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

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

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

14) for any continuous function h(t) ∈ F with h(t0 ) = h(t1 ) = 0.1: Variational Calculus 3 We now consider the question for which functions the functionals of the type (1. This condition is stated in the following theorem. y(x + dx)) is ds = (dx)2 + ( dy dx)2 = dx dx 1+( dy 2 ) .   ∂qj dt ∂ qj ˙ t0 j=1 (1. According to the Lemma above follows then (1. M . Let us assume that the two points are connected by the path y(x).12) for j = 1. y1 ) and (x2 .6) which reads in the present case     M ∂L t1 d ∂L δS[q(t). hence. . δq(t)] = dt − δqj (t) . M and δqj (t0 ) = δqj (t1 ) = 0 follows according to (1. if and only if it satisfies the conditions (j = 1.13) dt f (t)h(t) = 0 t0 (1. . t1 ] ⊂ R → R holds t1 (1. y(x1 ) = y1 . . y(x)) to (x + dx. .16) This property holds for any δqj with δq(t) ∈ F .5). 2.11) The extremals qe (t) can be identified through a condition which provides a suitable differential equation for this purpose. · ] ≡ 0 and.15) We will not provide a proof for this Lemma.12) for j = 1. t1 ]. On the other side. y2 ). (1.5) assume extreme values. . δq(t)] = 0 for all δq(t) ∈ F . then f (t) ≡ 0 on [t0 . y(x2 ) = y2 . . . The proof of the above theorem starts from (1. Theorem: Euler–Lagrange Condition For the functional defined through (1. . it holds in case δq(t0 ) = δq(t1 ) = 0 that qe (t) is an extremal.17) .16) the property δS[qe (t). . . For this purpose we define Definition: An extremal of a differentiable functional S[ ] is a function qe (t) with the property δS[qe (t). (1. 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 . M ) d dt ∂L ∂ qj ˙ − ∂L = 0 ∂qj (1. the above theorem.1. The length of such path can be determined starting from the fact that the incremental length ds in going from point (x. 2. 2. from (1. dx (1.12) The proof of this theorem is based on the property Lemma: If for a continuous function f(t) f : [t0 .

q(t).4 The total path length is then given by the integral x1 Lagrangian Mechanics s = x0 dx 1+( dy 2 ) . dx ) = 1 + (dy/dx)2 . (1..23) Presently we consider only velocity–independent potentials. q(t). 1.19) From this follows y / 1 + (y )2 = const and.22) ˙ where L(q(t). j = 1. The constants a and b are readily identified through the conditons y(x1 ) = y1 and y(x2 ) = y2 . t) = j=1 1 mj qj − U (q1 . qM ) . .1: Show that the shortest path between two points on a sphere are great circles. 2. . The shortest path is an extremal of s[y(x)] which must. . .18) dy s is a functional of y(x) of the type (1. i. according to the theorems above. q2 . M (1. ˙2 2 (1. obey the Euler–Lagrange dy condition.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). One obtains y 1 − y2 y(x) = (x − x2 ) + y2 . .20) x1 − x2 Exercise 1. .1.e. circles whose centers lie at the center of the sphere. the so-called action integral. t) is the so-called Lagrangian M ˙ L(q(t). t) (1. This in turn yields y(x) = ax + b. hence. . 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 . t1 S[q(t)] = t0 ˙ dt L(q(t). (1. Velocity–dependent potentials which describe particles moving in electromagnetic fields will be considered below. Using y = dx the condition reads d dx ∂L ∂y = d dx y 1 + (y )2 = 0.21) are extremals of the functional. . y = const.5) with L(y(x). dx (1.

25) = 0 = 0 4π J c = 4πρ = (1.2: Lagrangian 5 For a proof of the Hamiltonian Principle of Least Action we inspect the Euler–Lagrange conditions associated with the action integral defined through (1.29) where ρ(r. Equations (1. t) = V (r. t) 1∂ K(r. t) and B(r.22. 1.24) ∂qj dt ∂ qj ˙ ∂qj dt which are obviously equivalent to the Newtonian equations of motion.26) (1.25) of a particle moving in the field.32) (1. These conditions read in the present case ∂L d ∂L ∂U d − = 0 → − − (mj qj ) = 0 ˙ (1.27) (1. c ∂t (1. t) .31) = − V − Gauge Symmetry of the Electromagnetic Field It is well known that the relationship between fields and potentials (1. t) and a vector potential A(r. t) − K(r. t).31) allows one to transform the potentials without affecting the fields and without affecting the equations of motion (1.30) (1. x3 (t)) moving in an electromagnetic field described through the electrical and magnetic field components E(r. The transformation which leaves the fields invariant is A (r.28) (1. t) and B(r. t) + v × B(r.30. t) is the Lorentz force. t) + V (r.23). t) = q E(r. t) describes the charge density present in the field and J(r. The equations of motion for such a particle are d q ˙ (mr) = F (r. dt The fields E(r. t) dt c where dr = v and where F (r. respectively.33) .1. t) as follows B E = ×A 1 ∂A . F (r.28) can be satisfied implicitly if one represents the fields through a scalar potential V (r. 1. t) describes the charge current density. t) c ∂t (1.27) and (1. x2 (t). t) obey the Maxwell equations ×E + ·B ×B − ·E 1 ∂ c ∂t E 1 ∂ c ∂t B (1. Particle Moving in an Electromagnetic Field We will now consider the Newtonian equations of motion for a single particle of charge q with a trajectory r(t) = (x1 (t). t) = A(r.

36) This expression allows us to show that (1. (1. 1. . 1.30). t) · v .38.38) ∂L ∂ x1 ˙ − ∂L = 0.37) The results (1.35) is equivalent to the Euler–Lagrange condition d dt The second term in (1.40.39) 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 (1.40) (1.36) in the Newtonian equation of motion. (1. ∂x1 (1. e.41) which is identical to the term (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 . dt ∂x1 c We notice using (1. in fact.34) For this purpose we consider only one component of the equation of motion (1. namely.35) − ∂A1 ∂x2 [v × B]1 = x2 B3 − x3 B2 = x2 ˙ ˙ ˙ ∂A2 ∂A1 − ∂x1 ∂x2 − x3 ˙ ∂A1 ∂A3 − ∂x3 ∂x1 .41) with (1.37) is ∂L ∂V q = −q + ∂x1 ∂x1 c The first term in (1. Comparing then (1. (1. t) = 1 q mv 2 − q V (r.39) ∂A1 ∂A2 ∂A3 x1 + ˙ x2 + ˙ x3 ˙ ∂x1 ∂x1 ∂x1 .35) shows that the Newtonian equations of motion and the Euler–Lagrange conditions are.25) follows from the Hamiltonian Principle of Least Action.. if one assumes for a particle the Lagrangian ˙ L(r.g. equivalent. 2 c (1. t) + A(r. B3 = ∂A2 ∂x1 (1.25). d ∂V q (mv1 ) = F1 = −q + [v × B]1 .6 Lagrangian of Particle Moving in Electromagnetic Field Lagrangian Mechanics We want to show now that the equation of motion (1. r.

32.32. q. q. equivalent to the gauge transformation (1. Note that one adds the total time derivative of a function K(r. and since these properties allow one often to simplify mathematical descriptions. appear in a different.32. t) . 1. i. This term is d ∂K ∂K ∂K ∂K ∂K K(r.e. We will consider in the following two symmetries. q.33) of electyromagnetic potentials.33) of electromagnetic potentials. t) c t1 t0 = S[q(t)] + (1.34) q 1 q d ˙ L (r.3: Symmetry Properties 7 1. t) c t1 t0 t1 t0 ˙ dtL(q. t) + A (r. t) · v . a term not affected by the variations allowed.45) . gauge symmetry and symmetries with respect to spatial transformations. t) − + 2 c ∂t c 1 q = mv 2 − q V (r. 2 c c dt Inserting (1.32. t) = mv 2 − q V (r. t) + dt c (1. One can conclude then immediately that any extremal of S [q(t)] is also an extremal of S[q(t)]. t) · v + K(r. t) + q K(q.46) (1.33) ˙ L (r. 1. Eq.33) of electromagnetic potentials. in fact. t) = x1 + ˙ x2 + ˙ x3 + ˙ = ( K) · v + .43) into (1. t) = L(q. t) L (q. t) 1 1 ∂K q mv 2 − q V (r. encountered above in connection with the transformations (1.42) This transformation is termed gauge transformation.44) implies that the gauge transformation amounts to adding a constant term to the action integral. (1. r. The factor q has been introduced to make this c transformation equivalent to the gauge transformation (1. t) = q K(q. dt ∂x1 ∂x2 ∂x3 ∂t ∂t (1. The gauge symmetry. 1. They are the subject of the following theorem. t) the Lagrangian. Theorem: Gauge Transformation of Lagrangian The equation of motion (Euler–Lagrange conditions) of a classical mechanical system are unaffected by the following transformation of its Lagrangian d q ˙ ˙ K(q. t) + K ·v (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.44) Since the condition δq(t0 ) = δq(t1 ) = 0 holds for the variational functions of Lagrangian Mechanics. t) + A(r. 2 c = A(r.42) is. 1.1. q.. surprisingly simple fashion in Lagrangian Mechanics.43) To prove this theorem we determine the action integral corresponding to the transformed Lagrangian t1 S [q(t)] = t0 ˙ dtL (q. r. We want to demonstrate now that the transformation (1.45) and reordering terms yields using (1. For this purpose we consider the transformation of the single particle Lagrangian (1.

52) x3 0 0 1 x3 .47) (1.48) In the following we will denote unit vectors as a. t). the equivalence of (1. In this case the transformation can be written in matrix form      cos −sin 0 x1 x1  x2  =  sin cos 0   x2  (1. R(r) = through ∂r ∂ (1. i. We consider now invariance properties connected with coordinate transformations. the transformation (1. ˆ (1. ˆ (1.42) and (1. ) .51) We would like to derive this transformation in a somewhat complicated. for example. In case of a translation transformation in the direction e nothing new is gained. r ∈ R .49) =0 ∈ R (1. t) by gauge transformed potentials V (r. in the case of central force fields which are invariant with respect to rotations of coordinates around the center. However.32. r. The transformations considered are specified in the following definition.e. t). A(r. t)..8 Lagrangian Mechanics Obviously. and for the introduction of infinitesimal transformations which will provide a crucial method for the study of symmetry in Quantum Mechanics. 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. Definition: Infinitesimal One-Parameter Coordinate Transformations A one-parameter coordinate transformation is decribed through r = r (r.50) A non-trivial example is furnished by the infinitesimal rotation around axis e ˆ r = r + e×r . for the connection between invariance properties and constants of motion. ˆ ˆ ˆ Examples of Infinitesimal Transformations The beauty of infinitesimal transformations is that they can be stated in a very simple manner. therefore. A (r. we like to ˆ provide the transformation here anyway for later reference r = r + e. but nevertheless instructive way considering rotations around the x3 –axis. for such vectors holds a · a = 1. ) = r + R(r) + O( 2 ) . 1. Such invariance properties are very familiar. where the origin of is chosen such that r (r. The corresponding infinitesimal transformation is defined for small r (r. which has an important analogy in Quantum Mechanics.42) corresponds to replacing in the Lagrangian potentials V (r.33). The following description of spatial symmetry is important in two respects. The transformations we consider are the most simple kind. 0) = r . We have proven.

Theorem: Noether’s Theorem ˙ If L(q. ˙ We have generalized in this theorem the definition of infinitesimal coordinate transformation to M –dimensional vectors q.k=1 ∂L ∂Qj qk ˙ ∂ qj ∂qk ˙ (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) (1.55) (1. Invariance implies L = L. by Noether. r) which is constant along the classical path of the system. (1.54) which is equivalent to (1.57) must cancel each other or both vanish. i.e.51) reads r = r − x2 e1 + x1 e2 ˆ ˆ (1. q.58) From this follows the statement of the theorem. q. then M ∂L j=1 Qj ∂ xj is a constant of motion. Before the embark on this theorem we will comment on what is meant by the statement that a classical mechanical system is invariant under a coordinate transformation. neglecting terms of order O( M ˙ ˙ L (q. In order to prove Noether’s theorem we note qj qj ˙ = qj + = qj + ˙ k=1 Qj (q) M (1. t) is invariant with respect to an infinitesimal transformation q = q + Q(q).. expansion. namely for fields rather than only for particle motion. a classical mechanical system is invariant with respect to a coordinate transformation ˙ a constant of motion exists.56) ∂Qj qk .57) ∂ d where we used dt Qj = M ( ∂qk Qj )qk . We have used here the notation corresponding to single particle motion. the second and third term ˙ k=1 in (1.. . The property has been shown to hold in a more general context. a quantity C(r.53). q. t) yields after Taylor ˙ 2 ).3: Symmetry Properties 9 In case of small this transformation can be written neglecting terms O( 2 ) using cos = 1 + O( 2 ). t) = L(q. t) + j=1 ∂L Qj + ∂qj M j. Using the fact.1. i.57)   M Qj j=1 d dt ∂L ∂ qj ˙ + ∂L d Qj ∂ qj dt ˙ = d  dt M Qj j=1 ∂L  = 0 ∂ qj ˙ (1. We consider here only the ‘particle version’ of the theorem. In the context of Lagrangian Mechanics this implies that such transformation leaves the Lagrangian of the system unchanged. ˙ ∂qk ˙ Inserting these infinitesimal changes of qj and qj into the Lagrangian L(q. q. however. sin = + O( 2 )        0 − 0 x1 x1 x1  x2  =  x2  +  0 0   x2  + O( 2 ) .e. Anytime. the property holds for any system. 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.

3 and. v) = 1 mv 2 − U (r). Noether’s theorem yields ˆ ˆ the constant of motion (qj = xj . ˆ ˙ ˆ j=1 (1. In this case we will use the same notation as in (1.50). is conserved. j = 1.59) yields the constant of motion ˆ e (ˆ × r) · mv.51). 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.59).59) We obtain the well known result that in this case the momentum in the direction. It is not necessary to investigate the consequences of global. i. . 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. A calculation similar to that in (1.e. We will now investigate the consequence of rotational invariance as described according to the infinitesimal transformation (1. j = 1.10 Application of Noether’s Theorem Lagrangian Mechanics We consider briefly two examples of invariances with respect to coordinate transformations for the Lagrangian L(r. not infinitesimal transformations. It was. for which translational invariance holds. except using now Qj = ej · (ˆ × r). of course. 2. to be expected that this is the constant of motion. 2 We first determine the constant of motion in case of invariance with respect to translations as defined in (1. In this case we have Qj = ej · e. 3) 3 Qj j=1 ∂L = e· ˆ ∂ xj ˙ 3 ej mxj = e · mv . hence. 2.

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

t ) φ(r. ψ(r0 .. t0 ) . t|r . requires then according to (2.10) We have introduced here a new symbol. (2. t|r0 . t ) φ(r . t1 ]) d3 r φ(r. 8. t). tN −1 |rN −2 . t ∈ (2. .e. t) = Ω d3 r φ(r.9) Employing a continuum of intermediate times t ∈ [t0 . t|r . t ) Ωd 3 r |ψ(r.8) 7.6) φ(r. The generalization of the completeness property to N − 1 intermediate points t > tN −1 > tN −2 > . t ∈ [t0 . The time evolution of ψ(r. t = )|2 = 1 . namely the point on the classical path at time t . t. t|r0 . t|r .4) holds for all times. .12 Quantum Mechanical Path Integral 5. hence. t|r0 ..e. i. t1 ]) Ωd Ωd 3r Ωd 3r Ωd 3r 3 r |ψ(r. t0 ) Ω This relationship has the following interpretation: Assume that at time t0 a particle is generated by a source at one point r0 in space. t ) which requires d3 r φ(r. t ∈ [t0 . tN −2 ) · · · φ(r1 . t0 ) = φ(r. t1 |r0 . t )φ(r. t0 ) = r(t0 )=r0 d[r(t)] Φ[r(t)] . (t > t .8) a knowledge of the state at all space points r ∈ Ω at some intermediate time t . the propagator is unitary. the path integral r(tN )=rN d[r(t)] · · · r(t0 )=r0 (2. t ) t > t . t1 ] (2. t ) ψ(r . The state of a system at time t. tN −1 ) φ(rN −1 .7) t. t ) ψ(r . t|r . .9) for N → ∞. This symbol will be defined further below. (2. Since (2. t1 ] yields an expression of the form r(tN )=rN φ(r.5) 6. described by ψ(r. t ) = δ(r − r ) Ω (2. The definition will actually assume an infinite number of intermediate times and express the path integral through integrals of the type (2. t. i.11) which denotes an integral over all paths r(t) with end points r(t0 ) = r0 and r(tN ) = rN . t0 ) = Ωd 3r N −1 Ωd 3r N −2 ··· Ωd 3r 1 φ(r. This is different from the classical situation where the particle follows a discrete path and. t| rN −1 . at any intermediate time the particle needs only be known at one space point. This must hold for any ψ(r . t ) ψ(r . > t1 > t0 is φ(r. t) is described by a linear map of the type ψ(r. t|r . t0 ) = δ(r − r0 ). The following so-called completeness relationship holds for the propagator (t > t [t0 . t|r . t)|2 = (2. t |r0 .

not only for the classical path. interferences should be much less dramatic than in case of the macroscopic particle.12.13) are built on action integrals for all possible paths. expressions (2. 2 2 (2. Situations which are well described classically will be distinguished through the property that the classical path gives the dominant.15) The exponent of (2.. i.13) L(r. . 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. However.11) is Φ[r(t)] = exp where S[r(t)] is the classical action integral tN 13 i S[r(t)] (2. t) is the Lagrangian of the classical particle. Its value is extremely small in comparision with typical values for action integrals of macroscopic particles. a very large number. 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. t) (2. Only for paths close to the classical path is such interference ruled out.0545 · 10−27 erg s . 2. r.12. in complete distinction from Classical Mechanics.12) is then Scl / ≈ 0. actually often essentially exclusive. the exponent is a very large imaginary number.2: Axioms 9. r. for the electron many paths contribute and the action integrals for non-classical paths need to be known.12) S[r(t)] = t0 ˙ dt L(r.e. it is comparable to action integrals as they arise for microscopic particles under typical circumstances. However.. such that destructive interference of the contributions of such paths does not occur. However. The functional Φ[r(t)] in (2. i.13) and = 1.5 · 1027 .14) has the same dimension as the action integral S[r(t)]. (2.2. for microscopic particles like the electron this is by no means the case. accordingly. Any variations of Scl would then lead to strong oscillations of the contributions exp( i S) to the path integral and one can expect destructive interference betwen these contributions.e. namely due to the property of the classical path to be an extremal of the action integral. 2. The constant given in (2. contribution to the path integral (2. One would still expect significant descructive interference between contributions of different paths since the value calculated is comparable to 2π. The value of S[r(t)] is then Scl = 1 1 m v 2 t = erg s .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. Since this number is multiplied by ‘i’. However.13). This implies that small variations of the path near the classical path alter the value of the action integral by very little. In case of a proton with mass m = 1. The situation is very different for microscopic particles. Scl / ≈ 8.14) ˙ In (2.

(xN . ∞[. t) = ˙ 1 mx2 − U (x) . 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 . 2 2m N (2. .16) In order to define the path integral we assume. For the sake of simplicity we will consider the case of particles moving in one dimension labelled by the position coordinate x. provides an avenue to evaluate path integrals.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. (2.9). respectively. at the same time. t1 ). The particles have associated with them a Lagrangian L(x.20) can be properly taken. 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. however. tN )} . . etc. (xN −1 . .12.14 Quantum Mechanical Path Integral 2. tN −1 ). namely 1 m(xj+1 − xj )2 / 2 and U (xj ). 2.3 How to Evaluate the Path Integral In this section we will provide an explicit algorithm which defines the path integral (2.21) N 2. . tj+1 ) we assume that kinetic energy and potential energy are constant. . Its value is CN = m 2πi N 2 (2.10) and (2. as in (2. The spacings between the times tj+1 and tj will all be identical.13) and. x. x2 . ˙ 2 (2.20) Here. This allows us to write the evolution operator using (2. (2. In passing from one space–time instance (xj . tj ) to the next (xj+1 . (2. . . tN |x0 .19) The main idea is that one can replace the path integral now by a multiple integral over x1 .17) The discretization in time leads to a discretization of the paths x(t) which will be represented through the series of space–time points {(x0 . t0 ). the xj values are not. (x1 . > t1 > t0 letting N go to infinity later. t0 ) = limN →∞ CN exp i N N −1 j=0 +∞ +∞ +∞ −∞ dx1 −∞ dx2 .12) φ(xN . . a series of times tN > tN −1 > tN −2 > . −∞ dxN −1 (xj+1 −xj )2 1 − U (xj ) .18) The time instances are fixed. namely tj+1 − tj = (tN − t0 )/N = N . They can be anywhere in the allowed volume which we will choose to be the interval ] − ∞.

x(tN ) = xN ] is S[x(t)|x(t0 ) = x0 . 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. which can also be written φ(xN . using (2. tN dt t0 1 1 (xN − x0 )2 m x2 = m ˙ cl . due to (2. x(tN ) = 0] .12) yields x(tN )=0 im (xN − x0 )2 i φ(xN .22) where xcl (t) is the classical path which connects the space–time points (x0 . x(tN ) = xN ] 1 (xN − x0 )2 m + 0 + 2 tN − t 0 + S[x(t)|x(t0 ) = 0.4: Propagator for a Free Particle 15 Rather then using the integration variables xj . = (2.23). The ˙ 1 S[x(t)|x(t ) = x . since x(t0 ) = xcl (t0 ) = x0 and x(tN ) = xcl (tN ) = xN . tN ) is to be evaluated. tN |x0 . The resulting expression for S[x(t)|x(t0 ) = x0 .24) implies for the second term on the r. the origin of which coincides with the classical path of the particle. One obtains S[x(t)|x(t0 ) = x0 . 2 2 tN − t0 (2. (2.26) The first term on the r.h. ˙ t0 (2. (2.. tN |0. xN − x0 xcl (t) = x0 + ( t − t0 ) .e. Inserting the result into (2. tN ). can be expressed through a path integral with endpoints x(t0 ) = 0. it holds y(t0 ) = y(tN ) = 0 .s. .28) i. ˙ ˙ ˙ 2 + y2) = ˙ (2. 2.25) + The condition (2. t0 ) and (xN . tN dt y = y(tN ) − y(t0 ) = 0 . To see the benifit of such approach we define a path y(t) as follows x(t) = xcl (t) + y(t) (2. t0 ) = exp 1 im (xN − x0 )2 2 tN − t0 φ(0.2. x(tN ) = 0. t0 ) (2.h.31) We have denoted explicitly that the action integral for a path connecting the space–time points (x0 .25) is. t0 ) = exp S[x(t)] (2.s. ˙ 2 t0 (2.30) d[x(t)] exp 2 tN − t0 x(t0 )=0 a result. 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 .10. it is more suitable to define new integration variables yj .27) The third term can be written in the notation introduced tN 1 dt my 2 = S[x(t)|x(t0 ) = 0. tN |x0 .24) Also we use the fact that the velocity of the classical path xcl = (xN −x0 )/(tn −t0 ) is constant. namely. x(tN ) = 0] .24).13).29) This expression corresponds to the action integral in (2. of (2.23) tN − t0 It is essential for the following to note that. t0 ) and (xN .

35) must be determined.  .36) which holds for a d-dimensional. (2.  . In the appendix we prove  +∞ +∞ d j. . (2. symmetric matrix (bjk ) and det(bjk ) = 0.k=1 yj ajk yk   (2. 2 N  .38) . .  0 0 0 0 0 0 The following integral +∞ +∞ (2.  . t0 ) = limN →∞ × +∞ +∞ −∞ dy1 · · · −∞ dyN −1 exp i N m 2πi N 2 N × (2. (2.33) where ajk are the elements of the following symmetric (N  2 −1 0  −1 2 −1  2 m  0 −1  ajk =  . . 0 0   . .k=1 yj ajk yk − 1) × (N − 1) matrix  . ..32) we split off the factor 2 mN in the definition (2. . 2 −1  −1 2 (2.32) (yj+1 − yj )2 N −1 1 2 j=0 2 m N The exponent E can be written.. . as the quadratic form E = im 2 2 ( 2y1 − y1 y2 − y2 y1 + 2y2 − y2 y3 − y3 y2 2 N 2 2 + 2y3 − · · · − yN −2 yN −1 − yN −1 yN −2 + 2yN −1 ) N −1 = i j. In order to complete the evaluation of (2.34) of (ajk ) defining a new matrix (Ajk ) through ajk = Using det(ajk ) = m 2 N N −1 m 2 N Ajk . ..34) dy1 · · · −∞ −∞ dyN −1 exp i  N −1 j.. . 0 0   . 0 0 ..  . real.37) det(Ajk ) .16 Evaluation of the necessary path integral Quantum Mechanical Path Integral To determine the propagator (2.. tN |0..31) for a free particle one needs to evaluate the following path integral φ(0. noting y0 = yN = 0.. .k=1 dy1 · · · −∞ −∞ dyN −1 exp i yj bjk yk  =  (iπ)d det(bjk ) 1 2 .. .

41) D2 = 2 −1 −1 2 = 3 (2.31) and obtain.4: Propagator for a Free Particle a property which follows from det(cB) = cn detB for any n × n matrix B.. t0 ) = m 2πi (tN − t0 ) 1 2 .42) m 2πi NN 1 2 (2. t0 ) = m 2πi (t − t0 ) 1 2 exp im (x − x0 )2 2 t − t0 . which follows from (2. One can readily verify that this procedure leads to the following recursion equation for the determinants Dn = 2 Dn−1 − Dn−2 . .46) .39) In order to determine det(Ajk ) we consider the dimension n of (Ajk )..40) in terms of subdeterminants along the last column. variable... 0 0     0 −1 2 . t0 ) = limN →∞ and with NN (2. defining t = tN . t0 ) = limN →∞ m 2πi N 2 17 N 2πi m N N −1 2 1 .  . To solve this three term recursion relationship one needs two starting values.41) the same result for D2 . (2. D1 = 2 and obtain from (2. 0 0    Dn =  . . (2. let say n. (2. . one can readily verify Dn = n + 1 .. . x = xN φ(x. .. det(Ajk ) (2. . . . . . we obtain φ(0.40) . .45) Expressions for Free Particle Propagator We have now collected all pieces for the final expression of the propagator (2. We seek then to evaluate the determinant of the n × n matrix   2 −1 0 . tN |0. . . . presently N − 1.43) We like to note here for further use below that one might as well employ the ‘artificial’ starting values D0 = 1. . .2.   .44) = tN − t0 . tN |0. (2. Inserting this into (2.39) yields φ(0. . . D3 . 0 0  −1 2 −1 . Our derivation has provided us with the value det(Ajk ) = N .18) we obtain φ(0. 2 −1  0 0 0 −1 2 For this purpose we expand (2. t|x0 . . .. 2.. tN |0. . n = 1.    0 0 0 . Using D1 = |(2)| = 2 .. .

48) Obviously.5) allows us to predict the time evolution of any state function ψ(x. x) + E(x) in (2. according to (2. t0 )| = 2 1 πδ 2 1 2 exp − x2 0 δ2 (2.51) For this evaluation we adopt the strategy of combining in the exponential the terms quadratic (∼ x2 ) and linear (∼ x0 ) in the integration variable to a complete square 0 ax2 + 2bx0 = a 0 x0 + b a 2 − b2 a (2. centered around x0 = 0.46). We will obtain. t0 ) = 1 πδ 2 1 4 exp − x2 po 0 +i x 2 2δ (2. the wave function of the particle at later times.10) φ(r.48) as the initial state using the propagator (2. (2. We need to evaluate for this purpose the integral ψ(x.53) . x) + E(x) (2. and describes a single particle since 1 πδ 2 1 2 +∞ dx0 exp −∞ − x2 0 δ2 = 1. x) = im 2 t x2 1 + i o t mδ 2 − 2xo x − po t m + f (x) (2. t) of a free particle.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 .46) can be generalized to three dimensions in a rather obvious way.49) is Gaussian of width δ. t0 ) = m 2πi (t − t0 ) 3 2 exp im (r − r0 )2 2 t − t0 .247).52) and applying (2. thereby.50) One refers to such states as wave packets. the associated probability distribution |ψ(x0 .5) to (2. Below we will apply this to a particle at rest and a particle forming a so-called wave packet.18 Quantum Mechanical Path Integral This propagator. One obtains then for the propagator (2. We devide the contributions to the exponent Eo (xo . t|r0 . (2. t) = 1 πδ 2 1 4 m 2πi t 1 2 +∞ dx0 exp −∞ im (x − x0 )2 x2 po − 02 + i xo 2 t 2δ Eo (xo . The result (2.51) as follows Eo (xo . We want to apply axiom (2.

(2.52). according to (2. We proceed as follows.4: Propagator for a Free Particle im x2 − f (x) .60) is then represented by real ρ values. x) = One can write then (2.x)   im  xo 2 t  t − mδ 2 x− 1+ x− 1+ po m = −∞ +∞ dx0 exp  1+i t 2    2 t i mδ2 = −∞ im dx0 exp  2 t t 1+i 2 mδ xo − po m t t i mδ2  . the square in (2. making certain.51) ψ(x. t) = 1 πδ 2 1 4 2 t 1 + i mδ2  .2. (2. z2 = +∞ × i 1−i t mδ 2 (2.56) m 2πi t 1 2 +∞ eE(x) −∞ dx0 eEo (xo . however.57) and needs to determine the integral +∞ I = −∞ +∞ dx0 eEo (xo .53) E(x) =  x− po mt 19 (2.58) The integrand is an analytical function everywhere in the complex plane and we can alter the integration path.59) An integration path in the complex x0 –plane along the direction i 1−i t mδ 2 (2. We consider a transformation to a new integration variable ρ defined through i 1−i t mδ 2 ρ = x0 − x− 1+ po m t t i mδ2 . (2. The beginning and the end of such path are the points z1 = −∞ × i 1−i t mδ 2 .54) f (x) =  This yields Eo (xo .55) im  xo 2 t  1+i t − mδ 2 x− 1+ po m t t i mδ2  . 2 t One chooses then f (x) to complete. that the new path does not lead to additional contributions to the integral.x) (2. 2 (2.61) .

20 whereas the original path in (2.58) has the end points z1 = −∞ . the paths between z1 → z1 and z2 → z2 have a real part of x0 of ±∞.69) t)2 t 2δ 2 (1 + i mδ2 ) + i po x − i p2 o t 2m (2.57) reads then 1 πδ 2 1 4 2πi t .62) If one can show that an integration of (2.58) by t mδ 2 +∞ I = i 1−i dρ exp − −∞ m 2 t 1+ t mδ 2 2 ρ2 (2. z2 = +∞ .58) vanishes 0 for Re x0 → ±∞.66) yields ψ(x. one can show that z1 lies at −∞ − i × ∞ and z2 at +∞ + i × ∞. t m(1 + i mδ2 ) (2.68) and. Hence. 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 (2.64) ψ(x. 1+b 1+b finally E(x) = − (x − po m (2. Quantum Mechanical Path Integral (2. We can conclude then that (2. (2.63) holds and. In fact. 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) ] . (2.54) using (2.63) which can be readily evaluated.65) t 1 − i mδ2 t 1 + i mδ2 . (2.70) . using a ab = a − .55). accordingly.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. Since the exponent in (2. I = Equation (2. t) = Seperating the phase factor 1 t 1 + i mδ2 1 4 1 2 exp [ E(x) ] .58) has a leading contribution in x0 of −x2 /δ 2 the integrand of (2.

48)] to = 0.74) remains invariant in time. the case of arbitrary to is recovered iby replacing t → to in (2. This yields.. However. coinciding at t = 0 with the width of the initial distribution (2. t0 ) = exp i po i p2 o xo − to m 2m .72) 2 t2 δ 1+ 2 4 (2.75) i.72). The center of the distribution (2.71) conserves the phase factor exp[i(po / )x] of the original wave function (2. t) = exp i i p2 po o x − (t − to ) m 2m . the spatial dependence of the initial state (2.71. We had assumed above [c. In this case holds ψ(x.4: Propagator for a Free Particle which inserted in (2.77) . instead of (2.48) for δ → ∞. 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 . t0 ) = exp i po xo m . Another interesting observation is that the wave function (2. respectively. The corresponding probability distribution is |ψ(x.72) moves in the direction of the positive x-axis with velocity vo = po /m which identifies po as the momentum of the particle. (2.75) ψ(x. t)| = 2 1 πδ 2 (1 + 2 t2 m2 δ 4 1 2 ) exp − (x − δ 2 (1 + po m t)2 ) 2 t2 m2 δ 4 . 2. t) = t 1 − i mδ2 t 1 + i mδ2 1 4 21 1 πδ 2 (1 + 2 t2 ) m2 δ 4 1 4 × po i p2 t o ) + i x − t mδ 2 2m (2.71) × exp − (x − po m t)2 ) 2δ 2 (1 + 2 t2 m2 δ 4 (1 − i . The width of the distribution (2.72) for the initial state (2.72) Comparision of Moving Wave Packet with Classical Motion It is revealing to compare the probability distributions (2.2.f.49).74) which corresponds to (2. t) = exp i po i p2 o x − t m 2m .48) and for the final state (2. (2. The conservation of this factor is particularly striking for the (unnormalized) initial wave function ψ(x0 . (2.e.49).72). (2.73) m δ increases with time. (2. This ‘spreading’ of the wave function is a genuine quantum phenomenon. (2. (2.67) provides the complete expression of the wave function at all times t ψ(x.71).76) From this we conclude that an initial wave function ψ(xo .48) and that the respective phase factor is related with the velocity of the classical particle and of the center of the distribution (2.

83) (2. One refers to the respective states as stationary states. the spatial as well as the temporal dependence of the wave function remains invariant in this case. t) = exp i Quantum Mechanical Path Integral po i p2 o x − t m 2m . xcl + y(t). 2. t) + δL ˙ ˙ ˙ ˙ where L(xcl . t) + L (y. t) ˙ δL 1 1 mx2 − c(t)x2 − e(t)xcl ˙ cl cl 2 2 1 1 = my 2 − c(t)y 2 ˙ 2 2 = mxcl y(t) − c(t)xcl y − e(t)y .5 Propagator for a Quadratic Lagrangian x(tN )=xN We will now determine the propagator (2.79) for a quadratic Lagrangian L(x. For this purpose we use d xcl y = ˙ ˙ (xcl y) − xcl y ˙ ¨ (2. xcl .10.84) . y(t).e. Obviously.78) i. t) = ˙ 1 1 mx2 − c(t)x2 − e(t)x . ˙ 2 2 tN (2. y(tN ) = 0 .81) for an arbitrary path x(t) with end points x(t0 ) = x0 and x(tN ) = xN . Such states play a cardinal role in quantum mechanics. x. (2.12. t) = L(xcl . the end points of y(t) are y(t0 ) = 0 Inserting (2. y(t). (2.22 becomes at t > to ψ(x..13) φ(xN . 2. t0 ) = x(t0 )=x0 d[x(t)] exp i S[x(t)] (2. 2 . t) ˙ (2. 1.81) vanishes2 .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 .82) one obtains L(xcl + y.82) which describes the deviation from the classical path xcl (t) with end points xcl (t0 ) = x0 and xcl (tN ) = xN . x. In order to simplify this task we define again a new path y(t) x(t) = xcl (t) + y(t) (2.80) For this purpose we need to determine the action integral S[x(t)] = t0 dt L(x.85) We want to show now that the contribution of δL to the action integral (2. tN |x0 .80) into (2. 2. ˙ ˙ = (2. t) ˙ L (y. xcl . y(t)] which vanishes according to the Hamiltonian principle as discussed in Sect.

. .  . tN |0.  0 0 0 . x t0 (2. . 2 −1  0 0 0 −1 2   c1 0 0 . .. t0 ) (2. .   . .88) vanishes.88) and. . . . 0 0  −1 2 −1 .k=1 yj ajk yk (2. also the second contribution on the r. .91) through the quadratic N −1 E = i j. hence. y.  . The discretization scheme adopted above yields in the present case ˜ φ(0.93) .s. . . One can then express the propagator (2. (2. 2 .2. . 0 0   N  0 −  . . tN |0.. .87) According to (2.   . . . . .h. .32). vanishes. .. .. cN −2 0  0 0 0 0 cN −1 (2. 0 0     0 c3 .24) to the Lagrangian (2. One can express the exponent E in (2.s. . t0 ) we will adopt a strategy which is similar to that used for the evaluation of (2.5: Propagator for a Quadratic Lagrangian and obtain tN t0 23 dt δL = m [xcl y]|t0 − ˙ t N tN dt [ m¨cl (t) + c(t) xcl (t) + e(t) ] y(t) ..80) yields for the classical path m¨cl + c(t) xcl + e(t) = 0 x (2.91) (yj+1 − yj )2 1 2 2m N where cj = c(tj ). . Applying the Euler–Lagrange conditions (1.31) for the free particle propagator. 0 0  m   ajk =  . tN |x0 . t0 ) = exp S[xcl (t)] φ(0. . . . .79) i ˜ φ(xN . . 0 0     0 −1 2 . . .92) where ajk are the elements of the following (N − 1) × (N − 1) matrix   2 −1 0 . . 0 0  0 c2 0 . tN |0. .h.89) where ˜ φ(0. t0 ) = limN →∞ × +∞ +∞ −∞ dy1 · · · −∞ dyN −1 exp N i N N −1 j=0 m 2πi N 2 N × − 1 2 cj 2 yj (2. . For the evaluation of φ(0.    0 0 0 . . t) ˙ t0 .83) the first term on the r.  2 N  . of (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. tj = t0 + form j. tN |0.. . t0 ) = y(tN )=0 d[y(t)] exp y(t0 )=0 i tN dt L (y.. .

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

187) or 1 ˜ exp[−iω(n + 2 ) t] φn (x) +∞ = −∞ ˜ dxo φ(x. .183) where the expansion coefficients are functions of x.. t|xo . xo . .186) Fourier transform.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) = n=0 1 ˜ exp[ − iω(n + 1 ) t ] φn (x) 2 n! (2. 2. to ) exp[−iω(n + 1 ) to ] φn (xo ) . Accoordingly. to ) we express (2. i. t − to ) 1 ˜ exp[−iω(m + 1 ) to ] φm (xo ) 2 m! (2. n = 1. 1 exp[iω(n + 2 ) to ] · · · .36 Altogether. Noting that the propagator (2.172) in the form ∞ n=0 (2. t − to ) = φ(x. We want to inspect the consequences of the invariance property (2. In contrast to W (x.172). one can expand (2. 2. xo .172) is a function of t − to and defining accordingly Φ(x.171) ∞ W (x. to ).172) I = exp 2 mω x e−iωt − e−2iωt − Quantum Mechanical Path Integral mω 2 1 x − iωt 2 2 .171) can be expanded in terms of e−inωt . . . t) = exp[−iω(n + 1 ) t] φn (x) as invariants under the 2 action of the propagator φ(x. t − to ) 1 ˜ φn (xo ) n! (2.h. (2. but not of t. results in 1 ˜ exp[−iω(n + 1 ) t] φn (x) 2 n! +∞ = −∞ dxo Φ(x.. one obtains for the r.184) 1 ˜ exp[−iω(n + 1 ) t] φn (x) 2 n! ∞ +∞ = m=0 −∞ dxo Φ(x.188) ˜ ˜ Equation (2. t) +∞ −∞ dto 1 ˜ exp[−iω(m + 1 ) to ] φm (xo ) 2 m! (2.172). t|xo . We note that the factor exp(2 mω/ xe−iωt − e−2iωt ) in (2. xo .171) concludes the proof of (2. t). which also exhibits such invariance.171. 2 (2.s.147) in (2. of (2. t|xo . xo .e.188) identifies the functions ψn (x.182) Comparision with (2.

2.5: Propagator for a Quadratic Lagrangian 37 ˜ ˜ the functions ψn (x.196) where Hn (y) = ey 2 /2 (2.192) w(y.183). through the expansion coefficients φn (x) in (2.195) zn Hn (y) n! ˜ φn (y) . In the following we will characterize the functions φn (x) and determine the normalization constants Nn . . z) = Hn (y) (2. z) z 2 e−y or w(y. We will also argue that the functions ψn (x.197) The expansion coefficients Hn (y) in (2. To obtain ˜n (x) we simplify the expansion (2.189) ˜ dx φ2 (x) = 1 n (2.198) ∂z n z=0 . t)|2 = Nn +∞ −∞ n = 0. in a ‘nutshell’ all information on the Hermite polynomials. 1. . t) are associated with a time-independent probablity density |ψn (x. according to (2.193) (2. t) = z 2 e−y where w(y. t) = exp[−iω(n + 1 ) t] Nn φn (x) . 2. ˜ |φ stationary wave functions ψn (x. t) such that +∞ −∞ 2 dx |ψ(x.190) ˜ is obeyed. 2 Here Nn are constants which normalize ψn (x.183).171) W (x.197) are called Hermite polynomials which are polynomials of degree n which will be evaluated below. . we have identified then. z) (2.194) plays a central role for the Hermite polynomials since it contains.183).191) (2. t) of the quantum mechanical harmonic oscillator ˜ ψn (x. Actually. characterizes the wave functions φn (x). (2. (2.183) reads then w(y. This follows from ∂n w(y. through expansion (2. t) provide all stationary states of the quantum mechanical harmonic oscillator. Expression (2.171). For this purpose we introduce first closed expressions for φ the new variables y z and write (2. t)|2 = ˜n (x)|2 . z) = n=0 1 2 /2 1 2 /2 = = mω e−iωt x (2. The Hermite Polynomials ˜ The function (2. Expansion (2. z) = exp(2yz − z 2 ) .194) = z2 n=0 ∞ 1 ∞ zn ˜ φn (y) n! (2.194).

normalization factors for φ(y). As will become evident in the present case generating functions provide an extremely elegant access to the special functions of Mathematical Physics3 .1: Schematic representation of change of summation variables ν and µ to n = ν + µ and m = ν −µ. Reading. and O. µ lattice (left diagram) corresponds to a summation over only every other point in an n. ν     H2 (y) = 4y 2 − 2. and recursion equations for the efficient evaluation of Hn (y).201) ν−µ     µ µ ν−µ Figure 2.38 Quantum Mechanical Path Integral which is a direct consequence of (2. (2..202) generatingfunctionology by H.198) for the Hermite polynomials can be expressed in a more convenient form employing definition (2.196) ∂n w(y. The diagrams also identify the areas over which the summation is to be carried out. D.199) Comparision with (2. 1989).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). Massachusetts.Wilf (Academic Press. For this purpose we expand the generating function (2.E. . The diagrams illustrate that a summation over all points in a ν. Inc. m lattice (right diagram).196) results in the so-called Rodrigues formula for the Hermite polynomials Hn (y) = (−1)n ey 2 ∂ n −y2 e ..196). . closed expressions for Hn (y). One calls w(y. that Hn (y) is a polynomial of degree n. z) the generating function for the Hermite polynomials.S.Patashnik (AddisonWesley.Graham. The identity (2. ∂y n (2. 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. We want to derive now explicit expressions for the Hermite polynomials. We will employ (2. H1 (y) = 2y. .200) One can deduce from this expression the polynomial character of Hn (y).196) to ˜ derive. Boston.L. ν+µ   (2. 1990) is a useful introduction to this tool as is a chapter in the eminently useful Concrete Mathematics by R.200) yields for the first Hermite polynomials H0 (y) = 1. ν+µ ν   H3 (y) = 8y 3 − 12y.Knuth. among other properties. 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.e. . i. 2.194.

k! (n − 2k)! (2. 0 ≤ k ≤ [n/2] where [x] denotes the largest integer p. it contains only odd powers. µ are integers the summation over n.194) holds w(−y. (2. 2 (2. In case of even n . (2. With this restriction in mind one can express (2.203) The old summation variables ν. z) = w(y. for odd n. it holds Hn (−y) = (−1)n Hn (y) . µ expressend in terms of n. the sum in (2. Hence. 2 µ = n−m .204) Since ν. in fact. m must be restricted such that either both n and m are even or both n and m are odd. . 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 . From (2. otherwise. k! (n − 2k)! = Hn (y) (2. −z) and.202) e 2yz − z 2 ∞ = n=0 zn n! ≤n m≥0 n! (−1) n−m 2 n−m 2 ! m! (2y)m . i. 0 ≤ m ≤ n . 2.207) one can deduce that Hn (y).208) which agrees with the expressions in (2.197) ∞ n=0 zn Hn (−y) = n! ∞ n=0 (−z)n Hn (y) = n! ∞ n=0 zn (−1)n Hn (y) n! (2. p ≤ x.209).201). (2. . hence. m are ν = n+m . According to (2.5: Propagator for a Quadratic Lagrangian and introduce now new summation variables n = ν + µ. The lattices representing the summation terms are shown in Fig.207) This expression yields for the first four Hermite polynomials H0 (y) = 1.e.1.207) contains only even powers.206) From this expansion we can identify Hn (y) [n/2] Hn (y) = k=0 (−1)k n! (2y)n−2k .. H3 (y) = 8y 3 − 12y.210) from which one can conclude the property (2. 39 (2.209) This property follows also from the generating function.2. is a polynomial of degree n. H1 (y) = 2y. . according to 2. . H2 (y) = 4y 2 − 2. m = ν − µ 0 ≤ n < ∞.205) Since (n − m)/2 is an integer we can introduce now the summation variable k = (n − m)/2 .

. z) = 0 (2.218) 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. . n = 1.215) ∂z which can be readily verified using (2.233) below. Hn+1 (y) − 2y Hn (y) + 2n Hn−1 (y) = 0 . . For this purpose one considers w(0.196) into the differential equation (2. . (2. .217) gives H1 (y) − 2y H0 (y) = 0. namely. (2. n = 0.209) since odd functions have a node at the origin. .40 Quantum Mechanical Path Integral The generating function allows one to determine the values of Hn (y) at y = 0. . as given in (2. n! (2. . z) ∂ w(y.194).211) m! n! m=0 n=0 Comparing terms on both sides of the equation yields (2n)! . . Hn+1 (y) − 2y Hn (y) + 2n Hn−1 (y) = 0. . z) should be equivalent to the relationship (??). n = 1. Starting point of the derivation is the property of w(y.212) n! This implies that stationary states of the harmonic oscillator φ2n+1 (x). (2. (2. One can then readily state which differential equation of w(y.208). Another relationship is d Hn (y) = 2n Hn−1 (y).216) Combining the sums and collecting terms with identical powers of z ∞ n=1 zn Hn+1 (y) − 2y Hn (y) + 2n Hn−1 (y) n! + H1 (y) − 2yH0 (y) = 0 (2. 2. For Hermite polynomials holds. a factor z reduces the order of Hn by one and introduces also a factor n. a property which is consistent with (2.213) which allow one to evaluate Hn (y) from H0 (y) and H1 (y) given by (2. as defined through (2. z) = exp(−z 2 ) and carries out the Taylor expansion on both sides of this expression resulting in ∞ ∞ (−1)m z 2m zn = Hn (0) . 1. 2. H2n (0) = (−1)n Recursion Relationships A useful set of properties for special functions are the so-called recursion relationships. 2.188.197) above and given by (2. 2. (2. 2. .194).214) We want to derive (??) using the generating function. The reader may verify that w(y.215) yields ∞ n=1 z n−1 Hn (y) − 2y (n − 1)! ∞ n=0 zn Hn (y) + 2 n! ∞ n=0 z n+1 Hn (y) = 0 . . . for example. dy n = 1. z). indeed satisfies the latter relationship. Substituting expansion (2. dw/dy − 2zw = 0. have a node at y = 0. H2n+1 (0) = 0 . z) − (2y − 2z) w(y.

196) the orthogonality properties of the Hermite polynomials.228) .h.s.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 .226a). (2.200) yields Hn (y) = The identity (−1)n y2 √ e π +∞ dt −∞ dn 2iyt − t2 e .194.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. For this purpose we consider the integral +∞ −∞ dy w(y.219) which can be written I(y) = e−y 2 +∞ −∞ dte−(t−iy) = e−y √ 2 2 +∞ −∞ dze−z . 2 √ π e2 z z (2. 1.5: 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 . z ) e−y 2 +∞ = e2 z z = √ ∞ dy e−(y−z−z ) = .h.221) 1 = √ π +∞ −∞ dt e2iyt − t .247) for a = i one obtains I(y) = and.196) yields ∞ n. in the integral representation of the Hermite polynomials 2n (−i)n ey √ Hn (y) = π Orthonormality Properties 2 (2. 2 (2. acording to the definition (2.194. 2 (2. 2.223) dn 2iy t − t2 2 e = (2 i t)n e2iy t − t n dy results. 2.225) We want to derive from the generating function (2.s. 2 (2.2. dy n (2. .224) +∞ −∞ dt tn e2iy t − t .220) Using (2. (2. 2. . of the Rodrigues formula (2. . 2 n = 0. finally.226) π −∞ 2n z n z n n=0 n! Expressing the l.222) Employing this expression now on the r. e−y 2 π e−y 2 (2. z) w(y. through a double series over Hermite polynomials using (2.

the value of the wave function at y = 0 is positive for n = 0. 1. 3.229) Hn (y) .232) Hn (y) . ¡ ¢ ¡ -4   -2 0 2 4 y (2. The normalized states are [c.231) √ π (2.f. negative for n = 2 and vanishes for n = 1. 4 and odd for n = 1.197) holds 2 ˜ φn (y) = e−y /2 Hn (y) . 3. 2. 2. in harmony with (??).233) . that the wave functions are even for n = 0.228) allow us to construct normalized stationary states of the harmonic oscillator.2: Stationary states φn (y) of the harmonic oscillator for n = 0. One can recognize. 2.2. 3.228) +∞ 2 Nn −∞ √ 2 2 2 dy e−y Hn (y) = Nn 2n n! π = 1 1 2n n! 1 2n n! We conclude Nn = and can finally state the explicit form of the normalized stationary states φn (y) = −y √ e π 2 /2 The stationary states (2. (2. (2.189.233) are presented for n = 0. (2. One can also recognize that n is equal to the number of nodes of the wave function. Normalized Stationary States The orthonormality conditions (2. According to (2. 2.42 Quantum Mechanical Path Integral E 9/2 h ω ¥ ¥ ¥ ¥ ¥ 7/2 h ω 5/2 h ω ¤ 3/2 h ω £ 1/2 h ω ¦   Figure 2.190)] φn (y) = Nn e−y 2 /2 and for the normalization constants Nn follows from (2. 4. 4 in Fig.230) (2. Furthermore. 3. 4. in agreement with our above discussions. 1. 2.

g ∈ F .237) for Ω = R. (10. using (2. denoted by Ln (x) and encountered in case of the stationary states of the nonrelativistic [see Sect. 5 below [c. f.237) and where In denotes some constants. 5. i. ??? and eq. w(x) = 0 only at a discrete set of points xk ∈ Ω (2. (2. The latter are written for polynomials pn (x) in the general form dx pn (x) pm (x) w(x) = In δnm Ω (2.e.189) φn (x) = √ 1 2n n! mω π 1 4 e− mωx2 2 Hn ( mω x) .235) Completeness of the Hermite Polynomials The Hermite polynomials are the first members of a large class of special functions which one encounters in the course of describing stationary quantum states for various potentials and in spaces of different dimensions. denoted by P (x) and introduced in Sect.236) where w(x) is a so-called weight function with the property w(x) ≥ 0. The orthogonal polynomials pn mentioned above have the important property that they form a complete basis in the space F of normalizable functions.233) and (2.5: Propagator for a Quadratic Lagrangian 43 The normalization condition (2.e.2.228) shows that the orthonogality condition of the Hermite polynomials is in complience with (2. 5. and In = 2n n! π. 10.228). w(x) = xα e−x .f.234) The explicit form of the stationary states of the harmonic oscillator in terms of the position variable x is then.179] holds Ω = [−1.150 . 5. 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. by dx = dy mω 1 4 . Other examples of orthogonal polynomials are the Legendre and Jacobi polynomials which arise in solving three-dimensional stationary Schr¨dinger equations.231) of the wave functions differs from that postulated in (2. Comparision with (2.236 . In case of the Legendre polynomials. i. 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. w(x) = √ exp(−x2 ).10 and eq. (2. 2.156.238) where the space is endowed with the scalar product (f |g) = Ω dx f (x) g(x) w(x) = < ∞ . In = Γ(n+α+1)/n! where Γ(z) is the so-called Gamma function. ???] and relativistic [see Sect. of functions which obey dx f 2 (x) w(x) = < ∞ . +∞[. (5.151. holds Ω = [0.239) . The Hermite polynomials are so-called orthonogal polynomials since they obey the conditions (2. Ω (2.459] hydrogen atom... In case of the associated Laguerre (α) polynomials.189) by the Jacobian dx/dy. 1]. (2. and I = 2/(2 + 1). w(x) ≡ 1.

The only possibility for this to be true is dn = 0. det(a) (2. If one replaces for all f ∈ F: f (x) → w(x) f (x) and. t) = n=0 1 2 dn exp[−iω(n + )t] e−y /2 Hn (y) . a state for which (2. in particular.172) holds. aT = a. i.. t) which is stationary under the action of the harmonic oscillator propagator (2. the states (2.197) it must hold cn (t) = dn exp[−iω(n + 1 )t] and.k yj ajk yk = (iπ)n . 2 (2. i. symmetric a. .e.e..147).e. (2. pn (x) → w(x) pn (x) the scalar product (2.188. n 2 (2.233). Appendix: Exponential Integral We want to prove +∞ +∞ I = dy1 . i..242) and the existence of a normalizable state ψ(y. Ω (2. hence.239) becomes the conventional scalar product of quantum mechanics f |g = Ω dx f (x) g(x) .246) for det(a) = 0 and real. the 2 stationary state ψ(y. Therefore. 2.e. Since the Hermite polynomials form a complete basis for such states we can expand ∞ ψ(y. −∞ dyn ei −∞ n j. In case of n = 1 this reads +∞ −∞ dx ei a x = 2 iπ .247) .233) exhaust all stationary states of the harmonic oscillator.243) To be consistent with(2. . i. ψ(y.m=0 d∗ dm exp[iω(m − n)t] e−y Hn (y)Hm (y) .44 As a result holds for any f ∈ F f (x) = n Quantum Mechanical Path Integral cn pn (x) (2.245) must be time-independent. except for a single n = no . t) is ∞ ψ(y.236). a (2. t)|2 . t) must be identical to one of the stationary states (2.240) where cn = 1 In dx w(x) f (x) pn (x) . t) = n=0 cn (t) e−y 2 /2 Hn (y) . ∞ n.242) Let us assume now the case of a function space governed by the norm (2.244) For the state to be stationary |ψ(x. (2.241) The latter identity follows from (2.

k n m n y Sj ajk Skm ym ˜ ˜ y (S T ) j ajk Skm ym ˜ ˜ j.. . . i.2. symmetric matrix exists a similarity transformation such that   a11 0 . ST S = 1 1 or S = S−1 . 2. ˜ (2.250) The bilinear form in (2.246) reads then in terms of yj ˜ n n j.k yj ajk yk n = j. yk = k Skj yj .248. .252) ˜ For the determinant of a holds det(˜) = a n ajj ˜ j=1 (2. The proof of (2. (2.253) as well as det(˜) = det(S−1 aS) = det(S−1 ) det(a) . . 0 0 .246) exploits that for any real. det(S) a = (det(S))−1 det(a) det(S) = det(a) . .255) . ann ˜ where S can be chosen as an orthonormal transformation. This property allows one to simplify the bilinear ˜ form n yj ajk yk by introducing new integration variables j..248) .k n n yj = ˜ k (S −1 )jk yk .k yj ajk yk ˜˜ ˜ (2. . according to (2.254) ajj .  .249) The akk are the eigenvalues of a and are real.251) where. . 0  ˜   ˜ S−1 a S = a =  .m (S T )jl alm Smk . (2.249) n ajk = ˜ l. . . . 0 ˜  0 a22 . . . . (2.   .k n m = = j. One can conclude det(a) = j=1 n (2.e.7: Appendix / Exponential Integral 45 which holds for a ∈ C as long as a = 0. ˜ (2. .

(2. . .265) k=1 −∞ which leaves us to determine integrals of the type +∞ dx eicx −∞ 2 (2. .256) such that none of the eigenvalues of a vanishes. .266) .257) Substitution of the integration variables (2. . . n (2. yn ) ∂(˜1 . . . . .e. yn ) y ˜ ∂yj .250) allows one to express (2.249) follows 1 = det ST S such that one can conclude det S = ±1 One can right then (2.250) +∞ +∞ I = d˜1 . y −∞ +∞ d˜n ei y −∞ n k akk yk ˜ ˜2 +∞ i˜11 y1 a ˜2 n i˜nn yn a ˜2 +∞ a ˜ d˜k ei˜kk yk y 2 = d˜1 e y −∞ .262) (2. . . . i.250) holds J = S and.263) (2. y −∞ d˜n det y −∞ ∂(y1 . . 2.258) +∞ +∞ ∂(y1 ..46 We have assumed det(a) = 0. yn ) y ˜ = ( det S )2 (2. . ajj = 0 .258) where we introduced the Jacobian matrix J = with elements Jjs = According to (2.. yn ) y ˜ ei n k akk yk ˜ ˜2 . .259) (2. Accordingly. . ˜ for j = 1. . . .260) (2. ∂(˜1 .264) I = d˜1 . holds n Quantum Mechanical Path Integral ajj = 0 ˜ j=1 (2.. . . . d˜n e y −∞ = (2. . . . yn ) ∂(˜1 . .261) ∂(y1 . hence. . . . det( From (2. . ∂ ys ˜ (2. . yn ) ) = det(S) .

7: Appendix / Exponential Integral 47 where.267) by considering the contour integral J = γ dz eicz = 0 2 (2.257) holds c = 0. The contour intergral (2. We consider first the case c > 0 and discuss the case c < 0 further below. 2 since eicz is a holomorphic function.2. Substituting −x for x into integral J4 one obtains 0 J4 = p p (−i) dx eic(−ix+p) 2 = 0 i dx eic(ix−p) 2 = J2 . (2. the integrand does not exhibit any singularities anywhere in C.271) . c (2.270) dx e−cx 2 2 − 2p √ γ4 : z = ix − p J4 = −p i dx eic(ix−p) .c > 0 . One can relate integral (2.269) The contributions Jk can be expressed through integrals along a real coordinate axis by realizing that the paths γk can be parametrized by real coordinates x p γ1 : z = x γ2 : z = ix + p γ3 : z = √ ix J1 = −p p dx eicx 2 J2 = J3 = = i dx eic(ix+p) 0√ − 2p √ √ 2p 2 i dx eic( 2p √ ix)2 √ − i 0 √ (2. Jk = γk dz eicz 2 (2.266) to the well-known Gaussian integral +∞ dx e−cx −∞ 2 = π ..268) along the path γ = γ1 + γ2 + γ3 + γ4 displayed in Figure 2. The contour integral (2.e.3.268) can be written as the sum of the following path integrals J = J1 + J2 + J3 + J4 . for x. p ∈ IR. i.268) vanishes. according to (2.

(2. p p→+∞ lim |J2 or 4 | ≤ = p→+∞ 0 lim dx |e−2cxp | 1 − e−2cp 2cp = 0. Hence.273) p→+∞ lim J2 and J4 do not contribute then to integral (2.272) = 0 . 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. (2. c   ¡ -p γ1 p Re(z) 2 | |e−2cxp | .3: Contour path γ in the complex plain.268) for p = +∞. (2. We will now show that the two integrals J2 and J4 vanish for p → +∞.274) (2.275) . One can state accordingly ∞ J = dx e −∞ icx2 − √ ∞ i dx e−cx −∞ 2 Using 2.48 Quantum Mechanical Path Integral Im(z) ip γ2 ¡ γ4 ¡ γ3 ¡ -i p Figure 2.267) one has shown then ∞ dx eicx −∞ 2 = iπ .

275.265). n ˜ j=1 ajj (2.277) to (2.7: Appendix / Exponential Integral 49 One can derive the same result for c < 0.268).278) Noting (2.2.276) and (c < 0) ∞ dx eicx −∞ 2 = −iπ = −|c| iπ . if one chooses the same contour integral as (2. This leads to ∞ icx2 J = dx e −∞ + √ −∞ −i ∞ dx ecx = 0 2 (2. (iπ)n det(a) (2. c (2.255) this result can be expressed in terms of the matrix a I = which concludes our proof. 2. It holds n I = k=1 iπ = akk ˜ (iπ)n .277) We apply the above results (2.279) . but with a path γ that is reflected at the real axis.

50 Quantum Mechanical Path Integral .

20. t + |r0 .4) − U (r. t + ) = × exp i 2 +∞ +∞ +∞ −∞ dx1 −∞ dx2 −∞ dx3 m x2 + x2 + x2 1 2 3 m 2πi 3 2 × (3. t) satisfies a partial differential equation.2) d 3 r0 m 2πi 3 2 exp i m (r − r0 )2 − 2 U (r.1 Derivation of the Schr¨dinger Equation o We will consider now the propagation of a wave function ψ(r. Generalizing (2. Notable exceptions are non-stationary systems involving time-dependent linear and quadratic Lagrangians. t) by an infinitesimal time step . It holds then according to (2. (3. of this equation in terms of powers of and we will demonstrate that the terms of order require that ψ(r.h. t + |r0 . We will denote the components of s by (x1 . the Schr¨dinger o o equation provides the simpler avenue towards describing quantum systems than the path ingral formulation of Section 2. For many situations. t) ψ(r + s. t) ψ(r0 . t) . x2 . namely the Schr¨dinger equation.Chapter 3 The Schr¨dinger Equation o 3. (3.h. t) = From this follows ψ(r. (3. t) . t + ) = Ω d3 r0 φ(r. and the r. but by no means all. x3 )T .20) to R one obtains then for small φ(r. . x3 51 . even in x1 .20) to evaluate the propagator. t) . t) ψ(r0 . This yields ψ(r.5) ψ(r. The propagator in (3.3) In order to carry out the integration we set r0 = r + s and use s as the new integration variable. 2. t + ) = Ω m 2πi 3 2 exp i m (r − r0 )2 − 2 U (r. x2 . In the limit of very small it is sufficient to employ a single discretization interval in (2.s.1) can be expressed through the discretization scheme (2. t) .s.1) We will expand the l.21).

11) and. n = 0..4) only for small x2 + x2 + x2 values 1 2 3 we expand 3 ψ(r + s. t) + j=1 xj ∂ 1 ψ(r.52 Schr¨dinger Equation o It is important to note that the integration is not over r.8) one can evaluate recursively all integrals In (a). Obviously. a supposition which will be examined below. 1. It holds In+1 (a) = I1 (a) = i 2a iπ . . t) and the kinetic energy term. of the expansion of ψ(r + s.k=1 xj xk ∂x2 ∂x2 ψ(r. i ∂a Starting from (3.13) 2πi a .8) (3.5) assuming that only the leading terms contribute. t) = ψ(r. e. O( 7 2 ) . (3. t) j k . U (r. t) + ∂xj 2 3 xj xk j.. only terms of the expansion (3.k=1 ∂2 ψ(r. x3 )T .7) According to (2.g. t) j .11) holds 3 m 3 iπ 2 2 × = 1 (3.7) shows 1 ∂ In (a). the terms collected in (3. O( 5 2 ) . consequently. .9) iπ a (3. x2 . According to (3.36) holds I0 (a) = Inspection of (3.6) make contributions of the order O( 3 2 ) .. Since the kinetic energy contribution in (3. t) . It is then sufficient to consider the terms ψ(r. 1 2 3 2 ∂2 j=1 xj ∂x2 ψ(r. (3. t) + . ∂xj ∂xk (3.4) one identifies 2 1 = a m = O( ) (3.5) which are even separately in all three coordinates yield non-vanishing contributions.10) It is now important to note that in case of integral (3. but over s = (x1 . we need then to evaluate integrals of the type +∞ In (a) = −∞ dx x2n exp i a x2 . t) is a constant with respect to this integration.. Since the latter contributes to (3. t). t) j 1 12 . a I2 (a) = − 3 4a2 iπ .6) 3 4 ∂4 j=1 xj ∂x4 ψ(r. 2 (3. a (3. The integration involves only the wave function ψ(r +s..12) Here one needs to note that we are actually dealing with a three-fold integral. 1 4 3 2 2 ∂4 j. x3 ..4) is even in all three coordinates x1 . x2 . O( 7 2 ).

ψ(r. This equation is often written in the o form ∂ ˆ i ψ(r. As we will discuss in Chapter 5 below the solutions of the Schr¨dinger equation are restricted to particular Hilbert spaces H which are o . Obviously. i. In order O( ) the equation reads i ∂ ψ(r.18) This is the celebrated time-dependent Schr¨dinger equation. t) + O( 2 ) . t) = H ψ(r. one specifies the so-called initial condition. (3. t) ψ(r. Usually. t) + O( 2 ) . We will derive briefly the type of boundary conditions encountered. t1 ) = f (r). t) + 1 2i 4 m 2 53 ψ(r. t) = ∂t 2 − 2 2m + U (r. t) (3.20) 2m 3.16) Inserting this into (3. (3. t) + ∂ ∂t ψ(r. t)ψ(r. t) + and exp − i U (r. 1st order in time. t) . t + ) = ψ(r. t) + O( 2 ) ∂t i U (r.14) results in ψ(r.19) ∂t where 2 2 ˆ H = − + U (r. t) ψ(r. (3. t) = ψ(r. The following general remarks can be made about the solution.17) 2m ψ(r..14) This expansion in terms of powers of suggests that we also expand ∂ ψ(r. The 1st order time derivative requires that for any solution a single temporal condition needs to be specified.10). t) + O( 2 ) . t + ) = exp − i U (r.g. a solution is thought for t ≥ t0 and the solution is specified at the intial time t0 . e. t) = 1 − (3.15) ψ(r. using (3.. (3.2: Boundary Conditions and one can conclude. 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. ψ(r. 2nd o order in the spatial variables and linear in the solution ψ(r. this equation is trivially satisfied to order O( 0 ).3. t).e. t) − + i 2 2 i U (r. t) .2 Boundary Conditions The time-dependent Schr¨dinger equation is a partial differential equation. t) (3. Due to its linear character any linear combination of solutions of the time-dependent Schr¨dinger o equation is also a solution.

for example. Since energy is a real ˆ ˆ quantity one needs to require that the eigenvalues of the operator H are real and. in principle.. 1975). (John Wiley. hence. In (3. g ∈ H is defined as follows f |g Ω = d3 rf ∗ (r)g(r) (3.22) and (3. Interchanging f ∗ (r) and g(r) results in g| H |f Ω = Ω ˆ d3 r g(r) Hf ∗ (r) . D. The hermitian property. . 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. t).22) ˆ where H is defined in (3. implies hermitian ˆ f | H |g 1 2 Ω ˆ = g| H |f Ω (3.24) Using Green’s theorem1 Ωd 3r f ∗ (r) 2 g(r) − g(r) 2 f ∗ (r) = one obtains the identity ∂Ω da · ( f ∗ (r) g(r) − g(r) f ∗ (r) ) (3. t) g(r) − Ω d3 rg(r) U (r. New York.27) ˆ We will postulate below that H is an operator in H which represents energy.28) See. The reader is advised to consult a reference text on ‘Linear Algebra’ to follow this argument.25) ˆ f | H |g Ω ˆ = g| H |f Ω + ∂Ω da · P (f ∗ .26) where ∂Ω da · A(r) denotes an integral over the surface ∂Ω of the volume Ω. g|r) = − 2 2m ( f ∗ (r) g(r) − g(r) f ∗ (r) ) (3.54 Schr¨dinger Equation o linear vector spaces of functions f (r) in which a scalar product between two elements f.21) Ω This leads one to consider the integral f | H |g Ω = Ω ˆ d3 rf ∗ (r) H g(r) (3. 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. Chapter 1. however. that H is 2 . Jackson. g|r) is called the ˆ concomitant of H and is P (f ∗ . f ∗ (r) d3 rg(r) Ω 2 = − 2m d3 rf ∗ (r) Ω 2 g(r) − f ∗ (r) (3. 2nd Ed.26) the vector–valued function P (f ∗ .20). differ from each other. J. (3.23).23) ˆ Since H is a differential operator the expressions (3. g|r) (3. Calssical Electrodynamics.

29..e. r ∈ ∂Ω or da · f (r) = 0 ∀r. in which case ∂Ω1 is the inner sphere and ∂Ω2 is the outer sphere. κ > 0 or like r−α . i. . i. Often the closed surface of a volume ∂Ω is the union of disconnected surfaces3 ∂Ωj .34) An example is a volume between two concentric spheres. t)∂t ψ ∗ (r.31) |r|→∞ (3.3: Particle Flux 55 and.. The latter property stems from the fact that the boundary condition (3.32) the wave function must decay for |r| → ∞ rapidly enough to be square integrable.e. It must hold then for all f ∈ H f (r) = 0 ∀r. we can only allow functions which make the differential da · P (f ∗ . However. 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. like exp(−κr).30) Note that these boundary conditions are linear in f . (3. therefore. to avoid discontinuities in ψ(r. 3. (3..30) each condition holding on an entire surface ∂Ωj . t).g. obey (3. . A most common boundary condition is encountered for the volume Ω = R3 in which case one postulates lim f (r) = 0 “natural boundary condition” . The observable directly linked to the wave function is the probability to find the particle at position r at time t.30) can be postulated. α > 2. In either case does f (r) and all of its derivatives vanish asymptotically. t)|2 . r ∈ ∂Ω (3. t) + ψ(r.32).29) In fact.e. (3. namely. in this case also all derivatives of f (r) vanish at infinity. ∂Ω = ∂Ω1 ∩ ∂Ω2 ∩ ∂Ω3 ∩ . if f and g satisfy these conditions than also does any linear combination αf + βg. . e. t) = ω 3 d3 r [ ψ ∗ (r. t) on a single connected surface ∂Ωj only either one of the conditions (3.3 Particle Flux and Schr¨dinger Equation o The solution of the Schr¨dinger equation is the wave function ψ(r. i.31) usually arises when a particle existing in a bound state is described. t) ] . and that the density decays rapidly when one moves away from that area.3. t)|2 .. 3. |ψ(r. t) = ω d3 r |ψ(r. The probability to observe the particle anywhere in the subvolume ω ⊂ Ω is p(ω. g|r) vanish on ∂Ω. t)∂t ψ(r. t) is ∂t p(ω.29. t) which describes the state of o a particle moving in the potential U (r. In this case one can postulate both conditions (3.33) The time derivative of p(ω. 3. Since the total probability of finding the particle anywhere in space is d3 r|f (r)|2 = 1 (3.

36) According to (3. ψ|r. t) − ψ ∗ (r.19) and its conjugate complex4 −i yields i ∂t p(ω.42) 2mi Since (3. t) = ∂ω da · P (ψ ∗ (r.41) where j(r.18) by any complex number and accordingly one can define ψ(r.35) d3 r . One can multiply the solution of (3. (3. The surface integral (3. . A natural choice for this constant is 1. ψ(r.37) can be expressed through a volume integral according to da · A(r) = ∂ω ω (3. Accordingly the probability to observe the particle anywhere in the total volume Ω is constant. t) = 0 (3. t) such that d3 r|ψ(r. (3. t)|2 is a probability density with units 1/volume.43) (3. t) ψ ∗ (r. t) − ψ(r. t) = [ ψ(r. t). We will assume in the remainder of this Section that the solutions discussed are normalized. t) = 0 . t) = H ψ ∗ (r.39) d3 r · A(r) (3. t)r. t) ψ(r. Note that for a normalized wave function the quantity ρ(r. One refers to such solution as normalized. t) = P (ψ ∗ .29. t) .37) If one applies this identity to ω = Ω one obtains according to (3. t) = |ψ(r. (3.44) ˆ Note that the Hamiltonian H involves only real terms. 3. t) + 4 · j(r. t) ∂t ˆ ˆ ψ ∗ (r.40) One can rewrite then (3. t) = 0. t) + ω · j(r.27) this can be written i ∂t p(ω.37) d3 r ∂t ρ(r. t) .27) one can express this j(r.26. t) ] . t) = ω Schr¨dinger Equation o ∂ ∗ ˆ ψ (r. t) H ψ ∗ (r. t) (3.30) ∂t p(Ω.41) holds for any volume ω ⊂ Ω one can conclude ∂t ρ(r.56 Using (3. 3. (3. Using (3.38) holds. t) H ψ(r. t)|2 = 1 Ω (3.

arises solely from the complex factor exp(i k · r). t) points in the direction to the outside of volume ω.50) 3 ˜ d3 k φ(k) exp i(k · r − ωt) Ω∞ (3. 2 for a free particle [c. t) (3. (2.48) that the general solution is of the form ψ(r. i. Equation (3.f. t) = − ∂ω da · j(r. Such case arose in Sect. t) = 0 2 ∂ 2 i ψ(r. It is of interest to note from (3. t0 ) determines φ(k). (3. (3. t) gives rise to a descrease of the total probability in volume ω due to the disappearence of probability density at the surface ∂ω.125)] and in a quadratic [c.f. t) = − ψ(r. 3.71)]. equal to the velocity of the particle.18) in Ω∞ = R in the case o U (r. t) = [2π]− 2 where the dispersion relationship holds ω = k2 .3. 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.51) . for f (r) ∈ R.46) i. 2. (3.4: Free Particle 57 The interpretation of j(r. Note that j(r.105.47) . 2m (3. 2.. and for particles moving in a linear [c.48. This finding is consistent with the present evaluation of the particle flux: a factor exp(ipo xo / ) gives rise to a flux po /m. m (3. t) is that of density flux. One can readily show by insertion into (3.48) ∂t 2m which describes the motion of free particles.. This follows directly from an inspection of Eq. (3.148. t0 ) = [2π]− 2 Ω∞ 3 ˜ d3 k φ(k) exp i(k · r − ωt0 ) . t) has vanishing flux anywhere.167)] potential.f. j(r.49) reads at t = t0 ψ(r. the initial condition at ψ(r.43) that any real wave function ψ(r.45) Obviously. t) . (2. (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.e. 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) .49) ˜ Obviously.4 Solution of the Free Particle Schr¨dinger Equation o We want to consider now solutions of the Schr¨dinger equation (3.e.41) written in the form ∂t ω d3 r ρ(r. In Sect. 2.

t|r0 .. In fact. that the “natural boundary condition” (3. Comparision with Path Integral Formulation One can write solution (3.55) 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. To show this one needs to note 1 2π = +∞ dk1 exp −∞ ik1 (x − x0 ) − 1 2 i 2 k2 1 2m . i.247).49). t) as given by (3.51.47).114) and (4. 3.49) is square integrable at all subsequent times and. t0 ) (3.54) yields (2. t) = Ω∞ d3 r0 φ(r. In case of the harmonoc oscillator the expansion can be stated in a closed form given in (4. in (2.52) We have not specified the spatial boundary condition in case of (3.70) derived below for a particle in a box and the harmonic oscillator. 3. t|r0 . We have identified then with (3. evaluating the integral in (3.54) valid for this case. (3.54) to the case of non-vanishing potentials U (r). t0 ) .47).52) above ψ(r. t0 ) ψ(r0 . derive an expression similar to (3. Below we will generalize the propagator (3..53) where φ(r.51) to be square integrable it follows according to the properties of the Fourier–transform that ψ(r.31) applies. If one chooses the initial state f (r) defined in (3. i.56) and using (2. (t − t0 ) (3. respectively.81) . t0 ) = 1 2π 3 d3 k exp Ω∞ ik · (r − r0 ) − i 2 k2 2m (t − t0 ) .58 The inverse Fourier transform yields 3 ˜ φ(k) = [2π]− 2 Schr¨dinger Equation o d3 r0 exp(−ik · r0 ) ψ(r0 .54) This expression obviously has the same form as postulated in the path integral formulation of Quantum Mechanics introduced above. hence.e. Ω∞ (3.54) an alternative representation of the propagator (2.e. The solution as stated is defined in the infinte space Ω∞ = R .5). The general form for this propagator involves an expansion in terms of a complete set of eigenfunctions as in (3. The ensuing solutions are called wave packets.49.

49.56)]. (3. (3. (3.63) .60) yields with t0 = 0 ψ(x.62) We will restrict the space of allowed solutions to a volume Ω such that the functions also make the concomitant (3.57) where φ(x.27) vanish on the surface ∂Ω of Ω.f. t) 2 k2 (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 (3.29. t|x0 .64) (3. 3. 3.4: Free Particle Free Particle at Rest 59 We want to apply solution (3.30).. The remaining integration over k leads to the integral +∞ −∞ dk exp ikx − 1 2 (k − po 2 2 ) δ − i k2 m (t − t0 ) (3.19.52) to the case that the initial state of a 1-dimensional free particle ψ(x0 . t|x0 .61) × exp − (x − po m t)2 2δ 2 (1 + 2 t2 ) m2 δ 4 (1 − i .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 (3.46). the functions obey boundary conditions of the type (3. the boundary conditions are φ(r) = 0 ∀r.59) which is solved through completion of the square in the exponent [c. i. The 1-dimensional version of (3.54) is +∞ ψ(x. r ∈ ∂Ω or da · φ(r) = 0 ∀r.60) =???? Combining (3. We have demonstrated then in this case that the Schr¨dinger formulation of Quantum o Mechanics is equivalent to the Feynman path integral formulation. t0 ) ψ(x0 .53.3.55. t) = f (t) φ(r) . t0 ) = 1 2π +∞ dk exp −∞ ik (x − x0 ) − i 2m (t − t0 ) . r ∈ ∂Ω (3. a result which is identical to the expression (??) obtained by means of the path integral propagator (2. (3. Stationary States We consider now solutions of the time-dependent Schr¨dinger equation (3. Accordingly.57–3.20) which are of o the form ψ(r. 3. t) is given by (??). t) = −∞ dx0 φ(x.e. 3.

69) exist. t)|2 = |f (t)|2 Ω d3 r |φ(r)|2 (3.70) The task of finding solutions φ(r) which solve (3. At this point we will accept that solutions φ(r) of the type (3. one factor depending only on the time variable and the other only on the space variables are called separable in space and time. We will encounter many such problems in the subsequent Sections.62) we insert it into (3. t) = f (0) exp − i E φE (r) where ˆ H φE (r) = E φE (r) .e. We want to demonstrate this property now. In fact. One can conclude that the probability density for the state (3.62) which consist of two factors.45) that the total probability d3 r ρ(r.62) is |ψ(r.68) can hold only for all t and all r if g1 (t) = E g2 (t) and E h1 (r) = h2 (r) for some E ∈ C. As pointed out above. for the eigenvalue problems in the confined function space. It follows then from (3.62) are special solutions of the timedependent Schr¨dinger equation. We may note in passing that solutions of the type (3. It is important to realize that the separable solutions (3. t)|2 = |φ(r)|2 . i. We will demonstrate that solutions of the type (3. One calls such states stationary states...e. (3. (3.62) exists.62) have the particular property that the associated probability distributions are independent of time.27) holds and.42) the flux j(r. At this point we state without proof that. It turns out that a solution for f (t) can be found for any E. by no means all solutions are of this type. h1 (r) = φ(r). 2.68) ˆ where g1 (t) = i ∂t f (t).69) is called an eigenvalue problem.69) If these two equations can be solved simultaneously a solution of the type (3. We have then shown that ψ(r. and h2 (r) = H φ(r). however. often only for a discrete set of values En . a common case is Ω = Ω∞ and the ‘natural boundary condition’ (3.62) do exist and we will characterize the two factors of the solution f (t) and φ(r).65) is constant.71) .67) (3. .63. for functions required to obey boundary conditions (3. g2 (t) = f (t). .31). namely f (t) = f (0) exp − i Et . the solutions o (3. (3. if f (t) = eiα . t) vanishes on the surface of ∂Ω. according to (3. In order to further characterize the solution (3. n = 1.60 Schr¨dinger Equation o and affect only the spatial wave function φ(r). is time-independent.e. α ∈ R . We must postulate therefore ∂t f (t) ˆ H φ(r) = E f (t) = E φ(r) . This can hold only if |f (t)| is time-independent..64). (3.19). The identity (3. i. t) = Ω Ω d3 r |ψ(r. solutions exist only for a set of discrete E ˆ values.66) i. This yields an expression g1 (t) h1 (r) = g2 (t) h2 (r) (3. the eigenvalues of the operator H. in general. 3. It follows from the observation that for the solution space considered (3. hence. We denote the corresponding solution by φE (r). .

