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

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

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

˙ Theorem: Let S[q(t)] = t0 dx dt t1 ˙ dt L(q(t). · . q(t). the velocity vector dt q(t) and time t. t) + j=1 ∂L ∂L δqj + δ qj ˙ ∂qj ∂ qj ˙ + O( 2 ) (1. We will later often assume that only variations of a trajectory q(t) are permitted for which δq(t0 ) = 0 and δq(t1 ) = 0 holds. t) of the conﬁguration d vector q(t). It holds M M ∂L t1 d ∂L ∂L δS[q(t). t0 (1. i. df = dx dx or.g. q(t). . df = M ∂xj dxj . e. dt q(t). j=1 We will now consider a particular class of functionals S[ ] which are expressed through an integral d over the the interval [t0 .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 .5) where L( · . at the ends of the time interval of the trajectories the variations vanish. For this purpose we consider M ˙ ˙ ˙ L(q(t) + δq(t). We focus on such functionals because they play a central role in the so-called action integrals of Classical Mechanics. i. t) (1.7) through Taylor expansion and identiﬁcation of terms linear in δqj (t).6) immediately. equating these terms with δS[q(t). q(t) + δ q(t).e. ˙ In the following we will often use the notation for velocities and other time derivatives d q(t) = q(t) and dtj = xj . · ) is a function diﬀerentiable in its three arguments. We attempt to evaluate t1 S[q(t) + δq(t)] = t0 ˙ ˙ dt L(q(t) + δq(t). δq(t)]. · ] in conventional diﬀerential calculus does not correspond to a diﬀerentiated function. but rather to a diﬀerential of the function which is simply the df diﬀerentiated function multiplied by the diﬀerential increment of the variable... q(t) + δ q(t). q(t) + δq(t) is a ‘slightly’ diﬀerent trajectory than q(t). t) (1. t1 ] where the integrand is a function L(q(t). (1. δq(t)] = dt − δqj (t) + δqj (t) ∂qj dt ∂ qj ˙ ∂ qj ˙ t0 j=1 j=1 t1 .6) is expressed in terms of ‘normal’ functions.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. t) = L(q(t).2 Lagrangian Mechanics Note: δq(t) describes small variations around the trajectory q(t).6) For a proof we can use conventional diﬀerential calculus since the functional (1. It is also important to appreciate that δS[ · .10) From this follows (1. ∂f in case of a function of M variables.

2. . .12) The proof of this theorem is based on the property Lemma: If for a continuous function f(t) f : [t0 .17) . . For this purpose we deﬁne Deﬁnition: An extremal of a diﬀerentiable functional S[ ] is a function qe (t) with the property δS[qe (t).1.1: Variational Calculus 3 We now consider the question for which functions the functionals of the type (1.5). On the other side. if and only if it satisﬁes the conditions (j = 1. y(x1 ) = y1 . The proof of the above theorem starts from (1. 2. . t1 ]. y(x2 ) = y2 .5) assume extreme values.11) The extremals qe (t) can be identiﬁed through a condition which provides a suitable diﬀerential equation for this purpose. M and δqj (t0 ) = δqj (t1 ) = 0 follows according to (1. δq(t)] = dt − δqj (t) . . M . This condition is stated in the following theorem.12) for j = 1. ∂qj dt ∂ qj ˙ t0 j=1 (1. dx (1. According to the Lemma above follows then (1. hence.16) This property holds for any δqj with δq(t) ∈ F . it holds in case δq(t0 ) = δq(t1 ) = 0 that qe (t) is an extremal.15) We will not provide a proof for this Lemma. then f (t) ≡ 0 on [t0 . Let us assume that the two points are connected by the path y(x). . M ) d dt ∂L ∂ qj ˙ − ∂L = 0 ∂qj (1. the above theorem.12) for j = 1. The length of such path can be determined starting from the fact that the incremental length ds in going from point (x. y(x + dx)) is ds = (dx)2 + ( dy dx)2 = dx dx 1+( dy 2 ) . . . t1 ] ⊂ R → R holds t1 (1.6) which reads in the present case M ∂L t1 d ∂L δS[q(t).16) the property δS[qe (t). δq(t)] = 0 for all δq(t) ∈ F . An Example As an application of the above rules of the variational calculus we like to prove the well-known result that a straight line in R is the shortest connection (geodesics) between two points (x1 . y2 ). from (1. y(x)) to (x + dx. . (1.14) for any continuous function h(t) ∈ F with h(t0 ) = h(t1 ) = 0. 2. .13) dt f (t)h(t) = 0 t0 (1. · ] ≡ 0 and. (1. y1 ) and (x2 . Theorem: Euler–Lagrange Condition For the functional deﬁned through (1.

.21) are extremals of the functional. . obey the Euler–Lagrange dy condition. qM ) . . 1. t) (1. . q(t).22) ˙ where L(q(t).23) Presently we consider only velocity–independent potentials. the so-called action integral. (1. according to the theorems above.. j = 1. dx (1. Threorem: Hamiltonian Principle of Least Action The trajectories q(t) of systems of particles described through the Newtonian equations of motion d ∂U (mj qj ) + ˙ = 0 dt ∂qj . circles whose centers lie at the center of the sphere. 2. (1. hence. Velocity–dependent potentials which describe particles moving in electromagnetic ﬁelds will be considered below. The constants a and b are readily identiﬁed through the conditons y(x1 ) = y1 and y(x2 ) = y2 .19) From this follows y / 1 + (y )2 = const and. t) is the so-called Lagrangian M ˙ L(q(t). dx ) = 1 + (dy/dx)2 .1: Show that the shortest path between two points on a sphere are great circles. . .4 The total path length is then given by the integral x1 Lagrangian Mechanics s = x0 dx 1+( dy 2 ) . t) = j=1 1 mj qj − U (q1 . q2 . . t1 S[q(t)] = t0 ˙ dt L(q(t).5) with L(y(x). Using y = dx the condition reads d dx ∂L ∂y = d dx y 1 + (y )2 = 0. q(t). The shortest path is an extremal of s[y(x)] which must.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. One obtains y 1 − y2 y(x) = (x − x2 ) + y2 . i.20) x1 − x2 Exercise 1. M (1. . q(t). This in turn yields y(x) = ax + b.e. y = const. ˙2 2 (1.1.18) dy s is a functional of y(x) of the type (1.

31) = − V − Gauge Symmetry of the Electromagnetic Field It is well known that the relationship between ﬁelds and potentials (1.31) allows one to transform the potentials without aﬀecting the ﬁelds and without aﬀecting the equations of motion (1. These conditions read in the present case ∂L d ∂L ∂U d − = 0 → − − (mj qj ) = 0 ˙ (1. t) is the Lorentz force.26) (1.23).28) can be satisﬁed implicitly if one represents the ﬁelds through a scalar potential V (r. t) and B(r.32) (1. t) and a vector potential A(r. t) describes the charge density present in the ﬁeld and J(r. The transformation which leaves the ﬁelds invariant is A (r.22. t) = q E(r. t) . 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).29) where ρ(r. t) describes the charge current density. 1. t) as follows B E = ×A 1 ∂A . F (r. t) and B(r.27) (1. x3 (t)) moving in an electromagnetic ﬁeld described through the electrical and magnetic ﬁeld components E(r.30) (1. t) + V (r. dt The ﬁelds E(r. t) = A(r.25) = 0 = 0 4π J c = 4πρ = (1. respectively. t) obey the Maxwell equations ×E + ·B ×B − ·E 1 ∂ c ∂t E 1 ∂ c ∂t B (1. x2 (t).28) (1. t). t) − K(r. t) + v × B(r.25) of a particle moving in the ﬁeld. c ∂t (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 (1.30. 1. Equations (1.24) ∂qj dt ∂ qj ˙ ∂qj dt which are obviously equivalent to the Newtonian equations of motion. t) c ∂t (1.27) and (1.1. t) dt c where dr = v and where F (r. t) = V (r. The equations of motion for such a particle are d q ˙ (mr) = F (r.33) . t) 1∂ K(r.

if one assumes for a particle the Lagrangian ˙ L(r. e.34) For this purpose we consider only one component of the equation of motion (1.37) is ∂L ∂V q = −q + ∂x1 ∂x1 c The ﬁrst term in (1. (1.41) which is identical to the term (1.25) follows from the Hamiltonian Principle of Least Action.37) is d dt ∂L ∂ x1 ˙ = d q dA1 d q (mx1 ) + ˙ = (mx1 ) + ˙ dt c dt dt c ∂A1 ∂A1 ∂A1 x1 + ˙ x2 + ˙ x3 ˙ ∂x1 ∂x2 ∂x3 . t) + A(r.35) − ∂A1 ∂x2 [v × B]1 = x2 B3 − x3 B2 = x2 ˙ ˙ ˙ ∂A2 ∂A1 − ∂x1 ∂x2 − x3 ˙ ∂A1 ∂A3 − ∂x3 ∂x1 .36) in the Newtonian equation of motion. (1.36) This expression allows us to show that (1.6 Lagrangian of Particle Moving in Electromagnetic Field Lagrangian Mechanics We want to show now that the equation of motion (1. namely. dt ∂x1 c We notice using (1. 1.35) shows that the Newtonian equations of motion and the Euler–Lagrange conditions are. r. t) · v . ∂x1 (1.35) is equivalent to the Euler–Lagrange condition d dt The second term in (1.30).25).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. in fact.g. B3 = ∂A2 ∂x1 (1.40.38. (1.38) ∂L ∂ x1 ˙ − ∂L = 0. 1. t) = 1 q mv 2 − q V (r.. d ∂V q (mv1 ) = F1 = −q + [v × B]1 . Comparing then (1.39) ∂A1 ∂A2 ∂A3 x1 + ˙ x2 + ˙ x3 ˙ ∂x1 ∂x1 ∂x1 .40) (1. 2 c (1.37) The results (1. equivalent.41) with (1. .

dt ∂x1 ∂x2 ∂x3 ∂t ∂t (1.33) of electromagnetic potentials. q. t) .e. This term is d ∂K ∂K ∂K ∂K ∂K K(r. t) − + 2 c ∂t c 1 q = mv 2 − q V (r. t) + A(r. t) c t1 t0 t1 t0 ˙ dtL(q. 2 c = A(r.32. t) + A (r. encountered above in connection with the transformations (1. surprisingly simple fashion in Lagrangian Mechanics. For this purpose we consider the transformation of the single particle Lagrangian (1. t) · v . 1. t) · v + K(r.46) (1.3: Symmetry Properties 7 1.44) implies that the gauge transformation amounts to adding a constant term to the action integral. They are the subject of the following theorem. t) = q K(q.32. We want to demonstrate now that the transformation (1.42) is. r. q. t) + q K(q. and since these properties allow one often to simplify mathematical descriptions. t) 1 1 ∂K q mv 2 − q V (r. t) + dt c (1.45) . t) the Lagrangian.33) of electyromagnetic potentials. a term not aﬀected by the variations allowed. i. q. in fact. One can conclude then immediately that any extremal of S [q(t)] is also an extremal of S[q(t)]. r. (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. Eq. 2 c c dt Inserting (1. q. equivalent to the gauge transformation (1.32. appear in a diﬀerent.33) ˙ L (r.44) Since the condition δq(t0 ) = δq(t1 ) = 0 holds for the variational functions of Lagrangian Mechanics. We will consider in the following two symmetries. t) = mv 2 − q V (r. t) = x1 + ˙ x2 + ˙ x3 + ˙ = ( K) · v + . gauge symmetry and symmetries with respect to spatial transformations. t) L (q.42) This transformation is termed gauge transformation. Note that one adds the total time derivative of a function K(r.34) q 1 q d ˙ L (r. The factor q has been introduced to make this c transformation equivalent to the gauge transformation (1.3 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 gauge symmetry.45) and reordering terms yields using (1. t) + K ·v (1.33) of electromagnetic potentials. 1.43) into (1. t) = L(q.32. t) c t1 t0 = S[q(t)] + (1. 1. 1.43) To prove this theorem we determine the action integral corresponding to the transformed Lagrangian t1 S [q(t)] = t0 ˙ dtL (q.1.

which has an important analogy in Quantum Mechanics..32. r ∈ R . ˆ (1.48) In the following we will denote unit vectors as a. 1. Deﬁnition: Inﬁnitesimal One-Parameter Coordinate Transformations A one-parameter coordinate transformation is decribed through r = r (r. r.42) and (1. we like to ˆ provide the transformation here anyway for later reference r = r + e. t). A (r. 0) = r . i. ) . ˆ (1. but nevertheless instructive way considering rotations around the x3 –axis. The following description of spatial symmetry is important in two respects. for example. t) by gauge transformed potentials V (r. R(r) = through ∂r ∂ (1. in the case of central force ﬁelds which are invariant with respect to rotations of coordinates around the center. the transformation (1. However. 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 corresponding inﬁnitesimal transformation is deﬁned for small r (r. t).42) corresponds to replacing in the Lagrangian potentials V (r. and for the introduction of inﬁnitesimal transformations which will provide a crucial method for the study of symmetry in Quantum Mechanics. We have proven. ) = r + R(r) + O( 2 ) . A(r.51) We would like to derive this transformation in a somewhat complicated. In case of a translation transformation in the direction e nothing new is gained.8 Lagrangian Mechanics Obviously. for the connection between invariance properties and constants of motion. t). The transformations considered are speciﬁed in the following deﬁnition. therefore.50) A non-trivial example is furnished by the inﬁnitesimal rotation around axis e ˆ r = r + e×r . The transformations we consider are the most simple kind.33). the equivalence of (1.49) =0 ∈ R (1. where the origin of is chosen such that r (r.e. ˆ ˆ ˆ Examples of Inﬁnitesimal Transformations The beauty of inﬁnitesimal transformations is that they can be stated in a very simple manner. for such vectors holds a · a = 1. We consider now invariance properties connected with coordinate transformations.47) (1.52) x3 0 0 1 x3 . In this case the transformation can be written in matrix form cos −sin 0 x1 x1 x2 = sin cos 0 x2 (1.

a classical mechanical system is invariant with respect to a coordinate transformation ˙ a constant of motion exists. i. In order to prove Noether’s theorem we note qj qj ˙ = qj + = qj + ˙ k=1 Qj (q) M (1.58) From this follows the statement of the theorem. i.1.51) reads r = r − x2 e1 + x1 e2 ˆ ˆ (1. the property holds for any system.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) (1.57) must cancel each other or both vanish.k=1 ∂L ∂Qj qk ˙ ∂ qj ∂qk ˙ (1.54) which is equivalent to (1. q. t) yields after Taylor ˙ 2 ).55) (1. (1. t) = L(q. however. The property has been shown to hold in a more general context. . 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. ˙ We have generalized in this theorem the deﬁnition of inﬁnitesimal coordinate transformation to M –dimensional vectors q. We have used here the notation corresponding to single particle motion. Theorem: Noether’s Theorem ˙ If L(q. t) + j=1 ∂L Qj + ∂qj M j. ˙ ∂qk ˙ Inserting these inﬁnitesimal changes of qj and qj into the Lagrangian L(q. Invariance implies L = L.e. by Noether. the second and third term ˙ k=1 in (1. Anytime. neglecting terms of order O( M ˙ ˙ L (q. We consider here only the ‘particle version’ of the theorem. namely for ﬁelds rather than only for particle motion. then M ∂L j=1 Qj ∂ xj is a constant of motion. 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. a quantity C(r. r) which is constant along the classical path of the system.3: Symmetry Properties 9 In case of small this transformation can be written neglecting terms O( 2 ) using cos = 1 + O( 2 ). q.. sin = + O( 2 ) 0 − 0 x1 x1 x1 x2 = x2 + 0 0 x2 + O( 2 ) . q.57) ∂ d where we used dt Qj = M ( ∂qk Qj )qk .53). t) is invariant with respect to an inﬁnitesimal transformation q = q + Q(q).e.56) ∂Qj qk .57) M Qj j=1 d dt ∂L ∂ qj ˙ + ∂L d Qj ∂ qj dt ˙ = d dt M Qj j=1 ∂L = 0 ∂ qj ˙ (1. In the context of Lagrangian Mechanics this implies that such transformation leaves the Lagrangian of the system unchanged. q.. expansion. Using the fact.

59) We obtain the well known result that in this case the momentum in the direction.59) yields the constant of motion ˆ e (ˆ × r) · mv. of course. not inﬁnitesimal transformations. . 3) 3 Qj j=1 ∂L = e· ˆ ∂ xj ˙ 3 ej mxj = e · mv . hence. It was. for which translational invariance holds. i. v) = 1 mv 2 − U (r).e. j = 1. It is not necessary to investigate the consequences of global. A calculation similar to that in (1. 2.50).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. except using now Qj = ej · (ˆ × r). ˆ The important result to be remembered for later considerations of symmetry transformations in the context of Quantum Mechanics is that it is suﬃcient to know the consequences of inﬁnitesimal transformations to predict the symmetry properties of Classical Mechanics. We will now investigate the consequence of rotational invariance as described according to the inﬁnitesimal transformation (1. 2 We ﬁrst determine the constant of motion in case of invariance with respect to translations as deﬁned in (1. Noether’s theorem yields ˆ ˆ the constant of motion (qj = xj . to be expected that this is the constant of motion. ˆ ˙ ˆ j=1 (1. In this case we have Qj = ej · e.51).59). 2. j = 1. 3 and. is conserved. 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. In this case we will use the same notation as in (1.

is unity. t is described in Quantum Mechanics. t) ψ : R3 ⊗ R → C. t is |ψ(r. 1. 2. t)ψ(r. t)|2 = ψ(r. The probability that the particle is detected at space–time point r. At present one may think of f (r) as a sum over δ–functions which represent a multi–slit screen. t) is normalized d3 r |ψ(r. t)|2 Ω (2.4) a condition which enforces that the probability of ﬁnding the particle somewhere in Ω at any particular time t in an interval [t0 . t). The wave function ψ(r.2) (2. We provide now a set of formal rules which state how the probability to observe such a particle at some space–time point r.2 Axioms for Quantum Mechanical Description of Single Particle ˙ Let us consider a particle which is described by a Lagrangian L(r. t ∈ [t0 . r. 4. The particle is described by a wave function ψ(r.1) (2. The probability to detect the particle with a detector of sensitivity f (r) is d3 r f (r) |ψ(r. 11 . t1 ] in which the particle is known to exist. t1 ] . (2. 3.1 The Double Slit Experiment Will be supplied at later date 2.3) where Ω is the space volume in which the particle can exist. placed into the space at some particular time and with a detector behind each slit.Chapter 2 Quantum Mechanical Path Integral 2. t) where z is the conjugate complex of z. t)|2 = 1 Ω ∀t.

t ) ψ(r . i. 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 deﬁnition will actually assume an inﬁnite number of intermediate times and express the path integral through integrals of the type (2. The state of a system at time t. . The generalization of the completeness property to N − 1 intermediate points t > tN −1 > tN −2 > . t ) Ωd 3 r |ψ(r. ψ(r0 .e. t1 ]) Ωd Ωd 3r Ωd 3r Ωd 3r 3 r |ψ(r. t0 ) . t ) which requires d3 r φ(r. at any intermediate time the particle needs only be known at one space point.12 Quantum Mechanical Path Integral 5. The time evolution of ψ(r. t|r . (2. t0 ) = δ(r − r0 ). tN −1 |rN −2 . t ∈ (2. . t) = Ω d3 r φ(r. t|r0 . t ) φ(r. t|r .9) for N → ∞. described by ψ(r. requires then according to (2. t = )|2 = 1 . the path integral r(tN )=rN d[r(t)] · · · r(t0 )=r0 (2. t1 ] yields an expression of the form r(tN )=rN φ(r. hence. t ∈ [t0 . t1 ] (2. t). t|r . Since (2. t)|2 = (2. t. t0 ) = Ωd 3r N −1 Ωd 3r N −2 ··· Ωd 3r 1 φ(r. The following so-called completeness relationship holds for the propagator (t > t [t0 . t0 ) = φ(r.6) φ(r. t1 |r0 . t0 ) = r(t0 )=r0 d[r(t)] Φ[r(t)] . This symbol will be deﬁned further below.8) a knowledge of the state at all space points r ∈ Ω at some intermediate time t . t|r .5) 6.11) which denotes an integral over all paths r(t) with end points r(t0 ) = r0 and r(tN ) = rN . t ∈ [t0 . t ) t > t . . (2. (t > t . t ) = δ(r − r ) Ω (2. t) is described by a linear map of the type ψ(r. t|r . t ) φ(r . t|r . t |r0 . namely the point on the classical path at time t .. t| rN −1 .8) 7.9) Employing a continuum of intermediate times t ∈ [t0 . t|r0 . t )φ(r.10) We have introduced here a new symbol. t1 ]) d3 r φ(r. > t1 > t0 is φ(r. tN −2 ) · · · φ(r1 .7) t. t ) ψ(r . the propagator is unitary.. This is diﬀerent from the classical situation where the particle follows a discrete path and. This must hold for any ψ(r .e. t|r0 . 8. t.4) holds for all times. t ) ψ(r . i. tN −1 ) φ(rN −1 .

t) is the Lagrangian of the classical particle. r. Since this number is multiplied by ‘i’. t) (2. Only for paths close to the classical path is such interference ruled out. However. The situation is very diﬀerent for microscopic particles. However. expressions (2. To show this we consider the value of the action integral for a particle of mass m = 1 g moving over a distance of 1 cm/s in a time period of 1 s. not only for the classical path. it is comparable to action integrals as they arise for microscopic particles under typical circumstances. i..e. i. 2.0545 · 10−27 erg s .2. r.12) S[r(t)] = t0 ˙ dt L(r. Situations which are well described classically will be distinguished through the property that the classical path gives the dominant.12. 2.14) ˙ In (2. 2 2 (2. for the electron many paths contribute and the action integrals for non-classical paths need to be known. Its value is extremely small in comparision with typical values for action integrals of macroscopic particles. namely due to the property of the classical path to be an extremal of the action integral. The value of S[r(t)] is then Scl = 1 1 m v 2 t = erg s .2: Axioms 9. 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. such that destructive interference of the contributions of such paths does not occur. The constant given in (2. interferences should be much less dramatic than in case of the macroscopic particle. The functional Φ[r(t)] in (2. the exponent is a very large imaginary number. However.13). in complete distinction from Classical Mechanics.12. 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. This implies that small variations of the path near the classical path alter the value of the action integral by very little. Scl / ≈ 8. actually often essentially exclusive. However.e. In case of a proton with mass m = 1.6725 · 10−24 g moving over a distance of 1 ˚ in a time period of 10−14 s the value of S[r(t)] is A Scl ≈ 10−26 erg s and. for microscopic particles like the electron this is by no means the case.13) L(r. One would still expect signiﬁcant descructive interference between contributions of diﬀerent paths since the value calculated is comparable to 2π. (2.15) The exponent of (2.5 · 1027 . contribution to the path integral (2.14) has the same dimension as the action integral S[r(t)].12) is then Scl / ≈ 0.11) is Φ[r(t)] = exp where S[r(t)] is the classical action integral tN 13 i S[r(t)] (2. accordingly.13) are built on action integrals for all possible paths. a very large number.13) and = 1. ..

For the sake of simplicity we will consider the case of particles moving in one dimension labelled by the position coordinate x. t) = ˙ 1 mx2 − U (x) . as in (2.13) and. (xN .12. (2. etc.19) The main idea is that one can replace the path integral now by a multiple integral over x1 . They can be anywhere in the allowed volume which we will choose to be the interval ] − ∞. (2. −∞ dxN −1 (xj+1 −xj )2 1 − U (xj ) .10) and (2. The spacings between the times tj+1 and tj will all be identical. t0 ) = limN →∞ CN exp i N N −1 j=0 +∞ +∞ +∞ −∞ dx1 −∞ dx2 . x2 . . 2. tj ) to the next (xj+1 .17) The discretization in time leads to a discretization of the paths x(t) which will be represented through the series of space–time points {(x0 . t0 ). > t1 > t0 letting N go to inﬁnity later. The particles have associated with them a Lagrangian L(x. . ˙ 2 (2. t1 ). however. . the xj values are not. .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. In passing from one space–time instance (xj . namely tj+1 − tj = (tN − t0 )/N = N . namely 1 m(xj+1 − xj )2 / 2 and U (xj ). tN )} .12) φ(xN . respectively. (2.3 How to Evaluate the Path Integral In this section we will provide an explicit algorithm which deﬁnes the path integral (2. x.14 Quantum Mechanical Path Integral 2. (x1 . 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. This allows us to write the evolution operator using (2. tN −1 ). 2 2m N (2. Its value is CN = m 2πi N 2 (2. . ∞[. tj+1 ) we assume that kinetic energy and potential energy are constant.20) Here.20) can be properly taken.18) The time instances are ﬁxed. at the same time.9). . tN |x0 . (xN −1 . a series of times tN > tN −1 > tN −2 > .21) N 2. These assumptions lead then to the following N 2 Riemann form for the action integral N −1 S[x(t)] = lim N →∞ N j=0 1 (xj+1 − xj )2 m − U (xj ) 2 2 N . provides an avenue to evaluate path integrals. .16) In order to deﬁne the path integral we assume. .

4: Propagator for a Free Particle 15 Rather then using the integration variables xj . t0 ) (2. The resulting expression for S[x(t)|x(t0 ) = x0 .24) implies for the second term on the r. the origin of which coincides with the classical path of the particle. (2.23). x(tN ) = xN ] 1 (xN − x0 )2 m + 0 + 2 tN − t 0 + S[x(t)|x(t0 ) = 0. ˙ 2 t0 (2.24) Also we use the fact that the velocity of the classical path xcl = (xN −x0 )/(tn −t0 ) is constant. which can also be written φ(xN . it is more suitable to deﬁne new integration variables yj .28) i. tN ). x(tN ) = 0] . t0 ) and (xN . 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.29) This expression corresponds to the action integral in (2. tN dt t0 1 1 (xN − x0 )2 m x2 = m ˙ cl . tN dt y = y(tN ) − y(t0 ) = 0 . namely. tN |x0 . x(tN ) = 0.25) is. One obtains S[x(t)|x(t0 ) = x0 .25) + The condition (2.h.s. can be expressed through a path integral with endpoints x(t0 ) = 0. 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 . t0 ) = exp S[x(t)] (2. since x(t0 ) = xcl (t0 ) = x0 and x(tN ) = xcl (tN ) = xN . ˙ ˙ ˙ 2 + y2) = ˙ (2.13). xN − x0 xcl (t) = x0 + ( t − t0 ) .h.26) The ﬁrst term on the r. x(tN ) = xN ] is S[x(t)|x(t0 ) = x0 . ˙ t0 (2. 2 2 tN − t0 (2.22) where xcl (t) is the classical path which connects the space–time points (x0 .s. it holds y(t0 ) = y(tN ) = 0 .10. Inserting the result into (2. tN ) is to be evaluated. t0 ) = exp 1 im (xN − x0 )2 2 tN − t0 φ(0. using (2.e. = (2. x(tN ) = 0] .27) The third term can be written in the notation introduced tN 1 dt my 2 = S[x(t)|x(t0 ) = 0. tN |0. 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 .. (2. t0 ) and (xN . tN |x0 . 2.24). due to (2. .2.23) tN − t0 It is essential for the following to note that. To see the beniﬁt of such approach we deﬁne a path y(t) as follows x(t) = xcl (t) + y(t) (2. of (2.30) d[x(t)] exp 2 tN − t0 x(t0 )=0 a result.12) yields x(tN )=0 im (xN − x0 )2 i φ(xN .

37) det(Ajk ) .31) for a free particle one needs to evaluate the following path integral φ(0. . 0 0 ..35) must be determined.36) which holds for a d-dimensional. In the appendix we prove +∞ +∞ d j. 0 0 . (2. . 0 0 0 0 0 0 The following integral +∞ +∞ (2.16 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.. . (2.. . real.33) where ajk are the elements of the following symmetric (N 2 −1 0 −1 2 −1 2 m 0 −1 ajk = .k=1 dy1 · · · −∞ −∞ dyN −1 exp i yj bjk yk = (iπ)d det(bjk ) 1 2 .34) of (ajk ) deﬁning a new matrix (Ajk ) through ajk = Using det(ajk ) = m 2 N N −1 m 2 N Ajk .. t0 ) = limN →∞ × +∞ +∞ −∞ dy1 · · · −∞ dyN −1 exp i N m 2πi N 2 N × (2. .. .32) (yj+1 − yj )2 N −1 1 2 j=0 2 m N The exponent E can be written.. . In order to complete the evaluation of (2. symmetric matrix (bjk ) and det(bjk ) = 0. ..34) dy1 · · · −∞ −∞ dyN −1 exp i N −1 j. . . . (2.k=1 yj ajk yk − 1) × (N − 1) matrix . 2 −1 −1 2 (2.38) . tN |0. 0 0 .k=1 yj ajk yk (2.. .32) we split oﬀ the factor 2 mN in the deﬁnition (2.. . noting y0 = yN = 0.

t0 ) = m 2πi (t − t0 ) 1 2 exp im (x − x0 )2 2 t − t0 . . det(Ajk ) (2.. . deﬁning t = tN ..45) Expressions for Free Particle Propagator We have now collected all pieces for the ﬁnal expression of the propagator (2.40) in terms of subdeterminants along the last column. 0 0 Dn = . . . . We seek then to evaluate the determinant of the n × n matrix 2 −1 0 . . let say n. Using D1 = |(2)| = 2 . .4: Propagator for a Free Particle a property which follows from det(cB) = cn detB for any n × n matrix B... (2. (2. . . Our derivation has provided us with the value det(Ajk ) = N ..39) In order to determine det(Ajk ) we consider the dimension n of (Ajk ). . n = 1.42) m 2πi NN 1 2 (2. x = xN φ(x. tN |0.40) . (2. tN |0.18) we obtain φ(0. one can readily verify Dn = n + 1 . .46) .. 0 0 −1 2 −1 . .2. 2. tN |0. t0 ) = m 2πi (tN − t0 ) 1 2 .43) We like to note here for further use below that one might as well employ the ‘artiﬁcial’ starting values D0 = 1. . .41) the same result for D2 . t|x0 . . t0 ) = limN →∞ and with NN (2. (2.31) and obtain. . . variable. .41) D2 = 2 −1 −1 2 = 3 (2. To solve this three term recursion relationship one needs two starting values. .. One can readily verify that this procedure leads to the following recursion equation for the determinants Dn = 2 Dn−1 − Dn−2 . 0 0 0 −1 2 . 2 −1 0 0 0 −1 2 For this purpose we expand (2. D3 . which follows from (2. . Inserting this into (2. presently N − 1. D1 = 2 and obtain from (2. . we obtain φ(0.39) yields φ(0. . ...44) = tN − t0 . t0 ) = limN →∞ m 2πi N 2 17 N 2πi m N N −1 2 1 . 0 0 0 . .

51) as follows Eo (xo . 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 . t0 ) = 1 πδ 2 1 4 exp − x2 po 0 +i x 2 2δ (2. (2. We need to evaluate for this purpose the integral ψ(x.48) as the initial state using the propagator (2.49) is Gaussian of width δ.50) One refers to such states as wave packets. t|r0 .5) allows us to predict the time evolution of any state function ψ(x. according to (2.5) to (2.52) and applying (2. the wave function of the particle at later times. t) of a free particle. t0 )| = 2 1 πδ 2 1 2 exp − x2 0 δ2 (2.47) One-Dimensional Free Particle Described by Wave Packet We assume a particle at time t = to = 0 is described by the wave function ψ(x0 . centered around x0 = 0. We will obtain.10) φ(r. x) + E(x) in (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. One obtains then for the propagator (2. x) + E(x) (2. We want to apply axiom (2.51) For this evaluation we adopt the strategy of combining in the exponential the terms quadratic (∼ x2 ) and linear (∼ x0 ) in the integration variable to a complete square 0 ax2 + 2bx0 = a 0 x0 + b a 2 − b2 a (2. x) = im 2 t x2 1 + i o t mδ 2 − 2xo x − po t m + f (x) (2.46) can be generalized to three dimensions in a rather obvious way.18 Quantum Mechanical Path Integral This propagator. the associated probability distribution |ψ(x0 .247). t0 ) = m 2πi (t − t0 ) 3 2 exp im (r − r0 )2 2 t − t0 . (2.46). thereby. We devide the contributions to the exponent Eo (xo .53) . and describes a single particle since 1 πδ 2 1 2 +∞ dx0 exp −∞ − x2 0 δ2 = 1.

2 (2.58) The integrand is an analytical function everywhere in the complex plane and we can alter the integration path.53) E(x) = x− po mt 19 (2. the square in (2.59) An integration path in the complex x0 –plane along the direction i 1−i t mδ 2 (2.60) is then represented by real ρ values.51) ψ(x.56) m 2πi t 1 2 +∞ eE(x) −∞ dx0 eEo (xo .2.x) im xo 2 t t − mδ 2 x− 1+ x− 1+ po m = −∞ +∞ dx0 exp 1+i t 2 2 t i mδ2 = −∞ im dx0 exp 2 t t 1+i 2 mδ xo − po m t t i mδ2 . that the new path does not lead to additional contributions to the integral. x) = One can write then (2.54) f (x) = This yields Eo (xo .52). z2 = +∞ × i 1−i t mδ 2 (2.x) (2. (2. t) = 1 πδ 2 1 4 2 t 1 + i mδ2 .57) and needs to determine the integral +∞ I = −∞ +∞ dx0 eEo (xo . We consider a transformation to a new integration variable ρ deﬁned through i 1−i t mδ 2 ρ = x0 − x− 1+ po m t t i mδ2 . (2. making certain. (2. according to (2. 2 t One chooses then f (x) to complete. however.61) .4: Propagator for a Free Particle im x2 − f (x) . We proceed as follows. The beginning and the end of such path are the points z1 = −∞ × i 1−i t mδ 2 .55) im xo 2 t 1+i t − mδ 2 x− 1+ po m t t i mδ2 .

using a ab = a − . I = Equation (2. We can conclude then that (2.20 whereas the original path in (2.63) which can be readily evaluated.58) vanishes 0 for Re x0 → ±∞.65) t 1 − i mδ2 t 1 + i mδ2 .63) holds and. 1+b 1+b ﬁnally E(x) = − (x − po m (2. accordingly.58) by t mδ 2 +∞ I = i 1−i dρ exp − −∞ m 2 t 1+ t mδ 2 2 ρ2 (2. (2. the paths between z1 → z1 and z2 → z2 have a real part of x0 of ±∞.62) If one can show that an integration of (2.64) ψ(x. z2 = +∞ .67) We need to determine ﬁnally (2.57) reads then 1 πδ 2 1 4 2πi t . In fact.58) has a leading contribution in x0 of −x2 /δ 2 the integrand of (2.66) yields ψ(x. one can show that z1 lies at −∞ − i × ∞ and z2 at +∞ + i × ∞.68) and.58) along the path z1 → z1 and along the path z2 → z2 gives only vanishing contributions one can replace (2.69) t)2 t 2δ 2 (1 + i mδ2 ) + i po x − i p2 o t 2m (2. One obtains i po i po x x2 2m t E(x) = − 2 + − t t t 2δ (1 + i mδ2 ) 1 + i mδ2 1 + i mδ2 2 (2. Since the exponent in (2.55). 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) ] . Quantum Mechanical Path Integral (2.54) using (2. t) = Seperating the phase factor 1 t 1 + i mδ2 1 4 1 2 exp [ E(x) ] .70) . (2. (2. t m(1 + i mδ2 ) (2. Hence.58) has the end points z1 = −∞ .

2. the case of arbitrary to is recovered iby replacing t → to in (2. t) = exp i i p2 po o x − (t − to ) m 2m . the spatial dependence of the initial state (2. t) = exp i po i p2 o x − t m 2m . (2. respectively. 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. instead of (2.49). 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 . coinciding at t = 0 with the width of the initial distribution (2.72). The width of the distribution (2.72) 2 t2 δ 1+ 2 4 (2. (2.76) From this we conclude that an initial wave function ψ(xo . t0 ) = exp i po xo m . (2.71) conserves the phase factor exp[i(po / )x] of the original wave function (2.71) × exp − (x − po m t)2 ) 2δ 2 (1 + 2 t2 m2 δ 4 (1 − i .f.72). Another interesting observation is that the wave function (2.74) which corresponds to (2.48) and for the ﬁnal state (2.72) for the initial state (2. t)| = 2 1 πδ 2 (1 + 2 t2 m2 δ 4 1 2 ) exp − (x − δ 2 (1 + po m t)2 ) 2 t2 m2 δ 4 .77) .48) and that the respective phase factor is related with the velocity of the classical particle and of the center of the distribution (2.71). We had assumed above [c. The conservation of this factor is particularly striking for the (unnormalized) initial wave function ψ(x0 .49). The corresponding probability distribution is |ψ(x.73) m δ increases with time. However. This ‘spreading’ of the wave function is a genuine quantum phenomenon. This yields.74) remains invariant in time. In this case holds ψ(x.75) i.e.72) Comparision of Moving Wave Packet with Classical Motion It is revealing to compare the probability distributions (2.67) provides the complete expression of the wave function at all times t ψ(x. 2. t0 ) = exp i po i p2 o xo − to m 2m . (2. The center of the distribution (2..48)] to = 0.75) ψ(x.71.72) moves in the direction of the positive x-axis with velocity vo = po /m which identiﬁes po as the momentum of the particle. (2.48) for δ → ∞.4: Propagator for a Free Particle which inserted in (2. (2. (2.

2.81) vanishes2 . 1.13) φ(xN .5 Propagator for a Quadratic Lagrangian x(tN )=xN We will now determine the propagator (2.12. y(tN ) = 0 . t) = exp i Quantum Mechanical Path Integral po i p2 o x − t m 2m . xcl . y(t). x. t) + δL ˙ ˙ ˙ ˙ where L(xcl . t) + L (y. (2. 2.80) into (2. the end points of y(t) are y(t0 ) = 0 Inserting (2. 2. 2 . xcl . the spatial as well as the temporal dependence of the wave function remains invariant in this case. t) ˙ L (y.78) i. ˙ ˙ = (2.83) (2.22 becomes at t > to ψ(x. One refers to the respective states as stationary states.10.86) dt The reader may want to verify that the contribution of δL to the action integral is actually equal to the diﬀerential δS[xcl . t) ˙ 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 . Obviously.82) which describes the deviation from the classical path xcl (t) with end points xcl (t0 ) = x0 and xcl (tN ) = xN . Such states play a cardinal role in quantum mechanics. t) = L(xcl . For this purpose we use d xcl y = ˙ ˙ (xcl y) − xcl y ˙ ¨ (2. tN |x0 .82) one obtains L(xcl + y. t) = ˙ 1 1 mx2 − c(t)x2 − e(t)x .80) For this purpose we need to determine the action integral S[x(t)] = t0 dt L(x.85) We want to show now that the contribution of δL to the action integral (2. In order to simplify this task we deﬁne again a new path y(t) x(t) = xcl (t) + y(t) (2. ˙ 2 2 tN (2.84) .e. (2. t0 ) = x(t0 )=x0 d[x(t)] exp i S[x(t)] (2. y(t).79) for a quadratic Lagrangian L(x. t) ˙ (2. xcl + y(t). x.81) for an arbitrary path x(t) with end points x(t0 ) = x0 and x(tN ) = xN . y(t)] which vanishes according to the Hamiltonian principle as discussed in Sect..

The discretization scheme adopted above yields in the present case ˜ φ(0. 0 0 m ajk = . hence. . .. 0 0 0 c2 0 . tN |x0 . . . y.88) vanishes. One can then express the propagator (2. t0 ) = exp S[xcl (t)] φ(0. One can express the exponent E in (2. . .. x t0 (2. .91) through the quadratic N −1 E = i j.92) where ajk are the elements of the following (N − 1) × (N − 1) matrix 2 −1 0 . 0 0 0 c3 . .91) (yj+1 − yj )2 1 2 2m N where cj = c(tj ). 2 −1 0 0 0 −1 2 c1 0 0 . .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. 0 0 N 0 − . . tN |0.s.5: Propagator for a Quadratic Lagrangian and obtain tN t0 23 dt δL = m [xcl y]|t0 − ˙ t N tN dt [ m¨cl (t) + c(t) xcl (t) + e(t) ] y(t) . of (2. . t0 ) (2. t0 ) we will adopt a strategy which is similar to that used for the evaluation of (2. t0 ) = y(tN )=0 d[y(t)] exp y(t0 )=0 i tN dt L (y. tN |0. ..32). . . tj = t0 + form j. t0 ) = limN →∞ × +∞ +∞ −∞ dy1 · · · −∞ dyN −1 exp N i N N −1 j=0 m 2πi N 2 N × − 1 2 cj 2 yj (2. 0 0 0 . Applying the Euler–Lagrange conditions (1. tN |0.. . tN |0. .k=1 yj ajk yk (2.31) for the free particle propagator. . ..83) the ﬁrst term on the r. .2. 2 N . . . . . . .88) and.s. 0 0 0 .24) to the Lagrangian (2. . .79) i ˜ φ(xN . For the evaluation of φ(0. . t) ˙ t0 . . . . . 0 0 −1 2 −1 . cN −2 0 0 0 0 0 cN −1 (2. (2. . also the second contribution on the r. vanishes. 0 0 0 −1 2 .h..87) According to (2. .h. . 2 .. .93) . .89) where ˜ φ(0.. .80) yields for the classical path m¨cl + c(t) xcl + e(t) = 0 x (2. . .

24

Quantum Mechanical Path Integral

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

N 2

N

(iπ)N −1 det(a) 1

1 2

= limN →∞

1

2

N

2

N

N −1

m

det(a)

.

(2.94)

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

m c1

2 N 2 N

2

N

N

N −1

m

det(a)

.

(2.95)

def

2

N

N −1

m

det(a) 0 0 0 . . .

m cN −2

2 N

(2.96) 0

−1 −1 2 − . . . 0 0

2 N

0 ... −1 . . . ... . .. . . .

m c3

−1 2 − 0 . . . 0 0

m c2

0 ... 2 − 0

−1 2 −

m cN −1

2 N

0 0 . . . −1

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

2 N

m

cn

Dn−1 − Dn−2

(2.97)

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

2 N

m

c1

(2.98)

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

2 N

= −

cn+1 Dn . m

(2.99)

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

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

df (t0 ,t) dt t=t0

25

= =

N N

D0 = 0 ; D1 − D0

N

= 2−

2 N

m

c1 − 1 = 1 .

(2.101)

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

1 2

exp

i

S[xcl (t)]

(2.102)

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

2.6

Wave Packet Moving in Homogeneous Force Field

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

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

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

1 2

exp

i

S[xcl (τ )] .

(2.106)

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

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

Quantum Mechanical Path Integral

(2.108)

x − xo 1 f − (t − to ) t − to 2m

τ +

1 f 2 τ 2m

(2.109)

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

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

t 2

=

+ 2f

x − xo 1 f − (t − to ) t − to 2m

τ (2.111)

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

to

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

2

=

1 m 2 +f +

(t − to ) (t − to )2

1 f2 (t − to )3 + xo f (t − to ) 3 m

=

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

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

1 2

×

(2.113)

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

2.5: Propagator for a Quadratic Lagrangian

27

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

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

1

(2.114)

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

E(x) =

f t2 x − f (x) m

1 f 2 t3 . 24 m

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

1 4 2 t2 ) m2 δ 4

(2.117)

po m

f t2 2m

2

.

(2.118)

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

t 1 − i mδ2 t 1 + i mδ2

1 4

1 πδ 2 (1 +

exp [ E(x) ]

(2.119)

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

2

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

2 2

(2.120)

1+i t mδ 2 −

− x−

x−

x− x

f t2 po t− m 2m

po f t2 t− m 2m

2

f t2 + 2x m

po f t2 t+ m 2m

f t2 po t+ m 2m

1+i

t mδ 2 (2.121)

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

po m

Quantum Mechanical Path Integral

E(x)

=

− +

t−

2δ 2 1 + i

f t2 2m t mδ 2

2

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

(po + f t) x −

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

t

dτ (po + f τ )2 .

0

(2.123)

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

t 1 − i mδ2 t 1 + i mδ2

1 4

ψ(x, t)

=

1 πδ 2 (1 + x− 2δ 2

po m

2 t2 ) m2 δ 4

1 4

×

× exp − × exp i

t−

f t2 2m

2

1+

2 t2 m2 δ 4

1−i

t

t mδ 2

× . (2.124)

(po + f t) x −

i

0

dτ

(po + f τ )2 2m

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

2 t2 m2 δ 4 1 2

)

x− t− exp − 2 2 δ 2 1 + m2tδ4

po m

f t2 2m

2

.

(2.125)

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

2.5: Propagator for a Quadratic Lagrangian

29

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

**Propagator of a Harmonic Oscillator
**

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

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

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

t

(2.129) (2.130)

(2.131)

A =

x − x0 cosω(t − to ) , sinω(t − t0 )

(2.132)

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

(2.133)

S[xcl (τ )] =

t0

dτ

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

(2.135)

30

Quantum Mechanical Path Integral

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

(2.137)

(2.138)

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

t 0 dτ g(τ )

(2.139) (2.140) (2.141)

(2.142)

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

t

requires the integrals (2.143) (2.144)

dτ cos2ωτ

0 t

= =

dτ sin2ωτ

0

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

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

(2.145)

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

31

(2.146)

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

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

1 2

× . (2.147)

(x2 + x2 ) cosω(t − t0 ) − 2x0 x 0

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

1 4

exp

−

mω(x0 − bo )2 i + po xo 2

.

(2.148)

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

o

(2.149)

=

energy

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

∞

ψ(x, t) =

−∞

dx0 φ(x, t|x0 , t0 ) ψ(x0 , t0 ) .

(2.150)

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

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

1 4

m 2πiω sinω(t − t0 )

1 2

∞

dx0 exp [ E0 + E ]

−∞

(2.152)

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

Quantum Mechanical Path Integral

x2 c − 2xo x + isx2 − 2isxo bo + o o

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

E(x)

=

x2 c + isb2 − f (x) o

.

c = cos ω(t − to ) ,

s = sin ω(t − to ) .

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

1 4

m 2πiω sinω(t − t0 )

1 2

eE(x)

∞

dx0 exp [ E0 (xo , x)]

−∞

(2.156)

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

2

(2.157)

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

1 2

(2.158)

.

(2.159)

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

+∞

dx0 e

−∞

E0 (x0 )

=

2πi sinω(t − t0 ) mω exp[iω(t − t0 )]

(2.160)

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

imω 2 s

(2.161)

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

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

s2 c + i mpo 2 s3 2ω

2

=

− mω 2

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

p2 o m2 ω 2 po s2 − 2i mω xc

2

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

=

− mω (x − cbo − 2 − i

2

po mω

s)2 +

i

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

po ( 2mω − 1 mωb2 )sc + o 2

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

t

33

dτ

to

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

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

o (t

− to ) −

1 2

p2 mωb2 o o − 2mω 2

sc +

1 bo po s2 2

(2.164)

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

t to

˙ dτ L[b(τ ), b(τ )] =

p2 mωb2 o o − 2mω 2

sc − bo po s

(2.165)

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

t to

˙ dτ L[b(τ ), b(τ )]

(2.166)

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

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

o

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

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

1 2

exp −

mω

[x − b(t)]2

(2.168)

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

34

Quantum Mechanical Path Integral

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

1 4

exp

−

mωx2 iω 0 − to 2 2

.

(2.169)

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

1 2

exp

−

mωx2 0

,

(2.170)

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

2.7

Stationary States of the Harmonic Oscillator

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

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

+∞

W (x, t) =

−∞

dxo φ(x, t|xo , to ) W (xo , to ) .

(2.172)

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

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

− 1 iωT 2

(2.173)

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

− 1 iωt 2

35

mω π (1 − e−2iωT )

1 2

+∞

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

−∞

(2.174)

(2.175)

+ 2xo E(x) = − mω 2 x2

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

2

(2.176)

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

2

2xe−iωT + 2 1 − e−2iωT

mω

e

−iωto

.

(2.177)

=

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

+∞

mω

e

−iωto

(2.178)

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

−∞ Eo (xo ,x)

=

π (1 − e−2iωT mω

1 2

(2.179)

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

1

(2.180)

mω

=

−

x2 − 4

mω mω

x e−iωt +

2 e−2iωt mω (2.181)

=

−

mω 2 x + 2 2

x e−iωt − e−2iωt

172). but not of t.36 Altogether. 1 exp[iω(n + 2 ) to ] · · · .172) is a function of t − to and deﬁning accordingly Φ(x. one can expand (2.182) Comparision with (2. xo . We note that the factor exp(2 mω/ xe−iωt − e−2iωt ) in (2.172) in the form ∞ n=0 (2. 2 (2. t − to ) = φ(x. Noting that the propagator (2. t|xo .s. t|xo .. .183) where the expansion coeﬃcients are functions of x. to ).188) identiﬁes the functions ψn (x. t) = n=0 1 ˜ exp[ − iω(n + 1 ) t ] φn (x) 2 n! (2. 2. which also exhibits such invariance.171) can be expanded in terms of e−inωt . xo . In contrast to W (x.e.186) Fourier transform. 2.188) ˜ ˜ Equation (2. t). xo . We want to inspect the consequences of the invariance property (2.171) concludes the proof of (2. one obtains for the r. . results in 1 ˜ exp[−iω(n + 1 ) t] φn (x) 2 n! +∞ = −∞ dxo Φ(x. to ) exp[−iω(n + 1 ) to ] φn (xo ) .187) or 1 ˜ exp[−iω(n + 2 ) t] φn (x) +∞ = −∞ ˜ dxo φ(x. t) +∞ −∞ dto 1 ˜ exp[−iω(m + 1 ) to ] φm (xo ) 2 m! (2. of (2.h.171) ∞ W (x. t − to ) 1 ˜ exp[−iω(m + 1 ) to ] φm (xo ) 2 m! (2.172) I = exp 2 mω x e−iωt − e−2iωt − Quantum Mechanical Path Integral mω 2 1 x − iωt 2 2 . Accoordingly.185) Replacing t → t + to yields ∞ n=0 1 ˜ exp[−iω(n + 1 ) (t + to )] φn (x) 2 n! ∞ +∞ = m=0 −∞ dxo Φ(x. n = 1.184) 1 ˜ exp[−iω(n + 1 ) t] φn (x) 2 n! ∞ +∞ = m=0 −∞ dxo Φ(x. t − to ) 1 ˜ φn (xo ) n! (2. .171.172). . (2. t|xo . xo . t) = exp[−iω(n + 1 ) t] φn (x) as invariants under the 2 action of the propagator φ(x. i. to ) we express (2..147) in (2.

t) = z 2 e−y where w(y. Expansion (2. z) = exp(2yz − z 2 ) .5: Propagator for a Quadratic Lagrangian 37 ˜ ˜ the functions ψn (x.194) plays a central role for the Hermite polynomials since it contains. in a ‘nutshell’ all information on the Hermite polynomials. . z) (2. 1.190) ˜ is obeyed. . Actually.195) zn Hn (y) n! ˜ φn (y) .198) ∂z n z=0 . characterizes the wave functions φn (x). z) z 2 e−y or w(y.171). z) = Hn (y) (2.2. through the expansion coeﬃcients φn (x) in (2.193) (2. t) provide all stationary states of the quantum mechanical harmonic oscillator. through expansion (2.171) W (x.183). we have identiﬁed then. Expression (2. z) = n=0 1 2 /2 1 2 /2 = = mω e−iωt x (2. In the following we will characterize the functions φn (x) and determine the normalization constants Nn .194). The Hermite Polynomials ˜ The function (2. (2.183).191) (2. t) of the quantum mechanical harmonic oscillator ˜ ψn (x. To obtain ˜n (x) we simplify the expansion (2. t) = exp[−iω(n + 1 ) t] Nn φn (x) . t)|2 = ˜n (x)|2 . This follows from ∂n w(y.197) The expansion coeﬃcients Hn (y) in (2. t) such that +∞ −∞ 2 dx |ψ(x.183). 2 Here Nn are constants which normalize ψn (x. 2. t) are associated with a time-independent probablity density |ψn (x. ˜ |φ stationary wave functions ψn (x.197) are called Hermite polynomials which are polynomials of degree n which will be evaluated below. according to (2. (2.189) ˜ dx φ2 (x) = 1 n (2.192) w(y.183) reads then w(y. .194) = z2 n=0 ∞ 1 ∞ zn ˜ φn (y) n! (2.196) where Hn (y) = ey 2 /2 (2. t)|2 = Nn +∞ −∞ n = 0. We will also argue that the functions ψn (x. For this purpose we introduce ﬁrst closed expressions for φ the new variables y z and write (2.

E. ν H2 (y) = 4y 2 − 2. that Hn (y) is a polynomial of degree n.196) results in the so-called Rodrigues formula for the Hermite polynomials Hn (y) = (−1)n ey 2 ∂ n −y2 e . and recursion equations for the eﬃcient evaluation of Hn (y). One calls w(y. Massachusetts. closed expressions for Hn (y). H1 (y) = 2y.. i. The diagrams also identify the areas over which the summation is to be carried out. . ν+µ (2. 1990) is a useful introduction to this tool as is a chapter in the eminently useful Concrete Mathematics by R. Inc.38 Quantum Mechanical Path Integral which is a direct consequence of (2. 2. among other properties. ν+µ ν H3 (y) = 8y 3 − 12y. 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. D. For this purpose we expand the generating function (2. (2.Patashnik (AddisonWesley.198) for the Hermite polynomials can be expressed in a more convenient form employing deﬁnition (2.201) ν−µ µ µ ν−µ Figure 2.Wilf (Academic Press.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). 1989).199) Comparision with (2. As will become evident in the present case generating functions provide an extremely elegant access to the special functions of Mathematical Physics3 . Boston. 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. .Knuth. µ lattice (left diagram) corresponds to a summation over only every other point in an n.196) to ˜ derive.L. m lattice (right diagram).194.200) One can deduce from this expression the polynomial character of Hn (y).196) ∂n w(y. z) the generating function for the Hermite polynomials.200) yields for the ﬁrst Hermite polynomials H0 (y) = 1. and O. normalization factors for φ(y).196). The identity (2. The diagrams illustrate that a summation over all points in a ν.Graham.202) generatingfunctionology by H.. We will employ (2.S.e. ∂y n (2. Reading. .1: Schematic representation of change of summation variables ν and µ to n = ν + µ and m = ν −µ. We want to derive now explicit expressions for the Hermite polynomials. .

m must be restricted such that either both n and m are even or both n and m are odd.197) ∞ n=0 zn Hn (−y) = n! ∞ n=0 (−z)n Hn (y) = n! ∞ n=0 zn (−1)n Hn (y) n! (2. Hence. according to 2.207) This expression yields for the ﬁrst four Hermite polynomials H0 (y) = 1. k! (n − 2k)! = Hn (y) (2. 2 (2. p ≤ x. 2.207) one can deduce that Hn (y). m are ν = n+m .1. 39 (2. H2 (y) = 4y 2 − 2.194) holds w(−y. (2. otherwise. µ are integers the summation over n. z) = w(y. In case of even n . k! (n − 2k)! (2.2.208) which agrees with the expressions in (2.203) The old summation variables ν. in fact. 0 ≤ k ≤ [n/2] where [x] denotes the largest integer p.202) e 2yz − z 2 ∞ = n=0 zn n! ≤n m≥0 n! (−1) n−m 2 n−m 2 ! m! (2y)m .209). .e. it holds Hn (−y) = (−1)n Hn (y) . hence. (2. H3 (y) = 8y 3 − 12y. (2.206) From this expansion we can identify Hn (y) [n/2] Hn (y) = k=0 (−1)k n! (2y)n−2k . The lattices representing the summation terms are shown in Fig. the sum in (2. .209) This property follows also from the generating function. m = ν − µ 0 ≤ n < ∞. From (2.. −z) and. .201). it contains only odd powers. H1 (y) = 2y. According to (2. i. With this restriction in mind one can express (2. µ expressend in terms of n. 2 µ = n−m . . 0 ≤ m ≤ n . is a polynomial of degree n.5: Propagator for a Quadratic Lagrangian and introduce now new summation variables n = ν + µ.204) Since ν. 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 .205) Since (n − m)/2 is an integer we can introduce now the summation variable k = (n − m)/2 .210) from which one can conclude the property (2. for odd n.207) contains only even powers.

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. Starting point of the derivation is the property of w(y. namely. . z) − (2y − 2z) w(y. .40 Quantum Mechanical Path Integral The generating function allows one to determine the values of Hn (y) at y = 0. H2n+1 (0) = 0 .218) The reader should recognize the connection between the pattern of the diﬀerential equation (??) and the pattern of the recursion equation (??): a diﬀerential operator d/dz increases the order n of Hn by one. . dw/dy − 2zw = 0. a property which is consistent with (2. z). For this purpose one considers w(0. n = 1. .213) which allow one to evaluate Hn (y) from H0 (y) and H1 (y) given by (2. (2. Another relationship is d Hn (y) = 2n Hn−1 (y). z) should be equivalent to the relationship (??). . 2. dy n = 1.215) 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. . .208).211) m! n! m=0 n=0 Comparing terms on both sides of the equation yields (2n)! . . Hn+1 (y) − 2y Hn (y) + 2n Hn−1 (y) = 0.212) n! This implies that stationary states of the harmonic oscillator φ2n+1 (x). z) = 0 (2. For Hermite polynomials holds. 2. n! (2. as deﬁned through (2.214) We want to derive (??) using the generating function. (2. (2. 2. z) = exp(−z 2 ) and carries out the Taylor expansion on both sides of this expression resulting in ∞ ∞ (−1)m z 2m zn = Hn (0) . indeed satisﬁes the latter relationship. 2. .194). a factor z reduces the order of Hn by one and introduces also a factor n. . The reader may verify that w(y. n = 0. One can then readily state which diﬀerential equation of w(y. .233) below. . have a node at y = 0. H2n (0) = (−1)n Recursion Relationships A useful set of properties for special functions are the so-called recursion relationships.196) into the diﬀerential equation (2. 1.215) ∂z which can be readily veriﬁed using (2. .217) gives H1 (y) − 2y H0 (y) = 0. z) ∂ w(y.188.194).197) above and given by (2. as given in (2. 2. (2.209) since odd functions have a node at the origin. Substituting expansion (2. n = 1. for example. Hn+1 (y) − 2y Hn (y) + 2n Hn−1 (y) = 0 .

ﬁnally. 2 (2. (2.h. acording to the deﬁnition (2.221) 1 = √ π +∞ −∞ dt e2iyt − t . 2 √ π e2 z z (2. 2 (2. e−y 2 π e−y 2 (2.226a). 1.220) Using (2.226) π −∞ 2n z n z n n=0 n! Expressing the l.194.228) .h.s.247) for a = i one obtains I(y) = and. 2.5: 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 .196) the orthogonality properties of the Hermite polynomials.200) yields Hn (y) = The identity (−1)n y2 √ e π +∞ dt −∞ dn 2iyt − t2 e .224) +∞ −∞ dt tn e2iy t − t . of the Rodrigues formula (2. . through a double series over Hermite polynomials using (2. 2.2. 2 (2. For this purpose we consider the integral +∞ −∞ dy w(y. dy n (2.223) dn 2iy t − t2 2 e = (2 i t)n e2iy t − t n dy results. 2 n = 0.196) yields ∞ n. z ) e−y 2 +∞ = e2 z z = √ ∞ dy e−(y−z−z ) = . 2. in the integral representation of the Hermite polynomials 2n (−i)n ey √ Hn (y) = π Orthonormality Properties 2 (2.225) We want to derive from the generating function (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. z) w(y. (2.s.194. .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 .219) which can be written I(y) = e−y 2 +∞ −∞ dte−(t−iy) = e−y √ 2 2 +∞ −∞ dze−z .222) Employing this expression now on the r.

(2.2. According to (2. Furthermore.232) Hn (y) . 2. 2.230) (2. 4. 4. 4 in Fig.229) Hn (y) . that the wave functions are even for n = 0. One can recognize. negative for n = 2 and vanishes for n = 1. 2. 1. 3.228) +∞ 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. 2.231) √ π (2.233) .190)] φn (y) = Nn e−y 2 /2 and for the normalization constants Nn follows from (2.228) allow us to construct normalized stationary states of the harmonic oscillator. (2. 3. 3. Normalized Stationary States The orthonormality conditions (2.189. ¡ ¢ ¡ -4 -2 0 2 4 y (2.197) holds 2 ˜ φn (y) = e−y /2 Hn (y) . 4 and odd for n = 1. in harmony with (??).f. One can also recognize that n is equal to the number of nodes of the wave function. The normalized states are [c.233) are presented for n = 0. (2. the value of the wave function at y = 0 is positive for n = 0. 3. in agreement with our above discussions. 2.2: Stationary states φn (y) of the harmonic oscillator for n = 0. 1.42 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.

w(x) = xα e−x . 5.235) Completeness of the Hermite Polynomials The Hermite polynomials are the ﬁrst members of a large class of special functions which one encounters in the course of describing stationary quantum states for various potentials and in spaces of diﬀerent dimensions. 5. 1].e.238) where the space is endowed with the scalar product (f |g) = Ω dx f (x) g(x) w(x) = < ∞ .239) . w(x) = √ exp(−x2 ).234) The explicit form of the stationary states of the harmonic oscillator in terms of the position variable x is then. and I = 2/(2 + 1). 5 below [c. The Hermite polynomials are so-called orthonogal polynomials since they obey the conditions (2. The orthogonal polynomials pn mentioned above have the important property that they form a complete basis in the space F of normalizable functions. 2. (2.. of functions which obey dx f 2 (x) w(x) = < ∞ .228) shows that the orthonogality condition of the Hermite polynomials is in complience with (2. denoted by P (x) and introduced in Sect. g ∈ F . w(x) = 0 only at a discrete set of points xk ∈ Ω (2. The latter are written for polynomials pn (x) in the general form dx pn (x) pm (x) w(x) = In δnm Ω (2. In case of the associated Laguerre (α) polynomials. 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. denoted by Ln (x) and encountered in case of the stationary states of the nonrelativistic [see Sect. +∞[. (2.236) where w(x) is a so-called weight function with the property w(x) ≥ 0. i.e.f.2. i. 10. using (2. ???] and relativistic [see Sect.233) and (2. Comparision with (2.189) φn (x) = √ 1 2n n! mω π 1 4 e− mωx2 2 Hn ( mω x) . (2.459] hydrogen atom. Ω (2. ??? and eq.237) for Ω = R. (5. holds Ω = [0.10 and eq.237) and where In denotes some constants.228). w(x) ≡ 1.151.179] holds Ω = [−1. (10. and In = 2n n! π.156. 5.150 ..231) of the wave functions diﬀers from that postulated in (2. The various orthonogal polynomials diﬀer in the spaces Ω ⊂ R over which they are deﬁned and diﬀer in a weight function w(y) which enter in their orthonogality conditions. f. by dx = dy mω 1 4 .5: Propagator for a Quadratic Lagrangian 43 The normalization condition (2.189) by the Jacobian dx/dy.236 . Other examples of orthogonal polynomials are the Legendre and Jacobi polynomials which arise in solving three-dimensional stationary Schr¨dinger equations. In = Γ(n+α+1)/n! where Γ(z) is the so-called Gamma function. In case of the Legendre polynomials.

k yj ajk yk = (iπ)n .233) exhaust all stationary states of the harmonic oscillator.244) For the state to be stationary |ψ(x. a state for which (2.246) for det(a) = 0 and real. the states (2.e.247) . Appendix: Exponential Integral We want to prove +∞ +∞ I = dy1 . In case of n = 1 this reads +∞ −∞ dx ei a x = 2 iπ .. t)|2 .242) and the existence of a normalizable state ψ(y.240) where cn = 1 In dx w(x) f (x) pn (x) . . ∞ n.172) holds. Since the Hermite polynomials form a complete basis for such states we can expand ∞ ψ(y. (2. hence. ψ(y. . t) which is stationary under the action of the harmonic oscillator propagator (2. det(a) (2. i.44 As a result holds for any f ∈ F f (x) = n Quantum Mechanical Path Integral cn pn (x) (2.241) The latter identity follows from (2. except for a single n = no . If one replaces for all f ∈ F: f (x) → w(x) f (x) and.239) becomes the conventional scalar product of quantum mechanics f |g = Ω dx f (x) g(x) .. t) is ∞ ψ(y. symmetric a.e. Ω (2.m=0 d∗ dm exp[iω(m − n)t] e−y Hn (y)Hm (y) . i. aT = a.188. a (2.242) Let us assume now the case of a function space governed by the norm (2. Therefore. i.245) must be time-independent. i. t) = n=0 cn (t) e−y 2 /2 Hn (y) .236).243) To be consistent with(2. n 2 (2. pn (x) → w(x) pn (x) the scalar product (2.e. (2.233). t) must be identical to one of the stationary states (2. 2 (2. t) = n=0 1 2 dn exp[−iω(n + )t] e−y /2 Hn (y) . The only possibility for this to be true is dn = 0.147).. 2.197) it must hold cn (t) = dn exp[−iω(n + 1 )t] and.e. the 2 stationary state ψ(y. in particular. −∞ dyn ei −∞ n j.

253) as well as det(˜) = det(S−1 aS) = det(S−1 ) det(a) . .248. . . det(S) a = (det(S))−1 det(a) det(S) = det(a) . (2.e.250) The bilinear form in (2. ST S = 1 1 or S = S−1 . The proof of (2. This property allows one to simplify the bilinear ˜ form n yj ajk yk by introducing new integration variables j.248) .255) .k n m n y Sj ajk Skm ym ˜ ˜ y (S T ) j ajk Skm ym ˜ ˜ j.2.. . . according to (2.251) where.k n n yj = ˜ k (S −1 )jk yk .246) exploits that for any real.k n m = = j. .252) ˜ For the determinant of a holds det(˜) = a n ajj ˜ j=1 (2. (2. .m (S T )jl alm Smk . 0 ˜ ˜ S−1 a S = a = . . ˜ (2.k yj ajk yk ˜˜ ˜ (2. i. . . ann ˜ where S can be chosen as an orthonormal transformation. . 0 0 .249) n ajk = ˜ l. ˜ (2. . (2.249) The akk are the eigenvalues of a and are real.246) reads then in terms of yj ˜ n n j. 2. 0 ˜ 0 a22 . .. . One can conclude det(a) = j=1 n (2.7: Appendix / Exponential Integral 45 which holds for a ∈ C as long as a = 0. . yk = k Skj yj .254) ajj . symmetric matrix exists a similarity transformation such that a11 0 . .k yj ajk yk n = j.

.258) +∞ +∞ ∂(y1 . ..259) (2. . . . yn ) y ˜ ∂yj . . y −∞ d˜n det y −∞ ∂(y1 . . 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 −∞ . .46 We have assumed det(a) = 0. . d˜n e y −∞ = (2. . yn ) ) = det(S) . . .250) allows one to express (2. . . 2. . holds n Quantum Mechanical Path Integral ajj = 0 ˜ j=1 (2. . i. (2.264) I = d˜1 . . . det( From (2..266) . ∂ ys ˜ (2.e. yn ) ∂(˜1 . . Accordingly. .258) where we introduced the Jacobian matrix J = with elements Jjs = According to (2. n (2.260) (2. . ajj = 0 . ∂(˜1 .261) ∂(y1 . . . ˜ for j = 1.250) holds J = S and.249) follows 1 = det ST S such that one can conclude det S = ±1 One can right then (2. . . .256) such that none of the eigenvalues of a vanishes. . yn ) y ˜ = ( det S )2 (2. .263) (2.257) Substitution of the integration variables (2. .265) k=1 −∞ which leaves us to determine integrals of the type +∞ dx eicx −∞ 2 (2.262) (2. .. hence. yn ) ∂(˜1 . yn ) y ˜ ei n k akk yk ˜ ˜2 . .250) +∞ +∞ I = d˜1 . .

7: Appendix / Exponential Integral 47 where.268) can be written as the sum of the following path integrals J = J1 + J2 + J3 + J4 .271) . (2. according to (2.267) by considering the contour integral J = γ dz eicz = 0 2 (2.2.. 2 since eicz is a holomorphic function. for x.c > 0 .e.266) to the well-known Gaussian integral +∞ dx e−cx −∞ 2 = π . the integrand does not exhibit any singularities anywhere in C. One can relate integral (2. Substituting −x for x into integral J4 one obtains 0 J4 = p p (−i) dx eic(−ix+p) 2 = 0 i dx eic(ix−p) 2 = J2 . i.269) The contributions Jk can be expressed through integrals along a real coordinate axis by realizing that the paths γk can be parametrized by real coordinates x p γ1 : z = x γ2 : z = ix + p γ3 : z = √ ix J1 = −p p dx eicx 2 J2 = J3 = = i dx eic(ix+p) 0√ − 2p √ √ 2p 2 i dx eic( 2p √ ix)2 √ − i 0 √ (2. The contour intergral (2. We consider ﬁrst the case c > 0 and discuss the case c < 0 further below.257) holds c = 0. p ∈ IR.3.268) vanishes.268) along the path γ = γ1 + γ2 + γ3 + γ4 displayed in Figure 2. Jk = γk dz eicz 2 (2. The contour integral (2. c (2.270) dx e−cx 2 2 − 2p √ γ4 : z = ix − p J4 = −p i dx eic(ix−p) .

(2.275) .268) for p = +∞. (2.48 Quantum Mechanical Path Integral Im(z) ip γ2 ¡ γ4 ¡ γ3 ¡ -i p Figure 2. c ¡ -p γ1 p Re(z) 2 | |e−2cxp | . We will now show that the two integrals J2 and J4 vanish for p → +∞. 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.273) p→+∞ lim J2 and J4 do not contribute then to integral (2.267) one has shown then ∞ dx eicx −∞ 2 = iπ . 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 = 0.272) = 0 . (2.3: Contour path γ in the complex plain. Hence.274) (2.

7: Appendix / Exponential Integral 49 One can derive the same result for c < 0. (iπ)n det(a) (2.275.277) We apply the above results (2. n ˜ j=1 ajj (2. c (2. This leads to ∞ icx2 J = dx e −∞ + √ −∞ −i ∞ dx ecx = 0 2 (2.265). 2.277) to (2.268). but with a path γ that is reﬂected at the real axis.255) this result can be expressed in terms of the matrix a I = which concludes our proof.278) Noting (2. It holds n I = k=1 iπ = akk ˜ (iπ)n . if one chooses the same contour integral as (2.279) .276) and (c < 0) ∞ dx eicx −∞ 2 = −iπ = −|c| iπ .2.

50 Quantum Mechanical Path Integral .

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

7) shows 1 ∂ In (a). x2 . Since the latter contributes to (3. n = 0.36) holds I0 (a) = Inspection of (3.k=1 xj xk ∂x2 ∂x2 ψ(r.5) assuming that only the leading terms contribute. 1. a I2 (a) = − 3 4a2 iπ . t) + . t) j 1 12 . e. a (3. It is then suﬃcient to consider the terms ψ(r. U (r.. Obviously.4) only for small x2 + x2 + x2 values 1 2 3 we expand 3 ψ(r + s. consequently. 2 (3. t) + ∂xj 2 3 xj xk j.11) holds 3 m 3 iπ 2 2 × = 1 (3. i ∂a Starting from (3. t) + j=1 xj ∂ 1 ψ(r.10) It is now important to note that in case of integral (3. (3.. t) and the kinetic energy term. O( 7 2 ) . x3 .7) According to (2.. only terms of the expansion (3.52 Schr¨dinger Equation o It is important to note that the integration is not over r. t) is a constant with respect to this integration. t) j . a supposition which will be examined below.8) one can evaluate recursively all integrals In (a). It holds In+1 (a) = I1 (a) = i 2a iπ .8) (3.4) one identiﬁes 2 1 = a m = O( ) (3. x3 )T .k=1 ∂2 ψ(r. According to (3.5) which are even separately in all three coordinates yield non-vanishing contributions. t) .. (3. 1 4 3 2 2 ∂4 j. ..6) make contributions of the order O( 3 2 ) .4) is even in all three coordinates x1 . t) = ψ(r. ∂xj ∂xk (3. The integration involves only the wave function ψ(r +s. O( 7 2 ).6) 3 4 ∂4 j=1 xj ∂x4 ψ(r.13) 2πi a . the terms collected in (3. t) j k . O( 5 2 ) . of the expansion of ψ(r + s. Since the kinetic energy contribution in (3.11) and. we need then to evaluate integrals of the type +∞ In (a) = −∞ dx x2n exp i a x2 . 1 2 3 2 ∂2 j=1 xj ∂x2 ψ(r.g. . t)..9) iπ a (3.12) Here one needs to note that we are actually dealing with a three-fold integral. x2 . but over s = (x1 .

t) + 1 2i 4 m 2 53 ψ(r. t) ψ(r.3.16) Inserting this into (3. (3. t) + O( 2 ) . t) = ψ(r. The 1st order time derivative requires that for any solution a single temporal condition needs to be speciﬁed. (3. Due to its linear character any linear combination of solutions of the time-dependent Schr¨dinger o equation is also a solution. t) (3. t). This equation is often written in the o form ∂ ˆ i ψ(r.. t) (3. one speciﬁes the so-called initial condition. t) + and exp − i U (r. t) = H ψ(r.20) 2m 3. The following general remarks can be made about the solution.14) This expansion in terms of powers of suggests that we also expand ∂ ψ(r. (3. t) = ∂t 2 − 2 2m + U (r.14) results in ψ(r.15) ψ(r. Usually. Obviously. We will derive brieﬂy the type of boundary conditions encountered.. t) + O( 2 ) . t) + O( 2 ) . ψ(r. t) ψ(r. (3.17) 2m ψ(r. t) . t + ) = ψ(r. t + ) = exp − i U (r. 2nd o order in the spatial variables and linear in the solution ψ(r. 1st order in time. t1 ) = f (r). As we will discuss in Chapter 5 below the solutions of the Schr¨dinger equation are restricted to particular Hilbert spaces H which are o . t) + O( 2 ) ∂t i U (r. t) . e. In order O( ) the equation reads i ∂ ψ(r.e. t) = 1 − (3.2 Boundary Conditions The time-dependent Schr¨dinger equation is a partial diﬀerential equation.19) ∂t where 2 2 ˆ H = − + U (r.10).2: Boundary Conditions and one can conclude. ψ(r. t) + ∂ ∂t ψ(r. The 2nd order spatial derivatives require that one speciﬁes also properties of the solution on a closed boundary ∂Ω surrounding the volume Ω in which a solution is to be determined.g. this equation is trivially satisﬁed to order O( 0 ). t) − + i 2 2 i U (r. t)ψ(r. i. a solution is thought for t ≥ t0 and the solution is speciﬁed at the intial time t0 .18) This is the celebrated time-dependent Schr¨dinger equation. using (3.

g|r) (3. Interchanging f ∗ (r) and g(r) results in g| H |f Ω = Ω ˆ d3 r g(r) Hf ∗ (r) . In (3. The diﬀerence between the integrals is g| H |f = Ω Ω = − g| H |f d3 rf ∗ (r) − Ω 2 2 2m g(r) − Ω d3 rg(r) 2 − 2 2m f ∗ (r) + Ω 2 d3 rf ∗ (r) U (r. Calssical Electrodynamics. (3.23). t).22) and (3. for example. Since energy is a real ˆ ˆ quantity one needs to require that the eigenvalues of the operator H are real and. however.27) ˆ We will postulate below that H is an operator in H which represents energy. that H is 2 . in principle. diﬀer from each other. t) g(r) − Ω d3 rg(r) U (r. Jackson.22) ˆ where H is deﬁned in (3. g|r) = − 2 2m ( f ∗ (r) g(r) − g(r) f ∗ (r) ) (3. implies hermitian ˆ f | H |g 1 2 Ω ˆ = g| H |f Ω (3. J. The hermitian property.26) the vector–valued function P (f ∗ . f ∗ (r) d3 rg(r) Ω 2 = − 2m d3 rf ∗ (r) Ω 2 g(r) − f ∗ (r) (3.28) See. (John Wiley. g ∈ H is deﬁned as follows f |g Ω = d3 rf ∗ (r)g(r) (3. 1975).20). New York. The reader is advised to consult a reference text on ‘Linear Algebra’ to follow this argument. hence. .54 Schr¨dinger Equation o linear vector spaces of functions f (r) in which a scalar product between two elements f.23) ˆ Since H is a diﬀerential operator the expressions (3.26) where ∂Ω da · A(r) denotes an integral over the surface ∂Ω of the volume Ω.. D. 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.21) Ω This leads one to consider the integral f | H |g Ω = Ω ˆ d3 rf ∗ (r) H g(r) (3. g|r) is called the ˆ concomitant of H and is P (f ∗ . Chapter 1. 2nd Ed.25) ˆ f | H |g Ω ˆ = g| H |f Ω + ∂Ω da · P (f ∗ .24) Using Green’s theorem1 Ωd 3r f ∗ (r) 2 g(r) − g(r) 2 f ∗ (r) = one obtains the identity ∂Ω da · ( f ∗ (r) g(r) − g(r) f ∗ (r) ) (3.

. t)|2 . The latter property stems from the fact that the boundary condition (3. t)∂t ψ(r.. In this case one can expect that the particle density is localized in the area where the energy of the particle exceeds the potential eenrgy. e. r ∈ ∂Ω (3. . t)|2 .34) An example is a volume between two concentric spheres. However.30) can be postulated. ∂Ω = ∂Ω1 ∩ ∂Ω2 ∩ ∂Ω3 ∩ . . Often the closed surface of a volume ∂Ω is the union of disconnected surfaces3 ∂Ωj .e.32). t) which describes the state of o a particle moving in the potential U (r. g|r) vanish on ∂Ω. 3. t) is ∂t p(ω.e.. . in which case ∂Ω1 is the inner sphere and ∂Ω2 is the outer sphere. in this case also all derivatives of f (r) vanish at inﬁnity. i. t) = ω 3 d3 r [ ψ ∗ (r. In this case one can postulate both conditions (3.29) In fact.e. α > 2. (3. t) = ω d3 r |ψ(r. r ∈ ∂Ω or da · f (r) = 0 ∀r. namely. like exp(−κr). κ > 0 or like r−α . It must hold then for all f ∈ H f (r) = 0 ∀r.3 Particle Flux and Schr¨dinger Equation o The solution of the Schr¨dinger equation is the wave function ψ(r. (3. therefore.3..3: Particle Flux 55 and.31) usually arises when a particle existing in a bound state is described. t) + ψ(r. and that the density decays rapidly when one moves away from that area. t)∂t ψ ∗ (r.29. t) ] . if f and g satisfy these conditions than also does any linear combination αf + βg. The observable directly linked to the wave function is the probability to ﬁnd the particle at position r at time t.32) the wave function must decay for |r| → ∞ rapidly enough to be square integrable. i. The probability to observe the particle anywhere in the subvolume ω ⊂ Ω is p(ω.g. In either case does f (r) and all of its derivatives vanish asymptotically.29. we can only allow functions which make the diﬀerential da · P (f ∗ . (3. 3.31) |r|→∞ (3. i. t) on a single connected surface ∂Ωj only either one of the conditions (3. Since the total probability of ﬁnding the particle anywhere in space is d3 r|f (r)|2 = 1 (3. to avoid discontinuities in 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” .33) The time derivative of p(ω. obey (3.30) Note that these boundary conditions are linear in f .30) each condition holding on an entire surface ∂Ωj . 3. t). |ψ(r.

t) ] . t) = 0 . t) − ψ ∗ (r. t).27) one can express this j(r. t) = 0.43) (3. Accordingly the probability to observe the particle anywhere in the total volume Ω is constant. t) ψ ∗ (r. t) H ψ(r.27) this can be written i ∂t p(ω. 3.44) ˆ Note that the Hamiltonian H involves only real terms. One can multiply the solution of (3. t) = [ ψ(r. t) (3. t) .42) 2mi Since (3. t) + 4 · j(r. t)|2 is a probability density with units 1/volume. ψ(r.37) can be expressed through a volume integral according to da · A(r) = ∂ω ω (3. t) H ψ ∗ (r. ψ|r.26.30) ∂t p(Ω. (3. t) = ∂ω da · P (ψ ∗ (r. Note that for a normalized wave function the quantity ρ(r.39) d3 r · A(r) (3. t) = |ψ(r. Using (3. (3.38) holds. t)|2 = 1 Ω (3. 3. t) = 0 (3. (3. (3. t) = P (ψ ∗ . t)r.41) where j(r. t) ψ(r.37) If one applies this identity to ω = Ω one obtains according to (3. t) . We will assume in the remainder of this Section that the solutions discussed are normalized.29. One refers to such solution as normalized. The surface integral (3. A natural choice for this constant is 1. t) = ω Schr¨dinger Equation o ∂ ∗ ˆ ψ (r. t) − ψ(r.56 Using (3. t) + ω · j(r. . t) such that d3 r|ψ(r.41) holds for any volume ω ⊂ Ω one can conclude ∂t ρ(r. t) ∂t ˆ ˆ ψ ∗ (r. t) = H ψ ∗ (r.36) According to (3.35) d3 r .37) d3 r ∂t ρ(r.40) One can rewrite then (3.19) and its conjugate complex4 −i yields i ∂t p(ω.18) by any complex number and accordingly one can deﬁne ψ(r.

t) = − ψ(r. 2. In Sect.4: Free Particle 57 The interpretation of j(r.47) . i.3. (3.45) Obviously.e. t) has vanishing ﬂux anywhere.50) 3 ˜ d3 k φ(k) exp i(k · r − ωt) Ω∞ (3.18) in Ω∞ = R in the case o U (r.f.167)] potential. equal to the velocity of the particle.48) ∂t 2m which describes the motion of free particles. t0 ) determines φ(k). (2. t) . t) = [2π]− 2 where the dispersion relationship holds ω = k2 . (2.4 Solution of the Free Particle Schr¨dinger Equation o We want to consider now solutions of the Schr¨dinger equation (3. One often encounters wave functions of the type φ(r) = f (r) ei k·r The corresponding ﬂux is j(r) = k 2 f (r) . the initial condition at ψ(r. for f (r) ∈ R.48) that the general solution is of the form ψ(r. Equation (3.43) that any real wave function ψ(r. t0 ) = [2π]− 2 Ω∞ 3 ˜ d3 k φ(k) exp i(k · r − ωt0 ) . t) gives rise to a descrease of the total probability in volume ω due to the disappearence of probability density at the surface ∂ω.48.49) ˜ Obviously. m (3.41) written in the form ∂t ω d3 r ρ(r.105. 2. (3. (3. t) = 0 2 ∂ 2 i ψ(r. It is of interest to note from (3.125)] and in a quadratic [c. Such case arose in Sect.f. 2m (3. One can readily show by insertion into (3.51) .. (3. 3. arises solely from the complex factor exp(i k · r). 2. t) points in the direction to the outside of volume ω. This ﬁnding is consistent with the present evaluation of the particle ﬂux: a factor exp(ipo xo / ) gives rise to a ﬂux po /m. j(r. (2.49) reads at t = t0 ψ(r..148. 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.e. and for particles moving in a linear [c. 2 for a free particle [c. 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. Note that j(r.46) i. t) (3.71)]. t) = − ∂ω da · j(r. t) is that of density ﬂux.f. This follows directly from an inspection of Eq.

hence.e. i. If one chooses the initial state f (r) deﬁned in (3. t0 ) ψ(r0 . Comparision with Path Integral Formulation One can write solution (3.47). (t − t0 ) (3.. In case of the harmonoc oscillator the expansion can be stated in a closed form given in (4. respectively. in (2.52) We have not speciﬁed the spatial boundary condition in case of (3.70) derived below for a particle in a box and the harmonic oscillator. We have identiﬁed then with (3.49). i.52) above ψ(r. The solution as stated is deﬁned in the inﬁnte space Ω∞ = R .81) .49..54) valid for this case. t|r0 . t|r0 . t0 ) (3. t0 ) .114) and (4. Ω∞ (3.49) is square integrable at all subsequent times and. evaluating the integral in (3. t0 ) = 1 2π 3 d3 k exp Ω∞ ik · (r − r0 ) − i 2 k2 2m (t − t0 ) . The ensuing solutions are called wave packets. 3. 3.47).54) This expression obviously has the same form as postulated in the path integral formulation of Quantum Mechanics introduced above.31) applies.54) to the case of non-vanishing potentials U (r).e.51.56) and using (2. (3. Below we will generalize the propagator (3. The general form for this propagator involves an expansion in terms of a complete set of eigenfunctions as in (3.54) yields (2.58 The inverse Fourier transform yields 3 ˜ φ(k) = [2π]− 2 Schr¨dinger Equation o d3 r0 exp(−ik · r0 ) ψ(r0 . t) = Ω∞ d3 r0 φ(r. that the “natural boundary condition” (3. t) as given by (3. To show this one needs to note 1 2π = +∞ dk1 exp −∞ ik1 (x − x0 ) − 1 2 i 2 k2 1 2m .247).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 (3.54) an alternative representation of the propagator (2. derive an expression similar to (3.5). In fact.51) to be square integrable it follows according to the properties of the Fourier–transform that ψ(r.53) where φ(r.

t) 2 k2 (3. (3. t) = −∞ dx0 φ(x. 3. t) = f (t) φ(r) .61) × exp − (x − po m t)2 2δ 2 (1 + 2 t2 ) m2 δ 4 (1 − i . a result which is identical to the expression (??) obtained by means of the path integral propagator (2.60) yields with t0 = 0 ψ(x.49. i.63) . t) is given by (??).4: Free Particle Free Particle at Rest 59 We want to apply solution (3. t|x0 . r ∈ ∂Ω or da · φ(r) = 0 ∀r. t0 ) ψ(x0 .60) =???? Combining (3.19. 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.3.30).64) (3.54) is +∞ ψ(x.29. r ∈ ∂Ω (3.62) We will restrict the space of allowed solutions to a volume Ω such that the functions also make the concomitant (3. 3. t|x0 . (3.57) where φ(x.55. 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 (3.59) which is solved through completion of the square in the exponent [c. (3.58) Integration over x0 leads to the integral +∞ dx0 exp −ikx0 + −∞ i po x0 − x2 0 2δ 2 = √ 2πδ 2 exp − (k − po 2 2 ) δ 2 (3. (3.27) vanish on the surface ∂Ω of Ω.20) which are of o the form ψ(r. The 1-dimensional version of (3.f.53.46). the functions obey boundary conditions of the type (3. Stationary States We consider now solutions of the time-dependent Schr¨dinger equation (3.e. 3. Accordingly.56)]. t0 ) = 1 2π +∞ dk exp −∞ ik (x − x0 ) − i 2m (t − t0 ) . the boundary conditions are φ(r) = 0 ∀r.57–3. The remaining integration over k leads to the integral +∞ −∞ dk exp ikx − 1 2 (k − po 2 2 ) δ − i k2 m (t − t0 ) (3..52) to the case that the initial state of a 1-dimensional free particle ψ(x0 .

