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

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

6) is expressed in terms of ‘normal’ functions. t) = L(q(t).. equating these terms with δS[q(t). ˙ In the following we will often use the notation for velocities and other time derivatives d q(t) = q(t) and dtj = xj . · . 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.e. but rather to a differential of the function which is simply the df differentiated function multiplied by the differential increment of the variable. · ) is a function differentiable in its three arguments. t1 ] where the integrand is a function L(q(t). at the ends of the time interval of the trajectories the variations vanish.10) From this follows (1. It holds   M  M ∂L  t1 d ∂L ∂L δS[q(t). t) of the configuration d vector q(t). q(t) + δ q(t). q(t) + δq(t) is a ‘slightly’ different trajectory than q(t). ∂f in case of a function of M variables.6) immediately.5) where L( · . We attempt to evaluate t1 S[q(t) + δq(t)] = t0 ˙ ˙ dt L(q(t) + δq(t). t) (1.2 Lagrangian Mechanics Note: δq(t) describes small variations around the trajectory q(t). δq(t)] = dt − δqj (t) + δqj (t)   ∂qj dt ∂ qj ˙ ∂ qj ˙ t0 j=1 j=1 t1 .e.8) We note then using d dt f (t)g(t) ˙ = f˙(t)g(t) + f (t)g(t) ∂L δqj ∂ qj ˙ − d ∂L dt ∂ qj ˙ δqj . t) (1. . the velocity vector dt q(t) and time t.7) through Taylor expansion and identification of terms linear in δqj (t). q(t). It is also important to appreciate that δS[ · . We focus on such functionals because they play a central role in the so-called action integrals of Classical Mechanics.g.6) For a proof we can use conventional differential calculus since the functional (1. df = dx dx or.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. δq(t)]. i. i. t0 (1. df = M ∂xj dxj . ˙ Theorem: Let S[q(t)] = t0 dx dt t1 ˙ dt L(q(t). (1. j=1 We will now consider a particular class of functionals S[ ] which are expressed through an integral d over the the interval [t0 . · ] in conventional differential calculus does not correspond to a differentiated function. For this purpose we consider M ˙ ˙ ˙ L(q(t) + δq(t).. e. q(t) + δ q(t). t) + j=1 ∂L ∂L δqj + δ qj ˙ ∂qj ∂ qj ˙ + O( 2 ) (1. q(t). dt q(t).

According to the Lemma above follows then (1. t1 ] ⊂ R → R holds t1 (1. . y(x1 ) = y1 .6) which reads in the present case     M ∂L t1 d ∂L δS[q(t). it holds in case δq(t0 ) = δq(t1 ) = 0 that qe (t) is an extremal. M . · ] ≡ 0 and. For this purpose we define Definition: An extremal of a differentiable functional S[ ] is a function qe (t) with the property δS[qe (t). .12) for j = 1.14) for any continuous function h(t) ∈ F with h(t0 ) = h(t1 ) = 0. . 2. An Example As an application of the above rules of the variational calculus we like to prove the well-known result that a straight line in R is the shortest connection (geodesics) between two points (x1 .1.12) for j = 1. This condition is stated in the following theorem. 2. y(x)) to (x + dx.5) assume extreme values. . . The length of such path can be determined starting from the fact that the incremental length ds in going from point (x. if and only if it satisfies the conditions (j = 1. (1. y2 ). t1 ]. M and δqj (t0 ) = δqj (t1 ) = 0 follows according to (1.16) This property holds for any δqj with δq(t) ∈ F . then f (t) ≡ 0 on [t0 . The proof of the above theorem starts from (1.17) . . . . dx (1. δq(t)] = 0 for all δq(t) ∈ F . 2. .11) The extremals qe (t) can be identified through a condition which provides a suitable differential equation for this purpose. δq(t)] = dt − δqj (t) . from (1. Theorem: Euler–Lagrange Condition For the functional defined through (1.   ∂qj dt ∂ qj ˙ t0 j=1 (1. On the other side.16) the property δS[qe (t).5). the above theorem. y1 ) and (x2 .15) We will not provide a proof for this Lemma. y(x + dx)) is ds = (dx)2 + ( dy dx)2 = dx dx 1+( dy 2 ) .12) The proof of this theorem is based on the property Lemma: If for a continuous function f(t) f : [t0 . . Let us assume that the two points are connected by the path y(x).1: Variational Calculus 3 We now consider the question for which functions the functionals of the type (1. M ) d dt ∂L ∂ qj ˙ − ∂L = 0 ∂qj (1. hence. (1. y(x2 ) = y2 .13) dt f (t)h(t) = 0 t0 (1.

. Threorem: Hamiltonian Principle of Least Action The trajectories q(t) of systems of particles described through the Newtonian equations of motion d ∂U (mj qj ) + ˙ = 0 dt ∂qj .1.. The constants a and b are readily identified through the conditons y(x1 ) = y1 and y(x2 ) = y2 .20) x1 − x2 Exercise 1. qM ) . the so-called action integral. dx (1. Using y = dx the condition reads d dx ∂L ∂y = d dx y 1 + (y )2 = 0. according to the theorems above. i. 1.21) are extremals of the functional. . t1 S[q(t)] = t0 ˙ dt L(q(t). 2.19) From this follows y / 1 + (y )2 = const and.e. Velocity–dependent potentials which describe particles moving in electromagnetic fields will be considered below. y = const. ˙2 2 (1.1: Show that the shortest path between two points on a sphere are great circles.22) ˙ where L(q(t). q2 .4 The total path length is then given by the integral x1 Lagrangian Mechanics s = x0 dx 1+( dy 2 ) . . . circles whose centers lie at the center of the sphere.5) with L(y(x). hence. (1. . obey the Euler–Lagrange dy condition. j = 1. . dx ) = 1 + (dy/dx)2 . q(t).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. t) (1. q(t). t) = j=1 1 mj qj − U (q1 . M (1. This in turn yields y(x) = ax + b. . (1.23) Presently we consider only velocity–independent potentials. The shortest path is an extremal of s[y(x)] which must. q(t). One obtains y 1 − y2 y(x) = (x − x2 ) + y2 . . t) is the so-called Lagrangian M ˙ L(q(t).18) dy s is a functional of y(x) of the type (1.

1.2: Lagrangian 5 For a proof of the Hamiltonian Principle of Least Action we inspect the Euler–Lagrange conditions associated with the action integral defined through (1. t) is the Lorentz force.31) = − V − Gauge Symmetry of the Electromagnetic Field It is well known that the relationship between fields and potentials (1. t) obey the Maxwell equations ×E + ·B ×B − ·E 1 ∂ c ∂t E 1 ∂ c ∂t B (1.31) allows one to transform the potentials without affecting the fields and without affecting the equations of motion (1.33) . t) and B(r. t) describes the charge current density.30. t) and B(r. Particle Moving in an Electromagnetic Field We will now consider the Newtonian equations of motion for a single particle of charge q with a trajectory r(t) = (x1 (t). respectively. t) describes the charge density present in the field and J(r. t) = q E(r. t) c ∂t (1.26) (1. x3 (t)) moving in an electromagnetic field described through the electrical and magnetic field components E(r. t) = A(r.24) ∂qj dt ∂ qj ˙ ∂qj dt which are obviously equivalent to the Newtonian equations of motion. t) and a vector potential A(r. t) 1∂ K(r.27) and (1.23).22. c ∂t (1. The transformation which leaves the fields invariant is A (r. t) as follows B E = ×A 1 ∂A . The equations of motion for such a particle are d q ˙ (mr) = F (r. t) + v × B(r. t) dt c where dr = v and where F (r. t) − K(r.25) = 0 = 0 4π J c = 4πρ = (1.28) (1. Equations (1. F (r.28) can be satisfied implicitly if one represents the fields through a scalar potential V (r. t) .32) (1.30) (1.1. These conditions read in the present case ∂L d ∂L ∂U d − = 0 → − − (mj qj ) = 0 ˙ (1. dt The fields E(r.29) where ρ(r. t) = V (r. x2 (t).25) of a particle moving in the field. t). t) + V (r. 1.27) (1.

30). (1.41) which is identical to the term (1. dt ∂x1 c We notice using (1.38. in fact. 1. d ∂V q (mv1 ) = F1 = −q + [v × B]1 .25) follows from the Hamiltonian Principle of Least Action.34) For this purpose we consider only one component of the equation of motion (1.25). t) + A(r.35) is equivalent to the Euler–Lagrange condition d dt The second term in (1. if one assumes for a particle the Lagrangian ˙ L(r. 2 c (1.40) (1. t) · v .36) This expression allows us to show that (1.40. .. e.36) in the Newtonian equation of motion.37) is ∂L ∂V q = −q + ∂x1 ∂x1 c The first term in (1.38) ∂L ∂ x1 ˙ − ∂L = 0.6 Lagrangian of Particle Moving in Electromagnetic Field Lagrangian Mechanics We want to show now that the equation of motion (1. (1. r.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 (1.35) − ∂A1 ∂x2 [v × B]1 = x2 B3 − x3 B2 = x2 ˙ ˙ ˙ ∂A2 ∂A1 − ∂x1 ∂x2 − x3 ˙ ∂A1 ∂A3 − ∂x3 ∂x1 .35) shows that the Newtonian equations of motion and the Euler–Lagrange conditions are.g. B3 = ∂A2 ∂x1 (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 . ∂x1 (1.37) The results (1. (1. equivalent. 1. namely.41) with (1. t) = 1 q mv 2 − q V (r. Comparing then (1.39) ∂A1 ∂A2 ∂A3 x1 + ˙ x2 + ˙ x3 ˙ ∂x1 ∂x1 ∂x1 .

The factor q has been introduced to make this c transformation equivalent to the gauge transformation (1.33) of electromagnetic potentials. surprisingly simple fashion in Lagrangian Mechanics. 2 c = A(r. 1. a term not affected by the variations allowed. The gauge symmetry. dt ∂x1 ∂x2 ∂x3 ∂t ∂t (1. For this purpose we consider the transformation of the single particle Lagrangian (1.43) To prove this theorem we determine the action integral corresponding to the transformed Lagrangian t1 S [q(t)] = t0 ˙ dtL (q.42) This transformation is termed gauge transformation.46) (1.e. Note that one adds the total time derivative of a function K(r. t) · v . i. t) + A(r. t) = x1 + ˙ x2 + ˙ x3 + ˙ = ( K) · v + .32. t) . t) + q K(q.1. r. One can conclude then immediately that any extremal of S [q(t)] is also an extremal of S[q(t)]. equivalent to the gauge transformation (1.34) q 1 q d ˙ L (r. t) = L(q.3: Symmetry Properties 7 1. t) c t1 t0 t1 t0 ˙ dtL(q.32. Eq. 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.45) . t) L (q.44) implies that the gauge transformation amounts to adding a constant term to the action integral. t) − + 2 c ∂t c 1 q = mv 2 − q V (r. 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. 1. We want to demonstrate now that the transformation (1. 1. t) + dt c (1. (1.42) is.43) into (1. t) the Lagrangian.32.33) of electyromagnetic potentials. t) + A (r. gauge symmetry and symmetries with respect to spatial transformations. We will consider in the following two symmetries. q.33) ˙ L (r. t) = q K(q. and since these properties allow one often to simplify mathematical descriptions. They are the subject of the following theorem. in fact. This term is d ∂K ∂K ∂K ∂K ∂K K(r. r. t) c t1 t0 = S[q(t)] + (1. t) + K ·v (1. 2 c c dt Inserting (1.44) Since the condition δq(t0 ) = δq(t1 ) = 0 holds for the variational functions of Lagrangian Mechanics. encountered above in connection with the transformations (1. appear in a different. q.32. t) 1 1 ∂K q mv 2 − q V (r.33) of electromagnetic potentials. q. t) = mv 2 − q V (r..45) and reordering terms yields using (1. t) · v + K(r. q.

t).e.32. where the origin of is chosen such that r (r.33).47) (1.42) corresponds to replacing in the Lagrangian potentials V (r.50) A non-trivial example is furnished by the infinitesimal rotation around axis e ˆ r = r + e×r . 0) = r .. ˆ (1. the reason being that our interest lies in achieving familiarity with the principles (just mentioned above ) of symmetry properties rather than in providing a general tool in the context of Classical Mechanics. and for the introduction of infinitesimal transformations which will provide a crucial method for the study of symmetry in Quantum Mechanics. therefore. r ∈ R . for such vectors holds a · a = 1. 1. We have proven. t). the transformation (1. for the connection between invariance properties and constants of motion. ) = r + R(r) + O( 2 ) . i. but nevertheless instructive way considering rotations around the x3 –axis. A(r. for example. However. r. the equivalence of (1.49) =0 ∈ R (1. The transformations considered are specified in the following definition. we like to ˆ provide the transformation here anyway for later reference r = r + e. ) .48) In the following we will denote unit vectors as a. in the case of central force fields which are invariant with respect to rotations of coordinates around the center. R(r) = through ∂r ∂ (1. t).52) x3 0 0 1 x3 . We consider now invariance properties connected with coordinate transformations. ˆ (1. ˆ ˆ ˆ Examples of Infinitesimal Transformations The beauty of infinitesimal transformations is that they can be stated in a very simple manner. The transformations we consider are the most simple kind.51) We would like to derive this transformation in a somewhat complicated. t) by gauge transformed potentials V (r. Definition: Infinitesimal One-Parameter Coordinate Transformations A one-parameter coordinate transformation is decribed through r = r (r. The following description of spatial symmetry is important in two respects.42) and (1. A (r. In this case the transformation can be written in matrix form      cos −sin 0 x1 x1  x2  =  sin cos 0   x2  (1.8 Lagrangian Mechanics Obviously. which has an important analogy in Quantum Mechanics. Such invariance properties are very familiar. In case of a translation transformation in the direction e nothing new is gained. The corresponding infinitesimal transformation is defined for small r (r.

. r) which is constant along the classical path of the system. a quantity C(r. a classical mechanical system is invariant with respect to a coordinate transformation ˙ a constant of motion exists. t) = L(q.e.57) must cancel each other or both vanish.57) ∂ d where we used dt Qj = M ( ∂qk Qj )qk . We consider here only the ‘particle version’ of the theorem. neglecting terms of order O( M ˙ ˙ L (q. that along the classical path holds d ∂L ∂L the Euler-Lagrange condition ∂qj = dt ( ∂ qj ) one can rewrite the sum of the second and third term ˙ in (1. We have used here the notation corresponding to single particle motion. t) is invariant with respect to an infinitesimal transformation q = q + Q(q). Using the fact. i. namely for fields rather than only for particle motion. expansion.3: Symmetry Properties 9 In case of small this transformation can be written neglecting terms O( 2 ) using cos = 1 + O( 2 ). q. t) + j=1 ∂L Qj + ∂qj M j. q. In order to prove Noether’s theorem we note qj qj ˙ = qj + = qj + ˙ k=1 Qj (q) M (1. the property holds for any system.56) ∂Qj qk . (1.k=1 ∂L ∂Qj qk ˙ ∂ qj ∂qk ˙ (1. Before the embark on this theorem we will comment on what is meant by the statement that a classical mechanical system is invariant under a coordinate transformation. sin = + O( 2 )        0 − 0 x1 x1 x1  x2  =  x2  +  0 0   x2  + O( 2 ) . The property has been shown to hold in a more general context.53). ˙ ∂qk ˙ Inserting these infinitesimal changes of qj and qj into the Lagrangian L(q. q. the second and third term ˙ k=1 in (1.e.57)   M Qj j=1 d dt ∂L ∂ qj ˙ + ∂L d Qj ∂ qj dt ˙ = d  dt M Qj j=1 ∂L  = 0 ∂ qj ˙ (1..51) reads r = r − x2 e1 + x1 e2 ˆ ˆ (1. Anytime.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. In the context of Lagrangian Mechanics this implies that such transformation leaves the Lagrangian of the system unchanged. however.58) From this follows the statement of the theorem.55) (1. Theorem: Noether’s Theorem ˙ If L(q. ˙ We have generalized in this theorem the definition of infinitesimal coordinate transformation to M –dimensional vectors q. Invariance implies L = L. t) yields after Taylor ˙ 2 ). i. q. then M ∂L j=1 Qj ∂ xj is a constant of motion. by Noether..54) which is equivalent to (1.1.

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

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

t.7) t. t ) ψ(r . namely the point on the classical path at time t . t|r .. t1 |r0 . This symbol will be defined further below.e. t ) ψ(r . tN −1 ) φ(rN −1 . t|r0 . . at any intermediate time the particle needs only be known at one space point.11) which denotes an integral over all paths r(t) with end points r(t0 ) = r0 and r(tN ) = rN . i. t). (2. t ) φ(r . tN −1 |rN −2 . This must hold for any ψ(r . t0 ) = Ωd 3r N −1 Ωd 3r N −2 ··· Ωd 3r 1 φ(r. t ∈ [t0 . . t. described by ψ(r. t|r . > t1 > t0 is φ(r. t |r0 . t ) φ(r.8) a knowledge of the state at all space points r ∈ Ω at some intermediate time t . i. t ) ψ(r . t|r . The generalization of the completeness property to N − 1 intermediate points t > tN −1 > tN −2 > . t = )|2 = 1 . t|r . t ) Ωd 3 r |ψ(r. t1 ] yields an expression of the form r(tN )=rN φ(r. The following so-called completeness relationship holds for the propagator (t > t [t0 . requires then according to (2.10) We have introduced here a new symbol. t1 ] (2. the propagator is unitary. The time evolution of ψ(r. The state of a system at time t. t1 ]) d3 r φ(r.4) holds for all times. tN −2 ) · · · φ(r1 . t )φ(r. ψ(r0 . t ∈ (2. . t)|2 = (2.8) 7.9) for N → ∞. 8. t ) = δ(r − r ) Ω (2. t|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) is described by a linear map of the type ψ(r.e. t1 ]) Ωd Ωd 3r Ωd 3r Ωd 3r 3 r |ψ(r. t0 ) = r(t0 )=r0 d[r(t)] Φ[r(t)] . The definition will actually assume an infinite number of intermediate times and express the path integral through integrals of the type (2. hence.9) Employing a continuum of intermediate times t ∈ [t0 . Since (2.5) 6. t ∈ [t0 . t|r0 . t) = Ω d3 r φ(r. This is different from the classical situation where the particle follows a discrete path and. (2. t| rN −1 . (t > t . t ) which requires d3 r φ(r. t0 ) = δ(r − r0 ). t ) t > t .6) φ(r.12 Quantum Mechanical Path Integral 5. t|r0 . t|r . t0 ) = φ(r. t0 ) . the path integral r(tN )=rN d[r(t)] · · · r(t0 )=r0 (2..

0545 · 10−27 erg s . a very large number. 2 2 (2. the exponent is a very large imaginary number. Scl / ≈ 8. 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.. r. Since this number is multiplied by ‘i’. 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.2.13) are built on action integrals for all possible paths. 2. The situation is very different for microscopic particles. 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. One would still expect significant descructive interference between contributions of different paths since the value calculated is comparable to 2π. t) (2.14) has the same dimension as the action integral S[r(t)].5 · 1027 . i. r. expressions (2. interferences should be much less dramatic than in case of the macroscopic particle.12) S[r(t)] = t0 ˙ dt L(r.. The functional Φ[r(t)] in (2. The constant given in (2. such that destructive interference of the contributions of such paths does not occur. i. for the electron many paths contribute and the action integrals for non-classical paths need to be known.2: Axioms 9. contribution to the path integral (2. Only for paths close to the classical path is such interference ruled out.14) ˙ In (2. However.12. However. Its value is extremely small in comparision with typical values for action integrals of macroscopic particles. The value of S[r(t)] is then Scl = 1 1 m v 2 t = erg s . accordingly.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. This implies that small variations of the path near the classical path alter the value of the action integral by very little. (2. not only for the classical path. in complete distinction from Classical Mechanics.12. actually often essentially exclusive. 2. t) is the Lagrangian of the classical particle.13) and = 1.13).e. it is comparable to action integrals as they arise for microscopic particles under typical circumstances.12) is then Scl / ≈ 0.15) The exponent of (2. . namely due to the property of the classical path to be an extremal of the action integral. for microscopic particles like the electron this is by no means the case. In case of a proton with mass m = 1.13) L(r. However. However.11) is Φ[r(t)] = exp where S[r(t)] is the classical action integral tN 13 i S[r(t)] (2.e. Situations which are well described classically will be distinguished through the property that the classical path gives the dominant.

t1 ). . (2. provides an avenue to evaluate path integrals. namely tj+1 − tj = (tN − t0 )/N = N . . (xN −1 . ˙ 2 (2. tN )} .13) and.17) The discretization in time leads to a discretization of the paths x(t) which will be represented through the series of space–time points {(x0 . the xj values are not.20) can be properly taken.19) The main idea is that one can replace the path integral now by a multiple integral over x1 . x. as in (2. ∞[. (xN . x2 . a series of times tN > tN −1 > tN −2 > . .3 How to Evaluate the Path Integral In this section we will provide an explicit algorithm which defines the path integral (2. (x1 . CN is a constant which depends on N (actually also on other constant in the exponent) which needs to be chosen to ascertain that the limit in (2. (2. . Its value is CN = m 2πi N 2 (2. t) = ˙ 1 mx2 − U (x) . t0 ) = limN →∞ CN exp i N N −1 j=0 +∞ +∞ +∞ −∞ dx1 −∞ dx2 . 2.16) In order to define the path integral we assume. The particles have associated with them a Lagrangian L(x. etc.21) N 2. For the sake of simplicity we will consider the case of particles moving in one dimension labelled by the position coordinate x.20) Here. however. . . This allows us to write the evolution operator using (2.4 Propagator for a Free Particle As a first example we will evaluate the path integral for a free particle following the algorithm introduced above.18) The time instances are fixed. at the same time. tj ) to the next (xj+1 . The spacings between the times tj+1 and tj will all be identical. (2. tN |x0 . tN −1 ). tj+1 ) we assume that kinetic energy and potential energy are constant.12. respectively. t0 ). −∞ dxN −1 (xj+1 −xj )2 1 − U (xj ) . > t1 > t0 letting N go to infinity later.12) φ(xN .10) and (2. namely 1 m(xj+1 − xj )2 / 2 and U (xj ). . They can be anywhere in the allowed volume which we will choose to be the interval ] − ∞. 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 . .14 Quantum Mechanical Path Integral 2. 2 2m N (2.9). In passing from one space–time instance (xj .

24).2. tN ). = (2.23) tN − t0 It is essential for the following to note that. tN |x0 . x(tN ) = 0.30) d[x(t)] exp 2 tN − t0 x(t0 )=0 a result. The ˙ 1 S[x(t)|x(t ) = x .25) is.24) implies for the second term on the r. xN − x0 xcl (t) = x0 + ( t − t0 ) . x(tN ) = xN ] 1 (xN − x0 )2 m + 0 + 2 tN − t 0 + S[x(t)|x(t0 ) = 0.12) yields x(tN )=0 im (xN − x0 )2 i φ(xN . 2.13).29) This expression corresponds to the action integral in (2. tN |0.27) The third term can be written in the notation introduced tN 1 dt my 2 = S[x(t)|x(t0 ) = 0. t0 ) = exp 1 im (xN − x0 )2 2 tN − t0 φ(0. . (2. using (2.s.h. t0 ) (2. x(tN ) = xN ] is S[x(t)|x(t0 ) = x0 .31) We have denoted explicitly that the action integral for a path connecting the space–time points (x0 . which can also be written φ(xN . To see the benifit of such approach we define a path y(t) as follows x(t) = xcl (t) + y(t) (2.24) Also we use the fact that the velocity of the classical path xcl = (xN −x0 )/(tn −t0 ) is constant.26) The first term on the r. tN dt y = y(tN ) − y(t0 ) = 0 . The resulting expression for S[x(t)|x(t0 ) = x0 .s.e. ˙ 2 t0 (2. t0 ) = exp S[x(t)] (2.23). it holds y(t0 ) = y(tN ) = 0 . t0 ) and (xN .22) where xcl (t) is the classical path which connects the space–time points (x0 . (2. tN |x0 . the origin of which coincides with the classical path of the particle. 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 . x(tN ) = 0] .10. Inserting the result into (2. tN dt t0 1 1 (xN − x0 )2 m x2 = m ˙ cl . tN ) is to be evaluated. namely.. can be expressed through a path integral with endpoints x(t0 ) = 0. of (2. since x(t0 ) = xcl (t0 ) = x0 and x(tN ) = xcl (tN ) = xN . ˙ ˙ ˙ 2 + y2) = ˙ (2. x(t ) = x ] for any path x(t) can then be expressed through action integral 0 0 N N an action integral over the path y(t) relative to the classical path.28) i. x(tN ) = 0] .h. it is more suitable to define new integration variables yj . One obtains S[x(t)|x(t0 ) = x0 . t0 ) and (xN . 2 2 tN − t0 (2. ˙ t0 (2.4: Propagator for a Free Particle 15 Rather then using the integration variables xj .25) + The condition (2. due to (2.

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

n = 1.43) We like to note here for further use below that one might as well employ the ‘artificial’ starting values D0 = 1. .. t0 ) = m 2πi (t − t0 ) 1 2 exp im (x − x0 )2 2 t − t0 . Using D1 = |(2)| = 2 .41) D2 = 2 −1 −1 2 = 3 (2. .46) ... tN |0.40) . We seek then to evaluate the determinant of the n × n matrix   2 −1 0 .  . . x = xN φ(x..42) m 2πi NN 1 2 (2. 2 −1  0 0 0 −1 2 For this purpose we expand (2.4: Propagator for a Free Particle a property which follows from det(cB) = cn detB for any n × n matrix B.39) yields φ(0.40) in terms of subdeterminants along the last column. 2. presently N − 1. Our derivation has provided us with the value det(Ajk ) = N . . defining t = tN .. 0 0     0 −1 2 . det(Ajk ) (2. . . let say n. . one can readily verify Dn = n + 1 .   . . . One can readily verify that this procedure leads to the following recursion equation for the determinants Dn = 2 Dn−1 − Dn−2 . variable. t|x0 . . . .. which follows from (2. D1 = 2 and obtain from (2. . t0 ) = m 2πi (tN − t0 ) 1 2 .31) and obtain.. . .41) the same result for D2 . t0 ) = limN →∞ m 2πi N 2 17 N 2πi m N N −1 2 1 . tN |0. . (2.44) = tN − t0 . we obtain φ(0.2. (2. . Inserting this into (2. (2.    0 0 0 . D3 .45) Expressions for Free Particle Propagator We have now collected all pieces for the final expression of the propagator (2. t0 ) = limN →∞ and with NN (2. 0 0    Dn =  . . . tN |0. (2. . To solve this three term recursion relationship one needs two starting values. 0 0  −1 2 −1 .39) In order to determine det(Ajk ) we consider the dimension n of (Ajk ). .18) we obtain φ(0... .

according to (2. t0 )| = 2 1 πδ 2 1 2 exp − x2 0 δ2 (2. We devide the contributions to the exponent Eo (xo .50) One refers to such states as wave packets. t0 ) = 1 πδ 2 1 4 exp − x2 po 0 +i x 2 2δ (2. x) = im 2 t x2 1 + i o t mδ 2 − 2xo x − po t m + f (x) (2.52) and applying (2. t) of a free particle.247).48) as the initial state using the propagator (2.48) Obviously.46). t0 ) = m 2πi (t − t0 ) 3 2 exp im (r − r0 )2 2 t − t0 . We want to apply axiom (2.51) For this evaluation we adopt the strategy of combining in the exponential the terms quadratic (∼ x2 ) and linear (∼ x0 ) in the integration variable to a complete square 0 ax2 + 2bx0 = a 0 x0 + b a 2 − b2 a (2.5) allows us to predict the time evolution of any state function ψ(x. One obtains then for the propagator (2. and describes a single particle since 1 πδ 2 1 2 +∞ dx0 exp −∞ − x2 0 δ2 = 1. x) + E(x) in (2. Below we will apply this to a particle at rest and a particle forming a so-called wave packet. (2.46) can be generalized to three dimensions in a rather obvious way. thereby.49) is Gaussian of width δ.18 Quantum Mechanical Path Integral This propagator. x) + E(x) (2.51) as follows Eo (xo . The result (2.53) .10) φ(r. (2. t|r0 .5) to (2. the associated probability distribution |ψ(x0 . centered around x0 = 0. the wave function of the particle at later times. 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 . We need to evaluate for this purpose the integral ψ(x. We will obtain.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 .

the square in (2.4: Propagator for a Free Particle im x2 − f (x) .56) m 2πi t 1 2 +∞ eE(x) −∞ dx0 eEo (xo . making certain.59) An integration path in the complex x0 –plane along the direction i 1−i t mδ 2 (2.55) im  xo 2 t  1+i t − mδ 2 x− 1+ po m t t i mδ2  .52).58) The integrand is an analytical function everywhere in the complex plane and we can alter the integration path.54) f (x) =  This yields Eo (xo .2. We proceed as follows.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  .53) E(x) =  x− po mt 19 (2.61) . 2 t One chooses then f (x) to complete.51) ψ(x. z2 = +∞ × i 1−i t mδ 2 (2. 2 (2. t) = 1 πδ 2 1 4 2 t 1 + i mδ2  . (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 . however. (2.60) is then represented by real ρ values. that the new path does not lead to additional contributions to the integral.x) (2. x) = One can write then (2.57) and needs to determine the integral +∞ I = −∞ +∞ dx0 eEo (xo . according to (2. (2. The beginning and the end of such path are the points z1 = −∞ × i 1−i t mδ 2 .

Quantum Mechanical Path Integral (2. z2 = +∞ . the paths between z1 → z1 and z2 → z2 have a real part of x0 of ±∞.58) has a leading contribution in x0 of −x2 /δ 2 the integrand of (2. One obtains i po i po x x2 2m t E(x) = − 2 + − t t t 2δ (1 + i mδ2 ) 1 + i mδ2 1 + i mδ2 2 (2.54) using (2.58) vanishes 0 for Re x0 → ±∞.63) which can be readily evaluated. using a ab = a − .57) reads then 1 πδ 2 1 4 2πi t .58) along the path z1 → z1 and along the path z2 → z2 gives only vanishing contributions one can replace (2.66) yields ψ(x.67) We need to determine finally (2. (2.62) If one can show that an integration of (2.20 whereas the original path in (2.63) holds and. one can show that z1 lies at −∞ − i × ∞ and z2 at +∞ + i × ∞. Since the exponent in (2.69) t)2 t 2δ 2 (1 + i mδ2 ) + i po x − i p2 o t 2m (2. Hence.64) ψ(x.65) t 1 − i mδ2 t 1 + i mδ2 .68) and. 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. t m(1 + i mδ2 ) (2. accordingly.58) has the end points z1 = −∞ .58) by t mδ 2 +∞ I = i 1−i dρ exp − −∞ m 2 t 1+ t mδ 2 2 ρ2 (2.70) .55). 1+b 1+b finally E(x) = − (x − po m (2. (2. I = Equation (2. t) = Seperating the phase factor 1 t 1 + i mδ2 1 4 1 2 exp [ E(x) ] . We can conclude then that (2. In fact.

(2.71.72).71) × exp − (x − po m t)2 ) 2δ 2 (1 + 2 t2 m2 δ 4 (1 − i .2. (2. t)| = 2 1 πδ 2 (1 + 2 t2 m2 δ 4 1 2 ) exp − (x − δ 2 (1 + po m t)2 ) 2 t2 m2 δ 4 .71) conserves the phase factor exp[i(po / )x] of the original wave function (2.72).74) which corresponds to (2. The conservation of this factor is particularly striking for the (unnormalized) initial wave function ψ(x0 .74) remains invariant in time.77) .72) for the initial state (2. We had assumed above [c. In this case holds ψ(x.49). t) = exp i i p2 po o x − (t − to ) m 2m . (2. Another interesting observation is that the wave function (2.72) Comparision of Moving Wave Packet with Classical Motion It is revealing to compare the probability distributions (2. (2.f.4: Propagator for a Free Particle which inserted in (2. The width of the distribution (2.75) i. the spatial dependence of the initial state (2. (2. This ‘spreading’ of the wave function is a genuine quantum phenomenon. This yields. 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..48) and that the respective phase factor is related with the velocity of the classical particle and of the center of the distribution (2.75) ψ(x.48)] to = 0. However. t0 ) = exp i po xo m .e.48) for δ → ∞. The center of the distribution (2.71). instead of (2.49). 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 (2.73) m δ increases with time. coinciding at t = 0 with the width of the initial distribution (2.72) 2 t2 δ 1+ 2 4 (2. t) = exp i po i p2 o x − t m 2m .48) and for the final state (2. 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 . the case of arbitrary to is recovered iby replacing t → to in (2. t0 ) = exp i po i p2 o xo − to m 2m . 2.76) From this we conclude that an initial wave function ψ(xo . (2. respectively.67) provides the complete expression of the wave function at all times t ψ(x.

x. t) = ˙ 1 1 mx2 − c(t)x2 − e(t)x .81) vanishes2 . 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 . xcl + y(t).83) (2.e. One refers to the respective states as stationary states.78) i.12.79) for a quadratic Lagrangian L(x. y(tN ) = 0 .. 2. t) = exp i Quantum Mechanical Path Integral po i p2 o x − t m 2m . the end points of y(t) are y(t0 ) = 0 Inserting (2. 2.80) For this purpose we need to determine the action integral S[x(t)] = t0 dt L(x. 1.82) one obtains L(xcl + y. t) + δL ˙ ˙ ˙ ˙ where L(xcl . ˙ 2 2 tN (2. Such states play a cardinal role in quantum mechanics. ˙ ˙ = (2. xcl . For this purpose we use d xcl y = ˙ ˙ (xcl y) − xcl y ˙ ¨ (2.13) φ(xN .80) into (2. (2.85) We want to show now that the contribution of δL to the action integral (2.84) .22 becomes at t > to ψ(x.5 Propagator for a Quadratic Lagrangian x(tN )=xN We will now determine the propagator (2. 2. In order to simplify this task we define again a new path y(t) x(t) = xcl (t) + y(t) (2. tN |x0 . the spatial as well as the temporal dependence of the wave function remains invariant in this case. t0 ) = x(t0 )=x0 d[x(t)] exp i S[x(t)] (2. 2 .10. t) + L (y. y(t)] which vanishes according to the Hamiltonian principle as discussed in Sect. y(t). t) ˙ L (y.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 . Obviously. (2. y(t). t) ˙ (2.81) for an arbitrary path x(t) with end points x(t0 ) = x0 and x(tN ) = xN . xcl . t) = L(xcl .82) which describes the deviation from the classical path xcl (t) with end points xcl (t0 ) = x0 and xcl (tN ) = xN . x.

.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. .88) vanishes. . t0 ) = y(tN )=0 d[y(t)] exp y(t0 )=0 i tN dt L (y. 0 0     0 −1 2 . t0 ) we will adopt a strategy which is similar to that used for the evaluation of (2.. 0 0  0 c2 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. . . . tN |0. tN |0. . . . tN |0.h. 0 0  m   ajk =  .32). . .92) where ajk are the elements of the following (N − 1) × (N − 1) matrix   2 −1 0 . .31) for the free particle propagator. . .   . .  . t) ˙ t0 .. .. t0 ) (2.  2 N  . vanishes.87) According to (2. . .24) to the Lagrangian (2.s. . .. x t0 (2. .83) the first term on the r. The discretization scheme adopted above yields in the present case ˜ φ(0. 0 0  −1 2 −1 . For the evaluation of φ(0. y. .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) .k=1 yj ajk yk (2.    0 0 0 .. . One can then express the propagator (2. .2. 0 0   N  0 −  . One can express the exponent E in (2. of (2.91) through the quadratic N −1 E = i j. also the second contribution on the r. . t0 ) = exp S[xcl (t)] φ(0.. . tj = t0 + form j.s. .79) i ˜ φ(xN .  . . tN |x0 . . hence. . . (2. Applying the Euler–Lagrange conditions (1. .. 2 ..88) and. . . cN −2 0  0 0 0 0 cN −1 (2.h.89) where ˜ φ(0.   . . tN |0. 2 −1  0 0 0 −1 2   c1 0 0 .80) yields for the classical path m¨cl + c(t) xcl + e(t) = 0 x (2.93) .  0 0 0 . 0 0     0 c3 .91) (yj+1 − yj )2 1 2 2m N where cj = c(tj ). .

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

1 exp[iω(n + 2 ) to ] · · · . results in 1 ˜ exp[−iω(n + 1 ) t] φn (x) 2 n! +∞ = −∞ dxo Φ(x.171) ∞ W (x. one can expand (2.172) is a function of t − to and defining accordingly Φ(x.172) I = exp 2 mω x e−iωt − e−2iωt − Quantum Mechanical Path Integral mω 2 1 x − iωt 2 2 . . i. 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! (2.171. which also exhibits such invariance.183) where the expansion coefficients are functions of x.184) 1 ˜ exp[−iω(n + 1 ) t] φn (x) 2 n! ∞ +∞ = m=0 −∞ dxo Φ(x. We want to inspect the consequences of the invariance property (2. We note that the factor exp(2 mω/ xe−iωt − e−2iωt ) in (2. . xo .172). t − to ) = φ(x. t) +∞ −∞ dto 1 ˜ exp[−iω(m + 1 ) to ] φm (xo ) 2 m! (2. to ) we express (2. but not of t. t|xo . . 2.s. of (2.186) Fourier transform. t|xo .147) in (2. (2.36 Altogether. . t). t|xo . to ). In contrast to W (x.e. 2 (2.171) concludes the proof of (2. xo .h.187) or 1 ˜ exp[−iω(n + 2 ) t] φn (x) +∞ = −∞ ˜ dxo φ(x. xo . 2. Accoordingly. to ) exp[−iω(n + 1 ) to ] φn (xo ) .172) in the form ∞ n=0 (2. n = 1. Noting that the propagator (2.188) identifies the functions ψn (x.. one obtains for the r.. xo .185) Replacing t → t + to yields ∞ n=0 1 ˜ exp[−iω(n + 1 ) (t + to )] φn (x) 2 n! ∞ +∞ = m=0 −∞ dxo Φ(x.172). t) = n=0 1 ˜ exp[ − iω(n + 1 ) t ] φn (x) 2 n! (2.188) ˜ ˜ Equation (2.171) can be expanded in terms of e−inωt .182) Comparision with (2. t − to ) 1 ˜ exp[−iω(m + 1 ) to ] φm (xo ) 2 m! (2.

198) ∂z n z=0 .195) zn Hn (y) n! ˜ φn (y) . The Hermite Polynomials ˜ The function (2.190) ˜ is obeyed. t)|2 = ˜n (x)|2 . z) (2. 2. z) = Hn (y) (2. 2 Here Nn are constants which normalize ψn (x. t) provide all stationary states of the quantum mechanical harmonic oscillator. through the expansion coefficients φn (x) in (2.196) where Hn (y) = ey 2 /2 (2. according to (2.194) plays a central role for the Hermite polynomials since it contains. In the following we will characterize the functions φn (x) and determine the normalization constants Nn . t) of the quantum mechanical harmonic oscillator ˜ ψn (x.194) = z2 n=0 ∞ 1 ∞ zn ˜ φn (y) n! (2. This follows from ∂n w(y. 1. To obtain ˜n (x) we simplify the expansion (2.183) reads then w(y. t) such that +∞ −∞ 2 dx |ψ(x. Actually. (2. z) = exp(2yz − z 2 ) .194). characterizes the wave functions φn (x). (2.2. We will also argue that the functions ψn (x. . .171). in a ‘nutshell’ all information on the Hermite polynomials. t) are associated with a time-independent probablity density |ψn (x.183).183). ˜ |φ stationary wave functions ψn (x.193) (2.191) (2. t)|2 = Nn +∞ −∞ n = 0.183). through expansion (2. z) z 2 e−y or w(y. t) = exp[−iω(n + 1 ) t] Nn φn (x) .171) W (x. we have identified then.192) w(y. . t) = z 2 e−y where w(y.197) The expansion coefficients Hn (y) in (2. z) = n=0 1 2 /2 1 2 /2 = = mω e−iωt x (2. Expression (2. For this purpose we introduce first closed expressions for φ the new variables y z and write (2. Expansion (2.197) are called Hermite polynomials which are polynomials of degree n which will be evaluated below.189) ˜ dx φ2 (x) = 1 n (2.5: Propagator for a Quadratic Lagrangian 37 ˜ ˜ the functions ψn (x.

200) yields for the first Hermite polynomials H0 (y) = 1. 1990) is a useful introduction to this tool as is a chapter in the eminently useful Concrete Mathematics by R. The identity (2. D. that Hn (y) is a polynomial of degree n.e.199) Comparision with (2.L. . . z) ∂z n = = z=0 2 ∂ n 2 y z − z2 e ∂z n e−(y−z) 2 ey z=0 2 ∂ n −(y−z)2 e ∂z n 2 z=0 2 (−1)n ey ∂n ∂y n = (−1)n ey z=0 ∂n ∂y n e−y (2. Boston. The diagrams also identify the areas over which the summation is to be carried out. Reading. among other properties. closed expressions for Hn (y).E. As will become evident in the present case generating functions provide an extremely elegant access to the special functions of Mathematical Physics3 .201) ν−µ     µ µ ν−µ Figure 2. Massachusetts.38 Quantum Mechanical Path Integral which is a direct consequence of (2. and recursion equations for the efficient evaluation of Hn (y). One calls w(y.200) One can deduce from this expression the polynomial character of Hn (y).194) in a Taylor series in terms of y p z q and identify the corresponding coefficient cpq with the coefficient of the p–th power of y in Hq (y). 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. For this purpose we expand the generating function (2. ν+µ   (2.. ν+µ ν   H3 (y) = 8y 3 − 12y.202) generatingfunctionology by H.196) results in the so-called Rodrigues formula for the Hermite polynomials Hn (y) = (−1)n ey 2 ∂ n −y2 e . H1 (y) = 2y. µ lattice (left diagram) corresponds to a summation over only every other point in an n.196) ∂n w(y.Graham.. 2.Knuth. ν     H2 (y) = 4y 2 − 2.194. .198) for the Hermite polynomials can be expressed in a more convenient form employing definition (2.1: Schematic representation of change of summation variables ν and µ to n = ν + µ and m = ν −µ. normalization factors for φ(y).Wilf (Academic Press. i. . (2.Patashnik (AddisonWesley. z) the generating function for the Hermite polynomials.196) to ˜ derive. m lattice (right diagram). The diagrams illustrate that a summation over all points in a ν. We want to derive now explicit expressions for the Hermite polynomials.S. We will employ (2. and O. Inc.196). ∂y n (2. 1989).

for odd n.209) This property follows also from the generating function. the sum in (2.205) Since (n − m)/2 is an integer we can introduce now the summation variable k = (n − m)/2 .209). according to 2. 2. m are ν = n+m . is a polynomial of degree n. −z) and. In case of even n . it holds Hn (−y) = (−1)n Hn (y) ..5: Propagator for a Quadratic Lagrangian and introduce now new summation variables n = ν + µ. 0 ≤ k ≤ [n/2] where [x] denotes the largest integer p.207) This expression yields for the first four Hermite polynomials H0 (y) = 1.207) contains only even powers.e. µ expressend in terms of n. 39 (2. otherwise. According to (2. H1 (y) = 2y.202) e 2yz − z 2 ∞ = n=0 zn n! ≤n m≥0 n! (−1) n−m 2 n−m 2 ! m! (2y)m . p ≤ x.1. i. 2 (2. . k! (n − 2k)! (2.2. in fact. (2.206) From this expansion we can identify Hn (y) [n/2] Hn (y) = k=0 (−1)k n! (2y)n−2k . . it contains only odd powers. m must be restricted such that either both n and m are even or both n and m are odd. 2 µ = n−m .207) one can deduce that Hn (y).197) ∞ n=0 zn Hn (−y) = n! ∞ n=0 (−z)n Hn (y) = n! ∞ n=0 zn (−1)n Hn (y) n! (2.210) from which one can conclude the property (2. µ are integers the summation over n.194) holds w(−y.203) The old summation variables ν. The lattices representing the summation terms are shown in Fig. Hence. hence. z) = w(y. (2. .201).208) which agrees with the expressions in (2. m = ν − µ 0 ≤ n < ∞. . k! (n − 2k)! = Hn (y) (2. From (2. 0 ≤ m ≤ n . H3 (y) = 8y 3 − 12y. (2. One can write then using m = n − 2k e 2yz − z 2 ∞ = n=0 zn n! [n/2] k=0 n! (−1)k (2y)n−2k . H2 (y) = 4y 2 − 2.204) Since ν. With this restriction in mind one can express (2.

. . as given in (2. H2n (0) = (−1)n Recursion Relationships A useful set of properties for special functions are the so-called recursion relationships. For Hermite polynomials holds. Hn+1 (y) − 2y Hn (y) + 2n Hn−1 (y) = 0. For this purpose one considers w(0. 2. H2n+1 (0) = 0 . dy n = 1. . 2. .217) gives H1 (y) − 2y H0 (y) = 0. 2.215) ∂z which can be readily verified using (2. . have a node at y = 0.233) below. a factor z reduces the order of Hn by one and introduces also a factor n. The reader may verify that w(y. . Starting point of the derivation is the property of w(y.216) Combining the sums and collecting terms with identical powers of z ∞ n=1 zn Hn+1 (y) − 2y Hn (y) + 2n Hn−1 (y) n! + H1 (y) − 2yH0 (y) = 0 (2. for example. z) ∂ w(y. One can then readily state which differential equation of w(y. . 2. .194).188. . indeed satisfies the latter relationship.208). Another relationship is d Hn (y) = 2n Hn−1 (y).209) since odd functions have a node at the origin. n = 1. namely. (2. a property which is consistent with (2. n = 0.40 Quantum Mechanical Path Integral The generating function allows one to determine the values of Hn (y) at y = 0. .197) above and given by (2.215) yields ∞ n=1 z n−1 Hn (y) − 2y (n − 1)! ∞ n=0 zn Hn (y) + 2 n! ∞ n=0 z n+1 Hn (y) = 0 . z). 1. Substituting expansion (2. (2. Hn+1 (y) − 2y Hn (y) + 2n Hn−1 (y) = 0 . .212) n! This implies that stationary states of the harmonic oscillator φ2n+1 (x).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. z) should be equivalent to the relationship (??). n! (2. n = 1. z) = 0 (2.213) which allow one to evaluate Hn (y) from H0 (y) and H1 (y) given by (2. (2. (2. .196) into the differential equation (2. (2.214) We want to derive (??) using the generating function. z) − (2y − 2z) w(y. 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) . dw/dy − 2zw = 0. as defined through (2.194). 2.211) m! n! m=0 n=0 Comparing terms on both sides of the equation yields (2n)! . .

h.200) yields Hn (y) = The identity (−1)n y2 √ e π +∞ dt −∞ dn 2iyt − t2 e . (2. 2 (2. 2.247) for a = i one obtains I(y) = and.s.226a).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. . 2. z) w(y. dy n (2. 2 (2.227) Comparing the terms of the expansions allows one to conclude the orthonormality conditions +∞ −∞ dy Hn (y) Hn (y) e−y 2 √ = 2n n! π δn n .222) Employing this expression now on the r.194.219) which can be written I(y) = e−y 2 +∞ −∞ dte−(t−iy) = e−y √ 2 2 +∞ −∞ dze−z .h.5: Propagator for a Quadratic Lagrangian Integral Representation of Hermite Polynomials An integral representation of the Hermite polynomials can be derived starting from the integral +∞ 41 I(y) = −∞ dt e2iy t − t .194. z ) e−y 2 +∞ = e2 z z = √ ∞ dy e−(y−z−z ) = . 1. 2 √ π e2 z z (2.196) yields ∞ n. .223) dn 2iy t − t2 2 e = (2 i t)n e2iy t − t n dy results. through a double series over Hermite polynomials using (2.2. For this purpose we consider the integral +∞ −∞ dy w(y. in the integral representation of the Hermite polynomials 2n (−i)n ey √ Hn (y) = π Orthonormality Properties 2 (2. e−y 2 π e−y 2 (2. 2 n = 0.224) +∞ −∞ dt tn e2iy t − t .196) the orthogonality properties of the Hermite polynomials.228) .226) π −∞ 2n z n z n n=0 n! Expressing the l.220) Using (2. of the Rodrigues formula (2. 2. acording to the definition (2. . (2.221) 1 = √ π +∞ −∞ dt e2iyt − t .225) We want to derive from the generating function (2. 2 (2. finally.s.

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 2.233) are presented for n = 0. (2. 4 in Fig. 2.231) √ π (2. 4.230) (2. 1.2: Stationary states φn (y) of the harmonic oscillator for n = 0. in harmony with (??). 3. 3. ¡ ¢ ¡ -4   -2 0 2 4 y (2. According to (2. 4. (2. that the wave functions are even for n = 0.228) allow us to construct normalized stationary states of the harmonic oscillator.197) holds 2 ˜ φn (y) = e−y /2 Hn (y) . 3.190)] φn (y) = Nn e−y 2 /2 and for the normalization constants Nn follows from (2. The normalized states are [c. Furthermore. 2. the value of the wave function at y = 0 is positive for n = 0.f. negative for n = 2 and vanishes for n = 1. 2.233) .2. Normalized Stationary States The orthonormality conditions (2. 1.229) Hn (y) . 3.232) Hn (y) . 4 and odd for n = 1. One can also recognize that n is equal to the number of nodes of the wave function.189.228) +∞ 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. in agreement with our above discussions. (2. 2. One can recognize. 2.

Comparision with (2. 5. i. 1]. denoted by Ln (x) and encountered in case of the stationary states of the nonrelativistic [see Sect.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. In case of the Legendre polynomials.. Other examples of orthogonal polynomials are the Legendre and Jacobi polynomials which arise in solving three-dimensional stationary Schr¨dinger equations. 2. g ∈ F .238) where the space is endowed with the scalar product (f |g) = Ω dx f (x) g(x) w(x) = < ∞ . The latter are written for polynomials pn (x) in the general form dx pn (x) pm (x) w(x) = In δnm Ω (2. The Hermite polynomials are so-called orthonogal polynomials since they obey the conditions (2. and In = 2n n! π. and I = 2/(2 + 1). +∞[.f.e. w(x) = 0 only at a discrete set of points xk ∈ Ω (2.228). In case of the associated Laguerre (α) polynomials.2.228) shows that the orthonogality condition of the Hermite polynomials is in complience with (2.179] holds Ω = [−1. f. denoted by P (x) and introduced in Sect. The orthogonal polynomials pn mentioned above have the important property that they form a complete basis in the space F of normalizable functions. Ω (2.156.236) where w(x) is a so-called weight function with the property w(x) ≥ 0. holds Ω = [0. (2. (2.459] hydrogen atom. w(x) ≡ 1. ??? and eq. 5 below [c. (5. w(x) = xα e−x .. i.189) φn (x) = √ 1 2n n! mω π 1 4 e− mωx2 2 Hn ( mω x) .239) . 5.189) by the Jacobian dx/dy.150 . (10.234) The explicit form of the stationary states of the harmonic oscillator in terms of the position variable x is then. (2. w(x) = √ exp(−x2 ).5: Propagator for a Quadratic Lagrangian 43 The normalization condition (2.237) for Ω = R.236 . In = Γ(n+α+1)/n! where Γ(z) is the so-called Gamma function.10 and eq. 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. 10. ???] and relativistic [see Sect. 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. 5. by dx = dy mω 1 4 .231) of the wave functions differs from that postulated in (2. of functions which obey dx f 2 (x) w(x) = < ∞ .151. using (2.e.237) and where In denotes some constants.233) and (2.

Since the Hermite polynomials form a complete basis for such states we can expand ∞ ψ(y..242) and the existence of a normalizable state ψ(y. aT = a. Therefore.243) To be consistent with(2. in particular. i. 2.147). pn (x) → w(x) pn (x) the scalar product (2. −∞ dyn ei −∞ n j. The only possibility for this to be true is dn = 0.242) Let us assume now the case of a function space governed by the norm (2. (2. i.240) where cn = 1 In dx w(x) f (x) pn (x) . ∞ n. t) which is stationary under the action of the harmonic oscillator propagator (2.241) The latter identity follows from (2.244) For the state to be stationary |ψ(x.239) becomes the conventional scalar product of quantum mechanics f |g = Ω dx f (x) g(x) .188. Appendix: Exponential Integral We want to prove +∞ +∞ I = dy1 . In case of n = 1 this reads +∞ −∞ dx ei a x = 2 iπ . Ω (2. If one replaces for all f ∈ F: f (x) → w(x) f (x) and.m=0 d∗ dm exp[iω(m − n)t] e−y Hn (y)Hm (y) . hence. the 2 stationary state ψ(y.197) it must hold cn (t) = dn exp[−iω(n + 1 )t] and.44 As a result holds for any f ∈ F f (x) = n Quantum Mechanical Path Integral cn pn (x) (2. n 2 (2.e.233). t) must be identical to one of the stationary states (2. 2 (2. a state for which (2.. t)|2 . the states (2. .236).e. ψ(y.k yj ajk yk = (iπ)n . t) is ∞ ψ(y. t) = n=0 cn (t) e−y 2 /2 Hn (y) .245) must be time-independent.e. det(a) (2. (2. a (2. i. symmetric a.172) holds. except for a single n = no .e. .233) exhaust all stationary states of the harmonic oscillator.247) ..246) for det(a) = 0 and real. t) = n=0 1 2 dn exp[−iω(n + )t] e−y /2 Hn (y) . i.

248.246) reads then in terms of yj ˜ n n j. .k yj ajk yk ˜˜ ˜ (2. . The proof of (2. . . (2.7: Appendix / Exponential Integral 45 which holds for a ∈ C as long as a = 0. 0 0 .  . 2.e. . ann ˜ where S can be chosen as an orthonormal transformation. . according to (2. . .248) . . ˜ (2.254) ajj .2. 0  ˜   ˜ S−1 a S = a =  .k yj ajk yk n = j.246) exploits that for any real..252) ˜ For the determinant of a holds det(˜) = a n ajj ˜ j=1 (2. . 0 ˜  0 a22 .251) where.249) The akk are the eigenvalues of a and are real. (2.253) as well as det(˜) = det(S−1 aS) = det(S−1 ) det(a) . i. yk = k Skj yj .. This property allows one to simplify the bilinear ˜ form n yj ajk yk by introducing new integration variables j.k n m = = j. ST S = 1 1 or S = S−1 .k n m n y Sj ajk Skm ym ˜ ˜ y (S T ) j ajk Skm ym ˜ ˜ j. symmetric matrix exists a similarity transformation such that   a11 0 . ˜ (2. One can conclude det(a) = j=1 n (2. .250) The bilinear form in (2.   .m (S T )jl alm Smk .k n n yj = ˜ k (S −1 )jk yk . det(S) a = (det(S))−1 det(a) det(S) = det(a) .249) n ajk = ˜ l. (2.255) . . . .

2. ˜ for j = 1. . i. . .e. (2. . . yn ) ∂(˜1 . yn ) ∂(˜1 . ∂ ys ˜ (2.261) ∂(y1 . . . . .46 We have assumed det(a) = 0. . .. . . . yn ) y ˜ ∂yj .262) (2. . . ∂(˜1 . .263) (2.249) follows 1 = det ST S such that one can conclude det S = ±1 One can right then (2.258) where we introduced the Jacobian matrix J = with elements Jjs = According to (2. det( From (2. .258) +∞ +∞ ∂(y1 .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 −∞ . . yn ) ) = det(S) . . . Accordingly. . y −∞ d˜n det y −∞ ∂(y1 . ..257) Substitution of the integration variables (2. d˜n e y −∞ = (2.256) such that none of the eigenvalues of a vanishes. holds n Quantum Mechanical Path Integral ajj = 0 ˜ j=1 (2. n (2. . yn ) y ˜ = ( det S )2 (2. .250) holds J = S and. .266) . yn ) y ˜ ei n k akk yk ˜ ˜2 .260) (2. ajj = 0 .265) k=1 −∞ which leaves us to determine integrals of the type +∞ dx eicx −∞ 2 (2. .250) allows one to express (2.259) (2. ..264) I = d˜1 . . . hence.

270) dx e−cx 2 2 − 2p √ γ4 : z = ix − p J4 = −p i dx eic(ix−p) . The contour intergral (2. for x. c (2.267) by considering the contour integral J = γ dz eicz = 0 2 (2.266) to the well-known Gaussian integral +∞ dx e−cx −∞ 2 = π .271) .c > 0 . One can relate integral (2.257) holds c = 0. p ∈ IR.268) along the path γ = γ1 + γ2 + γ3 + γ4 displayed in Figure 2.269) The contributions Jk can be expressed through integrals along a real coordinate axis by realizing that the paths γk can be parametrized by real coordinates x p γ1 : z = x γ2 : z = ix + p γ3 : z = √ ix J1 = −p p dx eicx 2 J2 = J3 = = i dx eic(ix+p) 0√ − 2p √ √ 2p 2 i dx eic( 2p √ ix)2 √ − i 0 √ (2.7: Appendix / Exponential Integral 47 where. the integrand does not exhibit any singularities anywhere in C.e.. Jk = γk dz eicz 2 (2. The contour integral (2.268) vanishes. 2 since eicz is a holomorphic function. (2. We consider first the case c > 0 and discuss the case c < 0 further below. according to (2.268) can be written as the sum of the following path integrals J = J1 + J2 + J3 + J4 .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 .3. i.

(2. c   ¡ -p γ1 p Re(z) 2 | |e−2cxp | . Hence.274) (2. (2.272) = 0 .275) . One can state accordingly ∞ J = dx e −∞ icx2 − √ ∞ i dx e−cx −∞ 2 Using 2. p p→+∞ lim |J2 or 4 | ≤ = p→+∞ 0 lim dx |e−2cxp | 1 − e−2cp 2cp = 0.3: Contour path γ in the complex plain. 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.48 Quantum Mechanical Path Integral Im(z) ip γ2 ¡ γ4 ¡ γ3 ¡ -i p Figure 2.268) for p = +∞. We will now show that the two integrals J2 and J4 vanish for p → +∞. (2.273) p→+∞ lim J2 and J4 do not contribute then to integral (2.267) one has shown then ∞ dx eicx −∞ 2 = iπ .

276) and (c < 0) ∞ dx eicx −∞ 2 = −iπ = −|c| iπ . if one chooses the same contour integral as (2.268).2. This leads to ∞ icx2 J = dx e −∞ + √ −∞ −i ∞ dx ecx = 0 2 (2. 2. It holds n I = k=1 iπ = akk ˜ (iπ)n . (iπ)n det(a) (2.277) to (2.7: Appendix / Exponential Integral 49 One can derive the same result for c < 0. n ˜ j=1 ajj (2.275.255) this result can be expressed in terms of the matrix a I = which concludes our proof.265). c (2. but with a path γ that is reflected at the real axis.279) .277) We apply the above results (2.278) Noting (2.

50 Quantum Mechanical Path Integral .

the Schr¨dinger o o equation provides the simpler avenue towards describing quantum systems than the path ingral formulation of Section 2. 2.20) to evaluate the propagator. In the limit of very small it is sufficient to employ a single discretization interval in (2.3) In order to carry out the integration we set r0 = r + s and use s as the new integration variable.5) ψ(r.1) can be expressed through the discretization scheme (2. even in x1 . This yields ψ(r. x2 . t) by an infinitesimal time step .2) d 3 r0 m 2πi 3 2 exp i m (r − r0 )2 − 2 U (r.21). .h.s.20.1 Derivation of the Schr¨dinger Equation o We will consider now the propagation of a wave function ψ(r. t + ) = Ω m 2πi 3 2 exp i m (r − r0 )2 − 2 U (r. t + |r0 . Generalizing (2. t + ) = × exp i 2 +∞ +∞ +∞ −∞ dx1 −∞ dx2 −∞ dx3 m x2 + x2 + x2 1 2 3 m 2πi 3 2 × (3.s. namely the Schr¨dinger equation.4) − U (r.1) We will expand the l. The propagator in (3. t + |r0 . t + ) = Ω d3 r0 φ(r. and the r. x3 51 . We will denote the components of s by (x1 . (3. but by no means all. t) = From this follows ψ(r. t) satisfies a partial differential equation. t) .20) to R one obtains then for small φ(r. t) .h. (3. t) ψ(r0 . t) ψ(r + s. For many situations. t) . x3 )T . (3. t) ψ(r0 . Notable exceptions are non-stationary systems involving time-dependent linear and quadratic Lagrangians. of this equation in terms of powers of and we will demonstrate that the terms of order require that ψ(r.Chapter 3 The Schr¨dinger Equation o 3. It holds then according to (2. t) . x2 .

O( 7 2 ).8) (3.7) According to (2. x2 .4) one identifies 2 1 = a m = O( ) (3. O( 7 2 ) . The integration involves only the wave function ψ(r +s. n = 0. i ∂a Starting from (3. t) + ∂xj 2 3 xj xk j.. e. x3 )T .52 Schr¨dinger Equation o It is important to note that the integration is not over r. x2 .. Since the latter contributes to (3. U (r.7) shows 1 ∂ In (a).12) Here one needs to note that we are actually dealing with a three-fold integral. we need then to evaluate integrals of the type +∞ In (a) = −∞ dx x2n exp i a x2 . a supposition which will be examined below. but over s = (x1 . consequently. t) and the kinetic energy term.. t). (3.k=1 xj xk ∂x2 ∂x2 ψ(r.13) 2πi a .5) assuming that only the leading terms contribute. 1. a I2 (a) = − 3 4a2 iπ .. 1 4 3 2 2 ∂4 j. a (3.4) is even in all three coordinates x1 .11) holds 3 m 3 iπ 2 2 × = 1 (3.9) iπ a (3. t) .6) make contributions of the order O( 3 2 ) . of the expansion of ψ(r + s. t) + j=1 xj ∂ 1 ψ(r..5) which are even separately in all three coordinates yield non-vanishing contributions. .36) holds I0 (a) = Inspection of (3.g.k=1 ∂2 ψ(r. x3 . (3. 2 (3.8) one can evaluate recursively all integrals In (a). t) j k . t) j 1 12 . It is then sufficient to consider the terms ψ(r. 1 2 3 2 ∂2 j=1 xj ∂x2 ψ(r. t) + . Since the kinetic energy contribution in (3. t) is a constant with respect to this integration. It holds In+1 (a) = I1 (a) = i 2a iπ . t) j .11) and. .10) It is now important to note that in case of integral (3. O( 5 2 ) . t) = ψ(r.4) only for small x2 + x2 + x2 values 1 2 3 we expand 3 ψ(r + s. Obviously.6) 3 4 ∂4 j=1 xj ∂x4 ψ(r. According to (3. the terms collected in (3. ∂xj ∂xk (3. only terms of the expansion (3..

The 2nd order spatial derivatives require that one specifies also properties of the solution on a closed boundary ∂Ω surrounding the volume Ω in which a solution is to be determined. t) ψ(r. The following general remarks can be made about the solution. t) . t) = ∂t 2 − 2 2m + U (r. using (3. t) = H ψ(r. (3. t) + ∂ ∂t ψ(r.. t) = ψ(r. t) ψ(r.e.g. 1st order in time. We will derive briefly the type of boundary conditions encountered.20) 2m 3.3. The 1st order time derivative requires that for any solution a single temporal condition needs to be specified. t). e. a solution is thought for t ≥ t0 and the solution is specified at the intial time t0 .14) results in ψ(r.19) ∂t where 2 2 ˆ H = − + U (r. t1 ) = f (r). t) + O( 2 ) . t) − + i 2 2 i U (r. t) + O( 2 ) .16) Inserting this into (3. (3. t) = 1 − (3. t) + 1 2i 4 m 2 53 ψ(r. 2nd o order in the spatial variables and linear in the solution ψ(r. t + ) = ψ(r.14) This expansion in terms of powers of suggests that we also expand ∂ ψ(r. In order O( ) the equation reads i ∂ ψ(r. Obviously. this equation is trivially satisfied to order O( 0 ).. i. t) . t) + O( 2 ) ∂t i U (r. Usually.15) ψ(r. t) + O( 2 ) . t)ψ(r. This equation is often written in the o form ∂ ˆ i ψ(r. 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 . ψ(r. t) + and exp − i U (r. t) (3. t + ) = exp − i U (r.17) 2m ψ(r. Due to its linear character any linear combination of solutions of the time-dependent Schr¨dinger o equation is also a solution. t) (3. (3.18) This is the celebrated time-dependent Schr¨dinger equation. (3. one specifies the so-called initial condition.10).2 Boundary Conditions The time-dependent Schr¨dinger equation is a partial differential equation.2: Boundary Conditions and one can conclude.

Chapter 1. 2nd Ed. that H is 2 . .27) ˆ We will postulate below that H is an operator in H which represents energy. implies hermitian ˆ f | H |g 1 2 Ω ˆ = g| H |f Ω (3. D. in principle. Calssical Electrodynamics. 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. g|r) is called the ˆ concomitant of H and is P (f ∗ . Interchanging f ∗ (r) and g(r) results in g| H |f Ω = Ω ˆ d3 r g(r) Hf ∗ (r) . however.26) the vector–valued function P (f ∗ .22) ˆ where H is defined in (3.22) and (3. The reader is advised to consult a reference text on ‘Linear Algebra’ to follow this argument. Jackson.23).. f ∗ (r) d3 rg(r) Ω 2 = − 2m d3 rf ∗ (r) Ω 2 g(r) − f ∗ (r) (3. g|r) (3. g|r) = − 2 2m ( f ∗ (r) g(r) − g(r) f ∗ (r) ) (3. The hermitian property. g ∈ H is defined as follows f |g Ω = d3 rf ∗ (r)g(r) (3. for example.23) ˆ Since H is a differential operator the expressions (3. New York. hence.26) where ∂Ω da · A(r) denotes an integral over the surface ∂Ω of the volume Ω. differ from each other. (3. Since energy is a real ˆ ˆ quantity one needs to require that the eigenvalues of the operator H are real and. t).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. (John Wiley.28) See.25) ˆ f | H |g Ω ˆ = g| H |f Ω + ∂Ω da · P (f ∗ . In (3. 1975).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) ) (3.20). J. The difference between the integrals is g| H |f = Ω Ω = − g| H |f d3 rf ∗ (r) − Ω 2 2 2m g(r) − Ω d3 rg(r) 2 − 2 2m f ∗ (r) + Ω 2 d3 rf ∗ (r) U (r. t) g(r) − Ω d3 rg(r) U (r.

(3. 3.30) can be postulated. t) is ∂t p(ω. to avoid discontinuities in ψ(r.g. t) + ψ(r. κ > 0 or like r−α . α > 2. t) = ω d3 r |ψ(r. t)∂t ψ(r. r ∈ ∂Ω or da · f (r) = 0 ∀r. i.32). i. 3.29.. we can only allow functions which make the differential da · P (f ∗ . i. In this case one can postulate both conditions (3. t)|2 . r ∈ ∂Ω (3.29) In fact. However. e. . g|r) vanish on ∂Ω. In this case one can expect that the particle density is localized in the area where the energy of the particle exceeds the potential eenrgy.e. in this case also all derivatives of f (r) vanish at infinity.e. 3. The latter property stems from the fact that the boundary condition (3. (3.31) usually arises when a particle existing in a bound state is described. and that the density decays rapidly when one moves away from that area. t) which describes the state of o a particle moving in the potential U (r. in which case ∂Ω1 is the inner sphere and ∂Ω2 is the outer sphere.e. .30) each condition holding on an entire surface ∂Ωj . if f and g satisfy these conditions than also does any linear combination αf + βg. The observable directly linked to the wave function is the probability to find the particle at position r at time t. like exp(−κr). t).3 Particle Flux and Schr¨dinger Equation o The solution of the Schr¨dinger equation is the wave function ψ(r. namely. t) ] . (3. |ψ(r. Often the closed surface of a volume ∂Ω is the union of disconnected surfaces3 ∂Ωj . Since the total probability of finding the particle anywhere in space is d3 r|f (r)|2 = 1 (3.. t) = ω 3 d3 r [ ψ ∗ (r. therefore. t) on a single connected surface ∂Ωj only either one of the conditions (3.30) Note that these boundary conditions are linear in f . t)|2 .32) the wave function must decay for |r| → ∞ rapidly enough to be square integrable. .33) The time derivative of p(ω. The probability to observe the particle anywhere in the subvolume ω ⊂ Ω is p(ω. A most common boundary condition is encountered for the volume Ω = R3 in which case one postulates lim f (r) = 0 “natural boundary condition” . ∂Ω = ∂Ω1 ∩ ∂Ω2 ∩ ∂Ω3 ∩ .3. In either case does f (r) and all of its derivatives vanish asymptotically. It must hold then for all f ∈ H f (r) = 0 ∀r. t)∂t ψ ∗ (r..29.34) An example is a volume between two concentric spheres. obey (3.31) |r|→∞ (3..3: Particle Flux 55 and.

t). (3.40) One can rewrite then (3. (3. A natural choice for this constant is 1. t) = [ ψ(r. t) . Note that for a normalized wave function the quantity ρ(r.56 Using (3. We will assume in the remainder of this Section that the solutions discussed are normalized.18) by any complex number and accordingly one can define ψ(r. t) = |ψ(r.29.26.37) If one applies this identity to ω = Ω one obtains according to (3.44) ˆ Note that the Hamiltonian H involves only real terms. One refers to such solution as normalized.43) (3.39) d3 r · A(r) (3.41) holds for any volume ω ⊂ Ω one can conclude ∂t ρ(r. t)|2 = 1 Ω (3. One can multiply the solution of (3. t) (3. t) = ∂ω da · P (ψ ∗ (r.42) 2mi Since (3. t) H ψ ∗ (r.27) this can be written i ∂t p(ω. t) = ω Schr¨dinger Equation o ∂ ∗ ˆ ψ (r.37) can be expressed through a volume integral according to da · A(r) = ∂ω ω (3. t) = 0. ψ|r. Accordingly the probability to observe the particle anywhere in the total volume Ω is constant. t) = H ψ ∗ (r. t) ψ(r. t) − ψ(r. t) + 4 · j(r. (3. t) ] . t) ∂t ˆ ˆ ψ ∗ (r. t) = 0 .27) one can express this j(r.30) ∂t p(Ω. t) . t) + ω · j(r. t) ψ ∗ (r.37) d3 r ∂t ρ(r.19) and its conjugate complex4 −i yields i ∂t p(ω.41) where j(r. The surface integral (3. (3. .38) holds. ψ(r. t) such that d3 r|ψ(r. t) = P (ψ ∗ . 3. Using (3.36) According to (3. t)|2 is a probability density with units 1/volume. t) − ψ ∗ (r. t) H ψ(r. 3. t) = 0 (3.35) d3 r . t)r.

One can readily show by insertion into (3.3. (2.47) . t0 ) = [2π]− 2 Ω∞ 3 ˜ d3 k φ(k) exp i(k · r − ωt0 ) . t0 ) determines φ(k).48. Such case arose in Sect.48) that the general solution is of the form ψ(r.f. 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.148.f. for f (r) ∈ R. t) is that of density flux. t) points in the direction to the outside of volume ω. (3.71)].50) 3 ˜ d3 k φ(k) exp i(k · r − ωt) Ω∞ (3. t) = − ∂ω da · j(r. Note that j(r.4: Free Particle 57 The interpretation of j(r.e. (3.18) in Ω∞ = R in the case o U (r. (2.43) that any real wave function ψ(r. the initial condition at ψ(r. 2. This finding is consistent with the present evaluation of the particle flux: a factor exp(ipo xo / ) gives rise to a flux po /m. In Sect. 2 for a free particle [c. 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) .46) i. t) = 0 2 ∂ 2 i ψ(r. t) (3. i.41) written in the form ∂t ω d3 r ρ(r.45) Obviously.49) reads at t = t0 ψ(r.51) .. It is of interest to note from (3. t) . 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. t) gives rise to a descrease of the total probability in volume ω due to the disappearence of probability density at the surface ∂ω.105. 2.167)] potential. j(r. t) = − ψ(r. equal to the velocity of the particle. This follows directly from an inspection of Eq. 2m (3. Equation (3.49) ˜ Obviously. t) has vanishing flux anywhere. (3.48) ∂t 2m which describes the motion of free particles.4 Solution of the Free Particle Schr¨dinger Equation o We want to consider now solutions of the Schr¨dinger equation (3. (2.f. and for particles moving in a linear [c. 2. m (3. arises solely from the complex factor exp(i k · r).125)] and in a quadratic [c. (3.. t) = [2π]− 2 where the dispersion relationship holds ω = k2 .

The solution as stated is defined in the infinte space Ω∞ = R . t|r0 . (t − t0 ) (3.58 The inverse Fourier transform yields 3 ˜ φ(k) = [2π]− 2 Schr¨dinger Equation o d3 r0 exp(−ik · r0 ) ψ(r0 . Comparision with Path Integral Formulation One can write solution (3. evaluating the integral in (3. in (2.56) and using (2. t) = Ω∞ d3 r0 φ(r. i. We have identified then with (3.47). To show this one needs to note 1 2π = +∞ dk1 exp −∞ ik1 (x − x0 ) − 1 2 i 2 k2 1 2m . The general form for this propagator involves an expansion in terms of a complete set of eigenfunctions as in (3..49.54) This expression obviously has the same form as postulated in the path integral formulation of Quantum Mechanics introduced above. In case of the harmonoc oscillator the expansion can be stated in a closed form given in (4. 3.e.51) to be square integrable it follows according to the properties of the Fourier–transform that ψ(r.54) to the case of non-vanishing potentials U (r).53) where φ(r.114) and (4. t|r0 . (3.e.247). t) as given by (3.31) applies.54) yields (2. Below we will generalize the propagator (3.54) valid for this case. In fact. t0 ) . The ensuing solutions are called wave packets.70) derived below for a particle in a box and the harmonic oscillator. derive an expression similar to (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 (3.81) .52) We have not specified the spatial boundary condition in case of (3.47). hence. t0 ) (3. If one chooses the initial state f (r) defined in (3.51.49) is square integrable at all subsequent times and. t0 ) = 1 2π 3 d3 k exp Ω∞ ik · (r − r0 ) − i 2 k2 2m (t − t0 ) . t0 ) ψ(r0 . respectively. 3.5).49).54) an alternative representation of the propagator (2. Ω∞ (3. i.. that the “natural boundary condition” (3.52) above ψ(r.

t) = f (t) φ(r) . the functions obey boundary conditions of the type (3.3. (3. t|x0 .e. 3.27) vanish on the surface ∂Ω of Ω.49.58) 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 (3.57–3. We have demonstrated then in this case that the Schr¨dinger formulation of Quantum o Mechanics is equivalent to the Feynman path integral formulation.46). t|x0 .55.. Stationary States We consider now solutions of the time-dependent Schr¨dinger equation (3.53.63) . t0 ) = 1 2π +∞ dk exp −∞ ik (x − x0 ) − i 2m (t − t0 ) .61) × exp − (x − po m t)2 2δ 2 (1 + 2 t2 ) m2 δ 4 (1 − i . i.f. Accordingly. a result which is identical to the expression (??) obtained by means of the path integral propagator (2.60) yields with t0 = 0 ψ(x.59) which is solved through completion of the square in the exponent [c.29. (3.52) to the case that the initial state of a 1-dimensional free particle ψ(x0 .20) which are of o the form ψ(r. (3. 3. t) = −∞ dx0 φ(x.4: Free Particle Free Particle at Rest 59 We want to apply solution (3. r ∈ ∂Ω (3. 3.57) where φ(x. (3.56)].62) We will restrict the space of allowed solutions to a volume Ω such that the functions also make the concomitant (3. The 1-dimensional version of (3.64) (3.60) =???? Combining (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. r ∈ ∂Ω or da · φ(r) = 0 ∀r. t) 2 k2 (3. The remaining integration over k leads to the integral +∞ −∞ dk exp ikx − 1 2 (k − po 2 2 ) δ − i k2 m (t − t0 ) (3. t) is given by (??).54) is +∞ ψ(x. t0 ) ψ(x0 .19. 3. the boundary conditions are φ(r) = 0 ∀r.30).

