wigner-Eckart theorem (6. I3 . The dot product here refers to the abstract isospin space. consider (I) p + p → d + π + (II) p + n → d + π 0 (12.53) 0 √ 1 2I (i) +1 I (f ) . which is an isoscalar. I3 (f ) = 1. γ (f ) ||T ||I (i) . 0 + |0.48) the initial states in (I) and (II) have isospin values |1. (12. √ According to (12. γ (i) = (f ) (i) (I (f ) I3 |00I (i) I3 ) (f ) (i) (−1)I −I (f ) (i) (12.38) and the fact that the deuteron is an isosinglet we know that the isospins of the final states in (I) and (II) are |1. such as the spatial dependence of the wave function and spin. For example. (f ) (i) (I (f ) I3 |00I (i) I3 ) is a Clebsch-Gordon coefficient and I (f ) .51) (12. respectively. The initial and final states can be denoted in more detail as | initial | final = = I (i) . The cross-section.50) . γ (i) . γ (f ) ||T0 ||I (i) . 0 . Consider the generalization of tensor operators discussed in section 6. where M is the scattering amplitude given by M = final | α I(1) · I(2) | initial .49) The only assumption that we put in will be that the interaction is of the form V = α I(1) · I(2) . Show that the expectation of I(1) · I(2) is state. 0 ). as discussed in the exercise above.52) where γ (i) and γ (f ) denote degrees of freedom other than isospin.380 Spinor Formulation As an exercise on the implications of isospin symmetry we will consider nucleon-nucleon scattering.259) in the present context: I (f ) I3 . I3 = 1.54) Here T00 ≡ V = α I(1) · I(2) . Exercise. We will now employ the (isospin analogue) of the Wigner-Eckart theorem (6.5 for the spin-analogue of the same problem.

but these are not dominant at relatively low energies. −1 2 3 2 − + .64) There are more exotic possibilities. . γ (f ) ||T00 || I (i) = 1.57) (12. I3 = 0. I3 = 0. We want to compute the ratio of total cross-sections assuming a similar interaction as in the previous example σ( π + + p → anything ) . (12. = 381 (12. which is also manifested by a vanishing Clebsch-Gordon prefactor. particles with strangeness.60) (12. 2 .61) MII Note that the second term in MII vanishes due to the isospin conservation. The relevant Clebsch-Gordon coefficients are easily evaluated: (11|0011) = (10|0010) = 1. It follows σI = 2. σII which is in approximate agreement with the observed ratio.56) (12. −2 1 1 √ . [2] As a further example.62) We can now write the ratio of the scattering amplitudes: MI 1 √ . γ (i) = √ 2 1 (f ) (i) +√ I (f ) = 1. (c) π − + p → π 0 + n .11. σ( π − + p → anything ) The possibilities are (a) π + + p → π + + p .5: Lorentz Invariant Field Equations 1 (11|0011) √ I (f ) = 1. 1 π − + p : |1. 1 1 1 1 3 √ 2. −2 2 1 1 3 2. we will consider pion-nucleon scattering. I3 = 0. γ (i) 2 1 = (10|0010) √ I (f ) = 1.66) = 2 3 1 3 2.65) (12.59) (12. γ (i) 3 1 (f ) (i) I (f ) = 1. γ (f ) ||T00 || I (i) = 1. 2 = 3 2. which are obtained by a standard Clebsch-Gordan expansion π + + p : |1. −1 π0 + n : |1. (b) π − + p → π − + p . 2 = 2. γ (i) 3 +0. γ (f ) T00 I (i) = 1. −2 . Once again we need the isospins for the initial and final states. (12. involving. γ (f ) T00 I (i) = 0. −2 1 1 2. = MII (1/ 2) (12. for example.63) where common dynamical factors (which would not be as easy to compute) have dropped out thanks to the Wigner-Eckart theorem. I3 = 0.58) (12. 0 1 1 3 3 2. (12.