In order to further characterize the solution (3.62) are special solutions of the timedependent Schr¨dinger equation.69) is called an eigenvalue problem.64). (3. One can conclude that the probability density for the state (3.69) exist. One calls such states stationary states.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.63. hence. is time-independent.66) i.69) If these two equations can be solved simultaneously a solution of the type (3. We may note in passing that solutions of the type (3. At this point we will accept that solutions φ(r) of the type (3.. if f (t) = eiα . (3.62) is |ψ(r. i. however. namely f (t) = f (0) exp − i Et . (3. The identity (3. In fact..19). solutions exist only for a set of discrete E ˆ values. in general. This can hold only if |f (t)| is time-independent. α ∈ R . according to (3. 2.70) The task of ﬁnding solutions φ(r) which solve (3. We will encounter many such problems in the subsequent Sections. n = 1.62) do exist and we will characterize the two factors of the solution f (t) and φ(r). As pointed out above. .e. We denote the corresponding solution by φE (r).e. one factor depending only on the time variable and the other only on the space variables are called separable in space and time.62) which consist of two factors.42) the ﬂux j(r.62) we insert it into (3. for functions required to obey boundary conditions (3. We must postulate therefore ∂t f (t) ˆ H φ(r) = E f (t) = E φ(r) . .62) have the particular property that the associated probability distributions are independent of time. the solutions o (3..31). (3.62) exists. h1 (r) = φ(r). t) vanishes on the surface of ∂Ω. by no means all solutions are of this type. It is important to realize that the separable solutions (3. At this point we state without proof that. t)|2 = |φ(r)|2 . i. t)|2 = |f (t)|2 Ω d3 r |φ(r)|2 (3. a common case is Ω = Ω∞ and the ‘natural boundary condition’ (3. We will demonstrate that solutions of the type (3.67) (3. We have then shown that ψ(r. It follows then from (3.71) . and h2 (r) = H φ(r). often only for a discrete set of values En . This yields an expression g1 (t) h1 (r) = g2 (t) h2 (r) (3.68) ˆ where g1 (t) = i ∂t f (t). We want to demonstrate this property now.45) that the total probability d3 r ρ(r.27) holds and. g2 (t) = f (t). . t) = f (0) exp − i E φE (r) where ˆ H φE (r) = E φE (r) . It follows from the observation that for the solution space considered (3.60 Schr¨dinger Equation o and aﬀect only the spatial wave function φ(r). t) = Ω Ω d3 r |ψ(r. It turns out that a solution for f (t) can be found for any E. 3.e.65) is constant. for the eigenvalue problems in the conﬁned function space. the eigenvalues of the operator H.

in fact. The ﬂux corresponding to (3. (3. = E.71) are. of course. E ∈ R.74) is actually best labelled by an index k. one can interpret then k as the magnitude of the momentum of the particle. We want to prove now that the eigenvalues E which arise in the eigenvalue problem (3. The resulting total energy values E are positive. particles scattered of a potential. t) = exp − E t 2 φE (r) . We will wave this assumption as we always need to do later whenever we deal with unbound particles. real.. o According to (3. The corresponding stationary solution ψ(r. 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) . We will show in Section 5 that E can be interprerted as the total energy of a stationary state.3. k ∈ R 2 k2 φk (r) = N exp i k · r .e.71) this yields E Ω d3 r φ∗ (r) φE (r) = E ∗ E Ω d3 r φE (r) φ∗ (r) . hence.19. . m (3. a property which is to be expected since the energy is purely kinetic energy which. should be positive. Stationary State of a Free Particle We consider now the stationary state of a free particle described by i ψ(r.73) from which follows E = E ∗ and. As a result we cannot postulate in the present case that wave functions are localized and normalizable.74) The classical free particle with constant energy E > 0 moves without bounds in the space Ω∞ .66) E must be real.20).76) according to (3.75) 2m One can ascertain this statement by inserting the expression for φk (r) into the eigenvalue problem posed in (3. 3.72) According to (3.4: Free Particle 61 is a solution of the time-dependent Schr¨dinger equation (3. E (3.g. Obviously. − 2 2m φ(r) = E φ(r) .28) for the special case that f and g in (3. e. i. t) = exp − i 2 k2 2m t exp i k · r (3. t) = |N |2 k . (3.28) both represent the state φE (r).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 .74) using exp i k · r = ik exp i k · r . E (3.43) is j(r. The solution φE (r) corresponding to the eigenvalue problem posed by (3.76) has kinetic energy 2 k 2 /2m.

3. 5 . f continuous. f ∈ F1 = {f : [−a. Solutions of the Schr¨dinger Equation in F1 o The time–dependent solutions satisfy i ∂t ψ(x. φ(±a) = 0 . 3. hence. a] ⊂ R → R.81. (3.82) have the property that either both boundary conditions are satisﬁed or none. (3. that the solutions obey this symmetry as well. so-called even solutions obeying φ(x) = φ(−x) φE (x) = A coskx . Hence.o) (a) = 0 and φE (e. .o) (−a) = 0 (3. we have to consider only one boundary condition.62 Schr¨dinger Equation o 3. t) = exp(−iEt/ )φE (x) where φE (x) is determined by 2 d2 − φE (x) = E φE (x) . i. 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.82) One can readily verify that (3.80) need to be satisﬁed are φE (e.81. therefore. a] ⊂ R assuming that the potential outside of this interval is inﬁnite.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. 2m dx2 2 (± ) = } (3.81) they are kn = nπ .80) 2m dx2 We note that the box is symmetric with respect to the origin. t) = − d2 ψ(x.82) satisfy the diﬀerential equation in (3. In case of the even solutions (3. 2a n = 1. n ∈ N.83) The solutions (3. We assume. a}.80). t) . and so-called odd solutions obeying φ(x) = −φ(−x) φE (x) = A sinkx.78) (3. two types of solutions. (3. We can expect. let say the one at x = a. (o) 2 k2 (e) 2 k2 2m = E (3.79) The stationary solutions have the form ψ(x.81) 2m = E. It turns out that this boundary condition can only be satisﬁed for a discrete set of k–values kn . We will refer to this as a particle in a one-dimensional ‘box’.84) .. The boundary conditions which according to (3. 3. . a] ⊂ R which vanish on the surface ∂Ω1 = {−a.e. .

This condition implies for the even states +a |An |2 −a dx cos2 nπx = |An |2 a = 1 . .. 6 . i.) The Energy Spectrum and Stationary State Wave Functions The energy values corresponding to the kn –values in (3. 3. 3 .5: Particle in One-Dimensional Box since for such kn 63 nπa nπ = cos 2a 2 In case of the odd solutions (3.92) and for the odd states +a |An |2 −a dx sin2 nπx = |An |2 a = 1 . (3. x) = An sin n nπx .89) nπx .82) only the kn –values cos(kn a) = cos kn = nπ . (3.84.82). 3. (3.88) where the energies for odd (even) n–values correspond to the even (odd) solutions given in (3. (3.93) The normalization constants are then An = (3.81) and (3.86). 2a n = 2. 3.o) (a.81.94) .91) holds. . . (3. (3.86) satisfy the boundary condition since for such kn nπa sin(kn a) = sin 2a = sin nπ 2 = 0. . n is assumed to be even. 2a n = 1. . . 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 . 5 .86). The wave functions for the ﬁve lowest energies En are presented in Fig. according to (3. x)|2 = 1 n (3.e. 2a 1 . a n = 2. 6 .87) (Note that.90) 2a The wave functions represent stationary states of the particle in a one-dimensional box. φ(e) (a. n = 1. x) = An cos n and φ(o) (a. are En = 2π2 8ma2 n2 . = 0. . (3.1). according to the dispersion relationships given in (3. . 3. 2. . respectively. . . By counting the number of their nodes one can determine the energy ordering of the wave functions. 4. 4.3. 6 . n = 2. It is desirable to normalize the wave functions such that +a −a dx |φ(e. .85) (3. 2a n = 1. 4. (3. . 5 .82).

5 of particle in a box.1: Eigenvalues En and eigenfunctions φn (e. 2. x) for n = 1. 3. 4.64 Schr¨dinger Equation o E h2 ] [ 8 m a2 25 ¢ 16 9 4 Figure 3. ¡ 1 -a a x .o) (a.

3. one obtains for n.90.101) +a −a dx cos (n−m)πx − cos (n+m)πx 2a 2a and. i. n (3. . the integrals vanish. the integral vanishes. 5 . 2. and (iii) the mixed case. 4. m odd. 3. it holds: f |f Ω1 = 0 → f (x) ≡ 0. f. . x) (3. n = m.96) f |g Ω1 = −a dxf (x) g(x) . In case of n. 2a 2a (3. the integrand is odd. .103) We will show in Section 5 that the property of a scalar product do indeed apply.102) holds according to (3. In particular.98) The latter property is obviously true for n = m.e. The latter case leads to integrals φn |φm Ω1 = 1 a +a dx cos −a nπx mπx sin . (3. (3. Similarly. x) .} where φn (a.5: Particle in One-Dimensional Box The Stationary States form a Complete Orthonormal Basis of F1 We want to demonstrate now that the set of solutions (3. In case of n = m we have to consider three cases. the integral arises φn |φm 1 a Ω1 = 1 a +a nπx −a dx cos 2a cos mπx = 2a (3. .97) form an orthonormal basis set.89. Obviously. i. .94) B1 = {φn (a. In fact.e. for n = 2. 6 . .98) ∞ f |f 5 Ω1 = n=1 d2 . m even φn |φm 1 a Ω1 = 1 a +a nπx −a dx sin 2a sin mπx = 2a (3. . for ∞ f (x) = n=1 dn φn (a. a] are N. Because of the property (3. n = 1.95) 1 a cos nπx 2a sin nπx 2a for n = 1. Hence we need to consider only the ﬁrst two cases. 3. it holds φn |φm Ω1 = δn m . this integral vanishes. (i) n. m both odd.3. . m both even.. N ≥ 1. x) = together with the scalar product5 +a 65 (3.99) Since in this case the integrand is a product of an even and of an odd function.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. g ∈ F1 (3.. too. hence.98) the elements of B1 must be linearly independent. 3. (ii) n.

t0 ) = +a ∞ i n=1 − En φn (a. We like to show ﬁnally that the basis (3. . in (3. 2. . t0 ) in terms of eigenfunctions φn (a. . Hence. n = 2. t0 ) = n=1 dn φn (a. 5 . . there exist real constants {an . i. x0 ) ψ(x0 . t0 ) (3. t0 ) . . . 3. Accordingly.} such that ∞ ∞ nπx nπx ˜ = + an sin (3. m = 1. −a (3.98) one obtains +a dx0 φm (a. t) = n=1 φn (a.107) Inserting this into (3. We have then shown that any f corresponding to elements of F1 can be expanded in terms of elements in B1 .105) to the interval [−a. the coeﬃcients an . For this purpose we expand ∞ ψ(x.95) is also complete. (3. and bn . . ( ) = ([ + ] − ) (3. t0 ) = dm . 3. 1.106) and generalizing to t ≥ t0 one can write ∞ +a ψ(x. x) .} and {bn ..108) into the Schr¨dinger equation yields o +a ∞ n=1 φn (a. .106) Using the orthonormality property (3. .108) where the functions cn (t) are to be determined from the Schr¨dinger equation (3. In the present case. x0 ) ψ(x0 . .95. x) cn (t) −a dx0 φn (a. however. .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. (3. 4. a] yields then also an expansion for any element in F1 and B1 is a complete basis for F1 . i. 3. (3. x0 ) ψ(x0 .. x0 ) ψ(x0 .e. Restricting the expansion (3.96).66 Schr¨dinger Equation o f (x) ≡ 0 implies f |f Ω1 = 0 which in turn implies dn = 0 since d2 ≥ 0. . This implies that only the trigonometric functions which are elements of B1 enter into the Fourier ˜ series.79) requiring the o initial condition cn (t0 ) = 1 . For this purpose we extend the deﬁnition of the elements of F1 to the whole real axis through ˜ f : R → R.105) must vanish. 2. they can be expanded in terms of a Fourier series.78) can be expressed as a linear combination of the elements of B1 deﬁned in (3. n = 0.110) . any element of the function space F1 deﬁned in (3. n = 1. n = 1.104) ˜ where [y]a = y mod2a. The functions f are periodic with period 2a. x) ∂t cn (t) −a dx0 φn (a.95) is an orthonormal basis. t) at times t > t0 . such demonstration can be based on the theory of Fourier series.109) Insertion of (3. Demonstration of completeness is a formidable task. . . x) cn (t) −a dx0 φn (a. x) to obtain an expression for ψ(x.e. Evaluating the Propagator We can now use the expansion of any initial wave function ψ(x. It follows that B1 n deﬁned in (3.

2 presents the probability distribution of the particle at subsequent times. The respective initial condition is φ(x. t|x0 . Figure 3.112) Equations (3. The interference pattern begins to ‘smear out’ ﬁrst. Note that the propagator has been evaluated in terms of elements of a particular function space F1 .e. k0 = . t0 ) (3.5). t0 ) = ∞ φn (a. The particle moves then to the left. rather than by an integral (3. x) and integrating over [−a.114) This expression has the same form as postulated in the path integral formulation of Quantum Mechanics introduced above.98).108. in (2. the elements of which satisfy the appropriate boundary conditions. a] yields. i. 3. σ = a 15 .5: Particle in One-Dimensional Box Multiplying both sides by φm (a. according to (3.112) determine now ψ(x.111) The solutions of these equations which satisfy (3. t0 ) ψ(x0 . . Often the propagator φ(x. being reﬂected at the right wall. 67 (3. t0 ) itself is a solution of the time-dependent Schr¨dinger o equation (3. (3.54) as in the case of the free particle system which does not exhibit any bound states. It is of interest to note that φ(x.79) which lies in the proper function space (3. x0 ) . t) = −a dx0 φ(x. i ∂t cm (t) = − Em cm (t) .3. but the collision with the left wall leads to new interference eﬀects. This solution can be written +a ψ(x. In the present system which is composed solely of bound states the propagator is given by a sum. cm (t0 ) = 1 . x) exp n=1 i − En (t − t0 ) φn (a. t0 ).109) are cm (t) = exp i − Em (t − t0 ) . In case that diﬀerent boundary conditions hold the propagator will be diﬀerent as well. The last frame shows the wave front reaching again the right wall and the onset of new interference.113) where φ(x. One can recognize that the particle moves ﬁrst to the left and that near the right wall of the box interference eﬀects develop.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. t0 ) is also referred to as a Greens function..113). t0 ) = 1 2πσ 2 1 4 exp − x2 0 + i k0 x0 2σ 2 . 4 a (3.78). t|x0 . t0 ) = δ(x − x0 ) as can be readily veriﬁed using (3. t) for any initial condition ψ(x. t|x0 .114) the representation of the propagator for a particle in a box with inﬁnite walls. t0 |x0 . Example of a Non-Stationary State As an illustration of a non-stationary state we consider a particle in an initial state ψ(x0 . t|x0 . We have identiﬁed then with (3. (3.

96 m a h2 |Ψ|2 2 t = 1.68 Schr¨dinger Equation o |Ψ|2 2 t = 0.00 m a h2 |Ψ|2 2 t = 0.48 m a h2 |Ψ|2 2 t = 0.72 m a h2 1 πσ 1 πσ -a a -a a |Ψ|2 2 t = 0.24 m a h2 1 πσ k = 15 a 1 πσ -a a -a a |Ψ|2 2 t = 0.68 m a h2 1 πσ 1 πσ -a a -a a |Ψ|2 2 t = 1. t)|2 for a particle in a box starting in a Gaussian distribution with momentum 15 /a.92 m a h2 |Ψ|2 2 t = 2.16 m a h2 1 πσ 1 πσ -a a -a a Figure 3.44 m a h2 |Ψ|2 2 t = 1.2: Stroboscopic views of the probability distribution |ψ(x. .20 m a h2 1 πσ 1 πσ -a a -a a |Ψ|2 2 t = 1.

one can express H 3 φ(x1 .95). f continuous. x3 )T ∈ Ω. a3 ] ⊂ R = {(x1 . x1 = ±a1 } ∪ {(x1 . a2 ] ⊗ [−a3 . We seek then solutions of the time-dependent Schr¨dinger equation o ˆ i ∂t ψ(x. i. . ˆ H = − 2 (3. i. a1 ] ⊗ [−a2 . 2.120) ˆ Since the Hamiltonian H is a sum of operators O(xj ) each dependent only on a single variable. x2 .e. )T ∈ ∂ } . ˆ = O(x1 ) + O(x2 ) + O(x3 ). x2 . hence. of the space F1 is that the elements of the basis set can be enumerated by integer numbers.117) 2m 2 2 2 ∂1 + ∂2 + ∂3 (3.122) . t) = H ψ(x1 . x2 . x3 ) = E φE (x1 .116) and obeys the partial diﬀerential equation ˆ H φE (x1 . x2 .119) where φE (x1 . t) . (3. x2 . Placing the origin at the center and aligning the x1 .e. x3 ) = j=1 φ(j) (xj ) (3. x3 )T ∈ Ω. (3. x2 = ±a2 } ∪ {(x1 . can be counted. The description employed a space of functions F1 deﬁned in (3. x3 = ±a3 } . x3 .121) where − 2 2m 2 ∂j φ(j) (xj ) = Ej φ(j) (xj ) . x3 )T ∈ Ω.78). 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 ) .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. A complete basis of F1 is given by the inﬁnite set B1 (3. ) = ∀( . This property allowed us to evaluate the propagator (3. . x3 ) (3.. x2 . 3 (3. 3. x2 . We have deﬁned in the space F1 a scalar product (3. j = 1. 2a3 . x2 .6 Particle in Three-Dimensional Box We consider now a particle moving in a three-dimensional rectangular box with side lengths 2a1 . ( . x2 . The corresponding solutions have the form i ψ(x1 .. t) = exp − E t φE (x1 .97) with respect to which the eigenfunctions are orthonormal. x3 . x2 . φ(j) (±aj ) = 0 .116) where Ω is the interior of the box and ∂Ω its surface Ω ∂Ω = [−a1 . An important property of this basis and. x3 ) is an element of the function space F3 deﬁned in (3.118) which are stationary states. 2a2 .114) in terms of which the solutions for all initial conditions can be expressed.3.

n3 = 1. 3. n1 .n2 . a1 = a2 = a3 = a then rotation around the x1 . x1 . . If two or all three orthogonal sides of the box have the same length further symmetry operations leave the system unaltered. x2 .. a2 .n2 . a3 .n3 ) (a1 .122) with (3. a2 . x1 .n2 . x3 ) = 1. . 2.80) shows that the solutions of (3.n2 . a2 . hence. n 2 . Comparing (3. in the present case . namely rotation by π around the x1 . n1 . a3 . . x3 –axes by π/2 also leaves the system unaltered. r) exp i − E(n1 n2 n3 ) (t − t0 ) φ(n1 .n2 . (3. x3 ) . This symmetry has been exploited in deriving the stationary states. . 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. For example. x2 . n2 . x2 . a2 . i.123) 8m n2 n2 n2 1 2 3 + 2 + 2 a2 a2 a3 1 . x2 ) φn3 (a3 .122) are given by (3. . a3 .70 Schr¨dinger Equation o and E1 + E2 + E3 = E.124) The same considerations as in the one-dimensional case allow one to show that B3 = {φ(n1 .126) Symmetries The three-dimensional box conﬁning a particle allows three symmetry operations that leave the box unchanged. (3. 3. . 3.} ∞ (3. . n2 . a2 . n3 = 1. n 3 and E(n1 . This additional symmetry is also reﬂected by degeneracies in the energy levels. r0 ) . (3. x3 ) n1 . .96) and.125) is a complete orthonormal basis of F3 and that the propagator for the three-dimensional box is φ(r. the solutions of (3.n3 ) (a1 . a3 . Actually.n3 ) = 2π2 = φn1 (a1 . t|r0 .120) can be written φ(n1 . t0 ) = n1 .n3 ) (a1 . x1 ) φn2 (a2 .n2 . if all three lengths a1 . a3 are identical. x3 –axes. Such degeneracies are always a signature of an underlying symmetry.e. . 2.n3 ) (a1 . 2.n3 =1 φ(n1 . x2 .

n3 ) = (3. However. Hence.2) (a. r) + γ φ(2.2) (a. however. a. a. . due to the hermitian character of ˆ H. r) + β φ(2. in case of degenerate states one cannot necessarily expect that stationary states are orthogonal. a. a.2.2. However. e. φ(1. n2 .2.1. 3. We consider the degeneracy of the energy E122 which is due to the identity E122 = E212 = E221 . a. 6 This is a result of linear algebra which the reader may ﬁnd in a respective textbook. 1.2) (a.6: Particle in Three-Dimensional Box 71 ‘accidental’ degeneracies also occur. a. 5) as shown in the Table above. to construct n orthogonal stationary states6 ..1) (a. One particular aspect of the degeneracies illustrated in the Table above is worth mentioning. The origin of this degeneracy is. 3) (1.g. r) (3. a. this linear combination o is not necessarily orthogonal to other degeneraste states.127) ˜ ˆ ˜ obeys the stationary Schr¨dinger equation H φ(r) = E122 φ(r). in case of an n–fold degeneracy it is always possible. Any linear combination of wave functions ˜ φ(r) = α φ(1. the identity 32 + 32 + 32 = 12 + 12 + 52 . r).3. a. for example. for (n1 .

72 Schr¨dinger Equation o .

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

is the commutation property 1 1 (4. This property states that p and x obey an ˆ ˆ algebra in which the two do not commute.7) which is why these operators are of interest to us.7) of the Hamiltonian. only for a discrete set of E values can the boundary conditions (4. beside the representation (4. In the following algebraic solution of (4. therefore. the spectrum of ˆ H in (4. t) where ˆ H φE (x) = E φE (x) .6) the ﬁrst being a diﬀerential operator and the second a multiplicative operator. If this restriction would not apply and all functions f : R → R would be admitted. lim ( ) = } (4. the commutator has a simple form. x] = p ˆ . The Hamiltonian H can be expressed in terms of the operators acting on the space (4. t) = exp(−iEt/ )φE (x. (4. however. i dx x = x ˆ (4.5) ˆ H = 1 1 2 p + m ω 2 x2 ˆ ˆ 2m 2 (4.1 Creation and Annihilation Operators The Hamiltonian operator (4. The stationary states of the harmonic oscillator have been considered already in Chapter 2 where the corresponding wave functions (2.235) had been determined. the natural boundary conditions apply x→±∞ lim φE (x) = 0 . It is important to keep in mind this restriction of the space. The operators act ˆ on the space of functions N1 deﬁned in (4.1) of the form ψ(x. In the framework of the Schr¨dinger equation the stationary o states are solutions of (4. We will point out explicitly where assumptions are made which built on this restriction. act.3) would be continuous and the eigenfunctions φE (x) would not be normalizable.5) →±∞ where C∞ denotes the set of functions which together with all of their derivatives are continuous.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.4) be satisﬁed.3) Due to the nature of the harmonic potential which does not allow a particle with ﬁnite energy to move to arbitrary large distances.4) Equation (4. however. In order [ˆ.3) can be solved for any E ∈ R. The cardinal property exploited below. ∈ C∞ .2) can be expressed in terms of the two operators p = ˆ d .5). 4. in which the operators used below. (4.8) i which holds for any position and momentum operator.74 Linear Harmonic Oscillator In the following we consider ﬁrst the stationary states of the linear harmonic oscillator and later consider the propagator which describes the time evolution of any initial state. all stationary states of the harmonic oscillator must be bound states and.

hence.15) 1 ˆ 1 H + 1 . a ˆ 1 (4.11) a− a+ = ˆ ˆ (4. x] f (x) = p ˆ d d x f (x) − x f (x) = f (x) = 1 f (x) .14) 1 i mω 2 x + ˆ p2 + ˆ [ˆ. 2mω dx a− = ˆ mω x + ˆ 2 d .16) (4.7) and (4. One obtains [ˆ. 1 ω 2 (4. 1 i dx i dx i i (4.8). 1 ω 2 (4. a+ ] = 1 .14) and (4. Being guided by the identity for scalar numbers (b − ic)(b + ic) = b2 − i(cb − bc) + c2 = b2 + c2 (4. 4. in fact. Since a+ and a− are ˆ ˆ + and a− since operators and not scalars we cannot simply expect that the identy (4. 1 2 (4. to factor H/ ω.10) we deﬁne a+ = ˆ mω i x − √ ˆ p.15) together lead to the commutation property (4.8) yield a− a+ = ˆ ˆ Similarly one can show a+ a− = ˆ ˆ 1 ˆ 1 H − 1 . ˆ ˆ .8) we recall that all operators act on functions in N1 and.8) a ˆ ˆ ˆ ˆ ˆ implies [ˆ− .9) From this follows (4.4. Before we continue we like to write (4.10) holds for a ˆ ˆ [ˆ− . p ˆ 2 2m ω 2 (4. x] .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 . the action of [ˆ. In fact. 2mω dx (4.14. ˆ 2 2m ω a− = ˆ mω i x + √ ˆ p.13) (4. a+ ] = a− a+ − a+ a− does not necessarily vanish.12) To prove this commutation property we determine using (4. x] p ˆ on such functions must be considered.1: Creation and Annihilation Operators 75 to derive (4. Our next step is an attempt to factorize the Hamiltonian (4.7) assuming that the factors are easier to handle than the Hamiltonian in yielding spectrum and eigenstates.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 . ˆ 2 2m ω (4.18) It is of interest to note that the operators a+ and a− are real diﬀerential operators. the commutator property (4.11) ˆ The reader may note that we have attempted.12).

4. ˆ (4. The derivation will be based solely on the properties (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) .17.22. 4. 4.76 Relationship between a+ and a− Linear Harmonic Oscillator The operators a+ and a− are related to each other by the following property which holds for all ˆ ˆ functions f. 4.17. This property of operators is investigated more systematically in Section 5. Using (??) we like to state (4.16.22) Similarly one can show using again (4. it holds that for a stationary state φE (x) of energy E deﬁned through (4. . ˆ (4.21) is available.12.19) This property states that the operators a+ and a− are the adjoints of each other. 4.19) in the form1 f |ˆ+ g a Ω∞ = g|ˆ− f a Ω∞ .20) ˆ In the following we will determine the spectrum of H and its eigenstates.23) Together.24) − m This property states that the operators in the function space N1 are the hermitian conjugate of each other.19).17.16. 4. 4.16. (4. (4. The property ˆ ˆ follows directly from (??). 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) 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) .21) ˆˆ H a+ φE (x) ω 1 ) a+ φE (x) 1 ˆ 2 ω = a+ ( ω a− a+ − ˆ ˆ ˆ 1 + ω 1 ) φE (x) 1 1 2 ˆ = a+ ( H + ω ) φE (x) ˆ = ( ω a+ a− + ˆ ˆ = (E + ω ) a+ φE (x) . one obtains using (4. g ∈ N1 +∞ +∞ dx f (x) a+ g(x) = −∞ −∞ dx g(x) a− f (x) . In fact.21) a− φE (x) is ˆ a stationary state of energy E − ω and a+ φE (x) is a stationary state of energy E + ω. ˆ The results (4.

be found.30) Assuming that φ0 (y) does not vanish anywhere in its domain ] − ∞. Another name is creation (ˆ+ ) and annihilation (ˆ− ) operators since these operators a a create and annihilate vibrational quanta ω.25) φ0 (y) = 0 (4. one expects that E = 0 should be a lower bound.6.25) can. (4. a state φ0 (x) which obeys the property a− φ0 (x) = 0 ˆ (4. two diﬃculties: (i) one needs to know at least one stationary state to start the construction.33) .25) needs to be normalizable in order to represent a bound state of the harmonic oscillator. (4. E = E . E . in fact. +∞[ one can write (4.2 Ground State of the Harmonic Oscillator d + y dy A suitable solution of (4.5). Using (4.25) on one side would lead to termination of the sequence E0 + m. however. on ˆ the other side such a state is itself an eigenstate of H as can be shown using (4. 4.) ˆ ˆ 0 = φE |HφE Ω∞ − φE |HφE Ω∞ = ( E − E ) φE |φE Ω∞ .11) one can rewrite (4. (4. 2 2 (4. (4. φ0 (y) dy dy the solution of which is 1 lnφ0 (y) = − y 2 + c0 2 φ0 (y) = c0 exp 1 − y2 2 . i.28) 4. In fact.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. φ0 (x) should be an element of the function space N1 deﬁned in (4.e.4. Let φE (x) and φE (x) be two normalized stationary states corresponding to two diﬀerent energies E.32) for some constant c0 or (4.27) Since E = E one can conclude φE |φE Ω∞ = 0. There arise. For f (x) = φE (x) and g(x) = φE (x) in (??) follows (Note that according to (??) the eigenvalue E is real. (ii) the construction appears to yield energy eigenvalues without lower bounds when..17) ˆ H φ0 (x) = ( ω a+ a− + ˆ ˆ 1 ω 1 ) φ0 (x) = 1 ω φ0 (x) .26) Of course. in fact. m ∈ Z when m is decreased.31) (4. the solution φ0 (x) of (4. It turns out that both diﬃculties can be resolved together.29) d 1 d φ0 (y) = = lnφ0 (y) = − y .29) where y = mω x. 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).

We a can. i.78 Linear Harmonic Oscillator This solution is obviously normalizable. it holds ˆ H φn (x) = ω(n + 1 ) φn (x) . a state conﬁned to the minimum corresponding to the classical state of lowest energy. the potential energy contribution which for a state δ(x). We will consider this point more systematically in Section 5.36) n times.e. The reason is that in the Hamiltonian (4.24) can then be written as follows En = ω(n + 1 ). m ∈ Z which properly terminate at some minimum value E + m0 ω ≥ 0. We recall that according to (4. (4.39) We notice that the ground state wave function φ0 (x) as well as the operators (ˆ+ )n are real. . (4.36) assumes a functional form such that both terms together assume a minimum value. i. .e. . 1.25) admits only one solution there is only one set of states with energy E + m ω. (4. 1. i.e. therefore. 2. The conventional normalization condition φ0 |φ0 reads +∞ +∞ Ω∞ = 1 (4. mω (4. . .38) These states correspond to the energy eigenvalues (4. 2 n = 0. the so-called excited states.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. Such states need to be suitably normalized for which purpose we introduce a normalization constant cn φn (x) = cn a+ ˆ n φ0 (x) .37) It is most remarkable that the energy 1 ω of the ground state is larger than the lowest classically 2 allowed energy E = 0. . . (4.37) above the ground state energy. 2.. The ground state (4. choose the normalization constants cn and the functions φn (x) real as well. 1.36) Since the ﬁrst order diﬀerential equation (4.. which for a narrowly localized state yields a large positive value.2) there are two competing contributions to the energy. 2. . we construct the states for n = 1. For this purpose we apply the operator a+ to the ˆ ground state (4. would yield a vanishing contribution. we can now construct the stationary states corresponding to energies (4. and the kinetic energy contribution. . .35) The appropriate ground state is φ0 (x) = mω π 1 4 exp − mωx2 2 . n = 0. 4. . The set of allowed energies of the oscillator according to (4.39).. .. 2. .26) the energy value associated with this state is 1 ω.34) dx |φ0 (x)|2 = |c0 |2 −∞ −∞ dx exp − mωx2 2 = |c0 |2 π . 2 n = 0.

ˆ (4.42) ˆ Using (4. (4.24) holds in analogy to (4. (4. they obey φn |φn = δn n . √ a+ φn (x) = n + 1 φn+1 (x) . βn must be real as well.3: Excited States 79 We need to determine now the normalization constants cn . (4. according to (4.40) φn−1 (x) = βn a− φn (x) ˆ (4. 1. 4.45) form an orthonormal set.40–4. (4.44) One can conclude then that the stationary states of the oscillator are described by the functions 1 φn (x) = √ a+ ˆ n! n φ0 (x) . 2.e. . √ a− φn (x) = n φn−1 (x) . Ω∞ 2 = βn a− φn |ˆ− φn ˆ a Ω∞ 2 = βn φn |ˆ+ a− φn a ˆ Ω∞ . . according to (4.48) (4. 2.4.14) together with H φn (x) = ω(n+ 1 ) φn (x) leads to the condition (note that we assumed 2 φn (x) to be normalized) 2 2 αn ( n + 1 ) φn |ˆ− a+ φn Ω∞ = αn ( n + 1 ) = 1 a ˆ √ From this follows αn = 1/ n + 1 and. For n = 0 holds c0 = 1. .28) and the construction (4.46) According to (4.47) for some suitable constants βn .50) φn (x) . 1.45) We like to note that these functions according to (4. . Since a− is a real diﬀerential operator [see (4. (4.49) (4..40).15 .39) yield √ From this follows βn = 1/ n and.47). 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. i. 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 = 0.40) = 1. n.41) Employing the adjoint property (4.20) yields 2 αn φn |ˆ− a+ φn a ˆ Ω∞ = 1. To determine βn we note using (4. .51) . (4. .20) 1 = φn−1 |φn−1 Equations (4. n = 0.43) (4.18)] and since the ˆ φn (x) are real functions. We consider then the situation that we have obtained a properly normalized state φn (x). To determine these constants we adopt a recursion scheme. .

(4. In terms of y the ground state wave function is φ0 (y) = π − 4 e− and a+ is ˆ 1 a+ = √ ˆ 2 y − d dy .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. For this purpose we start from expression (4.46) yields a set of normalized states.52). by introducing the variable y deﬁned in (4. (4. for example.35) by a constant. . by dx = dy mω 1 4 .7 and given. by the Rodrigues formula (2.56) can be expressed in terms of Hermite polynomials Hn (y) introduced in Sect.56) Relationship to Hermite Polynomials We want to demonstrate now that the expression (4. It should be noted that √ the normalization (4. (4.40–4.57) such that one can write the stationary state wave functions of the harmonic oscillator φn (y) = − √ e 2n n! π 1 y2 2 Hn (y) .58) This result agrees with the expression (2.e.30) and employing the normalization +∞ −∞ dy φ2 (y) = 1 n (4. n–independent factor.35) rather than (4..55) 1 y2 2 (4.28) of the ground state and the deﬁnition (4. namely the square root of the Jacobian dx/dy.39).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 .7. We will demonstrate below the identity Hn (y) = e y2 2 y − d dy n e− y2 2 (4. (4.52) This normalization of the wave functions diﬀers from that postulated in (4. 2. i. 2.233) derived in Sect.45). simplifying the calculation.200).53) We will later re-introduce this factor to account for the proper normalization (4.57) which includes the factor 1/ n! according to Eqs. (4. however.

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 .30) and return later to the coordinate x.106–3.e. therefore. t0 ) = n=0 dn φn (y0 ) .59) holds for n = 0.64) which implies that (4.57) of the hermite polynomials and the deﬁnition given by (2. 1.114)]. hence.59) hold. (4.59) by induction noting ﬁrst that (4. 4.39) ∞ ψ(y0 . dy dy (4. For the sake of notational simplicity we employ initially the coordinate y as deﬁned in (4.4 Propagator for the Harmonic Oscillator We consider now the solution of the time-dependent Schr¨dinger equation of the harmonic oscillator o (4. and showing then that the property also holds for n + 1 in case it holds for n. g(y) = (−1)n+1 e y2 2 dn+1 −y2 e = dy n+1 y − d dy n+1 e− y2 2 .4: Propagator 81 To verify the realtionship between the deﬁnition (4.65) . (4. Starting point of our derivation is the assumption that the initial condition can be expanded in terms of the eigenstates φn (y) (4.200).59) dn −y2 e dy n (4.4.200) we need to verify y − which implies Hn (y) = (−1)n ey 2 d dy n e− y2 2 = (−1)n e y2 2 dn −y2 e .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 (4.2) for an arbitrary initial wave function ψ(x0 .61) and. t0 ).61) One can factor g(y) and employ (4. dy n (4. the Rodrigues formula (2..60) and. (3. i. We prove (4. (4.63) one obtains g(y) = y − d dy f (y) (4.1. Our derivation will follow closely the procedure adopted for the case of a ‘particle in a box’ [see Eqs. 4.

5).. 68. sec:harm. t0 ) ψ(y0 .225) which allows one to derive a generating function for products of Hermite polynomials which can be applied to the r. Employing orthogonality condition (4.70) is the propagator of the linear harmonic oscillator.Lebedev (Prentice Hall. t1 ) = n=0 dn φn (y) cn (t) (4. 1 t = e−iω(t1 −t0 ) (4.h. y0 . . (4.66) One can extend expansion (4. pp. In order to prove the equivalence of (4.N.7. 4. t) = n=0 Hn (y)e−y /2 Hn (y0 )e−y0 /2 n √ t .70). t0 ) which is an element of N1 deﬁned in (4.39) holds cn (t1 ) = exp −i ω ( n + 1 ) ( t1 − t0 ) 2 . t|y0 . 1965) Sect. (4. t) = π −3/2 exp 2 2 y 2 y0 + 2 2 2 +∞ +∞ ∞ du −∞ −∞ dv n=0 1 ( −2tuv )n × n! (4. y0 . of (4. t0 ) = n=0 φn (y) φn (y0 ) tn+ 2 .108–3.46) the expansion coeﬃcients dn are +∞ dn = −∞ dy0 ψ(y0 . a supposition which is not proven here2 .71) Applying (??) to express Hn (y) and Hn (y0 ) yields exp(−2tuv) w(y. 2 A demonstration of this property can be found in Special Functions and their Applications by N.69) where ∞ φ(y. t) (4. 2. t1 |y0 .82 Linear Harmonic Oscillator Such expansion is possible for any f (y0 ) = ψ(y0 .15. We consider for this purpose the following expression for |t| < 1 ∞ w(y.65) to times t ≥ t0 through insertion of time-dependent coeﬃcients cn (t1 ) ∞ ψ(y.112) and (4. t0 ) φn (y0 ) . 2n n! π 2 2 (4.68) Altogether one can express then the solution +∞ ψ(y. We want to demonstrate now that this propagator is identical to the propagator (2. t1 ) = −∞ dy0 φ(y. this is an excellent textbook from which we have borrowed heavily in this Section.s.67) for which according to (3. Inc.70) and (2. Englewood Cliﬀs. For this purpose we start from the integral representation (2.J.147) we employ the technique of generating functions as in Sect.. N.72) × exp( −u − v + 2iyu + 2iy0 v ) .147) for the harmonic iscillator determined in Sect.

t0 ) = i φ(x. t0 ) = exp mω 2iπ sinω(t1 − t0 ) 1 2 × . y0 .70) is. t) = π −3/2 exp +∞ 83 y2 y2 + 0 2 2 +∞ dv exp( −v 2 + 2iy0 v ) × −∞ × −∞ du exp( −u2 − 2u(tv − iy) ) . t0 ) (4.75) Applying (4. a Re a2 > 0 (4. y0 . (4.78) in terms of the coordinates x and x0 .73) The incomplete Gaussian integral +∞ −∞ dx e−a 2 x2 √ − 2bx = π b2 /a2 e .77) The sum in (4. identical to the generating function (4. −∞ (4.e. t1 |y0 .4: Propagator Carrying out the sum on the r. This requires that we employ (4. t0 ) exp i 2 1 2 (y + y0 ) G(t1 .30) to replace y and y0 and that we multiply the propagator by mω/ . t0 ) = and 1 2 i π F (t1 .78) 1 − e−2iω(t1 −t0 ) = sinω(t1 − t0 ) 2 i e−iω(t1 −t0 ) (4. (4. it holds φ(y. indeed. i. t0 ) − i y y0 2 F (t1 .71) of w(y. i. .81) imω 2 sinω(t1 − t0 ) (x2 + x2 ) cosω(t1 − t0 ) − 2 x x0 0 This result agrees with the propagator (2. (4.76) One can express this in terms of the stationary states (4. t0 ) = where F (t1 .79) 1 + e−2iω(t1 −t0 ) cosω(t1 − t0 ) = .e. t|x0 .s.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 .147) derived by means of the path integral description. by a factor 4 mω/ for both φn (y) and φn (y0 ). √ πexp( t2 v 2 − 2iytv − y 2 ) (4. one obtains w(y. t) = π −1 exp − 2 y 2 y0 + 2 2 +∞ dv exp(−(1 − t2 ) v 2 − 2i (yt − y0 ) v) .4.80) −2iω(t1 −t0 ) sinω(t1 − t0 ) 1−e We can ﬁnally express the propagator (4.74) applied once results in w(y..74) a second time yields ﬁnally together with the deﬁnition (4. 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 = .h. The resulting propagator is G(t1 .77).. (4. y0 .

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

f (ˆ+ )] + f (ˆ+ ) a− ˆ a a a a ˆ which implies that in order to prove (4. It holds a a euˆ f (ˆ+ ) φ0 (y) = f (ˆ+ + u) φ0 (y) .88) we need to show actually [ˆ− . An example is the so-called shift operator exp(uˆ− ). Hence.94) This identity follows from (4. (ˆ+ )n a+ ] = (ˆ+ )n [ˆ− . ˆ a a a ˆ In particular. The following important property holds a a− f (ˆ+ ) = f (1) (ˆ+ ) + f (ˆ+ ) a− . The ﬁrst case is trivial.90) holds for fn . Note that ˆ a an immediate consequence of (4. f (ˆ+ )] = f (1) (ˆ+ ) . a a a 85 (4.83) and can be proven for all powers fn by induction and then inferred for all proper functions f (ˆ± ). For fn+1 follows then [ˆ− .87.88) is (ˆ− )n f (ˆ+ ) φo (y) = f (n) (ˆ+ ) φ0 (y) . The convergence of the Taylor expansion ascertains then that (4..93) one can reduce all operators acting on some proper state function by operators acting on the state φ0 (y). f1 (ˆ+ )] = [ˆ− . fn+1 (ˆ+ )] a a = [ˆ− . a a − (4. (ˆ+ )n ] a+ a a ˆ a a ˆ a a ˆ = (ˆ+ )n 1 + n (ˆ+ )n−1 a+ = (n + 1) (ˆ+ )n a 1 a ˆ a Since any function f (y) in the proper function space N1 can be expanded ∞ ∞ (1) (4. we will only assume operator functions acting on φ0 (y).89) (4. a a a ˆ 1 a Let us assume that (4. the second case follows from [ˆ− .95) (4.91) (4.90) holds for f (ˆ+ ). a a (4.96) . We note ˆ a− f (ˆ+ ) = [ˆ− .90) To prove (4. a+ ] = 1 = f1 (ˆ+ ) .87) (4. ˆ a We proceed by induction noticing ﬁrst that (4. a− f (ˆ+ ) φ0 (y) = f (1) (ˆ+ ) φ0 (y) ˆ a a which follows from a− φo (y) = 0.90) holds for f0 and for f1 .e.4.88) is a fundamental one and will be used in the following.88) (4. a+ ] + [ˆ− .90) we show that (4. property (4.90) holds for any function fn (ˆ+ ) = (ˆ+ )n which is a power of a a a+ .5: Working with Creation and Annihilation Operators and similarly for f (ˆ− ). i.92) f (y) = n=0 dn φn (y) = n=0 (ˆ+ )n a dn √ φ0 (y) n! (4. a a a We like to state the following property of the functions f (ˆ± ) a f (ˆ− ) φ|ψ = φ|f (ˆ+ ) ψ . 4. As long as the operators act on φ0 (y) one can state then that a− behaves like a diﬀerential operator with respect to functions f (ˆ+ ). a An important operator function is the exponential function.

98) for u = −α∗ and v = α yields e−α This together with (4. Below we will use the operator identity which follows for the choice of v = α in (4. hence.101) ˆ ˆ ˆ C(u) obviously obeys C(0) = 1 This results in D(0) = 1 and.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) .99) ˆ C(u) = zˆ a zˆ+ + z ∗ a− a ˆ + +z a ˆ ∗ − a euzˆ euz ˆ − u C(u) .101) 1.106) Applying (4. + (4. that the operators considered act on φ0 (y).96) is applied is a a a euˆ evˆ φ0 (y) = ev (ˆ − + + + u) a φ0 (y) = euv evˆ φ0 (y) . z ∈ C.108) . without explicitly stating this. 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.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) .103) (4.98) The related operators exp[vˆ+ ± v ∗ a− ] play also an important role.107) − α∗ a− ˆ φ0 (y) = e− αα∗ 2 a eαˆ φ0 (y) .105) a eαˆ + − α∗ a− ˆ φ0 (y) = e αα∗ 2 e−α ∗ a− ˆ a eαˆ φ0 (y) . + ∗ a− ˆ a − u euzˆ euz + ∗ a− ˆ = (4.106) yields a eαˆ + ∗ a− ˆ a a eαˆ φ0 (y) = e−αα eαˆ φ0 (y) . 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) . ∗ = exp(uzˆ a ˆ and determine its derivative d du ˆ C(u) a = zˆ+ euzˆ euz a + ∗ a− ˆ a + euzˆ z ∗ a− euz ˆ + ∗ a− ˆ = + zˆ+ + z ∗ a− a ˆ Using (4. + (4. (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) (4.97) An example in which (4. 1 + + uz ∗ a− ). We assume in the following a ˆ derivation.90) and zz ∗ = 1 we can write this d du e uzˆ+ a e uz ∗ a− ˆ a − z ∗ a− . + ∗ + (4. a a ν! (4.102) Similarly. euzˆ ˆ euz ∗ a− ˆ (4. the solution of (4. + (4.104) (4. one obtains a evˆ + − v ∗ a− ˆ a φ0 (y) = e− 2 vv evˆ e− v 1 ∗ + ∗ a− ˆ φ0 (y) (4.

s.h.111) yields π − 4 exp − 1 u2 2 a = euˆ φ0 (0) . Using (4. is the equivalent of the generating function (??).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! (4.111) the orthonormality properties of the wave functions φn (y). Setting y = 0 in (4. Expanding r.h. (??).4.95) one obtains a a a a euˆ φ0 |evˆ φ0 = φ0 |euv evˆ φ0 = euv evˆ φ0 |φ0 = euv − + + + − (4. We start from the generating function (??) and replace according to (4.110) √ y2 u2 + 2uy − 2 2 a = euˆ φ0 (y) . 4. .s. For this purpose we consider a a a a euˆ φ0 |evˆ φ0 = φ0 |euˆ evˆ φ0 + + − + (4. i.h.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.116) = δµ ν .s.109) e = π4 n! n=0 √ Using (4. and on the r. in the ladder operator calculus. we can reproduce by means of the generating function (4.95).h.s. We will derive the equivalent of a generating function and use it to rederive the values Hn (0) and the orthonormality properties of φn (y)..112) Expanding both sides of this equation and using (4. Similarly. + (4.111) This expression.114) where we have employed (4.84) and deﬁning 2t = u one can write then π − 4 exp − or π − 4 exp − 1 1 √ y2 u2 + 2uy − 2 2 ∞ = n=0 un + n (ˆ ) φ0 (y) a n! + (4. of (4. the expected from which follows by comparision of each term on both sides φµ |φν orthonormality property.e.113) Comparision of all terms on the l.115) a where the latter step follows after expansion of evˆ and using a− φ0 (y) = 0. provide the same values for Hn (0) as provided in Eq. and ˆ l. (4.98) and again (4.µ=0 uµ v ν √ φµ |φν µ!ν! ∞ = µ=0 uµ v µ µ! (4.84) yields ∞ ν.115) and using (4.58) the Hermite polynomials Hn (y) by the eigenstates φn (y) √ ∞ 1 ( 2t)n y2 /2 2yt − t2 √ e φn (y) (4. We want to derive now the values Hn (0).84.

e. (4. For example.123) (4. for an oscillator in a stationary state φn (x) as given by (2.6 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 eigenvalues are identical to those determined in the position representation. the probability density is P (x) = |φn (x)|2 . 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. the time-independent Schr¨dinger equation for the oscillator can be written o ˆ H = − mω 2 2 2 x = i ˆ d .118) d2 1 2 + p dp2 2m ˜ ˜ φE (p) = E φE (p) .7) is given by (4. (−i)n ˜ φn (p) = √ 2n n! m˜ ω π 1 4 exp − m˜ p2 ω 2 Hn ( m˜ ω p) .121) ω (n + ˜ 1 ) 2 . The momentum representation eigenfunctions are. (4. i. ω = 1 / m2 ω .. 2.e.119) 2m 2 dp2 Accordingly. As to be expected. 1. ˆ and the Hamiltonian (4. i. (4.7) is 1 2 m 2 ω 2 d2 p − .235).122) For a solution of the Schr¨dinger equation in the momentum representation. For this purpose we employ the Schr¨dinger equation in the momentum o representation. The eigenvalues.. using (2. . The solutions can be stated readily exploiting the earlier results.6) and the Hamiltonian (4. .2. the momentum and position operators are as stated in (4. is given by the ˜ function ψ(p). n = 0. are ˜ En = or.88 Linear Harmonic Oscillator 4.121).122). 2.125) .124) ). of (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. 1..120) Multiplying this equation by 1/m2 ω 2 yields − where ˜ E = E / m2 ω 2 . . according to (4.235).e.3). i.2). is given by a function ψ(x). . ˜ (4. n = 0. In the position representation the wave function is a function of x. (4. the momentum and position operators are p = p. En = ω (n + 1 ) 2 d2 1 + m˜ 2 p2 ω 2 2m dp 2 2 ˜ ˜˜ φE (p) = E φE (p) (4. using (4. . 4. In the momentum representation the wave function is a function of p. (4.37). . dp (4.

126) a phase factor (−i)n .122) (−i)n ˜ φn (p) = √ 2n n! 1 πm ω 1 4 89 exp − p2 2m ω Hn ( 1 p) . From (4. φn (k) = We want to ﬁnally apply the relationship 1 ˜ φn (p) = √ 2π +∞ 2 (−i)n −k √ e 2 Hn (k) .131) in the present case employing dimensionless units. i.4. α ∈ R.133) where we note that a change of the normalization of φn (y) [c. hence.130) dx exp(−ipx/ ) φn (x) −∞ (4. m ω (4. 2n n! π (4.126) in terms of dimensionless units following the procedure adopted for the position representation where we employed the substitution (4.. the correctness of the phase factors (−i)n .125.e.127) exp − p2 m ω 2 Hn ( 1 p) . Proceed as follows. of course. Deﬁning k = m˜ / p or ω 1 k = p (4.58).. instead of (4.f.235)] absorbs the replacement dx → dy. the Jacobian.134).30) and (4. (4. this allowed us to introduce in (4.6.6: Momentum Representation or with (4.132) dy exp(−iky) φn (y) −∞ (4. According to the theory of the momentum representation holds ˜ ˜ Pn (p) = |φn (p)|2 or 1 ˜ Pn (p) = n 2 n! 1 πm ω 1 2 (4. deﬁned only up to an arbitrary phase factor eiα .134) Exercise 4. 1 ˜ φn (k) = √ 2π +∞ (4.30) to obtain (4. ˜ We can now state the probability density Pn (p) for an oscillator with energy En to assume momentum p.126) Normalized eigenfunctions are.e.129) follows ky = px/ and. through direct integration.128) We may also express the eigenstates (4. . This implies the property of Hermite polynomials (−i)n e− k2 2 +∞ Hn (k) = −∞ dy exp(−iky) e− y2 2 Hn (y) (4.126). i. 4.129) m ω one obtains. m ω (4.1: Demonstrate the validity of (4.30) and (2.

140) An excellent textbook is Photons and Atoms: Introduction to Quantum Electrodynamics by C. except that α changes periodically in time. y) −∞ (4. New York. The answer is ‘yes’.C.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. namely through the eigenvalue problem for normalized states a− φ(α) (y) = α φ(α) (y) . ˆ α ∈ R. +∞ Linear Harmonic Oscillator dy exp(−iky) g(z. The motion of the probability distribution of such state is presented in Fig. The question arises if similar states exist also for the quantum oscillator.138) yields using a− φn (y) = n φn−1 (y) ˆ ∞ n=0 √ n d(α) φn−1 (y) − α d(α) φn (y) n n = 0 (4. J.90 (a) Prove ﬁrst the property 1 f (k. or Glauber states of the harmonic oscillator and they play. Inserting this expansion into √ (4. (4. 1989) which discusses quasi-classical states of the free electromagnetic ﬁeld in Sect. 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 .139) where we have added the condition that the states be normalized.— III. z) = √ 2π where f (k. n n=0 d(α) n 2 = 1 (4. Inc. carries out a periodic motion described by y(t) = Re y0 eiωt (4.138) We will show that a harmonic oscillator prepared in such a state will remain forever in such a state.7 Quasi-Classical States of the Harmonic Oscillator A classical harmonic oscillator.Dupont-Roc. z) = exp −2ikz + z 2 − k 2 /2 .135) g(k. 2. We ﬁrst construct the solution of (4. We will ﬁnd that α can be characterized as a displacement form the minimum of the oscillator and that the state φ(α) (y) displays the same spatial probability distribution as the ground state. coherent states. z) = exp 2yz − z 2 − y 2 /2 (4. For this purpose we assume such state exists and expand ∞ ∞ φ(α) (y) = n=0 d(α) φn (y) . φ(α) | φ(α) = .Chen-Tannoudji. 4. pp. for example. We want to construct and characterize such states.196) of Hermite polynomials and equate coeﬃcients of equal powers of z to prove (4..134).194.138). and G. Such states are referred to as quasi-classical states.Grynberg (John Wiley & Sons.136) (b) Employ the generating function (2.1 showing in its top row the attributes of such state just discussed. The quasi-classical states can be obtained by generalizing the construction of the ground state of the oscillator. a pendulum swinging at small amplitudes.4. 192 3 . 4. for example.

[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) . we will see below that α ∈ C are also admissible and will provide a corresponding interpretation.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. However.109).139)] 1 = n=0 d(α) n 2 = |c|2 n=0 α2 2 = |c|2 eα .142) The solution is αn d(α) = √ c n n! ∞ ∞ (4.144) Normalization requires c = exp(−α2 /2) and.147) √ which identiﬁes this state as a ground state of the oscillator displaced by 2α. One can also write (4. hence.146) √ 1 − (y − 2 α)2 2 (4. This interpretation justiﬁes the choice of α ∈ R in (4. Comparing (4.e.138).149) − α∗ a− ˆ (4. n! (4.141) Because of the linear independence of the states φn (y) all coeﬃcients multiplying φn (y) must vanish and on obtains the recursion relationship (α) dn+1 = √ α d(α) .145) Using the generating function in the form (4.149) with (4. n n+1 (4.147) allows one to interprete T (α) as an operator √ which shifts the ground state by 2α. n! n (4. .4. i. n! (4..145) using (4.84) φ(α) (y) = exp − α2 2 ∞ n=0 αα∗ (αˆ+ ) a a+ φ0 (y) = e− 2 eαˆ φ0 (y) .143) for a constant c which is determined through the normalization condition [see (4.7: Quasi-Classical States or ∞ n=0 91 √ n + 1 dn+1 − α d(α) n (α) φn (y) = 0 . n! (4.148) The identity (4. the quasi-classical states are φ (α) (y) = exp α2 − 2 ∞ n=0 αn √ φn (y) .150) for α presently real. (4.

it holds T + (α) φ0 |T + (α) φ0 = f |T (α) T + (α) g = φ0 |φ0 .145) is replaced by a complex number αu which at times t1 = t0 + 2πn/ω. one can write in case of T + (α) T + (α) + = T (α) = eα ∗ a− ˆ − αˆ+ a a = e − (αˆ + − α∗ a− ) ˆ . i. . is the adjoint of the operator function on the l.150) for complex α obeys the unitary property T + (α) T (α) = 1 1 (4.157) (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.e. n = 0. .156) (4.s.s.92 Linear Harmonic Oscillator The shift operator as deﬁned in (4.151) which follows from a generalization of (4.h. t1 ) = u exp 1 2 α2 − 2 (α u)n √ φn (y) .95) to linear combinations αˆ+ + βˆ− and reads then a a f (αˆ+ + βˆ− ) φ|ψ = φ|f (α∗ a− + β ∗ a+ ) ψ . the shift operator leaves the ground state normalized.153) This is obviously the inverse of T + (α) as deﬁned in (4.h.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 (4.152) T + (α) f |T + (α) g = f |T (α) T + (α) g = f |g .145) shows that α in (4. In particular. Following the procedure adopted in determining the propagator of the harmonic oscillator [see (4. α ∈ C (4.160) .67.109) serves again to simplify this sum √ y2 1 1 α2 (u2 + 1) φ(α) (y. n! u = e−iω(t1 − t0 ) .155) (4.154) . 1. Relationship (4.150) and one can conclude that T + (α) is a so-called unitary operator with the property T + (α) T (α) = T (α) T + (α) = 1 1 or applying this to a scalar product like (4.152) Since the operator function on the r. 4. t1 ) = π − 4 u 2 exp − + 2yαu − 2 2 E (4.68)] one can write the time-dependent solution with (4.145) as the initial state φ(α) (y.158) Comparision of this expression with (4. (4.. a a ˆ ˆ (4. becomes real. (4.

(4.164) where according to (4. The diagram illustrates that the wave function retains its Gaussian shape with constant width for all times. a complex α corresponds to a displaced oscillator which has initially a non-vanishing velocity. The time-dependent wave function (4.4.163) also through the shift operator. In Fig. moving solely its center of mass in an oscillator fashion around the minimum of the harmonic potential. This shows clearly that the Glauber state is a close analogue to the classical oscillator. this characterization corresponds closely to that of the possible initial states of a classical oscillator. The oscillations of the mean position y0 (t1 ) and of the mean momentum k0 (t1 ) are out of phase by π just as in the case of the classical oscillator. 4. It is of interest to note that the displacement y0 of the Gaussian as well as the momentum k0 and phase φ0 associated with (4.7: 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. 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.159) is then 1 1 i φ(α) (y.147).150) with (4. except that it is not pointlike. Comparing (4. in general. a momentum factor and a phase factor.162) 2 with a periodic change.158) allows one to write 1 φ(α) (y. obviously. One can express (4.138): a real α corresponds to a displacement such that initially the oscillator is at rest.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 ) (4.165) . (4. t1 ) = u 2 T + (α u) φ0 (y) (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.148.138. being displaced around the center with period 2π/ω and.1 (top row) we present the probability distribution of the Glauber state for various instances in time. Our interpretation 2 explains also the meaning of a complex α in (4. t1 ) = π − 4 exp − [y − y0 (t1 )]2 + iy k0 (t1 ) − iφ0 (t1 ) − ω(t1 − t0 ) 2 2 .163) This solution corresponds to the initial state given by (4. experiences a change of its width (relative to . 4.161) and is found to describe a displaced Gaussian. 4.

t0 ) = n=0 dn φn (y) .169) . that of the Glauber states) with a period π/ω. σ|y. t0 ) = [πσ 2 ]− 4 exp we expand ∞ 1 − (y − ao )2 2σ 2 (4. σ|y. 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. 2σ 2 q = a . 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) (4.94 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. σ|y. (4. This behaviour is illustrated in Fig. 4.1: Time dependence of Gaussian wave packets in harmonic oscillator potential.1 (second and third row).167) One can derive for the expansion coeﬃcients +∞ dn = −∞ dy φn (y) φ(ao . To describe such state which satisﬁes the initial condition φ(ao . 1 + σ2 (4.166) φ(ao .168) dy e−p(y−q) Hn (y) where p = 1 + σ2 .

p−1 1 − σ4 t = 1 − σ 2 −iω(t1 −to ) e 1 + σ2 (4. this 3 volume integral table is likely the most complete today. σ|y.7: Quasi-Classical States The solution of integral (4.176) 1 √ pσ π a 2 exp − 1 1+σ2 + 1 yo − 1 ω(t1 − to ) 2 2 2 2 × (4. namely $750 4 . unfortunately very expensive. σ|y. and O.171) lim p−1 p n 2 Hn (q p ) = lim (2q)n = an . a ground state shifted by ao = 2α. This expansion can be written φ(ao . σ→1 p−1 2 (4.170) In order to check the resulting coeﬃcients we consider the limit σ → 1. i. New York.P.174) Hn (q p p−1 ) φn (y) .143).168) is4 e−aq/2 (p − 1)n/2 dn = √ Hn (q 2n n!σ p(n+1)/2 p ).I.e. Yu.4. t1 ) = e−aq/2 √ pσ ∞ n=0 e −iω(n+ 1 )(t1 −to ) √ 2 1 2n n! p−1 p n/2 (4. Marichev (Wiley. 1 (4. p−1 95 (4..157) as follows φ(ao .173) which is. therefore. 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. t1 ) = √ × where yo = q p a = √ . In this limit the coeﬃcients dn should√ become identical to the coeﬃcients (4. 1990). a worthy successor of the famous Gradshteyn.175) ∞ tn n=0 2n n! Hn (yo ) e−yo /2 Hn (y) e−y 2 2 /2 The generating function (4. 2 by A.166) in analogy to (4. σ|y. vol. Prudnikov. σ→1 O(y n−2 ) for n ≥ 2 1 for n = 0.172) One can then conclude e−a /4 n lim dn = √ a σ→1 2n n! (4. 1 a2 2 1 + σ2 + 1 2 2 y o + 2 y yo This integral can be found in Integrals and Series. Brychkov.76) permits us to write this φ(ao . t1 ) = √ 2 × exp − 1 (y 2 + yo ) 1+t2 − 2 1−t 2 1 1 √ √ pσ π 1 − t2 × t 1 − t2 (4. in fact.143) of the Glauber state. Following the strategy adopted previously one can express the time-dependent state corresponding to the initial condition (4. in agreement with the expected result (4.177) − 1 ω(t1 − to ) 2 .

96 Linear Harmonic Oscillator .

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

x2 .6) follows immediatety for the determinant of R(ϑ) using detA B = detA detB and detAT = detA detR(ϑ) = ±1 .7) implies that a rotation is associated with an inversion.7) Let us consider brieﬂy an example to illustrate rotational transformations and to interpret the sign of detR(ϑ).3) Since this holds for any a and b.e. ϕ)T ) = ± sinϕ cosϕ 0 (5. .k. Let us deﬁne the rotated vector as r = R(ϑ) r . (5. it follows 3 [R(ϑ)]jk [R(ϑ)]j = δk j=1 . where ϑ denotes the rotation. they conserve a · b = 3 aj bj . can be represented by 3 × 3 matrices R(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. the matrix represents a proper rotation.6) (5. 2.4) 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 . (5. i. =1 [R(ϑ)]jk [R(ϑ)]j ak b = j=1 aj bj . ϕ)T ) is negative. x3 )T . In the latter case the determinant of R(ϑ = (0. In the following we want to exclude inversions and consider only rotation matrices with detR = 1. (5. A rotation around the x3 -axis by an angle ϕ is described by the matrix cosϕ − sinϕ 0 R(ϑ = (0. if the thumb in your right hand ﬁst points in the direction of ϑ. 0..2) Rotations conserve the scalar product between any pair of vectors a and b. i. a minus sign in Eq. From (5. ϑ parametrizes the rotations uniquely as long as |ϑ| < 2π. then the ﬁngers of your ﬁst point in the direction of the rotation.e. namely around an axis given by the direction of ϑ and by an angle |ϑ|. in Cartesian coordinates r = (x1 . (5.5) This equation states the key characteristic of rotation matrices. i. It holds then j=1 3 3 a ·b = j. (5. are linear and. 3 . in case of a prefactor −1 a rotation and an inversion at the origin.e. (5.(5. therefore. 0. We assume the convention that rotations are right-handed.98 Properties of Rotations in R Theory of Angular Momentum and Spin Rotational transformations of vectors r ∈ R ..

hence. (ii) The group element ‘e’ is played by the 3 × 3 identity matrix 1 1. one can show R3 R3 = 1 Furthermore.5. a ∈ G → ∃a−1 ∈ G with a ◦ a−1 = a−1 ◦ a = e. (5. We need to demonstrate that this inverse belongs also to −1 T SO(3). In the following we will assume R1 . (5. For this purpose we must demonstrate that the group properties (i–iv) hold.12) (iv) Since the associative law holds for multiplication of any square matrices this property holds for the elements of SO(3). 3 × 3–matrix. R3 = R1 R2 is a real. We have shown altogether that the elements of SO(3) form a group. R2 ∈ SO(3) and do not write out “◦” explicitly since it represents the matrix product. It holds T T T T R3 R3 = ( R1 R2 )T R1 R2 = R2 R1 R1 R2 = R2 R2 = 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.11) (iii) From detR1 = 0 follows that R1 is non-singular and. in fact. 99 (5.9) This set of matrices is closed under matrix multiplication and. (iii) ∀a. (ii) There exists an element e in G with the property ∀a ∈ G → e ◦ a = a ◦ e = a. (i) Obviously. there exists a real 3 × 3–matrix −1 R1 which is the inverse of R1 . T R1 R1 −1 = 1 −1 = 1 1 1 (5. It follows that R3 is also an element of SO(3). forms a group G satisfying the axioms (i) For any pair a. 1 T Similarly. We want to prove now that SO(3) as deﬁned in (5.13) . We would like to point out the obvious property that for all elements R of SO(3) holds R−1 = RT (5. R RT = RT R = 1 detR = 1 } 1.9) forms a group. it holds 1. (iv) For products of three elements holds the associative law a ◦ (b ◦ c) = (a ◦ b) ◦ c. b ∈ G the product c = a ◦ b is also in G.10) detR3 = det ( R1 R2 ) = detR1 detR2 = 1 . Since (R1 )T = (R1 )−1 it follows −1 −1 −1 T (R1 )T R1 = (R1 )−1 R1 = −1 which implies R1 ∈ SO(3).

establish if 1-1 homomorphic mappings exist.13) R(ϑk ek )−1 = R(ϑk ek )T = [exp ( ϑk Lk )]T = exp ϑk LT ˆ ˆ k and. we want to assume that they exist. (c) the set of four rotations in the x1 . Generators. Lie Group.19) (5. mappings between the sets which conserve the group properties: (a) the set of real and imaginary numbers {1.18) (5. −1. rotation by 180o . J2 .16) must be antisymmetric.1: Test if the following sets together with the stated binary operation ‘◦’ form groups. 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.e. identify subgroups. (b) the set of matrices {M1 . −i} together with multiplication as the binary operation ‘◦’. rotation by 270o } and consecutive execution of rotation as the binary operation ‘◦’. 0 1 −1 0 .15) Below we will construct appropriate matrices Jk . due to the uniqueness of the inverse. i. rotation by 90o .14) together with matrix multiplication as the binary operation ‘◦’. −1 0 0 −1 .17) . will play a central role in the representation of many other symmetry transformations in Physics. We want to show that for the property R RT = 1 to hold. the matrices i Lk = − Jk (5. k = exp ( −ϑk Lk ) (5. J3 as follows i R(ϑ) = exp − ϑ · J (5.100 Theory of Angular Momentum and Spin Exercise 5. Lie Algebra We want to consider now a most convenient.20) (5. M3 . M2 . compact representation of R(ϑ) which will be of general use and. 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. This representation expresses rotation matrices in terms of three 3 × 3– matrices J1 .1. 0 −1 1 0 (5. x2 –plane {rotation by Oo . in fact. M4 } = 1 0 0 1 . presently. we arrive at the property exp ϑk LT k from which follows LT = −Lk . i.

Using the deﬁnition of partial derivatives we can state L1 = lim ϑ−1 R(ϑ = (ϑ1 .27) .2: Determine the generator L1 according to Eq. (5.8) we can state cosϑ3 −sinϑ3 0 1 −ϑ3 0 1 0 . L2 .21) −b −c 0 It is this the reason why one can expect that three-dimensional rotation matrices can be expressed through (5.5.23) (5. Using (5. lim R[(0. 0)T the rotation is the identity. 0. the real matrix A is anti-symmetric. 0)T ) − 1 1 1 ϑ1 →0 (5.(5.25) 0 −ϑ3 0 0 −1 0 ϑ3 0 0 = 1 0 0 . ϑ2 .. Let us use this deﬁnition to evaluate L3 .1. (5.26) (5.22) for k = 3 yields L3 = ϑ−1 3 One obtains in this way 0 0 0 i − J1 = L1 = 0 0 −1 0 1 0 0 0 1 i − J2 = L2 = 0 0 0 −1 0 0 0 −1 0 i 1 0 0 − J3 = L3 = 0 0 0 The matrices Lk are called the generators of the group SO(3).22). ϑ2 . 0 0 0 0 0 0 (5.15) as the derivatives of R(ϑ) with respect to the Cartesian components ϑk taken at ϑ = (0. 0)T . 0. One can determine then the matrices Lk deﬁned in (5.e. Exercise 5. 0. This property gives A then only three independent matrix elements. We assume now that we parameterize rotations through a three-dimensional rotation vector ϑ = (ϑ1 .15) with three real parameters ϑ1 .22) and similar for L2 and L3 . 0.24) (5. ϑ3 )T ] = lim sinϑ3 cosϑ3 0 = ϑ3 ϑ3 →0 ϑ3 →0 0 0 1 0 0 1 Insertion into (5. i. L3 which are independent of the rotation angles. A must have the form 0 a b A = −a 0 c . ϑ3 )T such that for ϑ = (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 .

30) Groups with the property that their elements can be expressed like (5.102 Theory of Angular Momentum and Spin Sofar it is by no means obvious that the generators Lk allow one to represent the rotation matrices for any ﬁnite rotation. B]. The expression for Zn are actually rather complicated.28) The construction of the generators through the limit taken in Eq. 0)T ) = exp( 2 L2 ) and R(ϑ = (0.g. that the generators represent inﬁnitesimal rotations with 1 >> |ϑ| around the x1 -. = 2.. Z2 = 2 [A. e. 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. as we see later.33) We want to demonstrate now that exp(ϑ · L) yields the known rotational transformations. the property (5. = 2. it must hold 3 [Lk .29) where Z1 = A + B and where the remaining operators Zn are commutators of A and B. Of course. holds the algebra [Jk . 0.e.g. B] ). however. if the resulting products of exponential operators can be expressed in the form (5. commutators of commutators of A and B. = 2. 0. that the operators (5. (5. only the latter example and . [A. x2 -.15) and their generators obey the property (5. B]. L ] = k m Lm (5. In case of the group SO(3) the structure constants are particularly simple.. the matrix (5.8) in case of ϑ = (0. the elements of if any two indices are identical for k = 1. B]] + [[A.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.. From this result one can conclude that expressions of the type (5. m = 3 .30) are called Lie groups.32) For the matrices Jk = i Lk which.e. however. 0)T ) = exp( 1 L1 ). etc.22) implies. m = 3 for any odd permutation of k = 1. 3 )T ) = exp( 3 L3 ). We want to consider.30) is called the Lie algebra. An answer can be found considering the Baker-Campbell-Hausdorf expansion ∞ exp( λA ) exp( λB ) = exp( n=1 λn Zn ) (5. (5. J ] = i k m Jm . m = 3 for any even permutation of k = 1. 0. and the constants fk m are called the structure constants. the Lie algebra of SO(3) can be written [Lk . ϕ)T . 1 1 e. Lie groups can have any number of generators. L ] = m=1 fk m Lm . The question is then.15) obey the group property. in fact. are related to angular momentum operators. three being a special case. In fact.15). i. (5. Finite rotations can be obtained by applications of many inﬁnitesimal rotations of the type R(ϑ = ( 1 . R(ϑ = (0. i. (5. Z3 = 12 ( [A.31) where k m are the elements which are 0 1 k m = 1 −1 of the totally antisymmetric 3-dimensional tensor. 2 . and x3 -axes.

1.31).1: Matrix Representation of SO(3) determine exp(ϕL3 ). that the matrix ϕL3 .37) = (−1)n 1 0 0 1 . (5. We note.8). using (5. accordingly. can be written ϕL3 = 0 103 A 0 0 0 0 . (2n + 1)! (5.34) One obtains then for the rotational transformation exp A 0 0 0 0 0 1 0 0 1 0 0 0 1 0 0 1 0 0 0 1 0 0 1 ∞ ν=1 ∞ ν=1 0 0 1 = = = + + 1 ν! 1 ν! A 0 0 0 0 0 ν Aν 0 0 0 0 0 0 0 1 ∞ 1 ν ν=0 ν! A 0 0 = eA 0 0 . 0 −1 1 0 = (−1)n 0 −1 1 0 (5.3: Show that the generators Lk of SO(3) obey the commutation relationship (5. (5. according to (5.34. Exercise 5.35).5. eϕL3 which agrees with (5.27). (5.39) Recognizing the Taylor expansions of cos ϕ and sin ϕ one obtains eA = and. ∞ 2n = − 1 0 0 1 2n+1 (5.40) .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 . 5. cos ϕ − sin ϕ 0 = sin ϕ cos ϕ 0 0 0 1 (5.41) cos ϕ − sin ϕ sin ϕ cos ϕ (5. A = ϕ 0 −1 1 0 .36) The property 0 −1 1 0 allows one to write 0 −1 1 0 and.38) eA = n=0 (−1)n 2n ϕ (2n)! 1 0 0 1 ∞ + n=0 (−1)n ϕ2n+1 (2n + 1)! 0 −1 1 0 .

44) holds.47) = ρ R(ϑ1 ) R(ϑ2 ) . we deﬁne rotations R(ϑ) as linear maps C∞ ( ) → C∞ ( ). Since this holds for any ψ(r ) one can conclude ρ R(ϑ1 ) ρ R(ϑ2 ) group property (5. i.48) ..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).46) (5. B ∈ SO(3) holds ρ(A B) = ρ(A) ρ(B) .45) (5.e. complex-valued functions over the 3-dimensional space R which can be diﬀerentiated inﬁnitely often.1: Assume the deﬁnition ρ ρ (A B) = ρ (B) ρ (A).2.42) Obviously. the R(ϑ) ψ(r) = ψ(R(ϑ) r ). i. hence. In analogy to R(ϑ) being a linear map R → R . Show that in this case holds Generators One can assume then that R(ϑ) should also form a Lie group.e. (5.e. (5. namely R(ϑ)ψ(r ) = ψ(R−1 (ϑ) r ) . i. for A.104 Theory of Angular Momentum and Spin 5. Deﬁnition We ﬁrst deﬁne how a rotational transformation acts on a wavefunction ψ(r ). the transformation R(ϑ): C∞ ( ) → C∞ ( ) is related to the transformation R(ϑ): R → R .43) ρ( ) conserves the group property of SO(3). we will learn that transformation patterns and invariances are connected with the angular momentum states of quantum mechanics. its elements can be represented through an exponential R(ϑ) = exp i − ϑ·J (5.44) This important property which makes ρ( SO(3) ) also into a group can be proven by considering ρ R(ϑ1 ) R(ϑ2 ) ψ(r) = ψ([R(ϑ2 )]−1 [R(ϑ1 )]−1 r ) = ρ R(ϑ1 ) ρ R(ϑ2 ) ψ(r) . Exercise 5. = ψ([ R(ϑ1 ) R(ϑ2 ) ]−1 r) = ρ R(ϑ2 ) ψ([R(ϑ1 )]−1 r) (5. one isomorphic to SO(3) and. In particular. a relationship which we like to express as follows R(ϑ) = ρ R(ϑ) . (5. For this purpose we require stringent continuity properties of the wavefunction: the wavefunctions under consideration must be elements of the set C∞ ( ). in fact.

In fact. k = 1.50) lim ϑ−1 ( f ( R(0. 0 0 1 x3 x3 (5.57) Exercise 5.2: Function Space Representation of SO(3) where the generators can be determined in analogy to Eq. ϑ3 )T ) f (r) − f (r) 3 (5. show that (5. i. − 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) . (5.(5. 0. 3 1 ϑ3 0 x1 x1 + ϑ3 x2 R(0. J3 ] and [J3 .5. i − J3 f (r) = = Using (5. (5.49) (note the minus sign of −ϑ1 which originates from the inverse of the rotation in Eq.23.55) Not only is the group property of SO(3) conserved by ρ( ). −ϑ3 ) r ≈ −ϑ3 1 0 x2 = −ϑ3 x1 + x2 .49).56) We like to verify this property for [J1 .2: (a) Derive the generators Jk . J2 ] f (r) = − = − 2[ 2 [− i J .51) Inserting this into (5. (5.49) to a function f (r). 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 .22) according to i − J1 = lim ϑ−1 R(ϑ = (−ϑ1 . . 5.42)) and similarly for J2 and J3 . in case of J3 .56) holds for [J2 . (5.53–5. J2 ].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. 0.55) and f (r) = f [J1 . it holds [Jk . (5. 0)T ) − 1 1 ϑ1 →0 105 (5.54) (5. −ϑ3 ) r ) − f (r) ) . J1 ]. but also the Lie algebra of the generators. The generators Jk can be determined by applying (5. One obtains using (5.24) one can expand ϑ3 →0 ϑ3 →0 lim ϑ−1 R(ϑ = (0. 2 by means of limits as suggested in Eq. 0.2.52) ( x2 ∂1 − x1 ∂2 ) f (r) . 0. J ] = i k m Jm ..53) (5.e.

62) J1 [J1 . J3 ] + [J1 .58) i = r ×p ˆ (5. can be found. θ.55) can be written as a vector product of r and i − J = −r × From this follows. Using [AB. according to the correspondence principle between classical and quantum mechanics. [J 2 . and x3 = r cosθ. k = 1. 5.61) We demonstrate this property for k = 3. Equation (5. We will show now that the properties of spherical harmonics J2Y J3 Y m (θ.61) simultaneous eigenstates of J 2 and of one of the Jk . The commutation relationships imply that no simultaneous eigenstates of all three Jk exist. To show this we ﬁrst note that (5. i.. Show that for J3 as deﬁned above holds exp( iα J3 ) ψ(φ) = ψ(φ + α). 2. However. usually chosen as J3 . φ). Jk ] = 0 .3 Angular Momentum Operators The generators Jk can be readily recognized as the angular momentum operators of quantum mechanics.60) commutes with all three Jk . k = 1. x2 = r sinθ sinφ. (5. the operator 2 2 2 J 2 = J1 + J2 + J3 (5. the system related to the Cartesian coordinates through x1 = r sinθ cosφ. φ) as well as the eﬀect of the so-called raising and lowering operators J± = J1 ± i J2 (5.63) (5.65) .e. 3.59) which. φ) m (θ.106 Theory of Angular Momentum and Spin Exercise 5. These eigenstates are the well-known spherical harmonics Y m (θ. (5. φ) = = 2 ( + 1) Y m (θ.56) one obtains [J 2 . C] = A[B. φ).64) mY m (θ.56) are the famous commutation relationships of the quantium mechanical angular momentum operators. C] + [A. 2. φ) (5.2. J3 ] J1 + J2 [J2 . using the momentum operator p = ˆ J = r × i . J3 ] J2 −i J1 J2 − i J2 J1 + i J2 J1 + i J1 J2 = 0 According to (5. identiﬁes J as the quantum mechanical angular momentum operator. 3 .53–5. J3 ] (5.e. J3 ] = = = 2 2 [J1 + J2 . i. J3 ] + [J2 . C]B and (5.3: Consider a wave function ψ(φ) which depends only on the azimutal angle φ of the spherical coordinate system (r. .

