Notes on Quantum Mechanics

K. Schulten
Department of Physics and Beckman Institute University of Illinois at Urbana–Champaign 405 N. Mathews Street, Urbana, IL 61801 USA
(April 18, 2000)

Preface Preface

i

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

ii

Preface

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

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

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

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

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

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

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

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

iii

iv 5.4 5.5 5.6 5.7 5.8 5.9 5.10 5.11 5.12

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

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

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

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

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

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

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

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

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

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

vi

Contents

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

· ] in conventional differential calculus does not correspond to a differentiated function. t) (1. i. df = M ∂xj dxj . at the ends of the time interval of the trajectories the variations vanish. · ) is a function differentiable in its three arguments. It is also important to appreciate that δS[ · .. t) = L(q(t). We attempt to evaluate t1 S[q(t) + δq(t)] = t0 ˙ ˙ dt L(q(t) + δq(t).6) For a proof we can use conventional differential calculus since the functional (1. e. q(t). q(t) + δq(t) is a ‘slightly’ different trajectory than q(t). δq(t)] = dt − δqj (t) + δqj (t)   ∂qj dt ∂ qj ˙ ∂ qj ˙ t0 j=1 j=1 t1 .2 Lagrangian Mechanics Note: δq(t) describes small variations around the trajectory q(t). q(t) + δ q(t).5) where L( · . t1 ] where the integrand is a function L(q(t). t) (1. It holds   M  M ∂L  t1 d ∂L ∂L δS[q(t). i. q(t) + δ q(t). t) of the configuration d vector q(t). ∂f in case of a function of M variables. ˙ Theorem: Let S[q(t)] = t0 dx dt t1 ˙ dt L(q(t). We focus on such functionals because they play a central role in the so-called action integrals of Classical Mechanics. q(t).e. but rather to a differential of the function which is simply the df differentiated function multiplied by the differential increment of the variable. 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.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. (1.10) From this follows (1.6) immediately. t0 (1. For this purpose we consider M ˙ ˙ ˙ L(q(t) + δq(t). · .. dt q(t).8) We note then using d dt f (t)g(t) ˙ = f˙(t)g(t) + f (t)g(t) ∂L δqj ∂ qj ˙ − d ∂L dt ∂ qj ˙ δqj .e.7) through Taylor expansion and identification of terms linear in δqj (t).g. j=1 We will now consider a particular class of functionals S[ ] which are expressed through an integral d over the the interval [t0 . df = dx dx or.6) is expressed in terms of ‘normal’ functions. the velocity vector dt q(t) and time t. equating these terms with δS[q(t). ˙ In the following we will often use the notation for velocities and other time derivatives d q(t) = q(t) and dtj = xj . t) + j=1 ∂L ∂L δqj + δ qj ˙ ∂qj ∂ qj ˙ + O( 2 ) (1. δq(t)]. .

y(x1 ) = y1 . The proof of the above theorem starts from (1.11) The extremals qe (t) can be identified through a condition which provides a suitable differential equation for this purpose.16) This property holds for any δqj with δq(t) ∈ F .5) assume extreme values. it holds in case δq(t0 ) = δq(t1 ) = 0 that qe (t) is an extremal.6) which reads in the present case     M ∂L t1 d ∂L δS[q(t). . . The length of such path can be determined starting from the fact that the incremental length ds in going from point (x. y(x)) to (x + dx. . if and only if it satisfies the conditions (j = 1. . Let us assume that the two points are connected by the path y(x). δq(t)] = 0 for all δq(t) ∈ F . t1 ] ⊂ R → R holds t1 (1. hence. y2 ).12) The proof of this theorem is based on the property Lemma: If for a continuous function f(t) f : [t0 . dx (1. t1 ]. This condition is stated in the following theorem.12) for j = 1. M and δqj (t0 ) = δqj (t1 ) = 0 follows according to (1. (1. M ) d dt ∂L ∂ qj ˙ − ∂L = 0 ∂qj (1. δq(t)] = dt − δqj (t) .17) . then f (t) ≡ 0 on [t0 . Theorem: Euler–Lagrange Condition For the functional defined through (1. y(x + dx)) is ds = (dx)2 + ( dy dx)2 = dx dx 1+( dy 2 ) . 2. y(x2 ) = y2 .14) for any continuous function h(t) ∈ F with h(t0 ) = h(t1 ) = 0.16) the property δS[qe (t). M . For this purpose we define Definition: An extremal of a differentiable functional S[ ] is a function qe (t) with the property δS[qe (t).13) dt f (t)h(t) = 0 t0 (1. · ] ≡ 0 and. On the other side. the above theorem.1. .5).15) We will not provide a proof for this Lemma.12) for j = 1. 2. . .   ∂qj dt ∂ qj ˙ t0 j=1 (1. According to the Lemma above follows then (1. y1 ) and (x2 . . from (1.1: Variational Calculus 3 We now consider the question for which functions the functionals of the type (1. An Example As an application of the above rules of the variational calculus we like to prove the well-known result that a straight line in R is the shortest connection (geodesics) between two points (x1 . . (1. . 2.

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

t) as follows B E = ×A 1 ∂A .26) (1. t) .33) .24) ∂qj dt ∂ qj ˙ ∂qj dt which are obviously equivalent to the Newtonian equations of motion. c ∂t (1. Equations (1. t) dt c where dr = v and where F (r. The equations of motion for such a particle are d q ˙ (mr) = F (r. t) and a vector potential A(r. x3 (t)) moving in an electromagnetic field described through the electrical and magnetic field components E(r.28) can be satisfied implicitly if one represents the fields through a scalar potential V (r.1. These conditions read in the present case ∂L d ∂L ∂U d − = 0 → − − (mj qj ) = 0 ˙ (1. t) describes the charge current density. t) + V (r.28) (1. t) = V (r. x2 (t).22. t) describes the charge density present in the field and J(r.30) (1. t) 1∂ K(r.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. respectively. t) obey the Maxwell equations ×E + ·B ×B − ·E 1 ∂ c ∂t E 1 ∂ c ∂t B (1.25) = 0 = 0 4π J c = 4πρ = (1. t) and B(r. Particle Moving in an Electromagnetic Field We will now consider the Newtonian equations of motion for a single particle of charge q with a trajectory r(t) = (x1 (t).32) (1. t) = A(r. t) + v × B(r. 1. t) is the Lorentz force.30.31) = − V − Gauge Symmetry of the Electromagnetic Field It is well known that the relationship between fields and potentials (1. t).25) of a particle moving in the field. The transformation which leaves the fields invariant is A (r.23). t) = q E(r.29) where ρ(r.27) and (1.31) allows one to transform the potentials without affecting the fields and without affecting the equations of motion (1. t) and B(r.27) (1. t) − K(r. dt The fields E(r. t) c ∂t (1. F (r. 1.

dt ∂x1 c We notice using (1. if one assumes for a particle the Lagrangian ˙ L(r.25) follows from the Hamiltonian Principle of Least Action.37) The results (1.37) is d dt ∂L ∂ x1 ˙ = d q dA1 d q (mx1 ) + ˙ = (mx1 ) + ˙ dt c dt dt c ∂A1 ∂A1 ∂A1 x1 + ˙ x2 + ˙ x3 ˙ ∂x1 ∂x2 ∂x3 .g.38) ∂L ∂ x1 ˙ − ∂L = 0. equivalent.. .37) is ∂L ∂V q = −q + ∂x1 ∂x1 c The first term in (1. 1.36) in the Newtonian equation of motion.40) (1. in fact.35) shows that the Newtonian equations of motion and the Euler–Lagrange conditions are. 1. t) + A(r.35) is equivalent to the Euler–Lagrange condition d dt The second term in (1.6 Lagrangian of Particle Moving in Electromagnetic Field Lagrangian Mechanics We want to show now that the equation of motion (1. B3 = ∂A2 ∂x1 (1.36) This expression allows us to show that (1. r.35) − ∂A1 ∂x2 [v × B]1 = x2 B3 − x3 B2 = x2 ˙ ˙ ˙ ∂A2 ∂A1 − ∂x1 ∂x2 − x3 ˙ ∂A1 ∂A3 − ∂x3 ∂x1 . namely.38. t) = 1 q mv 2 − q V (r.34) For this purpose we consider only one component of the equation of motion (1.40. (1. (1.39) ∂A1 ∂A2 ∂A3 x1 + ˙ x2 + ˙ x3 ˙ ∂x1 ∂x1 ∂x1 . ∂x1 (1. Comparing then (1.41) with (1. (1.30).41) which is identical to the term (1. e.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. d ∂V q (mv1 ) = F1 = −q + [v × B]1 .25). t) · v . 2 c (1.

q. i. encountered above in connection with the transformations (1. t) 1 1 ∂K q mv 2 − q V (r.44) implies that the gauge transformation amounts to adding a constant term to the action integral.33) of electyromagnetic potentials. t) + K ·v (1. and since these properties allow one often to simplify mathematical descriptions. 1. t) the Lagrangian.32.43) To prove this theorem we determine the action integral corresponding to the transformed Lagrangian t1 S [q(t)] = t0 ˙ dtL (q. 2 c = A(r.34) q 1 q d ˙ L (r.1. For this purpose we consider the transformation of the single particle Lagrangian (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. r. t) + A(r.45) and reordering terms yields using (1. t) + A (r. Eq.32. We want to demonstrate now that the transformation (1. 1. One can conclude then immediately that any extremal of S [q(t)] is also an extremal of S[q(t)]. q. t) − + 2 c ∂t c 1 q = mv 2 − q V (r. q.45) .33) of electromagnetic potentials. The factor q has been introduced to make this c transformation equivalent to the gauge transformation (1. t) . a term not affected by the variations allowed.3: Symmetry Properties 7 1. 1. q. t) c t1 t0 t1 t0 ˙ dtL(q. dt ∂x1 ∂x2 ∂x3 ∂t ∂t (1.e. This term is d ∂K ∂K ∂K ∂K ∂K K(r. t) = mv 2 − q V (r. Note that one adds the total time derivative of a function K(r.32.43) into (1.42) is. 2 c c dt Inserting (1. t) = q K(q. t) c t1 t0 = S[q(t)] + (1. appear in a different. We will consider in the following two symmetries.32.44) Since the condition δq(t0 ) = δq(t1 ) = 0 holds for the variational functions of Lagrangian Mechanics. t) = x1 + ˙ x2 + ˙ x3 + ˙ = ( K) · v + . t) · v + K(r. equivalent to the gauge transformation (1. (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.46) (1. t) + q K(q.33) of electromagnetic potentials. t) + dt c (1. t) · v . r. in fact. t) = L(q. gauge symmetry and symmetries with respect to spatial transformations. t) L (q.42) This transformation is termed gauge transformation.33) ˙ L (r. They are the subject of the following theorem. 1.. The gauge symmetry. surprisingly simple fashion in Lagrangian Mechanics.

t).42) corresponds to replacing in the Lagrangian potentials V (r. therefore. Definition: Infinitesimal One-Parameter Coordinate Transformations A one-parameter coordinate transformation is decribed through r = r (r. ˆ (1. ˆ ˆ ˆ Examples of Infinitesimal Transformations The beauty of infinitesimal transformations is that they can be stated in a very simple manner. The corresponding infinitesimal transformation is defined for small r (r.33). t). However.49) =0 ∈ R (1. where the origin of is chosen such that r (r. In this case the transformation can be written in matrix form      cos −sin 0 x1 x1  x2  =  sin cos 0   x2  (1. in the case of central force fields which are invariant with respect to rotations of coordinates around the center. The transformations considered are specified in the following definition. A(r. the reason being that our interest lies in achieving familiarity with the principles (just mentioned above ) of symmetry properties rather than in providing a general tool in the context of Classical Mechanics. i. In case of a translation transformation in the direction e nothing new is gained.50) A non-trivial example is furnished by the infinitesimal rotation around axis e ˆ r = r + e×r . 0) = r . and for the introduction of infinitesimal transformations which will provide a crucial method for the study of symmetry in Quantum Mechanics. t).48) In the following we will denote unit vectors as a. r ∈ R . ) . ) = r + R(r) + O( 2 ) .51) We would like to derive this transformation in a somewhat complicated. A (r. R(r) = through ∂r ∂ (1. We have proven. The following description of spatial symmetry is important in two respects.8 Lagrangian Mechanics Obviously.32. for such vectors holds a · a = 1. we like to ˆ provide the transformation here anyway for later reference r = r + e. the equivalence of (1. We consider now invariance properties connected with coordinate transformations. r.e. t) by gauge transformed potentials V (r. for the connection between invariance properties and constants of motion. ˆ (1.47) (1. but nevertheless instructive way considering rotations around the x3 –axis..42) and (1. for example. The transformations we consider are the most simple kind. 1. Such invariance properties are very familiar.52) x3 0 0 1 x3 . which has an important analogy in Quantum Mechanics. the transformation (1.

53). We have used here the notation corresponding to single particle motion. ˙ ∂qk ˙ Inserting these infinitesimal changes of qj and qj into the Lagrangian L(q.56) ∂Qj qk . Invariance implies L = L.57)   M Qj j=1 d dt ∂L ∂ qj ˙ + ∂L d Qj ∂ qj dt ˙ = d  dt M Qj j=1 ∂L  = 0 ∂ qj ˙ (1. 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. the second and third term ˙ k=1 in (1. however. a classical mechanical system is invariant with respect to a coordinate transformation ˙ a constant of motion exists.55) (1. t) is invariant with respect to an infinitesimal transformation q = q + Q(q). that along the classical path holds d ∂L ∂L the Euler-Lagrange condition ∂qj = dt ( ∂ qj ) one can rewrite the sum of the second and third term ˙ in (1. q.e.51) reads r = r − x2 e1 + x1 e2 ˆ ˆ (1. The property has been shown to hold in a more general context. r) which is constant along the classical path of the system. expansion. the property holds for any system. by Noether.58) From this follows the statement of the theorem. a quantity C(r. then M ∂L j=1 Qj ∂ xj is a constant of motion.3: Symmetry Properties 9 In case of small this transformation can be written neglecting terms O( 2 ) using cos = 1 + O( 2 ). t) yields after Taylor ˙ 2 ).1.53) x3 x3 0 0 0 x3 One can readily verify that in case e = e3 (ˆj denoting the unit vector in the direction of the ˆ ˆ e xj –axis) (1. neglecting terms of order O( M ˙ ˙ L (q. We consider here only the ‘particle version’ of the theorem. t) + j=1 ∂L Qj + ∂qj M j. i. ˙ We have generalized in this theorem the definition of infinitesimal coordinate transformation to M –dimensional vectors q. q.54) which is equivalent to (1... i.57) must cancel each other or both vanish. In the context of Lagrangian Mechanics this implies that such transformation leaves the Lagrangian of the system unchanged. sin = + O( 2 )        0 − 0 x1 x1 x1  x2  =  x2  +  0 0   x2  + O( 2 ) . namely for fields rather than only for particle motion. . q.57) ∂ d where we used dt Qj = M ( ∂qk Qj )qk .e. In order to prove Noether’s theorem we note qj qj ˙ = qj + = qj + ˙ k=1 Qj (q) M (1. t) = L(q. q. (1.k=1 ∂L ∂Qj qk ˙ ∂ qj ∂qk ˙ (1. Anytime. Theorem: Noether’s Theorem ˙ If L(q. Using the fact.

j = 1. not infinitesimal transformations. is conserved. 3) 3 Qj j=1 ∂L = e· ˆ ∂ xj ˙ 3 ej mxj = e · mv . 3 and. 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.e. j = 1. A calculation similar to that in (1.51). 2. In this case we have Qj = ej · e. ˆ ˙ ˆ j=1 (1. ˆ The important result to be remembered for later considerations of symmetry transformations in the context of Quantum Mechanics is that it is sufficient to know the consequences of infinitesimal transformations to predict the symmetry properties of Classical Mechanics.10 Application of Noether’s Theorem Lagrangian Mechanics We consider briefly two examples of invariances with respect to coordinate transformations for the Lagrangian L(r. We will now investigate the consequence of rotational invariance as described according to the infinitesimal transformation (1. 2. It is not necessary to investigate the consequences of global. .50). 2 We first determine the constant of motion in case of invariance with respect to translations as defined in (1. In this case we will use the same notation as in (1. v) = 1 mv 2 − U (r). hence. of course.59) yields the constant of motion ˆ e (ˆ × r) · mv. except using now Qj = ej · (ˆ × r). It was. to be expected that this is the constant of motion. i. Noether’s theorem yields ˆ ˆ the constant of motion (qj = xj . for which translational invariance holds.59).59) We obtain the well known result that in this case the momentum in the direction.

1 The Double Slit Experiment Will be supplied at later date 2. t). t) is normalized d3 r |ψ(r. is unity. t ∈ [t0 . placed into the space at some particular time and with a detector behind each slit. r.1) (2. The probability that the particle is detected at space–time point r. t1 ] in which the particle is known to exist. t) where z is the conjugate complex of z. t) ψ : R3 ⊗ R → C. At present one may think of f (r) as a sum over δ–functions which represent a multi–slit screen. The probability to detect the particle with a detector of sensitivity f (r) is d3 r f (r) |ψ(r.2 Axioms for Quantum Mechanical Description of Single Particle ˙ Let us consider a particle which is described by a Lagrangian L(r. t is described in Quantum Mechanics. 11 . 4.4) a condition which enforces that the probability of finding the particle somewhere in Ω at any particular time t in an interval [t0 .Chapter 2 Quantum Mechanical Path Integral 2. t)|2 = 1 Ω ∀t. 1. The wave function ψ(r. t)|2 Ω (2. t is |ψ(r. We provide now a set of formal rules which state how the probability to observe such a particle at some space–time point r. t1 ] . t)|2 = ψ(r.3) where Ω is the space volume in which the particle can exist. 2. t)ψ(r. The particle is described by a wave function ψ(r.2) (2. 3. (2.

t ∈ (2. i. t )φ(r. i. the propagator is unitary.4) holds for all times.12 Quantum Mechanical Path Integral 5.8) 7. t.7) t. t|r0 . . t|r . t|r . described by ψ(r.8) a knowledge of the state at all space points r ∈ Ω at some intermediate time t . t0 ) = δ(r − r0 ). t ) = δ(r − r ) Ω (2. t0 ) = r(t0 )=r0 d[r(t)] Φ[r(t)] . t ) Ωd 3 r |ψ(r. hence.e. t1 ]) Ωd Ωd 3r Ωd 3r Ωd 3r 3 r |ψ(r. t ) φ(r .. t ∈ [t0 . t ) ψ(r . t|r . The following so-called completeness relationship holds for the propagator (t > t [t0 . (t > t . This is different from the classical situation where the particle follows a discrete path and. t0 ) = Ωd 3r N −1 Ωd 3r N −2 ··· Ωd 3r 1 φ(r. t ) t > t . t|r .9) for N → ∞. The time evolution of ψ(r.10) We have introduced here a new symbol.6) φ(r. t1 ] yields an expression of the form r(tN )=rN φ(r. t |r0 . 8.5) 6. ψ(r0 . t1 ] (2. The generalization of the completeness property to N − 1 intermediate points t > tN −1 > tN −2 > . t. > t1 > t0 is φ(r. tN −1 |rN −2 . t0 ) = φ(r. tN −1 ) φ(rN −1 . 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. namely the point on the classical path at time t . The state of a system at time t. This must hold for any ψ(r . t ) which requires d3 r φ(r. t1 |r0 .. . This symbol will be defined further below. t|r0 . . t) = Ω d3 r φ(r. t| rN −1 . t ∈ [t0 . Since (2. t ) φ(r. t1 ]) d3 r φ(r.e. at any intermediate time the particle needs only be known at one space point. The definition will actually assume an infinite number of intermediate times and express the path integral through integrals of the type (2. tN −2 ) · · · φ(r1 . t|r . t|r0 . the path integral r(tN )=rN d[r(t)] · · · r(t0 )=r0 (2. t). (2. t = )|2 = 1 . requires then according to (2. (2. t ) ψ(r .11) which denotes an integral over all paths r(t) with end points r(t0 ) = r0 and r(tN ) = rN . t0 ) . t)|2 = (2. t|r . t) is described by a linear map of the type ψ(r.9) Employing a continuum of intermediate times t ∈ [t0 . t ) ψ(r .

14) has the same dimension as the action integral S[r(t)]. interferences should be much less dramatic than in case of the macroscopic particle. Situations which are well described classically will be distinguished through the property that the classical path gives the dominant. 2.0545 · 10−27 erg s . such that destructive interference of the contributions of such paths does not occur. for microscopic particles like the electron this is by no means the case. actually often essentially exclusive. the exponent is a very large imaginary number. The constant given in (2. However.13). Scl / ≈ 8. 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. . in complete distinction from Classical Mechanics. In case of a proton with mass m = 1. One would still expect significant descructive interference between contributions of different paths since the value calculated is comparable to 2π.e. However.2: Axioms 9.14) ˙ In (2. t) is the Lagrangian of the classical particle. Since this number is multiplied by ‘i’. The situation is very different for microscopic particles.. i. The functional Φ[r(t)] in (2.11) is Φ[r(t)] = exp where S[r(t)] is the classical action integral tN 13 i S[r(t)] (2. r. r. Its value is extremely small in comparision with typical values for action integrals of macroscopic particles. contribution to the path integral (2.15) The exponent of (2. namely due to the property of the classical path to be an extremal of the action integral. 2 2 (2. 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.5 · 1027 .12) S[r(t)] = t0 ˙ dt L(r.e.12.2. a very large number.13) L(r.12.13) are built on action integrals for all possible paths. This implies that small variations of the path near the classical path alter the value of the action integral by very little. i. for the electron many paths contribute and the action integrals for non-classical paths need to be known. However. expressions (2. The value of S[r(t)] is then Scl = 1 1 m v 2 t = erg s . it is comparable to action integrals as they arise for microscopic particles under typical circumstances. 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. not only for the classical path. accordingly. (2..12) is then Scl / ≈ 0. However. 2.6725 · 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. Only for paths close to the classical path is such interference ruled out. t) (2.13) and = 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 . t) = ˙ 1 mx2 − U (x) . respectively. x2 .14 Quantum Mechanical Path Integral 2.4 Propagator for a Free Particle As a first example we will evaluate the path integral for a free particle following the algorithm introduced above. This allows us to write the evolution operator using (2.16) In order to define the path integral we assume. as in (2. 2. at the same time. x. tN )} . tN |x0 .12. provides an avenue to evaluate path integrals. > t1 > t0 letting N go to infinity later. t0 ). tj+1 ) we assume that kinetic energy and potential energy are constant.13) and.3 How to Evaluate the Path Integral In this section we will provide an explicit algorithm which defines the path integral (2.10) and (2. (xN −1 . t1 ).21) N 2. For the sake of simplicity we will consider the case of particles moving in one dimension labelled by the position coordinate x. . ∞[. (2. . etc. They can be anywhere in the allowed volume which we will choose to be the interval ] − ∞. (xN . ˙ 2 (2. the xj values are not. namely tj+1 − tj = (tN − t0 )/N = N . (2. The particles have associated with them a Lagrangian L(x. CN is a constant which depends on N (actually also on other constant in the exponent) which needs to be chosen to ascertain that the limit in (2. . t0 ) = limN →∞ CN exp i N N −1 j=0 +∞ +∞ +∞ −∞ dx1 −∞ dx2 . tN −1 ).12) φ(xN . In passing from one space–time instance (xj .20) can be properly taken. however. Its value is CN = m 2πi N 2 (2.19) The main idea is that one can replace the path integral now by a multiple integral over x1 . −∞ dxN −1 (xj+1 −xj )2 1 − U (xj ) .18) The time instances are fixed. 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 . .20) Here. 2 2m N (2. . The spacings between the times tj+1 and tj will all be identical. namely 1 m(xj+1 − xj )2 / 2 and U (xj ). (x1 .9). a series of times tN > tN −1 > tN −2 > . tj ) to the next (xj+1 . (2. .

e. 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 (2. The resulting expression for S[x(t)|x(t0 ) = x0 . tN |x0 .23). tN |0. 2 2 tN − t0 (2. of (2. t0 ) = exp 1 im (xN − x0 )2 2 tN − t0 φ(0. the origin of which coincides with the classical path of the particle. which can also be written φ(xN . x(tN ) = xN ] is S[x(t)|x(t0 ) = x0 . since x(t0 ) = xcl (t0 ) = x0 and x(tN ) = xcl (tN ) = 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. The ˙ 1 S[x(t)|x(t ) = x .h. t0 ) and (xN . due to (2.s. (2.24).22) where xcl (t) is the classical path which connects the space–time points (x0 . x(tN ) = 0] . t0 ) (2. ˙ 2 t0 (2. x(tN ) = 0] . tN dt t0 1 1 (xN − x0 )2 m x2 = m ˙ cl .25) is.30) d[x(t)] exp 2 tN − t0 x(t0 )=0 a result. tN ) is to be evaluated.h.. x(tN ) = xN ] 1 (xN − x0 )2 m + 0 + 2 tN − t 0 + S[x(t)|x(t0 ) = 0.12) yields x(tN )=0 im (xN − x0 )2 i φ(xN . tN |x0 . can be expressed through a path integral with endpoints x(t0 ) = 0. tN dt y = y(tN ) − y(t0 ) = 0 . t0 ) and (xN .10.27) The third term can be written in the notation introduced tN 1 dt my 2 = S[x(t)|x(t0 ) = 0.24) Also we use the fact that the velocity of the classical path xcl = (xN −x0 )/(tn −t0 ) is constant. = (2.2.13).24) implies for the second term on the r. namely.29) This expression corresponds to the action integral in (2. ˙ ˙ ˙ 2 + y2) = ˙ (2.28) i.26) The first term on the r.23) tN − t0 It is essential for the following to note that. tN ). Inserting the result into (2. t0 ) = exp S[x(t)] (2.31) We have denoted explicitly that the action integral for a path connecting the space–time points (x0 . it holds y(t0 ) = y(tN ) = 0 .25) + The condition (2.s. x(tN ) = 0.4: Propagator for a Free Particle 15 Rather then using the integration variables xj . xN − x0 xcl (t) = x0 + ( t − t0 ) . (2. 2. it is more suitable to define new integration variables yj . To see the benifit of such approach we define a path y(t) as follows x(t) = xcl (t) + y(t) (2. One obtains S[x(t)|x(t0 ) = x0 . using (2.

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

.39) yields φ(0. . . which follows from (2. let say n.44) = tN − t0 . (2. . . . 2 −1  0 0 0 −1 2 For this purpose we expand (2. variable. . We seek then to evaluate the determinant of the n × n matrix   2 −1 0 . tN |0. t0 ) = m 2πi (t − t0 ) 1 2 exp im (x − x0 )2 2 t − t0 .    0 0 0 . Inserting this into (2. t|x0 .. t0 ) = m 2πi (tN − t0 ) 1 2 . .45) Expressions for Free Particle Propagator We have now collected all pieces for the final expression of the propagator (2.   . 0 0  −1 2 −1 .43) We like to note here for further use below that one might as well employ the ‘artificial’ starting values D0 = 1. presently N − 1.39) In order to determine det(Ajk ) we consider the dimension n of (Ajk ). .41) D2 = 2 −1 −1 2 = 3 (2. . To solve this three term recursion relationship one needs two starting values.. .. ... tN |0. . (2. Our derivation has provided us with the value det(Ajk ) = N .2. . D3 . 0 0    Dn =  ..4: Propagator for a Free Particle a property which follows from det(cB) = cn detB for any n × n matrix B.18) we obtain φ(0. 2.46) . Using D1 = |(2)| = 2 . . (2. (2. . .. tN |0. . one can readily verify Dn = n + 1 .40) . 0 0     0 −1 2 . t0 ) = limN →∞ m 2πi N 2 17 N 2πi m N N −1 2 1 . x = xN φ(x. t0 ) = limN →∞ and with NN (2.41) the same result for D2 .42) m 2πi NN 1 2 (2. . .31) and obtain. n = 1. det(Ajk ) (2.. .40) in terms of subdeterminants along the last column. D1 = 2 and obtain from (2. we obtain φ(0. One can readily verify that this procedure leads to the following recursion equation for the determinants Dn = 2 Dn−1 − Dn−2 . defining t = tN ..  . .

x) = im 2 t x2 1 + i o t mδ 2 − 2xo x − po t m + f (x) (2. (2. according to (2.48) Obviously. We want to apply axiom (2. t0 )| = 2 1 πδ 2 1 2 exp − x2 0 δ2 (2. x) + E(x) (2. Below we will apply this to a particle at rest and a particle forming a so-called wave packet. t) of a free particle. 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 . x) + E(x) in (2. centered around x0 = 0. and describes a single particle since 1 πδ 2 1 2 +∞ dx0 exp −∞ − x2 0 δ2 = 1.46) can be generalized to three dimensions in a rather obvious way.51) as follows Eo (xo . t0 ) = 1 πδ 2 1 4 exp − x2 po 0 +i x 2 2δ (2.247).50) One refers to such states as wave packets. (2. The result (2.5) to (2.48) as the initial state using the propagator (2.46). the wave function of the particle at later times.5) allows us to predict the time evolution of any state function ψ(x. We devide the contributions to the exponent Eo (xo .51) For this evaluation we adopt the strategy of combining in the exponential the terms quadratic (∼ x2 ) and linear (∼ x0 ) in the integration variable to a complete square 0 ax2 + 2bx0 = a 0 x0 + b a 2 − b2 a (2. t|r0 . We will obtain. thereby. the associated probability distribution |ψ(x0 .18 Quantum Mechanical Path Integral This propagator. We need to evaluate for this purpose the integral ψ(x. t0 ) = m 2πi (t − t0 ) 3 2 exp im (r − r0 )2 2 t − t0 .49) is Gaussian of width δ. One obtains then for the propagator (2.10) φ(r.52) and applying (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 .53) .

54) f (x) =  This yields Eo (xo . making certain.51) ψ(x. 2 t One chooses then f (x) to complete. The beginning and the end of such path are the points z1 = −∞ × i 1−i t mδ 2 .60) is then represented by real ρ values. (2.61) .53) E(x) =  x− po mt 19 (2.52). 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 . z2 = +∞ × i 1−i t mδ 2 (2. (2. according to (2. however.57) and needs to determine the integral +∞ I = −∞ +∞ dx0 eEo (xo .58) The integrand is an analytical function everywhere in the complex plane and we can alter the integration path.55) im  xo 2 t  1+i t − mδ 2 x− 1+ po m t t i mδ2  .2. (2.x) (2.4: Propagator for a Free Particle im x2 − f (x) . x) = One can write then (2. 2 (2.59) An integration path in the complex x0 –plane along the direction i 1−i t mδ 2 (2. the square in (2. that the new path does not lead to additional contributions to the integral.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  . We proceed as follows.56) m 2πi t 1 2 +∞ eE(x) −∞ dx0 eEo (xo . t) = 1 πδ 2 1 4 2 t 1 + i mδ2  .

54) using (2. We can conclude then that (2. using a ab = a − .58) has the end points z1 = −∞ .70) . 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. 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) ] .57) reads then 1 πδ 2 1 4 2πi t . (2.68) and. t m(1 + i mδ2 ) (2.63) holds and.64) ψ(x. accordingly.69) t)2 t 2δ 2 (1 + i mδ2 ) + i po x − i p2 o t 2m (2.65) t 1 − i mδ2 t 1 + i mδ2 . 1+b 1+b finally E(x) = − (x − po m (2. t) = Seperating the phase factor 1 t 1 + i mδ2 1 4 1 2 exp [ E(x) ] . Since the exponent in (2. Quantum Mechanical Path Integral (2.63) which can be readily evaluated.58) vanishes 0 for Re x0 → ±∞. I = Equation (2. the paths between z1 → z1 and z2 → z2 have a real part of x0 of ±∞.20 whereas the original path in (2. (2.66) yields ψ(x.55).58) by t mδ 2 +∞ I = i 1−i dρ exp − −∞ m 2 t 1+ t mδ 2 2 ρ2 (2. In fact. Hence. one can show that z1 lies at −∞ − i × ∞ and z2 at +∞ + i × ∞.58) has a leading contribution in x0 of −x2 /δ 2 the integrand of (2. z2 = +∞ . (2.67) We need to determine finally (2.58) along the path z1 → z1 and along the path z2 → z2 gives only vanishing contributions one can replace (2.62) If one can show that an integration of (2.

(2. the case of arbitrary to is recovered iby replacing t → to in (2.72) Comparision of Moving Wave Packet with Classical Motion It is revealing to compare the probability distributions (2. We had assumed above [c.49). the spatial dependence of the initial state (2. respectively. However.f. This ‘spreading’ of the wave function is a genuine quantum phenomenon.48) and for the final state (2.4: Propagator for a Free Particle which inserted in (2. This yields. 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 . (2. In this case holds ψ(x.72).74) which corresponds to (2.71. coinciding at t = 0 with the width of the initial distribution (2.71).2.72) for the initial state (2. t0 ) = exp i po i p2 o xo − to m 2m ..48) and that the respective phase factor is related with the velocity of the classical particle and of the center of the distribution (2. (2.76) From this we conclude that an initial wave function ψ(xo .e.72) 2 t2 δ 1+ 2 4 (2. (2.73) m δ increases with time. 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 .77) .72).75) ψ(x. (2.67) provides the complete expression of the wave function at all times t ψ(x. t) = exp i po i p2 o x − t m 2m .71) conserves the phase factor exp[i(po / )x] of the original wave function (2. The width of the distribution (2.74) remains invariant in time. t) = t 1 − i mδ2 t 1 + i mδ2 1 4 21 1 πδ 2 (1 + 2 t2 ) m2 δ 4 1 4 × po i p2 t o ) + i x − t mδ 2 2m (2.48) for δ → ∞. t0 ) = exp i po xo m .71) × exp − (x − po m t)2 ) 2δ 2 (1 + 2 t2 m2 δ 4 (1 − i .49). instead of (2.48)] to = 0.72) moves in the direction of the positive x-axis with velocity vo = po /m which identifies po as the momentum of the particle.75) i. The corresponding probability distribution is |ψ(x. The center of the distribution (2. t) = exp i i p2 po o x − (t − to ) m 2m . The conservation of this factor is particularly striking for the (unnormalized) initial wave function ψ(x0 . Another interesting observation is that the wave function (2. (2. (2.

t) ˙ (2. 2 . For this purpose we use d xcl y = ˙ ˙ (xcl y) − xcl y ˙ ¨ (2.78) i.80) For this purpose we need to determine the action integral S[x(t)] = t0 dt L(x. ˙ ˙ = (2.5 Propagator for a Quadratic Lagrangian x(tN )=xN We will now determine the propagator (2. One refers to the respective states as stationary states.79) for a quadratic Lagrangian L(x. ˙ 2 2 tN (2. y(t)] which vanishes according to the Hamiltonian principle as discussed in Sect. y(tN ) = 0 . t0 ) = x(t0 )=x0 d[x(t)] exp i S[x(t)] (2.85) We want to show now that the contribution of δL to the action integral (2.81) for an arbitrary path x(t) with end points x(t0 ) = x0 and x(tN ) = xN . x. t) + δL ˙ ˙ ˙ ˙ where L(xcl . t) = L(xcl .13) φ(xN . xcl + y(t).10.83) (2. y(t).81) vanishes2 . t) + L (y. the end points of y(t) are y(t0 ) = 0 Inserting (2.82) 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.86) dt The reader may want to verify that the contribution of δL to the action integral is actually equal to the differential δS[xcl . xcl .84) . t) = exp i Quantum Mechanical Path Integral po i p2 o x − t m 2m . 2. 2. tN |x0 .22 becomes at t > to ψ(x. 1. 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 .12. t) ˙ L (y. In order to simplify this task we define again a new path y(t) x(t) = xcl (t) + y(t) (2. Obviously. xcl .82) one obtains L(xcl + y.e. y(t). (2. the spatial as well as the temporal dependence of the wave function remains invariant in this case. 2. t) = ˙ 1 1 mx2 − c(t)x2 − e(t)x .80) into (2.. (2. x.

80) yields for the classical path m¨cl + c(t) xcl + e(t) = 0 x (2.h.87) According to (2. tN |0. . t0 ) = exp S[xcl (t)] φ(0. .24) to the Lagrangian (2. 2 −1  0 0 0 −1 2   c1 0 0 . x t0 (2.88) and. One can express the exponent E in (2. 0 0  −1 2 −1 . . .. . . .  0 0 0 . of (2. y. vanishes.  . t0 ) = y(tN )=0 d[y(t)] exp y(t0 )=0 i tN dt L (y.91) (yj+1 − yj )2 1 2 2m N where cj = c(tj ).  2 N  .. . tj = t0 + form j. . tN |0. . .. 0 0     0 −1 2 . . .. hence.    0 0 0 . .32). 0 0     0 c3 .s. 2 . t0 ) = limN →∞ × +∞ +∞ −∞ dy1 · · · −∞ dyN −1 exp N i N N −1 j=0 m 2πi N 2 N × − 1 2 cj 2 yj (2. tN |0.   . For the evaluation of φ(0. 0 0  0 c2 0 .88) vanishes.93) . One can then express the propagator (2.2. . 0 0   N  0 −  . The discretization scheme adopted above yields in the present case ˜ φ(0. . .  . t) ˙ t0 . (2. . .   . . tN |0.s.83) the first term on the r. .. . also the second contribution on the r. . .k=1 yj ajk yk (2. cN −2 0  0 0 0 0 cN −1 (2. .90) Evaluation of the Necessary Path Integral We have achieved for the quadratic Lagrangian a separation in terms of a classical action integral and a propagator connecting the end points y(t0 ) = 0 and y(tN ) = 0 which is analogue to the ˜ result (2. t0 ) we will adopt a strategy which is similar to that used for the evaluation of (2. .89) where ˜ φ(0. tN |x0 . t0 ) (2.79) i ˜ φ(xN .92) where ajk are the elements of the following (N − 1) × (N − 1) matrix   2 −1 0 . . . Applying the Euler–Lagrange conditions (1. . .h.. .91) through the quadratic N −1 E = i j. .31) for the free particle propagator.5: Propagator for a Quadratic Lagrangian and obtain tN t0 23 dt δL = m [xcl y]|t0 − ˙ t N tN dt [ m¨cl (t) + c(t) xcl (t) + e(t) ] y(t) . . . 0 0  m   ajk =  .. .. .

24

Quantum Mechanical Path Integral

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

N

(iπ)N −1 det(a) 1

1 2

= limN →∞

1
2

N

2

N

N −1

m

det(a)

 

.

(2.94)

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

2
N

N

N −1

m

det(a)

.

(2.95)

def

2

N

N −1

m

det(a) 0 0 0 . . .
m cN −2
2 N

(2.96) 0 

−1 −1 2 − . . . 0 0
2 N

0 ... −1 . . . ... . .. . . .
m c3

−1 2 − 0 . . . 0 0

m c2

0 ... 2 − 0

−1 2 −

m cN −1

2 N

 0    0   .  .  .  −1  

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

m

cn

Dn−1 − Dn−2

(2.97)

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

m

c1

(2.98)

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

= −

cn+1 Dn . m

(2.99)

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

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

25

= =

N N

D0 = 0 ; D1 − D0
N

= 2−

2 N

m

c1 − 1 = 1 .

(2.101)

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

exp

i

S[xcl (t)]

(2.102)

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

2.6

Wave Packet Moving in Homogeneous Force Field

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

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

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

exp

i

S[xcl (τ )] .

(2.106)

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

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

Quantum Mechanical Path Integral

(2.108)

x − xo 1 f − (t − to ) t − to 2m

τ +

1 f 2 τ 2m

(2.109)

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

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

=

+ 2f

x − xo 1 f − (t − to ) t − to 2m

τ (2.111)

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

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

=

1 m 2 +f +

(t − to ) (t − to )2

1 f2 (t − to )3 + xo f (t − to ) 3 m

=

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

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

×

(2.113)

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

2.5: Propagator for a Quadratic Lagrangian

27

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

(2.114)

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

E(x) =

f t2 x − f (x) m

1 f 2 t3 . 24 m

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

(2.117)

po m

f t2 2m

2

 .

(2.118)

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

1 πδ 2 (1 +

exp [ E(x) ]

(2.119)

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

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

(2.120)

1+i t mδ 2 −

− x−

x−

x− x

f t2 po t− m 2m

po f t2 t− m 2m
2

f t2 + 2x m

po f t2 t+ m 2m

f t2 po t+ m 2m

1+i

t mδ 2 (2.121)

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

Quantum Mechanical Path Integral

E(x)

=

− +

t−

2δ 2 1 + i

f t2 2m t mδ 2

2

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

(po + f t) x −

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

dτ (po + f τ )2 .
0

(2.123)

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

ψ(x, t)

=

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

1 4

× 

 × exp  − × exp i

t−

f t2 2m

2

1+

2 t2 m2 δ 4

1−i
t

t mδ 2

  × . (2.124)

(po + f t) x −

i
0

(po + f τ )2 2m

The corresponding probablity distribution is |ψ(x, t)|2 = 1 πδ 2 (1 +
2 t2 m2 δ 4 1 2

)

x− t−  exp  − 2 2 δ 2 1 + m2tδ4

po m

f t2 2m

2

  .

(2.125)

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

2.5: Propagator for a Quadratic Lagrangian

29

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

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

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

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

(2.129) (2.130)

(2.131)

A =

x − x0 cosω(t − to ) , sinω(t − t0 )

(2.132)

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

(2.133)

S[xcl (τ )] =
t0

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

(2.135)

30

Quantum Mechanical Path Integral

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

(2.137)

(2.138)

Using 1 1 + cos2ωτ 2 2 1 1 2 sin ωτ = − cos2ωτ 2 2 1 cosωτ sinωτ = sin2ωτ 2 the Lagrangian, considered as a function of τ , reads cos2 ωτ g(τ ) = 1 1 mx2 (τ ) − mω 2 x2 (τ ) ˙ cl cl 2 2 = 1 (x − x0 c)2 mω 2 cos2ωτ 2 s2 x − x0 c −mω 2 xo sin2ωτ s 1 − mω 2 x2 cos2ωτ o 2
t 0 dτ g(τ )

(2.139) (2.140) (2.141)

(2.142)

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

requires the integrals (2.143) (2.144)

dτ cos2ωτ
0 t

= =

dτ sin2ωτ
0

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

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

(2.145)

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

31

(2.146)

For the propagator of the harmonic oscillator holds then φ(x, t|x0 , t0 ) = × exp
imω 2 sinω(t − t0 ) mω 2πi sinω(t − t0 )
1 2

× . (2.147)

(x2 + x2 ) cosω(t − t0 ) − 2x0 x 0

Quantum Pendulum or Coherent States As a demonstration of the application of the propagator (2.147) we use it to describe the time development of the wave function for a particle in an initial state ψ(x0 , t0 ) = mω π
1 4

exp

mω(x0 − bo )2 i + po xo 2

.

(2.148)

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

(2.149)

=

energy

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

ψ(x, t) =
−∞

dx0 φ(x, t|x0 , t0 ) ψ(x0 , t0 ) .

(2.150)

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

(2.147, 2.150, 2.151) can be written ψ(x, t) = mω π
1 4

m 2πiω sinω(t − t0 )

1 2

dx0 exp [ E0 + E ]
−∞

(2.152)

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

Quantum Mechanical Path Integral

x2 c − 2xo x + isx2 − 2isxo bo + o o

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

E(x)

=

x2 c + isb2 − f (x) o

.

c = cos ω(t − to ) ,

s = sin ω(t − to ) .

Here f (x) is a function which is introduced to complete the square in (2.153) for simplification of the Gaussian integral in x0 . Since E(x) is independent of xo (2.152) becomes ψ(x, t) = mω π
1 4

m 2πiω sinω(t − t0 )

1 2

eE(x)

dx0 exp [ E0 (xo , x)]
−∞

(2.156)

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

(2.157)

For f (x) to complete the square we choose f (x) = (x + isbo − One obtains for (2.157) E0 (xo , x) = imω po exp [iω(t − t0 )] x0 − (x + isbo − s) exp (−iω(t − t0 ) 2 s mω
1 2

(2.158)

.

(2.159)

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

dx0 e
−∞

E0 (x0 )

=

2πi sinω(t − t0 ) mω exp[iω(t − t0 )]

(2.160)

Inserting this into (2.156) yields mω 1 E(x) 4 e π For E(x) as defined in (2.154) one obtains, using exp[±iω(t − to )] = c ± is, ψ(x, t) = E(x) =
imω 2 s

(2.161)

x2 c + isb2 − x2 c + isx2 − 2isxbo c − 2s2 xbo o
po po + s2 b2 c − is3 b2 + 2 mω xsc + 2i mω bo s2 c o o po po − 2i mω xs2 + 2 mω bo s3 − p2 o m2 ω 2

s2 c + i mpo 2 s3 2ω

2

=

− mω 2

x2 + c2 b2 − 2xbo c + 2ixsbo − ib2 sc o o
p2 o m2 ω 2 po s2 − 2i mω xc
2

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

=

− mω (x − cbo − 2 − i
2

po mω

s)2 +

i

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

po ( 2mω − 1 mωb2 )sc + o 2

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

33


to

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

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

− to ) −

1 2

p2 mωb2 o o − 2mω 2

sc +

1 bo po s2 2

(2.164)

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

˙ dτ L[b(τ ), b(τ )] =

p2 mωb2 o o − 2mω 2

sc − bo po s

(2.165)

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

˙ dτ L[b(τ ), b(τ )]

(2.166)

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

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

o

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

Comparision of Moving Wave Packet with Classical Motion The probability distribution associated with (2.167) |ψ(x, t)|2 = mω π
1 2

exp −

[x − b(t)]2

(2.168)

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

34

Quantum Mechanical Path Integral

In case of the free particle wave packet (2.48, 2.71) the factor exp(ipo x) gives rise to the translational motion of the wave packet described by po t/m, i.e., po also corresponds to initial classical momentum. In case of a homogeneous force field the phase factor exp(ipo x) for the initial state (2.48) gives rise to a motion of the center of the propagating wave packet [c.f. (2.125)] described by 1 (po /m)t + 2 f t2 such that again po corresponds to the classical momentum. Similarly, one observes for all three cases (free particle, linear and quadratic potential) a phase factor exp[ip(t)x/ ] for the propagating wave packet where fp(t) corresponds to the initial classical momentum at time t. One can, hence, summarize that for the three cases studied (free particle, linear and quadratic potential) propagating wave packets show remarkably close analogies to classical motion. We like to consider finally the propagation of an initial state as in (2.148), but with bo = 0 and po = 0. Such state is given by the wave function ψ(x0 , t0 ) = mω π
1 4

exp

mωx2 iω 0 − to 2 2

.

(2.169)

where we added a phase factor exp(−iωto /2). According to (2.167) the state (2.169) reproduces itself at later times t and the probablity distribution remains at all times equal to mω π
1 2

exp

mωx2 0

,

(2.170)

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

2.7

Stationary States of the Harmonic Oscillator

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

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

W (x, t) =
−∞

dxo φ(x, t|xo , to ) W (xo , to ) .

(2.172)

We will demonstrate further below that (2.172) provides us in a nutshell with all the stationary states of the harmonic oscillator, i.e., with all the states with time-independent probability distribution. In order to prove (2.172) we express the propagator, using (2.147) and the notation T = t − to
2 mω φ(x, t|xo , to ) = e × π (1 − e−2iωT ) 1 + e−2iωT mω 2x xo e−iωT mω 2 × exp − (xo + x2 ) − 2 1 − e−2iωT ω 1 − e−2iωT 1

− 1 iωT 2

(2.173)

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

35

mω π (1 − e−2iωT )

1 2

+∞

dxo exp[ Eo (xo , x) + E(x) ]
−∞

(2.174)

(2.175)

+ 2xo E(x) = − mω 2 x2

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

(2.176)

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

2xe−iωT + 2 1 − e−2iωT

e

−iωto

.

(2.177)

=

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

e

−iωto

(2.178)

for some constant zo ∈ C. Using (2.247) one obtains dxo e
−∞ Eo (xo ,x)

=

π (1 − e−2iωT mω

1 2

(2.179)

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

(2.180)

=

x2 − 4

mω mω

x e−iωt +

2 e−2iωt mω (2.181)

=

mω 2 x + 2 2

x e−iωt − e−2iωt

172). In contrast to W (x.171. We want to inspect the consequences of the invariance property (2. results in 1 ˜ exp[−iω(n + 1 ) t] φn (x) 2 n! +∞ = −∞ dxo Φ(x.187) or 1 ˜ exp[−iω(n + 2 ) t] φn (x) +∞ = −∞ ˜ dxo φ(x. t) = exp[−iω(n + 1 ) t] φn (x) as invariants under the 2 action of the propagator φ(x.e. .184) 1 ˜ exp[−iω(n + 1 ) t] φn (x) 2 n! ∞ +∞ = m=0 −∞ dxo Φ(x. one obtains for the r. .188) identifies the functions ψn (x. t|xo .172) is a function of t − to and defining accordingly Φ(x. 2. 2 (2. Noting that the propagator (2. xo . xo . to ) we express (2. xo . n = 1. ..185) Replacing t → t + to yields ∞ n=0 1 ˜ exp[−iω(n + 1 ) (t + to )] φn (x) 2 n! ∞ +∞ = m=0 −∞ dxo Φ(x. t − to ) = φ(x. .171) concludes the proof of (2.147) in (2. which also exhibits such invariance. 1 exp[iω(n + 2 ) to ] · · · . xo .182) Comparision with (2.171) can be expanded in terms of e−inωt . t − to ) 1 ˜ exp[−iω(m + 1 ) to ] φm (xo ) 2 m! (2. t − to ) 1 ˜ φn (xo ) n! (2. but not of t.183) where the expansion coefficients are functions of x. to ).171) ∞ W (x. t|xo . t|xo .36 Altogether. of (2. 2. t). We note that the factor exp(2 mω/ xe−iωt − e−2iωt ) in (2. t) +∞ −∞ dto 1 ˜ exp[−iω(m + 1 ) to ] φm (xo ) 2 m! (2. (2.172). to ) exp[−iω(n + 1 ) to ] φn (xo ) .188) ˜ ˜ Equation (2.h.172) I = exp 2 mω x e−iωt − e−2iωt − Quantum Mechanical Path Integral mω 2 1 x − iωt 2 2 .s.186) Fourier transform. t) = n=0 1 ˜ exp[ − iω(n + 1 ) t ] φn (x) 2 n! (2..172) in the form ∞ n=0 (2. i. Accoordingly. one can expand (2.

˜ |φ stationary wave functions ψn (x.183) reads then w(y. t) provide all stationary states of the quantum mechanical harmonic oscillator.191) (2. t) = exp[−iω(n + 1 ) t] Nn φn (x) . z) (2.196) where Hn (y) = ey 2 /2 (2. t) of the quantum mechanical harmonic oscillator ˜ ψn (x. t) are associated with a time-independent probablity density |ψn (x.192) w(y. For this purpose we introduce first closed expressions for φ the new variables y z and write (2.197) are called Hermite polynomials which are polynomials of degree n which will be evaluated below. In the following we will characterize the functions φn (x) and determine the normalization constants Nn . z) = Hn (y) (2.183). t)|2 = ˜n (x)|2 . through the expansion coefficients φn (x) in (2.194) = z2 n=0 ∞ 1 ∞ zn ˜ φn (y) n! (2.194) plays a central role for the Hermite polynomials since it contains.197) The expansion coefficients Hn (y) in (2. z) z 2 e−y or w(y. z) = exp(2yz − z 2 ) .171).195) zn Hn (y) n! ˜ φn (y) . through expansion (2.183).189) ˜ dx φ2 (x) = 1 n (2. t) such that +∞ −∞ 2 dx |ψ(x. 2. .194). Expression (2. 2 Here Nn are constants which normalize ψn (x. (2.2. Expansion (2. t) = z 2 e−y where w(y. in a ‘nutshell’ all information on the Hermite polynomials.171) W (x. Actually. (2. we have identified then. . . according to (2. characterizes the wave functions φn (x).198) ∂z n z=0 . We will also argue that the functions ψn (x. 1.193) (2. z) = n=0 1 2 /2 1 2 /2 = = mω e−iωt x (2.190) ˜ is obeyed.183). The Hermite Polynomials ˜ The function (2. This follows from ∂n w(y. t)|2 = Nn +∞ −∞ n = 0.5: Propagator for a Quadratic Lagrangian 37 ˜ ˜ the functions ψn (x. To obtain ˜n (x) we simplify the expansion (2.

among other properties.196). (2. Reading. D. Boston. .S.200) yields for the first Hermite polynomials H0 (y) = 1. We will employ (2.1: Schematic representation of change of summation variables ν and µ to n = ν + µ and m = ν −µ. i. The identity (2. ν     H2 (y) = 4y 2 − 2..Patashnik (AddisonWesley.L. ν+µ ν   H3 (y) = 8y 3 − 12y. .196) to ˜ derive. and O. One calls w(y. The diagrams also identify the areas over which the summation is to be carried out. ∂y n (2. closed expressions for Hn (y). The diagrams illustrate that a summation over all points in a ν. We want to derive now explicit expressions for the Hermite polynomials. 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 (2.194. and recursion equations for the efficient evaluation of Hn (y). 1990) is a useful introduction to this tool as is a chapter in the eminently useful Concrete Mathematics by R.198) for the Hermite polynomials can be expressed in a more convenient form employing definition (2. z) the generating function for the Hermite polynomials. We start from 2 e2yz − z ∞ ν = ν=0 µ=0 ∞ ν 1 ν! 1 ν ν µ ν µ z 2µ (−1)µ (2y)ν−µ z ν−µ (−1)µ (2y)ν−µ z ν+µ = ν=0 µ=0 3 (2.196) ∂n w(y.201) ν−µ     µ µ ν−µ Figure 2.e.202) generatingfunctionology by H. 2.194) 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). . normalization factors for φ(y). For this purpose we expand the generating function (2.E. that Hn (y) is a polynomial of degree n. µ lattice (left diagram) corresponds to a summation over only every other point in an n.38 Quantum Mechanical Path Integral which is a direct consequence of (2. 1989).Knuth. m lattice (right diagram).200) One can deduce from this expression the polynomial character of Hn (y).Wilf (Academic Press.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 . Inc. Massachusetts.Graham. ν+µ   (2.196) results in the so-called Rodrigues formula for the Hermite polynomials Hn (y) = (−1)n ey 2 ∂ n −y2 e . H1 (y) = 2y.

µ expressend in terms of n.5: Propagator for a Quadratic Lagrangian and introduce now new summation variables n = ν + µ. it contains only odd powers. (2. .208) which agrees with the expressions in (2. From (2. m must be restricted such that either both n and m are even or both n and m are odd. in fact. Hence. k! (n − 2k)! = Hn (y) (2. 39 (2. otherwise. is a polynomial of degree n. m = ν − µ 0 ≤ n < ∞.207) This expression yields for the first four Hermite polynomials H0 (y) = 1. k! (n − 2k)! (2. 2.205) Since (n − m)/2 is an integer we can introduce now the summation variable k = (n − m)/2 . it holds Hn (−y) = (−1)n Hn (y) . (2. according to 2. .207) contains only even powers.2.206) From this expansion we can identify Hn (y) [n/2] Hn (y) = k=0 (−1)k n! (2y)n−2k . According to (2. 0 ≤ k ≤ [n/2] where [x] denotes the largest integer p.. m are ν = n+m . p ≤ x. (2. 0 ≤ m ≤ n . H1 (y) = 2y. −z) and. 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 .209). the sum in (2. 2 (2. 2 µ = n−m .1. .201). H3 (y) = 8y 3 − 12y.194) holds w(−y. hence. . With this restriction in mind one can express (2. H2 (y) = 4y 2 − 2. µ are integers the summation over n.210) from which one can conclude the property (2. z) = w(y.197) ∞ n=0 zn Hn (−y) = n! ∞ n=0 (−z)n Hn (y) = n! ∞ n=0 zn (−1)n Hn (y) n! (2.209) This property follows also from the generating function.202) e 2yz − z 2 ∞ = n=0 zn n! ≤n m≥0 n! (−1) n−m 2 n−m 2 ! m! (2y)m . The lattices representing the summation terms are shown in Fig.203) The old summation variables ν.e. In case of even n . for odd n.207) one can deduce that Hn (y). i.204) Since ν.

as defined through (2. z) − (2y − 2z) w(y. . n = 1. . as given in (2.215) ∂z which can be readily verified using (2. a factor z reduces the order of Hn by one and introduces also a factor n. dw/dy − 2zw = 0. Starting point of the derivation is the property of w(y.194). z) ∂ w(y. . . . . namely. n = 0. . z) should be equivalent to the relationship (??). Another relationship is d Hn (y) = 2n Hn−1 (y). (2. (2.196) into the differential equation (2. . have a node at y = 0.188. Hn+1 (y) − 2y Hn (y) + 2n Hn−1 (y) = 0 .216) Combining the sums and collecting terms with identical powers of z ∞ n=1 zn Hn+1 (y) − 2y Hn (y) + 2n Hn−1 (y) n! + H1 (y) − 2yH0 (y) = 0 (2. 2. . H2n+1 (0) = 0 .197) above and given by (2. Hn+1 (y) − 2y Hn (y) + 2n Hn−1 (y) = 0. H2n (0) = (−1)n Recursion Relationships A useful set of properties for special functions are the so-called recursion relationships. The reader may verify that w(y. 2. . a property which is consistent with (2. dy n = 1. . 1. One can then readily state which differential equation of w(y. n! (2. for example. (2. 2.233) below.194). (2. z) = 0 (2. (2.208). .209) since odd functions have a node at the origin.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.211) m! n! m=0 n=0 Comparing terms on both sides of the equation yields (2n)! . 2. For this purpose one considers w(0. 2. . Substituting expansion (2. z). z) = exp(−z 2 ) and carries out the Taylor expansion on both sides of this expression resulting in ∞ ∞ (−1)m z 2m zn = Hn (0) . For Hermite polynomials holds.214) We want to derive (??) using the generating function.217) gives H1 (y) − 2y H0 (y) = 0.212) n! This implies that stationary states of the harmonic oscillator φ2n+1 (x). n = 1.40 Quantum Mechanical Path Integral The generating function allows one to determine the values of Hn (y) at y = 0.215) 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 .213) which allow one to evaluate Hn (y) from H0 (y) and H1 (y) given by (2. indeed satisfies the latter relationship.

194.226) π −∞ 2n z n z n n=0 n! Expressing the l. .247) for a = i one obtains I(y) = and. 2 √ π e2 z z (2.196) yields ∞ n. 2.228) . z ) e−y 2 +∞ = e2 z z = √ ∞ dy e−(y−z−z ) = .225) We want to derive from the generating function (2.221) 1 = √ π +∞ −∞ dt e2iyt − t .s. 2. 2. finally.200) yields Hn (y) = The identity (−1)n y2 √ e π +∞ dt −∞ dn 2iyt − t2 e .220) Using (2. 2 n = 0.h.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 . (2. (2.2.227) Comparing the terms of the expansions allows one to conclude the orthonormality conditions +∞ −∞ dy Hn (y) Hn (y) e−y 2 √ = 2n n! π δn n .222) Employing this expression now on the r.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. through a double series over Hermite polynomials using (2. z) w(y.226a).219) which can be written I(y) = e−y 2 +∞ −∞ dte−(t−iy) = e−y √ 2 2 +∞ −∞ dze−z . dy n (2.194. 1. .s. acording to the definition (2.h. 2 (2. .196) the orthogonality properties of the Hermite polynomials.224) +∞ −∞ dt tn e2iy t − t . 2 (2.223) dn 2iy t − t2 2 e = (2 i t)n e2iy t − t n dy results. in the integral representation of the Hermite polynomials 2n (−i)n ey √ Hn (y) = π Orthonormality Properties 2 (2. 2 (2. For this purpose we consider the integral +∞ −∞ dy w(y. of the Rodrigues formula (2. e−y 2 π e−y 2 (2.

(2.228) allow us to construct normalized stationary states of the harmonic oscillator. 3.232) Hn (y) .190)] φn (y) = Nn e−y 2 /2 and for the normalization constants Nn follows from (2.233) . One can recognize. 2.230) (2. 1. (2. ¡ ¢ ¡ -4   -2 0 2 4 y (2. Normalized Stationary States The orthonormality conditions (2. 2. According to (2.229) Hn (y) .233) are presented for n = 0. 4 and odd for n = 1. 3. negative for n = 2 and vanishes for n = 1. The normalized states are [c. 2. 2. Furthermore.f.231) √ π (2. One can also recognize that n is equal to the number of nodes of the wave function.197) holds 2 ˜ φn (y) = e−y /2 Hn (y) .228) +∞ 2 Nn −∞ √ 2 2 2 dy e−y Hn (y) = Nn 2n n! π = 1 1 2n n! 1 2n n! We conclude Nn = and can finally state the explicit form of the normalized stationary states φn (y) = −y √ e π 2 /2 The stationary states (2.2. in agreement with our above discussions. 3.2: Stationary states φn (y) of the harmonic oscillator for n = 0. (2. 4 in Fig. 3. 4. 2. in harmony with (??).189. that the wave functions are even for n = 0.42 Quantum Mechanical Path Integral E 9/2 h ω ¥ ¥ ¥ ¥ ¥ 7/2 h ω 5/2 h ω ¤ 3/2 h ω £ 1/2 h ω ¦   Figure 2. the value of the wave function at y = 0 is positive for n = 0. 1. 4.

i.150 . w(x) ≡ 1.156.151. and In = 2n n! π.5: Propagator for a Quadratic Lagrangian 43 The normalization condition (2. using (2. holds Ω = [0.2. f. 5.228) shows that the orthonogality condition of the Hermite polynomials is in complience with (2.e. In case of the associated Laguerre (α) polynomials. 5. ??? and eq..10 and eq.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. ???] and relativistic [see Sect. Ω (2. The orthogonal polynomials pn mentioned above have the important property that they form a complete basis in the space F of normalizable functions. Comparision with (2. (2.e. i.237) for Ω = R. denoted by Ln (x) and encountered in case of the stationary states of the nonrelativistic [see Sect. In = Γ(n+α+1)/n! where Γ(z) is the so-called Gamma function. (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. The latter are written for polynomials pn (x) in the general form dx pn (x) pm (x) w(x) = In δnm Ω (2.189) φn (x) = √ 1 2n n! mω π 1 4 e− mωx2 2 Hn ( mω x) . g ∈ F . Other examples of orthogonal polynomials are the Legendre and Jacobi polynomials which arise in solving three-dimensional stationary Schr¨dinger equations.228).459] hydrogen atom. 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. and I = 2/(2 + 1). denoted by P (x) and introduced in Sect. (2.. +∞[. w(x) = xα e−x .231) of the wave functions differs from that postulated in (2. 2. In case of the Legendre polynomials. 10. w(x) = 0 only at a discrete set of points xk ∈ Ω (2.179] holds Ω = [−1. 1].236) where w(x) is a so-called weight function with the property w(x) ≥ 0.239) . (10. 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. 5 below [c.233) and (2. by dx = dy mω 1 4 .237) and where In denotes some constants. 5. The Hermite polynomials are so-called orthonogal polynomials since they obey the conditions (2.189) by the Jacobian dx/dy. of functions which obey dx f 2 (x) w(x) = < ∞ .f.236 .238) where the space is endowed with the scalar product (f |g) = Ω dx f (x) g(x) w(x) = < ∞ .

244) For the state to be stationary |ψ(x. i.44 As a result holds for any f ∈ F f (x) = n Quantum Mechanical Path Integral cn pn (x) (2.247) .242) Let us assume now the case of a function space governed by the norm (2. .242) and the existence of a normalizable state ψ(y.245) must be time-independent.233).k yj ajk yk = (iπ)n .e.240) where cn = 1 In dx w(x) f (x) pn (x) . Therefore. i. Appendix: Exponential Integral We want to prove +∞ +∞ I = dy1 . . In case of n = 1 this reads +∞ −∞ dx ei a x = 2 iπ .246) for det(a) = 0 and real.e. Since the Hermite polynomials form a complete basis for such states we can expand ∞ ψ(y.. except for a single n = no . 2 (2.233) exhaust all stationary states of the harmonic oscillator. t) which is stationary under the action of the harmonic oscillator propagator (2. a state for which (2.e. the states (2.188. −∞ dyn ei −∞ n j. a (2. t)|2 . ∞ n.. t) = n=0 cn (t) e−y 2 /2 Hn (y) .147). If one replaces for all f ∈ F: f (x) → w(x) f (x) and. The only possibility for this to be true is dn = 0. det(a) (2. (2. Ω (2.241) The latter identity follows from (2..m=0 d∗ dm exp[iω(m − n)t] e−y Hn (y)Hm (y) . i.243) To be consistent with(2. hence. pn (x) → w(x) pn (x) the scalar product (2.236). 2.172) holds.197) it must hold cn (t) = dn exp[−iω(n + 1 )t] and.e. the 2 stationary state ψ(y. symmetric a. in particular. ψ(y. t) = n=0 1 2 dn exp[−iω(n + )t] e−y /2 Hn (y) . (2. aT = a. i.239) becomes the conventional scalar product of quantum mechanics f |g = Ω dx f (x) g(x) . t) must be identical to one of the stationary states (2. t) is ∞ ψ(y. n 2 (2.

255) . .k n m n y Sj ajk Skm ym ˜ ˜ y (S T ) j ajk Skm ym ˜ ˜ j.246) exploits that for any real. (2.249) The akk are the eigenvalues of a and are real.252) ˜ For the determinant of a holds det(˜) = a n ajj ˜ j=1 (2. yk = k Skj yj . .248) . .k yj ajk yk n = j.e.253) as well as det(˜) = det(S−1 aS) = det(S−1 ) det(a) .7: Appendix / Exponential Integral 45 which holds for a ∈ C as long as a = 0. The proof of (2. . ST S = 1 1 or S = S−1 . . (2. i. . .246) reads then in terms of yj ˜ n n j.254) ajj . . (2.m (S T )jl alm Smk .k n n yj = ˜ k (S −1 )jk yk . .251) where. symmetric matrix exists a similarity transformation such that   a11 0 .k n m = = j. . ˜ (2.. 0  ˜   ˜ S−1 a S = a =  . det(S) a = (det(S))−1 det(a) det(S) = det(a) . ˜ (2. One can conclude det(a) = j=1 n (2. This property allows one to simplify the bilinear ˜ form n yj ajk yk by introducing new integration variables j.248.249) n ajk = ˜ l.   . .  . .. ann ˜ where S can be chosen as an orthonormal transformation. according to (2. .k yj ajk yk ˜˜ ˜ (2. 0 ˜  0 a22 .250) The bilinear form in (2. .2. 0 0 . 2.

250) holds J = S and. . . ∂(˜1 . .e.258) where we introduced the Jacobian matrix J = with elements Jjs = According to (2. . . . . . . . . . . n (2. . yn ) y ˜ ∂yj .261) ∂(y1 . 2. yn ) ) = det(S) . hence. (2.250) allows one to express (2. i. . d˜n e y −∞ = (2. .249) follows 1 = det ST S such that one can conclude det S = ±1 One can right then (2.250) +∞ +∞ I = d˜1 . . yn ) y ˜ ei n k akk yk ˜ ˜2 . . ˜ for j = 1. . .260) (2.266) . .. ∂ ys ˜ (2. . det( From (2..259) (2. Accordingly.264) I = d˜1 . .257) Substitution of the integration variables (2. y −∞ d˜n det y −∞ ∂(y1 . . .. yn ) y ˜ = ( det S )2 (2. . yn ) ∂(˜1 . .265) k=1 −∞ which leaves us to determine integrals of the type +∞ dx eicx −∞ 2 (2.46 We have assumed det(a) = 0. . .258) +∞ +∞ ∂(y1 . yn ) ∂(˜1 . . ajj = 0 .256) such that none of the eigenvalues of a vanishes.262) (2. y −∞ +∞ d˜n ei y −∞ n k akk yk ˜ ˜2 +∞ i˜11 y1 a ˜2 n i˜nn yn a ˜2 +∞ a ˜ d˜k ei˜kk yk y 2 = d˜1 e y −∞ . . .263) (2. holds n Quantum Mechanical Path Integral ajj = 0 ˜ j=1 (2.

271) . 2 since eicz is a holomorphic function. according to (2.266) to the well-known Gaussian integral +∞ dx e−cx −∞ 2 = π . Substituting −x for x into integral J4 one obtains 0 J4 = p p (−i) dx eic(−ix+p) 2 = 0 i dx eic(ix−p) 2 = J2 .e. One can relate integral (2. We consider first the case c > 0 and discuss the case c < 0 further below.268) along the path γ = γ1 + γ2 + γ3 + γ4 displayed in Figure 2.268) vanishes.2.. c (2. p ∈ IR.267) by considering the contour integral J = γ dz eicz = 0 2 (2.3.269) 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 √ (2.270) dx e−cx 2 2 − 2p √ γ4 : z = ix − p J4 = −p i dx eic(ix−p) . Jk = γk dz eicz 2 (2. the integrand does not exhibit any singularities anywhere in C.c > 0 . The contour intergral (2.257) holds c = 0. for x. The contour integral (2.268) can be written as the sum of the following path integrals J = J1 + J2 + J3 + J4 .7: Appendix / Exponential Integral 47 where. i. (2.

We will now show that the two integrals J2 and J4 vanish for p → +∞.274) (2. This follows from the following calculation p p→+∞ lim |J2 or 4 | = p→+∞ lim | 0 p i dx eic(ix+p) | ≤ It holds |e ic(p2 −x2 ) p→+∞ 0 lim |i| dx |eic(p 2 −x2 ) | = 1 since the exponent of e is purely imaginary. 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 = 0.273) p→+∞ lim J2 and J4 do not contribute then to integral (2. Hence.3: Contour path γ in the complex plain. (2. c   ¡ -p γ1 p Re(z) 2 | |e−2cxp | .272) = 0 .275) . (2. (2.268) for p = +∞.267) one has shown then ∞ dx eicx −∞ 2 = iπ .48 Quantum Mechanical Path Integral Im(z) ip γ2 ¡ γ4 ¡ γ3 ¡ -i p Figure 2.

255) this result can be expressed in terms of the matrix a I = which concludes our proof. This leads to ∞ icx2 J = dx e −∞ + √ −∞ −i ∞ dx ecx = 0 2 (2. c (2. n ˜ j=1 ajj (2.278) Noting (2.279) .276) and (c < 0) ∞ dx eicx −∞ 2 = −iπ = −|c| iπ .275.268). but with a path γ that is reflected at the real axis.277) We apply the above results (2. It holds n I = k=1 iπ = akk ˜ (iπ)n . 2.7: Appendix / Exponential Integral 49 One can derive the same result for c < 0.2.277) to (2.265). if one chooses the same contour integral as (2. (iπ)n det(a) (2.

50 Quantum Mechanical Path Integral .

The propagator in (3. the Schr¨dinger o o equation provides the simpler avenue towards describing quantum systems than the path ingral formulation of Section 2. x3 )T .20) to evaluate the propagator. t) .h. In the limit of very small it is sufficient to employ a single discretization interval in (2.s.s. t + ) = × exp i 2 +∞ +∞ +∞ −∞ dx1 −∞ dx2 −∞ dx3 m x2 + x2 + x2 1 2 3 m 2πi 3 2 × (3.Chapter 3 The Schr¨dinger Equation o 3.5) ψ(r.20) to R one obtains then for small φ(r. .21).4) − U (r. t + |r0 .1 Derivation of the Schr¨dinger Equation o We will consider now the propagation of a wave function ψ(r. t) satisfies a partial differential equation. t) = From this follows ψ(r. t) . t + |r0 .1) can be expressed through the discretization scheme (2. (3.20. t) . We will denote the components of s by (x1 . For many situations. Generalizing (2. 2.3) In order to carry out the integration we set r0 = r + s and use s as the new integration variable. Notable exceptions are non-stationary systems involving time-dependent linear and quadratic Lagrangians. It holds then according to (2. namely the Schr¨dinger equation.h. t) . t + ) = Ω m 2πi 3 2 exp i m (r − r0 )2 − 2 U (r. t) ψ(r0 .2) d 3 r0 m 2πi 3 2 exp i m (r − r0 )2 − 2 U (r. (3. and the r. (3. x2 . t) by an infinitesimal time step . but by no means all. t) ψ(r0 . x3 51 .1) We will expand the l. t) ψ(r + s. x2 . This yields ψ(r. of this equation in terms of powers of and we will demonstrate that the terms of order require that ψ(r. even in x1 . t + ) = Ω d3 r0 φ(r.

t) j 1 12 ..8) (3. x2 . t) + ∂xj 2 3 xj xk j. the terms collected in (3.7) shows 1 ∂ In (a).4) one identifies 2 1 = a m = O( ) (3. e. but over s = (x1 . n = 0. 1 2 3 2 ∂2 j=1 xj ∂x2 ψ(r.. t). .4) is even in all three coordinates x1 .5) which are even separately in all three coordinates yield non-vanishing contributions. 1. t) + . 2 (3. U (r. ∂xj ∂xk (3. Since the latter contributes to (3. According to (3. t) is a constant with respect to this integration.6) 3 4 ∂4 j=1 xj ∂x4 ψ(r. Since the kinetic energy contribution in (3. x2 .. It is then sufficient to consider the terms ψ(r. The integration involves only the wave function ψ(r +s.52 Schr¨dinger Equation o It is important to note that the integration is not over r.36) holds I0 (a) = Inspection of (3. t) . of the expansion of ψ(r + s. O( 7 2 ) ... (3. i ∂a Starting from (3.k=1 xj xk ∂x2 ∂x2 ψ(r. t) j . x3 )T . 1 4 3 2 2 ∂4 j. t) and the kinetic energy term. Obviously. It holds In+1 (a) = I1 (a) = i 2a iπ . . a I2 (a) = − 3 4a2 iπ .11) holds 3 m 3 iπ 2 2 × = 1 (3.k=1 ∂2 ψ(r.9) iπ a (3.g.11) and. x3 . consequently.5) assuming that only the leading terms contribute. (3. O( 7 2 ).13) 2πi a .8) one can evaluate recursively all integrals In (a).10) It is now important to note that in case of integral (3. O( 5 2 ) . t) + j=1 xj ∂ 1 ψ(r.4) only for small x2 + x2 + x2 values 1 2 3 we expand 3 ψ(r + s. we need then to evaluate integrals of the type +∞ In (a) = −∞ dx x2n exp i a x2 .12) Here one needs to note that we are actually dealing with a three-fold integral. a supposition which will be examined below. t) j k .6) make contributions of the order O( 3 2 ) .7) According to (2. only terms of the expansion (3. t) = ψ(r. a (3..

19) ∂t where 2 2 ˆ H = − + U (r. t) + O( 2 ) .3. t) (3. t) = ∂t 2 − 2 2m + U (r. t) ψ(r. t)ψ(r. The 1st order time derivative requires that for any solution a single temporal condition needs to be specified. t) = H ψ(r.14) This expansion in terms of powers of suggests that we also expand ∂ 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.2 Boundary Conditions The time-dependent Schr¨dinger equation is a partial differential equation.e.2: Boundary Conditions and one can conclude. one specifies the so-called initial condition. Due to its linear character any linear combination of solutions of the time-dependent Schr¨dinger o equation is also a solution. 2nd o order in the spatial variables and linear in the solution ψ(r. t) + O( 2 ) ∂t i U (r. t) + and exp − i U (r. e. (3. t) + O( 2 ) . this equation is trivially satisfied to order O( 0 ).14) results in ψ(r.. t) . This equation is often written in the o form ∂ ˆ i ψ(r. ψ(r. t) = ψ(r. t) ψ(r. t) = 1 − (3. In order O( ) the equation reads i ∂ ψ(r. a solution is thought for t ≥ t0 and the solution is specified at the intial time t0 . t) .17) 2m ψ(r. t) − + i 2 2 i U (r. t) + 1 2i 4 m 2 53 ψ(r. (3. The following general remarks can be made about the solution. t + ) = ψ(r. t) + ∂ ∂t ψ(r. t + ) = exp − i U (r. t). t) + O( 2 ) .20) 2m 3. ψ(r.10). t) (3.18) This is the celebrated time-dependent Schr¨dinger equation. i. (3. using (3. 1st order in time. t1 ) = f (r).g. Usually. 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 . We will derive briefly the type of boundary conditions encountered..15) ψ(r. Obviously. (3.16) Inserting this into (3.

differ from each other. g ∈ H is defined as follows f |g Ω = d3 rf ∗ (r)g(r) (3. Calssical Electrodynamics. In (3. the surface elements da pointing out of the surface in a direction normal to the surface and the vector–valued function A(r) is taken at points r ∈ ∂Ω. The hermitian property.26) where ∂Ω da · A(r) denotes an integral over the surface ∂Ω of the volume Ω. New York. J. Interchanging f ∗ (r) and g(r) results in g| H |f Ω = Ω ˆ d3 r g(r) Hf ∗ (r) .24) Using Green’s theorem1 Ωd 3r f ∗ (r) 2 g(r) − g(r) 2 f ∗ (r) = one obtains the identity ∂Ω da · ( f ∗ (r) g(r) − g(r) f ∗ (r) ) (3. implies hermitian ˆ f | H |g 1 2 Ω ˆ = g| H |f Ω (3. hence. 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. in principle. t). The reader is advised to consult a reference text on ‘Linear Algebra’ to follow this argument. Chapter 1. t) g(r) − Ω d3 rg(r) U (r. .21) Ω This leads one to consider the integral f | H |g Ω = Ω ˆ d3 rf ∗ (r) H g(r) (3. Since energy is a real ˆ ˆ quantity one needs to require that the eigenvalues of the operator H are real and. D. Jackson. (John Wiley. that H is 2 .23) ˆ Since H is a differential operator the expressions (3.22) and (3. 1975). g|r) = − 2 2m ( f ∗ (r) g(r) − g(r) f ∗ (r) ) (3.26) the vector–valued function P (f ∗ . f ∗ (r) d3 rg(r) Ω 2 = − 2m d3 rf ∗ (r) Ω 2 g(r) − f ∗ (r) (3.. g|r) is called the ˆ concomitant of H and is P (f ∗ . for example. g|r) (3. however.22) ˆ where H is defined in (3.27) ˆ We will postulate below that H is an operator in H which represents energy.23).20).25) ˆ f | H |g Ω ˆ = g| H |f Ω + ∂Ω da · P (f ∗ .28) See.54 Schr¨dinger Equation o linear vector spaces of functions f (r) in which a scalar product between two elements f. (3. 2nd Ed.

The latter property stems from the fact that the boundary condition (3. t) is ∂t p(ω. In this case one can postulate both conditions (3. 3. obey (3. κ > 0 or like r−α .3: Particle Flux 55 and.. In either case does f (r) and all of its derivatives vanish asymptotically. i. 3. we can only allow functions which make the differential da · P (f ∗ .. e. like exp(−κr).30) can be postulated. . to avoid discontinuities in ψ(r. Since the total probability of finding the particle anywhere in space is d3 r|f (r)|2 = 1 (3. t). and that the density decays rapidly when one moves away from that area.e. . namely. t) + ψ(r. t) on a single connected surface ∂Ωj only either one of the conditions (3. (3. i. (3. (3.31) usually arises when a particle existing in a bound state is described..29) In fact.32) the wave function must decay for |r| → ∞ rapidly enough to be square integrable. .g.30) each condition holding on an entire surface ∂Ωj .33) The time derivative of p(ω.31) |r|→∞ (3. in which case ∂Ω1 is the inner sphere and ∂Ω2 is the outer sphere. t)|2 . |ψ(r.29.e. However. t)∂t ψ(r. i. The probability to observe the particle anywhere in the subvolume ω ⊂ Ω is p(ω. ∂Ω = ∂Ω1 ∩ ∂Ω2 ∩ ∂Ω3 ∩ .30) Note that these boundary conditions are linear in f . r ∈ ∂Ω (3. g|r) vanish on ∂Ω.. t) = ω d3 r |ψ(r. t)∂t ψ ∗ (r. t) which describes the state of o a particle moving in the potential U (r. if f and g satisfy these conditions than also does any linear combination αf + βg.29. r ∈ ∂Ω or da · f (r) = 0 ∀r. t) = ω 3 d3 r [ 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. 3. α > 2.34) An example is a volume between two concentric spheres. It must hold then for all f ∈ H f (r) = 0 ∀r. therefore. The observable directly linked to the wave function is the probability to find the particle at position r at time t. t)|2 .32). in this case also all derivatives of f (r) vanish at infinity. Often the closed surface of a volume ∂Ω is the union of disconnected surfaces3 ∂Ωj . A most common boundary condition is encountered for the volume Ω = R3 in which case one postulates lim f (r) = 0 “natural boundary condition” . t) ] .3.3 Particle Flux and Schr¨dinger Equation o The solution of the Schr¨dinger equation is the wave function ψ(r.

