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

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

e.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 . ∂f in case of a function of M variables. q(t).2 Lagrangian Mechanics Note: δq(t) describes small variations around the trajectory q(t). but rather to a differential of the function which is simply the df differentiated function multiplied by the differential increment of the variable. It holds   M  M ∂L  t1 d ∂L ∂L δS[q(t). t0 (1. We focus on such functionals because they play a central role in the so-called action integrals of Classical Mechanics. δq(t)]. ˙ Theorem: Let S[q(t)] = t0 dx dt t1 ˙ dt L(q(t).7) through Taylor expansion and identification of terms linear in δqj (t). df = M ∂xj dxj . equating these terms with δS[q(t).e. q(t) + δ q(t). t) of the configuration d vector q(t). It is also important to appreciate that δS[ · .6) immediately. 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. the velocity vector dt q(t) and time t. t) (1.6) is expressed in terms of ‘normal’ functions. e.. q(t). δq(t)] = dt − δqj (t) + δqj (t)   ∂qj dt ∂ qj ˙ ∂ qj ˙ t0 j=1 j=1 t1 . t) + j=1 ∂L ∂L δqj + δ qj ˙ ∂qj ∂ qj ˙ + O( 2 ) (1.. (1. i.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. t1 ] where the integrand is a function L(q(t). For this purpose we consider M ˙ ˙ ˙ L(q(t) + δq(t).6) For a proof we can use conventional differential calculus since the functional (1.10) From this follows (1.5) where L( · . We attempt to evaluate t1 S[q(t) + δq(t)] = t0 ˙ ˙ dt L(q(t) + δq(t). j=1 We will now consider a particular class of functionals S[ ] which are expressed through an integral d over the the interval [t0 . ˙ In the following we will often use the notation for velocities and other time derivatives d q(t) = q(t) and dtj = xj . q(t) + δ q(t). q(t) + δq(t) is a ‘slightly’ different trajectory than q(t). · ) is a function differentiable in its three arguments. · ] in conventional differential calculus does not correspond to a differentiated function. i. . at the ends of the time interval of the trajectories the variations vanish.g. · . t) = L(q(t). dt q(t). t) (1. df = dx dx or.

it holds in case δq(t0 ) = δq(t1 ) = 0 that qe (t) is an extremal. from (1. · ] ≡ 0 and. 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 .13) dt f (t)h(t) = 0 t0 (1. . y(x1 ) = y1 . Let us assume that the two points are connected by the path y(x). hence. On the other side. According to the Lemma above follows then (1. then f (t) ≡ 0 on [t0 .15) We will not provide a proof for this Lemma.14) for any continuous function h(t) ∈ F with h(t0 ) = h(t1 ) = 0. δq(t)] = 0 for all δq(t) ∈ F .1. . the above theorem.17) . y1 ) and (x2 . y(x2 ) = y2 .12) for j = 1.6) which reads in the present case     M ∂L t1 d ∂L δS[q(t). M . 2.16) the property δS[qe (t). .1: Variational Calculus 3 We now consider the question for which functions the functionals of the type (1. .12) The proof of this theorem is based on the property Lemma: If for a continuous function f(t) f : [t0 . y2 ). 2. The length of such path can be determined starting from the fact that the incremental length ds in going from point (x.12) for j = 1.11) The extremals qe (t) can be identified through a condition which provides a suitable differential equation for this purpose. dx (1. This condition is stated in the following theorem. M ) d dt ∂L ∂ qj ˙ − ∂L = 0 ∂qj (1. y(x)) to (x + dx. if and only if it satisfies the conditions (j = 1.   ∂qj dt ∂ qj ˙ t0 j=1 (1. t1 ]. y(x + dx)) is ds = (dx)2 + ( dy dx)2 = dx dx 1+( dy 2 ) .5). 2. (1. For this purpose we define Definition: An extremal of a differentiable functional S[ ] is a function qe (t) with the property δS[qe (t).5) assume extreme values. . . . The proof of the above theorem starts from (1. (1. .16) This property holds for any δqj with δq(t) ∈ F . . t1 ] ⊂ R → R holds t1 (1. . M and δqj (t0 ) = δqj (t1 ) = 0 follows according to (1. Theorem: Euler–Lagrange Condition For the functional defined through (1. δq(t)] = dt − δqj (t) .

M (1. ˙2 2 (1. . q(t). Using y = dx the condition reads d dx ∂L ∂y = d dx y 1 + (y )2 = 0. obey the Euler–Lagrange dy condition. 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 . q(t).1: Show that the shortest path between two points on a sphere are great circles. j = 1. y = const.22) ˙ where L(q(t). 2.18) dy s is a functional of y(x) of the type (1.19) From this follows y / 1 + (y )2 = const and.4 The total path length is then given by the integral x1 Lagrangian Mechanics s = x0 dx 1+( dy 2 ) . q2 . t) (1. t) = j=1 1 mj qj − U (q1 .21) are extremals of the functional. One obtains y 1 − y2 y(x) = (x − x2 ) + y2 . circles whose centers lie at the center of the sphere. The shortest path is an extremal of s[y(x)] which must. . The constants a and b are readily identified through the conditons y(x1 ) = y1 and y(x2 ) = y2 . . Velocity–dependent potentials which describe particles moving in electromagnetic fields will be considered below. i. This in turn yields y(x) = ax + b. dx (1. .23) Presently we consider only velocity–independent potentials. . . t) is the so-called Lagrangian M ˙ L(q(t). (1. . t1 S[q(t)] = t0 ˙ dt L(q(t). .. hence.1. 1. qM ) . the so-called action integral. according to the theorems above.5) with L(y(x).e.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. dx ) = 1 + (dy/dx)2 .20) x1 − x2 Exercise 1. q(t). (1.

30.24) ∂qj dt ∂ qj ˙ ∂qj dt which are obviously equivalent to the Newtonian equations of motion. t) 1∂ K(r. t) as follows B E = ×A 1 ∂A . t) obey the Maxwell equations ×E + ·B ×B − ·E 1 ∂ c ∂t E 1 ∂ c ∂t B (1.30) (1.32) (1. t) dt c where dr = v and where F (r. The transformation which leaves the fields invariant is A (r. t) + v × B(r. x2 (t).25) of a particle moving in the field. t). These conditions read in the present case ∂L d ∂L ∂U d − = 0 → − − (mj qj ) = 0 ˙ (1.22.25) = 0 = 0 4π J c = 4πρ = (1. x3 (t)) moving in an electromagnetic field described through the electrical and magnetic field components E(r.23).27) and (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). F (r. t) and B(r. dt The fields E(r. t) describes the charge current density. t) describes the charge density present in the field and J(r. t) .28) (1.31) = − V − Gauge Symmetry of the Electromagnetic Field It is well known that the relationship between fields and potentials (1. respectively.26) (1. t) + V (r. 1.2: Lagrangian 5 For a proof of the Hamiltonian Principle of Least Action we inspect the Euler–Lagrange conditions associated with the action integral defined through (1. t) and a vector potential A(r.28) can be satisfied implicitly if one represents the fields through a scalar potential V (r. t) = A(r. Equations (1. The equations of motion for such a particle are d q ˙ (mr) = F (r. 1. c ∂t (1.27) (1.1.29) where ρ(r.33) .31) allows one to transform the potentials without affecting the fields and without affecting the equations of motion (1. t) is the Lorentz force. t) and B(r. t) c ∂t (1. t) = V (r. t) − K(r. t) = q E(r.

namely. r. t) = 1 q mv 2 − q V (r.35) shows that the Newtonian equations of motion and the Euler–Lagrange conditions are.41) which is identical to the term (1. (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. dt ∂x1 c We notice using (1.g.25).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 .25) follows from the Hamiltonian Principle of Least Action. t) + A(r. .38) ∂L ∂ x1 ˙ − ∂L = 0. in fact.40) (1. ∂x1 (1.40. Comparing then (1.38.35) − ∂A1 ∂x2 [v × B]1 = x2 B3 − x3 B2 = x2 ˙ ˙ ˙ ∂A2 ∂A1 − ∂x1 ∂x2 − x3 ˙ ∂A1 ∂A3 − ∂x3 ∂x1 . if one assumes for a particle the Lagrangian ˙ L(r.34) For this purpose we consider only one component of the equation of motion (1. e.37) is ∂L ∂V q = −q + ∂x1 ∂x1 c The first term in (1. (1. 1.41) with (1.39) 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 (1.6 Lagrangian of Particle Moving in Electromagnetic Field Lagrangian Mechanics We want to show now that the equation of motion (1.36) in the Newtonian equation of motion..36) This expression allows us to show that (1.39) ∂A1 ∂A2 ∂A3 x1 + ˙ x2 + ˙ x3 ˙ ∂x1 ∂x1 ∂x1 .37) The results (1.30). 1. equivalent. (1. B3 = ∂A2 ∂x1 (1. t) · v .

q. (1. t) + A (r. q.42) is. equivalent to the gauge transformation (1.33) of electromagnetic potentials. t) = q K(q. 1. We will consider in the following two symmetries. r. t) + dt c (1.44) Since the condition δq(t0 ) = δq(t1 ) = 0 holds for the variational functions of Lagrangian Mechanics.32.45) .32. t) = L(q. t) + K ·v (1.32. This term is d ∂K ∂K ∂K ∂K ∂K K(r. gauge symmetry and symmetries with respect to spatial transformations. One can conclude then immediately that any extremal of S [q(t)] is also an extremal of S[q(t)]. For this purpose we consider the transformation of the single particle Lagrangian (1. encountered above in connection with the transformations (1. t) = x1 + ˙ x2 + ˙ x3 + ˙ = ( K) · v + . They are the subject of the following theorem.46) (1.43) To prove this theorem we determine the action integral corresponding to the transformed Lagrangian t1 S [q(t)] = t0 ˙ dtL (q. 2 c = A(r. t) − + 2 c ∂t c 1 q = mv 2 − q V (r. 2 c c dt Inserting (1. 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. Note that one adds the total time derivative of a function K(r. i.e. t) L (q. t) c t1 t0 t1 t0 ˙ dtL(q.33) of electyromagnetic potentials. in fact.34) q 1 q d ˙ L (r.3 Symmetry Properties in Lagrangian Mechanics Symmetry properties play an eminent role in Quantum Mechanics since they reflect the properties of the elementary constituents of physical systems. The factor q has been introduced to make this c transformation equivalent to the gauge transformation (1.32..33) ˙ L (r. q.43) into (1. dt ∂x1 ∂x2 ∂x3 ∂t ∂t (1. and since these properties allow one often to simplify mathematical descriptions. t) 1 1 ∂K q mv 2 − q V (r. t) · v . 1. surprisingly simple fashion in Lagrangian Mechanics. The gauge symmetry. r. Eq. a term not affected by the variations allowed. 1. t) the Lagrangian. t) + q K(q.33) of electromagnetic potentials.45) and reordering terms yields using (1.3: Symmetry Properties 7 1. t) c t1 t0 = S[q(t)] + (1. t) .42) This transformation is termed gauge transformation. t) + A(r.1. q.44) implies that the gauge transformation amounts to adding a constant term to the action integral. 1. appear in a different. t) · v + K(r. We want to demonstrate now that the transformation (1. t) = mv 2 − q V (r.

t) by gauge transformed potentials V (r.42) and (1. 1.8 Lagrangian Mechanics Obviously. We consider now invariance properties connected with coordinate transformations. The transformations considered are specified in the following definition. r ∈ R .48) In the following we will denote unit vectors as a. ) .50) A non-trivial example is furnished by the infinitesimal rotation around axis e ˆ r = r + e×r . 0) = r . ˆ ˆ ˆ Examples of Infinitesimal Transformations The beauty of infinitesimal transformations is that they can be stated in a very simple manner. In this case the transformation can be written in matrix form      cos −sin 0 x1 x1  x2  =  sin cos 0   x2  (1.. t). The corresponding infinitesimal transformation is defined for small r (r. t). therefore.42) corresponds to replacing in the Lagrangian potentials V (r. In case of a translation transformation in the direction e nothing new is gained.47) (1. which has an important analogy in Quantum Mechanics.51) We would like to derive this transformation in a somewhat complicated. for such vectors holds a · a = 1. 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. Such invariance properties are very familiar. the transformation (1. Definition: Infinitesimal One-Parameter Coordinate Transformations A one-parameter coordinate transformation is decribed through r = r (r. in the case of central force fields which are invariant with respect to rotations of coordinates around the center.e. ˆ (1.32. i. but nevertheless instructive way considering rotations around the x3 –axis. the equivalence of (1. A (r. A(r. and for the introduction of infinitesimal transformations which will provide a crucial method for the study of symmetry in Quantum Mechanics. for the connection between invariance properties and constants of motion.52) x3 0 0 1 x3 . r. ) = r + R(r) + O( 2 ) . ˆ (1. for example.49) =0 ∈ R (1. we like to ˆ provide the transformation here anyway for later reference r = r + e. The following description of spatial symmetry is important in two respects. R(r) = through ∂r ∂ (1.33). We have proven. where the origin of is chosen such that r (r. The transformations we consider are the most simple kind. t). However.

57) ∂ d where we used dt Qj = M ( ∂qk Qj )qk . a quantity C(r. the property holds for any system. the second and third term ˙ k=1 in (1. (1. q. neglecting terms of order O( M ˙ ˙ L (q.53). ˙ We have generalized in this theorem the definition of infinitesimal coordinate transformation to M –dimensional vectors q. expansion. Invariance implies L = L. however.58) From this follows the statement of the theorem. We consider here only the ‘particle version’ of the theorem.54) which is equivalent to (1. i.k=1 ∂L ∂Qj qk ˙ ∂ qj ∂qk ˙ (1.. We have used here the notation corresponding to single particle motion.e. 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 (1.55) (1. a classical mechanical system is invariant with respect to a coordinate transformation ˙ a constant of motion exists. In order to prove Noether’s theorem we note qj qj ˙ = qj + = qj + ˙ k=1 Qj (q) M (1.57)   M Qj j=1 d dt ∂L ∂ qj ˙ + ∂L d Qj ∂ qj dt ˙ = d  dt M Qj j=1 ∂L  = 0 ∂ qj ˙ (1. ˙ ∂qk ˙ Inserting these infinitesimal changes of qj and qj into the Lagrangian L(q. q.57) must cancel each other or both vanish. Using the fact. i. t) yields after Taylor ˙ 2 ). The property has been shown to hold in a more general context. t) = L(q. 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.51) reads r = r − x2 e1 + x1 e2 ˆ ˆ (1. Theorem: Noether’s Theorem ˙ If L(q..e.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. namely for fields rather than only for particle motion. then M ∂L j=1 Qj ∂ xj is a constant of motion.56) ∂Qj qk .1. . q. Anytime.3: Symmetry Properties 9 In case of small this transformation can be written neglecting terms O( 2 ) using cos = 1 + O( 2 ). r) which is constant along the classical path of the system. sin = + O( 2 )        0 − 0 x1 x1 x1  x2  =  x2  +  0 0   x2  + O( 2 ) . by Noether. t) is invariant with respect to an infinitesimal transformation q = q + Q(q). In the context of Lagrangian Mechanics this implies that such transformation leaves the Lagrangian of the system unchanged.

for which translational invariance holds. 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. We will now investigate the consequence of rotational invariance as described according to the infinitesimal transformation (1.59) We obtain the well known result that in this case the momentum in the direction. In this case we will use the same notation as in (1. 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. i. to be expected that this is the constant of motion. 2. j = 1.e. not infinitesimal transformations.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. j = 1. 2 We first determine the constant of motion in case of invariance with respect to translations as defined in (1.59). It was. In this case we have Qj = ej · e. ˆ ˙ ˆ j=1 (1. It is not necessary to investigate the consequences of global.50). 3) 3 Qj j=1 ∂L = e· ˆ ∂ xj ˙ 3 ej mxj = e · mv . of course.51). A calculation similar to that in (1. Noether’s theorem yields ˆ ˆ the constant of motion (qj = xj . .59) yields the constant of motion ˆ e (ˆ × r) · mv. is conserved. v) = 1 mv 2 − U (r). hence. except using now Qj = ej · (ˆ × r). 3 and. 2.

Chapter 2 Quantum Mechanical Path Integral 2. At present one may think of f (r) as a sum over δ–functions which represent a multi–slit screen.1 The Double Slit Experiment Will be supplied at later date 2. t) where z is the conjugate complex of z. t). 11 . t)|2 Ω (2. The particle is described by a wave function ψ(r.2 Axioms for Quantum Mechanical Description of Single Particle ˙ Let us consider a particle which is described by a Lagrangian L(r.3) where Ω is the space volume in which the particle can exist. 4. (2. t) is normalized d3 r |ψ(r.2) (2. t ∈ [t0 . is unity. placed into the space at some particular time and with a detector behind each slit.4) a condition which enforces that the probability of finding the particle somewhere in Ω at any particular time t in an interval [t0 . t1 ] . The probability that the particle is detected at space–time point r. 3.1) (2. t) ψ : R3 ⊗ R → C. 1. t1 ] in which the particle is known to exist. We provide now a set of formal rules which state how the probability to observe such a particle at some space–time point r. The wave function ψ(r. The probability to detect the particle with a detector of sensitivity f (r) is d3 r f (r) |ψ(r. 2. t)|2 = 1 Ω ∀t. t)ψ(r. t is |ψ(r. r. t)|2 = ψ(r. t is described in Quantum Mechanics.

This symbol will be defined further below.7) t. t|r . Since (2. The generalization of the completeness property to N − 1 intermediate points t > tN −1 > tN −2 > . t. t ) φ(r . > t1 > t0 is φ(r. t ∈ (2.10) We have introduced here a new symbol. namely the point on the classical path at time t . (t > t . t1 ] (2. t ) ψ(r . t|r0 . t ∈ [t0 . t ) t > t . t ) ψ(r . t0 ) Ω This relationship has the following interpretation: Assume that at time t0 a particle is generated by a source at one point r0 in space. The time evolution of ψ(r. t|r0 .4) holds for all times.. the propagator is unitary. the path integral r(tN )=rN d[r(t)] · · · r(t0 )=r0 (2. t|r . t0 ) = r(t0 )=r0 d[r(t)] Φ[r(t)] . tN −2 ) · · · φ(r1 . t)|2 = (2. t0 ) = Ωd 3r N −1 Ωd 3r N −2 ··· Ωd 3r 1 φ(r. t0 ) = δ(r − r0 ). t| rN −1 . t. i. t )φ(r. t|r . i. This must hold for any ψ(r . The definition will actually assume an infinite number of intermediate times and express the path integral through integrals of the type (2. t ) = δ(r − r ) Ω (2. . ψ(r0 .e.9) Employing a continuum of intermediate times t ∈ [t0 . requires then according to (2. The state of a system at time t.e. t |r0 . . The following so-called completeness relationship holds for the propagator (t > t [t0 . t|r0 . t|r . t) = Ω d3 r φ(r. . tN −1 ) φ(rN −1 . t ) φ(r. at any intermediate time the particle needs only be known at one space point. t ) which requires d3 r φ(r. t1 |r0 . t ) ψ(r . t1 ] yields an expression of the form r(tN )=rN φ(r.6) φ(r. t|r . described by ψ(r. 8. tN −1 |rN −2 . t1 ]) d3 r φ(r. This is different from the classical situation where the particle follows a discrete path and. hence. t0 ) .9) for N → ∞. t = )|2 = 1 . (2. t1 ]) Ωd Ωd 3r Ωd 3r Ωd 3r 3 r |ψ(r. t0 ) = φ(r. t) is described by a linear map of the type ψ(r. t ∈ [t0 . t).8) 7.11) which denotes an integral over all paths r(t) with end points r(t0 ) = r0 and r(tN ) = rN . (2. t|r .8) a knowledge of the state at all space points r ∈ Ω at some intermediate time t . t ) Ωd 3 r |ψ(r.5) 6.12 Quantum Mechanical Path Integral 5..

not only for the classical path. i. for microscopic particles like the electron this is by no means the case. t) is the Lagrangian of the classical particle. The situation is very different for microscopic particles.14) has the same dimension as the action integral S[r(t)]. actually often essentially exclusive.12. t) (2. accordingly. namely due to the property of the classical path to be an extremal of the action integral. r. 2.15) The exponent of (2.14) ˙ In (2. a very large number. However. i. Situations which are well described classically will be distinguished through the property that the classical path gives the dominant. 2 2 (2.. The constant given in (2. However. Scl / ≈ 8.2: Axioms 9. such that destructive interference of the contributions of such paths does not occur.0545 · 10−27 erg s . r. Since this number is multiplied by ‘i’. the exponent is a very large imaginary number. The functional Φ[r(t)] in (2. interferences should be much less dramatic than in case of the macroscopic particle. 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.e.13) L(r. 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.13).e. expressions (2.2.5 · 1027 . Its value is extremely small in comparision with typical values for action integrals of macroscopic particles.11) is Φ[r(t)] = exp where S[r(t)] is the classical action integral tN 13 i S[r(t)] (2. The value of S[r(t)] is then Scl = 1 1 m v 2 t = erg s . One would still expect significant descructive interference between contributions of different paths since the value calculated is comparable to 2π. (2. in complete distinction from Classical Mechanics.13) are built on action integrals for all possible paths.13) and = 1. However.12) S[r(t)] = t0 ˙ dt L(r. 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. This implies that small variations of the path near the classical path alter the value of the action integral by very little. 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.. contribution to the path integral (2. .12) is then Scl / ≈ 0.12. it is comparable to action integrals as they arise for microscopic particles under typical circumstances. 2. In case of a proton with mass m = 1. However.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.

. t1 ). Its value is CN = m 2πi N 2 (2. This allows us to write the evolution operator using (2.20) Here.16) In order to define the path integral we assume.14 Quantum Mechanical Path Integral 2. . as in (2. The spacings between the times tj+1 and tj will all be identical. For the sake of simplicity we will consider the case of particles moving in one dimension labelled by the position coordinate x. respectively. 2. etc.19) The main idea is that one can replace the path integral now by a multiple integral over x1 . .20) can be properly taken. (xN −1 . −∞ dxN −1 (xj+1 −xj )2 1 − U (xj ) . . namely tj+1 − tj = (tN − t0 )/N = N . provides an avenue to evaluate path integrals. (2. tj ) to the next (xj+1 .12) φ(xN . ∞[. . (2. tN )} . ˙ 2 (2.9). x. (xN . 2 2m N (2.21) N 2.13) and. > t1 > t0 letting N go to infinity later. at the same time. 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 (2.3 How to Evaluate the Path Integral In this section we will provide an explicit algorithm which defines the path integral (2. t0 ). 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 . . the xj values are not. t) = ˙ 1 mx2 − U (x) .10) and (2.18) The time instances are fixed. t0 ) = limN →∞ CN exp i N N −1 j=0 +∞ +∞ +∞ −∞ dx1 −∞ dx2 . .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 . The particles have associated with them a Lagrangian L(x. (2. however. They can be anywhere in the allowed volume which we will choose to be the interval ] − ∞. a series of times tN > tN −1 > tN −2 > . In passing from one space–time instance (xj . tj+1 ) we assume that kinetic energy and potential energy are constant.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. tN −1 ). (x1 . namely 1 m(xj+1 − xj )2 / 2 and U (xj ). .12. x2 . tN |x0 .

29) This expression corresponds to the action integral in (2. of (2. The resulting expression for S[x(t)|x(t0 ) = x0 . 2. since x(t0 ) = xcl (t0 ) = x0 and x(tN ) = xcl (tN ) = xN . tN ).s.25) is. 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 . To see the benifit of such approach we define a path y(t) as follows x(t) = xcl (t) + y(t) (2. namely.30) d[x(t)] exp 2 tN − t0 x(t0 )=0 a result. it is more suitable to define new integration variables yj .13). (2. xN − x0 xcl (t) = x0 + ( t − t0 ) . The ˙ 1 S[x(t)|x(t ) = x .31) We have denoted explicitly that the action integral for a path connecting the space–time points (x0 . tN dt y = y(tN ) − y(t0 ) = 0 . x(tN ) = 0] .24). using (2. due to (2. tN |x0 . tN ) is to be evaluated. t0 ) and (xN . can be expressed through a path integral with endpoints x(t0 ) = 0. the origin of which coincides with the classical path of the particle. t0 ) = exp S[x(t)] (2. tN |0. it holds y(t0 ) = y(tN ) = 0 .23).27) The third term can be written in the notation introduced tN 1 dt my 2 = S[x(t)|x(t0 ) = 0. 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. Inserting the result into (2.24) Also we use the fact that the velocity of the classical path xcl = (xN −x0 )/(tn −t0 ) is constant. ˙ t0 (2.22) where xcl (t) is the classical path which connects the space–time points (x0 . x(tN ) = xN ] is S[x(t)|x(t0 ) = x0 . tN |x0 . (2.28) i.23) tN − t0 It is essential for the following to note that.e. x(tN ) = 0.10. t0 ) and (xN . One obtains S[x(t)|x(t0 ) = x0 . 2 2 tN − t0 (2.4: Propagator for a Free Particle 15 Rather then using the integration variables xj . x(tN ) = 0] .s. t0 ) (2.2.h. = (2.h.12) yields x(tN )=0 im (xN − x0 )2 i φ(xN . x(tN ) = xN ] 1 (xN − x0 )2 m + 0 + 2 tN − t 0 + S[x(t)|x(t0 ) = 0.. tN dt t0 1 1 (xN − x0 )2 m x2 = m ˙ cl .25) + The condition (2. which can also be written φ(xN . ˙ ˙ ˙ 2 + y2) = ˙ (2. t0 ) = exp 1 im (xN − x0 )2 2 tN − t0 φ(0. . ˙ 2 t0 (2.26) The first term on the r.

(2.. .36) which holds for a d-dimensional. 0 0 .34) dy1 · · · −∞ −∞ dyN −1 exp i  N −1 j. real.  . .34) of (ajk ) defining a new matrix (Ajk ) through ajk = Using det(ajk ) = m 2 N N −1 m 2 N Ajk .31) for a free particle one needs to evaluate the following path integral φ(0. 0 0   .38) . (2.33) where ajk are the elements of the following symmetric (N  2 −1 0  −1 2 −1  2 m  0 −1  ajk =  . . . . In the appendix we prove  +∞ +∞ d j.35) must be determined. t0 ) = limN →∞ × +∞ +∞ −∞ dy1 · · · −∞ dyN −1 exp i N m 2πi N 2 N × (2.k=1 dy1 · · · −∞ −∞ dyN −1 exp i yj bjk yk  =  (iπ)d det(bjk ) 1 2 .. . ... .37) det(Ajk ) .16 Evaluation of the necessary path integral Quantum Mechanical Path Integral To determine the propagator (2.. . 2 N  . 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.k=1 yj ajk yk   (2. 0 0   . ...32) (yj+1 − yj )2 N −1 1 2 j=0 2 m N The exponent E can be written.. .  0 0 0 0 0 0 The following integral +∞ +∞ (2. symmetric matrix (bjk ) and det(bjk ) = 0.  .32) we split off the factor 2 mN in the definition (2. noting y0 = yN = 0.. 2 −1  −1 2 (2.k=1 yj ajk yk − 1) × (N − 1) matrix  .  . In order to complete the evaluation of (2. (2.

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 .44) = tN − t0 . D1 = 2 and obtain from (2. 0 0  −1 2 −1 . tN |0. Inserting this into (2.4: Propagator for a Free Particle a property which follows from det(cB) = cn detB for any n × n matrix B. . variable.. ..46) . one can readily verify Dn = n + 1 . . tN |0. 2 −1  0 0 0 −1 2 For this purpose we expand (2. t0 ) = m 2πi (tN − t0 ) 1 2 . . x = xN φ(x.. (2. . . To solve this three term recursion relationship one needs two starting values.41) the same result for D2 . . (2. . One can readily verify that this procedure leads to the following recursion equation for the determinants Dn = 2 Dn−1 − Dn−2 .31) and obtain. 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. D3 . (2. We seek then to evaluate the determinant of the n × n matrix   2 −1 0 .39) yields φ(0. . . tN |0.42) m 2πi NN 1 2 (2.39) In order to determine det(Ajk ) we consider the dimension n of (Ajk ). n = 1.45) Expressions for Free Particle Propagator We have now collected all pieces for the final expression of the propagator (2. we obtain φ(0. 0 0    Dn =  .2. . . ..41) D2 = 2 −1 −1 2 = 3 (2.18) we obtain φ(0.40) in terms of subdeterminants along the last column. .40) . t|x0 .  . ... t0 ) = limN →∞ and with NN (2. det(Ajk ) (2. Using D1 = |(2)| = 2 . 0 0     0 −1 2 . 2. .. (2. which follows from (2. defining t = tN . let say n. .    0 0 0 .. . . ..   . t0 ) = m 2πi (t − t0 ) 1 2 exp im (x − x0 )2 2 t − t0 . .

49) is Gaussian of width δ. the wave function of the particle at later times. (2.53) .46).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 . centered around x0 = 0. We need to evaluate for this purpose the integral ψ(x. thereby. 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 . One obtains then for the propagator (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. x) = im 2 t x2 1 + i o t mδ 2 − 2xo x − po t m + f (x) (2.5) to (2.46) can be generalized to three dimensions in a rather obvious way.247). the associated probability distribution |ψ(x0 . x) + E(x) in (2. (2. t0 ) = m 2πi (t − t0 ) 3 2 exp im (r − r0 )2 2 t − t0 . We will obtain. x) + E(x) (2. Below we will apply this to a particle at rest and a particle forming a so-called wave packet.48) Obviously. The result (2.10) φ(r. t|r0 .18 Quantum Mechanical Path Integral This propagator. and describes a single particle since 1 πδ 2 1 2 +∞ dx0 exp −∞ − x2 0 δ2 = 1.52) and applying (2.50) One refers to such states as wave packets. We devide the contributions to the exponent Eo (xo .5) allows us to predict the time evolution of any state function ψ(x.51) as follows Eo (xo . t0 ) = 1 πδ 2 1 4 exp − x2 po 0 +i x 2 2δ (2. We want to apply axiom (2. t0 )| = 2 1 πδ 2 1 2 exp − x2 0 δ2 (2. according to (2.48) as the initial state using the propagator (2. t) of a free particle.

(2. making certain.60) is then represented by real ρ values.59) An integration path in the complex x0 –plane along the direction i 1−i t mδ 2 (2. 2 (2.61) . We proceed as follows.53) E(x) =  x− po mt 19 (2.4: Propagator for a Free Particle im x2 − f (x) .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  .55) im  xo 2 t  1+i t − mδ 2 x− 1+ po m t t i mδ2  .x) (2. t) = 1 πδ 2 1 4 2 t 1 + i mδ2  . 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 . x) = One can write then (2.52). however.56) m 2πi t 1 2 +∞ eE(x) −∞ dx0 eEo (xo . The beginning and the end of such path are the points z1 = −∞ × i 1−i t mδ 2 . according to (2.51) ψ(x. that the new path does not lead to additional contributions to the integral.2. 2 t One chooses then f (x) to complete.58) The integrand is an analytical function everywhere in the complex plane and we can alter the integration path. z2 = +∞ × i 1−i t mδ 2 (2.54) f (x) =  This yields Eo (xo . the square in (2.57) and needs to determine the integral +∞ I = −∞ +∞ dx0 eEo (xo . (2. (2.

58) has the end points z1 = −∞ . Since the exponent in (2. t m(1 + i mδ2 ) (2.58) has a leading contribution in x0 of −x2 /δ 2 the integrand of (2.54) using (2. (2. t) = t 1 − i mδ2 t 1 + i mδ2 1 4 1 πδ 2 (1 + 2 t2 m2 δ 4 1 4 ) exp [ E(x) ] .69) t)2 t 2δ 2 (1 + i mδ2 ) + i po x − i p2 o t 2m (2. We can conclude then that (2. Hence.58) along the path z1 → z1 and along the path z2 → z2 gives only vanishing contributions one can replace (2.66) yields ψ(x.63) holds and. (2. one can show that z1 lies at −∞ − i × ∞ and z2 at +∞ + i × ∞.63) which can be readily evaluated. Quantum Mechanical Path Integral (2.70) .65) t 1 − i mδ2 t 1 + i mδ2 .58) vanishes 0 for Re x0 → ±∞. using a ab = a − .55). t) = Seperating the phase factor 1 t 1 + i mδ2 1 4 1 2 exp [ E(x) ] .20 whereas the original path in (2.64) ψ(x. (2.62) If one can show that an integration of (2.68) and. 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. z2 = +∞ . I = Equation (2. 1+b 1+b finally E(x) = − (x − po m (2.67) We need to determine finally (2.58) by t mδ 2 +∞ I = i 1−i dρ exp − −∞ m 2 t 1+ t mδ 2 2 ρ2 (2. accordingly.57) reads then 1 πδ 2 1 4 2πi t . In fact. the paths between z1 → z1 and z2 → z2 have a real part of x0 of ±∞.

72) 2 t2 δ 1+ 2 4 (2.2. (2.76) From this we conclude that an initial wave function ψ(xo . (2.72).f. 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 . Another interesting observation is that the wave function (2.75) ψ(x.e. However. (2.72).77) . In this case holds ψ(x. the spatial dependence of the initial state (2. (2. This ‘spreading’ of the wave function is a genuine quantum phenomenon. the case of arbitrary to is recovered iby replacing t → to in (2.49). The corresponding probability distribution is |ψ(x.71) conserves the phase factor exp[i(po / )x] of the original wave function (2.4: Propagator for a Free Particle which inserted in (2. The width of the distribution (2.71).71. This yields.72) moves in the direction of the positive x-axis with velocity vo = po /m which identifies po as the momentum of the particle.48) for δ → ∞. (2. t) = exp i po i p2 o x − t m 2m . The center of the distribution (2. 2.. We had assumed above [c.74) which corresponds to (2. coinciding at t = 0 with the width of the initial distribution (2.48) and for the final state (2.48)] to = 0. t0 ) = exp i po xo m . (2.72) Comparision of Moving Wave Packet with Classical Motion It is revealing to compare the probability distributions (2. t0 ) = exp i po i p2 o xo − to m 2m . t) = exp i i p2 po o x − (t − to ) m 2m .71) × exp − (x − po m t)2 ) 2δ 2 (1 + 2 t2 m2 δ 4 (1 − i .73) m δ increases with time.72) for the initial state (2.75) i. respectively.49).48) and that the respective phase factor is related with the velocity of the classical particle and of the center of the distribution (2. instead of (2.74) remains invariant in time.67) provides the complete expression of the wave function at all times t ψ(x. t)| = 2 1 πδ 2 (1 + 2 t2 m2 δ 4 1 2 ) exp − (x − δ 2 (1 + po m t)2 ) 2 t2 m2 δ 4 . The conservation of this factor is particularly striking for the (unnormalized) initial wave function ψ(x0 . (2. 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.

y(t)] which vanishes according to the Hamiltonian principle as discussed in Sect. t) = L(xcl .e. t) + L (y. 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 .81) vanishes2 .80) into (2.. y(tN ) = 0 .78) i. Obviously.80) For this purpose we need to determine the action integral S[x(t)] = t0 dt L(x.82) one obtains L(xcl + y.85) We want to show now that the contribution of δL to the action integral (2.13) φ(xN .83) (2. t) + δL ˙ ˙ ˙ ˙ where L(xcl .82) which describes the deviation from the classical path xcl (t) with end points xcl (t0 ) = x0 and xcl (tN ) = xN . In order to simplify this task we define again a new path y(t) x(t) = xcl (t) + y(t) (2.79) for a quadratic Lagrangian L(x. ˙ ˙ = (2.86) dt The reader may want to verify that the contribution of δL to the action integral is actually equal to the differential δS[xcl .12. x.84) . xcl . 2. xcl . (2. y(t). the spatial as well as the temporal dependence of the wave function remains invariant in this case.10. the end points of y(t) are y(t0 ) = 0 Inserting (2. Such states play a cardinal role in quantum mechanics. One refers to the respective states as stationary states. 1. t) = exp i Quantum Mechanical Path Integral po i p2 o x − t m 2m . For this purpose we use d xcl y = ˙ ˙ (xcl y) − xcl y ˙ ¨ (2.22 becomes at t > to ψ(x. 2 .81) for an arbitrary path x(t) with end points x(t0 ) = x0 and x(tN ) = xN . t) ˙ L (y. t) ˙ (2. y(t). t0 ) = x(t0 )=x0 d[x(t)] exp i S[x(t)] (2. xcl + y(t).5 Propagator for a Quadratic Lagrangian x(tN )=xN We will now determine the propagator (2. 2. x. tN |x0 . (2. 2. ˙ 2 2 tN (2. t) = ˙ 1 1 mx2 − c(t)x2 − e(t)x .

x t0 (2. .   . .  0 0 0 .88) and. tN |0. tN |x0 . .. . 2 . .h. . For the evaluation of φ(0.88) vanishes. .s. . .  .92) where ajk are the elements of the following (N − 1) × (N − 1) matrix   2 −1 0 . 0 0   N  0 −  .24) to the Lagrangian (2.5: 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) .. 0 0  0 c2 0 . cN −2 0  0 0 0 0 cN −1 (2. . 2 −1  0 0 0 −1 2   c1 0 0 .2. . t0 ) (2. . .. . t0 ) = limN →∞ × +∞ +∞ −∞ dy1 · · · −∞ dyN −1 exp N i N N −1 j=0 m 2πi N 2 N × − 1 2 cj 2 yj (2. tN |0. y. . . vanishes. .  2 N  . 0 0  m   ajk =  . One can express the exponent E in (2. .89) where ˜ φ(0. . (2. 0 0     0 −1 2 . One can then express the propagator (2. . .s. .. tN |0.    0 0 0 . .h. . . t0 ) = exp S[xcl (t)] φ(0. Applying the Euler–Lagrange conditions (1. of (2. .87) According to (2. 0 0  −1 2 −1 . . 0 0     0 c3 .32). .  .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. t0 ) = y(tN )=0 d[y(t)] exp y(t0 )=0 i tN dt L (y.   .k=1 yj ajk yk (2.79) i ˜ φ(xN ..91) through the quadratic N −1 E = i j. t0 ) we will adopt a strategy which is similar to that used for the evaluation of (2. The discretization scheme adopted above yields in the present case ˜ φ(0. hence. .. .. t) ˙ t0 . tN |0.93) .80) yields for the classical path m¨cl + c(t) xcl + e(t) = 0 x (2. .. . tj = t0 + form j. also the second contribution on the r. .31) for the free particle propagator. . .91) (yj+1 − yj )2 1 2 2m N where cj = c(tj ). .83) the first term on the r.

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

Noting that the propagator (2. t − to ) = φ(x.e. (2. . 2 (2. results in 1 ˜ exp[−iω(n + 1 ) t] φn (x) 2 n! +∞ = −∞ dxo Φ(x. t) = n=0 1 ˜ exp[ − iω(n + 1 ) t ] φn (x) 2 n! (2. xo .183) where the expansion coefficients are functions of x. 2.182) Comparision with (2. . xo . We note that the factor exp(2 mω/ xe−iωt − e−2iωt ) in (2.185) Replacing t → t + to yields ∞ n=0 1 ˜ exp[−iω(n + 1 ) (t + to )] φn (x) 2 n! ∞ +∞ = m=0 −∞ dxo Φ(x. i.172) is a function of t − to and defining accordingly Φ(x. 1 exp[iω(n + 2 ) to ] · · · .h. one obtains for the r. t) = exp[−iω(n + 1 ) t] φn (x) as invariants under the 2 action of the propagator φ(x..187) or 1 ˜ exp[−iω(n + 2 ) t] φn (x) +∞ = −∞ ˜ dxo φ(x. 2.172) in the form ∞ n=0 (2.172) I = exp 2 mω x e−iωt − e−2iωt − Quantum Mechanical Path Integral mω 2 1 x − iωt 2 2 .188) identifies the functions ψn (x.36 Altogether.172). xo . t).s.171) can be expanded in terms of e−inωt . t|xo . but not of t. . t − to ) 1 ˜ exp[−iω(m + 1 ) to ] φm (xo ) 2 m! (2. t − to ) 1 ˜ φn (xo ) n! (2. which also exhibits such invariance. of (2. one can expand (2..171) ∞ W (x. . t) +∞ −∞ dto 1 ˜ exp[−iω(m + 1 ) to ] φm (xo ) 2 m! (2.172). to ) we express (2. In contrast to W (x. n = 1.171. xo .184) 1 ˜ exp[−iω(n + 1 ) t] φn (x) 2 n! ∞ +∞ = m=0 −∞ dxo Φ(x. to ) exp[−iω(n + 1 ) to ] φn (xo ) .186) Fourier transform.171) concludes the proof of (2. t|xo . to ). Accoordingly. We want to inspect the consequences of the invariance property (2.147) in (2. t|xo .188) ˜ ˜ Equation (2.

197) The expansion coefficients Hn (y) in (2. . through expansion (2. t) provide all stationary states of the quantum mechanical harmonic oscillator. 2. t) are associated with a time-independent probablity density |ψn (x. We will also argue that the functions ψn (x. The Hermite Polynomials ˜ The function (2. z) z 2 e−y or w(y. z) = Hn (y) (2.198) ∂z n z=0 . through the expansion coefficients φn (x) in (2. Actually.195) zn Hn (y) n! ˜ φn (y) .183).171) W (x.193) (2. Expression (2. in a ‘nutshell’ all information on the Hermite polynomials. 2 Here Nn are constants which normalize ψn (x. t)|2 = Nn +∞ −∞ n = 0. t) such that +∞ −∞ 2 dx |ψ(x.183).183) reads then w(y. . according to (2. t) of the quantum mechanical harmonic oscillator ˜ ψn (x. To obtain ˜n (x) we simplify the expansion (2. characterizes the wave functions φn (x).191) (2. Expansion (2.196) where Hn (y) = ey 2 /2 (2.197) are called Hermite polynomials which are polynomials of degree n which will be evaluated below. For this purpose we introduce first closed expressions for φ the new variables y z and write (2. ˜ |φ stationary wave functions ψn (x.194) plays a central role for the Hermite polynomials since it contains. 1. In the following we will characterize the functions φn (x) and determine the normalization constants Nn .194). t) = z 2 e−y where w(y.190) ˜ is obeyed. t)|2 = ˜n (x)|2 . t) = exp[−iω(n + 1 ) t] Nn φn (x) .189) ˜ dx φ2 (x) = 1 n (2. . z) = n=0 1 2 /2 1 2 /2 = = mω e−iωt x (2. This follows from ∂n w(y. z) (2.192) w(y. (2.171). (2. z) = exp(2yz − z 2 ) .5: Propagator for a Quadratic Lagrangian 37 ˜ ˜ the functions ψn (x.2. we have identified then.183).194) = z2 n=0 ∞ 1 ∞ zn ˜ φn (y) n! (2.

Reading. 1989).S. i. 2. D. closed expressions for Hn (y). (2. ∂y n (2.201) ν−µ     µ µ ν−µ Figure 2.200) One can deduce from this expression the polynomial character of Hn (y). Boston.38 Quantum Mechanical Path Integral which is a direct consequence of (2.198) for the Hermite polynomials can be expressed in a more convenient form employing definition (2. The identity (2..Wilf (Academic Press.196) ∂n w(y.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).196) to ˜ derive.199) Comparision with (2. Inc. m lattice (right diagram).Graham. . We will employ (2.194. and O. For this purpose we expand the generating function (2. The diagrams illustrate that a summation over all points in a ν. normalization factors for φ(y).e. that Hn (y) is a polynomial of degree n. One calls w(y. The diagrams also identify the areas over which the summation is to be carried out.1: Schematic representation of change of summation variables ν and µ to n = ν + µ and m = ν −µ. .E. z) the generating function for the Hermite polynomials. Massachusetts. among other properties. We want to derive now explicit expressions for the Hermite polynomials. and recursion equations for the efficient evaluation of Hn (y). 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.Patashnik (AddisonWesley. ν+µ   (2.L.. As will become evident in the present case generating functions provide an extremely elegant access to the special functions of Mathematical Physics3 . µ lattice (left diagram) corresponds to a summation over only every other point in an n.196).Knuth. ν+µ ν   H3 (y) = 8y 3 − 12y. H1 (y) = 2y.200) yields for the first Hermite polynomials H0 (y) = 1. We start from 2 e2yz − z ∞ ν = ν=0 µ=0 ∞ ν 1 ν! 1 ν ν µ ν µ z 2µ (−1)µ (2y)ν−µ z ν−µ (−1)µ (2y)ν−µ z ν+µ = ν=0 µ=0 3 (2. 1990) is a useful introduction to this tool as is a chapter in the eminently useful Concrete Mathematics by R.202) generatingfunctionology by H. . 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 . .

With this restriction in mind one can express (2. From (2. . .210) from which one can conclude the property (2. 0 ≤ k ≤ [n/2] where [x] denotes the largest integer p. (2. µ expressend in terms of n.206) From this expansion we can identify Hn (y) [n/2] Hn (y) = k=0 (−1)k n! (2y)n−2k . Hence.207) This expression yields for the first four Hermite polynomials H0 (y) = 1. it holds Hn (−y) = (−1)n Hn (y) .209). H1 (y) = 2y. 0 ≤ m ≤ n .204) Since ν. k! (n − 2k)! = Hn (y) (2.208) which agrees with the expressions in (2. m are ν = n+m . for odd n. (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 . is a polynomial of degree n. p ≤ x. H2 (y) = 4y 2 − 2. hence.205) Since (n − m)/2 is an integer we can introduce now the summation variable k = (n − m)/2 . (2. m = ν − µ 0 ≤ n < ∞. 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). The lattices representing the summation terms are shown in Fig. In case of even n . 2.e.201). −z) and.203) The old summation variables ν. i.194) holds w(−y.2. z) = w(y.197) ∞ n=0 zn Hn (−y) = n! ∞ n=0 (−z)n Hn (y) = n! ∞ n=0 zn (−1)n Hn (y) n! (2. in fact. 2 µ = n−m . 39 (2.207) contains only even powers. it contains only odd powers.1. otherwise. m must be restricted such that either both n and m are even or both n and m are odd. the sum in (2. H3 (y) = 8y 3 − 12y. 2 (2. according to 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.209) This property follows also from the generating function. k! (n − 2k)! (2.

Another relationship is d Hn (y) = 2n Hn−1 (y). . Hn+1 (y) − 2y Hn (y) + 2n Hn−1 (y) = 0 . (2. .209) since odd functions have a node at the origin. (2. z) − (2y − 2z) w(y. 2.188.194).194).212) n! This implies that stationary states of the harmonic oscillator φ2n+1 (x). Substituting expansion (2. dw/dy − 2zw = 0.208). 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) . For this purpose one considers w(0. . indeed satisfies the latter relationship. have a node at y = 0. (2. . a factor z reduces the order of Hn by one and introduces also a factor n. for example. z) = 0 (2. z) ∂ w(y. (2. Starting point of the derivation is the property of w(y. . . H2n (0) = (−1)n Recursion Relationships A useful set of properties for special functions are the so-called recursion relationships.215) ∂z which can be readily verified using (2. For Hermite polynomials holds. Hn+1 (y) − 2y Hn (y) + 2n Hn−1 (y) = 0. as given in (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 . . 2. 2. n! (2.213) which allow one to evaluate Hn (y) from H0 (y) and H1 (y) given by (2. a property which is consistent with (2.233) below.40 Quantum Mechanical Path Integral The generating function allows one to determine the values of Hn (y) at y = 0. as defined through (2. z) should be equivalent to the relationship (??).214) We want to derive (??) using the generating function. z). . dy n = 1. The reader may verify that w(y.217) gives H1 (y) − 2y H0 (y) = 0. . One can then readily state which differential equation of w(y.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 (2.211) m! n! m=0 n=0 Comparing terms on both sides of the equation yields (2n)! . . namely. n = 1.218) 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. . 1. 2. n = 1. (2. 2. . n = 0. .197) above and given by (2. H2n+1 (0) = 0 .196) into the differential equation (2.

s.224) +∞ −∞ dt tn e2iy t − t . finally.196) the orthogonality properties of the Hermite polynomials. (2.219) which can be written I(y) = e−y 2 +∞ −∞ dte−(t−iy) = e−y √ 2 2 +∞ −∞ dze−z . dy n (2. e−y 2 π e−y 2 (2. acording to the definition (2. . of the Rodrigues formula (2. .s. in the integral representation of the Hermite polynomials 2n (−i)n ey √ Hn (y) = π Orthonormality Properties 2 (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! (2.196) yields ∞ n. 2.222) Employing this expression now on the r.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 .194. 2 (2.200) yields Hn (y) = The identity (−1)n y2 √ e π +∞ dt −∞ dn 2iyt − t2 e . 2 (2.223) dn 2iy t − t2 2 e = (2 i t)n e2iy t − t n dy results.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 (2.247) for a = i one obtains I(y) = and. 1.220) Using (2.221) 1 = √ π +∞ −∞ dt e2iyt − t . z) w(y.h.225) We want to derive from the generating function (2. 2. 2 (2. .226a).194.h. z ) e−y 2 +∞ = e2 z z = √ ∞ dy e−(y−z−z ) = .226) π −∞ 2n z n z n n=0 n! Expressing the l.2. For this purpose we consider the integral +∞ −∞ dy w(y. 2.228) . 2 √ π e2 z z (2. 2 n = 0. (2.

the value of the wave function at y = 0 is positive for n = 0. 3. According to (2. negative for n = 2 and vanishes for n = 1. 4 in Fig. 3. 2.42 Quantum Mechanical Path Integral E 9/2 h ω ¥ ¥ ¥ ¥ ¥ 7/2 h ω 5/2 h ω ¤ 3/2 h ω £ 1/2 h ω ¦   Figure 2.231) √ π (2.233) .197) holds 2 ˜ φn (y) = e−y /2 Hn (y) . 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. The normalized states are [c. in agreement with our above discussions. (2.189. 1. Furthermore. One can also recognize that n is equal to the number of nodes of the wave function. (2. that the wave functions are even for n = 0. 2. (2. ¡ ¢ ¡ -4   -2 0 2 4 y (2. 4. 3.229) Hn (y) . 2.232) Hn (y) .233) are presented for n = 0.f.190)] φn (y) = Nn e−y 2 /2 and for the normalization constants Nn follows from (2.230) (2.2.228) allow us to construct normalized stationary states of the harmonic oscillator. 3. 4 and odd for n = 1.2: Stationary states φn (y) of the harmonic oscillator for n = 0. in harmony with (??). 4. One can recognize. 1. 2. 2.

231) of the wave functions differs from that postulated in (2.e. Ω (2. (2.179] holds Ω = [−1. In case of the associated Laguerre (α) polynomials. 1]. 2. 5. ??? and eq.e.239) . and In = 2n n! π. w(x) ≡ 1.5: Propagator for a Quadratic Lagrangian 43 The normalization condition (2. i. i..228).10 and eq. w(x) = √ exp(−x2 ).189) by the Jacobian dx/dy. of functions which obey dx f 2 (x) w(x) = < ∞ .150 . +∞[. by dx = dy mω 1 4 . 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. 5 below [c. and I = 2/(2 + 1). In = Γ(n+α+1)/n! where Γ(z) is the so-called Gamma function. 5.236) where w(x) is a so-called weight function with the property w(x) ≥ 0. 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. In case of the Legendre polynomials. (5.228) shows that the orthonogality condition of the Hermite polynomials is in complience with (2. (2. g ∈ F . 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.. f.237) and where In denotes some constants. (10.238) where the space is endowed with the scalar product (f |g) = Ω dx f (x) g(x) w(x) = < ∞ . denoted by P (x) and introduced in Sect.f. w(x) = xα e−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.237) for Ω = R. w(x) = 0 only at a discrete set of points xk ∈ Ω (2. 5.459] hydrogen atom. ???] and relativistic [see Sect.233) and (2.156. denoted by Ln (x) and encountered in case of the stationary states of the nonrelativistic [see Sect.151. using (2. holds Ω = [0.236 . 10. Comparision with (2. The Hermite polynomials are so-called orthonogal polynomials since they obey the conditions (2.189) φn (x) = √ 1 2n n! mω π 1 4 e− mωx2 2 Hn ( mω x) . (2. The orthogonal polynomials pn mentioned above have the important property that they form a complete basis in the space F of normalizable functions.234) The explicit form of the stationary states of the harmonic oscillator in terms of the position variable x is then.

(2.242) and the existence of a normalizable state ψ(y. Therefore. t) is ∞ ψ(y.197) it must hold cn (t) = dn exp[−iω(n + 1 )t] and. a (2.188. t) must be identical to one of the stationary states (2.k yj ajk yk = (iπ)n . det(a) (2. t) = n=0 cn (t) e−y 2 /2 Hn (y) .247) .e. ψ(y.240) where cn = 1 In dx w(x) f (x) pn (x) .. the states (2. n 2 (2. hence. i.233) exhaust all stationary states of the harmonic oscillator.e.241) The latter identity follows from (2.m=0 d∗ dm exp[iω(m − n)t] e−y Hn (y)Hm (y) .44 As a result holds for any f ∈ F f (x) = n Quantum Mechanical Path Integral cn pn (x) (2. In case of n = 1 this reads +∞ −∞ dx ei a x = 2 iπ . Appendix: Exponential Integral We want to prove +∞ +∞ I = dy1 . symmetric a. −∞ dyn ei −∞ n j. t) = n=0 1 2 dn exp[−iω(n + )t] e−y /2 Hn (y) .245) must be time-independent. Since the Hermite polynomials form a complete basis for such states we can expand ∞ ψ(y. t)|2 . Ω (2.242) Let us assume now the case of a function space governed by the norm (2. a state for which (2. t) which is stationary under the action of the harmonic oscillator propagator (2. the 2 stationary state ψ(y. pn (x) → w(x) pn (x) the scalar product (2.246) for det(a) = 0 and real.e.233). ∞ n.236). The only possibility for this to be true is dn = 0.243) To be consistent with(2.e.239) becomes the conventional scalar product of quantum mechanics f |g = Ω dx f (x) g(x) . i.172) holds. . 2 (2. except for a single n = no . If one replaces for all f ∈ F: f (x) → w(x) f (x) and.147). . aT = a. i.. 2. (2. in particular.244) For the state to be stationary |ψ(x.. i.

.2.248.k yj ajk yk n = j. ˜ (2. according to (2. (2. (2.k yj ajk yk ˜˜ ˜ (2. i.246) exploits that for any real. symmetric matrix exists a similarity transformation such that   a11 0 .   .k n n yj = ˜ k (S −1 )jk yk . 0 ˜  0 a22 . .k n m = = j. .7: Appendix / Exponential Integral 45 which holds for a ∈ C as long as a = 0. det(S) a = (det(S))−1 det(a) det(S) = det(a) . ˜ (2. The proof of (2. .253) as well as det(˜) = det(S−1 aS) = det(S−1 ) det(a) .  . . . . 2.246) reads then in terms of yj ˜ n n j.249) n ajk = ˜ l.249) The akk are the eigenvalues of a and are real. .m (S T )jl alm Smk . 0 0 .. 0  ˜   ˜ S−1 a S = a =  . . yk = k Skj yj .251) where. .e. (2. One can conclude det(a) = j=1 n (2. This property allows one to simplify the bilinear ˜ form n yj ajk yk by introducing new integration variables j. ST S = 1 1 or S = S−1 . .250) The bilinear form in (2. .248) .k n m n y Sj ajk Skm ym ˜ ˜ y (S T ) j ajk Skm ym ˜ ˜ j. . ann ˜ where S can be chosen as an orthonormal transformation.255) . .254) ajj ..252) ˜ For the determinant of a holds det(˜) = a n ajj ˜ j=1 (2.

. ∂(˜1 ..46 We have assumed det(a) = 0. y −∞ d˜n det y −∞ ∂(y1 . holds n Quantum Mechanical Path Integral ajj = 0 ˜ j=1 (2. . 2. yn ) ) = det(S) .249) follows 1 = det ST S such that one can conclude det S = ±1 One can right then (2.250) allows one to express (2. . . y −∞ +∞ d˜n ei y −∞ n k akk yk ˜ ˜2 +∞ i˜11 y1 a ˜2 n i˜nn yn a ˜2 +∞ a ˜ d˜k ei˜kk yk y 2 = d˜1 e y −∞ .257) Substitution of the integration variables (2. ajj = 0 . d˜n e y −∞ = (2. .266) . ∂ ys ˜ (2. yn ) ∂(˜1 . .265) k=1 −∞ which leaves us to determine integrals of the type +∞ dx eicx −∞ 2 (2. . ˜ for j = 1.258) where we introduced the Jacobian matrix J = with elements Jjs = According to (2.264) I = d˜1 .259) (2. .261) ∂(y1 . . . .262) (2. hence.258) +∞ +∞ ∂(y1 . . . . . . . . . yn ) y ˜ = ( det S )2 (2. n (2. . i. . .260) (2.263) (2. det( From (2. (2. .. . .. .256) such that none of the eigenvalues of a vanishes. .250) +∞ +∞ I = d˜1 . . . . yn ) y ˜ ∂yj . Accordingly. yn ) ∂(˜1 . . . yn ) y ˜ ei n k akk yk ˜ ˜2 .250) holds J = S and.e.

e. for x.270) dx e−cx 2 2 − 2p √ γ4 : z = ix − p J4 = −p i dx eic(ix−p) .266) to the well-known Gaussian integral +∞ dx e−cx −∞ 2 = π .7: Appendix / Exponential Integral 47 where. i. We consider first the case c > 0 and discuss the case c < 0 further below.271) .257) holds c = 0.267) by considering the contour integral J = γ dz eicz = 0 2 (2. The contour integral (2. 2 since eicz is a holomorphic function.268) vanishes. One can relate integral (2. The contour intergral (2. p ∈ IR. 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 . the integrand does not exhibit any singularities anywhere in C..3.268) along the path γ = γ1 + γ2 + γ3 + γ4 displayed in Figure 2.269) The contributions Jk can be expressed through integrals along a real coordinate axis by realizing that the paths γk can be parametrized by real coordinates x p γ1 : z = x γ2 : z = ix + p γ3 : z = √ ix J1 = −p p dx eicx 2 J2 = J3 = = i dx eic(ix+p) 0√ − 2p √ √ 2p 2 i dx eic( 2p √ ix)2 √ − i 0 √ (2.c > 0 . (2. c (2.268) can be written as the sum of the following path integrals J = J1 + J2 + J3 + J4 . Jk = γk dz eicz 2 (2.2. according to (2.

3: Contour path γ in the complex plain. c   ¡ -p γ1 p Re(z) 2 | |e−2cxp | .272) = 0 . Hence.274) (2.48 Quantum Mechanical Path Integral Im(z) ip γ2 ¡ γ4 ¡ γ3 ¡ -i p Figure 2. (2. We will now show that the two integrals J2 and J4 vanish for p → +∞.268) for p = +∞. One can state accordingly ∞ J = dx e −∞ icx2 − √ ∞ i dx e−cx −∞ 2 Using 2. p p→+∞ lim |J2 or 4 | ≤ = p→+∞ 0 lim dx |e−2cxp | 1 − e−2cp 2cp = 0.275) .267) one has shown then ∞ dx eicx −∞ 2 = iπ .273) p→+∞ lim J2 and J4 do not contribute then to integral (2. (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.

279) .278) Noting (2. 2.265). n ˜ j=1 ajj (2.268). This leads to ∞ icx2 J = dx e −∞ + √ −∞ −i ∞ dx ecx = 0 2 (2.275. but with a path γ that is reflected at the real axis. (iπ)n det(a) (2. c (2. if one chooses the same contour integral as (2.2.7: Appendix / Exponential Integral 49 One can derive the same result for c < 0.277) to (2.255) this result can be expressed in terms of the matrix a I = which concludes our proof. It holds n I = k=1 iπ = akk ˜ (iπ)n .277) We apply the above results (2.276) and (c < 0) ∞ dx eicx −∞ 2 = −iπ = −|c| iπ .

50 Quantum Mechanical Path Integral .

t + ) = Ω m 2πi 3 2 exp i m (r − r0 )2 − 2 U (r. The propagator in (3. We will denote the components of s by (x1 .h.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) . For many situations. t) . namely the Schr¨dinger equation.1) can be expressed through the discretization scheme (2.4) − U (r. t) = From this follows ψ(r. t + ) = Ω d3 r0 φ(r.20.s. x2 . This yields ψ(r. x2 .h. t) satisfies a partial differential equation. t) ψ(r + s. Notable exceptions are non-stationary systems involving time-dependent linear and quadratic Lagrangians. Generalizing (2. x3 )T .20) to evaluate the propagator. (3. and the r.20) to R one obtains then for small φ(r. 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. . t) ψ(r0 . the Schr¨dinger o o equation provides the simpler avenue towards describing quantum systems than the path ingral formulation of Section 2. x3 51 . (3.3) In order to carry out the integration we set r0 = r + s and use s as the new integration variable. 2.s.5) ψ(r.2) d 3 r0 m 2πi 3 2 exp i m (r − r0 )2 − 2 U (r. t) . (3. t + ) = × exp i 2 +∞ +∞ +∞ −∞ dx1 −∞ dx2 −∞ dx3 m x2 + x2 + x2 1 2 3 m 2πi 3 2 × (3. t + |r0 .21). t) . It holds then according to (2. t) by an infinitesimal time step . t) ψ(r0 . t + |r0 . even in x1 .Chapter 3 The Schr¨dinger Equation o 3.

(3. n = 0. 1 2 3 2 ∂2 j=1 xj ∂x2 ψ(r.. t) + .7) According to (2. t) + ∂xj 2 3 xj xk j.36) holds I0 (a) = Inspection of (3. 2 (3. t) j k . . a (3.4) is even in all three coordinates x1 .12) Here one needs to note that we are actually dealing with a three-fold integral.8) one can evaluate recursively all integrals In (a). ∂xj ∂xk (3.5) assuming that only the leading terms contribute. (3. According to (3. The integration involves only the wave function ψ(r +s. Since the latter contributes to (3. t) . only terms of the expansion (3. t) is a constant with respect to this integration. a I2 (a) = − 3 4a2 iπ . t) and the kinetic energy term. O( 5 2 ) .52 Schr¨dinger Equation o It is important to note that the integration is not over r. the terms collected in (3.8) (3. x3 .6) 3 4 ∂4 j=1 xj ∂x4 ψ(r.10) It is now important to note that in case of integral (3. of the expansion of ψ(r + s. x3 )T .k=1 ∂2 ψ(r. It holds In+1 (a) = I1 (a) = i 2a iπ . U (r. t) j 1 12 .4) only for small x2 + x2 + x2 values 1 2 3 we expand 3 ψ(r + s. 1 4 3 2 2 ∂4 j.. x2 . i ∂a Starting from (3. Since the kinetic energy contribution in (3. O( 7 2 ). consequently. x2 . we need then to evaluate integrals of the type +∞ In (a) = −∞ dx x2n exp i a x2 . O( 7 2 ) . e. Obviously. 1.11) and.. . t). but over s = (x1 . It is then sufficient to consider the terms ψ(r.g..9) iπ a (3.6) make contributions of the order O( 3 2 ) . t) + j=1 xj ∂ 1 ψ(r.k=1 xj xk ∂x2 ∂x2 ψ(r. a supposition which will be examined below. t) j .13) 2πi a .11) holds 3 m 3 iπ 2 2 × = 1 (3. t) = ψ(r.7) shows 1 ∂ In (a).4) one identifies 2 1 = a m = O( ) (3..5) which are even separately in all three coordinates yield non-vanishing contributions..

Usually. The 2nd order spatial derivatives require that one specifies also properties of the solution on a closed boundary ∂Ω surrounding the volume Ω in which a solution is to be determined. t) + and exp − i U (r. t) + O( 2 ) ∂t i U (r. t) = ψ(r.17) 2m ψ(r. e. t) + ∂ ∂t ψ(r. t) − + i 2 2 i U (r. t + ) = exp − i U (r. (3. t) + O( 2 ) .e. (3. t) .10)..2 Boundary Conditions The time-dependent Schr¨dinger equation is a partial differential equation. a solution is thought for t ≥ t0 and the solution is specified at the intial time t0 . one specifies the so-called initial condition. 2nd o order in the spatial variables and linear in the solution ψ(r.20) 2m 3. t) + O( 2 ) .2: Boundary Conditions and one can conclude. Obviously. t)ψ(r. The following general remarks can be made about the solution. 1st order in time.. 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 . This equation is often written in the o form ∂ ˆ i ψ(r.g. ψ(r.14) results in ψ(r.19) ∂t where 2 2 ˆ H = − + U (r. (3. (3.3. t) (3. Due to its linear character any linear combination of solutions of the time-dependent Schr¨dinger o equation is also a solution. t) = 1 − (3. t) + O( 2 ) . We will derive briefly the type of boundary conditions encountered. t). t + ) = ψ(r. t) ψ(r. t) = H ψ(r. i.16) Inserting this into (3.14) This expansion in terms of powers of suggests that we also expand ∂ ψ(r. t) + 1 2i 4 m 2 53 ψ(r. t) = ∂t 2 − 2 2m + U (r. this equation is trivially satisfied to order O( 0 ). In order O( ) the equation reads i ∂ ψ(r. t) ψ(r.15) ψ(r. The 1st order time derivative requires that for any solution a single temporal condition needs to be specified. t) (3. ψ(r. using (3. t) . t1 ) = f (r).18) This is the celebrated time-dependent Schr¨dinger equation.

hence. The hermitian property. however. (3. Interchanging f ∗ (r) and g(r) results in g| H |f Ω = Ω ˆ d3 r g(r) Hf ∗ (r) . In (3. Since energy is a real ˆ ˆ quantity one needs to require that the eigenvalues of the operator H are real and.22) and (3.23) ˆ Since H is a differential operator the expressions (3.23). t). f ∗ (r) d3 rg(r) Ω 2 = − 2m d3 rf ∗ (r) Ω 2 g(r) − f ∗ (r) (3. t) g(r) − Ω d3 rg(r) U (r. 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. D. that H is 2 . for example. (John Wiley. g ∈ H is defined as follows f |g Ω = d3 rf ∗ (r)g(r) (3. Calssical Electrodynamics. g|r) = − 2 2m ( f ∗ (r) g(r) − g(r) f ∗ (r) ) (3.28) See.. . in principle. J. 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. The reader is advised to consult a reference text on ‘Linear Algebra’ to follow this argument.26) the vector–valued function P (f ∗ . Jackson. New York. 1975). Chapter 1. differ from each other. g|r) (3.25) ˆ f | H |g Ω ˆ = g| H |f Ω + ∂Ω da · P (f ∗ .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) (3.22) ˆ where H is defined in (3. implies hermitian ˆ f | H |g 1 2 Ω ˆ = g| H |f Ω (3. g|r) is called the ˆ concomitant of H and is P (f ∗ . 2nd Ed.20).26) where ∂Ω da · A(r) denotes an integral over the surface ∂Ω of the volume Ω.27) ˆ We will postulate below that H is an operator in H which represents energy.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.

. Since the total probability of finding the particle anywhere in space is d3 r|f (r)|2 = 1 (3.30) can be postulated.31) |r|→∞ (3. if f and g satisfy these conditions than also does any linear combination αf + βg. t) + ψ(r.. In either case does f (r) and all of its derivatives vanish asymptotically. . to avoid discontinuities in ψ(r. (3. i. obey (3. 3. r ∈ ∂Ω (3. like exp(−κr).3 Particle Flux and Schr¨dinger Equation o The solution of the Schr¨dinger equation is the wave function ψ(r.30) each condition holding on an entire surface ∂Ωj . r ∈ ∂Ω or da · f (r) = 0 ∀r. The probability to observe the particle anywhere in the subvolume ω ⊂ Ω is p(ω. α > 2.29.29) In fact. i. ∂Ω = ∂Ω1 ∩ ∂Ω2 ∩ ∂Ω3 ∩ .34) An example is a volume between two concentric spheres. i. κ > 0 or like r−α . namely. t) ] . t)∂t ψ(r. t)|2 .3. in this case also all derivatives of f (r) vanish at infinity. t).32).33) The time derivative of p(ω.e. t) is ∂t p(ω.e. t)|2 ..32) the wave function must decay for |r| → ∞ rapidly enough to be square integrable. t) = ω 3 d3 r [ ψ ∗ (r. 3. we can only allow functions which make the differential da · P (f ∗ . in which case ∂Ω1 is the inner sphere and ∂Ω2 is the outer sphere. In this case one can expect that the particle density is localized in the area where the energy of the particle exceeds the potential eenrgy. 3.g.. The latter property stems from the fact that the boundary condition (3. t) which describes the state of o a particle moving in the potential U (r.e. It must hold then for all f ∈ H f (r) = 0 ∀r. t) on a single connected surface ∂Ωj only either one of the conditions (3.31) usually arises when a particle existing in a bound state is described. therefore. g|r) vanish on ∂Ω.29. t) = ω d3 r |ψ(r. . A most common boundary condition is encountered for the volume Ω = R3 in which case one postulates lim f (r) = 0 “natural boundary condition” . The observable directly linked to the wave function is the probability to find the particle at position r at time t. and that the density decays rapidly when one moves away from that area. However.3: Particle Flux 55 and.. t)∂t ψ ∗ (r. e.30) Note that these boundary conditions are linear in f . |ψ(r. In this case one can postulate both conditions (3. (3. Often the closed surface of a volume ∂Ω is the union of disconnected surfaces3 ∂Ωj . (3.

40) One can rewrite then (3.27) one can express this j(r.44) ˆ Note that the Hamiltonian H involves only real terms.29. ψ(r. t) ∂t ˆ ˆ ψ ∗ (r. The surface integral (3. t) + 4 · j(r. t)|2 = 1 Ω (3. t) ψ ∗ (r. t) = ω Schr¨dinger Equation o ∂ ∗ ˆ ψ (r. We will assume in the remainder of this Section that the solutions discussed are normalized. t) = 0 (3.43) (3.30) ∂t p(Ω.27) this can be written i ∂t p(ω. 3. t) . t) = ∂ω da · P (ψ ∗ (r.36) According to (3. .41) where j(r. One refers to such solution as normalized.38) holds.37) can be expressed through a volume integral according to da · A(r) = ∂ω ω (3. t) = 0. t)|2 is a probability density with units 1/volume. t) ] .35) d3 r . t) (3. t) = |ψ(r. t) ψ(r. (3. t) H ψ ∗ (r. t) + ω · j(r. t).26. Using (3.41) holds for any volume ω ⊂ Ω one can conclude ∂t ρ(r. t) such that d3 r|ψ(r. t) = [ ψ(r. (3. t) − ψ ∗ (r. t) = H ψ ∗ (r. t) . t) − ψ(r. A natural choice for this constant is 1. t)r. Accordingly the probability to observe the particle anywhere in the total volume Ω is constant. (3.18) by any complex number and accordingly one can define ψ(r. t) = P (ψ ∗ .37) d3 r ∂t ρ(r. t) = 0 . ψ|r. Note that for a normalized wave function the quantity ρ(r. (3.56 Using (3.19) and its conjugate complex4 −i yields i ∂t p(ω.39) d3 r · A(r) (3. t) H ψ(r. One can multiply the solution of (3. 3.42) 2mi Since (3.37) If one applies this identity to ω = Ω one obtains according to (3.

the initial condition at ψ(r. and for particles moving in a linear [c..e..49) ˜ Obviously. t) = 0 2 ∂ 2 i ψ(r.105. 3. t) (3.167)] potential. (2. t0 ) = [2π]− 2 Ω∞ 3 ˜ d3 k φ(k) exp i(k · r − ωt0 ) . 2. 2 for a free particle [c.46) i. 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.4 Solution of the Free Particle Schr¨dinger Equation o We want to consider now solutions of the Schr¨dinger equation (3.e. In Sect. Equation (3.f. (2. 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. equal to the velocity of the particle. t) = − ∂ω da · j(r. (3.43) that any real wave function ψ(r. for f (r) ∈ R.3. 2m (3. t) points in the direction to the outside of volume ω. arises solely from the complex factor exp(i k · r).125)] and in a quadratic [c. Such case arose in Sect.45) Obviously. It is of interest to note from (3. i. t) = [2π]− 2 where the dispersion relationship holds ω = k2 . 2.48.18) in Ω∞ = R in the case o U (r. t) .f.50) 3 ˜ d3 k φ(k) exp i(k · r − ωt) Ω∞ (3.41) written in the form ∂t ω d3 r ρ(r. t0 ) determines φ(k). (2.48) ∂t 2m which describes the motion of free particles. t) has vanishing flux anywhere. t) gives rise to a descrease of the total probability in volume ω due to the disappearence of probability density at the surface ∂ω. t) = − ψ(r.71)].148.47) .51) . (3. One often encounters wave functions of the type φ(r) = f (r) ei k·r The corresponding flux is j(r) = k 2 f (r) .f. 2. (3. (3. This finding is consistent with the present evaluation of the particle flux: a factor exp(ipo xo / ) gives rise to a flux po /m.48) that the general solution is of the form ψ(r.4: Free Particle 57 The interpretation of j(r. t) is that of density flux.49) reads at t = t0 ψ(r. m (3. This follows directly from an inspection of Eq. Note that j(r. j(r. One can readily show by insertion into (3.

Ω∞ (3. We have identified then with (3. respectively. i. Below we will generalize the propagator (3. evaluating the integral in (3.51.54) to the case of non-vanishing potentials U (r). The solution as stated is defined in the infinte space Ω∞ = R .e.51) to be square integrable it follows according to the properties of the Fourier–transform that ψ(r.58 The inverse Fourier transform yields 3 ˜ φ(k) = [2π]− 2 Schr¨dinger Equation o d3 r0 exp(−ik · r0 ) ψ(r0 . To show this one needs to note 1 2π = +∞ dk1 exp −∞ ik1 (x − x0 ) − 1 2 i 2 k2 1 2m . hence. The general form for this propagator involves an expansion in terms of a complete set of eigenfunctions as in (3. t) = Ω∞ d3 r0 φ(r. In fact.54) valid for this case. t|r0 .. If one chooses the initial state f (r) defined in (3. 3.114) and (4.5). t0 ) ψ(r0 .54) yields (2. that the “natural boundary condition” (3. in (2.56) and using (2. The ensuing solutions are called wave packets. In case of the harmonoc oscillator the expansion can be stated in a closed form given in (4. t) as given by (3. t0 ) (3. t|r0 .54) an alternative representation of the propagator (2.70) derived below for a particle in a box and the harmonic oscillator.e. t0 ) = 1 2π 3 d3 k exp Ω∞ ik · (r − r0 ) − i 2 k2 2m (t − t0 ) ..52) We have not specified the spatial boundary condition in case of (3. t0 ) .247).49.53) where φ(r.47).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 (3.31) applies. Comparision with Path Integral Formulation One can write solution (3. (t − t0 ) (3. 3. i. derive an expression similar to (3.54) This expression obviously has the same form as postulated in the path integral formulation of Quantum Mechanics introduced above. (3.49) is square integrable at all subsequent times and.47).81) .49).52) above ψ(r.

57–3. r ∈ ∂Ω or da · φ(r) = 0 ∀r.29.f.30).63) .49. (3.53.46). (3. t|x0 .61) × exp − (x − po m t)2 2δ 2 (1 + 2 t2 ) m2 δ 4 (1 − i . Accordingly.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. t) = −∞ dx0 φ(x. t0 ) = 1 2π +∞ dk exp −∞ ik (x − x0 ) − i 2m (t − t0 ) .56)]. i.52) to the case that the initial state of a 1-dimensional free particle ψ(x0 .55. t0 ) ψ(x0 .3.64) (3. t) = f (t) φ(r) . 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 (3.54) is +∞ ψ(x.59) which is solved through completion of the square in the exponent [c. 3.62) We will restrict the space of allowed solutions to a volume Ω such that the functions also make the concomitant (3. 3.20) which are of o the form ψ(r.57) where φ(x. 3.60) yields with t0 = 0 ψ(x. the functions obey boundary conditions of the type (3.e. The 1-dimensional version of (3. 3. t) is given by (??). a result which is identical to the expression (??) obtained by means of the path integral propagator (2. the boundary conditions are φ(r) = 0 ∀r. t) 2 k2 (3.4: Free Particle Free Particle at Rest 59 We want to apply solution (3. Stationary States We consider now solutions of the time-dependent Schr¨dinger equation (3. We have demonstrated then in this case that the Schr¨dinger formulation of Quantum o Mechanics is equivalent to the Feynman path integral formulation. The remaining integration over k leads to the integral +∞ −∞ dk exp ikx − 1 2 (k − po 2 2 ) δ − i k2 m (t − t0 ) (3.27) vanish on the surface ∂Ω of Ω. (3.60) =???? Combining (3.19. r ∈ ∂Ω (3. t|x0 . (3..

65) is constant.66) i. In order to further characterize the solution (3.. It turns out that a solution for f (t) can be found for any E. It follows then from (3. namely f (t) = f (0) exp − i Et . . We want to demonstrate this property now. often only for a discrete set of values En . We must postulate therefore ∂t f (t) ˆ H φ(r) = E f (t) = E φ(r) .62) do exist and we will characterize the two factors of the solution f (t) and φ(r).. hence. i. (3. if f (t) = eiα . We will demonstrate that solutions of the type (3. This yields an expression g1 (t) h1 (r) = g2 (t) h2 (r) (3.62) exists. is time-independent.69) is called an eigenvalue problem. i. . t)|2 = |φ(r)|2 . 3.69) exist. We will encounter many such problems in the subsequent Sections. according to (3. This can hold only if |f (t)| is time-independent. (3.42) the flux j(r.60 Schr¨dinger Equation o and affect only the spatial wave function φ(r). h1 (r) = φ(r).e.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.31).e. As pointed out above. and h2 (r) = H φ(r). In fact. t)|2 = |f (t)|2 Ω d3 r |φ(r)|2 (3.64). We may note in passing that solutions of the type (3. in general. One can conclude that the probability density for the state (3. for the eigenvalue problems in the confined function space. one factor depending only on the time variable and the other only on the space variables are called separable in space and time.63. n = 1. .e. 2. (3. t) = Ω Ω d3 r |ψ(r. At this point we state without proof that. a common case is Ω = Ω∞ and the ‘natural boundary condition’ (3. for functions required to obey boundary conditions (3.62) are special solutions of the timedependent Schr¨dinger equation.27) holds and.67) (3. α ∈ R .62) have the particular property that the associated probability distributions are independent of time. (3.62) we insert it into (3. One calls such states stationary states. solutions exist only for a set of discrete E ˆ values. We have then shown that ψ(r. t) vanishes on the surface of ∂Ω. It is important to realize that the separable solutions (3. The identity (3. by no means all solutions are of this type.62) is |ψ(r. We denote the corresponding solution by φE (r).68) ˆ where g1 (t) = i ∂t f (t).19).71) . t) = f (0) exp − i E φE (r) where ˆ H φE (r) = E φE (r) . the eigenvalues of the operator H.62) which consist of two factors. the solutions o (3. g2 (t) = f (t). It follows from the observation that for the solution space considered (3. however.69) If these two equations can be solved simultaneously a solution of the type (3. At this point we will accept that solutions φ(r) of the type (3.70) The task of finding solutions φ(r) which solve (3..45) that the total probability d3 r ρ(r.