62) which consist of two factors. n = 1. 2. (3. It follows from the observation that for the solution space considered (3. namely f (t) = f (0) exp − i Et . . in general. (3. t) = f (0) exp − i E φE (r) where ˆ H φE (r) = E φE (r) . .62) we insert it into (3.42) the flux j(r.70) The task of finding solutions φ(r) which solve (3. We want to demonstrate this property now. however. We will encounter many such problems in the subsequent Sections. .71) . one factor depending only on the time variable and the other only on the space variables are called separable in space and time.62) do exist and we will characterize the two factors of the solution f (t) and φ(r).66) i.e. the solutions o (3. by no means all solutions are of this type. according to (3.63. if f (t) = eiα . We must postulate therefore ∂t f (t) ˆ H φ(r) = E f (t) = E φ(r) . solutions exist only for a set of discrete E ˆ values. t) = Ω Ω d3 r |ψ(r. (3. It is important to realize that the separable solutions (3. One can conclude that the probability density for the state (3.. α ∈ R . a common case is Ω = Ω∞ and the ‘natural boundary condition’ (3. At this point we will accept that solutions φ(r) of the type (3.69) If these two equations can be solved simultaneously a solution of the type (3. t)|2 = |f (t)|2 Ω d3 r |φ(r)|2 (3.69) exist. This can hold only if |f (t)| is time-independent. t) vanishes on the surface of ∂Ω.19). often only for a discrete set of values En .45) that the total probability d3 r ρ(r.e. the eigenvalues of the operator H. 3.60 Schr¨dinger Equation o and affect only the spatial wave function φ(r).65) is constant.. hence.27) holds and.62) have the particular property that the associated probability distributions are independent of time. for the eigenvalue problems in the confined function space.67) (3. In fact.62) is |ψ(r.69) is called an eigenvalue problem. As pointed out above.e.64). It follows then from (3. We denote the corresponding solution by φE (r).. for functions required to obey boundary conditions (3. (3.31). We have then shown that 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. is time-independent. At this point we state without proof that.62) are special solutions of the timedependent Schr¨dinger equation. t)|2 = |φ(r)|2 . We will demonstrate that solutions of the type (3. One calls such states stationary states.62) exists. g2 (t) = f (t). The identity (3. It turns out that a solution for f (t) can be found for any E. In order to further characterize the solution (3.68) ˆ where g1 (t) = i ∂t f (t). i. and h2 (r) = H φ(r). We may note in passing that solutions of the type (3. This yields an expression g1 (t) h1 (r) = g2 (t) h2 (r) (3. h1 (r) = φ(r). i.

t) = exp − i 2 k2 2m t exp i k · r (3. t) = exp − E t 2 φE (r) .28) both represent the state φE (r). of course. E (3.71) are. i.76) has kinetic energy 2 k 2 /2m.e. should be positive.. E ∈ R.75) 2m One can ascertain this statement by inserting the expression for φk (r) into the eigenvalue problem posed in (3. − 2 2m φ(r) = E φ(r) . k ∈ R 2 k2 φk (r) = N exp i k · r . ˆ d3 r φ∗ (r) H φE (r) = E Ω Ω ˆ d3 r φE (r) H φ∗ (r) . The flux corresponding to (3. m (3. a property which is to be expected since the energy is purely kinetic energy which. Stationary State of a Free Particle We consider now the stationary state of a free particle described by i ψ(r. As a result we cannot postulate in the present case that wave functions are localized and normalizable. t) = |N |2 k . in fact.43) is j(r.28) for the special case that f and g in (3.71) this yields E Ω d3 r φ∗ (r) φE (r) = E ∗ E Ω d3 r φE (r) φ∗ (r) . = E. E (3. .74) The classical free particle with constant energy E > 0 moves without bounds in the space Ω∞ . particles scattered of a potential.4: Free Particle 61 is a solution of the time-dependent Schr¨dinger equation (3.3.g. o According to (3. The solution φE (r) corresponding to the eigenvalue problem posed by (3. hence.72) According to (3.74) using exp i k · r = ik exp i k · r . real. 3.73) from which follows E = E ∗ and.74) is actually best labelled by an index k. The corresponding stationary solution ψ(r.20). We start our proof using the property (3.19. (3. The resulting total energy values E are positive. one can interpret then k as the magnitude of the momentum of the particle. We will show in Section 5 that E can be interprerted as the total energy of a stationary state. e. We want to prove now that the eigenvalues E which arise in the eigenvalue problem (3.77) Noting that k can be interpreted as the magnitude of the momentum of the particle the flux is equal to the velocity of the particle v = k/m multiplied by |N |2 .66) E must be real. We will wave this assumption as we always need to do later whenever we deal with unbound particles. (3. Obviously.76) according to (3.

(3. 3. (o) 2 k2 (e) 2 k2 2m = E (3.80) need to be satisfied are φE (e. . t) = − d2 ψ(x. 2a n = 1. φ(±a) = 0 ..83) The solutions (3.78) (3.82) One can readily verify that (3. In case of the even solutions (3. (3. n ∈ N. hence. 3. 5 .e. a}. a] ⊂ R which vanish on the surface ∂Ω1 = {−a. so-called even solutions obeying φ(x) = φ(−x) φE (x) = A coskx .5 Particle in One-Dimensional Box As an example of a situation in which only bound states exist in a quantum system we consider the stationary states of a particle confined to a one-dimensional interval [−a.79) The stationary solutions have the form ψ(x. We assume. and so-called odd solutions obeying φ(x) = −φ(−x) φE (x) = A sinkx. The boundary conditions which according to (3.81) 2m = E. that the solutions obey this symmetry as well. Solutions of the Schr¨dinger Equation in F1 o The time–dependent solutions satisfy i ∂t ψ(x.80) 2m dx2 We note that the box is symmetric with respect to the origin. f continuous.84) . a] ⊂ R → R. t) = exp(−iEt/ )φE (x) where φE (x) is determined by 2 d2 − φE (x) = E φE (x) .o) (−a) = 0 (3. 2m dx2 2 (± ) = } (3.81.80). 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. It turns out that this boundary condition can only be satisfied for a discrete set of k–values kn . .81) they are kn = nπ .82) have the property that either both boundary conditions are satisfied or none. We can expect. Hence. (3. two types of solutions. therefore. t) .82) satisfy the differential equation in (3. let say the one at x = a.o) (a) = 0 and φE (e. 3.62 Schr¨dinger Equation o 3. a] ⊂ R assuming that the potential outside of this interval is infinite.81. we have to consider only one boundary condition. i. f ∈ F1 = {f : [−a. . We will refer to this as a particle in a one-dimensional ‘box’.

(3.91) holds. 4. respectively.o) (a. Notice that the number of nodes of the wave functions increase by one in going from one state to the state with the next higher energy En . . 3. . (3.86). (3.82).5: Particle in One-Dimensional Box since for such kn 63 nπa nπ = cos 2a 2 In case of the odd solutions (3. By counting the number of their nodes one can determine the energy ordering of the wave functions.94) .e. φ(e) (a. 3. (3. This condition implies for the even states +a |An |2 −a dx cos2 nπx = |An |2 a = 1 . according to (3. x) = An cos n and φ(o) (a. 2a 1 . (3. . 3.84. 3. .86) satisfy the boundary condition since for such kn nπa sin(kn a) = sin 2a = sin nπ 2 = 0.81.82).81) and (3. 4. n = 2. 3 . a n = 2. 4.87) (Note that.) The Energy Spectrum and Stationary State Wave Functions The energy values corresponding to the kn –values in (3.88) where the energies for odd (even) n–values correspond to the even (odd) solutions given in (3.90) 2a The wave functions represent stationary states of the particle in a one-dimensional box. 2a n = 1. x) = An sin n nπx . . 5 . 2a n = 1. . . . are En = 2π2 8ma2 n2 . . according to the dispersion relationships given in (3. The wave functions for the five lowest energies En are presented in Fig. n = 1. . 2. = 0.82) only the kn –values cos(kn a) = cos kn = nπ . (3. 6 .. (3. 5 . . i. .93) The normalization constants are then An = (3. It is desirable to normalize the wave functions such that +a −a dx |φ(e. (3. n is assumed to be even. x)|2 = 1 n (3. 6 .1).86).92) and for the odd states +a |An |2 −a dx sin2 nπx = |An |2 a = 1 .89) nπx . 2a n = 2. 6 . .85) (3.3.

4. x) for n = 1. 3.o) (a. 2. 5 of particle in a box.64 Schr¨dinger Equation o E h2 ] [ 8 m a2 25 ¢ 16 9 4 Figure 3.1: Eigenvalues En and eigenfunctions φn (e. ¡   1 -a a x .

Because of the property (3. .98) the elements of B1 must be linearly independent.3.96) f |g Ω1 = −a dxf (x) g(x) . In fact.99) Since in this case the integrand is a product of an even and of an odd function.90. 3. m odd. (3. In case of n. . it holds: f |f Ω1 = 0 → f (x) ≡ 0. x) . . m even φn |φm 1 a Ω1 = 1 a +a nπx −a dx sin 2a sin mπx = 2a (3. the integral arises φn |φm 1 a Ω1 = 1 a +a nπx −a dx cos 2a cos mπx = 2a (3. it holds φn |φm Ω1 = δn m . (ii) n.5: Particle in One-Dimensional Box The Stationary States form a Complete Orthonormal Basis of F1 We want to demonstrate now that the set of solutions (3. a] are N. m both even.98) The latter property is obviously true for n = m. 3. . 3.95) 1 a cos nπx 2a sin nπx 2a for n = 1. i. 5 . n = 1. x) = together with the scalar product5 +a 65 (3.101) +a −a dx cos (n−m)πx − cos (n+m)πx 2a 2a and.e. .. .} where φn (a. . Similarly. m both odd. (3.102) holds according to (3. 3. Hence we need to consider only the first two cases. Obviously.97) form an orthonormal basis set. and (iii) the mixed case. 4.. 2. (i) n. the integral vanishes. x) (3.100) +a −a dx cos (n−m)πx + cos (n+m)πx 2a 2a The periods of the two cos-functions in the interval [−a. n = m. 2a 2a (3.98) ∞ f |f 5 Ω1 = n=1 d2 . one obtains for n. the integrals vanish. N ≥ 1. this integral vanishes. the integrand is odd. n (3. 6 . The latter case leads to integrals φn |φm Ω1 = 1 a +a dx cos −a nπx mπx sin . f. g ∈ F1 (3. too. In particular. for ∞ f (x) = n=1 dn φn (a. i. . hence. In case of n = m we have to consider three cases.94) B1 = {φn (a.89.e. for n = 2.103) We will show in Section 5 that the property of a scalar product do indeed apply.

the coefficients an . 1. . n = 1. i.98) one obtains +a dx0 φm (a. . Hence. t) at times t > t0 . such demonstration can be based on the theory of Fourier series. Restricting the expansion (3. 2.78) can be expressed as a linear combination of the elements of B1 defined in (3..108) into the Schr¨dinger equation yields o +a ∞ n=1 φn (a.66 Schr¨dinger Equation o f (x) ≡ 0 implies f |f Ω1 = 0 which in turn implies dn = 0 since d2 ≥ 0. Evaluating the Propagator We can now use the expansion of any initial wave function ψ(x. x) to obtain an expression for ψ(x. In the present case.95. in (3. This implies that only the trigonometric functions which are elements of B1 enter into the Fourier ˜ series. t0 ) = dm . 3.95) is also complete.110) . . 2. x0 ) ψ(x0 .106) and generalizing to t ≥ t0 one can write ∞ +a ψ(x. t0 ) (3. n = 2. .} and {bn .105) to the interval [−a. Accordingly. We have then shown that any f corresponding to elements of F1 can be expanded in terms of elements in B1 .79) requiring the o initial condition cn (t0 ) = 1 . (3. 4. any element of the function space F1 defined in (3. t0 ) = n=1 dn φn (a. ( ) = ([ + ] − ) (3.106) Using the orthonormality property (3. . 5 .104) ˜ where [y]a = y mod2a. t0 ) = +a ∞ i n=1 − En φn (a. m = 1. Demonstration of completeness is a formidable task. t) = n=1 φn (a. they can be expanded in terms of a Fourier series. (3. We like to show finally that the basis (3.e. x) .95) is an orthonormal basis. . and bn . (3. however. −a (3.109) Insertion of (3.105) must vanish.e.105) f an cos 2a 2a n=0 n=1 ˜ The functions f corresponding to the functions in the space F1 have zeros at x = ±m a. .. x) cn (t) −a dx0 φn (a. there exist real constants {an . For this purpose we extend the definition of the elements of F1 to the whole real axis through ˜ f : R → R. .96). x) ∂t cn (t) −a dx0 φn (a. . It follows that B1 n defined in (3. i. a] yields then also an expansion for any element in F1 and B1 is a complete basis for F1 . x0 ) ψ(x0 . 3. x) cn (t) −a dx0 φn (a. . . t0 ) in terms of eigenfunctions φn (a. The functions f are periodic with period 2a. For this purpose we expand ∞ ψ(x. 3. . n = 0. x0 ) ψ(x0 .108) where the functions cn (t) are to be determined from the Schr¨dinger equation (3.} such that ∞ ∞ nπx nπx ˜ = + an sin (3. n = 1. t0 ) . x0 ) ψ(x0 .107) Inserting this into (3. . .

3.98).114) This expression has the same form as postulated in the path integral formulation of Quantum Mechanics introduced above. a] yields. t) = −a dx0 φ(x. 67 (3. 4 a (3. t0 ) = ∞ φn (a. according to (3. t|x0 . t0 |x0 .e. but the collision with the left wall leads to new interference effects.79) which lies in the proper function space (3. This solution can be written +a ψ(x. σ = a 15 . (3. .113). t|x0 . In the present system which is composed solely of bound states the propagator is given by a sum. In case that different boundary conditions hold the propagator will be different as well. It is of interest to note that φ(x. k0 = .108. 3.54) as in the case of the free particle system which does not exhibit any bound states. t0 ) = 1 2πσ 2 1 4 exp − x2 0 + i k0 x0 2σ 2 . t0 ) ψ(x0 .109) are cm (t) = exp i − Em (t − t0 ) . x0 ) .113) where φ(x. Example of a Non-Stationary State As an illustration of a non-stationary state we consider a particle in an initial state ψ(x0 . x) and integrating over [−a.2 presents the probability distribution of the particle at subsequent times. cm (t0 ) = 1 . (3. i. in (2. Often the propagator φ(x. t0 ).5). t0 ) (3. rather than by an integral (3. i ∂t cm (t) = − Em cm (t) . being reflected at the right wall. t|x0 . The last frame shows the wave front reaching again the right wall and the onset of new interference.5: Particle in One-Dimensional Box Multiplying both sides by φm (a. The particle moves then to the left. t|x0 . One can recognize that the particle moves first to the left and that near the right wall of the box interference effects develop. The respective initial condition is φ(x. x) exp n=1 i − En (t − t0 ) φn (a. the elements of which satisfy the appropriate boundary conditions.78).112) determine now ψ(x. Figure 3. t0 ) = δ(x − x0 ) as can be readily verified using (3. t) for any initial condition ψ(x. We have identified then with (3. The interference pattern begins to ‘smear out’ first. Note that the propagator has been evaluated in terms of elements of a particular function space F1 . t0 ) itself is a solution of the time-dependent Schr¨dinger o equation (3.112) Equations (3.115) This initial state corresponds to the particle being localized initially near x0 = 0 with a velocity v0 = 15 /m a in the direction of the positive x-axis..114) the representation of the propagator for a particle in a box with infinite walls. t0 ) is also referred to as a Greens function.111) The solutions of these equations which satisfy (3.

48 m a h2 |Ψ|2 2 t = 0.96 m a h2   |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.44 m a h2 |Ψ|2 2 t = 1.00 m a h2 |Ψ|2 2 t = 0.68 Schr¨dinger Equation o |Ψ|2 2 t = 0.2: Stroboscopic views of the probability distribution |ψ(x. .92 m a h2 |Ψ|2 2 t = 2.20 m a h2 1 πσ 1 πσ -a a -a a |Ψ|2 2 t = 1.24 m a h2 1 πσ k = 15 a 1 πσ -a a -a a |Ψ|2 2 t = 0.68 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.

We seek then solutions of the time-dependent Schr¨dinger equation o ˆ i ∂t ψ(x. φ(j) (±aj ) = 0 . x3 ) (3.97) with respect to which the eigenfunctions are orthonormal. x3 = ±a3 } . hence. (3.120) ˆ Since the Hamiltonian H is a sum of operators O(xj ) each dependent only on a single variable. This property allowed us to evaluate the propagator (3. 2. x3 ) . j = 1. ) = ∀( . x2 . We have defined in the space F1 a scalar product (3.114) in terms of which the solutions for all initial conditions can be expressed. The description employed a space of functions F1 defined in (3. t) . x3 ) is an element of the function space F3 defined in (3. x2 . i. . x2 . x2 = ±a2 } ∪ {(x1 .117) 2m 2 2 2 ∂1 + ∂2 + ∂3 (3.e. (3. x3 ) = j=1 φ(j) (xj ) (3.6: 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 5.119) where φE (x1 . t) = H ψ(x1 . ( . can be counted.118) which are stationary states. x2 . ˆ H = − 2 (3..6 Particle in Three-Dimensional Box We consider now a particle moving in a three-dimensional rectangular box with side lengths 2a1 . x2 . a2 ] ⊗ [−a3 . x2 . )T ∈ ∂ } .116) where Ω is the interior of the box and ∂Ω its surface Ω ∂Ω = [−a1 . x3 ) = E φE (x1 . x3 . 3.3. x2 . . x3 –axes with the edges of the box yields spatial boundary conditions which are obeyed by the elements of the function space F3 = {f : Ω → R. A complete basis of F1 is given by the infinite set B1 (3. a3 ] ⊂ R = {(x1 .78).116) and obeys the partial differential equation ˆ H φE (x1 . f continuous. x2 . x3 . t) = exp − E t φE (x1 . x3 )T ∈ Ω. 3 (3. i. x2 . ˆ = O(x1 ) + O(x2 ) + O(x3 ). Placing the origin at the center and aligning the x1 . a1 ] ⊗ [−a2 . of the space F1 is that the elements of the basis set can be enumerated by integer numbers. x2 .122) .121) where − 2 2m 2 ∂j φ(j) (xj ) = Ej φ(j) (xj ) . x1 = ±a1 } ∪ {(x1 . x3 )T ∈ Ω. x2 . The corresponding solutions have the form i ψ(x1 . x3 )T ∈ Ω. one can express H 3 φ(x1 . An important property of this basis and. 2a3 .95).. 2a2 .e.

} ∞ (3.n3 ) = 2π2 = φn1 (a1 .80) shows that the solutions of (3.126) Symmetries The three-dimensional box confining a particle allows three symmetry operations that leave the box unchanged. n2 . .n2 .e. i.120) can be written φ(n1 .. n 2 . hence. (3. r0 ) . . x2 ) φn3 (a3 .70 Schr¨dinger Equation o and E1 + E2 + E3 = E. If two or all three orthogonal sides of the box have the same length further symmetry operations leave the system unaltered. x3 –axes. x1 . a3 are identical. x1 . n1 .n3 ) (a1 .n2 . (3. a2 . a2 .96) and. a2 . a2 .n3 ) (a1 . a3 . . a3 . x2 . a2 .n2 . n3 = 1. a1 = a2 = a3 = a then rotation around the x1 . 3. x1 ) φn2 (a2 . . n 3 and E(n1 . . . r) exp i − E(n1 n2 n3 ) (t − t0 ) φ(n1 . t|r0 . a3 . t0 ) = n1 . . n3 = 1. 2. For example. 3. x3 ) .123) 8m n2 n2 n2 1 2 3 + 2 + 2 a2 a2 a3 1 . .n2 .125) is a complete orthonormal basis of F3 and that the propagator for the three-dimensional box is φ(r. n1 . the solutions of (3.124) The same considerations as in the one-dimensional case allow one to show that B3 = {φ(n1 . .n2 . Actually.n2 . 3.122) with (3. 2. This additional symmetry is also reflected by degeneracies in the energy levels.n3 ) (a1 . if all three lengths a1 . Comparing (3. (3. x2 . x2 .122) are given by (3. Such degeneracies are always a signature of an underlying symmetry.n3 =1 φ(n1 . x2 . The energies and corresponding degeneracies of the particle in the three-dimensional box with all side lengths equal to 2a are given in the following Table: n1 1 1 1 1 2 1 2 1 2 3 1 n2 1 1 2 1 2 2 2 1 3 3 1 n3 1 2 2 3 2 3 3 4 3 3 5 E/[ 2 π 2 /8ma2 ] 3 6 9 11 12 14 17 18 22 27 27 degeneracy single three-fold three-fold three-fold single six-fold three-fold three-fold three-fold single three-fold One can readily verify that the symmetry of the box leads to three-fold and six-fold degeneracies. x3 ) = 1. a3 . n2 . This symmetry has been exploited in deriving the stationary states. in the present case .n3 ) (a1 . 2. x3 –axes by π/2 also leaves the system unaltered. x3 ) n1 . namely rotation by π around the x1 .

due to the hermitian character of ˆ H. to construct n orthogonal stationary states6 .. n2 . We consider the degeneracy of the energy E122 which is due to the identity E122 = E212 = E221 . One particular aspect of the degeneracies illustrated in the Table above is worth mentioning. n3 ) = (3. r) + γ φ(2.3. a. a. a. However. φ(1. r) + β φ(2. However. r). for (n1 . in case of degenerate states one cannot necessarily expect that stationary states are orthogonal. however.2) (a.1) (a.1. a. the identity 32 + 32 + 32 = 12 + 12 + 52 . a.2.g. in case of an n–fold degeneracy it is always possible.2. 3. The origin of this degeneracy is. 5) as shown in the Table above. e. 1. Hence. a.6: Particle in Three-Dimensional Box 71 ‘accidental’ degeneracies also occur. r) (3. a. for example.2) (a.127) ˜ ˆ ˜ obeys the stationary Schr¨dinger equation H φ(r) = E122 φ(r).2. 6 This is a result of linear algebra which the reader may find in a respective textbook. 3) (1. .2) (a. a. this linear combination o is not necessarily orthogonal to other degeneraste states. Any linear combination of wave functions ˜ φ(r) = α φ(1.

72 Schr¨dinger Equation o .

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

is the commutation property 1 1 (4.4) be satisfied. beside the representation (4.3) Due to the nature of the harmonic potential which does not allow a particle with finite energy to move to arbitrary large distances.8) i which holds for any position and momentum operator. In the framework of the Schr¨dinger equation the stationary o states are solutions of (4. 4. the commutator has a simple form. all stationary states of the harmonic oscillator must be bound states and. It is important to keep in mind this restriction of the space.6) the first being a differential operator and the second a multiplicative operator. ∈ C∞ . lim ( ) = } (4. This property states that p and x obey an ˆ ˆ algebra in which the two do not commute. only for a discrete set of E values can the boundary conditions (4.1 Creation and Annihilation Operators The Hamiltonian operator (4. If this restriction would not apply and all functions f : R → R would be admitted. We will point out explicitly where assumptions are made which built on this restriction. the spectrum of ˆ H in (4. the natural boundary conditions apply x→±∞ lim φE (x) = 0 .4) Equation (4. in which the operators used below.5) ˆ H = 1 1 2 p + m ω 2 x2 ˆ ˆ 2m 2 (4. The stationary states of the harmonic oscillator have been considered already in Chapter 2 where the corresponding wave functions (2.7) which is why these operators are of interest to us. t) where ˆ H φE (x) = E φE (x) .235) had been determined. In order [ˆ. therefore. act. In the following algebraic solution of (4. The operators act ˆ on the space of functions N1 defined in (4.1) of the form ψ(x. however.5).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.3) can be solved for any E ∈ R. The cardinal property exploited below. (4.3) would be continuous and the eigenfunctions φE (x) would not be normalizable. however. x] = p ˆ . The Hamiltonian H can be expressed in terms of the operators acting on the space (4. i dx x = x ˆ (4.2) can be expressed in terms of the two operators p = ˆ d . (4. t) = exp(−iEt/ )φE (x.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.5) →±∞ where C∞ denotes the set of functions which together with all of their derivatives are continuous.7) of the Hamiltonian.

8) a ˆ ˆ ˆ ˆ ˆ implies [ˆ− .10) we define a+ = ˆ mω i x − √ ˆ p.14) 1 i mω 2 x + ˆ p2 + ˆ [ˆ.15) together lead to the commutation property (4.14) and (4.14. a+ ] = a− a+ − a+ a− does not necessarily vanish.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 .7) assuming that the factors are easier to handle than the Hamiltonian in yielding spectrum and eigenstates.9) From this follows (4. a+ ] = 1 .18) It is of interest to note that the operators a+ and a− are real differential operators.1: Creation and Annihilation Operators 75 to derive (4. 2mω dx (4. In fact. to factor H/ ω.8) we recall that all operators act on functions in N1 and.12). the action of [ˆ. Being guided by the identity for scalar numbers (b − ic)(b + ic) = b2 − i(cb − bc) + c2 = b2 + c2 (4.8). in fact. x] p ˆ on such functions must be considered.8) yield a− a+ = ˆ ˆ Similarly one can show a+ a− = ˆ ˆ 1 ˆ 1 H − 1 .16) (4.11) a− a+ = ˆ ˆ (4. p ˆ 2 2m ω 2 (4.7) and (4. One obtains [ˆ. ˆ 2 2m ω a− = ˆ mω i x + √ ˆ p. ˆ ˆ . 1 ω 2 (4. hence.4.13) (4. 2mω dx a− = ˆ mω x + ˆ 2 d . 1 i dx i dx i i (4.11) ˆ The reader may note that we have attempted. 1 2 (4. 1 ω 2 (4.15) in a form which will be useful below ˆ H = ˆ H = ω a− a+ − ˆ ˆ ω a+ a− + ˆ ˆ ω 1 1 2 ω 1 . 4. Before we continue we like to write (4.10) holds for a ˆ ˆ [ˆ− . Our next step is an attempt to factorize the Hamiltonian (4. a ˆ 1 (4. Since a+ and a− are ˆ ˆ + and a− since operators and not scalars we cannot simply expect that the identy (4. x] f (x) = p ˆ d d x f (x) − x f (x) = f (x) = 1 f (x) .15) 1 ˆ 1 H + 1 . the commutator property (4. x] . ˆ 2 2m ω (4.12) To prove this commutation property we determine using (4.

20) ˆ In the following we will determine the spectrum of H and its eigenstates. .21) is available. In fact.17. 4.17.19) This property states that the operators a+ and a− are the adjoints of each other. 4. (4.16.22) Similarly one can show using again (4. 4. The derivation will be based solely on the properties (4.19) in the form1 f |ˆ+ g a Ω∞ = g|ˆ− f a Ω∞ .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.24) − m This property states that the operators in the function space N1 are the hermitian conjugate of each other. 4. one obtains using (4. This property of operators is investigated more systematically in Section 5. Using (??) we like to state (4.17. a+ and a− as Ladder Operators ˆ ˆ The operators a+ and a− allow one to generate all stationary states of a harmonic oscillator once ˆ ˆ one such state φE (x) ˆ H φE (x) = E φE (x) (4. The property ˆ ˆ follows directly from (??). 4. ˆ The results (4.19).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) . g ∈ N1 +∞ +∞ dx f (x) a+ g(x) = −∞ −∞ dx g(x) a− f (x) . 4.23) Together.22.16. (4. ˆ (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.16. 4.12.21) 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. it holds that for a stationary state φE (x) of energy E defined through (4. 4. ˆ (4.21) ˆˆ H a+ φE (x) ω 1 ) a+ φE (x) 1 ˆ 2 ω = a+ ( ω a− a+ − ˆ ˆ ˆ 1 + ω 1 ) φE (x) 1 1 2 ˆ = a+ ( H + ω ) φE (x) ˆ = ( ω a+ a− + ˆ ˆ = (E + ω ) a+ φE (x) .

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 (4.25) needs to be normalizable in order to represent a bound state of the harmonic oscillator. i. (ii) the construction appears to yield energy eigenvalues without lower bounds when.31) (4.25) can. φ0 (x) should be an element of the function space N1 defined in (4. For f (x) = φE (x) and g(x) = φE (x) in (??) follows (Note that according to (??) the eigenvalue E is real. one expects that E = 0 should be a lower bound.25) on one side would lead to termination of the sequence E0 + m. in fact.) ˆ ˆ 0 = φE |HφE Ω∞ − φE |HφE Ω∞ = ( E − E ) φE |φE Ω∞ .26) Of course. on ˆ the other side such a state is itself an eigenstate of H as can be shown using (4.. two difficulties: (i) one needs to know at least one stationary state to start the construction.28) 4.33) . (4. however. be found. For obvious reasons one refers to a+ and a− as ladder ˆ ˆ operators. (4. Another name is creation (ˆ+ ) and annihilation (ˆ− ) operators since these operators a a create and annihilate vibrational quanta ω. The property (??) has an important consequence for the stationary states φE (x).17) ˆ H φ0 (x) = ( ω a+ a− + ˆ ˆ 1 ω 1 ) φ0 (x) = 1 ω φ0 (x) .29) where y = mω x. E .27) Since E = E one can conclude φE |φE Ω∞ = 0. There arise.e. m ∈ Z when m is decreased. 4.5).6. +∞[ one can write (4.4. (4. Let φE (x) and φE (x) be two normalized stationary states corresponding to two different energies E.11) one can rewrite (4. a state φ0 (x) which obeys the property a− φ0 (x) = 0 ˆ (4.30) Assuming that φ0 (y) does not vanish anywhere in its domain ] − ∞. It turns out that both difficulties can be resolved together.25) φ0 (y) = 0 (4. E = E . Using (4. (4.29) d 1 d φ0 (y) = = lnφ0 (y) = − y . the solution φ0 (x) of (4. in fact.2 Ground State of the Harmonic Oscillator d + y dy A suitable solution of (4. In fact.32) for some constant c0 or (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.

39).34) dx |φ0 (x)|2 = |c0 |2 −∞ −∞ dx exp − mωx2 2 = |c0 |2 π .38) These states correspond to the energy eigenvalues (4. which for a narrowly localized state yields a large positive value. 2. a state confined to the minimum corresponding to the classical state of lowest energy.35) The appropriate ground state is φ0 (x) = mω π 1 4 exp − mωx2 2 . The reason is that in the Hamiltonian (4. . . 2. we construct the states for n = 1. choose the normalization constants cn and the functions φn (x) real as well. .25) admits only one solution there is only one set of states with energy E + m ω. The conventional normalization condition φ0 |φ0 reads +∞ +∞ Ω∞ = 1 (4.78 Linear Harmonic Oscillator This solution is obviously normalizable.24) can then be written as follows En = ω(n + 1 ).. 4. .. .39) We notice that the ground state wave function φ0 (x) as well as the operators (ˆ+ )n are real. 1. m ∈ Z which properly terminate at some minimum value E + m0 ω ≥ 0. 2.37) above the ground state energy. 1. . the potential energy contribution which for a state δ(x). We recall that according to (4. . . the so-called excited states. (4. therefore. (4. mω (4.37) It is most remarkable that the energy 1 ω of the ground state is larger than the lowest classically 2 allowed energy E = 0.36) n times..e. i. (4. n = 0. The set of allowed energies of the oscillator according to (4. (4. 2.36) Since the first order differential equation (4. i. . 2 This state of lowest energy is called the ground state of the oscillator. Such states need to be suitably normalized for which purpose we introduce a normalization constant cn φn (x) = cn a+ ˆ n φ0 (x) . 2 n = 0. .26) the energy value associated with this state is 1 ω.e. 2 n = 0.e. The ground state (4. and the kinetic energy contribution. For this purpose we apply the operator a+ to the ˆ ground state (4. We a can. would yield a vanishing contribution. it holds ˆ H φn (x) = ω(n + 1 ) φn (x) . .3 Excited States of the Harmonic Oscillator Having obtained a suitable stationary state with lowest energy.. . We will consider this point more systematically in Section 5. .36) assumes a functional form such that both terms together assume a minimum value. i. 1.2) there are two competing contributions to the energy. we can now construct the stationary states corresponding to energies (4.

18)] and since the ˆ φn (x) are real functions.40). (4.24) holds in analogy to (4. i.20) 1 = φn−1 |φn−1 Equations (4. 2. . n = 0..43) (4.28) and the construction (4. Ω∞ 2 = βn a− φn |ˆ− φn ˆ a Ω∞ 2 = βn φn |ˆ+ a− φn a ˆ Ω∞ . (4. To determine these constants we adopt a recursion scheme.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. (4. (4. .40) φn−1 (x) = βn a− φn (x) ˆ (4. (4. For n = 0 holds c0 = 1. they obey φn |φn = δn n .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.e. βn must be real as well.46) According to (4.20) yields 2 αn φn |ˆ− a+ φn a ˆ Ω∞ = 1. (4.40–4. n. ˆ (4.48) (4. 4. ˆ Repeated application of this relationship yields φn−s (x) = (n − s)! − a ˆ n! s 2 1 = βn n φn |φn Ω∞ 2 = βn n . . n = 0.44) 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) . √ a+ φn (x) = n + 1 φn+1 (x) .42) ˆ Using (4.47).3: Excited States 79 We need to determine now the normalization constants cn . .45) We like to note that these functions according to (4.15 . .47) for some suitable constants βn . To determine βn we note using (4.49) (4. 2. according to (4.41) Employing the adjoint property (4.40) = 1.45) form an orthonormal set. 1. We consider then the situation that we have obtained a properly normalized state φn (x). . .39) yield √ From this follows βn = 1/ n and. √ a− φn (x) = n φn−1 (x) . 1.51) . according to (4.50) φn (x) . Since a− is a real differential operator [see (4.

by dx = dy mω 1 4 .7 and given. however.46) yields a set of normalized states.56) can be expressed in terms of Hermite polynomials Hn (y) introduced in Sect.45).35) by a constant.30) and employing the normalization +∞ −∞ dy φ2 (y) = 1 n (4..52).57) which includes the factor 1/ n! according to Eqs.233) derived in Sect.7. (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 . simplifying the calculation. (4. by the Rodrigues formula (2. We will demonstrate below the identity Hn (y) = e y2 2 y − d dy n e− y2 2 (4.53) We will later re-introduce this factor to account for the proper normalization (4.58) This result agrees with the expression (2. (4. For this purpose we start from expression (4.40–4. .e.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) .39). namely the square root of the Jacobian dx/dy. n–independent factor.80 Evaluating the Stationary States Linear Harmonic Oscillator We want to derive now an analytical expression for the stationary state wave functions φn (x) defined through (4. for example. (4.200). i.28) of the ground state and the definition (4. It should be noted that √ the normalization (4. 2. In terms of y the ground state wave function is φ0 (y) = π − 4 e− and a+ is ˆ 1 a+ = √ ˆ 2 y − d dy .52) This normalization of the wave functions differs from that postulated in (4. by introducing the variable y defined in (4. (4.55) 1 y2 2 (4. 2.35) rather than (4.56) Relationship to Hermite Polynomials We want to demonstrate now that the expression (4.

60) and.e. t0 ) = n=0 dn φn (y0 ) . Our derivation will follow closely the procedure adopted for the case of a ‘particle in a box’ [see Eqs.114)]. dy dy (4.. 1. Starting point of our derivation is the assumption that the initial condition can be expanded in terms of the eigenstates φn (y) (4.2) for an arbitrary initial wave function ψ(x0 . For the sake of notational simplicity we employ initially the coordinate y as defined in (4. and showing then that the property also holds for n + 1 in case it holds for n. t0 ). g(y) = (−1)n+1 e y2 2 dn+1 −y2 e = dy n+1 y − d dy n+1 e− y2 2 .4: Propagator 81 To verify the realtionship between the definition (4.64) which implies that (4.61) and.57) of the hermite polynomials and the definition given by (2.59) dn −y2 e dy n (4. (4.63) one obtains g(y) = y − d dy f (y) (4. dy n (4.65) .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 .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. We prove (4.4 Propagator for the Harmonic Oscillator We consider now the solution of the time-dependent Schr¨dinger equation of the harmonic oscillator o (4.200). hence. (3. the Rodrigues formula (2.59) hold. 4.1. (4.59) holds for n = 0.39) ∞ ψ(y0 .106–3. i.30) and return later to the coordinate x.59) by induction noting first that (4. 4.4. therefore.61) One can factor g(y) and employ (4.200) we need to verify y − which implies Hn (y) = (−1)n ey 2 d dy n e− y2 2 = (−1)n e y2 2 dn −y2 e .

We want to demonstrate now that this propagator is identical to the propagator (2. this is an excellent textbook from which we have borrowed heavily in this Section.225) which allows one to derive a generating function for products of Hermite polynomials which can be applied to the r. t1 |y0 . For this purpose we start from the integral representation (2.Lebedev (Prentice Hall. y0 . 2n n! π 2 2 (4. t0 ) ψ(y0 .46) the expansion coefficients dn are +∞ dn = −∞ dy0 ψ(y0 .112) and (4.15. t0 ) = n=0 φn (y) φn (y0 ) tn+ 2 . y0 .70) is the propagator of the linear harmonic oscillator. We consider for this purpose the following expression for |t| < 1 ∞ w(y. Employing orthogonality condition (4.68) Altogether one can express then the solution +∞ ψ(y.5). 1965) Sect.. pp. .72) × exp( −u − v + 2iyu + 2iy0 v ) . sec:harm.66) One can extend expansion (4. a supposition which is not proven here2 . 68. N. t1 ) = −∞ dy0 φ(y. of (4.. (4. t0 ) which is an element of N1 defined in (4.147) we employ the technique of generating functions as in Sect. 4.70) and (2. In order to prove the equivalence of (4. Englewood Cliffs. t) (4. 2. t|y0 .82 Linear Harmonic Oscillator Such expansion is possible for any f (y0 ) = ψ(y0 .67) for which according to (3.J.39) holds cn (t1 ) = exp −i ω ( n + 1 ) ( t1 − t0 ) 2 .65) to times t ≥ t0 through insertion of time-dependent coefficients cn (t1 ) ∞ ψ(y. t1 ) = n=0 dn φn (y) cn (t) (4.N. 2 A demonstration of this property can be found in Special Functions and their Applications by N. Inc.69) where ∞ φ(y. t) = n=0 Hn (y)e−y /2 Hn (y0 )e−y0 /2 n √ t .108–3. t) = π −3/2 exp 2 2 y 2 y0 + 2 2 2 +∞ +∞ ∞ du −∞ −∞ dv n=0 1 ( −2tuv )n × n! (4. (4.h.70). t0 ) φn (y0 ) .s.71) Applying (??) to express Hn (y) and Hn (y0 ) yields exp(−2tuv) w(y.147) for the harmonic iscillator determined in Sect. 1 t = e−iω(t1 −t0 ) (4.7.

t|x0 . a Re a2 > 0 (4. t0 ) = exp mω 2iπ sinω(t1 − t0 ) 1 2 × . by a factor 4 mω/ for both φn (y) and φn (y0 ). t0 ) − i y y0 2 F (t1 .147) derived by means of the path integral description. y0 ..70) is. t0 ) exp i 2 1 2 (y + y0 ) G(t1 .74) applied once results in w(y. i. (4. (4.4. t) √ 1 π(1−t2 ) 2 exp − 1 (y 2 + y0 ) 1 + t2 + 2 y y0 2 1−t Hn (y)e−y /2 Hn (y0 )e−y0 /2 n ∞ √ t n=0 2n n! π 2 2 2 t 1 − t2 = .. t) = π −1 exp − 2 y 2 y0 + 2 2 +∞ dv exp(−(1 − t2 ) v 2 − 2i (yt − y0 ) v) . t) = π −3/2 exp +∞ 83 y2 y2 + 0 2 2 +∞ dv exp( −v 2 + 2iy0 v ) × −∞ × −∞ du exp( −u2 − 2u(tv − iy) ) .e.77) The sum in (4.80) −2iω(t1 −t0 ) sinω(t1 − t0 ) 1−e We can finally express the propagator (4.30) to replace y and y0 and that we multiply the propagator by mω/ .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 . t1 |y0 . The resulting propagator is G(t1 . t0 ) = i φ(x. (4. −∞ (4.76) One can express this in terms of the stationary states (4. √ πexp( t2 v 2 − 2iytv − y 2 ) (4.4: Propagator Carrying out the sum on the r.79) 1 + e−2iω(t1 −t0 ) cosω(t1 − t0 ) = . indeed. . one obtains w(y. t0 ) = and 1 2 i π F (t1 .75) Applying (4.81) imω 2 sinω(t1 − t0 ) (x2 + x2 ) cosω(t1 − t0 ) − 2 x x0 0 This result agrees with the propagator (2.77). it holds φ(y. t0 ) (4. i. t0 ) = where F (t1 . identical to the generating function (4.78) in terms of the coordinates x and x0 . y0 .74) a second time yields finally together with the definition (4.78) 1 − e−2iω(t1 −t0 ) = sinω(t1 − t0 ) 2 i e−iω(t1 −t0 ) (4. (4.e. This requires that we employ (4.73) The incomplete Gaussian integral +∞ −∞ dx e−a 2 x2 √ − 2bx = π b2 /a2 e . y0 .71) of w(y.s.h.

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

92) f (y) = n=0 dn φn (y) = n=0 (ˆ+ )n a dn √ φ0 (y) n! (4. ˆ a a a ˆ In particular.e.88) we need to show actually [ˆ− .88) (4.90) To prove (4. f (ˆ+ )] = f (1) (ˆ+ ) . a− f (ˆ+ ) φ0 (y) = f (1) (ˆ+ ) φ0 (y) ˆ a a which follows from a− φo (y) = 0. f1 (ˆ+ )] = [ˆ− . a+ ] + [ˆ− . a+ ] = 1 = f1 (ˆ+ ) . For fn+1 follows then [ˆ− .90) we show that (4. property (4.83) and can be proven for all powers fn by induction and then inferred for all proper functions f (ˆ± ). 4.93) one can reduce all operators acting on some proper state function by operators acting on the state φ0 (y).96) . The following important property holds a a− f (ˆ+ ) = f (1) (ˆ+ ) + f (ˆ+ ) a− . (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) (4.87) (4. fn+1 (ˆ+ )] a a = [ˆ− .88) is a fundamental one and will be used in the following.95) (4.94) This identity follows from (4.91) (4. a a (4. f (ˆ+ )] + f (ˆ+ ) a− ˆ a a a a ˆ which implies that in order to prove (4. ˆ a We proceed by induction noticing first that (4. As long as the operators act on φ0 (y) one can state then that a− behaves like a differential operator with respect to functions f (ˆ+ ). It holds a a euˆ f (ˆ+ ) φ0 (y) = f (ˆ+ + u) φ0 (y) .4.90) holds for f0 and for f1 .87. We note ˆ a− f (ˆ+ ) = [ˆ− . The convergence of the Taylor expansion ascertains then that (4.88) is (ˆ− )n f (ˆ+ ) φo (y) = f (n) (ˆ+ ) φ0 (y) .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. we will only assume operator functions acting on φ0 (y).89) (4.90) holds for fn . a a a ˆ 1 a Let us assume that (4.90) holds for f (ˆ+ ). An example is the so-called shift operator exp(uˆ− ). (ˆ+ )n a+ ] = (ˆ+ )n [ˆ− . Hence.. a a − (4. The first case is trivial.5: Working with Creation and Annihilation Operators and similarly for f (ˆ− ). i. Note that ˆ a an immediate consequence of (4. a a a 85 (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 . the second case follows from [ˆ− .

102) Similarly.98) for u = −α∗ and v = α yields e−α This together with (4.106) yields a eαˆ + ∗ a− ˆ a a eαˆ φ0 (y) = e−αα eαˆ φ0 (y) .101) 1.97) An example in which (4. Below we will use the operator identity which follows for the choice of v = α in (4. + (4.101) ˆ ˆ ˆ C(u) obviously obeys C(0) = 1 This results in D(0) = 1 and.108) .103) (4.90) and zz ∗ = 1 we can write this d du e uzˆ+ a e uz ∗ a− ˆ a − z ∗ a− .96) is applied is a a a euˆ evˆ φ0 (y) = ev (ˆ − + + + u) a φ0 (y) = euv evˆ φ0 (y) . hence. We assume in the following a ˆ derivation. (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. + ∗ a− ˆ a − u euzˆ euz + ∗ a− ˆ = (4. + (4.99) ˆ C(u) = zˆ a zˆ+ + z ∗ a− a ˆ + +z a ˆ ∗ − a euzˆ euz ˆ − u C(u) .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. 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) .105) a eαˆ + − α∗ a− ˆ φ0 (y) = e αα∗ 2 e−α ∗ a− ˆ a eαˆ φ0 (y) .104) (4. that the operators considered act on φ0 (y). a a ν! (4. + (4.98) The related operators exp[vˆ+ ± v ∗ a− ] play also an important role. z ∈ C. the solution of (4. one obtains a evˆ + − v ∗ a− ˆ a φ0 (y) = e− 2 vv evˆ e− v 1 ∗ + ∗ a− ˆ φ0 (y) (4. 1 + + uz ∗ a− ). ∗ = 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 (4.105) 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) .106) Applying (4. euzˆ ˆ euz ∗ a− ˆ (4. without explicitly stating this.107) − α∗ a− ˆ φ0 (y) = e− αα∗ 2 a 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) . + ∗ + (4.

s. For this purpose we consider a a a a euˆ φ0 |evˆ φ0 = φ0 |euˆ evˆ φ0 + + − + (4.111) the orthonormality properties of the wave functions φn (y).98) and again (4.h. in the ladder operator calculus.116) = δµ ν . Using (4.115) a where the latter step follows after expansion of evˆ and using a− φ0 (y) = 0. and on the r. We want to derive now the values Hn (0)..113) Comparision of all terms on the l.s.h. Similarly.58) the Hermite polynomials Hn (y) by the eigenstates φn (y) √ ∞ 1 ( 2t)n y2 /2 2yt − t2 √ e φn (y) (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). and ˆ l. 4. (4.h.115) and using (4.95) one obtains a a a a euˆ φ0 |evˆ φ0 = φ0 |euv evˆ φ0 = euv evˆ φ0 |φ0 = euv − + + + − (4.111) This expression.4.84. of (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. .s. + (4. Setting y = 0 in (4.s. the expected from which follows by comparision of each term on both sides φµ |φν orthonormality property.112) Expanding both sides of this equation and using (4. (??).111) yields π − 4 exp − 1 u2 2 a = euˆ φ0 (0) .109) e = π4 n! n=0 √ Using (4. provide the same values for Hn (0) as provided in Eq.5: Working with Creation and Annihilation Operators Generating Function in Terms of a+ ˆ 87 We want to demonstrate now that generating functions are as useful a tool in the calculus of the ladder operators as they are in the calculus of differential operators.µ=0 uµ v ν √ φµ |φν µ!ν! ∞ = µ=0 uµ v µ µ! (4. Expanding r. We start from the generating function (??) and replace according to (4. is the equivalent of the generating function (??).84) yields ∞ ν.114) where we have employed (4.e.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! (4.110) √ y2 u2 + 2uy − 2 2 a = euˆ φ0 (y) . i.h. we can reproduce by means of the generating function (4.95).

e.235).. (−i)n ˜ φn (p) = √ 2n n! m˜ ω π 1 4 exp − m˜ p2 ω 2 Hn ( m˜ ω p) .124) ). n = 0. n = 0..120) Multiplying this equation by 1/m2 ω 2 yields − where ˜ E = E / m2 ω 2 . the eigenvalues are identical to those determined in the position representation. The eigenvalues.7) is given by (4.3). En = ω (n + 1 ) 2 d2 1 + m˜ 2 p2 ω 2 2m dp 2 2 ˜ ˜˜ φE (p) = E φE (p) (4.2). For this purpose we employ the Schr¨dinger equation in the momentum o representation.125) .37). dp (4. 1.122) For a solution of the Schr¨dinger equation in the momentum representation.e.121) ω (n + ˜ 1 ) 2 . 2.6 Momentum Representation for the Harmonic Oscillator The description of the harmonic oscillator provided so far allows one to determine the probability density P (x) of finding an oscillator at position x.235). .. is given by the ˜ function ψ(p).122).123) (4. the time-independent Schr¨dinger equation for the oscillator can be written o ˆ H = − mω 2 2 2 x = i ˆ d . 2. using (4.117) In this section we want to provide a representation for the harmonic oscillator which is most natural ˜ if one wishes to determine for a stationary state of the system the probability density P (p) of finding the system at momentum p. the momentum and position operators are as stated in (4. The solutions can be stated readily exploiting the earlier results. 4. . .88 Linear Harmonic Oscillator 4. . The momentum representation eigenfunctions are.7) is 1 2 m 2 ω 2 d2 p − . (4. ω = 1 / m2 ω . are ˜ En = or. of (4.121). according to (4. For example. 1. (4. (4. the probability density is P (x) = |φn (x)|2 .6) and the Hamiltonian (4. the momentum and position operators are p = p.2. (4. (4. using (2.118) d2 1 2 + p dp2 2m ˜ ˜ φE (p) = E φE (p) . In the momentum representation the wave function is a function of p.e. is given by a function ψ(x). i.119) 2m 2 dp2 Accordingly. i. ˆ and the Hamiltonian (4. for an oscillator in a stationary state φn (x) as given by (2. i. . . As to be expected. 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. ˜ (4. In the position representation the wave function is a function of x.

134) Exercise 4. 1 ˜ φn (k) = √ 2π +∞ (4.. φn (k) = We want to finally apply the relationship 1 ˜ φn (p) = √ 2π +∞ 2 (−i)n −k √ e 2 Hn (k) .235)] absorbs the replacement dx → dy.126) Normalized eigenfunctions are. From (4. α ∈ R. i. instead of (4.1: Demonstrate the validity of (4.133) where we note that a change of the normalization of φn (y) [c.126).4.f.. through direct integration.e. m ω (4. This implies the property of Hermite polynomials (−i)n e− k2 2 +∞ Hn (k) = −∞ dy exp(−iky) e− y2 2 Hn (y) (4. ˜ We can now state the probability density Pn (p) for an oscillator with energy En to assume momentum p. Defining k = m˜ / p or ω 1 k = p (4.30) and (4. this allowed us to introduce in (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 (4.30) and (2.129) follows ky = px/ and.127) exp − p2 m ω 2 Hn ( 1 p) . 4.132) dy exp(−iky) φn (y) −∞ (4.e.58). i. . the Jacobian. the correctness of the phase factors (−i)n .6: Momentum Representation or with (4. of course.131) in the present case employing dimensionless units.6.125.126) in terms of dimensionless units following the procedure adopted for the position representation where we employed the substitution (4.134).122) (−i)n ˜ φn (p) = √ 2n n! 1 πm ω 1 4 89 exp − p2 2m ω Hn ( 1 p) .129) m ω one obtains.130) dx exp(−ipx/ ) φn (x) −∞ (4.30) to obtain (4.126) a phase factor (−i)n . m ω (4. Proceed as follows. defined only up to an arbitrary phase factor eiα . 2n n! π (4.128) We may also express the eigenstates (4. hence. (4.

138). 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 . The motion of the probability distribution of such state is presented in Fig.— III. Inc.Grynberg (John Wiley & Sons. φ(α) | φ(α) = . coherent states. New York. for example. Such states are referred to as quasi-classical states. 2. 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.135) g(k.1 showing in its top row the attributes of such state just discussed. We first construct the solution of (4. y) −∞ (4. (4.136) (b) Employ the generating function (2. ˆ α ∈ R.196) of Hermite polynomials and equate coefficients of equal powers of z to prove (4. The answer is ‘yes’. Inserting this expansion into √ (4.137) 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. The quasi-classical states can be obtained by generalizing the construction of the ground state of the oscillator.194. and G.134).7 Quasi-Classical States of the Harmonic Oscillator A classical harmonic oscillator.. pp. a pendulum swinging at small amplitudes.4. n n=0 d(α) n 2 = 1 (4. J. except that α changes periodically in time. 4.Chen-Tannoudji.90 (a) Prove first the property 1 f (k.140) An excellent textbook is Photons and Atoms: Introduction to Quantum Electrodynamics by C. carries out a periodic motion described by y(t) = Re y0 eiωt (4. z) = exp 2yz − z 2 − y 2 /2 (4.138) We will show that a harmonic oscillator prepared in such a state will remain forever in such a state. 192 3 . z) = exp −2ikz + z 2 − k 2 /2 . We want to construct and characterize such states.139) where we have added the condition that the states be normalized. +∞ Linear Harmonic Oscillator dy exp(−iky) g(z. For this purpose we assume such state exists and expand ∞ ∞ φ(α) (y) = n=0 d(α) φn (y) . or Glauber states of the harmonic oscillator and they play.C.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. The question arises if similar states exist also for the quantum oscillator. 4. z) = √ 2π where f (k.Dupont-Roc. namely through the eigenvalue problem for normalized states a− φ(α) (y) = α φ(α) (y) . for example. 1989) which discusses quasi-classical states of the free electromagnetic field in Sect.

(4. However.4.. This interpretation justifies the choice of α ∈ R in (4. we will see below that α ∈ C are also admissible and will provide a corresponding interpretation.149) − α∗ a− ˆ (4.141) Because of the linear independence of the states φn (y) all coefficients multiplying φn (y) must vanish and on obtains the recursion relationship (α) dn+1 = √ α d(α) .e.142) The solution is αn d(α) = √ c n n! ∞ ∞ (4.146) √ 1 − (y − 2 α)2 2 (4.138).148) The identity (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 (4.143) for a constant c which is determined through the normalization condition [see (4.7: Quasi-Classical States or ∞ n=0 91 √ n + 1 dn+1 − α d(α) n (α) φn (y) = 0 . hence. n n+1 (4. Comparing (4. n! (4. i.145) Using the generating function in the form (4. n! n (4.145) using (4.139)] 1 = n=0 d(α) n 2 = |c|2 n=0 α2 2 = |c|2 eα . the quasi-classical states are φ (α) (y) = exp α2 − 2 ∞ n=0 αn √ φn (y) .149) with (4.147) √ which identifies this state as a ground state of the oscillator displaced by 2α.84) φ(α) (y) = exp − α2 2 ∞ n=0 αα∗ (αˆ+ ) a a+ φ0 (y) = e− 2 eαˆ φ0 (y) . One can also write (4.150) for α presently real.109). n! (4. . 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) .144) Normalization requires c = exp(−α2 /2) and.147) allows one to interprete T (α) as an operator √ which shifts the ground state by 2α.

t1 ) = exp or φ (α) − α2 2 ∞ exp n=0 −i ω ( n + ∞ n=0 1 ) ( t1 − t0 ) 2 αn √ φn (y) n! (4.145) as the initial state φ(α) (y.160) .s. Following the procedure adopted in determining the propagator of the harmonic oscillator [see (4.145) is replaced by a complex number αu which at times t1 = t0 + 2πn/ω. becomes real. is the adjoint of the operator function on the l.152) T + (α) f |T + (α) g = f |T (α) T + (α) g = f |g . it holds T + (α) φ0 |T + (α) φ0 = f |T (α) T + (α) g = φ0 |φ0 . .109) serves again to simplify this sum   √  y2 1 1 α2 (u2 + 1)   φ(α) (y.152) Since the operator function on the r.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. n = 0. t1 ) = π − 4 u 2 exp  − + 2yαu −  2  2 E (4. a a ˆ ˆ (4.155) (4.150) for complex α obeys the unitary property T + (α) T (α) = 1 1 (4. 1.156) (4.151) which follows from a generalization of (4. (4. .67.95) to linear combinations αˆ+ + βˆ− and reads then a a f (αˆ+ + βˆ− ) φ|ψ = φ|f (α∗ a− + β ∗ a+ ) ψ .h. Relationship (4. In particular..e.92 Linear Harmonic Oscillator The shift operator as defined in (4. (4.158) Comparision of this expression with (4. α ∈ C (4. i.h.154) . n! u = e−iω(t1 − t0 ) .68)] one can write the time-dependent solution with (4. the shift operator leaves the ground state normalized.s.157) (y. .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. 4. t1 ) = u exp 1 2 α2 − 2 (α u)n √ φn (y) .145) shows that α in (4.153) This is obviously the inverse of T + (α) as defined in (4. one can write in case of T + (α) T + (α) + = T (α) = eα ∗ a− ˆ − αˆ+ a a = e − (αˆ + − α∗ a− ) ˆ .

except that it is not pointlike. being displaced around the center with period 2π/ω and.164) where according to (4.161) and is found to describe a displaced Gaussian. 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.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 ) (4.7: 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 (4.159) is then 1 1 i φ(α) (y. This shows clearly that the Glauber state is a close analogue to the classical oscillator. t1 ) = u 2 T + (α u) φ0 (y) (4. moving solely its center of mass in an oscillator fashion around the minimum of the harmonic potential. this characterization corresponds closely to that of the possible initial states of a classical oscillator.138.138): a real α corresponds to a displacement such that initially the oscillator is at rest. 4. It is of interest to note that the displacement y0 of the Gaussian as well as the momentum k0 and phase φ0 associated with (4. in general. (4. Our interpretation 2 explains also the meaning of a complex α in (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.150) with (4. t1 ) = π − 4 exp − [y − y0 (t1 )]2 + iy k0 (t1 ) − iφ0 (t1 ) − ω(t1 − t0 ) 2 2 . In Fig.158) allows one to write 1 φ(α) (y. obviously. Arbitrary Gaussian Wave Packet Moving in a Harmonic Potential We want to demonstrate now that any initial state described by a Gaussian shows a time dependence very similar to that of the Glauber states in that such state remains Gaussian.165) . a momentum factor and a phase factor.148.163) also through the shift operator. Comparing (4. The time-dependent wave function (4. 4.162) 2 with a periodic change. (4. experiences a change of its width (relative to . 4. One can express (4.163) This solution corresponds to the initial state given by (4.147). a complex α corresponds to a displaced oscillator which has initially a non-vanishing velocity.4.1 (top row) we present the probability distribution of the Glauber state for various instances in time. The diagram illustrates that the wave function retains its Gaussian shape with constant width for all times.

94 E E E E Linear Harmonic Oscillator E 2h ω 2h ω 2h ω 2h ω 2h ω π t= 0ω σ2 = 1 ¥ ¦ £   y ¨ ¡ ¢ ¡   π t = 1/4 ω σ2 = 1 ¥ ¦ £   ¡ y ¨ ¢ ¡   π t = 1/2 ω σ2 = 1 ¥ ¦ £   ¡ y ¨ ¢ ¡   π t = 3/4 ω σ2 = 1 ¥ ¦ £   ¡ y ¨ ¢ ¡   π t= 1 ω σ2 = 1 ¥ ¦ £   ¤ ¤ ¤ ¤ ¤ y ¡ ¢ ¡ ¨ -4 -2 0 2 4 -4 -2 0 2 4 -4 -2 0 2 4 -4 -2 0 2 4 -4 -2 0 2 4 E E E E E 2h ω π t = 0ω σ2 = 1/3 ¥ ¤ § £ ¢ 2h ω π t = 1/4 ω σ2 = 1/3 ¥ § £ 2h ω π t = 1/2 ω σ2 = 1/3 ¥ § £ 2h ω π t = 3/4 ω σ2 = 1/3 ¥ § £ 2h ω π t= 1 ω σ2 = 1/3 ¥ § £ ¤ ¤ ¤ y ¨ y ¨ ¢ y ¨ ¢ y ¨ ¢ ¤ y 0 ¢ -4 -2 0 2 4 -4 -2 0 2 4 -4 -2 0 2 4 -4 -2 0 2 4 -4 -2 2 4 © © © © E E E E E 2h ω π t = 0ω σ2 = 2 ¥ £   2h ω π t = 1/4 ω σ2 = 2 ¥ £ 2h ω π t = 1/2 ω σ2 = 2 ¥ £ 2h ω π t = 3/4 ω σ2 = 2 ¥ £ © 2h ω π t= 1 ω σ2 = 2 ¥ ¤ ¤ ¤ ¤ y ¨ ¡ ¢ ¡   y ¨ ¢ ¡   y ¨ ¢ ¡   y ¨ ¢ ¡   ¤ y ¡ ¢ ¡ ¨   ¡   ¡   ¡ £   ¨ -4 -2 0 2 4 -4 -2 0 2 4 -4 -2 0 2 4 -4 -2 0 2 4 -4 -2 0 2 4 Figure 4. σ|y.1 (second and third row). 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) (4.1: Time dependence of Gaussian wave packets in harmonic oscillator potential. t0 ) = [πσ 2 ]− 4 exp we expand ∞ 1 − (y − ao )2 2σ 2     (4. 4.166) φ(ao . (4. 2σ 2 q = a . To describe such state which satisfies the initial condition φ(ao . t0 ) = n=0 dn φn (y) .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 . that of the Glauber states) with a period π/ω. σ|y. One can recognize the cyclic displacement of the wave packet and the cyclic change of its width with a period of twice that of the oscillator.169) . This behaviour is illustrated in Fig. σ|y. 1 + σ2 (4.

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. Prudnikov. Marichev (Wiley. namely $750 4 . 1 a2 2 1 + σ2 + 1 2 2 y o + 2 y yo This integral can be found in Integrals and Series.171) lim p−1 p n 2 Hn (q p ) = lim (2q)n = an .175) ∞ tn n=0 2n n! Hn (yo ) e−yo /2 Hn (y) e−y 2 2 /2 The generating function (4.170) In order to check the resulting coefficients we consider the limit σ → 1. in agreement with the expected result (4.76) permits us to write this φ(ao . In this limit the coefficients dn should√ become identical to the coefficients (4.177) − 1 ω(t1 − to ) 2 . t1 ) = √ × where yo = q p a = √ . Yu.P. σ→1 p−1 2 (4.172) One can then conclude e−a /4 n lim dn = √ a σ→1 2n n! (4.168) is4 e−aq/2 (p − 1)n/2 dn = √ Hn (q 2n n!σ p(n+1)/2 p ).173) which is. and O. a worthy successor of the famous Gradshteyn. therefore. σ|y.176) 1 √ pσ π a 2 exp − 1 1+σ2 + 1 yo − 1 ω(t1 − to ) 2 2 2 2 × (4. σ→1 O(y n−2 ) for n ≥ 2 1 for n = 0.143) of the Glauber state. 2 by A. 1990). vol.7: Quasi-Classical States The solution of integral (4. Brychkov.4. in fact. p−1 95 (4. For this purpose we note that according to the explicit expresssion (??) for Hn (y) the leading power of the Hermite polynomial is Hn (y) = (2 y)n + and. Following the strategy adopted previously one can express the time-dependent state corresponding to the initial condition (4. this 3 volume integral table is likely the most complete today.157) as follows φ(ao . a ground state shifted by ao = 2α.143). This expansion can be written φ(ao . σ|y. p−1 1 − σ4 t = 1 − σ 2 −iω(t1 −to ) e 1 + σ2 (4.I. New York.174) Hn (q p p−1 ) φn (y) . unfortunately very expensive..e. t1 ) = e−aq/2 √ pσ ∞ n=0 e −iω(n+ 1 )(t1 −to ) √ 2 1 2n n! p−1 p n/2 (4.166) in analogy to (4. 1 (4. i.

96 Linear Harmonic Oscillator .

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

. =1 [R(ϑ)]jk [R(ϑ)]j ak b = j=1 aj bj . .e. (5. Let us define the rotated vector as r = R(ϑ) r . where ϑ denotes the rotation. ϕ)T ) is negative. i.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 .6) (5.1) In Cartesian coordinates this reads 3 xk = j=1 [R(ϑ)]kj xj k = 1.2) Rotations conserve the scalar product between any pair of vectors a and b. We assume the convention that rotations are right-handed. (5. (5. namely around an axis given by the direction of ϑ and by an angle |ϑ|.5) This equation states the key characteristic of rotation matrices.6) follows immediatety for the determinant of R(ϑ) using detA B = detA detB and detAT = detA detR(ϑ) = ±1 . 2. 0. it follows 3 [R(ϑ)]jk [R(ϑ)]j = δk j=1 .e. can be represented by 3 × 3 matrices R(ϑ). (5. i. in case of a prefactor −1 a rotation and an inversion at the origin. ϕ)T ) = ±  sinϕ cosϕ 0  (5.7) implies that a rotation is associated with an inversion. ϑ parametrizes the rotations uniquely as long as |ϑ| < 2π. in Cartesian coordinates r = (x1 . In the following we want to exclude inversions and consider only rotation matrices with detR = 1.e.8) 0 0 1 In case of a prefactor +1. the matrix represents a proper rotation. are linear and. 3 .3) Since this holds for any a and b. i.7) Let us consider briefly an example to illustrate rotational transformations and to interpret the sign of detR(ϑ). a minus sign in Eq. 0.k. x3 )T . therefore.98 Properties of Rotations in R Theory of Angular Momentum and Spin Rotational transformations of vectors r ∈ R . then the fingers of your fist point in the direction of the rotation. In the latter case the determinant of R(ϑ = (0. It holds then j=1 3 3 a ·b = j.(5. A rotation around the x3 -axis by an angle ϕ is described by the matrix   cosϕ − sinϕ 0 R(ϑ = (0. (5. x2 .. they conserve a · b = 3 aj bj . From (5. if the thumb in your right hand fist points in the direction of ϑ. (5.

12) (iv) Since the associative law holds for multiplication of any square matrices this property holds for the elements of SO(3).9) forms a group. Since (R1 )T = (R1 )−1 it follows −1 −1 −1 T (R1 )T R1 = (R1 )−1 R1 = −1 which implies R1 ∈ SO(3). 1 T Similarly. We need to demonstrate that this inverse belongs also to −1 T SO(3). We want to prove now that SO(3) as defined in (5.10) detR3 = det ( R1 R2 ) = detR1 detR2 = 1 . it holds 1. there exists a real 3 × 3–matrix −1 R1 which is the inverse of R1 .9) This set of matrices is closed under matrix multiplication and. R RT = RT R = 1 detR = 1 } 1.13) . (ii) There exists an element e in G with the property ∀a ∈ G → e ◦ a = a ◦ e = a. (iv) For products of three elements holds the associative law a ◦ (b ◦ c) = (a ◦ b) ◦ c. (5. It follows that R3 is also an element of SO(3). a ∈ G → ∃a−1 ∈ G with a ◦ a−1 = a−1 ◦ a = e. We have shown altogether that the elements of SO(3) form a group. It holds T T T T R3 R3 = ( R1 R2 )T R1 R2 = R2 R1 R1 R2 = R2 R2 = 1 .5. hence.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. We would like to point out the obvious property that for all elements R of SO(3) holds R−1 = RT (5. In the following we will assume R1 . one can show R3 R3 = 1 Furthermore. 3 × 3–matrix. T R1 R1 −1 = 1 −1 = 1 1 1 (5. R3 = R1 R2 is a real.11) (iii) From detR1 = 0 follows that R1 is non-singular and. in fact. (ii) The group element ‘e’ is played by the 3 × 3 identity matrix 1 1. (5. 99 (5. R2 ∈ SO(3) and do not write out “◦” explicitly since it represents the matrix product. (iii) ∀a. b ∈ G the product c = a ◦ b is also in G. For this purpose we must demonstrate that the group properties (i–iv) hold. forms a group G satisfying the axioms (i) For any pair a. (i) Obviously.

e. 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 (5.13) R(ϑk ek )−1 = R(ϑk ek )T = [exp ( ϑk Lk )]T = exp ϑk LT ˆ ˆ k and. k = exp ( −ϑk Lk ) (5. −1. M4 } = 1 0 0 1 . −1 0 0 −1 .100 Theory of Angular Momentum and Spin Exercise 5. the matrices i Lk = − Jk (5.19) (5. x2 –plane {rotation by Oo . This representation expresses rotation matrices in terms of three 3 × 3– matrices J1 .15) Below we will construct appropriate matrices Jk . 0 1 −1 0 . 0 −1 1 0 (5. M3 . compact representation of R(ϑ) which will be of general use and. rotation by 90o . Lie Group. (b) the set of matrices {M1 . we arrive at the property exp ϑk LT k from which follows LT = −Lk .1. rotation by 180o . We want to show that for the property R RT = 1 to hold.16) must be antisymmetric.18) (5. J3 as follows i R(ϑ) = exp − ϑ · J (5. rotation by 270o } and consecutive execution of rotation as the binary operation ‘◦’. identify subgroups. Generators.20) (5. we want to assume that they exist. M2 . establish if 1-1 homomorphic mappings exist. mappings between the sets which conserve the group properties: (a) the set of real and imaginary numbers {1. Lie Algebra We want to consider now a most convenient. −i} together with multiplication as the binary operation ‘◦’. (c) the set of four rotations in the x1 . presently. i.17) . J2 .14) together with matrix multiplication as the binary operation ‘◦’. To demonstrate this property consider rotations with ϑ = ϑk ek where ek ˆ ˆ is a unit vector in the direction of the Cartesian k–axis.1: Test if the following sets together with the stated binary operation ‘◦’ form groups. in fact. will play a central role in the representation of many other symmetry transformations in Physics. due to the uniqueness of the inverse. i.

e. L3 which are independent of the rotation angles.26) (5.27) .15) with three real parameters ϑ1 .23) (5.24) (5. lim R[(0. A must have the form   0 a b A =  −a 0 c  . ϑ3 and three matrices L1 . 0. ϑ2 . We assume now that we parameterize rotations through a three-dimensional rotation vector ϑ = (ϑ1 .1. (5. ϑ3 )T ] = lim  sinϑ3 cosϑ3 0  =  ϑ3 ϑ3 →0 ϑ3 →0 0 0 1 0 0 1 Insertion into (5. 0)T . Using (5. 0.1: Matrix Representation of SO(3) 101 We can conclude then that if elements of SO(3) can be expressed as R = eA .5. Let us use this definition to evaluate L3 .(5.        (5. Using the definition of partial derivatives we can state L1 = lim ϑ−1 R(ϑ = (ϑ1 . ϑ2 . ϑ3 )T such that for ϑ = (0. the real matrix A is anti-symmetric. 0. i. 0 0 0 0 0 0  (5. 0.8) we can state    cosϑ3 −sinϑ3 0 1 −ϑ3 0 1 0  .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). L2 .22).25)     0 −ϑ3 0 0 −1 0  ϑ3 0 0  =  1 0 0  .. One can determine then the matrices Lk defined in (5. 0)T the rotation is the identity. 0)T ) − 1 1 1 ϑ1 →0 (5. Exercise 5.15) as the derivatives of R(ϑ) with respect to the Cartesian components ϑk taken at ϑ = (0. This property gives A then only three independent matrix elements.2: Determine the generator L1 according to Eq.22) and similar for L2 and L3 .21) −b −c 0 It is this the reason why one can expect that three-dimensional rotation matrices can be expressed through (5.

m = 3 for any even permutation of k = 1. 0. holds the algebra [Jk . B]] + [[A. it must hold 3 [Lk . B].15) and their generators obey the property (5.8) in case of ϑ = (0. 3 )T ) = exp( 3 L3 ).15) obey the group property. the matrix (5.. L ] = m=1 fk m Lm ..30) are called Lie groups. the Lie algebra of SO(3) can be written [Lk . L ] = k m Lm (5. (5. = 2. Finite rotations can be obtained by applications of many infinitesimal rotations of the type R(ϑ = ( 1 . are related to angular momentum operators. 0)T ) = exp( 2 L2 ) and R(ϑ = (0.. i. (5. e. [A. ϕ)T . The question is then. B] ). 1 1 e.g.31) where k m are the elements which are   0   1 k m =  1   −1 of the totally antisymmetric 3-dimensional tensor. Z2 = 2 [A.28) The construction of the generators through the limit taken in Eq. = 2.33) We want to demonstrate now that exp(ϑ · L) yields the known rotational transformations.102 Theory of Angular Momentum and Spin Sofar it is by no means obvious that the generators Lk allow one to represent the rotation matrices for any finite rotation. 2 . that the operators (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. (5. = 2. the elements of if any two indices are identical for k = 1. that the generators represent infinitesimal rotations with 1 >> |ϑ| around the x1 -. commutators of commutators of A and B. Lie groups can have any number of generators.32) For the matrices Jk = i Lk which.e. however. The expression for Zn are actually rather complicated.22) implies. 0. m = 3 for any odd permutation of k = 1.30) is called the Lie algebra. R(ϑ = (0. however. We want to consider.g. only the latter example and . and the constants fk m are called the structure constants. x2 -. i. the property (5. In fact.e. three being a special case. in fact. From this result one can conclude that expressions of the type (5. In case of the group SO(3) the structure constants are particularly simple. Z3 = 12 ( [A. An answer can be found considering the Baker-Campbell-Hausdorf expansion ∞ exp( λA ) exp( λB ) = exp( n=1 λn Zn ) (5.30) Groups with the property that their elements can be expressed like (5. J ] = i k m Jm .15). B]. m = 3 .29) where Z1 = A + B and where the remaining operators Zn are commutators of A and B. if the resulting products of exponential operators can be expressed in the form (5. and x3 -axes. (5. 0)T ) = exp( 1 L1 ). Of course. 0. as we see later. etc.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.

34. Exercise 5.  cos ϕ − sin ϕ 0 =  sin ϕ cos ϕ 0  0 0 1  (5. (5. according to (5.27). can be written ϕL3 = 0 103 A 0 0 0 0 .36) The property 0 −1 1 0 allows one to write 0 −1 1 0 and. eϕL3 which agrees with (5. that the matrix ϕL3 .35). We note.35) To determine expA = ∞ ν ν=1 A /ν! we split its Taylor expansion into even and odd powers 1 A2n + (2n)! 2 ∞ n=0 ∞ eA = n=0 1 A2n+1 . 5. A = ϕ 0 −1 1 0 . accordingly. using (5.38) eA = n=0 (−1)n 2n ϕ (2n)! 1 0 0 1 ∞ + n=0 (−1)n ϕ2n+1 (2n + 1)! 0 −1 1 0 .37) = (−1)n 1 0 0 1 .1.31). 0 −1 1 0 = (−1)n 0 −1 1 0 (5.5.39) Recognizing the Taylor expansions of cos ϕ and sin ϕ one obtains eA = and.1: Matrix Representation of SO(3) determine exp(ϕL3 ).40) . ∞ 2n = − 1 0 0 1 2n+1 (5.3: Show that the generators Lk of SO(3) obey the commutation relationship (5.8).41) cos ϕ − sin ϕ sin ϕ cos ϕ (5. (5. (2n + 1)! (5.34) One obtains then for the rotational transformation   exp A 0 0 0 0 0 1 0 0 1 0 0 0 1 0 0 1 0 0 0 1 0 0 1 ∞ ν=1 ∞ ν=1 0 0 1 =  =  =  +  + 1 ν! 1 ν! A 0 0 0 0 0 ν Aν 0 0 0 0 0 0 0 1 ∞ 1 ν ν=0 ν! A 0 0 = eA 0 0 . (5.

e.2. the R(ϑ) ψ(r) = ψ(R(ϑ) r ). = ψ([ R(ϑ1 ) R(ϑ2 ) ]−1 r) = ρ R(ϑ2 ) ψ([R(ϑ1 )]−1 r) (5. complex-valued functions over the 3-dimensional space R which can be differentiated infinitely often. namely R(ϑ)ψ(r ) = ψ(R−1 (ϑ) r ) . i.. (5.47) = ρ R(ϑ1 ) R(ϑ2 ) . i. a relationship which we like to express as follows R(ϑ) = ρ R(ϑ) . for A. Show that in this case holds Generators One can assume then that R(ϑ) should also form a Lie group. In analogy to R(ϑ) being a linear map R → R .44) holds.46) (5. Definition We first define how a rotational transformation acts on a wavefunction ψ(r ). For this purpose we require stringent continuity properties of the wavefunction: the wavefunctions under consideration must be elements of the set C∞ ( ).42) Obviously. we will learn that transformation patterns and invariances are connected with the angular momentum states of quantum mechanics.48) .2 Function space representation of the group SO(3) In this section we will investigate how rotational transformations act on single particle wavefunctions ψ(r).43) ρ( ) conserves the group property of SO(3). (5. In particular.1: Assume the definition ρ ρ (A B) = ρ (B) ρ (A). B ∈ SO(3) holds ρ(A B) = ρ(A) ρ(B) . (5. Exercise 5.104 Theory of Angular Momentum and Spin 5. its elements can be represented through an exponential R(ϑ) = exp i − ϑ·J (5.e. i. one isomorphic to SO(3) and.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) .e. hence. in fact. we define rotations R(ϑ) as linear maps C∞ ( ) → C∞ ( ). the transformation R(ϑ): C∞ ( ) → C∞ ( ) is related to the transformation R(ϑ): R → R .45) (5. Since this holds for any ψ(r ) one can conclude ρ R(ϑ1 ) ρ R(ϑ2 ) group property (5.

J3 ] and [J3 .2: (a) Derive the generators Jk .56) We like to verify this property for [J1 .56) holds for [J2 .2: Function Space Representation of SO(3) where the generators can be determined in analogy to Eq.54) (5. J2 ].49) (note the minus sign of −ϑ1 which originates from the inverse of the rotation in Eq. .2.49). 0.55) and f (r) = f [J1 . but also the Lie algebra of the generators. 0.5. −ϑ3 ) r ) − f (r) ) .  0 0 1 x3 x3 (5.22) according to i − J1 = lim ϑ−1 R(ϑ = (−ϑ1 . 0)T ) − 1 1 ϑ1 →0 105 (5. −ϑ3 ) r ≈  −ϑ3 1 0   x2  =  −ϑ3 x1 + x2 .(5. − 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..53) (5.55) Not only is the group property of SO(3) conserved by ρ( ). J2 ] f (r) = − = − 2[ 2 [− i J . 0. (5.53–5. ϑ3 )T ) f (r) − f (r) 3 (5. it holds [Jk . (5.57) Exercise 5. 2 by means of limits as suggested in Eq. show that (5.24) one can expand  ϑ3 →0 ϑ3 →0 lim ϑ−1 R(ϑ = (0. k = 1.49) to a function f (r).50) lim ϑ−1 ( f ( R(0. The generators Jk can be determined by applying (5.42)) and similarly for J2 and J3 . (5.51) Inserting this into (5. 0.23. in case of J3 .52) ( x2 ∂1 − x1 ∂2 ) f (r) . In fact. i − J3 f (r) = = Using (5. 5. (5. 3     1 ϑ3 0 x1 x1 + ϑ3 x2 R(0. J1 ].e.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. i. One obtains using (5. J ] = i k m Jm . 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 .