28) for the special case that f and g in (3.75) 2m One can ascertain this statement by inserting the expression for φk (r) into the eigenvalue problem posed in (3. We will show in Section 5 that E can be interprerted as the total energy of a stationary state. − 2 2m φ(r) = E φ(r) . of course. The corresponding stationary solution ψ(r. m (3. As a result we cannot postulate in the present case that wave functions are localized and normalizable.20). i. t) = exp − i 2 k2 2m t exp i k · r (3.76) has kinetic energy 2 k 2 /2m.28) both represent the state φE (r).3.73) from which follows E = E ∗ and.74) using exp i k · r = ik exp i k · r .74) The classical free particle with constant energy E > 0 moves without bounds in the space Ω∞ . one can interpret then k as the magnitude of the momentum of the particle. E ∈ R. . k ∈ R 2 k2 φk (r) = N exp i k · r .71) this yields E Ω d3 r φ∗ (r) φE (r) = E ∗ E Ω d3 r φE (r) φ∗ (r) .e. particles scattered of a potential. = E. E (3. ˆ d3 r φ∗ (r) H φE (r) = E Ω Ω ˆ d3 r φE (r) H φ∗ (r) . (3.43) is j(r. a property which is to be expected since the energy is purely kinetic energy which. in fact.. hence.66) E must be real. Obviously. should be positive.72) According to (3.71) are.g.74) is actually best labelled by an index k. 3. We want to prove now that the eigenvalues E which arise in the eigenvalue problem (3. o According to (3.19. The resulting total energy values E are positive. The solution φE (r) corresponding to the eigenvalue problem posed by (3.4: Free Particle 61 is a solution of the time-dependent Schr¨dinger equation (3. e. We start our proof using the property (3. E (3. (3. Stationary State of a Free Particle We consider now the stationary state of a free particle described by i ψ(r. We will wave this assumption as we always need to do later whenever we deal with unbound particles. The flux corresponding to (3.76) according to (3. t) = |N |2 k . real. t) = exp − E t 2 φE (r) .77) Noting that k can be interpreted as the magnitude of the momentum of the particle the flux is equal to the velocity of the particle v = k/m multiplied by |N |2 .

We assume.78) (3. . (3. Setting up the Space F1 of Proper Spatial Functions The presence of the infinite energy wall is accounted for by restricting the spatial dependence of the solutions to functions f (x) defined in the domain Ω1 = [−a.84) .81) 2m = E. 2m dx2 2 (± ) = } (3.82) have the property that either both boundary conditions are satisfied or none.o) (a) = 0 and φE (e. . so-called even solutions obeying φ(x) = φ(−x) φE (x) = A coskx .82) One can readily verify that (3.80) 2m dx2 We note that the box is symmetric with respect to the origin. (3. Solutions of the Schr¨dinger Equation in F1 o The time–dependent solutions satisfy i ∂t ψ(x. In case of the even solutions (3.81. t) . let say the one at x = a.81) they are kn = nπ . f continuous. hence.82) satisfy the differential equation in (3.83) The solutions (3. we have to consider only one boundary condition. Hence.79) The stationary solutions have the form ψ(x. We can expect. two types of solutions.o) (−a) = 0 (3. (3. t) = − d2 ψ(x. (o) 2 k2 (e) 2 k2 2m = E (3.81. i.62 Schr¨dinger Equation o 3. n ∈ N. a] ⊂ R assuming that the potential outside of this interval is infinite. that the solutions obey this symmetry as well. a}. 5 . t) = exp(−iEt/ )φE (x) where φE (x) is determined by 2 d2 − φE (x) = E φE (x) . a] ⊂ R which vanish on the surface ∂Ω1 = {−a.80).e. φ(±a) = 0 .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. .80) need to be satisfied are φE (e. a] ⊂ R → R.. It turns out that this boundary condition can only be satisfied for a discrete set of k–values kn . 3. 2a n = 1. We will refer to this as a particle in a one-dimensional ‘box’. The boundary conditions which according to (3. 3. f ∈ F1 = {f : [−a. therefore. 3. and so-called odd solutions obeying φ(x) = −φ(−x) φE (x) = A sinkx.

) The Energy Spectrum and Stationary State Wave Functions The energy values corresponding to the kn –values in (3. 4. x) = An cos n and φ(o) (a.81. n is assumed to be even. This condition implies for the even states +a |An |2 −a dx cos2 nπx = |An |2 a = 1 . are En = 2π2 8ma2 n2 .88) where the energies for odd (even) n–values correspond to the even (odd) solutions given in (3. 3.90) 2a The wave functions represent stationary states of the particle in a one-dimensional box.84. It is desirable to normalize the wave functions such that +a −a dx |φ(e.92) and for the odd states +a |An |2 −a dx sin2 nπx = |An |2 a = 1 . 2a n = 1. The wave functions for the five lowest energies En are presented in Fig. . 4. . 5 .91) holds.82). 6 . according to (3. n = 2. . 2a 1 . .3. 3.94) .87) (Note that. . . 6 . 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 . a n = 2. respectively. . . 6 . 2a n = 1. 3 .e. 3. x) = An sin n nπx .86). (3. By counting the number of their nodes one can determine the energy ordering of the wave functions. n = 1. 2a n = 2. (3. . (3.82) only the kn –values cos(kn a) = cos kn = nπ .1).o) (a.93) The normalization constants are then An = (3. (3. φ(e) (a.82). .86). . 3.89) nπx . (3.5: Particle in One-Dimensional Box since for such kn 63 nπa nπ = cos 2a 2 In case of the odd solutions (3. x)|2 = 1 n (3. = 0. (3. 5 . according to the dispersion relationships given in (3.. 2. i. (3.86) satisfy the boundary condition since for such kn nπa sin(kn a) = sin 2a = sin nπ 2 = 0.81) and (3. 4.85) (3. (3. . .

2. 5 of particle in a box. 4.1: Eigenvalues En and eigenfunctions φn (e.o) (a.64 Schr¨dinger Equation o E h2 ] [ 8 m a2 25 ¢ 16 9 4 Figure 3. 3. ¡   1 -a a x . x) for n = 1.

hence. a] are N. .102) holds according to (3. it holds φn |φm Ω1 = δn m . (i) n.e. . . m odd.101) +a −a dx cos (n−m)πx − cos (n+m)πx 2a 2a and. In particular. N ≥ 1. x) (3. i. .3. this integral vanishes. it holds: f |f Ω1 = 0 → f (x) ≡ 0. 5 . i.98) the elements of B1 must be linearly independent. m both odd. 6 . 3.97) form an orthonormal basis set. .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. (ii) n. the integrals vanish.94) B1 = {φn (a. . n = m. m even φn |φm 1 a Ω1 = 1 a +a nπx −a dx sin 2a sin mπx = 2a (3. In case of n = m we have to consider three cases. . one obtains for n.98) ∞ f |f 5 Ω1 = n=1 d2 . and (iii) the mixed case. 3.96) f |g Ω1 = −a dxf (x) g(x) . In fact. In case of n.89.. Hence we need to consider only the first two cases. too. The latter case leads to integrals φn |φm Ω1 = 1 a +a dx cos −a nπx mπx sin . 3. Similarly. Because of the property (3. x) = together with the scalar product5 +a 65 (3.99) Since in this case the integrand is a product of an even and of an odd function. 4. (3. 2. Obviously. the integral vanishes. n (3. for ∞ f (x) = n=1 dn φn (a.90. (3. f.. the integral arises φn |φm 1 a Ω1 = 1 a +a nπx −a dx cos 2a cos mπx = 2a (3. 3. x) . the integrand is odd.95) 1 a cos nπx 2a sin nπx 2a for n = 1.} where φn (a.98) The latter property is obviously true for n = m. 2a 2a (3.103) We will show in Section 5 that the property of a scalar product do indeed apply.e.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. n = 1. g ∈ F1 (3. . for n = 2. m both even.

m = 1.95.98) one obtains +a dx0 φm (a. (3.95) is also complete. . Demonstration of completeness is a formidable task.106) Using the orthonormality property (3. x) ∂t cn (t) −a dx0 φn (a. The functions f are periodic with period 2a.108) into the Schr¨dinger equation yields o +a ∞ n=1 φn (a. It follows that B1 n defined in (3. the coefficients an . x0 ) ψ(x0 .} such that ∞ ∞ nπx nπx ˜ = + an sin (3. We like to show finally that the basis (3. (3. . . n = 0. 5 . . x) cn (t) −a dx0 φn (a. . ( ) = ([ + ] − ) (3.105) f an cos 2a 2a n=0 n=1 ˜ The functions f corresponding to the functions in the space F1 have zeros at x = ±m a. 4. 2. n = 1. however. Evaluating the Propagator We can now use the expansion of any initial wave function ψ(x.79) requiring the o initial condition cn (t0 ) = 1 . n = 2. 3. . Hence. .109) Insertion of (3. x) . 3.104) ˜ where [y]a = y mod2a.66 Schr¨dinger Equation o f (x) ≡ 0 implies f |f Ω1 = 0 which in turn implies dn = 0 since d2 ≥ 0. In the present case.106) and generalizing to t ≥ t0 one can write ∞ +a ψ(x. x0 ) ψ(x0 . there exist real constants {an . a] yields then also an expansion for any element in F1 and B1 is a complete basis for F1 .110) . t0 ) .e. any element of the function space F1 defined in (3. We have then shown that any f corresponding to elements of F1 can be expanded in terms of elements in B1 .95) is an orthonormal basis.78) can be expressed as a linear combination of the elements of B1 defined in (3. Accordingly.} and {bn . 1. t) at times t > t0 . . .e. 2. Restricting the expansion (3. t0 ) (3. t0 ) = n=1 dn φn (a. t0 ) in terms of eigenfunctions φn (a.108) where the functions cn (t) are to be determined from the Schr¨dinger equation (3. x0 ) ψ(x0 . x) cn (t) −a dx0 φn (a. . . t0 ) = dm .107) Inserting this into (3.96). i. For this purpose we extend the definition of the elements of F1 to the whole real axis through ˜ f : R → R. (3.. −a (3. . they can be expanded in terms of a Fourier series. . 3. n = 1. t) = n=1 φn (a. such demonstration can be based on the theory of Fourier series. and bn .. in (3. This implies that only the trigonometric functions which are elements of B1 enter into the Fourier ˜ series. t0 ) = +a ∞ i n=1 − En φn (a.105) must vanish. x0 ) ψ(x0 . i. x) to obtain an expression for ψ(x.105) to the interval [−a. . For this purpose we expand ∞ ψ(x.

t0 ) itself is a solution of the time-dependent Schr¨dinger o equation (3. the elements of which satisfy the appropriate boundary conditions. 4 a (3.78).5). (3. t0 ) ψ(x0 .111) The solutions of these equations which satisfy (3. t|x0 .79) which lies in the proper function space (3. being reflected at the right wall. k0 = .2 presents the probability distribution of the particle at subsequent times. t) for any initial condition ψ(x. t0 |x0 . Note that the propagator has been evaluated in terms of elements of a particular function space F1 .109) are cm (t) = exp i − Em (t − t0 ) . The last frame shows the wave front reaching again the right wall and the onset of new interference.3. 67 (3. x) exp n=1 i − En (t − t0 ) φn (a. In the present system which is composed solely of bound states the propagator is given by a sum. Often the propagator φ(x. This solution can be written +a ψ(x. t|x0 . t0 ) (3.5: Particle in One-Dimensional Box Multiplying both sides by φm (a. t0 ) is also referred to as a Greens function. Example of a Non-Stationary State As an illustration of a non-stationary state we consider a particle in an initial state ψ(x0 .108.112) determine now ψ(x. t0 ) = ∞ φn (a. cm (t0 ) = 1 . i ∂t cm (t) = − Em cm (t) . One can recognize that the particle moves first to the left and that near the right wall of the box interference effects develop. t|x0 . 3.. but the collision with the left wall leads to new interference effects. t0 ) = 1 2πσ 2 1 4 exp − x2 0 + i k0 x0 2σ 2 .114) This expression has the same form as postulated in the path integral formulation of Quantum Mechanics introduced above. The interference pattern begins to ‘smear out’ first.113). t|x0 . In case that different boundary conditions hold the propagator will be different as well.113) where φ(x. i. t) = −a dx0 φ(x.115) This initial state corresponds to the particle being localized initially near x0 = 0 with a velocity v0 = 15 /m a in the direction of the positive x-axis. t0 ). It is of interest to note that φ(x. Figure 3. x) and integrating over [−a. (3. in (2. x0 ) . σ = a 15 .114) the representation of the propagator for a particle in a box with infinite walls. rather than by an integral (3. .112) Equations (3. The particle moves then to the left. We have identified then with (3. according to (3.54) as in the case of the free particle system which does not exhibit any bound states. a] yields.98). t0 ) = δ(x − x0 ) as can be readily verified using (3.e. The respective initial condition is φ(x.

68 Schr¨dinger Equation o |Ψ|2 2 t = 0.92 m a h2 |Ψ|2 2 t = 2.48 m a h2 |Ψ|2 2 t = 0.96 m a h2   |Ψ|2 2 t = 1.44 m a h2 |Ψ|2 2 t = 1.24 m a h2 1 πσ k = 15 a 1 πσ -a a -a a |Ψ|2 2 t = 0. .2: Stroboscopic views of the probability distribution |ψ(x.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 = 0.00 m a h2 |Ψ|2 2 t = 0.20 m a h2 1 πσ 1 πσ -a a -a a |Ψ|2 2 t = 1.16 m a h2   1 πσ 1 πσ -a a -a a Figure 3. t)|2 for a particle in a box starting in a Gaussian distribution with momentum 15 /a.

x2 .. ( . 3 (3. .78). t) = exp − E t φE (x1 .121) where − 2 2m 2 ∂j φ(j) (xj ) = Ej φ(j) (xj ) . x2 = ±a2 } ∪ {(x1 . a1 ] ⊗ [−a2 . one can express H 3 φ(x1 . x1 = ±a1 } ∪ {(x1 . i.6: Particle in Three-Dimensional Box Summary: Particle in One-Dimensional Box 69 We like to summarize our description of the particle in the one-dimensional box from a point of view which will be elaborated further in Section 5. x3 . i. x3 ) = j=1 φ(j) (xj ) (3. x3 ) .116) where Ω is the interior of the box and ∂Ω its surface Ω ∂Ω = [−a1 . φ(j) (±aj ) = 0 . can be counted. f continuous. 2a2 . (3. x2 .3. ˆ H = − 2 (3. x2 . The corresponding solutions have the form i ψ(x1 . (3. )T ∈ ∂ } .e. x3 )T ∈ Ω. x3 )T ∈ Ω.117) 2m 2 2 2 ∂1 + ∂2 + ∂3 (3.6 Particle in Three-Dimensional Box We consider now a particle moving in a three-dimensional rectangular box with side lengths 2a1 . hence. x3 . ) = ∀( . This property allowed us to evaluate the propagator (3. 2. x3 ) is an element of the function space F3 defined in (3. x2 . The description employed a space of functions F1 defined in (3. x2 . We seek then solutions of the time-dependent Schr¨dinger equation o ˆ i ∂t ψ(x.119) where φE (x1 . 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. We have defined in the space F1 a scalar product (3. Placing the origin at the center and aligning the x1 .122) . x2 .97) with respect to which the eigenfunctions are orthonormal. a3 ] ⊂ R = {(x1 . t) .120) ˆ Since the Hamiltonian H is a sum of operators O(xj ) each dependent only on a single variable.e.118) which are stationary states. of the space F1 is that the elements of the basis set can be enumerated by integer numbers. . j = 1. x3 ) = E φE (x1 .. A complete basis of F1 is given by the infinite set B1 (3.116) and obeys the partial differential equation ˆ H φE (x1 . x2 . x3 = ±a3 } . a2 ] ⊗ [−a3 . x3 ) (3.114) in terms of which the solutions for all initial conditions can be expressed. x2 . An important property of this basis and. x3 )T ∈ Ω. x2 . x2 .95). x2 . 2a3 . 3. t) = H ψ(x1 . ˆ = O(x1 ) + O(x2 ) + O(x3 ).

124) The same considerations as in the one-dimensional case allow one to show that B3 = {φ(n1 .e. if all three lengths a1 . n1 . 2. 2.96) and. . x1 . r) exp i − E(n1 n2 n3 ) (t − t0 ) φ(n1 . For example. .n2 . Such degeneracies are always a signature of an underlying symmetry. .122) are given by (3. 2. 3.126) Symmetries The three-dimensional box confining a particle allows three symmetry operations that leave the box unchanged. a2 . t|r0 .n3 ) (a1 .n3 ) (a1 . x2 . a2 . .120) can be written φ(n1 . This additional symmetry is also reflected by degeneracies in the energy levels. . r0 ) . t0 ) = n1 . x2 ) φn3 (a3 .122) with (3. a2 . . . n2 . 3.125) is a complete orthonormal basis of F3 and that the propagator for the three-dimensional box is φ(r. x1 . a3 .n3 ) = 2π2 = φn1 (a1 .n3 =1 φ(n1 ..n2 . n3 = 1.70 Schr¨dinger Equation o and E1 + E2 + E3 = E. a3 . Comparing (3. 3. n 3 and E(n1 . i. a2 .n2 .123) 8m n2 n2 n2 1 2 3 + 2 + 2 a2 a2 a3 1 .80) shows that the solutions of (3. a1 = a2 = a3 = a then rotation around the x1 .n2 . (3. x3 ) = 1. x3 ) n1 . a3 . 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. a3 . x3 ) . n3 = 1. (3. x1 ) φn2 (a2 . n1 . x3 –axes. namely rotation by π around the x1 . n 2 . in the present case .n2 .n3 ) (a1 . hence. the solutions of (3. This symmetry has been exploited in deriving the stationary states.} ∞ (3. (3. x2 . x3 –axes by π/2 also leaves the system unaltered.n2 . . If two or all three orthogonal sides of the box have the same length further symmetry operations leave the system unaltered. Actually.n3 ) (a1 . a2 . x2 . x2 . a3 are identical. n2 . .

due to the hermitian character of ˆ H. Hence. for example. n2 . We consider the degeneracy of the energy E122 which is due to the identity E122 = E212 = E221 . However. n3 ) = (3. 3. r) + β φ(2. 3) (1. However. a.. φ(1.1. the identity 32 + 32 + 32 = 12 + 12 + 52 . r) + γ φ(2.1) (a. 1. however.g. a. a. to construct n orthogonal stationary states6 .2. a. Any linear combination of wave functions ˜ φ(r) = α φ(1.2) (a.6: Particle in Three-Dimensional Box 71 ‘accidental’ degeneracies also occur. One particular aspect of the degeneracies illustrated in the Table above is worth mentioning. r). r) (3. for (n1 . 6 This is a result of linear algebra which the reader may find in a respective textbook.2. 5) as shown in the Table above. a. a.2) (a. this linear combination o is not necessarily orthogonal to other degeneraste states.127) ˜ ˆ ˜ obeys the stationary Schr¨dinger equation H φ(r) = E122 φ(r). a.2) (a.3. in case of an n–fold degeneracy it is always possible. e. The origin of this degeneracy is. in case of degenerate states one cannot necessarily expect that stationary states are orthogonal. . a.2.

72 Schr¨dinger Equation o .

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

1) of the form ψ(x. We will point out explicitly where assumptions are made which built on this restriction.3) can be solved for any E ∈ R. The stationary states of the harmonic oscillator have been considered already in Chapter 2 where the corresponding wave functions (2.3) Due to the nature of the harmonic potential which does not allow a particle with finite energy to move to arbitrary large distances.5). This property states that p and x obey an ˆ ˆ algebra in which the two do not commute.5) →±∞ where C∞ denotes the set of functions which together with all of their derivatives are continuous.5) ˆ H = 1 1 2 p + m ω 2 x2 ˆ ˆ 2m 2 (4. In the framework of the Schr¨dinger equation the stationary o states are solutions of (4. act. The Hamiltonian H can be expressed in terms of the operators acting on the space (4. all stationary states of the harmonic oscillator must be bound states and. 4.7) which is why these operators are of interest to us.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. If this restriction would not apply and all functions f : R → R would be admitted. In order [ˆ. i dx x = x ˆ (4. however. (4.3) would be continuous and the eigenfunctions φE (x) would not be normalizable. is the commutation property 1 1 (4.1 Creation and Annihilation Operators The Hamiltonian operator (4. lim ( ) = } (4. x] = p ˆ . beside the representation (4. the natural boundary conditions apply x→±∞ lim φE (x) = 0 .8) i which holds for any position and momentum operator.6) the first being a differential operator and the second a multiplicative operator. t) where ˆ H φE (x) = E φE (x) . in which the operators used below.235) had been determined. The cardinal property exploited below.2) can be expressed in terms of the two operators p = ˆ d . It is important to keep in mind this restriction of the space. however.4) Equation (4.74 Linear Harmonic Oscillator In the following we consider first the stationary states of the linear harmonic oscillator and later consider the propagator which describes the time evolution of any initial state. the commutator has a simple form.4) be satisfied. t) = exp(−iEt/ )φE (x. ∈ C∞ . In the following algebraic solution of (4. the spectrum of ˆ H in (4.7) of the Hamiltonian. therefore. (4. only for a discrete set of E values can the boundary conditions (4. The operators act ˆ on the space of functions N1 defined in (4.

Our next step is an attempt to factorize the Hamiltonian (4. in fact.16) (4.13) (4.4. p ˆ 2 2m ω 2 (4.9) From this follows (4.14. 2mω dx (4.8) we recall that all operators act on functions in N1 and.12). ˆ 2 2m ω (4.15) together lead to the commutation property (4. One obtains [ˆ. Being guided by the identity for scalar numbers (b − ic)(b + ic) = b2 − i(cb − bc) + c2 = b2 + c2 (4.18) It is of interest to note that the operators a+ and a− are real differential operators. In fact.1: Creation and Annihilation Operators 75 to derive (4. x] . a+ ] = 1 .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 . to factor H/ ω. Before we continue we like to write (4. a ˆ 1 (4. the commutator property (4.14) and (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 . Since a+ and a− are ˆ ˆ + and a− since operators and not scalars we cannot simply expect that the identy (4. 4.15) 1 ˆ 1 H + 1 . 1 ω 2 (4.14) 1 i mω 2 x + ˆ p2 + ˆ [ˆ. hence. the action of [ˆ. ˆ ˆ . ˆ 2 2m ω a− = ˆ mω i x + √ ˆ p.8).10) we define a+ = ˆ mω i x − √ ˆ p.10) holds for a ˆ ˆ [ˆ− .12) To prove this commutation property we determine using (4. a+ ] = a− a+ − a+ a− does not necessarily vanish.7) and (4. 2mω dx a− = ˆ mω x + ˆ 2 d . x] f (x) = p ˆ d d x f (x) − x f (x) = f (x) = 1 f (x) .8) a ˆ ˆ ˆ ˆ ˆ implies [ˆ− .11) ˆ The reader may note that we have attempted. 1 2 (4. 1 ω 2 (4.7) assuming that the factors are easier to handle than the Hamiltonian in yielding spectrum and eigenstates.11) a− a+ = ˆ ˆ (4.8) yield a− a+ = ˆ ˆ Similarly one can show a+ a− = ˆ ˆ 1 ˆ 1 H − 1 . 1 i dx i dx i i (4. x] p ˆ on such functions must be considered.

21) ˆˆ H a− φE (x) ω 1 ) a− φE (x) 1 ˆ 2 ω = a− ( ω a+ a− + ˆ ˆ ˆ 1 − ω 1 ) φE (x) 1 1 2 ˆ = a− ( H − ω ) φE (x) ˆ = ( ω a− a+ − ˆ ˆ = (E − ω ) a− φE (x) . (4. 4.16.16.16. 4.17. 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) .23) can be generalized to m-fold application of the operators a+ and a− ˆ ˆ ˆ a m H (ˆ+ ) φE (x) ˆ a m H (ˆ− ) φE (x) 1 = ( E + m ω ) a+ ˆ = (E − m ω) a ˆ m φE (x) φE (x) .19) in the form1 f |ˆ+ g a Ω∞ = g|ˆ− f a Ω∞ . ˆ (4.19).22. one obtains using (4. The property ˆ ˆ follows directly from (??).17. 4. 4. 4.22) Similarly one can show using again (4.21) a− φE (x) is ˆ a stationary state of energy E − ω and a+ φE (x) is a stationary state of energy E + ω. it holds that for a stationary state φE (x) of energy E defined through (4. .21) is available. 4. The derivation will be based solely on the properties (4.24) − m This property states that the operators in the function space N1 are the hermitian conjugate of each other. This property of operators is investigated more systematically in Section 5. In fact. 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. g ∈ N1 +∞ +∞ dx f (x) a+ g(x) = −∞ −∞ dx g(x) a− f (x) . ˆ (4. (4.76 Relationship between a+ and a− Linear Harmonic Oscillator The operators a+ and a− are related to each other by the following property which holds for all ˆ ˆ functions f. ˆ The results (4.23) Together.19) This property states that the operators a+ and a− are the adjoints of each other.20) ˆ In the following we will determine the spectrum of H and its eigenstates. (4. Using (??) we like to state (4.12.17. 4.

2: Ground State 77 One can use these relationships to construct an infinite number of stationary states by stepping up and down the spectrum in steps of ± ω..32) for some constant c0 or (4.26) Of course. E = E . (4. in fact.25) needs to be normalizable in order to represent a bound state of the harmonic oscillator. (ii) the construction appears to yield energy eigenvalues without lower bounds when. 4. i.e. m ∈ Z when m is decreased. in fact.) ˆ ˆ 0 = φE |HφE Ω∞ − φE |HφE Ω∞ = ( E − E ) φE |φE Ω∞ . one expects that E = 0 should be a lower bound. The property (??) has an important consequence for the stationary states φE (x). For obvious reasons one refers to a+ and a− as ladder ˆ ˆ operators. the solution φ0 (x) of (4. a state φ0 (x) which obeys the property a− φ0 (x) = 0 ˆ (4. φ0 (y) dy dy the solution of which is 1 lnφ0 (y) = − y 2 + c0 2 φ0 (y) = c0 exp 1 − y2 2 . 2 2 (4.5).29) where y = mω x.28) 4. (4. It turns out that both difficulties can be resolved together. Another name is creation (ˆ+ ) and annihilation (ˆ− ) operators since these operators a a create and annihilate vibrational quanta ω.25) φ0 (y) = 0 (4.6.25) can. For f (x) = φE (x) and g(x) = φE (x) in (??) follows (Note that according to (??) the eigenvalue E is real.25) on one side would lead to termination of the sequence E0 + m. (4.30) Assuming that φ0 (y) does not vanish anywhere in its domain ] − ∞. two difficulties: (i) one needs to know at least one stationary state to start the construction. however. be found.27) Since E = E one can conclude φE |φE Ω∞ = 0. on ˆ the other side such a state is itself an eigenstate of H as can be shown using (4.29) d 1 d φ0 (y) = = lnφ0 (y) = − y . E .17) ˆ H φ0 (x) = ( ω a+ a− + ˆ ˆ 1 ω 1 ) φ0 (x) = 1 ω φ0 (x) . (4. In fact. φ0 (x) should be an element of the function space N1 defined in (4. There arise.4.33) .2 Ground State of the Harmonic Oscillator d + y dy A suitable solution of (4.31) (4.11) one can rewrite (4. Let φE (x) and φE (x) be two normalized stationary states corresponding to two different energies E. Using (4. +∞[ one can write (4.

2 n = 0. mω (4.36) assumes a functional form such that both terms together assume a minimum value. For this purpose we apply the operator a+ to the ˆ ground state (4. a state confined to the minimum corresponding to the classical state of lowest energy. i.e. would yield a vanishing contribution. . n = 0.2) there are two competing contributions to the energy. .39).e. 2. the potential energy contribution which for a state δ(x). 2 This state of lowest energy is called the ground state of the oscillator. i. m ∈ Z which properly terminate at some minimum value E + m0 ω ≥ 0.37) It is most remarkable that the energy 1 ω of the ground state is larger than the lowest classically 2 allowed energy E = 0. . . .25) admits only one solution there is only one set of states with energy E + m ω. the so-called excited states. 1.3 Excited States of the Harmonic Oscillator Having obtained a suitable stationary state with lowest energy. (4. (4.. choose the normalization constants cn and the functions φn (x) real as well. 1.26) the energy value associated with this state is 1 ω. 2 n = 0. it holds ˆ H φn (x) = ω(n + 1 ) φn (x) . The ground state (4. 2.. Such states need to be suitably normalized for which purpose we introduce a normalization constant cn φn (x) = cn a+ ˆ n φ0 (x) . . . We a can. we construct the states for n = 1. . (4.39) We notice that the ground state wave function φ0 (x) as well as the operators (ˆ+ )n are real. . 2.38) These states correspond to the energy eigenvalues (4. . The set of allowed energies of the oscillator according to (4. therefore.36) n times. We recall that according to (4. we can now construct the stationary states corresponding to energies (4. 1..37) above the ground state energy.24) can then be written as follows En = ω(n + 1 ). (4. 2.78 Linear Harmonic Oscillator This solution is obviously normalizable. and the kinetic energy contribution. . which for a narrowly localized state yields a large positive value.35) The appropriate ground state is φ0 (x) = mω π 1 4 exp − mωx2 2 . 4. . i.34) dx |φ0 (x)|2 = |c0 |2 −∞ −∞ dx exp − mωx2 2 = |c0 |2 π . The reason is that in the Hamiltonian (4. The conventional normalization condition φ0 |φ0 reads +∞ +∞ Ω∞ = 1 (4.e. We will consider this point more systematically in Section 5. .36) Since the first order differential equation (4..

40–4. √ a+ φn (x) = n + 1 φn+1 (x) . according to (4. n = 0. they obey φn |φn = δn 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) .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. To determine these constants we adopt a recursion scheme.51) . .40). . √ a− φn (x) = n φn−1 (x) .50) φn (x) .4. ˆ (4.47) for some suitable constants βn .45) form an orthonormal set.40) φn−1 (x) = βn a− φn (x) ˆ (4.48) (4.18)] and since the ˆ φn (x) are real functions.15 .41) Employing the adjoint property (4. To determine βn we note using (4. 1. . (4. . (4.45) We like to note that these functions according to (4.39) yield √ From this follows βn = 1/ n and.42) ˆ Using (4..43) (4. according to (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 . Since a− is a real differential operator [see (4. n.3: Excited States 79 We need to determine now the normalization constants cn . 2. Ω∞ 2 = βn a− φn |ˆ− φn ˆ a Ω∞ 2 = βn φn |ˆ+ a− φn a ˆ Ω∞ .20) yields 2 αn φn |ˆ− a+ φn a ˆ Ω∞ = 1. 1.28) and the construction (4. (4.24) holds in analogy to (4.20) 1 = φn−1 |φn−1 Equations (4. 2. n = 0.e. . i.40) = 1. For n = 0 holds c0 = 1.46) According to (4. (4. (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∞ (4.47). βn must be real as well. 4. .49) (4. We consider then the situation that we have obtained a properly normalized state φn (x). (4.

In terms of y the ground state wave function is φ0 (y) = π − 4 e− and a+ is ˆ 1 a+ = √ ˆ 2 y − d dy . For this purpose we start from expression (4.56) Relationship to Hermite Polynomials We want to demonstrate now that the expression (4. It should be noted that √ the normalization (4.80 Evaluating the Stationary States Linear Harmonic Oscillator We want to derive now an analytical expression for the stationary state wave functions φn (x) defined through (4.54) 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 . by dx = dy mω 1 4 . however. i.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) .40–4.. (4. (4.45). (4.58) This result agrees with the expression (2.e.28) of the ground state and the definition (4. n–independent factor.52).7 and given.7.52) This normalization of the wave functions differs from that postulated in (4. by the Rodrigues formula (2. (4. by introducing the variable y defined in (4.39). (4.200).35) by a constant.53) We will later re-introduce this factor to account for the proper normalization (4. simplifying the calculation. namely the square root of the Jacobian dx/dy.57) which includes the factor 1/ n! according to Eqs. 2.55) 1 y2 2 (4. 2.56) can be expressed in terms of Hermite polynomials Hn (y) introduced in Sect. We will demonstrate below the identity Hn (y) = e y2 2 y − d dy n e− y2 2 (4.233) derived in Sect. for example.30) and employing the normalization +∞ −∞ dy φ2 (y) = 1 n (4.46) yields a set of normalized states. .35) rather than (4.

dy n (4. 1. Our derivation will follow closely the procedure adopted for the case of a ‘particle in a box’ [see Eqs. hence.200). (3.30) and return later to the coordinate x.57) of the hermite polynomials and the definition given by (2.200) 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 .106–3.4.4 Propagator for the Harmonic Oscillator We consider now the solution of the time-dependent Schr¨dinger equation of the harmonic oscillator o (4.62) Denoting f (y) = (y − d/dy)exp(−y 2 /2) and employing y2 y2 d d − y2 e 2 f (y) = −y e− 2 f (y) + e− 2 f (y) dy dy (4. (4. (4.65) . For the sake of notational simplicity we employ initially the coordinate y as defined in (4.114)]. t0 ) = n=0 dn φn (y0 ) .61) One can factor g(y) and employ (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 . (4..61) and.59) dn −y2 e dy n (4. therefore. t0 ).60) and.59) by induction noting first that (4.1. dy dy (4. i. We prove (4. g(y) = (−1)n+1 e y2 2 dn+1 −y2 e = dy n+1 y − d dy n+1 e− y2 2 . 4.59) holds for n = 0.2) for an arbitrary initial wave function ψ(x0 .e. and showing then that the property also holds for n + 1 in case it holds for n.64) which implies that (4.39) ∞ ψ(y0 .4: Propagator 81 To verify the realtionship between the definition (4. 4.59) hold.63) one obtains g(y) = y − d dy f (y) (4. the Rodrigues formula (2. Starting point of our derivation is the assumption that the initial condition can be expanded in terms of the eigenstates φn (y) (4.

For this purpose we start from the integral representation (2. t1 ) = n=0 dn φn (y) cn (t) (4. 2 A demonstration of this property can be found in Special Functions and their Applications by N. N. Englewood Cliffs. 2.Lebedev (Prentice Hall.68) Altogether one can express then the solution +∞ ψ(y.65) to times t ≥ t0 through insertion of time-dependent coefficients cn (t1 ) ∞ ψ(y. 1965) Sect.108–3. 2n n! π 2 2 (4.7.69) where ∞ φ(y. 68. t0 ) which is an element of N1 defined in (4. t|y0 . this is an excellent textbook from which we have borrowed heavily in this Section. (4. a supposition which is not proven here2 .147) for the harmonic iscillator determined in Sect. t1 |y0 .. t) = n=0 Hn (y)e−y /2 Hn (y0 )e−y0 /2 n √ t .67) for which according to (3. of (4.5).46) the expansion coefficients dn are +∞ dn = −∞ dy0 ψ(y0 .h. (4.112) and (4. 4. pp. t0 ) = n=0 φn (y) φn (y0 ) tn+ 2 .66) One can extend expansion (4. t0 ) φn (y0 ) . 1 t = e−iω(t1 −t0 ) (4. Employing orthogonality condition (4.70) is the propagator of the linear harmonic oscillator. sec:harm.82 Linear Harmonic Oscillator Such expansion is possible for any f (y0 ) = ψ(y0 .147) we employ the technique of generating functions as in Sect. In order to prove the equivalence of (4.N.70) and (2. t1 ) = −∞ dy0 φ(y. t0 ) ψ(y0 .39) holds cn (t1 ) = exp −i ω ( n + 1 ) ( t1 − t0 ) 2 . t) = π −3/2 exp 2 2 y 2 y0 + 2 2 2 +∞ +∞ ∞ du −∞ −∞ dv n=0 1 ( −2tuv )n × n! (4.s.70).J. We consider for this purpose the following expression for |t| < 1 ∞ w(y. y0 . Inc.225) which allows one to derive a generating function for products of Hermite polynomials which can be applied to the r. y0 . .71) Applying (??) to express Hn (y) and Hn (y0 ) yields exp(−2tuv) w(y. We want to demonstrate now that this propagator is identical to the propagator (2.72) × exp( −u − v + 2iyu + 2iy0 v ) . t) (4.15..

(4.78) in terms of the coordinates x and x0 . t) = π −1 exp − 2 y 2 y0 + 2 2 +∞ dv exp(−(1 − t2 ) v 2 − 2i (yt − y0 ) v) . t) √ 1 π(1−t2 ) 2 exp − 1 (y 2 + y0 ) 1 + t2 + 2 y y0 2 1−t Hn (y)e−y /2 Hn (y0 )e−y0 /2 n ∞ √ t n=0 2n n! π 2 2 2 t 1 − t2 = . (4. a Re a2 > 0 (4. t0 ) = and 1 2 i π F (t1 . t0 ) = i φ(x. by a factor 4 mω/ for both φn (y) and φn (y0 ).74) a second time yields finally together with the definition (4.71) of w(y. √ πexp( t2 v 2 − 2iytv − y 2 ) (4.58) of the harmonic oscillator ∞ φn (y) φn (y0 ) tn = n=0 2 1 t 2 1 + t exp − (y 2 + y0 ) + 2 y y0 2 2) 2 1−t 1 − t2 π(1 − t 1 . t) = π −3/2 exp +∞ 83 y2 y2 + 0 2 2 +∞ dv exp( −v 2 + 2iy0 v ) × −∞ × −∞ du exp( −u2 − 2u(tv − iy) ) .70) is. y0 . it holds φ(y.79) 1 + e−2iω(t1 −t0 ) cosω(t1 − t0 ) = . t|x0 .4. i... t0 ) − i y y0 2 F (t1 .80) −2iω(t1 −t0 ) sinω(t1 − t0 ) 1−e We can finally express the propagator (4.s.30) to replace y and y0 and that we multiply the propagator by mω/ .73) The incomplete Gaussian integral +∞ −∞ dx e−a 2 x2 √ − 2bx = π b2 /a2 e . (4.78) 1 − e−2iω(t1 −t0 ) = sinω(t1 − t0 ) 2 i e−iω(t1 −t0 ) (4. . one obtains w(y. −∞ (4.75) Applying (4. t0 ) (4. (4. y0 .e.81) imω 2 sinω(t1 − t0 ) (x2 + x2 ) cosω(t1 − t0 ) − 2 x x0 0 This result agrees with the propagator (2. This requires that we employ (4. indeed. The resulting propagator is G(t1 . y0 .74) applied once results in w(y.h. i.147) derived by means of the path integral description. t0 ) = exp mω 2iπ sinω(t1 − t0 ) 1 2 × . t0 ) exp i 2 1 2 (y + y0 ) G(t1 . t0 ) = where F (t1 .e.76) One can express this in terms of the stationary states (4.77). identical to the generating function (4. t1 |y0 .77) The sum in (4.4: Propagator Carrying out the sum on the r.

(4. ˆ ˆ Calculus of Creation and Annihilation Operators We summarize first the key properties of the operators a+ and a− ˆ ˆ [ˆ− .82) a+ = √ ˆ dy dy 2 2 √ √ Obviously.84) = 1 1 = 0 √ = n + 1 φn+1 (y) √ = n φn−1 (y) = f |ˆ+ g a Ω∞ . In this case we define ∞ f (ν) (0) + ν (4.86) f (ˆ+ ) = a a ˆ ν! ν=0 . a− must be equivalent. i. 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 − . We want to introduce now techniques based on the ladder operators a+ and ˆ a− ..e.85) is convergent everywhere in R. 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. a (4.g. d/dy and a+ .84 Linear Harmonic Oscillator 4. the approaches a ˆ a ˆ using y. each corresponding to a single harmonic oscillator of the type studied presently by us. e. the use of generating functions. phonons.. hence. a+ ] a ˆ a− φ0 (y) ˆ a+ φn (y) ˆ a− φn (y) ˆ a− f |g Ω∞ ˆ We note that these properties imply √ φn (y) = (ˆ+ )n φ0 (y)/ n! . It is with quantum electrodynamics and solid state physics in mind that we cease the opportunity of the single quantum mechanical harmonic oscillator to develop a working knowledge for a+ and a− in the most simple setting.5 Working with Ladder Operators In the last section we have demonstrated the use of differential equation techniques. 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. and of the modes of the quantized electromagnetic field. (4.83) We will encounter below functions of a+ and a− . both quantum systems are endowed with a large number of modes. a− = √ ˆ y + . one can express y = 2(ˆ− + a+ ) and d/dy = 2(ˆ− − a+ ) and. Here f (ν) (y0 ) denotes the ν-th derivative of f (y) taken at y = y0 . 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.

An example is the so-called shift operator exp(uˆ− ). Hence. The following important property holds a a− f (ˆ+ ) = f (1) (ˆ+ ) + f (ˆ+ ) a− .87) (4.95) (4.89) (4. fn+1 (ˆ+ )] a a = [ˆ− .91) (4..4. For fn+1 follows then [ˆ− .87. a a (4. a− f (ˆ+ ) φ0 (y) = f (1) (ˆ+ ) φ0 (y) ˆ a a which follows from a− φo (y) = 0. property (4. a a a 85 (4. (ˆ+ )n ] a+ a a ˆ a a ˆ a a ˆ = (ˆ+ )n 1 + n (ˆ+ )n−1 a+ = (n + 1) (ˆ+ )n a 1 a ˆ a Since any function f (y) in the proper function space N1 can be expanded ∞ ∞ (1) (4. It holds a a euˆ f (ˆ+ ) φ0 (y) = f (ˆ+ + u) φ0 (y) . f1 (ˆ+ )] = [ˆ− . i.88) we need to show actually [ˆ− . 4. (ˆ+ )n a+ ] = (ˆ+ )n [ˆ− .88) is a fundamental one and will be used in the following.83) and can be proven for all powers fn by induction and then inferred for all proper functions f (ˆ± ).90) holds for f0 and for f1 .90) we show that (4.90) holds for any function fn (ˆ+ ) = (ˆ+ )n which is a power of a a a+ . a An important operator function is the exponential function.e.5: Working with Creation and Annihilation Operators and similarly for f (ˆ− ).90) holds for f (ˆ+ ). ˆ a We proceed by induction noticing first that (4.92) f (y) = n=0 dn φn (y) = n=0 (ˆ+ )n a dn √ φ0 (y) n! (4. The first case is trivial. ˆ a a a ˆ In particular. Note that ˆ a an immediate consequence of (4. a+ ] + [ˆ− .88) is (ˆ− )n f (ˆ+ ) φo (y) = f (n) (ˆ+ ) φ0 (y) . 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+ ).90) holds for fn . a a a ˆ 1 a Let us assume that (4. The convergence of the Taylor expansion ascertains then that (4. the second case follows from [ˆ− .94) This identity follows from (4. we will only assume operator functions acting on φ0 (y).88) (4. f (ˆ+ )] + f (ˆ+ ) a− ˆ a a a a ˆ which implies that in order to prove (4. a a a We like to state the following property of the functions f (ˆ± ) a f (ˆ− ) φ|ψ = φ|f (ˆ+ ) ψ . We note ˆ a− f (ˆ+ ) = [ˆ− . a a − (4. a+ ] = 1 = f1 (ˆ+ ) .93) one can reduce all operators acting on some proper state function by operators acting on the state φ0 (y).90) To prove (4.96) . f (ˆ+ )] = f (1) (ˆ+ ) .

euzˆ ˆ euz ∗ a− ˆ (4. + (4.100) ˆ ˆ To solve this differential equation we define C(u) = D(u) exp(−u2 /2) which leads to d ˆ D(u) = du zˆ+ + z ∗ a− a ˆ ˆ D(u) (4.105) a eαˆ + − α∗ a− ˆ φ0 (y) = e αα∗ 2 e−α ∗ a− ˆ a eαˆ φ0 (y) . Below we will use the operator identity which follows for the choice of v = α in (4.104) (4.99) ˆ C(u) = zˆ a zˆ+ + z ∗ a− a ˆ + +z a ˆ ∗ − a euzˆ euz ˆ − u C(u) .101) 1. a a ν! (4. + (4.98) for u = −α∗ and v = α yields e−α This together with (4. ∗ = exp(uzˆ a ˆ and determine its derivative d du ˆ C(u) a = zˆ+ euzˆ euz a + ∗ a− ˆ a + euzˆ z ∗ a− euz ˆ + ∗ a− ˆ = + zˆ+ + z ∗ a− a ˆ Using (4.102) Similarly. z ∈ C.90) and zz ∗ = 1 we can write this d du e uzˆ+ a e uz ∗ a− ˆ a − z ∗ a− . that the operators considered act on φ0 (y). + ∗ 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) .103) (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) .98) The related operators exp[vˆ+ ± v ∗ a− ] play also an important role. + (4. hence. 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) . (4. We assume in the following a ˆ derivation. one obtains a evˆ + − v ∗ a− ˆ a φ0 (y) = e− 2 vv evˆ e− v 1 ∗ + ∗ a− ˆ φ0 (y) (4. the solution of (4. 1 + + uz ∗ a− ).107) − α∗ a− ˆ φ0 (y) = e− αα∗ 2 a eαˆ φ0 (y) . + ∗ + (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.101) ˆ ˆ ˆ C(u) obviously obeys C(0) = 1 This results in D(0) = 1 and. without explicitly stating this.106) Applying (4.96) is applied is a a a euˆ evˆ φ0 (y) = ev (ˆ − + + + u) a φ0 (y) = euv evˆ φ0 (y) .108) .97) An example in which (4.

95) one obtains a a a a euˆ φ0 |evˆ φ0 = φ0 |euv evˆ φ0 = euv evˆ φ0 |φ0 = euv − + + + − (4. i. . Using (4.h.110) √ y2 u2 + 2uy − 2 2 a = euˆ φ0 (y) .s.84. We want to derive now the values Hn (0).4. Setting y = 0 in (4.98) and again (4. we can reproduce by means of the generating function (4. and ˆ l. the expected from which follows by comparision of each term on both sides φµ |φν orthonormality property. is the equivalent of the generating function (??).. provide the same values for Hn (0) as provided in Eq.58) one obtains π− 4 1 ∞ ν=0 (−1)ν u2ν = 2ν ν! ∞ ν=0 uν √ φν (0) = ν! ∞ ν=0 uν Hν (0) √ 2ν π ν! (4.115) and using (4. Expanding r.58) the Hermite polynomials Hn (y) by the eigenstates φn (y) √ ∞ 1 ( 2t)n y2 /2 2yt − t2 √ e φn (y) (4. 4. of (4.s.h. For this purpose we consider a a a a euˆ φ0 |evˆ φ0 = φ0 |euˆ evˆ φ0 + + − + (4. 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).111) yields π − 4 exp − 1 u2 2 a = euˆ φ0 (0) .84) yields ∞ ν.h.5: Working with Creation and Annihilation Operators Generating Function in Terms of a+ ˆ 87 We want to demonstrate now that generating functions are as useful a tool in the calculus of the ladder operators as they are in the calculus of differential operators. We start from the generating function (??) and replace according to (4.114) where we have employed (4.s.111) the orthonormality properties of the wave functions φn (y).109) e = π4 n! n=0 √ Using (4.84) and defining 2t = u one can write then π − 4 exp − or π − 4 exp − 1 1 √ y2 u2 + 2uy − 2 2 ∞ = n=0 un + n (ˆ ) φ0 (y) a n! + (4. in the ladder operator calculus. (4.95).e.µ=0 uµ v ν √ φµ |φν µ!ν! ∞ = µ=0 uµ v µ µ! (4.116) = δµ ν .113) Comparision of all terms on the l. + (4.112) Expanding both sides of this equation and using (4. Similarly.111) This expression. and on the r.s. (??).h.115) a where the latter step follows after expansion of evˆ and using a− φ0 (y) = 0.

For example.122). are ˜ En = or.124) ). for an oscillator in a stationary state φn (x) as given by (2. (4.. i.7) is 1 2 m 2 ω 2 d2 p − . using (2. ˆ and the Hamiltonian (4. ω = 1 / m2 ω .e. The eigenvalues. (4. i. The momentum representation eigenfunctions are. En = ω (n + 1 ) 2 d2 1 + m˜ 2 p2 ω 2 2m dp 2 2 ˜ ˜˜ φE (p) = E φE (p) (4.e.6 Momentum Representation for the Harmonic Oscillator The description of the harmonic oscillator provided so far allows one to determine the probability density P (x) of finding an oscillator at position x. (−i)n ˜ φn (p) = √ 2n n! m˜ ω π 1 4 exp − m˜ p2 ω 2 Hn ( m˜ ω p) . ˜ (4. the probability density is P (x) = |φn (x)|2 .37).6) and the Hamiltonian (4.117) In this section we want to provide a representation for the harmonic oscillator which is most natural ˜ if one wishes to determine for a stationary state of the system the probability density P (p) of finding the system at momentum p. As to be expected. n = 0. the momentum and position operators are as stated in (4. (4. 4. .235). .121) ω (n + ˜ 1 ) 2 . . 2. the time-independent Schr¨dinger equation for the oscillator can be written o ˆ H = − mω 2 2 2 x = i ˆ d .88 Linear Harmonic Oscillator 4.125) . of (4. .2.3). 1. (4..122) For a solution of the Schr¨dinger equation in the momentum representation. using (4. The solutions can be stated readily exploiting the earlier results.235). is given by the ˜ function ψ(p).2). (4. . is given by a function ψ(x).7) is given by (4. In the position representation the wave function is a function of x. according to (4.121).123) (4. 2. For this purpose we employ the Schr¨dinger equation in the momentum o representation. the momentum and position operators are p = p. n = 0. the eigenvalues are identical to those determined in the position representation.e. . 1. i.119) 2m 2 dp2 Accordingly. dp (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. In the momentum representation the wave function is a function of p.120) Multiplying this equation by 1/m2 ω 2 yields − where ˜ E = E / m2 ω 2 .118) d2 1 2 + p dp2 2m ˜ ˜ φE (p) = E φE (p) .

.128) We may also express the eigenstates (4.131) in the present case employing dimensionless units.6: Momentum Representation or with (4.58).e. defined only up to an arbitrary phase factor eiα .134). instead of (4.30) to obtain (4.130) dx exp(−ipx/ ) φn (x) −∞ (4. m ω (4. through direct integration. i. Proceed as follows.30) and (4.4. . According to the theory of the momentum representation holds ˜ ˜ Pn (p) = |φn (p)|2 or 1 ˜ Pn (p) = n 2 n! 1 πm ω 1 2 (4.125. ˜ We can now state the probability density Pn (p) for an oscillator with energy En to assume momentum p. m ω (4. of course. α ∈ R. From (4. 1 ˜ φn (k) = √ 2π +∞ (4.235)] absorbs the replacement dx → dy.134) Exercise 4.30) and (2. φn (k) = We want to finally apply the relationship 1 ˜ φn (p) = √ 2π +∞ 2 (−i)n −k √ e 2 Hn (k) .. the Jacobian.127) exp − p2 m ω 2 Hn ( 1 p) .129) follows ky = px/ and. 2n n! π (4.129) m ω one obtains. hence.132) dy exp(−iky) φn (y) −∞ (4. i.e.6.133) where we note that a change of the normalization of φn (y) [c. this allowed us to introduce in (4.122) (−i)n ˜ φn (p) = √ 2n n! 1 πm ω 1 4 89 exp − p2 2m ω Hn ( 1 p) .126) in terms of dimensionless units following the procedure adopted for the position representation where we employed the substitution (4.1: Demonstrate the validity of (4. the correctness of the phase factors (−i)n . 4.126) Normalized eigenfunctions are.f.126).126) a phase factor (−i)n . Defining k = m˜ / p or ω 1 k = p (4. (4. This implies the property of Hermite polynomials (−i)n e− k2 2 +∞ Hn (k) = −∞ dy exp(−iky) e− y2 2 Hn (y) (4.

134). pp.138) We will show that a harmonic oscillator prepared in such a state will remain forever in such a state. z) = exp 2yz − z 2 − y 2 /2 (4. a pendulum swinging at small amplitudes. The question arises if similar states exist also for the quantum oscillator. z) = √ 2π where f (k.— III.140) An excellent textbook is Photons and Atoms: Introduction to Quantum Electrodynamics by C. a useful role in Quantum Electrodynamics since they allow to reproduce with a quantized field as closely as possible the properties of a classical field3 . Such states are referred to as quasi-classical states. Inserting this expansion into √ (4. or Glauber states of the harmonic oscillator and they play.139) where we have added the condition that the states be normalized. We want to construct and characterize such states. +∞ Linear Harmonic Oscillator dy exp(−iky) g(z.Grynberg (John Wiley & Sons. For this purpose we assume such state exists and expand ∞ ∞ φ(α) (y) = n=0 d(α) φn (y) . The answer is ‘yes’.Chen-Tannoudji. The quasi-classical states can be obtained by generalizing the construction of the ground state of the oscillator. except that α changes periodically in time. 4. z) = exp −2ikz + z 2 − k 2 /2 . Inc.138). coherent states. φ(α) | φ(α) = .Dupont-Roc. 192 3 . We first construct the solution of (4. ˆ α ∈ R. 4. (4.4. J..1 showing in its top row the attributes of such state just discussed. for example.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.136) (b) Employ the generating function (2.138) 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 (4.7 Quasi-Classical States of the Harmonic Oscillator A classical harmonic oscillator.194. The motion of the probability distribution of such state is presented in Fig. namely through the eigenvalue problem for normalized states a− φ(α) (y) = α φ(α) (y) . 1989) which discusses quasi-classical states of the free electromagnetic field in Sect.196) of Hermite polynomials and equate coefficients of equal powers of z to prove (4.C. 2.90 (a) Prove first the property 1 f (k. for example. New York. and G. y) −∞ (4. We will find that α can be characterized as a displacement form the minimum of the oscillator and that the state φ(α) (y) displays the same spatial probability distribution as the ground state. n n=0 d(α) n 2 = 1 (4.135) g(k. carries out a periodic motion described by y(t) = Re y0 eiωt (4.

.7: Quasi-Classical States or ∞ n=0 91 √ n + 1 dn+1 − α d(α) n (α) φn (y) = 0 .141) Because of the linear independence of the states φn (y) all coefficients multiplying φn (y) must vanish and on obtains the recursion relationship (α) dn+1 = √ α d(α) .138). n! (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) . we will see below that α ∈ C are also admissible and will provide a corresponding interpretation. n! n (4.144) Normalization requires c = exp(−α2 /2) and. This interpretation justifies the choice of α ∈ R in (4.4. n! (4.143) for a constant c which is determined through the normalization condition [see (4.84) φ(α) (y) = exp − α2 2 ∞ n=0 αα∗ (αˆ+ ) a a+ φ0 (y) = e− 2 eαˆ φ0 (y) . However. .150) for α presently real. hence. n! (4.146) √ 1 − (y − 2 α)2 2 (4.108) allows one to write φ(α) (y) = T + (α) φ0 (y) = T + (α) [π]− 4 e−y where a T + (α) = eαˆ + 1 2 /2 (4.149) − α∗ a− ˆ (4.142) The solution is αn d(α) = √ c n n! ∞ ∞ (4.e. One can also write (4. Comparing (4.139)] 1 = n=0 d(α) n 2 = |c|2 n=0 α2 2 = |c|2 eα .148) The identity (4.109). i.147) √ which identifies this state as a ground state of the oscillator displaced by 2α.149) with (4. (4.145) using (4. n n+1 (4.145) Using the generating function in the form (4. the quasi-classical states are φ (α) (y) = exp α2 − 2 ∞ n=0 αn √ φn (y) .147) allows one to interprete T (α) as an operator √ which shifts the ground state by 2α.

In particular.154) .h.h.145) shows that α in (4.95) to linear combinations αˆ+ + βˆ− and reads then a a f (αˆ+ + βˆ− ) φ|ψ = φ|f (α∗ a− + β ∗ a+ ) ψ .152) Since the operator function on the r.109) serves again to simplify this sum   √  y2 1 1 α2 (u2 + 1)   φ(α) (y.160) . .145) as the initial state φ(α) (y.s.145) is replaced by a complex number αu which at times t1 = t0 + 2πn/ω.e.152) T + (α) f |T + (α) g = f |T (α) T + (α) g = f |g . n = 0. t1 ) = π − 4 u 2 exp  − + 2yαu −  2  2 E (4.158) Comparision of this expression with (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 . becomes real. (4. (4.150) and one can conclude that T + (α) is a so-called unitary operator with the property T + (α) T (α) = T (α) T + (α) = 1 1 or applying this to a scalar product like (4.155) (4.156) (4. n! u = e−iω(t1 − t0 ) . . 4.150) for complex α obeys the unitary property T + (α) T (α) = 1 1 (4. one can write in case of T + (α) T + (α) + = T (α) = eα ∗ a− ˆ − αˆ+ a a = e − (αˆ + − α∗ a− ) ˆ . i. 1.151) which follows from a generalization of (4.s. . t1 ) = exp or φ (α) − α2 2 ∞ exp n=0 −i ω ( n + ∞ n=0 1 ) ( t1 − t0 ) 2 αn √ φn (y) n! (4.67. α ∈ C (4. Following the procedure adopted in determining the propagator of the harmonic oscillator [see (4.153) This is obviously the inverse of T + (α) as defined in (4. Relationship (4. t1 ) = u exp 1 2 α2 − 2 (α u)n √ φn (y) . is the adjoint of the operator function on the l..92 Linear Harmonic Oscillator The shift operator as defined in (4.68)] one can write the time-dependent solution with (4.159) Noticing the identity which holds for |u|2 = 1 u2 + 1 = (Re u)2 − (Im u)2 + 1 + 2 i Re u Im u = 2 (Re u)2 + 2 i Re u Im u (4. a a ˆ ˆ (4.157) (y.

138): a real α corresponds to a displacement such that initially the oscillator is at rest.150) with (4.158) allows one to write 1 φ(α) (y. The diagram illustrates that the wave function retains its Gaussian shape with constant width for all times. this characterization corresponds closely to that of the possible initial states of a classical oscillator. 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.1 (top row) we present the probability distribution of the Glauber state for various instances in time.159) is then 1 1 i φ(α) (y.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. 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. The time-dependent wave function (4. t1 ) = π − 4 exp − [y − y0 (t1 )]2 + iy k0 (t1 ) − iφ0 (t1 ) − ω(t1 − t0 ) 2 2 . Arbitrary Gaussian Wave Packet Moving in a Harmonic Potential We want to demonstrate now that any initial state described by a Gaussian shows a time dependence very similar to that of the Glauber states in that such state remains Gaussian. in general. t1 ) = u 2 T + (α u) φ0 (y) (4. (4.163) This solution corresponds to the initial state given by (4. In Fig. Comparing (4.162) 2 with a periodic change.165) .163) also through the shift operator. 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. a momentum factor and a phase factor.138. being displaced around the center with period 2π/ω and. 4.4. experiences a change of its width (relative to . (4.164) where according to (4.150) T + (α u) = exp α ( e−iω(t1 −t0 ) a+ − eiω(t1 −t0 ) a− ) ˆ ˆ This provides a very compact description of the quasi-classical state. This shows clearly that the Glauber state is a close analogue to the classical oscillator. obviously. moving solely its center of mass in an oscillator fashion around the minimum of the harmonic potential. One can express (4.147). 4.148.7: Quasi-Classical States the exponent E can be written E = − √ √ y2 + 2 αy Reu − α2 (Re u)2 + i 2 αy Imu − i α2 Reu Imu 2 √ √ 2 1 = − y − 2αReu + i 2yα Imu − i α2 Reu Imu 2 phase term momentum displaced Gaussian 93 (4. 4. a complex α corresponds to a displaced oscillator which has initially a non-vanishing velocity.

2σ 2 q = a . σ|y.1: Time dependence of Gaussian wave packets in harmonic oscillator potential.166) φ(ao . One can recognize the cyclic displacement of the wave packet and the cyclic change of its width with a period of twice that of the oscillator.1 (second and third row).169) . t0 ) = n=0 dn φn (y) . To describe such state which satisfies the initial condition φ(ao . that of the Glauber states) with a period π/ω. 1 + σ2 (4. 4. t0 ) +∞ = √ 1 n n!πσ 2 e−ao q/2 = √ 2n n!πσ dy exp −∞ +∞ −∞ − (y − ao )2 y2 − 2σ 2 2 2 Hn (y) (4. σ|y.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. (4. This behaviour is illustrated in Fig.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 . σ|y. t0 ) = [πσ 2 ]− 4 exp we expand ∞ 1 − (y − ao )2 2σ 2     (4.

Marichev (Wiley.166) in analogy to (4. 1 (4.157) as follows φ(ao .I. 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.172) One can then conclude e−a /4 n lim dn = √ a σ→1 2n n! (4. p−1 1 − σ4 t = 1 − σ 2 −iω(t1 −to ) e 1 + σ2 (4. In this limit the coefficients dn should√ become identical to the coefficients (4.171) lim p−1 p n 2 Hn (q p ) = lim (2q)n = an . σ→1 O(y n−2 ) for n ≥ 2 1 for n = 0. t1 ) = √ × where yo = q p a = √ .177) − 1 ω(t1 − to ) 2 . p−1 95 (4. 1990). therefore. New York. i. Following the strategy adopted previously one can express the time-dependent state corresponding to the initial condition (4.P.174) Hn (q p p−1 ) φn (y) . 1 a2 2 1 + σ2 + 1 2 2 y o + 2 y yo This integral can be found in Integrals and Series. σ|y. vol. this 3 volume integral table is likely the most complete today..170) In order to check the resulting coefficients we consider the limit σ → 1. and O. t1 ) = e−aq/2 √ pσ ∞ n=0 e −iω(n+ 1 )(t1 −to ) √ 2 1 2n n! p−1 p n/2 (4. σ|y.76) permits us to write this φ(ao . a worthy successor of the famous Gradshteyn.4.e. σ→1 p−1 2 (4.143) of the Glauber state. in agreement with the expected result (4. Brychkov. Prudnikov.168) is4 e−aq/2 (p − 1)n/2 dn = √ Hn (q 2n n!σ p(n+1)/2 p ). This expansion can be written φ(ao . namely $750 4 . 2 by A. unfortunately very expensive. Yu.175) ∞ tn n=0 2n n! Hn (yo ) e−yo /2 Hn (y) e−y 2 2 /2 The generating function (4. in fact.143). σ|y. t1 ) = √ 2 × exp − 1 (y 2 + yo ) 1+t2 − 2 1−t 2 1 1 √ √ pσ π 1 − t2 × t 1 − t2 (4.173) which is.176) 1 √ pσ π a 2 exp − 1 1+σ2 + 1 yo − 1 ω(t1 − to ) 2 2 2 2 × (4. a ground state shifted by ao = 2α.7: Quasi-Classical States The solution of integral (4.

96 Linear Harmonic Oscillator .

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

From (5.7) Let us consider briefly an example to illustrate rotational transformations and to interpret the sign of detR(ϑ).(5.. i. =1 [R(ϑ)]jk [R(ϑ)]j ak b = j=1 aj bj . 0.3) Since this holds for any a and b. ϕ)T ) is negative. x2 .k. ϑ parametrizes the rotations uniquely as long as |ϑ| < 2π. (5. i.6) (5. where ϑ denotes the rotation. the matrix represents a proper rotation.e. x3 )T . It holds then j=1 3 3 a ·b = j.5) This equation states the key characteristic of rotation matrices. 2. in Cartesian coordinates r = (x1 .7) implies that a rotation is associated with an inversion. 3 . are linear and. i. therefore. (5. 0. (5.e. they conserve a · b = 3 aj bj . (5. (5. In the latter case the determinant of R(ϑ = (0.e.2) Rotations conserve the scalar product between any pair of vectors a and b. then the fingers of your fist point in the direction of the rotation. namely around an axis given by the direction of ϑ and by an angle |ϑ|. We assume the convention that rotations are right-handed. In the following we want to exclude inversions and consider only rotation matrices with detR = 1.6) follows immediatety for the determinant of R(ϑ) using detA B = detA detB and detAT = detA detR(ϑ) = ±1 .1) In Cartesian coordinates this reads 3 xk = j=1 [R(ϑ)]kj xj k = 1. a minus sign in Eq. in case of a prefactor −1 a rotation and an inversion at the origin. if the thumb in your right hand fist points in the direction of ϑ. .4) With the definition of the transposed matrix RT [RT ]jk ≡ Rkj this property can be written R(ϑ) RT (ϑ) = RT (ϑ) R(ϑ) = 1 1 . ϕ)T ) = ±  sinϕ cosϕ 0  (5.8) 0 0 1 In case of a prefactor +1. can be represented by 3 × 3 matrices R(ϑ). (5.98 Properties of Rotations in R Theory of Angular Momentum and Spin Rotational transformations of vectors r ∈ R . Let us define the rotated vector as r = R(ϑ) r .. it follows 3 [R(ϑ)]jk [R(ϑ)]j = δk j=1 . A rotation around the x3 -axis by an angle ϕ is described by the matrix   cosϕ − sinϕ 0 R(ϑ = (0.

(ii) The group element ‘e’ is played by the 3 × 3 identity matrix 1 1. b ∈ G the product c = a ◦ b is also in G. We would like to point out the obvious property that for all elements R of SO(3) holds R−1 = RT (5. It follows that R3 is also an element of SO(3).10) detR3 = det ( R1 R2 ) = detR1 detR2 = 1 . (iii) ∀a. a ∈ G → ∃a−1 ∈ G with a ◦ a−1 = a−1 ◦ a = e. For this purpose we must demonstrate that the group properties (i–iv) hold.5. T R1 R1 −1 = 1 −1 = 1 1 1 (5. (ii) There exists an element e in G with the property ∀a ∈ G → e ◦ a = a ◦ e = a. In the following we will assume R1 . R RT = RT R = 1 detR = 1 } 1.13) . R3 = R1 R2 is a real. We need to demonstrate that this inverse belongs also to −1 T SO(3). R2 ∈ SO(3) and do not write out “◦” explicitly since it represents the matrix product. it holds 1. (5. there exists a real 3 × 3–matrix −1 R1 which is the inverse of R1 . We have shown altogether that the elements of SO(3) form a group. forms a group G satisfying the axioms (i) For any pair a.1: Matrix Representation of SO(3) Definition of the Group SO(3) We will consider then the following set SO(3) = { real 3 × 3 matrices R.9) This set of matrices is closed under matrix multiplication and.12) (iv) Since the associative law holds for multiplication of any square matrices this property holds for the elements of SO(3). We want to prove now that SO(3) as defined in (5. in fact. It holds T T T T R3 R3 = ( R1 R2 )T R1 R2 = R2 R1 R1 R2 = R2 R2 = 1 . 99 (5.9) forms a group. hence. (iv) For products of three elements holds the associative law a ◦ (b ◦ c) = (a ◦ b) ◦ c. one can show R3 R3 = 1 Furthermore. Since (R1 )T = (R1 )−1 it follows −1 −1 −1 T (R1 )T R1 = (R1 )−1 R1 = −1 which implies R1 ∈ SO(3). (i) Obviously. (5.11) (iii) From detR1 = 0 follows that R1 is non-singular and. 1 T Similarly. 3 × 3–matrix.

16) must be antisymmetric. i. we arrive at the property exp ϑk LT k from which follows LT = −Lk . −1 0 0 −1 .19) (5. rotation by 90o .20) (5. Geometric intuition tells us R(ϑk ek )−1 = R(−ϑk ek ) = exp ( −ϑk Lk ) ˆ ˆ Using the property which holds for matrix functions [f (R)]T = f (RT ) we can employ (5. Lie Group. 0 1 −1 0 . mappings between the sets which conserve the group properties: (a) the set of real and imaginary numbers {1.14) together with matrix multiplication as the binary operation ‘◦’. k = exp ( −ϑk Lk ) (5. 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.1.17) .e. J2 . compact representation of R(ϑ) which will be of general use and. Generators. −1. x2 –plane {rotation by Oo . 0 −1 1 0 (5. rotation by 270o } and consecutive execution of rotation as the binary operation ‘◦’. M2 . Lie Algebra We want to consider now a most convenient. the matrices i Lk = − Jk (5. we want to assume that they exist. M4 } = 1 0 0 1 . in fact. establish if 1-1 homomorphic mappings exist.15) Below we will construct appropriate matrices Jk . i.100 Theory of Angular Momentum and Spin Exercise 5. −i} together with multiplication as the binary operation ‘◦’.18) (5. will play a central role in the representation of many other symmetry transformations in Physics. identify subgroups.13) R(ϑk ek )−1 = R(ϑk ek )T = [exp ( ϑk Lk )]T = exp ϑk LT ˆ ˆ k and. (b) the set of matrices {M1 . J3 as follows i R(ϑ) = exp − ϑ · J (5. rotation by 180o . We want to show that for the property R RT = 1 to hold. This representation expresses rotation matrices in terms of three 3 × 3– matrices J1 .1: Test if the following sets together with the stated binary operation ‘◦’ form groups. presently. M3 . (c) the set of four rotations in the x1 . due to the uniqueness of the inverse.

Exercise 5.24) (5. i. (5. 0)T the rotation is the identity.22).25)     0 −ϑ3 0 0 −1 0  ϑ3 0 0  =  1 0 0  .1: Matrix Representation of SO(3) 101 We can conclude then that if elements of SO(3) can be expressed as R = eA . L3 which are independent of the rotation angles. 0)T ..27) . We assume now that we parameterize rotations through a three-dimensional rotation vector ϑ = (ϑ1 . This property gives A then only three independent matrix elements. ϑ3 )T such that for ϑ = (0. L2 . ϑ3 and three matrices L1 .22) for k = 3 yields L3 = ϑ−1 3 One obtains in this way 0 0 0 i − J1 = L1 =  0 0 −1 0 1 0  0 0 1 i − J2 = L2 =  0 0 0 −1 0 0  0 −1 0 i  1 0 0 − J3 = L3 = 0 0 0 The matrices Lk are called the generators of the group SO(3). 0.15) as the derivatives of R(ϑ) with respect to the Cartesian components ϑk taken at ϑ = (0.15) with three real parameters ϑ1 . 0. Using the definition of partial derivatives we can state L1 = lim ϑ−1 R(ϑ = (ϑ1 . Let us use this definition to evaluate L3 .5.21) −b −c 0 It is this the reason why one can expect that three-dimensional rotation matrices can be expressed through (5.(5.2: Determine the generator L1 according to Eq.e. Using (5. 0. One can determine then the matrices Lk defined in (5. ϑ2 . ϑ3 )T ] = lim  sinϑ3 cosϑ3 0  =  ϑ3 ϑ3 →0 ϑ3 →0 0 0 1 0 0 1 Insertion into (5.8) we can state    cosϑ3 −sinϑ3 0 1 −ϑ3 0 1 0  . A must have the form   0 a b A =  −a 0 c  . 0. ϑ2 . 0 0 0 0 0 0  (5.26) (5. 0)T ) − 1 1 1 ϑ1 →0 (5.23) (5. the real matrix A is anti-symmetric.        (5. lim R[(0.1.22) and similar for L2 and L3 .

In fact. commutators of commutators of A and B. J ] = i k m Jm .102 Theory of Angular Momentum and Spin Sofar it is by no means obvious that the generators Lk allow one to represent the rotation matrices for any finite rotation. B].28) The construction of the generators through the limit taken in Eq.8) in case of ϑ = (0. = 2. B]. 0)T ) = exp( 1 L1 ). Z3 = 12 ( [A. e.e.15) and their generators obey the property (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. [A. and the constants fk m are called the structure constants. however. The question is then. as we see later. in fact. the elements of if any two indices are identical for k = 1. (5. Finite rotations can be obtained by applications of many infinitesimal rotations of the type R(ϑ = ( 1 . holds the algebra [Jk . and x3 -axes. Z2 = 2 [A.33) We want to demonstrate now that exp(ϑ · L) yields the known rotational transformations. however. The expression for Zn are actually rather complicated.g. R(ϑ = (0. L ] = k m Lm (5. the property (5.32) For the matrices Jk = i Lk which. m = 3 . B]] + [[A. 0. i. the matrix (5. are related to angular momentum operators.29) where Z1 = A + B and where the remaining operators Zn are commutators of A and B.30) Groups with the property that their elements can be expressed like (5. only the latter example and . m = 3 for any odd permutation of k = 1.31) where k m are the elements which are   0   1 k m =  1   −1 of the totally antisymmetric 3-dimensional tensor. = 2.g.. We want to consider. 3 )T ) = exp( 3 L3 ).. that the generators represent infinitesimal rotations with 1 >> |ϑ| around the x1 -. (5. ϕ)T . (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. An answer can be found considering the Baker-Campbell-Hausdorf expansion ∞ exp( λA ) exp( λB ) = exp( n=1 λn Zn ) (5. From this result one can conclude that expressions of the type (5. 2 . it must hold 3 [Lk .30) are called Lie groups. 0. i. (5. In case of the group SO(3) the structure constants are particularly simple. x2 -.30) is called the Lie algebra. if the resulting products of exponential operators can be expressed in the form (5.15).e. m = 3 for any even permutation of k = 1.15) obey the group property.22) implies. L ] = m=1 fk m Lm . 0)T ) = exp( 2 L2 ) and R(ϑ = (0. 1 1 e. Of course. 0.. = 2. three being a special case. the Lie algebra of SO(3) can be written [Lk . that the operators (5. Lie groups can have any number of generators. B] ). etc.

(5. accordingly.35). We note.41) cos ϕ − sin ϕ sin ϕ cos ϕ (5. 0 −1 1 0 = (−1)n 0 −1 1 0 (5. Exercise 5.27).  cos ϕ − sin ϕ 0 =  sin ϕ cos ϕ 0  0 0 1  (5.5.36) The property 0 −1 1 0 allows one to write 0 −1 1 0 and. using (5.8). (5. that the matrix ϕL3 .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 .34.31).1. A = ϕ 0 −1 1 0 .40) . according to (5.3: Show that the generators Lk of SO(3) obey the commutation relationship (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 .39) Recognizing the Taylor expansions of cos ϕ and sin ϕ one obtains eA = and. 5. (2n + 1)! (5. ∞ 2n = − 1 0 0 1 2n+1 (5. eϕL3 which agrees with (5. (5. can be written ϕL3 = 0 103 A 0 0 0 0 .37) = (−1)n 1 0 0 1 .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 .1: Matrix Representation of SO(3) determine exp(ϕL3 ).

its elements can be represented through an exponential R(ϑ) = exp i − ϑ·J (5.2. we define rotations R(ϑ) as linear maps C∞ ( ) → C∞ ( ). = ψ([ R(ϑ1 ) R(ϑ2 ) ]−1 r) = ρ R(ϑ2 ) ψ([R(ϑ1 )]−1 r) (5.44) holds. 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. i.104 Theory of Angular Momentum and Spin 5.e. i.42) Obviously. for A. (5. namely R(ϑ)ψ(r ) = ψ(R−1 (ϑ) r ) .45) (5. Definition We first define how a rotational transformation acts on a wavefunction ψ(r ).e.1: Assume the definition ρ ρ (A B) = ρ (B) ρ (A). one isomorphic to SO(3) and. B ∈ SO(3) holds ρ(A B) = ρ(A) ρ(B) .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) .. For this purpose we require stringent continuity properties of the wavefunction: the wavefunctions under consideration must be elements of the set C∞ ( ). (5.47) = ρ R(ϑ1 ) R(ϑ2 ) . Exercise 5. (5. in fact.43) ρ( ) conserves the group property of SO(3). hence. the R(ϑ) ψ(r) = ψ(R(ϑ) r ). the transformation R(ϑ): C∞ ( ) → C∞ ( ) is related to the transformation R(ϑ): R → R . i.e. a relationship which we like to express as follows R(ϑ) = ρ R(ϑ) .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). complex-valued functions over the 3-dimensional space R which can be differentiated infinitely often. we will learn that transformation patterns and invariances are connected with the angular momentum states of quantum mechanics.46) (5. In particular. Since this holds for any ψ(r ) one can conclude ρ R(ϑ1 ) ρ R(ϑ2 ) group property (5.48) .

show that (5. .49) (note the minus sign of −ϑ1 which originates from the inverse of the rotation in Eq. The generators Jk can be determined by applying (5. (5.(5.52) ( x2 ∂1 − x1 ∂2 ) f (r) .56) holds for [J2 . 0. J ] = i k m Jm ..2. J2 ] f (r) = − = − 2[ 2 [− i J .24) one can expand  ϑ3 →0 ϑ3 →0 lim ϑ−1 R(ϑ = (0.53) (5.51) Inserting this into (5.56) We like to verify this property for [J1 . 5.42)) and similarly for J2 and J3 . ϑ3 )T ) f (r) − f (r) 3 (5.e. (5. 0.  0 0 1 x3 x3 (5.49).23.2: (a) Derive the generators Jk . J2 ]. in case of J3 .55) Not only is the group property of SO(3) conserved by ρ( ). −ϑ3 ) r ) − f (r) ) .5. (5. 3     1 ϑ3 0 x1 x1 + ϑ3 x2 R(0.50) and Taylor expansion yields i − J3 f (r) = = ϑ3 →0 lim ϑ−1 ( f (r) + ϑ3 x2 ∂1 f (r) − ϑ3 x1 ∂2 f (r) − f (r) ) 3 (5. J1 ]. but also the Lie algebra of the generators. i − J3 f (r) = = Using (5. −ϑ3 ) r ≈  −ϑ3 1 0   x2  =  −ϑ3 x1 + x2 .53–5. One obtains using (5.22) according to i − J1 = lim ϑ−1 R(ϑ = (−ϑ1 .54) (5. it holds [Jk . k = 1.55) and f (r) = f [J1 . 0. 0.50) lim ϑ−1 ( f ( R(0. Carrying out similar calculations for J1 and J2 one can derive i − J1 i − J2 i − J3 = = = x3 ∂2 − x2 ∂3 x1 ∂3 − x3 ∂1 x2 ∂1 − x1 ∂2 .49) to a function f (r). − i J ] f (r) 1 2 =− 2× [ (x3 ∂2 − x2 ∂3 ) (x1 ∂3 f − x3 ∂1 f ) − (x1 ∂3 − x3 ∂1 ) (x3 ∂2 f − x2 ∂3 f ) ] 2 +x1 x3 ∂2 ∂3 f − x1 x2 ∂3 f − x2 ∂1 ∂2 f + x2 x3 ∂1 ∂3 f + x2 ∂1 f 3 2(x ∂ 2 1 2 −x1 ∂2 f − x1 x3 ∂2 ∂3 f + x2 ∂1 ∂2 f + x1 x2 ∂3 f − x2 x3 ∂1 ∂3 f ] 3 = − − x1 ∂2 ) f (r) = i J3 f (r) . (5. J3 ] and [J3 .57) Exercise 5.2: Function Space Representation of SO(3) where the generators can be determined in analogy to Eq. i. 0)T ) − 1 1 ϑ1 →0 105 (5. 2 by means of limits as suggested in Eq. (5. In fact.