E (3.71) are. t) = exp − E t 2 φE (r) .28) for the special case that f and g in (3.19.71) this yields E Ω d3 r φ∗ (r) φE (r) = E ∗ E Ω d3 r φE (r) φ∗ (r) . Obviously. (3.e. e. hence. As a result we cannot postulate in the present case that wave functions are localized and normalizable. i. o According to (3. The resulting total energy values E are positive.73) from which follows E = E ∗ and. (3. should be positive.. E ∈ R.4: Free Particle 61 is a solution of the time-dependent Schr¨dinger equation (3.g. . m (3. = E. real. The solution φE (r) corresponding to the eigenvalue problem posed by (3. The corresponding stationary solution ψ(r.74) The classical free particle with constant energy E > 0 moves without bounds in the space Ω∞ .28) both represent the state φE (r). We will wave this assumption as we always need to do later whenever we deal with unbound particles. We want to prove now that the eigenvalues E which arise in the eigenvalue problem (3.43) is j(r.76) according to (3.3. k ∈ R 2 k2 φk (r) = N exp i k · r . The flux corresponding to (3.74) is actually best labelled by an index k. t) = exp − i 2 k2 2m t exp i k · r (3. particles scattered of a potential.76) has kinetic energy 2 k 2 /2m. in fact. We start our proof using the property (3. ˆ d3 r φ∗ (r) H φE (r) = E Ω Ω ˆ d3 r φE (r) H φ∗ (r) . one can interpret then k as the magnitude of the momentum of the particle.66) E must be real. of course.75) 2m One can ascertain this statement by inserting the expression for φk (r) into the eigenvalue problem posed in (3. Stationary State of a Free Particle We consider now the stationary state of a free particle described by i ψ(r. E (3. t) = |N |2 k .74) using exp i k · r = ik exp i k · r . − 2 2m φ(r) = E φ(r) . We will show in Section 5 that E can be interprerted as the total energy of a stationary state. a property which is to be expected since the energy is purely kinetic energy which. 3.20).72) According to (3.77) Noting that k can be interpreted as the magnitude of the momentum of the particle the flux is equal to the velocity of the particle v = k/m multiplied by |N |2 .

a] ⊂ R which vanish on the surface ∂Ω1 = {−a. 5 . 3. let say the one at x = a. t) .o) (−a) = 0 (3. We can expect.82) satisfy the differential equation in (3. (o) 2 k2 (e) 2 k2 2m = E (3.82) have the property that either both boundary conditions are satisfied or none. φ(±a) = 0 . . two types of solutions.80) need to be satisfied are φE (e.80).84) . a}. so-called even solutions obeying φ(x) = φ(−x) φE (x) = A coskx .80) 2m dx2 We note that the box is symmetric with respect to the origin. (3. In case of the even solutions (3.78) (3.81.o) (a) = 0 and φE (e. t) = exp(−iEt/ )φE (x) where φE (x) is determined by 2 d2 − φE (x) = E φE (x) . . 3.83) The solutions (3. . we have to consider only one boundary condition. f ∈ F1 = {f : [−a.81) 2m = E. a] ⊂ R assuming that the potential outside of this interval is infinite.82) One can readily verify that (3. It turns out that this boundary condition can only be satisfied for a discrete set of k–values kn . The boundary conditions which according to (3. n ∈ N. therefore. hence. (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. 2m dx2 2 (± ) = } (3.62 Schr¨dinger Equation o 3. f continuous. We assume.79) The stationary solutions have the form ψ(x. 3.81) they are kn = nπ . i.81.e. t) = − d2 ψ(x. Hence.. a] ⊂ R → R. (3. and so-called odd solutions obeying φ(x) = −φ(−x) φE (x) = A sinkx.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. that the solutions obey this symmetry as well. We will refer to this as a particle in a one-dimensional ‘box’. Solutions of the Schr¨dinger Equation in F1 o The time–dependent solutions satisfy i ∂t ψ(x. 2a n = 1.

a n = 2.81) and (3. n is assumed to be even.86) satisfy the boundary condition since for such kn nπa sin(kn a) = sin 2a = sin nπ 2 = 0. . 2a 1 . The wave functions for the five lowest energies En are presented in Fig. 4. This condition implies for the even states +a |An |2 −a dx cos2 nπx = |An |2 a = 1 . x) = An sin n nπx .82). according to (3. according to the dispersion relationships given in (3. 2a n = 1. . are En = 2π2 8ma2 n2 .94) .1). (3. . By counting the number of their nodes one can determine the energy ordering of the wave functions. (3. 5 . . (3. n = 1. .o) (a. φ(e) (a. 6 . .93) The normalization constants are then An = (3. .82) only the kn –values cos(kn a) = cos kn = nπ .e. 3.91) holds. 2a n = 2.87) (Note that. 6 . 5 .86).85) (3. (3. 4. 3. . 4. It is desirable to normalize the wave functions such that +a −a dx |φ(e. 6 . (3. .86). x) = An cos n and φ(o) (a. = 0. 2a n = 1.89) nπx . n = 2. 3.88) where the energies for odd (even) n–values correspond to the even (odd) solutions given in (3. 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 . (3.3. (3. .. x)|2 = 1 n (3.5: Particle in One-Dimensional Box since for such kn 63 nπa nπ = cos 2a 2 In case of the odd solutions (3. (3. 3 .81. 3.84.92) and for the odd states +a |An |2 −a dx sin2 nπx = |An |2 a = 1 . .82).90) 2a The wave functions represent stationary states of the particle in a one-dimensional box. i. . respectively. 2. .) The Energy Spectrum and Stationary State Wave Functions The energy values corresponding to the kn –values in (3.

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

(3.e.90. m even φn |φm 1 a Ω1 = 1 a +a nπx −a dx sin 2a sin mπx = 2a (3.} where φn (a. n (3. for n = 2. f. n = m. n = 1.103) We will show in Section 5 that the property of a scalar product do indeed apply. i.3.e. In case of n = m we have to consider three cases. (ii) n. it holds: f |f Ω1 = 0 → f (x) ≡ 0. . too. The latter case leads to integrals φn |φm Ω1 = 1 a +a dx cos −a nπx mπx sin .89.100) +a −a dx cos (n−m)πx + cos (n+m)πx 2a 2a The periods of the two cos-functions in the interval [−a.97) form an orthonormal basis set. Similarly. hence.96) f |g Ω1 = −a dxf (x) g(x) .102) holds according to (3.101) +a −a dx cos (n−m)πx − cos (n+m)πx 2a 2a and. . 3.95) 1 a cos nπx 2a sin nπx 2a for n = 1. In case of n. . x) = together with the scalar product5 +a 65 (3. 3. In particular. x) (3. m both odd.. g ∈ F1 (3. the integral arises φn |φm 1 a Ω1 = 1 a +a nπx −a dx cos 2a cos mπx = 2a (3. (3. N ≥ 1. 3.5: 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 odd.98) The latter property is obviously true for n = m. (i) n.99) Since in this case the integrand is a product of an even and of an odd function. for ∞ f (x) = n=1 dn φn (a. . this integral vanishes. m both even. a] are N. Hence we need to consider only the first two cases. the integral vanishes. and (iii) the mixed case. 3. 2a 2a (3. In fact.. the integrals vanish.94) B1 = {φn (a. 6 . Because of the property (3. Obviously. 5 . . i.98) ∞ f |f 5 Ω1 = n=1 d2 . . it holds φn |φm Ω1 = δn m . the integrand is odd. . x) . 2. 4. .98) the elements of B1 must be linearly independent. one obtains for n.

.98) one obtains +a dx0 φm (a. a] yields then also an expansion for any element in F1 and B1 is a complete basis for F1 . (3. the coefficients an . x) cn (t) −a dx0 φn (a.66 Schr¨dinger Equation o f (x) ≡ 0 implies f |f Ω1 = 0 which in turn implies dn = 0 since d2 ≥ 0.. For this purpose we extend the definition of the elements of F1 to the whole real axis through ˜ f : R → R. . 2. . such demonstration can be based on the theory of Fourier series. . there exist real constants {an . n = 2. 4.105) 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. . i.} such that ∞ ∞ nπx nπx ˜ = + an sin (3. x0 ) ψ(x0 . t0 ) in terms of eigenfunctions φn (a. . −a (3. t0 ) = dm . x0 ) ψ(x0 . This implies that only the trigonometric functions which are elements of B1 enter into the Fourier ˜ series.108) into the Schr¨dinger equation yields o +a ∞ n=1 φn (a. t0 ) ..108) where the functions cn (t) are to be determined from the Schr¨dinger equation (3. they can be expanded in terms of a Fourier series.95) is also complete. (3. t0 ) (3.106) Using the orthonormality property (3. t) = n=1 φn (a. 3.109) Insertion of (3. x) to obtain an expression for ψ(x. . We like to show finally that the basis (3. Restricting the expansion (3. .107) Inserting this into (3. 3. .} and {bn . x) . In the present case. Hence. x) ∂t cn (t) −a dx0 φn (a.e. . x0 ) ψ(x0 . Evaluating the Propagator We can now use the expansion of any initial wave function ψ(x. .95) is an orthonormal basis. . For this purpose we expand ∞ ψ(x.e. n = 0.96). The functions f are periodic with period 2a. in (3. ( ) = ([ + ] − ) (3. t0 ) = +a ∞ i n=1 − En φn (a. t) at times t > t0 . 1.104) ˜ where [y]a = y mod2a. however. It follows that B1 n defined in (3. n = 1. t0 ) = n=1 dn φn (a. 5 . Accordingly. 2.95. x0 ) ψ(x0 . x) cn (t) −a dx0 φn (a.110) . and bn .79) requiring the o initial condition cn (t0 ) = 1 . any element of the function space F1 defined in (3.105) must vanish. (3. We have then shown that any f corresponding to elements of F1 can be expanded in terms of elements in B1 . i.78) can be expressed as a linear combination of the elements of B1 defined in (3. n = 1.105) to the interval [−a. 3.106) and generalizing to t ≥ t0 one can write ∞ +a ψ(x. . . m = 1. Demonstration of completeness is a formidable task.

We have identified then with (3. t0 ) = ∞ φn (a. the elements of which satisfy the appropriate boundary conditions. t|x0 . The respective initial condition is φ(x. The last frame shows the wave front reaching again the right wall and the onset of new interference. t|x0 . The particle moves then to the left. (3. t0 ) (3.5). cm (t0 ) = 1 . t|x0 . t0 ) itself is a solution of the time-dependent Schr¨dinger o equation (3. . In case that different boundary conditions hold the propagator will be different as well. This solution can be written +a ψ(x. x0 ) . t0 ) ψ(x0 .. t0 ) = δ(x − x0 ) as can be readily verified using (3. t0 ) is also referred to as a Greens function. a] yields. t) = −a dx0 φ(x. Example of a Non-Stationary State As an illustration of a non-stationary state we consider a particle in an initial state ψ(x0 . k0 = . (3.112) Equations (3. It is of interest to note that φ(x.78). Often the propagator φ(x. σ = a 15 . t|x0 .114) This expression has the same form as postulated in the path integral formulation of Quantum Mechanics introduced above. i. x) and integrating over [−a. t) for any initial condition ψ(x.115) This initial state corresponds to the particle being localized initially near x0 = 0 with a velocity v0 = 15 /m a in the direction of the positive x-axis.114) the representation of the propagator for a particle in a box with infinite walls.2 presents the probability distribution of the particle at subsequent times. x) exp n=1 i − En (t − t0 ) φn (a.113). in (2.108. 67 (3.54) as in the case of the free particle system which does not exhibit any bound states. t0 ).113) where φ(x.5: Particle in One-Dimensional Box Multiplying both sides by φm (a. t0 |x0 . according to (3.109) are cm (t) = exp i − Em (t − t0 ) . 3. but the collision with the left wall leads to new interference effects.112) determine now ψ(x. Note that the propagator has been evaluated in terms of elements of a particular function space F1 .3. The interference pattern begins to ‘smear out’ first. being reflected at the right wall. Figure 3.111) The solutions of these equations which satisfy (3.79) which lies in the proper function space (3. In the present system which is composed solely of bound states the propagator is given by a sum. 4 a (3.e. One can recognize that the particle moves first to the left and that near the right wall of the box interference effects develop. t0 ) = 1 2πσ 2 1 4 exp − x2 0 + i k0 x0 2σ 2 .98). rather than by an integral (3. i ∂t cm (t) = − Em cm (t) .

44 m a h2 |Ψ|2 2 t = 1.24 m a h2 1 πσ k = 15 a 1 πσ -a a -a a |Ψ|2 2 t = 0.20 m a h2 1 πσ 1 πσ -a a -a a |Ψ|2 2 t = 1.68 Schr¨dinger Equation o |Ψ|2 2 t = 0.16 m a h2   1 πσ 1 πσ -a a -a a Figure 3. t)|2 for a particle in a box starting in a Gaussian distribution with momentum 15 /a.92 m a h2 |Ψ|2 2 t = 2.00 m a h2 |Ψ|2 2 t = 0.2: Stroboscopic views of the probability distribution |ψ(x.96 m a h2   |Ψ|2 2 t = 1.68 m a h2 1 πσ 1 πσ -a a -a a |Ψ|2 2 t = 1. .48 m a h2 |Ψ|2 2 t = 0.72 m a h2 1 πσ 1 πσ -a a -a a |Ψ|2 2 t = 0.

x3 ) = j=1 φ(j) (xj ) (3. x2 . a1 ] ⊗ [−a2 .117) 2m 2 2 2 ∂1 + ∂2 + ∂3 (3. f continuous. one can express H 3 φ(x1 . We have defined in the space F1 a scalar product (3. ˆ H = − 2 (3. x2 .121) where − 2 2m 2 ∂j φ(j) (xj ) = Ej φ(j) (xj ) . hence..120) ˆ Since the Hamiltonian H is a sum of operators O(xj ) each dependent only on a single variable.114) in terms of which the solutions for all initial conditions can be expressed. . An important property of this basis and.e.122) . 2a2 . Placing the origin at the center and aligning the x1 . x3 )T ∈ Ω. t) = H ψ(x1 . A complete basis of F1 is given by the infinite set B1 (3. x1 = ±a1 } ∪ {(x1 .118) which are stationary states. x2 . x3 ) = E φE (x1 . x2 ..e. 2a3 . x2 . x3 –axes with the edges of the box yields spatial boundary conditions which are obeyed by the elements of the function space F3 = {f : Ω → R. x3 )T ∈ Ω. of the space F1 is that the elements of the basis set can be enumerated by integer numbers. (3. x3 )T ∈ Ω. The corresponding solutions have the form i ψ(x1 .116) where Ω is the interior of the box and ∂Ω its surface Ω ∂Ω = [−a1 . ( . 3 (3.3. ˆ = O(x1 ) + O(x2 ) + O(x3 ). a3 ] ⊂ R = {(x1 . We seek then solutions of the time-dependent Schr¨dinger equation o ˆ i ∂t ψ(x. j = 1. This property allowed us to evaluate the propagator (3.97) with respect to which the eigenfunctions are orthonormal.6 Particle in Three-Dimensional Box We consider now a particle moving in a three-dimensional rectangular box with side lengths 2a1 .6: Particle in Three-Dimensional Box Summary: Particle in One-Dimensional Box 69 We like to summarize our description of the particle in the one-dimensional box from a point of view which will be elaborated further in Section 5. x3 ) (3. x3 . x3 .116) and obeys the partial differential equation ˆ H φE (x1 . i.95). can be counted. a2 ] ⊗ [−a3 . t) . i.78). x3 ) is an element of the function space F3 defined in (3. x2 . x2 . x2 = ±a2 } ∪ {(x1 . . )T ∈ ∂ } . ) = ∀( . 3. x2 . x2 . x2 . x3 = ±a3 } . x3 ) . 2. (3. φ(j) (±aj ) = 0 . t) = exp − E t φE (x1 . x2 . The description employed a space of functions F1 defined in (3.119) where φE (x1 .

a3 .n3 ) (a1 . a2 . n2 . n3 = 1. a3 .70 Schr¨dinger Equation o and E1 + E2 + E3 = E. i. x1 . . t|r0 . x2 . x1 ) φn2 (a2 . For example.124) The same considerations as in the one-dimensional case allow one to show that B3 = {φ(n1 .123) 8m n2 n2 n2 1 2 3 + 2 + 2 a2 a2 a3 1 .n2 . 3. (3.n3 ) (a1 .n3 ) (a1 . 3. x2 . a2 .122) with (3.. x3 –axes by π/2 also leaves the system unaltered. Actually.n3 ) (a1 . 2. a1 = a2 = a3 = a then rotation around the x1 .n2 . (3. r) exp i − E(n1 n2 n3 ) (t − t0 ) φ(n1 . (3.n3 =1 φ(n1 .e. hence. Such degeneracies are always a signature of an underlying symmetry. the solutions of (3. 2. .96) and. If two or all three orthogonal sides of the box have the same length further symmetry operations leave the system unaltered. a2 . if all three lengths a1 . a3 . x2 .n2 . . x3 ) .126) Symmetries The three-dimensional box confining a particle allows three symmetry operations that leave the box unchanged. n3 = 1.122) are given by (3. n 2 . . 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. x3 ) = 1. . 3. . x2 ) φn3 (a3 .} ∞ (3. t0 ) = n1 . This symmetry has been exploited in deriving the stationary states.80) shows that the solutions of (3. n1 . in the present case . This additional symmetry is also reflected by degeneracies in the energy levels.n2 . namely rotation by π around the x1 . x2 .125) is a complete orthonormal basis of F3 and that the propagator for the three-dimensional box is φ(r. x3 ) n1 . a3 . x1 . a3 are identical. .n2 . 2. r0 ) . .120) can be written φ(n1 . a2 .n2 . n 3 and E(n1 . Comparing (3. . a2 . n2 . x3 –axes.n3 ) = 2π2 = φn1 (a1 . n1 .

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

72 Schr¨dinger Equation o .

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

however. (4. the natural boundary conditions apply x→±∞ lim φE (x) = 0 .5) →±∞ where C∞ denotes the set of functions which together with all of their derivatives are continuous.5) ˆ H = 1 1 2 p + m ω 2 x2 ˆ ˆ 2m 2 (4.3) would be continuous and the eigenfunctions φE (x) would not be normalizable. We will point out explicitly where assumptions are made which built on this restriction.5). x] = p ˆ . t) = exp(−iEt/ )φE (x. the commutator has a simple form. all stationary states of the harmonic oscillator must be bound states and. In order [ˆ. beside the representation (4. is the commutation property 1 1 (4.235) had been determined. ∈ C∞ . lim ( ) = } (4. only for a discrete set of E values can the boundary conditions (4.7) which is why these operators are of interest to us. If this restriction would not apply and all functions f : R → R would be admitted.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. in which the operators used below. (4.1 Creation and Annihilation Operators The Hamiltonian operator (4. therefore.6) the first being a differential operator and the second a multiplicative operator.4) be satisfied. t) where ˆ H φE (x) = E φE (x) .3) can be solved for any E ∈ R. The operators act ˆ on the space of functions N1 defined in (4. the spectrum of ˆ H in (4.74 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.8) i which holds for any position and momentum operator.2) can be expressed in terms of the two operators p = ˆ d . In the following algebraic solution of (4. act.1) of the form ψ(x. The Hamiltonian H can be expressed in terms of the operators acting on the space (4.3) Due to the nature of the harmonic potential which does not allow a particle with finite energy to move to arbitrary large distances. This property states that p and x obey an ˆ ˆ algebra in which the two do not commute. The cardinal property exploited below. It is important to keep in mind this restriction of the space. however. 4.4) Equation (4. In the framework of the Schr¨dinger equation the stationary o states are solutions of (4. i dx x = x ˆ (4. The stationary states of the harmonic oscillator have been considered already in Chapter 2 where the corresponding wave functions (2.7) of the Hamiltonian.

the action of [ˆ. to factor H/ ω. x] . a+ ] = a− a+ − a+ a− does not necessarily vanish.11) ˆ The reader may note that we have attempted.18) It is of interest to note that the operators a+ and a− are real differential operators.8). 2mω dx a− = ˆ mω x + ˆ 2 d .7) assuming that the factors are easier to handle than the Hamiltonian in yielding spectrum and eigenstates. p ˆ 2 2m ω 2 (4. the commutator property (4. in fact. 4.12). One obtains [ˆ. ˆ 2 2m ω a− = ˆ mω i x + √ ˆ p. Being guided by the identity for scalar numbers (b − ic)(b + ic) = b2 − i(cb − bc) + c2 = b2 + c2 (4. 1 ω 2 (4.1: Creation and Annihilation Operators 75 to derive (4.17) d dx We also express a+ and a− directly in terms of the coordinate x and the differential operator ˆ ˆ a+ = ˆ mω x − 2 d . x] p ˆ on such functions must be considered. 1 i dx i dx i i (4. ˆ 2 2m ω (4. Before we continue we like to write (4.8) yield a− a+ = ˆ ˆ Similarly one can show a+ a− = ˆ ˆ 1 ˆ 1 H − 1 . In fact.9) From this follows (4. a ˆ 1 (4.10) we define a+ = ˆ mω i x − √ ˆ p. 2mω dx (4. Since a+ and a− are ˆ ˆ + and a− since operators and not scalars we cannot simply expect that the identy (4.15) in a form which will be useful below ˆ H = ˆ H = ω a− a+ − ˆ ˆ ω a+ a− + ˆ ˆ ω 1 1 2 ω 1 . hence. ˆ ˆ .7) and (4.15) together lead to the commutation property (4.16) (4. Our next step is an attempt to factorize the Hamiltonian (4. a+ ] = 1 .12) To prove this commutation property we determine using (4. x] f (x) = p ˆ d d x f (x) − x f (x) = f (x) = 1 f (x) .11) a− a+ = ˆ ˆ (4.10) holds for a ˆ ˆ [ˆ− .8) we recall that all operators act on functions in N1 and.15) 1 ˆ 1 H + 1 .14) 1 i mω 2 x + ˆ p2 + ˆ [ˆ.4. 1 2 (4.14) and (4. 1 ω 2 (4.14.13) (4.8) a ˆ ˆ ˆ ˆ ˆ implies [ˆ− .

19).17.17. (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) . In fact. The derivation will be based solely on the properties (4.16.21) 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) .19) This property states that the operators a+ and a− are the adjoints of each other. ˆ (4. g ∈ N1 +∞ +∞ dx f (x) a+ g(x) = −∞ −∞ dx g(x) a− f (x) .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.24) − m This property states that the operators in the function space N1 are the hermitian conjugate of each other.16. (4. it holds that for a stationary state φE (x) of energy E defined through (4. ˆ The results (4.22.16.21) is available. 4.19) in the form1 f |ˆ+ g a Ω∞ = g|ˆ− f a Ω∞ . ˆ (4.20) ˆ In the following we will determine the spectrum of H and its eigenstates.23) Together. This property of operators is investigated more systematically in Section 5.21) 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. 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. 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. 4.22) Similarly one can show using again (4.12. 4.17. (4. Using (??) we like to state (4. 4. 4. The property ˆ ˆ follows directly from (??). 4. one obtains using (4.

on ˆ the other side such a state is itself an eigenstate of H as can be shown using (4.. in fact.33) . φ0 (y) dy dy the solution of which is 1 lnφ0 (y) = − y 2 + c0 2 φ0 (y) = c0 exp 1 − y2 2 .11) one can rewrite (4.) ˆ ˆ 0 = φE |HφE Ω∞ − φE |HφE Ω∞ = ( E − E ) φE |φE Ω∞ .31) (4.6.29) where y = mω x.30) Assuming that φ0 (y) does not vanish anywhere in its domain ] − ∞.27) Since E = E one can conclude φE |φE Ω∞ = 0. i.25) needs to be normalizable in order to represent a bound state of the harmonic oscillator.32) for some constant c0 or (4. E = E .25) on one side would lead to termination of the sequence E0 + m. +∞[ one can write (4.29) d 1 d φ0 (y) = = lnφ0 (y) = − y .26) Of course. Another name is creation (ˆ+ ) and annihilation (ˆ− ) operators since these operators a a create and annihilate vibrational quanta ω.2 Ground State of the Harmonic Oscillator d + y dy A suitable solution of (4. in fact.28) 4. Using (4.e. E . 4. For obvious reasons one refers to a+ and a− as ladder ˆ ˆ operators. The property (??) has an important consequence for the stationary states φE (x). be found. (4.5).4. (4. For f (x) = φE (x) and g(x) = φE (x) in (??) follows (Note that according to (??) the eigenvalue E is real. There arise. It turns out that both difficulties can be resolved together. the solution φ0 (x) of (4. m ∈ Z when m is decreased.25) can. (4. 2 2 (4. (4. In fact.17) ˆ H φ0 (x) = ( ω a+ a− + ˆ ˆ 1 ω 1 ) φ0 (x) = 1 ω φ0 (x) . φ0 (x) should be an element of the function space N1 defined in (4.2: 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. one expects that E = 0 should be a lower bound. however. two difficulties: (i) one needs to know at least one stationary state to start the construction. Let φE (x) and φE (x) be two normalized stationary states corresponding to two different energies E. (ii) the construction appears to yield energy eigenvalues without lower bounds when. a state φ0 (x) which obeys the property a− φ0 (x) = 0 ˆ (4.25) φ0 (y) = 0 (4.

36) assumes a functional form such that both terms together assume a minimum value. it holds ˆ H φn (x) = ω(n + 1 ) φn (x) .. (4. which for a narrowly localized state yields a large positive value. We recall that according to (4. i.37) above the ground state energy. a state confined to the minimum corresponding to the classical state of lowest energy.78 Linear Harmonic Oscillator This solution is obviously normalizable. . .36) Since the first order differential equation (4. . we construct the states for n = 1. Such states need to be suitably normalized for which purpose we introduce a normalization constant cn φn (x) = cn a+ ˆ n φ0 (x) . (4. 2. i. (4. (4. and the kinetic energy contribution.e. The conventional normalization condition φ0 |φ0 reads +∞ +∞ Ω∞ = 1 (4. For this purpose we apply the operator a+ to the ˆ ground state (4. choose the normalization constants cn and the functions φn (x) real as well. The set of allowed energies of the oscillator according to (4.34) dx |φ0 (x)|2 = |c0 |2 −∞ −∞ dx exp − mωx2 2 = |c0 |2 π . mω (4. .. 2.e.e. . m ∈ Z which properly terminate at some minimum value E + m0 ω ≥ 0. would yield a vanishing contribution. 2 n = 0. 2 n = 0.. We a can.39) We notice that the ground state wave function φ0 (x) as well as the operators (ˆ+ )n are real. 1. 4. .3 Excited States of the Harmonic Oscillator Having obtained a suitable stationary state with lowest energy. 2 This state of lowest energy is called the ground state of the oscillator. therefore. the so-called excited states. the potential energy contribution which for a state δ(x).2) there are two competing contributions to the energy. The reason is that in the Hamiltonian (4. . n = 0. we can now construct the stationary states corresponding to energies (4. 2. i.39).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. 1.36) n times. .26) the energy value associated with this state is 1 ω.25) admits only one solution there is only one set of states with energy E + m ω. .38) These states correspond to the energy eigenvalues (4..24) can then be written as follows En = ω(n + 1 ). . .35) The appropriate ground state is φ0 (x) = mω π 1 4 exp − mωx2 2 . . 1. . 2. We will consider this point more systematically in Section 5. The ground state (4.

e. To determine these constants we adopt a recursion scheme. (4.50) φn (x) . √ a− φn (x) = n φn−1 (x) . 1.47).42) ˆ Using (4.49) (4. Since a− is a real differential operator [see (4. Ω∞ 2 = βn a− φn |ˆ− φn ˆ a Ω∞ 2 = βn φn |ˆ+ a− φn a ˆ Ω∞ .15 . (4. according to (4.43) (4. 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 . . n.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) . We consider then the situation that we have obtained a properly normalized state φn (x). βn must be real as well. i. .46) According to (4.20) 1 = φn−1 |φn−1 Equations (4. (4.18)] and since the ˆ φn (x) are real functions.48) (4. 2.3: Excited States 79 We need to determine now the normalization constants cn .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.40–4. according to (4. (4. .40). .24) holds in analogy to (4. ˆ (4..20) yields 2 αn φn |ˆ− a+ φn a ˆ Ω∞ = 1. 2. n = 0. . (4.41) Employing the adjoint property (4.4.51) .40) = 1. . √ a+ φn (x) = n + 1 φn+1 (x) . they obey φn |φn = δn n . For n = 0 holds c0 = 1.39) yield √ From this follows βn = 1/ n and.47) for some suitable constants βn . n = 0. (4.40) φn−1 (x) = βn a− φn (x) ˆ (4. 1. 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. 4.45) form an orthonormal set.45) We like to note that these functions according to (4. To determine βn we note using (4. .28) and the construction (4.

53) We will later re-introduce this factor to account for the proper normalization (4.56) Relationship to Hermite Polynomials We want to demonstrate now that the expression (4.45). (4. however. i. by dx = dy mω 1 4 .56) can be expressed in terms of Hermite polynomials Hn (y) introduced in Sect. simplifying the calculation.55) 1 y2 2 (4.30) and employing the normalization +∞ −∞ dy φ2 (y) = 1 n (4.35) by a constant.7. For this purpose we start from expression (4.52) This normalization of the wave functions differs from that postulated in (4.80 Evaluating the Stationary States Linear Harmonic Oscillator We want to derive now an analytical expression for the stationary state wave functions φn (x) defined through (4. We will demonstrate below the identity Hn (y) = e y2 2 y − d dy n e− y2 2 (4. . 2.58) This result agrees with the expression (2.e.40–4. by introducing the variable y defined in (4. (4.233) derived in Sect.7 and given. 2.28) of the ground state and the definition (4.200). namely the square root of the Jacobian dx/dy.35) rather than (4.54) 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 . 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 . by the Rodrigues formula (2.57) which includes the factor 1/ n! according to Eqs.. (4. It should be noted that √ the normalization (4. n–independent factor.52). (4.46) yields a set of normalized states.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) . (4.39). for example.

t0 ).59) dn −y2 e dy n (4.59) holds for n = 0. dy n (4. i.64) which implies that (4. For the sake of notational simplicity we employ initially the coordinate y as defined in (4. (4. the Rodrigues formula (2. (4.61) and.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 .4 Propagator for the Harmonic Oscillator We consider now the solution of the time-dependent Schr¨dinger equation of the harmonic oscillator o (4.65) . t0 ) = n=0 dn φn (y0 ) .106–3.1.63) one obtains g(y) = y − d dy f (y) (4. and showing then that the property also holds for n + 1 in case it holds for n.61) One can factor g(y) and employ (4.57) of the hermite polynomials and the definition given by (2.62) Denoting f (y) = (y − d/dy)exp(−y 2 /2) and employing y2 y2 d d − y2 e 2 f (y) = −y e− 2 f (y) + e− 2 f (y) dy dy (4. Starting point of our derivation is the assumption that the initial condition can be expanded in terms of the eigenstates φn (y) (4. (3. Our derivation will follow closely the procedure adopted for the case of a ‘particle in a box’ [see Eqs.59) by induction noting first that (4. 4. therefore.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 ..2) for an arbitrary initial wave function ψ(x0 . 1.60) and.30) and return later to the coordinate x. (4.e. hence.4. dy dy (4. g(y) = (−1)n+1 e y2 2 dn+1 −y2 e = dy n+1 y − d dy n+1 e− y2 2 . We prove (4.39) ∞ ψ(y0 .59) hold. 4.4: Propagator 81 To verify the realtionship between the definition (4.114)].200).

82 Linear Harmonic Oscillator Such expansion is possible for any f (y0 ) = ψ(y0 .70) and (2.147) we employ the technique of generating functions as in Sect.72) × exp( −u − v + 2iyu + 2iy0 v ) .15. t1 ) = n=0 dn φn (y) cn (t) (4. t|y0 . sec:harm.67) for which according to (3. t0 ) = n=0 φn (y) φn (y0 ) tn+ 2 . N. We want to demonstrate now that this propagator is identical to the propagator (2. We consider for this purpose the following expression for |t| < 1 ∞ w(y. y0 . 2 A demonstration of this property can be found in Special Functions and their Applications by N. 2n n! π 2 2 (4. t0 ) ψ(y0 . For this purpose we start from the integral representation (2.65) to times t ≥ t0 through insertion of time-dependent coefficients cn (t1 ) ∞ ψ(y.108–3. t) = π −3/2 exp 2 2 y 2 y0 + 2 2 2 +∞ +∞ ∞ du −∞ −∞ dv n=0 1 ( −2tuv )n × n! (4. Employing orthogonality condition (4.7. (4.N. a supposition which is not proven here2 .69) where ∞ φ(y.. t) (4.h. t1 ) = −∞ dy0 φ(y.70) is the propagator of the linear harmonic oscillator. t0 ) which is an element of N1 defined in (4. 1965) Sect. . 4.J.68) Altogether one can express then the solution +∞ ψ(y. t0 ) φn (y0 ) . this is an excellent textbook from which we have borrowed heavily in this Section. In order to prove the equivalence of (4. y0 . pp. t1 |y0 .225) which allows one to derive a generating function for products of Hermite polynomials which can be applied to the r.112) and (4.70). (4.Lebedev (Prentice Hall. Inc. of (4.5). t) = n=0 Hn (y)e−y /2 Hn (y0 )e−y0 /2 n √ t .147) for the harmonic iscillator determined in Sect.71) Applying (??) to express Hn (y) and Hn (y0 ) yields exp(−2tuv) w(y.46) the expansion coefficients dn are +∞ dn = −∞ dy0 ψ(y0 .66) One can extend expansion (4. 1 t = e−iω(t1 −t0 ) (4.s. 68.39) holds cn (t1 ) = exp −i ω ( n + 1 ) ( t1 − t0 ) 2 .. Englewood Cliffs. 2.

a Re a2 > 0 (4.77) The sum in (4. one obtains w(y. t|x0 . t0 ) exp i 2 1 2 (y + y0 ) G(t1 .58) of the harmonic oscillator ∞ φn (y) φn (y0 ) tn = n=0 2 1 t 2 1 + t exp − (y 2 + y0 ) + 2 y y0 2 2) 2 1−t 1 − t2 π(1 − t 1 . (4. t) = π −3/2 exp +∞ 83 y2 y2 + 0 2 2 +∞ dv exp( −v 2 + 2iy0 v ) × −∞ × −∞ du exp( −u2 − 2u(tv − iy) ) .75) Applying (4. indeed.76) One can express this in terms of the stationary states (4. t0 ) = exp mω 2iπ sinω(t1 − t0 ) 1 2 × . . i. t1 |y0 . 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 = .78) in terms of the coordinates x and x0 .80) −2iω(t1 −t0 ) sinω(t1 − t0 ) 1−e We can finally express the propagator (4.4: Propagator Carrying out the sum on the r. √ πexp( t2 v 2 − 2iytv − y 2 ) (4. it holds φ(y. y0 . −∞ (4. y0 . t0 ) (4.73) The incomplete Gaussian integral +∞ −∞ dx e−a 2 x2 √ − 2bx = π b2 /a2 e .74) a second time yields finally together with the definition (4. t0 ) = where F (t1 .30) to replace y and y0 and that we multiply the propagator by mω/ .. (4.e. This requires that we employ (4.. t0 ) = and 1 2 i π F (t1 . identical to the generating function (4.79) 1 + e−2iω(t1 −t0 ) cosω(t1 − t0 ) = . t) = π −1 exp − 2 y 2 y0 + 2 2 +∞ dv exp(−(1 − t2 ) v 2 − 2i (yt − y0 ) v) .70) is. (4.74) applied once results in w(y. (4.4.78) 1 − e−2iω(t1 −t0 ) = sinω(t1 − t0 ) 2 i e−iω(t1 −t0 ) (4.81) imω 2 sinω(t1 − t0 ) (x2 + x2 ) cosω(t1 − t0 ) − 2 x x0 0 This result agrees with the propagator (2. by a factor 4 mω/ for both φn (y) and φn (y0 ).71) of w(y.e. i.s.147) derived by means of the path integral description.77). y0 . The resulting propagator is G(t1 .h. t0 ) = i φ(x. t0 ) − i y y0 2 F (t1 .

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

a+ ] + [ˆ− . a a − (4. fn+1 (ˆ+ )] a a = [ˆ− . 4.5: Working with Creation and Annihilation Operators and similarly for f (ˆ− ).93) one can reduce all operators acting on some proper state function by operators acting on the state φ0 (y). the second case follows from [ˆ− .96) . a a a We like to state the following property of the functions f (ˆ± ) a f (ˆ− ) φ|ψ = φ|f (ˆ+ ) ψ .90) holds for any function fn (ˆ+ ) = (ˆ+ )n which is a power of a a a+ .90) holds for f0 and for f1 .90) To prove (4.95) (4.87) (4.88) we need to show actually [ˆ− .88) is (ˆ− )n f (ˆ+ ) φo (y) = f (n) (ˆ+ ) φ0 (y) . f (ˆ+ )] = f (1) (ˆ+ ) . a An important operator function is the exponential function. (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. The first case is trivial. a+ ] = 1 = f1 (ˆ+ ) . a− f (ˆ+ ) φ0 (y) = f (1) (ˆ+ ) φ0 (y) ˆ a a which follows from a− φo (y) = 0.83) and can be proven for all powers fn by induction and then inferred for all proper functions f (ˆ± ). a a (4. a a a 85 (4. The convergence of the Taylor expansion ascertains then that (4.90) holds for f (ˆ+ ). property (4. a a a ˆ 1 a Let us assume that (4.4.89) (4.92) f (y) = n=0 dn φn (y) = n=0 (ˆ+ )n a dn √ φ0 (y) n! (4.94) This identity follows from (4. ˆ a We proceed by induction noticing first that (4.87. The following important property holds a a− f (ˆ+ ) = f (1) (ˆ+ ) + f (ˆ+ ) a− .. It holds a a euˆ f (ˆ+ ) φ0 (y) = f (ˆ+ + u) φ0 (y) .88) (4. We note ˆ a− f (ˆ+ ) = [ˆ− . Hence. we will only assume operator functions acting on φ0 (y).88) is a fundamental one and will be used in the following. An example is the so-called shift operator exp(uˆ− ). For fn+1 follows then [ˆ− .90) holds for fn .91) (4.e. f1 (ˆ+ )] = [ˆ− . ˆ a a a ˆ In particular. 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+ ). f (ˆ+ )] + f (ˆ+ ) a− ˆ a a a a ˆ which implies that in order to prove (4. i. (ˆ+ )n a+ ] = (ˆ+ )n [ˆ− . Note that ˆ a an immediate consequence of (4.90) we show that (4.

+ (4. the solution of (4. one obtains a evˆ + − v ∗ a− ˆ a φ0 (y) = e− 2 vv evˆ e− v 1 ∗ + ∗ a− ˆ φ0 (y) (4. (4. + ∗ a− ˆ a − u euzˆ euz + ∗ a− ˆ = (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. + ∗ + (4.99) ˆ C(u) = zˆ a zˆ+ + z ∗ a− a ˆ + +z a ˆ ∗ − a euzˆ euz ˆ − u C(u) . z ∈ C.97) An example in which (4. a a ν! (4.106) Applying (4.108) . We assume in the following a ˆ derivation. 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) .103) (4.96) is applied is a a a euˆ evˆ φ0 (y) = ev (ˆ − + + + u) a φ0 (y) = euv evˆ φ0 (y) .101) ˆ ˆ ˆ C(u) obviously obeys C(0) = 1 This results in D(0) = 1 and.105) a eαˆ + − α∗ a− ˆ φ0 (y) = e αα∗ 2 e−α ∗ a− ˆ a eαˆ φ0 (y) .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) . that the operators considered act on φ0 (y). euzˆ ˆ euz ∗ a− ˆ (4. To express ˆ these operators as products of operators exp(vˆ+ ) and exp(v ∗ a− ) we consider the operator C(u) = a ˆ + ) exp(uz ∗ a− ) where u ∈ R.106) yields a eαˆ + ∗ a− ˆ a a eαˆ φ0 (y) = e−αα eαˆ φ0 (y) . + (4.107) − α∗ a− ˆ φ0 (y) = e− αα∗ 2 a eαˆ φ0 (y) .102) Similarly.98) for u = −α∗ and v = α yields e−α This together with (4. 1 + + uz ∗ a− ). hence.98) The related operators exp[vˆ+ ± v ∗ a− ] play also an important role. + (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. Below we will use the operator identity which follows for the choice of v = α in (4.101) 1.104) (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) . without explicitly stating this.90) and zz ∗ = 1 we can write this d du e uzˆ+ a e uz ∗ a− ˆ a − z ∗ a− .

95) one obtains a a a a euˆ φ0 |evˆ φ0 = φ0 |euv evˆ φ0 = euv evˆ φ0 |φ0 = euv − + + + − (4.113) Comparision of all terms on the l.h.98) and again (4.111) This expression.s. is the equivalent of the generating function (??). . (4. and on the r.110) √ y2 u2 + 2uy − 2 2 a = euˆ φ0 (y) .84. Setting y = 0 in (4.h. We start from the generating function (??) and replace according to (4. We want to derive now the values Hn (0). + (4. of (4.112) Expanding both sides of this equation and using (4. we can reproduce by means of the generating function (4.e.h.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.µ=0 uµ v ν √ φµ |φν µ!ν! ∞ = µ=0 uµ v µ µ! (4.116) = δµ ν .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. 4.115) and using (4.4.111) yields π − 4 exp − 1 u2 2 a = euˆ φ0 (0) . Using (4.s.58) the Hermite polynomials Hn (y) by the eigenstates φn (y) √ ∞ 1 ( 2t)n y2 /2 2yt − t2 √ e φn (y) (4.h. provide the same values for Hn (0) as provided in Eq.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. in the ladder operator calculus. 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).109) e = π4 n! n=0 √ Using (4. i. and ˆ l.114) where we have employed (4. the expected from which follows by comparision of each term on both sides φµ |φν orthonormality property. (??). Similarly. For this purpose we consider a a a a euˆ φ0 |evˆ φ0 = φ0 |euˆ evˆ φ0 + + − + (4.111) the orthonormality properties of the wave functions φn (y).s.95). Expanding r..115) a where the latter step follows after expansion of evˆ and using a− φ0 (y) = 0.s.84) yields ∞ ν.

ˆ and the Hamiltonian (4. n = 0. for an oscillator in a stationary state φn (x) as given by (2. using (4.117) 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..121).7) is 1 2 m 2 ω 2 d2 p − .88 Linear Harmonic Oscillator 4.122). n = 0. is given by the ˜ function ψ(p).119) 2m 2 dp2 Accordingly. (−i)n ˜ φn (p) = √ 2n n! m˜ ω π 1 4 exp − m˜ p2 ω 2 Hn ( m˜ ω p) .125) . i. . i.2. 2. .e.124) ). the probability density is P (x) = |φn (x)|2 .122) For a solution of the Schr¨dinger equation in the momentum representation.2). of (4.235). the time-independent Schr¨dinger equation for the oscillator can be written o ˆ H = − mω 2 2 2 x = i ˆ d . the momentum and position operators are as stated in (4. En = ω (n + 1 ) 2 d2 1 + m˜ 2 p2 ω 2 2m dp 2 2 ˜ ˜˜ φE (p) = E φE (p) (4. are ˜ En = or. i. (4...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. The solutions can be stated readily exploiting the earlier results.121) ω (n + ˜ 1 ) 2 . 4. . ω = 1 / m2 ω . For example. As to be expected.123) (4. the eigenvalues are identical to those determined in the position representation. the momentum and position operators are p = p. according to (4.120) Multiplying this equation by 1/m2 ω 2 yields − where ˜ E = E / m2 ω 2 . 2. using (2.118) d2 1 2 + p dp2 2m ˜ ˜ φE (p) = E φE (p) . For this purpose we employ the Schr¨dinger equation in the momentum o representation.37). .6) and the Hamiltonian (4. 1.e. (4. 1.e. (4.3).7) is given by (4. In the position representation the wave function is a function of x. . dp (4. (4. In the momentum representation the wave function is a function of p. (4. The momentum representation eigenfunctions are. 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. is given by a function ψ(x). .235). ˜ (4. The eigenvalues.

126) in terms of dimensionless units following the procedure adopted for the position representation where we employed the substitution (4. 1 ˜ φn (k) = √ 2π +∞ (4.129) m ω one obtains.133) where we note that a change of the normalization of φn (y) [c. This implies the property of Hermite polynomials (−i)n e− k2 2 +∞ Hn (k) = −∞ dy exp(−iky) e− y2 2 Hn (y) (4.30) and (4.58).1: Demonstrate the validity of (4.4.134) Exercise 4.6: Momentum Representation or with (4. Defining k = m˜ / p or ω 1 k = p (4.6.128) We may also express the eigenstates (4. the Jacobian.131) in the present case employing dimensionless units.122) (−i)n ˜ φn (p) = √ 2n n! 1 πm ω 1 4 89 exp − p2 2m ω Hn ( 1 p) .129) follows ky = px/ and.134). 2n n! π (4. ˜ We can now state the probability density Pn (p) for an oscillator with energy En to assume momentum p. m ω (4. of course. instead of (4. m ω (4. From (4.130) dx exp(−ipx/ ) φn (x) −∞ (4.126) a phase factor (−i)n .f. i.30) and (2. φn (k) = We want to finally apply the relationship 1 ˜ φn (p) = √ 2π +∞ 2 (−i)n −k √ e 2 Hn (k) .127) exp − p2 m ω 2 Hn ( 1 p) .e.125.126). hence. α ∈ R.. i. 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. the correctness of the phase factors (−i)n . this allowed us to introduce in (4. 4. . Proceed as follows.126) Normalized eigenfunctions are. defined only up to an arbitrary phase factor eiα .132) dy exp(−iky) φn (y) −∞ (4.30) to obtain (4. (4.235)] absorbs the replacement dx → dy.e. through direct integration..

The quasi-classical states can be obtained by generalizing the construction of the ground state of the oscillator.140) An excellent textbook is Photons and Atoms: Introduction to Quantum Electrodynamics by C.138).Chen-Tannoudji.1 showing in its top row the attributes of such state just discussed. n n=0 d(α) n 2 = 1 (4. carries out a periodic motion described by y(t) = Re y0 eiωt (4.C. 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. New York. y) −∞ (4.4. a useful role in Quantum Electrodynamics since they allow to reproduce with a quantized field as closely as possible the properties of a classical field3 . 2.136) (b) Employ the generating function (2. Such states are referred to as quasi-classical states. 192 3 . The answer is ‘yes’.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. We first construct the solution of (4. and G. 4.196) of Hermite polynomials and equate coefficients of equal powers of z to prove (4.138) yields using a− φn (y) = n φn−1 (y) ˆ ∞ n=0 √ n d(α) φn−1 (y) − α d(α) φn (y) n n = 0 (4. +∞ Linear Harmonic Oscillator dy exp(−iky) g(z. namely through the eigenvalue problem for normalized states a− φ(α) (y) = α φ(α) (y) . for example.139) where we have added the condition that the states be normalized. J. The motion of the probability distribution of such state is presented in Fig.135) g(k. ˆ α ∈ R. a pendulum swinging at small amplitudes. 1989) which discusses quasi-classical states of the free electromagnetic field in Sect. z) = √ 2π where f (k. Inc.134).7 Quasi-Classical States of the Harmonic Oscillator A classical harmonic oscillator. z) = exp 2yz − z 2 − y 2 /2 (4. Inserting this expansion into √ (4..194. except that α changes periodically in time. or Glauber states of the harmonic oscillator and they play. z) = exp −2ikz + z 2 − k 2 /2 . 4.— III. φ(α) | φ(α) = .90 (a) Prove first the property 1 f (k. For this purpose we assume such state exists and expand ∞ ∞ φ(α) (y) = n=0 d(α) φn (y) . pp. We want to construct and characterize such states.Dupont-Roc. coherent states. for example.Grynberg (John Wiley & Sons. (4. The question arises if similar states exist also for the quantum oscillator.138) We will show that a harmonic oscillator prepared in such a state will remain forever in such a state.

n! (4.138). n! (4.148) The identity (4.143) for a constant c which is determined through the normalization condition [see (4. However. (4.109). the quasi-classical states are φ (α) (y) = exp α2 − 2 ∞ n=0 αn √ φn (y) . n! n (4.146) √ 1 − (y − 2 α)2 2 (4. n n+1 (4.84) φ(α) (y) = exp − α2 2 ∞ n=0 αα∗ (αˆ+ ) a a+ φ0 (y) = e− 2 eαˆ φ0 (y) . .. we will see below that α ∈ C are also admissible and will provide a corresponding interpretation.4.e. This interpretation justifies the choice of α ∈ R in (4.7: Quasi-Classical States or ∞ n=0 91 √ n + 1 dn+1 − α d(α) n (α) φn (y) = 0 .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) .139)] 1 = n=0 d(α) n 2 = |c|2 n=0 α2 2 = |c|2 eα .147) allows one to interprete T (α) as an operator √ which shifts the ground state by 2α.149) − α∗ a− ˆ (4. One can also write (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 (4.150) for α presently real.144) Normalization requires c = exp(−α2 /2) and. i. [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) . hence.142) The solution is αn d(α) = √ c n n! ∞ ∞ (4. n! (4. Comparing (4.147) √ which identifies this state as a ground state of the oscillator displaced by 2α.145) using (4.145) Using the generating function in the form (4.149) with (4.

α ∈ C (4. n = 0.160) . (4.153) This is obviously the inverse of T + (α) as defined in (4. . 4. is the adjoint of the operator function on the l.145) shows that α in (4.67.157) (y. it holds T + (α) φ0 |T + (α) φ0 = f |T (α) T + (α) g = φ0 |φ0 . i.s. 1. becomes real.92 Linear Harmonic Oscillator The shift operator as defined in (4.s. Relationship (4.154) .95) to linear combinations αˆ+ + βˆ− and reads then a a f (αˆ+ + βˆ− ) φ|ψ = φ|f (α∗ a− + β ∗ a+ ) ψ .h.156) (4.e.68)] one can write the time-dependent solution with (4.109) serves again to simplify this sum   √  y2 1 1 α2 (u2 + 1)   φ(α) (y. . t1 ) = u exp 1 2 α2 − 2 (α u)n √ φn (y) . t1 ) = exp or φ (α) − α2 2 ∞ exp n=0 −i ω ( n + ∞ n=0 1 ) ( t1 − t0 ) 2 αn √ φn (y) n! (4.145) as the initial state φ(α) (y. Following the procedure adopted in determining the propagator of the harmonic oscillator [see (4. t1 ) = π − 4 u 2 exp  − + 2yαu −  2  2 E (4.150) for complex α obeys the unitary property T + (α) T (α) = 1 1 (4.155) (4. the shift operator leaves the ground state normalized.151) which follows from a generalization of (4. In particular. . (4.152) Since the operator function on the r..h. one can write in case of T + (α) T + (α) + = T (α) = eα ∗ a− ˆ − αˆ+ a a = e − (αˆ + − α∗ a− ) ˆ . a a ˆ ˆ (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 (4.145) is replaced by a complex number αu which at times t1 = t0 + 2πn/ω.152) T + (α) f |T + (α) g = f |T (α) T + (α) g = f |g .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. n! u = e−iω(t1 − t0 ) .

138): a real α corresponds to a displacement such that initially the oscillator is at rest. In Fig. Comparing (4.147).138. The time-dependent wave function (4. in general.1 (top row) we present the probability distribution of the Glauber state for various instances in time. 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. 4.161) are time-dependent √ y0 (t1 ) = 2 α cosω(t1 − t0 ) √ k0 (t1 ) = − 2 α sinω(t1 − t0 ) 1 φ0 (t1 ) = − α2 sin2ω(t1 − t0 ) (4. a complex α corresponds to a displaced oscillator which has initially a non-vanishing velocity.4. t1 ) = π − 4 exp − [y − y0 (t1 )]2 + iy k0 (t1 ) − iφ0 (t1 ) − ω(t1 − t0 ) 2 2 .159) is then 1 1 i φ(α) (y. being displaced around the center with period 2π/ω and.163) This solution corresponds to the initial state given by (4. t1 ) = u 2 T + (α u) φ0 (y) (4. The diagram illustrates that the wave function retains its Gaussian shape with constant width for all times.164) where according to (4. (4.163) also through the shift operator. except that it is not pointlike.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 (4.161) and is found to describe a displaced Gaussian. obviously. This shows clearly that the Glauber state is a close analogue to the classical oscillator. a momentum factor and a phase factor.150) with (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. experiences a change of its width (relative to .165) .148. 4. Arbitrary Gaussian Wave Packet Moving in a Harmonic Potential We want to demonstrate now that any initial state described by a Gaussian shows a time dependence very similar to that of the Glauber states in that such state remains Gaussian. moving solely its center of mass in an oscillator fashion around the minimum of the harmonic potential. One can express (4.150) 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. this characterization corresponds closely to that of the possible initial states of a classical oscillator.162) 2 with a periodic change.158) allows one to write 1 φ(α) (y. 4. Our interpretation 2 explains also the meaning of a complex α in (4.

166) φ(ao .167) One can derive for the expansion coefficients +∞ dn = −∞ dy φn (y) φ(ao . 4. 2σ 2 q = a .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. 1 + σ2 (4.168) dy e−p(y−q) Hn (y) where p = 1 + σ2 . σ|y. This behaviour is illustrated in Fig. 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. To describe such state which satisfies the initial condition φ(ao .1 (second and third row). t0 ) = [πσ 2 ]− 4 exp we expand ∞ 1 − (y − ao )2 2σ 2     (4. (4. 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. that of the Glauber states) with a period π/ω. σ|y.1: Time dependence of Gaussian wave packets in harmonic oscillator potential. t0 ) = n=0 dn φn (y) . σ|y.169) .

7: Quasi-Classical States The solution of integral (4. in agreement with the expected result (4. σ|y. p−1 1 − σ4 t = 1 − σ 2 −iω(t1 −to ) e 1 + σ2 (4. i.76) permits us to write this φ(ao .173) which is. this 3 volume integral table is likely the most complete today.177) − 1 ω(t1 − to ) 2 .157) as follows φ(ao . Prudnikov. σ|y.e. t1 ) = √ × where yo = q p a = √ . New York. Following the strategy adopted previously one can express the time-dependent state corresponding to the initial condition (4. Brychkov. Yu. 2 by A. a worthy successor of the famous Gradshteyn. 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.143) of the Glauber state. 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. namely $750 4 .171) lim p−1 p n 2 Hn (q p ) = lim (2q)n = an . t1 ) = e−aq/2 √ pσ ∞ n=0 e −iω(n+ 1 )(t1 −to ) √ 2 1 2n n! p−1 p n/2 (4. 1 (4.174) Hn (q p p−1 ) φn (y) . therefore.172) One can then conclude e−a /4 n lim dn = √ a σ→1 2n n! (4. σ→1 p−1 2 (4. vol. Marichev (Wiley. σ→1 O(y n−2 ) for n ≥ 2 1 for n = 0.4. This expansion can be written φ(ao . In this limit the coefficients dn should√ become identical to the coefficients (4. 1990). p−1 95 (4. in fact. unfortunately very expensive.143).175) ∞ tn n=0 2n n! Hn (yo ) e−yo /2 Hn (y) e−y 2 2 /2 The generating function (4.P.I.166) in analogy to (4. t1 ) = √ 2 × exp − 1 (y 2 + yo ) 1+t2 − 2 1−t 2 1 1 √ √ pσ π 1 − t2 × t 1 − t2 (4. and O. σ|y.168) is4 e−aq/2 (p − 1)n/2 dn = √ Hn (q 2n n!σ p(n+1)/2 p )..170) In order to check the resulting coefficients we consider the limit σ → 1.176) 1 √ pσ π a 2 exp − 1 1+σ2 + 1 yo − 1 ω(t1 − to ) 2 2 2 2 × (4.

96 Linear Harmonic Oscillator .

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

(5. ϕ)T ) = ±  sinϕ cosϕ 0  (5. a minus sign in Eq. then the fingers of your fist point in the direction of the rotation.3) Since this holds for any a and b.k.e. therefore. (5. (5. in Cartesian coordinates r = (x1 . x3 )T . It holds then j=1 3 3 a ·b = j. Let us define the rotated vector as r = R(ϑ) r . i.6) follows immediatety for the determinant of R(ϑ) using detA B = detA detB and detAT = detA detR(ϑ) = ±1 .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 . ϕ)T ) is negative. the matrix represents a proper rotation. i.e. .7) Let us consider briefly an example to illustrate rotational transformations and to interpret the sign of detR(ϑ).. 0.2) Rotations conserve the scalar product between any pair of vectors a and b. From (5. (5. if the thumb in your right hand fist points in the direction of ϑ.1) In Cartesian coordinates this reads 3 xk = j=1 [R(ϑ)]kj xj k = 1.8) 0 0 1 In case of a prefactor +1.. in case of a prefactor −1 a rotation and an inversion at the origin.7) implies that a rotation is associated with an inversion. ϑ parametrizes the rotations uniquely as long as |ϑ| < 2π. it follows 3 [R(ϑ)]jk [R(ϑ)]j = δk j=1 .(5.98 Properties of Rotations in R Theory of Angular Momentum and Spin Rotational transformations of vectors r ∈ R . 3 .5) This equation states the key characteristic of rotation matrices. In the latter case the determinant of R(ϑ = (0. We assume the convention that rotations are right-handed. namely around an axis given by the direction of ϑ and by an angle |ϑ|. 0. they conserve a · b = 3 aj bj . (5. i. are linear and. =1 [R(ϑ)]jk [R(ϑ)]j ak b = j=1 aj bj . 2. can be represented by 3 × 3 matrices R(ϑ). x2 . where ϑ denotes the rotation. (5. A rotation around the x3 -axis by an angle ϕ is described by the matrix   cosϕ − sinϕ 0 R(ϑ = (0.e. In the following we want to exclude inversions and consider only rotation matrices with detR = 1.6) (5.

It follows that R3 is also an element of SO(3). there exists a real 3 × 3–matrix −1 R1 which is the inverse of R1 . We have shown altogether that the elements of SO(3) form a group. (ii) The group element ‘e’ is played by the 3 × 3 identity matrix 1 1. It holds T T T T R3 R3 = ( R1 R2 )T R1 R2 = R2 R1 R1 R2 = R2 R2 = 1 . hence. it holds 1. Since (R1 )T = (R1 )−1 it follows −1 −1 −1 T (R1 )T R1 = (R1 )−1 R1 = −1 which implies R1 ∈ SO(3).12) (iv) Since the associative law holds for multiplication of any square matrices this property holds for the elements of SO(3). R2 ∈ SO(3) and do not write out “◦” explicitly since it represents the matrix product. in fact. R RT = RT R = 1 detR = 1 } 1.1: Matrix Representation of SO(3) Definition of the Group SO(3) We will consider then the following set SO(3) = { real 3 × 3 matrices R.9) This set of matrices is closed under matrix multiplication and.5. T R1 R1 −1 = 1 −1 = 1 1 1 (5.9) forms a group.13) . We need to demonstrate that this inverse belongs also to −1 T SO(3). We would like to point out the obvious property that for all elements R of SO(3) holds R−1 = RT (5. For this purpose we must demonstrate that the group properties (i–iv) hold. 3 × 3–matrix. b ∈ G the product c = a ◦ b is also in G. forms a group G satisfying the axioms (i) For any pair a.10) detR3 = det ( R1 R2 ) = detR1 detR2 = 1 . one can show R3 R3 = 1 Furthermore. In the following we will assume R1 . 1 T Similarly. (5. (i) Obviously. (ii) There exists an element e in G with the property ∀a ∈ G → e ◦ a = a ◦ e = a. a ∈ G → ∃a−1 ∈ G with a ◦ a−1 = a−1 ◦ a = e. (5. We want to prove now that SO(3) as defined in (5. (iv) For products of three elements holds the associative law a ◦ (b ◦ c) = (a ◦ b) ◦ c. 99 (5.11) (iii) From detR1 = 0 follows that R1 is non-singular and. (iii) ∀a. R3 = R1 R2 is a real.

16) must be antisymmetric. we arrive at the property exp ϑk LT k from which follows LT = −Lk . mappings between the sets which conserve the group properties: (a) the set of real and imaginary numbers {1. Lie Algebra We want to consider now a most convenient. compact representation of R(ϑ) which will be of general use and. presently.100 Theory of Angular Momentum and Spin Exercise 5.13) R(ϑk ek )−1 = R(ϑk ek )T = [exp ( ϑk Lk )]T = exp ϑk LT ˆ ˆ k and.18) (5.e. 0 −1 1 0 (5. (b) the set of matrices {M1 . we want to assume that they exist.19) (5.17) . rotation by 270o } and consecutive execution of rotation as the binary operation ‘◦’. 0 1 −1 0 . x2 –plane {rotation by Oo . (c) the set of four rotations in the x1 . M3 . Generators. rotation by 180o . M2 . −i} together with multiplication as the binary operation ‘◦’.1: Test if the following sets together with the stated binary operation ‘◦’ form groups. i.20) (5. This representation expresses rotation matrices in terms of three 3 × 3– matrices J1 .1. establish if 1-1 homomorphic mappings exist. We want to show that for the property R RT = 1 to hold. i. −1 0 0 −1 . J2 . M4 } = 1 0 0 1 . 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. J3 as follows i R(ϑ) = exp − ϑ · J (5. k = exp ( −ϑk Lk ) (5. Lie Group. the matrices i Lk = − Jk (5.14) together with matrix multiplication as the binary operation ‘◦’. −1. in fact. identify subgroups. 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 90o . due to the uniqueness of the inverse.15) Below we will construct appropriate matrices Jk . will play a central role in the representation of many other symmetry transformations in Physics.

0)T ) − 1 1 1 ϑ1 →0 (5. Exercise 5. 0. This property gives A then only three independent matrix elements. A must have the form   0 a b A =  −a 0 c  . ϑ2 .. 0. the real matrix A is anti-symmetric.5. 0)T the rotation is the identity.e.15) with three real parameters ϑ1 . Let us use this definition to evaluate L3 .2: Determine the generator L1 according to Eq. 0. (5.22) and similar for L2 and L3 . L2 .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). 0 0 0 0 0 0  (5. 0.26) (5.        (5. ϑ3 )T ] = lim  sinϑ3 cosϑ3 0  =  ϑ3 ϑ3 →0 ϑ3 →0 0 0 1 0 0 1 Insertion into (5. L3 which are independent of the rotation angles.15) as the derivatives of R(ϑ) with respect to the Cartesian components ϑk taken at ϑ = (0.21) −b −c 0 It is this the reason why one can expect that three-dimensional rotation matrices can be expressed through (5.25)     0 −ϑ3 0 0 −1 0  ϑ3 0 0  =  1 0 0  .1. i. lim R[(0. 0)T . Using (5. We assume now that we parameterize rotations through a three-dimensional rotation vector ϑ = (ϑ1 .8) we can state    cosϑ3 −sinϑ3 0 1 −ϑ3 0 1 0  . ϑ3 and three matrices L1 .1: Matrix Representation of SO(3) 101 We can conclude then that if elements of SO(3) can be expressed as R = eA .23) (5. ϑ3 )T such that for ϑ = (0.22).27) . One can determine then the matrices Lk defined in (5.24) (5. ϑ2 .(5. Using the definition of partial derivatives we can state L1 = lim ϑ−1 R(ϑ = (ϑ1 .

. = 2.15) and their generators obey the property (5. m = 3 . The expression for Zn are actually rather complicated. the elements of if any two indices are identical for k = 1. (5. B]. (5. m = 3 for any odd permutation of k = 1. three being a special case. In fact.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. 3 )T ) = exp( 3 L3 ). ϕ)T . 0.29) where Z1 = A + B and where the remaining operators Zn are commutators of A and B. From this result one can conclude that expressions of the type (5. B] ).30) is called the Lie algebra.. that the operators (5. Lie groups can have any number of generators. [A. L ] = m=1 fk m Lm . if the resulting products of exponential operators can be expressed in the form (5. = 2.30) are called Lie groups.g.e. In case of the group SO(3) the structure constants are particularly simple. 1 1 e. the Lie algebra of SO(3) can be written [Lk ..e.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. commutators of commutators of A and B. as we see later. 2 . Z2 = 2 [A.15).28) The construction of the generators through the limit taken in Eq. B]] + [[A.15) obey the group property. (5.30) Groups with the property that their elements can be expressed like (5. in fact.32) For the matrices Jk = i Lk which. that the generators represent infinitesimal rotations with 1 >> |ϑ| around the x1 -. the matrix (5. B]. The latter property implies that for any ϑ1 and ϑ2 there exists a ϑ3 such that exp i − ϑ1 · J exp i − ϑ2 · J = exp i − ϑ3 · J (5. 0. x2 -.8) in case of ϑ = (0. e. Of course. R(ϑ = (0. however. 0)T ) = exp( 2 L2 ) and R(ϑ = (0. holds the algebra [Jk . only the latter example and . = 2. and x3 -axes. J ] = i k m Jm .31) where k m are the elements which are   0   1 k m =  1   −1 of the totally antisymmetric 3-dimensional tensor. the property (5. etc. m = 3 for any even permutation of k = 1. Z3 = 12 ( [A.g. Finite rotations can be obtained by applications of many infinitesimal rotations of the type R(ϑ = ( 1 . The question is then. i. L ] = k m Lm (5.22) implies. 0)T ) = exp( 1 L1 ). i.33) We want to demonstrate now that exp(ϑ · L) yields the known rotational transformations. and the constants fk m are called the structure constants. 0. An answer can be found considering the Baker-Campbell-Hausdorf expansion ∞ exp( λA ) exp( λB ) = exp( n=1 λn Zn ) (5. We want to consider. it must hold 3 [Lk . (5. however. are related to angular momentum operators.

40) . that the matrix ϕL3 .8). using (5. 0 −1 1 0 = (−1)n 0 −1 1 0 (5. ∞ 2n = − 1 0 0 1 2n+1 (5.  cos ϕ − sin ϕ 0 =  sin ϕ cos ϕ 0  0 0 1  (5.37) = (−1)n 1 0 0 1 .39) Recognizing the Taylor expansions of cos ϕ and sin ϕ one obtains eA = and. 5. (5. can be written ϕL3 = 0 103 A 0 0 0 0 . accordingly.36) The property 0 −1 1 0 allows one to write 0 −1 1 0 and. (5.41) cos ϕ − sin ϕ sin ϕ cos ϕ (5. (2n + 1)! (5.38) eA = n=0 (−1)n 2n ϕ (2n)! 1 0 0 1 ∞ + n=0 (−1)n ϕ2n+1 (2n + 1)! 0 −1 1 0 .35).31).27).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 . A = ϕ 0 −1 1 0 .3: Show that the generators Lk of SO(3) obey the commutation relationship (5. eϕL3 which agrees with (5. We note.34. according to (5. (5.1.5.35) To determine expA = ∞ ν ν=1 A /ν! we split its Taylor expansion into even and odd powers 1 A2n + (2n)! 2 ∞ n=0 ∞ eA = n=0 1 A2n+1 .1: Matrix Representation of SO(3) determine exp(ϕL3 ). Exercise 5.

2.e. Since this holds for any ψ(r ) one can conclude ρ R(ϑ1 ) ρ R(ϑ2 ) group property (5. its elements can be represented through an exponential R(ϑ) = exp i − ϑ·J (5. = ψ([ R(ϑ1 ) R(ϑ2 ) ]−1 r) = ρ R(ϑ2 ) ψ([R(ϑ1 )]−1 r) (5.e. (5.42) Obviously. we define rotations R(ϑ) as linear maps C∞ ( ) → C∞ ( ). namely R(ϑ)ψ(r ) = ψ(R−1 (ϑ) r ) . Exercise 5.2 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 fact. the transformation R(ϑ): C∞ ( ) → C∞ ( ) is related to the transformation R(ϑ): R → R . (5.46) (5. the R(ϑ) ψ(r) = ψ(R(ϑ) r ). (5. For this purpose we require stringent continuity properties of the wavefunction: the wavefunctions under consideration must be elements of the set C∞ ( ). i.45) (5. we will learn that transformation patterns and invariances are connected with the angular momentum states of quantum mechanics. for A.104 Theory of Angular Momentum and Spin 5. complex-valued functions over the 3-dimensional space R which can be differentiated infinitely often. B ∈ SO(3) holds ρ(A B) = ρ(A) ρ(B) . In particular. Show that in this case holds Generators One can assume then that R(ϑ) should also form a Lie group.48) .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) . a relationship which we like to express as follows R(ϑ) = ρ R(ϑ) . hence..43) ρ( ) conserves the group property of SO(3).e. Definition We first define how a rotational transformation acts on a wavefunction ψ(r ).47) = ρ R(ϑ1 ) R(ϑ2 ) . i. one isomorphic to SO(3) and. In analogy to R(ϑ) being a linear map R → R .44) holds.1: Assume the definition ρ ρ (A B) = ρ (B) ρ (A). i.

.49).55) and f (r) = f [J1 . 0. (5.53–5.e.49) (note the minus sign of −ϑ1 which originates from the inverse of the rotation in Eq.2: Function Space Representation of SO(3) where the generators can be determined in analogy to Eq. 0.54) (5.24) one can expand  ϑ3 →0 ϑ3 →0 lim ϑ−1 R(ϑ = (0. J2 ] f (r) = − = − 2[ 2 [− i J . −ϑ3 ) r ) − f (r) ) . in case of J3 . (5. . (5.  0 0 1 x3 x3 (5. 0.56) holds for [J2 .57) Exercise 5.2: (a) Derive the generators Jk . show that (5. Carrying out similar calculations for J1 and J2 one can derive i − J1 i − J2 i − J3 = = = x3 ∂2 − x2 ∂3 x1 ∂3 − x3 ∂1 x2 ∂1 − x1 ∂2 . In fact.53) (5.2.42)) and similarly for J2 and J3 . but also the Lie algebra of the generators. (5. J1 ]. 3     1 ϑ3 0 x1 x1 + ϑ3 x2 R(0.50) lim ϑ−1 ( f ( R(0.50) and Taylor expansion yields i − J3 f (r) = = ϑ3 →0 lim ϑ−1 ( f (r) + ϑ3 x2 ∂1 f (r) − ϑ3 x1 ∂2 f (r) − f (r) ) 3 (5.52) ( x2 ∂1 − x1 ∂2 ) f (r) . 5. i. One obtains using (5. ϑ3 )T ) f (r) − f (r) 3 (5. 0. − i J ] f (r) 1 2 =− 2× [ (x3 ∂2 − x2 ∂3 ) (x1 ∂3 f − x3 ∂1 f ) − (x1 ∂3 − x3 ∂1 ) (x3 ∂2 f − x2 ∂3 f ) ] 2 +x1 x3 ∂2 ∂3 f − x1 x2 ∂3 f − x2 ∂1 ∂2 f + x2 x3 ∂1 ∂3 f + x2 ∂1 f 3 2(x ∂ 2 1 2 −x1 ∂2 f − x1 x3 ∂2 ∂3 f + x2 ∂1 ∂2 f + x1 x2 ∂3 f − x2 x3 ∂1 ∂3 f ] 3 = − − x1 ∂2 ) f (r) = i J3 f (r) . 0)T ) − 1 1 ϑ1 →0 105 (5.23. J ] = i k m Jm . k = 1.55) Not only is the group property of SO(3) conserved by ρ( ).22) according to i − J1 = lim ϑ−1 R(ϑ = (−ϑ1 . 2 by means of limits as suggested in Eq. (5. −ϑ3 ) r ≈  −ϑ3 1 0   x2  =  −ϑ3 x1 + x2 . J2 ].49) to a function f (r). J3 ] and [J3 . i − J3 f (r) = = Using (5.5.51) Inserting this into (5.(5. it holds [Jk .56) We like to verify this property for [J1 . The generators Jk can be determined by applying (5.

These eigenstates are the well-known spherical harmonics Y m (θ.3 Angular Momentum Operators The generators Jk can be readily recognized as the angular momentum operators of quantum mechanics. i. C] + [A..65) . To show this we first note that (5.106 Theory of Angular Momentum and Spin Exercise 5.56) are the famous commutation relationships of the quantium mechanical angular momentum operators.58) i = r ×p ˆ (5. according to the correspondence principle between classical and quantum mechanics. θ. [J 2 .56) one obtains [J 2 .3: Consider a wave function ψ(φ) which depends only on the azimutal angle φ of the spherical coordinate system (r. Show that for J3 as defined above holds exp( iα J3 ) ψ(φ) = ψ(φ + α).63) (5. We will show now that the properties of spherical harmonics J2Y J3 Y m (θ. and x3 = r cosθ. The commutation relationships imply that no simultaneous eigenstates of all three Jk exist. 5.64) mY m (θ. φ) m (θ.2. C] = A[B. can be found. i. x2 = r sinθ sinφ. 2. k = 1. φ) (5. J3 ] + [J2 .e. Using [AB. φ).61) We demonstrate this property for k = 3. C]B and (5. 3. φ). Jk ] = 0 . Equation (5. .55) can be written as a vector product of r and i − J = −r × From this follows. using the momentum operator p = ˆ J = r × i . 3 . usually chosen as J3 . J3 ] J1 + J2 [J2 .60) commutes with all three Jk . (5. J3 ] J2 −i J1 J2 − i J2 J1 + i J2 J1 + i J1 J2 = 0 According to (5.62) J1 [J1 . However. J3 ] + [J1 . the operator 2 2 2 J 2 = J1 + J2 + J3 (5. J3 ] (5. k = 1. J3 ] = = = 2 2 [J1 + J2 .e. (5. φ) as well as the effect of the so-called raising and lowering operators J± = J1 ± i J2 (5. identifies J as the quantum mechanical angular momentum operator. the system related to the Cartesian coordinates through x1 = r sinθ cosφ.59) which.53–5. 2.61) simultaneous eigenstates of J 2 and of one of the Jk . φ) = = 2 ( + 1) Y m (θ.