J3 ] = = = 2 2 [J1 + J2 . Show that for J3 as defined above holds exp( iα J3 ) ψ(φ) = ψ(φ + α).63) (5. J3 ] + [J1 . θ. (5. J3 ] (5. C] + [A. We will show now that the properties of spherical harmonics J2Y J3 Y m (θ.61) We demonstrate this property for k = 3. the operator 2 2 2 J 2 = J1 + J2 + J3 (5. x2 = r sinθ sinφ. φ) m (θ. J3 ] + [J2 .55) can be written as a vector product of r and i − J = −r × From this follows. according to the correspondence principle between classical and quantum mechanics. Jk ] = 0 . using the momentum operator p = ˆ J = r × i . 2. 2. C]B and (5.65) .59) which.3: Consider a wave function ψ(φ) which depends only on the azimutal angle φ of the spherical coordinate system (r. i. These eigenstates are the well-known spherical harmonics Y m (θ. k = 1. To show this we first note that (5. J3 ] J2 −i J1 J2 − i J2 J1 + i J2 J1 + i J1 J2 = 0 According to (5. However.106 Theory of Angular Momentum and Spin Exercise 5.58) i = r ×p ˆ (5.e. φ). usually chosen as J3 . φ) (5. Equation (5.56) are the famous commutation relationships of the quantium mechanical angular momentum operators. Using [AB. and x3 = r cosθ. identifies J as the quantum mechanical angular momentum operator. [J 2 . φ) = = 2 ( + 1) Y m (θ.2..61) simultaneous eigenstates of J 2 and of one of the Jk .e. 3 . φ). 3. k = 1.62) J1 [J1 . φ) as well as the effect of the so-called raising and lowering operators J± = J1 ± i J2 (5. . C] = A[B. can be found. 5. the system related to the Cartesian coordinates through x1 = r sinθ cosφ. The commutation relationships imply that no simultaneous eigenstates of all three Jk exist.64) mY m (θ.56) one obtains [J 2 . i.60) commutes with all three Jk .53–5. J3 ] J1 + J2 [J2 . (5.3 Angular Momentum Operators The generators Jk can be readily recognized as the angular momentum operators of quantum mechanics.

71) To prove this theorem we first show by induction that (i) holds. L− ] + L− L+ )| . L1 − iL2 ] = L2 + iL1 = i (L1 − iL2 ) = iL− and L3 | = −i | we obtain L3 L− | = ( iL− + L− L3 )| = i(− + 1) L− | .72) = 0 = −i | (5. L ] = k m Lm and let L± = L1 ± i L2 . Noting [L+ . k = 1. i. L1 − iL2 ] = −2iL3 and L+ | = 0 one obtains L+ L− | = −2iL3 | = −2 | .69) J+ Y (θ. φ) 107 = = 0 = = 0 ( + m + 1)( − m) Y ( + m + 1)( − m) Y m+1 (θ. For this purpose we first demonstrate that the property holds for m = − 1 by considering (i) for the state L− | ∼ | − 1 . 2. An Important Theorem Theorem 1. We will show that this property holds then also for m − 1.73) (5.e. φ) (5. φ) − m (θ. 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.1 Let Lk . in fact. J+ Y J− Y J− Y m (θ.76) We now show that property (ii) also holds for m = − 1 proceeding in a similar way. L− ] = [L3 .e. L− ] + L− L+ )| m = ( −2iL3 + . φ) are a consequence of the Lie algebra (5.70) (5. L2 = L2 + L2 + L2 . Let there exist states | in 1 2 3 H with the property L+ | L3 | then the states defined through L− | m + 1 = −i βm | m have the following properties (i) L+ | m (ii) (iii) L3 | m L2 | m = −i αm | m + 1 = −i m | m = − ( + 1) | m (5. The definitions of the coefficients α −1 and β −1 yield L+ L− | = L+ ( iβ −1 | − 1 = − α −1 β −1 | .5.74) (5. α −1 β −1 = 2 . φ) (θ. namely. We consider L+ L− | m = ( [L+ .66) (5. 3 be operators acting on a Hilbert space H which obey the algebra [Lk . φ) m+1 (θ. It holds L+ L− | = (L+ L− − L− L+ + L− L+ )| = ([L+ . L− ] = [L1 + iL2 . Using [L3 . For this purpose we prove the following theorem.67) (5.3: Angular Momentum Operators on the spherical harmonics. L− ] + L− L3 ) | .68) (5. L+ raises the m-value of | − 1 from m = − 1 to m = .56).. We note L3 L− | = ( [L3 . (∗) (5. i. and since we have already shown | − 1 = (iβ −1 )−1 L− | we can state L3 | − 1 = −i( − 1)| − 1 .75) (5. From this follows that L− | is an eigenstate of L3 .

θ. We will have various opportunities to employ the Theorem just derived. 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 . For this purpose we need to express the angular momentum operators in terms of spherical coordinates (r. 5. (5. i.79) We can now show that (iii) holds for all proper m–values.64).66. Before we can finally show that property (iii) holds as well for all − ≤ m ≤ we need to determine the coefficients αm and βm . (5. This completes the proof of the theorem. 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 .73.e. A first application will involve the construction of the eigenstates of the angular momentum operators as defined in (5.81) i. 5. 5.72.79) read L+ | m L− | m + 1 = −i = −i ( + m + 1)( − m) | m + 1 ( + m + 1)( − m) | m .84) . it also holds for m − 1.. We note L3 L− | m = ( [L3 .80) (5..53–5. in particular. 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 . finally αm = βm = (5.e. We note that we can write L2 = 1 1 1 1 2 2 2 2 L+ L− + 2 L− L+ + L3 . One can normalize the states | m such that αm = βm . given in (5. yield properties (5.82) (5. This implies that (i) holds for m − 1. from L+ L− | m = − αm−1 βm−1 | m follows the recursion relationship αm−1 βm−1 = 2m + αm βm .69). it holds αm βm = 2 one obtains αm βm = = ( + 1) k=m+1 k . We can deduce from (∗) and (∗∗) α α α −1 β −1 −2 β −2 −3 β −3 = 2 = 2( − 1 + ) = 2( − 2 + · · −1 + ) Obviously.77) We will show now that if (ii) holds for m. φ. Using the familiar formula + m − n k=0 k = n (n + 1)/2 − (m + 1) m = ( + 1) m − (m + 1) m (5.83) (5.63. in terms of spherical coordinates defined through x1 x2 x3 = = = r sin θ cos φ r sin θ sin φ r cos θ (5. 2 We note that Eqs.70.78) ( + m + 1)( − m) ( + m + 1)( − m) .55). 5. (∗∗) (5. Again the argumemts are similar to the ones used above.

90) This yields. using (5. in terms of spherical coordinates. In fact.92) (5. Using (5.93) and similarly.3: Angular Momentum Operators 109 We first like to demonstrate that the generators Jk . According to (5. φ) = ( x3 ∂2 − x2 ∂3 ) f (θ. φ) (5. φ) sin θ ∂θ sin θ ∂θ L1 f (θ. φ).91) which agrees with expression (5.88) (5. using (5.89) = ∂ x2 x2 ∂ x2 + x2 ∂ x3 − 33 − x2 1 3 2 x1 ∂ tan φ r ∂ cos θ r ∂ cos θ f (θ. the latter defined in (5.5.86). actually.82–5. (L1 )2 = = ∂ 2 ∂ + cot θ cos φ ∂θ ∂φ 2 ∂ ∂ 1 ∂2 sinφ cosφ sin2 φ 2 − + 2 cotθ sinφ cosφ ∂θ ∂φ ∂θ∂φ sin2 θ ∂2 ∂ + cot2 θ cos2 φ 2 + cotθ cos2 φ ∂θ ∂φ sin φ i ∂φ . φ) ∂x2 /x1 ∂ ∂x3 /r ∂ = x3 + x3 ∂x2 ∂ tan φ ∂x2 ∂ cos θ ∂x2 /x1 /r ∂ ∂ − x2 − x2 ∂x3 3 ∂ cos θ f (θ. (L2 )2 = − cos φ ∂ ∂ + cot θ sin φ ∂θ ∂φ 2 . involve only the angular variables θ and φ and not the radius r.87) holds J3 = To determine J 2 we note. using ∂/∂tanφ = cos2 φ ∂/∂φ and ∂/∂cosθ = −(1/sinθ) ∂/∂θ.85). for example.85).60). We like to express now the operators J3 and J 2 . L1 f (θ.86) (5. φ) = (5. cos θ ∂ cos2 φ + sin θ cos φ ∂φ cos2 θ sin θ sin φ ∂ sin θ sin φ ∂ + + sin2 θ f (θ. (5. φ) = (x3 ∂2 − x2 ∂3 )f (θ. φ) ∂x ∂x3 ∂ tan φ =0 (5.87) To derive these properties we consider. ∂φ = sin φ (5.85) (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 ∂ = − . applying repeatedly the chain rule. L1 f (θ.84 ) one obtains.

5.85.98) where Jclass = r × mv is the angular momentum and pr = mr is the radial momentum.94) It follows (L1 )2 + (L2 )2 + (L3 )2 = With cot2 θ + 1 = 1/sin2 θ.64) can be constructed as follows.4 Angular Momentum Eigenstates According to the theorem on page 107 above the eigenfunctions (5. 5. 2m r ∂r2 2m r2 2 (5.95) sinθ ∂ ∂θ = ∂2 ∂ + cotθ .100) (5.85.110 = 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 . These generators act on a subspace of C∞ ( ).97) Kinetic Energy Operator The kinetic energy of a classical particle can be expressed J2 p2 p2 = r + class 2m 2m 2m r2 (5. namely on the space of complex-valued functions on the unit 3-dim. ∂θ2 ∂θ ∂φ (5.63. 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 (5.86.86.101) 1 ∂2 r r ∂r2 and comparision with (5. C∞ (S ).87). One first identifies the Hilbert space H on which the generators Lk operate.97).96) and the relationship between Lk and Jk as given in (5. The operators in the present case are Lk = − i Jk where the Jk are differential operators in θ and φ given in (5. 5.e. sphere S2 .87). The functions in C∞ (S ) have real variables . 5. 1 sin2 θ sinθ ∂ ∂θ (5. 5. ∂θ ∂φ ∂2 ∂ ∂2 + cotθ + ( cot2 θ + 1 ) 2 . one can conclude J2 = − 2 sin θ 2 sinθ ∂ ∂θ 2 + ∂2 ∂φ2 . (5. i. 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 + . 2 ∂θ ∂θ (5.

104. φ) = = 0 −i Y (θ. 5. φ) (5. φ) . . The norm in H is defined through |f |2 = S2 dΩ f ∗ (θ.103) iteratively for m = − 1. φ) = 2 + 1 ( − m)! P 4π ( + m)! m (cosθ) eimφ . by carrying out the construction according to (5.108) m (θ. The function space endowed with this norm and including only functions for which the integral (5. (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. normalizing these functions.85–5.106) The latter property is obviously satisfied for the chosen φ-dependence. φ) = −i ( + m + 1)( − m)Y m (θ. } applying L− Y m+1 (θ. φ) (5. (5. can be expressed L± = Accordingly. (5. − + 1.102) where dΩ = sinθ dθ dφ is the volume element of S2 . φ. φ). φ) (5.107) (5. φ) = −i Y (θ. The eigenfunctions (5. . (5. .63.87). One obtains in this way Y m (θ. . 0 ≤ θ < π.102) exists. φ) = 0 as well as L3 Y (θ. and then determining the family {Y m (θ.4: Angular Momentum Eigenstates 111 θ.105) explicitly. − 2. m) = 1 −m J− 1 2 Y (θ. and to be admissable for description of quantum states.109) using (5. 0 ≤ φ < 2π. The raising and lowering operators L± = L1 ± i L2 .104) (5.105) ∆( . .5. φ). is indeed a Hilbert space. φ) in H which satisfy L+ Y (θ. . φ) L3 Y (θ. m) Constructing Y → Y →→ ··· Y ( + m)! (2 )! ( − m)! −1 − We like to characterize the resulting eigenfunctions Y m (θ. m = − . φ)f (θ. 5. the so-called spherical harmonics. φ). .107) reads ∂ ∂ + i cotθ ∂θ ∂φ P (cosθ) ei φ i e±iφ ∂ ∂ ± i cotθ ∂θ ∂φ .111) . φ) = ∆( . − . .64) can be constructed then by seeking first functions Y (θ. must be cyclic in φ with period 2π. (5.

114..e. 5. A proper solution of this equation.119) provide the following expression for Y Y (θ. one for which π dθ sinθ |P (cosθ)|2 (5. (5. 1975) . (5. (5. = = x (1 − x2 ) 2 x3 3 + 2 −1 +1 −1 −1 dx x2 (1 − x2 ) + 2 · (2 − 2) 1·3 −1 +1 (1 − x2 ) dx x4 (1 − x2 ) −1 −2 +1 2 · (2 − 2) · · · 2 dx x2 1 · 3 · 5 · · · (2 − 1) −1 2 · (2 − 2) · · · 2 2 (2 )! 2 = 2 2 +1 1 · 3 · 5 · · · (2 − 1) 2 + 1 [1 · 3 · 5 · · · (2 − 1)] (5.118) 2 +1 1 1 sin θ ei 4π (2 )! 2 ! φ .D. One obtains. . for example.115) holds.106.112 or Theory of Angular Momentum and Spin ∂ − ∂θ cotθ P (cosθ) = 0 (5.114) dθ sinθ |Y (cosθ)|2 = 1 0 (5.116) The integral appearing in the last expression can be evaluated by repeated integration by parts. ”Classical Electrodynamics.113) 0 is finite. The normalization factor N is chosen such that π (5.” by J. (5.112) where we employed the definition of P (cosθ) in (5.120) See.119) where the factor (−1) has been included to agree with convention1 . π +1 dθ sin2 0 +1 θ = −1 +1 −1 dx (1 − x2 ) +1 = = . 5. We note that (5. φ) = (−1) 1 1 = 1 [1 · 3 · 5 · · · (2 − 1)]2 (5. Jackson (John Wiley.114) yield |N |2 2 +1 1 2π 4π (2 )! π dθ sin2 0 +1 θ = 1.116) implies |N |2 from which we conclude N = (−1) 1 · 3 · 5 · · · (2 − 1) (5. i.106.107). New York. 2nd Ed.117) Accordingly. 5. using the new integration variable x = cos θ. . is P (cos θ) = N sin θ as one can readily verify.

5.125) which agrees with (5. employing again the variable x = cos θ. φ).126) Then holds.124) reads P (x) = = (2 )! (1 − x2 )− 2 (x2 − 1) 2 ! (−1) 1 · 3 · 5 · · · (2 − 1) sin θ (5.122) The latter identity can be written.124).103. P m (x) = −1 ( + m)! 2 ! ( − m)! × (1 − x2 )− m+1 2 −(1 − x2 ) 2 1 ∂ x + (m + 1) 1 ∂x (1 − x2 ) 2 × (5.124) we proceed by induction.127) ∂ −m−1 (x2 − 1) . .123) leads to the expression P m (x) = −m m ∂ 1 ( + m)! (1 − x2 )− 2 (x2 − 1) 2 ! ( − m)! ∂x −m (5.114. and can obtain now. .63. through repeated application of (5. (5.123). every application of L− reduces the power of eiφ by one. φ).110) together with (5. use (5. 5.106).124) called the associated Legendre polynomials.123) P m+1 (x) ∂ x + (m + 1) 1 ∂x (1 − x2 ) 2 .64). To prove (5. m = − 1. Let us then assume that (5.124) holds for m + 1. For m = (5. describes the φ-dependence of Y m (θ. − 2. (5. P m+1 (x) = −m−1 m+1 ∂ 1 ( + m + 1)! (1 − x2 )− 2 (x2 − 1) . Actually. ∂x −m−1 . . 2 ! ( − m − 1)! ∂x −m−1 (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) . The reader should note that the associated Legendre polynomials. namely Y (θ. 5.e.121) This expressions shows that the factor eimφ . we seek to determine the functions P m (cos θ) and. 5.119). φ). are real. i. indeed.. 5.103. according to (5. − defined in (5.4: Angular Momentum Eigenstates 113 We have constructed a normalized solution of (5.110). therefore. the eigenfunctions Y m (θ. as specified in (5. P m (x) = −1 × ( + m + 1) ( − m) × −(1 − x2 ) 2 1 (5. m (cos θ) e (5. We want to demonstrate now that the recursion equation (5.106) i e−iφ ∂ ∂ − i cotθ ∂θ ∂φ 2 + 1 ( − m − 1)! P 4π ( + m + 1)! 2 + 1 ( − m)! P 4π ( + m)! i(m+1)φ m+1 (cos θ) e = −i ( + m + 1)( − m) imφ .

expression (5. using x = cos θ. therefore. 5. (5. However.132). from such construction emerges an important symmetry property of Y m (θ. φ) using (5.134) (x) = 1 2 ! (1 − x2 ) 2 . (5. Such construction reproduces the eigenfunctions Y m (θ.106. and constructs then the eigenfunctions Y − +1 . i. therefore. P − (5. hence. therefore.128) one can show that (5. we embark on this construction in order to prove (5.135) . the derivative ∂ 2 +1 /∂x2 +1 of this expression vanishes.132) which reduces the number of spherical harmonics which need to be evaluated independently roughly by half . Since (5. if L− Y (θ. therefore. φ) L3 f (θ.131) identifies f (θ. In fact.124).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) holds for m if it holds for m + 1.124). Y − +2 .129) holds. P − (x) = 1 ∂2 (1 − x2 ) 2 (x2 − 1) . etc. Constructing Y − → Y − +1 → ···Y m (θ. Y ∗ (θ. by repeated application of the operator L+ . 5. We want to test if the recursion (5. appears not very interesting. it holds then for all m. 2 ! (2 )! ∂x2 1 (5. φ) as given in (5.133) Since the term with the highest power of (x2 − 1) is x2 . φ). φ) = (−1)m Y m −m (θ. it holds ∂2 (x2 − 1) = (2 )! ∂x2 and. φ) = 0 (5. yields a vanishing expression for m = − − 1 as long as is a non-negative integer.124) and. which is equivalent to recursive application of L− . φ) (5.124) and. namely.e. φ) . φ) An alternative route to construct the eigenfunctions Y of L− f (θ.124) holds for m = . φ) = = determines first a normalized solution 0 i f (θ.. We first determine Y − (θ. φ) = Y − (θ. that expression (5.127) reproduces (5.130) (5. In that case (1 − x2 ) is a polynomial of degree 2 and.123) terminates for m = − . φ) choosing the proper sign.106. It holds.

(5.106.120) for Y obey the postulated relationship (5. We first note that for m = − (5. φ) = 2 +1 1 1 sin θ e−i 4π (2 )! 2 ! φ 115 − .136) to construct all other Y recursion eiφ ∂ ∂ + i cotθ ∂θ ∂φ Y m (θ.142) yields P − (x) = (−1) 1 (1 − x2 )− 2 (x2 − 1) = (1 − x2 ) 2 2 ! 2 ! (5.136) is normalized and has the proper sign. the expression (5. 5.141) We want to demonstrate now that this recursion equation leads to the expression P m (x) = +m m ∂ (−1)m (1 − x2 ) 2 (x2 − 1) 2 ! ∂x +m (5. One can readily verify that this expression provides a solution of (5. According to (5.143) .122. 5.123)] P m+1 (x) = −(1 − x2 ) 2 1 ∂ x − m 1 ∂x (1 − x2 ) 2 P m (x) . a sign consistent with the family of functions Y m constructed above.110) applies the = ( + m + 1)( − m) Y m+1 (θ.138) Employing (5.e. Obviously.80) and (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.. (5.129). (5. following closely the proof of eq.4: Angular Momentum Eigenstates Due to (5.124) one arrives at Y (θ.136) We note in passing that this expression and the expression (5.5.106). (5.142) For this purpose we proceed by induction. φ) . (5.139) = ∂ − m cotθ ∂θ P m (cos θ) . (5. i. This follows from the identity ∂ ∂ − i cotθ ∂θ ∂φ One can use (5.124).f.132). (5.140) Introducing again the variable x = cos θ leads to the recursion equation [c.137) m (θ. φ). φ) sin θ e−i φ = 0.

s. The construction beginning with Y − and continuing with Y − +1 . of which agrees with the r.144) ∂ x − m 1 ∂x (1 − x2 ) 2 +m m ∂ (−1)m (1 − x2 ) 2 (x2 − 1) . one notes that application of (5.147) the r. (5.149) (5. one can conclude P m (x) = (−1)m m (θ. of (5.124) yields (−1)m ( − m)! P ( + m)! −m (x) = +m m ∂ (−1)m (1 − x2 ) 2 (x2 − 1) . Since (5. However. According to both (5.132). φ) ( − m)! P ( + m)! −m (x) .142) holds for m + 1 if it holds for m. (5. (5. 1 (1 − x ) ∂x +m (1 − x2 ) 2 1 (5. 2 ! ∂x +m −(1 − x2 ) 2 Replacing in (5. i.146) i.150) . are identical.142) one can state P (x) = ∂ (x2 − 1) .148) According to (5. P (x) = P 0 (x) are called Legendre polynomials.142) holds for m = − we verified that it holds for all m. yields the same eigenfunctions as the previous construction beginning with Y and stepping down the series of functions Y −1 . Y − +2 . 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 . Y −2 .128) m + 1 → −m or. etc.124).145) Hence..144) reads P m+1 (x) = +m+1 m+1 ∂ (−1)m+1 (1 − x2 ) 2 (x2 − 1) .142) and by (5. Accordingly. We then assume that (5. Hence.141) reads then P m+1 (x) = 1 (5. equivalently.116 Theory of Angular Momentum and Spin wich agrees with (5.124) and (5. 2 ! ∂x +m+1 (5. 2 ! ∂x +m (5. 2 ! ∂x 1 (5. etc. also the associated Legendre polynomials determined this way.s.h.106) this implies for Y The Legendre Polynomials The functions the identity (5.142) holds for m. (5.e.135). given by (5..h.142).e.

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. accordingly. 2 3 1 1 = = 2 1−t 1 − 2t + t ∞ t .156) yields P (1) = 1 . 1..157) = 0. . P3 (x) = 1 2 117 P2 (x) = − 3x) 1 2 2 (3x − 1) . the spherical harmonics for m ≥ 0 Y m (θ. The spherical harmonics for m < 0 can be obtained using (5. t) = √ 1 − 2xt + t2 ∞ 1 − 2 x t + t2 . pp. ”Classical Electrodynamics. N.N. Lebedev (Prentice-Hall. (5. For example. (5. 2nd Ed. . Using x = cos γ. 2. 92 To prove this property see.132).. φ) = 2 + 1 ( − m)! (−1)m sinm θ × 4π ( + m)! × ∂ ∂ cos θ m (5.g.153) P (cos θ) eimφ .J.152) and. 1965) which is an excellent compendium on special functions employed in physics.158) see.150) and (5.155) P (x) t . t) is called a generating function of the Legendre polynomials3 . 1975). =0 (5. t = r2 /r1 . Jackson (John Wiley. (5.5.D.4: Angular Momentum Eigenstates The first few polynomials are P0 (x) = 1 . . . Englewood Cliffs. New York. in case of x = 1 holds w(1. =0 (5. pp. and |r1 − r2 | = r1 (5.” by J. r1 is the point where the potential is measured. P1 (x) = x . e. t) = √ Comparision with (5.154) can be written 1 = w(x.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) . In case q = 1 and |r2 | < |r1 | holds the identity2 ∞ 1 r2 = P (cos γ) (5.154) |r1 − r2 | r +1 =0 1 where γ is the angle between r1 and r2 .156) w(x. for example. m ≥ 0 ∂xm (5. 45 of “Special Functions and their Applications” by N. The generating function allows one to derive useful properties of the Legendre polynomials.151) (5x3 Comparision of (5. and r2 denotes the location of the charge.

164) which. . 1.168) mY m (θ. If the spherical coordinates of r are (r. θ. ∂t 1 − 2xt + t2 (5.161) Collecting coefficients with equal powers t yields 0 = P1 (x) − x P0 (x) ∞ (5. π − θ. The spherical harmonics are eigenfunctions of the angular momentum operators J2Y J3 Y m (θ. then the coordinates of − r are (r. φ) .166) We want to summarize the properties of the spherical harmonics Y derived above. Inversion Symmetry of Y m (θ. Due to cos(π − θ) = − cos θ the replacement r → −r alters P m (x) into P m (−x). φ). using P0 (x) = 1 [c. Accordingly.118 For w(x. φ) m (π − θ. t) holds Theory of Angular Momentum and Spin ∂lnw(x. φ) m (θ. Inspection of (5. φ) = = 2 ( + 1) Y m (θ.159) by w(1 − 2xt + t2 ) leads to the differential equation obeyed by w(x.163) (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. t) (1 − 2xt + t2 ) Employing (5.159) Using ∂lnw/∂t = (1/w) ∂w/∂t and multiplying (5.151)] allows one to determine P (x) for = 1. φ) (5.142) allows one to conclude P m (−x) = (−1) +m P m (x) (5. . holds P1 (x) ( + 1) P +1 (x) = = x P0 (x) (2 + 1) x P (x) − P −1 (x) (5. . . (5. t) x−t = .156) and (∂/∂t) to P (x)t = ∞ ∂w + (t − x) w = 0 . 2.. π + φ). φ) goes over to Y m (π − θ. φ) . π + φ) = (−1) Y m (θ. φ) m (θ. ∂t P (x)t −1 −1 (5.165) since ∂ n /∂(−x)n = (−1)n ∂ n /∂xn .f.160) this differential equation is equivalent ∞ (1 − 2xt + t2 ) =0 P (x) t + (t − x) =0 P (x) t = 0 (5. under inversion Y m (θ. Accordingly. φ) Under inversion at the origin vectors r are replaced by − r. π + φ). (5.167) (5. Noting exp[im(π + φ)] = (−1)m exp(imφ) we determine Y Properties of Y m (θ.

except for the factor exp(imφ). For the Legendre polynomials holds the orthogonality property +1 dx P (x) P ( (x) = −1 2 δ 2 +1 .176) = 1 +1 [ (2 + 1) x P (x) − are the Legendre polynomials. φ) C = =0 m=− π C sin θdθ 0 m Y m (θ.5. φ) = δ δm m .171) m = dφ Y ∗ (θ.e. (5. ..177) Note that the spherical harmonics are real. φ) 2π 0 (5.169) 3.) P0 (x) = 1 . φ) = 2 +1 P (cos θ) . P −1 (x) ] (5. φ)f (θ. 6. φ) ∈ C∞ (S2 ) holds ∞ f (θ. 5. 2. . The spherical harmonics for m < 0 are given by Y −m (θ.179) . The spherical harmonics form. φ) = 2 + 1 ( − m)! (−1)m sinm θ × 4π ( + m)! 2 ! × ∂ ∂ cos θ m (5. . P +1 (x) P1 (x) = x . for any f (θ. The spherical harmonics are given by the formula Y m (θ. φ) Y m (θ. φ) = = ( + m + 1)( − m) Y ( + m + 1)( − m) Y m+1 (θ. φ) = (−1)m Y ∗ (θ. m (5.173) m+1 (θ. φ) m (θ. φ) .178) 7.174) m ≥ 0 P (cos θ) eimφ . The spherical harmonics form an orthonormal basis of the space C∞ (S2 ) of normalizable.170) (5.172) (5. φ) where J± = J1 ± iJ2 . where ( = 1. φ) (5. (5.175) (5. a complete basis of C∞ (S2 ). uniquely defined functions over the unit sphere S2 which are infinitely often differentiable π 2π sin θdθ 0 0 dφY ∗m (θ. in fact. m 4. 4π (5. The spherical harmonics Y 0 are Y 0 (θ. i.4: Angular Momentum Eigenstates 119 2. The spherical harmonics obey the recursion relationships J+ Y J− Y m (θ. φ) .

188) Exercise 5.181) = 0. 11.184) Y22 Y21 Y20 together with (5.h. (5.87). (5. φ).120 8.189) evaluate the integral on the l.s.h. For the Laplacian holds 2 = = = (5. The spherical harmonics for θ = 0 are Y m (θ Theory of Angular Momentum and Spin = 0.179) using the generating function w(x.4. Exercise 5.2: Derive the orthogonality property for the Legendre polynomials (5.185) (5.86).3: Construct all spherical harmonics Y3m (θ. 4π (5.177). The spherical harmonics for m (θ. t) stated in (5.s. φ) = δm0 2 +1 . . =0 t + −1 dxP (x) P (x) .180) 9.189).183) (5.182) Y11 Y10 = = − (5.1: Derive expressions (5.4. φ) = (−1) Y m (π − θ. (5. (5.186) (5. For this purpose start from the identity +1 ∞ +1 dx w(x. π + φ) .156). of (5. φ) . expand the result in powers of t and equate the resulting powers to those arising on the r.187) h(r) Y m (θ. t)2 = −1 .4.. The spherical harmonics obey the inversion symmetry Y 10. Exercise 5. 1. φ) = 1 ∂2 r − r ∂r2 ( + 1) r2 h(r) Y m (θ. 2 are given by Y00 = 1 √ 4π 3 sin θ eiφ 8π 3 cos θ 4π 1 4 − 15 sin2 θ e2iφ 2π 15 sin θ cos θ eiφ 8π 5 4π 3 1 cos2 θ − 2 2 (5.

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

e.n2 . with blocks of dimension 1. one concludes for the matrix elements (5. i. . One = contribute can conclude then that in the expansion (5. One proceeds by expanding first ϑ · J in terms of J+ .195) [D(ϑ)] m.e.e. 5.194) only spherical harmonics with or.n3 J+ 1 J− 2 J3 3 . (5. in case m = − .   ..201) From expression (5. . (5.199) [J+ . .. i. one can conclude that J+ leaves X invariant. φ) . If one orders the basis according to the partitoning (5. equivalently. φ) exp m i − ϑ·J Y m (θ. Similarly. J− . produces 0.202) D(ϑ) =           (2 + 1) × (2 + 1)               . n The action of J− on Y m produces Y m−1 or. and J3 1 1 ϑ · J = ϑ− J+ + ϑ+ J− + ϑ3 J3 (5. i. .200) of the rotational transformations one can also conclude that the X are the smallest invariant subspaces of C∞ (S ).200) leaves X invariant. leaves X invariant.197) of C∞ (S ) the matrix representation of D(ϑ) assumes a block-diagonal form. J− ] = 2 J3 . 3.. Since the Y m are n3 eigenfunctions of J3 the action of J3 reproduces the basis functions of X .198) 2 2 where ϑ± = ϑ1 ( 1 ϑ− J+ 2 + 1 2 ϑ+ J− ϑ2 . The exponential operator exp − i ϑ · J + ϑ3 J3 )n . m = δ S2 dΩ Y ∗ (θ.56. J± ] = ± J± (5.     (5.122 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 . J− 2 also leaves n1 X invariant.n3 n n n Accordingly. .200) in (5. Since any additive term n n n cn1 ...65).n2 . the transformation exp − i ϑ · J leaves X invariant as well.   1×1         3×3           . Application of (5. 5.. one needs to consider the action of J+ 1 J− 2 J3 3 on the subspaces X .199) allows one to express exp i − ϑ·J = n1 . can be expanded in powers One employs then the commutation properties [J3 . which follow readily from (5.

(f) The transformation ρ(ϕ) as defined in (e) leaves the following subspace of functions considered in (e) Xm = { f (α) = A e−imα . Exercise 5. The representation is referred to as the irreducible representation.22. f (α + 2π) = f (α). other representations are termed reducible representations. This step reconsiders the parametrization of rotations by the vector ϑ assumed sofar. (ii) a second rotation around the new x2 –axis by an angle β. this parametrization.6: Wigner Rotation Matrices 123 This representation of D(ϑ) is the one wich has blocks of the lowest dimensions possible. (d) Show that the similarity transformation T (ϕ) defined in the space of non-singular. for which holds RT R = R RT = 1 1. i. (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). 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.e. and for which detR = 1. and periodic functions f (α). is a group.e. does not provide the simplest mathematical description of rotations. (e) In the space C∞ ( ) of infinitely often differentiable. (b) Prove that the elements of SO(2) can be completely characterized through a single parameter.49). 5. though seemingly natural. ∈ N} invariant.5.6 Wigner Rotation Matrices We will now take an important first step to determine the functional form of the matrix elements of D(ϑ). 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). 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). However. 3 certainly allow one to describe any rotation around the origin. Determine the generator of ρ(ϕ) in analogy to (5. 2. (iii) a third rotation around the new x3 –axis by an angle γ. . A ∈ C. k = 1. Give the corresponding expression of ρ(ϕ). The three components ϑk . i. 5. (a) Show that SO(2) with the group operation ◦ defined as matrix multiplication.1: Representations of the group SO(2) SO(2) is the set of all 2×2 matrices R which are orthogonal.5.

the transformation from the original frame to the rotated frame. m i Y m (θ. the transformation from the rotated frame to the original frame.207) which replaces J2 by the inverse transformation (i). β. Obviously. φ) = S2 dΩ Y ∗m (θ.207) are equivalent. in terms of rotations defined with respect to the original frame. φ) (5. For any similarity transformation involving operators A and S holds eS A S Accordingly. i.e.204) is satisfied.124 Theory of Angular Momentum and Spin The angles α. γ)] = . i (5. of (5.s. in analogy to (5.209) in this expression one obtains e− i γJ3 = e− i αJ3 e− i βJ2 γJ3 e i αJ3 e− i i αJ3 e− i i γJ3 e i i αJ3 γJ3 e− e i i αJ3 e i i βJ2 e i αJ3 = e− i = e− αJ3 e− βJ2 e− βJ2 e αJ3 (5.209).h.203).208) αJ3 βJ2 = e− αJ3 e− i βJ2 e i (5. For this purpose we notice that J2 can be expressed through the similarity transformation J2 = e− i αJ3 J2 e i αJ3 (5. the axis x3 is defined in the coordinate frame which is related to the original frame by the consecutive rotations (i) and (ii). followed by J2 in the original frame. (5. (5. m [D(α. however. is e− i γJ3 = e− i βJ2 e− i αJ3 e− i γJ3 e i αJ3 e i βJ2 (5. We will demonstrate that (5. Accordingly.209) The first two rotations in (5. β.203) has the disadvantage that it employs rotations defined with respect to three different frames of reference.203) can be expressed.212) . The Euler rotation replaces exp − i ϑ · J by e− i γJ3 − i βJ2 − i αJ3 e e . we can write e− i −1 = S eA S −1 .211) Using (5. The axis x2 is defined in the coordinate frame which is related to the original frame by rotation (i). β. φ) [D(α. the l. i. can then be written e− i βJ2 − i αJ3 e = e− i αJ3 e− i βJ2 (5.e. γ will be referred to as Euler angles... followed by transformation (i). and the r. 5. γ)] m.e.205) Y m (θ.. φ) e− γJ3 − i βJ2 − i αJ3 e e .206) The expression (5.210) The third rotation in (5.203).m m. i. γ ∈ R such that exp i − ϑ·J = e− i γJ3 − i βJ2 − i αJ3 e e (5.203) For any ϑ ∈ R one can find Euler angles α.195) by ˜ Y m (θ.194. β. (i) and (ii). one can replace (5.h.s.

γ)] = .217) Accordingly.310)] an explicit expression for the Wigner rotation matrix (5. β. 5.64) yields dΩ f (θ.7 Spin 1 2 and the group SU(2) The spin describes a basic and fascinating property of matter. φ) . φ) e− S2 i γJ3 Y m (θ. in addition. (5. φ) i i (5.203) with respect to the original (unprimed) frame one simply needs to reverse the order of the rotations.7: Spin 1 2 and the group SU(2) 125 Multiplication from the left with (5.216) Using.220) We will derive below [see Eqs. the self-adjointness of the operator J3 one can state dΩ Y ∗ (θ. m (5. (5.218) Defining the so-called Wigner rotation matrix dmm (β) = one can express the rotation matrices [D(α. φ) [D(α.. φ) = e−im α S2 dΩ Y ∗ (θ. to redefine the three rotations in (5. Examples are the other five members of the lepton family to which the electron belongs.215) benefits from the choice of Euler angles for the parametrization of rotational transformations. Best known is the spin of the electron.219). the muon µ and its neutrino νµ of the second .309. but many other elementary components of matter are endowed with spin-like properties.219) = e−iαm dm m (β) e−iγm δ ( ) . one can write [D(α. (5. β. φ) (5.205. φ) e− αJ3 e− βJ2 Y m (θ.5.m m. 5. m = e−iαm S2 dΩ Y ∗ (θ. φ) e− m i βJ2 Y m (θ.210) yields the simple result e− i γJ3 − i βJ2 − i αJ3 e e = e− i αJ3 e− i βJ2 e− i γJ3 . 5. γ)] m. m i Y m (θ. This allows one to express the rotational trasnformation (5.e. φ) Y m (θ. φ) e−iγm δ (5. the electron e and the electron neutrino νe of the first generation. φ) e− m S2 i αJ3 f (θ.215) The evaluation of the matrix elements (5. γ)] m. φ) = S2 dΩ f (θ. β.213) i. φ) f (θ. The eigenvalue property (5. m ( ) S2 dΩ Y ∗ (θ. (5. φ) .214) e− γJ3 = S2 dΩ Y ∗m (θ. γ)] m. (5. φ) e− m i βJ2 Y m (θ.206) by ˜ Y m (θ. m [D(α. β. φ) e−imγ .

This specification implies that we consider transformations save for overall factors eiφ since such factors are known not to affect any observable properties. 3 representing a 2 × 2 matrix. We will consider first only the so-called spin– 1 . So do the three generations of six quarks. in fact.g. We may finally mention that the spin is at the heart of many properties of condensed matter systems. It appears to be rather impossible for a Physicist not to be enamored with the spin property. which might be the reason why consideration of rotational symmetry is so often fruitful in the study of matter. the gluon. the quarks. as long as |c+ |2 + |c− |2 = 1. there are together five conditions in terms of real quantities to be met by the matrix elements and. the tau τ and its neutrino ντ of the third generation and their antiparticles carry a 1 spin. eight real numbers to specify the matrix elements.1 systems 2 2 can be related to two states which we denote by χ+ and χ− . four real and four imaginary parts of Ujk . Conservation of the scalar product under symmetry transformations requires ∗ the property U U † = U † U = 1 where U † denotes the adjoint matrix with elements U † jk = Ukj . the photon. If we identify the states χ+ and χ− with the basis of a Hilbert space. i. e. and because of detU = 1. carry other spin-like properties which together.221) One can show readily that this set forms a group with the groups binary operation being matrix multiplication. The transformation matrices are then elements of the set SU (2) = { complex 2 × 2 matrices U . U U † = U † U = 1 detU = 1 } 1. The particles mentioned. The important feature is that all U ∈ SU(2) can be parametrized by an exponential operator U = exp − i ϑ · S (5. 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− .3 . k = 1. There .222) where the vector S has three components. We will find that the spin in its transformation behaviour is closely related to angular momentum states. c± ∈ C. particles which mediate the strong and the electro-weak interactions. leaving its imprint on the properties of nuclei. As complex 2 × 2 matrices one needs. 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. and which carry spin 1. There are also the mediators. in principle. 2.126 Theory of Angular Momentum and Spin generation. there is nothing more elementary to matter than the spin property. j.2 . The presence of this property permeates matter also at larger scales than those of the elementary particles mentioned. Spin. atoms and molecules. 2. the spin of the electron is likely the most important property in Chemistry. then allowed + − symmetry transformations of spin states are described by 2 × 2-matrices U with complex-valued matrix elements Ujk . k = 1. Such systems can also assume any linear combination c+ χ+ + c− χ− . two of which (in certain linear combinations) make up the vector mesons which carry spin 1. In any case. (5.e. have properties beyond those of single spins. a relationship. hence.1 and spin. the degrees of freedom of the matrices U are three real quantities. like superconductivity and magnetism. each component Sk . however. one for each matrix element of U † U . How can the elements of SU(2) be parametrized. [Sk ]mn = ([Sk ]nm )∗ . and the condition detU = 1 requires the Sk to have vanishing trace. One can show that the unitarity condition requires these matrices to be hermitian. the two W± and the 2 2 Zo . and three of which make up the baryons which carry spin. generalizing then further below.

2. 2 S3 = 1 σ3 2 (5.229) = 1 . 1 σj σk = − σk σj for j = k which can also be readily verified. y.k=1 σ σ 3 j k j.7: Spin 1 2 and the group SU(2) 127 exist three such linear independent matrices. 3 are the Pauli matrices σ1 = 0 1 1 0 . (5. but their comonents must not necessarily commute with each other. b are vectors commuting with σ. they are associated with the important fermion property of matter. (σ · a) (σ · b) = a · b 1 + iσ · (a × b) 1 (5. One can readily verify the commutation property σj σk − σk σj = [σj . According to this property the Pauli matrices generate a 3dimensional Clifford algebra C3 . σk ]+ = 2δjk 1 . z A = In fact. e.k=1 j<k + σ j σ k aj bk .. (5.225) σ (5. one obtains (σ · a) (σ · b) = 3 j 2 j=1 (σ ) aj bj 3 j k j.223) where σk . namely. The proof of this relation rests on the commutation relationship and the anti-commutation relationship (5. ∈ R . σk ]− = 2i jk z x − iy x + iy −z .226) The Pauli matrices satisfy very special commutation and anti-commutation relationships. A = x σ 1 + y σ 2 + z σ3 .224) Algebraic Properties of the Pauli Matrices The Pauli matrices (5.224) provide a basis in terms of which all traceless. x.e. the simplest choice being S1 = 1 σ1 .228) (5. it holds.k=1 σ σ aj bk + j>k aj bk = 3 j.230) where a. (5. k = 1. y. At this point we will state a useful property. σ2 = 0 −i i 0 . hermitian 2 × 2–matrices A can be expandend.e. it might hold aj bk − bk aj = 0.31) of the group SO(3). 1 i. 5. 2 S2 = 1 σ2 . σ3 = 1 0 0 −1 .227) which is essentially identical to the Lie algebra (5. The Pauli matrices obey the following anti-commutation properties σj σk + σk σj = [σj . Any such A can be expressed in terms of three real parameters x. σj 2 (5. In fact. z. (5.227.5. We will show below that a 2-1–homomorphic mapping exists between SU(2) and SO(3) which establishes the close relationship between the two groups. i.231) .g. Clifford algebras play an important role in the mathematical structure of physics.229) and avoids commuting the components aj and bk .

however. 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. however. (5. We like to derive this result now by evaluating the transformations of SU(2) explicitly. to rotations in space. Well almost like it. For this purpose we choose to replace (5. through the algebra obeyed by the operators Sk [Sk . hence.k=1 j>k σ j σ k (aj bk − ak bj ) − 1 2 3 j k j. spin changes sign. S3 χ± = ± 1 χ± 2 .e. yields a · b. let say around the x2 –axis by 360o . i.k=1 j>k σ j σ k (aj bk − ak bj ) = 1 2 3 j. In the special case a = b ∈ R holds (σ · a)2 = a 2 1 . 1 (5.k=1 (σ σ j>k σ k σ j (aj bk − ak bj ) = (5.238) S 2 χ± = . (5. This relationship emerges. The commutation property (5.237) + 1)1 and.s. (5. only a 720o rotation leaves the spin invariant. The two remaining terms yield using σ j σ k = −σ k σ j for j = k and altering ‘dummy’ summation indices 3 j.231) proves (5.k=1 j>k i σl (aj bk − ak bj ) = i =1 σ a×b .232) − σ k σ j ) (aj bk − ak bj ) .222) by the Euler form exp(−iγS3 ) exp(−iβS2 ) exp(−iαS3 ) . (5.230).222).128 Theory of Angular Momentum and Spin Using (σ j )2 = 1 the first term on the r.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 . 1 1 1 ( + 1) χ± 2 2 .h.236) The matrix elements of this 2 × 2 operator can be labelled by the basis states χ+ and χ− . since there is a slight difference: when you interprete the 3-component ϑ in (5.233) This result together with (5. S ] = i k m Sm (5. Noticing the idempotence of the Sk ’s 2 Sk = 3 one obtains S 2 = 4 1 = 1 1 1 2(2 1 4 1 0 0 1 (5.234) 5. we like to draw in this respect also on a close analogy to angular momentum states which developes 2 2 2 if one considers the operators S 2 = S1 + S2 + S3 and S3 .235) which is identical to that of the generators of SO(3).227) leads to 3 3 jk j.k=1 j>k 1 2 3 j.

We may. − + 1. namely. .245) 5.219).5. i. i. 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. 1 .239) where we made use of the notation (5. 3 .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). β. . 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 .e.9 Constructing Spin States with Larger Quantum Numbers Through Spinor Operators In this section we like to demonstrate. The complete matrix elements (5.240) . (5. 2.243) We note the property of this rotation matrix 2 dmm (2π) (1) = −1 1 . .236) 1 2m | exp(−iγS3 ) exp(−iβS2 ) exp(−iαS3 ) | 1 m 2 1 2m = e−iγm | exp(−iβS2 ) | 1 m e−iαm 2 (1) 2 = e−iγm dmm (β)e−iαm (5.242) The expressions in brackets [. .219.e. .. 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.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. γ)] m.244) i. . In order to determine the Wigner matrix in (5. use the label 2 2 | 1 ± 1 for the states χ± . therefore. .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 . For this purpose one needs to determine the powers of −iβS2 . no two such particles can exist in the same state. identical particles any number of which can exist in the same state χ+ and χ− . 2 2 We obtain with this notation for the transformations (5. following Jourdan and Schwinger. 1. (5. (5.220) are [D(α.239 ) one expands the exponential operator exp(−iβS2 ).239) in the notation (5. m = − . The idempotence property of Sk yields particularly simple expressions for these powers.5.e. rotation by 360o changes the sign of the spin state. . . ( − 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.] can be identified with the Taylor expansion of the cos. (5. that states | m for higher quantum numbers = 0.

−) bζ . b† |Ψ0 = χ+ + . 2 + 2 − † (5. We consider. b† )T and b = (b+ .130 Theory of Angular Momentum and Spin Definition of Spinor Creation and Annihilation Operators We present the states χ± through creation operators b† which when applied to a formal vacuum ± state |Ψ0 generate χ± . b † ζ ζ =0 (5. b± |Ψ0 = 0 (5.253) = δζζ .250) which for x∗ x = x∗ x+ + x∗ x− = 1 represent spinor operators in an arbitrary reference system. ζ = +. j = 0. (5. . . . m) † and bζ obey the commutation (5. . b† |Ψ0 = χ− − . i.247. 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. x∗ b = x∗ b+ + x∗ b− + − (5.254) We will show below that these states are orthonormal and form spin states with higher quantum numbers.251) One can show using this property and (5. 1. m = −j. + − For example. b ζ The States |Ψ(j. i. b† )T and b = (b+ . b ζ = b ζ. also operators of the type x b† = x+ b† + x− b† + − . 1 . using (5.248) bζ . m) = b† − j−m (j + m)! (j − m)! |Ψ0 . These operators are associated with a given spatial reference system.246) The corresponding adjoint operators are denoted by b+ and b− . b− )T is expressed through the commutation relationships + − (ζ. 3 . b− † = − sin β † β b + cos b† .248) that bζ relationships † † b ζ.e.5. b− )T as spinor creation and annihilation + − operators. . b † ζ For the vacuum state |Ψ0 holds = δζζ . therefore.247) (5. (5. −j + 1.252) (5. 2 2 .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 − .249) In the following will refer to b† = (b† . j. . b ζ = 0 b ζ.e. b ζ = b† . The boson character of the operators b† = (b† . .

248) we obtain b+ .249) b+ f (b† ) |Ψ0 = + b+ .260) . 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.257) motivate us to determine the commutator (5. b† + = n b† + n−1 .10: Algebraic Properties of Spinor Operators 131 5. We will derive the result. 5. f (b† ) + (5.249). b† + − well-known for the quantum mechanical harmonic oscillator. (5. We will show that for this commutator holds n b+ .255) In order to determine how this operator is modified by multiplication from the left by b+ we note b+ f (b† ) |state = + b+ . in b† and b† . b† + b+ . For this purpose we consider the following polynomial of creation operators N j+m j−m f (b† ) = + n=0 cn b† + n . (5. An example + − is the monomial b† which generates the states (5. we need to evaluate b+ . + b+ . b† + n = n cn b+ . b† + n+1 = = b+ .5.10 Algebraic Properties of Spinor Operators We want to establish first a few important and useful algebraic properties which result from the commutation relationships (5. b† + n . possibly infinite power series.257) which for (5. f (b† ) |state + f (b† ) b+ |state . (5.248) and from (5.259) The property is obviously true for n = 0.256) and (5.258) Equations (5.247. f b† Obviously. f (b† ) |Ψ0 . + + (5. b† + n b† + b† + b† + + n n n b+ . that the annihilation operators play a role similar to the differential operator in calculus.254). If it is true for n then using (5. .255) is b+ . b † + = n b† + n−1 b† + b † + + n = (n + 1) b† + .256) In the special case |state = |Ψ0 this reads using (5.

b† ) |Ψ0 = + − ∂ ∂b† ζ f (b† .262) Because of the commutation relationships (5.267) exp xb† |Ψ0 is called a generating function of |Ψ(j. + − (5. f (b† ) − = f (b† ) − (5. b† ) |Ψ0 . f (x)] = ∂k f (x) . b† ) + − = ∂ ∂b† ζ f (b† .263) According to (5..261) Similarly.266) we compare the terms x+ b† + j+m φjm (x) Ψ (j.. f (b† . (5.132 Theory of Angular Momentum and Spin i.e. the property holds also for n + 1. (5.. By induction we can conclude that (5.264) This demonstrates the equivalence of the spinor operators bζ and the derivative operation.250) and where φjm (x) represents the function of the two variables x+ and x− closely related to |Ψ(j. b† ) . 2 . + − (5.255) b+ .247) the more general property for polynomials f (b† . m) φjm (x) = j−m xj+m x− + (j + m)! (j − m)! . In order to derive (5. One can finally conclude for polynomials (5. φjm (x) Ψ (j. m) We want to prove now the property ∞ j exp xb † |Ψ0 = 1 j=0. f (x)] g(x) = (∂k f (x))g(x) or [∂k . Equation (5.266) where xb† has been defined in (5. (5. = f (b† ) + (5. f (b† ) + where f (x) = df dx .265) Generating Function of the States |Ψ(j. one can prove b− . b† ) in b† and b† holds (ζ = +.259) holds for all n ∈ N.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 . −) + − + − bζ . m) . m) m=−j (5.268) .257) we can conclude in particular bζ f (b† . m) = x− b† − j−m (j + m)! (j − m)! |Ψ0 .1.

s! (n − s)! Defining j+m = N −s j−m = s.270) s x+ b† + N −s x− b† − s |Ψ0 .1. 1 xb† u! |Ψ0 (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. .. chosing the summation index j = 0. 1.269) (5. m) = = |Ψ0 (N − s)! s! N −s 1 N! x+ b† x− b† + − N ! (N − s)!s! N −s 133 (5.273) Change of the summation indices allows one to conclude j φjm (x) Ψ (j... . 2. (2j)! 2j |Ψ0 (5.270) N 2j j → s=0 j−m=0 → m=−j . . m=−j = j=0. .. 2 2 leads to ∞ j ∞ φjm (x) |Ψ (j. 1 . . (5..1. (5.274) Summing this expression over 2j = 0. 3 .272) The summation over s can be written in terms of j and m using (5.. 2 ∞ 1 xb† 2j! u 2j |Ψ0 = u=0.e. 1 .10: Algebraic Properties of Spinor Operators with the s–th term in the binomial expansion of (a + b)n n! an−s bs .275) = exp xb† |Ψ0 . 2 . i.1. m) m=−j = = 1 x+ b† + x− b† + − (2j)! 2j 1 xb† |Ψ0 .. . m) 1 j=0. 1..5.. one obtains φjm (x) Ψ (j... 2.

(5. b− )e(yb ) |Ψ0 = f (y+ . 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 (5.279) One can generalize this to † † s (bζ )s e(yb ) |Ψ0 = yζ e(yb ) |Ψ0 .264) implies † † bζ e(yb ) |Ψ0 = yζ e(yb ) |Ψ0 .254) are orthonormal. For this purpose we consider the inner product † † e(xb ) Ψ0 |e(yb ) Ψ0 = δjj δmm (5.134 Theory of Angular Momentum and Spin which concludes our derivation. m . Orthonormality of the States |Ψ (j. hence. In order to evaluate the ∗ b) (yb† ) operator e(x e we notice that the derivative property (5.284) . b† ] = 0 allow one to state that for + − + any polynomial f (b+ .277) by Ψ0 |e(x b) e(yb ) Ψ0 . m) We want to show now that the states (5.281) and.m φjm (x∗ ) φj m (y) Ψ (j. we can conclude † † ∗ e(xb ) Ψ0 |e(yb ) Ψ0 = e(x y) (5. therefore. m) and will be used for this purpose below. b† ] = 0 and [b− .280) The commutation properties [b+ .e. y− )e(yb ) |Ψ0 (5. m holds. that Ψ (j. m) |Ψ j . b− ] = 0. b− ) holds † † f (b+ . of (5. The generating function allows one to derive various properties of the states |Ψ (j.283) (5.s. According to (5.278) † ∗ which allows us to replace the l. (5. [b† . (5.277) To evaluate this expression we first notice † e(xb ) † = e(x ∗ b) . i. (5.m.h.249) for any non-vanishing integer s holds + − Ψ0 |bs Ψ0 = 0 and. m) |Ψ j .276) = j.282) where we defined x∗ y = x∗ y+ + x∗ y− .j .

k = 1. 3 .m ijk Jk .m.n .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 5.285) Following steps similar to those in Eqs. In our present notation the matrix elements < ζ |σk | ζ > for ζ = ± correspond to the matrix elements < ζ |σk | ζ > for σ = ± 1 .282) and (5.224). b† bm n n n.m . One defines corresponding operators Jk through 1 Jk = b† < ζ |σk | ζ > bζ .n . (5.289) < n |σi | m >< n |σj | m > {b† bm .10: Algebraic Properties of Spinor Operators and comparing (5. 3 denote the Pauli spin matrices (5.287) ζ 2 ζ. (5.268–5. (5.j . 2 . bm bm } n n = 1 4 < n |σi | m >< n |σj | m > {b† bm δmn − b† bm δm n } n n n. 1 New Representation of Spin 0.276).288) The three operators obey the Lie algebra of SU(2) [Ji . Jj ] = i This property is derived as follows [Ji . . 1.m. m j. The 2 operators are explicitly.m +b† b† . . < + |σ1 | − > = < − |σ1 | + > = 1. m) as given in (5. using < + |σ1 | + > = < − |σ1 | − > = 0.286) from which follows immediately the orthonormality property (5.275) one can show e(x ∗ y) = j..m.223. (5.277) φjm (x∗ ) φj m (y) Ψ (j. − 2 + (5.m φjm (x∗ ) φjm (y) . 5. 2.m 135 = e(x ∗ y) . (5. b† bm n n n.ζ where σk .n .224).5. etc. Jj ] = = 1 4 1 4 < n |σi | m >< n |σj | m > b† bm . m) |Ψ j . . States 2 We want to demonstrate now that the states |Ψ (j. J1 = J2 = J3 = 1 † b b− + b† b+ − 2 + 1 † b b− − b† b+ − 2i + 1 † b b+ − b† b− .m.

(5.291) Here we have defined J+ J− = J1 + iJ2 = J1 − iJ2 = b† b− + = b† b+ . J1 ] + i [J1 .m < n |[σi . J2 ] = −2 J3 from which we can conclude the property 1 1 J− J+ = J+ J− − J3 2 2 Hence. (5.297) . m) as Eigenstates of J 2 and J3 We wish to show now that the states |Ψ (j.289) yield [J− . (5. (5. σj ]| m > b† bm n n. 2 J 2 = J+ J− + J3 − J3 = J3 (J3 − 1) + J+ J− (5.m.294) . To this end we note 2 2 2 J 2 = J1 + J2 + J3 = 1 1 2 J+ J− + J− J+ + J3 2 2 .k ijk ijk σk | m > b † bm n = i = i 1 2 < n |σk | m > b† bm n n. J1 + iJ2 ] = −i [J2 .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 .m 1 4 < n |σj | m >< m |σi | m > b† bm n n .292) The commutation relationships (5.293) . − (5. m) are eigenstates of J 2 and J3 . (5.295) The last result together with (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.292).m.290) |Ψ (j. J+ ] = [J1 − iJ2 .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.m ijk Jk .m < n |2i n.

304) (5. −λ + 1. 2 . i. . I3 ||λ. . 1. (b) Show that the three operators 1 + 1 1 + (b1 b2 + b+ b1 ) . the states (5.298) (5. 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† .301) Obviously.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) . [Ij . m) J3 |Ψ (j. I2 = (b1 b2 − b+ b1 ) ..e. 1 2. m = λ||λ.5. n2 . 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. . I3 = (b+ b1 + b+ b2 ) 2 2 2 2 2i 2 1 satisfy the Lie algebra of SU(2). b † + b + + b † + b+ b † − b − 1 1 b† + b + + b † + b + b † − b− + b † − b − b † + b + . λ. m) = = (j + m + 1) (j − m) |Ψ (j. ˆ respectively.s. Show λ = n1 +n2 n1 +n2 2 ( 2 + 1) and m = 1 2 (n1 − n2 ). . n = n1 + n2 fixed. n} eigenstates of I 2 = I1 + I2 + I3 and of I3 I 2 ||λ. .10: Algebraic Properties of Spinor Operators The last term on the r.e.305) = = j ( j + 1 ) |Ψ (j. b+ ] = δjk . 1 2 k (a) Show that the eigenstates are given by 1 |n1 . m . .299) (5. m = m||λ. m I1 = where λ = 0. m + 1) (j + m) (j − m + 1)|Ψ (j. (5. m − 1) . m) One can furthermore derive readily J+ |Ψ (j. i. m) . . .10. . m) J− |Ψ (j. . . State the corresponding eigenvalues and the degree of degeneracy. b† . 2 2 137 (5. One can then conclude J 2 |Ψ (j. [bj . using operators I± = I1 ± iI2 2 2 2 and the subspace {|n1 .h. m = −λ.302) (5. m) m |Ψ (j. j = 1.254) are eigenstates of + b+ and − b− with eigenvalues j + m and j − m. . Ik ] = i jk I .303) Exercise 5. (5. . the number of states to the possible energy eigenvalues. 2. n1 = 0. and eigenstates of k with eigenvalues j. Construct.300) (5.

The combination of σ. m) . m = m=−j dm m (β) |Ψ (j. On the other side the states |Ψ (j.243).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. both conditions are satisfied for σ = j + m − σ.306. The prefactor of (b† )j+m (b† )j−m which according to (5. For this purpose we expand + − + − b† + cos β b† + sin β b† 2 + 2 − j+m σ =0 j−m σ=0 j+m j+m b† − j−m = j−m −sin β b† + cos β b† 2 + 2 − cos β 2 j−m −σ σ = j+m σ σ sin β 2 σ +σ j+m −σ j−m σ −sin β 2 cos β 2 b† + b† − 2j−σ −σ (5.307) shows that the elements of the rotation matrix can be obtained by binomial expansion of b † and b † in terms of b† and b† . for example. In case j = 1 this expression yields.306) and (5. . σ values which yields (b† )j+m (b† )j−m is + − then σ = j + m − σ. However.308) The latter sum involves terms (b† )j+m (b† )j−m for σ + σ = j + m and 2j − σ − σ = j − m.310) it reduces to (5. m = b† + j+m b† − j−m (j + m )! (j − m )! |Ψ0 (5.138 Theory of Angular Momentum and Spin 5. m ) are related to the + − states |Ψ (j. + − One may expect that these two conditions restrict both σ and σ.  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. For this purpose we note that the states |Ψ(jm) in a rotated coordinate system according m to (5.254) are |Ψ j.309) .306) where b † and b † are given by (5. (j) (5.5. m) in the original coordinate system by j |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).307) Comparision of (5.

x3 ) ∈ R .308) and had the form (5. e.5.e. . U =  0 1 1 i √ ( x1 − ix2 ) √ x3 − √2 0 2 2 (5. that the the representation (5.314) leaves (5. The SU(2) transformation entered in (5.. namely that the component of the vector on the l.312) The choice of this transformation derives from a property shown further below.309) does not distinguish between SU(2) transformations (5.309) and the particular matrix (5. however. i.h.314) in case of integer j–values. It is of interest to trace the mapping from SU(2) onto SO(3) as represented by (5. Replacing the latter transformation by its negative form. b+ † = −cos β † β b+ − sin b† 2 2 − .313) × 0 0 2  =  0 sinβ 0 cosβ 1 −i 0 which.. 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. conclude that the mapping from SU(2) onto SO(3) is a 2–1 mapping. Hence.310).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  . in case of j = 1.12 Mapping of SU(2) onto SO(3) The representation (5.12: Mapping of SU(2) onto SO(3) 139 5. This factor implies. in fact. is the expected element of SO(3). One can.310) establishes a mapping of SU(2) onto SO(3). 2 2 − (5.312) transforms like an angular momentum state ˜ |1 m . the matrix A should represent a rotation around the x2 –axis in the space of vectors ˜ (x1 .309).s.e. b− † = sin β † β b+ − cos b† . i.251). x2 . the orthogonal 3 × 3 matrix which describes a rotation around the x2 –axis.251) and (5.g.311) (5.   = U  x2  .308) unaltered except for a factor (−1)2j which multiplies then also the final result (5. This can be shown by (1) applying to the matrix A = (dm m ) in (5.310) to the SU(2) transformation assumed in deriving the general result (5. of (5. therefore.

140 Theory of Angular Momentum and Spin .

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

max d(B) = 1 =0 (2 1 + 1) 2 =0 (2 2 + 1) = ( 1. n = 1. the dimensions d(Bk ) of these subspaces does not exceed 100. that only states within the subspaces Bk are coupled in the scattering process. composite systems behaving like elementary objects are common. = 0. . . . − 2 ≤ m1 ≤ (6. The rotational symmetry of the interaction between the particles allows one. .m1 2 . and by the moment of inertia IA−B of the diatomic molecule A–B approximately as follows 1. . rather than by Classical Mechanics.e. (6.max + 1)2 ( 2.m1 . mA mB + mB mC + mA mC 2.1) where 1.max = 10 one obtains d(B) = 14 641. however. 1. . .2) The dimension d(B) of B is 1. 2 . the values of which are determined by the size of the interaction domain ∆V . since further important degrees of freedom. each of which stands for two angles. .max = 1 2 IA−B E . by the masses mA . as we will demonstrate below.max . .m2 Y 1 m1 (ˆAB ) Y 2 m2 (ˆC ). Such large number of dynamically coupled states would constitute a serious problem in any detailed description of the scattering process.max denote the largest orbital and rotational angular momentum values. 14 641} .max 2. Such extremely useful transformation of the problem can be achieved through the choice of a new basis set B = { 1 . in fact. i. 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. 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. and the following mathematical description will shed light on their ubiquitous appearence in physics.3) For rather moderate values 1. mB . In fact.max = ∆ 2 mA mB mC E . to separate the 14 641 dimensional space of rotational states Y 1 m1 (ˆAB ) Y 2 m2 (ˆC ) into r ρ subspaces Bk . 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 ). .m2 c (n) r ρ 1. ρ 1 = 0. have not even be considered. mC .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. . = B such. albeit puzzling. 2. 2.max and 2. by the total energy E. (6. will make their appearence a natural consequence of the symmetry of the building blocks of matter. vibrations and rearrangement of the particles in reactions like AB + C → A + BC.. 1. Obviously. .max = 2. .max + 1)2 (6. B1 ⊕ B2 ⊕ .max . a very large number.142 Addition of Angular Momenta and Spin ρC . in particular. The latter often arrives at the physical properties of composite systems by adding the . − 1 ≤ m1 ≤ 1. 2} 1.

5) do not affect the Hamiltonian. One can equivalently express therefore (6.6. 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. the commutation relationships (1) Jp . in fact. R−1 (ϑ) ρC ) (6. 5. q = 1. connected with the fact that the Hamiltonian H does not depend on the overall orientation of the three interacting particles. the existence of a basis (6. 3 (6. the reader will learn in the following section how to add and subtract in the microscopic world of Quantum Physics. 2.7) i R(ϑ) = exp − ϑ · J (6. For ˆ ˆ example. but solely under simultaneous and identical rotations of rAB or ρC .0: Addition of Angular Momenta and Spin 143 corresponding physical properties of the elementary components. Jq(2) = 0 for p. ρC ) defined through R(ϑ) ψ(rAB . In this sense.. (6.55) holds i (1) − J1 = zAB ∂ ∂ − yAB . ρC ) it follows H R(ϑ) = R(ϑ) H. ∂ρx ∂ρy (6. i. Following our description of rotations of single particle wave functions we express (6. assuming presently that H might.6) For the following it is essential to note that H is not invariant under rotations of only rAB or ρC . ρC ) = ψ(R−1 (ϑ) rAB .8) Obviously.7) where the generators J (k) are differential operators acting on rAB (k = 1) and on ρC (k = 2).10) .48) i i R(ϑ) = exp − ϑ · J (1) exp − ϑ · J (2) (6. from which follows in turn the well-known result that H is related to H through the similarity transformation H = R(ϑ) H R−1 (ϑ).e. according to (5. It holds then H R(ϑ) ψ = R(ϑ) H ψ.53. ∂yAB ∂zAB i (2) ∂ ∂ − J3 = ρy − ρx . The invariance of the Hamiltonian under overall rotations of the three particle system implies then H = R(ϑ) H R−1 (ϑ) .4) which decouples rotational states is connected with the rotational symmetry of the Hamiltonian of the three particle system considered. presumably a facility the reader would like to acquire with great eagerness. 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. be different from H.9) hold since the components of J (k) are differential operators with respect to different variables. Hence.5) according to (5. Rotational Symmetry of the Hamiltonian As pointed out already. To specify this property mathematically let us denote by H the Hamiltonian in the rotated frame. Since this is true for any ψ(rAB . rotations R(ϑ) of the wave functions ψ(rAB .

2. For a proof one uses (6.81)]. k = 1. (5. (6.71–5. r ρ Definition of Total Angular Momentum States The commutation property (6.144 where J = J (1) + J (2) Addition of Angular Momenta and Spin . and the eigenvalues of J2 and J3 are good quantum numbers.11) we can write the condition (6. (6.71–5.81) we want to consider the relationship between YJM and Y 1 m1 (ˆAB ) Y 2 m2 (ˆC ).15) are satisfied. J ] = i k m Jm (6. (6. In the following we will use the notation r ρ Ω1 = rAB . Jk ] = 0 . We suspect.14) implies that the components of the total angular momentum operator (6. i. however. that energy. B + C] = [A. The property (6.f. [H..11) By means of (6.16) following the procedure stated in the theorem above [c. (6. 3 do not commute. the (n) (n) (n) properties [Jk . the important fact that the Jk obey the Lie algebra of SO(3). k = 1. the operators Jk .e.13) Hϑ · J − Since this holds for any ϑ it follows HJ − J H = 0 or. C]. 2. 3 as the three components of the total angular momentum operator. We recognize. 2 together with the property [A. i. Eqs. Before we apply the procedure (5.e. we will find that the states yield the basis B with the desired property of a maximal uncoupling of rotational states. for |ϑ| H ≈ 1 − 1 i ϑ·J H 1 + 1 i ϑ·J ≈ H + i 1. 2. that the components Jk . k = 1. One can show readily that the commutation relationships [Jk ..14) We will refer in the following to Jk .12) each individually can have simultaneous eigenstates with the Hamiltonian. a supposition which can be tested through the commutation properties of these operators. 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. componentwise. of course. 3 cannot have simultaneous eigenstates among each other. To order O(|ϑ|) one obtains i ϑ · JH .e. 3 .14) implies that the total angular momentum is conserved during the scattering process.6) for rotational invariance of the Hamiltonian in the form i i H = exp − ϑ · J H exp + ϑ · J . ˆ Ω 2 = ρC . k = 1. 2. J ] = i k m Jm for n = 1. In fact.9).12) We consider this equation for infinitesimal rotations. 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 ). i. ˆ (6. B] + [A.17) .

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

25) i.27) together with (6. . . (6. . 2 ) has (2 1 + 1)(2 2 + 1) elements.h. |J (1) | + |J (2) |]. M = −J. J} . Ω2 ) = δJJ δM M (6. 2 ) and B ( 1 . −J + 1. in fact. the basis B ( 1 .h. m2 = − 2 . to a new basis B ( 1 . − + 1.18) constitutes a change of an orthonormal basis B( 1 . .2: Prove Eq.28) This property follows from the fact that the basis elements are eigenstates of hermitian operators with different eigenvalues. .23) corresponding to the r.26) We will show below in an explicit construction of the Clebsch-Gordan coefficients that. .. 1 + 2. J.1. 2) . it is not immediately obvious what the J– values are. . This obviously corresponds quantum mechanically to a range of J–values J = | 1 − 2 |.s. Exercise 6. (6. Our derivation below will also yield real values for the Clebsch-Gordan coefficients. For each quantum number J there should be 2J + 1 elements YJM with M = −J. . .m2 2 M and with (6. 2) = {Y 1 m1 (Ω1 ) Y 2 2 m2 (Ω2 ). . 2 |Ω1 . The property dΩ1 dΩ2 Y∗ ( 1 .146 Theory of Angular Momentum and Spin The expansion (6. (6. the range of values assumed for J is correct. . = {YJM ( 1 . . .24) yields 1 m1 2 m2 ) ∗ 1 m1 2 m2 ) (J M| = δJJ δM M . However. The basis B ( 1 . . | 1 − 2 | + 1. | 1 − 2| + 1. In this case the range of |J|–values would be the interval [ |J (1) | − |J (2) | . (6. (6. B2 2) should be 2 |Ω1 .. 2 ). J = | 1 − 2 |. The orthonormality (Ω1 ) Y 2 m2 (Ω2 )Y 1 m1 Ω1 ) Y 2 m2 (Ω2 ) = δ 1 1 δm1 m1 δ 2 2 δm2 m2 . 1.18) applied to Y∗ and to YJ JM (J M | m1 . and that the states can be normalized. JM 2 |Ω1 . In fact. . 2 ) is also orthonormal2 and must have the same number of elements. . m1 = − 1 . Since YJM represents the total angular momentum state and Y 1 m1 (Ω1 ) and Y 2 m2 (Ω2 ) the individual angular momenta one may start from one’s classical notion that these states represent angular momentum vectors J. Ω2 ) YJ M ( 1. − 2} 1 + 1. .25) We want to state now two summation conditions which follow from the orthonormality of the two basis sets B( 1 . 1 + 2 .24) The basis B( 1 . . J (1) and J (2) . −J + 1. .e. Ω2 ). property implies dΩ1 dΩ2 Y 1 m1 corresponding to the l. it holds 1+ 2 ( 2J + 1 ) = (2 J=| 1 − 2 | 1 + 1) (2 2 + 1) . respectively. (6.s. .

. 2 |Ω1 . or baryons multiplets involving three quarks. 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 ( 1.28) from (6.g. (6. 2 ) and B ( 1 . Ω2 ).28) allows one to conclude that the coefficients cJ (J M | 1 m1 2 m2 )∗ .71–5. therefore. A similar 2 construction will also be applied to composite systems governed by other symmetry groups.e. 2 |Ω1 . 2 |Ω1 . the group SU(3) in case of meson multiplets involving two quarks. 2 ) span the same function space.33) JM The latter summation has not been restricted explicitly to allowed J–values. i.29) where the expansion coefficients are given by the respective scalar products in function space cJ M = dΩ1 dΩ2 Y∗ J M ( 1. We can. At this point we like to stress that the construction will be based on the theorems (5.34) 6.32) which complements (6.m2 M ( 1. Ω2 ) and integrating.18).81) stated above. i. The result of this construction will include all the properties previewed above.30) The latter property follows from multiplying (6. 1 m1 2 m2 ) = 0 if J < | 1 − 2| .27) yields δJJ δM M = m1 .. i. will be based solely on the commutation properties of the operators J and J (k) . The (J M | 1 m1 2 m2 ) cJ M . it is possible to expand Y 1 m1 (Ω1 )Y 2 m2 (Ω2 ) in terms of YJM ( 1 . Hence. One can show readily using the same reasoning as applied in the derivation of (6.e..2: Construction of Clebsch-Gordan Coefficients 147 The second summation condition starts from the fact that the basis sets B( 1 . (6. (6. 2 |Ω1 . e. 2 |Ω1 ..e. and actually also the properties of Clebsch-Gordan coefficients stated above. (6. rather the convention (J M | has been assumed.31) are identical to Comparision with (6.18) that the Clebsch-Gordan coefficients obey the second summation condition (J M | 1 m1 2 m2 )∗ (J M | 1 m1 2 m2 ) = δm1 m1 δm2 m2 . Ω2 )Y 1 m1 (Ω1 )Y 2 m2 (Ω2 ) .. also apply the results.18) by Y∗ J orthogonality property (6. (6. Ω2 ) .6.2 Construction of Clebsch-Gordan Coefficients We will now construct the Clebsch-Gordan coefficients. or J > 1 + 2 (6. to composite systems involving spin. 1+ 2 J Y 1 m1 (Ω1 )Y 2 m2 (Ω2 ) = J =| 1 − 2 | M =−J cJ M YJ M ( 1.1 states. Ω2 ) .

Similarly. . (6. 2 |Ω1 . The construction assumes a particular choice of J ∈ {| such J–value seeks an expansion (6. 2 |Ω1 . 2 |Ω1 . We embark on the suggested construction for the choice J = 1 + 2 . −J. We first seek an unnormalized solution GJJ and later normalize. (6. 2 |Ω1 . 1+ 2 ( 1. 2 |Ω1 .36) = 0 = J YJJ ( 1 . . Ω2 ) = Y 1 1 (Ω1 ) Y 2 2 (Ω2 ) (6. (6.5.39) which we will try for a solution of (6. 1 + 2} and for (6. For this purpose we insert (6. .11) and (5. have shown Y 1+ 2. . Ω2 ) = Y 1 1 (Ω1 ) Y 2 2 (Ω2 ) .18) which satisfies J+ YJJ ( 1 . To find GJJ we start from the observation that GJJ represents the state with the largest possible quantum number J = 1 + 2 with the largest possible component M = 1 + 2 along the z–axis. 1+ 2.64) 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 ) (6.25) using (6. | 1 − 2| + 1. 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. 2 |Ω1 . Having determined such YJJ we then construct the whole family of functions XJ = {YJM ( 1 . 2 |Ω1 . J3 YJJ ( 1 .37) The solution needs to be normalized. −J + 1.68) J+ = (1) + J+ (1) (2) Y 1 1 (Ω1 ) Y 2 2 2 2 (Ω2 ) 1 1 (6. 1+ 2 ( 1. Ω2 ) (Ω1 ) dΩ2 Y 2 2 (Ω2 ) = 1. M = −J. . . 2 |Ω1 . Ω2 ).66. (6. 2 |Ω1 . Ω2 ) . 1+ 2 In fact. Ω2 ) is normalized dΩ2 G dΩ1 Y 1 1 1+ 2. Ω2 ) = (J + M + 1)(J − M )YJM ( 1 .42) ( 1.43) .11) J+ by J+ + J+ . Ω2 ) 2 |Ω1 . 1+ 2 ( 1. we can demonstrate condition (6.38) for M = J − 1. .148 For the construction of YJM we will need the operators J± = J1 + iJ2 . .39) into (6. .37) and replace (1) (2) according to (6. We obtain using (5. therefore.41) 2) Y (Ω1 ) Y (Ω2 ) . J} by applying repeatedly J− YJM +1 ( 1 .37). . Ω2 ) Addition of Angular Momenta and Spin (6. Ω2 ) .35) 1 − 2 |. we can also demonstrate using (??) that G dΩ1 = We. Quantum mechanically this corresponds to a state G 1+ 2. J − 2.40) (Ω1 ) J+ Y (2) 2 2 J+ Y 1 1 (Ω1 ) Y (Ω2 ) + Y (Ω2 ) = 0 .

s. M = −( + − ). . . Show that the resulting functions are orthonormal.2. in particular.38) to construct the family of functions B + = {Y + M ( . Normalization yields furthermore Y = 1 1 + 2 −1 1 + 2 −1 We have thereby determined ( 1 2 |Ω1 Ω2 ) (6. Y 1 1 −1 (Ω1 )Y (Ω2 ) + Y 1 1 This construction can be continued to obtain all 2( 1 + 2 ) + 1 elements of B + and. 2 = 1 in Table 1. 1 + 2 − 3. −( 1 + 2 − 1). ( + )}. One can obviously continue the construction outlined to determine Y 1 + 2 −2 1 + 2 −2 .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. We like to construct now the family of total angular momentum functions B + − = {Y + − M ( . This expression is orthogonal to (6. .1: Construct following the procedure above the three functions YJM ( 1 . . of (6. . 1 + 2 − 1. | . thereby. 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. .6. 6. ). 2 |Ω1 .h. | .39) as required by (6.45) for some unknown constant c. and all total angular momentum functions for a given choice of 1 and 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. We demonstrate the procedure ex(1) (1) plicitly only for M = 1 + 2 − 1. Having constructed this family we have determined the Clebsch-Gordan coefficients ( 1 + 2 − 1M | 1 m1 2 m2 ).47) (Ω2 ) − 1 1 2 + Y 2 1 1 −1 (Ω1 )Y 2 2 + Y 2 1 1 (Ω1 )Y 2 2 −1 (Ω2 ) . The reader should familiarize himself with the entries of the Table.s. Ω2 ) 2 1+ 2 = (Ω1 )Y 2 2 −1 (6. We seek for this purpose first an unnormalized solution G 1 + 2 −1 1 + 2 −1 of (6. . .37).46) 2 1 +c Y (Ω1 ) Y 2 2 (Ω2 ) = 0 . To satisfy this equation one needs to choose c = − 1 / 2 .27). . Expression (6. 2 |Ω1 .38) yields pression 2( 1 + 2 )Y 1 + 2 1 + 2 −1 ( 1 . ( + − )}.36. The r.2: Construction of Clebsch-Gordan Coefficients 149 We now employ property (6. Exercise 6. One obtains then Y 1 1+ 2 1 + 2 1 + 2 −1 2 2 ( 1. . According to the condition (6. To demonstrate (6. M = −( + ). The l. One can readily show that (6. Ω2 ). etc. all the Clebsch-Gordan coefficients ( 1 + 2 M | 1 m1 2 m2 ). The result is illustrated for the case 1 = 2. . G 1 + 2 −1 1 + 2 −1 in (6. ).45). of (6. 1 + 2 − 2.38) yields with J− = J− + J− the ex√ √ 2 1 Y 1 1 −1 (Ω1 )Y 2 2 (Ω2 ) + 2 2 Y 1 1 −1 (Ω1 )Y 2 2 −1 (Ω2 ). with the symmetry pattern and with the terms Y 1 m1 Y 2 m2 contributing to each YJ M . .47) can serve now to obtain recursively the elements of the family B + − for M = 1 + 2 − 2.44) (Ω2 ) . 2 |Ω1 . Ω2) for M = 1 + 2 − 2 and J = 1 + 2 .37) is satisfied.h. .