1 Let Lk . Using [L3 .68) (5.72) = 0 = −i | (5. φ) are a consequence of the Lie algebra (5. φ) m+1 (θ. φ) (5. in fact. L− ] + L− L+ )| .74) (5. k = 1. L− ] + L− L+ )| m = ( −2iL3 + . The deﬁnitions of the coeﬃcients α −1 and β −1 yield L+ L− | = L+ ( iβ −1 | − 1 = − α −1 β −1 | .70) (5. namely. 3 be operators acting on a Hilbert space H which obey the algebra [Lk .e.75) (5.66) (5. L− ] = [L1 + iL2 . An Important Theorem Theorem 1. We will show that this property holds then also for m − 1.. L− ] = [L3 . From this follows that L− | is an eigenstate of L3 .71) To prove this theorem we ﬁrst show by induction that (i) holds.5. L1 − iL2 ] = L2 + iL1 = i (L1 − iL2 ) = iL− and L3 | = −i | we obtain L3 L− | = ( iL− + L− L3 )| = i(− + 1) L− | . Continuing our proof through induction we assume now that property (i) holds for m. For this purpose we prove the following theorem. L2 = L2 + L2 + L2 . L− ] + L− L3 ) | . We consider L+ L− | m = ( [L+ .56). For this purpose we ﬁrst demonstrate that the property holds for m = − 1 by considering (i) for the state L− | ∼ | − 1 . Noting [L+ . J+ Y J− Y J− Y m (θ.73) (5. We note L3 L− | = ( [L3 . and since we have already shown | − 1 = (iβ −1 )−1 L− | we can state L3 | − 1 = −i( − 1)| − 1 . L+ raises the m-value of | − 1 from m = − 1 to m = . φ) (θ. i.67) (5.76) We now show that property (ii) also holds for m = − 1 proceeding in a similar way. α −1 β −1 = 2 . Let there exist states | in 1 2 3 H with the property L+ | L3 | then the states deﬁned through L− | m + 1 = −i βm | m have the following properties (i) L+ | m (ii) (iii) L3 | m L2 | m = −i αm | m + 1 = −i m | m = − ( + 1) | m (5. The arguments are very similar to the ones used above and we can be brief. φ) − m (θ. i.3: Angular Momentum Operators on the spherical harmonics. It holds L+ L− | = (L+ L− − L− L+ + L− L+ )| = ([L+ . (∗) (5.69) J+ Y (θ.e. 2. φ) 107 = = 0 = = 0 ( + m + 1)( − m) Y ( + m + 1)( − m) Y m+1 (θ. L ] = k m Lm and let L± = L1 ± i L2 . L1 − iL2 ] = −2iL3 and L+ | = 0 one obtains L+ L− | = −2iL3 | = −2 | .

64).e. 5. We note L3 L− | m = ( [L3 .82) (5.81) i. 5..69). A ﬁrst application will involve the construction of the eigenstates of the angular momentum operators as deﬁned in (5. For this purpose we need to express the angular momentum operators in terms of spherical coordinates (r. from L+ L− | m = − αm−1 βm−1 | m follows the recursion relationship αm−1 βm−1 = 2m + αm βm . 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 . 2 We note that Eqs. Using the familiar formula + m − n k=0 k = n (n + 1)/2 − (m + 1) m = ( + 1) m − (m + 1) m (5.83) (5. ﬁnally αm = βm = (5. it holds αm βm = 2 one obtains αm βm = = ( + 1) k=m+1 k .66. (5. (∗∗) (5. in terms of spherical coordinates deﬁned through x1 x2 x3 = = = r sin θ cos φ r sin θ sin φ r cos θ (5. yield properties (5. We can deduce from (∗) and (∗∗) α α α −1 β −1 −2 β −2 −3 β −3 = 2 = 2( − 1 + ) = 2( − 2 + · · −1 + ) Obviously.84) .72. in particular. We note that we can write L2 = 1 1 1 1 2 2 2 2 L+ L− + 2 L− L+ + L3 . i. This completes the proof of the theorem. Angular Momentum Operators in Spherical Coordinates We want to express now the generators Jk . it also holds for m − 1. (5. Before we can ﬁnally show that property (iii) holds as well for all − ≤ m ≤ we need to determine the coeﬃcients αm and βm . 5.. Again the argumemts are similar to the ones used above.79) We can now show that (iii) holds for all proper m–values. θ.73. 5. 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 . given in (5.77) We will show now that if (ii) holds for m.79) read L+ | m L− | m + 1 = −i = −i ( + m + 1)( − m) | m + 1 ( + m + 1)( − m) | m .108 Theory of Angular Momentum and Spin L− L+ )| m = (−2m − αm βm )| m . 5. One can normalize the states | m such that αm = βm .55). We will have various opportunities to employ the Theorem just derived.70.e.80) (5. φ.63.53–5.78) ( + m + 1)( − m) ( + m + 1)( − m) .

(L2 )2 = − cos φ ∂ ∂ + cot θ sin φ ∂θ ∂φ 2 . cos θ ∂ cos2 φ + sin θ cos φ ∂φ cos2 θ sin θ sin φ ∂ sin θ sin φ ∂ + + sin2 θ f (θ.85). In fact. using ∂/∂tanφ = cos2 φ ∂/∂φ and ∂/∂cosθ = −(1/sinθ) ∂/∂θ. φ) ∂x ∂x3 ∂ tan φ =0 (5. φ) ∂x2 /x1 ∂ ∂x3 /r ∂ = x3 + x3 ∂x2 ∂ tan φ ∂x2 ∂ cos θ ∂x2 /x1 /r ∂ ∂ − x2 − x2 ∂x3 3 ∂ cos θ f (θ.84 ) one obtains.82–5. Using (5. (L1 )2 = = ∂ 2 ∂ + cot θ cos φ ∂θ ∂φ 2 ∂ ∂ 1 ∂2 sinφ cosφ sin2 φ 2 − + 2 cotθ sinφ cosφ ∂θ ∂φ ∂θ∂φ sin2 θ ∂2 ∂ + cot2 θ cos2 φ 2 + cotθ cos2 φ ∂θ ∂φ sin φ i ∂φ .90) This yields.85). L1 f (θ. φ) (5. we will prove − − − i i i J1 J2 J3 = L1 = L2 = L3 ∂ ∂ + cot θ cos φ ∂θ ∂φ ∂ ∂ = − cos φ + cot θ sin φ ∂θ ∂φ ∂ = − .3: Angular Momentum Operators 109 We ﬁrst like to demonstrate that the generators Jk .5. using (5.86) (5. for example. (5.92) (5. involve only the angular variables θ and φ and not the radius r.88) (5. ∂φ = sin φ (5.60). φ) = ( x3 ∂2 − x2 ∂3 ) f (θ.93) and similarly.87) To derive these properties we consider.91) which agrees with expression (5. the latter deﬁned in (5. actually.85) (5. applying repeatedly the chain rule. using (5.86).87) holds J3 = To determine J 2 we note.89) = ∂ x2 x2 ∂ x2 + x2 ∂ x3 − 33 − x2 1 3 2 x1 ∂ tan φ r ∂ cos θ r ∂ cos θ f (θ. φ) = (5. in terms of spherical coordinates. According to (5. L1 f (θ. φ). We like to express now the operators J3 and J 2 . φ) = (x3 ∂2 − x2 ∂3 )f (θ. φ) sin θ ∂θ sin θ ∂θ L1 f (θ.

2m r ∂r2 2m r2 2 (5. 2 ∂θ ∂θ (5.100) (5.85.96) and the relationship between Lk and Jk as given in (5. The ˙ corresponding expression for the quantum mechanical kinetic energy operator is ˆ T = − which follows from the identity 2 2 2 2m = − 1 ∂2 J2 r + .63.87).95) sinθ ∂ ∂θ = ∂2 ∂ + cotθ . 5. sphere S2 . 1 sin2 θ sinθ ∂ ∂θ (5. namely on the space of complex-valued functions on the unit 3-dim. C∞ (S ). ∂θ2 ∂θ ∂φ (5. 5.97).98) where Jclass = r × mv is the angular momentum and pr = mr is the radial momentum. 5. 5.86. The functions in C∞ (S ) have real variables . One ﬁrst identiﬁes the Hilbert space H on which the generators Lk operate.e. These generators act on a subspace of C∞ ( ).94) It follows (L1 )2 + (L2 )2 + (L3 )2 = With cot2 θ + 1 = 1/sin2 θ.101) 1 ∂2 r r ∂r2 and comparision with (5.64) can be constructed as follows.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. (5. ∂θ ∂φ ∂2 ∂ ∂2 + cotθ + ( cot2 θ + 1 ) 2 .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. 5.97) Kinetic Energy Operator The kinetic energy of a classical particle can be expressed J2 p2 p2 = r + class 2m 2m 2m r2 (5. one can conclude J2 = − 2 sin θ 2 sinθ ∂ ∂θ 2 + ∂2 ∂φ2 . 5.87).85.4 Angular Momentum Eigenstates According to the theorem on page 107 above the eigenfunctions (5.86. The operators in the present case are Lk = − i Jk where the Jk are diﬀerential operators in θ and φ given in (5.

(5.108) m (θ.107) (5. .110) = 0 (5.102) exists. φ)f (θ. φ) L3 Y (θ. φ) = 2 + 1 ( − m)! P 4π ( + m)! m (cosθ) eimφ . can be expressed L± = Accordingly.102) where dΩ = sinθ dθ dφ is the volume element of S2 . (5. .111) . φ) = = 0 −i Y (θ. φ) = −i ( + m + 1)( − m)Y m (θ. . m) Constructing Y → Y →→ ··· Y ( + m)! (2 )! ( − m)! −1 − We like to characterize the resulting eigenfunctions Y m (θ. φ) = ∆( . (5. φ) in H which satisfy L+ Y (θ. φ) (5. φ) (5.107) reads ∂ ∂ + i cotθ ∂θ ∂φ P (cosθ) ei φ i e±iφ ∂ ∂ ± i cotθ ∂θ ∂φ . is indeed a Hilbert space. m) = 1 −m J− 1 2 Y (θ.109) using (5. normalizing these functions.85–5. φ).103) iteratively for m = − 1.105) explicitly. . (5. φ.5.105) ∆( . . . The eigenfunctions (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. φ) (5. . The norm in H is deﬁned through |f |2 = S2 dΩ f ∗ (θ.106) The latter property is obviously satisﬁed for the chosen φ-dependence.63. 5. 0 ≤ φ < 2π. by carrying out the construction according to (5. φ). φ). − . and to be admissable for description of quantum states. φ) = 0 as well as L3 Y (θ.64) can be constructed then by seeking ﬁrst functions Y (θ. − 2. − + 1. The function space endowed with this norm and including only functions for which the integral (5. . the so-called spherical harmonics.104. The raising and lowering operators L± = L1 ± i L2 . must be cyclic in φ with period 2π. m = − . φ) = −i Y (θ. φ) . } applying L− Y m+1 (θ.4: Angular Momentum Eigenstates 111 θ. 5. 0 ≤ θ < π. One obtains in this way Y m (θ. and then determining the family {Y m (θ.87).104) (5. (5.

1975) . (5.106. i.115) holds.116) The integral appearing in the last expression can be evaluated by repeated integration by parts.117) Accordingly.106. one for which π dθ sinθ |P (cosθ)|2 (5.119) provide the following expression for Y Y (θ. 2nd Ed. (5. 5. using the new integration variable x = cos θ. π +1 dθ sin2 0 +1 θ = −1 +1 −1 dx (1 − x2 ) +1 = = . ”Classical Electrodynamics.119) where the factor (−1) has been included to agree with convention1 . A proper solution of this equation. New York. for example.114) yield |N |2 2 +1 1 2π 4π (2 )! π dθ sin2 0 +1 θ = 1. φ) = (−1) 1 1 = 1 [1 · 3 · 5 · · · (2 − 1)]2 (5. (5. 5.112 or Theory of Angular Momentum and Spin ∂ − ∂θ cotθ P (cosθ) = 0 (5. We note that (5.114) dθ sinθ |Y (cosθ)|2 = 1 0 (5.107).114.118) 2 +1 1 1 sin θ ei 4π (2 )! 2 ! φ . is P (cos θ) = N sin θ as one can readily verify.116) implies |N |2 from which we conclude N = (−1) 1 · 3 · 5 · · · (2 − 1) (5. . . The normalization factor N is chosen such that π (5. (5.112) where we employed the deﬁnition of P (cosθ) in (5.” by J.. = = 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)] (5.113) 0 is ﬁnite. 5. Jackson (John Wiley. One obtains.120) See.D.e.

e.110). employing again the variable x = cos θ.63. .127) ∂ −m−1 (x2 − 1) .121) This expressions shows that the factor eimφ .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) . describes the φ-dependence of Y m (θ. through repeated application of (5. every application of L− reduces the power of eiφ by one. For m = (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 × (5.4: Angular Momentum Eigenstates 113 We have constructed a normalized solution of (5. 5.106). .124). The reader should note that the associated Legendre polynomials. ∂x −m−1 .125) which agrees with (5.123). 5.122) The latter identity can be written. indeed.. 5.103. We want to demonstrate now that the recursion equation (5. therefore. we seek to determine the functions P m (cos θ) and. according to (5. use (5.126) Then holds.124) we proceed by induction. φ). − deﬁned in (5.5. i. (5. and can obtain now.124) reads P (x) = = (2 )! (1 − x2 )− 2 (x2 − 1) 2 ! (−1) 1 · 3 · 5 · · · (2 − 1) sin θ (5.64).124) holds for m + 1. Actually. φ). m = − 1.114.124) called the associated Legendre polynomials. are real. P m (x) = −1 × ( + m + 1) ( − m) × −(1 − x2 ) 2 1 (5. m (cos θ) e (5. .106) i e−iφ ∂ ∂ − i cotθ ∂θ ∂φ 2 + 1 ( − m − 1)! P 4π ( + m + 1)! 2 + 1 ( − m)! P 4π ( + m)! i(m+1)φ m+1 (cos θ) e = −i ( + m + 1)( − m) imφ .123) leads to the expression P m (x) = −m m ∂ 1 ( + m)! (1 − x2 )− 2 (x2 − 1) 2 ! ( − m)! ∂x −m (5.119). the eigenfunctions Y m (θ. − 2. (5. as speciﬁed in (5. 5. To prove (5. 2 ! ( − m − 1)! ∂x −m−1 (5.103. P m+1 (x) = −m−1 m+1 ∂ 1 ( + m + 1)! (1 − x2 )− 2 (x2 − 1) . namely Y (θ. Let us then assume that (5.123) P m+1 (x) ∂ x + (m + 1) 1 ∂x (1 − x2 ) 2 .110) together with (5. φ).

106. φ) = = determines ﬁrst a normalized solution 0 i f (θ.123) terminates for m = − . and constructs then the eigenfunctions Y − +1 .130) (5.124) holds for m = . P − (x) = 1 ∂2 (1 − x2 ) 2 (x2 − 1) . Y − +2 .135) . therefore.132). In that case (1 − x2 ) is a polynomial of degree 2 and. therefore. the derivative ∂ 2 +1 /∂x2 +1 of this expression vanishes. φ) = Y − (θ. therefore. 5. yields a vanishing expression for m = − − 1 as long as is a non-negative integer. namely.133) Since the term with the highest power of (x2 − 1) is x2 . 5. φ) (5. φ) = (−1)m Y m −m (θ. φ) L3 f (θ. However.128) one can show that (5.132) which reduces the number of spherical harmonics which need to be evaluated independently roughly by half .. if L− Y (θ. we embark on this construction in order to prove (5. 2 ! (2 )! ∂x2 1 (5. φ) choosing the proper sign. P − (5. (5.124) and. from such construction emerges an important symmetry property of Y m (θ. Such construction reproduces the eigenfunctions Y m (θ. Constructing Y − → Y − +1 → ···Y m (θ. using x = cos θ. We ﬁrst determine Y − (θ. It holds. i.134) (x) = 1 2 ! (1 − x2 ) 2 . by repeated application of the operator L+ .129) holds. We want to test if the recursion (5. φ) An alternative route to construct the eigenfunctions Y of L− f (θ. Since (5.131) identiﬁes f (θ.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. etc.127) reproduces (5. hence. φ) using (5.124).124). appears not very interesting.124) and. φ) = 0 (5.124) holds for m if it holds for m + 1.e. φ) as given in (5. which is equivalent to recursive application of L− . expression (5. φ). it holds then for all m. (5. φ) . therefore.106. Y ∗ (θ. that expression (5. it holds ∂2 (x2 − 1) = (2 )! ∂x2 and. In fact.

120) for Y obey the postulated relationship (5.. φ) = 2 +1 1 1 sin θ e−i 4π (2 )! 2 ! φ 115 − .e.124). φ) . following closely the proof of eq.124) one arrives at Y (θ. According to (5.137) m (θ. (5.142) yields P − (x) = (−1) 1 (1 − x2 )− 2 (x2 − 1) = (1 − x2 ) 2 2 ! 2 ! (5.110) applies the = ( + m + 1)( − m) Y m+1 (θ. One can readily verify that this expression provides a solution of (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. 5.123)] P m+1 (x) = −(1 − x2 ) 2 1 ∂ x − m 1 ∂x (1 − x2 ) 2 P m (x) . (5.4: Angular Momentum Eigenstates Due to (5.140) Introducing again the variable x = cos θ leads to the recursion equation [c.136) to construct all other Y recursion eiφ ∂ ∂ + i cotθ ∂θ ∂φ Y m (θ. (5.106).138) Employing (5.f.5. (5. (5.143) . φ) sin θ e−i φ = 0.122. 5.136) is normalized and has the proper sign.136) We note in passing that this expression and the expression (5.106.142) For this purpose we proceed by induction. (5.129).132). This follows from the identity ∂ ∂ − i cotθ ∂θ ∂φ One can use (5.139) = ∂ − m cotθ ∂θ P m (cos θ) . 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. i. We ﬁrst note that for m = − (5. φ). a sign consistent with the family of functions Y m constructed above. the expression (5. (5.80) and (5. Obviously.

(5.144) reads P m+1 (x) = +m+1 m+1 ∂ (−1)m+1 (1 − x2 ) 2 (x2 − 1) . etc. yields the same eigenfunctions as the previous construction beginning with Y and stepping down the series of functions Y −1 . According to both (5. equivalently. one notes that application of (5. (5. of (5.e. 2 ! ∂x +m (5.141) reads then P m+1 (x) = 1 (5.145) Hence. i. Hence. P (x) = P 0 (x) are called Legendre polynomials.142) holds for m + 1 if it holds for m.142) holds for m = − we veriﬁed that it holds for all m. We then assume that (5. 1 (1 − x ) ∂x +m (1 − x2 ) 2 1 (5.147) the r.h. of which agrees with the r. also the associated Legendre polynomials determined this way..149) (5.h.124) and (5.146) i. Y − +2 . Since (5.e.142) one can state P (x) = ∂ (x2 − 1) .128) m + 1 → −m or. Y −2 ..142) and by (5.106) this implies for Y The Legendre Polynomials The functions the identity (5. However.124) yields (−1)m ( − m)! P ( + m)! −m (x) = +m m ∂ (−1)m (1 − x2 ) 2 (x2 − 1) .142) holds for m.124).135).s. φ) ( − m)! P ( + m)! −m (x) . given by (5. m → −m − 1 yields −(1 − x2 ) 2 = +m m ∂ ∂ (1 − x2 ) 2 (x2 − 1) ∂x ∂x +m +m+1 m+1 ∂ −(1 − x2 ) 2 (x2 − 1) ∂x +m+1 +m x 2 m ∂ 2 +m . Accordingly. (5. 2 ! ∂x 1 (5.132). (5.116 Theory of Angular Momentum and Spin wich agrees with (5. 2 ! ∂x +m −(1 − x2 ) 2 Replacing in (5.142).150) .148) According to (5.144) ∂ x − m 1 ∂x (1 − x2 ) 2 +m m ∂ (−1)m (1 − x2 ) 2 (x2 − 1) . etc.s. are identical. The construction beginning with Y − and continuing with Y − +1 . one can conclude P m (x) = (−1)m m (θ. 2 ! ∂x +m+1 (5.

92 To prove this property see.152) and.157) = 0. pp. Lebedev (Prentice-Hall. 1965) which is an excellent compendium on special functions employed in physics. In case q = 1 and |r2 | < |r1 | holds the identity2 ∞ 1 r2 = P (cos γ) (5. Englewood Cliﬀs.158) see. the spherical harmonics for m ≥ 0 Y m (θ.N. in case of x = 1 holds w(1. e. =0 (5.J. 2 3 1 1 = = 2 1−t 1 − 2t + t ∞ t . =0 (5. Jackson (John Wiley.D. and r2 denotes the location of the charge. 2nd Ed.155) P (x) t . t) = √ 1 − 2xt + t2 ∞ 1 − 2 x t + t2 .151) (5x3 Comparision of (5. P1 (x) = x .150) and (5. t) is called a generating function of the Legendre polynomials3 . pp. 1975). (5.5. t) = √ Comparision with (5. for example.156) yields P (1) = 1 . The spherical harmonics for m < 0 can be obtained using (5. r1 is the point where the potential is measured. N.153) P (cos θ) eimφ . The generating function allows one to derive useful properties of the Legendre polynomials. Using x = cos γ.. (5.” by J. φ) = 2 + 1 ( − m)! (−1)m sinm θ × 4π ( + m)! × ∂ ∂ cos θ m (5. The Legendre polynomials arise in classical electrodynamics as the expansion coeﬃcients of the electrostatic potential around a point charge q/|r1 − r2 | where q is the charge.132). 45 of “Special Functions and their Applications” by N. . accordingly. New York.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) . 1.g.4: Angular Momentum Eigenstates The ﬁrst few polynomials are P0 (x) = 1 . .154) can be written 1 = w(x. m ≥ 0 ∂xm (5. . . and |r1 − r2 | = r1 (5. ”Classical Electrodynamics.156) w(x. (5.154) |r1 − r2 | r +1 =0 1 where γ is the angle between r1 and r2 . P3 (x) = 1 2 117 P2 (x) = − 3x) 1 2 2 (3x − 1) . t = r2 /r1 .. For example. 2.

θ. φ) m (θ. π + φ) = (−1) Y m (θ. The spherical harmonics are eigenfunctions of the angular momentum operators J2Y J3 Y m (θ. Inversion Symmetry of Y m (θ. π + φ). Accordingly. Accordingly.159) by w(1 − 2xt + t2 ) leads to the diﬀerential equation obeyed by w(x. . ∂t 1 − 2xt + t2 (5.118 For w(x.151)] allows one to determine P (x) for = 1.167) (5.160) this diﬀerential equation is equivalent ∞ (1 − 2xt + t2 ) =0 P (x) t + (t − x) =0 P (x) t = 0 (5.162) − (2 + 1)x P (x) + P −1 (x) ] t + =1 [( + 1) P +1 (x) The coeﬃcients for all powers t must vanish individually.165) since ∂ n /∂(−x)n = (−1)n ∂ n /∂xn . . holds P1 (x) ( + 1) P +1 (x) = = x P0 (x) (2 + 1) x P (x) − P −1 (x) (5. t) holds Theory of Angular Momentum and Spin ∂lnw(x.166) We want to summarize the properties of the spherical harmonics Y derived above. φ) .164) which.163) (5. (5. π − θ. φ) m (θ. If the spherical coordinates of r are (r. φ) (5. t) x−t = . using P0 (x) = 1 [c. Inspection of (5.161) Collecting coeﬃcients with equal powers t yields 0 = P1 (x) − x P0 (x) ∞ (5. φ) goes over to Y m (π − θ. . (5.142) allows one to conclude P m (−x) = (−1) +m P m (x) (5.156) and (∂/∂t) to P (x)t = ∞ ∂w + (t − x) w = 0 . .159) Using ∂lnw/∂t = (1/w) ∂w/∂t and multiplying (5. φ) Under inversion at the origin vectors r are replaced by − r. then the coordinates of − r are (r. 1.168) mY m (θ. 2.. φ) . Noting exp[im(π + φ)] = (−1)m exp(imφ) we determine Y Properties of Y m (θ. Due to cos(π − θ) = − cos θ the replacement r → −r alters P m (x) into P m (−x).f. φ) = = 2 ( + 1) Y m (θ. under inversion Y m (θ. φ) m (π − θ. ∂t P (x)t −1 −1 (5. t) (1 − 2xt + t2 ) Employing (5. π + φ). φ).

(5.) P0 (x) = 1 . φ) (5.174) m ≥ 0 P (cos θ) eimφ . φ) = 2 + 1 ( − m)! (−1)m sinm θ × 4π ( + m)! 2 ! × ∂ ∂ cos θ m (5. . 6. .171) m = dφ Y ∗ (θ. φ) .170) (5. φ) = δ δm m . φ) = 2 +1 P (cos θ) .175) (5.179) .e.176) = 1 +1 [ (2 + 1) x P (x) − are the Legendre polynomials. The spherical harmonics for m < 0 are given by Y −m (θ. in fact. 2. 5. φ) .173) m+1 (θ. φ) Y m (θ. P −1 (x) ] (5. where ( = 1. The spherical harmonics form. . φ) where J± = J1 ± iJ2 .4: Angular Momentum Eigenstates 119 2.177) Note that the spherical harmonics are real. m 4. For the Legendre polynomials holds the orthogonality property +1 dx P (x) P ( (x) = −1 2 δ 2 +1 . The spherical harmonics form an orthonormal basis of the space C∞ (S2 ) of normalizable.169) 3. φ) m (θ. The spherical harmonics obey the recursion relationships J+ Y J− Y m (θ. m (5. (5.5. φ) = = ( + m + 1)( − m) Y ( + m + 1)( − m) Y m+1 (θ. except for the factor exp(imφ). The spherical harmonics Y 0 are Y 0 (θ.178) 7..172) (5. φ) ∈ C∞ (S2 ) holds ∞ f (θ. a complete basis of C∞ (S2 ). 4π (5. φ) 2π 0 (5. 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. φ) = (−1)m Y ∗ (θ. for any f (θ. φ) C = =0 m=− π C sin θdθ 0 m Y m (θ. The spherical harmonics are given by the formula Y m (θ. i. P +1 (x) P1 (x) = x .

(5. t)2 = −1 .179) using the generating function w(x.120 8. (5. =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.189) evaluate the integral on the l. The spherical harmonics for θ = 0 are Y m (θ Theory of Angular Momentum and Spin = 0.s.181) = 0.180) 9.156).h. The spherical harmonics obey the inversion symmetry Y 10.4. π + φ) . Exercise 5.s. φ). φ) = 1 ∂2 r − r ∂r2 ( + 1) r2 h(r) Y m (θ. For this purpose start from the identity +1 ∞ +1 dx w(x.h. 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 (5.87).. 11. (5.4.188) Exercise 5.189).1: Derive expressions (5.184) Y22 Y21 Y20 together with (5. (5. φ) .187) h(r) Y m (θ.86). φ) = δm0 2 +1 .177).3: Construct all spherical harmonics Y3m (θ.4. t) stated in (5.182) Y11 Y10 = = − (5. The spherical harmonics for m (θ. For the Laplacian holds 2 = = = (5.186) (5. of (5. .2: Derive the orthogonality property for the Legendre polynomials (5.185) (5. φ) = (−1) Y m (π − θ. 1.183) (5. 4π (5. Exercise 5.

. − + 1.. orthonormal basis for the function c m Y m (θ. }] (5.. φ) D(ϑ) = . (5. φ) c = . m = − .m m m (θ. . therefore. ..197) The subspaces X have the important property that (i) they are invariant under rotations exp − i ϑ · J .5: Irreducible Representations 121 5. m = S2 i dΩ Y ∗m (θ. 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 . .. φ) = exp Since the spherical harmonics Y space C∞ (S ) one can expand ˜ f (θ.192) = S2 dΩ Y ∗ (θ . .e..− + .5 Irreducible Representations We will consider now the eﬀect of the rotational transformations introduced in (5... ˜ f (θ.. .42)..e. φ) i − ϑ·J f (θ. .. ∞. Comparision with (5. φ) . . assumes a particularly simple form. φ). . φ) exp − ϑ · J . }. For this purpose we consider the subspaces of C∞ (S ) X = [ {Y . . are characterized completely if one speciﬁes the transformation of any Y m (θ.193) shows ∞ = − .195) i. φ ) (5.193) where [ { } ] denotes closure of a set by taking all possible linear combinations of the elements of the set.195). φ) (5. (5. i. i. = 0. = − .m D(ϑ) m. = . exp − i ϑ · J (X ) = X . φ) m. φ) ˜ under such rotational transformation by f (θ. The transformations exp − i ϑ · J . . . on functions f (θ. .e. We like to argue that the matrix representing the rotational transformations of the function space C∞ (S ). m Y m (θ. }] .191) i − ϑ·J f (θ . (5. φ) ˜ Y m (θ.196) C∞ (S ) = = X . φ) (5.194) Y m (θ. We denote the image of a function f (θ. exp − i ϑ · J . ∞.. (5.190) provide a complete. called the irreducible representation. and (ii) they form the lowest dimensional sets X obeying C∞ (S ) = X which have this invariance property. m .. with elements given by (5. = .− + .5. 1. . φ ) exp m One can also represent C∞ (S ) C∞ (S ) = [ {Y . φ) ∈ C∞ (S ). . . if one speciﬁes the functional form of the coeﬃcients [D(ϑ)] m .

. with blocks of dimension 1. J− 2 also leaves n1 X invariant. (5.199) [J+ . equivalently. can be expanded in powers One employs then the commutation properties [J3 ..e.. Since any additive term n n n cn1 . which follow readily from (5. m = δ S2 dΩ Y ∗ (θ.195) [D(ϑ)] m. 5.. one concludes for the matrix elements (5. 1×1 3×3 .198) 2 2 where ϑ± = ϑ1 ( 1 ϑ− J+ 2 + 1 2 ϑ+ J− ϑ2 . Since the Y m are n3 eigenfunctions of J3 the action of J3 reproduces the basis functions of X . φ) exp m i − ϑ·J Y m (θ. . J− ] = 2 J3 . If one orders the basis according to the partitoning (5.65). .e. Application of (5. One = contribute can conclude then that in the expansion (5.199) allows one to express exp i − ϑ·J = n1 . . J± ] = ± J± (5. one needs to consider the action of J+ 1 J− 2 J3 3 on the subspaces X . one can conclude that J+ leaves X invariant.n2 . produces 0. φ) . (5. .201) From expression (5.. One proceeds by expanding ﬁrst ϑ · J in terms of J+ . i.56. n The action of J− on Y m produces Y m−1 or. i. 5.n3 J+ 1 J− 2 J3 3 .200) in (5.202) D(ϑ) = (2 + 1) × (2 + 1) . 3.197) of C∞ (S ) the matrix representation of D(ϑ) assumes a block-diagonal form.e. Similarly. leaves X invariant.n3 n n n Accordingly.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 . i. .n2 . (5.194) only spherical harmonics with or. and J3 1 1 ϑ · J = ϑ− J+ + ϑ+ J− + ϑ3 J3 (5.200) leaves X invariant.200) of the rotational transformations one can also conclude that the X are the smallest invariant subspaces of C∞ (S )... J− . in case m = − . the transformation exp − i ϑ · J leaves X invariant as well. The exponential operator exp − i ϑ · J + ϑ3 J3 )n .

i. f (α + 2π) = f (α). . though seemingly natural.49). i.e. Exercise 5. 3 certainly allow one to describe any rotation around the origin. for which holds RT R = R RT = 1 1.22. This step reconsiders the parametrization of rotations by the vector ϑ assumed sofar. However. other representations are termed reducible representations. A ∈ C. k = 1. and for which detR = 1.1: Representations of the group SO(2) SO(2) is the set of all 2×2 matrices R which are orthogonal. 5. 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). The representation is referred to as the irreducible representation.5. 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.6: Wigner Rotation Matrices 123 This representation of D(ϑ) is the one wich has blocks of the lowest dimensions possible. (iii) a third rotation around the new x3 –axis by an angle γ.e.6 Wigner Rotation Matrices We will now take an important ﬁrst step to determine the functional form of the matrix elements of D(ϑ). Give the corresponding expression of ρ(ϕ). (ii) a second rotation around the new x2 –axis by an angle β. 5. (a) Show that SO(2) with the group operation ◦ deﬁned as matrix multiplication. 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 . and periodic functions f (α). (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). ∈ N} invariant. 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). this parametrization. 2. does not provide the simplest mathematical description of rotations. The three components ϑk .5. (e) In the space C∞ ( ) of inﬁnitely often diﬀerentiable. (b) Prove that the elements of SO(2) can be completely characterized through a single parameter. (d) Show that the similarity transformation T (ϕ) deﬁned in the space of non-singular. Determine the generator of ρ(ϕ) in analogy to (5.

209) The ﬁrst two rotations in (5. For this purpose we notice that J2 can be expressed through the similarity transformation J2 = e− i αJ3 J2 e i αJ3 (5. β.e. however.. followed by J2 in the original frame.207) are equivalent.210) The third rotation in (5. one can replace (5. Obviously. followed by transformation (i). (i) and (ii). the transformation from the rotated frame to the original frame. For any similarity transformation involving operators A and S holds eS A S Accordingly. γ ∈ R such that exp i − ϑ·J = e− i γJ3 − i βJ2 − i αJ3 e e (5.s. Accordingly.e. i.195) by ˜ Y m (θ. φ) = S2 dΩ Y ∗m (θ. we can write e− i −1 = S eA S −1 . We will demonstrate that (5.203) has the disadvantage that it employs rotations deﬁned with respect to three diﬀerent frames of reference.h.211) Using (5.204) is satisﬁed. γ will be referred to as Euler angles. γ)] = . of (5. γ)] m. φ) e− γJ3 − i βJ2 − i αJ3 e e .h.209).212) . β.205) Y m (θ.208) αJ3 βJ2 = e− αJ3 e− i βJ2 e i (5.206) The expression (5.194. φ) [D(α. 5.e.. The axis x2 is deﬁned in the coordinate frame which is related to the original frame by rotation (i). i.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. φ) (5..203) For any ϑ ∈ R one can ﬁnd Euler angles α. in analogy to (5.s. β. can then be written e− i βJ2 − i αJ3 e = e− i αJ3 e− i βJ2 (5.m m. the axis x3 is deﬁned in the coordinate frame which is related to the original frame by the consecutive rotations (i) and (ii). (5. m [D(α. the transformation from the original frame to the rotated frame.207) which replaces J2 by the inverse transformation (i). and the r. (5. β.124 Theory of Angular Momentum and Spin The angles α. the l.203).203) can be expressed. m i Y m (θ.203). is e− i γJ3 = e− i βJ2 e− i αJ3 e− i γJ3 e i αJ3 e i βJ2 (5. in terms of rotations deﬁned with respect to the original frame. i. The Euler rotation replaces exp − i ϑ · J by e− i γJ3 − i βJ2 − i αJ3 e e . i (5.

5.218) Deﬁning the so-called Wigner rotation matrix dmm (β) = one can express the rotation matrices [D(α. This allows one to express the rotational trasnformation (5. φ) e−imγ . (5.217) Accordingly. γ)] m. φ) e− αJ3 e− βJ2 Y m (θ.214) e− γJ3 = S2 dΩ Y ∗m (θ.206) by ˜ Y m (θ. (5. φ) . The eigenvalue property (5. φ) e−iγm δ (5.m m. φ) = e−im α S2 dΩ Y ∗ (θ. φ) = S2 dΩ f (θ. β.205.215) The evaluation of the matrix elements (5.219) = e−iαm dm m (β) e−iγm δ ( ) . m ( ) S2 dΩ Y ∗ (θ. to redeﬁne the three rotations in (5. 5.. φ) e− m i βJ2 Y m (θ. (5. but many other elementary components of matter are endowed with spin-like properties. γ)] = . in addition.e. Examples are the other ﬁve members of the lepton family to which the electron belongs. 5.64) yields dΩ f (θ. φ) e− m S2 i αJ3 f (θ. β. γ)] m. m = e−iαm S2 dΩ Y ∗ (θ. the muon µ and its neutrino νµ of the second .213) i. φ) [D(α. β.7: Spin 1 2 and the group SU(2) 125 Multiplication from the left with (5. φ) . φ) f (θ. γ)] m. 5. m (5. (5.215) beneﬁts from the choice of Euler angles for the parametrization of rotational transformations.220) We will derive below [see Eqs. φ) e− S2 i γJ3 Y m (θ. m [D(α. m i Y m (θ.310)] an explicit expression for the Wigner rotation matrix (5. the electron e and the electron neutrino νe of the ﬁrst generation. φ) e− m i βJ2 Y m (θ. one can write [D(α. Best known is the spin of the electron. φ) i i (5. φ) Y m (θ.219). the self-adjointness of the operator J3 one can state dΩ Y ∗ (θ. (5. β.203) with respect to the original (unprimed) frame one simply needs to reverse the order of the rotations.7 Spin 1 2 and the group SU(2) The spin describes a basic and fascinating property of matter.309.216) Using. φ) (5.210) yields the simple result e− i γJ3 − i βJ2 − i αJ3 e e = e− i αJ3 e− i βJ2 e− i γJ3 .

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

(5.225) σ (5. it holds.k=1 σ σ 3 j k j.k=1 j<k + σ j σ k aj bk . σ2 = 0 −i i 0 .229) = 1 .31) of the group SO(3). y. e. x.e. but their comonents must not necessarily commute with each other. The Pauli matrices obey the following anti-commutation properties σj σk + σk σj = [σj . (5. one obtains (σ · a) (σ · b) = 3 j 2 j=1 (σ ) aj bj 3 j k j. σk ]− = 2i jk z x − iy x + iy −z .230) where a. ∈ R . (σ · a) (σ · b) = a · b 1 + iσ · (a × b) 1 (5. σ3 = 1 0 0 −1 .k=1 σ σ aj bk + j>k aj bk = 3 j. Cliﬀord algebras play an important role in the mathematical structure of physics.e.224) Algebraic Properties of the Pauli Matrices The Pauli matrices (5. it might hold aj bk − bk aj = 0. i. b are vectors commuting with σ.. namely. 2. hermitian 2 × 2–matrices A can be expandend. 1 σj σk = − σk σj for j = k which can also be readily veriﬁed. 2 S3 = 1 σ3 2 (5. z.227. Any such A can be expressed in terms of three real parameters x. z A = In fact.229) and avoids commuting the components aj and bk .227) which is essentially identical to the Lie algebra (5. (5.231) . According to this property the Pauli matrices generate a 3dimensional Cliﬀord algebra C3 . y. (5. 1 i. 2 S2 = 1 σ2 . σk ]+ = 2δjk 1 .5. One can readily verify the commutation property σj σk − σk σj = [σj . the simplest choice being S1 = 1 σ1 .226) The Pauli matrices satisfy very special commutation and anti-commutation relationships.g.7: Spin 1 2 and the group SU(2) 127 exist three such linear independent matrices. they are associated with the important fermion property of matter.224) provide a basis in terms of which all traceless. At this point we will state a useful property.223) where σk . 5. A = x σ 1 + y σ 2 + z σ3 . k = 1. 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. The proof of this relation rests on the commutation relationship and the anti-commutation relationship (5. 3 are the Pauli matrices σ1 = 0 1 1 0 . σj 2 (5. In fact.228) (5.

230).e. yields a · b. This relationship emerges. (5.k=1 j>k 1 2 3 j. let say around the x2 –axis by 360o . i.k=1 j>k i σl (aj bk − ak bj ) = i =1 σ a×b . In the special case a = b ∈ R holds (σ · a)2 = a 2 1 .236) The matrix elements of this 2 × 2 operator can be labelled by the basis states χ+ and χ− .235) which is identical to that of the generators of SO(3). Well almost like it. 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. For this purpose we choose to replace (5. (5.k=1 (σ σ j>k σ k σ j (aj bk − ak bj ) = (5. S3 χ± = ± 1 χ± 2 .128 Theory of Angular Momentum and Spin Using (σ j )2 = 1 the ﬁrst term on the r. We like to derive this result now by evaluating the transformations of SU(2) explicitly.231) proves (5.h. S ] = i k m Sm (5. since there is a slight diﬀerence: when you interprete the 3-component ϑ in (5. 1 1 1 ( + 1) χ± 2 2 . 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 .234) 5. (5.222). however. The commutation property (5.k=1 j>k σ j σ k (aj bk − ak bj ) = 1 2 3 j. 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 (5.233) This result together with (5.238) S 2 χ± = . however. hence. to rotations in space.237) + 1)1 and. 1 (5.222) as a rotation vector and you rotate once.222) by the Euler form exp(−iγS3 ) exp(−iβS2 ) exp(−iαS3 ) . (5.232) − σ k σ j ) (aj bk − ak bj ) . only a 720o rotation leaves the spin invariant. The two remaining terms yield using σ j σ k = −σ k σ j for j = k and altering ‘dummy’ summation indices 3 j.s.8 Generators and Rotation Matrices of SU(2) Sofar there is nothing which relates the transformations U .227) leads to 3 3 jk j.k=1 j>k σ j σ k (aj bk − ak bj ) − 1 2 3 j k j. spin changes sign.

β. can be constructed formally from spin 2 2 states χ± if one considers the two properties χ± to be carried by two kinds of bosons. . 2 2 We obtain with this notation for the transformations (5.9: 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). use the label 2 2 | 1 ± 1 for the states χ± . . (5. .242) The expressions in brackets [. (5. . i. (5. .219). therefore.9 Constructing Spin States with Larger Quantum Numbers Through Spinor Operators In this section we like to demonstrate.243) We note the property of this rotation matrix 2 dmm (2π) (1) = −1 1 . rotation by 360o changes the sign of the spin state.240) . γ)] m. One cannot consider the 1 entities carrying the spin– 2 to be particles in the ordinary sense for it can be shown that spin– 1 2 particles have fermion character. 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 .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. .e.] can be identiﬁed with the Taylor expansion of the cos. 2. m = − . 1 . − + 1.244) i. We may.5.245) 5.236) 1 2m | exp(−iγS3 ) exp(−iβS2 ) exp(−iαS3 ) | 1 m 2 1 2m = e−iγm | exp(−iβS2 ) | 1 m e−iαm 2 (1) 2 = e−iγm dmm (β)e−iαm (5.e. . that states | m for higher quantum numbers = 0. The complete matrix elements (5. . 1. .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 .239) in the notation (5. i. no two such particles can exist in the same state.220) are [D(α. following Jourdan and Schwinger.219. In order to determine the Wigner matrix in (5.239) where we made use of the notation (5.e.239 ) one expands the exponential operator exp(−iβS2 ). (5. ( − β S2 )2n = (−1)n ( − β S2 )2n+1 = (−1)n β 2 β 2 2n 1 1 0 1 −1 0 2n+1 (5. 3 . identical particles any number of which can exist in the same state χ+ and χ− .5. namely. The idempotence property of Sk yields particularly simple expressions for these powers. For this purpose one needs to determine the powers of −iβS2 ..

using (5. b− )T is expressed through the commutation relationships + − (ζ.250) which for x∗ x = x∗ x+ + x∗ x− = 1 represent spinor operators in an arbitrary reference system. b ζ = b† . These operators are associated with a given spatial reference system. We consider. b † ζ ζ =0 (5. (5. b− )T as spinor creation and annihilation + − operators. j. The boson character of the operators b† = (b† .248) bζ .5. m) † and bζ obey the commutation (5.254) We will show below that these states are orthonormal and form spin states with higher quantum numbers. b± |Ψ0 = 0 (5. b† |Ψ0 = χ− − . 1. . −) bζ .248) that bζ relationships † † b ζ. −j + 1. 1 . .e.252) (5.251) One can show using this property and (5. m = −j. therefore. 2 2 .e.243) the creation operators in a coordinate system rotated by an angle β around the y–axis are b+ † = cos β β † b + sin b† 2 + 2 − . b ζ = 0 b ζ. (5.249) In the following will refer to b† = (b† . m) = b† − j−m (j + m)! (j − m)! |Ψ0 . . b† )T and b = (b+ . . 2 + 2 − † (5. b† )T and b = (b+ . b ζ = b ζ. i. ζ = +. b ζ The States |Ψ(j. 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.246) The corresponding adjoint operators are denoted by b+ and b− . . b† |Ψ0 = χ+ + . .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 For the vacuum state |Ψ0 holds = δζζ .247) (5. b− † = − sin β † β b + cos b† . i. j = 0. also operators of the type x b† = x+ b† + x− b† + − .253) = δζζ .247. 3 . + − For example. x∗ b = x∗ b+ + x∗ b− + − (5. .

We will show that for this commutator holds n b+ .259) The property is obviously true for n = 0. b† + n b† + b† + b† + + n n n b+ .256) In the special case |state = |Ψ0 this reads using (5. The Spinor Derivative Operator A most important operation is the action of the annihilation operators b+ and b− on operators which can be expressed as polynomials. in b† and b† . + + (5.254).249) b+ f (b† ) |Ψ0 = + b+ .260) . + b+ . If it is true for n then using (5. b† + b+ . f (b† ) |state + f (b† ) b+ |state . For this purpose we consider the following polynomial of creation operators N j+m j−m f (b† ) = + n=0 cn b† + n .248) and from (5.10: Algebraic Properties of Spinor Operators 131 5. (5.258) Equations (5. . (5. b† + n+1 = = b+ . b† + − well-known for the quantum mechanical harmonic oscillator. An example + − is the monomial b† which generates the states (5.249). we need to evaluate b+ . b† + n .257) motivate us to determine the commutator (5. possibly inﬁnite power series.247. that the annihilation operators play a role similar to the diﬀerential operator in calculus.257) which for (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+ . 5. b † + = n b† + n−1 b† + b † + + n = (n + 1) b† + .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. (5. b† + = n b† + n−1 . f b† Obviously.256) and (5.5. b† + n = n cn b+ .255) is b+ . f (b† ) + (5.248) we obtain b+ . We will derive the result. f (b† ) |Ψ0 .

φjm (x) Ψ (j.263) According to (5. f (b† .259) holds for all n ∈ N.266) where xb† has been deﬁned in (5.247) the more general property for polynomials f (b† . = f (b† ) + (5. one can prove b− . f (x)] = ∂k f (x) . (5.264) This demonstrates the equivalence of the spinor operators bζ and the derivative operation. 2 . (5.. the property holds also for n + 1. f (b† ) − = f (b† ) − (5. b† ) in b† and b† holds (ζ = +.e.250) and where φjm (x) represents the function of the two variables x+ and x− closely related to |Ψ(j..261) Similarly. In order to derive (5. + − (5. (5.262) Because of the commutation relationships (5.1.268) . −) + − + − bζ .255) b+ .267) exp xb† |Ψ0 is called a generating function of |Ψ(j. m) m=−j (5. m) φjm (x) = j−m xj+m x− + (j + m)! (j − m)! .263) corresponds to the product rule of calculus ∂k f (x)g(x) = (∂k f (x))g(x) + f (x)(∂k g(x)) which can be written ( ∂k f (x) − f (x)∂k ) g(x) = [∂k . f (b† ) + where f (x) = df dx .132 Theory of Angular Momentum and Spin i. m) We want to prove now the property ∞ j exp xb † |Ψ0 = 1 j=0. b† ) + − = ∂ ∂b† ζ f (b† . f (x)] g(x) = (∂k f (x))g(x) or [∂k . b† ) . Equation (5. m) = x− b† − j−m (j + m)! (j − m)! |Ψ0 . One can ﬁnally conclude for polynomials (5. By induction we can conclude that (5. + − (5..257) we can conclude in particular bζ f (b† .265) Generating Function of the States |Ψ(j. m) . b† ) |Ψ0 = + − ∂ ∂b† ζ f (b† . b† ) |Ψ0 .266) we compare the terms x+ b† + j+m φjm (x) Ψ (j.

m) = = |Ψ0 (N − s)! s! N −s 1 N! x+ b† x− b† + − N ! (N − s)!s! N −s 133 (5. m) 1 j=0.5.272) The summation over s can be written in terms of j and m using (5.275) = exp xb† |Ψ0 . 1 . 2 2 leads to ∞ j ∞ φjm (x) |Ψ (j.273) Change of the summation indices allows one to conclude j φjm (x) Ψ (j. 1 . 2 .274) Summing this expression over 2j = 0. . i.10: Algebraic Properties of Spinor Operators with the s–th term in the binomial expansion of (a + b)n n! an−s bs . s! (n − s)! Deﬁning j+m = N −s j−m = s. 1..1. one obtains φjm (x) Ψ (j.269) (5. 1 xb† u! |Ψ0 (5.. m) m=−j = = 1 x+ b† + x− b† + − (2j)! 2j 1 xb† |Ψ0 .. . (5. . ... 2... . 2. .270) s x+ b† + N −s x− b† − s |Ψ0 . 2 ∞ 1 xb† 2j! u 2j |Ψ0 = u=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. (2j)! 2j |Ψ0 (5. 1... (5. m=−j = j=0.e.270) N 2j j → s=0 j−m=0 → m=−j .1. 3 ..1.. chosing the summation index j = 0.

b− )e(yb ) |Ψ0 = f (y+ .279) One can generalize this to † † s (bζ )s e(yb ) |Ψ0 = yζ e(yb ) |Ψ0 .134 Theory of Angular Momentum and Spin which concludes our derivation. For this purpose we consider the inner product † † e(xb ) Ψ0 |e(yb ) Ψ0 = δjj δmm (5. hence. therefore. 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 (5.249) for any non-vanishing integer s holds + − Ψ0 |bs Ψ0 = 0 and. (5. m) |Ψ j . b† ] = 0 and [b− . b− ] = 0.m.254) are orthonormal. m) and will be used for this purpose below.284) . (5.283) (5.264) implies † † bζ e(yb ) |Ψ0 = yζ e(yb ) |Ψ0 .j . In order to evaluate the ∗ b) (yb† ) operator e(x e we notice that the derivative property (5.s. The generating function allows one to derive various properties of the states |Ψ (j.277) by Ψ0 |e(x b) e(yb ) Ψ0 .281) and.h. (5. that Ψ (j. m . (5. According to (5.e.278) † ∗ which allows us to replace the l.277) To evaluate this expression we ﬁrst notice † e(xb ) † = e(x ∗ b) .276) = j. m) |Ψ j .280) The commutation properties [b+ . [b† . Orthonormality of the States |Ψ (j. m holds. b† ] = 0 allow one to state that for + − + any polynomial f (b+ . y− )e(yb ) |Ψ0 (5.282) where we deﬁned x∗ y = x∗ y+ + x∗ y− . i. we can conclude † † ∗ e(xb ) Ψ0 |e(yb ) Ψ0 = e(x y) (5. b− ) holds † † f (b+ . of (5.m φjm (x∗ ) φj m (y) Ψ (j.

J1 = J2 = J3 = 1 † b b− + b† b+ − 2 + 1 † b b− − b† b+ − 2i + 1 † b b+ − b† b− .m.282) and (5. 2.m. 3 . bm bm } n n = 1 4 < n |σi | m >< n |σj | m > {b† bm δmn − b† bm δm n } n n n.268–5.n . m) |Ψ j . Jj ] = = 1 4 1 4 < n |σi | m >< n |σj | m > b† bm .j .10: Algebraic Properties of Spinor Operators and comparing (5.288) The three operators obey the Lie algebra of SU(2) [Ji . k = 1.224).m 135 = e(x ∗ y) . (5. (5.224). m) as given in (5. (5.5.m +b† b† . . m j.m.254) are eigenstates of operators J 2 and J3 with eigenvalues j(j + 1) and m where J 2 and J3 are representations of the spin 2 2 2 operators S 2 = S1 + S2 + S3 and S3 deﬁned in 5.m.223.289) < n |σi | m >< n |σj | m > {b† bm .276).287) ζ 2 ζ.m φjm (x∗ ) φjm (y) . 2 .277) φjm (x∗ ) φj m (y) Ψ (j. . (5. 1.m ..m ijk Jk . etc. 1 New Representation of Spin 0.286) from which follows immediately the orthonormality property (5. Jj ] = i This property is derived as follows [Ji . One deﬁnes corresponding operators Jk through 1 Jk = b† < ζ |σk | ζ > bζ . The 2 operators are explicitly.ζ where σk . States 2 We want to demonstrate now that the states |Ψ (j. < + |σ1 | − > = < − |σ1 | + > = 1. b† bm n n n.285) Following steps similar to those in Eqs.275) one can show e(x ∗ y) = j. (5. 3 denote the Pauli spin matrices (5. 5. b† bm n n n.n . In our present notation the matrix elements < ζ |σk | ζ > for ζ = ± correspond to the matrix elements < ζ |σk | ζ > for σ = ± 1 .n . using < + |σ1 | + > = < − |σ1 | − > = 0. − 2 + (5. .

2 J 2 = J+ J− + J3 − J3 = J3 (J3 − 1) + J+ J− (5. (5.m < n |[σi .292).m. (5. m) are eigenstates of J 2 and J3 . (5.m. J1 + iJ2 ] = −i [J2 .m ijk Jk .289) yield [J− .293) . J+ ] = [J1 − iJ2 .136 = 1 4 − = = 1 4 1 4 Theory of Angular Momentum and Spin < n |σi | m >< m |σj | m > b† bm n n.297) .k ijk ijk σk | m > b † bm n = i = i 1 2 < n |σk | m > b† bm n n. (5. σj ]| m > b† bm n n.291) Here we have deﬁned J+ J− = J1 + iJ2 = J1 − iJ2 = b† b− + = b† b+ .288) yields ﬁrst the expression J2 = which results in J2 = 1 4 1 † ( b + b + − b † − b− ) ( b † + b+ − b† − b − − 2 ) + b † + b − b † − b + 4 .295) The last result together with (5. m) as Eigenstates of J 2 and J3 We wish to show now that the states |Ψ (j.m 1 4 < n |σj | m >< m |σi | m > b† bm n n .290) |Ψ (j. J1 ] + i [J1 . J2 ] = −2 J3 from which we can conclude the property 1 1 J− J+ = J+ J− − J3 2 2 Hence. To this end we note 2 2 2 J 2 = J1 + J2 + J3 = 1 1 2 J+ J− + J− J+ + J3 2 2 . (5.296) ( b † + b + b † + b+ + b † + b + b † − b − − 2 b † + b + − b † − b − b † + b + + b † − b − b † − b − + 2 b † − b − + 4 b † + b − b† − b + ) (5.292) The commutation relationships (5. (5.m. − (5.m < n |2i n.294) .

2 2 137 (5. n = n1 + n2 ﬁxed. m I1 = where λ = 0.300) (5. (5. .298) (5. i. m) J− |Ψ (j. Ik ] = i jk I . . and eigenstates of k with eigenvalues j. 2.10.e. using operators I± = I1 ± iI2 2 2 2 and the subspace {|n1 . m) One can furthermore derive readily J+ |Ψ (j. Construct. the states (5. . m = λ||λ. .302) (5. . n} eigenstates of I 2 = I1 + I2 + I3 and of I3 I 2 ||λ. . b+ ] = δjk .254) are eigenstates of + b+ and − b− with eigenvalues j + m and j − m. . . 1 2. m = m||λ. I2 = (b1 b2 − b+ b1 ) . . n2 . State the corresponding eigenvalues and the degree of degeneracy. −λ + 1. m + 1) (j + m) (j − m + 1)|Ψ (j. m . m) m |Ψ (j. λ. . I3 = (b+ b1 + b+ b2 ) 2 2 2 2 2i 2 1 satisfy the Lie algebra of SU(2).h.301) Obviously.303) Exercise 5. 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. .5. i. m) J3 |Ψ (j. [Ij .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) . (b) Show that the three operators 1 + 1 1 + (b1 b2 + b+ b1 ) . m) = = (j + m + 1) (j − m) |Ψ (j. 2 .305) = = j ( j + 1 ) |Ψ (j. . j = 1.e. 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† .. 1. m) .304) (5.s.10: Algebraic Properties of Spinor Operators The last term on the r. m = −λ. Show λ = n1 +n2 n1 +n2 2 ( 2 + 1) and m = 1 2 (n1 − n2 ). n1 = 0. I3 ||λ. (5. . the number of states to the possible energy eigenvalues. ˆ respectively. b † + b + + b † + b+ b † − b − 1 1 b† + b + + b † + b + b † − b− + b † − b − b † + b + .299) (5. 1 2 k (a) Show that the eigenstates are given by 1 |n1 . One can then conclude J 2 |Ψ (j. b† . [bj . m − 1) .

m = b† + j+m b† − j−m (j + m )! (j − m )! |Ψ0 (5.251). (j) (5.307) shows that the elements of the rotation matrix can be obtained by binomial expansion of b † and b † in terms of b† and b† . for example.243).310) it reduces to (5. + − One may expect that these two conditions restrict both σ and σ.306) where b † and b † are given by (5. On the other side the states |Ψ (j.309) . m ) are related to the + − states |Ψ (j. .254) are |Ψ j. 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. m = m=−j dm m (β) |Ψ (j. For this purpose we note that the states |Ψ(jm) in a rotated coordinate system according m to (5.308) The latter sum involves terms (b† )j+m (b† )j−m for σ + σ = j + m and 2j − σ − σ = j − m.307) Comparision of (5. The combination of σ.306.306) and (5. For this purpose we expand + − + − b† + cos β b† + sin β b† 2 + 2 − j+m σ =0 j−m σ=0 j+m j+m b† − j−m = j−m −sin β b† + cos β b† 2 + 2 − cos β 2 j−m −σ σ = j+m σ σ sin β 2 σ +σ j+m −σ j−m σ −sin β 2 cos β 2 b† + b† − 2j−σ −σ (5. m) .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). However.138 Theory of Angular Momentum and Spin 5.5. both conditions are satisﬁed for σ = j + m − σ. The prefactor of (b† )j+m (b† )j−m which according to (5. σ values which yields (b† )j+m (b† )j−m is + − then σ = j + m − σ. In case j = 1 this expression yields. m) in the original coordinate system by j |Ψ j.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.

308) and had the form (5. the matrix A should represent a rotation around the x2 –axis in the space of vectors ˜ (x1 .251). Evaluation of A yields √ −1 0 1 1 √ cosβ + 2 sinβ 1√ cosβ − 1 ˜ 0 A = 4 i √ i − 2 sinβ 2 cosβ 2 sinβ × √ 0 2 0 1 − cosβ − 2 sinβ 1 + cosβ √ −1 −i √ 0 −2 cosβ −2 2 sinβ 2 cosβ 2i 2i × × 0 0 2 = 1 4 √0 1 −i 0 −2 sinβ 2 2 cosβ 2 sinβ −1 −i √ 0 cosβ 0 −sinβ 1 0 (5. The SU(2) transformation entered in (5. It is of interest to trace the mapping from SU(2) onto SO(3) as represented by (5..312) The choice of this transformation derives from a property shown further below.251) and (5. 2 2 − (5.12: Mapping of SU(2) onto SO(3) 139 5. One can. conclude that the mapping from SU(2) onto SO(3) is a 2–1 mapping.312) transforms like an angular momentum state ˜ |1 m .309) and the particular matrix (5. b− † = sin β † β b+ − cos b† .g.314) leaves (5. namely that the component of the vector on the l.. = U x2 .e.12 Mapping of SU(2) onto SO(3) The representation (5. b+ † = −cos β † β b+ − sin b† 2 2 − . the orthogonal 3 × 3 matrix which describes a rotation around the x2 –axis.308) unaltered except for a factor (−1)2j which multiplies then also the ﬁnal result (5. This can be shown by (1) applying to the matrix A = (dm m ) in (5. This factor implies.310) to the SU(2) transformation assumed in deriving the general result (5. U = 0 1 1 i √ ( x1 − ix2 ) √ x3 − √2 0 2 2 (5. x3 ) ∈ R . is the expected element of SO(3). in case of j = 1. .s.313) × 0 0 2 = 0 sinβ 0 cosβ 1 −i 0 which.314) in case of integer j–values.311) (5. of (5. x2 . however.5. Hence.310) establishes a mapping of SU(2) onto SO(3).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 . e. that the the representation (5. i. in fact. i. Replacing the latter transformation by its negative form. therefore.309).e.h.310).309) does not distinguish between SU(2) transformations (5.

140 Theory of Angular Momentum and Spin .

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

. 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.max + 1)2 ( 2. The rotational symmetry of the interaction between the particles allows one.max = 10 one obtains d(B) = 14 641. 2. mB .max d(B) = 1 =0 (2 1 + 1) 2 =0 (2 2 + 1) = ( 1. composite systems behaving like elementary objects are common. that only states within the subspaces Bk are coupled in the scattering process. . and the following mathematical description will shed light on their ubiquitous appearence in physics.max + 1)2 (6. the dimensions d(Bk ) of these subspaces does not exceed 100. 1. . 1. − 1 ≤ m1 ≤ 1. as we will demonstrate below. albeit puzzling. . by the total energy E. 2} 1. mC . − 2 ≤ m1 ≤ (6.m2 c (n) r ρ 1. ρ 1 = 0.1) where 1. since further important degrees of freedom.max . by the masses mA .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. . = 0. The latter often arrives at the physical properties of composite systems by adding the .2) The dimension d(B) of B is 1. Obviously.3) For rather moderate values 1. 2 . rather than by Classical Mechanics. B1 ⊕ B2 ⊕ .max = ∆ 2 mA mB mC E . . 14 641} . .m2 Y 1 m1 (ˆAB ) Y 2 m2 (ˆC ). i. . In fact.max = 2. (6. .max = 1 2 IA−B E . have not even be considered. vibrations and rearrangement of the particles in reactions like AB + C → A + BC. and by the moment of inertia IA−B of the diatomic molecule A–B approximately as follows 1. the values of which are determined by the size of the interaction domain ∆V . . = B such. 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 ). . in fact. will make their appearence a natural consequence of the symmetry of the building blocks of matter. a very large number. 2. mA mB + mB mC + mA mC 2.m1 2 . Such extremely useful transformation of the problem can be achieved through the choice of a new basis set B = { 1 . each of which stands for two angles. n = 1.e.max denote the largest orbital and rotational angular momentum values.m1 . however. in particular. .max and 2. Such large number of dynamically coupled states would constitute a serious problem in any detailed description of the scattering process. (6.. 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.142 Addition of Angular Momenta and Spin ρC .max .max 2. to separate the 14 641 dimensional space of rotational states Y 1 m1 (ˆAB ) Y 2 m2 (ˆC ) into r ρ subspaces Bk . .

ρC ) deﬁned through R(ϑ) ψ(rAB . ρC ) it follows H R(ϑ) = R(ϑ) H.55) holds i (1) − J1 = zAB ∂ ∂ − yAB . In this sense.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) .48) i i R(ϑ) = exp − ϑ · J (1) exp − ϑ · J (2) (6. according to (5.7) i R(ϑ) = exp − ϑ · J (6. Hence.10) .6. q = 1. assuming presently that H might.5) according to (5. the existence of a basis (6. 3 (6.9) hold since the components of J (k) are diﬀerential operators with respect to diﬀerent variables. 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.53. 2. To specify this property mathematically let us denote by H the Hamiltonian in the rotated frame. the commutation relationships (1) Jp . 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. It holds then H R(ϑ) ψ = R(ϑ) H ψ. ∂ρx ∂ρy (6. ∂yAB ∂zAB i (2) ∂ ∂ − J3 = ρy − ρx . For ˆ ˆ example.4) which decouples rotational states is connected with the rotational symmetry of the Hamiltonian of the three particle system considered. 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 .e. Following our description of rotations of single particle wave functions we express (6.8) Obviously.0: Addition of Angular Momenta and Spin 143 corresponding physical properties of the elementary components. but solely under simultaneous and identical rotations of rAB or ρC .. (6. ρC ) = ψ(R−1 (ϑ) rAB . presumably a facility the reader would like to acquire with great eagerness. R−1 (ϑ) ρC ) (6. in fact. connected with the fact that the Hamiltonian H does not depend on the overall orientation of the three interacting particles.7) where the generators J (k) are diﬀerential operators acting on rAB (k = 1) and on ρC (k = 2). Jq(2) = 0 for p. be diﬀerent from H. Rotational Symmetry of the Hamiltonian As pointed out already. rotations R(ϑ) of the wave functions ψ(rAB . 5. the reader will learn in the following section how to add and subtract in the microscopic world of Quantum Physics. i. from which follows in turn the well-known result that H is related to H through the similarity transformation H = R(ϑ) H R−1 (ϑ). Since this is true for any ψ(rAB .

3 do not commute.12) We consider this equation for inﬁnitesimal rotations. The property (6. (6. componentwise.6) for rotational invariance of the Hamiltonian in the form i i H = exp − ϑ · J H exp + ϑ · J . ˆ Ω 2 = ρC . J ] = i k m Jm for n = 1. 3 cannot have simultaneous eigenstates among each other. J ] = i k m Jm (6. however. In the following we will use the notation r ρ Ω1 = rAB .144 where J = J (1) + J (2) Addition of Angular Momenta and Spin . B + C] = [A. 2. B] + [A. k = 1.15) are satisﬁed. i. r ρ Deﬁnition of Total Angular Momentum States The commutation property (6. (5.11) we can write the condition (6. we will ﬁnd that the states yield the basis B with the desired property of a maximal uncoupling of rotational states.71–5. (6. k = 1.f.12) each individually can have simultaneous eigenstates with the Hamiltonian.13) Hϑ · J − Since this holds for any ϑ it follows HJ − J H = 0 or.71–5.e.81)]. ˆ (6. a supposition which can be tested through the commutation properties of these operators. One can show readily that the commutation relationships [Jk . i. the (n) (n) (n) properties [Jk .14) We will refer in the following to Jk . 3 as the three components of the total angular momentum operator. the operators Jk . We suspect. To order O(|ϑ|) one obtains i ϑ · JH . (6. (6. Before we apply the procedure (5. and the eigenvalues of J2 and J3 are good quantum numbers.14) implies that the components of the total angular momentum operator (6.81) we want to consider the relationship between YJM and Y 1 m1 (ˆAB ) Y 2 m2 (ˆC ).e. the important fact that the Jk obey the Lie algebra of SO(3).16) following the procedure stated in the theorem above [c. 2. C]..14) implies that the total angular momentum is conserved during the scattering process. 2. In fact. [H.17) . i. We recognize. Eqs.e. for |ϑ| H ≈ 1 − 1 i ϑ·J H 1 + 1 i ϑ·J ≈ H + i 1. 2. 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 ). Jk ] = 0 . 2 together with the property [A.11) By means of (6. 3 . of course. 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. that energy. that the components Jk . k = 1.9).. For a proof one uses (6. k = 1.

m1 .s. hence. 2 |Ω1 . 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.m2 (J M | 1 m1 2 m2 ) Y 1 m1 (Ω1 ) Y 2 m2 (Ω2 ) (6.18) where the states YJM ( 1 . and J (1) · J (2) commute..18) to avoid summation of vanishing terms YJM ( 1 .18) yields using the property (k) J3 Y k mk (Ωk ) = mk Y k mk (Ωk ) . and J3 .6.h. . therefore. (6. (6. Ω2 ) = m1 (J M | 1 m1 2 M − m1 ) Y 1 m1 (Ω1 ) Y 2 m2 (Ω2 ) .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) . Exercise 6.e. (6. i. i. J3 YJM = M YJM . play a cardinal role in the mathematical description of microscopic physical systems. or the closely related Wigner 3j-coeﬃcients introduced further below.e.18) is an eigenfunction of J3 . (6.1: Show that J2 .m2 2 |Ω1 . The expansion coeﬃcients (J M | 1 m1 2 m2 ) ˆ are called the Clebsch-Gordan coeﬃcients which we seek to determine now. restrict the sum in (6. J (2) . an important example being the symmetry groups SU(N) governing elementary particles made up of two quarks. ρC ) are normalized. These coeﬃcients. J3 . mesons. 2 |Ω1 . 2 |Ω1 . baryons. The two missing quantum 2 2 numbers are 1 and 2 corresponding to the eigenvalues of J (1) and J (2) . the reader should always keep in mind that conditions equivalent to (6.21) One can.20) This equation can be satisﬁed only if the Clebsch-Gordan coeﬃcients satisfy (J M | 1 m1 2 m2 ) = 0 for m1 + m2 = M . two further quantum numbers need to be speciﬁed for a complete characterization of the total angular momentum states. Equivalent coeﬃcients exist for other symmetry properties of multi–component systems.. J (1) . J (2) . We.1: Clebsch-Gordan Coeﬃcients 145 6. Ω2 ) (1) (2) = 1 m1 2 m2 ) Y 1 m1 (Ω1 ) Y 2 m2 (Ω2 ) (m1 + m2 ) (J M | . of (6. We ﬁrst notice that YJM in (6. 2 M YJM ( 1 .19) Noting J3 = J3 + J3 and applying this to the l. Since YJM sofar speciﬁes solely two quantum numbers. k = 1. assume the expansion YJM ( 1 .e.1. Ω2 ) = m1 . i. However.22) We will not adopt such explicit summation since it leads to cumbersum notation. the eigenvalue being speciﬁed by the quantum number M .21) hold. and three quarks. J3 .

m2 2 M and with (6. it is not immediately obvious what the J– values are. = {YJM ( 1 . −J + 1. .s.. −J + 1. |J (1) | + |J (2) |]. (6. This obviously corresponds quantum mechanically to a range of J–values J = | 1 − 2 |. . 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. In fact. and that the states can be normalized. . Ω2 ) YJ M ( 1. . in fact. M = −J. the range of values assumed for J is correct. (6. J (1) and J (2) . The property dΩ1 dΩ2 Y∗ ( 1 . . . . Ω2 ). . (6. Exercise 6. 2 ).23) corresponding to the r. J. it holds 1+ 2 ( 2J + 1 ) = (2 J=| 1 − 2 | 1 + 1) (2 2 + 1) . 2 ) is also orthonormal2 and must have the same number of elements. (6. the basis B ( 1 .e. . . (6.s. 2 ) and B ( 1 . Our derivation below will also yield real values for the Clebsch-Gordan coeﬃcients. | 1 − 2| + 1. . . | 1 − 2 | + 1.25) i.18) applied to Y∗ and to YJ JM (J M | m1 . J = | 1 − 2 |. . − 2} 1 + 1.28) This property follows from the fact that the basis elements are eigenstates of hermitian operators with diﬀerent eigenvalues.1.146 Theory of Angular Momentum and Spin The expansion (6. .18) constitutes a change of an orthonormal basis B( 1 . (6.2: Prove Eq. 1 + 2 . Ω2 ) = δJJ δM M (6. In this case the range of |J|–values would be the interval [ |J (1) | − |J (2) | . . 1 + 2. 2 ) has (2 1 + 1)(2 2 + 1) elements. J} . 2) = {Y 1 m1 (Ω1 ) Y 2 2 m2 (Ω2 ). . to a new basis B ( 1 . . For each quantum number J there should be 2J + 1 elements YJM with M = −J. However.h.27) together with (6. m1 = − 1 . . . The basis B ( 1 . JM 2 |Ω1 ..h. respectively.24) The basis B( 1 . B2 2) should be 2 |Ω1 . 2) .24) yields 1 m1 2 m2 ) ∗ 1 m1 2 m2 ) (J M| = δJJ δM M . .25) We want to state now two summation conditions which follow from the orthonormality of the two basis sets B( 1 . 1.26) We will show below in an explicit construction of the Clebsch-Gordan coeﬃcients that. − + 1. property implies dΩ1 dΩ2 Y 1 m1 corresponding to the l. 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 . 2 |Ω1 . m2 = − 2 .

2 |Ω1 . (6. 2 ) and B ( 1 .29) where the expansion coeﬃcients are given by the respective scalar products in function space cJ M = dΩ1 dΩ2 Y∗ J M ( 1. to composite systems involving spin.18). Ω2 ) and integrating. A similar 2 construction will also be applied to composite systems governed by other symmetry groups. i. rather the convention (J M | has been assumed. therefore. the group SU(3) in case of meson multiplets involving two quarks.31) are identical to Comparision with (6. Ω2 )Y 1 m1 (Ω1 )Y 2 m2 (Ω2 ) . will be based solely on the commutation properties of the operators J and J (k) . and actually also the properties of Clebsch-Gordan coeﬃcients stated above. or J > 1 + 2 (6.32) which complements (6.. Y = J =| 1 − 2 | M =−J 1 m1 M (Ω1 )Y 1+ 2 2 m2 (Ω2 ) (J M | ∗ 1 m1 2 m2 ) YJ M J ( 1. (6.m2 M ( 1. 2 |Ω1 .2 Construction of Clebsch-Gordan Coeﬃcients We will now construct the Clebsch-Gordan coeﬃcients. i.. 1 m1 2 m2 ) = 0 if J < | 1 − 2| .e. Ω2 ) . 2 |Ω1 . The result of this construction will include all the properties previewed above. also apply the results. Ω2 ). We can. Ω2 ) .30) The latter property follows from multiplying (6.34) 6.27) yields δJJ δM M = m1 . 2 |Ω1 . i.81) stated above. (6. 2 |Ω1 .18) by Y∗ J orthogonality property (6. At this point we like to stress that the construction will be based on the theorems (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 . ..e. (6. Hence.2: Construction of Clebsch-Gordan Coeﬃcients 147 The second summation condition starts from the fact that the basis sets B( 1 . 1+ 2 J Y 1 m1 (Ω1 )Y 2 m2 (Ω2 ) = J =| 1 − 2 | M =−J cJ M YJ M ( 1.28) allows one to conclude that the coeﬃcients cJ (J M | 1 m1 2 m2 )∗ .1 states.e. or baryons multiplets involving three quarks. it is possible to expand Y 1 m1 (Ω1 )Y 2 m2 (Ω2 ) in terms of YJM ( 1 . 2 ) span the same function space.28) from (6.6. The (J M | 1 m1 2 m2 ) cJ M .. One can show readily using the same reasoning as applied in the derivation of (6.33) JM The latter summation has not been restricted explicitly to allowed J–values. e.71–5.g. (6.

have shown Y 1+ 2.37) and replace (1) (2) according to (6. 2 |Ω1 . Ω2 ) 2 |Ω1 . we can demonstrate condition (6.40) (Ω1 ) J+ Y (2) 2 2 J+ Y 1 1 (Ω1 ) Y (Ω2 ) + Y (Ω2 ) = 0 . Ω2 ).35) 1 − 2 |.37) The solution needs to be normalized. −J + 1. (6.43) . 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. 1 + 2} and for (6.41) 2) Y (Ω1 ) Y (Ω2 ) . (6. Ω2 ) (Ω1 ) dΩ2 Y 2 2 (Ω2 ) = 1. J} by applying repeatedly J− YJM +1 ( 1 . .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. Ω2 ) .39) into (6. We embark on the suggested construction for the choice J = 1 + 2 . Similarly. 2 |Ω1 . 2 |Ω1 . .148 For the construction of YJM we will need the operators J± = J1 + iJ2 . 2 |Ω1 . (6.25) using (6. | 1 − 2| + 1. 2 |Ω1 . Ω2 ) .39) which we will try for a solution of (6. M = −J. . 1+ 2 In fact.5. 1+ 2 ( 1. We ﬁrst seek an unnormalized solution GJJ and later normalize. .68) J+ = (1) + J+ (1) (2) Y 1 1 (Ω1 ) Y 2 2 2 2 (Ω2 ) 1 1 (6. The construction assumes a particular choice of J ∈ {| such J–value seeks an expansion (6. . . 2 |Ω1 . .38) for M = J − 1. Ω2 ) = Y 1 1 (Ω1 ) Y 2 2 (Ω2 ) (6. .66. . . we can also demonstrate using (??) that G dΩ1 = We. therefore. Ω2 ) = Y 1 1 (Ω1 ) Y 2 2 (Ω2 ) . 1+ 2 ( 1. J3 YJJ ( 1 .37).18) which satisﬁes J+ YJJ ( 1 .42) ( 1. Quantum mechanically this corresponds to a state G 1+ 2. Having determined such YJJ we then construct the whole family of functions XJ = {YJM ( 1 . We obtain using (5. Ω2 ) = (J + M + 1)(J − M )YJM ( 1 . 2 |Ω1 . −J. For this purpose we insert (6. (6. 2 |Ω1 .11) and (5. Ω2 ) Addition of Angular Momenta and Spin (6. 1+ 2 ( 1. 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. Ω2 ) is normalized dΩ2 G dΩ1 Y 1 1 1+ 2. 2 |Ω1 . J − 2.11) J+ by J+ + J+ . 1+ 2.36) = 0 = J YJJ ( 1 .

M = −( + ). One obtains then Y 1 1+ 2 1 + 2 1 + 2 −1 2 2 ( 1. One can obviously continue the construction outlined to determine Y 1 + 2 −2 1 + 2 −2 .45). ).47) can serve now to obtain recursively the elements of the family B + − for M = 1 + 2 − 2. We have provided in Table 1 the explicit form of YJ M ( 1 2 |Ω1 Ω2 ) for 1 = 2 and 2 = 1 to illustrate the construction. 1 + 2 − 2. all the Clebsch-Gordan coeﬃcients ( 1 + 2 M | 1 m1 2 m2 ). 2 = 1 in Table 1. G 1 + 2 −1 1 + 2 −1 in (6.s.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 ). | . ( + )}.47) (Ω2 ) − 1 1 2 + Y 2 1 1 −1 (Ω1 )Y 2 2 + Y 2 1 1 (Ω1 )Y 2 2 −1 (Ω2 ) . . Expression (6. Exercise 6. Ω2 ) 2 1+ 2 = (Ω1 )Y 2 2 −1 (6.46) 2 1 +c Y (Ω1 ) Y 2 2 (Ω2 ) = 0 . −( 1 + 2 − 1). 2 |Ω1 . of (6. 1 + 2 − 3. 1 + 2 − 1.36) we proceed as above and obtain J+ Y = (1) 1 1 −1 (Ω1 ) Y 2 2 2 2 (Ω2 ) + c Y 1 1 1 1 (Ω1 ) J+ Y (2) 2 2 −1 (Ω2 ) (6. thereby. Normalization yields furthermore Y = 1 1 + 2 −1 1 + 2 −1 We have thereby determined ( 1 2 |Ω1 Ω2 ) (6.38) to construct the family of functions B + = {Y + M ( . Show that the resulting functions are orthonormal. of (6. . The reader should familiarize himself with the entries of the Table. . . 2 |Ω1 . . . This expression is orthogonal to (6.37). Having constructed this family we have determined the Clebsch-Gordan coeﬃcients ( 1 + 2 − 1M | 1 m1 2 m2 ). . M = −( + − ). Ω2) for M = 1 + 2 − 2 and J = 1 + 2 .21) we set G 1 + 2 −1 1 + 2 −1 ( 1 2 |Ω1 Ω2 ) = Y 1 1 −1 (Ω1 )Y 2 2 (Ω2 ) + c Y 1 1 (Ω1 )Y 2 2 −1 (Ω2 ) (6.36.1: Construct following the procedure above the three functions YJM ( 1 . 6.45) for some unknown constant c. and all total angular momentum functions for a given choice of 1 and 2 .2.39) as required by (6. To satisfy this equation one needs to choose c = − 1 / 2 . .2: Construction of Clebsch-Gordan Coeﬃcients 149 We now employ property (6. etc. We demonstrate the procedure ex(1) (1) plicitly only for M = 1 + 2 − 1.37) is satisﬁed.h.38) yields pression 2( 1 + 2 )Y 1 + 2 1 + 2 −1 ( 1 . The r. ). 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. . The result is illustrated for the case 1 = 2.h. To demonstrate (6.6. We like to construct now the family of total angular momentum functions B + − = {Y + − M ( . . ( + − )}. . The l. with the symmetry pattern and with the terms Y 1 m1 Y 2 m2 contributing to each YJ M . Ω2 ). According to the condition (6. One can readily show that (6. in particular.27). We seek for this purpose ﬁrst an unnormalized solution G 1 + 2 −1 1 + 2 −1 of (6. . .44) (Ω2 ) . 2 |Ω1 .s. | .

Ω2) Y21 (2.816497 0.707107 0.316228 0. Ω2) Y30 (2. 1|Ω1 .1: Some explicit analytical and numerical values of Clebsch-Gordan coeﬃcients and their relationship to the total angular momentum wave functions and single particle angular momentum wave functions. 1|Ω1 .57735 0. Ω2) Y11 (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. 1|Ω1 .547723 0.632456 −0.730297 0. 1|Ω1 .316228 0. Ω2) Y3−1 (2. 1|Ω1 . Ω2) Y3−3 (2. Ω2) Y20 (2.707107 0.816497 Y2−2 (Ω1 )Y11 (Ω2 ) Y2−1 (Ω1 )Y10 (Ω2 ) Y20 (Ω1 )Y1−1 (Ω2 ) Y2−2 (Ω1 )Y10 (Ω2 ) − 2 3 Y2−1 (Ω1 )Y1−1 (Ω2 ) Y2−2 (Ω1 )Y1−1 (Ω2 ) 1 Table 6. Ω2) Y32 (2.57735 −0.447214 0.57735 0.258199 0.774597 − − − − 0.408248 −0.632456 0. 1|Ω1 .774597 0. Ω2) Y2−1 (2.632456 0. Ω2) Y2−2 (2. Ω2) Y1−1 (2.57735 3 10 3 5 Y2−1 (Ω1 )Y11 (Ω2 ) Y20 (Ω1 )Y10 (Ω2 ) Y21 (Ω1 )Y1−1 (Ω2 ) −0. 1|Ω1 . Ω2) Y22 (2. .816497 Y22 (Ω1 )Y1−1 (Ω2 ) 1 15 1 3 3 5 1 5 1 2 3 10 2 5 1 2 1 10 2 3 1 3 Y20 (Ω1 )Y11 (Ω2 ) Y21 (Ω1 )Y10 (Ω2 ) 0. 1|Ω1 . 1|Ω1 .547723 0. Ω2) Y31 (2. 1|Ω1 .150 Addition of Angular Momenta and Spin Y33 (2. 1|Ω1 .816497 −0.57735 0.730297 −0.258199 −0.57735 0. 1|Ω1 . Ω2) Y3−2 (2.707107 0. Ω2) Y10 (2.408248 −0.547723 0.774597 0 2 5 8 15 1 6 3 10 1 3 0.547723 0. 1|Ω1 . 1|Ω1 . 1|Ω1 .707107 0.447214 −0.