74) (5. The definitions of the coefficients α −1 and β −1 yield L+ L− | = L+ ( iβ −1 | − 1 = − α −1 β −1 | .3: Angular Momentum Operators on the spherical harmonics.67) (5. 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. 2. Using [L3 . (∗) (5. i.68) (5. L ] = k m Lm and let L± = L1 ± i L2 . in fact.70) (5. The arguments are very similar to the ones used above and we can be brief. For this purpose we prove the following theorem. φ) (θ. L− ] + L− L+ )| m = ( −2iL3 + .5. It holds L+ L− | = (L+ L− − L− L+ + L− L+ )| = ([L+ . L− ] = [L3 . We note L3 L− | = ( [L3 .e. Noting [L+ .56).1 Let Lk . and since we have already shown | − 1 = (iβ −1 )−1 L− | we can state L3 | − 1 = −i( − 1)| − 1 . We consider L+ L− | m = ( [L+ .75) (5. L1 − iL2 ] = L2 + iL1 = i (L1 − iL2 ) = iL− and L3 | = −i | we obtain L3 L− | = ( iL− + L− L3 )| = i(− + 1) L− | . For this purpose we first demonstrate that the property holds for m = − 1 by considering (i) for the state L− | ∼ | − 1 .71) To prove this theorem we first show by induction that (i) holds. φ) (5. L+ raises the m-value of | − 1 from m = − 1 to m = .76) We now show that property (ii) also holds for m = − 1 proceeding in a similar way. We will show that this property holds then also for m − 1. φ) m+1 (θ. 3 be operators acting on a Hilbert space H which obey the algebra [Lk . φ) 107 = = 0 = = 0 ( + m + 1)( − m) Y ( + m + 1)( − m) Y m+1 (θ. L2 = L2 + L2 + L2 .. L1 − iL2 ] = −2iL3 and L+ | = 0 one obtains L+ L− | = −2iL3 | = −2 | . φ) − m (θ. An Important Theorem Theorem 1. J+ Y J− Y J− Y m (θ. L− ] = [L1 + iL2 .72) = 0 = −i | (5. α −1 β −1 = 2 . φ) are a consequence of the Lie algebra (5.69) J+ Y (θ. k = 1. namely. Continuing our proof through induction we assume now that property (i) holds for m. From this follows that L− | is an eigenstate of L3 . L− ] + L− L3 ) | . L− ] + L− L+ )| .73) (5.66) (5.e. i.

it also holds for m − 1. A first application will involve the construction of the eigenstates of the angular momentum operators as defined in (5. i.80) (5.81) i. in terms of spherical coordinates defined through x1 x2 x3 = = = r sin θ cos φ r sin θ sin φ r cos θ (5..79) read L+ | m L− | m + 1 = −i = −i ( + m + 1)( − m) | m + 1 ( + m + 1)( − m) | m .55).108 Theory of Angular Momentum and Spin L− L+ )| m = (−2m − αm βm )| m . Angular Momentum Operators in Spherical Coordinates We want to express now the generators Jk .66. it holds αm βm = 2 one obtains αm βm = = ( + 1) k=m+1 k . We will have various opportunities to employ the Theorem just derived. (∗∗) (5. 5. (5. Again the argumemts are similar to the ones used above. from L+ L− | m = − αm−1 βm−1 | m follows the recursion relationship αm−1 βm−1 = 2m + αm βm . finally αm = βm = (5.63. We note that we can write L2 = 1 1 1 1 2 2 2 2 L+ L− + 2 L− L+ + L3 .84) . This implies that (i) holds for m − 1. 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 . This completes the proof of the theorem. We can deduce from (∗) and (∗∗) α α α −1 β −1 −2 β −2 −3 β −3 = 2 = 2( − 1 + ) = 2( − 2 + · · −1 + ) Obviously. yield properties (5. 5.e. θ.e.69). One can normalize the states | m such that αm = βm ..77) We will show now that if (ii) holds for m. 5. Using the familiar formula + m − n k=0 k = n (n + 1)/2 − (m + 1) m = ( + 1) m − (m + 1) m (5.82) (5.64). We note L3 L− | m = ( [L3 .53–5. 5. L− ] + L− L3 )| m = ( iL− − i m L− )| m = −i (m − 1) L− | m from which we can deduce L3 | m − 1 = −i(m − 1)| m − 1 . (5.79) We can now show that (iii) holds for all proper m–values.70. Before we can finally show that property (iii) holds as well for all − ≤ m ≤ we need to determine the coefficients αm and βm . For this purpose we need to express the angular momentum operators in terms of spherical coordinates (r.78) ( + m + 1)( − m) ( + m + 1)( − m) .72. in particular.83) (5. given in (5. 5.73. φ. 2 We note that Eqs.

using ∂/∂tanφ = cos2 φ ∂/∂φ and ∂/∂cosθ = −(1/sinθ) ∂/∂θ. φ) = (x3 ∂2 − x2 ∂3 )f (θ.85). (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 . According to (5. using (5. (L2 )2 = − cos φ ∂ ∂ + cot θ sin φ ∂θ ∂φ 2 .93) and similarly.82–5. φ) ∂x2 /x1 ∂ ∂x3 /r ∂ = x3 + x3 ∂x2 ∂ tan φ ∂x2 ∂ cos θ ∂x2 /x1 /r ∂ ∂ − x2 − x2 ∂x3 3 ∂ cos θ f (θ.87) To derive these properties we consider.87) holds J3 = To determine J 2 we note. using (5.86). cos θ ∂ cos2 φ + sin θ cos φ ∂φ cos2 θ sin θ sin φ ∂ sin θ sin φ ∂ + + sin2 θ f (θ. we will prove − − − i i i J1 J2 J3 = L1 = L2 = L3 ∂ ∂ + cot θ cos φ ∂θ ∂φ ∂ ∂ = − cos φ + cot θ sin φ ∂θ ∂φ ∂ = − . L1 f (θ. φ).60). for example. L1 f (θ. φ) ∂x ∂x3 ∂ tan φ =0 (5. φ) (5. In fact. applying repeatedly the chain rule.91) which agrees with expression (5.85) (5.86) (5. We like to express now the operators J3 and J 2 .85). in terms of spherical coordinates. (5.90) This yields. φ) = (5.88) (5. φ) = ( x3 ∂2 − x2 ∂3 ) f (θ.89) = ∂ x2 x2 ∂ x2 + x2 ∂ x3 − 33 − x2 1 3 2 x1 ∂ tan φ r ∂ cos θ r ∂ cos θ f (θ. ∂φ = sin φ (5.84 ) one obtains. φ) sin θ ∂θ sin θ ∂θ L1 f (θ. the latter defined in (5.5. Using (5. actually. involve only the angular variables θ and φ and not the radius r.3: Angular Momentum Operators 109 We first like to demonstrate that the generators Jk .92) (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 (5.85.97) Kinetic Energy Operator The kinetic energy of a classical particle can be expressed J2 p2 p2 = r + class 2m 2m 2m r2 (5.96) and the relationship between Lk and Jk as given in (5.85.95) sinθ ∂ ∂θ = ∂2 ∂ + cotθ . 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 + . C∞ (S ).86. 2m r ∂r2 2m r2 2 (5.98) where Jclass = r × mv is the angular momentum and pr = mr is the radial momentum. (5. One first identifies the Hilbert space H on which the generators Lk operate. 5. 5. 2 ∂θ ∂θ (5.100) (5.97). 1 sin2 θ sinθ ∂ ∂θ (5. 5. The functions in C∞ (S ) have real variables .94) It follows (L1 )2 + (L2 )2 + (L3 )2 = With cot2 θ + 1 = 1/sin2 θ. namely on the space of complex-valued functions on the unit 3-dim.86.87). ∂θ2 ∂θ ∂φ (5.87). sphere S2 . The operators in the present case are Lk = − i Jk where the Jk are differential operators in θ and φ given in (5. These generators act on a subspace of C∞ ( ). ∂θ ∂φ ∂2 ∂ ∂2 + cotθ + ( cot2 θ + 1 ) 2 . 5.63.4 Angular Momentum Eigenstates According to the theorem on page 107 above the eigenfunctions (5. 5.110 = cos2 φ Theory of Angular Momentum and Spin ∂2 1 ∂ ∂2 + sinφ cosφ − 2 cotθ sinφ cosφ ∂θ2 ∂φ ∂θ∂φ sin2 θ 2 ∂ ∂ + cotθ sin2 φ + cot2 θ sin2 φ 2 . i. 5.e. one can conclude J2 = − 2 sin θ 2 sinθ ∂ ∂θ 2 + ∂2 ∂φ2 .64) can be constructed as follows.101) 1 ∂2 r r ∂r2 and comparision with (5.

63.106) The latter property is obviously satisfied for the chosen φ-dependence.4: Angular Momentum Eigenstates 111 θ. (5.102) where dΩ = sinθ dθ dφ is the volume element of S2 .85–5. − 2. φ) (5. − . φ) = −i ( + m + 1)( − m)Y m (θ.105) explicitly. (5. . One obtains in this way Y m (θ. is indeed a Hilbert space. φ) = 0 as well as L3 Y (θ.107) (5. (5.105) ∆( . (5. the so-called spherical harmonics. m = − . .107) reads ∂ ∂ + i cotθ ∂θ ∂φ P (cosθ) ei φ i e±iφ ∂ ∂ ± i cotθ ∂θ ∂φ . φ). } applying L− Y m+1 (θ. by carrying out the construction according to (5. The function space endowed with this norm and including only functions for which the integral (5. φ). φ) (5. 0 ≤ φ < 2π. must be cyclic in φ with period 2π.64) can be constructed then by seeking first functions Y (θ. and to be admissable for description of quantum states. φ) L3 Y (θ. φ) = ∆( .110) = 0 (5. φ) .104. φ) (5. m) Constructing Y → Y →→ ··· Y ( + m)! (2 )! ( − m)! −1 − We like to characterize the resulting eigenfunctions Y m (θ. − + 1. φ) = = 0 −i Y (θ.104) (5. φ) in H which satisfy L+ Y (θ. normalizing these functions. . .111) .5. (5. 5.87). can be expressed L± = Accordingly. φ)f (θ. . and then determining the family {Y m (θ. The raising and lowering operators L± = L1 ± i L2 .103) iteratively for m = − 1. The eigenfunctions (5. φ) = 2 + 1 ( − m)! P 4π ( + m)! m (cosθ) eimφ . . . 5. .109) using (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.108) m (θ. m) = 1 −m J− 1 2 Y (θ. 0 ≤ θ < π. The norm in H is defined through |f |2 = S2 dΩ f ∗ (θ.102) exists. φ. φ) = −i Y (θ. φ).

φ) = (−1) 1 1 = 1 [1 · 3 · 5 · · · (2 − 1)]2 (5. π +1 dθ sin2 0 +1 θ = −1 +1 −1 dx (1 − x2 ) +1 = = . 2nd Ed.113) 0 is finite. using the new integration variable x = cos θ. . 5. The normalization factor N is chosen such that π (5.114..119) where the factor (−1) has been included to agree with convention1 .116) implies |N |2 from which we conclude N = (−1) 1 · 3 · 5 · · · (2 − 1) (5.120) See.106. i.106.” by J. (5. .112) where we employed the definition of P (cosθ) in (5. We note that (5. (5. (5. (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. is P (cos θ) = N sin θ as one can readily verify. for example.107). ”Classical Electrodynamics.116) The integral appearing in the last expression can be evaluated by repeated integration by parts.119) provide the following expression for Y Y (θ.e.117) Accordingly. New York. One obtains. 5. Jackson (John Wiley.D.114) dθ sinθ |Y (cosθ)|2 = 1 0 (5. 1975) .112 or Theory of Angular Momentum and Spin ∂ − ∂θ cotθ P (cosθ) = 0 (5.115) holds. one for which π dθ sinθ |P (cosθ)|2 (5.114) yield |N |2 2 +1 1 2π 4π (2 )! π dθ sin2 0 +1 θ = 1.118) 2 +1 1 1 sin θ ei 4π (2 )! 2 ! φ . A proper solution of this equation. 5.

m = − 1. To prove (5. as specified in (5. m (cos θ) e (5.126) Then holds.124) reads P (x) = = (2 )! (1 − x2 )− 2 (x2 − 1) 2 ! (−1) 1 · 3 · 5 · · · (2 − 1) sin θ (5. are real.5.124) called the associated Legendre polynomials. Actually.103.124) holds for m + 1. Let us then assume that (5.4: Angular Momentum Eigenstates 113 We have constructed a normalized solution of (5.123) P m+1 (x) ∂ x + (m + 1) 1 ∂x (1 − x2 ) 2 . we seek to determine the functions P m (cos θ) and. every application of L− reduces the power of eiφ by one.114. employing again the variable x = cos θ. ∂x −m−1 .63. i.e. 5. 5. indeed.123). . use (5.125) which agrees with (5. . .106).127) ∂ −m−1 (x2 − 1) . 2 ! ( − m − 1)! ∂x −m−1 (5. (5. − 2..64). the eigenfunctions Y m (θ.110) together with (5. therefore.124). We want to demonstrate now that the recursion equation (5.119). φ).123) leads to the expression P m (x) = −m m ∂ 1 ( + m)! (1 − x2 )− 2 (x2 − 1) 2 ! ( − m)! ∂x −m (5.121) states then for the functions P m (cos θ) ∂ + (m + 1) cotθ ∂θ P m+1 (cos θ) = − ( + m + 1) ( − m) P m (cos θ) . 5. and can obtain now.103. 5. namely Y (θ.124) we proceed by induction. φ). P m+1 (x) = −m−1 m+1 ∂ 1 ( + m + 1)! (1 − x2 )− 2 (x2 − 1) . − defined in (5. P m (x) = −1 ( + m)! 2 ! ( − m)! × (1 − x2 )− m+1 2 −(1 − x2 ) 2 1 ∂ x + (m + 1) 1 ∂x (1 − x2 ) 2 × (5. through repeated application of (5. The reader should note that the associated Legendre polynomials. P m (x) = −1 × ( + m + 1) ( − m) × −(1 − x2 ) 2 1 (5. according to (5.121) This expressions shows that the factor eimφ .122) The latter identity can be written. describes the φ-dependence of Y m (θ. For m = (5.110).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 . φ). (5.

therefore. we embark on this construction in order to prove (5.134) (x) = 1 2 ! (1 − x2 ) 2 . We first determine Y − (θ. therefore. P − (x) = 1 ∂2 (1 − x2 ) 2 (x2 − 1) . φ) = Y − (θ. (5.106.106.133) Since the term with the highest power of (x2 − 1) is x2 . therefore.129) holds. φ) . expression (5. Since (5.127) reproduces (5. φ) = (−1)m Y m −m (θ. Y ∗ (θ. if L− Y (θ. In fact. φ) = = determines first a normalized solution 0 i f (θ. from such construction emerges an important symmetry property of Y m (θ. that expression (5. φ). etc. it holds ∂2 (x2 − 1) = (2 )! ∂x2 and. hence. However.124). appears not very interesting.114 By means of −(1 − x2 ) 2 = Theory of Angular Momentum and Spin −m−1 m+1 ∂ ∂ (1 − x2 )− 2 (x2 − 1) ∂x ∂x −m−1 −m m ∂ −(1 − x2 )− 2 (x2 − 1) ∂x −m −m−1 m+1 ∂ x (1 − x2 )− 2 − (m + 1) 1 ∂x −m−1 (1 − x2 ) 2 1 (5. It holds.130) (5. and constructs then the eigenfunctions Y − +1 .124) holds for m = . φ) using (5. φ) choosing the proper sign. by repeated application of the operator L+ .124) and. φ) L3 f (θ.135) . φ) as given in (5. 2 ! (2 )! ∂x2 1 (5. φ) = 0 (5. Y − +2 . therefore. it holds then for all m. φ) (5.. 5. yields a vanishing expression for m = − − 1 as long as is a non-negative integer. φ) An alternative route to construct the eigenfunctions Y of L− f (θ. In that case (1 − x2 ) is a polynomial of degree 2 and. Such construction reproduces the eigenfunctions Y m (θ.132). using x = cos θ. Constructing Y − → Y − +1 → ···Y m (θ. (5.124) holds for m if it holds for m + 1.124). We want to test if the recursion (5.131) identifies f (θ.128) one can show that (5. namely. i.124) and.e.132) which reduces the number of spherical harmonics which need to be evaluated independently roughly by half . P − (5. which is equivalent to recursive application of L− . 5.123) terminates for m = − . the derivative ∂ 2 +1 /∂x2 +1 of this expression vanishes.

(5.110) applies the = ( + m + 1)( − m) Y m+1 (θ.4: Angular Momentum Eigenstates Due to (5.5..142) For this purpose we proceed by induction. following closely the proof of eq.139) = ∂ − m cotθ ∂θ P m (cos θ) . According to (5. a sign consistent with the family of functions Y m constructed above. (5. (5. Obviously. (5.143) .106).120) for Y obey the postulated relationship (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. φ). (5. (5.124). We first note that for m = − (5.138) Employing (5.137) m (θ. (5.136) is normalized and has the proper sign. φ) .106. φ) sin θ e−i φ = 0. 5.129).132).80) and (5.136) We note in passing that this expression and the expression (5.123)] P m+1 (x) = −(1 − x2 ) 2 1 ∂ x − m 1 ∂x (1 − x2 ) 2 P m (x) .122.124) one arrives at Y (θ.140) Introducing again the variable x = cos θ leads to the recursion equation [c.e. One can readily verify that this expression provides a solution of (5. the expression (5. This follows from the identity ∂ ∂ − i cotθ ∂θ ∂φ One can use (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. φ) = 2 +1 1 1 sin θ e−i 4π (2 )! 2 ! φ 115 − . 5. i.136) to construct all other Y recursion eiφ ∂ ∂ + i cotθ ∂θ ∂φ Y m (θ.142) yields P − (x) = (−1) 1 (1 − x2 )− 2 (x2 − 1) = (1 − x2 ) 2 2 ! 2 ! (5.

According to both (5. Y − +2 . Since (5.124) yields (−1)m ( − m)! P ( + m)! −m (x) = +m m ∂ (−1)m (1 − x2 ) 2 (x2 − 1) . φ) ( − m)! P ( + m)! −m (x) .142) holds for m.132). are identical. etc.s. Accordingly. (5. of which agrees with the r.s.144) reads P m+1 (x) = +m+1 m+1 ∂ (−1)m+1 (1 − x2 ) 2 (x2 − 1) .e..142) and by (5..150) . Hence. of (5.144) ∂ x − m 1 ∂x (1 − x2 ) 2 +m m ∂ (−1)m (1 − x2 ) 2 (x2 − 1) .142) holds for m = − we verified that it holds for all m. (5. However. 1 (1 − x ) ∂x +m (1 − x2 ) 2 1 (5. one can conclude P m (x) = (−1)m m (θ.135).h.146) i. 2 ! ∂x 1 (5.e. The construction beginning with Y − and continuing with Y − +1 .141) reads then P m+1 (x) = 1 (5. Y −2 .142). P (x) = P 0 (x) are called Legendre polynomials.142) holds for m + 1 if it holds for m.148) According to (5. equivalently. etc. i.142) one can state P (x) = ∂ (x2 − 1) . 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 .147) the r.149) (5.124).h.128) m + 1 → −m or.116 Theory of Angular Momentum and Spin wich agrees with (5. (5. one notes that application of (5. We then assume that (5. 2 ! ∂x +m (5. given by (5. also the associated Legendre polynomials determined this way. 2 ! ∂x +m −(1 − x2 ) 2 Replacing in (5.124) and (5. yields the same eigenfunctions as the previous construction beginning with Y and stepping down the series of functions Y −1 .145) Hence. 2 ! ∂x +m+1 (5. (5.106) this implies for Y The Legendre Polynomials The functions the identity (5.

(5. e.158) see. The generating function allows one to derive useful properties of the Legendre polynomials. pp. accordingly. r1 is the point where the potential is measured.g.152) and. pp. t) is called a generating function of the Legendre polynomials3 . 92 To prove this property see. the spherical harmonics for m ≥ 0 Y m (θ. 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. New York. Jackson (John Wiley. for example. Lebedev (Prentice-Hall. =0 (5. t = r2 /r1 . P3 (x) = 1 2 117 P2 (x) = − 3x) 1 2 2 (3x − 1) .142) allows one to express the associated Legendre polynomials in terms of Legendre polynomials P m (x) = (−1)m (1 − x2 ) 2 m ∂m P (x) . t) = √ 1 − 2xt + t2 ∞ 1 − 2 x t + t2 . 2nd Ed. P1 (x) = x .156) w(x.. ”Classical Electrodynamics.D. . (5.J. N.4: Angular Momentum Eigenstates The first few polynomials are P0 (x) = 1 . In case q = 1 and |r2 | < |r1 | holds the identity2 ∞ 1 r2 = P (cos γ) (5. Englewood Cliffs. (5. 2.157) = 0. 1. 1975). . The spherical harmonics for m < 0 can be obtained using (5. and |r1 − r2 | = r1 (5.154) can be written 1 = w(x.” by J.153) P (cos θ) eimφ . φ) = 2 + 1 ( − m)! (−1)m sinm θ × 4π ( + m)! × ∂ ∂ cos θ m (5. =0 (5.5. m ≥ 0 ∂xm (5. . in case of x = 1 holds w(1.155) P (x) t .150) and (5. 45 of “Special Functions and their Applications” by N. .132).156) yields P (1) = 1 .N.. For example. and r2 denotes the location of the charge. Using x = cos γ. 1965) which is an excellent compendium on special functions employed in physics. 2 3 1 1 = = 2 1−t 1 − 2t + t ∞ t . t) = √ Comparision with (5.154) |r1 − r2 | r +1 =0 1 where γ is the angle between r1 and r2 .151) (5x3 Comparision of (5.

φ) m (π − θ.. Inversion Symmetry of Y m (θ.159) Using ∂lnw/∂t = (1/w) ∂w/∂t and multiplying (5. (5. (5. φ) . t) x−t = . Accordingly.151)] allows one to determine P (x) for = 1. t) holds Theory of Angular Momentum and Spin ∂lnw(x. φ) .166) We want to summarize the properties of the spherical harmonics Y derived above. Accordingly. φ) goes over to Y m (π − θ. 1.167) (5. Noting exp[im(π + φ)] = (−1)m exp(imφ) we determine Y Properties of Y m (θ.163) (5.160) this differential equation is equivalent ∞ (1 − 2xt + t2 ) =0 P (x) t + (t − x) =0 P (x) t = 0 (5.118 For w(x. θ. ∂t 1 − 2xt + t2 (5. If the spherical coordinates of r are (r. then the coordinates of − r are (r. π + φ) = (−1) Y m (θ. 2. φ).165) since ∂ n /∂(−x)n = (−1)n ∂ n /∂xn .159) by w(1 − 2xt + t2 ) leads to the differential equation obeyed by w(x. .164) which. φ) m (θ. ∂t P (x)t −1 −1 (5.142) allows one to conclude P m (−x) = (−1) +m P m (x) (5. . The spherical harmonics are eigenfunctions of the angular momentum operators J2Y J3 Y m (θ.f.168) mY m (θ. φ) Under inversion at the origin vectors r are replaced by − r. π + φ).162) − (2 + 1)x P (x) + P −1 (x) ] t + =1 [( + 1) P +1 (x) The coefficients for all powers t must vanish individually. holds P1 (x) ( + 1) P +1 (x) = = x P0 (x) (2 + 1) x P (x) − P −1 (x) (5. . under inversion Y m (θ. φ) = = 2 ( + 1) Y m (θ. π − θ. t) (1 − 2xt + t2 ) Employing (5. φ) m (θ. Due to cos(π − θ) = − cos θ the replacement r → −r alters P m (x) into P m (−x). Inspection of (5. using P0 (x) = 1 [c.156) and (∂/∂t) to P (x)t = ∞ ∂w + (t − x) w = 0 . φ) (5. π + φ). .161) Collecting coefficients with equal powers t yields 0 = P1 (x) − x P0 (x) ∞ (5.

φ) ∈ C∞ (S2 ) holds ∞ f (θ. 5. m (5. m 4. 4π (5. The spherical harmonics form an orthonormal basis of the space C∞ (S2 ) of normalizable. where ( = 1. (5.179) .175) (5. P −1 (x) ] (5. φ) = (−1)m Y ∗ (θ. . uniquely defined functions over the unit sphere S2 which are infinitely often differentiable π 2π sin θdθ 0 0 dφY ∗m (θ. φ)f (θ. a complete basis of C∞ (S2 ).169) 3. φ) = δ δm m .. .174) m ≥ 0 P (cos θ) eimφ . φ) C = =0 m=− π C sin θdθ 0 m Y m (θ. The spherical harmonics are given by the formula Y m (θ. in fact.176) = 1 +1 [ (2 + 1) x P (x) − are the Legendre polynomials. P +1 (x) P1 (x) = x . (5.) P0 (x) = 1 .5. φ) where J± = J1 ± iJ2 . 6. for any f (θ.173) m+1 (θ. φ) (5. φ) Y m (θ. . φ) . φ) 2π 0 (5. i. 2.172) (5.4: Angular Momentum Eigenstates 119 2. φ) m (θ.171) m = dφ Y ∗ (θ. The spherical harmonics Y 0 are Y 0 (θ.170) (5. φ) = 2 + 1 ( − m)! (−1)m sinm θ × 4π ( + m)! 2 ! × ∂ ∂ cos θ m (5. The spherical harmonics form.178) 7. The spherical harmonics obey the recursion relationships J+ Y J− Y m (θ. φ) = 2 +1 P (cos θ) . The spherical harmonics for m < 0 are given by Y −m (θ. except for the factor exp(imφ).e. For the Legendre polynomials holds the orthogonality property +1 dx P (x) P ( (x) = −1 2 δ 2 +1 .177) Note that the spherical harmonics are real. φ) = = ( + m + 1)( − m) Y ( + m + 1)( − m) Y m+1 (θ. φ) .

(5. φ).s. t) stated in (5.188) Exercise 5.120 8.h.3: Construct all spherical harmonics Y3m (θ. The spherical harmonics for m (θ. Exercise 5. The spherical harmonics for θ = 0 are Y m (θ Theory of Angular Momentum and Spin = 0.187) h(r) Y m (θ. 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 (5.189).2: Derive the orthogonality property for the Legendre polynomials (5.180) 9.4.181) = 0.189) evaluate the integral on the l.186) (5.s.185) (5.1: Derive expressions (5.86). of (5.179) using the generating function w(x. φ) = δm0 2 +1 .. 4π (5. φ) = 1 ∂2 r − r ∂r2 ( + 1) r2 h(r) Y m (θ. (5. π + φ) . =0 t + −1 dxP (x) P (x) . expand the result in powers of t and equate the resulting powers to those arising on the r. For this purpose start from the identity +1 ∞ +1 dx w(x.4. (5.4.h. For the Laplacian holds 2 = = = (5.183) (5. t)2 = −1 .87).156). φ) = (−1) Y m (π − θ.177). 1. The spherical harmonics obey the inversion symmetry Y 10. . φ) .184) Y22 Y21 Y20 together with (5.182) Y11 Y10 = = − (5. (5. 11. Exercise 5.

. exp − i ϑ · J . . exp − i ϑ · J (X ) = X ... ∞. 1. φ ) (5. . (5. φ). assumes a particularly simple form.192) = S2 dΩ Y ∗ (θ .197) The subspaces X have the important property that (i) they are invariant under rotations exp − i ϑ · J . }] (5. . Comparision with (5.m m m (θ. . φ) ∈ C∞ (S ). .m D(ϑ) m.196) C∞ (S ) = = X . m = S2 i dΩ Y ∗m (θ. φ) (5.e. i. φ) D(ϑ) = . therefore. φ) exp − ϑ · J .193) shows ∞ = − . φ) . φ) m. φ) i − ϑ·J f (θ..195) i.193) where [ { } ] denotes closure of a set by taking all possible linear combinations of the elements of the set. m = − . φ ) exp m One can also represent C∞ (S ) C∞ (S ) = [ {Y .5 Irreducible Representations We will consider now the effect of the rotational transformations introduced in (5.195). are characterized completely if one specifies the transformation of any Y m (θ.− + . and (ii) they form the lowest dimensional sets X obeying C∞ (S ) = X which have this invariance property.. ∞. . }] .. . . (5. ˜ f (θ.190) provide a complete. = − . φ) ˜ under such rotational transformation by f (θ.. . .194) Y m (θ.. with elements given by (5. if one specifies the functional form of the coefficients [D(ϑ)] m . φ) c = . .. The transformations exp − i ϑ · J . .. φ) ˜ Y m (θ. . = . }. We like to argue that the matrix representing the rotational transformations of the function space C∞ (S ). called the irreducible representation.e. = 0.. = .− + .5: Irreducible Representations 121 5. . (5. (5. φ) = exp Since the spherical harmonics Y space C∞ (S ) one can expand ˜ f (θ.191) i − ϑ·J f (θ . m . These coefficients can be considered the elements of an infinite-dimensional matrix which provides the representation of exp − i ϑ · J in the basis {Y m ... on functions f (θ. For this purpose we consider the subspaces of C∞ (S ) X = [ {Y .e.5. We denote the image of a function f (θ. m Y m (θ. i. orthonormal basis for the function c m Y m (θ. . φ) (5.42). − + 1.

Application of (5. 5.. leaves X invariant.199) [J+ .200) leaves X invariant. J− . J− 2 also leaves n1 X invariant. one needs to consider the action of J+ 1 J− 2 J3 3 on the subspaces X . n The action of J− on Y m produces Y m−1 or. produces 0. Similarly. Since any additive term n n n cn1 .197) of C∞ (S ) the matrix representation of D(ϑ) assumes a block-diagonal form.200) in (5. . with blocks of dimension 1.198) 2 2 where ϑ± = ϑ1 ( 1 ϑ− J+ 2 + 1 2 ϑ+ J− ϑ2 ..   . 3. . (5. J± ] = ± J± (5.65).n3 n n n Accordingly.200) of the rotational transformations one can also conclude that the X are the smallest invariant subspaces of C∞ (S ). If one orders the basis according to the partitoning (5.. i.56.e. One = contribute can conclude then that in the expansion (5.n2 .n3 J+ 1 J− 2 J3 3 .   1×1         3×3           . in case m = − . and J3 1 1 ϑ · J = ϑ− J+ + ϑ+ J− + ϑ3 J3 (5. (5.195) [D(ϑ)] m.     (5. The exponential operator exp − i ϑ · J + ϑ3 J3 )n . one concludes for the matrix elements (5. .e. 5.201) From expression (5. J− ] = 2 J3 .n2 . φ) . φ) exp m i − ϑ·J Y m (θ.202) D(ϑ) =           (2 + 1) × (2 + 1)               . i.. Since the Y m are n3 eigenfunctions of J3 the action of J3 reproduces the basis functions of X .199) allows one to express exp i − ϑ·J = n1 .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 . equivalently. can be expanded in powers One employs then the commutation properties [J3 .. One proceeds by expanding first ϑ · J in terms of J+ .194) only spherical harmonics with or. which follow readily from (5. one can conclude that J+ leaves X invariant. . m = δ S2 dΩ Y ∗ (θ. . the transformation exp − i ϑ · J leaves X invariant as well.e.. i.

e. i. 2. 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. k = 1. and for which detR = 1. 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). (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).22. (b) Prove that the elements of SO(2) can be completely characterized through a single parameter. Determine the generator of ρ(ϕ) in analogy to (5. (ii) a second rotation around the new x2 –axis by an angle β. this parametrization. (iii) a third rotation around the new x3 –axis by an angle γ.5. The representation is referred to as the irreducible representation. (a) Show that SO(2) with the group operation ◦ defined as matrix multiplication.e. real 2 × 2– matrices O through O = T (ϕ)O = R(ϕ)OR−1 (ϕ) with R(ϕ) as in (c) is also a representation of SO(2). 5. 5. . f (α + 2π) = f (α). is a group. (f) The transformation ρ(ϕ) as defined in (e) leaves the following subspace of functions considered in (e) Xm = { f (α) = A e−imα . though seemingly natural. other representations are termed reducible representations. A ∈ C. (e) In the space C∞ ( ) of infinitely often differentiable. Exercise 5. 3 certainly allow one to describe any rotation around the origin. However. ∈ N} invariant. Give the corresponding expression of ρ(ϕ).5. i.1: Representations of the group SO(2) SO(2) is the set of all 2×2 matrices R which are orthogonal. does not provide the simplest mathematical description of rotations.6: Wigner Rotation Matrices 123 This representation of D(ϑ) is the one wich has blocks of the lowest dimensions possible. The three components ϑk .49). (d) Show that the similarity transformation T (ϕ) defined in the space of non-singular. for which holds RT R = R RT = 1 1.6 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). This step reconsiders the parametrization of rotations by the vector ϑ assumed sofar.

209) The first two rotations in (5. β. m i Y m (θ. followed by J2 in the original frame.e. of (5.m m. γ)] = .124 Theory of Angular Momentum and Spin The angles α. Accordingly. Obviously. the l.207) are equivalent. (5.203) can be expressed.h. is e− i γJ3 = e− i βJ2 e− i αJ3 e− i γJ3 e i αJ3 e i βJ2 (5.203) has the disadvantage that it employs rotations defined with respect to three different frames of reference. i.. i (5. β. one can replace (5. β.e.205) Y m (θ. m [D(α.207) which replaces J2 by the inverse transformation (i).e. For this purpose we notice that J2 can be expressed through the similarity transformation J2 = e− i αJ3 J2 e i αJ3 (5. β. γ ∈ R such that exp i − ϑ·J = e− i γJ3 − i βJ2 − i αJ3 e e (5.212) .204) is satisfied.195) by ˜ Y m (θ. The Euler rotation replaces exp − i ϑ · J by e− i γJ3 − i βJ2 − i αJ3 e e . φ) [D(α.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 (5.194. The axis x2 is defined in the coordinate frame which is related to the original frame by rotation (i). (5.208) αJ3 βJ2 = e− αJ3 e− i βJ2 e i (5. in terms of rotations defined with respect to the original frame.211) Using (5.206) The expression (5.203) For any ϑ ∈ R one can find Euler angles α.203).. We will demonstrate that (5. For any similarity transformation involving operators A and S holds eS A S Accordingly. followed by transformation (i). however.203). the axis x3 is defined in the coordinate frame which is related to the original frame by the consecutive rotations (i) and (ii). φ) = S2 dΩ Y ∗m (θ. φ) (5.h.. i. and the r. i. can then be written e− i βJ2 − i αJ3 e = e− i αJ3 e− i βJ2 (5. in analogy to (5. φ) e− γJ3 − i βJ2 − i αJ3 e e .s. γ will be referred to as Euler angles. 5. the transformation from the rotated frame to the original frame. (i) and (ii).210) The third rotation in (5. we can write e− i −1 = S eA S −1 . γ)] m.s. the transformation from the original frame to the rotated frame.209).

Best known is the spin of the electron. (5. β.215) benefits from the choice of Euler angles for the parametrization of rotational transformations.215) The evaluation of the matrix elements (5. φ) e−imγ . γ)] m. to redefine the three rotations in (5.210) yields the simple result e− i γJ3 − i βJ2 − i αJ3 e e = e− i αJ3 e− i βJ2 e− i γJ3 . (5. φ) i i (5.213) i.206) by ˜ Y m (θ.217) Accordingly. The eigenvalue property (5.m m. (5. φ) e−iγm δ (5. m ( ) S2 dΩ Y ∗ (θ. γ)] m.218) Defining the so-called Wigner rotation matrix dmm (β) = one can express the rotation matrices [D(α. m (5. φ) .310)] an explicit expression for the Wigner rotation matrix (5.219) = e−iαm dm m (β) e−iγm δ ( ) . 5. φ) (5.. β.309. φ) e− αJ3 e− βJ2 Y m (θ. in addition. β. 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.219). φ) [D(α.e. φ) = e−im α S2 dΩ Y ∗ (θ.64) yields dΩ f (θ.7: Spin 1 2 and the group SU(2) 125 Multiplication from the left with (5. 5. the self-adjointness of the operator J3 one can state dΩ Y ∗ (θ. (5. φ) . one can write [D(α. the muon µ and its neutrino νµ of the second . (5. φ) e− m i βJ2 Y m (θ. φ) Y m (θ. φ) e− m i βJ2 Y m (θ. m [D(α.216) Using. γ)] = .214) e− γJ3 = S2 dΩ Y ∗m (θ.203) with respect to the original (unprimed) frame one simply needs to reverse the order of the rotations. m = e−iαm S2 dΩ Y ∗ (θ. φ) e− m S2 i αJ3 f (θ. Examples are the other five members of the lepton family to which the electron belongs. γ)] m. β. but many other elementary components of matter are endowed with spin-like properties. 5. m i Y m (θ.205. φ) = S2 dΩ f (θ.220) We will derive below [see Eqs. φ) f (θ.5. This allows one to express the rotational trasnformation (5. φ) e− S2 i γJ3 Y m (θ.

the quarks. then allowed + − symmetry transformations of spin states are described by 2 × 2-matrices U with complex-valued matrix elements Ujk . Spin.1 systems 2 2 can be related to two states which we denote by χ+ and χ− .1 and spin. which might be the reason why consideration of rotational symmetry is so often fruitful in the study of matter. It appears to be rather impossible for a Physicist not to be enamored with the spin property. the two W± and the 2 2 Zo . The presence of this property permeates matter also at larger scales than those of the elementary particles mentioned. in principle. however. In any case. one for each matrix element of U † U . two of which (in certain linear combinations) make up the vector mesons which carry spin 1. leaving its imprint on the properties of nuclei. We will consider first only the so-called spin– 1 . a relationship. there are together five conditions in terms of real quantities to be met by the matrix elements and. How can the elements of SU(2) be parametrized. the tau τ and its neutrino ντ of the third generation and their antiparticles carry a 1 spin. there is nothing more elementary to matter than the spin property. 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 . carry other spin-like properties which together. 3 representing a 2 × 2 matrix. We will find that the spin in its transformation behaviour is closely related to angular momentum states. As complex 2 × 2 matrices one needs. and because of detU = 1. and which carry spin 1. the gluon. There . 1 We will specify for the transformations considered detU = 1. e. particles which mediate the strong and the electro-weak interactions. four real and four imaginary parts of Ujk . k = 1. U U † = U † U = 1 detU = 1 } 1. atoms and molecules. as long as |c+ |2 + |c− |2 = 1. Such systems can also assume any linear combination c+ χ+ + c− χ− . in fact. like superconductivity and magnetism. eight real numbers to specify the matrix elements. This specification implies that we consider transformations save for overall factors eiφ since such factors are known not to affect any observable properties. There are also the mediators. the spin of the electron is likely the most important property in Chemistry.e.g. and the condition detU = 1 requires the Sk to have vanishing trace. 2.221) One can show readily that this set forms a group with the groups binary operation being matrix multiplication. The particles mentioned. k = 1. Because of the unitarity condition U † U = 1 which are really four equations in terms of real 1 quantities. (5. have properties beyond those of single spins. hence.126 Theory of Angular Momentum and Spin generation. The transformation matrices are then elements of the set SU (2) = { complex 2 × 2 matrices U . One can show that the unitarity condition requires these matrices to be hermitian. the degrees of freedom of the matrices U are three real quantities. and three of which make up the baryons which carry spin. The important feature is that all U ∈ SU(2) can be parametrized by an exponential operator U = exp − i ϑ · S (5. i. generalizing then further below. [Sk ]mn = ([Sk ]nm )∗ . We may finally mention that the spin is at the heart of many properties of condensed matter systems.3 . 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− . So do the three generations of six quarks.2 . 2. the photon. each component Sk . If we identify the states χ+ and χ− with the basis of a Hilbert space. c± ∈ C.222) where the vector S has three components. j.

223) where σk . According to this property the Pauli matrices generate a 3dimensional Clifford algebra C3 . One can readily verify the commutation property σj σk − σk σj = [σj .225) σ (5. it holds. y. A = x σ 1 + y σ 2 + z σ3 . the simplest choice being S1 = 1 σ1 .31) of the group SO(3). k = 1. 2 S3 = 1 σ3 2 (5. but their comonents must not necessarily commute with each other. σk ]− = 2i jk z x − iy x + iy −z . σ2 = 0 −i i 0 . ∈ R . z A = In fact..7: Spin 1 2 and the group SU(2) 127 exist three such linear independent matrices.227) which is essentially identical to the Lie algebra (5. 3 are the Pauli matrices σ1 = 0 1 1 0 . hermitian 2 × 2–matrices A can be expandend. The Pauli matrices obey the following anti-commutation properties σj σk + σk σj = [σj .e. z. (5. (5. The proof of this relation rests on the commutation relationship and the anti-commutation relationship (5. b are vectors commuting with σ. namely.k=1 σ σ 3 j k j. σ3 = 1 0 0 −1 .224) provide a basis in terms of which all traceless.k=1 σ σ aj bk + j>k aj bk = 3 j. 2.e. e.5. (5. one obtains (σ · a) (σ · b) = 3 j 2 j=1 (σ ) aj bj 3 j k j. (5. i. Any such A can be expressed in terms of three real parameters x. 5. σk ]+ = 2δjk 1 . x. y. (σ · a) (σ · b) = a · b 1 + iσ · (a × b) 1 (5.k=1 j<k + σ j σ k aj bk .224) Algebraic Properties of the Pauli Matrices The Pauli matrices (5.229) and avoids commuting the components aj and bk .g.230) where a. 1 i.226) The Pauli matrices satisfy very special commutation and anti-commutation relationships.227.231) . 1 σj σk = − σk σj for j = k which can also be readily verified. Clifford algebras play an important role in the mathematical structure of physics.228) (5. 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. σj 2 (5. they are associated with the important fermion property of matter. it might hold aj bk − bk aj = 0. 2 S2 = 1 σ2 . In fact.229) = 1 . At this point we will state a useful property.

S3 χ± = ± 1 χ± 2 .231) proves (5. (5.222).k=1 (σ σ j>k σ k σ j (aj bk − ak bj ) = (5. We like to derive this result now by evaluating the transformations of SU(2) explicitly. i. This relationship emerges. yields a · b.234) 5.222) as a rotation vector and you rotate once. In the special case a = b ∈ R holds (σ · a)2 = a 2 1 .k=1 j>k σ j σ k (aj bk − ak bj ) = 1 2 3 j. spin changes sign.k=1 j>k i σl (aj bk − ak bj ) = i =1 σ a×b . hence. (5.s.222) by the Euler form exp(−iγS3 ) exp(−iβS2 ) exp(−iαS3 ) .h. (5. to rotations in space. only a 720o rotation leaves the spin invariant. The commutation property (5.238) S 2 χ± = .k=1 j>k 1 2 3 j.235) which is identical to that of the generators of SO(3).230). Well almost like it. since there is a slight difference: when you interprete the 3-component ϑ in (5. The algebra of the generators of a transformation is such a basic property that the ‘accident’ that the generators of spin transformations behave in this respect like the generators of 3-dimensional rotations makes spins appear in their physical behaviour like rotations.128 Theory of Angular Momentum and Spin Using (σ j )2 = 1 the first term on the r. however. The two remaining terms yield using σ j σ k = −σ k σ j for j = k and altering ‘dummy’ summation indices 3 j. (5. however. For this purpose we choose to replace (5.227) leads to 3 3 jk j.236) The matrix elements of this 2 × 2 operator can be labelled by the basis states χ+ and χ− .k=1 j>k σ j σ k (aj bk − ak bj ) − 1 2 3 j k j. let say around the x2 –axis by 360o .e.237) + 1)1 and.232) − σ k σ j ) (aj bk − ak bj ) . 1 1 1 ( + 1) χ± 2 2 . through the algebra obeyed by the operators Sk [Sk . 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. 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 .233) This result together with (5.8 Generators and Rotation Matrices of SU(2) Sofar there is nothing which relates the transformations U . S ] = i k m Sm (5.

. The idempotence property of Sk yields particularly simple expressions for these powers.243) We note the property of this rotation matrix 2 dmm (2π) (1) = −1 1 .9 Constructing Spin States with Larger Quantum Numbers Through Spinor Operators In this section we like to demonstrate.239 ) one expands the exponential operator exp(−iβS2 ). m = − . no two such particles can exist in the same state. . (5. In order to determine the Wigner matrix in (5.239) in the notation (5. (5. i.219). . identical particles any number of which can exist in the same state χ+ and χ− .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. i. 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. . namely. that states | m for higher quantum numbers = 0.9: 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).240) .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 (5.242) The expressions in brackets [.244) i.e. 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 . 2 2 We obtain with this notation for the transformations (5.e.239) where we made use of the notation (5. . therefore. use the label 2 2 | 1 ± 1 for the states χ± . For this purpose one needs to determine the powers of −iβS2 .5. β.. One cannot consider the 1 entities carrying the spin– 2 to be particles in the ordinary sense for it can be shown that spin– 1 2 particles have fermion character.and sinfunctions and one obtains for the rotation matrices 2 dmm (β) (1) = cos β 2 sin β 2 − sin β 2 cos β 2 .5. 2. We may. 3 . 1. .219. ( − 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. . The complete matrix elements (5. following Jourdan and Schwinger.e. (5.245) 5.220) are [D(α. − + 1. 1 .] can be identified with the Taylor expansion of the cos. (5. γ)] m. . rotation by 360o changes the sign of the spin state. .

m) = b† − j−m (j + m)! (j − m)! |Ψ0 . b− )T is expressed through the commutation relationships + − (ζ. (5. b† |Ψ0 = χ+ + .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 − . m = −j. −) bζ .247. also operators of the type x b† = x+ b† + x− b† + − .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± . b ζ The States |Ψ(j. .254) We will show below that these states are orthonormal and form spin states with higher quantum numbers.246) The corresponding adjoint operators are denoted by b+ and b− .251) One can show using this property and (5.e. . . j = 0.253) = δζζ . b− )T as spinor creation and annihilation + − operators. . b± |Ψ0 = 0 (5.252) (5. 3 . b ζ = b ζ. 1. . We consider. −j + 1. x∗ b = x∗ b+ + x∗ b− + − (5.e. .250) which for x∗ x = x∗ x+ + x∗ x− = 1 represent spinor operators in an arbitrary reference system. b− † = − sin β † β b + cos b† . b ζ = 0 b ζ. b† )T and b = (b+ . i. + − For example. b † ζ ζ =0 (5. The operators b† allow one to construct a set of states which represent j + m–fold and j − m–fold ± χ+ and χ− states as follows b† + j+m |Ψ (j.248) bζ .5.249) In the following will refer to b† = (b† . 2 + 2 − † (5. i. b† |Ψ0 = χ− − . ζ = +. (5. using (5. b† )T and b = (b+ . j. 1 . 2 2 . b † ζ For the vacuum state |Ψ0 holds = δζζ .248) that bζ relationships † † b ζ. . These operators are associated with a given spatial reference system. The boson character of the operators b† = (b† . therefore. m) † and bζ obey the commutation (5. b ζ = b† .247) (5.

247. b† + b+ . (5.10: Algebraic Properties of Spinor Operators 131 5.5.249).10 Algebraic Properties of Spinor Operators We want to establish first a few important and useful algebraic properties which result from the commutation relationships (5. in b† and b† . b † + = n b† + n−1 b† + b † + + n = (n + 1) b† + . b† + n . 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+ . + + (5. f (b† ) + (5. We will derive the result.249) b+ f (b† ) |Ψ0 = + b+ . The Spinor Derivative Operator A most important operation is the action of the annihilation operators b+ and b− on operators which can be expressed as polynomials.258) Equations (5.255) In order to determine how this operator is modified by multiplication from the left by b+ we note b+ f (b† ) |state = + b+ . b† + = n b† + n−1 . b† + n = n cn b+ .256) and (5. we need to evaluate b+ . 5. + b+ .255) is b+ .248) and from (5.248) we obtain b+ .257) which for (5. that the annihilation operators play a role similar to the differential operator in calculus. f b† Obviously. An example + − is the monomial b† which generates the states (5. We will show that for this commutator holds n b+ .260) .256) In the special case |state = |Ψ0 this reads using (5. (5.257) motivate us to determine the commutator (5. f (b† ) |Ψ0 . b† + − well-known for the quantum mechanical harmonic oscillator. b† + n+1 = = b+ . f (b† ) |state + f (b† ) b+ |state .259) The property is obviously true for n = 0. (5. .254). possibly infinite power series. If it is true for n then using (5.

247) the more general property for polynomials f (b† . 2 ..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 . (5. m) = x− b† − j−m (j + m)! (j − m)! |Ψ0 . m) m=−j (5.266) where xb† has been defined in (5.. f (b† ) − = f (b† ) − (5. b† ) in b† and b† holds (ζ = +.266) we compare the terms x+ b† + j+m φjm (x) Ψ (j. the property holds also for n + 1. b† ) |Ψ0 = + − ∂ ∂b† ζ f (b† . m) We want to prove now the property ∞ j exp xb † |Ψ0 = 1 j=0. b† ) |Ψ0 . One can finally conclude for polynomials (5. b† ) + − = ∂ ∂b† ζ f (b† . f (b† .265) Generating Function of the States |Ψ(j.263) According to (5.261) Similarly.255) b+ . f (x)] g(x) = (∂k f (x))g(x) or [∂k . m) φjm (x) = j−m xj+m x− + (j + m)! (j − m)! . one can prove b− . In order to derive (5. + − (5.264) This demonstrates the equivalence of the spinor operators bζ and the derivative operation.259) holds for all n ∈ N. + − (5. Equation (5.267) exp xb† |Ψ0 is called a generating function of |Ψ(j. m) .257) we can conclude in particular bζ f (b† . By induction we can conclude that (5.1. b† ) .262) Because of the commutation relationships (5.250) and where φjm (x) represents the function of the two variables x+ and x− closely related to |Ψ(j. = f (b† ) + (5. f (x)] = ∂k f (x) . (5.. (5.e. −) + − + − bζ . φjm (x) Ψ (j.132 Theory of Angular Momentum and Spin i. f (b† ) + where f (x) = df dx .268) .

2.274) Summing this expression over 2j = 0. (5. 1 . 1 . 3 .. . 2.272) The summation over s can be written in terms of j and m using (5. .1. one obtains φjm (x) Ψ (j.275) = exp xb† |Ψ0 . 2 . i... 1. . chosing the summation index j = 0.1. . 1.1.269) (5. 2 ∞ 1 xb† 2j! u 2j |Ψ0 = u=0...e.270) N 2j j → s=0 j−m=0 → m=−j . m) m=−j = = 1 x+ b† + x− b† + − (2j)! 2j 1 xb† |Ψ0 ... (5. m) 1 j=0.271) 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)! (5.5. (2j)! 2j |Ψ0 (5. 2 2 leads to ∞ j ∞ φjm (x) |Ψ (j.10: Algebraic Properties of Spinor Operators with the s–th term in the binomial expansion of (a + b)n n! an−s bs .. 1 xb† u! |Ψ0 (5. . .270) s x+ b† + N −s x− b† − s |Ψ0 . m) = = |Ψ0 (N − s)! s! N −s 1 N! x+ b† x− b† + − N ! (N − s)!s! N −s 133 (5.273) Change of the summation indices allows one to conclude j φjm (x) Ψ (j.... s! (n − s)! Defining j+m = N −s j−m = s. m=−j = j=0.

In order to evaluate the ∗ b) (yb† ) operator e(x e we notice that the derivative property (5. (5. [b† . hence. of (5. b− ) holds † † f (b+ . m) |Ψ j .279) One can generalize this to † † s (bζ )s e(yb ) |Ψ0 = yζ e(yb ) |Ψ0 . m) |Ψ j .278) † ∗ which allows us to replace the l. The generating function allows one to derive various properties of the states |Ψ (j.e.m φjm (x∗ ) φj m (y) Ψ (j. For this purpose we consider the inner product † † e(xb ) Ψ0 |e(yb ) Ψ0 = δjj δmm (5.s. b† ] = 0 and [b− . (5.254) are orthonormal. According to (5. (5.264) implies † † bζ e(yb ) |Ψ0 = yζ e(yb ) |Ψ0 .282) where we defined x∗ y = x∗ y+ + x∗ y− . we can conclude † † ∗ e(xb ) Ψ0 |e(yb ) Ψ0 = e(x y) (5.h.276) = j. i.134 Theory of Angular Momentum and Spin which concludes our derivation.249) for any non-vanishing integer s holds + − Ψ0 |bs Ψ0 = 0 and.m.277) by Ψ0 |e(x b) e(yb ) Ψ0 . b− ] = 0. b− )e(yb ) |Ψ0 = f (y+ . (5.281) and.284) . m . m holds. m) and will be used for this purpose below. that Ψ (j. m) We want to show now that the states (5. 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. Orthonormality of the States |Ψ (j.277) To evaluate this expression we first notice † e(xb ) † = e(x ∗ b) . y− )e(yb ) |Ψ0 (5. b† ] = 0 allow one to state that for + − + any polynomial f (b+ . therefore.283) (5.280) The commutation properties [b+ .j .

5.m .m. k = 1.268–5.m ijk Jk . One defines corresponding operators Jk through 1 Jk = b† < ζ |σk | ζ > bζ . 1. 2 .n . bm bm } n n = 1 4 < n |σi | m >< n |σj | m > {b† bm δmn − b† bm δm n } n n n..m. In our present notation the matrix elements < ζ |σk | ζ > for ζ = ± correspond to the matrix elements < ζ |σk | ζ > for σ = ± 1 . Jj ] = i This property is derived as follows [Ji .254) are eigenstates of operators J 2 and J3 with eigenvalues j(j + 1) and m where J 2 and J3 are representations of the spin 2 2 2 operators S 2 = S1 + S2 + S3 and S3 defined in 5. Jj ] = = 1 4 1 4 < n |σi | m >< n |σj | m > b† bm . J1 = J2 = J3 = 1 † b b− + b† b+ − 2 + 1 † b b− − b† b+ − 2i + 1 † b b+ − b† b− . m) as given in (5. < + |σ1 | − > = < − |σ1 | + > = 1. States 2 We want to demonstrate now that the states |Ψ (j. (5.277) φjm (x∗ ) φj m (y) Ψ (j.5.285) Following steps similar to those in Eqs.n .282) and (5. 1 New Representation of Spin 0.ζ where σk . . etc. b† bm n n n. .286) from which follows immediately the orthonormality property (5. 2. (5.223.m +b† b† . − 2 + (5. (5.m φjm (x∗ ) φjm (y) .m 135 = e(x ∗ y) . The 2 operators are explicitly.n .287) ζ 2 ζ.j .276).275) one can show e(x ∗ y) = j. (5. 3 denote the Pauli spin matrices (5. (5.224).289) < n |σi | m >< n |σj | m > {b† bm . 3 .288) The three operators obey the Lie algebra of SU(2) [Ji .m.224).10: Algebraic Properties of Spinor Operators and comparing (5. b† bm n n n. . m j. using < + |σ1 | + > = < − |σ1 | − > = 0.m. m) |Ψ j .

292) The commutation relationships (5. (5.m 1 4 < n |σj | m >< m |σi | m > b† bm n n .288) yields first the expression J2 = which results in J2 = 1 4 1 † ( b + b + − b † − b− ) ( b † + b+ − b† − b − − 2 ) + b † + b − b † − b + 4 .k ijk ijk σk | m > b † bm n = i = i 1 2 < n |σk | m > b† bm n n. (5. m) as Eigenstates of J 2 and J3 We wish to show now that the states |Ψ (j.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 + ) (5.136 = 1 4 − = = 1 4 1 4 Theory of Angular Momentum and Spin < n |σi | m >< m |σj | m > b† bm n n.m. − (5.m < n |2i n.297) . J1 + iJ2 ] = −i [J2 .m ijk Jk . (5.m. σj ]| m > b† bm n n. m) are eigenstates of J 2 and J3 .293) .294) .290) |Ψ (j. 2 J 2 = J+ J− + J3 − J3 = J3 (J3 − 1) + J+ J− (5. J2 ] = −2 J3 from which we can conclude the property 1 1 J− J+ = J+ J− − J3 2 2 Hence.292).289) yield [J− .295) The last result together with (5.m < n |[σi . (5. (5. To this end we note 2 2 2 J 2 = J1 + J2 + J3 = 1 1 2 J+ J− + J− J+ + J3 2 2 . (5.m. J+ ] = [J1 − iJ2 . J1 ] + i [J1 .291) Here we have defined J+ J− = J1 + iJ2 = J1 − iJ2 = b† b− + = b† b+ .

m I1 = where λ = 0.1: 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) . 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† . ˆ respectively. using operators I± = I1 ± iI2 2 2 2 and the subspace {|n1 . .5. . I3 = (b+ b1 + b+ b2 ) 2 2 2 2 2i 2 1 satisfy the Lie algebra of SU(2). the states (5. . . and eigenstates of k with eigenvalues j. b† . −λ + 1. the number of states to the possible energy eigenvalues. [bj . (b) Show that the three operators 1 + 1 1 + (b1 b2 + b+ b1 ) .304) (5. Show λ = n1 +n2 n1 +n2 2 ( 2 + 1) and m = 1 2 (n1 − n2 ).s. .e. λ. m = −λ. m = λ||λ. i. m) J− |Ψ (j. b+ ] = δjk . n} eigenstates of I 2 = I1 + I2 + I3 and of I3 I 2 ||λ.10: Algebraic Properties of Spinor Operators The last term on the r. .299) (5.298) (5. m) = = (j + m + 1) (j − m) |Ψ (j. i. m) One can furthermore derive readily J+ |Ψ (j. .302) (5. b † + b + + b † + b+ b † − b − 1 1 b† + b + + b † + b + b † − b− + b † − b − b † + b + . m) J3 |Ψ (j. n = n1 + n2 fixed.h. One can then conclude J 2 |Ψ (j. 2 2 137 (5. [Ij .254) are eigenstates of + b+ and − b− with eigenvalues j + m and j − m.300) (5. I3 ||λ. . 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. n1 = 0. m) . State the corresponding eigenvalues and the degree of degeneracy. m − 1) . I2 = (b1 b2 − b+ b1 ) .303) Exercise 5. 1 2 k (a) Show that the eigenstates are given by 1 |n1 . (5. m = m||λ.. .305) = = j ( j + 1 ) |Ψ (j. 1 2. j = 1. 2. . Ik ] = i jk I . 2 . m .301) Obviously. 1. .10. n2 . .e. m) m |Ψ (j. (5. . Construct. m + 1) (j + m) (j − m + 1)|Ψ (j.

308) The latter sum involves terms (b† )j+m (b† )j−m for σ + σ = j + m and 2j − σ − σ = j − m.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 × (5.5.306) and (5. On the other side the states |Ψ (j.310) it reduces to (5.243).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). for example. For this purpose we note that the states |Ψ(jm) in a rotated coordinate system according m 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    (5.251). (j) (5.254) are |Ψ j. m ) are related to the + − states |Ψ (j.306. m) in the original coordinate system by j |Ψ j. In case j = 1 this expression yields. However.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† .306) where b † and b † are given by (5. m = m=−j dm m (β) |Ψ (j. + − One may expect that these two conditions restrict both σ and σ. σ values which yields (b† )j+m (b† )j−m is + − then σ = j + m − σ.309) . m = b† + j+m b† − j−m (j + m )! (j − m )! |Ψ0 (5. m) . 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. The prefactor of (b† )j+m (b† )j−m which according to (5.138 Theory of Angular Momentum and Spin 5. both conditions are satisfied for σ = j + m − σ. The combination of σ. .307) Comparision of (5.

313) × 0 0 2  =  0 sinβ 0 cosβ 1 −i 0 which.12 Mapping of SU(2) onto SO(3) The representation (5.309) does not distinguish between SU(2) transformations (5.310) establishes a mapping of SU(2) onto SO(3). however.s. U =  0 1 1 i √ ( x1 − ix2 ) √ x3 − √2 0 2 2 (5. 2 2 − (5.e. in fact.308) and had the form (5. Hence.309). x3 ) ∈ R . in case of j = 1.314) leaves (5.312) The choice of this transformation derives from a property shown further below.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  . .h.. This factor implies. namely that the component of the vector on the l. i.308) unaltered except for a factor (−1)2j which multiplies then also the final result (5. is the expected element of SO(3). The SU(2) transformation entered in (5. b− † = sin β † β b+ − cos b† . of (5.314) in case of integer j–values.251).5.e. e. It is of interest to trace the mapping from SU(2) onto SO(3) as represented by (5.310). One can. This can be shown by (1) applying to the matrix A = (dm m ) in (5.12: Mapping of SU(2) onto SO(3) 139 5.311) (5.312) transforms like an angular momentum state ˜ |1 m .. that the the representation (5. the orthogonal 3 × 3 matrix which describes a rotation around the x2 –axis. x2 . b+ † = −cos β † β b+ − sin b† 2 2 − .251) and (5. Evaluation of A yields √    −1 0 1 1 √ cosβ + 2 sinβ 1√ cosβ − 1 ˜ 0 A = 4  i √ i   − 2 sinβ 2 cosβ 2 sinβ  × √ 0 2 0 1 − cosβ − 2 sinβ 1 + cosβ √     −1 −i √ 0 −2 cosβ −2 2 sinβ 2 cosβ 2i 2i  × × 0 0 2  = 1  4 √0 1 −i 0 −2 sinβ 2 2 cosβ 2 sinβ     −1 −i √ 0 cosβ 0 −sinβ  1 0 (5. therefore. i. the matrix A should represent a rotation around the x2 –axis in the space of vectors ˜ (x1 .g. Replacing the latter transformation by its negative form.   = U  x2  .310) to the SU(2) transformation assumed in deriving the general result (5. conclude that the mapping from SU(2) onto SO(3) is a 2–1 mapping.309) and the particular matrix (5.

140 Theory of Angular Momentum and Spin .

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

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

. 3 (6. from which follows in turn the well-known result that H is related to H through the similarity transformation H = R(ϑ) H R−1 (ϑ). the existence of a basis (6. 5.53. It holds then H R(ϑ) ψ = R(ϑ) H ψ. To specify this property mathematically let us denote by H the Hamiltonian in the rotated frame.0: Addition of Angular Momenta and Spin 143 corresponding physical properties of the elementary components. ρC ) it follows H R(ϑ) = R(ϑ) H. Rotational Symmetry of the Hamiltonian As pointed out already. the commutation relationships (1) Jp .8) Obviously.5) do not affect the Hamiltonian. The invariance of the Hamiltonian under overall rotations of the three particle system implies then H = R(ϑ) H R−1 (ϑ) .9) hold since the components of J (k) are differential operators with respect to different variables.10) . connected with the fact that the Hamiltonian H does not depend on the overall orientation of the three interacting particles. i. Following our description of rotations of single particle wave functions we express (6. according to (5. 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. Hence. ρC ) = ψ(R−1 (ϑ) rAB . 2. the reader will learn in the following section how to add and subtract in the microscopic world of Quantum Physics. assuming presently that H might.48) i i R(ϑ) = exp − ϑ · J (1) exp − ϑ · J (2) (6. be different from H.4) which decouples rotational states is connected with the rotational symmetry of the Hamiltonian of the three particle system considered. but solely under simultaneous and identical rotations of rAB or ρC . presumably a facility the reader would like to acquire with great eagerness. ∂yAB ∂zAB i (2) ∂ ∂ − J3 = ρy − ρx . 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. (6. q = 1.55) holds i (1) − J1 = zAB ∂ ∂ − yAB . rotations R(ϑ) of the wave functions ψ(rAB . R−1 (ϑ) ρC ) (6.5) according to (5. in fact. For ˆ ˆ example. Jq(2) = 0 for p. ρC ) defined through R(ϑ) ψ(rAB .6. One can equivalently express therefore (6.6) For the following it is essential to note that H is not invariant under rotations of only rAB or ρC . Since this is true for any ψ(rAB .7) where the generators J (k) are differential operators acting on rAB (k = 1) and on ρC (k = 2).7) i R(ϑ) = exp − ϑ · J (6. ∂ρx ∂ρy (6. In this sense.e.

C].6) for rotational invariance of the Hamiltonian in the form i i H = exp − ϑ · J H exp + ϑ · J . that energy.13) Hϑ · J − Since this holds for any ϑ it follows HJ − J H = 0 or. that the components Jk . the important fact that the Jk obey the Lie algebra of SO(3). One can show readily that the commutation relationships [Jk .81)]. (6. (6. i. Eqs. 3 cannot have simultaneous eigenstates among each other.144 where J = J (1) + J (2) Addition of Angular Momenta and Spin . k = 1. 2.f. The property (6.14) We will refer in the following to Jk . the operators Jk . r ρ Definition of Total Angular Momentum States The commutation property (6. ˆ (6. Jk ] = 0 .. the (n) (n) (n) properties [Jk . i.12) We consider this equation for infinitesimal rotations. J ] = i k m Jm (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 ).12) each individually can have simultaneous eigenstates with the Hamiltonian.11) we can write the condition (6. For a proof one uses (6. componentwise.17) . (5. however. We recognize. for |ϑ| H ≈ 1 − 1 i ϑ·J H 1 + 1 i ϑ·J ≈ H + i 1.14) implies that the total angular momentum is conserved during the scattering process. 2.81) we want to consider the relationship between YJM and Y 1 m1 (ˆAB ) Y 2 m2 (ˆC ). In fact. a supposition which can be tested through the commutation properties of these operators.71–5. B] + [A. 2 together with the property [A. In the following we will use the notation r ρ Ω1 = rAB . To order O(|ϑ|) one obtains i ϑ · JH .e. k = 1.16) following the procedure stated in the theorem above [c. we will find that the states yield the basis B with the desired property of a maximal uncoupling of rotational states. [H. (6. ˆ Ω 2 = ρC . (6. of course. 3 as the three components of the total angular momentum operator.e. Before we apply the procedure (5. i. B + C] = [A.9). We suspect.e. 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. 3 . k = 1.. and the eigenvalues of J2 and J3 are good quantum numbers. k = 1. 2. 2. J ] = i k m Jm for n = 1.15) are satisfied.71–5.11) By means of (6.14) implies that the components of the total angular momentum operator (6. 3 do not commute.

the reader should always keep in mind that conditions equivalent to (6.s.. 2 |Ω1 .. ρC ) are normalized. and J (1) · J (2) commute. 2 M YJM ( 1 .21) One can.m2 2 |Ω1 . the eigenvalue being specified by the quantum number M .19) Noting J3 = J3 + J3 and applying this to the l.1: Show that J2 .21) hold. Ω2 ) (1) (2) = 1 m1 2 m2 ) Y 1 m1 (Ω1 ) Y 2 m2 (Ω2 ) (m1 + m2 ) (J M | . (6. 2 |Ω1 .e. J3 . The expansion coefficients (J M | 1 m1 2 m2 ) ˆ are called the Clebsch-Gordan coefficients which we seek to determine now. or the closely related Wigner 3j-coefficients introduced further below.6.e.18) is an eigenfunction of J3 . hence. Ω2 ) = m1 . and J3 . Equivalent coefficients exist for other symmetry properties of multi–component systems. . Exercise 6.18) yields using the property (k) J3 Y k mk (Ωk ) = mk Y k mk (Ωk ) .20) This equation can be satisfied only if the Clebsch-Gordan coefficients satisfy (J M | 1 m1 2 m2 ) = 0 for m1 + m2 = M . These coefficients. J (1) . k = 1.18) where the states YJM ( 1 .18) to avoid summation of vanishing terms YJM ( 1 . The two missing quantum 2 2 numbers are 1 and 2 corresponding to the eigenvalues of J (1) and J (2) . However. (6. play a cardinal role in the mathematical description of microscopic physical systems. therefore. i. Why can states YJM then not be specified by 5 quantum numbers? 2 2 Properties of Clebsch-Gordan Coefficients A few important properties of Clebsch-Gordan coefficients can be derived rather easily. i. two further quantum numbers need to be specified for a complete characterization of the total angular momentum states. J3 YJM = M YJM . 2 |Ω1 .h. J (2) . an important example being the symmetry groups SU(N) governing elementary particles made up of two quarks.1: Clebsch-Gordan Coefficients 145 6.e. m1 .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) . restrict the sum in (6. J (2) . Since YJM sofar specifies solely two quantum numbers. of (6. and three quarks. We. (6. J3 . mesons. (6.22) We will not adopt such explicit summation since it leads to cumbersum notation. assume the expansion YJM ( 1 . We first notice that YJM in (6. i.m2 (J M | 1 m1 2 m2 ) Y 1 m1 (Ω1 ) Y 2 m2 (Ω2 ) (6.1. baryons. Ω2 ) = m1 (J M | 1 m1 2 M − m1 ) Y 1 m1 (Ω1 ) Y 2 m2 (Ω2 ) .

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 . For each quantum number J there should be 2J + 1 elements YJM with M = −J. M = −J. | 1 − 2| + 1. .25) i.23) corresponding to the r. . . J (1) and J (2) . and that the states can be normalized. 1. . 2) . to a new basis B ( 1 . Ω2 ). .s. it holds 1+ 2 ( 2J + 1 ) = (2 J=| 1 − 2 | 1 + 1) (2 2 + 1) . . 1 + 2 .2: Prove Eq. = {YJM ( 1 . J = | 1 − 2 |. (6. . 2 ) and B ( 1 . Our derivation below will also yield real values for the Clebsch-Gordan coefficients. . However. . B2 2) should be 2 |Ω1 . 2 ) is also orthonormal2 and must have the same number of elements. J} . the basis B ( 1 . property implies dΩ1 dΩ2 Y 1 m1 corresponding to the l. −J + 1. Exercise 6. . The property dΩ1 dΩ2 Y∗ ( 1 .146 Theory of Angular Momentum and Spin The expansion (6. − + 1. 2 ) has (2 1 + 1)(2 2 + 1) elements. it is not immediately obvious what the J– values are. J.18) constitutes a change of an orthonormal basis B( 1 .. −J + 1.s.1. the range of values assumed for J is correct. | 1 − 2 | + 1. Since YJM represents the total angular momentum state and Y 1 m1 (Ω1 ) and Y 2 m2 (Ω2 ) the individual angular momenta one may start from one’s classical notion that these states represent angular momentum vectors J.18) applied to Y∗ and to YJ JM (J M | m1 . (6. in fact.27) together with (6. In this case the range of |J|–values would be the interval [ |J (1) | − |J (2) | . (6. |J (1) | + |J (2) |]. m2 = − 2 . . . 1 + 2.25) We want to state now two summation conditions which follow from the orthonormality of the two basis sets B( 1 . 2) = {Y 1 m1 (Ω1 ) Y 2 2 m2 (Ω2 ).m2 2 M and with (6.28) This property follows from the fact that the basis elements are eigenstates of hermitian operators with different eigenvalues.26) We will show below in an explicit construction of the Clebsch-Gordan coefficients that. Ω2 ) = δJJ δM M (6. (6.e. In fact. .h. . Ω2 ) YJ M ( 1. 2 |Ω1 . The basis B ( 1 . This obviously corresponds quantum mechanically to a range of J–values J = | 1 − 2 |. − 2} 1 + 1.24) yields 1 m1 2 m2 ) ∗ 1 m1 2 m2 ) (J M| = δJJ δM M . m1 = − 1 .24) The basis B( 1 .. . respectively. . . .h. . 2 ). . (6. JM 2 |Ω1 . (6.

(6. A similar 2 construction will also be applied to composite systems governed by other symmetry groups. therefore. 2 ) span the same function space.28) allows one to conclude that the coefficients cJ (J M | 1 m1 2 m2 )∗ . 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 ( 1.32) which complements (6..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 . to composite systems involving spin. Ω2 ) and integrating. At this point we like to stress that the construction will be based on the theorems (5.. or J > 1 + 2 (6. 2 |Ω1 . Ω2 ).m2 M ( 1.71–5.18) that the Clebsch-Gordan coefficients obey the second summation condition (J M | 1 m1 2 m2 )∗ (J M | 1 m1 2 m2 ) = δm1 m1 δm2 m2 .2 Construction of Clebsch-Gordan Coefficients We will now construct the Clebsch-Gordan coefficients. One can show readily using the same reasoning as applied in the derivation of (6. i.e.e.28) from (6. 2 |Ω1 .1 states. (6.6. i. (6. 2 ) and B ( 1 . or baryons multiplets involving three quarks.18) by Y∗ J orthogonality property (6.e.30) The latter property follows from multiplying (6. 2 |Ω1 . 1+ 2 J Y 1 m1 (Ω1 )Y 2 m2 (Ω2 ) = J =| 1 − 2 | M =−J cJ M YJ M ( 1. will be based solely on the commutation properties of the operators J and J (k) . Hence.31) are identical to Comparision with (6. e. Ω2 ) . it is possible to expand Y 1 m1 (Ω1 )Y 2 m2 (Ω2 ) in terms of YJM ( 1 . 2 |Ω1 .g. We can. The result of this construction will include all the properties previewed above.27) yields δJJ δM M = m1 . rather the convention (J M | has been assumed. (6.2: Construction of Clebsch-Gordan Coefficients 147 The second summation condition starts from the fact that the basis sets B( 1 . Ω2 ) .33) JM The latter summation has not been restricted explicitly to allowed J–values. also apply the results. the group SU(3) in case of meson multiplets involving two quarks. 1 m1 2 m2 ) = 0 if J < | 1 − 2| .. and actually also the properties of Clebsch-Gordan coefficients stated above.34) 6. The (J M | 1 m1 2 m2 ) cJ M . Ω2 )Y 1 m1 (Ω1 )Y 2 m2 (Ω2 ) . (6.18).81) stated above.. i. .

Ω2 ) 2 |Ω1 . Ω2 ) . 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. J3 YJJ ( 1 . For this purpose we insert (6. 2 |Ω1 .42) ( 1. 1+ 2 ( 1. 2 |Ω1 .18) which satisfies J+ YJJ ( 1 .35) 1 − 2 |. Ω2 ) = Y 1 1 (Ω1 ) Y 2 2 (Ω2 ) . −J + 1. | 1 − 2| + 1. (6.64) J3 = = (1) + J3 (1) 1 (2) Y 1 1 (Ω1 ) Y 2 2 2 2 (Ω2 ) 1 1 J3 Y ( + 1 1 (Ω1 ) Y 1 1 (Ω2 ) + Y 2 2 (Ω1 ) J3 Y (2) 2 2 (Ω2 ) (6.41) 2) Y (Ω1 ) Y (Ω2 ) . we can also demonstrate using (??) that G dΩ1 = We. . M = −J. .36) = 0 = J YJJ ( 1 . .25) using (6.11) and (5.37).37) and replace (1) (2) according to (6. 1+ 2 ( 1. Ω2 ) is normalized dΩ2 G dΩ1 Y 1 1 1+ 2.37) The solution needs to be normalized. (6. J} by applying repeatedly J− YJM +1 ( 1 .11) J+ by J+ + J+ .43) . . Ω2 ) = Y 1 1 (Ω1 ) Y 2 2 (Ω2 ) (6.148 For the construction of YJM we will need the operators J± = J1 + iJ2 . . Ω2 ) = (J + M + 1)(J − M )YJM ( 1 .68) J+ = (1) + J+ (1) (2) Y 1 1 (Ω1 ) Y 2 2 2 2 (Ω2 ) 1 1 (6. Quantum mechanically this corresponds to a state G 1+ 2. . We embark on the suggested construction for the choice J = 1 + 2 . (6.5. Having determined such YJJ we then construct the whole family of functions XJ = {YJM ( 1 . −J. 1+ 2 In fact. .66. 2 |Ω1 . We obtain using (5. 1+ 2. 2 |Ω1 . have shown Y 1+ 2. . (6.38) for M = J − 1. Similarly. J − 2. 2 |Ω1 . 1+ 2 ( 1. Ω2 ) .39) which we will try for a solution of (6. 2 |Ω1 . Ω2 ).40) (Ω1 ) J+ Y (2) 2 2 J+ Y 1 1 (Ω1 ) Y (Ω2 ) + Y (Ω2 ) = 0 . 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 can demonstrate condition (6. . 1 + 2} and for (6. 2 |Ω1 . therefore.39) into (6. Ω2 ) Addition of Angular Momenta and Spin (6. 2 |Ω1 . 2 |Ω1 . Ω2 ) (Ω1 ) dΩ2 Y 2 2 (Ω2 ) = 1. We first seek an unnormalized solution GJJ and later normalize. The construction assumes a particular choice of J ∈ {| such J–value seeks an expansion (6. .

One obtains then Y 1 1+ 2 1 + 2 1 + 2 −1 2 2 ( 1. Exercise 6.s.6. 1 + 2 − 2. M = −( + ). | .37). The result is illustrated for the case 1 = 2. ). with the symmetry pattern and with the terms Y 1 m1 Y 2 m2 contributing to each YJ M . 2 |Ω1 .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 ). According to the condition (6.2: Construction of Clebsch-Gordan Coefficients 149 We now employ property (6. thereby.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. . . . . Show that the resulting functions are orthonormal. . 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. .44) (Ω2 ) . 2 |Ω1 . The r. The reader should familiarize himself with the entries of the Table. all the Clebsch-Gordan coefficients ( 1 + 2 M | 1 m1 2 m2 ). . . 1 + 2 − 1.38) to construct the family of functions B + = {Y + M ( .39) as required by (6. Ω2 ) 2 1+ 2 = (Ω1 )Y 2 2 −1 (6. Ω2) for M = 1 + 2 − 2 and J = 1 + 2 .1: Construct following the procedure above the three functions YJM ( 1 . . | . 6. 2 = 1 in Table 1.45). We like to construct now the family of total angular momentum functions B + − = {Y + − M ( .38) yields pression 2( 1 + 2 )Y 1 + 2 1 + 2 −1 ( 1 . of (6. Normalization yields furthermore Y = 1 1 + 2 −1 1 + 2 −1 We have thereby determined ( 1 2 |Ω1 Ω2 ) (6. −( 1 + 2 − 1).46) 2 1 +c Y (Ω1 ) Y 2 2 (Ω2 ) = 0 . etc. . The l. This expression is orthogonal to (6. One can readily show that (6.2.47) can serve now to obtain recursively the elements of the family B + − for M = 1 + 2 − 2.45) for some unknown constant c.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. ).37) is satisfied. 2 |Ω1 . To demonstrate (6. 1 + 2 − 3. ( + − )}. ( + )}.h.36. . We demonstrate the procedure ex(1) (1) plicitly only for M = 1 + 2 − 1.21) 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. . G 1 + 2 −1 1 + 2 −1 in (6. Expression (6. M = −( + − ). in particular. Ω2 ).s. and all total angular momentum functions for a given choice of 1 and 2 . We seek for this purpose first an unnormalized solution G 1 + 2 −1 1 + 2 −1 of (6.47) (Ω2 ) − 1 1 2 + Y 2 1 1 −1 (Ω1 )Y 2 2 + Y 2 1 1 (Ω1 )Y 2 2 −1 (Ω2 ) . One can obviously continue the construction outlined to determine Y 1 + 2 −2 1 + 2 −2 . To satisfy this equation one needs to choose c = − 1 / 2 . of (6.27). . Having constructed this family we have determined the Clebsch-Gordan coefficients ( 1 + 2 − 1M | 1 m1 2 m2 ).