φ). (5. . J3 ] = = = 2 2 [J1 + J2 . the system related to the Cartesian coordinates through x1 = r sinθ cosφ.58) i = r ×p ˆ (5. Using [AB. [J 2 . Show that for J3 as defined above holds exp( iα J3 ) ψ(φ) = ψ(φ + α). J3 ] + [J2 . 2. To show this we first note that (5. (5. φ) = = 2 ( + 1) Y m (θ. J3 ] (5. φ) as well as the effect of the so-called raising and lowering operators J± = J1 ± i J2 (5.59) which.61) We demonstrate this property for k = 3.e. i. C] + [A.55) can be written as a vector product of r and i − J = −r × From this follows. k = 1. k = 1. We will show now that the properties of spherical harmonics J2Y J3 Y m (θ. C]B and (5.106 Theory of Angular Momentum and Spin Exercise 5. identifies J as the quantum mechanical angular momentum operator.63) (5. according to the correspondence principle between classical and quantum mechanics. The commutation relationships imply that no simultaneous eigenstates of all three Jk exist. Jk ] = 0 . J3 ] + [J1 .62) J1 [J1 .64) mY m (θ. J3 ] J2 −i J1 J2 − i J2 J1 + i J2 J1 + i J1 J2 = 0 According to (5. x2 = r sinθ sinφ.60) commutes with all three Jk . J3 ] J1 + J2 [J2 . φ). However..61) simultaneous eigenstates of J 2 and of one of the Jk . 2. i.53–5. using the momentum operator p = ˆ J = r × i . θ. 5. can be found.2. and x3 = r cosθ. 3 .56) one obtains [J 2 . 3. Equation (5.e. φ) m (θ.3 Angular Momentum Operators The generators Jk can be readily recognized as the angular momentum operators of quantum mechanics. C] = A[B.56) are the famous commutation relationships of the quantium mechanical angular momentum operators. usually chosen as J3 .3: Consider a wave function ψ(φ) which depends only on the azimutal angle φ of the spherical coordinate system (r.65) . the operator 2 2 2 J 2 = J1 + J2 + J3 (5. φ) (5. These eigenstates are the well-known spherical harmonics Y m (θ.

and since we have already shown | − 1 = (iβ −1 )−1 L− | we can state L3 | − 1 = −i( − 1)| − 1 .68) (5. namely. It holds L+ L− | = (L+ L− − L− L+ + L− L+ )| = ([L+ . L− ] + L− L3 ) | .e.73) (5. For this purpose we first demonstrate that the property holds for m = − 1 by considering (i) for the state L− | ∼ | − 1 . k = 1.69) J+ Y (θ. We note L3 L− | = ( [L3 . φ) are a consequence of the Lie algebra (5.75) (5. For this purpose we prove the following theorem. 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.e.70) (5. 3 be operators acting on a Hilbert space H which obey the algebra [Lk . L1 − iL2 ] = −2iL3 and L+ | = 0 one obtains L+ L− | = −2iL3 | = −2 | . L− ] + L− L+ )| . φ) 107 = = 0 = = 0 ( + m + 1)( − m) Y ( + m + 1)( − m) Y m+1 (θ. L− ] = [L3 . φ) (5. Using [L3 . An Important Theorem Theorem 1.1 Let Lk . (∗) (5. L1 − iL2 ] = L2 + iL1 = i (L1 − iL2 ) = iL− and L3 | = −i | we obtain L3 L− | = ( iL− + L− L3 )| = i(− + 1) L− | . Noting [L+ . The definitions of the coefficients α −1 and β −1 yield L+ L− | = L+ ( iβ −1 | − 1 = − α −1 β −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. 2. L+ raises the m-value of | − 1 from m = − 1 to m = . i.66) (5.56). The arguments are very similar to the ones used above and we can be brief. Continuing our proof through induction we assume now that property (i) holds for m. L2 = L2 + L2 + L2 .71) To prove this theorem we first show by induction that (i) holds.72) = 0 = −i | (5.74) (5. φ) m+1 (θ. L− ] = [L1 + iL2 . φ) − m (θ. L− ] + L− L+ )| m = ( −2iL3 + ..67) (5. We consider L+ L− | m = ( [L+ .3: Angular Momentum Operators on the spherical harmonics. J+ Y J− Y J− Y m (θ. We will show that this property holds then also for m − 1. φ) (θ. in fact. α −1 β −1 = 2 . i. From this follows that L− | is an eigenstate of L3 .5.

i..55). One can normalize the states | m such that αm = βm .e.72.82) (5. from L+ L− | m = − αm−1 βm−1 | m follows the recursion relationship αm−1 βm−1 = 2m + αm βm . We note that we can write L2 = 1 1 1 1 2 2 2 2 L+ L− + 2 L− L+ + L3 . 5. (5. Again the argumemts are similar to the ones used above. Using the familiar formula + m − n k=0 k = n (n + 1)/2 − (m + 1) m = ( + 1) m − (m + 1) m (5.84) ..77) We will show now that if (ii) holds for m. We note L3 L− | m = ( [L3 . 5. This implies that (i) holds for m − 1.83) (5. L− ] + L− L3 )| m = ( iL− − i m L− )| m = −i (m − 1) L− | m from which we can deduce L3 | m − 1 = −i(m − 1)| m − 1 . it also holds for m − 1. 5. (∗∗) (5. A first application will involve the construction of the eigenstates of the angular momentum operators as defined in (5. φ.66. 5.64).73. in terms of spherical coordinates defined through x1 x2 x3 = = = r sin θ cos φ r sin θ sin φ r cos θ (5.53–5. (5.70. it holds αm βm = 2 one obtains αm βm = = ( + 1) k=m+1 k .e. For this purpose we need to express the angular momentum operators in terms of spherical coordinates (r. This completes the proof of the theorem. It follows then L | m = − 2 αm−1 βm−1 + 2 αm βm + m | m = 1 − 2 ( + m + 1)( − m) + 1 ( + m)( − m + 1) + m2 = − ( + 1)| m . We will have various opportunities to employ the Theorem just derived.80) (5. yield properties (5.79) read L+ | m L− | m + 1 = −i = −i ( + m + 1)( − m) | m + 1 ( + m + 1)( − m) | m .79) We can now show that (iii) holds for all proper m–values. given in (5.78) ( + m + 1)( − m) ( + m + 1)( − m) . finally αm = βm = (5.81) i.63. We can deduce from (∗) and (∗∗) α α α −1 β −1 −2 β −2 −3 β −3 = 2 = 2( − 1 + ) = 2( − 2 + · · −1 + ) Obviously. in particular. Before we can finally show that property (iii) holds as well for all − ≤ m ≤ we need to determine the coefficients αm and βm . 5. Angular Momentum Operators in Spherical Coordinates We want to express now the generators Jk .108 Theory of Angular Momentum and Spin L− L+ )| m = (−2m − αm βm )| m .69). θ. 2 We note that Eqs.

3: Angular Momentum Operators 109 We first like to demonstrate that the generators Jk . L1 f (θ. (L2 )2 = − cos φ ∂ ∂ + cot θ sin φ ∂θ ∂φ 2 .93) and similarly.82–5. in terms of spherical coordinates.87) holds J3 = To determine J 2 we note. (5.87) To derive these properties we consider. using (5. applying repeatedly the chain rule. φ) = (x3 ∂2 − x2 ∂3 )f (θ. using ∂/∂tanφ = cos2 φ ∂/∂φ and ∂/∂cosθ = −(1/sinθ) ∂/∂θ. φ) (5.90) This yields. using (5. Using (5. According to (5. involve only the angular variables θ and φ and not the radius r. φ) sin θ ∂θ sin θ ∂θ L1 f (θ.85). (L1 )2 = = ∂ 2 ∂ + cot θ cos φ ∂θ ∂φ 2 ∂ ∂ 1 ∂2 sinφ cosφ sin2 φ 2 − + 2 cotθ sinφ cosφ ∂θ ∂φ ∂θ∂φ sin2 θ ∂2 ∂ + cot2 θ cos2 φ 2 + cotθ cos2 φ ∂θ ∂φ sin φ i ∂φ .85).86). We like to express now the operators J3 and J 2 . φ) = (5. we will prove − − − i i i J1 J2 J3 = L1 = L2 = L3 ∂ ∂ + cot θ cos φ ∂θ ∂φ ∂ ∂ = − cos φ + cot θ sin φ ∂θ ∂φ ∂ = − .86) (5. cos θ ∂ cos2 φ + sin θ cos φ ∂φ cos2 θ sin θ sin φ ∂ sin θ sin φ ∂ + + sin2 θ f (θ. φ).92) (5. actually. φ) ∂x ∂x3 ∂ tan φ =0 (5. φ) ∂x2 /x1 ∂ ∂x3 /r ∂ = x3 + x3 ∂x2 ∂ tan φ ∂x2 ∂ cos θ ∂x2 /x1 /r ∂ ∂ − x2 − x2 ∂x3 3 ∂ cos θ f (θ.91) which agrees with expression (5.5.60). ∂φ = sin φ (5. In fact.84 ) one obtains.85) (5. the latter defined in (5. L1 f (θ. φ) = ( x3 ∂2 − x2 ∂3 ) f (θ. for example.89) = ∂ x2 x2 ∂ x2 + x2 ∂ x3 − 33 − x2 1 3 2 x1 ∂ tan φ r ∂ cos θ r ∂ cos θ f (θ.88) (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 + .96) and the relationship between Lk and Jk as given in (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 . 2m r ∂r2 2m r2 2 (5.99) = = ∂2 2 ∂ 1 + + 2 2 2 ∂r r ∂r r sin θ ∂2 2 ∂ + 2 ∂r r ∂r sinθ ∂ ∂θ 2 + ∂2 ∂φ2 (5.64) can be constructed as follows.101) 1 ∂2 r r ∂r2 and comparision with (5.e. ∂θ2 ∂θ ∂φ (5. (5.97). sphere S2 . C∞ (S ).85.4 Angular Momentum Eigenstates According to the theorem on page 107 above the eigenfunctions (5. 5. i.87).94) It follows (L1 )2 + (L2 )2 + (L3 )2 = With cot2 θ + 1 = 1/sin2 θ. 5. These generators act on a subspace of C∞ ( ).85. namely on the space of complex-valued functions on the unit 3-dim. 5. 2 ∂θ ∂θ (5.86. The operators in the present case are Lk = − i Jk where the Jk are differential operators in θ and φ given in (5. One first identifies the Hilbert space H on which the generators Lk operate.95) sinθ ∂ ∂θ = ∂2 ∂ + cotθ .87).86. ∂θ ∂φ ∂2 ∂ ∂2 + cotθ + ( cot2 θ + 1 ) 2 . The functions in C∞ (S ) have real variables . one can conclude J2 = − 2 sin θ 2 sinθ ∂ ∂θ 2 + ∂2 ∂φ2 .97) Kinetic Energy Operator The kinetic energy of a classical particle can be expressed J2 p2 p2 = r + class 2m 2m 2m r2 (5. 5. 1 sin2 θ sinθ ∂ ∂θ (5.63. 5. 5.98) where Jclass = r × mv is the angular momentum and pr = mr is the radial momentum.100) (5.

109) using (5. 5. m) = 1 −m J− 1 2 Y (θ.64) can be constructed then by seeking first functions Y (θ.5.105) explicitly.103) iteratively for m = − 1. φ). .85–5. (5.111) . φ) (5. 0 ≤ φ < 2π. (5.4: Angular Momentum Eigenstates 111 θ.107) reads ∂ ∂ + i cotθ ∂θ ∂φ P (cosθ) ei φ i e±iφ ∂ ∂ ± i cotθ ∂θ ∂φ . φ) in H which satisfy L+ Y (θ. The eigenfunctions (5. φ) .108) m (θ. The norm in H is defined through |f |2 = S2 dΩ f ∗ (θ.63. by carrying out the construction according to (5. − 2. − + 1. . φ) = = 0 −i Y (θ. 5. 0 ≤ θ < π. φ) = −i Y (θ. The raising and lowering operators L± = L1 ± i L2 . − . and to be admissable for description of quantum states. . .87). the so-called spherical harmonics.105) ∆( . . is indeed a Hilbert space. φ). φ)f (θ. must be cyclic in φ with period 2π. m) Constructing Y → Y →→ ··· Y ( + m)! (2 )! ( − m)! −1 − We like to characterize the resulting eigenfunctions Y m (θ. and then determining the family {Y m (θ.104. φ) L3 Y (θ. .102) exists.106) The latter property is obviously satisfied for the chosen φ-dependence. φ) = ∆( .104) (5. For this purpose we split off a suitable normalization factor as well as introduce the assumption that the dependence on φ is described by a factor exp(imφ) Y It must hold L+ Y (θ.110) = 0 (5. m = − . φ). φ) = −i ( + m + 1)( − m)Y m (θ.102) where dΩ = sinθ dθ dφ is the volume element of S2 . (5. φ. φ) (5. (5. (5. φ) = 0 as well as L3 Y (θ. normalizing these functions. can be expressed L± = Accordingly. φ) (5.107) (5. φ) = 2 + 1 ( − m)! P 4π ( + m)! m (cosθ) eimφ . The function space endowed with this norm and including only functions for which the integral (5. . One obtains in this way Y m (θ. } applying L− Y m+1 (θ. .

114) yield |N |2 2 +1 1 2π 4π (2 )! π dθ sin2 0 +1 θ = 1.106.113) 0 is finite. One obtains. π +1 dθ sin2 0 +1 θ = −1 +1 −1 dx (1 − x2 ) +1 = = . 1975) .” by J. . (5. . The normalization factor N is chosen such that π (5. for example. 5.114. 5. i. ”Classical Electrodynamics. one for which π dθ sinθ |P (cosθ)|2 (5.117) Accordingly.106..116) implies |N |2 from which we conclude N = (−1) 1 · 3 · 5 · · · (2 − 1) (5.D.115) holds.116) The integral appearing in the last expression can be evaluated by repeated integration by parts.114) dθ sinθ |Y (cosθ)|2 = 1 0 (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)] (5.112) where we employed the definition of P (cosθ) in (5. φ) = (−1) 1 1 = 1 [1 · 3 · 5 · · · (2 − 1)]2 (5.119) where the factor (−1) has been included to agree with convention1 . (5.118) 2 +1 1 1 sin θ ei 4π (2 )! 2 ! φ . (5. 5. A proper solution of this equation. 2nd Ed. (5. using the new integration variable x = cos θ. is P (cos θ) = N sin θ as one can readily verify. We note that (5.107).120) See. Jackson (John Wiley.e.119) provide the following expression for Y Y (θ.112 or Theory of Angular Momentum and Spin ∂ − ∂θ cotθ P (cosθ) = 0 (5. New York.

the eigenfunctions Y m (θ. φ).121) This expressions shows that the factor eimφ . as specified in (5. i.110) together with (5. φ). To prove (5. .114.106) 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 . therefore. (5. Let us then assume that (5. namely Y (θ. 5. indeed.122) The latter identity can be written. 5.103. and can obtain now. 5. − defined in (5. 5.63.64).127) ∂ −m−1 (x2 − 1) . use (5.. ∂x −m−1 .126) Then holds. .103. .125) which agrees with (5. P m (x) = −1 ( + m)! 2 ! ( − m)! × (1 − x2 )− m+1 2 −(1 − x2 ) 2 1 ∂ x + (m + 1) 1 ∂x (1 − x2 ) 2 × (5.123) leads to the expression P m (x) = −m m ∂ 1 ( + m)! (1 − x2 )− 2 (x2 − 1) 2 ! ( − m)! ∂x −m (5.124) we proceed by induction. m = − 1.4: Angular Momentum Eigenstates 113 We have constructed a normalized solution of (5. according to (5. through repeated application of (5.121) states then for the functions P m (cos θ) ∂ + (m + 1) cotθ ∂θ P m+1 (cos θ) = − ( + m + 1) ( − m) P m (cos θ) .106).124).124) reads P (x) = = (2 )! (1 − x2 )− 2 (x2 − 1) 2 ! (−1) 1 · 3 · 5 · · · (2 − 1) sin θ (5.110). m (cos θ) e (5. − 2. Actually.5. For m = (5. every application of L− reduces the power of eiφ by one. 2 ! ( − m − 1)! ∂x −m−1 (5.123) P m+1 (x) ∂ x + (m + 1) 1 ∂x (1 − x2 ) 2 . P m+1 (x) = −m−1 m+1 ∂ 1 ( + m + 1)! (1 − x2 )− 2 (x2 − 1) . The reader should note that the associated Legendre polynomials.124) holds for m + 1. P m (x) = −1 × ( + m + 1) ( − m) × −(1 − x2 ) 2 1 (5. employing again the variable x = cos θ. are real.119). φ). describes the φ-dependence of Y m (θ.123). (5. we seek to determine the functions P m (cos θ) and. We want to demonstrate now that the recursion equation (5.e.124) called the associated Legendre polynomials.

P − (x) = 1 ∂2 (1 − x2 ) 2 (x2 − 1) . it holds ∂2 (x2 − 1) = (2 )! ∂x2 and.134) (x) = 1 2 ! (1 − x2 ) 2 . φ) = 0 (5.114 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 (5.124) and. Since (5. φ) L3 f (θ. 2 ! (2 )! ∂x2 1 (5. φ) .131) identifies f (θ. It holds. We want to test if the recursion (5. we embark on this construction in order to prove (5.133) Since the term with the highest power of (x2 − 1) is x2 . using x = cos θ. φ) choosing the proper sign. therefore. Constructing Y − → Y − +1 → ···Y m (θ.124). etc. which is equivalent to recursive application of L− .127) reproduces (5.130) (5. the derivative ∂ 2 +1 /∂x2 +1 of this expression vanishes. We first determine Y − (θ. therefore.129) holds. Y − +2 . Such construction reproduces the eigenfunctions Y m (θ. from such construction emerges an important symmetry property of Y m (θ. and constructs then the eigenfunctions Y − +1 .132). namely. 5.124) holds for m = . φ) using (5. that expression (5. Y ∗ (θ.124) holds for m if it holds for m + 1.e. yields a vanishing expression for m = − − 1 as long as is a non-negative integer. φ) = = determines first a normalized solution 0 i f (θ. In fact. 5. In that case (1 − x2 ) is a polynomial of degree 2 and. it holds then for all m. P − (5. hence. However. φ). φ) = (−1)m Y m −m (θ. by repeated application of the operator L+ .135) .128) one can show that (5.106. if L− Y (θ.124).123) terminates for m = − . therefore. i. φ) (5. (5. expression (5. φ) An alternative route to construct the eigenfunctions Y of L− f (θ. appears not very interesting.124) and.. therefore. φ) as given in (5. φ) = Y − (θ. (5.106.132) which reduces the number of spherical harmonics which need to be evaluated independently roughly by half .

(5.140) Introducing again the variable x = cos θ leads to the recursion equation [c. One can readily verify that this expression provides a solution of (5. (5.143) . This follows from the identity ∂ ∂ − i cotθ ∂θ ∂φ One can use (5.129). (5. (5. According to (5.123)] P m+1 (x) = −(1 − x2 ) 2 1 ∂ x − m 1 ∂x (1 − x2 ) 2 P m (x) . i. (5. (5.136) is normalized and has the proper sign.142) yields P − (x) = (−1) 1 (1 − x2 )− 2 (x2 − 1) = (1 − x2 ) 2 2 ! 2 ! (5.e.124). 5. a sign consistent with the family of functions Y m constructed above.141) We want to demonstrate now that this recursion equation leads to the expression P m (x) = +m m ∂ (−1)m (1 − x2 ) 2 (x2 − 1) 2 ! ∂x +m (5.5.f. 5. following closely the proof of eq.4: Angular Momentum Eigenstates Due to (5. the expression (5.106.142) For this purpose we proceed by induction.136) to construct all other Y recursion eiφ ∂ ∂ + i cotθ ∂θ ∂φ Y m (θ.106).122. φ). φ) sin θ e−i φ = 0.110) applies the = ( + m + 1)( − m) Y m+1 (θ.139) = ∂ − m cotθ ∂θ P m (cos θ) . φ) = 2 +1 1 1 sin θ e−i 4π (2 )! 2 ! φ 115 − .138) Employing (5.80) and (5.137) m (θ.. Obviously.120) for Y obey the postulated relationship (5. We first note that for m = − (5. one can conclude for the associated Legendre polynomials ( − m)! ( + m)! = or P m+1 (cos θ) ∂ − m cotθ ∂θ P m (cos θ) ( + m + 1)( − m) ( − m − 1)! P ( + m + 1)! m+1 (cos θ) (5. (5. φ) .124) one arrives at Y (θ.132).136) We note in passing that this expression and the expression (5.

124) and (5. are identical.150) . 2 ! ∂x +m+1 (5. 1 (1 − x ) ∂x +m (1 − x2 ) 2 1 (5.135).. (5. We then assume that (5. (5. Accordingly. 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 .149) (5.124). 2 ! ∂x +m (5. (5.141) reads then P m+1 (x) = 1 (5. P (x) = P 0 (x) are called Legendre polynomials. given by (5. one notes that application of (5. Y −2 .142) holds for m = − we verified that it holds for all m.142) holds for m.h.147) the r.s.e.142) and by (5. However.128) m + 1 → −m or. (5. Y − +2 . etc.s.124) yields (−1)m ( − m)! P ( + m)! −m (x) = +m m ∂ (−1)m (1 − x2 ) 2 (x2 − 1) . 2 ! ∂x +m −(1 − x2 ) 2 Replacing in (5.148) According to (5. 2 ! ∂x 1 (5.146) i.144) reads P m+1 (x) = +m+1 m+1 ∂ (−1)m+1 (1 − x2 ) 2 (x2 − 1) . of which agrees with the r. φ) ( − m)! P ( + m)! −m (x) . The construction beginning with Y − and continuing with Y − +1 .145) Hence. etc. Hence.142).132).116 Theory of Angular Momentum and Spin wich agrees with (5. Since (5.142) one can state P (x) = ∂ (x2 − 1) .144) ∂ x − m 1 ∂x (1 − x2 ) 2 +m m ∂ (−1)m (1 − x2 ) 2 (x2 − 1) . also the associated Legendre polynomials determined this way.e. of (5.106) this implies for Y The Legendre Polynomials The functions the identity (5. one can conclude P m (x) = (−1)m m (θ. i. According to both (5.142) holds for m + 1 if it holds for m. equivalently. yields the same eigenfunctions as the previous construction beginning with Y and stepping down the series of functions Y −1 .h..

pp. the spherical harmonics for m ≥ 0 Y m (θ. φ) = 2 + 1 ( − m)! (−1)m sinm θ × 4π ( + m)! × ∂ ∂ cos θ m (5.132). ”Classical Electrodynamics. =0 (5. t = r2 /r1 . accordingly. 1965) which is an excellent compendium on special functions employed in physics.154) |r1 − r2 | r +1 =0 1 where γ is the angle between r1 and r2 . (5..156) w(x. .J.158) see. r1 is the point where the potential is measured. and |r1 − r2 | = r1 (5.152) and. The generating function allows one to derive useful properties of the Legendre polynomials. pp.153) P (cos θ) eimφ . t) = √ Comparision with (5. The spherical harmonics for m < 0 can be obtained using (5.150) and (5. 92 To prove this property see. m ≥ 0 ∂xm (5. (5. in case of x = 1 holds w(1. 2.154) can be written 1 = w(x.g. In case q = 1 and |r2 | < |r1 | holds the identity2 ∞ 1 r2 = P (cos γ) (5.” by J. P1 (x) = x . t) = √ 1 − 2xt + t2 ∞ 1 − 2 x t + t2 . N.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) . Englewood Cliffs.5.157) = 0.151) (5x3 Comparision of (5. .D. 2nd Ed. Lebedev (Prentice-Hall. P3 (x) = 1 2 117 P2 (x) = − 3x) 1 2 2 (3x − 1) . Jackson (John Wiley. t) is called a generating function of the Legendre polynomials3 . 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. 1975). 1. Using x = cos γ.N. 2 3 1 1 = = 2 1−t 1 − 2t + t ∞ t . and r2 denotes the location of the charge.156) yields P (1) = 1 . New York.4: Angular Momentum Eigenstates The first few polynomials are P0 (x) = 1 . For example. 45 of “Special Functions and their Applications” by N. e. (5. =0 (5. for example.155) P (x) t .. . .

θ. φ) m (π − θ.f.159) Using ∂lnw/∂t = (1/w) ∂w/∂t and multiplying (5. t) holds Theory of Angular Momentum and Spin ∂lnw(x. Due to cos(π − θ) = − cos θ the replacement r → −r alters P m (x) into P m (−x).168) mY m (θ.163) (5.167) (5. holds P1 (x) ( + 1) P +1 (x) = = x P0 (x) (2 + 1) x P (x) − P −1 (x) (5.162) − (2 + 1)x P (x) + P −1 (x) ] t + =1 [( + 1) P +1 (x) The coefficients for all powers t must vanish individually.142) allows one to conclude P m (−x) = (−1) +m P m (x) (5. Accordingly. φ) (5. φ) . (5.165) since ∂ n /∂(−x)n = (−1)n ∂ n /∂xn . t) (1 − 2xt + t2 ) Employing (5. φ) m (θ.161) Collecting coefficients with equal powers t yields 0 = P1 (x) − x P0 (x) ∞ (5.159) by w(1 − 2xt + t2 ) leads to the differential equation obeyed by w(x.156) and (∂/∂t) to P (x)t = ∞ ∂w + (t − x) w = 0 . Inversion Symmetry of Y m (θ.151)] allows one to determine P (x) for = 1. using P0 (x) = 1 [c. Inspection of (5. ∂t 1 − 2xt + t2 (5. If the spherical coordinates of r are (r. The spherical harmonics are eigenfunctions of the angular momentum operators J2Y J3 Y m (θ.. Noting exp[im(π + φ)] = (−1)m exp(imφ) we determine Y Properties of Y m (θ. π − θ. π + φ). . 2. π + φ) = (−1) Y m (θ. . φ) goes over to Y m (π − θ.118 For w(x.160) this differential equation is equivalent ∞ (1 − 2xt + t2 ) =0 P (x) t + (t − x) =0 P (x) t = 0 (5.164) which. then the coordinates of − r are (r.166) We want to summarize the properties of the spherical harmonics Y derived above. . π + φ). φ). t) x−t = . under inversion Y m (θ. ∂t P (x)t −1 −1 (5. (5. 1. φ) m (θ. . φ) . φ) = = 2 ( + 1) Y m (θ. Accordingly. φ) Under inversion at the origin vectors r are replaced by − r.

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) .172) (5. for any f (θ. 5.4: Angular Momentum Eigenstates 119 2. φ) = 2 + 1 ( − m)! (−1)m sinm θ × 4π ( + m)! 2 ! × ∂ ∂ cos θ m (5. φ)f (θ.169) 3. P −1 (x) ] (5.173) m+1 (θ.177) Note that the spherical harmonics are real. where ( = 1. 6. 2.176) = 1 +1 [ (2 + 1) x P (x) − are the Legendre polynomials. φ) = δ δm m . φ) Y m (θ. φ) . The spherical harmonics form. 4π (5. a complete basis of C∞ (S2 ).5. φ) (5. . . (5. The spherical harmonics for m < 0 are given by Y −m (θ. P +1 (x) P1 (x) = x . φ) m (θ..e. except for the factor exp(imφ).) P0 (x) = 1 .178) 7. φ) C = =0 m=− π C sin θdθ 0 m Y m (θ. φ) ∈ C∞ (S2 ) holds ∞ f (θ.179) . in fact. m 4. The spherical harmonics Y 0 are Y 0 (θ. The spherical harmonics obey the recursion relationships J+ Y J− Y m (θ. φ) = = ( + m + 1)( − m) Y ( + m + 1)( − m) Y m+1 (θ.175) (5. (5. i. The spherical harmonics form an orthonormal basis of the space C∞ (S2 ) of normalizable. . φ) = (−1)m Y ∗ (θ. uniquely defined functions over the unit sphere S2 which are infinitely often differentiable π 2π sin θdθ 0 0 dφY ∗m (θ. φ) 2π 0 (5. The spherical harmonics are given by the formula Y m (θ.171) m = dφ Y ∗ (θ.170) (5.174) m ≥ 0 P (cos θ) eimφ . m (5. φ) where J± = J1 ± iJ2 . φ) .

(5.2: Derive the orthogonality property for the Legendre polynomials (5.177). =0 t + −1 dxP (x) P (x) .s. φ) .86). (5. .4. Exercise 5.h.120 8.1: Derive expressions (5.4.189). (5.183) (5. The spherical harmonics for m (θ.87). For the Laplacian holds 2 = = = (5.188) Exercise 5. t) stated in (5. Exercise 5. φ).189) evaluate the integral on the l.187) h(r) Y m (θ.186) (5. 1. The spherical harmonics obey the inversion symmetry Y 10. π + φ) .3: Construct all spherical harmonics Y3m (θ. expand the result in powers of t and equate the resulting powers to those arising on the r.184) Y22 Y21 Y20 together with (5.180) 9. The spherical harmonics for θ = 0 are Y m (θ Theory of Angular Momentum and Spin = 0. For this purpose start from the identity +1 ∞ +1 dx w(x. (5.181) = 0. φ) = δm0 2 +1 .179) using the generating function w(x. 2 are given by Y00 = 1 √ 4π 3 sin θ eiφ 8π 3 cos θ 4π 1 4 − 15 sin2 θ e2iφ 2π 15 sin θ cos θ eiφ 8π 5 4π 3 1 cos2 θ − 2 2 (5.156). φ) = (−1) Y m (π − θ.s.4. φ) = 1 ∂2 r − r ∂r2 ( + 1) r2 h(r) Y m (θ. 4π (5. of (5. t)2 = −1 .. 11.h.185) (5.182) Y11 Y10 = = − (5.

= 0.190) provide a complete. ∞. (5. φ ) (5. assumes a particularly simple form. φ) ˜ under such rotational transformation by f (θ. φ) c = . (5. m = − . }.42)..− + . = − .196) C∞ (S ) = = X . For this purpose we consider the subspaces of C∞ (S ) X = [ {Y .193) where [ { } ] denotes closure of a set by taking all possible linear combinations of the elements of the set. exp − i ϑ · J (X ) = X . φ) ˜ Y m (θ.195) i.5. .m D(ϑ) m.. (5. (5. m . φ) (5. φ) exp − ϑ · J . on functions f (θ. . 1. .. φ) i − ϑ·J f (θ.192) = S2 dΩ Y ∗ (θ . therefore. and (ii) they form the lowest dimensional sets X obeying C∞ (S ) = X which have this invariance property. i. ˜ f (θ... These coefficients can be considered the elements of an infinite-dimensional matrix which provides the representation of exp − i ϑ · J in the basis {Y m . φ) = exp Since the spherical harmonics Y space C∞ (S ) one can expand ˜ f (θ. . . . We denote the image of a function f (θ. .5: Irreducible Representations 121 5. φ) ∈ C∞ (S ). m = S2 i dΩ Y ∗m (θ...5 Irreducible Representations We will consider now the effect of the rotational transformations introduced in (5.194) Y m (θ. . φ) . ∞.. φ) D(ϑ) = . m Y m (θ.− + .197) The subspaces X have the important property that (i) they are invariant under rotations exp − i ϑ · J .193) shows ∞ = − . called the irreducible representation.m m m (θ.e. = . with elements given by (5. φ) (5. exp − i ϑ · J . The transformations exp − i ϑ · J . . .e. orthonormal basis for the function c m Y m (θ. . }] . i. ..191) i − ϑ·J f (θ .195)... if one specifies the functional form of the coefficients [D(ϑ)] m . . φ ) exp m One can also represent C∞ (S ) C∞ (S ) = [ {Y . φ). . = .e. }] (5. Comparision with (5. . are characterized completely if one specifies the transformation of any Y m (θ. − + 1. φ) m. We like to argue that the matrix representing the rotational transformations of the function space C∞ (S ). ..

.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 . .     (5. one concludes for the matrix elements (5.195) [D(ϑ)] m. one can conclude that J+ leaves X invariant.n2 . One = contribute can conclude then that in the expansion (5. . φ) ..   . in case m = − .e. .n2 .198) 2 2 where ϑ± = ϑ1 ( 1 ϑ− J+ 2 + 1 2 ϑ+ J− ϑ2 . J− 2 also leaves n1 X invariant. J− . produces 0.197) of C∞ (S ) the matrix representation of D(ϑ) assumes a block-diagonal form. equivalently.. and J3 1 1 ϑ · J = ϑ− J+ + ϑ+ J− + ϑ3 J3 (5. Application of (5..65). with blocks of dimension 1. Since the Y m are n3 eigenfunctions of J3 the action of J3 reproduces the basis functions of X . n The action of J− on Y m produces Y m−1 or. Similarly. φ) exp m i − ϑ·J Y m (θ.199) allows one to express exp i − ϑ·J = n1 . i.200) leaves X invariant. Since any additive term n n n cn1 . J± ] = ± J± (5. 3. J− ] = 2 J3 . One proceeds by expanding first ϑ · J in terms of J+ .e. the transformation exp − i ϑ · J leaves X invariant as well. i. which follow readily from (5.. 5.n3 J+ 1 J− 2 J3 3 . leaves X invariant. .n3 n n n Accordingly. m = δ S2 dΩ Y ∗ (θ.201) From expression (5.e. one needs to consider the action of J+ 1 J− 2 J3 3 on the subspaces X .200) in (5.56.202) D(ϑ) =           (2 + 1) × (2 + 1)               . (5. (5.200) of the rotational transformations one can also conclude that the X are the smallest invariant subspaces of C∞ (S ). i.199) [J+ .   1×1         3×3           . can be expanded in powers One employs then the commutation properties [J3 . The exponential operator exp − i ϑ · J + ϑ3 J3 )n . .194) only spherical harmonics with or. 5. If one orders the basis according to the partitoning (5..

(e) In the space C∞ ( ) of infinitely often differentiable.22. (a) Show that SO(2) with the group operation ◦ defined as matrix multiplication. 5. this parametrization. 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. The three components ϑk . Determine the generator of ρ(ϕ) in analogy to (5. other representations are termed reducible representations. does not provide the simplest mathematical description of rotations. for which holds RT R = R RT = 1 1.e.5. This step reconsiders the parametrization of rotations by the vector ϑ assumed sofar. Give the corresponding expression of ρ(ϕ). The representation is referred to as the irreducible representation.6: Wigner Rotation Matrices 123 This representation of D(ϑ) is the one wich has blocks of the lowest dimensions possible. f (α + 2π) = f (α). and for which detR = 1.e. is a group. (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 5.1: 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. (d) Show that the similarity transformation T (ϕ) defined in the space of non-singular. 2. (ii) a second rotation around the new x2 –axis by an angle β. However. 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).6 Wigner Rotation Matrices We will now take an important first step to determine the functional form of the matrix elements of D(ϑ).5. (b) Prove that the elements of SO(2) can be completely characterized through a single parameter. though seemingly natural. A ∈ C. (f) The transformation ρ(ϕ) as defined in (e) leaves the following subspace of functions considered in (e) Xm = { f (α) = A e−imα . i. 5. .49). i. and periodic functions f (α). k = 1. 3 certainly allow one to describe any rotation around the origin. 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.

e.. The Euler rotation replaces exp − i ϑ · J by e− i γJ3 − i βJ2 − i αJ3 e e .203) For any ϑ ∈ R one can find Euler angles α. m i Y m (θ. The axis x2 is defined in the coordinate frame which is related to the original frame by rotation (i). φ) [D(α. one can replace (5. the axis x3 is defined in the coordinate frame which is related to the original frame by the consecutive rotations (i) and (ii)..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. (5.204) is satisfied. can then be written e− i βJ2 − i αJ3 e = e− i αJ3 e− i βJ2 (5. i (5. i. φ) e− γJ3 − i βJ2 − i αJ3 e e . followed by J2 in the original frame. i. For any similarity transformation involving operators A and S holds eS A S Accordingly. γ)] m. of (5. in analogy to (5.209). β. however.e. γ ∈ R such that exp i − ϑ·J = e− i γJ3 − i βJ2 − i αJ3 e e (5.m m.207) which replaces J2 by the inverse transformation (i). β. We will demonstrate that (5..194.205) Y m (θ.124 Theory of Angular Momentum and Spin The angles α. followed by transformation (i). (i) and (ii). φ) = S2 dΩ Y ∗m (θ.210) The third rotation in (5. i. 5. in terms of rotations defined with respect to the original frame. (5.208) αJ3 βJ2 = e− αJ3 e− i βJ2 e i (5. Accordingly. and the r. Obviously.s.e.211) Using (5.209) The first two rotations in (5. the transformation from the original frame to the rotated frame. φ) (5. β. we can write e− i −1 = S eA S −1 .206) The expression (5.h. γ will be referred to as Euler angles. m [D(α.203) has the disadvantage that it employs rotations defined with respect to three different frames of reference.203). the transformation from the rotated frame to the original frame. is e− i γJ3 = e− i βJ2 e− i αJ3 e− i γJ3 e i αJ3 e i βJ2 (5.h. For this purpose we notice that J2 can be expressed through the similarity transformation J2 = e− i αJ3 J2 e i αJ3 (5.207) are equivalent. the l.203) can be expressed. γ)] = .195) by ˜ Y m (θ.s.212) .203). β.

219) = e−iαm dm m (β) e−iγm δ ( ) .7: Spin 1 2 and the group SU(2) 125 Multiplication from the left with (5. m ( ) S2 dΩ Y ∗ (θ.210) yields the simple result e− i γJ3 − i βJ2 − i αJ3 e e = e− i αJ3 e− i βJ2 e− i γJ3 . Examples are the other five members of the lepton family to which the electron belongs. (5. φ) . in addition.216) Using. the self-adjointness of the operator J3 one can state dΩ Y ∗ (θ. φ) e− m S2 i αJ3 f (θ. φ) [D(α. 5. γ)] m. The eigenvalue property (5. φ) = e−im α S2 dΩ Y ∗ (θ. β.m m.e. φ) Y m (θ. φ) i i (5.206) by ˜ Y m (θ.219).310)] an explicit expression for the Wigner rotation matrix (5. to redefine the three rotations in (5. β. φ) e− αJ3 e− βJ2 Y m (θ. one can write [D(α.215) The evaluation of the matrix elements (5. m = e−iαm S2 dΩ Y ∗ (θ. β. This allows one to express the rotational trasnformation (5. 5. γ)] = .5.7 Spin 1 2 and the group SU(2) The spin describes a basic and fascinating property of matter. γ)] m. φ) e− m i βJ2 Y m (θ.203) with respect to the original (unprimed) frame one simply needs to reverse the order of the rotations.214) e− γJ3 = S2 dΩ Y ∗m (θ. (5.218) Defining the so-called Wigner rotation matrix dmm (β) = one can express the rotation matrices [D(α. φ) (5. φ) .220) We will derive below [see Eqs. γ)] m. Best known is the spin of the electron. φ) e− m i βJ2 Y m (θ.. φ) = S2 dΩ f (θ. φ) e−iγm δ (5. but many other elementary components of matter are endowed with spin-like properties. φ) e− S2 i γJ3 Y m (θ.309.217) Accordingly. 5. (5. (5. β. φ) f (θ.64) yields dΩ f (θ.205. (5. m (5. m [D(α. φ) e−imγ . m i Y m (θ.215) benefits from the choice of Euler angles for the parametrization of rotational transformations. the muon µ and its neutrino νµ of the second .213) i. the electron e and the electron neutrino νe of the first generation.

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

x. 2 S3 = 1 σ3 2 (5. 1 i. namely. σj 2 (5. z. At this point we will state a useful property.224) Algebraic Properties of the Pauli Matrices The Pauli matrices (5.227) which is essentially identical to the Lie algebra (5.230) where a. e. hermitian 2 × 2–matrices A can be expandend. k = 1. A = x σ 1 + y σ 2 + z σ3 . σk ]+ = 2δjk 1 . 5. (5. the simplest choice being S1 = 1 σ1 . 1 σj σk = − σk σj for j = k which can also be readily verified.. σ3 = 1 0 0 −1 . 2 S2 = 1 σ2 . Any such A can be expressed in terms of three real parameters x. but their comonents must not necessarily commute with each other.7: Spin 1 2 and the group SU(2) 127 exist three such linear independent matrices. it holds. it might hold aj bk − bk aj = 0. (5. σk ]− = 2i jk z x − iy x + iy −z . i. σ2 = 0 −i i 0 . y.5. ∈ R . b are vectors commuting with σ.227. One can readily verify the commutation property σj σk − σk σj = [σj .k=1 σ σ aj bk + j>k aj bk = 3 j.226) The Pauli matrices satisfy very special commutation and anti-commutation relationships. (σ · a) (σ · b) = a · b 1 + iσ · (a × b) 1 (5. According to this property the Pauli matrices generate a 3dimensional Clifford algebra C3 .225) σ (5.229) = 1 .e. y. one obtains (σ · a) (σ · b) = 3 j 2 j=1 (σ ) aj bj 3 j k j.g. z A = In fact.231) .229) and avoids commuting the components aj and bk .k=1 σ σ 3 j k j. 3 are the Pauli matrices σ1 = 0 1 1 0 . they are associated with the important fermion property of matter.224) provide a basis in terms of which all traceless. In fact.31) of the group SO(3). 2. 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.k=1 j<k + σ j σ k aj bk . The Pauli matrices obey the following anti-commutation properties σj σk + σk σj = [σj . (5. The proof of this relation rests on the commutation relationship and the anti-commutation relationship (5. (5. Clifford algebras play an important role in the mathematical structure of physics.e.223) where σk .228) (5.

222) by the Euler form exp(−iγS3 ) exp(−iβS2 ) exp(−iαS3 ) . This relationship emerges. The commutation property (5.230).128 Theory of Angular Momentum and Spin Using (σ j )2 = 1 the first term on the r.231) proves (5. S ] = i k m Sm (5.k=1 j>k 1 2 3 j. For this purpose we choose to replace (5. (5. yields a · b. (5.236) The matrix elements of this 2 × 2 operator can be labelled by the basis states χ+ and χ− . however. 1 1 1 ( + 1) χ± 2 2 .h.e. In the special case a = b ∈ R holds (σ · a)2 = a 2 1 .k=1 j>k σ j σ k (aj bk − ak bj ) = 1 2 3 j.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. through the algebra obeyed by the operators Sk [Sk . 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 ) . hence. (5.222) as a rotation vector and you rotate once.8 Generators and Rotation Matrices of SU(2) Sofar there is nothing which relates the transformations U . Well almost like it.237) + 1)1 and. spin changes sign. (5. to rotations in space.235) which is identical to that of the generators of SO(3). We like to derive this result now by evaluating the transformations of SU(2) explicitly.s. i. The two remaining terms yield using σ j σ k = −σ k σ j for j = k and altering ‘dummy’ summation indices 3 j.k=1 j>k σ j σ k (aj bk − ak bj ) − 1 2 3 j k j. however. Noticing the idempotence of the Sk ’s 2 Sk = 3 one obtains S 2 = 4 1 = 1 1 1 2(2 1 4 1 0 0 1 (5. 1 (5. only a 720o rotation leaves the spin invariant. let say around the x2 –axis by 360o .k=1 (σ σ j>k σ k σ j (aj bk − ak bj ) = (5.222).238) S 2 χ± = .233) This result together with (5. since there is a slight difference: when you interprete the 3-component ϑ in (5.227) leads to 3 3 jk j. S3 χ± = ± 1 χ± 2 .234) 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 . therefore.5. The idempotence property of Sk yields particularly simple expressions for these powers. identical particles any number of which can exist in the same state χ+ and χ− . 1. .244) i. . following Jourdan and Schwinger. .239 ) one expands the exponential operator exp(−iβS2 ). namely. 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. In order to determine the Wigner matrix in (5. (5. . use the label 2 2 | 1 ± 1 for the states χ± .e.219).239) where we made use of the notation (5. rotation by 360o changes the sign of the spin state. 3 .220) are [D(α.. .240) . (5. ( − β S2 )2n = (−1)n ( − β S2 )2n+1 = (−1)n β 2 β 2 2n 1 1 0 1 −1 0 2n+1 (5. . (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. β. 2. that states | m for higher quantum numbers = 0. (5.242) The expressions in brackets [. .9 Constructing Spin States with Larger Quantum Numbers Through Spinor Operators In this section we like to demonstrate. .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 . The complete matrix elements (5.241) 2n+1 Taylor expansion of the exponential operator yields then ∞ exp(−iβS2 ) = n=0 (−1)n β 2 2n ∞ 1 + 1 n=0 (−1)n β 2 0 1 −1 0 (5.9: 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).245) 5.e. γ)] m. We may. i.] can be identified with the Taylor expansion of the cos.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. 2 2 We obtain with this notation for the transformations (5.219.239) in the notation (5. no two such particles can exist in the same state. m = − . 1 . − + 1. i.e.243) We note the property of this rotation matrix 2 dmm (2π) (1) = −1 1 .5. For this purpose one needs to determine the powers of −iβS2 .

2 2 . 2 + 2 − † (5. We consider. −j + 1.e. b ζ The States |Ψ(j.246) The corresponding adjoint operators are denoted by b+ and b− . . b− )T is expressed through the commutation relationships + − (ζ. + − For example. j.248) that bζ relationships † † b ζ. b− )T as spinor creation and annihilation + − operators. j = 0. 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. 3 . 1 . These operators are associated with a given spatial reference system. b† |Ψ0 = χ+ + .5.247.252) (5. b± |Ψ0 = 0 (5. m) = b† − j−m (j + m)! (j − m)! |Ψ0 .248) bζ . m) † and bζ obey the commutation (5.251) One can show using this property and (5. b ζ = 0 b ζ. (5. .243) the creation operators in a coordinate system rotated by an angle β around the y–axis are b+ † = cos β β † b + sin b† 2 + 2 − . b† )T and b = (b+ . i.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± . . therefore. i. b † ζ For the vacuum state |Ψ0 holds = δζζ . . −) bζ .247) (5. also operators of the type x b† = x+ b† + x− b† + − . x∗ b = x∗ b+ + x∗ b− + − (5. b− † = − sin β † β b + cos b† .253) = δζζ . . m = −j. b† |Ψ0 = χ− − .250) which for x∗ x = x∗ x+ + x∗ x− = 1 represent spinor operators in an arbitrary reference system. 1. b† )T and b = (b+ . b † ζ ζ =0 (5. . using (5. b ζ = b ζ. .e. The boson character of the operators b† = (b† . b ζ = b† .249) In the following will refer to b† = (b† .254) We will show below that these states are orthonormal and form spin states with higher quantum numbers. ζ = +. (5.

We will derive the result. that the annihilation operators play a role similar to the differential operator in calculus. If it is true for n then using (5. b † + = n b† + n−1 b† + b † + + n = (n + 1) b† + . For this purpose we consider the following polynomial of creation operators N j+m j−m f (b† ) = + n=0 cn b† + n .256) In the special case |state = |Ψ0 this reads using (5.259) The property is obviously true for n = 0. 5. . b† + b+ . f (b† ) |Ψ0 . + b+ .5.258) Equations (5.260) . b† + = n b† + n−1 .249).10: Algebraic Properties of Spinor Operators 131 5.257) motivate us to determine the commutator (5.248) we obtain b+ . (5.254). possibly infinite power series. b† + − well-known for the quantum mechanical harmonic oscillator. b† + n = n cn b+ .248) and from (5. b† + n . We will show that for this commutator holds n b+ . f (b† ) + (5. The Spinor Derivative Operator A most important operation is the action of the annihilation operators b+ and b− on operators which can be expressed as polynomials.256) and (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+ . we need to evaluate b+ . in b† and b† . (5. f b† Obviously. f (b† ) |state + f (b† ) b+ |state .247.249) b+ f (b† ) |Ψ0 = + b+ .257) which for (5.255) is 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. An example + − is the monomial b† which generates the states (5. (5. b† + n b† + b† + b† + + n n n b+ . b† + n+1 = = b+ . + + (5.

.262) Because of the commutation relationships (5.266) we compare the terms x+ b† + j+m φjm (x) Ψ (j.261) Similarly. = f (b† ) + (5. m) We want to prove now the property ∞ j exp xb † |Ψ0 = 1 j=0.263) According to (5..255) b+ .264) This demonstrates the equivalence of the spinor operators bζ and the derivative operation.1. 2 . f (x)] = ∂k f (x) . b† ) . f (x)] g(x) = (∂k f (x))g(x) or [∂k . m) = x− b† − j−m (j + m)! (j − m)! |Ψ0 . m) φjm (x) = j−m xj+m x− + (j + m)! (j − m)! .132 Theory of Angular Momentum and Spin i. In order to derive (5. m) m=−j (5.e. m) .247) the more general property for polynomials f (b† ..267) exp xb† |Ψ0 is called a generating function of |Ψ(j.268) .266) where xb† has been defined in (5. (5. b† ) in b† and b† holds (ζ = +. f (b† ) + where f (x) = df dx .259) holds for all n ∈ N. Equation (5.250) and where φjm (x) represents the function of the two variables x+ and x− closely related to |Ψ(j.257) we can conclude in particular bζ f (b† . the property holds also for n + 1. φjm (x) Ψ (j. b† ) |Ψ0 . b† ) |Ψ0 = + − ∂ ∂b† ζ f (b† . + − (5. f (b† ) − = f (b† ) − (5. One can finally conclude for polynomials (5. f (b† . (5. b† ) + − = ∂ ∂b† ζ f (b† .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 . By induction we can conclude that (5. + − (5. (5. −) + − + − bζ .265) Generating Function of the States |Ψ(j. one can prove b− .

1 . .e.. (2j)! 2j |Ψ0 (5..274) Summing this expression over 2j = 0.. (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)! (5.273) Change of the summation indices allows one to conclude j φjm (x) Ψ (j.. m) m=−j = = 1 x+ b† + x− b† + − (2j)! 2j 1 xb† |Ψ0 .275) = exp xb† |Ψ0 . 1 . . 2.270) s x+ b† + N −s x− b† − s |Ψ0 . chosing the summation index j = 0..269) (5. . . . 2 ∞ 1 xb† 2j! u 2j |Ψ0 = u=0.1.. one obtains φjm (x) Ψ (j. . 3 . s! (n − s)! Defining j+m = N −s j−m = s.270) N 2j j → s=0 j−m=0 → m=−j . m) = = |Ψ0 (N − s)! s! N −s 1 N! x+ b† x− b† + − N ! (N − s)!s! N −s 133 (5... (5.10: Algebraic Properties of Spinor Operators with the s–th term in the binomial expansion of (a + b)n n! an−s bs . 1. i. m) 1 j=0.5.1.. m=−j = j=0. 1.. 2 2 leads to ∞ j ∞ φjm (x) |Ψ (j. 2 .. 1 xb† u! |Ψ0 (5.1.272) The summation over s can be written in terms of j and m using (5. 2.

b† ] = 0 allow one to state that for + − + any polynomial f (b+ .280) The commutation properties [b+ .254) are orthonormal.282) where we defined x∗ y = x∗ y+ + x∗ y− .277) To evaluate this expression we first notice † e(xb ) † = e(x ∗ b) .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 . b− ) holds † † f (b+ . of (5. (5. b− ] = 0. m) |Ψ j .277) by Ψ0 |e(x b) e(yb ) Ψ0 . [b† . m) and will be used for this purpose below.h.283) (5. m) We want to show now that the states (5.284) . m holds. b− )e(yb ) |Ψ0 = f (y+ .276) = j.m φjm (x∗ ) φj m (y) Ψ (j. The generating function allows one to derive various properties of the states |Ψ (j. m) |Ψ j . we can conclude † † ∗ e(xb ) Ψ0 |e(yb ) Ψ0 = e(x y) (5. b† ] = 0 and [b− . therefore.e. (5.m. 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.j . y− )e(yb ) |Ψ0 (5. For this purpose we consider the inner product † † e(xb ) Ψ0 |e(yb ) Ψ0 = δjj δmm (5. that Ψ (j.278) † ∗ which allows us to replace the l.281) and. hence.249) for any non-vanishing integer s holds + − Ψ0 |bs Ψ0 = 0 and. In order to evaluate the ∗ b) (yb† ) operator e(x e we notice that the derivative property (5. According to (5. i. m .s. (5. Orthonormality of the States |Ψ (j. (5.264) implies † † bζ e(yb ) |Ψ0 = yζ e(yb ) |Ψ0 .

.275) one can show e(x ∗ y) = j.288) The three operators obey the Lie algebra of SU(2) [Ji . b† bm n n n.268–5. b† bm n n n.223. J1 = J2 = J3 = 1 † b b− + b† b+ − 2 + 1 † b b− − b† b+ − 2i + 1 † b b+ − b† b− . 5. 3 . (5.277) φjm (x∗ ) φj m (y) Ψ (j. k = 1.m. m) |Ψ j . (5. etc.289) < n |σi | m >< n |σj | m > {b† bm . 2 .m. − 2 + (5.n .224).m.. m j. (5.n . Jj ] = i This property is derived as follows [Ji .j .5. bm bm } n n = 1 4 < n |σi | m >< n |σj | m > {b† bm δmn − b† bm δm n } n n n. In our present notation the matrix elements < ζ |σk | ζ > for ζ = ± correspond to the matrix elements < ζ |σk | ζ > for σ = ± 1 . (5.m φjm (x∗ ) φjm (y) . 1 New Representation of Spin 0.276). < + |σ1 | − > = < − |σ1 | + > = 1.m +b† b† . 2. using < + |σ1 | + > = < − |σ1 | − > = 0.285) Following steps similar to those in Eqs.m ijk Jk .10: Algebraic Properties of Spinor Operators and comparing (5. One defines corresponding operators Jk through 1 Jk = b† < ζ |σk | ζ > bζ .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. .224).m 135 = e(x ∗ y) .m . 3 denote the Pauli spin matrices (5. m) as given in (5.282) and (5. The 2 operators are explicitly. (5.286) from which follows immediately the orthonormality property (5. States 2 We want to demonstrate now that the states |Ψ (j.287) ζ 2 ζ.ζ where σk . . Jj ] = = 1 4 1 4 < n |σi | m >< n |σj | m > b† bm .n .m. 1.

J1 ] + i [J1 .m. J2 ] = −2 J3 from which we can conclude the property 1 1 J− J+ = J+ J− − J3 2 2 Hence.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. (5.290) |Ψ (j. J+ ] = [J1 − iJ2 .295) The last result together with (5.288) 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 .291) Here we have defined J+ J− = J1 + iJ2 = J1 − iJ2 = b† b− + = b† b+ . m) as Eigenstates of J 2 and J3 We wish to show now that the states |Ψ (j.k ijk ijk σk | m > b † bm n = i = i 1 2 < n |σk | m > b† bm n n.m 1 4 < n |σj | m >< m |σi | m > b† bm n n .m. (5. σj ]| m > b† bm n n. To this end we note 2 2 2 J 2 = J1 + J2 + J3 = 1 1 2 J+ J− + J− J+ + J3 2 2 . (5. (5.292) The commutation relationships (5.m.297) .m < n |[σi .289) yield [J− .294) . 2 J 2 = J+ J− + J3 − J3 = J3 (J3 − 1) + J+ J− (5.293) .m ijk Jk .292). m) are eigenstates of J 2 and J3 . (5. J1 + iJ2 ] = −i [J2 . − (5.136 = 1 4 − = = 1 4 1 4 Theory of Angular Momentum and Spin < n |σi | m >< m |σj | m > b† bm n n.m < n |2i n. (5.

m + 1) (j + m) (j − m + 1)|Ψ (j. (5.304) (5. [Ij . Ik ] = i jk I . and eigenstates of k with eigenvalues j. . 1 2.s.300) (5. m) . b† . m) J3 |Ψ (j. m) m |Ψ (j. One can then conclude J 2 |Ψ (j. 1 2 k (a) Show that the eigenstates are given by 1 |n1 . . . State the corresponding eigenvalues and the degree of degeneracy.254) are eigenstates of + b+ and − b− with eigenvalues j + m and j − m. I2 = (b1 b2 − b+ b1 ) . Show λ = n1 +n2 n1 +n2 2 ( 2 + 1) and m = 1 2 (n1 − n2 ). . n2 .e. m) = = (j + m + 1) (j − m) |Ψ (j. n2 = √ b+ n1 ! 1 n1 |0 1 1 √ b+ n2 ! 2 n2 |0 2 where the vacuum states are defined through bj |0 j = 0. . the number of states to the possible energy eigenvalues. I3 = (b+ b1 + b+ b2 ) 2 2 2 2 2i 2 1 satisfy the Lie algebra of SU(2).305) = = j ( j + 1 ) |Ψ (j. 2. . (5. ˆ respectively.298) (5. . [bj . the states (5. . m I1 = where λ = 0. n1 = 0.5. m) J− |Ψ (j. . n = n1 + n2 fixed. (b) Show that the three operators 1 + 1 1 + (b1 b2 + b+ b1 ) . b † + b + + b † + b+ b † − b − 1 1 b† + b + + b † + b + b † − b− + b † − b − b † + b + . j = 1. m) One can furthermore derive readily J+ |Ψ (j. m = −λ. .303) Exercise 5. m − 1) .10. λ. m = λ||λ. n} eigenstates of I 2 = I1 + I2 + I3 and of I3 I 2 ||λ.299) (5. i. . i. 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† .10: Algebraic Properties of Spinor Operators The last term on the r. ..301) Obviously. b+ ] = δjk .h.e. Construct. using operators I± = I1 ± iI2 2 2 2 and the subspace {|n1 . −λ + 1. 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) . .302) (5. 2 2 137 (5. I3 ||λ. m . 2 . m = m||λ.

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 (5.306) and (5.243).306) where b † and b † are given by (5. for example.307) Comparision of (5.254) are |Ψ j.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    (5.308) The latter sum involves terms (b† )j+m (b† )j−m for σ + σ = j + m and 2j − σ − σ = j − m. In case j = 1 this expression yields.138 Theory of Angular Momentum and Spin 5. σ values which yields (b† )j+m (b† )j−m is + − then σ = j + m − σ. . The prefactor of (b† )j+m (b† )j−m which according to (5. m ) are related to the + − states |Ψ (j.310) it reduces to (5. m = b† + j+m b† − j−m (j + m )! (j − m )! |Ψ0 (5.309) . The combination of σ. both conditions are satisfied for σ = j + m − σ. However.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. (j) (5. + − One may expect that these two conditions restrict both σ and σ. On the other side the states |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).251). For this purpose we note that the states |Ψ(jm) in a rotated coordinate system according m to (5. m) . m = m=−j dm m (β) |Ψ (j.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† . m) in the original coordinate system by j |Ψ j.306.

308) unaltered except for a factor (−1)2j which multiplies then also the final result (5. This factor implies. however.s. x2 . i. b+ † = −cos β † β b+ − sin b† 2 2 − . 2 2 − (5. in case of j = 1.309). U =  0 1 1 i √ ( x1 − ix2 ) √ x3 − √2 0 2 2 (5. Replacing the latter transformation by its negative form. the orthogonal 3 × 3 matrix which describes a rotation around the x2 –axis. b− † = sin β † β b+ − cos b† .e.251).. namely that the component of the vector on the l.312) The choice of this transformation derives from a property shown further below. i. .311) (5.251) and (5. the matrix A should represent a rotation around the x2 –axis in the space of vectors ˜ (x1 .309) does not distinguish between SU(2) transformations (5. conclude that the mapping from SU(2) onto SO(3) is a 2–1 mapping.   = U  x2  .12: Mapping of SU(2) onto SO(3) 139 5.12 Mapping of SU(2) onto SO(3) The representation (5. It is of interest to trace the mapping from SU(2) onto SO(3) as represented by (5. of (5.314) leaves (5. The SU(2) transformation entered in (5.e.h. therefore.. x3 ) ∈ R . that the the representation (5. This can be shown by (1) applying to the matrix A = (dm m ) in (5. in fact.309) and the particular matrix (5.314) in case of integer j–values.312) transforms like an angular momentum state ˜ |1 m .5. e.g.310) establishes a mapping of SU(2) onto 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. Hence. One can. is the expected element of SO(3).310).308) and had the form (5.313) × 0 0 2  =  0 sinβ 0 cosβ 1 −i 0 which.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  .310) to the SU(2) transformation assumed in deriving the general result (5.

140 Theory of Angular Momentum and Spin .

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

each of which stands for two angles. 1. to separate the 14 641 dimensional space of rotational states Y 1 m1 (ˆAB ) Y 2 m2 (ˆC ) into r ρ subspaces Bk . . 14 641} . 1. 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 ).max . . n = 1. and by the moment of inertia IA−B of the diatomic molecule A–B approximately as follows 1. 2} 1.m1 .max d(B) = 1 =0 (2 1 + 1) 2 =0 (2 2 + 1) = ( 1. since further important degrees of freedom. . (6. . One may consider then to describe the motion in terms ˆ of products of spherical harmonics Y 1 m1 (ˆAB ) Y 2 m2 (ˆC ) describing rotation of the compound AB r ρ and the orbital angular momentum of C around AB. however. vibrations and rearrangement of the particles in reactions like AB + C → A + BC. = B such.max = 2.max + 1)2 (6. . .max . the dimensions d(Bk ) of these subspaces does not exceed 100.3) For rather moderate values 1. Obviously.1) where 1. .e. .max = ∆ 2 mA mB mC E . in particular. mC . In fact. 2 . There is yet another important reason why the following section is of fundamental importance for the theory of the microscopic world governed by Quantum Mechanics. .142 Addition of Angular Momenta and Spin ρC . . a very large number. .. − 1 ≤ m1 ≤ 1. in fact.m1 2 . composite systems behaving like elementary objects are common. and the following mathematical description will shed light on their ubiquitous appearence in physics. by the masses mA .4) The basis set which provides a maximum degree of decoupling between rotational states is of great principle interest since the new states behave in many respects like states with the attributes of a single angular momentum state: to an observer the three particle system prepared in such states my look like a two particle system governed by a single angular momentum state. will make their appearence a natural consequence of the symmetry of the building blocks of matter. B1 ⊕ B2 ⊕ . by the total energy E. that only states within the subspaces Bk are coupled in the scattering process. Such extremely useful transformation of the problem can be achieved through the choice of a new basis set B = { 1 . have not even be considered.max = 1 2 IA−B E . i. mA mB + mB mC + mA mC 2.max = 10 one obtains d(B) = 14 641. ρ 1 = 0. (6. 2. mB . Such large number of dynamically coupled states would constitute a serious problem in any detailed description of the scattering process.max + 1)2 ( 2. The latter often arrives at the physical properties of composite systems by adding the . rather than by Classical Mechanics. .m2 c (n) r ρ 1. the values of which are determined by the size of the interaction domain ∆V .max denote the largest orbital and rotational angular momentum values.max and 2. as we will demonstrate below. = 0.2) The dimension d(B) of B is 1.m2 Y 1 m1 (ˆAB ) Y 2 m2 (ˆC ). albeit puzzling. .max 2. − 2 ≤ m1 ≤ (6. The rotational symmetry of the interaction between the particles allows one. 2.

∂yAB ∂zAB i (2) ∂ ∂ − J3 = ρy − ρx . Rotational Symmetry of the Hamiltonian As pointed out already.48) i i R(ϑ) = exp − ϑ · J (1) exp − ϑ · J (2) (6. rotations R(ϑ) of the wave functions ψ(rAB . The invariance of the Hamiltonian under overall rotations of the three particle system implies then H = R(ϑ) H R−1 (ϑ) . It holds then H R(ϑ) ψ = R(ϑ) H ψ.5) do not affect the Hamiltonian. 5. according to (5. R−1 (ϑ) ρC ) (6. presumably a facility the reader would like to acquire with great eagerness.4) which decouples rotational states is connected with the rotational symmetry of the Hamiltonian of the three particle system considered.5) according to (5.6. Jq(2) = 0 for p. the commutation relationships (1) Jp . 2.7) where the generators J (k) are differential operators acting on rAB (k = 1) and on ρC (k = 2).10) . In this sense.e. q = 1. 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. from which follows in turn the well-known result that H is related to H through the similarity transformation H = R(ϑ) H R−1 (ϑ). the existence of a basis (6. in fact. Since this is true for any ψ(rAB . assuming presently that H might. Hence. Describing quantum mechanically a property of a composite object as a whole and relating this property to the properties of the elementary building blocks is then the quantum mechanical equivalent of the important operation of addittion..6) For the following it is essential to note that H is not invariant under rotations of only rAB or ρC . i. ρC ) = ψ(R−1 (ϑ) rAB . but solely under simultaneous and identical rotations of rAB or ρC . 3 (6. One can equivalently express therefore (6.9) hold since the components of J (k) are differential operators with respect to different variables.55) holds i (1) − J1 = zAB ∂ ∂ − yAB .7) i R(ϑ) = exp − ϑ · J (6. For ˆ ˆ example.0: Addition of Angular Momenta and Spin 143 corresponding physical properties of the elementary components.8) Obviously. ρC ) it follows H R(ϑ) = R(ϑ) H. Following our description of rotations of single particle wave functions we express (6. be different from H.53. ρC ) defined through R(ϑ) ψ(rAB . (6. To specify this property mathematically let us denote by H the Hamiltonian in the rotated frame. ∂ρx ∂ρy (6. the reader will learn in the following section how to add and subtract in the microscopic world of Quantum Physics. connected with the fact that the Hamiltonian H does not depend on the overall orientation of the three interacting particles.

17) . ˆ Ω 2 = ρC . 2.13) Hϑ · J − Since this holds for any ϑ it follows HJ − J H = 0 or.71–5.. One can show readily that the commutation relationships [Jk .e. 3 cannot have simultaneous eigenstates among each other. componentwise. k = 1. for |ϑ| H ≈ 1 − 1 i ϑ·J H 1 + 1 i ϑ·J ≈ H + i 1. the important fact that the Jk obey the Lie algebra of SO(3). 2. we will find that the states yield the basis B with the desired property of a maximal uncoupling of rotational states. C].71–5. i. i. For a proof one uses (6. We suspect. (6. 2. k = 1.14) implies that the components of the total angular momentum operator (6. J ] = i k m Jm (6. [H. and the eigenvalues of J2 and J3 are good quantum numbers. J ] = i k m Jm for n = 1. We recognize. 3 do not commute. To describe the scattering process of AB + C most concisely one seeks eigenstates YJM of J2 and J3 which can be expressed in terms of Y 1 m1 (ˆAB ) Y 2 m2 (ˆC ).9). (6.e. Jk ] = 0 .f.14) We will refer in the following to Jk . 3 .11) we can write the condition (6. In fact.15) are satisfied.16) following the procedure stated in the theorem above [c. (5. To order O(|ϑ|) one obtains i ϑ · JH .144 where J = J (1) + J (2) Addition of Angular Momenta and Spin . that energy. the (n) (n) (n) properties [Jk . B] + [A. k = 1. In the following we will use the notation r ρ Ω1 = rAB .12) each individually can have simultaneous eigenstates with the Hamiltonian. 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 (6.11) By means of (6.. (6.12) We consider this equation for infinitesimal rotations. that the components Jk .14) implies that the total angular momentum is conserved during the scattering process. 3 as the three components of the total angular momentum operator. B + C] = [A. Before we apply the procedure (5. of course.81)].e.81) we want to consider the relationship between YJM and Y 1 m1 (ˆAB ) Y 2 m2 (ˆC ). ˆ (6. r ρ Definition of Total Angular Momentum States The commutation property (6. i. (6. a supposition which can be tested through the commutation properties of these operators. The property (6. k = 1. 2.6) for rotational invariance of the Hamiltonian in the form i i H = exp − ϑ · J H exp + ϑ · J . Eqs. the operators Jk . 2 together with the property [A. however.

two further quantum numbers need to be specified for a complete characterization of the total angular momentum states.18) is an eigenfunction of J3 . J (1) .e.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) . m1 . i. . and J3 . Since YJM sofar specifies solely two quantum numbers.18) to avoid summation of vanishing terms YJM ( 1 .. Ω2 ) = m1 (J M | 1 m1 2 M − m1 ) Y 1 m1 (Ω1 ) Y 2 m2 (Ω2 ) .e. (6. 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. assume the expansion YJM ( 1 . 2 |Ω1 .1: Clebsch-Gordan Coefficients 145 6.s. mesons.. However. and J (1) · J (2) commute.18) where the states YJM ( 1 . the eigenvalue being specified by the quantum number M . We first notice that YJM in (6. i. The two missing quantum 2 2 numbers are 1 and 2 corresponding to the eigenvalues of J (1) and J (2) . 2 |Ω1 .m2 2 |Ω1 . k = 1.m2 (J M | 1 m1 2 m2 ) Y 1 m1 (Ω1 ) Y 2 m2 (Ω2 ) (6.h. hence. J (2) . J3 .20) This equation can be satisfied only if the Clebsch-Gordan coefficients satisfy (J M | 1 m1 2 m2 ) = 0 for m1 + m2 = M . restrict the sum in (6.e. The expansion coefficients (J M | 1 m1 2 m2 ) ˆ are called the Clebsch-Gordan coefficients which we seek to determine now. These coefficients. (6. of (6.19) Noting J3 = J3 + J3 and applying this to the l. or the closely related Wigner 3j-coefficients introduced further below.18) yields using the property (k) J3 Y k mk (Ωk ) = mk Y k mk (Ωk ) . (6. Ω2 ) (1) (2) = 1 m1 2 m2 ) Y 1 m1 (Ω1 ) Y 2 m2 (Ω2 ) (m1 + m2 ) (J M | . baryons.6. J3 YJM = M YJM . 2 |Ω1 . J3 . (6. Ω2 ) = m1 .1. Exercise 6.22) We will not adopt such explicit summation since it leads to cumbersum notation. and three quarks.1: Show that J2 . 2 M YJM ( 1 .21) hold. i. ρC ) are normalized. therefore. Equivalent coefficients exist for other symmetry properties of multi–component systems.21) One can. We. an important example being the symmetry groups SU(N) governing elementary particles made up of two quarks. the reader should always keep in mind that conditions equivalent to (6. play a cardinal role in the mathematical description of microscopic physical systems. J (2) .

In fact.h. However. The property dΩ1 dΩ2 Y∗ ( 1 . .25) i. . The basis B ( 1 . . . . m1 = − 1 . M = −J.s. . J} . . in fact.1.23) corresponding to the r.e. . (6. 2 ) has (2 1 + 1)(2 2 + 1) elements. 1 + 2.. − + 1.18) applied to Y∗ and to YJ JM (J M | m1 .24) yields 1 m1 2 m2 ) ∗ 1 m1 2 m2 ) (J M| = δJJ δM M .18) constitutes a change of an orthonormal basis B( 1 . and that the states can be normalized. −J + 1. respectively. In this case the range of |J|–values would be the interval [ |J (1) | − |J (2) | . . 2 |Ω1 . 1. the basis B ( 1 .28) This property follows from the fact that the basis elements are eigenstates of hermitian operators with different eigenvalues. J = | 1 − 2 |.26) We will show below in an explicit construction of the Clebsch-Gordan coefficients that. (6. . 2 ). 2 ) is also orthonormal2 and must have the same number of elements. 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. Ω2 ).25) We want to state now two summation conditions which follow from the orthonormality of the two basis sets B( 1 . . (6. This obviously corresponds quantum mechanically to a range of J–values J = | 1 − 2 |. |J (1) | + |J (2) |]. . Ω2 ) YJ M ( 1. 2 ) and B ( 1 . .m2 2 M and with (6. B2 2) should be 2 |Ω1 .. = {YJM ( 1 . . JM 2 |Ω1 . 2) = {Y 1 m1 (Ω1 ) Y 2 2 m2 (Ω2 ).24) The basis B( 1 . property implies dΩ1 dΩ2 Y 1 m1 corresponding to the l. | 1 − 2 | + 1.s. 1 + 2 . J (1) and J (2) . − 2} 1 + 1. the range of values assumed for J is correct. m2 = − 2 . (6. Ω2 ) = δJJ δM M (6. (6. −J + 1. J. . it is not immediately obvious what the J– values are. it holds 1+ 2 ( 2J + 1 ) = (2 J=| 1 − 2 | 1 + 1) (2 2 + 1) .146 Theory of Angular Momentum and Spin The expansion (6. Our derivation below will also yield real values for the Clebsch-Gordan coefficients. . . . Exercise 6. The orthonormality (Ω1 ) Y 2 m2 (Ω2 )Y 1 m1 Ω1 ) Y 2 m2 (Ω2 ) = δ 1 1 δm1 m1 δ 2 2 δm2 m2 . | 1 − 2| + 1. . . to a new basis B ( 1 .h.27) together with (6. (6. 2) . For each quantum number J there should be 2J + 1 elements YJM with M = −J.

i.1 states.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 . We can. Ω2 ). Ω2 )Y 1 m1 (Ω1 )Y 2 m2 (Ω2 ) . A similar 2 construction will also be applied to composite systems governed by other symmetry groups.18). 2 |Ω1 . (6. or J > 1 + 2 (6.2: Construction of Clebsch-Gordan Coefficients 147 The second summation condition starts from the fact that the basis sets B( 1 . One can show readily using the same reasoning as applied in the derivation of (6. (6.e. 2 |Ω1 .81) stated above.29) where the expansion coefficients are given by the respective scalar products in function space cJ M = dΩ1 dΩ2 Y∗ J M ( 1. also apply the results. 2 ) span the same function space. 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. 2 ) and B ( 1 .. . (6. i.e. to composite systems involving spin.2 Construction of Clebsch-Gordan Coefficients We will now construct the Clebsch-Gordan coefficients.6. or baryons multiplets involving three quarks.28) from (6.e. The result of this construction will include all the properties previewed above.34) 6. 2 |Ω1 ..33) JM The latter summation has not been restricted explicitly to allowed J–values. Ω2 ) . Ω2 ) . 1+ 2 J Y 1 m1 (Ω1 )Y 2 m2 (Ω2 ) = J =| 1 − 2 | M =−J cJ M YJ M ( 1.m2 M ( 1. e. therefore.30) The latter property follows from multiplying (6. 1 m1 2 m2 ) = 0 if J < | 1 − 2| .18) by Y∗ J orthogonality property (6.71–5.28) allows one to conclude that the coefficients cJ (J M | 1 m1 2 m2 )∗ ..32) which complements (6. i. (6. Hence. and actually also the properties of Clebsch-Gordan coefficients stated above. 2 |Ω1 . it is possible to expand Y 1 m1 (Ω1 )Y 2 m2 (Ω2 ) in terms of YJM ( 1 .31) are identical to Comparision with (6. 2 |Ω1 . The (J M | 1 m1 2 m2 ) cJ M . will be based solely on the commutation properties of the operators J and J (k) . rather the convention (J M | has been assumed. At this point we like to stress that the construction will be based on the theorems (5..g. the group SU(3) in case of meson multiplets involving two quarks. (6.27) yields δJJ δM M = m1 . Ω2 ) and integrating.

42) ( 1.25) using (6. . Ω2 ) = (J + M + 1)(J − M )YJM ( 1 . we can also demonstrate using (??) that G dΩ1 = We. Ω2 ) = Y 1 1 (Ω1 ) Y 2 2 (Ω2 ) . (6. We first seek an unnormalized solution GJJ and later normalize. 1+ 2 ( 1. Similarly.40) (Ω1 ) J+ Y (2) 2 2 J+ Y 1 1 (Ω1 ) Y (Ω2 ) + Y (Ω2 ) = 0 .37) The solution needs to be normalized. M = −J. . Quantum mechanically this corresponds to a state G 1+ 2. J3 YJJ ( 1 . Ω2 ). The construction assumes a particular choice of J ∈ {| such J–value seeks an expansion (6.11) J+ by J+ + J+ . . 2 |Ω1 . Ω2 ) . To find GJJ we start from the observation that GJJ represents the state with the largest possible quantum number J = 1 + 2 with the largest possible component M = 1 + 2 along the z–axis.38) for M = J − 1.35) 1 − 2 |.37) and replace (1) (2) according to (6. Ω2 ) 2 |Ω1 . Ω2 ) . . 2 |Ω1 . 2 |Ω1 .41) 2) Y (Ω1 ) Y (Ω2 ) . J − 2. 1 + 2} and for (6. 1+ 2 ( 1.11) and (5.148 For the construction of YJM we will need the operators J± = J1 + iJ2 . Ω2 ) (Ω1 ) dΩ2 Y 2 2 (Ω2 ) = 1. −J + 1.39) into (6. The corresponding classical total angular momentum vector Jclass (1) (2) would be obtained by aligning both Jclass and Jclass also along the z–axis and adding these two vectors.68) J+ = (1) + J+ (1) (2) Y 1 1 (Ω1 ) Y 2 2 2 2 (Ω2 ) 1 1 (6. For this purpose we insert (6. . . Ω2 ) = Y 1 1 (Ω1 ) Y 2 2 (Ω2 ) (6. (6.66.64) 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+ 2.18) which satisfies J+ YJJ ( 1 . 2 |Ω1 . Having determined such YJJ we then construct the whole family of functions XJ = {YJM ( 1 . . therefore.37).43) . 2 |Ω1 . We obtain using (5. . 2 |Ω1 . Ω2 ) Addition of Angular Momenta and Spin (6. −J. 2 |Ω1 . 2 |Ω1 . Ω2 ) is normalized dΩ2 G dΩ1 Y 1 1 1+ 2. . J} by applying repeatedly J− YJM +1 ( 1 . (6. We embark on the suggested construction for the choice J = 1 + 2 . 1+ 2 ( 1. . 2 |Ω1 .5. we can demonstrate condition (6.39) which we will try for a solution of (6. 1+ 2 In fact. | 1 − 2| + 1.36) = 0 = J YJJ ( 1 . have shown Y 1+ 2. (6.

39) as required by (6.38) to construct the family of functions B + = {Y + M ( . thereby.2. Ω2 ) 2 1+ 2 = (Ω1 )Y 2 2 −1 (6. 1 + 2 − 1. 2 |Ω1 . . 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.s.27).45) for some unknown constant c.37) is satisfied.h. . . The reader should familiarize himself with the entries of the Table. Ω2) for M = 1 + 2 − 2 and J = 1 + 2 .36. Normalization yields furthermore Y = 1 1 + 2 −1 1 + 2 −1 We have thereby determined ( 1 2 |Ω1 Ω2 ) (6. in particular. Having constructed this family we have determined the Clebsch-Gordan coefficients ( 1 + 2 − 1M | 1 m1 2 m2 ).44) (Ω2 ) . 6. We like to construct now the family of total angular momentum functions B + − = {Y + − M ( .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 ). of (6. . all the Clebsch-Gordan coefficients ( 1 + 2 M | 1 m1 2 m2 ). . To satisfy this equation one needs to choose c = − 1 / 2 .21) we set G 1 + 2 −1 1 + 2 −1 ( 1 2 |Ω1 Ω2 ) = Y 1 1 −1 (Ω1 )Y 2 2 (Ω2 ) + c Y 1 1 (Ω1 )Y 2 2 −1 (Ω2 ) (6. with the symmetry pattern and with the terms Y 1 m1 Y 2 m2 contributing to each YJ M . The r. G 1 + 2 −1 1 + 2 −1 in (6. . Exercise 6. . The result is illustrated for the case 1 = 2. One obtains then Y 1 1+ 2 1 + 2 1 + 2 −1 2 2 ( 1.2: Construction of Clebsch-Gordan Coefficients 149 We now employ property (6.h. 2 |Ω1 . and all total angular momentum functions for a given choice of 1 and 2 . . The l.45). 1 + 2 − 2. We demonstrate the procedure ex(1) (1) plicitly only for M = 1 + 2 − 1. | . 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. ). This expression is orthogonal to (6. ). 1 + 2 − 3. Show that the resulting functions are orthonormal.6.47) can serve now to obtain recursively the elements of the family B + − for M = 1 + 2 − 2. . M = −( + − ). ( + − )}. . To demonstrate (6.1: Construct following the procedure above the three functions YJM ( 1 .47) (Ω2 ) − 1 1 2 + Y 2 1 1 −1 (Ω1 )Y 2 2 + Y 2 1 1 (Ω1 )Y 2 2 −1 (Ω2 ) . | . of (6. We seek for this purpose first an unnormalized solution G 1 + 2 −1 1 + 2 −1 of (6.s. ( + )}.36) we proceed as above and obtain J+ Y = (1) 1 1 −1 (Ω1 ) Y 2 2 2 2 (Ω2 ) + c Y 1 1 1 1 (Ω1 ) J+ Y (2) 2 2 −1 (Ω2 ) (6. . . 2 = 1 in Table 1. Ω2 ). etc. M = −( + ). According to the condition (6.38) yields pression 2( 1 + 2 )Y 1 + 2 1 + 2 −1 ( 1 . −( 1 + 2 − 1). Expression (6. One can readily show that (6.37). .46) 2 1 +c Y (Ω1 ) Y 2 2 (Ω2 ) = 0 . 2 |Ω1 . One can obviously continue the construction outlined to determine Y 1 + 2 −2 1 + 2 −2 .