(3.41) holds for any volume ω ⊂ Ω one can conclude ∂t ρ(r. 3.35) d3 r . t) (3. t) such that d3 r|ψ(r. t) = |ψ(r.44) ˆ Note that the Hamiltonian H involves only real terms.37) d3 r ∂t ρ(r. t) = H ψ ∗ (r. t) = 0 (3. Using (3.39) d3 r · A(r) (3. The surface integral (3. t) + ω · j(r.37) can be expressed through a volume integral according to da · A(r) = ∂ω ω (3. One can multiply the solution of (3. t) .27) this can be written i ∂t p(ω.27) one can express this j(r. t) H ψ ∗ (r. t) − ψ(r. t) ψ ∗ (r. (3. ψ(r. Note that for a normalized wave function the quantity ρ(r. A natural choice for this constant is 1. We will assume in the remainder of this Section that the solutions discussed are normalized. t) = 0 . Accordingly the probability to observe the particle anywhere in the total volume Ω is constant. t) = ∂ω da · P (ψ ∗ (r.19) and its conjugate complex4 −i yields i ∂t p(ω. t)r. t)|2 is a probability density with units 1/volume. t)|2 = 1 Ω (3. t) − ψ ∗ (r.29.37) If one applies this identity to ω = Ω one obtains according to (3. ψ|r.26. t).36) According to (3. t) = P (ψ ∗ . One refers to such solution as normalized. t) ] .42) 2mi Since (3. t) = [ ψ(r. t) ψ(r. t) = ω Schr¨dinger Equation o ∂ ∗ ˆ ψ (r.56 Using (3. 3.30) ∂t p(Ω.41) where j(r. t) + 4 · j(r. t) = 0. t) ∂t ˆ ˆ ψ ∗ (r.18) by any complex number and accordingly one can define ψ(r. (3. (3.40) One can rewrite then (3. . t) .38) holds. t) H ψ(r.43) (3.

46) i.125)] and in a quadratic [c.167)] potential.3. t) = − ψ(r.f. t0 ) determines φ(k). t) is that of density flux. (2. i. Such case arose in Sect. 2. The generalization to three dimenisons implies then that the factor exp(i k · r) corresponds to an intial velocit k/m and a flux of the same magnitude. t) points in the direction to the outside of volume ω..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) . t) = 0 2 ∂ 2 i ψ(r. Equation (3. the initial condition at ψ(r.48. for f (r) ∈ R. 2.105. Note that j(r. (3.49) reads at t = t0 ψ(r. arises solely from the complex factor exp(i k · r).4: Free Particle 57 The interpretation of j(r. (3. t) = [2π]− 2 where the dispersion relationship holds ω = k2 . One can readily show by insertion into (3. 2. 2 we had demonstrated that a factor exp(ipo xo / ) induces a motion of 1-dimensional wave packets such that po /m corresponds to the initial velocity. (3. (3.f.71)].18) in Ω∞ = R in the case o U (r.45) Obviously.48) that the general solution is of the form ψ(r.e. t) (3. j(r. 3. and for particles moving in a linear [c. t) = − ∂ω da · j(r. This finding is consistent with the present evaluation of the particle flux: a factor exp(ipo xo / ) gives rise to a flux po /m.148.e. 2 for a free particle [c. (2. t) has vanishing flux anywhere.49) ˜ Obviously. In Sect.47) . This follows directly from an inspection of Eq.51) . t) .41) written in the form ∂t ω d3 r ρ(r. (2.50) 3 ˜ d3 k φ(k) exp i(k · r − ωt) Ω∞ (3. equal to the velocity of the particle. t) gives rise to a descrease of the total probability in volume ω due to the disappearence of probability density at the surface ∂ω.43) that any real wave function ψ(r. m (3.48) ∂t 2m which describes the motion of free particles. t0 ) = [2π]− 2 Ω∞ 3 ˜ d3 k φ(k) exp i(k · r − ωt0 ) . It is of interest to note from (3..f. 2m (3.

e. In fact. (3. Below we will generalize the propagator (3.e.54) valid for this case.52) above ψ(r.54) to the case of non-vanishing potentials U (r). Ω∞ (3.31) applies. evaluating the integral in (3.49). in (2. i. 3. t) = Ω∞ d3 r0 φ(r. respectively. In case of the harmonoc oscillator the expansion can be stated in a closed form given in (4.56) and using (2.54) This expression obviously has the same form as postulated in the path integral formulation of Quantum Mechanics introduced above. derive an expression similar to (3. We have identified then with (3.114) and (4. Comparision with Path Integral Formulation One can write solution (3.49. To show this one needs to note 1 2π = +∞ dk1 exp −∞ ik1 (x − x0 ) − 1 2 i 2 k2 1 2m .. 3.47).54) an alternative representation of the propagator (2. If one chooses the initial state f (r) defined in (3. t) as given by (3. that the “natural boundary condition” (3.47).49) is square integrable at all subsequent times and.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 (3.51) to be square integrable it follows according to the properties of the Fourier–transform that ψ(r.58 The inverse Fourier transform yields 3 ˜ φ(k) = [2π]− 2 Schr¨dinger Equation o d3 r0 exp(−ik · r0 ) ψ(r0 . i.81) . (t − t0 ) (3. The general form for this propagator involves an expansion in terms of a complete set of eigenfunctions as in (3. t0 ) . t0 ) = 1 2π 3 d3 k exp Ω∞ ik · (r − r0 ) − i 2 k2 2m (t − t0 ) . The solution as stated is defined in the infinte space Ω∞ = R . t0 ) (3.70) derived below for a particle in a box and the harmonic oscillator. t|r0 .53) where φ(r.247). The ensuing solutions are called wave packets.52) We have not specified the spatial boundary condition in case of (3. hence.5). t0 ) ψ(r0 .54) yields (2..51. t|r0 .

.54) is +∞ ψ(x. r ∈ ∂Ω or da · φ(r) = 0 ∀r. (3.57) where φ(x. i.49.56)].57–3. t0 ) = 1 2π +∞ dk exp −∞ ik (x − x0 ) − i 2m (t − t0 ) . The 1-dimensional version of (3.52) to the case that the initial state of a 1-dimensional free particle ψ(x0 .27) vanish on the surface ∂Ω of Ω.46).e.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 (3.64) (3.60) yields with t0 = 0 ψ(x.29. 3.20) which are of o the form ψ(r. the boundary conditions are φ(r) = 0 ∀r. 3. t) = t 1 − i mδ2 t 1 + i mδ2 1 4 1 πδ 2 (1 + 2 t2 m2 δ 4 1 4 ) × t po i p2 o ) + i x − t mδ 2 2m (3.19.53. t) is given by (??). r ∈ ∂Ω (3.59) which is solved through completion of the square in the exponent [c. t) = f (t) φ(r) .30).62) We will restrict the space of allowed solutions to a volume Ω such that the functions also make the concomitant (3.4: Free Particle Free Particle at Rest 59 We want to apply solution (3.f. t|x0 . We have demonstrated then in this case that the Schr¨dinger formulation of Quantum o Mechanics is equivalent to the Feynman path integral formulation.55.63) . a result which is identical to the expression (??) obtained by means of the path integral propagator (2. t|x0 . Stationary States We consider now solutions of the time-dependent Schr¨dinger equation (3. the functions obey boundary conditions of the type (3. 3. (3. (3.60) =???? Combining (3. (3. t0 ) ψ(x0 . 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. t) 2 k2 (3. t) = −∞ dx0 φ(x. Accordingly.61) × exp − (x − po m t)2 2δ 2 (1 + 2 t2 ) m2 δ 4 (1 − i .

. This yields an expression g1 (t) h1 (r) = g2 (t) h2 (r) (3. We will encounter many such problems in the subsequent Sections. It follows from the observation that for the solution space considered (3. t)|2 = |f (t)|2 Ω d3 r |φ(r)|2 (3. t) = f (0) exp − i E φE (r) where ˆ H φE (r) = E φE (r) . according to (3.64). t)|2 = |φ(r)|2 . (3.42) the flux j(r.62) have the particular property that the associated probability distributions are independent of time.69) is called an eigenvalue problem.45) that the total probability d3 r ρ(r.69) exist. We want to demonstrate this property now. (3. the eigenvalues of the operator H. by no means all solutions are of this type. 3. We have then shown that ψ(r. if f (t) = eiα .66) i.e. a common case is Ω = Ω∞ and the ‘natural boundary condition’ (3.e.68) can hold only for all t and all r if g1 (t) = E g2 (t) and E h1 (r) = h2 (r) for some E ∈ C.. t) = Ω Ω d3 r |ψ(r. We must postulate therefore ∂t f (t) ˆ H φ(r) = E f (t) = E φ(r) .62) do exist and we will characterize the two factors of the solution f (t) and φ(r). It turns out that a solution for f (t) can be found for any E. .65) is constant. one factor depending only on the time variable and the other only on the space variables are called separable in space and time. n = 1. namely f (t) = f (0) exp − i Et . and h2 (r) = H φ(r). in general. At this point we will accept that solutions φ(r) of the type (3.62) we insert it into (3. i.62) are special solutions of the timedependent Schr¨dinger equation. We denote the corresponding solution by φE (r). for functions required to obey boundary conditions (3.27) holds and. We may note in passing that solutions of the type (3. α ∈ R . hence.. One calls such states stationary states.60 Schr¨dinger Equation o and affect only the spatial wave function φ(r). solutions exist only for a set of discrete E ˆ values. One can conclude that the probability density for the state (3. the solutions o (3. It follows then from (3. At this point we state without proof that.63.e. In fact. .71) . is time-independent. often only for a discrete set of values En . 2. g2 (t) = f (t). (3. t) vanishes on the surface of ∂Ω.70) The task of finding solutions φ(r) which solve (3.68) ˆ where g1 (t) = i ∂t f (t). however. for the eigenvalue problems in the confined function space.31). i. h1 (r) = φ(r).62) which consist of two factors. In order to further characterize the solution (3. This can hold only if |f (t)| is time-independent. The identity (3.19). . We will demonstrate that solutions of the type (3.69) If these two equations can be solved simultaneously a solution of the type (3.62) is |ψ(r. As pointed out above.67) (3. (3.62) exists. It is important to realize that the separable solutions (3.

k ∈ R 2 k2 φk (r) = N exp i k · r .3. E (3. The flux corresponding to (3. real.66) E must be real. (3.74) is actually best labelled by an index k. t) = exp − i 2 k2 2m t exp i k · r (3.. We will wave this assumption as we always need to do later whenever we deal with unbound particles. of course. hence. should be positive. As a result we cannot postulate in the present case that wave functions are localized and normalizable.e.43) is j(r. in fact. a property which is to be expected since the energy is purely kinetic energy which. m (3. particles scattered of a potential. (3.71) this yields E Ω d3 r φ∗ (r) φE (r) = E ∗ E Ω d3 r φE (r) φ∗ (r) . The corresponding stationary solution ψ(r. We start our proof using the property (3. E ∈ R. e. 3.28) both represent the state φE (r).71) are. one can interpret then k as the magnitude of the momentum of the particle.20).77) Noting that k can be interpreted as the magnitude of the momentum of the particle the flux is equal to the velocity of the particle v = k/m multiplied by |N |2 . The resulting total energy values E are positive.75) 2m One can ascertain this statement by inserting the expression for φk (r) into the eigenvalue problem posed in (3. Obviously. t) = exp − E t 2 φE (r) . We will show in Section 5 that E can be interprerted as the total energy of a stationary state.73) from which follows E = E ∗ and. E (3. Stationary State of a Free Particle We consider now the stationary state of a free particle described by i ψ(r.74) using exp i k · r = ik exp i k · r . t) = |N |2 k . We want to prove now that the eigenvalues E which arise in the eigenvalue problem (3. − 2 2m φ(r) = E φ(r) .72) According to (3.4: Free Particle 61 is a solution of the time-dependent Schr¨dinger equation (3. i.g. The solution φE (r) corresponding to the eigenvalue problem posed by (3.76) according to (3.28) for the special case that f and g in (3.19. o According to (3.74) The classical free particle with constant energy E > 0 moves without bounds in the space Ω∞ . ˆ d3 r φ∗ (r) H φE (r) = E Ω Ω ˆ d3 r φE (r) H φ∗ (r) . = E.76) has kinetic energy 2 k 2 /2m. .

We can expect.e. t) = exp(−iEt/ )φE (x) where φE (x) is determined by 2 d2 − φE (x) = E φE (x) . i. φ(±a) = 0 . (o) 2 k2 (e) 2 k2 2m = E (3. and so-called odd solutions obeying φ(x) = −φ(−x) φE (x) = A sinkx.80) 2m dx2 We note that the box is symmetric with respect to the origin. It turns out that this boundary condition can only be satisfied for a discrete set of k–values kn .80) need to be satisfied are φE (e.81. therefore. f continuous. t) = − d2 ψ(x.81) 2m = E. In case of the even solutions (3.84) . so-called even solutions obeying φ(x) = φ(−x) φE (x) = A coskx . 3. we have to consider only one boundary condition.80).78) (3. let say the one at x = a.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. Hence. two types of solutions. f ∈ F1 = {f : [−a.62 Schr¨dinger Equation o 3. We will refer to this as a particle in a one-dimensional ‘box’. 3. . (3. that the solutions obey this symmetry as well.82) have the property that either both boundary conditions are satisfied or none.81.81) they are kn = nπ . The boundary conditions which according to (3.79) The stationary solutions have the form ψ(x. 2m dx2 2 (± ) = } (3. 2a n = 1. 5 . a] ⊂ R → R. (3.o) (−a) = 0 (3. Solutions of the Schr¨dinger Equation in F1 o The time–dependent solutions satisfy i ∂t ψ(x.82) One can readily verify that (3. a] ⊂ R assuming that the potential outside of this interval is infinite.83) The solutions (3. hence. 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. a}.82) satisfy the differential equation in (3. We assume. 3. .. (3.o) (a) = 0 and φE (e. . a] ⊂ R which vanish on the surface ∂Ω1 = {−a. n ∈ N. t) .

85) (3. 4. = 0.86). . 3 . . a n = 2.92) and for the odd states +a |An |2 −a dx sin2 nπx = |An |2 a = 1 .1). (3.5: Particle in One-Dimensional Box since for such kn 63 nπa nπ = cos 2a 2 In case of the odd solutions (3.82) only the kn –values cos(kn a) = cos kn = nπ . (3. (3. 2a n = 1. 2a n = 2. φ(e) (a. according to the dispersion relationships given in (3. 5 . n = 1. . 2a 1 . 4.90) 2a The wave functions represent stationary states of the particle in a one-dimensional box. 6 . i. . 4.87) (Note that.86). 2.93) The normalization constants are then An = (3. .86) satisfy the boundary condition since for such kn nπa sin(kn a) = sin 2a = sin nπ 2 = 0. 3.81) and (3. 6 .) The Energy Spectrum and Stationary State Wave Functions The energy values corresponding to the kn –values in (3. .e.94) . . By counting the number of their nodes one can determine the energy ordering of the wave functions. . x) = An sin n nπx . (3. respectively. n is assumed to be even.91) holds. 2a n = 1. This condition implies for the even states +a |An |2 −a dx cos2 nπx = |An |2 a = 1 . according to (3.3.81.o) (a. Notice that the number of nodes of the wave functions increase by one in going from one state to the state with the next higher energy En . (3. (3. x) = An cos n and φ(o) (a. . x)|2 = 1 n (3.89) nπx .82). . (3. It is desirable to normalize the wave functions such that +a −a dx |φ(e. (3. 6 .84. are En = 2π2 8ma2 n2 .88) where the energies for odd (even) n–values correspond to the even (odd) solutions given in (3.. 5 . 3. .82). The wave functions for the five lowest energies En are presented in Fig. n = 2. 3. 3. . .

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

.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. x) = together with the scalar product5 +a 65 (3. for n = 2.98) The latter property is obviously true for n = m. the integrand is odd. too.89. . . 5 . In fact. and (iii) the mixed case. 3. the integral arises φn |φm 1 a Ω1 = 1 a +a nπx −a dx cos 2a cos mπx = 2a (3. the integral vanishes. . (i) n. x) (3. the integrals vanish. x) . m both even. it holds: f |f Ω1 = 0 → f (x) ≡ 0. .96) f |g Ω1 = −a dxf (x) g(x) . (ii) n. Similarly. 2a 2a (3. i. m even φn |φm 1 a Ω1 = 1 a +a nπx −a dx sin 2a sin mπx = 2a (3. n = 1.102) holds according to (3.} where φn (a. n (3.101) +a −a dx cos (n−m)πx − cos (n+m)πx 2a 2a and. Hence we need to consider only the first two cases. .98) ∞ f |f 5 Ω1 = n=1 d2 . . this integral vanishes. 2. (3.98) the elements of B1 must be linearly independent.103) We will show in Section 5 that the property of a scalar product do indeed apply. it holds φn |φm Ω1 = δn m . f.e..90.95) 1 a cos nπx 2a sin nπx 2a for n = 1. In case of n.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. hence. m both odd. 3. for ∞ f (x) = n=1 dn φn (a. In particular. n = m. Because of the property (3.97) form an orthonormal basis set.3. i. In case of n = m we have to consider three cases. . a] are N. 3.99) Since in this case the integrand is a product of an even and of an odd function. 6 . Obviously. N ≥ 1. 3. one obtains for n. The latter case leads to integrals φn |φm Ω1 = 1 a +a dx cos −a nπx mπx sin . m odd.e.. (3. 4. g ∈ F1 (3.94) B1 = {φn (a.

Demonstration of completeness is a formidable task. Hence.108) into the Schr¨dinger equation yields o +a ∞ n=1 φn (a. .107) Inserting this into (3. x0 ) ψ(x0 . 1. t0 ) .109) Insertion of (3.106) and generalizing to t ≥ t0 one can write ∞ +a ψ(x. t0 ) in terms of eigenfunctions φn (a. . x) ∂t cn (t) −a dx0 φn (a. t0 ) = dm .105) must vanish. (3. .78) can be expressed as a linear combination of the elements of B1 defined in (3. For this purpose we expand ∞ ψ(x. . n = 1.106) Using the orthonormality property (3. i. (3. .95) is an orthonormal basis.} and {bn . x) . t0 ) = +a ∞ i n=1 − En φn (a. We have then shown that any f corresponding to elements of F1 can be expanded in terms of elements in B1 . i. 2.66 Schr¨dinger Equation o f (x) ≡ 0 implies f |f Ω1 = 0 which in turn implies dn = 0 since d2 ≥ 0. t0 ) = n=1 dn φn (a.. . n = 2. For this purpose we extend the definition of the elements of F1 to the whole real axis through ˜ f : R → R. there exist real constants {an . x0 ) ψ(x0 . . In the present case.95) is also complete. 2. they can be expanded in terms of a Fourier series.95. Accordingly. x0 ) ψ(x0 . ( ) = ([ + ] − ) (3. −a (3. .e. 3.96). any element of the function space F1 defined in (3. . the coefficients an . . a] yields then also an expansion for any element in F1 and B1 is a complete basis for F1 . 4. and bn . This implies that only the trigonometric functions which are elements of B1 enter into the Fourier ˜ series. .} such that ∞ ∞ nπx nπx ˜ = + an sin (3. The functions f are periodic with period 2a. . such demonstration can be based on the theory of Fourier series.108) where the functions cn (t) are to be determined from the Schr¨dinger equation (3. x) to obtain an expression for ψ(x. t) at times t > t0 . Evaluating the Propagator We can now use the expansion of any initial wave function ψ(x. x) cn (t) −a dx0 φn (a. 3. 3. It follows that B1 n defined in (3. t) = n=1 φn (a. n = 1. Restricting the expansion (3. (3. We like to show finally that the basis (3.e.105) f an cos 2a 2a n=0 n=1 ˜ The functions f corresponding to the functions in the space F1 have zeros at x = ±m a. in (3. . x) cn (t) −a dx0 φn (a.98) one obtains +a dx0 φm (a.104) ˜ where [y]a = y mod2a.79) requiring the o initial condition cn (t0 ) = 1 . however.110) . n = 0. 5 .105) to the interval [−a. . t0 ) (3. m = 1. x0 ) ψ(x0 ..

.114) the representation of the propagator for a particle in a box with infinite walls. t0 ) (3. In case that different boundary conditions hold the propagator will be different as well. In the present system which is composed solely of bound states the propagator is given by a sum. The particle moves then to the left. being reflected at the right wall. a] yields. rather than by an integral (3.113) where φ(x. t|x0 . x0 ) . i ∂t cm (t) = − Em cm (t) . (3. k0 = .78). t0 |x0 .111) The solutions of these equations which satisfy (3. The interference pattern begins to ‘smear out’ first.5).109) are cm (t) = exp i − Em (t − t0 ) . (3.5: Particle in One-Dimensional Box Multiplying both sides by φm (a. t0 ) = 1 2πσ 2 1 4 exp − x2 0 + i k0 x0 2σ 2 .98). t0 ) ψ(x0 . The last frame shows the wave front reaching again the right wall and the onset of new interference. t0 ) is also referred to as a Greens function.112) determine now ψ(x. t|x0 . 4 a (3. Note that the propagator has been evaluated in terms of elements of a particular function space F1 .114) This expression has the same form as postulated in the path integral formulation of Quantum Mechanics introduced above. σ = a 15 .112) Equations (3. t) = −a dx0 φ(x. in (2. t|x0 . t0 ) itself is a solution of the time-dependent Schr¨dinger o equation (3. t) for any initial condition ψ(x. Example of a Non-Stationary State As an illustration of a non-stationary state we consider a particle in an initial state ψ(x0 .79) which lies in the proper function space (3. t0 ) = ∞ φn (a. One can recognize that the particle moves first to the left and that near the right wall of the box interference effects develop. cm (t0 ) = 1 .e.2 presents the probability distribution of the particle at subsequent times.54) as in the case of the free particle system which does not exhibit any bound states. This solution can be written +a ψ(x.113).3. but the collision with the left wall leads to new interference effects. x) exp n=1 i − En (t − t0 ) φn (a. The respective initial condition is φ(x. 67 (3. Often the propagator φ(x. It is of interest to note that φ(x. the elements of which satisfy the appropriate boundary conditions. Figure 3. t0 ).. 3. according to (3. x) and integrating over [−a. i. t0 ) = δ(x − x0 ) as can be readily verified using (3.108. t|x0 .115) This initial state corresponds to the particle being localized initially near x0 = 0 with a velocity v0 = 15 /m a in the direction of the positive x-axis. We have identified then with (3.

16 m a h2   1 πσ 1 πσ -a a -a a Figure 3. .2: Stroboscopic views of the probability distribution |ψ(x. t)|2 for a particle in a box starting in a Gaussian distribution with momentum 15 /a.20 m a h2 1 πσ 1 πσ -a a -a a |Ψ|2 2 t = 1.48 m a h2 |Ψ|2 2 t = 0.44 m a h2 |Ψ|2 2 t = 1.92 m a h2 |Ψ|2 2 t = 2.00 m a h2 |Ψ|2 2 t = 0.68 Schr¨dinger Equation o |Ψ|2 2 t = 0.96 m a h2   |Ψ|2 2 t = 1.68 m a h2 1 πσ 1 πσ -a a -a a |Ψ|2 2 t = 1.72 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.

(3. x3 ) = E φE (x1 .6 Particle in Three-Dimensional Box We consider now a particle moving in a three-dimensional rectangular box with side lengths 2a1 . 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. one can express H 3 φ(x1 . (3. 3. x2 .117) 2m 2 2 2 ∂1 + ∂2 + ∂3 (3.3. x3 . . x2 .114) in terms of which the solutions for all initial conditions can be expressed. We have defined in the space F1 a scalar product (3. x2 = ±a2 } ∪ {(x1 . x2 . ˆ H = − 2 (3. a3 ] ⊂ R = {(x1 . i. can be counted. x2 .120) ˆ Since the Hamiltonian H is a sum of operators O(xj ) each dependent only on a single variable. x3 . j = 1. x2 . x1 = ±a1 } ∪ {(x1 . x2 .118) which are stationary states. x3 )T ∈ Ω. x2 . hence. The description employed a space of functions F1 defined in (3.e. ˆ = O(x1 ) + O(x2 ) + O(x3 ). a1 ] ⊗ [−a2 . x2 . x3 )T ∈ Ω. i. a2 ] ⊗ [−a3 . 2. The corresponding solutions have the form i ψ(x1 . x3 = ±a3 } . This property allowed us to evaluate the propagator (3. x3 ) (3. 2a2 . f continuous. t) . x2 . x3 ) . x3 )T ∈ Ω. 2a3 . An important property of this basis and. x2 . of the space F1 is that the elements of the basis set can be enumerated by integer numbers. ( . x3 ) is an element of the function space F3 defined in (3.122) .119) where φE (x1 .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. t) = exp − E t φE (x1 . x2 . )T ∈ ∂ } ..78). t) = H ψ(x1 . Placing the origin at the center and aligning the x1 .e.97) with respect to which the eigenfunctions are orthonormal. A complete basis of F1 is given by the infinite set B1 (3.116) where Ω is the interior of the box and ∂Ω its surface Ω ∂Ω = [−a1 . x3 ) = j=1 φ(j) (xj ) (3.121) where − 2 2m 2 ∂j φ(j) (xj ) = Ej φ(j) (xj ) . We seek then solutions of the time-dependent Schr¨dinger equation o ˆ i ∂t ψ(x.95).116) and obeys the partial differential equation ˆ H φE (x1 . 3 (3.. . ) = ∀( . φ(j) (±aj ) = 0 .

n2 .122) with (3. r) exp i − E(n1 n2 n3 ) (t − t0 ) φ(n1 . r0 ) . a1 = a2 = a3 = a then rotation around the x1 . x2 ) φn3 (a3 .96) and. namely rotation by π around the x1 . 2. 3. hence. x3 –axes by π/2 also leaves the system unaltered. . a2 . t0 ) = n1 . x3 ) n1 . a3 are identical. x2 .125) is a complete orthonormal basis of F3 and that the propagator for the three-dimensional box is φ(r. the solutions of (3. . . . Comparing (3.124) The same considerations as in the one-dimensional case allow one to show that B3 = {φ(n1 .} ∞ (3. . a3 .123) 8m n2 n2 n2 1 2 3 + 2 + 2 a2 a2 a3 1 . This symmetry has been exploited in deriving the stationary states.n2 . x1 ) φn2 (a2 .n3 =1 φ(n1 . 3. n2 . The energies and corresponding degeneracies of the particle in the three-dimensional box with all side lengths equal to 2a are given in the following Table: n1 1 1 1 1 2 1 2 1 2 3 1 n2 1 1 2 1 2 2 2 1 3 3 1 n3 1 2 2 3 2 3 3 4 3 3 5 E/[ 2 π 2 /8ma2 ] 3 6 9 11 12 14 17 18 22 27 27 degeneracy single three-fold three-fold three-fold single six-fold three-fold three-fold three-fold single three-fold One can readily verify that the symmetry of the box leads to three-fold and six-fold degeneracies.e. a2 . x3 ) .122) are given by (3.n2 . If two or all three orthogonal sides of the box have the same length further symmetry operations leave the system unaltered.70 Schr¨dinger Equation o and E1 + E2 + E3 = E. x1 . (3. a2 . a2 . x2 . 3. Such degeneracies are always a signature of an underlying symmetry. in the present case . . n1 . For example.120) can be written φ(n1 . t|r0 . (3. n1 . n3 = 1. n3 = 1.n3 ) (a1 . x2 . 2. n2 . Actually. if all three lengths a1 . x2 .n3 ) (a1 . n 3 and E(n1 . 2.n3 ) = 2π2 = φn1 (a1 .n3 ) (a1 . x3 ) = 1. ..80) shows that the solutions of (3. . x3 –axes. (3.126) Symmetries The three-dimensional box confining a particle allows three symmetry operations that leave the box unchanged. a3 .n2 . a3 . . i. a3 . n 2 . x1 .n3 ) (a1 . This additional symmetry is also reflected by degeneracies in the energy levels. a2 .n2 .n2 .

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

72 Schr¨dinger Equation o .

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

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. the natural boundary conditions apply x→±∞ lim φE (x) = 0 .4) Equation (4. x] = p ˆ . however. in which the operators used below.5) →±∞ where C∞ denotes the set of functions which together with all of their derivatives are continuous. In the following algebraic solution of (4.3) Due to the nature of the harmonic potential which does not allow a particle with finite energy to move to arbitrary large distances.4) be satisfied. The operators act ˆ on the space of functions N1 defined in (4. however. ∈ C∞ . (4.2) can be expressed in terms of the two operators p = ˆ d . This property states that p and x obey an ˆ ˆ algebra in which the two do not commute. t) where ˆ H φE (x) = E φE (x) .7) of the Hamiltonian.5). It is important to keep in mind this restriction of the space. beside the representation (4.3) would be continuous and the eigenfunctions φE (x) would not be normalizable. The cardinal property exploited below. is the commutation property 1 1 (4.7) which is why these operators are of interest to us. the commutator has a simple form. We will point out explicitly where assumptions are made which built on this restriction.1) of the form ψ(x.6) the first being a differential operator and the second a multiplicative operator. The stationary states of the harmonic oscillator have been considered already in Chapter 2 where the corresponding wave functions (2.74 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. 4. all stationary states of the harmonic oscillator must be bound states and. only for a discrete set of E values can the boundary conditions (4. i dx x = x ˆ (4.8) i which holds for any position and momentum operator. act. The Hamiltonian H can be expressed in terms of the operators acting on the space (4. therefore. If this restriction would not apply and all functions f : R → R would be admitted.235) had been determined. In order [ˆ. t) = exp(−iEt/ )φE (x.3) can be solved for any E ∈ R.5) ˆ H = 1 1 2 p + m ω 2 x2 ˆ ˆ 2m 2 (4. In the framework of the Schr¨dinger equation the stationary o states are solutions of (4. (4. lim ( ) = } (4.1 Creation and Annihilation Operators The Hamiltonian operator (4. the spectrum of ˆ H in (4.

10) holds for a ˆ ˆ [ˆ− .12) To prove this commutation property we determine using (4.17) d dx We also express a+ and a− directly in terms of the coordinate x and the differential operator ˆ ˆ a+ = ˆ mω x − 2 d .15) together lead to the commutation property (4.8) we recall that all operators act on functions in N1 and. Our next step is an attempt to factorize the Hamiltonian (4.11) a− a+ = ˆ ˆ (4.14) 1 i mω 2 x + ˆ p2 + ˆ [ˆ. ˆ 2 2m ω (4.1: Creation and Annihilation Operators 75 to derive (4. 4. p ˆ 2 2m ω 2 (4.14) and (4. ˆ 2 2m ω a− = ˆ mω i x + √ ˆ p. hence. a+ ] = a− a+ − a+ a− does not necessarily vanish. a ˆ 1 (4.7) and (4. In fact.7) assuming that the factors are easier to handle than the Hamiltonian in yielding spectrum and eigenstates. x] f (x) = p ˆ d d x f (x) − x f (x) = f (x) = 1 f (x) . ˆ ˆ . 2mω dx a− = ˆ mω x + ˆ 2 d . Being guided by the identity for scalar numbers (b − ic)(b + ic) = b2 − i(cb − bc) + c2 = b2 + c2 (4. in fact. Before we continue we like to write (4. 1 i dx i dx i i (4.8).11) ˆ The reader may note that we have attempted. Since a+ and a− are ˆ ˆ + and a− since operators and not scalars we cannot simply expect that the identy (4. 1 ω 2 (4. x] . the action of [ˆ. a+ ] = 1 .13) (4.15) 1 ˆ 1 H + 1 . One obtains [ˆ. the commutator property (4.8) a ˆ ˆ ˆ ˆ ˆ implies [ˆ− .15) in a form which will be useful below ˆ H = ˆ H = ω a− a+ − ˆ ˆ ω a+ a− + ˆ ˆ ω 1 1 2 ω 1 .16) (4.9) From this follows (4. 2mω dx (4.4.10) we define a+ = ˆ mω i x − √ ˆ p. 1 2 (4.14.8) yield a− a+ = ˆ ˆ Similarly one can show a+ a− = ˆ ˆ 1 ˆ 1 H − 1 . 1 ω 2 (4.12).18) It is of interest to note that the operators a+ and a− are real differential operators. x] p ˆ on such functions must be considered. to factor H/ ω.

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. This property of operators is investigated more systematically in Section 5. it holds that for a stationary state φE (x) of energy E defined through (4. g ∈ N1 +∞ +∞ dx f (x) a+ g(x) = −∞ −∞ dx g(x) a− f (x) .17.24) − m This property states that the operators in the function space N1 are the hermitian conjugate of each other. 4.19) This property states that the operators a+ and a− are the adjoints of each other.16.19) in the form1 f |ˆ+ g a Ω∞ = g|ˆ− f a Ω∞ . Using (??) we like to state (4.23) Together.19). 4.20) ˆ In the following we will determine the spectrum of H and its eigenstates. ˆ (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) . The derivation will be based solely on the properties (4. (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. In fact.22. 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) .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. .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) . The property ˆ ˆ follows directly from (??). 4. one obtains using (4.16. 4. 4.21) is available.16. ˆ (4.12.17. 4.22) Similarly one can show using again (4. (4. (4.17. ˆ The results (4.

(4. +∞[ one can write (4.25) can. It turns out that both difficulties can be resolved together.25) needs to be normalizable in order to represent a bound state of the harmonic oscillator.31) (4.26) Of course. For obvious reasons one refers to a+ and a− as ladder ˆ ˆ operators. φ0 (x) should be an element of the function space N1 defined in (4.29) d 1 d φ0 (y) = = lnφ0 (y) = − y . There arise.28) 4. (ii) the construction appears to yield energy eigenvalues without lower bounds when. φ0 (y) dy dy the solution of which is 1 lnφ0 (y) = − y 2 + c0 2 φ0 (y) = c0 exp 1 − y2 2 . one expects that E = 0 should be a lower bound. For f (x) = φE (x) and g(x) = φE (x) in (??) follows (Note that according to (??) the eigenvalue E is real. (4.27) Since E = E one can conclude φE |φE Ω∞ = 0. The property (??) has an important consequence for the stationary states φE (x). the solution φ0 (x) of (4. however. Let φE (x) and φE (x) be two normalized stationary states corresponding to two different energies E.25) φ0 (y) = 0 (4. be found. Another name is creation (ˆ+ ) and annihilation (ˆ− ) operators since these operators a a create and annihilate vibrational quanta ω.2 Ground State of the Harmonic Oscillator d + y dy A suitable solution of (4.11) one can rewrite (4. (4. i.e..17) ˆ H φ0 (x) = ( ω a+ a− + ˆ ˆ 1 ω 1 ) φ0 (x) = 1 ω φ0 (x) . (4.25) on one side would lead to termination of the sequence E0 + m. Using (4.4.32) for some constant c0 or (4. E = E .30) Assuming that φ0 (y) does not vanish anywhere in its domain ] − ∞.29) where y = mω x. 2 2 (4. two difficulties: (i) one needs to know at least one stationary state to start the construction. on ˆ the other side such a state is itself an eigenstate of H as can be shown using (4. in fact.6.2: Ground State 77 One can use these relationships to construct an infinite number of stationary states by stepping up and down the spectrum in steps of ± ω.) ˆ ˆ 0 = φE |HφE Ω∞ − φE |HφE Ω∞ = ( E − E ) φE |φE Ω∞ . m ∈ Z when m is decreased. 4.33) .5). In fact. a state φ0 (x) which obeys the property a− φ0 (x) = 0 ˆ (4. E . in fact.

39) We notice that the ground state wave function φ0 (x) as well as the operators (ˆ+ )n are real. (4. (4. (4. . . 2. .36) assumes a functional form such that both terms together assume a minimum value. 2. 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) .e. .78 Linear Harmonic Oscillator This solution is obviously normalizable. . 2 n = 0. 4. and the kinetic energy contribution.25) admits only one solution there is only one set of states with energy E + m ω. .2) there are two competing contributions to the energy. mω (4. 2. . The conventional normalization condition φ0 |φ0 reads +∞ +∞ Ω∞ = 1 (4. The set of allowed energies of the oscillator according to (4. n = 0. choose the normalization constants cn and the functions φn (x) real as well.37) above the ground state energy.. 1.e. m ∈ Z which properly terminate at some minimum value E + m0 ω ≥ 0. i.36) n times. . . We a can. 1.26) the energy value associated with this state is 1 ω. i.35) The appropriate ground state is φ0 (x) = mω π 1 4 exp − mωx2 2 ... 2.37) It is most remarkable that the energy 1 ω of the ground state is larger than the lowest classically 2 allowed energy E = 0.36) Since the first order differential equation (4. 2 n = 0. . 1. .24) can then be written as follows En = ω(n + 1 ).34) dx |φ0 (x)|2 = |c0 |2 −∞ −∞ dx exp − mωx2 2 = |c0 |2 π . a state confined to the minimum corresponding to the classical state of lowest energy. For this purpose we apply the operator a+ to the ˆ ground state (4. We recall that according to (4. therefore. The reason is that in the Hamiltonian (4. we can now construct the stationary states corresponding to energies (4. . (4. the potential energy contribution which for a state δ(x). which for a narrowly localized state yields a large positive value. would yield a vanishing contribution. we construct the states for n = 1. the so-called excited states.e. it holds ˆ H φn (x) = ω(n + 1 ) φn (x) . i. We will consider this point more systematically in Section 5.39). .. The ground state (4.3 Excited States of the Harmonic Oscillator Having obtained a suitable stationary state with lowest energy.38) These states correspond to the energy eigenvalues (4.

ˆ Repeated application of this relationship yields φn−s (x) = (n − s)! − a ˆ n! s 2 1 = βn n φn |φn Ω∞ 2 = βn n . 4.47). i.18)] and since the ˆ φn (x) are real functions. (4. . n = 0. 1. (4. √ a− φn (x) = n φn−1 (x) . . 2.3: Excited States 79 We need to determine now the normalization constants cn .40) = 1. n = 0. 1. .46) According to (4.51) . For n = 0 holds c0 = 1. We consider then the situation that we have obtained a properly normalized state φn (x).40). (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.20) yields 2 αn φn |ˆ− a+ φn a ˆ Ω∞ = 1.43) (4. n.44) One can conclude then that the stationary states of the oscillator are described by the functions 1 φn (x) = √ a+ ˆ n! n φ0 (x) . ˆ (4. (4.24) holds in analogy to (4.49) (4. (4. To determine these constants we adopt a recursion scheme. To determine βn we note using (4. 2.42) ˆ Using (4. they obey φn |φn = δn n . Since a− is a real differential operator [see (4. according to (4.40–4.20) 1 = φn−1 |φn−1 Equations (4. Ω∞ 2 = βn a− φn |ˆ− φn ˆ a Ω∞ 2 = βn φn |ˆ+ a− φn a ˆ Ω∞ .15 .39) yield √ From this follows βn = 1/ n and.50) φn (x) .4. (4.45) We like to note that these functions according to (4.e. .47) for some suitable constants βn . according to (4.40) φn−1 (x) = βn a− φn (x) ˆ (4. √ a+ φn (x) = n + 1 φn+1 (x) . βn must be real as well.48) (4. .41) Employing the adjoint property (4. A properly normalized state φn+1 (x) is then of the form φn+1 (x) = αn a+ φn (x) ˆ for some real constant αn which is chosen to satisfy φn+1 |φn+1 Ω∞ 2 = αn a+ φn |ˆ+ φn ˆ a Ω∞ (4.28) and the construction (4. .45) form an orthonormal set.

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. n–independent factor.58) This result agrees with the expression (2.233) derived in Sect.55) 1 y2 2 (4. i.52) This normalization of the wave functions differs from that postulated in (4. It should be noted that √ the normalization (4.35) by a constant. For this purpose we start from expression (4. for example. by dx = dy mω 1 4 .57) which includes the factor 1/ n! according to Eqs.52). by introducing the variable y defined in (4. . simplifying the calculation.46) yields a set of normalized states.53) We will later re-introduce this factor to account for the proper normalization (4.7 and given.56) Relationship to Hermite Polynomials We want to demonstrate now that the expression (4.54) The eigenstates of the Hamiltonian are then given by φn (y) = − √ e n n! π 2 1 y2 2 e y2 2 y − d dy n e− y2 2 .e.56) can be expressed in terms of Hermite polynomials Hn (y) introduced in Sect.7.40–4. (4.45). In terms of y the ground state wave function is φ0 (y) = π − 4 e− and a+ is ˆ 1 a+ = √ ˆ 2 y − d dy .28) of the ground state and the definition (4.200)..57) such that one can write the stationary state wave functions of the harmonic oscillator φn (y) = − √ e 2n n! π 1 y2 2 Hn (y) . (4. however. (4. (4.35) rather than (4. 2. 2. namely the square root of the Jacobian dx/dy. (4. We will demonstrate below the identity Hn (y) = e y2 2 y − d dy n e− y2 2 (4.30) and employing the normalization +∞ −∞ dy φ2 (y) = 1 n (4.39). by the Rodrigues formula (2.

Our derivation will follow closely the procedure adopted for the case of a ‘particle in a box’ [see Eqs.106–3. g(y) = (−1)n+1 e y2 2 dn+1 −y2 e = dy n+1 y − d dy n+1 e− y2 2 . and showing then that the property also holds for n + 1 in case it holds for n. hence. We prove (4.39) ∞ ψ(y0 . t0 ).59) hold. 1.30) and return later to the coordinate x. t0 ) = n=0 dn φn (y0 ) .63) one obtains g(y) = y − d dy f (y) (4. 4. (4.59) dn −y2 e dy n (4.61) and.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 (4.200).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 .114)].61) One can factor g(y) and employ (4. (4.4: Propagator 81 To verify the realtionship between the definition (4.4.64) which implies that (4. i.4 Propagator for the Harmonic Oscillator We consider now the solution of the time-dependent Schr¨dinger equation of the harmonic oscillator o (4.60) and.. Starting point of our derivation is the assumption that the initial condition can be expanded in terms of the eigenstates φn (y) (4.57) of the hermite polynomials and the definition given by (2. the Rodrigues formula (2. therefore.65) .2) for an arbitrary initial wave function ψ(x0 .59) holds for n = 0.59) by induction noting first that (4. dy n (4.e. (4. (3. dy dy (4.1. 4.59) as follows g(y) = −e = −e y2 2 y2 2 n y2 d d − y2 2 e 2 (−1)n e 2 e−y n dy dy d − y2 d n − y2 e 2 y − e 2 . For the sake of notational simplicity we employ initially the coordinate y as defined in (4.

t1 ) = −∞ dy0 φ(y.h.70) is the propagator of the linear harmonic oscillator.108–3. sec:harm. N. t) = π −3/2 exp 2 2 y 2 y0 + 2 2 2 +∞ +∞ ∞ du −∞ −∞ dv n=0 1 ( −2tuv )n × n! (4. t) = n=0 Hn (y)e−y /2 Hn (y0 )e−y0 /2 n √ t . We consider for this purpose the following expression for |t| < 1 ∞ w(y. t|y0 .70). t0 ) = n=0 φn (y) φn (y0 ) tn+ 2 . We want to demonstrate now that this propagator is identical to the propagator (2. Employing orthogonality condition (4.71) Applying (??) to express Hn (y) and Hn (y0 ) yields exp(−2tuv) w(y. 2n n! π 2 2 (4. y0 .82 Linear Harmonic Oscillator Such expansion is possible for any f (y0 ) = ψ(y0 . 4. 2. 1965) Sect. For this purpose we start from the integral representation (2.147) for the harmonic iscillator determined in Sect. . 68.7. Englewood Cliffs. (4. Inc.65) to times t ≥ t0 through insertion of time-dependent coefficients cn (t1 ) ∞ ψ(y. t) (4.39) holds cn (t1 ) = exp −i ω ( n + 1 ) ( t1 − t0 ) 2 ..70) and (2. of (4.5). In order to prove the equivalence of (4.68) Altogether one can express then the solution +∞ ψ(y.Lebedev (Prentice Hall. a supposition which is not proven here2 . 2 A demonstration of this property can be found in Special Functions and their Applications by N.69) where ∞ φ(y.J.112) and (4.147) we employ the technique of generating functions as in Sect. this is an excellent textbook from which we have borrowed heavily in this Section. t1 ) = n=0 dn φn (y) cn (t) (4. (4. t0 ) which is an element of N1 defined in (4.s.72) × exp( −u − v + 2iyu + 2iy0 v ) . t0 ) φn (y0 ) .225) which allows one to derive a generating function for products of Hermite polynomials which can be applied to the r.N. pp. 1 t = e−iω(t1 −t0 ) (4. t0 ) ψ(y0 .66) One can extend expansion (4..67) for which according to (3.46) the expansion coefficients dn are +∞ dn = −∞ dy0 ψ(y0 .15. y0 . t1 |y0 .

√ πexp( t2 v 2 − 2iytv − y 2 ) (4. t0 ) = and 1 2 i π F (t1 .147) derived by means of the path integral description.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 .77) The sum in (4. (4.4.s.30) to replace y and y0 and that we multiply the propagator by mω/ .78) 1 − e−2iω(t1 −t0 ) = sinω(t1 − t0 ) 2 i e−iω(t1 −t0 ) (4.4: Propagator Carrying out the sum on the r.e. t) √ 1 π(1−t2 ) 2 exp − 1 (y 2 + y0 ) 1 + t2 + 2 y y0 2 1−t Hn (y)e−y /2 Hn (y0 )e−y0 /2 n ∞ √ t n=0 2n n! π 2 2 2 t 1 − t2 = .81) imω 2 sinω(t1 − t0 ) (x2 + x2 ) cosω(t1 − t0 ) − 2 x x0 0 This result agrees with the propagator (2.78) in terms of the coordinates x and x0 .71) of w(y. i.75) Applying (4.77). y0 .70) is. i. The resulting propagator is G(t1 . t0 ) exp i 2 1 2 (y + y0 ) G(t1 . t0 ) (4. t0 ) − i y y0 2 F (t1 . it holds φ(y.. t0 ) = where F (t1 .74) applied once results in w(y.e. (4.79) 1 + e−2iω(t1 −t0 ) cosω(t1 − t0 ) = . one obtains w(y. This requires that we employ (4. identical to the generating function (4. t) = π −3/2 exp +∞ 83 y2 y2 + 0 2 2 +∞ dv exp( −v 2 + 2iy0 v ) × −∞ × −∞ du exp( −u2 − 2u(tv − iy) ) .80) −2iω(t1 −t0 ) sinω(t1 − t0 ) 1−e We can finally express the propagator (4. . y0 . (4.74) a second time yields finally together with the definition (4. t|x0 .h. (4. −∞ (4. t0 ) = i φ(x. by a factor 4 mω/ for both φn (y) and φn (y0 ). t0 ) = exp mω 2iπ sinω(t1 − t0 ) 1 2 × .76) One can express this in terms of the stationary states (4. t) = π −1 exp − 2 y 2 y0 + 2 2 +∞ dv exp(−(1 − t2 ) v 2 − 2i (yt − y0 ) v) . a Re a2 > 0 (4.. t1 |y0 . indeed.73) The incomplete Gaussian integral +∞ −∞ dx e−a 2 x2 √ − 2bx = π b2 /a2 e . y0 .

84 Linear Harmonic Oscillator 4. a+ ] a ˆ a− φ0 (y) ˆ a+ φn (y) ˆ a− φn (y) ˆ a− f |g Ω∞ ˆ We note that these properties imply √ φn (y) = (ˆ+ )n φ0 (y)/ n! . (4.e.85) is convergent everywhere in R.g. d/dy and a+ . (4..86) f (ˆ+ ) = a a ˆ ν! ν=0 .5 Working with Ladder Operators In the last section we have demonstrated the use of differential equation techniques. 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 − . the approaches a ˆ a ˆ using y. For the present there is actually no pressing need to apply such techniques since the techniques ˆ borrowed from the theory of differential equations serve us well in describing harmonic oscillator type quantum systems. 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. We want to introduce now techniques based on the ladder operators a+ and ˆ a− . Here f (ν) (y0 ) denotes the ν-th derivative of f (y) taken at y = y0 . one can express y = 2(ˆ− + a+ ) and d/dy = 2(ˆ− − a+ ) and. e. each corresponding to a single harmonic oscillator of the type studied presently by us.. hence.84) = 1 1 = 0 √ = n + 1 φn+1 (y) √ = n φn−1 (y) = f |ˆ+ g a Ω∞ . ˆ ˆ Calculus of Creation and Annihilation Operators We summarize first the key properties of the operators a+ and a− ˆ ˆ [ˆ− . In this case we define ∞ f (ν) (0) + ν (4.82) a+ = √ ˆ dy dy 2 2 √ √ Obviously. and of the modes of the quantized electromagnetic field. a− = √ ˆ y + . phonons. The reason for introducing the calculus of the operators a+ and a− is that ˆ ˆ this calculus proves to be useful for the description of vibrations in crystals.83) We will encounter below functions of a+ and a− . i. a− must be equivalent. f (ˆ+ ). 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. a (4. both quantum systems are endowed with a large number of modes. the use of generating functions.

The first case is trivial.90) To prove (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+ ). For fn+1 follows then [ˆ− .88) is a fundamental one and will be used in the following. a An important operator function is the exponential function. f (ˆ+ )] + f (ˆ+ ) a− ˆ a a a a ˆ which implies that in order to prove (4.90) holds for fn . a a a 85 (4.90) holds for f (ˆ+ ). i.83) and can be proven for all powers fn by induction and then inferred for all proper functions f (ˆ± ).95) (4. We note ˆ a− f (ˆ+ ) = [ˆ− . f (ˆ+ )] = f (1) (ˆ+ ) . a− f (ˆ+ ) φ0 (y) = f (1) (ˆ+ ) φ0 (y) ˆ a a which follows from a− φo (y) = 0.88) we need to show actually [ˆ− . An example is the so-called shift operator exp(uˆ− ). we will only assume operator functions acting on φ0 (y).90) holds for any function fn (ˆ+ ) = (ˆ+ )n which is a power of a a a+ . a a a We like to state the following property of the functions f (ˆ± ) a f (ˆ− ) φ|ψ = φ|f (ˆ+ ) ψ . a a (4. a a − (4.96) .87. (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) (4.90) we show that (4. It holds a a euˆ f (ˆ+ ) φ0 (y) = f (ˆ+ + u) φ0 (y) .92) f (y) = n=0 dn φn (y) = n=0 (ˆ+ )n a dn √ φ0 (y) n! (4. The convergence of the Taylor expansion ascertains then that (4.4. ˆ a We proceed by induction noticing first that (4. (ˆ+ )n a+ ] = (ˆ+ )n [ˆ− .e. ˆ a a a ˆ In particular.91) (4. Hence. 4. fn+1 (ˆ+ )] a a = [ˆ− .93) one can reduce all operators acting on some proper state function by operators acting on the state φ0 (y). The following important property holds a a− f (ˆ+ ) = f (1) (ˆ+ ) + f (ˆ+ ) a− . a+ ] + [ˆ− .5: Working with Creation and Annihilation Operators and similarly for f (ˆ− ). Note that ˆ a an immediate consequence of (4..89) (4. a a a ˆ 1 a Let us assume that (4. a+ ] = 1 = f1 (ˆ+ ) . f1 (ˆ+ )] = [ˆ− . property (4.94) This identity follows from (4.90) holds for f0 and for f1 . the second case follows from [ˆ− .87) (4.88) (4.88) is (ˆ− )n f (ˆ+ ) φo (y) = f (n) (ˆ+ ) φ0 (y) .

101) ˆ ˆ ˆ C(u) obviously obeys C(0) = 1 This results in D(0) = 1 and.108) . + (4. To express ˆ these operators as products of operators exp(vˆ+ ) and exp(v ∗ a− ) we consider the operator C(u) = a ˆ + ) exp(uz ∗ a− ) where u ∈ R. one obtains a evˆ + − v ∗ a− ˆ a φ0 (y) = e− 2 vv evˆ e− v 1 ∗ + ∗ a− ˆ φ0 (y) (4. + (4. 1 + + uz ∗ a− ). without explicitly stating this.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. the solution of (4. + ∗ + (4.105) a eαˆ + − α∗ a− ˆ φ0 (y) = e αα∗ 2 e−α ∗ a− ˆ a eαˆ φ0 (y) .98) for u = −α∗ and v = α yields e−α This together with (4. hence. + (4.106) yields a eαˆ + ∗ a− ˆ a a eαˆ φ0 (y) = e−αα eαˆ φ0 (y) .86 To prove this property we expand exp(uˆ− ) a ∞ ν=0 Linear Harmonic Oscillator uν − ν (ˆ ) f (ˆ+ ) φ0 (y) = a a ν! ∞ ν=0 uν (ν) + f (ˆ ) φ0 (y) = f (ˆ+ + u) φ0 (y) .104) (4.98) The related operators exp[vˆ+ ± v ∗ a− ] play also an important role. a a ν! (4. We assume in the following a ˆ derivation.96) is applied is a a a euˆ evˆ φ0 (y) = ev (ˆ − + + + u) a φ0 (y) = euv evˆ 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 (4.103) (4. + ∗ a− ˆ a − u euzˆ euz + ∗ a− ˆ = (4.105) e e vˆ+ a + v ∗ a− ˆ φ0 (y) = e − 1 vv ∗ 2 e v ∗ a− ˆ e vˆ+ a φ0 (y) vˆ+ a − v ∗ a− ˆ φ0 (y) = e 1 vv ∗ 2 e −v ∗ a− ˆ e vˆ+ a φ0 (y) .90) and zz ∗ = 1 we can write this d du e uzˆ+ a e uz ∗ a− ˆ a − z ∗ a− .107) − α∗ a− ˆ φ0 (y) = e− αα∗ 2 a eαˆ φ0 (y) . euzˆ ˆ euz ∗ a− ˆ (4.99) ˆ C(u) = zˆ a zˆ+ + z ∗ a− a ˆ + +z a ˆ ∗ − a euzˆ euz ˆ − u C(u) .102) Similarly.101) 1.106) Applying (4. z ∈ C. (4. that the operators considered act on φ0 (y).97) An example in which (4. Below we will use the operator identity which follows for the choice of v = α in (4.

98) and again (4.111) yields π − 4 exp − 1 u2 2 a = euˆ φ0 (0) .84. i.h.110) √ y2 u2 + 2uy − 2 2 a = euˆ φ0 (y) .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. 4.114) where we have employed (4.s.µ=0 uµ v ν √ φµ |φν µ!ν! ∞ = µ=0 uµ v µ µ! (4. Using (4. We start from the generating function (??) and replace according to (4. For this purpose we consider a a a a euˆ φ0 |evˆ φ0 = φ0 |euˆ evˆ φ0 + + − + (4. . We want to derive now the values Hn (0). and ˆ l.95) one obtains a a a a euˆ φ0 |evˆ φ0 = φ0 |euv evˆ φ0 = euv evˆ φ0 |φ0 = euv − + + + − (4.s. the expected from which follows by comparision of each term on both sides φµ |φν orthonormality property. Expanding r.95). Setting y = 0 in (4.112) Expanding both sides of this equation and using (4.5: 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.109) e = π4 n! n=0 √ Using (4.e.h.s.58) one obtains π− 4 1 ∞ ν=0 (−1)ν u2ν = 2ν ν! ∞ ν=0 uν √ φν (0) = ν! ∞ ν=0 uν Hν (0) √ 2ν π ν! (4.111) the orthonormality properties of the wave functions φn (y). is the equivalent of the generating function (??).115) and using (4. of (4. in the ladder operator calculus. and on the r. 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). provide the same values for Hn (0) as provided in Eq.84) yields ∞ ν.h.58) the Hermite polynomials Hn (y) by the eigenstates φn (y) √ ∞ 1 ( 2t)n y2 /2 2yt − t2 √ e φn (y) (4. Similarly.113) Comparision of all terms on the l.116) = δµ ν . we can reproduce by means of the generating function (4..4.h.111) This expression.s. (??).115) a where the latter step follows after expansion of evˆ and using a− φ0 (y) = 0. + (4. (4.

using (4.123) (4. 1. dp (4.2. according to (4. (−i)n ˜ φn (p) = √ 2n n! m˜ ω π 1 4 exp − m˜ p2 ω 2 Hn ( m˜ ω p) . The eigenvalues. In the position representation the wave function is a function of x. .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. 2. (4.121). .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.125) . the momentum and position operators are as stated in (4.118) d2 1 2 + p dp2 2m ˜ ˜ φE (p) = E φE (p) . are ˜ En = or. i. the momentum and position operators are p = p.235). . . n = 0. . is given by the ˜ function ψ(p). In the momentum representation the wave function is a function of p.37).88 Linear Harmonic Oscillator 4.e. for an oscillator in a stationary state φn (x) as given by (2. ˆ and the Hamiltonian (4.120) Multiplying this equation by 1/m2 ω 2 yields − where ˜ E = E / m2 ω 2 . ˜ (4. . the time-independent Schr¨dinger equation for the oscillator can be written o ˆ H = − mω 2 2 2 x = i ˆ d . (4. we o note that the posed eigenvalue problem is formally identical to the Schr¨dinger equation in the o position representation as given by (4. ω = 1 / m2 ω . i.124) ). For this purpose we employ the Schr¨dinger equation in the momentum o representation. of (4.121) ω (n + ˜ 1 ) 2 . The momentum representation eigenfunctions are.. i..119) 2m 2 dp2 Accordingly. 1.3). 2.6) and the Hamiltonian (4.7) is given by (4.e. En = ω (n + 1 ) 2 d2 1 + m˜ 2 p2 ω 2 2m dp 2 2 ˜ ˜˜ φE (p) = E φE (p) (4. using (2. As to be expected. For example.e. (4. the eigenvalues are identical to those determined in the position representation. (4..122) For a solution of the Schr¨dinger equation in the momentum representation.2). The solutions can be stated readily exploiting the earlier results. the probability density is P (x) = |φn (x)|2 .7) is 1 2 m 2 ω 2 d2 p − . n = 0. is given by a function ψ(x).122).235). (4. 4.

. 4. φn (k) = We want to finally apply the relationship 1 ˜ φn (p) = √ 2π +∞ 2 (−i)n −k √ e 2 Hn (k) . of course. ˜ We can now state the probability density Pn (p) for an oscillator with energy En to assume momentum p. From (4.129) follows ky = px/ and.134).131) in the present case employing dimensionless units. 1 ˜ φn (k) = √ 2π +∞ (4.129) m ω one obtains. 2n n! π (4.30) and (4.4.1: Demonstrate the validity of (4. Proceed as follows. α ∈ R.122) (−i)n ˜ φn (p) = √ 2n n! 1 πm ω 1 4 89 exp − p2 2m ω Hn ( 1 p) . i.f. m ω (4. m ω (4. .126) a phase factor (−i)n . this allowed us to introduce in (4.235)] absorbs the replacement dx → dy. defined only up to an arbitrary phase factor eiα .30) to obtain (4. i.58). Defining k = m˜ / p or ω 1 k = p (4. the correctness of the phase factors (−i)n .132) dy exp(−iky) φn (y) −∞ (4.130) dx exp(−ipx/ ) φn (x) −∞ (4..6.30) and (2.e. instead of (4. through direct integration. This implies the property of Hermite polynomials (−i)n e− k2 2 +∞ Hn (k) = −∞ dy exp(−iky) e− y2 2 Hn (y) (4. hence.128) We may also express the eigenstates (4.e.6: Momentum Representation or with (4.126) in terms of dimensionless units following the procedure adopted for the position representation where we employed the substitution (4. (4. the Jacobian. 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 (4.126).133) where we note that a change of the normalization of φn (y) [c.134) Exercise 4.126) Normalized eigenfunctions are.125.127) exp − p2 m ω 2 Hn ( 1 p) .

Inc.135) g(k. a pendulum swinging at small amplitudes.Dupont-Roc.140) An excellent textbook is Photons and Atoms: Introduction to Quantum Electrodynamics by C. For this purpose we assume such state exists and expand ∞ ∞ φ(α) (y) = n=0 d(α) φn (y) . coherent states.134).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. ˆ α ∈ R. J. (4. The answer is ‘yes’.90 (a) Prove first the property 1 f (k.— III. or Glauber states of the harmonic oscillator and they play.C.1 showing in its top row the attributes of such state just discussed. namely through the eigenvalue problem for normalized states a− φ(α) (y) = α φ(α) (y) . We first construct the solution of (4. y) −∞ (4. except that α changes periodically in time. z) = exp 2yz − z 2 − y 2 /2 (4.196) of Hermite polynomials and equate coefficients of equal powers of z to prove (4. 4. The quasi-classical states can be obtained by generalizing the construction of the ground state of the oscillator.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. z) = exp −2ikz + z 2 − k 2 /2 . Such states are referred to as quasi-classical states. 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 .136) (b) Employ the generating function (2. 4. n n=0 d(α) n 2 = 1 (4. 2. and G.4. New York. 1989) which discusses quasi-classical states of the free electromagnetic field in Sect.138).139) where we have added the condition that the states be normalized. 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. The question arises if similar states exist also for the quantum oscillator. 192 3 .194. for example.138) We will show that a harmonic oscillator prepared in such a state will remain forever in such a state. z) = √ 2π where f (k. Inserting this expansion into √ (4.7 Quasi-Classical States of the Harmonic Oscillator A classical harmonic oscillator.Chen-Tannoudji. pp. +∞ Linear Harmonic Oscillator dy exp(−iky) g(z. for example. φ(α) | φ(α) = . The motion of the probability distribution of such state is presented in Fig. carries out a periodic motion described by y(t) = Re y0 eiωt (4.Grynberg (John Wiley & Sons. We want to construct and characterize such states..

150) for α presently real.145) Using the generating function in the form (4.147) √ which identifies this state as a ground state of the oscillator displaced by 2α.4.109). the quasi-classical states are φ (α) (y) = 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. This interpretation justifies the choice of α ∈ R in (4.147) allows one to interprete T (α) as an operator √ which shifts the ground state by 2α.139)] 1 = n=0 d(α) n 2 = |c|2 n=0 α2 2 = |c|2 eα . n! n (4.146) √ 1 − (y − 2 α)2 2 (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) . n! (4.148) The identity (4.145) using (4.e. n! (4. (4. [π]− 4 e− one can write φ(α) (y) = [π]− 4 exp 1 1 y2 2 + √ 2 α y − α2 = exp − α2 2 ∞ n=0 αn √ φn (y) .149) with (4.149) − α∗ a− ˆ (4.144) Normalization requires c = exp(−α2 /2) and. Comparing (4. One can also write (4. hence. .143) for a constant c which is determined through the normalization condition [see (4. n! (4.138).142) The solution is αn d(α) = √ c n n! ∞ ∞ (4.7: Quasi-Classical States or ∞ n=0 91 √ n + 1 dn+1 − α d(α) n (α) φn (y) = 0 .84) φ(α) (y) = exp − α2 2 ∞ n=0 αα∗ (αˆ+ ) a a+ φ0 (y) = e− 2 eαˆ φ0 (y) . However. n n+1 (4. i.

150) for complex α obeys the unitary property T + (α) T (α) = 1 1 (4.67.150) 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 (4.157) (y.109) serves again to simplify this sum   √  y2 1 1 α2 (u2 + 1)   φ(α) (y.s.156) (4. the shift operator leaves the ground state normalized. it holds T + (α) φ0 |T + (α) φ0 = f |T (α) T + (α) g = φ0 |φ0 .s. Following the procedure adopted in determining the propagator of the harmonic oscillator [see (4.159) Noticing the identity which holds for |u|2 = 1 u2 + 1 = (Re u)2 − (Im u)2 + 1 + 2 i Re u Im u = 2 (Re u)2 + 2 i Re u Im u (4. t1 ) = exp or φ (α) − α2 2 ∞ exp n=0 −i ω ( n + ∞ n=0 1 ) ( t1 − t0 ) 2 αn √ φn (y) n! (4.155) (4. t1 ) = π − 4 u 2 exp  − + 2yαu −  2  2 E (4.154) .158) Comparision of this expression with (4. 4. 1. In particular. n = 0.145) shows that α in (4. .153) This is obviously the inverse of T + (α) as defined in (4. .e. (4.152) Since the operator function on the r.151) which follows from a generalization of (4.160) .95) to linear combinations αˆ+ + βˆ− and reads then a a f (αˆ+ + βˆ− ) φ|ψ = φ|f (α∗ a− + β ∗ a+ ) ψ . α ∈ C (4.92 Linear Harmonic Oscillator The shift operator as defined in (4. (4. n! u = e−iω(t1 − t0 ) . one can write in case of T + (α) T + (α) + = T (α) = eα ∗ a− ˆ − αˆ+ a a = e − (αˆ + − α∗ a− ) ˆ . a a ˆ ˆ (4.h. .152) T + (α) f |T + (α) g = f |T (α) T + (α) g = f |g .145) is replaced by a complex number αu which at times t1 = t0 + 2πn/ω. becomes real. t1 ) = u exp 1 2 α2 − 2 (α u)n √ φn (y) . is the adjoint of the operator function on the l.68)] one can write the time-dependent solution with (4..145) as the initial state φ(α) (y. Relationship (4.h. i.

this characterization corresponds closely to that of the possible initial states of a classical oscillator.158) allows one to write 1 φ(α) (y. (4.150) with (4. It is of interest to note that the displacement y0 of the Gaussian as well as the momentum k0 and phase φ0 associated with (4.161) are time-dependent √ y0 (t1 ) = 2 α cosω(t1 − t0 ) √ k0 (t1 ) = − 2 α sinω(t1 − t0 ) 1 φ0 (t1 ) = − α2 sin2ω(t1 − t0 ) (4.138): a real α corresponds to a displacement such that initially the oscillator is at rest. a momentum factor and a phase factor.148. moving solely its center of mass in an oscillator fashion around the minimum of the harmonic potential. 4. The diagram illustrates that the wave function retains its Gaussian shape with constant width for all times. in general.1 (top row) we present the probability distribution of the Glauber state for various instances in time. being displaced around the center with period 2π/ω and. This shows clearly that the Glauber state is a close analogue to the 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.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. experiences a change of its width (relative to . except that it is not pointlike. Our interpretation 2 explains also the meaning of a complex α in (4.161) and is found to describe a displaced Gaussian.147). 4.4. In Fig.150) T + (α u) = exp α ( e−iω(t1 −t0 ) a+ − eiω(t1 −t0 ) a− ) ˆ ˆ This provides a very compact description of the quasi-classical state. obviously. The time-dependent wave function (4. t1 ) = π − 4 exp − [y − y0 (t1 )]2 + iy k0 (t1 ) − iφ0 (t1 ) − ω(t1 − t0 ) 2 2 . t1 ) = u 2 T + (α u) φ0 (y) (4. Comparing (4. 4.138.162) 2 with a periodic change. 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.165) .163) This solution corresponds to the initial state given by (4.163) also through the shift operator. One can express (4. a complex α corresponds to a displaced oscillator which has initially a non-vanishing velocity. (4.164) where according to (4.159) is then 1 1 i φ(α) (y.

t0 ) = n=0 dn φn (y) .169) . 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. 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 . 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. To describe such state which satisfies the initial condition φ(ao . t0 ) = [πσ 2 ]− 4 exp we expand ∞ 1 − (y − ao )2 2σ 2     (4. 1 + σ2 (4. (4.1: Time dependence of Gaussian wave packets in harmonic oscillator potential.166) φ(ao . σ|y.1 (second and third row).168) dy e−p(y−q) Hn (y) where p = 1 + σ2 .167) One can derive for the expansion coefficients +∞ dn = −∞ dy φn (y) φ(ao . This behaviour is illustrated in Fig. σ|y. that of the Glauber states) with a period π/ω. σ|y.

P. p−1 95 (4.177) − 1 ω(t1 − to ) 2 . Prudnikov.168) is4 e−aq/2 (p − 1)n/2 dn = √ Hn (q 2n n!σ p(n+1)/2 p ). namely $750 4 .171) lim p−1 p n 2 Hn (q p ) = lim (2q)n = an . and O.170) In order to check the resulting coefficients we consider the limit σ → 1. t1 ) = √ 2 × exp − 1 (y 2 + yo ) 1+t2 − 2 1−t 2 1 1 √ √ pσ π 1 − t2 × t 1 − t2 (4.175) ∞ tn n=0 2n n! Hn (yo ) e−yo /2 Hn (y) e−y 2 2 /2 The generating function (4. Marichev (Wiley. this 3 volume integral table is likely the most complete today.76) permits us to write this φ(ao .176) 1 √ pσ π a 2 exp − 1 1+σ2 + 1 yo − 1 ω(t1 − to ) 2 2 2 2 × (4. This expansion can be written φ(ao .157) as follows φ(ao . 2 by A. σ→1 p−1 2 (4. 1 a2 2 1 + σ2 + 1 2 2 y o + 2 y yo This integral can be found in Integrals and Series. Yu. σ|y.e.143). 1 (4. t1 ) = e−aq/2 √ pσ ∞ n=0 e −iω(n+ 1 )(t1 −to ) √ 2 1 2n n! p−1 p n/2 (4. 1990).143) of the Glauber state.172) One can then conclude e−a /4 n lim dn = √ a σ→1 2n n! (4.4. σ|y.7: Quasi-Classical States The solution of integral (4. t1 ) = √ × where yo = q p a = √ . p−1 1 − σ4 t = 1 − σ 2 −iω(t1 −to ) e 1 + σ2 (4.. unfortunately very expensive. 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. i. in agreement with the expected result (4. New York. σ|y. a ground state shifted by ao = 2α. In this limit the coefficients dn should√ become identical to the coefficients (4. Brychkov. a worthy successor of the famous Gradshteyn. therefore. σ→1 O(y n−2 ) for n ≥ 2 1 for n = 0.I. in fact. Following the strategy adopted previously one can express the time-dependent state corresponding to the initial condition (4.174) Hn (q p p−1 ) φn (y) .173) which is.166) in analogy to (4. vol.

96 Linear Harmonic Oscillator .

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

k.7) Let us consider briefly an example to illustrate rotational transformations and to interpret the sign of detR(ϑ). (5. the matrix represents a proper rotation. 3 . where ϑ denotes the rotation.6) follows immediatety for the determinant of R(ϑ) using detA B = detA detB and detAT = detA detR(ϑ) = ±1 . We assume the convention that rotations are right-handed. 0.6) (5. are linear and.(5. ϕ)T ) = ±  sinϕ cosϕ 0  (5.e. therefore. x2 . in Cartesian coordinates r = (x1 .98 Properties of Rotations in R Theory of Angular Momentum and Spin Rotational transformations of vectors r ∈ R .e. x3 )T . a minus sign in Eq. A rotation around the x3 -axis by an angle ϕ is described by the matrix   cosϕ − sinϕ 0 R(ϑ = (0. they conserve a · b = 3 aj bj . From (5. . i.. then the fingers of your fist point in the direction of the rotation.7) implies that a rotation is associated with an inversion. can be represented by 3 × 3 matrices R(ϑ).2) Rotations conserve the scalar product between any pair of vectors a and b. (5. ϑ parametrizes the rotations uniquely as long as |ϑ| < 2π. In the following we want to exclude inversions and consider only rotation matrices with detR = 1. 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.5) This equation states the key characteristic of rotation matrices. =1 [R(ϑ)]jk [R(ϑ)]j ak b = j=1 aj bj . (5.e. i. Let us define the rotated vector as r = R(ϑ) r . ϕ)T ) is negative.1) In Cartesian coordinates this reads 3 xk = j=1 [R(ϑ)]kj xj k = 1. (5.8) 0 0 1 In case of a prefactor +1. 0. In the latter case the determinant of R(ϑ = (0. if the thumb in your right hand fist points in the direction of ϑ. in case of a prefactor −1 a rotation and an inversion at the origin. It holds then j=1 3 3 a ·b = j. namely around an axis given by the direction of ϑ and by an angle |ϑ|. (5. it follows 3 [R(ϑ)]jk [R(ϑ)]j = δk j=1 .3) Since this holds for any a and b.

1 T Similarly. there exists a real 3 × 3–matrix −1 R1 which is the inverse of R1 . 3 × 3–matrix.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. R3 = R1 R2 is a real. (5. It holds T T T T R3 R3 = ( R1 R2 )T R1 R2 = R2 R1 R1 R2 = R2 R2 = 1 . (iv) For products of three elements holds the associative law a ◦ (b ◦ c) = (a ◦ b) ◦ c. (5. one can show R3 R3 = 1 Furthermore. R2 ∈ SO(3) and do not write out “◦” explicitly since it represents the matrix product. R RT = RT R = 1 detR = 1 } 1. hence. (iii) ∀a. We need to demonstrate that this inverse belongs also to −1 T SO(3). We have shown altogether that the elements of SO(3) form a group.12) (iv) Since the associative law holds for multiplication of any square matrices this property holds for the elements of SO(3). It follows that R3 is also an element of SO(3).13) . (ii) There exists an element e in G with the property ∀a ∈ G → e ◦ a = a ◦ e = a. forms a group G satisfying the axioms (i) For any pair a.9) forms a group. In the following we will assume R1 . (ii) The group element ‘e’ is played by the 3 × 3 identity matrix 1 1. 99 (5. (i) Obviously. We want to prove now that SO(3) as defined in (5. T R1 R1 −1 = 1 −1 = 1 1 1 (5. a ∈ G → ∃a−1 ∈ G with a ◦ a−1 = a−1 ◦ a = e. 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).9) This set of matrices is closed under matrix multiplication and.11) (iii) From detR1 = 0 follows that R1 is non-singular and. For this purpose we must demonstrate that the group properties (i–iv) hold. We would like to point out the obvious property that for all elements R of SO(3) holds R−1 = RT (5. in fact. it holds 1.5.

14) together with matrix multiplication as the binary operation ‘◦’.15) Below we will construct appropriate matrices Jk . rotation by 180o . 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. Lie Algebra We want to consider now a most convenient. Generators.16) must be antisymmetric. we want to assume that they exist. we arrive at the property exp ϑk LT k from which follows LT = −Lk . M3 . will play a central role in the representation of many other symmetry transformations in Physics. compact representation of R(ϑ) which will be of general use and. (b) the set of matrices {M1 . −1. 0 −1 1 0 (5. J2 . the matrices i Lk = − Jk (5. J3 as follows i R(ϑ) = exp − ϑ · J (5.13) R(ϑk ek )−1 = R(ϑk ek )T = [exp ( ϑk Lk )]T = exp ϑk LT ˆ ˆ k and.20) (5. −1 0 0 −1 . M2 . presently. i.17) .100 Theory of Angular Momentum and Spin Exercise 5. mappings between the sets which conserve the group properties: (a) the set of real and imaginary numbers {1. To demonstrate this property consider rotations with ϑ = ϑk ek where ek ˆ ˆ is a unit vector in the direction of the Cartesian k–axis.1: Test if the following sets together with the stated binary operation ‘◦’ form groups. −i} together with multiplication as the binary operation ‘◦’.18) (5. rotation by 270o } and consecutive execution of rotation as the binary operation ‘◦’. establish if 1-1 homomorphic mappings exist.e.19) (5. We want to show that for the property R RT = 1 to hold. i. This representation expresses rotation matrices in terms of three 3 × 3– matrices J1 . M4 } = 1 0 0 1 . x2 –plane {rotation by Oo . Lie Group. 0 1 −1 0 . in fact. rotation by 90o . due to the uniqueness of the inverse.1. (c) the set of four rotations in the x1 . k = exp ( −ϑk Lk ) (5. identify subgroups.

the real matrix A is anti-symmetric. A must have the form   0 a b A =  −a 0 c  . i. ϑ3 )T ] = lim  sinϑ3 cosϑ3 0  =  ϑ3 ϑ3 →0 ϑ3 →0 0 0 1 0 0 1 Insertion into (5. ϑ3 )T such that for ϑ = (0.15) as the derivatives of R(ϑ) with respect to the Cartesian components ϑk taken at ϑ = (0.        (5. Let us use this definition to evaluate L3 .22) and similar for L2 and L3 . 0. ϑ2 .22). Using the definition of partial derivatives we can state L1 = lim ϑ−1 R(ϑ = (ϑ1 .26) (5. One can determine then the matrices Lk defined in (5.21) −b −c 0 It is this the reason why one can expect that three-dimensional rotation matrices can be expressed through (5. ϑ2 . 0. Exercise 5.15) with three real parameters ϑ1 . (5.8) we can state    cosϑ3 −sinϑ3 0 1 −ϑ3 0 1 0  .1: Matrix Representation of SO(3) 101 We can conclude then that if elements of SO(3) can be expressed as R = eA .e.25)     0 −ϑ3 0 0 −1 0  ϑ3 0 0  =  1 0 0  . 0)T ) − 1 1 1 ϑ1 →0 (5. ϑ3 and three matrices L1 . 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 . lim R[(0.24) (5.1.22) for k = 3 yields L3 = ϑ−1 3 One obtains in this way 0 0 0 i − J1 = L1 =  0 0 −1 0 1 0  0 0 1 i − J2 = L2 =  0 0 0 −1 0 0  0 −1 0 i  1 0 0 − J3 = L3 = 0 0 0 The matrices Lk are called the generators of the group SO(3).5. 0)T . 0)T the rotation is the identity. 0.27) .23) (5. 0.(5. Using (5. L3 which are independent of the rotation angles.2: Determine the generator L1 according to Eq.. L2 . 0 0 0 0 0 0  (5.