1 .10. (6. It is actually possible to state explicit expressions for any single coeﬃcient (JM | 1 m1 2 m2 ). Deﬁnition of Spinor Operators for Two Particles In contrast to Sections 5. 5. bζ . For this purpsose we will employ the spinor operators introduced in Sections 5. hence. 1 (6. aζ bζ .3 Explicit Expression for the Clebsch–Gordan Coeﬃcients We want to establish in this Section an explicit expression for the Clebsch–Gordan coeﬃcients (JM | 1 m1 2 m2 ).10 where we studied single particle angular momentum and spin.49) where ζ. b † ζ = δζζ = δζζ .10 [cf.ζ ζ. M | .2.50) (6. we are dealing now with two particles carrying angular momentum or spin. aζ . b† ζ ζ = aζ . Accordingly. (6. a† ζ . (5.54) 2 + = √ b† ( 2 +m2 )! √− ( 2 −m2 )! . b† ζ ζ = 0 = 0 . b† refer to diﬀerent particles and.51) The operators aζ .9. M | . 5. commute with each other ζ ζ aζ .6.2: Use the construction for Clebsch-Gordan coeﬃcients above to prove the following formulas J+M 1 for = J − 2 2J 1 1 1 = J.287) to two particles (1) J k k = = (2) J 1 2 1 2 ζ. m + 1 . 6.3: Explicit Expressions of CG–Coeﬃcients 151 Exercise 6. we extent deﬁnition (5. b† ζ = 0.48) (6. a† ζ ζ b† .10. These expressions will be derived now.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 . bζ = = a† . J. ζ = ± and the matrix elements < ζ |σk | ζ > are as deﬁned in Section 5. bζ = a† . a† and bζ .53) (6.ζ a† < ζ |σk | ζ > aζ ζ b† < ζ |σk | ζ > bζ ζ (6. The creation and annihilation operators are again of the boson type with commutation properties aζ . m − 2 . 2 .52) and | 2 m2 2 According to Section 5.9. − 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. + 2 − J−M +1 for = J + 1 2J+2 2 J−M for = J − 1 2J 2 = .

(2) J 2 . 2. m1 1 | 2 . k = 1. J3 . M ( 2) 2) = J(J + 1) |J. as usual are denoted by their respective quantum numbers J.62) . = M |J. (6. M ( 1 . M ( 1 . (6. we like to recall for future reference that the operators (5.54).59) J± = J1 ± i J2 .60) (6. M ( 1 . for such states should hold J2 |J. ˆ For the operators kj holds ˆ kj | j . (1) J (2) J 3 | 1 m1 1 3 | 2 m2 2 = m1 | 1 m1 = m2 | 2 m2 1 2 (6. m1 1 | 2 .300). 2) We will also require below the raising and lowering operators associated with the total angular momentum operator (6. . in analogy to Eqs. are | 1 . + 1) |J. m1 . (2) J which.18) as follows = | 1 m1 .69) − 2| ≤ J ≤ + 2 . 2) . M ( 1 . 2 )) . M ( 1 . (1) J 2 . can be expressed in terms of the number operators 1 † 1 ˆ ˆ a† a+ + a† a− . (1) J 2 |J.58) Jk + (2) Jk . 3 . 2 ) which are simultaneous eigenstates of the operators 2 2 J2 . (1) J . = = 1( 1 2( 2 2) 2) (6. 1. 2) 1. i. (6. m2 2 . The operator of the total angular momentum/spin of the two particle system is J = with Cartesian components Jk = (1) J + J (6.m2 | 1 .152 It holds. mj j and. 2 )). −J ≤ M ≤ J .59) We seek to determine states |J. 2) 2) can be expressed in terms of Clebsch-Gordan coeﬃcients (6.65) (6. m1 1 | 2 . (6. j = 1. . m2 a† + ( 1 2 = a† − ( 1 1 −m1 (6. The aim of the present Section is to determine closed expressions for the Clebsch–Gordan coeﬃcents ( 1 . M ( 1 . m2 |J.63) according to At this point. M ( 1 . (j) 2 ˆ ˆ J = kj ( kj + 1 ) . k2 = b b+ + b† b− k1 = + − − 2 2 + namely.68) The states |J. M ( 1 .53. J3 |J.61) (6. M ( 1 . M ( 1 . (1) J 2 | (2) J 2 | 1 m1 1 2 m2 2 Addition of Angular Momentum and Spin = = 1( 1 2( 2 + 1) | 1 m1 + 1) | 2 m2 1 2 . 2 . 2 . 2) + 1) |J.e. M ( 1 .. (5.66) (6.57) b† + ( (2) 2 2 +m2 1 +m1 b† − ( 2 2 −m2 + m1 )! − m1 )! (1) + m2 )! − m2 )! |Ψo . 1 .56) The states | 1 . hence. 6. 2 . M ( 1 .67) = j |J. m2 1 2 ( 1 . 5. |J.303). 2) 2) (6.55) . m1 . which describe a two particle system according to (6. M ( 1 .302. 2 . M. m2 |J. (6.64) (6. mj j = j | j . ˆ kj |J. M ( (2) J 2 |J.

65) and (6. K † 1 a† a+ − a† a− + b† b+ − b† b− .72) (6. a† b† + + − − + 2 2 − 1 1 = a† b† − a† b† = K† . K † (j) J 2 . 2 2 3 ˆ = K † kj + K † . We note that. K † J± . + − − + 2 2 = 1 † 2K .6. (6. + − − + 153 (6. K† J3 . K † (kj + 1) 2 2 1ˆ † 1 ˆ kj K + K † (kj + 1) 2 2 1 ˆ 1 ˆ K † kj + kj . K † + K † 2 2 3 † †ˆ K kj + K .71) (6. a† b† + a† a− . 6.73) can be readily proven. K † = 0. K † + kj .288). ˆ A similar calculation yields [k2 .3: Explicit Expressions of CG–Coeﬃcients The Operator K † The following operator K † = a† b† − a† b† . a† b† + − − + 2 + 2 − 1 1 − b† a† b+ . expressing (k) J 3 through the creation and annihilation operators according to (5. and applying the relationships (6.74) imply 3 2 J2 .71) one can show K† = = = = = 1 ˆ ˆ kj (kj + 1).75) The relationships (6. a† b† − a† b† + − + − + − − + 2 1 1 = a† a+ . using (6. b† − b† a† b− . 4 Using J3 = (1) J 3 + (2) J 3 . K † ] = (j) J 2 .70) will play a crucial role in the evaluation of the Clebsch-Gordan-Coeﬃcients. 2 4 = 0 = = 0.73) yields J3 . This operator obeys the following commutation relationships with the other pertinent angular momentum / spin operators ˆ kj . a† b† − a† a− . K † 2 1ˆ ˆ 1 ˆ ˆ kj kj + 1.51) one obtains ˆ k1 .50. (6. j = 1. + − + − 2 2 − + = . 6.73. 6. the relationships (6. b† = 0 .74) + 1 J− J+ + J2 . For example.71–6.71–6. Employing (6.64. due to J2 = K† 1 2 J+ J− 1 † K .73) (6. j = 1. a† b† − a† b† − + − − + 2 + 1 1 = a† a+ . K † 1 † a a+ + a† a− .

292) one can derive similarly J+ .154 Starting from (5. K † = Addition of Angular Momentum and Spin a† a− + b† b− . M ( 1 .75) ascertain that under the action of K † the states |J. 6.e. The commutation properties (6. j1 1 + .78) a state which has been used already in the construction of Clebsch-Gordan coeﬃcients in Section 6. a† b− − a† b† + + + − + = 0. 2) We want to demonstrate now that the action of K † on the states |J. M ( 1 . 2) = N |J. = − a† a− . on the state |j1 + j2 . 2) = = (j) 2 J . |j1 + j2 . M ( 1 + n . M ( 1 + .77). Application of (K † )n to this state yields. M ( 1. j1 + j2 (j1 + n n .72) and (6. 2) 2) ˆ K † kj + = K† = j 3 † K + K † j ( j + 1) |J. j2 ) = |j1 . 2 (6. j2 ) . 1 2 + 2 .2. M ( 1 .67) (j) J 2 K † |J. M ( 1 . M ( 1 . 2 ) produces again total angular momentum eigenstates to the same J and M quantum numbers of J2 and J3 . j2 2 . j1 + j2 (j1 . M ( 1 . j1 1 |j2 .77) where N is another normalization constant. 2) is a state with quantum numbers 1 1 + 1 2 and = N |J. 4 3 + j ( j + 1) |J. (6. a† b† + b† a† b− . j1 + j2 (j1 . j1 + ) K † 2 2 (6. To demonstrate that the resulting states are eigenstates of (1) J 2 and (2) J 2 we exploit (6.76) and state K† n |J. K† + K † (j) J 2 |J.63. 2 2 + n ) 2 (6. this result implies that K † |J. M ( 1 . Action of K † on the states |J. j + 2 2 2) However. 6. j2 + ) 2 2 n n = N (n. K † ] = 0 is demonstrated in an analoguous way. M ( 1 .. We consider now the case that (K † )n acts on the simplest total angular momentum / spin state.79) n |j1 + j2 . M ( 1 . One can generalize property (6. but for diﬀerent 1 and 2 quantum numbers of the operators (1) J 2 and (2) J 2 .62. namely. it holds K † |J.76) Here N is an unknown normalization constant. M ( 1 . 2 ) j + 4 1 3 + K † |J. b† + − + + + − The property [J− . 2 ) remain eigenstates of J2 and J3 with the same quantum numbers. 2 ) .73. 2 2 1 + ) . i. 6. according to (6.

Strategy for Generating the States |J. For this purpose one needs to choose in (6. holds |J. allow one to obtain the states |J.79) n. 2) 2) . J( 1 .s. 2 2 It is now important to notice that any state of the type |J. j1 + n . j2 as follows J = j1 + j2 which is equivalent to n j1 j2 Accodingly. K† 1 2 1 + 2 −J × × . J( 1 .83) The derivation of this expression will be provided further below (see page 158 ﬀ). 2 ) can be expressed through the r. 2) Our construction of the states |J.82) 1 × | (J + 2 1 × | (J + 2 1 (J + 2 1 − 1 ) . M ( 1 .74)] yields |J. M ( 1 .h. 2) . M ) (J + × K† 1 − 2 )! (J + 2 − 1 )! J+ 1 − 1 + 2 −J (J− )J−M a† + × 2 (6.84) (6. in analogy to the construction (5. The latter states. 5. M ( 1 .80) = = = 1 1 2 2 − J 1 n − = (J + 2 2 n 1 − = (J + 2 2 + 1 1 − − 2) 2) . 2 ) for −J ≤ M ≤ J as follows |J. (J + 2 − − 2) 1 1) 2 The normalization constant appearing here is actually N( + − J. 2 = j2 + n 2 (6. 1 = j1 + n 2 . M ) . J( 1 . M ( 1 .85) ∆(J. Combining (6.f.3: Explicit Expressions of CG–Coeﬃcients 155 where we denoted the associated normalization constant by N (n. j2 + n ). 1.79). 2) = N( 1 + 2 − J.6. (6. 2) = ∆(J. (6.57) and exploiting the fact that J− and K † commute [c. 6.82. (6.105) of the spherical harmonics. 2 ) .84) with (6.86) b† + J+ 2 − 1 |Ψo .81) = N( 1 + 2 1 2 − J. 2) 1 2 = ( 1 + 2 (2J + 1)! − J)! ( 1 + 2 + J + 1)! 1 2 .104.82) for |J. j1 . 2 ) ∆(J. 2 ) exploits the expression (6. of (6. 1 . (6. M ) (J− )J−M |J. J( 1 . − 1. = (J + M )! (2J)! (J − M )! 1 2 2) (6. M ( 1 .

in terms of | 1 . (6. a† ] = [b+ .s.87) + − + − i. + − + − .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 .50–6. x. b† ] = 0 [cf. One obtains then using J− = a† a+ + b† b+ (6. r!s!t! r.e.. 2 [cf. x. in analogy to (5.92) − − and noting [a+ .264).91) which.52). + + Taylor expansion of the two exponential operators yields immediately I(λ.87).90) (6. For this purpose we introduce the generating function I(λ. I(λ.88).69) yields then the Clebsch- Expansion of an Intermediate State We ﬁrst consider the expansion of the following factor appearing in (6. m1 1 | 2 .86) |Gr = Jr a† st − + s b† + t |Ψo (6. m2 Gordan-coeﬃcients. (6.88) in terms of monomials (6. (6.t (6.. y) = λr xs y t r |Gst .57)]. y) is a generating function for the states |Gr deﬁned in (6.93) = f (a† + λ a† ) g(b† + λ b† ) |Ψo .89) i. x.e.86) in terms of monomials 1 +m1 1 −m1 2 +m2 2 −m2 a† a† b† b† |Ψo . follows from the commutation properties (6. Comparision with (6. st The desired expansion of |Gr can be obtained from an alternate evaluation of I(λ. y) which is st based on the properties aζ f (a† ) |Ψo = ζ ∂ ∂a† ζ f (a† ) |Ψo ζ bζ f (b† ) |Ψo = ζ ∂ ∂b† ζ f (b† ) |Ψo ζ (6.50)] − − 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. x.

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 commutation property (6. y) = exp a† + λ a† + − exp b† + λ b† + − |Ψo . |J. 2 ) in terms of states | 1 . m1 1 | 2 .94) One can infer from this result the desired expressions for |Gr . x. 157 (6.88).94) yields I(λ.86) (J− )J−M a† + = q J+ 1 − 2 b† + J+ 2 − 1 |Ψo 1 (6. M ( 1 . y) = s. M ( 1 .q (−1)s (6.70) holds K† 1 + 2 −J = s 1 + 2 −J b† − s 1 + 2 −J−s (−1)s 1 + 2 −J−s (6. 2) = s. one can write the right factor in (6. to obtain the desired expansion of |J.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.6. x. using the deﬁnition (6.3: Explicit Expressions of CG–Coeﬃcients We conclude I(λ. Expanding the exponentials in st (6.t × a† + s−q s q b† + t r−q t−r+q a† − q b† − |Ψo (6.90) allows one to infer |Gr s.98) a† − s a† + b† + s . With K † given by (6.t × λ v b† − × r−q v × = t s−u v λu a† − r! q |Ψo λr xs y t r! s! t! r. Operation of this operator on (6.s.97) yields.97).95) Comparision with (6.99) . m2 2 .50).t = q r! × a† + s q s−q t r−q a† − q × b† + t−r+q b† − r−q |Ψo (6.96) and.

101). 1 . The Clebsch-Gordan coeﬃcents are then ﬁnally ( 1 . n)|ψ(j1 .54) between creation operator monomials and angular momentum states allow one to write this |J. 2 ) − 2) . − 1 . The summation over q corresponds then to the summation over m1 . 2 ) we consider the scalar product J. M ) (J + 2 )! (J + 2 − 1 )! s.82). 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. 1 . n = 1 + 2 −J.103) = 1.158 Addition of Angular Momentum and Spin ( 1 + 2 − J)!(J − M )!(J + 1 − 2 )!(J + 2 − 1 )! s!( 1 + 2 − J − s)!q!(J + 1 − 2 − q)!(J − M − q)!(M + 2 − a† + 2 1 −q−s 1 + q)! a† − q+s b† + M + 2 − 1 +q+s b† − 1 + 2 −M −q−s |Ψo The relationships (6. 2 ) ∆(J. J( 1 .103) this can be written 1 = N 2 ψ(j1 .104) . × 2) = N( 1 + 2 1 − J. m2 in (6.53.6. according to (6. m1 . For this purpose we introduce j1 = 1 (J + 2 1 2 − J. To determine N = N ( 1 + 2 −J. q = 1 − m1 − s and m2 = M − m1 .69) since. j2 . − q − s)!(q + s)! 2 × (M + 1 − 1 + q + s)!( 1 1 + 2 − M − q − s)! 2 − q − s 1 | 2. m2 |J.q (−1)s × (6. (6.69) if one identiﬁes m1 = 1 −q − 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 . 2 ) (6. m2 = M − 1 + q + s. (6. Using (6. J( 1 .101) Note that m1 + m2 = M holds. j2 = 1 (J + 2 2 − 1) . j2 . n) 2 )|J.82) and (6. M ( 1 . √ 2J + 1 1 2 .83) of the normalization constant N ( 1 + deﬁned through (6. M − +q+s One can conclude that this expression reproduces (6.102) We want to determine now the expression (6.

108).112) s . m1 1 |j2 .m2 n1 + m1 n1 n2 + m2 n2 λm1 +m2 (6.70) for the operator K † .110) which can be written 1 1−λ n1 +n2 +2 = r s n1 + r − s n1 n2 + s n2 λr (6. Accordingly.3: Explicit Expressions of CG–Coeﬃcients where |ψ(j1 . n) = 1 (2j1 )!(2j2 )! a† + n! (2j1 )!(2j2 )! a† + 2j1 −n−s s 2j1 s n s 2j2 a† b† + − |Ψo = n−s (−1)s a† b† − + s b† + (−1)s a† − s (2j1 + n − s)!s!(2j2 + s)!(n − s)! s!(n − s)! 2j2 +s b† + b† − n−s (2j1 + n − s)!s!(2j2 + s)!(n − s)! |Ψo .106) The orthonormality of the states occurring in the last expression allows one to write (6. j2 .109) and Taylor expansion of the left hand side of (6.108) twice.6.111) Comparision with (6. We employ the expression (6. One obtains then. n) in terms of states |j1 . applying (6. m2 2 . 1 1−λ n1 +1 1 1−λ n2 +1 = m1 .105) The ﬁrst step of our calculation is the expansion of ψ(j1 .57) for these states and the expression (6.107) s = (n!)2 s 2j1 + n − s 2j1 The latter sum can be evaluated using 1 1−λ a property which follows from ∂ν ∂λν 1 1−λ n1 +1 n1 +1 = m1 n1 + m1 n1 λm1 (6. (6. j2 . (6. j1 1 |j2 .108) yields the identity n1 + r − s n1 n2 + s n2 = n1 + n2 + r + 1 n1 + n2 + 1 .108) = λ=0 (n1 + ν)! n1 ! (6. j2 2 . we obtain |ψ(j1 . n) = K† n 159 |j1 .104) 1 = N2 (n!)2 (2j1 )!(2j2 )! (2j1 + n − s)!(2j2 + s)! s!(n − s)! 2j2 + s 2j2 (6. (6. j2 .

120) can be readily derived from the expression (6. and J (2) are vectors in R . −m1 . m2 |J. M ) .102) of the Clebsch-Gordan coeﬃcients. For the quantum mechanical addition of angular momenta the Clebsch Gordan coeﬃcients ( 1 . m1 . J (1) on the r. m1 . (6. e. of (6. M ) corresponding to (6.4 Symmetries of the Clebsch-Gordan Coeﬃcients The Clebsch-Gordan coeﬃcients obey symmetry properties which reﬂect geometrical aspects of the operator relationship (6. m1 ). namely. 2 . −m1 2|J. of (6.s. 2 .116) If one takes the negatives of the operators in (6. 6. m2 . M ) corresponding to (6.116). and (6. −m2 .h. (6. −M ) = (−1) 1 + 2 −J ( 1 . 2 .107) yields 1 = N 2 (n!)2 Theory of Angular Momentum and Spin 2j1 + 2j2 + n + 1 2j1 + 2j2 + 1 n!(2j1 + 2j2 + n + 1)! = N2 . (6.. M | 1 . m2 |J. m2 ) by ( 1 . m2 . m1 . −m2 |J. interchanging the operators J (1) and J (2) results in J = J (2) + J (1) . one can interchange also the operator J on the l. m1 |J. (6. 2 .114) as long as J. m2 |J. vice versa. m1 .h. 1 . m1 |J. ( 2 . M ) . 1 .117) are again related in a simple 2 . J (1) . −m1 . (6. M ) = (−1) 2 +m2 2J + 1 ( 2 . of this equation J (1) = J (2) − J .114) one obtains − J = − J (1) − J (2) The respective Clebsch-Gordan coeﬃcients ( 1 .114) show a simple relationship to the Clebsch Gordan coeﬃcients ( 2 . To derive relationship (6. −M ) . M ) ( 1 .160 Applying this to (6.113) Using the identities (6.116) through formula (6.119) The corresponding symmetry property of the Clebsch-Gordan coeﬃcients is ( 1 .115). 2 1+1 (6. (6.118) Finally.120) The symmetry properties (6.s.103) one obtains the desired result (6. m2 ) and. m2 |J. ( 1 .115) This relationship is a trivial consequence of (6.114) For example.g.s. m1 ) . and .11) J = J (1) + J (2) . 2 . (6. m2 |J.116) one expresses the Clebsch-Gordan coeﬃcient on the r.h. m1 . 2 .114) by. M ) = (−1) 1 + 2 −J ( 2 .83).102) by replacing ( 1 . We will demonstrate this now.118). (2j1 + 2j2 + 1)! (6. J. m1 ) by ( 2 . manner to the coeﬃcients ( 1 .

6.4: Symmetries of the Clebsch-Gordan Coeﬃcients

161

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

2 , m2 |J, M )

=

s

(−1)s s!( 1 − m1 − s)!( 2 + m2 − s)!

2

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

+ m1 + s)!

(6.121)

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

2 , m2 |J, M )

= (−1)

1 + 2 −J

S( 2 , m2 ,

1 , m1 |J, M )

.

(6.122)

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

1 , m1 |J, M )

=

s

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

1

×

(

1+

2 − J − s)!(J −

+ m2 + s)!

.

(6.123)

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

1 1

+

2

−J −s

(6.124)

+

2

− J − s ,

−s = J − −

1

−

2

+ s

(6.125)

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

1 , m1 |J, M )

1

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

2 2

+

2

−J s

+ m1 + s )! (6.126)

×

s !(

1 − m1 − s )!( 1

+ m2 − s )!

.

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

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

162

Theory of Angular Momentum and Spin

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

2 , m2 |J, M )

= (−1)

1 + 2 −J

S( 1 , −m1 ,

2 , −m1 2|J, −M )

.

(6.128)

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

2 , −m1 2|J, −M ) = s

(−1)s s!( 1 + m1 − s)!( 2 − m2 − s)!

2

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

− m1 + s)!

.

(6.129)

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

2 , −m2 |J, −M )

=

1

1 + 2 −J

s

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

1

− m2 + s )! (6.130)

×

1 s !( 2 + m2 − s )!(

− m1 − s )!

.

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

s 2 , m2 |J, M )

= (−1)

2 +m2

S( 2 , −m2 , J, M | 1 , m1 ) .

(6.131)

(−1)s s!( 2 + m2 − s)!(J + M − s)!

1

×

1 ( 2 + J − 1 − s)!( 1 − 2 − M + s)!(

− J − m2 + s)!

.

(6.132)

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

2 2

+ m2 − s

(6.133)

+ m2 − s ,

−s = −

2

− m2 + s

(6.134)

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

2 , −m1 2|J, −M )

163

=

2

2 +m2

s

(−1)−s ( 2 + m2 − s )!s !(J −

+ m1 + s )!

1

×

1 (J − 1 − m2 + s )!( 1 − m1 − s )!(

+

2

− J − s )!

.

(6.135)

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

6.5

Example: Spin–Orbital Angular Momentum States

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

2

m ,σ

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

2 2

(tot)

2

σ

(6.138)

space C χ1 1 =

2 2

1 0

,

χ1−1 =

2 2

0 1

.

(6.139)

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

1 σ = ±2

(6.140)

+ ( , m + 1 , − 1 , 1 |j, m) Y 2 2 2

r m+ 1 (ˆ) 2

164 or Yjm ( , |ˆ) = r

1 2

Addition of Angular Momentum and Spin

( , m − 1 , 1 , 1 |j, m) Y 2 2 2

r m− 1 (ˆ) 2 r m+ 1 (ˆ) 2

1 ( , m + 1 , − 1 , 2 |j, m) Y 2 2

.

(6.142)

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

1 (j − 2 , m − 1 , 1 , 1 |j, m) 2 2 2

=

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

(6.143)

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

=

(6.144)

=

(6.145)

=

(6.146)

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

2 2

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

2

1 2 1 2 1 2

, 1 |ˆ) r 2 , 1 |ˆ) r 2 , |ˆ) r

1 2

= = =

2

j(j + 1) Yjm (j

1 2

1 2

, 1 |ˆ) r 2

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

1 2

m Yjm (j

2

, 1 |ˆ) r 2

1 2

(j

2

1 2

)(j

1 2

+ 1) Yjm (j

, |ˆ) r

1 2

1 2

, 1 |ˆ) r 2

=

3 4

Yjm (j

, 1 |ˆ) r 2

(6.152)

6.5: Spin–Orbital Angular Momentum States

165

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

2 2 2

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

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

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

2

(6.156)

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

+π 2π

sin θdθ

−π 0

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

(6.157)

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

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

j = 1, 2 assumes the block-diagonal form

3 2

. . . ; m = −j, −j + 1, . . . + j }

(6.159)

0 −1 −1 0 0 −1 σ·r = ˆ −1 0

(6.160)

..

.

166

Addition of Angular Momentum and Spin

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

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

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

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

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

(tot)

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

(6.167)

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

2

1 σ · r Yjm (j ± 2 , 1 |ˆ) ˆ r 2

(tot)

, σ · r] = 0 , ˆ

k = 1, 2 3 .

(6.168)

σ · r J (tot) ˆ

2

2

1 Yjm (j ± 2 , 1 |ˆ) r 2

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

(6.169)

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

(6.170)

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

6.5: Spin–Orbital Angular Momentum States

167

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

(tot)

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

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

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

2 2 2 2 1

(6.174)

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

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

(6.175)

(6.176)

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

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

(6.178)

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

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

(6. For this purpose we consider the application of σ · r in the case of θ = 0. 1 |ˆ) + f (r) σ · p Yjm (j ± 1 . 1 |ˆ) r 2 (6. 1 |θ = 0. According to (5. therefore. φ) = − Yj 1 (j + 1 . (5. have proven (6.174–5.f. (6.180) where the sign could depend on j. Noting that p = −i r is a ﬁrst order 2 2 diﬀerential operator it holds ˆ ˆ ˆ σ · p f (r) Yjm (j ± 1 .180) and (6. ˆ (6.186) Here (( · · · )) denotes again conﬁnement of the diﬀusion operator ∂r to within the double bracket.185) We have.180). 1 |θ. 2 |ˆ) and Yj 1 (j + 1 .147. 6.188) . We want to determine ˆ its action on the wave functions f (r) Yjm (j ± 1 . identiﬁed the sign of (6. for θ = 0 (6. ˆ (6.184) 0 1 space χ 1±1 2 2 one can conclude from (6.182) σ · r Yj 1 (j − 1 .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.187) 1 2 . 6. 1 |ˆ) at θ = 0 are r r 2 2 2 2 Yj 1 (j − 1 . 1 |ˆ) = ((σ · p f (r)))Yjm (j ± 1 .168 Addition of Angular Momentum and Spin 1 Yjm (j ± 2 . 1 |θ = 0. φ) and (5. φ) . 1 |θ = 0. 1 |ˆ) r r r 2 2 2 2 2 2 (6.148) the particular ˆ 1 1 states Yj 1 (j − 2 . Since f (r) is independent of θ and φ follows ˆ ((σ · p f (r))) = −i (( ∂r f (r) )) σ · r .181. 1 |ˆ) = −Yjm (j ˆ r 2 2 ˆ The Operator σ · p ˆ The operator σ · p plays an important role in relativistic quantum mechanics. We want to demonstrate ﬁnally that the “−”-sign holds in (6. φ) 2 2 2 = − j+ 1 2 4π (6. hence.1 states χ 1 ± 1 2 2 2 [c.182) 0 Since σ · r.224)] 1 0 σ·r = ˆ .183) in case θ = 0 becomes in the standard representation with respect to the spin.160). 1 |θ = 0. 1 |ˆ).181) 0 = j+ 1 2 4π Yj 1 (j + 1 . The result can also be stated in the compact form σ · r Yjm (j ± 1 . φ) 2 2 2 . ˆ 2 2 2 2 2 2 for θ = 0. given by ˆ σ · r = σ1 sin θ cos φ + σ2 sin θ sin φ + σ3 cos θ .180) and. One can conclude then ˆ σ·r = ± ˆ 0 1 1 0 (6.

187) one obtains ˆ σ · p f (r) Yjm (j ± 1 .195) . 1 |ˆ) = 0 and σ = 2S/ .194) ˆ ˆ where J1 is deﬁned in (5.230) allows one to express ˆ ˆ ˆ ˆ ˆ σ · p σ · r = p · r + i σ · (p × r) .196) = i j+ ∂r + r 1 2 3 2 1 f (r) Yjm (j − 2 .198) . 1 |ˆ) r 2 (6. e.53).155. r 2 (6.191) (6. hence.6. r Altogether we obtain. x3 x2 ˆ ˆ ( p × r )1 h = −i (∂2 − ∂3 ) h (6. 1 |ˆ) r 2 1 2 (6.5: Spin–Orbital Angular Momentum States Using (6. 6. 1 |ˆ) r 2 (6. ˆ For the test function h(r) holds ˆ ˆ p·rh = = Using (1/r) = −r/r3 and r · ˆ −i −i · r h r = −i h r ·r − i r· h. using ∂r Yjm (j ± 1 . 1 |ˆ) = i [ ∂r + r + ] f (r) Yjm (j 2 r r Using (6.193) r r 1 1 1 = −i (∂2 x3 − ∂3 x2 ) h − i h (x3 ∂2 − x2 ∂3 ) .190) can be related to the angular momentum operator.186) for both terms in (6. we conclude the intuitively expected identity 1 ˆ ˆ p×r = − J .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 (6.197) 1 ˆ σ · p g(r) Yjm (j − 2 . ∂3 (1/r) = −x3 /r3 and ∂2 x3 = x3 ∂2 .156) this yields ﬁnally 1 ˆ σ · p f (r) Yjm (j + 2 .189) The celebrated property of the Pauli matrices (5.g. 1 |ˆ) r 2 (6. 1 |ˆ) = r 2 2 ˆ i ∂r f (r) − f (r) σ · p σ · r Yjm (j ˆ 1 2 169 . 1 |ˆ) . ∂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. Corresponding results are obtained for the other components of p × r and. r 2 2 1 J ·S 2 1 ˆ σ · p f (r) Yjm (j ± 2 . h r (6. 1 |ˆ) r 2 = i ∂r + −j r 1 g(r) Yjm (j + 2 . r r r Using ∂2 (1/r) = −x2 /r3 ..192) ˆ ˆ The operator p × r in (6. To demonstrate this we consider one of its cartision components.

45). 1 . 1 |ˆ) r 2 2 r r2 and. 1 |ˆ) = Yj− 1 j− 1 (ˆ) χ 1 1 .43). 1 |ˆ) = J 2 Yjm (j + 1 . For this purpose we use the construction method introduced in Sec.230. . 6. as well as (6. j) 1 2 1 2 1 2 1 2 = = 1 0 (6. 6. j 2 + 2j + 2 3 4 (6. 6.146). j + .2.e. according to (6. r 2 2 2 2 2 2 − r 2 ∂r r + J2 r2 f (r) Yjm (j + 1 .99) holds ˆ (σ · p)2 = − 2 2 = ∂2 J2 r + 2 . adopting the method in Sec.101). We begin with the coeﬃcients (6. j) 2 2 2 2 (j − . 5. r ∂r2 r 2 (6. 1 |ˆ) r 2 2 r r2 r2 j 2 + 2j + 3 2 2 2 4 = [ −∂r − ∂r + ] f (r) Yjm (j + 1 . using (5. We note ˆ (σ · p)2 f (r) Yjm (j + 1 . i.199) We want to show that eqs. For m = j − 1 one can state. 1 |ˆ) r 2 2 r 2 3 i i 1 2 1 r = [ i∂r + ( − j) ] [ i∂r + (j + ) ] f (r) Yjm (j + 2 .143.2.203) (6. − |j.144) and. 1 |ˆ) r 2 2 i 3 ˆ = σ · p [ i∂r + (j + ) ] f (r) Yjm (j − 1 . (6.201) 1 3 (j + )(j + ) Yjm (j + 1 . are consistent with this identity. 1 |ˆ) r 2 2 The Clebsch-Gordan coeﬃcients are then (j − 1 . In this case holds.200) = (j + 1 )(j + 3 ).197.206) .202) Yjj (j − 1 . 6..143–6. Evaluation of Relevant Clebsch-Gordan Coeﬃcients We want to determine now the Clebsch-Gordan coeﬃcients (6. r (6.151).198). 1 |ˆ) 2 r 2 r 2 j+3 ( 1 − j)(j + 3 ) 2 2 2 2 2 = [ −∂r − ∂r + − 2 ] f (r) Yjm (j + 1 . 1 |ˆ) r r 2 2 2 2 2 2 2 yields ˆ (σ · p)2 f (r) Yjm (j + 1 . 1 |j. 6. 2 2 (6. Yjj−1 (j − 1 .170 Addition of Angular Momentum and Spin To demonstrate the validity of this key result we note that according to (5.199). consider ﬁrst the case of the largest m-value m = j. in fact. according to (6. 1 |ˆ) = r 2 2 which agrees with (6.144) for m = j.205) which agrees with the expressions (6.143. j − 1 . 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 (6.204) (6.

1 |ˆ) = r 2 2 Applying J− (tot) j+m Yj− 1 m− 1 χ 1 1 + 2 2 2 2 2j j−m Yj− 1 m+ 1 χ 1 − 1 . 6. Let us verify then the expression (6.6. 2 2 2 2 2j (6.2. (6.144) for the Clebsch-Gordan coeﬃcients by induction.. 6.5: Spin–Orbital Angular Momentum States The corresponding Clebsch-Gordan coeﬃcients are then 1 (j − 1 .h.s. − 1 |j. 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 .210)] yields (j + m)(j − m + 1) Yjm−1 (j − 1 . and J− + S− to the r. 1 . j − 1 . (6. j − 3 . j − 1) 2 2 2 2 which again agrees with the expressions (6.143. 1 . (6. are obtained by iterating the application of (6. 1 |j.144) implies for j = m Yjm (j − 1 . 1 |ˆ) = r 2 2 j+m 2j + (j + m − 1)(j − m + 1) Yj− 1 m− 3 χ 1 1 2 2 2 2 j+m Yj− 1 m− 1 χ 1 − 1 2 2 2 2 2j j−m 2j (j + m)(j − m) Yj− 1 m− 1 χ 1 − 1 2 2 2 2 + or Yjm−1 (j − 1 .211) to the l.202).210) to (6. 2 2 2 2 2j . 6.144) for m = j − 1. 2 |j. 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 2 171 = = 2j − 1 2j 1 2j (6.137)] J− (tot) = J− + S− (6. 1 . (· · · |jj − 3).143.207) (6. Expression (6. m − 3 . m − 1) 2 2 2 2 = j +m−1 2j (6.206).h.208) (6.143. is obtained by applying the operator [c.209) (j − 1 .s.f. as described in Sec. [c.210). etc.f. The further Clebsch-Gordan coeﬃcients (· · · |jj − 2).

m1 . The Clebsch-Gordan coeﬃcients (6. − 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. the relationship (6. by the set j1 . 1 .143). m − 1) 2 2 2 2 = (6. 2j + 1 (6. m2 1 (6. In this context Jclass and Jclass play the same role. 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 . m3 ) = (−1) × 2 + 3 − 1 + 2 +m2 × 2 . 6. m + 1 ) = 2 2 2 2 and. j3 . −m2 | 1 .217) 2j + 2 (j + 1 . 6. m. 1+1 3 For 1 (j. We have.144) for j = m − 1. M. (1) (2) (−tot) Obviously. m1 . m. 1 |j + 1 . m2 . m1 ) (6. (1) (2) (−tot) The coeﬃcients which are the quantum mechanical equivalent to Jclass + Jclass + Jclass = 0 are the 3j–coeﬃcients introduced by Wigner. j2 m2 ) 2 .146) can be obtained from (6. all three vectors Jclass . m3 .144) by induction. m2 | 3 .172 Addition of Angular Momentum and Spin j −m+1 2j (j − 1 . 1 . m + 1 .143) by applying the symmetry relationships (6. 2 .214) 2 2 +1 ( 3 . using (6.214) yields (j. m3 . However. 2j + 2 (6.218) where we have replaced the quantum numbers J.145) from (6. one obtains (j + 1 . − 1 |j.144). − 1 |j. j2 .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 .215).120). proven (6.213) which is in agreement with (6.116. m − 1 . to reﬂect in the notation the symmetry of these quantities.215) which follows from (6. m + 1 ) = 2 2 2 j+m+1 . 1 . 6. 1 |j + 1 . 1 . The latter relationships applied together read ( 1 . They are related in a simple manner to the ClebschGordan coeﬃcients j1 j2 j3 m1 m2 m3 = (−1)j1 −j2 +m3 (2j3 + 1)− 2 (j3 − m3 |j1 m1 .216) 6. one can obtain expression (6. m1 . 2 .143. 1 . m) = 2 2 2 2 Similarly. m) 2 2 2 2 2j + 1 (6. Jclass and Jclass play equivalent roles. j+m+1 . hence.143. m + 1 .

m1 . 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. e. as follows −j1 + j2 + j3 j1 − j2 + j3 j1 + j2 − j3 j1 j2 j3 j1 − m1 j2 − m2 j3 − m3 . The reader may note that the entries of the Regge-symbol.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 . m3 .223) (Σ + 1)!n11 !n22 !n33 !n23 !n32 !)) n=0 n31 n32 n33 . m2 and j3 . discovered by Regge. = (6.102) are obtained each through the diﬀerence of two entries of the Regge-symbol. j2 . for which no known classical analogue exists. In case of a non-cyclic exchange of rows and columns the 3j–coeﬃcent assumes a prefactor (−1)Σ . According to the deﬁnition of the 3j–coeﬃcients one would expect symmetry properties with respect to exchange of j1 . The two integer entries J − 1 − m2 and J − 2 + m1 in (6. √ safe for the prefactor 2j3 + 1 which is cancelled according to the deﬁnition (6.102) of the Clebsch-Gordan coeﬃcients. The Regge symbol reﬂects a remarkable degree of symmetry of the related 3j–coeﬃcients: One can exchange rows.g.6: The 3j–Coeﬃcients We ﬁrst like to point out that conditions (6. In this way the values of 12 3j-coeﬃcients are related. one can exchange columns and one can reﬂect at the diagonal (transposition). j2 .e. m3 and with respect to sign reversals of all three values m1 . For this purpose one can use the symmetry transformations to place the smallest element into the upper left corner of the Regge symbol. except when all elements are non-negative integers and each row and column has the same integer value Σ = j1 + j2 + j3 . j3 form the sides of a triangle and the condition is symmetric in the three quantum numbers. states that j1 . Because of its high degree of symmetry the Regge symbol is very suited for numerical evaluations of the 3j–coeﬃcents. anti-cyclic exchange and of a sign reversal are stated.221) j1 j2 j3 −m1 −m2 −m3 where the the results of a cyclic. The Regge symbol also vanishes in case that two rows or columns are identical and Σ is an odd integer. 173 (6. i. m2 .21. 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. there exist even further symmetry properties. These symmetry operations relate altogether 72 3j–coeﬃcients.219) (6. with respect to alltogether 12 symmetry operations. the Regge-symbol..220) The latter condition |j1 − j2 | ≤ j3 ≤ j1 + j2 . 6. To represent the full symmetry one expresses the 3j–coeﬃcients through a 3 × 3–matrix.6. However.218) relating 3jcoeﬃcients and Clebsch-Gordan coeﬃcients. the so-called triangle condition. −j1 + j2 + j3 . are identical to the integer arguments which enter the analytical expression (6.222) m1 m2 m3 j1 + m1 j2 + m2 j3 + m3 The Regge symbol vanishes.

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 0.224) (6.227) (2m|2m1 1m2 ) = (−1)m+m1 (m + 2m2 ) δ(m.232) .10752786393409395427444450130056540562826159542886 We also illustrate the numerical values of a sequence of 3j-coeﬃcients in Figure 6.225) sn−1 . m2 ) √ 2 1 ≥ |m1 |) 2 (6.1.226) We like to state ﬁnally a few explicit analytical expressions for Clebsch-Gordan coeﬃcients which were actually obtained using (6.226) by means of a symbolic manipulation package (Mathematica) (1m|2m1 1m2 ) = (−1)1+m+m1 δ(m.223-6.174 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.229) 1 1 (0m| m1 m2 ) = 2 2 i(−1)1−m2 δ(0. m1 + m2 ) (1 − m)! (1 + m)! √ 1 1 1 1 6 2 − m1 ! 2 + m1 ! 2 − m2 ! 2 + m2 ! 1 1 ≥ |m1 | ∧ ≥ |m2 | ∧ 1 ≥ |m| 2 2 (6. m1 + m2 ) (3 − m)! (3 + m)! (3m|2m1 1m2 ) = √ 105 (2 − m1 )! (2 + m1 )! (1 − m2 )! (1 + m2 )! 1 ≥ |m2 |) ∧ 2 ≥ |m1 | ∧ 3 ≥ |m| (6. (6. (6.228) √ (−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. m) δ(−m1 . m1 + m2 ) (2 − m1 )! (2 + m1 )! √ 10 (1 − m)! (1 + m)! (1 − m2 )! (1 + m2 )! 1 ≥ |m| ∧ 1 ≥ |m2 | ∧ 2 ≥ |m1 | (6.

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

each of the blocks in (6. . 2 = 1. 2. . . M = −J. (6. (2( 1 + 2 ) + 1) ×(2( 1 + 2 ) + 1) Partitioning in smaller blocks is not possible. .234) Exercise 6. Exercise 6. .233. 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. 2 |Ω1 .233) is further block-diagonalized as follows + 2. . 1). Show that the probability of detecting a triplet substate |1m1 (1) for arbitrary polarization (m2 – 1 value) of the other triplet is 3 . . . Ω2 ).176 Addition of Angular Momentum and Spin 1 For the basis {YJM ( 1 . Exercise 6.6.6. . J = | 1 − 2 |. . . .3: Two triplet states |1m1 (1) |1m2 (2) are coupled to an overall singlet state Y00 (1..2: How many overall singlet states can be constructed from four spin– 1 states 2 | 1 m1 (1) | 1 m2 (2) | 1 m3 (3) | 1 m4 (4) ? Construct these singlet states in terms of the product wave 2 2 2 2 functions above.6.6. 1 .234) .1: Prove Eqs. .

i. . −k + 1.g. (k) (6. The operator T may also act on multi-particle states like Y 1 m1 (ˆ1 )Y 2 m2 (ˆ2 ).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.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 . that in case (i) T is an operator C∞ ( ) → C∞ ( ) acting on single particle wave functions. e. The latter can be written T |ψ where T denotes T in the rotated frame given by T = R(ϑ) T R−1 (ϑ) . 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). −k + 1.238) . . e. In fact. (k) (6. S 2 = S1 + S2 + S3 or any other polynomial of Sk . the operator (5. q = −k. Note that in the examples mentioned the operator T would be deﬁned within the same representation as R(ϑ). for example. some of the r r examples considered below involve tensor operators T of this type. (6.42) in case (i) of the position representation of single particle wave functions or the operator (5.g.6.e. The latter implies that T transforms like an angular momentum or spin state | m . ∂x2 3 1 2 2 2 2 case (ii) the operator T could be a spin operator Sk deﬁned in (5. k} such that k Tkq = q =−k (k) Dq q (ϑ) Tkq . . 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. q = −k. .236. This implies. .237) are called tensor operators of rank k. k} with the transformation property (6. Such operators T can be characterized through their behaviour under rotational transformations. 6.222) in case (ii) of single particle spin operators. that T belongs to a family of operators {Tkq . Examples of Tensor Operators The multiplicative operators C∞ ( ) → C∞ ( ) Y ( ) = def Y ( ) (6. . e. Rotations transform |ψ as |ψ = R(ϑ)|ψ and T |ψ as R(ϑ)T |ψ .237) The operators T ∈ {Tkq .g.223).7: Tensor Operators 177 6.236) In this equation Dq q (ϑ) denotes the rotation matrix Dq q (ϑ) = kq |R(ϑ)|kq .

3. 2. Exercise 6. 3 follows from the transformation properties of the spherical harmonics derived in Section 1. This choice of parametrization allows us to derive conditions which are equivalent to the property (6. but are far easier to ascertain for any particular operator.235 . but rather any rotation and any parametrization can be assumed. The fact that these operators form tensor operators of rank 1. Exercise 6. For the following it is important to note that the rotation matrix elements (6. 1.7.239) These operators can be expressed in terms of the coordinates x1 . .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.7.1: 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 0.178 are tensor operators of rank k. x2 .203). x3 .6. we will assume presently that the rotation is chosen as follows R(ϑ) = exp ( ϑ+ L+ + ϑ− L− + ϑ3 L3 ) (6.240) 1 where we have deﬁned ϑ± = 2 (ϑ1 iϑ2 ) and L± = L1 ± iL2 .237) do not require that the rotation R−1 (ϑ) is expressed in terms of Euler angles according to (5. Examples are Y00 Y1±1 Y10 Y2±2 Y2±1 Y20 1 = √ 4π Theory of Angular Momentum and Spin 3 3 r sinθ e±iφ = (x1 ± ix2 ) 8π 8π 3 3 = r cosθ = x3 4π 4π 1 15 2 2 ±2iφ 1 15 = r sin θ e = (x1 ± ix2 )2 4 2π 4 2π 15 2 15 = r sinθ e±iφ = (x1 ± ix2 ) x3 8π 8π 5 2 3 2 1 5 3 2 r2 = r cos θ − = x3 − 4π 2 2 4π 2 2 = (6. In fact. 2.237).

246.6. 0)T in case of small ϑ+ . Exercise 6. ϑ− . 0. (6.3: Derive Eqs.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 .6.7. J3 yields [J+ .235 . Tkq ] = −iTk q+1 (k + q + 1)(k − q) . Tkq ] [J3 .7: Tensor Operators 179 The conditions can be derived if we consider the property (6.80) follows kq |L+ |kq = −i (k + q + 1)(k − q)δq and. ϑ3 .4: Is the 1-particle Hamiltonian 2 H = − 2 2m + V (|r|) (6. Exercise 6. For this 2 2 purpose state ﬁrst the proper rotation operator R(ϑ) and note then that S (1) · S (2) commutes with 1 the generators of the rotation of | 2 m1 (1) | 1 m2 (2) . The property (6. . Tkq ] = = δqq and by subtracting Tkq on both sides of the k ϑ+ kq |L+ |kq Tkq . We ﬁrst consider a rotation with ϑ = (ϑ+ . (6.6.246) (6. ϑ− .7.7.1 particles for which the ﬁrst spin is described by 2 the operator S (1) and the second spin by the operator S (2) . 2 Exercise 6.7. (0. hence.235 . Tkq ] = = = Tkq+1 Tkq−1 q Tkq . 0)T .243) From (5. Tkq ] [J− .244) [L+ .235 .247) These properties often can be readily demonstrated for operators and the transformation properties (6. q =−k q+1 (6. (6. (k + q + 1)(k − q) (k + q)(k − q + 1) (6.5: Consider a system of two spin. 1 (6.248) a tensor operator? Exercise 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) .247).237) for transformations characterized by inﬁnitesimal values of ϑ+ .6.237) yields ﬁrst k R(ϑ) Tkq R −1 (ϑ) = q =−k kq |R(ϑ)|kq Tkq .245) (6.237) be assumed then. J− . 6. Similar equations can be derived for inﬁnitesimal rotations of the form ϑ = (0.242) from which follows by means of kq |1 1|kq equation ϑ+ [L+ . ϑ3 )T .6: Form tensor operators of rank 1 in terms of the three components of acting on the space of 1-particle wave functions. Expressing the results in terms of the angular momentum operators J+ . 0.

(6. 1 m1 |Φ 1 m1 (k.253) Exercise 6.237) is that their matrix elements 1 m1 γ1 |Tkq | 2 m2 γ2 obey simple relationships expressed in terms of Clebsch-Gordan coeﬃcients. These states according to (6.250) q.249) which demonstrates. Before we proceed we like to point out that the states | 2 m2 γ2 are also eigenstates of J 2 .21) of Clebsch-Gordan coeﬃcients J3 Φ 1 m1 (k. i.1: Show that Φ 1 m1 (k. 2 |γ2 ) = q.e. 2 |γ2 ) is an eigenstate of J 2 with eigenvalue + 1). i. which is characterized also by a set of other quantum numbers γ1 which are not aﬀected by the rotational transformation R(ϑ). J3 | 2 m2 γ2 = m2 γ2 | 2 m2 and the property (6. .m2 ( 1 m1 |kq m1 q.247).251) The corresponding property for Φ 1 m1 (k. hence. possibly the total angular momentum– spin state of a compositie system.m2 ( 1 m1 |kq 2 m2 ) (J3 Tkq − Tkq J3 + Tkq J3 ) | 2 m2 γ2 = qTkq = q. construct states Φ 1 m1 (k.252) ∼δm1 q+m2 = ( 1 m1 |kq 2 m2 ) Tkq Similarly.e. J3 | 2 m2 γ2 = m2 | 2 m2 γ2 . the stated property. 2 |γ2 ) = Cδ 1 1 δm1 m1 2 1( 1 (6.18) are deﬁned through Φ 1 m1 (k. J3 . i. (6. m1 .8. 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.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. The hermitian property of J3 and J 2 implies that states Φ 1 m1 (k.m2 2 m2 ) (q + m2 ) Tkq | 2 m2 γ2 | 2 m2 γ2 (6.8 Wigner-Eckart Theorem A second important property of tensor operators Tkq beside (6. J2 . 2 |γ2 ) are orthogonal to | 1 m1 in case of diﬀerent quantum numbers 1 . One can. | 1 m1 γ1 denotes an angular momentum (spin) state. J 2 | 2 m2 γ2 = 2 2( 2 + 1) | 2 m2 γ2 . in fact.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). 2 |γ2 ) can be shown readily as follows using (6. one can show that Φ 1 m1 (k. behave like |kq | 1 m1 .e. 2 |γ2 ) = ( 1 m1 |kq 2 m2 ) Tkq | 2 m2 γ2 .180 Theory of Angular Momentum and Spin 6.235 . J3 are the generators of the rotation R(ϑ). 2 |γ2 ) which correspond to total angular momentum states.

h.259) The socalled reduced matrix element 1 . Using | 1 m1 + 1γ1 = 1 ( 1 + m1 + 1)( 1 − m1 ) J+ | 1 m1 γ1 (6. the matrix elements 1 m1 γ1 |Tkq | 2 m2 γ2 can be reduced to an m1 –independent factor. γ2 (−1)k− 2 + 1 ( 1 m1 |kq 2 m2 ) (6.2: Determine the matrix elements of the gradient operator d3 rF (r) G(r) of the type (6.259) to a combination of magnetic quantum numbers m1 .260) Exercise 6. (6.253) 1 m1 γ1 |Tkq | 2 m2 γ2 1 m1 γ1 |Tkq | 2 m2 γ2 we express using the equivalent of 1 m1 ( 1 m1 |kq 2 m2 ) Φ 1 m1 (k 2 |γ2 ) (6.257) which establishes the m1 –independence of 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) .261) . say m1 + 1. γ2 k− 2 + 1 √ 1 2 m2 ) (−1) 2 1 +1 = 1 .s.255) At this point the important property can be proven that 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) is independent of m1 . q . γ1 ||Tk || 2 .8: Wigner-Eckart Theorem 181 In order to evaluate the matrix elements (6. can be evaluated as easily as possible.254) = 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) ( 1 m1 |kq 2 m2 ) . γ1 ||Tk || 2 . γ1 |Tkq | 2 m2 . In order to express the m1 – independence explicitly we adopt the following notation 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) = (−1)k− 2+ 1 √ 1 2 1 +1 1 . its m1 –dependence being expressed solely through a Clebsch-Gordan coeﬃcient. γ2 = 2 1 +1 1 m1 . m1 = q = m2 = 0. 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 ) = (6.8. One can then evaluate also the corresponding Clebsch-Gordan coeﬃcient ( 1 m1 |kq 2 m2 ) and determine 1 .256) and noting that the operator adjoint to J+ is J− . γ2 is determined by applying (6. γ2 ( 1 m1 |kq (6. for which the l. γ2 .32) Tkq | 2 m2 γ2 = and orthogonality property (6. (6. γ1 |Tkq | 2 m2 .6. γ1 ||Tk || 2 . γ1 ||Tk || 2 . To prove this property we consider 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) for a diﬀerent m1 value. m2 .e. i.g. e.258) We can then ﬁnally express the matrix elements of the tensor operators Tkq as follows 1 m1 .

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

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

θ for the distance from the origin and for the angular position.15) (7.m (r) . J3 ψE.6) ˙ Since Jk is also equal to the poisson bracket {H.e. k = 1. 3. 2 . ˆ A classical particle moving in a potential V (r) is governed by the Newtonian equation of motion (7.f.m (r) m ψE. stationary states ψE.85– 5. Jk } where H is the Hamiltonian.5) and r = v. d ˙ J = v × mv + r × mv = 0 + r × er (−∂r V (r)) = 0 .1 Radial Schr¨dinger Equation o ˙ mv = − er ∂r V (r) . k = 1.13) . Jk ] = 0 . one can conclude ˙ Jk = {H. 2. i.5) In the case of an angular independent potential angular momentum J = m r × v is a constant of ˙ motion. (7. The expression of the kinetic energy 1 1 1 J2 p2 ˙ = m r2 + m r2 θ2 = m r2 + ˙ ˙ 2m 2 2 2 2mr2 and conservation of energy yield J2 1 m r2 + ˙ + V (r) = E .m (r) . .7) ˆ The correspondence principle dictates then for the quantum mechanical Hamiltonian operator H and angular momentum operators Jk ˆ [H.184 Spherically Symmetric Potentials 7.10) (7. Employing in this plane the coordinates r. In fact.99)] 2 1 2 J2 2 2 = − ∂r r + . the time variation of J can be written. 2 .14) (7.m (r) . (5. (7. Jk } = 0 .9) − 2m 2m r 2mr2 the expressions (5. Accordingly.m (r) = = = E 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. . J ψE. ˆ H ψE.m (r) (7.m (r) can be chosen as simultaneous eigenstates of the Hamiltonian ˆ operator H as well as of J 2 and J3 . one can state ˙ J = m r2 θ . 3 .11) (7. 2.12) .8) This property can also be proven readily employing the expression of the kinetic energy operator [c.61) of J 2 and Jk . ˆ dt (7. 2 2mr2 (7.87) for Jk as well as the commutation property (5. For this purpose one concludes ﬁrst that the conservation of angular momentum J implies a motion of the particle conﬁned to a plane.. using (7. ( + 1) ψE.

16) m 2mr2 Once. It can be solved by integration of 1 2 2 J2 r = ± ˙ E − V (r) − .m .20) Since this equation is independent of the quantum number m we drop the index m on the radial wave function vE. . . (r) = 0 .10) − 1 2 ∂ r + 2m r r 2 2 ( + 1) + V (r) − E 2mr2 .11). .4. . φ) are the angular momentum eigenstates deﬁned in Section 5.m (r) 1 2 J2 ∂r r + + V (r) − E 2m r 2mr2 the functional form ψE.1: Radial Schr¨dinger Equation o 185 This is a diﬀerential equation which governs solely the radial coordinate.11.7.19) where Y m (θ. (7.m (r) .m (r) 2 . by integration of ˙ θ = J .m ψE. mr2 (t) (7.m (r) Y m (θ. .m (r) = 0. ( + 1) 2mr2 .12) are obeyed and one obtains for (7. (7. ∞[ governed by the eﬀective potential (7.20) in the form of the one-dimensional Schr¨dinger equation o 2 − where 2m 2 ∂r + Veﬀ (r) − E φE (r) = 0 . (7.22) which is the original potential 2 ( +1) V (r) with an added rotational barrier potential 2mr2 . φ) (7.24) .20) by −2mr/ 2 yields the so-called radial Schr¨dinger equation o 2 ∂r − ( + 1) − U (r) − κ2 r2 r vκ.23) This demonstrates that the function r vE. r(t) is determined the angular motion follows from (7. Employing the kinetic energy operator in the form (7. can exclude particles from the space with small r values.13). (7. i.m vE. but can also trap particles in the latter space giving rise to strong scattering resonances (see Section ??). .18) = vE.m (r) = 0.m (r) describes the radial motion as a one-dimensional motion in the interval [0. − Adopting for ψE. 7. in the present case.17) In analogy to the classical description one can derivem. One can write (7.. for the wave function of a quantum mechanical particle a diﬀerential equation which governs solely the r-dependence.10). . This barrier. (7. Multiplying (7. equations (7. 2 (7.e. using (7.m (r) and E . together with the original potential.22) (7.9) one can write the stationary Schr¨dinger o equation (7.21) Veﬀ (r) φE (r) = = V (r) + r vE.

28) (7. 4π|r| (7. 1975).186 where we deﬁned U (r) κ2 = = − − 2m 2 Spherically Symmetric Potentials V (r) E (7. hence. . assumes the general form r vκ. For = 0 the contribution of Br−( function is.D. (r) ∼ A r + B r− −1 +1 + B r− (7. (r) ∼ A r or vκ. Pascual (Springer. for > 0 is not integrable. New York. according to a well-known result in Classical Electromagnetism3 . r → 0 (7.33) A detailed discussion of the proper boundary conditions.31) to the complete wave which. We replaced in (7.25) (7. the term Br−( +1) would contribute |B|2 0 dr r−2 +1) (7. ψE.32) The total kinetic energy resulting from this contribution is. Boundary Conditions In order to solve (7. 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). The radial part of this integral is ∞ 0 2 dr r2 vκ. This follows for > 0 from consideration of the integral which measures the total particle density. ”Classical Electrodynamics.26) 2m 2 In case E < 0. at r = 0 is found in the excellent monographs Quantum Mechanics I.30) and.” by J.182) for Y00 . Berlin. Jackson (John Wiley. For r → 0 one may assume that the term ( + 1)/r2 becomes larger than the potential U (r). 2nd Ed. Galindo and P. In this case the solution is governed my ( + 1) 2 r vκ. using the expression (5. (r) = 0 .20) one needs to specify proper boundary conditions2 . accordingly.29) Only the ﬁrst term is admissable. . II by A. (7. κ assumes real values. . in particular. for example.m (r) ∼ √ B .m (r) ∼ √ 4π B δ(x1 )δ(x2 )δ(x3 ) .20) the index E by the equivalent index κ. see.27) ∂r − r2 and. (r) (7. 2 2 2m 2 ψE.

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).

s. (7. in case of a free particle moving along the x3 -axis one expands eik3 x3 = .178) are given in terms of Legendre polynomials P (cos θ) one can replace the expansion in (7. One can show d −1 2 (x − 1) ∼ (x2 − 1) × polynomial in x dx −1 (7. One can expand the former solution in terms of solutions (7. For example. hence. φ) . according to (5.s. The orthogonality properties (5. (7. In this case the expansion on the r. can be written exp(ikr cos θ). of (7. φ).h.194 Generating Function of Spherical Bessel Functions Spherically Symmetric Potentials The stationary Schr¨dinger equation of free particles (7. −1 .. the wave function does not depend on φ.98) The l.e. φ) for non-vanishing m has a non-trivial φ-dependence as described by (5. Since the spherical harmonics Y 0 (θ. z = kr. This holds for −1 and one can conclude +1 consecutive integrations by part d izx e dx (7. one given o by (7. the surface term ∼ [· · ·]+1 vanishes.100) Deﬁning x = cos θ.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 (7.99) +1 d cos θ eikr cos θ P (cos θ) = b j (kr) −1 2 .50).99) We want to determine the expansion coeﬃcients b . and using the Rodrigues formula for Legendre polynomials (5.106).44. 7.45) and one given by (7.50). namely.43) has two solutions.98) does not involve any non-vanishing m-values since Y m (θ.179) yield from (7. 2 +1 (7. (7.150) one obtains +1 +1 1 ∂ dx eizx P (x) = dx eizx (x2 − 1) .98) by ∞ eik3 x3 = =0 b j (kr) P (cos θ) . i.102) −1 d −1 +1 −1 dx −1 (x2 − 1) .104) dx eizx P (x) −1 = (−1) 2 ! (iz) = 2 ! +1 dx (x2 − 1) −1 +1 dx (1 − x2 ) eizx .m a mj (kr) Y m (θ.103) and.h.

110) One refers to the l.105) must hold for all powers of z.f.105) This expression allows one to determine the expansion coeﬃcients b . for the leading power x [c. 1 (7.h.109) eikr cos θ = =0 i (2 + 1) j (kr) P (cos θ) .100) gives b j (kr) 2 (iz) = 2 +1 2 ! +1 195 dx (1 − x2 ) eizx . (7.s.113) dt (1 − t2 )ν− 2 eizt .h.96) one can express this.117) one can write the r. z i or i (2 + 1) z (2 )! 2 2 2 +1 2 ! [1 · 3 · 5 · · · (2 − 1)] 1 (7.108) where the last factor is equal to unity. Integral Representation of Bessel Functions Combining (7. ∞ (7. Comparision with the l. 1 (7. The identity (7.h.111) Employing (7. −1 (7.107) 1 2 (2 )! 1 · 3 · 5 · · · (2 + 1) 2 + 1 1 · 2 · 3 · 4 · · · (2 − 1) · 2 (7. 2 +1 −1 dx (1 − x2 )ν− 2 eizx .7. using ν = Jν (z) = √ 1 πΓ(ν + 1 ) 2 z 2 ν + 1.106) yields ﬁnally b = i (2 + 1) or. after insertion into (7. in particular.69)] b z 2 (iz) = 1 · 3 · 5 · · · (2 + 1) 2 + 1 2 ! +1 dx (1 − x2 ) .106) Employing (5. −1 (7. (7.112) We want to consider the expression Gν (z) = aν z ν fν (z) where we deﬁne fν (z) = C (7.105) and (7. as the generating function of the spherical Bessel functions.99).2: Free Particle Comparision with (7. −1 (7.114) . of (7.109) results in the integral representation of j (z) j (z) = (z) 2 +1 ! +1 dx (1 − x2 ) eizx .s.s.

it holds for the derivatives of Gν (z) Gν (z) G (z) = = ν aν z ν fν (z) + aν z ν fν (z) z 2ν ν(ν − 1) aν z ν fν (z) + aν z ν fν (z) + aν z ν fν (z) 2 z z (7. (7. We can now demonstrate that Gν (z) deﬁned in (7. Gν (z).196 Spherically Symmetric Potentials Here C is an integration path in the complex plane with endpoints t1 . C 1 (7.116) fν (z) = − or fν (z) = − From this we can conclude fν (z) + 2ν + 1 fν (z) + fν (z) = 0 . 4.120) z 2ν + 1 dt (1 − t2 )ν− 2 eizt + C C 1 1 t2 t1 = 0 (7. We consider now the functions u(j) (z) = √ 1 πΓ(ν + 1 ) 2 z 2 ν Cj dx (1 − x2 )ν− 2 eizx .121). 2. obeys Bessel’s equation (7.114. To prove this we note fν (z) = i C dt (1 − t2 )ν− 2 t eizt .2.4: Prove (7.117).118) z 2ν + 1 fν (z) + fν (z) .122) (7.123) Insertion of these identities into Bessel’s equation leads to a diﬀerential equation for fν (z) which is identical to (7.121) (7.124) .113) satisﬁes (7. 1 j = 1. In fact.93). (7.115) 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 . 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. 3.119) We note that equations (7.120) such that we can conclude that Gν (z) for proper integration paths is a solution of (7. 1 (7.120) imply also the property (2ν + 1) fν (z) + z fν+1 (z) = 0 .117) dt (1 − t2 )ν− 2 t2 eizt 1 (7. Exercise 7.93) as long as the integration path in (7. z (7.113) obeys the Bessel equation (7. t2 of the integration paths C. t2 . for properly chosen endpoints t1 .

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 . using t = 1 + is. we introduce the so-called Hankel functions (1) Hν (z) = −2 u(2) (z) .135) (7.125) 0 ≤ s < ∞ (7. u(3) (z) ≡ 0.112) shows u(1) (z) = Jν (z).93).132) For Hν (z) one derives.129) The integrand in (7.128) C = C1 ∪ C2 ∪ C3 ∪ C4 is a closed path. Accordingly.130) holds Jν (z) = (1) 1 2 1 (1) (2) Hν (z) + Hν (z) . (7.135) are known as the Poisson integrals of the Bessel functions.124) vanishes along the whole path C3 and. one can state Jν (z) = − u(2) (z) + u(4) (z) . (7. the integral expression (1) Hν (z) 1 1 1 1 (7. dt = i ds.7.131) According to (7. one can derive (2) Hν (z) = e −i(z−πν/2−π/4) ∞ 0 √ ds [s(1 − is/2)]ν− 2 e−zs .126) −1 ≤ s ≤ 1 (7. 1 (7.127) 0 ≤ s < ∞ (7. for Rez > 0 and ν ∈ R obey (7. 1 (7. . therefore. Comparision with (7.117) and. u(2) (z) and u(2) (z) are both solutions of Bessel’s equation (7. (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. Since the integrand in (7. hence. Following convention.124) is analytical in the part of the complex plane surrounded by the path C we conclude 4 u(j) (z) ≡ 0 j=1 (7.134) Similarly.134) and (7. (2) Hν (z) = −2 u(4) (z) .130) We note that the endpoints of the integration paths C2 and C4 .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 .

h(z) to be linearly independent.135) which read for ν = + 1 2 H where f The binomial formula yields f (1) 2 (1) 2 1 +2 (1. The general solution of (7.e.140) Bessel Functions with Negative Index Since ν enters the Bessel equation (7..141) (1) (1) the derivation of which follows the derivation on page 192 for the Wronskian of the radial Schr¨dinger equation. (7. Hν (z) as well as H−ν (z) are solutions of this equation. We employ for this purpose the Poisson integrals (7. For the Wronskian connected with the Bessel equation (7.134) and (7. z (7. we obtain H (1) 2 1 +2 (1) 2 (z) = r=0 ( + r)! ! r!( − r)! ± i 2 r 1 z +r+1 (7.93) only as ν 2 . πz (7. As a second order diﬀerential equation the Bessel equation has two linearly independent solutions.143) .140).138) The formula for the Laplace transform of sn leads to f and. h) must be a non-vanishing function.141) is o W (z) = − In case of g(z) = H (1) (z) +1 2 c . One obtains W (z) = − 4i .93) holds the identity 1 W = − W z (7.198 Asymptotic Behaviour of Bessel Functions We want to obtain now expansions of the Hankel functions H Spherically Symmetric Potentials the expansions converge fast asymptotically. the Wronskian W (g. i.137) (z) = r=0 r ± i 2 r 0 ∞ ds s +r e−zs .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. For such solutions g(z).139) (z) = 2 ±i[z − ( e πz +1) π ] 2 r=0 ( + r)! r!( − r)! i ± 2z r (7.136) (1) 2 (z) = 0 ds [s(s ± is/2)] e−zs . (7.142) and h(z) = H (2) (z) +1 2 one can identify the constant c by using the leading terms in the expansions (7. converge fast for |z| → ∞. hence.

140). linearly independent. (7.144) The expansion coeﬃcients A. 7.. We conclude 2 H−( Simarly. one can show H−( Spherical Hankel Functions In analogy to equations (7.97). 7.2) (2) +1) 2 (1) +1) 2 (z) = i (−1) H (1) (z) +1 2 .147) (z) = π (1.132) follows n (z) = 1 2 π (−1) 2z +1 H−( (1) 1 +2) (z) + H−( (2) +1) 2 (z) .149) We want to establish also the relationship between h (z) and the irregular spherical Bessel function n (z) deﬁned in (7.151) are equivalent to h h (1) (2) (z) (z) = = j (z) + i n (z) j (z) − i n (z) .149. B can be obtained from the asymptotic expansion (7.2) (7.2) (7.97.153) .e. 2z 2 (1.2: Free Particle i.145) This equation can hold only for B = 0 and A = exp[i( + 1 )π].96.2) H + 1 (z) . (7.7. (7.151) Equations (7.152) (7.146) (z) = −i (−1) H (2) 1 (z) +2 . 7. (7. For |z| → ∞ the leading terms yield π π π π π 1 1 1 √ ei(z+ 2 ) = √ A ei(z− 2 − 2 ) + √ B e−i(z− 2 − 2 ) z z z |z| → ∞ .147) this can be written n (z) = 1 2i h (1) (z) − h (2) (z) .97) one deﬁnes the spherical Hankel functions h (1. 7. 7. in fact.148) Following the arguments provided above (see page 193) the functions h (z) are solutions of the radial Schr¨dinger equation of free particles (7.146.53). 7.148) holds for the o regular spherical Bessel function j (z) = 1 (1) (2) [ h (z) + h (z) ] . (7. (7. H (1.150) According to (7. H−ν (z) = A H (1) (1) (z) +1 2 One can then expand + BH (2) (z) +1 2 . 2 (1.2) (z) +1 2 199 are. (7.132. From (7.90. According to (7.

as given by (7.155) (z) and To determine j (z) and n (z) we note that for z ∈ R the spherical Hankel functions h (2) h (z).164) .163) (7. Hence. 2z 2 z sin(z − π/2) z cos(z − π/2) − . = sin(z − π/2) and sin[z − ( + 1) π ] .156) ( + r)! r!( − r)! ( + r)! r!( − r)! i 2z r [ /2] = r=0 ( + 2r)! (2r)!( − 2r)! −1 4z 2 r (7. at large |z|. 2 2z z (7. (1) (7. 7.160) Employing cos[z − ( + 1) π ] .140.140) and (7.161) (7. j (z) (z) = ( i) z +1 e±iz r=0 ( + r)! r!( − r)! ± i 2z r (7. (7. it follows z ∈ R : Using p (z) = Re r=0 j (z) = Re[h (1) (z)] .154) the identities j (z) n (z) = = sin[z − ( + 1) π ] cos[z − ( + 1) π ] 2 2 p (z) − q (z) z 2z 2 cos[z − ( + 1) π ] cos(z − π/2) 2 q (z) + p (z) .2) (z). are complex conjugates.162) The leading terms in these expansions.148). at large |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) .148) one obtains readily h (1) 2 (1.154) The leading term in this expansion. From (7.159) (7. z (7. is h (1) 2 (z) = ( i) z +1 e±iz .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). 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 (7.158) one can derive then from (7. are j (z) n (z) = = (7.

174) We want to prove these relationships.174) and from (7. deﬁning a = −1 and employing fν (z) as deﬁned in (7.169) (7.7.131.171.168) (7. 7.151) follows then that the relatiosnhips hold also for g (z) = j (z) and g (z) = n (z).172) where g (z) is either of the functions h obtain the recursion relationship g g (z). 7. One can combine (7.172) to +1 (z) +1 (z) = A (z) 1− g (z) g (z) −1 (7.162) allow one to provide explicit expressions for j (z) and n (z). (7. 7.167) (7. 7.171–7.2) g (z) − g (z) z 2 +1 g (z) − g z (7. j (z) and n (z). 1. One obtains for = 0.2) (z) = −2 π 1 √ 2z πΓ(ν + 1 ) 2 z 2 ν C2. (1.176) .171) −1 (z) (7.2) relationships hold for g (z) = h (z).149. 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. 7.2: Free Particle Expressions for the Spherical bessel Functions j (z) and n (z) 201 The identities (7. 1 (7.4 dx (1 − x2 )ν− 2 eizx .124.148) and express g (z) = h (1. From the linearity of the relationships (7.173) A (z) = z ( +1) z2 +2 z (7.114) we can write g (z) = a z f 2 ! + 1 (z) 2 .170) Recursion Formulas of Spherical Bessel Functions The spherical Bessel functions obey the recursion relationships g g +1 (z) +1 (z) = = (1.175) Using Γ( + 1) = !.2) To demonstrate that (7. For this purpose we need to demonstrate only that the (1.165) (7.161.171) holds for g (z) = h (z) we employ (7.166) (7.

Exercise 7.165).179) g (z) = − z2 g (z) + g (z) − g +1 (z) z . f yields.171). and (b) n1 (z). in fact. is equivalent to (7. (7.166). i. (7. In order to prove (7.171).173). j2 (z) from j0 (z).179) from which follows (7.173.182) Replacing all ﬁrst derivatives employing (7. The ﬁrst component of (7. z2 (7.6: Employ the recursion relationship (7. together with (7.172). n2 (z) from n0 (z). To prove (7.176) g (z) = z g (z) − g +1 (z) +1 2 Spherically Symmetric Potentials z g (z) + a z f 2 ! +1 2 (z) .121). 7. j0 (z) using (7.173) is equivalent to (7.172) we diﬀerentiate (7.182).173.. z (7.171).2.180) yields g +1 (z) = g (z) − ( + 2) g (z) + z2 +2 g (z) .172).177) (z) = − z f 2 +2 + 3 (z) 2 .5: Provide a detailed derivation of (7.178) (7.181) Using this identity to replace the second derivative in (7. The second component of (7.180) Since g (z) is a solution of the radial Schr¨dinger equation (7.2. . Exercise 7. 7.174) we start from (7. n0 (z) using (7.e.58) it holds o 2 g (z) = − g (z) + z ( + 1) g (z) − g (z) .202 The derivative of this expression is g (z) = Employing (7.179) leads to (7.174) to determine (a) j1 (z). (7.

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

t) = 4 π J(r. t) = − V (r. In the non-relativistic limit holds p ≈ mr and (8. t) − ∂t χ(r. t) − ∂t A(r. t) which is solved by E(r. t) + ∂t A(r.9) (8. obeys the equation of motion d p = q dt E[ro (t). t) through the scalar and vector potentials V (r. t) in (8.2) In addition. t)] and. The following substitutions. t) + χ(r. (8.1) the term qδ(r − ro (t)) and to the current density ˙ J(r. solves implicitly (8. t) and A(r. Equation (8. t] + v × B[ro (t). The particle. t) . called gauge transformations. We assume so-called Gaussian units. called the scalar potential. t) + ∂t B(r.7) Gauge Transformations We have expressed now the electric and magnetic ﬁelds E(r. t) in (8. t) (8.1) (8. t) . t) + ∂t A(r.10) (8.5) where p is the momentum of the particle and ro (t) it’s position at time t. t). t) where V (r. t) = − V (r. the relationship between ﬁelds and potentials is not unique.3) (8. called the vector potential. (8. 1975) for a discussion of these and other conventional units. From this follows E(r. The reader is referred to the well-known textbook ”Classical Electrodynamics”. t) (8. t) = 0 = 0. Jackson (John Wiley & Sons. the two homogeneous Maxwell equations hold × E(r. but leave the ﬁelds unaltered: A(r. t) × B(r. accordingly. New York. t) · B(r.25).2) the term q r˙o δ(r − ro (t)). As is well known. by J.11) V (r.6) for some vector-valued function A(r.8) = 0 (8. t). t] (8. 2nd Edition. in turn. t) + v × B(r. t) −→ V (r. (8. t) is a scalar function.204 Maxwell equations1 · E(r. t) − ∂t E(r. . t) = × A(r. contributes to the charge density ρ(r. t) and B(r.5) above agrees with the equation of motion as given in (1. alter the potentials.4).4) Lorentz Force A classical particle with charge q moving in the electromagnetic ﬁeld experiences the so-called Lorentz force q[E(r. t) Interaction of Radiation with Matter = 4 π ρ(r.3) reads then × E(r. t) . Scalar and Vector Potential Setting B(r. D. t) −→ 1 A(r.

t) . t) can be obtained employing (8. equation (8. t = 4 π J(r. t) = 0 . such that the latter describes the electromagnetic radiation. t) and the current due to moving net charges. t) − ∂t V (r. t) |r − r | (8. The vector potential A(r. t) = 1 ∂t V (r. t) .19) couples the vector potential A(r. t) Using · ∂t A(r. this potential results from inserting (8. t) in (8. (8.2).17) A(r. 4π (8. be achieved.19) and deﬁne J (r. t) . t). t) = 0 the solution is given by the Coulomb integral V (r. t) .1) · − V (r. a name which is due to the form of the resulting scalar potential V (r.6) and (8. t) − 2 × A(r. (8. t) and V (r. t). t) − ∂t A(r.20) . In fact.12) yields then the Poisson equation 2 V (r.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. in fact. t) + ∂t A(r.19) Unfortunately. t) + ∂t V (r.18) together with condition (8. and a component due to the remaining currents. t) . t) = Ω∞ for r ∈ ∂Ω∞ (8. t) = ∂t = 4 π ρ(r. the second inhomogeneous Maxwell equation.8. t) = · A(r. For this purpose we examine the oﬀending term ∂t V (r.9) for the ﬁelds results in × The identity × × A(r. t) = − 4 π J(r. (8. Using the expressions (8.12) leads us to 2 2 A(r.12) The corresponding gauge is referred to as the Coulomb gauge.13) · A(r.14) In case of the boundary condition V (r.9) into (8. Such description can. (8.15) d3 r ρ(r . t) (8. (8. One would prefer a description in which the Coulomb potential (8. t) − ∂t A(r.16) and the vector potential are uncoupled. t) together with (8. the gauge freedom allows us to impose on the vector potential A(r. t) the condition · A(r. t) = − 4 π ρ(r.16) This is the potential commonly employed in quantum mechanical calculations for the description of Coulomb interactions between charged particles. In fact. and the former the Coulomb interactions in the unperturbed bound particle system.

29) · Et (r.12) and the Coulomb potential (8. t) − ∂t A(r. (8.14).16) yield ﬁnally the electric and magnetic ﬁelds. the name longitudinal electric ﬁeld. (8. hence. Let J(r. t) = 0). t) = − ∂t ρ(r.22) 1 ∂t 4π · J (r. t) = 0 (8. no divergence-free currents. in a ring-shaped antenna which exhibits no net charge.26) and (8.19) yield 2 2 A(r. t) Bt (r. using ∂t · J (r. yet a current.. t) + ∂t ρ(r. For J also holds the continuity equation · J(r. For the divergence of J (r. t). t) = or = 2 Interaction of Radiation with Matter (8.24) (8. V (r.25) one refers to J and Jt as the longitudinal and the transverse currents.25) Because of properties (8.26) turns into the well-known wave equation 2 2 A(r. = − ∂t A(r.23) This continuity equation identiﬁes J (r. V (r. exist in the space considered. the name transverse ﬁelds. 8.28) (8. i. t) = × A(r.. t) (8.27) which is obviously curl-free ( × E (r. The deﬁnitions of J and Jt applied to (8.206 For the curl of J holds × J (r. t) holds. t) These ﬁelds are obviously divergence -free (e. t) = 0).30) . (8. A(r. The vector potential determined through (8. t) = 0 . respectively. t) + ∂t ρ(r.21) and (8. (8. we want to assume that no ring-type currents. t) contributes solely an electric ﬁeld component E (r. t) contributes an electrical ﬁeld component as well as the total magnetic ﬁeld Et (r. t) (8. for example. t) = − 4 π Jt (r. such current densities exist.21) ∂t and the Poisson equation (8.2 Planar Electromagnetic Waves The current density Jt describes ring-type currents in the space under consideration. t) . hence. t) = − V (r. In this case (8. t) = 0 . Presently. t) − ∂t A(r. t) be the total current of the system under investigation and let Jt = J − J . t) as the current due to the time-dependence of the charge distribution ρ(r. t) .g.e.26) This equation does not couple anymore scalar and vector potentials. t) = 0 . t) = 0 and from this follows · Jt (r.

Note that in the units chosen the velocity of light is c = 1. 8. t) . there exist two linearly ˆ u independent orientations for u corresponding to two independent planes of polarization. is orthogonal to k.29). (8. 2 .2: Planar Electromagnetic Waves 207 which describes electromagnetic ﬁelds in vacuum. This ﬂux is given by 1 S(r. t) × Re B(r. hence.34) (8.37) ˆ ˆ where k is the unit vector k = k/|k|.35) one obtains S(r.158). t) = ±i ω A(r. Et (r.32) holds. Obviously.34. are orthogonal to each other and are both orthogonal to the wave vector k.35). (8. t) = Ao u exp i(k · r ˆ where the dispersion relationship |k| = ω (8. obviously.35) The vector potential in (8. 8. at each point r and moment t. ω2 = |Ao |2 (8.38) ˆ In this expression for the energy ﬂux one can interprete k as the propagation velocity (note c = 1) and.31. ˆ We want to characterize now the radiation ﬁeld connected with the plane wave solutions (8.32.39) 8π The deﬁnition incoming waves and outgoing waves is rationalized below in the discussion following Eq.38. In applying the potential and ﬁelds to physical observables and processes we will only employ the real parts. 8.34. The corresponding electric and magnetic ﬁelds. (8. The latter vector describes the direction of propagation of the energy ﬂux connected with the plane wave electromagnetic radiation. Time average over one period 2π/ω yields S(r.33) ωt) (8.28.36) 4π Using the identity a × (b × c) = b (a · c) − c (a · b) and (8. A complete set of solutions is given by the so-called plane waves A(r. t) = ± ω2 ˆ |Ao |2 k sin2 ( k · r − ωt ) 4π (8. t) Bt (r. are Et (r. The Coulomb gauge condition (8. t) = ± ω2 ˆ |Ao |2 k .8. t) .31) and the resulting ﬁelds (8. 8.39). Here the “-” sign corresponds to so-called incoming waves and the “+” sign to outgoing waves2 . t) in (8.31) u is a unit vector (|ˆ| = 1) which. t) and Bt (r. 8π (8. t) = i k × A(r.34.31). 8. the constant k is referred to as the wave vector.12) yields u·k = 0. see also the comment below Eqs. (8. t) = Re Et (r. 8. ˆ (8. 8. according to (8. accordingly.35) are complex-valued quantities.

41) ωV Inserting this into (8.8. t) ˆ replacing the classical momentum p by the diﬀerential operator p = of the particle is then described by the Schr¨dinger equation o ˆ i ∂t Ψ(r. (8. t) 2m 2 + qV (r) . deﬁned below Eqs. We are left with a classical Hamiltonian function which has an obvious quantum mechanical analogue ˆ H = ˆ p − q A(r.45) . The sign in (8. The integrals express the integration over the energy density of the ﬁelds.42) 8. |k| = ω .28.43) Here the ﬁelds are deﬁned through Eqs. t) = H Ψ(r.38) implies that for incoming waves. Comparision of (8. (8.31). is p − q A(r. We assume that the energy content of the ﬁelds is not altered signiﬁcantly in the processes described and. A correct description of the electromagnetic ﬁeld requires that the ﬁeld be quantized.44) . 8. 8.208 Interaction of Radiation with Matter as the energy density.39) and (8.31) allows one ﬁnally to state for the planar wave vector potential A(r. t) are complex leading to the diﬀerence of a factor 2 in the energy densities of the lontitudinal and transverse components of the ﬁelds. t) = 8πNω u exp i(k · r − ωt) ˆ ωV . The wave function Ψ(r. respectively. If we consider a volume V with a number of photons Nω the energy density is obviously Nω ω .16. u · k = 0 . (8. t).27. 8.40) = V It should be pointed out that Nω represents the number of photons for a speciﬁc frequency ω. Bt (r. t) 2m + 1 8π Ω∞ 2 H = + qV (r) 1 16π d3 r |Et |2 + |Bt |2 Ω∞ d3 r E 2 + .39) through the density of photons connected with the planar waves (8.31). a ˆ speciﬁc k and a speciﬁc u. i (8.31. the energy of the plane wave is transported in the direction of −k.40) allows one to express then the ﬁeld ˆ amplitudes 8πNω Ao = . ˆ (8. t) .32). whereas in the case of outgoing waves the energy is transported in the direction of k.43). (8. hence.29) together with the potentials (8. (8.3 Hamilton Operator The classical Hamiltonian for a particle of charge q in a scalar and vector potential V (r) and A(r. t) is real 1 and that Et (r. (8. These photons each carry the energy ω. Note that E (r. t). A ‘poor man’s’ quantization of the ﬁeld is possible at this point by expressing the energy density (8. we will neglect the respective terms in the Hamiltonian (8.