408248 −0. 1|Ω1 .707107 0.730297 −0.774597 0 2 5 8 15 1 6 3 10 1 3 0.150 Addition of Angular Momenta and Spin Y33 (2. 1|Ω1 .316228 0. Ω2) Y31 (2.57735 0. 1|Ω1 . Ω2) Y2−2 (2.632456 0.447214 0. Ω2) Y3−1 (2.57735 0.547723 0. Ω2) Y1−1 (2.632456 0.1: Some explicit analytical and numerical values of Clebsch-Gordan coefficients and their relationship to the total angular momentum wave functions and single particle angular momentum wave functions.707107 0.57735 3 10 3 5 Y2−1 (Ω1 )Y11 (Ω2 ) Y20 (Ω1 )Y10 (Ω2 ) Y21 (Ω1 )Y1−1 (Ω2 ) −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. Ω2) Y11 (2. Ω2) Y30 (2.774597 − − − − 0. Ω2) Y10 (2. Ω2) Y3−2 (2.258199 −0.816497 0. 1|Ω1 . 1|Ω1 .774597 0. . 1|Ω1 . Ω2) Y22 (2. 1|Ω1 .816497 Y22 (Ω1 )Y1−1 (Ω2 ) 1 15 1 3 3 5 1 5 1 2 3 10 2 5 1 2 1 10 2 3 1 3 Y20 (Ω1 )Y11 (Ω2 ) Y21 (Ω1 )Y10 (Ω2 ) 0. Ω2) Y20 (2. Ω2) Y21 (2. 1|Ω1 . 1|Ω1 .730297 0. Ω2) Y32 (2. 1|Ω1 .316228 0.408248 −0. 1|Ω1 .707107 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. Ω2) Y3−3 (2.632456 −0.447214 −0.258199 0. 1|Ω1 .547723 0.707107 0.547723 0. 1|Ω1 . Ω2) Y2−1 (2.57735 −0.57735 0.547723 0. 1|Ω1 .57735 0.816497 −0.

commute with each other ζ ζ aζ .9. For this purpsose we will employ the spinor operators introduced in Sections 5. 6.2: Use the construction for Clebsch-Gordan coefficients above to prove the following formulas  J+M 1  for = J − 2 2J 1 1 1 = J. 5.52) and | 2 m2 2 According to Section 5. aζ bζ . aζ . we extent definition (5. m + 1 .3: Explicit Expressions of CG–Coefficients 151 Exercise 6.287) to two particles (1) J k k = = (2) J 1 2 1 2 ζ.10. J. bζ = a† . M | . These expressions will be derived now.50) (6. bζ = = a† .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 .6.10 [cf. hence. we are dealing now with two particles carrying angular momentum or spin.48) (6. a† ζ ζ b† . b† ζ ζ = 0 = 0 .ζ a† < ζ |σk | ζ > aζ ζ b† < ζ |σk | ζ > bζ ζ (6.49) where ζ. bζ . b† refer to different particles and.53) (6.51) The operators aζ . Definition of Spinor Operators for Two Particles In contrast to Sections 5. + 2  − J−M +1 for = J + 1 2J+2 2  J−M  for = J − 1 2J 2 = . (6.10.10 where we studied single particle angular momentum and spin.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 ). 5. M | .54) 2 + = √ b† ( 2 +m2 )! √− ( 2 −m2 )! . It is actually possible to state explicit expressions for any single coefficient (JM | 1 m1 2 m2 ). (5. b† ζ = 0. 1 (6. 2 . Accordingly. b † ζ = δζζ = δζζ . (6. The creation and annihilation operators are again of the boson type with commutation properties aζ . b† ζ ζ = aζ .2. 1 . a† and bζ . m − 2 . a† 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.9. ζ = ± and the matrix elements < ζ |σk | ζ > are as defined in Section 5.ζ ζ.

152 It holds.59) J± = J1 ± i J2 . we like to recall for future reference that the operators (5. (j) 2 ˆ ˆ J = kj ( kj + 1 ) . The operator of the total angular momentum/spin of the two particle system is J = with Cartesian components Jk = (1) J + J (6. can be expressed in terms of the number operators 1 † 1 ˆ ˆ a† a+ + a† a− .54). M ( 1 . . 2) .55) . M ( 1 .57) b† + ( (2) 2 2 +m2 1 +m1 b† − ( 2 2 −m2 + m1 )! − m1 )! (1) + m2 )! − m2 )! |Ψo . (6. ˆ kj |J.303). J3 . |J.66) (6. 2 .18) as follows = | 1 m1 . 3 . 2 .60) (6. m2 2 . are | 1 . k = 1. 2 )). hence. (1) J 2 |J. 2 ) which are simultaneous eigenstates of the operators 2 2 J2 . = M |J. 1.. (1) J 2 | (2) J 2 | 1 m1 1 2 m2 2 Addition of Angular Momentum and Spin = = 1( 1 2( 2 + 1) | 1 m1 + 1) | 2 m2 1 2 .e. (1) J .53. m1 1 | 2 . 1 . 2 )) . −J ≤ M ≤ J . mj j = j | j . m2 |J. 2 . 6. M ( 2) 2) = J(J + 1) |J. M ( 1 . as usual are denoted by their respective quantum numbers J.302. M ( 1 . M ( 1 . 2) + 1) |J. J3 |J. .m2 | 1 . m1 . (6. (6. M ( 1 . (2) J 2 . mj j and.62) . + 1) |J.68) The states |J. M ( 1 . (1) J (2) J 3 | 1 m1 1 3 | 2 m2 2 = m1 | 1 m1 = m2 | 2 m2 1 2 (6. m2 |J. m2 a† + ( 1 2 = a† − ( 1 1 −m1 (6. i. M ( 1 . M ( 1 .61) (6.63) according to At this point.65) (6. m1 . M. 2) 2) can be expressed in terms of Clebsch-Gordan coefficients (6. (5. 2) 1. (1) J 2 . M ( 1 .58) Jk + (2) Jk .69) − 2| ≤ J ≤ + 2 . M ( 1 . m1 1 | 2 . 2) 2) (6. 5. 2. m2 1 2 ( 1 .56) The states | 1 . which describe a two particle system according to (6. j = 1.59) We seek to determine states |J. M ( 1 . The aim of the present Section is to determine closed expressions for the Clebsch–Gordan coefficents ( 1 . in analogy to Eqs. = = 1( 1 2( 2 2) 2) (6. (6. (6. ˆ For the operators kj holds ˆ kj | j . 2 . (2) J which. for such states should hold J2 |J. k2 = b b+ + b† b− k1 = + − − 2 2 + namely. M ( (2) J 2 |J. m1 1 | 2 .64) (6.300). M ( 1 .67) = j |J. 2) We will also require below the raising and lowering operators associated with the total angular momentum operator (6.

75) The relationships (6.51) one obtains ˆ k1 . K † (kj + 1) 2 2 1ˆ † 1 ˆ kj K + K † (kj + 1) 2 2 1 ˆ 1 ˆ K † kj + kj .71) (6. K † 1 † a a+ + a† a− . a† b† − a† a− .73. 6. Employing (6. K † J± . 2 4 = 0 = = 0. K † = 0. 2 2 3 ˆ = K † kj + K † . 6.71–6.74) imply 3 2 J2 . + − − + 153 (6. j = 1. 6. K † 2 1ˆ ˆ 1 ˆ ˆ kj kj + 1.71–6. This operator obeys the following commutation relationships with the other pertinent angular momentum / spin operators ˆ kj .71) one can show K† = = = = = 1 ˆ ˆ kj (kj + 1). expressing (k) J 3 through the creation and annihilation operators according to (5. a† b† + + − − + 2 2 − 1 1 = a† b† − a† b† = K† . K † (j) J 2 . K † + K † 2 2 3 † †ˆ K kj + K .73) (6.72) (6.64. + − + − 2 2 − + = .74) + 1 J− J+ + J2 . We note that. using (6. j = 1. b† − b† a† b− .3: Explicit Expressions of CG–Coefficients The Operator K † The following operator K † = a† b† − a† b† . (6. ˆ A similar calculation yields [k2 . a† b† + − − + 2 + 2 − 1 1 − b† a† b+ . For example. (6. + − − + 2 2 = 1 † 2K . K † + kj .65) and (6.6.288). K † 1 a† a+ − a† a− + b† b+ − b† b− . K † ] = (j) J 2 . a† b† − a† b† − + − − + 2 + 1 1 = a† a+ . and applying the relationships (6.70) will play a crucial role in the evaluation of the Clebsch-Gordan-Coefficients. a† b† + a† a− . a† b† − a† b† + − + − + − − + 2 1 1 = a† a+ .50. the relationships (6. 4 Using J3 = (1) J 3 + (2) J 3 .73) yields J3 . K† J3 .73) can be readily proven. due to J2 = K† 1 2 J+ J− 1 † K . b† = 0 .

2 (6. 2 ) .62. K † = Addition of Angular Momentum and Spin a† a− + b† b− . |j1 + j2 . M ( 1 . M ( 1 .76) Here N is an unknown normalization constant. We consider now the case that (K † )n acts on the simplest total angular momentum / spin state. Action of K † on the states |J.77).2. (6. One can generalize property (6. j1 + j2 (j1 . according to (6. M ( 1. K † ] = 0 is demonstrated in an analoguous way. j1 + j2 (j1 + n n . 2) is a state with quantum numbers 1 1 + 1 2 and = N |J. a† b− − a† b† + + + − + = 0. j1 + ) K † 2 2 (6. M ( 1 . 2) We want to demonstrate now that the action of K † on the states |J. j2 + ) 2 2 n n = N (n. 1 2 + 2 . = − a† a− . M ( 1 + . it holds K † |J.154 Starting from (5. 6. 2) = N |J. 4 3 + j ( j + 1) |J. M ( 1 . i. a† b† + b† a† b− . this result implies that K † |J. on the state |j1 + j2 . j2 2 ..79) n |j1 + j2 . b† + − + + + − The property [J− .78) a state which has been used already in the construction of Clebsch-Gordan coefficients in Section 6. M ( 1 . M ( 1 . j + 2 2 2) However. M ( 1 .75) ascertain that under the action of K † the states |J. Application of (K † )n to this state yields. j1 + j2 (j1 . 6. 6. 2 ) j + 4 1 3 + K † |J.77) where N is another normalization constant.292) one can derive similarly J+ . M ( 1 . j1 1 |j2 . The commutation properties (6. 2 ) produces again total angular momentum eigenstates to the same J and M quantum numbers of J2 and J3 . 2 ) remain eigenstates of J2 and J3 with the same quantum numbers. j2 ) . To demonstrate that the resulting states are eigenstates of (1) J 2 and (2) J 2 we exploit (6.63. M ( 1 .72) and (6. namely. j1 1 + . M ( 1 . K† + K † (j) J 2 |J. 2) 2) ˆ K † kj + = K† = j 3 † K + K † j ( j + 1) |J. 2) = = (j) 2 J . j2 ) = |j1 .e. but for different 1 and 2 quantum numbers of the operators (1) J 2 and (2) J 2 . 2 2 + n ) 2 (6.67) (j) J 2 K † |J.73. M ( 1 + n . 2 2 1 + ) .76) and state K† n |J.

K† 1 2 1 + 2 −J × × . J( 1 . 2 ) exploits the expression (6. For this purpose one needs to choose in (6. 2) = N( 1 + 2 − J. 2) 2) . 2 2 It is now important to notice that any state of the type |J. M ( 1 . Strategy for Generating the States |J. 2) . M ( 1 .74)] yields |J.83) The derivation of this expression will be provided further below (see page 158 ff).82.82) for |J. M ( 1 .84) (6. − 1. 5. (J + 2 − − 2) 1 1) 2 The normalization constant appearing here is actually N( + − J. The latter states. M ( 1 . j2 + n ).84) with (6.86) b† + J+ 2 − 1 |Ψo .79). M ) (J− )J−M |J. (6. 1 = j1 + n 2 . j2 as follows J = j1 + j2 which is equivalent to n j1 j2 Accodingly. J( 1 . M ) . J( 1 .57) and exploiting the fact that J− and K † commute [c. 2) = ∆(J.81) = N( 1 + 2 1 2 − J. 2 ) ∆(J. 1. (6.82) 1 × | (J + 2 1 × | (J + 2 1 (J + 2 1 − 1 ) .6. holds |J. 2 = j2 + n 2 (6. 2 ) for −J ≤ M ≤ J as follows |J.3: Explicit Expressions of CG–Coefficients 155 where we denoted the associated normalization constant by N (n. M ( 1 .105) of the spherical harmonics.85) ∆(J. of (6. 2 ) can be expressed through the r.104. j1 . 2) Our construction of the states |J. 2) 1 2 = ( 1 + 2 (2J + 1)! − J)! ( 1 + 2 + J + 1)! 1 2 .79) n. j1 + n . (6. Combining (6. 1 .80) = = = 1 1 2 2 − J 1 n − = (J + 2 2 n 1 − = (J + 2 2 + 1 1 − − 2) 2) .f.h. = (J + M )! (2J)! (J − M )! 1 2 2) (6. J( 1 .s. allow one to obtain the states |J. (6. in analogy to the construction (5. M ) (J + × K† 1 − 2 )! (J + 2 − 1 )! J+ 1 − 1 + 2 −J (J− )J−M a† + × 2 (6. 2 ) . 6.

m1 1 | 2 .e. x. (6. in terms of | 1 . a† ] = [b+ . (6. I(λ.69) yields then the Clebsch- Expansion of an Intermediate State We first consider the expansion of the following factor appearing in (6.86) in terms of monomials 1 +m1 1 −m1 2 +m2 2 −m2 a† a† b† b† |Ψo . m2 Gordan-coefficients. y) = λr xs y t r |Gst . (6.89) i.87) + − + − i. + + Taylor expansion of the two exponential operators yields immediately I(λ. x..57)].87). Comparision with (6.91) which. 2 [cf.50–6. st The desired expansion of |Gr can be obtained from an alternate evaluation of I(λ. One obtains then using J− = a† a+ + b† b+ (6.88).93) = f (a† + λ a† ) g(b† + λ b† ) |Ψo . 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.t (6.264). follows from the commutation properties (6.156 Addition of Angular Momentum and Spin Our strategy for the evaluation of the Clebsch-Gordan-coefficients is to expand (6.92) − − and noting [a+ .90) (6. in analogy to (5. + − + − .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. y) is a generating function for the states |Gr defined in (6.88) in terms of monomials (6. y) = exp (λJ− ) exp x a† exp x b† |Ψo .. r!s!t! r.52).s. x. b† ] = 0 [cf. For this purpose we introduce the generating function I(λ.e.86) |Gr = Jr a† st − + s b† + t |Ψo (6. x.

|J. m2 2 . using the commutation property (6. one can write the right factor in (6.95) Comparision with (6. Expanding the exponentials in st (6. y) = s. y) = exp a† + λ a† + − exp b† + λ b† + − |Ψo .86) (J− )J−M a† + = q J+ 1 − 2 b† + J+ 2 − 1 |Ψo 1 (6.97). m1 1 | 2 .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.98) a† − s a† + b† + s . x.97) + q)! |Ψo × (J − M )!(J + 1 − 2 )!(J + 2 − 1 )! q!(J + 1 − 2 − q)!(J − M − q)!(M + 2 − J+ 1 − 2 −q × Final Result a† + a† − q b† + M + 2 − 1 +q b† − J−M −q Our last step is to apply the operator (K † ) 1 + 2 −J to expression (6. to obtain the desired expansion of |J.97) yields. 157 (6.50).s.3: Explicit Expressions of CG–Coefficients We conclude I(λ.t × λ v b† − × r−q v × = t s−u v λu a† − r! q |Ψo λr xs y t r! s! t! r. 2) = s. M ( 1 .70) holds K† 1 + 2 −J = s 1 + 2 −J b† − s 1 + 2 −J−s (−1)s 1 + 2 −J−s (6.6.90) allows one to infer |Gr s. using the definition (6.99) . x. 2 ) in terms of states | 1 .94) yields I(λ. Operation of this operator on (6.88).q (−1)s (6. With K † given by (6.t × a† + s−q s q b† + t r−q t−r+q a† − q b† − |Ψo (6. M ( 1 .96) and.t = q r! × a† + s q s−q t r−q a† − q × b† + t−r+q b† − r−q |Ψo (6.94) One can infer from this result the desired expressions for |Gr .

101) Note that m1 + m2 = M holds.83) of the normalization constant N ( 1 + defined through (6. − q − s)!(q + s)! 2 × (M + 1 − 1 + q + s)!( 1 1 + 2 − M − q − s)! 2 − q − s 1 | 2. × 2) = N( 1 + 2 1 − J. j2 . J( 1 . M ) 1 = + J)!(− 1 + 2 + J + 1)! 2 2 + 2 − J)!( 1 − ( 1+ + J)! 1 2 × [( × s + m1 )!( − m1 )!( 2 + m2 )!( 2 − m2 )!(J + M )!(J − M )!] 2 (−1)s s!( 1 − m1 − s)!( 2 + m2 − s)! 1 2 1 × The Normalization 1 ( 1 + 2 − J − s)!(J − 1 − m2 + s)!(J − + m1 + s)! (6.54) between creation operator monomials and angular momentum states allow one to write this |J.69) if one identifies m1 = 1 −q − s . Using (6. To determine N = N ( 1 + 2 −J.6. (6.104) . (6. The Clebsch-Gordan coefficents are then finally ( 1 . according to (6. m2 |J.103) this can be written 1 = N 2 ψ(j1 .82).69) since. j2 . − 1 .101).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 ) we consider the scalar product J. m2 in (6.82) and (6.53. J( 1 . M − +q+s One can conclude that this expression reproduces (6.q (−1)s × (6.102) We want to determine now the expression (6. 2 ) − 2) . 1 . √ 2J + 1 1 2 . M ( 1 . n) 2 )|J.158 Addition of Angular Momentum and Spin ( 1 + 2 − J)!(J − M )!(J + 1 − 2 )!(J + 2 − 1 )! s!( 1 + 2 − J − s)!q!(J + 1 − 2 − q)!(J − M − q)!(M + 2 − a† + 2 1 −q−s 1 + q)! a† − q+s b† + M + 2 − 1 +q+s b† − 1 + 2 −M −q−s |Ψo The relationships (6. n)|ψ(j1 . The summation over q corresponds then to the summation over m1 . 2 ) ∆(J.103) = 1. m1 . q = 1 − m1 − s and m2 = M − m1 . 2 ) (6. m2 = M − 1 + q + s. n = 1 + 2 −J. M ) (J + 2 )! (J + 2 − 1 )! s. For this purpose we introduce j1 = 1 (J + 2 1 2 − J. j2 = 1 (J + 2 2 − 1) .

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. We employ the expression (6. n) = 1 (2j1 )!(2j2 )! a† + n! (2j1 )!(2j2 )! a† + 2j1 −n−s s 2j1 s n s 2j2 a† b† + − |Ψo = n−s (−1)s a† b† − + s b† + (−1)s a† − s (2j1 + n − s)!s!(2j2 + s)!(n − s)! s!(n − s)! 2j2 +s b† + b† − n−s (2j1 + n − s)!s!(2j2 + s)!(n − s)! |Ψo .70) for the operator K † . 1 1−λ n1 +1 1 1−λ n2 +1 = m1 .111) Comparision with (6.57) for these states and the expression (6.106) The orthonormality of the states occurring in the last expression allows one to write (6. Accordingly.6. m2 2 .109) and Taylor expansion of the left hand side of (6.110) which can be written 1 1−λ n1 +n2 +2 = r s n1 + r − s n1 n2 + s n2 λr (6.105) The first step of our calculation is the expansion of ψ(j1 .3: Explicit Expressions of CG–Coefficients where |ψ(j1 .108) twice. we obtain |ψ(j1 .108) yields the identity n1 + r − s n1 n2 + s n2 = n1 + n2 + r + 1 n1 + n2 + 1 . (6.108). n) = K† n 159 |j1 . n) in terms of states |j1 . j1 1 |j2 . One obtains then.m2 n1 + m1 n1 n2 + m2 n2 λm1 +m2 (6. applying (6. m1 1 |j2 . j2 .108) = λ=0 (n1 + ν)! n1 ! (6. (6.104) 1 = N2 (n!)2 (2j1 )!(2j2 )! (2j1 + n − s)!(2j2 + s)! s!(n − s)! 2j2 + s 2j2 (6. j2 . j2 2 . j2 .112) s . (6.

m1 ) by ( 2 .120) The symmetry properties (6. 1 .118). and (6. m1 |J. J. m1 ). (2j1 + 2j2 + 1)! (6.h. M ) . For the quantum mechanical addition of angular momenta the Clebsch Gordan coefficients ( 1 .114) For example.113) Using the identities (6. of (6.115) This relationship is a trivial consequence of (6. (6. (6.116) through formula (6. m2 |J. −m1 . m2 ) by ( 1 .103) one obtains the desired result (6. m2 .h. (6. m2 |J.83). 2 .160 Applying this to (6. ( 2 . m1 . 2 . ( 1 .119) The corresponding symmetry property of the Clebsch-Gordan coefficients is ( 1 . one can interchange also the operator J on the l. 6. To derive relationship (6. 2 . vice versa. m1 . and . M ) = (−1) 2 +m2 2J + 1 ( 2 .115).114) one obtains − J = − J (1) − J (2) The respective Clebsch-Gordan coefficients ( 1 . of this equation J (1) = J (2) − J .. e. −m1 2|J.118) Finally. M | 1 .4 Symmetries of the Clebsch-Gordan Coefficients The Clebsch-Gordan coefficients obey symmetry properties which reflect geometrical aspects of the operator relationship (6. m2 |J.h.s. (6. m1 .120) can be readily derived from the expression (6. m1 ) . We will demonstrate this now. M ) corresponding to (6.117) are again related in a simple 2 . J (1) on the r. m1 . −m2 .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 . M ) corresponding to (6. −m1 . and J (2) are vectors in R . of (6. m2 ) and.g. 2 . 2 . 2 . −M ) . manner to the coefficients ( 1 .102) by replacing ( 1 . M ) . M ) ( 1 . (6. −m2 |J. m1 .s. m2 |J.114) by. namely.114) as long as J.s.11) J = J (1) + J (2) . (6. (6.116) one expresses the Clebsch-Gordan coefficient on the r. J (1) . interchanging the operators J (1) and J (2) results in J = J (2) + J (1) .116). 1 .114) show a simple relationship to the Clebsch Gordan coefficients ( 2 .102) of the Clebsch-Gordan coefficients. −M ) = (−1) 1 + 2 −J ( 1 . M ) = (−1) 1 + 2 −J ( 2 . m1 |J. m2 . m2 |J.116) If one takes the negatives of the operators in (6. 2 1+1 (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

1 |θ.187) 1 2 . 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. for θ = 0 (6. Since f (r) is independent of θ and φ follows ˆ ((σ · p f (r))) = −i (( ∂r f (r) )) σ · r . 1 |ˆ) r 2 (6. given by ˆ σ · r = σ1 sin θ cos φ + σ2 sin θ sin φ + σ3 cos θ . (6.f. The result can also be stated in the compact form σ · r Yjm (j ± 1 .168 Addition of Angular Momentum and Spin 1 Yjm (j ± 2 . have proven (6. 1 |θ = 0. ˆ (6.180) where the sign could depend on j. One can conclude then ˆ σ·r = ± ˆ 0 1 1 0 (6.147.182) σ · r Yj 1 (j − 1 . For this purpose we consider the application of σ · r in the case of θ = 0. identified the sign of (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 . 1 |ˆ). We want to determine ˆ its action on the wave functions f (r) Yjm (j ± 1 .186) Here (( · · · )) denotes again confinement of the diffusion operator ∂r to within the double bracket. We want to demonstrate finally that the “−”-sign holds in (6. 6. 2 |ˆ) and Yj 1 (j + 1 . φ) . ˆ (6.1 states χ 1 ± 1 2 2 2 [c. 1 |ˆ) + f (r) σ · p Yjm (j ± 1 . (5.180). 1 |θ = 0.180) and (6.183) in case θ = 0 becomes in the standard representation with respect to the spin. (6.184) 0 1 space χ 1±1 2 2 one can conclude from (6. φ) 2 2 2 = − j+ 1 2 4π (6. ˆ 2 2 2 2 2 2 for θ = 0.181) 0 = j+ 1 2 4π Yj 1 (j + 1 .224)] 1 0 σ·r = ˆ . φ) 2 2 2 . therefore. 1 |θ = 0.174–5. hence. 1 |ˆ) = ((σ · p f (r)))Yjm (j ± 1 .181.177) one notes that for φ = 0 the spin-angular momentum states 2 are entirely real such that σ · r must be real as well.160). 1 |ˆ) r r r 2 2 2 2 2 2 (6. According to (5. φ) and (5. φ) = − Yj 1 (j + 1 .188) . 6.148) the particular ˆ 1 1 states Yj 1 (j − 2 .180) and. 1 |θ = 0.185) We have.182) 0 Since σ · r. 1 |ˆ) at θ = 0 are r r 2 2 2 2 Yj 1 (j − 1 .

ˆ For the test function h(r) holds ˆ ˆ p·rh = = Using (1/r) = −r/r3 and r · ˆ −i −i · r h r = −i h r ·r − i r· h.189) The celebrated property of the Pauli matrices (5. 1 |ˆ) r 2 (6. e. Corresponding results are obtained for the other components of p × r and.192) ˆ ˆ The operator p × r in (6. 1 |ˆ) = i [ ∂r + r + ] f (r) Yjm (j 2 r r Using (6. 1 |ˆ) r 2 (6. 1 |ˆ) = 0 and σ = 2S/ .6.190) can be related to the angular momentum operator.191) (6. r 2 2 1 J ·S 2 1 ˆ σ · p f (r) Yjm (j ± 2 . x3 x2 ˆ ˆ ( p × r )1 h = −i (∂2 − ∂3 ) h (6. using ∂r Yjm (j ± 1 ..53).g.195) . 1 |ˆ) r 2 = i ∂r + −j r 1 g(r) Yjm (j + 2 .193) r r 1 1 1 = −i (∂2 x3 − ∂3 x2 ) h − i h (x3 ∂2 − x2 ∂3 ) . 1 |ˆ) . 1 |ˆ) r 2 (6.190) 3 h − i hr · r 1 − i r· ˆ r h = ∂r h one can conclude ˆ ˆ p · r h = −i 2 + ∂r r h (6.196) = i j+ ∂r + r 1 2 3 2 1 f (r) Yjm (j − 2 . we conclude the intuitively expected identity 1 ˆ ˆ p×r = − J .155.187) one obtains ˆ σ · p f (r) Yjm (j ± 1 .5: Spin–Orbital Angular Momentum States Using (6. To demonstrate this we consider one of its cartision components. r r r Using ∂2 (1/r) = −x2 /r3 .186) for both terms 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. h r (6. r 2 (6.194) ˆ ˆ where J1 is defined in (5. 6. 1 |ˆ) r 2 1 2 (6.230) allows one to express ˆ ˆ ˆ ˆ ˆ σ · p σ · r = p · r + i σ · (p × r) . ∂3 (1/r) = −x3 /r3 and ∂2 x3 = x3 ∂2 .156) this yields finally 1 ˆ σ · p f (r) Yjm (j + 2 . 1 |ˆ) = r 2 2 ˆ i ∂r f (r) − f (r) σ · p σ · r Yjm (j ˆ 1 2 169 . r Altogether we obtain.197) 1 ˆ σ · p g(r) Yjm (j − 2 . hence.198) .

203) (6. 1 |ˆ) = Yj− 1 j− 1 (ˆ) χ 1 1 . j − 1 .144) for m = j.206) . 1 |ˆ) r 2 2 r r2 and. in fact. 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 . adopting the method in Sec. We begin with the coefficients (6.143–6. − |j.197.202) Yjj (j − 1 . For m = j − 1 one can state. 1 |ˆ) r 2 2 i 3 ˆ = σ · p [ i∂r + (j + ) ] f (r) Yjm (j − 1 .e. 1 .230.200) = (j + 1 )(j + 3 ). 1 |ˆ) = r 2 2 which agrees with (6. (6. 2 2 (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 .143. are consistent with this identity. 6. . j 2 + 2j + 2 3 4 (6. 1 |j. consider first the case of the largest m-value m = j. In this case holds.101).45).143. 1 |ˆ) r 2 2 The Clebsch-Gordan coefficients are then (j − 1 .199) We want to show that eqs. 1 |ˆ) = J 2 Yjm (j + 1 . 6.2. 6. r ∂r2 r 2 (6. 6. according to (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. Yjj−1 (j − 1 .99) holds ˆ (σ · p)2 = − 2 2 = ∂2 J2 r + 2 .151).205) which agrees with the expressions (6. 1 |ˆ) r 2 2 r r2 r2 j 2 + 2j + 3 2 2 2 4 = [ −∂r − ∂r + ] f (r) Yjm (j + 1 . We note ˆ (σ · p)2 f (r) Yjm (j + 1 . r (6. according to (6.43).2.146). r 2 2 2 2 2 2 − r 2 ∂r r + J2 r2 f (r) Yjm (j + 1 . j + .198). as well as (6.144) and. 5. 1 |ˆ) r r 2 2 2 2 2 2 2 yields ˆ (σ · p)2 f (r) Yjm (j + 1 . For this purpose we use the construction method introduced in Sec. j) 1 2 1 2 1 2 1 2 = = 1 0 (6. Evaluation of Relevant Clebsch-Gordan Coefficients We want to determine now the Clebsch-Gordan coefficients (6.201) 1 3 (j + )(j + ) Yjm (j + 1 .199).170 Addition of Angular Momentum and Spin To demonstrate the validity of this key result we note that according to (5. using (5. i.204) (6. j) 2 2 2 2 (j − .

212) 1 + 2j(j − m + 1) 1 2j(j − m + 1) Yj− 1 m− 1 χ 1 − 1 2 2 2 2 j+m−1 Yj− 1 m− 3 χ 1 1 + 2 2 2 2 2j j−m+1 Yj− 1 m− 1 χ 1 − 1 . 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 .208) (6. 1 .2. (6.143.202). Let us verify then the expression (6. m − 3 . j − 1 . 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 . are obtained by iterating the application of (6. − 1 |j.143. [c. 6.210). 2 2 2 2 2j (6.206).s. 6. 2 2 2 2 2j .f. m − 1) 2 2 2 2 = j +m−1 2j (6. 1 . j − 1) 2 2 2 2 which again agrees with the expressions (6.144) for m = j − 1. 1 .144) for the Clebsch-Gordan coefficients by induction.h.143. 6.137)] J− (tot) = J− + S− (6. is obtained by applying the operator [c.210)] yields (j + m)(j − m + 1) Yjm−1 (j − 1 .. (6. (6.211) to the l. and J− + S− to the r. 6.207) (6. etc.6. Expression (6. 1 |j. as described in Sec. 2 |j. j − 1) 2 2 2 171 = = 2j − 1 2j 1 2j (6.209) (j − 1 . (· · · |jj − 3).h.s. 1 |ˆ) = r 2 2 j+m 2j + (j + m − 1)(j − m + 1) Yj− 1 m− 3 χ 1 1 2 2 2 2 j+m Yj− 1 m− 1 χ 1 − 1 2 2 2 2 2j j−m 2j (j + m)(j − m) Yj− 1 m− 1 χ 1 − 1 2 2 2 2 + or Yjm−1 (j − 1 .f.144) implies for j = m Yjm (j − 1 .5: Spin–Orbital Angular Momentum States The corresponding Clebsch-Gordan coefficients are then 1 (j − 1 . j − 3 .210) to (6. The further Clebsch-Gordan coefficients (· · · |jj − 2).

leading quantum mechanically to a symmetry of the Clebsch-Gordan coefficients (JM | 1 m1 2 m2 ) with respect to exchange of 1 m1 and 2 m2 . one can obtain expression (6. j+m+1 . However. 2 . one obtains (j + 1 .144) by induction.116. to reflect in the notation the symmetry of these quantities. (1) (2) (−tot) The coefficients which are the quantum mechanical equivalent to Jclass + Jclass + Jclass = 0 are the 3j–coefficients introduced by Wigner. M. m1 . the relationship (6. j2 m2 ) 2 .144) for j = m − 1. m1 ) (6. m3 ) = (−1) × 2 + 3 − 1 + 2 +m2 × 2 . − 1 |j. proven (6. m1 . m − 1) 2 2 2 2 = (6.216) 6. (1) (2) (−tot) Obviously. all three vectors Jclass . 1+1 3 For 1 (j. −m2 | 1 . j3 . In this context Jclass and Jclass play the same role. Jclass and Jclass play equivalent roles. − 1 |j. m + 1 ) = 2 2 2 j+m+1 . m.144). − 1 |j. hence. 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 = 0.145) from (6.143.215) which follows from (6. 1 . m + 1 . 6.214) 2 2 +1 ( 3 .217) 2j + 2 (j + 1 . m) 2 2 2 2 2j + 1 (6.146) can be obtained from (6.214) yields (j. 1 . m) = 2 2 2 2 Similarly. m3 .215).143. 6. m + 1 . m + 1 ) = 2 2 2 2 and. 1 .120). m2 1 (6. using (6. 2 . 1 |j + 1 . m3 . m1 . 1 .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 .213) which is in agreement with (6. The latter relationships applied together read ( 1 . 1 .218) where we have replaced the quantum numbers J. 2j + 1 (6. m2 . The Clebsch-Gordan coefficients (6.143) by applying the symmetry relationships (6. m. 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 . m − 1 . by the set j1 . j2 . 2j + 2 (6. m2 | 3 . We have.172 Addition of Angular Momentum and Spin j −m+1 2j (j − 1 .143). 6. 1 |j + 1 .

218) relating 3jcoefficients and Clebsch-Gordan coefficients.6: The 3j–Coefficients We first like to point out that conditions (6. √ safe for the prefactor 2j3 + 1 which is cancelled according to the definition (6. e. discovered by Regge. j2 . The two integer entries J − 1 − m2 and J − 2 + m1 in (6. =  (6. For this purpose one can use the symmetry transformations to place the smallest element into the upper left corner of the Regge symbol. anti-cyclic exchange and of a sign reversal are stated. for which no known classical analogue exists.102) of the Clebsch-Gordan coefficients. states that j1 . are identical to the integer arguments which enter the analytical expression (6. the Regge-symbol. with respect to alltogether 12 symmetry operations. These symmetry operations relate altogether 72 3j–coefficients.g. 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. m3 . −j1 + j2 + j3 . i.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 . one can exchange columns and one can reflect at the diagonal (transposition).21. To represent the full symmetry one expresses the 3j–coefficients through a 3 × 3–matrix. as follows   −j1 + j2 + j3 j1 − j2 + j3 j1 + j2 − j3 j1 j2 j3 j1 − m1 j2 − m2 j3 − m3  .. 173 (6.6. the so-called triangle condition.223) (Σ + 1)!n11 !n22 !n33 !n23 !n32 !)) n=0 n31 n32 n33 .222) m1 m2 m3 j1 + m1 j2 + m2 j3 + m3 The Regge symbol vanishes. The reader may note that the entries of the Regge-symbol. The Regge symbol also vanishes in case that two rows or columns are identical and Σ is an odd integer. j3 form the sides of a triangle and the condition is symmetric in the three quantum numbers. except when all elements are non-negative integers and each row and column has the same integer value Σ = j1 + j2 + j3 . m3 and with respect to sign reversals of all three values m1 .220) The latter condition |j1 − j2 | ≤ j3 ≤ j1 + j2 . m1 . According to the definition of the 3j–coefficients one would expect symmetry properties with respect to exchange of j1 .219) (6.102) are obtained each through the difference of two entries of the Regge-symbol. 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. there exist even further symmetry properties. m2 and j3 .e. In this way the values of 12 3j-coefficients are related. In case of a non-cyclic exchange of rows and columns the 3j–coefficent assumes a prefactor (−1)Σ . j2 . The Regge symbol reflects a remarkable degree of symmetry of the related 3j–coefficients: One can exchange rows. m2 . However.221) j1 j2 j3 −m1 −m2 −m3 where the the results of a cyclic. Because of its high degree of symmetry the Regge symbol is very suited for numerical evaluations of the 3j–coefficents.

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 (6.230) 1 1 (1m| m1 m2 ) = 2 2 √ (−1)2m1 −2m2 3 δ(m.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. m1 + m2 ) (2 − m1 )! (2 + m1 )! √ 10 (1 − m)! (1 + m)! (1 − m2 )! (1 + m2 )! 1 ≥ |m| ∧ 1 ≥ |m2 | ∧ 2 ≥ |m1 | (6.10752786393409395427444450130056540562826159542886 We also illustrate the numerical values of a sequence of 3j-coefficients in Figure 6.224) (6. m2 ) √ 2 1 ≥ |m1 |) 2 (6.225) sn−1 . 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 (6.1. m) δ(−m1 .229) 1 1 (0m| m1 m2 ) = 2 2 i(−1)1−m2 δ(0. m1 + m2 ) (3 − m)! (3 + m)! (3m|2m1 1m2 ) = √ 105 (2 − m1 )! (2 + m1 )! (1 − m2 )! (1 + m2 )! 1 ≥ |m2 |) ∧ 2 ≥ |m1 | ∧ 3 ≥ |m| (6.228) √ (−1)2m1 −2m2 7 δ(m. (6.223-6. m1 + m2 ) (2 − m1 )! (2 + m1 )! √ 6 (2 − m)! (2 + m)! (1 − m2 )! (1 + m2 )! 1 ≥ |m2 | ∧ 2 ≥ |m| ∧ 2 ≥ |m1 | (6.226) by means of a symbolic manipulation package (Mathematica) (1m|2m1 1m2 ) = (−1)1+m+m1 δ(m.231) Here is an explicit value of a Clebsch-Gordan coefficient: 1 1 (70 2 −15 1 |120 −10 60 2 −5 1 ) = 2 2 √ √ √ 4793185293503147294940209340 127 142 35834261990081573635135027068718971996984731222241334046198355 0. (6.232) .

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

. 2 = 1.. . J = | 1 − 2 |.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 |Ω1 . Exercise 6. Exercise 6.3: Two triplet states |1m1 (1) |1m2 (2) are coupled to an overall singlet state Y00 (1. .234) Exercise 6.233. each of the blocks in (6.1: Prove Eqs.176 Addition of Angular Momentum and Spin 1 For the basis {YJM ( 1 . J} (2 (2 1 2 + 1) (2 + 1) × (2 1 2 + 1) + 1) = (2| 1 − 2 | + 1) ×(2| 1 − 2 | + 1) (2| 1 − 2 | + 3) ×(2| 1 − 2 | + 3) (6. 1 . . . . .234) . .233) is further block-diagonalized as follows + 2. 2. . Ω2 ).6. . (6. . (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 . . 1).6. .6. M = −J.

−k + 1. .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 . e.7: Tensor Operators 177 6.237) The operators T ∈ {Tkq . S 2 = S1 + S2 + S3 or any other polynomial of Sk .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} with the transformation property (6. . for example. Note that in the examples mentioned the operator T would be defined within the same representation as R(ϑ).g. that T belongs to a family of operators {Tkq . Let T |ψ denote the state obtained after the operator T has been applied and let R(ϑ) denote a rotation in the representation of SO(3) or SU(2) which describes rotational transformations of the quantum states under consideration. . . ∂x2 3 1 2 2 2 2 case (ii) the operator T could be a spin operator Sk defined in (5.e. 6. a mul∂2 ∂2 tiplicative operator T ψ(r) = f (r) ψ(r) or a differential operator T ψ(r) = ( ∂x2 + ∂x2 + ∂2 ) ψ(r). −k + 1. i. that in case (i) T is an operator C∞ ( ) → C∞ ( ) acting on single particle wave functions. This implies.238) . (6.222) in case (ii) of single particle spin operators. q = −k. The operator T may also act on multi-particle states like Y 1 m1 (ˆ1 )Y 2 m2 (ˆ2 ). . The latter implies that T transforms like an angular momentum or spin state | m . In fact. e.g.236. some of the r r examples considered below involve tensor operators T of this type.236) In this equation Dq q (ϑ) denotes the rotation matrix Dq q (ϑ) = kq |R(ϑ)|kq . the operator (5.223). (k) (6. q = −k. Rotations transform |ψ as |ψ = R(ϑ)|ψ and T |ψ as R(ϑ)T |ψ . Examples of Tensor Operators The multiplicative operators C∞ ( ) → C∞ ( ) Y ( ) = def Y ( ) (6. .6.42) in case (i) of the position representation of single particle wave functions or the operator (5. (k) (6. e. k} such that k Tkq = q =−k (k) Dq q (ϑ) Tkq .237) are called tensor operators of rank k. The latter can be written T |ψ where T denotes T in the rotated frame given by T = R(ϑ) T R−1 (ϑ) .g. Such operators T can be characterized through their behaviour under rotational transformations.

7. but rather any rotation and any parametrization can be assumed.7. x3 .235 . but are far easier to ascertain for any particular operator. 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 = (6.237) do not require that the rotation R−1 (ϑ) is expressed in terms of Euler angles according to (5.2: 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. .239) These operators can be expressed in terms of the coordinates x1 . 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 . we will assume presently that the rotation is chosen as follows R(ϑ) = exp ( ϑ+ L+ + ϑ− L− + ϑ3 L3 ) (6. Exercise 6. In fact. Exercise 6. 2.3.203).237).6. For the following it is important to note that the rotation matrix elements (6.1: 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. x2 . 2. This choice of parametrization allows us to derive conditions which are equivalent to the property (6. 1. 3 follows from the transformation properties of the spherical harmonics derived in Section 1.178 are tensor operators of rank k.

. (0.246. 0)T in case of small ϑ+ .243) From (5.237) yields first k R(ϑ) Tkq R −1 (ϑ) = q =−k kq |R(ϑ)|kq Tkq .7: Tensor Operators 179 The conditions can be derived if we consider the property (6. Similar equations can be derived for infinitesimal rotations of the form ϑ = (0. The property (6.235 .6: Form tensor operators of rank 1 in terms of the three components of acting on the space of 1-particle wave functions. ϑ3 )T .245) (6.247).235 .246) (6. 6.237) for transformations characterized by infinitesimal values of ϑ+ .4: Is the 1-particle Hamiltonian 2 H = − 2 2m + V (|r|) (6.242) from which follows by means of kq |1 1|kq equation ϑ+ [L+ . We first consider a rotation with ϑ = (ϑ+ . Expressing the results in terms of the angular momentum operators J+ . 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) .247) These properties often can be readily demonstrated for operators and the transformation properties (6.7. J3 yields [J+ . Tkq ] = = δqq and by subtracting Tkq on both sides of the k ϑ+ kq |L+ |kq Tkq .237) be assumed then.1 particles for which the first spin is described by 2 the operator S (1) and the second spin by the operator S (2) . Tkq ] [J3 .6. 1 (6. Tkq ] = −iTk q+1 (k + q + 1)(k − q) .7.6.235 . (6.244) [L+ . q =−k q+1 (6. ϑ3 . (6.248) a tensor operator? Exercise 6. Tkq ] = = = Tkq+1 Tkq−1 q Tkq . 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) . Exercise 6. J− . 0. (k + q + 1)(k − q) (k + q)(k − q + 1) (6. 0.7. Tkq ] [J− . ϑ− .5: Consider a system of two spin. ϑ− .6. 2 Exercise 6.241) Using R(ϑ) = 1 + ϑ+ L+ + O(ϑ2 ) this equation can be rewritten neglecting terms of order O(ϑ2 ) 1 + + k (1 + ϑ+ L+ ) Tkq (1 − ϑ+ L+ ) = 1 1 q =−k kq |1 + ϑ+ L+ |kq Tkq .6.80) follows kq |L+ |kq = −i (k + q + 1)(k − q)δq and. (6.7. Exercise 6. hence.3: Derive Eqs. 0)T .

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.8. .21) of Clebsch-Gordan coefficients J3 Φ 1 m1 (k. m1 .8 Wigner-Eckart Theorem A second important property of tensor operators Tkq beside (6. 1 m1 |Φ 1 m1 (k. One can.m2 ( 1 m1 |kq 2 m2 ) (J3 Tkq − Tkq J3 + Tkq J3 ) | 2 m2 γ2 = qTkq = q. 2 |γ2 ) which correspond to total angular momentum states. J2 . 2 |γ2 ) is an eigenstate of J 2 with eigenvalue + 1). The hermitian property of J3 and J 2 implies that states Φ 1 m1 (k. 2 |γ2 ) = q. 2 |γ2 ) is an eigenstate of J 2 with eigenvalue 2 1 ( 1 + 1). Before we proceed we like to point out that the states | 2 m2 γ2 are also eigenstates of J 2 .m2 ( 1 m1 |kq m1 q.251) The corresponding property for Φ 1 m1 (k. behave like |kq | 1 m1 . (6. construct states Φ 1 m1 (k. | 1 m1 γ1 denotes an angular momentum (spin) state. i. 2 |γ2 ) can be shown readily as follows using (6.18) are defined through Φ 1 m1 (k. i. J3 | 2 m2 γ2 = m2 | 2 m2 γ2 .6. 2 |γ2 ) = Cδ 1 1 δm1 m1 2 1( 1 (6.e. These states according to (6. J3 are the generators of the rotation R(ϑ). J 2 | 2 m2 γ2 = 2 2( 2 + 1) | 2 m2 γ2 .237) is that their matrix elements 1 m1 γ1 |Tkq | 2 m2 γ2 obey simple relationships expressed in terms of Clebsch-Gordan coefficients. possibly the total angular momentum– spin state of a compositie system.1: Show that Φ 1 m1 (k. 2 |γ2 ) = ( 1 m1 |kq 2 m2 ) Tkq | 2 m2 γ2 . (6. To prove this we consider the transformation of Tkq | 2 m2 γ2 R(ϑ) Tkq | 2 m2 γ2 = R(ϑ) Tkq R−1 (ϑ) R(ϑ) | 2 m2 γ2 = q m2 Dq q Tkq Dm2 m2 | 2 m2 γ2 2 (k) ( ) (6. Starting point of the derivation is the fact that the states Tkq | 2 m2 γ2 behave like angular momentum states of a composite system of two particles each carrying angular momentum or spin. J3 .249) which demonstrates. 2 |γ2 ) are orthogonal to | 1 m1 in case of different quantum numbers 1 .253) Exercise 6. J3 | 2 m2 γ2 = m2 γ2 | 2 m2 and the property (6.235 .252) ∼δm1 q+m2 = ( 1 m1 |kq 2 m2 ) Tkq Similarly. in fact. which is characterized also by a set of other quantum numbers γ1 which are not affected by the rotational transformation R(ϑ).e. hence.e.247).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 .250) q.m2 2 m2 ) (q + m2 ) Tkq | 2 m2 γ2 | 2 m2 γ2 (6. i. one can show that Φ 1 m1 (k.180 Theory of Angular Momentum and Spin 6. the stated property.

(6. say m1 + 1.256) and noting that the operator adjoint to J+ is J− .6.2: Determine the matrix elements of the gradient operator d3 rF (r) G(r) of the type (6.h.32) Tkq | 2 m2 γ2 = and orthogonality property (6.259) The socalled reduced matrix element 1 . γ1 ||Tk || 2 .8. γ2 . for which the l. γ1 ||Tk || 2 .258) We can then finally express the matrix elements of the tensor operators Tkq as follows 1 m1 . m1 = q = m2 = 0. One can then evaluate also the corresponding Clebsch-Gordan coefficient ( 1 m1 |kq 2 m2 ) and determine 1 . e. 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 . (6.254) = 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) ( 1 m1 |kq 2 m2 ) . q . can be evaluated as easily as possible. γ2 k− 2 + 1 √ 1 2 m2 ) (−1) 2 1 +1 = 1 . γ1 ||Tk || 2 . its m1 –dependence being expressed solely through a Clebsch-Gordan coefficient.260) Exercise 6. γ2 = 2 1 +1 1 m1 .8: Wigner-Eckart Theorem 181 In order to evaluate the matrix elements (6. γ1 |Tkq | 2 m2 . γ2 ( 1 m1 |kq (6. To prove this property we consider 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) for a different m1 value. γ2 (−1)k− 2 + 1 ( 1 m1 |kq 2 m2 ) (6. Using | 1 m1 + 1γ1 = 1 ( 1 + m1 + 1)( 1 − m1 ) J+ | 1 m1 γ1 (6. γ1 |Tkq | 2 m2 . 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 ) = (6.253) 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.s. the matrix elements 1 m1 γ1 |Tkq | 2 m2 γ2 can be reduced to an m1 –independent factor. i.e.255) At this point the important property can be proven that 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) is independent of m1 .259) to a combination of magnetic quantum numbers m1 . γ2 is determined by applying (6.257) which establishes the m1 –independence of 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) . m2 .g.261) .

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

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

3.m (r) = = = E ψE.87) for Jk as well as the commutation property (5. 2 . J3 ψE. For this purpose one concludes first that the conservation of angular momentum J implies a motion of the particle confined to a plane. k = 1.8) This property can also be proven readily employing the expression of the kinetic energy operator [c. θ for the distance from the origin and for the angular position. 2 2mr2 (7. Employing in this plane the coordinates r.m (r) . J ψE. (7.e. one can state ˙ J = m r2 θ .184 Spherically Symmetric Potentials 7.11) (7. the time variation of J can be written.m (r) .13) .10) (7.6) ˙ Since Jk is also equal to the poisson bracket {H.7) ˆ The correspondence principle dictates then for the quantum mechanical Hamiltonian operator H and angular momentum operators Jk ˆ [H.14) (7. stationary states ψE.85– 5. one can conclude ˙ Jk = {H. Jk } = 0 . ˆ H ψE. .m (r) (7. Jk ] = 0 . (7. Accordingly.1 Radial Schr¨dinger Equation o ˙ mv = − er ∂r V (r) .m (r) .15) (7.f. using (7. k = 1. In fact. 3 .. (5.99)] 2 1 2 J2 2 2 = − ∂r r + . ˆ dt (7. i.m (r) m ψE. ( + 1) ψE.m (r) can be chosen as simultaneous eigenstates of the Hamiltonian ˆ operator H as well as of J 2 and J3 . (7. 2.61) of J 2 and Jk . 2 . d ˙ J = v × mv + r × mv = 0 + r × er (−∂r V (r)) = 0 . Jk } where H is the Hamiltonian. 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 .9) − 2m 2m r 2mr2 the expressions (5. 2.12) . 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.5) and r = v.5) In the case of an angular independent potential angular momentum J = m r × v is a constant of ˙ motion. . ˆ A classical particle moving in a potential V (r) is governed by the Newtonian equation of motion (7.

22) which is the original potential 2 ( +1) V (r) with an added rotational barrier potential 2mr2 . φ) (7.m (r) 1 2 J2 ∂r r + + V (r) − E 2m r 2mr2 the functional form ψE. φ) are the angular momentum eigenstates defined in Section 5.4. using (7. equations (7.21) Veff (r) φE (r) = = V (r) + r vE.m (r) = 0.9) one can write the stationary Schr¨dinger o equation (7. by integration of ˙ θ = J . . . One can write (7.19) where Y m (θ.20) Since this equation is independent of the quantum number m we drop the index m on the radial wave function vE. i.m (r) 2 . r(t) is determined the angular motion follows from (7.23) This demonstrates that the function r vE. for the wave function of a quantum mechanical particle a differential equation which governs solely the r-dependence. (7.m (r) = 0. (7. This barrier. It can be solved by integration of 1 2 2 J2 r = ± ˙ E − V (r) − . can exclude particles from the space with small r values. in the present case. . together with the original potential. . − Adopting for ψE.22) (7. 7.11). ∞[ governed by the effective potential (7. .11.m (r) Y m (θ.m (r) and E .m (r) .18) = vE.10). (7.20) by −2mr/ 2 yields the so-called radial Schr¨dinger equation o 2 ∂r − ( + 1) − U (r) − κ2 r2 r vκ. Employing the kinetic energy operator in the form (7.7. .m vE.m . . but can also trap particles in the latter space giving rise to strong scattering resonances (see Section ??). ( + 1) 2mr2 ..10) − 1 2 ∂ r + 2m r r 2 2 ( + 1) + V (r) − E 2mr2 .m (r) describes the radial motion as a one-dimensional motion in the interval [0. 2 (7.20) in the form of the one-dimensional Schr¨dinger equation o 2 − where 2m 2 ∂r + Veff (r) − E φE (r) = 0 .17) In analogy to the classical description one can derivem.13).e. mr2 (t) (7.16) m 2mr2 Once. (r) = 0 . (7.12) are obeyed and one obtains for (7.m ψE. Multiplying (7.1: Radial Schr¨dinger Equation o 185 This is a differential equation which governs solely the radial coordinate.24) . (7.

the term Br−( +1) would contribute |B|2 0 dr r−2 +1) (7. assumes the general form r vκ. for > 0 is not integrable. II by A.28) (7. 1975). Pascual (Springer. This follows for > 0 from consideration of the integral which measures the total particle density. (7. Boundary Conditions In order to solve (7.33) A detailed discussion of the proper boundary conditions. . (r) (7. ”Classical Electrodynamics. Galindo and P. The radial part of this integral is ∞ 0 2 dr r2 vκ. r → 0 (7.29) Only the first term is admissable. For = 0 the contribution of Br−( function is. 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). for example.30) and.26) 2m 2 In case E < 0. In this case the solution is governed my ( + 1) 2 r vκ.m (r) ∼ √ B . We replaced in (7. ψE. (r) = 0 .186 where we defined U (r) κ2 = = − − 2m 2 Spherically Symmetric Potentials V (r) E (7.D.20) one needs to specify proper boundary conditions2 . 4π|r| (7. using the expression (5.25) (7. Jackson (John Wiley.31) to the complete wave which. at r = 0 is found in the excellent monographs Quantum Mechanics I. κ assumes real values.182) for Y00 . 2nd Ed. (r) ∼ A r or vκ.20) the index E by the equivalent index κ.27) ∂r − r2 and. hence. Berlin. New York.” by J. accordingly. in particular. (r) ∼ A r + B r− −1 +1 + B r− (7. according to a well-known result in Classical Electromagnetism3 . For r → 0 one may assume that the term ( + 1)/r2 becomes larger than the potential U (r).32) The total kinetic energy resulting from this contribution is. . 2 2 2m 2 ψE.m (r) ∼ √ 4π B δ(x1 )δ(x2 )δ(x3 ) . see. .

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).

hence. (7.150) one obtains +1 +1 1 ∂ dx eizx P (x) = dx eizx (x2 − 1) . Since the spherical harmonics Y 0 (θ.s.s.100) Defining x = cos θ.179) yield from (7.e.99) We want to determine the expansion coefficients b . of (7.50). according to (5.44.98) The l.45) and one given by (7. One can show d −1 2 (x − 1) ∼ (x2 − 1) × polynomial in x dx −1 (7. For example.98) does not involve any non-vanishing m-values since Y m (θ. This holds for −1 and one can conclude +1 consecutive integrations by part d izx e dx (7. The orthogonality properties (5. One can expand the former solution in terms of solutions (7.194 Generating Function of Spherical Bessel Functions Spherically Symmetric Potentials The stationary Schr¨dinger equation of free particles (7. φ). i. In this case the expansion on the r.102) −1 d −1 +1 −1 dx −1 (x2 − 1) . and using the Rodrigues formula for Legendre polynomials (5. the wave function does not depend on φ. can be written exp(ikr cos θ). −1 . 7. (7.104) 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. one given o by (7.h. 2 +1 (7. the surface term ∼ [· · ·]+1 vanishes.98) by ∞ eik3 x3 = =0 b j (kr) P (cos θ) .43) has two solutions.50). (7.103) and.106). namely.m a mj (kr) Y m (θ. in case of a free particle moving along the x3 -axis one expands eik3 x3 = . φ) .99) +1 d cos θ eikr cos θ P (cos θ) = b j (kr) −1 2 . z = kr.h.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..178) are given in terms of Legendre polynomials P (cos θ) one can replace the expansion in (7.

106) yields finally b = i (2 + 1) or.109) results in the integral representation of j (z) j (z) = (z) 2 +1 ! +1 dx (1 − x2 ) eizx . The identity (7.s. ∞ (7. −1 (7.7.s.105) and (7. Integral Representation of Bessel Functions Combining (7.107) 1 2 (2 )! 1 · 3 · 5 · · · (2 + 1) 2 + 1 1 · 2 · 3 · 4 · · · (2 − 1) · 2 (7.f.113) dt (1 − t2 )ν− 2 eizt .h. (7. in particular.114) .105) This expression allows one to determine the expansion coefficients b . −1 (7.109) eikr cos θ = =0 i (2 + 1) j (kr) P (cos θ) .s.h. −1 (7. (7.96) one can express this. 1 (7. Comparision with the l. after insertion into (7. as the generating function of the spherical Bessel functions.69)] b z 2 (iz) = 1 · 3 · 5 · · · (2 + 1) 2 + 1 2 ! +1 dx (1 − x2 ) .105) must hold for all powers of z. 1 (7.2: Free Particle Comparision with (7.117) one can write the r. for the leading power x [c.110) One refers to the l. of (7. using ν = Jν (z) = √ 1 πΓ(ν + 1 ) 2 z 2 ν + 1.108) where the last factor is equal to unity.112) We want to consider the expression Gν (z) = aν z ν fν (z) where we define fν (z) = C (7.111) Employing (7.106) Employing (5.h.100) gives b j (kr) 2 (iz) = 2 +1 2 ! +1 195 dx (1 − x2 ) eizx . 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 .99).

120) z 2ν + 1 dt (1 − t2 )ν− 2 eizt + C C 1 1 t2 t1 = 0 (7. 1 (7. Exercise 7.93).117) dt (1 − t2 )ν− 2 t2 eizt 1 (7. 1 j = 1. (7.122) (7. We can now demonstrate that Gν (z) defined in (7.116) In case that the endpoints of the integration path C satisfy (1 − t2 )ν+ 2 eizt one can write (7. We consider now the functions u(j) (z) = √ 1 πΓ(ν + 1 ) 2 z 2 ν Cj dx (1 − x2 )ν− 2 eizx .120) imply also the property (2ν + 1) fν (z) + z fν+1 (z) = 0 . t2 of the integration paths C.121). t2 . In fact.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 . To prove this we note fν (z) = i C dt (1 − t2 )ν− 2 t eizt .116) fν (z) = − or fν (z) = − From this we can conclude fν (z) + 2ν + 1 fν (z) + fν (z) = 0 .117).196 Spherically Symmetric Potentials Here C is an integration path in the complex plane with endpoints t1 . 4.113) obeys the Bessel equation (7.123) Insertion of these identities into Bessel’s equation leads to a differential equation for fν (z) which is identical to (7.93) for arbitrary ν. 7. obeys Bessel’s equation (7.2.93) as long as the integration path in (7. C 1 (7. (7.120) such that we can conclude that Gν (z) for proper integration paths is a solution of (7.118) z 2ν + 1 fν (z) + fν (z) .119) We note that equations (7. 3. z (7.4: Prove (7. for properly chosen endpoints t1 . Gν (z).114.124) .113) satisfies (7. 2. it holds for the derivatives of Gν (z) Gν (z) G (z) = = ν aν z ν fν (z) + aν z ν fν (z) z 2ν ν(ν − 1) aν z ν fν (z) + aν z ν fν (z) + aν z ν fν (z) 2 z z (7.121) (7.

135) are known as the Poisson integrals of the Bessel functions. for Rez > 0 and ν ∈ R obey (7.130) We note that the endpoints of the integration paths C2 and C4 .131) According to (7. using t = 1 + is.124) vanishes along the whole path C3 and. . u(2) (z) and u(2) (z) are both solutions of Bessel’s equation (7.128) C = C1 ∪ C2 ∪ C3 ∪ C4 is a closed path. (7. Comparision with (7. hence.127) 0 ≤ s < ∞ (7.129) The integrand in (7. (2) Hν (z) = −2 u(4) (z) .135) (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. Since the integrand in (7. we introduce the so-called Hankel functions (1) Hν (z) = −2 u(2) (z) .134) and (7.132) For Hν (z) one derives.93).112) shows u(1) (z) = Jν (z).126) −1 ≤ s ≤ 1 (7.7. (7. Following convention. Accordingly.117) and. therefore. 1 (7.134) Similarly.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 (7.133) = e i(z−πν/2−π/4) √ 2 zν π Γ(ν + 1 ) 2 2 zν π Γ(ν + 1 ) 2 √ √ ∞ 0 ds [s(1 + is/2)]ν− 2 e−zs .130) holds Jν (z) = (1) 1 2 1 (1) (2) Hν (z) + Hν (z) . 1 (7. and (1 − t2 )ν− 2 eizt = [(1 − t)(1 + t)]ν− 2 eizt = [−is(2 + is)]ν− 2 eiz e−zs = ei(z−πν/2+π/4) 2ν− 2 [s(1 + is/2)]ν− 2 e−zs . (7. one can derive (2) Hν (z) = e −i(z−πν/2−π/4) ∞ 0 √ ds [s(1 − is/2)]ν− 2 e−zs . u(3) (z) ≡ 0. dt = i ds.125) 0 ≤ s < ∞ (7. one can state Jν (z) = − u(2) (z) + u(4) (z) . the integral expression (1) Hν (z) 1 1 1 1 (7.

142) and h(z) = H (2) (z) +1 2 one can identify the constant c by using the leading terms in the expansions (7.134) and (7. Hν (z) as well as H−ν (z) are solutions of this equation.e.141) is o W (z) = − In case of g(z) = H (1) (z) +1 2 c .136) (1) 2 (z) = 0 ds [s(s ± is/2)] e−zs . 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 (7.93) only as ν 2 . One obtains W (z) = − 4i .139) (z) = 2 ±i[z − ( e πz +1) π ] 2 r=0 ( + r)! r!( − r)! i ± 2z r (7. converge fast for |z| → ∞. (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. h(z) to be linearly independent. For such solutions g(z).140) Bessel Functions with Negative Index Since ν enters the Bessel equation (7.137) (z) = r=0 r ± i 2 r 0 ∞ ds s +r e−zs .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. We employ for this purpose the Poisson integrals (7.135) which read for ν = + 1 2 H where f The binomial formula yields f (1) 2 (1) 2 1 +2 (1.141) (1) (1) the derivation of which follows the derivation on page 192 for the Wronskian of the radial Schr¨dinger equation.. πz (7. h) must be a non-vanishing function. i. hence.140). the Wronskian W (g.143) . As a second order differential equation the Bessel equation has two linearly independent solutions. z (7.93) holds the identity 1 W = − W z (7. The general solution of (7.138) The formula for the Laplace transform of sn leads to f and. (7.

7.132. (7.148) Following the arguments provided above (see page 193) the functions h (z) are solutions of the radial Schr¨dinger equation of free particles (7.2) H + 1 (z) .53).144) The expansion coefficients A. 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| → ∞ .96. 2z 2 (1.149.2) (z) +1 2 199 are.146) (z) = −i (−1) H (2) 1 (z) +2 . (7.2) (7.153) .7.2) (2) +1) 2 (1) +1) 2 (z) = i (−1) H (1) (z) +1 2 .147) (z) = π (1. (7.147) this can be written n (z) = 1 2i h (1) (z) − h (2) (z) . in fact.148) holds for the o regular spherical Bessel function j (z) = 1 (1) (2) [ h (z) + h (z) ] . 7.140).145) This equation can hold only for B = 0 and A = exp[i( + 1 )π].e. 7. According to (7. (7. one can show H−( Spherical Hankel Functions In analogy to equations (7. H (1.97) one defines the spherical Hankel functions h (1. H−ν (z) = A H (1) (1) (z) +1 2 One can then expand + BH (2) (z) +1 2 . linearly independent. (7. 2 (1.152) (7.151) are equivalent to h h (1) (2) (z) (z) = = j (z) + i n (z) j (z) − i n (z) . (7. B can be obtained from the asymptotic expansion (7. (7. 7. We conclude 2 H−( Simarly.149) We want to establish also the relationship between h (z) and the irregular spherical Bessel function n (z) defined in (7..132) follows n (z) = 1 2 π (−1) 2z +1 H−( (1) 1 +2) (z) + H−( (2) +1) 2 (z) . 7.90.150) According to (7.146.97.2: Free Particle i. From (7. 7.151) Equations (7.97).2) (7.

158) one can derive then from (7.157) 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. Hence. 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).140. as given by (7. (1) (7.156) ( + r)! r!( − r)! ( + r)! r!( − r)! i 2z r [ /2] = r=0 ( + 2r)! (2r)!( − 2r)! −1 4z 2 r (7.162) The leading terms in these expansions.2) (z). 7.154) The leading term in this expansion.163) (7.161) (7. From (7. j (z) (z) = ( i) z +1 e±iz r=0 ( + r)! r!( − r)! ± i 2z r (7. = −cos(z − π/2) results in 2 2 the alternative expressions j (z) n (z) = = sin(z − π/2) cos(z − π/2) p (z) + q (z) z 2z 2 cos(z − π/2) sin(z − π/2) q (z) − p (z) . are j (z) n (z) = = (7.154) 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) . (7.148). 2 2z z (7. it follows z ∈ R : Using p (z) = Re r=0 j (z) = Re[h (1) (z)] .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).160) Employing cos[z − ( + 1) π ] .140) and (7. ne ll(z) = Im[h (1) (z)] . is h (1) 2 (z) = ( i) z +1 e±iz . at large |z|. are complex conjugates.164) .159) (7. 2z 2 z sin(z − π/2) z cos(z − π/2) − . z (7. = sin(z − π/2) and sin[z − ( + 1) π ] .148) one obtains readily h (1) 2 (1.

2) g (z) − g (z) z 2 +1 g (z) − g z (7.175) Using Γ( + 1) = !.171–7.171) −1 (z) (7.174) We want to prove these relationships. 1 (7.114) we can write g (z) = a z f 2 ! + 1 (z) 2 . (1. 1.148) and express g (z) = h (1. defining a = −1 and employing fν (z) as defined in (7. 2 j0 (z) n0 (z) j1 (z) n1 (z) j2 (z) n2 (z) = = = = = = sin z z cos z − z sin z cos z − z2 z cos z sin z − 2 − z z 3 1 3 − sin z − 2 cos z 3 z z z 3 1 3 − 3 + cos z − 2 sin z z z z (7.174) and from (7. One obtains for = 0. One can combine (7. 7. From the linearity of the relationships (7.165) (7.172) where g (z) is either of the functions h obtain the recursion relationship g g (z).176) .171) holds for g (z) = h (z) we employ (7.2) To demonstrate that (7.2) relationships hold for g (z) = h (z).7. 7.124.2: Free Particle Expressions for the Spherical bessel Functions j (z) and n (z) 201 The identities (7. 7.171.168) (7.173) A (z) =  z ( +1) z2 +2 z (7.166) (7.4 dx (1 − x2 )ν− 2 eizx .169) (7.167) (7. 7.131.151) follows then that the relatiosnhips hold also for g (z) = j (z) and g (z) = n (z).2) (z) = −2 π 1 √ 2z πΓ(ν + 1 ) 2 z 2 ν C2. 7. For this purpose we need to demonstrate only that the (1.161.170) Recursion Formulas of Spherical Bessel Functions The spherical Bessel functions obey the recursion relationships g g +1 (z) +1 (z) = = (1.172) to +1 (z) +1 (z) = A (z)  1− g (z) g (z) −1   (7. j (z) and n (z).149.162) allow one to provide explicit expressions for j (z) and n (z). (7.

179) from which follows (7.172) we differentiate (7.173. i. f yields.5: Provide a detailed derivation of (7.e.2. . (7.176) g (z) = z g (z) − g +1 (z) +1 2 Spherically Symmetric Potentials z g (z) + a z f 2 ! +1 2 (z) .166).172). 7.2.180) yields g +1 (z) = g (z) − ( + 2) g (z) + z2 +2 g (z) .165). (7.173). Exercise 7.182) Replacing all first derivatives employing (7. Exercise 7.171).177) (z) = − z f 2 +2 + 3 (z) 2 . j0 (z) using (7.171).171).180) Since g (z) is a solution of the radial Schr¨dinger equation (7. z (7.121). To prove (7.173.202 The derivative of this expression is g (z) = Employing (7.174) to determine (a) j1 (z).179) leads to (7. The second component of (7. 7.173) is equivalent to (7. (7. j2 (z) from j0 (z).179) g (z) = − z2 g (z) + g (z) − g +1 (z) z . In order to prove (7.6: Employ the recursion relationship (7.182).58) it holds o 2 g (z) = − g (z) + z ( + 1) g (z) − g (z) . together with (7. n2 (z) from n0 (z).172). and (b) n1 (z).178) (7.181) Using this identity to replace the second derivative in (7. n0 (z) using (7. The first component of (7. z2 (7.174) we start from (7. in fact. is equivalent to (7..

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

the relationship between fields and potentials is not unique. t) = − V (r.11) V (r. t) where V (r. t) − ∂t A(r. t) = 4 π J(r. t) Interaction of Radiation with Matter = 4 π ρ(r. t) . t) − ∂t χ(r. t). From this follows E(r. t) . t) (8. the two homogeneous Maxwell equations hold × E(r. called the scalar potential. t] + v × B[ro (t). (8. Jackson (John Wiley & Sons. t) which is solved by E(r. New York.3) reads then × E(r. t) in (8. t) + ∂t A(r. t) · B(r. t) = × A(r. t) −→ 1 A(r.8) = 0 (8. t) in (8. t) = − V (r. t] (8. 1975) for a discussion of these and other conventional units. t) + v × B(r. t) + χ(r. t) = 0 = 0. . in turn.4) Lorentz Force A classical particle with charge q moving in the electromagnetic field experiences the so-called Lorentz force q[E(r.1) the term qδ(r − ro (t)) and to the current density ˙ J(r. D. In the non-relativistic limit holds p ≈ mr and (8. (8.7) Gauge Transformations We have expressed now the electric and magnetic fields E(r. accordingly.6) for some vector-valued function A(r. The following substitutions.10) (8. contributes to the charge density ρ(r. called the vector potential.3) (8. t) . but leave the fields unaltered: A(r. 2nd Edition.25). called gauge transformations. alter the potentials. The reader is referred to the well-known textbook ”Classical Electrodynamics”. by J. As is well known. The particle. solves implicitly (8.1) (8. (8. t) through the scalar and vector potentials V (r. t) + ∂t A(r. t)] and. Scalar and Vector Potential Setting B(r. t) × B(r. obeys the equation of motion d p = q dt E[ro (t). Equation (8. t) (8.4). t) and A(r.2) In addition.204 Maxwell equations1 · E(r. t) and B(r. t) + ∂t B(r.5) above agrees with the equation of motion as given in (1.5) where p is the momentum of the particle and ro (t) it’s position at time t. t). t) −→ V (r.2) the term q r˙o δ(r − ro (t)). t) − ∂t E(r. t) is a scalar function. We assume so-called Gaussian units.9) (8.

13) · A(r. t) . such that the latter describes the electromagnetic radiation.19) couples the vector potential A(r. equation (8. t) − 2 × A(r.15) d3 r ρ(r . t). t) − ∂t V (r. (8. t) . t) . t) can be obtained employing (8. One would prefer a description in which the Coulomb potential (8.19) and define J (r.14) In case of the boundary condition V (r. Using the expressions (8. in fact. (8.19) Unfortunately. t) = 0 the solution is given by the Coulomb integral V (r. (8.6) and (8. t = 4 π J(r.16) This is the potential commonly employed in quantum mechanical calculations for the description of Coulomb interactions between charged particles. t) . t) = − 4 π J(r. t) + ∂t A(r.12) yields then the Poisson equation 2 V (r. t) = ∂t = 4 π ρ(r. t) + ∂t V (r. t) (8. the gauge freedom allows us to impose on the vector potential A(r. a name which is due to the form of the resulting scalar potential V (r. (8. For this purpose we examine the offending term ∂t V (r.17) A(r.16) and the vector potential are uncoupled. be achieved. In fact. t) |r − r | (8. t) together with (8.9) for the fields results in × The identity × × A(r.12) The corresponding gauge is referred to as the Coulomb gauge.20) . and the former the Coulomb interactions in the unperturbed bound particle system. t) = Ω∞ for r ∈ ∂Ω∞ (8. t) the condition · A(r. t) . t).1) · − V (r. t) Using · ∂t A(r. t) in (8. t) and V (r. (8. t) − ∂t A(r. and a component due to the remaining currents.9) into (8. the second inhomogeneous Maxwell equation. The vector potential A(r. Such description can. t) − ∂t A(r.12) leads us to 2 2 A(r. 4π (8. In fact.8. t) = 0 . t) = 1 ∂t V (r. t) = − 4 π ρ(r. this potential results from inserting (8. t) = · A(r.2). t) and the current due to moving net charges.1: Electromagnetic Field 205 This gauge freedom will be exploited now to introduce potentials which are most suitable for the purpose of separating the electromagnetic field into a component arising from the Coulomb potential connected with the charge distribution ρ(r.18) together with condition (8.

t) contributes solely an electric field component E (r.g. (8.14).29) · Et (r. exist in the space considered. In this case (8. the name longitudinal electric field. The definitions of J and Jt applied to (8..30) . t) = 0 (8. t) contributes an electrical field component as well as the total magnetic field Et (r. respectively.28) (8. such current densities exist. t) These fields are obviously divergence -free (e.206 For the curl of J holds × J (r. (8. t) = − ∂t ρ(r. t) be the total current of the system under investigation and let Jt = J − J . hence. (8. t).23) This continuity equation identifies J (r.22) 1 ∂t 4π · J (r. for example. t) = − V (r.16) yield finally the electric and magnetic fields. t) Bt (r. A(r. t) − ∂t A(r. = − ∂t A(r.. we want to assume that no ring-type currents. no divergence-free currents. Let J(r. t) = 0 .21) and (8. t) = × A(r.24) (8. t) (8.26) turns into the well-known wave equation 2 2 A(r. V (r. t) (8. Presently.12) and the Coulomb potential (8. For the divergence of J (r. t) = 0 . i.27) which is obviously curl-free ( × E (r. V (r.19) yield 2 2 A(r.21) ∂t and the Poisson equation (8. t) holds. t) as the current due to the time-dependence of the charge distribution ρ(r.25) one refers to J and Jt as the longitudinal and the transverse currents. t) + ∂t ρ(r. t) .e. (8.2 Planar Electromagnetic Waves The current density Jt describes ring-type currents in the space under consideration. hence. in a ring-shaped antenna which exhibits no net charge. t) + ∂t ρ(r. t) = − 4 π Jt (r. t) = 0). 8. using ∂t · J (r. t) . t) − ∂t A(r. t) = 0). For J also holds the continuity equation · J(r. t) = 0 . yet a current. The vector potential determined through (8.26) This equation does not couple anymore scalar and vector potentials.26) and (8.25) Because of properties (8. t) = or = 2 Interaction of Radiation with Matter (8. t) = 0 and from this follows · Jt (r. the name transverse fields.

t) = ±i ω A(r. hence.31) and the resulting fields (8. according to (8.37) ˆ ˆ where k is the unit vector k = k/|k|.36) 4π Using the identity a × (b × c) = b (a · c) − c (a · b) and (8. Here the “-” sign corresponds to so-called incoming waves and the “+” sign to outgoing waves2 . 2 . Time average over one period 2π/ω yields S(r. (8.28. (8. t) = Re Et (r. see also the comment below Eqs. 8. 8. is orthogonal to k.31) u is a unit vector (|ˆ| = 1) which.32) holds. the constant k is referred to as the wave vector. ˆ We want to characterize now the radiation field connected with the plane wave solutions (8. The corresponding electric and magnetic fields. accordingly.32. 8. 8π (8.29). t) × Re B(r. Et (r. are orthogonal to each other and are both orthogonal to the wave vector k. A complete set of solutions is given by the so-called plane waves A(r. 8. The Coulomb gauge condition (8.31. at each point r and moment t. t) = i k × A(r.35) are complex-valued quantities. Note that in the units chosen the velocity of light is c = 1.35) one obtains S(r.35).34.158).38) ˆ In this expression for the energy flux one can interprete k as the propagation velocity (note c = 1) and. This flux is given by 1 S(r. Obviously.34) (8. 8. ω2 = |Ao |2 (8. In applying the potential and fields to physical observables and processes we will only employ the real parts.39). 8. t) and Bt (r.34. t) . obviously. ˆ (8.31). The latter vector describes the direction of propagation of the energy flux connected with the plane wave electromagnetic radiation.38. t) = Ao u exp i(k · r ˆ where the dispersion relationship |k| = ω (8.12) yields u·k = 0. t) = ± ω2 ˆ |Ao |2 k sin2 ( k · r − ωt ) 4π (8. (8. there exist two linearly ˆ u independent orientations for u corresponding to two independent planes of polarization. t) .39) 8π The definition incoming waves and outgoing waves is rationalized below in the discussion following Eq. t) in (8.33) ωt) (8. t) Bt (r. (8.8.35) The vector potential in (8.34. t) = ± ω2 ˆ |Ao |2 k . 8.2: Planar Electromagnetic Waves 207 which describes electromagnetic fields in vacuum. are Et (r.

16.40) = V It should be pointed out that Nω represents the number of photons for a specific frequency ω. i (8. t) . (8. t). We assume that the energy content of the fields is not altered significantly in the processes described and. Comparision of (8.43) Here the fields are defined through Eqs.39) and (8. t) is real 1 and that Et (r.8.41) ωV Inserting this into (8. (8.32).43).27. A ‘poor man’s’ quantization of the field is possible at this point by expressing the energy density (8. t) 2m + 1 8π Ω∞ 2 H = + qV (r) 1 16π d3 r |Et |2 + |Bt |2 Ω∞ d3 r E 2 + . (8. (8. t) = H Ψ(r. 8. The sign in (8.208 Interaction of Radiation with Matter as the energy density. These photons each carry the energy ω. hence. the energy of the plane wave is transported in the direction of −k. whereas in the case of outgoing waves the energy is transported in the direction of k. t) 2m 2 + qV (r) . The wave function Ψ(r. we will neglect the respective terms in the Hamiltonian (8. Bt (r. If we consider a volume V with a number of photons Nω the energy density is obviously Nω ω . a ˆ specific k and a specific u. respectively. Note that E (r. A correct description of the electromagnetic field requires that the field be quantized. 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. t) = 8πNω u exp i(k · r − ωt) ˆ ωV .45) .44) .3 Hamilton Operator The classical Hamiltonian for a particle of charge q in a scalar and vector potential V (r) and A(r. (8.42) 8. u · k = 0 . defined below Eqs. ˆ (8. |k| = ω . The integrals express the integration over the energy density of the fields.31. t) are complex leading to the difference of a factor 2 in the energy densities of the lontitudinal and transverse components of the fields. 8. We are left with a classical Hamiltonian function which has an obvious quantum mechanical analogue ˆ H = ˆ p − q A(r. (8.39) through the density of photons connected with the planar waves (8.38) implies that for incoming waves. t). is p − q A(r.29) together with the potentials (8.31).28. 8.40) allows one to express then the field ˆ amplitudes 8πNω Ao = .31) allows one finally to state for the planar wave vector potential A(r.31).

t)|2 and.t)/ should not affect a system and that. an element of the group U(1).51) 2m the potentials A(r.11) of the potentials is equivalent to multiplying the wave function Ψ(r.47. 8. (8.50) The equivalence of (8. t) are necessary to compensate terms which arise through the phase factor. . t) .46) the fields enter.44. It should be noted. Elementary constituents of matter which are governed by other symmetry groups. that this principle applies only to fundamental interactions.47) is equivalent to   2 ˆ p − qA   i ∂t eiqχ(r. An important conceptual step of modern physics has been to turn the derivation given around and to state that introduction of a local phase factor eiqχ(r. In the classical case such question is mute since the gauge transformations do not alter the fields and. t) = eiqχ(r. t) by a local and time-dependent phase factor eiqχ(r. t) ˆ = eiqχ(r.. by the group 3 The effect on other expectation values is not discussed here.g.e.t)/ .48) 2m For this purpose one notes i ∂t eiqχ(r. t) and A(r. This leads to the question in how far the gauge transformations (8.t)/ [ i ∂t − q((∂t χ)) ] Ψ(r. (8. in the Schr¨dinger equation o   2 ˆ p − qA   i ∂t Ψ(r.t)/ p + q(( χ)) Ψ(r.t)/ Ψ(r.49) (8. t) ˆ p eiqχ(r.10. does not change expectation values which contain the probability densities3 . hence.t)/ Ψ(r.45) leads to the Schr¨dinger equation o   2 ˆ p − q A − q(( χ))   + qV − q((∂t χ))  Ψ(r. (8.48) implies that the gauge transformation (8. have no effect on the motion of the particle described by (8. t) .. t) =  + qV  eiqχ(r. t) r (8.t)/ constitutes a unitary transformation of a scalar quantity. One can show that (8. e. Obviously.46). t) enter whereas in the classical equations of motion ˙ m¨ = q E(r. t) =  + qV  Ψ(r.3: Hamilton Operator 209 Gauge Transformations It is interesting to note that in the quantum mechanical description of a charged particle the potentials V (r. accordingly.10.t)/ Ψ(r. t) (8. such phase factor does not change the probability density |Ψ(r.10. t) =  2m where ((· · ·)) denotes derivatives in (( χ)) and ((∂t χ)) which are confined to the function χ(r. however. hence. 8.11) affect the quantum mechanical description.t)/ Ψ(r. The idea just stated can be generalized by noting that multiplication by a phase factor eiqχ(r. t) inside the double brackets. t) .11) to (8. t) + q r × B(r. t) and V (r. i. not to phenomenological interactions like the molecular van der Waals interaction.47) i ∂t Ψ(r. Applying the gauge transformations (8. 8. 8. 8.8.