1|Ω1 .258199 −0.57735 0. Ω2) Y3−3 (2.707107 0.57735 −0.707107 0. 1|Ω1 .632456 −0. Ω2) Y10 (2. Ω2) Y22 (2. 1|Ω1 . 1|Ω1 . 1|Ω1 .547723 0.447214 0. 1|Ω1 . 1|Ω1 .150 Addition of Angular Momenta and Spin Y33 (2. 1|Ω1 . 1|Ω1 . Ω2) Y2−2 (2.316228 0. Ω2) Y1−1 (2.57735 0.707107 0.774597 − − − − 0.707107 0. .632456 0.816497 Y22 (Ω1 )Y1−1 (Ω2 ) 1 15 1 3 3 5 1 5 1 2 3 10 2 5 1 2 1 10 2 3 1 3 Y20 (Ω1 )Y11 (Ω2 ) Y21 (Ω1 )Y10 (Ω2 ) 0. Ω2) Y30 (2. 1|Ω1 .816497 0. Ω2) Y3−2 (2.408248 −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) Y3−1 (2.447214 −0.774597 0 2 5 8 15 1 6 3 10 1 3 0.547723 0.730297 −0.316228 0. Ω2) Y20 (2. 1|Ω1 .408248 −0.632456 0.57735 0. 1|Ω1 .547723 0. 1|Ω1 . Ω2) Y32 (2. Ω2) Y31 (2.774597 0. Ω2) Y11 (2.258199 0.816497 Y2−2 (Ω1 )Y11 (Ω2 ) Y2−1 (Ω1 )Y10 (Ω2 ) Y20 (Ω1 )Y1−1 (Ω2 ) Y2−2 (Ω1 )Y10 (Ω2 ) − 2 3 Y2−1 (Ω1 )Y1−1 (Ω2 ) Y2−2 (Ω1 )Y1−1 (Ω2 ) 1 Table 6. Ω2) Y21 (2. 1|Ω1 . 1|Ω1 .730297 0.547723 0.57735 0.57735 3 10 3 5 Y2−1 (Ω1 )Y11 (Ω2 ) Y20 (Ω1 )Y10 (Ω2 ) Y21 (Ω1 )Y1−1 (Ω2 ) −0. Ω2) Y2−1 (2.816497 −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.

+ 2  − J−M +1 for = J + 1 2J+2 2  J−M  for = J − 1 2J 2 = . 5. − 1 2 2 2 J+M +1  for = J + 1 2J+2 2 The construction described provides a very cumbersome route to the analytical and numerical values of the Clebsch-Gordan coefficients. we extent definition (5. aζ .9. b † ζ = δζζ = δζζ . (5. M | .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 . 6.287) to two particles (1) J k k = = (2) J 1 2 1 2 ζ. 1 (6.2. bζ = = a† .10. It is actually possible to state explicit expressions for any single coefficient (JM | 1 m1 2 m2 ). 1 . a† ζ ζ b† . For this purpsose we will employ the spinor operators introduced in Sections 5. (6.10 [cf. b† refer to different particles and. These expressions will be derived now. b† ζ ζ = aζ . bζ = a† .50) (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. ζ = ± and the matrix elements < ζ |σk | ζ > are as defined in Section 5.48) (6. Definition of Spinor Operators for Two Particles In contrast to Sections 5. hence.3: Explicit Expressions of CG–Coefficients 151 Exercise 6. J.51) The operators aζ . bζ .ζ ζ. commute with each other ζ ζ aζ . M | . m − 2 .49) where ζ. m + 1 .54) 2 + = √ b† ( 2 +m2 )! √− ( 2 −m2 )! . we are dealing now with two particles carrying angular momentum or spin. b† ζ = 0. a† and bζ . a† ζ .52) and | 2 m2 2 According to Section 5.6. aζ bζ .3 Explicit Expression for the Clebsch–Gordan Coefficients We want to establish in this Section an explicit expression for the Clebsch–Gordan coefficients (JM | 1 m1 2 m2 ).10. The creation and annihilation operators are again of the boson type with commutation properties aζ .53) (6.ζ a† < ζ |σk | ζ > aζ ζ b† < ζ |σk | ζ > bζ ζ (6. 2 . b† ζ ζ = 0 = 0 . (6.9.10 where we studied single particle angular momentum and spin. Accordingly. 5.

(1) J (2) J 3 | 1 m1 1 3 | 2 m2 2 = m1 | 1 m1 = m2 | 2 m2 1 2 (6. can be expressed in terms of the number operators 1 † 1 ˆ ˆ a† a+ + a† a− . which describe a two particle system according to (6.65) (6. 6. 5. M ( 1 . i. M ( 1 . J3 |J. M ( 1 . m2 |J. M ( 1 . 3 . ˆ kj |J. = = 1( 1 2( 2 2) 2) (6. M ( 1 . m2 1 2 ( 1 .69) − 2| ≤ J ≤ + 2 . 2 . hence. 2. M ( 1 . ˆ For the operators kj holds ˆ kj | j . (2) J 2 . (6.68) The states |J. (6.55) . for such states should hold J2 |J. (j) 2 ˆ ˆ J = kj ( kj + 1 ) . we like to recall for future reference that the operators (5. 2) 1. 1. m2 |J. 2 .53. 1 . (6. M ( 1 .54). M ( 1 .64) (6.62) . 2) 2) can be expressed in terms of Clebsch-Gordan coefficients (6. M ( 1 . m1 . 2) 2) (6.66) (6. M ( 1 . k2 = b b+ + b† b− k1 = + − − 2 2 + namely. m1 1 | 2 . (6.56) The states | 1 . m2 2 . as usual are denoted by their respective quantum numbers J. M ( 1 .58) Jk + (2) Jk .63) according to At this point. (1) J 2 | (2) J 2 | 1 m1 1 2 m2 2 Addition of Angular Momentum and Spin = = 1( 1 2( 2 + 1) | 1 m1 + 1) | 2 m2 1 2 . (2) J which. in analogy to Eqs. −J ≤ M ≤ J . mj j and. j = 1.m2 | 1 . M ( 2) 2) = J(J + 1) |J.59) We seek to determine states |J. = M |J. .61) (6. 2 ) which are simultaneous eigenstates of the operators 2 2 J2 . + 1) |J. 2 )). (1) J 2 |J. M ( 1 .18) as follows = | 1 m1 . The aim of the present Section is to determine closed expressions for the Clebsch–Gordan coefficents ( 1 .59) J± = J1 ± i J2 . k = 1. are | 1 . 2) .67) = j |J.302. |J. 2) We will also require below the raising and lowering operators associated with the total angular momentum operator (6. M ( (2) J 2 |J. .152 It holds. The operator of the total angular momentum/spin of the two particle system is J = with Cartesian components Jk = (1) J + J (6.303). m1 1 | 2 . m1 . J3 . M.57) b† + ( (2) 2 2 +m2 1 +m1 b† − ( 2 2 −m2 + m1 )! − m1 )! (1) + m2 )! − m2 )! |Ψo . (6.e. 2) + 1) |J. 2 )) . 2 . m2 a† + ( 1 2 = a† − ( 1 1 −m1 (6..300). mj j = j | j . m1 1 | 2 . 2 . (5.60) (6. (1) J 2 . M ( 1 . (1) J .

6. using (6. K † 1 a† a+ − a† a− + b† b+ − b† b− .51) one obtains ˆ k1 .71) (6.71–6. 2 4 = 0 = = 0.64. + − − + 153 (6.75) The relationships (6. For example. K† J3 . K † + kj . We note that.71–6. K † (kj + 1) 2 2 1ˆ † 1 ˆ kj K + K † (kj + 1) 2 2 1 ˆ 1 ˆ K † kj + kj . K † (j) J 2 . the relationships (6.72) (6. a† b† + − − + 2 + 2 − 1 1 − b† a† b+ .74) + 1 J− J+ + J2 . K † = 0. 6. 2 2 3 ˆ = K † kj + K † . K † + K † 2 2 3 † †ˆ K kj + K .3: Explicit Expressions of CG–Coefficients The Operator K † The following operator K † = a† b† − a† b† . Employing (6.288). b† − b† a† b− . a† b† − a† b† − + − − + 2 + 1 1 = a† a+ . a† b† − a† b† + − + − + − − + 2 1 1 = a† a+ . ˆ A similar calculation yields [k2 . due to J2 = K† 1 2 J+ J− 1 † K . 4 Using J3 = (1) J 3 + (2) J 3 . + − − + 2 2 = 1 † 2K .50. b† = 0 .73) can be readily proven.71) one can show K† = = = = = 1 ˆ ˆ kj (kj + 1). 6.6.74) imply 3 2 J2 .73. + − + − 2 2 − + = . a† b† − a† a− . K † 1 † a a+ + a† a− .70) will play a crucial role in the evaluation of the Clebsch-Gordan-Coefficients. (6. a† b† + a† a− . This operator obeys the following commutation relationships with the other pertinent angular momentum / spin operators ˆ kj . K † ] = (j) J 2 . expressing (k) J 3 through the creation and annihilation operators according to (5. and applying the relationships (6.73) yields J3 . j = 1. a† b† + + − − + 2 2 − 1 1 = a† b† − a† b† = K† . K † 2 1ˆ ˆ 1 ˆ ˆ kj kj + 1. j = 1. K † J± .73) (6. (6.65) and (6.

2) = = (j) 2 J .72) and (6. but for different 1 and 2 quantum numbers of the operators (1) J 2 and (2) J 2 . 6.75) ascertain that under the action of K † the states |J.2.77).. Action of K † on the states |J. j2 2 .73. j1 + j2 (j1 + n n . 2) is a state with quantum numbers 1 1 + 1 2 and = N |J. (6. namely. M ( 1 .78) a state which has been used already in the construction of Clebsch-Gordan coefficients in Section 6. M ( 1 + n . a† b− − a† b† + + + − + = 0. j1 1 |j2 . M ( 1. We consider now the case that (K † )n acts on the simplest total angular momentum / spin state. M ( 1 . j1 + j2 (j1 . One can generalize property (6.292) one can derive similarly J+ . 2 ) . K † = Addition of Angular Momentum and Spin a† a− + b† b− .154 Starting from (5. |j1 + j2 . j1 1 + . M ( 1 . j1 + j2 (j1 . 6. j + 2 2 2) However. 2) = N |J. 6. according to (6.e.79) n |j1 + j2 . To demonstrate that the resulting states are eigenstates of (1) J 2 and (2) J 2 we exploit (6. 2 2 1 + ) . M ( 1 . i. j2 + ) 2 2 n n = N (n. b† + − + + + − The property [J− . M ( 1 . M ( 1 .76) Here N is an unknown normalization constant. 2) We want to demonstrate now that the action of K † on the states |J. K † ] = 0 is demonstrated in an analoguous way. 2 (6. a† b† + b† a† b− . M ( 1 + . M ( 1 . Application of (K † )n to this state yields.63. K† + K † (j) J 2 |J. M ( 1 . this result implies that K † |J. 2 2 + n ) 2 (6. = − a† a− .77) where N is another normalization constant. The commutation properties (6. it holds K † |J. j2 ) . 1 2 + 2 . 2 ) remain eigenstates of J2 and J3 with the same quantum numbers.76) and state K† n |J. 2) 2) ˆ K † kj + = K† = j 3 † K + K † j ( j + 1) |J. 2 ) j + 4 1 3 + K † |J.67) (j) J 2 K † |J. on the state |j1 + j2 . 2 ) produces again total angular momentum eigenstates to the same J and M quantum numbers of J2 and J3 . j1 + ) K † 2 2 (6. M ( 1 . M ( 1 .62. 4 3 + j ( j + 1) |J. j2 ) = |j1 .

J( 1 . 2) = N( 1 + 2 − J. j1 + n . j1 . allow one to obtain the states |J. Strategy for Generating the States |J.6. 2 ) exploits the expression (6.74)] yields |J.105) of the spherical harmonics.57) and exploiting the fact that J− and K † commute [c. 2) Our construction of the states |J.86) b† + J+ 2 − 1 |Ψo .104. j2 as follows J = j1 + j2 which is equivalent to n j1 j2 Accodingly.81) = N( 1 + 2 1 2 − J. 1. Combining (6.82. M ) . M ( 1 . 6. in analogy to the construction (5. 2 = j2 + n 2 (6. (6. M ( 1 . 1 = j1 + n 2 .3: Explicit Expressions of CG–Coefficients 155 where we denoted the associated normalization constant by N (n. K† 1 2 1 + 2 −J × × . 2 ) for −J ≤ M ≤ J as follows |J. (6. (6. = (J + M )! (2J)! (J − M )! 1 2 2) (6. M ( 1 . J( 1 .84) with (6. of (6. The latter states. 2) 1 2 = ( 1 + 2 (2J + 1)! − J)! ( 1 + 2 + J + 1)! 1 2 . M ( 1 . For this purpose one needs to choose in (6.79) n. (6. 2) . j2 + n ).84) (6.h.80) = = = 1 1 2 2 − J 1 n − = (J + 2 2 n 1 − = (J + 2 2 + 1 1 − − 2) 2) . 2 ) . M ) (J + × K† 1 − 2 )! (J + 2 − 1 )! J+ 1 − 1 + 2 −J (J− )J−M a† + × 2 (6. 2 ) can be expressed through the r. 2 2 It is now important to notice that any state of the type |J. 1 .83) The derivation of this expression will be provided further below (see page 158 ff). M ) (J− )J−M |J.82) for |J.82) 1 × | (J + 2 1 × | (J + 2 1 (J + 2 1 − 1 ) . 2) 2) .79). (J + 2 − − 2) 1 1) 2 The normalization constant appearing here is actually N( + − J. 2) = ∆(J.f. − 1. 2 ) ∆(J. M ( 1 . 5. J( 1 .85) ∆(J.s. J( 1 . holds |J.

m1 1 | 2 . 2 [cf. in analogy to (5.. x.52).86) in terms of monomials 1 +m1 1 −m1 2 +m2 2 −m2 a† a† b† b† |Ψo .264). (6. y) = exp (λJ− ) exp x a† exp x b† |Ψo . b† ] = 0 [cf.87) + − + − i. in terms of | 1 . y) is a generating function for the states |Gr defined in (6.90) (6.s. x. x.89) i.87).156 Addition of Angular Momentum and Spin Our strategy for the evaluation of the Clebsch-Gordan-coefficients is to expand (6.86) |Gr = Jr a† st − + s b† + t |Ψo (6. (6. y) which is st based on the properties aζ f (a† ) |Ψo = ζ ∂ ∂a† ζ f (a† ) |Ψo ζ bζ f (b† ) |Ψo = ζ ∂ ∂b† ζ f (b† ) |Ψo ζ (6. follows from the commutation properties (6. Comparision with (6.50–6. x. (6.91) which.93) = f (a† + λ a† ) g(b† + λ b† ) |Ψo . m2 Gordan-coefficients. + + Taylor expansion of the two exponential operators yields immediately I(λ. For this purpose we introduce the generating function I(λ. + − + − . r!s!t! r.88) in terms of monomials (6.57)]..t (6.e. One obtains then using J− = a† a+ + b† b+ (6. st The desired expansion of |Gr can be obtained from an alternate evaluation of I(λ. y) = λr xs y t r |Gst .88).92) − − and noting [a+ . a† ] = [b+ .69) yields then the Clebsch- Expansion of an Intermediate State We first consider the expansion of the following factor appearing in (6.e.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. I(λ.

2 ) in terms of states | 1 . x.99) . 2) = s.90) allows one to infer |Gr s. using the definition (6.94) yields I(λ.3: Explicit Expressions of CG–Coefficients We conclude I(λ.88).t × a† + s−q s q b† + t r−q t−r+q a† − q b† − |Ψo (6. using the commutation property (6.t xs y t s! t! xs y t s! t! a† + a† + λ a† + − s u u s b† + λ b† + − t v b† + t−v t |Ψo = s.97) + 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 (6.50). m2 2 .96) and. 157 (6.94) One can infer from this result the desired expressions for |Gr .t × λ v b† − × r−q v × = t s−u v λu a† − r! q |Ψo λr xs y t r! s! t! r. x. M ( 1 .86) (J− )J−M a† + = q J+ 1 − 2 b† + J+ 2 − 1 |Ψo 1 (6.97). |J. Operation of this operator on (6. Expanding the exponentials in st (6. one can write the right factor in (6. With K † given by (6.95) Comparision with (6. m1 1 | 2 . M ( 1 .98) a† − s a† + b† + s .q (−1)s (6. to obtain the desired expansion of |J. y) = exp a† + λ a† + − exp b† + λ b† + − |Ψo . y) = s.97) yields.70) holds K† 1 + 2 −J = s 1 + 2 −J b† − s 1 + 2 −J−s (−1)s 1 + 2 −J−s (6.t = q r! × a† + s q s−q t r−q a† − q × b† + t−r+q b† − r−q |Ψo (6.s.6.

√ 2J + 1 1 2 . (6.101). j2 .69) if one identifies m1 = 1 −q − s .54) between creation operator monomials and angular momentum states allow one to write this |J.103) = 1.101) Note that m1 + m2 = M holds.158 Addition of Angular Momentum and Spin ( 1 + 2 − J)!(J − M )!(J + 1 − 2 )!(J + 2 − 1 )! s!( 1 + 2 − J − s)!q!(J + 1 − 2 − q)!(J − M − q)!(M + 2 − a† + 2 1 −q−s 1 + q)! a† − q+s b† + M + 2 − 1 +q+s b† − 1 + 2 −M −q−s |Ψo The relationships (6. 1 .102) We want to determine now the expression (6. j2 . M ( 1 . 2 ) ∆(J.83) of the normalization constant N ( 1 + defined through (6. n) 2 )|J. q = 1 − m1 − s and m2 = M − m1 .53. To determine N = N ( 1 + 2 −J. m2 in (6. m2 |J. m2 = M − 1 + q + s. The Clebsch-Gordan coefficents are then finally ( 1 . J( 1 . (6. n = 1 + 2 −J. × 2) = N( 1 + 2 1 − J.104) . The summation over q corresponds then to the summation over m1 . m1 . M − +q+s One can conclude that this expression reproduces (6. M ) (J + 2 )! (J + 2 − 1 )! s.100) ( 1 + 2 − J)!(J − M )!(J + 1 − 2 )!(J + 2 − 1 )! s!( 1 + 2 − J − s)!q!(J + 1 − 2 − q)!(J − M − q)!(M + 2 − 1 1 + q)! × (2 ×| 1 . − q − s)!(q + s)! 2 × (M + 1 − 1 + q + s)!( 1 1 + 2 − M − q − s)! 2 − q − s 1 | 2.82). J( 1 . For this purpose we introduce j1 = 1 (J + 2 1 2 − J. according to (6.103) this can be written 1 = N 2 ψ(j1 . Using (6. 1 . j2 = 1 (J + 2 2 − 1) . 2 ) − 2) .6.82) and (6. − 1 .69) since.q (−1)s × (6. n)|ψ(j1 . 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. 2 ) (6. 2 ) we consider the scalar product J.

70) for the operator K † .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 (6.104) 1 = N2 (n!)2 (2j1 )!(2j2 )! (2j1 + n − s)!(2j2 + s)! s!(n − s)! 2j2 + s 2j2 (6. n) = K† n 159 |j1 .m2 n1 + m1 n1 n2 + m2 n2 λm1 +m2 (6. One obtains then.108) twice. m2 2 .112) s . m1 1 |j2 . 1 1−λ n1 +1 1 1−λ n2 +1 = m1 .6.107) s = (n!)2 s 2j1 + n − s 2j1 The latter sum can be evaluated using 1 1−λ a property which follows from ∂ν ∂λν 1 1−λ n1 +1 n1 +1 = m1 n1 + m1 n1 λm1 (6. (6.108) yields the identity n1 + r − s n1 n2 + s n2 = n1 + n2 + r + 1 n1 + n2 + 1 . j1 1 |j2 . j2 . we obtain |ψ(j1 . j2 . j2 .57) for these states and the expression (6. (6.3: Explicit Expressions of CG–Coefficients where |ψ(j1 . Accordingly.108). n) = 1 (2j1 )!(2j2 )! a† + n! (2j1 )!(2j2 )! a† + 2j1 −n−s s 2j1 s n s 2j2 a† b† + − |Ψo = n−s (−1)s a† b† − + s b† + (−1)s a† − s (2j1 + n − s)!s!(2j2 + s)!(n − s)! s!(n − s)! 2j2 +s b† + b† − n−s (2j1 + n − s)!s!(2j2 + s)!(n − s)! |Ψo .106) The orthonormality of the states occurring in the last expression allows one to write (6.105) The first step of our calculation is the expansion of ψ(j1 . applying (6. j2 2 .108) = λ=0 (n1 + ν)! n1 ! (6.111) Comparision with (6. n) in terms of states |j1 . (6. We employ the expression (6.

M ) corresponding to (6.s. −m1 . m1 .102) of the Clebsch-Gordan coefficients.s. M ) . m1 ) by ( 2 . interchanging the operators J (1) and J (2) results in J = J (2) + J (1) . −m2 |J.116) one expresses the Clebsch-Gordan coefficient on the r. m2 |J. 2 ..116). J. (6. m2 |J. 2 . 2 . −m1 . m1 .11) J = J (1) + J (2) . m2 . For the quantum mechanical addition of angular momenta the Clebsch Gordan coefficients ( 1 . (6. −M ) = (−1) 1 + 2 −J ( 1 . ( 1 .114) one obtains − J = − J (1) − J (2) The respective Clebsch-Gordan coefficients ( 1 . 1 . m1 .119) The corresponding symmetry property of the Clebsch-Gordan coefficients is ( 1 .h.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 . one can interchange also the operator J on the l. and . 2 . and J (2) are vectors in R .102) by replacing ( 1 . 6. m1 . m1 |J. m1 . m2 |J. M ) corresponding to (6. −m2 . m2 ) and.4 Symmetries of the Clebsch-Gordan Coefficients The Clebsch-Gordan coefficients obey symmetry properties which reflect geometrical aspects of the operator relationship (6.114) show a simple relationship to the Clebsch Gordan coefficients ( 2 . −m1 2|J. m2 |J.115) This relationship is a trivial consequence of (6.114) by.120) The symmetry properties (6. (6.h. manner to the coefficients ( 1 .160 Applying this to (6. and (6. M ) = (−1) 1 + 2 −J ( 2 .118) Finally. vice versa. (6. (6. (6.83). m2 . m1 |J.114) as long as J.117) are again related in a simple 2 . m1 ). of (6. 2 . m2 ) by ( 1 .113) Using the identities (6.s. of (6. To derive relationship (6. M ) ( 1 .118). J (1) on the r. −M ) . M | 1 . We will demonstrate this now. m1 ) . ( 2 .114) For example.h. 2 . M ) . (2j1 + 2j2 + 1)! (6.120) can be readily derived from the expression (6. namely. (6. 1 . J (1) .115). m2 |J. M ) = (−1) 2 +m2 2J + 1 ( 2 .116) through formula (6. 2 1+1 (6.116) If one takes the negatives of the operators in (6. of this equation J (1) = J (2) − J .103) one obtains the desired result (6.g. e.

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

According to (5. 1 |θ. 6.147.184) 0 1 space χ 1±1 2 2 one can conclude from (6. Since f (r) is independent of θ and φ follows ˆ ((σ · p f (r))) = −i (( ∂r f (r) )) σ · r .183) in case θ = 0 becomes in the standard representation with respect to the spin.160).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. 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.224)] 1 0 σ·r = ˆ . (5. The result can also be stated in the compact form σ · r Yjm (j ± 1 . for θ = 0 (6.180) where the sign could depend on j. given by ˆ σ · r = σ1 sin θ cos φ + σ2 sin θ sin φ + σ3 cos θ . 1 |θ = 0. ˆ (6. One can conclude then ˆ σ·r = ± ˆ 0 1 1 0 (6. identified the sign of (6.168 Addition of Angular Momentum and Spin 1 Yjm (j ± 2 . hence.181) 0 = j+ 1 2 4π Yj 1 (j + 1 .180) and. 1 |ˆ) = ((σ · p f (r)))Yjm (j ± 1 .186) Here (( · · · )) denotes again confinement of the diffusion operator ∂r to within the double bracket. 1 |ˆ). have proven (6. 1 |ˆ) + f (r) σ · p Yjm (j ± 1 .182) 0 Since σ · r. 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 . φ) = − Yj 1 (j + 1 . 1 |θ = 0.185) We have. ˆ 2 2 2 2 2 2 for θ = 0. 1 |ˆ) r 2 (6.f.148) the particular ˆ 1 1 states Yj 1 (j − 2 .180). φ) 2 2 2 . 1 |θ = 0. 6. φ) and (5. φ) . We want to determine ˆ its action on the wave functions f (r) Yjm (j ± 1 . therefore. (6. φ) 2 2 2 = − j+ 1 2 4π (6. (6.1 states χ 1 ± 1 2 2 2 [c. ˆ (6. 1 |θ = 0. For this purpose we consider the application of σ · r in the case of θ = 0.187) 1 2 .188) .182) σ · r Yj 1 (j − 1 . 1 |ˆ) at θ = 0 are r r 2 2 2 2 Yj 1 (j − 1 .181. We want to demonstrate finally that the “−”-sign holds in (6. 1 |ˆ) r r r 2 2 2 2 2 2 (6.180) and (6. 2 |ˆ) and Yj 1 (j + 1 .174–5.

1 |ˆ) r 2 (6. hence. r r r Using ∂2 (1/r) = −x2 /r3 . we conclude the intuitively expected identity 1 ˆ ˆ p×r = − J .193) r r 1 1 1 = −i (∂2 x3 − ∂3 x2 ) h − i h (x3 ∂2 − x2 ∂3 ) . r Altogether we obtain.197) 1 ˆ σ · p g(r) Yjm (j − 2 .191) (6.5: Spin–Orbital Angular Momentum States Using (6. r 2 (6. ∂3 (1/r) = −x3 /r3 and ∂2 x3 = x3 ∂2 .186) for both terms in (6. 1 |ˆ) r 2 1 2 (6.194) ˆ ˆ where J1 is defined in (5.230) allows one to express ˆ ˆ ˆ ˆ ˆ σ · p σ · r = p · r + i σ · (p × r) .190) can be related to the angular momentum operator. h r (6. ∂3 x2 = x2 ∂3 we obtain ˆ ˆ ( p × r )1 h = −i 1 1 (x3 ∂2 − x2 ∂3 ) h = − J1 h r r (6. To demonstrate this we consider one of its cartision components. x3 x2 ˆ ˆ ( p × r )1 h = −i (∂2 − ∂3 ) h (6. 1 |ˆ) r 2 (6..6.156) this yields finally 1 ˆ σ · p f (r) Yjm (j + 2 . e. r 2 2 1 J ·S 2 1 ˆ σ · p f (r) Yjm (j ± 2 .192) ˆ ˆ The operator p × r in (6.189) The celebrated property of the Pauli matrices (5. 1 |ˆ) = i [ ∂r + r + ] f (r) Yjm (j 2 r r Using (6.53). 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.g.196) = i j+ ∂r + r 1 2 3 2 1 f (r) Yjm (j − 2 .155. 1 |ˆ) .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. 6. 1 |ˆ) r 2 = i ∂r + −j r 1 g(r) Yjm (j + 2 .187) one obtains ˆ σ · p f (r) Yjm (j ± 1 . 1 |ˆ) = 0 and σ = 2S/ . Corresponding results are obtained for the other components of p × r and. using ∂r Yjm (j ± 1 . 1 |ˆ) = r 2 2 ˆ i ∂r f (r) − f (r) σ · p σ · r Yjm (j ˆ 1 2 169 .198) . 1 |ˆ) r 2 (6.195) .

1 |ˆ) r 2 2 The Clebsch-Gordan coefficients are then (j − 1 .197. 6. 1 |j. 6.99) holds ˆ (σ · p)2 = − 2 2 = ∂2 J2 r + 2 . according to (6. Yjj−1 (j − 1 . r 2 2 2 2 2 2 − r 2 ∂r r + J2 r2 f (r) Yjm (j + 1 . 6.43).144) and. 5.e.146). adopting the method in Sec.200) = (j + 1 )(j + 3 ).199). j − 1 . j 2 + 2j + 2 3 4 (6. j) 2 2 2 2 (j − . j) 1 2 1 2 1 2 1 2 = = 1 0 (6. 1 |ˆ) r 2 2 r r2 r2 j 2 + 2j + 3 2 2 2 4 = [ −∂r − ∂r + ] f (r) Yjm (j + 1 . − |j.199) We want to show that eqs.204) (6.143–6. 1 |ˆ) r 2 2 i 3 ˆ = σ · p [ i∂r + (j + ) ] f (r) Yjm (j − 1 .2. 6.2.144) for m = j.202) Yjj (j − 1 . Evaluation of Relevant Clebsch-Gordan Coefficients We want to determine now the Clebsch-Gordan coefficients (6. i. 1 |ˆ) r 2 2 r 2 3 i i 1 2 1 r = [ i∂r + ( − j) ] [ i∂r + (j + ) ] f (r) Yjm (j + 2 .143. 2 2 (6. as well as (6. We note ˆ (σ · p)2 f (r) Yjm (j + 1 .. .203) (6. 6. are consistent with this identity.151). 1 |ˆ) r r 2 2 2 2 2 2 2 yields ˆ (σ · p)2 f (r) Yjm (j + 1 .206) . consider first the case of the largest m-value m = j.198). For this purpose we use the construction method introduced in Sec. r ∂r2 r 2 (6.230. 1 |ˆ) = r 2 2 which agrees with (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 .45). using (5. In this case holds. 1 . 1 |ˆ) r 2 2 r r2 and. j + . according to (6.143. 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. We begin with the coefficients (6. 1 |ˆ) = J 2 Yjm (j + 1 .201) 1 3 (j + )(j + ) Yjm (j + 1 . For m = j − 1 one can state.170 Addition of Angular Momentum and Spin To demonstrate the validity of this key result we note that according to (5. (6.101). in fact. 1 |ˆ) = Yj− 1 j− 1 (ˆ) χ 1 1 .205) which agrees with the expressions (6. r (6.

1 .f.143.210). Expression (6. (6.144) implies for j = m Yjm (j − 1 . 6. 1 |j. 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 .144) for the Clebsch-Gordan coefficients by induction. etc. The further Clebsch-Gordan coefficients (· · · |jj − 2). 6. j − 1 . Let us verify then the expression (6.210) to (6.2. is obtained by applying the operator [c.137)] J− (tot) = J− + S− (6. 2 2 2 2 2j (6.5: Spin–Orbital Angular Momentum States The corresponding Clebsch-Gordan coefficients are then 1 (j − 1 . [c. j − 3 . 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 .206). (6. 2 2 2 2 2j .143. m − 1) 2 2 2 2 = j +m−1 2j (6.202). j − 1) 2 2 2 2 which again agrees with the expressions (6. and J− + S− to the r. 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 .207) (6.212) 1 + 2j(j − m + 1) 1 2j(j − m + 1) Yj− 1 m− 1 χ 1 − 1 2 2 2 2 j+m−1 Yj− 1 m− 3 χ 1 1 + 2 2 2 2 2j j−m+1 Yj− 1 m− 1 χ 1 − 1 .h.143. (6.s. 6.h. 1 . (· · · |jj − 3). as described in Sec. j − 1) 2 2 2 171 = = 2j − 1 2j 1 2j (6. m − 3 .f. 6.210)] yields (j + m)(j − m + 1) Yjm−1 (j − 1 .208) (6..s.209) (j − 1 . are obtained by iterating the application of (6. − 1 |j.211) to the l. 2 |j. 1 .6.144) for m = j − 1.

m3 . using (6. m1 .144) by induction. all three vectors Jclass . 1 . 2j + 2 (6. m2 .213) which is in agreement with (6. Jclass and Jclass play equivalent roles. a higher degree of symmetry is obtained if one rather (−tot) (1) (2) (−tot) considers classically to obtain a vector Jclass with the property Jclass + Jclass + Jclass = 0. m) = 2 2 2 2 Similarly. proven (6. 1 . 1 . 2j + 1 (6.146) can be obtained from (6. The latter relationships applied together read ( 1 .116. M. 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 . 1+1 3 For 1 (j. −m2 | 1 . − 1 |j. The Clebsch-Gordan coefficients (6. 2 . In this context Jclass and Jclass play the same role. the relationship (6.215) which follows from (6. m1 ) (6. m1 . m + 1 . m2 1 (6. m3 ) = (−1) × 2 + 3 − 1 + 2 +m2 × 2 . They are related in a simple manner to the ClebschGordan coefficients j1 j2 j3 m1 m2 m3 = (−1)j1 −j2 +m3 (2j3 + 1)− 2 (j3 − m3 |j1 m1 .216) 6. (1) (2) (−tot) Obviously. m. − 1 |j. m − 1) 2 2 2 2 = (6. 6. by the set j1 .145) from (6. 1 |j + 1 .214) 2 2 +1 ( 3 .143. 6. j2 m2 ) 2 .143) by applying the symmetry relationships (6.218) where we have replaced the quantum numbers J. However. m + 1 ) = 2 2 2 2 and. one can obtain expression (6. m2 | 3 . 6. m. m3 .217) 2j + 2 (j + 1 . 2 . m + 1 .215). 1 |j + 1 .120).172 Addition of Angular Momentum and Spin j −m+1 2j (j − 1 . to reflect in the notation the symmetry of these quantities.144).144) for j = m − 1. one obtains (j + 1 . We have. m − 1 . j3 . j+m+1 . m) 2 2 2 2 2j + 1 (6. (1) (2) (−tot) The coefficients which are the quantum mechanical equivalent to Jclass + Jclass + Jclass = 0 are the 3j–coefficients introduced by Wigner.143).6 The 3j–Coefficients The Clebsch-Gordan coefficients describe the quantum mechanical equivalent of the addition of two (1) (2) classical angular momentum vectors Jclass and Jclass to obtain the total angular momentum vector (tot) (1) (2) (1) (2) Jclass = Jclass + Jclass . 1 . hence. j2 . m1 . 1 .214) yields (j. − 1 |j.143. m + 1 ) = 2 2 2 j+m+1 .

222) m1 m2 m3 j1 + m1 j2 + m2 j3 + m3 The Regge symbol vanishes.218) relating 3jcoefficients and Clebsch-Gordan coefficients. −j1 + j2 + j3 . 6. m3 .6: The 3j–Coefficients We first like to point out that conditions (6. the so-called triangle condition. In this way the values of 12 3j-coefficients are related. states that j1 . In case of a non-cyclic exchange of rows and columns the 3j–coefficent assumes a prefactor (−1)Σ . 173 (6. the Regge-symbol. with respect to alltogether 12 symmetry operations. The Regge symbol reflects a remarkable degree of symmetry of the related 3j–coefficients: One can exchange rows. except when all elements are non-negative integers and each row and column has the same integer value Σ = j1 + j2 + j3 . anti-cyclic exchange and of a sign reversal are stated.219) (6. for which no known classical analogue exists.6.221) j1 j2 j3 −m1 −m2 −m3 where the the results of a cyclic. j2 . there exist even further symmetry properties. For this purpose one can use the symmetry transformations to place the smallest element into the upper left corner of the Regge symbol. e. √ safe for the prefactor 2j3 + 1 which is cancelled according to the definition (6. These symmetry operations relate altogether 72 3j–coefficients.g. The Regge symbol also vanishes in case that two rows or columns are identical and Σ is an odd integer. discovered by Regge. According to the definition of the 3j–coefficients one would expect symmetry properties with respect to exchange of j1 . To represent the full symmetry one expresses the 3j–coefficients through a 3 × 3–matrix. m3 and with respect to sign reversals of all three values m1 . Because of its high degree of symmetry the Regge symbol is very suited for numerical evaluations of the 3j–coefficents. These symmetries follow the equations j1 j2 j3 m1 m2 m3 = (−1)j1 +j2 +j3 = j2 j3 j1 m2 m3 m1 = (−1)j1 +j2 +j3 j2 j1 j3 m2 m1 m3 (6. The reader may note that the entries of the Regge-symbol. Assuming this placement the Regge symbol can be determined as follows (n11 is the smallest element!)   n11 n11 n12 n13 n12 !n13 !n21 !n31 !  n21 n22 n23  = (−1)n23 +n32 sn (6. j3 form the sides of a triangle and the condition is symmetric in the three quantum numbers. =  (6.102) are obtained each through the difference of two entries of the Regge-symbol..e.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 . are identical to the integer arguments which enter the analytical expression (6. m2 . as follows   −j1 + j2 + j3 j1 − j2 + j3 j1 + j2 − j3 j1 j2 j3 j1 − m1 j2 − m2 j3 − m3  . m1 .223) (Σ + 1)!n11 !n22 !n33 !n23 !n32 !)) n=0 n31 n32 n33 . one can exchange columns and one can reflect at the diagonal (transposition).102) of the Clebsch-Gordan coefficients. i. However. The two integer entries J − 1 − m2 and J − 2 + m1 in (6.220) The latter condition |j1 − j2 | ≤ j3 ≤ j1 + j2 . j2 . m2 and j3 .21.

m) δ(−m1 .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. m2 ) √ 2 1 ≥ |m1 |) 2 (6.229) 1 1 (0m| m1 m2 ) = 2 2 i(−1)1−m2 δ(0.227) (2m|2m1 1m2 ) = (−1)m+m1 (m + 2m2 ) δ(m.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 ) (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 (6.223-6.228) √ (−1)2m1 −2m2 7 δ(m. (6.10752786393409395427444450130056540562826159542886 We also illustrate the numerical values of a sequence of 3j-coefficients in Figure 6. (6. m1 + m2 ) (2 − m1 )! (2 + m1 )! √ 10 (1 − m)! (1 + m)! (1 − m2 )! (1 + m2 )! 1 ≥ |m| ∧ 1 ≥ |m2 | ∧ 2 ≥ |m1 | (6. m1 + m2 ) (2 − m1 )! (2 + m1 )! √ 6 (2 − m)! (2 + m)! (1 − m2 )! (1 + m2 )! 1 ≥ |m2 | ∧ 2 ≥ |m| ∧ 2 ≥ |m1 | (6.1.230) 1 1 (1m| m1 m2 ) = 2 2 √ (−1)2m1 −2m2 3 δ(m.224) (6.225) sn−1 . m1 + m2 ) (3 − m)! (3 + m)! (3m|2m1 1m2 ) = √ 105 (2 − m1 )! (2 + m1 )! (1 − m2 )! (1 + m2 )! 1 ≥ |m2 |) ∧ 2 ≥ |m1 | ∧ 3 ≥ |m| (6.226) We like to state finally a few explicit analytical expressions for Clebsch-Gordan coefficients which were actually obtained using (6.232) .226) by means of a symbolic manipulation package (Mathematica) (1m|2m1 1m2 ) = (−1)1+m+m1 δ(m.

. (2 (2 1 2 + 1) (2 + 1) × (2 1 2 + 1) + 1) . . Ω2 ) provide the irreducible representation for the rotation R(ϑ) defined in (6. m2 = − 2 .             (6.233) . . then for a basis {Y 1 m1 (Ω1 )Y 2 m2 (Ω2 ). . If we define the matrix representation of R(ϑ) by D(ϑ). .+ 2 } the matrix has the blockdiagonal form             D(ϑ) =               1·1×1·1 ¤ -10 60 700 m 10-m -40 -20 0 20 40 60 m  1·3×1·3 . . i. .5).6. 1 . Similarly..1: Oscillatory behavior of 3j-coefficients. eigenfunctions of the single particle angular momentum operators J 2 and J3 .. . . . rotations in 2–particle function space. 2 |Ω1 . 2.6: The 3j–Coefficients 10 -3 ¥ ¢ £ 175 3j coefficient 20 120 18 16 14 12 10 8 6 4   2 0 -2 -4 -6 -8 -10 -12 -14 -16 ¡ ¢ -60 ¡ Figure 6.+ 1 . the 2–particle total angular momentum wave functions YJM ( 1 . . Irreducible Representation We had stated above that the spherical harmonics Y m (Ω)).e. m1 = − 1 .e. 2 = 1. i. the rotations in single particle function space. provide the irreducible representation for D(ϑ). .              .

2 |Ω1 .6. 1). . . . each of the blocks in (6. 2.233. . 2 = 1.6. . (6.. .2: How many overall singlet states can be constructed from four spin– 1 states 2 | 1 m1 (1) | 1 m2 (2) | 1 m3 (3) | 1 m4 (4) ? Construct these singlet states in terms of the product wave 2 2 2 2 functions above. .233) is further block-diagonalized as follows + 2. Exercise 6.6. (2( 1 + 2 ) + 1) ×(2( 1 + 2 ) + 1) Partitioning in smaller blocks is not possible. . . J = | 1 − 2 |.234) .6.3: Two triplet states |1m1 (1) |1m2 (2) are coupled to an overall singlet state Y00 (1.1: Prove Eqs. 1 . . Ω2 ). M = −J.234) Exercise 6. . .176 Addition of Angular Momentum and Spin 1 For the basis {YJM ( 1 . Exercise 6. J} (2 (2 1 2 + 1) (2 + 1) × (2 1 2 + 1) + 1) = (2| 1 − 2 | + 1) ×(2| 1 − 2 | + 1) (2| 1 − 2 | + 3) ×(2| 1 − 2 | + 3) (6. Show that the probability of detecting a triplet substate |1m1 (1) for arbitrary polarization (m2 – 1 value) of the other triplet is 3 . .

g. i. . (k) (6.g. (6. for example. q = −k.223). 6. Let T |ψ denote the state obtained after the operator T has been applied and let R(ϑ) denote a rotation in the representation of SO(3) or SU(2) which describes rotational transformations of the quantum states under consideration. .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. 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) (6. S 2 = S1 + S2 + S3 or any other polynomial of Sk . Examples of Tensor Operators The multiplicative operators C∞ ( ) → C∞ ( ) Y ( ) = def Y ( ) (6.222) in case (ii) of single particle spin operators. e.237) The operators T ∈ {Tkq . The latter can be written T |ψ where T denotes T in the rotated frame given by T = R(ϑ) T R−1 (ϑ) .236) In this equation Dq q (ϑ) denotes the rotation matrix Dq q (ϑ) = kq |R(ϑ)|kq .6. k} with the transformation property (6. k} such that k Tkq = q =−k (k) Dq q (ϑ) Tkq . . e.e. In fact. that in case (i) T is an operator C∞ ( ) → C∞ ( ) acting on single particle wave functions. e. Note that in the examples mentioned the operator T would be defined within the same representation as R(ϑ).237) are called tensor operators of rank k.238) .236. Rotations transform |ψ as |ψ = R(ϑ)|ψ and T |ψ as R(ϑ)T |ψ . some of the r r examples considered below involve tensor operators T of this type. The operator T may also act on multi-particle states like Y 1 m1 (ˆ1 )Y 2 m2 (ˆ2 ). the operator (5. Such operators T can be characterized through their behaviour under rotational transformations. . . ∂x2 3 1 2 2 2 2 case (ii) the operator T could be a spin operator Sk defined in (5. −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 . . This implies. that T belongs to a family of operators {Tkq .7: Tensor Operators 177 6. q = −k. −k + 1.g.42) in case (i) of the position representation of single particle wave functions or the operator (5. The latter implies that T transforms like an angular momentum or spin state | m .

but rather any rotation and any parametrization can be assumed.3.6. For the following it is important to note that the rotation matrix elements (6. we will assume presently that the rotation is chosen as follows R(ϑ) = exp ( ϑ+ L+ + ϑ− L− + ϑ3 L3 ) (6.203).7.240) 1 where we have defined ϑ± = 2 (ϑ1 iϑ2 ) and L± = L1 ± iL2 . x3 . x2 .7. 3 follows from the transformation properties of the spherical harmonics derived in Section 1. 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. This choice of parametrization allows us to derive conditions which are equivalent to the property (6. 1. Exercise 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. The fact that these operators form tensor operators of rank 1. but are far easier to ascertain for any particular operator.239) These operators can be expressed in terms of the coordinates x1 .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. In fact.237). 2. 2.178 are tensor operators of rank k. .235 . Exercise 6.

ϑ− .235 . q =−k q+1 (6.235 . 0)T in case of small ϑ+ .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 .3: Derive Eqs. The property (6. Exercise 6.246. (0.242) from which follows by means of kq |1 1|kq equation ϑ+ [L+ .235 .246) (6. Expressing the results in terms of the angular momentum operators J+ .6. (6. (6. hence.6. Tkq ] = = δqq and by subtracting Tkq on both sides of the k ϑ+ kq |L+ |kq Tkq . ϑ− . 2 Exercise 6.243) From (5. 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 (6.5: Consider a system of two spin.248) a tensor operator? Exercise 6.247). 0.7. . Tkq ] = −iTk q+1 (k + q + 1)(k − q) .4: Is the 1-particle Hamiltonian 2 H = − 2 2m + V (|r|) (6. ϑ3 )T .237) yields first k R(ϑ) Tkq R −1 (ϑ) = q =−k kq |R(ϑ)|kq Tkq . Tkq ] [J3 .6. 1 (6. 6. J3 yields [J+ . 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) .237) be assumed then.7: Tensor Operators 179 The conditions can be derived if we consider the property (6. J− .1 particles for which the first spin is described by 2 the operator S (1) and the second spin by the operator S (2) . (6. ϑ3 .80) follows kq |L+ |kq = −i (k + q + 1)(k − q)δq and.6: Form tensor operators of rank 1 in terms of the three components of acting on the space of 1-particle wave functions.7.237) for transformations characterized by infinitesimal values of ϑ+ .245) (6. Tkq ] = = = Tkq+1 Tkq−1 q Tkq . Similar equations can be derived for infinitesimal rotations of the form ϑ = (0.6. Exercise 6. Tkq ] [J− . 0)T . (k + q + 1)(k − q) (k + q)(k − q + 1) (6. We first consider a rotation with ϑ = (ϑ+ .244) [L+ . 0.7.7.

m1 .253) Exercise 6. Before we proceed we like to point out that the states | 2 m2 γ2 are also eigenstates of J 2 .247). i.18) are defined through Φ 1 m1 (k. One can.249) which demonstrates.180 Theory of Angular Momentum and Spin 6. which is characterized also by a set of other quantum numbers γ1 which are not affected by the rotational transformation R(ϑ).m2 2 2 2 We want to show now that these states are eigenstates of J 2 = J1 + J2 + J3 and of J3 where J1 . i. J3 | 2 m2 γ2 = m2 | 2 m2 γ2 . 2 |γ2 ) = q.m2 ( 1 m1 |kq m1 q. the stated property.m2 2 m2 ) (q + m2 ) Tkq | 2 m2 γ2 | 2 m2 γ2 (6.251) The corresponding property for Φ 1 m1 (k. | 1 m1 γ1 denotes an angular momentum (spin) state. 2 |γ2 ) which correspond to total angular momentum states.21) of Clebsch-Gordan coefficients J3 Φ 1 m1 (k. 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. . These states according to (6. 2 |γ2 ) can be shown readily as follows using (6.6. possibly the total angular momentum– spin state of a compositie system. construct states Φ 1 m1 (k. 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. 2 |γ2 ) = ( 1 m1 |kq 2 m2 ) Tkq | 2 m2 γ2 .252) ∼δm1 q+m2 = ( 1 m1 |kq 2 m2 ) Tkq Similarly. in fact. J 2 | 2 m2 γ2 = 2 2( 2 + 1) | 2 m2 γ2 . one can show that Φ 1 m1 (k. J3 are the generators of the rotation R(ϑ). behave like |kq | 1 m1 . (6.e. hence. i.8. The hermitian property of J3 and J 2 implies that states Φ 1 m1 (k. 1 m1 |Φ 1 m1 (k. 2 |γ2 ) is an eigenstate of J 2 with eigenvalue + 1).e.m2 ( 1 m1 |kq 2 m2 ) (J3 Tkq − Tkq J3 + Tkq J3 ) | 2 m2 γ2 = qTkq = q.237) is that their matrix elements 1 m1 γ1 |Tkq | 2 m2 γ2 obey simple relationships expressed in terms of Clebsch-Gordan coefficients.235 .250) q.e. 2 |γ2 ) = Cδ 1 1 δm1 m1 2 1( 1 (6. J2 . (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. 2 |γ2 ) are orthogonal to | 1 m1 in case of different quantum numbers 1 . 2 |γ2 ) is an eigenstate of J 2 with eigenvalue 2 1 ( 1 + 1). J3 | 2 m2 γ2 = m2 γ2 | 2 m2 and the property (6.8 Wigner-Eckart Theorem A second important property of tensor operators Tkq beside (6.1: Show that Φ 1 m1 (k. J3 .

γ1 ||Tk || 2 . i. To prove this property we consider 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) for a different m1 value.6. Using | 1 m1 + 1γ1 = 1 ( 1 + m1 + 1)( 1 − m1 ) J+ | 1 m1 γ1 (6. for which the l. γ2 .259) to a combination of magnetic quantum numbers m1 . γ2 k− 2 + 1 √ 1 2 m2 ) (−1) 2 1 +1 = 1 . γ1 |Tkq | 2 m2 .261) . say m1 + 1.258) We can then finally express the matrix elements of the tensor operators Tkq as follows 1 m1 . q . the matrix elements 1 m1 γ1 |Tkq | 2 m2 γ2 can be reduced to an m1 –independent factor. One can then evaluate also the corresponding Clebsch-Gordan coefficient ( 1 m1 |kq 2 m2 ) and determine 1 . m2 . e.2: Determine the matrix elements of the gradient operator d3 rF (r) G(r) of the type (6. m1 = q = m2 = 0.h. one obtains 1 m1 + 1γ1 |Φ 1 m1 +1 (k 2 |γ2 ) = 1 √ 1 m1 γ1 |J− Φ 1 m1 +1 (k 2 |γ2 ) ( 1 +m1 +1)( 1 −m1 ) 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) = (6.8.s. γ2 (−1)k− 2 + 1 ( 1 m1 |kq 2 m2 ) (6.259) The socalled reduced matrix element 1 .255) At this point the important property can be proven that 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) is independent of m1 . γ1 ||Tk || 2 .32) Tkq | 2 m2 γ2 = and orthogonality property (6. γ2 = 2 1 +1 1 m1 .e. can be evaluated as easily as possible.254) = 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) ( 1 m1 |kq 2 m2 ) .257) which establishes the m1 –independence of 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) . γ1 ||Tk || 2 . (6. γ2 ( 1 m1 |kq (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. γ1 ||Tk || 2 .260) Exercise 6.256) and noting that the operator adjoint to J+ is J− . 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 . γ2 is determined by applying (6. γ1 |Tkq | 2 m2 .8: Wigner-Eckart Theorem 181 In order to evaluate the matrix elements (6. (6.g. its m1 –dependence being expressed solely through a Clebsch-Gordan coefficient.

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

molecules. 7000 Chur. 7.2) serve as schematic descriptions of quantum particles.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 . Switzerland. in potentials which are solely a function of r and are independent of the angles θ.4) governs the motion of electrons in hydrogen-type atoms.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).1) confines a freely moving particle to a spherical box of radius R. for example. and crystals. It leads to the stationary electronic states of the hydrogen atom (Z = 1). Simple Models of Complex Nuclei / The Shell Model and the Interacting Boson Model by I.4) is by far the most relevant of the four choices. The potentials (7.2) V (r) = 0 r > R The third potential V (r) = 1 mω 2 r2 2 (7. φ.1.3) describes an isotropic harmonic oscillator . See. The potential (7. for example. i.e. in case of the so-called bag model of hadronic matter. The second potential is of the square well type −Vo 0 ≤ r ≤ R . Four examples will be studied. The corresponding wave functions serve basis functions for multi-electron systems in atoms. Talmi (Harwood Academic Publishers. The first potential V (r) = 0 0 ≤ r ≤ R ∞ r > R (7. Potential (7. 1993) 1 183 .. (7. The fourth potential V (r) = − Z e2 r (7. Poststrasse 22.

11) (7. ( + 1) ψE. ˆ A classical particle moving in a potential V (r) is governed by the Newtonian equation of motion (7. k = 1. . 2 2mr2 (7.5) In the case of an angular independent potential angular momentum J = m r × v is a constant of ˙ motion.m (r) (7. J ψE.5) and r = v. one can conclude ˙ Jk = {H.f. k = 1. In fact. (7.1 Radial Schr¨dinger Equation o ˙ mv = − er ∂r V (r) .m (r) can be chosen as simultaneous eigenstates of the Hamiltonian ˆ operator H as well as of J 2 and J3 . 3 .m (r) . (7. For this purpose one concludes first that the conservation of angular momentum J implies a motion of the particle confined to a plane. ˆ H ψE.87) for Jk as well as the commutation property (5.e. 3.m (r) . Accordingly. 2.m (r) = = = E ψE.99)] 2 1 2 J2 2 2 = − ∂r r + . the time variation of J can be written.85– 5. d ˙ J = v × mv + r × mv = 0 + r × er (−∂r V (r)) = 0 . using (7.. (7. 2 .10) (7.m (r) .14) (7. 2 . 2.7) ˆ The correspondence principle dictates then for the quantum mechanical Hamiltonian operator H and angular momentum operators Jk ˆ [H.6) ˙ Since Jk is also equal to the poisson bracket {H. stationary states ψE. J3 ψE. Jk ] = 0 .12) .m (r) m ψE. 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 . Employing in this plane the coordinates r. Jk } where H is the Hamiltonian. . θ for the distance from the origin and for the angular position. i. 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.61) of J 2 and Jk .184 Spherically Symmetric Potentials 7.8) This property can also be proven readily employing the expression of the kinetic energy operator [c.15) (7.9) − 2m 2m r 2mr2 the expressions (5.13) . (5. ˆ dt (7. Jk } = 0 . one can state ˙ J = m r2 θ .

Multiplying (7.22) which is the original potential 2 ( +1) V (r) with an added rotational barrier potential 2mr2 . but can also trap particles in the latter space giving rise to strong scattering resonances (see Section ??).m .22) (7. ( + 1) 2mr2 . One can write (7. 2 (7. (7. .16) m 2mr2 Once. ∞[ governed by the effective potential (7. r(t) is determined the angular motion follows from (7. . Employing the kinetic energy operator in the form (7.m (r) = 0.13).m (r) and E . φ) (7. for the wave function of a quantum mechanical particle a differential equation which governs solely the r-dependence. This barrier. .20) Since this equation is independent of the quantum number m we drop the index m on the radial wave function vE. .m (r) 1 2 J2 ∂r r + + V (r) − E 2m r 2mr2 the functional form ψE. It can be solved by integration of 1 2 2 J2 r = ± ˙ E − V (r) − .. (r) = 0 . − Adopting for ψE.12) are obeyed and one obtains for (7.m (r) .m (r) describes the radial motion as a one-dimensional motion in the interval [0.1: Radial Schr¨dinger Equation o 185 This is a differential equation which governs solely the radial coordinate.m (r) = 0. . mr2 (t) (7. by integration of ˙ θ = J .7.m (r) Y m (θ.17) In analogy to the classical description one can derivem.10) − 1 2 ∂ r + 2m r r 2 2 ( + 1) + V (r) − E 2mr2 .10).m vE.4. (7.m ψE.21) Veff (r) φE (r) = = V (r) + r vE.23) This demonstrates that the function r vE.19) where Y m (θ. φ) are the angular momentum eigenstates defined in Section 5.20) in the form of the one-dimensional Schr¨dinger equation o 2 − where 2m 2 ∂r + Veff (r) − E φE (r) = 0 .m (r) 2 . i. using (7.9) one can write the stationary Schr¨dinger o equation (7.24) .11.11).18) = vE.e. in the present case.20) by −2mr/ 2 yields the so-called radial Schr¨dinger equation o 2 ∂r − ( + 1) − U (r) − κ2 r2 r vκ. together with the original potential. 7. equations (7. . (7. (7. can exclude particles from the space with small r values. (7. .

(r) ∼ A r or vκ.33) A detailed discussion of the proper boundary conditions. Pascual (Springer. see. (r) = 0 .D. . hence.20) the index E by the equivalent index κ. 2 2 2m 2 ψE. For r → 0 one may assume that the term ( + 1)/r2 becomes larger than the potential U (r). ψE.20) one needs to specify proper boundary conditions2 . 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).26) 2m 2 In case E < 0. II by A.28) (7. 1975). using the expression (5. (r) (7.27) ∂r − r2 and.182) for Y00 . for example. Berlin.” by J. according to a well-known result in Classical Electromagnetism3 . for > 0 is not integrable. κ assumes real values. (r) ∼ A r + B r− −1 +1 + B r− (7. Jackson (John Wiley. r → 0 (7.30) and.29) Only the first term is admissable. in particular. the term Br−( +1) would contribute |B|2 0 dr r−2 +1) (7. Galindo and P.32) The total kinetic energy resulting from this contribution is. In this case the solution is governed my ( + 1) 2 r vκ.m (r) ∼ √ 4π B δ(x1 )δ(x2 )δ(x3 ) . This follows for > 0 from consideration of the integral which measures the total particle density. at r = 0 is found in the excellent monographs Quantum Mechanics I. We replaced in (7. accordingly. New York. For = 0 the contribution of Br−( function is. (7.25) (7. 2nd Ed. 4π|r| (7. Boundary Conditions In order to solve (7.31) to the complete wave which.186 where we defined U (r) κ2 = = − − 2m 2 Spherically Symmetric Potentials V (r) E (7. The radial part of this integral is ∞ 0 2 dr r2 vκ. . . assumes the general form r vκ. ”Classical Electrodynamics.m (r) ∼ √ B .

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).

e. the surface term ∼ [· · ·]+1 vanishes. namely.s. in case of a free particle moving along the x3 -axis one expands eik3 x3 = . 2 +1 (7. This holds for −1 and one can conclude +1 consecutive integrations by part d izx e dx (7. 7. φ) .99) We want to determine the expansion coefficients b .178) are given in terms of Legendre polynomials P (cos θ) one can replace the expansion in (7.99) +1 d cos θ eikr cos θ P (cos θ) = b j (kr) −1 2 .43) has two solutions. of (7. z = kr. i.98) The l.104) dx eizx P (x) −1 = (−1) 2 ! (iz) = 2 ! +1 dx (x2 − 1) −1 +1 dx (1 − x2 ) eizx .s.194 Generating Function of Spherical Bessel Functions Spherically Symmetric Potentials The stationary Schr¨dinger equation of free particles (7.102) −1 d −1 +1 −1 dx −1 (x2 − 1) . −1 . the wave function does not depend on φ. φ) for non-vanishing m has a non-trivial φ-dependence as described by (5.101) 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 (7.50). One can expand the former solution in terms of solutions (7. For example. The orthogonality properties (5. φ). (7.45) and one given by (7. (7.98) by ∞ eik3 x3 = =0 b j (kr) P (cos θ) . according to (5. and using the Rodrigues formula for Legendre polynomials (5. one given o by (7. (7.100) Defining x = cos θ.h.103) and.98) does not involve any non-vanishing m-values since Y m (θ.h. hence. In this case the expansion on the r.150) one obtains +1 +1 1 ∂ dx eizx P (x) = dx eizx (x2 − 1) .106).50).44. One can show d −1 2 (x − 1) ∼ (x2 − 1) × polynomial in x dx −1 (7..m a mj (kr) Y m (θ. can be written exp(ikr cos θ). Since the spherical harmonics Y 0 (θ.179) yield from (7.

s.109) results in the integral representation of j (z) j (z) = (z) 2 +1 ! +1 dx (1 − x2 ) eizx . in particular. of (7.105) This expression allows one to determine the expansion coefficients b .s.110) One refers to the l.106) Employing (5.96) one can express this.108) where the last factor is equal to unity.105) must hold for all powers of z.112) We want to consider the expression Gν (z) = aν z ν fν (z) where we define fν (z) = C (7. 1 (7.7.105) and (7. using ν = Jν (z) = √ 1 πΓ(ν + 1 ) 2 z 2 ν + 1.114) . (7. 2 +1 −1 dx (1 − x2 )ν− 2 eizx .f.2: Free Particle Comparision with (7. Integral Representation of Bessel Functions Combining (7.99). Comparision with the l.107) 1 2 (2 )! 1 · 3 · 5 · · · (2 + 1) 2 + 1 1 · 2 · 3 · 4 · · · (2 − 1) · 2 (7.h. for the leading power x [c. −1 (7. after insertion into (7.113) dt (1 − t2 )ν− 2 eizt .111) Employing (7. −1 (7. The identity (7. −1 (7. z i or i (2 + 1) z (2 )! 2 2 2 +1 2 ! [1 · 3 · 5 · · · (2 − 1)] 1 (7.69)] b z 2 (iz) = 1 · 3 · 5 · · · (2 + 1) 2 + 1 2 ! +1 dx (1 − x2 ) .h.h. 1 (7.100) gives b j (kr) 2 (iz) = 2 +1 2 ! +1 195 dx (1 − x2 ) eizx .106) yields finally b = i (2 + 1) or.117) one can write the r. as the generating function of the spherical Bessel functions. (7. ∞ (7.109) eikr cos θ = =0 i (2 + 1) j (kr) P (cos θ) .s.

113) obeys the Bessel equation (7. (7.115) Integration by part yields fν (z) = − i 2ν + 1 (1 − t2 )ν+ 2 eizt 1 t2 t1 − z 2ν + 1 dt (1 − t2 )ν+ 2 eizt .2. 7.117). In fact.4: Prove (7.117) dt (1 − t2 )ν− 2 t2 eizt 1 (7. We consider now the functions u(j) (z) = √ 1 πΓ(ν + 1 ) 2 z 2 ν Cj dx (1 − x2 )ν− 2 eizx . To prove this we note fν (z) = i C dt (1 − t2 )ν− 2 t eizt .93) as long as the integration path in (7. t2 .124) .121).120) imply also the property (2ν + 1) fν (z) + z fν+1 (z) = 0 .120) z 2ν + 1 dt (1 − t2 )ν− 2 eizt + C C 1 1 t2 t1 = 0 (7. for properly chosen endpoints t1 . 1 j = 1.118) z 2ν + 1 fν (z) + fν (z) .121) (7.116) fν (z) = − or fν (z) = − From this we can conclude fν (z) + 2ν + 1 fν (z) + fν (z) = 0 .119) We note that equations (7.93) for arbitrary ν.122) (7. We can now demonstrate that Gν (z) defined in (7. obeys Bessel’s equation (7.113) satisfies (7. 4. 2. 1 (7.196 Spherically Symmetric Potentials Here C is an integration path in the complex plane with endpoints t1 .93).123) Insertion of these identities into Bessel’s equation leads to a differential equation for fν (z) which is identical to (7. C 1 (7. z (7. Exercise 7.120) such that we can conclude that Gν (z) for proper integration paths is a solution of (7.114. Gν (z). 3. 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. t2 of the integration paths C. (7.116) In case that the endpoints of the integration path C satisfy (1 − t2 )ν+ 2 eizt one can write (7.

Comparision with (7.125) 0 ≤ s < ∞ (7. for Rez > 0 and ν ∈ R obey (7.131) According to (7. .129) The integrand in (7.127) 0 ≤ s < ∞ (7.124) vanishes along the whole path C3 and.117) and. therefore. u(3) (z) ≡ 0. (7.132) For Hν (z) one derives. the integral expression (1) Hν (z) 1 1 1 1 (7.135) are known as the Poisson integrals of the Bessel functions. dt = i ds.7. u(2) (z) and u(2) (z) are both solutions of Bessel’s equation (7. (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. Following convention. one can state Jν (z) = − u(2) (z) + u(4) (z) .130) We note that the endpoints of the integration paths C2 and C4 . 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 .2: Free Particle 197 for the four integration paths in the complex plane parametrized as follows through a real path length s C1 (t1 = −1 → t2 = +1) : C2 (t1 = 1 → t2 = 1 + i ∞) : C3 (t1 = 1 + i ∞ → t2 = −1 + i ∞) : C4 (t1 = −1 + i ∞ → t1 = −1) : t = s t = 1 + is t = 1 + is t = −1 + is −1 ≤ s ≤ 1 (7.128) C = C1 ∪ C2 ∪ C3 ∪ C4 is a closed path. using t = 1 + is. (7.112) shows u(1) (z) = Jν (z). Accordingly.135) (7.134) and (7.93). Since the integrand in (7. one can derive (2) Hν (z) = e −i(z−πν/2−π/4) ∞ 0 √ ds [s(1 − is/2)]ν− 2 e−zs . 1 (7. we introduce the so-called Hankel functions (1) Hν (z) = −2 u(2) (z) . (2) Hν (z) = −2 u(4) (z) .130) holds Jν (z) = (1) 1 2 1 (1) (2) Hν (z) + Hν (z) . hence.134) Similarly.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 .126) −1 ≤ s ≤ 1 (7. 1 (7.

πz (7. converge fast for |z| → ∞. One obtains W (z) = − 4i . h(z) to be linearly independent. We employ for this purpose the Poisson integrals (7.134) and (7.136) (1) 2 (z) = 0 ds [s(s ± is/2)] e−zs . h) must be a non-vanishing function. Hν (z) as well as H−ν (z) are solutions of this equation.93) only as ν 2 .e.140). i.138) The formula for the Laplace transform of sn leads to f and.135) which read for ν = + 1 2 H where f The binomial formula yields f (1) 2 (1) 2 1 +2 (1. we obtain H (1) 2 1 +2 (1) 2 (z) = r=0 ( + r)! ! r!( − r)! ± i 2 r 1 z +r+1 (7. (7. (7.139) (z) = 2 ±i[z − ( e πz +1) π ] 2 r=0 ( + r)! r!( − r)! i ± 2z r (7.137) (z) = r=0 r ± i 2 r 0 ∞ ds s +r e−zs . For such solutions g(z). For the Wronskian connected with the Bessel equation (7. the Wronskian W (g.140) Bessel Functions with Negative Index Since ν enters the Bessel equation (7. The general solution of (7.141) (1) (1) the derivation of which follows the derivation on page 192 for the Wronskian of the radial Schr¨dinger equation.. hence.141) is o W (z) = − In case of g(z) = H (1) (z) +1 2 c .142) and h(z) = H (2) (z) +1 2 one can identify the constant c by using the leading terms in the expansions (7. z (7. As a second order differential equation the Bessel equation has two linearly independent solutions.143) .198 Asymptotic Behaviour of Bessel Functions We want to obtain now expansions of the Hankel functions H Spherically Symmetric Potentials the expansions converge fast asymptotically.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 ! (7.93) holds the identity 1 W = − W z (7.

132.152) (7.146.90.153) . 7. B can be obtained from the asymptotic expansion (7.151) are equivalent to h h (1) (2) (z) (z) = = j (z) + i n (z) j (z) − i n (z) .151) Equations (7.145) This equation can hold only for B = 0 and A = exp[i( + 1 )π].7.2: Free Particle i. in fact.2) H + 1 (z) .132) follows n (z) = 1 2 π (−1) 2z +1 H−( (1) 1 +2) (z) + H−( (2) +1) 2 (z) .2) (2) +1) 2 (1) +1) 2 (z) = i (−1) H (1) (z) +1 2 . We conclude 2 H−( Simarly.97) one defines the spherical Hankel functions h (1.149.144) The expansion coefficients A. H (1. 7. 2z 2 (1.2) (7.. linearly independent.150) According to (7.2) (7. For |z| → ∞ the leading terms yield π π π π π 1 1 1 √ ei(z+ 2 ) = √ A ei(z− 2 − 2 ) + √ B e−i(z− 2 − 2 ) z z z |z| → ∞ . From (7. 7. (7. 7. 7.147) (z) = π (1. According to (7. (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.e. (7. 7. (7. (7. (7. one can show H−( Spherical Hankel Functions In analogy to equations (7.2) (z) +1 2 199 are.140).53). (7. H−ν (z) = A H (1) (1) (z) +1 2 One can then expand + BH (2) (z) +1 2 .148) holds for the o regular spherical Bessel function j (z) = 1 (1) (2) [ h (z) + h (z) ] .149) We want to establish also the relationship between h (z) and the irregular spherical Bessel function n (z) defined in (7.97).96.146) (z) = −i (−1) H (2) 1 (z) +2 . 2 (1.97.147) this can be written n (z) = 1 2i h (1) (z) − h (2) (z) .

it follows z ∈ R : Using p (z) = Re r=0 j (z) = Re[h (1) (z)] .148) one obtains readily h (1) 2 (1.160) Employing cos[z − ( + 1) π ] . = −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) .156) ( + r)! r!( − r)! ( + r)! r!( − r)! i 2z r [ /2] = r=0 ( + 2r)! (2r)!( − 2r)! −1 4z 2 r (7. 2 2z z (7.161) (7.155) (z) and To determine j (z) and n (z) we note that for z ∈ R the spherical Hankel functions h (2) h (z).140) and (7.2) (z). at large |z|. 7.154) The leading term in this expansion. at large |z|. z (7.159) (7.140. j (z) (z) = ( i) z +1 e±iz r=0 ( + r)! r!( − r)! ± i 2z r (7. 2z 2 z sin(z − π/2) z cos(z − π/2) − . is h (1) 2 (z) = ( i) z +1 e±iz . (1) (7. (7.164) .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 (7.162) The leading terms in these expansions. ne ll(z) = Im[h (1) (z)] . are j (z) n (z) = = (7.148). From (7. Hence. = sin(z − π/2) and sin[z − ( + 1) π ] .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).163) (7. as given by (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) . are complex conjugates.158) one can derive then from (7.

114) we can write g (z) = a z f 2 ! + 1 (z) 2 .173) A (z) =  z ( +1) z2 +2 z (7.124.149. From the linearity of the relationships (7.168) (7.171) −1 (z) (7.174) We want to prove these relationships.171) holds for g (z) = h (z) we employ (7.162) allow one to provide explicit expressions for j (z) and n (z). 7. One obtains for = 0. For this purpose we need to demonstrate only that the (1. defining a = −1 and employing fν (z) as defined in (7. (7. (1. j (z) and n (z).166) (7.170) Recursion Formulas of Spherical Bessel Functions The spherical Bessel functions obey the recursion relationships g g +1 (z) +1 (z) = = (1.131.161.4 dx (1 − x2 )ν− 2 eizx . One can combine (7.151) follows then that the relatiosnhips hold also for g (z) = j (z) and g (z) = n (z).7.172) where g (z) is either of the functions h obtain the recursion relationship g g (z).2) To demonstrate that (7. 1. 7.167) (7.172) to +1 (z) +1 (z) = A (z)  1− g (z) g (z) −1   (7.169) (7.2) (z) = −2 π 1 √ 2z πΓ(ν + 1 ) 2 z 2 ν C2. 7. 1 (7.171–7. 7.176) . 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 (7.175) Using Γ( + 1) = !.174) and from (7.2) g (z) − g (z) z 2 +1 g (z) − g z (7.148) and express g (z) = h (1.171.2: Free Particle Expressions for the Spherical bessel Functions j (z) and n (z) 201 The identities (7. 7.165) (7.2) relationships hold for g (z) = h (z).

together with (7.182) Replacing all first derivatives employing (7. 7.165).174) we start from (7.58) it holds o 2 g (z) = − g (z) + z ( + 1) g (z) − g (z) .121).166). To prove (7.202 The derivative of this expression is g (z) = Employing (7.178) (7.182). is equivalent to (7.171).172).6: Employ the recursion relationship (7. 7. j0 (z) using (7. Exercise 7.173) is equivalent to (7.2.173. in fact. (7.5: Provide a detailed derivation of (7.174) to determine (a) j1 (z). n2 (z) from n0 (z).179) from which follows (7.179) leads to (7.180) Since g (z) is a solution of the radial Schr¨dinger equation (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) .180) yields g +1 (z) = g (z) − ( + 2) g (z) + z2 +2 g (z) . j2 (z) from j0 (z). Exercise 7..172) we differentiate (7.173).171). (7. The first component of (7.e.171). The second component of (7. n0 (z) using (7. z2 (7. and (b) n1 (z).2. In order to prove (7. f yields. (7. z (7.181) Using this identity to replace the second derivative in (7. .179) g (z) = − z2 g (z) + g (z) − g +1 (z) z .173.172).177) (z) = − z f 2 +2 + 3 (z) 2 . i.

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

5) above agrees with the equation of motion as given in (1. In the non-relativistic limit holds p ≈ mr and (8. Scalar and Vector Potential Setting B(r. t) and A(r. called the scalar potential. t). t) (8. Equation (8.4) Lorentz Force A classical particle with charge q moving in the electromagnetic field experiences the so-called Lorentz force q[E(r. New York. t) = − V (r. t) is a scalar function. .10) (8.9) (8. t) + ∂t A(r.1) the term qδ(r − ro (t)) and to the current density ˙ J(r. but leave the fields unaltered: A(r. t) = 4 π J(r.7) Gauge Transformations We have expressed now the electric and magnetic fields E(r.204 Maxwell equations1 · E(r. t) Interaction of Radiation with Matter = 4 π ρ(r.11) V (r.3) (8. t) + ∂t A(r. t)] and.6) for some vector-valued function A(r. t) −→ 1 A(r. t) −→ V (r. the two homogeneous Maxwell equations hold × E(r. t) = − V (r. t) + χ(r. t) and B(r. t) + v × B(r. From this follows E(r. t) − ∂t E(r. t) × B(r. t) in (8. t) where V (r. We assume so-called Gaussian units.4). by J. in turn. As is well known. t) which is solved by E(r. Jackson (John Wiley & Sons. 2nd Edition. contributes to the charge density ρ(r. The following substitutions.1) (8. t) = × A(r. t) + ∂t B(r. obeys the equation of motion d p = q dt E[ro (t).5) where p is the momentum of the particle and ro (t) it’s position at time t. t) · B(r. t) − ∂t A(r. t) through the scalar and vector potentials V (r. solves implicitly (8. t) .3) reads then × E(r.2) In addition.2) the term q r˙o δ(r − ro (t)). called the vector potential.8) = 0 (8. t] (8. t) in (8.25). t] + v × B[ro (t). The reader is referred to the well-known textbook ”Classical Electrodynamics”. D. The particle. t) − ∂t χ(r. the relationship between fields and potentials is not unique. accordingly. alter the potentials. t) . t). 1975) for a discussion of these and other conventional units. t) . t) (8. called gauge transformations. (8. (8. t) = 0 = 0. (8.