15) will be closed under matrix multiplication only in the case that commutators of Lk can be expressed in terms of linear combinations of generators. three being a special case. commutators of commutators of A and B. the Lie algebra of SO(3) can be written [Lk . The question is then. 0)T ) = exp( 1 L1 ).. 0.15) and their generators obey the property (5. (5. An answer can be found considering the Baker-Campbell-Hausdorf expansion ∞ exp( λA ) exp( λB ) = exp( n=1 λn Zn ) (5. L ] = m=1 fk m Lm . 2 . Lie groups can have any number of generators. Finite rotations can be obtained by applications of many infinitesimal rotations of the type R(ϑ = ( 1 . (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. = 2. as we see later. = 2.15) obey the group property.22) implies.. J ] = i k m Jm .g. B]. are related to angular momentum operators. the property (5. i. however. i. only the latter example and . 1 1 e. Of course. B]. if the resulting products of exponential operators can be expressed in the form (5. the matrix (5. The latter property implies that for any ϑ1 and ϑ2 there exists a ϑ3 such that exp i − ϑ1 · J exp i − ϑ2 · J = exp i − ϑ3 · J (5.e. B] ). R(ϑ = (0. 0. in fact. = 2. m = 3 for any odd permutation of k = 1. 0.32) For the matrices Jk = i Lk which. Z3 = 12 ( [A. 0)T ) = exp( 2 L2 ) and R(ϑ = (0. We want to consider. that the generators represent infinitesimal rotations with 1 >> |ϑ| around the x1 -.15). that the operators (5. x2 -.30) is called the Lie algebra. (5. however.e. B]] + [[A. it must hold 3 [Lk . Z2 = 2 [A..g.33) We want to demonstrate now that exp(ϑ · L) yields the known rotational transformations. The expression for Zn are actually rather complicated. From this result one can conclude that expressions of the type (5. L ] = k m Lm (5. [A. and x3 -axes. 3 )T ) = exp( 3 L3 ). m = 3 . m = 3 for any even permutation of k = 1. ϕ)T . and the constants fk m are called the structure constants. holds the algebra [Jk .8) in case of ϑ = (0.31) where k m are the elements which are   0   1 k m =  1   −1 of the totally antisymmetric 3-dimensional tensor.29) where Z1 = A + B and where the remaining operators Zn are commutators of A and B. etc. In fact. In case of the group SO(3) the structure constants are particularly simple.30) Groups with the property that their elements can be expressed like (5.30) are called Lie groups. the elements of if any two indices are identical for k = 1. e.28) The construction of the generators through the limit taken in Eq. (5.

eϕL3 which agrees with (5.38) eA = n=0 (−1)n 2n ϕ (2n)! 1 0 0 1 ∞ + n=0 (−1)n ϕ2n+1 (2n + 1)! 0 −1 1 0 .3: Show that the generators Lk of SO(3) obey the commutation relationship (5. using (5.  cos ϕ − sin ϕ 0 =  sin ϕ cos ϕ 0  0 0 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 . Exercise 5. (5.1.40) . (5.5.36) The property 0 −1 1 0 allows one to write 0 −1 1 0 and. can be written ϕL3 = 0 103 A 0 0 0 0 . A = ϕ 0 −1 1 0 . 0 −1 1 0 = (−1)n 0 −1 1 0 (5.37) = (−1)n 1 0 0 1 .39) Recognizing the Taylor expansions of cos ϕ and sin ϕ one obtains eA = and.35). accordingly. ∞ 2n = − 1 0 0 1 2n+1 (5.41) cos ϕ − sin ϕ sin ϕ cos ϕ (5. (2n + 1)! (5.8).34. 5.31). We note. (5. according to (5.35) To determine expA = ∞ ν ν=1 A /ν! we split its Taylor expansion into even and odd powers 1 A2n + (2n)! 2 ∞ n=0 ∞ eA = n=0 1 A2n+1 . that the matrix ϕL3 .1: Matrix Representation of SO(3) determine exp(ϕL3 ).27).

a relationship which we like to express as follows R(ϑ) = ρ R(ϑ) .47) = ρ R(ϑ1 ) R(ϑ2 ) .104 Theory of Angular Momentum and Spin 5.46) (5. the R(ϑ) ψ(r) = ψ(R(ϑ) r ). the transformation R(ϑ): C∞ ( ) → C∞ ( ) is related to the transformation R(ϑ): R → R . we will learn that transformation patterns and invariances are connected with the angular momentum states of quantum mechanics. hence. complex-valued functions over the 3-dimensional space R which can be differentiated infinitely often. namely R(ϑ)ψ(r ) = ψ(R−1 (ϑ) r ) . Exercise 5. (5.42) Obviously.e.45) (5. its elements can be represented through an exponential R(ϑ) = exp i − ϑ·J (5.2 Function space representation of the group SO(3) In this section we will investigate how rotational transformations act on single particle wavefunctions ψ(r). one isomorphic to SO(3) and.2.44) holds.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) . B ∈ SO(3) holds ρ(A B) = ρ(A) ρ(B) . i. For this purpose we require stringent continuity properties of the wavefunction: the wavefunctions under consideration must be elements of the set C∞ ( ).e. Since this holds for any ψ(r ) one can conclude ρ R(ϑ1 ) ρ R(ϑ2 ) group property (5.1: Assume the definition ρ ρ (A B) = ρ (B) ρ (A).. in fact. i. (5. we define rotations R(ϑ) as linear maps C∞ ( ) → C∞ ( ). for A. (5. In particular. i. = ψ([ R(ϑ1 ) R(ϑ2 ) ]−1 r) = ρ R(ϑ2 ) ψ([R(ϑ1 )]−1 r) (5. Definition We first define how a rotational transformation acts on a wavefunction ψ(r ).48) .43) ρ( ) conserves the group property of SO(3). 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.

49) (note the minus sign of −ϑ1 which originates from the inverse of the rotation in Eq. 0.50) and Taylor expansion yields i − J3 f (r) = = ϑ3 →0 lim ϑ−1 ( f (r) + ϑ3 x2 ∂1 f (r) − ϑ3 x1 ∂2 f (r) − f (r) ) 3 (5.42)) and similarly for J2 and J3 .52) ( x2 ∂1 − x1 ∂2 ) f (r) ..54) (5. J3 ] and [J3 . J2 ].51) Inserting this into (5. (5. (5. k = 1.57) Exercise 5. 0)T ) − 1 1 ϑ1 →0 105 (5. J2 ] f (r) = − = − 2[ 2 [− i J .2: Function Space Representation of SO(3) where the generators can be determined in analogy to Eq.56) holds for [J2 .5.53–5. but also the Lie algebra of the generators. (5. 0. in case of J3 . J ] = i k m Jm . i.e.22) according to i − J1 = lim ϑ−1 R(ϑ = (−ϑ1 . 2 by means of limits as suggested in Eq.2: (a) Derive the generators Jk . 3     1 ϑ3 0 x1 x1 + ϑ3 x2 R(0. 0. (5. ϑ3 )T ) f (r) − f (r) 3 (5. −ϑ3 ) r ≈  −ϑ3 1 0   x2  =  −ϑ3 x1 + x2 .49) to a function f (r).(5. i − J3 f (r) = = Using (5.53) (5.24) one can expand  ϑ3 →0 ϑ3 →0 lim ϑ−1 R(ϑ = (0. it holds [Jk . −ϑ3 ) r ) − 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 . show that (5. One obtains using (5.55) Not only is the group property of SO(3) conserved by ρ( ). J1 ].56) We like to verify this property for [J1 .2. 0.23.  0 0 1 x3 x3 (5. (5.55) and f (r) = f [J1 .49). In fact. . The generators Jk can be determined by applying (5. 5.50) lim ϑ−1 ( f ( R(0. − i J ] f (r) 1 2 =− 2× [ (x3 ∂2 − x2 ∂3 ) (x1 ∂3 f − x3 ∂1 f ) − (x1 ∂3 − x3 ∂1 ) (x3 ∂2 f − x2 ∂3 f ) ] 2 +x1 x3 ∂2 ∂3 f − x1 x2 ∂3 f − x2 ∂1 ∂2 f + x2 x3 ∂1 ∂3 f + x2 ∂1 f 3 2(x ∂ 2 1 2 −x1 ∂2 f − x1 x3 ∂2 ∂3 f + x2 ∂1 ∂2 f + x1 x2 ∂3 f − x2 x3 ∂1 ∂3 f ] 3 = − − x1 ∂2 ) f (r) = i J3 f (r) .

56) one obtains [J 2 .61) simultaneous eigenstates of J 2 and of one of the Jk . J3 ] + [J2 . k = 1. 5. (5.60) commutes with all three Jk .e. 2.53–5.56) are the famous commutation relationships of the quantium mechanical angular momentum operators. . x2 = r sinθ sinφ. usually chosen as J3 . according to the correspondence principle between classical and quantum mechanics. i. C]B and (5. Jk ] = 0 . k = 1. [J 2 . 3. φ).e. J3 ] J2 −i J1 J2 − i J2 J1 + i J2 J1 + i J1 J2 = 0 According to (5. We will show now that the properties of spherical harmonics J2Y J3 Y m (θ. To show this we first note that (5. Using [AB. φ) m (θ. using the momentum operator p = ˆ J = r × i ..58) i = r ×p ˆ (5.63) (5. J3 ] (5.55) can be written as a vector product of r and i − J = −r × From this follows.3 Angular Momentum Operators The generators Jk can be readily recognized as the angular momentum operators of quantum mechanics.64) mY m (θ. identifies J as the quantum mechanical angular momentum operator. J3 ] = = = 2 2 [J1 + J2 . C] + [A. J3 ] + [J1 . J3 ] J1 + J2 [J2 . the operator 2 2 2 J 2 = J1 + J2 + J3 (5. Equation (5. φ) = = 2 ( + 1) Y m (θ. can be found.2. C] = A[B. 3 . These eigenstates are the well-known spherical harmonics Y m (θ. Show that for J3 as defined above holds exp( iα J3 ) ψ(φ) = ψ(φ + α). φ) (5.62) J1 [J1 . φ).3: Consider a wave function ψ(φ) which depends only on the azimutal angle φ of the spherical coordinate system (r. i. θ. The commutation relationships imply that no simultaneous eigenstates of all three Jk exist. However. 2. the system related to the Cartesian coordinates through x1 = r sinθ cosφ. and x3 = r cosθ.106 Theory of Angular Momentum and Spin Exercise 5.61) We demonstrate this property for k = 3. φ) as well as the effect of the so-called raising and lowering operators J± = J1 ± i J2 (5. (5.65) .59) which.

φ) 107 = = 0 = = 0 ( + m + 1)( − m) Y ( + m + 1)( − m) Y m+1 (θ. φ) are a consequence of the Lie algebra (5. i. Noting [L+ . 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) (5. k = 1.71) To prove this theorem we first show by induction that (i) holds. We note L3 L− | = ( [L3 . i. For this purpose we first demonstrate that the property holds for m = − 1 by considering (i) for the state L− | ∼ | − 1 .5. Continuing our proof through induction we assume now that property (i) holds for m.66) (5.67) (5. We consider L+ L− | m = ( [L+ . φ) − m (θ.68) (5. J+ Y J− Y J− Y m (θ.74) (5. From this follows that L− | is an eigenstate of L3 .1 Let Lk . An Important Theorem Theorem 1. in fact.72) = 0 = −i | (5. φ) (θ. For this purpose we prove the following theorem. L− ] = [L1 + iL2 . L1 − iL2 ] = L2 + iL1 = i (L1 − iL2 ) = iL− and L3 | = −i | we obtain L3 L− | = ( iL− + L− L3 )| = i(− + 1) L− | .e.73) (5. φ) m+1 (θ. 3 be operators acting on a Hilbert space H which obey the algebra [Lk . L− ] = [L3 . 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. It holds L+ L− | = (L+ L− − L− L+ + L− L+ )| = ([L+ . We will show that this property holds then also for m − 1. L ] = k m Lm and let L± = L1 ± i L2 .76) We now show that property (ii) also holds for m = − 1 proceeding in a similar way.56). α −1 β −1 = 2 .69) J+ Y (θ. L− ] + L− L3 ) | . namely.70) (5. The arguments are very similar to the ones used above and we can be brief. L2 = L2 + L2 + L2 . L− ] + L− L+ )| .75) (5. The definitions of the coefficients α −1 and β −1 yield L+ L− | = L+ ( iβ −1 | − 1 = − α −1 β −1 | . L1 − iL2 ] = −2iL3 and L+ | = 0 one obtains L+ L− | = −2iL3 | = −2 | .e. L− ] + L− L+ )| m = ( −2iL3 + .. Using [L3 . (∗) (5.3: Angular Momentum Operators on the spherical harmonics.

Before we can finally show that property (iii) holds as well for all − ≤ m ≤ we need to determine the coefficients αm and βm . yield properties (5. i. We note L3 L− | m = ( [L3 .63. in terms of spherical coordinates defined through x1 x2 x3 = = = r sin θ cos φ r sin θ sin φ r cos θ (5. 5.. This implies that (i) holds for m − 1.70.81) i.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 . 5.83) (5.55). finally αm = βm = (5. For this purpose we need to express the angular momentum operators in terms of spherical coordinates (r. φ. A first application will involve the construction of the eigenstates of the angular momentum operators as defined in (5. We will have various opportunities to employ the Theorem just derived.e.. One can normalize the states | m such that αm = βm . We note that we can write L2 = 1 1 1 1 2 2 2 2 L+ L− + 2 L− L+ + L3 .108 Theory of Angular Momentum and Spin L− L+ )| m = (−2m − αm βm )| m .82) (5.66. in particular.e.79) We can now show that (iii) holds for all proper m–values.78) ( + m + 1)( − m) ( + m + 1)( − m) . θ. it also holds for m − 1. Angular Momentum Operators in Spherical Coordinates We want to express now the generators Jk .64).72. 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 . from L+ L− | m = − αm−1 βm−1 | m follows the recursion relationship αm−1 βm−1 = 2m + αm βm .69). We can deduce from (∗) and (∗∗) α α α −1 β −1 −2 β −2 −3 β −3 = 2 = 2( − 1 + ) = 2( − 2 + · · −1 + ) Obviously.53–5.73. (5. Using the familiar formula + m − n k=0 k = n (n + 1)/2 − (m + 1) m = ( + 1) m − (m + 1) m (5. 2 We note that Eqs.84) . 5. given in (5. it holds αm βm = 2 one obtains αm βm = = ( + 1) k=m+1 k . (∗∗) (5.80) (5. Again the argumemts are similar to the ones used above. L− ] + L− L3 )| m = ( iL− − i m L− )| m = −i (m − 1) L− | m from which we can deduce L3 | m − 1 = −i(m − 1)| m − 1 . 5. This completes the proof of the theorem. 5. (5.

82–5.92) (5. cos θ ∂ cos2 φ + sin θ cos φ ∂φ cos2 θ sin θ sin φ ∂ sin θ sin φ ∂ + + sin2 θ f (θ.3: Angular Momentum Operators 109 We first like to demonstrate that the generators Jk . for example. applying repeatedly the chain rule. Using (5. φ) = (x3 ∂2 − x2 ∂3 )f (θ.93) and similarly. L1 f (θ. (5. using ∂/∂tanφ = cos2 φ ∂/∂φ and ∂/∂cosθ = −(1/sinθ) ∂/∂θ. φ) ∂x2 /x1 ∂ ∂x3 /r ∂ = x3 + x3 ∂x2 ∂ tan φ ∂x2 ∂ cos θ ∂x2 /x1 /r ∂ ∂ − x2 − x2 ∂x3 3 ∂ cos θ f (θ.85). In fact. φ) = ( x3 ∂2 − x2 ∂3 ) f (θ. involve only the angular variables θ and φ and not the radius r. φ) sin θ ∂θ sin θ ∂θ L1 f (θ.60). (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 .89) = ∂ x2 x2 ∂ x2 + x2 ∂ x3 − 33 − x2 1 3 2 x1 ∂ tan φ r ∂ cos θ r ∂ cos θ f (θ. L1 f (θ. in terms of spherical coordinates. We like to express now the operators J3 and J 2 .84 ) one obtains. According to (5.91) which agrees with expression (5.90) This yields.87) To derive these properties we consider. actually. (L2 )2 = − cos φ ∂ ∂ + cot θ sin φ ∂θ ∂φ 2 . using (5. the latter defined in (5. using (5. φ) (5.86).85). ∂φ = sin φ (5. φ) ∂x ∂x3 ∂ tan φ =0 (5. φ). 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 ∂ = − .86) (5. φ) = (5.5.87) holds J3 = To determine J 2 we note.85) (5.88) (5.

one can conclude J2 = − 2 sin θ 2 sinθ ∂ ∂θ 2 + ∂2 ∂φ2 .85.87).85.99) = = ∂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. 5. One first identifies the Hilbert space H on which the generators Lk operate. 5. 2 ∂θ ∂θ (5. 1 sin2 θ sinθ ∂ ∂θ (5.4 Angular Momentum Eigenstates According to the theorem on page 107 above the eigenfunctions (5.64) can be constructed as follows. 5. (5. 2m r ∂r2 2m r2 2 (5.100) (5. 5.98) where Jclass = r × mv is the angular momentum and pr = mr is the radial momentum.95) sinθ ∂ ∂θ = ∂2 ∂ + cotθ . 5.110 = cos2 φ Theory of Angular Momentum and Spin ∂2 1 ∂ ∂2 + sinφ cosφ − 2 cotθ sinφ cosφ ∂θ2 ∂φ ∂θ∂φ sin2 θ 2 ∂ ∂ + cotθ sin2 φ + cot2 θ sin2 φ 2 . C∞ (S ).86. ∂θ2 ∂θ ∂φ (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 + .86.97) Kinetic Energy Operator The kinetic energy of a classical particle can be expressed J2 p2 p2 = r + class 2m 2m 2m r2 (5.96) and the relationship between Lk and Jk as given in (5. The functions in C∞ (S ) have real variables . namely on the space of complex-valued functions on the unit 3-dim.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 (5.63.e.87). ∂θ ∂φ ∂2 ∂ ∂2 + cotθ + ( cot2 θ + 1 ) 2 . These generators act on a subspace of C∞ ( ). sphere S2 .97). i. The operators in the present case are Lk = − i Jk where the Jk are differential operators in θ and φ given in (5.

63.103) iteratively for m = − 1.107) (5. .105) ∆( . φ) (5. and then determining the family {Y m (θ. . One obtains in this way Y m (θ. . 0 ≤ θ < π.85–5. must be cyclic in φ with period 2π.102) where dΩ = sinθ dθ dφ is the volume element of S2 . (5.104. 0 ≤ φ < 2π.105) explicitly. is indeed a Hilbert space. m = − . } applying L− Y m+1 (θ.111) . φ) L3 Y (θ. φ) = ∆( . φ) = = 0 −i Y (θ. φ) = 2 + 1 ( − m)! P 4π ( + m)! m (cosθ) eimφ .87). 5. φ) = 0 as well as L3 Y (θ. For this purpose we split off a suitable normalization factor as well as introduce the assumption that the dependence on φ is described by a factor exp(imφ) Y It must hold L+ Y (θ.106) The latter property is obviously satisfied for the chosen φ-dependence. The eigenfunctions (5.109) using (5.5.4: Angular Momentum Eigenstates 111 θ.64) can be constructed then by seeking first functions Y (θ. m) = 1 −m J− 1 2 Y (θ. . − .108) m (θ. the so-called spherical harmonics.102) exists. The function space endowed with this norm and including only functions for which the integral (5. can be expressed L± = Accordingly. (5. The norm in H is defined through |f |2 = S2 dΩ f ∗ (θ. . φ) (5. The raising and lowering operators L± = L1 ± i L2 . φ). 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) in H which satisfy L+ Y (θ. normalizing these functions. (5.107) reads ∂ ∂ + i cotθ ∂θ ∂φ P (cosθ) ei φ i e±iφ ∂ ∂ ± i cotθ ∂θ ∂φ . . 5. φ. φ). φ) = −i Y (θ. . and to be admissable for description of quantum states. (5. . φ) (5. − + 1. − 2. φ) .110) = 0 (5.104) (5. by carrying out the construction according to (5. (5. φ). φ)f (θ.

(5.116) The integral appearing in the last expression can be evaluated by repeated integration by parts. ”Classical Electrodynamics. is P (cos θ) = N sin θ as one can readily verify. (5. = = x (1 − x2 ) 2 x3 3 + 2 −1 +1 −1 −1 dx x2 (1 − x2 ) + 2 · (2 − 2) 1·3 −1 +1 (1 − x2 ) dx x4 (1 − x2 ) −1 −2 +1 2 · (2 − 2) · · · 2 dx x2 1 · 3 · 5 · · · (2 − 1) −1 2 · (2 − 2) · · · 2 2 (2 )! 2 = 2 2 +1 1 · 3 · 5 · · · (2 − 1) 2 + 1 [1 · 3 · 5 · · · (2 − 1)] (5. New York.114) yield |N |2 2 +1 1 2π 4π (2 )! π dθ sin2 0 +1 θ = 1.118) 2 +1 1 1 sin θ ei 4π (2 )! 2 ! φ . i.107). for example. We note that (5.112) where we employed the definition of P (cosθ) in (5..119) provide the following expression for Y Y (θ.119) where the factor (−1) has been included to agree with convention1 .115) holds. (5. π +1 dθ sin2 0 +1 θ = −1 +1 −1 dx (1 − x2 ) +1 = = . 5. A proper solution of this equation.106. . 2nd Ed. one for which π dθ sinθ |P (cosθ)|2 (5.117) Accordingly.113) 0 is finite. 1975) . 5.116) implies |N |2 from which we conclude N = (−1) 1 · 3 · 5 · · · (2 − 1) (5. One obtains. φ) = (−1) 1 1 = 1 [1 · 3 · 5 · · · (2 − 1)]2 (5.” by J.D.114.112 or Theory of Angular Momentum and Spin ∂ − ∂θ cotθ P (cosθ) = 0 (5. . (5.e.106. 5. The normalization factor N is chosen such that π (5. using the new integration variable x = cos θ. Jackson (John Wiley.114) dθ sinθ |Y (cosθ)|2 = 1 0 (5.120) See.

m = − 1. i. and can obtain now.123) P m+1 (x) ∂ x + (m + 1) 1 ∂x (1 − x2 ) 2 .126) Then holds.124) holds for m + 1. φ). 5. the eigenfunctions Y m (θ. through repeated application of (5.5. namely Y (θ. To prove (5. every application of L− reduces the power of eiφ by one. P m+1 (x) = −m−1 m+1 ∂ 1 ( + m + 1)! (1 − x2 )− 2 (x2 − 1) . employing again the variable x = cos θ.124) reads P (x) = = (2 )! (1 − x2 )− 2 (x2 − 1) 2 ! (−1) 1 · 3 · 5 · · · (2 − 1) sin θ (5. 5. For m = (5. according to (5.123).e. φ). (5. describes the φ-dependence of Y m (θ. indeed. 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. The reader should note that the associated Legendre polynomials. − 2. . use (5.124) called the associated Legendre polynomials.4: Angular Momentum Eigenstates 113 We have constructed a normalized solution of (5.103.106) 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 .125) which agrees with (5. P m (x) = −1 × ( + m + 1) ( − m) × −(1 − x2 ) 2 1 (5.119). . we seek to determine the functions P m (cos θ) and.124) we proceed by induction.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) .114. 5. Actually.110) together with (5.. 5. m (cos θ) e (5. ∂x −m−1 .124). 2 ! ( − m − 1)! ∂x −m−1 (5. . Let us then assume that (5. therefore.123) leads to the expression P m (x) = −m m ∂ 1 ( + m)! (1 − x2 )− 2 (x2 − 1) 2 ! ( − m)! ∂x −m (5.127) ∂ −m−1 (x2 − 1) . We want to demonstrate now that the recursion equation (5. are real. as specified in (5. − defined in (5.64).122) The latter identity can be written.106).63. φ).121) This expressions shows that the factor eimφ . (5.103.110).

Such construction reproduces the eigenfunctions Y m (θ.127) reproduces (5. that expression (5. from such construction emerges an important symmetry property of Y m (θ.129) holds. (5.135) .124). φ) = = determines first a normalized solution 0 i f (θ. However. (5.134) (x) = 1 2 ! (1 − x2 ) 2 . φ).106.124) holds for m if it holds for m + 1. In fact. φ) An alternative route to construct the eigenfunctions Y of L− f (θ. therefore. φ) (5. using x = cos θ. etc. we embark on this construction in order to prove (5. We first determine Y − (θ. yields a vanishing expression for m = − − 1 as long as is a non-negative integer.124). φ) choosing the proper sign. hence. it holds then for all m. therefore. therefore. φ) . φ) L3 f (θ. We want to test if the recursion (5. expression (5. 5.130) (5. therefore. Since (5. 2 ! (2 )! ∂x2 1 (5.106.132). by repeated application of the operator L+ .. φ) = 0 (5.132) which reduces the number of spherical harmonics which need to be evaluated independently roughly by half .124) and. and constructs then the eigenfunctions Y − +1 .e. Y − +2 . appears not very interesting. it holds ∂2 (x2 − 1) = (2 )! ∂x2 and. In that case (1 − x2 ) is a polynomial of degree 2 and. 5. φ) = Y − (θ. namely. i.133) Since the term with the highest power of (x2 − 1) is x2 .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 (5.128) one can show that (5. which is equivalent to recursive application of L− .123) terminates for m = − . φ) using (5. Constructing Y − → Y − +1 → ···Y m (θ. Y ∗ (θ. P − (x) = 1 ∂2 (1 − x2 ) 2 (x2 − 1) .131) identifies f (θ. φ) as given in (5. the derivative ∂ 2 +1 /∂x2 +1 of this expression vanishes. φ) = (−1)m Y m −m (θ.124) holds for m = . It holds. P − (5.124) and. if L− Y (θ.

φ) sin θ e−i φ = 0. Obviously. (5.5. φ).124).136) to construct all other Y recursion eiφ ∂ ∂ + i cotθ ∂θ ∂φ Y m (θ.140) Introducing again the variable x = cos θ leads to the recursion equation [c. the expression (5.106. (5. (5. (5.f..143) . one can conclude for the associated Legendre polynomials ( − m)! ( + m)! = or P m+1 (cos θ) ∂ − m cotθ ∂θ P m (cos θ) ( + m + 1)( − m) ( − m − 1)! P ( + m + 1)! m+1 (cos θ) (5. following closely the proof of eq.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 (5.80) and (5. 5. We first note that for m = − (5.142) yields P − (x) = (−1) 1 (1 − x2 )− 2 (x2 − 1) = (1 − x2 ) 2 2 ! 2 ! (5.132).123)] P m+1 (x) = −(1 − x2 ) 2 1 ∂ x − m 1 ∂x (1 − x2 ) 2 P m (x) . a sign consistent with the family of functions Y m constructed above. According to (5.120) for Y obey the postulated relationship (5.138) Employing (5.137) m (θ.136) is normalized and has the proper sign.136) We note in passing that this expression and the expression (5.139) = ∂ − m cotθ ∂θ P m (cos θ) .142) For this purpose we proceed by induction. One can readily verify that this expression provides a solution of (5.110) applies the = ( + m + 1)( − m) Y m+1 (θ. (5. (5. φ) . This follows from the identity ∂ ∂ − i cotθ ∂θ ∂φ One can use (5.4: Angular Momentum Eigenstates Due to (5. 5. (5.124) one arrives at Y (θ.106). i. φ) = 2 +1 1 1 sin θ e−i 4π (2 )! 2 ! φ 115 − .e.129).122.

are identical. We then assume that (5.s. 2 ! ∂x 1 (5.128) m + 1 → −m or.142) and by (5.106) this implies for Y The Legendre Polynomials The functions the identity (5.144) ∂ x − m 1 ∂x (1 − x2 ) 2 +m m ∂ (−1)m (1 − x2 ) 2 (x2 − 1) .149) (5.145) Hence.144) reads P m+1 (x) = +m+1 m+1 ∂ (−1)m+1 (1 − x2 ) 2 (x2 − 1) . According to both (5.142) one can state P (x) = ∂ (x2 − 1) .147) the r. (5.142). Since (5..141) reads then P m+1 (x) = 1 (5.h.124) and (5. Accordingly.150) .142) holds for m. etc. (5. 2 ! ∂x +m+1 (5. Hence. However.124) yields (−1)m ( − m)! P ( + m)! −m (x) = +m m ∂ (−1)m (1 − x2 ) 2 (x2 − 1) .h. also the associated Legendre polynomials determined this way.146) i. equivalently. given by (5.. one notes that application of (5.135). one can conclude P m (x) = (−1)m m (θ.116 Theory of Angular Momentum and Spin wich agrees with (5.e. 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 .148) According to (5. The construction beginning with Y − and continuing with Y − +1 . φ) ( − m)! P ( + m)! −m (x) . 2 ! ∂x +m (5.s. yields the same eigenfunctions as the previous construction beginning with Y and stepping down the series of functions Y −1 . P (x) = P 0 (x) are called Legendre polynomials.e. i. (5. of (5.124).142) holds for m = − we verified that it holds for all m. Y − +2 . (5. etc. Y −2 . 1 (1 − x ) ∂x +m (1 − x2 ) 2 1 (5. of which agrees with the r. 2 ! ∂x +m −(1 − x2 ) 2 Replacing in (5.142) holds for m + 1 if it holds for m.132).

N. 1975). t) = √ Comparision with (5. r1 is the point where the potential is measured..4: Angular Momentum Eigenstates The first few polynomials are P0 (x) = 1 . ”Classical Electrodynamics..158) see. accordingly.D. 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. 92 To prove this property see.150) and (5. 2. and r2 denotes the location of the charge.” by J. m ≥ 0 ∂xm (5. 2nd Ed.151) (5x3 Comparision of (5.154) can be written 1 = w(x. (5. . The generating function allows one to derive useful properties of the Legendre polynomials. In case q = 1 and |r2 | < |r1 | holds the identity2 ∞ 1 r2 = P (cos γ) (5.157) = 0. pp. φ) = 2 + 1 ( − m)! (−1)m sinm θ × 4π ( + m)! × ∂ ∂ cos θ m (5.152) and. P1 (x) = x . New York. t) is called a generating function of the Legendre polynomials3 . Englewood Cliffs.J. 2 3 1 1 = = 2 1−t 1 − 2t + t ∞ t . =0 (5. (5. Using x = cos γ. . pp. The spherical harmonics for m < 0 can be obtained using (5. Jackson (John Wiley.5. 1965) which is an excellent compendium on special functions employed in physics. (5. . .156) yields P (1) = 1 . for example. Lebedev (Prentice-Hall.132).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) .154) |r1 − r2 | r +1 =0 1 where γ is the angle between r1 and r2 . the spherical harmonics for m ≥ 0 Y m (θ. 1. e.156) w(x. =0 (5. For example. N. and |r1 − r2 | = r1 (5. t = r2 /r1 . in case of x = 1 holds w(1.g.153) P (cos θ) eimφ .155) P (x) t . 45 of “Special Functions and their Applications” by N. t) = √ 1 − 2xt + t2 ∞ 1 − 2 x t + t2 . P3 (x) = 1 2 117 P2 (x) = − 3x) 1 2 2 (3x − 1) .

159) Using ∂lnw/∂t = (1/w) ∂w/∂t and multiplying (5.151)] allows one to determine P (x) for = 1. φ) .166) We want to summarize the properties of the spherical harmonics Y derived above. Inspection of (5. π + φ).f.167) (5. . Noting exp[im(π + φ)] = (−1)m exp(imφ) we determine Y Properties of Y m (θ. φ) goes over to Y m (π − θ.156) and (∂/∂t) to P (x)t = ∞ ∂w + (t − x) w = 0 . .165) since ∂ n /∂(−x)n = (−1)n ∂ n /∂xn .162) − (2 + 1)x P (x) + P −1 (x) ] t + =1 [( + 1) P +1 (x) The coefficients for all powers t must vanish individually. φ). π + φ) = (−1) Y m (θ. t) (1 − 2xt + t2 ) Employing (5. holds P1 (x) ( + 1) P +1 (x) = = x P0 (x) (2 + 1) x P (x) − P −1 (x) (5. φ) = = 2 ( + 1) Y m (θ. π + φ). φ) . Due to cos(π − θ) = − cos θ the replacement r → −r alters P m (x) into P m (−x). Accordingly. t) holds Theory of Angular Momentum and Spin ∂lnw(x.161) Collecting coefficients with equal powers t yields 0 = P1 (x) − x P0 (x) ∞ (5. (5. 1.168) mY m (θ. π − θ. then the coordinates of − r are (r. (5.159) by w(1 − 2xt + t2 ) leads to the differential equation obeyed by w(x. The spherical harmonics are eigenfunctions of the angular momentum operators J2Y J3 Y m (θ. ∂t P (x)t −1 −1 (5. φ) m (θ.118 For w(x. φ) (5.142) allows one to conclude P m (−x) = (−1) +m P m (x) (5.160) this differential equation is equivalent ∞ (1 − 2xt + t2 ) =0 P (x) t + (t − x) =0 P (x) t = 0 (5. Inversion Symmetry of Y m (θ. under inversion Y m (θ. θ. using P0 (x) = 1 [c. φ) m (θ. φ) m (π − θ.164) which.163) (5. ∂t 1 − 2xt + t2 (5. If the spherical coordinates of r are (r. . 2. Accordingly. φ) Under inversion at the origin vectors r are replaced by − r. t) x−t = . ..

175) (5.179) .170) (5. φ) C = =0 m=− π C sin θdθ 0 m Y m (θ. in fact. except for the factor exp(imφ). φ) = 2 + 1 ( − m)! (−1)m sinm θ × 4π ( + m)! 2 ! × ∂ ∂ cos θ m (5.174) m ≥ 0 P (cos θ) eimφ . φ) = = ( + m + 1)( − m) Y ( + m + 1)( − m) Y m+1 (θ. uniquely defined functions over the unit sphere S2 which are infinitely often differentiable π 2π sin θdθ 0 0 dφY ∗m (θ. a complete basis of C∞ (S2 ). φ)f (θ. (5. The spherical harmonics obey the recursion relationships J+ Y J− Y m (θ. The spherical harmonics form an orthonormal basis of the space C∞ (S2 ) of normalizable. .177) Note that the spherical harmonics are real. ..) P0 (x) = 1 . φ) Y m (θ.171) m = dφ Y ∗ (θ. 5. The spherical harmonics are given by the formula Y m (θ.178) 7. P +1 (x) P1 (x) = x . φ) .173) m+1 (θ. m 4. for any f (θ.169) 3. The spherical harmonics Y 0 are Y 0 (θ.172) (5. m (5. φ) = (−1)m Y ∗ (θ.4: Angular Momentum Eigenstates 119 2. φ) (5. φ) ∈ C∞ (S2 ) holds ∞ f (θ. where ( = 1. The spherical harmonics for m < 0 are given by Y −m (θ. φ) m (θ. 6.176) = 1 +1 [ (2 + 1) x P (x) − are the Legendre polynomials. For the Legendre polynomials holds the orthogonality property +1 dx P (x) P ( (x) = −1 2 δ 2 +1 . φ) = 2 +1 P (cos θ) . φ) = δ δm m . φ) .5. φ) where J± = J1 ± iJ2 . P −1 (x) ] (5. i.e. (5. 2. φ) 2π 0 (5. . 4π (5. The spherical harmonics form.

s.h.4. π + φ) .3: Construct all spherical harmonics Y3m (θ.187) h(r) Y m (θ. t) stated in (5. φ) = (−1) Y m (π − θ. (5.180) 9. 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. (5.186) (5. 11.179) using the generating function w(x. Exercise 5. =0 t + −1 dxP (x) P (x) ..188) Exercise 5.189).s.182) Y11 Y10 = = − (5. 1.4.87). (5.2: Derive the orthogonality property for the Legendre polynomials (5.h.181) = 0.1: Derive expressions (5.189) evaluate the integral on the l.156). φ) = δm0 2 +1 . For the Laplacian holds 2 = = = (5. For this purpose start from the identity +1 ∞ +1 dx w(x. φ). of (5. 4π (5. t)2 = −1 . The spherical harmonics for m (θ.177). Exercise 5. expand the result in powers of t and equate the resulting powers to those arising on the r. (5.183) (5. . The spherical harmonics obey the inversion symmetry Y 10.86).120 8.184) Y22 Y21 Y20 together with (5. The spherical harmonics for θ = 0 are Y m (θ Theory of Angular Momentum and Spin = 0.4.185) (5. φ) = 1 ∂2 r − r ∂r2 ( + 1) r2 h(r) Y m (θ. φ) .

(5..m m m (θ. . φ) (5.192) = S2 dΩ Y ∗ (θ .. i. .42). These coefficients can be considered the elements of an infinite-dimensional matrix which provides the representation of exp − i ϑ · J in the basis {Y m ..196) C∞ (S ) = = X .195) i. φ) = exp Since the spherical harmonics Y space C∞ (S ) one can expand ˜ f (θ.. The transformations exp − i ϑ · J .5. . m = − . We like to argue that the matrix representing the rotational transformations of the function space C∞ (S ).e.195). if one specifies the functional form of the coefficients [D(ϑ)] m . We denote the image of a function f (θ. (5. m Y m (θ. with elements given by (5. . − + 1. exp − i ϑ · J (X ) = X . .. φ) D(ϑ) = .. . . φ) i − ϑ·J f (θ.194) Y m (θ.e. }] (5. . φ) ˜ Y m (θ. . . Comparision with (5. orthonormal basis for the function c m Y m (θ. = 0.193) where [ { } ] denotes closure of a set by taking all possible linear combinations of the elements of the set.− + .197) The subspaces X have the important property that (i) they are invariant under rotations exp − i ϑ · J . φ) m. φ ) (5...191) i − ϑ·J f (θ . 1. . (5. φ) (5..193) shows ∞ = − . φ).. φ) exp − ϑ · J . = .5 Irreducible Representations We will consider now the effect of the rotational transformations introduced in (5. }] . . φ) . φ) ˜ under such rotational transformation by f (θ. i. }. therefore. exp − i ϑ · J . . . on functions f (θ. are characterized completely if one specifies the transformation of any Y m (θ. φ ) exp m One can also represent C∞ (S ) C∞ (S ) = [ {Y . . For this purpose we consider the subspaces of C∞ (S ) X = [ {Y . m . (5. = .m D(ϑ) m. ..− + .e. m = S2 i dΩ Y ∗m (θ.5: Irreducible Representations 121 5. ˜ f (θ. ∞. assumes a particularly simple form.190) provide a complete. φ) c = . φ) ∈ C∞ (S ). and (ii) they form the lowest dimensional sets X obeying C∞ (S ) = X which have this invariance property. ∞. = − . called the irreducible representation..

J± ] = ± J± (5. One proceeds by expanding first ϑ · J in terms of J+ . produces 0. one needs to consider the action of J+ 1 J− 2 J3 3 on the subspaces X .202) D(ϑ) =           (2 + 1) × (2 + 1)               . with blocks of dimension 1.199) [J+ .     (5. which follow readily from (5.65).e.200) in (5. the transformation exp − i ϑ · J leaves X invariant as well. If one orders the basis according to the partitoning (5.198) 2 2 where ϑ± = ϑ1 ( 1 ϑ− J+ 2 + 1 2 ϑ+ J− ϑ2 .195) [D(ϑ)] m. .194) only spherical harmonics with or. one concludes for the matrix elements (5.e. in case m = − . J− 2 also leaves n1 X invariant.n3 n n n Accordingly. 5. φ) . one can conclude that J+ leaves X invariant.122 Theory of Angular Momentum and Spin The invariance of the subspaces X under rotations follows from a Taylor expansion of the exponential operator exp − i ϑ · J .. The exponential operator exp − i ϑ · J + ϑ3 J3 )n . Since any additive term n n n cn1 . and J3 1 1 ϑ · J = ϑ− J+ + ϑ+ J− + ϑ3 J3 (5.   .. i. φ) exp m i − ϑ·J Y m (θ. J− ] = 2 J3 .   1×1         3×3           . m = δ S2 dΩ Y ∗ (θ. (5.n2 .200) of the rotational transformations one can also conclude that the X are the smallest invariant subspaces of C∞ (S ).. . 5.200) leaves X invariant. . (5. leaves X invariant. n The action of J− on Y m produces Y m−1 or. 3.. can be expanded in powers One employs then the commutation properties [J3 .n3 J+ 1 J− 2 J3 3 . i. .. Since the Y m are n3 eigenfunctions of J3 the action of J3 reproduces the basis functions of X .n2 . J− .197) of C∞ (S ) the matrix representation of D(ϑ) assumes a block-diagonal form. .201) From expression (5. Similarly..199) allows one to express exp i − ϑ·J = n1 . equivalently.e.56. i. Application of (5. One = contribute can conclude then that in the expansion (5.

other representations are termed reducible representations. f (α + 2π) = 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). Exercise 5. 5.5. for which holds RT R = R RT = 1 1. does not provide the simplest mathematical description of rotations. though seemingly natural.22. The representation is referred to as the irreducible representation. A ∈ C. i.49). This step reconsiders the parametrization of rotations by the vector ϑ assumed sofar. . (d) Show that the similarity transformation T (ϕ) defined in the space of non-singular. 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). ∈ N} invariant.6 Wigner Rotation Matrices We will now take an important first step to determine the functional form of the matrix elements of D(ϑ). this parametrization.e.5. 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). i. However. Determine the generator of ρ(ϕ) in analogy to (5. 5. is a group. k = 1.1: Representations of the group SO(2) SO(2) is the set of all 2×2 matrices R which are orthogonal. (iii) a third rotation around the new x3 –axis by an angle γ. (ii) a second rotation around the new x2 –axis by an angle β. 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. 2. Give the corresponding expression of ρ(ϕ). The three components ϑk . (a) Show that SO(2) with the group operation ◦ defined as matrix multiplication.6: Wigner Rotation Matrices 123 This representation of D(ϑ) is the one wich has blocks of the lowest dimensions possible. and for which detR = 1.e. and periodic functions f (α). (f) The transformation ρ(ϕ) as defined in (e) leaves the following subspace of functions considered in (e) Xm = { f (α) = A e−imα . (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. 3 certainly allow one to describe any rotation around the origin.

φ) e− γJ3 − i βJ2 − i αJ3 e e . i.h. can then be written e− i βJ2 − i αJ3 e = e− i αJ3 e− i βJ2 (5.195) by ˜ Y m (θ. we can write e− i −1 = S eA S −1 . γ will be referred to as Euler angles.203).211) Using (5.204) is satisfied. φ) (5. γ)] = . i.e.207) are equivalent.h.210) The third rotation in (5.. β. one can replace (5.s. the axis x3 is defined in the coordinate frame which is related to the original frame by the consecutive rotations (i) and (ii).203).203) can be expressed.m m. and the r. however. φ) [D(α. For this purpose we notice that J2 can be expressed through the similarity transformation J2 = e− i αJ3 J2 e i αJ3 (5.209) 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. i. followed by transformation (i).203) has the disadvantage that it employs rotations defined with respect to three different frames of reference. in terms of rotations defined with respect to the original frame. φ) = S2 dΩ Y ∗m (θ. i (5.194.124 Theory of Angular Momentum and Spin The angles α..e. the transformation from the rotated frame to the original frame. β.s. of (5.207) which replaces J2 by the inverse transformation (i).212) . We will demonstrate that (5.209) The first two rotations in (5. Obviously. The Euler rotation replaces exp − i ϑ · J by e− i γJ3 − i βJ2 − i αJ3 e e . is e− i γJ3 = e− i βJ2 e− i αJ3 e− i γJ3 e i αJ3 e i βJ2 (5. The axis x2 is defined in the coordinate frame which is related to the original frame by rotation (i). m i Y m (θ.. followed by J2 in the original frame.203) For any ϑ ∈ R one can find Euler angles α. (5. β.205) Y m (θ. (5. β.e. 5.206) The expression (5. γ)] m. the transformation from the original frame to the rotated frame. m [D(α. Accordingly. For any similarity transformation involving operators A and S holds eS A S Accordingly. (i) and (ii). in analogy to (5. γ ∈ R such that exp i − ϑ·J = e− i γJ3 − i βJ2 − i αJ3 e e (5.208) αJ3 βJ2 = e− αJ3 e− i βJ2 e i (5.209). the l.

206) by ˜ Y m (θ. The eigenvalue property (5. m [D(α. φ) e− m i βJ2 Y m (θ. to redefine the three rotations in (5. β. (5. 5. φ) Y m (θ.310)] an explicit expression for the Wigner rotation matrix (5. m i Y m (θ. 5. 5. φ) e− m S2 i αJ3 f (θ.219) = e−iαm dm m (β) e−iγm δ ( ) . φ) (5.e. m (5. φ) f (θ. φ) e− αJ3 e− βJ2 Y m (θ.216) Using. This allows one to express the rotational trasnformation (5. the electron e and the electron neutrino νe of the first generation.64) yields dΩ f (θ. φ) e−iγm δ (5. φ) [D(α. (5. γ)] m. (5.217) Accordingly.7: Spin 1 2 and the group SU(2) 125 Multiplication from the left with (5.218) Defining the so-called Wigner rotation matrix dmm (β) = one can express the rotation matrices [D(α. one can write [D(α. φ) . m = e−iαm S2 dΩ Y ∗ (θ. Best known is the spin of the electron.309. in addition. the muon µ and its neutrino νµ of the second .220) We will derive below [see Eqs. Examples are the other five members of the lepton family to which the electron belongs.7 Spin 1 2 and the group SU(2) The spin describes a basic and fascinating property of matter. γ)] m.214) e− γJ3 = S2 dΩ Y ∗m (θ.215) The evaluation of the matrix elements (5. (5. φ) i i (5.219). the self-adjointness of the operator J3 one can state dΩ Y ∗ (θ. γ)] = . φ) . φ) e− m i βJ2 Y m (θ. φ) e− S2 i γJ3 Y m (θ..205.203) with respect to the original (unprimed) frame one simply needs to reverse the order of the rotations. φ) = e−im α S2 dΩ Y ∗ (θ. β. γ)] m.m m.5. φ) e−imγ . but many other elementary components of matter are endowed with spin-like properties.213) i. β. β. φ) = S2 dΩ f (θ. (5.215) benefits from the choice of Euler angles for the parametrization of rotational transformations.210) yields the simple result e− i γJ3 − i βJ2 − i αJ3 e e = e− i αJ3 e− i βJ2 e− i γJ3 . m ( ) S2 dΩ Y ∗ (θ.

221) One can show readily that this set forms a group with the groups binary operation being matrix multiplication. and which carry spin 1. each component Sk . like superconductivity and magnetism. two of which (in certain linear combinations) make up the vector mesons which carry spin 1. as long as |c+ |2 + |c− |2 = 1.3 . The important feature is that all U ∈ SU(2) can be parametrized by an exponential operator U = exp − i ϑ · S (5. the quarks. there are together five conditions in terms of real quantities to be met by the matrix elements and. So do the three generations of six quarks. the spin of the electron is likely the most important property in Chemistry.e. [Sk ]mn = ([Sk ]nm )∗ . one for each matrix element of U † U . the degrees of freedom of the matrices U are three real quantities. j. then allowed + − symmetry transformations of spin states are described by 2 × 2-matrices U with complex-valued matrix elements Ujk . One can show that the unitarity condition requires these matrices to be hermitian.222) where the vector S has three components.1 systems 2 2 can be related to two states which we denote by χ+ and χ− . there is nothing more elementary to matter than the spin property. It appears to be rather impossible for a Physicist not to be enamored with the spin property. We will find that the spin in its transformation behaviour is closely related to angular momentum states. hence. 3 representing a 2 × 2 matrix. This specification implies that we consider transformations save for overall factors eiφ since such factors are known not to affect any observable properties. Such systems can also assume any linear combination c+ χ+ + c− χ− . i. 2. and because of detU = 1. Conservation of the scalar product under symmetry transformations requires ∗ the property U U † = U † U = 1 where U † denotes the adjoint matrix with elements U † jk = Ukj . have properties beyond those of single spins. carry other spin-like properties which together.126 Theory of Angular Momentum and Spin generation. If we identify the states χ+ and χ− with the basis of a Hilbert space. eight real numbers to specify the matrix elements. c± ∈ C. k = 1. atoms and molecules. We may finally mention that the spin is at the heart of many properties of condensed matter systems. the tau τ and its neutrino ντ of the third generation and their antiparticles carry a 1 spin. k = 1. The transformation matrices are then elements of the set SU (2) = { complex 2 × 2 matrices U . leaving its imprint on the properties of nuclei. (5. in principle. the photon. 2. in fact. How can the elements of SU(2) be parametrized. e. Because of the unitarity condition U † U = 1 which are really four equations in terms of real 1 quantities. There . particles which mediate the strong and the electro-weak interactions. There are also the mediators. the two W± and the 2 2 Zo . 1 We will specify for the transformations considered detU = 1.2 . the gluon. Spin. however. We will consider first only the so-called spin– 1 . U U † = U † U = 1 detU = 1 } 1. and the condition detU = 1 requires the Sk to have vanishing trace. The particles mentioned.g. in which we define the scalar product between any state |1 = a+ χ+ + a− χ− and |2 = b+ χ+ + b− χ− as 1|2 = a∗ b+ + a∗ b− . The presence of this property permeates matter also at larger scales than those of the elementary particles mentioned. and three of which make up the baryons which carry spin.1 and spin. a relationship. which might be the reason why consideration of rotational symmetry is so often fruitful in the study of matter. four real and four imaginary parts of Ujk . As complex 2 × 2 matrices one needs. In any case. generalizing then further below.

A = x σ 1 + y σ 2 + z σ3 . b are vectors commuting with σ. In fact. they are associated with the important fermion property of matter.231) . According to this property the Pauli matrices generate a 3dimensional Clifford algebra C3 .7: Spin 1 2 and the group SU(2) 127 exist three such linear independent matrices. One can readily verify the commutation property σj σk − σk σj = [σj .228) (5. (σ · a) (σ · b) = a · b 1 + iσ · (a × b) 1 (5. 2 S2 = 1 σ2 .229) and avoids commuting the components aj and bk . σk ]+ = 2δjk 1 . σk ]− = 2i jk z x − iy x + iy −z . (5. (5. (5.k=1 σ σ aj bk + j>k aj bk = 3 j.224) Algebraic Properties of the Pauli Matrices The Pauli matrices (5. The Pauli matrices obey the following anti-commutation properties σj σk + σk σj = [σj . it holds. Clifford algebras play an important role in the mathematical structure of physics.227. 3 are the Pauli matrices σ1 = 0 1 1 0 . 2 S3 = 1 σ3 2 (5. 1 σj σk = − σk σj for j = k which can also be readily verified.g. At this point we will state a useful property. The proof of this relation rests on the commutation relationship and the anti-commutation relationship (5.226) The Pauli matrices satisfy very special commutation and anti-commutation relationships. y. 2. σ2 = 0 −i i 0 . hermitian 2 × 2–matrices A can be expandend. σj 2 (5. (5.31) of the group SO(3). namely. Any such A can be expressed in terms of three real parameters x.227) which is essentially identical to the Lie algebra (5. y. σ3 = 1 0 0 −1 . z.k=1 j<k + σ j σ k aj bk . 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.229) = 1 .223) where σk . x. ∈ R . e. one obtains (σ · a) (σ · b) = 3 j 2 j=1 (σ ) aj bj 3 j k j. 5. z A = In fact. but their comonents must not necessarily commute with each other.e.k=1 σ σ 3 j k j.224) provide a basis in terms of which all traceless.230) where a.e.. i.225) σ (5.5. the simplest choice being S1 = 1 σ1 . it might hold aj bk − bk aj = 0. 1 i.

through the algebra obeyed by the operators Sk [Sk . S ] = i k m Sm (5.235) which is identical to that of the generators of SO(3).h.k=1 j>k σ j σ k (aj bk − ak bj ) − 1 2 3 j k j. This relationship emerges. however. For this purpose we choose to replace (5. S3 χ± = ± 1 χ± 2 .8 Generators and Rotation Matrices of SU(2) Sofar there is nothing which relates the transformations U . 1 (5. (5.231) proves (5. We like to derive this result now by evaluating the transformations of SU(2) explicitly.e.s. i. (5. In the special case a = b ∈ R holds (σ · a)2 = a 2 1 .k=1 j>k i σl (aj bk − ak bj ) = i =1 σ a×b . 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. spin changes sign. (5.222).238) S 2 χ± = . 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 .232) − σ k σ j ) (aj bk − ak bj ) . however.k=1 j>k 1 2 3 j. to rotations in space. hence.k=1 (σ σ j>k σ k σ j (aj bk − ak bj ) = (5. only a 720o rotation leaves the spin invariant.236) The matrix elements of this 2 × 2 operator can be labelled by the basis states χ+ and χ− . 1 1 1 ( + 1) χ± 2 2 .128 Theory of Angular Momentum and Spin Using (σ j )2 = 1 the first term on the r.222) by the Euler form exp(−iγS3 ) exp(−iβS2 ) exp(−iαS3 ) . The commutation property (5. since there is a slight difference: when you interprete the 3-component ϑ in (5. yields a · b. Noticing the idempotence of the Sk ’s 2 Sk = 3 one obtains S 2 = 4 1 = 1 1 1 2(2 1 4 1 0 0 1 (5.230). let say around the x2 –axis by 360o .233) This result together with (5. The two remaining terms yield using σ j σ k = −σ k σ j for j = k and altering ‘dummy’ summation indices 3 j. Well almost like it.227) leads to 3 3 jk j.234) 5.k=1 j>k σ j σ k (aj bk − ak bj ) = 1 2 3 j. (5.222) as a rotation vector and you rotate once.237) + 1)1 and.

244) i.239) where we made use of the notation (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. − + 1. (5. 1.241) 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 (5.9: 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). .243) We note the property of this rotation matrix 2 dmm (2π) (1) = −1 1 . use the label 2 2 | 1 ± 1 for the states χ± . (5. γ)] m.9 Constructing Spin States with Larger Quantum Numbers Through Spinor Operators In this section we like to demonstrate. following Jourdan and Schwinger. The complete matrix elements (5. 3 .e.e. .219).239 ) one expands the exponential operator exp(−iβS2 ).220) are [D(α.. namely. In order to determine the Wigner matrix in (5. identical particles any number of which can exist in the same state χ+ and χ− .5. 1 . 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. .] can be identified with the Taylor expansion of the cos.5. β. (5.239) in the notation (5.245) 5. m = cos β e−i( 2 + 2 2 γ α sin β ei(− 2 + 2 2 α γ − sin β ei( 2 − 2 2 α γ cos β ei( 2 + 2 2 α γ . 2. .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 . m = − . i. The idempotence property of Sk yields particularly simple expressions for these powers. . 2 2 We obtain with this notation for the transformations (5.240) . ( − 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. .219. i. . rotation by 360o changes the sign of the spin state. .e.242) The expressions in brackets [. no two such particles can exist in the same state. For this purpose one needs to determine the powers of −iβS2 . therefore. . that states | m for higher quantum numbers = 0. (5. 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. We may.

The boson character of the operators b† = (b† . ζ = +. b ζ = 0 b ζ. m) † and bζ obey the commutation (5. −j + 1. b− † = − sin β † β b + cos b† . b† |Ψ0 = χ+ + .246) The corresponding adjoint operators are denoted by b+ and b− . m) = b† − j−m (j + m)! (j − m)! |Ψ0 . b± |Ψ0 = 0 (5. . . b † ζ For the vacuum state |Ψ0 holds = δζζ . .5.251) One can show using this property and (5. b− )T is expressed through the commutation relationships + − (ζ. i. j = 0. 3 . b† |Ψ0 = χ− − . . 2 + 2 − † (5. . 1. b− )T as spinor creation and annihilation + − operators. 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. (5.252) (5. using (5. b† )T and b = (b+ .249) In the following will refer to b† = (b† . These operators are associated with a given spatial reference system. 2 2 .254) We will show below that these states are orthonormal and form spin states with higher quantum numbers. b ζ = b† . i.253) = δζζ . −) bζ . b† )T and b = (b+ .248) bζ .e.247. (5.247) (5. therefore. m = −j. j. . x∗ b = x∗ b+ + x∗ b− + − (5. b † ζ ζ =0 (5. . b ζ = b ζ.243) the creation operators in a coordinate system rotated by an angle β around the y–axis are b+ † = cos β β † b + sin b† 2 + 2 − .e.250) which for x∗ x = x∗ x+ + x∗ x− = 1 represent spinor operators in an arbitrary reference system. We consider. 1 .130 Theory of Angular Momentum and Spin Definition of Spinor Creation and Annihilation Operators We present the states χ± through creation operators b† which when applied to a formal vacuum ± state |Ψ0 generate χ± .248) that bζ relationships † † b ζ. also operators of the type x b† = x+ b† + x− b† + − . + − For example. b ζ The States |Ψ(j.

We will show that for this commutator holds n b+ .255) is b+ . f (b† ) |Ψ0 .248) we obtain b+ .258) Equations (5. An example + − is the monomial b† which generates the states (5. in b† and b† . (5. possibly infinite power series. we need to evaluate b+ . + + (5.260) . b † + = n b† + n−1 b† + b † + + n = (n + 1) b† + .5. b† + = n b† + n−1 . b† + n b† + b† + b† + + n n n b+ .259) The property is obviously true for n = 0.247. (5. + b+ .248) and from (5.256) In the special case |state = |Ψ0 this reads using (5. b† + n = n cn b+ .257) motivate us to determine the commutator (5. b† + b+ . .257) which for (5. b† + n . 5. f (b† ) + (5. b† + n+1 = = b+ . f b† Obviously. 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.249). For this purpose we consider the following polynomial of creation operators N j+m j−m f (b† ) = + n=0 cn b† + n . (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+ . that the annihilation operators play a role similar to the differential operator in calculus. If it is true for n then using (5. f (b† ) |state + f (b† ) b+ |state .256) and (5. We will derive the result. b† + − well-known for the quantum mechanical harmonic oscillator.10: Algebraic Properties of Spinor Operators 131 5.254).249) b+ f (b† ) |Ψ0 = + 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† ) in b† and b† holds (ζ = +.267) exp xb† |Ψ0 is called a generating function of |Ψ(j..263) corresponds to the product rule of calculus ∂k f (x)g(x) = (∂k f (x))g(x) + f (x)(∂k g(x)) which can be written ( ∂k f (x) − f (x)∂k ) g(x) = [∂k . (5.259) holds for all n ∈ N.266) we compare the terms x+ b† + j+m φjm (x) Ψ (j. Equation (5.. one can prove b− . b† ) |Ψ0 = + − ∂ ∂b† ζ f (b† . b† ) . b† ) + − = ∂ ∂b† ζ f (b† .262) Because of the commutation relationships (5.1. f (x)] = ∂k f (x) . m) m=−j (5.264) This demonstrates the equivalence of the spinor operators bζ and the derivative operation. One can finally conclude for polynomials (5. + − (5. 2 . m) = x− b† − j−m (j + m)! (j − m)! |Ψ0 . (5.261) Similarly. In order to derive (5. the property holds also for n + 1. By induction we can conclude that (5.250) and where φjm (x) represents the function of the two variables x+ and x− closely related to |Ψ(j. b† ) |Ψ0 . m) . = f (b† ) + (5.265) Generating Function of the States |Ψ(j.268) . m) φjm (x) = j−m xj+m x− + (j + m)! (j − m)! . f (b† ) + where f (x) = df dx . m) We want to prove now the property ∞ j exp xb † |Ψ0 = 1 j=0. −) + − + − bζ . (5. f (b† ..e.266) where xb† has been defined in (5.255) b+ . φjm (x) Ψ (j. f (x)] g(x) = (∂k f (x))g(x) or [∂k .263) According to (5.132 Theory of Angular Momentum and Spin i.257) we can conclude in particular bζ f (b† . f (b† ) − = f (b† ) − (5.247) the more general property for polynomials f (b† .

1 ... (5.. . 1. . chosing the summation index j = 0.270) s x+ b† + N −s x− b† − s |Ψ0 . (2j)! 2j |Ψ0 (5. 2. 2 2 leads to ∞ j ∞ φjm (x) |Ψ (j.1.272) The summation over s can be written in terms of j and m using (5. 2 ∞ 1 xb† 2j! u 2j |Ψ0 = u=0. m) = = |Ψ0 (N − s)! s! N −s 1 N! x+ b† x− b† + − N ! (N − s)!s! N −s 133 (5.274) Summing this expression over 2j = 0. 1.1. 2 .1. 1 ... . m) m=−j = = 1 x+ b† + x− b† + − (2j)! 2j 1 xb† |Ψ0 ..5.e. . 3 . 1 xb† u! |Ψ0 (5. m) 1 j=0. one obtains φjm (x) Ψ (j.270) N 2j j → s=0 j−m=0 → m=−j ...275) = exp xb† |Ψ0 . m=−j = j=0. s! (n − s)! Defining j+m = N −s j−m = s. .10: Algebraic Properties of Spinor Operators with the s–th term in the binomial expansion of (a + b)n n! an−s bs .. i. .273) Change of the summation indices allows one to conclude j φjm (x) Ψ (j. 2.269) (5.. (5.271) Summation over s from s = 0 to s = N yields 1 N! = = N s=0 N! (N − s)!s! x+ b† + x− b† − s |Ψ0 N 1 x+ b† + x− b† |Ψ0 + − N! 2j 1 x+ b† + x− b† |Ψ0 + − (2j)! (5..

s. [b† . b† ] = 0 allow one to state that for + − + any polynomial f (b+ .m φjm (x∗ ) φj m (y) Ψ (j. y− )e(yb ) |Ψ0 (5. of (5.134 Theory of Angular Momentum and Spin which concludes our derivation.279) One can generalize this to † † s (bζ )s e(yb ) |Ψ0 = yζ e(yb ) |Ψ0 . hence.281) and. m) |Ψ j . m .j . m) We want to show now that the states (5. The generating function allows one to derive various properties of the states |Ψ (j. i. m) and will be used for this purpose below.277) To evaluate this expression we first notice † e(xb ) † = e(x ∗ b) . we can conclude † † ∗ e(xb ) Ψ0 |e(yb ) Ψ0 = e(x y) (5. b− ] = 0. m) |Ψ j .278) † ∗ which allows us to replace the l.249) for any non-vanishing integer s holds + − Ψ0 |bs Ψ0 = 0 and. (5. According to (5.254) are orthonormal. b− )e(yb ) |Ψ0 = f (y+ . b− ) holds † † f (b+ . m holds.276) = j. (5. (5.264) implies † † bζ e(yb ) |Ψ0 = yζ e(yb ) |Ψ0 .m.283) (5. (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 (5.280) The commutation properties [b+ .284) . therefore.e. that Ψ (j. For this purpose we consider the inner product † † e(xb ) Ψ0 |e(yb ) Ψ0 = δjj δmm (5.h.282) where we defined x∗ y = x∗ y+ + x∗ y− . b† ] = 0 and [b− . Orthonormality of the States |Ψ (j. In order to evaluate the ∗ b) (yb† ) operator e(x e we notice that the derivative property (5.277) by Ψ0 |e(x b) e(yb ) Ψ0 .

.287) ζ 2 ζ. States 2 We want to demonstrate now that the states |Ψ (j.10: Algebraic Properties of Spinor Operators and comparing (5.224).m.289) < n |σi | m >< n |σj | m > {b† bm . (5. Jj ] = i This property is derived as follows [Ji .5.276).n . 2. (5. 2 . etc. (5. bm bm } n n = 1 4 < n |σi | m >< n |σj | m > {b† bm δmn − b† bm δm n } n n n. m) as given in (5.n . b† bm n n n.j .m +b† b† .m 135 = e(x ∗ y) . k = 1.268–5. 3 denote the Pauli spin matrices (5.223. 1.m.. m) |Ψ j .277) φjm (x∗ ) φj m (y) Ψ (j.m. − 2 + (5.m.275) one can show e(x ∗ y) = j. (5. m j. 1 New Representation of Spin 0.m ijk Jk . The 2 operators are explicitly. One defines corresponding operators Jk through 1 Jk = b† < ζ |σk | ζ > bζ . 3 .ζ where σk . b† bm n n n. J1 = J2 = J3 = 1 † b b− + b† b+ − 2 + 1 † b b− − b† b+ − 2i + 1 † b b+ − b† b− .224).288) The three operators obey the Lie algebra of SU(2) [Ji .286) from which follows immediately the orthonormality property (5. . In our present notation the matrix elements < ζ |σk | ζ > for ζ = ± correspond to the matrix elements < ζ |σk | ζ > for σ = ± 1 .m . using < + |σ1 | + > = < − |σ1 | − > = 0. Jj ] = = 1 4 1 4 < n |σi | m >< n |σj | m > b† bm . < + |σ1 | − > = < − |σ1 | + > = 1.254) 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. (5.m φjm (x∗ ) φjm (y) .285) Following steps similar to those in Eqs. .n . 5.282) and (5.

297) .k ijk ijk σk | m > b † bm n = i = i 1 2 < n |σk | m > b† bm n n. (5. (5.m < n |[σi .294) .292) The commutation relationships (5.m < n |2i n.m. To this end we note 2 2 2 J 2 = J1 + J2 + J3 = 1 1 2 J+ J− + J− J+ + J3 2 2 . m) are eigenstates of J 2 and J3 . J1 ] + i [J1 . − (5. J+ ] = [J1 − iJ2 .288) yields first the expression J2 = which results in J2 = 1 4 1 † ( b + b + − b † − b− ) ( b † + b+ − b† − b − − 2 ) + b † + b − b † − b + 4 .296) ( b † + b + b † + b+ + b † + b + b † − b − − 2 b † + b + − b † − b − b † + b + + b † − b − b † − b − + 2 b † − b − + 4 b † + b − b† − b + ) (5. 2 J 2 = J+ J− + J3 − J3 = J3 (J3 − 1) + J+ J− (5. m) as Eigenstates of J 2 and J3 We wish to show now that the states |Ψ (j.293) .m.m. J1 + iJ2 ] = −i [J2 .m 1 4 < n |σj | m >< m |σi | m > b† bm n n .295) The last result together with (5. J2 ] = −2 J3 from which we can conclude the property 1 1 J− J+ = J+ J− − J3 2 2 Hence. (5. (5. (5.m ijk Jk .292).290) |Ψ (j.289) yield [J− . (5.136 = 1 4 − = = 1 4 1 4 Theory of Angular Momentum and Spin < n |σi | m >< m |σj | m > b† bm n n.291) Here we have defined J+ J− = J1 + iJ2 = J1 − iJ2 = b† b− + = b† b+ . σj ]| m > b† bm n n.

.10: Algebraic Properties of Spinor Operators The last term on the r. using operators I± = I1 ± iI2 2 2 2 and the subspace {|n1 . m + 1) (j + m) (j − m + 1)|Ψ (j. m) J− |Ψ (j. m . I3 = (b+ b1 + b+ b2 ) 2 2 2 2 2i 2 1 satisfy the Lie algebra of SU(2). b† . (5. i. [Ij . (b) Show that the three operators 1 + 1 1 + (b1 b2 + b+ b1 ) . Ik ] = i jk I . m = m||λ. .e.300) (5.298) (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 = 0.10. . . m − 1) .1: The system investigated in this section is formally identical to a 2-dimensional isotropic harmonic oscillator governed by the Hamiltonian H = ω(b+ b1 + b+ b2 + 1) . . n} eigenstates of I 2 = I1 + I2 + I3 and of I3 I 2 ||λ. .s. Construct. 1. the number of states to the possible energy eigenvalues.h. . n2 . n1 = 0. .e. m) One can furthermore derive readily J+ |Ψ (j. can be written b† + b − b † − b + = = One can then state J2 = 1 (b† + b+ )2 + b† + b+ b† − b− + b† − b− b† + b+ 4 + (b† − b− )2 + 2 b† + b+ + 2 b† − b− Defining the operator 1 ˆ k = ( b† + b+ + b† − b− ) 2 one can write finally ˆ ˆ J2 = k ( k + 1 ) b† . m I1 = where λ = 0. λ. −λ + 1. . State the corresponding eigenvalues and the degree of degeneracy. (5. .5.301) Obviously. . j = 1. One can then conclude J 2 |Ψ (j. b † + b + + b † + b+ b † − b − 1 1 b† + b + + b † + b + b † − b− + b † − b − b † + b + . m) m |Ψ (j. i. m) . and eigenstates of k with eigenvalues j. ˆ respectively.. I2 = (b1 b2 − b+ b1 ) . m) J3 |Ψ (j. m = −λ.305) = = j ( j + 1 ) |Ψ (j. 1 2. . b+ ] = δjk . n = n1 + n2 fixed. 2. [bj . 2 . m) = = (j + m + 1) (j − m) |Ψ (j. Show λ = n1 +n2 n1 +n2 2 ( 2 + 1) and m = 1 2 (n1 − n2 ).254) are eigenstates of + b+ and − b− with eigenvalues j + m and j − m. . m = λ||λ. 2 2 137 (5. I3 ||λ.304) (5. the states (5.303) Exercise 5. 1 2 k (a) Show that the eigenstates are given by 1 |n1 .302) (5.299) (5.

In case j = 1 this expression yields. However. for example. On the other side the states |Ψ (j. σ values which yields (b† )j+m (b† )j−m is + − then σ = j + m − σ.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† .138 Theory of Angular Momentum and Spin 5. (j) (5. + − One may expect that these two conditions restrict both σ and σ.5. m = b† + j+m b† − j−m (j + m )! (j − m )! |Ψ0 (5.306) where b † and b † are given by (5.243).307) Comparision of (5.308) The latter sum involves terms (b† )j+m (b† )j−m for σ + σ = j + m and 2j − σ − σ = j − m.306) and (5. For this purpose we note that the states |Ψ(jm) in a rotated coordinate system according m to (5.254) are |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. The combination of σ. . m) . m) in the original coordinate system by j |Ψ j.309) . m ) are related to the + − states |Ψ (j.310) it reduces to (5.306. both conditions are satisfied for σ = j + m − σ. 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.  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.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). The prefactor of (b† )j+m (b† )j−m which according to (5.251).

309) does not distinguish between SU(2) transformations (5.s.. that the the representation (5.310) the similarity transformation ˜ A = U† A U where U is the 3 × 3 unitary matrix which establishes the transformation  1      1 i √ ( −x1 − ix2 ) − √2 − √2 0 x1 2     x3 0 1  . .251) and (5. U =  0 1 1 i √ ( x1 − ix2 ) √ x3 − √2 0 2 2 (5.308) unaltered except for a factor (−1)2j which multiplies then also the final result (5.   = U  x2  . the matrix A should represent a rotation around the x2 –axis in the space of vectors ˜ (x1 . b+ † = −cos β † β b+ − sin b† 2 2 − .251). 2 2 − (5. Hence.310) to the SU(2) transformation assumed in deriving the general result (5.312) The choice of this transformation derives from a property shown further below.h. Replacing the latter transformation by its negative form. b− † = sin β † β b+ − cos b† . is the expected element of SO(3). 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. i.310). the orthogonal 3 × 3 matrix which describes a rotation around the x2 –axis. It is of interest to trace the mapping from SU(2) onto SO(3) as represented by (5.g. therefore.310) establishes a mapping of SU(2) onto SO(3).308) and had the form (5.e. One can. i. namely that the component of the vector on the l. in case of j = 1. This can be shown by (1) applying to the matrix A = (dm m ) in (5. x2 . The SU(2) transformation entered in (5.5. of (5. This factor implies.311) (5.309) and the particular matrix (5. however.12: Mapping of SU(2) onto SO(3) 139 5.12 Mapping of SU(2) onto SO(3) The representation (5.314) in case of integer j–values.312) transforms like an angular momentum state ˜ |1 m . conclude that the mapping from SU(2) onto SO(3) is a 2–1 mapping..313) × 0 0 2  =  0 sinβ 0 cosβ 1 −i 0 which.314) leaves (5. in fact. x3 ) ∈ R . e.309).e.

140 Theory of Angular Momentum and Spin .

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

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

presumably a facility the reader would like to acquire with great eagerness.6) For the following it is essential to note that H is not invariant under rotations of only rAB or ρC . Following our description of rotations of single particle wave functions we express (6. the commutation relationships (1) Jp . in fact. It holds then H R(ϑ) ψ = R(ϑ) H ψ. The invariance of the Hamiltonian under overall rotations of the three particle system implies then H = R(ϑ) H R−1 (ϑ) . In this sense. ∂ρx ∂ρy (6. 3 (6. connected with the fact that the Hamiltonian H does not depend on the overall orientation of the three interacting particles. ρC ) it follows H R(ϑ) = R(ϑ) H.5) according to (5. (6.7) where the generators J (k) are differential operators acting on rAB (k = 1) and on ρC (k = 2).6. One can equivalently express therefore (6. be different from H. Since this is true for any ψ(rAB .e.. the reader will learn in the following section how to add and subtract in the microscopic world of Quantum Physics.8) Obviously. 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. according to (5.10) . but solely under simultaneous and identical rotations of rAB or ρC . ρC ) = ψ(R−1 (ϑ) rAB . i.48) i i R(ϑ) = exp − ϑ · J (1) exp − ϑ · J (2) (6.5) do not affect the Hamiltonian.0: Addition of Angular Momenta and Spin 143 corresponding physical properties of the elementary components.9) hold since the components of J (k) are differential operators with respect to different variables.55) holds i (1) − J1 = zAB ∂ ∂ − yAB . 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. ρC ) defined through R(ϑ) ψ(rAB .7) i R(ϑ) = exp − ϑ · J (6. To specify this property mathematically let us denote by H the Hamiltonian in the rotated frame. ∂yAB ∂zAB i (2) ∂ ∂ − J3 = ρy − ρx . 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). assuming presently that H might. q = 1. Jq(2) = 0 for p. Rotational Symmetry of the Hamiltonian As pointed out already. R−1 (ϑ) ρC ) (6. For ˆ ˆ example. 5. the existence of a basis (6. 2. rotations R(ϑ) of the wave functions ψ(rAB . Hence.4) which decouples rotational states is connected with the rotational symmetry of the Hamiltonian of the three particle system considered.53.

16) following the procedure stated in the theorem above [c..14) implies that the components of the total angular momentum operator (6. To describe the scattering process of AB + C most concisely one seeks eigenstates YJM of J2 and J3 which can be expressed in terms of Y 1 m1 (ˆAB ) Y 2 m2 (ˆC ). B + C] = [A.f. ˆ Ω 2 = ρC . We recognize.17) . k = 1. The property (6. We suspect. that the components Jk . (6. (6. (6. J ] = i k m Jm (6. 3 . J ] = i k m Jm for n = 1. r ρ Definition of Total Angular Momentum States The commutation property (6. the operators Jk .13) Hϑ · J − Since this holds for any ϑ it follows HJ − J H = 0 or. (5. In fact.14) We will refer in the following to Jk . of course.11) we can write the condition (6.15) are satisfied.6) for rotational invariance of the Hamiltonian in the form i i H = exp − ϑ · J H exp + ϑ · J .12) each individually can have simultaneous eigenstates with the Hamiltonian.71–5. 3 as the three components of the total angular momentum operator. Jk ] = 0 .81) we want to consider the relationship between YJM and Y 1 m1 (ˆAB ) Y 2 m2 (ˆC ). B] + [A.9).81)]. 3 cannot have simultaneous eigenstates among each other. k = 1. 2. One can show readily that the commutation relationships [Jk .. To order O(|ϑ|) one obtains i ϑ · JH . In the following we will use the notation r ρ Ω1 = rAB . C].14) implies that the total angular momentum is conserved during the scattering process. For a proof one uses (6. According to the theorem above this property implies that one can construct eigenstates YJM of J3 and of J2 = J2 + J2 + J2 1 2 3 (6. 2 together with the property [A. and the eigenvalues of J2 and J3 are good quantum numbers.71–5. ˆ (6. the (n) (n) (n) properties [Jk .e. the important fact that the Jk obey the Lie algebra of SO(3). i. [H. i. (6. 2.11) By means of (6. i. 2. Eqs. that energy.144 where J = J (1) + J (2) Addition of Angular Momenta and Spin . k = 1. Before we apply the procedure (5. k = 1.e.e. a supposition which can be tested through the commutation properties of these operators. we will find that the states yield the basis B with the desired property of a maximal uncoupling of rotational states. 3 do not commute.12) We consider this equation for infinitesimal rotations. for |ϑ| H ≈ 1 − 1 i ϑ·J H 1 + 1 i ϑ·J ≈ H + i 1. componentwise. 2. however.

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

24) The basis B( 1 . property implies dΩ1 dΩ2 Y 1 m1 corresponding to the l. it is not immediately obvious what the J– values are. . (6. in fact. 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. .2: Prove Eq. 2) = {Y 1 m1 (Ω1 ) Y 2 2 m2 (Ω2 ). M = −J. and that the states can be normalized.h. 2) . Exercise 6. However. J = | 1 − 2 |.s. For each quantum number J there should be 2J + 1 elements YJM with M = −J. In this case the range of |J|–values would be the interval [ |J (1) | − |J (2) | . .. Ω2 ) YJ M ( 1. | 1 − 2 | + 1.26) We will show below in an explicit construction of the Clebsch-Gordan coefficients that. This obviously corresponds quantum mechanically to a range of J–values J = | 1 − 2 |. . 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 . = {YJM ( 1 .1. (6. . . respectively. (6. 2 ) and B ( 1 . .m2 2 M and with (6. it holds 1+ 2 ( 2J + 1 ) = (2 J=| 1 − 2 | 1 + 1) (2 2 + 1) . J. Ω2 ) = δJJ δM M (6. B2 2) should be 2 |Ω1 . . to a new basis B ( 1 . −J + 1.146 Theory of Angular Momentum and Spin The expansion (6. m2 = − 2 . (6. . 1 + 2 . 2 ) has (2 1 + 1)(2 2 + 1) elements. JM 2 |Ω1 .18) constitutes a change of an orthonormal basis B( 1 . The basis B ( 1 .25) i. . . 2 ) is also orthonormal2 and must have the same number of elements. . . . Ω2 ). (6. . the range of values assumed for J is correct. (6. the basis B ( 1 .28) This property follows from the fact that the basis elements are eigenstates of hermitian operators with different eigenvalues. | 1 − 2| + 1.e. In fact. .24) yields 1 m1 2 m2 ) ∗ 1 m1 2 m2 ) (J M| = δJJ δM M . 1. |J (1) | + |J (2) |].. . 2 |Ω1 . Our derivation below will also yield real values for the Clebsch-Gordan coefficients. The property dΩ1 dΩ2 Y∗ ( 1 . − 2} 1 + 1.25) We want to state now two summation conditions which follow from the orthonormality of the two basis sets B( 1 .h. m1 = − 1 . − + 1. 1 + 2.s.27) together with (6. J (1) and J (2) .23) corresponding to the r. −J + 1. J} .18) applied to Y∗ and to YJ JM (J M | m1 . . . . 2 ).

2 ) span the same function space. (6.18) by Y∗ J orthogonality property (6. i. 2 |Ω1 . 1+ 2 J Y 1 m1 (Ω1 )Y 2 m2 (Ω2 ) = J =| 1 − 2 | M =−J cJ M YJ M ( 1. 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. it is possible to expand Y 1 m1 (Ω1 )Y 2 m2 (Ω2 ) in terms of YJM ( 1 .. Ω2 ) . or J > 1 + 2 (6.31) are identical to Comparision with (6. A similar 2 construction will also be applied to composite systems governed by other symmetry groups. therefore.6. i.. and actually also the properties of Clebsch-Gordan coefficients stated above. 2 |Ω1 .1 states.e. Ω2 ) and integrating. The (J M | 1 m1 2 m2 ) cJ M .. will be based solely on the commutation properties of the operators J and J (k) . the group SU(3) in case of meson multiplets involving two quarks.28) from (6..81) stated above. Hence.71–5.2: Construction of Clebsch-Gordan Coefficients 147 The second summation condition starts from the fact that the basis sets B( 1 . 1 m1 2 m2 ) = 0 if J < | 1 − 2| .34) 6. (6. also apply the results. Ω2 ) . to composite systems involving spin. The result of this construction will include all the properties previewed above. 2 |Ω1 .28) allows one to conclude that the coefficients cJ (J M | 1 m1 2 m2 )∗ . Ω2 )Y 1 m1 (Ω1 )Y 2 m2 (Ω2 ) . (6.e. 2 ) and B ( 1 . We can. Ω2 ). One can show readily using the same reasoning as applied in the derivation of (6. or baryons multiplets involving three quarks.2 Construction of Clebsch-Gordan Coefficients We will now construct the Clebsch-Gordan coefficients. 2 |Ω1 . 2 |Ω1 . . At this point we like to stress that the construction will be based on the theorems (5.m2 M ( 1. e.32) which complements (6.e.33) JM The latter summation has not been restricted explicitly to allowed J–values. rather the convention (J M | has been assumed.30) The latter property follows from multiplying (6.18). i.27) yields δJJ δM M = m1 .29) where the expansion coefficients are given by the respective scalar products in function space cJ M = dΩ1 dΩ2 Y∗ J M ( 1.18) that the Clebsch-Gordan coefficients obey the second summation condition (J M | 1 m1 2 m2 )∗ (J M | 1 m1 2 m2 ) = δm1 m1 δm2 m2 . (6.g.