632456 0.547723 0.730297 −0. Ω2) Y30 (2.408248 −0. 1|Ω1 . Ω2) Y22 (2.547723 0.774597 0 2 5 8 15 1 6 3 10 1 3 0. Ω2) Y10 (2.408248 −0. Ω2) Y2−2 (2.258199 −0.150 Addition of Angular Momenta and Spin Y33 (2.816497 0. 1|Ω1 . Ω2) Y1−1 (2.316228 0. Ω2) Y3−1 (2.707107 0. 1|Ω1 .632456 −0.57735 3 10 3 5 Y2−1 (Ω1 )Y11 (Ω2 ) Y20 (Ω1 )Y10 (Ω2 ) Y21 (Ω1 )Y1−1 (Ω2 ) −0.707107 0. Ω2) Y3−2 (2.816497 −0. 1|Ω1 .707107 0.57735 −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.57735 0. 1|Ω1 .57735 0.774597 − − − − 0. 1|Ω1 .447214 0. 1|Ω1 . Ω2) Y20 (2. 1|Ω1 . Ω2) Y3−3 (2.57735 0.730297 0.258199 0.447214 −0.57735 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 ) 0.707107 0. .547723 0.316228 0.816497 Y2−2 (Ω1 )Y11 (Ω2 ) Y2−1 (Ω1 )Y10 (Ω2 ) Y20 (Ω1 )Y1−1 (Ω2 ) Y2−2 (Ω1 )Y10 (Ω2 ) − 2 3 Y2−1 (Ω1 )Y1−1 (Ω2 ) Y2−2 (Ω1 )Y1−1 (Ω2 ) 1 Table 6.632456 0. 1|Ω1 .774597 0. Ω2) Y2−1 (2. 1|Ω1 . 1|Ω1 . 1|Ω1 . Ω2) Y31 (2. 1|Ω1 . Ω2) Y21 (2. 1|Ω1 . 1|Ω1 .547723 0. Ω2) Y11 (2. Ω2) Y32 (2. 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.

we are dealing now with two particles carrying angular momentum or spin. J.10.6. bζ = a† . (5. m + 1 . Accordingly.9. + 2  − J−M +1 for = J + 1 2J+2 2  J−M  for = J − 1 2J 2 = .2. bζ . 1 . a† ζ .10. 6. commute with each other ζ ζ aζ . b† refer to different particles and. Definition of Spinor Operators for Two Particles In contrast to Sections 5. (6. (6.53) (6. b † ζ = δζζ = δζζ .54) 2 + = √ b† ( 2 +m2 )! √− ( 2 −m2 )! . b† ζ ζ = 0 = 0 . For this purpsose we will employ the spinor operators introduced in Sections 5. 5.49) where ζ. ζ = ± and the matrix elements < ζ |σk | ζ > are as defined in Section 5. aζ .50) (6.10 where we studied single particle angular momentum and spin. 1 (6. bζ = = a† . b† ζ = 0. b† ζ ζ = aζ .ζ ζ. 2 . 5. m − 2 .52) and | 2 m2 2 According to Section 5. − 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† and bζ .254)] the angular momentum / spin eigenstates | 1 m1 of the two particles are | 1 m1 | 2 m2 1 + = √ a† 1 +m1 ( 1 +m1 )! 2 +m2 √− b† a† 1 −m1 ( 1 −m1 )! 2 −m2 |Ψ0 |Ψ0 .10 [cf.287) to two particles (1) J k k = = (2) J 1 2 1 2 ζ. The creation and annihilation operators are again of the boson type with commutation properties aζ . These expressions will be derived now. M | .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. It is actually possible to state explicit expressions for any single coefficient (JM | 1 m1 2 m2 ).9. M | .3: Explicit Expressions of CG–Coefficients 151 Exercise 6.3 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 ). we extent definition (5.ζ a† < ζ |σk | ζ > aζ ζ b† < ζ |σk | ζ > bζ ζ (6.48) (6. aζ bζ .51) The operators aζ . hence. a† ζ ζ b† .

m1 1 | 2 . = = 1( 1 2( 2 2) 2) (6. 2) 2) (6. (2) J 2 . (6. M ( 1 . m1 1 | 2 .300). 2) . m2 |J. we like to recall for future reference that the operators (5. 2 . 2 ) which are simultaneous eigenstates of the operators 2 2 J2 . M ( 1 . M. in analogy to Eqs. (j) 2 ˆ ˆ J = kj ( kj + 1 ) . (1) J 2 |J.69) − 2| ≤ J ≤ + 2 . m2 2 . can be expressed in terms of the number operators 1 † 1 ˆ ˆ a† a+ + a† a− . m1 .60) (6. 2) + 1) |J. M ( 1 . + 1) |J. M ( 1 .59) J± = J1 ± i J2 . 2) We will also require below the raising and lowering operators associated with the total angular momentum operator (6. J3 |J. M ( 2) 2) = J(J + 1) |J. (1) J . (6. M ( 1 . M ( 1 .m2 | 1 . 5.62) .64) (6. M ( 1 .65) (6. as usual are denoted by their respective quantum numbers J.53. 2) 1.. m1 . |J. k = 1. (1) J 2 . (5. m2 |J. M ( (2) J 2 |J.66) (6. 2 )) .302. J3 .55) . M ( 1 . hence. −J ≤ M ≤ J .59) We seek to determine states |J. j = 1. ˆ kj |J. M ( 1 . (6. 1 . 2 . The operator of the total angular momentum/spin of the two particle system is J = with Cartesian components Jk = (1) J + J (6. 2 .67) = j |J. 1. m1 1 | 2 . (6.152 It holds.68) The states |J. . k2 = b b+ + b† b− k1 = + − − 2 2 + namely. M ( 1 . m2 a† + ( 1 2 = a† − ( 1 1 −m1 (6.303).61) (6. 3 . m2 1 2 ( 1 . (1) J (2) J 3 | 1 m1 1 3 | 2 m2 2 = m1 | 1 m1 = m2 | 2 m2 1 2 (6. for such states should hold J2 |J. (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 . i. are | 1 .54). . 2) 2) can be expressed in terms of Clebsch-Gordan coefficients (6. mj j = j | j .e.58) Jk + (2) Jk . which describe a two particle system according to (6.56) The states | 1 . M ( 1 . 2. 2 )). The aim of the present Section is to determine closed expressions for the Clebsch–Gordan coefficents ( 1 .18) as follows = | 1 m1 . M ( 1 . mj j and. ˆ For the operators kj holds ˆ kj | j . = M |J.57) b† + ( (2) 2 2 +m2 1 +m1 b† − ( 2 2 −m2 + m1 )! − m1 )! (1) + m2 )! − m2 )! |Ψo . 2 . M ( 1 . 6. (2) J which. (6.63) according to At this point.

K † (j) J 2 . For example. 6. the relationships (6. and applying the relationships (6.73. K † = 0. 2 4 = 0 = = 0. 4 Using J3 = (1) J 3 + (2) J 3 . K † + K † 2 2 3 † †ˆ K kj + K . Employing (6. b† = 0 . due to J2 = K† 1 2 J+ J− 1 † K .73) (6.50.73) yields J3 . b† − b† a† b− . (6. K † 2 1ˆ ˆ 1 ˆ ˆ kj kj + 1. a† b† − a† a− . We note that.71–6. 6. K † ] = (j) J 2 . K † (kj + 1) 2 2 1ˆ † 1 ˆ kj K + K † (kj + 1) 2 2 1 ˆ 1 ˆ K † kj + kj .71) one can show K† = = = = = 1 ˆ ˆ kj (kj + 1). K † + kj . using (6. a† b† − a† b† + − + − + − − + 2 1 1 = a† a+ .64.74) + 1 J− J+ + J2 . j = 1. expressing (k) J 3 through the creation and annihilation operators according to (5. (6.72) (6.65) and (6. a† b† + − − + 2 + 2 − 1 1 − b† a† b+ .6.75) The relationships (6.288). K† J3 . j = 1.71–6.3: Explicit Expressions of CG–Coefficients The Operator K † The following operator K † = a† b† − a† b† . + − − + 2 2 = 1 † 2K .51) one obtains ˆ k1 .73) can be readily proven. + − + − 2 2 − + = .74) imply 3 2 J2 . a† b† − a† b† − + − − + 2 + 1 1 = a† a+ . ˆ A similar calculation yields [k2 . a† b† + a† a− . K † J± . a† b† + + − − + 2 2 − 1 1 = a† b† − a† b† = K† . This operator obeys the following commutation relationships with the other pertinent angular momentum / spin operators ˆ kj . + − − + 153 (6.71) (6.70) will play a crucial role in the evaluation of the Clebsch-Gordan-Coefficients. K † 1 a† a+ − a† a− + b† b+ − b† b− . K † 1 † a a+ + a† a− . 6. 2 2 3 ˆ = K † kj + K † .

2) 2) ˆ K † kj + = K† = j 3 † K + K † j ( j + 1) |J. M ( 1 + n .78) a state which has been used already in the construction of Clebsch-Gordan coefficients in Section 6. K† + K † (j) J 2 |J. j2 + ) 2 2 n n = N (n. j1 + ) K † 2 2 (6. M ( 1 . 2 2 + n ) 2 (6. 2 ) j + 4 1 3 + K † |J. a† b− − a† b† + + + − + = 0. M ( 1 .67) (j) J 2 K † |J. 6. but for different 1 and 2 quantum numbers of the operators (1) J 2 and (2) J 2 . it holds K † |J. One can generalize property (6. 1 2 + 2 . j1 1 + . 6. j1 + j2 (j1 . The commutation properties (6. j + 2 2 2) However. j2 ) .292) one can derive similarly J+ . M ( 1. K † ] = 0 is demonstrated in an analoguous way. M ( 1 . j2 2 . i. K † = Addition of Angular Momentum and Spin a† a− + b† b− .76) and state K† n |J. on the state |j1 + j2 . M ( 1 .75) ascertain that under the action of K † the states |J. 4 3 + j ( j + 1) |J. j1 + j2 (j1 . 2) is a state with quantum numbers 1 1 + 1 2 and = N |J. 2 ) remain eigenstates of J2 and J3 with the same quantum numbers. b† + − + + + − The property [J− .154 Starting from (5. M ( 1 . Application of (K † )n to this state yields. = − a† a− . a† b† + b† a† b− . 6. |j1 + j2 . according to (6.2. 2 ) . 2 ) produces again total angular momentum eigenstates to the same J and M quantum numbers of J2 and J3 .77). (6.77) where N is another normalization constant. Action of K † on the states |J. M ( 1 . M ( 1 . j1 1 |j2 . 2 2 1 + ) .72) and (6. To demonstrate that the resulting states are eigenstates of (1) J 2 and (2) J 2 we exploit (6. j2 ) = |j1 . M ( 1 + . 2) = N |J. We consider now the case that (K † )n acts on the simplest total angular momentum / spin state. this result implies that K † |J. namely.79) n |j1 + j2 .63.76) Here N is an unknown normalization constant.73.e. M ( 1 . 2) = = (j) 2 J . M ( 1 . 2 (6. M ( 1 .62. j1 + j2 (j1 + n n . 2) We want to demonstrate now that the action of K † on the states |J..

− 1. 2) = ∆(J. (6. For this purpose one needs to choose in (6.s.84) with (6.3: Explicit Expressions of CG–Coefficients 155 where we denoted the associated normalization constant by N (n. 2) . 6.104.81) = N( 1 + 2 1 2 − J. allow one to obtain the states |J.85) ∆(J.h. 1 = j1 + n 2 . M ) (J + × K† 1 − 2 )! (J + 2 − 1 )! J+ 1 − 1 + 2 −J (J− )J−M a† + × 2 (6. 5.83) The derivation of this expression will be provided further below (see page 158 ff).82.6.74)] yields |J. M ( 1 . j2 as follows J = j1 + j2 which is equivalent to n j1 j2 Accodingly. J( 1 . 1 . 2 ) for −J ≤ M ≤ J as follows |J. j1 . M ) . M ( 1 . The latter states. M ) (J− )J−M |J. 2) = N( 1 + 2 − J.82) for |J. 2 ) can be expressed through the r.84) (6. j1 + n . 2) 1 2 = ( 1 + 2 (2J + 1)! − J)! ( 1 + 2 + J + 1)! 1 2 . (6. holds |J. 2 ) exploits the expression (6. M ( 1 . 2 2 It is now important to notice that any state of the type |J. (6. Combining (6. J( 1 . j2 + n ). 2 = j2 + n 2 (6.86) b† + J+ 2 − 1 |Ψo .79). 2 ) ∆(J. 2) Our construction of the states |J.79) n. in analogy to the construction (5. 1. 2) 2) . (J + 2 − − 2) 1 1) 2 The normalization constant appearing here is actually N( + − J. K† 1 2 1 + 2 −J × × .80) = = = 1 1 2 2 − J 1 n − = (J + 2 2 n 1 − = (J + 2 2 + 1 1 − − 2) 2) . M ( 1 . = (J + M )! (2J)! (J − M )! 1 2 2) (6.105) of the spherical harmonics.57) and exploiting the fact that J− and K † commute [c. 2 ) . J( 1 .f. Strategy for Generating the States |J. of (6. J( 1 .82) 1 × | (J + 2 1 × | (J + 2 1 (J + 2 1 − 1 ) . (6. M ( 1 .

x.88).86) in terms of monomials 1 +m1 1 −m1 2 +m2 2 −m2 a† a† b† b† |Ψo . m2 Gordan-coefficients.90) (6.52). y) = λr xs y t r |Gst .87) + − + − i. x. b† ] = 0 [cf.87). in terms of | 1 . follows from the commutation properties (6. + − + − .69) yields then the Clebsch- Expansion of an Intermediate State We first consider the expansion of the following factor appearing in (6. y) is a generating function for the states |Gr defined in (6. a† ] = [b+ .. st The desired expansion of |Gr can be obtained from an alternate evaluation of I(λ. in analogy to (5.91) which.92) − − and noting [a+ .88) in terms of monomials (6. One obtains then using J− = a† a+ + b† b+ (6. Comparision with (6. r!s!t! r. (6.s.156 Addition of Angular Momentum and Spin Our strategy for the evaluation of the Clebsch-Gordan-coefficients is to expand (6. y) = exp (λJ− ) exp x a† exp x b† |Ψo .50)] − − exp ( λJ− ) f (a† ) g(b† ) |Ψo + + = exp a† a+ − = u f (a† ) exp b† b+ + u g(b† ) |Ψo + λu u! a† a+ − λv v! u f (a† ) × + v × v b† b+ − ∂ g(b† ) |Ψo + u = u λ a† − u! × v λ b† − v! ∂a† + v ∂ ∂b† + f (a† ) × + v g(b† ) |Ψo + (6. + + Taylor expansion of the two exponential operators yields immediately I(λ.e.86) |Gr = Jr a† st − + s b† + t |Ψo (6. 2 [cf.t (6. x..e.93) = f (a† + λ a† ) g(b† + λ b† ) |Ψo . (6. (6. 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. m1 1 | 2 .50–6. For this purpose we introduce the generating function I(λ.89) i. x.264). I(λ.57)].

97). y) = exp a† + λ a† + − exp b† + λ b† + − |Ψo . 157 (6. y) = s.3: Explicit Expressions of CG–Coefficients We conclude I(λ.70) holds K† 1 + 2 −J = s 1 + 2 −J b† − s 1 + 2 −J−s (−1)s 1 + 2 −J−s (6.6.96) and. Operation of this operator on (6. one can write the right factor in (6. With K † given by (6. 2 ) in terms of states | 1 .t × a† + s−q s q b† + t r−q t−r+q a† − q b† − |Ψo (6.97) + q)! |Ψo × (J − M )!(J + 1 − 2 )!(J + 2 − 1 )! q!(J + 1 − 2 − q)!(J − M − q)!(M + 2 − J+ 1 − 2 −q × Final Result a† + a† − q b† + M + 2 − 1 +q b† − J−M −q Our last step is to apply the operator (K † ) 1 + 2 −J to expression (6. using the definition (6. M ( 1 .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. m2 2 . to obtain the desired expansion of |J. x. |J.t × λ v b† − × r−q v × = t s−u v λu a† − r! q |Ψo λr xs y t r! s! t! r. m1 1 | 2 .94) yields I(λ. Expanding the exponentials in st (6. M ( 1 .94) One can infer from this result the desired expressions for |Gr .95) Comparision with (6.s.50).97) yields.t = q r! × a† + s q s−q t r−q a† − q × b† + t−r+q b† − r−q |Ψo (6. x.90) allows one to infer |Gr s.86) (J− )J−M a† + = q J+ 1 − 2 b† + J+ 2 − 1 |Ψo 1 (6.98) a† − s a† + b† + s .88).q (−1)s (6.99) . 2) = s. using the commutation property (6.

6. j2 . 1 . m2 = M − 1 + q + s. 2 ) we consider the scalar product J.83) of the normalization constant N ( 1 + defined through (6. M ( 1 . 1 . Using (6.69) if one identifies m1 = 1 −q − s . m1 . The Clebsch-Gordan coefficents are then finally ( 1 .69) since. − 1 .q (−1)s × (6. 2 ) (6. 2 ) ∆(J.82) and (6. m2 |J. − q − s)!(q + s)! 2 × (M + 1 − 1 + q + s)!( 1 1 + 2 − M − q − s)! 2 − q − s 1 | 2. × 2) = N( 1 + 2 1 − J.103) = 1.103) this can be written 1 = N 2 ψ(j1 . J( 1 . The summation over q corresponds then to the summation over m1 . 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. m2 in (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. M − +q+s One can conclude that this expression reproduces (6.102) We want to determine now the expression (6. (6.54) between creation operator monomials and angular momentum states allow one to write this |J. (6.82). n = 1 + 2 −J. j2 .101). J( 1 . q = 1 − m1 − s and m2 = M − m1 . j2 = 1 (J + 2 2 − 1) . 2 ) − 2) . according to (6. n)|ψ(j1 .104) . √ 2J + 1 1 2 . M ) (J + 2 )! (J + 2 − 1 )! s.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 . n) 2 )|J.101) Note that m1 + m2 = M holds.53. To determine N = N ( 1 + 2 −J. For this purpose we introduce j1 = 1 (J + 2 1 2 − J.

110) which can be written 1 1−λ n1 +n2 +2 = r s n1 + r − s n1 n2 + s n2 λr (6. (6. m1 1 |j2 . One obtains then. 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. j2 .57) for these states and the expression (6.70) for the operator K † . j1 1 |j2 .3: Explicit Expressions of CG–Coefficients where |ψ(j1 . applying (6.105) The first step of our calculation is the expansion of ψ(j1 .108). 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 . Accordingly.108) twice. n) in terms of states |j1 . (6. we obtain |ψ(j1 .109) and Taylor expansion of the left hand side of (6. m2 2 .104) 1 = N2 (n!)2 (2j1 )!(2j2 )! (2j1 + n − s)!(2j2 + s)! s!(n − s)! 2j2 + s 2j2 (6.108) yields the identity n1 + r − s n1 n2 + s n2 = n1 + n2 + r + 1 n1 + n2 + 1 .112) s .111) Comparision with (6. j2 2 .108) = λ=0 (n1 + ν)! n1 ! (6.6. n) = K† n 159 |j1 . 1 1−λ n1 +1 1 1−λ n2 +1 = m1 . (6.m2 n1 + m1 n1 n2 + m2 n2 λm1 +m2 (6.106) The orthonormality of the states occurring in the last expression allows one to write (6. j2 . We employ the expression (6.

M ) corresponding to (6.116) If one takes the negatives of the operators in (6. m2 ) by ( 1 . J (1) on the r. For the quantum mechanical addition of angular momenta the Clebsch Gordan coefficients ( 1 .114) show a simple relationship to the Clebsch Gordan coefficients ( 2 . m1 . of (6. one can interchange also the operator J on the l. m1 ) . m2 .160 Applying this to (6.117) are again related in a simple 2 . J (1) . m2 .s. manner to the coefficients ( 1 . M ) .113) Using the identities (6. To derive relationship (6. 2 .118) Finally. −m2 . 2 .114) For example.114) by. 1 .115). and (6. m1 . namely.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 .118). of this equation J (1) = J (2) − J . 2 . m1 .4 Symmetries of the Clebsch-Gordan Coefficients The Clebsch-Gordan coefficients obey symmetry properties which reflect geometrical aspects of the operator relationship (6.116) through formula (6. J. M ) corresponding to (6. and .h.g. 2 .115) This relationship is a trivial consequence of (6.102) of the Clebsch-Gordan coefficients.h.114) as long as J. −m1 . 1 . and J (2) are vectors in R . We will demonstrate this now.103) one obtains the desired result (6.120) The symmetry properties (6. vice versa. m2 |J. (6.s. M ) . m2 ) and. m1 |J.119) The corresponding symmetry property of the Clebsch-Gordan coefficients is ( 1 . m1 .h. −m2 |J. ( 1 . 6.120) can be readily derived from the expression (6. m1 ) by ( 2 .116) one expresses the Clebsch-Gordan coefficient on the r. 2 . M ) = (−1) 2 +m2 2J + 1 ( 2 . M ) = (−1) 1 + 2 −J ( 2 . 2 . M | 1 . −M ) = (−1) 1 + 2 −J ( 1 . 2 1+1 (6. (6. (6. m1 |J.83). m1 . M ) ( 1 .114) one obtains − J = − J (1) − J (2) The respective Clebsch-Gordan coefficients ( 1 . −m1 . ( 2 . −M ) . m2 |J. m2 |J.102) by replacing ( 1 .. −m1 2|J. (6. of (6. m2 |J.s. (2j1 + 2j2 + 1)! (6.11) J = J (1) + J (2) . m1 ). e. interchanging the operators J (1) and J (2) results in J = J (2) + J (1) . (6. (6. m2 |J. (6.116).

6.4: Symmetries of the Clebsch-Gordan Coefficients

161

seeks then to relate the resulting expression to the original expression (6.102) to prove identity with the l.h.s. Inspecting (6.102) one recognizes that only the sum S( 1 , m1 , ×
2 , m2 |J, M )

=
s

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

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

+ m1 + s)!

(6.121)

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

= (−1)

1 + 2 −J

S( 2 , m2 ,

1 , m1 |J, M )

.

(6.122)

To prove this we note that S on the r.h.s. reads, according to (6.121), S( 2 , m2 ,
1 , m1 |J, M )

=
s

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

×

(

1+

2 − J − s)!(J −

+ m2 + s)!

.

(6.123)

Introducing the new summation index s = and using the equivalent relationships s =
1 1

+

2

−J −s

(6.124)

+

2

− J − s ,

−s = J − −

1

2

+ s

(6.125)

to express s in terms of s in (6.123) one obtains S( 2 , m2 , (−1)
1 , m1 |J, M )
1

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

+

2

−J s

+ m1 + s )! (6.126)

×

s !(

1 − m1 − s )!( 1

+ m2 − s )!

.

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

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

162

Theory of Angular Momentum and Spin

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

= (−1)

1 + 2 −J

S( 1 , −m1 ,

2 , −m1 2|J, −M )

.

(6.128)

We note that according to (6.121) holds S( 1 , −m1 , ×
2 , −m1 2|J, −M ) = s

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

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

− m1 + s)!

.

(6.129)

Introducing the new summation index s as defined in (6.124) and using the relationships (6.125) to replace, in (6.129), s by s one obtains S( 1 , −m1 , (−1)
2 , −m2 |J, −M )

=
1

1 + 2 −J

s

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

− m2 + s )! (6.130)

×

1 s !( 2 + m2 − s )!(

− m1 − s )!

.

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

= (−1)

2 +m2

S( 2 , −m2 , J, M | 1 , m1 ) .

(6.131)

(−1)s s!( 2 + m2 − s)!(J + M − s)!
1

×

1 ( 2 + J − 1 − s)!( 1 − 2 − M + s)!(

− J − m2 + s)!

.

(6.132)

Introducing the new summation index s = and, using the equivalent relationships s =
2 2

+ m2 − s

(6.133)

+ m2 − s ,

−s = −

2

− m2 + s

(6.134)

6.5: Spin–Orbital Angular Momentum States to replace s by s in (6.132), one obtains S( 1 , −m1 , (−1)
2 , −m1 2|J, −M )

163

=
2

2 +m2

s

(−1)−s ( 2 + m2 − s )!s !(J −

+ m1 + s )!
1

×

1 (J − 1 − m2 + s )!( 1 − m1 − s )!(

+

2

− J − s )!

.

(6.135)

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

6.5

Example: Spin–Orbital Angular Momentum States

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

m ,σ

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

(tot)

2

σ

(6.138)

space C χ1 1 =
2 2

1 0

,

χ1−1 =
2 2

0 1

.

(6.139)

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

(6.140)

+ ( , m + 1 , − 1 , 1 |j, m) Y 2 2 2

r m+ 1 (ˆ) 2

164 or Yjm ( , |ˆ) = r
1 2

Addition of Angular Momentum and Spin

( , m − 1 , 1 , 1 |j, m) Y 2 2 2

r m− 1 (ˆ) 2 r m+ 1 (ˆ) 2

1 ( , m + 1 , − 1 , 2 |j, m) Y 2 2

.

(6.142)

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

=

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

(6.143)

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

=

(6.144)

=

(6.145)

=

(6.146)

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

Eigenvalues For the states (6.147, 6.148) holds (J + S)2 Yjm (j (J + S)3 Yjm (j J Yjm (j S 2 Yjm (j Furthermore, using (J (tot) )2 = ( J + S )2 = J 2 + S 2 + 2J · S or, equivalently, 2J · S = (J (tot) )2 − J 2 − S 2 (6.154) (6.153)
2
1 2 1 2 1 2

, 1 |ˆ) r 2 , 1 |ˆ) r 2 , |ˆ) r
1 2

= = =

2

j(j + 1) Yjm (j
1 2

1 2

, 1 |ˆ) r 2

(6.149) (6.150) (6.151)
1 2

m Yjm (j
2

, 1 |ˆ) r 2
1 2

(j
2

1 2

)(j
1 2

+ 1) Yjm (j

, |ˆ) r
1 2

1 2

, 1 |ˆ) r 2

=

3 4

Yjm (j

, 1 |ˆ) r 2

(6.152)

6.5: Spin–Orbital Angular Momentum States

165

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

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

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

2

(6.156)

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

sin θdθ
−π 0

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

(6.157)

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

j = 1, 2 assumes the block-diagonal form

3 2

. . . ; m = −j, −j + 1, . . . + j }

(6.159)

0 −1  −1 0   0 −1 σ·r =  ˆ  −1 0 

       (6.160)

..

.

166

Addition of Angular Momentum and Spin

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

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

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

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

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

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

(6.167)

One can carry out this algebra in a similar way for the k = 2, 3 and, hence, prove (6.161). (tot) Since the differential operators in Jk do not contain derivatives with respect to r, the property (6.161) applies also to σ · r, i.e., it holds ˆ [Jk From this follows J (tot) = =
2
1 σ · r Yjm (j ± 2 , 1 |ˆ) ˆ r 2

(tot)

, σ · r] = 0 , ˆ

k = 1, 2 3 .

(6.168)

σ · r J (tot) ˆ
2

2

1 Yjm (j ± 2 , 1 |ˆ) r 2

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

(6.169)

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

(6.170)

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

6.5: Spin–Orbital Angular Momentum States

167

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

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

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

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

(6.174)

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

(6.175)

(6.176)

We note that Yjm (j + 1 , 1 |ˆ) and Yjm (j − 1 , 1 |ˆ) have opposite parity. Since σ · r has odd parity, r r ˆ 2 2 2 2 i.e., when applied to the states Yjm (j ± 1 , 1 |ˆ) changes their parity, we can conclude α++ (j) = r 2 2 α−− (j) = 0. The operator σ· r in the two-dimensional subspace spanned by Yjm (j ± 1 , 1 |ˆ) assumes ˆ r 2 2 then the form 0 α+− (j) σ·r = ˆ . (6.177) α−+ (j) 0
∗ Since σ · r must be a hermitean operator it must hold α−+ (j) = α+− (j). ˆ According to (5.230) one obtains ( σ · r )2 = 1 . ˆ 1

(6.178)

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

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

1 |θ = 0. identified the sign of (6. 1 |θ. 1 |ˆ) = ((σ · p f (r)))Yjm (j ± 1 .224)] 1 0 σ·r = ˆ . φ) 2 2 2 = − j+ 1 2 4π (6.160).147.1 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 . According to (5. (6. 1 |ˆ) + f (r) σ · p Yjm (j ± 1 . ˆ (6. 1 |ˆ) r r r 2 2 2 2 2 2 (6. 2 |ˆ) and Yj 1 (j + 1 . ˆ 2 2 2 2 2 2 for θ = 0. have proven (6. φ) 2 2 2 .183) in case θ = 0 becomes in the standard representation with respect to the spin. φ) = − Yj 1 (j + 1 .181) 0 = j+ 1 2 4π Yj 1 (j + 1 .185) We have. for θ = 0 (6.180) where the sign could depend on j.182) 0 Since σ · r. 1 |ˆ) r 2 (6. φ) .186) Here (( · · · )) denotes again confinement of the diffusion operator ∂r to within the double bracket. given by ˆ σ · r = σ1 sin θ cos φ + σ2 sin θ sin φ + σ3 cos θ .187) 1 2 .168 Addition of Angular Momentum and Spin 1 Yjm (j ± 2 .174–5.182) σ · r Yj 1 (j − 1 . The result can also be stated in the compact form σ · r Yjm (j ± 1 . φ) and (5. 1 |θ = 0.188) . 1 |ˆ) at θ = 0 are r r 2 2 2 2 Yj 1 (j − 1 .181.184) 0 1 space χ 1±1 2 2 one can conclude from (6. We want to demonstrate finally that the “−”-sign holds in (6. ˆ (6. therefore.180). 1 |θ = 0. For this purpose we consider the application of σ · r in the case of θ = 0.f. Since f (r) is independent of θ and φ follows ˆ ((σ · p f (r))) = −i (( ∂r f (r) )) σ · r . (5.180) and. hence. 6.148) the particular ˆ 1 1 states Yj 1 (j − 2 . 1 |ˆ).180) and (6. 6. One can conclude then ˆ σ·r = ± ˆ 0 1 1 0 (6. (6. 1 |θ = 0.177) one notes that for φ = 0 the spin-angular momentum states 2 are entirely real such that σ · r must be real as well. We want to determine ˆ its action on the wave functions f (r) Yjm (j ± 1 . 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.

∂3 (1/r) = −x3 /r3 and ∂2 x3 = x3 ∂2 . 1 |ˆ) = i [ ∂r + r + ] f (r) Yjm (j 2 r r Using (6.198) .187) one obtains ˆ σ · p f (r) Yjm (j ± 1 . h r (6. r 2 (6.g. 1 |ˆ) r 2 (6. To demonstrate this we consider one of its cartision components.194) ˆ ˆ where J1 is defined in (5.186) for both terms in (6.190) can be related to the angular momentum operator.191) (6..155. ∂3 x2 = x2 ∂3 we obtain ˆ ˆ ( p × r )1 h = −i 1 1 (x3 ∂2 − x2 ∂3 ) h = − J1 h r r (6.53).196) = i j+ ∂r + r 1 2 3 2 1 f (r) Yjm (j − 2 . 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. 1 |ˆ) = 0 and σ = 2S/ .6. Corresponding results are obtained for the other components of p × r and. 1 |ˆ) r 2 = i ∂r + −j r 1 g(r) Yjm (j + 2 .230) allows one to express ˆ ˆ ˆ ˆ ˆ σ · p σ · r = p · r + i σ · (p × r) . we conclude the intuitively expected identity 1 ˆ ˆ p×r = − J .192) ˆ ˆ The operator p × r in (6.193) r r 1 1 1 = −i (∂2 x3 − ∂3 x2 ) h − i h (x3 ∂2 − x2 ∂3 ) . using ∂r Yjm (j ± 1 .195) .5: Spin–Orbital Angular Momentum States Using (6.189) The celebrated property of the Pauli matrices (5. r Altogether we obtain.156) this yields finally 1 ˆ σ · p 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) r 2 1 2 (6. e.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. r 2 2 1 J ·S 2 1 ˆ σ · p f (r) Yjm (j ± 2 . r r r Using ∂2 (1/r) = −x2 /r3 . 1 |ˆ) r 2 (6. 1 |ˆ) . 6.197) 1 ˆ σ · p g(r) Yjm (j − 2 . 1 |ˆ) r 2 (6.

45).200) = (j + 1 )(j + 3 ). . 1 |ˆ) r 2 2 r r2 r2 j 2 + 2j + 3 2 2 2 4 = [ −∂r − ∂r + ] f (r) Yjm (j + 1 . consider first the case of the largest m-value m = j.43).143. 6. 1 |ˆ) = r 2 2 which agrees with (6. 1 |ˆ) r 2 2 The Clebsch-Gordan coefficients are then (j − 1 .199) We want to show that eqs.101).202) Yjj (j − 1 . 1 |j.99) holds ˆ (σ · p)2 = − 2 2 = ∂2 J2 r + 2 . adopting the method in Sec. We note ˆ (σ · p)2 f (r) Yjm (j + 1 .204) (6.151). 1 |ˆ) = J 2 Yjm (j + 1 . using (5. We begin with the coefficients (6. r 2 2 2 2 2 2 − r 2 ∂r r + J2 r2 f (r) Yjm (j + 1 .144) for m = j. j + . 6. r (6.2. 1 .206) . j − 1 . 6. 1 |ˆ) r 2 2 r r2 and.143.198). For m = j − 1 one can state.144) and.203) (6.199).146). j 2 + 2j + 2 3 4 (6. 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 . 2 2 (6. are consistent with this identity.170 Addition of Angular Momentum and Spin To demonstrate the validity of this key result we note that according to (5. j) 2 2 2 2 (j − . 1 |ˆ) r r 2 2 2 2 2 2 2 yields ˆ (σ · p)2 f (r) Yjm (j + 1 . For this purpose we use the construction method introduced in Sec. 5.197. 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. Yjj−1 (j − 1 . (6..230. In this case holds. j) 1 2 1 2 1 2 1 2 = = 1 0 (6. r ∂r2 r 2 (6. 1 |ˆ) r 2 2 i 3 ˆ = σ · p [ i∂r + (j + ) ] f (r) Yjm (j − 1 . 6. as well as (6.205) which agrees with the expressions (6.2. 1 |ˆ) = Yj− 1 j− 1 (ˆ) χ 1 1 .e. i. 6. according to (6. in fact. − |j. Evaluation of Relevant Clebsch-Gordan Coefficients We want to determine now the Clebsch-Gordan coefficients (6.201) 1 3 (j + )(j + ) Yjm (j + 1 .143–6. 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 . according to (6.

m − 3 . 2 |j.2.207) (6.210) to (6.144) for the Clebsch-Gordan coefficients by induction. (6. 6. 2 2 2 2 2j . 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 . is obtained by applying the operator [c.202). 6. 1 .206). 6. 6.143.143. Let us verify then the expression (6. j − 1) 2 2 2 171 = = 2j − 1 2j 1 2j (6. and J− + S− to the r.211) to the l. 1 |j. (6. m − 1) 2 2 2 2 = j +m−1 2j (6. j − 1 .f. [c.. 2 2 2 2 2j (6.210)] yields (j + m)(j − m + 1) Yjm−1 (j − 1 .6.h. 1 .208) (6.212) 1 + 2j(j − m + 1) 1 2j(j − m + 1) Yj− 1 m− 1 χ 1 − 1 2 2 2 2 j+m−1 Yj− 1 m− 3 χ 1 1 + 2 2 2 2 2j j−m+1 Yj− 1 m− 1 χ 1 − 1 . j − 1) 2 2 2 2 which again agrees with the expressions (6.143.h. (· · · |jj − 3).5: Spin–Orbital Angular Momentum States The corresponding Clebsch-Gordan coefficients are then 1 (j − 1 .137)] J− (tot) = J− + S− (6. The further Clebsch-Gordan coefficients (· · · |jj − 2). Expression (6.210). 1 .209) (j − 1 . − 1 |j. etc. 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 .144) implies for j = m Yjm (j − 1 . j − 3 .f. (6. as described in Sec.144) for m = j − 1.s.s. 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 . are obtained by iterating the application of (6.

172 Addition of Angular Momentum and Spin j −m+1 2j (j − 1 . j+m+1 . proven (6.143. (1) (2) (−tot) Obviously. 6.6 The 3j–Coefficients The Clebsch-Gordan coefficients describe the quantum mechanical equivalent of the addition of two (1) (2) classical angular momentum vectors Jclass and Jclass to obtain the total angular momentum vector (tot) (1) (2) (1) (2) Jclass = Jclass + Jclass . m2 | 3 . m. to reflect in the notation the symmetry of these quantities. 6. We have.214) yields (j. m − 1 .216) 6. j2 .144) by induction. 1 . 2 . j2 m2 ) 2 . 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 . the relationship (6. all three vectors Jclass . 1 . 1 . 1 . M.215). 2j + 2 (6.146) can be obtained from (6.143.116. m3 ) = (−1) × 2 + 3 − 1 + 2 +m2 × 2 .217) 2j + 2 (j + 1 . m. one can obtain expression (6. m) = 2 2 2 2 Similarly. m + 1 ) = 2 2 2 j+m+1 . 1 |j + 1 . −m2 | 1 . m + 1 . m3 . 6.215) which follows from (6. m) 2 2 2 2 2j + 1 (6. m + 1 ) = 2 2 2 2 and. Jclass and Jclass play equivalent roles. j3 .143) by applying the symmetry relationships (6. m1 ) (6. m2 . one obtains (j + 1 .213) which is in agreement with (6. m1 . 1 |j + 1 . by the set j1 . − 1 |j. However.144) for j = m − 1.120). m3 . − 1 |j. m2 1 (6. using (6. hence. The Clebsch-Gordan coefficients (6. 2 . m + 1 .214) 2 2 +1 ( 3 . 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 .218) where we have replaced the quantum numbers J. 1+1 3 For 1 (j. 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. m − 1) 2 2 2 2 = (6. m1 . (1) (2) (−tot) The coefficients which are the quantum mechanical equivalent to Jclass + Jclass + Jclass = 0 are the 3j–coefficients introduced by Wigner. − 1 |j.144).143).145) from (6. m1 . 1 . In this context Jclass and Jclass play the same role. The latter relationships applied together read ( 1 . 2j + 1 (6.

In this way the values of 12 3j-coefficients are related. To represent the full symmetry one expresses the 3j–coefficients through a 3 × 3–matrix. However. The Regge symbol reflects a remarkable degree of symmetry of the related 3j–coefficients: One can exchange rows.6: The 3j–Coefficients We first like to point out that conditions (6. states that j1 . are identical to the integer arguments which enter the analytical expression (6. m3 and with respect to sign reversals of all three values m1 .223) (Σ + 1)!n11 !n22 !n33 !n23 !n32 !)) n=0 n31 n32 n33 . i. These symmetry operations relate altogether 72 3j–coefficients. for which no known classical analogue exists. m3 .218) relating 3jcoefficients and Clebsch-Gordan coefficients. The two integer entries J − 1 − m2 and J − 2 + m1 in (6.221) j1 j2 j3 −m1 −m2 −m3 where the the results of a cyclic. j2 . discovered by Regge.g. there exist even further symmetry properties. m2 and j3 . The reader may note that the entries of the Regge-symbol.. 6. According to the definition of the 3j–coefficients one would expect symmetry properties with respect to exchange of j1 . For this purpose one can use the symmetry transformations to place the smallest element into the upper left corner of the Regge symbol. m2 . e.220) The latter condition |j1 − j2 | ≤ j3 ≤ j1 + j2 . 173 (6. one can exchange columns and one can reflect at the diagonal (transposition). In case of a non-cyclic exchange of rows and columns the 3j–coefficent assumes a prefactor (−1)Σ . except when all elements are non-negative integers and each row and column has the same integer value Σ = j1 + j2 + j3 . with respect to alltogether 12 symmetry operations.e. 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. anti-cyclic exchange and of a sign reversal are stated. j2 .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.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 .21. m1 .219) (6. 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. √ safe for the prefactor 2j3 + 1 which is cancelled according to the definition (6.6. −j1 + j2 + j3 . =  (6. the Regge-symbol. j3 form the sides of a triangle and the condition is symmetric in the three quantum numbers. the so-called triangle condition. Because of its high degree of symmetry the Regge symbol is very suited for numerical evaluations of the 3j–coefficents. as follows   −j1 + j2 + j3 j1 − j2 + j3 j1 + j2 − j3 j1 j2 j3 j1 − m1 j2 − m2 j3 − m3  . The Regge symbol also vanishes in case that two rows or columns are identical and Σ is an odd integer.102) of the Clebsch-Gordan coefficients.

225) sn−1 .227) (2m|2m1 1m2 ) = (−1)m+m1 (m + 2m2 ) δ(m.224) (6.10752786393409395427444450130056540562826159542886 We also illustrate the numerical values of a sequence of 3j-coefficients in Figure 6.1.232) .229) 1 1 (0m| m1 m2 ) = 2 2 i(−1)1−m2 δ(0. (6. m) δ(−m1 . m2 ) √ 2 1 ≥ |m1 |) 2 (6.223-6. (6.226) We like to state finally a few explicit analytical expressions for Clebsch-Gordan coefficients which were actually obtained using (6. m1 + m2 ) (1 − m)! (1 + m)! √ 1 1 1 1 6 2 − m1 ! 2 + m1 ! 2 − m2 ! 2 + m2 ! 1 1 ≥ |m1 | ∧ ≥ |m2 | ∧ 1 ≥ |m| 2 2 (6. m1 + m2 ) (3 − m)! (3 + m)! (3m|2m1 1m2 ) = √ 105 (2 − m1 )! (2 + m1 )! (1 − m2 )! (1 + m2 )! 1 ≥ |m2 |) ∧ 2 ≥ |m1 | ∧ 3 ≥ |m| (6.174 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) (6.230) 1 1 (1m| m1 m2 ) = 2 2 √ (−1)2m1 −2m2 3 δ(m.226) by means of a symbolic manipulation package (Mathematica) (1m|2m1 1m2 ) = (−1)1+m+m1 δ(m.228) √ (−1)2m1 −2m2 7 δ(m. m1 + m2 ) (2 − m1 )! (2 + m1 )! √ 6 (2 − m)! (2 + m)! (1 − m2 )! (1 + m2 )! 1 ≥ |m2 | ∧ 2 ≥ |m| ∧ 2 ≥ |m1 | (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. m1 + m2 ) (2 − m1 )! (2 + m1 )! √ 10 (1 − m)! (1 + m)! (1 − m2 )! (1 + m2 )! 1 ≥ |m| ∧ 1 ≥ |m2 | ∧ 2 ≥ |m1 | (6.

.5). . .6. .. i. m2 = − 2 . . provide the irreducible representation for D(ϑ). 2. Ω2 ) provide the irreducible representation for the rotation R(ϑ) defined in (6. If we define the matrix representation of R(ϑ) by D(ϑ). rotations in 2–particle function space. then for a basis {Y 1 m1 (Ω1 )Y 2 m2 (Ω2 ).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 6.1: Oscillatory behavior of 3j-coefficients. . . the rotations in single particle function space. (2 (2 1 2 + 1) (2 + 1) × (2 1 2 + 1) + 1) . 2 |Ω1 . i.             (6.233) . Similarly. . .. m1 = − 1 . . . the 2–particle total angular momentum wave functions YJM ( 1 .              .+ 1 . Irreducible Representation We had stated above that the spherical harmonics Y m (Ω)). 1 .e. . 2 = 1.+ 2 } the matrix has the blockdiagonal form             D(ϑ) =               1·1×1·1 ¤ -10 60 700 m 10-m -40 -20 0 20 40 60 m  1·3×1·3 .e. eigenfunctions of the single particle angular momentum operators J 2 and J3 .

.234) .1: Prove Eqs.234) Exercise 6. 2. . . (2( 1 + 2 ) + 1) ×(2( 1 + 2 ) + 1) Partitioning in smaller blocks is not possible. .2: How many overall singlet states can be constructed from four spin– 1 states 2 | 1 m1 (1) | 1 m2 (2) | 1 m3 (3) | 1 m4 (4) ? Construct these singlet states in terms of the product wave 2 2 2 2 functions above. each of the blocks in (6.6.3: Two triplet states |1m1 (1) |1m2 (2) are coupled to an overall singlet state Y00 (1.6. . . M = −J. . (6.6. 2 |Ω1 . 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) (6.176 Addition of Angular Momentum and Spin 1 For the basis {YJM ( 1 . . Ω2 ). . Show that the probability of detecting a triplet substate |1m1 (1) for arbitrary polarization (m2 – 1 value) of the other triplet is 3 .233.. 2 = 1. 1). .6.233) is further block-diagonalized as follows + 2. . 1 . J = | 1 − 2 |. . . Exercise 6.

g. e. . Rotations transform |ψ as |ψ = R(ϑ)|ψ and T |ψ as R(ϑ)T |ψ . i. In fact. The latter implies that T transforms like an angular momentum or spin state | m . . e. .g.6.222) in case (ii) of single particle spin operators. −k + 1.g. that in case (i) T is an operator C∞ ( ) → C∞ ( ) acting on single particle wave functions. Such operators T can be characterized through their behaviour under rotational transformations. the operator (5. (6.42) in case (i) of the position representation of single particle wave functions or the operator (5. Examples of Tensor Operators The multiplicative operators C∞ ( ) → C∞ ( ) Y ( ) = def Y ( ) (6.235) 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 .223). (k) (6. k} with the transformation property (6.236. q = −k.237) The operators T ∈ {Tkq .e. . some of the r r examples considered below involve tensor operators T of this type. 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. ∂x2 3 1 2 2 2 2 case (ii) the operator T could be a spin operator Sk defined in (5. .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. (k) (6. This implies. The latter can be written T |ψ where T denotes T in the rotated frame given by T = R(ϑ) T R−1 (ϑ) .238) .7: Tensor Operators 177 6.237) are called tensor operators of rank k. q = −k. 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). 6. . The operator T may also act on multi-particle states like Y 1 m1 (ˆ1 )Y 2 m2 (ˆ2 ). for example. k} such that k Tkq = q =−k (k) Dq q (ϑ) Tkq . e. that T belongs to a family of operators {Tkq . S 2 = S1 + S2 + S3 or any other polynomial of Sk .236) In this equation Dq q (ϑ) denotes the rotation matrix Dq q (ϑ) = kq |R(ϑ)|kq . −k + 1. Note that in the examples mentioned the operator T would be defined within the same representation as R(ϑ).

This choice of parametrization allows us to derive conditions which are equivalent to the property (6.7. In fact. we will assume presently that the rotation is chosen as follows R(ϑ) = exp ( ϑ+ L+ + ϑ− L− + ϑ3 L3 ) (6. 1.7.240) 1 where we have defined ϑ± = 2 (ϑ1 iϑ2 ) and L± = L1 ± iL2 . 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. x2 .235 . Exercise 6. x3 . The fact that these operators form tensor operators of rank 1.237). .237) do not require that the rotation R−1 (ϑ) is expressed in terms of Euler angles according to (5. 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.3. 2. 3 follows from the transformation properties of the spherical harmonics derived in Section 1. For the following it is important to note that the rotation matrix elements (6.203).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 0.6.239) These operators can be expressed in terms of the coordinates x1 . but rather any rotation and any parametrization can be assumed. but are far easier to ascertain for any particular operator. 2.178 are tensor operators of rank k.

1 particles for which the first spin is described by 2 the operator S (1) and the second spin by the operator S (2) .248) a tensor operator? Exercise 6.247) These properties often can be readily demonstrated for operators and the transformation properties (6. . (0.7.6: Form tensor operators of rank 1 in terms of the three components of acting on the space of 1-particle wave functions. Tkq ] = −iTk q+1 (k + q + 1)(k − q) .245) (6. For this 2 2 purpose state first the proper rotation operator R(ϑ) and note then that S (1) · S (2) commutes with 1 the generators of the rotation of | 2 m1 (1) | 1 m2 (2) . We first consider a rotation with ϑ = (ϑ+ .244) [L+ . (k + q + 1)(k − q) (k + q)(k − q + 1) (6.241) Using R(ϑ) = 1 + ϑ+ L+ + O(ϑ2 ) this equation can be rewritten neglecting terms of order O(ϑ2 ) 1 + + k (1 + ϑ+ L+ ) Tkq (1 − ϑ+ L+ ) = 1 1 q =−k kq |1 + ϑ+ L+ |kq Tkq . Tkq ] = = δqq and by subtracting Tkq on both sides of the k ϑ+ kq |L+ |kq Tkq . Tkq ] [J− . Exercise 6.7: Tensor Operators 179 The conditions can be derived if we consider the property (6. 1 (6. ϑ3 .80) follows kq |L+ |kq = −i (k + q + 1)(k − q)δq and.246) (6. (6.6.235 . 6.237) for transformations characterized by infinitesimal values of ϑ+ .246.6. 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) . (6. Similar equations can be derived for infinitesimal rotations of the form ϑ = (0.247).6.237) yields first k R(ϑ) Tkq R −1 (ϑ) = q =−k kq |R(ϑ)|kq Tkq . ϑ3 )T .3: Derive Eqs. 0)T .235 .5: Consider a system of two spin.4: Is the 1-particle Hamiltonian 2 H = − 2 2m + V (|r|) (6.243) From (5. Tkq ] = = = Tkq+1 Tkq−1 q Tkq .7. J− . ϑ− .6.7. Exercise 6.7. 2 Exercise 6. hence. 0)T in case of small ϑ+ .235 .237) be assumed then. Tkq ] [J3 . The property (6. q =−k q+1 (6. ϑ− .242) from which follows by means of kq |1 1|kq equation ϑ+ [L+ . 0. Expressing the results in terms of the angular momentum operators J+ . (6. 0. J3 yields [J+ .

the stated property. 2 |γ2 ) = ( 1 m1 |kq 2 m2 ) Tkq | 2 m2 γ2 . hence.8. 2 |γ2 ) are orthogonal to | 1 m1 in case of different quantum numbers 1 . possibly the total angular momentum– spin state of a compositie system. i.e.247). 2 |γ2 ) = Cδ 1 1 δm1 m1 2 1( 1 (6. which is characterized also by a set of other quantum numbers γ1 which are not affected by the rotational transformation R(ϑ). One can.180 Theory of Angular Momentum and Spin 6. J3 are the generators of the rotation R(ϑ).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 . 2 |γ2 ) is an eigenstate of J 2 with eigenvalue 2 1 ( 1 + 1). J2 . J 2 | 2 m2 γ2 = 2 2( 2 + 1) | 2 m2 γ2 .253) Exercise 6.6. 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. 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. i. 2 |γ2 ) which correspond to total angular momentum states.235 . J3 | 2 m2 γ2 = m2 γ2 | 2 m2 and the property (6.250) q. J3 . . The hermitian property of J3 and J 2 implies that states Φ 1 m1 (k. 1 m1 |Φ 1 m1 (k. behave like |kq | 1 m1 .21) of Clebsch-Gordan coefficients J3 Φ 1 m1 (k. in fact.e.237) is that their matrix elements 1 m1 γ1 |Tkq | 2 m2 γ2 obey simple relationships expressed in terms of Clebsch-Gordan coefficients.e.252) ∼δm1 q+m2 = ( 1 m1 |kq 2 m2 ) Tkq Similarly. (6.251) The corresponding property for Φ 1 m1 (k. 2 |γ2 ) can be shown readily as follows using (6.18) are defined through Φ 1 m1 (k.m2 ( 1 m1 |kq 2 m2 ) (J3 Tkq − Tkq J3 + Tkq J3 ) | 2 m2 γ2 = qTkq = q. (6.1: Show that Φ 1 m1 (k. Before we proceed we like to point out that the states | 2 m2 γ2 are also eigenstates of J 2 . 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.m2 ( 1 m1 |kq m1 q.m2 2 m2 ) (q + m2 ) Tkq | 2 m2 γ2 | 2 m2 γ2 (6. i. 2 |γ2 ) = q. J3 | 2 m2 γ2 = m2 | 2 m2 γ2 . m1 . These states according to (6. construct states Φ 1 m1 (k.8 Wigner-Eckart Theorem A second important property of tensor operators Tkq beside (6. one can show that Φ 1 m1 (k.249) which demonstrates. | 1 m1 γ1 denotes an angular momentum (spin) state. 2 |γ2 ) is an eigenstate of J 2 with eigenvalue + 1).

γ1 |Tkq | 2 m2 .e.260) Exercise 6.259) The socalled reduced matrix element 1 .8: Wigner-Eckart Theorem 181 In order to evaluate the matrix elements (6. m1 = q = m2 = 0. γ2 = 2 1 +1 1 m1 .8. Using | 1 m1 + 1γ1 = 1 ( 1 + m1 + 1)( 1 − m1 ) J+ | 1 m1 γ1 (6.256) and noting that the operator adjoint to J+ is J− . m2 . (6. γ2 . One can then evaluate also the corresponding Clebsch-Gordan coefficient ( 1 m1 |kq 2 m2 ) and determine 1 . for which the l. 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 ) = (6.255) At this point the important property can be proven that 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) is independent of m1 .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 ) (6.2: Determine the matrix elements of the gradient operator d3 rF (r) G(r) of the type (6.s. γ2 (−1)k− 2 + 1 ( 1 m1 |kq 2 m2 ) (6.h.254) = 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) ( 1 m1 |kq 2 m2 ) . γ1 ||Tk || 2 . γ1 |Tkq | 2 m2 . its m1 –dependence being expressed solely through a Clebsch-Gordan coefficient. (6.259) to a combination of magnetic quantum numbers m1 .261) .258) We can then finally express the matrix elements of the tensor operators Tkq as follows 1 m1 . can be evaluated as easily as possible. 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 . γ1 ||Tk || 2 . i.32) Tkq | 2 m2 γ2 = and orthogonality property (6.g. the matrix elements 1 m1 γ1 |Tkq | 2 m2 γ2 can be reduced to an m1 –independent factor. γ2 k− 2 + 1 √ 1 2 m2 ) (−1) 2 1 +1 = 1 . To prove this property we consider 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) for a different m1 value.6. e. say m1 + 1. γ2 ( 1 m1 |kq (6. γ2 is determined by applying (6. q .257) which establishes the m1 –independence of 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) . γ1 ||Tk || 2 .

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

The potential (7. See.4) governs the motion of electrons in hydrogen-type atoms. The first potential V (r) = 0 0 ≤ r ≤ R ∞ r > R (7.e. in potentials which are solely a function of r and are independent of the angles θ. φ.2) V (r) = 0 r > R The third potential V (r) = 1 mω 2 r2 2 (7. i. Talmi (Harwood Academic Publishers. The potentials (7. 7000 Chur.. (7. 7. Simple Models of Complex Nuclei / The Shell Model and the Interacting Boson Model by I. The second potential is of the square well type −Vo 0 ≤ r ≤ R . molecules.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). Four examples will be studied. The fourth potential V (r) = − Z e2 r (7. It leads to the stationary electronic states of the hydrogen atom (Z = 1). Switzerland. for example. for example.2) serve as schematic descriptions of quantum particles. Poststrasse 22.1.4) is by far the most relevant of the four choices.1) confines a freely moving particle to a spherical box of radius R.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 corresponding wave functions serve basis functions for multi-electron systems in atoms. Potential (7. in case of the so-called bag model of hadronic matter.3) describes an isotropic harmonic oscillator . and crystals. 1993) 1 183 .

184 Spherically Symmetric Potentials 7.15) (7. (7.m (r) . using (7. (5.m (r) . 3.m (r) . J3 ψE. θ for the distance from the origin and for the angular position.12) . k = 1. Jk } where H is the Hamiltonian.87) for Jk as well as the commutation property (5.85– 5.m (r) m ψE. d ˙ J = v × mv + r × mv = 0 + r × er (−∂r V (r)) = 0 . 2.8) This property can also be proven readily employing the expression of the kinetic energy operator [c. . the time variation of J can be written. ( + 1) ψE. 2 .m (r) can be chosen as simultaneous eigenstates of the Hamiltonian ˆ operator H as well as of J 2 and J3 . 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 .5) In the case of an angular independent potential angular momentum J = m r × v is a constant of ˙ motion. stationary states ψE. 2 . 2. Jk ] = 0 .m (r) = = = E ψE. one can state ˙ J = m r2 θ .11) (7.5) and r = v. 3 .. Employing in this plane the coordinates r. .e. Jk } = 0 . 2 2mr2 (7. i. k = 1. For this purpose one concludes first that the conservation of angular momentum J implies a motion of the particle confined to a plane. (7. In fact. Accordingly. J ψE. (7. In classical mechanics one can exploit the conservation of angular momentum to reduce the equation of motion to an equation governing solely the radial coordinate of the particle.1 Radial Schr¨dinger Equation o ˙ mv = − er ∂r V (r) .14) (7.61) of J 2 and Jk .10) (7.7) ˆ The correspondence principle dictates then for the quantum mechanical Hamiltonian operator H and angular momentum operators Jk ˆ [H. one can conclude ˙ Jk = {H. ˆ dt (7.99)] 2 1 2 J2 2 2 = − ∂r r + .13) .f.6) ˙ Since Jk is also equal to the poisson bracket {H.9) − 2m 2m r 2mr2 the expressions (5. ˆ A classical particle moving in a potential V (r) is governed by the Newtonian equation of motion (7.m (r) (7. ˆ H ψE.

for the wave function of a quantum mechanical particle a differential equation which governs solely the r-dependence.m (r) 1 2 J2 ∂r r + + V (r) − E 2m r 2mr2 the functional form ψE.m (r) = 0.e. Multiplying (7. ( + 1) 2mr2 . together with the original potential. ∞[ governed by the effective potential (7.22) which is the original potential 2 ( +1) V (r) with an added rotational barrier potential 2mr2 . 2 (7.11).23) This demonstrates that the function r vE.7.17) In analogy to the classical description one can derivem. . This barrier.11.24) . .13).m vE.1: Radial Schr¨dinger Equation o 185 This is a differential equation which governs solely the radial coordinate.12) are obeyed and one obtains for (7. (r) = 0 . (7. (7. in the present case. can exclude particles from the space with small r values.20) in the form of the one-dimensional Schr¨dinger equation o 2 − where 2m 2 ∂r + Veff (r) − E φE (r) = 0 .m ψE.22) (7. − Adopting for ψE.4. 7.16) m 2mr2 Once.10).20) by −2mr/ 2 yields the so-called radial Schr¨dinger equation o 2 ∂r − ( + 1) − U (r) − κ2 r2 r vκ.m (r) Y m (θ.m (r) 2 . (7. i.m (r) . Employing the kinetic energy operator in the form (7. using (7.18) = vE. φ) are the angular momentum eigenstates defined in Section 5.20) Since this equation is independent of the quantum number m we drop the index m on the radial wave function vE. .10) − 1 2 ∂ r + 2m r r 2 2 ( + 1) + V (r) − E 2mr2 . (7. but can also trap particles in the latter space giving rise to strong scattering resonances (see Section ??).21) Veff (r) φE (r) = = V (r) + r vE. One can write (7. φ) (7. It can be solved by integration of 1 2 2 J2 r = ± ˙ E − V (r) − . equations (7.9) one can write the stationary Schr¨dinger o equation (7. mr2 (t) (7. .m (r) and E . .m (r) = 0. .m (r) describes the radial motion as a one-dimensional motion in the interval [0. . r(t) is determined the angular motion follows from (7.m .. by integration of ˙ θ = J . (7.19) where Y m (θ.

for example. 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). hence. Berlin. (r) (7. . 2 2 2m 2 ψE. κ assumes real values.32) The total kinetic energy resulting from this contribution is. r → 0 (7.m (r) ∼ √ B . Boundary Conditions In order to solve (7. accordingly. 2nd Ed. This follows for > 0 from consideration of the integral which measures the total particle density.186 where we defined U (r) κ2 = = − − 2m 2 Spherically Symmetric Potentials V (r) E (7.28) (7. see. (r) ∼ A r or vκ. For r → 0 one may assume that the term ( + 1)/r2 becomes larger than the potential U (r).30) and. We replaced in (7.26) 2m 2 In case E < 0.31) to the complete wave which. For = 0 the contribution of Br−( function is. according to a well-known result in Classical Electromagnetism3 . Jackson (John Wiley. The radial part of this integral is ∞ 0 2 dr r2 vκ. for > 0 is not integrable. Pascual (Springer.27) ∂r − r2 and.20) one needs to specify proper boundary conditions2 . In this case the solution is governed my ( + 1) 2 r vκ.D. ”Classical Electrodynamics.182) for Y00 . using the expression (5. (7. II by A. the term Br−( +1) would contribute |B|2 0 dr r−2 +1) (7. New York. (r) ∼ A r + B r− −1 +1 + B r− (7.25) (7. 1975). . . ψE. assumes the general form r vκ. in particular. Galindo and P.29) Only the first term is admissable. 4π|r| (7.33) A detailed discussion of the proper boundary conditions.m (r) ∼ √ 4π B δ(x1 )δ(x2 )δ(x3 ) . at r = 0 is found in the excellent monographs Quantum Mechanics I. (r) = 0 .” by J.20) the index E by the equivalent index κ.

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).

z = kr.s.100) Defining x = cos θ. φ) for non-vanishing m has a non-trivial φ-dependence as described by (5. (7..150) one obtains +1 +1 1 ∂ dx eizx P (x) = dx eizx (x2 − 1) .98) The l. 2 +1 (7.s.h. For example.194 Generating Function of Spherical Bessel Functions Spherically Symmetric Potentials The stationary Schr¨dinger equation of free particles (7. One can show d −1 2 (x − 1) ∼ (x2 − 1) × polynomial in x dx −1 (7. −1 .50). can be written exp(ikr cos θ).98) by ∞ eik3 x3 = =0 b j (kr) P (cos θ) .178) are given in terms of Legendre polynomials P (cos θ) one can replace the expansion in (7. the surface term ∼ [· · ·]+1 vanishes. 7. (7.102) −1 d −1 +1 −1 dx −1 (x2 − 1) .h.50). The orthogonality properties (5. of (7.106). Since the spherical harmonics Y 0 (θ. according to (5. and using the Rodrigues formula for Legendre polynomials (5.98) does not involve any non-vanishing m-values since Y m (θ. one given o by (7. φ) . In this case the expansion on the r.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 (7.179) yield from (7.104) dx eizx P (x) −1 = (−1) 2 ! (iz) = 2 ! +1 dx (x2 − 1) −1 +1 dx (1 − x2 ) eizx . namely. φ).e. the wave function does not depend on φ.103) and. (7. i. in case of a free particle moving along the x3 -axis one expands eik3 x3 = .99) +1 d cos θ eikr cos θ P (cos θ) = b j (kr) −1 2 . hence. This holds for −1 and one can conclude +1 consecutive integrations by part d izx e dx (7.m a mj (kr) Y m (θ.43) has two solutions. One can expand the former solution in terms of solutions (7.99) We want to determine the expansion coefficients b .44.45) and one given by (7.

1 (7. The identity (7. after insertion into (7.108) where the last factor is equal to unity.111) Employing (7. for the leading power x [c. Integral Representation of Bessel Functions Combining (7.69)] b z 2 (iz) = 1 · 3 · 5 · · · (2 + 1) 2 + 1 2 ! +1 dx (1 − x2 ) .h.117) one can write the r.s.105) This expression allows one to determine the expansion coefficients b . −1 (7.112) We want to consider the expression Gν (z) = aν z ν fν (z) where we define fν (z) = C (7. using ν = Jν (z) = √ 1 πΓ(ν + 1 ) 2 z 2 ν + 1. 2 +1 −1 dx (1 − x2 )ν− 2 eizx .h.106) Employing (5.109) results in the integral representation of j (z) j (z) = (z) 2 +1 ! +1 dx (1 − x2 ) eizx . of (7.105) and (7.105) must hold for all powers of z.106) yields finally b = i (2 + 1) or.2: Free Particle Comparision with (7. as the generating function of the spherical Bessel functions. −1 (7.f.107) 1 2 (2 )! 1 · 3 · 5 · · · (2 + 1) 2 + 1 1 · 2 · 3 · 4 · · · (2 − 1) · 2 (7. in particular. −1 (7.100) gives b j (kr) 2 (iz) = 2 +1 2 ! +1 195 dx (1 − x2 ) eizx . Comparision with the l. (7.h.s. 1 (7.114) .109) eikr cos θ = =0 i (2 + 1) j (kr) P (cos θ) . ∞ (7. z i or i (2 + 1) z (2 )! 2 2 2 +1 2 ! [1 · 3 · 5 · · · (2 − 1)] 1 (7.96) one can express this.7.110) One refers to the l.99).s.113) dt (1 − t2 )ν− 2 eizt . (7.

2. z (7. 3. Gν (z).120) z 2ν + 1 dt (1 − t2 )ν− 2 eizt + C C 1 1 t2 t1 = 0 (7. obeys Bessel’s equation (7.118) z 2ν + 1 fν (z) + fν (z) . C 1 (7. 4. 1 j = 1. We can now demonstrate that Gν (z) defined in (7. t2 of the integration paths C.120) imply also the property (2ν + 1) fν (z) + z fν+1 (z) = 0 .113) obeys the Bessel equation (7. (7.114.117). for properly chosen endpoints t1 .2. We consider now the functions u(j) (z) = √ 1 πΓ(ν + 1 ) 2 z 2 ν Cj dx (1 − x2 )ν− 2 eizx . Exercise 7.116) In case that the endpoints of the integration path C satisfy (1 − t2 )ν+ 2 eizt one can write (7.93) for arbitrary ν.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 . 1 (7. In fact.117) dt (1 − t2 )ν− 2 t2 eizt 1 (7. (7.121) (7.93) as long as the integration path in (7.124) .116) fν (z) = − or fν (z) = − From this we can conclude fν (z) + 2ν + 1 fν (z) + fν (z) = 0 .113) satisfies (7.93).196 Spherically Symmetric Potentials Here C is an integration path in the complex plane with endpoints t1 . t2 . 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 (7.121). To prove this we note fν (z) = i C dt (1 − t2 )ν− 2 t eizt .4: Prove (7.123) Insertion of these identities into Bessel’s equation leads to a differential equation for fν (z) which is identical to (7.119) We note that equations (7.122) (7.120) such that we can conclude that Gν (z) for proper integration paths is a solution of (7. 7.

130) holds Jν (z) = (1) 1 2 1 (1) (2) Hν (z) + Hν (z) .117) and. Following convention. we introduce the so-called Hankel functions (1) Hν (z) = −2 u(2) (z) .7.126) −1 ≤ s ≤ 1 (7. the integral expression (1) Hν (z) 1 1 1 1 (7.125) 0 ≤ s < ∞ (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. therefore.124) vanishes along the whole path C3 and. one can derive (2) Hν (z) = e −i(z−πν/2−π/4) ∞ 0 √ ds [s(1 − is/2)]ν− 2 e−zs . .130) We note that the endpoints of the integration paths C2 and C4 .134) and (7. u(3) (z) ≡ 0. (7. hence.128) C = C1 ∪ C2 ∪ C3 ∪ C4 is a closed path.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 .134) Similarly. (7.135) are known as the Poisson integrals of the Bessel functions.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. dt = i ds.93). 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 (7.132) For Hν (z) one derives.131) According to (7. u(2) (z) and u(2) (z) are both solutions of Bessel’s equation (7. using t = 1 + is.112) shows u(1) (z) = Jν (z). Since the integrand in (7. Accordingly. for Rez > 0 and ν ∈ R obey (7. (7.135) (7.127) 0 ≤ s < ∞ (7.129) The integrand in (7. 1 (7. Comparision with (7. (2) Hν (z) = −2 u(4) (z) . one can state Jν (z) = − u(2) (z) + u(4) (z) .

πz (7.140) Bessel Functions with Negative Index Since ν enters the Bessel equation (7.134) and (7.141) is o W (z) = − In case of g(z) = H (1) (z) +1 2 c .143) .135) which read for ν = + 1 2 H where f The binomial formula yields f (1) 2 (1) 2 1 +2 (1. We employ for this purpose the Poisson integrals (7.136) (1) 2 (z) = 0 ds [s(s ± is/2)] e−zs . i. converge fast for |z| → ∞.. As a second order differential equation the Bessel equation has two linearly independent solutions.140). For the Wronskian connected with the Bessel equation (7.198 Asymptotic Behaviour of Bessel Functions We want to obtain now expansions of the Hankel functions H Spherically Symmetric Potentials the expansions converge fast asymptotically. For such solutions g(z). Hν (z) as well as H−ν (z) are solutions of this equation. z (7.141) (1) (1) the derivation of which follows the derivation on page 192 for the Wronskian of the radial Schr¨dinger equation. The general solution of (7.138) The formula for the Laplace transform of sn leads to f and.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. (7.93) holds the identity 1 W = − W z (7.e.142) and h(z) = H (2) (z) +1 2 one can identify the constant c by using the leading terms in the expansions (7. h(z) to be linearly independent. hence. (7. the Wronskian W (g. h) must be a non-vanishing function.93) only as ν 2 .137) (z) = r=0 r ± i 2 r 0 ∞ ds s +r e−zs . One obtains W (z) = − 4i .139) (z) = 2 ±i[z − ( e πz +1) π ] 2 r=0 ( + r)! r!( − r)! i ± 2z r (7. we obtain H (1) 2 1 +2 (1) 2 (z) = r=0 ( + r)! ! r!( − r)! ± i 2 r 1 z +r+1 (7.

2 (1. (7. H (1.146) (z) = −i (−1) H (2) 1 (z) +2 . B can be obtained from the asymptotic expansion (7.96.148) holds for the o regular spherical Bessel function j (z) = 1 (1) (2) [ h (z) + h (z) ] .2) (2) +1) 2 (1) +1) 2 (z) = i (−1) H (1) (z) +1 2 .132) follows n (z) = 1 2 π (−1) 2z +1 H−( (1) 1 +2) (z) + H−( (2) +1) 2 (z) .90.2) (7.148) Following the arguments provided above (see page 193) the functions h (z) are solutions of the radial Schr¨dinger equation of free particles (7.2) (z) +1 2 199 are.151) are equivalent to h h (1) (2) (z) (z) = = j (z) + i n (z) j (z) − i n (z) . We conclude 2 H−( Simarly. 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| → ∞ .152) (7.97). 7. 7.147) (z) = π (1. 2z 2 (1.e.97.2: Free Particle i.146. 7.149. 7.151) Equations (7.7. linearly independent.132.53).145) This equation can hold only for B = 0 and A = exp[i( + 1 )π]. (7. (7. 7.153) . (7. one can show H−( Spherical Hankel Functions In analogy to equations (7. (7.140). (7.150) According to (7.97) one defines the spherical Hankel functions h (1.2) H + 1 (z) . 7.2) (7. H−ν (z) = A H (1) (1) (z) +1 2 One can then expand + BH (2) (z) +1 2 . in fact.147) this can be written n (z) = 1 2i h (1) (z) − h (2) (z) . According to (7.149) We want to establish also the relationship between h (z) and the irregular spherical Bessel function n (z) defined in (7.. (7. From (7.144) The expansion coefficients A.

ne ll(z) = Im[h (1) (z)] .157) i q (z) 2z = i Im r=0 i 2z i 2z r = [ −1/2] ( +2r+1)! r=0 (2r+1)!( −2r−1)! −1 r 4z 2 0 ≥ 1 = 0 (7. are j (z) n (z) = = (7.154) The leading term in this expansion.159) (7.160) Employing cos[z − ( + 1) π ] . Hence.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) . = −cos(z − π/2) results in 2 2 the alternative expressions j (z) n (z) = = sin(z − π/2) cos(z − π/2) p (z) + q (z) z 2z 2 cos(z − π/2) sin(z − π/2) q (z) − p (z) . z (7.140. j (z) (z) = ( i) z +1 e±iz r=0 ( + r)! r!( − r)! ± i 2z r (7. at large |z|. as given by (7.140) and (7. at large |z|. (1) (7.155) (z) and To determine j (z) and n (z) we note that for z ∈ R the spherical Hankel functions h (2) h (z).148).158) one can derive then from (7. it follows z ∈ R : Using p (z) = Re r=0 j (z) = Re[h (1) (z)] .161) (7.162) The leading terms in these expansions. (7. 7. From (7. are complex conjugates. = sin(z − π/2) and sin[z − ( + 1) π ] . 2z 2 z sin(z − π/2) z cos(z − π/2) − .163) (7.2) (z). 2 2z z (7.156) ( + r)! r!( − r)! ( + r)! r!( − r)! i 2z r [ /2] = r=0 ( + 2r)! (2r)!( − 2r)! −1 4z 2 r (7.164) .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). is h (1) 2 (z) = ( i) z +1 e±iz .148) one obtains readily h (1) 2 (1.

175) Using Γ( + 1) = !.151) follows then that the relatiosnhips hold also for g (z) = j (z) and g (z) = n (z). (1.171) −1 (z) (7.171–7.124. (7.168) (7.148) and express g (z) = h (1.4 dx (1 − x2 )ν− 2 eizx .149.2) g (z) − g (z) z 2 +1 g (z) − g z (7. defining a = −1 and employing fν (z) as defined in (7. 7. 1 (7.174) and from (7.2: Free Particle Expressions for the Spherical bessel Functions j (z) and n (z) 201 The identities (7.174) We want to prove these relationships.7. 7. From the linearity of the relationships (7.171) holds for g (z) = h (z) we employ (7.2) relationships hold for g (z) = h (z). 7. One obtains for = 0.170) Recursion Formulas of Spherical Bessel Functions The spherical Bessel functions obey the recursion relationships g g +1 (z) +1 (z) = = (1.2) (z) = −2 π 1 √ 2z πΓ(ν + 1 ) 2 z 2 ν C2.162) allow one to provide explicit expressions for j (z) and n (z). j (z) and n (z).172) to +1 (z) +1 (z) = A (z)  1− g (z) g (z) −1   (7.167) (7.176) .161. One can combine (7.172) where g (z) is either of the functions h obtain the recursion relationship g g (z).131.114) we can write g (z) = a z f 2 ! + 1 (z) 2 . 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 (7.165) (7.169) (7. 7. 1.171.166) (7.2) To demonstrate that (7. For this purpose we need to demonstrate only that the (1.173) A (z) =  z ( +1) z2 +2 z (7. 7.

6: Employ the recursion relationship (7.179) leads to (7. 7. z (7.182) Replacing all first derivatives employing (7.165).179) from which follows (7.172).177) (z) = − z f 2 +2 + 3 (z) 2 .173) is equivalent to (7. (7.171).172).180) yields g +1 (z) = g (z) − ( + 2) g (z) + z2 +2 g (z) . z2 (7. j0 (z) using (7. n0 (z) using (7. in fact.176) g (z) = z g (z) − g +1 (z) +1 2 Spherically Symmetric Potentials z g (z) + a z f 2 ! +1 2 (z) .166).174) to determine (a) j1 (z). To prove (7.181) Using this identity to replace the second derivative in (7. is equivalent to (7.5: Provide a detailed derivation of (7.171). and (b) n1 (z). In order to prove (7.e. The second component of (7. n2 (z) from n0 (z). 7. (7.174) we start from (7.171).178) (7. together with (7. .173.2. (7.182).179) g (z) = − z2 g (z) + g (z) − g +1 (z) z .121).172) we differentiate (7. j2 (z) from j0 (z). Exercise 7.2. The first component of (7.. i.58) it holds o 2 g (z) = − g (z) + z ( + 1) g (z) − g (z) .173.202 The derivative of this expression is g (z) = Employing (7. Exercise 7. f yields.180) Since g (z) is a solution of the radial Schr¨dinger equation (7.173).

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

t) = − V (r.2) In addition. the two homogeneous Maxwell equations hold × E(r.25).5) where p is the momentum of the particle and ro (t) it’s position at time t. obeys the equation of motion d p = q dt E[ro (t). Equation (8. From this follows E(r. t) − ∂t χ(r. (8. t) is a scalar function. t] (8. As is well known.204 Maxwell equations1 · E(r. t) where V (r. called the vector potential. t) which is solved by E(r.2) the term q r˙o δ(r − ro (t)). 1975) for a discussion of these and other conventional units. but leave the fields unaltered: A(r.4). t) + v × B(r. by J. t) + ∂t A(r. New York. t) (8. t) = − V (r.7) Gauge Transformations We have expressed now the electric and magnetic fields E(r. D. t) = 0 = 0. t] + v × B[ro (t). solves implicitly (8. The following substitutions.3) (8. t). t) + ∂t B(r.5) above agrees with the equation of motion as given in (1. t) × B(r. The particle. accordingly. t) −→ 1 A(r. t) = × A(r. contributes to the charge density ρ(r.6) for some vector-valued function A(r. t) . t) + ∂t A(r.1) (8. (8. t) = 4 π J(r.4) Lorentz Force A classical particle with charge q moving in the electromagnetic field experiences the so-called Lorentz force q[E(r. The reader is referred to the well-known textbook ”Classical Electrodynamics”. t) . t) in (8. In the non-relativistic limit holds p ≈ mr and (8.9) (8. (8. We assume so-called Gaussian units. 2nd Edition.1) the term qδ(r − ro (t)) and to the current density ˙ J(r. t) in (8. called the scalar potential.11) V (r. t) −→ V (r. the relationship between fields and potentials is not unique. Jackson (John Wiley & Sons. t) . t). t) through the scalar and vector potentials V (r. . t) − ∂t A(r. t) Interaction of Radiation with Matter = 4 π ρ(r. t) + χ(r.8) = 0 (8. t) (8. alter the potentials. Scalar and Vector Potential Setting B(r. called gauge transformations. t)] and. t) · B(r. t) and A(r.3) reads then × E(r. t) − ∂t E(r. in turn. t) and B(r.10) (8.

t) (8. t) Using · ∂t A(r. (8.20) . In fact.16) This is the potential commonly employed in quantum mechanical calculations for the description of Coulomb interactions between charged particles. t) = − 4 π J(r.8. t) and the current due to moving net charges.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. this potential results from inserting (8.9) into (8. t) . and the former the Coulomb interactions in the unperturbed bound particle system. t) . and a component due to the remaining currents. t) can be obtained employing (8. Using the expressions (8. t) = Ω∞ for r ∈ ∂Ω∞ (8.1) · − V (r.2).14) In case of the boundary condition V (r. t) and V (r. t) = 1 ∂t V (r. t) . (8.15) d3 r ρ(r . The vector potential A(r. the second inhomogeneous Maxwell equation. t) + ∂t A(r. t = 4 π J(r.12) yields then the Poisson equation 2 V (r. 4π (8. For this purpose we examine the offending term ∂t V (r. equation (8. the gauge freedom allows us to impose on the vector potential A(r.9) for the fields results in × The identity × × A(r. t) − ∂t V (r. t) .12) The corresponding gauge is referred to as the Coulomb gauge.6) and (8. t) − ∂t A(r. (8.16) and the vector potential are uncoupled. Such description can.17) A(r. be achieved.13) · A(r. t). t) = ∂t = 4 π ρ(r.19) Unfortunately.19) couples the vector potential A(r.12) leads us to 2 2 A(r.18) together with condition (8. t) together with (8. t) − ∂t A(r. (8. a name which is due to the form of the resulting scalar potential V (r. t) |r − r | (8.19) and define J (r. (8. such that the latter describes the electromagnetic radiation. t). t) + ∂t V (r. in fact. t) in (8. In fact. t) . t) − 2 × A(r. t) = 0 the solution is given by the Coulomb integral V (r. t) = 0 . t) the condition · A(r. t) = · A(r. One would prefer a description in which the Coulomb potential (8. t) = − 4 π ρ(r.

= − ∂t A(r.g.30) .25) one refers to J and Jt as the longitudinal and the transverse currents.27) which is obviously curl-free ( × E (r. the name transverse fields. such current densities exist.12) and the Coulomb potential (8. t) = 0 .21) ∂t and the Poisson equation (8. (8. A(r. yet a current.16) yield finally the electric and magnetic fields. 8.24) (8. hence. the name longitudinal electric field.29) · Et (r. For the divergence of J (r.. t) + ∂t ρ(r. (8. t) = − V (r. no divergence-free currents. t) = − 4 π Jt (r. (8. t) = 0 (8. respectively. t) (8. t) . V (r. for example. t) be the total current of the system under investigation and let Jt = J − J . in a ring-shaped antenna which exhibits no net charge.26) This equation does not couple anymore scalar and vector potentials. t). t) Bt (r. t) − ∂t A(r.22) 1 ∂t 4π · J (r. hence.14). t) + ∂t ρ(r. t) = 0 . t) − ∂t A(r.28) (8. The definitions of J and Jt applied to (8.25) Because of properties (8. t) = or = 2 Interaction of Radiation with Matter (8. t) as the current due to the time-dependence of the charge distribution ρ(r. t) . For J also holds the continuity equation · J(r. (8. i. t) = 0 and from this follows · Jt (r.e. t) = 0 . t) contributes solely an electric field component E (r. t) holds. t) = 0).21) and (8. t) These fields are obviously divergence -free (e. Let J(r. In this case (8. The vector potential determined through (8.26) turns into the well-known wave equation 2 2 A(r.2 Planar Electromagnetic Waves The current density Jt describes ring-type currents in the space under consideration. we want to assume that no ring-type currents.. V (r. exist in the space considered.19) yield 2 2 A(r. t) = × A(r.206 For the curl of J holds × J (r. t) = − ∂t ρ(r. Presently.23) This continuity equation identifies J (r. t) contributes an electrical field component as well as the total magnetic field Et (r.26) and (8. t) = 0). t) (8. using ∂t · J (r.