11) to (8. t) are necessary to compensate terms which arise through the phase factor. t) (8.47.10. in the Schr¨dinger equation o 2 ˆ p − qA i ∂t Ψ(r. This leads to the question in how far the gauge transformations (8. t)|2 and.10.47) i ∂t Ψ(r.48) 2m For this purpose one notes i ∂t eiqχ(r.50) The equivalence of (8. t) . 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. (8.8.49) (8.t)/ Ψ(r.11) aﬀect the quantum mechanical description.51) 2m the potentials A(r. t) by a local and time-dependent phase factor eiqχ(r. hence. Obviously. i. t) = 2m where ((· · ·)) denotes derivatives in (( χ)) and ((∂t χ)) which are conﬁned to the function χ(r. 8. 8. that this principle applies only to fundamental interactions. t) = eiqχ(r.g.44.48) implies that the gauge transformation (8. such phase factor does not change the probability density |Ψ(r. In the classical case such question is mute since the gauge transformations do not alter the ﬁelds and. e. t) . t) and V (r. t) and A(r.t)/ p + q(( χ)) Ψ(r.t)/ Ψ(r.e. t) ˆ = eiqχ(r. t) enter whereas in the classical equations of motion ˙ m¨ = q E(r. hence. 8. t) .t)/ . t) inside the double brackets.. Elementary constituents of matter which are governed by other symmetry groups. 8. Applying the gauge transformations (8. . One can show that (8.11) of the potentials is equivalent to multiplying the wave function Ψ(r.47) is equivalent to 2 ˆ p − qA i ∂t eiqχ(r.3: Hamilton Operator 209 Gauge Transformations It is interesting to note that in the quantum mechanical description of a charged particle the potentials V (r. t) + q r × B(r.46). however. not to phenomenological interactions like the molecular van der Waals interaction. accordingly. 8. (8. (8.10. t) r (8. t) ˆ p eiqχ(r.t)/ Ψ(r. does not change expectation values which contain the probability densities3 . have no eﬀect on the motion of the particle described by (8.46) the ﬁelds enter. t) = + qV eiqχ(r. It should be noted.t)/ [ i ∂t − q((∂t χ)) ] Ψ(r.t)/ should not aﬀect a system and that. The idea just stated can be generalized by noting that multiplication by a phase factor eiqχ(r.. by the group 3 The eﬀect on other expectation values is not discussed here.t)/ Ψ(r.45) leads to the Schr¨dinger equation o 2 ˆ p − q A − q(( χ)) + qV − q((∂t χ)) Ψ(r.t)/ constitutes a unitary transformation of a scalar quantity. t) = + qV Ψ(r. an element of the group U(1).

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. The choice of a vector potential aﬀects the form of the wave functions describing the eigenstates and. x3 ) = exp(ik2 x2 + ik3 x3 ) φE (x1 ) . holds (8.g. the generators of SU(2). j = 1.57) with a homogeneous magnetic ﬁeld Bo . The resulting ﬁelds are called Yang-Mills ﬁelds. the so-called Landau gauge ˆ associates the vector potential AL (r) = Bo x1 e2 ˆ (8. therefore. In this o case holds V (r. consequently. t). x2 . for Bo = Bo e3 in (8. likewise can demand the existence of ﬁelds which compensate local transformations described by eiσ·χ(r.55). 3.t) where σ is the vector of Pauli matrices. The stationary homogeneous magnetic ﬁeld B(r. In case of a homogeneous potential pointing in the x3 -direction. We want to describe the stationary states corresponding to the Hamiltonian (8. Solution for Landau Gauge A particularly convenient form for the Hamiltonian results for a choice of a so-called Landau gauge for the vector potential A(r.59) . 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. 2. (8.55) 8.12)]. t) = Bo . (8.210 Interaction of Radiation with Matter SU(2).45) with Hamiltonian (8. The Hamiltonian (8. can be described by various vector potentials. (8. For this purpose we use the wave function in the form Ψ(E.53) ˆ ˆ p·A + A·p + q2 2 A + qV 2m (8.. one can employ the Hamiltonian (8.4 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. Accordingly. x1 . (8. H = 2m m 2m (8. k2 .57) satisﬁes · A = 0 and.52) This expression vanishes in the present case since since ˆ ˆ p · Af = A · pf and. t) ≡ 0. · A = 0 [cf. k3 .58).56) due to the gauge freedom. e.54) ˆ2 p q ˆ q2 2 − p·A + A + qV .55). aﬀects the complexity of the mathematical derivation of the wave functions. thereby. The vector potential (8.58) where ∂j = (∂/∂xj ).56).

k2 . We want to employ . From this observation one can immediately conclude that the wave shifted (by e 3 function of the system.61) leads to − where 2 2me 2 ∂1 + 2 k2 1 3 me ω 2 ( x1 + x1o )2 + 2 2me φE (x1 ) = E φE (x1 ) .59).56) through the vector potential A(r) = 1 Bo × r .65) which localize the electrons.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. However. the familiar harmonic oscillator quantum number. is Ψ(n. is not symmetric in the x1 .59). according to (8.62) x1o = and where ω = k2 eBo eBo me (8. x2 . the so-called symmetric gauge which expresses the homogeneous potential (8.67) . k3 . according to (8.and x2 -coordinates.62) is that of a displaced (by x1o ) harmonic oscillator with o 2 k 2 /2m ) energies. Unfortunately.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 ) . k2 . like the corresponding gauge (8. Without aﬀecting the energy one can form wave packets in terms of the solutions (8. therefore. 2 2me (8.57).65). The energies corresponding to these states are E(n. 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. Solution for Symmetric Gauge The solution obtained above has the advantage that the derivation is comparatively simple. the wave function (8.8. the states are degenerate in the quantum number k2 describing displacement along the x2 coordinate. 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. 2 (8. (8.65) where we replaced the parameter E by the integer n. x1 . k3 ) = ω(n + 2 k2 1 3 ) + .63) this induces a spread of the wave function in the x1 direction. (8.64) is the classical Larmor frequency (c = 1).66) Obviously. The stationary Schr¨dinger equation (8.63) (8.

The theory of classical electromagnetism predicts an analogue energy contribution . (8. (8.71) (8. for any (8. (8.74) is in the present case 1 µclass = − e r v e3 .69) × u and × r = 0.52) 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 . Accordingly.73. Vmag = e 2me class = r me v e3 of the electron ˆ (8. Vmag = −µclass · Bo (8.72) to Hamiltonian (8. In this case the current density measures −e v oriented tangentially to the ring.76) e 2me class class · Bo . y-plane with constant velocity v on a ring of radius r. 8.73) where µclass is the magnetic moment connected with a current density j µclass = 1 2 r × j(r) dr (8.212 Interaction of Radiation with Matter One can readily verify that this vector potential satisﬁes the condition (8.71).67) one can write ˆ p·A = Using · (u × v) = − u · function f (r) ×v + v· 2i · Bo × r . namely.72) allows one to interpret µ = − e L 2me (8.12) for the Coulomb gauge.77) Comparision with (8. For the vector potential (8. an electron moving in the x. the Hamiltonian (8.68) × u yields. using × (uf ) = −u × f +f ˆ ˆ Bo · p × r f = Bo · r × p f . in the present case of constant Bo . accordingly. The latter can be rewritten.75) 2 The latter can be related to the angular momentum µclass = − and. the magnetic moment (8.72) and (8.78) .74) We consider a simple case to relate (8. (8. ˆ (8. ˆ Identifying r × p with the angular momentum operator L. namely.74).70) ˆ ˆ p · A f = − Bo · p × r f .

e. x2 . Accordingly.8. (8. x3 ) = E ΨE (x1 . E = E − (8. x2 ) .56) pointing in the x3 -direction the symmetric gauge (8. For a magnetic ﬁeld (8.82) 1 ω (x1 ∂2 − x2 ∂1 ) i 2me + 1 me ω 2 x2 + x2 1 2 2 2 k2 3 + (8. 10. A derivation of his property and the value of g. 2 .79) To obtain the stationary states.e. x3 ) = exp(ik3 x3 ) ψ(x1 . x2 ) = E ψ(x1 . x2 from x3 setting ΨE (x1 .84) . i.80) (8. the Hamiltonian (8. x2 . 2me (8.81) (8. To obtain these eigenstates we introduce the customary dimensionless variables of the harmonic oscillator 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. we separate the variable x1 .72) with an associated magnetic moment − g 2me S where g ≈ 2. In this case.71) is ˆ H = 2 ˆ2 p e2 Bo + 2me 8me x2 + x2 + 1 2 eBo L3 .67) yields a more symmetric solution which decays to zero along both the ±x1 . described by the operator S. the so-called gyromagnetic ratio of the electron. 2me We have used here the expression for the angular momentum operator ˆ L3 = ( /i)(x1 ∂2 − x2 ∂1 ) .87) . the solutions of ˆ H ΨE (x1 .83) describes two identical oscillators along the x1 -and x2 -directions which ˆ are coupled through the angular momentum operator L3 . x3 ) .4: Electron in Homogenous Magnetic Field 213 as the quantum mechanical magnetic moment operator for the electron (charge −e). ˆ i..86) 1 2 2me 1 2 ˆ as well as of the angular momentum operator L3 . e gives rise to an energy contribution (8. requires a Lorentz-invariant quantum mechanical description as provided in Sect. for Bo = Bo e3 . 10 that the spin of the electron. x2 ) obey then ˆ Ho ψ(x1 . x2 . The functions ψ(x1 . x2 ) where ˆ Ho = − 2 2 2 ∂1 + ∂2 (8.and the ±x2 -direction.83) (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 .85) The Hamiltonian (8.88) me ω xj . (8. We will demonstrate in Sect. likewise. j = 1.

94) is an eigenstate of the vibrational Hamiltonian. (8. . x1 . (8. it holds a† a1 + a† a2 + 1 1 1 2 Ψ (j. 1. . (j) (8. x1 . x2 ) .89) .92) are not eigenstates of L3 . m . (8.214 Employing the creation and annihilation operators 1 a† = √ j 2 and the identity ˆ ω L3 = which can readily be proven. m. . 2 (8.90) a† a2 − a† a1 1 2 . m . x2 ) = a† 1 j−m (j + m)! (j − m)! Ψ(0. m as follows: j = 0. however.94) is an eigenstate of Ho ˆ Ho Ψ (j.. therefore. x2 ) (8.93) ˆ The states (8. aj = √ 2 Interaction of Radiation with Matter Xj + ∂ ∂Xj .92) where Ψ(0. x2 ) = m=−j αmm Ψ(j. x1 . . x2 ) . +j . x2 ) is the wave function for the state with zero vibrational quanta for the x1 .91) We note that the operator a† a2 − a† a1 leaves the total number of vibrational quanta invariant. m . 1 3 . 1 2 since one phonon is annihilated and one created. m . i. . (8. one obtains 1 ˆ 1 H = a† a1 + a† a2 + 1 + 1 1 2 ω i 1 i a† a2 − a† a1 1 2 Xj − ∂ ∂Xj 1 . x1 . attempt to express eigenstates in terms of vibrational wave functions a† 1 j+m Ψ(j. (8. In order to cover all posible vibrational quantum numbers one needs to choose j. m.94) is also an eigenstate of L3 . m = −j..94) Since this state is a linear combination of states which all have vibrational energy (2j + 1) ω.e. (8. through a combination of states j Ψ (j. x1 . such that 1 i a† a2 − a† a1 1 2 Ψ (j. (8. x1 . x1 . the total vibrational energy being ω(2j + 1). m . i.e.92) represents a state with j + m quanta in the x1 -oscillator and j − m quanta in the x2 -oscillator. x1 .97) . 0. x1 . (8.95) (j) ˆ We want to choose the coeﬃcients αmm such that (8. m . obeyed. We. . x1 .96) ˆ holds. If this property is. x2 ) = 2m Ψ (j. x2 ) = ω 2j + 2m + 1 Ψ (j. j = 1. 2 2 . x2 ) . in fact. x2 ) (8. Such eigenstates can be expressed. m . −j + 1. . x1 . 0. x2 ) = ( 2j + 1 ) Ψ (j. .as well as for the x2 -oscillator.

102) We have identiﬁed.9. is identical to the operator L3 introduced in (8. a2 1 2 present notation ←→ b† . m. 5. a1 .5. b+ . (8. the second rotation a factor dmm ( π ).8. a transformation moving the x3 –axis into the x2 – 2 axis.11. x1 . x2 ) (8.10.10.101) 2 2 where Dmm ( π .e. m.10. According to the derivation given there. the 2 2 latter representing the Wigner rotation matrix of Sect. t) introduced above. The connection with the present problem arises due to the fact that the operator J2 in Sect. x2 ) . an example is the Coulomb potential V (r.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. 5. This implies that ˆ we can obtain eigenstates of L3 by rotation of the states Ψ(j. x1 . If we identify 2 a† . x1 . m. We assume that the scalar potential V in (8. x2 ) deﬁned in (8. b† . except for a factor 2 .309) yields ﬁnally Ψ (j. x1 . is in the notation of the present section 1 ˆ a† a2 − a† a1 .5.84) above. m . The required rotation must transform the x3 –axis into the x2 –axis.99) with eigenvalue m. Eq. m.10. (8.288)] 1 ˆ J3 = 2 a† a1 − a† a2 1 2 (8. 5. i. b− + − notation in Sects. x1 .e.55). In the simplest case the radiation ﬁeld consists of a single planar electromagnetic wave described through the potential (8. 0) Ψ(j.11 (8. 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.97). Using the explicit form of the Wigner rotation matrix as given in (5.(5. x2 ). 8.5: Time-Dependent Perturbation Theory (j) 215 In order to obtain coeﬃcients αmm we can proﬁtably employ the construction of angular momentum states in terms of spin– 1 states as presented in Sects. According to Sect. thus.55) conﬁnes the particle to stationary bound states.9. 5. 5. the eigenstates of (8. x1 . . The external radiation ﬁeld is accounted for by the vector potential A(r.83) and conﬁrmed the eigenvalues stated in (8.98) then the states Ψ(j.5. π . cf.. 5.31). the states are eigenstates of the operator [we use for the operator the notation of Sect.9. 5. t) = 1/4πr conﬁning an electron with energy E < 0 to move in the well known orbitals of the hydrogen atom.5. The ﬁrst rotation contributes a factor exp(−im π ). which corresponds there to the angular momentum in the x2 –direction. m) in Sect. a† .100) J2 = 1 2 2i 1 ˆ i.11.92) correspond to the eigenstates |Ψ(j.11 such transformation is provided through π π (j) Ψ (j. 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 . Other radiation ﬁelds can .. x2 ) = Dmm ( . m .

105) in further calculations and to expand the wave function in terms of powers of VS (t) in a perturbation calculation. i.103) (8.104) (8.41) e ˆ e p2 p·A ∼ 2me me me 2me 4 1 2 8πNω . is much smaller than the eigenvalue diﬀerences near ˆ typical atomic bound states. .105) for the case of radiationinduced transitions in atomic systems. We will see below that the perturbation resulting from a ‘pure’ plane wave radiation ﬁeld will serve us to describe also the perturbation resulting from a radiation ﬁeld made up of a superposition of many planar waves.e. using (8.105) where A(r.e.108) We assume for the sake of simplicity that the eigenstates of Ho can be labeled through integers. This result will allow us to neglect the larger than the second term.106) where we adopted the Dirac notation for the states of the quantum system. Estimate of the Magnitude of VS We want to demonstrate now that the interaction VS (t). t) + A (r. in this case. Here the so-called unperturbed system is governed by the Hamiltonian Ho with stationary states deﬁned through the eigenvalue problem Ho |n = n |n . i.105). 1. ωV (8. . the term ∼ A second term in (8.109) The reader should note that the estimates are very crude since we are establishing an order of magnitude estimate only.e. can be considered a weak perturbation. 2 .e. However. t) − m 2m (8. n = 0. i. (8. is much 2 (r. orthonormal basis. t).. t) is given by (8. t). 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. The states |n are thought to form a complete. For an electron charge q = −e and an electron mass m = me one can provide the estimate for the ﬁrst term of (8. one can estimate that the perturbation. In fact. this assumption can be waved as our results below will not depend on it.216 Interaction of Radiation with Matter be expanded through Fourier analysis in terms of such plane waves.42).105) as follows4 . (8.107) 1 = 1 n=0 |n n| .. i. and that the ﬁrst term in (8... We ﬁrst note. as given in (8. we assume n|m = δnm and for the identity ∞ (8. . the term ∼ p · A(r. we discount the possibility of a continuum of eigenstates.

Consequently. Hence. altogether. one can neglect this term as long as the ﬁrst term gives non-vanishing contributions. replace the perturbation (8.105).114) Employing for the second factor the estimate as stated in (8.. λ = 2πc/λ.105) due to a radiation ﬁeld by VS = − q ˆ p · A(r. i.116) This term is obviously much smaller than the ﬁrst term. We want to estimate now the second term in (8.5 ˚. t) .112) π λ me c2 and with e2 /ao ≈ 27 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.. i. Nω = 1.111) For ω = 3 eV and a corresponding λ = 4000 ˚ one obtains. Assuming a single photon. e ˆ p · A ∼ 10 eV · 10−4 = 10−3 eV .110) where ao is the Bohr radius. one obtains for (8. m (8.ˆ u 4πNk α(k.115) (8.8. hence. and me c2 ≈ A A 500 keV 2 ao ω ≈ 10−8 (8. and as long as the photon densities Nω /V are small. me (8.42) it holds VS ≈ e m k.117) In case that such perturbation acts on an electron and is due to superpositions of planar waves described through the vector potential (8.113) 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. We can.e. u) p · u exp i(k · r − ωt) ˆ ˆ ˆ kV . the ﬁrst term in (8.105) for radiation ﬁelds can be considered a small perturbation.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 .118) .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 (8.112) yields e2 A2 ∼ 10 eV · 10−8 = 10−7 eV . 2me (8. (8. a volume V = λ3 where λ is the wave length corresponding to a plane wave with frequency ω. Using again (8. with ao ≈ 0.e. (8.

The sum runs over all possible k vectors and might actually be an integral. . n = 1.123) i = Ho exp − Ho (t − to ) (8.127) i ∂t |ΨD (t) = VD (t) |ΨD (t) . (8. Perturbation Expansion The generic situation we attempt to describe entails a particle at time t = to in a state |0 and a radiation ﬁeld beginning to act at t = to on the particle promoting it into some of the other states |n . 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 . . under ordinary circumstances.124) (8.119) one obtains i exp − Ho (t − to ) ( Ho + i ∂t ) |ΨD (t) i = [ Ho + VS (t) ] exp − Ho (t − to ) |ΨD (t) from which follows i i exp − Ho (t − to ) i ∂t |ΨD (t) = VS (t) exp − Ho (t − to ) |ΨD (t) . . A factor α(k. u) has been added to describe eliptically or circularly polarized ˆ waves.42) through k = |k| = ω. The states |0 .118) is the form of the perturbation which. One seeks to predict the probability to observe the particle in one of the states |n . |Ψ(to ) = |0 (8.125) (8.. 2. For this purpose one needs to determine the state |ΨS (t) of the particle. Equation (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 .121) (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) (8.108) as the eigenstates of the unperturbed Hamiltonian Ho .122) where |ΨD (to ) = |0 .120) (8. Multiplying the latter equation by the operator exp i (8. Using i i ∂t exp − Ho (t − to ) and (8. describes the eﬀect of a radiation ﬁeld on an electron system and which will be assumed below to describe radiative transitions.218 Interaction of Radiation with Matter where we have replaced ω in (8. |n are deﬁned in (8. the sum over u involves the two possible polarizations of planar ˆ electromagnetic waves.

136) together with the initial conditions |ψD (to ) = |0 0 for n = 0 for n = 1. = . From (8.128) We note that the transition probability (8.137) |ΨD (t) = (1) (0) (8.137) One can readily verify that (8.135) i ∂t |ΨD (t) (n) VD (t) |ΨD (n−1) (t) (8.5: Time-Dependent Perturbation Theory where VD (t) = exp 219 i i Ho (t − to ) VS (t) exp − Ho (t − to ) .121) expressed in terms of ΨD (t) is i p0→n (t) = | n|exp − Ho (t − to ) |ΨD (t) |2 .132–8. to (8. we deﬁne |ΨD (t) through the evolution equations i ∂t |ΨD (t) i ∂t |ΨD (t) i (2) ∂t |ΨD (t) (1) (0) = = = . Employing this result one obtains for (8. In order to determine |ΨD (t) described through (8. consequently. (8.133. Equations (8.134. . .139) .133) (8.134) (8.137) are consistent with (8.132) where |ΨD (t) accounts for the contribution due to n-fold products of VD (t) to |ΨD (t) .133–8. 2. using |exp[− i n (t n (8.129) n| and. 8. (8. . 0 VD (t) |ΨD (t) (1) VD (t) |ΨD (t) (0) (8. (8. .131) p0→n (t) = | n |ΨD (t) |2 .128).130) = exp − i n (t − to ) − to )]| = 1. Accord(n) ingly. . 8.8.127.137) can be solved recursively. 8.138) 1 i t dt VD (t ) |0 .137) follows |ΨD (t) = |0 . We will consider here only the two leading contributions to |ΨD (t) .127) we assume the expansion ∞ |ΨD (t) = n=0 (n) |ΨD (t) (n) (8. Due to the Hermitean property of the Hamiltonian Ho holds n|Ho = i n|exp − Ho (t − to ) from which we conclude. . . n| (8.

144) r is the position of the electron and p = ( /i) is the momentum operator of the electron. is VS = ˆ ˆ V1 exp(−iω1 t) + V2 exp( iω2 t) eλt . such that the combined radiation ﬁeld is decribed by the vector potential A(r. yields for (8. 8.6 Perturbations due to Electromagnetic Radiation We had identiﬁed in Eq. . however. We will demonstrate below that the the sign of iωt determines if the energy of the planar wave is absorbed (“-” sign) or emitted (“+” sign) by ˆ the quantum system. In (8.. For the sake of including the eﬀect of superpositions of plane waves we will assume. according to (8.142). We want to apply now the perturbation expansion derived to such perturbations.140) Altogether we have provided the formal expansion for the transition amplitude n|ΨD (t) = n|0 + ∞ 1 i 1 i t dt n|VD (t ) |0 to 2 t t (8. to → −∞ (8. + m=0 dt to to dt n|VD (t )|m m|VD (t ) |0 8. 8. The resulting perturbation on an electron system. i.144) describes the propagation of the planar wave. in turn. non-singular. deﬁnes the direction of the E-ﬁeld of the ˆ radiation.137) |ΨD (t) = (2) Interaction of Radiation with Matter 1 i 2 t t dt to to dt VD (t ) VD (t ) |0 .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).41). A factor exp(λt).141) + . A2 are deﬁned through (8.e.135.144) 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. according to (8. (8.220 This result. This factor will serve mainly the purpose of keeping in the following derivation all mathematical quantities properly behaved. (8. that two planar waves simulataneously interact with an electronic system.142) combining an incoming and an incoming or outgoing wave. λ → 0+ . II III (8. The coeﬃcients A1 ..118). note that uj . the direction of ˆ the propagation being determined by k = k/|k|.34. t) = + A1 u1 exp i (k1 · r − ω1 t) ˆ A2 u2 exp i (k2 · r ˆ ω2 t) incoming wave incoming or outgoing wave (8.. The factor III in (8. λ → 0+ has been introduced which describes that at time to → −∞ the perturbation is turned on gradually.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 .

(8.141). (8.145) we obtain then. using (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 (8.143).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 .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 .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 .150) and.143) to the expansion (8.149) Using the deﬁnition of VD stated in (8. For the 1st order contribution to the transition amplitude n|ΨD (t) = (1) 1 i t dt n|VD (t ) |0 to (8. (8.8.128).140. (8.145) n|ΨD (t) (1) m (t − to ) |m . According to (8.146) = 1 λ→0+ t→−∞ i lim lim × t dt exp to i ( n − o − i λ) t iω2 t × . employing the perturbation (8. (8. 8.130) and (for m = 0) i i exp − Ho (t − to ) |m = exp − for (8.6: Perturbations due to Electromagnetic Radiation 1st Order Processes 221 We employ now the perturbation (8.148) We consider now the 2nd order contribution to the transition amplitude.128) one obtains k|VD (t) | = k|VS (t)| exp i ( k − ) (8.

148) and ∗ |z1 + z2 |2 = |z1 |2 + |z2 |2 + 2Re(z1 z2 ) (1) (8.222 ˆ ˆ + n|V1 |m m|V2 |0 exp ˆ ˆ + n|V2 |m m|V1 |0 exp i i ( ( − − Interaction of Radiation with Matter − ω1 − i λ)t exp ω2 − i λ)t exp i i ( ( − − ω2 − i λ)t − ω1 − i λ)t n m m o n m m o Carrying out the integrations and the limit limt→−∞ provides the result (2) n|ΨD (t) = − 1 ∞ 2 λ→0+ lim (8. provide now the transition probability p0→n (t) according to Eq. i. (8. respectively. in which case one can expect that it is larger than the 2nd (2) order contribution n|ΨD (t) which we will neglect.. For this rate holds .131).155) λ→0+ ( o We are actually interested in the transition rate. We assume ﬁrst that the ﬁrst order transi(1) tion amplitude n|ΨD (t) is non-zero. Using (8.154) 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) (8.148) and (8. We also assume for the ﬁnal state n = 0 such that n|0 = 0 and p0→n (t) = | n|ΨD (t) |2 holds. the time derivative of p0→n (t).152).152) m=0 i ˆ ˆ n|V1 |m m|V1 |0 exp m − o − ω1 − i λ + − 2 ω1 − 2i λ)t n − o − 2 ω1 − 2i λ ( n o − ˆ ˆ n|V2 |m m|V2 |0 exp ω2 − i λ m− o ˆ1 |m m|V2 |0 exp ˆ n|V + ω2 − i λ m− o ˆ2 |m m|V1 |0 exp ˆ n|V + m − o − ω1 − i λ 1st Order Radiative Transitions i 2 ω2 − 2i λ)t 2 ω2 − 2i λ n− o i ( n − o − ω1 ω2 − 2i λ)t ω2 − 2i λ n − o − ω1 ( n o i − − ω1 ω2 − 2i λ)t ω2 − 2i λ n − o − ω1 ( n o − The 1st and 2nd order transition amplitudes (8.e.

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. ˆ (8.142) ˆ ˆ The matrix elements n|V1 |0 and n|V2 |0 in (8. The result supports our deﬁnition of incoming and outgoing waves in (8. the two terms apearing on the rhs.156) such that the 1st order contributions of the two planar waves of the perturbation simply add.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 (8.158) ˆ + | n|V2 |0 |2 δ( − ω2 ) Obviously. in fact.e.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+ (8.160) leads to ﬁnal states |n with energy n = o ± ω2 . a property. one should average the rate over many periods of the radiation. the matrix elements can vary strongly for diﬀerent states |n of the system. In case the matrix elements are non-zero.156) ( o The period of electromagnetic radiation absorbed by electronic systems in atoms is of the order 10−17 s.31) and (8.143. The result of such average is. i. of this expression describe the individual eﬀects of the two planar wave contributions of the perturbation (8. any attempt to do so.158) play an essential role for the transition rates of radiative transitions. For the resulting expression the limit limλ→0+ can be taken. Using lim x2 + 2 δ→0+ = π δ(x) (8. is much shorter than could be resolved in any observation.8. (8. The time average will be denoted by · · · t . .. however. The δ-functions appearing in this expression reﬂect energy conservation: the incoming plane wave contribution of (8. The second contribution to (8. 8. which is observed through the so-called spectral intensities of transitions |0 → |n . First.144).142–8.159) leads to ﬁnal states |n with energy n = o + ω1 . due to the vector potential A1 u1 exp i (k1 · r − ω1 t) . due to the uncertainty relationship would introduce a considerable perturbation to the system. to cancel the third term in (8. Hence.144). describing either an incoming or an outgoing plane wave due to the vector potential A2 u2 exp i (k1 · r ˆ ω2 t) .158).

157) yields.164) (8. it holds p0→n (t) = | n|ΨD (t) |2 . in analogy to (8. i. e.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. Time average · · · t of expression (8.k=1 j=k 2 2 2 2 ∗ zj zk (8.161) To determine the transition rate we proceed again..163) × (8.158). 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..168) absorption of 2 photons ω1 .158). |z1 + z2 + z3 |2 t = |z1 |2 + |z2 |2 + |z3 |2 (8.224 2nd Order Radiative Transitions Interaction of Radiation with Matter We now consider situations where the ﬁrst order transition amplitude in (8.165) × ∞ 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. (8.167) Taking now the limit limλ→0+ and using (8.152).153) vanishes such that the leading contribution to the transition probability p0→n (t) arises from the 2nd order amplitude (8. We deﬁne ∞ z1 = m=0 ˆ ˆ n|V1 |m m|V1 |0 m − o − ω1 − i λ exp i × (8. hence. similarly.166) over many periods of the radiation yields exp[ i (±ω2 − ω1 )] t = 0 and.162) × and.e.e. ∞ − 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. exp[ i (±ω2 − ω1 )]. i. as we did in the the case of 1st order transitions. in Eqs. (2) (8.153–8.g..

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

8.174) We seek to evaluate the matrix element ˆ M = n| p eik·r |0 . φ) = 2 − 1 2 1 −r/ao e Y00 (θ.e. the second term can be discounted in case of absorption. and the integral is =1.178) These wave functions make signiﬁcant contributions to this integral only for r-values in the range r < 10 ao . For example.m (r. φ) × × eik·r e−r/ao Y00 (θ.158). θ.2 = m ωV Only the ﬁrst term of (8. t) = 8πN ωV u exp ˆ ı (k · r − ωt) . =0. φ) a3 o 1s (8.172) δ( n − o − ω) (8.m=0 (r. in this range one can expand eik·r ≈ 1 + ik · r + . . (8. φ) (8.143) will contribute to the absorption process. φ) a3 ao o 2p (8.170) We will employ only the real part of this potential. However. in case of a radiative transition from the 1s state of hydrogen to one of its three 2p states. . The absorption rate.226 Absorption of a Plane Polarized Wave Interaction of Radiation with Matter We consider ﬁrst the case of absorption of a monochromatic. (8. according to (8. the wave functions are (n. θ.177) √ M = 6 0 ∞ 1 2π ia4 o r2 dr −1 dcosθ 0 ∗ dφ r e−r/2ao Y1m (θ. to → −∞ (8. t) = 2πN ωV u exp ˆ ı (k · r − ωt) + 2πN ωV u exp ˆ ı (−k · r + ωt) . i..173) V1. ˆ (8.171) The perturbation on an atomic electron system is then according to (8. . plane polarized wave described through the complex vector potential A(r.144) VS = where e 2πN ˆ ˆ p · u e±ik·r . the vector potential actually assumed is A(r. m denote the relevant quantum numbers) ψn=1. 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 .175) The matrix element involves a spatial integral over the electronic wave functions associated with states |n and |0 . (8.176) ψn=2. (8.143. φ) = 6 r −r/2ao e Y1m (θ.179) . λ → 0+ .

182) one obtains M ≈ Insertion into (8. V (r) ] 2me =0 = Using r = 3 ˆ j=1 xj ej 1 2me 3 pk [ r. pj ] = i δkj one obtains ˆ [ r. (8. Ho ] = i m 3 pk ej δjk = ˆ j. We are now in a position to obtain an alternative expression for the matrix element (8. pk ] + ˆ ˆ k=1 1 2me 3 [ r. Using |k| = 2π/λ.182).175).179) and other terms are never larger than 20π ao /λ.183) (8.181) For the commutator in (8.181) and (8. Ho ] i where Ho is the Hamiltonian given by (8.187) .k=1 i ˆ p m (8.k=1 i m 3 pk ek = ˆ j.174) yields kabs = 4π 2 e2 N ω V u · n| ˆ |0 ˆ r 2 m( o − m n| [r. V (r) = − .185) from which follows (8. 2me r (8.182) one ﬁnds [ r.104) and. Using (8.186) δ( n − o − ω) (8. Transition Dipole Moment A further simpliﬁcation of the matrix element (8. This simpliﬁcation results from the identity m ˆ p = [ r. in case of the hydrogen atom.182) (8.180) ∆E2p−1s and ao = 0.178) holds eik·r ≈ A 1 1 + O( 50 ) such that one can approximate eik·r ≈ 1 .529 ˚ one concludes that in the signiﬁcant integration range in (8. One refers to this approximation as the dipole approximation.7: One-Photon Emission and Absorption 227 One can estimate that the absolute magnitude of the second term in (8. Ho ] = ˆ2 p [ r. is Ho = ˆ (p)2 e2 + V (r) . ] + [ r.184) and the commutation property [xk . Ho ] |0 = i i n) n|r |0 . pk ] pk ˆ k=1 (8.8.175) can then be ˆ achieved and the diﬀerential operator p = i replaced by by the simpler multiplicative operator r. the value of the wave length for the 1s → 2p transition 2π c λ = = 1216 ˚ A (8.

For the two angles α = ∠(ˆ1 . Integrating and summing accordingly over all contributions as given by (8.187) 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. ϕ1 = 0 and of u2 by ϑ2 = π/2. ˆ In order to carry out the integral dk we note that u describes the possible polarizations of the ˆ ˆ and u. a very hot oven. such that the thermal radiation present supplies a continuum of frequencies. three-dimensional vector. u2 . k frame is described by the angles ϑ. ρ) holds then u cosα = cosϑ1 cosϑ + sinϑ1 sinϑ cos(ϕ1 − ϕ) = sinϑcosϕ and cosβ = cosϑ2 cosϑ + sinϑ2 sinϑ cos(ϕ2 − ϕ) = sinϑsinϕ . 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 (8.33) are orthogonal to each other. 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.. ϕ2 = π/2. k ˆ As a result. hence.190) (8. x3 -axes of a right-handed cartesian vectors u1 . will actually be replaced by a distribution function which reﬂects (1) the ﬁnite life time of the states |n . obtain the transition rate in the present case by adding the individual transition rates of all planar waves present in the oven. Accordingly. Absorption of Thermal Radiation We want to assume now that the hydrogen atom is placed in an evironment which is suﬃciently hot. according to (8. We have demonstrated in our derivation of the rate of one-photon processes (8. The δ-function appearing in this expression.158) above that in ﬁrst order the contributions of all components of the radiation ﬁeld add.193) .35). ρ) and ˆ u β = ∠(ˆ2 .188) ˆ Here dk is the integral over all orientations of k. there are ony two linearly independent directions of u possible. ϕ. The unit ˆ ˆ ˆ ˆ can be chosen to point along the x1 . planar waves as deﬁned in (8. ˆ u ˆ (8.191) (8. u2 and k ˆ ˆ coordinate system.31–8. |0 . The direction of this vector in the u1 . the direction of u1 is described by the ˆ ˆ ˆ ˆ angles ϑ1 = π/2. in practical situations. and all polarizations of the radiation.189) where the factor 1/ arose from the integral over the δ-function. one can express | u · n| r |0 |2 = |ρ|2 ( cos2 α + cos2 β ) = sin2 θ . i.228 Interaction of Radiation with Matter where we used the fact that due to the δ-function factor in (8.174) one can replace n − o by ω. say u1 and u2 . directions. We can.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. and (2) the fact that strictly monochromatic radiation cannot be prepared such that any radiation source provides radiation with a frequency distribution. x2 .e.

for atoms or symmetric molecules. This dependence predicts. (8. that emission occurs even if no photon is present in the environment.e. For practical evaluations we provide an expression which eliminates the physical constants and allows one to determine numerical values readily. Emission of Radiation We now consider the rate of emission of a photon. 2 .173). The most intensely absorbing molecules are long. Nω photons before absorption. 8.172. i. This poses. in agreement with observations. . (8. i. The quantum nature of radiation manifests itself in that the number of photons Nω msut be an integer. ˆ In case of emission only the second term V2 exp(+iωt) in (8. However. 1. for Nω = 0. there is also a contribution to the emission rate which is proportional to Nω . . In case of a classical radiation ﬁeld one would expect that emission cannot occur. Transition Dipole Moment The expression (8. the calculation of the emission rate proceeds as in the case of absorption. However. as for the absorption process. by planar waves with vector potential (8.189) to yield the total absorption rate kabs (tot) = Nω 4 e2 ω 3 | n|r|0 |2 3 c3 .196) The last two factors in (8.8. Accordingly.173) contributes.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 |.197) o (tot) (8. a quantum mechanical treatment of the radiation ﬁeld leads to a total emission rate which is proportional to Nω + 1 where Nω is the number of photons before emission. the absorption rate is of the order of 109 s−1 or 1/nanosecond.37 × 1019 1 ao | n|r|0 |2 × s λ λ2 .195) = Nω 1. This diﬀerence between emission and absorption is due to the quantum nature of the radiation ﬁeld..171) and perturbation (8.e.194) combined are typically somewhat smaller than (1 ˚/1000 ˚)3 = A A 10−9 . For this purpose we use ω/c = 2π/λ and obtain 4 e2 ω 3 32π 3 e2 ao 1 ao = = 1.37 × 1019 × 3 3 c3 3 ao λ3 s λ and kabs where λ = 2πc n − (8. . Otherwise.7: One-Photon Emission and Absorption 229 This geometrical average. however. can be inserted into (8.. in particular. The corresponding process is termed spontaneous emission. The value of | n| r |0 | determines the strength of an optical transition.. the resulting total rate of emission bears a diﬀerent dependence on the number of photons present in the environment. a problem in case of emission by quantum systems in complete darkness. Nω = 0. linear molecules. The transition dipole moment can vanish for many transitions between stationary states of a quantum system. ﬁnally.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 . However. The radiation ﬁeld is described.

195) and (8.202) It follows.. it must hold Nω ksp N0 = (Nω + 1) ksp Nn (8. exp[− ω/kB T ] = This equation yields Nω = 1 . Deﬁning the rate of spontaneous emission 4 e2 ω 3 | n|r|0 |2 (8.g. We assume n − o = ω. 8.198) (8.203) i. is (tot) kem 4 e2 ω 3 | n|r|0 |2 (spontaneous emission) 3 3c 4 e2 ω 3 | n|r|0 |2 (induced emission) + Nω 3 3c 4 e2 ω 3 = ( Nω + 1 ) | n|r|0 |2 3 3c Nω photons before emission.200) where kB is the Boltzmann constant. using (8. = (8. respectively. accordingly. 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.201) ksp = 3 c3 the rates of absorption and emission are Nω ksp and (Nω + 1)ksp .198) is consistent with Planck’s radiation law which reﬂects the (boson) quantum nature of the radiation ﬁeld.8 Two-Photon Processes In many important processes induced by interactions between radiation and matter two or more photons participate.204) Nω . e.198) to determine the stationary distribution of photons ω in an oven of temperature T . 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 . To demonstrate this property we apply the transition rates (8. exp[ ω/kB T ] − 1 (8. The number of atoms undergoing absorption in unit time are Nω ksp No and undergoing emission are (Nω +1)ksp Nn . the well-known Planck radiation formula.230 Interaction of Radiation with Matter which is termed induced emission since it can be induced through radiation provided. In the following we discuss several examples. Let No and Nn denote the number of atoms in state |0 and |n .. For these numbers holds Nn / No = exp[−( n − o )/kB T ] (8. in lasers.199) Planck’s Radiation Law The postulate of the Nω + 1 dependence of the rate of emission as given in (8.e. Hence. Nω + 1 (8. .200). respectively. The total rate of emission.

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. even if n| r |0 vanishes.104) with stationary states |n deﬁned through (8.207) Expression (8. (8. a transition between the states |0 and |n may be possible. ﬁnally.205) The respective radiative transition is of 2nd order as described by the transition rate (8.172. ˆ (8. k = Nω V 2 8π 3 e4 × δ( m− ∞ m=0 o ( n − ˆ m) u · n|ˆ |m ( m − o ) u · m|ˆ |0 r ˆ r ω( m − o − ω − i λ) 2 − 2 ω) .173). (8. (8. which is not necessarily the case. One obtains.181) and using (8. but then requires the absorption of two photons.208) has been prepared. t) = Ao1 u1 cos(k1 · r − ω1 t) . 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. k = 2π e2 2πNω m2 ωV e × δ( m 2 ∞ m=0 n|ˆ · p eik·r |m m|ˆ · p eik·r |0 u ˆ u ˆ m − o − ω1 − i λ 2 × (8. The electron absorbs the radiation and emits immediately a second photon. induce single photon absorption processes as described by the transition rate Eq.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.106). Scattering of Photons at Electrons – Kramers-Heisenberg Cross Section We consider in the following the scattering of a photon at an electron governed by the Hamiltonian Ho as given in (8.207) one is.8. If the sum in (8. however.8: Two-Photon Processes Two-Photon Absorption 231 The interaction of electrons with radiation. However.207) does not converge rapidly.182) yields.207) does not provide a suitable avenue of computing the rates of 2-photon transitions. . as described through the vector potential ˆ A(r. In this case one needs to choose the energy of the photons to obey n = o + 2 ω. then expression (8. 8. The intense radiation ﬁelds of lasers allow one to increase transition rates to levels which can readily be observed in the laboratory.168) where the ﬁrst term describes the relevant contribution. The transition requires that the transition dipole moment n| r |0 does not vanish for two states |0 and |n . Employing the dipole approximation (8. however. under ordinary circumstances. We assume that a planar wave with wave vector k1 and polarization u1 . faced with the dilemma of having to sum over all intermediate states |m of the system. In applying (8.187). in case ˆ of absorption only V1 contributes.206) − o − 2 ω) . dropping the index 1 characterizing the radiation. The resulting rate of the transition depends on 2 Nω .

213) where N2 /V is the density of photons characterized through k2 . We know already from our description in Section 8. 8. o1 ˆ o2 ˆ (8. u2 .208.210). hence. In the present case we consider only scattering processes which absorb radiation corresponding to the vector potential (8.211) are given by VS (t) = e ˆ p · A+ u1 exp[i(k1 · r − ω1 t)] + A− u2 exp[−i(k2 · r − ω2 t)] o1 ˆ o2 ˆ 2me contributes in 2nd order + e2 4me A+ A− u1 · u2 exp{i[(k1 − k2 ) · r − (ω1 − ω2 ) t]} ˆ o1 o2 ˆ contributes in 1st order The eﬀect of the perturbation on the state of the electronic system is as stated in the perturbation expansion (8. This expansion yields.209) cosθ2 where the value of k2 has been ﬁxed.210) A(r.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 .e. i.208).210) is accounted for by the following contributions of (8. for the components of the wave (8.208) and emit radiation corresponding to the vector potential (8. the wave characterized through k1 . ˆ The perturbation which arises due to the vector potential (8..210) The vector potential which describes both incoming wave and outgoing wave is a superposition of the potentials in (8. (8. the amplitude A+ is given by o1 A+ = o1 8πN1 ω1 V (8.105).212) where N1 /V is the density of photons for the wave described by (8. the amplititude A− deﬁned in (8. u1 . 8. The relevant terms of the perturbation (8. t) = Ao2 u2 cos(k2 · r − ω2 t) .141).208) and the emission of the radiation in (8.105) using the vector potantial (8. t) = A+ u1 exp[ i (k1 · r − ω1 t) ] + A− u2 exp[ i (k2 · r − ω2 t) ] .6 above that the absorption of the radiation in (8.211) accounts for the emitted wave and. The vector potential describing the emitted plane wave is then A(r.211) is o2 A− = o2 8π (N2 + 1) ω1 V (8.211) is stated in Eq. in the present case. later we will allow the quantum system to select appropriate values. ˆ (8.208. according to the ˆ description of emission processes on page 229.211) The ﬁrst term describes the absorption of a photon and. however.214) . 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 second term in (8.

e.8 · 10−15 m .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 (8.216) 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 .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 (8.218) . Only the second (1st order) and the third (2nd order) terms in (8. me c2 (8.217) 2 Here ro denotes the classical electron radius ro = e2 = 2. one needs to integrate the rate as given by (8.215) 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. Inserting also the values (8.212. i. only these two terms contribute to the scattering amplitude.8.215) correspond to scattering processes in which the radiation ﬁeld ‘looses’ a photon ω1 and ‘gains’ a photon ω2 .. Accordingly. Hence. evaluating the time integrals in (8.e. i.181) in stating this result.. Following closely the procedures adopted in evaluating the rates of 1st order and 2nd order radiative transitions on page 222–225.8: Two-Photon Processes function accounting for absorption and re-emission of a photon n|ΨD (t) = + + m=0 233 n|0 + 1 e2 A+ A− u1 · u2 n|0 ˆ ˆ i 4me o1 o2 ∞ (8. 8.216) over all available modes of the ﬁeld. one needs to carry out the integration 2 V(2π)−3 k2 dk2 · · ·.

221) suggests to expand S( ω) S( ω) = S(0) + S (0) ω + Using 1 − o ± ω one can readily determine m (8.227) u ˆ 0|ˆ1 · p |m m|ˆ2 · p |0 u ˆ u ˆ 0|ˆ2 · p |m m|ˆ1 · p |0 u ˆ + me ( m − o )3 me ( m − o )3 (8.223) m Condition (8. it follows 2 dσ = ro (N2 + 1) dΩ2 |ˆ1 · u2 − S( ω) |2 u ˆ (8. consequently. Rayleigh scattering is deﬁned as the limit in which the wave length of the scattered radiation is so long that none of the quantum states of the electronic system can be excited. Accordingly. one assumes the even stronger condition ω1 << | o − m| .228) S (0) = m S (0) = 2 m .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 (8.226) 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. .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. i. for all states |m of the electronic system (8. 2 . (8.222) where S( ω) = 1 me 0|ˆ1 · p |m m|ˆ2 · p |0 u ˆ u ˆ 0|ˆ2 · p |m m|ˆ1 · p |0 u ˆ u ˆ + m − o + ω m − o − ω 1 S (0)( ω)2 + .. one can relate (8. − o )2 ( m − o )3 (8. ω1 = ω2 .219) (8.e. Rayleigh Scattering We turn ﬁrst to an example of so-called elastic scattering. a process in which the electronic state remains unaltered after the scattering.. in fact.234 Interaction of Radiation with Matter The factor N1 c/V can be interpreted as the ﬂux of incoming photons.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. .221) Using |n = |0 and..224) = m 1 − o ( m ω ( ω)2 + + .

as a compass.8: Two-Photon Processes 235 ˆ These three expressions can be simpliﬁed using the expression (8.182) for p and the expression (8. u ˆ ˆ ˆ ˆˆ i The commutator property [xj . Expression (8.229) both for the u1 · p and the to u2 · p terms in (8. (8.8.182) 0|ˆ1 · p |m u ˆ 1 = 0|ˆ1 ·r |m u me ( m − o ) i This transforms (8.229) both to u1 · p and to u2 · p which results in ˆ ˆ ˆ ˆ me S (0) = 2 ( 0|ˆ2 ·r |m m|ˆ1 ·r |0 − 0|ˆ1 ·r |m m|ˆ2 ·r |0 ) .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 (8.226). 8. ˆ ˆ S (0) given in (8. i.232.233) m Employing again (8.234) where we used for the second identity the fact that u1 ·r and u2 ·r commute.230) According to (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 . this term cancels the u1 · u2 term in (8.236) is of great practical importance.227) vanishes. ˆ We want to simplify ﬁrst (8.232) Obviously. . for example. 8.. honey bees. m|ˆ1 · p |0 u ˆ 1 = − m|ˆ1 ·r |0 u me ( m − o ) i (8. u u u u (8.222) of the cross section for Rayleigh scattering dσ = 2 ro m2 ω 4 (N2 + 1) dΩ2 × e (8.108) for the identity operator.224.e. For this purpose we replace p using (8.k=1 (ˆ1 )j (ˆ2 )k δjk = u1 · u2 u u ˆ ˆ (8. pk ] = i δjk yields ﬁnally ˆ S(0) = 1 S(0) = i 3 3 (8.236) 2 × m 0|ˆ∗ ·r |m m|ˆ2 ·r |0 u1 u 0|ˆ∗ ·r |m m|ˆ1 ·r |0 u2 u + m − o m − o We have applied here a modiﬁcation which arises in case of complex polarization vectors u which ˆ describe circular and elliptical polarizaed light. 8.108) yields S (0) = me 2 0| [ˆ2 ·r.222).222). the blue color of the sky and the polarization pattern in the sky which serves many animals.234.k=1 j. ˆ ˆ We want to prove now that expression (8.229) (8.231) (ˆ1 )j (ˆ2 )k 0|[xj . For this purpose we apply (8. pk ]|0 = u u ˆ j. It explains. u1 ·r ] |0 = 0 u ˆ (8.235) and obtain the leading contribution to the expression (8.235) We can now combine eqs. Using again (8.108) this is 1 0|ˆ1 ·r u2 · p − u2 · p u1 ·r |0 .228) provides then the ﬁrst non-vanishing contribution to the scattering cross section (8.

238) (8.241) 0 0 Similarly.243) cos2 θ2 cos2 φ2 for u1 = u1 . |n = |0 and ω1 = ω2 = ω in (8. (8.240) cosθ2 We choose for the polarization of the incoming radiation the directions along the x1 .242) |k2 × k1 | 0 where the second identity follows readily from k1 = e3 and from (8. To evaluate |ˆ1 · u2 |2 we assume that k1 is oriented along the x3 -axis and. u2 = u2 ˆ ˆ ˆ ˆ cos2 φ2 cos2 θ2 sin2 φ2 for u1 = u1 . We want to assume. that the dipole approximation is still valid which restricts the applicability of the following derivation to k1 >> One obtains immediately from (8.239) We will show below that this expression decribes the non-relativistic limit of Compton scattering. for all states |m of the electronic system . ao Bohr radius . 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 . u ˆ 1 ao . (8.244) .220). though.237) The resulting scattering is called Thomson scattering. however.236 Thomson Scattering Interaction of Radiation with Matter We consider again elastic scattering. 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 . u2 = u2 ˆ ˆ ˆ ˆ (8. u2 = u2 ˆ ˆ ˆ ˆ (1) (2) (2) (1) (1) (2) (1) (2) (2) (1) needs to be (8..and the x2 -axes 1 0 (1) (2) u1 = 0 . assume now that the scattered radiation has very short wave length such that ω >> | o − m| . hence. u1 = 1 ˆ ˆ (8. i. The ﬁrst choice is ˆ sinφ2 k2 × k1 (1) u2 = ˆ = −cosφ2 (8.240).220) 2 dσ = ro (N2 + 1) dΩ2 |ˆ1 · u2 |2 . u2 = u2 ˆ ˆ ˆ ˆ for u1 = u1 . the emitted u ˆ radiation is decribed by the wave vector sinθ2 cosφ2 k2 = k1 sinθ2 sinφ2 (8.e..

i.e. Since the scattering is inelastic.8. the states |n etc. 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.. one speaks of Stokes scattering.)n . For electronic systems in molecules or crystals the degrees of freedom excited are nuclear motions. deﬁned in (8. (8. 8. In case that the nuclear degrees of freedom excited absorb very little energy.247) becomes identical with the Thomson scattering cross section (8.246) 2 This expression is the non-relativistic limit of the cross section of Compton scattering. Such scattering is called Raman scattering. one speaks of anti-Stokes scattering. or in case of translational motion of molecules in liquids.e. e.)n .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 . molecular vibrations or crystal vibrational modes. In the case of such scattering an electron system absorbs and re-emits radiation without ending up in the initial state. the ﬁrst term in (8. 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 (8.249) dσ = ro (N2 + 1) ω1 −1 −1 ω2 − ω1 = . but most often involves other degrees of freedom.248) me c2 One can readily show that in the non-relativistic limit.. in case of a diatomic molecule |n = |φ(elect. The excitation can be electronic. the scattered radiation is polarized to some degree. Raman Scattering and Brillouin Scattering We now consider ineleastic scattering described by the Kramers-Heisenberg formula. In the case that the scattering excites other than electronic degrees of freedom. the resulting scattering cross section is 2 ro (N2 + 1) ( 1 + cos2 θ2 ) dΩ2 .246). The radiation scattered at right angles is even completely polarized (1) along the u2 -direction. as in the case of excitations of accustical modes of crystals. ˆ In case that one measures the scattered radiation irrespective of its polarization. if energy is released.247. φ(vibr. i.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. for c → ∞ the Compton scattering cross section (8.247) ( 1 − cosθ2 ) (8.g.220) vanishes and one obtains in case of Raman scattering ω2 2 dΩ2 | u2 · R · u1 |2 ˆ ˆ (8.. the scattering is termed Brillouin scattering.220) represent actually electronic as well as nuclear motions. If energy is absorbed by the system. The energy deﬁcit is used to excite the system.

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

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

. However. . . namely implicitly the group property. particles 4 and 5. and the second row showing the numbers 1 to N again. i. . 4) T (4. P2 = T (4. xN ) = Ψ(xP (1) . The elements P ∈ SN can be represented conveniently as 2 × N matrices. 8) . Denoting by P (j) the image of j. the top row representing the numbers 1 to N labelling the N particles under consideration. the representations in the space of N –particle wave functions are extremely useful in dealing with N –particle systems. xj . . . but in a diﬀerent order. xP (N ) ) .4) both permutations being obviously odd. .e. The permutations we need to consider for a system of N particles are the elements of the group SN . in particular. For the permutations P1 and P2 given above holds P1 = T (3.2). .. (9. . . i. 6) T (6. a given P can either only be presented by even numbers of transpositions or by odd numbers of transpositions. We can then state P2 = T (4. We will not discuss here at any depth the properties of the permutation groups SN .5) . . one can state P Ψ(x1 . 3) T (7. . k) = (4. 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. leaving all other particles unchanged. 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 = . Transpositions denoted by T (j. . and explicitly the fact that any P ∈ SN can be given as a product of transpositions T (j. k) of the two particles transposed. 5) (9. xP (2) . This property will be established now. P2 = . x2 . k) are characterized by the two indices (j. 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. is called a transposition.. For the following we will require only two properties of SN . 5) T (5. . rather only states which obey a certain transformation property under permutations are acceptable. One calls permutations of the ﬁrst type even permutations and permutations of the second type odd permutations. xP (j) . . A reader interested in the quantum mechanical description of N particle systems is strongly encouraged to study these representations. The latter factorization has the essential property that the number of factors is either even or odd. . (9.. . in notation (9. 5). .3) P2 which aﬀects only two particles. i. The group SN is then composed of two disjunct classes of even and of odd permutations. k). (j. . even though these properties. the set of all permutations of N objects.e. 5) in case of P2 .. not any state can be chosen as to represent the many– particle system.e.e. number k under the entry j of the ﬁrst row indicating that particle j is switched with particle k.240 Many–Particle Systems repeatedly like degenerate states.3) the j-th index in the second row. i. .

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

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

1: (a) Show that the states (9. . (9.. . nS ) holds S Ψ (N )|Ψ (N ) = j=1 B B δnj nj . . .15) obey the boson symmetry (9. . λ2 . . φλ2 (x1 ) φλ2 (x2 ) . nS ) and N = (n1 . . nS ). . . . . i. The fermion states (9..20) . n2 . x2 . x2 . φλN (x1 ) φλN (x2 ) . φλN (xN ) Here. . These states form an orthonormal basis. . for example. . .19) φλ1 (xN ) φλ2 (xN ) .e. x2 . . . correspond to a particular choice of placing N particles consistent with the occupation numbers (n1 . n2 . . for two sets of occupation numbers N = (n1 . . xN |ΨF (n1 . (b) Show that for the states (9. .. . . These states form an orthonormal basis. a molecule or a crystal. √1 N! (9. For any value nj > 1 two or more of the columns of the Slater matrix are identical and the wave function vanishes. . . nS ) and N = (n1 . .. . xN |ΨF (n1 . nS ) = √1 N! P ∈SN P P x1 . nS ) holds S ΨF (N )|ΨF (N ) = j=1 δnj nj . . n2 . Such wave function is x1 . . .λN ) (9. n2 .. .1: Bosons and Fermions 243 The wave functions (9.1. . . . the photons of the electromagnetic ﬁeld or the phonons in a crystal. electrons in an atom.λN ) on the rhs. .17) describe. . . .8). .18) this wave function can also be expressed through the so-called Slater determinant x1 . The important property holds that nj can only assume the two values nj = 0 or nj = 1.9). xN |ΨFock (λ1 . (9. . . . λ2 .e. i. . for example.17) Here the states (λ1 . . . . Wave Functions with Fermion Symmetry in the Occupation Number Representation One can construct similarly a wave function which obeys the fermion symmetry property (9. . .15) holds (9. .. . for two sets of occupation numbers N = (n1 . .16) Exercise 9.15) describe.16). . . . the integers nj denote the occupancy of the single–particle state |j . Using the identity governing the determinant of N × N –matrices A N det(A) = P ∈SN P j=1 Aj P (j) (9. . n2 . . .9. n2 . . n2 . nS ) = φλ1 (x1 ) φλ1 (x2 ) . .

. nj + 1. . .244 Many–Particle Systems The representation of wave functions (9. . .22) The operators aj and a† obey the commutation properties j [aj . . Employing the superindices N and N introduced above and using the orthonormality property (9. . . (9. . n ) and the N − 1-particle wave function ΨB (n . . These operators allow one to construct and manipulate many–particle wave functions while preserving permutation symmetry. . .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 . n .2: (a) Show that the states (9. . . nS ). For the adjoint operator a† holds j a† ΨB (n1 . nj − 1.17) obey the orthonormality property (9. . . . To prove that (9. since both the N -particle wave function B (n .16) one obtains (Aj )N = ΨB (N )| aj |ΨB (N ) = √ S N nj δnj nj −1 k=1 k=j δnk nk . . . (9.24) [aj . . .17) obey the fermion symmetry (9. . . guaranteeing the correct permutation properties of the many–particle states. . . . . . . Creation and Annihilation Operators for Bosons We consider the operator aj deﬁned through aj ΨB (n1 . . nj .15). . are the so-called operators of 2nd quantization. n ) are normalΨ 1 j 1 j S S ized according to (9.25) . Exercise 9. . 9. . n2 .9).2 Operators of 2nd Quantization An important tool to describe mathematically systems of many bosons or fermions. . .23) (9. (b) Show that the states (9. [a† . . nS ) = nj ΨB (n1 . . a† ] = 0 j k (9.1. .20). . . 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 .21) √ The factor nj appears here on the rhs. nS ) = j nj + 1 ΨB (n1 . . nS ) . . . . . . a† ] = δj k . . . .15). . nj . (9.15) and (9.21) we consider the matrix Aj corresponding to aj in the basis of many–particle states (9. . n − 1.22) follows from (9. nS ) nj ≥ 1 0 nj = 0 √ . . ak ] = 0 .

nS ) = 0 .e. a† ] = 1 For j = k follows similarly 1. The creation operators a† allow one to construct boson states ΨB (N ) from the vacuum state j |0 = |ΨB (n1 = 0.22) ˆ Nj |ΨB (N ) = nj |ΨB (N ) . . j (aj a† − a† aj )|Ψ(N ) = k k ( nj (nk + 1) − (nk + 1)nj )|Ψ(n1 . . . the properties (9. . .e. nS ) N |ΨB (n1 . ..29) |ΨB (n1 . . n2 . .27) Equations (9.24). .31) This operator is diagonal in the occupation number representation.2: Operators of 2nd Quantization Let A† be the matrix adjoint to Aj .30) Of particular interest is the operator ˆ Nj = a† aj .. n2 .28) (9. nS ) . . . (9. n2 . .33) is called the particle number operator since. One can readily show using (9. . . . (9. for basis states |ΨB (n1 . . . 9. . . n2 . .21) and vice versa. 9.34) = . . . (9.9. . n2 . .22) follows from (9. j j Since this holds for any |Ψ(N ) it follows [aj .27. One refers to Nj as the occupation number operator. . . j (9. (9. The related operator S ˆ N = j=1 ˆ Nj (9.23) follow in an analogous way.26) From this result one can immediately conclude that (9. nS = 0) as follows S (9. the eigenvalues of Nj are the occupation numbers nj .29) yield the commutation relationship (9. nS ) = j=1 a† j nj nj ! |0 . obviously. nk + 1 . It holds for j = k (aj a† − a† aj )|Ψ(N ) = (nj + 1 − nj )|Ψ(N ) = |Ψ(N ) . nj − 1 . (9.21. n2 = 0. .32) ˆ ˆ i. . nS ) = j=1 nj |ΨB (n1 . 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 .24). We will prove the commutation properties (9. S ˆ N |ΨB (n1 . . i. nS ) = |ΨB (N ) . . .

21. xN |ΨF (n1 . nS ) = . . . .. .36) Creation and Annihilation Operators for Fermions We want to consider now creation and annihilation operators for fermions. xN |ΨF (n1 . in general.. the fermion wave function changes sign when one exchanges the order of the occupancy. x2 .. .38) . . showing that the property holds for nj = 0 and. (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 .24) and the property aj |0 = 0. .. . operators which can alter the occupancy of the wave function (9..23. . Obviously. . x2 . . nj . it also holds for nj + 1. . nj+1 .2. 9. i. . nj (9.. of (9. . x2 . . Exercise 9. φj (x1 ) φj (x2 ) . f (a† )] j [a† . .29. nS ) = − x1 .24) and aj |0 = 0 imply that the the properties (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. nj+1 .19) without aﬀecting the fermion symmetry (9. As long as one starts from a wave function with the proper boson symmetry. one needs to apply only the commutation propertes (9.30). nj+1 .. . . i.2. (n − 1)! (9. from the vacuum state |0 or states |ΨB (N ) .2: Show that the boson operators a† and aj satisfy j [aj . since they are induced through the algebra of a† and aj . .22.h. . x1 . . . . 9. . .246 Many–Particle Systems In using the boson creation and annihilation operators.e. .37) (9. .23) hold for the state deﬁned through (9.s. √1 N! ..17. To prove this property we notice that the l. φj (xN ) φj+1 (xN ) .38) corresponds to x1 .35) and where the derivative operation is deﬁned through ∂f (A) = ∂A ∞ n=1 1 f (n) (0) An−1 . . . . j Exercise 9. 9.9). xN |ΨF (n 1 .. 9. nj+1 = 1. . . . nj . one does not need to worry ever about proper symmetries of wave functions. nS ) . Prove this by induction. 9. . in case that the single particle states |j and |j + 1 are both occupied. 9. .e. .1: The commutation relationships (9. e.. .39) . . . φj+1 (x1 ) φj+1 (x2 ) . in case it holds for nj .g.. . .23.

. x2 . . .19). in order to be consistent with this convention.43) . We are now ready to deﬁne the annihilation operator cj for fermions in the single–particle state |j as follows cj |ΨF (n1 . . . i. . vanishes. n3 . . . that occupancies are always ordered according to a monotonous increase of the single–particle state index. x2 |ΨF (n1 = 0.2: Operators of 2nd Quantization The r. . n6 = n7 = . φj (x1 ) φj (x2 ) .42) i.. . This requires one. . at the ﬁrst column of the Slater determinant (9. . . it is necessary to specify for a fermion wave function the order in which the single– particle states |λj are occupied. . . nS ). n2 = 1. to move the particle to the ﬁrst position (to be annihilated there) or to move it from the ﬁrst position to its canonical position (after it had been created at the ﬁrst position). . the expressions (9. xN |ΨF (n1 . .h. the rhs. . . . nj+1 .. nj → 0. . n5 = 1. nj . One adopts the convention that particles are created and annihilated by these operators always at the ﬁrst position of the wave function.e.e. φj+1 (xN ) φj (xN ) . .s. . 2. i. . Obviously. (9.40) . . .38) reads − x1 . (9. For this purpose one must adhere to a strict convention: the labelling of single–particle states by indices j = 1. . . n2 .. . (9.. nj = 0. .39) and (9. nS = 0) φ2 (x1 ) φ5 (x1 ) 1 1 = √2 = √2 [φ2 (x1 )φ5 (x2 ) − φ5 (x1 )φ2 (x2 )] . nS ) qj states occupied In case that the single particle state |j is not occupied.9.40) are identical. We will be adopting below the notational convention that qj corresponds to occupancies N = (n1 . . nj+1 . φj+1 (x1 ) φj+1 (x2 ) . . . . . . n2 . . nS ) as follows: j−1 qj = k=1 nk . .e. .. n3 = 0. . 247 − √1 ! N .e. n2 . . however. of (9. i. nj−1 ... nS ) = . particles are being created or annihilated. A proper example is the two particle fermion wave function x1 .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.. .. nj . .19). . This change of position is connected with a possible sign change (−1)qj where qj is deﬁned for a given N = (n1 . . ... qj is the number of states |k with k < j which are occupied in a fermion wave function. nS ) = nj (−1)qj |ΨF ( n1 . n2 . 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. φ2 (x2 ) φ5 (x2 ) (9. . Because of the property of the determinant to change sign when two columns are interchanged. at which column of the Slater determinant (9. n4 = 0.

(9. We ﬁrst show that (9. of (9. . 9. . It holds c† cj |ΨF (N ) = c† (−1)qj nj |ΨF (n1 . 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 .49) The derivation involves realization of the fact that a non-zero result is obtained only in case nj = 1. . nS ) = j (−1)qj (nj + 1))(−1)qj (1 − nj ) |ΨF (N ) = (1 − nj ) |ΨF (N ) (9. .47) Let C† be the matrix adjoint to Cj . . . Let Cj be the matrix corresponding to the operator cj in the basis of fermion states (9.46). ck ]+ = 0 . nS ) = j j (−1)qj (1 − nj + 1)(−1)qj nj |ΨF (N ) = nj ΨF (N ) (9. We consider (9. nS ) qj states occupied The operators thus deﬁned obey the commutation properties [cj . n2 .48) From this follows (9. .248 The operator c† adjoint to cj exhibits the following property j c† |ΨF (n1 . . .50) . 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 . (9. The anti-commutation properties (9.46) [cj . nj−1 . We will prove now the anti-commutation property (9.19). . c† ]+ = 0 j k (9.45) follow in an analogous way and will not be derived here. otherwise.48) vanishes. the last factor on the rhs. .44) (9. B]+ = AB + BA.44) follows from the deﬁnition (9. c† ]+ = δj k k where we have used the deﬁnition of the so-called anti-commutators [A. .17.43). . nj → 1. nS ) = j Many–Particle Systems (−1)qj (1 − nj ) |ΨF ( n1 .44). nj → 1 . n3 . We have used here the deﬁnition qj = k<j nk along with qj = qj . Similarly one obtains cj c† |ΨF (N ) = cj (−1)qj (1 − nj ) |ΨF (n1 . nj → 0 . . . . [c† .45) (9. . nj .46) ﬁrst for the case j = k. nj+1 . n2 .

s. i. coordinates x and index j into a space part and a spin part.. in the basis |ΨF (N ) . nλ2 .e.55) is called the particle number operator. . We j will often separate the states |j .49) shows that the operator ˆ Nj = c† cj j (9. . Correspondingly. 1 In the following we will apply fermion operators c† and cj only to electrons which carry spin 2 . of this equation we meant to indicate only those N occupation numbers nλj which are non-vanishing. nk = 1 .56) Here we have made use of cj cj = 0 and c† c† = 0 which follows from nj ∈ {0. . an operator cj must be on the left of an operator ck for j < k. . S (9. nk = 1 . . 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. j → j.2: Operators of 2nd Quantization (9. i.46). Equation (9. .50) yield ( cj c† + c† cj ) |ΨF (N ) = |ΨF (N ) . . . x|j → φj (r) | 1 σ . 1}. ( cj c† + c† cj ) |ΨF (N ) = 0 . On the r. with eigenvalues equal to the occupation numbers nj ˆ Nj |ΨF (N ) = nj |ΨF (N ) ˆ Nj . nS ) k c† cj |ΨF (N ) = (−1)qj +qk −1 nj (1 − nk )|ΨF (n1 .e. hence.53) is diagonal in the occupation number representation.51.29)) |ΨF (.. . i.51) (9. nλN .46). . nj = 0 .52) One can obtain similarly the same relationship in the case j > k.) = c† 1 c† 2 · · · c† N |0 . .h. λ λ λ (9. Since (9. . the creation operators must operate in the proper order. 2 (9.49) and (9.9. σ 1 (σ = ± 2 ) . . j j j j j j j (9. is referred to as the occupation number operator.52) hold for any |ΨF (N ) one can conclude (9. hence. .57) On the l. nS ) k and. k k 249 (9. .s.e. for which purpose one needs to envoke only the algebraic (anti-commutation) properties (9. . nj = 0 .. ˆ It is an interesting exercise to demonstrate that Nj only has eigenvalues 0 or 1. nλ1 . . . j j For j < k one obtains cj c† |ΨF (N ) = (−1)qj +qk nj (1 − nk )|ΨF (n1 . 9. 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 . .58) . .h.54) ˆ N = j=1 ˆ Nj (9.

60) is that the sum over particles in (9. ˆ r|f |s = 1 ∗ ˆ d3 r ψr (r. . 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 ) . σr ) f (x) ψs (r.F (N )|G|ΨB. .17). ˆ . . r = 1. σs ) (9.6) above.F (N )|F |ΨB. j = 1. The operators f (x) which constitute F are called the associated generic single–particle operators1 . The operators g (x.k=1 j=k (9. .62) These matrix elements can be obtained by chosing the operators and many-particle states in the position representation. 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. 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 (9.3 One– and Two–Particle Operators Deﬁnition Operators acting on N particle systems of the type N ˆ F = j=1 ˆ f (xj ) (9.15. x ) which constitute G are ˆ called the associated generic two–particle operators. (9. S of single–particle spinorbital states (see page 241) and the operators are speciﬁed through the associated generic operators ˆ f and g . xN .61) is not a one particle operator as long as the N operators rj (x). . (9.F (N ) . . . xk ) ˆ j.59.59. These operators had been introduced already in Eq. ˆ ΨB. The essential aspect of deﬁnition (9. Since the many–particle states are built-up from a basis |r . N are not all identical. respectively – which are all identical.15.59) ˆ ˆ are called single–particle operators.60) ˆ correspondingly are called two–particle operators. An operator N ˆ R = j=1 rj (xj ) ˆ (9. We adopt a similar expression to distinguish two–particle operators G and g . xk . .60) involves generic operators – acting on xj and on xj . 2.17) and integrating over all particle space-spin coordinates x1 . The action of the operators boson and many–fermion states |Ψ is obviously described through the many–particle state matrix elements ˆ ΨB.59) and (9. . 9. 2.63) ˆ ˆ This expression has been speciﬁed for the purpose of these notes to distinguish F and f and is not common ˆ terminology.e.59) and over pairs of particles in (9. Operators of the type 1 ˆ G = 2 N g (xj . x2 .F (N ) . namely. i..250 Many–Particle Systems 9. (9. 9. respectively.

69).67) ˆ In this case the generic single–particle operator. u = s.64) r.68) Another single–particle operator is the one–particle density operator N ρ(r ) = ˆ˜ j=1 δ(r − rj ) .66) Here we adopt the notation for the Laplacian ∂2 ∂2 ∂2 ∂2 = + + . An example for a two–particle operator is the Coulomb repulsion operator 1 ˆ V = 2 N j. σr )ψs (r . s|ˆ|t. g g (9.63. u. σu ) . u = g d3 r ∗ ∗ d3 r ψr (r. The matrix elements of the generic operator ˜ in the one–particle basis are ˆ rσ|f |sσ = δσσ d3 r φ∗ (r) δ(r − r) φs (r) = δσσ φ∗ (r) φs (r) δσσ ˜ ˜ r r ˜ (9.69) ˆ In this case the generic operator is f (x) = δ(r − r). 9. ∂r 2 (9.59). t .66) and (9. s|ˆ|t. s and t. 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 . is t(x) = (− 2 /2m)(∂ 2 /∂r 2 ). (9.65) This symmetry will be exploited repeatedly below. ˜ (9. one cannot switch the indices of one pair individually. u which follows directly g from the deﬁnition (9.62) in terms of the matrix elements of single–particle states (9.64). i. 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) . σs ) g (x. σt )ψu (r .k=1 j=k q2 |rj − rk | (9. Notice..3: One– and Two–Particle Operators r. ˆ 251 (9.9. s|ˆ|t. ∂rj2 (9.64) We like to note an important symmetry of the matrix element r. x ) ψt (r. r|ˆ|u.70) Both operators. 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. according to (9. are spin-independent as is evident from the factor δσσ . ∂r 2 ∂x2 ∂x2 ∂x2 1 2 3 (9. We will show that the boson and fermion creation and annihilation operators serve this purpose. that the symmetry implies that one can switch simultaneously the pairs of indices r.71) .e.

α s r.+ + S1. u.252 Many–Particle Systems which is also spin-independent. β|ˆ3 |s.e. (9. i. The operator N S = j=1 Sj (9. The generic operator is s12 = S1 · S2 = ˆ 1 1 S1.3 S2.74) = N j=1 Sj = 2 N Sj · Sk j.73) is a one-particle operator. s. . Sj. β|ˆ− |s.+ .76) the single–particle state matrix elements of which are r. u.74) and (9.72) As a ﬁnal example of one– and two–particle operators we consider operators of total spin. σ. σ .3 2 2 (9.− . In this case the generic operator is v (x. s = 1 σ. σ. α s r.75) is a two–particle operator. β s r. however.77) δσ σ ( δσσ − δ−σσ ) The operators (9. σ ˆ + 1 δ− 1 σ δ 1 σ δ 1 σ δ− 1 σ 2 2 2 2 2 = δrt δsu + 1 4 δσσ 1 2 δ 1 σ δ− 1 σ 2 2 δ− 1 σ δ 1 σ 2 2 + . The ˆ matrix elements of the generic operator in terms of single–particle states are ˆ r.− + S1.. Sj.k=1 (9. x ) = q 2 /|r1 − r2 |. β s The operator S 2 = δrs = δrs = = + − 1 2 1 2 δrs δrs (9. Sj is the spin operator for particle j with components (in the spherical representation) Sj. not in the strict sense of our deﬁnition above since the restriction 2 j = k does not apply in the summation. σ | t |t. However. α|ˆ3 |s. The non-vanishing matrix elements of the spherical components of s are (α = + 1 . α|ˆ+ |s. σ | s12 |t. s.77) are orbital-independent as evidenced by the factors δrs and δrt δsu in the respective formulas. The generic operator is given through the Pauli matrices. β = ˆ 2 2 1 −2) r.− S2. σ . 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.+ S2. one can obviously extract the term j Sj to be left with a two–particle operator in the strict sense of our deﬁnition. Here.3 .

pp. respectively.176 where the matrix elements of fermion states can be found.F (N ) have the same values as the respective matrix elements (9.63. 9.82) ΨF (N )|c† cs |ΨF (N ) = r cr 2 (9.s=1 ˆ r|f |s a† as bosons r ˆ r|f |s c† cs fermions r r.80) (9. 9.81) For the latter matrix elements simple rules can be derived from the algebraic properties of the boson and fermion operators (9..46).78) 1 2 g (xj . r ΨB (N )|a† a† at au |ΨB (N ) r s ΨF (N )|c† c† ct cu |ΨF (N ) r s bosons fermions (9.60) as follows N ˆ f (xj ) → j=1 N 1 2 1 2 S r.45. To show that the resulting values of the matrix elements (9.79) It is crucial to notice in the expression for the fermion two–particle operator that the order of the annihilation operators in (9. respectively.23. r ΨF (N )|c† cs |ΨF (N ) .59) and G in (9. g and not ct cu . s|ˆ|t.9. r S Ψ (N In case of r = s one obtains F )|c† cr |ΨF (N ) r = nr r=1 δnr nr . 9.s. The Matrix Elements ΨF (N )|c† cs |ΨF (N ) r We consider ﬁrst the case r = s.79) is opposite to that in the matrix element r. Expressions (9.64) one replaces the operators F in (9.t. 9. u a† a† at au bosons g r s † † r.62) are correct one needs to compare the result derived with conventional derivations of the matrix elements2 .3: One– and Two–Particle Operators Deﬁnition in Terms of Creation and Annihilation Operators 253 In order to express many–particle state matrix elements (9.62).s. u cr cs cu ct g fermions (9.79) have the following implication: The operators (9. xk ) → ˆ j.62) through single–particle state matrix ˆ ˆ elements (9.s=1 S r.79) in the basis of many–particle states |ΨB.83) t=1 t=r. (9. s|ˆ|t.e.171 and pp. s|ˆ|t. .u=1 S r. The matrix elements are then actually those of the number operator c† cr which is diagonal in the occupation number representation. u . u as deﬁned in (9.s ΨF (N )|cs ΨF (N ) = nr (−1)qr ns (−1)qs δnr −1 nr δns ns −1 δnt nt We refer the reader to Condon and Shortley’s ‘Theory of Atomic Spectra’.u=1 (9.78) and (9.k=1 j=k S r.t.63) and (9.78. i. In order to determine these matrix elements one needs to evaluate ﬁrst the matrix elements ˆ of the generic operators r|f |s r.24) and (9. and then in g a second step the matrix elements ΨB (N )|a† as |ΨB (N ) . namely cu ct . s|ˆ|t.64). These rules will be provided below only for the case of fermions.

1 2 S r s|ˆ|s r c† c† cr cs . G = G0 + G1 + G2 . s.i. namely. t.t. Comparision with the results in Condon&Shortley (pp. t.d. t=u 2 = 0. s. t. c2 = 0 r (9.d. ˆ The ﬁrst contribution G0 is 1 ˆ G0 = 2 S r s|ˆ|r s c† c† cs cr + g r s r.u r=s. where each contribution corresponds to one of the three cases mentioned. i. are ˆ ˆ ΨF (N )|F |ΨF (N ) = (1 − ns )(−1)qs nt (−1)qt s|f |t (9. u) are included in the summation for which all indices are diﬀerent.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..86) For the combination of indices r.s=1 a.87) Here and in the following we denote by ‘a.s=1 a. Case 2 all four indices are diﬀerent.i. ˆ ˆ Nr Ns . Starting from the deﬁnition (9.84) The oﬀ-diagonal elements with nr = nr except for r = s. in which case ns = ns + 1 and nt = nt − 1 holds. s.79) and using c† r we can write S 1 ˆ G = rs|ˆ|tu c† c† cu ct g r s 2 r. (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. g r s r.i.88) . 171) shows that the operator deﬁned in (9.59) when the matrix elements are determined between Slater determinant wave functions. t).78) does yield the same results as the operator deﬁned in (9.53) 1 ˆ G0 = 2 S r s|ˆ|r s − r s|ˆ|s r g g r. (9. u in this sum three possibilities exist Case 0 two of the four indices are diﬀerent.d. (r. (r. for which holds r = s. those for which N and N diﬀer in more than two occupation numbers.’ (all indices diﬀerent) that only tuples (r. r = t etc. t.e.e. Case 1 three of the four indices are diﬀerent. ˆ ˆ ˆ ˆ Accordingly. i.86) into three contributions.82. s)..254 From (9. The anti-commutation property (9.85) All other matrix elements vanish.s.45) yields together with the deﬁnition of the occupation number operator (9.d.s=1 a. we split the sum in (9. s.i. 9.

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

3: 9. One obtains ˆ ΨF (N )|G|ΨF (N ) = ± (−1)qs +qr +qu +qv (1 − ns )(1 − nr )nu nv r. t for which holds ˆ ns = ns + 1 and nt = nt − 1. nu = nu − 1. s|ˆ|r.89).91).73) and S 2 given in (9. i.95) from (9. s|ˆ|v.75) which are. 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 (9.90).3.92).3. One obtains ˆ ΨF (N )|G|ΨF (N ) = (−1)qs +qt (1 − ns )nt r r=s.1: 9.3. u < v or s > r. We like to express these operators through ˆ fermion creation and annihilation operators.. t. respectively. Derive the non-vanishing matrix elements (9. the ‘−’-sign applies for s < r. for ˆ non-vanishing matrix elements ΨF (N )|G|ΨF (N ) at most four occupation numbers nr and nr can diﬀer. u > v. ˆ ˆ Show that the matrix elements of F and of G conserve particle number. particle number is conserved. u < v. given in (9. Only G2 . given in (9. s|ˆ|u. u g g (9.90) from (9. contributes in this case. s|ˆ|t.94) Case 2 This case covers oﬀ-diagonal elements with nr = nr except for r = s. nt = nt + 1. t − r.e. one–particle and two–particle operators. u > v or s > r.5: Derive (9.76).74) and (9. In particular.92) from (9.3. r g g (9. v − r.3. Spin Operators ˆ ˆ In this section we will brieﬂy consider the spin operators S given in (9. The matrix elements of the corresponding generic operators are (9.95) In the latter formula the ‘+’-sign applies for s < r. Furthermore.t nr r. r=1 r=1 Exercise Exercise Exercise Exercise Exercise 9. All matrix elements which are not covered by the three cases above vanish. Derive (9.256 Many–Particle Systems Case 1 This case covers oﬀ-diagonal elements with nr = nr except for r = s. Derive (9.96) . v for which holds ˆ ns = ns + 1.2: 9.4: 9.92).62) for bosons. Only G1 . u. nv = nv − 1. it holds S nr = S nr . contributes in this case.

s|ˆ|u.96) and (9. atoms and nuclei.s=1 σ 1 ˆ rσ r|t|s c† csσ + 2 S r.s.u=1 σ.t. 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.96.9.σ (9. v r s 2 r.6: Derive (9. which represents the interactions between particles. Actually. the systems investigated involve a macroscopic number of components such that statistical mechanical approaches are envoked. t. s|ˆ|t. however.s. superconductivity. 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.3. e. the concerted behaviour of interacting systems can deviate qualitatively from that of systems made up of independent components. In the latter case the indices r. v c† c† cu ct . u c† c† cuσ ctσ v rσ sσ r. the description of many–particle systems.s=1 r.4 Independent-Particle Models In the remaining part of this chapter we will apply the technique of operators of 2nd quantization to the description of many–fermion systems as they arise. The outcome of these studies is that even when interactions between components (particles) are simple. the study of problems governed by Hamiltonians of the type (9. evaluation of their stationary states and excitation energies. Often. e.97) is over the orbital states. rα sβ (9. molecules.97). in crystals.u S (9. 9.98. u refer only to the orbital part of the single–particle basis. that the two–particle operator representing interactions between particles induces a variety of interesting phenomena.g. If it were not for the two–particle contribution to the Hamiltonian.g. 9. The great generality of the concepts developed in the context of many–particle systems can be judged from the fact . 9..4: Independent-Particle Models ˆ For S 2 one obtains 1ˆ 1 ˆ S2 = N + 2 4 S S 257 ˆ ˆ ˆ ˆ (Nrα − Nrβ )(Nsα − Nsβ ) − r.97) The summation in (9. the spin part of the single–particle states is represented by α = 1 and β = − 1 .. Unfortunately.99) will be used. would be a simple exercise in linear algebra. 2 2 Exercise 9. the two–particle operator makes the description of many–particle systems a very hard problem.99) has preoccupied an important part of all intellectual eﬀorts in Theoretical Physics during the past ﬁfty years. but also ordinary phenomena like freezing and evaporation. s.s=1 c† c† crβ csα .98) In many cases the Hamiltonian is spin-independent and the equivalent notation S H = r. The fortunate side of this is. In fact. Examples is the laser action.t.