In case of a homogeneous potential pointing in the x3 -direction. (8. e.59) .52) This expression vanishes in the present case since since ˆ ˆ p · Af = A · pf and. H = 2m m 2m (8. holds (8.44) can be expanded ˆ2 p q H = − 2m 2m For any function f (r) holds ˆ ˆ p · A − A · p f (r) = i A· f + f ·A − A· f = i f ·A. For this purpose we use the wave function in the form Ψ(E. affects the complexity of the mathematical derivation of the wave functions. the generators of SU(2). j = 1. likewise can demand the existence of fields which compensate local transformations described by eiσ·χ(r. k2 . The Hamiltonian (8. We want to describe the stationary states corresponding to the Hamiltonian (8.57) with a homogeneous magnetic field Bo .g.58). 3.45) with Hamiltonian (8. 2.58) where ∂j = (∂/∂xj ). x3 ) = exp(ik2 x2 + ik3 x3 ) φE (x1 ) . x1 . x2 . k3 .210 Interaction of Radiation with Matter SU(2). t) ≡ 0. (8. (8. Accordingly. t) = Bo .56). consequently. (8.53) ˆ ˆ p·A + A·p + q2 2 A + qV 2m (8. t).t) where σ is the vector of Pauli matrices. · A = 0 [cf. for Bo = Bo e3 in (8.55) 8. The vector potential (8.56) due to the gauge freedom. The stationary homogeneous magnetic field B(r. The choice of a vector potential affects the form of the wave functions describing the eigenstates and. thereby.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 (8.57) satisfies · A = 0 and. 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. one can employ the Hamiltonian (8. Solution for Landau Gauge A particularly convenient form for the Hamiltonian results for a choice of a so-called Landau gauge for the vector potential A(r.55)..12)]. therefore. In this o case holds V (r.55). can be described by various vector potentials. the so-called Landau gauge ˆ associates the vector potential AL (r) = Bo x1 e2 ˆ (8.54) ˆ2 p q ˆ q2 2 − p·A + A + qV . The resulting fields are called Yang-Mills fields.

56) through the vector potential A(r) = 1 Bo × r . Solution for Symmetric Gauge The solution obtained above has the advantage that the derivation is comparatively simple.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. like the corresponding gauge (8. (8. the states are degenerate in the quantum number k2 describing displacement along the x2 coordinate.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 ) .59). Without affecting the energy one can form wave packets in terms of the solutions (8. x1 .65) where we replaced the parameter E by the integer n. x2 . Unfortunately.66) Obviously.57).63) (8. 2 (8. is Ψ(n. therefore. k3 ) = ω(n + 2 k2 1 3 ) + .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 ) . From this observation one can immediately conclude that the wave shifted (by e 3 function of the system. the familiar harmonic oscillator quantum number.62) x1o = and where ω = k2 eBo eBo me (8. k2 . 2 2me (8.67) .64) is the classical Larmor frequency (c = 1). However.62) is that of a displaced (by x1o ) harmonic oscillator with o 2 k 2 /2m ) energies. 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. is not symmetric in the x1 .65) which localize the electrons. The energies corresponding to these states are E(n.65). the so-called symmetric gauge which expresses the homogeneous potential (8. k2 .63) this induces a spread of the wave function in the x1 direction. k3 .and x2 -coordinates. (8. the wave function (8. The stationary Schr¨dinger equation (8. according to (8. according to (8. We want to employ .59).8. 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.

Accordingly.12) for the Coulomb gauge. (8.74) We consider a simple case to relate (8.69) × u and × r = 0. using × (uf ) = −u × f +f ˆ ˆ Bo · p × r f = Bo · r × p f . For the vector potential (8. (8.78) .68) × u yields.70) ˆ ˆ p · A f = − Bo · p × r f . namely.72) allows one to interpret µ = − e L 2me (8. Vmag = e 2me class = r me v e3 of the electron ˆ (8. ˆ Identifying r × p with the angular momentum operator L. y-plane with constant velocity v on a ring of radius r. Vmag = −µclass · Bo (8.74). ˆ (8. an electron moving in the x.77) Comparision with (8. in the present case of constant Bo .67) one can write ˆ p·A = Using · (u × v) = − u · function f (r) ×v + v· 2i · Bo × r . the magnetic moment (8.72) and (8.76) e 2me class class · Bo .73. (8.212 Interaction of Radiation with Matter One can readily verify that this vector potential satisfies the condition (8. In this case the current density measures −e v oriented tangentially to the ring.73) where µclass is the magnetic moment connected with a current density j µclass = 1 2 r × j(r) dr (8. the Hamiltonian (8.52) becomes ˆ2 e e2 p + Bo · L + H = 2me 2me 8me Of particular interest is the contribution Vmag = e L · Bo 2me Bo × r 2 .72) to Hamiltonian (8. 8. (8.74) is in the present case 1 µclass = − e r v e3 .71). for any (8. The latter can be rewritten. accordingly.71) (8. namely. The theory of classical electromagnetism predicts an analogue energy contribution .75) 2 The latter can be related to the angular momentum µclass = − and.

A derivation of his property and the value of g.85) The Hamiltonian (8. x3 ) = exp(ik3 x3 ) ψ(x1 .86) 1 2 2me 1 2 ˆ as well as of the angular momentum operator L3 .83) (8.79) To obtain the stationary states. For a magnetic field (8. the so-called gyromagnetic ratio of the electron. x2 .e. 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. requires a Lorentz-invariant quantum mechanical description as provided in Sect. for Bo = Bo e3 . 2 . x3 ) .72) with an associated magnetic moment − g 2me S where g ≈ 2. likewise.8. 10.and the ±x2 -direction.87) . described by the operator S. We will demonstrate in Sect. (8. The functions ψ(x1 . x2 . 2me We have used here the expression for the angular momentum operator ˆ L3 = ( /i)(x1 ∂2 − x2 ∂1 ) .80) (8. 2me (8.88) me ω xj .84) .e.67) yields a more symmetric solution which decays to zero along both the ±x1 . we separate the variable x1 .71) is ˆ H = 2 ˆ2 p e2 Bo + 2me 8me x2 + x2 + 1 2 eBo L3 . (8. x2 from x3 setting ΨE (x1 . Accordingly. the Hamiltonian (8. E = E − (8.83) can then be expressed 1 ˆ 1 Ho = − ω 2 ∂2 ∂2 + 2 2 ∂X1 ∂X2 + 1 2 2 X1 + X2 2 + 1 i X1 ∂ ∂ − X2 ∂X2 ∂X1 . x2 ) where ˆ Ho = − 2 2 2 ∂1 + ∂2 (8. x3 ) = E ΨE (x1 . 10 that the spin of the electron. x2 ) = E ψ(x1 . To obtain these eigenstates we introduce the customary dimensionless variables of the harmonic oscillator Xj = (8. x2 . x2 ) .4: Electron in Homogenous Magnetic Field 213 as the quantum mechanical magnetic moment operator for the electron (charge −e).81) (8. ˆ i.83) describes two identical oscillators along the x1 -and x2 -directions which ˆ are coupled through the angular momentum operator L3 .56) pointing in the x3 -direction the symmetric gauge (8.82) 1 ω (x1 ∂2 − x2 ∂1 ) i 2me + 1 me ω 2 x2 + x2 1 2 2 2 k2 3 + (8. x2 ) obey then ˆ Ho ψ(x1 . In this case. the solutions of ˆ H ΨE (x1 . e gives rise to an energy contribution (8. j = 1. i..

. x2 ) = ( 2j + 1 ) Ψ (j. therefore.91) We note that the operator a† a2 − a† a1 leaves the total number of vibrational quanta invariant. x1 . 1 3 .95) (j) ˆ We want to choose the coefficients αmm such that (8.94) is an eigenstate of Ho ˆ Ho Ψ (j. (j) (8. m . +j . 2 2 . aj = √ 2 Interaction of Radiation with Matter Xj + ∂ ∂Xj . (8. x2 ) . m = −j. (8. 2 (8. however. We. it holds a† a1 + a† a2 + 1 1 1 2 Ψ (j. x1 .92) are not eigenstates of L3 . x2 ) (8. x1 . . .. in fact. x1 .214 Employing the creation and annihilation operators 1 a† = √ j 2 and the identity ˆ ω L3 = which can readily be proven.97) . such that 1 i a† a2 − a† a1 1 2 Ψ (j. m . If this property is. x2 ) = ω 2j + 2m + 1 Ψ (j. (8. i.. x2 ) is the wave function for the state with zero vibrational quanta for the x1 . x2 ) = 2m Ψ (j. m . x1 . x2 ) .as well as for the x2 -oscillator. x1 . through a combination of states j Ψ (j. 1 2 since one phonon is annihilated and one created.94) is an eigenstate of the vibrational Hamiltonian.89) . the total vibrational energy being ω(2j + 1). . x2 ) = m=−j αmm Ψ(j. i. x1 . .94) is also an eigenstate of L3 . −j + 1. (8. 0. (8. x2 ) . m . attempt to express eigenstates in terms of vibrational wave functions a† 1 j+m Ψ(j. m . j = 1. x1 . (8.94) Since this state is a linear combination of states which all have vibrational energy (2j + 1) ω. 1. obeyed. m. x1 .e.96) ˆ holds. In order to cover all posible vibrational quantum numbers one needs to choose j. m.90) a† a2 − a† a1 1 2 . x2 ) = a† 1 j−m (j + m)! (j − m)! Ψ(0. 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 . . 0. . x1 . (8.93) ˆ The states (8. x2 ) (8.92) where Ψ(0. m . x1 . m as follows: j = 0.92) represents a state with j + m quanta in the x1 -oscillator and j − m quanta in the x2 -oscillator.e. (8. m . Such eigenstates can be expressed. .

84) above. thus. 5.10. (8. the 2 2 latter representing the Wigner rotation matrix of Sect. 0) Ψ(j. the states are eigenstates of the operator [we use for the operator the notation of Sect. 5. π . Eq. 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 (8.55). The connection with the present problem arises due to the fact that the operator J2 in Sect. We assume that the scalar potential V in (8.100) J2 = 1 2 2i 1 ˆ i.98) then the states Ψ(j. x1 . x1 .. a1 . m. the eigenstates of (8. The first rotation contributes a factor exp(−im π ). t) introduced above. This implies that ˆ we can obtain eigenstates of L3 by rotation of the states Ψ(j. an example is the Coulomb potential V (r.31). cf.11. 8. According to the derivation given there. i.101) 2 2 where Dmm ( π . The required rotation must transform the x3 –axis into the x2 –axis. b− + − notation in Sects.288)] 1 ˆ J3 = 2 a† a1 − a† a2 1 2 (8. a† .8. 5.5: 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. 5.. Other radiation fields can . is identical to the operator L3 introduced in (8.11 such transformation is provided through π π (j) Ψ (j. x2 ) defined in (8.55) confines the particle to stationary bound states.102) We have identified.83) and confirmed the eigenvalues stated in (8.5. Using the explicit form of the Wigner rotation matrix as given in (5. x1 . m) in Sect. In the simplest case the radiation field consists of a single planar electromagnetic wave described through the potential (8.11. According to Sect.(5. the second rotation a factor dmm ( π ).9. m.5. which corresponds there to the angular momentum in the x2 –direction. If we identify 2 a† . b+ . 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. x1 . m .10. x2 ) .9.99) with eigenvalue m.97).5. a transformation moving the x3 –axis into the x2 – 2 axis.11 (8. x1 .5.92) correspond to the eigenstates |Ψ(j. except for a factor 2 .10.e. a2 1 2 present notation ←→ b† . x1 . 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 . x2 ) = Dmm ( . 5. m. . b† . is in the notation of the present section 1 ˆ a† a2 − a† a1 .309) yields finally Ψ (j. t) = 1/4πr confining an electron with energy E < 0 to move in the well known orbitals of the hydrogen atom.e. 5. m.10.9. The external radiation field is accounted for by the vector potential A(r. x2 ). 5. x2 ) (8. (8.

103) (8. i. t) + A (r.42).216 Interaction of Radiation with Matter be expanded through Fourier analysis in terms of such plane waves. The states |n are thought to form a complete.105) for the case of radiationinduced transitions in atomic systems. Estimate of the Magnitude of VS We want to demonstrate now that the interaction VS (t). we discount the possibility of a continuum of eigenstates. the term ∼ p · A(r. We will see below that the perturbation resulting from a ‘pure’ plane wave radiation field will serve us to describe also the perturbation resulting from a radiation field made up of a superposition of many planar waves. i. we assume n|m = δnm and for the identity ∞ (8. . .41) e ˆ e p2 p·A ∼ 2me me me 2me 4 1 2 8πNω . .e. t).107) 1 = 1 n=0 |n n| .105) where A(r.. (8. This result will allow us to neglect the larger than the second term.105). 2 . For an electron charge q = −e and an electron mass m = me one can provide the estimate for the first term of (8..104) (8. i. 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. as given in (8. one can estimate that the perturbation.105) in further calculations and to expand the wave function in terms of powers of VS (t) in a perturbation calculation. the term ∼ A second term in (8. t) is given by (8.108) We assume for the sake of simplicity that the eigenstates of Ho can be labeled through integers. (8.105) as follows4 . in this case. this assumption can be waved as our results below will not depend on it. and that the first term in (8. orthonormal basis..e. t). 1. is much 2 (r. i.109) The reader should note that the estimates are very crude since we are establishing an order of magnitude estimate only. is much smaller than the eigenvalue differences near ˆ typical atomic bound states. t) − m 2m (8.e. We first note. ωV (8.. In fact. using (8. n = 0. Here the so-called unperturbed system is governed by the Hamiltonian Ho with stationary states defined through the eigenvalue problem Ho |n = n |n .106) where we adopted the Dirac notation for the states of the quantum system. However.e.

We can. λ = 2πc/λ..42) it holds VS ≈ e m k. Nω = 1. hence.105) for radiation fields can be considered a small perturbation. Using again (8.114) Employing for the second factor the estimate as stated in (8.112) π λ me c2 and with e2 /ao ≈ 27 eV. e ˆ p · A ∼ 10 eV · 10−4 = 10−3 eV . Consequently.8. one obtains for (8. Hence..ˆ u 4πNk α(k. and me c2 ≈ A A 500 keV 2 ao ω ≈ 10−8 (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. u) p · u exp i(k · r − ωt) ˆ ˆ ˆ kV .105).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 (8.118) .115) (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 . i. altogether.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. with ao ≈ 0.e.105) due to a radiation field by VS = − q ˆ p · A(r. t) . i. me (8. m (8. one can neglect this term as long as the first term gives non-vanishing contributions. replace the perturbation (8. We want to estimate now the second term in (8.117) In case that such perturbation acts on an electron and is due to superpositions of planar waves described through the vector potential (8.5 ˚.112) yields e2 A2 ∼ 10 eV · 10−8 = 10−7 eV .110) where ao is the Bohr radius. 2me (8. (8. a volume V = λ3 where λ is the wave length corresponding to a plane wave with frequency ω.e. (8.111) For ω = 3 eV and a corresponding λ = 4000 ˚ one obtains. the first term in (8.116) This term is obviously much smaller than the first term. Assuming a single photon. and as long as the photon densities Nω /V are small.

The states |0 . Multiplying the latter equation by the operator exp i (8. Equation (8. describes the effect of a radiation field on an electron system and which will be assumed below to describe radiative transitions. (8. A factor α(k.123) i = Ho exp − Ho (t − to ) (8. The sum runs over all possible k vectors and might actually be an integral.42) through k = |k| = ω.120) (8.125) (8. u) has been added to describe eliptically or circularly polarized ˆ waves. |Ψ(to ) = |0 (8. n = 0 at some later time t ≥ to . under ordinary circumstances. Perturbation Expansion The generic situation we attempt to describe entails a particle at time t = to in a state |0 and a radiation field beginning to act at t = to on the particle promoting it into some of the other states |n . . the sum over u involves the two possible polarizations of planar ˆ electromagnetic waves. Using i i ∂t exp − Ho (t − to ) and (8. The probability to find the particle in the state |n at time t is then p0→n (t) = | n|ΨS (t) |2 ..106–8.127) i ∂t |ΨD (t) = VD (t) |ΨD (t) . One seeks to predict the probability to observe the particle in one of the states |n . For this purpose one needs to determine the state |ΨS (t) of the particle.122) where |ΨD (to ) = |0 .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) . n = 1. This state obeys the Schr¨dinger equation o i ∂t |ΨS (t) = [ Ho + VS (t) ] |ΨS (t) subject to the initial condition |ΨS (to ) = |0 . |n are defined in (8. .124) (8. .118) is the form of the perturbation which.108) as the eigenstates of the unperturbed Hamiltonian Ho .121) (8.218 Interaction of Radiation with Matter where we have replaced ω in (8. 2.119) 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) (8.126) Ho (t − to ) yields finally .

Employing this result one obtains for (8. 2.137) One can readily verify that (8.136) together with the initial conditions |ψD (to ) = |0 0 for n = 0 for n = 1.133. (8.130) = exp − i n (t − to ) − to )]| = 1.128) We note that the transition probability (8.137) |ΨD (t) = (1) (0) (8. .128).132–8. (8. Equations (8.8.127) we assume the expansion ∞ |ΨD (t) = n=0 (n) |ΨD (t) (n) (8.137) follows |ΨD (t) = |0 .127. We will consider here only the two leading contributions to |ΨD (t) .132) where |ΨD (t) accounts for the contribution due to n-fold products of VD (t) to |ΨD (t) . n| (8. Accord(n) ingly. we define |ΨD (t) through the evolution equations i ∂t |ΨD (t) i ∂t |ΨD (t) i (2) ∂t |ΨD (t) (1) (0) = = = .134. using |exp[− i n (t n (8. consequently. . 8. From (8.129) n| and. to (8. 8.139) . .137) are consistent with (8.133) (8. = .137) can be solved recursively. . 8. . 0 VD (t) |ΨD (t) (1) VD (t) |ΨD (t) (0) (8.5: Time-Dependent Perturbation Theory where VD (t) = exp 219 i i Ho (t − to ) VS (t) exp − Ho (t − to ) . In order to determine |ΨD (t) described through (8. Due to the Hermitean property of the Hamiltonian Ho holds n|Ho = i n|exp − Ho (t − to ) from which we conclude. .133–8.121) expressed in terms of ΨD (t) is i p0→n (t) = | n|exp − Ho (t − to ) |ΨD (t) |2 .134) (8. (8.135) i ∂t |ΨD (t) (n) VD (t) |ΨD (n−1) (t) (8.131) p0→n (t) = | n |ΨD (t) |2 . .138) 1 i t dt VD (t ) |0 .

yields for (8.144) describes the propagation of the planar wave. defines the direction of the E-field of the ˆ radiation. however. + m=0 dt to to dt n|VD (t )|m m|VD (t ) |0 8. We will demonstrate below that the the sign of iωt determines if the energy of the planar wave is absorbed (“-” sign) or emitted (“+” sign) by ˆ the quantum system. is VS = ˆ ˆ V1 exp(−iω1 t) + V2 exp( iω2 t) eλt . note that uj .. A2 are defined through (8. to → −∞ (8. non-singular. The coefficients A1 . λ → 0+ .6 Perturbations due to Electromagnetic Radiation We had identified in Eq. according to (8. This factor will serve mainly the purpose of keeping in the following derivation all mathematical quantities properly behaved.34.41). in turn. A factor exp(λt).142) combining an incoming and an incoming or outgoing wave. We want to apply now the perturbation expansion derived to such perturbations. The factor III in (8. that two planar waves simulataneously interact with an electronic system.. The resulting perturbation on an electron system. (8.143) ˆ ˆ where V1 and V2 are time-independent operators defined as e ˆ Vj = m 8πNj ωj V I ˆ ˆ p · uj eik·r .135.137) |ΨD (t) = (2) Interaction of Radiation with Matter 1 i 2 t t dt to to dt VD (t ) VD (t ) |0 . i. In (8. 8.. II III (8. 8. (8.118) above that the effect of a radiation field on an electronic system is accounted for by perturbations with a so-called harmonic time dependence ∼ exp(−iωt).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. such that the combined radiation field is decribed by the vector potential A(r. λ → 0+ has been introduced which describes that at time to → −∞ the perturbation is turned on gradually.142).144) Here the factor I describes the strength of the radiation field (for the specified planar wave) as determined through the photon density Nj /V and the factor II describes the polarization of the planar wave. the direction of ˆ the propagation being determined by k = k/|k|. 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.e.220 This result. according to (8.144) r is the position of the electron and p = ( /i) is the momentum operator of the electron. For the sake of including the effect of superpositions of plane waves we will assume.141) + . .118).

149) Using the definition of VD stated in (8.145) we obtain then. 8.6: Perturbations due to Electromagnetic Radiation 1st Order Processes 221 We employ now the perturbation (8. (8. (8.128) one obtains k|VD (t) | = k|VS (t)| exp i ( k − ) (8.150) and. According to (8.145) n|ΨD (t) (1) m (t − to ) |m .151) ˆ ˆ n|V1 |m m|V1 |0 exp ˆ ˆ + n|V2 |m m|V2 |0 exp i i n − − m − ω1 − i λ)t exp ω2 − i λ)t exp m o n m m o .143) to the expansion (8. (8.140.143). 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 (8.147) 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 .146) = 1 λ→0+ t→−∞ i lim lim × t dt exp to i ( n − o − i λ) t iω2 t × .141). For the 1st order contribution to the transition amplitude n|ΨD (t) = (1) 1 i t dt n|VD (t ) |0 to (8. employing the perturbation (8.128). (8.130) and (for m = 0) i i exp − Ho (t − to ) |m = exp − for (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. (8.8.

i. For this rate holds .e.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.. We assume first that the first order transi(1) tion amplitude n|ΨD (t) is non-zero.152). We also assume for the final state n = 0 such that n|0 = 0 and p0→n (t) = | n|ΨD (t) |2 holds.155) λ→0+ ( o We are actually interested in the transition rate. (8. in which case one can expect that it is larger than the 2nd (2) order contribution n|ΨD (t) which we will neglect. provide now the transition probability p0→n (t) according to Eq.153) (8.152) m=0 i ˆ ˆ n|V1 |m m|V1 |0 exp m − o − ω1 − i λ + − 2 ω1 − 2i λ)t n − o − 2 ω1 − 2i λ ( n o − ˆ ˆ n|V2 |m m|V2 |0 exp ω2 − i λ m− o ˆ1 |m m|V2 |0 exp ˆ n|V + ω2 − i λ m− o ˆ2 |m m|V1 |0 exp ˆ n|V + m − o − ω1 − i λ 1st Order Radiative Transitions i 2 ω2 − 2i λ)t 2 ω2 − 2i λ n− o i ( n − o − ω1 ω2 − 2i λ)t ω2 − 2i λ n − o − ω1 ( n o i − − ω1 ω2 − 2i λ)t ω2 − 2i λ n − o − ω1 ( n o − The 1st and 2nd order transition amplitudes (8. respectively.148) and (8. the time derivative of p0→n (t). Using (8.131).222 ˆ ˆ + n|V1 |m m|V2 |0 exp ˆ ˆ + n|V2 |m m|V1 |0 exp i i ( ( − − Interaction of Radiation with Matter − ω1 − i λ)t exp ω2 − i λ)t exp i i ( ( − − ω2 − i λ)t − ω1 − i λ)t n m m o n m m o Carrying out the integrations and the limit limt→−∞ provides the result (2) n|ΨD (t) = − 1 ∞ 2 λ→0+ lim (8.148) and ∗ |z1 + z2 |2 = |z1 |2 + |z2 |2 + 2Re(z1 z2 ) (1) (8.

142–8. any attempt to do so.160) leads to final states |n with energy n = o ± ω2 .31) and (8. the two terms apearing on the rhs. In case the matrix elements are non-zero.143. to cancel the third term in (8. which is observed through the so-called spectral intensities of transitions |0 → |n . due to the uncertainty relationship would introduce a considerable perturbation to the system. describing either an incoming or an outgoing plane wave due to the vector potential A2 u2 exp i (k1 · r ˆ ω2 t) .159) leads to final states |n with energy n = o + ω1 . The δ-functions appearing in this expression reflect energy conservation: the incoming plane wave contribution of (8. The result of such average is.e.156) ( o The period of electromagnetic radiation absorbed by electronic systems in atoms is of the order 10−17 s. ˆ (8. one should average the rate over many periods of the radiation. the matrix elements can vary strongly for different states |n of the system.158). is much shorter than could be resolved in any observation.. The result supports our definition of incoming and outgoing waves in (8. (8.144). 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. however.142) ˆ ˆ The matrix elements n|V1 |0 and n|V2 |0 in (8. i. The time average will be denoted by · · · 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+ (8.156) such that the 1st order contributions of the two planar waves of the perturbation simply add. due to the vector potential A1 u1 exp i (k1 · r − ω1 t) . Hence.8. a property.158) play an essential role for the transition rates of radiative transitions. Using lim x2 + 2 δ→0+ = π δ(x) (8.158) ˆ + | n|V2 |0 |2 δ( − ω2 ) Obviously.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. The second contribution to (8. of this expression describe the individual effects of the two planar wave contributions of the perturbation (8. . in fact. 8. For the resulting expression the limit limλ→0+ can be taken.144). First.

157) yields.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. e.162) × and.164) (8.168) absorption of 2 photons ω1 .153–8.167) Taking now the limit limλ→0+ and using (8. as we did in the the case of 1st order transitions.152).224 2nd Order Radiative Transitions Interaction of Radiation with Matter We now consider situations where the first order transition amplitude in (8. exp[ i (±ω2 − ω1 )].e. ∞ − 2 ω1 − 2i λ)t n − o − 2 ω1 − 2i λ ( n o − z2 = m=0 ˆ ˆ n|V2 |m m|V2 |0 ω2 − i λ m− o exp i × (8..163) × (8. (8.166) In this expression the terms |zj |2 exhibit only a time dependence through a factor e2λt whereas the terms zj z ∗ k for j = k have also time-dependent phase factors.153) vanishes such that the leading contribution to the transition probability p0→n (t) arises from the 2nd order amplitude (8. hence.158). similarly..e. We define ∞ z1 = m=0 ˆ ˆ n|V1 |m m|V1 |0 m − o − ω1 − i λ exp i × (8. i. 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. Time average · · · t of expression (8.161) To determine the transition rate we proceed again.g. |z1 + z2 + z3 |2 t = |z1 |2 + |z2 |2 + |z3 |2 (8. it holds p0→n (t) = | n|ΨD (t) |2 . i. in analogy to (8.158). (2) (8.166) over many periods of the radiation yields exp[ i (±ω2 − ω1 )] t = 0 and.k=1 j=k 2 2 2 2 ∗ zj zk (8.. in Eqs.

In case of the first term in (8. the second term describes the processes leading to final states |n with energy n = o ± 2 ω2 and. according to its δ-function factor. this state is only virtually excited.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. it must hold n = o + 2 ω1 . |n ← |m . i.e. Most remarkably. each of energy ω1 . the system is perturbed through absorption of a photon with energy ω1 from the initial state |0 into a state |m . interference between the contributions from all intermediate states |m can arise. Similarly. promotes the system then to the state |n . which is stationary and energy is conserved. |m ← |0 (8.168) describe similar histories of the excitation process. The first term. 8.168) describes two intermediate histories. through absorption of a photon.. For this purpose we will employ the transition rate as given in Eq. m = o + ω1 ) and the evolution of state |m is described by a factor 1/( m − o − ω1 −i λ).169) pert. namely absorption/ emission first of photon ω2 and then absorption of photon ω1 and. describes processes which lead to final states |n with energy n = o + 2 ω1 and.. i. first absorption of photon ω1 and then absorption/ emission of photon ω2 .168) describe the time sequence of the two photon absorption/ emission processes. accordingly.e. i. The remaining two contributions in (8. the third term in (8.158) which accounts for such transitions. i.e.8.. (8. each of energy ω2 . describe the absorption of two photons. accordingly.e.7 One-Photon Absorption and Emission in Atoms We finally can apply the results derived to describe transition processes which involve the absorption or emission of a single photon. The expression sums over all possible virtually occupied states |m and takes the absolute value of this sum. a second perturbation.168) the interpretation is ∞ ˆ n|V1 |m m=0 1 m − o − ω1 − i λ virtually occupied state |m ˆ m|V1 |0 pert. The factors | · · · |2 in (8. the third term describes processes in which a photon of energy ω1 is absorbed and a second photon of energy ω2 is absorbed/emitted. vice versa. .. Similarly. describe the absorption/emission of two photons. there is no energy conservation necessary (in general.

For example. θ.143. the second term can be discounted in case of absorption. and the integral is =1. is then kabs = Dipole Approximation 2π e2 2πN m2 ωV e ˆ u · n| p eik·r |0 ˆ 2 ˆ ˆ V1 exp(−iωt) + V2 exp(+iωt) eλt . (8. t) = 8πN ωV u exp ˆ ı (k · r − ωt) .226 Absorption of a Plane Polarized Wave Interaction of Radiation with Matter We consider first the case of absorption of a monochromatic.172) δ( n − o − ω) (8. plane polarized wave described through the complex vector potential A(r. The absorption rate. i.158). in case of a radiative transition from the 1s state of hydrogen to one of its three 2p states. the wave functions are (n.m (r.178) These wave functions make significant contributions to this integral only for r-values in the range r < 10 ao . φ) (8. (8. =0. φ) = 6 r −r/2ao e Y1m (θ.143) will contribute to the absorption process. φ) a3 o 1s (8. to → −∞ (8. .174) We seek to evaluate the matrix element ˆ M = n| p eik·r |0 . .171) The perturbation on an atomic electron system is then according to (8. in this range one can expand eik·r ≈ 1 + ik · r + . (8.176) ψn=2.177) √ M = 6 0 ∞ 1 2π ia4 o r2 dr −1 dcosθ 0 ∗ dφ r e−r/2ao Y1m (θ. φ) × × eik·r e−r/ao Y00 (θ. m denote the relevant quantum numbers) ψn=1.179) .173) V1. according to (8. .175) The matrix element involves a spatial integral over the electronic wave functions associated with states |n and |0 .144) VS = where e 2πN ˆ ˆ p · u e±ik·r . φ) a3 ao o 2p (8.e. (8. ˆ (8. φ) = 2 − 1 2 1 −r/ao e Y00 (θ.170) We will employ only the real part of this potential.. λ → 0+ . 8.m=0 (r. the vector potential actually assumed is A(r. t) = 2πN ωV u exp ˆ ı (k · r − ωt) + 2πN ωV u exp ˆ ı (−k · r + ωt) . θ. However.2 = m ωV Only the first term of (8.

Ho ] i where Ho is the Hamiltonian given by (8. Transition Dipole Moment A further simplification of the matrix element (8.178) holds eik·r ≈ A 1 1 + O( 50 ) such that one can approximate eik·r ≈ 1 .104) and.182) one finds [ r.174) yields kabs = 4π 2 e2 N ω V u · n| ˆ |0 ˆ r 2 m( o − m n| [r.181) and (8.187) . Ho ] |0 = i i n) n|r |0 .181) For the commutator in (8.k=1 i m 3 pk ek = ˆ j.182) one obtains M ≈ Insertion into (8.7: One-Photon Emission and Absorption 227 One can estimate that the absolute magnitude of the second term in (8. (8. ] + [ r. Using (8.179) and other terms are never larger than 20π ao /λ. Ho ] = ˆ2 p [ r. V (r) = − .529 ˚ one concludes that in the significant integration range in (8.8.182). Ho ] = i m 3 pk ej δjk = ˆ j.175). One refers to this approximation as the dipole approximation. in case of the hydrogen atom.186) δ( n − o − ω) (8.184) and the commutation property [xk .175) can then be ˆ achieved and the differential operator p = i replaced by by the simpler multiplicative operator r.k=1 i ˆ p m (8. This simplification results from the identity m ˆ p = [ r. the value of the wave length for the 1s → 2p transition 2π c λ = = 1216 ˚ A (8. V (r) ] 2me =0 = Using r = 3 ˆ j=1 xj ej 1 2me 3 pk [ r. 2me r (8.185) from which follows (8. pj ] = i δkj one obtains ˆ [ r. We are now in a position to obtain an alternative expression for the matrix element (8. pk ] pk ˆ k=1 (8.182) (8. is Ho = ˆ (p)2 e2 + V (r) . Using |k| = 2π/λ.183) (8. pk ] + ˆ ˆ k=1 1 2me 3 [ r.180) ∆E2p−1s and ao = 0.

Integrating and summing accordingly over all contributions as given by (8. u2 and k ˆ ˆ coordinate system. say u1 and u2 . For the two angles α = ∠(ˆ1 . ˆ u ˆ (8. The unit ˆ ˆ ˆ ˆ can be chosen to point along the x1 . directions. three-dimensional vector. We can.158) above that in first order the contributions of all components of the radiation field add. in practical situations. ϕ2 = π/2. The δ-function appearing in this expression. i. The direction of this vector in the u1 . ϕ.35). hence. Accordingly.190) (8. k frame is described by the angles ϑ. x2 .. 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. obtain the transition rate in the present case by adding the individual transition rates of all planar waves present in the oven.228 Interaction of Radiation with Matter where we used the fact that due to the δ-function factor in (8. such that the thermal radiation present supplies a continuum of frequencies.174) one can replace n − o by ω. the direction of u1 is described by the ˆ ˆ ˆ ˆ angles ϑ1 = π/2.33) are orthogonal to each other. will actually be replaced by a distribution function which reflects (1) the finite life time of the states |n . ϕ1 = 0 and of u2 by ϑ2 = π/2. there are ony two linearly independent directions of u possible.188) ˆ Here dk is the integral over all orientations of k.193) .191) (8.187) and using k c = ω results in the total absorption rate kabs (tot) = e2 Nω ω 3 2π c3 ˆ dk u ˆ u · n| ˆ |0 ˆ r 2 (8. |0 . ˆ In order to carry out the integral dk we note that u describes the possible polarizations of the ˆ ˆ and u. ρ) and ˆ u β = ∠(ˆ2 . u2 . Instead of adding the components of all possible k values we integrate over all k using the following rule +∞ =⇒ k u ˆ V −∞ k 2 dk (2π)3 ˆ dk u ˆ (8. We have demonstrated in our derivation of the rate of one-photon processes (8. and all polarizations of the radiation. planar waves as defined in (8.189) where the factor 1/ arose from the integral over the δ-function.e. k ˆ As a result. ρ) holds then u cosα = cosϑ1 cosϑ + sinϑ1 sinϑ cos(ϕ1 − ϕ) = sinϑcosϕ and cosβ = cosϑ2 cosϑ + sinϑ2 sinϑ cos(ϕ2 − ϕ) = sinϑsinϕ . according to (8. a very hot oven.31–8. Absorption of Thermal Radiation We want to assume now that the hydrogen atom is placed in an evironment which is sufficiently hot. and (2) the fact that strictly monochromatic radiation cannot be prepared such that any radiation source provides radiation with a frequency distribution. x3 -axes of a right-handed cartesian vectors u1 . one can express | u · n| r |0 |2 = |ρ|2 ( cos2 α + cos2 β ) = sin2 θ .192) and obtain ˆ dk u ˆ u · n| ˆ |0 ˆ r 2 2π 1 = |ρ|2 0 −1 dcosϑ (1 − cos2 ϑ) = 8π 3 (8.

a problem in case of emission by quantum systems in complete darkness.194) combined are typically somewhat smaller than (1 ˚/1000 ˚)3 = A A 10−9 . in particular.172. (8. . ˆ In case of emission only the second term V2 exp(+iωt) in (8. i. The quantum nature of radiation manifests itself in that the number of photons Nω msut be an integer.37 × 1019 1 ao | n|r|0 |2 × s λ λ2 . Otherwise. the calculation of the emission rate proceeds as in the case of absorption.173). can be inserted into (8. i. The value of | n| r |0 | determines the strength of an optical transition.171) and perturbation (8. Emission of Radiation We now consider the rate of emission of a photon.7: One-Photon Emission and Absorption 229 This geometrical average. for Nω = 0. However. The radiation field is described. The transition dipole moment can vanish for many transitions between stationary states of a quantum system. In case of a classical radiation field one would expect that emission cannot occur.194) For absorption processes involving the electronic degrees of freedom of atoms and molecules this radiation rate is typicaly of the order of 109 s−1 . This poses. 1. However.196) The last two factors in (8.. for atoms or symmetric molecules.195) = Nω 1. For practical evaluations we provide an expression which eliminates the physical constants and allows one to determine numerical values readily. 2 .189) to yield the total absorption rate kabs (tot) = Nω 4 e2 ω 3 | n|r|0 |2 3 c3 . (8.8. 8.37 × 1019 × 3 3 c3 3 ao λ3 s λ and kabs where λ = 2πc n − (8.. finally.. The most intensely absorbing molecules are long. by planar waves with vector potential (8. there is also a contribution to the emission rate which is proportional to Nω . This difference between emission and absorption is due to the quantum nature of the radiation field. However. .197) o (tot) (8. in agreement with observations.194) for the absorption rate shows that the essential property of a molecule which determines the absorption rate is the so-called transition dipole moment | n| r |0 |. Transition Dipole Moment The expression (8.e.e. the absorption rate is of the order of 109 s−1 or 1/nanosecond. The corresponding process is termed spontaneous emission. Nω = 0. however.173) contributes. that emission occurs even if no photon is present in the environment. This dependence predicts. linear molecules. as for the absorption process. For this purpose we use ω/c = 2π/λ and obtain 4 e2 ω 3 32π 3 e2 ao 1 ao = = 1. the resulting total rate of emission bears a different dependence on the number of photons present in the environment. . Nω photons before absorption. Accordingly. a quantum mechanical treatment of the radiation field leads to a total emission rate which is proportional to Nω + 1 where Nω is the number of photons before emission.

8. accordingly. For these numbers holds Nn / No = exp[−( n − o )/kB T ] (8. in lasers. respectively.g.230 Interaction of Radiation with Matter which is termed induced emission since it can be induced through radiation provided. respectively. 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 Two-Photon Processes In many important processes induced by interactions between radiation and matter two or more photons participate.199) Planck’s Radiation Law The postulate of the Nω + 1 dependence of the rate of emission as given in (8. Defining the rate of spontaneous emission 4 e2 ω 3 | n|r|0 |2 (8. e. = (8. Nω + 1 (8. The number of atoms undergoing absorption in unit time are Nω ksp No and undergoing emission are (Nω +1)ksp Nn .. Let No and Nn denote the number of atoms in state |0 and |n . In the following we discuss several examples.202) It follows.203) i. using (8.198) to determine the stationary distribution of photons ω in an oven of temperature T . The total rate of emission..200) where kB is the Boltzmann constant. Examples are radiative transitions in which two photons are absorbed or emitted or scattering of radiation by matter in which a photon is aborbed and another re-emitted.201) ksp = 3 c3 the rates of absorption and emission are Nω ksp and (Nω + 1)ksp .200). We assume n − o = ω.198) (8. it must hold Nω ksp N0 = (Nω + 1) ksp Nn (8.e.195) and (8. exp[− ω/kB T ] = This equation yields Nω = 1 . the well-known Planck radiation formula. Under stationary conditions the number of hydrogen atoms undergoing an absorption process |0 → |n must be the same as the number of atoms undergoing an emission process |n → |0 . To demonstrate this property we apply the transition rates (8. exp[ ω/kB T ] − 1 (8.204) Nω . . Hence.198) is consistent with Planck’s radiation law which reflects the (boson) quantum nature of the radiation field.

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. ˆ (8.106).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.181) and using (8.208) has been prepared. 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. under ordinary circumstances. The resulting rate of the transition depends on 2 Nω . In this case one needs to choose the energy of the photons to obey n = o + 2 ω. (8. even if n| r |0 vanishes.187).104) with stationary states |n defined through (8. finally. but then requires the absorption of two photons. The intense radiation fields of lasers allow one to increase transition rates to levels which can readily be observed in the laboratory. which is not necessarily the case. If the sum in (8.207) does not converge rapidly. faced with the dilemma of having to sum over all intermediate states |m of the system.172. dropping the index 1 characterizing the radiation. however. Employing the dipole approximation (8. We assume that a planar wave with wave vector k1 and polarization u1 . However.207) one is.206) − o − 2 ω) .207) Expression (8. 8. t) = Ao1 u1 cos(k1 · r − ω1 t) . however.8: Two-Photon Processes Two-Photon Absorption 231 The interaction of electrons with radiation.173). then expression (8. The electron absorbs the radiation and emits immediately a second photon. in case ˆ of absorption only V1 contributes. .8. In applying (8.168) where the first term describes the relevant contribution. 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. One obtains. (8. The transition requires that the transition dipole moment n| r |0 does not vanish for two states |0 and |n . (8.182) yields. a transition between the states |0 and |n may be possible. k = Nω V 2 8π 3 e4 × δ( m− ∞ m=0 o ( n − ˆ m) u · n|ˆ |m ( m − o ) u · m|ˆ |0 r ˆ r ω( m − o − ω − i λ) 2 − 2 ω) .207) does not provide a suitable avenue of computing the rates of 2-photon transitions. 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. induce single photon absorption processes as described by the transition rate Eq. as described through the vector potential ˆ A(r.205) The respective radiative transition is of 2nd order as described by the transition rate (8.

8.105) using the vector potantial (8.211) is o2 A− = o2 8π (N2 + 1) ω1 V (8.210) The vector potential which describes both incoming wave and outgoing wave is a superposition of the potentials in (8.209) cosθ2 where the value of k2 has been fixed. for the components of the wave (8.211) is stated in Eq. hence. In the present case we consider only scattering processes which absorb radiation corresponding to the vector potential (8.212) where N1 /V is the density of photons for the wave described by (8. according to the ˆ description of emission processes on page 229. (8. t) = Ao2 u2 cos(k2 · r − ω2 t) .105).208) and the emission of the radiation in (8. the wave characterized through k1 .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. u1 . 8. the amplitude A+ is given by o1 A+ = o1 8πN1 ω1 V (8.211) accounts for the emitted wave and. ˆ The perturbation which arises due to the vector potential (8. o1 ˆ o2 ˆ (8.6 above that the absorption of the radiation in (8.141). This expansion yields.208.208).210) is accounted for by the following contributions of (8. t) = A+ u1 exp[ i (k1 · r − ω1 t) ] + A− u2 exp[ i (k2 · r − ω2 t) ] .210) A(r.208.208) and emit radiation corresponding to the vector potential (8. i. ˆ (8. in the present case..214) .213) where N2 /V is the density of photons characterized through k2 . the amplititude A− defined in (8.211) The first term describes the absorption of a photon and. The relevant terms of the perturbation (8.e.210). We know already from our description in Section 8. u2 . The second term in (8. The vector potential describing the emitted plane wave is then A(r. later we will allow the quantum system to select appropriate values.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 . however.210). 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.

8.e. evaluating the time integrals in (8.181) in stating this result. 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 (8.216) over all available modes of the field. one needs to integrate the rate as given by (8. only these two terms contribute to the scattering amplitude.215) and taking the limits limto →−∞ and limλ→0+ yields the transition rate k − m = 2π δ( n − o − ω1 + ω2 ) e2 A+ A− u1 · u2 n|0 ˆ ˆ 4m2 o1 o2 e 2 (8.215) t dt ei( to n− o− ω1 + ω2 +i λ)t 1 i 2 e2 4m2 e A+ A− × o1 o2 × ˆ ˆ u1 · n| p |m u2 · m| p |0 × ˆ ˆ t t × to dt to dt ei( n− m− ω1 +i λ)t i( e m− o+ ω2 +i λ)t ˆ ˆ + u2 · n| p |m u1 · m| p |0 × ˆ ˆ t t × to dt to dt ei( n− m+ ω2 +i λ)t i( e m− o− ω1 +i λ)t We have adopted the dipole approximation (8. Inserting also the values (8.218) ..215) correspond to scattering processes in which the radiation field ‘looses’ a photon ω1 and ‘gains’ a photon ω2 . Accordingly. Following closely the procedures adopted in evaluating the rates of 1st order and 2nd order radiative transitions on page 222–225.217) 2 Here ro denotes the classical electron radius ro = e2 = 2. one needs to carry out the integration 2 V(2π)−3 k2 dk2 · · ·.e. me c2 (8. Only the second (1st order) and the third (2nd order) terms in (8. Hence.212. i.8: Two-Photon Processes function accounting for absorption and re-emission of a photon n|ΨD (t) = + + m=0 233 n|0 + 1 e2 A+ A− u1 · u2 n|0 ˆ ˆ i 4me o1 o2 ∞ (8.. 8.216) e2 + − A A 4me o1 o2 n|ˆ1 · p |m m|ˆ2 · p |0 u ˆ u ˆ n|ˆ2 · p |m m|ˆ1 · p |0 u ˆ u ˆ + m − o + ω2 m − o − ω1 We now note that the quantum system has the freedom to interact with any component of the radiation field to produce the emitted photon ω2 .8 · 10−15 m .

. consequently. one can relate (8. it follows 2 dσ = ro (N2 + 1) dΩ2 |ˆ1 · u2 − S( ω) |2 u ˆ (8.224) = m 1 − o ( m ω ( ω)2 + + . Accordingly. i. ω1 = ω2 .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 (8.225) S(0) = m u ˆ 0|ˆ1 · p |m m|ˆ2 · p |0 u ˆ u ˆ 0|ˆ2 · p |m m|ˆ1 · p |0 u ˆ + me ( m − o ) me ( m − o ) (8.228) S (0) = m S (0) = 2 m . in fact. 2 . .234 Interaction of Radiation with Matter The factor N1 c/V can be interpreted as the flux of incoming photons.221) suggests to expand S( ω) S( ω) = S(0) + S (0) ω + Using 1 − o ± ω one can readily determine m (8. (8.219) (8. a process in which the electronic state remains unaltered after the scattering. one assumes the even stronger condition ω1 << | o − m| .. .227) u ˆ 0|ˆ1 · p |m m|ˆ2 · p |0 u ˆ u ˆ 0|ˆ2 · p |m m|ˆ1 · p |0 u ˆ + me ( m − o )3 me ( m − o )3 (8.e. 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.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 + . Rayleigh Scattering We turn first to an example of so-called elastic scattering. − o )2 ( m − o )3 (8.220) n|ˆ2 · p |m m|ˆ1 · p |0 u ˆ u ˆ + m − o − ω1 2 − − o + ω2 In the following we want to consider various applications of this formula.223) m Condition (8. for all states |m of the electronic system (8.217) to the scattering cross section defined through dσ = It holds then dσ = 2 ro rate of photons arriving in the the solid angle element dΩ2 flux of incoming photons ω2 (N2 + 1) dΩ2 u1 · u2 n|0 ˆ ˆ ω1 1 n|ˆ1 · p |m m|ˆ2 · p |0 u ˆ u ˆ me m m (8.221) Using |n = |0 and..

222).227) vanishes.224. honey bees.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. ˆ ˆ S (0) given in (8. u1 ·r ] |0 = 0 u ˆ (8. this term cancels the u1 · u2 term in (8. ˆ We want to simplify first (8. ˆ ˆ We want to prove now that expression (8.235) and obtain the leading contribution to the expression (8.232) Obviously. . It explains.228) provides then the first non-vanishing contribution to the scattering cross section (8. For this purpose we replace p using (8. 8.k=1 (ˆ1 )j (ˆ2 )k δjk = u1 · u2 u u ˆ ˆ (8.108) for the identity operator.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 (8.233) m Employing again (8.235) We can now combine eqs.e.8.226) into S(0) = 1 i ( 0|ˆ1 ·r |m m|ˆ2 · p |0 − 0|ˆ2 · p |m m|ˆ1 ·r |0 ) u u ˆ u ˆ u m .108) yields S (0) = me 2 0| [ˆ2 ·r. the blue color of the sky and the polarization pattern in the sky which serves many animals. u ˆ ˆ ˆ ˆˆ i The commutator property [xj . Using again (8. 8.234. (8.234) where we used for the second identity the fact that u1 ·r and u2 ·r commute. i.k=1 j.231) (ˆ1 )j (ˆ2 )k 0|[xj . pk ]|0 = u u ˆ j.229) both for the u1 · p and the to u2 · p terms in (8. m|ˆ1 · p |0 u ˆ 1 = − m|ˆ1 ·r |0 u me ( m − o ) i (8.230) According to (8. as a compass.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 ) .236) is of great practical importance.8: Two-Photon Processes 235 ˆ These three expressions can be simplified using the expression (8. for example..182) for p and the expression (8. 8. Expression (8.232.222).229) (8.236) 2 × m 0|ˆ∗ ·r |m m|ˆ2 ·r |0 u1 u 0|ˆ∗ ·r |m m|ˆ1 ·r |0 u2 u + m − o m − o We have applied here a modification which arises in case of complex polarization vectors u which ˆ describe circular and elliptical polarizaed light. For this purpose we apply (8. u u u u (8.226). pk ] = i δjk yields finally ˆ S(0) = 1 S(0) = i 3 3 (8.

(8.241) 0 0 Similarly.244) . that the dipole approximation is still valid which restricts the applicability of the following derivation to k1 >> One obtains immediately from (8. u1 =  1  ˆ ˆ (8. (8. u2 = u2 ˆ ˆ ˆ ˆ cos2 φ2 cos2 θ2 sin2 φ2 for u1 = u1 .242) |k2 × k1 | 0 where the second identity follows readily from k1 = e3 and from (8. We want to assume. i. The first choice is ˆ   sinφ2 k2 × k1 (1) u2 = ˆ =  −cosφ2  (8. Since u2 ˆ ˆ (1) orthogonal to k2 as well as to u2 the sole choice is ˆ   cosθ2 cosφ2 (1) k2 × u2 ˆ (2) u2 = ˆ =  cosθ2 sinφ2  (1) |k2 × u2 | ˆ −sinθ2 The resulting scattering cross sections         2 dσ = ro (N2 + 1) dΩ2 ×        for the various choices of polarizations are sin2 φ2 for u1 = u1 . 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 . however. ao Bohr radius . u2 = u2 ˆ ˆ ˆ ˆ (8.243) cos2 θ2 cos2 φ2 for u1 = u1 . we choose for the polarization of the emitted radiation two perpendicular directions u2 ˆ (2) and u2 which are also orthogonal to the direction of k2 .237) The resulting scattering is called Thomson scattering. the emitted u ˆ radiation is decribed by the wave vector   sinθ2 cosφ2 k2 = k1  sinθ2 sinφ2  (8.e. for all states |m of the electronic system . hence.220). |n = |0 and ω1 = ω2 = ω in (8. assume now that the scattered radiation has very short wave length such that ω >> | o − m| .240). though..239) We will show below that this expression decribes the non-relativistic limit of Compton scattering.238) (8. u2 = u2 ˆ ˆ ˆ ˆ for u1 = u1 . u2 = u2 ˆ ˆ ˆ ˆ (1) (2) (2) (1) (1) (2) (1) (2) (2) (1) needs to be (8. u ˆ 1 ao .240) cosθ2 We choose for the polarization of the incoming radiation the directions along the x1 .and the x2 -axes     1 0 (1) (2) u1 =  0  ..236 Thomson Scattering Interaction of Radiation with Matter We consider again elastic scattering.

For electronic systems in molecules or crystals the degrees of freedom excited are nuclear motions. i.246). The excitation can be electronic.249) dσ = ro (N2 + 1) ω1 −1 −1 ω2 − ω1 = . defined in (8.)n . In case that the nuclear degrees of freedom excited absorb very little energy. as in the case of excitations of accustical modes of crystals. molecular vibrations or crystal vibrational modes.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 .. The energy deficit is used to excite the system.247) ( 1 − cosθ2 ) (8. the scattered radiation is polarized to some degree.)n . (8. the resulting scattering cross section is 2 ro (N2 + 1) ( 1 + cos2 θ2 ) dΩ2 . for c → ∞ the Compton scattering cross section (8. Raman Scattering and Brillouin Scattering We now consider ineleastic scattering described by the Kramers-Heisenberg formula. ˆ In case that one measures the scattered radiation irrespective of its polarization. The radiation scattered at right angles is even completely polarized (1) along the u2 -direction. the states |n etc.8.220) vanishes and one obtains in case of Raman scattering ω2 2 dΩ2 | u2 · R · u1 |2 ˆ ˆ (8.e. the scattering is termed Brillouin scattering. the first term in (8. If energy is absorbed by the system.. e. in case of a diatomic molecule |n = |φ(elect.220) represent actually electronic as well as nuclear motions. or in case of translational motion of molecules in liquids.247. φ(vibr. 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. but most often involves other degrees of freedom. if energy is released.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 of such scattering an electron system absorbs and re-emits radiation without ending up in the initial state. Such scattering is called Raman scattering.. 8.246) 2 This expression is the non-relativistic limit of the cross section of Compton scattering. i.247) becomes identical with the Thomson scattering cross section (8. 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.e. In the case that the scattering excites other than electronic degrees of freedom. one speaks of Stokes scattering.248) me c2 One can readily show that in the non-relativistic limit. one speaks of anti-Stokes scattering.g. Since the scattering is inelastic.

..251) ω1 − ( We define x · R · y = j. a case called resonant 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 )/ (8.250) (8. ω1 ≈ m − o for a particular state |m . Of course. 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. a property which can be exploited to selectively probe suitable molecules of low concentration in bulk matter. In case that the incoming photon energy ω1 is chosen to match one of the electronic excitations. Nevertheless. e.g. the Raman scattering cross section will be much enhanced.k xj Rjk yk .

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

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

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. In the latter case all even permutations are represented by ‘1’ and all odd permutations by ‘-1’. particles with integer spin values. Since permutations do not necessarily commute.. a diagonal representation can only berealized in a simpler group. the second group to fermions. hence. provide a representation of the group structure represented by the multiplication table of even and odd permutations given above.e. i. S where S denotes the number of available single particle states. it is essential that the functions F ( ) and G( . Obviously. all particles being governed by the same interactions. P ] = 0 .6) which describes one–particle interactions and two–particle interactions of the system. or the corresponding group of two elements 1.7) This property. −1. i.. for the scalar product holds j|k = δj k . G(xj . the representation is either isomorphic to the group composed of only the ‘1’. In fact. . Presently. The terms of the Hamiltonian will be discussed further below.11) .. the Hamiltonian for permuted indices P HP −1 is identical to H.k=1 (9. 2. In fact.10) We will construct now wave functions which exhibit the appropriate properties. The first case applies to bosons. in turn.9) = 1 for P even −1 for P odd (9. i.8) |Ψ (9.-1}. k = 1.e. xk ) j. both groups. usually a very large number. From this property follows that H is independent of any specific assignment of indices to the particles (note that (9. (9. In fact. it is a postulate of quantum mechanics that for any description of many–particle systems the permutation operators must be diagonal. i. xk ) is symmetric in its two arguments. The boson and fermion property stated implies that for a system of N bosons holds P |Ψ = |Ψ For fermions holds P |Ψ = where P P ∀P ∈ SN . particles with half–integer spin values. implies that the permutation operators can be chosen in a diagonal representation. 1 together with multiplication. {1} (trivial) and {1. (9. .e. ) are the same for all particles and for all pairs of particles.5) implies that a permutation effectively changes the indices of the particles) and. 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 . Fock-Space Our derivation will start by placing each particle in a set of S orthonormal single particle states x|k = φk (x).e. ∀P ∈ SN . i. From the latter results [H.

x2 . xN |ΨB (n1 .15) with λk = j. . . . . . λ2 . The integers nj are equal to the number of λk in (9.242 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. therefore. .14) Obviously. x2 . xN |ΨFock (λ1 . λN ) ⊂ SN is the set of all permutations involving only particles j with different λj . characterize the wave function (9. . . the states |ΨFock (λ1 .11. .λN ) Here the states (λ1 . . . . . 9. . . It holds N ΨFock (λ1 . i.8.λN ) P x1 . the reason is that such transposition duplicates the state which would. .15). . . . A wave function which obeys the boson symmetry (9. . S} of the type (9.8) is x1 . . λ2 .. . .e. . . λ2 . . λN ) = j=1 φλj (xj ) . We first consider orthonormal many–particle wave functions. . λN ) only if λj = λk . . hence. . . .. .g. . is a particular choice of placing N particles consistent with the occupation numbers (n1 . x2 . Λ(λ1 . 2. n2 .12) where the second factor represents the scalar product between spin states. . T (j. .12). with λj = λk = . λj ∈ {1. N x1 .λN ) on the rhs. 9. . . λ2 . .. xN |Ψ Fock (λ1 . i. (9. each particle j into a specific state |λj .15) completely and. This detail is connected with the fact that a two–particle state of the type ΨFock (λ. 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. . . . |λ2 .. nS ). .9). An important detail of the definition (9. . e. . . the so-called Fock-states. λ2 ..15) is that the sum over permutations does not involve permutations among indices j. . The numbers nj . λN ) do not exhibit the symmetries dictated by (9. . |λN . are essential. 9. . λ) in which two particles occupy the same single– particle state |λ does not allow transposition of particles. λN ) = j=1 δλj λj . referred to as the occupation numbers. k. . . . λN )|ΨFock (λ1 .13) These states form an orthonormal basis of N -particle states. n2 .15) (nj !) N! 1 2 P ∈Λ(λ1 . . appear severalfold in (9. (9. . .8. The Fock states represent N particles which are placed into S states |λ1 .e. .9). which do not yet obey the symmetries (9. . In a second step we form linear combinations of Fock states with the desired symmetries. . . the nj specify how often a single particle state |j is occupied. . k) ∈ Λ(λ1 . The reader is well advised to pause and grasp the definition of the nj .. . . . λ2 . nS ) = S j=1 (9.

. . λ2 .8). the integers nj denote the occupancy of the single–particle state |j .. electrons in an atom.. for example. . n2 . .16).. . . n2 . . . . ..λN ) on the rhs. nS ) holds S Ψ (N )|Ψ (N ) = j=1 B B δnj nj . λ2 . . . .. . nS ) = φλ1 (x1 ) φλ1 (x2 ) . √1 N! (9. . .e. . . . nS ). . for two sets of occupation numbers N = (n1 . nS ) and N = (n1 . . (b) Show that for the states (9. φλN (xN ) Here. . . n2 .9). .19) φλ1 (xN ) φλ2 (xN ) . nS ) and N = (n1 . x2 .15) describe.17) Here the states (λ1 . . . . x2 . φλN (x1 ) φλN (x2 ) . (9.20) .1. .. . The fermion states (9. nS ) = √1 N! P ∈SN P P x1 . . Such wave function is x1 . n2 . . . for example. correspond to a particular choice of placing N particles consistent with the occupation numbers (n1 .15) holds (9. . . These states form an orthonormal basis. n2 .e. .9. xN |ΨFock (λ1 .15) obey the boson symmetry (9. n2 . . . . . . i. . . .18) this wave function can also be expressed through the so-called Slater determinant x1 . x2 . . .16) Exercise 9. These states form an orthonormal basis. the photons of the electromagnetic field or the phonons in a crystal. xN |ΨF (n1 . Wave Functions with Fermion Symmetry in the Occupation Number Representation One can construct similarly a wave function which obeys the fermion symmetry property (9. xN |ΨF (n1 . .1: Bosons and Fermions 243 The wave functions (9. For any value nj > 1 two or more of the columns of the Slater matrix are identical and the wave function vanishes. . . . (9. . nS ) holds S ΨF (N )|ΨF (N ) = j=1 δnj nj . . i. . . Using the identity governing the determinant of N × N –matrices A N det(A) = P ∈SN P j=1 Aj P (j) (9.λN ) (9.1: (a) Show that the states (9. φλ2 (x1 ) φλ2 (x2 ) . for two sets of occupation numbers N = (n1 . The important property holds that nj can only assume the two values nj = 0 or nj = 1. . n2 . .17) describe. . . a molecule or a crystal. .

. n2 . . (9. . n ) are normalΨ 1 j 1 j S S ized according to (9. . .15). nS ) . . [a† . a† ] = δj k . guaranteeing the correct permutation properties of the many–particle states. .9). . . . . .22) follows from (9. .17) obey the orthonormality property (9. Employing the superindices N and N introduced above and using the orthonormality property (9. nS ) = j nj + 1 ΨB (n1 . . . ak ] = 0 . are the so-called operators of 2nd quantization. . . . .17) is commonly referred to as the occupation number representation since the wave functions are uniquely specified by the set of occupation numbers N = (n1 . nS ). . n . .1. . (9. . . . . . (9. nj − 1.21) we consider the matrix Aj corresponding to aj in the basis of many–particle states (9. .15) and (9.25) . nj . . n − 1. . nS ) nj ≥ 1 0 nj = 0 √ . since both the N -particle wave function B (n . 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 . Exercise 9. (b) Show that the states (9. nj . . .23) (9. 9. . . To prove that (9. . For the adjoint operator a† holds j a† ΨB (n1 . . These operators allow one to construct and manipulate many–particle wave functions while preserving permutation symmetry. . . . . n ) and the N − 1-particle wave function ΨB (n . . .2 Operators of 2nd Quantization An important tool to describe mathematically systems of many bosons or fermions.24) [aj . .2: (a) Show that the states (9.16) one obtains (Aj )N = ΨB (N )| aj |ΨB (N ) = √ S N nj δnj nj −1 k=1 k=j δnk nk .20). a† ] = 0 j k (9. . . nS ) = nj ΨB (n1 .244 Many–Particle Systems The representation of wave functions (9.17) obey the fermion symmetry (9. . . . nj + 1. .21) √ The factor nj appears here on the rhs. . Creation and Annihilation Operators for Bosons We consider the operator aj defined through aj ΨB (n1 . .15). .22) The operators aj and a† obey the commutation properties j [aj .

The related operator S ˆ N = j=1 ˆ Nj (9. . n2 .29) |ΨB (n1 . .24). n2 . .. a† ] = 1 For j = k follows similarly 1.27) Equations (9. n2 . nS ) = j=1 nj |ΨB (n1 . . . . . We will prove the commutation properties (9. One refers to Nj as the occupation number operator. obviously. (9.e. . . the properties (9. nj − 1 . for basis states |ΨB (n1 . . .33) is called the particle number operator since. The creation operators a† allow one to construct boson states ΨB (N ) from the vacuum state j |0 = |ΨB (n1 = 0. One can readily show using (9. .34) = .32) ˆ ˆ i. n2 .27. . . (9. nS ) = 0 . j (9. (9. nS = 0) as follows S (9.24). .2: Operators of 2nd Quantization Let A† be the matrix adjoint to Aj . nS ) = j=1 a† j nj nj ! |0 . S ˆ N |ΨB (n1 . nS ) N |ΨB (n1 . 9. i. .. (9. It holds for j = k (aj a† − a† aj )|Ψ(N ) = (nj + 1 − nj )|Ψ(N ) = |Ψ(N ) .30) Of particular interest is the operator ˆ Nj = a† aj . 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 .9. . nk + 1 . n2 = 0.22) follows from (9.e. . . . .21) and vice versa. j (aj a† − a† aj )|Ψ(N ) = k k ( nj (nk + 1) − (nk + 1)nj )|Ψ(n1 . . nS ) = |ΨB (N ) .22) ˆ Nj |ΨB (N ) = nj |ΨB (N ) . . 9.21. (9. .23) follow in an analogous way.26) From this result one can immediately conclude that (9.31) This operator is diagonal in the occupation number representation. nS ) . n2 . . . . j j Since this holds for any |Ψ(N ) it follows [aj . . .28) (9. . the eigenvalues of Nj are the occupation numbers nj .29) yield the commutation relationship (9. .

. . . .1: The commutation relationships (9. nj+1 = 1.. . 9. from the vacuum state |0 or states |ΨB (N ) . nj+1 .35) and where the derivative operation is defined through ∂f (A) = ∂A ∞ n=1 1 f (n) (0) An−1 .e.g. Obviously. . Before proceeding in this respect we need to account for the following property of the fermion wave function which applies in the case nj . it also holds for nj + 1.. nS ) = . .39) .246 Many–Particle Systems In using the boson creation and annihilation operators. . one needs to apply only the commutation propertes (9. since they are induced through the algebra of a† and aj . . . .2: Show that the boson operators a† and aj satisfy j [aj . . φj (x1 ) φj (x2 ) . . x2 . x1 . . x2 . e. ..23.24) and aj |0 = 0 imply that the the properties (9. . xN |ΨF (n1 . f (a† )] j [a† .. . Prove this by induction.17. . .23. φj+1 (x1 ) φj+1 (x2 ) .38) corresponds to x1 . .2. . .21. . .h. . . in case it holds for nj .38) . x2 . . showing that the property holds for nj = 0 and. As long as one starts from a wave function with the proper boson symmetry. . .36) Creation and Annihilation Operators for Fermions We want to consider now creation and annihilation operators for fermions. i. 9. 9.30). . the fermion wave function changes sign when one exchanges the order of the occupancy. .. .. . √1 N! . nS ) . . j Exercise 9. operators which can alter the occupancy of the wave function (9. one does not need to worry ever about proper symmetries of wave functions. . nj .. nj+1 . nj (9.19) without affecting the fermion symmetry (9. φj (xN ) φj+1 (xN ) . 9. .9).. To prove this property we notice that the l. xN |ΨF (n 1 . . 9. xN |ΨF (n1 .24) and the property aj |0 = 0. .e. nj+1 . . . . nS ) = − x1 . in general. . . (n − 1)! (9. of (9. nj . 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. i.23) hold for the state defined through (9. Exercise 9. (9. 9. in case that the single particle states |j and |j + 1 are both occupied. .22.2..s..29.37) (9.

. n5 = 1. . . . Because of the property of the determinant to change sign when two columns are interchanged. .. .. This requires one. qj is the number of states |k with k < j which are occupied in a fermion wave function. xN |ΨF (n1 . . . n2 = 1. . nj+1 . Obviously. . n3 . .38) reads − x1 . .. nj = 0. . x2 |ΨF (n1 = 0.h. For this purpose one must adhere to a strict convention: the labelling of single–particle states by indices j = 1.. . i. ..40) are identical. n3 = 0. .. i. . (9.43) . . This change of position is connected with a possible sign change (−1)qj where qj is defined for a given N = (n1 . n2 . . the expressions (9. nS ) as follows: j−1 qj = k=1 nk . (9. x2 . . n2 . 247 − √1 ! N . . n2 . φj (x1 ) φj (x2 ) . .. in order to be consistent with this convention. .9. .s.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.e. at the first column of the Slater determinant (9.2: Operators of 2nd Quantization The r.. . 2. that occupancies are always ordered according to a monotonous increase of the single–particle state index. φj+1 (xN ) φj (xN ) . nS ) = .19). We are now ready to define the annihilation operator cj for fermions in the single–particle state |j as follows cj |ΨF (n1 . n2 . nj → 0. the rhs. nj+1 . . vanishes. nS ) = nj (−1)qj |ΨF ( n1 . . must be chosen once and for all at the beginning of a calculation and these states must be occupied always in the order of increasing indices. nS ).e. φ2 (x2 ) φ5 (x2 ) (9.40) . however. 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).e. . One adopts the convention that particles are created and annihilated by these operators always at the first position of the wave function.42) i. . We will be adopting below the notational convention that qj corresponds to occupancies N = (n1 . i. nj .. . A proper example is the two particle fermion wave function x1 . φj+1 (x1 ) φj+1 (x2 ) . . . of (9.e. . nS = 0) φ2 (x1 ) φ5 (x1 ) 1 1 = √2 = √2 [φ2 (x1 )φ5 (x2 ) − φ5 (x1 )φ2 (x2 )] . . . . at which column of the Slater determinant (9. nj . nS ) qj states occupied In case that the single particle state |j is not occupied. . . nj−1 ...19). . .39) and (9. . it is necessary to specify for a fermion wave function the order in which the single– particle states |λj are occupied. . . . n4 = 0.. n6 = n7 = . . particles are being created or annihilated. (9. .

(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 . .45) follow in an analogous way and will not be derived here.46). . otherwise. nS ) qj states occupied The operators thus defined obey the commutation properties [cj .19). (9. n3 .44) (9. Similarly one obtains cj c† |ΨF (N ) = cj (−1)qj (1 − nj ) |ΨF (n1 .46) first for the case j = k. nj+1 .43). 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 .48) From this follows (9. 9. . B]+ = AB + BA.44). n2 . . . nS ) = j (−1)qj (nj + 1))(−1)qj (1 − nj ) |ΨF (N ) = (1 − nj ) |ΨF (N ) (9. .17.49) The derivation involves realization of the fact that a non-zero result is obtained only in case nj = 1. . We will prove now the anti-commutation property (9. It holds c† cj |ΨF (N ) = c† (−1)qj nj |ΨF (n1 . .248 The operator c† adjoint to cj exhibits the following property j c† |ΨF (n1 . the last factor on the rhs. . nj−1 . We first show that (9.48) vanishes.44) follows from the definition (9. . n2 . c† ]+ = 0 j k (9. nj → 0 . . nj . The anti-commutation properties (9. We consider (9.50) .47) Let C† be the matrix adjoint to Cj . . nj → 1.46) [cj . . . nj → 1 . c† ]+ = δj k k where we have used the definition of the so-called anti-commutators [A. ck ]+ = 0 . nS ) = j j (−1)qj (1 − nj + 1)(−1)qj nj |ΨF (N ) = nj ΨF (N ) (9.45) (9. . Let Cj be the matrix corresponding to the operator cj in the basis of fermion states (9. We have used here the definition qj = k<j nk along with qj = qj . [c† . nS ) = j Many–Particle Systems (−1)qj (1 − nj ) |ΨF ( n1 . .

.) = c† 1 c† 2 · · · c† N |0 .. We j will often separate the states |j . . . coordinates x and index j into a space part and a spin part. an operator cj must be on the left of an operator ck for j < k. . Correspondingly. j → j.54) ˆ N = j=1 ˆ Nj (9.56) Here we have made use of cj cj = 0 and c† c† = 0 which follows from nj ∈ {0. . The stated property ˆ follows from the idempotence of Nj which can be derived as follows † ˆ2 ˆ Nj = c† cj c† cj = c† (1 − c† cj )cj = c† cj − c† c† cj cj = cj cj = Nj . Since (9.52) One can obtain similarly the same relationship in the case j > k.49) and (9. nj = 0 . . i.2: Operators of 2nd Quantization (9.e. ˆ It is an interesting exercise to demonstrate that Nj only has eigenvalues 0 or 1. . with eigenvalues equal to the occupation numbers nj ˆ Nj |ΨF (N ) = nj |ΨF (N ) ˆ Nj .h. σ 1 (σ = ± 2 ) .51) (9. .h.51.e.55) is called the particle number operator. nk = 1 . . ( cj c† + c† cj ) |ΨF (N ) = 0 .50) yield ( cj c† + c† cj ) |ΨF (N ) = |ΨF (N ) .. j j j j j j j (9.9. nλN . . . k k 249 (9. in the basis |ΨF (N ) . hence. i. . nS ) k and.49) shows that the operator ˆ Nj = c† cj j (9. 1}.57) On the l. .46). of this equation we meant to indicate only those N occupation numbers nλj which are non-vanishing. nj = 0 .e. is referred to as the occupation number operator. nk = 1 . S (9. .53) is diagonal in the occupation number representation. 1 In the following we will apply fermion operators c† and cj only to electrons which carry spin 2 . .58) .52) hold for any |ΨF (N ) one can conclude (9.29)) |ΨF (. . .46). j j For j < k one obtains cj c† |ΨF (N ) = (−1)qj +qk nj (1 − nk )|ΨF (n1 . λ λ λ (9. for which purpose one needs to envoke only the algebraic (anti-commutation) properties (9. nλ2 . the creation operators must operate in the proper order. . . 2 (9. x|j → φj (r) | 1 σ . Equation (9. nS ) k c† cj |ΨF (N ) = (−1)qj +qk −1 nj (1 − nk )|ΨF (n1 .s. On the r. 9. . hence. 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. i. nλ1 ..s.

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

are spin-independent as is evident from the factor δσσ . r|ˆ|u. We will show that the boson and fermion creation and annihilation operators serve this purpose. (9. 9. σu ) .66) Here we adopt the notation for the Laplacian ∂2 ∂2 ∂2 ∂2 = + + .70) Both operators.66) and (9. u = g d3 r ∗ ∗ d3 r ψr (r.71) . ˆ 251 (9. g g (9. u = s. 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. σr )ψs (r .69) ˆ In this case the generic operator is f (x) = δ(r − r).63.64) r.69).62) in terms of the matrix elements of single–particle states (9. ∂rj2 (9. t .9. An example for a two–particle operator is the Coulomb repulsion operator 1 ˆ V = 2 N j.65) This symmetry will be exploited repeatedly below. The matrix elements of this operator in a single–particle basis are ˆ rσ|t|sσ = δσσ d3 r φ∗ (r) r 2 − 2m ∂2 φs (r) . s|ˆ|t. one cannot switch the indices of one pair individually.64). ∂r 2 ∂x2 ∂x2 ∂x2 1 2 3 (9. x ) ψt (r. Notice. according to (9. Examples of One– and Two–Particle Operators An example for a single–particle operator is the kinetic energy operator N ˆ T = j=1 2 − 2m ∂2 .3: One– and Two–Particle Operators r.67) ˆ In this case the generic single–particle operator. u which follows directly g from the definition (9. is t(x) = (− 2 /2m)(∂ 2 /∂r 2 ).k=1 j=k q2 |rj − rk | (9. σt )ψu (r . ˜ (9.68) Another single–particle operator is the one–particle density operator N ρ(r ) = ˆ˜ j=1 δ(r − rj ) . ∂r 2 (9. i. s|ˆ|t.. s and t. that the symmetry implies that one can switch simultaneously the pairs of indices r.59).64) We like to note an important symmetry of the matrix element 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. u. σs ) g (x.e. s|ˆ|t.

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

To show that the resulting values of the matrix elements (9.64).e. The Matrix Elements ΨF (N )|c† cs |ΨF (N ) r We consider first the case r = s.171 and pp.62) are correct one needs to compare the result derived with conventional derivations of the matrix elements2 ..79) in the basis of many–particle states |ΨB.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’. r ΨB (N )|a† a† at au |ΨB (N ) r s ΨF (N )|c† c† ct cu |ΨF (N ) r s bosons fermions (9.F (N ) have the same values as the respective matrix elements (9. namely cu ct . u cr cs cu ct g fermions (9.78. g and not ct cu . xk ) → ˆ j. 9.80) (9.t.79) is opposite to that in the matrix element r.63.24) and (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 (9.83) t=1 t=r. s|ˆ|t.81) For the latter matrix elements simple rules can be derived from the algebraic properties of the boson and fermion operators (9.s=1 S r.79) have the following implication: The operators (9. u a† a† at au bosons g r s † † r. s|ˆ|t. pp.60) as follows N ˆ f (xj ) → j=1 N 1 2 1 2 S r.u=1 S r. The matrix elements are then actually those of the number operator c† cr which is diagonal in the occupation number representation.9.23.62).62) through single–particle state matrix ˆ ˆ elements (9. s|ˆ|t.82) ΨF (N )|c† cs |ΨF (N ) = r cr 2 (9. . Expressions (9.45.s=1 ˆ r|f |s a† as bosons r ˆ r|f |s c† cs fermions r r.46).79) It is crucial to notice in the expression for the fermion two–particle operator that the order of the annihilation operators in (9.t.78) 1 2 g (xj .s. s|ˆ|t. 9.59) and G in (9.s. 9.64) one replaces the operators F in (9. These rules will be provided below only for the case of fermions. and then in g a second step the matrix elements ΨB (N )|a† as |ΨB (N ) .k=1 j=k S r. respectively. i. u . respectively. r S Ψ (N In case of r = s one obtains F )|c† cr |ΨF (N ) r = nr r=1 δnr nr . 9.78) and (9. r ΨF (N )|c† cs |ΨF (N ) .176 where the matrix elements of fermion states can be found.63) and (9. u as defined in (9.u=1 (9. (9. In order to determine these matrix elements one needs to evaluate first the matrix elements ˆ of the generic operators r|f |s r.

86) into three contributions. (9.84) The off-diagonal elements with nr = nr except for r = s.s=1 a. (9. 171) shows that the operator defined in (9.s=1 a. s. u) are included in the summation for which all indices are different.86) For the combination of indices r. s. G = G0 + G1 + G2 .88) . in which case ns = ns + 1 and nt = nt − 1 holds.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.d.87) Here and in the following we denote by ‘a.d. t). u in this sum three possibilities exist Case 0 two of the four indices are different.45) yields together with the definition of the occupation number operator (9. s).’ (all indices different) that only tuples (r.u r=s..53) 1 ˆ G0 = 2 S r s|ˆ|r s − r s|ˆ|s r g g r.e.s=1 a.i.59) when the matrix elements are determined between Slater determinant wave functions.78) does yield the same results as the operator defined in (9. Case 2 all four indices are different. ˆ ˆ ˆ ˆ Accordingly.i. s. Comparision with the results in Condon&Shortley (pp. i. 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.85) All other matrix elements vanish. ˆ The first contribution G0 is 1 ˆ G0 = 2 S r s|ˆ|r s c† c† cs cr + g r s r. t=u 2 = 0.t. i. r = t etc.s. where each contribution corresponds to one of the three cases mentioned.d. t. t.e. Case 1 three of the four indices are different. Starting from the definition (9. ˆ ˆ Nr Ns .79) and using c† r we can write S 1 ˆ G = rs|ˆ|tu c† c† cu ct g r s 2 r. are ˆ ˆ ΨF (N )|F |ΨF (N ) = (1 − ns )(−1)qs nt (−1)qt s|f |t (9. those for which N and N differ in more than two occupation numbers. g r s r. t. (r.254 From (9. s. The anti-commutation property (9.d.82. we split the sum in (9.i. 1 2 S r s|ˆ|s r c† c† cr cs . (r. u need to be considered. 9.i. for which holds r = s. c2 = 0 r (9. namely. t..

d.t. S r. i.s.9. Only G0 . (9. As we will find.t. given in (9. r s|ˆ|r t c† c† ct cr g r s r s|ˆ|t s c† c† cs ct g r s r s|ˆ|s t c† c† ct cs g r s r s|ˆ|t r c† c† cr ct . One obtains 1 ˆ ΨF (N )|G|ΨF (N ) = 2 S nr ns r.t. s − r.s. (9. ns . relabelling ˆ summation indices and using c† c† ct cr = c† c† cr ct = c† ct Nr yields r s s r s S ˆ G1 = r. s|ˆ|r.89) S r. r g g .t=1 a.91) ˆ G2 = r. contributes in this case.t=1 a.86).i.i.t>u S r.u r>s. 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|ˆ|s.t=1 a.i. r s|ˆ|r t − r s|ˆ|t r g g ˆ c† ct Nr .t=1 a. F (N )|G|ΨF (N ) .d.93) .s.s.u r<s.t=1 a.u r>s. N = N or nr = nr for all r.t>u S r.i.84) ˆ of the one–particle operator F .d. s (9.s.i. this contribution obviously behaves similarly to the off-diagonal part (9. nt . S r. nu .t<u S r s|ˆ|t u c† c† cu ct g r s r s|ˆ|t u c† c† cu ct g r s r s|ˆ|t u c† c† cu ct g r s r s|ˆ|t u c† c† cu ct g r s (9.t. ˆ Case 0 This case covers diagonal matrix elements.t. r s (9.s r.s.e.45). ˆ The second contribution G1 is ˆ G1 = + + + 1 2 1 2 1 2 1 2 S r.65). employing the symmetry property (9.92) ˆ This contribution to G has non-vanishing matrix elements only when N differs from N in four ˆ occupation numbers nr .t>u r s|ˆ|t u − r s|ˆ|u t g g c† c† cu ct .d. A similar contribution does not arise for F .s.s. Commutation of ct cu according to (9.t<u S r. this contribution to the two–particle operator is diagonal in the occupation number ˆ representation.u r>s.d.s.90) ˆ Since Nr is diagonal. this part accounts for all diagonal contributions to G.88). 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. g r s (9.3: One– and Two–Particle Operators 255 Obviously.s..u r<s.