We obtain using (5. Ω2 ) = Y 1 1 (Ω1 ) Y 2 2 (Ω2 ) .36) = 0 = J YJJ ( 1 . Ω2 ) (Ω1 ) dΩ2 Y 2 2 (Ω2 ) = 1. . Ω2 ) . M = −J. 2 |Ω1 . 1+ 2 In fact. Ω2 ) 2 |Ω1 . The corresponding classical total angular momentum vector Jclass (1) (2) would be obtained by aligning both Jclass and Jclass also along the z–axis and adding these two vectors. Ω2 ) . The construction assumes a particular choice of J ∈ {| such J–value seeks an expansion (6. we can demonstrate condition (6. 2 |Ω1 .68) J+ = (1) + J+ (1) (2) Y 1 1 (Ω1 ) Y 2 2 2 2 (Ω2 ) 1 1 (6.37) and replace (1) (2) according to (6. We first seek an unnormalized solution GJJ and later normalize. (6.38) for M = J − 1. 2 |Ω1 . 2 |Ω1 . . .42) ( 1.43) .11) and (5.37).18) which satisfies J+ YJJ ( 1 . −J + 1.37) The solution needs to be normalized. We embark on the suggested construction for the choice J = 1 + 2 . 2 |Ω1 . 1 + 2} and for (6.40) (Ω1 ) J+ Y (2) 2 2 J+ Y 1 1 (Ω1 ) Y (Ω2 ) + Y (Ω2 ) = 0 . we can also demonstrate using (??) that G dΩ1 = We.41) 2) Y (Ω1 ) Y (Ω2 ) . .64) J3 = = (1) + J3 (1) 1 (2) Y 1 1 (Ω1 ) Y 2 2 2 2 (Ω2 ) 1 1 J3 Y ( + 1 1 (Ω1 ) Y 1 1 (Ω2 ) + Y 2 2 (Ω1 ) J3 Y (2) 2 2 (Ω2 ) (6. | 1 − 2| + 1. .35) 1 − 2 |. . 2 |Ω1 . . 1+ 2 ( 1. 2 |Ω1 . Quantum mechanically this corresponds to a state G 1+ 2. 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.11) J+ by J+ + J+ . Ω2 ).39) into (6.66. (6. 2 |Ω1 . Having determined such YJJ we then construct the whole family of functions XJ = {YJM ( 1 . J} by applying repeatedly J− YJM +1 ( 1 .39) which we will try for a solution of (6. (6. therefore. J − 2. −J.25) using (6. 2 |Ω1 .148 For the construction of YJM we will need the operators J± = J1 + iJ2 . . 1+ 2. 1+ 2 ( 1. Ω2 ) = (J + M + 1)(J − M )YJM ( 1 .5. (6. Ω2 ) is normalized dΩ2 G dΩ1 Y 1 1 1+ 2. Similarly. J3 YJJ ( 1 . . . have shown Y 1+ 2. Ω2 ) = Y 1 1 (Ω1 ) Y 2 2 (Ω2 ) (6. Ω2 ) Addition of Angular Momenta and Spin (6. For this purpose we insert (6. 1+ 2 ( 1.

in particular. Normalization yields furthermore Y = 1 1 + 2 −1 1 + 2 −1 We have thereby determined ( 1 2 |Ω1 Ω2 ) (6.45). One can obviously continue the construction outlined to determine Y 1 + 2 −2 1 + 2 −2 . thereby. One can readily show that (6.21) 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. | . Show that the resulting functions are orthonormal. .38) yields pression 2( 1 + 2 )Y 1 + 2 1 + 2 −1 ( 1 . We seek for this purpose first an unnormalized solution G 1 + 2 −1 1 + 2 −1 of (6.36) 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. The result is illustrated for the case 1 = 2. 6. etc.47) can serve now to obtain recursively the elements of the family B + − for M = 1 + 2 − 2. 1 + 2 − 3. −( 1 + 2 − 1). . . . . To demonstrate (6.46) 2 1 +c Y (Ω1 ) Y 2 2 (Ω2 ) = 0 .27). Ω2 ) 2 1+ 2 = (Ω1 )Y 2 2 −1 (6.37) is satisfied. G 1 + 2 −1 1 + 2 −1 in (6.h.39) as required by (6. . The reader should familiarize himself with the entries of the Table.2: Construction of Clebsch-Gordan Coefficients 149 We now employ property (6. This expression is orthogonal to (6. Having constructed this family we have determined the Clebsch-Gordan coefficients ( 1 + 2 − 1M | 1 m1 2 m2 ). all the Clebsch-Gordan coefficients ( 1 + 2 M | 1 m1 2 m2 ). . Y 1 1 −1 (Ω1 )Y (Ω2 ) + Y 1 1 This construction can be continued to obtain all 2( 1 + 2 ) + 1 elements of B + and. ).38) to construct the family of functions B + = {Y + M ( . | . 2 |Ω1 . 2 = 1 in Table 1.45) for some unknown constant c. Ω2) for M = 1 + 2 − 2 and J = 1 + 2 .37). Exercise 6. The r. 1 + 2 − 1. of (6.38) 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 ). . ).1: Construct following the procedure above the three functions YJM ( 1 .6. 2 |Ω1 . . 2 |Ω1 . We demonstrate the procedure ex(1) (1) plicitly only for M = 1 + 2 − 1. of (6. . M = −( + ). 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. According to the condition (6. Expression (6. One obtains then Y 1 1+ 2 1 + 2 1 + 2 −1 2 2 ( 1.44) (Ω2 ) .s. The l.s. 1 + 2 − 2.h.2.47) (Ω2 ) − 1 1 2 + Y 2 1 1 −1 (Ω1 )Y 2 2 + Y 2 1 1 (Ω1 )Y 2 2 −1 (Ω2 ) . .36. ( + )}. We like to construct now the family of total angular momentum functions B + − = {Y + − M ( . M = −( + − ). To satisfy this equation one needs to choose c = − 1 / 2 . and all total angular momentum functions for a given choice of 1 and 2 . Ω2 ). . with the symmetry pattern and with the terms Y 1 m1 Y 2 m2 contributing to each YJ M . ( + − )}. .

632456 0.57735 −0. Ω2) Y3−1 (2. 1|Ω1 .730297 0. Ω2) Y11 (2.447214 0.316228 0.57735 0.258199 −0. 1|Ω1 .547723 0.774597 − − − − 0.316228 0.57735 0.816497 0. Ω2) Y1−1 (2.258199 0. Ω2) Y32 (2.632456 −0.408248 −0.707107 0. Ω2) Y2−2 (2. 1|Ω1 .707107 0.447214 −0. 1|Ω1 . 1|Ω1 .408248 −0.547723 0. Ω2) − − − − Y22 (Ω1 )Y11 (Ω2 ) 1 Y21 (Ω1 )Y11 (Ω2 ) 2 3 1 3 2 5 1 2 1 10 1 5 1 2 3 10 1 15 1 3 3 5 Y22 (Ω1 )Y10 (Ω2 ) 1 3 2 3 8 15 1 6 0.774597 0 2 5 8 15 1 6 3 10 1 3 0. 1|Ω1 .816497 Y2−2 (Ω1 )Y11 (Ω2 ) Y2−1 (Ω1 )Y10 (Ω2 ) Y20 (Ω1 )Y1−1 (Ω2 ) Y2−2 (Ω1 )Y10 (Ω2 ) − 2 3 Y2−1 (Ω1 )Y1−1 (Ω2 ) Y2−2 (Ω1 )Y1−1 (Ω2 ) 1 Table 6. 1|Ω1 . Ω2) Y22 (2. 1|Ω1 . Ω2) Y31 (2. Ω2) Y20 (2.57735 0. Ω2) Y30 (2. Ω2) Y21 (2. 1|Ω1 .707107 0. Ω2) Y2−1 (2. Ω2) Y3−3 (2. 1|Ω1 . Ω2) Y3−2 (2. 1|Ω1 .774597 0.150 Addition of Angular Momenta and Spin Y33 (2.632456 0. .57735 0. 1|Ω1 .57735 3 10 3 5 Y2−1 (Ω1 )Y11 (Ω2 ) Y20 (Ω1 )Y10 (Ω2 ) Y21 (Ω1 )Y1−1 (Ω2 ) −0. Ω2) Y10 (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 .730297 −0.1: Some explicit analytical and numerical values of Clebsch-Gordan coefficients and their relationship to the total angular momentum wave functions and single particle angular momentum wave functions.547723 0.816497 −0.707107 0.547723 0. 1|Ω1 .

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 ). hence.54) 2 + = √ b† ( 2 +m2 )! √− ( 2 −m2 )! . + 2  − J−M +1 for = J + 1 2J+2 2  J−M  for = J − 1 2J 2 = .10.9. b† refer to different particles and. bζ = = a† .3: Explicit Expressions of CG–Coefficients 151 Exercise 6.287) to two particles (1) J k k = = (2) J 1 2 1 2 ζ. − 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. we extent definition (5.10 [cf. 1 (6.6. M | .ζ a† < ζ |σk | ζ > aζ ζ b† < ζ |σk | ζ > bζ ζ (6. M | . a† ζ .53) (6.49) where ζ. It is actually possible to state explicit expressions for any single coefficient (JM | 1 m1 2 m2 ).52) and | 2 m2 2 According to Section 5.2: Use the construction for Clebsch-Gordan coefficients above to prove the following formulas  J+M 1  for = J − 2 2J 1 1 1 = J. Definition of Spinor Operators for Two Particles In contrast to Sections 5. 5. aζ bζ .254)] the angular momentum / spin eigenstates | 1 m1 of the two particles are | 1 m1 | 2 m2 1 + = √ a† 1 +m1 ( 1 +m1 )! 2 +m2 √− b† a† 1 −m1 ( 1 −m1 )! 2 −m2 |Ψ0 |Ψ0 . m + 1 . bζ .50) (6. ζ = ± and the matrix elements < ζ |σk | ζ > are as defined in Section 5. b† ζ ζ = aζ . 5. a† ζ ζ b† . Accordingly.9. m − 2 .51) The operators aζ . b† ζ ζ = 0 = 0 . 1 . For this purpsose we will employ the spinor operators introduced in Sections 5.10. b† ζ = 0.2. The creation and annihilation operators are again of the boson type with commutation properties aζ . (6. J. 6. a† and bζ .48) (6.10 where we studied single particle angular momentum and spin.ζ ζ. commute with each other ζ ζ aζ . 2 . (5. These expressions will be derived now. we are dealing now with two particles carrying angular momentum or spin. (6. aζ . bζ = a† . b † ζ = δζζ = δζζ .

(1) J 2 |J. we like to recall for future reference that the operators (5. m1 . i. M ( 1 .152 It holds. (1) J 2 . 2 )).68) The states |J. (5. k2 = b b+ + b† b− k1 = + − − 2 2 + namely. (j) 2 ˆ ˆ J = kj ( kj + 1 ) . 2 . M ( 1 .65) (6. ˆ kj |J. m1 1 | 2 . m2 1 2 ( 1 . (6. 2) 1. M ( 1 . = = 1( 1 2( 2 2) 2) (6.59) We seek to determine states |J. J3 |J.55) . can be expressed in terms of the number operators 1 † 1 ˆ ˆ a† a+ + a† a− .61) (6.60) (6.. m1 1 | 2 . 3 . M ( 1 . which describe a two particle system according to (6. m2 2 .54). (1) J (2) J 3 | 1 m1 1 3 | 2 m2 2 = m1 | 1 m1 = m2 | 2 m2 1 2 (6. M ( 1 . (2) J 2 . 2 . 5. 2) We will also require below the raising and lowering operators associated with the total angular momentum operator (6. m2 a† + ( 1 2 = a† − ( 1 1 −m1 (6.302. (6. M ( 1 . 1. M ( 1 . are | 1 . m2 |J. 2) . 2 )) . (6. . M ( 1 .18) as follows = | 1 m1 .57) b† + ( (2) 2 2 +m2 1 +m1 b† − ( 2 2 −m2 + m1 )! − m1 )! (1) + m2 )! − m2 )! |Ψo . 2. M ( 1 . (1) J .303). mj j = j | j . k = 1. 2) + 1) |J. = M |J. (2) J which.59) J± = J1 ± i J2 . for such states should hold J2 |J. 2 ) which are simultaneous eigenstates of the operators 2 2 J2 .m2 | 1 . M ( (2) J 2 |J. mj j and. M ( 2) 2) = J(J + 1) |J. 2) 2) (6. hence.53. M ( 1 . in analogy to Eqs. M ( 1 .69) − 2| ≤ J ≤ + 2 . .e.66) (6.56) The states | 1 . M ( 1 .58) Jk + (2) Jk . as usual are denoted by their respective quantum numbers J. 2) 2) can be expressed in terms of Clebsch-Gordan coefficients (6.300). ˆ For the operators kj holds ˆ kj | j . −J ≤ M ≤ J . 2 . m1 . m2 |J. j = 1. J3 .64) (6.63) according to At this point. m1 1 | 2 . (6. 6. 2 . M ( 1 . |J. (6. (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 . The operator of the total angular momentum/spin of the two particle system is J = with Cartesian components Jk = (1) J + J (6. The aim of the present Section is to determine closed expressions for the Clebsch–Gordan coefficents ( 1 . 1 .67) = j |J. + 1) |J. M.62) .

ˆ A similar calculation yields [k2 . K † J± . 6. a† b† − a† b† + − + − + − − + 2 1 1 = a† a+ . 6. a† b† − a† b† − + − − + 2 + 1 1 = a† a+ . (6. b† − b† a† b− .74) imply 3 2 J2 .74) + 1 J− J+ + J2 . (6. We note that. 2 4 = 0 = = 0. K † ] = (j) J 2 .51) one obtains ˆ k1 . K † + K † 2 2 3 † †ˆ K kj + K . K † + kj . a† b† + + − − + 2 2 − 1 1 = a† b† − a† b† = K† .71–6. 6.71) (6. 4 Using J3 = (1) J 3 + (2) J 3 . K† J3 . K † 1 † a a+ + a† a− . K † 1 a† a+ − a† a− + b† b+ − b† b− . b† = 0 . K † = 0.73. K † (kj + 1) 2 2 1ˆ † 1 ˆ kj K + K † (kj + 1) 2 2 1 ˆ 1 ˆ K † kj + kj . a† b† − a† a− .73) (6. + − − + 153 (6. and applying the relationships (6.3: Explicit Expressions of CG–Coefficients The Operator K † The following operator K † = a† b† − a† b† . Employing (6. a† b† + − − + 2 + 2 − 1 1 − b† a† b+ . This operator obeys the following commutation relationships with the other pertinent angular momentum / spin operators ˆ kj .288).65) and (6. a† b† + a† a− .70) will play a crucial role in the evaluation of the Clebsch-Gordan-Coefficients.64.6. the relationships (6. For example. 2 2 3 ˆ = K † kj + K † .72) (6.75) The relationships (6.73) yields J3 . + − − + 2 2 = 1 † 2K . due to J2 = K† 1 2 J+ J− 1 † K . j = 1.73) can be readily proven. expressing (k) J 3 through the creation and annihilation operators according to (5. K † 2 1ˆ ˆ 1 ˆ ˆ kj kj + 1. + − + − 2 2 − + = . using (6. j = 1.71–6.71) one can show K† = = = = = 1 ˆ ˆ kj (kj + 1). K † (j) J 2 .50.

this result implies that K † |J. j2 ) = |j1 .67) (j) J 2 K † |J.76) and state K† n |J. M ( 1 . M ( 1 . K † = Addition of Angular Momentum and Spin a† a− + b† b− .154 Starting from (5. on the state |j1 + j2 . M ( 1 . it holds K † |J. 2 ) j + 4 1 3 + K † |J. (6. j2 2 . Action of K † on the states |J. 2 ) remain eigenstates of J2 and J3 with the same quantum numbers. 2) = N |J. 2) = = (j) 2 J . j + 2 2 2) However. 2) We want to demonstrate now that the action of K † on the states |J. M ( 1 .77) where N is another normalization constant. a† b† + b† a† b− . j1 + j2 (j1 . j1 + j2 (j1 . j1 + j2 (j1 + n n . M ( 1 + n . 2 ) . 2 2 1 + ) . 6. M ( 1 . j1 1 + . 1 2 + 2 . but for different 1 and 2 quantum numbers of the operators (1) J 2 and (2) J 2 . The commutation properties (6. K † ] = 0 is demonstrated in an analoguous way.e. 2) is a state with quantum numbers 1 1 + 1 2 and = N |J.292) one can derive similarly J+ .76) Here N is an unknown normalization constant. j1 1 |j2 . M ( 1 . M ( 1 .73.79) n |j1 + j2 . One can generalize property (6. We consider now the case that (K † )n acts on the simplest total angular momentum / spin state. 2 (6. M ( 1 .62. Application of (K † )n to this state yields. 2) 2) ˆ K † kj + = K† = j 3 † K + K † j ( j + 1) |J. M ( 1.. b† + − + + + − The property [J− . M ( 1 .78) a state which has been used already in the construction of Clebsch-Gordan coefficients in Section 6. To demonstrate that the resulting states are eigenstates of (1) J 2 and (2) J 2 we exploit (6. = − a† a− . a† b− − a† b† + + + − + = 0. according to (6.72) and (6. 6. M ( 1 + . |j1 + j2 . namely. M ( 1 . 4 3 + j ( j + 1) |J. j2 + ) 2 2 n n = N (n. i. 6.63.75) ascertain that under the action of K † the states |J.77).2. j2 ) . K† + K † (j) J 2 |J. j1 + ) K † 2 2 (6. 2 ) produces again total angular momentum eigenstates to the same J and M quantum numbers of J2 and J3 . 2 2 + n ) 2 (6.

105) of the spherical harmonics. 6. M ( 1 . M ( 1 . 2) Our construction of the states |J. For this purpose one needs to choose in (6. 2 ) for −J ≤ M ≤ J as follows |J. The latter states. in analogy to the construction (5.80) = = = 1 1 2 2 − J 1 n − = (J + 2 2 n 1 − = (J + 2 2 + 1 1 − − 2) 2) . = (J + M )! (2J)! (J − M )! 1 2 2) (6.85) ∆(J. − 1.84) (6. 2) 2) . J( 1 .s.104. 2) = N( 1 + 2 − J. J( 1 .74)] yields |J.82. M ( 1 . 1.79). 2) . allow one to obtain the states |J. (6. j2 as follows J = j1 + j2 which is equivalent to n j1 j2 Accodingly. M ( 1 . 1 = j1 + n 2 . (6. 2 ) ∆(J. 2 ) exploits the expression (6. M ) . J( 1 .83) The derivation of this expression will be provided further below (see page 158 ff).57) and exploiting the fact that J− and K † commute [c. (J + 2 − − 2) 1 1) 2 The normalization constant appearing here is actually N( + − J. 2 = j2 + n 2 (6. K† 1 2 1 + 2 −J × × . j1 + n . (6. Strategy for Generating the States |J.79) n. (6. 2 2 It is now important to notice that any state of the type |J.h. holds |J. J( 1 .f. 2 ) . of (6. M ) (J− )J−M |J.86) b† + J+ 2 − 1 |Ψo .82) for |J. 2) = ∆(J.82) 1 × | (J + 2 1 × | (J + 2 1 (J + 2 1 − 1 ) . 1 .84) with (6. 2 ) can be expressed through the r. M ) (J + × K† 1 − 2 )! (J + 2 − 1 )! J+ 1 − 1 + 2 −J (J− )J−M a† + × 2 (6.3: Explicit Expressions of CG–Coefficients 155 where we denoted the associated normalization constant by N (n. Combining (6. M ( 1 . j1 . 2) 1 2 = ( 1 + 2 (2J + 1)! − J)! ( 1 + 2 + J + 1)! 1 2 . 5.81) = N( 1 + 2 1 2 − J.6. j2 + n ).

s. (6. I(λ.92) − − and noting [a+ .50–6.87) + − + − i. r!s!t! r.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. in analogy to (5.86) |Gr = Jr a† st − + s b† + t |Ψo (6. st The desired expansion of |Gr can be obtained from an alternate evaluation of I(λ. x. y) = λr xs y t r |Gst .69) yields then the Clebsch- Expansion of an Intermediate State We first consider the expansion of the following factor appearing in (6. 2 [cf. For this purpose we introduce the generating function I(λ. (6.88) in terms of monomials (6.. a† ] = [b+ . Comparision with (6. m2 Gordan-coefficients. b† ] = 0 [cf. x.57)].t (6.86) in terms of monomials 1 +m1 1 −m1 2 +m2 2 −m2 a† a† b† b† |Ψo . x.93) = f (a† + λ a† ) g(b† + λ b† ) |Ψo . m1 1 | 2 .52). One obtains then using J− = a† a+ + b† b+ (6.264). + + Taylor expansion of the two exponential operators yields immediately I(λ.e.156 Addition of Angular Momentum and Spin Our strategy for the evaluation of the Clebsch-Gordan-coefficients is to expand (6. y) which is st based on the properties aζ f (a† ) |Ψo = ζ ∂ ∂a† ζ f (a† ) |Ψo ζ bζ f (b† ) |Ψo = ζ ∂ ∂b† ζ f (b† ) |Ψo ζ (6.90) (6. (6..e.89) i. in terms of | 1 . + − + − .88). y) = exp (λJ− ) exp x a† exp x b† |Ψo . x. follows from the commutation properties (6.87). y) is a generating function for the states |Gr defined in (6.91) which.

y) = exp a† + λ a† + − exp b† + λ b† + − |Ψo . using the definition (6. m1 1 | 2 .86) (J− )J−M a† + = q J+ 1 − 2 b† + J+ 2 − 1 |Ψo 1 (6.97) + q)! |Ψo × (J − M )!(J + 1 − 2 )!(J + 2 − 1 )! q!(J + 1 − 2 − q)!(J − M − q)!(M + 2 − J+ 1 − 2 −q × Final Result a† + a† − q b† + M + 2 − 1 +q b† − J−M −q Our last step is to apply the operator (K † ) 1 + 2 −J to expression (6. 2) = s.99) . Expanding the exponentials in st (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.96) and. 157 (6.3: Explicit Expressions of CG–Coefficients We conclude I(λ. x. 2 ) in terms of states | 1 . one can write the right factor in (6.94) One can infer from this result the desired expressions for |Gr .s.50). M ( 1 .97) yields.6.t × a† + s−q s q b† + t r−q t−r+q a† − q b† − |Ψo (6. With K † given by (6.70) holds K† 1 + 2 −J = s 1 + 2 −J b† − s 1 + 2 −J−s (−1)s 1 + 2 −J−s (6. |J. using the commutation property (6.90) allows one to infer |Gr s.88).94) yields I(λ. M ( 1 .t = q r! × a† + s q s−q t r−q a† − q × b† + t−r+q b† − r−q |Ψo (6.q (−1)s (6. to obtain the desired expansion of |J. Operation of this operator on (6.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.97).98) a† − s a† + b† + s .95) Comparision with (6. x. m2 2 . y) = s.

83) of the normalization constant N ( 1 + defined through (6. J( 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. − 1 . m2 in (6. m1 .69) since. (6. 2 ) (6. M ) 1 = + J)!(− 1 + 2 + J + 1)! 2 2 + 2 − J)!( 1 − ( 1+ + J)! 1 2 × [( × s + m1 )!( − m1 )!( 2 + m2 )!( 2 − m2 )!(J + M )!(J − M )!] 2 (−1)s s!( 1 − m1 − s)!( 2 + m2 − s)! 1 2 1 × The Normalization 1 ( 1 + 2 − J − s)!(J − 1 − m2 + s)!(J − + m1 + s)! (6. 1 . Using (6.102) We want to determine now the expression (6. √ 2J + 1 1 2 .101) Note that m1 + m2 = M holds.82) and (6. m2 = M − 1 + q + s. The summation over q corresponds then to the summation over m1 .104) . 2 ) ∆(J.6. j2 = 1 (J + 2 2 − 1) . − q − s)!(q + s)! 2 × (M + 1 − 1 + q + s)!( 1 1 + 2 − M − q − s)! 2 − q − s 1 | 2.100) ( 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 . 1 . 2 ) − 2) . M ( 1 .53.101). n)|ψ(j1 . q = 1 − m1 − s and m2 = M − m1 . (6. × 2) = N( 1 + 2 1 − J. m2 |J. n = 1 + 2 −J.103) this can be written 1 = N 2 ψ(j1 . according to (6. M − +q+s One can conclude that this expression reproduces (6.q (−1)s × (6. j2 .82). The Clebsch-Gordan coefficents are then finally ( 1 . 2 ) we consider the scalar product J. n) 2 )|J. To determine N = N ( 1 + 2 −J. J( 1 .54) between creation operator monomials and angular momentum states allow one to write this |J.103) = 1.69) if one identifies m1 = 1 −q − s . M ) (J + 2 )! (J + 2 − 1 )! s. For this purpose we introduce j1 = 1 (J + 2 1 2 − J. j2 .

applying (6. n) = K† n 159 |j1 .106) The orthonormality of the states occurring in the last expression allows one to write (6. m2 2 .104) 1 = N2 (n!)2 (2j1 )!(2j2 )! (2j1 + n − s)!(2j2 + s)! s!(n − s)! 2j2 + s 2j2 (6. 1 1−λ n1 +1 1 1−λ n2 +1 = m1 . One obtains then. We employ the expression (6.110) which can be written 1 1−λ n1 +n2 +2 = r s n1 + r − s n1 n2 + s n2 λr (6. j2 . n) in terms of states |j1 .105) The first step of our calculation is the expansion of ψ(j1 .6. j2 .109) and Taylor expansion of the left hand side of (6. Accordingly. j2 .108) yields the identity n1 + r − s n1 n2 + s n2 = n1 + n2 + r + 1 n1 + n2 + 1 .3: Explicit Expressions of CG–Coefficients where |ψ(j1 .107) s = (n!)2 s 2j1 + n − s 2j1 The latter sum can be evaluated using 1 1−λ a property which follows from ∂ν ∂λν 1 1−λ n1 +1 n1 +1 = m1 n1 + m1 n1 λm1 (6.108) twice.m2 n1 + m1 n1 n2 + m2 n2 λm1 +m2 (6.57) for these states and the expression (6. we obtain |ψ(j1 .108).112) s . (6.111) Comparision with (6. (6. (6.108) = λ=0 (n1 + ν)! n1 ! (6.70) for the operator K † . m1 1 |j2 . j2 2 . j1 1 |j2 . 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 .

and J (2) are vectors in R . m2 |J.115). To derive relationship (6. manner to the coefficients ( 1 .s. 1 .113) Using the identities (6. J (1) on the r. m1 ) . J (1) . m1 . M ) = (−1) 2 +m2 2J + 1 ( 2 .114) one obtains − J = − J (1) − J (2) The respective Clebsch-Gordan coefficients ( 1 . (6. namely. (6. For the quantum mechanical addition of angular momenta the Clebsch Gordan coefficients ( 1 .s.s. M ) .103) one obtains the desired result (6. and .114) by. m2 ) and.h.4 Symmetries of the Clebsch-Gordan Coefficients The Clebsch-Gordan coefficients obey symmetry properties which reflect geometrical aspects of the operator relationship (6. M ) ( 1 .117) are again related in a simple 2 .120) can be readily derived from the expression (6.114) For example. M ) corresponding to (6. 1 .116) If one takes the negatives of the operators in (6.g. and (6. (6. m1 |J. −m2 . m2 .114) show a simple relationship to the Clebsch Gordan coefficients ( 2 . M | 1 .116). −M ) = (−1) 1 + 2 −J ( 1 . m1 . of (6.83).102) by replacing ( 1 .118).118) Finally. m2 |J. 6. m1 ). vice versa. −m1 2|J. 2 . of (6.114) as long as J.160 Applying this to (6. m2 |J. m1 . 2 .102) of the Clebsch-Gordan coefficients. −m1 .107) yields 1 = N 2 (n!)2 Theory of Angular Momentum and Spin 2j1 + 2j2 + n + 1 2j1 + 2j2 + 1 n!(2j1 + 2j2 + n + 1)! = N2 . 2 . 2 1+1 (6. ( 2 . of this equation J (1) = J (2) − J .119) The corresponding symmetry property of the Clebsch-Gordan coefficients is ( 1 . We will demonstrate this now.h. 2 . m1 |J. m1 ) by ( 2 . −m1 . M ) = (−1) 1 + 2 −J ( 2 .116) through formula (6.116) one expresses the Clebsch-Gordan coefficient on the r. m1 . interchanging the operators J (1) and J (2) results in J = J (2) + J (1) .120) The symmetry properties (6. m1 . J.11) J = J (1) + J (2) . (6. ( 1 . (6. −M ) . e. 2 . (6. m2 |J. m2 |J. one can interchange also the operator J on the l.115) This relationship is a trivial consequence of (6.. 2 . (2j1 + 2j2 + 1)! (6. m2 ) by ( 1 .h. (6. M ) . m2 . −m2 |J. M ) corresponding to (6.

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

(5.180) and. 1 |ˆ). 1 |ˆ) r r r 2 2 2 2 2 2 (6. For this purpose we consider the application of σ · r in the case of θ = 0. 1 |ˆ) = ((σ · p f (r)))Yjm (j ± 1 .181) 0 = j+ 1 2 4π 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. given by ˆ σ · r = σ1 sin θ cos φ + σ2 sin θ sin φ + σ3 cos θ . hence.182) 0 Since σ · r. 1 |ˆ) + f (r) σ · p Yjm (j ± 1 . ˆ (6. Since f (r) is independent of θ and φ follows ˆ ((σ · p f (r))) = −i (( ∂r f (r) )) σ · r . 6.185) We have.148) the particular ˆ 1 1 states Yj 1 (j − 2 . (6. 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 . One can conclude then ˆ σ·r = ± ˆ 0 1 1 0 (6. for θ = 0 (6.184) 0 1 space χ 1±1 2 2 one can conclude from (6. 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.186) Here (( · · · )) denotes again confinement of the diffusion operator ∂r to within the double bracket. φ) and (5. According to (5. 1 |θ. 6.147. 1 |ˆ) r 2 (6.187) 1 2 .f.188) .182) σ · r Yj 1 (j − 1 .160).1 states χ 1 ± 1 2 2 2 [c.180) and (6. We want to demonstrate finally that the “−”-sign holds in (6. 1 |θ = 0. (6. therefore. φ) . φ) = − Yj 1 (j + 1 . 1 |θ = 0. φ) 2 2 2 = − j+ 1 2 4π (6. identified the sign of (6.180).174–5. 1 |θ = 0.180) where the sign could depend on j. We want to determine ˆ its action on the wave functions f (r) Yjm (j ± 1 .183) in case θ = 0 becomes in the standard representation with respect to the spin. have proven (6. 1 |θ = 0. 2 |ˆ) and Yj 1 (j + 1 . 1 |ˆ) at θ = 0 are r r 2 2 2 2 Yj 1 (j − 1 . ˆ (6. φ) 2 2 2 .168 Addition of Angular Momentum and Spin 1 Yjm (j ± 2 .181. ˆ 2 2 2 2 2 2 for θ = 0.224)] 1 0 σ·r = ˆ . The result can also be stated in the compact form σ · r Yjm (j ± 1 .

196) = i j+ ∂r + r 1 2 3 2 1 f (r) Yjm (j − 2 .189) The celebrated property of the Pauli matrices (5.198) . r Altogether we obtain.5: Spin–Orbital Angular Momentum States Using (6. using ∂r Yjm (j ± 1 .6.g.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. To demonstrate this we consider one of its cartision components.193) r r 1 1 1 = −i (∂2 x3 − ∂3 x2 ) h − i h (x3 ∂2 − x2 ∂3 ) . e. 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. 1 |ˆ) r 2 (6.191) (6. x3 x2 ˆ ˆ ( p × r )1 h = −i (∂2 − ∂3 ) h (6.156) this yields finally 1 ˆ σ · p f (r) Yjm (j + 2 . 1 |ˆ) . 6. hence. h r (6. 1 |ˆ) r 2 1 2 (6. r 2 (6.197) 1 ˆ σ · p g(r) Yjm (j − 2 . r 2 2 1 J ·S 2 1 ˆ σ · p f (r) Yjm (j ± 2 .192) ˆ ˆ The operator p × r in (6. ∂3 x2 = x2 ∂3 we obtain ˆ ˆ ( p × r )1 h = −i 1 1 (x3 ∂2 − x2 ∂3 ) h = − J1 h r r (6. 1 |ˆ) r 2 (6.230) allows one to express ˆ ˆ ˆ ˆ ˆ σ · p σ · r = p · r + i σ · (p × r) . 1 |ˆ) = 0 and σ = 2S/ . Corresponding results are obtained for the other components of p × r and. r r r Using ∂2 (1/r) = −x2 /r3 .187) one obtains ˆ σ · p f (r) Yjm (j ± 1 ..195) .190) can be related to the angular momentum operator.155. ∂3 (1/r) = −x3 /r3 and ∂2 x3 = x3 ∂2 . 1 |ˆ) = r 2 2 ˆ i ∂r f (r) − f (r) σ · p σ · r Yjm (j ˆ 1 2 169 .53). we conclude the intuitively expected identity 1 ˆ ˆ p×r = − J .194) ˆ ˆ where J1 is defined in (5. 1 |ˆ) r 2 = i ∂r + −j r 1 g(r) Yjm (j + 2 . 1 |ˆ) = i [ ∂r + r + ] f (r) Yjm (j 2 r r Using (6.186) for both terms in (6.

j 2 + 2j + 2 3 4 (6.101). as well as (6.198). j − 1 .170 Addition of Angular Momentum and Spin To demonstrate the validity of this key result we note that according to (5. . i.202) Yjj (j − 1 . 1 |j.143.205) which agrees with the expressions (6. 1 |ˆ) r 2 2 r 2 3 i i 1 2 1 r = [ i∂r + ( − j) ] [ i∂r + (j + ) ] f (r) Yjm (j + 2 . j) 2 2 2 2 (j − .201) 1 3 (j + )(j + ) Yjm (j + 1 .2. 1 |ˆ) = r 2 2 which agrees with (6.199) We want to show that eqs.. (6. according to (6. 5. We note ˆ (σ · p)2 f (r) Yjm (j + 1 . in fact. 2 2 (6. r ∂r2 r 2 (6.203) (6. according to (6.230. We begin with the coefficients (6. r 2 2 2 2 2 2 − r 2 ∂r r + J2 r2 f (r) Yjm (j + 1 . are consistent with this identity.204) (6. For this purpose we use the construction method introduced in Sec.199).45). Yjj−1 (j − 1 . adopting the method in Sec. consider first the case of the largest m-value m = j. j + .e. Evaluation of Relevant Clebsch-Gordan Coefficients We want to determine now the Clebsch-Gordan coefficients (6.200) = (j + 1 )(j + 3 ). For m = j − 1 one can state.146).43). 6. 1 |ˆ) 2 r 2 r 2 j+3 ( 1 − j)(j + 3 ) 2 2 2 2 2 = [ −∂r − ∂r + − 2 ] f (r) Yjm (j + 1 . r (6. 6. − |j. using (5.197. In this case holds.2.151). 1 . 6. 1 |ˆ) r 2 2 i 3 ˆ = σ · p [ i∂r + (j + ) ] f (r) Yjm (j − 1 .143–6. 1 |ˆ) = r 2 2 2j − 1 Yj− 1 j− 3 χ 1 1 + 2 2 2 2 2j 1 Y 1 1 χ1 1 2j j− 2 j− 2 2 − 2 (6. 1 |ˆ) r 2 2 The Clebsch-Gordan coefficients are then (j − 1 . 6. 1 |ˆ) r r 2 2 2 2 2 2 2 yields ˆ (σ · p)2 f (r) Yjm (j + 1 .144) and. 6. j) 1 2 1 2 1 2 1 2 = = 1 0 (6.99) holds ˆ (σ · p)2 = − 2 2 = ∂2 J2 r + 2 . 1 |ˆ) = J 2 Yjm (j + 1 .206) . 1 |ˆ) r 2 2 r r2 r2 j 2 + 2j + 3 2 2 2 4 = [ −∂r − ∂r + ] f (r) Yjm (j + 1 .144) for m = j.143. 1 |ˆ) = Yj− 1 j− 1 (ˆ) χ 1 1 . 1 |ˆ) r 2 2 r r2 and.

211) to the l.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 .s.6. 1 |j. [c. j − 1) 2 2 2 2 which again agrees with the expressions (6. etc.137)] J− (tot) = J− + S− (6. m − 1) 2 2 2 2 = j +m−1 2j (6. 1 . (6. Let us verify then the expression (6.210).2. m − 3 . 6. j − 1) 2 2 2 171 = = 2j − 1 2j 1 2j (6. 6.210)] yields (j + m)(j − m + 1) Yjm−1 (j − 1 . is obtained by applying the operator [c. 2 2 2 2 2j (6. 1 |ˆ) = r 2 2 Applying J− (tot) j+m Yj− 1 m− 1 χ 1 1 + 2 2 2 2 2j j−m Yj− 1 m+ 1 χ 1 − 1 . 2 |j.5: Spin–Orbital Angular Momentum States The corresponding Clebsch-Gordan coefficients are then 1 (j − 1 .143. (6. and J− + S− to the r.202). 6.f.209) (j − 1 .. j − 3 . − 1 |j. 1 .207) (6.206). are obtained by iterating the application of (6. 1 |ˆ) = r 2 2 j+m 2j + (j + m − 1)(j − m + 1) Yj− 1 m− 3 χ 1 1 2 2 2 2 j+m Yj− 1 m− 1 χ 1 − 1 2 2 2 2 2j j−m 2j (j + m)(j − m) Yj− 1 m− 1 χ 1 − 1 2 2 2 2 + or Yjm−1 (j − 1 . 1 .h.144) implies for j = m Yjm (j − 1 .144) for the Clebsch-Gordan coefficients by induction.143. 2 2 2 2 2j . 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 .f. The further Clebsch-Gordan coefficients (· · · |jj − 2).s.143. Expression (6.208) (6. as described in Sec. 6. (· · · |jj − 3).h. (6.210) to (6.144) for m = j − 1. j − 1 .

2 . 2j + 2 (6. M. 2 . 6.143) by applying the symmetry relationships (6. m1 .144) for j = m − 1.216) 6. m2 | 3 . to reflect in the notation the symmetry of these quantities. m3 . m) = 2 2 2 2 Similarly. 1 .217) 2j + 2 (j + 1 . m + 1 . m3 . −m2 | 1 . The Clebsch-Gordan coefficients (6. In this context Jclass and Jclass play the same role. hence.172 Addition of Angular Momentum and Spin j −m+1 2j (j − 1 .215) which follows from (6. m − 1 . − 1 |j. j3 . However. j+m+1 . m1 ) (6. 1 . 1 . m1 . We have.143.215).145) from (6. m. j2 m2 ) 2 . − 1 |j. m. Jclass and Jclass play equivalent roles. The latter relationships applied together read ( 1 . all three vectors Jclass .213) which is in agreement with (6. m1 . 1 |j + 1 . They are related in a simple manner to the ClebschGordan coefficients j1 j2 j3 m1 m2 m3 = (−1)j1 −j2 +m3 (2j3 + 1)− 2 (j3 − m3 |j1 m1 . 1 . 1+1 3 For 1 (j. 1 .143). 6. proven (6.146) can be obtained from (6. m + 1 . m2 . using (6. m + 1 ) = 2 2 2 2 and.144) by induction. m) 2 2 2 2 2j + 1 (6. (1) (2) (−tot) Obviously. one can obtain expression (6. 6. by the set j1 . one obtains (j + 1 . m3 ) = (−1) × 2 + 3 − 1 + 2 +m2 × 2 . 2j + 1 (6. m + 1 ) = 2 2 2 j+m+1 .116.218) where we have replaced the quantum numbers 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. m2 1 (6. j2 . (1) (2) (−tot) The coefficients which are the quantum mechanical equivalent to Jclass + Jclass + Jclass = 0 are the 3j–coefficients introduced by Wigner. m − 1) 2 2 2 2 = (6. 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 .120). − 1 |j.214) yields (j.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 . 1 |j + 1 .214) 2 2 +1 ( 3 .144). the relationship (6.143.

j3 form the sides of a triangle and the condition is symmetric in the three quantum numbers. there exist even further symmetry properties. In this way the values of 12 3j-coefficients are related. m3 and with respect to sign reversals of all three values m1 . 173 (6. However. √ safe for the prefactor 2j3 + 1 which is cancelled according to the definition (6.221) j1 j2 j3 −m1 −m2 −m3 where the the results of a cyclic.e. are identical to the integer arguments which enter the analytical expression (6. states that j1 . with respect to alltogether 12 symmetry operations. −j1 + j2 + j3 . The reader may note that the entries of the Regge-symbol. To represent the full symmetry one expresses the 3j–coefficients through a 3 × 3–matrix. m2 .223) (Σ + 1)!n11 !n22 !n33 !n23 !n32 !)) n=0 n31 n32 n33 . m1 . The Regge symbol also vanishes in case that two rows or columns are identical and Σ is an odd integer. for which no known classical analogue exists. Because of its high degree of symmetry the Regge symbol is very suited for numerical evaluations of the 3j–coefficents.219) (6.21. 6. For this purpose one can use the symmetry transformations to place the smallest element into the upper left corner of the Regge symbol. The Regge symbol reflects a remarkable degree of symmetry of the related 3j–coefficients: One can exchange rows. In case of a non-cyclic exchange of rows and columns the 3j–coefficent assumes a prefactor (−1)Σ .6: The 3j–Coefficients We first like to point out that conditions (6. i. The two integer entries J − 1 − m2 and J − 2 + m1 in (6. These symmetries follow the equations j1 j2 j3 m1 m2 m3 = (−1)j1 +j2 +j3 = j2 j3 j1 m2 m3 m1 = (−1)j1 +j2 +j3 j2 j1 j3 m2 m1 m3 (6. the so-called triangle condition. j2 .g. as follows   −j1 + j2 + j3 j1 − j2 + j3 j1 + j2 − j3 j1 j2 j3 j1 − m1 j2 − m2 j3 − m3  . j2 .34) imply j1 j2 j3 m1 m2 m3 j1 j2 j3 m1 m2 m3 = 0 if not m1 + m2 + m3 = 0 = 0 if not |j1 − j2 | ≤ j3 ≤ j1 + j2 .218) relating 3jcoefficients and Clebsch-Gordan coefficients. According to the definition of the 3j–coefficients one would expect symmetry properties with respect to exchange of j1 . anti-cyclic exchange and of a sign reversal are stated.102) of the Clebsch-Gordan coefficients.220) The latter condition |j1 − j2 | ≤ j3 ≤ j1 + j2 . except when all elements are non-negative integers and each row and column has the same integer value Σ = j1 + j2 + j3 . m2 and j3 . m3 . the Regge-symbol.222) m1 m2 m3 j1 + m1 j2 + m2 j3 + m3 The Regge symbol vanishes. discovered by Regge.102) are obtained each through the difference of two entries of the Regge-symbol.. 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. one can exchange columns and one can reflect at the diagonal (transposition).6. These symmetry operations relate altogether 72 3j–coefficients. e. =  (6.

226) by means of a symbolic manipulation package (Mathematica) (1m|2m1 1m2 ) = (−1)1+m+m1 δ(m. m2 ) √ 2 1 ≥ |m1 |) 2 (6.10752786393409395427444450130056540562826159542886 We also illustrate the numerical values of a sequence of 3j-coefficients in Figure 6.1. (6.223-6.231) Here is an explicit value of a Clebsch-Gordan coefficient: 1 1 (70 2 −15 1 |120 −10 60 2 −5 1 ) = 2 2 √ √ √ 4793185293503147294940209340 127 142 35834261990081573635135027068718971996984731222241334046198355 0.174 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) (6.224) (6. m1 + m2 ) (2 − m1 )! (2 + m1 )! √ 6 (2 − m)! (2 + m)! (1 − m2 )! (1 + m2 )! 1 ≥ |m2 | ∧ 2 ≥ |m| ∧ 2 ≥ |m1 | (6.230) 1 1 (1m| m1 m2 ) = 2 2 √ (−1)2m1 −2m2 3 δ(m.225) sn−1 . m1 + m2 ) (2 − m1 )! (2 + m1 )! √ 10 (1 − m)! (1 + m)! (1 − m2 )! (1 + m2 )! 1 ≥ |m| ∧ 1 ≥ |m2 | ∧ 2 ≥ |m1 | (6.229) 1 1 (0m| m1 m2 ) = 2 2 i(−1)1−m2 δ(0.227) (2m|2m1 1m2 ) = (−1)m+m1 (m + 2m2 ) δ(m.226) We like to state finally a few explicit analytical expressions for Clebsch-Gordan coefficients which were actually obtained using (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. m) δ(−m1 . (6. m1 + m2 ) (1 − m)! (1 + m)! √ 1 1 1 1 6 2 − m1 ! 2 + m1 ! 2 − m2 ! 2 + m2 ! 1 1 ≥ |m1 | ∧ ≥ |m2 | ∧ 1 ≥ |m| 2 2 (6.232) .228) √ (−1)2m1 −2m2 7 δ(m.

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

1: Prove Eqs. M = −J. (2( 1 + 2 ) + 1) ×(2( 1 + 2 ) + 1) Partitioning in smaller blocks is not possible. Show that the probability of detecting a triplet substate |1m1 (1) for arbitrary polarization (m2 – 1 value) of the other triplet is 3 . . . . . . . . Exercise 6.6.3: Two triplet states |1m1 (1) |1m2 (2) are coupled to an overall singlet state Y00 (1.234) Exercise 6. 1 .. J} (2 (2 1 2 + 1) (2 + 1) × (2 1 2 + 1) + 1) = (2| 1 − 2 | + 1) ×(2| 1 − 2 | + 1) (2| 1 − 2 | + 3) ×(2| 1 − 2 | + 3) (6. (6. J = | 1 − 2 |.6.233.234) .233) is further block-diagonalized as follows + 2. Ω2 ). 1). 2 |Ω1 . .176 Addition of Angular Momentum and Spin 1 For the basis {YJM ( 1 . . each of the blocks in (6. .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. 2 = 1. 2. .6. Exercise 6. . .

−k + 1. The operator T may also act on multi-particle states like Y 1 m1 (ˆ1 )Y 2 m2 (ˆ2 ).g. This implies. 6.237) are called tensor operators of rank k.236) In this equation Dq q (ϑ) denotes the rotation matrix Dq q (ϑ) = kq |R(ϑ)|kq . i.6.7 Tensor Operators and Wigner-Eckart Theorem In this Section we want to discuss operators which have the property that they impart angular momentum and spin properties onto a quantum state. (k) (6. In fact. (k) (6.7: Tensor Operators 177 6. for example. . Such operators T can be characterized through their behaviour under rotational transformations.g.236.237) The operators T ∈ {Tkq .222) in case (ii) of single particle spin operators. The latter can be written T |ψ where T denotes T in the rotated frame given by T = R(ϑ) T R−1 (ϑ) . . some of the r r examples considered below involve tensor operators T of this type. a mul∂2 ∂2 tiplicative operator T ψ(r) = f (r) ψ(r) or a differential operator T ψ(r) = ( ∂x2 + ∂x2 + ∂2 ) ψ(r). .g. The latter implies that T transforms like an angular momentum or spin state | m . e. .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 . Examples of Tensor Operators The multiplicative operators C∞ ( ) → C∞ ( ) Y ( ) = def Y ( ) (6.e. ∂x2 3 1 2 2 2 2 case (ii) the operator T could be a spin operator Sk defined in (5. Rotations transform |ψ as |ψ = R(ϑ)|ψ and T |ψ as R(ϑ)T |ψ . k} such that k Tkq = q =−k (k) Dq q (ϑ) Tkq . Note that in the examples mentioned the operator T would be defined within the same representation as R(ϑ). (6. 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.238) .223). the operator (5. k} with the transformation property (6. e. . q = −k. . S 2 = S1 + S2 + S3 or any other polynomial of Sk .42) in case (i) of the position representation of single particle wave functions or the operator (5. that T belongs to a family of operators {Tkq . e. q = −k. −k + 1. that in case (i) T is an operator C∞ ( ) → C∞ ( ) acting on single particle wave functions.

In fact.1: Show that the following set of spin operators T00 T1±1 T10 T2±2 T2±1 T20 = 1 = 1 √ S± 2 S± 2 = S3 = = = (S3 S ± + S ± S3 ) 2 2 (3S3 − S 2 ) 3 are tensor operators of rank 0.7. 2.2: Express the transformed versions of the following operators (a) T = x2 − x2 and 1 2 (b) S2 S3 in terms of Wigner rotation matrices and untransformed operators. 2.235 . x2 . we will assume presently that the rotation is chosen as follows R(ϑ) = exp ( ϑ+ L+ + ϑ− L− + ϑ3 L3 ) (6.3. x3 . 3 follows from the transformation properties of the spherical harmonics derived in Section 1. Exercise 6. 1. but are far easier to ascertain for any particular operator.237). but rather any rotation and any parametrization can be assumed. This choice of parametrization allows us to derive conditions which are equivalent to the property (6. Exercise 6. For the following it is important to note that the rotation matrix elements (6.178 are tensor operators of rank k. 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 = (6.239) These operators can be expressed in terms of the coordinates x1 .7. The fact that these operators form tensor operators of rank 1.240) 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 (5.6.203). .

Exercise 6.237) be assumed then.242) from which follows by means of kq |1 1|kq equation ϑ+ [L+ .7.6: Form tensor operators of rank 1 in terms of the three components of acting on the space of 1-particle wave functions. Tkq ] [J− . (k + q + 1)(k − q) (k + q)(k − q + 1) (6. We first consider a rotation with ϑ = (ϑ+ . .244) [L+ . 1 (6.80) follows kq |L+ |kq = −i (k + q + 1)(k − q)δq and. Expressing the results in terms of the angular momentum operators J+ .245) (6.235 . 0.241) Using R(ϑ) = 1 + ϑ+ L+ + O(ϑ2 ) this equation can be rewritten neglecting terms of order O(ϑ2 ) 1 + + k (1 + ϑ+ L+ ) Tkq (1 − ϑ+ L+ ) = 1 1 q =−k kq |1 + ϑ+ L+ |kq Tkq . Tkq ] = = δqq and by subtracting Tkq on both sides of the k ϑ+ kq |L+ |kq Tkq .237) for transformations characterized by infinitesimal values of ϑ+ . For this 2 2 purpose state first the proper rotation operator R(ϑ) and note then that S (1) · S (2) commutes with 1 the generators of the rotation of | 2 m1 (1) | 1 m2 (2) .243) From (5. Tkq ] [J3 .235 . Similar equations can be derived for infinitesimal rotations of the form ϑ = (0. 2 Exercise 6. ϑ− . (6. 6. Exercise 6. ϑ3 )T .7.6.237) yields first k R(ϑ) Tkq R −1 (ϑ) = q =−k kq |R(ϑ)|kq Tkq .247).246) (6.3: Derive Eqs. ϑ3 .248) a tensor operator? Exercise 6.7. ϑ− .6.4: Is the 1-particle Hamiltonian 2 H = − 2 2m + V (|r|) (6. (6.235 . The property (6.246. Show that S (1) · S (2) is a tensor operator of rank 0 in the space of the products of the corresponding spin states | 1 m1 (1) | 1 m2 (2) .6.247) These properties often can be readily demonstrated for operators and the transformation properties (6.1 particles for which the first spin is described by 2 the operator S (1) and the second spin by the operator S (2) . J3 yields [J+ . (6.7: Tensor Operators 179 The conditions can be derived if we consider the property (6. 0. 0)T in case of small ϑ+ . q =−k q+1 (6. Tkq ] = −iTk q+1 (k + q + 1)(k − q) . hence. 0)T .6. (0. J− . Tkq ] = = = Tkq+1 Tkq−1 q Tkq .7.5: Consider a system of two spin.

J 2 | 2 m2 γ2 = 2 2( 2 + 1) | 2 m2 γ2 .8. i.e. 2 |γ2 ) = Cδ 1 1 δm1 m1 2 1( 1 (6. 2 |γ2 ) is an eigenstate of J 2 with eigenvalue + 1). in fact.e.8 Wigner-Eckart Theorem A second important property of tensor operators Tkq beside (6.247). The hermitian property of J3 and J 2 implies that states Φ 1 m1 (k. the stated property. m1 . 1 m1 |Φ 1 m1 (k.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 . J3 | 2 m2 γ2 = m2 | 2 m2 γ2 . i. J3 . (6.249) which demonstrates. J3 | 2 m2 γ2 = m2 γ2 | 2 m2 and the property (6. 2 |γ2 ) are orthogonal to | 1 m1 in case of different quantum numbers 1 . J2 . These states according to (6. Starting point of the derivation is the fact that the states Tkq | 2 m2 γ2 behave like angular momentum states of a composite system of two particles each carrying angular momentum or spin. 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) ( ) (6.1: Show that Φ 1 m1 (k. Before we proceed we like to point out that the states | 2 m2 γ2 are also eigenstates of J 2 . which is characterized also by a set of other quantum numbers γ1 which are not affected by the rotational transformation R(ϑ).m2 ( 1 m1 |kq 2 m2 ) (J3 Tkq − Tkq J3 + Tkq J3 ) | 2 m2 γ2 = qTkq = q. 2 |γ2 ) is an eigenstate of J 2 with eigenvalue 2 1 ( 1 + 1). . behave like |kq | 1 m1 .6.250) q. 2 |γ2 ) = q. one can show that Φ 1 m1 (k. 2 |γ2 ) = ( 1 m1 |kq 2 m2 ) Tkq | 2 m2 γ2 . construct states Φ 1 m1 (k. J3 are the generators of the rotation R(ϑ).251) The corresponding property for Φ 1 m1 (k. (6.m2 ( 1 m1 |kq m1 q. One can.e. hence.235 . 2 |γ2 ) which correspond to total angular momentum states.18) are defined through Φ 1 m1 (k.21) of Clebsch-Gordan coefficients J3 Φ 1 m1 (k.237) is that their matrix elements 1 m1 γ1 |Tkq | 2 m2 γ2 obey simple relationships expressed in terms of Clebsch-Gordan coefficients. 2 |γ2 ) can be shown readily as follows using (6. | 1 m1 γ1 denotes an angular momentum (spin) state.252) ∼δm1 q+m2 = ( 1 m1 |kq 2 m2 ) Tkq Similarly. i. possibly the total angular momentum– spin state of a compositie system. The relationships among the matrix elements 1 m1 γ1 |Tkq | 2 m2 γ2 are stated by the Wigner–Eckart theorem which we will derive now.m2 2 m2 ) (q + m2 ) Tkq | 2 m2 γ2 | 2 m2 γ2 (6.253) Exercise 6.180 Theory of Angular Momentum and Spin 6.

m1 = q = m2 = 0. say m1 + 1. γ1 |Tkq | 2 m2 . can be evaluated as easily as possible. γ2 = 2 1 +1 1 m1 . i.258) We can then finally express the matrix elements of the tensor operators Tkq as follows 1 m1 .g.257) which establishes the m1 –independence of 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) .259) to a combination of magnetic quantum numbers m1 . γ1 ||Tk || 2 . 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.255) At this point the important property can be proven that 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) is independent of m1 . (6. 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 ) = (6.8.261) .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. γ1 ||Tk || 2 . Using | 1 m1 + 1γ1 = 1 ( 1 + m1 + 1)( 1 − m1 ) J+ | 1 m1 γ1 (6.260) Exercise 6. γ2 is determined by applying (6. γ2 k− 2 + 1 √ 1 2 m2 ) (−1) 2 1 +1 = 1 . To prove this property we consider 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) for a different m1 value.32) Tkq | 2 m2 γ2 = and orthogonality property (6. γ2 (−1)k− 2 + 1 ( 1 m1 |kq 2 m2 ) (6.259) The socalled reduced matrix element 1 .h.254) = 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) ( 1 m1 |kq 2 m2 ) .2: Determine the matrix elements of the gradient operator d3 rF (r) G(r) of the type (6. for which the l. m2 .256) and noting that the operator adjoint to J+ is J− .s.e. e.6.8: Wigner-Eckart Theorem 181 In order to evaluate the matrix elements (6. One can then evaluate also the corresponding Clebsch-Gordan coefficient ( 1 m1 |kq 2 m2 ) and determine 1 . the matrix elements 1 m1 γ1 |Tkq | 2 m2 γ2 can be reduced to an m1 –independent factor. q . γ2 ( 1 m1 |kq (6. γ1 |Tkq | 2 m2 . γ1 ||Tk || 2 . (6. γ1 ||Tk || 2 . γ2 .

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

1993) 1 183 .2) serve as schematic descriptions of quantum particles.4) is by far the most relevant of the four choices.Chapter 7 Motion in Spherically Symmetric Potentials We describe in this section the stationary bound states of quantum mechanical particles in spherically symmetric potentials V (r). The potential (7. 7. The second potential is of the square well type −Vo 0 ≤ r ≤ R .4) governs the motion of electrons in hydrogen-type atoms.1) confines a freely moving particle to a spherical box of radius R.2) V (r) = 0 r > R The third potential V (r) = 1 mω 2 r2 2 (7.3) describes an isotropic harmonic oscillator . The corresponding wave functions serve basis functions for multi-electron systems in atoms. Switzerland. in potentials which are solely a function of r and are independent of the angles θ.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 . Four examples will be studied. Potential (7.1. It leads to the stationary electronic states of the hydrogen atom (Z = 1). 7000 Chur. Poststrasse 22. for example. i.e. See. The first potential V (r) = 0 0 ≤ r ≤ R ∞ r > R (7. Simple Models of Complex Nuclei / The Shell Model and the Interacting Boson Model by I. The potentials (7. Talmi (Harwood Academic Publishers. molecules. φ. and crystals. (7. in case of the so-called bag model of hadronic matter.. The fourth potential V (r) = − Z e2 r (7. for example.

.87) for Jk as well as the commutation property (5. the time variation of J can be written. (5.m (r) can be chosen as simultaneous eigenstates of the Hamiltonian ˆ operator H as well as of J 2 and J3 . . 3.f.5) In the case of an angular independent potential angular momentum J = m r × v is a constant of ˙ motion. (7. 2 2mr2 (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 . using (7. Jk } where H is the Hamiltonian. (7.m (r) m ψE.85– 5.m (r) (7. 2 . In fact. ˆ A classical particle moving in a potential V (r) is governed by the Newtonian equation of motion (7.m (r) .14) (7. 3 . J ψE.99)] 2 1 2 J2 2 2 = − ∂r r + . i. stationary states ψE.12) .7) ˆ The correspondence principle dictates then for the quantum mechanical Hamiltonian operator H and angular momentum operators Jk ˆ [H. Jk ] = 0 .10) (7. 2 . 2.m (r) = = = E ψE. k = 1. For this purpose one concludes first that the conservation of angular momentum J implies a motion of the particle confined to a plane. J3 ψE.15) (7.. 2. Jk } = 0 . In classical mechanics one can exploit the conservation of angular momentum to reduce the equation of motion to an equation governing solely the radial coordinate of the particle.m (r) . θ for the distance from the origin and for the angular position. Accordingly. (7.61) of J 2 and Jk . ˆ dt (7. ( + 1) ψE. k = 1.e.8) This property can also be proven readily employing the expression of the kinetic energy operator [c.9) − 2m 2m r 2mr2 the expressions (5. one can state ˙ J = m r2 θ .13) . d ˙ J = v × mv + r × mv = 0 + r × er (−∂r V (r)) = 0 .5) and r = v.6) ˙ Since Jk is also equal to the poisson bracket {H. Employing in this plane the coordinates r. ˆ H ψE. one can conclude ˙ Jk = {H.184 Spherically Symmetric Potentials 7.m (r) .11) (7.1 Radial Schr¨dinger Equation o ˙ mv = − er ∂r V (r) .

equations (7.18) = vE. It can be solved by integration of 1 2 2 J2 r = ± ˙ E − V (r) − .16) m 2mr2 Once.1: Radial Schr¨dinger Equation o 185 This is a differential equation which governs solely the radial coordinate. can exclude particles from the space with small r values. . in the present case.20) in the form of the one-dimensional Schr¨dinger equation o 2 − where 2m 2 ∂r + Veff (r) − E φE (r) = 0 . .12) are obeyed and one obtains for (7. (7.m (r) and E .m (r) = 0.11).11.m .21) Veff (r) φE (r) = = V (r) + r vE.24) . (7.4.7.10) − 1 2 ∂ r + 2m r r 2 2 ( + 1) + V (r) − E 2mr2 . mr2 (t) (7.m (r) 2 . φ) (7. (r) = 0 ..19) where Y m (θ.13). together with the original potential. − Adopting for ψE. .10). . . Multiplying (7.m (r) 1 2 J2 ∂r r + + V (r) − E 2m r 2mr2 the functional form ψE. Employing the kinetic energy operator in the form (7. r(t) is determined the angular motion follows from (7. (7.m (r) Y m (θ.m (r) describes the radial motion as a one-dimensional motion in the interval [0.m (r) . ( + 1) 2mr2 . φ) are the angular momentum eigenstates defined in Section 5. by integration of ˙ θ = J .22) (7. 2 (7.20) by −2mr/ 2 yields the so-called radial Schr¨dinger equation o 2 ∂r − ( + 1) − U (r) − κ2 r2 r vκ. for the wave function of a quantum mechanical particle a differential equation which governs solely the r-dependence. but can also trap particles in the latter space giving rise to strong scattering resonances (see Section ??). (7. 7. . (7.20) Since this equation is independent of the quantum number m we drop the index m on the radial wave function vE.e. i.m vE. This barrier.23) This demonstrates that the function r vE.m ψE.22) which is the original potential 2 ( +1) V (r) with an added rotational barrier potential 2mr2 . ∞[ governed by the effective potential (7. .9) one can write the stationary Schr¨dinger o equation (7. using (7. One can write (7.17) In analogy to the classical description one can derivem.m (r) = 0.

New York. .” by J. see.182) for Y00 .26) 2m 2 In case E < 0. Berlin. 1975). . assumes the general form r vκ. Galindo and P. ψE.30) and. according to a well-known result in Classical Electromagnetism3 . using the expression (5. in particular.28) (7. r → 0 (7. (r) ∼ A r + B r− −1 +1 + B r− (7.186 where we defined U (r) κ2 = = − − 2m 2 Spherically Symmetric Potentials V (r) E (7. hence. 2nd Ed. We replaced in (7. In this case the solution is governed my ( + 1) 2 r vκ. for example. at r = 0 is found in the excellent monographs Quantum Mechanics I. For r → 0 one may assume that the term ( + 1)/r2 becomes larger than the potential U (r).20) one needs to specify proper boundary conditions2 . the term Br−( +1) would contribute |B|2 0 dr r−2 +1) (7.29) Only the first term is admissable.31) to the complete wave which.m (r) ∼ √ 4π B δ(x1 )δ(x2 )δ(x3 ) .20) the index E by the equivalent index κ.m (r) ∼ √ B .27) ∂r − r2 and. (r) ∼ A r or vκ. κ assumes real values. The radial part of this integral is ∞ 0 2 dr r2 vκ. Pascual (Springer. 2 2 2m 2 ψE. (r) = 0 . 4π|r| (7. (7. II by A. Boundary Conditions In order to solve (7. ”Classical Electrodynamics.33) A detailed discussion of the proper boundary conditions. Jackson (John Wiley.32) The total kinetic energy resulting from this contribution is. for > 0 is not integrable. 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). This follows for > 0 from consideration of the integral which measures the total particle density. For = 0 the contribution of Br−( function is.D. . (r) (7.25) (7. accordingly.

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).

For example. Since the spherical harmonics Y 0 (θ. (7. and using the Rodrigues formula for Legendre polynomials (5.103) and.e.106). z = kr. can be written exp(ikr cos θ). hence.99) +1 d cos θ eikr cos θ P (cos θ) = b j (kr) −1 2 . 7..s. (7.h.50). namely.98) by ∞ eik3 x3 = =0 b j (kr) P (cos θ) . φ).98) does not involve any non-vanishing m-values since Y m (θ. The orthogonality properties (5.h. In this case the expansion on the r.150) one obtains +1 +1 1 ∂ dx eizx P (x) = dx eizx (x2 − 1) .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.m a mj (kr) Y m (θ. One can expand the former solution in terms of solutions (7.102) −1 d −1 +1 −1 dx −1 (x2 − 1) . of (7. One can show d −1 2 (x − 1) ∼ (x2 − 1) × polynomial in x dx −1 (7.43) has two solutions.104) dx eizx P (x) −1 = (−1) 2 ! (iz) = 2 ! +1 dx (x2 − 1) −1 +1 dx (1 − x2 ) eizx . φ) for non-vanishing m has a non-trivial φ-dependence as described by (5.45) and one given by (7. the wave function does not depend on φ.s. 2 +1 (7.98) The l. −1 .179) yield from (7.44. This holds for −1 and one can conclude +1 consecutive integrations by part d izx e dx (7. the surface term ∼ [· · ·]+1 vanishes. i. (7. according to (5.178) are given in terms of Legendre polynomials P (cos θ) one can replace the expansion in (7. in case of a free particle moving along the x3 -axis one expands eik3 x3 = .50). φ) . one given o by (7.99) We want to determine the expansion coefficients b .194 Generating Function of Spherical Bessel Functions Spherically Symmetric Potentials The stationary Schr¨dinger equation of free particles (7.100) Defining x = cos θ.

96) one can express this. for the leading power x [c.106) yields finally b = i (2 + 1) or.s. z i or i (2 + 1) z (2 )! 2 2 2 +1 2 ! [1 · 3 · 5 · · · (2 − 1)] 1 (7. 2 +1 −1 dx (1 − x2 )ν− 2 eizx . −1 (7.113) dt (1 − t2 )ν− 2 eizt . Comparision with the l.99).105) and (7.100) gives b j (kr) 2 (iz) = 2 +1 2 ! +1 195 dx (1 − x2 ) eizx .110) One refers to the l.h.108) where the last factor is equal to unity.111) Employing (7. in particular.7. 1 (7. after insertion into (7.109) eikr cos θ = =0 i (2 + 1) j (kr) P (cos θ) .106) Employing (5.69)] b z 2 (iz) = 1 · 3 · 5 · · · (2 + 1) 2 + 1 2 ! +1 dx (1 − x2 ) . −1 (7.s. of (7.114) . ∞ (7. 1 (7.107) 1 2 (2 )! 1 · 3 · 5 · · · (2 + 1) 2 + 1 1 · 2 · 3 · 4 · · · (2 − 1) · 2 (7.109) results in the integral representation of j (z) j (z) = (z) 2 +1 ! +1 dx (1 − x2 ) eizx .h. Integral Representation of Bessel Functions Combining (7.f.112) We want to consider the expression Gν (z) = aν z ν fν (z) where we define fν (z) = C (7.2: Free Particle Comparision with (7. The identity (7.117) one can write the r. (7.h. using ν = Jν (z) = √ 1 πΓ(ν + 1 ) 2 z 2 ν + 1. (7.105) This expression allows one to determine the expansion coefficients b . as the generating function of the spherical Bessel functions.s. −1 (7.105) must hold for all powers of z.

114. (7. To prove this we note fν (z) = i C dt (1 − t2 )ν− 2 t eizt . (7.124) . 1 (7.93) for arbitrary ν. t2 of the integration paths C.120) imply also the property (2ν + 1) fν (z) + z fν+1 (z) = 0 . Gν (z). 2. We can now demonstrate that Gν (z) defined in (7.4: Prove (7.116) fν (z) = − or fν (z) = − From this we can conclude fν (z) + 2ν + 1 fν (z) + fν (z) = 0 . t2 .121) (7.2.116) In case that the endpoints of the integration path C satisfy (1 − t2 )ν+ 2 eizt one can write (7.93).113) obeys the Bessel equation (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 .113) satisfies (7. 3.122) (7.119) We note that equations (7. 7. z (7. for properly chosen endpoints t1 . 1 j = 1.123) Insertion of these identities into Bessel’s equation leads to a differential equation for fν (z) which is identical to (7. In fact. 4. Exercise 7.93) as long as the integration path in (7.120) z 2ν + 1 dt (1 − t2 )ν− 2 eizt + C C 1 1 t2 t1 = 0 (7.121). C 1 (7.118) z 2ν + 1 fν (z) + fν (z) .117) dt (1 − t2 )ν− 2 t2 eizt 1 (7. it holds for the derivatives of Gν (z) Gν (z) G (z) = = ν aν z ν fν (z) + aν z ν fν (z) z 2ν ν(ν − 1) aν z ν fν (z) + aν z ν fν (z) + aν z ν fν (z) 2 z z (7.120) such that we can conclude that Gν (z) for proper integration paths is a solution of (7.196 Spherically Symmetric Potentials Here C is an integration path in the complex plane with endpoints t1 . We consider now the functions u(j) (z) = √ 1 πΓ(ν + 1 ) 2 z 2 ν Cj dx (1 − x2 )ν− 2 eizx . obeys Bessel’s equation (7.117).

130) holds Jν (z) = (1) 1 2 1 (1) (2) Hν (z) + Hν (z) . for Rez > 0 and ν ∈ R obey (7.124) vanishes along the whole path C3 and.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 . u(3) (z) ≡ 0.134) and (7. . (7. one can state Jν (z) = − u(2) (z) + u(4) (z) . (7.124) is analytical in the part of the complex plane surrounded by the path C we conclude 4 u(j) (z) ≡ 0 j=1 (7. one can derive (2) Hν (z) = e −i(z−πν/2−π/4) ∞ 0 √ ds [s(1 − is/2)]ν− 2 e−zs . dt = i ds. 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 . therefore.130) We note that the endpoints of the integration paths C2 and C4 .132) For Hν (z) one derives.112) shows u(1) (z) = Jν (z). 1 (7.135) (7.7. u(2) (z) and u(2) (z) are both solutions of Bessel’s equation (7.93).134) Similarly. the integral expression (1) Hν (z) 1 1 1 1 (7.126) −1 ≤ s ≤ 1 (7. we introduce the so-called Hankel functions (1) Hν (z) = −2 u(2) (z) .117) and. Comparision with (7. Since the integrand in (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 (7.128) C = C1 ∪ C2 ∪ C3 ∪ C4 is a closed path.125) 0 ≤ s < ∞ (7. using t = 1 + is. 1 (7.131) According to (7.135) are known as the Poisson integrals of the Bessel functions.127) 0 ≤ s < ∞ (7. (2) Hν (z) = −2 u(4) (z) .129) The integrand in (7. Accordingly. hence. (7. Following convention.

(7.138) The formula for the Laplace transform of sn leads to f and. h(z) to be linearly independent. h) must be a non-vanishing function.141) (1) (1) the derivation of which follows the derivation on page 192 for the Wronskian of the radial Schr¨dinger equation.93) holds the identity 1 W = − W z (7.135) which read for ν = + 1 2 H where f The binomial formula yields f (1) 2 (1) 2 1 +2 (1. hence. We employ for this purpose the Poisson integrals (7.143) .198 Asymptotic Behaviour of Bessel Functions We want to obtain now expansions of the Hankel functions H Spherically Symmetric Potentials the expansions converge fast asymptotically.134) and (7.2) (z) +1 2 in terms of z −1 such that (z) = e±i[z − ( ∞ +1) π ] 2 2z z ( 1 ) f 2 (z) π ! (7.139) (z) = 2 ±i[z − ( e πz +1) π ] 2 r=0 ( + r)! r!( − r)! i ± 2z r (7. z (7. One obtains W (z) = − 4i . For such solutions g(z).137) (z) = r=0 r ± i 2 r 0 ∞ ds s +r e−zs ..93) only as ν 2 .142) and h(z) = H (2) (z) +1 2 one can identify the constant c by using the leading terms in the expansions (7.140).140) Bessel Functions with Negative Index Since ν enters the Bessel equation (7. converge fast for |z| → ∞.e. the Wronskian W (g. As a second order differential equation the Bessel equation has two linearly independent solutions. i. The general solution of (7. Hν (z) as well as H−ν (z) are solutions of this equation. For the Wronskian connected with the Bessel equation (7. we obtain H (1) 2 1 +2 (1) 2 (z) = r=0 ( + r)! ! r!( − r)! ± i 2 r 1 z +r+1 (7. (7.141) is o W (z) = − In case of g(z) = H (1) (z) +1 2 c . πz (7.136) (1) 2 (z) = 0 ds [s(s ± is/2)] e−zs .

2) (z) +1 2 199 are.e. (7.140). (7. 7.152) (7.146.144) The expansion coefficients A. H (1.2) (7.97).97) one defines the spherical Hankel functions h (1.148) Following the arguments provided above (see page 193) the functions h (z) are solutions of the radial Schr¨dinger equation of free particles (7. B can be obtained from the asymptotic expansion (7.146) (z) = −i (−1) H (2) 1 (z) +2 .90.150) According to (7. H−ν (z) = A H (1) (1) (z) +1 2 One can then expand + BH (2) (z) +1 2 . 2z 2 (1. (7. From (7.145) This equation can hold only for B = 0 and A = exp[i( + 1 )π].151) are equivalent to h h (1) (2) (z) (z) = = j (z) + i n (z) j (z) − i n (z) .148) holds for the o regular spherical Bessel function j (z) = 1 (1) (2) [ h (z) + h (z) ] . 2 (1.149) We want to establish also the relationship between h (z) and the irregular spherical Bessel function n (z) defined in (7.147) this can be written n (z) = 1 2i h (1) (z) − h (2) (z) .149. 7.132.132) follows n (z) = 1 2 π (−1) 2z +1 H−( (1) 1 +2) (z) + H−( (2) +1) 2 (z) . one can show H−( Spherical Hankel Functions In analogy to equations (7. (7. According to (7. (7. (7.97.2) (7. For |z| → ∞ the leading terms yield π π π π π 1 1 1 √ ei(z+ 2 ) = √ A ei(z− 2 − 2 ) + √ B e−i(z− 2 − 2 ) z z z |z| → ∞ . 7.7. 7. 7. We conclude 2 H−( Simarly.96.153) . (7. 7.151) Equations (7. linearly independent.2) (2) +1) 2 (1) +1) 2 (z) = i (−1) H (1) (z) +1 2 .147) (z) = π (1. in fact.2) H + 1 (z) ..2: Free Particle i.53).

2z 2 z sin(z − π/2) z cos(z − π/2) − .140) and (7.156) ( + r)! r!( − r)! ( + r)! r!( − r)! i 2z r [ /2] = r=0 ( + 2r)! (2r)!( − 2r)! −1 4z 2 r (7.159) (7. 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) . j (z) (z) = ( i) z +1 e±iz r=0 ( + r)! r!( − r)! ± i 2z r (7. 7.200 Asymptotic Behaviour of Spherical Bessel Functions Spherically Symmetric Potentials We want to derive now the asymptotic behaviour of the spherical Bessel functions h and n (z). = sin(z − π/2) and sin[z − ( + 1) π ] .162) The leading terms in these expansions. From (7. z (7. (7.161) (7.2) (z). 2 2z z (7.154) The leading term in this expansion. Hence.154) the identities j (z) n (z) = = sin[z − ( + 1) π ] cos[z − ( + 1) π ] 2 2 p (z) − q (z) z 2z 2 cos[z − ( + 1) π ] cos(z − π/2) 2 q (z) + p (z) . at large |z|.155) (z) and To determine j (z) and n (z) we note that for z ∈ R the spherical Hankel functions h (2) h (z).148) one obtains readily h (1) 2 (1. it follows z ∈ R : Using p (z) = Re r=0 j (z) = Re[h (1) (z)] .163) (7.160) Employing cos[z − ( + 1) π ] . are j (z) n (z) = = (7.164) .148). are complex conjugates. 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 (7.140. as given by (7.158) one can derive then from (7. (1) (7. is h (1) 2 (z) = ( i) z +1 e±iz .

151) follows then that the relatiosnhips hold also for g (z) = j (z) and g (z) = n (z).173) A (z) =  z ( +1) z2 +2 z (7.149.172) to +1 (z) +1 (z) = A (z)  1− g (z) g (z) −1   (7.171) −1 (z) (7.172) where g (z) is either of the functions h obtain the recursion relationship g g (z).131. 7. From the linearity of the relationships (7. 7.170) Recursion Formulas of Spherical Bessel Functions The spherical Bessel functions obey the recursion relationships g g +1 (z) +1 (z) = = (1.176) .2) (z) = −2 π 1 √ 2z πΓ(ν + 1 ) 2 z 2 ν C2.175) Using Γ( + 1) = !.2: Free Particle Expressions for the Spherical bessel Functions j (z) and n (z) 201 The identities (7.165) (7.148) and express g (z) = h (1.168) (7. 7.2) relationships hold for g (z) = h (z). j (z) and n (z).167) (7.4 dx (1 − x2 )ν− 2 eizx .169) (7. 1 (7.174) We want to prove these relationships. 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 (7.2) g (z) − g (z) z 2 +1 g (z) − g z (7.7.171.124.162) allow one to provide explicit expressions for j (z) and n (z).114) we can write g (z) = a z f 2 ! + 1 (z) 2 .2) To demonstrate that (7. 1. One can combine (7. (1. One obtains for = 0.166) (7. 7.174) and from (7.161.171) holds for g (z) = h (z) we employ (7. 7. For this purpose we need to demonstrate only that the (1.171–7. (7. defining a = −1 and employing fν (z) as defined in (7.

166). j0 (z) using (7. Exercise 7.202 The derivative of this expression is g (z) = Employing (7.179) g (z) = − z2 g (z) + g (z) − g +1 (z) z . and (b) n1 (z).181) Using this identity to replace the second derivative in (7.171). i.173. Exercise 7.173).178) (7.182) Replacing all first derivatives employing (7.180) yields g +1 (z) = g (z) − ( + 2) g (z) + z2 +2 g (z) .173. The second component of (7. The first component of (7.2.58) it holds o 2 g (z) = − g (z) + z ( + 1) g (z) − g (z) .180) Since g (z) is a solution of the radial Schr¨dinger equation (7.174) we start from (7..182). In order to prove (7.e.179) leads to (7.173) is equivalent to (7. n2 (z) from n0 (z).177) (z) = − z f 2 +2 + 3 (z) 2 .171). (7.165). together with (7. 7. in fact. To prove (7. z2 (7.2.174) to determine (a) j1 (z). (7.172).176) g (z) = z g (z) − g +1 (z) +1 2 Spherically Symmetric Potentials z g (z) + a z f 2 ! +1 2 (z) . (7. j2 (z) from j0 (z). f yields. is equivalent to (7.171).121).172). z (7.5: Provide a detailed derivation of (7.6: Employ the recursion relationship (7. 7. n0 (z) using (7.179) from which follows (7. .172) we differentiate (7.