was motivated by the existence of particles that are produced strongly but decay only weakly.d =  1  . (12. The classification of the newly found particles as members of some higher multiplet structure was less obvious then the case of isospin. . 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.s =  0  . It was assumed that strong interactions conserved S (at least approximately). m V V 3 2. For instance. m 1 2. has a lifetime which is comparable to that of π + albeit being more than three times heavier. Hence Gell-Mann and independently Nishijima postulated the existence of a separate quantum number. [2] (12. while weak interactions did not.67) which are independent of m. A computation similar to the previous example of nucleon-nucleon scattering yields (apart from common prefactors) the following amplitudes for the reactions in (12. are up to 40% heavier. it appears an obvious generalization to add another component for an extra quark to the isospin vector space       1 0 0 (12. such that S(K + ) = 1. which can be produced by π − + p → K + + Σ− . S. 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. giving it a higher than usual lifetime. In the light of the quark model.65) Ma = M 3 2 Mb = 1 M 3 + 2 M 1 3 3 √ 2 Mc = 2 3 M3 2 √ 2 (12. K + .69) 12. Hence the strangeness changing strong decays of K + (or Σ− ) were forbidden. etc.3 The Eightfold Way and the flavor SU (3) symmetry The discovery of the concept of strangeness. m 1 2.382 Spinor Formulation As in the example of nucleon-meson scattering we define the relevant matrix elements M3 = 2 M1 = 2 3 2. S(Σ− ) = −1 and S(π) = S(N ) = 0. called strangeness. however. Strange partners of the familiar nucleons. mentioned in the previous section.70) (12.68) − 2 3 M1 2 Guided by empirical data we will further assume that M 3 >> M 1 .71) u =  0  . 0 0 1 . m . for example.

we observe that (U V )† = V † U † = V −1 U −1 = (U V )−1 . (The ‘bare’ mass of the charm quark is already heavier than the two nucleons. called quantum chromodynamics. ak . However. (12. Then came the question of what the fundamental representation. we want to find those transformations ak → Ukl al that preserve the norm. Multiplying U by a phase.74) for any two unitary matrices U and V . The unitarity relation imposes 9 constraints on the total of 18 real degrees of freedom of a 3 × 3 complex matrix. history followed the reverse of this path. The bottom and top are even heavier. should correspond to. becomes increasingly inaccurate for Nf > 3. bottom (beauty) and top (truth) are significantly heavier than the hadronic energy scale. The reason is that the other known quarks. which set the hadronic energy scale. of three dimensions. Therefore U (3) can be decomposed into a direct product U (1) × SU (3) where SU (3) consists of 3 × 3 unitary matrices of unit determinant. eiφ . However SU (3) symmetry appears in another and much more fundamental context in strong interaction physics. in fact modern theory of strong interactions is a ‘gauge theory’ of this color group.72) a∗ ak . The reader is also invited to revisit section 5. but there are a number of subtle differences. [2] ) Before discussing the significance and the physical implications of the quark model. the same way nucleons and pions were identified as representations of the isospin SU (2) symmetry. SU (N ). as in (12. of a.5: Lorentz Invariant Field Equations 383 In this case the transformations that ‘rotate’ the components of (12.1 whenever necessary. which again form representations of an SU (3) group. Hence the group U (3) has 9 dimensions. giving rise to the quark model. Flavor SU (Nf ) symmetry on the other hand. (12. called color. (The reader is referred to section 8. for a period they were considered as useful bookkeeping devices without physical content. In many respects it will resemble the more familiar group SU (2) discussed in some detail in chapter 5.11.73) 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). This group of 3 × 3 unitary matrices is denoted by U (3). The quarks posses another quantum number. First particle multiplets were identified as representations of the SU (3) group. namely charm. Because of this additional constraint SU (3) has 8 dimensions. while preserving the norm. It is seen that such a matrix U has to satisfy the following k U † = U −1 . it is the group . The reader shall note that most of what is being said trivially generalizes to other unitary groups. where Nf is the number quark flavors. In this perspective the flavor SU (3) symmetry may appear to be mainly of historical interest. Given a complex vector.71). in reference to Lie groups. As quarks were never directly observed. but we will stick to N = 3 in the following. unitary relation k (12. we will establish some mathematical preliminaries about the group SU (3). Lie algebras and related concepts. Since arbitrary phase factors are of no physical interest. still leaves the norm invariant. This is believed to be an exact symmetry of strong interactions.3 for a brief discussion of gauge transformations).71) into each other.