Et (r. (8. 8. see also the comment below Eqs. t) . Note that in the units chosen the velocity of light is c = 1. 2 . 8.38) ˆ In this expression for the energy flux one can interprete k as the propagation velocity (note c = 1) and.158).34) (8.31) u is a unit vector (|ˆ| = 1) which. at each point r and moment t.34.34. t) = i k × A(r. t) and Bt (r. t) = ±i ω A(r. 8.35) The vector potential in (8. The corresponding electric and magnetic fields. are Et (r. is orthogonal to k.31). 8. The latter vector describes the direction of propagation of the energy flux connected with the plane wave electromagnetic radiation.31. 8π (8. ˆ (8.37) ˆ ˆ where k is the unit vector k = k/|k|. there exist two linearly ˆ u independent orientations for u corresponding to two independent planes of polarization. ω2 = |Ao |2 (8.2: Planar Electromagnetic Waves 207 which describes electromagnetic fields in vacuum.28. Obviously.8. accordingly.35) one obtains S(r. according to (8. obviously.32.39) 8π The definition incoming waves and outgoing waves is rationalized below in the discussion following Eq. 8. t) = Re Et (r.12) yields u·k = 0.29).35) are complex-valued quantities. the constant k is referred to as the wave vector. 8.34. t) . (8. ˆ We want to characterize now the radiation field connected with the plane wave solutions (8. This flux is given by 1 S(r. t) = Ao u exp i(k · r ˆ where the dispersion relationship |k| = ω (8. t) × Re B(r.39).38.35). In applying the potential and fields to physical observables and processes we will only employ the real parts. A complete set of solutions is given by the so-called plane waves A(r. Time average over one period 2π/ω yields S(r. t) = ± ω2 ˆ |Ao |2 k sin2 ( k · r − ωt ) 4π (8.32) holds.36) 4π Using the identity a × (b × c) = b (a · c) − c (a · b) and (8. hence. are orthogonal to each other and are both orthogonal to the wave vector k. t) = ± ω2 ˆ |Ao |2 k . 8. The Coulomb gauge condition (8.31) and the resulting fields (8. t) in (8. (8.33) ωt) (8. Here the “-” sign corresponds to so-called incoming waves and the “+” sign to outgoing waves2 . (8. t) Bt (r.

43) Here the fields are defined through Eqs. 8. t) = 8πNω u exp i(k · r − ωt) ˆ ωV . we will neglect the respective terms in the Hamiltonian (8. The sign in (8. u · k = 0 . t) are complex leading to the difference of a factor 2 in the energy densities of the lontitudinal and transverse components of the fields. (8. t) . hence. defined below Eqs. t) = H Ψ(r.43). Bt (r. 8.41) ωV Inserting this into (8. a ˆ specific k and a specific u.29) together with the potentials (8. We assume that the energy content of the fields is not altered significantly in the processes described and. (8. Comparision of (8.31. ˆ (8. t) 2m 2 + qV (r) . These photons each carry the energy ω. Note that E (r.208 Interaction of Radiation with Matter as the energy density. A ‘poor man’s’ quantization of the field is possible at this point by expressing the energy density (8.31).27.42) 8.40) allows one to express then the field ˆ amplitudes 8πNω Ao = .31). (8. The integrals express the integration over the energy density of the fields.44) . 8. respectively.38) implies that for incoming waves. t).39) through the density of photons connected with the planar waves (8.8. (8.40) = V It should be pointed out that Nω represents the number of photons for a specific frequency ω. If we consider a volume V with a number of photons Nω the energy density is obviously Nω ω . The wave function Ψ(r. t).32).45) . |k| = ω . (8. i (8. is p − q A(r. t) is real 1 and that Et (r. We are left with a classical Hamiltonian function which has an obvious quantum mechanical analogue ˆ H = ˆ p − q A(r.39) and (8. whereas in the case of outgoing waves the energy is transported in the direction of k. 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.28.16. t) 2m + 1 8π Ω∞ 2 H = + qV (r) 1 16π d3 r |Et |2 + |Bt |2 Ω∞ d3 r E 2 + . A correct description of the electromagnetic field requires that the field be quantized.3 Hamilton Operator The classical Hamiltonian for a particle of charge q in a scalar and vector potential V (r) and A(r.31) allows one finally to state for the planar wave vector potential A(r. (8. the energy of the plane wave is transported in the direction of −k.

8. not to phenomenological interactions like the molecular van der Waals interaction. 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. Obviously. e. It should be noted.11) affect the quantum mechanical description. In the classical case such question is mute since the gauge transformations do not alter the fields and.49) (8. hence. t) enter whereas in the classical equations of motion ˙ m¨ = q E(r.e.t)/ Ψ(r. by the group 3 The effect on other expectation values is not discussed here.t)/ should not affect a system and that. t) inside the double brackets.46). 8. (8. (8. t) are necessary to compensate terms which arise through the phase factor.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..51) 2m the potentials A(r. accordingly. t)|2 and. t) (8.t)/ [ i ∂t − q((∂t χ)) ] Ψ(r. 8..11) to (8. .g. One can show that (8. t) ˆ p eiqχ(r. (8. The idea just stated can be generalized by noting that multiplication by a phase factor eiqχ(r. t) ˆ = eiqχ(r. t) . t) . that this principle applies only to fundamental interactions. have no effect on the motion of the particle described by (8.t)/ constitutes a unitary transformation of a scalar quantity. t) = eiqχ(r. 8. t) and A(r. i.11) of the potentials is equivalent to multiplying the wave function Ψ(r.t)/ p + q(( χ)) Ψ(r.10. t) =  + qV  eiqχ(r. Elementary constituents of matter which are governed by other symmetry groups.50) The equivalence of (8.47.t)/ Ψ(r. t) r (8. does not change expectation values which contain the probability densities3 .t)/ .10. t) and V (r. however. t) + q r × B(r. such phase factor does not change the probability density |Ψ(r.t)/ Ψ(r. in the Schr¨dinger equation o   2 ˆ p − qA   i ∂t Ψ(r.44.45) leads to the Schr¨dinger equation o   2 ˆ p − q A − q(( χ))   + qV − q((∂t χ))  Ψ(r.47) is equivalent to   2 ˆ p − qA   i ∂t eiqχ(r. hence. an element of the group U(1).47) i ∂t Ψ(r.46) the fields enter. This leads to the question in how far the gauge transformations (8. t) . t) by a local and time-dependent phase factor eiqχ(r.t)/ Ψ(r.10. t) =  + qV  Ψ(r. Applying the gauge transformations (8.48) 2m For this purpose one notes i ∂t eiqχ(r. 8. t) =  2m where ((· · ·)) denotes derivatives in (( χ)) and ((∂t χ)) which are confined to the function χ(r.48) implies that the gauge transformation (8. 8.

(8.52) This expression vanishes in the present case since since ˆ ˆ p · Af = A · pf and.59) . In this o case holds V (r.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 (8.55). likewise can demand the existence of fields which compensate local transformations described by eiσ·χ(r. x2 . We want to describe the stationary states corresponding to the Hamiltonian (8. x1 .12)]. The choice of a vector potential affects the form of the wave functions describing the eigenstates and. thereby.56). therefore.58) where ∂j = (∂/∂xj ).55). t). k3 . affects the complexity of the mathematical derivation of the wave functions.t) where σ is the vector of Pauli matrices. 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.55) 8. t) = Bo . one can employ the Hamiltonian (8. The Hamiltonian (8. x3 ) = exp(ik2 x2 + ik3 x3 ) φE (x1 ) . H = 2m m 2m (8. can be described by various vector potentials. t) ≡ 0.44) 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. j = 1. for Bo = Bo e3 in (8. 2. the generators of SU(2). 3. (8. holds (8.54) ˆ2 p q ˆ q2 2 − p·A + A + qV . The vector potential (8.. For this purpose we use the wave function in the form Ψ(E.210 Interaction of Radiation with Matter SU(2).g. Accordingly. 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. The stationary homogeneous magnetic field B(r.57) satisfies · A = 0 and. the so-called Landau gauge ˆ associates the vector potential AL (r) = Bo x1 e2 ˆ (8. k2 . consequently. e.53) ˆ ˆ p·A + A·p + q2 2 A + qV 2m (8. The resulting fields are called Yang-Mills fields.56) due to the gauge freedom. In case of a homogeneous potential pointing in the x3 -direction. · A = 0 [cf.57) with a homogeneous magnetic field Bo . (8.45) with Hamiltonian (8.58).

2 2me (8. However. The stationary Schr¨dinger equation (8. x2 .65). like the corresponding gauge (8.60) Completing the square 2 2 e2 Bo 2 eBo k2 e2 Bo x1 + x1 = 2me me 2me x + k2 eBo 2 − 2 k2 2 2me (8. (8.62) x1o = and where ω = k2 eBo eBo me (8.67) . Without affecting the energy one can form wave packets in terms of the solutions (8. From this observation one can immediately conclude that the wave shifted (by e 3 function of the system. 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. according to (8.8.63) this induces a spread of the wave function in the x1 direction. is Ψ(n. 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 (8.63) (8. k2 . (8. according to (8.57).62) is that of a displaced (by x1o ) harmonic oscillator with o 2 k 2 /2m ) energies. k2 .61) leads to − where 2 2me 2 ∂1 + 2 k2 1 3 me ω 2 ( x1 + x1o )2 + 2 2me φE (x1 ) = E φE (x1 ) .64) is the classical Larmor frequency (c = 1).and x2 -coordinates.65) which localize the electrons. therefore. x1 . Solution for Symmetric Gauge The solution obtained above has the advantage that the derivation is comparatively simple. We want to employ .56) through the vector potential A(r) = 1 Bo × r . k3 .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 ) . Unfortunately. the so-called symmetric gauge which expresses the homogeneous potential (8. the states are degenerate in the quantum number k2 describing displacement along the x2 coordinate.65) where we replaced the parameter E by the integer n.59). The energies corresponding to these states are E(n.66) Obviously. the familiar harmonic oscillator quantum number. the wave function (8. 2 (8. is not symmetric in the x1 .59). k3 ) = ω(n + 2 k2 1 3 ) + .

Accordingly.12) for the Coulomb gauge.74) is in the present case 1 µclass = − e r v e3 .68) × u yields.67) one can write ˆ p·A = Using · (u × v) = − u · function f (r) ×v + v· 2i · Bo × r . For the vector potential (8.77) Comparision with (8.74) We consider a simple case to relate (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. (8. In this case the current density measures −e v oriented tangentially to the ring.72) to Hamiltonian (8.73. an electron moving in the x.74).70) ˆ ˆ p · A f = − Bo · p × r f .69) × u and × r = 0.72) and (8.78) .71) (8. (8.72) allows one to interpret µ = − e L 2me (8. 8. the magnetic moment (8.75) 2 The latter can be related to the angular momentum µclass = − and. ˆ Identifying r × p with the angular momentum operator L. in the present case of constant Bo . the Hamiltonian (8. Vmag = −µclass · Bo (8. The latter can be rewritten.76) e 2me class class · Bo .73) where µclass is the magnetic moment connected with a current density j µclass = 1 2 r × j(r) dr (8. for any (8. using × (uf ) = −u × f +f ˆ ˆ Bo · p × r f = Bo · r × p f . namely. namely.71).212 Interaction of Radiation with Matter One can readily verify that this vector potential satisfies the condition (8. accordingly. ˆ (8. The theory of classical electromagnetism predicts an analogue energy contribution . y-plane with constant velocity v on a ring of radius r. Vmag = e 2me class = r me v e3 of the electron ˆ (8. (8.

In this case. ˆ i. 10 that the spin of the electron. for Bo = Bo e3 .86) 1 2 2me 1 2 ˆ as well as of the angular momentum operator L3 . The functions ψ(x1 . 10.83) (8.71) is ˆ H = 2 ˆ2 p e2 Bo + 2me 8me x2 + x2 + 1 2 eBo L3 . x2 . e gives rise to an energy contribution (8. the solutions of ˆ H ΨE (x1 .88) me ω xj . (8. we seek stationary states which are simultaneous eigenstates of the Hamiltonian of the two-dimensional isotropic harmonic oscillator 2 1 2 ˆ Hosc = − ∂ 2 + ∂2 + me ω 2 x2 + x2 (8.e.83) describes two identical oscillators along the x1 -and x2 -directions which ˆ are coupled through the angular momentum operator L3 . x2 from x3 setting ΨE (x1 . x2 . (8.and the ±x2 -direction.67) yields a more symmetric solution which decays to zero along both the ±x1 .82) 1 ω (x1 ∂2 − x2 ∂1 ) i 2me + 1 me ω 2 x2 + x2 1 2 2 2 k2 3 + (8.8. 2me We have used here the expression for the angular momentum operator ˆ L3 = ( /i)(x1 ∂2 − x2 ∂1 ) . 2me (8. 2 . x2 ) obey then ˆ Ho ψ(x1 . we separate the variable x1 .80) (8.83) 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 ..72) with an associated magnetic moment − g 2me S where g ≈ 2. A derivation of his property and the value of g. x2 ) where ˆ Ho = − 2 2 2 ∂1 + ∂2 (8. the Hamiltonian (8. x2 ) . requires a Lorentz-invariant quantum mechanical description as provided in Sect. E = E − (8. x2 . likewise. x2 ) = E ψ(x1 .56) pointing in the x3 -direction the symmetric gauge (8. x3 ) = E ΨE (x1 . j = 1. i. To obtain these eigenstates we introduce the customary dimensionless variables of the harmonic oscillator Xj = (8. x3 ) = exp(ik3 x3 ) ψ(x1 . the so-called gyromagnetic ratio of the electron. We will demonstrate in Sect.84) .85) The Hamiltonian (8. For a magnetic field (8.87) .81) (8. Accordingly.79) To obtain the stationary states. x3 ) . described by the operator S.4: Electron in Homogenous Magnetic Field 213 as the quantum mechanical magnetic moment operator for the electron (charge −e).e.

−j + 1. 1 2 since one phonon is annihilated and one created. x2 ) (8. x1 . x1 .91) We note that the operator a† a2 − a† a1 leaves the total number of vibrational quanta invariant.92) where Ψ(0. x1 . .95) (j) ˆ We want to choose the coefficients αmm such that (8. x1 . .93) ˆ The states (8.94) is an eigenstate of the vibrational Hamiltonian. 2 2 . m . it holds a† a1 + a† a2 + 1 1 1 2 Ψ (j. x1 .97) . (8. 2 (8.94) is an eigenstate of Ho ˆ Ho Ψ (j. j = 1. (8. m = −j. Such eigenstates can be expressed. i. m . x2 ) (8. . m. x1 . . x1 . 0. . m . m. .90) a† a2 − a† a1 1 2 . +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 . (8. attempt to express eigenstates in terms of vibrational wave functions a† 1 j+m Ψ(j. x2 ) = ( 2j + 1 ) Ψ (j. m . 1.e.e.94) is also an eigenstate of L3 . x2 ) = m=−j αmm Ψ(j. m . x2 ) = a† 1 j−m (j + m)! (j − m)! Ψ(0.89) . x2 ) is the wave function for the state with zero vibrational quanta for the x1 . If this property is..96) ˆ holds. 1 3 . x1 . such that 1 i a† a2 − a† a1 1 2 Ψ (j. aj = √ 2 Interaction of Radiation with Matter Xj + ∂ ∂Xj . x1 . through a combination of states j Ψ (j. however.. obeyed. in fact. m . (8. . x2 ) . (8. x2 ) = ω 2j + 2m + 1 Ψ (j. We. m . x1 . therefore. x2 ) . In order to cover all posible vibrational quantum numbers one needs to choose j. (8.as well as for the x2 -oscillator. x1 .214 Employing the creation and annihilation operators 1 a† = √ j 2 and the identity ˆ ω L3 = which can readily be proven. the total vibrational energy being ω(2j + 1). i. (8. x2 ) .92) represents a state with j + m quanta in the x1 -oscillator and j − m quanta in the x2 -oscillator.94) Since this state is a linear combination of states which all have vibrational energy (2j + 1) ω. m as follows: j = 0. .92) are not eigenstates of L3 . (8. 0. x2 ) = 2m Ψ (j. (j) (8.

(8.55) confines the particle to stationary bound states. The required rotation must transform the x3 –axis into the x2 –axis. x1 . Eq. t) = 1/4πr confining an electron with energy E < 0 to move in the well known orbitals of the hydrogen atom. m . which corresponds there to the angular momentum in the x2 –direction. x1 . (8. Using the explicit form of the Wigner rotation matrix as given in (5.11.97). π . a† .101) 2 2 where Dmm ( π . except for a factor 2 . m) in Sect.. m .10. is identical to the operator L3 introduced in (8.9. x2 ) . x2 ) defined in (8.9. t) introduced above. The first rotation contributes a factor exp(−im π ).5. If we identify 2 a† .11 (8. an example is the Coulomb potential V (r. . This implies that ˆ we can obtain eigenstates of L3 by rotation of the states Ψ(j. m. m. x2 ).102) We have identified.9. 5. the second rotation a factor dmm ( π ). 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 . i. a2 1 2 present notation ←→ b† .31). We assume that the scalar potential V in (8. m.84) above.55). The external radiation field is accounted for by the vector potential A(r. x2 ) (8. According to the derivation given there.5.10.11 such transformation is provided through π π (j) Ψ (j.100) J2 = 1 2 2i 1 ˆ i. Other radiation fields can .5.e. According to Sect. b† . 5. 5.e. cf. 5. 5. m. is in the notation of the present section 1 ˆ a† a2 − a† a1 . 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.11. b− + − notation in Sects.10. a1 . x1 . the eigenstates of (8.92) correspond to the eigenstates |Ψ(j. 5. x1 .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. 8. x1 .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. b+ .10. 0) Ψ(j. 5. x2 ) = Dmm ( .83) and confirmed the eigenvalues stated in (8.98) then the states Ψ(j. In the simplest case the radiation field consists of a single planar electromagnetic wave described through the potential (8.5. The connection with the present problem arises due to the fact that the operator J2 in Sect.(5. the states are eigenstates of the operator [we use for the operator the notation of Sect. the 2 2 latter representing the Wigner rotation matrix of Sect. x1 .8.309) yields finally Ψ (j.. thus.99) with eigenvalue m. a transformation moving the x3 –axis into the x2 – 2 axis.288)] 1 ˆ J3 = 2 a† a1 − a† a2 1 2 (8.

In fact..e. in this case. (8. i.41) e ˆ e p2 p·A ∼ 2me me me 2me 4 1 2 8πNω .105) for the case of radiationinduced transitions in atomic systems.103) (8. orthonormal basis. this assumption can be waved as our results below will not depend on it.105) in further calculations and to expand the wave function in terms of powers of VS (t) in a perturbation calculation.e. ωV (8. the term ∼ p · A(r. t) + A (r... as given in (8.107) 1 = 1 n=0 |n n| . t) − m 2m (8.e. Estimate of the Magnitude of VS We want to demonstrate now that the interaction VS (t).216 Interaction of Radiation with Matter be expanded through Fourier analysis in terms of such plane waves. . 1.105) where A(r.105). using (8. we discount the possibility of a continuum of eigenstates. t) is given by (8. Here the so-called unperturbed system is governed by the Hamiltonian Ho with stationary states defined through the eigenvalue problem Ho |n = n |n .104) (8. However. . i.. is much 2 (r. n = 0. (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.105) as follows4 .109) The reader should note that the estimates are very crude since we are establishing an order of magnitude estimate only. i. is much smaller than the eigenvalue differences near ˆ typical atomic bound states. and that the first term in (8. i.e. The states |n are thought to form a complete. 2 .42). can be considered a weak perturbation. one can estimate that the perturbation. we assume n|m = δnm and for the identity ∞ (8. For an electron charge q = −e and an electron mass m = me one can provide the estimate for the first term of (8. t). We first note.106) where we adopted the Dirac notation for the states of the quantum system. the term ∼ A second term in (8. t). 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. This result will allow us to neglect the larger than the second term.108) We assume for the sake of simplicity that the eigenstates of Ho can be labeled through integers.

We want to estimate now the second term in (8. u) p · u exp i(k · r − ωt) ˆ ˆ ˆ kV . one obtains for (8.. and me c2 ≈ A A 500 keV 2 ao ω ≈ 10−8 (8. i.42) it holds VS ≈ e m k. altogether.e.111) For ω = 3 eV and a corresponding λ = 4000 ˚ one obtains.117) In case that such perturbation acts on an electron and is due to superpositions of planar waves described through the vector potential (8.ˆ u 4πNk α(k.e. Using again (8. Nω = 1. with ao ≈ 0. Hence. i. (8.112) π λ me c2 and with e2 /ao ≈ 27 eV. one can neglect this term as long as the first term gives non-vanishing contributions..105) due to a radiation field by VS = − q ˆ p · A(r.105). and as long as the photon densities Nω /V are small. (8.112) yields e2 A2 ∼ 10 eV · 10−8 = 10−7 eV . me (8.118) . t) .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 . the first term in (8. 2me (8.110) where ao is the Bohr radius. hence. We can. a volume V = λ3 where λ is the wave length corresponding to a plane wave with frequency ω. Consequently.114) Employing for the second factor the estimate as stated in (8.116) This term is obviously much smaller than the first term. m (8. λ = 2πc/λ. e ˆ p · A ∼ 10 eV · 10−4 = 10−3 eV .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 (8.113) 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. Assuming a single photon.115) (8.105) for radiation fields can be considered a small perturbation.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. replace the perturbation (8.8.5 ˚.

u) has been added to describe eliptically or circularly polarized ˆ waves. Multiplying the latter equation by the operator exp i (8.118) is the form of the perturbation which.218 Interaction of Radiation with Matter where we have replaced ω in (8.120) (8. |Ψ(to ) = |0 (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) . For this purpose one needs to determine the state |ΨS (t) of the particle. describes the effect of a radiation field on an electron system and which will be assumed below to describe radiative transitions. n = 1.42) through k = |k| = ω. Perturbation Expansion The generic situation we attempt to describe entails a particle at time t = to in a state |0 and a radiation field beginning to act at t = to on the particle promoting it into some of the other states |n . The sum runs over all possible k vectors and might actually be an integral.122) where |ΨD (to ) = |0 .123) i = Ho exp − Ho (t − to ) (8.127) i ∂t |ΨD (t) = VD (t) |ΨD (t) . 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 . |n are defined in (8. A factor α(k.124) (8.106–8. The probability to find the particle in the state |n at time t is then p0→n (t) = | n|ΨS (t) |2 . Equation (8. .108) as the eigenstates of the unperturbed Hamiltonian Ho . (8.125) (8.. n = 0 at some later time t ≥ to . .126) Ho (t − to ) yields finally .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. under ordinary circumstances. Using i i ∂t exp − Ho (t − to ) and (8.121) (8. The states |0 . One seeks to predict the probability to observe the particle in one of the states |n . the sum over u involves the two possible polarizations of planar ˆ electromagnetic waves. . 2.

8. we define |ΨD (t) through the evolution equations i ∂t |ΨD (t) i ∂t |ΨD (t) i (2) ∂t |ΨD (t) (1) (0) = = = .137) |ΨD (t) = (1) (0) (8.137) One can readily verify that (8. to (8. consequently. In order to determine |ΨD (t) described through (8.137) follows |ΨD (t) = |0 . 8.133) (8.136) together with the initial conditions |ψD (to ) = |0 0 for n = 0 for n = 1.137) are consistent with (8. Employing this result one obtains for (8.121) expressed in terms of ΨD (t) is i p0→n (t) = | n|exp − Ho (t − to ) |ΨD (t) |2 .132–8.5: Time-Dependent Perturbation Theory where VD (t) = exp 219 i i Ho (t − to ) VS (t) exp − Ho (t − to ) . .128) We note that the transition probability (8.130) = exp − i n (t − to ) − to )]| = 1. We will consider here only the two leading contributions to |ΨD (t) . . 0 VD (t) |ΨD (t) (1) VD (t) |ΨD (t) (0) (8.138) 1 i t dt VD (t ) |0 .139) .133–8. From (8.132) where |ΨD (t) accounts for the contribution due to n-fold products of VD (t) to |ΨD (t) .127.135) i ∂t |ΨD (t) (n) VD (t) |ΨD (n−1) (t) (8. using |exp[− i n (t n (8.8. (8. 8.133. (8. n| (8. . .134.134) (8. = .127) we assume the expansion ∞ |ΨD (t) = n=0 (n) |ΨD (t) (n) (8. .128).131) p0→n (t) = | n |ΨD (t) |2 . (8. 2.129) n| and. . .137) can be solved recursively. Equations (8. Accord(n) ingly. Due to the Hermitean property of the Hamiltonian Ho holds n|Ho = i n|exp − Ho (t − to ) from which we conclude.

. + m=0 dt to to dt n|VD (t )|m m|VD (t ) |0 8. II III (8. The factor III in (8.142) combining an incoming and an incoming or outgoing wave. The resulting perturbation on an electron system. defines the direction of the E-field of the ˆ radiation. 8.220 This result. however. This factor will serve mainly the purpose of keeping in the following derivation all mathematical quantities properly behaved.. non-singular. is VS = ˆ ˆ V1 exp(−iω1 t) + V2 exp( iω2 t) eλt . (8. i. A factor exp(λt). 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. to → −∞ (8.135. λ → 0+ has been introduced which describes that at time to → −∞ the perturbation is turned on gradually.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.. in turn.141) + .e.6 Perturbations due to Electromagnetic Radiation We had identified in Eq. We want to apply now the perturbation expansion derived to such perturbations. note that uj .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 .41). A2 are defined through (8. such that the combined radiation field is decribed by the vector potential A(r.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).34.144) r is the position of the electron and p = ( /i) is the momentum operator of the electron. yields for (8. according to (8. that two planar waves simulataneously interact with an electronic system.142). t) = + A1 u1 exp i (k1 · r − ω1 t) ˆ A2 u2 exp i (k2 · r ˆ ω2 t) incoming wave incoming or outgoing wave (8..118). according to (8. 8. (8.144) 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. The coefficients A1 . λ → 0+ .144) describes the propagation of the planar wave.137) |ΨD (t) = (2) Interaction of Radiation with Matter 1 i 2 t t dt to to dt VD (t ) VD (t ) |0 . the direction of ˆ the propagation being determined by k = k/|k|. In (8. For the sake of including the effect of superpositions of plane waves we will assume.

140.8. yields n|ΨD (t) = − (2) 1 ∞ t t 2 λ→0+ t→−∞ lim lim ( ( dt m=0 to to dt i i ( ( − − − ω1 − i λ)t ω2 − i λ)t (8. 8.128).149) Using the definition of VD stated in (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 .143).146) = 1 λ→0+ t→−∞ i lim lim × t dt exp to i ( n − o − i λ) t iω2 t × .6: Perturbations due to Electromagnetic Radiation 1st Order Processes 221 We employ now the perturbation (8. For the 1st order contribution to the transition amplitude n|ΨD (t) = (1) 1 i t dt n|VD (t ) |0 to (8. According to (8.148) We consider now the 2nd order contribution to the transition amplitude. using (8. (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 .145) we obtain then. employing the perturbation (8. (8. (8. (8.143) to the expansion (8.141).130) and (for m = 0) i i exp − Ho (t − to ) |m = exp − for (8. (8.145) n|ΨD (t) (1) m (t − to ) |m .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 .128) one obtains k|VD (t) | = k|VS (t)| exp i ( k − ) (8.150) and.

We assume first that the first order transi(1) tion amplitude n|ΨD (t) is non-zero.e.152). in which case one can expect that it is larger than the 2nd (2) order contribution n|ΨD (t) which we will neglect.148) and (8.152) m=0 i ˆ ˆ n|V1 |m m|V1 |0 exp m − o − ω1 − i λ + − 2 ω1 − 2i λ)t n − o − 2 ω1 − 2i λ ( n o − ˆ ˆ n|V2 |m m|V2 |0 exp ω2 − i λ m− o ˆ1 |m m|V2 |0 exp ˆ n|V + ω2 − i λ m− o ˆ2 |m m|V1 |0 exp ˆ n|V + m − o − ω1 − i λ 1st Order Radiative Transitions i 2 ω2 − 2i λ)t 2 ω2 − 2i λ n− o i ( n − o − ω1 ω2 − 2i λ)t ω2 − 2i λ n − o − ω1 ( n o i − − ω1 ω2 − 2i λ)t ω2 − 2i λ n − o − ω1 ( n o − The 1st and 2nd order transition amplitudes (8. We also assume for the final state n = 0 such that n|0 = 0 and p0→n (t) = | n|ΨD (t) |2 holds.148) and ∗ |z1 + z2 |2 = |z1 |2 + |z2 |2 + 2Re(z1 z2 ) (1) (8.155) λ→0+ ( o We are actually interested in the transition rate. (8. the time derivative of p0→n (t)..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 ) (8.153) (8. For this rate holds . Using (8. respectively.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. i.131). provide now the transition probability p0→n (t) according to Eq.

Hence. Using lim x2 + 2 δ→0+ = π δ(x) (8. i. a property.159) leads to final states |n with energy n = o + ω1 . The δ-functions appearing in this expression reflect energy conservation: the incoming plane wave contribution of (8. (8. First.142) ˆ ˆ The matrix elements n|V1 |0 and n|V2 |0 in (8. due to the uncertainty relationship would introduce a considerable perturbation to the system. The result supports our definition of incoming and outgoing waves in (8.158) ˆ + | n|V2 |0 |2 δ( − ω2 ) Obviously. which is observed through the so-called spectral intensities of transitions |0 → |n .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+ (8.8. to cancel the third term in (8. in fact.144).160) leads to final states |n with energy n = o ± ω2 . The time average will be denoted by · · · t . The second contribution to (8.31) and (8. one should average the rate over many periods of the radiation.144).143.158) play an essential role for the transition rates of radiative transitions.142–8.156) such that the 1st order contributions of the two planar waves of the perturbation simply add. 8. For the resulting expression the limit limλ→0+ can be taken.156) ( o The period of electromagnetic radiation absorbed by electronic systems in atoms is of the order 10−17 s. any attempt to do so. is much shorter than could be resolved in any observation.157) one can conclude for the average transition rate k = d p0→n (t) dt t = 2π ˆ | n|V1 |0 |2 δ( n − n o − ω1 ) o (8. however. In case the matrix elements are non-zero. The result of such average is. the matrix elements can vary strongly for different states |n of the system. the two terms apearing on the rhs. . ˆ (8. describing either an incoming or an outgoing plane wave due to the vector potential A2 u2 exp i (k1 · r ˆ ω2 t) . due to the vector potential A1 u1 exp i (k1 · r − ω1 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..e. of this expression describe the individual effects of the two planar wave contributions of the perturbation (8.158).

166) over many periods of the radiation yields exp[ i (±ω2 − ω1 )] t = 0 and.168) absorption of 2 photons ω1 ..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.152). hence.e. We define ∞ z1 = m=0 ˆ ˆ n|V1 |m m|V1 |0 m − o − ω1 − i λ exp i × (8.161) To determine the transition rate we proceed again. |z1 + z2 + z3 |2 t = |z1 |2 + |z2 |2 + |z3 |2 (8.157) yields. in Eqs. i.158).. 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 ) (8.165) × ∞ 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. it holds p0→n (t) = | n|ΨD (t) |2 .k=1 j=k 2 2 2 2 ∗ zj zk (8. as we did in the the case of 1st order transitions.162) × and. similarly. exp[ i (±ω2 − ω1 )].g.224 2nd Order Radiative Transitions Interaction of Radiation with Matter We now consider situations where the first order transition amplitude in (8.163) × (8.167) Taking now the limit limλ→0+ and using (8.158). (2) (8.e.153) vanishes such that the leading contribution to the transition probability p0→n (t) arises from the 2nd order amplitude (8. e. Time average · · · t of expression (8.153–8.164) (8. in analogy to (8. (8.. ∞ − 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.

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

in case of a radiative transition from the 1s state of hydrogen to one of its three 2p states. and the integral is =1. according to (8. . λ → 0+ .172) δ( n − o − ω) (8. t) = 8πN ωV u exp ˆ ı (k · r − ωt) . =0. θ. in this range one can expand eik·r ≈ 1 + ik · r + . φ) × × eik·r e−r/ao Y00 (θ. 8.177) √ M = 6 0 ∞ 1 2π ia4 o r2 dr −1 dcosθ 0 ∗ dφ r e−r/2ao Y1m (θ. . plane polarized wave described through the complex vector potential A(r. the vector potential actually assumed is A(r. φ) = 6 r −r/2ao e Y1m (θ. θ. ˆ (8.m=0 (r. φ) = 2 − 1 2 1 −r/ao e Y00 (θ.179) .2 = m ωV Only the first term of (8.143) will contribute to the absorption process. 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 . For example. the second term can be discounted in case of absorption.173) V1. φ) a3 ao o 2p (8. φ) a3 o 1s (8. (8. .175) The matrix element involves a spatial integral over the electronic wave functions associated with states |n and |0 . the wave functions are (n. i. (8.174) We seek to evaluate the matrix element ˆ M = n| p eik·r |0 . (8. m denote the relevant quantum numbers) ψn=1.m (r. (8. t) = 2πN ωV u exp ˆ ı (k · r − ωt) + 2πN ωV u exp ˆ ı (−k · r + ωt) .226 Absorption of a Plane Polarized Wave Interaction of Radiation with Matter We consider first the case of absorption of a monochromatic. φ) (8.143.171) The perturbation on an atomic electron system is then according to (8. The absorption rate.178) These wave functions make significant contributions to this integral only for r-values in the range r < 10 ao .170) We will employ only the real part of this potential. to → −∞ (8.144) VS = where e 2πN ˆ ˆ p · u e±ik·r . However..158).176) ψn=2.e.

Using |k| = 2π/λ. V (r) ] 2me =0 = Using r = 3 ˆ j=1 xj ej 1 2me 3 pk [ r.182) one finds [ r. in case of the hydrogen atom. ] + [ r. V (r) = − . pk ] + ˆ ˆ k=1 1 2me 3 [ r.182).k=1 i m 3 pk ek = ˆ j.104) and.7: One-Photon Emission and Absorption 227 One can estimate that the absolute magnitude of the second term in (8. One refers to this approximation as the dipole approximation.183) (8. pk ] pk ˆ k=1 (8.174) yields kabs = 4π 2 e2 N ω V u · n| ˆ |0 ˆ r 2 m( o − m n| [r. (8. Using (8.8.175) can then be ˆ achieved and the differential operator p = i replaced by by the simpler multiplicative operator r.181) For the commutator in (8. This simplification results from the identity m ˆ p = [ r. Ho ] i where Ho is the Hamiltonian given by (8.k=1 i ˆ p m (8.182) (8.181) and (8. Ho ] |0 = i i n) n|r |0 . pj ] = i δkj one obtains ˆ [ r. Ho ] = ˆ2 p [ r.186) δ( n − o − ω) (8.178) holds eik·r ≈ A 1 1 + O( 50 ) such that one can approximate eik·r ≈ 1 . the value of the wave length for the 1s → 2p transition 2π c λ = = 1216 ˚ A (8.184) and the commutation property [xk . We are now in a position to obtain an alternative expression for the matrix element (8.185) from which follows (8. is Ho = ˆ (p)2 e2 + V (r) .182) one obtains M ≈ Insertion into (8.175).180) ∆E2p−1s and ao = 0. Ho ] = i m 3 pk ej δjk = ˆ j. 2me r (8.529 ˚ one concludes that in the significant integration range in (8.179) and other terms are never larger than 20π ao /λ.187) . Transition Dipole Moment A further simplification of the matrix element (8.

k frame is described by the angles ϑ.188) ˆ Here dk is the integral over all orientations of k.31–8.. ϕ2 = π/2. u2 . 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 (8.174) one can replace n − o by ω. there are ony two linearly independent directions of u possible.190) (8. u2 and k ˆ ˆ coordinate system. x3 -axes of a right-handed cartesian vectors u1 . We can. directions.35). one can express | u · n| r |0 |2 = |ρ|2 ( cos2 α + cos2 β ) = sin2 θ .192) and obtain ˆ dk u ˆ u · n| ˆ |0 ˆ r 2 2π 1 = |ρ|2 0 −1 dcosϑ (1 − cos2 ϑ) = 8π 3 (8. i. Integrating and summing accordingly over all contributions as given by (8. We have demonstrated in our derivation of the rate of one-photon processes (8. planar waves as defined in (8. The direction of this vector in the u1 .33) are orthogonal to each other. say u1 and u2 . three-dimensional vector. and all polarizations of the radiation. For the two angles α = ∠(ˆ1 . ϕ1 = 0 and of u2 by ϑ2 = π/2. ˆ u ˆ (8. k ˆ As a result. in practical situations.191) (8. ρ) and ˆ u β = ∠(ˆ2 . 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. The δ-function appearing in this expression.158) above that in first order the contributions of all components of the radiation field add. x2 . The unit ˆ ˆ ˆ ˆ can be chosen to point along the x1 . ˆ In order to carry out the integral dk we note that u describes the possible polarizations of the ˆ ˆ and u.189) where the factor 1/ arose from the integral over the δ-function. according to (8.193) . Accordingly. a very hot oven. Absorption of Thermal Radiation We want to assume now that the hydrogen atom is placed in an evironment which is sufficiently hot. the direction of u1 is described by the ˆ ˆ ˆ ˆ angles ϑ1 = π/2. such that the thermal radiation present supplies a continuum of frequencies.e. ρ) holds then u cosα = cosϑ1 cosϑ + sinϑ1 sinϑ cos(ϕ1 − ϕ) = sinϑcosϕ and cosβ = cosϑ2 cosϑ + sinϑ2 sinϑ cos(ϕ2 − ϕ) = sinϑsinϕ . will actually be replaced by a distribution function which reflects (1) the finite life time of the states |n .228 Interaction of Radiation with Matter where we used the fact that due to the δ-function factor in (8. obtain the transition rate in the present case by adding the individual transition rates of all planar waves present in the oven. ϕ. |0 . and (2) the fact that strictly monochromatic radiation cannot be prepared such that any radiation source provides radiation with a frequency distribution. hence.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.

. The value of | n| r |0 | determines the strength of an optical transition. there is also a contribution to the emission rate which is proportional to Nω . The quantum nature of radiation manifests itself in that the number of photons Nω msut be an integer.37 × 1019 1 ao | n|r|0 |2 × s λ λ2 . for Nω = 0.172.171) and perturbation (8. . by planar waves with vector potential (8. The transition dipole moment can vanish for many transitions between stationary states of a quantum system. For practical evaluations we provide an expression which eliminates the physical constants and allows one to determine numerical values readily. in particular.194) combined are typically somewhat smaller than (1 ˚/1000 ˚)3 = A A 10−9 . linear molecules. The radiation field is described. a problem in case of emission by quantum systems in complete darkness. finally. This difference between emission and absorption is due to the quantum nature of the radiation field. Otherwise. as for the absorption process. Nω photons before absorption. .197) o (tot) (8.194) 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 .173) contributes. the resulting total rate of emission bears a different dependence on the number of photons present in the environment. The corresponding process is termed spontaneous emission.195) = Nω 1.. (8. 2 .e.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 |. i. that emission occurs even if no photon is present in the environment.e..8.37 × 1019 × 3 3 c3 3 ao λ3 s λ and kabs where λ = 2πc n − (8. In case of a classical radiation field one would expect that emission cannot occur.189) to yield the total absorption rate kabs (tot) = Nω 4 e2 ω 3 | n|r|0 |2 3 c3 . However. However. the calculation of the emission rate proceeds as in the case of absorption. for atoms or symmetric molecules. However. The most intensely absorbing molecules are long. 1.173). the absorption rate is of the order of 109 s−1 or 1/nanosecond. Nω = 0. Emission of Radiation We now consider the rate of emission of a photon. in agreement with observations.196) The last two factors in (8. ˆ In case of emission only the second term V2 exp(+iωt) in (8. This poses. 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. 8. can be inserted into (8. This dependence predicts. however.. Accordingly. For this purpose we use ω/c = 2π/λ and obtain 4 e2 ω 3 32π 3 e2 ao 1 ao = = 1. (8.7: One-Photon Emission and Absorption 229 This geometrical average. i. Transition Dipole Moment The expression (8.

198) to determine the stationary distribution of photons ω in an oven of temperature T . exp[− ω/kB T ] = This equation yields Nω = 1 .199) Planck’s Radiation Law The postulate of the Nω + 1 dependence of the rate of emission as given in (8. using (8. e.203) i. Defining the rate of spontaneous emission 4 e2 ω 3 | n|r|0 |2 (8. Under stationary conditions the number of hydrogen atoms undergoing an absorption process |0 → |n must be the same as the number of atoms undergoing an emission process |n → |0 . accordingly. The number of atoms undergoing absorption in unit time are Nω ksp No and undergoing emission are (Nω +1)ksp Nn . = (8.. the well-known Planck radiation formula. 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.g. 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. The total rate of emission.230 Interaction of Radiation with Matter which is termed induced emission since it can be induced through radiation provided.198) (8. in lasers.e. . respectively. Nω + 1 (8.8 Two-Photon Processes In many important processes induced by interactions between radiation and matter two or more photons participate. We assume n − o = ω. To demonstrate this property we apply the transition rates (8. it must hold Nω ksp N0 = (Nω + 1) ksp Nn (8.200) where kB is the Boltzmann constant.202) It follows. exp[ ω/kB T ] − 1 (8. Let No and Nn denote the number of atoms in state |0 and |n .200).195) and (8.198) is consistent with Planck’s radiation law which reflects the (boson) quantum nature of the radiation field. For these numbers holds Nn / No = exp[−( n − o )/kB T ] (8. 8. respectively.. Hence.204) Nω .201) ksp = 3 c3 the rates of absorption and emission are Nω ksp and (Nω + 1)ksp . In the following we discuss several examples.

8. t) = Ao1 u1 cos(k1 · r − ω1 t) . under ordinary circumstances.173). finally. which is not necessarily the case. in case ˆ of absorption only V1 contributes. In this case one needs to choose the energy of the photons to obey n = o + 2 ω. The transition requires that the transition dipole moment n| r |0 does not vanish for two states |0 and |n . One obtains.106). dropping the index 1 characterizing the radiation. faced with the dilemma of having to sum over all intermediate states |m of the system.182) yields.8. (8. induce single photon absorption processes as described by the transition rate Eq.207) does not converge rapidly. even if n| r |0 vanishes. 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) .207) Expression (8. then expression (8. however. 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.205) The respective radiative transition is of 2nd order as described by the transition rate (8. ˆ (8. In applying (8. as described through the vector potential ˆ A(r.181) and using (8. (8. However. The resulting rate of the transition depends on 2 Nω . (8.208) has been prepared. The electron absorbs the radiation and emits immediately a second photon. 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. 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. If the sum in (8.207) does not provide a suitable avenue of computing the rates of 2-photon transitions. . however. Employing the dipole approximation (8.8: Two-Photon Processes Two-Photon Absorption 231 The interaction of electrons with radiation.206) − o − 2 ω) .172.207) one is.168) where the first term describes the relevant contribution.104) with stationary states |n defined through (8.207) 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. We assume that a planar wave with wave vector k1 and polarization u1 . 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. 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 (8.187). a transition between the states |0 and |n may be possible.

212) where N1 /V is the density of photons for the wave described by (8. later we will allow the quantum system to select appropriate values. The second term in (8. in the present case. In the present case we consider only scattering processes which absorb radiation corresponding to the vector potential (8..211) accounts for the emitted wave and.210) is accounted for by the following contributions of (8.105). o1 ˆ o2 ˆ (8.211) The first term describes the absorption of a photon and. 8. The relevant terms of the perturbation (8.e.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. ˆ The perturbation which arises due to the vector potential (8. hence.208) and the emission of the radiation in (8.210) A(r. however. u2 .208. (8. We know already from our description in Section 8. Let us assume for the present that the emitted photon has a wave vector ˆ k2 with cartesian components   sinθ2 cosφ2 k2 = k2  sinθ2 sinφ2  (8.210) The vector potential which describes both incoming wave and outgoing wave is a superposition of the potentials in (8. 8. The vector potential describing the emitted plane wave is then A(r.214) . according to the ˆ description of emission processes on page 229. t) = A+ u1 exp[ i (k1 · r − ω1 t) ] + A− u2 exp[ i (k2 · r − ω2 t) ] .208) and emit radiation corresponding to the vector potential (8.141).211) is stated in Eq. the amplitude A+ is given by o1 A+ = o1 8πN1 ω1 V (8. for the components of the wave (8. u1 .6 above that the absorption of the radiation in (8.208.208).211) is o2 A− = o2 8π (N2 + 1) ω1 V (8.209) cosθ2 where the value of k2 has been fixed.210). the wave characterized through k1 .210).105) using the vector potantial (8. t) = Ao2 u2 cos(k2 · r − ω2 t) . i.213) where N2 /V is the density of photons characterized through k2 . ˆ (8. the amplititude A− defined in (8.232 Interaction of Radiation with Matter We assume that the experimental set-up also includes a polarizer which selects only radiation with a certain polarization u2 . This expansion yields.

216) over all available modes of the field. Accordingly.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 . one needs to integrate the rate as given by (8.212.e. only these two terms contribute to the scattering amplitude. Only the second (1st order) and the third (2nd order) terms in (8.. me c2 (8.e. evaluating the time integrals in (8. Following closely the procedures adopted in evaluating the rates of 1st order and 2nd order radiative transitions on page 222–225. i. Inserting also the values (8. i.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. Hence.181) in stating this result. 8..213) for the amplitudes A+ and A− results o1 o2 in the Kramers-Heisenberg formula for the scattering rate k = N1 c 2 ω2 r (N2 + 1) dΩ2 u1 · u2 n|0 ˆ ˆ V o ω1 1 n|ˆ2 · p |m m|ˆ1 · p |0 u ˆ u ˆ n|ˆ1 · p |m m|ˆ2 · p |0 u ˆ u ˆ − + me m m − o + ω2 m − o − ω1 (8.8 · 10−15 m . one needs to carry out the integration 2 V(2π)−3 k2 dk2 · · ·.217) 2 Here ro denotes the classical electron radius ro = e2 = 2.218) .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 (8.215) correspond to scattering processes in which the radiation field ‘looses’ a photon ω1 and ‘gains’ a photon ω2 .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.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..228) S (0) = m S (0) = 2 m . in fact. − o )2 ( m − o )3 (8. a process in which the electronic state remains unaltered after the scattering.223) m Condition (8.. 2 . 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. .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 (8.227) 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. one can relate (8.219) (8.234 Interaction of Radiation with Matter The factor N1 c/V can be interpreted as the flux of incoming photons. it follows 2 dσ = ro (N2 + 1) dΩ2 |ˆ1 · u2 − S( ω) |2 u ˆ (8. Rayleigh Scattering We turn first to an example of so-called elastic scattering.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 + . consequently.e..221) Using |n = |0 and.224) = m 1 − o ( m ω ( ω)2 + + .221) suggests to expand S( ω) S( ω) = S(0) + S (0) ω + Using 1 − o ± ω one can readily determine m (8. for all states |m of the electronic system (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. ω1 = ω2 . . one assumes the even stronger condition ω1 << | o − m| . i. Accordingly. (8.

u ˆ ˆ ˆ ˆˆ i The commutator property [xj .232. Using again (8.182) 0|ˆ1 · p |m u ˆ 1 = 0|ˆ1 ·r |m u me ( m − o ) i This transforms (8.228) we obtain ˆ ˆ ˆ ˆ S (0) = 2me 2 m 0|ˆ2 ·r |m m|ˆ1 ·r |0 u u 0|ˆ1 ·r |m m|ˆ2 ·r |0 u u + m − o m − o (8. 8. Expression (8.108) for the identity operator. ˆ ˆ We want to prove now that expression (8.230) According to (8. pk ]|0 = u u ˆ j. as a compass. 8.234.233) m Employing again (8.227) vanishes. It explains. . u1 ·r ] |0 = 0 u ˆ (8. u u u u (8. For this purpose we replace p using (8.222) of the cross section for Rayleigh scattering dσ = 2 ro m2 ω 4 (N2 + 1) dΩ2 × e (8. for example.k=1 j. honey bees. 8. ˆ ˆ S (0) given in (8.228) provides then the first non-vanishing contribution to the scattering cross section (8.236) is of great practical importance.229) both for the u1 · p and the to u2 · p terms in (8.229) (8.e.232) Obviously. (8.235) and obtain the leading contribution to the expression (8.235) We can now combine eqs. ˆ We want to simplify first (8.k=1 (ˆ1 )j (ˆ2 )k δjk = u1 · u2 u u ˆ ˆ (8.222).234) where we used for the second identity the fact that u1 ·r and u2 ·r commute. pk ] = i δjk yields finally ˆ S(0) = 1 S(0) = i 3 3 (8..8: Two-Photon Processes 235 ˆ These three expressions can be simplified using the expression (8.222).224.229) 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 ) .226). For this purpose we apply (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 . the blue color of the sky and the polarization pattern in the sky which serves many animals. i. this term cancels the u1 · u2 term in (8.231) (ˆ1 )j (ˆ2 )k 0|[xj .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.108) yields S (0) = me 2 0| [ˆ2 ·r. m|ˆ1 · p |0 u ˆ 1 = − m|ˆ1 ·r |0 u me ( m − o ) i (8.182) for p and the expression (8.108) this is 1 0|ˆ1 ·r u2 · p − u2 · p u1 ·r |0 .8.

The first choice is ˆ   sinφ2 k2 × k1 (1) u2 = ˆ =  −cosφ2  (8.239) We will show below that this expression decribes the non-relativistic limit of Compton scattering. hence.236 Thomson Scattering Interaction of Radiation with Matter We consider again elastic scattering. however.244) .237) The resulting scattering is called Thomson scattering.243) cos2 θ2 cos2 φ2 for u1 = u1 .e. 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. u ˆ 1 ao .220) 2 dσ = ro (N2 + 1) dΩ2 |ˆ1 · u2 |2 . ao Bohr radius . u1 =  1  ˆ ˆ (8. i.238) (8. u2 = u2 ˆ ˆ ˆ ˆ for u1 = u1 . (8. |n = |0 and ω1 = ω2 = ω in (8.240). assume now that the scattered radiation has very short wave length such that ω >> | o − m| .220).. though.240) cosθ2 We choose for the polarization of the incoming radiation the directions along the x1 . To evaluate |ˆ1 · u2 |2 we assume that k1 is oriented along the x3 -axis and. u2 = u2 ˆ ˆ ˆ ˆ (1) (2) (2) (1) (1) (2) (1) (2) (2) (1) needs to be (8. for all states |m of the electronic system .and the x2 -axes     1 0 (1) (2) u1 =  0  .241) 0 0 Similarly.. u2 = u2 ˆ ˆ ˆ ˆ cos2 φ2 cos2 θ2 sin2 φ2 for u1 = u1 . u2 = u2 ˆ ˆ ˆ ˆ (8. (8. 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 .242) |k2 × k1 | 0 where the second identity follows readily from k1 = e3 and from (8. We want to assume. the emitted u ˆ radiation is decribed by the wave vector   sinθ2 cosφ2 k2 = k1  sinθ2 sinφ2  (8.

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

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

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

x2 . . The elements P ∈ SN can be represented conveniently as 2 × N matrices. 8) . .4) both permutations being obviously odd. The latter factorization has the essential property that the number of factors is either even or odd. 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 = . and explicitly the fact that any P ∈ SN can be given as a product of transpositions T (j. in particular. namely implicitly the group property. i. a given P can either only be presented by even numbers of transpositions or by odd numbers of transpositions. 5). For the following we will require only two properties of SN . . leaving all other particles unchanged. k) of the two particles transposed. k) are characterized by the two indices (j. the set of all permutations of N objects. For the permutations P1 and P2 given above holds P1 = T (3. 4) T (4. the top row representing the numbers 1 to N labelling the N particles under consideration. The group SN is then composed of two disjunct classes of even and of odd permutations. One calls permutations of the first type even permutations and permutations of the second type odd permutations.e. We can then state P2 = T (4. . xP (N ) ) . However. We will not discuss here at any depth the properties of the permutation groups SN .240 Many–Particle Systems repeatedly like degenerate states. 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. Transpositions denoted by T (j. A reader interested in the quantum mechanical description of N particle systems is strongly encouraged to study these representations. not any state can be chosen as to represent the many– particle system.3) the j-th index in the second row. 6) T (6.2). P2 = . i. (9. P2 = T (4.. is called a transposition. 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. (9. i. This property will be established now. . 3) T (7. . but in a different order. number k under the entry j of the first row indicating that particle j is switched with particle k. the representations in the space of N –particle wave functions are extremely useful in dealing with N –particle systems..e. The permutations we need to consider for a system of N particles are the elements of the group SN . and the second row showing the numbers 1 to N again. xP (2) . in notation (9.e.. . even though these properties. one can state P Ψ(x1 . k) = (4. .5) . xj . . . . . . xP (j) . (j. 5) T (5.e. ..3) P2 which affects only two particles. . . particles 4 and 5. . . . 5) (9. 5) in case of P2 . rather only states which obey a certain transformation property under permutations are acceptable. . k). i. Denoting by P (j) the image of j. xN ) = Ψ(xP (1) .

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

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

. .. . . . .1: Bosons and Fermions 243 The wave functions (9. nS ) and N = (n1 ..e. ..18) this wave function can also be expressed through the so-called Slater determinant x1 . (b) Show that for the states (9.1. a molecule or a crystal.17) Here the states (λ1 . . n2 . . (9.8).λN ) (9. . . nS ) holds S Ψ (N )|Ψ (N ) = j=1 B B δnj nj . nS ) = √1 N! P ∈SN P P x1 . n2 . . electrons in an atom..15) describe. . . Wave Functions with Fermion Symmetry in the Occupation Number Representation One can construct similarly a wave function which obeys the fermion symmetry property (9. . . . nS ) holds S ΨF (N )|ΨF (N ) = j=1 δnj nj . .20) . . the integers nj denote the occupancy of the single–particle state |j . . . . The important property holds that nj can only assume the two values nj = 0 or nj = 1. . . . for example.17) describe. .9). n2 . √1 N! (9. n2 . . . . . . . . φλ2 (x1 ) φλ2 (x2 ) . . . for two sets of occupation numbers N = (n1 . . These states form an orthonormal basis. . . n2 . i. n2 . xN |ΨF (n1 . . . x2 . Such wave function is x1 . . . The fermion states (9.16). x2 . xN |ΨFock (λ1 . Using the identity governing the determinant of N × N –matrices A N det(A) = P ∈SN P j=1 Aj P (j) (9. x2 . . . .e. nS ). These states form an orthonormal basis.. . for example. xN |ΨF (n1 . .16) Exercise 9. n2 . i. φλN (xN ) Here. λ2 . (9. . φλN (x1 ) φλN (x2 ) .9.λN ) on the rhs. . for two sets of occupation numbers N = (n1 . nS ) = φλ1 (x1 ) φλ1 (x2 ) . the photons of the electromagnetic field or the phonons in a crystal. λ2 .19) φλ1 (xN ) φλ2 (xN ) . . . . . . . For any value nj > 1 two or more of the columns of the Slater matrix are identical and the wave function vanishes. . . correspond to a particular choice of placing N particles consistent with the occupation numbers (n1 . ..15) holds (9.15) obey the boson symmetry (9. nS ) and N = (n1 .1: (a) Show that the states (9.

nj + 1. 9.16) one obtains (Aj )N = ΨB (N )| aj |ΨB (N ) = √ S N nj δnj nj −1 k=1 k=j δnk nk . . guaranteeing the correct permutation properties of the many–particle states.22) The operators aj and a† obey the commutation properties j [aj . . nS ) = j nj + 1 ΨB (n1 .15). . Creation and Annihilation Operators for Bosons We consider the operator aj defined through aj ΨB (n1 . . . nj − 1. . These operators allow one to construct and manipulate many–particle wave functions while preserving permutation symmetry.9).15). . nj . . . . . n2 .2: (a) Show that the states (9.21) we consider the matrix Aj corresponding to aj in the basis of many–particle states (9. (b) Show that the states (9.17) obey the orthonormality property (9. . n − 1. .15) and (9. . . [a† . . nS ) nj ≥ 1 0 nj = 0 √ . nj . .20). .244 Many–Particle Systems The representation of wave functions (9. . nS ) . . . n .24) [aj . . . . . . nS ). . nS ) = nj ΨB (n1 . .1. .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 . since both the N -particle wave function B (n . . . . Exercise 9. (9. . . are the so-called operators of 2nd quantization. . . . . .17) obey the fermion symmetry (9. To prove that (9.21) √ The factor nj appears here on the rhs. (9. .22) follows from (9. . . a† ] = 0 j k (9. a† ] = δj k . . 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 . .23) (9. For the adjoint operator a† holds j a† ΨB (n1 . . . . n ) are normalΨ 1 j 1 j S S ized according to (9. . Employing the superindices N and N introduced above and using the orthonormality property (9. . . n ) and the N − 1-particle wave function ΨB (n .25) .2 Operators of 2nd Quantization An important tool to describe mathematically systems of many bosons or fermions. . ak ] = 0 . (9. .

24). j (aj a† − a† aj )|Ψ(N ) = k k ( nj (nk + 1) − (nk + 1)nj )|Ψ(n1 . n2 . . . .29) |ΨB (n1 . . nS ) = |ΨB (N ) . n2 . n2 = 0. nS ) = 0 . . 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 .e.9. . 9. . the properties (9. S ˆ N |ΨB (n1 . (9.30) Of particular interest is the operator ˆ Nj = a† aj . We will prove the commutation properties (9. One can readily show using (9. 9. The creation operators a† allow one to construct boson states ΨB (N ) from the vacuum state j |0 = |ΨB (n1 = 0. (9. It holds for j = k (aj a† − a† aj )|Ψ(N ) = (nj + 1 − nj )|Ψ(N ) = |Ψ(N ) . . j j Since this holds for any |Ψ(N ) it follows [aj . .24).. . . . . (9.31) This operator is diagonal in the occupation number representation. nS ) . . . n2 .22) follows from (9. . .28) (9. . . . The related operator S ˆ N = j=1 ˆ Nj (9. nj − 1 . . . .e. nS = 0) as follows S (9. nS ) = j=1 a† j nj nj ! |0 . n2 . nk + 1 . .26) From this result one can immediately conclude that (9. . nS ) = j=1 nj |ΨB (n1 .27) Equations (9.29) yield the commutation relationship (9. n2 . i. a† ] = 1 For j = k follows similarly 1.27.33) is called the particle number operator since. (9.2: Operators of 2nd Quantization Let A† be the matrix adjoint to Aj .23) follow in an analogous way. . obviously. . .. j (9. the eigenvalues of Nj are the occupation numbers nj . for basis states |ΨB (n1 .34) = . .32) ˆ ˆ i. . (9. One refers to Nj as the occupation number operator.21) and vice versa.21. .22) ˆ Nj |ΨB (N ) = nj |ΨB (N ) . nS ) N |ΨB (n1 .

. nj+1 . nj+1 . in general.35) and where the derivative operation is defined through ∂f (A) = ∂A ∞ n=1 1 f (n) (0) An−1 .. .s. . .17. . Obviously..e.38) . . . . . 9.24) and aj |0 = 0 imply that the the properties (9. i.h.. nj+1 = 1.19) without affecting the fermion symmetry (9. nS ) = . . one needs to apply only the commutation propertes (9. . . Prove this by induction. . (9.30). . nj+1 .. . . 9.37) (9. . . (n − 1)! (9.24) and the property aj |0 = 0. . . φj (x1 ) φj (x2 ) . Exercise 9. . . . . xN |ΨF (n1 . j Exercise 9. nj . i.2: Show that the boson operators a† and aj satisfy j [aj . nS ) . 9. .38) corresponds to x1 . .. . xN |ΨF (n1 .. the fermion wave function changes sign when one exchanges the order of the occupancy. since they are induced through the algebra of a† and aj . showing that the property holds for nj = 0 and. . one does not need to worry ever about proper symmetries of wave functions.. .29.e. ..39) . in case that the single particle states |j and |j + 1 are both occupied. x2 . nS ) = − x1 . .9). 9. Before proceeding in this respect we need to account for the following property of the fermion wave function which applies in the case nj . .. .23) hold for the state defined through (9. . x2 .23. φj (xN ) φj+1 (xN ) . √1 N! . . f (a† )] j [a† . operators which can alter the occupancy of the wave function (9. x1 . As long as one starts from a wave function with the proper boson symmetry. in case it holds for nj .2. .246 Many–Particle Systems In using the boson creation and annihilation operators. To prove this property we notice that the l. from the vacuum state |0 or states |ΨB (N ) . nj (9.21. 9. nj .23. of (9. 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. .22. . x2 . .1: The commutation relationships (9.36) Creation and Annihilation Operators for Fermions We want to consider now creation and annihilation operators for fermions. it also holds for nj + 1. .2. φj+1 (x1 ) φj+1 (x2 ) .g. .. .. xN |ΨF (n 1 . . e.

. that occupancies are always ordered according to a monotonous increase of the single–particle state index. . . xN |ΨF (n1 .. . qj is the number of states |k with k < j which are occupied in a fermion wave function. . . . nj = 0. . We are now ready to define the annihilation operator cj for fermions in the single–particle state |j as follows cj |ΨF (n1 . . n2 = 1. n2 . vanishes. . nj+1 . 247 − √1 ! N . (9. .. of (9.. This requires one. . (9. . however.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.. . .19). . One adopts the convention that particles are created and annihilated by these operators always at the first position of the wave function. . . A proper example is the two particle fermion wave function x1 .2: Operators of 2nd Quantization The r. n3 = 0. . (9. nS ) as follows: j−1 qj = k=1 nk ... . .43) . n3 . nS ) = nj (−1)qj |ΨF ( n1 . nj+1 .e. . . . ..40) are identical. . x2 . This change of position is connected with a possible sign change (−1)qj where qj is defined for a given N = (n1 . nj .9. n2 .e. the expressions (9.19). 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). nj−1 . . in order to be consistent with this convention. n2 . . For this purpose one must adhere to a strict convention: the labelling of single–particle states by indices j = 1. n4 = 0. at the first column of the Slater determinant (9. . nS ) qj states occupied In case that the single particle state |j is not occupied. i. . i.e. . φj (x1 ) φj (x2 ) . Obviously. Because of the property of the determinant to change sign when two columns are interchanged... i. φj+1 (x1 ) φj+1 (x2 ) . .. . We will be adopting below the notational convention that qj corresponds to occupancies N = (n1 . n5 = 1. nj .42) i. . x2 |ΨF (n1 = 0. . . .. nS = 0) φ2 (x1 ) φ5 (x1 ) 1 1 = √2 = √2 [φ2 (x1 )φ5 (x2 ) − φ5 (x1 )φ2 (x2 )] .39) and (9. .s. 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. at which column of the Slater determinant (9. .e. particles are being created or annihilated. . the rhs. n2 .38) reads − x1 . φ2 (x2 ) φ5 (x2 ) (9. 2. . it is necessary to specify for a fermion wave function the order in which the single– particle states |λj are occupied. . . . nS ) = .. n6 = n7 = .h. φj+1 (xN ) φj (xN ) .40) . . nS ). . nj → 0.

We will prove now the anti-commutation property (9. We first show that (9. nj . . . . . 9. .19).44). . ck ]+ = 0 . The anti-commutation properties (9.49) The derivation involves realization of the fact that a non-zero result is obtained only in case nj = 1.46) first for the case j = k. Similarly one obtains cj c† |ΨF (N ) = cj (−1)qj (1 − nj ) |ΨF (n1 . . of (9.48) vanishes. nj+1 . otherwise. (9.17. n2 . . 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 . 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 . .45) follow in an analogous way and will not be derived here. . nj → 0 . (9. [c† .47) Let C† be the matrix adjoint to Cj . nj → 1 . the last factor on the rhs.46). We have used here the definition qj = k<j nk along with qj = qj . . c† ]+ = δj k k where we have used the definition of the so-called anti-commutators [A.44) (9.43). It holds c† cj |ΨF (N ) = c† (−1)qj nj |ΨF (n1 .248 The operator c† adjoint to cj exhibits the following property j c† |ΨF (n1 . n2 . . nS ) qj states occupied The operators thus defined obey the commutation properties [cj . .45) (9. We consider (9. nS ) = j j (−1)qj (1 − nj + 1)(−1)qj nj |ΨF (N ) = nj ΨF (N ) (9.48) From this follows (9. n3 . B]+ = AB + BA. . nj−1 . nS ) = j Many–Particle Systems (−1)qj (1 − 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 → 1. c† ]+ = 0 j k (9.46) [cj . nS ) = j (−1)qj (nj + 1))(−1)qj (1 − nj ) |ΨF (N ) = (1 − nj ) |ΨF (N ) (9.50) . .

. with eigenvalues equal to the occupation numbers nj ˆ Nj |ΨF (N ) = nj |ΨF (N ) ˆ Nj . x|j → φj (r) | 1 σ .. . . . coordinates x and index j into a space part and a spin part. σ 1 (σ = ± 2 ) .s.55) is called the particle number operator. i. .51.) = c† 1 c† 2 · · · c† N |0 . The stated property ˆ follows from the idempotence of Nj which can be derived as follows † ˆ2 ˆ Nj = c† cj c† cj = c† (1 − c† cj )cj = c† cj − c† c† cj cj = cj cj = Nj .e. is referred to as the occupation number operator.46). .9. .54) ˆ N = j=1 ˆ Nj (9. .50) yield ( cj c† + c† cj ) |ΨF (N ) = |ΨF (N ) .52) One can obtain similarly the same relationship in the case j > k. 2 (9. . .s..h. Equation (9. nλ2 . . nj = 0 . ( cj c† + c† cj ) |ΨF (N ) = 0 . i.57) On the l.58) . of this equation we meant to indicate only those N occupation numbers nλj which are non-vanishing. nk = 1 .h. 1}. nS ) k and. .56) Here we have made use of cj cj = 0 and c† c† = 0 which follows from nj ∈ {0. in the basis |ΨF (N ) . Correspondingly. nλ1 .51) (9. j → j.49) shows that the operator ˆ Nj = c† cj j (9. k k 249 (9.53) is diagonal in the occupation number representation. We j will often separate the states |j . hence. j j j j j j j (9.e. nk = 1 . . nλN . . hence.46).29)) |ΨF (. i. an operator cj must be on the left of an operator ck for j < k.e. 9. . . S (9.. nS ) k c† cj |ΨF (N ) = (−1)qj +qk −1 nj (1 − nk )|ΨF (n1 . j j For j < k one obtains cj c† |ΨF (N ) = (−1)qj +qk nj (1 − nk )|ΨF (n1 .2: Operators of 2nd Quantization (9.49) and (9.52) hold for any |ΨF (N ) one can conclude (9. Since (9. . ˆ It is an interesting exercise to demonstrate that Nj only has eigenvalues 0 or 1. 1 In the following we will apply fermion operators c† and cj only to electrons which carry spin 2 . λ λ λ (9. j j We finally note that the creation operators c† allow one to construct any fermion state |ΨF (N ) j from the vacuum state |0 defined as above (see (9. On the r. nj = 0 . the creation operators must operate in the proper order. for which purpose one needs to envoke only the algebraic (anti-commutation) properties (9. . . .

σr ) f (x) ψs (r.k=1 j=k (9. j = 1.60) involves generic operators – acting on xj and on xj . .59) and (9. (9. x ) which constitute G are ˆ called the associated generic two–particle operators.62) These matrix elements can be obtained by chosing the operators and many-particle states in the position representation. . r = 1. 2.F (N ) . i. ˆ ΨB. ˆ r|f |s = 1 ∗ ˆ d3 r ψr (r.59) and over pairs of particles in (9. N are not all identical. xN .6) above..60) is that the sum over particles in (9. one expects that the matrix elements can actually be expressed in terms of matrix elements ˆ involving solely the generic operators and the single–particle states. 9. . σs ) (9.250 Many–Particle Systems 9. The essential aspect of definition (9. These operators had been introduced already in Eq.F (N )|F |ΨB. The action of the operators boson and many–fermion states |Ψ is obviously described through the many–particle state matrix elements ˆ ΨB. S of single–particle spinorbital states (see page 241) and the operators are specified through the associated generic operators ˆ f and g . xk . Operators of the type 1 ˆ G = 2 N g (xj . 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. An operator N ˆ R = j=1 rj (xj ) ˆ (9. 9.3 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. . 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 ) . respectively – which are all identical. ˆ .F (N )|G|ΨB.F (N ) .e. xk ) ˆ j. 2. We adopt a similar expression to distinguish two–particle operators G and g .59) ˆ ˆ are called single–particle operators. respectively. . namely.60) ˆ correspondingly are called two–particle operators.59. .63) ˆ ˆ This expression has been specified for the purpose of these notes to distinguish F and f and is not common ˆ terminology. . 9.59.17). . x2 . The operators f (x) which constitute F are called the associated generic single–particle operators1 . (9.15. 9. . Since the many–particle states are built-up from a basis |r .17) and integrating over all particle space-spin coordinates x1 .61) is not a one particle operator as long as the N operators rj (x). . The operators g (x.15.

9. Notice. is t(x) = (− 2 /2m)(∂ 2 /∂r 2 ). s|ˆ|t. σr )ψs (r .. ˜ (9.k=1 j=k q2 |rj − rk | (9. are spin-independent as is evident from the factor δσσ . s|ˆ|t.71) .63. ∂r 2 ∂x2 ∂x2 ∂x2 1 2 3 (9.64) r.66) Here we adopt the notation for the Laplacian ∂2 ∂2 ∂2 ∂2 = + + . r|ˆ|u. (9. x ) ψt (r.67) ˆ In this case the generic single–particle operator. σt )ψu (r .64) We like to note an important symmetry of the matrix element r. t . An example for a two–particle operator is the Coulomb repulsion operator 1 ˆ V = 2 N j. ˆ 251 (9. ∂rj2 (9.66) and (9. u = s.65) This symmetry will be exploited repeatedly below. 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. one cannot switch the indices of one pair individually.62) in terms of the matrix elements of single–particle states (9. We will show that the boson and fermion creation and annihilation operators serve this purpose.59). σu ) . 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 . ∂r 2 (9. g g (9. s and t. 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. u which follows directly g from the definition (9. u.9. according to (9.64). 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) . i. s|ˆ|t.e.3: One– and Two–Particle Operators r. σs ) g (x. u = g d3 r ∗ ∗ d3 r ψr (r.70) Both operators. that the symmetry implies that one can switch simultaneously the pairs of indices r.69).69) ˆ In this case the generic operator is f (x) = δ(r − r).68) Another single–particle operator is the one–particle density operator N ρ(r ) = ˆ˜ j=1 δ(r − rj ) .

3 2 2 (9. Here. The generic operator is s12 = S1 · S2 = ˆ 1 1 S1. The non-vanishing matrix elements of the spherical components of s are (α = + 1 . s. Sj.77) δσ σ ( δσσ − δ−σσ ) The operators (9.74) =   N j=1 Sj  = 2 N Sj · Sk j. σ . α s r.3 .76) the single–particle state matrix elements of which are r. β|ˆ− |s.+ + S1. α|ˆ3 |s.− + S1. however. The ˆ matrix elements of the generic operator in terms of single–particle states are ˆ r. Sj is the spin operator for particle j with components (in the spherical representation) Sj. x ) = q 2 /|r1 − r2 |.73) is a one-particle operator. u. s = 1 σ..+ S2. The generic operator is given through the Pauli matrices.e. one can obviously extract the term j Sj to be left with a two–particle operator in the strict sense of our definition.− . α|ˆ+ |s. σ.− S2. σ | t |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. not in the strict sense of our definition above since the restriction 2 j = k does not apply in the summation.74) and (9.252 Many–Particle Systems which is also spin-independent. α s r. σ.75) is a two–particle operator. β = ˆ 2 2 1 −2) r. β s r. . β|ˆ3 |s.k=1 (9. σ .+ . (9. However. σ | s12 |t.72) As a final example of one– and two–particle operators we consider operators of total spin. u.3 S2. β s The operator S 2 = δrs = δrs = = + − 1 2 1 2 δrs δrs (9. The operator N S = j=1 Sj (9. i.77) are orbital-independent as evidenced by the factors δrs and δrt δsu in the respective formulas. σ ˆ + 1 δ− 1 σ δ 1 σ δ 1 σ δ− 1 σ 2 2 2 2 2 = δrt δsu + 1 4 δσσ 1 2 δ 1 σ δ− 1 σ 2 2 δ− 1 σ δ 1 σ 2 2 + . s. Sj. In this case the generic operator is v (x.