For this purpose we examine the offending term ∂t V (r. t) |r − r | (8.8. (8.1: Electromagnetic Field 205 This gauge freedom will be exploited now to introduce potentials which are most suitable for the purpose of separating the electromagnetic field into a component arising from the Coulomb potential connected with the charge distribution ρ(r. t) in (8. t) − 2 × A(r. t) − ∂t V (r. t) the condition · A(r. the second inhomogeneous Maxwell equation.9) into (8.19) couples the vector potential A(r. t). 4π (8. and a component due to the remaining currents. t) = 0 the solution is given by the Coulomb integral V (r. In fact. t) . t) + ∂t V (r. t) = 0 .19) and define J (r.12) The corresponding gauge is referred to as the Coulomb gauge.2). a name which is due to the form of the resulting scalar potential V (r.12) yields then the Poisson equation 2 V (r.1) · − V (r. (8.20) . (8. t) − ∂t A(r.12) leads us to 2 2 A(r. t) can be obtained employing (8. Such description can. t) = Ω∞ for r ∈ ∂Ω∞ (8. t) = ∂t = 4 π ρ(r. t) . in fact. Using the expressions (8. t). t) = − 4 π J(r. be achieved. this potential results from inserting (8.14) In case of the boundary condition V (r.6) and (8. t) and the current due to moving net charges. the gauge freedom allows us to impose on the vector potential A(r.15) d3 r ρ(r .13) · A(r.17) A(r.16) and the vector potential are uncoupled.19) Unfortunately. and the former the Coulomb interactions in the unperturbed bound particle system. equation (8.18) together with condition (8. such that the latter describes the electromagnetic radiation. t) . t) = · A(r. t = 4 π J(r. t) − ∂t A(r. t) and V (r. t) = − 4 π ρ(r. t) . One would prefer a description in which the Coulomb potential (8. t) = 1 ∂t V (r.9) for the fields results in × The identity × × A(r. t) (8. (8. t) Using · ∂t A(r.16) This is the potential commonly employed in quantum mechanical calculations for the description of Coulomb interactions between charged particles. The vector potential A(r. In fact. t) . (8. t) together with (8. t) + ∂t A(r.

26) turns into the well-known wave equation 2 2 A(r. hence. (8. t) + ∂t ρ(r.28) (8. the name transverse fields. t). t) = 0 .g.26) This equation does not couple anymore scalar and vector potentials. The definitions of J and Jt applied to (8. t) . (8. Let J(r.27) which is obviously curl-free ( × E (r.2 Planar Electromagnetic Waves The current density Jt describes ring-type currents in the space under consideration. t) = − ∂t ρ(r. using ∂t · J (r. t) holds. the name longitudinal electric field. For the divergence of J (r. t) + ∂t ρ(r. t) = 0). t) = 0 . V (r.e. Presently. t) (8. t) = 0). respectively. A(r.24) (8. V (r. t) . For J also holds the continuity equation · J(r. such current densities exist.22) 1 ∂t 4π · J (r. t) contributes solely an electric field component E (r. The vector potential determined through (8.30) . for example. i.29) · Et (r. t) = − V (r.21) ∂t and the Poisson equation (8. yet a current. = − ∂t A(r. In this case (8. t) − ∂t A(r. (8. exist in the space considered.21) and (8. hence.206 For the curl of J holds × J (r. t) = 0 and from this follows · Jt (r. t) = 0 .. in a ring-shaped antenna which exhibits no net charge. 8.14). t) be the total current of the system under investigation and let Jt = J − J . t) Bt (r. t) = or = 2 Interaction of Radiation with Matter (8. t) These fields are obviously divergence -free (e.12) and the Coulomb potential (8.26) and (8.16) yield finally the electric and magnetic fields. t) = − 4 π Jt (r.19) yield 2 2 A(r. t) contributes an electrical field component as well as the total magnetic field Et (r.. we want to assume that no ring-type currents. no divergence-free currents.25) Because of properties (8. t) = 0 (8. t) (8. t) = × A(r. t) − ∂t A(r.25) one refers to J and Jt as the longitudinal and the transverse currents.23) This continuity equation identifies J (r. t) as the current due to the time-dependence of the charge distribution ρ(r. (8.

ω2 = |Ao |2 (8.35) The vector potential in (8.34) (8. t) = ± ω2 ˆ |Ao |2 k . (8.39) 8π The definition incoming waves and outgoing waves is rationalized below in the discussion following Eq. Here the “-” sign corresponds to so-called incoming waves and the “+” sign to outgoing waves2 .35) one obtains S(r. at each point r and moment t. t) .35).2: Planar Electromagnetic Waves 207 which describes electromagnetic fields in vacuum. In applying the potential and fields to physical observables and processes we will only employ the real parts. 8.32.34.38) ˆ In this expression for the energy flux one can interprete k as the propagation velocity (note c = 1) and. t) = ± ω2 ˆ |Ao |2 k sin2 ( k · r − ωt ) 4π (8. The latter vector describes the direction of propagation of the energy flux connected with the plane wave electromagnetic radiation.158). The Coulomb gauge condition (8.31). see also the comment below Eqs. Et (r.39).31. t) Bt (r. ˆ (8. This flux is given by 1 S(r. the constant k is referred to as the wave vector. t) and Bt (r. are Et (r. Note that in the units chosen the velocity of light is c = 1. 2 . t) = Ao u exp i(k · r ˆ where the dispersion relationship |k| = ω (8.31) and the resulting fields (8.33) ωt) (8. accordingly. Time average over one period 2π/ω yields S(r. t) in (8. 8. (8. according to (8.8. 8. t) . are orthogonal to each other and are both orthogonal to the wave vector k.12) yields u·k = 0. obviously. 8.32) holds.35) are complex-valued quantities.29). A complete set of solutions is given by the so-called plane waves A(r. t) = i k × A(r. is orthogonal to k. Obviously. The corresponding electric and magnetic fields. 8. 8π (8. 8. (8.28.36) 4π Using the identity a × (b × c) = b (a · c) − c (a · b) and (8.31) u is a unit vector (|ˆ| = 1) which.37) ˆ ˆ where k is the unit vector k = k/|k|. (8. there exist two linearly ˆ u independent orientations for u corresponding to two independent planes of polarization. 8. t) = Re Et (r. t) × Re B(r. hence.34.34.38. t) = ±i ω A(r. ˆ We want to characterize now the radiation field connected with the plane wave solutions (8.

defined below Eqs. t) = 8πNω u exp i(k · r − ωt) ˆ ωV . t) 2m 2 + qV (r) .43) Here the fields are defined through Eqs.3 Hamilton Operator The classical Hamiltonian for a particle of charge q in a scalar and vector potential V (r) and A(r.39) and (8.44) . i (8. u · k = 0 .8.43). 8. whereas in the case of outgoing waves the energy is transported in the direction of k.31). is p − q A(r.42) 8. (8.27. We assume that the energy content of the fields is not altered significantly in the processes described and. t) 2m + 1 8π Ω∞ 2 H = + qV (r) 1 16π d3 r |Et |2 + |Bt |2 Ω∞ d3 r E 2 + . (8.29) together with the potentials (8. |k| = ω . t). 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. a ˆ specific k and a specific u.31) allows one finally to state for the planar wave vector potential A(r.31). The wave function Ψ(r.31.39) through the density of photons connected with the planar waves (8. We are left with a classical Hamiltonian function which has an obvious quantum mechanical analogue ˆ H = ˆ p − q A(r. t) are complex leading to the difference of a factor 2 in the energy densities of the lontitudinal and transverse components of the fields. respectively. The sign in (8.32). Note that E (r. If we consider a volume V with a number of photons Nω the energy density is obviously Nω ω . t) is real 1 and that Et (r. (8. ˆ (8. t). These photons each carry the energy ω. t) = H Ψ(r. A ‘poor man’s’ quantization of the field is possible at this point by expressing the energy density (8.40) allows one to express then the field ˆ amplitudes 8πNω Ao = .40) = V It should be pointed out that Nω represents the number of photons for a specific frequency ω. (8. 8.208 Interaction of Radiation with Matter as the energy density.41) ωV Inserting this into (8. t) . Comparision of (8. hence.28. 8. The integrals express the integration over the energy density of the fields. Bt (r. (8.38) implies that for incoming waves. we will neglect the respective terms in the Hamiltonian (8. (8. A correct description of the electromagnetic field requires that the field be quantized.16. the energy of the plane wave is transported in the direction of −k.45) .

50) The equivalence of (8.t)/ . In the classical case such question is mute since the gauge transformations do not alter the fields and. does not change expectation values which contain the probability densities3 .t)/ should not affect a system and that.47) is equivalent to   2 ˆ p − qA   i ∂t eiqχ(r. however. not to phenomenological interactions like the molecular van der Waals interaction. t) inside the double brackets.t)/ p + q(( χ)) Ψ(r.11) to (8.. One can show that (8. t) =  + qV  eiqχ(r.10. Obviously. t) . t) =  + qV  Ψ(r. hence. by the group 3 The effect on other expectation values is not discussed here. Applying the gauge transformations (8. An important conceptual step of modern physics has been to turn the derivation given around and to state that introduction of a local phase factor eiqχ(r. hence.49) (8.11) of the potentials is equivalent to multiplying the wave function Ψ(r.48) 2m For this purpose one notes i ∂t eiqχ(r. such phase factor does not change the probability density |Ψ(r. t) by a local and time-dependent phase factor eiqχ(r.51) 2m the potentials A(r.10. 8. that this principle applies only to fundamental interactions.t)/ [ i ∂t − q((∂t χ)) ] Ψ(r.10.t)/ Ψ(r. t) and V (r. This leads to the question in how far the gauge transformations (8.47. Elementary constituents of matter which are governed by other symmetry groups.8.t)/ Ψ(r.47) i ∂t Ψ(r. (8. t) ˆ = eiqχ(r. t) enter whereas in the classical equations of motion ˙ m¨ = q E(r. t) + q r × B(r. t) (8. t) r (8. t) =  2m where ((· · ·)) denotes derivatives in (( χ)) and ((∂t χ)) which are confined to the function χ(r.46) the fields enter.t)/ Ψ(r. 8. an element of the group U(1). t) ..e. t)|2 and. in the Schr¨dinger equation o   2 ˆ p − qA   i ∂t Ψ(r. It should be noted. t) and A(r.t)/ Ψ(r.11) affect the quantum mechanical description. 8.3: Hamilton Operator 209 Gauge Transformations It is interesting to note that in the quantum mechanical description of a charged particle the potentials V (r.44. The idea just stated can be generalized by noting that multiplication by a phase factor eiqχ(r. have no effect on the motion of the particle described by (8. 8. (8. 8.45) leads to the Schr¨dinger equation o   2 ˆ p − q A − q(( χ))   + qV − q((∂t χ))  Ψ(r.t)/ constitutes a unitary transformation of a scalar quantity. accordingly. t) ˆ p eiqχ(r. t) . t) are necessary to compensate terms which arise through the phase factor.46). (8. i. t) = eiqχ(r.g. e.48) implies that the gauge transformation (8. .

x2 . x3 ) = exp(ik2 x2 + ik3 x3 ) φE (x1 ) . The stationary homogeneous magnetic field B(r. x1 . k2 .t) where σ is the vector of Pauli matrices.58) where ∂j = (∂/∂xj ).55) 8.59) . t). (8. (8. can be described by various vector potentials. k3 ..54) ˆ2 p q ˆ q2 2 − p·A + A + qV .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. · A = 0 [cf. 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. j = 1.g. H = 2m m 2m (8.55).53) ˆ ˆ p·A + A·p + q2 2 A + qV 2m (8. (8. for Bo = Bo e3 in (8.58).52) This expression vanishes in the present case since since ˆ ˆ p · Af = A · pf and.45) with Hamiltonian (8.56) due to the gauge freedom.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. t) = Bo . affects the complexity of the mathematical derivation of the wave functions. the generators of SU(2). Accordingly. t) ≡ 0. 2. The vector potential (8. holds (8. Using Cartesian coordinates this yields 2 H = − 2me 2 2 2 ∂1 + ∂2 + ∂3 + 2 eBo e2 Bo 2 x1 ∂2 + x i me 2me 1 (8.55). The Hamiltonian (8.210 Interaction of Radiation with Matter SU(2). likewise can demand the existence of fields which compensate local transformations described by eiσ·χ(r. For this purpose we use the wave function in the form Ψ(E. We want to describe the stationary states corresponding to the Hamiltonian (8. 3. The choice of a vector potential affects the form of the wave functions describing the eigenstates and. (8. consequently.57) satisfies · A = 0 and. one can employ the Hamiltonian (8. e. thereby. the so-called Landau gauge ˆ associates the vector potential AL (r) = Bo x1 e2 ˆ (8. therefore.12)]. The resulting fields are called Yang-Mills fields.57) with a homogeneous magnetic field Bo . In this o case holds V (r. In case of a homogeneous potential pointing in the x3 -direction.56).

65) which localize the electrons.8. according to (8. 2 2me (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. (8. Without affecting the energy one can form wave packets in terms of the solutions (8. the states are degenerate in the quantum number k2 describing displacement along the x2 coordinate.62) is that of a displaced (by x1o ) harmonic oscillator with o 2 k 2 /2m ) energies. the so-called symmetric gauge which expresses the homogeneous potential (8.64) is the classical Larmor frequency (c = 1).66) Obviously. Unfortunately.63) this induces a spread of the wave function in the x1 direction. Solution for Symmetric Gauge The solution obtained above has the advantage that the derivation is comparatively simple. is not symmetric in the x1 . according to (8.63) (8. k3 . The energies corresponding to these states are E(n.62) x1o = and where ω = k2 eBo eBo me (8.59). The stationary Schr¨dinger equation (8. However.65) where we replaced the parameter E by the integer n.59).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 (8.67) . k2 . the wave function (8. x1 . is Ψ(n. (8. k3 ) = ω(n + 2 k2 1 3 ) + . We want to employ . From this observation one can immediately conclude that the wave shifted (by e 3 function of the system. like the corresponding gauge (8. therefore.and x2 -coordinates.57).4: Electron in Homogenous Magnetic Field This results in a stationary Schr¨dinger equation o 2 211 − 2me 2 ∂1 + 2 k2 2 2me + 2 k2 3 2me + 2 eBo k2 e2 Bo 2 x1 + x φE (x1 ) me 2me 1 = E φE (x1 ) . the familiar harmonic oscillator quantum number. k2 . x2 .56) through the vector potential A(r) = 1 Bo × r . 2 (8.61) leads to − where 2 2me 2 ∂1 + 2 k2 1 3 me ω 2 ( x1 + x1o )2 + 2 2me φE (x1 ) = E φE (x1 ) . 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 (8.65).

namely.75) 2 The latter can be related to the angular momentum µclass = − and. (8.68) × u yields. Accordingly.78) . ˆ (8. an electron moving in the x. In this case the current density measures −e v oriented tangentially to the ring.72) and (8. (8.12) for the Coulomb gauge.72) to Hamiltonian (8. (8.72) allows one to interpret µ = − e L 2me (8.67) one can write ˆ p·A = Using · (u × v) = − u · function f (r) ×v + v· 2i · Bo × r . ˆ Identifying r × p with the angular momentum operator L.74) is in the present case 1 µclass = − e r v e3 . for any (8. (8.73.74) We consider a simple case to relate (8. y-plane with constant velocity v on a ring of radius r.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 .73) where µclass is the magnetic moment connected with a current density j µclass = 1 2 r × j(r) dr (8. Vmag = −µclass · Bo (8. The theory of classical electromagnetism predicts an analogue energy contribution . 8. in the present case of constant Bo .212 Interaction of Radiation with Matter One can readily verify that this vector potential satisfies the condition (8.71).71) (8.74).77) Comparision with (8. namely.69) × u and × r = 0. The latter can be rewritten. using × (uf ) = −u × f +f ˆ ˆ Bo · p × r f = Bo · r × p f .76) e 2me class class · Bo . accordingly. the Hamiltonian (8. For the vector potential (8. the magnetic moment (8.70) ˆ ˆ p · A f = − Bo · p × r f . Vmag = e 2me class = r me v e3 of the electron ˆ (8.

likewise.87) . e gives rise to an energy contribution (8. 2 . Accordingly.85) The Hamiltonian (8. We will demonstrate in Sect. x2 ) .56) pointing in the x3 -direction the symmetric gauge (8. x2 ) where ˆ Ho = − 2 2 2 ∂1 + ∂2 (8. In this case.86) 1 2 2me 1 2 ˆ as well as of the angular momentum operator L3 . 2me We have used here the expression for the angular momentum operator ˆ L3 = ( /i)(x1 ∂2 − x2 ∂1 ) . x3 ) .e. To obtain these eigenstates we introduce the customary dimensionless variables of the harmonic oscillator Xj = (8. The functions ψ(x1 . i. (8.82) 1 ω (x1 ∂2 − x2 ∂1 ) i 2me + 1 me ω 2 x2 + x2 1 2 2 2 k2 3 + (8.81) (8. 2me (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 . we separate the variable x1 . x2 . A derivation of his property and the value of g.71) is ˆ H = 2 ˆ2 p e2 Bo + 2me 8me x2 + x2 + 1 2 eBo L3 . (8. x2 ) = E ψ(x1 . j = 1. described by the operator S.67) yields a more symmetric solution which decays to zero along both the ±x1 . x2 . x2 ) obey then ˆ Ho ψ(x1 . 10. the solutions of ˆ H ΨE (x1 .8.80) (8. x2 .88) me ω xj . ˆ i. requires a Lorentz-invariant quantum mechanical description as provided in Sect.83) describes two identical oscillators along the x1 -and x2 -directions which ˆ are coupled through the angular momentum operator L3 . 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.and the ±x2 -direction. the Hamiltonian (8. x2 from x3 setting ΨE (x1 . x3 ) = E ΨE (x1 .e.. E = E − (8. 10 that the spin of the electron. for Bo = Bo e3 .84) .79) To obtain the stationary states.4: Electron in Homogenous Magnetic Field 213 as the quantum mechanical magnetic moment operator for the electron (charge −e). For a magnetic field (8.83) (8. x3 ) = exp(ik3 x3 ) ψ(x1 . the so-called gyromagnetic ratio of the electron.72) with an associated magnetic moment − g 2me S where g ≈ 2.

2 2 . x1 .94) Since this state is a linear combination of states which all have vibrational energy (2j + 1) ω. in fact. (8. +j .92) represents a state with j + m quanta in the x1 -oscillator and j − m quanta in the x2 -oscillator.. x2 ) = m=−j αmm Ψ(j. obeyed. x2 ) .94) is an eigenstate of the vibrational Hamiltonian. it holds a† a1 + a† a2 + 1 1 1 2 Ψ (j.90) a† a2 − a† a1 1 2 . 0.92) where Ψ(0. x1 . x1 . . j = 1. m . m . 1 3 .94) is also an eigenstate of L3 . x1 .89) . m . (8. (8. x2 ) . x2 ) . . x1 . x2 ) (8. x2 ) (8. . m . x1 . . (8.97) .. such that 1 i a† a2 − a† a1 1 2 Ψ (j. −j + 1. . m = −j. m. x2 ) = 2m Ψ (j. 1 2 since one phonon is annihilated and one created. (8.e. (j) (8. x1 .96) ˆ holds. Such eigenstates can be expressed. In order to cover all posible vibrational quantum numbers one needs to choose j. attempt to express eigenstates in terms of vibrational wave functions a† 1 j+m Ψ(j. however.95) (j) ˆ We want to choose the coefficients αmm such that (8. x1 . aj = √ 2 Interaction of Radiation with Matter Xj + ∂ ∂Xj . (8. . therefore. If this property is. i. x2 ) is the wave function for the state with zero vibrational quanta for the x1 . through a combination of states j Ψ (j. x1 . the total vibrational energy being ω(2j + 1). 1. x1 . We. x2 ) = a† 1 j−m (j + m)! (j − m)! Ψ(0.91) We note that the operator a† a2 − a† a1 leaves the total number of vibrational quanta invariant.e. m as follows: j = 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 . (8. x2 ) = ( 2j + 1 ) Ψ (j.as well as for the x2 -oscillator. . (8. i. 2 (8. m.93) ˆ The states (8. m . m .94) is an eigenstate of Ho ˆ Ho Ψ (j.214 Employing the creation and annihilation operators 1 a† = √ j 2 and the identity ˆ ω L3 = which can readily be proven. m .92) are not eigenstates of L3 . x1 . . x2 ) = ω 2j + 2m + 1 Ψ (j. 0.

an example is the Coulomb potential V (r. the eigenstates of (8. x1 .11. According to Sect. the 2 2 latter representing the Wigner rotation matrix of Sect.9.e. t) introduced above.84) above.11.(5.102) We have identified.99) with eigenvalue 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. the states are eigenstates of the operator [we use for the operator the notation of Sect. cf.e. 5. x2 ). Other radiation fields can . x1 . 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 .9. 8. According to the derivation given there. x1 .100) J2 = 1 2 2i 1 ˆ i.10. m . In the simplest case the radiation field consists of a single planar electromagnetic wave described through the potential (8. b† . π . m.55). b+ .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. m. 5. t) = 1/4πr confining an electron with energy E < 0 to move in the well known orbitals of the hydrogen atom.11 (8.5. a transformation moving the x3 –axis into the x2 – 2 axis.11 such transformation is provided through π π (j) Ψ (j. 5.55) confines the particle to stationary bound states. m.288)] 1 ˆ J3 = 2 a† a1 − a† a2 1 2 (8. The required rotation must transform the x3 –axis into the x2 –axis.9.5. b− + − notation in Sects. a1 . 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 . m . m. Eq. 5. This implies that ˆ we can obtain eigenstates of L3 by rotation of the states Ψ(j. We assume that the scalar potential V in (8.101) 2 2 where Dmm ( π . (8. 5. except for a factor 2 .5.10. x1 . which corresponds there to the angular momentum in the x2 –direction. x2 ) . is in the notation of the present section 1 ˆ a† a2 − a† a1 . i. a2 1 2 present notation ←→ b† . the second rotation a factor dmm ( π ). 5.98) then the states Ψ(j. x1 . thus. Using the explicit form of the Wigner rotation matrix as given in (5..8. x2 ) defined in (8. a† .. is identical to the operator L3 introduced in (8. (8.97).5. 5. If we identify 2 a† . The external radiation field is accounted for by the vector potential A(r. . The first rotation contributes a factor exp(−im π ).10.10. x2 ) = Dmm ( . The connection with the present problem arises due to the fact that the operator J2 in Sect. x2 ) (8.92) correspond to the eigenstates |Ψ(j.83) and confirmed the eigenvalues stated in (8. 0) Ψ(j. m) in Sect.31).309) yields finally Ψ (j.

105) in further calculations and to expand the wave function in terms of powers of VS (t) in a perturbation calculation. i. the term ∼ p · A(r. t). i.105) as follows4 . we assume n|m = δnm and for the identity ∞ (8. We will see below that the perturbation resulting from a ‘pure’ plane wave radiation field will serve us to describe also the perturbation resulting from a radiation field made up of a superposition of many planar waves.107) 1 = 1 n=0 |n n| .42)..109) The reader should note that the estimates are very crude since we are establishing an order of magnitude estimate only.104) (8. i. 2 . This result will allow us to neglect the larger than the second term.105) where A(r..e.41) e ˆ e p2 p·A ∼ 2me me me 2me 4 1 2 8πNω . one can estimate that the perturbation. . The states |n are thought to form a complete. t) + A (r. However. For an electron charge q = −e and an electron mass m = me one can provide the estimate for the first term of (8.106) where we adopted the Dirac notation for the states of the quantum system. the term ∼ A second term in (8. Estimate of the Magnitude of VS We want to demonstrate now that the interaction VS (t).105).e. and that the first term in (8. Here the so-called unperturbed system is governed by the Hamiltonian Ho with stationary states defined through the eigenvalue problem Ho |n = n |n . in this case. (8. . n = 0. We first note. is much 2 (r. t) − m 2m (8.. can be considered a weak perturbation. t) is given by (8.e. orthonormal basis. this assumption can be waved as our results below will not depend on it. .216 Interaction of Radiation with Matter be expanded through Fourier analysis in terms of such plane waves. 1.103) (8. using (8.105) for the case of radiationinduced transitions in atomic systems.. we discount the possibility of a continuum of eigenstates. (8.108) We assume for the sake of simplicity that the eigenstates of Ho can be labeled through integers. ωV (8. The Hamiltonian of the particle in the radiation field is then described through the Hamiltonian H Ho VS = = = Ho + VS ˆ2 p + qV 2m q ˆ q2 2 p · A(r. t). as given in (8. In fact. i. is much smaller than the eigenvalue differences near ˆ typical atomic bound states.e.

hence. Assuming a single photon.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.105) due to a radiation field by VS = − q ˆ p · A(r. and me c2 ≈ A A 500 keV 2 ao ω ≈ 10−8 (8. one can neglect this term as long as the first term gives non-vanishing contributions. and as long as the photon densities Nω /V are small.111) For ω = 3 eV and a corresponding λ = 4000 ˚ one obtains. (8. 2me (8. Consequently. one obtains for (8. (8. m (8. replace the perturbation (8..105).e.105) for radiation fields can be considered a small perturbation.110) where ao is the Bohr radius.114) Employing for the second factor the estimate as stated in (8.116) This term is obviously much smaller than the first term. altogether. e ˆ p · A ∼ 10 eV · 10−4 = 10−3 eV . the first term in (8.ˆ u 4πNk α(k. me (8. Using again (8.115) (8.42) it holds VS ≈ e m k.e.5 ˚. We want to estimate now the second term in (8.. We can.8. a volume V = λ3 where λ is the wave length corresponding to a plane wave with frequency ω. t) .112) π λ me c2 and with e2 /ao ≈ 27 eV. with ao ≈ 0.109) 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. λ = 2πc/λ.112) yields e2 A2 ∼ 10 eV · 10−8 = 10−7 eV .117) In case that such perturbation acts on an electron and is due to superpositions of planar waves described through the vector potential (8. Hence.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. i.118) . Nω = 1. u) p · u exp i(k · r − ωt) ˆ ˆ ˆ kV .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.

The sum runs over all possible k vectors and might actually be an integral. Equation (8.120) (8. Multiplying the latter equation by the operator exp i (8.108) as the eigenstates of the unperturbed Hamiltonian Ho . One seeks to predict the probability to observe the particle in one of the states |n .126) Ho (t − to ) yields finally . A factor α(k. . 2. The probability to find the particle in the state |n at time t is then p0→n (t) = | n|ΨS (t) |2 . 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 . Using i i ∂t exp − Ho (t − to ) and (8.122) where |ΨD (to ) = |0 . under ordinary circumstances.125) (8. n = 1. describes the effect of a radiation field on an electron system and which will be assumed below to describe radiative transitions. 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 . (8. n = 0 at some later time t ≥ to .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) (8.218 Interaction of Radiation with Matter where we have replaced ω in (8.123) i = Ho exp − Ho (t − to ) (8. |n are defined in (8. the sum over u involves the two possible polarizations of planar ˆ electromagnetic waves.42) through k = |k| = ω.127) i ∂t |ΨD (t) = VD (t) |ΨD (t) . . |Ψ(to ) = |0 (8. For this purpose one needs to determine the state |ΨS (t) of the particle.106–8. . The states |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) . u) has been added to describe eliptically or circularly polarized ˆ waves.118) is the form of the perturbation which.121) (8..124) (8.

137) |ΨD (t) = (1) (0) (8. we define |ΨD (t) through the evolution equations i ∂t |ΨD (t) i ∂t |ΨD (t) i (2) ∂t |ΨD (t) (1) (0) = = = .8. 8. = . We will consider here only the two leading contributions to |ΨD (t) . Due to the Hermitean property of the Hamiltonian Ho holds n|Ho = i n|exp − Ho (t − to ) from which we conclude. In order to determine |ΨD (t) described through (8. 8.130) = exp − i n (t − to ) − to )]| = 1.137) can be solved recursively. to (8. . .121) expressed in terms of ΨD (t) is i p0→n (t) = | n|exp − Ho (t − to ) |ΨD (t) |2 . Accord(n) ingly.135) i ∂t |ΨD (t) (n) VD (t) |ΨD (n−1) (t) (8. Employing this result one obtains for (8.131) p0→n (t) = | n |ΨD (t) |2 . .134. 8. Equations (8.132) where |ΨD (t) accounts for the contribution due to n-fold products of VD (t) to |ΨD (t) .127. using |exp[− i n (t n (8. 2. n| (8.127) we assume the expansion ∞ |ΨD (t) = n=0 (n) |ΨD (t) (n) (8.133–8.137) One can readily verify that (8. (8. .139) .133) (8. . (8.134) (8. From (8.5: Time-Dependent Perturbation Theory where VD (t) = exp 219 i i Ho (t − to ) VS (t) exp − Ho (t − to ) . consequently.128) We note that the transition probability (8.128).129) n| and.138) 1 i t dt VD (t ) |0 .136) together with the initial conditions |ψD (to ) = |0 0 for n = 0 for n = 1.133.137) are consistent with (8. . 0 VD (t) |ΨD (t) (1) VD (t) |ΨD (t) (0) (8.137) follows |ΨD (t) = |0 . (8.132–8. .

This factor will serve mainly the purpose of keeping in the following derivation all mathematical quantities properly behaved. that two planar waves simulataneously interact with an electronic system. (8.142). The coefficients A1 . according to (8. the direction of ˆ the propagation being determined by k = k/|k|. + m=0 dt to to dt n|VD (t )|m m|VD (t ) |0 8.144) describes the propagation of the planar wave. non-singular.141) + .220 This result. defines the direction of the E-field of the ˆ radiation. λ → 0+ .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 ..34. λ → 0+ has been introduced which describes that at time to → −∞ the perturbation is turned on gradually.41). . such that the combined radiation field is decribed by the vector potential A(r. II III (8. 8. yields for (8. For the sake of including the effect of superpositions of plane waves we will assume.142) combining an incoming and an incoming or outgoing wave.e. We want to apply now the perturbation expansion derived to such perturbations.118).137) |ΨD (t) = (2) Interaction of Radiation with Matter 1 i 2 t t dt to to dt VD (t ) VD (t ) |0 . however. A2 are defined through (8.. The factor III in (8. in turn. 8. A factor exp(λt). according to (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).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. i. 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..144) r is the position of the electron and p = ( /i) is the momentum operator of the electron.6 Perturbations due to Electromagnetic Radiation We had identified in Eq.140) Altogether we have provided the formal expansion for the transition amplitude n|ΨD (t) = n|0 + ∞ 1 i 1 i t dt n|VD (t ) |0 to 2 t t (8. In (8. t) = + A1 u1 exp i (k1 · r − ω1 t) ˆ A2 u2 exp i (k2 · r ˆ ω2 t) incoming wave incoming or outgoing wave (8. The resulting perturbation on an electron system. is VS = ˆ ˆ V1 exp(−iω1 t) + V2 exp( iω2 t) eλt . note that uj . to → −∞ (8.135.

140.145) we obtain then.143).141). using (8. For the 1st order contribution to the transition amplitude n|ΨD (t) = (1) 1 i t dt n|VD (t ) |0 to (8.128).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 . employing the perturbation (8.150) and.6: Perturbations due to Electromagnetic Radiation 1st Order Processes 221 We employ now the perturbation (8. (8.149) Using the definition of VD stated in (8.145) n|ΨD (t) (1) m (t − to ) |m . (8.8. (8.143) to the expansion (8.128) one obtains k|VD (t) | = k|VS (t)| exp i ( k − ) (8.146) = 1 λ→0+ t→−∞ i lim lim × t dt exp to i ( n − o − i λ) t iω2 t × . According to (8.151) ˆ ˆ n|V1 |m m|V1 |0 exp ˆ ˆ + n|V2 |m m|V2 |0 exp i i n − − m − ω1 − i λ)t exp ω2 − i λ)t exp m o n m m o . 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 (8.141) this is n|ΨD (t) = − (2) 1 2 ∞ m=0 t t dt to to dt n|VD (t ) |m m|VD (t ) |0 .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. (8. 8. (8.

222 ˆ ˆ + n|V1 |m m|V2 |0 exp ˆ ˆ + n|V2 |m m|V1 |0 exp i i ( ( − − Interaction of Radiation with Matter − ω1 − i λ)t exp ω2 − i λ)t exp i i ( ( − − ω2 − i λ)t − ω1 − i λ)t n m m o n m m o Carrying out the integrations and the limit limt→−∞ provides the result (2) n|ΨD (t) = − 1 ∞ 2 λ→0+ lim (8.148) and ∗ |z1 + z2 |2 = |z1 |2 + |z2 |2 + 2Re(z1 z2 ) (1) (8. We also assume for the final state n = 0 such that n|0 = 0 and p0→n (t) = | n|ΨD (t) |2 holds.152).152) m=0 i ˆ ˆ n|V1 |m m|V1 |0 exp m − o − ω1 − i λ + − 2 ω1 − 2i λ)t n − o − 2 ω1 − 2i λ ( n o − ˆ ˆ n|V2 |m m|V2 |0 exp ω2 − i λ m− o ˆ1 |m m|V2 |0 exp ˆ n|V + ω2 − i λ m− o ˆ2 |m m|V1 |0 exp ˆ n|V + m − o − ω1 − i λ 1st Order Radiative Transitions i 2 ω2 − 2i λ)t 2 ω2 − 2i λ n− o i ( n − o − ω1 ω2 − 2i λ)t ω2 − 2i λ n − o − ω1 ( n o i − − ω1 ω2 − 2i λ)t ω2 − 2i λ n − o − ω1 ( n o − The 1st and 2nd order transition amplitudes (8.e.131). in which case one can expect that it is larger than the 2nd (2) order contribution n|ΨD (t) which we will neglect..153) (8. i. (8.148) and (8. provide now the transition probability p0→n (t) according to Eq. the time derivative of p0→n (t). respectively. Using (8.155) λ→0+ ( o We are actually interested in the transition rate.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. For this rate holds .

158) ˆ + | n|V2 |0 |2 δ( − ω2 ) Obviously.142) ˆ ˆ The matrix elements n|V1 |0 and n|V2 |0 in (8.160) leads to final states |n with energy n = o ± ω2 . (8. The result supports our definition of incoming and outgoing waves in (8. a property. describing either an incoming or an outgoing plane wave due to the vector potential A2 u2 exp i (k1 · r ˆ ω2 t) .e..142–8. First.144).8. 8. The δ-functions appearing in this expression reflect energy conservation: the incoming plane wave contribution of (8. however. The result of such average is.158).156) ( o The period of electromagnetic radiation absorbed by electronic systems in atoms is of the order 10−17 s. the two terms apearing on the rhs.31) and (8. i.159) leads to final states |n with energy n = o + ω1 . The second contribution to (8. one should average the rate over many periods of the radiation.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. For the resulting expression the limit limλ→0+ can be taken. to cancel the third term in (8. due to the vector potential A1 u1 exp i (k1 · r − ω1 t) . is much shorter than could be resolved in any observation. of this expression describe the individual effects of the two planar wave contributions of the perturbation (8. In case the matrix elements are non-zero. 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. in fact. the matrix elements can vary strongly for different states |n of the system.156) such that the 1st order contributions of the two planar waves of the perturbation simply add. Using lim x2 + 2 δ→0+ = π δ(x) (8.143. Hence. ˆ (8. any attempt to do so. The time average will be denoted by · · · t .158) play an essential role for the transition rates of radiative transitions. due to the uncertainty relationship would introduce a considerable perturbation to the system. which is observed through the so-called spectral intensities of transitions |0 → |n .144). .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.

e. i.224 2nd Order Radiative Transitions Interaction of Radiation with Matter We now consider situations where the first order transition amplitude in (8... in Eqs. i.g. 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. We define ∞ z1 = m=0 ˆ ˆ n|V1 |m m|V1 |0 m − o − ω1 − i λ exp i × (8.158).164) (8.152). ∞ − 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 × (8.162) × and.166) over many periods of the radiation yields exp[ i (±ω2 − ω1 )] t = 0 and.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.. as we did in the the case of 1st order transitions.k=1 j=k 2 2 2 2 ∗ zj zk (8. (8.161) To determine the transition rate we proceed again.e. it holds p0→n (t) = | n|ΨD (t) |2 .163) × (8.157) yields. |z1 + z2 + z3 |2 t = |z1 |2 + |z2 |2 + |z3 |2 (8. hence.153–8. (2) (8. Time average · · · t of expression (8.167) Taking now the limit limλ→0+ and using (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.168) absorption of 2 photons ω1 .153) vanishes such that the leading contribution to the transition probability p0→n (t) arises from the 2nd order amplitude (8. similarly. in analogy to (8.158). exp[ i (±ω2 − ω1 )].

In case of the first term in (8. |m ← |0 (8. i.e. describe the absorption/emission of two photons. there is no energy conservation necessary (in general..168) describes two intermediate histories.e. i. the third term describes processes in which a photon of energy ω1 is absorbed and a second photon of energy ω2 is absorbed/emitted. describe the absorption of two photons.168) describe the time sequence of the two photon absorption/ emission processes. accordingly.e. this state is only virtually excited.168) describe similar histories of the excitation process. vice versa. it must hold n = o + 2 ω1 . Similarly. first absorption of photon ω1 and then absorption/ emission of photon ω2 . Most remarkably. 8. according to its δ-function factor. a second perturbation. the system is perturbed through absorption of a photon with energy ω1 from the initial state |0 into a state |m . the second term describes the processes leading to final states |n with energy n = o ± 2 ω2 and.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.e. 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. (8.. The factors | · · · |2 in (8..7: One-Photon Emission and Absorption 2π ∞ m=0 225 2 + ˆ ˆ n|V2 |m m|V2 |0 ω2 − i λ m− o δ( m − o 2 ω2 ) absorption/emission of 2 photons ω2 + 2π ∞ m=0 ˆ ˆ n|V2 |m m|V1 |0 + m − o − ω1 − i λ + ˆ ˆ n|V2 |m m|V1 |0 ω2 − i λ m− o 2 δ( n − o − ω1 ω2 ) absorption of a photon ω1 and absorption/emission of a photon ω2 This transition rate is to be interpreted as follows. . The expression sums over all possible virtually occupied states |m and takes the absolute value of this sum.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. promotes the system then to the state |n . i. The first term.158) which accounts for such transitions. interference between the contributions from all intermediate states |m can arise. m = o + ω1 ) and the evolution of state |m is described by a factor 1/( m − o − ω1 −i λ). accordingly. The remaining two contributions in (8. i.169) pert. each of energy ω1 . each of energy ω2 . through absorption of a photon. For this purpose we will employ the transition rate as given in Eq. Similarly. the third term in (8. which is stationary and energy is conserved.. |n ← |m .8.

(8. the wave functions are (n. 8.179) . φ) × × eik·r e−r/ao Y00 (θ.2 = m ωV Only the first term of (8. ˆ (8. φ) (8. 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 . plane polarized wave described through the complex vector potential A(r.144) VS = where e 2πN ˆ ˆ p · u e±ik·r . (8. the vector potential actually assumed is A(r.m=0 (r. θ. according to (8. (8.177) √ M = 6 0 ∞ 1 2π ia4 o r2 dr −1 dcosθ 0 ∗ dφ r e−r/2ao Y1m (θ. φ) a3 o 1s (8.178) These wave functions make significant contributions to this integral only for r-values in the range r < 10 ao . θ. λ → 0+ . to → −∞ (8. φ) a3 ao o 2p (8. φ) = 6 r −r/2ao e Y1m (θ.172) δ( n − o − ω) (8.170) We will employ only the real part of this potential. However. and the integral is =1. the second term can be discounted in case of absorption. m denote the relevant quantum numbers) ψn=1. =0.e. t) = 2πN ωV u exp ˆ ı (k · r − ωt) + 2πN ωV u exp ˆ ı (−k · r + ωt) . φ) = 2 − 1 2 1 −r/ao e Y00 (θ. . The absorption rate.143.171) The perturbation on an atomic electron system is then according to (8.158).226 Absorption of a Plane Polarized Wave Interaction of Radiation with Matter We consider first the case of absorption of a monochromatic.143) will contribute to the absorption process. i.176) ψn=2. in this range one can expand eik·r ≈ 1 + ik · r + .173) V1. in case of a radiative transition from the 1s state of hydrogen to one of its three 2p states.. . . For example.175) The matrix element involves a spatial integral over the electronic wave functions associated with states |n and |0 . t) = 8πN ωV u exp ˆ ı (k · r − ωt) .m (r.174) We seek to evaluate the matrix element ˆ M = n| p eik·r |0 . (8.

180) ∆E2p−1s and ao = 0.529 ˚ one concludes that in the significant integration range in (8. pk ] pk ˆ k=1 (8. Ho ] = ˆ2 p [ r. One refers to this approximation as the dipole approximation. in case of the hydrogen atom.179) and other terms are never larger than 20π ao /λ.178) holds eik·r ≈ A 1 1 + O( 50 ) such that one can approximate eik·r ≈ 1 .181) and (8.186) δ( n − o − ω) (8. We are now in a position to obtain an alternative expression for the matrix element (8. Transition Dipole Moment A further simplification of the matrix element (8. the value of the wave length for the 1s → 2p transition 2π c λ = = 1216 ˚ A (8.183) (8.175). ] + [ r.187) .174) yields kabs = 4π 2 e2 N ω V u · n| ˆ |0 ˆ r 2 m( o − m n| [r.8. Ho ] i where Ho is the Hamiltonian given by (8. (8. pj ] = i δkj one obtains ˆ [ r. Ho ] |0 = i i n) n|r |0 .182) (8.184) and the commutation property [xk .k=1 i m 3 pk ek = ˆ j.7: One-Photon Emission and Absorption 227 One can estimate that the absolute magnitude of the second term in (8. V (r) ] 2me =0 = Using r = 3 ˆ j=1 xj ej 1 2me 3 pk [ r. Ho ] = i m 3 pk ej δjk = ˆ j.185) from which follows (8. Using (8. is Ho = ˆ (p)2 e2 + V (r) . This simplification results from the identity m ˆ p = [ r.182).181) For the commutator in (8. 2me r (8.182) one finds [ r.182) one obtains M ≈ Insertion into (8. pk ] + ˆ ˆ k=1 1 2me 3 [ r.175) can then be ˆ achieved and the differential operator p = i replaced by by the simpler multiplicative operator r. Using |k| = 2π/λ.k=1 i ˆ p m (8. V (r) = − .104) and.

The unit ˆ ˆ ˆ ˆ can be chosen to point along the x1 .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. u2 and k ˆ ˆ coordinate system. x3 -axes of a right-handed cartesian vectors u1 . a very hot oven. u2 .35). We can.31–8. k ˆ As a result.e. ϕ1 = 0 and of u2 by ϑ2 = π/2. ρ) holds then u cosα = cosϑ1 cosϑ + sinϑ1 sinϑ cos(ϕ1 − ϕ) = sinϑcosϕ and cosβ = cosϑ2 cosϑ + sinϑ2 sinϑ cos(ϕ2 − ϕ) = sinϑsinϕ . Accordingly. one can express | u · n| r |0 |2 = |ρ|2 ( cos2 α + cos2 β ) = sin2 θ . obtain the transition rate in the present case by adding the individual transition rates of all planar waves present in the oven. 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. the direction of u1 is described by the ˆ ˆ ˆ ˆ angles ϑ1 = π/2.187) 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. directions. hence. ˆ u ˆ (8. ϕ.190) (8. We have demonstrated in our derivation of the rate of one-photon processes (8. |0 . 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.33) are orthogonal to each other. in practical situations.193) .158) above that in first order the contributions of all components of the radiation field add. and (2) the fact that strictly monochromatic radiation cannot be prepared such that any radiation source provides radiation with a frequency distribution.191) (8. k frame is described by the angles ϑ. x2 . three-dimensional vector. ˆ In order to carry out the integral dk we note that u describes the possible polarizations of the ˆ ˆ and u. according to (8.228 Interaction of Radiation with Matter where we used the fact that due to the δ-function factor in (8. Integrating and summing accordingly over all contributions as given by (8. The δ-function appearing in this expression. ρ) and ˆ u β = ∠(ˆ2 . there are ony two linearly independent directions of u possible. will actually be replaced by a distribution function which reflects (1) the finite life time of the states |n . ϕ2 = π/2.. Absorption of Thermal Radiation We want to assume now that the hydrogen atom is placed in an evironment which is sufficiently hot. planar waves as defined in (8. For the two angles α = ∠(ˆ1 .188) ˆ Here dk is the integral over all orientations of k. The direction of this vector in the u1 .174) one can replace n − o by ω. and all polarizations of the radiation. say u1 and u2 . such that the thermal radiation present supplies a continuum of frequencies.189) where the factor 1/ arose from the integral over the δ-function. i.

In case of a classical radiation field one would expect that emission cannot occur.172.7: One-Photon Emission and Absorption 229 This geometrical average. The most intensely absorbing molecules are long. the calculation of the emission rate proceeds as in the case of absorption. can be inserted into (8.. for Nω = 0. The radiation field is described. as for the absorption process.173) contributes. Transition Dipole Moment The expression (8. The transition dipole moment can vanish for many transitions between stationary states of a quantum system. Emission of Radiation We now consider the rate of emission of a photon. a quantum mechanical treatment of the radiation field leads to a total emission rate which is proportional to Nω + 1 where Nω is the number of photons before emission. in agreement with observations. (8.195) = Nω 1. However. . 8. in particular. ˆ In case of emission only the second term V2 exp(+iωt) in (8.37 × 1019 × 3 3 c3 3 ao λ3 s λ and kabs where λ = 2πc n − (8. a problem in case of emission by quantum systems in complete darkness.194) for the absorption rate shows that the essential property of a molecule which determines the absorption rate is the so-called transition dipole moment | n| r |0 |. i. however. 2 .8. for atoms or symmetric molecules. there is also a contribution to the emission rate which is proportional to Nω .e. by planar waves with vector potential (8. linear molecules. . 1. Nω photons before absorption. (8.171) and perturbation (8. that emission occurs even if no photon is present in the environment.173). The quantum nature of radiation manifests itself in that the number of photons Nω msut be an integer. For this purpose we use ω/c = 2π/λ and obtain 4 e2 ω 3 32π 3 e2 ao 1 ao = = 1.189) to yield the total absorption rate kabs (tot) = Nω 4 e2 ω 3 | n|r|0 |2 3 c3 .e. the resulting total rate of emission bears a different dependence on the number of photons present in the environment..194) For absorption processes involving the electronic degrees of freedom of atoms and molecules this radiation rate is typicaly of the order of 109 s−1 . i. The value of | n| r |0 | determines the strength of an optical transition. For practical evaluations we provide an expression which eliminates the physical constants and allows one to determine numerical values readily.. The corresponding process is termed spontaneous emission. Accordingly. Otherwise. This difference between emission and absorption is due to the quantum nature of the radiation field. Nω = 0. This dependence predicts.196) The last two factors in (8.37 × 1019 1 ao | n|r|0 |2 × s λ λ2 . This poses. However.194) combined are typically somewhat smaller than (1 ˚/1000 ˚)3 = A A 10−9 . finally. the absorption rate is of the order of 109 s−1 or 1/nanosecond. .197) o (tot) (8. However.

Let No and Nn denote the number of atoms in state |0 and |n .198) (8.201) ksp = 3 c3 the rates of absorption and emission are Nω ksp and (Nω + 1)ksp . e.e.195) and (8.200) where kB is the Boltzmann constant. For these numbers holds Nn / No = exp[−( n − o )/kB T ] (8.200)..230 Interaction of Radiation with Matter which is termed induced emission since it can be induced through radiation provided. The number of atoms undergoing absorption in unit time are Nω ksp No and undergoing emission are (Nω +1)ksp Nn . it must hold Nω ksp N0 = (Nω + 1) ksp Nn (8. In the following we discuss several examples.198) is consistent with Planck’s radiation law which reflects the (boson) quantum nature of the radiation field. To demonstrate this property we apply the transition rates (8. = (8. Examples are radiative transitions in which two photons are absorbed or emitted or scattering of radiation by matter in which a photon is aborbed and another re-emitted. is (tot) kem 4 e2 ω 3 | n|r|0 |2 (spontaneous emission) 3 3c 4 e2 ω 3 | n|r|0 |2 (induced emission) + Nω 3 3c 4 e2 ω 3 = ( Nω + 1 ) | n|r|0 |2 3 3c Nω photons before emission.g. Hence. Defining the rate of spontaneous emission 4 e2 ω 3 | n|r|0 |2 (8. . exp[− ω/kB T ] = This equation yields Nω = 1 .202) It follows.198) to determine the stationary distribution of photons ω in an oven of temperature T . using (8. respectively. the well-known Planck radiation formula. The total rate of emission. Nω + 1 (8. We assume n − o = ω.8 Two-Photon Processes In many important processes induced by interactions between radiation and matter two or more photons participate. respectively. exp[ ω/kB T ] − 1 (8.203) i. 8. accordingly..199) Planck’s Radiation Law The postulate of the Nω + 1 dependence of the rate of emission as given in (8.204) Nω . 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 . in lasers.

induce single photon absorption processes as described by the transition rate Eq. which is not necessarily the case. . The intense radiation fields of lasers allow one to increase transition rates to levels which can readily be observed in the laboratory. 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. a transition between the states |0 and |n may be possible. 8. The perturbation which accounts for the coupling of the electronic system and the radiation field is the same as in case of 1st order absorption processes and given by (8. We assume that a planar wave with wave vector k1 and polarization u1 . One obtains.205) The respective radiative transition is of 2nd order as described by the transition rate (8.206) − o − 2 ω) . (8. however. under ordinary circumstances.207) Expression (8. k = Nω V 2 8π 3 e4 × δ( m− ∞ m=0 o ( n − ˆ m) u · n|ˆ |m ( m − o ) u · m|ˆ |0 r ˆ r ω( m − o − ω − i λ) 2 − 2 ω) . finally. even if n| r |0 vanishes. dropping the index 1 characterizing the radiation. Employing the dipole approximation (8. In this case one needs to choose the energy of the photons to obey n = o + 2 ω.207) one is. (8. as described through the vector potential ˆ A(r.168) where the first term describes the relevant contribution.104) with stationary states |n defined through (8. ˆ (8. The transition requires that the transition dipole moment n| r |0 does not vanish for two states |0 and |n .173).207) does not provide a suitable avenue of computing the rates of 2-photon transitions. However.8.207) does not converge rapidly.172. (8. We wish to describe an observation in which a detector is placed at a solid angle element dΩ2 = sinθ2 dθ2 dφ2 with respect to the origin of the coordinate system in which the electron is described. t) = Ao1 u1 cos(k1 · r − ω1 t) . The resulting rate of the transition depends on 2 Nω . faced with the dilemma of having to sum over all intermediate states |m of the system. The electron absorbs the radiation and emits immediately a second photon.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. in case ˆ of absorption only V1 contributes. In applying (8.106).182) yields. If the sum in (8.181) and using (8.8: Two-Photon Processes Two-Photon Absorption 231 The interaction of electrons with radiation.187). but then requires the absorption of two photons. then expression (8. however.208) has been prepared. 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.

210). The relevant terms of the perturbation (8.211) are given by VS (t) = e ˆ p · A+ u1 exp[i(k1 · r − ω1 t)] + A− u2 exp[−i(k2 · r − ω2 t)] o1 ˆ o2 ˆ 2me contributes in 2nd order + e2 4me A+ A− u1 · u2 exp{i[(k1 − k2 ) · r − (ω1 − ω2 ) t]} ˆ o1 o2 ˆ contributes in 1st order The effect of the perturbation on the state of the electronic system is as stated in the perturbation expansion (8.210) A(r.210) is accounted for by the following contributions of (8.232 Interaction of Radiation with Matter We assume that the experimental set-up also includes a polarizer which selects only radiation with a certain polarization u2 . The vector potential describing the emitted plane wave is then A(r.208) and the emission of the radiation in (8.211) is stated in Eq. the amplitude A+ is given by o1 A+ = o1 8πN1 ω1 V (8.6 above that the absorption of the radiation in (8. We know already from our description in Section 8. u1 .208.210) The vector potential which describes both incoming wave and outgoing wave is a superposition of the potentials in (8.208). The second term in (8. the amplititude A− defined in (8. This expansion yields. i.209) cosθ2 where the value of k2 has been fixed.141). according to the ˆ description of emission processes on page 229. (8.105).e..211) is o2 A− = o2 8π (N2 + 1) ω1 V (8.211) The first term describes the absorption of a photon and. in the present case. later we will allow the quantum system to select appropriate values. t) = A+ u1 exp[ i (k1 · r − ω1 t) ] + A− u2 exp[ i (k2 · r − ω2 t) ] .212) where N1 /V is the density of photons for the wave described by (8. ˆ (8. Let us assume for the present that the emitted photon has a wave vector ˆ k2 with cartesian components   sinθ2 cosφ2 k2 = k2  sinθ2 sinφ2  (8. t) = Ao2 u2 cos(k2 · r − ω2 t) . 8. hence. o1 ˆ o2 ˆ (8. ˆ The perturbation which arises due to the vector potential (8.105) using the vector potantial (8. the wave characterized through k1 . for the components of the wave (8. In the present case we consider only scattering processes which absorb radiation corresponding to the vector potential (8.213) where N2 /V is the density of photons characterized through k2 .208) and emit radiation corresponding to the vector potential (8. u2 .210). 8.208.214) . however.211) accounts for the emitted wave and.

i.215) correspond to scattering processes in which the radiation field ‘looses’ a photon ω1 and ‘gains’ a photon ω2 . one needs to carry out the integration 2 V(2π)−3 k2 dk2 · · ·. one needs to integrate the rate as given by (8. Accordingly.e.. Hence. me c2 (8. i.212.218) ..216) over all available modes of the field.213) for the amplitudes A+ and A− results o1 o2 in the Kramers-Heisenberg formula for the scattering rate k = N1 c 2 ω2 r (N2 + 1) dΩ2 u1 · u2 n|0 ˆ ˆ V o ω1 1 n|ˆ2 · p |m m|ˆ1 · p |0 u ˆ u ˆ n|ˆ1 · p |m m|ˆ2 · p |0 u ˆ u ˆ − + me m m − o + ω2 m − o − ω1 (8.8 · 10−15 m . only these two terms contribute to the scattering amplitude. Inserting also the values (8. evaluating the time integrals in (8.215) t dt ei( to n− o− ω1 + ω2 +i λ)t 1 i 2 e2 4m2 e A+ A− × o1 o2 × ˆ ˆ u1 · n| p |m u2 · m| p |0 × ˆ ˆ t t × to dt to dt ei( n− m− ω1 +i λ)t i( e m− o+ ω2 +i λ)t ˆ ˆ + u2 · n| p |m u1 · m| p |0 × ˆ ˆ t t × to dt to dt ei( n− m+ ω2 +i λ)t i( e m− o− ω1 +i λ)t We have adopted the dipole approximation (8. Following closely the procedures adopted in evaluating the rates of 1st order and 2nd order radiative transitions on page 222–225.181) in stating this result.e.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 ∞ (8.216) e2 + − A A 4me o1 o2 n|ˆ1 · p |m m|ˆ2 · p |0 u ˆ u ˆ n|ˆ2 · p |m m|ˆ1 · p |0 u ˆ u ˆ + m − o + ω2 m − o − ω1 We now note that the quantum system has the freedom to interact with any component of the radiation field to produce the emitted photon ω2 .215) and taking the limits limto →−∞ and limλ→0+ yields the transition rate k − m = 2π δ( n − o − ω1 + ω2 ) e2 A+ A− u1 · u2 n|0 ˆ ˆ 4m2 o1 o2 e 2 (8.217) 2 Here ro denotes the classical electron radius ro = e2 = 2.8. Only the second (1st order) and the third (2nd order) terms in (8. 8.

− o )2 ( m − o )3 (8. one assumes the even stronger condition ω1 << | o − m| . . 2 .228) S (0) = m S (0) = 2 m . for all states |m of the electronic system (8. a process in which the electronic state remains unaltered after the scattering. ω1 = ω2 .227) u ˆ 0|ˆ1 · p |m m|ˆ2 · p |0 u ˆ u ˆ 0|ˆ2 · p |m m|ˆ1 · p |0 u ˆ + me ( m − o )3 me ( m − o )3 (8..223) m Condition (8. Rayleigh Scattering We turn first to an example of so-called elastic scattering. . in fact..221) Using |n = |0 and.225) S(0) = m u ˆ 0|ˆ1 · p |m m|ˆ2 · p |0 u ˆ u ˆ 0|ˆ2 · p |m m|ˆ1 · p |0 u ˆ + me ( m − o ) me ( m − o ) (8.221) suggests to expand S( ω) S( ω) = S(0) + S (0) ω + Using 1 − o ± ω one can readily determine m (8. Rayleigh scattering is defined as the limit in which the wave length of the scattered radiation is so long that none of the quantum states of the electronic system can be excited. one can relate (8. it follows 2 dσ = ro (N2 + 1) dΩ2 |ˆ1 · u2 − S( ω) |2 u ˆ (8.220) n|ˆ2 · p |m m|ˆ1 · p |0 u ˆ u ˆ + m − o − ω1 2 − − o + ω2 In the following we want to consider various applications of this formula.224) = m 1 − o ( m ω ( ω)2 + + .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. (8.222) where S( ω) = 1 me 0|ˆ1 · p |m m|ˆ2 · p |0 u ˆ u ˆ 0|ˆ2 · p |m m|ˆ1 · p |0 u ˆ u ˆ + m − o + ω m − o − ω 1 S (0)( ω)2 + . consequently.219) (8. i. Accordingly..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.e.234 Interaction of Radiation with Matter The factor N1 c/V can be interpreted as the flux of incoming photons.

108) for the identity operator.236) is of great practical importance. ˆ ˆ S (0) given in (8.k=1 j.228) provides then the first non-vanishing contribution to the scattering cross section (8. the blue color of the sky and the polarization pattern in the sky which serves many animals. pk ] = i δjk yields finally ˆ S(0) = 1 S(0) = i 3 3 (8. u1 ·r ] |0 = 0 u ˆ (8. For this purpose we replace p using (8.222).232. ˆ We want to simplify first (8.231) (ˆ1 )j (ˆ2 )k 0|[xj . (8. . ˆ ˆ We want to prove now that expression (8.224.226) 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 .226).228) we obtain ˆ ˆ ˆ ˆ S (0) = 2me 2 m 0|ˆ2 ·r |m m|ˆ1 ·r |0 u u 0|ˆ1 ·r |m m|ˆ2 ·r |0 u u + m − o m − o (8.229) (8.232) Obviously.8. for example. pk ]|0 = u u ˆ j. u ˆ ˆ ˆ ˆˆ i The commutator property [xj . u u u u (8. 8.e.233) m Employing again (8.234) where we used for the second identity the fact that u1 ·r and u2 ·r commute.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 (8.108) this is 1 0|ˆ1 ·r u2 · p − u2 · p u1 ·r |0 . i.k=1 (ˆ1 )j (ˆ2 )k δjk = u1 · u2 u u ˆ ˆ (8.230) According to (8. 8.. this term cancels the u1 · u2 term in (8.108) yields S (0) = me 2 0| [ˆ2 ·r.182) 0|ˆ1 · p |m u ˆ 1 = 0|ˆ1 ·r |m u me ( m − o ) i This transforms (8.229) both to u1 · p and to u2 · p which results in ˆ ˆ ˆ ˆ me S (0) = 2 ( 0|ˆ2 ·r |m m|ˆ1 ·r |0 − 0|ˆ1 ·r |m m|ˆ2 ·r |0 ) .222) of the cross section for Rayleigh scattering dσ = 2 ro m2 ω 4 (N2 + 1) dΩ2 × e (8. as a compass.235) and obtain the leading contribution to the expression (8.182) for p and the expression (8. Expression (8.235) We can now combine eqs.234. 8. Using again (8.8: Two-Photon Processes 235 ˆ These three expressions can be simplified using the expression (8. m|ˆ1 · p |0 u ˆ 1 = − m|ˆ1 ·r |0 u me ( m − o ) i (8.227) vanishes. It explains.229) both for the u1 · p and the to u2 · p terms in (8. honey bees.222).

the emitted u ˆ radiation is decribed by the wave vector   sinθ2 cosφ2 k2 = k1  sinθ2 sinφ2  (8.. u2 = u2 ˆ ˆ ˆ ˆ (1) (2) (2) (1) (1) (2) (1) (2) (2) (1) needs to be (8.244) . (8. u ˆ 1 ao . u2 = u2 ˆ ˆ ˆ ˆ for u1 = u1 .237) The resulting scattering is called Thomson scattering. for all states |m of the electronic system . 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 . assume now that the scattered radiation has very short wave length such that ω >> | o − m| .240) cosθ2 We choose for the polarization of the incoming radiation the directions along the x1 . however. 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 .and the x2 -axes     1 0 (1) (2) u1 =  0  .241) 0 0 Similarly.220) 2 dσ = ro (N2 + 1) dΩ2 |ˆ1 · u2 |2 .239) We will show below that this expression decribes the non-relativistic limit of Compton scattering. We want to assume. (8.240). hence. u1 =  1  ˆ ˆ (8.238) (8.220). To evaluate |ˆ1 · u2 |2 we assume that k1 is oriented along the x3 -axis and. |n = |0 and ω1 = ω2 = ω in (8.243) cos2 θ2 cos2 φ2 for u1 = u1 . that the dipole approximation is still valid which restricts the applicability of the following derivation to k1 >> One obtains immediately from (8.242) |k2 × k1 | 0 where the second identity follows readily from k1 = e3 and from (8. ao Bohr radius . i. u2 = u2 ˆ ˆ ˆ ˆ (8. though.236 Thomson Scattering Interaction of Radiation with Matter We consider again elastic scattering. u2 = u2 ˆ ˆ ˆ ˆ cos2 φ2 cos2 θ2 sin2 φ2 for u1 = u1 ..e. The first choice is ˆ   sinφ2 k2 × k1 (1) u2 = ˆ =  −cosφ2  (8.

if energy is released. In the case that the scattering excites other than electronic degrees of freedom.249) dσ = ro (N2 + 1) ω1 −1 −1 ω2 − ω1 = . In the case of such scattering an electron system absorbs and re-emits radiation without ending up in the initial state.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.e. 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.. Since the scattering is inelastic.246) 2 This expression is the non-relativistic limit of the cross section of Compton scattering. The Compton scattering cross section which is derived from a model which treats photons and electrons as colliding relativistic particles is dσ tot = dσ tot where (rel) = 2 ro (N2 + 1) 2 ω2 ω1 2 ω1 ω2 + − sin2 θ2 ω2 ω1 dΩ2 (8. the resulting scattering cross section is 2 ro (N2 + 1) ( 1 + cos2 θ2 ) dΩ2 . in case of a diatomic molecule |n = |φ(elect. The radiation scattered at right angles is even completely polarized (1) along the u2 -direction. the states |n etc.247.)n . for c → ∞ the Compton scattering cross section (8. defined in (8.220) represent actually electronic as well as nuclear motions. the first term in (8. but most often involves other degrees of freedom.. If energy is absorbed by the system.246).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 .247) ( 1 − cosθ2 ) (8. φ(vibr. e. The energy deficit is used to excite the system. For electronic systems in molecules or crystals the degrees of freedom excited are nuclear motions. Such scattering is called Raman scattering. The excitation can be electronic. ˆ In case that one measures the scattered radiation irrespective of its polarization.220) vanishes and one obtains in case of Raman scattering ω2 2 dΩ2 | u2 · R · u1 |2 ˆ ˆ (8.248) me c2 One can readily show that in the non-relativistic limit. molecular vibrations or crystal vibrational modes. Raman Scattering and Brillouin Scattering We now consider ineleastic scattering described by the Kramers-Heisenberg formula.e..)n .247) becomes identical with the Thomson scattering cross section (8. one speaks of anti-Stokes scattering. i. 8. i. one speaks of Stokes scattering. the scattering is termed Brillouin scattering. or in case of translational motion of molecules in liquids.8.g. In case that the nuclear degrees of freedom excited absorb very little energy. the scattered radiation is polarized to some degree. (8. as in the case of excitations of accustical modes of crystals.

250) (8.238 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. 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. ω1 ≈ m − o for a particular state |m .g. Of course. a case called resonant Raman scattering.251) ω1 − ( We define x · R · y = j. . the Raman scattering cross section will be much enhanced. e.k xj Rjk yk . Nevertheless. In case that the incoming photon energy ω1 is chosen to match one of the electronic excitations. a property which can be exploited to selectively probe suitable molecules of low concentration in bulk matter..

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

. The elements P ∈ SN can be represented conveniently as 2 × N matrices. number k under the entry j of the first row indicating that particle j is switched with particle k. . x2 . 6) T (6. i. 5) T (5. We can then state P2 = T (4. in notation (9. . the top row representing the numbers 1 to N labelling the N particles under consideration..3) the j-th index in the second row. . The permutations we need to consider for a system of N particles are the elements of the group SN .. namely implicitly the group property. xP (2) . 4) T (4.4) both permutations being obviously odd. i. xP (j) . . For the following we will require only two properties of SN . k). P2 = . not any state can be chosen as to represent the many– particle system. 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 = . . We will not discuss here at any depth the properties of the permutation groups SN . the set of all permutations of N objects. xP (N ) ) . .. . and the second row showing the numbers 1 to N again. i. P2 = T (4. 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. one can state P Ψ(x1 . and explicitly the fact that any P ∈ SN can be given as a product of transpositions T (j.3) P2 which affects only two particles. is called a transposition. . even though these properties. xj . xN ) = Ψ(xP (1) .e. . One calls permutations of the first type even permutations and permutations of the second type odd permutations. . particles 4 and 5. 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. Denoting by P (j) the image of j.e. (9. . . 8) . . Transpositions denoted by T (j. . A reader interested in the quantum mechanical description of N particle systems is strongly encouraged to study these representations. 3) T (7. This property will be established now.240 Many–Particle Systems repeatedly like degenerate states. . The group SN is then composed of two disjunct classes of even and of odd permutations. 5) (9. The latter factorization has the essential property that the number of factors is either even or odd.e. (j. i. . 5) in case of P2 .e. . 5).5) . However. the representations in the space of N –particle wave functions are extremely useful in dealing with N –particle systems. rather only states which obey a certain transformation property under permutations are acceptable. k) are characterized by the two indices (j. (9. . but in a different order.2). For the permutations P1 and P2 given above holds P1 = T (3. . k) = (4. in particular. k) of the two particles transposed. . a given P can either only be presented by even numbers of transpositions or by odd numbers of transpositions.

(9. The single particle states are assumed to be orthonormal.e. k = 1. −1... a diagonal representation can only berealized in a simpler group.e. the representation is either isomorphic to the group composed of only the ‘1’. G(xj .e.. ) are the same for all particles and for all pairs of particles.k=1 (9.5) implies that a permutation effectively changes the indices of the particles) and. From this property follows that H is independent of any specific assignment of indices to the particles (note that (9. In fact. implies that the permutation operators can be chosen in a diagonal representation. in turn. . particles with integer spin values. 1 together with multiplication. hence. i. all particles being governed by the same interactions. the Hamiltonian for permuted indices P HP −1 is identical to H. In the latter case all even permutations are represented by ‘1’ and all odd permutations by ‘-1’. In fact. xk ) is symmetric in its two arguments. for the scalar product holds j|k = δj k . it is a postulate of quantum mechanics that for any description of many–particle systems the permutation operators must be diagonal. .7) This property. Since permutations do not necessarily commute. Fock-Space Our derivation will start by placing each particle in a set of S orthonormal single particle states x|k = φk (x). (9. the Hamiltonian must be of the type N H = j=1 1 F (xj ) + 2 N G(xj .9. i. particles with half–integer spin values.6) which describes one–particle interactions and two–particle interactions of the system.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. S where S denotes the number of available single particle states. The terms of the Hamiltonian will be discussed further below.9) = 1 for P even −1 for P odd (9.e. provide a representation of the group structure represented by the multiplication table of even and odd permutations given above. P ] = 0 . From the latter results [H.. it is essential that the functions F ( ) and G( . (9. Presently. i. the second group to fermions. xk ) j. ∀P ∈ SN . Obviously. 2.-1}. The first case applies to bosons. i. In fact. usually a very large number. {1} (trivial) and {1. 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 .8) |Ψ (9. or the corresponding group of two elements 1. .10) We will construct now wave functions which exhibit the appropriate properties. i. both groups.11) .

. xN |ΨFock (λ1 . In a second step we form linear combinations of Fock states with the desired symmetries. λN ) = j=1 δλj λj . . . λN ) ⊂ SN is the set of all permutations involving only particles j with different λj .. We first consider orthonormal many–particle wave functions. . . .12).15) (nj !) N! 1 2 P ∈Λ(λ1 .15) is that the sum over permutations does not involve permutations among indices j. . .15) with λk = j. . λ2 . . . k. . .13) These states form an orthonormal basis of N -particle states. . .8) is x1 . nS ).e. . the nj specify how often a single particle state |j is occupied. N x1 . . with λj = λk = . . . (9. . . .14) Obviously. .e. . . nS ) = S j=1 (9. are essential. 9. . . k) ∈ Λ(λ1 .λN ) on the rhs. .. The reader is well advised to pause and grasp the definition of the nj . appear severalfold in (9. λ2 .. 2. The numbers nj . xN |Ψ Fock (λ1 . n2 . . It holds N ΨFock (λ1 . |λN . . .9). λ2 . . . . (9. . λ2 . λN ) = j=1 φλj (xj ) . λ) in which two particles occupy the same single– particle state |λ does not allow transposition of particles.11. referred to as the occupation numbers. 9.15). .g. . . Λ(λ1 . . xN |ΨB (n1 . The Fock states represent N particles which are placed into S states |λ1 . . . λ2 . T (j. . λN ) only if λj = λk . i.15) completely and.. λ2 . . which do not yet obey the symmetries (9. . |λ2 .. . 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. . . . . x2 . n2 . . is a particular choice of placing N particles consistent with the occupation numbers (n1 . .λN ) P x1 . λj ∈ {1. x2 . 9. S} of the type (9. . . . λN )|ΨFock (λ1 . The integers nj are equal to the number of λk in (9.12) where the second factor represents the scalar product between spin states. An important detail of the definition (9. the reason is that such transposition duplicates the state which would. each particle j into a specific state |λj .. hence. .8. This detail is connected with the fact that a two–particle state of the type ΨFock (λ. λN ) do not exhibit the symmetries dictated by (9. . characterize the wave function (9. the so-called Fock-states. x2 . .9). .8. . .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. . . i. e. therefore. the states |ΨFock (λ1 .λN ) Here the states (λ1 . A wave function which obeys the boson symmetry (9.

x2 .17) describe. . . φλN (x1 ) φλN (x2 ) . . . . . . . nS ) and N = (n1 . . (9.1. . . . . the integers nj denote the occupancy of the single–particle state |j ..1: Bosons and Fermions 243 The wave functions (9. . √1 N! (9. n2 .λN ) on the rhs.16) Exercise 9.18) this wave function can also be expressed through the so-called Slater determinant x1 . .20) . λ2 . .. Wave Functions with Fermion Symmetry in the Occupation Number Representation One can construct similarly a wave function which obeys the fermion symmetry property (9. The important property holds that nj can only assume the two values nj = 0 or nj = 1. . . . λ2 . . . . .e. (9. . nS ) holds S Ψ (N )|Ψ (N ) = j=1 B B δnj nj .15) obey the boson symmetry (9. n2 . . . for example. .λN ) (9. nS ) = φλ1 (x1 ) φλ1 (x2 ) . electrons in an atom. The fermion states (9. . . For any value nj > 1 two or more of the columns of the Slater matrix are identical and the wave function vanishes.e.8). nS ) = √1 N! P ∈SN P P x1 . nS ) holds S ΨF (N )|ΨF (N ) = j=1 δnj nj . n2 . n2 . . . .. i.16).15) holds (9. . xN |ΨF (n1 . . . n2 . . (b) Show that for the states (9. for two sets of occupation numbers N = (n1 . These states form an orthonormal basis. x2 . nS ) and N = (n1 . . . . correspond to a particular choice of placing N particles consistent with the occupation numbers (n1 . φλ2 (x1 ) φλ2 (x2 ) . . . nS ). . .19) φλ1 (xN ) φλ2 (xN ) . . . n2 . Such wave function is x1 . . . Using the identity governing the determinant of N × N –matrices A N det(A) = P ∈SN P j=1 Aj P (j) (9. . . for example.9.15) describe. .9). . the photons of the electromagnetic field or the phonons in a crystal.17) Here the states (λ1 . xN |ΨF (n1 . i. . . x2 . . .. .. . a molecule or a crystal.. φλN (xN ) Here. n2 . xN |ΨFock (λ1 .1: (a) Show that the states (9. . These states form an orthonormal basis. for two sets of occupation numbers N = (n1 .

. .15). (b) Show that the states (9. k The operators a† and aj are referred to as the creation and annihilation operators of bosons for the j orthonormal single–particle states |j . Exercise 9. . . . These operators allow one to construct and manipulate many–particle wave functions while preserving permutation symmetry. . . . Creation and Annihilation Operators for Bosons We consider the operator aj defined through aj ΨB (n1 . . . guaranteeing the correct permutation properties of the many–particle states. n − 1. Employing the superindices N and N introduced above and using the orthonormality property (9. . . . . ak ] = 0 . .23) (9. .244 Many–Particle Systems The representation of wave functions (9. . (9. nS ) .24) [aj .22) follows from (9. . . nS ) = nj ΨB (n1 . are the so-called operators of 2nd quantization. . . . nS ). . (9. .21) √ The factor nj appears here on the rhs. .25) .21) we consider the matrix Aj corresponding to aj in the basis of many–particle states (9.17) is commonly referred to as the occupation number representation since the wave functions are uniquely specified by the set of occupation numbers N = (n1 . a† ] = 0 j k (9. nj .9). 9.2 Operators of 2nd Quantization An important tool to describe mathematically systems of many bosons or fermions. . (9.15).16) one obtains (Aj )N = ΨB (N )| aj |ΨB (N ) = √ S N nj δnj nj −1 k=1 k=j δnk nk . . n . since both the N -particle wave function B (n . . n2 . . nj + 1. .17) obey the orthonormality property (9. . . nS ) nj ≥ 1 0 nj = 0 √ . For the adjoint operator a† holds j a† ΨB (n1 . . .2: (a) Show that the states (9. . . . .15) and (9. . . . . . .20). n ) and the N − 1-particle wave function ΨB (n . nS ) = j nj + 1 ΨB (n1 .17) obey the fermion symmetry (9. . . [a† . . a† ] = δj k . .22) The operators aj and a† obey the commutation properties j [aj . To prove that (9. nj . nj − 1. n ) are normalΨ 1 j 1 j S S ized according to (9. . . .1. .

n2 . . .2: Operators of 2nd Quantization Let A† be the matrix adjoint to Aj .24). n2 = 0. . n2 .. the eigenvalues of Nj are the occupation numbers nj . n2 . .29) yield the commutation relationship (9. . . . nk + 1 .27. .24). a† ] = 1 For j = k follows similarly 1. One can readily show using (9. the properties (9. (9.23) follow in an analogous way. 9. .31) This operator is diagonal in the occupation number representation. One refers to Nj as the occupation number operator. obviously.27) Equations (9. nS ) = |ΨB (N ) . . j j Since this holds for any |Ψ(N ) it follows [aj . We will prove the commutation properties (9. It holds for j = k (aj a† − a† aj )|Ψ(N ) = (nj + 1 − nj )|Ψ(N ) = |Ψ(N ) . .21) and vice versa..22) ˆ Nj |ΨB (N ) = nj |ΨB (N ) . .21. .32) ˆ ˆ i. nS ) = j=1 nj |ΨB (n1 . .22) follows from (9. nS = 0) as follows S (9. S ˆ N |ΨB (n1 . . .9. . . . . 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 . . for basis states |ΨB (n1 . nS ) N |ΨB (n1 .30) Of particular interest is the operator ˆ Nj = a† aj . i. . . n2 . .e.29) |ΨB (n1 .33) is called the particle number operator since. j (aj a† − a† aj )|Ψ(N ) = k k ( nj (nk + 1) − (nk + 1)nj )|Ψ(n1 . . . The related operator S ˆ N = j=1 ˆ Nj (9. The creation operators a† allow one to construct boson states ΨB (N ) from the vacuum state j |0 = |ΨB (n1 = 0. nS ) = 0 . .34) = . j (9. (9. nS ) . (9. nj − 1 . (9. . . nS ) = j=1 a† j nj nj ! |0 .26) From this result one can immediately conclude that (9. 9. . (9. . n2 .e.28) (9.

x2 . . nj . .e. 9. . e. j Exercise 9.23.29. . . nj+1 . one needs to apply only the commutation propertes (9.24) and the property aj |0 = 0.s. . of (9. .24) and aj |0 = 0 imply that the the properties (9. ..22. . √1 N! . one does not need to worry ever about proper symmetries of wave functions. As long as one starts from a wave function with the proper boson symmetry. . it also holds for nj + 1. the fermion wave function changes sign when one exchanges the order of the occupancy. xN |ΨF (n1 . f (a† )] j [a† . nj+1 . Obviously. . . 9. nj (9.g.. φj+1 (x1 ) φj+1 (x2 ) . . Exercise 9. .30). . (n − 1)! (9. .23) hold for the state defined through (9. 9. xN |ΨF (n 1 . since they are induced through the algebra of a† and aj .h..35) and where the derivative operation is defined through ∂f (A) = ∂A ∞ n=1 1 f (n) (0) An−1 . .. .19) without affecting the fermion symmetry (9. in general. .246 Many–Particle Systems In using the boson creation and annihilation operators.2. . 9. Prove this by induction. nj+1 = 1.39) .37) (9. . .. .17. nj .e. . . i. φj (xN ) φj+1 (xN ) . .2: Show that the boson operators a† and aj satisfy j [aj . . showing that the property holds for nj = 0 and. i.. 9. .. Before proceeding in this respect we need to account for the following property of the fermion wave function which applies in the case nj . . nj+1 . operators which can alter the occupancy of the wave function (9. . .21.. nS ) = − x1 .. 9. x2 .2.9). nS ) . . . nS ) = . from the vacuum state |0 or states |ΨB (N ) ..38) corresponds to x1 . . . .36) Creation and Annihilation Operators for Fermions We want to consider now creation and annihilation operators for fermions.. in case that the single particle states |j and |j + 1 are both occupied.1: The commutation relationships (9.23. . .38) . To prove this property we notice that the l. in case it holds for nj . . xN |ΨF (n1 . . x1 . x2 . φj (x1 ) φj (x2 ) . . . . (9. f (aj )] j = ∂f (a† ) j ∂a† j ∂f (aj ) ∂aj = − where the operator function is assumed to have a convergent Taylor expansion ∞ f (A) = n=0 1 (n) f (0) An n! (9.