As in the case of SU (2) higher dimensional representations obeying the same structure can be found. λk ] = 2ifjkl λl . can be written in the form U = eiH where H is a hermitian matrix. obey the following relation tr(λj λk ) = 2δjk . (12.77)  λ1  . (12. λ7 =   0 0  . λk ]+ = δjk + 2δjkl λl . λ2 =     . 3 (12.79) (12. . λk . As discussed in section 5. any unitary matrix. (We shall at times refer to the Lie algebra with the name of the group. We can also introduce the constants. δjkl .1.) The unit determinant condition requires that all λk are traceless. the meaning being apparent from the context.384 Spinor Formulation SU (3) and not U (3) that is of main interest. regardless of the dimension of the representation. since det(eA ) = etrA . The reader is invited to compare the structure of SU (3) to that of SU (2) discussed in section 5.75) αk λk (12.79).   . −2 λ4 λ6 λ8 The generators. 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  . which can be verified explicitly as matrix identities. λ3 =  0 −1 0  . 0 0 0   . U . Therefore we will express elements of SU (3) as U = ei k (12.80) This is the fundamental or defining representation of SU (3).7. The fundamental relation to be preserved is (12.32). λ5 =   . The Lie algebra structure is given by the commutators of λk [λj .76) where λk are 8 linearly independent matrices forming the basis of the Lie algebra of SU (3).78) where fjkl are the antisymmetric structure constants similar to the jkl of SU (2) given in (5. via the anticommutator relation 4 [λj .

[Y.94) . Exercise. we have [Y. First. T0 ] = 0. which satisfied an ‘eigenvalue equation’ [J0 . (12. 2 Y = √ F8 . Y is called the hypercharge.85) (12. V± ] = ±V± . V± ] = ± V± . 2 1 [T0 . (12.92) (12. (Recall that the strange quark has S = −1 by convention). We first introduce the F-spin operators 1 Fi = λ i . 3 (12. U± = F6 ± iF7 .83) (12. (12.5: Lorentz Invariant Field Equations 385 As it turns out the set of generators in (12.86) (12.84) (12. starting with the observation that Y = B + S. [Y.90) (12. J− } proved useful. Using the convention in (12. V± = F4 ± iF5 . In the case of SU (2) the identification of ‘ladder’ operators {J+ . Derive the Gell-Mann–Nishijima relation (12. 2 (12. I3 . J± ] = ± J± . these relations have been used to construct the angular momentum spectrum as well as the function space representation of the rotation algebra. The generators of SU (3) can be arranged into a very similar form to that of SU (2).91) [T0 . U± ] = ±U± .93) (12.89) (12. T± ] = 0.71) show that T0 is the isospin operator. T± ] = ±T± .77) is not the most useful basis in the study of SU (3). 2 (12. T0 } and [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 . In the basis (12.87) Exercise. U± ] = U± .87) the commutation relations between the generators can be expressed in a succinct manner.88) which defines (not uniquely) a maximal set of mutually commuting operators {Y. T 0 = F3 .11.81) In chapter 5. namely spherical harmonics. 1 [T0 .44).

The interested reader is instead referred to a very readable account given in chapter 12 of [4]. the representations will now lie in a T0 − Y -plane. V+ ] = 0. it is said to be in Cartan-Weyl form. Another important property of SU (2) is the existence of an operator (namely the total angular momentum. An operator which commutes with all generators of a Lie group is called a Casimir operator. However.5.99) (12.97) which relate commutators of generators with opposite eigenvalues to the maximal set {Y. [T+ . [U+ . V− ] = 2 (12. V− ] = −U− . jkl (12. V+ ] = 0.101) (12. U+ ] = V+ . 2 3 Y + T0 = 2V0 . A formal definition and a detailed discussion of the Cartan-Weyl form is beyond the scope of this chapter.103) Any remaining commutators follow from hermiticity. as the relation between the states in a given representation can be conveniently expressed in terms of ladder operators. Note that. similar to the way it was done in section 5. 3 [U+ . As in the case of J 2 and SU (2). Thus. T− ] = 2T0 . we have [T+ .105) C2 = jkl djkl λj λk λl In general the number of independent Casimir operators of a Lie group is equal to its rank. U± } to this maximal set by ‘eigenvalue equations’ and [T+ .102) (12. . U− ] = 0. Finally. J 2 ) which commutes with all of the generators. while SU (3) has rank 2. U0 and V0 are linear combinations of T0 and Y . V± . This form is essential for easy labeling of the representations of the group. [T+ . Casimir operators can be used to label irreducible representations of the Lie algebra. We can construct two such independent Casimir operators for the group SU (3). C1 = k λ2 = − k 2i 3 fjkl λj λk λl . [T+ .100) (12. T0 }. When the basis of a Lie algebra is expressed in such a way to satisfy the form of the eigenvalue relations as given above. since there are two mutually commuting generators in SU (3) as given in (12. (12. [V+ . The maximum number of mutually commuting generators of a Lie algebra is called its rank. V− ] = T− . SU (2) has rank 1. In the case of SU (2) the representations lay on a line on the J0 axis. U− ] = Y − T0 = 2U0 . we will use these commutation relations to construct representations of SU (3).104) (12. [U+ .98) (12.96) (12.386 Spinor Formulation which relate the remaining generators {T± .88). The same way the angular momentum ladder operators have been used to construct the representations of SU (2).95) (12.