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

t) + ∂t A(r. t) = × A(r.2) the term q r˙o δ(r − ro (t)). The reader is referred to the well-known textbook ”Classical Electrodynamics”. Equation (8.204 Maxwell equations1 · E(r. t) (8. but leave the fields unaltered: A(r. t) through the scalar and vector potentials V (r. In the non-relativistic limit holds p ≈ mr and (8.25).4). accordingly. t) = − V (r. t) (8. the relationship between fields and potentials is not unique. . t] + v × B[ro (t). solves implicitly (8. t) in (8. contributes to the charge density ρ(r. t) .1) the term qδ(r − ro (t)) and to the current density ˙ J(r. t)] and. t) = − V (r. t) and B(r. called gauge transformations. t) = 0 = 0. t) −→ 1 A(r. obeys the equation of motion d p = q dt E[ro (t). t) −→ V (r.8) = 0 (8. (8. t) in (8.11) V (r. t) + χ(r.2) In addition.9) (8.4) Lorentz Force A classical particle with charge q moving in the electromagnetic field experiences the so-called Lorentz force q[E(r. 2nd Edition. t) = 4 π J(r. t). The following substitutions. t) − ∂t A(r. (8. 1975) for a discussion of these and other conventional units.1) (8. called the scalar potential. the two homogeneous Maxwell equations hold × E(r. t) . t) − ∂t E(r. Scalar and Vector Potential Setting B(r. t) + v × B(r. t) × B(r. From this follows E(r.7) Gauge Transformations We have expressed now the electric and magnetic fields E(r. t) . New York. As is well known.3) (8. by J. t) · B(r.3) reads then × E(r. Jackson (John Wiley & Sons. t) which is solved by E(r. t) and A(r. We assume so-called Gaussian units.10) (8. t) + ∂t A(r. t) Interaction of Radiation with Matter = 4 π ρ(r. called the vector potential. t) where V (r. t) + ∂t B(r. The particle. alter the potentials.5) where p is the momentum of the particle and ro (t) it’s position at time t. t) is a scalar function. in turn. t) − ∂t χ(r. D. t). t] (8. (8.5) above agrees with the equation of motion as given in (1.6) for some vector-valued function A(r.

t) and V (r. (8. t = 4 π J(r. t) (8. t) = 1 ∂t V (r.13) · A(r. t) can be obtained employing (8. t) − 2 × A(r. (8.20) . the gauge freedom allows us to impose on the vector potential A(r. t) |r − r | (8. t) = 0 .19) Unfortunately.14) In case of the boundary condition V (r.8. t) . t) = Ω∞ for r ∈ ∂Ω∞ (8. t) − ∂t V (r.2).6) and (8. t) = · A(r.18) together with condition (8. such that the latter describes the electromagnetic radiation.12) The corresponding gauge is referred to as the Coulomb gauge. One would prefer a description in which the Coulomb potential (8. a name which is due to the form of the resulting scalar potential V (r. t) . be achieved. For this purpose we examine the offending term ∂t V (r. t) in (8. this potential results from inserting (8. Such description can. t) = 0 the solution is given by the Coulomb integral V (r. and a component due to the remaining currents. t) the condition · A(r. t) .17) A(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. t). t) and the current due to moving net charges. t) + ∂t V (r. the second inhomogeneous Maxwell equation.9) for the fields results in × The identity × × A(r.19) and define J (r. and the former the Coulomb interactions in the unperturbed bound particle system.1) · − V (r. t) together with (8. t). equation (8.9) into (8. t) . in fact. In fact.12) leads us to 2 2 A(r. The vector potential A(r. In fact. (8. t) .15) d3 r ρ(r .16) and the vector potential are uncoupled. t) − ∂t A(r. (8. t) + ∂t A(r.12) yields then the Poisson equation 2 V (r. 4π (8. t) − ∂t A(r. t) = − 4 π J(r. (8. t) = ∂t = 4 π ρ(r. Using the expressions (8. t) = − 4 π ρ(r.19) couples the vector potential A(r. t) Using · ∂t A(r.16) This is the potential commonly employed in quantum mechanical calculations for the description of Coulomb interactions between charged particles.

t) . t) as the current due to the time-dependence of the charge distribution ρ(r. i. t) (8.19) yield 2 2 A(r. t) − ∂t A(r.27) which is obviously curl-free ( × E (r. t) .25) Because of properties (8. hence. no divergence-free currents. = − ∂t A(r. t) be the total current of the system under investigation and let Jt = J − J . t) + ∂t ρ(r.2 Planar Electromagnetic Waves The current density Jt describes ring-type currents in the space under consideration.. (8. t) These fields are obviously divergence -free (e.e. (8. t) = 0). yet a current. (8. t). A(r. t) − ∂t A(r. t) = × A(r. t) = or = 2 Interaction of Radiation with Matter (8. t) contributes solely an electric field component E (r. in a ring-shaped antenna which exhibits no net charge.25) one refers to J and Jt as the longitudinal and the transverse currents. (8. t) = 0 .21) ∂t and the Poisson equation (8.206 For the curl of J holds × J (r. t) = − V (r. using ∂t · J (r.14).30) . t) = − 4 π Jt (r.21) and (8. t) contributes an electrical field component as well as the total magnetic field Et (r. t) Bt (r. The definitions of J and Jt applied to (8. t) = 0 . V (r.22) 1 ∂t 4π · J (r. t) + ∂t ρ(r.26) and (8.26) turns into the well-known wave equation 2 2 A(r. the name longitudinal electric field. such current densities exist. exist in the space considered. t) = 0 and from this follows · Jt (r.24) (8. t) = 0 . the name transverse fields.26) This equation does not couple anymore scalar and vector potentials.. Presently. Let J(r. t) holds. hence. respectively. For J also holds the continuity equation · J(r. we want to assume that no ring-type currents. 8. The vector potential determined through (8. In this case (8. t) = 0). t) (8. t) = − ∂t ρ(r.28) (8.23) This continuity equation identifies J (r.16) yield finally the electric and magnetic fields. V (r.12) and the Coulomb potential (8. t) = 0 (8.g. For the divergence of J (r.29) · Et (r. for example.

8. t) = ± ω2 ˆ |Ao |2 k .38) ˆ In this expression for the energy flux one can interprete k as the propagation velocity (note c = 1) and. hence.29).31) u is a unit vector (|ˆ| = 1) which.35) one obtains S(r. Note that in the units chosen the velocity of light is c = 1.31) and the resulting fields (8.35). t) × Re B(r. ω2 = |Ao |2 (8. A complete set of solutions is given by the so-called plane waves A(r. 8. are Et (r.28. ˆ We want to characterize now the radiation field connected with the plane wave solutions (8.34) (8. t) = i k × A(r. according to (8. (8. t) . t) . In applying the potential and fields to physical observables and processes we will only employ the real parts. 2 .31).37) ˆ ˆ where k is the unit vector k = k/|k|. at each point r and moment t.8.31. t) = ±i ω A(r. 8.2: Planar Electromagnetic Waves 207 which describes electromagnetic fields in vacuum.33) ωt) (8. The Coulomb gauge condition (8.39). 8π (8. are orthogonal to each other and are both orthogonal to the wave vector k. ˆ (8. 8. accordingly. This flux is given by 1 S(r. t) = Ao u exp i(k · r ˆ where the dispersion relationship |k| = ω (8. t) Bt (r.36) 4π Using the identity a × (b × c) = b (a · c) − c (a · b) and (8.35) are complex-valued quantities. (8.32) holds. 8.35) The vector potential in (8.34. t) = Re Et (r.12) yields u·k = 0. Et (r.34. see also the comment below Eqs.38. Obviously. The latter vector describes the direction of propagation of the energy flux connected with the plane wave electromagnetic radiation. t) = ± ω2 ˆ |Ao |2 k sin2 ( k · r − ωt ) 4π (8. the constant k is referred to as the wave vector. obviously. t) and Bt (r.32. Time average over one period 2π/ω yields S(r. The corresponding electric and magnetic fields.39) 8π The definition incoming waves and outgoing waves is rationalized below in the discussion following Eq. there exist two linearly ˆ u independent orientations for u corresponding to two independent planes of polarization. (8. 8.34.158). is orthogonal to k. (8. Here the “-” sign corresponds to so-called incoming waves and the “+” sign to outgoing waves2 . t) in (8. 8.

40) = V It should be pointed out that Nω represents the number of photons for a specific frequency ω. (8. hence.44) . A correct description of the electromagnetic field requires that the field be quantized. t) = H Ψ(r. Note that E (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. the energy of the plane wave is transported in the direction of −k. t) ˆ replacing the classical momentum p by the differential operator p = of the particle is then described by the Schr¨dinger equation o ˆ i ∂t Ψ(r.43). i (8.29) together with the potentials (8.3 Hamilton Operator The classical Hamiltonian for a particle of charge q in a scalar and vector potential V (r) and A(r.31) allows one finally to state for the planar wave vector potential A(r. respectively. t) 2m 2 + qV (r) .42) 8.39) and (8. We assume that the energy content of the fields is not altered significantly in the processes described and. a ˆ specific k and a specific u.208 Interaction of Radiation with Matter as the energy density. These photons each carry the energy ω. u · k = 0 .31). A ‘poor man’s’ quantization of the field is possible at this point by expressing the energy density (8.40) allows one to express then the field ˆ amplitudes 8πNω Ao = . 8. The integrals express the integration over the energy density of the fields. we will neglect the respective terms in the Hamiltonian (8. t) . whereas in the case of outgoing waves the energy is transported in the direction of k.8. t) 2m + 1 8π Ω∞ 2 H = + qV (r) 1 16π d3 r |Et |2 + |Bt |2 Ω∞ d3 r E 2 + . |k| = ω . Bt (r.28.27.31. (8. ˆ (8. 8. If we consider a volume V with a number of photons Nω the energy density is obviously Nω ω . defined below Eqs. (8.39) through the density of photons connected with the planar waves (8.16. t) is real 1 and that Et (r. 8.38) implies that for incoming waves.45) . (8. Comparision of (8. t) = 8πNω u exp i(k · r − ωt) ˆ ωV . t).43) Here the fields are defined through Eqs. We are left with a classical Hamiltonian function which has an obvious quantum mechanical analogue ˆ H = ˆ p − q A(r. is p − q A(r.31). (8. (8. The wave function Ψ(r.32).41) ωV Inserting this into (8. The sign in (8. t).

t) . In the classical case such question is mute since the gauge transformations do not alter the fields and. 8.50) The equivalence of (8. (8.11) to (8. t) and V (r. . Obviously. i.46).46) the fields enter. t) ˆ p eiqχ(r..47. (8. hence. 8.t)/ Ψ(r.10.11) affect the quantum mechanical description. t) . have no effect on the motion of the particle described by (8. not to phenomenological interactions like the molecular van der Waals interaction. e. however.t)/ Ψ(r.t)/ Ψ(r.48) implies that the gauge transformation (8. t) are necessary to compensate terms which arise through the phase factor.t)/ should not affect a system and that. t) and A(r.44. t) (8. 8. t) r (8.45) leads to the Schr¨dinger equation o   2 ˆ p − q A − q(( χ))   + qV − q((∂t χ))  Ψ(r.48) 2m For this purpose one notes i ∂t eiqχ(r. t) . (8. 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. that this principle applies only to fundamental interactions. t) = eiqχ(r. t) enter whereas in the classical equations of motion ˙ m¨ = q E(r. The idea just stated can be generalized by noting that multiplication by a phase factor eiqχ(r. accordingly.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)/ [ i ∂t − q((∂t χ)) ] Ψ(r. t) by a local and time-dependent phase factor eiqχ(r. 8.t)/ Ψ(r. by the group 3 The effect on other expectation values is not discussed here. t)|2 and.47) i ∂t Ψ(r. Applying the gauge transformations (8.47) is equivalent to   2 ˆ p − qA   i ∂t eiqχ(r.e. t) ˆ = eiqχ(r. an element of the group U(1). Elementary constituents of matter which are governed by other symmetry groups.51) 2m the potentials A(r. t) inside the double brackets. t) + q r × B(r. This leads to the question in how far the gauge transformations (8.g. such phase factor does not change the probability density |Ψ(r. t) =  + qV  eiqχ(r.t)/ p + q(( χ)) Ψ(r..11) of the potentials is equivalent to multiplying the wave function Ψ(r.t)/ . 8.8. It should be noted. t) =  + qV  Ψ(r.t)/ constitutes a unitary transformation of a scalar quantity. does not change expectation values which contain the probability densities3 . in the Schr¨dinger equation o   2 ˆ p − qA   i ∂t Ψ(r.10.49) (8. One can show that (8. t) =  2m where ((· · ·)) denotes derivatives in (( χ)) and ((∂t χ)) which are confined to the function χ(r. hence.10.

3. for Bo = Bo e3 in (8.57) satisfies · A = 0 and.52) This expression vanishes in the present case since since ˆ ˆ p · Af = A · pf and. The stationary homogeneous magnetic field B(r. (8. therefore. x1 .g. thereby.58) where ∂j = (∂/∂xj ). Accordingly.53) ˆ ˆ p·A + A·p + q2 2 A + qV 2m (8. j = 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. k2 .54) ˆ2 p q ˆ q2 2 − p·A + A + qV . consequently. We want to describe the stationary states corresponding to the Hamiltonian (8.57) with a homogeneous magnetic field Bo . t) = Bo .45) with Hamiltonian (8. holds (8. The vector potential (8. likewise can demand the existence of fields which compensate local transformations described by eiσ·χ(r. Using Cartesian coordinates this yields 2 H = − 2me 2 2 2 ∂1 + ∂2 + ∂3 + 2 eBo e2 Bo 2 x1 ∂2 + x i me 2me 1 (8. the so-called Landau gauge ˆ associates the vector potential AL (r) = Bo x1 e2 ˆ (8. The resulting fields are called Yang-Mills fields.55). The choice of a vector potential affects the form of the wave functions describing the eigenstates and. affects the complexity of the mathematical derivation of the wave functions. 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. In case of a homogeneous potential pointing in the x3 -direction. t). (8.55) 8. · A = 0 [cf.210 Interaction of Radiation with Matter SU(2). (8. In this o case holds V (r.. x2 .59) . x3 ) = exp(ik2 x2 + ik3 x3 ) φE (x1 ) . For this purpose we use the wave function in the form Ψ(E.56) due to the gauge freedom. The Hamiltonian (8.12)].55). t) ≡ 0. 2.4 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. (8.56). can be described by various vector potentials.t) where σ is the vector of Pauli matrices. e. the generators of SU(2).58). H = 2m m 2m (8. k3 . one can employ the Hamiltonian (8.

Unfortunately. Solution for Symmetric Gauge The solution obtained above has the advantage that the derivation is comparatively simple. like the corresponding gauge (8. 2 2me (8. The stationary Schr¨dinger equation (8. the wave function (8. (8. x1 .and x2 -coordinates.66) Obviously.65) where we replaced the parameter E by the integer n. k3 ) = ω(n + 2 k2 1 3 ) + . k3 .56) through the vector potential A(r) = 1 Bo × r .57). is not symmetric in the x1 . From this observation one can immediately conclude that the wave shifted (by e 3 function of the system.63) this induces a spread of the wave function in the x1 direction. 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 (8.64) is the classical Larmor frequency (c = 1).62) x1o = and where ω = k2 eBo eBo me (8. the familiar harmonic oscillator quantum number.65). (8. The energies corresponding to these states are E(n. therefore.8.60) Completing the square 2 2 e2 Bo 2 eBo k2 e2 Bo x1 + x1 = 2me me 2me x + k2 eBo 2 − 2 k2 2 2me (8. according to (8. according to (8.59). k2 . 2 (8.4: 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 ) . is Ψ(n. the states are degenerate in the quantum number k2 describing displacement along the x2 coordinate.65) which localize the electrons.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 ) .67) . the so-called symmetric gauge which expresses the homogeneous potential (8.62) is that of a displaced (by x1o ) harmonic oscillator with o 2 k 2 /2m ) energies.59). 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. However. Without affecting the energy one can form wave packets in terms of the solutions (8. We want to employ . x2 .63) (8. k2 .

72) to Hamiltonian (8. ˆ Identifying r × p with the angular momentum operator L.67) one can write ˆ p·A = Using · (u × v) = − u · function f (r) ×v + v· 2i · Bo × r .71). In this case the current density measures −e v oriented tangentially to the ring.73. y-plane with constant velocity v on a ring of radius r. ˆ (8. accordingly. (8. Vmag = −µclass · Bo (8.74).68) × u yields.212 Interaction of Radiation with Matter One can readily verify that this vector potential satisfies the condition (8.69) × u and × r = 0. an electron moving in the x.74) We consider a simple case to relate (8. for any (8.74) is in the present case 1 µclass = − e r v e3 .76) e 2me class class · Bo .72) allows one to interpret µ = − e L 2me (8. the Hamiltonian (8. Vmag = e 2me class = r me v e3 of the electron ˆ (8. The theory of classical electromagnetism predicts an analogue energy contribution .73) where µclass is the magnetic moment connected with a current density j µclass = 1 2 r × j(r) dr (8.71) (8. (8.12) for the Coulomb gauge. 8.70) ˆ ˆ p · A f = − Bo · p × r f . namely.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 .78) . the magnetic moment (8.77) Comparision with (8.75) 2 The latter can be related to the angular momentum µclass = − and. using × (uf ) = −u × f +f ˆ ˆ Bo · p × r f = Bo · r × p f . The latter can be rewritten. (8. in the present case of constant Bo . Accordingly. namely.72) and (8. For the vector potential (8. (8.

10.72) with an associated magnetic moment − g 2me S where g ≈ 2. likewise. (8. x3 ) = exp(ik3 x3 ) ψ(x1 .83) (8.88) me ω xj . x2 ) .56) pointing in the x3 -direction the symmetric gauge (8. (8. described by the operator S.85) The Hamiltonian (8. 2me We have used here the expression for the angular momentum operator ˆ L3 = ( /i)(x1 ∂2 − x2 ∂1 ) .86) 1 2 2me 1 2 ˆ as well as of the angular momentum operator L3 .71) is ˆ H = 2 ˆ2 p e2 Bo + 2me 8me x2 + x2 + 1 2 eBo L3 .79) To obtain the stationary states. we separate the variable x1 . 10 that the spin of the electron. x2 ) = E ψ(x1 . the solutions of ˆ H ΨE (x1 .87) .80) (8. for Bo = Bo e3 . requires a Lorentz-invariant quantum mechanical description as provided in Sect.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 .4: Electron in Homogenous Magnetic Field 213 as the quantum mechanical magnetic moment operator for the electron (charge −e).82) 1 ω (x1 ∂2 − x2 ∂1 ) i 2me + 1 me ω 2 x2 + x2 1 2 2 2 k2 3 + (8.8. the Hamiltonian (8. x2 from x3 setting ΨE (x1 . e gives rise to an energy contribution (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. In this case. For a magnetic field (8. Accordingly. x2 ) obey then ˆ Ho ψ(x1 . We will demonstrate in Sect. x2 ) where ˆ Ho = − 2 2 2 ∂1 + ∂2 (8. x3 ) .84) . 2 . 2me (8. x2 .e. ˆ i. j = 1. x2 . The functions ψ(x1 . x2 .83) describes two identical oscillators along the x1 -and x2 -directions which ˆ are coupled through the angular momentum operator L3 . x3 ) = E ΨE (x1 . the so-called gyromagnetic ratio of the electron. To obtain these eigenstates we introduce the customary dimensionless variables of the harmonic oscillator Xj = (8. E = E − (8.e.81) (8.. i. A derivation of his property and the value of g.and the ±x2 -direction.67) yields a more symmetric solution which decays to zero along both the ±x1 .

0. x2 ) . j = 1.e. +j . x1 . 1 3 . through a combination of states j Ψ (j.94) is also an eigenstate of L3 . i. . 1. 2 2 . x2 ) (8. m . aj = √ 2 Interaction of Radiation with Matter Xj + ∂ ∂Xj .94) is an eigenstate of the vibrational Hamiltonian. .90) a† a2 − a† a1 1 2 .97) . (8. (8. x1 ..92) are not eigenstates of L3 . m ..94) is an eigenstate of Ho ˆ Ho Ψ (j. therefore. m as follows: j = 0. In order to cover all posible vibrational quantum numbers one needs to choose j. . We. (8.214 Employing the creation and annihilation operators 1 a† = √ j 2 and the identity ˆ ω L3 = which can readily be proven. (8. . x1 .91) We note that the operator a† a2 − a† a1 leaves the total number of vibrational quanta invariant.94) Since this state is a linear combination of states which all have vibrational energy (2j + 1) ω.92) represents a state with j + m quanta in the x1 -oscillator and j − m quanta in the x2 -oscillator. m. (8. x2 ) (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. If this property is. x2 ) = m=−j αmm Ψ(j. x2 ) . i. m . x2 ) = a† 1 j−m (j + m)! (j − m)! Ψ(0. however. . 1 2 since one phonon is annihilated and one created. x2 ) = ω 2j + 2m + 1 Ψ (j. x1 . obeyed.93) ˆ The states (8. m .92) where Ψ(0. x1 . x2 ) . m . (8. x1 .96) ˆ holds. (8. m . m = −j. x1 .89) . 2 (8.95) (j) ˆ We want to choose the coefficients αmm such that (8. m . m.e. in fact. x1 . x1 . Such eigenstates can be expressed. . x2 ) = 2m Ψ (j. it holds a† a1 + a† a2 + 1 1 1 2 Ψ (j. .as well as for the x2 -oscillator. x1 . x2 ) = ( 2j + 1 ) Ψ (j. (j) (8. the total vibrational energy being ω(2j + 1). −j + 1. such that 1 i a† a2 − a† a1 1 2 Ψ (j. (8. . x2 ) is the wave function for the state with zero vibrational quanta for the x1 . 0.

a transformation moving the x3 –axis into the x2 – 2 axis.11. x2 ) .288)] 1 ˆ J3 = 2 a† a1 − a† a2 1 2 (8.31).5. except for a factor 2 . the eigenstates of (8. is identical to the operator L3 introduced in (8. an example is the Coulomb potential V (r. x2 ) = Dmm ( .. i. x1 . cf.5: Time-Dependent Perturbation Theory (j) 215 In order to obtain coefficients αmm we can profitably employ the construction of angular momentum states in terms of spin– 1 states as presented in Sects. (8. According to Sect.5. The required rotation must transform the x3 –axis into the x2 –axis. 5.84) above. If we identify 2 a† . x1 . 8. The external radiation field is accounted for by the vector potential A(r. thus. π . m. m) in Sect.8. x2 ) defined in (8.9.11.e. This implies that ˆ we can obtain eigenstates of L3 by rotation of the states Ψ(j. Other radiation fields can . Using the explicit form of the Wigner rotation matrix as given in (5.10. 5.11 (8.9. b− + − notation in Sects.83) and confirmed the eigenvalues stated in (8.5. 5. 5. m .10. x2 ). m.5.10. 5. Eq. x1 .10. x2 ) (8. The first rotation contributes a factor exp(−im π ). t) = 1/4πr confining an electron with energy E < 0 to move in the well known orbitals of the hydrogen atom. According to the derivation given there.92) correspond to the eigenstates |Ψ(j. .101) 2 2 where Dmm ( π .55) confines the particle to stationary bound states. m.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 (8.98) then the states Ψ(j. x1 .102) We have identified.. m.309) yields finally Ψ (j. In the simplest case the radiation field consists of a single planar electromagnetic wave described through the potential (8. b+ . 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 . x1 . which corresponds there to the angular momentum in the x2 –direction. the 2 2 latter representing the Wigner rotation matrix of Sect. the states are eigenstates of the operator [we use for the operator the notation of Sect. m .55). 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. a1 . a† . is in the notation of the present section 1 ˆ a† a2 − a† a1 .99) with eigenvalue m.100) J2 = 1 2 2i 1 ˆ i.97).9. 5. (8. 0) Ψ(j. The connection with the present problem arises due to the fact that the operator J2 in Sect. t) introduced above.(5.11 such transformation is provided through π π (j) Ψ (j. a2 1 2 present notation ←→ b† . We assume that the scalar potential V in (8. b† .e. 5. x1 . the second rotation a factor dmm ( π ).

t) − m 2m (8. In fact. in this case.e. Here the so-called unperturbed system is governed by the Hamiltonian Ho with stationary states defined through the eigenvalue problem Ho |n = n |n . (8. the term ∼ p · A(r. t). .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. the term ∼ A second term in (8. Estimate of the Magnitude of VS We want to demonstrate now that the interaction VS (t). We first note. t) + A (r.e. 2 . . orthonormal basis. n = 0. as given in (8. is much 2 (r.103) (8. (8.e. we discount the possibility of a continuum of eigenstates.e.. i. ωV (8. i. .108) We assume for the sake of simplicity that the eigenstates of Ho can be labeled through integers. The states |n are thought to form a complete.42).107) 1 = 1 n=0 |n n| . t). and that the first term in (8. For an electron charge q = −e and an electron mass m = me one can provide the estimate for the first term of (8.216 Interaction of Radiation with Matter be expanded through Fourier analysis in terms of such plane waves.. 1.105) where A(r. using (8. can be considered a weak perturbation. The Hamiltonian of the particle in the radiation field is then described through the Hamiltonian H Ho VS = = = Ho + VS ˆ2 p + qV 2m q ˆ q2 2 p · A(r.105) as follows4 . However.105) in further calculations and to expand the wave function in terms of powers of VS (t) in a perturbation calculation. i... t) is given by (8.105). one can estimate that the perturbation. This result will allow us to neglect the larger than the second term. we assume n|m = δnm and for the identity ∞ (8.109) The reader should note that the estimates are very crude since we are establishing an order of magnitude estimate only. i.106) where we adopted the Dirac notation for the states of the quantum system. is much smaller than the eigenvalue differences near ˆ typical atomic bound states.104) (8.41) e ˆ e p2 p·A ∼ 2me me me 2me 4 1 2 8πNω . this assumption can be waved as our results below will not depend on it.

m (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..111) For ω = 3 eV and a corresponding λ = 4000 ˚ one obtains.112) π λ me c2 and with e2 /ao ≈ 27 eV. and me c2 ≈ A A 500 keV 2 ao ω ≈ 10−8 (8.110) where ao is the Bohr radius. e ˆ p · A ∼ 10 eV · 10−4 = 10−3 eV .105) for radiation fields can be considered a small perturbation. one can neglect this term as long as the first term gives non-vanishing contributions. i.118) .105) due to a radiation field by VS = − q ˆ p · A(r.42) it holds VS ≈ e m k. (8. u) p · u exp i(k · r − ωt) ˆ ˆ ˆ kV .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. a volume V = λ3 where λ is the wave length corresponding to a plane wave with frequency ω.. me (8. λ = 2πc/λ. Hence. 2me (8. Using again (8.41) one can state e2 e2 1 8πNω ω A2 ∼ 2me 2me ω 2 V For the same assumptions as above one obtains e2 e2 A2 ∼ · 2me 8πao ao 4 ω λ me c2 .116) This term is obviously much smaller than the first term. Consequently. and as long as the photon densities Nω /V are small.114) Employing for the second factor the estimate as stated in (8. Nω = 1.e. Assuming a single photon. We can. i.117) In case that such perturbation acts on an electron and is due to superpositions of planar waves described through the vector potential (8. replace the perturbation (8.ˆ u 4πNk α(k.e.8. hence. with ao ≈ 0. We want to estimate now the second term in (8.115) (8. t) . (8. altogether. the first term in (8.5: 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 (8.5 ˚. one obtains for (8.105).112) yields e2 A2 ∼ 10 eV · 10−8 = 10−7 eV .

|Ψ(to ) = |0 (8. u) has been added to describe eliptically or circularly polarized ˆ waves. Multiplying the latter equation by the operator exp i (8. n = 1.120) (8. .. .123) i = Ho exp − Ho (t − to ) (8. 2. (8. Using i i ∂t exp − Ho (t − to ) and (8.218 Interaction of Radiation with Matter where we have replaced ω in (8. |n are defined in (8.126) 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. describes the effect of a radiation field on an electron system and which will be assumed below to describe radiative transitions.42) through k = |k| = ω.121) (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) .127) i ∂t |ΨD (t) = VD (t) |ΨD (t) .106–8. One seeks to predict the probability to observe the particle in one of the states |n . The sum runs over all possible k vectors and might actually be an integral. Equation (8. For this purpose one needs to determine the state |ΨS (t) of the particle. n = 0 at some later time t ≥ to .125) (8. Perturbation Expansion The generic situation we attempt to describe entails a particle at time t = to in a state |0 and a radiation field beginning to act at t = to on the particle promoting it into some of the other states |n . The probability to find the particle in the state |n at time t is then p0→n (t) = | n|ΨS (t) |2 .124) (8. This state obeys the Schr¨dinger equation o i ∂t |ΨS (t) = [ Ho + VS (t) ] |ΨS (t) subject to the initial condition |ΨS (to ) = |0 . A factor α(k.118) is the form of the perturbation which.122) where |ΨD (to ) = |0 . . under ordinary circumstances. The states |0 . the sum over u involves the two possible polarizations of planar ˆ electromagnetic waves.108) as the eigenstates of the unperturbed Hamiltonian Ho .

Equations (8. 0 VD (t) |ΨD (t) (1) VD (t) |ΨD (t) (0) (8.133. .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. 8.137) are consistent with (8.133) (8.133–8. (8. 2.136) together with the initial conditions |ψD (to ) = |0 0 for n = 0 for n = 1. From (8. . n| (8. . consequently.127.139) . .137) |ΨD (t) = (1) (0) (8.132–8.131) p0→n (t) = | n |ΨD (t) |2 . We will consider here only the two leading contributions to |ΨD (t) . Accord(n) ingly. In order to determine |ΨD (t) described through (8. we define |ΨD (t) through the evolution equations i ∂t |ΨD (t) i ∂t |ΨD (t) i (2) ∂t |ΨD (t) (1) (0) = = = .128).138) 1 i t dt VD (t ) |0 .130) = exp − i n (t − to ) − to )]| = 1.8. 8. .137) One can readily verify that (8. 8.128) We note that the transition probability (8.121) expressed in terms of ΨD (t) is i p0→n (t) = | n|exp − Ho (t − to ) |ΨD (t) |2 .137) follows |ΨD (t) = |0 .5: Time-Dependent Perturbation Theory where VD (t) = exp 219 i i Ho (t − to ) VS (t) exp − Ho (t − to ) .134) (8. to (8. using |exp[− i n (t n (8. . = .132) where |ΨD (t) accounts for the contribution due to n-fold products of VD (t) to |ΨD (t) . (8. .137) can be solved recursively.134. Employing this result one obtains for (8. (8.129) n| and.135) i ∂t |ΨD (t) (n) VD (t) |ΨD (n−1) (t) (8.

. in turn..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 .135. λ → 0+ . (8. yields for (8. defines the direction of the E-field of the ˆ radiation. II III (8. is VS = ˆ ˆ V1 exp(−iω1 t) + V2 exp( iω2 t) eλt .. non-singular. according to (8.220 This result. The factor III in (8. This factor will serve mainly the purpose of keeping in the following derivation all mathematical quantities properly behaved.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. 8.41).142) combining an incoming and an incoming or outgoing wave. according to (8. (8. note that uj . 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. For the sake of including the effect of superpositions of plane waves we will assume. 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.118) above that the effect of a radiation field on an electronic system is accounted for by perturbations with a so-called harmonic time dependence ∼ exp(−iωt).142).e. We want to apply now the perturbation expansion derived to such perturbations. The resulting perturbation on an electron system.118). i. . + m=0 dt to to dt n|VD (t )|m m|VD (t ) |0 8.140) 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. In (8.6 Perturbations due to Electromagnetic Radiation We had identified in Eq. The coefficients A1 . 8. A2 are defined through (8.34. the direction of ˆ the propagation being determined by k = k/|k|.141) + . that two planar waves simulataneously interact with an electronic system. A factor exp(λt).144) describes the propagation of the planar wave.137) |ΨD (t) = (2) Interaction of Radiation with Matter 1 i 2 t t dt to to dt VD (t ) VD (t ) |0 . λ → 0+ has been introduced which describes that at time to → −∞ the perturbation is turned on gradually. such that the combined radiation field is decribed by the vector potential A(r. however.144) r is the position of the electron and p = ( /i) is the momentum operator of the electron.

6: Perturbations due to Electromagnetic Radiation 1st Order Processes 221 We employ now the perturbation (8.140. According to (8.141). 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 (8.150) and.149) Using the definition of VD stated in (8. (8.8.128).145) n|ΨD (t) (1) m (t − to ) |m . employing the perturbation (8.143).128) one obtains k|VD (t) | = k|VS (t)| exp i ( k − ) (8.147) ˆ ˆ n|V1 |0 e−iω1 t + n|V2 |0 e Carrying out the time integration and taking the limit limt→−∞ yields (1) n|ΨD (t) = λ→0+ lim e λt exp ˆ n|V1 |0 o − o − ω1 ) t + + ω1 − n + iλ ( n i i ˆ + n|V2 |0 2nd Order Processes exp o − o ω2 ) t ± ω2 − n + iλ ( n . 8.145) we obtain then. (8.148) We consider now the 2nd order contribution to the transition amplitude.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 . using (8.130) and (for m = 0) i i exp − Ho (t − to ) |m = exp − for (8. (8.151) ˆ ˆ n|V1 |m m|V1 |0 exp ˆ ˆ + n|V2 |m m|V2 |0 exp i i n − − m − ω1 − i λ)t exp ω2 − i λ)t exp m o n m m o .146) = 1 λ→0+ t→−∞ i lim lim × t dt exp to i ( n − o − i λ) t iω2 t × . For the 1st order contribution to the transition amplitude n|ΨD (t) = (1) 1 i t dt n|VD (t ) |0 to (8. (8.143) to the expansion (8. (8.

e.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. (8.152).148) and (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. the time derivative of p0→n (t).155) λ→0+ ( o We are actually interested in the transition rate..153) (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 ) (8.131). Using (8. i.148) and ∗ |z1 + z2 |2 = |z1 |2 + |z2 |2 + 2Re(z1 z2 ) (1) (8. For this rate holds . We also assume for the final state n = 0 such that n|0 = 0 and p0→n (t) = | n|ΨD (t) |2 holds. We assume first that the first order transi(1) tion amplitude n|ΨD (t) is non-zero. provide now the transition probability p0→n (t) according to Eq. in which case one can expect that it is larger than the 2nd (2) order contribution n|ΨD (t) which we will neglect. respectively.

In case the matrix elements are non-zero.143. of this expression describe the individual effects of the two planar wave contributions of the perturbation (8. due to the vector potential A1 u1 exp i (k1 · r − ω1 t) .6: Perturbations due to Electromagnetic Radiation d p0→n (t) dt lim e2λt + × 2 Re ˆ 2 λ n|V1 |0 |2 ( o + ω1 − n )2 + (λ )2 ˆ 2 λ n|V2 |0 |2 d + 2λ + × ( o ± ω1 − n )2 + (λ )2 dt ˆ ˆ n|V1 |0 0|V2 |n exp i (± ω2 − ω1 ) t + ω1 − n + iλ ) ( o ± ω2 − n − iλ ) 223 = λ→0+ (8. The second contribution to (8.142) ˆ ˆ The matrix elements n|V1 |0 and n|V2 |0 in (8.e. however. The result supports our definition of incoming and outgoing waves in (8. to cancel the third term in (8. .158) play an essential role for the transition rates of radiative transitions. Using lim x2 + 2 δ→0+ = π δ(x) (8.157) 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 (8.159) leads to final states |n with energy n = o + ω1 .156) such that the 1st order contributions of the two planar waves of the perturbation simply add. The time average will be denoted by · · · t . describing either an incoming or an outgoing plane wave due to the vector potential A2 u2 exp i (k1 · r ˆ ω2 t) . The δ-functions appearing in this expression reflect energy conservation: the incoming plane wave contribution of (8.158) ˆ + | n|V2 |0 |2 δ( − ω2 ) Obviously. Hence.160) leads to final states |n with energy n = o ± ω2 .156) ( o The period of electromagnetic radiation absorbed by electronic systems in atoms is of the order 10−17 s..8.31) and (8. 8. which is observed through the so-called spectral intensities of transitions |0 → |n . ˆ (8. For the resulting expression the limit limλ→0+ can be taken. a property. First.142–8.144).158). 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. i. The result of such average is. (8. the two terms apearing on the rhs. one should average the rate over many periods of the radiation. in fact. any attempt to do so.144). the matrix elements can vary strongly for different states |n of the system. due to the uncertainty relationship would introduce a considerable perturbation to the system. is much shorter than could be resolved in any observation.

.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. hence. |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 (8.158).157) yields. it holds p0→n (t) = | n|ΨD (t) |2 .167) Taking now the limit limλ→0+ and using (8.163) × (8.162) × and. i.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.. as we did in the the case of 1st order transitions. (8.e.153) vanishes such that the leading contribution to the transition probability p0→n (t) arises from the 2nd order amplitude (8.153–8. in analogy to (8.k=1 j=k 2 2 2 2 ∗ zj zk (8..158). ∞ − 2 ω1 − 2i λ)t n − o − 2 ω1 − 2i λ ( n o − z2 = m=0 ˆ ˆ n|V2 |m m|V2 |0 ω2 − i λ m− o exp i × (8.166) over many periods of the radiation yields exp[ i (±ω2 − ω1 )] t = 0 and. i.168) absorption of 2 photons ω1 . exp[ i (±ω2 − ω1 )].164) (8.152).g. We define ∞ z1 = m=0 ˆ ˆ n|V1 |m m|V1 |0 m − o − ω1 − i λ exp i × (8. Time average · · · t of expression (8.e.161) To determine the transition rate we proceed again. 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. similarly. (2) (8. in Eqs. e.

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

φ) = 2 − 1 2 1 −r/ao e Y00 (θ. . according to (8. However. plane polarized wave described through the complex vector potential A(r. to → −∞ (8.226 Absorption of a Plane Polarized Wave Interaction of Radiation with Matter We consider first the case of absorption of a monochromatic. (8. the second term can be discounted in case of absorption. =0. 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 .170) We will employ only the real part of this potential. i. . .143) will contribute to the absorption process.178) These wave functions make significant contributions to this integral only for r-values in the range r < 10 ao .e.173) V1.144) VS = where e 2πN ˆ ˆ p · u e±ik·r .175) The matrix element involves a spatial integral over the electronic wave functions associated with states |n and |0 . For example. in case of a radiative transition from the 1s state of hydrogen to one of its three 2p states.171) The perturbation on an atomic electron system is then according to (8. φ) × × eik·r e−r/ao Y00 (θ. and the integral is =1. φ) = 6 r −r/2ao e Y1m (θ.143. the wave functions are (n.m=0 (r. The absorption rate.172) δ( n − o − ω) (8. (8. ˆ (8.177) √ M = 6 0 ∞ 1 2π ia4 o r2 dr −1 dcosθ 0 ∗ dφ r e−r/2ao Y1m (θ. t) = 8πN ωV u exp ˆ ı (k · r − ωt) . φ) a3 ao o 2p (8. (8. t) = 2πN ωV u exp ˆ ı (k · r − ωt) + 2πN ωV u exp ˆ ı (−k · r + ωt) . in this range one can expand eik·r ≈ 1 + ik · r + . φ) (8. the vector potential actually assumed is A(r.176) ψn=2. φ) a3 o 1s (8. m denote the relevant quantum numbers) ψn=1.174) We seek to evaluate the matrix element ˆ M = n| p eik·r |0 .. λ → 0+ .158). θ.2 = m ωV Only the first term of (8. θ. 8. (8.179) .m (r.

k=1 i m 3 pk ek = ˆ j. 2me r (8.183) (8.104) and. is Ho = ˆ (p)2 e2 + V (r) . Using (8. V (r) = − . Transition Dipole Moment A further simplification of the matrix element (8.181) and (8. Ho ] |0 = i i n) n|r |0 .187) . Ho ] = i m 3 pk ej δjk = ˆ j. Ho ] = ˆ2 p [ r. We are now in a position to obtain an alternative expression for the matrix element (8. in case of the hydrogen atom. pk ] pk ˆ k=1 (8. V (r) ] 2me =0 = Using r = 3 ˆ j=1 xj ej 1 2me 3 pk [ r.182) (8.529 ˚ one concludes that in the significant integration range in (8.182).7: One-Photon Emission and Absorption 227 One can estimate that the absolute magnitude of the second term in (8.180) ∆E2p−1s and ao = 0.184) and the commutation property [xk . pk ] + ˆ ˆ k=1 1 2me 3 [ r. the value of the wave length for the 1s → 2p transition 2π c λ = = 1216 ˚ A (8.186) δ( n − o − ω) (8. One refers to this approximation as the dipole approximation. Using |k| = 2π/λ. Ho ] i where Ho is the Hamiltonian given by (8.k=1 i ˆ p m (8.182) one obtains M ≈ Insertion into (8.179) and other terms are never larger than 20π ao /λ.175).174) yields kabs = 4π 2 e2 N ω V u · n| ˆ |0 ˆ r 2 m( o − m n| [r. ] + [ r.178) holds eik·r ≈ A 1 1 + O( 50 ) such that one can approximate eik·r ≈ 1 .8. (8.182) one finds [ r.175) can then be ˆ achieved and the differential operator p = i replaced by by the simpler multiplicative operator r. This simplification results from the identity m ˆ p = [ r.185) from which follows (8.181) For the commutator in (8. pj ] = i δkj one obtains ˆ [ r.

a very hot oven. and (2) the fact that strictly monochromatic radiation cannot be prepared such that any radiation source provides radiation with a frequency distribution.174) one can replace n − o by ω. 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.191) (8. x3 -axes of a right-handed cartesian vectors u1 . u2 and k ˆ ˆ coordinate system. Absorption of Thermal Radiation We want to assume now that the hydrogen atom is placed in an evironment which is sufficiently hot.193) . hence.35). The unit ˆ ˆ ˆ ˆ can be chosen to point along the x1 . ρ) and ˆ u β = ∠(ˆ2 . i. obtain the transition rate in the present case by adding the individual transition rates of all planar waves present in the oven. The δ-function appearing in this expression.e. ϕ1 = 0 and of u2 by ϑ2 = π/2. according to (8. Accordingly. three-dimensional vector..228 Interaction of Radiation with Matter where we used the fact that due to the δ-function factor in (8.189) where the factor 1/ arose from the integral over the δ-function. will actually be replaced by a distribution function which reflects (1) the finite life time of the states |n . The direction of this vector in the u1 . the direction of u1 is described by the ˆ ˆ ˆ ˆ angles ϑ1 = π/2.33) are orthogonal to each other. ρ) holds then u cosα = cosϑ1 cosϑ + sinϑ1 sinϑ cos(ϕ1 − ϕ) = sinϑcosϕ and cosβ = cosϑ2 cosϑ + sinϑ2 sinϑ cos(ϕ2 − ϕ) = sinϑsinϕ . ˆ u ˆ (8. We can. 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. planar waves as defined in (8. We have demonstrated in our derivation of the rate of one-photon processes (8. one can express | u · n| r |0 |2 = |ρ|2 ( cos2 α + cos2 β ) = sin2 θ . directions.158) above that in first order the contributions of all components of the radiation field add.190) (8. k frame is described by the angles ϑ. in practical situations. |0 .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. and all polarizations of the radiation. k ˆ As a result. there are ony two linearly independent directions of u possible. x2 . such that the thermal radiation present supplies a continuum of frequencies. Integrating and summing accordingly over all contributions as given by (8.192) 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 (8.31–8. ϕ. ϕ2 = π/2. u2 .188) ˆ Here dk is the integral over all orientations of k. ˆ In order to carry out the integral dk we note that u describes the possible polarizations of the ˆ ˆ and u. say u1 and u2 . For the two angles α = ∠(ˆ1 .

2 . For practical evaluations we provide an expression which eliminates the physical constants and allows one to determine numerical values readily. a quantum mechanical treatment of the radiation field leads to a total emission rate which is proportional to Nω + 1 where Nω is the number of photons before emission. In case of a classical radiation field one would expect that emission cannot occur. there is also a contribution to the emission rate which is proportional to Nω .. that emission occurs even if no photon is present in the environment. However.37 × 1019 × 3 3 c3 3 ao λ3 s λ and kabs where λ = 2πc n − (8. This dependence predicts.37 × 1019 1 ao | n|r|0 |2 × s λ λ2 .197) o (tot) (8.194) combined are typically somewhat smaller than (1 ˚/1000 ˚)3 = A A 10−9 . 8. The quantum nature of radiation manifests itself in that the number of photons Nω msut be an integer. .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 . a problem in case of emission by quantum systems in complete darkness. by planar waves with vector potential (8. The most intensely absorbing molecules are long.7: One-Photon Emission and Absorption 229 This geometrical average. The corresponding process is termed spontaneous emission. Accordingly.. Emission of Radiation We now consider the rate of emission of a photon. However.195) = Nω 1. can be inserted into (8. finally.171) and perturbation (8..194) for the absorption rate shows that the essential property of a molecule which determines the absorption rate is the so-called transition dipole moment | n| r |0 |. for Nω = 0.e. . in particular.189) to yield the total absorption rate kabs (tot) = Nω 4 e2 ω 3 | n|r|0 |2 3 c3 .e. The value of | n| r |0 | determines the strength of an optical transition. (8.173) contributes. This poses. For this purpose we use ω/c = 2π/λ and obtain 4 e2 ω 3 32π 3 e2 ao 1 ao = = 1. the calculation of the emission rate proceeds as in the case of absorption. i.8. Nω = 0. however. Otherwise. i. for atoms or symmetric molecules. linear molecules. ˆ In case of emission only the second term V2 exp(+iωt) in (8. Nω photons before absorption.196) The last two factors in (8. in agreement with observations. Transition Dipole Moment The expression (8. as for the absorption process. the resulting total rate of emission bears a different dependence on the number of photons present in the environment. This difference between emission and absorption is due to the quantum nature of the radiation field. 1. The transition dipole moment can vanish for many transitions between stationary states of a quantum system. The radiation field is described.173). (8. .172. the absorption rate is of the order of 109 s−1 or 1/nanosecond. However.

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

dropping the index 1 characterizing the radiation.207) Expression (8.207) does not converge rapidly. The resulting rate of the transition depends on 2 Nω . (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 × (8.205) The respective radiative transition is of 2nd order as described by the transition rate (8. The electron absorbs the radiation and emits immediately a second photon.207) one is.8.206) − o − 2 ω) .181) and using (8.207) does not provide a suitable avenue of computing the rates of 2-photon transitions. however. but then requires the absorption of two photons. ˆ (8. 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 (8.207) 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. (8. . We wish to describe an observation in which a detector is placed at a solid angle element dΩ2 = sinθ2 dθ2 dφ2 with respect to the origin of the coordinate system in which the electron is described. (8. Employing the dipole approximation (8. faced with the dilemma of having to sum over all intermediate states |m of the system. 8.187). however. If the sum in (8. in case ˆ of absorption only V1 contributes. under ordinary circumstances.104) with stationary states |n defined through (8. In this case one needs to choose the energy of the photons to obey n = o + 2 ω. induce single photon absorption processes as described by the transition rate Eq. The intense radiation fields of lasers allow one to increase transition rates to levels which can readily be observed in the laboratory. However.168) where the first term describes the relevant contribution. as described through the vector potential ˆ A(r. a transition between the states |0 and |n may be possible.208) has been prepared.173). finally.8: Two-Photon Processes Two-Photon Absorption 231 The interaction of electrons with radiation. One obtains. In applying (8.172. 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 (8.106).182) yields. 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) . even if n| r |0 vanishes. which is not necessarily the case. We assume that a planar wave with wave vector k1 and polarization u1 . The transition requires that the transition dipole moment n| r |0 does not vanish for two states |0 and |n . then expression (8.

Let us assume for the present that the emitted photon has a wave vector ˆ k2 with cartesian components   sinθ2 cosφ2 k2 = k2  sinθ2 sinφ2  (8.6 above that the absorption of the radiation in (8.211) is o2 A− = o2 8π (N2 + 1) ω1 V (8. for the components of the wave (8.211) accounts for the emitted wave and. i. the amplititude A− defined in (8. The second term in (8.209) cosθ2 where the value of k2 has been fixed.212) where N1 /V is the density of photons for the wave described by (8.211) is stated in Eq.214) . u2 .210) The vector potential which describes both incoming wave and outgoing wave is a superposition of the potentials in (8. ˆ The perturbation which arises due to the vector potential (8. however.210) A(r. in the present case.211) The first term describes the absorption of a photon and. ˆ (8. (8.210). 8. 8.232 Interaction of Radiation with Matter We assume that the experimental set-up also includes a polarizer which selects only radiation with a certain polarization u2 .e.213) where N2 /V is the density of photons characterized through k2 .208. t) = Ao2 u2 cos(k2 · r − ω2 t) .210) is accounted for by the following contributions of (8.210). The relevant terms of the perturbation (8. The vector potential describing the emitted plane wave is then A(r..141). u1 . later we will allow the quantum system to select appropriate values.105) using the vector potantial (8. This expansion yields. the amplitude A+ is given by o1 A+ = o1 8πN1 ω1 V (8.208) and the emission of the radiation in (8. according to the ˆ description of emission processes on page 229. o1 ˆ o2 ˆ (8. In the present case we consider only scattering processes which absorb radiation corresponding to the vector potential (8. We know already from our description in Section 8.208).208. hence.105).208) and emit 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. t) = A+ u1 exp[ i (k1 · r − ω1 t) ] + A− u2 exp[ i (k2 · r − ω2 t) ] . the wave characterized through k1 .

Accordingly.217) 2 Here ro denotes the classical electron radius ro = e2 = 2. one needs to carry out the integration 2 V(2π)−3 k2 dk2 · · ·. Only the second (1st order) and the third (2nd order) terms in (8.215) t dt ei( to n− o− ω1 + ω2 +i λ)t 1 i 2 e2 4m2 e A+ A− × o1 o2 × ˆ ˆ u1 · n| p |m u2 · m| p |0 × ˆ ˆ t t × to dt to dt ei( n− m− ω1 +i λ)t i( e m− o+ ω2 +i λ)t ˆ ˆ + u2 · n| p |m u1 · m| p |0 × ˆ ˆ t t × to dt to dt ei( n− m+ ω2 +i λ)t i( e m− o− ω1 +i λ)t We have adopted the dipole approximation (8.215) correspond to scattering processes in which the radiation field ‘looses’ a photon ω1 and ‘gains’ a photon ω2 . Hence.. only these two terms contribute to the scattering amplitude. 8.218) . me c2 (8. Inserting also the values (8.8. i.213) for the amplitudes A+ and A− results o1 o2 in the Kramers-Heisenberg formula for the scattering rate k = N1 c 2 ω2 r (N2 + 1) dΩ2 u1 · u2 n|0 ˆ ˆ V o ω1 1 n|ˆ2 · p |m m|ˆ1 · p |0 u ˆ u ˆ n|ˆ1 · p |m m|ˆ2 · p |0 u ˆ u ˆ − + me m m − o + ω2 m − o − ω1 (8.e.216) over all available modes of the field.181) in stating this result. one needs to integrate the rate as given by (8. i..8 · 10−15 m .212. 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. evaluating the time integrals in (8.e.216) e2 + − A A 4me o1 o2 n|ˆ1 · p |m m|ˆ2 · p |0 u ˆ u ˆ n|ˆ2 · p |m m|ˆ1 · p |0 u ˆ u ˆ + m − o + ω2 m − o − ω1 We now note that the quantum system has the freedom to interact with any component of the radiation field to produce the emitted photon ω2 .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.

− o )2 ( m − o )3 (8.223) m Condition (8. 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.228) S (0) = m S (0) = 2 m . it follows 2 dσ = ro (N2 + 1) dΩ2 |ˆ1 · u2 − S( ω) |2 u ˆ (8. in fact. consequently. i.221) suggests to expand S( ω) S( ω) = S(0) + S (0) ω + Using 1 − o ± ω one can readily determine m (8.e.. .. one can relate (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.224) = m 1 − o ( m ω ( ω)2 + + .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. ω1 = ω2 .219) (8.221) Using |n = |0 and. 2 .234 Interaction of Radiation with Matter The factor N1 c/V can be interpreted as the flux of incoming photons. for all states |m of the electronic system (8.217) to the scattering cross section defined through dσ = It holds then dσ = 2 ro rate of photons arriving in the the solid angle element dΩ2 flux of incoming photons ω2 (N2 + 1) dΩ2 u1 · u2 n|0 ˆ ˆ ω1 1 n|ˆ1 · p |m m|ˆ2 · p |0 u ˆ u ˆ me m m (8.225) 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. (8. Accordingly.222) where S( ω) = 1 me 0|ˆ1 · p |m m|ˆ2 · p |0 u ˆ u ˆ 0|ˆ2 · p |m m|ˆ1 · p |0 u ˆ u ˆ + m − o + ω m − o − ω 1 S (0)( ω)2 + . .226) u ˆ 0|ˆ2 · p |m m|ˆ1 · p |0 u ˆ u ˆ 0|ˆ1 · p |m m|ˆ2 · p |0 u ˆ − me ( m − o )2 me ( m − o )2 (8. 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. one assumes the even stronger condition ω1 << | o − m| .

e.227) vanishes. honey bees.233) m Employing again (8.108) this is 1 0|ˆ1 ·r u2 · p − u2 · p u1 ·r |0 .222) of the cross section for Rayleigh scattering dσ = 2 ro m2 ω 4 (N2 + 1) dΩ2 × e (8.182) 0|ˆ1 · p |m u ˆ 1 = 0|ˆ1 ·r |m u me ( m − o ) i This transforms (8.228) we obtain ˆ ˆ ˆ ˆ S (0) = 2me 2 m 0|ˆ2 ·r |m m|ˆ1 ·r |0 u u 0|ˆ1 ·r |m m|ˆ2 ·r |0 u u + m − o m − o (8.229) both for the u1 · p and the to u2 · p terms in (8.222). For this purpose we replace p using (8. It explains. ˆ ˆ S (0) given in (8. as a compass.234.224.232.226) into S(0) = 1 i ( 0|ˆ1 ·r |m m|ˆ2 · p |0 − 0|ˆ2 · p |m m|ˆ1 ·r |0 ) u u ˆ u ˆ u m .8. pk ] = i δjk yields finally ˆ S(0) = 1 S(0) = i 3 3 (8. u u u u (8. u1 ·r ] |0 = 0 u ˆ (8. i.230) According to (8.234) where we used for the second identity the fact that u1 ·r and u2 ·r commute. this term cancels the u1 · u2 term in (8.108) yields S (0) = me 2 0| [ˆ2 ·r. Using again (8.232) Obviously.229) (8.231) (ˆ1 )j (ˆ2 )k 0|[xj . u ˆ ˆ ˆ ˆˆ i The commutator property [xj .k=1 (ˆ1 )j (ˆ2 )k δjk = u1 · u2 u u ˆ ˆ (8.182) for p and the expression (8. 8. ˆ ˆ We want to prove now that expression (8. the blue color of the sky and the polarization pattern in the sky which serves many animals.226).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 ) . For this purpose we apply (8. ˆ We want to simplify first (8.228) provides then the first non-vanishing contribution to the scattering cross section (8. m|ˆ1 · p |0 u ˆ 1 = − m|ˆ1 ·r |0 u me ( m − o ) i (8. Expression (8.236) is of great practical importance.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.222).. (8.235) We can now combine eqs. pk ]|0 = u u ˆ j.235) and obtain the leading contribution to the expression (8.108) for the identity operator.8: Two-Photon Processes 235 ˆ These three expressions can be simplified using the expression (8. 8. . 8.k=1 j. for example.

We want to assume. u2 = u2 ˆ ˆ ˆ ˆ (8. To evaluate |ˆ1 · u2 |2 we assume that k1 is oriented along the x3 -axis and.220) 2 dσ = ro (N2 + 1) dΩ2 |ˆ1 · u2 |2 .220).244) ..242) |k2 × k1 | 0 where the second identity follows readily from k1 = e3 and from (8. (8. u1 =  1  ˆ ˆ (8. assume now that the scattered radiation has very short wave length such that ω >> | o − m| . The first choice is ˆ   sinφ2 k2 × k1 (1) u2 = ˆ =  −cosφ2  (8. |n = |0 and ω1 = ω2 = ω in (8. though.238) (8.241) 0 0 Similarly. u2 = u2 ˆ ˆ ˆ ˆ (1) (2) (2) (1) (1) (2) (1) (2) (2) (1) needs to be (8..237) The resulting scattering is called Thomson scattering.e. hence. for all states |m of the electronic system . 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 . however. u2 = u2 ˆ ˆ ˆ ˆ for u1 = u1 .243) cos2 θ2 cos2 φ2 for u1 = u1 .239) We will show below that this expression decribes the non-relativistic limit of Compton scattering. the emitted u ˆ radiation is decribed by the wave vector   sinθ2 cosφ2 k2 = k1  sinθ2 sinφ2  (8. (8. that the dipole approximation is still valid which restricts the applicability of the following derivation to k1 >> One obtains immediately from (8.240). 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 .236 Thomson Scattering Interaction of Radiation with Matter We consider again elastic scattering. i.and the x2 -axes     1 0 (1) (2) u1 =  0  . u2 = u2 ˆ ˆ ˆ ˆ cos2 φ2 cos2 θ2 sin2 φ2 for u1 = u1 .240) cosθ2 We choose for the polarization of the incoming radiation the directions along the x1 .

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

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

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

One calls permutations of the first type even permutations and permutations of the second type odd permutations. . This property will be established now. namely implicitly the group property.240 Many–Particle Systems repeatedly like degenerate states. . but in a different order. We can then state P2 = T (4. i. the set of all permutations of N objects. . . . 5) in case of P2 . . i. . the top row representing the numbers 1 to N labelling the N particles under consideration. (9. For the permutations P1 and P2 given above holds P1 = T (3. 5) T (5..2). 3) T (7. one can state P Ψ(x1 .e. 6) T (6. 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 = . . (9. 4) T (4. xP (N ) ) . The latter factorization has the essential property that the number of factors is either even or odd. . Denoting by P (j) the image of j. P2 = . and explicitly the fact that any P ∈ SN can be given as a product of transpositions T (j. a given P can either only be presented by even numbers of transpositions or by odd numbers of transpositions. and the second row showing the numbers 1 to N again. . not any state can be chosen as to represent the many– particle system. 5). . i.. in particular. The permutations we need to consider for a system of N particles are the elements of the group SN . P2 = T (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.e.e. xN ) = Ψ(xP (1) . However. x2 . k) = (4. . k) of the two particles transposed.5) . k) are characterized by the two indices (j. . . xP (j) . particles 4 and 5. the representations in the space of N –particle wave functions are extremely useful in dealing with N –particle systems. rather only states which obey a certain transformation property under permutations are acceptable. is called a transposition. (j. .. . xj . . xP (2) .3) the j-th index in the second row. i.3) P2 which affects only two particles. even though these properties. 8) . . in notation (9. 5) (9. For the following we will require only two properties of SN . leaving all other particles unchanged. 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 (9.4) both permutations being obviously odd. . We will not discuss here at any depth the properties of the permutation groups SN . The elements P ∈ SN can be represented conveniently as 2 × N matrices. number k under the entry j of the first row indicating that particle j is switched with particle k. The group SN is then composed of two disjunct classes of even and of odd permutations. . A reader interested in the quantum mechanical description of N particle systems is strongly encouraged to study these representations. Transpositions denoted by T (j. k).e..

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

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

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

. nS ) = nj ΨB (n1 . are the so-called operators of 2nd quantization. These operators allow one to construct and manipulate many–particle wave functions while preserving permutation symmetry. . . . since both the N -particle wave function B (n . . nS ) . nj . a† ] = 0 j k (9. .2 Operators of 2nd Quantization An important tool to describe mathematically systems of many bosons or fermions. .15). n . . . . . n − 1. . 9. n2 . nS ). n ) are normalΨ 1 j 1 j S S ized according to (9. . . To prove that (9. . .15) and (9. .22) follows from (9. nj . . . (9. nS ) nj ≥ 1 0 nj = 0 √ . . (b) Show that the states (9. . k The operators a† and aj are referred to as the creation and annihilation operators of bosons for the j orthonormal single–particle states |j . . .1. .2: (a) Show that the states (9. .22) The operators aj and a† obey the commutation properties j [aj . . nS ) = j nj + 1 ΨB (n1 . . .21) we consider the matrix Aj corresponding to aj in the basis of many–particle states (9. . guaranteeing the correct permutation properties of the many–particle states. . . a† ] = δj k . For the adjoint operator a† holds j a† ΨB (n1 . . Exercise 9.17) obey the fermion symmetry (9.16) one obtains (Aj )N = ΨB (N )| aj |ΨB (N ) = √ S N nj δnj nj −1 k=1 k=j δnk nk . .244 Many–Particle Systems The representation of wave functions (9. (9.15). . (9. Employing the superindices N and N introduced above and using the orthonormality property (9.17) is commonly referred to as the occupation number representation since the wave functions are uniquely specified by the set of occupation numbers N = (n1 .25) .20). . .23) (9. . . . .17) obey the orthonormality property (9. . .21) √ The factor nj appears here on the rhs. n ) and the N − 1-particle wave function ΨB (n . . . . . . ak ] = 0 .24) [aj . . . [a† . nj − 1. Creation and Annihilation Operators for Bosons We consider the operator aj defined through aj ΨB (n1 . nj + 1. .9). . .

n2 . . nS ) = 0 . . . . . n2 . j (aj a† − a† aj )|Ψ(N ) = k k ( nj (nk + 1) − (nk + 1)nj )|Ψ(n1 . obviously.21) and vice versa.24). i.29) |ΨB (n1 . The related operator S ˆ N = j=1 ˆ Nj (9. nj − 1 . nk + 1 . the eigenvalues of Nj are the occupation numbers nj .21. S ˆ N |ΨB (n1 . . (9. (9. . . . . nS = 0) as follows S (9. . We will prove the commutation properties (9. (9. .27) Equations (9.24).34) = . n2 = 0. One can readily show using (9. .e.. .9. . . . for basis states |ΨB (n1 . .28) (9. .23) follow in an analogous way. . . nS ) N |ΨB (n1 . . .27. . nS ) . The creation operators a† allow one to construct boson states ΨB (N ) from the vacuum state j |0 = |ΨB (n1 = 0. .32) ˆ ˆ i.e.22) ˆ Nj |ΨB (N ) = nj |ΨB (N ) . One obtains for its matrix elements j A† j N N 245 = ΨB (N )| a† |ΨB (N ) = (Aj )N N j S k=1 k=j nj δnj nj −1 δnk nk = nj + 1 δnj nj +1 S k=1 k=j δnk nk .26) From this result one can immediately conclude that (9.31) This operator is diagonal in the occupation number representation. the properties (9. nS ) = |ΨB (N ) . (9.33) is called the particle number operator since. . (9. 9. . j j Since this holds for any |Ψ(N ) it follows [aj .29) yield the commutation relationship (9. It holds for j = k (aj a† − a† aj )|Ψ(N ) = (nj + 1 − nj )|Ψ(N ) = |Ψ(N ) . . j (9.22) follows from (9. . nS ) = j=1 a† j nj nj ! |0 . n2 .. One refers to Nj as the occupation number operator. .2: Operators of 2nd Quantization Let A† be the matrix adjoint to Aj . n2 . a† ] = 1 For j = k follows similarly 1. n2 . 9. .30) Of particular interest is the operator ˆ Nj = a† aj . nS ) = j=1 nj |ΨB (n1 .

..2. . Obviously. nS ) = − x1 . . . in case that the single particle states |j and |j + 1 are both occupied. xN |ΨF (n 1 . (9. . . .19) without affecting the fermion symmetry (9. operators which can alter the occupancy of the wave function (9. nj+1 . in general... nS ) = .38) .2. since they are induced through the algebra of a† and aj . . 9. i.21. . .24) and the property aj |0 = 0. . . one does not need to worry ever about proper symmetries of wave functions. one needs to apply only the commutation propertes (9. x2 . . it also holds for nj + 1. .23. in case it holds for nj . .e. . nj (9.2: Show that the boson operators a† and aj satisfy j [aj . the fermion wave function changes sign when one exchanges the order of the occupancy. i. 9. . . xN |ΨF (n1 . 9.e. √1 N! .36) Creation and Annihilation Operators for Fermions We want to consider now creation and annihilation operators for fermions. . φj (xN ) φj+1 (xN ) . x2 ..30). . 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 . . φj+1 (x1 ) φj+1 (x2 ) .29. . nj . .38) corresponds to x1 .. 9. .s. 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. nS ) . nj . . .. . . . . To prove this property we notice that the l. (n − 1)! (9. . . .17.23. .22.37) (9.9). xN |ΨF (n1 . x1 . . j Exercise 9.1: The commutation relationships (9.246 Many–Particle Systems In using the boson creation and annihilation operators.h. .g... nj+1 = 1.. Exercise 9.. φj (x1 ) φj (x2 ) . from the vacuum state |0 or states |ΨB (N ) . showing that the property holds for nj = 0 and. of (9. e.23) hold for the state defined through (9.24) and aj |0 = 0 imply that the the properties (9.. f (a† )] j [a† . nj+1 . As long as one starts from a wave function with the proper boson symmetry.39) . . .35) and where the derivative operation is defined through ∂f (A) = ∂A ∞ n=1 1 f (n) (0) An−1 . x2 . . . 9. nj+1 . . Prove this by induction. . .

. . . however.. nj+1 . n3 . .42) i. . ..19). nj .40) . We will be adopting below the notational convention that qj corresponds to occupancies N = (n1 . φj+1 (xN ) φj (xN ) . in order to be consistent with this convention. . Obviously. . . φj+1 (x1 ) φj+1 (x2 ) . i. . . φj (x1 ) φj (x2 ) .. n2 . 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). x2 |ΨF (n1 = 0. of (9. .e. . the rhs. vanishes. qj is the number of states |k with k < j which are occupied in a fermion wave function. .. . n6 = n7 = .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.. i. . .9. nj → 0. that occupancies are always ordered according to a monotonous increase of the single–particle state index.43) . . We are now ready to define the annihilation operator cj for fermions in the single–particle state |j as follows cj |ΨF (n1 . the expressions (9. . (9. . . .. n2 .. x2 . nS ).e. nS ) = nj (−1)qj |ΨF ( n1 . . A proper example is the two particle fermion wave function x1 . . nj . One adopts the convention that particles are created and annihilated by these operators always at the first position of the wave function. . 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. .e. . . This change of position is connected with a possible sign change (−1)qj where qj is defined for a given N = (n1 .e. . 2. . xN |ΨF (n1 . n2 = 1. (9.2: Operators of 2nd Quantization The r.40) are identical. For this purpose one must adhere to a strict convention: the labelling of single–particle states by indices j = 1. . .19). . at the first column of the Slater determinant (9. .. . n3 = 0.. i.h. n4 = 0.. . . φ2 (x2 ) φ5 (x2 ) (9.. nj−1 . . nS ) = . nS ) qj states occupied In case that the single particle state |j is not occupied. particles are being created or annihilated. This requires one. (9. at which column of the Slater determinant (9. Because of the property of the determinant to change sign when two columns are interchanged. it is necessary to specify for a fermion wave function the order in which the single– particle states |λj are occupied. 247 − √1 ! N .39) and (9. . . . nS ) as follows: j−1 qj = k=1 nk . .. n5 = 1.38) reads − x1 . n2 . n2 . . nS = 0) φ2 (x1 ) φ5 (x1 ) 1 1 = √2 = √2 [φ2 (x1 )φ5 (x2 ) − φ5 (x1 )φ2 (x2 )] . nj = 0.s. . . . nj+1 .

otherwise. B]+ = AB + BA.248 The operator c† adjoint to cj exhibits the following property j c† |ΨF (n1 . The anti-commutation properties (9. (9.45) follow in an analogous way and will not be derived here.19).44). . ck ]+ = 0 . .17. . . . . nS ) = j j (−1)qj (1 − nj + 1)(−1)qj nj |ΨF (N ) = nj ΨF (N ) (9.49) The derivation involves realization of the fact that a non-zero result is obtained only in case nj = 1. 9. n2 . n2 . . Similarly one obtains cj c† |ΨF (N ) = cj (−1)qj (1 − nj ) |ΨF (n1 . (9. . . nS ) = j (−1)qj (nj + 1))(−1)qj (1 − nj ) |ΨF (N ) = (1 − nj ) |ΨF (N ) (9.44) follows from the definition (9. . [c† . . It holds c† cj |ΨF (N ) = c† (−1)qj nj |ΨF (n1 . nj . . . . .47) Let C† be the matrix adjoint to Cj . nS ) qj states occupied The operators thus defined obey the commutation properties [cj .48) vanishes.46) first for the case j = k. Let Cj be the matrix corresponding to the operator cj in the basis of fermion states (9. We first show that (9. n3 . c† ]+ = δj k k where we have used the definition of the so-called anti-commutators [A. nS ) = j Many–Particle Systems (−1)qj (1 − nj ) |ΨF ( n1 .48) From this follows (9. nj → 0 . We consider (9. nj−1 . the last factor on the rhs. .43).45) (9. of (9. The elements of Cj are then ( note that nj only assumes values 0 or 1) S (Cj )N N = Ψ (N )| cj |Ψ (N ) = nj (−1) δnj nj −1 k=1 k=j F F qj δnk nk .46). We have used here the definition qj = k<j nk along with qj = qj .46) [cj . nj+1 . nj → 1. 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 . .50) . nj → 1 . c† ]+ = 0 j k (9.44) (9. We will prove now the anti-commutation property (9.

56) Here we have made use of cj cj = 0 and c† c† = 0 which follows from nj ∈ {0. with eigenvalues equal to the occupation numbers nj ˆ Nj |ΨF (N ) = nj |ΨF (N ) ˆ Nj . . 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. nk = 1 . . nλ1 . k k 249 (9. . .54) ˆ N = j=1 ˆ Nj (9. j j For j < k one obtains cj c† |ΨF (N ) = (−1)qj +qk nj (1 − nk )|ΨF (n1 . . i. .s.58) . 1 In the following we will apply fermion operators c† and cj only to electrons which carry spin 2 .9.) = c† 1 c† 2 · · · c† N |0 . nλ2 .46).52) One can obtain similarly the same relationship in the case j > k. of this equation we meant to indicate only those N occupation numbers nλj which are non-vanishing. . i.49) and (9. .49) shows that the operator ˆ Nj = c† cj j (9. nj = 0 . nj = 0 . . nλN . . is referred to as the occupation number operator.51) (9.50) yield ( cj c† + c† cj ) |ΨF (N ) = |ΨF (N ) . x|j → φj (r) | 1 σ .e. . On the r. in the basis |ΨF (N ) . We j will often separate the states |j .. . . 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 . j j j j j j j (9. nS ) k c† cj |ΨF (N ) = (−1)qj +qk −1 nj (1 − nk )|ΨF (n1 . j → j. 9. coordinates x and index j into a space part and a spin part. Equation (9. the creation operators must operate in the proper order. Since (9.52) hold for any |ΨF (N ) one can conclude (9. i.57) On the l. nS ) k and. ( cj c† + c† cj ) |ΨF (N ) = 0 . for which purpose one needs to envoke only the algebraic (anti-commutation) properties (9. hence. an operator cj must be on the left of an operator ck for j < k.e.53) is diagonal in the occupation number representation. . .46). . . nk = 1 .29)) |ΨF (. .h. 2 (9. .s. σ 1 (σ = ± 2 ) .. 1}.55) is called the particle number operator. S (9.h. λ λ λ (9. . hence.2: Operators of 2nd Quantization (9. ˆ It is an interesting exercise to demonstrate that Nj only has eigenvalues 0 or 1.51. Correspondingly..e.

(9.59) and over pairs of particles in (9. ˆ ΨB.62) These matrix elements can be obtained by chosing the operators and many-particle states in the position representation.15. (9. i.6) above.59. 2.F (N )|F |ΨB. 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 ) .59) and (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 ) (9.63) ˆ ˆ This expression has been specified for the purpose of these notes to distinguish F and f and is not common ˆ terminology. xk .61) is not a one particle operator as long as the N operators rj (x).F (N )|G|ΨB. 9. ˆ r|f |s = 1 ∗ ˆ d3 r ψr (r. j = 1. 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.F (N ) . xk ) ˆ j. namely. The action of the operators boson and many–fermion states |Ψ is obviously described through the many–particle state matrix elements ˆ ΨB. one expects that the matrix elements can actually be expressed in terms of matrix elements ˆ involving solely the generic operators and the single–particle states.17) and integrating over all particle space-spin coordinates x1 . respectively.17). 9. x ) which constitute G are ˆ called the associated generic two–particle operators.60) is that the sum over particles in (9. An operator N ˆ R = j=1 rj (xj ) ˆ (9. r = 1. We adopt a similar expression to distinguish two–particle operators G and g . .e. . .59.15. The essential aspect of definition (9. 2. (9. σr ) f (x) ψs (r. The operators f (x) which constitute F are called the associated generic single–particle operators1 . . These operators had been introduced already in Eq. N are not all identical.k=1 j=k (9. Since the many–particle states are built-up from a basis |r . . .60) involves generic operators – acting on xj and on xj .F (N ) .250 Many–Particle Systems 9. 9. ˆ . The operators g (x. 9.60) ˆ correspondingly are called two–particle operators.. . respectively – which are all identical. Operators of the type 1 ˆ G = 2 N g (xj . xN . σs ) (9. .59) ˆ ˆ are called single–particle operators. . S of single–particle spinorbital states (see page 241) and the operators are specified through the associated generic operators ˆ f and g . . x2 .

64). (9. are spin-independent as is evident from the factor δσσ ..64) We like to note an important symmetry of the matrix element r.62) in terms of the matrix elements of single–particle states (9.67) ˆ In this case the generic single–particle operator.66) Here we adopt the notation for the Laplacian ∂2 ∂2 ∂2 ∂2 = + + . i. s|ˆ|t.65) This symmetry will be exploited repeatedly below. one cannot switch the indices of one pair individually.64) r.68) Another single–particle operator is the one–particle density operator N ρ(r ) = ˆ˜ j=1 δ(r − rj ) . 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 .e. r|ˆ|u.71) . ∂rj2 (9. ˆ 251 (9. σs ) g (x. 9. is t(x) = (− 2 /2m)(∂ 2 /∂r 2 ). s and t. σt )ψu (r . The aim of the present section is to express the matrix elements of single– and two–particle operators in a basis of many–particle states (9.70) Both operators. u = s. according to (9. u.9. An example for a two–particle operator is the Coulomb repulsion operator 1 ˆ V = 2 N j. s|ˆ|t. 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) . u which follows directly g from the definition (9.69) ˆ In this case the generic operator is f (x) = δ(r − r). The matrix elements of the generic operator ˜ in the one–particle basis are ˆ rσ|f |sσ = δσσ d3 r φ∗ (r) δ(r − r) φs (r) = δσσ φ∗ (r) φs (r) δσσ ˜ ˜ r r ˜ (9. t . σr )ψs (r . that the symmetry implies that one can switch simultaneously the pairs of indices r.66) and (9. ∂r 2 ∂x2 ∂x2 ∂x2 1 2 3 (9.63. ˜ (9. s|ˆ|t. x ) ψt (r. σu ) . ∂r 2 (9.3: One– and Two–Particle Operators r.k=1 j=k q2 |rj − rk | (9. Notice. u = g d3 r ∗ ∗ d3 r ψr (r. We will show that the boson and fermion creation and annihilation operators serve this purpose.59).69). g g (9.

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

60) as follows N ˆ f (xj ) → j=1 N 1 2 1 2 S r.78) 1 2 g (xj .t.s.78) and (9.59) and G in (9.62).s=1 ˆ r|f |s a† as bosons r ˆ r|f |s c† cs fermions r r.45.24) and (9. s|ˆ|t. The Matrix Elements ΨF (N )|c† cs |ΨF (N ) r We consider first the case r = s.63.79) It is crucial to notice in the expression for the fermion two–particle operator that the order of the annihilation operators in (9. To show that the resulting values of the matrix elements (9.3: One– and Two–Particle Operators Definition in Terms of Creation and Annihilation Operators 253 In order to express many–particle state matrix elements (9.t.63) and (9. s|ˆ|t.171 and pp. . and then in g a second step the matrix elements ΨB (N )|a† as |ΨB (N ) . (9.62) through single–particle state matrix ˆ ˆ elements (9. The matrix elements are then actually those of the number operator c† cr which is diagonal in the occupation number representation. s|ˆ|t.F (N ) have the same values as the respective matrix elements (9. These rules will be provided below only for the case of fermions.82) ΨF (N )|c† cs |ΨF (N ) = r cr 2 (9.s. u as defined in (9. xk ) → ˆ j. u .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 S r.83) t=1 t=r.79) have the following implication: The operators (9. respectively. r ΨB (N )|a† a† at au |ΨB (N ) r s ΨF (N )|c† c† ct cu |ΨF (N ) r s bosons fermions (9.78.79) is opposite to that in the matrix element r.e.46).9.64) one replaces the operators F in (9.62) are correct one needs to compare the result derived with conventional derivations of the matrix elements2 . In order to determine these matrix elements one needs to evaluate first the matrix elements ˆ of the generic operators r|f |s r.23. 9. respectively.u=1 (9. g and not ct cu .81) For the latter matrix elements simple rules can be derived from the algebraic properties of the boson and fermion operators (9. i.79) in the basis of many–particle states |ΨB.s=1 S r. namely cu ct . r ΨF (N )|c† cs |ΨF (N ) . s|ˆ|t.. r S Ψ (N In case of r = s one obtains F )|c† cr |ΨF (N ) r = nr r=1 δnr nr . 9. u a† a† at au bosons g r s † † r. 9. Expressions (9.64). 9. pp.176 where the matrix elements of fermion states can be found.80) (9.k=1 j=k S r. u cr cs cu ct g fermions (9.

84) we can conclude that for diagonal elements holds S Many–Particle Systems ˆ ΨF (N )|F |ΨF (N ) = r=1 ˆ nr r|f |r (9. ˆ ˆ Nr Ns .88) .86) For the combination of indices r. The anti-commutation property (9. 1 2 S r s|ˆ|s r c† c† cr cs .s.d. those for which N and N differ in more than two occupation numbers. 171) shows that the operator defined in (9. t).d. s). G = G0 + G1 + G2 . s. ˆ ˆ ˆ ˆ Accordingly. t. s.45) yields together with the definition of the occupation number operator (9.86) into three contributions.78) does yield the same results as the operator defined in (9. Comparision with the results in Condon&Shortley (pp.79) and using c† r we can write S 1 ˆ G = rs|ˆ|tu c† c† cu ct g r s 2 r. u in this sum three possibilities exist Case 0 two of the four indices are different. ˆ The first contribution G0 is 1 ˆ G0 = 2 S r s|ˆ|r s c† c† cs cr + g r s r.s=1 a.e. Case 2 all four indices are different. in which case ns = ns + 1 and nt = nt − 1 holds. s.d. (r.’ (all indices different) that only tuples (r. we split the sum in (9. The Matrix Elements ΨF (N )|c† c† ct cu |ΨF (N ) r s Before we determine these matrix elements we will investigate which possible combination of indices r. 9. u need to be considered.254 From (9.87) Here and in the following we denote by ‘a.s=1 a. t. Case 1 three of the four indices are different.u r=s.s=1 a. Starting from the definition (9. i.i.85) All other matrix elements vanish. (r. are ˆ ˆ ΨF (N )|F |ΨF (N ) = (1 − ns )(−1)qs nt (−1)qt s|f |t (9. g r s r. namely. s. t. c2 = 0 r (9.82.d. where each contribution corresponds to one of the three cases mentioned. t.i.53) 1 ˆ G0 = 2 S r s|ˆ|r s − r s|ˆ|s r g g r. u) are included in the summation for which all indices are different.e. for which holds r = s.59) when the matrix elements are determined between Slater determinant wave functions. (9.. t=u 2 = 0. i.84) The off-diagonal elements with nr = nr except for r = s.i.. r = t etc.t. (9.i.

t=1 a. ns .s. this part accounts for all diagonal contributions to G.t.92) ˆ This contribution to G has non-vanishing matrix elements only when N differs from N in four ˆ occupation numbers nr . A similar contribution does not arise for F .84) ˆ of the one–particle operator F . s|ˆ|s. S r.91) ˆ G2 = r.s. r s|ˆ|r t − r s|ˆ|t r g g ˆ c† ct Nr . r s|ˆ|r t c† c† ct cr g r s r s|ˆ|t s c† c† cs ct g r s r s|ˆ|s t c† c† ct cs g r s r s|ˆ|t r c† c† cr ct .65).t>u S r.93) .t.s.i.s.45). r g g .d.i. One obtains 1 ˆ ΨF (N )|G|ΨF (N ) = 2 S nr ns r.u r<s.u r>s. r s (9.s r. (9. ˆ The second contribution G1 is ˆ G1 = + + + 1 2 1 2 1 2 1 2 S r. S r.t<u S r. s − r. s (9.t.t.s.s.s.9. contributes in this case.i. this contribution obviously behaves similarly to the off-diagonal part (9.t=1 a.i. i. 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.3: One– and Two–Particle Operators 255 Obviously.t>u r s|ˆ|t u − r s|ˆ|u t g g c† c† cu ct . given in (9.t<u S r s|ˆ|t u c† c† cu ct g r s r s|ˆ|t u c† c† cu ct g r s r s|ˆ|t u c† c† cu ct g r s r s|ˆ|t u c† c† cu ct g r s (9.90) ˆ Since Nr is diagonal.s. F (N )|G|ΨF (N ) .s. Only G0 .u r>s.. nu .t>u S r.d.i. this contribution to the two–particle operator is diagonal in the occupation number ˆ representation. N = N or nr = nr for all r.u r>s.d.t=1 a. Commutation of ct cu according to (9. g r s (9.89) S r.t=1 a.86).t. (9. employing the symmetry property (9. s|ˆ|r.88). As we will find. 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. ˆ Case 0 This case covers diagonal matrix elements. nt .u r<s.d.e.d.t=1 a. 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.s.

s|ˆ|t.3. The matrix elements of the corresponding generic operators are (9. the ‘−’-sign applies for s < r.3. u. for ˆ non-vanishing matrix elements ΨF (N )|G|ΨF (N ) at most four occupation numbers nr and nr can differ.3: 9.3. u g g (9. contributes in this case.1: 9. We like to express these operators through ˆ fermion creation and annihilation operators. In particular. ˆ ˆ Show that the matrix elements of F and of G conserve particle number.3. One obtains for the three components of S ˆ S3 ˆ S+ ˆ S− = 1 2 S ˆ ˆ Nrα − Nrβ = = r=1 S c† crβ rα r=1 S c† crα rβ r=1 (9. particle number is conserved. One obtains ˆ ΨF (N )|G|ΨF (N ) = ± (−1)qs +qr +qu +qv (1 − ns )(1 − nr )nu nv r.73) and S 2 given in (9.75) which are. u < v. Derive (9.96) .94) Case 2 This case covers off-diagonal elements with nr = nr except for r = s.91). nt = nt + 1.3.89).95) In the latter formula the ‘+’-sign applies for s < r. Derive the non-vanishing matrix elements (9.5: Derive (9. t − r.92) from (9. v − r. u < v or s > r. v for which holds ˆ ns = ns + 1. nv = nv − 1. Derive (9.95) from (9. Only G1 . one–particle and two–particle operators.74) and (9. respectively. s|ˆ|r.92). r=1 r=1 Exercise Exercise Exercise Exercise Exercise 9. contributes in this case. One obtains ˆ ΨF (N )|G|ΨF (N ) = (−1)qs +qt (1 − ns )nt r r=s. t for which holds ˆ ns = ns + 1 and nt = nt − 1. r g g (9. s|ˆ|u. given in (9.62) for bosons.76).90) from (9. given in (9. Furthermore.2: 9.e.4: 9.92). s|ˆ|v. u > v.. nu = nu − 1. u > v or s > r. All matrix elements which are not covered by the three cases above vanish. it holds S nr = S nr .t nr r. Only G2 .90).256 Many–Particle Systems Case 1 This case covers off-diagonal elements with nr = nr except for r = s. i. Spin Operators ˆ ˆ In this section we will briefly consider the spin operators S given in (9. t.