62).62) are correct one needs to compare the result derived with conventional derivations of the matrix elements2 . u . respectively.79) is opposite to that in the matrix element r. (9.3: One– and Two–Particle Operators Definition in Terms of Creation and Annihilation Operators 253 In order to express many–particle state matrix elements (9. i. The Matrix Elements ΨF (N )|c† cs |ΨF (N ) r We consider first the case r = s. Expressions (9. 9.176 where the matrix elements of fermion states can be found.64). namely cu ct . r S Ψ (N In case of r = s one obtains F )|c† cr |ΨF (N ) r = nr r=1 δnr nr . 9. s|ˆ|t. To show that the resulting values of the matrix elements (9.78.64) one replaces the operators F in (9.e.s.63) and (9.59) and G in (9. These rules will be provided below only for the case of fermions. pp.24) and (9.81) For the latter matrix elements simple rules can be derived from the algebraic properties of the boson and fermion operators (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.s=1 ˆ r|f |s a† as bosons r ˆ r|f |s c† cs fermions r r.u=1 S r.78) and (9.171 and pp.s.k=1 j=k S r. s|ˆ|t.t.80) (9. r ΨF (N )|c† cs |ΨF (N ) .60) as follows N ˆ f (xj ) → j=1 N 1 2 1 2 S r.u=1 (9.63.62) through single–particle state matrix ˆ ˆ elements (9.79) in the basis of many–particle states |ΨB.. u as defined in (9. s|ˆ|t.t. s|ˆ|t.83) t=1 t=r.79) have the following implication: The operators (9.23. The matrix elements are then actually those of the number operator c† cr which is diagonal in the occupation number representation. g and not ct cu .45. u cr cs cu ct g fermions (9. respectively. 9. In order to determine these matrix elements one needs to evaluate first the matrix elements ˆ of the generic operators r|f |s r. xk ) → ˆ j. 9.78) 1 2 g (xj .46). u a† a† at au bosons g r s † † r.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’.79) It is crucial to notice in the expression for the fermion two–particle operator that the order of the annihilation operators in (9. .9.F (N ) have the same values as the respective matrix elements (9.82) ΨF (N )|c† cs |ΨF (N ) = r cr 2 (9. and then in g a second step the matrix elements ΨB (N )|a† as |ΨB (N ) .s=1 S r.

u in this sum three possibilities exist Case 0 two of the four indices are different. ˆ ˆ Nr Ns .i.88) . we split the sum in (9.. Starting from the definition (9.’ (all indices different) that only tuples (r.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. u need to be considered.78) does yield the same results as the operator defined in (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. t).i.s=1 a. s. Case 2 all four indices are different.86) For the combination of indices r.i. s. (9. are ˆ ˆ ΨF (N )|F |ΨF (N ) = (1 − ns )(−1)qs nt (−1)qt s|f |t (9. for which holds r = s.84) The off-diagonal elements with nr = nr except for r = s. The anti-commutation property (9. 1 2 S r s|ˆ|s r c† c† cr cs . s.s=1 a.87) Here and in the following we denote by ‘a. (9. in which case ns = ns + 1 and nt = nt − 1 holds.e. those for which N and N differ in more than two occupation numbers.79) and using c† r we can write S 1 ˆ G = rs|ˆ|tu c† c† cu ct g r s 2 r. s. t=u 2 = 0.85) All other matrix elements vanish.d.e. 171) shows that the operator defined in (9. i. g r s r. c2 = 0 r (9. t. t.86) into three contributions. Comparision with the results in Condon&Shortley (pp.. t. 9. ˆ ˆ ˆ ˆ Accordingly. r = t etc. G = G0 + G1 + G2 . ˆ The first contribution G0 is 1 ˆ G0 = 2 S r s|ˆ|r s c† c† cs cr + g r s r. t.254 From (9.d. i. where each contribution corresponds to one of the three cases mentioned.i.45) yields together with the definition of the occupation number operator (9. u) are included in the summation for which all indices are different.59) when the matrix elements are determined between Slater determinant wave functions. (r.u r=s.d.s.t. Case 1 three of the four indices are different. (r.53) 1 ˆ G0 = 2 S r s|ˆ|r s − r s|ˆ|s r g g r. namely.d. s).82.s=1 a.

s.t.d. this contribution obviously behaves similarly to the off-diagonal part (9.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. A similar contribution does not arise for F .s. 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.d.d.3: One– and Two–Particle Operators 255 Obviously.t<u S r. N = N or nr = nr for all r.s.d.t.u r<s. ˆ Case 0 This case covers diagonal matrix elements.t=1 a.d. s|ˆ|s. As we will find. S r.84) ˆ of the one–particle operator F . 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.s.t=1 a.88).t>u r s|ˆ|t u − r s|ˆ|u t g g c† c† cu ct . s − r.i.t=1 a.s r.u r>s. contributes in this case. r s (9. One obtains 1 ˆ ΨF (N )|G|ΨF (N ) = 2 S nr ns r.i. nt . s (9.s. r s|ˆ|r t − r s|ˆ|t r g g ˆ c† ct Nr . Commutation of ct cu according to (9. Only G0 .i.e.91) ˆ G2 = r. ns .92) ˆ This contribution to G has non-vanishing matrix elements only when N differs from N in four ˆ occupation numbers nr . r s|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 .t=1 a. s|ˆ|r. (9.t>u S r.s.45).93) . S r.u r>s. nu .t.90) ˆ Since Nr is diagonal. i. this part accounts for all diagonal contributions to G. ˆ The second contribution G1 is ˆ G1 = + + + 1 2 1 2 1 2 1 2 S r.t.s.86). r g g .u r>s. this contribution to the two–particle operator is diagonal in the occupation number ˆ representation.s.89) S r.s.t=1 a. F (N )|G|ΨF (N ) .. given in (9. For this purpose we consider three ˆ We can state now the matrix elements Ψ cases which actually correspond to the three cases considered below Eq.i.s. employing the symmetry property (9. g r s (9.65).t.u r<s.i. (9.t>u S r.9.

u < v. respectively. s|ˆ|u. t − r.75) which are.96) .73) and S 2 given in (9. Spin Operators ˆ ˆ In this section we will briefly consider the spin operators S given in (9.90). the ‘−’-sign applies for s < r.90) from (9. v for which holds ˆ ns = ns + 1. r g g (9.1: 9. s|ˆ|r.3. Only G1 .2: 9. u < v or s > r. Derive the non-vanishing matrix elements (9. contributes in this case. t. u > v or s > r. nv = nv − 1. particle number is conserved.3. 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. nt = nt + 1. Only G2 .94) Case 2 This case covers off-diagonal elements with nr = nr except for r = s. s|ˆ|v. given in (9. nu = nu − 1. contributes in this case.3. u. Derive (9. u > v.3: 9.e.92).91).3. Furthermore. The matrix elements of the corresponding generic operators are (9. One obtains ˆ ΨF (N )|G|ΨF (N ) = (−1)qs +qt (1 − ns )nt r r=s. Derive (9.95) from (9.92).74) and (9. one–particle and two–particle operators.256 Many–Particle Systems Case 1 This case covers off-diagonal elements with nr = nr except for r = s. u g g (9.89).95) In the latter formula the ‘+’-sign applies for s < r.62) for bosons. given in (9.76).3.5: Derive (9.4: 9.. r=1 r=1 Exercise Exercise Exercise Exercise Exercise 9. ˆ ˆ Show that the matrix elements of F and of G conserve particle number. i. One obtains ˆ ΨF (N )|G|ΨF (N ) = ± (−1)qs +qr +qu +qv (1 − ns )(1 − nr )nu nv r. for ˆ non-vanishing matrix elements ΨF (N )|G|ΨF (N ) at most four occupation numbers nr and nr can differ.92) from (9. t for which holds ˆ ns = ns + 1 and nt = nt − 1. All matrix elements which are not covered by the three cases above vanish. We like to express these operators through ˆ fermion creation and annihilation operators. it holds S nr = S nr . s|ˆ|t. In particular.t nr r. v − r.

atoms and nuclei.98.96. v r s 2 r.3.99) will be used. The outcome of these studies is that even when interactions between components (particles) are simple. v c† c† cu ct . that the two–particle operator representing interactions between particles induces a variety of interesting phenomena. the concerted behaviour of interacting systems can deviate qualitatively from that of systems made up of independent components.t. In fact. the spin part of the single–particle states is represented by α = 1 and β = − 1 . Examples is the laser action. u refer only to the orbital part of the single–particle basis.s=1 r.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.. 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.s=1 c† c† crβ csα .97). If it were not for the two–particle contribution to the Hamiltonian. superconductivity. s|ˆ|u. evaluation of their stationary states and excitation energies. in crystals. the two–particle operator makes the description of many–particle systems a very hard problem. 9. e. rα sβ (9. however. Actually. but also ordinary phenomena like freezing and evaporation. e.s.u=1 σ. 9. s|ˆ|t.6: Derive (9. molecules.g. the study of problems governed by Hamiltonians of the type (9.9.g. 2 2 Exercise 9.s=1 σ 1 ˆ rσ r|t|s c† csσ + 2 S r.σ (9. Often.97) The summation in (9.4 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.99) has preoccupied an important part of all intellectual efforts in Theoretical Physics during the past fifty years. the description of many–particle systems. The fortunate side of this is.98) In many cases the Hamiltonian is spin-independent and the equivalent notation S H = r. u c† c† cuσ ctσ v rσ sσ r. would be a simple exercise in linear algebra. 9. 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. In the latter case the indices r.96) and (9.97) is over the orbital states.t. the systems investigated involve a macroscopic number of components such that statistical mechanical approaches are envoked. t. The great generality of the concepts developed in the context of many–particle systems can be judged from the fact . Unfortunately. which represents the interactions between particles. s.u S (9.s..

e.102) U ∗rm Urn = δmn . electrons and nucleons.. m = 1.J. would not exist. What might happen? Independent-Particle Hamiltonian A many–fermion system governed by an independent-particle Hamiltonian. Chemistry essentially would know only one element and Life would not exist.. . r=1 3 Urm U ∗sm = δrs m=1 (9. where S |m) = r=1 Urm |r . ranging from nuclei to the molecules of living systems . 1989) . . 2. namely of neurons3 . i. The electronic properties of atoms with different numbers of electrons would differ little. S}. We will restrict our description in the following to systems composed of an even number (2N ) of particles and to spin-independent interactions. . The Hamiltonian of such system is S Ho = r. Exercise 9.258 Many–Particle Systems that today these concepts are providing insight into the functioning of biological brains.e. another prototypical systems of many strongly interacting components.103) An account of some of these efforts can be found in Modelling Brain Function – The World of Attractor Neural Networks by D. New York. (9. Amit (Cambridge University Press. Before continuing one may finally point out that the material world as we know it and as it establishes the varieties of natural substances. can be described in a rather straightforward way. water-filled jar all electrons of water turn their fermion nature into a boson nature.100) in a form in which the factors c† csσ become rσ occupation number operators.4. The Aufbau principle. If it were’nt for the fermion nature. all material systems would condensate into states which would depend little on particle number. a Hamiltonian without a two–particle operator contribution accounting for interactions among the particles.101) The unitary property of Urn U† U = U U† = 1 1 or S S (9.s=1 σ ˆ rσ r|t|s c† csσ (9. according to which nuclei and atoms change their qualitative properties when they grow larger. ultimately depend in a crucial way on the fermion character of its constituent building blocks. In such a representation Ho is diagonal and the stationary states can be stated readily.1: Imagine that in a closed.100) ˆ ˆ We will denote the matrix elements of t as r|t|s = trs . Transformation to a New Set of Creation and Annihilation Operators Our aim is to represent the Hamiltonian (9. 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). i. {|m).

103) allows one to express [c. d† ]+ = sσ nσ r. d† ]+ = 0 (9. (9. of (9. behave exactly like the operators c† and crσ behave for states |ΨF (N ) .4: Independent-Particle Models allows one to express conversely |r in terms of |m) S 259 |r = m=1 ∗ Urm |m) .f. (9.h.110) can be demonstrated similarly.46).110) mσ nσ [dmσ . they are Fermion creation and annihilation operators. csσ = n=1 Usn dnσ . ∗ ∗ Urm Urn . i.111) can be written.s.111).112) . nσ We demonstrate here the anti-commutation property (9.105) where ˜ ˆ tmn = (m|t|n) = S ∗ trs Urm Usn = r. the operators d† and dnσ .45. accordingly.101) are occupied.46) as c† and crσ . d† ]+ = δmn δσσ nσ (9..9.s=1 ˆ U † tU mn (9.and two-particle operators hold in an analogous way for d† and dnσ . . The l. (9. [dmσ .111) and.108) obey the same anti-commutation properties (9. dnσ = r=1 ∗ Urn crσ (9.e.104)] S S ∗ Urm d† mσ m=1 c† rσ = .100) yields ˜ Ho = tmn d† dnσ mσ where d† nσ = r=1 S S (9. the expressions drived above for the matrix elements of one. using (9. hence.s r The unitarity property r ∗ Urm Urn = δmn yields immediately identity (??).101) of states |r . rσ namely.107) Urn c† rσ . 9. dnσ ]+ = 0 . of N -Fermion states in which single particle states |nj ) as defined in (9.106) which together with (9.113) Urm Usn [crσ . In a basis of states N d† j σj |0 n j=1 (9. the anti-commutation properties (9.108) and (9.104) One can. expand S trs = mn ∗ ˜ tmn Urm Usn (9.109) The operators (9. The unitary property (9. (9. for nσ rσ example. [d† .108) These operators describe the creation and annihilation operators of fermions in states |n) which are linear combinations (9. c† ]+ = δσσ [dmσ .

.. 1 as well as a fermion in a spin state | 1 . − 1 .115) ˜ and involves solely occupation number operators N nσ = d† dnσ . 1 and |β = | 1 . It is.e. n = 1. 2. σ2 . (9.115) with eigenvalues 2N E(n1 . . 2 2 2 2 i.117) are eigenstates of (9.. j j (9. This equation poses the eigenvalue problem for the hermitian matrix (trs ). The condition (9. i.116) which follows from (9. We apply for this purpose (9.115) can be stated immediately since any many–particle wave function in the occupation number representation is suitable. .260 Diagonal Representation Many–Particle Systems The transformation (9.. is obtained. . − 1 is called a closed 2 2 2 2 . Vn = (U1n . S (9. ˜ tmn = In this case Ho has the desired form S n δmn . USn ). . transformation matrices Urn and energy values n .114) Ho = n=1 † n dnσ dnσ (9. i.e.106) diagonal. n2N . .106) and the unitary property of Urn .119) In this state the N lowest orbital eigenstates of trs are occupied each by an electron with spin |α = | 1 . . .106) is equivalent to S trs Usn = s=1 n Urn ∀r.e. σ1 . The corresponding eigenvalues n . Before proceeding we like to address the issue how the new representation. is N |Φo = j=1 d† α d† β |0 . Eigenstates of (9. the so-called ground state.57) to the case of 2N particles. σj ) = (nk . i. .e. The 2N fermion state 2N d† j σj |0 n j=1 where (nj . S are T real. r = 1.114) together with (9. . ..101) gives us the freedom to choose the matrix (9. A non-degenerate orbital state which is occupied by two fermions. σk ) for j = k (9. . (9. σ2N . U2n . a simple matter to nσ state many–particle states in the new representation. hence. 2. n2 . the associated ( properly normalized) eigenvectors Vn are the columns of (Urn ). ) = j=1 nj .118) Ground State In case of an ordering m < n for m < n the state of lowest energy. a fermion in a spin state | 1 .

120). β = 2 |α.n=1 d† d† dmβ dnα mα nβ (9. i.n=1 m. α .e. ˆ S 2 ||β.124) . α |β. We will demonstrate now that this property endows the ground state with total spin zero. 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. β ˆ are eigenstates of S 2 given in (9.123) ˆ which acts only on the N − 1 closed shells (except for N ) of the two states and has an eigenvalue ˆ N + 0 − (N − 1) = 1 in both cases. α N +1 − N .120) where the occupation number operators refer to the single-particle states |m). ˆ one can conclude that |Φo is an eigenstate of S 2 with eigenvalue zero. α = 2 |β.n=N (9.4: Independent-Particle Models 261 shell. The two states |α. α |α. namely. However. is a singlet state.e. 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. the third term − S m. both with eigenvalues 2 ˆ S 2 |α.120) is particularly simple for closed shells.n=1 m. i. There are four such states.n=1 d† d† dmβ dnα mα nβ (9. The ground state has only closed shells. of ∆E = and |β.121) All four states have the same excitation energy. |α. Alltogether.97). (9. energy above the ground state. 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. 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.9.n=1 cmα cnβ cmβ cnα does not give a contribution if m = n and either |m) or |n) are closed shells. 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). The action of the operator (9. In case of m = n it gives a contribution ‘1’ for each closed shell which cancels the contribution of the first term.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.97).n=1 † † if either |m) or |n) are closed shells. Obviously. β |β.100). Similarly.. the states differ in their spin character.. The spin operator in the present representation of single-particle states is 1 1ˆ ˆ S2 = N + 2 4 S S ˜ ˜ ˜ ˜ (N mα − N mβ )(N nα − N nβ ) − m. β . 1 S 4 m.

. the linear combinations 1/2(|α. c† +1 σ = c† σ 1 2N (9.125) ˆ Contributions to S 2 acting on states |N + 2 . 2.130) . however.e. β and |β. r = 1. . β both with eigenvalues 1. m |Φo m m m γσ. .122) is correct. 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. mα nβ (9. One can. . for example. etc. for |α. The action of Σ N +1 ˆ Σ3 |α.126) which follows from the occurence of squares of fermion operators which. (9. .100) outlined above we consider a system of 2N electrons which move in a set of atomic orbitals |r . The system is assumed to posses a 2N –fold symmetry axis and interactions connect only orbital states |r → |r ± 1 . The same result holds for |β.4. 2N ) drσ = √ 1 2N 2N e−irnπ/N cnσ n=1 (9.2: Construct four operators O ˆ O ˆ such that O ˆ S3 . We identify the states |2N + 1 = |1 to avoid cumbersome summation limits. . . .262 and ˆ Σ3 = − m. α ± |β.n=N Many–Particle Systems N +1 d† d† dmβ dnα . vanish. . α . The two states |α. hence. ˆ The remaining states |α. |β. conclude that (9. 2. α and |β. β .σ d† +1. also eigenstates of Σ2 . β are not eigenstates of S 2 .n=N d† d† dmβ dnα d† +1α dN β |Φo = 0 mα nβ N (9. α as triplet states.σ = (9. of course.. 2N which are located on a ring.σ dN σ N σ. β =− m. do not need to be considered since they ˆ give vanishing contributions.129) The symmetry of the Hamiltonian suggests to choose a new representation defined through the operators (r = 1. β qualify as eigenstates. of course. one with eigenvalue zero (singlet) and one with eigenvalue ‘2’ (triplet). i. 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 . α are. ˆ Exercise 9.127) ˆ are triplet and singlet states. ˆ 3 on the two states produces . |N + 3 . ˆ An eigenvalue ‘2’ of the spin operator S 2 identifies the states |α.128) The cyclic property of the system is expressed through c2N +1 σ = c1 σ .

.138) ˜ trs = from which follows ˆ Ho = eirπ/N + e−irπ/N 2N δrs (9. vanishes for integer s.130. . 9.h.s.135) c† nσ 1 = √ 2N e−irnπ/N d† rσ r=1 (9. 2N.132) an identity which can be derived employing the well-known expression for a finite geometric series m as = a s=1 1 − am .140) The Hamiltonian is now diagonal and the construction outlined above can be applied. 1−a (9.132) one can conclude 2N ei(s−r)nπ/N + eisπ/N n=1 1 2N 2N ei(r−s)nπ/N .4: Independent-Particle Models d† = √ rσ 1 2N 2N 263 eirnπ/N c† nσ n=1 √ 1 eirnπ/N .139) −2t cos r=1 σ rπ N d† drσ .134) which.131) constitute a unitary transformation U defined through Urn = unitarity property follows from 2N eirsπ/N = 0 r=1 for 0 < s < 2N . . thereby.132) gives for r = 0. one obtains the new representation of Ho 2N ˆ Ho = −t r. (9. indeed.131) The (9. . .109)] ¸ cnσ = √ 1 2N 2N eirnπ/N drσ r=1 2N (9.136) ˆ and. 4N. n=1 (9. rσ (9. One can then express[. eisπ/N 1 − e2N isπ/N 1 − eisπ/N (9. (9.s=1 σ ˜ rσ trs d† dsσ (9. of (9.133) Application to the l.9. 2N (9.137) where 1 ˜ trs = e−irπ/N 2N Using (9.f.

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

The states |m are connected with the states |r through (m = 1.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. u refer to the initial basis of single-particle states |r . . s|ˆ|t.σ and derive a single-particle operator which approximates this Hamiltonian. 2. |r = m=1 ∗ Urm |m . This is done as follows: ˆ Vmf (|Φo (U) ) = 1 2 S r. q. as in in (9. 2. . S is a unitary matrix. . m = 1. The states |m serve to define a reference state of the system ˜ N |Φo (U) = j=1 d† α d† β |0 . .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. p. t. S on which (9.145) which will play a pivotal role: it is this the state in which the fermions are assumed to be moving and relative to which the mean interactions acting on individual fermions is determined. 2.145). U = ( Urm ) as defined in (9. 2. r = 1. We will adopt a second set of single-particle states {|m .144) defined as in (9. . S} the elements of which will be labelled by indices ˜ m.142) r|t|s c† csσ + rσ sσ 2 r. j j (9. .146) . . the corresponding U = ( Urm ) forms an S × N matrix and. .s. naturally.143) with r. 2. . . .u=1 σ. u v r. . s. ˜ (9.u=1 σ. S) ˜ S S |m = ˜ r=1 Urm |r .s=1 σ r.t. . m = 1.t.108). .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 = (9.142) over this reference state to turn the contribution into an effective one-particle operator.9. s|ˆ|t. .s. N . 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. is not unitary.142) is based. . u c† c† cuσ ctσ v (9. If one restricts m = 1. n. In our formulas below we will adopt strictly the convention that indices r.

n. n c† ˜ v ˜ sσ m.146) accordingly. ˆ (9.u=1 σ.n=1 σ.σ S dmσ dnσ + r. n = ˜ ˆ ˜ s.n.150) r. u c† d† dnσ cuσ ˜ vn rσ mσ r. u ˜ ˜v m. n c† ctσ δσ σ ˜ v ˜ rσ m≤N r.266 Here ··· denotes the average O = Φo (U)| O |Φo (U) . for example. m| v |t.m.t.u=1 σ. For the averages ··· in (9.147) 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.n=1 σ. m|ˆ|˜ . s|ˆ|m.148) are obtained in a similar way by transforming only two of the four single-particle states into the new representation.m.s.m. m|ˆ|t.σ S 1 − 2 where.u=1 σ.n. 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.s. Many–Particle Systems (9.n. n c† d† dnσ ˜ v ˜ rσ mσ r. s|ˆ|˜ . n c† c† v ˜ ˜ rσ sσ r. s| v t.148) r. n|ˆ|t.σ . m.σ S d† d† mσ nσ cuσ ctσ + r.t. by means of the rules (9.t=1 σ.σ S d† dnσ mσ ctσ (9. s|ˆ|t.σ S ctσ + m.u ∗ r. d† d† mσ nσ dmσ dnσ d† dnσ mσ = = = = and.148) one obtains.149) The remaining matrix elements appearing in (9.s. m|ˆ|t.t=1 σ.84). hence. u c† ˜ vn sσ m. u Usm Uun . 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.σ S d† dnσ cuσ mσ − r.

u=1 σ.u ( 2 r.152) By means of expression (9. 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.2.u=1 σ.m. ˜ v ˜ sσ m≤N s.σ r.. u c† cuσ ˜ vn rσ (9. s )  ˆ ˆ  m≤N ∗ Utm Uum   (9.153) rσ r.s=1 σ ˆ r|t|s + r| vmf (|Φo (U) ) |s ˆ c† csσ rσ (9. s|ˆ|˜ ..S. as introduced here.151) Carrying out the sum σ δσσ = 2 leads to a factor 2 for the first two terms. m|ˆ|˜ .t=1 σ (9. t| v |u.σ S r. t| v |s.142) by Hmf (U) defined through S ˆ Hmf (U) = r. s|ˆ|t. The one-particle operator (9. n c† ctσ δσ σ ˜ v ˜ rσ − r. .154) ˆ The mean field approximation replaces then the Hamiltonian (9. m|ˆ|t.152) can then be written in the form ˆ Vmf (|Φo (U) ) = r| vmf (|Φo (U) ) |s c† csσ ˆ (9. n c† ctσ . At this point this state is completely arbitrary. u − r.m.s σ where r| vmf (|Φo (U) ) |s = ˆ t.149) for the matrix elements in the mixed representation one can state the right hand side explicitly in terms of Urm .u=1 σ S − m. m|ˆ|˜ . u c† cuσ ˜ vn rσ m≤N r... m=1.155) which is a function of the S × N –matrix U = ( Urm )r=1.. The mean field approximation. u = s. s| v |t.t=1 σ.156) which defines the reference state |Φo (U) .N (9. leaves open the question how the reference state should be chosen.2.σ S + − r. Exploiting the symmetry r. u c† cuσ δσσ ˜ vn sσ m≤N s.. r| v |u. We will address now a proper choice of the reference state.5: Self-Consistent Field Theory 1 2 1 2 1 2 S 267 m.9.

160) . often also referred to as the Hartree-Fock approximation. . This procedure. r = 1. .e.145). 2. is the lowest energy independent-particle ground state one can construct. The state thus determined is referred to as the self-consistent independent-particle ground state. . the independent-particle nature stemming from the fact that the functional form of the ground state. can achieve this goal only iteratively.142). 2. (9. m = 1. . i. We will argue below that the self-consistent field ground state. . m = 1.145). steps (2) and (3) are being repeated M times until the procedure converges.s=1 σ S ˆ rσ r|t|s c† csσ (9.6 Self-Consistent Field Algorithm The Self-Consistent Field (SCF) approximation. 2. . . Let us state now the construction of the self-consistent field ground state in more detail. M = 1.: 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 . The procedure determines the reference state as the ground state of the independent-particle Hamiltonian (9.158) where the labels m are ordered to obey the condition (1) m < (1) n for m < n .e.159) One defines the reference state |Φo (U(1) ) using the definition (9.155) following the method outlined in Section 9. however. S . 2. . and obtaining in a step (3) the ground state of this one-particle Hamiltonian according to the construction in Section 9.. (9. . is exact only for an independent-particle Hamiltonian. . S . 2. i.4 and defining this as the new reference state.4. determining in a step (2) the corresponding mean field Hamiltonian (9.119. assuming in an intial step (1) a properly chosen reference state.s=1 σ S ˆ r|t|s + r| vmf |Φo (U(M ) ) ˆ |s c† csσ rσ = r. 9. under conditions which often are realized. S .. (9. .155).s=1 σ ˆ r|hmf |s c† csσ rσ (M ) (9. SCF-Algorithm. the reference state resulting from step 2M + 1 (within numerical error) is equal to the reference state resulting from step 2M − 1. i. SCF-Algorithm.268 Many–Particle Systems 9.157) and determines the associated diagonal representation defined through S (1) ˆ r|t|s Usm = s=1 (1) (1) m Urm . Step 2M. S. r = 1. is based on the mean field approach and provides an algorithm to construct a reference state |Φo (U) for a 2N fermion system described by a spin-independent Hamiltonian (9. . . . . Step 1: Choosing an Initial Reference State One defines the Hamiltonian ˆ (1) Hmf = r.e.

ˆ (b) How can the expectation values of the operator Prr be interpreted. i. s ) ˆ ˆ m=1 Utm ∗(M ) (M Uum ) (9. ˆ Exercise 9. r = 1. i.163) The result yields the reference state |Φo (U(M +1) ) using the definition (9. U(M +1) is the (M +1) S × N –matrix of the first N column vectors of Urm . . (9. step 2M + 2 is carried out.e.6.6: Self-Consistent Field Algorithm is evaluated.162) is solved adopting an ordering of the labels m which obeys the condition (M +1) m < (M +1) n for m < n .167) . 2. ˆ t. . S. ..9. . . t| v |u. S do not exceed a preset threshold. . 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. (9. M = 1. m = 1. 2. . SCF-Algorithm: Continuation and Convergence Condition When step 2M + 1 is completed.145).s=1 σ ˆ r|f |s c† csσ rσ (9. σ|ˆrs |u. etc. (e) Consider the creation operators N † gqσ = m=1 Vmq d† mσ (9. 2. m = 1. . 2. i. r = 1.u=1 ( 2 r. the S × S–matrix S (M ) r|hmf |s N 269 ˆ = r|t|s + t.166) ˆ (d) Define a similar two-particle operator Prstu . S . S (9. t| v |s.e.: 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 . Step 2M + 1. σ p = δtr δsu δσσ .145). S . .. .e.165) can be written ˆ F = S ˆ ˆ r|f |s Prs r. . SCF-Algorithm. . . .164) ˆ (a) Determine the expectation value of Prs for the state as defined in (9.s=1 (9.1: Let Prs be the one-particle operator with the generic operator prs = |r s|. . indicating that the state converged.161) is calculated. . (c) Show that any spin-independent one-particle operator S ˆ F = r. 2 . . u − r. 2. s = 1.

7 Properties of the SCF Ground State We want to investigate now the properties of the SCF ground state. We denote the representation which results from the SCF algorithm after its convergence as follows: S |m = ˜ r=1 (SCF Urm ) |r .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. S denotes the initial single-particle states and Urm is the unitary S × S– transformation matrix obtained in (9.171) where the convention (9.σ + 2v ( c† c1α c† c1β + c† c2α c† c2β ) + v 1α 2α 1β 2β + c† c2σ c† 2σ 1σ c1σ ) . Equation (9. r = 1. We begin by summarizing the result of the SCF algorithm. .170) (SCF ) Here |r .145) and (9. (9. (9. (SCF ) m < (SCF ) n for m < n (9.169) 9. .163). (9. For this representation holds N m| t |˜ + ˜ ˆn m 2 mm | v |˜ m ˜ ˜ ˆ n˜ − mm | v |m n ˜˜ ˆ ˜ ˜ = (SCF ) δmn m (9. i. Note that this is not an mσ S × S–matrix! Show that the reference state defined through N † † gqα gqβ |0 q=1 (9.173) .162.9.270 Many–Particle Systems which are connected with d† through a unitary N × N –matrix (Vqm )..145).168) ˆ ˆ has the same expectation values for Prs and Prstu as the reference state (9.171) implies that for the mean field Hamiltonian holds S ˆ Hmf (U(SCF ) ) = m=1 σ (SCF ) † dmσ dmσ m .168) through a physical observation? Exercise 9.172) is obeyed.e. 2.163). . (f) Can one distinguish the states (9.6.

175) However.7: Properties of the SCF Ground State 271 i.177).174) The first property we consider is the total spin character of SCF . one can also determine the expectation value of SCF for the Hamiltonian (9. once as a member of the ground state.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. r. 2 m=1 m|t|m + ˜ ˆ˜ m. s|ˆ|t. (SCF ) ˆ Exercise 9.142).84) and (9. the SCF ground state is N SCF = m=1 d† d† |0 . . and once as a probe particle being subject to the mean pair interaction. This over-counting leads to the correction term in (9.e.171) in terms of matrix elements r|t|s .173) holds N ˆ SCF Hmf (U(SCF ) ) SCF = 2 r=1 (SCF ) m . the mean field potential ˆ Vmf (|Φo (U) ) defined in (9.177) 2 mm | v |mm ˜˜ ˆ ˜˜ − mm | v |m m ˜˜ ˆ ˜ ˜ . 2 m=1 (SCF ) m − m.93) for the expectation values (diagonal elements) of the one-particle and two-particle part of the Hamiltonian (9.2: Derive (9.142).146) counts a particle twice.4 that SCF is a total singlet state. v (SCF ) Show that the resulting eigenvalue problem of the type (9. Exercise 9.9. Within the mean field approximation as described by (9.7.176) 2 mm | v |mm ˜˜ ˆ ˜˜ − mm | v |m m ˜˜ ˆ ˜ ˜ .7. one obtains ˆ SCF H SCF N = N (9.1: Reformulate (9. mα mβ (9.. exploiting the spinindependence of (9. Since the system under consideration has only 2N particles altogether. u and Urm .162) is non-linear in Urm .171) yields ˆ SCF H SCF N = N (9.177).142). Since this state is composed only of closed shells one can conclude following Section 9. the mean-field Hamiltonian is diagonal in the SCF representation. ˆ SCF H SCF .. Employing expressions (9. i.e. We like to determine next the energy expectation value of SCF .m =1 Comparision with (9. (9. Finally.

178). in the two-dimensional copper oxide lattices of high temperature superconductors. v of the Hubbard Hamiltonian (9.8 Mean Field Theory for Macroscopic Systems The SCF algorithm has been developed in Sections 9. cannot be solved exactly. The Hubbard Model The Hubbard model serves today the important role as the simplest manifestation of strongly correlated electron systems which arise in many instances in molecular and solid state physics. Due to the lack of an exact solution in dimensions higher than one.u σσ where r. † ˆ Ho = −t c† . labeled r = −N0 + 1. the second term on ther.4–9. is characterized through the parameters t. real number. reads 1 ˆ r.179) r+1.t. The case n = 1 is termed the half filled band case. in the usual two–particle operator form.e.e.s. In the present Section we want to adapt the algorithm to systems containing a macroscopically large number of fermions. referring to the fact that in this case the band of single particle energies (9. of (9. S N0 (9. r ∈ Z.178).272 Many–Particle Systems 9. i. t = v δru δst δrs .178) v contributes only in case that two electrons occupy the same lattice site. t is assumed to be a positive. (9. We assume that there are altogether S = 2N0 lattice sites.178) and through the so-called filling factor n n= 2N N = . β for ‘up’ and ‘down’ rσ spins. According to (9.σ crσ + crσ cr+1.s.180) V = rσ sσ 2 r.σ crσ + crσ cr+1. The operator c† (crσ ) creates (destroys) an electron with spin σ (σ = α. i. approximation schemes like the self-consistent field approximation play an important role. .128).7 for finite systems. s|ˆ|u.σ rσ The potential energy term. t describes the coupling of state |r to states |r ± 1 at the two neighboring lattice sites. v (9. v (9. . In spite of the simplicity of its Hamiltonian the Hubbard model. of (9. The one-dimensional Hubbard model is described by the Hamiltonian ˆ H = −t rσ † c† r+1. except in case of one-dimensional systems. Both N0 and N are macroscopically large numbers. s|ˆ|t.s. rσ rσ r σσ (9.h. for example.192) derived below is filled half. . The first term on the r. describes the sites of a linear lattice. v represents the ‘on-site’ Coulomb repulsion.181) The Hubbard model. as stated through Hamiltonian (9. We consider as an example the Hubbard model of an infinite linear lattice and apply the model to describe magnetic instabilities in metals..h.178) Here the index r. is identical to that of the independent–particle Hamiltonian H0 (9. . i.σ + v 2 c† c† crσ crσ . N0 which are populated by 2N electrons (we choose an even number of electrons for convenience). respectively) in state |r at lattice site r..e.182) n can assume values 0 ≤ n ≤ 2. . u c† c† cuσ ctσ ..178).

one obtains for the mean field Hamiltonian N −1 ˆ ˆ Hmf (U) = H0 + v r. the operator H0 can be re-written in the diagonal form N0 ˆ H0 = m=−N0 +1 σ m d† dmσ mσ (9. 9.187) The energy levels are labeled such that 0 < ±1 < . i. Accordingly. 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. 2N0 (9. we assume that the SCF ground state is given by (9. Using (9.179). 9.185) ˆ According to (9.8: Macroscopic Systems 273 Since we are dealing presently with a macroscopic system. Accordingly. We can now embark on step 2 of the SCF algorithm. we identify |r + N0 = |r and later let N0 go to infinity.186) where m = −2t cos mπ .144).184) and using (9. SCF Ground State for the Hubbard Model We seek to determine the ground state of the Hubbard model stated by (9.184) ˆ We are now ready to apply step 1 of the SCF algorithm and diagonalize term H0 . by N −1 ||SCF = m=0 d† d† |0 .183) where the operators d† are related to the original creation operators c† through the unitary mσ rσ ˆ transformation U defined in (9. . rσ (9. This task has ˆ been solved already on page 262.186) one obtains for the mean field Hamiltonian N0 ˆ (2) Hmf = m=−N0 +1 σ m d† dmσ + v mσ N 2N0 c† crσ . rσ rσ (9. boundary effects are assumed to be negligible. .σ m=0 ∗ Urm Urm c† crσ .155.154).e. in fact. < ±(N0 −1) < N0 holds as can be readily verified. The self-consistent field Hamiltonian Hmf which corresponds to the Hubbard Hamiltonian (9. mα mβ (9.174). Inserting (9.178) can be determined by applying (9.185) into (9.181..6.140).178) within the framework of the self-consistent field theory.188) . N0 (9. is Urm = √ 1 exp (irmπ/N0 ) . For this purpose we apply the SCF theory presented in Section 9. The unitary transformation which diagonalizes H0 .9.

The ground state of the system. i.190) We have used in the latter expression the definition (9. hence. the results of different formulations agree qualitatively. π]. ˆ The Hamiltonian Hmf .. . ..190). 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. (9.191) where m is given in (9. as given in (9. −π ≤k ≤π .192) This dispersion law is presented in Fig. dnσ according to (9. the electrons in the present description behave like independent particles with energies ¯m = m + vn . 9. according to the Pauli principle. The energy m can be expressed as a function of a continuous variable k ∈] − π. (9. can be obtained. called the Fermi energy. The bottom of the energy band corresponds to (0) = −2t and the width of the energy band is 4t. 9. Apparently. .178).182) of the filling factor n [c.187) form a quasi1 2 continuous energy band. The energy Emf of this state can be calculated readily by using (9. but may differ quantitatively. through the so-called dispersion law (k) = −2t cos k . i. Alternative Description of the Mean Field Approximation The state (9.187). π]. Indeed. . In other words.f. −π + N0 . the self–consistency condition is exactly met and the SCF algorithm converged after two steps.189) and. holds k ≡ km ∈] − π.177) and mm |ˆ|mm = mm |ˆ|m m = ˜˜ v ˜˜ ˜˜ v ˜ ˜ which holds in the present case. even for one and the same Hamiltonian.274 Many–Particle Systems Expressing c† .1. denoted by F (see Fig.109) and using (9.183) is not the only candidate for the mean field ground state of the Hubbard model. there exist several ways of formulating mean field approximations.185) one can prove rσ mσ N0 c† crσ = rσ rσ m=−N0 +1 σ d† dmσ . namely. mσ (9. in the limit N0 → ∞.182)]. When N0 becomes macroscopically large the single particle discrete energy levels (9.e. One can see that −2t ≤ (k) ≤ 2t holds.e. To demonstrate this we consider an alternative formulation of the mean field approximation for Hamiltonian (9. for km = mπ/N0 holds km ∈ −π + N0 . the ground state of the Hubbard model in the self-consistent field approximation can be determined exactly. N0 ˆ (2) ˆ Hmf = Hmf = m=−N0 +1 σ m + vn † dmσ dmσ 2 (9. π and. the state with the lowest possible energy. 4 So far we have not exploited the fact that N0 and N are macroscopically large quantities. crσ through d† . Usually. by filling up this energy band with electrons from the bottom of the band (which corresponds to k = 0) to a maximum energy. Indeed.1).

rσ rσ rσ (9.56).194) ˆ For a given reference state. in the last term on the right hand side.183).45. The mean field approximation neglects the . the operator Nrσ can be written ˆ Nrσ = Nrσ + δ (Nrσ ) . The anti-commutation properties (9. First.195) ˆ ˆ ˆ and where δ (Nrσ ) = Nrσ − Nrσ describes the fluctuations of Nrσ .. we have employed (9.178) in terms of the occupation number operaˆ tors Nrσ = c† crσ . F = (±kF ) denotes the Fermi energy.193) into (9. (9.e.178) and performing the summation over the spin indices results in ˆ ˆ H = H0 + v r ˆ ˆ Nrα Nrβ . i. 9.193) where. (9.8: Macroscopic Systems 275 Figure 9. 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 . the deviation of Nrσ from its mean value Nrσ for the corresponding reference state. Inserting (9. Nrσ = SCF ||Nrσ ||SCF . say the one given in (9.9.. let us express the interaction term in (9. where Nrσ is the mean occupation number of the one particle state |rσ .196) (9.1: The dispersion law (k) for independent electrons on an infinite one dimensional lattice.e.

..199) which is identical to expression (9. i.198) as the one obtained by using the mean field potential (9. non-vanishing magnetization.276 Many–Particle Systems fluctuations to second order and the mean field Hamiltonian is obtained from (9.183) . 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. 2N0 2 holds ˆ Hmf = = mσ mv ˆ H0 + 2 vn2 2 rσ nv † vn dmσ dmσ − N . (9. Because of Nrσ = c† crσ rσ N −1 = m.198) Let us assume that the ground state of the Hubbard model. separating the occupation number operator into a mean value plus fluctuation and neglecting the fluctuations in 2nd order. Therefore.194) by dropping all the terms which contain the fluctuations to second order. yields essentially the same mean field Hamiltonian (9.m ∗ Urm Urm d† dm σ mσ = m=0 ∗ Urm Urm = n N = . Note that the procedure employed in (9. In case nα = nβ the following construction will lead to a ground state which coincides with the non-magnetic ground state (9.197) and the corresponding mean field Hamiltonian becomes ˆ ˆ Hmf = H0 + v r ˆ ˆ Nrα Nrβ + Nrβ Nrα − v r Nrα Nrβ .195).146).e. the ground state of the system can have a non-zero local spin and. m+ 2 2 c† crσ − 2N0 rσ (9.191) for the ground state energy.183). Spin-Polarized Mean Field Ground State ˆ We want to consider now a ground state for Hmf defined through Nrσ = nσ = const (9.200) allowing. but the mean number of particles with spin α can be different from the mean number of particles with spin β. consequently. however. that nα and nβ assume different values. in the mean field approximation. 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 . is given by (9. For such state the mean occupation number nrσ is uniform at all lattice sites.

208) .206). The values of nα and nβ are determined by minimizing the energy density (9. hence. (9.202) This Hamiltonian describes a system of non-interacting electrons with a spin-dependent dispersion law (9.205) is met. In fact. Since β (k) − α (k) = v(nα − nβ ). a spin–polarized ground state. the smaller is σ (k) for a given k. (9.204) Emf ≡ 2N0 2N0 m. if nα > nβ then β (k) > α (k).9.206) ˜mf is minimized where µ is the Lagrangian multiplier considered an independent variable. In the limit N0 → ∞ this sum can be expressed as an integral. (9. such that the larger nσ . In this limit the discrete energy spectrum (9. E with respect to nα .192). subject to the constraint n = nα + nβ . β) are determined by minimizing the energy density of the ground state Hmf 1 = [ mσ Nmσ ] − vnα nβ (9..206) needs to be evaluated.. e.206) ascertains that condition (9. m (9. The ground state (corresponding to the lowest possible energy) of the many electron system is obtained by filling these two energy sub–bands with electrons up to the same maximum energy value. the so-called Fermi energy F = µ (see below).2a.g. The actual values of nσ (σ = α. At this point we exploit the fact that our system is macroscopically large.. 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 ).σ with respect to nα and nβ . as is shown schematically in Figure 9.8: Macroscopic Systems from which we obtain nσ = The mean field Hamiltonian is then ˆ Hmf = m 277 1 2N0 Nmσ .201) ( m ˆ + vnβ ) Nmα + ( m ˆ + vnα ) Nmβ − 2N0 vnα nβ . Therefore.207) where (k) is given by (9. N0 ∼ 1023 . i. (9.e.e. The additional term in (9. i. 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.203) mσ = m + vn−σ . nβ and µ.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) .205) Such minimization is realized through the method of Lagrangian multipliers. 2π (9. According to this method one minimizes ˜ Emf = Emf + µ (n − nα − nβ ) . one can expect the system to lower its energy by assuming nα = nβ and. For this purpose the sum in (9.

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

220) µ can be readily obtained µ= 1 ( 2 α + β) + vn . (9. From (9. namely. (9.213) The presence of the θ function in the above formula guaranties that the density of states vanishes for | | > 2t.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.8: Macroscopic Systems 279 where θ(x) is the step function. namely. outside the energy band (k). in principle. Replacing again the sum by an integral over energy one can write σ nσ = −2t ρ( )d (9.f. β . F = α + vnβ = β + vnα . (9.206)] depends only on continuous quantities. the necessary ˜ Since Emf [c.222) . (9. + vnα − µ = 0 .218) The derivatives can be readily determined and the conditions (9. to determine the unknown quantities α .2].221) ∂µ Equations (9.9. of t.204) through an integral expression. Fig. i. ∂ β ˜ ∂ Emf =0.219..213) can simply be replaced by 1..e. one arrives at the following expression of the density of states ρ( ) = θ(2t − | |) π (2t)2 − 2 .219) (9.216) Using (9.217) and µ.215) nα and nβ in (9. (9.216) allow one.f. µ.e. ∂µ (9. 2t[ then the θ function in (9. if we assume that is restricted to the interval ] − 2t.214) are given by the sum (9. ∂ α ˜ ∂ Emf =0. π α. i. 2 (9.221) and (9. i.220) β ˜ ∂ Emf = n − nα − nβ = 0 . β (9.201). v and n.213) and carrying out the resulting integral yields nσ = 1 arcsin( σ /2t) . We can employ the results obtained to express the ground state energy density (9. θ(x) = 1 if x > 0 and θ(x) = 0 if x < 0. nα and nβ as a function of the parameters of the Hubbard model. 9.e. (9. One obtains ˜ Emf = α β ρ( )d + −2t −2t ρ( )d + vnα nβ + µ(n − nα − nβ ) .. on conditions for a ground state of minimum energy are ˜ ∂ Emf =0.219–9. 9.218) read ˜ ∂ Emf = ∂ α ˜ ∂ Emf = ∂ β and α + vnβ − µ = 0 .

(9. We assume in the following discussion that the implicit ∆ dependence of β in f (∆) can be neglected. From the definition (9. i.225) Defining the function v∆ f (∆) = 0 ρ( β + )d . (9. i.223 one obtains ∆= and second.f.227) can be solved graphically. (9.226) follows that f (∆) is a monotonically increasing function of ∆. (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.f. if. The Stoner criterion (9. condition (9. One still needs to determine α and β . condition (9.227). (9. from (9. The slope of f (∆) at the origin depends on the Hubbard model parameter v.215) reveals that µ is indeed equal to the Fermi energy F (see also Figure 9.229) has a wider range of validity than one can infer from the present analysis of the Hubbard model. For v above a critical value vc . Figure 9. the Hubbard model has a ground state with ∆ = 0. If the slope of f (∆) at the origin is less than 1. For this purpose one plots f (∆) versus ∆ as is shown schematically in Figure 9.226) ∆ = f (∆) . (9. from (9.222) and (9.223) follows α α − v β . For this purpose we consider the magnetization of the mean field ground state ∆ ≡ nα − nβ .227) has only the trivial solution ∆ = 0.224) v∆ ∆= β ρ( )d = 0 ρ( β + )d . as reflected by the expression d[f (∆)] = vρ( d∆ β + ∆) = π (2t)2 v − (v∆ + 2 β) >0. through the condition [c.e. (9. non trivial.217) and (9.280 Many–Particle Systems Comparision of (9. the slope of f (∆) is positive for all ∆ ≥ 0. .3]. Accordingly.228)] vc ρ ( β ) = 1 . One can determine vc by equating the slope of f (∆) at the origin to 1.e. One can express ∆ as a function of α and β in two different ways. (9. the slope of f (∆) at the origin is larger than 1 [c. so called ferromagnetic solution.3. Equation (9.223) ∆ = 0 corresponds to the ground state (9.226) one can obtain ∆ by solving (cf.227) will have a second. f (∆) converges to a maximum value as ∆ → ∆max = n.220) and (9. First.219–9.227) Since f (0) = 0 one can infer that ∆ = 0 is always a solution (the so called paramagnetic solution) of condition (9.228) Since the total number of states per site is finite.. (9.2b). a magnetic ground state.. and only if. Indeed.183) without magnetization.

(9. Since ∆ vanishes near vc we set x = y + δ where δ is a small quantity. n.233) yield the desired expression vc π = π cos (n − 1) .. 9. is presented in Figure 9.231) follows y= δ π π (n − 1) − ≈ (n − 1) . the plot of vc vs.232) states that x − y is proportional to ∆. 2 2 2 (9.205) the following set of equations determining x. the magnetization ∆ of the ground state vanishes and remains zero for values v < vc . Accordingly. We like to determine finally the dependence of vc on the filling factor n. β = 2t sin y .232) together with (9.219. i. (9. the ground state is characterized by the quantities (v.4.e.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. By parameterizing α and β α = 2t sin x .8: Macroscopic Systems 281 Figure 9.233) Employing the approximation sin(y + δ) − sin y ≈ δ cos y for small δ. 2t As already mentioned. any .231) (9. 9.234) 2t 2 The resulting phase diagram of the ground state. The corresponding 2t analytical expression provides the phase diagram of the ground state: when v approaches vc from above. β x+y x−y = = π(n − 1) 2t π (sin x − sin y) = π∆ .220. (9. t. n).9.230) one obtains from (9.3: Graphical solution of the mean field equation ∆ = f (∆). hence. From (9. v (9. α. y and. Equation (9.

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

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

284 Many–Particle Systems .

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

t) as components. x2 . i. we will provide a more basic definition of the transformations. for the transformed space-time–cordinates x µ = (x 0 . x 3 ) holds (x0 )2 − (x1 )2 − (x2 )2 − (x3 )2 = (x )2 − (x )2 − (x )2 − (x )2 .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..3) where x0 = ct describes the time coordinate and (x1 . This space is called the Minkowski space. We will denote these vectors as follows def xµ = (x0 .4 to cover fields. Before introduco ing these general Lorentz–invariant field equations we will provide in Sects. x 2 . Rather than starting from (10. x1 .1). Note that the components of xµ all have the same dimension. We will find that this definition will lead us back to the transformation law (10. The elements L ∈ L act on 4–dimensional vectors of position– and time–coordinates. wave functions ψ(r.e. namely the Klein–Gordon (Sect. x3 ) = r describes the space coordinates. henceforth. 10.4) invariant. however. 10. This will provide us with a general set of Lorentz–invariant equations which for various particles take the place of the Schr¨dinger equation. t) and vectors with functions ψ(r.5) √ One can interprete the quantity −s2 as a distance in a 4–dimensional Euclidean space if one chooses the time component purely imaginary. 10. 10. assume new units for time such that the velocity of light c becomes c = 1. x 1 . namely that of length. Rather than following this avenue we will introduce Lorentz transformations within a setting which does not require real and imaginary coordinates.1 Natural Representation of the Lorentz Group In this Section we consider the natural representation of the Lorentz group L. Denoting by xµ the . In such a space Lorentz transformations correspond to 4-dimensional rotations. the group of Lorentz transformations (10. 5 to the groups SO(3) and SU(2) of rotation transformations of space coordinates and of spin. 0 1 2 3 (10.. 10. 10.1 will be extended in Sect. x3 ) (10.286 Relativistic Quantum Mechanics in Sect. x2 .1). i. i. 10.e.1). the Lorentz transformations were formulated in a space in which the time component of xµ was chosen as a purely imaginary number and the space components real.5) and the Dirac (Sect.e.7 a heuristic derivation of the two most widely used and best known Lorentz–invariant field equations.1) leave the quantity s2 = (x0 )2 − (x1 )2 − (x2 )2 − (x3 )2 (10. The reason for this choice is that the transformations (10. We will. This choice implies dim(time) = dim(length). Minkowski Space Historically. but in a setting of representation theory methods as applied in Secti.5.7) equation. The Group of Lorentz Transformations L = O(3.

10. Aµ ν . the . xµ .6) Here Lµ ν are the elements of a 4 × 4–matrix representing the Lorentz transformation.. the latter set constituting a group.9)  1 0 0  0 −1 0 0   ( gµν ) =   0 0 −1 0  = g . the notation in (10. Aµν . Aµ ν xν = µ=0 new old Aµ ν xν .6) allows us to introduce the so-called summation conventon: any time the same index appears in an upper and a lower position. (10. The upper index closer to ‘L’ denotes the first index of the matrix and the lower index ν further away from ‘L’ denotes the second index. 1. 0 0 0 −1 (10.6) x µ = Lµ ν xν . R). Since this holds for any xµ it must be true Lµ ρ gµν Lν σ = gρσ . the latter notation will be used for different quantities further below. L ∈ GL(4. [ A more conventional notation would be Lµν . Aµ ν B ν ρ = ν=0 Aµ ν B ν ρ . This condition can be written xµ gµν xν = x gµν x where  µ ν (10. the Lorentz transformations constitute a linear transformation which we denote by 3 x µ = ν=0 Lµ ν xν . ν=0 (10. R). This will be explained further below. 4 × 4 tensor: Aµ ν . 2..8) new old old The summation convention allows us to write (10. The Lorentz transformations form the subgroup of all matrices which leave the expression (10.6) 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. (10. 3 3 3 yµ xµ = new yµ xµ . The subset of GL(4. ν = 0.] The following possibilities exist for the positioning of the indices µ.9) and (10. We will exploit this property below to determine the general form of the Lorentz transformations. summation over that index is assumed without explicitly noting it.e.11) This condition specifies the key property of Lorentz transformations. 3: 4-vector: xµ .e. however.10) Combining condition (10. i.12) (10. First. i.5) into scalar products. Aµν . The second reason is that upper and lower positions allow us to accomodate the expression (10. The Lorentz transformations are non-singular 4 × 4–matrices with real coefficients.7) The reason for the notation is two-fold. (10.5) invariant.

From this one can obtain using g2 = 1 1 (gLT g)L = 1 and. i.1). 10.288 Relativistic Quantum Mechanics elements of which satisfy this condition.18) (10. employing expressions (10.19) i. One can also show that if L ∈ O(3.1).15) (10. of (10.. LT gL = g } . We want to show now L = O(3.10) of g one can rewrite the invariance property (10.16) The corresponding expression for (LT )−1 is obviously (LT )−1 = (L−1 )T = g L g .1) since it satisfies (10.17) we note first that the identity matrix 1 is an element of O(3. (10. i.. To demonstrate the group property of O(3. L2 ∈ O(3.17) one obtains (L−1 )T g L−1 = gLgggLT g = gLgLT g .h. .e. L−1 ∈ O(3. In fact. For L3 = L1 L2 holds LT g L3 = LT LT g L1 L2 = LT (LT gL1 ) L2 = LT g L2 = g . is called O(3.14) holds for the inverse of L. 1) = { L.20) results in the desired property (L−1 )T g L−1 = g .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. = g LT g =  0 1 2  −L 2 L 2 L 2 L3 2  −L0 3 L1 3 L2 3 L3 3 (10.14)  L0 0 −L1 0 −L2 0 −L3 0  −L0 1 L1 1 L2 1 L3 1   .e. of O(3.e.1).1). property (10. To simplify the following proof of the key group properties we like to adopt the conventional matrix notation for Lµ ν   0 L 0 L0 1 L0 2 L0 3  L1 L1 1 L1 2 L1 3   (10.14) from the right by gLT and using (10.14). This set is identical with the set of all Lorentz transformations L. in fact.20) Multiplying (10.1).1) ⊂ GL(4. L3 ∈ O(3.1) is.. a group. hence. L ∈ GL(4.1).12) LT gL = g .16.. R). L3 0 L3 1 L3 2 L3 3 Using the definition (10. 3 2 1 2 1 2 (10. (10. This stipulates that O(3. We consider 1 then L1 . the inverse of L 1  L−1 (10.22) (10.13) L = ( Lµ ν ) =  2 0  L 0 L2 1 L2 2 L3 3  . R) is a group.s.21) i. the associated inverse obeys (10. (10.14).e.

+ + ↑ ↑ We can also demonstrate that for A. 1). Considering in (10.e. = { L.24) or since (L1 0 )2 + (L2 0 )2 + (L3 0 )2 ≥ 0 it holds (L0 0 )2 ≥ 1.31) (10. (10. det L = 1. 1). A second class property follows from (10.14). 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 .12). hence.10. (10.34) j=1 j=1 .28) (10.14) which we employ in the formulation (10.14). L ∈ O(3. L↓ are disjunct sets defined as follows + + − − L↑ + L↓ + L↑ − L↓ − = { L. One can also 1 verify that one can write L↑ − L↓ + L↓ − = gL↑ + = = −L↑ . (10.32) (10. L0 0 ≥ 1} . B ∈ L+ holds C = AB ∈ L+ . = { L. Properties (10. = { L. ∃M2 . + = − L↑ g + (10. 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↑ . Lorentz transformations in L↑ . M1 = a M2 }.1: Natural Representation of the Lorentz Group Classification of Lorentz Transformations 289 We like to classify now the elements of L = O(3. Using det AB = det A det B and det AT = det A yields (det L)2 = 1 or det L = ±1 . det L = −1. L ∈ O(3. A statement on this value can be made on account of property (10.25) i. L↓ .. det L = −1. entirely suitable to investigate first only 1. It holds g ∈ L and −1 ∈ L as one can readily verify testing for property (10. σ = 0 yields L0 0 2 − L1 0 2 − L2 0 2 − L3 0 2 = 1.12) the case ρ = 0.29) (10. there exist two other distinct classes. Schwartz’s inequality yields j=1  2 3 3 3  j=1 A jB 0  ≤ 0 j A 0 2 j Bj 0 2 . 1). det L = 1. L0 0 ≥ 1} . L ∈ O(3. The above shows that the set of proper Lorentz transformations L↑ allows one to generate all Lorentz transformations.1). (10.23) and (10. 1).33) where we used the definition aM = {M1 .23) One can classify Lorentz transformations according to the value of the determinant into two distinct classes. M2 ∈ M. L0 0 (10. L0 0 ≤ −1} .27) (10. + − gL↑ + = L↑ g . Obviously.26) (10. L ∈ O(3. L↑ contains 1 1. For this purpose we consider first the value of det L. + except for the trivial factors g and −1 It is.30) ≤ −1} . + We start our investigation by demonstrating that L↑ forms a group. From this we can conclude L0 0 ≥ 1 or L0 0 ≤ −1 .

. δ00 = 1 and It should be noted that according to our present definition holds δµν = gµρ δ δ11 = δ22 = δ33 = −1.1) ascertains CT gC = g it must hold C 0 0 ≥ 1. hence.37) (10. 1 0 0 = ≥ 1 and.40) and.12) follows (B 0 0 )2 − 3 (B j 0 )2 = 1 or 3 (B j 0 )2 = (B 0 0 )2 − 1. i.1). 1 ∈ L↑ . In the following we + will consider solely this subgroup of SO(3.14) actually L0 0 ≥ 1 and det L = 1 must hold. + Accordingly. (10.34) provides then the estimate conclude from (10. accordingly. since the group property of O(3. | 3 A0 j B j 0 | < A0 0 B 0 0 . Similarly.1).290 Relativistic Quantum Mechanics From (10. from which one can conclude L−1 ∈ L↑ . orthochronous Lorentz transformations. For the inverse of + ↑ any L ∈ L+ holds (10. .16).14) implies 1 + T g (1 + ) = g 1 1 (10.21) 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 .36) It holds. (10. obviously. obviously j=1 A0 0 B 0 0 ≥ 1. one can j=1 j=1 3 0 )2 = (A0 )2 − 1. L↑ is called the subgroup of proper. holds L0 0 > 0 and the value of the determinant is close to unity.35) One can conclude. we consider transformations Lµ ν = δ µ ν + µ ν . 1 We will then employ the Lie group properties to generate all transformations in L↑ . 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. Since the associative property holds for matrix multipli1 1 + ↑ cation we have verified that L+ is indeed a subgroup of SO(3. (10. In fact.. therefore. Property (10.38) For these transformations. ρ ν (10. The next group property of L↑ to be demonstrated is the existence of the inverse.e. µ ν small .39) where we have employed the matrix form T 1 defined as in (10. This relationship shows (L−1 )0 0 = L0 0 . Infinitesimal Lorentz transformations The transformations in L↑ have the property that they are continously connected to the identity 1 1. Since A0 0 ≥ 1 and B 0 0 ≥ 1. if we enforce (10. Using the above expression for C 0 0 one can state C 0 0 > 0.e. 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. + transformations in a small neighborhood of 1 which we parametrize by infinitesimal parameters. + i.13). To order O( 2 ) holds g + g = 0.

w3 ) =   −w2 ϑ3 0 −ϑ1  −w3 −ϑ2 ϑ1 0 (10. k j j = 1.50) .  (10. K ] [ Jk .1: Natural Representation of the Lorentz Group Using (10. J2 =    0 0 0 −1  0 0 0 0   0 0 0 0 0  0 0 −1 0 1  .43) k = − Inspection shows that the matrix has 6 independent elements and can be written   0 −w1 −w2 −w3  −w1 0 −ϑ3 ϑ2   . ϑ3 . 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. 2. J =  0  3  0 1 0 0 0 0 0 0 0 0 0 0 0     (10. other five parameters zero) . 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 .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    .46)  0 0   0 0  . k = 1. w1 . J ] [ Kk .42) This relationship implies µ 0 j µ j = 0 = j 0 .49) . 2. j.15) one can conclude T 291 = −g g (10. (ϑ1 . 3 (10. other five parameters zero) Kk = (wk = 1. 3 .45) (10. = − k m Km (10. K2 =  .44) This result allows us now to define six generators for the Lorentz transformations(k = 1.48) (10. 3) Jk = (ϑk = 1.47)     0 −1 0 0 0 0 −1 0 0  −1   0 0   0 0 0 0  0 0  K1 =  . 2. w2 .10. ϑ2 . The generators are explicitly  0 0 0 0  0 0 0 0 J1 =   0 0 0 −1 0 0 1 0  (10.

57) . In analogy to equation (5. 3 (10.55) L 0 0 0 0  exp (w1 K1 ) = exp   Using the idempotence property 0 −1 −1 0 2 0 0 1 0 (10. k = 1. however. 2.51) One can readily show. 2.54) which. Kk . 3 is closed. L(ϑ. To determine the explicit form of this transformation we evaluate the exponential operator by Taylor expansion. + 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’. k = 1. 2π[ . Following the treatment of the rotation group SO(3) one can express the elements of L↑ through + the exponential operators L(ϑ. w ∈ R3 3 k=1 wk Kk .. constitutes an overcomplete parametrization of the rotations (see Chapter 5). Relativistic Quantum Mechanics The commutation relationships (10.e. and using the relationship Jk = 0 0 0 Lk (10. 10. (10. 0  1 n (10. For the parameters ϑk of the Lorentz transformations holds obviously ϑk ∈ [0.51) for w = 0 correspond to rotations of the spatial coordinates.292 as can be readily verified.56) = 1 0 0 1 = 1 1 (10. A boost in the x1 –direction is L = exp(w1 K1 ). w = 0) = 0 0 0 R(ϑ) .1: Demonstrate the commutation relationships (10. 3 where we have defined ϑ · J = k=1 ϑk Jk and w · K = following the algebra in Chapter 5. (10. that the transformations (10. Exercise 7. i.50).53) Here R(ϑ) are the 3 × 3–rotation matrices constructed in Chapter 5. ϑ.49) define the Lie algebra associated with the Lie group L↑ .52) where the 3 × 3–matrices Lk are the generators of SO(3) defined in Chapter 5. w) = exp ϑ · J + w · K .49.35) 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   .

x3 = x3 (10. From (10. (10. (10.10.59) v1 = These two equations yield cosh w1 = 1/ and (10. vk defined in (10. Accordingly.62) According to (10. cosh2 w1 − 1 and coshw1 = cosh2 w1 − 1 = cosh w1 sinh w1 sinh2 w1 + 1 .59).6. consistent with (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.64) We note that the range of wk -values is not a compact set even though the range of vk -values is compact.3. in fact.61) √ √−v1 √   exp (w1 K1 ) =    2 1 − v1 1 2 1 − v1 0 0 0 0 0 0  0 0    1 0  0 1  (10.58) = cosh w1 1 + sinh w1 1 0 −1 −1 0 The conventional form (10. 10. Correspondingly.63) which agrees with (10. This property is. we can state vk ∈ ] − 1. however. for wk wk ∈ ] − ∞.60) sinh w1 = v1 / 2 1 − v1 .1). t = t − v1 x1 2 1 − v1 . (10.1: 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 . The range of the parameters wk can now be specified.59) can be written  1 2 1 − v1 √−v1 2 1 − v1 2 1 − v1 . 1[.59) cosh w1 Using cosh2 w1 − sinh2 w1 = 1 one can identify sinhw1 = sinh2 w1 + 1. We expect that vk can only assume values less than the velocity of light c which in the present units is c = 1.1) of the Lorentz transformations is obtained through the parameter change sinh w1 v1 = = tanh w1 (10. ∞[ .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. 10. (10.56. x2 = x2 . one obtains from (10. 10.59) follows.

294 Relativistic Quantum Mechanics 10. Another example is the potential 4-vector Aµ = (V. f = f. A) (10. tensor operators Tkm . A third 4-vector is the so-called current vector J µ = (ρ. the square of the electric field |E(r.. p) . we will encounter an impressive example of physical properties. r2 ) and. Presently. t) are the charge density and the current density. namely. 2 |ˆ1 . not any physical property f ∈ R is a scalar.67) the transformation behaviour of which we will demonstrate further below. The 4-vector character of J µ and of Aµ will be demonstrated further below. Some of these r ˆ quantities were actually defined with respect to representations of the rotation group in function spaces. our definition of quantities with special properties under Lorentz transformations presently is confined to the natural representation. of a system of charges. scalars like r = x2 + x2 + x2 . another example is the rest mass m of a particle. However. finally. E = √ m mv . J) (10.3) are the so-called 4–vectors aµ . We have encountered examples in connection with rotational transformations.65) An example is s2 defined in (10. x2 . the charge density. . total angular momentum r 1 2 3 states of composite systems like Y m ( 1 . the z–component x3 of a particle. p = √ 2 1−v 1 − v2 (10. vectors like r = (x1 . 4-Vectors The quantities with the transformation behaviour like that of the position–time vector xµ defined in (10. Counterexamples are the energy. These quantites always come as four components (a0 . a1 . a2 . respectively.4). a3 )T and transform according to a µ = Lµ ν aν . i. t)|2 or the scalar product r1 · r2 of two particle positions. Hence. t) and A(r. (10. x3 )T . not in the so-called natural representation associated with the 3–dimensional Euclidean space E3 . t) and J(r.68) where ρ(r.2 Scalars.69) where V (r. spherical harmonics Y m (ˆ). 4–Vectors and Tensors In this Section we define quantities according to their behaviour under Lorentz transformations. Nevertheless. Scalars The quantities with the simplest transformation behaviour are so-called scalars f ∈ R which are invariant under transformations.e.66) Examples of 4-vectors beside xµ are the momentum 4-vector pµ = (E. (10. t) are the electrical and the vector potential of an electromagnetic field. Such quantities appear in the description of physical systems and statements about transformation properties are often extremely helpful and usually provide important physical insight. we have not yet defined representations of Lorentz transformations beyond the ‘natural’ representation acting in the 4–dimensional space of position– and time–coordinates. We will see below how true scalars under Lorentz transformations can be constructed.

17) Lµ σ is the transformation inverse to Lσ µ .74) g µν aν .72) where aν is a vector with transformation behaviour as stated in (10.70) is a scalar. aµ (10... −a3 ) (10.72) of the covariant 4-vector follows = In µν and g µ as well. and .10. Multiplication (and summation) of x µ = Lµ ν xν by Lρ σ gρµ yields. lower.e. 4–Vectors and Tensors Scalar Product then 295 4-vectors allow one to construct scalar quantities. We like to point out that from definition (10. We do not fact. −a1 . i.71) Contravariant and Covariant 4-Vectors It is convenient to define a second class of 4-vectors. the relationship being aµ = gµν aν = (a0 . i. ∂x µ (10.2: Scalars. xν = g νσ Lρ σ gρµ x µ . accordingly. In fact.76) An important example of a covariant 4-vector is the differential operator ∂µ = ∂ = ∂xµ ∂ . (10. The respective vectors aµ are associated with the 4-vectors aµ .79) This is the inverse Lorentz transformation consistent with (10.66).12). This property follows from (10.66) together with (10. (10. In any case one can express (10. Covariant 4-vectors transform like a µ = gµν Lν ρ g ρσ aσ where we defined g µν = gµν . We have duplicated the expression for the inverse of Lµ ν to obtain the correct notation in terms of covariant. one can express [(L−1 )T ]µ ν = (L−1 )ν µ and. One calls 4-vectors aµ covariant and 4-vectors aµ contravariant. −a2 . gσν xν = Lρ σ gρµ x µ and g µσ gσν = δ µ ν .73) aµ = Lµ σ aσ .e. We start from x µ = Lµ ν xν .17) can be written (L−1 )ν µ = Lµ ν . using (10.75) (10. one can employ the tensors g µν to raise and lower indices of L ν establish here the consistency of the ensuing notation.73) Note that according to (10.16).78) To prove the 4-vector property of ∂µ we will show that g µν ∂ν transforms like a contravariant 4vector. ∂t (10.12) a gµν b µ ν = Lµ ρ gµν Lν σ aρ bσ = aρ gρσ bσ (10. (10.77) The transformed differential operator will be denoted by ∂µ = def ∂ . The 4-Vector ∂µ (10. g µν ∂ν = Lµ ρ g ρσ ∂σ . If aµ and bµ are 4-vectors aµ gµν bν (10.

p1 = √ m v1 m dx1 = √ . d’Alembert Operator We want to construct now a scalar differential operator.83) is a scalar under Lorentz transformations. dτ 1 − v 2 dt p0 = E = √ m m dt = √ . i.g.− ∂t .84) = 1 − (v)2 (10. 2 dt dτ 1 − v (10.67) transforms like (10.. Proof that pµ is a 4-vector We will demonstrate now that the momentum 4-vector pµ defined in (10.86) One can write (10.66).82) (10. ∂µ does indeed transform like a covariant vector.85) 1 d d = √ . i. indices. (10.. (10. upper. 1 − v2 1 − v 2 dt (10.296 Relativistic Quantum Mechanics contravariant. 2 2 dt 1 − v 1 − v (10. this operator is equal to the d’Alembert operator which is known to be Lorentz-invariant. 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 (10.89) .79) allows us to determine the connection between ∂µ and ∂µ .e..81) ∂ .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 . 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. e. In fact.87) The remaining components of pµ can be written.88) One can express then the momentum vector pµ = √ d m dxµ = m xµ .e. For this purpose we define first the contravariant differential operator ∂ µ = g µν ∂ν = Then the operator 2 ∂µ ∂ µ = ∂t − 2 (10.

96) The Lorentz-invariance of this gauge.96) is a scalar and. t) = 0 . t).97) and. (10. in (10. · A = 0.3 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.s. Since xµ transforms like a contravariant 4-vector.e.s. If we can show that Aµ defined in (10.h. in fact.h.3: Relativistic Electrodynamics 297 d The operator m dτ transforms like a scalar. and. We like to rewrite the last result E 2 = p 2 + m2 or E = ± p 2 + m2 . 8. kinetic energy ± 2m and relativistic corrections.91) (10.95) 2 p This obviously describes the energy of a free particle with rest energy ±m.94) 1 m In the non-relativistic limit the rest energy m is the dominant contribution to E. t) and A(r. is a scalar.77) which allow one to express (10. This gauge is not Lorentz-invariant and we will adopt here another gauge.s.69) is.s.96) in the form ∂µ Aµ = 0 . namely pµ pµ = pµ gµν pν = E 2 − p 2 Evaluation of the r.67) E2 − p 2 = or pµ pµ = m2 which. t) + · A(r. in fact.90) (10. can be demonstrated readily using the 4-vector notation (10. Lorentz Gauge In our previous description of the electrodynamic field we had introduced the scalar and vector potential V (r. the r.h.92) (10. m2 m2 v 2 − = m2 1 − v2 1 − v2 (10. for these potentials.93) (10. . the so-called Lorentz gauge. (10. in (10. (10. pµ on the l. of (10.69) for the electrodynamic potential and the 4-vector derivative (10. One obtains E = ±m ± p2 2m (p 2 )2 + O 4m3 (p 2 )3 4m5 .89) must be a 4-vector. Expansion in should then be rapidly convergent. respectively. a contravariant 4-vector then the l. namely.12). hence.89) alltogether transforms like a contravariant 4-vector. yields according to (10. of (10. The momentum 4-vector allows us to construct a scalar quantity.97) We have proven already that ∂µ is a contravariant 4-vector. and had chosen the so-called Coulomb gauge (8. We will demonstrate now the 4-vector property of Aµ . equivalently. Lorentz-invariant. 10. hence.h.. ∂t V (r.10. i.

68). that F µν can be expressed through the potential Aµ in (10.98) which reflects the principle of charge conservation.96).69) and ∂ µ in (10. t) = 4 π J(r. Since ∂µ transforms like a covariant 4-vector. using (10. using (8. t) − 2 A(r. 3 (10.98) can be written. The l. t) + · J(r. in fact. has to transform like a contravariant 4-vector.13).s.17) using the identity (8. F mn = − mn (10. · A(r. t) = ∂t · A(r. Using · ∂t A(r. must transform likewise.9). t) = 0 (10.h. m.102) In this equation the r.18) and. i.69). 10. (10. (8. transforms like a 4-vector.96). a 4-vector. F µν =  x  Ey B z 0 −Bx  Ez −By Bx 0 Alternatively. ∂µ J µ (xµ ) = 0 . according to (10. 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   .s.e. it follows that J µ . also the l. t) from (8. in fact. k.101). t) = −∂t V (r. t) . t) = −∂t V (r. The respective equation for A0 = V can be obtained from Eq.e. t) . · A(r.. one obtains for A(r.. of (10. one obtains 2 ∂t V (r. t) = 4πρ(r.100) Similarly.32).101) Combining equations (10.77) and (10. using (10. One can readily verify.97) and. t) are known to obey the continuity property ∂t ρ(r. i. (10. n = 1.6) and (8. (10.103) B . must be a scalar. t) − 2 V (r. t) and current density J(r. thereby. 2. Equation (10.s.104) where mn = mn is the totally anti-symmetric three-dimensional tensor defined in (5. t) together with (10. this can be stated F k0 = −F 0k = E k .105) .99) Since this equation must be true in any frame of reference the right hand side must vanish in all frames. (10.100.82) as follows F µν = ∂ µ Aν − ∂ ν Aµ .h. t).83) and (10. yields ∂µ ∂ µ Aν (xσ ) = 4 π J ν (xσ ) .298 Transformation Properties of J µ and Aµ Relativistic Quantum Mechanics The charge density ρ(r. Since ∂µ ∂ µ transforms like a scalar one can conclude that Aν (xσ ) must transform like a 4-vector. Consequently. prove that Aµ is. This principle should hold in any frame of reference. . t) 2 ∂t A(r. We want to derive now the differential equations which determine the 4-potential Aµ in the Lorentz gauge (10.h. (10.99) must be a scalar.

108) yields then the expressions for the transformed fields E and B . equivalently. 8.10. by (10. In a new frame of reference holds F µν = Lµ α Lν β F αβ (10. .109) (10. (10.107)  y 1 z  Bz −v1 Ey √ 2  √ 2  0 −Bx   1−v1 1−v1   Ez +v1 By By +v1 Ez √ 2 − √ 2 Bx 0 1−v1 1−v1 Comparision with F µν  0 −Ex −Ey −Ez  E 0 −Bz By   =  x  Ey B z 0 −Bx  Ez −By Bx 0  (10.97). 10. Noting ∂µ F µν = ∂µ ∂ µ Aν − ∂µ ∂ ν Aµ = ∂µ ∂ µ Aν − ∂ ν ∂µ Aµ = ∂µ ∂ µ Aν .3.63). the components of the fields parallel and perpendicular to the velocity v which determines the Lorentz transformation.102) ∂µ F µν = 4π J ν . These equations show that under Lorentz transformations electric and magnetic fields convert into one another. where we used (10.110) where E .103. t) and B(r. From the definition (10.112) (10.3: Relativistic Electrodynamics 299 The relationships (10. B . Maxwell Equations in Lorentz-Invariant Form One can express the Maxwell equations in terms of the tensor F µν in Lorentz-invariant form.113) which can be shown to be equivalent to the two homogeneous Maxwell equations (8.106) In case that the Lorentz transformation Lµ ν is given by (10. one can conclude from (10.111) One can readily prove that this equation is equivalent to the two inhomogeneous Maxwell equations (8. one obtains  Ey −v B Ez +v B  0 −Ex − √ 1 2z − √ 1 2y 1−v1 1−v1    Bz −v1 Ey By +v1 Ez   √ 2  Ex 0 − √ 2   1−v1 1−v1  F µν =  E −v B (10.105) of the tensor F µν one can conclude the property ∂ σ F µν + ∂ µ F νσ + ∂ ν F σµ = 0 (10.62) or. E⊥ = B⊥ E⊥ + v × B √ 1 − v2 B⊥ − v × E = √ 1 − v2 (10.105) and (10.104) establishe the transformation behaviour of E(r. 8.2). respectively. B and E⊥ .1. t). The results can be put into the more general form E B = = E .4). B⊥ are.

104).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...5). 1 dE 2 1 dp 2 = = qp·E 2 dt 2 dt (10.86) dE q = p·E .5) where p = mv/ 1 − v 2 .119) which is the x-component of the equation of motion (8. We want to √ demonstrate now that this equation is equivalent to the equation of motion (8. for µ = 1.115) This equation follows.5) taking the scalar product with p dp = qp · E (10. i. dt E From this one can conclude. hence. also from the equation of motion (8. of (10.5). employing (10.86). To avoid √ confusion we will employ in the following for the energy of the particle the notation E = m/ 1 − v 2 [see (10. . however.120) is referred to as the Lorentz force. The term on the r. (10.93). Equation (10.117) (10. It holds for a particle with 4-momentum pµ as given by (10.87)] and retain the definition E for the electric field.116) (10.114). dE q = p·E dτ m or with (10.s. provides an alternative description of the action of the Lorentz force. yields dpx = q dt Ex + (v × B)x (10. We have.114) reads then.114) reads using (10.114) is. hence. equivalent to (8.g. For the spatial components.67) and charge q dpµ q = pν F µν dτ m (10. We want to express this force through the tensor F µν . using (10.86) and (10.67).h.e.103) dpx q = ( EEx + py Bz − pz By ) . in fact. The µ = 0 component of (10.120) (10. demonstrated that (10. E = m/ 1 − v 2 . dτ m √ Employing again (10.118) p· dt √ where we exploited the fact that according to p = mv/ 1 − v 2 holds p v.114) where d/dτ is given by (10. e.

i. ψ ∈ C∞ (4). We will denote the intended representation by 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.125)].48) of SO(3). i. J3 ) and K = (K1 .125) def (10.123) This definition corresponds closely to the function space representation (5. w) = ρ(Lµ ν (ϑ.1) is a generalization (rotation + boosts) of the group SO(3).10.47). and which can be constructed following the procedure adopted for the function space representation of SO(3). J2 . in as far as SO(3.121) 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 . Accordingly.. K3 ) are the generators of L(ϑ. w) = exp ϑ · J + w · K . 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 (10.127) .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 ) . def (10. (10.121) covers solely the transformation behaviour of scalar functions. The resulting expression should be a generalization of the function space representation (5. the charge-current density in case of Sect. w. In this section we will investigate the transformation properties of scalar functions ψ(xµ ). obey a different transformation law.121) by (L−1 )µ ν xν and obtains ψ (xµ ) = ψ((L−1 )µ ν xν ) . K2 . (10.e.4: Function Space Representation of Lorentz Group 301 10.3 or the bi-spinor wave function of electron-positron pairs in Sect.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 such functions holds in the transformed frame ψ (Lµ ν xν ) = ψ(xµ ) (10.124) In this expression J = (J1 . We need to emphasize that (10. e. 10. we seek an expression which corresponds to (10. However.7. Functions which represent 4-vectorsor other non-scalar entities. We like to express now ψ (x µ ) in terms of the old coordinates xµ . w)) = ρ eϑ·J + w·K which we present in the form L(ϑ. (10. w) which correspond to the generators Jk and Kk in (10. xµ ).g.51) for the natural representation of the Lorentz group.. (5.e.48) and (10. 10.42) of SO(3). taken at the pairs of points (x µ = Lµ ν xν .126) ρ ewk Kk − 1 1 . in the present case we exclude the factor ‘−i’ [cf. For this purpose one replaces xµ in (10..

K2 ] = −J3 . reproduces the expression for J1 in (10. x1 ∂3 − x3 ∂1 ] = [x3 ∂2 .126).132) and. x2 . K ] [ Jk .302 One obtains J1 = x3 ∂2 − x2 ∂3 .130) ew1 K1 −1 = e−w1 K1 (10. K ] = = k m Jm k m Jm = − k m Km . obviously.134) .131) This yields for small w1 ρ ew1 K1 ψ(xµ ) = ψ(coshw1 x0 + sinhw1 x1 .49) as the generators of the natural representation. cosϑ1 x2 + sinϑ1 x3 . [ Jk . (10. 0  1 (10. x3 ∂1 ] = −x1 ∂2 + x2 ∂1 = J3 . [K1 .128).133) We demonstrate this for three cases. J ] [ Kk . K obey the same Lie algebra (10. namely [J1 .128) eϑ1 J1 = e−ϑ1 J1 1  0 =   0 0   0 0 0 1 0 0   0 cosϑ1 sinϑ1  0 −sinϑ1 cosϑ1 (10. sinhw1 x0 + coshw1 x1 . J2 ] = [x3 ∂2 − x2 ∂3 . which we like to demonstrate for J1 and K1 . x1 ∂3 ] − [x2 ∂3 . i.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.e. (10. −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. J3 = x2 ∂1 − x1 ∂2 . x1 . 1 This result. One can determine similarly K1 starting from coshw1 sinhw1  sinhw1 coshw1 =   0 0 0 0  0 0 1 0  0 0   . the expression for K1 in (10. and [J1 . The generators J . J2 = x1 ∂3 − x3 ∂1 . J2 ] = J3 .129) which yields for small ϑ1 ρ eϑ1 J1 ψ(xµ ) = ψ(x0 . K2 ] = K3 : [ J1 . In order to evaluate (10. obviously.