For example the pion octet (or any other meson octet) is therefore realized as states of a quark . y U± |t3 . For example.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 : T0 |t3 . y .112) The effect of these operators to the states in the y − t3 plane have been outlined in Fig. 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. 1) as an example.111) (12. 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. especially chapters 7 and 8. This property can be used to label representations in terms of the values of the Casimir operators. The interested reader is referred to [4].106) (12. Figure (12. (See Fig. respectively. |t3 . 1) correspond to the triplets of quarks and antiquarks. y V0 |t3 . The representations for hexagons with sides of length p and q in the T0 − Y -plane. 2 1 = β t3 ± . y V± |t3 . 2 1 = γ t3 . 2 (12. For example. 0) and D(0. U± . (12.11. y . we have U0 |t3 . 1) representation. 2 1 t3 ) |t3 . the pion family forms part of an octet corresponding to the D(1. . (12. y 3 = ( y− 4 3 = ( y+ 4 1 t3 ) |t3 . Now we will construct explicit representations of SU (3). Such a representation is labeled as 1 D(p. y 1 = α t3 ± .110) (12. V± . y . y . All other states of the representation are then constructed by successive application of ladder operators T± . (12.107) 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 . y ± 1 . Because of (12.3).) All other representations can be constructed by combining these two conjugate representations. = y |t3 .108) (12. As another example for the representations of SU (3).y ± 1 . it was shown in section 5. y .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. q) and it has a dimensionality of 2 (p + 1)(q + 1)(p + q + 2). For example.88) we can label states by the eigenvalues of T0 and Y operators.2) shows the representation D(2. 2 (12. y Y |t3 . y = t3 |t3 . The representations of SU (3) are constructed analogous to those of SU (2) by identifying the ‘boundary’ states annihilated by raising (or lowering) operators. The representations D(1.antiquark pair. y . A notation suggestive of the dimensionality of the representation can be used to identify representations. we have T± |t3 .1).

The states in the inner triangle are doubly degenerate.1: The effect of SU (3) ladder operators on the y − t3 plane.388 £ Spinor Formulation Figure 12. 1) representation of SU (3). £ Figure 12. ¡ ¢  ¡ ¢  8 6 2 '745301) ¥¦¤   ¤   #"  ! (& % '$   © ¨     ¦ §¥ ©§ ¨ .2: D(2.

5: Lorentz Invariant Field Equations 389 D(1. for example. £ (12. 2 3 2 0. where we would write [3] ⊗ [3] = [1] ⊕ [3] ⊕ [5]. 0) or the fundamental representation of flavor SU (3) symmetry. 3]. in the case of SU (2).113) The additional singlet state corresponds to the η meson. This expansion can be compared. . 0) and D(0. This way the quark-antiquark states can be represented as follows [3] ⊗ [¯ = [8] ⊕ [1] 3] (12. 3 (12. 1) are denoted by [3] and [¯ respectively.116) (12. . 2 3 1 1 − .118) ¡ ¢    ¨ ©§ ¨¥¦¤ "   ! .11.3: D(1. 2 (12. The three quarks in the fundamental representation can now be written as u = d = s = 1 1 .117) The Gell-Mann–Nishijima relation can then be succinctly expressed as 1 Q = y + t3 .115) (12.114) Figure 12. The octet D(1. − . 1) is written as [8] etc. to the case of adding two spin triplet states.

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

Sign up to vote on this title
UsefulNot useful