. .s. . This requires one. nS ). nj+1 . nj+1 . . φ2 (x2 ) φ5 (x2 ) (9. . .40) . . . . nS = 0) φ2 (x1 ) φ5 (x1 ) 1 1 = √2 = √2 [φ2 (x1 )φ5 (x2 ) − φ5 (x1 )φ2 (x2 )] . 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). A proper example is the two particle fermion wave function x1 . . . 2. xN |ΨF (n1 . One adopts the convention that particles are created and annihilated by these operators always at the first position of the wave function. of (9. . n6 = n7 = .. n3 . .9. .. at which column of the Slater determinant (9. . Because of the property of the determinant to change sign when two columns are interchanged. .. . . . particles are being created or annihilated. . Obviously. x2 |ΨF (n1 = 0.. This change of position is connected with a possible sign change (−1)qj where qj is defined for a given N = (n1 . n5 = 1. . i. i. . (9. . the expressions (9. . For this purpose one must adhere to a strict convention: the labelling of single–particle states by indices j = 1. .39) and (9. that occupancies are always ordered according to a monotonous increase of the single–particle state index. 247 − √1 ! N .h. nj = 0.. nS ) = . i. n2 . at the first column of the Slater determinant (9. (9. . qj is the number of states |k with k < j which are occupied in a fermion wave function. . (9. . nj−1 ..2: Operators of 2nd Quantization The r. We are now ready to define the annihilation operator cj for fermions in the single–particle state |j as follows cj |ΨF (n1 . ..e. .. nS ) qj states occupied In case that the single particle state |j is not occupied. x2 . nj . vanishes. . . 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. . . it is necessary to specify for a fermion wave function the order in which the single– particle states |λj are occupied.38) reads − x1 . nj . . . . . .e..e.19). . . . n2 = 1. . . We will be adopting below the notational convention that qj corresponds to occupancies N = (n1 . n2 ... n3 = 0.42) i. n4 = 0. φj+1 (xN ) φj (xN ) . n2 . . φj+1 (x1 ) φj+1 (x2 ) .43) . however. nS ) as follows: j−1 qj = k=1 nk .e.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.. n2 . the rhs. nS ) = nj (−1)qj |ΨF ( n1 .40) are identical.19). in order to be consistent with this convention. φj (x1 ) φj (x2 ) . nj → 0. .

We first show that (9. 9. One obtains for its matrix elements j C† j NN = ΨF (N )| c† |ΨF (N ) = (Cj )N j S N = nj (−1)qj δnj nj −1 k=1 k=j δnk nk S = (−1) (1 − nj ) δnj nj +1 k=1 k=j qj δnk nk .44). Similarly one obtains cj c† |ΨF (N ) = cj (−1)qj (1 − nj ) |ΨF (n1 . nj → 1 . . . B]+ = AB + BA. The anti-commutation properties (9. (9.46) first for the case j = k. nS ) qj states occupied The operators thus defined obey the commutation properties [cj . . . We will prove now the anti-commutation property (9. . nj . . n2 . nj−1 .248 The operator c† adjoint to cj exhibits the following property j c† |ΨF (n1 .48) From this follows (9. c† ]+ = 0 j k (9. We have used here the definition qj = k<j nk along with qj = qj . nj → 0 . .50) . nS ) = j Many–Particle Systems (−1)qj (1 − nj ) |ΨF ( n1 . 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 . nS ) = j (−1)qj (nj + 1))(−1)qj (1 − nj ) |ΨF (N ) = (1 − nj ) |ΨF (N ) (9. .46) [cj . . nj → 1. It holds c† cj |ΨF (N ) = c† (−1)qj nj |ΨF (n1 .43).17. . (9. Let Cj be the matrix corresponding to the operator cj in the basis of fermion states (9. nj+1 .47) Let C† be the matrix adjoint to Cj . . nS ) = j j (−1)qj (1 − nj + 1)(−1)qj nj |ΨF (N ) = nj ΨF (N ) (9. c† ]+ = δj k k where we have used the definition of the so-called anti-commutators [A. .19).45) follow in an analogous way and will not be derived here.49) The derivation involves realization of the fact that a non-zero result is obtained only in case nj = 1. [c† . We consider (9. .44) follows from the definition (9. . . ck ]+ = 0 .46).44) (9. n2 .48) vanishes.45) (9. the last factor on the rhs. otherwise. . n3 . .

nλ1 .e. . ..s.51) (9. nλN . an operator cj must be on the left of an operator ck for j < k.. nS ) k and. nS ) k c† cj |ΨF (N ) = (−1)qj +qk −1 nj (1 − nk )|ΨF (n1 . ˆ It is an interesting exercise to demonstrate that Nj only has eigenvalues 0 or 1. Correspondingly.52) One can obtain similarly the same relationship in the case j > k. On the r.9.h. . ( cj c† + c† cj ) |ΨF (N ) = 0 .s. . . is referred to as the occupation number operator.56) Here we have made use of cj cj = 0 and c† c† = 0 which follows from nj ∈ {0. 1 In the following we will apply fermion operators c† and cj only to electrons which carry spin 2 .) = c† 1 c† 2 · · · c† N |0 .51. for which purpose one needs to envoke only the algebraic (anti-commutation) properties (9. 2 (9. 1}. . . nj = 0 .52) hold for any |ΨF (N ) one can conclude (9. nk = 1 . We j will often separate the states |j . . . Equation (9.50) yield ( cj c† + c† cj ) |ΨF (N ) = |ΨF (N ) . . coordinates x and index j into a space part and a spin part. in the basis |ΨF (N ) . .58) . j → j. The stated property ˆ follows from the idempotence of Nj which can be derived as follows † ˆ2 ˆ Nj = c† cj c† cj = c† (1 − c† cj )cj = c† cj − c† c† cj cj = cj cj = Nj .53) is diagonal in the occupation number representation. .49) shows that the operator ˆ Nj = c† cj j (9.e. . λ λ λ (9. with eigenvalues equal to the occupation numbers nj ˆ Nj |ΨF (N ) = nj |ΨF (N ) ˆ Nj . i. σ 1 (σ = ± 2 ) . . k k 249 (9.h. the creation operators must operate in the proper order. x|j → φj (r) | 1 σ . . S (9. nj = 0 . i. Since (9.55) is called the particle number operator. nλ2 .46).57) On the l.49) and (9. . .e. j j For j < k one obtains cj c† |ΨF (N ) = (−1)qj +qk nj (1 − nk )|ΨF (n1 .2: Operators of 2nd Quantization (9. . j j We finally note that the creation operators c† allow one to construct any fermion state |ΨF (N ) j from the vacuum state |0 defined as above (see (9. nk = 1 . hence. . i. j j j j j j j (9. . hence.46).29)) |ΨF (.. of this equation we meant to indicate only those N occupation numbers nλj which are non-vanishing.54) ˆ N = j=1 ˆ Nj (9. 9.

.60) is that the sum over particles in (9.15. 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 ) . ˆ ΨB. xk .62) These matrix elements can be obtained by chosing the operators and many-particle states in the position representation.250 Many–Particle Systems 9.63) ˆ ˆ This expression has been specified for the purpose of these notes to distinguish F and f and is not common ˆ terminology.F (N ) . The operators g (x. 9.F (N )|F |ΨB. The essential aspect of definition (9. x2 . . N are not all identical. 9. . 2. 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. The action of the operators boson and many–fermion states |Ψ is obviously described through the many–particle state matrix elements ˆ ΨB. . . .17).17) and integrating over all particle space-spin coordinates x1 . r = 1. j = 1. Since the many–particle states are built-up from a basis |r . These operators had been introduced already in Eq. namely.k=1 j=k (9. one expects that the matrix elements can actually be expressed in terms of matrix elements ˆ involving solely the generic operators and the single–particle states. S of single–particle spinorbital states (see page 241) and the operators are specified through the associated generic operators ˆ f and g .6) above. . An operator N ˆ R = j=1 rj (xj ) ˆ (9.59) ˆ ˆ are called single–particle operators.60) ˆ correspondingly are called two–particle operators.15. (9. We adopt a similar expression to distinguish two–particle operators G and g . ˆ . . (9.59) and over pairs of particles in (9. 2. (9.F (N )|G|ΨB. .e. Operators of the type 1 ˆ G = 2 N g (xj .59.59. The operators f (x) which constitute F are called the associated generic single–particle operators1 .61) is not a one particle operator as long as the N operators rj (x). σr ) f (x) ψs (r.F (N ) . 9. xN .59) and (9. ˆ r|f |s = 1 ∗ ˆ d3 r ψr (r. x ) which constitute G are ˆ called the associated generic two–particle operators.. i. xk ) ˆ j.60) involves generic operators – acting on xj and on xj . respectively. σs ) (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. respectively – which are all identical. 9. .

. σt )ψu (r . according to (9.59). s and t. is t(x) = (− 2 /2m)(∂ 2 /∂r 2 ). σu ) .67) ˆ In this case the generic single–particle operator.e. x ) ψt (r.64) r. ∂r 2 ∂x2 ∂x2 ∂x2 1 2 3 (9. are spin-independent as is evident from the factor δσσ .63.69). i.66) and (9.69) ˆ In this case the generic operator is f (x) = δ(r − r). We will show that the boson and fermion creation and annihilation operators serve this purpose.66) Here we adopt the notation for the Laplacian ∂2 ∂2 ∂2 ∂2 = + + . 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. (9. 9.3: One– and Two–Particle Operators r. 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 .68) Another single–particle operator is the one–particle density operator N ρ(r ) = ˆ˜ j=1 δ(r − rj ) . s|ˆ|t. σr )ψs (r . u = g d3 r ∗ ∗ d3 r ψr (r.64) We like to note an important symmetry of the matrix element r.k=1 j=k q2 |rj − rk | (9.71) . 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) . u = s. ∂rj2 (9. ˆ 251 (9. u.62) in terms of the matrix elements of single–particle states (9. u which follows directly g from the definition (9. that the symmetry implies that one can switch simultaneously the pairs of indices r. σs ) g (x.9. ∂r 2 (9.64). s|ˆ|t. Notice. t . one cannot switch the indices of one pair individually. g g (9. ˜ (9. s|ˆ|t. 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. r|ˆ|u.70) Both operators.

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

23. pp.83) t=1 t=r. 9.64) one replaces the operators F in (9.46)..u=1 S r. u as defined in (9.62) through single–particle state matrix ˆ ˆ elements (9.63) and (9. respectively. r ΨF (N )|c† cs |ΨF (N ) .64).63.79) in the basis of many–particle states |ΨB. namely cu ct . (9.82) ΨF (N )|c† cs |ΨF (N ) = r cr 2 (9. .3: One– and Two–Particle Operators Definition in Terms of Creation and Annihilation Operators 253 In order to express many–particle state matrix elements (9.t. The matrix elements are then actually those of the number operator c† cr which is diagonal in the occupation number representation. u . s|ˆ|t.s.60) as follows N ˆ f (xj ) → j=1 N 1 2 1 2 S r.80) (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. The Matrix Elements ΨF (N )|c† cs |ΨF (N ) r We consider first the case r = s. s|ˆ|t. 9.45.78) and (9.62). 9.k=1 j=k S r. Expressions (9.176 where the matrix elements of fermion states can be found.171 and pp.59) and G in (9. respectively.s ΨF (N )|cs ΨF (N ) = nr (−1)qr ns (−1)qs δnr −1 nr δns ns −1 δnt nt We refer the reader to Condon and Shortley’s ‘Theory of Atomic Spectra’.u=1 (9.s=1 ˆ r|f |s a† as bosons r ˆ r|f |s c† cs fermions r r.s=1 S r. i.9. r ΨB (N )|a† a† at au |ΨB (N ) r s ΨF (N )|c† c† ct cu |ΨF (N ) r s bosons fermions (9.81) For the latter matrix elements simple rules can be derived from the algebraic properties of the boson and fermion operators (9. s|ˆ|t.F (N ) have the same values as the respective matrix elements (9. To show that the resulting values of the matrix elements (9.24) and (9.79) It is crucial to notice in the expression for the fermion two–particle operator that the order of the annihilation operators in (9.78. u a† a† at au bosons g r s † † r. xk ) → ˆ j.t. These rules will be provided below only for the case of fermions.e.s. r S Ψ (N In case of r = s one obtains F )|c† cr |ΨF (N ) r = nr r=1 δnr nr .78) 1 2 g (xj .79) have the following implication: The operators (9. 9.79) is opposite to that in the matrix element r. g and not ct cu . u cr cs cu ct g fermions (9.62) are correct one needs to compare the result derived with conventional derivations of the matrix elements2 . and then in g a second step the matrix elements ΨB (N )|a† as |ΨB (N ) . s|ˆ|t.

t.87) Here and in the following we denote by ‘a. Comparision with the results in Condon&Shortley (pp.e. are ˆ ˆ ΨF (N )|F |ΨF (N ) = (1 − ns )(−1)qs nt (−1)qt s|f |t (9. u) are included in the summation for which all indices are different. (r. 1 2 S r s|ˆ|s r c† c† cr cs .s=1 a. u need to be considered.e. t=u 2 = 0.t. (9. c2 = 0 r (9.i.d. Case 1 three of the four indices are different. 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. i.84) The off-diagonal elements with nr = nr except for r = s. those for which N and N differ in more than two occupation numbers. u in this sum three possibilities exist Case 0 two of the four indices are different.53) 1 ˆ G0 = 2 S r s|ˆ|r s − r s|ˆ|s r g g r. (9.d. ˆ ˆ ˆ ˆ Accordingly. we split the sum in (9. for which holds r = s.86) into three contributions. s.. r = t etc.82. Case 2 all four indices are different.i. i. t.d.u r=s. t. t.86) For the combination of indices r. The anti-commutation property (9.s.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 (9.254 From (9.i.45) yields together with the definition of the occupation number operator (9.s=1 a.79) and using c† r we can write S 1 ˆ G = rs|ˆ|tu c† c† cu ct g r s 2 r. Starting from the definition (9. in which case ns = ns + 1 and nt = nt − 1 holds.59) when the matrix elements are determined between Slater determinant wave functions. ˆ The first contribution G0 is 1 ˆ G0 = 2 S r s|ˆ|r s c† c† cs cr + g r s r.’ (all indices different) that only tuples (r. s. t).85) All other matrix elements vanish. g r s r. where each contribution corresponds to one of the three cases mentioned. s). G = G0 + G1 + G2 . 9..s=1 a.i. (r.88) .78) does yield the same results as the operator defined in (9.d. ˆ ˆ Nr Ns . s. namely. s. 171) shows that the operator defined in (9.

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

1: 9.91).90) from (9.5: Derive (9. contributes in this case.3.62) for bosons.96) .3. i. particle number is conserved.73) and S 2 given in (9.92). u < v. t − r. t for which holds ˆ ns = ns + 1 and nt = nt − 1. Only G1 .92) from (9. u > v or s > r. nv = nv − 1.89). v − r. for ˆ non-vanishing matrix elements ΨF (N )|G|ΨF (N ) at most four occupation numbers nr and nr can differ.92).94) Case 2 This case covers off-diagonal elements with nr = nr except for r = s..74) and (9. One obtains ˆ ΨF (N )|G|ΨF (N ) = (−1)qs +qt (1 − ns )nt r r=s. r g g (9. Derive (9. it holds S nr = S nr . One obtains ˆ ΨF (N )|G|ΨF (N ) = ± (−1)qs +qr +qu +qv (1 − ns )(1 − nr )nu nv 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–particle and two–particle operators. u < v or s > r. s|ˆ|u. u > v. The matrix elements of the corresponding generic operators are (9. Derive the non-vanishing matrix elements (9. u g g (9. s|ˆ|v.95) from (9. nu = nu − 1.3.t nr r. v for which holds ˆ ns = ns + 1. r=1 r=1 Exercise Exercise Exercise Exercise Exercise 9.2: 9. t.3: 9. In particular. s|ˆ|t.3.4: 9.3. given in (9. respectively. Derive (9. ˆ ˆ Show that the matrix elements of F and of G conserve particle number.76). nt = nt + 1.75) which are.256 Many–Particle Systems Case 1 This case covers off-diagonal elements with nr = nr except for r = s.95) In the latter formula the ‘+’-sign applies for s < r. s|ˆ|r. given in (9. All matrix elements which are not covered by the three cases above vanish. We like to express these operators through ˆ fermion creation and annihilation operators. the ‘−’-sign applies for s < r. u.90). Furthermore. contributes in this case. Spin Operators ˆ ˆ In this section we will briefly consider the spin operators S given in (9. Only G2 .e.

s|ˆ|u. u c† c† cuσ ctσ v rσ sσ r. the study of problems governed by Hamiltonians of the type (9. In fact. 9.u=1 σ. rα sβ (9. In all these systems the Hamiltonian is a sum of one– and two–particle operators S H = r=1 1 ˆ r r|t|s c† cs + r. the spin part of the single–particle states is represented by α = 1 and β = − 1 .6: Derive (9. The fortunate side of this is.s=1 σ 1 ˆ rσ r|t|s c† csσ + 2 S r. Examples is the laser action.s=1 r. Actually. molecules. The great generality of the concepts developed in the context of many–particle systems can be judged from the fact .s. 9.97) The summation in (9. in crystals.g. the concerted behaviour of interacting systems can deviate qualitatively from that of systems made up of independent components.96.4: 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. but also ordinary phenomena like freezing and evaporation.99) has preoccupied an important part of all intellectual efforts in Theoretical Physics during the past fifty years.3. superconductivity. it is one of the main intellectual achievements of Physics among the Sciences to have addressed systematically the study of systems composed of many strongly interacting components.9. however.s=1 c† c† crβ csα .98.97) is over the orbital states. the two–particle operator makes the description of many–particle systems a very hard problem. which represents the interactions between particles. the systems investigated involve a macroscopic number of components such that statistical mechanical approaches are envoked.98) In many cases the Hamiltonian is spin-independent and the equivalent notation S H = r. In the latter case the indices r..σ (9. 9. would be a simple exercise in linear algebra. e.96) and (9. e. s|ˆ|t. 2 2 Exercise 9. s.97). that the two–particle operator representing interactions between particles induces a variety of interesting phenomena. evaluation of their stationary states and excitation energies. u refer only to the orbital part of the single–particle basis. Often.t.g..t. t.u S (9. the description of many–particle systems. If it were not for the two–particle contribution to the Hamiltonian.s. The outcome of these studies is that even when interactions between components (particles) are simple. atoms and nuclei. Unfortunately.4 Independent-Particle Models In the remaining part of this chapter we will apply the technique of operators of 2nd quantization to the description of many–fermion systems as they arise.99) will be used. v c† c† cu ct . v r s 2 r.

e. another prototypical systems of many strongly interacting components.. electrons and nucleons. m = 1. What might happen? Independent-Particle Hamiltonian A many–fermion system governed by an independent-particle Hamiltonian. If it were’nt for the fermion nature.100) ˆ ˆ We will denote the matrix elements of t as r|t|s = trs . ultimately depend in a crucial way on the fermion character of its constituent building blocks. The Hamiltonian of such system is S Ho = r. namely of neurons3 . according to which nuclei and atoms change their qualitative properties when they grow larger. i. 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).. Before continuing one may finally point out that the material world as we know it and as it establishes the varieties of natural substances. . all material systems would condensate into states which would depend little on particle number. {|m). can be described in a rather straightforward way.100) in a form in which the factors c† csσ become rσ occupation number operators. ranging from nuclei to the molecules of living systems .J. would not exist.1: Imagine that in a closed.4. In such a representation Ho is diagonal and the stationary states can be stated readily. The Aufbau principle. Transformation to a New Set of Creation and Annihilation Operators Our aim is to represent the Hamiltonian (9. 1989) .102) U ∗rm Urn = δmn .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.258 Many–Particle Systems that today these concepts are providing insight into the functioning of biological brains.101) The unitary property of Urn U† U = U U† = 1 1 or S S (9. Exercise 9. .103) An account of some of these efforts can be found in Modelling Brain Function – The World of Attractor Neural Networks by D. The electronic properties of atoms with different numbers of electrons would differ little. 2. i. where S |m) = r=1 Urm |r . a Hamiltonian without a two–particle operator contribution accounting for interactions among the particles. (9. S}. New York. Chemistry essentially would know only one element and Life would not exist. r=1 3 Urm U ∗sm = δrs m=1 (9. . Amit (Cambridge University Press. We will restrict our description in the following to systems composed of an even number (2N ) of particles and to spin-independent interactions.

behave exactly like the operators c† and crσ behave for states |ΨF (N ) . of N -Fermion states in which single particle states |nj ) as defined in (9. d† ]+ = 0 (9. (9.104) One can. nσ We demonstrate here the anti-commutation property (9.111) and. rσ namely. they are Fermion creation and annihilation operators. The l. accordingly. .100) yields ˜ Ho = tmn d† dnσ mσ where d† nσ = r=1 S S (9. the anti-commutation properties (9.h.110) can be demonstrated similarly.108) obey the same anti-commutation properties (9.4: Independent-Particle Models allows one to express conversely |r in terms of |m) S 259 |r = m=1 ∗ Urm |m) . dnσ = r=1 ∗ Urn crσ (9.and two-particle operators hold in an analogous way for d† and dnσ .101) of states |r . The unitary property (9. (9.. the operators d† and dnσ . d† ]+ = sσ nσ r. using (9.101) are occupied. hence.108) These operators describe the creation and annihilation operators of fermions in states |n) which are linear combinations (9.s r The unitarity property r ∗ Urm Urn = δmn yields immediately identity (??).112) .46) as c† and crσ .e.46). In a basis of states N d† j σj |0 n j=1 (9.110) mσ nσ [dmσ .104)] S S ∗ Urm d† mσ m=1 c† rσ = . 9.s=1 ˆ U † tU mn (9. i.107) Urn c† rσ . dnσ ]+ = 0 .f. d† ]+ = δmn δσσ nσ (9.9. csσ = n=1 Usn dnσ .111) can be written. of (9. (9.105) where ˜ ˆ tmn = (m|t|n) = S ∗ trs Urm Usn = r. [d† . (9.108) and (9. for nσ rσ example.103) allows one to express [c.45.111). c† ]+ = δσσ [dmσ .109) The operators (9. the expressions drived above for the matrix elements of one.106) which together with (9. [dmσ . expand S trs = mn ∗ ˜ tmn Urm Usn (9.s.113) Urm Usn [crσ . ∗ ∗ Urm Urn .

The corresponding eigenvalues n . σ2 .e. is obtained. r = 1.57) to the case of 2N particles. − 1 .115) with eigenvalues 2N E(n1 . transformation matrices Urn and energy values n . 2 2 2 2 i. 1 and |β = | 1 .117) are eigenstates of (9.115) ˜ and involves solely occupation number operators N nσ = d† dnσ .106) is equivalent to S trs Usn = s=1 n Urn ∀r.e.. .e.118) Ground State In case of an ordering m < n for m < n the state of lowest energy. n2 . i. . .. ) = j=1 nj . i. hence. (9.e. S are T real.115) can be stated immediately since any many–particle wave function in the occupation number representation is suitable. S (9. Eigenstates of (9. This equation poses the eigenvalue problem for the hermitian matrix (trs ). A non-degenerate orbital state which is occupied by two fermions. . is N |Φo = j=1 d† α d† β |0 . a simple matter to nσ state many–particle states in the new representation. i. σ1 .114) together with (9. − 1 is called a closed 2 2 2 2 . the so-called ground state. σk ) for j = k (9. . Before proceeding we like to address the issue how the new representation. the associated ( properly normalized) eigenvectors Vn are the columns of (Urn ). n = 1. The 2N fermion state 2N d† j σj |0 n j=1 where (nj . σ2N .260 Diagonal Representation Many–Particle Systems The transformation (9. .114) Ho = n=1 † n dnσ dnσ (9. It is. . σj ) = (nk . (9.. . 2. U2n . Vn = (U1n . USn ). ˜ tmn = In this case Ho has the desired form S n δmn . . 2. a fermion in a spin state | 1 . n2N . . .116) which follows from (9.106) diagonal.. j j (9.101) gives us the freedom to choose the matrix (9. . We apply for this purpose (9.106) and the unitary property of Urn . The condition (9.119) In this state the N lowest orbital eigenstates of trs are occupied each by an electron with spin |α = | 1 . 1 as well as a fermion in a spin state | 1 .

(9. ˆ one can conclude that |Φo is an eigenstate of S 2 with eigenvalue zero. β ˆ are eigenstates of S 2 given in (9.n=1 cmα cnβ cmβ cnα does not give a contribution if m = n and either |m) or |n) are closed shells. α . |α. the third term − S m. α |β.e. of ∆E = and |β. β |β. namely.e.97).n=1 † † if either |m) or |n) are closed shells.4: Independent-Particle Models 261 shell.n=N (9. α = 2 |β. 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 = 2 |α. We will demonstrate now that this property endows the ground state with total spin zero. the states differ in their spin character.124) .n=1 d† d† dmβ dnα mα nβ (9. 1 S 4 m. is a singlet state.n=1 m. α |α. energy above the ground state.n=1 m. α N +1 − N . i. ˆ S 2 ||β.122) This property can be derived as follows: The two states are obviously diagonal with respect to the ˆ following contribution to S 2 1ˆ 1 ˆ Σ1 = N + 2 4 N −1 N −1 ˆ ˆ ˆ ˆ (Nmα − Nmβ )(Nnα − Nnβ ) − m. However. β . Similarly.123) ˆ which acts only on the N − 1 closed shells (except for N ) of the two states and has an eigenvalue ˆ N + 0 − (N − 1) = 1 in both cases.9. both with eigenvalues 2 ˆ S 2 |α.120) where the occupation number operators refer to the single-particle states |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).120) is particularly simple for closed shells. There are four such states. Excited States We want to construct now excited states for the system governed by the independent-particle Hamiltonian (9. The action of the operator (9. One can conclude immediately that the ˜ ˜ ˜ ˜ (N mα − N mβ )(N nα − N nβ ) does not give any contribution second term in (9.120).. The ground state has only closed shells. The two states |α. Obviously. Alltogether.100). The spin operator in the present representation of single-particle states is 1 1ˆ ˆ S2 = N + 2 4 S S ˜ ˜ ˜ ˜ (N mα − N mβ )(N nα − N nβ ) − m..121) All four states have the same excitation energy.n=1 d† d† dmβ dnα mα nβ (9. namely. 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 (9.97). In case of m = n it gives a contribution ‘1’ for each closed shell which cancels the contribution of the first term. i. 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.

r = 1. The two states |α. . . . The same result holds for |β. α as triplet states. c† +1 σ = c† σ 1 2N (9. conclude that (9. mα nβ (9. 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. α ± |β. . m |Φo m m m γσ. of course. for |α. of course.σ d† +1. vanish.127) ˆ are triplet and singlet states. One can.100) outlined above we consider a system of 2N electrons which move in a set of atomic orbitals |r . β qualify as eigenstates. do not need to be considered since they ˆ give vanishing contributions. one with eigenvalue zero (singlet) and one with eigenvalue ‘2’ (triplet). 2.σ = (9. |N + 3 . 2N which are located on a ring.125) ˆ Contributions to S 2 acting on states |N + 2 .130) . The system is assumed to posses a 2N –fold symmetry axis and interactions connect only orbital states |r → |r ± 1 . ˆ The remaining states |α. 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 . 2N ) drσ = √ 1 2N 2N e−irnπ/N cnσ n=1 (9.122) is correct. (9. We identify the states |2N + 1 = |1 to avoid cumbersome summation limits. .n=N d† d† dmβ dnα d† +1α dN β |Φo = 0 mα nβ N (9. .128) The cyclic property of the system is expressed through c2N +1 σ = c1 σ . β both with eigenvalues 1. β =− m.σ dN σ N σ.2: Construct four operators O ˆ O ˆ such that O ˆ S3 .. β are not eigenstates of S 2 . α . ˆ Exercise 9. etc. |β.126) which follows from the occurence of squares of fermion operators which. The action of Σ N +1 ˆ Σ3 |α. ˆ 3 on the two states produces . hence.129) The symmetry of the Hamiltonian suggests to choose a new representation defined through the operators (r = 1. however. α and |β. α are. β . . i.e. also eigenstates of Σ2 . 2.262 and ˆ Σ3 = − m.4. . . β and |β. ˆ An eigenvalue ‘2’ of the spin operator S 2 identifies the states |α. for example.n=N Many–Particle Systems N +1 d† d† dmβ dnα . the linear combinations 1/2(|α.

s.140) The Hamiltonian is now diagonal and the construction outlined above can be applied.132) one can conclude 2N ei(s−r)nπ/N + eisπ/N n=1 1 2N 2N ei(r−s)nπ/N .132) gives for r = 0. thereby.131) The (9.4: Independent-Particle Models d† = √ rσ 1 2N 2N 263 eirnπ/N c† nσ n=1 √ 1 eirnπ/N .133) Application to the l.131) constitute a unitary transformation U defined through Urn = unitarity property follows from 2N eirsπ/N = 0 r=1 for 0 < s < 2N .h.s=1 σ ˜ rσ trs d† dsσ (9. . (9.139) −2t cos r=1 σ rπ N d† drσ .137) where 1 ˜ trs = e−irπ/N 2N Using (9. 2N. 9.130. one obtains the new representation of Ho 2N ˆ Ho = −t r. rσ (9. 4N. .134) which. vanishes for integer s. n=1 (9. 2N (9. One can then express[.9.f.135) c† nσ 1 = √ 2N e−irnπ/N d† rσ r=1 (9. .136) ˆ and. eisπ/N 1 − e2N isπ/N 1 − eisπ/N (9. of (9. 1−a (9. .109)] ¸ cnσ = √ 1 2N 2N eirnπ/N drσ r=1 2N (9. indeed. .132) an identity which can be derived employing the well-known expression for a finite geometric series m as = a s=1 1 − am . (9.138) ˜ trs = from which follows ˆ Ho = eirπ/N + e−irπ/N 2N δrs (9.

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

. S} the elements of which will be labelled by indices ˜ m. If one restricts m = 1.145).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. .142) is based. . t.142) r|t|s c† csσ + rσ sσ 2 r. . . The states |m serve to define a reference state of the system ˜ N |Φo (U) = j=1 d† α d† β |0 . S) ˜ S S |m = ˜ r=1 Urm |r .u=1 σ. 2. u v r. 2. q. m = 1. s|ˆ|t. . . s|ˆ|t. |r = m=1 ∗ Urm |m .143) These S states are associated with the creation and annihilation operators S d† mσ = r=1 S Urm c† rσ ∗ Urm crσ r=1 dmσ = (9. j j (9.144) defined as in (9. S on which (9. .σ and derive a single-particle operator which approximates this Hamiltonian. . 2. u refer to the initial basis of single-particle states |r . The states |m are connected with the states |r through (m = 1. . . n.u=1 σ. Mean-Field Potential The mean field for the reference state |Φo (U) is defined in a straightforward way by averaging the two-particle contribution to (9.143) with r.146) . This is done as follows: ˆ Vmf (|Φo (U) ) = 1 2 S r.142) over this reference state to turn the contribution into an effective one-particle operator.s.s=1 σ r. . In our formulas below we will adopt strictly the convention that indices r. We will adopt a second set of single-particle states {|m . 2. U = ( Urm ) as defined in (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. r = 1.t. ˜ (9. the corresponding U = ( Urm ) forms an S × N matrix and. .108). s.s. N . . p. u c† c† cuσ ctσ v (9. S is a unitary matrix. as in in (9. 2. is not unitary. .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 (9.9.t. m = 1. naturally. .

n. n c† ˜ v ˜ sσ m. m|ˆ|t. u c† d† dnσ cuσ ˜ vn rσ mσ r. s|ˆ|m. n c† ctσ δσ σ ˜ v ˜ rσ m≤N r. by means of the rules (9. u ˜ ˜v m. m| v |t. 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. hence.147) Since the reference state |Φo (U) is defined in terms of the single-particle states |m it is preferable ˜ to switch the representation of (9.s.u ∗ r.m.σ S d† d† mσ nσ cuσ ctσ + r.σ S dmσ dnσ + r. u Usm Uun .σ S d† dnσ mσ ctσ (9.n.σ S 1 − 2 where. 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.m. ˆ (9.149) The remaining matrix elements appearing in (9. for example. m|ˆ|˜ .t.σ S d† dnσ cuσ mσ − r.u=1 σ.m.148) r.148) one obtains.t.148) are obtained in a similar way by transforming only two of the four single-particle states into the new representation.u=1 σ. m|ˆ|t. n c† c† v ˜ ˜ rσ sσ r.u=1 σ. s| v t. For the averages ··· in (9.n=1 σ.150) r.σ S ctσ + m. n c† d† dnσ ˜ v ˜ rσ mσ r.s.s.84).t=1 σ.n.n=1 σ. s|ˆ|t.146) accordingly.266 Here ··· denotes the average O = Φo (U)| O |Φo (U) . s|ˆ|˜ . d† d† mσ nσ dmσ dnσ d† dnσ mσ = = = = and. u c† ˜ vn sσ m. n = ˜ ˆ ˜ s. Many–Particle Systems (9. m.σ . n|ˆ|t.t=1 σ.n.

2..m.154) ˆ The mean field approximation replaces then the Hamiltonian (9. u = s. t| v |u. n c† ctσ δσ σ ˜ v ˜ rσ − r. n c† ctσ . s| v |t.. s|ˆ|t. u c† cuσ δσσ ˜ vn sσ m≤N s.9.5: Self-Consistent Field Theory 1 2 1 2 1 2 S 267 m.152) can then be written in the form ˆ Vmf (|Φo (U) ) = r| vmf (|Φo (U) ) |s c† csσ ˆ (9. u c† cuσ ˜ vn rσ (9..u=1 σ.156) which defines the reference state |Φo (U) .152) By means of expression (9. The one-particle operator (9. The mean field approximation.σ S + − r. m=1. m|ˆ|˜ .151) Carrying out the sum σ δσσ = 2 leads to a factor 2 for the first two terms.S.142) by Hmf (U) defined through S ˆ Hmf (U) = r.s=1 σ ˆ r|t|s + r| vmf (|Φo (U) ) |s ˆ c† csσ rσ (9. m|ˆ|t. At this point this state is completely arbitrary..t=1 σ.t=1 σ (9. as introduced here.. .149) for the matrix elements in the mixed representation one can state the right hand side explicitly in terms of Urm .s σ where r| vmf (|Φo (U) ) |s = ˆ t. leaves open the question how the reference state should be chosen.m.σ r. r| v |u.N (9.. ˜ v ˜ sσ m≤N s. t| v |s. Exploiting the symmetry r. u − r.u=1 σ S − m. s )  ˆ ˆ  m≤N ∗ Utm Uum   (9. We will address now a proper choice of the reference state. 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.155) which is a function of the S × N –matrix U = ( Urm )r=1.2. m|ˆ|˜ .u ( 2 r.u=1 σ. u c† cuσ ˜ vn rσ m≤N r.σ S r. s|ˆ|˜ .153) rσ r.

119. U(1) is the S ×N –matrix (1) of the first N column vectors of Urm . We will argue below that the self-consistent field ground state. i. is the lowest energy independent-particle ground state one can construct.142). 2. S. . . Step 1: Choosing an Initial Reference State One defines the Hamiltonian ˆ (1) Hmf = r. i. The state thus determined is referred to as the self-consistent independent-particle ground state. S . is based on the mean field approach and provides an algorithm to construct a reference state |Φo (U) for a 2N fermion system described by a spin-independent Hamiltonian (9. (9.268 Many–Particle Systems 9.155). S . . often also referred to as the Hartree-Fock approximation. 2.145).4. Step 2M. This procedure.s=1 σ S ˆ r|t|s + r| vmf |Φo (U(M ) ) ˆ |s c† csσ rσ = r. . . SCF-Algorithm. SCF-Algorithm. the reference state resulting from step 2M + 1 (within numerical error) is equal to the reference state resulting from step 2M − 1.. Let us state now the construction of the self-consistent field ground state in more detail. . assuming in an intial step (1) a properly chosen reference state. determining in a step (2) the corresponding mean field Hamiltonian (9. r = 1. 9. and obtaining in a step (3) the ground state of this one-particle Hamiltonian according to the construction in Section 9. S . is exact only for an independent-particle Hamiltonian. . under conditions which often are realized.155) following the method outlined in Section 9. i. can achieve this goal only iteratively. r = 1.145). 2. 2.6 Self-Consistent Field Algorithm The Self-Consistent Field (SCF) approximation. however. .4 and defining this as the new reference state.157) and determines the associated diagonal representation defined through S (1) ˆ r|t|s Usm = s=1 (1) (1) m Urm . (9.e.e. . . . 2.e. .s=1 σ ˆ r|hmf |s c† csσ rσ (M ) (9.160) . .. m = 1.s=1 σ S ˆ rσ r|t|s c† csσ (9. The procedure determines the reference state as the ground state of the independent-particle Hamiltonian (9. steps (2) and (3) are being repeated M times until the procedure converges. m = 1. the independent-particle nature stemming from the fact that the functional form of the ground state. M = 1.158) where the labels m are ordered to obey the condition (1) m < (1) n for m < n . . .159) One defines the reference state |Φo (U(1) ) using the definition (9.: Determine Mean Field Hamiltonian In this step the mean field Hamiltonian S ˆ (M ) Hmf = r. (9.

.167) .e.6.: 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 . .164) ˆ (a) Determine the expectation value of Prs for the state as defined in (9. i. S (9. . U(M +1) is the (M +1) S × N –matrix of the first N column vectors of Urm . .e. S . i. 2.145). the S × S–matrix S (M ) r|hmf |s N 269 ˆ = r|t|s + t.165) can be written ˆ F = S ˆ ˆ r|f |s Prs r. . . . SCF-Algorithm: Continuation and Convergence Condition When step 2M + 1 is completed. .145). ˆ t.. i. σ p = δtr δsu δσσ . (9. s ) ˆ ˆ m=1 Utm ∗(M ) (M Uum ) (9. m = 1.161) is calculated.e. (c) Show that any spin-independent one-particle operator S ˆ F = r. . s = 1. t| v |u.166) ˆ (d) Define a similar two-particle operator Prstu . SCF-Algorithm. .163) The result yields the reference state |Φo (U(M +1) ) using the definition (9.u=1 ( 2 r. . S do not exceed a preset threshold.6: Self-Consistent Field Algorithm is evaluated. (9.s=1 σ ˆ r|f |s c† csσ rσ (9. 2. u − r. 2. . etc..s=1 (9. . σ|ˆrs |u. S. indicating that the state converged. step 2M + 2 is carried out. S . ˆ (b) How can the expectation values of the operator Prr be interpreted. . .9. (e) Consider the creation operators N † gqσ = m=1 Vmq d† mσ (9. . m = 1. . ˆ Exercise 9.162) is solved adopting an ordering of the labels m which obeys the condition (M +1) m < (M +1) n for m < n . t| v |s. r = 1.1: Let Prs be the one-particle operator with the generic operator prs = |r s|. 2. 2. r = 1. M = 1. The procedure is continued until it ∗(M ) (M ) ∗(M +1) (M +1) Uum − N Urm Uum | is detected that the one-particle density differences | N Urm m=1 m=1 for r. 2 . Step 2M + 1.

. (SCF ) m < (SCF ) n for m < n (9. (9. (f) Can one distinguish the states (9.163). Note that this is not an mσ S × S–matrix! Show that the reference state defined through N † † gqα gqβ |0 q=1 (9.6.171) implies that for the mean field Hamiltonian holds S ˆ Hmf (U(SCF ) ) = m=1 σ (SCF ) † dmσ dmσ m .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.173) . (9.172) is obeyed. i.171) where the convention (9.170) (SCF ) Here |r . r = 1. Equation (9. ..σ + 2v ( c† c1α c† c1β + c† c2α c† c2β ) + v 1α 2α 1β 2β + c† c2σ c† 2σ 1σ c1σ ) . 2. We begin by summarizing the result of the SCF algorithm. .e. We denote the representation which results from the SCF algorithm after its convergence as follows: S |m = ˜ r=1 (SCF Urm ) |r .168) through a physical observation? Exercise 9.163).9.145).270 Many–Particle Systems which are connected with d† through a unitary N × N –matrix (Vqm ).145) and (9.168) ˆ ˆ has the same expectation values for Prs and Prstu as the reference state (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. S denotes the initial single-particle states and Urm is the unitary S × S– transformation matrix obtained in (9.7 Properties of the SCF Ground State We want to investigate now the properties of the SCF ground state.169) 9.162. (9.

and once as a probe particle being subject to the mean pair interaction.146) counts a particle twice.162) is non-linear in Urm .175) However. Finally..1: Reformulate (9.177) 2 mm | v |mm ˜˜ ˆ ˜˜ − mm | v |m m ˜˜ ˆ ˜ ˜ . s|ˆ|t. the mean-field Hamiltonian is diagonal in the SCF representation. r. the SCF ground state is N SCF = m=1 d† d† |0 . the mean field potential ˆ Vmf (|Φo (U) ) defined in (9. once as a member of the ground state. one can also determine the expectation value of SCF for the Hamiltonian (9. (SCF ) ˆ Exercise 9. (9.177).142).173) holds N ˆ SCF Hmf (U(SCF ) ) SCF = 2 r=1 (SCF ) m .e.177).142). 2 m=1 m|t|m + ˜ ˆ˜ m. mα mβ (9.171) yields ˆ SCF H SCF N = N (9. exploiting the spinindependence of (9.84) and (9. . i.93) for the expectation values (diagonal elements) of the one-particle and two-particle part of the Hamiltonian (9.e. Since the system under consideration has only 2N particles altogether.176) 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. This over-counting leads to the correction term in (9..174) The first property we consider is the total spin character of SCF .7: Properties of the SCF Ground State 271 i. one obtains ˆ SCF H SCF N = N (9.7.m =1 Comparision with (9. We like to determine next the energy expectation value of SCF .4 that SCF is a total singlet state. 2 m=1 (SCF ) m − m. Employing expressions (9.171) in terms of matrix elements r|t|s .7.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. v (SCF ) Show that the resulting eigenvalue problem of the type (9.2: Derive (9. Within the mean field approximation as described by (9. u and Urm .9. ˆ SCF H SCF .142). Exercise 9.

272 Many–Particle Systems 9.h. The one-dimensional Hubbard model is described by the Hamiltonian ˆ H = −t rσ † c† r+1.e. v (9.178) and through the so-called filling factor n n= 2N N = . in the usual two–particle operator form.h. In spite of the simplicity of its Hamiltonian the Hubbard model. s|ˆ|u.128). labeled r = −N0 + 1. v of the Hubbard Hamiltonian (9. v (9. rσ rσ r σσ (9. r ∈ Z.181) The Hubbard model. s|ˆ|t. the second term on ther. In the present Section we want to adapt the algorithm to systems containing a macroscopically large number of fermions.8 Mean Field Theory for Macroscopic Systems The SCF algorithm has been developed in Sections 9. in the two-dimensional copper oxide lattices of high temperature superconductors. The first term on the r. u c† c† cuσ ctσ .179) r+1. is identical to that of the independent–particle Hamiltonian H0 (9. Both N0 and N are macroscopically large numbers. of (9. t = v δru δst δrs . † ˆ Ho = −t c† . respectively) in state |r at lattice site r.178) v contributes only in case that two electrons occupy the same lattice site..178).u σσ where r.σ rσ The potential energy term.s. v represents the ‘on-site’ Coulomb repulsion. reads 1 ˆ r. . According to (9. Due to the lack of an exact solution in dimensions higher than one. approximation schemes like the self-consistent field approximation play an important role. is characterized through the parameters t.σ + v 2 c† c† crσ crσ . S N0 (9. . The Hubbard Model The Hubbard model serves today the important role as the simplest manifestation of strongly correlated electron systems which arise in many instances in molecular and solid state physics. The case n = 1 is termed the half filled band case. The operator c† (crσ ) creates (destroys) an electron with spin σ (σ = α.s. t is assumed to be a positive.178).σ crσ + crσ cr+1.192) derived below is filled half.σ crσ + crσ cr+1. N0 which are populated by 2N electrons (we choose an even number of electrons for convenience). .s.182) n can assume values 0 ≤ n ≤ 2. t describes the coupling of state |r to states |r ± 1 at the two neighboring lattice sites. describes the sites of a linear lattice. real number. except in case of one-dimensional systems. cannot be solved exactly. .e. We assume that there are altogether S = 2N0 lattice sites... referring to the fact that in this case the band of single particle energies (9.180) V = rσ sσ 2 r.7 for finite systems. We consider as an example the Hubbard model of an infinite linear lattice and apply the model to describe magnetic instabilities in metals. .178). i. i. of (9. as stated through Hamiltonian (9.4–9. (9.178) Here the index r.e. β for ‘up’ and ‘down’ rσ spins. i. for example.t.

The unitary transformation which diagonalizes H0 . the operator H0 can be re-written in the diagonal form N0 ˆ H0 = m=−N0 +1 σ m d† dmσ mσ (9.154).178) within the framework of the self-consistent field theory.144).140). Accordingly.σ m=0 ∗ Urm Urm c† crσ . 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. We can now embark on step 2 of the SCF algorithm. by N −1 ||SCF = m=0 d† d† |0 .6.. . in fact.155.186) where m = −2t cos mπ . one obtains for the mean field Hamiltonian N −1 ˆ ˆ Hmf (U) = H0 + v r. 9. boundary effects are assumed to be negligible.183) where the operators d† are related to the original creation operators c† through the unitary mσ rσ ˆ transformation U defined in (9. rσ rσ (9.188) .185) into (9. For this purpose we apply the SCF theory presented in Section 9. Inserting (9. we assume that the SCF ground state is given by (9. The self-consistent field Hamiltonian Hmf which corresponds to the Hubbard Hamiltonian (9. 2N0 (9.185) ˆ According to (9.181.174). Accordingly. This task has ˆ been solved already on page 262. < ±(N0 −1) < N0 holds as can be readily verified. SCF Ground State for the Hubbard Model We seek to determine the ground state of the Hubbard model stated by (9. mα mβ (9. N0 (9. 9. Using (9. rσ (9. we identify |r + N0 = |r and later let N0 go to infinity.8: Macroscopic Systems 273 Since we are dealing presently with a macroscopic system. i.9.179). .187) The energy levels are labeled such that 0 < ±1 < .186) one obtains for the mean field Hamiltonian N0 ˆ (2) Hmf = m=−N0 +1 σ m d† dmσ + v mσ N 2N0 c† crσ .184) and using (9.178) can be determined by applying (9. is Urm = √ 1 exp (irmπ/N0 ) .e.184) ˆ We are now ready to apply step 1 of the SCF algorithm and diagonalize term H0 .

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.. Alternative Description of the Mean Field Approximation The state (9.182)]. Indeed. One can see that −2t ≤ (k) ≤ 2t holds. 4 So far we have not exploited the fact that N0 and N are macroscopically large quantities. (9. by filling up this energy band with electrons from the bottom of the band (which corresponds to k = 0) to a maximum energy. crσ through d† . .187) form a quasi1 2 continuous energy band. 9. Indeed. −π + N0 . can be obtained.187).190) We have used in the latter expression the definition (9. To demonstrate this we consider an alternative formulation of the mean field approximation for Hamiltonian (9.191) where m is given in (9. π].f. hence. denoted by F (see Fig.178). .1. through the so-called dispersion law (k) = −2t cos k . the results of different formulations agree qualitatively. The ground state of the system. . the ground state of the Hubbard model in the self-consistent field approximation can be determined exactly. The energy m can be expressed as a function of a continuous variable k ∈] − π. π]. The bottom of the energy band corresponds to (0) = −2t and the width of the energy band is 4t. the state with the lowest possible energy..e.190). . dnσ according to (9. even for one and the same Hamiltonian. i.e.274 Many–Particle Systems Expressing c† . In other words. 9. the electrons in the present description behave like independent particles with energies ¯m = m + vn . When N0 becomes macroscopically large the single particle discrete energy levels (9. i.192) This dispersion law is presented in Fig. holds k ≡ km ∈] − π. as given in (9. for km = mπ/N0 holds km ∈ −π + N0 . −π ≤k ≤π . Usually. Apparently. The energy Emf of this state can be calculated readily by using (9. in the limit N0 → ∞.183) is not the only candidate for the mean field ground state of the Hubbard model.189) and.109) and using (9.1). N0 ˆ (2) ˆ Hmf = Hmf = m=−N0 +1 σ m + vn † dmσ dmσ 2 (9. according to the Pauli principle.185) one can prove rσ mσ N0 c† crσ = rσ rσ m=−N0 +1 σ d† dmσ . but may differ quantitatively. there exist several ways of formulating mean field approximations.177) and mm |ˆ|mm = mm |ˆ|m m = ˜˜ v ˜˜ ˜˜ v ˜ ˜ which holds in the present case. called the Fermi energy. is already diagonal. (9.182) of the filling factor n [c. mσ (9. namely. π and. ˆ The Hamiltonian Hmf . the self–consistency condition is exactly met and the SCF algorithm converged after two steps.

178) in terms of the occupation number operaˆ tors Nrσ = c† crσ .1: The dispersion law (k) for independent electrons on an infinite one dimensional lattice.45.183). 9. the operator Nrσ can be written ˆ Nrσ = Nrσ + δ (Nrσ ) . F = (±kF ) denotes the Fermi energy.194) ˆ For a given reference state.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 .9.e. the deviation of Nrσ from its mean value Nrσ for the corresponding reference state. Inserting (9.8: Macroscopic Systems 275 Figure 9. where Nrσ is the mean occupation number of the one particle state |rσ .193) into (9.195) ˆ ˆ ˆ and where δ (Nrσ ) = Nrσ − Nrσ describes the fluctuations of Nrσ . (9. we have employed (9.. say the one given in (9. i. (9. rσ rσ rσ (9. The anti-commutation properties (9. The mean field approximation neglects the ..196) (9. in the last term on the right hand side. First. Nrσ = SCF ||Nrσ ||SCF .56).178) and performing the summation over the spin indices results in ˆ ˆ H = H0 + v r ˆ ˆ Nrα Nrβ .e.193) where. let us express the interaction term in (9. i.

Spin-Polarized Mean Field Ground State ˆ We want to consider now a ground state for Hmf defined through Nrσ = nσ = const (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 . For such state the mean occupation number nrσ is uniform at all lattice sites.198) as the one obtained by using the mean field potential (9. .183) ..m ∗ Urm Urm d† dm σ mσ = m=0 ∗ Urm Urm = n N = . Because of Nrσ = c† crσ rσ N −1 = m.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. that nα and nβ assume different values. (9.199) which is identical to expression (9.191) for the ground state energy.195). 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 . m+ 2 2 c† crσ − 2N0 rσ (9. however. the ground state of the system can have a non-zero local spin and.276 Many–Particle Systems fluctuations to second order and the mean field Hamiltonian is obtained from (9. separating the occupation number operator into a mean value plus fluctuation and neglecting the fluctuations in 2nd order. consequently.183). non-vanishing magnetization.194) by dropping all the terms which contain the fluctuations to second order.197) and the corresponding mean field Hamiltonian becomes ˆ ˆ Hmf = H0 + v r ˆ ˆ Nrα Nrβ + Nrβ Nrα − v r Nrα Nrβ .146). Therefore.200) allowing. in the mean field approximation. yields essentially the same mean field Hamiltonian (9. Note that the procedure employed in (9.e. is given by (9. i. In case nα = nβ the following construction will lead to a ground state which coincides with the non-magnetic ground state (9.

E with respect to nα . According to this method one minimizes ˜ Emf = Emf + µ (n − nα − nβ ) . β) are determined by minimizing the energy density of the ground state Hmf 1 = [ mσ Nmσ ] − vnα nβ (9.g.205) is met..202) This Hamiltonian describes a system of non-interacting electrons with a spin-dependent dispersion law (9. (9. (9. In the limit N0 → ∞ this sum can be expressed as an integral. the so-called Fermi energy F = µ (see below). m (9. the smaller is σ (k) for a given k.σ with respect to nα and nβ .206) needs to be evaluated. At this point we exploit the fact that our system is macroscopically large. Since β (k) − α (k) = v(nα − nβ ). For this purpose the sum in (9.e. nβ and µ. The values of nα and nβ are determined by minimizing the energy density (9. 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 ). as is shown schematically in Figure 9. e.192). The actual values of nσ (σ = α.2a. (9. such that the larger nσ .203) is provided by the continuous function σ (k) = (k) + vn−σ . i. one can see that an uneven occupation by electrons of the two energy bands yields a relative shift of the energy bands with respect to each other.201) ( m ˆ + vnβ ) Nmα + ( m ˆ + vnα ) Nmβ − 2N0 vnα nβ .9.206) ˜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.207) where (k) is given by (9. for any function f mπ holds (compare with the definition of a definite integral N0 as the limit of the corresponding Riemann sum) 1 N0 →∞ 2N0 lim N0 f m=−N0 +1 mπ N0 π = −π dk f (k) .203) mσ = m + vn−σ . The additional term in (9.206). (9. N0 ∼ 1023 .206) ascertains that condition (9.. a spin–polarized ground state. 2π (9. i. In this limit the discrete energy spectrum (9. In fact.204) Emf ≡ 2N0 2N0 m. subject to the constraint n = nα + nβ .205) Such minimization is realized through the method of Lagrangian multipliers.e. one can expect the system to lower its energy by assuming nα = nβ and. Therefore.208) . hence. if nα > nβ then β (k) > α (k)..8: Macroscopic Systems from which we obtain nσ = The mean field Hamiltonian is then ˆ Hmf = m 277 1 2N0 Nmσ .

209) where ρ( ) = −π dk δ ( − (k)) . which is usually called the density of states. For a given dispersion law (k) the density of states ρ( ) can be calculated by employing equation (9.211) and using π dkδ(k ± (π − arccos( /2t))) = θ(2t − | |) . therefore. 2π (9. (b) Graphical definition of the limiting energies α and β . and noting |f (k)| = |2t sin k| = 2t one obtains δ( − (k)) = 1 − cos2 k = (2t)2 − 2 . If the function f depends explicitly only on (k) one can state π −π dk f ( (k)) = 2π π ∞ ρ( )d f ( ) −∞ (9. namely k1.192) and.2: (a) Relative position of the two energy sub–bands α (k) and β (k) for ∆ = nα − nβ > 0. 2π (9. unit energy and a given spin orientation of the particle.211) In order to calculate the integral we recall the following property of the Dirac–delta function δ [f (x)] = i δ(x − xi ) . Inserting this last result into (9.210). δ(k + π − arccos( /2t)) + δ(k − π + arccos( /2t)) (2t)2 − 2 .2 = ±(π − arccos( /2t)).210) Here δ(x) is the Dirac–delta function and ρ( ). gives the the number of available one particle states per site. Since the equation f (k) ≡ 2t cos k + = 0 has two simple roots.212) where xi are the simple roots of the function f (x). −π . holds π ρ( ) = −π dk δ( + 2t cos k) .278 Many–Particle Systems (a) (b) Figure 9. The shaded areas represent the filled portions of these bands by electrons. In our case (k) is given by (9. |f (xi )| (9.

. v and n. (9. (9. ∂ β ˜ ∂ Emf =0. We can employ the results obtained to express the ground state energy density (9. namely.221) and (9.. From (9.f.219) (9. one arrives at the following expression of the density of states ρ( ) = θ(2t − | |) π (2t)2 − 2 . 9.9. i.220) µ can be readily obtained µ= 1 ( 2 α + β) + vn .222) .2]. on conditions for a ground state of minimum energy are ˜ ∂ Emf =0.e. (9.216) Using (9. θ(x) = 1 if x > 0 and θ(x) = 0 if x < 0.220) β ˜ ∂ Emf = n − nα − nβ = 0 . F = α + vnβ = β + vnα . β . (9.204) through an integral expression. if we assume that is restricted to the interval ] − 2t. + vnα − µ = 0 .e. i. to determine the unknown quantities α .221) ∂µ Equations (9. ∂µ (9. µ.215) nα and nβ in (9.213) can simply be replaced by 1.214) where σ denotes the value of (k) which corresponds to the top of the filled portion of the energy sub–band σ (k) [c.214) are given by the sum (9. (9.219. outside the energy band (k). nα and nβ as a function of the parameters of the Hubbard model.8: Macroscopic Systems 279 where θ(x) is the step function.219–9.218) read ˜ ∂ Emf = ∂ α ˜ ∂ Emf = ∂ β and α + vnβ − µ = 0 .206)] depends only on continuous quantities. in principle.213) The presence of the θ function in the above formula guaranties that the density of states vanishes for | | > 2t.. i. Replacing again the sum by an integral over energy one can write σ nσ = −2t ρ( )d (9. namely. ∂ α ˜ ∂ Emf =0. β (9. the necessary ˜ Since Emf [c. of t.f. π α.201).213) and carrying out the resulting integral yields nσ = 1 arcsin( σ /2t) .e. (9. 2 (9. Fig. 9.218) The derivatives can be readily determined and the conditions (9.216) allow one. 2t[ then the θ function in (9.217) and µ. One obtains ˜ Emf = α β ρ( )d + −2t −2t ρ( )d + vnα nβ + µ(n − nα − nβ ) .

from (9.224) v∆ ∆= β ρ( )d = 0 ρ( β + )d .227) Since f (0) = 0 one can infer that ∆ = 0 is always a solution (the so called paramagnetic solution) of condition (9. (9. and only if. (9.e. One still needs to determine α and β .227). First.227) will have a second.226) ∆ = f (∆) .f. If the slope of f (∆) at the origin is less than 1.3. (9.. the Hubbard model has a ground state with ∆ = 0. Figure 9. Indeed.228)] vc ρ ( β ) = 1 .215) reveals that µ is indeed equal to the Fermi energy F (see also Figure 9. i.228) Since the total number of states per site is finite. condition (9. from (9.223 one obtains ∆= and second. For v above a critical value vc .2b). The slope of f (∆) at the origin depends on the Hubbard model parameter v.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.229) has a wider range of validity than one can infer from the present analysis of the Hubbard model.223) follows α α − v β .222) and (9.217) and (9. a magnetic ground state. We assume in the following discussion that the implicit ∆ dependence of β in f (∆) can be neglected. Accordingly. if. f (∆) converges to a maximum value as ∆ → ∆max = n. condition (9.f. From the definition (9. For this purpose we consider the magnetization of the mean field ground state ∆ ≡ nα − nβ . Equation (9. For this purpose one plots f (∆) versus ∆ as is shown schematically in Figure 9. non trivial. The Stoner criterion (9.183) without magnetization.227) has only the trivial solution ∆ = 0.227) can be solved graphically. the slope of f (∆) is positive for all ∆ ≥ 0. through the condition [c. (9. as reflected by the expression d[f (∆)] = vρ( d∆ β + ∆) = π (2t)2 v − (v∆ + 2 β) >0.219–9. (9. One can express ∆ as a function of α and β in two different ways. (9.3].220) and (9.225) Defining the function v∆ f (∆) = 0 ρ( β + )d .. the slope of f (∆) at the origin is larger than 1 [c. (9.280 Many–Particle Systems Comparision of (9.226) follows that f (∆) is a monotonically increasing function of ∆. (9.e. so called ferromagnetic solution. . (9.223) ∆ = 0 corresponds to the ground state (9. One can determine vc by equating the slope of f (∆) at the origin to 1. i.226) one can obtain ∆ by solving (cf.

220. the magnetization ∆ of the ground state vanishes and remains zero for values v < vc . Equation (9. any . n).205) the following set of equations determining x.233) yield the desired expression vc π = π cos (n − 1) . hence.231) (9. (9. α.9. 2 2 2 (9. β = 2t sin y .230) one obtains from (9. i.8: Macroscopic Systems 281 Figure 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.e. the ground state is characterized by the quantities (v.232) together with (9. From (9. The corresponding 2t analytical expression provides the phase diagram of the ground state: when v approaches vc from above.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. (9. is presented in Figure 9. y and. 9. 9. 2t As already mentioned. (9. n. Accordingly. We like to determine finally the dependence of vc on the filling factor n.4. the plot of vc vs. v (9.. By parameterizing α and β α = 2t sin x .3: Graphical solution of the mean field equation ∆ = f (∆). t.219.234) 2t 2 The resulting phase diagram of the ground state. β x+y x−y = = π(n − 1) 2t π (sin x − sin y) = π∆ .232) states that x − y is proportional to ∆.233) Employing the approximation sin(y + δ) − sin y ≈ δ cos y for small δ.

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

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

284 Many–Particle Systems .

v = v1 x1 . We will find that actually several Lorentz–invariant equations which replace (10.e. e. spin– 1 particles. (refeq:ham2. x2 . the equations need to be interpreted as actually describing many–particle-systems: the equivalence of mass and energy in relativistic formulations of physics allows that energy converts into particles such that any particle described will have ‘companions’ which assume at least a virtual existence. x3 = x3 (10. t) ∂ i ψ(r. any of these equations being specific for certain classes of particles. t) c 2 i + qV (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. t) = ∂t 1 2m q − A(r.2) The replacement of (10.f. t) are vectors of dimension larger one. the components representing the spin attribute of particles and also representing together with a particle its anti-particle.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.. x3 . Furthermore. The transformations beween such frames according to the Theory of Special Relativity are described by Lorentz transformations. The representation 285 . In case that v is oriented along the x1 –axis.. etc. x2 . t) ψ(r. x2 = x2 . t) and a vector potential A(r. t) (10.g..1) which connect space time coordinates (x1 . 2 As mentioned.45)] described by an electrical potential V (r. some of the equations describe a particle together with its anti-particle. Obviously. t ) in another frame.2) will result. It is not possible to uncouple the equations to describe only a single type particle without affecting negatively the Lorentz invariance of the equations. t = t− 1 − v1 x c2 1 v1 2 c . i. these ˆ transformations are x1 = x1 − v1 t 1 − v1 2 c . i. 8. t) in one frame with space time coordinates (x1 . 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. x3 .2) by Lorentz–invariant equations will have two surprising and extremely important consequences: some of the equations need to be formulated in a representation for which the wave functions ψ(r.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. spin–0 particles. Here c denotes the velocity of light.

The reason for this choice is that the transformations (10. but in a setting of representation theory methods as applied in Secti. t) as components. however. The elements L ∈ L act on 4–dimensional vectors of position– and time–coordinates. wave functions ψ(r. We will denote these vectors as follows def xµ = (x0 .7) equation. Denoting by xµ the . i. x 1 . the Lorentz transformations were formulated in a space in which the time component of xµ was chosen as a purely imaginary number and the space components real.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. the group of Lorentz transformations (10.3) where x0 = ct describes the time coordinate and (x1 .1). i. x 2 . 10.5) and the Dirac (Sect. 10. for the transformed space-time–cordinates x µ = (x 0 .e.1) leave the quantity s2 = (x0 )2 − (x1 )2 − (x2 )2 − (x3 )2 (10. x2 . x3 ) = r describes the space coordinates. We will. x3 ) (10. x 3 ) holds (x0 )2 − (x1 )2 − (x2 )2 − (x3 )2 = (x )2 − (x )2 − (x )2 − (x )2 .7 a heuristic derivation of the two most widely used and best known Lorentz–invariant field equations.1). In such a space Lorentz transformations correspond to 4-dimensional rotations. This space is called the Minkowski space. we will provide a more basic definition of the transformations. This will provide us with a general set of Lorentz–invariant equations which for various particles take the place of the Schr¨dinger equation.1 Natural Representation of the Lorentz Group In this Section we consider the natural representation of the Lorentz group L. t) and vectors with functions ψ(r.286 Relativistic Quantum Mechanics in Sect. Rather than starting from (10.. henceforth. Rather than following this avenue we will introduce Lorentz transformations within a setting which does not require real and imaginary coordinates.e. namely the Klein–Gordon (Sect. 10. x2 . The Group of Lorentz Transformations L = O(3. Before introduco ing these general Lorentz–invariant field equations we will provide in Sects. assume new units for time such that the velocity of light c becomes c = 1. 10. 0 1 2 3 (10. Minkowski Space Historically. 10. Note that the components of xµ all have the same dimension.1). This choice implies dim(time) = dim(length).5) √ One can interprete the quantity −s2 as a distance in a 4–dimensional Euclidean space if one chooses the time component purely imaginary.1 will be extended in Sect.. i. We will find that this definition will lead us back to the transformation law (10.4) invariant.5. 10.e. x1 .4 to cover fields. namely that of length. 10. 5 to the groups SO(3) and SU(2) of rotation transformations of space coordinates and of spin.

Since this holds for any xµ it must be true Lµ ρ gµν Lν σ = gρσ . Aµ ν B ν ρ = ν=0 Aµ ν B ν ρ . (10. (10.9) and (10. 3 3 3 yµ xµ = new yµ xµ .6) x µ = Lµ ν xν .6) allows us to introduce the so-called summation conventon: any time the same index appears in an upper and a lower position.e.12) (10. the Lorentz transformations constitute a linear transformation which we denote by 3 x µ = ν=0 Lµ ν xν . R). ν = 0. the latter set constituting a group.5) invariant. [ A more conventional notation would be Lµν . L ∈ GL(4.6) Here Lµ ν are the elements of a 4 × 4–matrix representing the Lorentz transformation. Aµ ν . 0 0 0 −1 (10. i.. This will be explained further below. R).10. Aµν .10) Combining condition (10. First.e. the . the notation in (10.9)  1 0 0  0 −1 0 0   ( gµν ) =   0 0 −1 0  = g .] The following possibilities exist for the positioning of the indices µ. the latter notation will be used for different quantities further below. 2. 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 subset of GL(4. We will exploit this property below to determine the general form of the Lorentz transformations. ν=0 (10. xµ . 4 × 4 tensor: Aµ ν . however. Aµν . This condition can be written xµ gµν xν = x gµν x where  µ ν (10. Aµ ν xν = µ=0 new old Aµ ν xν .6) yields Lµ ρ gµν Lν σ xρ xσ = gρσ xρ xσ . The Lorentz transformations form the subgroup of all matrices which leave the expression (10.8) new old old The summation convention allows us to write (10. The Lorentz transformations are non-singular 4 × 4–matrices with real coefficients.5) into scalar products.11) This condition specifies the key property of Lorentz transformations.7) The reason for the notation is two-fold. (10.1: Natural Representation of the Lorentz Group 287 coordinates in one reference frame and by x µ the coordinates in the other reference frame. The second reason is that upper and lower positions allow us to accomodate the expression (10. 3: 4-vector: xµ . summation over that index is assumed without explicitly noting it. 1. i.

17) one obtains (L−1 )T g L−1 = gLgggLT g = gLgLT g . This stipulates that O(3.20) Multiplying (10. of O(3.15) (10.1). = g LT g =  0 1 2  −L 2 L 2 L 2 L3 2  −L0 3 L1 3 L2 3 L3 3 (10. In fact.16) The corresponding expression for (LT )−1 is obviously (LT )−1 = (L−1 )T = g L g . property (10. is called O(3. 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   (10.1) is.16. a group.e.e.1) ⊂ GL(4.. the associated inverse obeys (10.. For L3 = L1 L2 holds LT g L3 = LT LT g L1 L2 = LT (LT gL1 ) L2 = LT g L2 = g . We consider 1 then L1 .14).1) since it satisfies (10.1).14)  L0 0 −L1 0 −L2 0 −L3 0  −L0 1 L1 1 L2 1 L3 1   .h.12) LT gL = g . . To demonstrate the group property of O(3.21) i. This set is identical with the set of all Lorentz transformations L. R) is a group.18) (10.10) of g one can rewrite the invariance property (10. (10. (10.1).14) holds for the inverse of L.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. (10.. i.s. We want to show now L = O(3. 1) = { L.14). 3 2 1 2 1 2 (10. employing expressions (10..22) (10. 10.e. LT gL = g } . L−1 ∈ O(3. in fact.1).20) results in the desired property (L−1 )T g L−1 = g . L ∈ GL(4. L3 ∈ O(3. hence.e. One can also show that if L ∈ O(3. L2 ∈ O(3. R).1).17) we note first that the identity matrix 1 is an element of O(3. the inverse of L 1  L−1 (10.1).288 Relativistic Quantum Mechanics elements of which satisfy this condition. i. From this one can obtain using g2 = 1 1 (gLT g)L = 1 and.13) L = ( Lµ ν ) =  2 0  L 0 L2 1 L2 2 L3 3  .14) from the right by gLT and using (10.19) i. L3 0 L3 1 L3 2 L3 3 Using the definition (10. of (10.

14). M1 = a M2 }. L ∈ O(3. L0 0 ≥ 1} .1). det L = −1. Schwartz’s inequality yields j=1  2 3 3 3  j=1 A jB 0  ≤ 0 j A 0 2 j Bj 0 2 . It holds g ∈ L and −1 ∈ L as one can readily verify testing for property (10. B ∈ L+ holds C = AB ∈ L+ . Considering in (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 . The above shows that the set of proper Lorentz transformations L↑ allows one to generate all Lorentz transformations. + We start our investigation by demonstrating that L↑ forms a group. A second class property follows from (10. L↓ .30) ≤ −1} . L↑ . 1). L↓ are disjunct sets defined as follows + + − − L↑ + L↓ + L↑ − L↓ − = { L. L0 0 ≥ 1} . L0 0 ≤ −1} . (10. A statement on this value can be made on account of property (10.14) which we employ in the formulation (10. From this we can conclude L0 0 ≥ 1 or L0 0 ≤ −1 .34) j=1 j=1 . Using det AB = det A det B and det AT = det A yields (det L)2 = 1 or det L = ±1 . 1).12). M2 ∈ M.26) (10. = { L. + except for the trivial factors g and −1 It is. 1). = { L.31) (10. det L = 1. (10.25) can be stated as follows: The set of all Lorentz transformations L is given as the union L = L↑ ∪ L↓ ∪ L↑ ∪ L↓ + + − − where L↑ . Properties (10. + + ↑ ↑ We can also demonstrate that for A. det L = −1. σ = 0 yields L0 0 2 − L1 0 2 − L2 0 2 − L3 0 2 = 1. there exist two other distinct classes.24) or since (L1 0 )2 + (L2 0 )2 + (L3 0 )2 ≥ 0 it holds (L0 0 )2 ≥ 1.12) the case ρ = 0. Obviously. One can also 1 verify that one can write L↑ − L↓ + L↓ − = gL↑ + = = −L↑ . (10. + = − L↑ g + (10.10. For this purpose we consider first the value of det L. 1).27) (10.23) and (10.14). Lorentz transformations in L↑ . (10.32) (10.. L ∈ O(3.28) (10.1: Natural Representation of the Lorentz Group Classification of Lorentz Transformations 289 We like to classify now the elements of L = O(3. = { L. hence. L0 0 (10. entirely suitable to investigate first only 1. L↑ contains 1 1.29) (10.25) i.23) One can classify Lorentz transformations according to the value of the determinant into two distinct classes. L ∈ O(3. ∃M2 . + − gL↑ + = L↑ g . det L = 1.33) where we used the definition aM = {M1 . L ∈ O(3.e.

therefore. L↑ is called the subgroup of proper.1).12) follows (B 0 0 )2 − 3 (B j 0 )2 = 1 or 3 (B j 0 )2 = (B 0 0 )2 − 1. (10. 1 We will then employ the Lie group properties to generate all transformations in L↑ . obviously j=1 A0 0 B 0 0 ≥ 1. .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 . 1 ∈ L↑ . + Accordingly. Similarly. ρ ν (10..14) actually L0 0 ≥ 1 and det L = 1 must hold.35) One can conclude. For the inverse of + ↑ any L ∈ L+ holds (10. δ00 = 1 and It should be noted that according to our present definition holds δµν = gµρ δ δ11 = δ22 = δ33 = −1.34) provides then the estimate conclude from (10. Since the associative property holds for matrix multipli1 1 + ↑ cation we have verified that L+ is indeed a subgroup of SO(3..39) where we have employed the matrix form T 1 defined as in (10.290 Relativistic Quantum Mechanics From (10. from which one can conclude L−1 ∈ L↑ . obviously. orthochronous Lorentz transformations.1). Since A0 0 ≥ 1 and B 0 0 ≥ 1. i. In fact. 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 (10.40) and. To order O( 2 ) holds g + g = 0.1) ascertains CT gC = g it must hold C 0 0 ≥ 1.36) It holds. 1 0 0 = ≥ 1 and. In the following we + will consider solely this subgroup of SO(3. since the group property of O(3.16). This relationship shows (L−1 )0 0 = L0 0 . (10. accordingly. The next group property of L↑ to be demonstrated is the existence of the inverse. 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 1.14) implies 1 + T g (1 + ) = g 1 1 (10. + i.13). if we enforce (10. we consider transformations Lµ ν = δ µ ν + µ ν .37) (10. Using the above expression for C 0 0 one can state C 0 0 > 0. one can j=1 j=1 3 0 )2 = (A0 )2 − 1. Infinitesimal Lorentz transformations The transformations in L↑ have the property that they are continously connected to the identity 1 1.e. (10. | 3 A0 j B j 0 | < A0 0 B 0 0 .38) For these transformations.e. hence. µ ν small . Property (10. + transformations in a small neighborhood of 1 which we parametrize by infinitesimal parameters. holds L0 0 > 0 and the value of the determinant is close to unity.

2. w1 .45) (10.41) which reads explicitly     0 0 0 0 0 1 2 3 1 1 1 1 0 1 2 3 2 2 2 2 0 1 2 3 3 3 3 3 0 1 2 3     =     − 0 1 2 3 0 0 0 0 0 1 1 1 1 0 2 2 2 2 0 3 3 3 3 − − − 1 2 3 − − − 1 2 3 − − − 1 2 3    .49) . 3) Jk = (ϑk = 1. other five parameters zero) Kk = (wk = 1. ϑ3 . 2.47)     0 −1 0 0 0 0 −1 0 0  −1   0 0   0 0 0 0  0 0  K1 =  . 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. w2 . k j j = 1. (ϑ1 . J =  0  3  0 1 0 0 0 0 0 0 0 0 0 0 0     (10.10. K3 =   0  −1 0  0 0 0 0  0 0  0 0 0 0 −1 0 0 0 0 These commutators obey the following commutation relationships [ Jk .48) (10. K2 =  . The generators are explicitly  0 0 0 0  0 0 0 0 J1 =   0 0 0 −1 0 0 1 0  (10.46)  0 0   0 0  . = − k m Km (10. J ] [ Kk . 2. 3 (10.1: Natural Representation of the Lorentz Group Using (10.42) This relationship implies µ 0 j µ j = 0 = j 0 .43) k = − Inspection shows that the matrix has 6 independent elements and can be written   0 −w1 −w2 −w3  −w1 0 −ϑ3 ϑ2   .  (10. w3 ) =   −w2 ϑ3 0 −ϑ1  −w3 −ϑ2 ϑ1 0 (10.15) one can conclude T 291 = −g g (10.50) . other five parameters zero) . k = 1. K ] [ Jk .44) This result allows us now to define six generators for the Lorentz transformations(k = 1. ϑ2 . j. J2 =    0 0 0 −1  0 0 0 0   0 0 0 0 0  0 0 −1 0 1  . 3 .

To determine the explicit form of this transformation we evaluate the exponential operator by Taylor expansion. w ∈ R3 3 k=1 wk Kk . + The commutation relationships imply that the algebra of the generators Jk . k = 1. Kk . ϑ. i. Following the treatment of the rotation group SO(3) one can express the elements of L↑ through + the exponential operators L(ϑ. In analogy to equation (5. For the parameters ϑk of the Lorentz transformations holds obviously ϑk ∈ [0.. Relativistic Quantum Mechanics The commutation relationships (10. however.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   . 2π[ . Exercise 7. 2. (10.51) for w = 0 correspond to rotations of the spatial coordinates.e. A boost in the x1 –direction is L = exp(w1 K1 ). w) = exp ϑ · J + w · K .52) where the 3 × 3–matrices Lk are the generators of SO(3) defined in Chapter 5. L(ϑ. w = 0) = 0 0 0 R(ϑ) .54) which. 3 (10. that the transformations (10.50). 2. 10.49. 3 is closed.53) Here R(ϑ) are the 3 × 3–rotation matrices constructed in Chapter 5.51) One can readily show.292 as can be readily verified.1: Demonstrate the commutation relationships (10. 3 where we have defined ϑ · J = k=1 ϑk Jk and w · K = following the algebra in Chapter 5. k = 1.56) = 1 0 0 1 = 1 1 (10. and using the relationship Jk = 0 0 0 Lk (10. (10.49) define the Lie algebra associated with the Lie group L↑ .55) L 0 0 0 0  exp (w1 K1 ) = exp   Using the idempotence property 0 −1 −1 0 2 0 0 1 0 (10. constitutes an overcomplete parametrization of the rotations (see Chapter 5). 0  1 n (10. We consider now the Lorentz transformations for ϑ = 0 which are referred to as ‘boosts’.57) .