. a Hamiltonian without a two–particle operator contribution accounting for interactions among the particles. electrons and nucleons. New York. namely of neurons3 . 1989) .258 Many–Particle Systems that today these concepts are providing insight into the functioning of biological brains. m = 1. Chemistry essentially would know only one element and Life would not exist. Amit (Cambridge University Press. water-ﬁlled jar all electrons of water turn their fermion nature into a boson nature. The electronic properties of atoms with diﬀerent numbers of electrons would diﬀer little. S}.102) U ∗rm Urn = δmn . ranging from nuclei to the molecules of living systems .. If it were’nt for the fermion nature. Exercise 9.100) in a form in which the factors c† csσ become rσ occupation number operators. In such a representation Ho is diagonal and the stationary states can be stated readily. Before continuing one may ﬁnally point out that the material world as we know it and as it establishes the varieties of natural substances. Transformation to a New Set of Creation and Annihilation Operators Our aim is to represent the Hamiltonian (9..s=1 σ ˆ rσ r|t|s c† csσ (9.103) An account of some of these eﬀorts can be found in Modelling Brain Function – The World of Attractor Neural Networks by D. i. The Hamiltonian of such system is S Ho = r.100) ˆ ˆ We will denote the matrix elements of t as r|t|s = trs . would not exist. i. We will restrict our description in the following to systems composed of an even number (2N ) of particles and to spin-independent interactions.e. r=1 3 Urm U ∗sm = δrs m=1 (9. all material systems would condensate into states which would depend little on particle number. ultimately depend in a crucial way on the fermion character of its constituent building blocks.e. .4. . 2.101) The unitary property of Urn U† U = U U† = 1 1 or S S (9. (9.1: Imagine that in a closed. What might happen? Independent-Particle Hamiltonian A many–fermion system governed by an independent-particle Hamiltonian.J. where S |m) = r=1 Urm |r . The Aufbau principle. To alter the representation of Ho we apply a unitary transformation U to the single–particle states |r to obtain a new basis of states |m). according to which nuclei and atoms change their qualitative properties when they grow larger. another prototypical systems of many strongly interacting components. {|m). can be described in a rather straightforward way.

the anti-commutation properties (9.113) Urm Usn [crσ .101) of states |r .112) .e. In a basis of states N d† j σj |0 n j=1 (9. (9. c† ]+ = δσσ [dmσ ..46).h. hence. using (9.109) The operators (9.103) allows one to express [c.100) yields ˜ Ho = tmn d† dnσ mσ where d† nσ = r=1 S S (9.s=1 ˆ U † tU mn (9.110) mσ nσ [dmσ .104)] S S ∗ Urm d† mσ m=1 c† rσ = .4: Independent-Particle Models allows one to express conversely |r in terms of |m) S 259 |r = m=1 ∗ Urm |m) .108) These operators describe the creation and annihilation operators of fermions in states |n) which are linear combinations (9. expand S trs = mn ∗ ˜ tmn Urm Usn (9. [d† .106) which together with (9. accordingly. The unitary property (9. the expressions drived above for the matrix elements of one.f.111) can be written.110) can be demonstrated similarly. of N -Fermion states in which single particle states |nj ) as deﬁned in (9.46) as c† and crσ . rσ namely. csσ = n=1 Usn dnσ .s.101) are occupied.108) and (9. behave exactly like the operators c† and crσ behave for states |ΨF (N ) . d† ]+ = sσ nσ r. d† ]+ = 0 (9.107) Urn c† rσ . The l. ∗ ∗ Urm Urn . (9. of (9. dnσ ]+ = 0 . . (9.111) and. nσ We demonstrate here the anti-commutation property (9.45.s r The unitarity property r ∗ Urm Urn = δmn yields immediately identity (??). i.104) One can. 9.and two-particle operators hold in an analogous way for d† and dnσ .105) where ˜ ˆ tmn = (m|t|n) = S ∗ trs Urm Usn = r. the operators d† and dnσ . for nσ rσ example. [dmσ . they are Fermion creation and annihilation operators. (9. dnσ = r=1 ∗ Urn crσ (9.111). d† ]+ = δmn δσσ nσ (9.108) obey the same anti-commutation properties (9.9.

USn ).116) which follows from (9.57) to the case of 2N particles. .. i.101) gives us the freedom to choose the matrix (9. − 1 . σj ) = (nk .. 1 and |β = | 1 . .e. . . .e. a simple matter to nσ state many–particle states in the new representation. σ2 .106) is equivalent to S trs Usn = s=1 n Urn ∀r.106) and the unitary property of Urn . σ2N . j j (9.e. The condition (9.e. i. It is. n = 1.119) In this state the N lowest orbital eigenstates of trs are occupied each by an electron with spin |α = | 1 .114) together with (9. Vn = (U1n . i. is N |Φo = j=1 d† α d† β |0 .260 Diagonal Representation Many–Particle Systems The transformation (9. Eigenstates of (9. is obtained. S (9. The 2N fermion state 2N d† j σj |0 n j=1 where (nj . . The corresponding eigenvalues n . ˜ tmn = In this case Ho has the desired form S n δmn . r = 1. n2 . transformation matrices Urn and energy values n . hence. . − 1 is called a closed 2 2 2 2 . the so-called ground state. 2. . . ) = j=1 nj . (9. Before proceeding we like to address the issue how the new representation. σk ) for j = k (9.115) can be stated immediately since any many–particle wave function in the occupation number representation is suitable. 2. . 2 2 2 2 i. (9. 1 as well as a fermion in a spin state | 1 .115) with eigenvalues 2N E(n1 .106) diagonal.. n2N . the associated ( properly normalized) eigenvectors Vn are the columns of (Urn ).114) Ho = n=1 † n dnσ dnσ (9.115) ˜ and involves solely occupation number operators N nσ = d† dnσ . S are T real. a fermion in a spin state | 1 . This equation poses the eigenvalue problem for the hermitian matrix (trs ).117) are eigenstates of (9.. U2n . . . We apply for this purpose (9. σ1 . A non-degenerate orbital state which is occupied by two fermions.118) Ground State In case of an ordering m < n for m < n the state of lowest energy.

124) . is a singlet state. We will demonstrate now that this property endows the ground state with total spin zero.n=N (9. namely. i.e. Alltogether. |α. (9. α = 2 |β.n=1 d† d† dmβ dnα mα nβ (9. However.n=1 d† d† dmβ dnα mα nβ (9. the states diﬀer in their spin character. the third term − S m. both with eigenvalues 2 ˆ S 2 |α. energy above the ground state. 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. Obviously.97). of ∆E = and |β. One can conclude immediately that the ˜ ˜ ˜ ˜ (N mα − N mβ )(N nα − N nβ ) does not give any contribution second term in (9. The action of the operator (9..100).e. β ˆ are eigenstates of S 2 given in (9.9. α |α. The spin operator in the present representation of single-particle states is 1 1ˆ ˆ S2 = N + 2 4 S S ˜ ˜ ˜ ˜ (N mα − N mβ )(N nα − N nβ ) − m.121) All four states have the same excitation energy. β .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.4: Independent-Particle Models 261 shell. 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. α |β. namely. The ground state has only closed shells. α . the states with energy closest to the ground state are those in which a particle is promoted from the highest occupied state |N ) to the lowest unoccupied state |N + 1). Similarly. 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. ˆ S 2 ||β.n=1 † † if either |m) or |n) are closed shells. The two states |α..n=1 cmα cnβ cmβ cnα does not give a contribution if m = n and either |m) or |n) are closed shells.n=1 m. β = 2 |α.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. There are four such states. In case of m = n it gives a contribution ‘1’ for each closed shell which cancels the contribution of the ﬁrst term. 1 S 4 m.n=1 m. i.120).120) is particularly simple for closed shells. ˆ one can conclude that |Φo is an eigenstate of S 2 with eigenvalue zero.97). Excited States We want to construct now excited states for the system governed by the independent-particle Hamiltonian (9.120) where the occupation number operators refer to the single-particle states |m). β |β. α N +1 − N .

.129) The symmetry of the Hamiltonian suggests to choose a new representation deﬁned through the operators (r = 1. ˆ Exercise 9. 2N which are located on a ring. One can. 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 . 2..125) ˆ Contributions to S 2 acting on states |N + 2 .n=N Many–Particle Systems N +1 d† d† dmβ dnα . ˆ 3 on the two states produces . 2N ) drσ = √ 1 2N 2N e−irnπ/N cnσ n=1 (9. m |Φo m m m γσ. . do not need to be considered since they ˆ give vanishing contributions. . ˆ The remaining states |α.126) which follows from the occurence of squares of fermion operators which. . We identify the states |2N + 1 = |1 to avoid cumbersome summation limits. β .262 and ˆ Σ3 = − m. α ± |β. one with eigenvalue zero (singlet) and one with eigenvalue ‘2’ (triplet). r = 1.2: Construct four operators O ˆ O ˆ such that O ˆ S3 . The action of Σ N +1 ˆ Σ3 |α. α are. also eigenstates of Σ2 .127) ˆ are triplet and singlet states. |β. for |α. β both with eigenvalues 1.122) is correct. |N + 3 . conclude that (9. β qualify as eigenstates.σ dN σ N σ. of course. etc.4. for example. β and |β. β =− m. however.130) .e. (9. α . . 2.σ = (9.100) outlined above we consider a system of 2N electrons which move in a set of atomic orbitals |r . . vanish. ˆ An eigenvalue ‘2’ of the spin operator S 2 identiﬁes the states |α. apropriate eigenstates of the operators S 2 and Example: 2N Independent Electrons on a Ring In order to illustrate the procedure to obtain eigenstates of one-particle Hamiltonians (9.128) The cyclic property of the system is expressed through c2N +1 σ = c1 σ . α and |β. c† +1 σ = c† σ 1 2N (9. α as triplet states.n=N d† d† dmβ dnα d† +1α dN β |Φo = 0 mα nβ N (9. hence. . The same result holds for |β. . mα nβ (9. the linear combinations 1/2(|α. The two states |α. β are not eigenstates of S 2 . of course.σ d† +1. . The system is assumed to posses a 2N –fold symmetry axis and interactions connect only orbital states |r → |r ± 1 . i.

134) which. rσ (9. indeed.131) The (9. .137) where 1 ˜ trs = e−irπ/N 2N Using (9. 4N. 9.139) −2t cos r=1 σ rπ N d† drσ .132) gives for r = 0.4: Independent-Particle Models d† = √ rσ 1 2N 2N 263 eirnπ/N c† nσ n=1 √ 1 eirnπ/N .109)] ¸ cnσ = √ 1 2N 2N eirnπ/N drσ r=1 2N (9. thereby.132) an identity which can be derived employing the well-known expression for a ﬁnite geometric series m as = a s=1 1 − am . 1−a (9. 2N (9.133) Application to the l. (9.9.135) c† nσ 1 = √ 2N e−irnπ/N d† rσ r=1 (9.136) ˆ and.h. n=1 (9. of (9.140) The Hamiltonian is now diagonal and the construction outlined above can be applied. eisπ/N 1 − e2N isπ/N 1 − eisπ/N (9. 2N. .s=1 σ ˜ rσ trs d† dsσ (9. (9. vanishes for integer s. one obtains the new representation of Ho 2N ˆ Ho = −t r.130. .132) one can conclude 2N ei(s−r)nπ/N + eisπ/N n=1 1 2N 2N ei(r−s)nπ/N .s.131) constitute a unitary transformation U deﬁned through Urn = unitarity property follows from 2N eirsπ/N = 0 r=1 for 0 < s < 2N .f. . . One can then express[.138) ˜ trs = from which follows ˆ Ho = eirπ/N + e−irπ/N 2N δrs (9.

in states which are constructed according to the ‘Aufbauprinzip’ in a manner which neglects the mutual repulsion between electrons.e.. Since independent-particle behaviour is so universal one may wonder why the presence of particleparticle interactions. We will present in this and the following section a method which constructs such optimal single-fermion states. if one can choose the singleparticle states such that the residual perturbations are small. i. assuming that electrons move independently from each other in so-called bands. The reason is mainly connected with the fact that most many-fermion systems are found in their ground state or removed from it through low energy excitations. The properties of molecules can be understood largely in terms of models which place electrons into so-called molecular orbitals in which the electrons move as if they were not interacting with other electrons. These states do not altogether neglect the pair interactions.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 ..e. The regularities of the elements can be rationalized in terms of electrons moving in hydrogen atom-type orbitals. 2N with interactions between neighbouring (n. .e.141) State the excitation energy ∆E.5 Self-Consistent Field Theory Observations of many-fermion systems show that properties of such systems often can be explained to a surprisingly good degree on the basis of the assumption that the fermions move like independent particles. Hence. one expects that mean-ﬁeld descriptions should be rather suﬃcient. The slow decay of the interaction makes the motion of any one electron rather independent of the exact position of most other electrons. .. and that independent-particle behaviour surfaces mainly because it applies well to the ground state. does not spoil it.e. if one places fermions into an independent-particle ground state of the type (9. 2. Of course. A similar principle accounts approximately for properties of nuclei as described by the nuclear shell model. the fermion nature restricts the possibility for perturbations on the system. A second reason why independent-particle models can be successful is that the pair interaction for electrons is the slowly decaying Coulomb interaction.e.g. i. rather they assume that each particle experiences the average interaction due to the fermions frozen into the ground state. n = 1. e.4. in particular. i.264 Many–Particle Systems Exercise 9. properties of many-fermion properties deviant from such simple description is then of as much interest as independent fermion descriptions are useful. 9. i. valence bands or conduction bands. The most famous example is the periodic table of the elements. those with energy above that of the (energetically) highest occupied single-particle state.119) then perturbations due to pair interactions according to the Pauli exclusion principle can only involve independent particle states not occupied in the ground state. Many electronic properties of solids can be understood in a similar way... i. the everpresent Coulomb repulsion between electrons. 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 . (9. There are two reasons why this is so: ﬁrst.. . .

s|ˆ|t.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.t. . We will adopt a second set of single-particle states {|m . r = 1.s=1 σ r.u=1 σ.u=1 σ. . u refer to the initial basis of single-particle states |r .t. N . . S} the elements of which will be labelled by indices ˜ m. In our formulas below we will adopt strictly the convention that indices r. . the corresponding U = ( Urm ) forms an S × N matrix and. S on which (9.143) with r. . . p. m = 1. j j (9. ˜ (9. m = 1. .σ and derive a single-particle operator which approximates this Hamiltonian.s.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. S) ˜ S S |m = ˜ r=1 Urm |r . U = ( Urm ) as deﬁned in (9. n. . t.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.144) deﬁned as in (9. . 2.9. as in in (9.142) over this reference state to turn the contribution into an eﬀective one-particle operator. 2. q. The states |m are connected with the states |r through (m = 1. u c† c† cuσ ctσ v (9. 2. 2. . S is a unitary matrix. . If one restricts m = 1. naturally. |r = m=1 ∗ Urm |m . .145). This is done as follows: ˆ Vmf (|Φo (U) ) = 1 2 S r.142) r|t|s c† csσ + rσ sσ 2 r. 2. . s. .108). 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. .s. is not unitary.146) .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. The states |m serve to deﬁne a reference state of the system ˜ N |Φo (U) = j=1 d† α d† β |0 . . s|ˆ|t. u v r.142) is based.

ˆ (9.n.s.t=1 σ. s| v t.148) one obtains. m. ˆ Vmf (|Φo (U) ) = 1 2 S Φo (U)|d† d† |Φo (U) = 0 mσ nσ Φo (U)|dmσ dnσ |Φo (U) = 0 Φo (U)|d† dnσ mσ δmn δσσ 0 |Φo (U) m ≤ N m > N (9.u ∗ r. m|ˆ|t. u c† d† dnσ cuσ ˜ vn rσ mσ r.σ S 1 − 2 where.m. n = ˜ ˆ ˜ s.266 Here ··· denotes the average O = Φo (U)| O |Φo (U) .146) accordingly. u c† ˜ vn sσ m.u=1 σ. m|ˆ|t.148) are obtained in a similar way by transforming only two of the four single-particle states into the new representation.n.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 c† ctσ δσ σ ˜ v ˜ rσ m≤N r. m| v |t.σ S ctσ + m. d† d† mσ nσ dmσ dnσ d† dnσ mσ = = = = and.σ S d† dnσ cuσ mσ − r.m.u=1 σ. 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.148) r. n c† c† v ˜ ˜ rσ sσ r. u ˜ ˜v m.s.σ S d† dnσ mσ ctσ (9. u Usm Uun . m|ˆ|˜ .n=1 σ.149) The remaining matrix elements appearing in (9. n|ˆ|t.t.84). s|ˆ|t. hence. for example.σ S dmσ dnσ + r.σ . For the averages ··· in (9.150) r. Many–Particle Systems (9. s|ˆ|m.u=1 σ.m.t. n c† d† dnσ ˜ v ˜ rσ mσ r.n. n c† ˜ v ˜ sσ m.n.s. by means of the rules (9. s|ˆ|˜ .n=1 σ.t=1 σ.σ S d† d† mσ nσ cuσ ctσ + r.

154) ˆ The mean ﬁeld approximation replaces then the Hamiltonian (9. t| v |u.2.t=1 σ (9. m|ˆ|˜ . s ) ˆ ˆ m≤N ∗ Utm Uum (9.. We will address now a proper choice of the reference state. n c† ctσ δσ σ ˜ v ˜ rσ − r.m. s|ˆ|t. m|ˆ|t.151) Carrying out the sum σ δσσ = 2 leads to a factor 2 for the ﬁrst two terms. m|ˆ|˜ . m=1.155) which is a function of the S × N –matrix U = ( Urm )r=1.. . leaves open the question how the reference state should be chosen.153) rσ r.149) for the matrix elements in the mixed representation one can state the right hand side explicitly in terms of Urm . At this point this state is completely arbitrary.s σ where r| vmf (|Φo (U) ) |s = ˆ t.σ S + − r.u ( 2 r. u c† cuσ ˜ vn rσ m≤N r.2. r| v |u. n c† ctσ . u = s. u c† cuσ δσσ ˜ vn sσ m≤N s.s=1 σ ˆ r|t|s + r| vmf (|Φo (U) ) |s ˆ c† csσ rσ (9. as introduced here.N (9..5: Self-Consistent Field Theory 1 2 1 2 1 2 S 267 m.u=1 σ S − m.u=1 σ.m. s| v |t. t| v |s. The one-particle operator (9. s|ˆ|˜ ...152) can then be written in the form ˆ Vmf (|Φo (U) ) = r| vmf (|Φo (U) ) |s c† csσ ˆ (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.9.142) by Hmf (U) deﬁned through S ˆ Hmf (U) = r.u=1 σ. u − r.σ S r.σ r. The mean ﬁeld approximation. Exploiting the symmetry r.152) By means of expression (9.t=1 σ.156) which deﬁnes the reference state |Φo (U) .S. ˜ v ˜ sσ m≤N s. u c† cuσ ˜ vn rσ (9.

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. . m = 1.160) . m = 1. i.: Determine Mean Field Hamiltonian In this step the mean ﬁeld Hamiltonian S ˆ (M ) Hmf = r. . (9. .4.6 Self-Consistent Field Algorithm The Self-Consistent Field (SCF) approximation. 2. . the reference state resulting from step 2M + 1 (within numerical error) is equal to the reference state resulting from step 2M − 1. We will argue below that the self-consistent ﬁeld ground state. i.155).e.157) and determines the associated diagonal representation deﬁned through S (1) ˆ r|t|s Usm = s=1 (1) (1) m Urm .s=1 σ S ˆ rσ r|t|s c† csσ (9. This procedure.s=1 σ ˆ r|hmf |s c† csσ rσ (M ) (9. i.145).4 and deﬁning this as the new reference state. r = 1. steps (2) and (3) are being repeated M times until the procedure converges. . .142). 2. is the lowest energy independent-particle ground state one can construct.268 Many–Particle Systems 9. . can achieve this goal only iteratively. Let us state now the construction of the self-consistent ﬁeld ground state in more detail. 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. determining in a step (2) the corresponding mean ﬁeld Hamiltonian (9.145). 9. S . under conditions which often are realized. . . Step 2M. however. . . . 2. (9.159) One deﬁnes the reference state |Φo (U(1) ) using the deﬁnition (9. U(1) is the S ×N –matrix (1) of the ﬁrst N column vectors of Urm .e. r = 1.155) following the method outlined in Section 9.119. Step 1: Choosing an Initial Reference State One deﬁnes the Hamiltonian ˆ (1) Hmf = r. . SCF-Algorithm. The procedure determines the reference state as the ground state of the independent-particle Hamiltonian (9. S . (9. and obtaining in a step (3) the ground state of this one-particle Hamiltonian according to the construction in Section 9.e.s=1 σ S ˆ r|t|s + r| vmf |Φo (U(M ) ) ˆ |s c† csσ rσ = r. S .158) where the labels m are ordered to obey the condition (1) m < (1) n for m < n . often also referred to as the Hartree-Fock approximation. S. assuming in an intial step (1) a properly chosen reference state. SCF-Algorithm. is exact only for an independent-particle Hamiltonian. 2. . M = 1. . 2...

step 2M + 2 is carried out.145). (9. 2.6. S . SCF-Algorithm. . .1: Let Prs be the one-particle operator with the generic operator prs = |r s|. S do not exceed a preset threshold. ˆ Exercise 9.s=1 σ ˆ r|f |s c† csσ rσ (9. 2. s ) ˆ ˆ m=1 Utm ∗(M ) (M Uum ) (9. .162) is solved adopting an ordering of the labels m which obeys the condition (M +1) m < (M +1) n for m < n .6: Self-Consistent Field Algorithm is evaluated. Step 2M + 1.165) can be written ˆ F = S ˆ ˆ r|f |s Prs r. t| v |u. s = 1.164) ˆ (a) Determine the expectation value of Prs for the state as deﬁned in (9. m = 1. ˆ (b) How can the expectation values of the operator Prr be interpreted. i. i. SCF-Algorithm: Continuation and Convergence Condition When step 2M + 1 is completed.: 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 . indicating that the state converged. . .145). . ..e. (e) Consider the creation operators N † gqσ = m=1 Vmq d† mσ (9.s=1 (9. .161) is calculated. 2. .e. . 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. . the S × S–matrix S (M ) r|hmf |s N 269 ˆ = r|t|s + t. S. (9. M = 1. i. t| v |s. (c) Show that any spin-independent one-particle operator S ˆ F = r. σ|ˆrs |u. m = 1. 2. 2 . r = 1.e.u=1 ( 2 r.163) The result yields the reference state |Φo (U(M +1) ) using the deﬁnition (9. r = 1. σ p = δtr δsu δσσ . ˆ t.9. 2. U(M +1) is the (M +1) S × N –matrix of the ﬁrst N column vectors of Urm . etc. u − r. .167) . . . . S (9.. S .166) ˆ (d) Deﬁne a similar two-particle operator Prstu . . .

170) (SCF ) Here |r .173) . r = 1. .163).9.172) is obeyed. We denote the representation which results from the SCF algorithm after its convergence as follows: S |m = ˜ r=1 (SCF Urm ) |r . For this representation holds N m| t |˜ + ˜ ˆn m 2 mm | v |˜ m ˜ ˜ ˆ n˜ − mm | v |m n ˜˜ ˆ ˜ ˜ = (SCF ) δmn m (9. Note that this is not an mσ S × S–matrix! Show that the reference state deﬁned through N † † gqα gqβ |0 q=1 (9.σ + 2v ( c† c1α c† c1β + c† c2α c† c2β ) + v 1α 2α 1β 2β + c† c2σ c† 2σ 1σ c1σ ) .168) ˆ ˆ has the same expectation values for Prs and Prstu as the reference state (9.145). (SCF ) m < (SCF ) n for m < n (9.169) 9. (9.162.6.171) implies that for the mean ﬁeld Hamiltonian holds S ˆ Hmf (U(SCF ) ) = m=1 σ (SCF ) † dmσ dmσ m . Equation (9. (9. (9. (f) Can one distinguish the states (9.168) through a physical observation? Exercise 9.145) and (9. .. S denotes the initial single-particle states and Urm is the unitary S × S– transformation matrix obtained in (9.270 Many–Particle Systems which are connected with d† through a unitary N × N –matrix (Vqm ).e. . 2.7 Properties of the SCF Ground State We want to investigate now the properties of the SCF ground state. i. We begin by summarizing the result of the SCF algorithm.171) where the convention (9.163).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.

177) 2 mm | v |mm ˜˜ ˆ ˜˜ − mm | v |m m ˜˜ ˆ ˜ ˜ .171) in terms of matrix elements r|t|s . Finally.e.177). ˆ SCF H SCF .. Employing expressions (9. u and Urm .177). the SCF ground state is N SCF = m=1 d† d† |0 .174) The ﬁrst property we consider is the total spin character of SCF . exploiting the spinindependence of (9. and once as a probe particle being subject to the mean pair interaction. once as a member of the ground state.84) and (9.9.176) 2 mm | v |mm ˜˜ ˆ ˜˜ − mm | v |m m ˜˜ ˆ ˜ ˜ . s|ˆ|t. v (SCF ) Show that the resulting eigenvalue problem of the type (9. 2 m=1 m|t|m + ˜ ˆ˜ m.146) counts a particle twice. i. . Since the system under consideration has only 2N particles altogether.e. one obtains ˆ SCF H SCF N = N (9. (SCF ) ˆ Exercise 9. the mean ﬁeld potential ˆ Vmf (|Φo (U) ) deﬁned in (9.4 that SCF is a total singlet state.93) for the expectation values (diagonal elements) of the one-particle and two-particle part of the Hamiltonian (9.. Exercise 9. Within the mean ﬁeld approximation as described by (9.2: Derive (9.162) is non-linear in Urm . mα mβ (9.173) holds N ˆ SCF Hmf (U(SCF ) ) SCF = 2 r=1 (SCF ) m .142). 2 m=1 (SCF ) m − m.142). one can also determine the expectation value of SCF for the Hamiltonian (9. r.1: Reformulate (9.m =1 The second term originates from the fact that in the mean ﬁeld approximation one assumes that each of the fermions is subject to an average pair interaction involving a 2N -particle reference state.175) However.171) yields ˆ SCF H SCF N = N (9.7.7: Properties of the SCF Ground State 271 i. Since this state is composed only of closed shells one can conclude following Section 9. We like to determine next the energy expectation value of SCF . the mean-ﬁeld Hamiltonian is diagonal in the SCF representation.7.142). (9. This over-counting leads to the correction term in (9.m =1 Comparision with (9.

s. We consider as an example the Hubbard model of an inﬁnite linear lattice and apply the model to describe magnetic instabilities in metals. v represents the ‘on-site’ Coulomb repulsion.. .4–9.128).180) V = rσ sσ 2 r.. is characterized through the parameters t. in the two-dimensional copper oxide lattices of high temperature superconductors. v (9. S N0 (9. Both N0 and N are macroscopically large numbers.σ crσ + crσ cr+1. the second term on ther. t = v δru δst δrs .182) n can assume values 0 ≤ n ≤ 2. rσ rσ r σσ (9. u c† c† cuσ ctσ .e. v (9. We assume that there are altogether S = 2N0 lattice sites.178) and through the so-called ﬁlling factor n n= 2N N = . referring to the fact that in this case the band of single particle energies (9.. reads 1 ˆ r.178). approximation schemes like the self-consistent ﬁeld approximation play an important role. β for ‘up’ and ‘down’ rσ spins. † ˆ Ho = −t c† .272 Many–Particle Systems 9.σ crσ + crσ cr+1.178) v contributes only in case that two electrons occupy the same lattice site. respectively) in state |r at lattice site r.181) The Hubbard model. describes the sites of a linear lattice.s. The one-dimensional Hubbard model is described by the Hamiltonian ˆ H = −t rσ † c† r+1.t. In the present Section we want to adapt the algorithm to systems containing a macroscopically large number of fermions. r ∈ Z. . The ﬁrst term on the r. s|ˆ|t.178). .7 for ﬁnite systems. The operator c† (crσ ) creates (destroys) an electron with spin σ (σ = α.178). The case n = 1 is termed the half ﬁlled band case. of (9.h. 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.8 Mean Field Theory for Macroscopic Systems The SCF algorithm has been developed in Sections 9.h.s.u σσ where r. N0 which are populated by 2N electrons (we choose an even number of electrons for convenience).e.e. is identical to that of the independent–particle Hamiltonian H0 (9. . . in the usual two–particle operator form. cannot be solved exactly. i.σ rσ The potential energy term. i. In spite of the simplicity of its Hamiltonian the Hubbard model.192) derived below is ﬁlled half.178) Here the index r. of (9. According to (9. i. real number.σ + v 2 c† c† crσ crσ . labeled r = −N0 + 1. s|ˆ|u. t is assumed to be a positive.179) r+1. as stated through Hamiltonian (9. except in case of one-dimensional systems. for example. t describes the coupling of state |r to states |r ± 1 at the two neighboring lattice sites. v of the Hubbard Hamiltonian (9. (9.

6. N0 (9. .8: Macroscopic Systems 273 Since we are dealing presently with a macroscopic system.155. 2N0 (9. boundary eﬀects are assumed to be negligible. 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. The unitary transformation which diagonalizes H0 . The self-consistent ﬁeld Hamiltonian Hmf which corresponds to the Hubbard Hamiltonian (9. we assume that the SCF ground state is given by (9. SCF Ground State for the Hubbard Model We seek to determine the ground state of the Hubbard model stated by (9.154).178) can be determined by applying (9.144). 9.186) where m = −2t cos mπ .e.183) where the operators d† are related to the original creation operators c† through the unitary mσ rσ ˆ transformation U deﬁned in (9.187) The energy levels are labeled such that 0 < ±1 < . mα mβ (9.185) ˆ According to (9. < ±(N0 −1) < N0 holds as can be readily veriﬁed.184) and using (9.179). Using (9. 9.181. the operator H0 can be re-written in the diagonal form N0 ˆ H0 = m=−N0 +1 σ m d† dmσ mσ (9. we identify |r + N0 = |r and later let N0 go to inﬁnity.184) ˆ We are now ready to apply step 1 of the SCF algorithm and diagonalize term H0 . i. in fact. by N −1 ||SCF = m=0 d† d† |0 .9. This task has ˆ been solved already on page 262. Accordingly. is Urm = √ 1 exp (irmπ/N0 ) . .188) .174). We can now embark on step 2 of the SCF algorithm. rσ rσ (9. Accordingly.140).σ m=0 ∗ Urm Urm c† crσ ..185) into (9.178) within the framework of the self-consistent ﬁeld theory. rσ (9. one obtains for the mean ﬁeld Hamiltonian N −1 ˆ ˆ Hmf (U) = H0 + v r. For this purpose we apply the SCF theory presented in Section 9. Inserting (9.186) one obtains for the mean ﬁeld Hamiltonian N0 ˆ (2) Hmf = m=−N0 +1 σ m d† dmσ + v mσ N 2N0 c† crσ .

holds k ≡ km ∈] − π. One can see that −2t ≤ (k) ≤ 2t holds.274 Many–Particle Systems Expressing c† .190) We have used in the latter expression the deﬁnition (9. π].1. π and. but may diﬀer quantitatively. there exist several ways of formulating mean ﬁeld approximations.191) where m is given in (9. −π ≤k ≤π .f. the results of diﬀerent formulations agree qualitatively. In other words. Apparently.e. −π + N0 . ˆ The Hamiltonian Hmf . 9. (9. i. 4 So far we have not exploited the fact that N0 and N are macroscopically large quantities. .187).185) one can prove rσ mσ N0 c† crσ = rσ rσ m=−N0 +1 σ d† dmσ . 9. The energy Emf of this state can be calculated readily by using (9.. the self–consistency condition is exactly met and the SCF algorithm converged after two steps. according to the Pauli principle. Usually. as given in (9. Emf is then N −1 v 2N0 Emf = 2 m=0 vn vn −N = 2 m+ 2 2 N −1 m m=0 + vn 4 (9. by ﬁlling up this energy band with electrons from the bottom of the band (which corresponds to k = 0) to a maximum energy.187) form a quasi1 2 continuous energy band. N0 ˆ (2) ˆ Hmf = Hmf = m=−N0 +1 σ m + vn † dmσ dmσ 2 (9.109) and using (9. .. crσ through d† . Indeed.e. dnσ according to (9. is already diagonal.178).177) and mm |ˆ|mm = mm |ˆ|m m = ˜˜ v ˜˜ ˜˜ v ˜ ˜ which holds in the present case.190). the ground state of the Hubbard model in the self-consistent ﬁeld approximation can be determined exactly. When N0 becomes macroscopically large the single particle discrete energy levels (9. i. namely. the electrons in the present description behave like independent particles with energies ¯m = m + vn . Alternative Description of the Mean Field Approximation The state (9. To demonstrate this we consider an alternative formulation of the mean ﬁeld approximation for Hamiltonian (9. even for one and the same Hamiltonian. The energy m can be expressed as a function of a continuous variable k ∈] − π. The ground state of the system. in the limit N0 → ∞.182) of the ﬁlling factor n [c. called the Fermi energy. . for km = mπ/N0 holds km ∈ −π + N0 . the state with the lowest possible energy.182)]. denoted by F (see Fig. can be obtained. through the so-called dispersion law (k) = −2t cos k .189) and. hence.1). (9. Indeed.192) This dispersion law is presented in Fig. . π].183) is not the only candidate for the mean ﬁeld ground state of the Hubbard model. mσ (9. The bottom of the energy band corresponds to (0) = −2t and the width of the energy band is 4t.

in the last term on the right hand side.1: The dispersion law (k) for independent electrons on an inﬁnite one dimensional lattice. let us express the interaction term in (9. i. i..56).8: Macroscopic Systems 275 Figure 9.45. Inserting (9. The anti-commutation properties (9.46) of the fermionic creation and rσ annihilation operators yield c† c† crσ crσ = −c† c† crσ crσ = rσ rσ rσ rσ ˆ ˆ ˆ2 c† crσ c† crσ − c† crσ δσσ = Nrσ Nrσ − Nrσ δσσ . rσ rσ rσ (9. The mean ﬁeld approximation neglects the .178) and performing the summation over the spin indices results in ˆ ˆ H = H0 + v r ˆ ˆ Nrα Nrβ .196) (9. say the one given in (9.183).e.178) in terms of the occupation number operaˆ tors Nrσ = c† crσ .9.195) ˆ ˆ ˆ and where δ (Nrσ ) = Nrσ − Nrσ describes the ﬂuctuations of Nrσ . First. where Nrσ is the mean occupation number of the one particle state |rσ .194) ˆ For a given reference state. the deviation of Nrσ from its mean value Nrσ for the corresponding reference state. (9.e. Nrσ = SCF ||Nrσ ||SCF ..193) where. 9. the operator Nrσ can be written ˆ Nrσ = Nrσ + δ (Nrσ ) . F = (±kF ) denotes the Fermi energy.193) into (9. we have employed (9. (9.

the ground state of the system can have a non-zero local spin and. i. that nα and nβ assume diﬀerent values. is given by (9. . Spin-Polarized Mean Field Ground State ˆ We want to consider now a ground state for Hmf deﬁned through Nrσ = nσ = const (9.276 Many–Particle Systems ﬂuctuations to second order and the mean ﬁeld Hamiltonian is obtained from (9. in the mean ﬁeld approximation..191) for the ground state energy. Because of Nrσ = c† crσ rσ N −1 = m. consequently. m+ 2 2 c† crσ − 2N0 rσ (9.m ∗ Urm Urm d† dm σ mσ = m=0 ∗ Urm Urm = n N = .183). separating the occupation number operator into a mean value plus ﬂuctuation and neglecting the ﬂuctuations in 2nd order.198) as the one obtained by using the mean ﬁeld potential (9. To determine a magnetic ground state we note ˆ Nrσ = r mm r ∗ Urm Urm d† dm σ = mσ mm ˆ δmm Nmσ = m ˆ Nmσ . 2N0 2 holds ˆ Hmf = = mσ mv ˆ H0 + 2 vn2 2 rσ nv † vn dmσ dmσ − N . however. Note that the procedure employed in (9. For such state the mean occupation number nrσ is uniform at all lattice sites. Therefore.e. 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 . (9.195).146).199) which is identical to expression (9. yields essentially the same mean ﬁeld Hamiltonian (9.200) allowing. In this approximation one can express ˆ ˆ Nrα Nrβ = ≈ = ( Nrα + δ (Nrα )) ( Nrβ + δ (Nrβ )) Nrα Nrβ + Nrα δ (Nrβ ) + Nrβ δ (Nrα ) ˆ ˆ Nrα Nrβ + Nrβ Nrα − Nrα Nrβ (9.183) . but the mean number of particles with spin α can be diﬀerent from the mean number of particles with spin β. non-vanishing magnetization.198) Let us assume that the ground state of the Hubbard model.194) by dropping all the terms which contain the ﬂuctuations to second order.

9.e. 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) .σ with respect to nα and nβ .208) . In this limit the discrete energy spectrum (9.8: Macroscopic Systems from which we obtain nσ = The mean ﬁeld Hamiltonian is then ˆ Hmf = m 277 1 2N0 Nmσ . as is shown schematically in Figure 9.. Therefore. 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 smaller is σ (k) for a given k. This last equation tells us that the electrons with spin α (β) are accommodated by an energy sub–band α (k) ( β (k)) which is obtained from the dispersion law of the non-interactive electrons (k) by an overall shift of vnβ (vnα ). For this purpose the sum in (9. if nα > nβ then β (k) > α (k).2a. (9.201) ( m ˆ + vnβ ) Nmα + ( m ˆ + vnα ) Nmβ − 2N0 vnα nβ .203) is provided by the continuous function σ (k) = (k) + vn−σ . In the limit N0 → ∞ this sum can be expressed as an integral.207) where (k) is given by (9.202) This Hamiltonian describes a system of non-interacting electrons with a spin-dependent dispersion law (9.192).203) mσ = m + vn−σ .e. e. The actual values of nσ (σ = α. (9. According to this method one minimizes ˜ Emf = Emf + µ (n − nα − nβ ) .206) ascertains that condition (9.204) Emf ≡ 2N0 2N0 m. (9. i. subject to the constraint n = nα + nβ . i. (9. At this point we exploit the fact that our system is macroscopically large.206).205) Such minimization is realized through the method of Lagrangian multipliers. the so-called Fermi energy F = µ (see below). Since β (k) − α (k) = v(nα − nβ ). The values of nα and nβ are determined by minimizing the energy density (9..206) needs to be evaluated. nβ and µ. 2π (9.206) ˜mf is minimized where µ is the Lagrangian multiplier considered an independent variable. one can see that an uneven occupation by electrons of the two energy bands yields a relative shift of the energy bands with respect to each other. N0 ∼ 1023 . one can expect the system to lower its energy by assuming nα = nβ and. a spin–polarized ground state.205) is met. E with respect to nα ..g. In fact. such that the larger nσ . β) are determined by minimizing the energy density of the ground state Hmf 1 = [ mσ Nmσ ] − vnα nβ (9. The additional term in (9. hence.

211) and using π dkδ(k ± (π − arccos( /2t))) = θ(2t − | |) . |f (xi )| (9. Inserting this last result into (9. therefore. For a given dispersion law (k) the density of states ρ( ) can be calculated by employing equation (9. which is usually called the density of states. Since the equation f (k) ≡ 2t cos k + = 0 has two simple roots. 2π (9.210). namely k1.192) and.212) where xi are the simple roots of the function f (x). holds π ρ( ) = −π dk δ( + 2t cos k) . and noting |f (k)| = |2t sin k| = 2t one obtains δ( − (k)) = 1 − cos2 k = (2t)2 − 2 . δ(k + π − arccos( /2t)) + δ(k − π + arccos( /2t)) (2t)2 − 2 .211) In order to calculate the integral we recall the following property of the Dirac–delta function δ [f (x)] = i δ(x − xi ) . unit energy and a given spin orientation of the particle.278 Many–Particle Systems (a) (b) Figure 9. gives the the number of available one particle states per site. The shaded areas represent the ﬁlled portions of these bands by electrons.2 = ±(π − arccos( /2t)). If the function f depends explicitly only on (k) one can state π −π dk f ( (k)) = 2π π ∞ ρ( )d f ( ) −∞ (9.210) Here δ(x) is the Dirac–delta function and ρ( ). (b) Graphical deﬁnition of the limiting energies α and β . In our case (k) is given by (9.209) where ρ( ) = −π dk δ ( − (k)) . 2π (9. −π .2: (a) Relative position of the two energy sub–bands α (k) and β (k) for ∆ = nα − nβ > 0.

218) read ˜ ∂ Emf = ∂ α ˜ ∂ Emf = ∂ β and α + vnβ − µ = 0 .219) (9.206)] depends only on continuous quantities. (9. 9. if we assume that is restricted to the interval ] − 2t.213) The presence of the θ function in the above formula guaranties that the density of states vanishes for | | > 2t. 2 (9.216) allow one.215) nα and nβ in (9. (9. in principle. on conditions for a ground state of minimum energy are ˜ ∂ Emf =0. π α... β (9. of t. outside the energy band (k). (9.f. One obtains ˜ Emf = α β ρ( )d + −2t −2t ρ( )d + vnα nβ + µ(n − nα − nβ ) .220) β ˜ ∂ Emf = n − nα − nβ = 0 . 2t[ then the θ function in (9. Fig. nα and nβ as a function of the parameters of the Hubbard model. ∂µ (9.e.221) ∂µ Equations (9. namely. We can employ the results obtained to express the ground state energy density (9.213) can simply be replaced by 1.e.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.221) and (9.9. i.222) . 9. β .218) The derivatives can be readily determined and the conditions (9. (9. + vnα − µ = 0 . ∂ β ˜ ∂ Emf =0. the necessary ˜ Since Emf [c. θ(x) = 1 if x > 0 and θ(x) = 0 if x < 0.. i.219. v and n.214) are given by the sum (9. µ.204) through an integral expression. ∂ α ˜ ∂ Emf =0. to determine the unknown quantities α . one arrives at the following expression of the density of states ρ( ) = θ(2t − | |) π (2t)2 − 2 .201).216) Using (9. Replacing again the sum by an integral over energy one can write σ nσ = −2t ρ( )d (9.219–9.213) and carrying out the resulting integral yields nσ = 1 arcsin( σ /2t) .f.e. F = α + vnβ = β + vnα .8: Macroscopic Systems 279 where θ(x) is the step function. From (9. namely. i.2].220) µ can be readily obtained µ= 1 ( 2 α + β) + vn . (9.217) and µ. (9.

226) ∆ = f (∆) . For this purpose one plots f (∆) versus ∆ as is shown schematically in Figure 9. For v above a critical value vc ..223) ∆ = 0 corresponds to the ground state (9.226) follows that f (∆) is a monotonically increasing function of ∆. Equation (9. through the condition [c. One can determine vc by equating the slope of f (∆) at the origin to 1. f (∆) converges to a maximum value as ∆ → ∆max = n. (9.280 Many–Particle Systems Comparision of (9. condition (9.222) and (9.227) Since f (0) = 0 one can infer that ∆ = 0 is always a solution (the so called paramagnetic solution) of condition (9.228)] vc ρ ( β ) = 1 .223 one obtains ∆= and second. so called ferromagnetic solution. the slope of f (∆) at the origin is larger than 1 [c.224) v∆ ∆= β ρ( )d = 0 ρ( β + )d .227) has only the trivial solution ∆ = 0.226) one can obtain ∆ by solving (cf.229) has a wider range of validity than one can infer from the present analysis of the Hubbard model. Accordingly. We assume in the following discussion that the implicit ∆ dependence of β in f (∆) can be neglected.183) without magnetization. From the deﬁnition (9.228) Since the total number of states per site is ﬁnite. First. (9. i. i.220) and (9. a magnetic ground state. (9. (9. the Hubbard model has a ground state with ∆ = 0.225) Deﬁning the function v∆ f (∆) = 0 ρ( β + )d .e. and only if. One still needs to determine α and β . from (9.e. If the slope of f (∆) at the origin is less than 1. as reﬂected by the expression d[f (∆)] = vρ( d∆ β + ∆) = π (2t)2 v − (v∆ + 2 β) >0. For this purpose we consider the magnetization of the mean ﬁeld ground state ∆ ≡ nα − nβ .219–9. Indeed. non trivial. (9. (9. The Stoner criterion (9. from (9.215) reveals that µ is indeed equal to the Fermi energy F (see also Figure 9.217) and (9.227). if. (9.227) can be solved graphically.. The slope of f (∆) at the origin depends on the Hubbard model parameter v. One can express ∆ as a function of α and β in two diﬀerent ways. .3]. (9.227) will have a second. condition (9. Figure 9.f. (9.223) follows α α − v β . the slope of f (∆) is positive for all ∆ ≥ 0.3.2b).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.

233) yield the desired expression vc π = π cos (n − 1) . 9. y and. By parameterizing α and β α = 2t sin x . n). β = 2t sin y . Equation (9.230) one obtains from (9. 2t As already mentioned.4. the plot of vc vs.232) together with (9. t.231) follows y= δ π π (n − 1) − ≈ (n − 1) . n.234) 2t 2 The resulting phase diagram of the ground state.e. From (9.8: Macroscopic Systems 281 Figure 9..205) the following set of equations determining x. is presented in Figure 9. The corresponding 2t analytical expression provides the phase diagram of the ground state: when v approaches vc from above.233) Employing the approximation sin(y + δ) − sin y ≈ δ cos y for small δ. (9. the magnetization ∆ of the ground state vanishes and remains zero for values v < vc .231) (9.232) For a given 2t the values of x and y can be obtained numerically and other relevant quantities v mentioned above can be calculated. α. Accordingly. i.232) states that x − y is proportional to ∆. 2 2 2 (9. (9.9. the ground state is characterized by the quantities (v.3: Graphical solution of the mean ﬁeld equation ∆ = f (∆). 9. (9. We like to determine ﬁnally the dependence of vc on the ﬁlling factor n. v (9. hence. any . β x+y x−y = = π(n − 1) 2t π (sin x − sin y) = π∆ . Since ∆ vanishes near vc we set x = y + δ where δ is a small quantity.219.220.

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

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

284 Many–Particle Systems .

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

5) √ One can interprete the quantity −s2 as a distance in a 4–dimensional Euclidean space if one chooses the time component purely imaginary.7) equation.5. This choice implies dim(time) = dim(length).4) invariant. i.5) and the Dirac (Sect. namely that of length.e. i. We will denote these vectors as follows def xµ = (x0 . The Group of Lorentz Transformations L = O(3. wave functions ψ(r. Rather than starting from (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. The elements L ∈ L act on 4–dimensional vectors of position– and time–coordinates.4 to cover ﬁelds.286 Relativistic Quantum Mechanics in Sect. 10. namely the Klein–Gordon (Sect. however. 10. x 1 . The reason for this choice is that the transformations (10.e. x 2 . x2 .7 a heuristic derivation of the two most widely used and best known Lorentz–invariant ﬁeld equations.. 10.3) where x0 = ct describes the time coordinate and (x1 .. Minkowski Space Historically. 10. t) and vectors with functions ψ(r.1 Natural Representation of the Lorentz Group In this Section we consider the natural representation of the Lorentz group L. We will. x1 . Denoting by xµ the .e. henceforth.1). x3 ) (10. but in a setting of representation theory methods as applied in Secti. Note that the components of xµ all have the same dimension. x2 . Rather than following this avenue we will introduce Lorentz transformations within a setting which does not require real and imaginary coordinates. This space is called the Minkowski space.1) The Lorentz transformations L describe the relationship between space-time coordinates xµ of two reference frames which move relative to each other with uniform ﬁxed velocity v and which might be reoriented relative to each other by a rotation around a common origin. We will ﬁnd that this deﬁnition will lead us back to the transformation law (10. 5 to the groups SO(3) and SU(2) of rotation transformations of space coordinates and of spin.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. 10.1 will be extended in Sect. In such a space Lorentz transformations correspond to 4-dimensional rotations. the group of Lorentz transformations (10. assume new units for time such that the velocity of light c becomes c = 1. we will provide a more basic deﬁnition of the transformations. x 3 ) holds (x0 )2 − (x1 )2 − (x2 )2 − (x3 )2 = (x )2 − (x )2 − (x )2 − (x )2 . 10. for the transformed space-time–cordinates x µ = (x 0 .1) leave the quantity s2 = (x0 )2 − (x1 )2 − (x2 )2 − (x3 )2 (10.1). t) as components. 0 1 2 3 (10. x3 ) = r describes the space coordinates. i. 10. Before introduco ing these general Lorentz–invariant ﬁeld equations we will provide in Sects.

0 0 0 −1 (10. ν=0 (10. Aµ ν . xµ . We will exploit this property below to determine the general form of the Lorentz transformations. The Lorentz transformations are non-singular 4 × 4–matrices with real coeﬃcients. Aµ ν B ν ρ = ν=0 Aµ ν B ν ρ .10. R). 3 3 3 yµ xµ = new yµ xµ .5) into scalar products.9) 1 0 0 0 −1 0 0 ( gµν ) = 0 0 −1 0 = g . the latter set constituting a group. This condition can be written xµ gµν xν = x gµν x where µ ν (10. summation over that index is assumed without explicitly noting it. i. [ A more conventional notation would be Lµν .6) Here Lµ ν are the elements of a 4 × 4–matrix representing the Lorentz transformation. the latter notation will be used for diﬀerent quantities further below.9) and (10.8) new old old The summation convention allows us to write (10. First. Aµν .6) x µ = Lµ ν xν .12) (10. (10. (10. 3: 4-vector: xµ . (10.6) yields Lµ ρ gµν Lν σ xρ xσ = gρσ xρ xσ . L ∈ GL(4. 2.e. This will be explained further below.. 4 × 4 tensor: Aµ ν ..10) Combining condition (10. the notation in (10. The Lorentz transformations form the subgroup of all matrices which leave the expression (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. Aµ ν xν = µ=0 new old Aµ ν xν .5) invariant.11) This condition speciﬁes the key property of Lorentz transformations. i. however.e.7) The reason for the notation is two-fold.1: Natural Representation of the Lorentz Group 287 coordinates in one reference frame and by x µ the coordinates in the other reference frame. the Lorentz transformations constitute a linear transformation which we denote by 3 x µ = ν=0 Lµ ν xν . Since this holds for any xµ it must be true Lµ ρ gµν Lν σ = gρσ . The second reason is that upper and lower positions allow us to accomodate the expression (10. The subset of GL(4. ν = 0. 1.] The following possibilities exist for the positioning of the indices µ. R). Aµν . the .6) allows us to introduce the so-called summation conventon: any time the same index appears in an upper and a lower position.

1).1) since it satisﬁes (10. L3 ∈ O(3.19) i.1).288 Relativistic Quantum Mechanics elements of which satisfy this condition. of O(3.h.17) one obtains (L−1 )T g L−1 = gLgggLT g = gLgLT g .1).s. We consider 1 then L1 .e.14). R).18) (10.1). One can also show that if L ∈ O(3.1) is...22) (10. L2 ∈ O(3.14).e. = g LT g = 0 1 2 −L 2 L 2 L 2 L3 2 −L0 3 L1 3 L2 3 L3 3 (10. .15) (10. property (10.21) i. LT gL = g } . This set is identical with the set of all Lorentz transformations L. In fact.. the inverse of L 1 L−1 (10.14) holds for the inverse of L. 1) = { L.16) The corresponding expression for (LT )−1 is obviously (LT )−1 = (L−1 )T = g L g . hence.16.14) from the right by gLT and using (10. This stipulates that O(3. L3 0 L3 1 L3 2 L3 3 Using the deﬁnition (10. L ∈ GL(4.e.1). L−1 ∈ O(3. i. 3 2 1 2 1 2 (10.12) LT gL = g . (10. We want to show now L = O(3. (10. of (10.13) L = ( Lµ ν ) = 2 0 L 0 L2 1 L2 2 L3 3 . (10. To demonstrate the group property of O(3.10) of g one can rewrite the invariance property (10. R) is a group. employing expressions (10.20) Multiplying (10.. in fact.14) L0 0 −L1 0 −L2 0 −L3 0 −L0 1 L1 1 L2 1 L3 1 .17) we note ﬁrst that the identity matrix 1 is an element of O(3. From this one can obtain using g2 = 1 1 (gLT g)L = 1 and. 10.e.1) ⊂ GL(4.15) 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. i. For L3 = L1 L2 holds LT g L3 = LT LT g L1 L2 = LT (LT gL1 ) L2 = LT g L2 = g . 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. the associated inverse obeys (10. is called O(3.1).20) results in the desired property (L−1 )T g L−1 = g . a group.

B ∈ L+ holds C = AB ∈ L+ . Using det AB = det A det B and det AT = det A yields (det L)2 = 1 or det L = ±1 . = { L. L ∈ O(3. = { L.24) or since (L1 0 )2 + (L2 0 )2 + (L3 0 )2 ≥ 0 it holds (L0 0 )2 ≥ 1. For this purpose we consider the value of C 0 0 = A0 µ B µ 0 = 3 A0 j B j 0 + A0 0 B 0 0 . A statement on this value can be made on account of property (10.34) j=1 j=1 .31) (10.14).14). 1).23) One can classify Lorentz transformations according to the value of the determinant into two distinct classes.23) and (10. L0 0 ≤ −1} . 1).30) ≤ −1} .1). ∃M2 .1: Natural Representation of the Lorentz Group Classiﬁcation of Lorentz Transformations 289 We like to classify now the elements of L = O(3.25) can be stated as follows: The set of all Lorentz transformations L is given as the union L = L↑ ∪ L↓ ∪ L↑ ∪ L↓ + + − − where L↑ .25) i.10. For this purpose we consider ﬁrst the value of det L.29) (10. L↓ . det L = −1. Lorentz transformations in L↑ . L0 0 (10. (10. hence. + = − L↑ g + (10. Obviously. (10. L↓ are disjunct sets deﬁned as follows + + − − L↑ + L↓ + L↑ − L↓ − = { L.. Properties (10. It holds g ∈ L and −1 ∈ L as one can readily verify testing for property (10. M1 = a M2 }. = { L.28) (10.12).e.27) (10. + We start our investigation by demonstrating that L↑ forms a group. + except for the trivial factors g and −1 It is. entirely suitable to investigate ﬁrst only 1. One can also 1 verify that one can write L↑ − L↓ + L↓ − = gL↑ + = = −L↑ .12) the case ρ = 0. σ = 0 yields L0 0 2 − L1 0 2 − L2 0 2 − L3 0 2 = 1. L ∈ O(3.33) where we used the deﬁnition aM = {M1 . L0 0 ≥ 1} .26) (10. From this we can conclude L0 0 ≥ 1 or L0 0 ≤ −1 . Considering in (10. det L = 1. + − gL↑ + = L↑ g . L↑ contains 1 1. A second class property follows from (10. L↑ . The above shows that the set of proper Lorentz transformations L↑ allows one to generate all Lorentz transformations. det L = −1. there exist two other distinct classes. L ∈ O(3. M2 ∈ M. + + ↑ ↑ We can also demonstrate that for A. 1). (10. L0 0 ≥ 1} .32) (10.14) which we employ in the formulation (10. (10. 1). L ∈ O(3. Schwartz’s inequality yields j=1 2 3 3 3 j=1 A jB 0 ≤ 0 j A 0 2 j Bj 0 2 . det L = 1.

therefore.e. In the following we + will consider solely this subgroup of SO(3. i. + i. since the group property of O(3.. we consider transformations Lµ ν = δ µ ν + µ ν . ρ ν (10.35) One can conclude.38) For these transformations.1).e. Inﬁnitesimal Lorentz transformations The transformations in L↑ have the property that they are continously connected to the identity 1 1. holds L0 0 > 0 and the value of the determinant is close to unity. 1 0 0 = ≥ 1 and. (10. . if we enforce (10.13). µ ν small . Since the associative property holds for matrix multipli1 1 + ↑ cation we have veriﬁed that L+ is indeed a subgroup of SO(3.12) follows (B 0 0 )2 − 3 (B j 0 )2 = 1 or 3 (B j 0 )2 = (B 0 0 )2 − 1. + transformations in a small neighborhood of 1 which we parametrize by inﬁnitesimal parameters. In fact. hence. 1 We will then employ the Lie group properties to generate all transformations in L↑ . To order O( 2 ) holds g + g = 0. 1 ∈ L↑ . orthochronous Lorentz transformations. one can j=1 j=1 3 0 )2 = (A0 )2 − 1.. + Accordingly.40) and. L↑ is called the subgroup of proper. 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. (10. (10.14) implies 1 + T g (1 + ) = g 1 1 (10.1). accordingly. from which one can conclude L−1 ∈ L↑ . The next group property of L↑ to be demonstrated is the existence of the inverse.14) actually L0 0 ≥ 1 and det L = 1 must hold.36) It holds. Similarly. Since A0 0 ≥ 1 and B 0 0 ≥ 1.1) ascertains CT gC = g it must hold C 0 0 ≥ 1. δ00 = 1 and It should be noted that according to our present deﬁnition holds δµν = gµρ δ δ11 = δ22 = δ33 = −1. | 3 A0 j B j 0 | < A0 0 B 0 0 . obviously j=1 A0 0 B 0 0 ≥ 1. 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 1.34) provides then the estimate conclude from (10. This relationship shows (L−1 )0 0 = L0 0 .16). Property (10.290 Relativistic Quantum Mechanics From (10.37) (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 . obviously.39) where we have employed the matrix form T 1 deﬁned as in (10. Using the above expression for C 0 0 one can state C 0 0 > 0. For the inverse of + ↑ any L ∈ L+ holds (10.

K ] [ Jk . (10. J = 0 3 0 1 0 0 0 0 0 0 0 0 0 0 0 (10. ϑ3 .46) 0 0 0 0 . ϑ2 . The generators are explicitly 0 0 0 0 0 0 0 0 J1 = 0 0 0 −1 0 0 1 0 (10. j.45) (10. k = 1. = − k m Km (10. w1 .42) This relationship implies µ 0 j µ j = 0 = j 0 . K2 = . K ] The operators also obey J · K = J1 J1 + J2 J2 + J3 J3 = 0 = = k m Jm k m Jm 0 −1 0 0 0 0 0 0 (10. 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 .15) one can conclude T 291 = −g g (10. 2. 3 (10. 3 .48) (10. w3 ) = −w2 ϑ3 0 −ϑ1 −w3 −ϑ2 ϑ1 0 (10. J ] [ Kk .1: Natural Representation of the Lorentz Group Using (10. 2. 2. w2 .47) 0 −1 0 0 0 0 −1 0 0 −1 0 0 0 0 0 0 0 0 K1 = . J2 = 0 0 0 −1 0 0 0 0 0 0 0 0 0 0 0 −1 0 1 .49) .50) . k j j = 1. other ﬁve parameters zero) Kk = (wk = 1.44) This result allows us now to deﬁne six generators for the Lorentz transformations(k = 1.10. (ϑ1 . other ﬁve parameters zero) .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 .43) k = − Inspection shows that the matrix has 6 independent elements and can be written 0 −w1 −w2 −w3 −w1 0 −ϑ3 ϑ2 . 3) Jk = (ϑk = 1.

w) = exp ϑ · J + w · K .56) = 1 0 0 1 = 1 1 (10. that the transformations (10.52) where the 3 × 3–matrices Lk are the generators of SO(3) deﬁned in Chapter 5. k = 1.49) deﬁne the Lie algebra associated with the Lie group L↑ . 2.51) for w = 0 correspond to rotations of the spatial coordinates. w ∈ R3 3 k=1 wk Kk . i.53) Here R(ϑ) are the 3 × 3–rotation matrices constructed in Chapter 5.35) 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 . and using the relationship Jk = 0 0 0 Lk (10. 2.1: Demonstrate the commutation relationships (10. To determine the explicit form of this transformation we evaluate the exponential operator by Taylor expansion. ϑ. 2π[ . A boost in the x1 –direction is L = exp(w1 K1 ). L(ϑ. constitutes an overcomplete parametrization of the rotations (see Chapter 5). (10. Relativistic Quantum Mechanics The commutation relationships (10. 0 1 n (10.292 as can be readily veriﬁed. 3 (10. however. For the parameters ϑk of the Lorentz transformations holds obviously ϑk ∈ [0. (10.51) One can readily show.54) which. We consider now the Lorentz transformations for ϑ = 0 which are referred to as ‘boosts’. 3 where we have deﬁned ϑ · J = k=1 ϑk Jk and w · K = following the algebra in Chapter 5. Kk .50).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. + The commutation relationships imply that the algebra of the generators Jk .. In analogy to equation (5.57) . 3 is closed. w = 0) = 0 0 0 R(ϑ) . Following the treatment of the rotation group SO(3) one can express the elements of L↑ through + the exponential operators L(ϑ.49. Exercise 7.e. 10.

vk deﬁned in (10. we can state vk ∈ ] − 1.63) which agrees with (10. Accordingly. (10.58) = cosh w1 1 + sinh w1 1 0 −1 −1 0 The conventional form (10.60) sinh w1 = v1 / 2 1 − v1 . ∞[ . cosh2 w1 − 1 and coshw1 = cosh2 w1 − 1 = cosh w1 sinh w1 sinh2 w1 + 1 . consistent with (10. one obtains from (10. (10. 10.51) the explicit transformation for space–time–coordinates is then x1 = x1 − v1 t 1 − 2 v1 .59) v1 = These two equations yield cosh w1 = 1/ and (10. 10.6. The range of the parameters wk can now be speciﬁed. Correspondingly. in fact.1: Natural Representation of the Lorentz Group one can carry out the Taylor expansion above: ∞ 293 L = n=0 2n w1 1 + 1 (2n)! ∞ n=0 2n+1 w1 (2n + 1)! 0 −1 −1 0 = cosh w1 −sinh w1 −sinh w1 cosh w1 . 10. x2 = x2 . . (10. t = t − v1 x1 2 1 − v1 .10. This property is.62) According to (10. 1[.1). x3 = x3 (10.59).59) can be written 1 2 1 − v1 √−v1 2 1 − v1 2 1 − v1 .59) for the case k = 1 corresponds to the relative velocity of two frames of reference.59) cosh w1 Using cosh2 w1 − sinh2 w1 = 1 one can identify sinhw1 = sinh2 w1 + 1. (10.59) follows. for wk wk ∈ ] − ∞.56.64) We note that the range of wk -values is not a compact set even though the range of vk -values is compact. however.3.1) of the Lorentz transformations is obtained through the parameter change sinh w1 v1 = = tanh w1 (10.61) √ √−v1 √ exp (w1 K1 ) = 2 1 − v1 1 2 1 − v1 0 0 0 0 0 0 0 0 1 0 0 1 (10. From (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. We expect that vk can only assume values less than the velocity of light c which in the present units is c = 1.

p) . vectors like r = (x1 . f = f. scalars like r = x2 + x2 + x2 . The 4-vector character of J µ and of Aµ will be demonstrated further below. a2 . However. another example is the rest mass m of a particle. A) (10.65) An example is s2 deﬁned in (10. Such quantities appear in the description of physical systems and statements about transformation properties are often extremely helpful and usually provide important physical insight. not in the so-called natural representation associated with the 3–dimensional Euclidean space E3 . we have not yet deﬁned representations of Lorentz transformations beyond the ‘natural’ representation acting in the 4–dimensional space of position– and time–coordinates. our deﬁnition of quantities with special properties under Lorentz transformations presently is conﬁned to the natural representation. t) are the charge density and the current density. a1 . E = √ m mv . the z–component x3 of a particle. not any physical property f ∈ R is a scalar.68) where ρ(r. spherical harmonics Y m (ˆ). .69) where V (r. x2 . we will encounter an impressive example of physical properties. t)|2 or the scalar product r1 · r2 of two particle positions. of a system of charges. x3 )T . Counterexamples are the energy. Scalars The quantities with the simplest transformation behaviour are so-called scalars f ∈ R which are invariant under transformations. t) and A(r.67) the transformation behaviour of which we will demonstrate further below.3) are the so-called 4–vectors aµ . J) (10. i. Nevertheless. (10. namely.e. 2 |ˆ1 . Presently. p = √ 2 1−v 1 − v2 (10. Another example is the potential 4-vector Aµ = (V.2 Scalars. t) are the electrical and the vector potential of an electromagnetic ﬁeld. the charge density. We will see below how true scalars under Lorentz transformations can be constructed. Some of these r ˆ quantities were actually deﬁned with respect to representations of the rotation group in function spaces.294 Relativistic Quantum Mechanics 10. 4-Vectors The quantities with the transformation behaviour like that of the position–time vector xµ deﬁned in (10. a3 )T and transform according to a µ = Lµ ν aν . r2 ) and. These quantites always come as four components (a0 .4). respectively. t) and J(r. total angular momentum r 1 2 3 states of composite systems like Y m ( 1 . (10. 4–Vectors and Tensors In this Section we deﬁne quantities according to their behaviour under Lorentz transformations.66) Examples of 4-vectors beside xµ are the momentum 4-vector pµ = (E. the square of the electric ﬁeld |E(r. We have encountered examples in connection with rotational transformations. A third 4-vector is the so-called current vector J µ = (ρ.. tensor operators Tkm . ﬁnally. Hence.

one can express [(L−1 )T ]µ ν = (L−1 )ν µ and.66). This property follows from (10. ∂t (10. −a2 . i.73) aµ = Lµ σ aσ .78) To prove the 4-vector property of ∂µ we will show that g µν ∂ν transforms like a contravariant 4vector. the relationship being aµ = gµν aν = (a0 . (10. We have duplicated the expression for the inverse of Lµ ν to obtain the correct notation in terms of covariant. In any case one can express (10. (10. The respective vectors aµ are associated with the 4-vectors aµ . In fact. gσν xν = Lρ σ gρµ x µ and g µσ gσν = δ µ ν . One calls 4-vectors aµ covariant and 4-vectors aµ contravariant. ∂x µ (10. (10.71) Contravariant and Covariant 4-Vectors It is convenient to deﬁne a second class of 4-vectors.74) g µν aν . aµ (10. −a1 .77) The transformed diﬀerential operator will be denoted by ∂µ = def ∂ .72) where aν is a vector with transformation behaviour as stated in (10. We do not fact. 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. i. Covariant 4-vectors transform like a µ = gµν Lν ρ g ρσ aσ where we deﬁned g µν = gµν . using (10..72) of the covariant 4-vector follows = In µν and g µ as well.2: Scalars. −a3 ) (10.e.76) An important example of a covariant 4-vector is the diﬀerential operator ∂µ = ∂ = ∂xµ ∂ .73) Note that according to (10.75) (10. If aµ and bµ are 4-vectors aµ gµν bν (10.16).17) can be written (L−1 )ν µ = Lµ ν . xν = g νσ Lρ σ gρµ x µ . accordingly. g µν ∂ν = Lµ ρ g ρσ ∂σ .79) This is the inverse Lorentz transformation consistent with (10. We start from x µ = Lµ ν xν . one can employ the tensors g µν to raise and lower indices of L ν establish here the consistency of the ensuing notation. Multiplication (and summation) of x µ = Lµ ν xν by Lρ σ gρµ yields.66) together with (10. lower. The 4-Vector ∂µ (10.e.12) a gµν b µ ν = Lµ ρ gµν Lν σ aρ bσ = aρ gρσ bσ (10.70) is a scalar.17) Lµ σ is the transformation inverse to Lσ µ . and .12).10.

e. Proof that pµ is a 4-vector We will demonstrate now that the momentum 4-vector pµ deﬁned in (10. dτ 1 − v 2 dt p0 = E = √ m m dt = √ . 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.− ∂t . 1 − v2 1 − v 2 dt (10. d’Alembert Operator We want to construct now a scalar diﬀerential operator.84) = 1 − (v)2 (10.80) Multiplication by g λµ (and summation over µ) together with g λµ gρµ = δ λ ρ yields g λµ ∂µ = Lλ σ g σν ∂ν . For this purpose we deﬁne ﬁrst the contravariant diﬀerential operator ∂ µ = g µν ∂ν = Then the operator 2 ∂µ ∂ µ = ∂t − 2 (10.67) transforms like (10.e. (10.87) The remaining components of pµ can be written.85) 1 d d = √ . (10.79) allows us to determine the connection between ∂µ and ∂µ . indices. ∂µ does indeed transform like a covariant vector. In fact. upper.82) (10. this operator is equal to the d’Alembert operator which is known to be Lorentz-invariant.81) ∂ ..83) is a scalar under Lorentz transformations. i.296 Relativistic Quantum Mechanics contravariant.88) One can express then the momentum vector pµ = √ d m dxµ = m xµ . 2 2 dt 1 − v 1 − v (10. For this purpose we consider the scalar diﬀerential (dτ )2 = dxµ dxµ = (dt)2 − (dr)2 It holds dτ dt from which follows 2 (10.86) One can write (10. p1 = √ m v1 m dx1 = √ .89) ..g.66).. i. e. 2 dt dτ 1 − v (10.

. Expansion in should then be rapidly convergent.h. namely. · A = 0. and had chosen the so-called Coulomb gauge (8. yields according to (10. i.96) is a scalar and.10. can be demonstrated readily using the 4-vector notation (10.89) alltogether transforms like a contravariant 4-vector. t) + · A(r. We like to rewrite the last result E 2 = p 2 + m2 or E = ± p 2 + m2 . hence. t) and A(r. The momentum 4-vector allows us to construct a scalar quantity. in (10. equivalently.s.s.92) (10.90) (10. t) = 0 .h. . pµ on the l. If we can show that Aµ deﬁned in (10. ∂t V (r. in fact. namely pµ pµ = pµ gµν pν = E 2 − p 2 Evaluation of the r.s. (10. 10.93) (10.12). kinetic energy ± 2m and relativistic corrections.96) The Lorentz-invariance of this gauge.h. (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 .s. Lorentz Gauge In our previous description of the electrodynamic ﬁeld we had introduced the scalar and vector potential V (r. (10. and.3: Relativistic Electrodynamics 297 d The operator m dτ transforms like a scalar.e. t). Since xµ transforms like a contravariant 4-vector. of (10. in fact. m2 m2 v 2 − = m2 1 − v2 1 − v2 (10.89) must be a 4-vector. of (10. is a scalar.96) in the form ∂µ Aµ = 0 .97) We have proven already that ∂µ is a contravariant 4-vector.3 Relativistic Electrodynamics In the following we summarize the Lorentz-invariant formulation of electrodynamics and demonstrate its connection to the conventional formulation as provided in Sect.91) (10. respectively. hence. the so-called Lorentz gauge.h.67) E2 − p 2 = or pµ pµ = m2 which.77) which allow one to express (10. for these potentials. in (10. the r. We will demonstrate now the 4-vector property of Aµ .97) and.69) is. Lorentz-invariant.95) 2 p This obviously describes the energy of a free particle with rest energy ±m. 8. This gauge is not Lorentz-invariant and we will adopt here another gauge.94) 1 m In the non-relativistic limit the rest energy m is the dominant contribution to E. a contravariant 4-vector then the l.

thereby. The l.77) and (10.h. a 4-vector. t) . 10. t) . 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 . one obtains 2 ∂t V (r. k. according to (10. of (10. The respective equation for A0 = V can be obtained from Eq. t) = 0 (10. F µν = x Ey B z 0 −Bx Ez −By Bx 0 Alternatively. t).98) which reﬂects the principle of charge conservation.13). . Equation (10. t) 2 ∂t A(r.102) In this equation the r. · A(r.96). F mn = − mn (10. t) = 4 π J(r.99) must be a scalar. must transform likewise. t) = ∂t · A(r.s.h. t) together with (10. it follows that J µ . t) are known to obey the continuity property ∂t ρ(r. Since ∂µ ∂ µ transforms like a scalar one can conclude that Aν (xσ ) must transform like a 4-vector.e. in fact. We want to derive now the diﬀerential equations which determine the 4-potential Aµ in the Lorentz gauge (10. using (10.96). that F µν can be expressed through the potential Aµ in (10.105) . must be a scalar. yields ∂µ ∂ µ Aν (xσ ) = 4 π J ν (xσ ) . t) − 2 V (r.69) and ∂ µ in (10. prove that Aµ is.101) Combining equations (10. t) − 2 A(r.18) and. (10. t) = 4πρ(r.. ∂µ J µ (xµ ) = 0 . has to transform like a contravariant 4-vector. Consequently. t) + · J(r.298 Transformation Properties of J µ and Aµ Relativistic Quantum Mechanics The charge density ρ(r. This principle should hold in any frame of reference.6) and (8. 2. (10. n = 1. (10. 3 (10.. · A(r. t) from (8.103) B . (10. (8.97) and. t) = −∂t V (r. using (8.68). One can readily verify.100.9).82) as follows F µν = ∂ µ Aν − ∂ ν Aµ .104) where mn = mn is the totally anti-symmetric three-dimensional tensor deﬁned in (5. Since ∂µ transforms like a covariant 4-vector. t) and current density J(r. i.17) using the identity (8.101).32).99) Since this equation must be true in any frame of reference the right hand side must vanish in all frames. in fact. one obtains for A(r.s. Using · ∂t A(r.69). transforms like a 4-vector. t) = −∂t V (r.83) and (10. i.s.e. also the l.h.100) Similarly. (10. this can be stated F k0 = −F 0k = E k .98) can be written. using (10. m.

t). respectively.4).97). 8. These equations show that under Lorentz transformations electric and magnetic ﬁelds convert into one another.111) One can readily prove that this equation is equivalent to the two inhomogeneous Maxwell equations (8.3: Relativistic Electrodynamics 299 The relationships (10. the components of the ﬁelds parallel and perpendicular to the velocity v which determines the Lorentz transformation.105) and (10. 8.107) 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 (10.105) of the tensor F µν one can conclude the property ∂ σ F µν + ∂ µ F νσ + ∂ ν F σµ = 0 (10. From the deﬁnition (10. .10.108) yields then the expressions for the transformed ﬁelds E and B .62) or. (10.106) In case that the Lorentz transformation Lµ ν is given by (10. 10.109) (10. B .113) which can be shown to be equivalent to the two homogeneous Maxwell equations (8. Noting ∂µ F µν = ∂µ ∂ µ Aν − ∂µ ∂ ν Aµ = ∂µ ∂ µ Aν − ∂ ν ∂µ Aµ = ∂µ ∂ µ Aν . where we used (10.63). B and E⊥ . one obtains Ey −v B Ez +v B 0 −Ex − √ 1 2z − √ 1 2y 1−v1 1−v1 Bz −v1 Ey By +v1 Ez √ 2 Ex 0 − √ 2 1−v1 1−v1 F µν = E −v B (10. by (10. t) and B(r. In a new frame of reference holds F µν = Lµ α Lν β F αβ (10. one can conclude from (10. The results can be put into the more general form E B = = E . B⊥ are.104) establishe the transformation behaviour of E(r.1.110) where E .3.102) ∂µ F µν = 4π J ν . equivalently. Maxwell Equations in Lorentz-Invariant Form One can express the Maxwell equations in terms of the tensor F µν in Lorentz-invariant form. E⊥ = B⊥ E⊥ + v × B √ 1 − v2 B⊥ − v × E = √ 1 − v2 (10.2).112) (10.103.

114) where d/dτ is given by (10. hence.5) where p = mv/ 1 − v 2 . Equation (10. also from the equation of motion (8. equivalent to (8.104).115) This equation follows.5).e. for µ = 1. (10.300 Lorentz Force Relativistic Quantum Mechanics One important property of the electromagnetic ﬁeld is the Lorentz force acting on charged particles moving through the ﬁeld.114) reads using (10.118) p· dt √ where we exploited the fact that according to p = mv/ 1 − v 2 holds p v. We want to √ demonstrate now that this equation is equivalent to the equation of motion (8.86).87)] and retain the deﬁnition E for the electric ﬁeld. The term on the r. dt E From this one can conclude. To avoid √ confusion we will employ in the following for the energy of the particle the notation E = m/ 1 − v 2 [see (10.114).86) dE q = p·E .h. i.103) dpx q = ( EEx + py Bz − pz By ) . 1 dE 2 1 dp 2 = = qp·E 2 dt 2 dt (10. dτ m √ Employing again (10. demonstrated that (10.. We have.117) (10. however.s. yields dpx = q dt Ex + (v × B)x (10.5). using (10. E = m/ 1 − v 2 . provides an alternative description of the action of the Lorentz force. The µ = 0 component of (10. e.67) and charge q dpµ q = pν F µν dτ m (10. . in fact. It holds for a particle with 4-momentum pµ as given by (10.120) is referred to as the Lorentz force. employing (10.114) is. of (10.120) (10.67).5) taking the scalar product with p dp = qp · E (10.119) which is the x-component of the equation of motion (8. dE q = p·E dτ m or with (10.86) and (10. hence. We want to express this force through the tensor F µν ..114) reads then. For the spatial components.g.116) (10.93).

w) = ρ(Lµ ν (ϑ. the charge-current density in case of Sect.125) def (10. (10.48) and (10. and which can be constructed following the procedure adopted for the function space representation of SO(3). K3 ) are the generators of L(ϑ. e. in as far as SO(3. w)) = ρ eϑ·J + w·K which we present in the form L(ϑ. For this purpose one replaces xµ in (10.3 or the bi-spinor wave function of electron-positron pairs in Sect.10.1) is a generalization (rotation + boosts) of the group SO(3).42) of SO(3). J2 . 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. K2 . i. However..121) covers solely the transformation behaviour of scalar functions. 10. In this section we will investigate the transformation properties of scalar functions ψ(xµ ). obey a diﬀerent transformation law. We need to emphasize that (10. We will denote the intended representation by L(ϑ.7. In analogy to the situation for SO(3) we seek an expression for ρ(Lµ ν ) in terms of an exponential operator and transformation parameters ϑ. (10.126) ρ ewk Kk − 1 1 .e. J3 ) and K = (K1 .4 Function Space Representation of Lorentz Group In the following it will be required to decribe the transformation of wave functions under Lorentz transformations.g. we seek an expression which corresponds to (10. (5. taken at the pairs of points (x µ = Lµ ν xν . We like to express now ψ (x µ ) in terms of the old coordinates xµ .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 . w) which correspond to the generators Jk and Kk in (10. w. xµ ). w) = exp ϑ · J + w · K . Accordingly.4: Function Space Representation of Lorentz Group 301 10.121) by (L−1 )µ ν xν and obtains ψ (xµ ) = ψ((L−1 )µ ν xν ) . The resulting expression should be a generalization of the function space representation (5.e.47).127) . Functions which represent 4-vectorsor other non-scalar entities.51) for the natural representation of the Lorentz group.123) This deﬁnition corresponds closely to the function space representation (5. ψ ∈ C∞ (4). (10.124) In this expression J = (J1 . 10. in the present case we exclude the factor ‘−i’ [cf.. i.122) This result gives rise to the deﬁnition of the function space representation ρ(Lµ ν ) of the Lorentz group (ρ(Lµ ν )ψ)(xµ ) = ψ((L−1 )µ ν xν ) .48) of SO(3). For such functions holds in the transformed frame ψ (Lµ ν xν ) = ψ(xµ ) (10..125)]. def (10.

0 1 (10. K ] [ Jk . (10. K obey the same Lie algebra (10. J ] [ Kk . x3 ∂1 ] = −x1 ∂2 + x2 ∂1 = J3 . [ Jk .128). −sinϑ1 x2 + cosϑ1 x3 ) = ψ(xµ ) + ϑ1 (x3 ∂2 − x2 ∂3 ) ψ(xµ ) + O(ϑ2 ) . x2 . J2 ] = J3 .130) ew1 K1 −1 = e−w1 K1 (10. x3 ) 2 = ψ(xµ ) + w1 (x1 ∂0 + x0 ∂1 ) ψ(xµ ) + O(w1 ) (10. sinhw1 x0 + coshw1 x1 . which we like to demonstrate for J1 and K1 . (10. the expression for K1 in (10.134) .128) eϑ1 J1 = e−ϑ1 J1 1 0 = 0 0 0 0 0 1 0 0 0 cosϑ1 sinϑ1 0 −sinϑ1 cosϑ1 (10.129) which yields for small ϑ1 ρ eϑ1 J1 ψ(xµ ) = ψ(x0 . J2 = x1 ∂3 − x3 ∂1 . x1 .49) as the generators of the natural representation. and [J1 . K2 ] = K3 : [ J1 . i. namely [J1 . J3 = x2 ∂1 − x1 ∂2 . obviously. The generators J . x1 ∂3 − x3 ∂1 ] = [x3 ∂2 . In order to evaluate (10. [K1 .302 One obtains J1 = x3 ∂2 − x2 ∂3 . 1 This result. One can determine similarly K1 starting from coshw1 sinhw1 sinhw1 coshw1 = 0 0 0 0 0 0 1 0 0 0 . reproduces the expression for J1 in (10. K ] = = k m Jm k m Jm = − k m Km .126).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. x1 ∂3 ] − [x2 ∂3 .e.132) and. obviously.131) This yields for small w1 ρ ew1 K1 ψ(xµ ) = ψ(coshw1 x0 + sinhw1 x1 . cosϑ1 x2 + sinϑ1 x3 . K2 ] = −J3 . J2 ] = [x3 ∂2 − x2 ∂3 .133) We demonstrate this for three cases.