4: Function Space Representation of Lorentz Group [ K1 . cothΩ = 3 .141) (10. [ J1 . x2 ∂0 ] − [x1 ∂0 . 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 . a relationship which can also be stated R = and + − (x0 )2 − (x3 )2 (x0 )2 − (x3 )2 if x0 ≥ 0 if x0 < 0 (10. The relationships (10. ∂Ω . x0 ∂2 + x2 ∂0 ] = [x3 ∂2 .142) (10. x2 ∂0 ] − [x2 ∂3 . can be simplified considerably. x0 ∂2 ] = x3 ∂0 + x0 ∂3 = K3 ..125) in the case of a one-dimensional transformation of the type L(w3 ) = exp w3 K3 . x3 as follows x0 = R coshΩ . One-Dimensional Function Space Representation The exponential operator (10.136) (10. Ω which are connected with x0 .140) allow one to ∂ ∂ express the derivatives ∂0 .e. K2 ] = [x0 ∂1 + x1 ∂0 .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. x2 + x2 + x2 1 2 3 in the case of SO(3) and (x0 )2 − (x3 )2 in case of transformation (10. (10. We note tanhΩ = ∂R x0 = . K2 ] = [x3 ∂2 − x2 ∂3 .85-5. In both cases the radial coordinate is the quantity conserved under the transformations.139) x3 = R sinhΩ (10. In the following we consider solely the case x0 ≥ 0. Eqs.135) (10.137) where K3 is given in (10. i. For this purpose one expresses K3 in terms of hyperbolic coordinates R. 10.10. x0 ∂2 ] = −x2 ∂1 + x1 ∂2 = − J3 .138) x3 x0 .128).143) . (5.87)]. x0 ∂2 + x2 ∂0 ] = [x0 ∂1 . ∂3 in terms of ∂R . 303 (10.139.137). ∂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 . 0 ∂cothΩ ∂x x = ∂R x0 = − ∂x3 R (10.

is not very insightful.122).145) ∂Ω 10.150) . In natural units the Klein–Gordon equation (10. in 4-vector notation reads pµ =⇒ pµ = i (∂t . Free Particle Case A quantum mechanical description of a relativistic free particle results from applying the correspondence principle. complex-valued function. ∂Ω Relativistic Quantum Mechanics (10.146) Applying the correspondence principle to (10.121. − ) = i ∂ µ . namely the Klein–Gordon and the Dirac equations. however. but certainly is short and. ) = i ∂µ .5 Klein–Gordon Equation In the following Sections we will provide a heuristic derivation of the two most widely used quantum mechanical descriptions in the relativistic regime. ˆ pµ =⇒ pµ = i(∂t . which allows one to replace classical observables by quantum mechanical operators acting on wave functions. extremely useful. ˆ (10.92) one obtains the wave equation − or 2 2 µ ∂ ∂µ ψ(xν ) = m2 ψ(xν ) ψ(xν ) = 0 . We will provide a ‘derivation’ of these two equations which stem from the historical development of relativistic quantum mechanics. The partial differential equation (10. In the following we will employ so-called natural units = c = 1. In these units the quantities energy.148) ∂ µ ∂µ + m2 (10. (10. The historic route to these two equations. therefore. and (time)−1 all have the same dimension.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 ) .151) is called the Klein-Gordon equation. The latter property implies that upon change of reference frame ψ(xµ ) transforms according to (10. Further below we will provide a more systematic.304 Inserting these results into the definition of K3 in (10.151) reads ∂µ ∂ µ + m2 ψ(xµ ) = 0 (10. momentum. (length)−1 .147) (10.128) yields K3 = x0 ∂3 + x3 ∂0 = ∂ . 10.144) The action of the exponential operator (10. In the position representation the correspondence principle states ˆ E =⇒ E = − ∂t i ˆ p =⇒ p = i which. mass.149) where ψ(xµ ) is a scalar. (10. representation theoretic treatment.

t) ] 2mi (10.5: Klein–Gordon Equation or 2 ∂t − 2 305 + m2 ψ(xµ ) = 0 .155) describes the positive definite probability to detect a particle at position r at time t and where jS (r. To 1 derive this conservation law one rewrites the Schr¨dinger equation in the form (i∂t − 2m 2 )ψ = 0 o and considers 1 2 Im ψ ∗ i∂t − ψ = 0 (10.157) 2m which is equivalent to (10.150) one considers Im ψ ∗ ∂µ ∂ µ + m2 ψ = 0 (10. and that (as postulated) ψ(xµ ) is a scalar. t) ψ(r. t) (10. t) = · jKG (r. t)∂t ψ(r. t) 1 [ ψ ∗ (r. Under Lorentz transformations the free particle Klein–Gordon equation (10.150) becomes ∂µ ∂ µ + m2 ψ (x µ ) = 0 (10. t) = (10.160) ∂t (ψ ∗ ∂t ψ which corresponds to − ψ∂t ψ∗) + · (ψ ψ∗ ∂t ρKG (r.161) .152) which has the same form as the Klein–Gordon equation in the original frame.158) which yields 2 2 ψ ∗ ∂t ψ − ψ∂t ψ ∗ − ψ ∗ 2ψ +ψ ψ∗ − 2ψ∗ = ψ) = 0 (10. t) ψ(r. t) ψ ∗ (r.153) is associated with a conservation law for particle probability ∂t ρS (r. t) ψ ∗ (r. In order to obtain the conservation law connected with the Klein–Gordon equation (10.154). t) + where ρS (r. 2mi (10. 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) − ψ(r.150) is invariant under Lorentz transformations. t) = 0 i ( ψ ∗ (r. t) ψ(r.10. t)∂t ψ ∗ (r. t) − ψ(r. t) = ψ ∗ (r. t) − ψ(r. t) ) .156) · jS (r.154) describes the current density connected with motion of the particle probability distribution. t) jKG (r. t) ) 2m 1 = ( ψ ∗ (r.151) One can notice immeadiately that (10. t) + where ρKG (r. (10. t) = 0 (10. t) = − 1 2m 2 ψ(r.159) (10. This follows from the fact that ∂µ ∂ µ and m2 are scalars.

(10. m 2 2 E0 = m2 + po . corresponding to λ = −. is actually that of a charge density. Generating a Solution Through Lorentz Transformation A particle at rest.306 Relativistic Quantum Mechanics This conservation law differs in one important aspect from that of the Schr¨dinger equation (10. λ|xµ )ψo (p. o namely.p ei(p·r − λEo (p)t) Eo (p) = The spectrum of the Klein–Gordon equation (10.165) This result together with (10.162) shows that the solutions of the free particle Klein-Gordon equation (10.150) are actually determined by po and by the choice of sign ±. t) = 0 (10. and of negative energies E ≤ −m.166) ψo (p. it had been realized later. 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. (10. λ = ± (10. corresponding to λ = +.166) is Eo (p) ∗ ρKG (p.163) which results in the expected [see (10.167) m which is positive for λ = + and negative for λ = −. The density ρKG (p.164) The corresponding energy is Eo (po .162) Inserting this into the Klein–Gordon equation (10. λ) associated with the corresponding wave functions ψo (p.93] dispersion relationship connecting E0 . λ) actually describes charge density rather than probability.168) .. with p = 0. λ|xµ ) = Nλ. and the density ρKG (p.161) and (10. in that the expression for ρKG is not positive definite. the anti-particles carry a charge opposite to that of the associated particles. Today. according to (??) is decribed by the r–independent wave function ψo (p = 0. p0 . The proper interpretation of ρKG (r. We denote this by summarizing the solutions as follows ∂µ ∂ µ + m2 ψo (p. (10. not of particle probability. λ|xµ ) (10. i. λ = ± .151) yields 2 2 Eo − p0 − m2 ψ(r. The proper interpretation of the two cases is that the Klein–Gordon equation describes particles as well as anti-particles. λ) = λ ψo (p. λ|xµ ) = 0 2 m2 + po .154). for example pions. t).150) is a continuum of positive energies E ≥ m. λ|xµ ) = N e−iλmt . the Klein-Gordon equation is considered as a suitable equation to describe spin–0 particles. λ|xµ ) according to (10.e. ±) = ± 2 m2 + po (10. 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) .

f. Ω.173) (10.e. of (10. Such wave function should be generated by applying the Lorentz transformation in the function space representation p (10.145) choosing m = sinhw3 .61) yield for this expression −iλ √ m mv x0 − iλ √ x3 . for the wave function (10. λ = −|xµ ) = N ei(px 3 + Eo (p)x0 . t)) . For this purpose we consider the wave function for a particle moving with momentum velocity v in the direction of the x3 –axis.168).177) . 2 2 1−v 1−v (10.171) One can interpret then for λ = +. −A(r..138) and the relationships (10. In case of λ = + one obtains finally L(w3 )ψo (p = 0.. In case of λ = −.10.172) as the energy [c. This yields.166)] of the particle.138) to replace t = x0 by hyperbolic coordinates R.h. λ = +|xµ ) = N ei(px 3 − Eo (p)x0 (10. L(w3 )ψo (p = 0.6 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. (10. for negative energy solutions. λ|xµ ) = = exp w3 ∂ ∂Ω N e−iλmRcoshΩ 3) N e−iλmRcosh(Ω+w .169) −iλ ( m coshw3 ) ( R coshΩ ) − iλ ( m sinhw3 ) ( R sinhΩ ) . for positive energy solutions.175) as the momentum of the particle and one obtains L(w3 )ψo (p = 0. Aµ (xµ ) = (V (r. A(r. (10. in fact.s. 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. t)) (10.169) The addition theorem of hyperbolic functions cosh(Ω + w3 ) = coshΩ coshw3 + sinhΩ sinhw3 allows us to rewrite the exponent on the r. (10.176) 10. i. one has to interprete p = mv/ 1 − v 2 (10. using (10.170) The coordinate transformation (10.166). i.e.168). t). t).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 (10.174) which agrees with the expression given in (10. (10.

1: Coupling of a particle of charge q to an electromagnetic field described by the 4-vector potential Aµ = (V. and focus on the remaining part of the wave function. t) 2 (10. that the mass m of the particle.180) + m2 ψ(r. i. m2 ) ψ(xµ ) = 0 . (10. albeit in a form. t) = −i − q A(r. In terms of space-time derivatives this reads (i∂t − qV (r. it’s rest energy. momenta for classical particles and the respective operators for quantum mechanical particles in the manner shown. We will also assume. For this purpose we split-off a factor exp(−imt) which describes the oscillations of the wave function due to the rest energy. t) .. t).147). is much larger than all other energy terms. A) or Aµ = (V. (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. which couples a particle of charge q to an electromagnetic field described through the potential Aµ (xν ). i. According to the so-called minimum coupling principle the presence of the field is accounted for by altering energy. t) = e−imt Ψ(r. (10. according to the rules shown in Table 10. in keeping withnthe non-relativistic limit. −A).182) and seek an equation for Ψ(r.1. (10.. i∂t and −i in (10.181) 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.93). i..150) −g µν (i∂µ )(i∂ν ) + m2 ψ(xµ ) = 0 .e.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 − .308 free classical particle classical particle in field (V. For this purpose one writes the Klein-Gordon equation (10.178) According to the replacements in Table 10.150). t))2 ψ(r.e. in .69)] one replaces the quantum mechanical operators. 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. A) free quantum particle quantum particle in field (V. According to the principle of minimal coupling [see (10. t) .e.179) (10. See also Eq. we define ψ(r.

t) ≡ 0 r + Ze2 r 2 + 2 − m2 π ϕ(r ) = 0 .s.186) then are slowed down. e.g. A stationary state of the Klein–Gordon equation is described by a wave function ψ(xµ ) = ϕ(r ) e−i t . Pionic Atoms To apply the Klein–Gordon equation (10.183).2) of a non-relativistic spin-0 particle o moving in an electromagnetic field. 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.184) where we neglected all terms which did not contain factors m. π − mesons are generated through inelastic proton–proton scattering p + p −→ p + p + π − + π + . t)]2 + qV (r.183) The term on the l. t) = ˆ [p − q A(r. 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− . identical to the Schr¨dinger equation (10.187) 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. t) 2m Ψ(r. (10. The Klein-Gordon equation (10.. The process of π − meson capture involves the so-called Auger effect. t) = − Ze and A(r.185) This is. | i∂t Ψ | << m . filtered out of the beam and finally fall as slow pions onto elements for which a pionic variant is to be studied. i. of (10. (10.181) to a physical system we consider pionic atoms.h. i. To ‘manufacture’ pionic atoms. This application demonstrates that the Klein–Gordon equation describes spin zero particles. atoms in which one or more electrons are replaced by π − mesons.e.189) . Ψ |q V | << m .10.. (10. The approximation is justified on the ground of the inequalities (10. 309 (10. however.. larger than |i∂t Ψ/Ψ| and alrger than qV .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.e.181) reads then i ∂t Ψ(r. (10. t) (10.188) Inserting this into (10.6: Klein–Gordon Equation with Electromagnetic Field particular.

φ) = ( + 1) Y m (θ. φ) (10. n = 1. .191) ˆ The operator L2 in this equation suggests to choose a solution of the type ϕ(r ) = where the functions Y in (10. i.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..195) The latter problem leads to the well-known spectrum En = − mπ (Ze2 )2 . (10. this quantity definitely must be an integer. n − 1 .310 Relativistic Quantum Mechanics Because of the radial symmetry of the Coulomb potential we express this equation in terms of spherical coordinates r.e. (10. .e. 2n2 = 0. . (10. 2. 1.194) 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. .193) (10. the eigenfunctions of the operator L2 ˆ L2 Y m (θ. i. . .192) leads then to the ordinary differential equation d2 − dr2 ( + 1) − Z 2 e4 2 Ze2 + + r2 r 2 − m2 π R (r) = 0 .197) counts the number of nodes of the wave function. 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 .192) ˆ are spherical harmonics. φ. (10.190) With this expression and after expanding ( + one obtains 2 ˆ d2 L2 − Z 2 e4 2 Ze2 − + + dr2 r2 r − m2 π r φ(r) = 0 . r r r r sin θ r sin θ Ze2 2 r ) (10. . φ) .191) m (θ. The similarity of (10. θ.. (10.199) ( + 1) − Z 2 e4 . φ) R (r) Y r m (θ.194) and (10.198) .196) In this expression the number n defined through n = n− −1 (10.

202) Solving this for state) yields (choosing the root which renders = 1 + mπ Z 2 e4 ( n + λ( )+1 )2 ≤ mπ .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 ) . . . . . . . 1. . .204) reads 1 (10. .10.197.e. . i.200) The bound state solutions of this equation should correspond to from (10. Introducing α = Z 2 e4 and 1 β = + 2 (10. EKG = Using (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 ).201) mπ Z 2 e4 m2 − m2 π π = − . 1. .. 1.6: Klein–Gordon Equation with Electromagnetic Field and one can write (10. . . . (10. n = 0. . which corresponds to a bound = 0. (10. . 10. . n − 1 m = − . − + 1. = 0.196) if one makes the replacement 2 values which can be obtained E −→ One obtains − m2 π . 2mπ 2 (n + λ( ) + 1)2 (10. . results in the (10.199) and defining EKG (n. 2mπ 2 −→ λ( ) . . 2 e2 −→ e2 mπ . 2. m) = mπ 1 + Z 2 e4 (n− − 1 2 + ( + 1 )2 − Z 2 e4 )2 2 n = 1. . + (10.203) spectrum .194) d2 λ( ) ( λ( ) + 1 ) 2 Ze2 − + + dr2 r2 r 2 311 − m2 π R (r) = 0 .

10. 10. a feature which is connected with the 2nd order time derivative in this equation. t) . in fact. however. Equations (10. the π − meson is a pseudoscalar particle. It should feature then a differential operator of the type D = iγ µ ∂µ . arises because the Klein–Gordon equation. it does not agree with observations of the hydrogen spectrum. however. one with a positive-definite density. for example.. t) = ± m2 − 2 Ψ(r. Also. The latter term agrees with observations of pionic atoms. 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. the Dirac equation introduced below.. but not an operator iγ · as suggested by the principle of relativity (equivalence of space and time). In order to describe electron spectra one must employ the Lorentz-invariant 1 wave equation for spin. 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. through the correspondence principle.e. 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 . However.146) leads to the wave equation i∂t Ψ(r.e.209).209) which.e. (10. i.208). His solution . and the third term gives the leading relativistic correction.312 ≈ From this results for (10. would have only a first order time derivative. is related to the equation E 2 = m2 + p 2 of the classical theory which involves a term E 2 . a splitting of the six n = 2. the wave function changes sign under reflection. are unsatisfactory since expansion of the square root operator involves all powers of the Laplace operator. Finally.207) Here the first term represents the rest energy. Dirac succeeded. = 1 states into groups of two and four degenerate states. is identical to the two equations (10. It must be pointed out here that does not denote energy.208.. 10. The latter spectrum shows. This derivative. m) ≈ m − mZ 2 e4 mZ 4 e8 − 2n2 2n3 1 + − 1− Relativistic Quantum Mechanics α α2 α2 α2 − + + + O(α3 ) . i. A more satisfactory Lorentz–invariant wave equation. the second term the non-relativistic energy.204) EKG (n.208. 2n2 2βn3 8n4 4n4 (10. Application of the correspondence principle (10.7 The Dirac Equation Historically. i.209). but for a long time to no avail. but in the present case rather the negative of the energy.206) 1 2 3 + O(Z 6 e12 ) . t) (10. in turn. 4n (10.208) These two equation can be combined i∂t + m2 − 2 i∂t − m2 − 2 Ψ(r. .2 particles.

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 (10.213) reads (γ µ )2 = 1 µ = 0 −1 µ = 1. The discovery by Dirac. but did not commute the. 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 . In fact.215) . However. ψ2 (xµ ). spin.210) where P = iγ µ ∂µ . therefore. It turns out that no 4-vector of real or complex coefficients can satisfy these conditions. quantities with the transformation law (10. γ ν ]+ = 2 g µν (10. these descriptions ultimately merging with the true theory of matter.e. 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 ). 3 . γ 1 . (10. the 4-dimensionsional representation discovered by Dirac involved new physical properties. We seek to identify the coefficients γ µ . For µ = ν condition (10. achieved through a beautiful mathematical theory. Initially.66). exploited ∂µ ∂ν = ∂ν ∂µ . i. .209).212) where we have changed ‘dummy’ summation indices. 2. Inserting (10. the quantities γ 0 .213) We want to determine now the simplest algebraic realization of γ µ . . γ 2 . therefore.211) into (10. one can impose the condition γ0 is hermitian .211) Obviously. according to (10. .10.. ψd (xµ ). unspecified algebraic objects γ µ and γ ν . the operator of the Klein–Gordon equation ∂µ ∂ µ + m2 = −( P + m ) ( P − m ) (10. the realization that the linearization can be carried out only if one assumes a 4-dimensional representation of the coefficients γ µ . Instead.1 2 and anti-particles.7: Dirac Equation 313 to the problem involved an ingenious step. j = 1. are associated with a positive-definite particle density.208). Properties of the Dirac Matrices Let us now trace Dirac’s steps in achieving the linearization of the ‘square root operator’ in (10.214) From this follows that γ 0 has real eigenvalues ±1 and γ j . one route to important discoveries in physics is the creation of new mathematical descriptions of nature. it was assumed that the 4-dimensional space introduced by Dirac could be linked to 4vectors. γ j . 2. yet to be discovered. 3 are anti-hermitian . (10. Accordingly. so far. Starting point is to boldly factorize. 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. j = 1. (10. 2. beautiful mathematical theory and that. this was not so. strengthens the believe of many theoretical physicists today that the properties of physical matter ultimately derive from a. 3 has imaginary eigenvalues ±i. namely.

is implied by (10. in fact.213) reads γ µ γ ν = −γ ν γ µ . to construct four matrices γ µ for the next possible dimension d = 4. (10.213) is satisfied.218) of the Pauli matrices one can readily prove that condition (10.220) where Ψ(xµ ) represents a 4-dimensional wave function.218) The Pauli matrices allow one. σ 2 .314 For µ = ν (10. (10.219) Using property (10. however.224) .216) i.223) Inserting this into (10. σ 3 for which. 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. ψ2 . σ j σ k = − σ k σ j for j = k . For d = 2 there exist only three anti-commuting matrices. as long as det(γ µ ) = 0 the dimension d of the square matrices γ µ must be even so that (−1)d = 1. namely the Pauli matrices σ 1 .221) =0 (10. From this one can conclude for the determinants of γ µ det(γ µ γ ν ) = det(−γ ν γ µ ) = (−1)d det(γ ν γ µ ) = (−1)d det(γ µ γ ν ) .215). 3 which.. ψ3 . A proper choice is γ0 = 1 1 0 0 −1 1 . We will argue further below that the choice f γ µ .e. the γ µ are anti-commuting. 2.222) and multiplication from the right by γ 0 yields the adjoint Dirac equation Ψ† (xµ ) γ 0 iγ µ ∂µ + m ← =0. ψ4 ) and where ∂µ denotes the differential operator ∂ µ operating to the left side. This property can also be written (γ µ )† = γ 0 γ µ γ 0 . rather than to the right side. rather than a scalar wave function. (10. Relativistic Quantum Mechanics (10. The Adjoint Dirac Equation The adjoint equation is Ψ† (xµ ) i(γ µ )† ∂µ + m ← ← (10. except for similarity trasnformations. is unique. 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. in fact.222) ∗ ∗ ∗ ∗ where we have defined Ψ† = (ψ1 . holds 2 σj = 1 . One can readily show using the hermitian property of the Pauli matrices (γ 0 )† = γ 0 and (γ j )† = −γ j for j = 1.217) Obviously. (10. γj = 0 −σ j σj 0 . 1 (10.

(10. (10. (10..233) The eigenstates and eigenvalues of H correspond to the stationary states and energies of the particles described by the Dirac equation.225) .1: Derive (refeq:Dirac-intro20a) from (10.222.7.e.228) (10. αj = ˆ 0 σj σj 0 .229) The similarity transformation (10. 3 and ˆ β = 1 1 0 0 −1 1 . (10.231) ˆ whereα has the three components αj .226) (10.232) 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 ) . 3 . 2. 10.227). [˜ µ . γ ν ]+ = 2 g µν .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. 3 . γ ˜ Exercise 10. j = 1. (10. j = 1. . i.221) from the left by γ 0 ˆ ˆ i ∂t + i α · − β m Ψ(r.230) Schr¨dinger Form of the Dirac Equation o Another form in which the Dirac equation is used often results from multiplying (10. j = 1. 2. t) = 0 (10.227) leaves the algebra of the Dirac matrices unaffected and commutation property (10.Chiral Representation 315 The Dirac equation can be subject to any similarity transformation defined through a non-singular ˜ 4 × 4–matrix S. ˆ ˆ ˆ Ho = α · p + β m . 2. γj = ˜ 0 σj 1 S = √ 2 −σ j 0 1 1 1 1 1 −1 1 1 (10. and γ0 = ˜ 0 1 1 1 0 1 .213).7: Dirac Equation Similarity Transformations of the Dirac Equation .219) is the socalled chiral representation defined through ˜ Ψ(xµ ) = S Ψ(xµ ) .10.213) still holds.

236) where ‘⊗’ denotes the Kronecker product between matrices. 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 . d4 is called a Clifford algebra C4 . For example. form a Clifford algebra C3 . d2 = . The three Pauli matrices also obey the property (10.234) (10. Γ±7 = ±d1 d3 . 27. Γ±9 = ±d2 d3 Γ±10 = ±d2 d4 . (1955).235) can differ only with respect to unitary similarity transformations.238) These elements form a group since obviously any product of two Γr ’s can be expressed in terms of a third Γr .235) and. There are (including the different signs) 32 elements in G4 which we define as follows Γ±1 = ±1 1 Γ±2 = ±d1 . Rev. hence. This property extends then to 4 × 4–matrices which obey (10. Γ±15 = ±d2 d3 d4 Γ±16 = ±d1 d2 d3 d4 (10. d2 . . Γ±11 = ±d3 d4 Γ±12 = ±d1 d2 d3 . 2.213)..Good.e.H.237) which are the ordered products of the operators ±1 and d1 . satisfy the anti-commutation property dj dk + dk dj = 2 for j = k 0 for j = k Relativistic Quantum Mechanics d4 = γ 0 (10.Phys.237) by means of the property (10. d4 . page 187.235) as can be readily verified from (10.213). The associative algebra generated by d1 .6 and R. i. Γ±14 = ±d1 d3 d4 . Obviously.. Γ±4 = ±d3 . j = 1. Γ±5 = ±d4 Γ±6 = ±d1 d2 . One can conclude then that also any set of 4 × 4–matrices obeying (10. m = Aj Bkm . To complete the proof in this section the reader may consult Miller ‘Symmetry Groups and their Application’ Chapter 9. The representations of this group are given by a set of 32 4 × 4–matrices which are equivalent with respect to similarity transformations. the matrix elements of C = A ⊗ B are Cjk. Γ±3 = ±d2 . The representations of Clifford algebras Cm are well established. 3 . Γ±13 = ±d1 d2 d4 . d3 . .235). Since the Γj are hermitian the similarity transformations are actually given in terms of unitary transformations.316 Clifford Algebra and Dirac Matrices The matrices defined through dj = iγ j . The reader may . j1 < j2 < · · · < js s ≤ 4 (10. any product 1 of the dj ’s can be brought to the form (10. to Dirac matrices. The Clifford algebra C4 entails a subgroup G4 of elements ± dj1 dj2 · · · djs .Mod. d4 = 1 0 0 −1 1 0 0 −1 0 1 1 0 0 i −i 0 (10. Γ±8 = ±d1 d4 . i.e. in case of C4 .

in a moving frame should hold iγ ∂µ − m Ψ (x ) = 0 .2: Demonstrate the anti-commutation relationships (10. and γ j . in fact. We can. Exercise 7. However. namely. exist.239) Form invariance implies that the matrices γ µ should have the same properties as γ µ .3: Demonstrate the anti-commutation relationships (10. 10. 3 has imaginary eigenvalues ±i and can be represented by an anti-hermitian matrix. the form of the Dirac equation is identical in equivalent frames of reference. Exercise 7.. it must hold γ µ = γ µ .241) .e.8: Lorentz Invariance of the Dirac Equation 317 also want to establish the unitary transformation which connects the Dirac matrices in the form (10.4: Show that from (10. 10. µ µ (10.6). therefore.. we do not know yet how to transform the 4-dimensional wave function Ψ and the Dirac matrices γ µ . i. i.219). 2.. a result one could have also obtained by applying the chain rule to (10.80) by Lµ ρ and using (10.214) follows that γ 0 has real eigenvalues ±1 and can be represented by a hermitian matrix. and derivatives according to (10.. x = Lµ ν xν .e.218) of the Dirac matrices γ µ .. Lorentz Transformation of the Bispinor State Actually. Multiplication and summation of (10.80). i. j = 1.76) yields ∂ρ = Lν ρ ∂ ν .213. i.218) of the Pauli matrices σ j . 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 holds iγ µ ∂µ − m Ψ (x ) = 0 . transform coordinates and derivatives of the Dirac equation.e. The latter transformations imply that coordinates transform according to (10. i.215).236) with the Dirac representation (10. Infinitesimal Bispinor State Transformation We want to show now that a suitable transformation of Ψ(xµ ) does. 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.e.8 Lorentz Invariance of the Dirac Equation We want to show now that the Dirac equation is invariant under Lorentz transformations.e.240) . Hence. these properties determine the matrices γ µ uniquely.6). Exercise 7. (10.10. x µ = Lµ ν xν (10. in frames connected by Lorentz transformations. Except for a similarity transformation.

it must hold (10.. One can finally conclude multiplying both sides by g ρµ S(Lη ξ )γ µ S −1 (Lη ξ ) = Lν µ γν . (10.e. µ g ρν = µν and. νµ (10.243) We want to determine now the 4 × 4–matrix S(Lη ξ ) which satisfies this condition. the transformation (10.12) in the form Lν µ gνσ Lσ ρ = gµρ = gρµ .249) from which we can conclude σµν = −σµν νµ = −σνµ i.221) upon Lorentz transformation yields iS(Lη ξ )γ µ S −1 (Lη ξ ) Lν µ ∂ν − m Ψ (x ) = 0 . i. For this purpose we exploit (10.246) (10.. S( µν ) should not change its value if one replaces in its argument µν by − νµ . One can.h.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. hence. the coefficients of which depend on the matrix Lµ ν defining the Lorentz transformation in such a way that S(Lµ ν ) = 1 for Lµ ν = δ µ ν holds. (10. g is an anti-symmetric matrix. .s. of the equation. The proper starting point for a constructiuon of S(Lη ξ ) is actually (10. We had introduced in (10. Multiplication of this property by g from the right yields ( g)T = − g. Ob1 viously. µ (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 . follows S(Lη ξ )γρ S −1 (Lη ξ ) = Lσ ρ γσ . It holds then S( µν i ) = 1 − σµν 1 4 µν µν = S(− νµ i ) = 1 + σµν 1 4 µν .247) the infinitesimal matrix µν is anti-symmetric.250) σµν = −σνµ .248) Here σµν denote 4 × 4–matrices operating in the 4-dimensional space of the wave functions Ψ. state that the Dirac equation (10. using gρµ γ µ = γρ .243) in a form in which the Lorentz transformation in the form Lµν is on the r.245) The construction of S(Lη ξ ) will proceed using the avenue of infinitesimal transformations. however.e.244) from which. The infinitesimal transformation S(Lρ σ ) which corresponds to (10. in the expression of the infinitesimal transformation ρ Lµν = g µν + µν (10.318 Relativistic Quantum Mechanics where S(Lµ ν ) is a non-singular 4 × 4–matrix. (10.243) from the left by Lσ ρ gσν yields S(Lη ξ )γ µ S −1 (Lη ξ ) gρµ = Lσ ρ gσν γσ .247) can be expanded S( µν i ) = 1 − σµν 1 4 µν (10.241) implies a similarity transformation Sγ µ S −1 . Multiplication of (10. The elements of g are. hence.

γβ ]− 2 (10. j = 1.h.246) results then in a condition for the generators σµν − i ( σαβ γ µ − γ µ σαβ ) 4 αβ = νµ γν .254). For this purpose we inspect the properties of the generators in a particular representation.. Comparing this with the l. namely. 10.213. a solution of condition (10. (10. (10.10. (10.254) is σαβ = i [ γα .255) which can be demonstrated using the properties (10.e. (10. of (10. 2.h. γk ] = ˜ γ ˜ jk σ 0 0 σ .251) S −1 ( µν ) = 1 + σµν µν 1 4 Inserting (10.8: Lorentz Invariance of the Dirac Equation 319 One can readily show expanding SS −1 = 1 to first order in µν that for the inverse infinitesimal 1 transformation holds i (10.251) into (10.216) of the Dirac matrices.255) satisfy condition (10.s. 10. β [ σαβ . For this purpose we need to verify that the algebra of the generators involving addition and multiplication is closed. ˜ ˜ σ only six generators need to be determined) σ0j = ˜ i [˜0 . i. of (10. (10.254) The proper σαβ must be anti-symmetric in the indices α. Furthermore.. β and operate in the same space as the Dirac matrices.252) Since six of the coefficients αβ can be chosen independently. −γ ) are ˜ γ ˜ γ0 = ˜ 0 1 1 1 0 1 . γ µ ]− = 2i ( δ µ β γα − δ µ α γβ ) . the expression on the r. Exercise 7.256) One can readily verify that the non-vanishing generators σµν are given by (note σµν = −˜νµ . In fact.252) results in the condition for each α. σjk = [˜j .252) also as a sum over both indices of αβ . the chiral representation introduced above in Eqs. For this purpose one needs to express formally the r. 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 (10.253) 1 2 αβ ( δ µ β γα − δ µ α γβ ) .s. this condition can actually be expressed through six independent conditions. like the expression on the l.257) . γj ] = γ ˜ 2 −iσ j 0 0 iσ j .229). In this representation the Dirac matrices γµ = (˜ 0 .s.h.s. must be symmetric with respect to interchange of the indices α and β.5: Show that the σαβ defined through (10.228. 3 . γj = ˜ 0 −σ j σj 0 .h. Algebra of Generators of Bispinor Transformation We want to construct the bispinor Lorentz transformation by exponentiating the generators σµν .248. 10.

In fact.. since both operations convert block-diagonal operators A 0 0 B (10. any representation of the generators. 2 0 −(w + iϑ) · σ 13 + σ23 ˜ 23 ) (10. µν in the exponential form i S = exp − σµν 4 µν . the algebra of these generators is closed under addition and multiplication.259) We had stated before that the transformation S is actually determined through the Lorentz transformation Lµ ν . in particular. ϑ) = e 2 (w − iϑ)·σ 0 − 1 (w + iϑ)·σ 0 e 2 1 (10. One should.44)  0 −w1 −w2 −w3  w 0 ϑ3 −ϑ2   =  1  w2 −ϑ3 0 ϑ1  w3 ϑ2 −ϑ1 0  µν (10. (10.258) again into block-diagonal operators. Finite Bispinor Transformation The closedness of the algebra of the generators σµν defined through (10. not necessarily infinitesimal. therefore. and since the algebra of the Pauli matrices is closed. .51).248) allows us to write the transformation S for any.e. i.51) and S. one can express the tensor µν through w and ϑ using µν = µ ρ g ρν and the expression (10.260) Inserting this into (10. We can finally note that the closedness of the algebra of the generators σµν is not affected by similarity transformations and that. the representation (10.262) This expression allows one to transform according to (10.241) bispinor wave functions from one frame of reference into another frame of reference. Since such operator does not change its block-diagonal form upon exponentiation the bispinor transformation (10.255) yields a closed algebra. In order to construct an explicit expression of S in terms of w and ϑ we employ again the chiral representation. 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 = .261) We note that this operator is block-diagonal. be able to state S in terms of the same parameters w and ϑ as the Lorentz transformation in (10. therefore.259) yields the desired connection between the Lorentz transformation (10.320 Relativistic Quantum Mechanics Obviously.259) becomes in the chiral representation ˜ S(w.

263) does hold. (10. and addition yields Ψ† (xµ )γ 0 The last result can be written ∂µ Ψ† (xν )γ 0 γ µ Ψ(xν ) = 0 .10.265) (10. Since ∂µ transforms like a covariant 4-vector. the conservation law (10. (10. The reason is that the r.e. j) = Ψ† (xµ )γ 0 γ µ Ψ(xµ ) . The time-like component ρ of j µ 4 iγ µ ∂µ + iγ µ ∂µ ← Ψ(xµ ) = 0 .266) ρ(xµ ) = Ψ† (xµ )Ψ(xµ ) = s=1 |ψs (xµ )|2 (10. This transformation behaviour can also be deduced from the transformation properties of the bispinor wave function Ψ(xµ ). ϑ) = S(−w. j µ must transform like a contravariant 4-vector.270) We conclude that (10.221) and the adjoint Dirac equation (10.268) We will prove this property in the chiral representation. 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 . for j µ (xµ ) = (ρ.263) allows one to conclude that j µ must transform like a contravariant 4-vector as the notation implies. We will now determine the relationship between the flux j µ = Ψ (x )γ 0 γ µ Ψ (x ) † µ (10. (10. of (10. (10.263) Starting point are the Dirac equation in the form (10. (10. the property applies then in any representation of S.. in fact. Obviously.221) from the left by Ψ† (xµ )γ 0 .h. For this purpose we prove first the relationship S −1 = γ 0 S † γ 0 .224) from the right by Ψ(xµ ).264) (10.s.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 . The conservation law (10. For our proof we note first ˜ ˜ S −1 (w.267) has the desired property of being positive definite. −ϑ) = e− 2 (w − iϑ)·σ 0 1 (w + iϑ)·σ 0 e2 1 (10.263) obviously is a scalar under Lorentz transformations and that the left hand side must then also transform like a scalar.269) One can readily show that the same operator is obtained evaluating γ S (w. Multiplying (10. i.224).268) holds for the bispinor Lorentz transformation.271) .

(10. the wave function has four components which invite further characterization of the free particle state. The solutions provide also a complete. t) and is determined through the momentum p ∈ R3 .233). These operaˆ tors.271) that the Dirac matrices are independent of the frame of reference.246) S −1 (Lη ξ )γ µ S ( Lη ξ ) = (L−1 )νµ γν = (L−1 )ν γ ν = Lµ ν γν . Like for the free particle wave functions of the non-relativistic Schr¨dinger and the Klein–Gordon o equations one expects that the space–time dependence is governed by a factor exp[i(p · r − t)].273) where we have employed (10. In the following we want to provide this characterization. For this purpose we consider .9 Solutions of the Free Particle Dirac Equation We want to determine now the wave functions of free particles described by the Dirac equation. only two degrees of freedom of the bispinor four degrees of freedom are independent [c.232. In relativistic quantum mechanics a Dirac particle can also be characterized through a momentum. commute with each other.272) With S −1 (Lη ξ ) = S((L−1 )η ξ ) one can restate (10. 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.268) µ j = Ψ† (xµ )S † γ 0 γ µ SΨ(xµ ) = Ψ† (xµ )γ 0 S −1 γ µ SΨ(xµ ) . (10. however. One obtains using (10. Note that we have assumed in (10. In case of non-relativistic quantum mechanics the free particle wave function has a single component ψ(r.322 Relativistic Quantum Mechanics in a moving frame of reference and the flux j µ in a frame at rest. 10. (10. As it turns out.274) 10.272) results in the expected transformation behaviour µ j = Lµ ν j ν . but also as providing a basis in which the wave functions of interacting particle systems can be expanded and characterized. µ (10. Like in non-relativistic quantum mechanics the free particle wave function plays a central role. i.e.275) The free particle wave function is an eigenfunction of Ho . as is required for the mentioned property. as just mentioned. not only as the most simple demonstration of the theory. 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.f. bispinor wave functions and we need to determine corresponding components of the wave function. additional characterizations. specific for the Dirac free particle.231.283)]. The additional degrees of freedom described by the four components of the bispinor wave function require. the identification of observables and their quantum mechanical operators. (10..76). the Dirac particles are described by 4-dimensional. As pointed out. The operators commute also with Ho in (10. Combining this with (10.282. a property which leads to the energy– momentum (dispersion) relationship of the Dirac particle. of which the wave functions are eigenfunctions as well.233) o i∂t Ψ(xµ ) = Ho Ψ(xµ ) . 10. We will start from the Dirac equation in the Schr¨dinger form (10.

namely φo and χo are related as follows φ0 χo = = σ·p χ0 −m σ·p φ0 . We want to show now that these degrees of freedom correspond to a spin-like property. one finds using (5. 10.94) Equation (10.282) (10. (10.281). from (10.278) Multiplication of the 1st equation by ( +m)1 and of the second equation by −σ · p and subtraction 1 of the results yields the 2-dimensional equation ( 2 − m2 ) 1 − (σ · p)2 φo = 0 . +m (10. and φo .278) provides us with information about the components of the bispinor wave function (10. hence.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. 10.277) To solve this problem we write (10.279) According to the property (5.232) leads to the 4-dimensional eigenvalue problem m σ·p σ · p −m φo χo = φo χo .276). 10.282. In fact. Inserting (10.276) into (10. (10. later to be identified with momentum and energy. These two relationships are consistent with each other. E(p) = m2 + p 2 .280) which has a positive and a negative solution = ± E(p) .276) where p and together represent four real constants.10.284) The relationships (10.283) where is defined in (10. conclude 1.231.279) the well-known relativistic dispersion relationship 2 = m2 + p 2 (10. 2 − m2 0 (10.280) σ·p (σ · p)2 φ0 = χ0 = +m ( + m) ( + m) p2 χ = χ0 . two-dimensional spinor state.234) and (10. .234) of Pauli matrices holds (σ · p)2 = p 2 1 One can.283) imply that the bispinor part of the wave function allows only two degrees of freedom to be chosen independently. reproduce the classical relativistic energy–momentum relationships (10. (10. χo each represent a constant.281) Obviously. like the Klein–Gordon equation. 1 (10. the Dirac equation.93. the socalled helicity of the particle.277) explicitly ( − m) 1 φo − σ · p χo 1 − σ · p φo + ( + m) 1 χo 1 = 0 = 0.

= +E(p) .290) should differ. i.e. the solution for = −E(p). One can see this as follows: Let Ψ(p. (10.219) and u† u = 1 [c. The form of this condition and of (10.e. +) = 1 . ±) S † γ 0 S Ψ(0. (10. we present φo through the normalized vector φo = u1 u2 = u ∈ C2 . +) γ 0 Ψ† (p. (10. the solution for = +E(p).287) in the minus sign on the r.324 Relativistic Quantum Mechanics For our further characterization we will deal with the positive and negative energy solutions [cf. 0) γ 0 u 0 = |N+ (0)|2 . is invariant under Lorentz transformations.h.281)] separately. −|xµ ) = N− (p) −σ·p E(p) + m u u ei(p·r − t) . Hence.285)].287.. u† u = |u1 |2 + |u2 |2 = 1 . i. ±) . i. (10. ±) γ 0 Ψ(0. Ψ† (0. S −1 = γ 0 S † γ 0 and. For this purpose we consider first the positive energy solution. we demonstrate that the product Ψ† (p. where S is given by (10.288) corresponding to the wave function Ψ(p. is Ψ(p.292) . 10.f. (10. First. of (10.289) Here N− (p) is a constant which will be chosen to satisfy the normalization condition Ψ† (p. ±). For the positive energy solution. 10. ±). through χo given by χo = u1 u2 = u ∈ C2 .287) will be justified now.241). Ψ† (p.259). u† u = |u1 |2 + |u2 |2 = 1 .286) Here N+ (p) is a constant which will be chosen to satisfy the normalization condition Ψ† (p. hence.268). ±) Ψ† (0. (10.290) which differs from the normalization condition (10.287) the form of which will be justified further below. = −E(p) .. and let Ψ(0. according to (10. ±) Ψ† (0. using γ 0 as given in (10. according to (10.286) for p = 0 yields.h.. +) γ 0 Ψ† (p. +) = |N+ (0)|2 (u† . ±) γ 0 S −1 γ 0 γ 0 S Ψ(0. (10. S † = γ 0 S −1 γ 0 .290). ±) = S Ψ(0. (10. ±)γ 0 Ψ(p.e. We want to demonstrate now that the signs on the r.h. the l. we present the negative energy solution. Employing (10. of (10.s. ±) = = = Ψ† (0. ±) γ 0 Ψ(p. +) = −1 .285) The corresponding free Dirac particle is then described through the wave function Ψ(p. ±) denote the solution of a free particle moving with momentum p in the laboratory frame.s.s. +) γ 0 Ψ(0.291) Note that we have used that. (10. Similarly. The connection between the solutions. (10. ±) denote the corresponding solution of a particle in its rest frame.287. +|xµ ) = N+ (p) u σ·p E(p) + m u ei(p·r − t) .

f. of (10. −) γ 0 Ψ(0.s.300) (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 (10.289) yields Ψ† (0.6: Show that the normalization condition 2 N+ (p) (u∗ )T . We consider first the positive energy solution.286). Condition (10.s. σ·p u∗ E(p) + m T γo u σ·p E(p) + m u = 1 (10.298) This result completes the expression for the wave function (10.295) Evaluating the l.9: Solutions of the Free Particle Dirac Equation The same calculation for the negative energy wave function as given in (10. We can also conclude from our derivation N± (0) = 1 . (5.293) Obviously. 325 (10.294) We want to determine now N± (p) for arbitrary p. (10. σ·p u∗ E(p) + m T u σ·p E(p) + m u = 1 (10. (10. 2m (10.h. this requires the choice of a negative side on the r.10.287) written explicitly using (10.286) is 2 N+ (p) (u∗ )T . u† ) γ 0 0 u = − |N− (0)|2 .299) ( m + E(p) )2 .302) . using γ 0 as given in (10. ( m + E(p) )2 − p 2 (10.285) results in 2 N+ (p) 1 − p2 (E(p) + m)2 = 1 (10. Exercise 7. 2 E(p) (10.h. −) = |N− (0)|2 (0.296) Replacing (σ · p)2 by p 2 [c.290) to assign a positive value to |N− (0)|2 .234)] and using the normalization of u in (10.301) yields the normalization coefficient N+ (p) = m + E(p) .219) yields 2 N+ (p) u† u − u† (σ · p)2 u (E(p) + m)2 = 1.298) becomes N+ (p) = m + E(p) .

289)..296) for the normalization constant N+ (p) of the positive energy solution. (10. This attribute is the so-called helicity. λ|xµ ) = λ E(p) Ψ(p. pj ] = 0 .275) yields Ho Ψ(p.300) have been constructed to satisfy the free particle Dirac equation (10. (u∗ )T γ o = −1 (10.e.307) we investigate first the observable due to the simpler operator σ · p. λ|xµ ) = pµ Ψ(p. The property [σ · p. This degree of freedom 1 describes a spin– 2 attribute. however..326 Relativistic Quantum Mechanics We consider now the negative energy solution.307) where pµ = ( ..303) E(p) + m u Evaluating the l. −p). We can. 3 is fairly obvious. the wave functions constructed represent eigenstates of Ho .275). not a constant vector.286. λ|xµ ) in 4-vector notation.289) are not completely specified. the two components of u indicate another degree of freedom which needs to be defined.289.s.304) This condition is. The corresponding operator which measures this observable is 1 p ˆ Λ = σ· .290) written explicitly using (10. i∂µ Ψ(p. The wave functions are also eigenstates of the momentum operator i∂µ . exp[i(p · r − t)] = exp(ipµ xµ ). Rather than considering the ˆ observable (10. Helicity The free Dirac particle wave functions (10. Inserting (10.e. conclude N− (p) = m + E(p) 2m (10. i.286.306) i.305) and. thereby.e. λ|xµ ) (10. 10. Condition (10.233) 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. (10. 10.289) is T −σ·p −σ · p 2 E(p) + m u N− (p) u∗ . The commutation property [σ · p.h. 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. hence. i.308) 2 |p| ˆ Note that p represents here an operator. The wave functions (10. 2.286) into (10. have completed the determination for the wave function (10. identical to the condition (10. (10. yields 2 N− (p) u† (σ · p)2 u − u† u (E(p) + m)2 = −1 . λ|xµ ) . defined as the component of the particle spin along the direction of motion. This can be verified expressing the space–time factor of Ψ(p.309) . 10. j = 1. Ho ] = 0 follows from (10.

can ˆ ˆ be simultaneously diagonal. p3 ).308) above.310) We have shown altogether that the operators p.286) are eigenfunctions of Λ as well will specify now the vectors u. +. 2 the wave functions which are eigenstates of the helicity operator. in principle. = −E(p). According to the definition (5. We consider first the simplest case that particles move along the x3 –direction. States which are simultaneously eigenvectors of these three operators are also simulteneously eigenvectors of the three operators p.e. 1 p = (0. ˆ The condition that the wave functions (10. Np = m + Ep . 0. 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 . + 1 |r.. 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..e.312). 0)T and (0. We assume first particles with positive energy. 327 0 (σ · p)2 ˆ 2 (σ · p) ˆ 0 0 (σ · p)2 ˆ 2 (σ · p) ˆ 0 (10. +. 2m (10. is independent of the direction of motion of the particle. . hence. i. i. − 1 |r. Ho and σ·p commute with each other and. Therefore. = +E(p).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 = 0.224) 1 of σ3 the two u vectors (1.e. t) = Np  e   1 0   0 p  m + Ep  i(px3 + E t) 1 p e Ψ(pˆ3 . + 2 |r. Ho and Λ defined in (10. −. t) = Np  e (10. are   1   i(px3 − E t) 0 p e Ψ(pˆ3 .313) 2   0 1 where Ep and Np are defined in (10. t) = Np  e (10.312) We assume now particles with negative energy.311) 2  −p  0 m + Ep 1 where e3 denotes the unit vector in the x3 -direction and where ˆ Ep = m2 + p2 .10. i.. 1)T are eigenstates of 2 σ 3 with eigenvalues ± 1 . t) = Np  e 2   1 p m + Ep 0   0   i(px3 − E t) 1 p e Ψ(pˆ3 . In this case Λ = 2 σ 3 .

226. 10.262) for the bispinor wave function and the transformation (10. p) e |p| (10.317) Ψ(p. 2  1  0  (10. 1 |t) = √2   e−imt . + 2 |r. − 2 |r. 5. 10.123).312) can be obtained also by means of the Lorentz transformation (10. − 1 |r. t) 2 = Np u(p.311. in which case the transformation is given by (5. (10.312).319) 1 Ψ(p. denoted by ˜. t) 2 = Np 1 u(p. + 1 ) 2 p 1 m + Ep u(p. γ and are  1  0  1 ˜ Ψ(p = 0. 1)T have to be replaced by eigenstates u± (p) of the spin operator along the direction of p.222. a wave function which represents free particles at rest.. −.e.311. + 1 |r.220).223) as follows 1 u(p. β. Generating Solutions Through Lorentz Transformation The solutions (10. + 2 ) u(p.316) describes a rotation which aligns the x3 –axis with the direction of p.318) 1 Ψ(p. 0)T and (0. i. +. The corresponding wave functions are determined through ˜ i˜ 0 ∂t − m Ψ(t) = 0 .320) where Ep and N are again given by (10. t) = Np ei(p·r + Ep t) (10. + ) 2 1 u(p. γ . These eigenstates are obtained through a rotational transformation (5.229).] The corresponding free particle wave functions are then Ψ(p.315) e3 × p ˆ ∠(ˆ3 . t) = N√ ei(p·r + Ep t) (10. for an r–independent wave function. 10. − 1 ) 2 −p 1 m + Ep u(p. − 2 ) u(p. [One can also express the rotation through Euler angles α. For this purpose one starts from the solutions of the Dirac equation in the chiral representation (10.313) except that the states (1.328 Relativistic Quantum Mechanics To obtain free particle wave functions for arbitrary directions of p one can employ the wave functions (10. + 2 ) ei(p·r − Ep t) (10. +. + 2 ) p 1 m + Ep u(p. − 2 ) −p 1 m + Ep u(p.314) (10. −. +. − ) 2 where ϑ(p) = = = exp exp i − ϑ(p) · σ 2 i − ϑ(p) · σ 2 1 0 0 1 . − 1 ) 2 ei(p·r − Ep t) (10.321) .

9: Solutions of the Free Particle Dirac Equation  0  1  1 ˜ Ψ(p = 0. For the resulting wave functions in the chiral representation one can use then a notation corresponding to that adopted in (10. yields for the exponential space–time dependence according to (10. +. 2 |t) = √2  .325) One should note that φo .174. ± 1 of the wave functions stated in (10.262) φo χo → e 2 w3 σ 0 1 3 Et ) (10.311.228) yields the corresponding solutions (10. − 1 |t) = √2  .. Applying (10. 10. +. 2 The solutions (10. for w = (0.311)  e 1 w 2 3 ˜ Ψ(p(w3 )ˆ3 . + 1 |r. − 1 |t) = √2   e−imt .313) in the p → 0 limit.325) to (10.322) can be written in spinor form 1 √ 2 φo χo e imt . −. χo are eigenstates of σ 3 with eigenvalues ±1. −. i. 2  0  1   1  0  +imt 1 1 ˜  Ψ(p = 0.10.322) The reader can readily verify that transformation of these solutions to the Dirac representationsas defined in (10. 0 1 (10. (10. +. 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 . φo .176) imt → i ( p3 x3 and for the bispinor part according to (10. t) e 2 e− 2 w3 1  = √  1 2  e 2 w3  i(px3 − E t) p e  . χo ∈ 1 0 . 10.322). 2  0 e −1  329 (10. 10.324.323) Transformation (10. − 1 |r.262) for a boost in the x3 –direction.323) should yield the solutions for non-vanishing momentum p in the x3 –direction. w3 ).324) 0 e − 1 w3 σ 3 2 φo χo = e 2 w3 σ φo 1 3 e− 2 w3 σ χo 1 3 .e.  −1  e 0   0  1  +imt 1 ˜  Ψ(p = 0. This correspondence justifies the characterization ±.

329) . 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  =    i(px3 + E t) p e  (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. − 2 |r. −. −. + 2 |r.326) where according to (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. +. − 1 |r. + 2 |r. t) e 2  =     i(px3 + E t) p e   1 Ψ(p(w3 )ˆ3 . t) e 2 e 1   = √  1 2 − e 2 w3  i(px3 + E t) p e  (10.328) the relationship (10.327) Employing the hyperbolic function properties cosh x = 2 coshx + 1 . t) e 1  = √  2   −e− 2 w3 − 1 w3 2 1  i(px3 + E t) p e   ˜ Ψ(p(w3 )ˆ3 .61) p(w3 ) = m sinhw3 . 2 (10. −. −.61) between the parameter w3 and boost velocity v3 .330  e 1 w 2 3 Relativistic Quantum Mechanics 1 0 1 0 0 1 0 1  1 ˜ Ψ(p(w3 )ˆ3 . 2 sinh x = 2 coshx − 1 . and the expression (10. Transformation to the Dirac representation by means of (10. + 1 |r. +.

221) using (10.243) and.331) is a solution as well. we will derive this equation only for infinitesimal transformations. when we generated the solutions Ψ(p. The reason why we provide another derivation of (10. = γν . The transformation is Ψ (xµ ) = S(Lη ξ ) ρ(Lη ξ ) Ψ(xµ ) (10.168–10.10. The change of sign for the latter solutions had to be expected as it was already noted for the negative energy solutions of the Klein–Gordon equation (10. 10. Making this expectation a postulate allows one to derive the condition (10.9: Invariance of Dirac Equation Revisited and similarly sinh w3 = 2 Ep − m = 2m p 2m (Ep + m) 331 (10.313) for −p. the proper transformation S(Lη ξ ). Actually. thereby. expresses the system in the old coordinates.311) as well as the negative energy solutions (10. refered to as the active transformation.243) considered solely the limit of infinitesimal transformations anyway. Λ|t). In the new derivation we consider the particle described by the wave function transformed. λ.123) above and characterized there. Λ|xµ ) from the solutions describing particles at rest Ψ(p = 0. indeed. Such description will be essential for the formal description of Lorentz invariant wave equations for arbitray spin further below.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. that if Ψ(xµ ) is a solution of the Dirac equation that Ψ (xµ ) as given in (10.332) Here we have made use of the fact that S(Lη ξ ) commutes with ∂µ and ρ(Lη ξ ) commutes with γ µ .331) i S(Lη ξ )γ µ S −1 (Lη ξ ) ρ(Lη ξ )∂µ ρ−1 (Lη ξ ) − m Ψ (xµ ) = 0 (10.334) (10. is sufficient since (1) it must hold then for any finite transformation. reproduces the positive energy wave functions (10. Such transformation had been applied by us. To show this we rewrite the Dirac equation (10. We expect. and since (2) the calculations following (10. in general. the essential property stating the Lorentz–invariance of the Dirac equation.333) .330) into (10. λ. This transformation. of course. but not the observer. ρ(Lη ξ ) has been defined in (10. (10.329. 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 . Invariance of Dirac Equation Revisited At this point we like to provide a variation of the derivation of (10.330) Inserting expressions (10. which however.327).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 ).176).243).

0). ∂µ ] = [x0 ∂2 + x2 ∂0 . ∂µ ] = (K )ν µ ∂ν . 0. [K . 0. 0).342) One can readily convince oneself that these results are consistent with (10.331) is again a solution of the Dirac equation. i. it is met by S(Lη ξ ) as given in the chiral representation by (10. . ∂µ ] = [x0 ∂1 + x1 ∂0 . therefore. 0).243) and. . . The second condition is identical to (10. 0. Inspection of (10.338) [J3 . ∂µ ] = (10. w = (0. 0).336).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 = (1.341) [K3 . ∂µ ] = (10.337) [J2 .334) for infinitesimal Lorentz transformations Lη ξ . ∂µ ] = (10..332 Relativistic Quantum Mechanics We will demonstrate now that the first condition is satisfied by ρ(Lη ξ ).336) We will proceed with this task considering all six cases: [J1 . ∂µ ] = (J )ν µ ∂ν . ∂µ ] = [x3 ∂2 − x2 ∂3 . ∂µ ] = (10. ϑ = (0.335) shows that we need to demonstrate [J .340) [K2 . ∂µ ] = (10. ∂µ ] = (10. ϑ = (0.339) [K1 . the Dirac equation is invariant under active Lorentz transformations. As mentioned already we will show condition (10. 0). ∂µ ] = [x1 ∂3 − x3 ∂1 . 1. that any solution Ψ(xµ ) transformed according to (10. 0.e. . w = (0. We have demonstrated. 0. of course. 1) .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.     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. .262). ∂µ ] = [x2 ∂1 − x1 ∂2 . ∂µ ] = [x0 ∂3 + x3 ∂0 . w = (0. We will proceed by employing the generators (10.

350) (10. For this purpose we choose the decomposition Ψ(xµ ) = φ(xµ ) χ(xµ ) .348) (10.344) Using the notation σ = (σ1 .345) in the so-called non-relativistic limit in which all energies are much smaller than m..232).221) reads then [ iγ µ (∂µ + iqAµ ) − m ] Ψ(xν ) = 0 (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 10.351) X (x ) µ .10: Dirac Particles in Electromagnetic Field 333 10.346) ˆ ˆ ˆ α · π + qV + β m Ψ(xµ ) (10. The Dirac equation in the Schr¨dinger form reads then o i∂t Ψ(xµ ) = ˆ ˆ where α and β are defined in (10.10 Dirac Particles in Electromagnetic Field We like to provide now a description for particles governed by the Dirac equation which includes the coupling to an electromagnetic field in the minimum coupling description. accordingly.6 we assume that the field is described through the 4-vector potential Aµ and.10.g. we define φ(xµ ) = e−imt Φ(xµ ) χ(x ) = e µ −imt (10. for the scalar field V in (10. e.1) i∂t − qV and p by ˆ ˆ π = p − qA . (10. Equivalently. Non-Relativistic Limit We want to consider now the Dirac equation (10.343) One may also include the electromagnetic field in the Dirac equation given in the Schr¨dinger form o ˆ (10. σ2 .6 above).1 in Sect 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 . 10.349) We want to focus on the stationary positive energy solution.345) holds |qV | << m . (10.233) by replacing i∂t by (see Table 10. we replace the operator ∂µ by ∂µ + iqAµ . The Dirac equation (10.345) (10. This solution exhibits a timedependence exp[−i(m + )t] where for holds in the non-relativistic limit | | << m. Accordingly.347) (10. Following the respective procedure developed for the Klein-Gordon equation in Sect.

Φ Using (10. does not need to be considered anymore. One obtains [πj .353) by X ≈ to obtain a closed equation for Φ ˆ σ·π 2m 2 (10. Ak ] ) f = ( ∂j Ak − Ak ∂j + Aj ∂k − ∂k Aj ) f . and.7. derived in Sect. which in the present case states ˆ ˆ2 ˆ ˆ (σ · π)2 = π + i σ · (π × π) .334 and assume that for the time-derivative of Φ and X holds ∂t Φ << m . ∂k + qAk ] i i 1 1 1 1 [ ∂j .348.352) allow one to approximate (10. Ak ] . 10.346.360) For an arbitrary function f (r) holds ( [Aj . ∂k ] + q [Aj . henceforth. Ak ] + q 2 [Aj . ∂k ] + [∂j .358) ( πj πk − πk πj ) = jk [πj . due to the m−1 factor. ˆ Equation (10. (10.353) (10.349) yields i∂t Φ i∂t X = = ∂t X X Relativistic Quantum Mechanics << m . πk ] = = 1 1 [ ∂j + qAj . For this purpose we employ the identity (5. Ak ] i i i i =0 =0 = q q [Aj . π ] .351) in (10. (10.356) i∂t Φ ≈ Φ + qV Φ .355) ˆ σ·π Φ 2m (10. ∂k ] + [∂j .357) for Φ can be reformulated by expansion of (σ · π)2 . ˆ ˆ For the components of π × π holds ˆ ˆ π×π = jk (10. 5.356).352) ˆ σ · π X + qV Φ ˆ σ · π Φ + qV X − 2mX . ∂k ] + q [ ∂j .230).357) Equation (10.350. i i (10. (10. (10. accordingly.354) The properties (10.359) We want to evaluate the latter commutator. 10. (10.361) . 10. one can replace X in (10. identifies X as the small component of the bi-spinor wave function which.354) ˆ 0 ≈ σ · π Φ − 2m X .

t) + q V (r.366) (10.365) Comparision of this expression with (5.222..364) which is referred to as the Pauli equation.223) shows that the propagator in (10. r (10.10. Equations (10. 0) .343) for the vector potential Aµ = (− Ze2 . does not leave the theory again when one takes the non-relativistic limit. Ak ] ) f = [ ((∂j Ak )) − ((∂k Aj )) ] f (10.e. using (10. Comparision of (10.359. 0. The formal solution of this equation is Φ(t) = e−iqtB·σ Φ(0) . In other words.2) it introduces the o extra term qσ · B Φ which describes the well-known interaction of a spin. [πj . 10.363) where we employed B(r. the interaction qσ · B induces a 1 precession of the spin.Spectrum We want to describe now the spectrum of a relativistic electron (q = −e) in the Coulomb field of a nucleus with charge Ze.360) and Aµ = (V. while agreeing in all other respects with the non-relativistic Schr¨dinger equation (10.1 which emerged in the Lorentz-invariant theory as an algebraic 2 necessity. (10.10: 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 ((· · ·)). 10. t) [see (8.1 particle with a magnetic 2 field B. 0.358. i. t) (10. πk ] = q q (( ∂j Ak − ∂k Aj )) = − i i ×A jk q = − B i jk (10. Let us consider briefly the consequences of the interaction of a spin. one obtains ( [Aj . the spin. t)]2 q − σ · B(r. t) = × A(r. 10.2) governing a one-dimensional wave function ψ ∈ C. For 2 this purpose we disregard the spatial degrees of freedom and assume the Schr¨dinger equation o i∂t Φ(t) = q σ · B Φ(t) .362) or. Dirac Particle in Coulomb Field . The respective bispinor wave function Ψ(xµ ) ∈ C4 is described as the stationary solution of the Dirac equation (10. −A).366) can be interpreted as a rotation around the field B by an angle qtB.344. but rather remains as a steady “guest” of non-relativistic physics with the proper interaction term. o reveals a stunning feature: the Pauli equation does justice to its two-dimensional character.364) governing a two-dimensional wave function Φ ∈ C2 with the corresponding non-relativistic Schr¨dinger equation (10.6)]. t) ≈ ˆ [p − q A(r.367) .363) allow us to write (10. 5.357) in the final form i∂t Φ(r.1 with the magnetic field. ∂k ] + [∂j . t) 2m 2m Φ(r.2 around the magnetic field.

369). µ = ν and altering ‘dummy’ summation ˜ ˜ ˜ ˜ .372) (10.231).229).369) from the left by γ ν πν + m.371) such that we can conclude that (10. ˜ (10. γ ˆ ˜ ˆ (10. 3 µ=0 ˆ2 (˜ µ )2 πµ = π0 − π .230) γ µ γ ν = − γ ν γ µ . For this purpose we ‘square’ the Dirac equation.370) is obtained then ˜ [ i˜ µ (∂µ + iqAµ ) + m ] Ψ(xµ ) γ is a solution of (10. indeed.368) ˜ ( γ µ πµ − m ) Ψ(xµ ) = 0 .ν=1 µ =ν γ µ γ ν πµ πν .375) Following the algebra that connected Eqs. once a solution Ψ(xµ ) of (10. Employing πµ as defined ˜ in Table 10.1 one can write (10. (5.228) and in (10. is a solution of (10.374) The first term on the r. i.368) ˜ where Ψ(xµ ) and γ µ are defined in (10. The difference arises from the term γ µ πµ γ ν πν in (10. 3. 5. (˜ 0 )2 = 1 and (˜ j )2 = γ 1 γ −1 j = 1.7 one can write the second term in (10. This follows from [ i˜ µ (∂µ + iqAµ ) + m ] [ i˜ µ (∂µ + iqAµ ) − m ] γ γ = [ i˜ µ (∂µ + iqAµ ) − m ] [ i˜ µ (∂µ + iqAµ ) + m ] γ γ according to which follows from (10.h.370) Any solution of (10.370) ˜ [ i˜ µ (∂µ + iqAµ ) − m ] [ i˜ µ (∂µ + iqAµ ) + m ] Ψ(xµ ) = 0 γ γ (10.e.232) in Sect. we solve ˜ [ i˜ µ (∂µ + iqAµ ) − m ] Ψ(xµ ) = 0 γ (10. multiplying (10.371).368) is also a solution of (10. 2. but the converse is not necessarily true.370) for which holds ˜ ˆ ˜ ˆ 3 γ πµ γ πν = ˜ ˆ ˜ ˆ µ=0 µ ν (˜ µ )2 πµ + γ ˆ2 µ. ˜ ˜ ˆ ˆ (10. according to (10.6.s.370). 10. 1.369). This yields ˜ [ i˜ µ (∂µ + iqAµ ) + m ] [ i˜ µ (∂µ + iqAµ ) − m ] Ψ(xµ ) γ γ ˜ = (˜ µ πµ γ ν πν − m2 ) Ψ(xµ ) = 0 .373) (10. Equation (10.374). (5.180).370) resembles closely the Klein-Gordon equation (10.336 Relativistic Quantum Mechanics For the purpose of the solution we assume the chiral representation. ˜ However. γ ˆ2 ˆ2 (10. noting from (10.230). can be rewritten using. respectively. but differs from it in an essential way.369) For our solution we will adopt presently a strategy which follows closely that for the spectrum of pionic atoms in Sect.

381) According to (10. γ 0 ] [ˆj . one can summarize (10.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.344). πν ] γ ˜ π ˆ µ. i..ν=1 µ =ν = [˜ µ .382) . γ ν ] [ˆµ . −i∂j ] = − [ (∂t + iqA0 ) ∂j − ∂j (∂t + iqA0 ) ] f i ((∂j qA0 )) f (10. γ j ] [ˆ0 .ν=1 µ =ν (10. due to A = 0.e. ν = 1. πj ] in (10.ν=1 µ =ν 3 j=1 3 3 = 1 4 1 2 1 [˜ .10. πj ] π ˆ = = (−i∂t + qA0 .229). Since [ˆµ . πj ] + γ ˜ π ˆ 4 0 j [˜ j .376–10. πj ] .377) which follows readily from the definition (10. γ ν ] [ˆµ . 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 337 1 2 ( γ µ γ ν πµ πν + γ ν γ µ πν πµ ) ˜ ˜ ˆ ˆ ˜ ˜ ˆ ˆ µ. γ ] [ˆ0 .379) The commutators [ˆ0 . the commutators [˜ 0 .378) can be evaluated using (10.10: Dirac Particles in Electromagnetic Field indices. 2. Altogether. γ µ γ ν πµ πν = ˜ ˜ ˆ ˆ µ.380) where f = f (r.378) According to the definition (10. r2 (10.344) π ˆ [ˆ0 . γ ˜ π ˆ j=1 (10.367) of Aµ .ν=1 µ =ν σ 0 0 −σ · (( qA0 )) (10.380) γ µ γ ν πµ πν = i ˜ ˜ ˆ ˆ µ. 3 (10. γ j ] γ ˜ = = 2 0 1 1 1 0 1 σj 0 0 −σ j 0 σj −σ j 0 − 0 σj −σ j 0 0 1 1 1 0 1 (10.367) and the definition (10.367) holds qA0 = r ˆ Ze2 . πν ] γ ˜ π ˆ µ. γ j ] are γ ˜ [˜ 0 .376) This expression can be simplified due to the special form (10. πν ] = 0 π ˆ for µ. it holds 1 4 [˜ µ .

e.5 which describe the coupling of orbital angular momentum and spin 1 {( Yjm (j − 2 . hence. couples the orbital angular momentum of the electron to its spin. 6. m values are coupled.383) We seek stationary solutions of this equation. .389) According to the results in Sect. a solution of the form ∼ Y m (ˆ) can be obtained.385) 1 We split the wave function into two spin. . . in r ˆ fact..368) reads − Ze2 ∂t − i r 2 + 2 + i σ·r ˆ 0 0 −σ · r ˆ Ze2 − m2 r2 Ψ(xµ ) = 0 (10.384) into (10. (10. Combining this result with (10.5 the operator iσ · r is block-diagonal in this basis such that only ˆ 1 the states for identical j. (10. Accordingly. 1 |ˆ). Yjm (j + r 2 1 1 .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 .6.148). (10.189) for the Laplacian and expansion of the term (· · ·)2 result in the twodimensional equation 2 ∂r − ˆ L2 − Z 2 e4 iσ · r Ze2 ˆ 2Ze2 + r2 r + 2 − m2 r φ± (r) = 0 .1 .388) Except for the term iσ · r this equation is identical to that posed by the one-dimensional Kleinˆ Gordon equation for pionic atoms (10. −j + 1. m = −j. 1 |ˆ).387) The expression (10.381). (10. In the latter case. The term iσ · r. only the two states {Yjm (j − 2 .191) solved in Sect.147. however. r r 2 2 2 j = 1.375) the ˆ ‘squared’ Dirac equation (10. Insertion of (10. 2 |ˆ)} as given in (6.384) can be interpreted as the energy of the stationary state and.2 components ˜ Ψ(r) = and obtain for the separate components φ± (r) Ze2 + r 2 ˜ φ+ (r) ˜ φ− (r) (10. 1 |ˆ) ) .374). We note that these states are also eigenstates of the angular r 2 . . + j } (10. Yjm (j + 1 . it is this quantity that we want to determine. 6.388) in terms of states introduced in Sect.338 Relativistic Quantum Mechanics where r = r/|r| is a unit vector. (10. 10. Such solutions are of the form ˜ ˜ Ψ(xµ ) = Φ(r) e−i t . . we express 2 the solution of (10. 2 3 2 . 6. . (10.386) + 2 ± iσ · r ˆ Ze2 − m2 r2 φ± (r) = 0 . is genuinely two-dimensional and. i.

396) and λ2 (j) [λ2 (j) + 1] (10. in the present treatment. .199).396) and (10. Any similarity transformation leaves the first term in (10. λ1 (j) and λ2 (j). . . rather than = 0. (6. as in the case of pionic atoms. the eigenvalues are independent of m. not the wave functions.395. 1 |ˆ) = − Yjm (j ˆ r 2 2 1 2 h± (r) g± (r) Yjm (j − 1 . 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 .390) (10. .200).151].393) but do not explicitly consider further the wavefunctions. (6. and except for the fact that 2 2 we have two sets of values for λ1. conclude that the .395) (10.2 (j) as given by (10. the missing additive term − 1 .2 (j) + 1) 2 Ze2 + + r2 r 2 − m2 f1. 1. namely. 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. 3 . property (6. Since. we want to determine solely the spectrum. the values of the argument of λ1. 1 |ˆ) r 2 2 2 2 r r . therefore.392) h± (r) g± (r) = 0. Obviously.. − 1 r2 (j − 1 )(j + 1 ) − Z 2 e4 2 2 ±i Ze2 (j + )(j + ) − Z 2 e4 We seek to bring (10. However. m and expand φ± (r) = Using σ · r Yjm (j ± 1 . involving the 2 × 2 unit matrix.5 [c. 6. 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. which states that the states Yjm (j ± 1 .391) derived in Sect. We select. such transformation can be chosen as to diagonalize the second term. namely.f.397) This equation is identical to the Klein-Gordon equation for pionic atoms written in the form (10.392) reads then in the diagonal representation 2 ∂r − λ1.186)].2 (j).394) Equation (10.392) into diagonal form. hence.10: Dirac Particles in Electromagnetic Field 339 ˆ momentum operator L2 [cf. 10. except for the slight difference in the expression of λ1. 1 |ˆ) r 2 (10.151). 1 |ˆ) are r 2 2 ˆ eigenfunctions of L2 . .2 (j) being 2 j = 1 . a specific pair of total spin-orbital angular momentum quantum numbers j. We can.10.2 (j)[λ1. . unaltered. (10.2 (r) = 0 (10. 1 |ˆ) + r Yjm (j + 1 .392).

j = 1 . 2. . ( +1) − Z 2 e4 )2 (10. 1. . . . Expanding the energies in terms of this parameter allows one to identify is determined by Z the relationship between the energies 1 and 2 and the non-relativistic spectrum. counts the nodes of the wave function. One may also state 2 this in the following way: (10. Using (10. . Obviously.397) is again given by eq. . One can equate (10. m) 2 n = 1.401) if one attributes to the respective 1 states the angular momentum quantum numbers = j + 2 and = j − 1 . m = − . . m = −j. . . (10. − + 1. . . 10. 2 2 m = −j. . One obtains in case of (10. . These considerations 2 are summarized in the following equations 1 ED (n. . . . .398. (10. j = + 1 . = 1 + m Z 2 e4 √ ( n − − 1 + 2 − Z 2 e4 )2 .396) we obtain. . 1. 1.400) corresponds to a non-relativistic state with quantum numbers n.402) In this expression n is the so-called main quantum number. . 2.2 . 3. 2 2 1 ≈ m − (10. . . accordingly. . (10.402) with (10. . 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 .203). .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. . 3 . . 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 .399) n = 0. m = −j. 1. . −j + 1. . .h. the stationary states have binding energies E = − mZ 2 e4 n = 1. 2. j . albeit with some modifications. . (10. . and spin-orbital angular momentum j = + 1 . . In case of non-relativistic hydrogen-type 1 atoms. . .401) corresponds to a non-relativistic state 2 with quantum numbers n.398) 2 = 1 + m . (10. . . n − 1 . . . . .340 Relativistic Quantum Mechanics spectrum of (10. . . . . . of these equations describe the binding energy.404) = 0. and spin-orbital angular momentum j = − 1 . 1.403) √ Z 2 e4 2 ED (n. j . the second term on the r. . . j where 1 corresponds to λ1 (j) as given in (10. 1 ms = ± 2 (10. 2n2 = 0. including spin.395) and 2 corresponds to λ2 (j) as given in (10. It is given by n = n + + 1 where is the orbital angular momentum quantum number and n = 0. m) = 1 + m (n− + . 10. . n − 1 .395. The magnitude of the relativistic effect 2 2 e4 . −j + 1. . . −j + 1. .s.401) These expressions can be equated with the non-relativistic spectrum.396). 2.400) (10. j = − 2 . .400) and (10. . j = 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. (10.10. r 2 2 2 . (10. j. notation 1s 1 2 non-rel. ..551177 ⇑ .403. . .1. Degeneracies are denoted by ⇑.551176 spectr. is in the range of 10 eV.40151 ⇑ .036 by means of Eqs.e.551178 ⇑ .404) finally into the single formula ED (n. the nonrelativistic degeneracy for the six 2p states of the hydrogen atom: in the present.40146 ⇑ ⇑ -1. i. 10.2 the non-relativistic [cf. 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. much less than the rest mass of the electron (511 keV). 2 |ˆ) for m = ± 1 . j 1 2 1 2 1 2 3 2 1 2 1 2 3 2 3 2 5 2 341 rel.405) In order to demonstrate relativistic effects in the spectrum of the hydrogen atom we compare in Table 10.402. for example. 10. binding energy / eV Eq. = 0. relativistic. (10. .405)] spectrum of the hydrogen atom. The energies were evaluated with m = 511. case these six states are split into energetically different 2p 1 and 2p 3 states. This is clearly demonstrated by the comparision of non-relativistic and relativistic spectra of a hydrogen-type atom with Z = 100 in Table 10. . . 2.60601 -3. 0 0 1 1 0 1 1 2 2 spinorbital ang. binding energy / eV Eq.2: Binding energies for the hydrogen (Z = 1) atom. However.. The table entries demonstrate that the energies as given by the expression (10.402)] and the relativistic [cf.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 10.405) in terms of the non-relativistic quantum numbers n. m) = m 1 + Z 2 e4 (n−j −1 + 2 (j+ 1 )2 − Z 2 e4 )2 2 n = 1. ± 3 . (10. 1 |ˆ) for m = ± 1 .405 ) -13.3.405). 1.e. .60583 -3. (10.0041 keV and e2 = 1/137.1.402) -13. in fact.10: Dirac Particles in Electromagnetic Field main quantum number n 1 2 2 2 3 3 3 3 3 orbital angular mom. relate closely to the corresponding nonrelativistic states. the 2p 3 states with r 2 2 2 2 1 involve four degenerate states corresponding to Y 3 m (1. . The reason is that the mean kinetic energy of the electron in the hydrogen atom. . −j + 1. One can combine the expressions (10. Of particular interest is the effect of spin-orbit coupling which removes. i.40147 -1. j (10.3. the non-relativistic and relativistic energies are hardly discernible. . n − 1 j = 1 2 1 2 ± m = −j. . mom.

m) ≈ m − mZ 2 e4 mZ 4 e8 − 2n2 2n3 1 j+ − 3 4n + O(Z 6 e12 ) .15. Degeneracies are denoted by ⇑.036 by means of Eqs.1 ⇑ .3: Binding energies for the hydrogen-type (Z = 100) atom.405) reads 1 (10.35.402) -136. 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. 1 2 2 2 2 E (2p 3 ) − E (2p 1 ) ≈ − 2 2 mZ 4 e8 2 · 23 1 −1 2 = mZ 4 e8 .15. 0.342 main quantum number n 1 2 2 2 3 3 3 3 3 orbital angular mom.0 ⇑ ⇑ -15.6 -42. Tables 10.405) to order O(Z 4 e8 ). (10.408) 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).206) provides in the present case ED (n.407) 1 2 This expression allows one.405 ) -161.3).0041 keV and e2 = 1/137.409) where V (r) is spherically symmetric.402. We assume for the wave function the stationary state . The energies were evaluated with m = 511. 0. .2. (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. binding energy / keV Eq.406) α √ 1+ (n − β + β 2 −α)2 The expansion (10. An example for such potential is the Coulomb potential V (r) = −Ze2 /r considered further below. for example. mom. (10.1 ⇑ ⇑ ⇑ ⇑ rel. In order to investigate further the deviation between relativistic and non-relativistic spectra of hydrogen-type atoms we expand the expression (10.3 spectr.2 -17.10. It holds for n = 2 and j = 3 .9 ⇑ . Introducing α = Z 2 e4 1 and β = j + 2 (10. to estimate the difference between the energies of the states 2p 3 and 2p 1 (cf. binding energy / keV Eq. 0) (10. (10.8 ⇑ . 32 (10.405). j.1 -34. 0 0 1 1 0 1 1 2 2 spinorbital ang. 10.