256 Many–Particle Systems Case 1 This case covers off-diagonal elements with nr = nr except for r = s. s|ˆ|v. u < v or s > r.62) for bosons.2: 9.89).74) and (9. r=1 r=1 Exercise Exercise Exercise Exercise Exercise 9.. Furthermore. All matrix elements which are not covered by the three cases above vanish. ˆ ˆ Show that the matrix elements of F and of G conserve particle number. t for which holds ˆ ns = ns + 1 and nt = nt − 1.3. u. it holds S nr = S nr . nv = nv − 1. contributes in this case.3: 9. Derive (9. s|ˆ|u.1: 9. i.92). Derive the non-vanishing matrix elements (9.96) . nt = nt + 1. Spin Operators ˆ ˆ In this section we will briefly consider the spin operators S given in (9. Only G2 . for ˆ non-vanishing matrix elements ΨF (N )|G|ΨF (N ) at most four occupation numbers nr and nr can differ. We like to express these operators through ˆ fermion creation and annihilation operators. u g g (9. u < v. the ‘−’-sign applies for s < r. given in (9.4: 9.92) from (9.3. In particular. contributes in this case. respectively. s|ˆ|t.92). s|ˆ|r.3.5: Derive (9.76).e.3. u > v.95) from (9.90). u > v or s > r. r g g (9. one–particle and two–particle operators. Only G1 .3.t nr r.91).75) which are. v for which holds ˆ ns = ns + 1.94) Case 2 This case covers off-diagonal elements with nr = nr except for r = s. Derive (9.95) In the latter formula the ‘+’-sign applies for s < r. nu = nu − 1. One obtains ˆ ΨF (N )|G|ΨF (N ) = ± (−1)qs +qr +qu +qv (1 − ns )(1 − nr )nu nv r. particle number is conserved. given in (9. v − r. 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. One obtains ˆ ΨF (N )|G|ΨF (N ) = (−1)qs +qt (1 − ns )nt r r=s. The matrix elements of the corresponding generic operators are (9.90) from (9. t − r.73) and S 2 given in (9. t.

Examples is the laser action. 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. in crystals. Unfortunately.u=1 σ. would be a simple exercise in linear algebra. rα sβ (9. evaluation of their stationary states and excitation energies.. s..96) and (9.97) The summation in (9. the spin part of the single–particle states is represented by α = 1 and β = − 1 . the description of many–particle systems. The great generality of the concepts developed in the context of many–particle systems can be judged from the fact .t. that the two–particle operator representing interactions between particles induces a variety of interesting phenomena.99) will be used. u refer only to the orbital part of the single–particle basis.s=1 σ 1 ˆ rσ r|t|s c† csσ + 2 S r. 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. which represents the interactions between particles.g. e. v c† c† cu ct .99) has preoccupied an important part of all intellectual efforts in Theoretical Physics during the past fifty years. superconductivity.u S (9.98) In many cases the Hamiltonian is spin-independent and the equivalent notation S H = r.s. The outcome of these studies is that even when interactions between components (particles) are simple.9. e. molecules.σ (9.4: Independent-Particle Models ˆ For S 2 one obtains 1ˆ 1 ˆ S2 = N + 2 4 S S 257 ˆ ˆ ˆ ˆ (Nrα − Nrβ )(Nsα − Nsβ ) − r. Actually. If it were not for the two–particle contribution to the Hamiltonian.97) is over the orbital states.t. In fact.s=1 r.s=1 c† c† crβ csα . but also ordinary phenomena like freezing and evaporation. 2 2 Exercise 9.6: Derive (9. the study of problems governed by Hamiltonians of the type (9.g. the concerted behaviour of interacting systems can deviate qualitatively from that of systems made up of independent components.97).4 Independent-Particle Models In the remaining part of this chapter we will apply the technique of operators of 2nd quantization to the description of many–fermion systems as they arise. the systems investigated involve a macroscopic number of components such that statistical mechanical approaches are envoked. v r s 2 r. Often. 9. the two–particle operator makes the description of many–particle systems a very hard problem. 9.98. In the latter case the indices r.96.s. s|ˆ|u. The fortunate side of this is. however. s|ˆ|t. t. atoms and nuclei. 9. u c† c† cuσ ctσ v rσ sσ r.3.

103) An account of some of these efforts can be found in Modelling Brain Function – The World of Attractor Neural Networks by D.100) in a form in which the factors c† csσ become rσ occupation number operators. New York.101) The unitary property of Urn U† U = U U† = 1 1 or S S (9. The Aufbau principle.1: Imagine that in a closed. Before continuing one may finally point out that the material world as we know it and as it establishes the varieties of natural substances. m = 1.4. i. ultimately depend in a crucial way on the fermion character of its constituent building blocks. Chemistry essentially would know only one element and Life would not exist. electrons and nucleons. If it were’nt for the fermion nature.s=1 σ ˆ rσ r|t|s c† csσ (9. water-filled jar all electrons of water turn their fermion nature into a boson nature.e. What might happen? Independent-Particle Hamiltonian A many–fermion system governed by an independent-particle Hamiltonian. The Hamiltonian of such system is S Ho = r. ranging from nuclei to the molecules of living systems . 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). Exercise 9. {|m).102) U ∗rm Urn = δmn . 2. can be described in a rather straightforward way. namely of neurons3 . according to which nuclei and atoms change their qualitative properties when they grow larger. S}.e. We will restrict our description in the following to systems composed of an even number (2N ) of particles and to spin-independent interactions. In such a representation Ho is diagonal and the stationary states can be stated readily. . another prototypical systems of many strongly interacting components. r=1 3 Urm U ∗sm = δrs m=1 (9. would not exist. all material systems would condensate into states which would depend little on particle number. i.258 Many–Particle Systems that today these concepts are providing insight into the functioning of biological brains. where S |m) = r=1 Urm |r . . The electronic properties of atoms with different numbers of electrons would differ little... 1989) . Transformation to a New Set of Creation and Annihilation Operators Our aim is to represent the Hamiltonian (9.100) ˆ ˆ We will denote the matrix elements of t as r|t|s = trs .J. (9. . a Hamiltonian without a two–particle operator contribution accounting for interactions among the particles. Amit (Cambridge University Press.

107) Urn c† rσ . they are Fermion creation and annihilation operators. d† ]+ = sσ nσ r.s. of N -Fermion states in which single particle states |nj ) as defined in (9. of (9. csσ = n=1 Usn dnσ .105) where ˜ ˆ tmn = (m|t|n) = S ∗ trs Urm Usn = r. hence. d† ]+ = δmn δσσ nσ (9. The l.4: Independent-Particle Models allows one to express conversely |r in terms of |m) S 259 |r = m=1 ∗ Urm |m) .h.106) which together with (9.9.100) yields ˜ Ho = tmn d† dnσ mσ where d† nσ = r=1 S S (9. nσ We demonstrate here the anti-commutation property (9. the anti-commutation properties (9. behave exactly like the operators c† and crσ behave for states |ΨF (N ) . c† ]+ = δσσ [dmσ .e.111) and.108) These operators describe the creation and annihilation operators of fermions in states |n) which are linear combinations (9.104)] S S ∗ Urm d† mσ m=1 c† rσ = . [dmσ .104) One can. dnσ = r=1 ∗ Urn crσ (9. (9.111) can be written. i. .103) allows one to express [c.46) as c† and crσ .s r The unitarity property r ∗ Urm Urn = δmn yields immediately identity (??). expand S trs = mn ∗ ˜ tmn Urm Usn (9.. rσ namely. (9. using (9.s=1 ˆ U † tU mn (9.108) and (9. (9.and two-particle operators hold in an analogous way for d† and dnσ . dnσ ]+ = 0 . ∗ ∗ Urm Urn .109) The operators (9. for nσ rσ example. accordingly.101) of states |r . [d† . the operators d† and dnσ .112) .113) Urm Usn [crσ .101) are occupied. d† ]+ = 0 (9.111). the expressions drived above for the matrix elements of one.45.110) can be demonstrated similarly. The unitary property (9.46).110) mσ nσ [dmσ . (9.f. 9.108) obey the same anti-commutation properties (9. In a basis of states N d† j σj |0 n j=1 (9.

Vn = (U1n .57) to the case of 2N particles. . σ2N . − 1 is called a closed 2 2 2 2 . r = 1. transformation matrices Urn and energy values n . a fermion in a spin state | 1 . the associated ( properly normalized) eigenvectors Vn are the columns of (Urn ). − 1 . U2n .e. Before proceeding we like to address the issue how the new representation. n2 .114) Ho = n=1 † n dnσ dnσ (9. S (9. ) = j=1 nj . . a simple matter to nσ state many–particle states in the new representation.106) diagonal.114) together with (9. .. hence.e. 2. It is.115) can be stated immediately since any many–particle wave function in the occupation number representation is suitable. σ2 . σk ) for j = k (9.115) with eigenvalues 2N E(n1 . Eigenstates of (9. We apply for this purpose (9. S are T real.101) gives us the freedom to choose the matrix (9. This equation poses the eigenvalue problem for the hermitian matrix (trs ). USn ). n = 1. .e. . is N |Φo = j=1 d† α d† β |0 . is obtained. .118) Ground State In case of an ordering m < n for m < n the state of lowest energy..260 Diagonal Representation Many–Particle Systems The transformation (9. σj ) = (nk . The corresponding eigenvalues n . A non-degenerate orbital state which is occupied by two fermions.117) are eigenstates of (9. (9. σ1 .116) which follows from (9.106) is equivalent to S trs Usn = s=1 n Urn ∀r.106) and the unitary property of Urn .. i.115) ˜ and involves solely occupation number operators N nσ = d† dnσ . 2 2 2 2 i. i. . The condition (9. .119) In this state the N lowest orbital eigenstates of trs are occupied each by an electron with spin |α = | 1 . The 2N fermion state 2N d† j σj |0 n j=1 where (nj . ˜ tmn = In this case Ho has the desired form S n δmn .e. n2N . i. the so-called ground state. . 1 and |β = | 1 . . j j (9. 1 as well as a fermion in a spin state | 1 . 2. (9.. . .

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

β are not eigenstates of S 2 .262 and ˆ Σ3 = − m.125) ˆ Contributions to S 2 acting on states |N + 2 . β qualify as eigenstates. also eigenstates of Σ2 . one with eigenvalue zero (singlet) and one with eigenvalue ‘2’ (triplet). ˆ An eigenvalue ‘2’ of the spin operator S 2 identifies the states |α. 2N ) drσ = √ 1 2N 2N e−irnπ/N cnσ n=1 (9.126) which follows from the occurence of squares of fermion operators which. ˆ 3 on the two states produces . for example. hence. The same result holds for |β.σ dN σ N σ. .σ d† +1. r = 1.122) is correct. do not need to be considered since they ˆ give vanishing contributions. One can. |N + 3 .130) . β . 2.n=N d† d† dmβ dnα d† +1α dN β |Φo = 0 mα nβ N (9. The action of Σ N +1 ˆ Σ3 |α. i. mα nβ (9. etc.128) The cyclic property of the system is expressed through c2N +1 σ = c1 σ . . however. for |α. apropriate eigenstates of the operators S 2 and Example: 2N Independent Electrons on a Ring In order to illustrate the procedure to obtain eigenstates of one-particle Hamiltonians (9. β and |β. c† +1 σ = c† σ 1 2N (9.2: Construct four operators O ˆ O ˆ such that O ˆ S3 . of course. α are. The system is assumed to posses a 2N –fold symmetry axis and interactions connect only orbital states |r → |r ± 1 . β =− m. of course. .σ = (9. α . . conclude that (9. m |Φo m m m γσ.n=N Many–Particle Systems N +1 d† d† dmβ dnα .4. . α ± |β. β both with eigenvalues 1. ˆ The remaining states |α. . |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 . (9.100) outlined above we consider a system of 2N electrons which move in a set of atomic orbitals |r . We identify the states |2N + 1 = |1 to avoid cumbersome summation limits.127) ˆ are triplet and singlet states. . α as triplet states.129) The symmetry of the Hamiltonian suggests to choose a new representation defined through the operators (r = 1.e. the linear combinations 1/2(|α. 2N which are located on a ring. α and |β. The two states |α. vanish. ˆ Exercise 9. 2. .

2N.132) one can conclude 2N ei(s−r)nπ/N + eisπ/N n=1 1 2N 2N ei(r−s)nπ/N . One can then express[.139) −2t cos r=1 σ rπ N d† drσ .140) The Hamiltonian is now diagonal and the construction outlined above can be applied.9.131) The (9. n=1 (9.132) gives for r = 0. 4N.130.f. 1−a (9.109)] ¸ cnσ = √ 1 2N 2N eirnπ/N drσ r=1 2N (9.133) Application to the l. of (9. . (9. (9.135) c† nσ 1 = √ 2N e−irnπ/N d† rσ r=1 (9. one obtains the new representation of Ho 2N ˆ Ho = −t r. thereby.s. 2N (9.136) ˆ and. .134) which. rσ (9. . 9. indeed. eisπ/N 1 − e2N isπ/N 1 − eisπ/N (9.h.4: Independent-Particle Models d† = √ rσ 1 2N 2N 263 eirnπ/N c† nσ n=1 √ 1 eirnπ/N .131) constitute a unitary transformation U defined through Urn = unitarity property follows from 2N eirsπ/N = 0 r=1 for 0 < s < 2N .132) an identity which can be derived employing the well-known expression for a finite geometric series m as = a s=1 1 − am . .138) ˜ trs = from which follows ˆ Ho = eirπ/N + e−irπ/N 2N δrs (9.s=1 σ ˜ rσ trs d† dsσ (9. .137) where 1 ˜ trs = e−irπ/N 2N Using (9. vanishes for integer s.

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

m = 1. . . The states |m are connected with the states |r through (m = 1. S} the elements of which will be labelled by indices ˜ m. S on which (9.s. r = 1. p. s|ˆ|t. n. . 2. This is done as follows: ˆ Vmf (|Φo (U) ) = 1 2 S r. u v r.s=1 σ r. . . u c† c† cuσ ctσ v (9.u=1 σ. S is a unitary matrix. . 2. |r = m=1 ∗ Urm |m .146) .145) which will play a pivotal role: it is this the state in which the fermions are assumed to be moving and relative to which the mean interactions acting on individual fermions is determined.t.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. .144) defined as in (9. .9. . .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.s. If one restricts m = 1.142) is based. u refer to the initial basis of single-particle states |r . . 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 (9.145). s|ˆ|t. as in in (9. In our formulas below we will adopt strictly the convention that indices r.108). N . We will adopt a second set of single-particle states {|m . 2. naturally. . The states |m serve to define a reference state of the system ˜ N |Φo (U) = j=1 d† α d† β |0 .σ and derive a single-particle operator which approximates this Hamiltonian.t.142) r|t|s c† csσ + rσ sσ 2 r. . t.143) with r. ˜ (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.u=1 σ. q. the corresponding U = ( Urm ) forms an S × N matrix and. U = ( Urm ) as defined in (9. 2. S) ˜ S S |m = ˜ r=1 Urm |r . s. 2. is not unitary.142) over this reference state to turn the contribution into an effective one-particle operator. . m = 1. . j j (9. .

t=1 σ.σ S ctσ + m. Many–Particle Systems (9. m|ˆ|t.σ S 1 − 2 where. m. for example.σ S d† dnσ mσ ctσ (9. n c† ctσ δσ σ ˜ v ˜ rσ m≤N r. by means of the rules (9.147) Since the reference state |Φo (U) is defined in terms of the single-particle states |m it is preferable ˜ to switch the representation of (9.146) accordingly.t=1 σ.n=1 σ. s|ˆ|t. u ˜ ˜v m.n.σ S d† dnσ cuσ mσ − r.s. m|ˆ|t. u Usm Uun . n = ˜ ˆ ˜ s.u=1 σ.u=1 σ.m.n=1 σ.u ∗ r.149) The remaining matrix elements appearing in (9. Since the averages affect only two creationannihilation operators actually a mixed representation is most suitable ˆ Vmf (|Φo (U) ) 1 = 2 1 2 1 2 1 2 1 2 S m. u c† ˜ vn sσ m. 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† d† dnσ ˜ v ˜ rσ mσ r.m.u=1 σ. s|ˆ|m. d† d† mσ nσ dmσ dnσ d† dnσ mσ = = = = and.σ S d† d† mσ nσ cuσ ctσ + r.t.σ .n.84).150) r.266 Here ··· denotes the average O = Φo (U)| O |Φo (U) . m|ˆ|˜ . s| v t. ˆ (9. hence.148) one obtains. s|ˆ|˜ .s. n c† ˜ v ˜ sσ m.n. For the averages ··· in (9. m| v |t.m. n c† c† v ˜ ˜ rσ sσ r. n|ˆ|t.148) r.n. u c† d† dnσ cuσ ˜ vn rσ mσ r.s.t.148) are obtained in a similar way by transforming only two of the four single-particle states into the new representation.σ S dmσ dnσ + r.

.u ( 2 r. u c† cuσ ˜ vn rσ m≤N r.151) Carrying out the sum σ δσσ = 2 leads to a factor 2 for the first two terms. u − r. .m.N (9. At this point this state is completely arbitrary. m=1. s|ˆ|˜ .2. ˜ v ˜ sσ m≤N s.. u = s.t=1 σ (9. The one-particle operator (9.u=1 σ.155) which is a function of the S × N –matrix U = ( Urm )r=1.t=1 σ. t| v |s. n c† ctσ ..σ S r. leaves open the question how the reference state should be chosen.152) can then be written in the form ˆ Vmf (|Φo (U) ) = r| vmf (|Φo (U) ) |s c† csσ ˆ (9.149) for the matrix elements in the mixed representation one can state the right hand side explicitly in terms of Urm . m|ˆ|˜ . s| v |t.S...m. The mean field approximation.σ S + − r. n c† ctσ δσ σ ˜ v ˜ rσ − r.5: Self-Consistent Field Theory 1 2 1 2 1 2 S 267 m.142) by Hmf (U) defined through S ˆ Hmf (U) = r.9. as introduced here.152) By means of expression (9.154) ˆ The mean field approximation replaces then the Hamiltonian (9.u=1 σ. We will address now a proper choice of the reference state. m|ˆ|t. r| v |u.σ r. t and renaming dummy summation indices one can demonstrate ˆ ˆ that term one and term two as well as term three and four are identical and one obtains S ˆ Vmf (|Φo (U) ) = 2 r.153) rσ r.. u c† cuσ δσσ ˜ vn sσ m≤N s.2. m|ˆ|˜ . Exploiting the symmetry r. u c† cuσ ˜ vn rσ (9. t| v |u.156) which defines the reference state |Φo (U) .s=1 σ ˆ r|t|s + r| vmf (|Φo (U) ) |s ˆ c† csσ rσ (9. s|ˆ|t.s σ where r| vmf (|Φo (U) ) |s = ˆ t.u=1 σ S − m. s )  ˆ ˆ  m≤N ∗ Utm Uum   (9.

and obtaining in a step (3) the ground state of this one-particle Hamiltonian according to the construction in Section 9. r = 1. SCF-Algorithm. . 2.142).158) where the labels m are ordered to obey the condition (1) m < (1) n for m < n . the independent-particle nature stemming from the fact that the functional form of the ground state. This procedure. 9. . S .4. U(1) is the S ×N –matrix (1) of the first N column vectors of Urm . Step 2M. . i. . S .268 Many–Particle Systems 9. Step 1: Choosing an Initial Reference State One defines the Hamiltonian ˆ (1) Hmf = r. The procedure determines the reference state as the ground state of the independent-particle Hamiltonian (9. i.145). however. 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 (9.155). . r = 1. determining in a step (2) the corresponding mean field Hamiltonian (9. M = 1. We will argue below that the self-consistent field ground state.155) following the method outlined in Section 9. . the reference state resulting from step 2M + 1 (within numerical error) is equal to the reference state resulting from step 2M − 1. (9. assuming in an intial step (1) a properly chosen reference state. 2. (9.159) One defines the reference state |Φo (U(1) ) using the definition (9. (9.: Determine Mean Field Hamiltonian In this step the mean field Hamiltonian S ˆ (M ) Hmf = r.s=1 σ ˆ r|hmf |s c† csσ rσ (M ) (9. . S . . S. 2.e..e. i. . . .160) .157) and determines the associated diagonal representation defined through S (1) ˆ r|t|s Usm = s=1 (1) (1) m Urm . . m = 1. steps (2) and (3) are being repeated M times until the procedure converges. . .s=1 σ S ˆ rσ r|t|s c† csσ (9. SCF-Algorithm. often also referred to as the Hartree-Fock approximation. m = 1. .145). is exact only for an independent-particle Hamiltonian. under conditions which often are realized. Let us state now the construction of the self-consistent field ground state in more detail. is the lowest energy independent-particle ground state one can construct. 2..6 Self-Consistent Field Algorithm The Self-Consistent Field (SCF) approximation. can achieve this goal only iteratively.119. The state thus determined is referred to as the self-consistent independent-particle ground state.s=1 σ S ˆ r|t|s + r| vmf |Φo (U(M ) ) ˆ |s c† csσ rσ = r.4 and defining this as the new reference state. 2.e.

163) The result yields the reference state |Φo (U(M +1) ) using the definition (9. 2. . 2. ˆ (b) How can the expectation values of the operator Prr be interpreted. . . i. m = 1. m = 1. M = 1.e.167) . . ˆ Exercise 9.145). . . the S × S–matrix S (M ) r|hmf |s N 269 ˆ = r|t|s + t. SCF-Algorithm: Continuation and Convergence Condition When step 2M + 1 is completed. Step 2M + 1. (e) Consider the creation operators N † gqσ = m=1 Vmq d† mσ (9. indicating that the state converged. i. S (9. .162) is solved adopting an ordering of the labels m which obeys the condition (M +1) m < (M +1) n for m < n . S . i..u=1 ( 2 r. step 2M + 2 is carried out. r = 1. ˆ t. . s ) ˆ ˆ m=1 Utm ∗(M ) (M Uum ) (9. 2. etc. σ p = δtr δsu δσσ .s=1 σ ˆ r|f |s c† csσ rσ (9. 2. r = 1. σ|ˆrs |u.1: Let Prs be the one-particle operator with the generic operator prs = |r s|.6: Self-Consistent Field Algorithm is evaluated. . .145). 2 . (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. . (9. S . s = 1. . . .e.164) ˆ (a) Determine the expectation value of Prs for the state as defined in (9. (c) Show that any spin-independent one-particle operator S ˆ F = r. t| v |s. .161) is calculated.. u − r. S do not exceed a preset threshold.: Diagonalize Mean Field Hamiltonian In this step the eigenvalue problem S (M r|hmf |s Usm +1) = s=1 (M ) (M +1) (M +1) Urm m . . t| v |u.166) ˆ (d) Define a similar two-particle operator Prstu . S.9. SCF-Algorithm. 2.e.165) can be written ˆ F = S ˆ ˆ r|f |s Prs r. .6.s=1 (9. U(M +1) is the (M +1) S × N –matrix of the first N column vectors of Urm .

. Equation (9.171) where the convention (9. (9.145) and (9.171) implies that for the mean field Hamiltonian holds S ˆ Hmf (U(SCF ) ) = m=1 σ (SCF ) † dmσ dmσ m . . (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. (f) Can one distinguish the states (9. We begin by summarizing the result of the SCF algorithm.173) .145).162.170) (SCF ) Here |r . . 2. (9.168) ˆ ˆ has the same expectation values for Prs and Prstu as the reference state (9. Note that this is not an mσ S × S–matrix! Show that the reference state defined through N † † gqα gqβ |0 q=1 (9.172) is obeyed. S denotes the initial single-particle states and Urm is the unitary S × S– transformation matrix obtained in (9.2: Determine the SCF ground state and its energy expectation value for a system of two particles described through the Hamiltonian (S = 2) H = σ ( c† c1σ + c† c2σ ) − t ( c† c2σ + c† c1σ ) 1σ 2σ 1σ 2σ ( c† c1σ c† c2σ 1σ 2σ σ.9. We denote the representation which results from the SCF algorithm after its convergence as follows: S |m = ˜ r=1 (SCF Urm ) |r . r = 1. .7 Properties of the SCF Ground State We want to investigate now the properties of the SCF ground state.168) through a physical observation? Exercise 9.σ + 2v ( c† c1α c† c1β + c† c2α c† c2β ) + v 1α 2α 1β 2β + c† c2σ c† 2σ 1σ c1σ ) .e.169) 9.270 Many–Particle Systems which are connected with d† through a unitary N × N –matrix (Vqm ).163). (SCF ) m < (SCF ) n for m < n (9.163).6. i.

Within the mean field approximation as described by (9.142).177).174) The first property we consider is the total spin character of SCF .e.7. Finally.177) 2 mm | v |mm ˜˜ ˆ ˜˜ − mm | v |m m ˜˜ ˆ ˜ ˜ . Since this state is composed only of closed shells one can conclude following Section 9.m =1 Comparision with (9.171) yields ˆ SCF H SCF N = N (9.4 that SCF is a total singlet state.e. the mean field potential ˆ Vmf (|Φo (U) ) defined in (9. Exercise 9. exploiting the spinindependence of (9.146) counts a particle twice.2: Derive (9. one obtains ˆ SCF H SCF N = N (9. the SCF ground state is N SCF = m=1 d† d† |0 . (9. mα mβ (9.173) holds N ˆ SCF Hmf (U(SCF ) ) SCF = 2 r=1 (SCF ) m .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. i.1: Reformulate (9. 2 m=1 m|t|m + ˜ ˆ˜ m.175) However. 2 m=1 (SCF ) m − m.9. ˆ SCF H SCF .162) is non-linear in Urm . We like to determine next the energy expectation value of SCF .177). Since the system under consideration has only 2N particles altogether. Employing expressions (9.142).93) for the expectation values (diagonal elements) of the one-particle and two-particle part of the Hamiltonian (9. This over-counting leads to the correction term in (9..7.142).. once as a member of the ground state. . r.176) 2 mm | v |mm ˜˜ ˆ ˜˜ − mm | v |m m ˜˜ ˆ ˜ ˜ .7: Properties of the SCF Ground State 271 i. (SCF ) ˆ Exercise 9. one can also determine the expectation value of SCF for the Hamiltonian (9.171) in terms of matrix elements r|t|s .84) and (9. and once as a probe particle being subject to the mean pair interaction. s|ˆ|t. v (SCF ) Show that the resulting eigenvalue problem of the type (9. the mean-field Hamiltonian is diagonal in the SCF representation. u and Urm .

We assume that there are altogether S = 2N0 lattice sites. is characterized through the parameters t. the second term on ther. v represents the ‘on-site’ Coulomb repulsion. t = v δru δst δrs . In the present Section we want to adapt the algorithm to systems containing a macroscopically large number of fermions. reads 1 ˆ r. u c† c† cuσ ctσ . in the two-dimensional copper oxide lattices of high temperature superconductors. except in case of one-dimensional systems. in the usual two–particle operator form. respectively) in state |r at lattice site r.e.179) r+1.128). The first term on the r.178) and through the so-called filling factor n n= 2N N = .. β for ‘up’ and ‘down’ rσ spins. N0 which are populated by 2N electrons (we choose an even number of electrons for convenience). We consider as an example the Hubbard model of an infinite linear lattice and apply the model to describe magnetic instabilities in metals.178). referring to the fact that in this case the band of single particle energies (9. (9. labeled r = −N0 + 1.s.192) derived below is filled half.178) Here the index r. rσ rσ r σσ (9. i. real number. The case n = 1 is termed the half filled band case. 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.s.σ rσ The potential energy term. s|ˆ|u. t is assumed to be a positive. The one-dimensional Hubbard model is described by the Hamiltonian ˆ H = −t rσ † c† r+1. r ∈ Z.h.180) V = rσ sσ 2 r.7 for finite systems. . Due to the lack of an exact solution in dimensions higher than one. for example.178). describes the sites of a linear lattice.σ crσ + crσ cr+1.σ + v 2 c† c† crσ crσ .8 Mean Field Theory for Macroscopic Systems The SCF algorithm has been developed in Sections 9.178) v contributes only in case that two electrons occupy the same lattice site. t describes the coupling of state |r to states |r ± 1 at the two neighboring lattice sites. cannot be solved exactly.σ crσ + crσ cr+1. of (9. approximation schemes like the self-consistent field approximation play an important role.e.. . . † ˆ Ho = −t c† . of (9. Both N0 and N are macroscopically large numbers. In spite of the simplicity of its Hamiltonian the Hubbard model. .e.h. . v (9.272 Many–Particle Systems 9.t.4–9. as stated through Hamiltonian (9.182) n can assume values 0 ≤ n ≤ 2. s|ˆ|t.181) The Hubbard model.u σσ where r. S N0 (9. The operator c† (crσ ) creates (destroys) an electron with spin σ (σ = α. i. is identical to that of the independent–particle Hamiltonian H0 (9. v of the Hubbard Hamiltonian (9. According to (9.. i. v (9.s.178).

is Urm = √ 1 exp (irmπ/N0 ) .187) The energy levels are labeled such that 0 < ±1 < . The unitary transformation which diagonalizes H0 .9. Inserting (9. by N −1 ||SCF = m=0 d† d† |0 . we identify |r + N0 = |r and later let N0 go to infinity.179).. 2N0 (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. Accordingly.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 . in fact.178) can be determined by applying (9.174). rσ (9.e.188) . We can now embark on step 2 of the SCF algorithm. The self-consistent field Hamiltonian Hmf which corresponds to the Hubbard Hamiltonian (9.178) within the framework of the self-consistent field theory.8: Macroscopic Systems 273 Since we are dealing presently with a macroscopic system. Using (9. one obtains for the mean field Hamiltonian N −1 ˆ ˆ Hmf (U) = H0 + v r. .181.140).σ m=0 ∗ Urm Urm c† crσ . For this purpose we apply the SCF theory presented in Section 9.155. the operator H0 can be re-written in the diagonal form N0 ˆ H0 = m=−N0 +1 σ m d† dmσ mσ (9. 9.183) where the operators d† are related to the original creation operators c† through the unitary mσ rσ ˆ transformation U defined in (9. mα mβ (9. < ±(N0 −1) < N0 holds as can be readily verified. N0 (9.186) where m = −2t cos mπ . SCF Ground State for the Hubbard Model We seek to determine the ground state of the Hubbard model stated by (9.144). we assume that the SCF ground state is given by (9.184) and using (9.6.154). 9.185) ˆ According to (9.185) into (9. i. rσ rσ (9. Accordingly. . This task has ˆ been solved already on page 262. boundary effects are assumed to be negligible.184) ˆ We are now ready to apply step 1 of the SCF algorithm and diagonalize term H0 .

f. The energy m can be expressed as a function of a continuous variable k ∈] − π. i.189) and. .182) of the filling factor n [c.190) We have used in the latter expression the definition (9. The energy Emf of this state can be calculated readily by using (9. denoted by F (see Fig.e. −π ≤k ≤π . can be obtained. 4 So far we have not exploited the fact that N0 and N are macroscopically large quantities. Indeed. Indeed. in the limit N0 → ∞. mσ (9. i. Alternative Description of the Mean Field Approximation The state (9.182)]. the results of different formulations agree qualitatively. namely. −π + N0 .e. for km = mπ/N0 holds km ∈ −π + N0 .192) This dispersion law is presented in Fig.109) and using (9. Apparently. The bottom of the energy band corresponds to (0) = −2t and the width of the energy band is 4t. ˆ The Hamiltonian Hmf .187). holds k ≡ km ∈] − π.187) form a quasi1 2 continuous energy band. . even for one and the same Hamiltonian. hence. is already diagonal.178). according to the Pauli principle. N0 ˆ (2) ˆ Hmf = Hmf = m=−N0 +1 σ m + vn † dmσ dmσ 2 (9. called the Fermi energy. . One can see that −2t ≤ (k) ≤ 2t holds.. When N0 becomes macroscopically large the single particle discrete energy levels (9. 9. Emf is then N −1 v 2N0 Emf = 2 m=0 vn vn −N = 2 m+ 2 2 N −1 m m=0 + vn 4 (9. dnσ according to (9. 9. the self–consistency condition is exactly met and the SCF algorithm converged after two steps. crσ through d† . (9. π]. 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. by filling up this energy band with electrons from the bottom of the band (which corresponds to k = 0) to a maximum energy. the state with the lowest possible energy. The ground state of the system.191) where m is given in (9.1. . Usually. π].. the ground state of the Hubbard model in the self-consistent field approximation can be determined exactly. but may differ quantitatively. π and.190).274 Many–Particle Systems Expressing c† . as given in (9. (9.185) one can prove rσ mσ N0 c† crσ = rσ rσ m=−N0 +1 σ d† dmσ . there exist several ways of formulating mean field approximations. In other words. through the so-called dispersion law (k) = −2t cos k .1). To demonstrate this we consider an alternative formulation of the mean field approximation for Hamiltonian (9.177) and mm |ˆ|mm = mm |ˆ|m m = ˜˜ v ˜˜ ˜˜ v ˜ ˜ which holds in the present case.

we have employed (9. say the one given in (9.195) ˆ ˆ ˆ and where δ (Nrσ ) = Nrσ − Nrσ describes the fluctuations of Nrσ .194) ˆ For a given reference state..8: Macroscopic Systems 275 Figure 9. 9. (9. (9.193) into (9..183).1: The dispersion law (k) for independent electrons on an infinite one dimensional lattice.45. Nrσ = SCF ||Nrσ ||SCF . where Nrσ is the mean occupation number of the one particle state |rσ .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 .196) (9. i.193) where. The mean field approximation neglects the . the deviation of Nrσ from its mean value Nrσ for the corresponding reference state. Inserting (9. i.56). F = (±kF ) denotes the Fermi energy.9. First. in the last term on the right hand side. let us express the interaction term in (9. the operator Nrσ can be written ˆ Nrσ = Nrσ + δ (Nrσ ) .e.178) and performing the summation over the spin indices results in ˆ ˆ H = H0 + v r ˆ ˆ Nrα Nrβ .e. The anti-commutation properties (9.178) in terms of the occupation number operaˆ tors Nrσ = c† crσ . rσ rσ rσ (9.

183). In this approximation one can express ˆ ˆ Nrα Nrβ = ≈ = ( Nrα + δ (Nrα )) ( Nrβ + δ (Nrβ )) Nrα Nrβ + Nrα δ (Nrβ ) + Nrβ δ (Nrα ) ˆ ˆ Nrα Nrβ + Nrβ Nrα − Nrα Nrβ (9. 2N0 2 holds ˆ Hmf = = mσ mv ˆ H0 + 2 vn2 2 rσ nv † vn dmσ dmσ − N . yields essentially the same mean field Hamiltonian (9. Because of Nrσ = c† crσ rσ N −1 = m.200) allowing. consequently.m ∗ Urm Urm d† dm σ mσ = m=0 ∗ Urm Urm = n N = .198) as the one obtained by using the mean field potential (9. Spin-Polarized Mean Field Ground State ˆ We want to consider now a ground state for Hmf defined through Nrσ = nσ = const (9.195). Note that the procedure employed in (9..e. however. Therefore. m+ 2 2 c† crσ − 2N0 rσ (9. For such state the mean occupation number nrσ is uniform at all lattice sites. non-vanishing magnetization. separating the occupation number operator into a mean value plus fluctuation and neglecting the fluctuations in 2nd order. In case nα = nβ the following construction will lead to a ground state which coincides with the non-magnetic ground state (9. in the mean field approximation. i. (9. that nα and nβ assume different values. is given by (9. the ground state of the system can have a non-zero local spin and. .194) by dropping all the terms which contain the fluctuations to second order.199) which is identical to expression (9.191) for the ground state energy.183) .197) and the corresponding mean field Hamiltonian becomes ˆ ˆ Hmf = H0 + v r ˆ ˆ Nrα Nrβ + Nrβ Nrα − v r Nrα Nrβ .276 Many–Particle Systems fluctuations to second order and the mean field Hamiltonian is obtained from (9. To determine a magnetic ground state we note ˆ Nrσ = r mm r ∗ Urm Urm d† dm σ = mσ mm ˆ δmm Nmσ = m ˆ Nmσ .146).198) Let us assume that the ground state of the Hubbard model. but the mean number of particles with spin α can be different from the mean number of particles with spin β.

one can expect the system to lower its energy by assuming nα = nβ and.203) is provided by the continuous function σ (k) = (k) + vn−σ . if nα > nβ then β (k) > α (k). 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.205) Such minimization is realized through the method of Lagrangian multipliers. N0 ∼ 1023 . 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 ). 2π (9. Therefore. (9.192). (9.203) mσ = m + vn−σ .206) ˜mf is minimized where µ is the Lagrangian multiplier considered an independent variable. 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. such that the larger nσ .e.206) needs to be evaluated. nβ and µ.201) ( m ˆ + vnβ ) Nmα + ( m ˆ + vnα ) Nmβ − 2N0 vnα nβ . i. the so-called Fermi energy F = µ (see below). i. e.206) ascertains that condition (9.206).9. subject to the constraint n = nα + nβ . hence..202) This Hamiltonian describes a system of non-interacting electrons with a spin-dependent dispersion law (9. E with respect to nα .8: Macroscopic Systems from which we obtain nσ = The mean field Hamiltonian is then ˆ Hmf = m 277 1 2N0 Nmσ . 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) . At this point we exploit the fact that our system is macroscopically large. The values of nα and nβ are determined by minimizing the energy density (9. (9. a spin–polarized ground state. m (9. For this purpose the sum in (9. The additional term in (9.207) where (k) is given by (9.205) is met. β) are determined by minimizing the energy density of the ground state Hmf 1 = [ mσ Nmσ ] − vnα nβ (9. the smaller is σ (k) for a given k. as is shown schematically in Figure 9. In fact. According to this method one minimizes ˜ Emf = Emf + µ (n − nα − nβ ) .204) Emf ≡ 2N0 2N0 m... Since β (k) − α (k) = v(nα − nβ ).g. In the limit N0 → ∞ this sum can be expressed as an integral. (9. In this limit the discrete energy spectrum (9. The actual values of nσ (σ = α.σ with respect to nα and nβ .e.208) .2a.

(b) Graphical definition of the limiting energies α and β . which is usually called the density of states.278 Many–Particle Systems (a) (b) Figure 9.211) and using π dkδ(k ± (π − arccos( /2t))) = θ(2t − | |) . Since the equation f (k) ≡ 2t cos k + = 0 has two simple roots. If the function f depends explicitly only on (k) one can state π −π dk f ( (k)) = 2π π ∞ ρ( )d f ( ) −∞ (9.212) where xi are the simple roots of the function f (x). In our case (k) is given by (9.210) Here δ(x) is the Dirac–delta function and ρ( ). 2π (9.209) where ρ( ) = −π dk δ ( − (k)) . −π . and noting |f (k)| = |2t sin k| = 2t one obtains δ( − (k)) = 1 − cos2 k = (2t)2 − 2 . namely k1. 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. |f (xi )| (9.192) and.211) In order to calculate the integral we recall the following property of the Dirac–delta function δ [f (x)] = i δ(x − xi ) . holds π ρ( ) = −π dk δ( + 2t cos k) .2 = ±(π − arccos( /2t)). unit energy and a given spin orientation of the particle. Inserting this last result into (9. therefore. 2π (9.210). gives the the number of available one particle states per site. δ(k + π − arccos( /2t)) + δ(k − π + arccos( /2t)) (2t)2 − 2 . For a given dispersion law (k) the density of states ρ( ) can be calculated by employing equation (9.

216) allow one.206)] depends only on continuous quantities. ∂ β ˜ ∂ Emf =0. (9.213) and carrying out the resulting integral yields nσ = 1 arcsin( σ /2t) . of t. (9. We can employ the results obtained to express the ground state energy density (9.221) ∂µ Equations (9. 9. if we assume that is restricted to the interval ] − 2t. ∂ α ˜ ∂ Emf =0. outside the energy band (k). ∂µ (9. Replacing again the sum by an integral over energy one can write σ nσ = −2t ρ( )d (9.220) µ can be readily obtained µ= 1 ( 2 α + β) + vn . (9.219–9.e. + vnα − µ = 0 . on conditions for a ground state of minimum energy are ˜ ∂ Emf =0.213) can simply be replaced by 1.213) The presence of the θ function in the above formula guaranties that the density of states vanishes for | | > 2t. β .219) (9.214) are given by the sum (9. 2 (9. 2t[ then the θ function in (9. namely. (9.222) . i. nα and nβ as a function of the parameters of the Hubbard model.f..e. (9. one arrives at the following expression of the density of states ρ( ) = θ(2t − | |) π (2t)2 − 2 .220) β ˜ ∂ Emf = n − nα − nβ = 0 . β (9.201). to determine the unknown quantities α . v and n. µ.217) and µ.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.216) Using (9.8: Macroscopic Systems 279 where θ(x) is the step function. θ(x) = 1 if x > 0 and θ(x) = 0 if x < 0. From (9.218) The derivatives can be readily determined and the conditions (9. i.2]. in principle. i. π α.215) nα and nβ in (9.204) through an integral expression.221) and (9. the necessary ˜ Since Emf [c. 9.f.219.. F = α + vnβ = β + vnα . namely. One obtains ˜ Emf = α β ρ( )d + −2t −2t ρ( )d + vnα nβ + µ(n − nα − nβ ) .218) read ˜ ∂ Emf = ∂ α ˜ ∂ Emf = ∂ β and α + vnβ − µ = 0 . (9. Fig..e.9.

219–9.215) reveals that µ is indeed equal to the Fermi energy F (see also Figure 9. One still needs to determine α and β .227) has only the trivial solution ∆ = 0. (9.226) ∆ = f (∆) . First. the slope of f (∆) is positive for all ∆ ≥ 0.222) and (9. from (9.3]. From the definition (9. The slope of f (∆) at the origin depends on the Hubbard model parameter v.2b). as reflected by the expression d[f (∆)] = vρ( d∆ β + ∆) = π (2t)2 v − (v∆ + 2 β) >0.226) one can obtain ∆ by solving (cf. i.226) follows that f (∆) is a monotonically increasing function of ∆. (9.217) and (9.e. Equation (9.183) without magnetization. (9. For this purpose we consider the magnetization of the mean field ground state ∆ ≡ nα − nβ . One can determine vc by equating the slope of f (∆) at the origin to 1. Indeed.223 one obtains ∆= and second.f. and only if. We assume in the following discussion that the implicit ∆ dependence of β in f (∆) can be neglected. (9. if.229) has a wider range of validity than one can infer from the present analysis of the Hubbard model.3.228)] vc ρ ( β ) = 1 .227) Since f (0) = 0 one can infer that ∆ = 0 is always a solution (the so called paramagnetic solution) of condition (9. (9.225) Defining the function v∆ f (∆) = 0 ρ( β + )d .224) v∆ ∆= β ρ( )d = 0 ρ( β + )d . condition (9.228) Since the total number of states per site is finite. (9. (9.e. For v above a critical value vc .220) and (9.280 Many–Particle Systems Comparision of (9. (9.. One can express ∆ as a function of α and β in two different ways.. Accordingly. (9. so called ferromagnetic solution. the slope of f (∆) at the origin is larger than 1 [c.223) follows α α − v β . If the slope of f (∆) at the origin is less than 1.229) 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. non trivial. from (9.227) can be solved graphically. through the condition [c. .223) ∆ = 0 corresponds to the ground state (9. condition (9.f. For this purpose one plots f (∆) versus ∆ as is shown schematically in Figure 9.227) will have a second. f (∆) converges to a maximum value as ∆ → ∆max = n. Figure 9. i. a magnetic ground state. the Hubbard model has a ground state with ∆ = 0. The Stoner criterion (9.227).

3: Graphical solution of the mean field equation ∆ = f (∆). the plot of vc vs.9. n. n). v (9. 2t As already mentioned. Accordingly. (9. β x+y x−y = = π(n − 1) 2t π (sin x − sin y) = π∆ . the magnetization ∆ of the ground state vanishes and remains zero for values v < vc . 9. Equation (9.205) the following set of equations determining x.232) For a given 2t the values of x and y can be obtained numerically and other relevant quantities v mentioned above can be calculated.233) Employing the approximation sin(y + δ) − sin y ≈ δ cos y for small δ. is presented in Figure 9. i. hence. any .e.8: Macroscopic Systems 281 Figure 9. β = 2t sin y .. t. (9.4.231) (9.233) yield the desired expression vc π = π cos (n − 1) . The corresponding 2t analytical expression provides the phase diagram of the ground state: when v approaches vc from above. 9. the ground state is characterized by the quantities (v. We like to determine finally the dependence of vc on the filling factor n. 2 2 2 (9. From (9. By parameterizing α and β α = 2t sin x .230) one obtains from (9. α.232) states that x − y is proportional to ∆. (9.231) follows y= δ π π (n − 1) − ≈ (n − 1) . Since ∆ vanishes near vc we set x = y + δ where δ is a small quantity. y and.219.234) 2t 2 The resulting phase diagram of the ground state.232) together with (9.220.

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

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

284 Many–Particle Systems .

Furthermore. We will find that actually several Lorentz–invariant equations which replace (10. The transformations beween such frames according to the Theory of Special Relativity are described by Lorentz transformations. t) c 2 i + qV (r. t ) in another frame. t) are vectors of dimension larger one. etc.2) The replacement of (10.2) will result. v = v1 x1 .1) which connect space time coordinates (x1 . x3 . Here c denotes the velocity of light. spin– 1 particles.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 description should not allow one to differentiate between frames of reference which are moving relative to each other with a constant uniform velocity v. it will be necessary to begin this Chapter with an investigation of the group of Lorentz transformations and their representation in the space of position r and time t. t) in one frame with space time coordinates (x1 . e. spin–0 particles.g. Obviously. t) = ∂t 1 2m q − A(r. i. t) ∂ i ψ(r.f. some of the equations describe a particle together with its anti-particle. x2 . x2 . any of these equations being specific for certain classes of particles. x3 . 8. these ˆ transformations are x1 = x1 − v1 t 1 − v1 2 c . i. t = t− 1 − v1 x c2 1 v1 2 c . the components representing the spin attribute of particles and also representing together with a particle its anti-particle. It is not possible to uncouple the equations to describe only a single type particle without affecting negatively the Lorentz invariance of the equations.45)] described by an electrical potential V (r.e.. x2 = x2 . (refeq:ham2. t) and a vector potential A(r. 2 As mentioned. 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. x3 = x3 (10. t) (10..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.. In case that v is oriented along the x1 –axis.e. We will introduce below Lorentz-invariant differential equations which take the place of the Schr¨dinger equation of a paro ticle of mass m and charge q in an electromagnetic field [c. t) ψ(r. The representation 285 .

5) and the Dirac (Sect.7 a heuristic derivation of the two most widely used and best known Lorentz–invariant field equations. x 1 . Before introduco ing these general Lorentz–invariant field equations we will provide in Sects.e. This space is called the Minkowski space. 10.e. The reason for this choice is that the transformations (10. x1 .1). t) as components. This choice implies dim(time) = dim(length).7) equation. Denoting by xµ the . Rather than starting from (10. 10. This will provide us with a general set of Lorentz–invariant equations which for various particles take the place of the Schr¨dinger equation. We will. assume new units for time such that the velocity of light c becomes c = 1. Rather than following this avenue we will introduce Lorentz transformations within a setting which does not require real and imaginary coordinates. The elements L ∈ L act on 4–dimensional vectors of position– and time–coordinates. x3 ) (10. however. 5 to the groups SO(3) and SU(2) of rotation transformations of space coordinates and of spin.3) where x0 = ct describes the time coordinate and (x1 . Minkowski Space Historically. Note that the components of xµ all have the same dimension. x2 . The Group of Lorentz Transformations L = O(3. henceforth.1) leave the quantity s2 = (x0 )2 − (x1 )2 − (x2 )2 − (x3 )2 (10. the group of Lorentz transformations (10. we will provide a more basic definition of the transformations.1) The Lorentz transformations L describe the relationship between space-time coordinates xµ of two reference frames which move relative to each other with uniform fixed velocity v and which might be reoriented relative to each other by a rotation around a common origin. 0 1 2 3 (10. In such a space Lorentz transformations correspond to 4-dimensional rotations. We will find that this definition will lead us back to the transformation law (10. 10. x 3 ) holds (x0 )2 − (x1 )2 − (x2 )2 − (x3 )2 = (x )2 − (x )2 − (x )2 − (x )2 . 10. for the transformed space-time–cordinates x µ = (x 0 .e. namely the Klein–Gordon (Sect.1 Natural Representation of the Lorentz Group In this Section we consider the natural representation of the Lorentz group L.1). namely that of length. We will denote these vectors as follows def xµ = (x0 . 10. i. x 2 .5.286 Relativistic Quantum Mechanics in Sect.. 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. 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.4 to cover fields. wave functions ψ(r. i.1 will be extended in Sect. x3 ) = r describes the space coordinates. but in a setting of representation theory methods as applied in Secti..4) invariant.1). i. 10. x2 . t) and vectors with functions ψ(r.

L ∈ GL(4. the latter set constituting a group.6) yields Lµ ρ gµν Lν σ xρ xσ = gρσ xρ xσ . First. (10. summation over that index is assumed without explicitly noting it. 3: 4-vector: xµ . 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. the notation in (10. the latter notation will be used for different quantities further below. Aµν . however.e.8) new old old The summation convention allows us to write (10. 4 × 4 tensor: Aµ ν .5) into scalar products. Since this holds for any xµ it must be true Lµ ρ gµν Lν σ = gρσ .10.. [ A more conventional notation would be Lµν . R).11) This condition specifies the key property of Lorentz transformations. R). the . ν=0 (10.12) (10. The subset of GL(4.9) and (10. 0 0 0 −1 (10.10) Combining condition (10. This will be explained further below. This condition can be written xµ gµν xν = x gµν x where  µ ν (10. Aµ ν B ν ρ = ν=0 Aµ ν B ν ρ .1: Natural Representation of the Lorentz Group 287 coordinates in one reference frame and by x µ the coordinates in the other reference frame.9)  1 0 0  0 −1 0 0   ( gµν ) =   0 0 −1 0  = g .7) The reason for the notation is two-fold. ν = 0.] The following possibilities exist for the positioning of the indices µ.6) x µ = Lµ ν xν . Aµ ν . (10. Aµν .. The Lorentz transformations are non-singular 4 × 4–matrices with real coefficients. Aµ ν xν = µ=0 new old Aµ ν xν .e.6) Here Lµ ν are the elements of a 4 × 4–matrix representing the Lorentz transformation.5) invariant. i.6) allows us to introduce the so-called summation conventon: any time the same index appears in an upper and a lower position. 3 3 3 yµ xµ = new yµ xµ . the Lorentz transformations constitute a linear transformation which we denote by 3 x µ = ν=0 Lµ ν xν . (10. 1. xµ . The second reason is that upper and lower positions allow us to accomodate the expression (10. We will exploit this property below to determine the general form of the Lorentz transformations. The Lorentz transformations form the subgroup of all matrices which leave the expression (10. i. 2.

e.18) (10. L ∈ GL(4.13) L = ( Lµ ν ) =  2 0  L 0 L2 1 L2 2 L3 3  . For L3 = L1 L2 holds LT g L3 = LT LT g L1 L2 = LT (LT gL1 ) L2 = LT g L2 = g . of (10. We want to show now L = O(3.20) Multiplying (10.14)  L0 0 −L1 0 −L2 0 −L3 0  −L0 1 L1 1 L2 1 L3 1   . 10. a group. 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. L3 ∈ O(3. L2 ∈ O(3. From this one can obtain using g2 = 1 1 (gLT g)L = 1 and.e.17) one obtains (L−1 )T g L−1 = gLgggLT g = gLgLT g . 1) = { L. .14) from the right by gLT and using (10. employing expressions (10.22) (10.14). LT gL = g } . i. In fact. the associated inverse obeys (10..e. property (10.e.1).. 3 2 1 2 1 2 (10.h. hence.1) is.21) i.10) of g one can rewrite the invariance property (10. = g LT g =  0 1 2  −L 2 L 2 L 2 L3 2  −L0 3 L1 3 L2 3 L3 3 (10. L−1 ∈ O(3. One can also show that if L ∈ O(3.1). of O(3.19) i.1). (10.15) 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.s.1) since it satisfies (10..288 Relativistic Quantum Mechanics elements of which satisfy this condition.16) The corresponding expression for (LT )−1 is obviously (LT )−1 = (L−1 )T = g L g . We consider 1 then L1 . This set is identical with the set of all Lorentz transformations L. This stipulates that O(3.16. L3 0 L3 1 L3 2 L3 3 Using the definition (10. To demonstrate the group property of O(3.1) ⊂ GL(4.14). is called O(3.1).20) results in the desired property (L−1 )T g L−1 = g . R). (10.. in fact.12) LT gL = g .17) we note first that the identity matrix 1 is an element of O(3. R) is a group.1).14) holds for the inverse of L.15) (10.1). i. the inverse of L 1  L−1 (10. (10.

L↓ are disjunct sets defined as follows + + − − L↑ + L↓ + L↑ − L↓ − = { L. L0 0 ≥ 1} . L ∈ O(3.12) the case ρ = 0.34) j=1 j=1 . + = − L↑ g + (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 . det L = −1. Obviously. M1 = a M2 }. entirely suitable to investigate first only 1. = { L. Lorentz transformations in L↑ . 1). One can also 1 verify that one can write L↑ − L↓ + L↓ − = gL↑ + = = −L↑ .14).29) (10. L ∈ O(3.14). σ = 0 yields L0 0 2 − L1 0 2 − L2 0 2 − L3 0 2 = 1. From this we can conclude L0 0 ≥ 1 or L0 0 ≤ −1 . 1).25) i. L ∈ O(3. Schwartz’s inequality yields j=1  2 3 3 3  j=1 A jB 0  ≤ 0 j A 0 2 j Bj 0 2 .10. ∃M2 . B ∈ L+ holds C = AB ∈ L+ . (10.14) which we employ in the formulation (10.1).32) (10. Using det AB = det A det B and det AT = det A yields (det L)2 = 1 or det L = ±1 . (10. A second class property follows from (10.e..1: Natural Representation of the Lorentz Group Classification of Lorentz Transformations 289 We like to classify now the elements of L = O(3. Considering in (10. = { L. + except for the trivial factors g and −1 It is. = { L. For this purpose we consider first the value of det L. + + ↑ ↑ We can also demonstrate that for A. L0 0 ≤ −1} . det L = −1. A statement on this value can be made on account of property (10. det L = 1. L ∈ O(3. det L = 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↑ .33) where we used the definition aM = {M1 . The above shows that the set of proper Lorentz transformations L↑ allows one to generate all Lorentz transformations. (10. + − gL↑ + = L↑ g .31) (10.28) (10. It holds g ∈ L and −1 ∈ L as one can readily verify testing for property (10. 1). Properties (10.24) or since (L1 0 )2 + (L2 0 )2 + (L3 0 )2 ≥ 0 it holds (L0 0 )2 ≥ 1. M2 ∈ M.30) ≤ −1} .23) One can classify Lorentz transformations according to the value of the determinant into two distinct classes. L↓ .12). L↑ . there exist two other distinct classes.27) (10. 1). + We start our investigation by demonstrating that L↑ forms a group.26) (10. L0 0 ≥ 1} .23) and (10. L0 0 (10. (10. hence. L↑ contains 1 1.

16).13). δ00 = 1 and It should be noted that according to our present definition holds δµν = gµρ δ δ11 = δ22 = δ33 = −1. since the group property of O(3. + Accordingly. Infinitesimal Lorentz transformations The transformations in L↑ have the property that they are continously connected to the identity 1 1. This relationship shows (L−1 )0 0 = L0 0 . i. we consider transformations Lµ ν = δ µ ν + µ ν . orthochronous Lorentz transformations. ρ ν (10.21) j=1 (A j 0   3 j=1 A0 j B j 0  ≤ 2 (A0 0 )2 − 1 (B 0 0 )2 − 1 < (A0 0 )2 (B 0 0 )2 . Property (10. . 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. | 3 A0 j B j 0 | < A0 0 B 0 0 . The next group property of L↑ to be demonstrated is the existence of the inverse. hence.34) provides then the estimate conclude from (10. holds L0 0 > 0 and the value of the determinant is close to unity. + i. (10. if we enforce (10.36) It holds. 1 0 0 = ≥ 1 and.37) (10.1) ascertains CT gC = g it must hold C 0 0 ≥ 1. from which one can conclude L−1 ∈ L↑ ..39) where we have employed the matrix form T 1 defined as in (10. We also note that the identity operator 1 has elements 1 + 1 µν = δµν 1 where we defined1 δµν = 1 for µ = ν 0 for µ = ν (10. therefore. µ ν small .38) For these transformations. In the following we + will consider solely this subgroup of SO(3.35) One can conclude.12) follows (B 0 0 )2 − 3 (B j 0 )2 = 1 or 3 (B j 0 )2 = (B 0 0 )2 − 1. Since the associative property holds for matrix multipli1 1 + ↑ cation we have verified that L+ is indeed a subgroup of SO(3. L↑ is called the subgroup of proper.e. (10. (10.14) implies 1 + T g (1 + ) = g 1 1 (10. To order O( 2 ) holds g + g = 0.. + transformations in a small neighborhood of 1 which we parametrize by infinitesimal parameters. accordingly. 1 ∈ L↑ .40) and. Since A0 0 ≥ 1 and B 0 0 ≥ 1. Similarly. one can j=1 j=1 3 0 )2 = (A0 )2 − 1.1). In fact.290 Relativistic Quantum Mechanics From (10.e. obviously j=1 A0 0 B 0 0 ≥ 1. For the inverse of + ↑ any L ∈ L+ holds (10. obviously. 1 We will then employ the Lie group properties to generate all transformations in L↑ .14) actually L0 0 ≥ 1 and det L = 1 must hold. Using the above expression for C 0 0 one can state C 0 0 > 0.1).

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 (10.42) This relationship implies µ 0 j µ j = 0 = j 0 .44) This result allows us now to define six generators for the Lorentz transformations(k = 1. J =  0  3  0 1 0 0 0 0 0 0 0 0 0 0 0     (10. K ] [ Jk .43) k = − Inspection shows that the matrix has 6 independent elements and can be written   0 −w1 −w2 −w3  −w1 0 −ϑ3 ϑ2   . 2. ϑ2 . w2 .45) (10. K2 =  . 2.41) 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    . K3 =   0  −1 0  0 0 0 0  0 0  0 0 0 0 −1 0 0 0 0 These commutators obey the following commutation relationships [ Jk . 2.46)  0 0   0 0  . 3 (10. w3 ) =   −w2 ϑ3 0 −ϑ1  −w3 −ϑ2 ϑ1 0 (10.47)     0 −1 0 0 0 0 −1 0 0  −1   0 0   0 0 0 0  0 0  K1 =  . The generators are explicitly  0 0 0 0  0 0 0 0 J1 =   0 0 0 −1 0 0 1 0  (10. (ϑ1 . w1 . 3 .15) one can conclude T 291 = −g g (10. k = 1. other five parameters zero) Kk = (wk = 1. J2 =    0 0 0 −1  0 0 0 0   0 0 0 0 0  0 0 −1 0 1  .49) . J ] [ Kk .48) (10.50) . j. ϑ3 . 3) Jk = (ϑk = 1. = − k m Km (10.10.1: Natural Representation of the Lorentz Group Using (10.  (10. k j j = 1. other five parameters zero) .

ϑ. however. constitutes an overcomplete parametrization of the rotations (see Chapter 5). + The commutation relationships imply that the algebra of the generators Jk . and using the relationship Jk = 0 0 0 Lk (10.52) 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. 3 (10.56) = 1 0 0 1 = 1 1 (10.55) L 0 0 0 0  exp (w1 K1 ) = exp   Using the idempotence property 0 −1 −1 0 2 0 0 1 0 (10. 0  1 n (10. i.53) Here R(ϑ) are the 3 × 3–rotation matrices constructed in Chapter 5. k = 1.57) . A boost in the x1 –direction is L = exp(w1 K1 ). 2.e. 10. We consider now the Lorentz transformations for ϑ = 0 which are referred to as ‘boosts’. L(ϑ. w = 0) = 0 0 0 R(ϑ) . 2π[ . (10.49) define the Lie algebra associated with the Lie group L↑ . In analogy to equation (5. Following the treatment of the rotation group SO(3) one can express the elements of L↑ through + the exponential operators L(ϑ.49.292 as can be readily verified. 2. w ∈ R3 3 k=1 wk Kk .. k = 1.1: Demonstrate the commutation relationships (10. 3 is closed.51) One can readily show. For the parameters ϑk of the Lorentz transformations holds obviously ϑk ∈ [0. Exercise 7.35) 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   .50). 3 where we have defined ϑ · J = k=1 ϑk Jk and w · K = following the algebra in Chapter 5. (10. w) = exp ϑ · J + w · K . Kk . that the transformations (10.51) for w = 0 correspond to rotations of the spatial coordinates.54) which. Relativistic Quantum Mechanics The commutation relationships (10.

for wk wk ∈ ] − ∞.59) follows.6. 10. however.10. (10.59) for the case k = 1 corresponds to the relative velocity of two frames of reference.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 . consistent with (10.58) = cosh w1 1 + sinh w1 1 0 −1 −1 0 The conventional form (10. Accordingly. We expect that vk can only assume values less than the velocity of light c which in the present units is c = 1. 1[.59) cosh w1 Using cosh2 w1 − sinh2 w1 = 1 one can identify sinhw1 = sinh2 w1 + 1. in fact.3. cosh2 w1 − 1 and coshw1 = cosh2 w1 − 1 = cosh w1 sinh w1 sinh2 w1 + 1 . From (10. 10.1).64) We note that the range of wk -values is not a compact set even though the range of vk -values is compact. vk defined in (10. 10. Correspondingly.59) can be written  1 2 1 − v1 √−v1 2 1 − v1 2 1 − v1 .51) the explicit transformation for space–time–coordinates is then x1 = x1 − v1 t 1 − 2 v1 . we can state vk ∈ ] − 1. (10. one obtains from (10.59). x3 = x3 (10.59) v1 = These two equations yield cosh w1 = 1/ and (10. x2 = x2 . 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.56.61) √ √−v1 √   exp (w1 K1 ) =    2 1 − v1 1 2 1 − v1 0 0 0 0 0 0  0 0    1 0  0 1  (10. This property is.1) of the Lorentz transformations is obtained through the parameter change sinh w1 v1 = = tanh w1 (10.63) which agrees with (10. t = t − v1 x1 2 1 − v1 . The range of the parameters wk can now be specified. ∞[ . (10.60) sinh w1 = v1 / 2 1 − v1 .62) According to (10. . (10.

not any physical property f ∈ R is a scalar. J) (10. Nevertheless. Some of these r ˆ quantities were actually defined with respect to representations of the rotation group in function spaces. t) are the charge density and the current density. E = √ m mv . a1 . These quantites always come as four components (a0 . finally. 4-Vectors The quantities with the transformation behaviour like that of the position–time vector xµ defined in (10. a3 )T and transform according to a µ = Lµ ν aν . i. 2 |ˆ1 . f = f. the z–component x3 of a particle. Another example is the potential 4-vector Aµ = (V.68) where ρ(r. However.67) the transformation behaviour of which we will demonstrate further below. scalars like r = x2 + x2 + x2 . we have not yet defined representations of Lorentz transformations beyond the ‘natural’ representation acting in the 4–dimensional space of position– and time–coordinates. Hence. t) and A(r. namely. . A third 4-vector is the so-called current vector J µ = (ρ. the square of the electric field |E(r. t)|2 or the scalar product r1 · r2 of two particle positions. p = √ 2 1−v 1 − v2 (10. (10. not in the so-called natural representation associated with the 3–dimensional Euclidean space E3 . r2 ) and. Presently. our definition of quantities with special properties under Lorentz transformations presently is confined to the natural representation. tensor operators Tkm .4). spherical harmonics Y m (ˆ). Counterexamples are the energy.69) where V (r. t) are the electrical and the vector potential of an electromagnetic field. the charge density. total angular momentum r 1 2 3 states of composite systems like Y m ( 1 . The 4-vector character of J µ and of Aµ will be demonstrated further below. of a system of charges.. Such quantities appear in the description of physical systems and statements about transformation properties are often extremely helpful and usually provide important physical insight. a2 .294 Relativistic Quantum Mechanics 10. Scalars The quantities with the simplest transformation behaviour are so-called scalars f ∈ R which are invariant under transformations.3) are the so-called 4–vectors aµ . (10.2 Scalars. p) . We will see below how true scalars under Lorentz transformations can be constructed. x3 )T . We have encountered examples in connection with rotational transformations.66) Examples of 4-vectors beside xµ are the momentum 4-vector pµ = (E. respectively.65) An example is s2 defined in (10. x2 . vectors like r = (x1 . A) (10. t) and J(r. we will encounter an impressive example of physical properties.e. another example is the rest mass m of a particle. 4–Vectors and Tensors In this Section we define quantities according to their behaviour under Lorentz transformations.

66) together with (10.e. We do not fact.72) where aν is a vector with transformation behaviour as stated in (10. the relationship being aµ = gµν aν = (a0 . We have duplicated the expression for the inverse of Lµ ν to obtain the correct notation in terms of covariant.10.78) To prove the 4-vector property of ∂µ we will show that g µν ∂ν transforms like a contravariant 4vector. (10. We like to point out that from definition (10. Multiplication (and summation) of x µ = Lµ ν xν by Lρ σ gρµ yields. In fact.12) a gµν b µ ν = Lµ ρ gµν Lν σ aρ bσ = aρ gρσ bσ (10.72) of the covariant 4-vector follows = In µν and g µ as well.e.2: Scalars. (10.12).79) This is the inverse Lorentz transformation consistent with (10. Covariant 4-vectors transform like a µ = gµν Lν ρ g ρσ aσ where we defined g µν = gµν .17) Lµ σ is the transformation inverse to Lσ µ ..71) Contravariant and Covariant 4-Vectors It is convenient to define a second class of 4-vectors.74) g µν aν .17) can be written (L−1 )ν µ = Lµ ν . g µν ∂ν = Lµ ρ g ρσ ∂σ . i.. We start from x µ = Lµ ν xν . In any case one can express (10. i. ∂t (10. one can employ the tensors g µν to raise and lower indices of L ν establish here the consistency of the ensuing notation. This property follows from (10.76) An important example of a covariant 4-vector is the differential operator ∂µ = ∂ = ∂xµ ∂ . and .16). xν = g νσ Lρ σ gρµ x µ . The respective vectors aµ are associated with the 4-vectors aµ . aµ (10. The 4-Vector ∂µ (10.77) The transformed differential operator will be denoted by ∂µ = def ∂ .73) aµ = Lµ σ aσ . (10. −a1 .66). If aµ and bµ are 4-vectors aµ gµν bν (10. ∂x µ (10. 4–Vectors and Tensors Scalar Product then 295 4-vectors allow one to construct scalar quantities. using (10.73) Note that according to (10. −a2 . gσν xν = Lρ σ gρµ x µ and g µσ gσν = δ µ ν . accordingly. lower. −a3 ) (10.70) is a scalar. one can express [(L−1 )T ]µ ν = (L−1 )ν µ and.75) (10. One calls 4-vectors aµ covariant and 4-vectors aµ contravariant.

d’Alembert Operator We want to construct now a scalar differential operator..87) The remaining components of pµ can be written. (10.86) One can write (10.e.88) One can express then the momentum vector pµ = √ d m dxµ = m xµ .83) is a scalar under Lorentz transformations. Using the chain rule of differential calculus we obtain 3 ∂µ = ν=0 ∂xν ∂ = g νσ Lρ σ gρµ ∂ν = Lµ ν ∂ν ∂x µ ∂xν (10.89) . i.66). (10.79) allows us to determine the connection between ∂µ and ∂µ ..84) = 1 − (v)2 (10. upper. For this purpose we consider the scalar differential (dτ )2 = dxµ dxµ = (dt)2 − (dr)2 It holds dτ dt from which follows 2 (10. Proof that pµ is a 4-vector We will demonstrate now that the momentum 4-vector pµ defined in (10. p1 = √ m v1 m dx1 = √ .85) 1 d d = √ . 1 − v2 1 − v 2 dt (10.296 Relativistic Quantum Mechanics contravariant. i.67) transforms like (10. dτ 1 − v 2 dt p0 = E = √ m m dt = √ . For this purpose we define first the contravariant differential operator ∂ µ = g µν ∂ν = Then the operator 2 ∂µ ∂ µ = ∂t − 2 (10. In fact.e.g.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 . 2 dt dτ 1 − v (10. e. this operator is equal to the d’Alembert operator which is known to be Lorentz-invariant. 2 2 dt 1 − v 1 − v (10.81) ∂ .− ∂t . indices.82) (10.. ∂µ does indeed transform like a covariant vector.

t) and A(r. and had chosen the so-called Coulomb gauge (8.97) and.67) E2 − p 2 = or pµ pµ = m2 which. (10. 10. This gauge is not Lorentz-invariant and we will adopt here another gauge. for these potentials. a contravariant 4-vector then the l.69) is. in fact. Lorentz-invariant.91) (10.3: Relativistic Electrodynamics 297 d The operator m dτ transforms like a scalar.69) for the electrodynamic potential and the 4-vector derivative (10. t) = 0 .e.77) which allow one to express (10.95) 2 p This obviously describes the energy of a free particle with rest energy ±m. in (10.89) must be a 4-vector.h. If we can show that Aµ defined in (10. m2 m2 v 2 − = m2 1 − v2 1 − v2 (10.. One obtains E = ±m ± p2 2m (p 2 )2 + O 4m3 (p 2 )3 4m5 .12).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. of (10. The momentum 4-vector allows us to construct a scalar quantity.h.s. t) + · A(r.97) We have proven already that ∂µ is a contravariant 4-vector.96) is a scalar and. hence. (10. is a scalar. namely pµ pµ = pµ gµν pν = E 2 − p 2 Evaluation of the r. pµ on the l. kinetic energy ± 2m and relativistic corrections. hence.90) (10. . i. Since xµ transforms like a contravariant 4-vector. in (10. of (10.89) alltogether transforms like a contravariant 4-vector. t).h. (10. We will demonstrate now the 4-vector property of Aµ . in fact. Lorentz Gauge In our previous description of the electrodynamic field we had introduced the scalar and vector potential V (r.96) The Lorentz-invariance of this gauge. yields according to (10.s. the r. equivalently.94) 1 m In the non-relativistic limit the rest energy m is the dominant contribution to E. and.h. · A = 0.10. respectively. We like to rewrite the last result E 2 = p 2 + m2 or E = ± p 2 + m2 .93) (10. 8. the so-called Lorentz gauge.92) (10.s. can be demonstrated readily using the 4-vector notation (10.s. ∂t V (r. namely. Expansion in should then be rapidly convergent.96) in the form ∂µ Aµ = 0 .

(10.e. 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   . according to (10.101) Combining equations (10..96). m. (10. must transform likewise. t) − 2 V (r. also the l.9). We want to derive now the differential equations which determine the 4-potential Aµ in the Lorentz gauge (10. (10. of (10.77) and (10.h. k. prove that Aµ is. n = 1. yields ∂µ ∂ µ Aν (xσ ) = 4 π J ν (xσ ) . · A(r. i. t) = −∂t V (r. t) = 0 (10. The l. t) 2 ∂t A(r.6) and (8. 10. 3 (10.103) B .s. a 4-vector.69) and ∂ µ in (10. t) .. has to transform like a contravariant 4-vector.97) and. transforms like a 4-vector. Since ∂µ ∂ µ transforms like a scalar one can conclude that Aν (xσ ) must transform like a 4-vector.s. The respective equation for A0 = V can be obtained from Eq. . (8. 2.13). Consequently. (10. it follows that J µ .17) using the identity (8. (10. t) + · J(r. using (10. t) from (8. t) = 4πρ(r.32). F mn = − mn (10.68). F µν =  x  Ey B z 0 −Bx  Ez −By Bx 0 Alternatively. t) − 2 A(r.104) where mn = mn is the totally anti-symmetric three-dimensional tensor defined in (5. one obtains 2 ∂t V (r. that F µν can be expressed through the potential Aµ in (10.s.105) .e.h. one obtains for A(r.298 Transformation Properties of J µ and Aµ Relativistic Quantum Mechanics The charge density ρ(r. t). · A(r. t) = 4 π J(r.h. i.99) must be a scalar. t) are known to obey the continuity property ∂t ρ(r. Since ∂µ transforms like a covariant 4-vector.69).99) Since this equation must be true in any frame of reference the right hand side must vanish in all frames. thereby. using (8.83) and (10.100) Similarly. this can be stated F k0 = −F 0k = E k . using (10.98) can be written. t) and current density J(r.82) as follows F µν = ∂ µ Aν − ∂ ν Aµ . Using · ∂t A(r. must be a scalar.18) and. Equation (10. One can readily verify. in fact. t) together with (10.101). t) = −∂t V (r. ∂µ J µ (xµ ) = 0 .96). t) . in fact.100.102) In this equation the r.98) which reflects the principle of charge conservation. t) = ∂t · A(r. This principle should hold in any frame of reference.

104) establishe the transformation behaviour of E(r.4).10.103. 8.109) (10.62) or. t) and B(r.2).106) In case that the Lorentz transformation Lµ ν is given by (10. The results can be put into the more general form E B = = E . These equations show that under Lorentz transformations electric and magnetic fields convert into one another. the components of the fields parallel and perpendicular to the velocity v which determines the Lorentz transformation.112) (10. From the definition (10.113) which can be shown to be equivalent to the two homogeneous Maxwell equations (8. E⊥ = B⊥ E⊥ + v × B √ 1 − v2 B⊥ − v × E = √ 1 − v2 (10.105) and (10.63).102) ∂µ F µν = 4π J ν . .108) yields then the expressions for the transformed fields E and B . 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. equivalently. 10. by (10.3: Relativistic Electrodynamics 299 The relationships (10. respectively.1. B⊥ are. (10.97). B and E⊥ .3. Noting ∂µ F µν = ∂µ ∂ µ Aν − ∂µ ∂ ν Aµ = ∂µ ∂ µ Aν − ∂ ν ∂µ Aµ = ∂µ ∂ µ Aν .105) of the tensor F µν one can conclude the property ∂ σ F µν + ∂ µ F νσ + ∂ ν F σµ = 0 (10. one can conclude from (10.107)  y 1 z  Bz −v1 Ey √ 2  √ 2  0 −Bx   1−v1 1−v1   Ez +v1 By By +v1 Ez √ 2 − √ 2 Bx 0 1−v1 1−v1 Comparision with F µν  0 −Ex −Ey −Ez  E 0 −Bz By   =  x  Ey B z 0 −Bx  Ez −By Bx 0  (10. In a new frame of reference holds F µν = Lµ α Lν β F αβ (10. t). where we used (10. Maxwell Equations in Lorentz-Invariant Form One can express the Maxwell equations in terms of the tensor F µν in Lorentz-invariant form. B . 8.110) where E .111) One can readily prove that this equation is equivalent to the two inhomogeneous Maxwell equations (8.

dτ m √ Employing again (10.104).117) (10. We want to express this force through the tensor F µν . employing (10.93). i.5) taking the scalar product with p dp = qp · E (10.103) dpx q = ( EEx + py Bz − pz By ) . Equation (10. We have. using (10.120) is referred to as the Lorentz force.5).86). dE q = p·E dτ m or with (10.5) where p = mv/ 1 − v 2 . for µ = 1. The µ = 0 component of (10. hence. hence. e.116) (10. equivalent to (8.86) dE q = p·E . in fact. (10.114). of (10.e. To avoid √ confusion we will employ in the following for the energy of the particle the notation E = m/ 1 − v 2 [see (10.g. however.115) This equation follows.114) reads using (10. E = m/ 1 − v 2 .87)] and retain the definition E for the electric field.5). It holds for a particle with 4-momentum pµ as given by (10.300 Lorentz Force Relativistic Quantum Mechanics One important property of the electromagnetic field is the Lorentz force acting on charged particles moving through the field..114) reads then.119) which is the x-component of the equation of motion (8. provides an alternative description of the action of the Lorentz force. For the spatial components.120) (10. also from the equation of motion (8. .86) and (10. The term on the r. yields dpx = q dt Ex + (v × B)x (10. dt E From this one can conclude.s. 1 dE 2 1 dp 2 = = qp·E 2 dt 2 dt (10. We want to √ demonstrate now that this equation is equivalent to the equation of motion (8. demonstrated that (10.118) p· dt √ where we exploited the fact that according to p = mv/ 1 − v 2 holds p v.114) is.67).114) where d/dτ is given by (10.67) and charge q dpµ q = pν F µν dτ m (10..h.

i. We will denote the intended representation by L(ϑ. (10. one can evaluate Jk as follows Jk = lim and Kk Kk = lim wk →0 ϑk →0 1 ϑ1 1 w1 ρ eϑk Jk −1 1 (10. the charge-current density in case of Sect.3 or the bi-spinor wave function of electron-positron pairs in Sect. e. in as far as SO(3. w) = exp ϑ · J + w · K . def (10.122) This result gives rise to the definition of the function space representation ρ(Lµ ν ) of the Lorentz group (ρ(Lµ ν )ψ)(xµ ) = ψ((L−1 )µ ν xν ) .7.4 Function Space Representation of Lorentz Group In the following it will be required to decribe the transformation of wave functions under Lorentz transformations. w) which correspond to the generators Jk and Kk in (10. 10.125)]. (10.126) ρ ewk Kk − 1 1 . taken at the pairs of points (x µ = Lµ ν xν .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 . Functions which represent 4-vectorsor other non-scalar entities. (5. and which can be constructed following the procedure adopted for the function space representation of SO(3).125) def (10.48) of SO(3). (10. J3 ) and K = (K1 . 10.121) covers solely the transformation behaviour of scalar functions.121) by (L−1 )µ ν xν and obtains ψ (xµ ) = ψ((L−1 )µ ν xν ) . i.48) and (10. However. xµ ). We like to express now ψ (x µ ) in terms of the old coordinates xµ .e.e.. J2 . K3 ) are the generators of L(ϑ. We need to emphasize that (10.1) is a generalization (rotation + boosts) of the group SO(3). w)) = ρ eϑ·J + w·K which we present in the form L(ϑ.51) for the natural representation of the Lorentz group. In this section we will investigate the transformation properties of scalar functions ψ(xµ ).127) . we seek an expression which corresponds to (10. For such functions holds in the transformed frame ψ (Lµ ν xν ) = ψ(xµ ) (10. w) = ρ(Lµ ν (ϑ.10. K2 .124) In this expression J = (J1 .4: Function Space Representation of Lorentz Group 301 10.123) This definition corresponds closely to the function space representation (5. obey a different transformation law.42) of SO(3).47). 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...g. The resulting expression should be a generalization of the function space representation (5. w. For this purpose one replaces xµ in (10. ψ ∈ C∞ (4). Accordingly. in the present case we exclude the factor ‘−i’ [cf.

reproduces the expression for J1 in (10. x1 ∂3 ] − [x2 ∂3 . [K1 . x2 .126). K2 ] = −J3 . The generators J . which we like to demonstrate for J1 and K1 .128) eϑ1 J1 = e−ϑ1 J1 1  0 =   0 0   0 0 0 1 0 0   0 cosϑ1 sinϑ1  0 −sinϑ1 cosϑ1 (10. (10. J ] [ Kk .302 One obtains J1 = x3 ∂2 − x2 ∂3 . J2 ] = J3 . the expression for K1 in (10. One can determine similarly K1 starting from coshw1 sinhw1  sinhw1 coshw1 =   0 0 0 0  0 0 1 0  0 0   . cosϑ1 x2 + sinϑ1 x3 .132) and. K obey the same Lie algebra (10. −sinϑ1 x2 + cosϑ1 x3 ) = ψ(xµ ) + ϑ1 (x3 ∂2 − x2 ∂3 ) ψ(xµ ) + O(ϑ2 ) .129) which yields for small ϑ1 ρ eϑ1 J1 ψ(xµ ) = ψ(x0 . obviously. J3 = x2 ∂1 − x1 ∂2 . x3 ) 2 = ψ(xµ ) + w1 (x1 ∂0 + x0 ∂1 ) ψ(xµ ) + O(w1 ) (10. and [J1 . 0  1 (10. namely [J1 . K2 ] = K3 : [ J1 . In order to evaluate (10.134) .e. i.130) ew1 K1 −1 = e−w1 K1 (10.49) as the generators of the natural representation. 1 This result. obviously. J2 = x1 ∂3 − x3 ∂1 .128). (10. J2 ] = [x3 ∂2 − x2 ∂3 . K ] = = k m Jm k m Jm = − k m Km . x1 . sinhw1 x0 + coshw1 x1 .131) This yields for small w1 ρ ew1 K1 ψ(xµ ) = ψ(coshw1 x0 + sinhw1 x1 . x1 ∂3 − x3 ∂1 ] = [x3 ∂2 . [ Jk . x3 ∂1 ] = −x1 ∂2 + x2 ∂1 = J3 .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 (10.133) We demonstrate this for three cases. K ] [ Jk .