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

r=1 3 Urm U ∗sm = δrs m=1 (9. {|m).e. 1989) . all material systems would condensate into states which would depend little on particle number. If it were’nt for the fermion nature. ultimately depend in a crucial way on the fermion character of its constituent building blocks.101) The unitary property of Urn U† U = U U† = 1 1 or S S (9. We will restrict our description in the following to systems composed of an even number (2N ) of particles and to spin-independent interactions. S}.e. Transformation to a New Set of Creation and Annihilation Operators Our aim is to represent the Hamiltonian (9. m = 1. 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).103) An account of some of these efforts can be found in Modelling Brain Function – The World of Attractor Neural Networks by D. electrons and nucleons. What might happen? Independent-Particle Hamiltonian A many–fermion system governed by an independent-particle Hamiltonian. namely of neurons3 . The electronic properties of atoms with different numbers of electrons would differ little.s=1 σ ˆ rσ r|t|s c† csσ (9.4. i. ranging from nuclei to the molecules of living systems .1: Imagine that in a closed. i. (9.258 Many–Particle Systems that today these concepts are providing insight into the functioning of biological brains.. water-filled jar all electrons of water turn their fermion nature into a boson nature. . Chemistry essentially would know only one element and Life would not exist. can be described in a rather straightforward way. where S |m) = r=1 Urm |r .J.100) ˆ ˆ We will denote the matrix elements of t as r|t|s = trs .102) U ∗rm Urn = δmn . New York. . In such a representation Ho is diagonal and the stationary states can be stated readily. 2. would not exist.100) in a form in which the factors c† csσ become rσ occupation number operators. another prototypical systems of many strongly interacting components. a Hamiltonian without a two–particle operator contribution accounting for interactions among the particles. according to which nuclei and atoms change their qualitative properties when they grow larger. The Aufbau principle. The Hamiltonian of such system is S Ho = r. Before continuing one may finally point out that the material world as we know it and as it establishes the varieties of natural substances.. . Exercise 9. Amit (Cambridge University Press.

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

. is N |Φo = j=1 d† α d† β |0 .e. . σk ) for j = k (9. ) = j=1 nj . . Vn = (U1n .e. (9. hence. transformation matrices Urn and energy values n . Eigenstates of (9.116) which follows from (9. 2. σ2 .e. n2 . .119) In this state the N lowest orbital eigenstates of trs are occupied each by an electron with spin |α = | 1 . We apply for this purpose (9. . the associated ( properly normalized) eigenvectors Vn are the columns of (Urn ).115) with eigenvalues 2N E(n1 . It is.106) diagonal. This equation poses the eigenvalue problem for the hermitian matrix (trs ). 1 and |β = | 1 .106) and the unitary property of Urn . . S are T real.57) to the case of 2N particles. The 2N fermion state 2N d† j σj |0 n j=1 where (nj . U2n . − 1 . . n = 1. A non-degenerate orbital state which is occupied by two fermions. The corresponding eigenvalues n . . i.. (9.e... . .260 Diagonal Representation Many–Particle Systems The transformation (9.118) Ground State In case of an ordering m < n for m < n the state of lowest energy.. σ1 . S (9. is obtained. .106) is equivalent to S trs Usn = s=1 n Urn ∀r. USn ).115) can be stated immediately since any many–particle wave function in the occupation number representation is suitable. σ2N . The condition (9.101) gives us the freedom to choose the matrix (9. i. σj ) = (nk . r = 1. a simple matter to nσ state many–particle states in the new representation. n2N . i.117) are eigenstates of (9.115) ˜ and involves solely occupation number operators N nσ = d† dnσ .114) together with (9. 1 as well as a fermion in a spin state | 1 . j j (9. 2. 2 2 2 2 i. − 1 is called a closed 2 2 2 2 . ˜ tmn = In this case Ho has the desired form S n δmn . a fermion in a spin state | 1 . Before proceeding we like to address the issue how the new representation. . the so-called ground state.114) Ho = n=1 † n dnσ dnσ (9.

the states differ in their spin character. However. |α.. 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.. Alltogether. One can conclude immediately that the ˜ ˜ ˜ ˜ (N mα − N mβ )(N nα − N nβ ) does not give any contribution second term in (9. Similarly. 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).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. α .n=N (9.n=1 m. i. ˆ S 2 ||β.e. 1 S 4 m. The ground state has only closed shells.n=1 cmα cnβ cmβ cnα does not give a contribution if m = n and either |m) or |n) are closed shells. We will demonstrate now that this property endows the ground state with total spin zero. (9. Obviously. β . ˆ one can conclude that |Φo is an eigenstate of S 2 with eigenvalue zero. β = 2 |α. both with eigenvalues 2 ˆ S 2 |α. the third term − S m. β ˆ are eigenstates of S 2 given in (9.97).4: Independent-Particle Models 261 shell. α |α. α |β.121) All four states have the same excitation energy.120) is particularly simple for closed shells. ˆ In order to investigate the total spin of |Φo we apply to this state the total spin operator S 2 as given by (9. α N +1 − N . namely. The two states |α. There are four such states.n=1 m.100).n=1 d† d† dmβ dnα mα nβ (9. The remaining contributions to S 2 which act on partially occupied orbitals |N and |N + 1 are 1 ˆ Σ2 = 4 N +1 ˆ ˆ ˆ ˆ (Nmα − Nmβ )(Nnα − Nnβ ) m. i.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.n=1 † † if either |m) or |n) are closed shells.120) where the occupation number operators refer to the single-particle states |m).9.120). α = 2 |β. of ∆E = and |β.124) . The spin operator in the present representation of single-particle states is 1 1ˆ ˆ S2 = N + 2 4 S S ˜ ˜ ˜ ˜ (N mα − N mβ )(N nα − N nβ ) − m. β |β. namely.97).n=1 d† d† dmβ dnα mα nβ (9. energy above the ground state. The action of the operator (9. is a singlet state.e. Excited States We want to construct now excited states for the system governed by the independent-particle Hamiltonian (9. In case of m = n it gives a contribution ‘1’ for each closed shell which cancels the contribution of the first term.

apropriate eigenstates of the operators S 2 and Example: 2N Independent Electrons on a Ring In order to illustrate the procedure to obtain eigenstates of one-particle Hamiltonians (9. . One can. . for |α. do not need to be considered since they ˆ give vanishing contributions. ˆ An eigenvalue ‘2’ of the spin operator S 2 identifies the states |α. of course.n=N d† d† dmβ dnα d† +1α dN β |Φo = 0 mα nβ N (9. . .129) The symmetry of the Hamiltonian suggests to choose a new representation defined through the operators (r = 1. conclude that (9.127) ˆ are triplet and singlet states.262 and ˆ Σ3 = − m. α as triplet states. . also eigenstates of Σ2 . The system is assumed to posses a 2N –fold symmetry axis and interactions connect only orbital states |r → |r ± 1 .σ dN σ N σ. mα nβ (9. of course. for example. 2N which are located on a ring. β both with eigenvalues 1. ˆ Exercise 9. . β are not eigenstates of S 2 .σ = (9. however. β and |β. The system is described by the Hamiltonian 2N ˆ Ho = − t n=1 σ c† σ cn σ + c† σ cn+1 σ n n+1 . the linear combinations 1/2(|α. The two states |α.130) . i. m |Φo m m m γσ.4. β qualify as eigenstates.122) is correct. 2N ) drσ = √ 1 2N 2N e−irnπ/N cnσ n=1 (9. one with eigenvalue zero (singlet) and one with eigenvalue ‘2’ (triplet). α and |β. 2. The action of Σ N +1 ˆ Σ3 |α. α ± |β.e. . The same result holds for |β.100) outlined above we consider a system of 2N electrons which move in a set of atomic orbitals |r . β =− m. c† +1 σ = c† σ 1 2N (9.125) ˆ Contributions to S 2 acting on states |N + 2 .2: Construct four operators O ˆ O ˆ such that O ˆ S3 . ˆ 3 on the two states produces .n=N Many–Particle Systems N +1 d† d† dmβ dnα . |N + 3 . .σ d† +1. hence. (9. α . 2. vanish. ˆ The remaining states |α. .. We identify the states |2N + 1 = |1 to avoid cumbersome summation limits.126) which follows from the occurence of squares of fermion operators which.128) The cyclic property of the system is expressed through c2N +1 σ = c1 σ . |β. r = 1. α are. etc. β .

132) an identity which can be derived employing the well-known expression for a finite geometric series m as = a s=1 1 − am . of (9. 9.4: Independent-Particle Models d† = √ rσ 1 2N 2N 263 eirnπ/N c† nσ n=1 √ 1 eirnπ/N .s. 2N. . One can then express[. vanishes for integer s.134) which. rσ (9. .137) where 1 ˜ trs = e−irπ/N 2N Using (9. .131) constitute a unitary transformation U defined through Urn = unitarity property follows from 2N eirsπ/N = 0 r=1 for 0 < s < 2N . eisπ/N 1 − e2N isπ/N 1 − eisπ/N (9.132) gives for r = 0. n=1 (9.s=1 σ ˜ rσ trs d† dsσ (9.131) The (9.139) −2t cos r=1 σ rπ N d† drσ .9.f.109)] ¸ cnσ = √ 1 2N 2N eirnπ/N drσ r=1 2N (9. one obtains the new representation of Ho 2N ˆ Ho = −t r.132) one can conclude 2N ei(s−r)nπ/N + eisπ/N n=1 1 2N 2N ei(r−s)nπ/N .h.138) ˜ trs = from which follows ˆ Ho = eirπ/N + e−irπ/N 2N δrs (9.130. . thereby. (9. 4N. 1−a (9.140) The Hamiltonian is now diagonal and the construction outlined above can be applied.135) c† nσ 1 = √ 2N e−irnπ/N d† rσ r=1 (9.136) ˆ and.133) Application to the l. (9. indeed. . 2N (9.

The most famous example is the periodic table of the elements. i.264 Many–Particle Systems Exercise 9. Many electronic properties of solids can be understood in a similar way. .e. valence bands or conduction bands. The slow decay of the interaction makes the motion of any one electron rather independent of the exact position of most other electrons.e.e. if one places fermions into an independent-particle ground state of the type (9. 9. We will present in this and the following section a method which constructs such optimal single-fermion states.. n ± 1) orbitals and Hamiltonian 2N −1 ˆ Ho = − t n=1 σ c† σ cn σ + c† σ cn+1 σ n n+1 .. i. does not spoil it.. 2. and that independent-particle behaviour surfaces mainly because it applies well to the ground state.5 Self-Consistent Field Theory Observations of many-fermion systems show that properties of such systems often can be explained to a surprisingly good degree on the basis of the assumption that the fermions move like independent particles. Of course.g. 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..e. . one expects that mean-field descriptions should be rather sufficient.141) State the excitation energy ∆E. in states which are constructed according to the ‘Aufbauprinzip’ in a manner which neglects the mutual repulsion between electrons. . i. n = 1.e. The regularities of the elements can be rationalized in terms of electrons moving in hydrogen atom-type orbitals. the fermion nature restricts the possibility for perturbations on the system. (9.3: Construct the ground state and the lowest energy excited state(s) for a system of 2N electrons moving in a linear chain of single-electron orbitals |n . in particular. Hence. the everpresent Coulomb repulsion between electrons. A similar principle accounts approximately for properties of nuclei as described by the nuclear shell model.. i. properties of many-fermion properties deviant from such simple description is then of as much interest as independent fermion descriptions are useful. There are two reasons why this is so: first. ..4. if one can choose the singleparticle states such that the residual perturbations are small. These states do not altogether neglect the pair interactions.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. those with energy above that of the (energetically) highest occupied single-particle state. A second reason why independent-particle models can be successful is that the pair interaction for electrons is the slowly decaying Coulomb interaction. assuming that electrons move independently from each other in so-called bands. e. i. Since independent-particle behaviour is so universal one may wonder why the presence of particleparticle interactions. 2N with interactions between neighbouring (n. rather they assume that each particle experiences the average interaction due to the fermions frozen into the ground state. The properties of molecules can be understood largely in terms of models which place electrons into so-called molecular orbitals in which the electrons move as if they were not interacting with other electrons.

s. . 2. S) ˜ S S |m = ˜ r=1 Urm |r .u=1 σ. .t. . p. u refer to the initial basis of single-particle states |r . as in in (9. t. . q.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 (9.143) with r. S} the elements of which will be labelled by indices ˜ m. . the corresponding U = ( Urm ) forms an S × N matrix and.s.t. ˜ (9. U = ( Urm ) as defined in (9. 2. j j (9. 2.145). S on which (9. 2.σ and derive a single-particle operator which approximates this Hamiltonian. The states |m are connected with the states |r through (m = 1. If one restricts m = 1. In our formulas below we will adopt strictly the convention that indices r. naturally. r = 1. s|ˆ|t. 2. u v r. . n. Mean-Field Potential The mean field for the reference state |Φo (U) is defined in a straightforward way by averaging the two-particle contribution to (9.143) 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. m = 1.142) is based. . .142) r|t|s c† csσ + rσ sσ 2 r. .s=1 σ r. .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.9.108). |r = m=1 ∗ Urm |m . u c† c† cuσ ctσ v (9.142) over this reference state to turn the contribution into an effective one-particle operator. . We will adopt a second set of single-particle states {|m .144) defined as in (9. . This is done as follows: ˆ Vmf (|Φo (U) ) = 1 2 S r.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.u=1 σ.146) . m = 1. . . s|ˆ|t. is not unitary. . S is a unitary matrix.s. N . . The states |m serve to define a reference state of the system ˜ N |Φo (U) = j=1 d† α d† β |0 .

σ S dmσ dnσ + r. n c† c† v ˜ ˜ rσ sσ r. u ˜ ˜v m. u c† ˜ vn sσ m.150) r. hence.t.m. m.s.84). ˆ Vmf (|Φo (U) ) = 1 2 S Φo (U)|d† d† |Φo (U) = 0 mσ nσ Φo (U)|dmσ dnσ |Φo (U) = 0 Φo (U)|d† dnσ mσ δmn δσσ 0 |Φo (U) m ≤ N m > N (9. n c† ctσ δσ σ ˜ v ˜ rσ m≤N r. n|ˆ|t. for example. s| v t. ˆ (9.σ S d† dnσ cuσ mσ − r.149) The remaining matrix elements appearing in (9. u Usm Uun . Since the averages affect only two creationannihilation operators actually a mixed representation is most suitable ˆ Vmf (|Φo (U) ) 1 = 2 1 2 1 2 1 2 1 2 S m. n c† ˜ v ˜ sσ m.266 Here ··· denotes the average O = Φo (U)| O |Φo (U) .146) accordingly. s|ˆ|t. m|ˆ|˜ . m| v |t.σ .s.n.u ∗ r.n.m.n. u c† d† dnσ cuσ ˜ vn rσ mσ r.n=1 σ.t=1 σ. m|ˆ|t.t=1 σ.u=1 σ.n.σ S ctσ + m.σ S 1 − 2 where.148) one obtains. For the averages ··· in (9. n c† d† dnσ ˜ v ˜ rσ mσ r. by means of the rules (9. s|ˆ|˜ .n=1 σ.u=1 σ. s|ˆ|m.148) r. m|ˆ|t. d† d† mσ nσ dmσ dnσ d† dnσ mσ = = = = and.148) are obtained in a similar way by transforming only two of the four single-particle states into the new representation.m. n = ˜ ˆ ˜ s.t.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.s.u=1 σ.σ S d† d† mσ nσ cuσ ctσ + r. Many–Particle Systems (9.σ S d† dnσ mσ ctσ (9.

s| v |t. n c† ctσ .152) By means of expression (9. u c† cuσ δσσ ˜ vn sσ m≤N s.s σ where r| vmf (|Φo (U) ) |s = ˆ t. u − r. m|ˆ|˜ .154) ˆ The mean field approximation replaces then the Hamiltonian (9. The mean field approximation. u = s. We will address now a proper choice of the reference state. Exploiting the symmetry r.t=1 σ (9. m|ˆ|t. leaves open the question how the reference state should be chosen.153) rσ r.. s|ˆ|t. The one-particle operator (9.5: Self-Consistent Field Theory 1 2 1 2 1 2 S 267 m..u=1 σ.9.σ r.156) which defines the reference state |Φo (U) .S. m=1.. t| v |u.σ S r.N (9.t=1 σ. At this point this state is completely arbitrary. u c† cuσ ˜ vn rσ (9.2.s=1 σ ˆ r|t|s + r| vmf (|Φo (U) ) |s ˆ c† csσ rσ (9.. s )  ˆ ˆ  m≤N ∗ Utm Uum   (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.m.u=1 σ..u ( 2 r.142) by Hmf (U) defined through S ˆ Hmf (U) = r. .149) for the matrix elements in the mixed representation one can state the right hand side explicitly in terms of Urm . t| v |s.σ S + − r. s|ˆ|˜ .152) can then be written in the form ˆ Vmf (|Φo (U) ) = r| vmf (|Φo (U) ) |s c† csσ ˆ (9. r| v |u. n c† ctσ δσ σ ˜ v ˜ rσ − r. m|ˆ|˜ .155) which is a function of the S × N –matrix U = ( Urm )r=1. u c† cuσ ˜ vn rσ m≤N r. as introduced here.2.m.151) Carrying out the sum σ δσσ = 2 leads to a factor 2 for the first two terms. ˜ v ˜ sσ m≤N s..u=1 σ S − m.

. 2. We will argue below that the self-consistent field ground state.155). the independent-particle nature stemming from the fact that the functional form of the ground state.s=1 σ S ˆ rσ r|t|s c† csσ (9. . M = 1.268 Many–Particle Systems 9.159) One defines the reference state |Φo (U(1) ) using the definition (9.145). . (9.4 and defining this as the new reference state. .s=1 σ S ˆ r|t|s + r| vmf |Φo (U(M ) ) ˆ |s c† csσ rσ = r.s=1 σ ˆ r|hmf |s c† csσ rσ (M ) (9. 2.6 Self-Consistent Field Algorithm The Self-Consistent Field (SCF) approximation. Step 1: Choosing an Initial Reference State One defines the Hamiltonian ˆ (1) Hmf = r. . r = 1.158) where the labels m are ordered to obey the condition (1) m < (1) n for m < n . . the reference state resulting from step 2M + 1 (within numerical error) is equal to the reference state resulting from step 2M − 1.160) . S . determining in a step (2) the corresponding mean field Hamiltonian (9. . .157) and determines the associated diagonal representation defined through S (1) ˆ r|t|s Usm = s=1 (1) (1) m Urm . 2. 2. 9.e. The procedure determines the reference state as the ground state of the independent-particle Hamiltonian (9. often also referred to as the Hartree-Fock approximation. . (9. (9. . SCF-Algorithm. S . SCF-Algorithm..e.: Determine Mean Field Hamiltonian In this step the mean field Hamiltonian S ˆ (M ) Hmf = r. can achieve this goal only iteratively. . under conditions which often are realized. Let us state now the construction of the self-consistent field ground state in more detail.145). The state thus determined is referred to as the self-consistent independent-particle ground state. r = 1. is the lowest energy independent-particle ground state one can construct. assuming in an intial step (1) a properly chosen reference state. i.119. . S . m = 1. steps (2) and (3) are being repeated M times until the procedure converges. however. m = 1.155) following the method outlined in Section 9..4. S. 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.e. 2.142). U(1) is the S ×N –matrix (1) of the first N column vectors of Urm . . i. and obtaining in a step (3) the ground state of this one-particle Hamiltonian according to the construction in Section 9. i. This procedure. . Step 2M. . is exact only for an independent-particle Hamiltonian.

indicating that the state converged. . (9.164) ˆ (a) Determine the expectation value of Prs for the state as defined in (9. .s=1 (9. S. s ) ˆ ˆ m=1 Utm ∗(M ) (M Uum ) (9. SCF-Algorithm. S do not exceed a preset threshold.145). t| v |u. . s = 1. (e) Consider the creation operators N † gqσ = m=1 Vmq d† mσ (9. 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.163) The result yields the reference state |Φo (U(M +1) ) using the definition (9.9. M = 1. (9. t| v |s. S .1: Let Prs be the one-particle operator with the generic operator prs = |r s|. (c) Show that any spin-independent one-particle operator S ˆ F = r. u − r.145). SCF-Algorithm: Continuation and Convergence Condition When step 2M + 1 is completed. . ˆ Exercise 9.: Diagonalize Mean Field Hamiltonian In this step the eigenvalue problem S (M r|hmf |s Usm +1) = s=1 (M ) (M +1) (M +1) Urm m . σ p = δtr δsu δσσ . 2. S . m = 1.. σ|ˆrs |u.165) can be written ˆ F = S ˆ ˆ r|f |s Prs r. the S × S–matrix S (M ) r|hmf |s N 269 ˆ = r|t|s + t. . . . . i. i. . 2.. ˆ t. . 2. . ˆ (b) How can the expectation values of the operator Prr be interpreted.6. i. 2 . Step 2M + 1. m = 1.6: Self-Consistent Field Algorithm is evaluated. U(M +1) is the (M +1) S × N –matrix of the first N column vectors of Urm . . 2.e. . r = 1. S (9. . step 2M + 2 is carried out. .e.167) .e.u=1 ( 2 r. etc.162) is solved adopting an ordering of the labels m which obeys the condition (M +1) m < (M +1) n for m < n . r = 1. . .166) ˆ (d) Define a similar two-particle operator Prstu .161) is calculated.s=1 σ ˆ r|f |s c† csσ rσ (9. 2.

Note that this is not an mσ S × S–matrix! Show that the reference state defined through N † † gqα gqβ |0 q=1 (9. .7 Properties of the SCF Ground State We want to investigate now the properties of the SCF ground state.163)..171) where the convention (9. Equation (9.168) through a physical observation? Exercise 9.172) is obeyed.6. (SCF ) m < (SCF ) n for m < n (9. (9. . We denote the representation which results from the SCF algorithm after its convergence as follows: S |m = ˜ r=1 (SCF Urm ) |r . We begin by summarizing the result of the SCF algorithm. i.145).162. r = 1.145) and (9.173) . S denotes the initial single-particle states and Urm is the unitary S × S– transformation matrix obtained in (9.169) 9. For this representation holds N m| t |˜ + ˜ ˆn m 2 mm | v |˜ m ˜ ˜ ˆ n˜ − mm | v |m n ˜˜ ˆ ˜ ˜ = (SCF ) δmn m (9. (9.171) implies that for the mean field Hamiltonian holds S ˆ Hmf (U(SCF ) ) = m=1 σ (SCF ) † dmσ dmσ m . (9.270 Many–Particle Systems which are connected with d† through a unitary N × N –matrix (Vqm ). 2. (f) Can one distinguish the states (9.σ + 2v ( c† c1α c† c1β + c† c2α c† c2β ) + v 1α 2α 1β 2β + c† c2σ c† 2σ 1σ c1σ ) .170) (SCF ) Here |r .9.168) ˆ ˆ has the same expectation values for Prs and Prstu as the reference state (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.e. .

r. one obtains ˆ SCF H SCF N = N (9.177) 2 mm | v |mm ˜˜ ˆ ˜˜ − mm | v |m m ˜˜ ˆ ˜ ˜ . u and Urm .171) in terms of matrix elements r|t|s . and once as a probe particle being subject to the mean pair interaction.142). s|ˆ|t. Employing expressions (9.93) for the expectation values (diagonal elements) of the one-particle and two-particle part of the Hamiltonian (9.e. once as a member of the ground state.84) and (9. ˆ SCF H SCF .175) However. Exercise 9.171) yields ˆ SCF H SCF N = N (9.176) 2 mm | v |mm ˜˜ ˆ ˜˜ − mm | v |m m ˜˜ ˆ ˜ ˜ . the SCF ground state is N SCF = m=1 d† d† |0 . the mean-field Hamiltonian is diagonal in the SCF representation. exploiting the spinindependence of (9.1: Reformulate (9.177).173) holds N ˆ SCF Hmf (U(SCF ) ) SCF = 2 r=1 (SCF ) m .7.e.m =1 Comparision with (9. (SCF ) ˆ Exercise 9. . one can also determine the expectation value of SCF for the Hamiltonian (9. v (SCF ) Show that the resulting eigenvalue problem of the type (9. Since the system under consideration has only 2N particles altogether.7: Properties of the SCF Ground State 271 i. 2 m=1 m|t|m + ˜ ˆ˜ m.7.177).174) The first property we consider is the total spin character of SCF .4 that SCF is a total singlet state. Since this state is composed only of closed shells one can conclude following Section 9.142).162) is non-linear in Urm . 2 m=1 (SCF ) m − m.. We like to determine next the energy expectation value of SCF . mα mβ (9. (9.m =1 The second term originates from the fact that in the mean field approximation one assumes that each of the fermions is subject to an average pair interaction involving a 2N -particle reference state. This over-counting leads to the correction term in (9.142). the mean field potential ˆ Vmf (|Φo (U) ) defined in (9..2: Derive (9.9. i.146) counts a particle twice. Finally. Within the mean field approximation as described by (9.

the second term on ther.h.. describes the sites of a linear lattice. real number.u σσ where r. We consider as an example the Hubbard model of an infinite linear lattice and apply the model to describe magnetic instabilities in metals. reads 1 ˆ r. i.181) The Hubbard model. t = v δru δst δrs . r ∈ Z. The case n = 1 is termed the half filled band case.178) Here the index r. N0 which are populated by 2N electrons (we choose an even number of electrons for convenience).8 Mean Field Theory for Macroscopic Systems The SCF algorithm has been developed in Sections 9. Both N0 and N are macroscopically large numbers.178). v of the Hubbard Hamiltonian (9. as stated through Hamiltonian (9. .σ crσ + crσ cr+1.178) v contributes only in case that two electrons occupy the same lattice site. S N0 (9.178) and through the so-called filling factor n n= 2N N = .179) r+1.4–9. v represents the ‘on-site’ Coulomb repulsion. v (9. of (9. (9. is identical to that of the independent–particle Hamiltonian H0 (9. The first term on the r.180) V = rσ sσ 2 r. labeled r = −N0 + 1.s. β for ‘up’ and ‘down’ rσ spins.e.σ rσ The potential energy term.128). The one-dimensional Hubbard model is described by the Hamiltonian ˆ H = −t rσ † c† r+1.182) n can assume values 0 ≤ n ≤ 2. . except in case of one-dimensional systems..178). u c† c† cuσ ctσ . i. In spite of the simplicity of its Hamiltonian the Hubbard model. for example.e.7 for finite systems. t is assumed to be a positive. respectively) in state |r at lattice site r.h. We assume that there are altogether S = 2N0 lattice sites. s|ˆ|u.σ + v 2 c† c† crσ crσ ..272 Many–Particle Systems 9. t describes the coupling of state |r to states |r ± 1 at the two neighboring lattice sites. † ˆ Ho = −t c† . In the present Section we want to adapt the algorithm to systems containing a macroscopically large number of fermions. in the usual two–particle operator form. rσ rσ r σσ (9. s|ˆ|t. v (9.s. . in the two-dimensional copper oxide lattices of high temperature superconductors. . i.e.σ crσ + crσ cr+1.t. .192) derived below is filled half. cannot be solved exactly.s. The Hubbard Model The Hubbard model serves today the important role as the simplest manifestation of strongly correlated electron systems which arise in many instances in molecular and solid state physics. The operator c† (crσ ) creates (destroys) an electron with spin σ (σ = α. According to (9. Due to the lack of an exact solution in dimensions higher than one. is characterized through the parameters t. approximation schemes like the self-consistent field approximation play an important role. referring to the fact that in this case the band of single particle energies (9.178). of (9.

For this purpose we apply the SCF theory presented in Section 9. boundary effects are assumed to be negligible. 2N0 (9. Inserting (9.. . 9.178) can be determined by applying (9. The unitary transformation which diagonalizes H0 .183) where the operators d† are related to the original creation operators c† through the unitary mσ rσ ˆ transformation U defined in (9.184) and using (9.187) The energy levels are labeled such that 0 < ±1 < .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 .e.179).140). N0 (9. a freedom which we employ to adopt so-called periodic boundary conditions like those for the example “2N independent electrons on a ring” on page 262.184) ˆ We are now ready to apply step 1 of the SCF algorithm and diagonalize term H0 . Accordingly. i. Accordingly. mα mβ (9. < ±(N0 −1) < N0 holds as can be readily verified. we assume that the SCF ground state is given by (9.185) into (9. is Urm = √ 1 exp (irmπ/N0 ) .σ m=0 ∗ Urm Urm c† crσ .178) within the framework of the self-consistent field theory.174).181. rσ (9. rσ rσ (9.188) .9. We can now embark on step 2 of the SCF algorithm.6. the operator H0 can be re-written in the diagonal form N0 ˆ H0 = m=−N0 +1 σ m d† dmσ mσ (9. 9.155.185) ˆ According to (9.186) where m = −2t cos mπ . . we identify |r + N0 = |r and later let N0 go to infinity. The self-consistent field Hamiltonian Hmf which corresponds to the Hubbard Hamiltonian (9. in fact.144).8: Macroscopic Systems 273 Since we are dealing presently with a macroscopic system. one obtains for the mean field Hamiltonian N −1 ˆ ˆ Hmf (U) = H0 + v r. This task has ˆ been solved already on page 262. SCF Ground State for the Hubbard Model We seek to determine the ground state of the Hubbard model stated by (9.154). by N −1 ||SCF = m=0 d† d† |0 . Using (9.

187) form a quasi1 2 continuous energy band. 4 So far we have not exploited the fact that N0 and N are macroscopically large quantities. the ground state of the Hubbard model in the self-consistent field approximation can be determined exactly.185) one can prove rσ mσ N0 c† crσ = rσ rσ m=−N0 +1 σ d† dmσ . mσ (9.190). namely. π]. The energy Emf of this state can be calculated readily by using (9. i.f. . Apparently. Indeed. the electrons in the present description behave like independent particles with energies ¯m = m + vn .183) is not the only candidate for the mean field ground state of the Hubbard model.. hence. When N0 becomes macroscopically large the single particle discrete energy levels (9. according to the Pauli principle.109) and using (9. (9. Usually. . N0 ˆ (2) ˆ Hmf = Hmf = m=−N0 +1 σ m + vn † dmσ dmσ 2 (9.1). The energy m can be expressed as a function of a continuous variable k ∈] − π. . in the limit N0 → ∞.192) This dispersion law is presented in Fig. In other words. for km = mπ/N0 holds km ∈ −π + N0 . Indeed. .177) and mm |ˆ|mm = mm |ˆ|m m = ˜˜ v ˜˜ ˜˜ v ˜ ˜ which holds in the present case.189) and.190) We have used in the latter expression the definition (9. The ground state of the system. through the so-called dispersion law (k) = −2t cos k .182) of the filling factor n [c.191) where m is given in (9. there exist several ways of formulating mean field approximations. the self–consistency condition is exactly met and the SCF algorithm converged after two steps.1.e. 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. as given in (9. −π ≤k ≤π . can be obtained. the results of different formulations agree qualitatively. even for one and the same Hamiltonian. is already diagonal. To demonstrate this we consider an alternative formulation of the mean field approximation for Hamiltonian (9. One can see that −2t ≤ (k) ≤ 2t holds. −π + N0 . (9. dnσ according to (9.182)]. called the Fermi energy. the state with the lowest possible energy. π]. 9.e. ˆ The Hamiltonian Hmf . denoted by F (see Fig. Alternative Description of the Mean Field Approximation The state (9. crσ through d† . by filling up this energy band with electrons from the bottom of the band (which corresponds to k = 0) to a maximum energy. i. π and.178).274 Many–Particle Systems Expressing c† .187). but may differ quantitatively.. holds k ≡ km ∈] − π. 9. The bottom of the energy band corresponds to (0) = −2t and the width of the energy band is 4t.

(9.45.178) in terms of the occupation number operaˆ tors Nrσ = c† crσ . F = (±kF ) denotes the Fermi energy. say the one given in (9.1: The dispersion law (k) for independent electrons on an infinite one dimensional lattice.193) where.196) (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 . the operator Nrσ can be written ˆ Nrσ = Nrσ + δ (Nrσ ) ..e. (9. i.194) ˆ For a given reference state. 9.183). we have employed (9.56). First. Nrσ = SCF ||Nrσ ||SCF .178) and performing the summation over the spin indices results in ˆ ˆ H = H0 + v r ˆ ˆ Nrα Nrβ .e.8: Macroscopic Systems 275 Figure 9. Inserting (9. where Nrσ is the mean occupation number of the one particle state |rσ . in the last term on the right hand side. let us express the interaction term in (9. i. rσ rσ rσ (9.195) ˆ ˆ ˆ and where δ (Nrσ ) = Nrσ − Nrσ describes the fluctuations of Nrσ .193) into (9. The mean field approximation neglects the . the deviation of Nrσ from its mean value Nrσ for the corresponding reference state.9. The anti-commutation properties (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 .198) as the one obtained by using the mean field potential (9. In case nα = nβ the following construction will lead to a ground state which coincides with the non-magnetic ground state (9. yields essentially the same mean field Hamiltonian (9..194) by dropping all the terms which contain the fluctuations to second order.198) Let us assume that the ground state of the Hubbard model. Therefore. In this approximation one can express ˆ ˆ Nrα Nrβ = ≈ = ( Nrα + δ (Nrα )) ( Nrβ + δ (Nrβ )) Nrα Nrβ + Nrα δ (Nrβ ) + Nrβ δ (Nrα ) ˆ ˆ Nrα Nrβ + Nrβ Nrα − Nrα Nrβ (9. m+ 2 2 c† crσ − 2N0 rσ (9. Spin-Polarized Mean Field Ground State ˆ We want to consider now a ground state for Hmf defined through Nrσ = nσ = const (9. that nα and nβ assume different values. Because of Nrσ = c† crσ rσ N −1 = m. separating the occupation number operator into a mean value plus fluctuation and neglecting the fluctuations in 2nd order. consequently. but the mean number of particles with spin α can be different from the mean number of particles with spin β.e. the ground state of the system can have a non-zero local spin and. in the mean field approximation. i. 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 .183). is given by (9.m ∗ Urm Urm d† dm σ mσ = m=0 ∗ Urm Urm = n N = . (9.191) for the ground state energy. however.183) .195). For such state the mean occupation number nrσ is uniform at all lattice sites. Note that the procedure employed in (9.200) allowing. 2N0 2 holds ˆ Hmf = = mσ mv ˆ H0 + 2 vn2 2 rσ nv † vn dmσ dmσ − N . . non-vanishing magnetization.276 Many–Particle Systems fluctuations to second order and the mean field Hamiltonian is obtained from (9.199) which is identical to expression (9.146).

a spin–polarized ground state.. hence.206) ˜mf is minimized where µ is the Lagrangian multiplier considered an independent variable. The actual values of nσ (σ = α.g. i. (9. At this point we exploit the fact that our system is macroscopically large. For this purpose the sum in (9.192).204) Emf ≡ 2N0 2N0 m.σ with respect to nα and nβ .202) This Hamiltonian describes a system of non-interacting electrons with a spin-dependent dispersion law (9. 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) . 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. as is shown schematically in Figure 9. Therefore. e.206) ascertains that condition (9. the so-called Fermi energy F = µ (see below)..203) mσ = m + vn−σ . In this limit the discrete energy spectrum (9.205) is met. subject to the constraint n = nα + nβ . β) are determined by minimizing the energy density of the ground state Hmf 1 = [ mσ Nmσ ] − vnα nβ (9.203) is provided by the continuous function σ (k) = (k) + vn−σ .8: Macroscopic Systems from which we obtain nσ = The mean field Hamiltonian is then ˆ Hmf = m 277 1 2N0 Nmσ .206). In the limit N0 → ∞ this sum can be expressed as an integral. 2π (9. i.e.207) where (k) is given by (9. In fact.206) needs to be evaluated.9. (9. (9. N0 ∼ 1023 .. Since β (k) − α (k) = v(nα − nβ ). if nα > nβ then β (k) > α (k). E with respect to nα . 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. The values of nα and nβ are determined by minimizing the energy density (9. m (9.208) . According to this method one minimizes ˜ Emf = Emf + µ (n − nα − nβ ) .205) Such minimization is realized through the method of Lagrangian multipliers.2a. 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 ).e. The additional term in (9. such that the larger nσ . one can expect the system to lower its energy by assuming nα = nβ and. (9.201) ( m ˆ + vnβ ) Nmα + ( m ˆ + vnα ) Nmβ − 2N0 vnα nβ . nβ and µ.

holds π ρ( ) = −π dk δ( + 2t cos k) . (b) Graphical definition of the limiting energies α and β . If the function f depends explicitly only on (k) one can state π −π dk f ( (k)) = 2π π ∞ ρ( )d f ( ) −∞ (9.192) and. Since the equation f (k) ≡ 2t cos k + = 0 has two simple roots. Inserting this last result into (9. The shaded areas represent the filled portions of these bands by electrons.2: (a) Relative position of the two energy sub–bands α (k) and β (k) for ∆ = nα − nβ > 0. gives the the number of available one particle states per site.211) In order to calculate the integral we recall the following property of the Dirac–delta function δ [f (x)] = i δ(x − xi ) . therefore. which is usually called the density of states.212) where xi are the simple roots of the function f (x).209) where ρ( ) = −π dk δ ( − (k)) . and noting |f (k)| = |2t sin k| = 2t one obtains δ( − (k)) = 1 − cos2 k = (2t)2 − 2 . In our case (k) is given by (9. namely k1.210) Here δ(x) is the Dirac–delta function and ρ( ). unit energy and a given spin orientation of the particle. 2π (9. δ(k + π − arccos( /2t)) + δ(k − π + arccos( /2t)) (2t)2 − 2 . |f (xi )| (9. 2π (9. −π .278 Many–Particle Systems (a) (b) Figure 9.210).2 = ±(π − arccos( /2t)). For a given dispersion law (k) the density of states ρ( ) can be calculated by employing equation (9.211) and using π dkδ(k ± (π − arccos( /2t))) = θ(2t − | |) .

222) . to determine the unknown quantities α . (9. Replacing again the sum by an integral over energy one can write σ nσ = −2t ρ( )d (9. θ(x) = 1 if x > 0 and θ(x) = 0 if x < 0. (9. β . 2t[ then the θ function in (9. ∂ α ˜ ∂ Emf =0.213) and carrying out the resulting integral yields nσ = 1 arcsin( σ /2t) .. One obtains ˜ Emf = α β ρ( )d + −2t −2t ρ( )d + vnα nβ + µ(n − nα − nβ ) .213) The presence of the θ function in the above formula guaranties that the density of states vanishes for | | > 2t. 9.220) β ˜ ∂ Emf = n − nα − nβ = 0 . 9. 2 (9. µ.214) are given by the sum (9. outside the energy band (k).206)] depends only on continuous quantities. ∂ β ˜ ∂ Emf =0.e. one arrives at the following expression of the density of states ρ( ) = θ(2t − | |) π (2t)2 − 2 . i. namely.219) (9.. i.219.f. on conditions for a ground state of minimum energy are ˜ ∂ Emf =0. (9. in principle.219–9. (9.8: Macroscopic Systems 279 where θ(x) is the step function.217) and µ. v and n.9. ∂µ (9. + vnα − µ = 0 .201).215) nα and nβ in (9.e.e. (9. nα and nβ as a function of the parameters of the Hubbard model.2]. of t. We can employ the results obtained to express the ground state energy density (9. (9. F = α + vnβ = β + vnα . i.216) Using (9.f.220) µ can be readily obtained µ= 1 ( 2 α + β) + vn .204) through an integral expression. Fig. β (9.218) The derivatives can be readily determined and the conditions (9. if we assume that is restricted to the interval ] − 2t.216) allow one.213) can simply be replaced by 1.214) where σ denotes the value of (k) which corresponds to the top of the filled portion of the energy sub–band σ (k) [c. the necessary ˜ Since Emf [c.221) and (9.218) read ˜ ∂ Emf = ∂ α ˜ ∂ Emf = ∂ β and α + vnβ − µ = 0 . From (9..221) ∂µ Equations (9. π α. namely.

222) and (9. (9. If the slope of f (∆) at the origin is less than 1. i.224) v∆ ∆= β ρ( )d = 0 ρ( β + )d . non trivial. From the definition (9.228) Since the total number of states per site is finite.f. from (9.227) can be solved graphically. (9. . One can determine vc by equating the slope of f (∆) at the origin to 1.280 Many–Particle Systems Comparision of (9..3]. Figure 9. For this purpose we consider the magnetization of the mean field ground state ∆ ≡ nα − nβ . the slope of f (∆) at the origin is larger than 1 [c.220) and (9.e. f (∆) converges to a maximum value as ∆ → ∆max = n. a magnetic ground state.223) ∆ = 0 corresponds to the ground state (9. (9.223 one obtains ∆= and second.227).3.e. The slope of f (∆) at the origin depends on the Hubbard model parameter v. the Hubbard model has a ground state with ∆ = 0.226) one can obtain ∆ by solving (cf. Accordingly. One can express ∆ as a function of α and β in two different ways. (9.225) Defining the function v∆ f (∆) = 0 ρ( β + )d .183) without magnetization. if. (9. as reflected by the expression d[f (∆)] = vρ( d∆ β + ∆) = π (2t)2 v − (v∆ + 2 β) >0. (9. Indeed.229) has a wider range of validity than one can infer from the present analysis of the Hubbard model. The Stoner criterion (9. from (9. One still needs to determine α and β . i. and only if. For v above a critical value vc .227) Since f (0) = 0 one can infer that ∆ = 0 is always a solution (the so called paramagnetic solution) of condition (9.229) This cindition is known as the Stoner criterion and gives the critical value of v which determines the onset of the ferromagnetic long range order. We assume in the following discussion that the implicit ∆ dependence of β in f (∆) can be neglected.219–9.217) and (9.2b). For this purpose one plots f (∆) versus ∆ as is shown schematically in Figure 9.f. through the condition [c.227) has only the trivial solution ∆ = 0.228)] vc ρ ( β ) = 1 .226) ∆ = f (∆) .223) follows α α − v β . the slope of f (∆) is positive for all ∆ ≥ 0.. so called ferromagnetic solution. condition (9.215) reveals that µ is indeed equal to the Fermi energy F (see also Figure 9. condition (9.227) will have a second.226) follows that f (∆) is a monotonically increasing function of ∆. (9. Equation (9. (9. (9. First.

233) Employing the approximation sin(y + δ) − sin y ≈ δ cos y for small δ.3: Graphical solution of the mean field equation ∆ = f (∆).232) states that x − y is proportional to ∆.9.4.231) (9. 2t As already mentioned. Since ∆ vanishes near vc we set x = y + δ where δ is a small quantity.219. v (9. (9. β x+y x−y = = π(n − 1) 2t π (sin x − sin y) = π∆ .205) the following set of equations determining x. i.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.232) together with (9.234) 2t 2 The resulting phase diagram of the ground state. the plot of vc vs. the magnetization ∆ of the ground state vanishes and remains zero for values v < vc . (9.. n). any .8: Macroscopic Systems 281 Figure 9. y and.220.231) follows y= δ π π (n − 1) − ≈ (n − 1) . t.230) one obtains from (9. 9.e. By parameterizing α and β α = 2t sin x . β = 2t sin y . α. From (9. hence. n. Equation (9. the ground state is characterized by the quantities (v. (9. The corresponding 2t analytical expression provides the phase diagram of the ground state: when v approaches vc from above.233) yield the desired expression vc π = π cos (n − 1) . is presented in Figure 9. 9. We like to determine finally the dependence of vc on the filling factor n. Accordingly. 2 2 2 (9.

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

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

284 Many–Particle Systems .

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

x3 ) = r describes the space coordinates. 10.5) and the Dirac (Sect. for the transformed space-time–cordinates x µ = (x 0 .4) invariant. 10. the group of Lorentz transformations (10.5) √ One can interprete the quantity −s2 as a distance in a 4–dimensional Euclidean space if one chooses the time component purely imaginary. x 2 . x3 ) (10. 5 to the groups SO(3) and SU(2) of rotation transformations of space coordinates and of spin. The elements L ∈ L act on 4–dimensional vectors of position– and time–coordinates.7 a heuristic derivation of the two most widely used and best known Lorentz–invariant field equations.e.1 will be extended in Sect. This choice implies dim(time) = dim(length)..1 Natural Representation of the Lorentz Group In this Section we consider the natural representation of the Lorentz group L. but in a setting of representation theory methods as applied in Secti.. In such a space Lorentz transformations correspond to 4-dimensional rotations. the Lorentz transformations were formulated in a space in which the time component of xµ was chosen as a purely imaginary number and the space components real. Before introduco ing these general Lorentz–invariant field equations we will provide in Sects.1). The reason for this choice is that the transformations (10.1) The Lorentz transformations L describe the relationship between space-time coordinates xµ of two reference frames which move relative to each other with uniform fixed velocity v and which might be reoriented relative to each other by a rotation around a common origin. We will denote these vectors as follows def xµ = (x0 . We will. x2 . Minkowski Space Historically. x 1 . we will provide a more basic definition of the transformations.1).3) where x0 = ct describes the time coordinate and (x1 .1). 10. henceforth. t) as components. 0 1 2 3 (10. 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. We will find that this definition will lead us back to the transformation law (10. Note that the components of xµ all have the same dimension. 10. wave functions ψ(r.4 to cover fields. x1 . 10. Rather than starting from (10. namely that of length. i. however. t) and vectors with functions ψ(r. 10.7) equation.286 Relativistic Quantum Mechanics in Sect. Denoting by xµ the . x2 . The Group of Lorentz Transformations L = O(3. i. 10.e.e. namely the Klein–Gordon (Sect.1) leave the quantity s2 = (x0 )2 − (x1 )2 − (x2 )2 − (x3 )2 (10. assume new units for time such that the velocity of light c becomes c = 1. This will provide us with a general set of Lorentz–invariant equations which for various particles take the place of the Schr¨dinger equation.5. x 3 ) holds (x0 )2 − (x1 )2 − (x2 )2 − (x3 )2 = (x )2 − (x )2 − (x )2 − (x )2 . i.

(10.9)  1 0 0  0 −1 0 0   ( gµν ) =   0 0 −1 0  = g . the notation in (10. ν = 0.6) allows us to introduce the so-called summation conventon: any time the same index appears in an upper and a lower position. (10. Aµ ν .11) This condition specifies the key property of Lorentz transformations. The subset of GL(4. (10. i. Aµ ν B ν ρ = ν=0 Aµ ν B ν ρ . We will exploit this property below to determine the general form of the Lorentz transformations. 3 3 3 yµ xµ = new yµ xµ .5) invariant. 2. 1. the .. Aµν . Aµ ν xν = µ=0 new old Aµ ν xν .e.5) into scalar products. the latter set constituting a group. The upper index closer to ‘L’ denotes the first index of the matrix and the lower index ν further away from ‘L’ denotes the second index.6) 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. The second reason is that upper and lower positions allow us to accomodate the expression (10. 3: 4-vector: xµ . R).6) yields Lµ ρ gµν Lν σ xρ xσ = gρσ xρ xσ .] The following possibilities exist for the positioning of the indices µ. This will be explained further below.7) The reason for the notation is two-fold. L ∈ GL(4. Since this holds for any xµ it must be true Lµ ρ gµν Lν σ = gρσ . [ A more conventional notation would be Lµν . Aµν ..12) (10.1: Natural Representation of the Lorentz Group 287 coordinates in one reference frame and by x µ the coordinates in the other reference frame.10.10) Combining condition (10. 4 × 4 tensor: Aµ ν . 0 0 0 −1 (10. First. R). xµ .8) new old old The summation convention allows us to write (10. summation over that index is assumed without explicitly noting it.e. This condition can be written xµ gµν xν = x gµν x where  µ ν (10.9) and (10. The Lorentz transformations are non-singular 4 × 4–matrices with real coefficients. the Lorentz transformations constitute a linear transformation which we denote by 3 x µ = ν=0 Lµ ν xν . i. The Lorentz transformations form the subgroup of all matrices which leave the expression (10.6) x µ = Lµ ν xν . ν=0 (10. however.

15) (10. a group. . the inverse of L 1  L−1 (10.18) (10.17) one obtains (L−1 )T g L−1 = gLgggLT g = gLgLT g . of (10. (10. (10.22) (10.12) LT gL = g .1) ⊂ GL(4.14).1).14) from the right by gLT and using (10. i. This stipulates that O(3. L3 ∈ O(3.15) one can derive LT gLgLT = LT and multiplying this from the left by by g(LT )−1 yields L g LT = g Using this result to simplify the r. hence.e. the associated inverse obeys (10.1) is. employing expressions (10. L−1 ∈ O(3.1) since it satisfies (10. = g LT g =  0 1 2  −L 2 L 2 L 2 L3 2  −L0 3 L1 3 L2 3 L3 3 (10.. For L3 = L1 L2 holds LT g L3 = LT LT g L1 L2 = LT (LT gL1 ) L2 = LT g L2 = g . From this one can obtain using g2 = 1 1 (gLT g)L = 1 and. property (10.288 Relativistic Quantum Mechanics elements of which satisfy this condition.14)  L0 0 −L1 0 −L2 0 −L3 0  −L0 1 L1 1 L2 1 L3 1   . of O(3. R).14) holds for the inverse of L.1). 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. is called O(3.e. To demonstrate the group property of O(3.10) of g one can rewrite the invariance property (10. in fact.20) Multiplying (10. One can also show that if L ∈ O(3.16.20) results in the desired property (L−1 )T g L−1 = g .e. L2 ∈ O(3.14).17) we note first that the identity matrix 1 is an element of O(3.s. L3 0 L3 1 L3 2 L3 3 Using the definition (10.19) i.. L ∈ GL(4.21) i. R) is a group.e..1). i.h. LT gL = g } .1).16) The corresponding expression for (LT )−1 is obviously (LT )−1 = (L−1 )T = g L g .13) L = ( Lµ ν ) =  2 0  L 0 L2 1 L2 2 L3 3  . 3 2 1 2 1 2 (10. In fact.1). 10. 1) = { L.1).. (10. We consider 1 then L1 . We want to show now L = O(3. This set is identical with the set of all Lorentz transformations L.

23) One can classify Lorentz transformations according to the value of the determinant into two distinct classes. 1). (10. L0 0 (10. 1). L ∈ O(3. For this purpose we consider first the value of det L.23) and (10. L0 0 ≥ 1} . The above shows that the set of proper Lorentz transformations L↑ allows one to generate all Lorentz transformations.. = { L. L↑ contains 1 1. L0 0 ≥ 1} . Obviously.10.1). 1). A second class property follows from (10.26) (10. For this purpose we consider the value of C 0 0 = A0 µ B µ 0 = 3 A0 j B j 0 + A0 0 B 0 0 . = { L.27) (10.1: Natural Representation of the Lorentz Group Classification of Lorentz Transformations 289 We like to classify now the elements of L = O(3. It holds g ∈ L and −1 ∈ L as one can readily verify testing for property (10.29) (10. σ = 0 yields L0 0 2 − L1 0 2 − L2 0 2 − L3 0 2 = 1. L↓ are disjunct sets defined as follows + + − − L↑ + L↓ + L↑ − L↓ − = { L. From this we can conclude L0 0 ≥ 1 or L0 0 ≤ −1 . Properties (10. + We start our investigation by demonstrating that L↑ forms a group.34) j=1 j=1 .12). there exist two other distinct classes.28) (10.24) or since (L1 0 )2 + (L2 0 )2 + (L3 0 )2 ≥ 0 it holds (L0 0 )2 ≥ 1.25) can be stated as follows: The set of all Lorentz transformations L is given as the union L = L↑ ∪ L↓ ∪ L↑ ∪ L↓ + + − − where L↑ . ∃M2 . L ∈ O(3. det L = −1. (10. L ∈ O(3.14). L0 0 ≤ −1} .32) (10.12) the case ρ = 0. entirely suitable to investigate first only 1. B ∈ L+ holds C = AB ∈ L+ . L↑ .33) where we used the definition aM = {M1 . Considering in (10. hence. + except for the trivial factors g and −1 It is.14). L↓ . M1 = a M2 }. 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 = −1.14) which we employ in the formulation (10. One can also 1 verify that one can write L↑ − L↓ + L↓ − = gL↑ + = = −L↑ . + + ↑ ↑ We can also demonstrate that for A.31) (10.25) i. A statement on this value can be made on account of property (10. + = − L↑ g + (10.30) ≤ −1} . = { L. 1). Lorentz transformations in L↑ . det L = 1. (10. (10. det L = 1. L ∈ O(3. M2 ∈ M. + − gL↑ + = L↑ g .e. Using det AB = det A det B and det AT = det A yields (det L)2 = 1 or det L = ±1 .

13).39) where we have employed the matrix form T 1 defined as in (10.14) actually L0 0 ≥ 1 and det L = 1 must hold. if we enforce (10. + i. Since the associative property holds for matrix multipli1 1 + ↑ cation we have verified that L+ is indeed a subgroup of SO(3. accordingly.e.290 Relativistic Quantum Mechanics From (10. + Accordingly.1) ascertains CT gC = g it must hold C 0 0 ≥ 1. from which one can conclude L−1 ∈ L↑ . In fact.1). (10. This relationship shows (L−1 )0 0 = L0 0 . holds L0 0 > 0 and the value of the determinant is close to unity. 1 ∈ L↑ . 1 0 0 = ≥ 1 and. since the group property of O(3. µ ν small .35) One can conclude. Since A0 0 ≥ 1 and B 0 0 ≥ 1.16)..14) implies 1 + T g (1 + ) = g 1 1 (10. therefore.40) and. orthochronous Lorentz transformations.34) provides then the estimate conclude from (10.36) It holds.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 . δ00 = 1 and It should be noted that according to our present definition holds δµν = gµρ δ δ11 = δ22 = δ33 = −1.38) For these transformations.e. For the inverse of + ↑ any L ∈ L+ holds (10. + transformations in a small neighborhood of 1 which we parametrize by infinitesimal parameters.12) follows (B 0 0 )2 − 3 (B j 0 )2 = 1 or 3 (B j 0 )2 = (B 0 0 )2 − 1. one can j=1 j=1 3 0 )2 = (A0 )2 − 1.. hence.37) (10.1). Infinitesimal Lorentz transformations The transformations in L↑ have the property that they are continously connected to the identity 1 1. obviously. Using the above expression for C 0 0 one can state C 0 0 > 0. obviously j=1 A0 0 B 0 0 ≥ 1. (10. Similarly. ρ ν (10. In the following we + will consider solely this subgroup of SO(3. 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. 1 We will then employ the Lie group properties to generate all transformations in L↑ . . L↑ is called the subgroup of proper. The next group property of L↑ to be demonstrated is the existence of the inverse. i. Property (10. To order O( 2 ) holds g + g = 0. | 3 A0 j B j 0 | < A0 0 B 0 0 . we consider transformations Lµ ν = δ µ ν + µ ν . 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.

The generators are explicitly  0 0 0 0  0 0 0 0 J1 =   0 0 0 −1 0 0 1 0  (10. k j j = 1. 3) Jk = (ϑk = 1. K2 =  .15) one can conclude T 291 = −g g (10.48) (10.45) (10. ϑ3 . J ] [ Kk . w3 ) =   −w2 ϑ3 0 −ϑ1  −w3 −ϑ2 ϑ1 0 (10. 2. 2. other five parameters zero) . J =  0  3  0 1 0 0 0 0 0 0 0 0 0 0 0     (10. w2 .46)  0 0   0 0  . J2 =    0 0 0 −1  0 0 0 0   0 0 0 0 0  0 0 −1 0 1  . (ϑ1 . k = 1.  (10.43) k = − Inspection shows that the matrix has 6 independent elements and can be written   0 −w1 −w2 −w3  −w1 0 −ϑ3 ϑ2   . 3 .47)     0 −1 0 0 0 0 −1 0 0  −1   0 0   0 0 0 0  0 0  K1 =  . 2.10.49) .50) .44) This result allows us now to define six generators for the Lorentz transformations(k = 1. K ] [ Jk .41) which reads explicitly     0 0 0 0 0 1 2 3 1 1 1 1 0 1 2 3 2 2 2 2 0 1 2 3 3 3 3 3 0 1 2 3     =     − 0 1 2 3 0 0 0 0 0 1 1 1 1 0 2 2 2 2 0 3 3 3 3 − − − 1 2 3 − − − 1 2 3 − − − 1 2 3    .42) This relationship implies µ 0 j µ j = 0 = j 0 . ϑ2 . other five parameters zero) Kk = (wk = 1. j. 3 (10. w1 . 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 .1: Natural Representation of the Lorentz Group Using (10. = − k m Km (10. 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.

+ The commutation relationships imply that the algebra of the generators Jk . (10. w ∈ R3 3 k=1 wk Kk .52) where the 3 × 3–matrices Lk are the generators of SO(3) defined in Chapter 5. To determine the explicit form of this transformation we evaluate the exponential operator by Taylor expansion.1: Demonstrate the commutation relationships (10.35) 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   . k = 1. w = 0) = 0 0 0 R(ϑ) .. however. 2π[ .49. ϑ.51) for w = 0 correspond to rotations of the spatial coordinates. Relativistic Quantum Mechanics The commutation relationships (10. In analogy to equation (5.55) L 0 0 0 0  exp (w1 K1 ) = exp   Using the idempotence property 0 −1 −1 0 2 0 0 1 0 (10. 0  1 n (10. constitutes an overcomplete parametrization of the rotations (see Chapter 5).53) Here R(ϑ) are the 3 × 3–rotation matrices constructed in Chapter 5. k = 1.54) which.56) = 1 0 0 1 = 1 1 (10. 3 is closed. 2. 3 (10. Following the treatment of the rotation group SO(3) one can express the elements of L↑ through + the exponential operators L(ϑ. 3 where we have defined ϑ · J = k=1 ϑk Jk and w · K = following the algebra in Chapter 5. that the transformations (10. 10. A boost in the x1 –direction is L = exp(w1 K1 ).50). w) = exp ϑ · J + w · K . and using the relationship Jk = 0 0 0 Lk (10. (10. i. Exercise 7. We consider now the Lorentz transformations for ϑ = 0 which are referred to as ‘boosts’.e. Kk . 2. L(ϑ. For the parameters ϑk of the Lorentz transformations holds obviously ϑk ∈ [0.51) One can readily show.57) .292 as can be readily verified.49) define the Lie algebra associated with the Lie group L↑ .

Correspondingly. vk defined in (10. From (10. 10. .59) cosh w1 Using cosh2 w1 − sinh2 w1 = 1 one can identify sinhw1 = sinh2 w1 + 1. This property is.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 .62) According to (10.58) = cosh w1 1 + sinh w1 1 0 −1 −1 0 The conventional form (10. we can state vk ∈ ] − 1.61) √ √−v1 √   exp (w1 K1 ) =    2 1 − v1 1 2 1 − v1 0 0 0 0 0 0  0 0    1 0  0 1  (10.51) the explicit transformation for space–time–coordinates is then x1 = x1 − v1 t 1 − 2 v1 . cosh2 w1 − 1 and coshw1 = cosh2 w1 − 1 = cosh w1 sinh w1 sinh2 w1 + 1 . consistent with (10.1) of the Lorentz transformations is obtained through the parameter change sinh w1 v1 = = tanh w1 (10. This property of the wk -values contrasts with the property of the parameters ϑk specifying rotational angles which assume only values in a compact range. one obtains from (10.59) for the case k = 1 corresponds to the relative velocity of two frames of reference. x3 = x3 (10.60) sinh w1 = v1 / 2 1 − v1 . The range of the parameters wk can now be specified. 10. (10. We expect that vk can only assume values less than the velocity of light c which in the present units is c = 1.59). x2 = x2 . 10. (10.59) v1 = These two equations yield cosh w1 = 1/ and (10. however. 1[. (10.3. t = t − v1 x1 2 1 − v1 .10. for wk wk ∈ ] − ∞.64) We note that the range of wk -values is not a compact set even though the range of vk -values is compact.59) follows.6. Accordingly.59) can be written  1 2 1 − v1 √−v1 2 1 − v1 2 1 − v1 .1).56.63) which agrees with (10. in fact. ∞[ . (10.

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

using (10.2: Scalars. (10. accordingly.76) An important example of a covariant 4-vector is the differential operator ∂µ = ∂ = ∂xµ ∂ ..74) g µν aν .e.73) aµ = Lµ σ aσ . one can employ the tensors g µν to raise and lower indices of L ν establish here the consistency of the ensuing notation.72) of the covariant 4-vector follows = In µν and g µ as well. one can express [(L−1 )T ]µ ν = (L−1 )ν µ and. lower. −a2 . Multiplication (and summation) of x µ = Lµ ν xν by Lρ σ gρµ yields. One calls 4-vectors aµ covariant and 4-vectors aµ contravariant. xν = g νσ Lρ σ gρµ x µ .66) together with (10.78) To prove the 4-vector property of ∂µ we will show that g µν ∂ν transforms like a contravariant 4vector. 4–Vectors and Tensors Scalar Product then 295 4-vectors allow one to construct scalar quantities. i.16).17) Lµ σ is the transformation inverse to Lσ µ .e. We do not fact.77) The transformed differential operator will be denoted by ∂µ = def ∂ . i. −a3 ) (10. gσν xν = Lρ σ gρµ x µ and g µσ gσν = δ µ ν . (10. (10. the relationship being aµ = gµν aν = (a0 . Covariant 4-vectors transform like a µ = gµν Lν ρ g ρσ aσ where we defined g µν = gµν .73) Note that according to (10.10. aµ (10. −a1 .66). We like to point out that from definition (10.72) where aν is a vector with transformation behaviour as stated in (10.71) Contravariant and Covariant 4-Vectors It is convenient to define a second class of 4-vectors. If aµ and bµ are 4-vectors aµ gµν bν (10.17) can be written (L−1 )ν µ = Lµ ν . In fact. ∂t (10.12). We have duplicated the expression for the inverse of Lµ ν to obtain the correct notation in terms of covariant. and .79) This is the inverse Lorentz transformation consistent with (10. The respective vectors aµ are associated with the 4-vectors aµ . g µν ∂ν = Lµ ρ g ρσ ∂σ .75) (10. In any case one can express (10. We start from x µ = Lµ ν xν . This property follows from (10..70) is a scalar. ∂x µ (10. The 4-Vector ∂µ (10.12) a gµν b µ ν = Lµ ρ gµν Lν σ aρ bσ = aρ gρσ bσ (10.

66)..84) = 1 − (v)2 (10.e.86) One can write (10. 2 dt dτ 1 − v (10. 2 2 dt 1 − v 1 − v (10. (10..83) is a scalar under Lorentz transformations.82) (10. (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..e.296 Relativistic Quantum Mechanics contravariant. indices. 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. dτ 1 − v 2 dt p0 = E = √ m m dt = √ . 1 − v2 1 − v 2 dt (10.80) Multiplication by g λµ (and summation over µ) together with g λµ gρµ = δ λ ρ yields g λµ ∂µ = Lλ σ g σν ∂ν .79) allows us to determine the connection between ∂µ and ∂µ . i.85) 1 d d = √ .81) ∂ . d’Alembert Operator We want to construct now a scalar differential operator.g. upper.87) The remaining components of pµ can be written.89) . For this purpose we define first the contravariant differential operator ∂ µ = g µν ∂ν = Then the operator 2 ∂µ ∂ µ = ∂t − 2 (10. i.67) transforms like (10.− ∂t . In fact. ∂µ does indeed transform like a covariant vector.88) One can express then the momentum vector pµ = √ d m dxµ = m xµ . p1 = √ m v1 m dx1 = √ . Proof that pµ is a 4-vector We will demonstrate now that the momentum 4-vector pµ defined in (10. e. this operator is equal to the d’Alembert operator which is known to be Lorentz-invariant.

We like to rewrite the last result E 2 = p 2 + m2 or E = ± p 2 + m2 .67) E2 − p 2 = or pµ pµ = m2 which. (10. Lorentz-invariant.95) 2 p This obviously describes the energy of a free particle with rest energy ±m. the so-called Lorentz gauge.92) (10.93) (10.s. in fact. of (10.h. (10. i.3: Relativistic Electrodynamics 297 d The operator m dτ transforms like a scalar. t) = 0 .96) is a scalar and. t) and A(r. The momentum 4-vector allows us to construct a scalar quantity.91) (10.10. namely.s.69) is. . kinetic energy ± 2m and relativistic corrections.77) which allow one to express (10.s. m2 m2 v 2 − = m2 1 − v2 1 − v2 (10. ∂t V (r.89) must be a 4-vector.69) for the electrodynamic potential and the 4-vector derivative (10.96) in the form ∂µ Aµ = 0 . yields according to (10. hence.h. We will demonstrate now the 4-vector property of Aµ . the r. t) + · A(r.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. · A = 0. Lorentz Gauge In our previous description of the electrodynamic field we had introduced the scalar and vector potential V (r. If we can show that Aµ defined in (10. and. in fact. in (10.97) We have proven already that ∂µ is a contravariant 4-vector.h.89) alltogether transforms like a contravariant 4-vector.97) and. respectively. is a scalar. namely pµ pµ = pµ gµν pν = E 2 − p 2 Evaluation of the r. and had chosen the so-called Coulomb gauge (8. 8.96) The Lorentz-invariance of this gauge.s.e. 10.h. Expansion in should then be rapidly convergent. This gauge is not Lorentz-invariant and we will adopt here another gauge. Since xµ transforms like a contravariant 4-vector.90) (10. equivalently. can be demonstrated readily using the 4-vector notation (10. (10.12). pµ on the l. hence.94) 1 m In the non-relativistic limit the rest energy m is the dominant contribution to E. t). One obtains E = ±m ± p2 2m (p 2 )2 + O 4m3 (p 2 )3 4m5 .. for these potentials. a contravariant 4-vector then the l. in (10. of (10.

∂µ J µ (xµ ) = 0 . · A(r. (10.. thereby.104) where mn = mn is the totally anti-symmetric three-dimensional tensor defined in (5. m. 10. 3 (10. t) = 0 (10. (8. has to transform like a contravariant 4-vector. n = 1.101).98) can be written. i. The l.97) and. Equation (10. according to (10.h. 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   . it follows that J µ . (10.s. t) − 2 V (r. t) − 2 A(r. using (10. t) and current density J(r.s. (10.6) and (8. a 4-vector. · A(r.98) which reflects the principle of charge conservation.100. t) together with (10. This principle should hold in any frame of reference.83) and (10. (10. k. F µν =  x  Ey B z 0 −Bx  Ez −By Bx 0 Alternatively. t) = −∂t V (r. t) 2 ∂t A(r.99) must be a scalar. yields ∂µ ∂ µ Aν (xσ ) = 4 π J ν (xσ ) . t) . t) = ∂t · A(r.18) and. One can readily verify..96). i. using (10. t) + · J(r.101) Combining equations (10. in fact.69) and ∂ µ in (10.h. Since ∂µ ∂ µ transforms like a scalar one can conclude that Aν (xσ ) must transform like a 4-vector.e.h.77) and (10.69). t) are known to obey the continuity property ∂t ρ(r. t) . F mn = − mn (10. Consequently.105) . (10.96).17) using the identity (8.102) In this equation the r. 2.103) B . The respective equation for A0 = V can be obtained from Eq.100) Similarly. .s. Since ∂µ transforms like a covariant 4-vector. t) from (8.32). using (8. t) = 4 π J(r. t).9).298 Transformation Properties of J µ and Aµ Relativistic Quantum Mechanics The charge density ρ(r.13). Using · ∂t A(r. in fact. transforms like a 4-vector. prove that Aµ is. must transform likewise. one obtains 2 ∂t V (r. must be a scalar. t) = 4πρ(r.82) as follows F µν = ∂ µ Aν − ∂ ν Aµ . that F µν can be expressed through the potential Aµ in (10. of (10. t) = −∂t V (r.e.99) Since this equation must be true in any frame of reference the right hand side must vanish in all frames. We want to derive now the differential equations which determine the 4-potential Aµ in the Lorentz gauge (10. also the l.68). one obtains for A(r. this can be stated F k0 = −F 0k = E k .

106) In case that the Lorentz transformation Lµ ν is given by (10.1. 8. (10. B⊥ are.3: Relativistic Electrodynamics 299 The relationships (10.62) or.102) ∂µ F µν = 4π J ν .110) where E .10. equivalently. t).108) yields then the expressions for the transformed fields E and B .63).3.107)  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. These equations show that under Lorentz transformations electric and magnetic fields convert into one another. In a new frame of reference holds F µν = Lµ α Lν β F αβ (10.109) (10.103. From the definition (10. Maxwell Equations in Lorentz-Invariant Form One can express the Maxwell equations in terms of the tensor F µν in Lorentz-invariant form.105) and (10. B and E⊥ .2).111) One can readily prove that this equation is equivalent to the two inhomogeneous Maxwell equations (8. . by (10.112) (10. where we used (10. B . E⊥ = B⊥ E⊥ + v × B √ 1 − v2 B⊥ − v × E = √ 1 − v2 (10. t) and B(r. one can conclude from (10.113) which can be shown to be equivalent to the two homogeneous Maxwell equations (8. 8.4). Noting ∂µ F µν = ∂µ ∂ µ Aν − ∂µ ∂ ν Aµ = ∂µ ∂ µ Aν − ∂ ν ∂µ Aµ = ∂µ ∂ µ Aν .104) establishe the transformation behaviour of E(r.97). 10. The results can be put into the more general form E B = = E .105) of the tensor F µν one can conclude the property ∂ σ F µν + ∂ µ F νσ + ∂ ν F σµ = 0 (10. one obtains  Ey −v B Ez +v B  0 −Ex − √ 1 2z − √ 1 2y 1−v1 1−v1    Bz −v1 Ey By +v1 Ez   √ 2  Ex 0 − √ 2   1−v1 1−v1  F µν =  E −v B (10. the components of the fields parallel and perpendicular to the velocity v which determines the Lorentz transformation. respectively.

e. equivalent to (8. (10.115) This equation follows. E = m/ 1 − v 2 .5) where p = mv/ 1 − v 2 .s.116) (10.g.5). for µ = 1. demonstrated that (10. dE q = p·E dτ m or with (10. The µ = 0 component of (10.67) and charge q dpµ q = pν F µν dτ m (10. Equation (10.. employing (10.118) p· dt √ where we exploited the fact that according to p = mv/ 1 − v 2 holds p v.87)] and retain the definition E for the electric field.114) reads using (10.114).86) dE q = p·E .114) is. of (10. dτ m √ Employing again (10. We want to √ demonstrate now that this equation is equivalent to the equation of motion (8.103) dpx q = ( EEx + py Bz − pz By ) . dt E From this one can conclude.114) reads then. We want to express this force through the tensor F µν .104).86). provides an alternative description of the action of the Lorentz force.h. It holds for a particle with 4-momentum pµ as given by (10.120) is referred to as the Lorentz force. hence. yields dpx = q dt Ex + (v × B)x (10.114) where d/dτ is given by (10. For the spatial components.e. in fact.93).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.67).119) which is the x-component of the equation of motion (8.86) and (10. . To avoid √ confusion we will employ in the following for the energy of the particle the notation E = m/ 1 − v 2 [see (10..5) taking the scalar product with p dp = qp · E (10. using (10. hence. The term on the r.117) (10. We have. also from the equation of motion (8. i. however. 1 dE 2 1 dp 2 = = qp·E 2 dt 2 dt (10.5).120) (10.

10. (10. ψ ∈ C∞ (4).g. We will denote the intended representation by L(ϑ. (5.127) . Functions which represent 4-vectorsor other non-scalar entities.. K3 ) are the generators of L(ϑ.123) This definition corresponds closely to the function space representation (5. 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.126) ρ ewk Kk − 1 1 . in the present case we exclude the factor ‘−i’ [cf. We like to express now ψ (x µ ) in terms of the old coordinates xµ .e.. For this purpose one replaces xµ in (10. i. obey a different transformation law. we seek an expression which corresponds to (10. xµ ). The resulting expression should be a generalization of the function space representation (5. (10. w) = ρ(Lµ ν (ϑ.48) of SO(3). def (10.47). in as far as SO(3. and which can be constructed following the procedure adopted for the function space representation of SO(3).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 ) .4 Function Space Representation of Lorentz Group In the following it will be required to decribe the transformation of wave functions under Lorentz transformations. w)) = ρ eϑ·J + w·K which we present in the form L(ϑ.7. However. For such functions holds in the transformed frame ψ (Lµ ν xν ) = ψ(xµ ) (10.51) for the natural representation of the Lorentz group.124) In this expression J = (J1 . 10.121) covers solely the transformation behaviour of scalar functions. e. (10.3 or the bi-spinor wave function of electron-positron pairs in Sect.10.e.4: Function Space Representation of Lorentz Group 301 10. Accordingly.121) by (L−1 )µ ν xν and obtains ψ (xµ ) = ψ((L−1 )µ ν xν ) . 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 (10.125)].1) is a generalization (rotation + boosts) of the group SO(3).48) and (10.. J2 .125) def (10. i. the charge-current density in case of Sect. J3 ) and K = (K1 .42) of SO(3). w) = exp ϑ · J + w · K . taken at the pairs of points (x µ = Lµ ν xν . We need to emphasize that (10.121) which states that the function values ψ (x µ ) at each point x µ in the new frame are identical to the function values ψ(xµ ) in the old frame taken at the same space–time point xµ . K2 . w) which correspond to the generators Jk and Kk in (10. w. In this section we will investigate the transformation properties of scalar functions ψ(xµ ).

K ] = = k m Jm k m Jm = − k m Km . J2 ] = J3 . K obey the same Lie algebra (10. 0  1 (10.132) and. −sinϑ1 x2 + cosϑ1 x3 ) = ψ(xµ ) + ϑ1 (x3 ∂2 − x2 ∂3 ) ψ(xµ ) + O(ϑ2 ) . obviously. sinhw1 x0 + coshw1 x1 . x3 ) 2 = ψ(xµ ) + w1 (x1 ∂0 + x0 ∂1 ) ψ(xµ ) + O(w1 ) (10. x1 ∂3 − x3 ∂1 ] = [x3 ∂2 .133) We demonstrate this for three cases. x2 .128) eϑ1 J1 = e−ϑ1 J1 1  0 =   0 0   0 0 0 1 0 0   0 cosϑ1 sinϑ1  0 −sinϑ1 cosϑ1 (10. K2 ] = −J3 .128). 1 This result. obviously. namely [J1 . One can determine similarly K1 starting from coshw1 sinhw1  sinhw1 coshw1 =   0 0 0 0  0 0 1 0  0 0   .131) This yields for small w1 ρ ew1 K1 ψ(xµ ) = ψ(coshw1 x0 + sinhw1 x1 . i.134) .129) which yields for small ϑ1 ρ eϑ1 J1 ψ(xµ ) = ψ(x0 . (10.302 One obtains J1 = x3 ∂2 − x2 ∂3 . K ] [ Jk . J3 = x2 ∂1 − x1 ∂2 . x1 ∂3 ] − [x2 ∂3 .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. The generators J .49) as the generators of the natural representation. K2 ] = K3 : [ J1 . reproduces the expression for J1 in (10. x3 ∂1 ] = −x1 ∂2 + x2 ∂1 = J3 . cosϑ1 x2 + sinϑ1 x3 . J2 ] = [x3 ∂2 − x2 ∂3 . [ Jk .130) ew1 K1 −1 = e−w1 K1 (10. [K1 .126). J2 = x1 ∂3 − x3 ∂1 .e. In order to evaluate (10. which we like to demonstrate for J1 and K1 . and [J1 . x1 . (10. J ] [ Kk . the expression for K1 in (10.

141) (10. ∂Ω . ∂3 in terms of ∂R . One-Dimensional Function Space Representation The exponential operator (10. In both cases the radial coordinate is the quantity conserved under the transformations.137) where K3 is given in (10.139) x3 = R sinhΩ (10. For this purpose one expresses K3 in terms of hyperbolic coordinates R. x2 ∂0 ] − [x1 ∂0 . In the following we consider solely the case x0 ≥ 0. x3 as follows x0 = R coshΩ .85-5.143) .140) allow one to ∂ ∂ express the derivatives ∂0 . x0 ∂2 + x2 ∂0 ] = [x3 ∂2 . i. 303 (10.e. cothΩ = 3 . x0 ∂2 ] = x3 ∂0 + x0 ∂3 = K3 . a relationship which can also be stated R = and + − (x0 )2 − (x3 )2 (x0 )2 − (x3 )2 if x0 ≥ 0 if x0 < 0 (10.137).136) (10.139.140) 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.10.4: Function Space Representation of Lorentz Group [ K1 . (10. x0 ∂2 ] = −x2 ∂1 + x1 ∂2 = − J3 . The relationships (10.135) (10. K2 ] = [x0 ∂1 + x1 ∂0 . ∂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 .125) in the case of a one-dimensional transformation of the type L(w3 ) = exp w3 K3 .. 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 . 10. can be simplified considerably. Eqs. x2 + x2 + x2 1 2 3 in the case of SO(3) and (x0 )2 − (x3 )2 in case of transformation (10. x2 ∂0 ] − [x2 ∂3 .138) x3 x0 . (5. Ω which are connected with x0 . x0 ∂2 + x2 ∂0 ] = [x0 ∂1 .128).142) (10. [ J1 .87)]. 0 ∂cothΩ ∂x x = ∂R x0 = − ∂x3 R (10. K2 ] = [x3 ∂2 − x2 ∂3 . We note tanhΩ = ∂R x0 = .