303 (10.85-5.137).142) (10. x0 ∂2 + x2 ∂0 ] = [x0 ∂1 . x0 ∂2 + x2 ∂0 ] = [x3 ∂2 .. ∂3 in terms of ∂R .137) where K3 is given in (10. ∂Ω .87)]. cothΩ = 3 . Eqs. ∂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 . x3 as follows x0 = R coshΩ . Ω which are connected with x0 .143) .4: Function Space Representation of Lorentz Group [ K1 . 10. x2 + x2 + x2 1 2 3 in the case of SO(3) and (x0 )2 − (x3 )2 in case of transformation (10. K2 ] = [x3 ∂2 − x2 ∂3 . We note tanhΩ = ∂R x0 = . x0 ∂2 ] = −x2 ∂1 + x1 ∂2 = − J3 . In the following we consider solely the case x0 ≥ 0. In both cases the radial coordinate is the quantity conserved under the transformations. 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 . The relationships (10.125) in the case of a one-dimensional transformation of the type L(w3 ) = exp w3 K3 . a relationship which can also be stated R = and + − (x0 )2 − (x3 )2 (x0 )2 − (x3 )2 if x0 ≥ 0 if x0 < 0 (10.139.135) (10.138) x3 x0 . i. K2 ] = [x0 ∂1 + x1 ∂0 .e. [ J1 . x2 ∂0 ] − [x1 ∂0 .140) allow one to ∂ ∂ express the derivatives ∂0 . One-Dimensional Function Space Representation The exponential operator (10. (5.128).136) (10. x2 ∂0 ] − [x2 ∂3 . (10.10. For this purpose one expresses K3 in terms of hyperbolic coordinates R.139) x3 = R sinhΩ (10.140) x0 x The transformation to hyperbolic coordinates closely resembles the transformation to radial coordinates for the generators of SO(3) in the function space representation [cf.141) (10. can be simpliﬁed considerably. x0 ∂2 ] = x3 ∂0 + x0 ∂3 = K3 . 0 ∂cothΩ ∂x x = ∂R x0 = − ∂x3 R (10.

but certainly is short and. ˆ pµ =⇒ pµ = i(∂t . In natural units the Klein–Gordon equation (10.149) where ψ(xµ ) is a scalar. momentum.304 Inserting these results into the deﬁnition of K3 in (10.122). ) = i ∂µ . We will provide a ‘derivation’ of these two equations which stem from the historical development of relativistic quantum mechanics. − ) = i ∂ µ . complex-valued function.148) ∂ µ ∂µ + m2 (10.144) The action of the exponential operator (10. (10. The latter property implies that upon change of reference frame ψ(xµ ) transforms according to (10. namely the Klein–Gordon and the Dirac equations.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.128) yields K3 = x0 ∂3 + x3 ∂0 = ∂ .145) ∂Ω 10. In the following we will employ so-called natural units = c = 1. Free Particle Case A quantum mechanical description of a relativistic free particle results from applying the correspondence principle. extremely useful. and (time)−1 all have the same dimension.151) is called the Klein-Gordon equation. The historic route to these two equations.121. Further below we will provide a more systematic. is not very insightful. which allows one to replace classical observables by quantum mechanical operators acting on wave functions. in 4-vector notation reads pµ =⇒ pµ = i (∂t . In the position representation the correspondence principle states ˆ E =⇒ E = − ∂t i ˆ p =⇒ p = i which. therefore. mass.147) (10. (length)−1 .150) . 10. The partial diﬀerential equation (10. representation theoretic treatment.146) Applying the correspondence principle to (10.92) one obtains the wave equation − or 2 2 µ ∂ ∂µ ψ(xν ) = m2 ψ(xν ) ψ(xν ) = 0 . ∂Ω Relativistic Quantum Mechanics (10. however.151) reads ∂µ ∂ µ + m2 ψ(xµ ) = 0 (10. ˆ (10. (10. In these units the quantities energy.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 ) .

t) 1 [ ψ ∗ (r. t) = 0 (10.154) describes the current density connected with motion of the particle probability distribution.152) which has the same form as the Klein–Gordon equation in the original frame. t) ψ(r.153) is associated with a conservation law for particle probability ∂t ρS (r. t)∂t ψ ∗ (r.159) (10. (10. t) + where ρS (r. t) − ψ(r. t) ψ(r. Current 4-Vector Associated with the Klein-Gordon Equation As is well-known the Schr¨dinger equation of a free particle o i∂t ψ(r.160) ∂t (ψ ∗ ∂t ψ which corresponds to − ψ∂t ψ∗) + · (ψ ψ∗ ∂t ρKG (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 (10.150) becomes ∂µ ∂ µ + m2 ψ (x µ ) = 0 (10. t)∂t ψ(r. t) = 0 i ( ψ ∗ (r. t) + where ρKG (r. t) = (10.5: Klein–Gordon Equation or 2 ∂t − 2 305 + m2 ψ(xµ ) = 0 . t) jKG (r. t) − ψ(r. In order to obtain the conservation law connected with the Klein–Gordon equation (10.154). t) (10. t) ) 2m 1 = ( ψ ∗ (r.155) describes the positive deﬁnite probability to detect a particle at position r at time t and where jS (r.157) 2m which is equivalent to (10. t) = ψ ∗ (r. t) ] 2mi (10.161) . t) = · jKG (r. t) ψ ∗ (r.150) one considers Im ψ ∗ ∂µ ∂ µ + m2 ψ = 0 (10. t) ) . t) − ψ(r.151) One can notice immeadiately that (10. t) ψ(r. t) ψ ∗ (r. 2mi (10.156) · jS (r. This follows from the fact that ∂µ ∂ µ and m2 are scalars.10. and that (as postulated) ψ(xµ ) is a scalar.150) is invariant under Lorentz transformations. t) = − 1 2m 2 ψ(r. Under Lorentz transformations the free particle Klein–Gordon equation (10.158) which yields 2 2 ψ ∗ ∂t ψ − ψ∂t ψ ∗ − ψ ∗ 2ψ +ψ ψ∗ − 2ψ∗ = ψ) = 0 (10.

it had been realized later. not of particle probability. λ) = λ ψo (p.165) This result together with (10.168) .. 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) . o namely. according to (??) is decribed by the r–independent wave function ψo (p = 0. ±) = ± 2 m2 + po (10. t).150) are actually determined by po and by the choice of sign ±. λ) associated with the corresponding wave functions ψo (p. and of negative energies E ≤ −m. λ) actually describes charge density rather than probability. corresponding to λ = −.p ei(p·r − λEo (p)t) Eo (p) = The spectrum of the Klein–Gordon equation (10. We denote this by summarizing the solutions as follows ∂µ ∂ µ + m2 ψo (p. is actually that of a charge density. The proper interpretation of ρKG (r. (10. m 2 2 E0 = m2 + po . λ|xµ ) = Nλ. λ|xµ )ψo (p. in that the expression for ρKG is not positive deﬁnite.166) ψo (p. λ|xµ ) (10.151) yields 2 2 Eo − p0 − m2 ψ(r.162) shows that the solutions of the free particle Klein-Gordon equation (10. Generating a Solution Through Lorentz Transformation A particle at rest. t) = 0 (10. corresponding to λ = +. The proper interpretation of the two cases is that the Klein–Gordon equation describes particles as well as anti-particles.154).166) is Eo (p) ∗ ρKG (p. λ = ± (10.167) m which is positive for λ = + and negative for λ = −.306 Relativistic Quantum Mechanics This conservation law diﬀers in one important aspect from that of the Schr¨dinger equation (10. λ|xµ ) = 0 2 m2 + po . (10.150) is a continuum of positive energies E ≥ m.e. and the density ρKG (p. Today. λ = ± . with p = 0.163) which results in the expected [see (10. the anti-particles carry a charge opposite to that of the associated particles. for example pions. 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. i. λ|xµ ) = N e−iλmt .162) Inserting this into the Klein–Gordon equation (10. The density ρKG (p.164) The corresponding energy is Eo (po . the Klein-Gordon equation is considered as a suitable equation to describe spin–0 particles.93] dispersion relationship connecting E0 . p0 .161) and (10. λ|xµ ) according to (10.

t). Such wave function should be generated by applying the Lorentz transformation in the function space representation p (10.10. t)) (10.177) .173) (10. For this purpose we consider the wave function for a particle moving with momentum velocity v in the direction of the x3 –axis. in fact. λ = +|xµ ) = N ei(px 3 − Eo (p)x0 (10.170) The coordinate transformation (10.h. In case of λ = −. This yields. (10. (10. Ω.166). λ = −|xµ ) = N ei(px 3 + Eo (p)x0 .s.e.176) 10. for negative energy solutions. i.138) to replace t = x0 by hyperbolic coordinates R. of (10. p = − mv/ 1 − v 2 as the momentum of the particle relative to the moving frame and √ m = 1 − v2 m2 = 1 − v2 m2 + m2 v 2 = 1 − v2 m2 + p2 = Eo (p) (10.168).138) and the relationships (10. t)) . (10. using (10.e.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.61) yield for this expression −iλ √ m mv x0 − iλ √ x3 . for the wave function (10.174) which agrees with the expression given in (10.168).166)] of the particle.f.. one has to interprete p = mv/ 1 − v 2 (10. Aµ (xµ ) = (V (r.175) as the momentum of the particle and one obtains L(w3 )ψo (p = 0. (10. A(r. 2 2 1−v 1−v (10.171) One can interpret then for λ = +.169) −iλ ( m coshw3 ) ( R coshΩ ) − iλ ( m sinhw3 ) ( R sinhΩ ) .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. λ|xµ ) = = exp w3 ∂ ∂Ω N e−iλmRcoshΩ 3) N e−iλmRcosh(Ω+w . i.145) choosing m = sinhw3 . t).. In case of λ = + one obtains ﬁnally L(w3 )ψo (p = 0.172) as the energy [c. L(w3 )ψo (p = 0. −A(r. for positive energy solutions.169) The addition theorem of hyperbolic functions cosh(Ω + w3 ) = coshΩ coshw3 + sinhΩ sinhw3 allows us to rewrite the exponent on the r.

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. t))2 ψ(r. (10. t) 2 (10.150) −g µν (i∂µ )(i∂ν ) + m2 ψ(xµ ) = 0 . A) or Aµ = (V. we deﬁne ψ(r. (10.1: Coupling of a particle of charge q to an electromagnetic ﬁeld described by the 4-vector potential Aµ = (V. t) = −i − q A(r.182) and seek an equation for Ψ(r.e. t) = e−imt Ψ(r. t). momenta for classical particles and the respective operators for quantum mechanical particles in the manner shown. i∂t and −i in (10. m2 ) ψ(xµ ) = 0 .e. it’s rest energy. which couples a particle of charge q to an electromagnetic ﬁeld described through the potential Aµ (xν ).. i. albeit in a form. 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 (10.179) (10.93). We will also assume.150). in keeping withnthe non-relativistic limit. i.178) According to the replacements in Table 10. For this purpose we split-oﬀ a factor exp(−imt) which describes the oscillations of the wave function due to the rest energy. and focus on the remaining part of the wave function. t) . t) . in ..69)] one replaces the quantum mechanical operators. According to the so-called minimum coupling principle the presence of the ﬁeld is accounted for by altering energy.147). according to the rules shown in Table 10.180) + m2 ψ(r. In terms of space-time derivatives this reads (i∂t − qV (r. (10. A) free quantum particle quantum particle in ﬁeld (V. that the mass m of the particle. According to the principle of minimal coupling [see (10.1 this becomes g µν (i∂µ − qAµ )(i∂ν − Aν ) ψ(xµ ) = m2 ψ(xµ ) which can also be written g µν πµ πν − .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. For this purpose one writes the Klein-Gordon equation (10. (10. i. See also Eq.1. is much larger than all other energy terms.308 free classical particle classical particle in ﬁeld (V..e. −A).

181) reads then i ∂t Ψ(r. atoms in which one or more electrons are replaced by π − mesons.10.181) 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. π − mesons are generated through inelastic proton–proton scattering p + p −→ p + p + π − + π + . Pionic Atoms To apply the Klein–Gordon equation (10. (10.. (10. This application demonstrates that the Klein–Gordon equation describes spin zero particles. spin-0 mesons..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.2) of a non-relativistic spin-0 particle o moving in an electromagnetic ﬁeld.6: Klein–Gordon Equation with Electromagnetic Field particular. | i∂t Ψ | << m . of (10. A stationary state of the Klein–Gordon equation is described by a wave function ψ(xµ ) = ϕ(r ) e−i t . t)]2 + qV (r. however.. (10. t) = ˆ [p − q A(r.g. i.e. larger than |i∂t Ψ/Ψ| and alrger than qV . ﬁltered out of the beam and ﬁnally fall as slow pions onto elements for which a pionic variant is to be studied. Ψ |q V | << m .h. The approximation is justiﬁed on the ground of the inequalities (10.183) The term on the l. t) = − Ze and A(r. identical to the Schr¨dinger equation (10.189) .184) where we neglected all terms which did not contain factors m. To ‘manufacture’ pionic atoms. (10. The process of π − meson capture involves the so-called Auger eﬀect.188) Inserting this into (10. 309 (10. t) (10.186) then are slowed down. t) 2m Ψ(r.181) yields (we assume now that the Klein–Gordon equation describes a 2 particle with mass mπ and charge −e) for qV (r. i.185) This is. t) ≡ 0 r + Ze2 r 2 + 2 − m2 π ϕ(r ) = 0 . The Klein-Gordon equation (10. 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− .183).181) to a physical system we consider pionic atoms. e.s.e.

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 . φ. r r r r sin θ r sin θ Ze2 2 r ) (10. . (10.195) can be made complete if one determines λ such that λ( ) (λ( ) + 1) = The suitable choice is λ( ) = − 1 + 2 ( + 1 2 ) − Z 2 e4 2 (10. 1. (10. . (10. n − 1 . φ) (10. 2n2 = 0. .191) ˆ The operator L2 in this equation suggests to choose a solution of the type ϕ(r ) = where the functions Y in (10.194) and (10. .194) Bound state solutions can be obtained readily noticing that this equation is essentially identical to 2 that posed by the Coulomb problem (potential − Ze ) for the Schr¨dinger equation o r d2 − dr2 ( + 1) 2mπ Ze2 + + 2mπ E r2 r R (r) = 0 (10. The similarity of (10.192) leads then to the ordinary diﬀerential equation d2 − dr2 ( + 1) − Z 2 e4 2 Ze2 + + r2 r 2 − m2 π R (r) = 0 .192) ˆ are spherical harmonics.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 .. . (10. i. . this quantity deﬁnitely must be an integer.196) In this expression the number n deﬁned through n = n− −1 (10..e. (10. i. θ. φ) = ( + 1) Y m (θ.191) m (θ.195) The latter problem leads to the well-known spectrum En = − mπ (Ze2 )2 .e. .310 Relativistic Quantum Mechanics Because of the radial symmetry of the Coulomb potential we express this equation in terms of spherical coordinates r.199) ( + 1) − Z 2 e4 . the eigenfunctions of the operator L2 ˆ L2 Y m (θ. 2.193) (10. φ) .198) . n = 1.197) counts the number of nodes of the wave function. φ) R (r) Y r m (θ.

204) reads 1 (10. which corresponds to a bound = 0. n = 0.196) if one makes the replacement 2 values which can be obtained E −→ One obtains − m2 π . n − 1 m = − . . 1. . . . = 0. . . .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 ). EKG = Using (10. i.202) Solving this for state) yields (choosing the root which renders = 1 + mπ Z 2 e4 ( n + λ( )+1 )2 ≤ mπ .e. . . 2 e2 −→ e2 mπ .6: Klein–Gordon Equation with Electromagnetic Field and one can write (10. (10. (10. + (10. .194) d2 λ( ) ( λ( ) + 1 ) 2 Ze2 − + + dr2 r2 r 2 311 − m2 π R (r) = 0 . 2mπ 2 (n + λ( ) + 1)2 (10. . . . . results in the (10.201) mπ Z 2 e4 m2 − m2 π π = − .199) and deﬁning EKG (n. 2. . . 10.200) The bound state solutions of this equation should correspond to from (10. Introducing α = Z 2 e4 and 1 β = + 2 (10. . m) = mπ 1 + Z 2 e4 (n− − 1 2 + ( + 1 )2 − Z 2 e4 )2 2 n = 1.205) 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. − + 1.203) spectrum . . . 1. .197. 2mπ 2 −→ λ( ) .

However. but for a long time to no avail. Many attempts were made by theoretical physicists to ‘linearize’ the square root operator in (10. . however. the Klein–Gordon equation had been rejected since it did not yield a positive-deﬁnite probability density.2 particles.208.208). one with a positive-deﬁnite density. It should feature then a diﬀerential operator of the type D = iγ µ ∂µ . 10. a feature which is connected with the 2nd order time derivative in this equation.. Finally. The latter spectrum shows. 4n (10.312 ≈ From this results for (10. is related to the equation E 2 = m2 + p 2 of the classical theory which involves a term E 2 . Application of the correspondence principle (10. 2n2 2βn3 8n4 4n4 (10... and the third term gives the leading relativistic correction.146) leads to the wave equation i∂t Ψ(r. however. a splitting of the six n = 2.204) EKG (n. the Dirac equation introduced below. are unsatisfactory since expansion of the square root operator involves all powers of the Laplace operator. 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 latter term agrees with observations of pionic atoms. i.208) These two equation can be combined i∂t + m2 − 2 i∂t − m2 − 2 Ψ(r. it does not agree with observations of the hydrogen spectrum. the second term the non-relativistic energy.207) Here the ﬁrst term represents the rest energy. It must be pointed out here that does not denote energy.208. 10. for example.e. A more satisfactory Lorentz–invariant wave equation. Also. m) ≈ m − mZ 2 e4 mZ 4 e8 − 2n2 2n3 1 + − 1− Relativistic Quantum Mechanics α α2 α2 α2 − + + + O(α3 ) . arises because the Klein–Gordon equation. 10. = 1 states into groups of two and four degenerate states. t) = ± m2 − 2 Ψ(r.e.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 .209) which. In order to describe electron spectra one must employ the Lorentz-invariant 1 wave equation for spin.206) 1 2 3 + O(Z 6 e12 ) . This derivative. (10. would have only a ﬁrst order time derivative. the π − meson is a pseudoscalar particle.7 The Dirac Equation Historically. His solution . in turn. is identical to the two equations (10. t) (10. through the correspondence principle.209). Equations (10.209). but in the present case rather the negative of the energy. Dirac succeeded. in fact. i. but not an operator iγ · as suggested by the principle of relativity (equivalence of space and time). the wave function changes sign under reﬂection. t) . i.

10.1 2 and anti-particles. Instead. namely. the 4-dimensionsional representation discovered by Dirac involved new physical properties. i.210) where P = iγ µ ∂µ . this would lead to the two wave equations (P − m)Ψ = 0 and (P + m)Ψ = 0 which have a ﬁrst order time derivative and. 3 are anti-hermitian .209). unspeciﬁed algebraic objects γ µ and γ ν .211) into (10. . Initially. the operator of the Klein–Gordon equation ∂µ ∂ µ + m2 = −( P + m ) ( P − m ) (10.208). The discovery by Dirac. Properties of the Dirac Matrices Let us now trace Dirac’s steps in achieving the linearization of the ‘square root operator’ in (10. γ ν ]+ = 2 g µν (10. the realization that the linearization can be carried out only if one assumes a 4-dimensional representation of the coeﬃcients γ µ . the quantities γ 0 . 2.213) reads (γ µ )2 = 1 µ = 0 −1 µ = 1. therefore. (10.66). 2. We seek to identify the coeﬃcients γ µ .. but did not commute the. yet to be discovered. Accordingly.211) Obviously. 3 has imaginary eigenvalues ±i.212) where we have changed ‘dummy’ summation indices. Inserting (10. one route to important discoveries in physics is the creation of new mathematical descriptions of nature.7: Dirac Equation 313 to the problem involved an ingenious step.215) . For µ = ν condition (10. γ j .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. . achieved through a beautiful mathematical theory. therefore.213) We want to determine now the simplest algebraic realization of γ µ . so far.e. . However. strengthens the believe of many theoretical physicists today that the properties of physical matter ultimately derive from a. γ 3 can only be realized by d×d–matrices requiring that the wave function Ψ(xµ ) is actually a d–dimensional vector of functions ψ1 (xµ ). γ 2 . according to (10. j = 1. spin. exploited ∂µ ∂ν = ∂ν ∂µ . j = 1. are associated with a positive-deﬁnite particle density. beautiful mathematical theory and that. 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 . It turns out that no 4-vector of real or complex coeﬃcients can satisfy these conditions. Starting point is to boldly factorize. it was assumed that the 4-dimensional space introduced by Dirac could be linked to 4vectors. ψ2 (xµ ). 2. (10. (10. this was not so. these descriptions ultimately merging with the true theory of matter. ψd (xµ ). In fact. quantities with the transformation law (10. γ 1 . 3 . one can impose the condition γ0 is hermitian .214) From this follows that γ 0 has real eigenvalues ±1 and γ j .

We will argue further below that the choice f γ µ . σ 3 for which.220) where Ψ(xµ ) represents a 4-dimensional wave function.219) Using property (10. γj = 0 −σ j σj 0 .218) The Pauli matrices allow one. in fact. to construct four matrices γ µ for the next possible dimension d = 4.217) Obviously. The Adjoint Dirac Equation The adjoint equation is Ψ† (xµ ) i(γ µ )† ∂µ + m ← ← (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 ) . (10.224) .223) Inserting this into (10..e. 1 (10.314 For µ = ν (10. ψ3 . is implied by (10. in fact. A proper choice is γ0 = 1 1 0 0 −1 1 .213) is satisﬁed.222) and multiplication from the right by γ 0 yields the adjoint Dirac equation Ψ† (xµ ) γ 0 iγ µ ∂µ + m ← =0. This property can also be written (γ µ )† = γ 0 γ µ γ 0 . For d = 2 there exist only three anti-commuting matrices. 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. is unique. however. One can readily show using the hermitian property of the Pauli matrices (γ 0 )† = γ 0 and (γ j )† = −γ j for j = 1. namely the Pauli matrices σ 1 . ψ4 ) and where ∂µ denotes the diﬀerential operator ∂ µ operating to the left side. rather than a scalar wave function. σ j σ k = − σ k σ j for j = k .215). (10. σ 2 .221) =0 (10.216) i. 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. (10.213) reads γ µ γ ν = −γ ν γ µ . except for similarity trasnformations. the γ µ are anti-commuting. rather than to the right side. Relativistic Quantum Mechanics (10.222) ∗ ∗ ∗ ∗ where we have deﬁned Ψ† = (ψ1 . holds 2 σj = 1 . (10. ψ2 . 2. as long as det(γ µ ) = 0 the dimension d of the square matrices γ µ must be even so that (−1)d = 1.218) of the Pauli matrices one can readily prove that condition (10. 3 which.

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.e.222. . 2.1: Derive (refeq:Dirac-intro20a) from (10.229) The similarity transformation (10.233) The eigenstates and eigenvalues of H correspond to the stationary states and energies of the particles described by the Dirac equation. j = 1.10. γj = ˜ 0 σj 1 S = √ 2 −σ j 0 1 1 1 1 1 −1 1 1 (10.7: Dirac Equation Similarity Transformations of the Dirac Equation . (10. 10. 3 and ˆ β = 1 1 0 0 −1 1 . (10. αj = ˆ 0 σj σj 0 .228) (10. and γ0 = ˜ 0 1 1 1 0 1 . γ ˜ Exercise 10. j = 1. [˜ µ . (10. ˆ ˆ ˆ Ho = α · p + β m . j = 1. (10.227) A representation often adopted beside the one given by (10. γ ν ]+ = 2 g µν .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 ) .213).7. 2. 2.227).Chiral Representation 315 The Dirac equation can be subject to any similarity transformation deﬁned through a non-singular ˜ 4 × 4–matrix S. 3 . (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. t) = 0 (10..226) (10.225) .227) leaves the algebra of the Dirac matrices unaﬀected and commutation property (10.231) ˆ whereα has the three components αj .213) still holds. 3 . i.219) is the socalled chiral representation deﬁned through ˜ Ψ(xµ ) = S Ψ(xµ ) .

any product 1 of the dj ’s can be brought to the form (10. 27. Γ±15 = ±d2 d3 d4 Γ±16 = ±d1 d2 d3 d4 (10.. Γ±11 = ±d3 d4 Γ±12 = ±d1 d2 d3 . 2.e. To complete the proof in this section the reader may consult Miller ‘Symmetry Groups and their Application’ Chapter 9. form a Cliﬀord algebra C3 . The associative algebra generated by d1 . i.. 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 .237) which are the ordered products of the operators ±1 and d1 . The representations of Cliﬀord algebras Cm are well established.237) by means of the property (10. Γ±7 = ±d1 d3 . The three Pauli matrices also obey the property (10.235) can diﬀer only with respect to unitary similarity transformations. For example. d2 = . i. satisfy the anti-commutation property dj dk + dk dj = 2 for j = k 0 for j = k Relativistic Quantum Mechanics d4 = γ 0 (10. There are (including the diﬀerent signs) 32 elements in G4 which we deﬁne as follows Γ±1 = ±1 1 Γ±2 = ±d1 . d4 is called a Cliﬀord algebra C4 . in case of C4 . The reader may . d4 .H. Γ±13 = ±d1 d2 d4 .235). Γ±4 = ±d3 . d3 . One can conclude then that also any set of 4 × 4–matrices obeying (10. Γ±5 = ±d4 Γ±6 = ±d1 d2 . Γ±8 = ±d1 d4 . to Dirac matrices.Good. Obviously. .235) as can be readily veriﬁed from (10.6 and R.238) These elements form a group since obviously any product of two Γr ’s can be expressed in terms of a third Γr .236) where ‘⊗’ denotes the Kronecker product between matrices. j = 1.Phys.235) and. Since the Γj are hermitian the similarity transformations are actually given in terms of unitary transformations.234) (10.213). j1 < j2 < · · · < js s ≤ 4 (10. . d4 = 1 0 0 −1 1 0 0 −1 0 1 1 0 0 i −i 0 (10.Mod. the matrix elements of C = A ⊗ B are Cjk. Γ±9 = ±d2 d3 Γ±10 = ±d2 d4 . Γ±14 = ±d1 d3 d4 . Rev. hence. 3 .e. The Cliﬀord algebra C4 entails a subgroup G4 of elements ± dj1 dj2 · · · djs . page 187. d2 .316 Cliﬀord Algebra and Dirac Matrices The matrices deﬁned through dj = iγ j . This property extends then to 4 × 4–matrices which obey (10. m = Aj Bkm . (1955).213). The representations of this group are given by a set of 32 4 × 4–matrices which are equivalent with respect to similarity transformations. Γ±3 = ±d2 .

236) with the Dirac representation (10. 2.2: Demonstrate the anti-commutation relationships (10.80) by Lµ ρ and using (10.. Lorentz Transformation of the Bispinor State Actually.213. 3 has imaginary eigenvalues ±i and can be represented by an anti-hermitian matrix. in a moving frame should hold iγ ∂µ − m Ψ (x ) = 0 ..239) Form invariance implies that the matrices γ µ should have the same properties as γ µ . Multiplication and summation of (10.219). it must hold γ µ = γ µ . i. 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.240) . we do not know yet how to transform the 4-dimensional wave function Ψ and the Dirac matrices γ µ .218) of the Pauli matrices σ j .6).4: Show that from (10.. 10. and derivatives according to (10.80). We can. the form of the Dirac equation is identical in equivalent frames of reference.6). exist.e. The transformation is assumed to be linear and of the form Ψ (x ) = S(Lµ ν ) Ψ(xµ ) µ µ (10. Exercise 7. 10. Exercise 7. transform coordinates and derivatives of the Dirac equation.3: Demonstrate the anti-commutation relationships (10. However. i. Exercise 7.218) of the Dirac matrices γ µ .e.. x = Lµ ν xν . these properties determine the matrices γ µ uniquely. i.8 Lorentz Invariance of the Dirac Equation We want to show now that the Dirac equation is invariant under Lorentz transformations. i. Inﬁnitesimal Bispinor State Transformation We want to show now that a suitable transformation of Ψ(xµ ) does. (10. The latter transformations imply that coordinates transform according to (10.e.8: Lorentz Invariance of the Dirac Equation 317 also want to establish the unitary transformation which connects the Dirac matrices in the form (10.214) follows that γ 0 has real eigenvalues ±1 and can be represented by a hermitian matrix. a result one could have also obtained by applying the chain rule to (10. namely.76) yields ∂ρ = Lν ρ ∂ ν . in fact. µ µ (10.e. in frames connected by Lorentz transformations. Hence.e. therefore. j = 1.10. x µ = Lµ ν xν (10.241) .215).. Except for a similarity transformation. i. and γ j . in a moving frame holds iγ µ ∂µ − m Ψ (x ) = 0 .

(10..250) σµν = −σνµ . One can.243) from the left by Lσ ρ gσν yields S(Lη ξ )γ µ S −1 (Lη ξ ) gρµ = Lσ ρ gσν γσ .12) in the form Lν µ gνσ Lσ ρ = gµρ = gρµ . hence. µ (10. follows S(Lη ξ )γρ S −1 (Lη ξ ) = Lσ ρ γσ . (10. The proper starting point for a constructiuon of S(Lη ξ ) is actually (10.245) The construction of S(Lη ξ ) will proceed using the avenue of inﬁnitesimal transformations. µ g ρν = µν and. Multiplication of this property by g from the right yields ( g)T = − g.248) Here σµν denote 4 × 4–matrices operating in the 4-dimensional space of the wave functions Ψ.318 Relativistic Quantum Mechanics where S(Lµ ν ) is a non-singular 4 × 4–matrix.243) in a form in which the Lorentz transformation in the form Lµν is on the r. The elements of g are.249) from which we can conclude σµν = −σµν νµ = −σνµ i. νµ (10. of the equation.221) upon Lorentz transformation yields iS(Lη ξ )γ µ S −1 (Lη ξ ) Lν µ ∂ν − m Ψ (x ) = 0 . in the expression of the inﬁnitesimal transformation ρ Lµν = g µν + µν (10. Ob1 viously. .e. (10.243) We want to determine now the 4 × 4–matrix S(Lη ξ ) which satisﬁes this condition.s. hence. One can ﬁnally conclude multiplying both sides by g ρµ S(Lη ξ )γ µ S −1 (Lη ξ ) = Lν µ γν .244) from which. using gρµ γ µ = γρ . it must hold (10. S( µν ) should not change its value if one replaces in its argument µν by − νµ .e. however. i.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 .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. the transformation (10.h.247) can be expanded S( µν i ) = 1 − σµν 1 4 µν (10.. g is an anti-symmetric matrix.246) (10.247) the inﬁnitesimal matrix µν is anti-symmetric. 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. Multiplication of (10. state that the Dirac equation (10. The inﬁnitesimal transformation S(Lρ σ ) which corresponds to (10. For this purpose we exploit (10. It holds then S( µν i ) = 1 − σµν 1 4 µν µν = S(− νµ i ) = 1 + σµν 1 4 µν . We had introduced in (10.241) implies a similarity transformation Sγ µ S −1 .

γj ] = γ ˜ 2 −iσ j 0 0 iσ j . like the expression on the l. σjk = [˜j . β [ σαβ . 3 . Algebra of Generators of Bispinor Transformation We want to construct the bispinor Lorentz transformation by exponentiating the generators σµν .s.216) of the Dirac matrices. γj = ˜ 0 −σ j σj 0 .s.. γ µ ]− = 2i ( δ µ β γα − δ µ α γβ ) . γβ ]− 2 (10.257) . Furthermore.254) is σαβ = i [ γα . of (10. Comparing this with the l.252) also as a sum over both indices of αβ .246) results then in a condition for the generators σµν − i ( σαβ γ µ − γ µ σαβ ) 4 αβ = νµ γν . In this representation the Dirac matrices γµ = (˜ 0 . this condition can actually be expressed through six independent conditions..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. must be symmetric with respect to interchange of the indices α and β.10.251) S −1 ( µν ) = 1 + σµν µν 1 4 Inserting (10. (10.h. (10. 10.h. namely. j = 1. of (10.253) 1 2 αβ ( δ µ β γα − δ µ α γβ ) . β and operate in the same space as the Dirac matrices.h.e.h. 10. i. 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 (10.5: Show that the σαβ deﬁned through (10.254) The proper σαβ must be anti-symmetric in the indices α.s. (10. 10. In fact.213.252) results in the condition for each α.252) Since six of the coeﬃcients αβ can be chosen independently.256) One can readily verify that the non-vanishing generators σµν are given by (note σµν = −˜νµ . For this purpose we need to verify that the algebra of the generators involving addition and multiplication is closed.s. For this purpose we inspect the properties of the generators in a particular representation.229).248. (10. −γ ) are ˜ γ ˜ γ0 = ˜ 0 1 1 1 0 1 .251) into (10. the expression on the r.254). Exercise 7.228. γk ] = ˜ γ ˜ jk σ 0 0 σ . For this purpose one needs to express formally the r. 2. (10.255) which can be demonstrated using the properties (10. ˜ ˜ σ only six generators need to be determined) σ0j = ˜ i [˜0 .255) satisfy condition (10. a solution of condition (10. the chiral representation introduced above in Eqs.

any representation of the generators. . be able to state S in terms of the same parameters w and ϑ as the Lorentz transformation in (10.44) 0 −w1 −w2 −w3 w 0 ϑ3 −ϑ2 = 1 w2 −ϑ3 0 ϑ1 w3 ϑ2 −ϑ1 0 µν (10. (10. not necessarily inﬁnitesimal.259) becomes in the chiral representation ˜ S(w. i. Finite Bispinor Transformation The closedness of the algebra of the generators σµν deﬁned through (10.51) and S.51). 2 0 −(w + iϑ) · σ 13 + σ23 ˜ 23 ) (10.261) We note that this operator is block-diagonal. therefore.262) This expression allows one to transform according to (10. In order to construct an explicit expression of S in terms of w and ϑ we employ again the chiral representation. the algebra of these generators is closed under addition and multiplication. µν in the exponential form i S = exp − σµν 4 µν . since both operations convert block-diagonal operators A 0 0 B (10. We can ﬁnally note that the closedness of the algebra of the generators σµν is not aﬀected by similarity transformations and that.255) yields a closed algebra.e.248) allows us to write the transformation S for any.258) again into block-diagonal operators.320 Relativistic Quantum Mechanics Obviously. In fact. and since the algebra of the Pauli matrices is closed.259) yields the desired connection between the Lorentz transformation (10. the representation (10..259) We had stated before that the transformation S is actually determined through the Lorentz transformation Lµ ν . 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 = . in particular. therefore. ϑ) = e 2 (w − iϑ)·σ 0 − 1 (w + iϑ)·σ 0 e 2 1 (10. one can express the tensor µν through w and ϑ using µν = µ ρ g ρν and the expression (10. One should. Since such operator does not change its block-diagonal form upon exponentiation the bispinor transformation (10.241) bispinor wave functions from one frame of reference into another frame of reference.260) Inserting this into (10.

ϑ)˜ ˜ γ 0 ˜† 0 = 0 1 1 1 0 1 e 2 (w + iϑ)·σ 0 − 1 (w − iϑ)·σ 0 e 2 1 0 1 1 1 0 1 . For our proof we note ﬁrst ˜ ˜ S −1 (w. The conservation law (10. j µ must transform like a contravariant 4-vector. Obviously.267) has the desired property of being positive deﬁnite.221) from the left by Ψ† (xµ )γ 0 .270) We conclude that (10. Multiplying (10.269) One can readily show that the same operator is obtained evaluating γ S (w. ϑ) = S(−w. For this purpose we prove ﬁrst the relationship S −1 = γ 0 S † γ 0 . and addition yields Ψ† (xµ )γ 0 The last result can be written ∂µ Ψ† (xν )γ 0 γ µ Ψ(xν ) = 0 .h. Since ∂µ transforms like a covariant 4-vector. (10. (10. This transformation behaviour can also be deduced from the transformation properties of the bispinor wave function Ψ(xµ ).224) from the right by Ψ(xµ ).271) .263) Starting point are the Dirac equation in the form (10.266) ρ(xµ ) = Ψ† (xµ )Ψ(xµ ) = s=1 |ψs (xµ )|2 (10.268) We will prove this property in the chiral representation. We will now determine the relationship between the ﬂux j µ = Ψ (x )γ 0 γ µ Ψ (x ) † µ (10. i. the conservation law (10.263) allows one to conclude that j µ must transform like a contravariant 4-vector as the notation implies.221) and the adjoint Dirac equation (10. (10.. j) = Ψ† (xµ )γ 0 γ µ Ψ(xµ ) .224). −ϑ) = e− 2 (w − iϑ)·σ 0 1 (w + iϑ)·σ 0 e2 1 (10. the property applies then in any representation of S.263) obviously is a scalar under Lorentz transformations and that the left hand side must then also transform like a scalar.268) holds for the bispinor Lorentz transformation.10.e. of (10.264) (10. The time-like component ρ of j µ 4 iγ µ ∂µ + iγ µ ∂µ ← Ψ(xµ ) = 0 .263) does hold. The reason is that the r. in fact.8: Lorentz Invariance of the Dirac Equation Current 4-Vector Associated with Dirac Equation 321 We like to derive now an expression for the current 4-vector j µ associated with the Dirac equation which satisﬁes the conservation law ∂µ j µ = 0 . (10.265) (10. (10. for j µ (xµ ) = (ρ.s.

232. The operators commute also with Ho in (10.283)]. not only as the most simple demonstration of the theory. As pointed out. Like in non-relativistic quantum mechanics the free particle wave function plays a central role. 10. These operaˆ tors. The additional degrees of freedom described by the four components of the bispinor wave function require. Combining this with (10. as just mentioned. however.e. additional characterizations. In the following we want to provide this characterization.272) results in the expected transformation behaviour µ j = Lµ ν j ν . a property which leads to the energy– momentum (dispersion) relationship of the Dirac particle. bispinor wave functions and we need to determine corresponding components of the wave function. speciﬁc for the Dirac free particle.233).282. µ (10. The solutions provide also a complete. t) and is determined through the momentum p ∈ R3 .231. One obtains using (10. Like for the free particle wave functions of the non-relativistic Schr¨dinger and the Klein–Gordon o equations one expects that the space–time dependence is governed by a factor exp[i(p · r − t)].275) The free particle wave function is an eigenfunction of Ho . the identiﬁcation of observables and their quantum mechanical operators. (10.273) where we have employed (10.246) S −1 (Lη ξ )γ µ S ( Lη ξ ) = (L−1 )νµ γν = (L−1 )ν γ ν = Lµ ν γν . In case of non-relativistic quantum mechanics the free particle wave function has a single component ψ(r. i. We will start from the Dirac equation in the Schr¨dinger form (10. (10. (10. Note that we have assumed in (10. In relativistic quantum mechanics a Dirac particle can also be characterized through a momentum.271) that the Dirac matrices are independent of the frame of reference. (10.274) 10. commute with each other.322 Relativistic Quantum Mechanics in a moving frame of reference and the ﬂux j µ in a frame at rest. the Dirac particles are described by 4-dimensional. the wave function has four components which invite further characterization of the free particle state.76).272) With S −1 (Lη ξ ) = S((L−1 )η ξ ) one can restate (10.233) o i∂t Ψ(xµ ) = Ho Ψ(xµ ) . only two degrees of freedom of the bispinor four degrees of freedom are independent [c.f.268) µ j = Ψ† (xµ )S † γ 0 γ µ SΨ(xµ ) = Ψ† (xµ )γ 0 S −1 γ µ SΨ(xµ ) . For this purpose we consider . As it turns out. as is required for the mentioned property.9 Solutions of the Free Particle Dirac Equation We want to determine now the wave functions of free particles described by the Dirac equation. 10. 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. 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.. but also as providing a basis in which the wave functions of interacting particle systems can be expanded and characterized. of which the wave functions are eigenfunctions as well.

94) Equation (10.280) σ·p (σ · p)2 φ0 = χ0 = +m ( + m) ( + m) p2 χ = χ0 . hence.282.278) provides us with information about the components of the bispinor wave function (10.279) According to the property (5. two-dimensional spinor state. E(p) = m2 + p 2 .284) The relationships (10.276) into (10. 1 (10.234) and (10. later to be identiﬁed with momentum and energy. 2 − m2 0 (10.276) where p and together represent four real constants. .232) leads to the 4-dimensional eigenvalue problem m σ·p σ · p −m φo χo = φo χo .280) which has a positive and a negative solution = ± E(p) . These two relationships are consistent with each other.231. (10.283) where is deﬁned in (10. (10.277) explicitly ( − m) 1 φo − σ · p χo 1 − σ · p φo + ( + m) 1 χo 1 = 0 = 0. the Dirac equation.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. namely φo and χo are related as follows φ0 χo = = σ·p χ0 −m σ·p φ0 .10. 10. In fact. like the Klein–Gordon equation.234) of Pauli matrices holds (σ · p)2 = p 2 1 One can.282) (10.93. from (10. conclude 1.281). reproduce the classical relativistic energy–momentum relationships (10. Inserting (10. 10.281) Obviously. and φo . +m (10. one ﬁnds using (5.283) imply that the bispinor part of the wave function allows only two degrees of freedom to be chosen independently. 10. (10. the socalled helicity of the particle.276).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 . We want to show now that these degrees of freedom correspond to a spin-like property.279) the well-known relativistic dispersion relationship 2 = m2 + p 2 (10.277) To solve this problem we write (10. χo each represent a constant.

0) γ 0 u 0 = |N+ (0)|2 .268).e. according to (10. Ψ† (p. First. = +E(p) . ±) γ 0 Ψ(0. 10. −|xµ ) = N− (p) −σ·p E(p) + m u u ei(p·r − t) . Hence. Employing (10. 10.h.286) Here N+ (p) is a constant which will be chosen to satisfy the normalization condition Ψ† (p.e. +) = |N+ (0)|2 (u† .s.287) will be justiﬁed now. ±) = S Ψ(0. ±)γ 0 Ψ(p. is Ψ(p. we present φo through the normalized vector φo = u1 u2 = u ∈ C2 . One can see this as follows: Let Ψ(p. Similarly.e. (10. of (10.287. ±) Ψ† (0. u† u = |u1 |2 + |u2 |2 = 1 . we present the negative energy solution.290) which diﬀers from the normalization condition (10. ±) S † γ 0 S Ψ(0. ±) = = = Ψ† (0.219) and u† u = 1 [c. (10.289) Here N− (p) is a constant which will be chosen to satisfy the normalization condition Ψ† (p.259).h. ±). i.287) the form of which will be justiﬁed further below.f. (10. The form of this condition and of (10.. hence. where S is given by (10.324 Relativistic Quantum Mechanics For our further characterization we will deal with the positive and negative energy solutions [cf. (10.288) corresponding to the wave function Ψ(p..290) should diﬀer. +) γ 0 Ψ(0. through χo given by χo = u1 u2 = u ∈ C2 .291) Note that we have used that. and let Ψ(0. +) = 1 . Ψ† (0. ±) denote the corresponding solution of a particle in its rest frame. (10. according to (10. For this purpose we consider ﬁrst the positive energy solution. u† u = |u1 |2 + |u2 |2 = 1 .286) for p = 0 yields. i. +) γ 0 Ψ† (p. ±) γ 0 S −1 γ 0 γ 0 S Ψ(0. ±).292) . (10.s. S † = γ 0 S −1 γ 0 . using γ 0 as given in (10. the solution for = +E(p). = −E(p) . (10.s. the l. of (10.285)].287. the solution for = −E(p). is invariant under Lorentz transformations. ±) Ψ† (0.. (10. we demonstrate that the product Ψ† (p. +|xµ ) = N+ (p) u σ·p E(p) + m u ei(p·r − t) . For the positive energy solution. (10. ±) . i.290). (10. ±) γ 0 Ψ(p.241). The connection between the solutions. +) γ 0 Ψ† (p. ±) denote the solution of a free particle moving with momentum p in the laboratory frame.285) The corresponding free Dirac particle is then described through the wave function Ψ(p.h. S −1 = γ 0 S † γ 0 and.281)] separately. We want to demonstrate now that the signs on the r.287) in the minus sign on the r. +) = −1 .

301) yields the normalization coeﬃcient N+ (p) = m + E(p) . −) = |N− (0)|2 (0. u† ) γ 0 0 u = − |N− (0)|2 . ( m + E(p) )2 − p 2 (10.289) yields Ψ† (0.286) is 2 N+ (p) (u∗ )T .290) to assign a positive value to |N− (0)|2 . σ·p u∗ E(p) + m T γo u σ·p E(p) + m u = 1 (10.286).9: Solutions of the Free Particle Dirac Equation The same calculation for the negative energy wave function as given in (10.294) We want to determine now N± (p) for arbitrary p. Condition (10.f. (5. −) γ 0 Ψ(0.s.302) .6: Show that the normalization condition 2 N+ (p) (u∗ )T . σ·p u∗ E(p) + m T u σ·p E(p) + m u = 1 (10.285) results in 2 N+ (p) 1 − p2 (E(p) + m)2 = 1 (10.s.10.296) Replacing (σ · p)2 by p 2 [c. this requires the choice of a negative side on the r.219) yields 2 N+ (p) u† u − u† (σ · p)2 u (E(p) + m)2 = 1.297) from which follows N+ (p) = Noting ( m + E(p) )2 − p 2 = m2 − p 2 + 2mE(p) + E 2 (p) = 2(m + E(p)) m the normalization coeﬃcient (10.h. We can also conclude from our derivation N± (0) = 1 . 2 E(p) (10.h. (10. (10. 2m (10. of (10.293) Obviously.300) (10. Exercise 7.295) Evaluating the l. 325 (10.299) ( m + E(p) )2 .287) written explicitly using (10. We consider ﬁrst the positive energy solution.298) This result completes the expression for the wave function (10.298) becomes N+ (p) = m + E(p) .234)] and using the normalization of u in (10. using γ 0 as given in (10.

λ|xµ ) in 4-vector notation.300) have been constructed to satisfy the free particle Dirac equation (10. This can be veriﬁed expressing the space–time factor of Ψ(p. This attribute is the so-called helicity.296) for the normalization constant N+ (p) of the positive energy solution. We ˆ want to show that this operator commutes with Ho and p to ascertain that the free particle wave ˆ function can be simultaneously an eigenvector of all three operators. 10.. i. exp[i(p · r − t)] = exp(ipµ xµ ).e. hence.308) 2 |p| ˆ Note that p represents here an operator. The commutation property [σ · p.307) where pµ = ( .289) are not completely speciﬁed. i∂µ Ψ(p.286. the wave functions constructed represent eigenstates of Ho . deﬁned as the component of the particle spin along the direction of motion.290) written explicitly using (10.309) .233) and ˆ ˆ ˆ from the two identities 1 1 0 0 −1 1 σ 0 0 σ ·p − ˆ σ 0 0 σ ·p ˆ 1 1 0 0 −1 1 = 0 (10. λ|xµ ) (10. λ|xµ ) = pµ Ψ(p. (u∗ )T γ o = −1 (10.289). however. The corresponding operator which measures this observable is 1 p ˆ Λ = σ· .h.s. The wave functions (10.. Inserting (10. have completed the determination for the wave function (10. (10. (10.. Condition (10.306) i. The property [σ · p. i. j = 1.289) is T −σ·p −σ · p 2 E(p) + m u N− (p) u∗ . This degree of freedom 1 describes a spin– 2 attribute.326 Relativistic Quantum Mechanics We consider now the negative energy solution. 3 is fairly obvious. Ho ] = 0 follows from (10. conclude N− (p) = m + E(p) 2m (10. the two components of u indicate another degree of freedom which needs to be deﬁned.275) yields Ho Ψ(p.e. 10.304) This condition is. λ|xµ ) .289. (10. 10. −p). Helicity The free Dirac particle wave functions (10. thereby. not a constant vector. We can. λ|xµ ) = λ E(p) Ψ(p.307) we investigate ﬁrst the observable due to the simpler operator σ · p.275).305) and. 2. pj ] = 0 .286) into (10. The wave functions are also eigenstates of the momentum operator i∂µ . Rather than considering the ˆ observable (10.e. yields 2 N− (p) u† (σ · p)2 u − u† u (E(p) + m)2 = −1 .286. identical to the condition (10.303) E(p) + m u Evaluating the l.

t) = Np e 2 1 p m + Ep 0 0 i(px3 − E t) 1 p e Ψ(pˆ3 .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. hence. 2m (10.312). + 1 |r. We assume ﬁrst particles with positive energy.e. i.310) We have shown altogether that the operators p. t) = Np e 1 0 0 p m + Ep i(px3 + E t) 1 p e Ψ(pˆ3 . −. States which are simultaneously eigenvectors of these three operators are also simulteneously eigenvectors of the three operators p. 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 . +. + 2 |r. is independent of the direction of motion of the particle.. = −E(p). −.312) We assume now particles with negative energy. − 1 |r. t) = Np e (10. are 1 i(px3 − E t) 0 p e Ψ(pˆ3 .286) are eigenfunctions of Λ as well will specify now the vectors u. − 1 |r.. According to the deﬁnition (5. 327 0 (σ · p)2 ˆ 2 (σ · p) ˆ 0 0 (σ · p)2 ˆ 2 (σ · p) ˆ 0 (10. 0. = +E(p). We consider ﬁrst the simplest case that particles move along the x3 –direction. +. can ˆ ˆ be simultaneously diagonal. 1)T are eigenstates of 2 σ 3 with eigenvalues ± 1 . p3 ). 1 p = (0. 0)T and (0. t) = Np e (10. Ho and σ·p commute with each other and. ˆ The condition that the wave functions (10. Since helicity is deﬁned relative to the direction of motion of a particle the characterization of u as an eigenvector of the helicity operator. Np = m + Ep . .224) 1 of σ3 the two u vectors (1.e. i. Ho and Λ deﬁned in (10. 2 the wave functions which are eigenstates of the helicity operator.10..311) 2 −p 0 m + Ep 1 where e3 denotes the unit vector in the x3 -direction and where ˆ Ep = m2 + p2 . Therefore. In this case Λ = 2 σ 3 .313) 2 0 1 where Ep and Np are deﬁned in (10. in principle.e. i.308) above.

− 1 |r.] The corresponding free particle wave functions are then Ψ(p.317) Ψ(p. − 1 ) 2 −p 1 m + Ep u(p.318) 1 Ψ(p. t) 2 = Np u(p. γ and are 1 0 1 ˜ Ψ(p = 0. 2 1 0 (10. 0)T and (0. a wave function which represents free particles at rest. +.312).311. 10.226. for an r–independent wave function. + ) 2 1 u(p.229). + 2 |r. − 2 ) u(p.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.e.313) except that the states (1.316) describes a rotation which aligns the x3 –axis with the direction of p.312) can be obtained also by means of the Lorentz transformation (10.223) as follows 1 u(p.319) 1 Ψ(p. γ .. These eigenstates are obtained through a rotational transformation (5. t) 2 = Np 1 u(p. (10.222. [One can also express the rotation through Euler angles α. in which case the transformation is given by (5. +. − ) 2 where ϑ(p) = = = exp exp i − ϑ(p) · σ 2 i − ϑ(p) · σ 2 1 0 0 1 . denoted by ˜. 1 |t) = √2 e−imt . + 2 ) p 1 m + Ep u(p. + 1 |r. t) = N√ ei(p·r + Ep t) (10. −. −. β. p) e |p| (10.314) (10.123). For this purpose one starts from the solutions of the Dirac equation in the chiral representation (10. − 2 ) −p 1 m + Ep u(p. 10. 10. +.311. 1)T have to be replaced by eigenstates u± (p) of the spin operator along the direction of p. + 2 ) u(p.220).262) for the bispinor wave function and the transformation (10. + 1 ) 2 p 1 m + Ep u(p.320) where Ep and N are again given by (10. − 1 ) 2 ei(p·r − Ep t) (10. − 2 |r. 5. Generating Solutions Through Lorentz Transformation The solutions (10. The corresponding wave functions are determined through ˜ i˜ 0 ∂t − m Ψ(t) = 0 .315) e3 × p ˆ ∠(ˆ3 . t) = Np ei(p·r + Ep t) (10. i.321) .

311.323) Transformation (10. +. for w = (0. + 1 |r. Applying (10.262) for a boost in the x3 –direction.324.325) to (10.322). −1 e 0 0 1 +imt 1 ˜ Ψ(p = 0. 10. i.176) imt → i ( p3 x3 and for the bispinor part according to (10.10. 2 |t) = √2 . 2 0 e −1 329 (10.325) One should note that φo .9: Solutions of the Free Particle Dirac Equation 0 1 1 ˜ Ψ(p = 0. 2 The solutions (10. −. φo . +. 10. − 1 |t) = √2 e−imt . 0. For the resulting wave functions in the chiral representation one can use then a notation corresponding to that adopted in (10.322) The reader can readily verify that transformation of these solutions to the Dirac representationsas deﬁned in (10. χo ∈ 1 0 . t) e 2 e− 2 w3 1 = √ 1 2 e 2 w3 i(px3 − E t) p e . − 1 |t) = √2 .174. ± 1 of the wave functions stated in (10.313) in the p → 0 limit.311) e 1 w 2 3 ˜ Ψ(p(w3 )ˆ3 .262) φo χo → e 2 w3 σ 0 1 3 Et ) (10. (10. χo are eigenstates of σ 3 with eigenvalues ±1. yields for the exponential space–time dependence according to (10.323) should yield the solutions for non-vanishing momentum p in the x3 –direction. 0 1 (10. +.324) 0 e − 1 w3 σ 3 2 φo χo = e 2 w3 σ φo 1 3 e− 2 w3 σ χo 1 3 . −.322) can be written in spinor form 1 √ 2 φo χo e imt . t) e 2 1 = √ 2 e− 2 w3 1 1 1 0 1 0 0 1 0 1 i(px3 − E t) p e ˜ Ψ(p(w3 )ˆ3 . 2 0 1 1 0 +imt 1 1 ˜ Ψ(p = 0.228) yields the corresponding solutions (10. 10..e. − 1 |r. w3 ). This correspondence justiﬁes the characterization ±.

−. 2 (10.327) Employing the hyperbolic function properties cosh x = 2 coshx + 1 . −. + 1 |r.61) between the parameter w3 and boost velocity v3 . Transformation to the Dirac representation by means of (10. − 2 |r. − 1 |r. + 2 |r.330 e 1 w 2 3 Relativistic Quantum Mechanics 1 0 1 0 0 1 0 1 1 ˜ Ψ(p(w3 )ˆ3 . −. t) e 2 = i(px3 + E t) p e 1 Ψ(p(w3 )ˆ3 . t) e 1 = √ 2 −e− 2 w3 − 1 w3 2 1 i(px3 + E t) p e ˜ Ψ(p(w3 )ˆ3 .61) p(w3 ) = m sinhw3 . + 2 |r. t) e 2 e 1 = √ 1 2 − e 2 w3 i(px3 + E t) p e (10. t) e = i(px3 + E t) p e (10. +.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. 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 .328) the relationship (10.329) .228) yields w3 cosh 2 = sinh w3 2 1 Ψ(p(w3 )ˆ3 . t) e = cosh w3 2 − sinh w3 2 sinh w3 2 cosh w3 2 − sinh w3 2 cosh w3 2 i(px3 − E t) p e Ψ(p(w3 )ˆ3 . and the expression (10. +. −. 2 sinh x = 2 coshx − 1 . − 2 |r.326) where according to (10.

In the new derivation we consider the particle described by the wave function transformed. when we generated the solutions Ψ(p.313) for −p. the proper transformation S(Lη ξ ).327). of course.311) as well as the negative energy solutions (10. is suﬃcient since (1) it must hold then for any ﬁnite transformation. Λ|t). λ. refered to as the active transformation.329. the essential property stating the Lorentz–invariance of the Dirac equation.333) .330) Inserting expressions (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 .243) considered solely the limit of inﬁnitesimal transformations anyway. ρ(Lη ξ ) has been deﬁned in (10.330) into (10. = γν .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 ). This transformation.123) above and characterized there. We expect. in general.168–10.334) (10. Actually. Such transformation had been applied by us. 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 (10.9: Invariance of Dirac Equation Revisited and similarly sinh w3 = 2 Ep − m = 2m p 2m (Ep + m) 331 (10.10.243) and. indeed. Λ|xµ ) from the solutions describing particles at rest Ψ(p = 0.221) using (10. (10. The fact that any such Ψ (xµ ) is a solution of the Dirac equation allows us to conclude S(Lη ξ )γ µ S −1 (Lη ξ ) ρ(Lη ξ )∂µ ρ−1 (Lη ξ ) = γ ν ∂ν which is satisﬁed in case that the following conditions are met ρ(Lη ξ )∂µ ρ−1 (Lη ξ ) S(Lη ξ )γ µ S −1 (Lη ξ ) Lν µ = Lν µ ∂ν .176).331) i S(Lη ξ )γ µ S −1 (Lη ξ ) ρ(Lη ξ )∂µ ρ−1 (Lη ξ ) − m Ψ (xµ ) = 0 (10. we will derive this equation only for inﬁnitesimal transformations. λ. The transformation is Ψ (xµ ) = S(Lη ξ ) ρ(Lη ξ ) Ψ(xµ ) (10. 10.331) is a solution as well. Making this expectation a postulate allows one to derive the condition (10.243). and since (2) the calculations following (10. but not the observer. Invariance of Dirac Equation Revisited At this point we like to provide a variation of the derivation of (10. expresses the system in the old coordinates. The reason why we provide another derivation of (10. reproduces the positive energy wave functions (10.243) 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. Such description will be essential for the formal description of Lorentz invariant wave equations for arbitray spin further below. that if Ψ(xµ ) is a solution of the Dirac equation that Ψ (xµ ) as given in (10. thereby. which however. To show this we rewrite the Dirac equation (10.

that any solution Ψ(xµ ) transformed according to (10.128) 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).336). 0).341) [K3 . ∂µ ] = (10. 1) . 0).262). ϑ = (0.. 0.334) for inﬁnitesimal Lorentz transformations Lη ξ . ∂µ ] = [x2 ∂1 − x1 ∂2 . ∂µ ] = [x0 ∂3 + x3 ∂0 . i. . of course. ∂µ ] = [x1 ∂3 − x3 ∂1 . ∂µ ] = (10.340) [K2 . ∂µ ] = (10. . 0 0 ∂3 −∂2 0 −∂3 0 ∂1 0 ∂2 −∂1 0 −∂1 −∂0 0 0 −∂2 0 −∂0 0 −∂3 0 0 −∂0 µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ = = = = = = = = = = = = = = = = = = = = = = = = 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 (10.336) We will proceed with this task considering all six cases: [J1 .335) shows that we need to demonstrate [J .342) One can readily convince oneself that these results are consistent with (10. ∂µ ] = [x0 ∂1 + x1 ∂0 . ∂µ ] = [x0 ∂2 + x2 ∂0 . 0. [K . w = (0.339) [K1 .243) and. We have demonstrated. ∂µ ] = (J )ν µ ∂ν .332 Relativistic Quantum Mechanics We will demonstrate now that the ﬁrst condition is satisﬁed by ρ(Lη ξ ). ∂µ ] = (10. .335) = ∂µ + M ν µ ∂ν + O( 2 ) The result will show that the matrix M ν µ is identical to the generators of Lν µ for the six choices ϑ = (1. 0). w = (0. therefore. it is met by S(Lη ξ ) as given in the chiral representation by (10. As mentioned already we will show condition (10. ∂µ ] = (K )ν µ ∂ν . 0. the Dirac equation is invariant under active Lorentz transformations. ∂µ ] = (10. 0. 0). Inspection of (10.e. ∂µ ] = (10. We will proceed by employing the generators (10.331) is again a solution of the Dirac equation. w = (0. 1. ∂µ ] = [x3 ∂2 − x2 ∂3 .337) [J2 . 0. ϑ = (0. . The second condition is identical to (10. .338) [J3 .

221) reads then [ iγ µ (∂µ + iqAµ ) − m ] Ψ(xν ) = 0 (10.343) One may also include the electromagnetic ﬁeld in the Dirac equation given in the Schr¨dinger form o ˆ (10. e. (10.345) holds |qV | << m .1) i∂t − qV and p by ˆ ˆ π = p − qA .351) X (x ) µ . The Dirac equation (10. (10. Following the respective procedure developed for the Klein-Gordon equation in Sect. 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 .6 we assume that the ﬁeld is described through the 4-vector potential Aµ and. we replace the operator ∂µ by ∂µ + iqAµ .346) ˆ ˆ ˆ α · π + qV + β m Ψ(xµ ) (10. accordingly. we deﬁne φ(xµ ) = e−imt Φ(xµ ) χ(x ) = e µ −imt (10. This solution exhibits a timedependence exp[−i(m + )t] where for holds in the non-relativistic limit | | << m.. σ2 .6 above).345) (10.10 Dirac Particles in Electromagnetic Field We like to provide now a description for particles governed by the Dirac equation which includes the coupling to an electromagnetic ﬁeld in the minimum coupling description.344) Using the notation σ = (σ1 . Accordingly. For this purpose we choose the decomposition Ψ(xµ ) = φ(xµ ) χ(xµ ) .348) (10.233) by replacing i∂t by (see Table 10.349) We want to focus on the stationary positive energy solution.1 in Sect 10.347) (10.350) (10. Non-Relativistic Limit We want to consider now the Dirac equation (10. 10.345) in the so-called non-relativistic limit in which all energies are much smaller than m.10. Equivalently.10: Dirac Particles in Electromagnetic Field 333 10. for the scalar ﬁeld V in (10.g. 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. The Dirac equation in the Schr¨dinger form reads then o i∂t Ψ(xµ ) = ˆ ˆ where α and β are deﬁned in (10.232).

henceforth. (10.359) We want to evaluate the latter commutator. ∂k + qAk ] i i 1 1 1 1 [ ∂j . 5. 10. (10.357) Equation (10.352) allow one to approximate (10. 10. which in the present case states ˆ ˆ2 ˆ ˆ (σ · π)2 = π + i σ · (π × π) . ∂k ] + [∂j . For this purpose we employ the identity (5. One obtains [πj . ˆ ˆ For the components of π × π holds ˆ ˆ π×π = jk (10. i i (10.357) for Φ can be reformulated by expansion of (σ · π)2 . ∂k ] + q [ ∂j .358) ( πj πk − πk πj ) = jk [πj . due to the m−1 factor. (10. and.353) (10.346.360) For an arbitrary function f (r) holds ( [Aj .356).353) by X ≈ to obtain a closed equation for Φ ˆ σ·π 2m 2 (10. ∂k ] + q [Aj . accordingly. Ak ] . π ] .348. Ak ] + q 2 [Aj . ∂k ] + [∂j .349) yields i∂t Φ i∂t X = = ∂t X X Relativistic Quantum Mechanics << m . derived in Sect. Ak ] i i i i =0 =0 = q q [Aj .361) .354) ˆ 0 ≈ σ · π Φ − 2m X .356) i∂t Φ ≈ Φ + qV Φ .7.351) in (10.352) ˆ σ · π X + qV Φ ˆ σ · π Φ + qV X − 2mX . (10.354) The properties (10. one can replace X in (10.355) ˆ σ·π Φ 2m (10. Φ Using (10. πk ] = = 1 1 [ ∂j + qAj .334 and assume that for the time-derivative of Φ and X holds ∂t Φ << m . ˆ Equation (10. does not need to be considered anymore. (10. Ak ] ) f = ( ∂j Ak − Ak ∂j + Aj ∂k − ∂k Aj ) f .350. 10. identiﬁes X as the small component of the bi-spinor wave function which.230).

0. (10. −A).367) . the spin. r (10.6)].363) where we employed B(r. t) + q V (r. Equations (10.2) it introduces the o extra term qσ · B Φ which describes the well-known interaction of a spin.366) can be interpreted as a rotation around the ﬁeld B by an angle qtB. t) = × A(r. 10. Comparision of (10. Let us consider brieﬂy the consequences of the interaction of a spin..2) governing a one-dimensional wave function ψ ∈ C. t) ≈ ˆ [p − q A(r. 10. but rather remains as a steady “guest” of non-relativistic physics with the proper interaction term.364) which is referred to as the Pauli equation.1 particle with a magnetic 2 ﬁeld B. 0.1 with the magnetic ﬁeld. t) (10. does not leave the theory again when one takes the non-relativistic limit.343) for the vector potential Aµ = (− Ze2 . the interaction qσ · B induces a 1 precession of the spin. In other words. while agreeing in all other respects with the non-relativistic Schr¨dinger equation (10.362) or. 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) . one obtains ( [Aj .10: Dirac Particles in Electromagnetic Field Using ∂j Ak f = ((∂j Ak )) f + Ak ∂j f ∂k Aj f = ((∂k Aj )) f + Aj ∂k f 335 where ((∂j · · ·)) denotes conﬁnement of the diﬀerential operator to within the brackets ((· · ·)). i.360) and Aµ = (V. 5. Dirac Particle in Coulomb Field . 10. Ak ] ) f = [ ((∂j Ak )) − ((∂k Aj )) ] f (10. πk ] = q q (( ∂j Ak − ∂k Aj )) = − i i ×A jk q = − B i jk (10. The respective bispinor wave function Ψ(xµ ) ∈ C4 is described as the stationary solution of the Dirac equation (10. [πj . The formal solution of this equation is Φ(t) = e−iqtB·σ Φ(0) .2 around the magnetic ﬁeld.363) allow us to write (10.359. 0) .e. t) [see (8.10.357) in the ﬁnal form i∂t Φ(r.223) shows that the propagator in (10. using (10.358.366) (10. ∂k ] + [∂j .1 which emerged in the Lorentz-invariant theory as an algebraic 2 necessity. t) 2m 2m Φ(r.365) Comparision of this expression with (5.364) governing a two-dimensional wave function Φ ∈ C2 with the corresponding non-relativistic Schr¨dinger equation (10.Spectrum We want to describe now the spectrum of a relativistic electron (q = −e) in the Coulomb ﬁeld of a nucleus with charge Ze. o reveals a stunning feature: the Pauli equation does justice to its two-dimensional character.222. t)]2 q − σ · B(r.344.

γ ˆ ˜ ˆ (10. noting from (10.231). (5.368) is also a solution of (10.228) and in (10.370) for which holds ˜ ˆ ˜ ˆ 3 γ πµ γ πν = ˜ ˆ ˜ ˆ µ=0 µ ν (˜ µ )2 πµ + γ ˆ2 µ.180). The diﬀerence arises from the term γ µ πµ γ ν πν in (10. 1.230) γ µ γ ν = − γ ν γ µ .s.229). ˜ ˜ ˆ ˆ (10. Equation (10.374).232) in Sect. once a solution Ψ(xµ ) of (10.369).370) ˜ [ i˜ µ (∂µ + iqAµ ) − m ] [ i˜ µ (∂µ + iqAµ ) + m ] Ψ(xµ ) = 0 γ γ (10.371). 10.6.370) Any solution of (10. indeed.230).370) resembles closely the Klein-Gordon equation (10. can be rewritten using. (˜ 0 )2 = 1 and (˜ j )2 = γ 1 γ −1 j = 1. ˜ (10. (5. but the converse is not necessarily true. This yields ˜ [ i˜ µ (∂µ + iqAµ ) + m ] [ i˜ µ (∂µ + iqAµ ) − m ] Ψ(xµ ) γ γ ˜ = (˜ µ πµ γ ν πν − m2 ) Ψ(xµ ) = 0 .368) ˜ where Ψ(xµ ) and γ µ are deﬁned in (10. i. respectively.371) such that we can conclude that (10. For this purpose we ‘square’ the Dirac equation. 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. Employing πµ as deﬁned ˜ in Table 10.369) For our solution we will adopt presently a strategy which follows closely that for the spectrum of pionic atoms in Sect.372) (10.ν=1 µ =ν γ µ γ ν πµ πν .e. ˜ However.7 one can write the second term in (10. 2. multiplying (10. 3 µ=0 ˆ2 (˜ µ )2 πµ = π0 − π . 3.370).369).373) (10.370) is obtained then ˜ [ i˜ µ (∂µ + iqAµ ) + m ] Ψ(xµ ) γ is a solution of (10.375) Following the algebra that connected Eqs. is a solution of (10. µ = ν and altering ‘dummy’ summation ˜ ˜ ˜ ˜ .1 one can write (10.369) from the left by γ ν πν + m.336 Relativistic Quantum Mechanics For the purpose of the solution we assume the chiral representation. according to (10.374) The ﬁrst term on the r. γ ˆ2 ˆ2 (10. we solve ˜ [ i˜ µ (∂µ + iqAµ ) − m ] Ψ(xµ ) = 0 γ (10. but diﬀers from it in an essential way.h.368) ˜ ( γ µ πµ − m ) Ψ(xµ ) = 0 . 5.

πν ] γ ˜ π ˆ µ.367) holds qA0 = r ˆ Ze2 . γ µ γ ν πµ πν = ˜ ˜ ˆ ˆ µ.ν=1 µ =ν (10. γ ν ] [ˆµ .ν=1 µ =ν 3 j=1 3 3 = 1 4 1 2 1 [˜ .10.376–10.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. π0 ] γ ˜ π ˆ j=1 = [˜ 0 .377) which follows readily from the deﬁnition (10. γ 0 ] [ˆj .378) According to the deﬁnition (10.381) According to (10.229). πj ] π ˆ = = (−i∂t + qA0 . Altogether.382) .379) The commutators [ˆ0 . due to A = 0. 3 (10.. one can summarize (10.e. Since [ˆµ .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 j ] [ˆ0 .10: Dirac Particles in Electromagnetic Field indices. πν ] γ ˜ π ˆ µ.ν=1 µ =ν = [˜ µ .367) of Aµ . r2 (10. γ ] [ˆ0 . γ j ] are γ ˜ [˜ 0 . γ ˜ π ˆ j=1 (10. πj ] + γ ˜ π ˆ 4 0 j [˜ j .376) This expression can be simpliﬁed due to the special form (10.378) can be evaluated using (10.344).367) and the deﬁnition (10. the commutators [˜ 0 .380) where f = f (r. γ ν ] [ˆµ . πj ] in (10. i. t) is a suitable test function and where ((· · ·)) denotes the range to which the derivative is limited.ν=1 µ =ν σ 0 0 −σ · (( qA0 )) (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. 2. πν ] = 0 π ˆ for µ.344) π ˆ [ˆ0 . −i∂j ] = − [ (∂t + iqA0 ) ∂j − ∂j (∂t + iqA0 ) ] f i ((∂j qA0 )) f (10. πj ] . ν = 1.380) γ µ γ ν πµ πν = i ˜ ˜ ˆ ˆ µ. it holds 1 4 [˜ µ .

.387) The expression (10.383) We seek stationary solutions of this equation. Combining this result with (10. 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 . . Accordingly. however.389) According to the results in Sect. (10.5 which describe the coupling of orbital angular momentum and spin 1 {( Yjm (j − 2 . . 2 |ˆ)} as given in (6. 6. (10. (10. (10. r r 2 2 2 j = 1.189) 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 . .388) Except for the term iσ · r this equation is identical to that posed by the one-dimensional Kleinˆ Gordon equation for pionic atoms (10. 10. .e. it is this quantity that we want to determine.374). Insertion of (10.384) into (10.338 Relativistic Quantum Mechanics where r = r/|r| is a unit vector.384) can be interpreted as the energy of the stationary state and. a solution of the form ∼ Y m (ˆ) can be obtained. m values are coupled. in r ˆ fact.148). only the two states {Yjm (j − 2 . −j + 1.5 the operator iσ · r is block-diagonal in this basis such that only ˆ 1 the states for identical j.386) + 2 ± iσ · r ˆ Ze2 − m2 r2 φ± (r) = 0 . (10. + j } (10. couples the orbital angular momentum of the electron to its spin.1 . hence.368) reads − Ze2 ∂t − i r 2 + 2 + i σ·r ˆ 0 0 −σ · r ˆ Ze2 − m2 r2 Ψ(xµ ) = 0 (10. . 1 |ˆ). In the latter case.147. is genuinely two-dimensional and. Such solutions are of the form ˜ ˜ Ψ(xµ ) = Φ(r) e−i t . 6. 2 3 2 . i. m = −j.381).388) in terms of states introduced in Sect. 1 |ˆ) ) ..385) 1 We split the wave function into two spin. Yjm (j + 1 . we express 2 the solution of (10.191) solved in Sect.2 components ˜ Ψ(r) = and obtain for the separate components φ± (r) Ze2 + r 2 ˜ φ+ (r) ˜ φ− (r) (10. (10. We note that these states are also eigenstates of the angular r 2 . 6.375) the ˆ ‘squared’ Dirac equation (10.6. The term iσ · r.

. Since.199).. which states that the states Yjm (j ± 1 . 1 |ˆ) = − Yjm (j ˆ r 2 2 1 2 h± (r) g± (r) Yjm (j − 1 . .392).151]. (6. We can. . (6. 1 |ˆ) r 2 2 2 2 r r . in the present treatment. 10.394) Equation (10.186)]. We select.10: Dirac Particles in Electromagnetic Field 339 ˆ momentum operator L2 [cf.397) This equation is identical to the Klein-Gordon equation for pionic atoms written in the form (10. the values of the argument of λ1. 1. 1 |ˆ) r 2 (10.5 [c.2 (r) = 0 (10. Obviously.396) and λ2 (j) [λ2 (j) + 1] (10.2 (j) being 2 j = 1 . not the wave functions. except for the slight diﬀerence in the expression of λ1. Any similarity transformation leaves the ﬁrst term in (10.200). hence.f. we want to determine solely the spectrum.392) into diagonal form. 1 |ˆ) + r Yjm (j + 1 .2 (j) + 1) 2 Ze2 + + r2 r 2 − m2 f1.2 (j) as given by (10.392) reads then in the diagonal representation 2 ∂r − λ1. 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 . − 1 r2 (j − 1 )(j + 1 ) − Z 2 e4 2 2 ±i Ze2 (j + )(j + ) − Z 2 e4 We seek to bring (10. conclude that the . .396) and (10. such transformation can be chosen as to diagonalize the second term. 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 (10.393) but do not explicitly consider further the wavefunctions. involving the 2 × 2 unit matrix.391) derived in Sect.395. rather than = 0. (10.395) (10.390) (10. .392) h± (r) g± (r) = 0. a speciﬁc pair of total spin-orbital angular momentum quantum numbers j. the missing additive term − 1 . 1 |ˆ) are r 2 2 ˆ eigenfunctions of L2 .2 (j)[λ1.151). 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. property (6.2 (j). and except for the fact that 2 2 we have two sets of values for λ1. However. namely. the eigenvalues are independent of m.10. namely. therefore. 3 . . λ1 (j) and λ2 (j). m and expand φ± (r) = Using σ · r Yjm (j ± 1 . unaltered. as in the case of pionic atoms.

. 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 . . (10. . j = + 1 . Using (10. −j + 1. j where 1 corresponds to λ1 (j) as given in (10. . . . . 2 2 1 ≈ m − (10. accordingly. . . 2.398) 2 = 1 + m . One obtains in case of (10. m) 2 n = 1. − + 1. . . m) = 1 + m (n− + . counts the nodes of the wave function. . .400) and (10. = 1 + m Z 2 e4 √ ( n − − 1 + 2 − Z 2 e4 )2 . .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. including spin. (10. One may also state 2 this in the following way: (10. . . Obviously. The magnitude of the relativistic eﬀect 2 2 e4 .395) and 2 corresponds to λ2 (j) as given in (10. m = −j.402) with (10.397) is again given by eq. albeit with some modiﬁcations. the second term on the r. 2.404) = 0. ( +1) − Z 2 e4 )2 (10. . For a given value of n the energies 1 and 2 for identical j-values correspond to mixtures of states 1 with orbital angular momentum = j − 2 and = j + 1 . . . j = 1. j .399) n = 0. . . 1.203). . j = − 2 . j = 1 . n − 1 . 10. . . . . −j + 1. In case of non-relativistic hydrogen-type 1 atoms.400) corresponds to a non-relativistic state with quantum numbers n. . .2 . These considerations 2 are summarized in the following equations 1 ED (n.396) we obtain. 1 ms = ± 2 (10. . and spin-orbital angular momentum j = + 1 . 3. 1. 3 . . .401) corresponds to a non-relativistic state 2 with quantum numbers n. . .401) if one attributes to the respective 1 states the angular momentum quantum numbers = j + 2 and = j − 1 . . (10.402) In this expression n is the so-called main quantum number. .400) (10. 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. . n − 1 . . −j + 1. and spin-orbital angular momentum j = − 1 . 1. 10.395. (10.401) These expressions can be equated with the non-relativistic spectrum.s. . . . . . . . . 2n2 = 0. . . .403) √ Z 2 e4 2 ED (n. m = −j. m = − . . 1. (10.340 Relativistic Quantum Mechanics spectrum of (10.h. 2 2 m = −j.396). 1. 2. . It is given by n = n + + 1 where is the orbital angular momentum quantum number and n = 0. j . One can equate (10. of these equations describe the binding energy. .398. the stationary states have binding energies E = − mZ 2 e4 n = 1. 2. . .

60601 -3. mom. binding energy / eV Eq.2: Binding energies for the hydrogen (Z = 1) atom. 1. . r 2 2 2 .036 by means of Eqs. notation 1s 1 2 non-rel. . .3.402) -13.40151 ⇑ .1.402)] and the 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 (1.40146 ⇑ ⇑ -1. (10. relate closely to the corresponding nonrelativistic states.405 ) -13. −j + 1. j (10. Degeneracies are denoted by ⇑. m) = m 1 + Z 2 e4 (n−j −1 + 2 (j+ 1 )2 − Z 2 e4 )2 2 n = 1.10: Dirac Particles in Electromagnetic Field main quantum number n 1 2 2 2 3 3 3 3 3 orbital angular mom. i. = 0.405) In order to demonstrate relativistic eﬀects in the spectrum of the hydrogen atom we compare in Table 10. the nonrelativistic degeneracy for the six 2p states of the hydrogen atom: in the present.0041 keV and e2 = 1/137. (10. case these six states are split into energetically diﬀerent 2p 1 and 2p 3 states. much less than the rest mass of the electron (511 keV). . . 0 0 1 1 0 1 1 2 2 spinorbital ang. in fact. 2.e.. relativistic.551176 spectr.1.404) ﬁnally into the single formula ED (n. j. Of particular interest is the eﬀect of spin-orbit coupling which removes. 10. The reason is that the mean kinetic energy of the electron in the hydrogen atom. the 2p 3 states with r 2 2 2 2 1 involve four degenerate states corresponding to Y 3 m (1.551178 ⇑ . . n − 1 j = 1 2 1 2 ± m = −j..402. the non-relativistic and relativistic energies are hardly discernible. The table entries demonstrate that the energies as given by the expression (10.40147 -1. 10. . 1 |ˆ) for m = ± 1 . is in the range of 10 eV. . . ± 3 . j 1 2 1 2 1 2 3 2 1 2 1 2 3 2 3 2 5 2 341 rel.60583 -3.3.2 the non-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. . (10.551177 ⇑ .405). binding energy / eV Eq.405) in terms of the non-relativistic quantum numbers n. i. This is clearly demonstrated by the comparision of non-relativistic and relativistic spectra of a hydrogen-type atom with Z = 100 in Table 10. .403. 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.405)] spectrum of the hydrogen atom. for example. The energies were evaluated with m = 511. 2 |ˆ) for m = ± 1 .10. (10. However. . (10.e. One can combine the expressions (10.

0 ⇑ ⇑ -15.10. We assume for the wave function the stationary state . 32 (10.402.206) provides in the present case ED (n. 0) (10.9 ⇑ . Degeneracies are denoted by ⇑.036 by means of Eqs.15. The energies were evaluated with m = 511.2.1 -34. (10. 0 0 1 1 0 1 1 2 2 spinorbital ang.407) 1 2 This expression allows one.6 -42.342 main quantum number n 1 2 2 2 3 3 3 3 3 orbital angular mom.3: Binding energies for the hydrogen-type (Z = 100) atom. binding energy / keV Eq. An example for such potential is the Coulomb potential V (r) = −Ze2 /r considered further below.406) α √ 1+ (n − β + β 2 −α)2 The expansion (10. Introducing α = Z 2 e4 1 and β = j + 2 (10.3). m) ≈ m − mZ 2 e4 mZ 4 e8 − 2n2 2n3 1 j+ − 3 4n + O(Z 6 e12 ) . (10. 10. Tables 10. 0. (10. to estimate the diﬀerence between the energies of the states 2p 3 and 2p 1 (cf. for example.2 -17. j.3 spectr. binding energy / keV Eq. j 1 2 1 2 1 2 3 2 1 2 1 2 3 2 3 2 5 2 Relativistic Quantum Mechanics non-rel.405 ) -161.1 ⇑ .405) to order O(Z 4 e8 ). (10. . notation 1s 1 2 2s 1 2 2p 1 2 2p 3 2 3s 1 2 3p 1 2 3p 3 2 3d 3 2 3d 5 2 Table 10.405).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). 0.409) where V (r) is spherically symmetric.8 ⇑ .35. It holds for n = 2 and j = 3 .402) -136. In order to investigate further the deviation between relativistic and non-relativistic spectra of hydrogen-type atoms we expand the expression (10.1 ⇑ ⇑ ⇑ ⇑ rel. mom.0041 keV and e2 = 1/137. 1 2 2 2 2 E (2p 3 ) − E (2p 1 ) ≈ − 2 2 mZ 4 e8 2 · 23 1 −1 2 = mZ 4 e8 .15.405) reads 1 (10.

2 |ˆ) and does r 2 ˆ has odd parity and it holds [c. 1 |ˆ) r 2 1 1 = ∂r r r r (10.410) 1 where Φ(r). 6.415) 1 ˆ σ · p rg(r) Yjm (j − 2 . according to (10.411) (10.414) These equations can be brought into a more symmetric form using ∂r + which allows one to write (10. r 2 (10. i. 2 |ˆ) r Φ(r) r . 6.198)] not couple states with diﬀerent j. (10. The arguments above allow one to eliminate the angular dependence by expanding Φ(r) and X (r) 1 in terms of Yjm (j + 2 . 6. 1 |ˆ) .413. 1 |ˆ) r 2 = i ∂r − 1 2 r g(r) Yjm (j + 1 . m-values. i. the term is block-diagonal in the space spanned by the states Yjm (j ± 1 . X (r) ∈ C2 describe the spatial and spin. 1 |ˆ) . 1 |ˆ) r 2 2 = i ∂r + −j r 1 g(r) Yjm (j + 2 .411.417) c(r) d(r) X (r) 1 1 1 1 Yjm (j + 2 . 1 |ˆ) introduced in r 2 2 1 Sect. 1 |ˆ). ˆ σ · p X + m Φ + V (r) Φ ˆ σ · p Φ − mX + V (r) X = = Φ X (10.. σ · p ˆ σ · p f (r) Yjm (j + 1 . 10. (6.232.e.412) ˆ In this equation a coupling between the wave functions Φ(r) and X (r) arises due to the term σ · p. = r (10.10.412) are four-dimensional with r-dependent wave functions.e. 2 |ˆ) r r r .413) ˆ σ · p g(r) Yjm (j − 1 . but are timeindependent.345). This term has been discussed in detail in Sect.10: Dirac Particles in Electromagnetic Field form Ψ(xµ ) = e−i t 343 Φ(r) X (r) . 10.416) The diﬀerential equations (10.2 degrees of freedom.414) ˆ σ · p rf (r) Yjm (j + 1 . 2 |ˆ) + r Yjm (j − 2 . 1 |ˆ) r 2 2 (10. 1 |ˆ) r 2 2 = i ∂r + j+ r 1 2 3 2 f (r) Yjm (j − 1 . r r 2 2 2 a(r) b(r) 1 1 1 1 Yjm (j + 2 . 168]: the term is a scalar (rank zero tensor) in the space of the spin-angular momentum states Yjm (j ± 1 . 10.5..5 [see. 1 |ˆ) and Yjm (j − 1 .f. 2 |ˆ) + r Yjm (j − 2 . The Dirac equation reads then.197. pp. r 2 2 (10. 1 |ˆ) r 2 2 = i j+ ∂r + r j+ r 1 2 1 r f (r) Yjm (j − 2 . in particular.