One-Dimensional Function Space Representation The exponential operator (10. a relationship which can also be stated R = and + − (x0 )2 − (x3 )2 (x0 )2 − (x3 )2 if x0 ≥ 0 if x0 < 0 (10. x2 + x2 + x2 1 2 3 in the case of SO(3) and (x0 )2 − (x3 )2 in case of transformation (10. x0 ∂2 ] = x3 ∂0 + x0 ∂3 = 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 .128). x2 ∂0 ] − [x1 ∂0 ..137) where K3 is given in (10.142) (10.139) x3 = R sinhΩ (10.140) x0 x The transformation to hyperbolic coordinates closely resembles the transformation to radial coordinates for the generators of SO(3) in the function space representation [cf. ∂3 in terms of ∂R . In both cases the radial coordinate is the quantity conserved under the transformations. ∂x0 R and ∂Ω ∂x3 ∂Ω ∂x0 The chain rule yields then ∂0 ∂3 ∂R ∂ ∂Ω ∂ x0 ∂ 1 ∂ + = − sinh2 Ω 3 0 ∂R 0 ∂Ω ∂x ∂x R ∂R x ∂Ω ∂R ∂ ∂Ω ∂ x3 ∂ 1 ∂ = + = − + cosh2 Ω 0 .10.139. The relationships (10. Ω which are connected with x0 .85-5.137).125) in the case of a one-dimensional transformation of the type L(w3 ) = exp w3 K3 .141) (10. In the following we consider solely the case x0 ≥ 0. x0 ∂2 + x2 ∂0 ] = [x0 ∂1 . Eqs. 0 ∂cothΩ ∂x x = ∂R x0 = − ∂x3 R (10. 303 (10. i. cothΩ = 3 . For this purpose one expresses K3 in terms of hyperbolic coordinates R. x0 ∂2 + x2 ∂0 ] = [x3 ∂2 . ∂Ω . (10.135) (10. 10. [ J1 . K2 ] = [x0 ∂1 + x1 ∂0 .138) x3 x0 . can be simplified considerably. x0 ∂2 ] = −x2 ∂1 + x1 ∂2 = − J3 . x2 ∂0 ] − [x2 ∂3 . We note tanhΩ = ∂R x0 = . x3 as follows x0 = R coshΩ .136) (10.140) allow one to ∂ ∂ express the derivatives ∂0 .e.87)].4: Function Space Representation of Lorentz Group [ K1 . K2 ] = [x3 ∂2 − x2 ∂3 . (5.143) .

146) Applying the correspondence principle to (10. (10.122). momentum. in 4-vector notation reads pµ =⇒ pµ = i (∂t . therefore.128) yields K3 = x0 ∂3 + x3 ∂0 = ∂ . In natural units the Klein–Gordon equation (10.92) one obtains the wave equation − or 2 2 µ ∂ ∂µ ψ(xν ) = m2 ψ(xν ) ψ(xν ) = 0 .145) ∂Ω 10. representation theoretic treatment. The latter property implies that upon change of reference frame ψ(xµ ) transforms according to (10. In the position representation the correspondence principle states ˆ E =⇒ E = − ∂t i ˆ p =⇒ p = i which. (length)−1 .151) reads ∂µ ∂ µ + m2 ψ(xµ ) = 0 (10. ) = i ∂µ .144) The action of the exponential operator (10. ∂Ω Relativistic Quantum Mechanics (10.149) where ψ(xµ ) is a scalar. which allows one to replace classical observables by quantum mechanical operators acting on wave functions.150) . namely the Klein–Gordon and the Dirac equations. 10. however.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. is not very insightful. but certainly is short and.148) ∂ µ ∂µ + m2 (10. The historic route to these two equations. We will provide a ‘derivation’ of these two equations which stem from the historical development of relativistic quantum mechanics.151) is called the Klein-Gordon equation. Free Particle Case A quantum mechanical description of a relativistic free particle results from applying the correspondence principle.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 ) . − ) = i ∂ µ . complex-valued function.121. The partial differential equation (10. mass. (10. Further below we will provide a more systematic. In the following we will employ so-called natural units = c = 1. extremely useful. and (time)−1 all have the same dimension.304 Inserting these results into the definition of K3 in (10. In these units the quantities energy. ˆ (10.147) (10. ˆ pµ =⇒ pµ = i(∂t .

154) describes the current density connected with motion of the particle probability distribution.150) becomes ∂µ ∂ µ + m2 ψ (x µ ) = 0 (10. t) = · jKG (r. t) ψ(r. t) + where ρS (r. (10. t) = − 1 2m 2 ψ(r. In order to obtain the conservation law connected with the Klein–Gordon equation (10. Under Lorentz transformations the free particle Klein–Gordon equation (10. t) ψ(r.155) describes the positive definite probability to detect a particle at position r at time t and where jS (r.150) is invariant under Lorentz transformations. t) − ψ(r.153) is associated with a conservation law for particle probability ∂t ρS (r. t) ) . t)∂t ψ(r. t) = (10. t) ) 2m 1 = ( ψ ∗ (r.10. t) jKG (r. t) ψ ∗ (r. t) ψ(r. and that (as postulated) ψ(xµ ) is a scalar. t) 1 [ ψ ∗ (r. t) (10. This follows from the fact that ∂µ ∂ µ and m2 are scalars. t)∂t ψ ∗ (r. t) − ψ(r.156) · jS (r.150) one considers Im ψ ∗ ∂µ ∂ µ + m2 ψ = 0 (10. t) = 0 i ( ψ ∗ (r. Current 4-Vector Associated with the Klein-Gordon Equation As is well-known the Schr¨dinger equation of a free particle o i∂t ψ(r. 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 (10.159) (10. t) = 0 (10.161) .151) One can notice immeadiately that (10.157) 2m which is equivalent to (10. t) + where ρKG (r. 2mi (10.152) which has the same form as the Klein–Gordon equation in the original frame. t) ] 2mi (10. t) = ψ ∗ (r.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.158) which yields 2 2 ψ ∗ ∂t ψ − ψ∂t ψ ∗ − ψ ∗ 2ψ +ψ ψ∗ − 2ψ∗ = ψ) = 0 (10.154).

with p = 0. 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) . λ = ± .p ei(p·r − λEo (p)t) Eo (p) = The spectrum of the Klein–Gordon equation (10.161) and (10. λ) actually describes charge density rather than probability.e. o namely. ±) = ± 2 m2 + po (10. p0 . λ|xµ ) = 0 2 m2 + po . not of particle probability. λ|xµ )ψo (p. λ) = λ ψo (p. is actually that of a charge density. λ) associated with the corresponding wave functions ψo (p. (10. λ|xµ ) according to (10. t). λ|xµ ) = Nλ. the Klein-Gordon equation is considered as a suitable equation to describe spin–0 particles.93] dispersion relationship connecting E0 . λ = ± (10. (10. according to (??) is decribed by the r–independent wave function ψo (p = 0..164) The corresponding energy is Eo (po . the anti-particles carry a charge opposite to that of the associated particles.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. m 2 2 E0 = m2 + po . λ|xµ ) = N e−iλmt . corresponding to λ = −.166) ψo (p. it had been realized later. The density ρKG (p. t) = 0 (10.150) is a continuum of positive energies E ≥ m. in that the expression for ρKG is not positive definite. and the density ρKG (p.162) Inserting this into the Klein–Gordon equation (10.306 Relativistic Quantum Mechanics This conservation law differs in one important aspect from that of the Schr¨dinger equation (10.167) m which is positive for λ = + and negative for λ = −.151) yields 2 2 Eo − p0 − m2 ψ(r.166) is Eo (p) ∗ ρKG (p.162) shows that the solutions of the free particle Klein-Gordon equation (10. for example pions. The proper interpretation of ρKG (r.163) which results in the expected [see (10. Generating a Solution Through Lorentz Transformation A particle at rest. We denote this by summarizing the solutions as follows ∂µ ∂ µ + m2 ψo (p. i. λ|xµ ) (10. corresponding to λ = +. The proper interpretation of the two cases is that the Klein–Gordon equation describes particles as well as anti-particles.150) are actually determined by po and by the choice of sign ±. and of negative energies E ≤ −m. Today. (10.168) .165) This result together with (10.

s. λ = +|xµ ) = N ei(px 3 − Eo (p)x0 (10.174) which agrees with the expression given in (10.. A(r.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.173) (10. i. Ω. t).177) . t)) . (10. in fact.168). λ = −|xµ ) = N ei(px 3 + Eo (p)x0 . one has to interprete p = mv/ 1 − v 2 (10. (10.e..175) as the momentum of the particle and one obtains L(w3 )ψo (p = 0.169) −iλ ( m coshw3 ) ( R coshΩ ) − iλ ( m sinhw3 ) ( R sinhΩ ) . Such wave function should be generated by applying the Lorentz transformation in the function space representation p (10. using (10. In case of λ = + one obtains finally L(w3 )ψo (p = 0.168). In case of λ = −. i. 2 2 1−v 1−v (10.138) and the relationships (10.f. for positive energy solutions.145) choosing m = sinhw3 . λ|xµ ) = = exp w3 ∂ ∂Ω N e−iλmRcoshΩ 3) N e−iλmRcosh(Ω+w .166)] of the particle.138) to replace t = x0 by hyperbolic coordinates R. −A(r. (10.6: Klein–Gordon Equation with Electromagnetic Field 307 We want to demonstrate now that the wave functions for p = 0 in (??) can be obtained through appropriate Lorentz transformation of (10. t).61) yield for this expression −iλ √ m mv x0 − iλ √ x3 .10. For this purpose we consider the wave function for a particle moving with momentum velocity v in the direction of the x3 –axis. This yields.h. for negative energy solutions.171) One can interpret then for λ = +. L(w3 )ψo (p = 0. of (10.172) as the energy [c. (10. for the wave function (10.170) The coordinate transformation (10.176) 10.e.166).169) The addition theorem of hyperbolic functions cosh(Ω + w3 ) = coshΩ coshw3 + sinhΩ sinhw3 allows us to rewrite the exponent on the r. Aµ (xµ ) = (V (r. p = − mv/ 1 − v 2 as the momentum of the particle relative to the moving frame and √ m = 1 − v2 m2 = 1 − v2 m2 + m2 v 2 = 1 − v2 m2 + p2 = Eo (p) (10. t)) (10.

In terms of space-time derivatives this reads (i∂t − qV (r. t) 2 (10.180) + m2 ψ(r.308 free classical particle classical particle in field (V.150) −g µν (i∂µ )(i∂ν ) + m2 ψ(xµ ) = 0 .. −A).e. i. i∂t and −i in (10.e.. in keeping withnthe non-relativistic limit. that the mass m of the particle.69)] one replaces the quantum mechanical operators. (10. According to the so-called minimum coupling principle the presence of the field is accounted for by altering energy. 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 this becomes g µν (i∂µ − qAµ )(i∂ν − Aν ) ψ(xµ ) = m2 ψ(xµ ) which can also be written g µν πµ πν − . According to the principle of minimal coupling [see (10. (10. we define ψ(r.147).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. See also Eq. A) or Aµ = (V. We will also assume. and focus on the remaining part of the wave function.178) According to the replacements in Table 10. A) free quantum particle quantum particle in field (V.150). which couples a particle of charge q to an electromagnetic field described through the potential Aµ (xν ).179) (10.e. (10. For this purpose one writes the Klein-Gordon equation (10. in . i. m2 ) ψ(xµ ) = 0 .182) and seek an equation for Ψ(r. 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. (10. t) = e−imt Ψ(r.93). according to the rules shown in Table 10. t). momenta for classical particles and the respective operators for quantum mechanical particles in the manner shown.1. t))2 ψ(r. t) . albeit in a form.. it’s rest energy.1: Coupling of a particle of charge q to an electromagnetic field described by the 4-vector potential Aµ = (V. is much larger than all other energy terms. For this purpose we split-off a factor exp(−imt) which describes the oscillations of the wave function due to the rest energy. t) = −i − q A(r. i. t) .

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

193) (10.191) m (θ. (10.194) and (10. φ) ..192) ˆ are spherical harmonics.199) ( + 1) − Z 2 e4 . 2. . this quantity definitely must be an integer. The similarity of (10.198) . .e. .310 Relativistic Quantum Mechanics Because of the radial symmetry of the Coulomb potential we express this equation in terms of spherical coordinates r. φ) = ( + 1) Y m (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. n − 1 .. (10. φ) R (r) Y r m (θ. (10. 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. . n = 1. The Laplacian is 2 = ˆ 1 2 1 1 1 2 L2 2 ∂r r + 2 2 ∂θ sinθ∂θ + 2 2 ∂φ = ∂r r − 2 .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 (10.195) The latter problem leads to the well-known spectrum En = − mπ (Ze2 )2 . θ.197) counts the number of nodes of the wave function.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 . 2n2 = 0.196) In this expression the number n defined through n = n− −1 (10. i. . (10. . (10.e. . the eigenfunctions of the operator L2 ˆ L2 Y m (θ.192) leads then to the ordinary differential equation d2 − dr2 ( + 1) − Z 2 e4 2 Ze2 + + r2 r 2 − m2 π R (r) = 0 . r r r r sin θ r sin θ Ze2 2 r ) (10. 1. φ. i.

. 10. . (10. . Introducing α = Z 2 e4 and 1 β = + 2 (10. . . m) = mπ 1 + Z 2 e4 (n− − 1 2 + ( + 1 )2 − Z 2 e4 )2 2 n = 1. .196) if one makes the replacement 2 values which can be obtained E −→ One obtains − m2 π . n = 0.194) d2 λ( ) ( λ( ) + 1 ) 2 Ze2 − + + dr2 r2 r 2 311 − m2 π R (r) = 0 . . which corresponds to a bound = 0. i. EKG = Using (10. 2.204) In order to compare this result with the spectrum of the non-relativistic hydrogen-like atom we expand in terms of the fine structure constant e2 to order O( 8 ). . + (10. . (10. . .199) and defining EKG (n.197. . 1. . 1. n − 1 m = − . .10.e.200) The bound state solutions of this equation should correspond to from (10.202) Solving this for state) yields (choosing the root which renders = 1 + mπ Z 2 e4 ( n + λ( )+1 )2 ≤ mπ . . 2mπ 2 (n + λ( ) + 1)2 (10.205) α √ 2 2 1+ (n − β + β −α) and one obtains the series of approximations 1 1+ ≈ ≈ ≈ 1+ ≈ 1+ (n − β + α √ β 2 −α)2 1 1+ (n − α 2β α + O(α2 ))2 1 1+ n2 − α n + O(α2 ) β α 1 α n2 + α2 βn3 + O(α3 ) − α2 8n4 1 α 2n2 + α2 2βn3 + O(α3 ) . . . = 0. . − + 1. . 2mπ 2 −→ λ( ) . results in the (10.204) reads 1 (10. . 2 e2 −→ e2 mπ .203) spectrum .6: Klein–Gordon Equation with Electromagnetic Field and one can write (10.. 1.201) mπ Z 2 e4 m2 − m2 π π = − .

It must be pointed out here that does not denote energy.209) which.208). but not an operator iγ · as suggested by the principle of relativity (equivalence of space and time). The latter term agrees with observations of pionic atoms. t) = ± m2 − 2 Ψ(r. it does not agree with observations of the hydrogen spectrum. Equations (10. a feature which is connected with the 2nd order time derivative in this equation. His solution .7 The Dirac Equation Historically. m) ≈ m − mZ 2 e4 mZ 4 e8 − 2n2 2n3 1 + − 1− Relativistic Quantum Mechanics α α2 α2 α2 − + + + O(α3 ) . Finally. i.. a splitting of the six n = 2. It should feature then a differential operator of the type D = iγ µ ∂µ .204) EKG (n. i.e. 10. however. Many attempts were made by theoretical physicists to ‘linearize’ the square root operator in (10. is related to the equation E 2 = m2 + p 2 of the classical theory which involves a term E 2 .. the Dirac equation introduced below. This derivative. because of the equivalence of space and time coordinates in the Minkowski space such equation necessarily can only have then first order derivatives with respect to spatial coordinates. one with a positive-definite density. 10.207) Here the first term represents the rest energy. in turn. for example. Dirac succeeded. arises because the Klein–Gordon equation.e. but for a long time to no avail. the Klein–Gordon equation had been rejected since it did not yield a positive-definite probability density. 4n (10.. are unsatisfactory since expansion of the square root operator involves all powers of the Laplace operator. Also. t) (10.209). through the correspondence principle.208.209). . = 1 states into groups of two and four degenerate states. Application of the correspondence principle (10. i.208) These two equation can be combined i∂t + m2 − 2 i∂t − m2 − 2 Ψ(r. in fact. 10. would have only a first order time derivative.206) 1 2 3 + O(Z 6 e12 ) .208. the π − meson is a pseudoscalar particle.2 particles. t) . A more satisfactory Lorentz–invariant wave equation.146) leads to the wave equation i∂t Ψ(r. (10. Heuristic Derivation Starting from the Klein-Gordon Equation An obvious starting point for a Lorentz-invariant wave equation with only a first order time derivative is E = ± m2 + p 2 .e. However. but in the present case rather the negative of the energy. is identical to the two equations (10. and the third term gives the leading relativistic correction.312 ≈ From this results for (10. the second term the non-relativistic energy. the wave function changes sign under reflection. however. 2n2 2βn3 8n4 4n4 (10. In order to describe electron spectra one must employ the Lorentz-invariant 1 wave equation for spin. The latter spectrum shows.

unspecified algebraic objects γ µ and γ ν .209). 2. Properties of the Dirac Matrices Let us now trace Dirac’s steps in achieving the linearization of the ‘square root operator’ in (10. one route to important discoveries in physics is the creation of new mathematical descriptions of nature.1 2 and anti-particles. one can impose the condition γ0 is hermitian . this was not so. j = 1. 2. . namely. For µ = ν condition (10. the quantities γ 0 . Initially. spin.212) where we have changed ‘dummy’ summation indices. γ 3 can only be realized by d×d–matrices requiring that the wave function Ψ(xµ ) is actually a d–dimensional vector of functions ψ1 (xµ ). γ ν ]+ = 2 g µν (10. achieved through a beautiful mathematical theory.211) into (10. γ j . i. 2. .215) . yet to be discovered. therefore. are associated with a positive-definite particle density. ψ2 (xµ ). this would lead to the two wave equations (P − m)Ψ = 0 and (P + m)Ψ = 0 which have a first order time derivative and.7: Dirac Equation 313 to the problem involved an ingenious step. 3 . The discovery by Dirac. In fact. (10. γ 1 .66). so far. according to (10. 3 has imaginary eigenvalues ±i. strengthens the believe of many theoretical physicists today that the properties of physical matter ultimately derive from a. Instead. . j = 1. γ 2 .211) Obviously..210) where P = iγ µ ∂µ . 3 are anti-hermitian . However. We seek to identify the coefficients γ µ . Comparing the left-most and the right-most side of the equations above one can conclude the following property of γµ γ µ γ ν + γ ν γ µ = [ γ µ . It turns out that no 4-vector of real or complex coefficients can satisfy these conditions. but did not commute the. Inserting (10.213) reads (γ µ )2 = 1 µ = 0 −1 µ = 1. beautiful mathematical theory and that. it was assumed that the 4-dimensional space introduced by Dirac could be linked to 4vectors. ψd (xµ ). the operator of the Klein–Gordon equation ∂µ ∂ µ + m2 = −( P + m ) ( P − m ) (10. the 4-dimensionsional representation discovered by Dirac involved new physical properties.208). therefore. Accordingly.213) We want to determine now the simplest algebraic realization of γ µ . (10. quantities with the transformation law (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. Starting point is to boldly factorize.e.214) From this follows that γ 0 has real eigenvalues ±1 and γ j . exploited ∂µ ∂ν = ∂ν ∂µ . these descriptions ultimately merging with the true theory of matter.10. (10. the realization that the linearization can be carried out only if one assumes a 4-dimensional representation of the coefficients γ µ .

213) reads γ µ γ ν = −γ ν γ µ . γj = 0 −σ j σj 0 .220) where Ψ(xµ ) represents a 4-dimensional wave function. 1 (10. (10. namely the Pauli matrices σ 1 . We will argue further below that the choice f γ µ .224) . This property can also be written (γ µ )† = γ 0 γ µ γ 0 . rather than a scalar wave function.215). σ j σ k = − σ k σ j for j = k . to construct four matrices γ µ for the next possible dimension d = 4. For d = 2 there exist only three anti-commuting matrices.213) is satisfied.219) Using property (10. however. A proper choice is γ0 = 1 1 0 0 −1 1 . the γ µ are anti-commuting. (10. One can readily show using the hermitian property of the Pauli matrices (γ 0 )† = γ 0 and (γ j )† = −γ j for j = 1.e.. ψ3 . ψ4 ) and where ∂µ denotes the differential operator ∂ µ operating to the left side.222) ∗ ∗ ∗ ∗ where we have defined Ψ† = (ψ1 . The Adjoint Dirac Equation The adjoint equation is Ψ† (xµ ) i(γ µ )† ∂µ + m ← ← (10. 2.223) Inserting this into (10.218) of the Pauli matrices one can readily prove that condition (10. holds 2 σj = 1 .221) =0 (10.218) The Pauli matrices allow one. in fact. Relativistic Quantum Mechanics (10. ψ2 . (10. is implied by (10. as long as det(γ µ ) = 0 the dimension d of the square matrices γ µ must be even so that (−1)d = 1. 3 which. From this equation one can conclude that also the following should hold ( iγ µ ∂µ − m ) Ψ(xµ ) = 0 which is the celebrated Dirac equation. is unique. in fact. σ 3 for which. (10. σ 2 .217) Obviously. 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 (10.314 For µ = ν (10.222) and multiplication from the right by γ 0 yields the adjoint Dirac equation Ψ† (xµ ) γ 0 iγ µ ∂µ + m ← =0. except for similarity trasnformations. rather than to the right side. 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 ) .216) i.

2.225) . . αj = ˆ 0 σj σj 0 .229) The similarity transformation (10.1: Derive (refeq:Dirac-intro20a) from (10.226) (10. 3 and ˆ β = 1 1 0 0 −1 1 . (10..221) from the left by γ 0 ˆ ˆ i ∂t + i α · − β m Ψ(r. γ ˜ Exercise 10.227). and γ0 = ˜ 0 1 1 1 0 1 .Chiral Representation 315 The Dirac equation can be subject to any similarity transformation defined through a non-singular ˜ 4 × 4–matrix S. γj = ˜ 0 σj 1 S = √ 2 −σ j 0 1 1 1 1 1 −1 1 1 (10. i.227) A representation often adopted beside the one given by (10. 2. (10.213) still holds. (10.233) The eigenstates and eigenvalues of H correspond to the stationary states and energies of the particles described by the Dirac equation.e.230) Schr¨dinger Form of the Dirac Equation o Another form in which the Dirac equation is used often results from multiplying (10. 3 .228) (10.232) This form of the Dirac equation is called the Schr¨dinger form since it can be written in analogy o to the time-dependent Schr¨dinger equation o i∂t Ψ(xµ ) = Ho Ψ(xµ ) . ˆ ˆ ˆ Ho = α · p + β m . (10. j = 1. j = 1. 3 . 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 ˜ (10.219) is the socalled chiral representation defined through ˜ Ψ(xµ ) = S Ψ(xµ ) . 2. (10.222. 10.7.7: Dirac Equation Similarity Transformations of the Dirac Equation .10.213). [˜ µ .227) leaves the algebra of the Dirac matrices unaffected and commutation property (10. γ ν ]+ = 2 g µν . j = 1.231) ˆ whereα has the three components αj . t) = 0 (10.

Good. in case of C4 .Phys. There are (including the different signs) 32 elements in G4 which we define as follows Γ±1 = ±1 1 Γ±2 = ±d1 .316 Clifford Algebra and Dirac Matrices The matrices defined through dj = iγ j .213). The representations of Clifford algebras Cm are well established. Γ±15 = ±d2 d3 d4 Γ±16 = ±d1 d2 d3 d4 (10. Obviously. d3 . d2 = . (1955). The associative algebra generated by d1 . the matrix elements of C = A ⊗ B are Cjk. Γ±5 = ±d4 Γ±6 = ±d1 d2 . Γ±9 = ±d2 d3 Γ±10 = ±d2 d4 . Since the Γj are hermitian the similarity transformations are actually given in terms of unitary transformations. The Clifford algebra C4 entails a subgroup G4 of elements ± dj1 dj2 · · · djs . i. The reader may .e. i. .236) where ‘⊗’ denotes the Kronecker product between matrices. d4 . hence.235) and. Γ±7 = ±d1 d3 . One can conclude then that also any set of 4 × 4–matrices obeying (10.. d4 is called a Clifford algebra C4 . The representations of this group are given by a set of 32 4 × 4–matrices which are equivalent with respect to similarity transformations.237) by means of the property (10.237) which are the ordered products of the operators ±1 and d1 .e. satisfy the anti-commutation property dj dk + dk dj = 2 for j = k 0 for j = k Relativistic Quantum Mechanics d4 = γ 0 (10. j = 1. Rev. 27. 2. To complete the proof in this section the reader may consult Miller ‘Symmetry Groups and their Application’ Chapter 9.H.Mod. 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 . This property extends then to 4 × 4–matrices which obey (10.213). page 187. Γ±13 = ±d1 d2 d4 .238) These elements form a group since obviously any product of two Γr ’s can be expressed in terms of a third Γr . to Dirac matrices.234) (10. any product 1 of the dj ’s can be brought to the form (10. 3 . d4 = 1 0 0 −1 1 0 0 −1 0 1 1 0 0 i −i 0 (10. For example. m = Aj Bkm . Γ±8 = ±d1 d4 . .. Γ±4 = ±d3 . The three Pauli matrices also obey the property (10. d2 . Γ±11 = ±d3 d4 Γ±12 = ±d1 d2 d3 .235). form a Clifford algebra C3 .6 and R. Γ±14 = ±d1 d3 d4 . j1 < j2 < · · · < js s ≤ 4 (10.235) as can be readily verified from (10. Γ±3 = ±d2 .235) can differ only with respect to unitary similarity transformations.

the form of the Dirac equation is identical in equivalent frames of reference. we do not know yet how to transform the 4-dimensional wave function Ψ and the Dirac matrices γ µ .76) yields ∂ρ = Lν ρ ∂ ν . namely. transform coordinates and derivatives of the Dirac equation.80) by Lµ ρ and using (10. µ µ (10. We can.6). Lorentz Transformation of the Bispinor State Actually. j = 1.80). a result one could have also obtained by applying the chain rule to (10. x µ = Lµ ν xν (10. Exercise 7. x = Lµ ν xν . it must hold γ µ = γ µ . and γ j . (10. in a moving frame should hold iγ ∂µ − m Ψ (x ) = 0 . therefore. in a moving frame holds iγ µ ∂µ − m Ψ (x ) = 0 . these properties determine the matrices γ µ uniquely.. Hence.. Multiplication and summation of (10.239) Form invariance implies that the matrices γ µ should have the same properties as γ µ .6).. 10. 10.213. in frames connected by Lorentz transformations. i..218) of the Dirac matrices γ µ .241) .8 Lorentz Invariance of the Dirac Equation We want to show now that the Dirac equation is invariant under Lorentz transformations. 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.e. and derivatives according to (10. Infinitesimal Bispinor State Transformation We want to show now that a suitable transformation of Ψ(xµ ) does. in fact. 2.240) . i.8: Lorentz Invariance of the Dirac Equation 317 also want to establish the unitary transformation which connects the Dirac matrices in the form (10. 3 has imaginary eigenvalues ±i and can be represented by an anti-hermitian matrix. i. Exercise 7.2: Demonstrate the anti-commutation relationships (10.10.e. The latter transformations imply that coordinates transform according to (10. exist.219).214) follows that γ 0 has real eigenvalues ±1 and can be represented by a hermitian matrix.e. The transformation is assumed to be linear and of the form Ψ (x ) = S(Lµ ν ) Ψ(xµ ) µ µ (10.4: Show that from (10.218) of the Pauli matrices σ j . However.e.e. Except for a similarity transformation.215).236) with the Dirac representation (10.3: Demonstrate the anti-commutation relationships (10. i. Exercise 7..

241) implies a similarity transformation Sγ µ S −1 . Multiplication of this property by g from the right yields ( g)T = − g. of the equation. νµ (10. the coefficients of which depend on the matrix Lµ ν defining the Lorentz transformation in such a way that S(Lµ ν ) = 1 for Lµ ν = δ µ ν holds.246) (10.247) the infinitesimal matrix µν is anti-symmetric.s.243) in a form in which the Lorentz transformation in the form Lµν is on the r.221) upon Lorentz transformation yields iS(Lη ξ )γ µ S −1 (Lη ξ ) Lν µ ∂ν − m Ψ (x ) = 0 . S( µν ) should not change its value if one replaces in its argument µν by − νµ . For this purpose we exploit (10. One can finally conclude multiplying both sides by g ρµ S(Lη ξ )γ µ S −1 (Lη ξ ) = Lν µ γν .318 Relativistic Quantum Mechanics where S(Lµ ν ) is a non-singular 4 × 4–matrix.h.e. (10. Ob1 viously.e. state that the Dirac equation (10. µ (10.12) in the form Lν µ gνσ Lσ ρ = gµρ = gρµ . in the expression of the infinitesimal transformation ρ Lµν = g µν + µν (10.245) The construction of S(Lη ξ ) will proceed using the avenue of infinitesimal transformations.. using gρµ γ µ = γρ . The proper starting point for a constructiuon of S(Lη ξ ) is actually (10. One can. g is an anti-symmetric matrix. however. follows S(Lη ξ )γρ S −1 (Lη ξ ) = Lσ ρ γσ . µ g ρν = µν and.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 infinitesimal transformation S(Lρ σ ) which corresponds to (10. (10. it must hold (10.248) Here σµν denote 4 × 4–matrices operating in the 4-dimensional space of the wave functions Ψ.243) We want to determine now the 4 × 4–matrix S(Lη ξ ) which satisfies this condition. It holds then S( µν i ) = 1 − σµν 1 4 µν µν = S(− νµ i ) = 1 + σµν 1 4 µν . The elements of g are. (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.244) from which. We had introduced in (10.247) can be expanded S( µν i ) = 1 − σµν 1 4 µν (10.249) from which we can conclude σµν = −σµν νµ = −σνµ i. i. Multiplication of (10. ..250) σµν = −σνµ . hence. the transformation (10. hence.243) from the left by Lσ ρ gσν yields S(Lη ξ )γ µ S −1 (Lη ξ ) gρµ = Lσ ρ gσν γσ .

Furthermore.256) One can readily verify that the non-vanishing generators σµν are given by (note σµν = −˜νµ .252) results in the condition for each α. (10.252) also as a sum over both indices of αβ .h. β [ σαβ ..228.s. 2. −γ ) are ˜ γ ˜ γ0 = ˜ 0 1 1 1 0 1 .254) The proper σαβ must be anti-symmetric in the indices α.. (10. γj ] = γ ˜ 2 −iσ j 0 0 iσ j . For this purpose we need to verify that the algebra of the generators involving addition and multiplication is closed.s. a solution of condition (10. must be symmetric with respect to interchange of the indices α and β.248. (10. like the expression on the l. (10.229).10. γβ ]− 2 (10. of (10. β and operate in the same space as the Dirac matrices.253) 1 2 αβ ( δ µ β γα − δ µ α γβ ) . namely.246) results then in a condition for the generators σµν − i ( σαβ γ µ − γ µ σαβ ) 4 αβ = νµ γν .255) which can be demonstrated using the properties (10. i.257) . (10. 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. γ µ ]− = 2i ( δ µ β γα − δ µ α γβ ) .213. For this purpose one needs to express formally the r. Comparing this with the l. j = 1. For this purpose we inspect the properties of the generators in a particular representation.5: Show that the σαβ defined through (10. σjk = [˜j . 10. In this representation the Dirac matrices γµ = (˜ 0 .s.252) Since six of the coefficients αβ can be chosen independently.251) S −1 ( µν ) = 1 + σµν µν 1 4 Inserting (10. of (10.e.254). ˜ ˜ σ only six generators need to be determined) σ0j = ˜ i [˜0 . the chiral representation introduced above in Eqs. γj = ˜ 0 −σ j σj 0 .255) satisfy condition (10. Exercise 7. In fact.216) of the Dirac matrices. 10. this condition can actually be expressed through six independent conditions. Algebra of Generators of Bispinor Transformation We want to construct the bispinor Lorentz transformation by exponentiating the generators σµν .s.254) is σαβ = i [ γα .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. 3 . 10. the expression on the r. γk ] = ˜ γ ˜ jk σ 0 0 σ .h.h.251) into (10.h.

one can express the tensor µν through w and ϑ using µν = µ ρ g ρν and the expression (10. therefore. the algebra of these generators is closed under addition and multiplication.260) Inserting this into (10. ϑ) = e 2 (w − iϑ)·σ 0 − 1 (w + iϑ)·σ 0 e 2 1 (10. any representation of the generators. Finite Bispinor Transformation The closedness of the algebra of the generators σµν defined through (10.44)  0 −w1 −w2 −w3  w 0 ϑ3 −ϑ2   =  1  w2 −ϑ3 0 ϑ1  w3 ϑ2 −ϑ1 0  µν (10. and since the algebra of the Pauli matrices is closed. µν in the exponential form i S = exp − σµν 4 µν . In order to construct an explicit expression of S in terms of w and ϑ we employ again the chiral representation. be able to state S in terms of the same parameters w and ϑ as the Lorentz transformation in (10. the representation (10.320 Relativistic Quantum Mechanics Obviously. since both operations convert block-diagonal operators A 0 0 B (10. 2 0 −(w + iϑ) · σ 13 + σ23 ˜ 23 ) (10. i.51) and S. therefore. One should.248) allows us to write the transformation S for any. in particular.e.262) This expression allows one to transform according to (10. (10.259) We had stated before that the transformation S is actually determined through the Lorentz transformation Lµ ν . In fact.259) yields the desired connection between the Lorentz transformation (10.258) again into block-diagonal operators..259) becomes in the chiral representation ˜ S(w.51).241) bispinor wave functions from one frame of reference into another frame of reference.261) We note that this operator is block-diagonal. We can finally note that the closedness of the algebra of the generators σµν is not affected by similarity transformations and that. not necessarily infinitesimal. Since such operator does not change its block-diagonal form upon exponentiation the bispinor transformation (10. . In this representation holds i ˜ − 4 σµν µν i = − (˜01 01 + σ02 02 + σ03 03 + σ12 12 + σ13 σ ˜ ˜ ˜ ˜ 2 1 (w − iϑ) · σ 0 = .255) yields a closed algebra.

s. For our proof we note first ˜ ˜ S −1 (w. of (10. and addition yields Ψ† (xµ )γ 0 The last result can be written ∂µ Ψ† (xν )γ 0 γ µ Ψ(xν ) = 0 .e.270) We conclude that (10. −ϑ) = e− 2 (w − iϑ)·σ 0 1 (w + iϑ)·σ 0 e2 1 (10.h. ϑ) = S(−w.263) does hold.8: Lorentz Invariance of the Dirac Equation Current 4-Vector Associated with Dirac Equation 321 We like to derive now an expression for the current 4-vector j µ associated with the Dirac equation which satisfies the conservation law ∂µ j µ = 0 . This transformation behaviour can also be deduced from the transformation properties of the bispinor wave function Ψ(xµ ).. 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 .263) allows one to conclude that j µ must transform like a contravariant 4-vector as the notation implies. (10. The conservation law (10.267) has the desired property of being positive definite.10.221) and the adjoint Dirac equation (10.271) . (10.264) (10. the conservation law (10. j) = Ψ† (xµ )γ 0 γ µ Ψ(xµ ) .224) from the right by Ψ(xµ ).266) ρ(xµ ) = Ψ† (xµ )Ψ(xµ ) = s=1 |ψs (xµ )|2 (10.263) obviously is a scalar under Lorentz transformations and that the left hand side must then also transform like a scalar. The reason is that the r.263) Starting point are the Dirac equation in the form (10.269) One can readily show that the same operator is obtained evaluating γ S (w.268) We will prove this property in the chiral representation. We will now determine the relationship between the flux j µ = Ψ (x )γ 0 γ µ Ψ (x ) † µ (10. (10. Obviously. (10. in fact.265) (10.224). 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 . Multiplying (10. Since ∂µ transforms like a covariant 4-vector.268) holds for the bispinor Lorentz transformation. The time-like component ρ of j µ 4 iγ µ ∂µ + iγ µ ∂µ ← Ψ(xµ ) = 0 . for j µ (xµ ) = (ρ. (10. j µ must transform like a contravariant 4-vector. the property applies then in any representation of S.

272) results in the expected transformation behaviour µ j = Lµ ν j ν . commute with each other. Combining this with (10. In the following we want to provide this characterization. (10. As it turns out. not only as the most simple demonstration of the theory. The additional degrees of freedom described by the four components of the bispinor wave function require.246) S −1 (Lη ξ )γ µ S ( Lη ξ ) = (L−1 )νµ γν = (L−1 )ν γ ν = Lµ ν γν . specific for the Dirac free particle. i. Note that we have assumed in (10. of which the wave functions are eigenfunctions as well.273) where we have employed (10. 10.272) With S −1 (Lη ξ ) = S((L−1 )η ξ ) one can restate (10.232. The solutions provide also a complete.. One obtains using (10.275) The free particle wave function is an eigenfunction of Ho . Like in non-relativistic quantum mechanics the free particle wave function plays a central role.233) o i∂t Ψ(xµ ) = Ho Ψ(xµ ) . (10. (10.271) that the Dirac matrices are independent of the frame of reference. bispinor wave functions and we need to determine corresponding components of the wave function. In relativistic quantum mechanics a Dirac particle can also be characterized through a momentum.274) 10. a property which leads to the energy– momentum (dispersion) relationship of the Dirac particle. For this purpose we consider . (10. however. 10. 10. additional characterizations. the identification of observables and their quantum mechanical operators. but also as providing a basis in which the wave functions of interacting particle systems can be expanded and characterized. the Dirac particles are described by 4-dimensional. These operaˆ tors. the wave function has four components which invite further characterization of the free particle state. 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)]. As pointed out.322 Relativistic Quantum Mechanics in a moving frame of reference and the flux j µ in a frame at rest.233).9 Solutions of the Free Particle Dirac Equation We want to determine now the wave functions of free particles described by the Dirac equation.231.e. In case of non-relativistic quantum mechanics the free particle wave function has a single component ψ(r. µ (10. The operators commute also with Ho in (10. t) and is determined through the momentum p ∈ R3 . The independent degrees of freedom allow one to choose the states of the free Dirac particle as eigenstates of the ˆ ˆ 4-momentum operator pµ and of the helicity operator Γ ∼ σ · p/|p| introduced below.f.76). as just mentioned. We will start from the Dirac equation in the Schr¨dinger form (10.282. as is required for the mentioned property.268) µ j = Ψ† (xµ )S † γ 0 γ µ SΨ(xµ ) = Ψ† (xµ )γ 0 S −1 γ µ SΨ(xµ ) .283)]. 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. only two degrees of freedom of the bispinor four degrees of freedom are independent [c.

281) Obviously.10.234) and (10. (10.281). like the Klein–Gordon equation. 10.232) leads to the 4-dimensional eigenvalue problem m σ·p σ · p −m φo χo = φo χo . reproduce the classical relativistic energy–momentum relationships (10. 10.94) Equation (10.282) (10. and φo .279) the well-known relativistic dispersion relationship 2 = m2 + p 2 (10. later to be identified with momentum and energy.278) provides us with information about the components of the bispinor wave function (10. hence. (10. These two relationships are consistent with each other.277) explicitly ( − m) 1 φo − σ · p χo 1 − σ · p φo + ( + m) 1 χo 1 = 0 = 0. one finds using (5. 2 − m2 0 (10. the Dirac equation.279) According to the property (5. E(p) = m2 + p 2 .276) where p and together represent four real constants.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 .284) The relationships (10.234) of Pauli matrices holds (σ · p)2 = p 2 1 One can. In fact.283) imply that the bispinor part of the wave function allows only two degrees of freedom to be chosen independently. two-dimensional spinor state.280) σ·p (σ · p)2 φ0 = χ0 = +m ( + m) ( + m) p2 χ = χ0 .280) which has a positive and a negative solution = ± E(p) . conclude 1.276) into (10.282.277) To solve this problem we write (10. . (10.93.231. 10.276). the socalled helicity of the particle.283) where is defined in (10. We want to show now that these degrees of freedom correspond to a spin-like property. 1 (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. namely φo and χo are related as follows φ0 χo = = σ·p χ0 −m σ·p φ0 . Inserting (10. +m (10. χo each represent a constant. from (10.

For the positive energy solution. using γ 0 as given in (10. Ψ† (0. One can see this as follows: Let Ψ(p. +) γ 0 Ψ† (p. of (10. +) = |N+ (0)|2 (u† . +) γ 0 Ψ(0. 10. where S is given by (10. ±) . ±) Ψ† (0.268). u† u = |u1 |2 + |u2 |2 = 1 . (10. and let Ψ(0. we present φo through the normalized vector φo = u1 u2 = u ∈ C2 .h.287.287. ±) = S Ψ(0. the l.287) in the minus sign on the r. i.s. −|xµ ) = N− (p) −σ·p E(p) + m u u ei(p·r − t) .286) Here N+ (p) is a constant which will be chosen to satisfy the normalization condition Ψ† (p. 10.285)]. (10. S † = γ 0 S −1 γ 0 . (10. hence. through χo given by χo = u1 u2 = u ∈ C2 . ±) denote the solution of a free particle moving with momentum p in the laboratory frame.287) the form of which will be justified further below.s. The form of this condition and of (10. is Ψ(p.287) will be justified now. The connection between the solutions.290). ±). (10.281)] separately. is invariant under Lorentz transformations. ±) Ψ† (0. according to (10.. ±) denote the corresponding solution of a particle in its rest frame.h. +) γ 0 Ψ† (p. i. (10.290) should differ. (10.259).219) and u† u = 1 [c.e. (10.289) Here N− (p) is a constant which will be chosen to satisfy the normalization condition Ψ† (p.324 Relativistic Quantum Mechanics For our further characterization we will deal with the positive and negative energy solutions [cf. (10. S −1 = γ 0 S † γ 0 and. ±) S † γ 0 S Ψ(0.. i.286) for p = 0 yields.f.285) The corresponding free Dirac particle is then described through the wave function Ψ(p.e. we present the negative energy solution. For this purpose we consider first the positive energy solution. the solution for = −E(p). ±) = = = Ψ† (0.s.292) . the solution for = +E(p).h.290) which differs from the normalization condition (10. Hence.. Similarly. ±)γ 0 Ψ(p.291) Note that we have used that. = +E(p) . 0) γ 0 u 0 = |N+ (0)|2 . (10. Ψ† (p. (10. +|xµ ) = N+ (p) u σ·p E(p) + m u ei(p·r − t) . = −E(p) .241). Employing (10. ±). according to (10.e.288) corresponding to the wave function Ψ(p. ±) γ 0 S −1 γ 0 γ 0 S Ψ(0. of (10. we demonstrate that the product Ψ† (p. ±) γ 0 Ψ(p. ±) γ 0 Ψ(0. u† u = |u1 |2 + |u2 |2 = 1 . +) = −1 . +) = 1 . We want to demonstrate now that the signs on the r. First.

−) γ 0 Ψ(0. −) = |N− (0)|2 (0.h.f.285) results in 2 N+ (p) 1 − p2 (E(p) + m)2 = 1 (10.299) ( m + E(p) )2 .9: Solutions of the Free Particle Dirac Equation The same calculation for the negative energy wave function as given in (10. u† ) γ 0 0 u = − |N− (0)|2 . 2 E(p) (10. We consider first the positive energy solution.h.296) Replacing (σ · p)2 by p 2 [c. using γ 0 as given in (10. σ·p u∗ E(p) + m T u σ·p E(p) + m u = 1 (10.290) to assign a positive value to |N− (0)|2 .234)] and using the normalization of u in (10.295) Evaluating the l. Condition (10.300) (10.294) We want to determine now N± (p) for arbitrary p.293) Obviously.s.298) This result completes the expression for the wave function (10.289) yields Ψ† (0.286). (5. ( m + E(p) )2 − p 2 (10.s.6: Show that the normalization condition 2 N+ (p) (u∗ )T .286) is 2 N+ (p) (u∗ )T .298) becomes N+ (p) = m + E(p) .301) yields the normalization coefficient N+ (p) = m + E(p) . of (10.302) . (10. Exercise 7. 325 (10.219) yields 2 N+ (p) u† u − u† (σ · p)2 u (E(p) + m)2 = 1.297) from which follows N+ (p) = Noting ( m + E(p) )2 − p 2 = m2 − p 2 + 2mE(p) + E 2 (p) = 2(m + E(p)) m the normalization coefficient (10. (10. σ·p u∗ E(p) + m T γo u σ·p E(p) + m u = 1 (10.10. We can also conclude from our derivation N± (0) = 1 . 2m (10.287) written explicitly using (10. this requires the choice of a negative side on the r.

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. j = 1.289) is T −σ·p −σ · p 2 E(p) + m u N− (p) u∗ . 2. λ|xµ ) in 4-vector notation.e. The wave functions (10. i∂µ Ψ(p. exp[i(p · r − t)] = exp(ipµ xµ ). λ|xµ ) = pµ Ψ(p.. conclude N− (p) = m + E(p) 2m (10. i. have completed the determination for the wave function (10. The commutation property [σ · p. This attribute is the so-called helicity.286.e. λ|xµ ) . λ|xµ ) = λ E(p) Ψ(p. (10.305) and. identical to the condition (10.306) i.300) have been constructed to satisfy the free particle Dirac equation (10. pj ] = 0 . Helicity The free Dirac particle wave functions (10.296) for the normalization constant N+ (p) of the positive energy solution.289) are not completely specified. Ho ] = 0 follows from (10. the two components of u indicate another degree of freedom which needs to be defined.275). 3 is fairly obvious. This degree of freedom 1 describes a spin– 2 attribute. The corresponding operator which measures this observable is 1 p ˆ Λ = σ· . 10.309) . λ|xµ ) (10. Rather than considering the ˆ observable (10.307) we investigate first the observable due to the simpler operator σ · p. thereby.275) yields Ho Ψ(p. (u∗ )T γ o = −1 (10.303) E(p) + m u Evaluating the l.308) 2 |p| ˆ Note that p represents here an operator.290) written explicitly using (10. 10.289).h.289. This can be verified expressing the space–time factor of Ψ(p..304) This condition is. Condition (10. yields 2 N− (p) u† (σ · p)2 u − u† u (E(p) + m)2 = −1 . defined as the component of the particle spin along the direction of motion. hence.s. Inserting (10.233) and ˆ ˆ ˆ from the two identities 1 1 0 0 −1 1 σ 0 0 σ ·p − ˆ σ 0 0 σ ·p ˆ 1 1 0 0 −1 1 = 0 (10. −p).286) into (10. We can. (10. the wave functions constructed represent eigenstates of Ho .326 Relativistic Quantum Mechanics We consider now the negative energy solution. The wave functions are also eigenstates of the momentum operator i∂µ . (10. i.307) where pµ = ( ..286. The property [σ · p. 10.e. not a constant vector. however.

t) = Np  e (10. t) = Np  e   1 0   0 p  m + Ep  i(px3 + E t) 1 p e Ψ(pˆ3 . 0.308) above.10. − 1 |r. p3 ). = +E(p). i.. can ˆ ˆ be simultaneously diagonal. i. −. Since helicity is defined relative to the direction of motion of a particle the characterization of u as an eigenvector of the helicity operator.286) are eigenfunctions of Λ as well will specify now the vectors u. is independent of the direction of motion of the particle.e.e.e. −. In this case Λ = 2 σ 3 .313) 2   0 1 where Ep and Np are defined in (10. We consider first the simplest case that particles move along the x3 –direction. t) = Np  e 2   1 p m + Ep 0   0   i(px3 − E t) 1 p e Ψ(pˆ3 . +.. Therefore. + 1 |r. +. . are   1   i(px3 − E t) 0 p e Ψ(pˆ3 . i.224) 1 of σ3 the two u vectors (1. 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. 1)T are eigenstates of 2 σ 3 with eigenvalues ± 1 .310) We have shown altogether that the operators p.. + 2 |r. Np = m + Ep . States which are simultaneously eigenvectors of these three operators are also simulteneously eigenvectors of the three operators p. = −E(p). Ho and σ·p commute with each other and. 1 p = (0.312). hence. According to the definition (5. 327 0 (σ · p)2 ˆ 2 (σ · p) ˆ 0 0 (σ · p)2 ˆ 2 (σ · p) ˆ 0 (10. ˆ The condition that the wave functions (10. t) = Np  e (10. 0)T and (0. 2m (10. in principle.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. We assume first particles with positive energy. − 1 |r.312) We assume now particles with negative energy. Ho and Λ defined in (10.311) 2  −p  0 m + Ep 1 where e3 denotes the unit vector in the x3 -direction and where ˆ Ep = m2 + p2 .

γ and are  1  0  1 ˜ Ψ(p = 0.315) e3 × p ˆ ∠(ˆ3 . −. (10. 0)T and (0.319) 1 Ψ(p. t) = N√ ei(p·r + Ep t) (10. Generating Solutions Through Lorentz Transformation The solutions (10. +.229). − 1 ) 2 ei(p·r − Ep t) (10. t) 2 = Np 1 u(p. [One can also express the rotation through Euler angles α. − 1 |r. β. − ) 2 where ϑ(p) = = = exp exp i − ϑ(p) · σ 2 i − ϑ(p) · σ 2 1 0 0 1 . 10.e.318) 1 Ψ(p. p) e |p| (10. + 1 ) 2 p 1 m + Ep u(p. 2  1  0  (10. γ .262) for the bispinor wave function and the transformation (10.314) (10. in which case the transformation is given by (5. These eigenstates are obtained through a rotational transformation (5. + 1 |r.223) as follows 1 u(p. + 2 ) u(p. t) = Np ei(p·r + Ep t) (10. For this purpose one starts from the solutions of the Dirac equation in the chiral representation (10. + 2 ) p 1 m + Ep u(p.] The corresponding free particle wave functions are then Ψ(p. − 2 |r.311. a wave function which represents free particles at rest. 10.316) describes a rotation which aligns the x3 –axis with the direction of p. t) 2 = Np u(p.317) Ψ(p. −.321) . + 2 ) ei(p·r − Ep t) (10. 5. for an r–independent wave function.226.311. i. +.. + 2 |r.123). − 2 ) u(p.312). + ) 2 1 u(p. The corresponding wave functions are determined through ˜ i˜ 0 ∂t − m Ψ(t) = 0 . denoted by ˜.313) except that the states (1.220). +.312) can be obtained also by means of the Lorentz transformation (10. 10. 1 |t) = √2   e−imt .320) where Ep and N are again given by (10. − 1 ) 2 −p 1 m + Ep u(p.222. 1)T have to be replaced by eigenstates u± (p) of the spin operator along the direction of p. − 2 ) −p 1 m + Ep u(p.328 Relativistic Quantum Mechanics To obtain free particle wave functions for arbitrary directions of p one can employ the wave functions (10.

313) in the p → 0 limit.311. 10.324. 10.311)  e 1 w 2 3 ˜ Ψ(p(w3 )ˆ3 .323) should yield the solutions for non-vanishing momentum p in the x3 –direction.e.176) imt → i ( p3 x3 and for the bispinor part according to (10.322) can be written in spinor form 1 √ 2 φo χo e imt . χo are eigenstates of σ 3 with eigenvalues ±1. +.324) 0 e − 1 w3 σ 3 2 φo χo = e 2 w3 σ φo 1 3 e− 2 w3 σ χo 1 3 . − 1 |r.325) to (10. i.325) One should note that φo . 2  0  1   1  0  +imt 1 1 ˜  Ψ(p = 0. t) e 2 1  = √  2   e− 2 w3 1 1 1 0 1 0 0 1 0 1   i(px3 − E t) p e   ˜ Ψ(p(w3 )ˆ3 . φo . This correspondence justifies the characterization ±.322). ± 1 of the wave functions stated in (10.262) φo χo → e 2 w3 σ 0 1 3 Et ) (10. w3 ). χo ∈ 1 0 .322) The reader can readily verify that transformation of these solutions to the Dirac representationsas defined in (10. For the resulting wave functions in the chiral representation one can use then a notation corresponding to that adopted in (10. for w = (0. 2  0 e −1  329 (10. Applying (10.10. (10. 0 1 (10. − 1 |t) = √2   e−imt .9: Solutions of the Free Particle Dirac Equation  0  1  1 ˜ Ψ(p = 0.  −1  e 0   0  1  +imt 1 ˜  Ψ(p = 0.228) yields the corresponding solutions (10. +.174. − 1 |t) = √2  . 2 |t) = √2  . −.. yields for the exponential space–time dependence according to (10.323) Transformation (10. +. −. t) e 2 e− 2 w3 1  = √  1 2  e 2 w3  i(px3 − E t) p e  . 10.262) for a boost in the x3 –direction. 2 The solutions (10. + 1 |r. 0.

Transformation to the Dirac representation by means of (10. 2 sinh x = 2 coshx − 1 . and the expression (10. t) e  =    i(px3 + E t) p e  (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 . t) e 1  = √  2   −e− 2 w3 − 1 w3 2 1  i(px3 + E t) p e   ˜ Ψ(p(w3 )ˆ3 .311) for Ep one obtains 1 = √ 2 1 1 √ 2 + 1 = 1 − v3 2 m2 + = 2 m2 v3 2 1 − v3 cosh w3 2 1+ 2 v3 2 + 1 1 − v3 +m = 2m Ep + m 2m (10. + 2 |r. − 1 |r. −. t) e 2  =     i(px3 + E t) p e   1 Ψ(p(w3 )ˆ3 . t) e  =     cosh w3 2 − sinh w3 2 sinh w3 2 cosh w3 2 − sinh w3 2 cosh w3 2  i(px3 − E t) p e   Ψ(p(w3 )ˆ3 .330  e 1 w 2 3 Relativistic Quantum Mechanics 1 0 1 0 0 1 0 1  1 ˜ Ψ(p(w3 )ˆ3 .329) .61) p(w3 ) = m sinhw3 .61) between the parameter w3 and boost velocity v3 .327) Employing the hyperbolic function properties cosh x = 2 coshx + 1 . 2 (10. +. + 1 |r. −. −. − 2 |r. t) e 2 e 1   = √  1 2 − e 2 w3  i(px3 + E t) p e  (10.326) where according to (10. − 2 |r. +.328) the relationship (10. + 2 |r.228) yields w3  cosh 2 =   sinh w3 2 1 Ψ(p(w3 )ˆ3 .

but not the observer. that if Ψ(xµ ) is a solution of the Dirac equation that Ψ (xµ ) as given in (10.243) is to familiarize ourselves with a formulation of Lorentz transformations of the bispinor wave finction Ψ(xµ ) which treats the spinor and the space-time part of the wave function on the same footing. in general. of course. λ. The reason why we provide another derivation of (10. Actually. the proper transformation S(Lη ξ ). We expect. we will derive this equation only for infinitesimal transformations. Such transformation had been applied by us. Making this expectation a postulate allows one to derive the condition (10.9: Invariance of Dirac Equation Revisited and similarly sinh w3 = 2 Ep − m = 2m p 2m (Ep + m) 331 (10.331) is a solution as well.221) using (10. and since (2) the calculations following (10.327). Such description will be essential for the formal description of Lorentz invariant wave equations for arbitray spin further below.331) i S(Lη ξ )γ µ S −1 (Lη ξ ) ρ(Lη ξ )∂µ ρ−1 (Lη ξ ) − m Ψ (xµ ) = 0 (10. This transformation. The transformation is Ψ (xµ ) = S(Lη ξ ) ρ(Lη ξ ) Ψ(xµ ) (10. 10. expresses the system in the old coordinates.168–10.243) and. refered to as the active transformation. when we generated the solutions Ψ(p.123) above and characterized there.333) .243). thereby.243) considered solely the limit of infinitesimal transformations anyway.10. Invariance of Dirac Equation Revisited At this point we like to provide a variation of the derivation of (10. λ. Λ|xµ ) from the solutions describing particles at rest Ψ(p = 0.311) 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 . 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 . To show this we rewrite the Dirac equation (10. ρ(Lη ξ ) has been defined in (10.330) into (10.313) for −p.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 ).334) (10. = γν . reproduces the positive energy wave functions (10. is sufficient since (1) it must hold then for any finite transformation.330) Inserting expressions (10. In the new derivation we consider the particle described by the wave function transformed. the essential property stating the Lorentz–invariance of the Dirac equation. indeed. Λ|t).176).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. (10. which however.

    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. ∂µ ] = (10. As mentioned already we will show condition (10. . 0). ∂µ ] = (K )ν µ ∂ν . 0).339) [K1 .338) [J3 .. ∂µ ] = [x1 ∂3 − x3 ∂1 . We will proceed by employing the generators (10. w = (0. ∂µ ] = (10. ∂µ ] = (J )ν µ ∂ν . 0. . 0. w = (0. ∂µ ] = (10. 0. 0). the Dirac equation is invariant under active Lorentz transformations. therefore. We have demonstrated. i. w = (0.336). 0. 1) . ϑ = (0. ∂µ ] = [x0 ∂2 + x2 ∂0 . The second condition is identical to (10. . . Inspection of (10. 0).340) [K2 .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 (10.341) [K3 .334) for infinitesimal Lorentz transformations Lη ξ .243) and. ∂µ ] = [x3 ∂2 − x2 ∂3 . ∂µ ] = (10. that any solution Ψ(xµ ) transformed according to (10. [K . ∂µ ] = [x0 ∂1 + x1 ∂0 . 1. 0. ∂µ ] = (10. it is met by S(Lη ξ ) as given in the chiral representation by (10. 0). ∂µ ] = [x0 ∂3 + x3 ∂0 .337) [J2 .331) is again a solution of the Dirac equation.335) shows that we need to demonstrate [J .262). ∂µ ] = [x2 ∂1 − x1 ∂2 .e. ∂µ ] = (10. .335) = ∂µ + M ν µ ∂ν + O( 2 ) The result will show that the matrix M ν µ is identical to the generators of Lν µ for the six choices ϑ = (1. of course.336) We will proceed with this task considering all six cases: [J1 .342) One can readily convince oneself that these results are consistent with (10. ϑ = (0.332 Relativistic Quantum Mechanics We will demonstrate now that the first condition is satisfied by ρ(Lη ξ ).

σ2 . Accordingly.233) by replacing i∂t by (see Table 10.1 in Sect 10. we define φ(xµ ) = e−imt Φ(xµ ) χ(x ) = e µ −imt (10.345) holds |qV | << m .343) One may also include the electromagnetic field in the Dirac equation given in the Schr¨dinger form o ˆ (10. accordingly.345) in the so-called non-relativistic limit in which all energies are much smaller than m. The Dirac equation in the Schr¨dinger form reads then o i∂t Ψ(xµ ) = ˆ ˆ where α and β are defined in (10.347) (10.10 Dirac Particles in Electromagnetic Field We like to provide now a description for particles governed by the Dirac equation which includes the coupling to an electromagnetic field in the minimum coupling description. This solution exhibits a timedependence exp[−i(m + )t] where for holds in the non-relativistic limit | | << m.346) ˆ ˆ ˆ α · π + qV + β m Ψ(xµ ) (10..348) (10. e.g. 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 .350) (10.351) X (x ) µ . (10.221) reads then [ iγ µ (∂µ + iqAµ ) − m ] Ψ(xν ) = 0 (10. For this purpose we choose the decomposition Ψ(xµ ) = φ(xµ ) χ(xµ ) .232).6 we assume that the field is described through the 4-vector potential Aµ and. we replace the operator ∂µ by ∂µ + iqAµ . 10. (10.10: Dirac Particles in Electromagnetic Field 333 10. Non-Relativistic Limit We want to consider now the Dirac equation (10. Following the respective procedure developed for the Klein-Gordon equation in Sect.345) (10. The Dirac equation (10.349) We want to focus on the stationary positive energy solution. for the scalar field V in (10.10. 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 10.344) Using the notation σ = (σ1 . Equivalently.6 above).1) i∂t − qV and p by ˆ ˆ π = p − qA .

For this purpose we employ the identity (5. ˆ Equation (10. Ak ] + q 2 [Aj .356) i∂t Φ ≈ Φ + qV Φ . and.348. Ak ] . ∂k ] + [∂j . 5. One obtains [πj . (10.352) allow one to approximate (10.357) Equation (10.358) ( πj πk − πk πj ) = jk [πj . 10. ˆ ˆ For the components of π × π holds ˆ ˆ π×π = jk (10. π ] .355) ˆ σ·π Φ 2m (10.354) The properties (10. derived in Sect. 10. ∂k ] + [∂j . (10. (10. accordingly. identifies X as the small component of the bi-spinor wave function which. due to the m−1 factor.356). 10.334 and assume that for the time-derivative of Φ and X holds ∂t Φ << m . ∂k ] + q [Aj .361) . (10. ∂k ] + q [ ∂j . Φ Using (10.353) by X ≈ to obtain a closed equation for Φ ˆ σ·π 2m 2 (10. (10. which in the present case states ˆ ˆ2 ˆ ˆ (σ · π)2 = π + i σ · (π × π) .350. henceforth.357) for Φ can be reformulated by expansion of (σ · π)2 .352) ˆ σ · π X + qV Φ ˆ σ · π Φ + qV X − 2mX . πk ] = = 1 1 [ ∂j + qAj .346.359) We want to evaluate the latter commutator. ∂k + qAk ] i i 1 1 1 1 [ ∂j .230). i i (10. one can replace X in (10.349) yields i∂t Φ i∂t X = = ∂t X X Relativistic Quantum Mechanics << m .7.351) in (10.360) For an arbitrary function f (r) holds ( [Aj .354) ˆ 0 ≈ σ · π Φ − 2m X . Ak ] ) f = ( ∂j Ak − Ak ∂j + Aj ∂k − ∂k Aj ) f . Ak ] i i i i =0 =0 = q q [Aj .353) (10. does not need to be considered anymore.

362) or. the interaction qσ · B induces a 1 precession of the spin.343) for the vector potential Aµ = (− Ze2 . −A). Comparision of (10.359. t) (10.344. while agreeing in all other respects with the non-relativistic Schr¨dinger equation (10.360) and Aµ = (V.2) governing a one-dimensional wave function ψ ∈ C.366) can be interpreted as a rotation around the field B by an angle qtB.1 which emerged in the Lorentz-invariant theory as an algebraic 2 necessity.e. In other words. The formal solution of this equation is Φ(t) = e−iqtB·σ Φ(0) .364) governing a two-dimensional wave function Φ ∈ C2 with the corresponding non-relativistic Schr¨dinger equation (10. Ak ] ) f = [ ((∂j Ak )) − ((∂k Aj )) ] f (10.367) .222. 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) .364) which is referred to as the Pauli equation. t) ≈ ˆ [p − q A(r.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 ((· · ·)).1 particle with a magnetic 2 field B.10. 0.2 around the magnetic field. t) [see (8. 0.2) it introduces the o extra term qσ · B Φ which describes the well-known interaction of a spin. 5. 10. but rather remains as a steady “guest” of non-relativistic physics with the proper interaction term.363) allow us to write (10.1 with the magnetic field.Spectrum We want to describe now the spectrum of a relativistic electron (q = −e) in the Coulomb field of a nucleus with charge Ze. the spin. o reveals a stunning feature: the Pauli equation does justice to its two-dimensional character. Equations (10.363) where we employed B(r. πk ] = q q (( ∂j Ak − ∂k Aj )) = − i i ×A jk q = − B i jk (10. 10. does not leave the theory again when one takes the non-relativistic limit. [πj . Dirac Particle in Coulomb Field . r (10. The respective bispinor wave function Ψ(xµ ) ∈ C4 is described as the stationary solution of the Dirac equation (10.358.357) in the final form i∂t Φ(r.366) (10. using (10.223) shows that the propagator in (10. one obtains ( [Aj . ∂k ] + [∂j . t) 2m 2m Φ(r. i. 10.365) Comparision of this expression with (5. Let us consider briefly the consequences of the interaction of a spin.6)]. t) = × A(r. t)]2 q − σ · B(r.. (10. t) + q V (r.

respectively. according to (10. ˜ (10.1 one can write (10.e.h.371).180). 3 µ=0 ˆ2 (˜ µ )2 πµ = π0 − π .370) is obtained then ˜ [ i˜ µ (∂µ + iqAµ ) + m ] Ψ(xµ ) γ is a solution of (10. For this purpose we ‘square’ the Dirac equation. 2. but differs from it in an essential way.369).373) (10. ˜ However.s.230) γ µ γ ν = − γ ν γ µ . µ = ν and altering ‘dummy’ summation ˜ ˜ ˜ ˜ .6.ν=1 µ =ν γ µ γ ν πµ πν . noting from (10.369) For our solution we will adopt presently a strategy which follows closely that for the spectrum of pionic atoms in Sect.370) ˜ [ i˜ µ (∂µ + iqAµ ) − m ] [ i˜ µ (∂µ + iqAµ ) + m ] Ψ(xµ ) = 0 γ γ (10.370) resembles closely the Klein-Gordon equation (10. multiplying (10. 5. 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 . ˜ ˜ ˆ ˆ (10.368) is also a solution of (10. Equation (10. 3.370). is a solution of (10. (˜ 0 )2 = 1 and (˜ j )2 = γ 1 γ −1 j = 1. but the converse is not necessarily true. 10. (5.230).232) in Sect.370) for which holds ˜ ˆ ˜ ˆ 3 γ πµ γ πν = ˜ ˆ ˜ ˆ µ=0 µ ν (˜ µ )2 πµ + γ ˆ2 µ. 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. indeed. we solve ˜ [ i˜ µ (∂µ + iqAµ ) − m ] Ψ(xµ ) = 0 γ (10. can be rewritten using. γ ˆ ˜ ˆ (10.370) Any solution of (10.229).231).228) and in (10. once a solution Ψ(xµ ) of (10.372) (10.374) The first term on the r.336 Relativistic Quantum Mechanics For the purpose of the solution we assume the chiral representation.375) Following the algebra that connected Eqs.368) ˜ where Ψ(xµ ) and γ µ are defined in (10. 1. The difference arises from the term γ µ πµ γ ν πν in (10.369) from the left by γ ν πν + m.371) such that we can conclude that (10. γ ˆ2 ˆ2 (10. i.368) ˜ ( γ µ πµ − m ) Ψ(xµ ) = 0 .374). Employing πµ as defined ˜ in Table 10.7 one can write the second term in (10. (5.369).

γ ˜ π ˆ j=1 (10.379) The commutators [ˆ0 . Altogether.381) According to (10.367) and the definition (10.378) According to the definition (10.344).e. ν = 1. πj ] π ˆ = = (−i∂t + qA0 . 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. γ ν ] [ˆµ . γ j ] are γ ˜ [˜ 0 .380) γ µ γ ν πµ πν = i ˜ ˜ ˆ ˆ µ. t) is a suitable test function and where ((· · ·)) denotes the range to which the derivative is limited. it holds 1 4 [˜ µ .ν=1 µ =ν 3 j=1 3 3 = 1 4 1 2 1 [˜ .376–10. πj ] in (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. 3 (10. one can summarize (10.367) of Aµ .ν=1 µ =ν 337 1 2 ( γ µ γ ν πµ πν + γ ν γ µ πν πµ ) ˜ ˜ ˆ ˆ ˜ ˜ ˆ ˆ µ.10. πν ] γ ˜ π ˆ µ..376) This expression can be simplified due to the special form (10.ν=1 µ =ν = [˜ µ . π0 ] γ ˜ π ˆ j=1 = [˜ 0 .ν=1 µ =ν σ 0 0 −σ · (( qA0 )) (10. Since [ˆµ . γ j ] [ˆ0 .378) can be evaluated using (10. πj ] . γ µ γ ν πµ πν = ˜ ˜ ˆ ˆ µ. πν ] = 0 π ˆ for µ.10: Dirac Particles in Electromagnetic Field indices.ν=1 µ =ν (10. γ ] [ˆ0 .377) which follows readily from the definition (10.229).380) where f = f (r. πν ] γ ˜ π ˆ µ. i.344) π ˆ [ˆ0 . r2 (10.382) . −i∂j ] = − [ (∂t + iqA0 ) ∂j − ∂j (∂t + iqA0 ) ] f i ((∂j qA0 )) f (10. 2.367) holds qA0 = r ˆ Ze2 . due to A = 0. γ 0 ] [ˆj . πj ] + γ ˜ π ˆ 4 0 j [˜ j . γ ν ] [ˆµ . the commutators [˜ 0 .

384) can be interpreted as the energy of the stationary state and. however.2 components ˜ Ψ(r) = and obtain for the separate components φ± (r) Ze2 + r 2 ˜ φ+ (r) ˜ φ− (r) (10.147. + j } (10.1 . (10.e.388) in terms of states introduced in Sect.387) The expression (10. 2 |ˆ)} as given in (6. we express 2 the solution of (10. m values are coupled.368) reads − Ze2 ∂t − i r 2 + 2 + i σ·r ˆ 0 0 −σ · r ˆ Ze2 − m2 r2 Ψ(xµ ) = 0 (10. . . it is this quantity that we want to determine.5 the operator iσ · r is block-diagonal in this basis such that only ˆ 1 the states for identical j. i. is genuinely two-dimensional and.389) According to the results in Sect. Combining this result with (10. r r 2 2 2 j = 1. 6.385) 1 We split the wave function into two spin.386) + 2 ± iσ · r ˆ Ze2 − m2 r2 φ± (r) = 0 . 1 |ˆ) ) . (10.338 Relativistic Quantum Mechanics where r = r/|r| is a unit vector.374). (10. a solution of the form ∼ Y m (ˆ) can be obtained. (10.383) yields the purely spatial four-dimensional differential equation + Ze2 r 2 + 2 + i σ·r ˆ 0 0 −σ · r ˆ Ze2 − m2 r2 Φ(r) = 0 . −j + 1. .375) the ˆ ‘squared’ Dirac equation (10. m = −j.5 which describe the coupling of orbital angular momentum and spin 1 {( Yjm (j − 2 .6.383) We seek stationary solutions of this equation.191) solved in Sect. hence. 6.384) into (10.. 1 |ˆ). 10. 2 3 2 .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 . Yjm (j + 1 . Yjm (j + r 2 1 1 . in r ˆ fact. . .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. Accordingly. Insertion of (10. 1 |ˆ). 6. We note that these states are also eigenstates of the angular r 2 . . couples the orbital angular momentum of the electron to its spin. In the latter case.381). (10. (10. The term iσ · r. only the two states {Yjm (j − 2 .148). Such solutions are of the form ˜ ˜ Ψ(xµ ) = Φ(r) e−i t .

λ1 (j) and λ2 (j). . the missing additive term − 1 .392). such transformation can be chosen as to diagonalize the second term. 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 (10.392) h± (r) g± (r) = 0.186)]. 6. Obviously. We can. 1 |ˆ) = − Yjm (j ˆ r 2 2 1 2 h± (r) g± (r) Yjm (j − 1 . . the eigenvalues are independent of m.f. conclude that the .199). Since.2 (j)[λ1.390) (10. m and expand φ± (r) = Using σ · r Yjm (j ± 1 .151]. 1. (6.395) (10. .2 (j) as given by (10..10: Dirac Particles in Electromagnetic Field 339 ˆ momentum operator L2 [cf.5 [c. (10.392) into diagonal form. not the wave functions. unaltered. and except for the fact that 2 2 we have two sets of values for λ1. − 1 r2 (j − 1 )(j + 1 ) − Z 2 e4 2 2 ±i Ze2 (j + )(j + ) − Z 2 e4 We seek to bring (10. We select.2 (j). in the present treatment.396) and (10. involving the 2 × 2 unit matrix. 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 (10.151). 3 . However. 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 . Any similarity transformation leaves the first term in (10. 10.391) derived in Sect. 1 |ˆ) r 2 (10.2 (r) = 0 (10. (6. 1 |ˆ) are r 2 2 ˆ eigenfunctions of L2 .10.2 (j) being 2 j = 1 . 1 |ˆ) + r Yjm (j + 1 .395. rather than = 0.200). . except for the slight difference in the expression of λ1. . which states that the states Yjm (j ± 1 .396) and λ2 (j) [λ2 (j) + 1] (10. hence. . namely. namely.392) reads then in the diagonal representation 2 ∂r − λ1. a specific pair of total spin-orbital angular momentum quantum numbers j. 1 |ˆ) r 2 2 2 2 r r . property (6.2 (j) + 1) 2 Ze2 + + r2 r 2 − m2 f1. therefore. we want to determine solely the spectrum. as in the case of pionic atoms. the values of the argument of λ1.397) This equation is identical to the Klein-Gordon equation for pionic atoms written in the form (10.394) Equation (10.393) but do not explicitly consider further the wavefunctions.

1. 3 . j = 1 .404) = 0. .401) These expressions can be equated with the non-relativistic spectrum. j . . 1. . .398. .401) corresponds to a non-relativistic state 2 with quantum numbers n. 1 ms = ± 2 (10. .399) n = 0.400) corresponds to a non-relativistic state with quantum numbers n. . . and spin-orbital angular momentum j = + 1 . . One may also state 2 this in the following way: (10. . Using (10. the second term on the r. . .403) √ Z 2 e4 2 ED (n. −j + 1.395) and 2 corresponds to λ2 (j) as given in (10. j . In case of non-relativistic hydrogen-type 1 atoms. −j + 1. j where 1 corresponds to λ1 (j) as given in (10.400) (10. 1 = 1 + m Z 2 e4 (n + 1 + (j+ 1 )2 − Z 2 e4 )2 2 Z 2 e4 (n + (j+ 1 )2 − Z 2 e4 )2 2 .396) we obtain. 2.402) In this expression n is the so-called main quantum number.401) if one attributes to the respective 1 states the angular momentum quantum numbers = j + 2 and = j − 1 . One obtains in case of (10. . 2 2 1 ≈ m − (10. . Expanding the energies in terms of this parameter allows one to identify is determined by Z the relationship between the energies 1 and 2 and the non-relativistic spectrum. accordingly. . . (10. .402) with (10. n − 1 . . . − + 1. . 3. 10. . of these equations describe the binding energy. .396). (10. . . m = −j. 1. m = −j. . Obviously. j = 1. . m) = 1 + m (n− + . 2 2 m = −j. . .203). 10. n − 1 .395. . .2 . the stationary states have binding energies E = − mZ 2 e4 n = 1. These considerations 2 are summarized in the following equations 1 ED (n.397) is again given by eq.s. . . counts the nodes of the wave function. including spin.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. . . . (10. The magnitude of the relativistic effect 2 2 e4 . . j = − 2 . . 2.340 Relativistic Quantum Mechanics spectrum of (10. 2. . m) 2 n = 1. .400) and (10. (10. . One can equate (10. . . m = − . albeit with some modifications. . . 1. . . . . . . 2n2 = 0. −j + 1. = 1 + m Z 2 e4 √ ( n − − 1 + 2 − Z 2 e4 )2 . .398) 2 = 1 + m . It is given by n = n + + 1 where is the orbital angular momentum quantum number and n = 0. . 1.h. 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 . . and spin-orbital angular momentum j = − 1 . j = + 1 . 2. (10. ( +1) − Z 2 e4 )2 (10.

3.551178 ⇑ .405) In order to demonstrate relativistic effects in the spectrum of the hydrogen atom we compare in Table 10. . (10.1.60583 -3. relativistic.3. The energies were evaluated with m = 511. much less than the rest mass of the electron (511 keV). 1 |ˆ) for m = ± 1 . .405) in terms of the non-relativistic quantum numbers n.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. j (10. The table entries demonstrate that the energies as given by the expression (10. One can combine the expressions (10. case these six states are split into energetically different 2p 1 and 2p 3 states.e.10.40151 ⇑ . binding energy / eV Eq. . 2. . . . (10. in case of heavier nuclei the kinetic energy of bound electrons in the ground state scales with the nuclear charge Z like Z 2 such that in case Z = 100 the kinetic energy is of the order of the rest mass and relativistic effects become important.60601 -3. for example. .10: Dirac Particles in Electromagnetic Field main quantum number n 1 2 2 2 3 3 3 3 3 orbital angular mom.1. However. .551177 ⇑ . (10. (10. r 2 2 2 . Of particular interest is the effect of spin-orbit coupling which removes. . m) = m 1 + Z 2 e4 (n−j −1 + 2 (j+ 1 )2 − Z 2 e4 )2 2 n = 1. j 1 2 1 2 1 2 3 2 1 2 1 2 3 2 3 2 5 2 341 rel. relate closely to the corresponding nonrelativistic states.405).. This is clearly demonstrated by the comparision of non-relativistic and relativistic spectra of a hydrogen-type atom with Z = 100 in Table 10. 0 0 1 1 0 1 1 2 2 spinorbital ang.40146 ⇑ ⇑ -1. (10. = 0. j. n − 1 j = 1 2 1 2 ± m = −j. 2 |ˆ) for m = ± 1 .036 by means of Eqs. Degeneracies are denoted by ⇑. ± 3 .404) finally into the single formula ED (n. mom.2 the non-relativistic [cf.e. is in the range of 10 eV.402. notation 1s 1 2 non-rel. in fact. −j + 1.. .403. The reason is that the mean kinetic energy of the electron in the hydrogen atom. 10. i. binding energy / eV Eq.40147 -1.551176 spectr. the 2p 3 states with r 2 2 2 2 1 involve four degenerate states corresponding to Y 3 m (1. the non-relativistic and relativistic energies are hardly discernible.0041 keV and e2 = 1/137.405)] spectrum of the hydrogen atom. . 1.402)] and the relativistic [cf.405 ) -13. 10.402) -13. 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. the nonrelativistic degeneracy for the six 2p states of the hydrogen atom: in the present. .2: Binding energies for the hydrogen (Z = 1) atom. i.