− ) = i ∂ µ .151) reads ∂µ ∂ µ + m2 ψ(xµ ) = 0 (10. We will provide a ‘derivation’ of these two equations which stem from the historical development of relativistic quantum mechanics. however. complex-valued function. ˆ (10. (10. (length)−1 . extremely useful. (10.137) 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 ) . In the following we will employ so-called natural units = c = 1.147) (10. momentum. ) = i ∂µ . mass. 10. In these units the quantities energy.122). Further below we will provide a more systematic. and (time)−1 all have the same dimension.5 Klein–Gordon Equation In the following Sections we will provide a heuristic derivation of the two most widely used quantum mechanical descriptions in the relativistic regime. ˆ pµ =⇒ pµ = i(∂t .146) Applying the correspondence principle to (10. The latter property implies that upon change of reference frame ψ(xµ ) transforms according to (10. Free Particle Case A quantum mechanical description of a relativistic free particle results from applying the correspondence principle. but certainly is short and.121.128) yields K3 = x0 ∂3 + x3 ∂0 = ∂ . In natural units the Klein–Gordon equation (10.151) is called the Klein-Gordon equation. representation theoretic treatment.148) ∂ µ ∂µ + m2 (10.92) one obtains the wave equation − or 2 2 µ ∂ ∂µ ψ(xν ) = m2 ψ(xν ) ψ(xν ) = 0 .304 Inserting these results into the definition of K3 in (10.145) ∂Ω 10. is not very insightful.150) . ∂Ω Relativistic Quantum Mechanics (10. which allows one to replace classical observables by quantum mechanical operators acting on wave functions. In the position representation the correspondence principle states ˆ E =⇒ E = − ∂t i ˆ p =⇒ p = i which. in 4-vector notation reads pµ =⇒ pµ = i (∂t . The historic route to these two equations.144) The action of the exponential operator (10. The partial differential equation (10. namely the Klein–Gordon and the Dirac equations. therefore.149) where ψ(xµ ) is a scalar.

161) . t) 1 [ ψ ∗ (r. t) ψ(r. t) ] 2mi (10.151) One can notice immeadiately that (10. t)∂t ψ ∗ (r. t) = ψ ∗ (r.150) one considers Im ψ ∗ ∂µ ∂ µ + m2 ψ = 0 (10. 2mi (10.158) which yields 2 2 ψ ∗ ∂t ψ − ψ∂t ψ ∗ − ψ ∗ 2ψ +ψ ψ∗ − 2ψ∗ = ψ) = 0 (10. t) − ψ(r. t) ψ ∗ (r. In order to obtain the conservation law connected with the Klein–Gordon equation (10. t) (10.157) 2m which is equivalent to (10. t) = 0 i ( ψ ∗ (r.150) becomes ∂µ ∂ µ + m2 ψ (x µ ) = 0 (10.156) · jS (r. t) + where ρKG (r. t) − ψ(r. t) ψ ∗ (r. To 1 derive this conservation law one rewrites the Schr¨dinger equation in the form (i∂t − 2m 2 )ψ = 0 o and considers 1 2 Im ψ ∗ i∂t − ψ = 0 (10. t) ψ(r.155) describes the positive definite probability to detect a particle at position r at time t and where jS (r.154). Under Lorentz transformations the free particle Klein–Gordon equation (10.10.154) describes the current density connected with motion of the particle probability distribution.150) is invariant under Lorentz transformations. t) ) . t) − ψ(r. t) + where ρS (r. t) = 0 (10.153) is associated with a conservation law for particle probability ∂t ρS (r.152) which has the same form as the Klein–Gordon equation in the original frame. This follows from the fact that ∂µ ∂ µ and m2 are scalars. t) = − 1 2m 2 ψ(r. t) ψ(r. t) jKG (r.159) (10. (10. Current 4-Vector Associated with the Klein-Gordon Equation As is well-known the Schr¨dinger equation of a free particle o i∂t ψ(r. t) ) 2m 1 = ( ψ ∗ (r. t)∂t ψ(r. and that (as postulated) ψ(xµ ) is a scalar.5: Klein–Gordon Equation or 2 ∂t − 2 305 + m2 ψ(xµ ) = 0 .160) ∂t (ψ ∗ ∂t ψ which corresponds to − ψ∂t ψ∗) + · (ψ ψ∗ ∂t ρKG (r. t) = · jKG (r. t) = (10.

150) is a continuum of positive energies E ≥ m.166) ψo (p. λ|xµ )ψo (p. Today. λ) = λ ψo (p..163) which results in the expected [see (10.162) Inserting this into the Klein–Gordon equation (10.154). 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. i. not of particle probability.166) is Eo (p) ∗ ρKG (p. Generating a Solution Through Lorentz Transformation A particle at rest. for example pions. (10. the anti-particles carry a charge opposite to that of the associated particles. λ|xµ ) = 0 2 m2 + po .162) shows that the solutions of the free particle Klein-Gordon equation (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) . and of negative energies E ≤ −m.151) yields 2 2 Eo − p0 − m2 ψ(r.167) m which is positive for λ = + and negative for λ = −. λ|xµ ) according to (10. λ) associated with the corresponding wave functions ψo (p. t) = 0 (10. and the density ρKG (p. λ|xµ ) = N e−iλmt .161) and (10. λ = ± . λ = ± (10. o namely. is actually that of a charge density. ±) = ± 2 m2 + po (10. t). p0 . λ) actually describes charge density rather than probability. the Klein-Gordon equation is considered as a suitable equation to describe spin–0 particles.165) This result together with (10. The proper interpretation of the two cases is that the Klein–Gordon equation describes particles as well as anti-particles. corresponding to λ = +. according to (??) is decribed by the r–independent wave function ψo (p = 0. The density ρKG (p.e. in that the expression for ρKG is not positive definite. (10.164) The corresponding energy is Eo (po .p ei(p·r − λEo (p)t) Eo (p) = The spectrum of the Klein–Gordon equation (10. The proper interpretation of ρKG (r. λ|xµ ) (10. m 2 2 E0 = m2 + po . λ|xµ ) = Nλ. We denote this by summarizing the solutions as follows ∂µ ∂ µ + m2 ψo (p.150) are actually determined by po and by the choice of sign ±.93] dispersion relationship connecting E0 . (10. with p = 0. it had been realized later. corresponding to λ = −.306 Relativistic Quantum Mechanics This conservation law differs in one important aspect from that of the Schr¨dinger equation (10.

168). (10.172) as the energy [c.175) as the momentum of the particle and one obtains L(w3 )ψo (p = 0. In case of λ = + one obtains finally L(w3 )ψo (p = 0. using (10.166)] of the particle. A(r.h. This yields..168).170) The coordinate transformation (10.174) which agrees with the expression given in (10.61) yield for this expression −iλ √ m mv x0 − iλ √ x3 . in fact.176) 10.10.169) The addition theorem of hyperbolic functions cosh(Ω + w3 ) = coshΩ coshw3 + sinhΩ sinhw3 allows us to rewrite the exponent on the r. one has to interprete p = mv/ 1 − v 2 (10. of (10. In case of λ = −. t)) (10.166).173) (10. Ω.169) −iλ ( m coshw3 ) ( R coshΩ ) − iλ ( m sinhw3 ) ( R sinhΩ ) . λ = +|xµ ) = N ei(px 3 − Eo (p)x0 (10. for positive energy solutions. −A(r. (10. For this purpose we consider the wave function for a particle moving with momentum velocity v in the direction of the x3 –axis.138) and the relationships (10..6 Klein–Gordon Equation for Particles in an Electromagnetic Field We consider now the quantum mechanical wave equation for a spin–0 particle moving in an electromagnetic field described by the 4-vector potential Aµ (xµ ) = (V (r.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. i.s. (10. L(w3 )ψo (p = 0.f.171) One can interpret then for λ = +.138) to replace t = x0 by hyperbolic coordinates R. Such wave function should be generated by applying the Lorentz transformation in the function space representation p (10. i. 2 2 1−v 1−v (10.e. λ = −|xµ ) = N ei(px 3 + Eo (p)x0 . for negative energy solutions. (10. Aµ (xµ ) = (V (r. t).145) choosing m = sinhw3 . λ|xµ ) = = exp w3 ∂ ∂Ω N e−iλmRcoshΩ 3) N e−iλmRcosh(Ω+w . for the wave function (10.177) . p = − mv/ 1 − v 2 as the momentum of the particle relative to the moving frame and √ m = 1 − v2 m2 = 1 − v2 m2 + m2 v 2 = 1 − v2 m2 + p2 = Eo (p) (10. t).e. t)) .

179) (10. t) . According to the so-called minimum coupling principle the presence of the field is accounted for by altering energy. that the mass m of the particle.e. t) 2 (10.308 free classical particle classical particle in field (V. i.e.e.1 this becomes g µν (i∂µ − qAµ )(i∂ν − Aν ) ψ(xµ ) = m2 ψ(xµ ) which can also be written g µν πµ πν − . −A). t) = −i − q A(r. For this purpose we split-off a factor exp(−imt) which describes the oscillations of the wave function due to the rest energy. in .1. is much larger than all other energy terms. which couples a particle of charge q to an electromagnetic field described through the potential Aµ (xν ). albeit in a form.. it’s rest energy. i. we define ψ(r.93).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.69)] one replaces the quantum mechanical operators. t))2 ψ(r.. t) . t) = e−imt Ψ(r.180) + m2 ψ(r.150) −g µν (i∂µ )(i∂ν ) + m2 ψ(xµ ) = 0 . in keeping withnthe non-relativistic limit. and focus on the remaining part of the wave function. See also Eq. A) free quantum particle quantum particle in field (V. i. According to the principle of minimal coupling [see (10. t). (10.182) and seek an equation for Ψ(r. (10. (10. A) Relativistic Quantum Mechanics energy momentum 4-vector E p pµ E − qV p − qA pµ − qAµ i∂t ˆ p = −i i∂µ i∂t − qV ˆ ˆ p − qA = π i∂µ − qAµ = πµ Table 10.1: Coupling of a particle of charge q to an electromagnetic field described by the 4-vector potential Aµ = (V.178) According to the replacements in Table 10. according to the rules shown in Table 10.150). A) or Aµ = (V.. i∂t and −i in (10. We will also assume. 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 (10.147). momenta for classical particles and the respective operators for quantum mechanical particles in the manner shown. (10. In terms of space-time derivatives this reads (i∂t − qV (r. For this purpose one writes the Klein-Gordon equation (10. m2 ) ψ(xµ ) = 0 .

spin-0 mesons.h. t) ≡ 0 r + Ze2 r 2 + 2 − m2 π ϕ(r ) = 0 .g.181) yields (we assume now that the Klein–Gordon equation describes a 2 particle with mass mπ and charge −e) for qV (r. of (10.e.186) then are slowed down. t) = ˆ [p − q A(r.. larger than |i∂t Ψ/Ψ| and alrger than qV . i. | i∂t Ψ | << m .183) The term on the l. The approximation is justified on the ground of the inequalities (10. Ψ |q V | << m . t) (10. The Klein-Gordon equation (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 (10.2) of a non-relativistic spin-0 particle o moving in an electromagnetic field.189) ..s. (10. π − mesons are generated through inelastic proton–proton scattering p + p −→ p + p + π − + π + . This application demonstrates that the Klein–Gordon equation describes spin zero particles. atoms in which one or more electrons are replaced by π − mesons. the binding of a negative charge (typically an electron) while at the same time a lower shell electron is being emitted π − + atom −→ (atom − e− + π − ) + e− .181) reads then i ∂t Ψ(r. (10. i.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. 309 (10.184) where we neglected all terms which did not contain factors m. t)]2 + qV (r..185) This is. The process of π − meson capture involves the so-called Auger effect. A stationary state of the Klein–Gordon equation is described by a wave function ψ(xµ ) = ϕ(r ) e−i t .181) to a physical system we consider pionic atoms. To ‘manufacture’ pionic atoms.10. t) = − Ze and A(r. e.6: Klein–Gordon Equation with Electromagnetic Field particular.188) Inserting this into (10. however. Pionic Atoms To apply the Klein–Gordon equation (10.e. (10. (10.183). filtered out of the beam and finally fall as slow pions onto elements for which a pionic variant is to be studied. identical to the Schr¨dinger equation (10. t) 2m Ψ(r.

e. i. (10. φ. . φ) R (r) Y r m (θ. The similarity of (10. 2.194) and (10.196) In this expression the number n defined through n = n− −1 (10.310 Relativistic Quantum Mechanics Because of the radial symmetry of the Coulomb potential we express this equation in terms of spherical coordinates r. n = 1.195) The latter problem leads to the well-known spectrum En = − mπ (Ze2 )2 . r r r r sin θ r sin θ Ze2 2 r ) (10.193) (10. .e. φ) (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. . 2n2 = 0. (10. this quantity definitely must be an integer.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 .197) counts the number of nodes of the wave function.. the eigenfunctions of the operator L2 ˆ L2 Y m (θ. The Laplacian is 2 = ˆ 1 2 1 1 1 2 L2 2 ∂r r + 2 2 ∂θ sinθ∂θ + 2 2 ∂φ = ∂r r − 2 .192) leads then to the ordinary differential equation d2 − dr2 ( + 1) − Z 2 e4 2 Ze2 + + r2 r 2 − m2 π R (r) = 0 . (10. θ. .191) ˆ The operator L2 in this equation suggests to choose a solution of the type ϕ(r ) = where the functions Y in (10. φ) . . 1.198) .191) m (θ. .192) ˆ are spherical harmonics. (10.. i. . (10.199) ( + 1) − Z 2 e4 .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. n − 1 . φ) = ( + 1) Y m (θ.

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 ). . results in the (10.197. .203) spectrum . .194) d2 λ( ) ( λ( ) + 1 ) 2 Ze2 − + + dr2 r2 r 2 311 − m2 π R (r) = 0 .205) α √ 2 2 1+ (n − β + β −α) and one obtains the series of approximations 1 1+ ≈ ≈ ≈ 1+ ≈ 1+ (n − β + α √ β 2 −α)2 1 1+ (n − α 2β α + O(α2 ))2 1 1+ n2 − α n + O(α2 ) β α 1 α n2 + α2 βn3 + O(α3 ) − α2 8n4 1 α 2n2 + α2 2βn3 + O(α3 ) .6: Klein–Gordon Equation with Electromagnetic Field and one can write (10. − + 1.200) The bound state solutions of this equation should correspond to from (10. Introducing α = Z 2 e4 and 1 β = + 2 (10. . . 10. .204) reads 1 (10. 2. 2mπ 2 −→ λ( ) . . 1. + (10. (10. . . (10. . . m) = mπ 1 + Z 2 e4 (n− − 1 2 + ( + 1 )2 − Z 2 e4 )2 2 n = 1. 1. . . 2 e2 −→ e2 mπ .e.201) mπ Z 2 e4 m2 − m2 π π = − .. which corresponds to a bound = 0. . 2mπ 2 (n + λ( ) + 1)2 (10. . . i. = 0. .199) and defining EKG (n. n − 1 m = − . .202) Solving this for state) yields (choosing the root which renders = 1 + mπ Z 2 e4 ( n + λ( )+1 )2 ≤ mπ . .196) if one makes the replacement 2 values which can be obtained E −→ One obtains − m2 π .10. 1. . EKG = Using (10. n = 0.

204) EKG (n.e.208) These two equation can be combined i∂t + m2 − 2 i∂t − m2 − 2 Ψ(r. however. m) ≈ m − mZ 2 e4 mZ 4 e8 − 2n2 2n3 1 + − 1− Relativistic Quantum Mechanics α α2 α2 α2 − + + + O(α3 ) .208). for example. would have only a first order time derivative.209). through the correspondence principle. but in the present case rather the negative of the energy. the Klein–Gordon equation had been rejected since it did not yield a positive-definite probability density. the second term the non-relativistic energy..208. Application of the correspondence principle (10.. and the third term gives the leading relativistic correction.2 particles. are unsatisfactory since expansion of the square root operator involves all powers of the Laplace operator. 2n2 2βn3 8n4 4n4 (10. His solution . the Dirac equation introduced below. (10. The latter term agrees with observations of pionic atoms. however. In order to describe electron spectra one must employ the Lorentz-invariant 1 wave equation for spin. Also. 10. because of the equivalence of space and time coordinates in the Minkowski space such equation necessarily can only have then first order derivatives with respect to spatial coordinates. it does not agree with observations of the hydrogen spectrum. i. 4n (10. This derivative. 10. A more satisfactory Lorentz–invariant wave equation. = 1 states into groups of two and four degenerate states.206) 1 2 3 + O(Z 6 e12 ) . in fact. Many attempts were made by theoretical physicists to ‘linearize’ the square root operator in (10. a splitting of the six n = 2.e. i. ..146) leads to the wave equation i∂t Ψ(r. Equations (10. the wave function changes sign under reflection. Heuristic Derivation Starting from the Klein-Gordon Equation An obvious starting point for a Lorentz-invariant wave equation with only a first order time derivative is E = ± m2 + p 2 . i. the π − meson is a pseudoscalar particle. one with a positive-definite density. t) = ± m2 − 2 Ψ(r.7 The Dirac Equation Historically.208. but not an operator iγ · as suggested by the principle of relativity (equivalence of space and time). arises because the Klein–Gordon equation. is identical to the two equations (10. a feature which is connected with the 2nd order time derivative in this equation.312 ≈ From this results for (10. Dirac succeeded.207) Here the first term represents the rest energy. t) . but for a long time to no avail. The latter spectrum shows. 10.209). However.209) which. in turn. t) (10. Finally.e. It should feature then a differential operator of the type D = iγ µ ∂µ . is related to the equation E 2 = m2 + p 2 of the classical theory which involves a term E 2 . It must be pointed out here that does not denote energy.

these descriptions ultimately merging with the true theory of matter. 2. In fact. are associated with a positive-definite particle density. 3 . γ j . the realization that the linearization can be carried out only if one assumes a 4-dimensional representation of the coefficients γ µ . (10.210) yields −g µν ∂µ ∂ν − m2 = (iγ µ ∂µ + m)(iγ µ ∂µ − m) 1 = −γ µ γ ν ∂µ ∂ν − m2 = − ( γ µ γ ν ∂µ ∂ν + γ ν γ µ ∂ν ∂µ ) − m2 2 1 µ ν ν µ = − ( γ γ + γ γ ) ∂µ ∂ν − m2 2 (10. unspecified algebraic objects γ µ and γ ν . . Starting point is to boldly factorize. spin. (10. It turns out that no 4-vector of real or complex coefficients can satisfy these conditions. therefore. 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 . . Instead. exploited ∂µ ∂ν = ∂ν ∂µ .214) From this follows that γ 0 has real eigenvalues ±1 and γ j . ψd (xµ ).215) . so far. the quantities γ 0 . this was not so.212) where we have changed ‘dummy’ summation indices. We seek to identify the coefficients γ µ . Initially. the operator of the Klein–Gordon equation ∂µ ∂ µ + m2 = −( P + m ) ( P − m ) (10. it was assumed that the 4-dimensional space introduced by Dirac could be linked to 4vectors. γ 1 . (10. 2. For µ = ν condition (10.209).210) where P = iγ µ ∂µ . ψ2 (xµ ). 3 has imaginary eigenvalues ±i. therefore.208). γ ν ]+ = 2 g µν (10.e. one can impose the condition γ0 is hermitian .211) Obviously. beautiful mathematical theory and that. namely. However. one route to important discoveries in physics is the creation of new mathematical descriptions of nature. but did not commute the.211) into (10. γ 3 can only be realized by d×d–matrices requiring that the wave function Ψ(xµ ) is actually a d–dimensional vector of functions ψ1 (xµ ). . 3 are anti-hermitian .. quantities with the transformation law (10. Accordingly. j = 1.213) We want to determine now the simplest algebraic realization of γ µ .7: Dirac Equation 313 to the problem involved an ingenious step. γ 2 . this would lead to the two wave equations (P − m)Ψ = 0 and (P + m)Ψ = 0 which have a first order time derivative and. achieved through a beautiful mathematical theory. 2.213) reads (γ µ )2 = 1 µ = 0 −1 µ = 1. strengthens the believe of many theoretical physicists today that the properties of physical matter ultimately derive from a.66).10.1 2 and anti-particles. according to (10. Inserting (10. the 4-dimensionsional representation discovered by Dirac involved new physical properties. The discovery by Dirac. i. j = 1. Properties of the Dirac Matrices Let us now trace Dirac’s steps in achieving the linearization of the ‘square root operator’ in (10. yet to be discovered.

219) Using property (10.215). in fact. 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 ) . as long as det(γ µ ) = 0 the dimension d of the square matrices γ µ must be even so that (−1)d = 1. the γ µ are anti-commuting. rather than a scalar wave function. 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. σ j σ k = − σ k σ j for j = k . is implied by (10.224) . (10. (10.213) is satisfied. namely the Pauli matrices σ 1 . One can readily show using the hermitian property of the Pauli matrices (γ 0 )† = γ 0 and (γ j )† = −γ j for j = 1.314 For µ = ν (10.213) reads γ µ γ ν = −γ ν γ µ .221) =0 (10. is unique. ψ2 . For d = 2 there exist only three anti-commuting matrices. We will argue further below that the choice f γ µ . 1 (10. This property can also be written (γ µ )† = γ 0 γ µ γ 0 . ψ4 ) and where ∂µ denotes the differential operator ∂ µ operating to the left side. ψ3 .223) Inserting this into (10. γj = 0 −σ j σj 0 .217) Obviously.. (10. holds 2 σj = 1 .218) of the Pauli matrices one can readily prove that condition (10. except for similarity trasnformations. A proper choice is γ0 = 1 1 0 0 −1 1 . 3 which. (10.218) The Pauli matrices allow one. however. 2.222) ∗ ∗ ∗ ∗ where we have defined Ψ† = (ψ1 . The Adjoint Dirac Equation The adjoint equation is Ψ† (xµ ) i(γ µ )† ∂µ + m ← ← (10.222) and multiplication from the right by γ 0 yields the adjoint Dirac equation Ψ† (xµ ) γ 0 iγ µ ∂µ + m ← =0.220) where Ψ(xµ ) represents a 4-dimensional wave function. Relativistic Quantum Mechanics (10. σ 3 for which.216) i.e. in fact. From this equation one can conclude that also the following should hold ( iγ µ ∂µ − m ) Ψ(xµ ) = 0 which is the celebrated Dirac equation. σ 2 . rather than to the right side. to construct four matrices γ µ for the next possible dimension d = 4.

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 ) .Chiral Representation 315 The Dirac equation can be subject to any similarity transformation defined through a non-singular ˜ 4 × 4–matrix S.e.7.219) is the socalled chiral representation defined through ˜ Ψ(xµ ) = S Ψ(xµ ) .230) Schr¨dinger Form of the Dirac Equation o Another form in which the Dirac equation is used often results from multiplying (10. (10. 10.213) still holds. 3 .. t) = 0 (10. (10.225) . 2. γ ˜ Exercise 10. (10.227) leaves the algebra of the Dirac matrices unaffected and commutation property (10.226) (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 ˜ (10.7: Dirac Equation Similarity Transformations of the Dirac Equation . j = 1. αj = ˆ 0 σj σj 0 .10.1: Derive (refeq:Dirac-intro20a) from (10. . j = 1.228) (10. i. 2.229) The similarity transformation (10.231) ˆ whereα has the three components αj . 3 and ˆ β = 1 1 0 0 −1 1 . ˆ ˆ ˆ Ho = α · p + β m . [˜ µ .213). (10. γj = ˜ 0 σj 1 S = √ 2 −σ j 0 1 1 1 1 1 −1 1 1 (10. (10. j = 1.227).233) The eigenstates and eigenvalues of H correspond to the stationary states and energies of the particles described by the Dirac equation. γ ν ]+ = 2 g µν . 3 . and γ0 = ˜ 0 1 1 1 0 1 .222.227) A representation often adopted beside the one given by (10.221) from the left by γ 0 ˆ ˆ i ∂t + i α · − β m Ψ(r. 2.

6 and R. d3 . in case of C4 . One can conclude then that also any set of 4 × 4–matrices obeying (10.234) (10.Phys. to Dirac matrices. d2 = . Γ±9 = ±d2 d3 Γ±10 = ±d2 d4 . Rev. the matrix elements of C = A ⊗ B are Cjk.Mod.H.Good. There are (including the different signs) 32 elements in G4 which we define as follows Γ±1 = ±1 1 Γ±2 = ±d1 . satisfy the anti-commutation property dj dk + dk dj = 2 for j = k 0 for j = k Relativistic Quantum Mechanics d4 = γ 0 (10..213). hence. d2 .237) by means of the property (10. form a Clifford algebra C3 . Γ±13 = ±d1 d2 d4 . page 187. d4 = 1 0 0 −1 1 0 0 −1 0 1 1 0 0 i −i 0 (10. This property extends then to 4 × 4–matrices which obey (10.e. 27.235) as can be readily verified from (10. i.e.238) These elements form a group since obviously any product of two Γr ’s can be expressed in terms of a third Γr . 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 . m = Aj Bkm . 3 .. The representations of Clifford algebras Cm are well established. Obviously. The reader may . The Clifford algebra C4 entails a subgroup G4 of elements ± dj1 dj2 · · · djs . The associative algebra generated by d1 . Since the Γj are hermitian the similarity transformations are actually given in terms of unitary transformations.237) which are the ordered products of the operators ±1 and d1 . Γ±14 = ±d1 d3 d4 . .213). i. Γ±8 = ±d1 d4 .235). j = 1. 2.235) can differ only with respect to unitary similarity transformations. For example. Γ±5 = ±d4 Γ±6 = ±d1 d2 . (1955).236) where ‘⊗’ denotes the Kronecker product between matrices. To complete the proof in this section the reader may consult Miller ‘Symmetry Groups and their Application’ Chapter 9. Γ±3 = ±d2 .316 Clifford Algebra and Dirac Matrices The matrices defined through dj = iγ j . Γ±15 = ±d2 d3 d4 Γ±16 = ±d1 d2 d3 d4 (10.235) and. Γ±4 = ±d3 . j1 < j2 < · · · < js s ≤ 4 (10. The representations of this group are given by a set of 32 4 × 4–matrices which are equivalent with respect to similarity transformations. d4 . . The three Pauli matrices also obey the property (10. Γ±7 = ±d1 d3 . d4 is called a Clifford algebra C4 . Γ±11 = ±d3 d4 Γ±12 = ±d1 d2 d3 . any product 1 of the dj ’s can be brought to the form (10.

218) of the Pauli matrices σ j .. x = Lµ ν xν . x µ = Lµ ν xν (10.80).. in a moving frame holds iγ µ ∂µ − m Ψ (x ) = 0 . Exercise 7. the form of the Dirac equation is identical in equivalent frames of reference. in a moving frame should hold iγ ∂µ − m Ψ (x ) = 0 .80) by Lµ ρ and using (10.. namely..e. The latter transformations imply that coordinates transform according to (10.e. a result one could have also obtained by applying the chain rule to (10. i. 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. Exercise 7. Multiplication and summation of (10. µ µ (10.236) with the Dirac representation (10. We can.3: Demonstrate the anti-commutation relationships (10.6). transform coordinates and derivatives of the Dirac equation. and derivatives according to (10. Lorentz Transformation of the Bispinor State Actually.219). i. 2. in fact.e.218) of the Dirac matrices γ µ .214) follows that γ 0 has real eigenvalues ±1 and can be represented by a hermitian matrix. However.240) . 10. in frames connected by Lorentz transformations. and γ j .239) Form invariance implies that the matrices γ µ should have the same properties as γ µ . i.215).8: Lorentz Invariance of the Dirac Equation 317 also want to establish the unitary transformation which connects the Dirac matrices in the form (10. Infinitesimal Bispinor State Transformation We want to show now that a suitable transformation of Ψ(xµ ) does.10. 10. Hence. Except for a similarity transformation.e. therefore. it must hold γ µ = γ µ . we do not know yet how to transform the 4-dimensional wave function Ψ and the Dirac matrices γ µ . i.76) yields ∂ρ = Lν ρ ∂ ν .6).213. 3 has imaginary eigenvalues ±i and can be represented by an anti-hermitian matrix.241) .e. j = 1. Exercise 7.8 Lorentz Invariance of the Dirac Equation We want to show now that the Dirac equation is invariant under Lorentz transformations. The transformation is assumed to be linear and of the form Ψ (x ) = S(Lµ ν ) Ψ(xµ ) µ µ (10. (10. these properties determine the matrices γ µ uniquely.4: Show that from (10.2: Demonstrate the anti-commutation relationships (10. exist.

318 Relativistic Quantum Mechanics where S(Lµ ν ) is a non-singular 4 × 4–matrix.243) from the left by Lσ ρ gσν yields S(Lη ξ )γ µ S −1 (Lη ξ ) gρµ = Lσ ρ gσν γσ .249) from which we can conclude σµν = −σµν νµ = −σνµ i. hence. (10. We had introduced in (10.h. in the expression of the infinitesimal transformation ρ Lµν = g µν + µν (10..243) We want to determine now the 4 × 4–matrix S(Lη ξ ) which satisfies this condition. 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.243) in a form in which the Lorentz transformation in the form Lµν is on the r. .246) (10. One can.s. For this purpose we exploit (10.247) the infinitesimal matrix µν is anti-symmetric.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 . The elements of g are.244) from which.e.241) implies a similarity transformation Sγ µ S −1 .e.250) σµν = −σνµ .248) Here σµν denote 4 × 4–matrices operating in the 4-dimensional space of the wave functions Ψ. The infinitesimal transformation S(Lρ σ ) which corresponds to (10. (10. follows S(Lη ξ )γρ S −1 (Lη ξ ) = Lσ ρ γσ . the transformation (10.38) the infinitesimal Lorentz transformations in the form Lµ ν = δ µ ν + µ ν where the infinitesimal operator µ ν obeyed T = − g g. i. The proper starting point for a constructiuon of S(Lη ξ ) is actually (10.221) upon Lorentz transformation yields iS(Lη ξ )γ µ S −1 (Lη ξ ) Lν µ ∂ν − m Ψ (x ) = 0 .12) in the form Lν µ gνσ Lσ ρ = gµρ = gρµ . Multiplication of this property by g from the right yields ( g)T = − g. µ g ρν = µν and.247) can be expanded S( µν i ) = 1 − σµν 1 4 µν (10. Multiplication of (10. µ (10.. hence. state that the Dirac equation (10. One can finally conclude multiplying both sides by g ρµ S(Lη ξ )γ µ S −1 (Lη ξ ) = Lν µ γν . g is an anti-symmetric matrix. It holds then S( µν i ) = 1 − σµν 1 4 µν µν = S(− νµ i ) = 1 + σµν 1 4 µν . of the equation. (10. Ob1 viously. νµ (10. using gρµ γ µ = γρ .245) The construction of S(Lη ξ ) will proceed using the avenue of infinitesimal transformations. it must hold (10. S( µν ) should not change its value if one replaces in its argument µν by − νµ . however.

h. 10. σjk = [˜j . 3 .s.10. ˜ ˜ σ only six generators need to be determined) σ0j = ˜ i [˜0 .255) which can be demonstrated using the properties (10.s.228. (10. γj = ˜ 0 −σ j σj 0 .255) satisfy condition (10. a solution of condition (10. this condition can actually be expressed through six independent conditions. Exercise 7.251) into (10. 10. (10. namely. (10. Furthermore.253) 1 2 αβ ( δ µ β γα − δ µ α γβ ) . γβ ]− 2 (10.257) . i.254) is σαβ = i [ γα .s. For this purpose we need to verify that the algebra of the generators involving addition and multiplication is closed. In this representation the Dirac matrices γµ = (˜ 0 . γ µ ]− = 2i ( δ µ β γα − δ µ α γβ ) .229). (10.248.246) results then in a condition for the generators σµν − i ( σαβ γ µ − γ µ σαβ ) 4 αβ = νµ γν .252) results in the condition for each α. 10. of (10. Algebra of Generators of Bispinor Transformation We want to construct the bispinor Lorentz transformation by exponentiating the generators σµν . of (10.252) Since six of the coefficients αβ can be chosen independently. β [ σαβ . γj ] = γ ˜ 2 −iσ j 0 0 iσ j . j = 1. Comparing this with the l.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. 2. −γ ) are ˜ γ ˜ γ0 = ˜ 0 1 1 1 0 1 . like the expression on the l.e. For this purpose one needs to express formally the r.254).251) S −1 ( µν ) = 1 + σµν µν 1 4 Inserting (10. β and operate in the same space as the Dirac matrices.254) The proper σαβ must be anti-symmetric in the indices α. For this purpose we inspect the properties of the generators in a particular representation. γk ] = ˜ γ ˜ jk σ 0 0 σ . 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. the chiral representation introduced above in Eqs.h. In fact.216) of the Dirac matrices.h.s.h.256) One can readily verify that the non-vanishing generators σµν are given by (note σµν = −˜νµ .5: Show that the σαβ defined through (10.252) also as a sum over both indices of αβ . the expression on the r. (10.213.

not necessarily infinitesimal. .51) and S. In fact. one can express the tensor µν through w and ϑ using µν = µ ρ g ρν and the expression (10. Finite Bispinor Transformation The closedness of the algebra of the generators σµν defined through (10.262) This expression allows one to transform according to (10.255) yields a closed algebra. Since such operator does not change its block-diagonal form upon exponentiation the bispinor transformation (10.259) yields the desired connection between the Lorentz transformation (10. since both operations convert block-diagonal operators A 0 0 B (10.241) bispinor wave functions from one frame of reference into another frame of reference.258) again into block-diagonal operators.248) allows us to write the transformation S for any. One should. In this representation holds i ˜ − 4 σµν µν i = − (˜01 01 + σ02 02 + σ03 03 + σ12 12 + σ13 σ ˜ ˜ ˜ ˜ 2 1 (w − iϑ) · σ 0 = .261) We note that this operator is block-diagonal. µν in the exponential form i S = exp − σµν 4 µν .51). 2 0 −(w + iϑ) · σ 13 + σ23 ˜ 23 ) (10. therefore. i..259) We had stated before that the transformation S is actually determined through the Lorentz transformation Lµ ν . therefore.44)  0 −w1 −w2 −w3  w 0 ϑ3 −ϑ2   =  1  w2 −ϑ3 0 ϑ1  w3 ϑ2 −ϑ1 0  µν (10. We can finally note that the closedness of the algebra of the generators σµν is not affected by similarity transformations and that. (10. and since the algebra of the Pauli matrices is closed. in particular.320 Relativistic Quantum Mechanics Obviously.259) becomes in the chiral representation ˜ S(w. ϑ) = e 2 (w − iϑ)·σ 0 − 1 (w + iϑ)·σ 0 e 2 1 (10. any representation of the generators. the algebra of these generators is closed under addition and multiplication.e. the representation (10.260) Inserting this into (10. be able to state S in terms of the same parameters w and ϑ as the Lorentz transformation in (10. In order to construct an explicit expression of S in terms of w and ϑ we employ again the chiral representation.

The conservation law (10. (10. This transformation behaviour can also be deduced from the transformation properties of the bispinor wave function Ψ(xµ ). Obviously. The reason is that the r.224). (10. ϑ)˜ ˜ γ 0 ˜† 0 = 0 1 1 1 0 1 e 2 (w + iϑ)·σ 0 − 1 (w − iϑ)·σ 0 e 2 1 0 1 1 1 0 1 .267) has the desired property of being positive definite.263) obviously is a scalar under Lorentz transformations and that the left hand side must then also transform like a scalar.e. in fact. We will now determine the relationship between the flux j µ = Ψ (x )γ 0 γ µ Ψ (x ) † µ (10.263) Starting point are the Dirac equation in the form (10. the property applies then in any representation of S. for j µ (xµ ) = (ρ.221) and the adjoint Dirac equation (10. ϑ) = S(−w.268) We will prove this property in the chiral representation. and addition yields Ψ† (xµ )γ 0 The last result can be written ∂µ Ψ† (xν )γ 0 γ µ Ψ(xν ) = 0 . i. For this purpose we prove first the relationship S −1 = γ 0 S † γ 0 .221) from the left by Ψ† (xµ )γ 0 . (10. The time-like component ρ of j µ 4 iγ µ ∂µ + iγ µ ∂µ ← Ψ(xµ ) = 0 . j µ must transform like a contravariant 4-vector. Multiplying (10. (10.271) . j) = Ψ† (xµ )γ 0 γ µ Ψ(xµ ) .264) (10.8: Lorentz Invariance of the Dirac Equation Current 4-Vector Associated with Dirac Equation 321 We like to derive now an expression for the current 4-vector j µ associated with the Dirac equation which satisfies the conservation law ∂µ j µ = 0 .224) from the right by Ψ(xµ ).265) (10.270) We conclude that (10. (10.263) does hold.263) allows one to conclude that j µ must transform like a contravariant 4-vector as the notation implies.h.268) holds for the bispinor Lorentz transformation. the conservation law (10. Since ∂µ transforms like a covariant 4-vector.269) One can readily show that the same operator is obtained evaluating γ S (w.s.10. For our proof we note first ˜ ˜ S −1 (w.266) ρ(xµ ) = Ψ† (xµ )Ψ(xµ ) = s=1 |ψs (xµ )|2 (10. of (10.. −ϑ) = e− 2 (w − iϑ)·σ 0 1 (w + iϑ)·σ 0 e2 1 (10.

µ (10. as just mentioned.e. commute with each other. The additional degrees of freedom described by the four components of the bispinor wave function require.274) 10. bispinor wave functions and we need to determine corresponding components of the wave function. of which the wave functions are eigenfunctions as well. additional characterizations.268) µ j = Ψ† (xµ )S † γ 0 γ µ SΨ(xµ ) = Ψ† (xµ )γ 0 S −1 γ µ SΨ(xµ ) . The solutions provide also a complete. Note that we have assumed in (10.272) With S −1 (Lη ξ ) = S((L−1 )η ξ ) one can restate (10.273) where we have employed (10. however. (10. Like in non-relativistic quantum mechanics the free particle wave function plays a central role. i. 10.282. 10.246) S −1 (Lη ξ )γ µ S ( Lη ξ ) = (L−1 )νµ γν = (L−1 )ν γ ν = Lµ ν γν .233). In relativistic quantum mechanics a Dirac particle can also be characterized through a momentum. but also as providing a basis in which the wave functions of interacting particle systems can be expanded and characterized. 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)]. One obtains using (10.283)].231.232. Combining this with (10.275) The free particle wave function is an eigenfunction of Ho . not only as the most simple demonstration of the theory. 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. We will start from the Dirac equation in the Schr¨dinger form (10. a property which leads to the energy– momentum (dispersion) relationship of the Dirac particle. as is required for the mentioned property.f. the identification of observables and their quantum mechanical operators. For this purpose we consider . (10.. In the following we want to provide this characterization. (10. only two degrees of freedom of the bispinor four degrees of freedom are independent [c. The operators commute also with Ho in (10.233) o i∂t Ψ(xµ ) = Ho Ψ(xµ ) . As it turns out. the Dirac particles are described by 4-dimensional.272) results in the expected transformation behaviour µ j = Lµ ν j ν .271) that the Dirac matrices are independent of the frame of reference. As pointed out.322 Relativistic Quantum Mechanics in a moving frame of reference and the flux j µ in a frame at rest. the wave function has four components which invite further characterization of the free particle state. t) and is determined through the momentum p ∈ R3 . These operaˆ tors. (10. In case of non-relativistic quantum mechanics the free particle wave function has a single component ψ(r.76). specific for the Dirac free particle. 10.9 Solutions of the Free Particle Dirac Equation We want to determine now the wave functions of free particles described by the Dirac equation.

one finds using (5.282. the socalled helicity of the particle.281). These two relationships are consistent with each other. E(p) = m2 + p 2 . In fact.10.9: Solutions of the Free Particle Dirac Equation the following form of the free Dirac particle wave function Ψ(xµ ) = φ(xµ ) χ(xµ ) = φo χo ei(p·r − t) 323 (10.279) According to the property (5.231.282) (10.279) the well-known relativistic dispersion relationship 2 = m2 + p 2 (10. (10.234) and (10. and φo .280) σ·p (σ · p)2 φ0 = χ0 = +m ( + m) ( + m) p2 χ = χ0 .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 . conclude 1. 1 (10.277) To solve this problem we write (10.93. Inserting (10. 2 − m2 0 (10.94) Equation (10.283) imply that the bispinor part of the wave function allows only two degrees of freedom to be chosen independently. 10.276) into (10.281) Obviously. χo each represent a constant.283) where is defined in (10. (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 .277) explicitly ( − m) 1 φo − σ · p χo 1 − σ · p φo + ( + m) 1 χo 1 = 0 = 0. (10. +m (10. from (10. later to be identified with momentum and energy. 10.278) provides us with information about the components of the bispinor wave function (10.280) which has a positive and a negative solution = ± E(p) . like the Klein–Gordon equation.234) of Pauli matrices holds (σ · p)2 = p 2 1 One can. We want to show now that these degrees of freedom correspond to a spin-like property. 10.276). two-dimensional spinor state. the Dirac equation. . reproduce the classical relativistic energy–momentum relationships (10. namely φo and χo are related as follows φ0 χo = = σ·p χ0 −m σ·p φ0 .284) The relationships (10. hence.

±) denote the corresponding solution of a particle in its rest frame. ±) γ 0 Ψ(p. +) = 1 . where S is given by (10. ±)γ 0 Ψ(p.286) Here N+ (p) is a constant which will be chosen to satisfy the normalization condition Ψ† (p. 0) γ 0 u 0 = |N+ (0)|2 . +) = |N+ (0)|2 (u† . is Ψ(p. ±) . We want to demonstrate now that the signs on the r.s. according to (10.s.290) which differs from the normalization condition (10. u† u = |u1 |2 + |u2 |2 = 1 . For this purpose we consider first the positive energy solution.h.e. ±) = = = Ψ† (0. = −E(p) . 10. we present the negative energy solution. the solution for = −E(p). is invariant under Lorentz transformations.219) and u† u = 1 [c. −|xµ ) = N− (p) −σ·p E(p) + m u u ei(p·r − t) .290). Ψ† (0. (10. (10.281)] separately.292) . of (10. ±) Ψ† (0. ±). The form of this condition and of (10. (10. (10..287) the form of which will be justified further below. ±) γ 0 Ψ(0.268). we present φo through the normalized vector φo = u1 u2 = u ∈ C2 . The connection between the solutions.259). = +E(p) . S −1 = γ 0 S † γ 0 and. i.. First.. of (10.s. +) = −1 . ±) Ψ† (0. and let Ψ(0. Ψ† (p. (10.285) The corresponding free Dirac particle is then described through the wave function Ψ(p.324 Relativistic Quantum Mechanics For our further characterization we will deal with the positive and negative energy solutions [cf.h.287) in the minus sign on the r. (10. ±) S † γ 0 S Ψ(0.288) corresponding to the wave function Ψ(p. (10.h. according to (10.289) Here N− (p) is a constant which will be chosen to satisfy the normalization condition Ψ† (p.e.241). hence. 10.287) will be justified now.286) for p = 0 yields. Similarly. +) γ 0 Ψ† (p.285)]. i. +) γ 0 Ψ(0.e. Hence.f. through χo given by χo = u1 u2 = u ∈ C2 . ±) γ 0 S −1 γ 0 γ 0 S Ψ(0. using γ 0 as given in (10.287. S † = γ 0 S −1 γ 0 . ±) = S Ψ(0. u† u = |u1 |2 + |u2 |2 = 1 .287. (10.291) Note that we have used that. Employing (10. +|xµ ) = N+ (p) u σ·p E(p) + m u ei(p·r − t) . (10. For the positive energy solution. we demonstrate that the product Ψ† (p. +) γ 0 Ψ† (p. ±) denote the solution of a free particle moving with momentum p in the laboratory frame. (10. ±). the l. One can see this as follows: Let Ψ(p. i. the solution for = +E(p).290) should differ.

We consider first the positive energy solution. this requires the choice of a negative side on the r.295) Evaluating the l.298) This result completes the expression for the wave function (10. (5.293) Obviously.289) yields Ψ† (0.234)] and using the normalization of u in (10. 2 E(p) (10. 325 (10.302) .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. 2m (10. Exercise 7.296) Replacing (σ · p)2 by p 2 [c.300) (10.s. ( m + E(p) )2 − p 2 (10. −) = |N− (0)|2 (0. We can also conclude from our derivation N± (0) = 1 .s.299) ( m + E(p) )2 .286) is 2 N+ (p) (u∗ )T .286).285) results in 2 N+ (p) 1 − p2 (E(p) + m)2 = 1 (10. using γ 0 as given in (10.f.h.h. of (10. u† ) γ 0 0 u = − |N− (0)|2 .287) written explicitly using (10.301) yields the normalization coefficient N+ (p) = m + E(p) . σ·p u∗ E(p) + m T γo u σ·p E(p) + m u = 1 (10.298) becomes N+ (p) = m + E(p) .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 (10.297) 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. −) γ 0 Ψ(0. (10. (10.10.219) yields 2 N+ (p) u† u − u† (σ · p)2 u (E(p) + m)2 = 1.290) to assign a positive value to |N− (0)|2 .

This can be verified expressing the space–time factor of Ψ(p.286) into (10. 3 is fairly obvious. have completed the determination for the wave function (10. Inserting (10. λ|xµ ) = pµ Ψ(p. (10.275). defined as the component of the particle spin along the direction of motion. The wave functions (10. Rather than considering the ˆ observable (10. The wave functions are also eigenstates of the momentum operator i∂µ .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. identical to the condition (10.307) where pµ = ( . This attribute is the so-called helicity.304) This condition is.305) and.286.296) for the normalization constant N+ (p) of the positive energy solution. j = 1.309) .306) i. λ|xµ ) in 4-vector notation.275) yields Ho Ψ(p.289) are not completely specified. (10.s. the two components of u indicate another degree of freedom which needs to be defined. The property [σ · p.290) written explicitly using (10. The commutation property [σ · p. not a constant vector. however. λ|xµ ) (10. i. Helicity The free Dirac particle wave functions (10. −p).326 Relativistic Quantum Mechanics We consider now the negative energy solution. yields 2 N− (p) u† (σ · p)2 u − u† u (E(p) + m)2 = −1 . 10.e. conclude N− (p) = m + E(p) 2m (10. The corresponding operator which measures this observable is 1 p ˆ Λ = σ· .303) E(p) + m u Evaluating the l. Condition (10. We ˆ want to show that this operator commutes with Ho and p to ascertain that the free particle wave ˆ function can be simultaneously an eigenvector of all three operators. We can. 10.e.307) we investigate first the observable due to the simpler operator σ · p.289) is T −σ·p −σ · p 2 E(p) + m u N− (p) u∗ . hence.289).308) 2 |p| ˆ Note that p represents here an operator. pj ] = 0 . i∂µ Ψ(p. 10. 2. exp[i(p · r − t)] = exp(ipµ xµ ). (u∗ )T γ o = −1 (10.289. Ho ] = 0 follows from (10. λ|xµ ) .. λ|xµ ) = λ E(p) Ψ(p.e. This degree of freedom 1 describes a spin– 2 attribute.286.. thereby. i.300) have been constructed to satisfy the free particle Dirac equation (10. the wave functions constructed represent eigenstates of Ho . (10.h..

.312) We assume now particles with negative energy.e.286) are eigenfunctions of Λ as well will specify now the vectors u. We assume first particles with positive energy. 2m (10. Therefore. t) = Np  e (10.224) 1 of σ3 the two u vectors (1. States which are simultaneously eigenvectors of these three operators are also simulteneously eigenvectors of the three operators p. t) = Np  e 2   1 p m + Ep 0   0   i(px3 − E t) 1 p e Ψ(pˆ3 .312). 1 p = (0. + 1 |r.10. Np = m + Ep .308) above. . hence. − 1 |r. 1)T are eigenstates of 2 σ 3 with eigenvalues ± 1 . i. Ho and σ·p commute with each other and. − 1 |r. = −E(p).e. t) = Np  e   1 0   0 p  m + Ep  i(px3 + E t) 1 p e Ψ(pˆ3 . +. i. are   1   i(px3 − E t) 0 p e Ψ(pˆ3 . According to the definition (5.9: 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 = 0.310) We have shown altogether that the operators p. 327 0 (σ · p)2 ˆ 2 (σ · p) ˆ 0 0 (σ · p)2 ˆ 2 (σ · p) ˆ 0 (10. Ho and Λ defined in (10. t) = Np  e (10.311) 2  −p  0 m + Ep 1 where e3 denotes the unit vector in the x3 -direction and where ˆ Ep = m2 + p2 . We consider first the simplest case that particles move along the x3 –direction. 0)T and (0. 0. i. In this case Λ = 2 σ 3 .313) 2   0 1 where Ep and Np are defined in (10. p3 ).. is independent of the direction of motion of the particle.e. +.. can ˆ ˆ be simultaneously diagonal. in principle. −. Since helicity is defined relative to the direction of motion of a particle the characterization of u as an eigenvector of the helicity operator. ˆ The condition that the wave functions (10. −. 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 the wave functions which are eigenstates of the helicity operator. = +E(p). + 2 |r.

+ 1 |r. + 2 |r.317) Ψ(p.316) describes a rotation which aligns the x3 –axis with the direction of p. 2  1  0  (10. (10.229).312) can be obtained also by means of the Lorentz transformation (10. t) 2 = Np 1 u(p. γ .319) 1 Ψ(p.315) e3 × p ˆ ∠(ˆ3 .320) where Ep and N are again given by (10. β..328 Relativistic Quantum Mechanics To obtain free particle wave functions for arbitrary directions of p one can employ the wave functions (10.e.313) except that the states (1.220). The corresponding wave functions are determined through ˜ i˜ 0 ∂t − m Ψ(t) = 0 .123). t) = Np ei(p·r + Ep t) (10. − 1 ) 2 −p 1 m + Ep u(p. − 1 ) 2 ei(p·r − Ep t) (10. 1 |t) = √2   e−imt . p) e |p| (10. − 2 ) u(p. − 2 |r. − 1 |r.262) for the bispinor wave function and the transformation (10. −.311. − 2 ) −p 1 m + Ep u(p. + 2 ) ei(p·r − Ep t) (10.] The corresponding free particle wave functions are then Ψ(p. for an r–independent wave function.318) 1 Ψ(p. +. + 2 ) u(p. a wave function which represents free particles at rest.226.314) (10. Generating Solutions Through Lorentz Transformation The solutions (10.321) . [One can also express the rotation through Euler angles α. i. 1)T have to be replaced by eigenstates u± (p) of the spin operator along the direction of p. γ and are  1  0  1 ˜ Ψ(p = 0.312). 5. − ) 2 where ϑ(p) = = = exp exp i − ϑ(p) · σ 2 i − ϑ(p) · σ 2 1 0 0 1 . t) 2 = Np u(p. 10. + ) 2 1 u(p.311. + 1 ) 2 p 1 m + Ep u(p. in which case the transformation is given by (5. denoted by ˜. These eigenstates are obtained through a rotational transformation (5. +.223) as follows 1 u(p. 0)T and (0. For this purpose one starts from the solutions of the Dirac equation in the chiral representation (10.222. +. 10. −. 10. t) = N√ ei(p·r + Ep t) (10. + 2 ) p 1 m + Ep u(p.

262) for a boost in the x3 –direction. Applying (10. for w = (0. For the resulting wave functions in the chiral representation one can use then a notation corresponding to that adopted in (10.  −1  e 0   0  1  +imt 1 ˜  Ψ(p = 0. 0. t) e 2 1  = √  2   e− 2 w3 1 1 1 0 1 0 0 1 0 1   i(px3 − E t) p e   ˜ Ψ(p(w3 )ˆ3 . −. 0 1 (10.325) One should note that φo . 10.322).262) φo χo → e 2 w3 σ 0 1 3 Et ) (10. i. χo are eigenstates of σ 3 with eigenvalues ±1.228) yields the corresponding solutions (10. w3 ).174. − 1 |r.325) to (10. +.323) Transformation (10. 10.10. 2  0  1   1  0  +imt 1 1 ˜  Ψ(p = 0.311)  e 1 w 2 3 ˜ Ψ(p(w3 )ˆ3 . +. 2 |t) = √2  .311. − 1 |t) = √2  . This correspondence justifies the characterization ±. −. yields for the exponential space–time dependence according to (10.9: Solutions of the Free Particle Dirac Equation  0  1  1 ˜ Ψ(p = 0.324. ± 1 of the wave functions stated in (10.322) can be written in spinor form 1 √ 2 φo χo e imt . + 1 |r. +. φo .323) should yield the solutions for non-vanishing momentum p in the x3 –direction. 2  0 e −1  329 (10. − 1 |t) = √2   e−imt . t) e 2 e− 2 w3 1  = √  1 2  e 2 w3  i(px3 − E t) p e  . χo ∈ 1 0 . 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 . (10.176) imt → i ( p3 x3 and for the bispinor part according to (10. 2 The solutions (10..313) in the p → 0 limit.e.322) The reader can readily verify that transformation of these solutions to the Dirac representationsas defined in (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  =    i(px3 + E t) p e  (10.228) yields w3  cosh 2 =   sinh w3 2 1 Ψ(p(w3 )ˆ3 . + 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 . −. − 2 |r.329) . Transformation to the Dirac representation by means of (10.328) the relationship (10. t) e 1  = √  2   −e− 2 w3 − 1 w3 2 1  i(px3 + E t) p e   ˜ Ψ(p(w3 )ˆ3 . + 2 |r. − 1 |r. and the expression (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 . − 2 |r.61) p(w3 ) = m sinhw3 . t) e  =     cosh w3 2 − sinh w3 2 sinh w3 2 cosh w3 2 − sinh w3 2 cosh w3 2  i(px3 − E t) p e   Ψ(p(w3 )ˆ3 . −. −. 2 (10. + 1 |r.326) where according to (10. 2 sinh x = 2 coshx − 1 .61) between the parameter w3 and boost velocity v3 .327) Employing the hyperbolic function properties cosh x = 2 coshx + 1 . t) e 2 e 1   = √  1 2 − e 2 w3  i(px3 + E t) p e  (10.

331) i S(Lη ξ )γ µ S −1 (Lη ξ ) ρ(Lη ξ )∂µ ρ−1 (Lη ξ ) − m Ψ (xµ ) = 0 (10. Invariance of Dirac Equation Revisited At this point we like to provide a variation of the derivation of (10. which however.333) .243). The transformation is Ψ (xµ ) = S(Lη ξ ) ρ(Lη ξ ) Ψ(xµ ) (10. Λ|t).123) above and characterized there. reproduces the positive energy wave functions (10. when we generated the solutions Ψ(p.330) Inserting expressions (10. we will derive this equation only for infinitesimal transformations.331) is a solution as well.176).311) as well as the negative energy solutions (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 . λ.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. 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 . but not the observer.243) and. the essential property stating the Lorentz–invariance of the Dirac equation. that if Ψ(xµ ) is a solution of the Dirac equation that Ψ (xµ ) as given in (10.331) where S(Lη ξ ) denotes again the transformation acting on the bispinor character of the wave function Ψ(xµ ) and where ρ(Lη ξ ) denotes the transformation acting on the space-time character of the wave function Ψ(xµ ). Λ|xµ ) from the solutions describing particles at rest Ψ(p = 0.243) considered solely the limit of infinitesimal transformations anyway. ρ(Lη ξ ) has been defined in (10. = γν . To show this we rewrite the Dirac equation (10. of course.330) into (10. and since (2) the calculations following (10. is sufficient since (1) it must hold then for any finite transformation.313) for −p.9: Invariance of Dirac Equation Revisited and similarly sinh w3 = 2 Ep − m = 2m p 2m (Ep + m) 331 (10. in general. 10. Actually. Making this expectation a postulate allows one to derive the condition (10.168–10. We expect.334) (10. Such description will be essential for the formal description of Lorentz invariant wave equations for arbitray spin further below.221) using (10. λ. indeed. refered to as the active transformation.329. 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. expresses the system in the old coordinates. The reason why we provide another derivation of (10. Such transformation had been applied by us. (10.10. In the new derivation we consider the particle described by the wave function transformed. This transformation. thereby.327). the proper transformation S(Lη ξ ).

342) One can readily convince oneself that these results are consistent with (10.340) [K2 .e.262). ∂µ ] = (10. it is met by S(Lη ξ ) as given in the chiral representation by (10. 1.338) [J3 . ∂µ ] = (10. ∂µ ] = [x1 ∂3 − x3 ∂1 . ∂µ ] = [x2 ∂1 − x1 ∂2 .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 (10.336).337) [J2 . . We will proceed by employing the generators (10. ∂µ ] = (J )ν µ ∂ν . 1) . As mentioned already we will show condition (10. w = (0.334) for infinitesimal Lorentz transformations Lη ξ . . the Dirac equation is invariant under active Lorentz transformations. 0). ϑ = (0.335) shows that we need to demonstrate [J . of course. . We have demonstrated. 0. ∂µ ] = [x0 ∂1 + x1 ∂0 . 0. [K . w = (0. 0).331) is again a solution of the Dirac equation. Inspection of (10. therefore. ∂µ ] = (10. w = (0. i. 0). that any solution Ψ(xµ ) transformed according to (10.336) We will proceed with this task considering all six cases: [J1 . ∂µ ] = (10.341) [K3 . ∂µ ] = (K )ν µ ∂ν . 0. ∂µ ] = (10. 0). ∂µ ] = [x3 ∂2 − x2 ∂3 .243) and.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. ϑ = (0.     0 0  ∂3   −∂2  0    −∂3 0    ∂1  0    ∂2  −∂1   0   −∂1   −∂0 0    0   −∂2   0   −∂0  0   −∂3   0 0    −∂0 µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ = = = = = = = = = = = = = = = = = = = = = = = = 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 (10.332 Relativistic Quantum Mechanics We will demonstrate now that the first condition is satisfied by ρ(Lη ξ ). 0. The second condition is identical to (10. ∂µ ] = (10. . ∂µ ] = [x0 ∂3 + x3 ∂0 .339) [K1 .. 0). . 0. ∂µ ] = [x0 ∂2 + x2 ∂0 .

344) Using the notation σ = (σ1 . Accordingly. accordingly. Equivalently.6 we assume that the field is described through the 4-vector potential Aµ and. 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.10. we replace the operator ∂µ by ∂µ + iqAµ .232). (10.343) One may also include the electromagnetic field in the Dirac equation given in the Schr¨dinger form o ˆ (10.347) (10. (10.345) in the so-called non-relativistic limit in which all energies are much smaller than m.349) We want to focus on the stationary positive energy solution.351) X (x ) µ .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.221) reads then [ iγ µ (∂µ + iqAµ ) − m ] Ψ(xν ) = 0 (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 (10.345) (10. σ3 )T one obtains then i∂t φ i∂t χ = = ˆ σ · π χ + qV φ + mφ ˆ σ · π φ + qV χ − mχ .1 in Sect 10. we replace in the Dirac equation the momentum operator pµ = i∂µ by i∂µ − qAµ where q is the charge of the respective particles ˆ (see Table 10.6 above).348) (10. σ2 .345) holds |qV | << m . For this purpose we choose the decomposition Ψ(xµ ) = φ(xµ ) χ(xµ ) .g. The Dirac equation (10. 10.10: Dirac Particles in Electromagnetic Field 333 10.350) (10. e.1) i∂t − qV and p by ˆ ˆ π = p − qA . for the scalar field V in (10.. Following the respective procedure developed for the Klein-Gordon equation in Sect. Non-Relativistic Limit We want to consider now the Dirac equation (10.233) by replacing i∂t by (see Table 10.346) ˆ ˆ ˆ α · π + qV + β m Ψ(xµ ) (10.

358) ( πj πk − πk πj ) = jk [πj .350.352) ˆ σ · π X + qV Φ ˆ σ · π Φ + qV X − 2mX . For this purpose we employ the identity (5.348. (10.361) . ∂k ] + [∂j . identifies X as the small component of the bi-spinor wave function which. 10. Ak ] . ˆ Equation (10. derived in Sect. Ak ] + q 2 [Aj .357) Equation (10. and. Ak ] i i i i =0 =0 = q q [Aj .334 and assume that for the time-derivative of Φ and X holds ∂t Φ << m .356). ˆ ˆ For the components of π × π holds ˆ ˆ π×π = jk (10.354) ˆ 0 ≈ σ · π Φ − 2m X .353) (10.353) by X ≈ to obtain a closed equation for Φ ˆ σ·π 2m 2 (10. One obtains [πj . accordingly. does not need to be considered anymore. ∂k + qAk ] i i 1 1 1 1 [ ∂j . ∂k ] + q [ ∂j .357) for Φ can be reformulated by expansion of (σ · π)2 . ∂k ] + [∂j . Φ Using (10. (10.7. πk ] = = 1 1 [ ∂j + qAj . 10.351) in (10.355) ˆ σ·π Φ 2m (10.354) The properties (10.352) allow one to approximate (10. Ak ] ) f = ( ∂j Ak − Ak ∂j + Aj ∂k − ∂k Aj ) f . i i (10.360) For an arbitrary function f (r) holds ( [Aj . 10. ∂k ] + q [Aj .230). which in the present case states ˆ ˆ2 ˆ ˆ (σ · π)2 = π + i σ · (π × π) . (10. π ] . due to the m−1 factor.346. one can replace X in (10.349) yields i∂t Φ i∂t X = = ∂t X X Relativistic Quantum Mechanics << m .356) i∂t Φ ≈ Φ + qV Φ . (10. (10.359) We want to evaluate the latter commutator. 5. henceforth.

πk ] = q q (( ∂j Ak − ∂k Aj )) = − i i ×A jk q = − B i jk (10. −A). 10. r (10.357) in the final form i∂t Φ(r. t) (10.2 around the magnetic field.344.366) (10. The respective bispinor wave function Ψ(xµ ) ∈ C4 is described as the stationary solution of the Dirac equation (10. the interaction qσ · B induces a 1 precession of the spin.360) and Aµ = (V.364) governing a two-dimensional wave function Φ ∈ C2 with the corresponding non-relativistic Schr¨dinger equation (10. t) ≈ ˆ [p − q A(r. [πj . t) [see (8.10. 5.2) governing a one-dimensional wave function ψ ∈ C. t)]2 q − σ · B(r.367) . 0) . ∂k ] + [∂j .6)].1 particle with a magnetic 2 field B.2) it introduces the o extra term qσ · B Φ which describes the well-known interaction of a spin.363) allow us to write (10. Equations (10. Comparision of (10.1 which emerged in the Lorentz-invariant theory as an algebraic 2 necessity. 0. 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 .1 with the magnetic field.365) Comparision of this expression with (5. i.362) or. t) + q V (r.363) where we employed B(r. The formal solution of this equation is Φ(t) = e−iqtB·σ Φ(0) .Spectrum We want to describe now the spectrum of a relativistic electron (q = −e) in the Coulomb field of a nucleus with charge Ze. 0. Dirac Particle in Coulomb Field . t) = × A(r. does not leave the theory again when one takes the non-relativistic limit.223) shows that the propagator in (10.366) can be interpreted as a rotation around the field B by an angle qtB.e. the spin. Ak ] ) f = [ ((∂j Ak )) − ((∂k Aj )) ] f (10.343) for the vector potential Aµ = (− Ze2 . t) 2m 2m Φ(r. using (10. but rather remains as a steady “guest” of non-relativistic physics with the proper interaction term. (10. o reveals a stunning feature: the Pauli equation does justice to its two-dimensional character.. In other words.359.364) which is referred to as the Pauli equation. 10.358. Let us consider briefly the consequences of the interaction of a spin.10: Dirac Particles in Electromagnetic Field Using ∂j Ak f = ((∂j Ak )) f + Ak ∂j f ∂k Aj f = ((∂k Aj )) f + Aj ∂k f 335 where ((∂j · · ·)) denotes confinement of the differential operator to within the brackets ((· · ·)). 10. while agreeing in all other respects with the non-relativistic Schr¨dinger equation (10.222.

γ ˆ ˜ ˆ (10. multiplying (10.370) resembles closely the Klein-Gordon equation (10.374) The first term on the r. Employing πµ as defined ˜ in Table 10.370) for which holds ˜ ˆ ˜ ˆ 3 γ πµ γ πν = ˜ ˆ ˜ ˆ µ=0 µ ν (˜ µ )2 πµ + γ ˆ2 µ.368) ˜ ( γ µ πµ − m ) Ψ(xµ ) = 0 . indeed. ˜ (10.368) ˜ where Ψ(xµ ) and γ µ are defined in (10.230). 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 .369). i.368) is also a solution of (10.370) Any solution of (10. 5. we solve ˜ [ i˜ µ (∂µ + iqAµ ) − m ] Ψ(xµ ) = 0 γ (10.231). 3.370). The difference arises from the term γ µ πµ γ ν πν in (10. This follows from [ i˜ µ (∂µ + iqAµ ) + m ] [ i˜ µ (∂µ + iqAµ ) − m ] γ γ = [ i˜ µ (∂µ + iqAµ ) − m ] [ i˜ µ (∂µ + iqAµ ) + m ] γ γ according to which follows from (10.180).336 Relativistic Quantum Mechanics For the purpose of the solution we assume the chiral representation. 2.371) such that we can conclude that (10.232) in Sect. 10.ν=1 µ =ν γ µ γ ν πµ πν . γ ˆ2 ˆ2 (10. (5. µ = ν and altering ‘dummy’ summation ˜ ˜ ˜ ˜ .7 one can write the second term in (10.370) ˜ [ i˜ µ (∂µ + iqAµ ) − m ] [ i˜ µ (∂µ + iqAµ ) + m ] Ψ(xµ ) = 0 γ γ (10. once a solution Ψ(xµ ) of (10. 1. respectively. ˜ However.s.373) (10. but the converse is not necessarily true.369) from the left by γ ν πν + m. ˜ ˜ ˆ ˆ (10.e. noting from (10.372) (10.374). is a solution of (10. 3 µ=0 ˆ2 (˜ µ )2 πµ = π0 − π . can be rewritten using. (˜ 0 )2 = 1 and (˜ j )2 = γ 1 γ −1 j = 1. For this purpose we ‘square’ the Dirac equation. according to (10. Equation (10. (5.369) For our solution we will adopt presently a strategy which follows closely that for the spectrum of pionic atoms in Sect.230) γ µ γ ν = − γ ν γ µ .369).371).h.229).1 one can write (10.228) and in (10. but differs from it in an essential way.370) is obtained then ˜ [ i˜ µ (∂µ + iqAµ ) + m ] Ψ(xµ ) γ is a solution of (10.6.375) Following the algebra that connected Eqs.

γ ν ] [ˆµ . γ j ] [ˆ0 . Altogether.367) and the definition (10.380) where f = f (r.229). πj ] π ˆ = = (−i∂t + qA0 .ν=1 µ =ν (10.367) holds qA0 = r ˆ Ze2 . −i∂j ] = − [ (∂t + iqA0 ) ∂j − ∂j (∂t + iqA0 ) ] f i ((∂j qA0 )) f (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. due to A = 0. ν = 1. γ j ] γ ˜ = = 2 0 1 1 1 0 1 σj 0 0 −σ j 0 σj −σ j 0 − 0 σj −σ j 0 0 1 1 1 0 1 (10.367) of Aµ .ν=1 µ =ν 3 j=1 3 3 = 1 4 1 2 1 [˜ . Since [ˆµ .380) γ µ γ ν πµ πν = i ˜ ˜ ˆ ˆ µ.e. γ ˜ π ˆ j=1 (10. t) is a suitable test function and where ((· · ·)) denotes the range to which the derivative is limited.344).377) which follows readily from the definition (10. γ ν ] [ˆµ .ν=1 µ =ν = [˜ µ . it holds 1 4 [˜ µ .376) This expression can be simplified due to the special form (10. πν ] γ ˜ π ˆ µ.ν=1 µ =ν σ 0 0 −σ · (( qA0 )) (10. γ 0 ] [ˆj .378) can be evaluated using (10. i. πj ] + γ ˜ π ˆ 4 0 j [˜ j . γ µ γ ν πµ πν = ˜ ˜ ˆ ˆ µ.10.379) The commutators [ˆ0 . the commutators [˜ 0 .382) . πν ] γ ˜ π ˆ µ.344) π ˆ [ˆ0 . πν ] = 0 π ˆ for µ. γ ] [ˆ0 . r2 (10.381) According to (10.10: Dirac Particles in Electromagnetic Field indices.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 ] are γ ˜ [˜ 0 . 3 (10. πj ] . 2.378) According to the definition (10. πj ] in (10. π0 ] γ ˜ π ˆ j=1 = [˜ 0 . one can summarize (10..376–10.

389) According to the results in Sect. is genuinely two-dimensional and.147. 6.148). couples the orbital angular momentum of the electron to its spin. −j + 1. .375) the ˆ ‘squared’ Dirac equation (10. . r r 2 2 2 j = 1.5 which describe the coupling of orbital angular momentum and spin 1 {( Yjm (j − 2 . We note that these states are also eigenstates of the angular r 2 . a solution of the form ∼ Y m (ˆ) can be obtained.374). it is this quantity that we want to determine. (10. The term iσ · r. Combining this result with (10. (10. . Yjm (j + r 2 1 1 .e.387) The expression (10. in r ˆ fact.385) 1 We split the wave function into two spin.384) into (10. hence. (10. . (10.383) We seek stationary solutions of this equation. . 1 |ˆ) ) . however.2 components ˜ Ψ(r) = and obtain for the separate components φ± (r) Ze2 + r 2 ˜ φ+ (r) ˜ φ− (r) (10. (10.1 .381). i. 1 |ˆ). In the latter case.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. 2 |ˆ)} as given in (6. we express 2 the solution of (10. Insertion of (10. (10. Such solutions are of the form ˜ ˜ Ψ(xµ ) = Φ(r) e−i t .386) + 2 ± iσ · r ˆ Ze2 − m2 r2 φ± (r) = 0 . Yjm (j + 1 .388) in terms of states introduced in Sect. 6.189) for the Laplacian and expansion of the term (· · ·)2 result in the twodimensional equation 2 ∂r − ˆ L2 − Z 2 e4 iσ · r Ze2 ˆ 2Ze2 + r2 r + 2 − m2 r φ± (r) = 0 . m values are coupled. 2 3 2 . 10.338 Relativistic Quantum Mechanics where r = r/|r| is a unit vector.5 the operator iσ · r is block-diagonal in this basis such that only ˆ 1 the states for identical j.. only the two states {Yjm (j − 2 . 6. + j } (10. Accordingly. m = −j.191) solved in Sect. .6. 1 |ˆ).384) can be interpreted as the energy of the stationary state and.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 .368) reads − Ze2 ∂t − i r 2 + 2 + i σ·r ˆ 0 0 −σ · r ˆ Ze2 − m2 r2 Ψ(xµ ) = 0 (10.

we want to determine solely the spectrum. such transformation can be chosen as to diagonalize the second term. hence.2 (j). namely. We select.199). as in the case of pionic atoms. 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. Since. .2 (j) being 2 j = 1 . the missing additive term − 1 . However. (6.396) and (10.2 (j) + 1) 2 Ze2 + + r2 r 2 − m2 f1.200). . 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 . .395) (10. 6. involving the 2 × 2 unit matrix.10: Dirac Particles in Electromagnetic Field 339 ˆ momentum operator L2 [cf.151). 1 |ˆ) r 2 2 2 2 r r . 3 .10. unaltered.393) but do not explicitly consider further the wavefunctions. in the present treatment. namely. the values of the argument of λ1. . which states that the states Yjm (j ± 1 . 1 |ˆ) r 2 (10. 10. 1.392) into diagonal form.2 (r) = 0 (10.397) This equation is identical to the Klein-Gordon equation for pionic atoms written in the form (10. a specific pair of total spin-orbital angular momentum quantum numbers j.151].2 (j) as given by (10.392). 1 |ˆ) + r Yjm (j + 1 .392) h± (r) g± (r) = 0. (6. (10.390) (10.f. therefore. λ1 (j) and λ2 (j). rather than = 0. and except for the fact that 2 2 we have two sets of values for λ1. . not the wave functions. 1 |ˆ) = − Yjm (j ˆ r 2 2 1 2 h± (r) g± (r) Yjm (j − 1 . Obviously.5 [c. Any similarity transformation leaves the first term in (10.395.391) derived in Sect. We can. − 1 r2 (j − 1 )(j + 1 ) − Z 2 e4 2 2 ±i Ze2 (j + )(j + ) − Z 2 e4 We seek to bring (10. 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 (6.392) reads then in the diagonal representation 2 ∂r − λ1. except for the slight difference in the expression of λ1. 1 |ˆ) are r 2 2 ˆ eigenfunctions of L2 .. .396) and λ2 (j) [λ2 (j) + 1] (10.186)].394) Equation (10. conclude that the . the eigenvalues are independent of m.2 (j)[λ1. m and expand φ± (r) = Using σ · r Yjm (j ± 1 .

. and spin-orbital angular momentum j = − 1 . 2 2 1 ≈ m − (10. . (10. . . . − + 1.400) (10. −j + 1. . . . . .401) These expressions can be equated with the non-relativistic spectrum. . m = − .396). 1.340 Relativistic Quantum Mechanics spectrum of (10. . 10. . −j + 1. j = − 2 .395) and 2 corresponds to λ2 (j) as given in (10. . . j where 1 corresponds to λ1 (j) as given in (10.398. 1 ms = ± 2 (10. j = 1 . One obtains in case of (10. . the second term on the r. = 1 + m Z 2 e4 √ ( n − − 1 + 2 − Z 2 e4 )2 . . 2 2 m = −j.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. 2n2 = 0.399) n = 0. 2. . . . . (10. 2. .396) we obtain.403) √ Z 2 e4 2 ED (n. albeit with some modifications.398) 2 = 1 + m . . n − 1 .397) is again given by eq. . 3. . The magnitude of the relativistic effect 2 2 e4 . . m = −j. j .400) and (10. accordingly. These considerations 2 are summarized in the following equations 1 ED (n. 2. .2 . Obviously. n − 1 . ( +1) − Z 2 e4 )2 (10. 3 . . . One may also state 2 this in the following way: (10. . In case of non-relativistic hydrogen-type 1 atoms. It is given by n = n + + 1 where is the orbital angular momentum quantum number and n = 0.402) In this expression n is the so-called main quantum number. Using (10.404) = 0.401) corresponds to a non-relativistic state 2 with quantum numbers n. .s. . . −j + 1. . 1. j .402) with (10. the stationary states have binding energies E = − mZ 2 e4 n = 1.395. . 10. .h. . (10. . of these equations describe the binding energy. 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. 1. and spin-orbital angular momentum j = + 1 . (10. . . m) 2 n = 1. . . (10. counts the nodes of the wave function. One can equate (10. . .400) corresponds to a non-relativistic state with quantum numbers n. 1. .203). 2. . j = + 1 . including spin.401) if one attributes to the respective 1 states the angular momentum quantum numbers = j + 2 and = j − 1 . m) = 1 + m (n− + . 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. 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 . m = −j. . . . . . 1.

10. the nonrelativistic degeneracy for the six 2p states of the hydrogen atom: in the present. 1 |ˆ) for m = ± 1 . j 1 2 1 2 1 2 3 2 1 2 1 2 3 2 3 2 5 2 341 rel. j.2 the non-relativistic [cf. . 2 |ˆ) for m = ± 1 . (10. binding energy / eV Eq.1. 1. However. .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 10.402. . i. r 2 2 2 . relate closely to the corresponding nonrelativistic states. One can combine the expressions (10. in fact.402) -13.405) in terms of the non-relativistic quantum numbers n. relativistic. 10. for example. (10. 2. . n − 1 j = 1 2 1 2 ± m = −j.405) In order to demonstrate relativistic effects in the spectrum of the hydrogen atom we compare in Table 10. the 2p 3 states with r 2 2 2 2 1 involve four degenerate states corresponding to Y 3 m (1.10. mom. . j (10. m) = m 1 + Z 2 e4 (n−j −1 + 2 (j+ 1 )2 − Z 2 e4 )2 2 n = 1.e.3.40146 ⇑ ⇑ -1. . notation 1s 1 2 non-rel.551177 ⇑ . i. (10. . ± 3 .551178 ⇑ .3.e. The table entries demonstrate that the energies as given by the expression (10. The energies were evaluated with m = 511. Of particular interest is the effect of spin-orbit coupling which removes.405).2: Binding energies for the hydrogen (Z = 1) atom. (10. the non-relativistic and relativistic energies are hardly discernible. binding energy / eV Eq.40151 ⇑ .. 0 0 1 1 0 1 1 2 2 spinorbital ang. . = 0.40147 -1.403.60583 -3. much less than the rest mass of the electron (511 keV). . Degeneracies are denoted by ⇑. case these six states are split into energetically different 2p 1 and 2p 3 states.60601 -3. is in the range of 10 eV.. .036 by means of Eqs. 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.404) finally into the single formula ED (n.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 (1.405)] spectrum of the hydrogen atom.551176 spectr.0041 keV and e2 = 1/137. (10.10: Dirac Particles in Electromagnetic Field main quantum number n 1 2 2 2 3 3 3 3 3 orbital angular mom. This is clearly demonstrated by the comparision of non-relativistic and relativistic spectra of a hydrogen-type atom with Z = 100 in Table 10. . .1. −j + 1.405 ) -13. The reason is that the mean kinetic energy of the electron in the hydrogen atom.

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). j.0 ⇑ ⇑ -15. (10. In order to investigate further the deviation between relativistic and non-relativistic spectra of hydrogen-type atoms we expand the expression (10. 32 (10. (10.402. 10. binding energy / keV Eq.15.405). 1 2 2 2 2 E (2p 3 ) − E (2p 1 ) ≈ − 2 2 mZ 4 e8 2 · 23 1 −1 2 = mZ 4 e8 .405) reads 1 (10. binding energy / keV Eq.1 ⇑ . . Degeneracies are denoted by ⇑. 0.036 by means of Eqs. 0) (10.407) 1 2 This expression allows one. (10. m) ≈ m − mZ 2 e4 mZ 4 e8 − 2n2 2n3 1 j+ − 3 4n + O(Z 6 e12 ) .409) where V (r) is spherically symmetric.406) α √ 1+ (n − β + β 2 −α)2 The expansion (10. We assume for the wave function the stationary state . (10.402) -136. to estimate the difference between the energies of the states 2p 3 and 2p 1 (cf. It holds for n = 2 and j = 3 .405 ) -161.9 ⇑ .1 ⇑ ⇑ ⇑ ⇑ rel. Introducing α = Z 2 e4 1 and β = j + 2 (10.3 spectr. mom.8 ⇑ .0041 keV and e2 = 1/137.342 main quantum number n 1 2 2 2 3 3 3 3 3 orbital angular mom. for example. 0 0 1 1 0 1 1 2 2 spinorbital ang. An example for such potential is the Coulomb potential V (r) = −Ze2 /r considered further below.1 -34.3). 0. j 1 2 1 2 1 2 3 2 1 2 1 2 3 2 3 2 5 2 Relativistic Quantum Mechanics non-rel.35. Tables 10.6 -42.2 -17.2. 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 10.3: Binding energies for the hydrogen-type (Z = 100) atom.10.15. The energies were evaluated with m = 511.206) provides in the present case ED (n.405) to order O(Z 4 e8 ).