6.59) for the case k = 1 corresponds to the relative velocity of two frames of reference. 1[.60) sinh w1 = v1 / 2 1 − v1 .59) cosh w1 Using cosh2 w1 − sinh2 w1 = 1 one can identify sinhw1 = sinh2 w1 + 1.59) follows. vk defined in (10. we can state vk ∈ ] − 1. 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.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. x3 = x3 (10. cosh2 w1 − 1 and coshw1 = cosh2 w1 − 1 = cosh w1 sinh w1 sinh2 w1 + 1 .62) According to (10. 10. . for wk wk ∈ ] − ∞.1).59) v1 = These two equations yield cosh w1 = 1/ and (10. Accordingly.51) the explicit transformation for space–time–coordinates is then x1 = x1 − v1 t 1 − 2 v1 .64) We note that the range of wk -values is not a compact set even though the range of vk -values is compact. From (10. (10. ∞[ .59) can be written  1 2 1 − v1 √−v1 2 1 − v1 2 1 − v1 . Correspondingly. (10. 10. (10. x2 = x2 . This property is. We expect that vk can only assume values less than the velocity of light c which in the present units is c = 1. one obtains from (10.1: Natural Representation of the Lorentz Group one can carry out the Taylor expansion above: ∞ 293 L = n=0 2n w1 1 + 1 (2n)! ∞ n=0 2n+1 w1 (2n + 1)! 0 −1 −1 0 = cosh w1 −sinh w1 −sinh w1 cosh w1 . 10. t = t − v1 x1 2 1 − v1 .10.59).63) which agrees with (10. The range of the parameters wk can now be specified.58) = cosh w1 1 + sinh w1 1 0 −1 −1 0 The conventional form (10.3. however. in fact.1) of the Lorentz transformations is obtained through the parameter change sinh w1 v1 = = tanh w1 (10. (10.56. consistent with (10.

namely. we have not yet defined representations of Lorentz transformations beyond the ‘natural’ representation acting in the 4–dimensional space of position– and time–coordinates. (10. Hence. a3 )T and transform according to a µ = Lµ ν aν . A) (10.69) where V (r. x2 . t) are the electrical and the vector potential of an electromagnetic field. 4–Vectors and Tensors In this Section we define quantities according to their behaviour under Lorentz transformations.68) where ρ(r. We have encountered examples in connection with rotational transformations. p = √ 2 1−v 1 − v2 (10. E = √ m mv . These quantites always come as four components (a0 . x3 )T . . of a system of charges. finally. We will see below how true scalars under Lorentz transformations can be constructed. 4-Vectors The quantities with the transformation behaviour like that of the position–time vector xµ defined in (10. t)|2 or the scalar product r1 · r2 of two particle positions. (10. a2 . scalars like r = x2 + x2 + x2 . f = f.4). 2 |ˆ1 . the charge density. our definition of quantities with special properties under Lorentz transformations presently is confined to the natural representation. Counterexamples are the energy. t) and J(r. t) and A(r. a1 . Some of these r ˆ quantities were actually defined with respect to representations of the rotation group in function spaces. not in the so-called natural representation associated with the 3–dimensional Euclidean space E3 . Another example is the potential 4-vector Aµ = (V. A third 4-vector is the so-called current vector J µ = (ρ. the z–component x3 of a particle. we will encounter an impressive example of physical properties. The 4-vector character of J µ and of Aµ will be demonstrated further below.2 Scalars.e. Scalars The quantities with the simplest transformation behaviour are so-called scalars f ∈ R which are invariant under transformations. Presently.3) are the so-called 4–vectors aµ . respectively. r2 ) and.65) An example is s2 defined in (10. t) are the charge density and the current density. J) (10. Nevertheless. the square of the electric field |E(r.67) the transformation behaviour of which we will demonstrate further below. i. However. total angular momentum r 1 2 3 states of composite systems like Y m ( 1 .66) Examples of 4-vectors beside xµ are the momentum 4-vector pµ = (E. Such quantities appear in the description of physical systems and statements about transformation properties are often extremely helpful and usually provide important physical insight. vectors like r = (x1 ..294 Relativistic Quantum Mechanics 10. not any physical property f ∈ R is a scalar. p) . tensor operators Tkm . another example is the rest mass m of a particle. spherical harmonics Y m (ˆ).

77) The transformed differential operator will be denoted by ∂µ = def ∂ .73) Note that according to (10.72) of the covariant 4-vector follows = In µν and g µ as well.76) An important example of a covariant 4-vector is the differential operator ∂µ = ∂ = ∂xµ ∂ .71) Contravariant and Covariant 4-Vectors It is convenient to define a second class of 4-vectors. (10. ∂x µ (10. the relationship being aµ = gµν aν = (a0 .70) is a scalar. accordingly. ∂t (10.66). One calls 4-vectors aµ covariant and 4-vectors aµ contravariant.e. We like to point out that from definition (10. In fact.75) (10. −a3 ) (10. i.16).e. xν = g νσ Lρ σ gρµ x µ ..2: Scalars. (10. one can express [(L−1 )T ]µ ν = (L−1 )ν µ and. We do not fact.79) This is the inverse Lorentz transformation consistent with (10. lower.12) a gµν b µ ν = Lµ ρ gµν Lν σ aρ bσ = aρ gρσ bσ (10. In any case one can express (10. one can employ the tensors g µν to raise and lower indices of L ν establish here the consistency of the ensuing notation.17) can be written (L−1 )ν µ = Lµ ν . −a1 . Covariant 4-vectors transform like a µ = gµν Lν ρ g ρσ aσ where we defined g µν = gµν . The respective vectors aµ are associated with the 4-vectors aµ .66) together with (10. −a2 . This property follows from (10. g µν ∂ν = Lµ ρ g ρσ ∂σ . We have duplicated the expression for the inverse of Lµ ν to obtain the correct notation in terms of covariant. (10. If aµ and bµ are 4-vectors aµ gµν bν (10. aµ (10.74) g µν aν . and . using (10. Multiplication (and summation) of x µ = Lµ ν xν by Lρ σ gρµ yields.12). gσν xν = Lρ σ gρµ x µ and g µσ gσν = δ µ ν .17) Lµ σ is the transformation inverse to Lσ µ . The 4-Vector ∂µ (10. 4–Vectors and Tensors Scalar Product then 295 4-vectors allow one to construct scalar quantities. We start from x µ = Lµ ν xν . i.10.72) where aν is a vector with transformation behaviour as stated in (10.78) To prove the 4-vector property of ∂µ we will show that g µν ∂ν transforms like a contravariant 4vector..73) aµ = Lµ σ aσ .

(10. upper.87) The remaining components of pµ can be written.80) Multiplication by g λµ (and summation over µ) together with g λµ gρµ = δ λ ρ yields g λµ ∂µ = Lλ σ g σν ∂ν . For this purpose we consider the scalar differential (dτ )2 = dxµ dxµ = (dt)2 − (dr)2 It holds dτ dt from which follows 2 (10.89) .e.81) ∂ .− ∂t . dτ 1 − v 2 dt p0 = E = √ m m dt = √ . (10.88) One can express then the momentum vector pµ = √ d m dxµ = m xµ . i.. In fact. i.296 Relativistic Quantum Mechanics contravariant. indices..86) One can write (10.. 2 2 dt 1 − v 1 − v (10. 2 dt dτ 1 − v (10. 1 − v2 1 − v 2 dt (10. d’Alembert Operator We want to construct now a scalar differential operator.66).82) (10.e. Proof that pµ is a 4-vector We will demonstrate now that the momentum 4-vector pµ defined in (10.85) 1 d d = √ . 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. For this purpose we define first the contravariant differential operator ∂ µ = g µν ∂ν = Then the operator 2 ∂µ ∂ µ = ∂t − 2 (10.67) transforms like (10. ∂µ does indeed transform like a covariant vector. p1 = √ m v1 m dx1 = √ .g.84) = 1 − (v)2 (10. e. this operator is equal to the d’Alembert operator which is known to be Lorentz-invariant.79) allows us to determine the connection between ∂µ and ∂µ .83) is a scalar under Lorentz transformations.

in fact. and.3: Relativistic Electrodynamics 297 d The operator m dτ transforms like a scalar. in fact. and had chosen the so-called Coulomb gauge (8.77) which allow one to express (10. If we can show that Aµ defined in (10. Since xµ transforms like a contravariant 4-vector.69) for the electrodynamic potential and the 4-vector derivative (10.e.97) We have proven already that ∂µ is a contravariant 4-vector. namely pµ pµ = pµ gµν pν = E 2 − p 2 Evaluation of the r.s. t) and A(r.s. respectively.96) is a scalar and. 8. pµ on the l. (10. in (10. the so-called Lorentz gauge. yields according to (10. hence. t) = 0 .97) and.89) must be a 4-vector. a contravariant 4-vector then the l.10. The momentum 4-vector allows us to construct a scalar quantity.96) The Lorentz-invariance of this gauge. We like to rewrite the last result E 2 = p 2 + m2 or E = ± p 2 + m2 .69) is. 10. One obtains E = ±m ± p2 2m (p 2 )2 + O 4m3 (p 2 )3 4m5 . This gauge is not Lorentz-invariant and we will adopt here another gauge.96) in the form ∂µ Aµ = 0 . We will demonstrate now the 4-vector property of Aµ .95) 2 p This obviously describes the energy of a free particle with rest energy ±m. namely. (10.3 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. m2 m2 v 2 − = m2 1 − v2 1 − v2 (10.89) alltogether transforms like a contravariant 4-vector. Expansion in should then be rapidly convergent. Lorentz-invariant. in (10. the r. equivalently. ∂t V (r. hence. (10. is a scalar. t).s.h.h.12). Lorentz Gauge In our previous description of the electrodynamic field we had introduced the scalar and vector potential V (r.92) (10. can be demonstrated readily using the 4-vector notation (10. · A = 0. .90) (10. for these potentials. of (10.h.h.s. i.67) E2 − p 2 = or pµ pµ = m2 which.93) (10. t) + · A(r.91) (10.94) 1 m In the non-relativistic limit the rest energy m is the dominant contribution to E.. kinetic energy ± 2m and relativistic corrections.

101) Combining equations (10. n = 1.e. (10.69). Using · ∂t A(r. Since ∂µ ∂ µ transforms like a scalar one can conclude that Aν (xσ ) must transform like a 4-vector. 2. . using (10. using (8. t) together with (10. Equation (10. (10. 3 (10. Since ∂µ transforms like a covariant 4-vector. that F µν can be expressed through the potential Aµ in (10. t) − 2 A(r.100.32).s. t) + · J(r. ∂µ J µ (xµ ) = 0 . (10. The Field Tensor The electric and magnetic fields can be collected into an anti-symmetric 4×4 tensor   0 −Ex −Ey −Ez  E 0 −Bz By   . (8. t) = 4 π J(r. (10. t) − 2 V (r.s. t) ..105) .298 Transformation Properties of J µ and Aµ Relativistic Quantum Mechanics The charge density ρ(r. t) 2 ∂t A(r. The l. must be a scalar. transforms like a 4-vector.9). t) . thereby..e.68). yields ∂µ ∂ µ Aν (xσ ) = 4 π J ν (xσ ) .13). it follows that J µ .99) must be a scalar. t) = 0 (10. t) from (8.102) In this equation the r.18) and.103) B . 10. t).97) and.h.s. also the l. using (10. according to (10. t) = −∂t V (r.100) Similarly. of (10.98) which reflects the principle of charge conservation.77) and (10. · A(r.83) and (10. We want to derive now the differential equations which determine the 4-potential Aµ in the Lorentz gauge (10. k. The respective equation for A0 = V can be obtained from Eq. i. a 4-vector.96). t) = −∂t V (r. F µν =  x  Ey B z 0 −Bx  Ez −By Bx 0 Alternatively.h.6) and (8.82) as follows F µν = ∂ µ Aν − ∂ ν Aµ . one obtains for A(r.96). One can readily verify. F mn = − mn (10. t) = ∂t · A(r.98) can be written. (10. has to transform like a contravariant 4-vector.17) using the identity (8. one obtains 2 ∂t V (r.69) and ∂ µ in (10. Consequently. prove that Aµ is.101).99) Since this equation must be true in any frame of reference the right hand side must vanish in all frames.104) where mn = mn is the totally anti-symmetric three-dimensional tensor defined in (5. this can be stated F k0 = −F 0k = E k . must transform likewise. · A(r. i. m. in fact. in fact. t) = 4πρ(r.h. t) and current density J(r. This principle should hold in any frame of reference. t) are known to obey the continuity property ∂t ρ(r.

105) and (10.113) which can be shown to be equivalent to the two homogeneous Maxwell equations (8. The results can be put into the more general form E B = = E . Noting ∂µ F µν = ∂µ ∂ µ Aν − ∂µ ∂ ν Aµ = ∂µ ∂ µ Aν − ∂ ν ∂µ Aµ = ∂µ ∂ µ Aν .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. E⊥ = B⊥ E⊥ + v × B √ 1 − v2 B⊥ − v × E = √ 1 − v2 (10.109) (10. Maxwell Equations in Lorentz-Invariant Form One can express the Maxwell equations in terms of the tensor F µν in Lorentz-invariant form. where we used (10. equivalently.102) ∂µ F µν = 4π J ν .2). . t) and B(r.103.112) (10. by (10. In a new frame of reference holds F µν = Lµ α Lν β F αβ (10.1. From the definition (10. 10. B⊥ are.108) yields then the expressions for the transformed fields E and B .104) establishe the transformation behaviour of E(r.10.106) In case that the Lorentz transformation Lµ ν is given by (10. 8. one can conclude from (10.62) or.63). (10.4). the components of the fields parallel and perpendicular to the velocity v which determines the Lorentz transformation. B .3.110) where E .3: Relativistic Electrodynamics 299 The relationships (10.105) of the tensor F µν one can conclude the property ∂ σ F µν + ∂ µ F νσ + ∂ ν F σµ = 0 (10.97). These equations show that under Lorentz transformations electric and magnetic fields convert into one another. t). respectively. B and E⊥ . 8.111) One can readily prove that this equation is equivalent to the two inhomogeneous Maxwell equations (8. one obtains  Ey −v B Ez +v B  0 −Ex − √ 1 2z − √ 1 2y 1−v1 1−v1    Bz −v1 Ey By +v1 Ez   √ 2  Ex 0 − √ 2   1−v1 1−v1  F µν =  E −v B (10.

. equivalent to (8. for µ = 1.114).116) (10. It holds for a particle with 4-momentum pµ as given by (10. 1 dE 2 1 dp 2 = = qp·E 2 dt 2 dt (10.86). We have.. The term on the r.103) dpx q = ( EEx + py Bz − pz By ) .115) This equation follows. The µ = 0 component of (10. i.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. We want to express this force through the tensor F µν .119) which is the x-component of the equation of motion (8.114) is.93).5) where p = mv/ 1 − v 2 .h.5) taking the scalar product with p dp = qp · E (10.117) (10. Equation (10. employing (10.120) (10. dτ m √ Employing again (10. e. in fact. We want to √ demonstrate now that this equation is equivalent to the equation of motion (8.118) p· dt √ where we exploited the fact that according to p = mv/ 1 − v 2 holds p v.87)] and retain the definition E for the electric field.114) reads using (10. To avoid √ confusion we will employ in the following for the energy of the particle the notation E = m/ 1 − v 2 [see (10. hence.5).5).g. however.120) is referred to as the Lorentz force.s. hence. For the spatial components.67). provides an alternative description of the action of the Lorentz force. . E = m/ 1 − v 2 . also from the equation of motion (8. of (10. (10.86) dE q = p·E . demonstrated that (10.114) where d/dτ is given by (10. dE q = p·E dτ m or with (10.104). dt E From this one can conclude.e.114) reads then. using (10.67) and charge q dpµ q = pν F µν dτ m (10.86) and (10. yields dpx = q dt Ex + (v × B)x (10.

obey a different transformation law. J3 ) and K = (K1 . w)) = ρ eϑ·J + w·K which we present in the form L(ϑ. (10. For this purpose one replaces xµ in (10. In this section we will investigate the transformation properties of scalar functions ψ(xµ ). def (10.1) is a generalization (rotation + boosts) of the group SO(3). 10. 10.. in the present case we exclude the factor ‘−i’ [cf.4: Function Space Representation of Lorentz Group 301 10. we seek an expression which corresponds to (10.51) for the natural representation of the Lorentz group. i.125)]. K2 .e.. w) = ρ(Lµ ν (ϑ.4 Function Space Representation of Lorentz Group In the following it will be required to decribe the transformation of wave functions under Lorentz transformations.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 ) . i.126) ρ ewk Kk − 1 1 . (5. We need to emphasize that (10.121) covers solely the transformation behaviour of scalar functions. xµ ).e.g. ψ ∈ C∞ (4). K3 ) are the generators of L(ϑ. and which can be constructed following the procedure adopted for the function space representation of SO(3). The resulting expression should be a generalization of the function space representation (5. the charge-current density in case of Sect.123) This definition corresponds closely to the function space representation (5. (10.124) In this expression J = (J1 . taken at the pairs of points (x µ = Lµ ν xν . Accordingly.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. w. w) = exp ϑ · J + w · K . one can evaluate Jk as follows Jk = lim and Kk Kk = lim wk →0 ϑk →0 1 ϑ1 1 w1 ρ eϑk Jk −1 1 (10.121) 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 .121) by (L−1 )µ ν xν and obtains ψ (xµ ) = ψ((L−1 )µ ν xν ) .. We like to express now ψ (x µ ) in terms of the old coordinates xµ . w) which correspond to the generators Jk and Kk in (10.127) .3 or the bi-spinor wave function of electron-positron pairs in Sect.48) and (10. J2 .10. (10. e. For such functions holds in the transformed frame ψ (Lµ ν xν ) = ψ(xµ ) (10. However.42) of SO(3). We will denote the intended representation by L(ϑ.125) def (10.7.48) of SO(3). in as far as SO(3. Functions which represent 4-vectorsor other non-scalar entities.

J2 ] = J3 . K2 ] = −J3 .134) . The generators J . J ] [ Kk . K ] = = k m Jm k m Jm = − k m Km .133) We demonstrate this for three cases. and [J1 . J2 ] = [x3 ∂2 − x2 ∂3 .130) ew1 K1 −1 = e−w1 K1 (10. reproduces the expression for J1 in (10.126) for J1 we consider first −1 Relativistic Quantum Mechanics K1 = x0 ∂1 + x1 ∂0 K2 = x0 ∂2 + x2 ∂0 K3 = x0 ∂3 + x3 ∂0 (10. which we like to demonstrate for J1 and K1 . J3 = x2 ∂1 − x1 ∂2 . One can determine similarly K1 starting from coshw1 sinhw1  sinhw1 coshw1 =   0 0 0 0  0 0 1 0  0 0   . (10. sinhw1 x0 + coshw1 x1 .129) which yields for small ϑ1 ρ eϑ1 J1 ψ(xµ ) = ψ(x0 . J2 = x1 ∂3 − x3 ∂1 . namely [J1 . K obey the same Lie algebra (10. x3 ∂1 ] = −x1 ∂2 + x2 ∂1 = J3 .132) and. [ Jk .126). x1 ∂3 ] − [x2 ∂3 . K2 ] = K3 : [ J1 . −sinϑ1 x2 + cosϑ1 x3 ) = ψ(xµ ) + ϑ1 (x3 ∂2 − x2 ∂3 ) ψ(xµ ) + O(ϑ2 ) .302 One obtains J1 = x3 ∂2 − x2 ∂3 .131) This yields for small w1 ρ ew1 K1 ψ(xµ ) = ψ(coshw1 x0 + sinhw1 x1 . (10.128). the expression for K1 in (10.e. x1 . x3 ) 2 = ψ(xµ ) + w1 (x1 ∂0 + x0 ∂1 ) ψ(xµ ) + O(w1 ) (10. 1 This result. obviously. In order to evaluate (10. i. cosϑ1 x2 + sinϑ1 x3 . 0  1 (10. obviously. x1 ∂3 − x3 ∂1 ] = [x3 ∂2 .49) as the generators of the natural representation.128) eϑ1 J1 = e−ϑ1 J1 1  0 =   0 0   0 0 0 1 0 0   0 cosϑ1 sinϑ1  0 −sinϑ1 cosϑ1 (10. K ] [ Jk . x2 . [K1 .

[ J1 .125) in the case of a one-dimensional transformation of the type L(w3 ) = exp w3 K3 . Ω which are connected with x0 . x2 ∂0 ] − [x1 ∂0 . We note tanhΩ = ∂R x0 = . ∂3 in terms of ∂R . K2 ] = [x0 ∂1 + x1 ∂0 .136) (10. cothΩ = 3 . 303 (10. Eqs.4: Function Space Representation of Lorentz Group [ K1 . can be simplified considerably.10.139) x3 = R sinhΩ (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. ∂Ω . 10.135) (10.e.139.85-5. The relationships (10. x0 ∂2 ] = x3 ∂0 + x0 ∂3 = K3 .128). x2 ∂0 ] − [x2 ∂3 . In both cases the radial coordinate is the quantity conserved under the transformations.143) . K2 ] = [x3 ∂2 − x2 ∂3 . ∂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 . One-Dimensional Function Space Representation The exponential operator (10.142) (10. x0 ∂2 + x2 ∂0 ] = [x0 ∂1 .141) (10.140) allow one to ∂ ∂ express the derivatives ∂0 . x3 as follows x0 = R coshΩ ..87)]. x2 + x2 + x2 1 2 3 in the case of SO(3) and (x0 )2 − (x3 )2 in case of transformation (10.137) where K3 is given in (10.138) x3 x0 . i. x0 ∂2 + x2 ∂0 ] = [x3 ∂2 . In the following we consider solely the case x0 ≥ 0. 3 ∂R 3 ∂Ω ∂x ∂x R ∂R x ∂Ω = ∂Ω ∂tanhΩ 1 = cosh2 Ω 0 3 ∂tanhΩ ∂x x ∂Ω ∂cothΩ 1 = = − sinh2 Ω 3 .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. 0 ∂cothΩ ∂x x = ∂R x0 = − ∂x3 R (10. (5. For this purpose one expresses K3 in terms of hyperbolic coordinates R.137). (10. x0 ∂2 ] = −x2 ∂1 + x1 ∂2 = − J3 .

ˆ pµ =⇒ pµ = i(∂t . The historic route to these two equations. Further below we will provide a more systematic.146) Applying the correspondence principle to (10.147) (10. which allows one to replace classical observables by quantum mechanical operators acting on wave functions.92) one obtains the wave equation − or 2 2 µ ∂ ∂µ ψ(xν ) = m2 ψ(xν ) ψ(xν ) = 0 . representation theoretic treatment. is not very insightful. complex-valued function. (10. (10. − ) = i ∂ µ . In these units the quantities energy. We will provide a ‘derivation’ of these two equations which stem from the historical development of relativistic quantum mechanics. The latter property implies that upon change of reference frame ψ(xµ ) transforms according to (10. Free Particle Case A quantum mechanical description of a relativistic free particle results from applying the correspondence principle.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 ) .128) yields K3 = x0 ∂3 + x3 ∂0 = ∂ . 10.144) The action of the exponential operator (10.122).150) . ˆ (10. in 4-vector notation reads pµ =⇒ pµ = i (∂t . and (time)−1 all have the same dimension. mass. extremely useful.145) ∂Ω 10.304 Inserting these results into the definition of K3 in (10.151) is called the Klein-Gordon equation. but certainly is short and. In natural units the Klein–Gordon equation (10. (length)−1 . however. namely the Klein–Gordon and the Dirac equations. In the position representation the correspondence principle states ˆ E =⇒ E = − ∂t i ˆ p =⇒ p = i which. momentum.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. ∂Ω Relativistic Quantum Mechanics (10. ) = i ∂µ .151) reads ∂µ ∂ µ + m2 ψ(xµ ) = 0 (10.149) where ψ(xµ ) is a scalar.121. In the following we will employ so-called natural units = c = 1. therefore. The partial differential equation (10.148) ∂ µ ∂µ + m2 (10.

and that (as postulated) ψ(xµ ) is a scalar. t) = 0 i ( ψ ∗ (r. t) ) 2m 1 = ( ψ ∗ (r. t) ψ ∗ (r.161) . t) − ψ(r. Under Lorentz transformations the free particle Klein–Gordon equation (10. t) ψ(r.10.154). t) = ψ ∗ (r. t) jKG (r. t) ] 2mi (10. (10.157) 2m which is equivalent to (10. t) (10. t) ψ ∗ (r.152) which has the same form as the Klein–Gordon equation in the original frame.151) One can notice immeadiately that (10. t) + where ρKG (r. t) = − 1 2m 2 ψ(r.159) (10. t) ψ(r. t)∂t ψ ∗ (r.158) which yields 2 2 ψ ∗ ∂t ψ − ψ∂t ψ ∗ − ψ ∗ 2ψ +ψ ψ∗ − 2ψ∗ = ψ) = 0 (10. t) = · jKG (r.153) is associated with a conservation law for particle probability ∂t ρS (r. t) = (10.150) is invariant under Lorentz transformations. To 1 derive this conservation law one rewrites the Schr¨dinger equation in the form (i∂t − 2m 2 )ψ = 0 o and considers 1 2 Im ψ ∗ i∂t − ψ = 0 (10. t) − ψ(r.155) describes the positive definite probability to detect a particle at position r at time t and where jS (r. t) = 0 (10. t)∂t ψ(r. Current 4-Vector Associated with the Klein-Gordon Equation As is well-known the Schr¨dinger equation of a free particle o i∂t ψ(r. This follows from the fact that ∂µ ∂ µ and m2 are scalars. t) ) .150) one considers Im ψ ∗ ∂µ ∂ µ + m2 ψ = 0 (10. t) 1 [ ψ ∗ (r.150) becomes ∂µ ∂ µ + m2 ψ (x µ ) = 0 (10. t) ψ(r. t) + where ρS (r.156) · jS (r. 2mi (10. In order to obtain the conservation law connected with the Klein–Gordon equation (10. t) − ψ(r.154) describes the current density connected with motion of the particle probability distribution.160) ∂t (ψ ∗ ∂t ψ which corresponds to − ψ∂t ψ∗) + · (ψ ψ∗ ∂t ρKG (r.5: Klein–Gordon Equation or 2 ∂t − 2 305 + m2 ψ(xµ ) = 0 .

t).166) ψo (p.168) . λ) associated with the corresponding wave functions ψo (p.154).e.164) The corresponding energy is Eo (po . Generating a Solution Through Lorentz Transformation A particle at rest.161) and (10.151) yields 2 2 Eo − p0 − m2 ψ(r. t) = 0 (10.163) which results in the expected [see (10. corresponding to λ = +.150) is a continuum of positive energies E ≥ m. is actually that of a charge density. not of particle probability.93] dispersion relationship connecting E0 . it had been realized later.p ei(p·r − λEo (p)t) Eo (p) = The spectrum of the Klein–Gordon equation (10.. i. with p = 0.167) m which is positive for λ = + and negative for λ = −. (10. λ|xµ ) (10. corresponding to λ = −. the Klein-Gordon equation is considered as a suitable equation to describe spin–0 particles. λ|xµ ) = Nλ. 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. λ) actually describes charge density rather than probability. λ = ± (10. in that the expression for ρKG is not positive definite. λ|xµ ) according to (10.162) Inserting this into the Klein–Gordon equation (10. The density ρKG (p. We denote this by summarizing the solutions as follows ∂µ ∂ µ + m2 ψo (p. and the density ρKG (p. (10. p0 .166) is Eo (p) ∗ ρKG (p.150) are actually determined by po and by the choice of sign ±. m 2 2 E0 = m2 + po . (10. λ|xµ ) = N e−iλmt . the anti-particles carry a charge opposite to that of the associated particles.165) This result together with (10.306 Relativistic Quantum Mechanics This conservation law differs in one important aspect from that of the Schr¨dinger equation (10. λ|xµ ) = 0 2 m2 + po . Today. according to (??) is decribed by the r–independent wave function ψo (p = 0. and of negative energies E ≤ −m. λ|xµ )ψo (p. for example pions. λ) = λ ψo (p. λ = ± . The proper interpretation of the two cases is that the Klein–Gordon equation describes particles as well as anti-particles. ±) = ± 2 m2 + po (10.162) shows that the solutions of the free particle Klein-Gordon equation (10. Solution of the Free Particle Klein–Gordon Equation Solutions of the free particle Klein–Gordon equation are ψ(xµ ) = N e−ipµ x µ = N ei(p0 ·r − Eo t) . The proper interpretation of ρKG (r. o namely.

176) 10.175) as the momentum of the particle and one obtains L(w3 )ψo (p = 0. In case of λ = −.. 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.s. Ω. Such wave function should be generated by applying the Lorentz transformation in the function space representation p (10. In case of λ = + one obtains finally L(w3 )ψo (p = 0. for negative energy solutions. using (10. t).169) The addition theorem of hyperbolic functions cosh(Ω + w3 ) = coshΩ coshw3 + sinhΩ sinhw3 allows us to rewrite the exponent on the r. −A(r. t)) (10.170) The coordinate transformation (10. i. t).h. for the wave function (10. Aµ (xµ ) = (V (r.166). in fact. λ = −|xµ ) = N ei(px 3 + Eo (p)x0 . (10.61) yield for this expression −iλ √ m mv x0 − iλ √ x3 .171) One can interpret then for λ = +.138) and the relationships (10.145) choosing m = sinhw3 .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 (10.168). A(r. one has to interprete p = mv/ 1 − v 2 (10. (10.172) as the energy [c.177) . This yields. (10.f.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. for positive energy solutions. λ = +|xµ ) = N ei(px 3 − Eo (p)x0 (10. i.174) which agrees with the expression given in (10.173) (10. 2 2 1−v 1−v (10. (10.169) −iλ ( m coshw3 ) ( R coshΩ ) − iλ ( m sinhw3 ) ( R sinhΩ ) .e. t)) . λ|xµ ) = = exp w3 ∂ ∂Ω N e−iλmRcoshΩ 3) N e−iλmRcosh(Ω+w . For this purpose we consider the wave function for a particle moving with momentum velocity v in the direction of the x3 –axis.10. of (10..138) to replace t = x0 by hyperbolic coordinates R.e.168). L(w3 )ψo (p = 0.166)] of the particle.

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 (10.150).308 free classical particle classical particle in field (V. (10.181) Non-Relativistic Limit of Free Particle Klein–Gordon Equation In order to consider further the interpretation of the positive and negative energy solutions of the Klein–Gordon equation one can consider the non-relativistic limit. (10. We will also assume. For this purpose one writes the Klein-Gordon equation (10. A) free quantum particle quantum particle in field (V. t) = e−imt Ψ(r. t). we define ψ(r. according to the rules shown in Table 10. it’s rest energy. in keeping withnthe non-relativistic limit. that the mass m of the particle.179) (10.180) + m2 ψ(r.1. t) 2 (10. A) Relativistic Quantum Mechanics energy momentum 4-vector E p pµ E − qV p − qA pµ − qAµ i∂t ˆ p = −i i∂µ i∂t − qV ˆ ˆ p − qA = π i∂µ − qAµ = πµ Table 10. −A). momenta for classical particles and the respective operators for quantum mechanical particles in the manner shown. is much larger than all other energy terms.69)] one replaces the quantum mechanical operators.e.e. (10. (10. i. t) . which couples a particle of charge q to an electromagnetic field described through the potential Aµ (xν ). See also Eq.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 − . t) = −i − q A(r.. In terms of space-time derivatives this reads (i∂t − qV (r. m2 ) ψ(xµ ) = 0 .. According to the so-called minimum coupling principle the presence of the field is accounted for by altering energy. According to the principle of minimal coupling [see (10.178) According to the replacements in Table 10. and focus on the remaining part of the wave function. i.150) −g µν (i∂µ )(i∂ν ) + m2 ψ(xµ ) = 0 . t) .147)..e.1: Coupling of a particle of charge q to an electromagnetic field described by the 4-vector potential Aµ = (V. albeit in a form.182) and seek an equation for Ψ(r. A) or Aµ = (V. t))2 ψ(r.93). i. i∂t and −i in (10. For this purpose we split-off a factor exp(−imt) which describes the oscillations of the wave function due to the rest energy. in .

(10. Pionic Atoms To apply the Klein–Gordon equation (10.g. t) = ˆ [p − q A(r. e. Ψ |q V | << m . (10.183). however.181) yields (we assume now that the Klein–Gordon equation describes a 2 particle with mass mπ and charge −e) for qV (r. larger than |i∂t Ψ/Ψ| and alrger than qV .181) reads then i ∂t Ψ(r. 309 (10. (10. i.s.184) where we neglected all terms which did not contain factors m. The approximation is justified on the ground of the inequalities (10. | i∂t Ψ | << m . identical to the Schr¨dinger equation (10..185) This is. t) (10.6: Klein–Gordon Equation with Electromagnetic Field particular. 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− .e.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. The process of π − meson capture involves the so-called Auger effect. This application demonstrates that the Klein–Gordon equation describes spin zero particles.181) to a physical system we consider pionic atoms.186) then are slowed down. π − mesons are generated through inelastic proton–proton scattering p + p −→ p + p + π − + π + . atoms in which one or more electrons are replaced by π − mesons. A stationary state of the Klein–Gordon equation is described by a wave function ψ(xµ ) = ϕ(r ) e−i t .188) Inserting this into (10. spin-0 mesons.h. t) 2m Ψ(r. i. filtered out of the beam and finally fall as slow pions onto elements for which a pionic variant is to be studied.10. t)]2 + qV (r.181) can then be approximated as follows: (i∂t − qV )2 e−imt Ψ = = (i∂t − qV ) (me−imt Ψ + e−imt i∂t Ψ − qV e−imt Ψ) m2 e−imt Ψ + me−imt i∂t Ψ − qV e−imt Ψ +me−imt i∂t Ψ − e−imt ∂ 2 Ψ − qV e−imt i∂t Ψ −me−imt qV Ψ − e−imt i∂t qV Ψ + q 2 V 2 e−imt Ψ ≈ m2 e−imt Ψ − 2mqV e−imt Ψ − 2me−imt i∂t Ψ (10.2) of a non-relativistic spin-0 particle o moving in an electromagnetic field. To ‘manufacture’ pionic atoms. (10. t) = − Ze and A(r... of (10.183) The term on the l. t) ≡ 0 r + Ze2 r 2 + 2 − m2 π ϕ(r ) = 0 .e.189) . The Klein-Gordon equation (10.

195) The latter problem leads to the well-known spectrum En = − mπ (Ze2 )2 . (10..199) ( + 1) − Z 2 e4 .194) Bound state solutions can be obtained readily noticing that this equation is essentially identical to 2 that posed by the Coulomb problem (potential − Ze ) for the Schr¨dinger equation o r d2 − dr2 ( + 1) 2mπ Ze2 + + 2mπ E r2 r R (r) = 0 (10. φ) R (r) Y r m (θ.198) . i.310 Relativistic Quantum Mechanics Because of the radial symmetry of the Coulomb potential we express this equation in terms of spherical coordinates r.192) ˆ are spherical harmonics. this quantity definitely must be an integer. θ. .196) In this expression the number n defined through n = n− −1 (10. φ) = ( + 1) Y m (θ. n = 1. φ) . .192) leads then to the ordinary differential equation d2 − dr2 ( + 1) − Z 2 e4 2 Ze2 + + r2 r 2 − m2 π R (r) = 0 . 1.. (10.191) ˆ The operator L2 in this equation suggests to choose a solution of the type ϕ(r ) = where the functions Y in (10. i. φ.e.191) m (θ. the eigenfunctions of the operator L2 ˆ L2 Y m (θ. The Laplacian is 2 = ˆ 1 2 1 1 1 2 L2 2 ∂r r + 2 2 ∂θ sinθ∂θ + 2 2 ∂φ = ∂r r − 2 . . φ) (10. 2. . r r r r sin θ r sin θ Ze2 2 r ) (10. n − 1 .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 .194) and (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 (10.193) (10. 2n2 = 0. . . . The similarity of (10. (10.e.197) counts the number of nodes of the wave function. (10. (10.

Introducing α = Z 2 e4 and 1 β = + 2 (10. . n − 1 m = − .204) reads 1 (10.6: Klein–Gordon Equation with Electromagnetic Field and one can write (10. i.196) if one makes the replacement 2 values which can be obtained E −→ One obtains − m2 π . (10. (10. 2mπ 2 (n + λ( ) + 1)2 (10. 1. . results in the (10.10. . .197. 1. . . . . 2. m) = mπ 1 + Z 2 e4 (n− − 1 2 + ( + 1 )2 − Z 2 e4 )2 2 n = 1. 10. − + 1.201) mπ Z 2 e4 m2 − m2 π π = − . . + (10. n = 0.204) 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 ).. .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. . . . . .200) The bound state solutions of this equation should correspond to from (10.194) d2 λ( ) ( λ( ) + 1 ) 2 Ze2 − + + dr2 r2 r 2 311 − m2 π R (r) = 0 . . .203) spectrum . which corresponds to a bound = 0.202) Solving this for state) yields (choosing the root which renders = 1 + mπ Z 2 e4 ( n + λ( )+1 )2 ≤ mπ . EKG = Using (10. . 1. 2mπ 2 −→ λ( ) .199) and defining EKG (n. 2 e2 −→ e2 mπ . .e.

Application of the correspondence principle (10. the Klein–Gordon equation had been rejected since it did not yield a positive-definite probability density. in turn. 4n (10. i. arises because the Klein–Gordon equation. The latter spectrum shows. The latter term agrees with observations of pionic atoms. It should feature then a differential operator of the type D = iγ µ ∂µ .208) These two equation can be combined i∂t + m2 − 2 i∂t − m2 − 2 Ψ(r.. it does not agree with observations of the hydrogen spectrum. 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. i. t) . the Dirac equation introduced below. t) = ± m2 − 2 Ψ(r.207) Here the first term represents the rest energy. a feature which is connected with the 2nd order time derivative in this equation. His solution . one with a positive-definite density.e. the second term the non-relativistic energy. for example. In order to describe electron spectra one must employ the Lorentz-invariant 1 wave equation for spin. 2n2 2βn3 8n4 4n4 (10. i.e. Heuristic Derivation Starting from the Klein-Gordon Equation An obvious starting point for a Lorentz-invariant wave equation with only a first order time derivative is E = ± m2 + p 2 . m) ≈ m − mZ 2 e4 mZ 4 e8 − 2n2 2n3 1 + − 1− Relativistic Quantum Mechanics α α2 α2 α2 − + + + O(α3 ) . is related to the equation E 2 = m2 + p 2 of the classical theory which involves a term E 2 . in fact. Finally. = 1 states into groups of two and four degenerate states.204) EKG (n. but for a long time to no avail.146) leads to the wave equation i∂t Ψ(r.209)..209).208. but in the present case rather the negative of the energy.206) 1 2 3 + O(Z 6 e12 ) .209) which.7 The Dirac Equation Historically. through the correspondence principle. but not an operator iγ · as suggested by the principle of relativity (equivalence of space and time). t) (10. .2 particles. Equations (10. 10. and the third term gives the leading relativistic correction. Dirac succeeded. however. the π − meson is a pseudoscalar particle. This derivative.. A more satisfactory Lorentz–invariant wave equation. Many attempts were made by theoretical physicists to ‘linearize’ the square root operator in (10. is identical to the two equations (10. It must be pointed out here that does not denote energy. the wave function changes sign under reflection. however. 10. (10. However. would have only a first order time derivative.208). are unsatisfactory since expansion of the square root operator involves all powers of the Laplace operator.e.208. 10. a splitting of the six n = 2.312 ≈ From this results for (10. Also.

Properties of the Dirac Matrices Let us now trace Dirac’s steps in achieving the linearization of the ‘square root operator’ in (10. the 4-dimensionsional representation discovered by Dirac involved new physical properties. Starting point is to boldly factorize. these descriptions ultimately merging with the true theory of matter. We seek to identify the coefficients γ µ . are associated with a positive-definite particle density.66). 3 . beautiful mathematical theory and that. γ 1 . strengthens the believe of many theoretical physicists today that the properties of physical matter ultimately derive from a. Instead. yet to be discovered. However. (10. (10.213) We want to determine now the simplest algebraic realization of γ µ . (10. so far. spin. one route to important discoveries in physics is the creation of new mathematical descriptions of nature. the quantities γ 0 . γ j .209). but did not commute the. j = 1.211) into (10. one can impose the condition γ0 is hermitian . ψ2 (xµ ). 2. In fact. achieved through a beautiful mathematical theory.211) Obviously. Initially. The discovery by Dirac. quantities with the transformation law (10. therefore. the operator of the Klein–Gordon equation ∂µ ∂ µ + m2 = −( P + m ) ( P − m ) (10.210) where P = iγ µ ∂µ . namely. Inserting (10. the realization that the linearization can be carried out only if one assumes a 4-dimensional representation of the coefficients γ µ . i.215) . . 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 ).10. therefore. . Accordingly. 2. γ ν ]+ = 2 g µν (10.e.214) From this follows that γ 0 has real eigenvalues ±1 and γ j .. γ 2 . 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 . exploited ∂µ ∂ν = ∂ν ∂µ . this would lead to the two wave equations (P − m)Ψ = 0 and (P + m)Ψ = 0 which have a first order time derivative and. according to (10. .7: Dirac Equation 313 to the problem involved an ingenious step. For µ = ν condition (10. it was assumed that the 4-dimensional space introduced by Dirac could be linked to 4vectors.1 2 and anti-particles. unspecified algebraic objects γ µ and γ ν .212) where we have changed ‘dummy’ summation indices. this was not so.208). ψd (xµ ). 3 has imaginary eigenvalues ±i.213) reads (γ µ )2 = 1 µ = 0 −1 µ = 1. It turns out that no 4-vector of real or complex coefficients can satisfy these conditions. 3 are anti-hermitian . j = 1. 2.210) yields −g µν ∂µ ∂ν − m2 = (iγ µ ∂µ + m)(iγ µ ∂µ − m) 1 = −γ µ γ ν ∂µ ∂ν − m2 = − ( γ µ γ ν ∂µ ∂ν + γ ν γ µ ∂ν ∂µ ) − m2 2 1 µ ν ν µ = − ( γ γ + γ γ ) ∂µ ∂ν − m2 2 (10.

(10.314 For µ = ν (10. A proper choice is γ0 = 1 1 0 0 −1 1 . 1 (10.216) i. to construct four matrices γ µ for the next possible dimension d = 4.222) ∗ ∗ ∗ ∗ where we have defined Ψ† = (ψ1 . The Dirac Equation Altogether we have shown that the Klein–Gordon equation can be factorized formally ( iγ µ ∂µ + m ) ( iγ µ ∂µ − m ) Ψ(xµ ) = 0 (10. σ j σ k = − σ k σ j for j = k . rather than to the right side. in fact. For d = 2 there exist only three anti-commuting matrices. holds 2 σj = 1 . The Adjoint Dirac Equation The adjoint equation is Ψ† (xµ ) i(γ µ )† ∂µ + m ← ← (10. 2. however.224) .213) is satisfied. as long as det(γ µ ) = 0 the dimension d of the square matrices γ µ must be even so that (−1)d = 1. ψ4 ) and where ∂µ denotes the differential operator ∂ µ operating to the left side. ψ3 .218) The Pauli matrices allow one. the γ µ are anti-commuting.215). namely the Pauli matrices σ 1 .213) reads γ µ γ ν = −γ ν γ µ . γj = 0 −σ j σj 0 . (10. (10.220) where Ψ(xµ ) represents a 4-dimensional wave function.e. One can readily show using the hermitian property of the Pauli matrices (γ 0 )† = γ 0 and (γ j )† = −γ j for j = 1. ψ2 . σ 2 .218) of the Pauli matrices one can readily prove that condition (10.223) Inserting this into (10. rather than a scalar wave function. except for similarity trasnformations. Relativistic Quantum Mechanics (10.219) Using property (10. in fact. is unique. σ 3 for which. 3 which.221) =0 (10. 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. This property can also be written (γ µ )† = γ 0 γ µ γ 0 .222) and multiplication from the right by γ 0 yields the adjoint Dirac equation Ψ† (xµ ) γ 0 iγ µ ∂µ + m ← =0. We will argue further below that the choice f γ µ . From this one can conclude for the determinants of γ µ det(γ µ γ ν ) = det(−γ ν γ µ ) = (−1)d det(γ ν γ µ ) = (−1)d det(γ µ γ ν ) . (10.217) Obviously. is implied by (10..

228) (10. 3 and ˆ β = 1 1 0 0 −1 1 . ˆ ˆ ˆ Ho = α · p + β m . t) = 0 (10. γ ˜ Exercise 10.227) leaves the algebra of the Dirac matrices unaffected and commutation property (10. 2. 2.230) Schr¨dinger Form of the Dirac Equation o Another form in which the Dirac equation is used often results from multiplying (10.231) ˆ whereα has the three components αj . (10.10.222.226) (10. γj = ˜ 0 σj 1 S = √ 2 −σ j 0 1 1 1 1 1 −1 1 1 (10. . 3 .219) is the socalled chiral representation defined through ˜ Ψ(xµ ) = S Ψ(xµ ) .213) still holds. (10. Defining a new representation of the wave function Ψ(xµ ) ˜ Ψ(xµ ) = S Ψ(xµ ) leads to the ‘new’ Dirac equation ˜ ( i˜ µ ∂µ − m ) Ψ(xµ ) = 0 γ where γ µ = S γ µ S −1 ˜ (10.227). and γ0 = ˜ 0 1 1 1 0 1 . j = 1. αj = ˆ 0 σj σj 0 . 10. γ ν ]+ = 2 g µν .221) from the left by γ 0 ˆ ˆ i ∂t + i α · − β m Ψ(r. i. (10.e. j = 1. j = 1.225) .227) A representation often adopted beside the one given by (10. 2. (10.229) The similarity transformation (10. (10..7: Dirac Equation Similarity Transformations of the Dirac Equation .7.232) 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 ) .233) The eigenstates and eigenvalues of H correspond to the stationary states and energies of the particles described by the Dirac equation.Chiral Representation 315 The Dirac equation can be subject to any similarity transformation defined through a non-singular ˜ 4 × 4–matrix S.213). [˜ µ . 3 .1: Derive (refeq:Dirac-intro20a) from (10.

The Clifford algebra C4 entails a subgroup G4 of elements ± dj1 dj2 · · · djs . j1 < j2 < · · · < js s ≤ 4 (10. Γ±5 = ±d4 Γ±6 = ±d1 d2 . (1955). j = 1. Γ±11 = ±d3 d4 Γ±12 = ±d1 d2 d3 . Γ±3 = ±d2 . 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 . d2 = . Γ±13 = ±d1 d2 d4 .237) which are the ordered products of the operators ±1 and d1 .213). in case of C4 . page 187. Obviously. The associative algebra generated by d1 .Mod. satisfy the anti-commutation property dj dk + dk dj = 2 for j = k 0 for j = k Relativistic Quantum Mechanics d4 = γ 0 (10. i. One can conclude then that also any set of 4 × 4–matrices obeying (10. form a Clifford algebra C3 . d2 . This property extends then to 4 × 4–matrices which obey (10. . The representations of this group are given by a set of 32 4 × 4–matrices which are equivalent with respect to similarity transformations.6 and R. Since the Γj are hermitian the similarity transformations are actually given in terms of unitary transformations. d4 = 1 0 0 −1 1 0 0 −1 0 1 1 0 0 i −i 0 (10. any product 1 of the dj ’s can be brought to the form (10. 27.. To complete the proof in this section the reader may consult Miller ‘Symmetry Groups and their Application’ Chapter 9.316 Clifford Algebra and Dirac Matrices The matrices defined through dj = iγ j .237) by means of the property (10. Γ±8 = ±d1 d4 .H.235) can differ only with respect to unitary similarity transformations. Γ±9 = ±d2 d3 Γ±10 = ±d2 d4 . For example. i. Rev..e.236) where ‘⊗’ denotes the Kronecker product between matrices.Phys.Good.235) and. . The three Pauli matrices also obey the property (10. d4 . Γ±7 = ±d1 d3 . The representations of Clifford algebras Cm are well established. hence. the matrix elements of C = A ⊗ B are Cjk.235). The reader may . Γ±15 = ±d2 d3 d4 Γ±16 = ±d1 d2 d3 d4 (10. Γ±4 = ±d3 . There are (including the different signs) 32 elements in G4 which we define as follows Γ±1 = ±1 1 Γ±2 = ±d1 .234) (10. 2. d4 is called a Clifford algebra C4 . 3 . d3 . Γ±14 = ±d1 d3 d4 . m = Aj Bkm .238) These elements form a group since obviously any product of two Γr ’s can be expressed in terms of a third Γr .235) as can be readily verified from (10.213).e. to Dirac matrices.

the form of the Dirac equation is identical in equivalent frames of reference. 10. therefore.e. Exercise 7. 3 has imaginary eigenvalues ±i and can be represented by an anti-hermitian matrix. Exercise 7.218) of the Dirac matrices γ µ .214) follows that γ 0 has real eigenvalues ±1 and can be represented by a hermitian matrix. The latter transformations imply that coordinates transform according to (10. these properties determine the matrices γ µ uniquely. transform coordinates and derivatives of the Dirac equation. Infinitesimal Bispinor State Transformation We want to show now that a suitable transformation of Ψ(xµ ) does. a result one could have also obtained by applying the chain rule to (10. namely.2: Demonstrate the anti-commutation relationships (10. Hence.4: Show that from (10.240) .3: Demonstrate the anti-commutation relationships (10. in frames connected by Lorentz transformations.e.80) by Lµ ρ and using (10.6). i.219). µ µ (10. Exercise 7. i. Lorentz Transformation of the Bispinor State Actually. and derivatives according to (10.241) . The transformation is assumed to be linear and of the form Ψ (x ) = S(Lµ ν ) Ψ(xµ ) µ µ (10. i. We can.8 Lorentz Invariance of the Dirac Equation We want to show now that the Dirac equation is invariant under Lorentz transformations. (10.236) with the Dirac representation (10. Except for a similarity transformation.215). in a moving frame holds iγ µ ∂µ − m Ψ (x ) = 0 . in fact.213. x = Lµ ν xν . x µ = Lµ ν xν (10.80)...10.e.. i. exist. we do not know yet how to transform the 4-dimensional wave function Ψ and the Dirac matrices γ µ . 2. i..e. Multiplication and summation of (10. 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.239) Form invariance implies that the matrices γ µ should have the same properties as γ µ . it must hold γ µ = γ µ . However. and γ j . j = 1.218) of the Pauli matrices σ j .76) yields ∂ρ = Lν ρ ∂ ν .. 10.e. in a moving frame should hold iγ ∂µ − m Ψ (x ) = 0 .6).8: Lorentz Invariance of the Dirac Equation 317 also want to establish the unitary transformation which connects the Dirac matrices in the form (10.

hence. i. it must hold (10. in the expression of the infinitesimal transformation ρ Lµν = g µν + µν (10.e.249) from which we can conclude σµν = −σµν νµ = −σνµ i.e. S( µν ) should not change its value if one replaces in its argument µν by − νµ .247) can be expanded S( µν i ) = 1 − σµν 1 4 µν (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.243) in a form in which the Lorentz transformation in the form Lµν is on the r.243) We want to determine now the 4 × 4–matrix S(Lη ξ ) which satisfies this condition. of the equation. Ob1 viously.221) upon Lorentz transformation yields iS(Lη ξ )γ µ S −1 (Lη ξ ) Lν µ ∂ν − m Ψ (x ) = 0 . (10.244) from which.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.. We had introduced in (10.248) Here σµν denote 4 × 4–matrices operating in the 4-dimensional space of the wave functions Ψ. One can finally conclude multiplying both sides by g ρµ S(Lη ξ )γ µ S −1 (Lη ξ ) = Lν µ γν .250) σµν = −σνµ . using gρµ γ µ = γρ .243) from the left by Lσ ρ gσν yields S(Lη ξ )γ µ S −1 (Lη ξ ) gρµ = Lσ ρ gσν γσ . hence. follows S(Lη ξ )γρ S −1 (Lη ξ ) = Lσ ρ γσ . (10.247) the infinitesimal matrix µν is anti-symmetric. One can. For this purpose we exploit (10. Multiplication of this property by g from the right yields ( g)T = − g. µ (10.h. The proper starting point for a constructiuon of S(Lη ξ ) is actually (10.318 Relativistic Quantum Mechanics where S(Lµ ν ) is a non-singular 4 × 4–matrix. νµ (10. It holds then S( µν i ) = 1 − σµν 1 4 µν µν = S(− νµ i ) = 1 + σµν 1 4 µν . g is an anti-symmetric matrix.s.12) in the form Lν µ gνσ Lσ ρ = gµρ = gρµ . however. (10. µ g ρν = µν and. Multiplication of (10. state that the Dirac equation (10.245) The construction of S(Lη ξ ) will proceed using the avenue of infinitesimal transformations.. .38) the infinitesimal Lorentz transformations in the form Lµ ν = δ µ ν + µ ν where the infinitesimal operator µ ν obeyed T = − g g. The elements of g are.241) implies a similarity transformation Sγ µ S −1 .246) (10. the transformation (10.

252) also as a sum over both indices of αβ .255) satisfy condition (10.251) S −1 ( µν ) = 1 + σµν µν 1 4 Inserting (10. 2. γj = ˜ 0 −σ j σj 0 .248. 10. ˜ ˜ σ only six generators need to be determined) σ0j = ˜ i [˜0 .h.s..255) which can be demonstrated using the properties (10. (10. (10.252) Since six of the coefficients αβ can be chosen independently.256) One can readily verify that the non-vanishing generators σµν are given by (note σµν = −˜νµ . β and operate in the same space as the Dirac matrices. γj ] = γ ˜ 2 −iσ j 0 0 iσ j .228. (10.5: Show that the σαβ defined through (10.257) .216) of the Dirac matrices. this condition can actually be expressed through six independent conditions. 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.h.253) 1 2 αβ ( δ µ β γα − δ µ α γβ ) .h.229).. γ µ ]− = 2i ( δ µ β γα − δ µ α γβ ) . For this purpose we need to verify that the algebra of the generators involving addition and multiplication is closed.252) results in the condition for each α. γβ ]− 2 (10. For this purpose one needs to express formally the r. must be symmetric with respect to interchange of the indices α and β.s.254). j = 1. the chiral representation introduced above in Eqs. a solution of condition (10. β [ σαβ .h.213. (10. i. the expression on the r. like the expression on the l. Furthermore. namely. 3 .e. In fact. Algebra of Generators of Bispinor Transformation We want to construct the bispinor Lorentz transformation by exponentiating the generators σµν .8: Lorentz Invariance of the Dirac Equation 319 One can readily show expanding SS −1 = 1 to first order in µν that for the inverse infinitesimal 1 transformation holds i (10.251) into (10. In this representation the Dirac matrices γµ = (˜ 0 . γk ] = ˜ γ ˜ jk σ 0 0 σ . Comparing this with the l. Exercise 7. of (10.254) The proper σαβ must be anti-symmetric in the indices α. −γ ) are ˜ γ ˜ γ0 = ˜ 0 1 1 1 0 1 .s. 10. of (10.s. 10.246) results then in a condition for the generators σµν − i ( σαβ γ µ − γ µ σαβ ) 4 αβ = νµ γν . (10.254) is σαβ = i [ γα . σjk = [˜j . For this purpose we inspect the properties of the generators in a particular representation.10.

.241) bispinor wave functions from one frame of reference into another frame of reference.. 2 0 −(w + iϑ) · σ 13 + σ23 ˜ 23 ) (10. (10.259) We had stated before that the transformation S is actually determined through the Lorentz transformation Lµ ν . the representation (10. In fact. i.51).320 Relativistic Quantum Mechanics Obviously.44)  0 −w1 −w2 −w3  w 0 ϑ3 −ϑ2   =  1  w2 −ϑ3 0 ϑ1  w3 ϑ2 −ϑ1 0  µν (10.255) yields a closed algebra. Finite Bispinor Transformation The closedness of the algebra of the generators σµν defined through (10. one can express the tensor µν through w and ϑ using µν = µ ρ g ρν and the expression (10.260) Inserting this into (10. and since the algebra of the Pauli matrices is closed. 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 = . any representation of the generators. not necessarily infinitesimal. Since such operator does not change its block-diagonal form upon exponentiation the bispinor transformation (10.51) and S.261) We note that this operator is block-diagonal.248) allows us to write the transformation S for any. in particular. therefore.262) This expression allows one to transform according to (10.259) yields the desired connection between the Lorentz transformation (10. therefore. We can finally note that the closedness of the algebra of the generators σµν is not affected by similarity transformations and that. since both operations convert block-diagonal operators A 0 0 B (10. In order to construct an explicit expression of S in terms of w and ϑ we employ again the chiral representation. One should. µν in the exponential form i S = exp − σµν 4 µν . ϑ) = e 2 (w − iϑ)·σ 0 − 1 (w + iϑ)·σ 0 e 2 1 (10.258) again into block-diagonal operators.e. be able to state S in terms of the same parameters w and ϑ as the Lorentz transformation in (10. the algebra of these generators is closed under addition and multiplication.259) becomes in the chiral representation ˜ S(w.

The conservation law (10.269) One can readily show that the same operator is obtained evaluating γ S (w. the property applies then in any representation of S.264) (10. (10.263) does hold. Obviously. This transformation behaviour can also be deduced from the transformation properties of the bispinor wave function Ψ(xµ ).263) obviously is a scalar under Lorentz transformations and that the left hand side must then also transform like a scalar. −ϑ) = e− 2 (w − iϑ)·σ 0 1 (w + iϑ)·σ 0 e2 1 (10.270) We conclude that (10.. For our proof we note first ˜ ˜ S −1 (w. The time-like component ρ of j µ 4 iγ µ ∂µ + iγ µ ∂µ ← Ψ(xµ ) = 0 . We will now determine the relationship between the flux j µ = Ψ (x )γ 0 γ µ Ψ (x ) † µ (10. i.h.221) and the adjoint Dirac equation (10. the conservation law (10.224). for j µ (xµ ) = (ρ. 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 . (10.8: Lorentz Invariance of the Dirac Equation Current 4-Vector Associated with Dirac Equation 321 We like to derive now an expression for the current 4-vector j µ associated with the Dirac equation which satisfies the conservation law ∂µ j µ = 0 .271) .268) We will prove this property in the chiral representation. and addition yields Ψ† (xµ )γ 0 The last result can be written ∂µ Ψ† (xν )γ 0 γ µ Ψ(xν ) = 0 .263) Starting point are the Dirac equation in the form (10.10. For this purpose we prove first the relationship S −1 = γ 0 S † γ 0 . ϑ) = S(−w. j) = Ψ† (xµ )γ 0 γ µ Ψ(xµ ) . of (10.e.263) allows one to conclude that j µ must transform like a contravariant 4-vector as the notation implies.266) ρ(xµ ) = Ψ† (xµ )Ψ(xµ ) = s=1 |ψs (xµ )|2 (10. j µ must transform like a contravariant 4-vector.265) (10. (10.224) from the right by Ψ(xµ ). (10. (10. Since ∂µ transforms like a covariant 4-vector.221) from the left by Ψ† (xµ )γ 0 . in fact. Multiplying (10.267) has the desired property of being positive definite. The reason is that the r.268) holds for the bispinor Lorentz transformation.s.

10. (10. as just mentioned. bispinor wave functions and we need to determine corresponding components of the wave function. a property which leads to the energy– momentum (dispersion) relationship of the Dirac particle. not only as the most simple demonstration of the theory. The operators commute also with Ho in (10. In case of non-relativistic quantum mechanics the free particle wave function has a single component ψ(r.275) The free particle wave function is an eigenfunction of Ho . 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.283)]..274) 10. of which the wave functions are eigenfunctions as well. i. µ (10.246) S −1 (Lη ξ )γ µ S ( Lη ξ ) = (L−1 )νµ γν = (L−1 )ν γ ν = Lµ ν γν . One obtains using (10.76).f. Combining this with (10. but also as providing a basis in which the wave functions of interacting particle systems can be expanded and characterized. as is required for the mentioned property.9 Solutions of the Free Particle Dirac Equation We want to determine now the wave functions of free particles described by the Dirac equation. As it turns out. Note that we have assumed in (10. 10.322 Relativistic Quantum Mechanics in a moving frame of reference and the flux j µ in a frame at rest. 10. These operaˆ tors.272) results in the expected transformation behaviour µ j = Lµ ν j ν .e. additional characterizations. (10. In relativistic quantum mechanics a Dirac particle can also be characterized through a momentum. specific for the Dirac free particle. the Dirac particles are described by 4-dimensional. In the following we want to provide this characterization. (10. t) and is determined through the momentum p ∈ R3 . Like for the free particle wave functions of the non-relativistic Schr¨dinger and the Klein–Gordon o equations one expects that the space–time dependence is governed by a factor exp[i(p · r − t)]. (10.268) µ j = Ψ† (xµ )S † γ 0 γ µ SΨ(xµ ) = Ψ† (xµ )γ 0 S −1 γ µ SΨ(xµ ) .282.233) o i∂t Ψ(xµ ) = Ho Ψ(xµ ) . the identification of observables and their quantum mechanical operators. We will start from the Dirac equation in the Schr¨dinger form (10.272) With S −1 (Lη ξ ) = S((L−1 )η ξ ) one can restate (10. As pointed out.232. The solutions provide also a complete. For this purpose we consider . however. commute with each other.231.273) where we have employed (10. The additional degrees of freedom described by the four components of the bispinor wave function require. Like in non-relativistic quantum mechanics the free particle wave function plays a central role. the wave function has four components which invite further characterization of the free particle state. only two degrees of freedom of the bispinor four degrees of freedom are independent [c. 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.271) that the Dirac matrices are independent of the frame of reference.233).

like the Klein–Gordon equation. hence. from (10.276) into (10. (10.280) which has a positive and a negative solution = ± E(p) .279) the well-known relativistic dispersion relationship 2 = m2 + p 2 (10. 10.93. and φo .283) imply that the bispinor part of the wave function allows only two degrees of freedom to be chosen independently. namely φo and χo are related as follows φ0 χo = = σ·p χ0 −m σ·p φ0 . two-dimensional spinor state.284) The relationships (10. 10.279) According to the property (5.278) provides us with information about the components of the bispinor wave function (10.278) Multiplication of the 1st equation by ( +m)1 and of the second equation by −σ · p and subtraction 1 of the results yields the 2-dimensional equation ( 2 − m2 ) 1 − (σ · p)2 φo = 0 .231. later to be identified with momentum and energy. (10.234) of Pauli matrices holds (σ · p)2 = p 2 1 One can.281).94) Equation (10. +m (10.282) (10.276) where p and together represent four real constants.280) σ·p (σ · p)2 φ0 = χ0 = +m ( + m) ( + m) p2 χ = χ0 . (10. These two relationships are consistent with each other.283) where is defined in (10. E(p) = m2 + p 2 . .232) leads to the 4-dimensional eigenvalue problem m σ·p σ · p −m φo χo = φo χo .281) Obviously.276). the Dirac equation. In fact.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. reproduce the classical relativistic energy–momentum relationships (10. 1 (10. χo each represent a constant.234) and (10. the socalled helicity of the particle.277) To solve this problem we write (10. 2 − m2 0 (10. conclude 1. Inserting (10. We want to show now that these degrees of freedom correspond to a spin-like property. one finds using (5.277) explicitly ( − m) 1 φo − σ · p χo 1 − σ · p φo + ( + m) 1 χo 1 = 0 = 0. 10.10.282.

The connection between the solutions.285)]. is Ψ(p.290).e. using γ 0 as given in (10.. and let Ψ(0.s. ±).241).268).287) in the minus sign on the r. we demonstrate that the product Ψ† (p. u† u = |u1 |2 + |u2 |2 = 1 . −|xµ ) = N− (p) −σ·p E(p) + m u u ei(p·r − t) . = −E(p) . +) = 1 . of (10. of (10. ±).290) should differ.324 Relativistic Quantum Mechanics For our further characterization we will deal with the positive and negative energy solutions [cf. ±) = S Ψ(0.h. (10. ±) denote the corresponding solution of a particle in its rest frame.s. 10. +) = |N+ (0)|2 (u† . ±) denote the solution of a free particle moving with momentum p in the laboratory frame.291) Note that we have used that. For the positive energy solution.. We want to demonstrate now that the signs on the r. we present φo through the normalized vector φo = u1 u2 = u ∈ C2 . (10. Similarly. (10.h. = +E(p) . i. u† u = |u1 |2 + |u2 |2 = 1 .285) The corresponding free Dirac particle is then described through the wave function Ψ(p.286) for p = 0 yields. (10..288) corresponding to the wave function Ψ(p. the l. +|xµ ) = N+ (p) u σ·p E(p) + m u ei(p·r − t) . (10. i.259). Ψ† (0. ±)γ 0 Ψ(p. The form of this condition and of (10.281)] separately.e. +) γ 0 Ψ† (p. we present the negative energy solution. +) = −1 . +) γ 0 Ψ† (p. ±) = = = Ψ† (0. 10.287. ±) Ψ† (0.219) and u† u = 1 [c. One can see this as follows: Let Ψ(p. Employing (10.e. ±) γ 0 S −1 γ 0 γ 0 S Ψ(0.289) Here N− (p) is a constant which will be chosen to satisfy the normalization condition Ψ† (p. ±) Ψ† (0. S † = γ 0 S −1 γ 0 . where S is given by (10. (10. +) γ 0 Ψ(0. 0) γ 0 u 0 = |N+ (0)|2 . (10. ±) γ 0 Ψ(0. Ψ† (p. i. First. the solution for = +E(p). according to (10.h.287. (10. For this purpose we consider first the positive energy solution.287) will be justified now.287) the form of which will be justified further below.290) which differs from the normalization condition (10. ±) .f. the solution for = −E(p). Hence. through χo given by χo = u1 u2 = u ∈ C2 .286) Here N+ (p) is a constant which will be chosen to satisfy the normalization condition Ψ† (p.292) . (10. is invariant under Lorentz transformations. ±) γ 0 Ψ(p.s. hence. according to (10. ±) S † γ 0 S Ψ(0. S −1 = γ 0 S † γ 0 and. (10.

( m + E(p) )2 − p 2 (10.h. 2 E(p) (10.219) yields 2 N+ (p) u† u − u† (σ · p)2 u (E(p) + m)2 = 1.296) Replacing (σ · p)2 by p 2 [c.h.298) becomes N+ (p) = m + E(p) .234)] and using the normalization of u in (10.286).300) (10. u† ) γ 0 0 u = − |N− (0)|2 .299) ( m + E(p) )2 .289) yields Ψ† (0. σ·p u∗ E(p) + m T u σ·p E(p) + m u = 1 (10. (5. this requires the choice of a negative side on the r.286) is 2 N+ (p) (u∗ )T . σ·p u∗ E(p) + m T γo u σ·p E(p) + m u = 1 (10. 2m (10.10.f. −) = |N− (0)|2 (0.294) We want to determine now N± (p) for arbitrary p. of (10. Exercise 7. (10. We can also conclude from our derivation N± (0) = 1 .285) results in 2 N+ (p) 1 − p2 (E(p) + m)2 = 1 (10.s.302) .301) yields the normalization coefficient N+ (p) = m + E(p) .293) Obviously.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.s.295) Evaluating the l.298) This result completes the expression for the wave function (10.6: Show that the normalization condition 2 N+ (p) (u∗ )T . 325 (10. using γ 0 as given in (10.9: Solutions of the Free Particle Dirac Equation The same calculation for the negative energy wave function as given in (10. −) γ 0 Ψ(0. (10.290) to assign a positive value to |N− (0)|2 .287) written explicitly using (10. We consider first the positive energy solution. Condition (10.

the two components of u indicate another degree of freedom which needs to be defined. (10. 10. j = 1. identical to the condition (10. however. 10. 2. i.s.290) written explicitly using (10. λ|xµ ) = λ E(p) Ψ(p.289) is T −σ·p −σ · p 2 E(p) + m u N− (p) u∗ . (u∗ )T γ o = −1 (10.286. hence.300) have been constructed to satisfy the free particle Dirac equation (10.e.275). defined as the component of the particle spin along the direction of motion.289).305) and.e.289) are not completely specified. The wave functions (10. λ|xµ ) in 4-vector notation.304) This condition is. have completed the determination for the wave function (10.e. This attribute is the so-called helicity. −p). i. yields 2 N− (p) u† (σ · p)2 u − u† u (E(p) + m)2 = −1 .h. The wave functions are also eigenstates of the momentum operator i∂µ . This can be verified expressing the space–time factor of Ψ(p. We can. The property [σ · p. exp[i(p · r − t)] = exp(ipµ xµ ). Inserting (10. (10.296) for the normalization constant N+ (p) of the positive energy solution.303) E(p) + m u Evaluating the l.326 Relativistic Quantum Mechanics We consider now the negative energy solution. Ho ] = 0 follows from (10. λ|xµ ) .308) 2 |p| ˆ Note that p represents here an operator. pj ] = 0 . 3 is fairly obvious. Condition (10. not a constant vector..306) i.289. the wave functions constructed represent eigenstates of Ho . This degree of freedom 1 describes a spin– 2 attribute. λ|xµ ) = pµ Ψ(p. 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.275) yields Ho Ψ(p. i∂µ Ψ(p. Rather than considering the ˆ observable (10. The corresponding operator which measures this observable is 1 p ˆ Λ = σ· .. (10.286) into (10.309) . thereby. 10. λ|xµ ) (10.307) where pµ = ( .286.. The commutation property [σ · p. Helicity The free Dirac particle wave functions (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 (10.307) we investigate first the observable due to the simpler operator σ · p. conclude N− (p) = m + E(p) 2m (10.

286) are eigenfunctions of Λ as well will specify now the vectors u. 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 .. i. in principle. Ho and σ·p commute with each other and.311) 2  −p  0 m + Ep 1 where e3 denotes the unit vector in the x3 -direction and where ˆ Ep = m2 + p2 . is independent of the direction of motion of the particle.. p3 ).e. Np = m + Ep . Ho and Λ defined in (10. t) = Np  e (10. t) = Np  e 2   1 p m + Ep 0   0   i(px3 − E t) 1 p e Ψ(pˆ3 . In this case Λ = 2 σ 3 . can ˆ ˆ be simultaneously diagonal. According to the definition (5. States which are simultaneously eigenvectors of these three operators are also simulteneously eigenvectors of the three operators p. are   1   i(px3 − E t) 0 p e Ψ(pˆ3 . 2 the wave functions which are eigenstates of the helicity operator. − 1 |r. 1 p = (0. = +E(p). − 1 |r.312). Therefore. Since helicity is defined relative to the direction of motion of a particle the characterization of u as an eigenvector of the helicity operator. 0)T and (0.10. −. i. + 2 |r.310) We have shown altogether that the operators p. 0. + 1 |r.e. +. −.312) We assume now particles with negative energy.224) 1 of σ3 the two u vectors (1. = −E(p). i.313) 2   0 1 where Ep and Np are defined in (10. hence. We consider first the simplest case that particles move along the x3 –direction. t) = Np  e (10. +. 1)T are eigenstates of 2 σ 3 with eigenvalues ± 1 . We assume first particles with positive energy.e. ˆ The condition that the wave functions (10. 2m (10.308) above.. . t) = Np  e   1 0   0 p  m + Ep  i(px3 + E t) 1 p e Ψ(pˆ3 . 327 0 (σ · p)2 ˆ 2 (σ · p) ˆ 0 0 (σ · p)2 ˆ 2 (σ · p) ˆ 0 (10.9: Solutions of the Free Particle Dirac Equation and 0 σ σ 0 = ·p ˆ σ 0 0 σ ·p − ˆ − σ 0 0 σ ·p ˆ 0 σ σ 0 ·p ˆ = 0.

−.226. +.222. γ and are  1  0  1 ˜ Ψ(p = 0. + 1 ) 2 p 1 m + Ep u(p. + ) 2 1 u(p. + 2 ) p 1 m + Ep u(p. + 1 |r.314) (10. − 1 |r.318) 1 Ψ(p.319) 1 Ψ(p. These eigenstates are obtained through a rotational transformation (5. − 1 ) 2 ei(p·r − Ep t) (10. Generating Solutions Through Lorentz Transformation The solutions (10. t) = N√ ei(p·r + Ep t) (10.312) can be obtained also by means of the Lorentz transformation (10. 1 |t) = √2   e−imt .313) except that the states (1. 10.123). i. a wave function which represents free particles at rest. For this purpose one starts from the solutions of the Dirac equation in the chiral representation (10. for an r–independent wave function.223) as follows 1 u(p.e. [One can also express the rotation through Euler angles α. 10. The corresponding wave functions are determined through ˜ i˜ 0 ∂t − m Ψ(t) = 0 . 1)T have to be replaced by eigenstates u± (p) of the spin operator along the direction of p.. γ .315) e3 × p ˆ ∠(ˆ3 . t) 2 = Np u(p. + 2 |r.220). β. 0)T and (0. − ) 2 where ϑ(p) = = = exp exp i − ϑ(p) · σ 2 i − ϑ(p) · σ 2 1 0 0 1 . t) = Np ei(p·r + Ep t) (10. + 2 ) u(p. + 2 ) ei(p·r − Ep t) (10. t) 2 = Np 1 u(p. − 2 |r. (10.311. denoted by ˜.311.262) for the bispinor wave function and the transformation (10.320) where Ep and N are again given by (10.316) describes a rotation which aligns the x3 –axis with the direction of p. 5. p) e |p| (10. in which case the transformation is given by (5. 2  1  0  (10. − 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 (10.312). +. −. 10.229).321) . − 2 ) u(p. +.317) Ψ(p.] The corresponding free particle wave functions are then Ψ(p. − 1 ) 2 −p 1 m + Ep u(p.

10. 0 1 (10.176) imt → i ( p3 x3 and for the bispinor part according to (10. 2 The solutions (10. w3 ). −. ± 1 of the wave functions stated in (10.322) can be written in spinor form 1 √ 2 φo χo e imt .  −1  e 0   0  1  +imt 1 ˜  Ψ(p = 0. yields for the exponential space–time dependence according to (10. i.324. 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 .262) for a boost in the x3 –direction. +.325) to (10. t) e 2 e− 2 w3 1  = √  1 2  e 2 w3  i(px3 − E t) p e  .324) 0 e − 1 w3 σ 3 2 φo χo = e 2 w3 σ φo 1 3 e− 2 w3 σ χo 1 3 . 10.325) One should note that φo .313) in the p → 0 limit. −. − 1 |r.. for w = (0. − 1 |t) = √2   e−imt . 10. − 1 |t) = √2  .323) should yield the solutions for non-vanishing momentum p in the x3 –direction.311. χo are eigenstates of σ 3 with eigenvalues ±1. φo . 2  0  1   1  0  +imt 1 1 ˜  Ψ(p = 0. χo ∈ 1 0 .228) yields the corresponding solutions (10.322) The reader can readily verify that transformation of these solutions to the Dirac representationsas defined in (10. + 1 |r.262) φo χo → e 2 w3 σ 0 1 3 Et ) (10. 2  0 e −1  329 (10. This correspondence justifies the characterization ±.10.323) Transformation (10.174. 0.e. +. Applying (10.322).311)  e 1 w 2 3 ˜ Ψ(p(w3 )ˆ3 . 2 |t) = √2  .9: Solutions of the Free Particle Dirac Equation  0  1  1 ˜ Ψ(p = 0. +. (10. For the resulting wave functions in the chiral representation one can use then a notation corresponding to that adopted in (10.

+ 2 |r.326) where according to (10. − 1 |r. t) e  =     cosh w3 2 − sinh w3 2 sinh w3 2 cosh w3 2 − sinh w3 2 cosh w3 2  i(px3 − E t) p e   Ψ(p(w3 )ˆ3 . and the expression (10. + 2 |r.330  e 1 w 2 3 Relativistic Quantum Mechanics 1 0 1 0 0 1 0 1  1 ˜ Ψ(p(w3 )ˆ3 .329) . t) e 1  = √  2   −e− 2 w3 − 1 w3 2 1  i(px3 + E t) p e   ˜ Ψ(p(w3 )ˆ3 .61) p(w3 ) = m sinhw3 . −. −. +. + 1 |r. +. t) e  =    i(px3 + E t) p e  (10. − 2 |r. t) e 2 e 1   = √  1 2 − e 2 w3  i(px3 + E t) p e  (10. −. − 2 |r.311) for Ep one obtains 1 = √ 2 1 1 √ 2 + 1 = 1 − v3 2 m2 + = 2 m2 v3 2 1 − v3 cosh w3 2 1+ 2 v3 2 + 1 1 − v3 +m = 2m Ep + m 2m (10. t) e 2  =     i(px3 + E t) p e   1 Ψ(p(w3 )ˆ3 . 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 . −.61) between the parameter w3 and boost velocity v3 . 2 (10. Transformation to the Dirac representation by means of (10.327) Employing the hyperbolic function properties cosh x = 2 coshx + 1 .228) yields w3  cosh 2 =   sinh w3 2 1 Ψ(p(w3 )ˆ3 . 2 sinh x = 2 coshx − 1 .328) the relationship (10.