198)] not couple states with different j. the term is block-diagonal in the space spanned by the states Yjm (j ± 1 . 168]: the term is a scalar (rank zero tensor) in the space of the spin-angular momentum states Yjm (j ± 1 . i.417)  c(r)  d(r) X (r) 1 1 1 1 Yjm (j + 2 .412) ˆ In this equation a coupling between the wave functions Φ(r) and X (r) arises due to the term σ · p. 6. 1 |ˆ) r 2 2 = i ∂r + j+ r 1 2 3 2 f (r) Yjm (j − 1 .345). 1 |ˆ) introduced in r 2 2 1 Sect. The Dirac equation reads then. 1 |ˆ) r 2 1 1 = ∂r r r r (10.416) The differential equations (10.2 degrees of freedom. 1 |ˆ) . 6. m-values.415) 1 ˆ σ · p rg(r) Yjm (j − 2 . i.414) These equations can be brought into a more symmetric form using ∂r + which allows one to write (10. 10.232. in particular. 1 |ˆ) r 2 2 = i j+ ∂r + r j+ r 1 2 1 r f (r) Yjm (j − 2 . 10..   =  r (10. 1 |ˆ) and Yjm (j − 1 . 2 |ˆ) + r Yjm (j − 2 .411) (10.197. X (r) ∈ C2 describe the spatial and spin.10.411.410) 1 where Φ(r). pp.414) ˆ σ · p rf (r) Yjm (j + 1 . 10. 1 |ˆ) r 2 = i ∂r − 1 2 r g(r) Yjm (j + 1 . 1 |ˆ).413) ˆ σ · p g(r) Yjm (j − 1 . but are timeindependent.10: Dirac Particles in Electromagnetic Field form Ψ(xµ ) = e−i t 343 Φ(r) X (r) . 6. (6.5 [see.e. ˆ σ · p X + m Φ + V (r) Φ ˆ σ · p Φ − mX + V (r) X = = Φ X (10. 2 |ˆ) + r Yjm (j − 2 . This term has been discussed in detail in Sect. 2 |ˆ) r r r . r r 2 2 2     a(r) b(r) 1 1 1 1 Yjm (j + 2 . r 2 (10. 1 |ˆ) r 2 2 = i ∂r + −j r 1 g(r) Yjm (j + 2 .5.f. σ · p ˆ σ · p f (r) Yjm (j + 1 . 1 |ˆ) r 2 2 (10. r 2 2 (10. The arguments above allow one to eliminate the angular dependence by expanding Φ(r) and X (r) 1 in terms of Yjm (j + 2 .e.412) are four-dimensional with r-dependent wave functions. 2 |ˆ)  r Φ(r)  r  . (10.. 2 |ˆ) and does r 2 ˆ has odd parity and it holds [c. 1 |ˆ) . according to (10.413.

c(r) are coupled. Inserting (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) = 0.157). m pair contribute. According to (10.420)   g1 (r) X (r) 1 1 − Yjm (j − 2 . using (10.418) holds for f1 (r). 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. 1 |ˆ)  r Φ(r) 2 2  r    =  (10. (10. 2 |ˆ) r r   f2 (r) 1 1 r Φ(r)  i r Yjm (j − 2 . there exist two independent solutions (10. the orthonormality property (6. 2 |ˆ)   =  (10.417) of the form     f1 (r) i Yjm (j + 1 . 10. d(r) are coupled and b(r).415.422) 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 (10.423) . m values. 2 |ˆ) r r where the factors i and −1 have been introduced for convenience.411. g1 (r) ∂r − j+ r j+ r 1 2 g1 (r) + [ − m − V (r) ] f1 (r) = 0 ∂r + and for f2 (r). such expansion must include states with all possible j. 10. Accordingly. we consider the case that only states for one specific j.419) Obviously.421)   g2 (r) X (r) 1 1 − Yjm (j + 2 . Presently. only a(r).344 Relativistic Quantum Mechanics In general. g2 (r) ∂r + 1 2 f1 (r) − [ + m − V (r) ] g1 (r) = 0 (10.417) into (10.416).412).

3s 1 .424) Ψ(xµ ) ≈  r 0 1 i. i. According to the discussion of the spectrum (10. etc. According to this procedure. Φ in (10. in the non-relativistic limit wave functions   f2 (r) 1 1 i Yjm (j − 2 . The solution of (10.420) is the large component and X is the small component.422). 3p 1 . 1. We note that (10.405). . Behaviour at r → 0 We consider first the behaviour of the solutions f1 (r) and g1 (r) of (10.423).420). correspodingly the states We consider first the solution of (10.425) r 0 covers states with angular momentum 1s 1 . .e.e. to states with angular momentum = j + 2 . together with the appropriate boundary conditions at r = 0 and r → ∞. 2s 1 .426) . determining wave functions of the type (10. 3d 5 .422). etc. and obtains finally a polynomial function p(r) such that rγ exp(−µr)p(r) solves (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. Hence. (10. Equation (10. the ones given in (10. . the radial wave functions for Dirac particles in the potential (10. = 0.422) describes the states 2p 1 . for small r. 10.367) which is spherically symmetric such that equations (10. In the non-relativistic limit.422).422) corresponds to states   f1 (r) Yjm (j + 1 . n − 1. Similarly. namely.423) are identical. (10. .10: Dirac Particles in Electromagnetic Field 345 Equations (10. 2 2 2 2 2 2 = j − 1 2 and.422) determines solutions of the form (10. 1 |ˆ)  r i 2 2 . 3d 3 . (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.423) apply for V (r) = −Ze2 /r. (10.421). . one demonstrates first that the wave function at r → 0 behaves as rγ for some suitable γ. . 2p 3 . . 2 2 2 (10. the 2 equations determine.. 3p 3 .409). . 2 |ˆ)  r Ψ(xµ ) ≈  . We assume the 4-vector potential of pure Coulomb type (10.422) near r = 0.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. 2.405). one demonstrates then that the wave functions for r → ∞ behaves as exp(−µr) for some suitable µ.10. Dirac Particle in Coulomb Field . (10.422) and (10. except for the opposite sign of the term (j + 1 ).405) of the relativistic hydrogen atom the corresponding states have quantum numbers n = 1. enforcing the polynomial to be of finite order leads to discrete eigenvalues .422.

427) makes only the positive solution 2 possible. assume the r-dependence in (10.346 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 .432) = = − ( − m. in case j + 1 )2 < Ze2 . hence. hence.433) = = ( m2 − 2 ) g1 (r) ( m2 − 2 ) f1 (r) (10.434) √ The solutions of these equations are f1 . determined that the solutions f1 (r) and g1 (r).422) becomes ∂r g1 (r) ∂r f1 (r) Iterating this equation once yields 2 ∂r g1 (r) 2 ∂r f1 (r) 1 . g1 (r) →∞ ∼ e−µr . (10. (10.430) 2 Note that the exponent in (10. 10. One obtains γ = ± (j + 1 )2 − Z 2 e4 . Such r-dependence 2 would make the expectation value of the potential r2 dr ρ(r) 1 r (10.427) with 1 γ = (j + )2 − Z 2 e4 .266. for small r.420). g1 ∼ exp(± m2 − 2 r). g1 (r) →0 ∼ b rγ (10. we conclude f1 (r) →∞ ∼ e−µr . r2 (10. ) f1 (r) ( + m ) g1 (r) (10. becomes imaginary. We have.428) = 0. according to (10. Only the exponentially decaying solution is admissable and.427). µ = m2 − 2 (10.267. for the particle density holds then ρ(r) ∼ |rγ−1 |2 = Behaviour at r → ∞ For very large r values (10.435) .431) infinite since. (10. 10.429) This equation poses an eigenvalue problem (eigenvalue −γ) for proper γ values. The assumed r-dependence in (10.427) = = 0 0.

10. (10. of both (10. we set f1 (r) g1 (r) = = √ √ ˜ m + e−µr f1 (r) −µr = = 0 0 (10.438) (10.434) results in √ √ (m − ) m + a − (m − ) m + a √ √ (m + ) m − a − (m + ) m − a which is obviously correct. µ is real.437) where they ˜ ˜ cancelled in case f1 = g1 = a. ρ m+ ρ [∂ρ − (10.439) − m− e g1 (r) ˜ where µ is given in (10.440) and (10. However.436) f1 (r) = Insertion into (10.10: Dirac Particles in Electromagnetic Field For bound states holds 347 < m and. We also introduce the new variable ρ = 2µ r .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 . without loss of generality we can choose ˜ f1 (r) = φ1 (r) + φ2 (r) . (10. √ g1 (r) = − m − a e−µr . hence.441) correspond to (10.445) .444) (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. Let us consider then for the solution of (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. Solution of the Radial Dirac Equation for a Coulomb Potential To solve (10. g1 (r) = φ1 (r) − φ2 (r) ˜ (10.440) (10. Accordingly.h.436).441) The last two terms on the l.442) which leads to a partial cancellation of the asymptotically dominant terms.435 ).437) (10.s. In the present case the functions f1 and g1 cannot be chosen ˜ ˜ identical due to the terms in the differential equations contributing for finite r. Equation (10.

450) −√ mZe2 m2 − 2 βs ρs+γ−1 = 0 mZe2 m2 − αs (s + γ) β2 + (j + 1 ) αs + √ 2 s 2 +√ From (10.446 .426–10.451) 1 √mZe − (j + 2 ) 2 2 αs .454) .453) (s + γ)2 + Z 2 e4 − (j + 1 )2 2 2 √ Ze m2 − 2 βs .430)].448. 10. (10.430) which conform to the proper r → 0 behaviour determined above [c.446.452) From (10.450) follows Ze2 m2 − 2 βs − βs−1 =0 (10.448) φ2 (ρ) = ρ (10.449) into (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. = m− Ze2 βs s + γ − √m2 − 2 2 (10. 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. Inserting (10. 10. Using (10.f.446) (10.430) one can write this βs = s+γ − 2 √ Ze m2 − 2 s (s + 2γ) βs−1 .447) We seek solutions of (10.447) leads to (s + γ) αs + (j + 1 ) βs − √ 2 s Ze2 m2 − 2 αs (10.447) of the form n φ1 (ρ) = ργ s=0 n γ s=0 αs ρ2 βs ρ2 (10.449) for γ given in (10.

(s − so ) β0 s! (2γ + 1)(2γ + 2) · · · (2γ + s) (10.e.458) or. 3d 3 holds n = 1. 10. .457) One can relate the polynomials φ1 (ρ) and φ2 (ρ) defined through (10. m (n ) = . We can. .405) and the ensuing so values ( 10.461) In order that the wave functions remain normalizable the power series (10.457) for βs and αs imply that so must then be an integer. 2. 10.455) this confinement of so implies discrete values for ..455) are finite.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 ) . 1. for the states 2p 1 . n It holds φ1 (ρ) φ2 (ρ) = = ργ F (−so . namely.461 for values given by (10..457) with the confluent hypergeometric functions F (a.449) and (10. c c(c + 1) 2! (10. conclude 2 2 2 that the polynomials in (10. 1.405).452) follows j+ αs = 1 2 349 Ze2 m2 − 2 −γ (10.456) − so 2 √mZe m2 − 2 (−so )(1 − so )(2 − so ) . 10. . 2γ + 1.. x) . 2. i.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. . 3p 1 .459) − 2 √mZe m2 − 2 (10.448. .456. α + 1. . x) = 1 + a a(a + 1) x2 x + + . . This requires that all coefficients αs and βs must vanish for s ≥ n for some n ∈ N. equivalently. According to the definitions (10. with the associated Laguerre polynomials L(α) = F (−n.430. . 2γ + 1.10.10: Dirac Particles in Electromagnetic Field Defining so = √ one obtains βs = = . 10. = From (10. (10.456) and (10. in fact. ρ) so β0 ργ F (1 − so .448.405) allows one to identify n = n − j + 2 which.460) (10. For example. ρ) . n = 0. (s − so ) β0 s! (2γ + 1)(2γ + 2) · · · (2γ + s) (10. . β0 j+ 1 2 (10. is an integer. . c.449) must be of finite order. The expressions (10. so = n . Comparision with (10. hence.

m (j + 1 . j.147. The coefficients β0 in (10.466) F1 (r) = F− (κ|r) . can follow the procedure adopted for the wave functions of the non-relativistic hydrogen atom and will not be carried out here.460.461).421) correspond to non-relativistic states with orbital angular momentum 1 1 = j + 2 . (Springer. 2γ + 1. The wave functions (10.460. 10. 2µr) (10.438.439). where2 F± (κ|r) = G1 (r) = F+ (κ|r) .442).421). (10. Berlin. 10. The complete wave function is given by the following set of formulas 1 Ψ(n. 2γ + 1.465) (10. (n + γ)m N (2µr)γ−1 e−µr − κ F (−n . g1 (r) determined by (10.148).464) 0 The evaluation of the integrals.463) where Φ and X . (10. 10. 1990). which involve the confluent hypergeometric functions in (10. = j + 2 .350 Relativistic Quantum Mechanics Altogether we have determined the stationary states of the type (10.468) −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 . Sect.410) of the stationary state wave function the density ρ(xµ ) of the states under consideration. Greiner. 10. = j + 2 .421)] ∞ dr (|f1 (r)|2 + |g1 (r)|2 ) = 1 (10. The orthonormality 2 2 r properties (6. 1 |ˆ) 2 2 r G1 (r) Yj. 9.463) and yield [note the 1/r factor in (10. j.267). 1 |ˆ) in (6. is time-independent. m.421) with radial wave functions f1 (r).158) of the latter states absorb the angular integral in (10. and (10.467) ± n F (1 − n .460.461).m (j − 1 . as in (10. They are described through quantum numbers n. given by expression (10. m|xµ ) = e−i t iF1 (r) Yj. 2 . The normalization integral is then ∞ 0 π 2π r2 dr 0 sin θdθ 0 dφ (|Φ(r)|2 + |X (r)|2 ) = 1 (10. 2µr) (2µ) 2 Γ(2γ + 1) 3 N = and m 4m )Γ(2γ + n + 1) (n +γ)m (n +γ)m (10. 6.6. Due to the form (10.157. 6. 1 |ˆ) 2 2 r κ = j+ 1 2 (10.469) This formula has been adapted from ”Relativistic Quantum Mechanics” by W. are two-dimensional vectors determined through the explicit form of the spin-orbital angular momentum states Yjm (j ± 1 .461) are to be chosen to satisfy a normalization condition and to assign an overall phase.

2s 1 . in the non-relativistic limit.421) are governed by the radial Dirac equation (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. 1 |ˆ) 2 2 r κ = −j − 1 2 (10.469).422) to (10. where F± (κ|r) are as given in (10. hence. not not 2 2 2 2 2 2 covered by the wave functions given by (10.470) r 0 Obviously. One can verify. 2p 3 .10: Dirac Particles in Electromagnetic Field 351 We want to consider now the stationary states of the type (10. become   f2 (r) 1 1 i Yjm (j − 2 . 1 |ˆ) 2 2 r G2 (r) Yj. j. describes the complementary set of states 1s 1 . tracing all steps which lead from (10. 3p 3 .421) which. 2 |ˆ)  r Ψ(xµ ) ≈ e−1 t  . 3s 1 .m (j + 1 .467–10. = j − µ 1 2 . obtained closed expressions for the wave functions of all the stationary bound states of relativistic hydrogen-type atoms. . G2 (r) = F+ (κ|r) . 2 accordingly. (10.471) (10.472) F2 (r) = F− (κ|r) .465–10.469) that the following wave functions result Ψ(n. m|x ) = e −i t iF2 (r) Yj.469). The radial wave functions f2 (r) and g2 (r) in (10.m (j − 1 . this wavefunction has an orbital angular momentum quantum number = j − 1 and. We have.10. etc. 3d 5 .

352 Relativistic Quantum Mechanics .

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. Starting point is the Lorentz transformation S(w.Chapter 11 Spinor Formulation of Relativistic Quantum Mechanics 11. may be exploited to express the Lorentz-invariant equations of relativistic quantum mechanics. in fact. 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. The lower dimensionality of a(z) and b(z) implies. in particular. i. ˜ We have altered here slightly our notation of S(w. ϑ)µ ν correspond to different a(z) and b(z).. ϑ through 3. what will be achieved in the following.225–10..229). This is.2) (11. ϑ). different L(w.3) (11. in a sense. i. 1 353 . indeed. ϑ)µ ν . must be two-dimensional irreducible representations of the Lorentz group. We will characterize the space on which the transformations a(z) We will see below that the representations a(z) and b(z) are. We will proceed by building as much as possible on the results obtained sofar in the chiral representation of the Dirac equation.e. expressing its dependence on w. 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. a(z) and b(z).4) = exp = w − iϑ . z ∈ C ˜ Because of its block-diagonal form each of the diagonal components of S(z). a complex variable z.1 The Lorentz Transformation of the Dirac Bispinor We will provide in the following a new formulation of the Dirac equation in the chiral representation ˜ defined through (10. ϑ) for the ˜ in the chiral representation as given by (10.262). isomorphic to the natural representation.e.1) (11.

b(z). (11.8) We like to stress that there exists. +1  2 2 |1. In this case a(z) and b(z) are identical and read a(i ϑ) = b(i ϑ) = exp 1 − ϑ·σ 2 . and the Klein–Gordon equation in the spinor representation.243). −1 2 2 z = iϑ.7) where “∼” stands for “transforms like”. 2 2 2 ˜ Since a(z) acts on the first two components of the solution Ψ of the Dirac equation.1 states as described by Dm m (ϑ).5) The transformations in this case. ϑ ∈ R3 . for z = iϑ. This characterization allows one to draw conclusions regarding the state space in which a(z) and b(z) operate. Ψ4 (xµ ) transform according to ˜ ˜ ˜ and Ψ b(z). neutrino equation. β.354 Spinor Formulation and b(z) act. Ψ2 (xµ ) transform according to a(z) whereas Ψ3 (xµ ). Here | 1 . In this case holds a(z) = b(z) ˜ 1 (xµ ). however. 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 . ϑ ∈ R3 . For ϑ = (0. ϑ ∈ R3 (11.e. β. are elements of SU(2) and correspond.9) a∗ (z) = exp 2 . ± 1 represents the familiar spin– 1 states. to 1 ∗ ∗ z ·σ (11. − 1 2 2 z = iϑ.6) as given by (5. (11.. −1  2 2  |1. i.e. A First Characterization of the Bispinor States We note that in case w = 0 the Dirac transformations are pure rotations. (11. and since b(z) ˜ ˜ acts on the third and fourth component of Ψ one can characterize Ψ  ˜ Ψ1 (xµ ) ˜  Ψ2 (xµ )   Ψ3 (xµ ) ˜ ˜ Ψ4 (xµ ) ∼ ∼ ∼ ∼  1 |2.. namely. finally. will establish the map between L(w. a distinct difference in the transformation behaviour of ˜ ˜ ˜ ˜ Ψ1 (xµ ). to the rotational transformations of spin. the so-called spinor space. in 2 fact. a space of vectors state1 for which holds state2 state 1 ∼ | 1 . + 1 2 2 state 2 ∼ | 1 . Aν . usually expressed 2 as product of rotations and of functions of Euler angles α. +1 2 |1. Ψ4 (xµ ) in case z = w + iϑ for w = 0. the operator ∂µ and the Pauli matrices σ in the new representation and. express 4-vectors Aµ . Relationship Between a(z) and b(z) The transformation b(z) can be related to the conjugate complex of the transformation a(z). Ψ2 (xµ ) and Ψ3 (xµ ). γ (see Chapter 5). ϑ)µ ν and a(z). formulate the Dirac equation. θ ∈ R3 . i.

−1 = ∗ −σ3 = −σ ∗ . one can express a∗ (z) −1 = exp 1 − z ∗· σ 2 = b(z) . using f (a) −1 = f ( a −1 ). The answer lies in the necessity that application of spatial inversion should transform a solution of the Dirac equation into another possible solution of the Dirac equation. 11. Ψ (xµ ) even though these pairs of ˜ ˜3 ˜4 featuring the components Ψ1 (x ˜ 2 components transform independently of each other. ∗ σ 3 = σ3 . One can readily verify [c. −1 given by (11.f.14) .12). hold −1 = 1 One notices then. however.11.12) yields σ1 σ3 or in short σ Similarly. (5. ∗ σ 2 = − σ2 . but leave rotation angles ϑ unaltered. therefore. in fact.11) 0 −1 1 0 = .4) and . (11.12) To prove (11. one can show −1 −1 −1 = −σ1 = σ2 = −σ3 ∗ = −σ1 ∗ = −σ2 σ2 −1 (11. that the transformation (11.15) (11.1: Lorentz Transformation of Dirac Bispinor ∗ ∗ ∗ where σ ∗ = (σ1 . Hence. (11.12) does. σ2 . σ3 ).18) with a(z) given by (11.10) From this one can derive b(z) = where = 0 1 −1 0 . This demonstrates that a(z) is the ˜ transformation which characterizes both components of S(z).17) We conclude. −1 a∗ (z) −1 (11. 11.224. a∗ (z) −1 = exp 1 ∗ z σ∗ 2 −1 . . 11. that they alter w into −w. The effect of inversion on Lorentz transformations is. a result to be used further below. Ψ (xµ ) as well as Ψ (xµ ).10. (11.13) Explicit matrix multiplication using (5. (11.2. µ ).1) can be written in the form ˜ S(z) = a(z) 0 0 a∗ (z) −1 (11. 1.224)] ∗ σ1 = σ1 .11) one first demonstrates that for and −1 as given in (11.16) σ = −σ ∗ . Spatial Inversion ˜ One may question from the form of S(z) why the Dirac equation needs to be four-dimensional. 355 (11.

This implies  ˜ Ψ1 ˜  Ψ2 P  ˜  Ψ3 ˜ Ψ4  ˜ Ψ3 ˜   Ψ4  =    Ψ1 ˜ ˜ Ψ2     .3.e. Exercise 1. P −1 = P. The transformation S(z) in the transformed space is then ˜ P 1. based on the Jordan–Chevalley theorem. In fact. the space spanned by only two of the components of Ψ is not invariant under spatial inversion and.f.22) which follows from the fact that for any complex. the Lorentz invariant equation for neutrinos is only 2–dimensional..224)] tr( σj ) = 0 .19) i.10.356 Spinor Formulation ˜ Let P denote the representation of spatial inversion in the space of the wave functions Ψ. However. can be found in G.1) We have pointed out that a(iϑ). does not suffice for particles like the electron which obey inversion symmetry.24) = etr(M ) (11.23) Exercise 11.21) = etr( 2 z·σ) = 1 1 (11. However. the transformations a(z) and b(z) become interchanged. M is a complex 2 × 2–matrix.G. Obviously. One can verify this by evaluating the determinant of a(z) det( a(z) ) = det e 2 z·σ 1 (11. ˜ ˜ P S(w + iϑ) P = S(−w + iϑ) = b(z) 0 0 a(z) . C) = { M. 1990). 11. 2 = 1 i.1: Show that SL(2. is an element of SU(2).A.21) together with matrix multiplication as the binary operation forms a group. B¨uerle and E. Part (Elsevier. which describes pure rotations.. (5. 2.2 Relationship Between the Lie Groups SL(2. a(w + iϑ) for w = 0 is an element of SL(2. C) defined in (11.C) and SO(3. 3 . det(M ) = 1 } . (11. ϑ ∈ R3 .  (11. j = 1. Amsterdam.20) ˜ Obviously. 2 . de Kerf Lie a Algebras. for particles like the neutrinos which do not obey inversion symmetry two components of the wave function are sufficient.A. (11. For the general case the proof. hence.2.e. non-singular matrix M holds2 det eM and from [c. The proof of this important property is straightforward in case of hermitian M (see Chapter 5).

(11. (11.31) holds.11.76). 0 −i i 0 .75) holds aν = Lν µ aµ one realizes that σµ transforms inversely to covariant 4–vectors.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 .27) requires then σ µ Aµ Lν µ σµ = σν (11. (11.25) of the matrix M (Aµ ) is independent of the frame of reference.25) can also be written M (Aµ ) = A0 + A3 A1 − iA2 A1 + iA2 A0 − A3 .31) Exercise 11.135)] this transformation behaviour.2. Noting that for covariant vectors according to (10.2: Lie Groups SL(2. σ1 . (11. (11. the matrix M (Aµ ) is hermitian as can be seen from inspection of (11. in fact. Demonstrate that (11.1) Mapping Aµ onto matrices M (Aµ ) 357 We want to establish now the relationship between SL(2. or+ thochronous Lorentz transformations.26) or from the fact that the matrices σ0 . We will prove below [cf. one can provide a simple expression for the inverse of M (Aµ ) M = M (Aµ ) ↔ Aµ = 1 tr M σµ 2 . Starting point is a bijective map between R4 and the set of two-dimensional hermitian matrices defined through M (Aµ ) = σµ Aµ where   (11. 1 0 0 −1    . −→ µ Consistency of (11. σ2 .25) σ0 σ1 σ2 σ3 The quantity σµ thus defined does not transform like a covariant 4–vector.. 0 1 1 0 .26) (11. The function M (Aµ ) is bijective. σ3 provide a linear–independent basis for the space of hermitian 2 × 2–matrices.30).e. Argue why M (Aµ ) = σµ Aµ provides a bijective map. i.32) . C) and the group L↑ of proper.  (11. Using Aµ = (L−1 )µ ν A ν one would expect in a transformed frame to hold Lρ ν σ (L−1 )µ ν A ν .28)   σµ =   1 0 0 1 . In fact.30) Since the components of Aµ are real. σ2 .28) and (11. σ1 . The following important property holds for M (Aµ ) det ( M (Aµ ) ) = Aµ Aµ which follows directly from (11.29) where we used (10. one wishes that the definition (11.C) and SO(3. σ3 are hermitian. M (Aµ ) according to (11.2: Show that σ0 .

33) det(M ) = = det( a M a† ) = det(a) det(a† ) det(M ) [det(a)]2 det(M ) = det(M ) . (11. Accordingly.38) . C) defines the transformation [c.32.37) which implies that (11. C) . it holds for the matrix M defined through (11. + . (11. Due to det(a) = 1 the transformation (11. 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† .39) Exercise 11. (11.25) Aµ = which allows us to express finally A µ = L(a)µ ν Aν .31). The linear character of the transformation becomes apparent expressing A µ as given in (11.40) is an element of L↑ . (11. In fact.2. L(a)µ ν = 1 tr a σν a† σµ 2 . a ∈ SL(2.33) to M (Aµ ) describing the action of the transformation in terms of transformations of Aµ .35) We now apply the transformation (11.33) conserves the determinant of M .f.31)] 1 a Aµ −→ A µ = tr a M (Aµ ) a† σµ 2 Because of (11.34) where we used the properties M † = M and (a† )† = a. (11.35) holds for this transformation A µ Aµ = Aµ Aµ (11.37) using (11.3: Show that L(a)µ ν defined in (11.37) defines actually a Lorentz transformation.36) The suitable A µ can readily be constructed using (11.40) 1 tr a σν a† σµ 2 Aν (11. C) and for any Aµ there exists an A µ such that M (A µ ) = a M (Aµ ) a† . In fact. any a ∈ SL(2. (11. for any a ∈ SL(2.33) This transformation conserves the hermitian property of M since (M )† = ( a M a† )† = (a† )† M † a† = a M a† = M (11.

47) This completes the proof of the homomorphic property of L(a)µ ν .δ 1 4 Aαβγδ Γβα Γγδ α.. .45) One can demonstrate through direct evaluation   2 α =    2 α =  2 α = =   2 α =    0 else β β γ γ = = = = γ = γ = 1. i. β 2.1) defined through L(a)µ ν [cf.43) (σν )αβ Γβα Γγδ (σν )δγ = ν.40)] respects the group property of SL(2.1).C) and SO(3.44) (σν )αβ (σν )δγ .2: Lie Groups SL(2.1) For this purpose one writes using tr(AB) = tr(BA) L(a1 )µ ν L(a2 )µ ρ = ν a1 a2 product in SL(2. β δ = δ = = δ = δ 1 2 = 2 = 1 Aαβγδ (11.41) results in 1 ¯ Lµ ρ = 4 where Aαβγδ = ν Γ = a2 σρ a† 2 (11. (11. C) and SO(3. (11.1) L(a)µ ν provides a homomorphism 359 We want to demonstrate now that the map between SL(2.42) = ν Defining Γ = a† σµ a1 .e.δ (11. ¯ Lµ ρ = L(a1 )µ ν L(a2 )µ ρ = L( product in SO(3. 2 (11.β γ. C) and of SO(3.α.β γ.41) 1 tr a1 σν a† σµ 1 2 1 tr σν a† σµ a1 1 2 1 tr a2 σρ a† σν 2 2 1 tr a2 σρ a† σν 2 2 .C) )µ ρ (11. 1 ¯ and using the definition of Lµ ρ in (11.11.46) 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 . (11.

0) ) σ0 A 1 − σ1 A 0 (11. M (Aµ + (K1 )µ ν Aν ) = = M (Aµ ) + M (Aµ ) − M ( (−A1 . J2 . 0.e. Because of the condition det(a) = 1 only six independent real numbers actually suffice for the definition of a. To this end we consider infinitesimal transformations and keep only terms of zero order and first order in the small variables. 0. K2 .52) where we employed (11. 0. 0).53) Equating (11.51) Insertion of (K1 )µ ν as given in (10.h. 0.52) and (11. 0. (11. of (11. eight real numbers. i.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. 1. 0).55) (11.48) We want to determine now the generators of the transformation a(z) as defined in (11.h. J Spinor Formulation The transformations a ∈ SL(2. (11.s.50) (11. C) as complex 2 × 2–matrices are defined through four complex or. 0).48). 1 (11. noting the linearity of M (Aµ ).36) M (Lµ ν Aν ) = a M (Aµ ) a† (11. J1 . −A0 . of (11.58) .56) (11. C) which correspond to K. J3 of the Lorentz transformations Lµ ν in the natural representations. (11.2) which correspond to the generators K1 .53) results in the condition σ 0 A1 − σ 1 A0 = ( κ 1 σ µ + σ µ κ † ) Aµ .48) yields for the l..57) .49) we obtain using (11. To obtain the generator of a(z) corresponding to the generator K1 . For the r.360 Generators for SL(2. Aµ = (0. K3 . denoted below as κ1 . 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  .54) This reads for the four cases Aµ = (1.47. 10. we write (11. 1.s.49). Aµ = (0. 0.25) in the last step. 1 (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 ) . correspondingly. correspond to the generators given by (10. Aµ = (0. 1) − σ1 σ0 0 0 One can verify readily that 1 κ1 = − σ 1 2 (11.49) assuming (note that g µ ν is just the familiar Kronecker δµν ) Lµ ν a = = gµν + 1 + 1 (K1 )µ ν κ1 . 0.

2 (11. λ2 . (11. J2 . def α = 1.1) acts on 4–vectors Aµ . 2 λ = i σ. K3 and λ1 . C) acts on spinors φα ∈ C2 (characterized below) 1 (11.4: Show that the generators (11. ϑ) = ew·K + ϑ·J ∈ SO(3. − 1 . J3 are given by 1 κ = − σ.62) We denote the general (z = w + iϑ) transformation by φ α = aα β φβ = aα 1 φ1 + aα 2 φ2 . Exercise 11.60) of a ∈ SL(2. C) correspond to the generators K and J of Lµ ν . Similarly. w describing boosts and ϑ describing rotations. 11.3 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. a1 1 a1 2 a2 1 a2 2 (11. the so-called contravariant spinors. a(w − iϑ) = e− 2 (w − iϑ)·σ ∈ SL(2.63) where we extended the summation convention of 4-vectors to spinors. 2 2 φ1 φ2 ∈ C2 . + 1 2 2 1 transforms under rotations (z = iϑ) like a spin– 2 state | 1 .61) This identifies the transformations a(z = w − iϑ) as representations of Lorentz transformations.64) . 2 (11.3: Spinors 361 obeys these conditions. 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 . Here a(z) = ( aα β ) = describes the matrix a(z). λ3 corresponding to J1 . one can show that the generators κ2 .11. hence. κ3 of a(z) corresponding to K2 .2. state that the following two transformations are equivalent L(w.60) We can.

e.66) should constitute a singlet spin state.69) (11. which 2 can be constructed from products φ1 χ2 . as contravariant spinors and to φα . 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.. The corresponding states are defined through φ1 χ2 − φ2 χ1 = φ1 χ1 + φ2 χ2 It obviously holds χα = χ1 χ2 = χ2 −χ1 . The relationship between the two can be expressed φ1 φ2 = = 3 φ1 φ2 0 1 −1 0 (11.68) (11. 1 . 1 . ± 1 . In fact.. 0. (11. The existence of a scalar product gives rise to the definition of a dual representation of the states φα denoted by φα . In the notation developed in Chapter 5 holds for the singlet state | . invariant under transformations a(z). . −m 2 2 2 2 1 1 1 1 2 (11. exchange of φα and χβ alters the sign of the expression. etc. ± 1 . −m) | . i. To arrive at a suitable scalar product we consider first only rotational transformations a(iϑ).67) We will refer to φα . as we will demonstrate below such states are invariant under general Lorentz transformations a(z). i. . . 1 ) stands for the Clebsch–Gordon coefficient. However.. Definition of covariant spinors Expression (11. is the singlet state. .66) is a bilinear form. χβ . .. . m 1 | . Such a scalar product does. as covariant spinors. Using (0. this scalar product is anti-symmetric.e.65) where | · · · 1 describes the spin state of “particle 1” and | · · · 2 describes the spin state of “particle 2” and (0.70) ‘Scalar’ implies invariance under rotations and is conventionally generalized to invariance under other symmetry trasnformations. indeed. those of “particle 2” with the spinor χβ one can state that the quantity 1 Σ = √ 2 φ1 χ2 − φ2 χ1 (11. should remain invariant under transformations a(iϑ). 0| . In this case spinors φα transform like spin– 1 states and an invariant. 0| 1 . i.e. 0 = 2 2 m=± 1 2 1 1 (0. . m. . invariant and as such has the necessary properties of a scalar3 product. χβ . 0| 1 . 1 ) = 2 2 2 2 2 2 2 2 √ ±1/ 2 and equating the spin states of “particle 1” with the spinor φα . .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.

−1 connecting contravariant and covariant spinors play the role of the metric tensors gµν .69.78) aβ γ γδ φδ (11.81) . 10.70) and (11. g µν of the Minkowski space [cf.72) 0 −1 1 0 Exercise 11. Accordingly.10.69. (11. (11. 11.1: Show that for any non-singular complex 2 × 2–matrix M holds M −1 = det(M ) M −1 T The matrices . For this scalar product holds φα χα = −χα φα . Proof that φα χα is Lorentz invariant We want to verify now that the scalar product (11. 11.77) (11.3.73) (11.72)] φα φ α = = = = αβ αβ φβ φβ 11 21 11 21 12 22 12 22 (11.74)].73.79) = a −1 T a κβ . In the transformed frame holds φ α χα = aα β αγ aγ δ δκ φβ χκ . 11. 11.63. The inverse of (11.3: Spinors as can be verified from (11.77) is Lorentz invariant.70) is φ1 φ2 −1 363 = = −1 φ1 φ2 .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 (11.11. (11.71.71) (11.75) (11.75) is given by φα = as can be readily verified.76) αβ αβ The scalar product (11.68). (10. we will express (11. (10.67) will be expressed as φα χα = φ1 χ1 + φ2 χ2 = φ1 χ2 − φ2 χ1 . The transformation behaviour of φα according to (11. 11.72) in a notation analogous to that chosen for contravariant and covariant 4–vectors [cf.

Hence.90) extending the summation convention to ‘dotted’ indices. the ˙ conjugate complex spinors φα need to be considered separately from the spinors φα . The transformation ˙ behaviour of φα is ˙ φ α = (aα β )∗ φβ ˙ (11.80) φ α χα = φβ χβ . it holds φk = (φk )∗ .83) into (11. We denote ˙ (aα β )∗ = aα β ˙ (11.63).87) which we will employ from now on. hence. a∗ (z) provides a representation of the Lorentz group which is distinct from that provided by a(z). from (11. σ † = (σ ∗ )T = σ. ˙ (11.82) and with f (A)T = f (AT ) for polynomial f (A) a −1 T = e− 2 z·(σ 1 ∗ )T = e− 2 z·σ = a−1 1 (11.2) and (11.85) φ1 φ2 ∗ ∗ . Insertion of (11.84) (11. (11.14) one can write a −1 Spinor Formulation = e 2 z·σ 1 −1 = e 2 z· 1 σ −1 = e− 2 z·σ 1 ∗ (11.81) yields aα β and.364 Using (11. i. As discussed above.86) A concise notation of the conjugate complex spinors is provided by ˙ (φα )∗ = φα = φ1 ˙ φ2 ˙ ˙ (11.88) which one verifies taking the conjugate complex of (11.91) .88) reads ˙ ˙ φ α = aα β φβ ˙ ˙ (11. ˙ We also define covariant versions of φα φα = (φα )∗ . Obviously. 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 (11.83) Here we have employed the hermitian property of σ. 2. k = 1.e.89) such that (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 )   =   = .76) φα ˙ ˙ φα αβ ˙ ˙ αβ ˙ ˙ = = = = αβ ˙ ˙ αβ ˙ ˙ αβ αβ φβ φβ ˙ ˙ (11. + 1 2 2 2 1 transforms under rotations (z = iϑ) like a spin– 2 state | 1 .97) implies then that φα transforms like Ψ ˙ φ1 ˙ φ2 ˙ transforms under rotations (z = iϑ) like a spin– 1 state | 1 . a∗ (z) −1 accounting for the ˜ transformation behaviour of the third and fourth spinor component of Ψ.e.11. one can state ˜ and (11.95) ˙ where αβ and αβ are the real matrices defined in (11. A tensor tα1 α2 ···αk β1 β2 ···β (11.75.92) (11.75).4 Spinor Tensors ˙ ˙ ˙ We generalize now our definition of spinors φα to tensors.100) . 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. 11. Ψ4 .97) The transformation.18) ˜ 3 . For the spinors φα and χα thus ˙ defined holds that the scalar product ˙ φα χα = φ1 χ1 + φ2 χ2 ˙ ˙ ˙ ˙ ˙ (11. in matrix notation. 11. i. The transformation behaviour of φα is ˙ φα = ˙ ˙ β γδ ˙˙ φδ ˙ ˙ αβ a γ ˙ ˙ . is governed by the operator a∗ (z) −1 which arises in ˜ the Lorentz transformation (11.99)  Ψ3 (xµ )   χ ˙ (xµ )  ˜ χβ (xµ ) ˙ 1 ˜ χ2 (xµ ) Ψ4 (xµ ) ˙ 11.96) is Lorentz invariant. a property which is rather evident.73.94) (11.93) (11. (11..4: Spinor Tensors 365 The relationship between contravariant and covariant conjugate complex spinors can be expressed in analogy to (11. A comparision of (11.18) of the bispinor wave function Ψ. − 1 . (11.

with ˙ contravariant indices αβ.102) = .m aj a∗ t km m .m aj akm t m (11.106) Exercise 11. ˙ ˙ ˙ ˙ (11. According to (11. i. k. this transformation behaviour is in harmony with the tensor notation adopted. using conventional matrix indices j.1: Prove equation (11.25] M (Aµ ) = σµ Aµ . it holds αβ . i. An example of a tensor is αβ and tions. (11.25) the tensor is explicitly Aαβ = ˙ A1 1 A1 2 ˙ ˙ A2 1 A2 2 ˙ ˙ = A0 + A3 A1 − iA2 A1 + iA2 A0 − A3 . 11. αβ ˙ αγ t ˙ γβ (11. we want to express Aµ through a spinor tensor.e.e. We had demonstrated that M (Aµ ) transforms according to M = M (Lµ ν Aν ) = a M (Aµ ) a† (11. This tensor transforms according to t αβ = aα γ aβ δ tγ δ ˙ This reads in matrix notation..106).102. (11. (11..109) ..105) This tensor is actually invariant under Lorentz transforma= αβ .108) Obviously. tjk = a t a† jk ˙ ˙ ˙ ˙ (11. An obvious candidate is [cf.e. The 4–vector Aµ in spinor form We want to provide now the spinor form of the 4-vector Aµ .103) A αβ = aα γ aβ δ Aγ δ .104) Indices on tensors can also be lowered employing the formula tα β = and generalizations thereof.366 is a quantity which under Lorentz transformations behaves as t ˙ ˙ ˙ α1 α2 ···αk β1 β2 ···β k Spinor Formulation = m=1 aαm γm n=1 aβn δn tγ1 ···γk δ1 ···δ . the transformation bevaviour of a tensor tαβ reads in spinor and matrix notation t αβ = aαγ aβδ tγδ .4. This task implies that we seek a tensor. i.103) Similarly. αβ = αβ (11. ˙ ˙ ˙ ˙ (11. m.107) which reads in spinor notation [cf.101) An example is the tensor tαβ which will play an important role in the spinor presentation of the Dirac equation. (11. . the elements of which are linear functions of Aµ . tjk = a t aT jk = .

(11. (11.114) We finally note that the 4–vector scalar product Aµ Bµ reads in spinor notation Aµ B µ = 1 αβ ˙ A B αβ . Aµ and tensors tαβ can be applied to the partial differential operator ∂µ . employing (11.109. ∂µ in spinor notation The relationship between 4–vectors Aµ .111) A11 A12 ˙ ˙ A21 A22 ˙ ˙ ˙ 0 −1 1 0 (11.116) . 11.11.115) is correct.4. Using (11. (11. Aµ can also be associated with tensors Aα β = ˙ This tensor reads in matrix notation A11 A12 ˙ ˙ A21 A22 ˙ ˙ = = 0 1 −1 0 ˙ ˙˙ αγ β δ A ˙ γδ .115) Exercise 11.2: Prove that (11. ˙ 2 (11.110) one obtains Aα β = ˙ A0 − A3 −A1 − iA2 −A1 + iA2 A0 + A3 A0 + A3 A1 + iA2 A1 − iA2 A0 − A3 .114) yields ∂αβ = ˙ ∂0 + ∂3 ∂1 + i∂2 ∂1 − i∂2 ∂0 − ∂3 . (11. (11.113) Aα β = ˙ (11.117) ˙ ˙ ∂0 − ∂3 −∂1 + i∂2 −∂1 − i∂2 ∂0 + ∂3 . Hence.112) A22 −A21 ˙ ˙ −A12 A1 1 . (11.4: Spinor Tensors One can express Aαβ also through Aµ Aα β = ˙ ˙ 367 A0 − A3 −A1 + iA2 −A1 − iA2 A0 + A3 .110) one can state ∂ αβ = Similarly.110) The 4-vectors Aµ .

1 0 0 −1 .e.124) constitutes the essential transformation property of σ µ and will be considered further. σµ . This is ˜ possible since γ µ can be expressed through σ µ .f. Comparision of (10. one can conclude aσ µ ∗ −1 Lν µ σν 0 = . hence.18) and γ µ by (11.121) One expects then that the transformation properties of σ µ should follow from the transformation properties established already for γ µ [c.121). .. Hence.122) aσ µ −1 a−1 −1 (a∗ )−1 0 Lν µ (11. −1 0 0 1 (11. . (11.243) in the chiral representation expressing S(Lη ξ ) by (11. 0 1 1 0 .118) 0 −1 −1 0 0 i −i 0 For this purpose we consider first the transformation behaviour of σ µ and σµ .s.h. We will obtain the transformation behaviour of σ µ .. (10.e.124). i. Equation (10. hence. i. (11. holds also in the chiral representation. 0 −i i 0 .243)]. of this equation is 0 ∗ (σ µ )∗ a and.119) (11. σµ building on the known transformation behaviour of γ µ . Note that (10. γµ = ˜ 0 µ )∗ (σ −1 0 σµ σµ 0 (σ µ )∗ −1 .368 σµ in Tensor Notation Spinor Formulation We want to develop now the tensor notation for σµ (11.125) is equivalent to µ ∗ −1 −1 a∗ (σ µ )∗ −1 −1 a Lν µ = (σ ν )∗ (11.125) a (σ ) Equation (11.124) . To obtain the transformation properties of σ µ we employ then (10. (11.120) σµ 0 .229) and (11.243) 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. .26) and its contravariant analogue σ µ σµ = σµ = 1 0 0 1 1 0 0 1 . (11.123) (a∗ )−1 Lν µ a L ν µ = = σν (σ ) ν ∗ −1 (11.124).118) yields ˜ γµ = ˜ Using (11.125) is equivalent to (11. (11. (11.15) one can write σµ = and.243) holds independently of the representation chosen.126) which is the complex conjugate of (11.

. therefore. 1 0 0 −1 µ=3   . 0 −i i 0 µ=2 .103) ˙ identifies σ µ as a tensor of type tαβ .124) a σ µ [a∗ ]T Lν µ = σ ν . −1 0 0 1 µ=3   . (11.   =  .124) using (11.129. under rotations the Pauli matrices transform like (j = 1. hence. 3) σj −→ a(iϑ) σj a(iϑ) † = a(iϑ) σj a∗ (iϑ) T . (11.  (11. 11.  .131) one can express the transformation behaviour of σ µ in the succinct form Lν µ σ µ = σ ν . 11.108) that the same transformation behaviour applies then for general Lorentz transformations. (σ ∗ )T = σ yields using (11.127) Exploiting the hermitian property of σ.e.132) 1 0 0 1 µ=0 . One can.g.4) with w = 0. e. (11.130) We argue in analogy to the logic applied in going from (11.102. i. 11.2) −1 (a∗ )−1 = e2z 1 ∗ ·(σ ∗ )T = e2z 1 ∗ ·σ ∗ T = [a∗ ]T .2. This transformation behaviour.11..2) −1 369 (a∗ )−1 = −1 − 1 z ∗ ·σ ∗ 2 e = e− 2 z 1 ∗ · −1 σ ∗ = e2z 1 ∗ ·σ . (11.2. state that σ µ in a new reference frame is σ µ = a σ µ a† (11.107) to (11. 11. It holds according to (11. according to (11.128) One can express. 0 −1 −1 0 µ=1 . (11. transformations (11. 0 1 1 0 µ=1 . equation (11. 0 i −i 0 µ=2 .4).133) Combining (11.131) where a is given by (11. In fact. 11.4: Spinor Tensors One can rewrite (11.16.134) .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. 2.118) (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.

=  σαβ ˙  ˙ ˙ ˙ ˙ (σ21 )11 (σ21 )12 (σ22 )11 (σ22 )12   ˙ ˙ ˙ ˙ ˙ ˙ ˙ ˙ 21 (σ )22 21 (σ )22 (σ21 ) (σ22 ) ˙ ˙ ˙ ˙ 21 22 Equating this with the r. γ. however.114).e. whereas ˙ account for the elements of the individual Pauli the ‘outer’ contravariant spinor indices. hence.. account for the 4-vector µ.141) σ αβ = σ0 − σ3 σ1 + iσ2 −σ1 − iσ2 σ0 + σ3  = 2  0 0 0 1 0 −1 0 0 . account for the 4–vector index µ.137) σαβ ˙ = δαγ δβ δ ..138) the ‘inner’ covariant spinor indices. We want to express now σ µ and σµ also with respect to the 4–vector index µ in spinor form employing (11. We can summarize that σ µ and σµ transform like a 4–vector.136) σαβ = = 2 ˙   σ1 − iσ2 σ0 − σ3 0 0 0 0 1 0 0 1 where on the rhs.29)]. µ = 0. We can. ˙ ˙ In (11. This yields   1 0 0 1   0 0 0 0 σ0 + σ3 σ1 + iσ2  (11.12) for of σµ yields together with (5. σµ in which the contravariant indices are moved ‘inside’. 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 . The ˙ desired change of representation(σµ )αβ −→ (σµ )αβ corresponds to a transformation of the basis of ˙ spin states f g f −→ = (11.370 Inverting contravariant and covariant indices one can also state Lν µ σµ = σν . the submatrices correspond to tαβ spinors.h.138) Note that all elements of σαβ are real and that there are only four non-vanishing elements. of (11. 2. matrices.135) This is the property surmised already above [cf.224) ˙ Using (11.140) where we employed the expressions (11.e. Spinor Formulation (11.e. ˙ ˙ 0 −1 1 0 ˙ (11.. We will now consider the representation of σ µ . i.110) to express σ αβ in terms 0 −1 1 0 0 0 0 1     (11. δ. 3 as well as through two spinor indices αβ.s. ˙˙ (11. state   ˙ ˙ ˙ ˙ (σ11 )11 (σ11 )12 (σ12 )11 (σ12 )12 ˙ ˙ ˙ ˙   ˙ ˙ ˙ ˙ ˙ γδ   (σ11 )21 (σ11 )22 (σ12 )21 (σ12 )22 ˙ ˙ ˙ ˙   . and the covariant indices are moved outside.132) and (11. α.133) is characterized through a 4-vector ˙ index µ. i. β.136) results in the succinct expression 1 2 ˙ γδ ˙ (11. the transformation is inverse to that of ordinary 4–vectors. i. (11. in fact. 1. Each of the 4 × 4 = 16 matrix elements in (11.139) g −f g and.

146) Exploiting the property (11.140) to transform each of the four submatrices.145) Let Aµ be a covariant 4–vector.138) one can conclude that the following property holds 1 2 ˙ γδ σ αβ ˙ = 1 2 σ αβ ˙ ˙ γδ = δαγ δβ δ . (11.11. (11.144) The Dirac Matrices γ µ in spinor notation We want to express now the Dirac matrices γ µ in spinor form. (11. Accordingly. which in ˙ (11.121) of γ ˜ ˜ ˙ basis |αβ whereas the element of γ µ given by σ µ is in the basis |αβ . ˙˙ (11.142) = This can be expressed 1 2 σ αβ ˙ ˙ γδ = δαγ δβ δ .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 .143) Combined with (11. One can write then the scalar product using (11.144) results in the simple relationship γ µ Aµ = ˜ 0 Aα β ˙ Aα β 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 ˙ ˙    .147) . employing then transformation (11. we write ˜ ˙ γµ = ˜ 0 (σ µ )αβ µ ) )∗ ((σ αβ 0 ˙ ˙ . For this purpose we start from the ˜ µ . This expression implies that the element of γ µ given by σ µ −1 is in the expression (11.4: Spinor Tensors 371 and. ˙˙ (11.  ˙ ˙ ˙ ˙       (11.

148) ˜ Employing Ψ(xµ ) in the form (11.99) yields the Dirac equation in spinor form i ∂ αβ χβ ˙ i ∂ αβ φ ˙ The simplicity of this equation is striking.5 Lorentz Invariant Field Equations in Spinor Form Dirac Equation (11.226) ˜ i γ µ ∂µ Ψ(xµ ) = 0 i ∂ αβ ˙ i ∂ αβ 0 ˙ ˜ ˜ Ψ(xµ ) = m Ψ(xµ ) .149) (11.147) allows us to rewrite the Dirac equation in the chiral representation (10. ˙ (11. (11.150) .372 Spinor Formulation 11. α ˙ = = m φα m χβ .

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

Sk . Classically. Sk . 2m r (12. For example... each energy level is (2l + 1)-fold degenerate. · · · .. (7. has been discussed.λn = E exp(iαk Sk ) ψE. . we first consider a particular example of the implications of symmetry. as studied in chapter 7. . ψE.. In order to illustrate this.7) 2 n where the orbital angular momentum. (12.m (r) Y m (θ.12).8) .. The energy levels are mk 2 En = − 2 2 .18).. the states 3s. We want to understand this extra degeneracy in terms of the extra symmetry of the hydrogen atom given by an additional set of symmetry generators introduced below. leaves the energy of the state invariant H exp(iαk Sk ) ψE. l and the corresponding energy levels are independent of m. generates a state which has the same energy as the original one. V (r).6) From elementary quantum mechanics we know the spectrum of the hydrogen atom. . As an example of this we will consider the Coulomb problem with the Hamiltonian H = p2 k − .39).λn . this implies a degeneracy in the spectrum.λ1 . Jk . namely motion in a spherically symmetric potential described by the group SO(3) (or its double covering SU (2) as discussed in section 5.. (12.λ1 . can then be reduced to a one-dimensional radial o equation.3) If the newly obtained state is linearly independent of the original one.374 Spinor Formulation The action of any symmetry generator.. We will investigate this shortly in more detail in the case of systems with spherical symmetry. which yields a set of eigenstates of the form ψE. This degeneracy follows from the fact that any rotation. k = 1.λn . the additional symmetry generators of the Coulomb problem are the three components of the so-called eccentricity vector discovered by Hamilton = 1 r p×J −k . .. (12.5) with m = −l. . φ). (7. Jk ] = 0 .4) The stationary Schr¨dinger equation. can be identified with the angular momentum operators... Presence of additional symmetries may further increase the degeneracy of the system. . (12.8) ˆ [H.λ1 . m r (12.m (r) = vE.. as represented by an element of SO(3).24). (7. 3p and 3d all have the same energy. . is allowed to take values in 0. The spherical symmetry implies the commutation of the Hamiltonian with angular momentum operators (7. 2. Lie groups play an essential role in the discussion of mass degeneracies in particle physics. Therefore. In chapter 7 the dynamics of a particle moving in three dimensions under the influence of a spherically symmetric potential. l. n − 1. The energy levels are totally independent of l. on an energy eigenstate. 3 . where the symmetry generators.

21) .18) which complement the familiar angular momentum algebra of section 5.11) (12. We will also need the following identity [4] 2 = 2H J2 + m 2 + k2 . The vector in (12. 2m r 1 x2 (J1 p3 − J3 p1 + i p2 ) − k . Aj ] = i ijk Ak . 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 . 2m r 1 x1 (J3 p2 − J2 p3 + i p1 ) − k . (12. Kj ] = i ijk Kk .12) (12. Now we scale the eccentricity vector as follows K= − m . 2 which can be shown to satisfy [Ai . We introduce the following new operators 1 A = (J + K).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.19) (12.8) is not a hermitian operator. In order to understand the aforementioned extra degeneracy.11.9) (12. In the following we consider the bound states.14) This follows from a · (a × b) = (a × b) · a = 0. which is valid even when a and b do not commute.20) (12. 2E (12. we will compute the hydrogen spectrum using the additional symmetry. (12.15) which can be proved after very considerable algebra.16) Through some algebra [4] the following commutation relations can be verified [Ki .10) (12. 2m r (12. in the subspace of the Hilbert space corresponding to a certain energy we can replace H by E. ijk Jk .17) (12. Therefore. (12.3. which have a negative energy E.13) It can be verified explicitly that its components commute with the Hamiltonian. (12. Jj ] = i [Ki . 2m r 1 x3 (J2 p1 − J1 p2 + i p3 ) − k . For this purpose we first note that J · = 0. 2 1 B = (J − K).

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 (12.34) It follows that l has to have values in {0 = |a − b|. . . Chaos is described classically as exponential . . l. 1. . . 1. .23) (12. ijk Bk .24) (12. This implies that a = b. . can be seen to follow from the triangle inequality as applied to J = A + B.21) and (12. . 2 (12. a = 0. .31) where we have used (12. a = 0.376 [Bi . Using this equation the energy eigenvalues can be written in terms of the eigenvalues of A2 and B 2 operators. Furthermore the bound on the orbital angular momentum.30) 2 2 2 1 . 1.28). which is identical to the the direct sum of the Lie algebra of of two SO(3) (or SU (2)) algebras.26) (12. a lack of symmetry implies a lack of degeneracy in the energy levels of a quantum mechanical system.33) (12. a + b = n − 1}. .22): A2 = a(a + 1) B 2 = b(b + 1) Following (12. . (12. H = 0. .29) (12.22) (12. Bj ] = 0. b = 0.25) So far we have shown that the symmetry generators form an algebra. H B. . By comparing to the rotation algebra introduced in chapter 5.32) A comparison with (12. . 2 1 . 1. 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. namely that J J > < A − B =0 A + B =2 A (12. In contrast to the discussion above about extra symmetries.14) we note that A2 − B 2 = J · = 0. Bj ] = i [Ai . Spinor Formulation (12. the energy eigenvalues are found to be E=− mk 2 1 . Noticing that A2 and B 2 have the same eigenvalues because of (12.7) tells us that (2a + 1) = n. . . 2 (2a + 1)2 2 2 (12.27) (12.15). . A. The most extreme case of this is the quantum analogue of a classically chaotic system. = 0. .

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

albeit it is a good approximate one.39) All other isomultiplets. as good quantum numbers. . −1 = = = 1 1 1 1 . 1 π 0 = |1. (12. spelling disaster for the stability of matter. I3 = 2 2 1 1 . including the spin and color quantum numbers of their constituent quarks.38) are ud. 2. u¯ and d¯ states. 0 . we refer the reader to [1]. the fundamental representation of the isospin symmetry corresponds to the doublet that contains the so-called up and down quarks u= 1 1 .6M eV /c2 . π − = |1. Therefore it remains a useful concept. as we shall see below. n = I = . the proton and the neutron can be written as totally symmetric uud and udd states.− . (12. The mass difference between the neutron and the proton could then be attributed to the charge content of the latter.0M eV /c2 ). including the proton and the neutron. are made up of these two quarks. They can be constructed with the same rules that have been used for angular momentum addition ¯ u in chapter 6.− 2 2 . Shortly we 1 will see how to describe both the pion and nucleon states as composites of more fundamental I = 2 states. mπ0 = 135. d= 1 1 . Further than that.− 2 2 . 2 2 1 1 1 . We can describe isospin multiplets the same way we have described the angular momentum and spin multiplets. . Isospin symmetry is not an exact symmetry of strong interactions. Denoting the total isospin. For example.43) Similarly.42) (12.40) + 2 1 1 . If the mass difference (or the energy difference) were to be to be purely electrostatic in nature. They u form an isotriplet: π + = |1. If it were otherwise the proton would be unstable by decaying into the neutron. the proton is the lighter of the two. sec. 1 we can re-write (12. 2 2 1 2 2 2 1 1 1 1 1 √ .− . π 0 = |1. I3 = − 2 2 .11. it can be seen as part of a larger (and more approximate) symmetry which is of great utility to classify observed particle families. For a precise description of the two nucleons as composite states. (12. (mπ± = 139. and its third component.37) As an example of a multiplet with I = 1 we have the three pions or π-mesons π + = |1. I. However. 2 (12. −1 . I3 .38) which have all nearly identical masses.− 2 2 1 2 2 2 1 1 1 1 . 2 2 . the three pions in (12. respectively. 2 2 1 2 2 2 (12. 0 π − = |1. In the framework of the quark of model.36) as a multiplet with I = 2 1 1 p = I = . 1 .41) (12. . the proton had to be heavier. .378 Spinor Formulation beta decay).

by accidental convention. a hydrogen isotope. 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).or flavors as they are commonly called. 0 ). a number strange particles have been found which presumably contained a third quark species: the strange quark. The possibilities are that of an isosinglet 1 |0.45) Is the deuteron an isosinglet state or an isotriplet? If it were an isotriplet (|1.37) and in analogy to (12. All other composite states have their strangeness given by the sum of the strangeness content of their constituents. −1 = |p > |p >.48) (12. baryon and meson.43). whereas mesons are q q states. Therefore it has to have isospin. mesons have intermediate mass ranges. the pions being examples thereof. I3 = 0. By convention ¯ baryons have baryon number 1. but at the time the empirical concepts like isospin and strangeness quantum numbers were in use. mesons have baryon number 0.47) (12. strongly interacting particles) which is about a GeV /c2 . baryons generally are heavy. Following (12. This is why isospin is such a good symmetry and why isomultiplets have nearly identical masses. we shall setup some terminology: baryons are qqq states.11. It shall be noted here that the quark model was not invented until 1960’s. The reader may know that the deuteron. All antiparticles have their 3 quantum numbers reversed. 2 = |n > |n > . but such states do not exist.e. such as proton and the neutron.) are light. the neutrinos etc. The names. . Therefore the deuteron has to be an isosinglet state (|0. 1 |1. consists of a proton and a neutron. (12. 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. where leptons (electron. The up and down quarks have strangeness zero. to be −1 for the strange quark. (12. Before proceeding further. and quarks have baryon number 1 . In the late 1940’s and early 1950’s. Naturally. Now let us consider another example of combining the isospins of two particles.46) (12. 0 ) we should have seen nn and pp bound states of comparable energy in nature (because of isospin symmetry). 0 |1. If taken literally. this will be mathematically identical to adding two spins. We will now try to describe its wave function in terms of its constituent nucleons. 1 = √ (|p > |n > +|n > |p >). 0 = √ (|p > |n > −|n > |p >) 2 and that of an isotriplet |1. The value of the strangeness quantum number is taken. this remains only an inaccurate naming convention today. the up and down quarks are not the only quark ‘species’ .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. as some mesons discovered later are heavier than some baryons and so on. muon. originally refer to the relative weight of particles.44) 2 where B is the baryon number and S is the strangeness. As it later turned out.

. (f ) (i) (I (f ) I3 |00I (i) I3 ) is a Clebsch-Gordon coefficient and I (f ) . where M is the scattering amplitude given by M = final | α I(1) · I(2) | initial . Show that the expectation of I(1) · I(2) is state. 0 ). We will eventually be able to compute ratios of scattering cross-sections between different processes. 0 + |0. √ According to (12.54) Here T00 ≡ V = α I(1) · I(2) . γ (f ) ||T ||I (i) . 1 4 in an isotriplet state and − 3 in an isosinglet 4 Using (12. The cross-section. is proportional to |M|2 . Consider the generalization of tensor operators discussed in section 6. Show that I(1) ·I(2) is an ‘isotensor’ of rank zero. γ (i) = (f ) (i) (I (f ) I3 |00I (i) I3 ) (f ) (i) (−1)I −I (f ) (i) (12. as discussed in the exercise above.52) where γ (i) and γ (f ) denote degrees of freedom other than isospin. respectively.259) discussed in detail in sections 6. such as the spatial dependence of the wave function and spin. γ (i) .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. Exercise. The dot product here refers to the abstract isospin space.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 .45) and (12.7 to the case of isospin. σ.7. Exercise. 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 ) = 1.53) 0 √ 1 2I (i) +1 I (f ) . I3 . γ (f ) (f ) (i) (12. For completeness let us start by restating the Wigner-Eckart theorem (6. I3 = 1. I3 .51) (12. 1 and (1/ 2)(|1. (12. For example. γ (i) (i) (12.50) .48) the initial states in (I) and (II) have isospin values |1. consider (I) p + p → d + π + (II) p + n → d + π 0 (12. γ (f ) T00 I (i) = 1. γ (f ) |T00 |I (i) I3 .8. We will now employ the (isospin analogue) of the Wigner-Eckart theorem (6. I3 (f ) = 1.55) .5 for the spin-analogue of the same problem.49) The only assumption that we put in will be that the interaction is of the form V = α I(1) · I(2) .8 to compute the ratio of the scattering amplitudes MI and MII . which is an isoscalar. (12. γ (i) is a reduced matrix element defined in the same sense as in section 6. 0 . γ (f ) ||T0 ||I (i) . γ (i) I (f ) . The initial and final states can be denoted in more detail as | initial | final = = I (i) . 1 and |1. Refer to exercise 6.7 and 6.

60) (12. [2] As a further example. which is also manifested by a vanishing Clebsch-Gordon prefactor. −2 1 1 √ . I3 = 0. γ (f ) ||T00 || I (i) = 1. we will consider pion-nucleon scattering.62) We can now write the ratio of the scattering amplitudes: MI 1 √ . γ (i) = √ 2 1 (f ) (i) +√ I (f ) = 1. Once again we need the isospins for the initial and final states. (12. −1 π0 + n : |1.59) (12. (12. γ (f ) T00 I (i) = 0. I3 = 0. involving. 2 = 2. −1 2 3 2 − + . σ( π − + p → anything ) The possibilities are (a) π + + p → π + + p . γ (i) 3 +0. I3 = 0.58) (12. (12. γ (i) 3 1 (f ) (i) I (f ) = 1.11. γ (f ) T00 I (i) = 1. γ (i) 2 1 = (10|0010) √ I (f ) = 1. The relevant Clebsch-Gordon coefficients are easily evaluated: (11|0011) = (10|0010) = 1. but these are not dominant at relatively low energies. I3 = 0.57) (12. σII which is in approximate agreement with the observed ratio. .64) There are more exotic possibilities.56) (12. (b) π − + p → π − + p . = 381 (12. 1 1 1 1 3 √ 2. −2 1 1 2. for example. particles with strangeness. which are obtained by a standard Clebsch-Gordan expansion π + + p : |1.5: Lorentz Invariant Field Equations 1 (11|0011) √ I (f ) = 1. γ (f ) ||T00 || I (i) = 1. (c) π − + p → π 0 + n . 1 π − + p : |1.61) MII Note that the second term in MII vanishes due to the isospin conservation.65) (12.66) = 2 3 1 3 2. −2 .63) where common dynamical factors (which would not be as easy to compute) have dropped out thanks to the Wigner-Eckart theorem. = MII (1/ 2) (12. 2 = 3 2. It follows σI = 2. 2 . −2 2 1 1 3 2. We want to compute the ratio of total cross-sections assuming a similar interaction as in the previous example σ( π + + p → anything ) . 0 1 1 3 3 2.

called strangeness.67) which are independent of m.3 The Eightfold Way and the flavor SU (3) symmetry The discovery of the concept of strangeness. which can be produced by π − + p → K + + Σ− . For instance.68) − 2 3 M1 2 Guided by empirical data we will further assume that M 3 >> M 1 . was motivated by the existence of particles that are produced strongly but decay only weakly. m 1 2. The classification of the newly found particles as members of some higher multiplet structure was less obvious then the case of isospin. for example.70) (12. making an identification of the underlying symmetry and the multiplet structure less straightforward.71) u =  0  . are up to 40% heavier. mentioned in the previous section. S(Σ− ) = −1 and S(π) = S(N ) = 0. Hence the strangeness changing strong decays of K + (or Σ− ) were forbidden. m . 0 0 1 . S. It was assumed that strong interactions conserved S (at least approximately). such that S(K + ) = 1. m V V 3 2. Strange partners of the familiar nucleons. while weak interactions did not. Hence Gell-Mann and independently Nishijima postulated the existence of a separate quantum number. . it appears an obvious generalization to add another component for an extra quark to the isospin vector space       1 0 0 (12.69) 12. 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 (12.382 Spinor Formulation As in the example of nucleon-meson scattering we define the relevant matrix elements M3 = 2 M1 = 2 3 2. (12. etc. m 1 2. K + . which leads to the following 2 2 ratios for the cross-sections σa : σb : σc = 9 : 1 : 2. [2] (12. however. 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. A computation similar to the previous example of nucleon-nucleon scattering yields (apart from common prefactors) the following amplitudes for the reactions in (12.s =  0  .d =  1  . giving it a higher than usual lifetime. has a lifetime which is comparable to that of π + albeit being more than three times heavier.

of a. bottom (beauty) and top (truth) are significantly heavier than the hadronic energy scale. becomes increasingly inaccurate for Nf > 3. for a period they were considered as useful bookkeeping devices without physical content. However. should correspond to. while preserving the norm. The reader shall note that most of what is being said trivially generalizes to other unitary groups. The quarks posses another quantum number. (The reader is referred to section 8.74) for any two unitary matrices U and V . Hence the group U (3) has 9 dimensions. The bottom and top are even heavier. (The ‘bare’ mass of the charm quark is already heavier than the two nucleons. SU (N ). Given a complex vector. Lie algebras and related concepts. This group of 3 × 3 unitary matrices is denoted by U (3). giving rise to the quark model.11. ak . it is the group .3 for a brief discussion of gauge transformations). unitary relation k (12. but we will stick to N = 3 in the following.72) a∗ ak . we observe that (U V )† = V † U † = V −1 U −1 = (U V )−1 . have to be elements of the group SU (3) (which we will investigate closely very soon). In many respects it will resemble the more familiar group SU (2) discussed in some detail in chapter 5. we want to find those transformations ak → Ukl al that preserve the norm. history followed the reverse of this path. Then came the question of what the fundamental representation. It is seen that such a matrix U has to satisfy the following k U † = U −1 . Flavor SU (Nf ) symmetry on the other hand. of three dimensions. we will establish some mathematical preliminaries about the group SU (3). However SU (3) symmetry appears in another and much more fundamental context in strong interaction physics. which again form representations of an SU (3) group.71). In this perspective the flavor SU (3) symmetry may appear to be mainly of historical interest. The unitarity relation imposes 9 constraints on the total of 18 real degrees of freedom of a 3 × 3 complex matrix. [2] ) Before discussing the significance and the physical implications of the quark model. The reader is also invited to revisit section 5. called quantum chromodynamics. where Nf is the number quark flavors. but there are a number of subtle differences.71) into each other. the same way nucleons and pions were identified as representations of the isospin SU (2) symmetry. called color. (12.1 whenever necessary. as in (12. namely charm. The reason is that the other known quarks. Since arbitrary phase factors are of no physical interest.73) To verify that all such matrices form a group. As quarks were never directly observed. Because of this additional constraint SU (3) has 8 dimensions. in reference to Lie groups. still leaves the norm invariant. This is believed to be an exact symmetry of strong interactions. (12. Multiplying U by a phase. which set the hadronic energy scale. eiφ . in fact modern theory of strong interactions is a ‘gauge theory’ of this color group. 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.5: Lorentz Invariant Field Equations 383 In this case the transformations that ‘rotate’ the components of (12. First particle multiplets were identified as representations of the SU (3) group.

−2 λ4 λ6 λ8 The generators.384 Spinor Formulation SU (3) and not U (3) that is of main interest. 0 0 0   . λk .77)  λ1  .75) αk λk (12.7.79) (12. The reader is invited to compare the structure of SU (3) to that of SU (2) discussed in section 5.79). (12.1. As discussed in section 5. since det(eA ) = etrA . λ3 =  0 −1 0  . λ7 =   0 0  . 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  . The Lie algebra structure is given by the commutators of λk [λj . (12. As in the case of SU (2) higher dimensional representations obeying the same structure can be found.80) This is the fundamental or defining representation of SU (3). any unitary matrix. obey the following relation tr(λj λk ) = 2δjk .78) where fjkl are the antisymmetric structure constants similar to the jkl of SU (2) given in (5. Therefore we will express elements of SU (3) as U = ei k (12. the meaning being apparent from the context.) The unit determinant condition requires that all λk are traceless. The fundamental relation to be preserved is (12. λk ]+ = δjk + 2δjkl λl . can be written in the form U = eiH where H is a hermitian matrix. which can be verified explicitly as matrix identities. λ2 =     .   . λ5 =   .32). . δjkl . We can also introduce the constants. 3 (12. regardless of the dimension of the representation.76) where λk are 8 linearly independent matrices forming the basis of the Lie algebra of SU (3). U . λk ] = 2ifjkl λl . (We shall at times refer to the Lie algebra with the name of the group. via the anticommutator relation 4 [λj .

89) (12. The generators of SU (3) can be arranged into a very similar form to that of SU (2). (12. We first introduce the F-spin operators 1 Fi = λ i .87) Exercise.93) (12. T0 } and [Y. (12. In the case of SU (2) the identification of ‘ladder’ operators {J+ . 2 1 [T0 . 2 (12. 2 (12. V± ] = ± V± . 1 [T0 . [Y.87) the commutation relations between the generators can be expressed in a succinct manner.85) (12. U± = F6 ± iF7 .5: Lorentz Invariant Field Equations 385 As it turns out the set of generators in (12. First. T0 ] = 0. Exercise. starting with the observation that Y = B + S. 2 Y = √ F8 .94) .86) (12.91) [T0 . V± ] = ±V± .90) (12. In the basis (12.71) show that T0 is the isospin operator.88) which defines (not uniquely) a maximal set of mutually commuting operators {Y.81) In chapter 5. J± ] = ± J± . we have [Y. T 0 = F3 . V± = F4 ± iF5 .84) (12.77) is not the most useful basis in the study of SU (3). U± ] = ±U± .44).83) (12. 3 (12. Using the convention in (12. T± ] = 0. [Y. U± ] = U± . T± ] = ±T± . (Recall that the strange quark has S = −1 by convention). I3 . (12.82) With another change of basis we arrive at the ‘standard’ form of the generators of the Lie algebra of SU (3) T± = F1 ± iF2 . Derive the Gell-Mann–Nishijima relation (12. these relations have been used to construct the angular momentum spectrum as well as the function space representation of the rotation algebra. namely spherical harmonics. Y is called the hypercharge.11.92) (12. which satisfied an ‘eigenvalue equation’ [J0 . J− } proved useful.

we have [T+ .104) (12.5. In the case of SU (2) the representations lay on a line on the J0 axis. Another important property of SU (2) is the existence of an operator (namely the total angular momentum.88). [T+ . U+ ] = V+ .102) (12. V− ] = T− . [T+ . SU (2) has rank 1. V+ ] = 0. the representations will now lie in a T0 − Y -plane.99) (12. U− ] = 0.386 Spinor Formulation which relate the remaining generators {T± . The interested reader is instead referred to a very readable account given in chapter 12 of [4].95) (12. we will use these commutation relations to construct representations of SU (3). Casimir operators can be used to label irreducible representations of the Lie algebra. T0 }. The same way the angular momentum ladder operators have been used to construct the representations of SU (2). 3 [U+ . it is said to be in Cartan-Weyl form. C1 = k λ2 = − k 2i 3 fjkl λj λk λl . However.105) C2 = jkl djkl λj λk λl In general the number of independent Casimir operators of a Lie group is equal to its rank.98) (12. T− ] = 2T0 . . U− ] = Y − T0 = 2U0 . U± } to this maximal set by ‘eigenvalue equations’ and [T+ . V+ ] = 0. as the relation between the states in a given representation can be conveniently expressed in terms of ladder operators. U0 and V0 are linear combinations of T0 and Y . J 2 ) which commutes with all of the generators. since there are two mutually commuting generators in SU (3) as given in (12. Finally.96) (12. This form is essential for easy labeling of the representations of the group. The maximum number of mutually commuting generators of a Lie algebra is called its rank. Thus.100) (12. [U+ . jkl (12. Note that. A formal definition and a detailed discussion of the Cartan-Weyl form is beyond the scope of this chapter. We can construct two such independent Casimir operators for the group SU (3). [U+ . 2 3 Y + T0 = 2V0 . An operator which commutes with all generators of a Lie group is called a Casimir operator. [V+ .97) which relate commutators of generators with opposite eigenvalues to the maximal set {Y. When the basis of a Lie algebra is expressed in such a way to satisfy the form of the eigenvalue relations as given above. (12.101) (12. As in the case of J 2 and SU (2). V± . [T+ . V− ] = 2 (12. similar to the way it was done in section 5. while SU (3) has rank 2.103) Any remaining commutators follow from hermiticity. V− ] = −U− .

y U± |t3 .5 how to label the irreducible representations of the angular momentum algebra SU (2) in terms of the value of the total angular momentum. 1) as an example. it was shown in section 5. y . y . 1) correspond to the triplets of quarks and antiquarks.1). The interested reader is referred to [4].110) (12.112) The effect of these operators to the states in the y − t3 plane have been outlined in Fig. A notation suggestive of the dimensionality of the representation can be used to identify representations.2) shows the representation D(2. |t3 .88) we can label states by the eigenvalues of T0 and Y operators. 2 1 = γ t3 .y ± 1 . The representations of SU (3) are constructed analogous to those of SU (2) by identifying the ‘boundary’ states annihilated by raising (or lowering) operators. y . This property can be used to label representations in terms of the values of the Casimir operators. As another example for the representations of SU (3). y V± |t3 . especially chapters 7 and 8.) All other representations can be constructed by combining these two conjugate representations. All other states of the representation are then constructed by successive application of ladder operators T± .109) The same way that J± |m is proportional to |m ± 1 in the case of the angular momentum algebra. y : T0 |t3 . 2 1 = β t3 ± . (See Fig. q) and it has a dimensionality of 2 (p + 1)(q + 1)(p + q + 2). 0) and D(0. (12.111) (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. y 1 = α t3 ± . . U± . y . The details of this procedure is beyond the scope of this chapter.11. 2 (12. The representations for hexagons with sides of length p and q in the T0 − Y -plane. y V0 |t3 . y ± 1 .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.108) (12. 2 1 t3 ) |t3 . we have U0 |t3 .3). For example the pion octet (or any other meson octet) is therefore realized as states of a quark .antiquark pair. V± . For example. the pion family forms part of an octet corresponding to the D(1. (12. we have T± |t3 . Such a representation is labeled as 1 D(p. 1) representation. Figure (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 . Because of (12. Now we will construct explicit representations of SU (3). respectively.106) (12. y . = y |t3 . y = t3 |t3 . y Y |t3 . y 3 = ( y− 4 3 = ( y+ 4 1 t3 ) |t3 . For example. y . The representations D(1. (12. For example. 2 (12.

1) representation of SU (3).388 £ Spinor Formulation Figure 12. £ Figure 12.1: The effect of SU (3) ladder operators on the y − t3 plane.2: D(2. The states in the inner triangle are doubly degenerate. ¡ ¢  ¡ ¢  8 6 2 '745301) ¥¦¤   ¤   #"  ! (& % '$   © ¨     ¦ §¥ ©§ ¨ .

. 2 (12.115) (12. where we would write [3] ⊗ [3] = [1] ⊕ [3] ⊕ [5]. for example.113) The additional singlet state corresponds to the η meson.118) ¡ ¢    ¨ ©§ ¨¥¦¤ "   ! . 3]. 0) or the fundamental representation of flavor SU (3) symmetry. 0) and D(0.116) (12. This way the quark-antiquark states can be represented as follows [3] ⊗ [¯ = [8] ⊕ [1] 3] (12.117) The Gell-Mann–Nishijima relation can then be succinctly expressed as 1 Q = y + t3 . in the case of SU (2). The three quarks in the fundamental representation can now be written as u = d = s = 1 1 . − . 1) are denoted by [3] and [¯ respectively.3: D(1. 2 3 2 0. 2 3 1 1 − .114) Figure 12. to the case of adding two spin triplet states. 1) is written as [8] etc. The octet D(1. 3 (12. £ (12. This expansion can be compared.5: Lorentz Invariant Field Equations 389 D(1. .11.

G. Oxford. Wiley. 133. K. [2] Introduction to elementary particles. 3 1 Qd = − . Li. 3 1 Qs = − . 3 £ Spinor Formulation (12.120) (12. D. Cheng. D. A. and spectroscopy. Wiley. 1989. T. 1987. North holland. Wiley. [6] Gauge theory of elementary particle physics. J. Greiner. L. Martin. Eisenberg. [7] Expansion method for stationary states of quantum billiards. 1984. L. [4] Quantum mechanics: symmetries. M. Springer-Verlag u [5] Principles of symmetry. F. B. ¡ ¢  S - S 0 ¤ ¥ S + . W.121) L 0 X - X 0 Figure 12. 1993. Kaufman. 1) representation of SU (3). p. D. J. References [1] Quarks and leptons: an introductory course in particle physics.-P. 1972. Harter. Schulten. M¨ller. Phys.390 from which the quark charges follow 2 Qu = .4: The baryon octet as a D(1. Kosztin. W. W. I. dynamics. 67 (2). Halzen. Greiner. [3] Microscopic theory of the nucleus.119) (12. Am. 1984. (1999).-F. Griffiths.

Sign up to vote on this title
UsefulNot useful