413) ˆ σ · p g(r) Yjm (j − 1 . 2 |ˆ)  r Φ(r)  r  .410) 1 where Φ(r). 6. σ · p ˆ σ · p f (r) Yjm (j + 1 .415) 1 ˆ σ · p rg(r) Yjm (j − 2 . 2 |ˆ) r r r . 2 |ˆ) + r Yjm (j − 2 .412) are four-dimensional with r-dependent wave functions.197.411) (10. pp. The arguments above allow one to eliminate the angular dependence by expanding Φ(r) and X (r) 1 in terms of Yjm (j + 2 .413.414) ˆ σ · p rf (r) Yjm (j + 1 . 2 |ˆ) and does r 2 ˆ has odd parity and it holds [c.e.e. 168]: the term is a scalar (rank zero tensor) in the space of the spin-angular momentum states Yjm (j ± 1 . 1 |ˆ) r 2 2 = i j+ ∂r + r j+ r 1 2 1 r f (r) Yjm (j − 2 . (10.10. 6..417)  c(r)  d(r) X (r) 1 1 1 1 Yjm (j + 2 . X (r) ∈ C2 describe the spatial and spin. 1 |ˆ). 6. in particular. 1 |ˆ) r 2 2 = i ∂r + −j r 1 g(r) Yjm (j + 2 .412) ˆ In this equation a coupling between the wave functions Φ(r) and X (r) arises due to the term σ · p.2 degrees of freedom. 2 |ˆ) + r Yjm (j − 2 .   =  r (10. 10. 10. 1 |ˆ) . r 2 (10. i. The Dirac equation reads then.345). 1 |ˆ) r 2 2 (10.416) The differential equations (10. 1 |ˆ) and Yjm (j − 1 . 1 |ˆ) r 2 1 1 = ∂r r r r (10. 1 |ˆ) .411.5. 1 |ˆ) r 2 = i ∂r − 1 2 r g(r) Yjm (j + 1 . m-values.10: Dirac Particles in Electromagnetic Field form Ψ(xµ ) = e−i t 343 Φ(r) X (r) . (6.232. This term has been discussed in detail in Sect. i.f.5 [see. 1 |ˆ) introduced in r 2 2 1 Sect. r r 2 2 2     a(r) b(r) 1 1 1 1 Yjm (j + 2 . the term is block-diagonal in the space spanned by the states Yjm (j ± 1 .414) These equations can be brought into a more symmetric form using ∂r + which allows one to write (10.. ˆ σ · p X + m Φ + V (r) Φ ˆ σ · p Φ − mX + V (r) X = = Φ X (10. 1 |ˆ) r 2 2 = i ∂r + j+ r 1 2 3 2 f (r) Yjm (j − 1 . according to (10. 10.198)] not couple states with different j. r 2 2 (10. but are timeindependent.

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. 2 |ˆ) r r where the factors i and −1 have been introduced for convenience.157).419) Obviously.411. m values.418) holds for f1 (r). Presently.412). 1 |ˆ)  r Φ(r) 2 2  r    =  (10. g1 (r) ∂r − j+ r j+ r 1 2 g1 (r) + [ − m − V (r) ] f1 (r) = 0 ∂r + and for f2 (r). g2 (r) ∂r + 1 2 f1 (r) − [ + m − V (r) ] g1 (r) = 0 (10.421)   g2 (r) X (r) 1 1 − Yjm (j + 2 . Inserting (10. Accordingly. 2 |ˆ)   =  (10. 2 |ˆ) r r   f2 (r) 1 1 r Φ(r)  i r Yjm (j − 2 .415. m pair contribute.417) into (10.423) .344 Relativistic Quantum Mechanics In general. only a(r). 10. According to (10. the orthonormality property (6. we consider the case that only states for one specific j.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. (10. d(r) are coupled and b(r). 10. using (10. c(r) are coupled.420)   g1 (r) X (r) 1 1 − Yjm (j − 2 .416). such expansion must include states with all possible j. 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 (10.417) of the form     f1 (r) i Yjm (j + 1 . there exist two independent solutions (10.

one demonstrates first that the wave function at r → 0 behaves as rγ for some suitable γ. 10.420). 2 |ˆ)  r Ψ(xµ ) ≈  .422. (10. 3p 1 .423). 2. can be written ∂r − j+ r j+ r 1 2 g1 (r) + Ze2 f1 (r) r Ze2 g1 (r) r = 0 ∂r + 1 2 f1 (r) − = 0.e. and obtains finally a polynomial function p(r) such that rγ exp(−µr)p(r) solves (10. We note that (10. Hence.405). . in the non-relativistic limit wave functions   f2 (r) 1 1 i Yjm (j − 2 . 1 |ˆ)  r i 2 2 . According to the discussion of the spectrum (10. . Φ in (10.420) is the large component and X is the small component. together with the appropriate boundary conditions at r = 0 and r → ∞. 2 2 2 2 2 2 = j − 1 2 and. .367) which is spherically symmetric such that equations (10.423) are identical.422). (10. n − 1. the 2 equations determine. = 0.424) Ψ(xµ ) ≈  r 0 1 i. etc.423) apply for V (r) = −Ze2 /r. 2 2 2 (10. (10. 3d 3 . 2p 3 . one demonstrates then that the wave functions for r → ∞ behaves as exp(−µr) for some suitable µ. correspodingly the states We consider first the solution of (10. etc. Equation (10.e. Similarly.10: Dirac Particles in Electromagnetic Field 345 Equations (10. .421).409).422) and (10. (10. . (10.422). determining wave functions of the type (10.422) determines solutions of the form (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.. In the non-relativistic limit. 1. for small r. except for the opposite sign of the term (j + 1 ). namely.. 2s 1 .425) r 0 covers states with angular momentum 1s 1 . 3s 1 . i.10. . Behaviour at r → 0 We consider first the behaviour of the solutions f1 (r) and g1 (r) of (10. The solution of (10. Dirac Particle in Coulomb Field . . . 3p 3 . Hence. According to this procedure.422) near r = 0. the ones given in (10.405) of the relativistic hydrogen atom the corresponding states have quantum numbers n = 1.Wave Functions We want to determine now the wave functions of the stationary states of hydrogen-type atoms which correspond to the energy levels (10.422).422) describes the states 2p 1 .405). We assume the 4-vector potential of pure Coulomb type (10.422) corresponds to states   f1 (r) Yjm (j + 1 . the radial wave functions for Dirac particles in the potential (10.426) . to states with angular momentum = j + 2 . enforcing the polynomial to be of finite order leads to discrete eigenvalues . 3d 5 .

427) = = 0 0. The assumed r-dependence in (10. g1 ∼ exp(± m2 − 2 r). 10.266. Such r-dependence 2 would make the expectation value of the potential r2 dr ρ(r) 1 r (10.432) = = − ( − m.267. (10.431) infinite since.429) This equation poses an eigenvalue problem (eigenvalue −γ) for proper γ values. µ = m2 − 2 (10.427). r2 (10. assume the r-dependence in (10. Only the exponentially decaying solution is admissable and.430) 2 Note that the exponent in (10.433) = = ( m2 − 2 ) g1 (r) ( m2 − 2 ) f1 (r) (10.428) = 0. One obtains γ = ± (j + 1 )2 − Z 2 e4 .346 Setting f1 (r) yields γ b rγ−1 − (j + 1 ) b rγ−1 + Ze2 a rγ−1 2 γ a rγ−1 + (j + 1 ) a rγ−1 − Ze2 b rγ−1 2 or γ + (j + 1 ) −Ze2 2 2 Ze γ − (j + 1 ) 2 a b →0 Relativistic Quantum Mechanics ∼ a rγ . hence. (10.427) makes only the positive solution 2 possible.422) becomes ∂r g1 (r) ∂r f1 (r) Iterating this equation once yields 2 ∂r g1 (r) 2 ∂r f1 (r) 1 . We have. we conclude f1 (r) →∞ ∼ e−µr . (10. ) f1 (r) ( + m ) g1 (r) (10. 10. hence. in case j + 1 )2 < Ze2 . g1 (r) →∞ ∼ e−µr . according to (10.434) √ The solutions of these equations are f1 .420). g1 (r) →0 ∼ b rγ (10. for small r. for the particle density holds then ρ(r) ∼ |rγ−1 |2 = Behaviour at r → ∞ For very large r values (10. becomes imaginary.435) . determined that the solutions f1 (r) and g1 (r).427) with 1 γ = (j + )2 − Z 2 e4 .

We also introduce the new variable ρ = 2µ r .435 ). Solution of the Radial Dirac Equation for a Coulomb Potential To solve (10.437) (10.10: Dirac Particles in Electromagnetic Field For bound states holds 347 < m and.440) and (10. without loss of generality we can choose ˜ f1 (r) = φ1 (r) + φ2 (r) . (10.445) .436) f1 (r) = Insertion into (10.440) (10.s.438) (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. However. hence.434) √ m + a e−µ r . we set f1 (r) g1 (r) = = √ √ ˜ m + e−µr f1 (r) −µr = = 0 0 (10.h. g1 (r) = φ1 (r) − φ2 (r) ˜ (10.441) The last two terms on the l. √ g1 (r) = − m − a e−µr .444) (10. Let us consider then for the solution of (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 . Equation (10.439) − m− e g1 (r) ˜ where µ is given in (10. (10.442) which leads to a partial cancellation of the asymptotically dominant terms.10. of both (10.434) results in √ √ (m − ) m + a − (m − ) m + a √ √ (m + ) m − a − (m + ) m − a which is obviously correct. µ is real. 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.422 ) leads to √ √ j+1 Ze2 ˜ 2 − m − [∂r − ] g1 + m + ˜ f1 + r √r √ ˜ (m − ) m + g1 − (m − ) m + f1 = 0 ˜ 1 2 √ √ j+2 ˜ Ze m + [∂r + ] f1 + m − g1 − ˜ r r √ √ ˜ (m + ) m − f1 + (m + ) m − g1 = 0 ˜ (10.437) where they ˜ ˜ cancelled in case f1 = g1 = a.436).441) correspond to (10. Accordingly. ρ m+ ρ [∂ρ − (10.

f. 10.446) (10.453) (s + γ)2 + Z 2 e4 − (j + 1 )2 2 2 √ Ze m2 − 2 βs .449) for γ given in (10. 10.348 Relativistic Quantum Mechanics Addition and subtraction of these equations leads finally to the following two coupled differential equations for φ1 and φ2 ∂ρ φ1 + ∂ρ φ2 + j+ ρ 1 2 Ze2 mZe2 φ1 − √ φ2 = 0 m2 − 2 ρ m2 − 2 ρ mZe2 Ze2 φ1 + √ φ1 + √ φ2 − φ2 = 0 m2 − 2 ρ m2 − 2 ρ j+ ρ φ2 − √ 1 2 (10.454) .447) of the form n φ1 (ρ) = ργ s=0 n γ s=0 αs ρ2 βs ρ2 (10.450) follows Ze2 m2 − 2 βs − βs−1 =0 (10. (10. (10.448. = m− Ze2 βs s + γ − √m2 − 2 2 (10.452) From (10.430) one can write this βs = s+γ − 2 √ Ze m2 − 2 s (s + 2γ) βs−1 .448) φ2 (ρ) = ρ (10.446. 10.451) follows βs−1 = = Ze2 s+γ + √ m2 − s+γ − βs + m2 Z 2 e4 m2 − 2 − (j + 1 )2 2 2 √ Ze m2 − 2 2 s+γ − βs (10. Using (10.447) We seek solutions of (10.446 .430)]. Inserting (10.451) 1 √mZe − (j + 2 ) 2 2 αs .449) into (10.426–10.430) which conform to the proper r → 0 behaviour determined above [c.447) leads to (s + γ) αs + (j + 1 ) βs − √ 2 s Ze2 m2 − 2 αs (10.450) −√ mZe2 m2 − 2 βs ρs+γ−1 = 0 mZe2 m2 − αs (s + γ) β2 + (j + 1 ) αs + √ 2 s 2 +√ From (10.

α + 1.456.461) In order that the wave functions remain normalizable the power series (10.461 for values given by (10.457) One can relate the polynomials φ1 (ρ) and φ2 (ρ) defined through (10... ρ) so β0 ργ F (1 − so . 10. in fact.10: Dirac Particles in Electromagnetic Field Defining so = √ one obtains βs = = . . β0 j+ 1 2 (10. According to the definitions (10. is an integer. x) .452) follows j+ αs = 1 2 349 Ze2 m2 − 2 −γ (10. so = n . (s − so ) β0 s! (2γ + 1)(2γ + 2) · · · (2γ + s) (10.455) are finite. 1.459) − 2 √mZe m2 − 2 (10. 10. . with the associated Laguerre polynomials L(α) = F (−n. . ρ) . = From (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 ) . Comparision with (10. n It holds φ1 (ρ) φ2 (ρ) = = ργ F (−so .e.456) − so 2 √mZe m2 − 2 (−so )(1 − so )(2 − so ) . x) = 1 + a a(a + 1) x2 x + + . 10. namely. c. .458) or.460) (10. 2. for the states 2p 1 .405) and the ensuing so values ( 10. 3d 3 holds n = 1.449) must be of finite order. 10. 2γ + 1. We can.448. .455) this confinement of so implies discrete values for . .449) and (10. (s − so ) β0 s! (2γ + 1)(2γ + 2) · · · (2γ + s) (10. equivalently. . . . The expressions (10.430. ..457) with the confluent hypergeometric functions F (a. 2. 2γ + 1. For example. 3p 1 .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. conclude 2 2 2 that the polynomials in (10. This requires that all coefficients αs and βs must vanish for s ≥ n for some n ∈ N. (10. 1.405) allows one to identify n = n − j + 2 which.405). hence.456) and (10.10. i. c c(c + 1) 2! (10. m (n ) = . n = 0.448.457) for βs and αs imply that so must then be an integer.

(10. can follow the procedure adopted for the wave functions of the non-relativistic hydrogen atom and will not be carried out here. 9. = j + 2 .421). 10.421) correspond to non-relativistic states with orbital angular momentum 1 1 = j + 2 . are two-dimensional vectors determined through the explicit form of the spin-orbital angular momentum states Yjm (j ± 1 . is time-independent.467) ± n F (1 − n . 10. 2µr) (2µ) 2 Γ(2γ + 1) 3 N = and m 4m )Γ(2γ + n + 1) (n +γ)m (n +γ)m (10. Sect.463) where Φ and X .469) This formula has been adapted from ”Relativistic Quantum Mechanics” by W. Greiner.438.148).m (j + 1 .350 Relativistic Quantum Mechanics Altogether we have determined the stationary states of the type (10.461) are to be chosen to satisfy a normalization condition and to assign an overall phase. 1 |ˆ) 2 2 r κ = j+ 1 2 (10. 6. 2µr) (10. m|xµ ) = e−i t iF1 (r) Yj.158) of the latter states absorb the angular integral in (10. The complete wave function is given by the following set of formulas 1 Ψ(n.147. The coefficients β0 in (10. The wave functions (10. 1 |ˆ) 2 2 r G1 (r) Yj. m.461). Berlin. 2 . 10. g1 (r) determined by (10. as in (10.464) 0 The evaluation of the integrals. j.461).465) (10.442). = j + 2 . (10.m (j − 1 . j.460.439).421) with radial wave functions f1 (r).421)] ∞ dr (|f1 (r)|2 + |g1 (r)|2 ) = 1 (10. 2γ + 1. which involve the confluent hypergeometric functions in (10. and (10.157. (Springer.267). The normalization integral is then ∞ 0 π 2π r2 dr 0 sin θdθ 0 dφ (|Φ(r)|2 + |X (r)|2 ) = 1 (10.6.468) −κ n! µ γ n = = = = (m − )(m + ) (j + 1 )2 − Z 2 e4 2 n−j − m 1+ 1 2 Z 2 e4 (n +γ)2 . 10. They are described through quantum numbers n. 2γ + 1. Due to the form (10. where2 F± (κ|r) = G1 (r) = F+ (κ|r) .460. The orthonormality 2 2 r properties (6.463) and yield [note the 1/r factor in (10. given by expression (10.466) F1 (r) = F− (κ|r) . 6. 1 |ˆ) in (6.410) of the stationary state wave function the density ρ(xµ ) of the states under consideration. (n + γ)m N (2µr)γ−1 e−µr − κ F (−n . 1990).460.

m|x ) = e −i t iF2 (r) Yj. become   f2 (r) 1 1 i Yjm (j − 2 . j. describes the complementary set of states 1s 1 .10: Dirac Particles in Electromagnetic Field 351 We want to consider now the stationary states of the type (10. (10.471) (10. We have. 2s 1 .472) F2 (r) = F− (κ|r) . One can verify. where F± (κ|r) are as given in (10. 2 |ˆ)  r Ψ(xµ ) ≈ e−1 t  . not not 2 2 2 2 2 2 covered by the wave functions given by (10.m (j + 1 .469). 2 accordingly.469) that the following wave functions result Ψ(n. tracing all steps which lead from (10.465–10. 3s 1 .m (j − 1 . The radial wave functions f2 (r) and g2 (r) in (10. 2p 3 . = j − µ 1 2 . 1 |ˆ) 2 2 r G2 (r) Yj. 3p 3 .467–10. in the non-relativistic limit. G2 (r) = F+ (κ|r) .470) r 0 Obviously. this wavefunction has an orbital angular momentum quantum number = j − 1 and.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. . 3d 5 . hence.422) to (10.421) are governed by the radial Dirac equation (10. 1 |ˆ) 2 2 r κ = −j − 1 2 (10.421) which. obtained closed expressions for the wave functions of all the stationary bound states of relativistic hydrogen-type atoms.469). etc.10.

352 Relativistic Quantum Mechanics .

must be two-dimensional irreducible representations of the Lorentz group..1) (11. The lower dimensionality of a(z) and b(z) implies. different L(w.e. expressing its dependence on w. isomorphic to the natural representation.2) (11.225–10. a(z) and b(z). what will be achieved in the following.229). z ∈ C ˜ Because of its block-diagonal form each of the diagonal components of S(z). 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. 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.. ϑ)µ ν correspond to different a(z) and b(z). in particular. ˜ We have altered here slightly our notation of S(w. ϑ) for the ˜ in the chiral representation as given by (10. 1 353 . i.e. ϑ through 3. This is. in a sense. in fact. a complex variable z. i.4) = exp = w − iϑ . indeed. may be exploited to express the Lorentz-invariant equations of relativistic quantum mechanics. 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.Chapter 11 Spinor Formulation of Relativistic Quantum Mechanics 11. We will proceed by building as much as possible on the results obtained sofar in the chiral representation of the Dirac equation. that the corresponding representation of the Lorentz group is more basic than the natural representation and may serve as a building block for all representations. Starting point is the Lorentz transformation S(w.3) (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.262). ϑ).

ϑ)µ ν and a(z). ϑ ∈ R3 . + 1 2 2 state 2 ∼ | 1 .243). 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 . β. namely. express 4-vectors Aµ . a space of vectors state1 for which holds state2 state 1 ∼ | 1 . to 1 ∗ ∗ z ·σ (11. a distinct difference in the transformation behaviour of ˜ ˜ ˜ ˜ Ψ1 (xµ ). Here | 1 . finally. Ψ2 (xµ ) and Ψ3 (xµ ).5) The transformations in this case. the so-called spinor space. γ (see Chapter 5). i.6) as given by (5. are elements of SU(2) and correspond. (11.e. Ψ4 (xµ ) in case z = w + iϑ for w = 0. +1  2 2 |1. In this case holds a(z) = b(z) ˜ 1 (xµ ).. 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 (11. for z = iϑ. to the rotational transformations of spin. For ϑ = (0. Ψ2 (xµ ) transform according to a(z) whereas Ψ3 (xµ ). Aν . and the Klein–Gordon equation in the spinor representation. A First Characterization of the Bispinor States We note that in case w = 0 the Dirac transformations are pure rotations. −1 2 2 z = iϑ. usually expressed 2 as product of rotations and of functions of Euler angles α. This characterization allows one to draw conclusions regarding the state space in which a(z) and b(z) operate. (11.. b(z).9) a∗ (z) = exp 2 . θ ∈ R3 . Ψ4 (xµ ) transform according to ˜ ˜ ˜ and Ψ b(z). however. 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. − 1 2 2 z = iϑ. −1  2 2  |1. (11. Relationship Between a(z) and b(z) The transformation b(z) can be related to the conjugate complex of the transformation a(z). ϑ ∈ R3 . β.8) We like to stress that there exists. +1 2 |1.7) where “∼” stands for “transforms like”.354 Spinor Formulation and b(z) act. the operator ∂µ and the Pauli matrices σ in the new representation and.e. 2 2 2 ˜ Since a(z) acts on the first two components of the solution Ψ of the Dirac equation. in 2 fact. ± 1 represents the familiar spin– 1 states.1 states as described by Dm m (ϑ). neutrino equation. formulate the Dirac equation. i. will establish the map between L(w.

µ ). 11. one can show −1 −1 −1 = −σ1 = σ2 = −σ3 ∗ = −σ1 ∗ = −σ2 σ2 −1 (11.15) (11. ∗ σ 3 = σ3 . but leave rotation angles ϑ unaltered. 355 (11. Spatial Inversion ˜ One may question from the form of S(z) why the Dirac equation needs to be four-dimensional. −1 given by (11.12) yields σ1 σ3 or in short σ Similarly.11) one first demonstrates that for and −1 as given in (11.12).11) 0 −1 1 0 = . 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. σ2 .224. Ψ (xµ ) even though these pairs of ˜ ˜3 ˜4 featuring the components Ψ1 (x ˜ 2 components transform independently of each other. a∗ (z) −1 = exp 1 ∗ z σ∗ 2 −1 . (11. Ψ (xµ ) as well as Ψ (xµ ).10.17) We conclude.18) with a(z) given by (11. (11. hold −1 = 1 One notices then. Hence.4) and . The effect of inversion on Lorentz transformations is. . This demonstrates that a(z) is the ˜ transformation which characterizes both components of S(z). that the transformation (11. σ3 ).10) From this one can derive b(z) = where = 0 1 −1 0 . (11. therefore. −1 = ∗ −σ3 = −σ ∗ . 11.224)] ∗ σ1 = σ1 . 1.2. using f (a) −1 = f ( a −1 ). however. (11.1: Lorentz Transformation of Dirac Bispinor ∗ ∗ ∗ where σ ∗ = (σ1 .16) σ = −σ ∗ .11. one can express a∗ (z) −1 = exp 1 − z ∗· σ 2 = b(z) .f. One can readily verify [c.12) does. that they alter w into −w. ∗ σ 2 = − σ2 .13) Explicit matrix multiplication using (5. a result to be used further below. (5. −1 a∗ (z) −1 (11.1) can be written in the form ˜ S(z) = a(z) 0 0 a∗ (z) −1 (11. in fact.14) .12) To prove (11.

24) = etr(M ) (11. Part (Elsevier. P −1 = P. Amsterdam. One can verify this by evaluating the determinant of a(z) det( a(z) ) = det e 2 z·σ 1 (11. Obviously. de Kerf Lie a Algebras. the transformations a(z) and b(z) become interchanged. C) defined in (11. 1990).1) We have pointed out that a(iϑ). (5.  (11. However. 2 = 1 i. (11. Exercise 1. In fact.224)] tr( σj ) = 0 . j = 1.2 Relationship Between the Lie Groups SL(2. hence.1: Show that SL(2. 2.10. a(w + iϑ) for w = 0 is an element of SL(2.22) which follows from the fact that for any complex.23) Exercise 11.G. 3 .f.A.21) = etr( 2 z·σ) = 1 1 (11. The transformation S(z) in the transformed space is then ˜ P 1. This implies  ˜ Ψ1 ˜  Ψ2 P  ˜  Ψ3 ˜ Ψ4  ˜ Ψ3 ˜   Ψ4  =    Ψ1 ˜ ˜ Ψ2     . 11. for particles like the neutrinos which do not obey inversion symmetry two components of the wave function are sufficient.e. B¨uerle and E. which describes pure rotations.C) and SO(3. based on the Jordan–Chevalley theorem.19) i. the space spanned by only two of the components of Ψ is not invariant under spatial inversion and.356 Spinor Formulation ˜ Let P denote the representation of spatial inversion in the space of the wave functions Ψ. C) = { M. M is a complex 2 × 2–matrix. can be found in G.2. For the general case the proof.. 2 . ϑ ∈ R3 . non-singular matrix M holds2 det eM and from [c. (11.20) ˜ Obviously. the Lorentz invariant equation for neutrinos is only 2–dimensional. does not suffice for particles like the electron which obey inversion symmetry. det(M ) = 1 } .21) together with matrix multiplication as the binary operation forms a group. The proof of this important property is straightforward in case of hermitian M (see Chapter 5).e.. However. is an element of SU(2). ˜ ˜ P S(w + iϑ) P = S(−w + iϑ) = b(z) 0 0 a(z) .A.3.

75) holds aν = Lν µ aµ one realizes that σµ transforms inversely to covariant 4–vectors.30).27) −→ Straightforward transformation into another frame of reference would replace σµ by σµ . one wishes that the definition (11.2: Show that σ0 . Demonstrate that (11. 0 −i i 0 . Argue why M (Aµ ) = σµ Aµ provides a bijective map. (11. i. The following important property holds for M (Aµ ) det ( M (Aµ ) ) = Aµ Aµ which follows directly from (11. the matrix M (Aµ ) is hermitian as can be seen from inspection of (11.2. We will prove below [cf.28) and (11. (11.26) (11. σ3 provide a linear–independent basis for the space of hermitian 2 × 2–matrices. 0 1 1 0 . (11.25) σ0 σ1 σ2 σ3 The quantity σµ thus defined does not transform like a covariant 4–vector. M (Aµ ) according to (11. σ3 are hermitian.135)] this transformation behaviour.1) Mapping Aµ onto matrices M (Aµ ) 357 We want to establish now the relationship between SL(2. −→ µ Consistency of (11.25) of the matrix M (Aµ ) is independent of the frame of reference. C) and the group L↑ of proper.29) where we used (10.. or+ thochronous Lorentz transformations.28)   σµ =   1 0 0 1 . (11. Noting that for covariant vectors according to (10.25) can also be written M (Aµ ) = A0 + A3 A1 − iA2 A1 + iA2 A0 − A3 . in fact.31) Exercise 11. The function M (Aµ ) is bijective.11. σ1 . In fact. σ2 . σ2 . in a transformed frame should hold Lρ ν σ µ Aµ σµ A µ . (11.e.C) and SO(3.32) .30) Since the components of Aµ are real.76). Starting point is a bijective map between R4 and the set of two-dimensional hermitian matrices defined through M (Aµ ) = σµ Aµ where   (11.27) requires then σ µ Aµ Lν µ σµ = σν (11.31) holds.26) or from the fact that the matrices σ0 . 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 . 1 0 0 −1    . σ1 .  (11.2: Lie Groups SL(2. one can provide a simple expression for the inverse of M (Aµ ) M = M (Aµ ) ↔ Aµ = 1 tr M σµ 2 .

31).3: Show that L(a)µ ν defined in (11. 11.33) conserves the determinant of M . (11.37) using (11. Accordingly. any a ∈ SL(2. a ∈ SL(2.31)] 1 a Aµ −→ A µ = tr a M (Aµ ) a† σµ 2 Because of (11.358 Transforming the matrices M (Aµ ) Spinor Formulation We define now a transformation of the matrix M (Aµ ) in the space of hermitian 2 × 2–matrices a M −→ M = a M a† .40) is an element of L↑ .37) which implies that (11.33) to M (Aµ ) describing the action of the transformation in terms of transformations of Aµ .f.35) We now apply the transformation (11. C) defines the transformation [c. (11.40) 1 tr a σν a† σµ 2 Aν (11. L(a)µ ν = 1 tr a σν a† σµ 2 . C) .36) The suitable A µ can readily be constructed using (11.34) where we used the properties M † = M and (a† )† = a.35) holds for this transformation A µ Aµ = Aµ Aµ (11.37) defines actually a Lorentz transformation. (11. (11.2. it holds for the matrix M defined through (11.25) Aµ = which allows us to express finally A µ = L(a)µ ν Aν . The linear character of the transformation becomes apparent expressing A µ as given in (11.33) det(M ) = = det( a M a† ) = det(a) det(a† ) det(M ) [det(a)]2 det(M ) = det(M ) . C) and for any Aµ there exists an A µ such that M (A µ ) = a M (Aµ ) a† . In fact.33) This transformation conserves the hermitian property of M since (M )† = ( a M a† )† = (a† )† M † a† = a M a† = M (11. for any a ∈ SL(2. (11. (11.38) . In fact.32. + .39) Exercise 11. Due to det(a) = 1 the transformation (11.

i. C) and of SO(3.2: Lie Groups SL(2.11..41) results in 1 ¯ Lµ ρ = 4 where Aαβγδ = ν Γ = a2 σρ a† 2 (11. ¯ Lµ ρ = L(a1 )µ ν L(a2 )µ ρ = L( product in SO(3.44) (σν )αβ (σν )δγ . (11. (11. C) and SO(3.40)] respects the group property of SL(2.43) (σν )αβ Γβα Γγδ (σν )δγ = ν.C) )µ ρ (11.C) and SO(3. 2 (11. (11.1).β γ. β δ = δ = = δ = δ 1 2 = 2 = 1 Aαβγδ (11.δ 1 4 Aαβγδ Γβα Γγδ α.β γ. β 2.42) = ν Defining Γ = a† σµ a1 .1) L(a)µ ν provides a homomorphism 359 We want to demonstrate now that the map between SL(2.δ (11.1) defined through L(a)µ ν [cf.e.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 .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) 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. 1 ¯ and using the definition of Lµ ρ in (11.47) This completes the proof of the homomorphic property of L(a)µ ν .45) One can demonstrate through direct evaluation   2 α =    2 α =  2 α = =   2 α =    0 else β β γ γ = = = = γ = γ = 1.α.

.h.52) and (11. J3 of the Lorentz transformations Lµ ν in the natural representations. −A0 . 0). i.25) in the last step. C) which correspond to K. denoted below as κ1 .s.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. of (11. we write (11. Aµ = (0.48) yields for the l. Because of the condition det(a) = 1 only six independent real numbers actually suffice for the definition of a. 1. M (Aµ + (K1 )µ ν Aν ) = = M (Aµ ) + M (Aµ ) − M ( (−A1 . (11. 0. correspondingly. 1 (11. 1 (11. Aµ = (0. Aµ = (0.55) (11. 10.54) This reads for the four cases Aµ = (1. correspond to the generators given by (10.51) Insertion of (K1 )µ ν as given in (10.58) . J1 . 1. of (11. 0. 0. To this end we consider infinitesimal transformations and keep only terms of zero order and first order in the small variables. J2 . K3 . 0.2) which correspond to the generators K1 .53) Equating (11.57) . C) as complex 2 × 2–matrices are defined through four complex or. 0.47. (11.52) where we employed (11.53) results in the condition σ 0 A1 − σ 1 A0 = ( κ 1 σ µ + σ µ κ † ) Aµ . (11.49) we obtain using (11.48) We want to determine now the generators of the transformation a(z) as defined in (11. J Spinor Formulation The transformations a ∈ SL(2.h. 1) − σ1 σ0 0 0 One can verify readily that 1 κ1 = − σ 1 2 (11.49) assuming (note that g µ ν is just the familiar Kronecker δµν ) Lµ ν a = = gµν + 1 + 1 (K1 )µ ν κ1 . 0. 0) ) σ0 A 1 − σ1 A 0 (11. noting the linearity of M (Aµ ).360 Generators for SL(2.e. K2 . 0).36) M (Lµ ν Aν ) = a M (Aµ ) a† (11. eight real numbers.50) (11. 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  .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 ) . For the r. 0).56) (11. To obtain the generator of a(z) corresponding to the generator K1 .48).49). 0.s.

C) correspond to the generators K and J of Lµ ν . J3 are given by 1 κ = − σ. (11.11.63) where we extended the summation convention of 4-vectors to spinors. Similarly.4: Show that the generators (11. κ3 of a(z) corresponding to K2 . Here a(z) = ( aα β ) = describes the matrix a(z).64) . 2 λ = i σ.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. 2 (11. one can show that the generators κ2 . a(w − iϑ) = e− 2 (w − iϑ)·σ ∈ SL(2.62) We denote the general (z = w + iϑ) transformation by φ α = aα β φβ = aα 1 φ1 + aα 2 φ2 . hence. λ2 .61) This identifies the transformations a(z = w − iϑ) as representations of Lorentz transformations. def α = 1. − 1 . ϑ) = ew·K + ϑ·J ∈ SO(3.2. λ3 corresponding to J1 . 2 2 φ1 φ2 ∈ C2 .60) of a ∈ SL(2. the so-called contravariant spinors. 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 . + 1 2 2 1 transforms under rotations (z = iϑ) like a spin– 2 state | 1 . 11. Exercise 11.1) acts on 4–vectors Aµ . C) acts on spinors φα ∈ C2 (characterized below) 1 (11. w describing boosts and ϑ describing rotations. K3 and λ1 . 2 (11. state that the following two transformations are equivalent L(w. a1 1 a1 2 a2 1 a2 2 (11.60) We can.3: Spinors 361 obeys these conditions. J2 .

as covariant spinors. i. Definition of covariant spinors Expression (11.. 1 ) stands for the Clebsch–Gordon coefficient. which 2 can be constructed from products φ1 χ2 .66) is a bilinear form. In fact.. However.66) should constitute a singlet spin state.e. The relationship between the two can be expressed φ1 φ2 = = 3 φ1 φ2 0 1 −1 0 (11. i. −m 2 2 2 2 1 1 1 1 2 (11. this scalar product is anti-symmetric. . To arrive at a suitable scalar product we consider first only rotational transformations a(iϑ).67) We will refer to φα . (11. those of “particle 2” with the spinor χβ one can state that the quantity 1 Σ = √ 2 φ1 χ2 − φ2 χ1 (11.70) ‘Scalar’ implies invariance under rotations and is conventionally generalized to invariance under other symmetry trasnformations. exchange of φα and χβ alters the sign of the expression. indeed. . ± 1 . The corresponding states are defined through φ1 χ2 − φ2 χ1 = φ1 χ1 + φ2 χ2 It obviously holds χα = χ1 χ2 = χ2 −χ1 . 1 ) = 2 2 2 2 2 2 2 2 √ ±1/ 2 and equating the spin states of “particle 1” with the spinor φα .68) (11. 0| . .e.69) (11. . etc. 0| 1 . i.. . . The existence of a scalar product gives rise to the definition of a dual representation of the states φα denoted by φα . In the notation developed in Chapter 5 holds for the singlet state | . In this case spinors φα transform like spin– 1 states and an invariant. Such a scalar product does. 1 . .65) where | · · · 1 describes the spin state of “particle 1” and | · · · 2 describes the spin state of “particle 2” and (0. as we will demonstrate below such states are invariant under general Lorentz transformations a(z). 0.362 Definition of a scalar product Spinor Formulation The question arises if for the states φα there exists a scalar product which is invariant under Lorentz transformations. exist and it plays a role for spinors which is as central as the role of the scalar product Aµ Aµ is for 4–vectors.. 0| 1 . as contravariant spinors and to φα . χβ . m. . Using (0. ± 1 . χβ . is the singlet state. invariant under transformations a(z). should remain invariant under transformations a(iϑ). m 1 | . invariant and as such has the necessary properties of a scalar3 product. . 1 .e. 0 = 2 2 m=± 1 2 1 1 (0. −m) | .

73) (11. For this scalar product holds φα χα = −χα φα .75) (11.71) (11.71.77) (11.3.78) aβ γ γδ φδ (11.10. (11.1: Show that for any non-singular complex 2 × 2–matrix M holds M −1 = det(M ) M −1 T The matrices .68). 11.74)].74) = = 0 1 −1 0 0 −1 1 0 (11. −1 connecting contravariant and covariant spinors play the role of the metric tensors gµν .81) .70) and (11. 11. (11.11. 11.72) in a notation analogous to that chosen for contravariant and covariant 4–vectors [cf.75) is given by φα = as can be readily verified.76) αβ αβ The scalar product (11. 11.3: Spinors as can be verified from (11. g µν of the Minkowski space [cf. Proof that φα χα is Lorentz invariant We want to verify now that the scalar product (11. 11. Accordingly.69. 10.72)] φα φ α = = = = αβ αβ φβ φβ 11 21 11 21 12 22 12 22 (11.67) will be expressed as φα χα = φ1 χ1 + φ2 χ2 = φ1 χ2 − φ2 χ1 .80) One can write in matrix notation aα β γ δκ αγ a δ αβ (11. (10.72) 0 −1 1 0 Exercise 11.69.70) is φ1 φ2 −1 363 = = −1 φ1 φ2 . (10.63.73.77) is Lorentz invariant. In the transformed frame holds φ α χα = aα β αγ aγ δ δκ φβ χκ .79) = a −1 T a κβ . The inverse of (11. The transformation behaviour of φα according to (11. (11. we will express (11.

σ † = (σ ∗ )T = σ.14) one can write a −1 Spinor Formulation = e 2 z·σ 1 −1 = e 2 z· 1 σ −1 = e− 2 z·σ 1 ∗ (11. it holds φk = (φk )∗ . We denote ˙ (aα β )∗ = aα β ˙ (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 (11.364 Using (11. Insertion of (11. (11.81) yields aα β and. ˙ (11.87) which we will employ from now on. As discussed above.. The transformation ˙ behaviour of φα is ˙ φ α = (aα β )∗ φβ ˙ (11.2) and (11. 2.84) (11.63). ˙ We also define covariant versions of φα φα = (φα )∗ .86) A concise notation of the conjugate complex spinors is provided by ˙ (φα )∗ = φα = φ1 ˙ φ2 ˙ ˙ (11.88) which one verifies taking the conjugate complex of (11. hence. the ˙ conjugate complex spinors φα need to be considered separately from the spinors φα .91) .90) extending the summation convention to ‘dotted’ indices.88) reads ˙ ˙ φ α = aα β φβ ˙ ˙ (11.89) such that (11.e. Hence. i. from (11.80) φ α χα = φβ χβ .83) Here we have employed the hermitian property of σ. k = 1. a∗ (z) provides a representation of the Lorentz group which is distinct from that provided by a(z).83) into (11. Obviously.85) φ1 φ2 ∗ ∗ .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.

11. − 1 .73. is governed by the operator a∗ (z) −1 which arises in ˜ the Lorentz transformation (11. The transformation behaviour of φα is ˙ φα = ˙ ˙ β γδ ˙˙ φδ ˙ ˙ αβ a γ ˙ ˙ . 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 =  (11.98) ˜ ∗ (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 )   =   = .92) (11. in matrix notation.97) The transformation.97) implies then that φα transforms like Ψ ˙ φ1 ˙ φ2 ˙ transforms under rotations (z = iϑ) like a spin– 1 state | 1 . For the spinors φα and χα thus ˙ defined holds that the scalar product ˙ φα χα = φ1 χ1 + φ2 χ2 ˙ ˙ ˙ ˙ ˙ (11.75.96) is Lorentz invariant. i.95) ˙ where αβ and αβ are the real matrices defined in (11.76) φα ˙ ˙ φα αβ ˙ ˙ αβ ˙ ˙ = = = = αβ ˙ ˙ αβ ˙ ˙ αβ αβ φβ φβ ˙ ˙ (11.4 Spinor Tensors ˙ ˙ ˙ We generalize now our definition of spinors φα to tensors.94) (11. one can state ˜ and (11. Ψ4 . 11.4: Spinor Tensors 365 The relationship between contravariant and covariant conjugate complex spinors can be expressed in analogy to (11. A tensor tα1 α2 ···αk β1 β2 ···β (11. a property which is rather evident.18) of the bispinor wave function Ψ.75). A comparision of (11. (11. (11.e.99)  Ψ3 (xµ )   χ ˙ (xµ )  ˜ χβ (xµ ) ˙ 1 ˜ χ2 (xµ ) Ψ4 (xµ ) ˙ 11. a∗ (z) −1 accounting for the ˜ transformation behaviour of the third and fourth spinor component of Ψ.93) (11. + 1 2 2 2 1 transforms under rotations (z = iϑ) like a spin– 2 state | 1 .18) ˜ 3 ..100) . 11.

11.1: Prove equation (11. (11.. We had demonstrated that M (Aµ ) transforms according to M = M (Lµ ν Aν ) = a M (Aµ ) a† (11. it holds αβ .103) Similarly. (11. the elements of which are linear functions of Aµ . with ˙ contravariant indices αβ. ˙ ˙ ˙ ˙ (11. αβ ˙ αγ t ˙ γβ (11.101) An example is the tensor tαβ which will play an important role in the spinor presentation of the Dirac equation. .e.e.106) Exercise 11. According to (11. using conventional matrix indices j.25) the tensor is explicitly Aαβ = ˙ A1 1 A1 2 ˙ ˙ A2 1 A2 2 ˙ ˙ = A0 + A3 A1 − iA2 A1 + iA2 A0 − A3 .102) = . ˙ ˙ ˙ ˙ (11. (11. αβ = αβ (11. m. An example of a tensor is αβ and tions. the transformation bevaviour of a tensor tαβ reads in spinor and matrix notation t αβ = aαγ aβδ tγδ .25] M (Aµ ) = σµ Aµ . tjk = a t aT jk = .105) This tensor is actually invariant under Lorentz transforma= αβ .106).109) .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 . (11. i. i.108) Obviously. This task implies that we seek a tensor.m aj akm t m (11. An obvious candidate is [cf. i.104) Indices on tensors can also be lowered employing the formula tα β = and generalizations thereof. tjk = a t a† jk ˙ ˙ ˙ ˙ (11.107) which reads in spinor notation [cf.102.. this transformation behaviour is in harmony with the tensor notation adopted. we want to express Aµ through a spinor tensor.4.103) A αβ = aα γ aβ δ Aγ δ .. This tensor transforms according to t αβ = aα γ aβ δ tγ δ ˙ This reads in matrix notation. k.e.m aj a∗ t km m . The 4–vector Aµ in spinor form We want to provide now the spinor form of the 4-vector Aµ .

11. (11.110) one obtains Aα β = ˙ A0 − A3 −A1 − iA2 −A1 + iA2 A0 + A3 A0 + A3 A1 + iA2 A1 − iA2 A0 − A3 .112) A22 −A21 ˙ ˙ −A12 A1 1 . Hence. (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 . (11.111) A11 A12 ˙ ˙ A21 A22 ˙ ˙ ˙ 0 −1 1 0 (11.114) yields ∂αβ = ˙ ∂0 + ∂3 ∂1 + i∂2 ∂1 − i∂2 ∂0 − ∂3 . (11.2: Prove that (11. Using (11. ∂µ in spinor notation The relationship between 4–vectors Aµ .110) one can state ∂ αβ = Similarly. 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 . (11.117) ˙ ˙ ∂0 − ∂3 −∂1 + i∂2 −∂1 − i∂2 ∂0 + ∂3 . Aµ and tensors tαβ can be applied to the partial differential operator ∂µ . ˙ 2 (11.4: Spinor Tensors One can express Aαβ also through Aµ Aα β = ˙ ˙ 367 A0 − A3 −A1 + iA2 −A1 − iA2 A0 + A3 . employing (11.116) .4.110) The 4-vectors Aµ .115) is correct.113) Aα β = ˙ (11.115) Exercise 11. (11. 11.109.

of this equation is 0 ∗ (σ µ )∗ a and.e.f.243) in the chiral representation expressing S(Lη ξ ) by (11. −1 0 0 1 (11. .119) (11. i.26) and its contravariant analogue σ µ σµ = σµ = 1 0 0 1 1 0 0 1 . σµ .120) σµ 0 .121).118) 0 −1 −1 0 0 i −i 0 For this purpose we consider first the transformation behaviour of σ µ and σµ . (11. Hence. i. holds also in the chiral representation.125) a (σ ) Equation (11. hence.123) (a∗ )−1 Lν µ a L ν µ = = σν (σ ) ν ∗ −1 (11.125) is equivalent to µ ∗ −1 −1 a∗ (σ µ )∗ −1 −1 a Lν µ = (σ ν )∗ (11.124).243)].125) is equivalent to (11. (10.h. (11. Equation (10. Comparision of (10.s.118) yields ˜ γµ = ˜ Using (11. To obtain the transformation properties of σ µ we employ then (10...121) One expects then that the transformation properties of σ µ should follow from the transformation properties established already for γ µ [c. 1 0 0 −1 . σµ building on the known transformation behaviour of γ µ .124) .15) one can write σµ = and.229) and (11.124).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. one can conclude aσ µ ∗ −1 Lν µ σν 0 = . (11.124) constitutes the essential transformation property of σ µ and will be considered further. (11. 0 −i i 0 .243) holds independently of the representation chosen. This is ˜ possible since γ µ can be expressed through σ µ . 0 1 1 0 . . γµ = ˜ 0 µ )∗ (σ −1 0 σµ σµ 0 (σ µ )∗ −1 . .368 σµ in Tensor Notation Spinor Formulation We want to develop now the tensor notation for σµ (11.126) which is the complex conjugate of (11. We will obtain the transformation behaviour of σ µ .18) and γ µ by (11. Note that (10.122) aσ µ −1 a−1 −1 (a∗ )−1 0 Lν µ (11. (11. hence. (11.e.

107) to (11. equation (11.2. 11.4) with w = 0.103) ˙ identifies σ µ as a tensor of type tαβ . One can.131) one can express the transformation behaviour of σ µ in the succinct form Lν µ σ µ = σ ν .130) We argue in analogy to the logic applied in going from (11.133) Combining (11. 0 1 1 0 µ=1 .4).11. (11.g. 2. 0 −1 −1 0 µ=1 . hence.131) where a is given by (11. state that σ µ in a new reference frame is σ µ = a σ µ a† (11.   =  .129) We want to demonstrate now that the expression aσ µ [a∗ ]T is to be interpreted as the transform of σ µ under Lorentz transformations.e.  .. (11. This transformation behaviour. 0 i −i 0 µ=2 .129.118) (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. It holds according to (11. according to (11. (11.2.4: Spinor Tensors One can rewrite (11..16.127) Exploiting the hermitian property of σ.102. transformations (11. 1 0 0 −1 µ=3   .2) −1 369 (a∗ )−1 = −1 − 1 z ∗ ·σ ∗ 2 e = e− 2 z 1 ∗ · −1 σ ∗ = e2z 1 ∗ ·σ .108) that the same transformation behaviour applies then for general Lorentz transformations.124) a σ µ [a∗ ]T Lν µ = σ ν .2) −1 (a∗ )−1 = e2z 1 ∗ ·(σ ∗ )T = e2z 1 ∗ ·σ ∗ T = [a∗ ]T .134) . therefore. (11. (σ ∗ )T = σ yields using (11. 0 −i i 0 µ=2 . 3) σj −→ a(iϑ) σj a(iϑ) † = a(iϑ) σj a∗ (iϑ) T .132) 1 0 0 1 µ=0 . e. 11.124) using (11. 11. −1 0 0 1 µ=3   .  (11. under rotations the Pauli matrices transform like (j = 1. i. 11. In fact. (11. 11.128) One can express.

=  σαβ ˙  ˙ ˙ ˙ ˙ (σ21 )11 (σ21 )12 (σ22 )11 (σ22 )12   ˙ ˙ ˙ ˙ ˙ ˙ ˙ ˙ 21 (σ )22 21 (σ )22 (σ21 ) (σ22 ) ˙ ˙ ˙ ˙ 21 22 Equating this with the r. δ. (11. This yields   1 0 0 1   0 0 0 0 σ0 + σ3 σ1 + iσ2  (11. of (11. γ.138) the ‘inner’ covariant spinor indices. whereas ˙ account for the elements of the individual Pauli the ‘outer’ contravariant spinor indices.e. account for the 4–vector index µ.140) where we employed the expressions (11. i.224) ˙ Using (11.e.138) Note that all elements of σαβ are real and that there are only four non-vanishing elements. the submatrices correspond to tαβ spinors. Each of the 4 × 4 = 16 matrix elements in (11..s. We can summarize that σ µ and σµ transform like a 4–vector. 2. Spinor Formulation (11. 1. σµ in which the contravariant indices are moved ‘inside’. µ = 0. 3 as well as through two spinor indices αβ. in fact.136) results in the succinct expression 1 2 ˙ γδ ˙ (11.110) to express σ αβ in terms 0 −1 1 0 0 0 0 1     (11. state   ˙ ˙ ˙ ˙ (σ11 )11 (σ11 )12 (σ12 )11 (σ12 )12 ˙ ˙ ˙ ˙   ˙ ˙ ˙ ˙ ˙ γδ   (σ11 )21 (σ11 )22 (σ12 )21 (σ12 )22 ˙ ˙ ˙ ˙   . The ˙ desired change of representation(σµ )αβ −→ (σµ )αβ corresponds to a transformation of the basis of ˙ spin states f g f −→ = (11. α.141) σ αβ = σ0 − σ3 σ1 + iσ2 −σ1 − iσ2 σ0 + σ3  = 2  0 0 0 1 0 −1 0 0 . hence. matrices..370 Inverting contravariant and covariant indices one can also state Lν µ σµ = σν .136) σαβ = = 2 ˙   σ1 − iσ2 σ0 − σ3 0 0 0 0 1 0 0 1 where on the rhs. 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 . We can. however.139) g −f g and. account for the 4-vector µ.e.12) for of σµ yields together with (5. the transformation is inverse to that of ordinary 4–vectors.137) σαβ ˙ = δαγ δβ δ .135) This is the property surmised already above [cf. i. We will now consider the representation of σ µ . ˙ ˙ In (11.h. β.. We want to express now σ µ and σµ also with respect to the 4–vector index µ in spinor form employing (11. ˙ ˙ 0 −1 1 0 ˙ (11.132) and (11.114). ˙˙ (11.29)].133) is characterized through a 4-vector ˙ index µ. and the covariant indices are moved outside. i.

employing then transformation (11. Accordingly. One can write then the scalar product using (11.146) Exploiting the property (11.11.137) 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 ˙ ˙    .140) to transform each of the four submatrices.144) results in the simple relationship γ µ Aµ = ˜ 0 Aα β ˙ Aα β 0 ˙ . we write ˜ ˙ γµ = ˜ 0 (σ µ )αβ µ ) )∗ ((σ αβ 0 ˙ ˙ . This expression implies that the element of γ µ given by σ µ −1 is in the expression (11. which in ˙ (11. ˙˙ (11.145) Let Aµ be a covariant 4–vector. (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 .142) = This can be expressed 1 2 σ αβ ˙ ˙ γδ = δαγ δβ δ .143) Combined with (11.121) of γ ˜ ˜ ˙ basis |αβ whereas the element of γ µ given by σ µ is in the basis |αβ .  ˙ ˙ ˙ ˙       (11.138) one can conclude that the following property holds 1 2 ˙ γδ σ αβ ˙ = 1 2 σ αβ ˙ ˙ γδ = δαγ δβ δ .144) The Dirac Matrices γ µ in spinor notation We want to express now the Dirac matrices γ µ in spinor form.147) . For this purpose we start from the ˜ µ . ˙˙ (11. (11.4: Spinor Tensors 371 and. (11.

150) .5 Lorentz Invariant Field Equations in Spinor Form Dirac Equation (11.372 Spinor Formulation 11. α ˙ = = m φα m χβ . ˙ (11. (11.148) ˜ Employing Ψ(xµ ) in the form (11.147) allows us to rewrite the Dirac equation in the chiral representation (10.226) ˜ i γ µ ∂µ Ψ(xµ ) = 0 i ∂ αβ ˙ i ∂ αβ 0 ˙ ˜ ˜ Ψ(xµ ) = m Ψ(xµ ) .149) (11.99) yields the Dirac equation in spinor form i ∂ αβ χβ ˙ i ∂ αβ φ ˙ The simplicity of this equation is striking.

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

k = 1. l and the corresponding energy levels are independent of m. As an example of this we will consider the Coulomb problem with the Hamiltonian H = p2 k − . In order to illustrate this. we first consider a particular example of the implications of symmetry..λn . can be identified with the angular momentum operators. 2m r (12. This degeneracy follows from the fact that any rotation. 3 .6) From elementary quantum mechanics we know the spectrum of the hydrogen atom. Jk .8) ˆ [H.m (r) Y m (θ. . Sk . Sk .. generates a state which has the same energy as the original one. this implies a degeneracy in the spectrum.λ1 . Classically. n − 1. Jk ] = 0 . the states 3s. .12).λn . (7. .24).. . We will investigate this shortly in more detail in the case of systems with spherical symmetry.4) The stationary Schr¨dinger equation. ..374 Spinor Formulation The action of any symmetry generator. (12. l. (12... leaves the energy of the state invariant H exp(iαk Sk ) ψE. The spherical symmetry implies the commutation of the Hamiltonian with angular momentum operators (7.λn = E exp(iαk Sk ) ψE.3) If the newly obtained state is linearly independent of the original one.8) . Lie groups play an essential role in the discussion of mass degeneracies in particle physics. (7.m (r) = vE. has been discussed.λ1 . as studied in chapter 7. V (r)...18). namely motion in a spherically symmetric potential described by the group SO(3) (or its double covering SU (2) as discussed in section 5.5) with m = −l. Therefore. is allowed to take values in 0. φ)..λ1 ... 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 . as represented by an element of SO(3). m r (12. each energy level is (2l + 1)-fold degenerate. 2. 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. . 3p and 3d all have the same energy. (7. . can then be reduced to a one-dimensional radial o equation. The energy levels are totally independent of l. (12. (12. · · · . In chapter 7 the dynamics of a particle moving in three dimensions under the influence of a spherically symmetric potential. The energy levels are mk 2 En = − 2 2 .. where the symmetry generators.39). For example. on an energy eigenstate. which yields a set of eigenstates of the form ψE. Presence of additional symmetries may further increase the degeneracy of the system.7) 2 n where the orbital angular momentum. ψE.

20) (12. 2m r 1 x2 (J1 p3 − J3 p1 + i p2 ) − k .13) It can be verified explicitly that its components commute with the Hamiltonian. Jj ] = i [Ki . In order to understand the aforementioned extra degeneracy.11. which have a negative energy E.18) which complement the familiar angular momentum algebra of section 5.3.17) (12. 2 which can be shown to satisfy [Ai . For this purpose we first note that J · = 0. (12. Therefore.15) which can be proved after very considerable algebra.10) (12.12) (12. In the following we consider the bound states.9) (12. ijk Jk .21) . 2 1 B = (J − K). Aj ] = i ijk Ak . which is valid even when a and b do not commute. (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 .19) (12.5: 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. 2m r 1 x3 (J2 p1 − J1 p2 + i p3 ) − k .8) is not a hermitian operator. 2m r 1 x1 (J3 p2 − J2 p3 + i p1 ) − k . in the subspace of the Hilbert space corresponding to a certain energy we can replace H by E. we will compute the hydrogen spectrum using the additional symmetry. We will also need the following identity [4] 2 = 2H J2 + m 2 + k2 . We introduce the following new operators 1 A = (J + K).11) (12. The vector in (12. (12. 2m r (12. 2E (12.14) This follows from a · (a × b) = (a × b) · a = 0.16) Through some algebra [4] the following commutation relations can be verified [Ki . Kj ] = i ijk Kk . Now we scale the eccentricity vector as follows K= − m .

the energy eigenvalues are found to be E=− mk 2 1 . . . In contrast to the discussion above about extra symmetries.23) (12. . H B. . (12. Bj ] = 0. .28). . 1. .15). . b = 0.31) where we have used (12. Using this equation the energy eigenvalues can be written in terms of the eigenvalues of A2 and B 2 operators. 1. . a = 0. . 2 (2a + 1)2 2 2 (12. . 2 (12. .26) (12. Furthermore the bound on the orbital angular momentum. This illustrates the effect of additional symmetries to the degeneracy structure of a quantum mechanical system.25) So far we have shown that the symmetry generators form an algebra.33) (12.21) and (12. a lack of symmetry implies a lack of degeneracy in the energy levels of a quantum mechanical system. . . Spinor Formulation (12. A.24) (12.34) It follows that l has to have values in {0 = |a − b|. 1. 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. = 0. can be seen to follow from the triangle inequality as applied to J = A + B.27) (12.376 [Bi . ijk Bk . l.22): A2 = a(a + 1) B 2 = b(b + 1) Following (12. H = 0. namely that J J > < A − B =0 A + B =2 A (12.29) (12. . By comparing to the rotation algebra introduced in chapter 5.14) we note that A2 − B 2 = J · = 0. 1. a + b = n − 1}.22) (12. . This implies that a = b. The most extreme case of this is the quantum analogue of a classically chaotic system. 2 1 .7) tells us that (2a + 1) = n.28) .32) A comparison with (12. . Chaos is described classically as exponential . . Bj ] = i [Ai .30) 2 2 2 1 . Noticing that A2 and B 2 have the same eigenvalues because of (12. a = 0. which is identical to the the direct sum of the Lie algebra of of two SO(3) (or SU (2)) algebras. we can read off the eigenvalues of A2 and B 2 from (12. .

57M eV /c2 . A typical example of this so called quantum chaos is the quantum billiard problem. in the sense that nearby trajectories in the phase space diverge from each other over time. This was motivated by the observation that their masses are approximately equal: mp = 938. which is a particle in box problem in two dimensions with a boundary which can be chosen arbitrarily. (12. (Strong interactions bind the atomic nucleus together. 12.) If the proton and the neutron are to be viewed as two linearly independent states of the same particle.36) into each other in abstract isospin space. This (approximate) degeneracy led into the idea of the existence of an (approximate) symmetry obeyed by the underlying nuclear interactions. after the discovery of the neutron in 1932. For example.28M eV /c2 . For a more detailed discussion of quantum chaos in billiard systems we refer the reader to [7] and the references therein. for the . which was discovered by observing degeneracies in the particle spectrum. This enables us to utilize what we already know about the SU (2) symmetry group from the study of angular momentum. If the chosen boundary is ‘irregular’ in a suitably defined sense.5: Lorentz Invariant Field Equations 377 sensitivity to initial conditions. another manifestation of chaos is the lack of independent operators commuting with the Hamiltonian. (12. the isospin symmetry is also governed by an SU (2) group ‘rotating’ components in (12. n= 0 1 . mp = 939. Following the mass-energy equivalence of special relativity E = mc2 .35) this mass equivalence can be viewed as an energy degeneracy of the underlying interactions. However. (Weak interactions are responsible. In the next section we will proceed with the discussion of a symmetry. that the proton and the neutron behave identically under the so-called strong interactions and that their difference is solely in their charge content.36) In analogy to the concept of spin regarding the rotations in 3-space as discussed in chapter 5. analogous to the 1 spin-up and spin-down states of a spin. for example. the classical trajectories will diverge from each other after successive bounces from the boundary. Originally it was believed that isospin was an exact symmetry of strong interactions and that it was violated by electromagnetic and weak interactions. This is known as level repulsion.2 system p= 1 0 . 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. who. It is important to place the isospin concept in its proper historical context. 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. namely.2 Isospin and the SU (2) flavor symmetry The concept of isospin goes back to Heisenberg.11. it is natural to represent them in terms of a two component vector. suggested that the proton and the neutron can be regarded as two states of a single particle.

11. Therefore it remains a useful concept. Denoting the total isospin. 1 π 0 = |1.378 Spinor Formulation beta decay).− .42) (12. However. I3 .− 2 2 . 1 . (12.43) Similarly. (mπ± = 139.− 2 2 . −1 . including the spin and color quantum numbers of their constituent quarks. 2 2 1 2 2 2 (12.39) All other isomultiplets.37) As an example of a multiplet with I = 1 we have the three pions or π-mesons π + = |1. The mass difference between the neutron and the proton could then be attributed to the charge content of the latter. π − = |1. mπ0 = 135. n = I = . and its third component. I3 = − 2 2 . Isospin symmetry is not an exact symmetry of strong interactions. 2 (12. We can describe isospin multiplets the same way we have described the angular momentum and spin multiplets. 0 . Shortly we 1 will see how to describe both the pion and nucleon states as composites of more fundamental I = 2 states.38) are ud. −1 = = = 1 1 1 1 . the proton is the lighter of the two. . . 2 2 1 2 2 2 1 1 1 1 1 √ . the proton had to be heavier. 1 we can re-write (12. 0 π − = |1. sec. In the framework of the quark of model. π 0 = |1. . the proton and the neutron can be written as totally symmetric uud and udd states.6M eV /c2 . albeit it is a good approximate one. are made up of these two quarks. it can be seen as part of a larger (and more approximate) symmetry which is of great utility to classify observed particle families. (12. (12. 2. They u form an isotriplet: π + = |1.38) which have all nearly identical masses. as good quantum numbers.− . If it were otherwise the proton would be unstable by decaying into the neutron.− 2 2 1 2 2 2 1 1 1 1 . spelling disaster for the stability of matter. 2 2 1 1 1 . d= 1 1 .0M eV /c2 ). as we shall see below. For a precise description of the two nucleons as composite states.41) (12. respectively. u¯ and d¯ states. 2 2 . For example. the three pions in (12. we refer the reader to [1]. If the mass difference (or the energy difference) were to be to be purely electrostatic in nature. I.36) as a multiplet with I = 2 1 1 p = I = . the fundamental representation of the isospin symmetry corresponds to the doublet that contains the so-called up and down quarks u= 1 1 . including the proton and the neutron. They can be constructed with the same rules that have been used for angular momentum addition ¯ u in chapter 6.40) + 2 1 1 . . Further than that. I3 = 2 2 1 1 .

I3 = 0. (12.48) (12. strongly interacting particles) which is about a GeV /c2 . This is why isospin is such a good symmetry and why isomultiplets have nearly identical masses.37) and in analogy to (12. 0 ).43). mesons have intermediate mass ranges. By convention ¯ baryons have baryon number 1. consists of a proton and a neutron. In the late 1940’s and early 1950’s. baryons generally are heavy. The up and down quarks have strangeness zero. If taken literally. As it later turned out. such as proton and the neutron. baryon and meson.44) 2 where B is the baryon number and S is the strangeness. 0 = √ (|p > |n > −|n > |p >) 2 and that of an isotriplet |1. 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. Following (12.) are light. The possibilities are that of an isosinglet 1 |0. and quarks have baryon number 1 . but at the time the empirical concepts like isospin and strangeness quantum numbers were in use. by accidental convention. −1 = |p > |p >. Therefore the deuteron has to be an isosinglet state (|0.11. The value of the strangeness quantum number is taken. 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). Naturally.46) (12. All antiparticles have their 3 quantum numbers reversed. mesons have baryon number 0.5: 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 reader may know that the deuteron. (12. Now let us consider another example of combining the isospins of two particles. this remains only an inaccurate naming convention today. Therefore it has to have isospin. We will now try to describe its wave function in terms of its constituent nucleons. the up and down quarks are not the only quark ‘species’ . to be −1 for the strange quark. the pions being examples thereof. 0 ) we should have seen nn and pp bound states of comparable energy in nature (because of isospin symmetry). 1 |1. originally refer to the relative weight of particles. 0 |1. this will be mathematically identical to adding two spins. a number strange particles have been found which presumably contained a third quark species: the strange quark. muon. 1 = √ (|p > |n > +|n > |p >).47) (12. as some mesons discovered later are heavier than some baryons and so on.or flavors as they are commonly called. It shall be noted here that the quark model was not invented until 1960’s. but such states do not exist. . the neutrinos etc. 2 = |n > |n > . whereas mesons are q q states.45) Is the deuteron an isosinglet state or an isotriplet? If it were an isotriplet (|1.e. The names. a hydrogen isotope. where leptons (electron. All other composite states have their strangeness given by the sum of the strangeness content of their constituents. we shall setup some terminology: baryons are qqq states. Before proceeding further.

. σ. where M is the scattering amplitude given by M = final | α I(1) · I(2) | initial . For example.7 to the case of isospin. consider (I) p + p → d + π + (II) p + n → d + π 0 (12. I3 (f ) = 1. The dot product here refers to the abstract isospin space. γ (i) is a reduced matrix element defined in the same sense as in section 6. We will now employ the (isospin analogue) of the Wigner-Eckart theorem (6.7. I3 . Consider the generalization of tensor operators discussed in section 6.49) The only assumption that we put in will be that the interaction is of the form V = α I(1) · I(2) .8. 0 ). 1 and |1. Show that the expectation of I(1) · I(2) is state. 0 . γ (f ) |T00 |I (i) I3 .55) .54) Here T00 ≡ V = α I(1) · I(2) . is proportional to |M|2 . γ (i) . (12.8 to compute the ratio of the scattering amplitudes MI and MII . Exercise. 1 and (1/ 2)(|1. which is an isoscalar.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. γ (f ) ||T0 ||I (i) .53) 0 √ 1 2I (i) +1 I (f ) .259) in the present context: I (f ) I3 .51) (12. γ (f ) T00 I (i) = 1.45) and (12.52) where γ (i) and γ (f ) denote degrees of freedom other than isospin. γ (f ) ||T ||I (i) . Show that I(1) ·I(2) is an ‘isotensor’ of rank zero. (12. γ (f ) (f ) (i) (12. γ (i) = (f ) (i) (I (f ) I3 |00I (i) I3 ) (f ) (i) (−1)I −I (f ) (i) (12.7 and 6. We will eventually be able to compute ratios of scattering cross-sections between different processes. The initial and final states can be denoted in more detail as | initial | final = = I (i) . 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. such as the spatial dependence of the wave function and spin.50) . γ (i) I (f ) . 0 + |0. γ (i) (i) (12. √ According to (12. 1 4 in an isotriplet state and − 3 in an isosinglet 4 Using (12. I3 . respectively.48) the initial states in (I) and (II) have isospin values |1.5 for the spin-analogue of the same problem. (f ) (i) (I (f ) I3 |00I (i) I3 ) is a Clebsch-Gordon coefficient and I (f ) . Exercise. The cross-section.380 Spinor Formulation As an exercise on the implications of isospin symmetry we will consider nucleon-nucleon scattering. For completeness let us start by restating the Wigner-Eckart theorem (6. I3 = 1. as discussed in the exercise above. Refer to exercise 6.259) discussed in detail in sections 6.

1 1 1 1 3 √ 2. σII which is in approximate agreement with the observed ratio.59) (12.5: Lorentz Invariant Field Equations 1 (11|0011) √ I (f ) = 1. but these are not dominant at relatively low energies. γ (f ) T00 I (i) = 1. I3 = 0. 2 = 2. I3 = 0. It follows σI = 2.60) (12. = MII (1/ 2) (12. γ (i) 3 +0. = 381 (12. −2 1 1 2. 1 π − + p : |1.56) (12. −2 1 1 √ . γ (f ) T00 I (i) = 0. (12. particles with strangeness. −1 2 3 2 − + . 0 1 1 3 3 2. The relevant Clebsch-Gordon coefficients are easily evaluated: (11|0011) = (10|0010) = 1. [2] As a further example.64) There are more exotic possibilities. 2 .62) We can now write the ratio of the scattering amplitudes: MI 1 √ . (12. which is also manifested by a vanishing Clebsch-Gordon prefactor.11. 2 = 3 2. γ (i) 3 1 (f ) (i) I (f ) = 1. I3 = 0.65) (12. for example. (c) π − + p → π 0 + n . σ( π − + p → anything ) The possibilities are (a) π + + p → π + + p . −2 . γ (f ) ||T00 || I (i) = 1. I3 = 0.57) (12. (b) π − + p → π − + p .61) MII Note that the second term in MII vanishes due to the isospin conservation. Once again we need the isospins for the initial and final states. (12.63) where common dynamical factors (which would not be as easy to compute) have dropped out thanks to the Wigner-Eckart theorem. . −2 2 1 1 3 2. we will consider pion-nucleon scattering. γ (i) 2 1 = (10|0010) √ I (f ) = 1. involving.66) = 2 3 1 3 2.58) (12. γ (i) = √ 2 1 (f ) (i) +√ I (f ) = 1. −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. which are obtained by a standard Clebsch-Gordan expansion π + + p : |1.

for example. Hence Gell-Mann and independently Nishijima postulated the existence of a separate quantum number. . was motivated by the existence of particles that are produced strongly but decay only weakly. which can be produced by π − + p → K + + Σ− . A computation similar to the previous example of nucleon-nucleon scattering yields (apart from common prefactors) the following amplitudes for the reactions in (12.71) u =  0  . It was assumed that strong interactions conserved S (at least approximately).3 The Eightfold Way and the flavor SU (3) symmetry The discovery of the concept of strangeness. (12. S. it appears an obvious generalization to add another component for an extra quark to the isospin vector space       1 0 0 (12. called strangeness.70) (12. K + . For instance. which leads to the following 2 2 ratios for the cross-sections σa : σb : σc = 9 : 1 : 2. are up to 40% heavier. Hence the strangeness changing strong decays of K + (or Σ− ) were forbidden. mentioned in the previous section. m 1 2. however. such that S(K + ) = 1. m 1 2. m . Strange partners of the familiar nucleons.65) Ma = M 3 2 Mb = 1 M 3 + 2 M 1 3 3 √ 2 Mc = 2 3 M3 2 √ 2 (12. 0 0 1 .69) 12. The classification of the newly found particles as members of some higher multiplet structure was less obvious then the case of isospin. In the light of the quark model. m V V 3 2.68) − 2 3 M1 2 Guided by empirical data we will further assume that M 3 >> M 1 . making an identification of the underlying symmetry and the multiplet structure less straightforward.s =  0  .382 Spinor Formulation As in the example of nucleon-meson scattering we define the relevant matrix elements M3 = 2 M1 = 2 3 2.d =  1  . has a lifetime which is comparable to that of π + albeit being more than three times heavier. while weak interactions did not. etc. giving it a higher than usual lifetime.67) which are independent of m. [2] (12. S(Σ− ) = −1 and S(π) = S(N ) = 0. 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.

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

which can be verified explicitly as matrix identities. regardless of the dimension of the representation. We can also introduce the constants. U . 3 (12. since det(eA ) = etrA .   . δjkl . As in the case of SU (2) higher dimensional representations obeying the same structure can be found.79) (12. λk ] = 2ifjkl λl . the meaning being apparent from the context. . any unitary matrix.75) αk λk (12. λ3 =  0 −1 0  .7.77)  λ1  . via the anticommutator relation 4 [λj . −2 λ4 λ6 λ8 The generators. λ7 =   0 0  . obey the following relation tr(λj λk ) = 2δjk .32). λk .78) where fjkl are the antisymmetric structure constants similar to the jkl of SU (2) given in (5. The fundamental relation to be preserved is (12.) The unit determinant condition requires that all λk are traceless. As discussed in section 5. (We shall at times refer to the Lie algebra with the name of the group. The Lie algebra structure is given by the commutators of λk [λj . (12. Therefore we will express elements of SU (3) as U = ei k (12.80) This is the fundamental or defining representation of SU (3).384 Spinor Formulation SU (3) and not U (3) that is of main interest. λk ]+ = δjk + 2δjkl λl . λ2 =     . 0 0 0   . The reader is invited to compare the structure of SU (3) to that of SU (2) discussed in section 5.76) where λk are 8 linearly independent matrices forming the basis of the Lie algebra of SU (3).79).1. 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  . λ5 =   . can be written in the form U = eiH where H is a hermitian matrix. (12.

(12. which satisfied an ‘eigenvalue equation’ [J0 .81) In chapter 5. J± ] = ± J± .92) (12. 1 [T0 . (12.88) which defines (not uniquely) a maximal set of mutually commuting operators {Y. 2 (12. Using the convention in (12. 2 1 [T0 . J− } proved useful. In the case of SU (2) the identification of ‘ladder’ operators {J+ . 3 (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 .90) (12.94) . I3 . T0 ] = 0. V± ] = ± V± .85) (12. U± ] = U± . V± ] = ±V± . In the basis (12.91) [T0 . U± ] = ±U± . (12. The generators of SU (3) can be arranged into a very similar form to that of SU (2).5: Lorentz Invariant Field Equations 385 As it turns out the set of generators in (12. U± = F6 ± iF7 . Y is called the hypercharge.87) Exercise.44). We first introduce the F-spin operators 1 Fi = λ i . [Y.84) (12. starting with the observation that Y = B + S.93) (12.83) (12.11. T± ] = 0. (Recall that the strange quark has S = −1 by convention). T0 } and [Y.86) (12. namely spherical harmonics. T± ] = ±T± . First. Exercise. T 0 = F3 . these relations have been used to construct the angular momentum spectrum as well as the function space representation of the rotation algebra. 2 (12.77) is not the most useful basis in the study of SU (3).71) show that T0 is the isospin operator. we have [Y.89) (12.87) the commutation relations between the generators can be expressed in a succinct manner. V± = F4 ± iF5 . [Y. 2 Y = √ F8 . Derive the Gell-Mann–Nishijima relation (12.

104) (12.99) (12.386 Spinor Formulation which relate the remaining generators {T± . V+ ] = 0.95) (12. Thus. we have [T+ . T0 }.103) Any remaining commutators follow from hermiticity. similar to the way it was done in section 5.96) (12.88). When the basis of a Lie algebra is expressed in such a way to satisfy the form of the eigenvalue relations as given above. The interested reader is instead referred to a very readable account given in chapter 12 of [4]. Another important property of SU (2) is the existence of an operator (namely the total angular momentum. U0 and V0 are linear combinations of T0 and Y . U± } to this maximal set by ‘eigenvalue equations’ and [T+ . However. V− ] = T− . C1 = k λ2 = − k 2i 3 fjkl λj λk λl . jkl (12. In the case of SU (2) the representations lay on a line on the J0 axis. V− ] = −U− .101) (12. we will use these commutation relations to construct representations of SU (3). . 2 3 Y + T0 = 2V0 . U− ] = 0. U− ] = Y − T0 = 2U0 . We can construct two such independent Casimir operators for the group SU (3). Note that. V± . As in the case of J 2 and SU (2).5.102) (12. [V+ . U+ ] = V+ . V− ] = 2 (12. A formal definition and a detailed discussion of the Cartan-Weyl form is beyond the scope of this chapter. the representations will now lie in a T0 − Y -plane. it is said to be in Cartan-Weyl form. as the relation between the states in a given representation can be conveniently expressed in terms of ladder operators. An operator which commutes with all generators of a Lie group is called a Casimir operator. The maximum number of mutually commuting generators of a Lie algebra is called its rank.100) (12. V+ ] = 0.97) which relate commutators of generators with opposite eigenvalues to the maximal set {Y. (12.98) (12. SU (2) has rank 1.105) C2 = jkl djkl λj λk λl In general the number of independent Casimir operators of a Lie group is equal to its rank. This form is essential for easy labeling of the representations of the group. [T+ . [T+ . Finally. [U+ . J 2 ) which commutes with all of the generators. [T+ . 3 [U+ . while SU (3) has rank 2. [U+ . The same way the angular momentum ladder operators have been used to construct the representations of SU (2). T− ] = 2T0 . since there are two mutually commuting generators in SU (3) as given in (12. Casimir operators can be used to label irreducible representations of the Lie algebra.

For example the pion octet (or any other meson octet) is therefore realized as states of a quark . y 3 = ( y− 4 3 = ( y+ 4 1 t3 ) |t3 . For example.5: Lorentz Invariant Field Equations 387 The utility of Casimir operators arises from the fact that all states in a given representation assume the same value for a Casimir operator. y 1 = α t3 ± .112) The effect of these operators to the states in the y − t3 plane have been outlined in Fig. it was shown in section 5. y . 2 1 = β t3 ± . y : T0 |t3 .5 how to label the irreducible representations of the angular momentum algebra SU (2) in terms of the value of the total angular momentum.antiquark pair. the pion family forms part of an octet corresponding to the D(1. = y |t3 .y ± 1 . y ± 1 . For example. A notation suggestive of the dimensionality of the representation can be used to identify representations. y . Now we will construct explicit representations of SU (3). V± . especially chapters 7 and 8. y U± |t3 . 0) and D(0. Such a representation is labeled as 1 D(p. Figure (12. (12.11. y Y |t3 .1). y V± |t3 .108) (12. 2 1 = γ t3 . y .106) (12. The details of this procedure is beyond the scope of this chapter. Because of (12. y = t3 |t3 . y . . The representations of SU (3) are constructed analogous to those of SU (2) by identifying the ‘boundary’ states annihilated by raising (or lowering) operators. The representations D(1. 2 (12.110) (12. All other states of the representation are then constructed by successive application of ladder operators T± . y .111) (12. y V0 |t3 . (12. The interested reader is referred to [4]. 1) as an example.) All other representations can be constructed by combining these two conjugate representations. 1) correspond to the triplets of quarks and antiquarks.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 .109) The same way that J± |m is proportional to |m ± 1 in the case of the angular momentum algebra. (See Fig. For example. we have U0 |t3 . respectively. U± .88) we can label states by the eigenvalues of T0 and Y operators. (12. |t3 . 2 1 t3 ) |t3 . This is because the states in a given representation are connected by the action of the generators of the Lie algebra and such generators commute with the Casimir operators. we have T± |t3 . The representations for hexagons with sides of length p and q in the T0 − Y -plane. q) and it has a dimensionality of 2 (p + 1)(q + 1)(p + q + 2). This property can be used to label representations in terms of the values of the Casimir operators.2) shows the representation D(2.3). As another example for the representations of SU (3). 1) representation. 2 (12. y .

2: D(2.1: The effect of SU (3) ladder operators on the y − t3 plane.388 £ Spinor Formulation Figure 12. 1) representation of SU (3). ¡ ¢  ¡ ¢  8 6 2 '745301) ¥¦¤   ¤   #"  ! (& % '$   © ¨     ¦ §¥ ©§ ¨ . £ Figure 12. The states in the inner triangle are doubly degenerate.

2 (12. . 3 (12. where we would write [3] ⊗ [3] = [1] ⊕ [3] ⊕ [5].114) Figure 12.11. 2 3 1 1 − .5: Lorentz Invariant Field Equations 389 D(1. .3: D(1.116) (12. 3]. £ (12.113) The additional singlet state corresponds to the η meson. 2 3 2 0. This expansion can be compared. to the case of adding two spin triplet states.117) The Gell-Mann–Nishijima relation can then be succinctly expressed as 1 Q = y + t3 . for example. 0) and D(0. 1) are denoted by [3] and [¯ respectively. This way the quark-antiquark states can be represented as follows [3] ⊗ [¯ = [8] ⊕ [1] 3] (12.115) (12. − .118) ¡ ¢    ¨ ©§ ¨¥¦¤ "   ! . 1) is written as [8] etc. in the case of SU (2). The octet D(1. 0) or the fundamental representation of flavor SU (3) symmetry. The three quarks in the fundamental representation can now be written as u = d = s = 1 1 .

D. A. W. G. 3 1 Qs = − . W. Martin. [3] Microscopic theory of the nucleus. 1987. T. 67 (2). D. Schulten. [7] Expansion method for stationary states of quantum billiards.-P. W. Halzen. (1999).120) (12.119) (12. Wiley. 1993. D.4: The baryon octet as a D(1. and spectroscopy. 3 £ Spinor Formulation (12. 1984. Cheng. Springer-Verlag u [5] Principles of symmetry. Wiley. B. Griffiths. Greiner. 1) representation of SU (3). [2] Introduction to elementary particles. J. dynamics. [4] Quantum mechanics: symmetries. 133. Harter. Phys. References [1] Quarks and leptons: an introductory course in particle physics. North holland. Eisenberg. [6] Gauge theory of elementary particle physics. Kaufman. L.390 from which the quark charges follow 2 Qu = . F. L.121) L 0 X - X 0 Figure 12. Am. M¨ller. Kosztin. p. 1989. K.-F. I. 3 1 Qd = − . ¡ ¢  S - S 0 ¤ ¥ S + . J. 1972. Li. Oxford. Wiley. Greiner. 1984. M.

Sign up to vote on this title
UsefulNot useful