Tables 10. j 1 2 1 2 1 2 3 2 1 2 1 2 3 2 3 2 5 2 Relativistic Quantum Mechanics non-rel. It holds for n = 2 and j = 3 . (10. to estimate the difference between the energies of the states 2p 3 and 2p 1 (cf.10. 0.0 ⇑ ⇑ -15.3: Binding energies for the hydrogen-type (Z = 100) atom. . (10.036 by means of Eqs.402.342 main quantum number n 1 2 2 2 3 3 3 3 3 orbital angular mom.405) to order O(Z 4 e8 ). We assume for the wave function the stationary state .0041 keV and e2 = 1/137. binding energy / keV Eq.15.1 ⇑ ⇑ ⇑ ⇑ rel.1 ⇑ . m) ≈ m − mZ 2 e4 mZ 4 e8 − 2n2 2n3 1 j+ − 3 4n + O(Z 6 e12 ) . j.6 -42. 0 0 1 1 0 1 1 2 2 spinorbital ang.9 ⇑ .406) α √ 1+ (n − β + β 2 −α)2 The expansion (10. An example for such potential is the Coulomb potential V (r) = −Ze2 /r considered further below. 0.408) 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).409) where V (r) is spherically symmetric. The energies were evaluated with m = 511.405) reads 1 (10.2 -17.3).206) provides in the present case ED (n.2.407) 1 2 This expression allows one. 0) (10. Degeneracies are denoted by ⇑.15.8 ⇑ . for example. 32 (10. binding energy / keV Eq. 1 2 2 2 2 E (2p 3 ) − E (2p 1 ) ≈ − 2 2 mZ 4 e8 2 · 23 1 −1 2 = mZ 4 e8 .3 spectr. (10.405 ) -161.35.1 -34. 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. Introducing α = Z 2 e4 1 and β = j + 2 (10.402) -136. (10. In order to investigate further the deviation between relativistic and non-relativistic spectra of hydrogen-type atoms we expand the expression (10. 10.405). mom.

168]: the term is a scalar (rank zero tensor) in the space of the spin-angular momentum states Yjm (j ± 1 . (6.412) are four-dimensional with r-dependent wave functions.410) 1 where Φ(r).e. 2 |ˆ) + r Yjm (j − 2 . 6. 2 |ˆ)  r Φ(r)  r  . 1 |ˆ) r 2 1 1 = ∂r r r r (10. σ · p ˆ σ · p f (r) Yjm (j + 1 . in particular.10. according to (10..232. 1 |ˆ) .10: Dirac Particles in Electromagnetic Field form Ψ(xµ ) = e−i t 343 Φ(r) X (r) . the term is block-diagonal in the space spanned by the states Yjm (j ± 1 .415) 1 ˆ σ · p rg(r) Yjm (j − 2 . 10. This term has been discussed in detail in Sect. pp. 1 |ˆ) r 2 2 = i ∂r + j+ r 1 2 3 2 f (r) Yjm (j − 1 . 1 |ˆ).411..f. 6. 1 |ˆ) and Yjm (j − 1 .   =  r (10.413) ˆ σ · p g(r) Yjm (j − 1 . r 2 (10.5.414) These equations can be brought into a more symmetric form using ∂r + which allows one to write (10.198)] not couple states with different j.197. i.345). 1 |ˆ) r 2 2 (10. 1 |ˆ) introduced in r 2 2 1 Sect. but are timeindependent.414) ˆ σ · p rf (r) Yjm (j + 1 . 6.411) (10. 1 |ˆ) r 2 = i ∂r − 1 2 r g(r) Yjm (j + 1 . 1 |ˆ) r 2 2 = i j+ ∂r + r j+ r 1 2 1 r f (r) Yjm (j − 2 . ˆ σ · p X + m Φ + V (r) Φ ˆ σ · p Φ − mX + V (r) X = = Φ X (10.416) The differential equations (10. i. The Dirac equation reads then. X (r) ∈ C2 describe the spatial and spin. 1 |ˆ) r 2 2 = i ∂r + −j r 1 g(r) Yjm (j + 2 . 1 |ˆ) .412) ˆ In this equation a coupling between the wave functions Φ(r) and X (r) arises due to the term σ · p. 10. 2 |ˆ) + r Yjm (j − 2 . 2 |ˆ) and does r 2 ˆ has odd parity and it holds [c.e.417)  c(r)  d(r) X (r) 1 1 1 1 Yjm (j + 2 . 10. r 2 2 (10. (10.5 [see.2 degrees of freedom. 2 |ˆ) r r r . m-values.413. The arguments above allow one to eliminate the angular dependence by expanding Φ(r) and X (r) 1 in terms of Yjm (j + 2 . r r 2 2 2     a(r) b(r) 1 1 1 1 Yjm (j + 2 .

c(r) are coupled. 2 |ˆ) r r   f2 (r) 1 1 r Φ(r)  i r Yjm (j − 2 . (10.423) .417) into (10. only a(r). Presently. m values.411. and multiplying by r results in the following two independent pairs of coupled differential equations i ∂r − j+ r j+ r 1 2 d(r) + [ m + V (r) − ] a(r) = 0 i and i ∂r + 1 2 a(r) + [ −m + V (r) − ] d(r) = 0 (10. 1 |ˆ)  r Φ(r) 2 2  r    =  (10.417) of the form     f1 (r) i Yjm (j + 1 . According to (10. such expansion must include states with all possible j.418) holds for f1 (r). g1 (r) ∂r − j+ r j+ r 1 2 g1 (r) + [ − m − V (r) ] f1 (r) = 0 ∂r + and for f2 (r). we consider the case that only states for one specific j. m pair contribute. Inserting (10. 2 |ˆ) r r where the factors i and −1 have been introduced for convenience. Accordingly. the orthonormality property (6. there exist two independent solutions (10.415.420)   g1 (r) X (r) 1 1 − Yjm (j − 2 . d(r) are coupled and b(r). using (10.418) ∂r + j+ r 1 2 c(r) + [ m + V (r) − ] b(r) = 0 i j+ ∂r − r 1 2 b(r) + [ −m + V (r) − ] c(r) = 0. g2 (r) ∂r + 1 2 f1 (r) − [ + m − V (r) ] g1 (r) = 0 (10.416).421)   g2 (r) X (r) 1 1 − Yjm (j + 2 .344 Relativistic Quantum Mechanics In general.412).157).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 (10.419) Obviously. 2 |ˆ)   =  (10. 10. 10.

422) and (10.424) Ψ(xµ ) ≈  r 0 1 i. except for the opposite sign of the term (j + 1 ). 2 2 2 2 2 2 = j − 1 2 and.e. n − 1. . We assume the 4-vector potential of pure Coulomb type (10.422) describes the states 2p 1 .367) which is spherically symmetric such that equations (10. Dirac Particle in Coulomb Field .405). etc. In the non-relativistic limit. 1.422) corresponds to states   f1 (r) Yjm (j + 1 .423). = 0. namely. Behaviour at r → 0 We consider first the behaviour of the solutions f1 (r) and g1 (r) of (10. together with the appropriate boundary conditions at r = 0 and r → ∞.425) r 0 covers states with angular momentum 1s 1 . (10. and obtains finally a polynomial function p(r) such that rγ exp(−µr)p(r) solves (10. The solution of (10. determining wave functions of the type (10. to states with angular momentum = j + 2 .422). 2s 1 . We note that (10. 1 |ˆ)  r i 2 2 . 2 |ˆ)  r Ψ(xµ ) ≈  . can be written ∂r − j+ r j+ r 1 2 g1 (r) + Ze2 f1 (r) r Ze2 g1 (r) r = 0 ∂r + 1 2 f1 (r) − = 0. Hence.422) near r = 0. 3d 5 . 3p 1 .423) are identical. . in the non-relativistic limit wave functions   f2 (r) 1 1 i Yjm (j − 2 . .409).420) is the large component and X is the small component. . 3d 3 . 10. (10.Wave Functions We want to determine now the wave functions of the stationary states of hydrogen-type atoms which correspond to the energy levels (10. Similarly. Equation (10.405). one demonstrates first that the wave function at r → 0 behaves as rγ for some suitable γ.422. (10. the ones given in (10. . the radial wave functions for Dirac particles in the potential (10.. Φ in (10.10: Dirac Particles in Electromagnetic Field 345 Equations (10. 3s 1 . 3p 3 . enforcing the polynomial to be of finite order leads to discrete eigenvalues .10. (10. (10. .e.422) determines solutions of the form (10.423) apply for V (r) = −Ze2 /r. 2 2 2 (10. one demonstrates then that the wave functions for r → ∞ behaves as exp(−µr) for some suitable µ.421).422) follows in this case from the same procedure as that adopted for the radial wave function of the non-relativistic hydrogen-type atom. According to this procedure. the 2 equations determine. .. . 2p 3 . According to the discussion of the spectrum (10.422). for small r. correspodingly the states We consider first the solution of (10. 2. Hence. etc. i.405) of the relativistic hydrogen atom the corresponding states have quantum numbers n = 1.426) .422).420).

10. We have. g1 ∼ exp(± m2 − 2 r).346 Setting f1 (r) yields γ b rγ−1 − (j + 1 ) b rγ−1 + Ze2 a rγ−1 2 γ a rγ−1 + (j + 1 ) a rγ−1 − Ze2 b rγ−1 2 or γ + (j + 1 ) −Ze2 2 2 Ze γ − (j + 1 ) 2 a b →0 Relativistic Quantum Mechanics ∼ a rγ . 10.266. according to (10.433) = = ( m2 − 2 ) g1 (r) ( m2 − 2 ) f1 (r) (10.427) = = 0 0.422) becomes ∂r g1 (r) ∂r f1 (r) Iterating this equation once yields 2 ∂r g1 (r) 2 ∂r f1 (r) 1 .420). (10.427) makes only the positive solution 2 possible. hence. assume the r-dependence in (10. r2 (10.432) = = − ( − m. (10. for the particle density holds then ρ(r) ∼ |rγ−1 |2 = Behaviour at r → ∞ For very large r values (10.428) = 0.434) √ The solutions of these equations are f1 . The assumed r-dependence in (10. (10. g1 (r) →∞ ∼ e−µr .267.435) . Such r-dependence 2 would make the expectation value of the potential r2 dr ρ(r) 1 r (10.429) This equation poses an eigenvalue problem (eigenvalue −γ) for proper γ values.430) 2 Note that the exponent in (10. becomes imaginary. in case j + 1 )2 < Ze2 .431) infinite since. Only the exponentially decaying solution is admissable and.427). µ = m2 − 2 (10. One obtains γ = ± (j + 1 )2 − Z 2 e4 . hence. g1 (r) →0 ∼ b rγ (10.427) with 1 γ = (j + )2 − Z 2 e4 . we conclude f1 (r) →∞ ∼ e−µr . ) f1 (r) ( + m ) g1 (r) (10. determined that the solutions f1 (r) and g1 (r). for small r.

422) for the Coulomb potential V (r) = −Ze2 /r We assume a form for the solution which is adopted to the asymptotic solution (10.442) which leads to a partial cancellation of the asymptotically dominant terms.10: Dirac Particles in Electromagnetic Field For bound states holds 347 < m and. g1 (r) = φ1 (r) − φ2 (r) ˜ (10. Accordingly.434) √ m + a e−µ r .10. We also introduce the new variable ρ = 2µ r .436).439) − m− e g1 (r) ˜ where µ is given in (10.422 ) leads to √ √ j+1 Ze2 ˜ 2 − m − [∂r − ] g1 + m + ˜ f1 + r √r √ ˜ (m − ) m + g1 − (m − ) m + f1 = 0 ˜ 1 2 √ √ j+2 ˜ Ze m + [∂r + ] f1 + m − g1 − ˜ r r √ √ ˜ (m + ) m − f1 + (m + ) m − g1 = 0 ˜ (10.440) (10. (10.441) correspond to (10.437) (10.445) .s. hence.441) The last two terms on the l. √ g1 (r) = − m − a e−µr .437) where they ˜ ˜ cancelled in case f1 = g1 = a. Let us consider then for the solution of (10. ρ m+ ρ [∂ρ − (10. Solution of the Radial Dirac Equation for a Coulomb Potential To solve (10. In the present case the functions f1 and g1 cannot be chosen ˜ ˜ identical due to the terms in the differential equations contributing for finite r. without loss of generality we can choose ˜ f1 (r) = φ1 (r) + φ2 (r) .444) (10. (10. Equation (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 .435 ).434) results in √ √ (m − ) m + a − (m − ) m + a √ √ (m + ) m − a − (m + ) m − a which is obviously correct.438) (10. we set f1 (r) g1 (r) = = √ √ ˜ m + e−µr f1 (r) −µr = = 0 0 (10.440) and (10.h. However. µ is real.436) f1 (r) = Insertion into (10. of both (10.

(10.426–10.447) of the form n φ1 (ρ) = ργ s=0 n γ s=0 αs ρ2 βs ρ2 (10.451) follows βs−1 = = Ze2 s+γ + √ m2 − s+γ − βs + m2 Z 2 e4 m2 − 2 − (j + 1 )2 2 2 √ Ze m2 − 2 2 s+γ − βs (10. = m− Ze2 βs s + γ − √m2 − 2 2 (10.446) (10. Using (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) .452) From (10.430) one can write this βs = s+γ − 2 √ Ze m2 − 2 s (s + 2γ) βs−1 . (10.451) 1 √mZe − (j + 2 ) 2 2 αs .453) (s + γ)2 + Z 2 e4 − (j + 1 )2 2 2 √ Ze m2 − 2 βs .449) into (10. Inserting (10. 10.447) leads to (s + γ) αs + (j + 1 ) βs − √ 2 s Ze2 m2 − 2 αs (10. 10.447) We seek solutions of (10.448.430) which conform to the proper r → 0 behaviour determined above [c.430)].449) for γ given in (10. 10.450) follows Ze2 m2 − 2 βs − βs−1 =0 (10.446.450) −√ mZe2 m2 − 2 βs ρs+γ−1 = 0 mZe2 m2 − αs (s + γ) β2 + (j + 1 ) αs + √ 2 s 2 +√ From (10.446 .f.448) φ2 (ρ) = ρ (10.

456) and (10. For example. c. 2γ + 1.430. According to the definitions (10.449) and (10. with the associated Laguerre polynomials L(α) = F (−n. is an integer.448. x) . 2.456. 3d 3 holds n = 1. 10. α + 1. namely. x) = 1 + a a(a + 1) x2 x + + . (10. This requires that all coefficients αs and βs must vanish for s ≥ n for some n ∈ N. β0 j+ 1 2 (10. m (n ) = .. 2γ + 1.456) − so 2 √mZe m2 − 2 (−so )(1 − so )(2 − so ) . n It holds φ1 (ρ) φ2 (ρ) = = ργ F (−so . conclude 2 2 2 that the polynomials in (10.460) (10.457) for βs and αs imply that so must then be an integer. hence. (s − so ) β0 s! (2γ + 1)(2γ + 2) · · · (2γ + s) (10. (s − so ) β0 s! (2γ + 1)(2γ + 2) · · · (2γ + s) (10.457) with the confluent hypergeometric functions F (a.10. . equivalently. ρ) .405) allows one to identify n = n − j + 2 which.. 2. . 10.e.461) In order that the wave functions remain normalizable the power series (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 ) . .458) or. 10.457) One can relate the polynomials φ1 (ρ) and φ2 (ρ) defined through (10. We can. = From (10.448. . i. n = 0. in fact.459) − 2 √mZe m2 − 2 (10. . 1.452) follows j+ αs = 1 2 349 Ze2 m2 − 2 −γ (10. . .455) this confinement of so implies discrete values for . . ρ) so β0 ργ F (1 − so ..455) are finite. 1. so = n . Comparision with (10.449) must be of finite order. 3p 1 . c c(c + 1) 2! (10. The expressions (10.462) Z 2 e4 1 + 1 (n + (j+ 2 )2 − Z 2 e4 )2 This expression agrees with the spectrum of relativistic hydrogen-type atoms derived above and 1 given by (10.405) and the ensuing so values ( 10.461 for values given by (10. for the states 2p 1 . .10: Dirac Particles in Electromagnetic Field Defining so = √ one obtains βs = = . .405). 10.

Berlin. Greiner. 9. The coefficients β0 in (10.421)] ∞ dr (|f1 (r)|2 + |g1 (r)|2 ) = 1 (10. = j + 2 .460.421). (Springer. 2µr) (10. are two-dimensional vectors determined through the explicit form of the spin-orbital angular momentum states Yjm (j ± 1 .147. which involve the confluent hypergeometric functions in (10.460.463) where Φ and X .463) and yield [note the 1/r factor in (10. j. (10. = j + 2 . They are described through quantum numbers n.465) (10.158) of the latter states absorb the angular integral in (10. is time-independent.410) of the stationary state wave function the density ρ(xµ ) of the states under consideration.267). given by expression (10. (n + γ)m N (2µr)γ−1 e−µr − κ F (−n . can follow the procedure adopted for the wave functions of the non-relativistic hydrogen atom and will not be carried out here.460.m (j − 1 . as in (10. 2γ + 1. (10. 10.439).350 Relativistic Quantum Mechanics Altogether we have determined the stationary states of the type (10.469) This formula has been adapted from ”Relativistic Quantum Mechanics” by W. and (10.461). Due to the form (10.421) correspond to non-relativistic states with orbital angular momentum 1 1 = j + 2 . 2 . 1 |ˆ) 2 2 r G1 (r) Yj.461). 10. g1 (r) determined by (10.6. j.157. 6. 1 |ˆ) 2 2 r κ = j+ 1 2 (10. Sect.438.442). m.148). The normalization integral is then ∞ 0 π 2π r2 dr 0 sin θdθ 0 dφ (|Φ(r)|2 + |X (r)|2 ) = 1 (10. 2µr) (2µ) 2 Γ(2γ + 1) 3 N = and m 4m )Γ(2γ + n + 1) (n +γ)m (n +γ)m (10.421) with radial wave functions f1 (r). 2γ + 1. The complete wave function is given by the following set of formulas 1 Ψ(n. The orthonormality 2 2 r properties (6. m|xµ ) = e−i t iF1 (r) Yj.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 . 1 |ˆ) in (6. 1990).m (j + 1 . 6. 10.461) are to be chosen to satisfy a normalization condition and to assign an overall phase.464) 0 The evaluation of the integrals. The wave functions (10. 10. where2 F± (κ|r) = G1 (r) = F+ (κ|r) .466) F1 (r) = F− (κ|r) .467) ± n F (1 − n .

471) (10. etc. j.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. in the non-relativistic limit. obtained closed expressions for the wave functions of all the stationary bound states of relativistic hydrogen-type atoms. 3d 5 . The radial wave functions f2 (r) and g2 (r) in (10. tracing all steps which lead from (10.421) are governed by the radial Dirac equation (10. this wavefunction has an orbital angular momentum quantum number = j − 1 and. 2p 3 .422) to (10. hence. = j − µ 1 2 .465–10. G2 (r) = F+ (κ|r) . 3s 1 .470) r 0 Obviously. become   f2 (r) 1 1 i Yjm (j − 2 .10. We have.m (j + 1 . 1 |ˆ) 2 2 r G2 (r) Yj. (10.469).467–10. 2s 1 .m (j − 1 .10: Dirac Particles in Electromagnetic Field 351 We want to consider now the stationary states of the type (10. where F± (κ|r) are as given in (10.472) F2 (r) = F− (κ|r) . 2 accordingly. not not 2 2 2 2 2 2 covered by the wave functions given by (10. m|x ) = e −i t iF2 (r) Yj.469) that the following wave functions result Ψ(n.421) which. describes the complementary set of states 1s 1 . One can verify. 3p 3 . 2 |ˆ)  r Ψ(xµ ) ≈ e−1 t  . 1 |ˆ) 2 2 r κ = −j − 1 2 (10.469). .

352 Relativistic Quantum Mechanics .

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

usually expressed 2 as product of rotations and of functions of Euler angles α. however.. 0)T the transformations are (1) a(i β e2 ) = b(i β e2 ) = ˆ ˆ 2 dmm (β) (1) = cos β 2 sin β 2 − sin β 2 cos β 2 .8) We like to stress that there exists. and the Klein–Gordon equation in the spinor representation. (11. Ψ2 (xµ ) and Ψ3 (xµ ). Here | 1 . 2 2 2 ˜ Since a(z) acts on the first two components of the solution Ψ of the Dirac equation. For ϑ = (0. namely. In this case a(z) and b(z) are identical and read a(i ϑ) = b(i ϑ) = exp 1 − ϑ·σ 2 . ϑ ∈ R3 . A First Characterization of the Bispinor States We note that in case w = 0 the Dirac transformations are pure rotations.243). a space of vectors state1 for which holds state2 state 1 ∼ | 1 . θ ∈ R3 . − 1 2 2 z = iϑ. to the rotational transformations of spin. Ψ4 (xµ ) in case z = w + iϑ for w = 0. Ψ4 (xµ ) transform according to ˜ ˜ ˜ and Ψ b(z).e. i. β. express 4-vectors Aµ . to 1 ∗ ∗ z ·σ (11.354 Spinor Formulation and b(z) act. 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. the so-called spinor space.7) where “∼” stands for “transforms like”. neutrino equation. (11. In this case holds a(z) = b(z) ˜ 1 (xµ ).9) a∗ (z) = exp 2 . i. ϑ)µ ν and a(z).. −1 2 2 z = iϑ. Ψ2 (xµ ) transform according to a(z) whereas Ψ3 (xµ ). b(z). Aν . γ (see Chapter 5). in 2 fact.e. finally. −1  2 2  |1. +1  2 2 |1. for z = iϑ. a distinct difference in the transformation behaviour of ˜ ˜ ˜ ˜ Ψ1 (xµ ). ϑ ∈ R3 . +1 2 |1. This characterization allows one to draw conclusions regarding the state space in which a(z) and b(z) operate. the operator ∂µ and the Pauli matrices σ in the new representation and. β. Relationship Between a(z) and b(z) The transformation b(z) can be related to the conjugate complex of the transformation a(z).6) as given by (5. are elements of SU(2) and correspond. formulate the Dirac equation.5) The transformations in this case. + 1 2 2 state 2 ∼ | 1 . ± 1 represents the familiar spin– 1 states.1 states as described by Dm m (ϑ). will establish the map between L(w. (11. ϑ ∈ R3 (11.

18) with a(z) given by (11. however. that the transformation (11. one can show −1 −1 −1 = −σ1 = σ2 = −σ3 ∗ = −σ1 ∗ = −σ2 σ2 −1 (11. therefore. hold −1 = 1 One notices then. −1 = ∗ −σ3 = −σ ∗ .15) (11. Spatial Inversion ˜ One may question from the form of S(z) why the Dirac equation needs to be four-dimensional.224)] ∗ σ1 = σ1 . but leave rotation angles ϑ unaltered. σ2 . one can express a∗ (z) −1 = exp 1 − z ∗· σ 2 = b(z) . (11. Ψ (xµ ) even though these pairs of ˜ ˜3 ˜4 featuring the components Ψ1 (x ˜ 2 components transform independently of each other.10. ∗ σ 2 = − σ2 . (11.14) .12) To prove (11. This demonstrates that a(z) is the ˜ transformation which characterizes both components of S(z).f. 11. (11. One can readily verify [c.1) can be written in the form ˜ S(z) = a(z) 0 0 a∗ (z) −1 (11. a result to be used further below. −1 given by (11.4) and .16) σ = −σ ∗ .13) Explicit matrix multiplication using (5. The effect of inversion on Lorentz transformations is. in fact. a∗ (z) −1 = exp 1 ∗ z σ∗ 2 −1 . Ψ (xµ ) as well as Ψ (xµ ). using f (a) −1 = f ( a −1 ). −1 a∗ (z) −1 (11. σ3 ).12) yields σ1 σ3 or in short σ Similarly. 1. µ ).1: Lorentz Transformation of Dirac Bispinor ∗ ∗ ∗ where σ ∗ = (σ1 . .10) From this one can derive b(z) = where = 0 1 −1 0 . 11. 355 (11. (11.17) We conclude.2.11) one first demonstrates that for and −1 as given in (11.12) does.11. Hence.12).224. ∗ σ 3 = σ3 . 11. The answer lies in the necessity that application of spatial inversion should transform a solution of the Dirac equation into another possible solution of the Dirac equation.11) 0 −1 1 0 = . that they alter w into −w. (5.

22) which follows from the fact that for any complex. is an element of SU(2). for particles like the neutrinos which do not obey inversion symmetry two components of the wave function are sufficient.2.24) = etr(M ) (11. C) defined in (11.C) and SO(3.2 Relationship Between the Lie Groups SL(2.A. det(M ) = 1 } ..224)] tr( σj ) = 0 .1) We have pointed out that a(iϑ).. Amsterdam.20) ˜ Obviously. j = 1. C) = { M. which describes pure rotations. does not suffice for particles like the electron which obey inversion symmetry. This implies  ˜ Ψ1 ˜  Ψ2 P  ˜  Ψ3 ˜ Ψ4  ˜ Ψ3 ˜   Ψ4  =    Ψ1 ˜ ˜ Ψ2     . can be found in G. In fact.10. B¨uerle and E.  (11. the transformations a(z) and b(z) become interchanged. The transformation S(z) in the transformed space is then ˜ P 1.G.21) together with matrix multiplication as the binary operation forms a group. 2 . based on the Jordan–Chevalley theorem. (5. P −1 = P. (11.e. Part (Elsevier. 2 = 1 i. However. One can verify this by evaluating the determinant of a(z) det( a(z) ) = det e 2 z·σ 1 (11. de Kerf Lie a Algebras. hence. However. a(w + iϑ) for w = 0 is an element of SL(2.f. The proof of this important property is straightforward in case of hermitian M (see Chapter 5). (11.3. Exercise 1. 2.23) Exercise 11.A.21) = etr( 2 z·σ) = 1 1 (11.1: Show that SL(2.356 Spinor Formulation ˜ Let P denote the representation of spatial inversion in the space of the wave functions Ψ. the Lorentz invariant equation for neutrinos is only 2–dimensional. 11. the space spanned by only two of the components of Ψ is not invariant under spatial inversion and. 1990).e.19) i. M is a complex 2 × 2–matrix. 3 . ϑ ∈ R3 . non-singular matrix M holds2 det eM and from [c. For the general case the proof. ˜ ˜ P S(w + iϑ) P = S(−w + iϑ) = b(z) 0 0 a(z) . Obviously.

σ1 .. Starting point is a bijective map between R4 and the set of two-dimensional hermitian matrices defined through M (Aµ ) = σµ Aµ where   (11.11.2. one can provide a simple expression for the inverse of M (Aµ ) M = M (Aµ ) ↔ Aµ = 1 tr M σµ 2 .76).29) where we used (10.C) and SO(3. (11. The function M (Aµ ) is bijective. −→ µ Consistency of (11.30) Since the components of Aµ are real. The following important property holds for M (Aµ ) det ( M (Aµ ) ) = Aµ Aµ which follows directly from (11.28) and (11. M (Aµ ) according to (11.30). (11. 1 0 0 −1    .31) Exercise 11. Demonstrate that (11. (11. In fact. 0 −i i 0 .27) −→ Straightforward transformation into another frame of reference would replace σµ by σµ . one wishes that the definition (11.26) (11. Noting that for covariant vectors according to (10.135)] this transformation behaviour. σ2 .28)   σµ =   1 0 0 1 . (11.27) requires then σ µ Aµ Lν µ σµ = σν (11.2: Show that 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 .25) can also be written M (Aµ ) = A0 + A3 A1 − iA2 A1 + iA2 A0 − A3 . the matrix M (Aµ ) is hermitian as can be seen from inspection of (11. in fact. i.2: Lie Groups SL(2. σ3 are hermitian. σ1 .  (11.1) Mapping Aµ onto matrices M (Aµ ) 357 We want to establish now the relationship between SL(2.32) . in a transformed frame should hold Lρ ν σ µ Aµ σµ A µ .26) or from the fact that the matrices σ0 . σ3 provide a linear–independent basis for the space of hermitian 2 × 2–matrices.25) σ0 σ1 σ2 σ3 The quantity σµ thus defined does not transform like a covariant 4–vector. C) and the group L↑ of proper. (11. σ2 . We will prove below [cf.75) holds aν = Lν µ aµ one realizes that σµ transforms inversely to covariant 4–vectors. or+ thochronous Lorentz transformations.e. Argue why M (Aµ ) = σµ Aµ provides a bijective map.25) of the matrix M (Aµ ) is independent of the frame of reference. 0 1 1 0 .31) holds.

37) which implies that (11.f. In fact.31). (11. C) . (11.25) Aµ = which allows us to express finally A µ = L(a)µ ν Aν . 11.3: Show that L(a)µ ν defined in (11.32.36) The suitable A µ can readily be constructed using (11.37) defines actually a Lorentz transformation.38) .40) 1 tr a σν a† σµ 2 Aν (11. C) and for any Aµ there exists an A µ such that M (A µ ) = a M (Aµ ) a† . (11. a ∈ SL(2. + .33) det(M ) = = det( a M a† ) = det(a) det(a† ) det(M ) [det(a)]2 det(M ) = det(M ) .33) to M (Aµ ) describing the action of the transformation in terms of transformations of Aµ . C) defines the transformation [c. Due to det(a) = 1 the transformation (11.31)] 1 a Aµ −→ A µ = tr a M (Aµ ) a† σµ 2 Because of (11. any a ∈ SL(2.34) where we used the properties M † = M and (a† )† = a.37) using (11.33) conserves the determinant of M . (11. L(a)µ ν = 1 tr a σν a† σµ 2 .35) We now apply the transformation (11.2. (11.35) holds for this transformation A µ Aµ = Aµ Aµ (11. Accordingly. In fact. it holds for the matrix M defined through (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† .33) This transformation conserves the hermitian property of M since (M )† = ( a M a† )† = (a† )† M † a† = a M a† = M (11.39) Exercise 11. for any a ∈ SL(2. (11. The linear character of the transformation becomes apparent expressing A µ as given in (11.40) is an element of L↑ .

2 (11.2: Lie Groups SL(2.α.46) which yields ¯ Lµ ρ = = = 1 1 Γ11 Γ11 + Γ22 Γ22 + Γ12 Γ21 + Γ21 Γ12 = tr ΓΓ 2 2 1 1 tr a† σµ a1 a2 σρ a† = tr σµ a1 a2 σρ a† a† 1 2 2 1 2 2 1 tr a1 a2 σρ (a1 a2 )† σµ = L(a1 a2 )µ ρ .1) defined through L(a)µ ν [cf. ¯ Lµ ρ = L(a1 )µ ν L(a2 )µ ρ = L( product in SO(3. (11.C) )µ ρ (11.δ 1 4 Aαβγδ Γβα Γγδ α.β γ. C) and of SO(3.1) For this purpose one writes using tr(AB) = tr(BA) L(a1 )µ ν L(a2 )µ ρ = ν a1 a2 product in SL(2.1) L(a)µ ν provides a homomorphism 359 We want to demonstrate now that the map between SL(2.1).40)] respects the group property of SL(2.e.δ (11.β γ.41) 1 tr a1 σν a† σµ 1 2 1 tr σν a† σµ a1 1 2 1 tr a2 σρ a† σν 2 2 1 tr a2 σρ a† σν 2 2 ..C) and SO(3.43) (σν )αβ Γβα Γγδ (σν )δγ = ν. i. β δ = δ = = δ = δ 1 2 = 2 = 1 Aαβγδ (11. 1 ¯ and using the definition of Lµ ρ in (11.47) This completes the proof of the homomorphic property of L(a)µ ν .41) results in 1 ¯ Lµ ρ = 4 where Aαβγδ = ν Γ = a2 σρ a† 2 (11.45) One can demonstrate through direct evaluation   2 α =    2 α =  2 α = =   2 α =    0 else β β γ γ = = = = γ = γ = 1. β 2. C) and SO(3.11. . (11.44) (σν )αβ (σν )δγ . (11.42) = ν Defining Γ = a† σµ a1 .

J1 .51) Insertion of (K1 )µ ν as given in (10.53) results in the condition σ 0 A1 − σ 1 A0 = ( κ 1 σ µ + σ µ κ † ) Aµ . 0.e. J3 of the Lorentz transformations Lµ ν in the natural representations.2) which correspond to the generators K1 . 0. 0.49) we obtain using (11. 0) ) σ0 A 1 − σ1 A 0 (11.59) = = = = κ1 σ 0 + σ 0 κ† = κ 1 + κ† 1 1 κ1 σ 1 + κ1 σ 2 + κ1 σ 3 + σ 1 κ† 1 σ 2 κ† 1 σ 3 κ† 1 (11. 1. J Spinor Formulation The transformations a ∈ SL(2.360 Generators for SL(2. 0.52) and (11. Because of the condition det(a) = 1 only six independent real numbers actually suffice for the definition of a. 0. For the r. (11.49). −A0 . M (Aµ + (K1 )µ ν Aν ) = = M (Aµ ) + M (Aµ ) − M ( (−A1 . of (11. 1) − σ1 σ0 0 0 One can verify readily that 1 κ1 = − σ 1 2 (11. 0). of (11. 0.s.52) where we employed (11. To obtain the generator of a(z) corresponding to the generator K1 . One expects then that six generators Gj and six real coordinates fj can be defined which allow one to represent a in the form   6 a = exp  j=1 fj Gj  .54) This reads for the four cases Aµ = (1. 1. K3 . (11.47. 10. 1 (11.50) (11. C) which correspond to K.49) assuming (note that g µ ν is just the familiar Kronecker δµν ) Lµ ν a = = gµν + 1 + 1 (K1 )µ ν κ1 .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 ) .55) (11. correspondingly. 0. i.36) M (Lµ ν Aν ) = a M (Aµ ) a† (11. denoted below as κ1 . correspond to the generators given by (10. noting the linearity of M (Aµ ).58) .25) in the last step. 0).h..48) We want to determine now the generators of the transformation a(z) as defined in (11. Aµ = (0. To this end we consider infinitesimal transformations and keep only terms of zero order and first order in the small variables. 0). 1 (11.h. C) as complex 2 × 2–matrices are defined through four complex or. K2 .56) (11.48). we write (11.53) Equating (11. eight real numbers.s. Aµ = (0.48) yields for the l. (11. Aµ = (0. J2 .57) .

− 1 . hence. λ2 . C) correspond to the generators K and J of Lµ ν .63) where we extended the summation convention of 4-vectors to spinors.1) acts on 4–vectors Aµ .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. Similarly. C) acts on spinors φα ∈ C2 (characterized below) 1 (11. λ3 corresponding to J1 . ϑ) = ew·K + ϑ·J ∈ SO(3. 2 (11. 2 2 φ1 φ2 ∈ C2 . def α = 1.60) of a ∈ SL(2. Here a(z) = ( aα β ) = describes the matrix a(z). state that the following two transformations are equivalent L(w.61) This identifies the transformations a(z = w − iϑ) as representations of Lorentz transformations. a1 1 a1 2 a2 1 a2 2 (11.3: Spinors 361 obeys these conditions. 2 λ = i σ. κ3 of a(z) corresponding to K2 .60) We can. w describing boosts and ϑ describing rotations. 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 .64) .62) We denote the general (z = w + iϑ) transformation by φ α = aα β φβ = aα 1 φ1 + aα 2 φ2 . J3 are given by 1 κ = − σ. K3 and λ1 . a(w − iϑ) = e− 2 (w − iϑ)·σ ∈ SL(2. one can show that the generators κ2 .4: Show that the generators (11. 11. the so-called contravariant spinors. J2 . (11. Exercise 11. 2 (11.11.2.

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

11.72)] φα φ α = = = = αβ αβ φβ φβ 11 21 11 21 12 22 12 22 (11. 11.74)].74) = = 0 1 −1 0 0 −1 1 0 (11.75) is given by φα = as can be readily verified. The transformation behaviour of φα according to (11.76) αβ αβ The scalar product (11.67) will be expressed as φα χα = φ1 χ1 + φ2 χ2 = φ1 χ2 − φ2 χ1 . we will express (11. g µν of the Minkowski space [cf. 10.70) is φ1 φ2 −1 363 = = −1 φ1 φ2 .71.68).77) is Lorentz invariant. (10. (11. Accordingly. 11.77) (11.70) and (11. For this scalar product holds φα χα = −χα φα . (11.72) 0 −1 1 0 Exercise 11. 11. The inverse of (11.11. 11.79) = a −1 T a κβ . (11.63. (10.81) . In the transformed frame holds φ α χα = aα β αγ aγ δ δκ φβ χκ .72) in a notation analogous to that chosen for contravariant and covariant 4–vectors [cf.71) (11.73.10.78) aβ γ γδ φδ (11.73) (11. Proof that φα χα is Lorentz invariant We want to verify now that the scalar product (11.3.80) One can write in matrix notation aα β γ δκ αγ a δ αβ (11. −1 connecting contravariant and covariant spinors play the role of the metric tensors gµν .3: Spinors as can be verified from (11.69.1: Show that for any non-singular complex 2 × 2–matrix M holds M −1 = det(M ) M −1 T The matrices .75) (11.69.

The transformation ˙ behaviour of φα is ˙ φ α = (aα β )∗ φβ ˙ (11.89) such that (11. from (11.86) A concise notation of the conjugate complex spinors is provided by ˙ (φα )∗ = φα = φ1 ˙ φ2 ˙ ˙ (11. the ˙ conjugate complex spinors φα need to be considered separately from the spinors φα . Obviously.87) which we will employ from now on.90) extending the summation convention to ‘dotted’ indices.2) and (11.364 Using (11. hence.84) (11.80) φ α χα = φβ χβ . a∗ (z) provides a representation of the Lorentz group which is distinct from that provided by a(z).81) yields aα β and.. As discussed above.88) reads ˙ ˙ φ α = aα β φβ ˙ ˙ (11.82) and with f (A)T = f (AT ) for polynomial f (A) a −1 T = e− 2 z·(σ 1 ∗ )T = e− 2 z·σ = a−1 1 (11. We denote ˙ (aα β )∗ = aα β ˙ (11.85) φ1 φ2 ∗ ∗ . ˙ (11. σ † = (σ ∗ )T = σ. ˙ We also define covariant versions of φα φα = (φα )∗ . 2.91) .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. (11.83) Here we have employed the hermitian property of σ.e. i. k = 1. it holds φk = (φk )∗ .88) which one verifies taking the conjugate complex of (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. Insertion of (11.83) into (11.63). Hence.

A comparision of (11. in matrix notation.18) ˜ 3 . A tensor tα1 α2 ···αk β1 β2 ···β (11. 11. (11.95) ˙ where αβ and αβ are the real matrices defined in (11.97) implies then that φα transforms like Ψ ˙ φ1 ˙ φ2 ˙ transforms under rotations (z = iϑ) like a spin– 1 state | 1 .93) (11.75.100) . 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. (11. 11.97) The transformation.4: Spinor Tensors 365 The relationship between contravariant and covariant conjugate complex spinors can be expressed in analogy to (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 )   =   = .11.4 Spinor Tensors ˙ ˙ ˙ We generalize now our definition of spinors φα to tensors..96) is Lorentz invariant.73. + 1 2 2 2 1 transforms under rotations (z = iϑ) like a spin– 2 state | 1 .94) (11. is governed by the operator a∗ (z) −1 which arises in ˜ the Lorentz transformation (11. Ψ4 . For the spinors φα and χα thus ˙ defined holds that the scalar product ˙ φα χα = φ1 χ1 + φ2 χ2 ˙ ˙ ˙ ˙ ˙ (11. a property which is rather evident.92) (11.76) φα ˙ ˙ φα αβ ˙ ˙ αβ ˙ ˙ = = = = αβ ˙ ˙ αβ ˙ ˙ αβ αβ φβ φβ ˙ ˙ (11. − 1 . a∗ (z) −1 accounting for the ˜ transformation behaviour of the third and fourth spinor component of Ψ.75).99)  Ψ3 (xµ )   χ ˙ (xµ )  ˜ χβ (xµ ) ˙ 1 ˜ χ2 (xµ ) Ψ4 (xµ ) ˙ 11.e. i.18) of the bispinor wave function Ψ. one can state ˜ and (11.

4. this transformation behaviour is in harmony with the tensor notation adopted. (11. i. with ˙ contravariant indices αβ. k. This task implies that we seek a tensor.e..25] M (Aµ ) = σµ Aµ . the transformation bevaviour of a tensor tαβ reads in spinor and matrix notation t αβ = aαγ aβδ tγδ .e. i. An example of a tensor is αβ and tions.m aj a∗ t km m .106) Exercise 11.107) which reads in spinor notation [cf. it holds αβ .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. using conventional matrix indices j.102) = . ˙ ˙ ˙ ˙ (11. An obvious candidate is [cf. According to (11.104) Indices on tensors can also be lowered employing the formula tα β = and generalizations thereof. (11.. m.m aj akm t m (11. the elements of which are linear functions of Aµ . αβ ˙ αγ t ˙ γβ (11..25) the tensor is explicitly Aαβ = ˙ A1 1 A1 2 ˙ ˙ A2 1 A2 2 ˙ ˙ = A0 + A3 A1 − iA2 A1 + iA2 A0 − A3 .101) An example is the tensor tαβ which will play an important role in the spinor presentation of the Dirac equation.108) Obviously.103) A αβ = aα γ aβ δ Aγ δ .103) Similarly. 11.106). i.1: Prove equation (11. αβ = αβ (11. (11.e.109) .105) This tensor is actually invariant under Lorentz transforma= αβ .102. We had demonstrated that M (Aµ ) transforms according to M = M (Lµ ν Aν ) = a M (Aµ ) a† (11. tjk = a t aT jk = . The 4–vector Aµ in spinor form We want to provide now the spinor form of the 4-vector Aµ . tjk = a t a† jk ˙ ˙ ˙ ˙ (11. ˙ ˙ ˙ ˙ (11. This tensor transforms according to t αβ = aα γ aβ δ tγ δ ˙ This reads in matrix notation. we want to express Aµ through a spinor tensor.

˙ 2 (11.4: Spinor Tensors One can express Aαβ also through Aµ Aα β = ˙ ˙ 367 A0 − A3 −A1 + iA2 −A1 − iA2 A0 + A3 . (11. (11.4.114) We finally note that the 4–vector scalar product Aµ Bµ reads in spinor notation Aµ B µ = 1 αβ ˙ A B αβ . Aµ 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. ∂µ in spinor notation The relationship between 4–vectors Aµ . Using (11. Hence.113) Aα β = ˙ (11.11. 11.111) A11 A12 ˙ ˙ A21 A22 ˙ ˙ ˙ 0 −1 1 0 (11.2: Prove that (11.110) one obtains Aα β = ˙ A0 − A3 −A1 − iA2 −A1 + iA2 A0 + A3 A0 + A3 A1 + iA2 A1 − iA2 A0 − A3 .116) . (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 .110) one can state ∂ αβ = Similarly.109.110) The 4-vectors Aµ .115) is correct.115) Exercise 11. (11. employing (11. (11.114) yields ∂αβ = ˙ ∂0 + ∂3 ∂1 + i∂2 ∂1 − i∂2 ∂0 − ∂3 .112) A22 −A21 ˙ ˙ −A12 A1 1 .

. one can conclude aσ µ ∗ −1 Lν µ σν 0 = .125) a (σ ) Equation (11. This is ˜ possible since γ µ can be expressed through σ µ .126) which is the complex conjugate of (11.124) constitutes the essential transformation property of σ µ and will be considered further.119) (11. . i.125) is equivalent to (11.123) (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 )∗ (11.118) yields ˜ γµ = ˜ Using (11.f.243) in the chiral representation expressing S(Lη ξ ) by (11.h. (11. σµ building on the known transformation behaviour of γ µ .. (11.18) and γ µ by (11. (11. . (11. holds also in the chiral representation.124) .368 σµ in Tensor Notation Spinor Formulation We want to develop now the tensor notation for σµ (11. 1 0 0 −1 .e. i. Note that (10.122) aσ µ −1 a−1 −1 (a∗ )−1 0 Lν µ (11. 0 1 1 0 . To obtain the transformation properties of σ µ we employ then (10.. Equation (10.26) and its contravariant analogue σ µ σµ = σµ = 1 0 0 1 1 0 0 1 . −1 0 0 1 (11. (11.243)].118) 0 −1 −1 0 0 i −i 0 For this purpose we consider first the transformation behaviour of σ µ and σµ .15) one can write σµ = and. Hence.121) One expects then that the transformation properties of σ µ should follow from the transformation properties established already for γ µ [c.s.124). of this equation is 0 ∗ (σ µ )∗ a and.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. (10. (11. σµ .120) σµ 0 . Comparision of (10.e. γµ = ˜ 0 µ )∗ (σ −1 0 σµ σµ 0 (σ µ )∗ −1 . We will obtain the transformation behaviour of σ µ .229) and (11.121). 0 −i i 0 .243) holds independently of the representation chosen. hence.124). hence.

2. 11.  . 3) σj −→ a(iϑ) σj a(iϑ) † = a(iϑ) σj a∗ (iϑ) T . under rotations the Pauli matrices transform like (j = 1.108) that the same transformation behaviour applies then for general Lorentz transformations. (11.. In fact.134) . (11. 1 0 0 −1 µ=3   .131) one can express the transformation behaviour of σ µ in the succinct form Lν µ σ µ = σ ν .g. One can. (11.132) 1 0 0 1 µ=0 . (11. (σ ∗ )T = σ yields using (11.4) with w = 0. e. It holds according to (11.118) (σ µ )11 (σ µ )12 ˙ ˙ (σ µ )21 (σ µ )22      and (σµ )11 (σµ )12 (σ µ )21 (σ µ )22 ˙ ˙      1 0 0 1 µ=0 ˙ ˙ ˙ ˙ =  (11. therefore.4).102.2) −1 (a∗ )−1 = e2z 1 ∗ ·(σ ∗ )T = e2z 1 ∗ ·σ ∗ T = [a∗ ]T .2) −1 369 (a∗ )−1 = −1 − 1 z ∗ ·σ ∗ 2 e = e− 2 z 1 ∗ · −1 σ ∗ = e2z 1 ∗ ·σ . −1 0 0 1 µ=3   . 0 i −i 0 µ=2 .16.127) Exploiting the hermitian property of σ.   =  .103) ˙ identifies σ µ as a tensor of type tαβ . state that σ µ in a new reference frame is σ µ = a σ µ a† (11.  (11. hence.11. transformations (11. 11.124) using (11.e. 0 −1 −1 0 µ=1 . i. (11.128) One can express.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.131) where a is given by (11. 11.107) to (11.130) We argue in analogy to the logic applied in going from (11. 11.2.. This transformation behaviour.133) Combining (11. 0 1 1 0 µ=1 . equation (11.129.4: Spinor Tensors One can rewrite (11. 0 −i i 0 µ=2 . 11.124) a σ µ [a∗ ]T Lν µ = σ ν . 2. according to (11.

. i. We can. 2. 1.135) This is the property surmised already above [cf.e. whereas ˙ account for the elements of the individual Pauli the ‘outer’ contravariant spinor indices.138) the ‘inner’ covariant spinor indices. Spinor Formulation (11. ˙ ˙ 0 −1 1 0 ˙ (11.141) σ αβ = σ0 − σ3 σ1 + iσ2 −σ1 − iσ2 σ0 + σ3  = 2  0 0 0 1 0 −1 0 0 . The ˙ desired change of representation(σµ )αβ −→ (σµ )αβ corresponds to a transformation of the basis of ˙ spin states f g f −→ = (11. and the covariant indices are moved outside.136) σαβ = = 2 ˙   σ1 − iσ2 σ0 − σ3 0 0 0 0 1 0 0 1 where on the rhs.136) results in the succinct expression 1 2 ˙ γδ ˙ (11.132) and (11.370 Inverting contravariant and covariant indices one can also state Lν µ σµ = σν ..110) to express σ αβ in terms 0 −1 1 0 0 0 0 1     (11.s.140) where we employed the expressions (11. state   ˙ ˙ ˙ ˙ (σ11 )11 (σ11 )12 (σ12 )11 (σ12 )12 ˙ ˙ ˙ ˙   ˙ ˙ ˙ ˙ ˙ γδ   (σ11 )21 (σ11 )22 (σ12 )21 (σ12 )22 ˙ ˙ ˙ ˙   . We will now consider the representation of σ µ . account for the 4-vector µ. corresponds to the transformation (σµ )11 (σµ )12 ˙ ˙ (σµ )21 (σµ )22 ˙ ˙ = 0 1 −1 0 (σµ )11 (σµ )12 ˙ ˙ (σµ )21 (σµ )22 and  −1 . σµ in which the contravariant indices are moved ‘inside’. δ. i.137) σαβ ˙ = δαγ δβ δ . =  σαβ ˙  ˙ ˙ ˙ ˙ (σ21 )11 (σ21 )12 (σ22 )11 (σ22 )12   ˙ ˙ ˙ ˙ ˙ ˙ ˙ ˙ 21 (σ )22 21 (σ )22 (σ21 ) (σ22 ) ˙ ˙ ˙ ˙ 21 22 Equating this with the r. ˙ ˙ In (11. This yields   1 0 0 1   0 0 0 0 σ0 + σ3 σ1 + iσ2  (11. of (11. the transformation is inverse to that of ordinary 4–vectors. in fact. matrices. ˙˙ (11. We can summarize that σ µ and σµ transform like a 4–vector. µ = 0. α..224) ˙ Using (11.114). 3 as well as through two spinor indices αβ.138) Note that all elements of σαβ are real and that there are only four non-vanishing elements. i. account for the 4–vector index µ. however.e.12) for of σµ yields together with (5.e.139) g −f g and.h. the submatrices correspond to tαβ spinors. hence. β. Each of the 4 × 4 = 16 matrix elements in (11.133) is characterized through a 4-vector ˙ index µ.29)]. γ. We want to express now σ µ and σµ also with respect to the 4–vector index µ in spinor form employing (11. (11.

138) one can conclude that the following property holds 1 2 ˙ γδ σ αβ ˙ = 1 2 σ αβ ˙ ˙ γδ = δαγ δβ δ . we write ˜ ˙ γµ = ˜ 0 (σ µ )αβ µ ) )∗ ((σ αβ 0 ˙ ˙ .115) γ µ Aµ ˜ = 0 ((σ µ )αβ )∗ Aµ ˙ 0 ˙ ∗ 1 γδ ˙ ˙ 2 ((σ )αβ ) Aγ δ (σ µ )αβ Aµ 0 ˙ ˙ αβ γ δ 1 ˙ 2 (σγ δ ) A ˙ = 0 .143) Combined with (11. which in ˙ (11.121) of γ ˜ ˜ ˙ basis |αβ whereas the element of γ µ given by σ µ is in the basis |αβ . Accordingly.142) = This can be expressed 1 2 σ αβ ˙ ˙ γδ = δαγ δβ δ . employing then transformation (11.140) to transform each of the four submatrices.144) results in the simple relationship γ µ Aµ = ˜ 0 Aα β ˙ Aα β 0 ˙ . For this purpose we start from the ˜ µ .11.145) Let Aµ be a covariant 4–vector. (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 ˙ ˙    .4: Spinor Tensors 371 and. (11.144) The Dirac Matrices γ µ in spinor notation We want to express now the Dirac matrices γ µ in spinor form. (11.  ˙ ˙ ˙ ˙       (11.147) . This expression implies that the element of γ µ given by σ µ −1 is in the expression (11. One can write then the scalar product using (11. ˙˙ (11.146) Exploiting the property (11. ˙˙ (11.

99) yields the Dirac equation in spinor form i ∂ αβ χβ ˙ i ∂ αβ φ ˙ The simplicity of this equation is striking. α ˙ = = m φα m χβ .149) (11.372 Spinor Formulation 11.226) ˜ i γ µ ∂µ Ψ(xµ ) = 0 i ∂ αβ ˙ i ∂ αβ 0 ˙ ˜ ˜ Ψ(xµ ) = m Ψ(xµ ) . ˙ (11. (11.148) ˜ Employing Ψ(xµ ) in the form (11.5 Lorentz Invariant Field Equations in Spinor Form Dirac Equation (11.147) allows us to rewrite the Dirac equation in the chiral representation (10.150) .

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

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

The vector in (12. ijk Jk .14) This follows from a · (a × b) = (a × b) · a = 0.13) It can be verified explicitly that its components commute with the Hamiltonian.12) (12. we will compute the hydrogen spectrum using the additional symmetry.9) (12. In order to understand the aforementioned extra degeneracy. (12. Kj ] = i ijk Kk . 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 . which have a negative energy E.10) (12. In the following we consider the bound states. We introduce the following new operators 1 A = (J + K).15) which can be proved after very considerable algebra. 2m r 1 x1 (J3 p2 − J2 p3 + i p1 ) − k .20) (12. in the subspace of the Hilbert space corresponding to a certain energy we can replace H by E. Therefore.17) (12. 2 1 B = (J − K).11.21) . Aj ] = i ijk Ak . (12. (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. which is valid even when a and b do not commute. 2 which can be shown to satisfy [Ai . 2m r 1 x2 (J1 p3 − J3 p1 + i p2 ) − k .19) (12.18) which complement the familiar angular momentum algebra of section 5. Jj ] = i [Ki .11) (12. 2m r 1 x3 (J2 p1 − J1 p2 + i p3 ) − k . For this purpose we first note that J · = 0. 2E (12. (12. Now we scale the eccentricity vector as follows K= − m . We will also need the following identity [4] 2 = 2H J2 + m 2 + k2 . 2m r (12.16) Through some algebra [4] the following commutation relations can be verified [Ki .3.8) is not a hermitian operator.

H = 0. . A. . This illustrates the effect of additional symmetries to the degeneracy structure of a quantum mechanical system.24) (12. can be seen to follow from the triangle inequality as applied to J = A + B. = 0. l. . 2 1 .26) (12. 1.22): A2 = a(a + 1) B 2 = b(b + 1) Following (12. . . . which is identical to the the direct sum of the Lie algebra of of two SO(3) (or SU (2)) algebras. The most extreme case of this is the quantum analogue of a classically chaotic system. . Noticing that A2 and B 2 have the same eigenvalues because of (12.14) we note that A2 − B 2 = J · = 0. . Using this equation the energy eigenvalues can be written in terms of the eigenvalues of A2 and B 2 operators. This implies that a = b. the energy eigenvalues are found to be E=− mk 2 1 .33) (12. we can read off the eigenvalues of A2 and B 2 from (12. 1. H B.15).21) and (12. a + b = n − 1}. .28) .29) (12.34) It follows that l has to have values in {0 = |a − b|. Chaos is described classically as exponential .22) (12. . a lack of symmetry implies a lack of degeneracy in the energy levels of a quantum mechanical system. Bj ] = 0. In contrast to the discussion above about extra symmetries. . . namely that J J > < A − B =0 A + B =2 A (12. . Spinor Formulation (12. In order to arrive at the spectrum a final bit algebra is needed A2 + B 2 = J 2 + K 2 m 2 = J2 − 2E mk 2 1 = − − 4E 2 (12. .25) So far we have shown that the symmetry generators form an algebra. ijk Bk .27) (12. 2 (12. 1. b = 0. a = 0. (12. .31) where we have used (12.30) 2 2 2 1 . . a = 0. Bj ] = i [Ai .7) tells us that (2a + 1) = n. 2 (2a + 1)2 2 2 (12.32) A comparison with (12.28).376 [Bi . By comparing to the rotation algebra introduced in chapter 5. . .23) (12. 1. Furthermore the bound on the orbital angular momentum. .

Following the mass-energy equivalence of special relativity E = mc2 . Originally it was believed that isospin was an exact symmetry of strong interactions and that it was violated by electromagnetic and weak interactions. It is important to place the isospin concept in its proper historical context. which is a particle in box problem in two dimensions with a boundary which can be chosen arbitrarily. which was discovered by observing degeneracies in the particle spectrum. (12. However. If the chosen boundary is ‘irregular’ in a suitably defined sense. after the discovery of the neutron in 1932.2 Isospin and the SU (2) flavor symmetry The concept of isospin goes back to Heisenberg.5: Lorentz Invariant Field Equations 377 sensitivity to initial conditions.36) into each other in abstract isospin space. For a more detailed discussion of quantum chaos in billiard systems we refer the reader to [7] and the references therein. for example. the classical trajectories will diverge from each other after successive bounces from the boundary. the isospin symmetry is also governed by an SU (2) group ‘rotating’ components in (12. 12. analogous to the 1 spin-up and spin-down states of a spin. (12.2 system p= 1 0 . it is natural to represent them in terms of a two component vector. For example. This was motivated by the observation that their masses are approximately equal: mp = 938. in the sense that nearby trajectories in the phase space diverge from each other over time.36) In analogy to the concept of spin regarding the rotations in 3-space as discussed in chapter 5.57M eV /c2 . mp = 939. another manifestation of chaos is the lack of independent operators commuting with the Hamiltonian. that the proton and the neutron behave identically under the so-called strong interactions and that their difference is solely in their charge content. This enables us to utilize what we already know about the SU (2) symmetry group from the study of angular momentum. In the next section we will proceed with the discussion of a symmetry.11.28M eV /c2 . namely. 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. 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. n= 0 1 . suggested that the proton and the neutron can be regarded as two states of a single particle. This (approximate) degeneracy led into the idea of the existence of an (approximate) symmetry obeyed by the underlying nuclear interactions.35) this mass equivalence can be viewed as an energy degeneracy of the underlying interactions. (Strong interactions bind the atomic nucleus together. who. (Weak interactions are responsible.) If the proton and the neutron are to be viewed as two linearly independent states of the same particle. A typical example of this so called quantum chaos is the quantum billiard problem. for the . This is known as level repulsion.

0M eV /c2 ). π − = |1. 2 (12. They can be constructed with the same rules that have been used for angular momentum addition ¯ u in chapter 6. it can be seen as part of a larger (and more approximate) symmetry which is of great utility to classify observed particle families.− 2 2 1 2 2 2 1 1 1 1 . . Shortly we 1 will see how to describe both the pion and nucleon states as composites of more fundamental I = 2 states. (12. The mass difference between the neutron and the proton could then be attributed to the charge content of the latter. d= 1 1 . π 0 = |1. 0 π − = |1. . 1 π 0 = |1. the three pions in (12. I3 = − 2 2 . including the spin and color quantum numbers of their constituent quarks.− 2 2 . They u form an isotriplet: π + = |1.36) as a multiplet with I = 2 1 1 p = I = . For example.6M eV /c2 .378 Spinor Formulation beta decay). 2. (12.43) Similarly.− 2 2 . For a precise description of the two nucleons as composite states.38) which have all nearly identical masses. n = I = . the proton is the lighter of the two. 0 . albeit it is a good approximate one. If the mass difference (or the energy difference) were to be to be purely electrostatic in nature. the proton and the neutron can be written as totally symmetric uud and udd states. mπ0 = 135. we refer the reader to [1].− . as we shall see below. sec. Further than that. (mπ± = 139. the proton had to be heavier. I. . However.41) (12. If it were otherwise the proton would be unstable by decaying into the neutron. . I3 . 1 . the fundamental representation of the isospin symmetry corresponds to the doublet that contains the so-called up and down quarks u= 1 1 . In the framework of the quark of model. −1 = = = 1 1 1 1 .39) All other isomultiplets.− . −1 . I3 = 2 2 1 1 .40) + 2 1 1 . 2 2 1 2 2 2 (12. and its third component.11. including the proton and the neutron.38) are ud. 2 2 1 2 2 2 1 1 1 1 1 √ . We can describe isospin multiplets the same way we have described the angular momentum and spin multiplets. (12. Denoting the total isospin. Isospin symmetry is not an exact symmetry of strong interactions.37) As an example of a multiplet with I = 1 we have the three pions or π-mesons π + = |1. respectively. Therefore it remains a useful concept. spelling disaster for the stability of matter.42) (12. 2 2 1 1 1 . as good quantum numbers. are made up of these two quarks. 1 we can re-write (12. 2 2 . u¯ and d¯ states.

2 = |n > |n > . Following (12. a hydrogen isotope. such as proton and the neutron. this will be mathematically identical to adding two spins. Now let us consider another example of combining the isospins of two particles. 0 ) we should have seen nn and pp bound states of comparable energy in nature (because of isospin symmetry).48) (12. but at the time the empirical concepts like isospin and strangeness quantum numbers were in use. 1 = √ (|p > |n > +|n > |p >). 1 |1. to be −1 for the strange quark. Therefore it has to have isospin. the up and down quarks are not the only quark ‘species’ . 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). strongly interacting particles) which is about a GeV /c2 . and quarks have baryon number 1 . The up and down quarks have strangeness zero. 0 = √ (|p > |n > −|n > |p >) 2 and that of an isotriplet |1. In the late 1940’s and early 1950’s. as some mesons discovered later are heavier than some baryons and so on. we shall setup some terminology: baryons are qqq states.46) (12. the pions being examples thereof.44) 2 where B is the baryon number and S is the strangeness. Before proceeding further. mesons have intermediate mass ranges. The names.or flavors as they are commonly called. a number strange particles have been found which presumably contained a third quark species: the strange quark. (12. 0 ).43). −1 = |p > |p >. 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.47) (12. As it later turned out. consists of a proton and a neutron. I3 = 0.45) Is the deuteron an isosinglet state or an isotriplet? If it were an isotriplet (|1. but such states do not exist. The value of the strangeness quantum number is taken.11. baryons generally are heavy. 0 |1. by accidental convention. this remains only an inaccurate naming convention today. mesons have baryon number 0. Therefore the deuteron has to be an isosinglet state (|0. . It shall be noted here that the quark model was not invented until 1960’s. All other composite states have their strangeness given by the sum of the strangeness content of their constituents. If taken literally.37) and in analogy to (12. whereas mesons are q q states. the neutrinos etc. The possibilities are that of an isosinglet 1 |0. We will now try to describe its wave function in terms of its constituent nucleons. muon. By convention ¯ baryons have baryon number 1. This is why isospin is such a good symmetry and why isomultiplets have nearly identical masses. (12.5: 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. baryon and meson. The reader may know that the deuteron. originally refer to the relative weight of particles. where leptons (electron.e. All antiparticles have their 3 quantum numbers reversed. Naturally.) are light.

53) 0 √ 1 2I (i) +1 I (f ) . (12. as discussed in the exercise above. Refer to exercise 6. 0 + |0. γ (i) (i) (12.49) The only assumption that we put in will be that the interaction is of the form V = α I(1) · I(2) . γ (i) I (f ) .8 to compute the ratio of the scattering amplitudes MI and MII . σ. (12. Exercise. We will now employ the (isospin analogue) of the Wigner-Eckart theorem (6. 1 and (1/ 2)(|1. 0 ).54) Here T00 ≡ V = α I(1) · I(2) . √ According to (12. γ (f ) (f ) (i) (12. Exercise. For completeness let us start by restating the Wigner-Eckart theorem (6. γ (i) is a reduced matrix element defined in the same sense as in section 6.48) the initial states in (I) and (II) have isospin values |1. respectively. (f ) (i) (I (f ) I3 |00I (i) I3 ) is a Clebsch-Gordon coefficient and I (f ) . Consider the generalization of tensor operators discussed in section 6. where M is the scattering amplitude given by M = final | α I(1) · I(2) | initial . Show that I(1) ·I(2) is an ‘isotensor’ of rank zero. The dot product here refers to the abstract isospin space. 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 ) = 1.7.50) .38) and the fact that the deuteron is an isosinglet we know that the isospins of the final states in (I) and (II) are |1.259) discussed in detail in sections 6.45) and (12. I3 = 1.55) .52) where γ (i) and γ (f ) denote degrees of freedom other than isospin.8.51) (12.5 for the spin-analogue of the same problem. such as the spatial dependence of the wave function and spin. The cross-section. 0 . For example. The initial and final states can be denoted in more detail as | initial | final = = I (i) .380 Spinor Formulation As an exercise on the implications of isospin symmetry we will consider nucleon-nucleon scattering.259) in the present context: I (f ) I3 . . γ (f ) ||T ||I (i) .7 and 6. γ (f ) ||T0 ||I (i) . γ (f ) |T00 |I (i) I3 . I3 . I3 . is proportional to |M|2 . γ (f ) T00 I (i) = 1. 1 and |1. which is an isoscalar. We will eventually be able to compute ratios of scattering cross-sections between different processes. γ (i) . consider (I) p + p → d + π + (II) p + n → d + π 0 (12. 1 4 in an isotriplet state and − 3 in an isosinglet 4 Using (12.7 to the case of isospin. I3 (f ) = 1. Show that the expectation of I(1) · I(2) is state. γ (i) = (f ) (i) (I (f ) I3 |00I (i) I3 ) (f ) (i) (−1)I −I (f ) (i) (12.

which is also manifested by a vanishing Clebsch-Gordon prefactor. for example. −2 1 1 √ . γ (f ) ||T00 || I (i) = 1.11.61) MII Note that the second term in MII vanishes due to the isospin conservation. I3 = 0.5: Lorentz Invariant Field Equations 1 (11|0011) √ I (f ) = 1.56) (12. γ (i) 3 +0. (b) π − + p → π − + p . We want to compute the ratio of total cross-sections assuming a similar interaction as in the previous example σ( π + + p → anything ) . 0 1 1 3 3 2. we will consider pion-nucleon scattering. γ (f ) ||T00 || I (i) = 1. I3 = 0.65) (12. (12. involving.60) (12. σ( π − + p → anything ) The possibilities are (a) π + + p → π + + p . (12. −1 2 3 2 − + .59) (12. γ (f ) T00 I (i) = 0. The relevant Clebsch-Gordon coefficients are easily evaluated: (11|0011) = (10|0010) = 1. I3 = 0. 1 π − + p : |1.66) = 2 3 1 3 2. −1 π0 + n : |1. 1 1 1 1 3 √ 2. (c) π − + p → π 0 + n . −2 .63) where common dynamical factors (which would not be as easy to compute) have dropped out thanks to the Wigner-Eckart theorem. It follows σI = 2.62) We can now write the ratio of the scattering amplitudes: MI 1 √ .57) (12. σII which is in approximate agreement with the observed ratio. γ (i) 3 1 (f ) (i) I (f ) = 1. (12. which are obtained by a standard Clebsch-Gordan expansion π + + p : |1. γ (f ) T00 I (i) = 1. Once again we need the isospins for the initial and final states.64) There are more exotic possibilities. I3 = 0. γ (i) 2 1 = (10|0010) √ I (f ) = 1. = 381 (12. 2 . but these are not dominant at relatively low energies. [2] As a further example. . γ (i) = √ 2 1 (f ) (i) +√ I (f ) = 1. 2 = 3 2. = MII (1/ 2) (12. 2 = 2. −2 1 1 2.58) (12. particles with strangeness. −2 2 1 1 3 2.

however.70) (12. Hence Gell-Mann and independently Nishijima postulated the existence of a separate quantum number.68) − 2 3 M1 2 Guided by empirical data we will further assume that M 3 >> M 1 . while weak interactions did not.65) Ma = M 3 2 Mb = 1 M 3 + 2 M 1 3 3 √ 2 Mc = 2 3 M3 2 √ 2 (12. such that S(K + ) = 1. 0 0 1 . called strangeness.d =  1  . S(Σ− ) = −1 and S(π) = S(N ) = 0. [2] (12.69) 12. m . A computation similar to the previous example of nucleon-nucleon scattering yields (apart from common prefactors) the following amplitudes for the reactions in (12. (12. Strange partners of the familiar nucleons. Hence the strangeness changing strong decays of K + (or Σ− ) were forbidden. K + . m 1 2.71) u =  0  . mentioned in the previous section.67) which are independent of m. The classification of the newly found particles as members of some higher multiplet structure was less obvious then the case of isospin. which leads to the following 2 2 ratios for the cross-sections σa : σb : σc = 9 : 1 : 2. In the light of the quark model. was motivated by the existence of particles that are produced strongly but decay only weakly. which can be produced by π − + p → K + + Σ− . it appears an obvious generalization to add another component for an extra quark to the isospin vector space       1 0 0 (12. has a lifetime which is comparable to that of π + albeit being more than three times heavier.382 Spinor Formulation As in the example of nucleon-meson scattering we define the relevant matrix elements M3 = 2 M1 = 2 3 2. S. m V V 3 2. For instance. . giving it a higher than usual lifetime.s =  0  . are up to 40% heavier. m 1 2. etc. for example. It was assumed that strong interactions conserved S (at least approximately). 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. making an identification of the underlying symmetry and the multiplet structure less straightforward.3 The Eightfold Way and the flavor SU (3) symmetry The discovery of the concept of strangeness.

Multiplying U by a phase. It is seen that such a matrix U has to satisfy the following k U † = U −1 . the same way nucleons and pions were identified as representations of the isospin SU (2) symmetry.5: Lorentz Invariant Field Equations 383 In this case the transformations that ‘rotate’ the components of (12. bottom (beauty) and top (truth) are significantly heavier than the hadronic energy scale.1 whenever necessary. where Nf is the number quark flavors. ak .71). called quantum chromodynamics. in fact modern theory of strong interactions is a ‘gauge theory’ of this color group. In many respects it will resemble the more familiar group SU (2) discussed in some detail in chapter 5. but there are a number of subtle differences. in reference to Lie groups. as in (12. Because of this additional constraint SU (3) has 8 dimensions. we will establish some mathematical preliminaries about the group SU (3). (The ‘bare’ mass of the charm quark is already heavier than the two nucleons. it is the group . SU (N ). Then came the question of what the fundamental representation. which set the hadronic energy scale. unitary relation k (12. should correspond to.72) a∗ ak . namely charm. However. but we will stick to N = 3 in the following.73) To verify that all such matrices form a group. have to be elements of the group SU (3) (which we will investigate closely very soon). (12.74) for any two unitary matrices U and V . becomes increasingly inaccurate for Nf > 3. we observe that (U V )† = V † U † = V −1 U −1 = (U V )−1 . The bottom and top are even heavier. (12. The reason is that the other known quarks. we want to find those transformations ak → Ukl al that preserve the norm. Hence the group U (3) has 9 dimensions. This is believed to be an exact symmetry of strong interactions. However SU (3) symmetry appears in another and much more fundamental context in strong interaction physics. 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. which again form representations of an SU (3) group. Given a complex vector. history followed the reverse of this path.11. First particle multiplets were identified as representations of the SU (3) group. The reader is also invited to revisit section 5. Since arbitrary phase factors are of no physical interest. giving rise to the quark model. for a period they were considered as useful bookkeeping devices without physical content. The unitarity relation imposes 9 constraints on the total of 18 real degrees of freedom of a 3 × 3 complex matrix. called color.71) into each other. of a. The quarks posses another quantum number. while preserving the norm. As quarks were never directly observed. still leaves the norm invariant. eiφ . (The reader is referred to section 8. The reader shall note that most of what is being said trivially generalizes to other unitary groups. of three dimensions.3 for a brief discussion of gauge transformations). This group of 3 × 3 unitary matrices is denoted by U (3). [2] ) Before discussing the significance and the physical implications of the quark model. Flavor SU (Nf ) symmetry on the other hand. Lie algebras and related concepts. In this perspective the flavor SU (3) symmetry may appear to be mainly of historical interest.

79) (12. via the anticommutator relation 4 [λj .32).77)  λ1  . As in the case of SU (2) higher dimensional representations obeying the same structure can be found. 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  . λ7 =   0 0  . λk ]+ = δjk + 2δjkl λl . As discussed in section 5.7. λ3 =  0 −1 0  . obey the following relation tr(λj λk ) = 2δjk . −2 λ4 λ6 λ8 The generators. λk ] = 2ifjkl λl . λ2 =     .76) where λk are 8 linearly independent matrices forming the basis of the Lie algebra of SU (3).80) This is the fundamental or defining representation of SU (3). the meaning being apparent from the context. δjkl . any unitary matrix. U . which can be verified explicitly as matrix identities.1. 3 (12.78) where fjkl are the antisymmetric structure constants similar to the jkl of SU (2) given in (5. We can also introduce the constants.   . regardless of the dimension of the representation. Therefore we will express elements of SU (3) as U = ei k (12.79). (12. λk . λ5 =   .75) αk λk (12. since det(eA ) = etrA . (12. The reader is invited to compare the structure of SU (3) to that of SU (2) discussed in section 5.384 Spinor Formulation SU (3) and not U (3) that is of main interest. can be written in the form U = eiH where H is a hermitian matrix. The fundamental relation to be preserved is (12. The Lie algebra structure is given by the commutators of λk [λj .) The unit determinant condition requires that all λk are traceless. . (We shall at times refer to the Lie algebra with the name of the group. 0 0 0   .

U± = F6 ± iF7 . [Y.44). T± ] = 0.93) (12.81) In chapter 5. In the basis (12. 3 (12. U± ] = ±U± .90) (12.77) is not the most useful basis in the study of SU (3).84) (12.5: Lorentz Invariant Field Equations 385 As it turns out the set of generators in (12. The generators of SU (3) can be arranged into a very similar form to that of SU (2). 2 (12.88) which defines (not uniquely) a maximal set of mutually commuting operators {Y.92) (12. we have [Y. namely spherical harmonics. Exercise. J− } proved useful. (12. V± = F4 ± iF5 .85) (12. V± ] = ±V± .89) (12. I3 .86) (12. 2 Y = √ F8 .87) the commutation relations between the generators can be expressed in a succinct manner. [Y. First. (Recall that the strange quark has S = −1 by convention). starting with the observation that Y = B + S. (12. 1 [T0 .94) .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 . T 0 = F3 . T± ] = ±T± .83) (12.71) show that T0 is the isospin operator. these relations have been used to construct the angular momentum spectrum as well as the function space representation of the rotation algebra. V± ] = ± V± . 2 1 [T0 . (12. T0 ] = 0. Using the convention in (12. In the case of SU (2) the identification of ‘ladder’ operators {J+ . which satisfied an ‘eigenvalue equation’ [J0 . T0 } and [Y. 2 (12.11. J± ] = ± J± . U± ] = U± . Derive the Gell-Mann–Nishijima relation (12. We first introduce the F-spin operators 1 Fi = λ i . Y is called the hypercharge.87) Exercise.91) [T0 .

386 Spinor Formulation which relate the remaining generators {T± . [V+ . 2 3 Y + T0 = 2V0 . This form is essential for easy labeling of the representations of the group.98) (12. since there are two mutually commuting generators in SU (3) as given in (12. V− ] = 2 (12. The maximum number of mutually commuting generators of a Lie algebra is called its rank. SU (2) has rank 1. V+ ] = 0. Note that. [T+ .100) (12. Finally.99) (12.97) which relate commutators of generators with opposite eigenvalues to the maximal set {Y. jkl (12. [U+ .101) (12.102) (12. The interested reader is instead referred to a very readable account given in chapter 12 of [4]. Thus. U− ] = 0. V± . The same way the angular momentum ladder operators have been used to construct the representations of SU (2). it is said to be in Cartan-Weyl form. Another important property of SU (2) is the existence of an operator (namely the total angular momentum. the representations will now lie in a T0 − Y -plane. When the basis of a Lie algebra is expressed in such a way to satisfy the form of the eigenvalue relations as given above. Casimir operators can be used to label irreducible representations of the Lie algebra. However. U− ] = Y − T0 = 2U0 . [T+ . U0 and V0 are linear combinations of T0 and Y . similar to the way it was done in section 5.104) (12. In the case of SU (2) the representations lay on a line on the J0 axis.5. we have [T+ . [U+ . T− ] = 2T0 .103) Any remaining commutators follow from hermiticity.95) (12. V− ] = −U− . As in the case of J 2 and SU (2). An operator which commutes with all generators of a Lie group is called a Casimir operator. [T+ . V− ] = T− .88). A formal definition and a detailed discussion of the Cartan-Weyl form is beyond the scope of this chapter. J 2 ) which commutes with all of the generators. U± } to this maximal set by ‘eigenvalue equations’ and [T+ . C1 = k λ2 = − k 2i 3 fjkl λj λk λl . (12. as the relation between the states in a given representation can be conveniently expressed in terms of ladder operators. V+ ] = 0. U+ ] = V+ . We can construct two such independent Casimir operators for the group SU (3). . we will use these commutation relations to construct representations of SU (3).105) C2 = jkl djkl λj λk λl In general the number of independent Casimir operators of a Lie group is equal to its rank. 3 [U+ .96) (12. while SU (3) has rank 2. T0 }.

y . All other states of the representation are then constructed by successive application of ladder operators T± . (12. 0) and D(0.11. y . For example. For example. we have U0 |t3 . especially chapters 7 and 8.antiquark pair. This property can be used to label representations in terms of the values of the Casimir operators. y Y |t3 . y .109) The same way that J± |m is proportional to |m ± 1 in the case of the angular momentum algebra. U± . The representations D(1. it was shown in section 5. 2 (12. 1) correspond to the triplets of quarks and antiquarks. y 1 = α t3 ± . y 3 = ( y− 4 3 = ( y+ 4 1 t3 ) |t3 . The details of this procedure is beyond the scope of this chapter.88) we can label states by the eigenvalues of T0 and Y operators. The representations for hexagons with sides of length p and q in the T0 − Y -plane. 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.112) The effect of these operators to the states in the y − t3 plane have been outlined in Fig. (See Fig. q) and it has a dimensionality of 2 (p + 1)(q + 1)(p + q + 2). y . y V± |t3 .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 .) All other representations can be constructed by combining these two conjugate representations.3). y V0 |t3 . V± . y U± |t3 . y .y ± 1 . Now we will construct explicit representations of SU (3). the pion family forms part of an octet corresponding to the D(1. y . y ± 1 . As another example for the representations of SU (3). |t3 . 2 1 = β t3 ± .110) (12. y = t3 |t3 . 1) as an example. The representations of SU (3) are constructed analogous to those of SU (2) by identifying the ‘boundary’ states annihilated by raising (or lowering) operators. Figure (12. Such a representation is labeled as 1 D(p.5 how to label the irreducible representations of the angular momentum algebra SU (2) in terms of the value of the total angular momentum.2) shows the representation D(2. A notation suggestive of the dimensionality of the representation can be used to identify representations. Because of (12. 1) representation. 2 1 = γ t3 .111) (12. 2 (12. For example the pion octet (or any other meson octet) is therefore realized as states of a quark .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. (12.108) (12. y : T0 |t3 . The interested reader is referred to [4]. we have T± |t3 . = y |t3 .106) (12.1). (12. . respectively. For example.

1) representation of SU (3).388 £ Spinor Formulation Figure 12.2: D(2.1: The effect of SU (3) ladder operators on the y − t3 plane. £ Figure 12. The states in the inner triangle are doubly degenerate. ¡ ¢  ¡ ¢  8 6 2 '745301) ¥¦¤   ¤   #"  ! (& % '$   © ¨     ¦ §¥ ©§ ¨ .

3 (12.5: Lorentz Invariant Field Equations 389 D(1. £ (12.117) The Gell-Mann–Nishijima relation can then be succinctly expressed as 1 Q = y + t3 . in the case of SU (2).116) (12.113) The additional singlet state corresponds to the η meson. The octet D(1.118) ¡ ¢    ¨ ©§ ¨¥¦¤ "   ! . 0) and D(0. This expansion can be compared. − . 3]. 1) are denoted by [3] and [¯ respectively. 1) is written as [8] etc. to the case of adding two spin triplet states. The three quarks in the fundamental representation can now be written as u = d = s = 1 1 . 0) or the fundamental representation of flavor SU (3) symmetry. 2 3 1 1 − .11. where we would write [3] ⊗ [3] = [1] ⊕ [3] ⊕ [5]. for example.3: D(1. This way the quark-antiquark states can be represented as follows [3] ⊗ [¯ = [8] ⊕ [1] 3] (12. .114) Figure 12. 2 (12. . 2 3 2 0.115) (12.

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

Master your semester with Scribd & The New York Times

Special offer for students: Only $4.99/month.

Master your semester with Scribd & The New York Times

Cancel anytime.