418) holds for f1 (r).421) g2 (r) X (r) 1 1 − Yjm (j + 2 .422) j+ r j+ r 1 2 g2 (r) + [ − m − V (r) ] f2 (r) = 0 ∂r − 1 2 f2 (r) − [ + m − V (r) ] g2 (r) = 0 (10. g2 (r) ∂r + 1 2 f1 (r) − [ + m − V (r) ] g1 (r) = 0 (10. only a(r).344 Relativistic Quantum Mechanics In general. and multiplying by r results in the following two independent pairs of coupled diﬀerential equations i ∂r − j+ r j+ r 1 2 d(r) + [ m + V (r) − ] a(r) = 0 i and i ∂r + 1 2 a(r) + [ −m + V (r) − ] d(r) = 0 (10.419) Obviously.411. d(r) are coupled and b(r).412). g1 (r) ∂r − j+ r j+ r 1 2 g1 (r) + [ − m − V (r) ] f1 (r) = 0 ∂r + and for f2 (r).415. 2 |ˆ) r r where the factors i and −1 have been introduced for convenience.416). there exist two independent solutions (10. According to (10.420) g1 (r) X (r) 1 1 − Yjm (j − 2 . c(r) are coupled. 10. 2 |ˆ) r r f2 (r) 1 1 r Φ(r) i r Yjm (j − 2 . Presently.417) of the form f1 (r) i Yjm (j + 1 .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).417) into (10. such expansion must include states with all possible j. m values. 1 |ˆ) r Φ(r) 2 2 r = (10. the orthonormality property (6. m pair contribute. (10.423) . Inserting (10. 10. we consider the case that only states for one speciﬁc j. 2 |ˆ) = (10. using (10. Accordingly.

1 |ˆ) r i 2 2 .422) describes the states 2p 1 . According to this procedure.367) which is spherically symmetric such that equations (10.422.405) of the relativistic hydrogen atom the corresponding states have quantum numbers n = 1.424) Ψ(xµ ) ≈ r 0 1 i. together with the appropriate boundary conditions at r = 0 and r → ∞.423). 3d 3 . except for the opposite sign of the term (j + 1 ). . enforcing the polynomial to be of ﬁnite order leads to discrete eigenvalues . determining wave functions of the type (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) − = 0.422) and (10.Wave Functions We want to determine now the wave functions of the stationary states of hydrogen-type atoms which correspond to the energy levels (10. . In the non-relativistic limit. (10.423) are identical. The solution of (10. etc. We assume the 4-vector potential of pure Coulomb type (10. to states with angular momentum = j + 2 . the radial wave functions for Dirac particles in the potential (10. the ones given in (10.422) follows in this case from the same procedure as that adopted for the radial wave function of the non-relativistic hydrogen-type atom. . (10. = 0. and obtains ﬁnally a polynomial function p(r) such that rγ exp(−µr)p(r) solves (10. the 2 equations determine. 1. Hence. 2 |ˆ) r Ψ(xµ ) ≈ .425) r 0 covers states with angular momentum 1s 1 . 2s 1 . .422). Similarly.405). . 3d 5 . Φ in (10. namely.422).10. .10: Dirac Particles in Electromagnetic Field 345 Equations (10. 2 2 2 2 2 2 = j − 1 2 and. We note that (10..420) is the large component and X is the small component. According to the discussion of the spectrum (10.423) apply for V (r) = −Ze2 /r. 3p 3 .421). 2p 3 . 3p 1 .405).e. Equation (10. 10. one demonstrates then that the wave functions for r → ∞ behaves as exp(−µr) for some suitable µ.409). 3s 1 .e.422) corresponds to states f1 (r) Yjm (j + 1 .420). i. . 2 2 2 (10.422). etc.422) near r = 0. (10. . Hence. 2. Dirac Particle in Coulomb Field . Behaviour at r → 0 We consider ﬁrst the behaviour of the solutions f1 (r) and g1 (r) of (10. n − 1.. correspodingly the states We consider ﬁrst the solution of (10.422) determines solutions of the form (10. one demonstrates ﬁrst that the wave function at r → 0 behaves as rγ for some suitable γ. (10. in the non-relativistic limit wave functions f2 (r) 1 1 i Yjm (j − 2 . (10. for small r.426) .

429) This equation poses an eigenvalue problem (eigenvalue −γ) for proper γ values.434) √ The solutions of these equations are f1 . g1 ∼ exp(± m2 − 2 r). according to (10.267. 10.427).428) = 0. determined that the solutions f1 (r) and g1 (r).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 .431) inﬁnite since.433) = = ( m2 − 2 ) g1 (r) ( m2 − 2 ) f1 (r) (10. µ = m2 − 2 (10. We have. g1 (r) →0 ∼ b rγ (10.435) .420). (10. in case j + 1 )2 < Ze2 . for the particle density holds then ρ(r) ∼ |rγ−1 |2 = Behaviour at r → ∞ For very large r values (10. ) f1 (r) ( + m ) g1 (r) (10. r2 (10. 10. (10. Only the exponentially decaying solution is admissable and.422) becomes ∂r g1 (r) ∂r f1 (r) Iterating this equation once yields 2 ∂r g1 (r) 2 ∂r f1 (r) 1 .427) = = 0 0. One obtains γ = ± (j + 1 )2 − Z 2 e4 . g1 (r) →∞ ∼ e−µr . hence.427) with 1 γ = (j + )2 − Z 2 e4 . The assumed r-dependence in (10. hence. for small r. assume the r-dependence in (10.266. we conclude f1 (r) →∞ ∼ e−µr . Such r-dependence 2 would make the expectation value of the potential r2 dr ρ(r) 1 r (10. (10.430) 2 Note that the exponent in (10.432) = = − ( − m.427) makes only the positive solution 2 possible. becomes imaginary.

Solution of the Radial Dirac Equation for a Coulomb Potential To solve (10.445) . µ is real. of both (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. Let us consider then for the solution of (10. (10.10: Dirac Particles in Electromagnetic Field For bound states holds 347 < m and.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. ρ m+ ρ [∂ρ − (10. However.441) The last two terms on the l.437) where they ˜ ˜ cancelled in case f1 = g1 = a. g1 (r) = φ1 (r) − φ2 (r) ˜ (10.438) (10.439) − m− e g1 (r) ˜ where µ is given in (10.434) √ m + a e−µ r .435 ).437) (10. Equation (10. In the present case the functions f1 and g1 cannot be chosen ˜ ˜ identical due to the terms in the diﬀerential equations contributing for ﬁnite r.10.441) correspond to (10.434) results in √ √ (m − ) m + a − (m − ) m + a √ √ (m + ) m − a − (m + ) m − a which is obviously correct. without loss of generality we can choose ˜ f1 (r) = φ1 (r) + φ2 (r) .h.436) f1 (r) = Insertion into (10. √ g1 (r) = − m − a e−µr .436). (10. hence.440) (10.s. we set f1 (r) g1 (r) = = √ √ ˜ m + e−µr f1 (r) −µr = = 0 0 (10.440) and (10.442) which leads to a partial cancellation of the asymptotically dominant terms. We also introduce the new variable ρ = 2µ r . Accordingly.444) (10.443) 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 .

450) −√ mZe2 m2 − 2 βs ρs+γ−1 = 0 mZe2 m2 − αs (s + γ) β2 + (j + 1 ) αs + √ 2 s 2 +√ From (10.430)].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 (10.447) leads to (s + γ) αs + (j + 1 ) βs − √ 2 s Ze2 m2 − 2 αs (10. 10.430) one can write this βs = s+γ − 2 √ Ze m2 − 2 s (s + 2γ) βs−1 .426–10.447) We seek solutions of (10. Inserting (10. 10.430) which conform to the proper r → 0 behaviour determined above [c.448) φ2 (ρ) = ρ (10.452) From (10.449) for γ given in (10.448. 10.454) . (10.446) (10. Using (10.453) (s + γ)2 + Z 2 e4 − (j + 1 )2 2 2 √ Ze m2 − 2 βs .446.348 Relativistic Quantum Mechanics Addition and subtraction of these equations leads ﬁnally to the following two coupled diﬀerential equations for φ1 and φ2 ∂ρ φ1 + ∂ρ φ2 + j+ ρ 1 2 Ze2 mZe2 φ1 − √ φ2 = 0 m2 − 2 ρ m2 − 2 ρ mZe2 Ze2 φ1 + √ φ1 + √ φ2 − φ2 = 0 m2 − 2 ρ m2 − 2 ρ j+ ρ φ2 − √ 1 2 (10.446 .450) follows Ze2 m2 − 2 βs − βs−1 =0 (10.449) into (10.447) of the form n φ1 (ρ) = ργ s=0 n γ s=0 αs ρ2 βs ρ2 (10.451) 1 √mZe − (j + 2 ) 2 2 αs . (10.f. = m− Ze2 βs s + γ − √m2 − 2 2 (10.

is an integer. x) = 1 + a a(a + 1) x2 x + + .457) One can relate the polynomials φ1 (ρ) and φ2 (ρ) deﬁned through (10.459) − 2 √mZe m2 − 2 (10. hence. (s − so ) β0 s! (2γ + 1)(2γ + 2) · · · (2γ + s) (10. . 1.449) and (10.455) are ﬁnite. 2. . 2γ + 1. i.456. . namely.448. 1. 10.10. 2. For example. .430. c.457) with the conﬂuent hypergeometric functions F (a. β0 j+ 1 2 (10.. According to the deﬁnitions (10. We can.e.. n It holds φ1 (ρ) φ2 (ρ) = = ργ F (−so . ρ) . (s − so ) β0 s! (2γ + 1)(2γ + 2) · · · (2γ + s) (10.461) In order that the wave functions remain normalizable the power series (10.456) − so 2 √mZe m2 − 2 (−so )(1 − so )(2 − so ) .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 (10.457) for βs and αs imply that so must then be an integer. 10.460) (10.405) and the ensuing so values ( 10. 3p 1 . (10.405) allows one to identify n = n − j + 2 which.448.461 for values given by (10. m (n ) = . The expressions (10. . for the states 2p 1 .452) follows j+ αs = 1 2 349 Ze2 m2 − 2 −γ (10. α + 1. Comparision with (10. 10. so = n . equivalently. .458) or. 2γ + 1.456) and (10. .455) this conﬁnement of so implies discrete values for .449) must be of ﬁnite order. in fact. conclude 2 2 2 that the polynomials in (10. x) . ρ) so β0 ργ F (1 − so .. This requires that all coeﬃcients αs and βs must vanish for s ≥ n for some n ∈ N. c c(c + 1) 2! (10. 10.455) s − so βs−1 s(s + 2γ) (s − 1 − so )(s − so ) βs−2 (s − 1)s (s − 1 + 2γ)(s + 2γ) (1 − so )(2 − so ) . . n = 0. = From (10.10: Dirac Particles in Electromagnetic Field Deﬁning so = √ one obtains βs = = . . . with the associated Laguerre polynomials L(α) = F (−n.405). 3d 3 holds n = 1.

460. and (10. 2 . (n + γ)m N (2µr)γ−1 e−µr − κ F (−n . = j + 2 . g1 (r) determined by (10. The coeﬃcients β0 in (10. 10. The complete wave function is given by the following set of formulas 1 Ψ(n. 2γ + 1.421)] ∞ dr (|f1 (r)|2 + |g1 (r)|2 ) = 1 (10. as in (10. 1 |ˆ) in (6.m (j − 1 . are two-dimensional vectors determined through the explicit form of the spin-orbital angular momentum states Yjm (j ± 1 .463) where Φ and X .158) of the latter states absorb the angular integral in (10. which involve the conﬂuent hypergeometric functions in (10.421) with radial wave functions f1 (r). where2 F± (κ|r) = G1 (r) = F+ (κ|r) . 10. The normalization integral is then ∞ 0 π 2π r2 dr 0 sin θdθ 0 dφ (|Φ(r)|2 + |X (r)|2 ) = 1 (10.460. 1 |ˆ) 2 2 r κ = j+ 1 2 (10. 1990). 6.463) and yield [note the 1/r factor in (10.438. m|xµ ) = e−i t iF1 (r) Yj.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 . They are described through quantum numbers n.467) ± n F (1 − n . (10. = j + 2 . j.466) F1 (r) = F− (κ|r) . The wave functions (10. Berlin. 2γ + 1.267).461).461). The orthonormality 2 2 r properties (6.148). Due to the form (10.439). 1 |ˆ) 2 2 r G1 (r) Yj. 10. (Springer.6.465) (10.460. 6.410) of the stationary state wave function the density ρ(xµ ) of the states under consideration. (10. is time-independent. Sect.469) This formula has been adapted from ”Relativistic Quantum Mechanics” by W. 10. j.m (j + 1 . given by expression (10.421) correspond to non-relativistic states with orbital angular momentum 1 1 = j + 2 .421).147. can follow the procedure adopted for the wave functions of the non-relativistic hydrogen atom and will not be carried out here.442).464) 0 The evaluation of the integrals. 9. m. 2µr) (2µ) 2 Γ(2γ + 1) 3 N = and m 4m )Γ(2γ + n + 1) (n +γ)m (n +γ)m (10.157.461) are to be chosen to satisfy a normalization condition and to assign an overall phase. 2µr) (10. Greiner.

The radial wave functions f2 (r) and g2 (r) in (10. j.m (j − 1 . obtained closed expressions for the wave functions of all the stationary bound states of relativistic hydrogen-type atoms. = j − µ 1 2 . where F± (κ|r) are as given in (10. 2p 3 . One can verify. 2 accordingly. not not 2 2 2 2 2 2 covered by the wave functions given by (10.465–10.470) r 0 Obviously. 3s 1 .m (j + 1 . hence. We have.471) (10. tracing all steps which lead from (10.422) to (10.10.423) which diﬀers from the radial Dirac 1 equation for f1 (r) and g1 (r) solely by the sign of the terms (j + 2 )/r.469).10: Dirac Particles in Electromagnetic Field 351 We want to consider now the stationary states of the type (10. 1 |ˆ) 2 2 r κ = −j − 1 2 (10. 1 |ˆ) 2 2 r G2 (r) Yj. G2 (r) = F+ (κ|r) .421) are governed by the radial Dirac equation (10.421) which. in the non-relativistic limit. etc.467–10. this wavefunction has an orbital angular momentum quantum number = j − 1 and. become f2 (r) 1 1 i Yjm (j − 2 . m|x ) = e −i t iF2 (r) Yj. describes the complementary set of states 1s 1 .469) that the following wave functions result Ψ(n.472) F2 (r) = F− (κ|r) . .469). 3d 5 . 2 |ˆ) r Ψ(xµ ) ≈ e−1 t . 3p 3 . (10. 2s 1 .

352 Relativistic Quantum Mechanics .

ϑ). ˜ We have altered here slightly our notation of S(w. indeed.Chapter 11 Spinor Formulation of Relativistic Quantum Mechanics 11. expressing its dependence on w. z ∈ C ˜ Because of its block-diagonal form each of the diagonal components of S(z). may be exploited to express the Lorentz-invariant equations of relativistic quantum mechanics. isomorphic to the natural representation. ϑ)µ ν correspond to diﬀerent a(z) and b(z). ϑ) for the ˜ in the chiral representation as given by (10.e. This is.2) (11. what will be achieved in the following. ϑ)µ ν .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. Starting point is the Lorentz transformation S(w.. a complex variable z. 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.225–10.1) (11. 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.. We will characterize the space on which the transformations a(z) We will see below that the representations a(z) and b(z) are. a(z) and b(z). i.e. in particular. 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. diﬀerent L(w. in a sense.262). in fact. 1 353 .229). The lower dimensionality of a(z) and b(z) implies.3) (11. We will proceed by building as much as possible on the results obtained sofar in the chiral representation of the Dirac equation. ϑ through 3. must be two-dimensional irreducible representations of the Lorentz group.4) = exp = w − iϑ . i.

for z = iϑ. b(z). Relationship Between a(z) and b(z) The transformation b(z) can be related to the conjugate complex of the transformation a(z). ± 1 represents the familiar spin– 1 states. (11. formulate the Dirac equation. 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. For ϑ = (0. usually expressed 2 as product of rotations and of functions of Euler angles α. ﬁnally.. β. i.1 states as described by Dm m (ϑ). ϑ ∈ R3 . Aν . a space of vectors state1 for which holds state2 state 1 ∼ | 1 . In this case holds a(z) = b(z) ˜ 1 (xµ ). however. namely. and the Klein–Gordon equation in the spinor representation. Ψ4 (xµ ) in case z = w + iϑ for w = 0. (11. Here | 1 . (11.243). + 1 2 2 state 2 ∼ | 1 . will establish the map between L(w. Ψ4 (xµ ) transform according to ˜ ˜ ˜ and Ψ b(z). the so-called spinor space.354 Spinor Formulation and b(z) act. are elements of SU(2) and correspond.5) The transformations in this case.7) where “∼” stands for “transforms like”. −1 2 2 z = iϑ. 2 2 2 ˜ Since a(z) acts on the ﬁrst two components of the solution Ψ of the Dirac equation. to 1 ∗ ∗ z ·σ (11. This characterization allows one to draw conclusions regarding the state space in which a(z) and b(z) operate. neutrino equation. In this case a(z) and b(z) are identical and read a(i ϑ) = b(i ϑ) = exp 1 − ϑ·σ 2 . ϑ ∈ R3 . ϑ)µ ν and a(z). A First Characterization of the Bispinor States We note that in case w = 0 the Dirac transformations are pure rotations. −1 2 2 |1. the operator ∂µ and the Pauli matrices σ in the new representation and. 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 . Ψ2 (xµ ) transform according to a(z) whereas Ψ3 (xµ ). γ (see Chapter 5). in 2 fact.6) as given by (5. i. express 4-vectors Aµ . ϑ ∈ R3 (11. to the rotational transformations of spin.e.. Ψ2 (xµ ) and Ψ3 (xµ ). θ ∈ R3 .8) We like to stress that there exists. +1 2 |1. +1 2 2 |1. − 1 2 2 z = iϑ. β.9) a∗ (z) = exp 2 .e. a distinct diﬀerence in the transformation behaviour of ˜ ˜ ˜ ˜ Ψ1 (xµ ).

One can readily verify [c. using f (a) −1 = f ( a −1 ). ∗ σ 2 = − σ2 . 11. 11. (11. . Spatial Inversion ˜ One may question from the form of S(z) why the Dirac equation needs to be four-dimensional. The eﬀect of inversion on Lorentz transformations is.f. in fact.12) does. Hence.2.224)] ∗ σ1 = σ1 . Ψ (xµ ) even though these pairs of ˜ ˜3 ˜4 featuring the components Ψ1 (x ˜ 2 components transform independently of each other. hold −1 = 1 One notices then.10) From this one can derive b(z) = where = 0 1 −1 0 . one can show −1 −1 −1 = −σ1 = σ2 = −σ3 ∗ = −σ1 ∗ = −σ2 σ2 −1 (11. but leave rotation angles ϑ unaltered.1) can be written in the form ˜ S(z) = a(z) 0 0 a∗ (z) −1 (11. a result to be used further below. 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) one ﬁrst demonstrates that for and −1 as given in (11. one can express a∗ (z) −1 = exp 1 − z ∗· σ 2 = b(z) . σ3 ). −1 = ∗ −σ3 = −σ ∗ . Ψ (xµ ) as well as Ψ (xµ ).224. a∗ (z) −1 = exp 1 ∗ z σ∗ 2 −1 . 11. (11. µ ).12) To prove (11.4) and .14) . −1 a∗ (z) −1 (11.1: Lorentz Transformation of Dirac Bispinor ∗ ∗ ∗ where σ ∗ = (σ1 .12) yields σ1 σ3 or in short σ Similarly.13) Explicit matrix multiplication using (5.12).17) We conclude. however. −1 given by (11.10.11) 0 −1 1 0 = . (11. 355 (11. (5.16) σ = −σ ∗ .11. therefore. 1.18) with a(z) given by (11. (11. that they alter w into −w. σ2 . that the transformation (11.15) (11. ∗ σ 3 = σ3 . This demonstrates that a(z) is the ˜ transformation which characterizes both components of S(z).

e. Obviously.f. However. For the general case the proof. However. can be found in G.22) which follows from the fact that for any complex. M is a complex 2 × 2–matrix. C) deﬁned in (11. 2 . 3 . In fact. (11. 1990)..C) and SO(3.21) = etr( 2 z·σ) = 1 1 (11.23) Exercise 11. det(M ) = 1 } . Exercise 1.2 Relationship Between the Lie Groups SL(2.224)] tr( σj ) = 0 . the space spanned by only two of the components of Ψ is not invariant under spatial inversion and. the transformations a(z) and b(z) become interchanged. 2 = 1 i.356 Spinor Formulation ˜ Let P denote the representation of spatial inversion in the space of the wave functions Ψ. (5. B¨uerle and E.2. Part (Elsevier. This implies ˜ Ψ1 ˜ Ψ2 P ˜ Ψ3 ˜ Ψ4 ˜ Ψ3 ˜ Ψ4 = Ψ1 ˜ ˜ Ψ2 .e. The proof of this important property is straightforward in case of hermitian M (see Chapter 5). the Lorentz invariant equation for neutrinos is only 2–dimensional. which describes pure rotations.1: Show that SL(2. does not suﬃce for particles like the electron which obey inversion symmetry. a(w + iϑ) for w = 0 is an element of SL(2.G. ϑ ∈ R3 . P −1 = P.20) ˜ Obviously. for particles like the neutrinos which do not obey inversion symmetry two components of the wave function are suﬃcient. based on the Jordan–Chevalley theorem. C) = { M.3.A. non-singular matrix M holds2 det eM and from [c. 11. The transformation S(z) in the transformed space is then ˜ P 1. de Kerf Lie a Algebras. is an element of SU(2). (11.19) i.24) = etr(M ) (11.10. (11. 2..1) We have pointed out that a(iϑ). One can verify this by evaluating the determinant of a(z) det( a(z) ) = det e 2 z·σ 1 (11. j = 1. hence. Amsterdam.A.21) together with matrix multiplication as the binary operation forms a group. ˜ ˜ P S(w + iϑ) P = S(−w + iϑ) = b(z) 0 0 a(z) .

25) can also be written M (Aµ ) = A0 + A3 A1 − iA2 A1 + iA2 A0 − A3 .76).26) or from the fact that the matrices σ0 . Starting point is a bijective map between R4 and the set of two-dimensional hermitian matrices deﬁned through M (Aµ ) = σµ Aµ where (11.2: Show that σ0 .29) where we used (10. σ3 are hermitian.30) Since the components of Aµ are real. We will prove below [cf.135)] this transformation behaviour. one wishes that the deﬁnition (11. Demonstrate that (11.27) requires then σ µ Aµ Lν µ σµ = σν (11. 1 0 0 −1 .28) and (11. (11.31) holds. 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 . the matrix M (Aµ ) is hermitian as can be seen from inspection of (11.e. σ2 . in a transformed frame should hold Lρ ν σ µ Aµ σµ A µ . one can provide a simple expression for the inverse of M (Aµ ) M = M (Aµ ) ↔ Aµ = 1 tr M σµ 2 . Argue why M (Aµ ) = σµ Aµ provides a bijective map.27) −→ Straightforward transformation into another frame of reference would replace σµ by σµ . C) and the group L↑ of proper. (11. σ1 . (11. (11..31) Exercise 11.11. 0 −i i 0 .30).28) σµ = 1 0 0 1 . M (Aµ ) according to (11. −→ µ Consistency of (11. (11. σ2 . σ1 .25) of the matrix M (Aµ ) is independent of the frame of reference.75) holds aν = Lν µ aµ one realizes that σµ transforms inversely to covariant 4–vectors. 0 1 1 0 . σ3 provide a linear–independent basis for the space of hermitian 2 × 2–matrices.26) (11. i. or+ thochronous Lorentz transformations. In fact.2: Lie Groups SL(2.32) . Noting that for covariant vectors according to (10. (11. The function M (Aµ ) is bijective.2. in fact.1) Mapping Aµ onto matrices M (Aµ ) 357 We want to establish now the relationship between SL(2.C) and SO(3.25) σ0 σ1 σ2 σ3 The quantity σµ thus deﬁned does not transform like a covariant 4–vector. The following important property holds for M (Aµ ) det ( M (Aµ ) ) = Aµ Aµ which follows directly from (11.

38) . L(a)µ ν = 1 tr a σν a† σµ 2 .31). In fact. (11. In fact. (11.3: Show that L(a)µ ν deﬁned in (11. The linear character of the transformation becomes apparent expressing A µ as given in (11.33) to M (Aµ ) describing the action of the transformation in terms of transformations of Aµ .34) where we used the properties M † = M and (a† )† = a.36) The suitable A µ can readily be constructed using (11.31)] 1 a Aµ −→ A µ = tr a M (Aµ ) a† σµ 2 Because of (11. Due to det(a) = 1 the transformation (11.35) holds for this transformation A µ Aµ = Aµ Aµ (11. a ∈ SL(2.35) We now apply the transformation (11.f.40) is an element of L↑ .33) conserves the determinant of M . (11. C) and for any Aµ there exists an A µ such that M (A µ ) = a M (Aµ ) a† .37) using (11.37) deﬁnes actually a Lorentz transformation.37) which implies that (11. (11.32.33) This transformation conserves the hermitian property of M since (M )† = ( a M a† )† = (a† )† M † a† = a M a† = M (11.2. 11. Accordingly. + .39) Exercise 11.40) 1 tr a σν a† σµ 2 Aν (11. (11. C) . C) deﬁnes the transformation [c. (11. for any a ∈ SL(2.358 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† . it holds for the matrix M deﬁned through (11.33) det(M ) = = det( a M a† ) = det(a) det(a† ) det(M ) [det(a)]2 det(M ) = det(M ) . any a ∈ SL(2.25) Aµ = which allows us to express ﬁnally A µ = L(a)µ ν Aν .

44) (σν )αβ (σν )δγ .β γ. i. (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 .1) deﬁned through L(a)µ ν [cf. C) and of SO(3.1) For this purpose one writes using tr(AB) = tr(BA) L(a1 )µ ν L(a2 )µ ρ = ν a1 a2 product in SL(2. (11.2: Lie Groups SL(2.C) )µ ρ (11.1) L(a)µ ν provides a homomorphism 359 We want to demonstrate now that the map between SL(2.40)] respects the group property of SL(2. β 2.47) This completes the proof of the homomorphic property of L(a)µ ν .C) and SO(3.δ 1 4 Aαβγδ Γβα Γγδ α. C) and SO(3. (11.41) results in 1 ¯ Lµ ρ = 4 where Aαβγδ = ν Γ = a2 σρ a† 2 (11.δ (11.42) = ν Deﬁning Γ = a† σµ a1 . β δ = δ = = δ = δ 1 2 = 2 = 1 Aαβγδ (11.11. .e.α.1).45) One can demonstrate through direct evaluation 2 α = 2 α = 2 α = = 2 α = 0 else β β γ γ = = = = γ = γ = 1.β γ. 1 ¯ and using the deﬁnition of Lµ ρ in (11.41) 1 tr a1 σν a† σµ 1 2 1 tr σν a† σµ a1 1 2 1 tr a2 σρ a† σν 2 2 1 tr a2 σρ a† σν 2 2 ..43) (σν )αβ Γβα Γγδ (σν )δγ = ν. 2 (11. ¯ Lµ ρ = L(a1 )µ ν L(a2 )µ ρ = L( product in SO(3.

s. correspondingly. C) which correspond to K. i.54) This reads for the four cases Aµ = (1. 0.h. Aµ = (0.25) in the last step. (11. M (Aµ + (K1 )µ ν Aν ) = = M (Aµ ) + M (Aµ ) − M ( (−A1 .47.s.36) M (Lµ ν Aν ) = a M (Aµ ) a† (11.53) results in the condition σ 0 A1 − σ 1 A0 = ( κ 1 σ µ + σ µ κ † ) Aµ . K3 .55) (11.49) we obtain using (11.58) . Aµ = (0.49) assuming (note that g µ ν is just the familiar Kronecker δµν ) Lµ ν a = = gµν + 1 + 1 (K1 )µ ν κ1 . 0. J3 of the Lorentz transformations Lµ ν in the natural representations.48) yields for the l. 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 .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. −A0 . C) as complex 2 × 2–matrices are deﬁned through four complex or. To this end we consider inﬁnitesimal transformations and keep only terms of zero order and ﬁrst order in the small variables. 0. 0.51) ( 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 ) . of (11. (11. 1. correspond to the generators given by (10. of (11. 0). 1) − σ1 σ0 0 0 One can verify readily that 1 κ1 = − σ 1 2 (11. 10.360 Generators for SL(2. J1 . For the r. 0. J Spinor Formulation The transformations a ∈ SL(2. 1 (11. 1 (11. we write (11. Aµ = (0. denoted below as κ1 .51) Insertion of (K1 )µ ν as given in (10.48) We want to determine now the generators of the transformation a(z) as deﬁned in (11. J2 . 0).e. K2 . (11. 1.57) .56) (11. Because of the condition det(a) = 1 only six independent real numbers actually suﬃce for the deﬁnition of a. 0. 0) ) σ0 A 1 − σ1 A 0 (11.53) Equating (11. 0.49). To obtain the generator of a(z) corresponding to the generator K1 .2) which correspond to the generators K1 .52) where we employed (11.h.52) and (11. eight real numbers.48).. noting the linearity of M (Aµ ). 0).50) (11.

the so-called contravariant spinors. Exercise 11. C) correspond to the generators K and J of Lµ ν . J2 .3 Spinors Deﬁnition of contravariant spinors We will now further characterize the states on which the transformation a(z) and its conjugate complex a∗ (z) act. − 1 . 2 (11. κ3 of a(z) corresponding to K2 . λ2 .61) This identiﬁes the transformations a(z = w − iϑ) as representations of Lorentz transformations. 2 (11.2. a1 1 a1 2 a2 1 a2 2 (11. Here a(z) = ( aα β ) = describes the matrix a(z). + 1 2 2 1 transforms under rotations (z = iϑ) like a spin– 2 state | 1 . J3 are given by 1 κ = − σ.11. λ3 corresponding to J1 . one can show that the generators κ2 .60) We can. hence. 11. K3 and λ1 . 2 λ = i σ.63) where we extended the summation convention of 4-vectors to spinors. state that the following two transformations are equivalent L(w. 2 2 φ1 φ2 ∈ C2 . def α = 1.4: Show that the generators (11.60) of a ∈ SL(2. (11.62) We denote the general (z = w + iϑ) transformation by φ α = aα β φβ = aα 1 φ1 + aα 2 φ2 . Similarly.64) . C) acts on spinors φα ∈ C2 (characterized below) 1 (11.1) acts on 4–vectors Aµ . ϑ) = ew·K + ϑ·J ∈ SO(3. We consider ﬁrst the transformation a(z) which acts on a 2-dimensional space of states denoted by φα = According to our earlier discussion holds φ1 φ2 1 transforms under rotations (z = iϑ) like a spin– 2 state | 1 . w describing boosts and ϑ describing rotations.3: Spinors 361 obeys these conditions. a(w − iϑ) = e− 2 (w − iϑ)·σ ∈ SL(2.

e. 1 ) stands for the Clebsch–Gordon coeﬃcient. i. 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 . as contravariant spinors and to φα .. etc. is the singlet state. .66) should constitute a singlet spin state. Using (0. Deﬁnition of covariant spinors Expression (11. i. . exchange of φα and χβ alters the sign of the expression. (11. χβ . m. indeed. −m) | . as we will demonstrate below such states are invariant under general Lorentz transformations a(z).65) where | · · · 1 describes the spin state of “particle 1” and | · · · 2 describes the spin state of “particle 2” and (0. . 1 . those of “particle 2” with the spinor χβ one can state that the quantity 1 Σ = √ 2 φ1 χ2 − φ2 χ1 (11. 0 = 2 2 m=± 1 2 1 1 (0. In the notation developed in Chapter 5 holds for the singlet state | . . . as covariant spinors. χβ . However.69) (11. The relationship between the two can be expressed φ1 φ2 = = 3 φ1 φ2 0 1 −1 0 (11. i. should remain invariant under transformations a(iϑ).67) We will refer to φα .e. 0| 1 .362 Deﬁnition of a scalar product Spinor Formulation The question arises if for the states φα there exists a scalar product which is invariant under Lorentz transformations. . ± 1 . this scalar product is anti-symmetric.. . Such a scalar product does. 1 ) = 2 2 2 2 2 2 2 2 √ ±1/ 2 and equating the spin states of “particle 1” with the spinor φα . ± 1 . In this case spinors φα transform like spin– 1 states and an invariant. 0| .68) (11. invariant under transformations a(z).e. The existence of a scalar product gives rise to the deﬁnition of a dual representation of the states φα denoted by φα . In fact. . 1 .70) ‘Scalar’ implies invariance under rotations and is conventionally generalized to invariance under other symmetry trasnformations. . m 1 | .. To arrive at a suitable scalar product we consider ﬁrst only rotational transformations a(iϑ). which 2 can be constructed from products φ1 χ2 . 0. 0| 1 .. invariant and as such has the necessary properties of a scalar3 product.66) is a bilinear form. −m 2 2 2 2 1 1 1 1 2 (11.

11. 10.3: Spinors as can be veriﬁed from (11. −1 connecting contravariant and covariant spinors play the role of the metric tensors gµν .75) (11.71.70) is φ1 φ2 −1 363 = = −1 φ1 φ2 .72)] φα φ α = = = = αβ αβ φβ φβ 11 21 11 21 12 22 12 22 (11.77) is Lorentz invariant. For this scalar product holds φα χα = −χα φα .63. Proof that φα χα is Lorentz invariant We want to verify now that the scalar product (11. (11.76) αβ αβ The scalar product (11.11.69.78) aβ γ γδ φδ (11.70) and (11.72) in a notation analogous to that chosen for contravariant and covariant 4–vectors [cf.72) 0 −1 1 0 Exercise 11.67) will be expressed as φα χα = φ1 χ1 + φ2 χ2 = φ1 χ2 − φ2 χ1 .71) (11.73. Accordingly.3.74) = = 0 1 −1 0 0 −1 1 0 (11. (11. 11.80) One can write in matrix notation aα β γ δκ αγ a δ αβ (11. In the transformed frame holds φ α χα = aα β αγ aγ δ δκ φβ χκ . (10.69.74)]. (10.1: Show that for any non-singular complex 2 × 2–matrix M holds M −1 = det(M ) M −1 T The matrices .68). 11.10.81) . we will express (11. 11. The inverse of (11. 11. (11.75) is given by φα = as can be readily veriﬁed.79) = a −1 T a κβ .73) (11. g µν of the Minkowski space [cf. The transformation behaviour of φα according to (11.77) (11.

˙ (11. hence. from (11.90) extending the summation convention to ‘dotted’ indices. a∗ (z) provides a representation of the Lorentz group which is distinct from that provided by a(z).87) which we will employ from now on. (11.364 Using (11.89) such that (11.86) A concise notation of the conjugate complex spinors is provided by ˙ (φα )∗ = φα = φ1 ˙ φ2 ˙ ˙ (11. σ † = (σ ∗ )T = σ. 2.81) yields aα β and. it holds φk = (φk )∗ .83) into (11. i.e.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. ˙ We also deﬁne covariant versions of φα φα = (φα )∗ .2) and (11..80) φ α χα = φβ χβ .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. k = 1. The transformation ˙ behaviour of φα is ˙ φ α = (aα β )∗ φβ ˙ (11. Insertion of (11.88) which one veriﬁes taking the conjugate complex of (11.84) (11. Obviously. As discussed above. Hence.88) reads ˙ ˙ φ α = aα β φβ ˙ ˙ (11. The complex conjugate spinors We consider now the conjugate complex spinors (φ ) = α ∗ γ δκ αγ a δ = a −1 T a κβ = a−1 a κβ = δκβ (11. We denote ˙ (aα β )∗ = aα β ˙ (11.91) .85) φ1 φ2 ∗ ∗ .83) Here we have employed the hermitian property of σ.63). the ˙ conjugate complex spinors φα need to be considered separately from the spinors φα .

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 ) = = .97) implies then that φα transforms like Ψ ˙ φ1 ˙ φ2 ˙ transforms under rotations (z = iϑ) like a spin– 1 state | 1 .94) (11. For the spinors φα and χα thus ˙ deﬁned holds that the scalar product ˙ φα χα = φ1 χ1 + φ2 χ2 ˙ ˙ ˙ ˙ ˙ (11.96) is Lorentz invariant.76) φα ˙ ˙ φα αβ ˙ ˙ αβ ˙ ˙ = = = = αβ ˙ ˙ αβ ˙ ˙ αβ αβ φβ φβ ˙ ˙ (11.11.18) ˜ 3 .97) The transformation. − 1 .100) . (11.75). + 1 2 2 2 1 transforms under rotations (z = iϑ) like a spin– 2 state | 1 .92) (11. A tensor tα1 α2 ···αk β1 β2 ···β (11.18) of the bispinor wave function Ψ. a∗ (z) −1 accounting for the ˜ transformation behaviour of the third and fourth spinor component of Ψ. 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 = (11.75. A comparision of (11.93) (11.. in matrix notation.99) Ψ3 (xµ ) χ ˙ (xµ ) ˜ χβ (xµ ) ˙ 1 ˜ χ2 (xµ ) Ψ4 (xµ ) ˙ 11. a property which is rather evident.4: Spinor Tensors 365 The relationship between contravariant and covariant conjugate complex spinors can be expressed in analogy to (11. The transformation behaviour of φα is ˙ φα = ˙ ˙ β γδ ˙˙ φδ ˙ ˙ αβ a γ ˙ ˙ . i.73.e. 11. Ψ4 .4 Spinor Tensors ˙ ˙ ˙ We generalize now our deﬁnition of spinors φα to tensors.95) ˙ where αβ and αβ are the real matrices deﬁned in (11. one can state ˜ and (11. is governed by the operator a∗ (z) −1 which arises in ˜ the Lorentz transformation (11. (11.

i. .m aj akm t m (11.108) Obviously. (11.106).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 .104) Indices on tensors can also be lowered employing the formula tα β = and generalizations thereof.105) This tensor is actually invariant under Lorentz transforma= αβ . According to (11. (11. it holds αβ .106) Exercise 11.e. ˙ ˙ ˙ ˙ (11. The 4–vector Aµ in spinor form We want to provide now the spinor form of the 4-vector Aµ .e.102) = . This tensor transforms according to t αβ = aα γ aβ δ tγ δ ˙ This reads in matrix notation.25] M (Aµ ) = σµ Aµ .25) the tensor is explicitly Aαβ = ˙ A1 1 A1 2 ˙ ˙ A2 1 A2 2 ˙ ˙ = A0 + A3 A1 − iA2 A1 + iA2 A0 − A3 . using conventional matrix indices j. αβ ˙ αγ t ˙ γβ (11.m aj a∗ t km m .109) . We had demonstrated that M (Aµ ) transforms according to M = M (Lµ ν Aν ) = a M (Aµ ) a† (11. ˙ ˙ ˙ ˙ (11.366 is a quantity which under Lorentz transformations behaves as t ˙ ˙ ˙ α1 α2 ···αk β1 β2 ···β k Spinor Formulation = m=1 aαm γm n=1 aβn δn tγ1 ···γk δ1 ···δ . tjk = a t a† jk ˙ ˙ ˙ ˙ (11.102... i.e. i. the transformation bevaviour of a tensor tαβ reads in spinor and matrix notation t αβ = aαγ aβδ tγδ . (11. This task implies that we seek a tensor.4.. this transformation behaviour is in harmony with the tensor notation adopted. m. k.1: Prove equation (11. tjk = a t aT jk = . with ˙ contravariant indices αβ.107) which reads in spinor notation [cf. An example of a tensor is αβ and tions. αβ = αβ (11.103) A αβ = aα γ aβ δ Aγ δ .103) Similarly. 11. we want to express Aµ through a spinor tensor. An obvious candidate is [cf.

Aµ and tensors tαβ can be applied to the partial diﬀerential operator ∂µ .115) is correct.116) . ∂µ in spinor notation The relationship between 4–vectors Aµ .117) ˙ ˙ ∂0 − ∂3 −∂1 + i∂2 −∂1 − i∂2 ∂0 + ∂3 .110) The 4-vectors Aµ .115) Exercise 11. (11. 11. (11. (11.110) one obtains Aα β = ˙ A0 − A3 −A1 − iA2 −A1 + iA2 A0 + A3 A0 + A3 A1 + iA2 A1 − iA2 A0 − A3 .4. Hence.11. Using (11. (11.111) A11 A12 ˙ ˙ A21 A22 ˙ ˙ ˙ 0 −1 1 0 (11.114) We ﬁnally note that the 4–vector scalar product Aµ Bµ reads in spinor notation Aµ B µ = 1 αβ ˙ A B αβ . (11.113) Aα β = ˙ (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 . (11.114) yields ∂αβ = ˙ ∂0 + ∂3 ∂1 + i∂2 ∂1 − i∂2 ∂0 − ∂3 . ˙ 2 (11.109.2: Prove that (11.110) one can state ∂ αβ = Similarly.112) A22 −A21 ˙ ˙ −A12 A1 1 .4: Spinor Tensors One can express Aαβ also through Aµ Aα β = ˙ ˙ 367 A0 − A3 −A1 + iA2 −A1 − iA2 A0 + A3 . employing (11.

(11.125) a (σ ) Equation (11. (11. γµ = ˜ 0 µ )∗ (σ −1 0 σµ σµ 0 (σ µ )∗ −1 . of this equation is 0 ∗ (σ µ )∗ a and.s.121). (11.. . hence.18) and γ µ by (11. Comparision of (10. This is ˜ possible since γ µ can be expressed through σ µ .124). (11.126) which is the complex conjugate of (11.368 σµ in Tensor Notation Spinor Formulation We want to develop now the tensor notation for σµ (11. 1 0 0 −1 .118) 0 −1 −1 0 0 i −i 0 For this purpose we consider ﬁrst the transformation behaviour of σ µ and σµ .. σµ . hence.118) yields ˜ γµ = ˜ Using (11.f. −1 0 0 1 (11. σµ building on the known transformation behaviour of γ µ .15) one can write σµ = and.121) One expects then that the transformation properties of σ µ should follow from the transformation properties established already for γ µ [c. Equation (10. 0 −i i 0 . Hence. (10.229) and (11. Note that (10. To obtain the transformation properties of σ µ we employ then (10.122) aσ µ −1 a−1 −1 (a∗ )−1 0 Lν µ (11. (11.123) (a∗ )−1 Lν µ a L ν µ = = σν (σ ) ν ∗ −1 (11. .h. one can conclude aσ µ ∗ −1 Lν µ σν 0 = .124) constitutes the essential transformation property of σ µ and will be considered further.243) in the chiral representation expressing S(Lη ξ ) by (11.124). We will obtain the transformation behaviour of σ µ . . holds also in the chiral representation.243) holds independently of the representation chosen.125) is equivalent to (11.124) . i.243)].e.e.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. 0 1 1 0 .119) (11.120) σµ 0 . (11. i.26) and its contravariant analogue σ µ σµ = σµ = 1 0 0 1 1 0 0 1 .125) is equivalent to µ ∗ −1 −1 a∗ (σ µ )∗ −1 −1 a Lν µ = (σ ν )∗ (11.

2. 0 −1 −1 0 µ=1 . equation (11.. hence. One can. 0 −i i 0 µ=2 .4). therefore.. (σ ∗ )T = σ yields using (11. according to (11.118) (σ µ )11 (σ µ )12 ˙ ˙ (σ µ )21 (σ µ )22 and (σµ )11 (σµ )12 (σ µ )21 (σ µ )22 ˙ ˙ 1 0 0 1 µ=0 ˙ ˙ ˙ ˙ = (11. 3) σj −→ a(iϑ) σj a(iϑ) † = a(iϑ) σj a∗ (iϑ) T .132) 1 0 0 1 µ=0 . (11. transformations (11. . 11. (11.107) to (11. 0 1 1 0 µ=1 . This transformation behaviour.e. 2.11. (11. 0 i −i 0 µ=2 . 11. (11.102.133) Combining (11. = .2) −1 369 (a∗ )−1 = −1 − 1 z ∗ ·σ ∗ 2 e = e− 2 z 1 ∗ · −1 σ ∗ = e2z 1 ∗ ·σ . 1 0 0 −1 µ=3 .131) one can express the transformation behaviour of σ µ in the succinct form Lν µ σ µ = σ ν . i.2.130) We argue in analogy to the logic applied in going from (11.2) −1 (a∗ )−1 = e2z 1 ∗ ·(σ ∗ )T = e2z 1 ∗ ·σ ∗ T = [a∗ ]T . It holds according to (11. e.127) Exploiting the hermitian property of σ.128) One can express.131) where a is given by (11. In fact.108) that the same transformation behaviour applies then for general Lorentz transformations.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.16. (11. state that σ µ in a new reference frame is σ µ = a σ µ a† (11.4) with w = 0. (11. 11.4: Spinor Tensors One can rewrite (11. 11. under rotations the Pauli matrices transform like (j = 1.103) ˙ identiﬁes σ µ as a tensor of type tαβ .124) using (11. −1 0 0 1 µ=3 .124) a σ µ [a∗ ]T Lν µ = σ ν .134) . 11.g.129.

˙ ˙ 0 −1 1 0 ˙ (11. 2. We will now consider the representation of σ µ .29)].110) to express σ αβ in terms 0 −1 1 0 0 0 0 1 (11. µ = 0. This yields 1 0 0 1 0 0 0 0 σ0 + σ3 σ1 + iσ2 (11. α. matrices.e. γ.135) This is the property surmised already above [cf. however. of (11. account for the 4–vector index µ.114). whereas ˙ account for the elements of the individual Pauli the ‘outer’ contravariant spinor indices.140) where we employed the expressions (11. i. state ˙ ˙ ˙ ˙ (σ11 )11 (σ11 )12 (σ12 )11 (σ12 )12 ˙ ˙ ˙ ˙ ˙ ˙ ˙ ˙ ˙ γδ (σ11 )21 (σ11 )22 (σ12 )21 (σ12 )22 ˙ ˙ ˙ ˙ . i.224) ˙ Using (11. hence. = σαβ ˙ ˙ ˙ ˙ ˙ (σ21 )11 (σ21 )12 (σ22 )11 (σ22 )12 ˙ ˙ ˙ ˙ ˙ ˙ ˙ ˙ 21 (σ )22 21 (σ )22 (σ21 ) (σ22 ) ˙ ˙ ˙ ˙ 21 22 Equating this with the r. (11.12) for of σµ yields together with (5. in fact. The ˙ desired change of representation(σµ )αβ −→ (σµ )αβ corresponds to a transformation of the basis of ˙ spin states f g f −→ = (11. and the covariant indices are moved outside. ˙ ˙ In (11.132) and (11.138) the ‘inner’ covariant spinor indices.141) σ αβ = σ0 − σ3 σ1 + iσ2 −σ1 − iσ2 σ0 + σ3 = 2 0 0 0 1 0 −1 0 0 . 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 . β.138) Note that all elements of σαβ are real and that there are only four non-vanishing elements.136) σαβ = = 2 ˙ σ1 − iσ2 σ0 − σ3 0 0 0 0 1 0 0 1 where on the rhs.137) σαβ ˙ = δαγ δβ δ . 1. account for the 4-vector µ. 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.136) results in the succinct expression 1 2 ˙ γδ ˙ (11. σµ in which the contravariant indices are moved ‘inside’.. the submatrices correspond to tαβ spinors. δ. the transformation is inverse to that of ordinary 4–vectors.133) is characterized through a 4-vector ˙ index µ...h.139) g −f g and. Spinor Formulation (11.370 Inverting contravariant and covariant indices one can also state Lν µ σµ = σν . Each of the 4 × 4 = 16 matrix elements in (11.e. We can summarize that σ µ and σµ transform like a 4–vector. ˙˙ (11. 3 as well as through two spinor indices αβ.e. i.s.

137) are in a basis (· · ·)αβ to a basis (· · ·)αβ results in ˙ σ αβ ˙ ˙ γδ = 2 (σ 11 )11 (σ 11 )12 ˙ ˙ ˙ ˙ (σ 11 )21 (σ 11 )22 ˙ ˙ (σ 21 )11 (σ 21 )12 ˙ ˙ ˙ ˙ 21 ) 21 ) (σ 21 (σ 22 ˙ ˙ 1 0 0 1 0 0 0 0 0 0 0 0 1 0 0 1 ˙ ˙ ˙ ˙ (σ 12 )11 (σ 12 )12 ˙ ˙ ˙ ˙ (σ 12 )21 (σ 12 )22 ˙ ˙ (σ 22 )11 (σ 22 )12 ˙ ˙ ˙ ˙ 22 ) 22 ) (σ 21 (σ 22 ˙ ˙ . ˙˙ (11.145) Let Aµ be a covariant 4–vector.121) of γ ˜ ˜ ˙ basis |αβ whereas the element of γ µ given by σ µ is in the basis |αβ . (11. One can write then the scalar product using (11. ˙ ˙ ˙ ˙ (11. (11. For this purpose we start from the ˜ µ .4: Spinor Tensors 371 and.146) Exploiting the property (11. employing then transformation (11.147) . Accordingly. This expression implies that the element of γ µ given by σ µ −1 is in the expression (11. (11.142) = This can be expressed 1 2 σ αβ ˙ ˙ γδ = δαγ δβ δ .138) one can conclude that the following property holds 1 2 ˙ γδ σ αβ ˙ = 1 2 σ αβ ˙ ˙ γδ = δαγ δβ δ .144) results in the simple relationship γ µ Aµ = ˜ 0 Aα β ˙ Aα β 0 ˙ .144) The Dirac Matrices γ µ in spinor notation We want to express now the Dirac matrices γ µ in spinor form. which in ˙ (11. ˙˙ (11.143) Combined with (11.115) γ µ Aµ ˜ = 0 ((σ µ )αβ )∗ Aµ ˙ 0 ˙ ∗ 1 γδ ˙ ˙ 2 ((σ )αβ ) Aγ δ (σ µ )αβ Aµ 0 ˙ ˙ αβ γ δ 1 ˙ 2 (σγ δ ) A ˙ = 0 .11. we write ˜ ˙ γµ = ˜ 0 (σ µ )αβ µ ) )∗ ((σ αβ 0 ˙ ˙ .140) to transform each of the four submatrices.

372 Spinor Formulation 11. ˙ (11. (11.149) (11.226) ˜ i γ µ ∂µ Ψ(xµ ) = 0 i ∂ αβ ˙ i ∂ αβ 0 ˙ ˜ ˜ Ψ(xµ ) = m Ψ(xµ ) . α ˙ = = m φα m χβ .150) .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 (11.147) allows us to rewrite the Dirac equation in the chiral representation (10.148) ˜ Employing Ψ(xµ ) in the form (11.

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

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

11) (12. 2m r 1 x1 (J3 p2 − J2 p3 + i p1 ) − k . we will compute the hydrogen spectrum using the additional symmetry. Now we scale the eccentricity vector as follows K= − m . In order to understand the aforementioned extra degeneracy. 2m r (12. The vector in (12. Therefore. For this purpose we ﬁrst note that J · = 0. In the following we consider the bound states. ijk Jk . in the subspace of the Hilbert space corresponding to a certain energy we can replace H by E.11. 2 which can be shown to satisfy [Ai .12) (12.15) which can be proved after very considerable algebra. 2m r 1 x3 (J2 p1 − J1 p2 + i p3 ) − k .8) is not a hermitian operator.3. Kj ] = i ijk Kk .5: Lorentz Invariant Field Equations 375 The vector points along the symmetry axis of the elliptical orbit and its length equals the eccentricity of the orbit. (12. Jj ] = i [Ki .13) It can be veriﬁed explicitly that its components commute with the Hamiltonian. which is valid even when a and b do not commute. We will also need the following identity [4] 2 = 2H J2 + m 2 + k2 . (12.16) Through some algebra [4] the following commutation relations can be veriﬁed [Ki . which have a negative energy E.14) This follows from a · (a × b) = (a × b) · a = 0.10) (12. (12. 2 1 B = (J − K). Aj ] = i ijk Ak . 2E (12.17) (12.21) .9) (12.20) (12. 2m r 1 x2 (J1 p3 − J3 p1 + i p2 ) − k .18) which complement the familiar angular momentum algebra of section 5.19) (12. (12. The corresponding quantum mechanical hermitian operator can be deﬁned by 1 1 2 3 = = = = x1 1 (p2 J3 − p3 J2 + J3 p2 − J2 p3 ) − k . We introduce the following new operators 1 A = (J + K).

. ijk Bk .22): A2 = a(a + 1) B 2 = b(b + 1) Following (12. By comparing to the rotation algebra introduced in chapter 5. . Spinor Formulation (12.33) (12. a = 0. b = 0.34) It follows that l has to have values in {0 = |a − b|. l. . 1. Chaos is described classically as exponential . 2 (12. Using this equation the energy eigenvalues can be written in terms of the eigenvalues of A2 and B 2 operators. . a = 0. which is identical to the the direct sum of the Lie algebra of of two SO(3) (or SU (2)) algebras.21) and (12. . 2 1 . .7) tells us that (2a + 1) = n. .15). . namely that J J > < A − B =0 A + B =2 A (12.14) we note that A2 − B 2 = J · = 0.26) (12.32) A comparison with (12.27) (12. . a lack of symmetry implies a lack of degeneracy in the energy levels of a quantum mechanical system. a + b = n − 1}. .30) 2 2 2 1 . . . Bj ] = i [Ai .28). .31) where we have used (12. . . the energy eigenvalues are found to be E=− mk 2 1 . 1.22) (12. . 1. H B. we can read oﬀ the eigenvalues of A2 and B 2 from (12. . Noticing that A2 and B 2 have the same eigenvalues because of (12. (12. The most extreme case of this is the quantum analogue of a classically chaotic system. This implies that a = b. In order to arrive at the spectrum a ﬁnal bit algebra is needed A2 + B 2 = J 2 + K 2 m 2 = J2 − 2E mk 2 1 = − − 4E 2 (12. H = 0.25) So far we have shown that the symmetry generators form an algebra. In contrast to the discussion above about extra symmetries.28) . can be seen to follow from the triangle inequality as applied to J = A + B.24) (12.376 [Bi . 1.29) (12. A. . 2 (2a + 1)2 2 2 (12. = 0. Bj ] = 0. . This illustrates the eﬀect of additional symmetries to the degeneracy structure of a quantum mechanical system.23) (12. Furthermore the bound on the orbital angular momentum.

) If the proton and the neutron are to be viewed as two linearly independent states of the same particle. 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. for example. which is a particle in box problem in two dimensions with a boundary which can be chosen arbitrarily.2 Isospin and the SU (2) ﬂavor symmetry The concept of isospin goes back to Heisenberg. that the proton and the neutron behave identically under the so-called strong interactions and that their diﬀerence is solely in their charge content. (12. (Strong interactions bind the atomic nucleus together. It is important to place the isospin concept in its proper historical context. who. in the sense that nearby trajectories in the phase space diverge from each other over time. analogous to the 1 spin-up and spin-down states of a spin. for the . (12. the isospin symmetry is also governed by an SU (2) group ‘rotating’ components in (12. we will be able to use the familiar Clebsch-Gordan coeﬃcients to combine the isospin of two particles the same way we added spin in chapter 6. This is known as level repulsion. Originally it was believed that isospin was an exact symmetry of strong interactions and that it was violated by electromagnetic and weak interactions.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.5: Lorentz Invariant Field Equations 377 sensitivity to initial conditions. mp = 939. In the next section we will proceed with the discussion of a symmetry. (Weak interactions are responsible. This was motivated by the observation that their masses are approximately equal: mp = 938.28M eV /c2 .36) into each other in abstract isospin space. 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. Following the mass-energy equivalence of special relativity E = mc2 . 12. A typical example of this so called quantum chaos is the quantum billiard problem.11. However. the classical trajectories will diverge from each other after successive bounces from the boundary. namely. after the discovery of the neutron in 1932. For example.57M eV /c2 . it is natural to represent them in terms of a two component vector. which was discovered by observing degeneracies in the particle spectrum. n= 0 1 .2 system p= 1 0 . another manifestation of chaos is the lack of independent operators commuting with the Hamiltonian. For a more detailed discussion of quantum chaos in billiard systems we refer the reader to [7] and the references therein.36) In analogy to the concept of spin regarding the rotations in 3-space as discussed in chapter 5. This enables us to utilize what we already know about the SU (2) symmetry group from the study of angular momentum.

43) Similarly. 1 we can re-write (12. . . 1 .40) + 2 1 1 .38) are ud. respectively. π − = |1. (mπ± = 139. I3 = 2 2 1 1 .6M eV /c2 . −1 . 2.− . are made up of these two quarks. −1 = = = 1 1 1 1 . including the spin and color quantum numbers of their constituent quarks.38) which have all nearly identical masses. u¯ and d¯ states. it can be seen as part of a larger (and more approximate) symmetry which is of great utility to classify observed particle families.− 2 2 . n = I = .378 Spinor Formulation beta decay). 0 π − = |1. Denoting the total isospin. Further than that. If the mass diﬀerence (or the energy diﬀerence) were to be to be purely electrostatic in nature. . Shortly we 1 will see how to describe both the pion and nucleon states as composites of more fundamental I = 2 states. spelling disaster for the stability of matter. the proton and the neutron can be written as totally symmetric uud and udd states. However. 0 . the three pions in (12. They u form an isotriplet: π + = |1.− 2 2 . π 0 = |1. They can be constructed with the same rules that have been used for angular momentum addition ¯ u in chapter 6.0M eV /c2 ). mπ0 = 135. 1 π 0 = |1.− . For a precise description of the two nucleons as composite states.39) All other isomultiplets. as good quantum numbers. the proton is the lighter of the two. 2 2 . For example. the fundamental representation of the isospin symmetry corresponds to the doublet that contains the so-called up and down quarks u= 1 1 . Therefore it remains a useful concept. 2 (12. We can describe isospin multiplets the same way we have described the angular momentum and spin multiplets. including the proton and the neutron. albeit it is a good approximate one. If it were otherwise the proton would be unstable by decaying into the neutron.42) (12. . 2 2 1 2 2 2 1 1 1 1 1 √ . 2 2 1 1 1 . I3 .36) as a multiplet with I = 2 1 1 p = I = .41) (12. 2 2 1 2 2 2 (12. the proton had to be heavier. Isospin symmetry is not an exact symmetry of strong interactions. I. and its third component. I3 = − 2 2 .− 2 2 1 2 2 2 1 1 1 1 .11. (12. as we shall see below. sec. (12. we refer the reader to [1]. (12. The mass diﬀerence between the neutron and the proton could then be attributed to the charge content of the latter. In the framework of the quark of model. d= 1 1 .37) As an example of a multiplet with I = 1 we have the three pions or π-mesons π + = |1.

Therefore the deuteron has to be an isosinglet state (|0. the up and down quarks are not the only quark ‘species’ . a number strange particles have been found which presumably contained a third quark species: the strange quark. such as proton and the neutron. 0 |1.45) Is the deuteron an isosinglet state or an isotriplet? If it were an isotriplet (|1. strongly interacting particles) which is about a GeV /c2 . baryon and meson. as some mesons discovered later are heavier than some baryons and so on. 0 = √ (|p > |n > −|n > |p >) 2 and that of an isotriplet |1. 0 ). By convention ¯ baryons have baryon number 1. we shall setup some terminology: baryons are qqq states. 1 = √ (|p > |n > +|n > |p >). The reader may know that the deuteron. muon. this remains only an inaccurate naming convention today. 0 ) we should have seen nn and pp bound states of comparable energy in nature (because of isospin symmetry). mesons have baryon number 0. In the next section we will be able to view the Gell-Mann–Nishijima relation in the light of the representation theory for the ﬂavor SU (3) symmetry. but such states do not exist.e. I3 = 0. The possibilities are that of an isosinglet 1 |0. If taken literally. to be −1 for the strange quark. the pions being examples thereof. We will now try to describe its wave function in terms of its constituent nucleons. mesons have intermediate mass ranges. The value of the strangeness quantum number is taken. .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. the neutrinos etc. All other composite states have their strangeness given by the sum of the strangeness content of their constituents. −1 = |p > |p >. 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). Therefore it has to have isospin. a hydrogen isotope.43). This is why isospin is such a good symmetry and why isomultiplets have nearly identical masses.47) (12. originally refer to the relative weight of particles. Following (12. It shall be noted here that the quark model was not invented until 1960’s. In the late 1940’s and early 1950’s. The names. this will be mathematically identical to adding two spins.37) and in analogy to (12.) are light. whereas mesons are q q states.or ﬂavors as they are commonly called. (12.11.46) (12. Now let us consider another example of combining the isospins of two particles. As it later turned out. where leptons (electron.48) (12.44) 2 where B is the baryon number and S is the strangeness. 2 = |n > |n > . (12. consists of a proton and a neutron. baryons generally are heavy. by accidental convention. and quarks have baryon number 1 . 1 |1. Before proceeding further. Naturally. but at the time the empirical concepts like isospin and strangeness quantum numbers were in use. The up and down quarks have strangeness zero. All antiparticles have their 3 quantum numbers reversed.

γ (f ) |T00 |I (i) I3 . which is an isoscalar. γ (i) is a reduced matrix element deﬁned in the same sense as in section 6. 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 ) = 1.49) The only assumption that we put in will be that the interaction is of the form V = α I(1) · I(2) .259) discussed in detail in sections 6.48) the initial states in (I) and (II) have isospin values |1. I3 = 1. γ (i) . respectively. 0 + |0. γ (i) (i) (12. We will eventually be able to compute ratios of scattering cross-sections between diﬀerent processes. Show that I(1) ·I(2) is an ‘isotensor’ of rank zero.259) in the present context: I (f ) I3 . 0 . For example.51) (12. such as the spatial dependence of the wave function and spin. (12. 1 4 in an isotriplet state and − 3 in an isosinglet 4 Using (12. Show that the expectation of I(1) · I(2) is state. where M is the scattering amplitude given by M = ﬁnal | α I(1) · I(2) | initial .45) and (12. γ (i) I (f ) . We will now employ the (isospin analogue) of the Wigner-Eckart theorem (6.8 to compute the ratio of the scattering amplitudes MI and MII . γ (i) = (f ) (i) (I (f ) I3 |00I (i) I3 ) (f ) (i) (−1)I −I (f ) (i) (12. 1 and |1. 0 ). σ. I3 (f ) = 1. (12. (f ) (i) (I (f ) I3 |00I (i) I3 ) is a Clebsch-Gordon coeﬃcient and I (f ) . γ (f ) T00 I (i) = 1. I3 . γ (f ) ||T ||I (i) . The cross-section. 1 and (1/ 2)(|1. Exercise. The dot product here refers to the abstract isospin space. γ (f ) ||T0 ||I (i) .7 and 6.50) .55) .7.54) Here T00 ≡ V = α I(1) · I(2) . For completeness let us start by restating the Wigner-Eckart theorem (6. I3 . consider (I) p + p → d + π + (II) p + n → d + π 0 (12.5 for the spin-analogue of the same problem.8. is proportional to |M|2 .53) 0 √ 1 2I (i) +1 I (f ) .38) and the fact that the deuteron is an isosinglet we know that the isospins of the ﬁnal states in (I) and (II) are |1.7 to the case of isospin. Consider the generalization of tensor operators discussed in section 6. . Refer to exercise 6.52) where γ (i) and γ (f ) denote degrees of freedom other than isospin. γ (f ) (f ) (i) (12. as discussed in the exercise above. √ According to (12.380 Spinor Formulation As an exercise on the implications of isospin symmetry we will consider nucleon-nucleon scattering. Exercise. The initial and ﬁnal states can be denoted in more detail as | initial | ﬁnal = = I (i) .

−2 2 1 1 3 2. The relevant Clebsch-Gordon coeﬃcients are easily evaluated: (11|0011) = (10|0010) = 1.63) 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. (c) π − + p → π 0 + n .60) (12. γ (f ) ||T00 || I (i) = 1. γ (i) 2 1 = (10|0010) √ I (f ) = 1. −2 . for example. γ (i) 3 +0. which is also manifested by a vanishing Clebsch-Gordon prefactor. [2] As a further example. 1 π − + p : |1. γ (i) = √ 2 1 (f ) (i) +√ I (f ) = 1. I3 = 0. (12.56) (12. −1 π0 + n : |1. We want to compute the ratio of total cross-sections assuming a similar interaction as in the previous example σ( π + + p → anything ) . γ (f ) T00 I (i) = 1. γ (i) 3 1 (f ) (i) I (f ) = 1.66) = 2 3 1 3 2.58) (12. γ (f ) ||T00 || I (i) = 1. −2 1 1 2. σII which is in approximate agreement with the observed ratio.64) There are more exotic possibilities.62) We can now write the ratio of the scattering amplitudes: MI 1 √ . −1 2 3 2 − + . I3 = 0. 0 1 1 3 3 2. Once again we need the isospins for the initial and ﬁnal states. I3 = 0.59) (12. σ( π − + p → anything ) The possibilities are (a) π + + p → π + + p . 2 = 3 2. . but these are not dominant at relatively low energies. (b) π − + p → π − + p . γ (f ) T00 I (i) = 0.57) (12. I3 = 0.65) (12.11. = 381 (12. 2 = 2. we will consider pion-nucleon scattering. particles with strangeness. involving.5: Lorentz Invariant Field Equations 1 (11|0011) √ I (f ) = 1. It follows σI = 2. (12. (12. 2 . 1 1 1 1 3 √ 2. which are obtained by a standard Clebsch-Gordan expansion π + + p : |1.61) MII Note that the second term in MII vanishes due to the isospin conservation. −2 1 1 √ .

m 1 2.68) − 2 3 M1 2 Guided by empirical data we will further assume that M 3 >> M 1 . K + . A computation similar to the previous example of nucleon-nucleon scattering yields (apart from common prefactors) the following amplitudes for the reactions in (12. for example.d = 1 . m . which can be produced by π − + p → K + + Σ− . S.69) 12. making an identiﬁcation of the underlying symmetry and the multiplet structure less straightforward.382 Spinor Formulation As in the example of nucleon-meson scattering we deﬁne the relevant matrix elements M3 = 2 M1 = 2 3 2.s = 0 . are up to 40% heavier. 0 0 1 . S(Σ− ) = −1 and S(π) = S(N ) = 0. however. (12.3 The Eightfold Way and the ﬂavor SU (3) symmetry The discovery of the concept of strangeness. Strange partners of the familiar nucleons. As the total cross-section is the sum of individual processes we obtain σ( π + + p ) σa = =3 σ( π − + p ) σb + σc again in approximate agreement with the observed value. while weak interactions did not. giving it a higher than usual lifetime. Hence Gell-Mann and independently Nishijima postulated the existence of a separate quantum number.71) u = 0 . m 1 2. was motivated by the existence of particles that are produced strongly but decay only weakly. it appears an obvious generalization to add another component for an extra quark to the isospin vector space 1 0 0 (12. It was assumed that strong interactions conserved S (at least approximately). such that S(K + ) = 1. which leads to the following 2 2 ratios for the cross-sections σa : σb : σc = 9 : 1 : 2.65) Ma = M 3 2 Mb = 1 M 3 + 2 M 1 3 3 √ 2 Mc = 2 3 M3 2 √ 2 (12. called strangeness. has a lifetime which is comparable to that of π + albeit being more than three times heavier. mentioned in the previous section. The classiﬁcation of the newly found particles as members of some higher multiplet structure was less obvious then the case of isospin. .67) which are independent of m. Hence the strangeness changing strong decays of K + (or Σ− ) were forbidden.70) (12. [2] (12. m V V 3 2. For instance. etc. In the light of the quark model.

The reason is that the other known quarks. namely charm. have to be elements of the group SU (3) (which we will investigate closely very soon). This is believed to be an exact symmetry of strong interactions. First particle multiplets were identiﬁed as representations of the SU (3) group. The unitarity relation imposes 9 constraints on the total of 18 real degrees of freedom of a 3 × 3 complex matrix. In this perspective the ﬂavor SU (3) symmetry may appear to be mainly of historical interest. which set the hadronic energy scale. ak . [2] ) Before discussing the signiﬁcance and the physical implications of the quark model.72) a∗ ak . This group of 3 × 3 unitary matrices is denoted by U (3). SU (N ). in reference to Lie groups. of a. called quantum chromodynamics. Given a complex vector.3 for a brief discussion of gauge transformations). 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. history followed the reverse of this path. for a period they were considered as useful bookkeeping devices without physical content. still leaves the norm invariant. It is seen that such a matrix U has to satisfy the following k U † = U −1 . The quarks posses another quantum number. called color.71) into each other.11.73) To verify that all such matrices form a group. The reader shall note that most of what is being said trivially generalizes to other unitary groups. The reader is also invited to revisit section 5. but there are a number of subtle diﬀerences. becomes increasingly inaccurate for Nf > 3. giving rise to the quark model. the same way nucleons and pions were identiﬁed as representations of the isospin SU (2) symmetry. which again form representations of an SU (3) group. However. where Nf is the number quark ﬂavors.71). (12. in fact modern theory of strong interactions is a ‘gauge theory’ of this color group. while preserving the norm. should correspond to. eiφ . Flavor SU (Nf ) symmetry on the other hand. Lie algebras and related concepts. Multiplying U by a phase. Then came the question of what the fundamental representation. However SU (3) symmetry appears in another and much more fundamental context in strong interaction physics. The bottom and top are even heavier. it is the group . Since arbitrary phase factors are of no physical interest. (The reader is referred to section 8. of three dimensions. Hence the group U (3) has 9 dimensions.1 whenever necessary. unitary relation k (12. as in (12. Because of this additional constraint SU (3) has 8 dimensions.74) for any two unitary matrices U and V . (12. bottom (beauty) and top (truth) are signiﬁcantly heavier than the hadronic energy scale. we want to ﬁnd those transformations ak → Ukl al that preserve the norm. we will establish some mathematical preliminaries about the group SU (3). but we will stick to N = 3 in the following. As quarks were never directly observed. In many respects it will resemble the more familiar group SU (2) discussed in some detail in chapter 5. we observe that (U V )† = V † U † = V −1 U −1 = (U V )−1 . (The ‘bare’ mass of the charm quark is already heavier than the two nucleons.5: Lorentz Invariant Field Equations 383 In this case the transformations that ‘rotate’ the components of (12.

via the anticommutator relation 4 [λj .78) where fjkl are the antisymmetric structure constants similar to the jkl of SU (2) given in (5. obey the following relation tr(λj λk ) = 2δjk . We can also introduce the constants. As in the case of SU (2) higher dimensional representations obeying the same structure can be found. λk ]+ = δjk + 2δjkl λl .79) (12. Therefore we will express elements of SU (3) as U = ei k (12.7. the meaning being apparent from the context. U . can be written in the form U = eiH where H is a hermitian matrix.76) where λk are 8 linearly independent matrices forming the basis of the Lie algebra of SU (3). . As discussed in section 5. 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 .79). The Lie algebra structure is given by the commutators of λk [λj . The reader is invited to compare the structure of SU (3) to that of SU (2) discussed in section 5. since det(eA ) = etrA .75) αk λk (12. . (12. The fundamental relation to be preserved is (12. which can be veriﬁed explicitly as matrix identities. λk ] = 2ifjkl λl . −2 λ4 λ6 λ8 The generators. λ2 = . λk . (We shall at times refer to the Lie algebra with the name of the group. δjkl .77) λ1 .1.80) This is the fundamental or deﬁning representation of SU (3). regardless of the dimension of the representation. λ7 = 0 0 . (12. λ3 = 0 −1 0 . any unitary matrix.32).) The unit determinant condition requires that all λk are traceless. λ5 = . 0 0 0 . 3 (12.384 Spinor Formulation SU (3) and not U (3) that is of main interest.

2 1 [T0 . In the case of SU (2) the identiﬁcation of ‘ladder’ operators {J+ .71) show that T0 is the isospin operator.87) Exercise.77) is not the most useful basis in the study of SU (3).87) the commutation relations between the generators can be expressed in a succinct manner.93) (12.91) [T0 . these relations have been used to construct the angular momentum spectrum as well as the function space representation of the rotation algebra. (12. V± ] = ± V± . 2 Y = √ F8 . T0 ] = 0. J− } proved useful. First. 2 (12. We ﬁrst introduce the F-spin operators 1 Fi = λ i .84) (12. namely spherical harmonics. In the basis (12. T 0 = F3 . T± ] = ±T± .89) (12. which satisﬁed an ‘eigenvalue equation’ [J0 . 3 (12. Exercise. T± ] = 0. 2 (12. we have [Y. (Recall that the strange quark has S = −1 by convention).85) (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 .86) (12. (12.92) (12. [Y.94) . V± = F4 ± iF5 . J± ] = ± J± . T0 } and [Y.81) In chapter 5. starting with the observation that Y = B + S. I3 . V± ] = ±V± . [Y.88) which deﬁnes (not uniquely) a maximal set of mutually commuting operators {Y.83) (12. U± ] = U± . Y is called the hypercharge.5: Lorentz Invariant Field Equations 385 As it turns out the set of generators in (12. Using the convention in (12. Derive the Gell-Mann–Nishijima relation (12. (12. U± ] = ±U± . U± = F6 ± iF7 . The generators of SU (3) can be arranged into a very similar form to that of SU (2).90) (12.11. 1 [T0 .44).

105) C2 = jkl djkl λj λk λl In general the number of independent Casimir operators of a Lie group is equal to its rank. The interested reader is instead referred to a very readable account given in chapter 12 of [4].104) (12. [T+ . we will use these commutation relations to construct representations of SU (3). V+ ] = 0. T− ] = 2T0 . U0 and V0 are linear combinations of T0 and Y . The same way the angular momentum ladder operators have been used to construct the representations of SU (2).98) (12. C1 = k λ2 = − k 2i 3 fjkl λj λk λl . while SU (3) has rank 2. In the case of SU (2) the representations lay on a line on the J0 axis.5. However. [V+ . . V− ] = T− .88). [U+ . When the basis of a Lie algebra is expressed in such a way to satisfy the form of the eigenvalue relations as given above. Another important property of SU (2) is the existence of an operator (namely the total angular momentum.97) which relate commutators of generators with opposite eigenvalues to the maximal set {Y. V− ] = −U− . similar to the way it was done in section 5. A formal deﬁnition and a detailed discussion of the Cartan-Weyl form is beyond the scope of this chapter. 3 [U+ . T0 }. V± . V+ ] = 0.96) (12.102) (12. As in the case of J 2 and SU (2). U± } to this maximal set by ‘eigenvalue equations’ and [T+ .103) Any remaining commutators follow from hermiticity. This form is essential for easy labeling of the representations of the group.101) (12. as the relation between the states in a given representation can be conveniently expressed in terms of ladder operators.95) (12. [U+ . J 2 ) which commutes with all of the generators. [T+ . The maximum number of mutually commuting generators of a Lie algebra is called its rank. the representations will now lie in a T0 − Y -plane. since there are two mutually commuting generators in SU (3) as given in (12.99) (12. Casimir operators can be used to label irreducible representations of the Lie algebra. U− ] = 0. Finally. U+ ] = V+ . we have [T+ . V− ] = 2 (12. 2 3 Y + T0 = 2V0 . (12. Note that.100) (12. [T+ . We can construct two such independent Casimir operators for the group SU (3). SU (2) has rank 1. jkl (12. U− ] = Y − T0 = 2U0 . it is said to be in Cartan-Weyl form.386 Spinor Formulation which relate the remaining generators {T± . An operator which commutes with all generators of a Lie group is called a Casimir operator. Thus.

y . For example. y .2) shows the representation D(2. 2 1 = γ t3 . The interested reader is referred to [4]. For example. y . 2 (12. The representations of SU (3) are constructed analogous to those of SU (2) by identifying the ‘boundary’ states annihilated by raising (or lowering) operators. U± . 2 (12. respectively. (12.5: Lorentz Invariant Field Equations 387 The utility of Casimir operators arises from the fact that all states in a given representation assume the same value for a Casimir operator. y 1 = α t3 ± .106) (12.109) The same way that J± |m is proportional to |m ± 1 in the case of the angular momentum algebra. 1) correspond to the triplets of quarks and antiquarks. 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. V± . For example.5 how to label the irreducible representations of the angular momentum algebra SU (2) in terms of the value of the total angular momentum. y Y |t3 . y . 2 1 = β t3 ± . 1) as an example. 0) and D(0. y : T0 |t3 . y . y = t3 |t3 . Because of (12. q) and it has a dimensionality of 2 (p + 1)(q + 1)(p + q + 2). y ± 1 . Now we will construct explicit representations of SU (3). the pion family forms part of an octet corresponding to the D(1.107) 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 .112) The eﬀect of these operators to the states in the y − t3 plane have been outlined in Fig. |t3 . The representations D(1. = y |t3 . y V± |t3 . A notation suggestive of the dimensionality of the representation can be used to identify representations.3).antiquark pair. . The representations for hexagons with sides of length p and q in the T0 − Y -plane. The details of this procedure is beyond the scope of this chapter.111) (12. we have U0 |t3 . Such a representation is labeled as 1 D(p.1). (See Fig.108) (12. it was shown in section 5.11. especially chapters 7 and 8.88) we can label states by the eigenvalues of T0 and Y operators. y V0 |t3 . All other states of the representation are then constructed by successive application of ladder operators T± .y ± 1 . 2 1 t3 ) |t3 . Figure (12.) All other representations can be constructed by combining these two conjugate representations. As another example for the representations of SU (3). y 3 = ( y− 4 3 = ( y+ 4 1 t3 ) |t3 . y U± |t3 . 1) representation. (12. we have T± |t3 . (12. y .110) (12. For example the pion octet (or any other meson octet) is therefore realized as states of a quark . This property can be used to label representations in terms of the values of the Casimir operators.

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

0) and D(0. This way the quark-antiquark states can be represented as follows [3] ⊗ [¯ = [8] ⊕ [1] 3] (12. for example. 2 3 1 1 − .3: D(1. .118) ¡ ¢ ¨ ©§ ¨¥¦¤ " ! . 1) are denoted by [3] and [¯ respectively. 0) or the fundamental representation of ﬂavor SU (3) symmetry. in the case of SU (2).114) Figure 12.5: Lorentz Invariant Field Equations 389 D(1. to the case of adding two spin triplet states. 1) is written as [8] etc.11. 3 (12. £ (12. The octet D(1.115) (12. where we would write [3] ⊗ [3] = [1] ⊕ [3] ⊕ [5]. 3].116) (12. This expansion can be compared.113) The additional singlet state corresponds to the η meson. 2 (12. − . . 2 3 2 0. The three quarks in the fundamental representation can now be written as u = d = s = 1 1 .117) The Gell-Mann–Nishijima relation can then be succinctly expressed as 1 Q = y + t3 .

Am. D. 3 £ Spinor Formulation (12.-F. Wiley. 1987. M¨ller.390 from which the quark charges follow 2 Qu = . [4] Quantum mechanics: symmetries. J. 1972. North holland. 3 1 Qd = − . Halzen. M. T.4: The baryon octet as a D(1. K. Kaufman. Phys. [7] Expansion method for stationary states of quantum billiards. [2] Introduction to elementary particles. References [1] Quarks and leptons: an introductory course in particle physics. Harter. L. Martin. G. 133. L. (1999). ¡ ¢ S - S 0 ¤ ¥ S + . Greiner. Wiley. F. Kosztin. D. Greiner. 3 1 Qs = − . Wiley. B. W.-P. Oxford. I. W. 1) representation of SU (3). 1989. Griﬃths. Schulten. W. 1993. 67 (2). Eisenberg. 1984.120) (12. J. [3] Microscopic theory of the nucleus.121) L 0 X - X 0 Figure 12. and spectroscopy.119) (12. p. Cheng. [6] Gauge theory of elementary particle physics. D. Li. A. 1984. dynamics. Springer-Verlag u [5] Principles of symmetry.

- Annihilation and Creation Operators.Chapter 8
- Relativistic Quantum Mechanics
- 51891292 Quantum Mechanics I
- Quanten_physics
- Sakurai - Modern Quantum Mechanics Rev Ed- Solutions Manual
- 611 Electromagnetic Theory II
- Relativistic Quantum Mechanics
- Quantum Mechanics Lecture
- Classical Electrodynamics 3rd Ed J.D. Jackson - Solutions - 214 Pg
- Kittel Charles - Introduction to Solid State Physics 8Th Edition - Solution Manual
- Advanced Quantum Mechanics
- Merzbacher Quantum Mechanics
- Cambridge.quantum.mechanics.apr.2009.eBook ELOHiM
- Quantum Mechanics Review
- Book
- Lectures on Quantum Mechanics - Gordon Baym
- Problems and Solutions Quantum Mechanics - Y. K. Lim, World Scientific (1998, 751p)
- Relativistic Quantum Mechanics-Bjorken j.d., Drell s.d.
- Peskin Chapter 4
- Solution Manual String Theory
- Wick Contractions in Latex
- [Ashcroft & Mermin]Solid State Physics Solution
- What is String Theory - J. Pol Chin Ski
- particle physics 3
- Fetter Walecka Quantum Theory
- 3063 Exam2 Solutions Sp06
- Lok C. Lew Yan Voon Electronic and Optical Properties of Semiconductors a Study Based on the Empirical Tight Binding Model
- Quantum Mechanics PyEd 342
- path-integral
- Advanced Quantum Mechanics
- Notes on Quantum Mechanics

Are you sure?

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

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

Get the full title to continue

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

scribd