10. Invariance of Dirac Equation Revisited At this point we like to provide a variation of the derivation of (10. This transformation. λ. indeed.334) (10. in general.333) . 10.9: Invariance of Dirac Equation Revisited and similarly sinh w3 = 2 Ep − m = 2m p 2m (Ep + m) 331 (10.221) using (10.243) and.243) considered solely the limit of infinitesimal transformations anyway. when we generated the solutions Ψ(p. Such transformation had been applied by us.176). The transformation is Ψ (xµ ) = S(Lη ξ ) ρ(Lη ξ ) Ψ(xµ ) (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.243). λ. reproduces the positive energy wave functions (10. Λ|t). 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. and since (2) the calculations following (10.329.330) Inserting expressions (10. expresses the system in the old coordinates. which however. is sufficient since (1) it must hold then for any finite transformation. (10.327). ρ(Lη ξ ) has been defined in (10. thereby.168–10. In the new derivation we consider the particle described by the wave function transformed. of course. refered to as the active transformation. 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 .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 ).330) into (10.331) i S(Lη ξ )γ µ S −1 (Lη ξ ) ρ(Lη ξ )∂µ ρ−1 (Lη ξ ) − m Ψ (xµ ) = 0 (10. we will derive this equation only for infinitesimal transformations. To show this we rewrite the Dirac equation (10. We expect. Λ|xµ ) from the solutions describing particles at rest Ψ(p = 0.311) as well as the negative energy solutions (10. Actually. but not the observer. The reason why we provide another derivation of (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 essential property stating the Lorentz–invariance of the Dirac equation.123) above and characterized there. the proper transformation S(Lη ξ ). that if Ψ(xµ ) is a solution of the Dirac equation that Ψ (xµ ) as given in (10. Making this expectation a postulate allows one to derive the condition (10.331) is a solution as well.313) for −p. Such description will be essential for the formal description of Lorentz invariant wave equations for arbitray spin further below. = γν .

∂µ ] = (10.334) for infinitesimal Lorentz transformations Lη ξ . ϑ = (0. w = (0. ∂µ ] = (J )ν µ ∂ν . 0).     0 0  ∂3   −∂2  0    −∂3 0    ∂1  0    ∂2  −∂1   0   −∂1   −∂0 0    0   −∂2   0   −∂0  0   −∂3   0 0    −∂0 µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ = = = = = = = = = = = = = = = = = = = = = = = = 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 (10. therefore. ∂µ ] = (K )ν µ ∂ν .337) [J2 . i. 0.. . w = (0. ∂µ ] = (10. the Dirac equation is invariant under active Lorentz transformations. As mentioned already we will show condition (10.338) [J3 .262). 0). We have demonstrated. 0). [K . 0. 0. .335) shows that we need to demonstrate [J . ∂µ ] = (10. ∂µ ] = (10.341) [K3 . 0). 1.339) [K1 .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. that any solution Ψ(xµ ) transformed according to (10.342) One can readily convince oneself that these results are consistent with (10.336) We will proceed with this task considering all six cases: [J1 .340) [K2 . of course. ∂µ ] = (10.e. ∂µ ] = (10. We will proceed by employing the generators (10.332 Relativistic Quantum Mechanics We will demonstrate now that the first condition is satisfied by ρ(Lη ξ ). ϑ = (0. ∂µ ] = [x0 ∂1 + x1 ∂0 .336). ∂µ ] = [x0 ∂2 + x2 ∂0 . . ∂µ ] = [x3 ∂2 − x2 ∂3 . 0. . ∂µ ] = [x0 ∂3 + x3 ∂0 . Inspection of (10. The second condition is identical to (10.243) and.331) is again a solution of the Dirac equation. ∂µ ] = [x2 ∂1 − x1 ∂2 . it is met by S(Lη ξ ) as given in the chiral representation by (10. ∂µ ] = [x1 ∂3 − x3 ∂1 . w = (0. 1) . 0.335) = ∂µ + M ν µ ∂ν + O( 2 ) The result will show that the matrix M ν µ is identical to the generators of Lν µ for the six choices ϑ = (1. . 0).

10. σ3 )T one obtains then i∂t φ i∂t χ = = ˆ σ · π χ + qV φ + mφ ˆ σ · π φ + qV χ − mχ .1 in Sect 10. accordingly.10. e. Accordingly. This solution exhibits a timedependence exp[−i(m + )t] where for holds in the non-relativistic limit | | << m.233) by replacing i∂t by (see Table 10. (10.6 we assume that the field is described through the 4-vector potential Aµ and. Equivalently.343) One may also include the electromagnetic field in the Dirac equation given in the Schr¨dinger form o ˆ (10.345) (10. for the scalar field V in (10.6 above).344) Using the notation σ = (σ1 . (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 10.232). we replace the operator ∂µ by ∂µ + iqAµ .345) in the so-called non-relativistic limit in which all energies are much smaller than m.350) (10.349) We want to focus on the stationary positive energy solution. 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. we define φ(xµ ) = e−imt Φ(xµ ) χ(x ) = e µ −imt (10. For this purpose we choose the decomposition Ψ(xµ ) = φ(xµ ) χ(xµ ) . Non-Relativistic Limit We want to consider now the Dirac equation (10.345) holds |qV | << m .351) X (x ) µ .10: Dirac Particles in Electromagnetic Field 333 10.348) (10.221) reads then [ iγ µ (∂µ + iqAµ ) − m ] Ψ(xν ) = 0 (10.347) (10.g.346) ˆ ˆ ˆ α · π + qV + β m Ψ(xµ ) (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. The Dirac equation (10..1) i∂t − qV and p by ˆ ˆ π = p − qA . Following the respective procedure developed for the Klein-Gordon equation in Sect. σ2 .

10.346. due to the m−1 factor.356) i∂t Φ ≈ Φ + qV Φ .356). πk ] = = 1 1 [ ∂j + qAj .352) ˆ σ · π X + qV Φ ˆ σ · π Φ + qV X − 2mX .357) for Φ can be reformulated by expansion of (σ · π)2 . Φ Using (10. ˆ Equation (10. does not need to be considered anymore. 10.354) The properties (10. (10. One obtains [πj . one can replace X in (10. identifies X as the small component of the bi-spinor wave function which. π ] . (10. ∂k ] + [∂j . ∂k ] + q [ ∂j .360) For an arbitrary function f (r) holds ( [Aj . (10. which in the present case states ˆ ˆ2 ˆ ˆ (σ · π)2 = π + i σ · (π × π) .359) We want to evaluate the latter commutator. derived in Sect.352) allow one to approximate (10.351) in (10. 5. For this purpose we employ the identity (5. Ak ] . henceforth. Ak ] ) f = ( ∂j Ak − Ak ∂j + Aj ∂k − ∂k Aj ) f . ∂k + qAk ] i i 1 1 1 1 [ ∂j .349) yields i∂t Φ i∂t X = = ∂t X X Relativistic Quantum Mechanics << m . ∂k ] + q [Aj . and. (10.350.358) ( πj πk − πk πj ) = jk [πj .353) by X ≈ to obtain a closed equation for Φ ˆ σ·π 2m 2 (10. accordingly.230).354) ˆ 0 ≈ σ · π Φ − 2m X .348. 10.357) Equation (10. Ak ] + q 2 [Aj .334 and assume that for the time-derivative of Φ and X holds ∂t Φ << m . Ak ] i i i i =0 =0 = q q [Aj . i i (10.7. ∂k ] + [∂j .355) ˆ σ·π Φ 2m (10.361) . ˆ ˆ For the components of π × π holds ˆ ˆ π×π = jk (10. (10.353) (10.

0. t) [see (8. the interaction qσ · B induces a 1 precession of the spin. but rather remains as a steady “guest” of non-relativistic physics with the proper interaction term.366) can be interpreted as a rotation around the field B by an angle qtB.360) and Aµ = (V. t) 2m 2m Φ(r.363) where we employed B(r.359. t) + q V (r. The formal solution of this equation is Φ(t) = e−iqtB·σ Φ(0) . 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.343) for the vector potential Aµ = (− Ze2 .357) in the final form i∂t Φ(r. 10..362) or.364) which is referred to as the Pauli equation.366) (10. t) (10. Equations (10.344.365) Comparision of this expression with (5. (10. t) = × A(r. r (10. 5. using (10. while agreeing in all other respects with the non-relativistic Schr¨dinger equation (10. Let us consider briefly the consequences of the interaction of a spin. 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) . Ak ] ) f = [ ((∂j Ak )) − ((∂k Aj )) ] f (10. one obtains ( [Aj . Comparision of (10. 0) . [πj .6)]. 0.363) allow us to write (10. 10. 10.2 around the magnetic field.364) governing a two-dimensional wave function Φ ∈ C2 with the corresponding non-relativistic Schr¨dinger equation (10.367) . −A).358.222. The respective bispinor wave function Ψ(xµ ) ∈ C4 is described as the stationary solution of the Dirac equation (10.Spectrum We want to describe now the spectrum of a relativistic electron (q = −e) in the Coulomb field of a nucleus with charge Ze. the spin.10. o reveals a stunning feature: the Pauli equation does justice to its two-dimensional character. does not leave the theory again when one takes the non-relativistic limit.223) shows that the propagator in (10. Dirac Particle in Coulomb Field . t)]2 q − σ · B(r. In other words.2) governing a one-dimensional wave function ψ ∈ C.1 which emerged in the Lorentz-invariant theory as an algebraic 2 necessity. ∂k ] + [∂j .e.2) it introduces the o extra term qσ · B Φ which describes the well-known interaction of a spin. πk ] = q q (( ∂j Ak − ∂k Aj )) = − i i ×A jk q = − B i jk (10. i.1 with the magnetic field.

3.370) is obtained then ˜ [ i˜ µ (∂µ + iqAµ ) + m ] Ψ(xµ ) γ is a solution of (10.371).370) Any solution of (10.369) from the left by γ ν πν + m.6. For this purpose we ‘square’ the Dirac equation. ˜ ˜ ˆ ˆ (10.369).369) For our solution we will adopt presently a strategy which follows closely that for the spectrum of pionic atoms in Sect.230). (5. but the converse is not necessarily true. is a solution of (10. we solve ˜ [ i˜ µ (∂µ + iqAµ ) − m ] Ψ(xµ ) = 0 γ (10.232) in Sect.230) γ µ γ ν = − γ ν γ µ . Employing πµ as defined ˜ in Table 10. noting from (10.368) is also a solution of (10. 2.370) ˜ [ i˜ µ (∂µ + iqAµ ) − m ] [ i˜ µ (∂µ + iqAµ ) + m ] Ψ(xµ ) = 0 γ γ (10. but differs from it in an essential way.372) (10. Equation (10.231).370) for which holds ˜ ˆ ˜ ˆ 3 γ πµ γ πν = ˜ ˆ ˜ ˆ µ=0 µ ν (˜ µ )2 πµ + γ ˆ2 µ. indeed.s. 10. according to (10.375) Following the algebra that connected Eqs. γ ˆ2 ˆ2 (10.370) resembles closely the Klein-Gordon equation (10.229).1 one can write (10. (˜ 0 )2 = 1 and (˜ j )2 = γ 1 γ −1 j = 1. The difference arises from the term γ µ πµ γ ν πν in (10.373) (10.371) such that we can conclude that (10.368) ˜ ( γ µ πµ − m ) Ψ(xµ ) = 0 . i. 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. once a solution Ψ(xµ ) of (10. multiplying (10.374).374) The first term on the r.228) and in (10. respectively. 1.h. µ = ν and altering ‘dummy’ summation ˜ ˜ ˜ ˜ .370).e.ν=1 µ =ν γ µ γ ν πµ πν . ˜ (10.369). can be rewritten using. 5. ˜ However.180). γ ˆ ˜ ˆ (10.336 Relativistic Quantum Mechanics For the purpose of the solution we assume the chiral representation.368) ˜ where Ψ(xµ ) and γ µ are defined in (10. (5. 3 µ=0 ˆ2 (˜ µ )2 πµ = π0 − π . 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 .7 one can write the second term in (10.

πj ] in (10. 3 (10.ν=1 µ =ν 337 1 2 ( γ µ γ ν πµ πν + γ ν γ µ πν πµ ) ˜ ˜ ˆ ˆ ˜ ˜ ˆ ˆ µ.367) and the definition (10.378) According to the definition (10.e.376–10. πν ] γ ˜ π ˆ µ.380) where f = f (r. −i∂j ] = − [ (∂t + iqA0 ) ∂j − ∂j (∂t + iqA0 ) ] f i ((∂j qA0 )) f (10. γ j ] are γ ˜ [˜ 0 .344) π ˆ [ˆ0 . due to A = 0.381) According to (10. one can summarize (10. Altogether.344). πj ] π ˆ = = (−i∂t + qA0 . 2. ν = 1.367) holds qA0 = r ˆ Ze2 . πν ] = 0 π ˆ for µ. πν ] γ ˜ π ˆ µ.ν=1 µ =ν (10..ν=1 µ =ν σ 0 0 −σ · (( qA0 )) (10.378) can be evaluated using (10. i. π0 ] γ ˜ π ˆ j=1 = [˜ 0 .380) γ µ γ ν πµ πν = i ˜ ˜ ˆ ˆ µ.ν=1 µ =ν 3 j=1 3 3 = 1 4 1 2 1 [˜ . γ 0 ] [ˆj . πj ] . t) is a suitable test function and where ((· · ·)) denotes the range to which the derivative is limited.367) of Aµ . γ ] [ˆ0 . r2 (10. γ ν ] [ˆµ .379) The commutators [ˆ0 .10. γ µ γ ν πµ πν = ˜ ˜ ˆ ˆ µ.377) which follows readily from the definition (10. Since [ˆµ .10: Dirac Particles in Electromagnetic Field indices. the commutators [˜ 0 . it holds 1 4 [˜ µ . πj ] + γ ˜ π ˆ 4 0 j [˜ j . γ ˜ π ˆ j=1 (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. γ j ] [ˆ0 .229). 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 (10.382) .376) This expression can be simplified due to the special form (10. γ ν ] [ˆµ .ν=1 µ =ν = [˜ µ .

6.5 which describe the coupling of orbital angular momentum and spin 1 {( Yjm (j − 2 . hence. 1 |ˆ) ) .375) the ˆ ‘squared’ Dirac equation (10. We note that these states are also eigenstates of the angular r 2 . only the two states {Yjm (j − 2 . (10. . is genuinely two-dimensional and. it is this quantity that we want to determine.383) We seek stationary solutions of this equation.387) The expression (10. m = −j.338 Relativistic Quantum Mechanics where r = r/|r| is a unit vector. 6. Yjm (j + r 2 1 1 .374).368) reads − Ze2 ∂t − i r 2 + 2 + i σ·r ˆ 0 0 −σ · r ˆ Ze2 − m2 r2 Ψ(xµ ) = 0 (10. . The term iσ · r.1 .388) in terms of states introduced in Sect.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 . (10. (10. a solution of the form ∼ Y m (ˆ) can be obtained. .5 the operator iσ · r is block-diagonal in this basis such that only ˆ 1 the states for identical j. 1 |ˆ). + j } (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 . 2 3 2 . r r 2 2 2 j = 1. 1 |ˆ).148). 2 |ˆ)} as given in (6. however.384) can be interpreted as the energy of the stationary state and. 6.384) into (10. i.6. 10. Accordingly. m values are coupled. (10. Combining this result with (10.381). in r ˆ fact. . −j + 1.386) + 2 ± iσ · r ˆ Ze2 − m2 r2 φ± (r) = 0 . In the latter case. couples the orbital angular momentum of the electron to its spin. (10. .2 components ˜ Ψ(r) = and obtain for the separate components φ± (r) Ze2 + r 2 ˜ φ+ (r) ˜ φ− (r) (10.191) solved in Sect.389) According to the results in Sect.385) 1 We split the wave function into two spin. .147.e. Insertion of (10. we express 2 the solution of (10.388) Except for the term iσ · r this equation is identical to that posed by the one-dimensional Kleinˆ Gordon equation for pionic atoms (10. Such solutions are of the form ˜ ˜ Ψ(xµ ) = Φ(r) e−i t . (10..

. 10. − 1 r2 (j − 1 )(j + 1 ) − Z 2 e4 2 2 ±i Ze2 (j + )(j + ) − Z 2 e4 We seek to bring (10.393) but do not explicitly consider further the wavefunctions.10: Dirac Particles in Electromagnetic Field 339 ˆ momentum operator L2 [cf.2 (j) as given by (10.391) derived in Sect. 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 . the eigenvalues are independent of m. conclude that the .395) (10. namely. the values of the argument of λ1. Obviously. (6. m and expand φ± (r) = Using σ · r Yjm (j ± 1 .392). .151). 1 |ˆ) = − Yjm (j ˆ r 2 2 1 2 h± (r) g± (r) Yjm (j − 1 . in the present treatment.2 (r) = 0 (10.397) This equation is identical to the Klein-Gordon equation for pionic atoms written in the form (10. 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. 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 (10.2 (j)[λ1. λ1 (j) and λ2 (j).392) reads then in the diagonal representation 2 ∂r − λ1. Since. property (6. and except for the fact that 2 2 we have two sets of values for λ1. 1. 3 . We select.151].f.5 [c.392) into diagonal form. 6.199). unaltered. .396) and (10. 1 |ˆ) r 2 (10. as in the case of pionic atoms.396) and λ2 (j) [λ2 (j) + 1] (10. . rather than = 0. therefore. such transformation can be chosen as to diagonalize the second term. not the wave functions. we want to determine solely the spectrum.. which states that the states Yjm (j ± 1 .2 (j) being 2 j = 1 . namely. except for the slight difference in the expression of λ1.10.186)]. 1 |ˆ) + r Yjm (j + 1 .2 (j) + 1) 2 Ze2 + + r2 r 2 − m2 f1.390) (10. involving the 2 × 2 unit matrix. Any similarity transformation leaves the first term in (10.395.200). (10. 1 |ˆ) r 2 2 2 2 r r . . . (6. a specific pair of total spin-orbital angular momentum quantum numbers j. hence. 1 |ˆ) are r 2 2 ˆ eigenfunctions of L2 .2 (j).392) h± (r) g± (r) = 0. However. We can. the missing additive term − 1 .394) Equation (10.

counts the nodes of the wave function. 2 2 1 ≈ m − (10. (10. 1. . − + 1. = 1 + m Z 2 e4 √ ( n − − 1 + 2 − Z 2 e4 )2 . j .400) and (10. .403) √ Z 2 e4 2 ED (n.397) is again given by eq. . . 2.s. One may also state 2 this in the following way: (10. 2. .396). m = −j. .395. Obviously. . . . and spin-orbital angular momentum j = − 1 . . j = 1 . One obtains in case of (10. For a given value of n the energies 1 and 2 for identical j-values correspond to mixtures of states 1 with orbital angular momentum = j − 2 and = j + 1 . 1. . . 3. These considerations 2 are summarized in the following equations 1 ED (n. 1. .399) n = 0.398) 2 = 1 + m .396) we obtain. One can equate (10. . . .h. −j + 1. . j where 1 corresponds to λ1 (j) as given in (10. . . . m = −j.400) corresponds to a non-relativistic state with quantum numbers n. It is given by n = n + + 1 where is the orbital angular momentum quantum number and n = 0. In case of non-relativistic hydrogen-type 1 atoms. . albeit with some modifications. . . .395) and 2 corresponds to λ2 (j) as given in (10.2 .398. . 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. . j . 1.401) These expressions can be equated with the non-relativistic spectrum.340 Relativistic Quantum Mechanics spectrum of (10. . . 3 . . −j + 1. .400) (10. 2. 1 ms = ± 2 (10.404) = 0. n − 1 . j = 1. m) 2 n = 1. ( +1) − Z 2 e4 )2 (10. (10. . 1. . . The magnitude of the relativistic effect 2 2 e4 . . . the stationary states have binding energies E = − mZ 2 e4 n = 1. −j + 1. (10. . . . . . .402) with (10. 1 = 1 + m Z 2 e4 (n + 1 + (j+ 1 )2 − Z 2 e4 )2 2 Z 2 e4 (n + (j+ 1 )2 − Z 2 e4 )2 2 . . n − 1 . .401) if one attributes to the respective 1 states the angular momentum quantum numbers = j + 2 and = j − 1 .203).401) corresponds to a non-relativistic state 2 with quantum numbers n. m) = 1 + m (n− + . . (10. of these equations describe the binding energy.402) In this expression n is the so-called main quantum number. . m = − . 2 2 m = −j. . . 2n2 = 0. (10. accordingly.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. . . the second term on the r. j = − 2 . 10. . . . including spin. . 10. Using (10. j = + 1 . and spin-orbital angular momentum j = + 1 . 2.

40151 ⇑ . (10. . mom.e.405 ) -13.402)] and the relativistic [cf. . The table entries demonstrate that the energies as given by the expression (10. (10. . binding energy / eV Eq. However. relate closely to the corresponding nonrelativistic states. .402) -13.405). . 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. j 1 2 1 2 1 2 3 2 1 2 1 2 3 2 3 2 5 2 341 rel.60601 -3.551177 ⇑ . is in the range of 10 eV. .1. The reason is that the mean kinetic energy of the electron in the hydrogen atom. 2. .10: Dirac Particles in Electromagnetic Field main quantum number n 1 2 2 2 3 3 3 3 3 orbital angular mom. relativistic. the 2p 3 states with r 2 2 2 2 1 involve four degenerate states corresponding to Y 3 m (1.10.036 by means of Eqs. much less than the rest mass of the electron (511 keV). . This is clearly demonstrated by the comparision of non-relativistic and relativistic spectra of a hydrogen-type atom with Z = 100 in Table 10. n − 1 j = 1 2 1 2 ± m = −j. . 2 |ˆ) for m = ± 1 . −j + 1. 10. (10. . 1 |ˆ) for m = ± 1 .551176 spectr.403. m) = m 1 + Z 2 e4 (n−j −1 + 2 (j+ 1 )2 − Z 2 e4 )2 2 n = 1. ± 3 . One can combine the expressions (10. .402. (10. .60583 -3. binding energy / eV Eq. 1.40147 -1. j (10. j.3.404) finally into the single formula ED (n.405)] spectrum of the hydrogen atom.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 10.40146 ⇑ ⇑ -1. notation 1s 1 2 non-rel..0041 keV and e2 = 1/137. Degeneracies are denoted by ⇑.405) In order to demonstrate relativistic effects in the spectrum of the hydrogen atom we compare in Table 10. in fact. 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 (1.2: Binding energies for the hydrogen (Z = 1) atom.551178 ⇑ . 0 0 1 1 0 1 1 2 2 spinorbital ang. (10. i. Of particular interest is the effect of spin-orbit coupling which removes.405) in terms of the non-relativistic quantum numbers n.2 the non-relativistic [cf. i. for example.e..1. 10. The energies were evaluated with m = 511. = 0.3. r 2 2 2 . the non-relativistic and relativistic energies are hardly discernible. the nonrelativistic degeneracy for the six 2p states of the hydrogen atom: in the present. case these six states are split into energetically different 2p 1 and 2p 3 states.

0041 keV and e2 = 1/137.407) 1 2 This expression allows one. 0) (10. binding energy / keV Eq.2 -17.15. to estimate the difference between the energies of the states 2p 3 and 2p 1 (cf.402. In order to investigate further the deviation between relativistic and non-relativistic spectra of hydrogen-type atoms we expand the expression (10.6 -42. (10.3 spectr.9 ⇑ .10.405) to order O(Z 4 e8 ).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). mom. Degeneracies are denoted by ⇑. (10. (10.2. j 1 2 1 2 1 2 3 2 1 2 1 2 3 2 3 2 5 2 Relativistic Quantum Mechanics non-rel. 0 0 1 1 0 1 1 2 2 spinorbital ang.406) α √ 1+ (n − β + β 2 −α)2 The expansion (10.1 ⇑ . 0. for example.405) reads 1 (10.1 ⇑ ⇑ ⇑ ⇑ rel. The energies were evaluated with m = 511. binding energy / keV Eq. notation 1s 1 2 2s 1 2 2p 1 2 2p 3 2 3s 1 2 3p 1 2 3p 3 2 3d 3 2 3d 5 2 Table 10.1 -34. We assume for the wave function the stationary state . 32 (10. 0. j.402) -136.8 ⇑ .206) provides in the present case ED (n.409) where V (r) is spherically symmetric. Tables 10. .405).0 ⇑ ⇑ -15. (10. It holds for n = 2 and j = 3 .3: Binding energies for the hydrogen-type (Z = 100) atom. 1 2 2 2 2 E (2p 3 ) − E (2p 1 ) ≈ − 2 2 mZ 4 e8 2 · 23 1 −1 2 = mZ 4 e8 .35. m) ≈ m − mZ 2 e4 mZ 4 e8 − 2n2 2n3 1 j+ − 3 4n + O(Z 6 e12 ) . 10.036 by means of Eqs.342 main quantum number n 1 2 2 2 3 3 3 3 3 orbital angular mom.405 ) -161. Introducing α = Z 2 e4 1 and β = j + 2 (10.15. An example for such potential is the Coulomb potential V (r) = −Ze2 /r considered further below.3).

2 |ˆ) r r r . 1 |ˆ) introduced in r 2 2 1 Sect.414) These equations can be brought into a more symmetric form using ∂r + which allows one to write (10. 2 |ˆ) + r Yjm (j − 2 . r r 2 2 2     a(r) b(r) 1 1 1 1 Yjm (j + 2 . (6.413) ˆ σ · p g(r) Yjm (j − 1 .5 [see. (10.   =  r (10. pp. r 2 2 (10. σ · p ˆ σ · p f (r) Yjm (j + 1 . 6. i.10: Dirac Particles in Electromagnetic Field form Ψ(xµ ) = e−i t 343 Φ(r) X (r) .417)  c(r)  d(r) X (r) 1 1 1 1 Yjm (j + 2 . 1 |ˆ) .414) ˆ σ · p rf (r) Yjm (j + 1 . m-values. 2 |ˆ) and does r 2 ˆ has odd parity and it holds [c.411) (10. The arguments above allow one to eliminate the angular dependence by expanding Φ(r) and X (r) 1 in terms of Yjm (j + 2 .e. according to (10. 1 |ˆ) and Yjm (j − 1 .10.411. 1 |ˆ) r 2 2 (10.345).413. 1 |ˆ) r 2 = i ∂r − 1 2 r g(r) Yjm (j + 1 . 1 |ˆ). 2 |ˆ) + r Yjm (j − 2 . i. 6.197. 6.232..5. The Dirac equation reads then. 1 |ˆ) r 2 1 1 = ∂r r r r (10. r 2 (10. in particular. 168]: the term is a scalar (rank zero tensor) in the space of the spin-angular momentum states Yjm (j ± 1 . 10. 1 |ˆ) r 2 2 = i ∂r + −j r 1 g(r) Yjm (j + 2 . 10.198)] not couple states with different j.410) 1 where Φ(r). 1 |ˆ) r 2 2 = i j+ ∂r + r j+ r 1 2 1 r f (r) Yjm (j − 2 . 2 |ˆ)  r Φ(r)  r  .412) ˆ In this equation a coupling between the wave functions Φ(r) and X (r) arises due to the term σ · p. 1 |ˆ) .416) The differential equations (10. 1 |ˆ) r 2 2 = i ∂r + j+ r 1 2 3 2 f (r) Yjm (j − 1 . This term has been discussed in detail in Sect.. X (r) ∈ C2 describe the spatial and spin.412) are four-dimensional with r-dependent wave functions.e. the term is block-diagonal in the space spanned by the states Yjm (j ± 1 . 10. ˆ σ · p X + m Φ + V (r) Φ ˆ σ · p Φ − mX + V (r) X = = Φ X (10.2 degrees of freedom.f.415) 1 ˆ σ · p rg(r) Yjm (j − 2 . but are timeindependent.

421)   g2 (r) X (r) 1 1 − Yjm (j + 2 . 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 (10.157).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. m values.420)   g1 (r) X (r) 1 1 − Yjm (j − 2 . Presently.417) of the form     f1 (r) i Yjm (j + 1 . 10.423) . m pair contribute.411. g1 (r) ∂r − j+ r j+ r 1 2 g1 (r) + [ − m − V (r) ] f1 (r) = 0 ∂r + and for f2 (r). c(r) are coupled.417) into (10. 10. we consider the case that only states for one specific j. According to (10.422) 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. 1 |ˆ)  r Φ(r) 2 2  r    =  (10.419) Obviously. Inserting (10. g2 (r) ∂r + 1 2 f1 (r) − [ + m − V (r) ] g1 (r) = 0 (10. the orthonormality property (6. d(r) are coupled and b(r). 2 |ˆ) r r   f2 (r) 1 1 r Φ(r)  i r Yjm (j − 2 . Accordingly.344 Relativistic Quantum Mechanics In general. using (10.416).412). (10. 2 |ˆ)   =  (10. such expansion must include states with all possible j. there exist two independent solutions (10.415. only a(r).418) holds for f1 (r). 2 |ˆ) r r where the factors i and −1 have been introduced for convenience.

409). 3s 1 . together with the appropriate boundary conditions at r = 0 and r → ∞. (10. Hence. one demonstrates first that the wave function at r → 0 behaves as rγ for some suitable γ.420). one demonstrates then that the wave functions for r → ∞ behaves as exp(−µr) for some suitable µ.422) and (10.423). 2 |ˆ)  r Ψ(xµ ) ≈  .405).e.420) is the large component and X is the small component. (10.e. 2s 1 . to states with angular momentum = j + 2 . = 0. namely. 3d 5 . for small r.422) determines solutions of the form (10. 2 2 2 (10. .. We note that (10.421). 3d 3 . correspodingly the states We consider first the solution of (10. Equation (10. the 2 equations determine. etc.423) are identical.422).405) of the relativistic hydrogen atom the corresponding states have quantum numbers n = 1. . Similarly. 2p 3 . n − 1.422. 3p 1 . Behaviour at r → 0 We consider first the behaviour of the solutions f1 (r) and g1 (r) of (10.422) describes the states 2p 1 .422) corresponds to states   f1 (r) Yjm (j + 1 .426) . 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. 1. In the non-relativistic limit.422) near r = 0. Φ in (10. i. According to the discussion of the spectrum (10. etc.. 10. and obtains finally a polynomial function p(r) such that rγ exp(−µr)p(r) solves (10. . in the non-relativistic limit wave functions   f2 (r) 1 1 i Yjm (j − 2 . . 2 2 2 2 2 2 = j − 1 2 and. the ones given in (10.367) which is spherically symmetric such that equations (10. except for the opposite sign of the term (j + 1 ). We assume the 4-vector potential of pure Coulomb type (10.405).422). enforcing the polynomial to be of finite order leads to discrete eigenvalues .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. 1 |ˆ)  r i 2 2 . According to this procedure.423) apply for V (r) = −Ze2 /r. .422) follows in this case from the same procedure as that adopted for the radial wave function of the non-relativistic hydrogen-type atom. Hence. determining wave functions of the type (10. .10.424) Ψ(xµ ) ≈  r 0 1 i. the radial wave functions for Dirac particles in the potential (10. (10. 3p 3 . . Dirac Particle in Coulomb Field . 2.10: Dirac Particles in Electromagnetic Field 345 Equations (10.422). (10.425) r 0 covers states with angular momentum 1s 1 . The solution of (10. (10. .

422) becomes ∂r g1 (r) ∂r f1 (r) Iterating this equation once yields 2 ∂r g1 (r) 2 ∂r f1 (r) 1 . We have.266. according to (10. hence.435) .427). r2 (10. One obtains γ = ± (j + 1 )2 − Z 2 e4 . 10. g1 (r) →0 ∼ b rγ (10.427) = = 0 0.427) makes only the positive solution 2 possible. The assumed r-dependence in (10.428) = 0. we conclude f1 (r) →∞ ∼ e−µr .429) This equation poses an eigenvalue problem (eigenvalue −γ) for proper γ values. for the particle density holds then ρ(r) ∼ |rγ−1 |2 = Behaviour at r → ∞ For very large r values (10. becomes imaginary.434) √ The solutions of these equations are f1 . assume the r-dependence in (10. ) f1 (r) ( + m ) g1 (r) (10.430) 2 Note that the exponent in (10. determined that the solutions f1 (r) and g1 (r). in case j + 1 )2 < Ze2 . (10. hence. for small 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 . µ = m2 − 2 (10. (10. Only the exponentially decaying solution is admissable and.431) infinite since. g1 ∼ exp(± m2 − 2 r). 10. (10. g1 (r) →∞ ∼ e−µr . Such r-dependence 2 would make the expectation value of the potential r2 dr ρ(r) 1 r (10.433) = = ( m2 − 2 ) g1 (r) ( m2 − 2 ) f1 (r) (10.267.432) = = − ( − m.420).427) with 1 γ = (j + )2 − Z 2 e4 .

434) √ m + a e−µ r . However. we set f1 (r) g1 (r) = = √ √ ˜ m + e−µr f1 (r) −µr = = 0 0 (10.439) − m− e g1 (r) ˜ where µ is given in (10. hence. Solution of the Radial Dirac Equation for a Coulomb Potential To solve (10.442) which leads to a partial cancellation of the asymptotically dominant terms. µ is real.441) correspond to (10.445) .436) f1 (r) = Insertion into (10. We also introduce the new variable ρ = 2µ r . Equation (10.10. without loss of generality we can choose ˜ f1 (r) = φ1 (r) + φ2 (r) . of both (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 .437) where they ˜ ˜ cancelled in case f1 = g1 = a.437) (10. Let us consider then for the solution of (10.444) (10. √ g1 (r) = − m − a e−µr . (10.436).422) for the Coulomb potential V (r) = −Ze2 /r We assume a form for the solution which is adopted to the asymptotic solution (10. g1 (r) = φ1 (r) − φ2 (r) ˜ (10. In the present case the functions f1 and g1 cannot be chosen ˜ ˜ identical due to the terms in the differential equations contributing for finite r.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. Accordingly.10: Dirac Particles in Electromagnetic Field For bound states holds 347 < m and. ρ m+ ρ [∂ρ − (10.441) The last two terms on the l.440) and (10.434) results in √ √ (m − ) m + a − (m − ) m + a √ √ (m + ) m − a − (m + ) m − a which is obviously correct.438) (10.s. (10.440) (10.h.435 ).

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. 10. = m− Ze2 βs s + γ − √m2 − 2 2 (10.450) −√ mZe2 m2 − 2 βs ρs+γ−1 = 0 mZe2 m2 − αs (s + γ) β2 + (j + 1 ) αs + √ 2 s 2 +√ From (10.446) (10.448.348 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 (10.448) φ2 (ρ) = ρ (10.450) follows Ze2 m2 − 2 βs − βs−1 =0 (10.449) into (10.453) (s + γ)2 + Z 2 e4 − (j + 1 )2 2 2 √ Ze m2 − 2 βs .446.447) leads to (s + γ) αs + (j + 1 ) βs − √ 2 s Ze2 m2 − 2 αs (10.f.452) From (10. Inserting (10.430) one can write this βs = s+γ − 2 √ Ze m2 − 2 s (s + 2γ) βs−1 .430) which conform to the proper r → 0 behaviour determined above [c.447) We seek solutions of (10.430)]. Using (10. 10.426–10.447) of the form n φ1 (ρ) = ργ s=0 n γ s=0 αs ρ2 βs ρ2 (10. (10.454) .446 .451) 1 √mZe − (j + 2 ) 2 2 αs .449) for γ given in (10. (10.

n It holds φ1 (ρ) φ2 (ρ) = = ργ F (−so . . . i. . (s − so ) β0 s! (2γ + 1)(2γ + 2) · · · (2γ + s) (10. in fact. α + 1. n = 0.455) are finite.461 for values given by (10. .10: Dirac Particles in Electromagnetic Field Defining so = √ one obtains βs = = .405). 10. .457) for βs and αs imply that so must then be an integer.e. According to the definitions (10. m (n ) = ..458) or.456. 2.462) 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 ( 10.452) follows j+ αs = 1 2 349 Ze2 m2 − 2 −γ (10. This requires that all coefficients αs and βs must vanish for s ≥ n for some n ∈ N. We can. . 2. so = n . for the states 2p 1 . β0 j+ 1 2 (10. 3d 3 holds n = 1. (s − so ) β0 s! (2γ + 1)(2γ + 2) · · · (2γ + s) (10. 10. . namely. 1. equivalently.405) allows one to identify n = n − j + 2 which.448. c.457) with the confluent hypergeometric functions F (a. is an integer.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 ) . ρ) . c c(c + 1) 2! (10. (10. . x) = 1 + a a(a + 1) x2 x + + . For example.460) (10. ρ) so β0 ργ F (1 − so .455) this confinement of so implies discrete values for . hence.. 2γ + 1. = From (10.. conclude 2 2 2 that the polynomials in (10. . with the associated Laguerre polynomials L(α) = F (−n.457) One can relate the polynomials φ1 (ρ) and φ2 (ρ) defined through (10.456) and (10.459) − 2 √mZe m2 − 2 (10.456) − so 2 √mZe m2 − 2 (−so )(1 − so )(2 − so ) . x) .449) must be of finite order.461) In order that the wave functions remain normalizable the power series (10. 3p 1 . 10.430. 10. 1.448.449) and (10. The expressions (10. 2γ + 1. Comparision with (10. .

460. The coefficients β0 in (10. are two-dimensional vectors determined through the explicit form of the spin-orbital angular momentum states Yjm (j ± 1 .350 Relativistic Quantum Mechanics Altogether we have determined the stationary states of the type (10.460.467) ± n F (1 − n . (10. 9.461) are to be chosen to satisfy a normalization condition and to assign an overall phase. is time-independent. 1 |ˆ) in (6.148). The orthonormality 2 2 r properties (6. (10.463) and yield [note the 1/r factor in (10. 2γ + 1.464) 0 The evaluation of the integrals. j. 1 |ˆ) 2 2 r G1 (r) Yj. Due to the form (10. The complete wave function is given by the following set of formulas 1 Ψ(n. given by expression (10. The wave functions (10. 6.469) This formula has been adapted from ”Relativistic Quantum Mechanics” by W. 10. Greiner.463) where Φ and X .421). can follow the procedure adopted for the wave functions of the non-relativistic hydrogen atom and will not be carried out here. m. (n + γ)m N (2µr)γ−1 e−µr − κ F (−n .438. where2 F± (κ|r) = G1 (r) = F+ (κ|r) .m (j − 1 . = j + 2 .6. j. Sect. 2µr) (10.421)] ∞ dr (|f1 (r)|2 + |g1 (r)|2 ) = 1 (10.157.421) with radial wave functions f1 (r). which involve the confluent hypergeometric functions in (10.m (j + 1 . 10. They are described through quantum numbers n. 10.468) −κ n! µ γ n = = = = (m − )(m + ) (j + 1 )2 − Z 2 e4 2 n−j − m 1+ 1 2 Z 2 e4 (n +γ)2 .421) correspond to non-relativistic states with orbital angular momentum 1 1 = j + 2 . Berlin. 1990).461). and (10. 6.466) F1 (r) = F− (κ|r) . g1 (r) determined by (10.465) (10.410) of the stationary state wave function the density ρ(xµ ) of the states under consideration. 2γ + 1. 1 |ˆ) 2 2 r κ = j+ 1 2 (10. m|xµ ) = e−i t iF1 (r) Yj.147.460. 2 . 10.439). as in (10. (Springer. = j + 2 .158) of the latter states absorb the angular integral in (10.442). 2µr) (2µ) 2 Γ(2γ + 1) 3 N = and m 4m )Γ(2γ + n + 1) (n +γ)m (n +γ)m (10.461).267). The normalization integral is then ∞ 0 π 2π r2 dr 0 sin θdθ 0 dφ (|Φ(r)|2 + |X (r)|2 ) = 1 (10.

2p 3 .470) r 0 Obviously. obtained closed expressions for the wave functions of all the stationary bound states of relativistic hydrogen-type atoms. 2s 1 . 2 |ˆ)  r Ψ(xµ ) ≈ e−1 t  . = j − µ 1 2 . 1 |ˆ) 2 2 r G2 (r) Yj. where F± (κ|r) are as given in (10.472) F2 (r) = F− (κ|r) . We have. G2 (r) = F+ (κ|r) .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. 2 accordingly.465–10. this wavefunction has an orbital angular momentum quantum number = j − 1 and.421) which. The radial wave functions f2 (r) and g2 (r) in (10. etc.469) that the following wave functions result Ψ(n. .m (j − 1 . (10.422) to (10. tracing all steps which lead from (10. in the non-relativistic limit. become   f2 (r) 1 1 i Yjm (j − 2 . hence. 3d 5 . 3s 1 .m (j + 1 .471) (10. One can verify.469). j.469). not not 2 2 2 2 2 2 covered by the wave functions given by (10. 1 |ˆ) 2 2 r κ = −j − 1 2 (10.467–10.421) are governed by the radial Dirac equation (10. describes the complementary set of states 1s 1 .10: Dirac Particles in Electromagnetic Field 351 We want to consider now the stationary states of the type (10. 3p 3 . m|x ) = e −i t iF2 (r) Yj.10.

352 Relativistic Quantum Mechanics .

ϑ)µ ν correspond to different a(z) and b(z). Starting point is the Lorentz transformation S(w. ϑ) for the ˜ in the chiral representation as given by (10. 1 353 . This is. in fact. ϑ through 3. what will be achieved in the following. must be two-dimensional irreducible representations of the Lorentz group. ϑ)µ ν . in a sense. 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 (11.229).2) (11.1) (11. ˜ We have altered here slightly our notation of S(w.e. isomorphic to the natural representation.4) = exp = w − iϑ . We will characterize the space on which the transformations a(z) We will see below that the representations a(z) and b(z) are. 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.262).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..3) (11.e. 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. different L(w.225–10. indeed. i. The lower dimensionality of a(z) and b(z) implies. a(z) and b(z). in particular. expressing its dependence on w.. We will proceed by building as much as possible on the results obtained sofar in the chiral representation of the Dirac equation. may be exploited to express the Lorentz-invariant equations of relativistic quantum mechanics. ϑ). z ∈ C ˜ Because of its block-diagonal form each of the diagonal components of S(z). i. a complex variable z.Chapter 11 Spinor Formulation of Relativistic Quantum Mechanics 11.

354 Spinor Formulation and b(z) act. ϑ ∈ R3 . ϑ ∈ R3 (11.9) a∗ (z) = exp 2 .1 states as described by Dm m (ϑ). Ψ2 (xµ ) and Ψ3 (xµ ). finally. however. A First Characterization of the Bispinor States We note that in case w = 0 the Dirac transformations are pure rotations. in 2 fact. (11. for z = iϑ. i. usually expressed 2 as product of rotations and of functions of Euler angles α. to the rotational transformations of spin. Aν . + 1 2 2 state 2 ∼ | 1 . γ (see Chapter 5). +1 2 |1. (11. ± 1 represents the familiar spin– 1 states. In this case a(z) and b(z) are identical and read a(i ϑ) = b(i ϑ) = exp 1 − ϑ·σ 2 . i. 2 2 2 ˜ Since a(z) acts on the first two components of the solution Ψ of the Dirac equation. +1  2 2 |1. Here | 1 . a distinct difference in the transformation behaviour of ˜ ˜ ˜ ˜ Ψ1 (xµ ). are elements of SU(2) and correspond. Ψ4 (xµ ) transform according to ˜ ˜ ˜ and Ψ b(z). β.243)..5) The transformations in this case. ϑ ∈ R3 .6) as given by (5. −1  2 2  |1. and the Klein–Gordon equation in the spinor representation. In this case holds a(z) = b(z) ˜ 1 (xµ ). β.e. will establish the map between L(w. to 1 ∗ ∗ z ·σ (11.. ϑ)µ ν and a(z).8) We like to stress that there exists. namely.e. 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. Ψ4 (xµ ) in case z = w + iϑ for w = 0. 0)T the transformations are (1) a(i β e2 ) = b(i β e2 ) = ˆ ˆ 2 dmm (β) (1) = cos β 2 sin β 2 − sin β 2 cos β 2 . Ψ2 (xµ ) transform according to a(z) whereas Ψ3 (xµ ). neutrino equation. Relationship Between a(z) and b(z) The transformation b(z) can be related to the conjugate complex of the transformation a(z). − 1 2 2 z = iϑ. For ϑ = (0.7) where “∼” stands for “transforms like”. express 4-vectors Aµ . the so-called spinor space. b(z). θ ∈ R3 . This characterization allows one to draw conclusions regarding the state space in which a(z) and b(z) operate. formulate the Dirac equation. (11. −1 2 2 z = iϑ. a space of vectors state1 for which holds state2 state 1 ∼ | 1 . the operator ∂µ and the Pauli matrices σ in the new representation and.

The effect of inversion on Lorentz transformations is. Spatial Inversion ˜ One may question from the form of S(z) why the Dirac equation needs to be four-dimensional.224)] ∗ σ1 = σ1 . therefore. σ3 ). σ2 .16) σ = −σ ∗ . −1 given by (11. (11. a result to be used further below.12) To prove (11. Ψ (xµ ) as well as Ψ (xµ ). a∗ (z) −1 = exp 1 ∗ z σ∗ 2 −1 . however.10. Hence. One can readily verify [c. one can express a∗ (z) −1 = exp 1 − z ∗· σ 2 = b(z) .14) . −1 a∗ (z) −1 (11. 1. but leave rotation angles ϑ unaltered.2.12) does. that the transformation (11. ∗ σ 3 = σ3 . 11.18) with a(z) given by (11. . −1 = ∗ −σ3 = −σ ∗ . that they alter w into −w.1) can be written in the form ˜ S(z) = a(z) 0 0 a∗ (z) −1 (11.17) We conclude. (11.13) Explicit matrix multiplication using (5. hold −1 = 1 One notices then. µ ). (5. (11. one can show −1 −1 −1 = −σ1 = σ2 = −σ3 ∗ = −σ1 ∗ = −σ2 σ2 −1 (11.11. 11.15) (11.f.11) 0 −1 1 0 = . 355 (11.4) and .12). This demonstrates that a(z) is the ˜ transformation which characterizes both components of S(z). Ψ (xµ ) even though these pairs of ˜ ˜3 ˜4 featuring the components Ψ1 (x ˜ 2 components transform independently of each other. using f (a) −1 = f ( a −1 ). in fact.224. 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.1: Lorentz Transformation of Dirac Bispinor ∗ ∗ ∗ where σ ∗ = (σ1 . ∗ σ 2 = − σ2 .11) one first demonstrates that for and −1 as given in (11. (11. 11.10) From this one can derive b(z) = where = 0 1 −1 0 .12) yields σ1 σ3 or in short σ Similarly.

One can verify this by evaluating the determinant of a(z) det( a(z) ) = det e 2 z·σ 1 (11.C) and SO(3. Amsterdam. P −1 = P. a(w + iϑ) for w = 0 is an element of SL(2. non-singular matrix M holds2 det eM and from [c. B¨uerle and E.e. Exercise 1. 2 = 1 i. the space spanned by only two of the components of Ψ is not invariant under spatial inversion and. 2 . the transformations a(z) and b(z) become interchanged. de Kerf Lie a Algebras.G..f.356 Spinor Formulation ˜ Let P denote the representation of spatial inversion in the space of the wave functions Ψ. However.3.20) ˜ Obviously. does not suffice for particles like the electron which obey inversion symmetry. C) = { M. Obviously. is an element of SU(2). ˜ ˜ P S(w + iϑ) P = S(−w + iϑ) = b(z) 0 0 a(z) . (11.2 Relationship Between the Lie Groups SL(2. However.  (11. (5.22) which follows from the fact that for any complex. Part (Elsevier. the Lorentz invariant equation for neutrinos is only 2–dimensional. for particles like the neutrinos which do not obey inversion symmetry two components of the wave function are sufficient. (11.21) = etr( 2 z·σ) = 1 1 (11.19) i.2. which describes pure rotations. based on the Jordan–Chevalley theorem. In fact.24) = etr(M ) (11. 3 . j = 1. 11.1) We have pointed out that a(iϑ).A. This implies  ˜ Ψ1 ˜  Ψ2 P  ˜  Ψ3 ˜ Ψ4  ˜ Ψ3 ˜   Ψ4  =    Ψ1 ˜ ˜ Ψ2     .1: Show that SL(2. 2. det(M ) = 1 } . C) defined in (11. 1990). For the general case the proof.21) together with matrix multiplication as the binary operation forms a group.224)] tr( σj ) = 0 . ϑ ∈ R3 . can be found in G. The transformation S(z) in the transformed space is then ˜ P 1. hence. M is a complex 2 × 2–matrix.23) Exercise 11.A. The proof of this important property is straightforward in case of hermitian M (see Chapter 5)..10.e.

In fact.1) Mapping Aµ onto matrices M (Aµ ) 357 We want to establish now the relationship between SL(2.31) holds.2: Lie Groups SL(2.26) or from the fact that the matrices σ0 . Noting that for covariant vectors according to (10.31) Exercise 11.  (11. Argue why M (Aµ ) = σµ Aµ provides a bijective map.e. σ3 are hermitian.28) and (11. σ2 . We will prove below [cf. σ2 .28)   σµ =   1 0 0 1 .29) where we used (10. (11. C) and the group L↑ of proper. 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 . M (Aµ ) according to (11.11.25) σ0 σ1 σ2 σ3 The quantity σµ thus defined does not transform like a covariant 4–vector.26) (11.C) and SO(3. Demonstrate that (11. The following important property holds for M (Aµ ) det ( M (Aµ ) ) = Aµ Aµ which follows directly from (11. one can provide a simple expression for the inverse of M (Aµ ) M = M (Aµ ) ↔ Aµ = 1 tr M σµ 2 .135)] this transformation behaviour.75) holds aν = Lν µ aµ one realizes that σµ transforms inversely to covariant 4–vectors. or+ thochronous Lorentz transformations. σ3 provide a linear–independent basis for the space of hermitian 2 × 2–matrices.32) . in a transformed frame should hold Lρ ν σ µ Aµ σµ A µ .2. in fact.25) can also be written M (Aµ ) = A0 + A3 A1 − iA2 A1 + iA2 A0 − A3 .2: Show that σ0 .. 0 −i i 0 . the matrix M (Aµ ) is hermitian as can be seen from inspection of (11. The function M (Aµ ) is bijective. 0 1 1 0 .76). (11. 1 0 0 −1    . one wishes that the definition (11.25) of the matrix M (Aµ ) is independent of the frame of reference. (11. (11.30) Since the components of Aµ are real. (11.27) requires then σ µ Aµ Lν µ σµ = σν (11. i.30).27) −→ Straightforward transformation into another frame of reference would replace σµ by σµ . −→ µ Consistency of (11. σ1 . 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.

any a ∈ SL(2.33) This transformation conserves the hermitian property of M since (M )† = ( a M a† )† = (a† )† M † a† = a M a† = M (11.37) defines actually a Lorentz transformation. In fact.40) 1 tr a σν a† σµ 2 Aν (11. a ∈ SL(2. In fact.33) to M (Aµ ) describing the action of the transformation in terms of transformations of Aµ . The linear character of the transformation becomes apparent expressing A µ as given in (11.37) using (11.36) The suitable A µ can readily be constructed using (11.f.34) where we used the properties M † = M and (a† )† = a. Accordingly. C) defines the transformation [c.31).38) .31)] 1 a Aµ −→ A µ = tr a M (Aµ ) a† σµ 2 Because of (11. 11. C) and for any Aµ there exists an A µ such that M (A µ ) = a M (Aµ ) a† .40) is an element of L↑ . (11. C) .3: Show that L(a)µ ν defined in (11.25) Aµ = which allows us to express finally A µ = L(a)µ ν Aν .358 Transforming the matrices M (Aµ ) Spinor Formulation We define now a transformation of the matrix M (Aµ ) in the space of hermitian 2 × 2–matrices a M −→ M = a M a† . it holds for the matrix M defined through (11.37) which implies that (11.33) conserves the determinant of M .35) We now apply the transformation (11. (11.33) det(M ) = = det( a M a† ) = det(a) det(a† ) det(M ) [det(a)]2 det(M ) = det(M ) . L(a)µ ν = 1 tr a σν a† σµ 2 . (11. for any a ∈ SL(2.2. + . (11.39) Exercise 11. (11. Due to det(a) = 1 the transformation (11.35) holds for this transformation A µ Aµ = Aµ Aµ (11. (11.32.

2 (11.δ 1 4 Aαβγδ Γβα Γγδ α. (11.43) (σν )αβ Γβα Γγδ (σν )δγ = ν.40)] respects the group property of SL(2. β 2. C) and SO(3.e. i.11.α.β γ. (11.1) L(a)µ ν provides a homomorphism 359 We want to demonstrate now that the map between SL(2.C) and SO(3.δ (11.45) One can demonstrate through direct evaluation   2 α =    2 α =  2 α = =   2 α =    0 else β β γ γ = = = = γ = γ = 1.41) results in 1 ¯ Lµ ρ = 4 where Aαβγδ = ν Γ = a2 σρ a† 2 (11. C) and of SO(3.2: Lie Groups SL(2.1).. (11.46) which yields ¯ Lµ ρ = = = 1 1 Γ11 Γ11 + Γ22 Γ22 + Γ12 Γ21 + Γ21 Γ12 = tr ΓΓ 2 2 1 1 tr a† σµ a1 a2 σρ a† = tr σµ a1 a2 σρ a† a† 1 2 2 1 2 2 1 tr a1 a2 σρ (a1 a2 )† σµ = L(a1 a2 )µ ρ .42) = ν Defining Γ = a† σµ a1 .47) This completes the proof of the homomorphic property of L(a)µ ν .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 . 1 ¯ and using the definition of Lµ ρ in (11.C) )µ ρ (11. β δ = δ = = δ = δ 1 2 = 2 = 1 Aαβγδ (11. . ¯ Lµ ρ = L(a1 )µ ν L(a2 )µ ρ = L( product in SO(3.β γ.1) defined through L(a)µ ν [cf.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(2.44) (σν )αβ (σν )δγ .

Aµ = (0. J Spinor Formulation The transformations a ∈ SL(2. 1) − σ1 σ0 0 0 One can verify readily that 1 κ1 = − σ 1 2 (11.49) assuming (note that g µ ν is just the familiar Kronecker δµν ) Lµ ν a = = gµν + 1 + 1 (K1 )µ ν κ1 . (11.57) .49) we obtain using (11. −A0 . denoted below as κ1 . M (Aµ + (K1 )µ ν Aν ) = = M (Aµ ) + M (Aµ ) − M ( (−A1 . C) which correspond to K. 0. For the r.h. 0.. noting the linearity of M (Aµ ).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 (11.52) where we employed (11. 0) ) σ0 A 1 − σ1 A 0 (11.53) Equating (11. C) as complex 2 × 2–matrices are defined through four complex or.48) We want to determine now the generators of the transformation a(z) as defined in (11. Aµ = (0.55) (11. Because of the condition det(a) = 1 only six independent real numbers actually suffice for the definition of a.360 Generators for SL(2.51) Insertion of (K1 )µ ν as given in (10.h. To obtain the generator of a(z) corresponding to the generator K1 . 1 (11. correspond to the generators given by (10. Aµ = (0. 0. 0). 0. 0).47. J3 of the Lorentz transformations Lµ ν in the natural representations.48) yields for the l.49). (11. J2 . 0.51) ( 1 + κ1 ) M (Aµ ) ( 1 + κ† ) 1 1 1 = = M (Aµ ) + M (Aµ ) + ( κ1 M (Aµ ) + M (Aµ )κ† ) + O( 2 ) 1 ( κ1 σµ + σµ κ† ) Aµ + O( 2 ) .s. i. 0). (11. To this end we consider infinitesimal transformations and keep only terms of zero order and first order in the small variables.25) in the last step. 1 (11.48). 10. eight real numbers.36) M (Lµ ν Aν ) = a M (Aµ ) a† (11. of (11. K2 . 1. we write (11. correspondingly.54) This reads for the four cases Aµ = (1. 1.e. 0.s.53) results in the condition σ 0 A1 − σ 1 A0 = ( κ 1 σ µ + σ µ κ † ) Aµ . J1 .58) .2) which correspond to the generators K1 .50) (11. of (11. 0. K3 . 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  .52) and (11.56) (11.

2.3 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.64) . − 1 . C) correspond to the generators K and J of Lµ ν . def α = 1. 2 (11. Exercise 11.3: Spinors 361 obeys these conditions.60) of a ∈ SL(2. Similarly. one can show that the generators κ2 .1) acts on 4–vectors Aµ . We consider first the transformation a(z) which acts on a 2-dimensional space of states denoted by φα = According to our earlier discussion holds φ1 φ2 1 transforms under rotations (z = iϑ) like a spin– 2 state | 1 . Here a(z) = ( aα β ) = describes the matrix a(z). 2 (11. a(w − iϑ) = e− 2 (w − iϑ)·σ ∈ SL(2. the so-called contravariant spinors.62) We denote the general (z = w + iϑ) transformation by φ α = aα β φβ = aα 1 φ1 + aα 2 φ2 .11. λ3 corresponding to J1 .60) We can. 11. hence. K3 and λ1 . J2 . C) acts on spinors φα ∈ C2 (characterized below) 1 (11. state that the following two transformations are equivalent L(w. 2 2 φ1 φ2 ∈ C2 . + 1 2 2 1 transforms under rotations (z = iϑ) like a spin– 2 state | 1 . (11.63) where we extended the summation convention of 4-vectors to spinors.61) This identifies the transformations a(z = w − iϑ) as representations of Lorentz transformations. λ2 . 2 λ = i σ. J3 are given by 1 κ = − σ. a1 1 a1 2 a2 1 a2 2 (11. κ3 of a(z) corresponding to K2 . ϑ) = ew·K + ϑ·J ∈ SO(3. w describing boosts and ϑ describing rotations.4: Show that the generators (11.

. To arrive at a suitable scalar product we consider first only rotational transformations a(iϑ). ± 1 . this scalar product is anti-symmetric. i. 0| 1 .65) where | · · · 1 describes the spin state of “particle 1” and | · · · 2 describes the spin state of “particle 2” and (0. . as we will demonstrate below such states are invariant under general Lorentz transformations a(z). Such a scalar product does. i.69) (11. The corresponding states are defined through φ1 χ2 − φ2 χ1 = φ1 χ1 + φ2 χ2 It obviously holds χα = χ1 χ2 = χ2 −χ1 . as contravariant spinors and to φα .e.. χβ . which 2 can be constructed from products φ1 χ2 . should remain invariant under transformations a(iϑ).67) We will refer to φα . ± 1 .e. The relationship between the two can be expressed φ1 φ2 = = 3 φ1 φ2 0 1 −1 0 (11.. 1 . . 0 = 2 2 m=± 1 2 1 1 (0. exist and it plays a role for spinors which is as central as the role of the scalar product Aµ Aµ is for 4–vectors. exchange of φα and χβ alters the sign of the expression. indeed. Using (0. However. 0. Definition of covariant spinors Expression (11. In fact. . . (11.. etc. invariant and as such has the necessary properties of a scalar3 product.66) should constitute a singlet spin state. χβ . 1 ) stands for the Clebsch–Gordon coefficient. . 0| .70) ‘Scalar’ implies invariance under rotations and is conventionally generalized to invariance under other symmetry trasnformations. −m) | . . In the notation developed in Chapter 5 holds for the singlet state | .e.68) (11. 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. −m 2 2 2 2 1 1 1 1 2 (11. The existence of a scalar product gives rise to the definition of a dual representation of the states φα denoted by φα .. is the singlet state. 1 ) = 2 2 2 2 2 2 2 2 √ ±1/ 2 and equating the spin states of “particle 1” with the spinor φα . m 1 | . In this case spinors φα transform like spin– 1 states and an invariant. m. those of “particle 2” with the spinor χβ one can state that the quantity 1 Σ = √ 2 φ1 χ2 − φ2 χ1 (11. as covariant spinors. . invariant under transformations a(z).66) is a bilinear form. . i. 0| 1 .

72) in a notation analogous to that chosen for contravariant and covariant 4–vectors [cf. −1 connecting contravariant and covariant spinors play the role of the metric tensors gµν . In the transformed frame holds φ α χα = aα β αγ aγ δ δκ φβ χκ . 10.81) .77) (11.68). g µν of the Minkowski space [cf. The inverse of (11.71) (11. 11.67) will be expressed as φα χα = φ1 χ1 + φ2 χ2 = φ1 χ2 − φ2 χ1 . (10.78) aβ γ γδ φδ (11.72) 0 −1 1 0 Exercise 11.71. (11. For this scalar product holds φα χα = −χα φα .76) αβ αβ The scalar product (11.70) is φ1 φ2 −1 363 = = −1 φ1 φ2 . (10.75) is given by φα = as can be readily verified.73) (11.11.72)] φα φ α = = = = αβ αβ φβ φβ 11 21 11 21 12 22 12 22 (11. (11.79) = a −1 T a κβ . Proof that φα χα is Lorentz invariant We want to verify now that the scalar product (11.70) and (11. Accordingly. (11.74) = = 0 1 −1 0 0 −1 1 0 (11.69.75) (11.10.69. 11. we will express (11.73.77) is Lorentz invariant.74)].3. 11.3: Spinors as can be verified from (11.80) One can write in matrix notation aα β γ δκ αγ a δ αβ (11. 11.1: Show that for any non-singular complex 2 × 2–matrix M holds M −1 = det(M ) M −1 T The matrices . The transformation behaviour of φα according to (11. 11.63.

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. As discussed above. Insertion of (11.84) (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 (11.2) and (11. (11.83) into (11. The transformation ˙ behaviour of φα is ˙ φ α = (aα β )∗ φβ ˙ (11. a∗ (z) provides a representation of the Lorentz group which is distinct from that provided by a(z).80) φ α χα = φβ χβ . We denote ˙ (aα β )∗ = aα β ˙ (11. ˙ (11.85) φ1 φ2 ∗ ∗ .81) yields aα β and.83) Here we have employed the hermitian property of σ. σ † = (σ ∗ )T = σ.14) one can write a −1 Spinor Formulation = e 2 z·σ 1 −1 = e 2 z· 1 σ −1 = e− 2 z·σ 1 ∗ (11.89) such that (11.88) which one verifies taking the conjugate complex of (11. ˙ We also define covariant versions of φα φα = (φα )∗ . Obviously.90) extending the summation convention to ‘dotted’ indices.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 .88) reads ˙ ˙ φ α = aα β φβ ˙ ˙ (11.364 Using (11. k = 1.91) .87) which we will employ from now on. it holds φk = (φk )∗ . hence.e. Hence. i. from (11.. 2.63).

99)  Ψ3 (xµ )   χ ˙ (xµ )  ˜ χβ (xµ ) ˙ 1 ˜ χ2 (xµ ) Ψ4 (xµ ) ˙ 11. is governed by the operator a∗ (z) −1 which arises in ˜ the Lorentz transformation (11.98) ˜ ∗ (z) −1 Ψ3 a ˜ Ψ 4 obviously implies that the solution of the Dirac equation in the chiral representation can be written in spinor form  ˜   1 µ  Ψ1 (xµ ) φ (x ) ˜  Ψ2 (xµ )   φ2 (xµ )  φα (xµ )   =   = .76) φα ˙ ˙ φα αβ ˙ ˙ αβ ˙ ˙ = = = = αβ ˙ ˙ αβ ˙ ˙ αβ αβ φβ φβ ˙ ˙ (11. a property which is rather evident.95) ˙ where αβ and αβ are the real matrices defined in (11.93) (11.18) ˜ 3 . Ψ4 .11.96) is Lorentz invariant.94) (11. a∗ (z) −1 accounting for the ˜ transformation behaviour of the third and fourth spinor component of Ψ.4: Spinor Tensors 365 The relationship between contravariant and covariant conjugate complex spinors can be expressed in analogy to (11.75. 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 =  (11.100) . − 1 . (11. The transformation behaviour of φα is ˙ φα = ˙ ˙ β γδ ˙˙ φδ ˙ ˙ αβ a γ ˙ ˙ .18) of the bispinor wave function Ψ. A comparision of (11. 11.4 Spinor Tensors ˙ ˙ ˙ We generalize now our definition of spinors φα to tensors. i. (11.73. A tensor tα1 α2 ···αk β1 β2 ···β (11.92) (11. 11. in matrix notation. one can state ˜ and (11.97) implies then that φα transforms like Ψ ˙ φ1 ˙ φ2 ˙ transforms under rotations (z = iϑ) like a spin– 1 state | 1 . + 1 2 2 2 1 transforms under rotations (z = iϑ) like a spin– 2 state | 1 .75).e..97) The transformation. For the spinors φα and χα thus ˙ defined holds that the scalar product ˙ φα χα = φ1 χ1 + φ2 χ2 ˙ ˙ ˙ ˙ ˙ (11.

the transformation bevaviour of a tensor tαβ reads in spinor and matrix notation t αβ = aαγ aβδ tγδ . using conventional matrix indices j. k.e. (11. (11.103) Similarly. i. 11.105) This tensor is actually invariant under Lorentz transforma= αβ .104) Indices on tensors can also be lowered employing the formula tα β = and generalizations thereof.. According to (11. This task implies that we seek a tensor.e.1: Prove equation (11.101) An example is the tensor tαβ which will play an important role in the spinor presentation of the Dirac equation.106) Exercise 11. The 4–vector Aµ in spinor form We want to provide now the spinor form of the 4-vector Aµ . m.m aj akm t m (11. we want to express Aµ through a spinor tensor... this transformation behaviour is in harmony with the tensor notation adopted.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 . i. An example of a tensor is αβ and tions. An obvious candidate is [cf.25] M (Aµ ) = σµ Aµ .107) which reads in spinor notation [cf.108) Obviously. tjk = a t a† jk ˙ ˙ ˙ ˙ (11. i.102. it holds αβ . the elements of which are linear functions of Aµ . (11. This tensor transforms according to t αβ = aα γ aβ δ tγ δ ˙ This reads in matrix notation. αβ = αβ (11.103) A αβ = aα γ aβ δ Aγ δ .4.25) the tensor is explicitly Aαβ = ˙ A1 1 A1 2 ˙ ˙ A2 1 A2 2 ˙ ˙ = A0 + A3 A1 − iA2 A1 + iA2 A0 − A3 .m aj a∗ t km m . . ˙ ˙ ˙ ˙ (11.102) = . ˙ ˙ ˙ ˙ (11. αβ ˙ αγ t ˙ γβ (11.109) . We had demonstrated that M (Aµ ) transforms according to M = M (Lµ ν Aν ) = a M (Aµ ) a† (11. tjk = a t aT jk = .106). (11. with ˙ contravariant indices αβ.e.

11.110) The 4-vectors Aµ .117) ˙ ˙ ∂0 − ∂3 −∂1 + i∂2 −∂1 − i∂2 ∂0 + ∂3 .109.110) one can state ∂ αβ = Similarly.116) . employing (11.114) We finally note that the 4–vector scalar product Aµ Bµ reads in spinor notation Aµ B µ = 1 αβ ˙ A B αβ . (11. ˙ 2 (11.2: Prove that (11. (11. (11.115) is correct.4: Spinor Tensors One can express Aαβ also through Aµ Aα β = ˙ ˙ 367 A0 − A3 −A1 + iA2 −A1 − iA2 A0 + A3 .115) Exercise 11. (11. ∂µ in spinor notation The relationship between 4–vectors Aµ . 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 .112) A22 −A21 ˙ ˙ −A12 A1 1 .111) A11 A12 ˙ ˙ A21 A22 ˙ ˙ ˙ 0 −1 1 0 (11. (11. Hence.113) Aα β = ˙ (11. (11.114) yields ∂αβ = ˙ ∂0 + ∂3 ∂1 + i∂2 ∂1 − i∂2 ∂0 − ∂3 .110) one obtains Aα β = ˙ A0 − A3 −A1 − iA2 −A1 + iA2 A0 + A3 A0 + A3 A1 + iA2 A1 − iA2 A0 − A3 . 11. Aµ and tensors tαβ can be applied to the partial differential operator ∂µ . Using (11.4.

−1 0 0 1 (11.368 σµ in Tensor Notation Spinor Formulation We want to develop now the tensor notation for σµ (11.120) σµ 0 . This is ˜ possible since γ µ can be expressed through σ µ .124). (11. . 0 1 1 0 .243) holds independently of the representation chosen. 1 0 0 −1 . 0 −i i 0 . Comparision of (10.125) a (σ ) Equation (11.. hence.15) one can write σµ = and.126) which is the complex conjugate of (11. σµ building on the known transformation behaviour of γ µ . To obtain the transformation properties of σ µ we employ then (10. i. (11. (11.e.s.121). Equation (10.18) and γ µ by (11. one can conclude aσ µ ∗ −1 Lν µ σν 0 = . i. (11.124) .243)].118) yields ˜ γµ = ˜ Using (11.243) reads then a 0 0 a∗ −1 0 (σ µ )∗ −1 σµ 0 a−1 0 0 a∗ −1 0 (σ ν )∗ −1 The l.26) and its contravariant analogue σ µ σµ = σµ = 1 0 0 1 1 0 0 1 . . (11.h. (10.f.121) One expects then that the transformation properties of σ µ should follow from the transformation properties established already for γ µ [c. γµ = ˜ 0 µ )∗ (σ −1 0 σµ σµ 0 (σ µ )∗ −1 . σµ .119) (11. Hence.118) 0 −1 −1 0 0 i −i 0 For this purpose we consider first the transformation behaviour of σ µ and σµ .123) (a∗ )−1 Lν µ a L ν µ = = σν (σ ) ν ∗ −1 (11.229) and (11.125) is equivalent to µ ∗ −1 −1 a∗ (σ µ )∗ −1 −1 a Lν µ = (σ ν )∗ (11. holds also in the chiral representation. hence.e.124) constitutes the essential transformation property of σ µ and will be considered further. of this equation is 0 ∗ (σ µ )∗ a and. Note that (10.125) is equivalent to (11.243) in the chiral representation expressing S(Lη ξ ) by (11. We will obtain the transformation behaviour of σ µ .124). . (11..122) aσ µ −1 a−1 −1 (a∗ )−1 0 Lν µ (11.

11. 3) σj −→ a(iϑ) σj a(iϑ) † = a(iϑ) σj a∗ (iϑ) T . In fact. state that σ µ in a new reference frame is σ µ = a σ µ a† (11. (11. 2. 11..107) to (11. 11.124) a σ µ [a∗ ]T Lν µ = σ ν .131) where a is given by (11.16. 0 1 1 0 µ=1 .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 ˙ ˙ ˙ ˙ =  (11.132) 1 0 0 1 µ=0 .128) One can express.   =  . One can. i. e. This transformation behaviour.124) using (11. 0 −1 −1 0 µ=1 . (11.4) with w = 0.11.133) Combining (11. (σ ∗ )T = σ yields using (11.2) −1 (a∗ )−1 = e2z 1 ∗ ·(σ ∗ )T = e2z 1 ∗ ·σ ∗ T = [a∗ ]T .127) Exploiting the hermitian property of σ. −1 0 0 1 µ=3   .2.e. (11.g.4).129) We want to demonstrate now that the expression aσ µ [a∗ ]T is to be interpreted as the transform of σ µ under Lorentz transformations.2.134) .108) that the same transformation behaviour applies then for general Lorentz transformations. 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 . according to (11. (11.103) ˙ identifies σ µ as a tensor of type tαβ .  (11. transformations (11. (11. It holds according to (11. 11. 0 i −i 0 µ=2 .. 0 −i i 0 µ=2 .129.2) −1 369 (a∗ )−1 = −1 − 1 z ∗ ·σ ∗ 2 e = e− 2 z 1 ∗ · −1 σ ∗ = e2z 1 ∗ ·σ . hence. 11.130) We argue in analogy to the logic applied in going from (11.  . therefore. equation (11.102. under rotations the Pauli matrices transform like (j = 1.4: Spinor Tensors One can rewrite (11.

The ˙ desired change of representation(σµ )αβ −→ (σµ )αβ corresponds to a transformation of the basis of ˙ spin states f g f −→ = (11.132) and (11.e. α. i.141) σ αβ = σ0 − σ3 σ1 + iσ2 −σ1 − iσ2 σ0 + σ3  = 2  0 0 0 1 0 −1 0 0 .137) σαβ ˙ = δαγ δβ δ . account for the 4–vector index µ.114). i. 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 .138) the ‘inner’ covariant spinor indices. We want to express now σ µ and σµ also with respect to the 4–vector index µ in spinor form employing (11.370 Inverting contravariant and covariant indices one can also state Lν µ σµ = σν . account for the 4-vector µ.133) is characterized through a 4-vector ˙ index µ. whereas ˙ account for the elements of the individual Pauli the ‘outer’ contravariant spinor indices.29)]. =  σαβ ˙  ˙ ˙ ˙ ˙ (σ21 )11 (σ21 )12 (σ22 )11 (σ22 )12   ˙ ˙ ˙ ˙ ˙ ˙ ˙ ˙ 21 (σ )22 21 (σ )22 (σ21 ) (σ22 ) ˙ ˙ ˙ ˙ 21 22 Equating this with the r. σµ in which the contravariant indices are moved ‘inside’.224) ˙ Using (11. 3 as well as through two spinor indices αβ.110) to express σ αβ in terms 0 −1 1 0 0 0 0 1     (11. ˙ ˙ In (11. We can.135) This is the property surmised already above [cf. β. (11. µ = 0. ˙˙ (11. i..e. Each of the 4 × 4 = 16 matrix elements in (11. We can summarize that σ µ and σµ transform like a 4–vector. γ.s.138) Note that all elements of σαβ are real and that there are only four non-vanishing elements.12) for of σµ yields together with (5.e..136) results in the succinct expression 1 2 ˙ γδ ˙ (11. however. in fact. 1. the transformation is inverse to that of ordinary 4–vectors. ˙ ˙ 0 −1 1 0 ˙ (11. hence. and the covariant indices are moved outside.h. 2. Spinor Formulation (11. of (11. the submatrices correspond to tαβ spinors. δ.136) σαβ = = 2 ˙   σ1 − iσ2 σ0 − σ3 0 0 0 0 1 0 0 1 where on the rhs. This yields   1 0 0 1   0 0 0 0 σ0 + σ3 σ1 + iσ2  (11.140) 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 .139) g −f g and. matrices.

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 . employing then transformation (11. (11.144) results in the simple relationship γ µ Aµ = ˜ 0 Aα β ˙ Aα β 0 ˙ . (11. ˙˙ (11.138) one can conclude that the following property holds 1 2 ˙ γδ σ αβ ˙ = 1 2 σ αβ ˙ ˙ γδ = δαγ δβ δ . This expression implies that the element of γ µ given by σ µ −1 is in the expression (11.145) Let Aµ be a covariant 4–vector. we write ˜ ˙ γµ = ˜ 0 (σ µ )αβ µ ) )∗ ((σ αβ 0 ˙ ˙ .143) Combined with (11.  ˙ ˙ ˙ ˙       (11.147) . For this purpose we start from the ˜ µ .144) The Dirac Matrices γ µ in spinor notation We want to express now the Dirac matrices γ µ in spinor form. One can write then the scalar product using (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 ˙ ˙    .11.146) Exploiting the property (11.121) of γ ˜ ˜ ˙ basis |αβ whereas the element of γ µ given by σ µ is in the basis |αβ .4: Spinor Tensors 371 and.140) to transform each of the four submatrices.142) = This can be expressed 1 2 σ αβ ˙ ˙ γδ = δαγ δβ δ . (11. which in ˙ (11. ˙˙ (11. Accordingly.

(11.372 Spinor Formulation 11.5 Lorentz Invariant Field Equations in Spinor Form Dirac Equation (11.149) (11.226) ˜ i γ µ ∂µ Ψ(xµ ) = 0 i ∂ αβ ˙ i ∂ αβ 0 ˙ ˜ ˜ Ψ(xµ ) = m Ψ(xµ ) .147) allows us to rewrite the Dirac equation in the chiral representation (10.150) . ˙ (11.99) yields the Dirac equation in spinor form i ∂ αβ χβ ˙ i ∂ αβ φ ˙ The simplicity of this equation is striking.148) ˜ Employing Ψ(xµ ) in the form (11. α ˙ = = m φα m χβ .

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

. V (r). Sk . Classically. Jk ] = 0 . For example. as represented by an element of SO(3). The energy levels are totally independent of l. ψE. .. can be identified with the angular momentum operators.. we first consider a particular example of the implications of symmetry. 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 . This degeneracy follows from the fact that any rotation.λ1 . 2. this implies a degeneracy in the spectrum.λn . which yields a set of eigenstates of the form ψE. as studied in chapter 7. φ)... each energy level is (2l + 1)-fold degenerate. k = 1. · · · .24).. (12. namely motion in a spherically symmetric potential described by the group SO(3) (or its double covering SU (2) as discussed in section 5. the states 3s. In chapter 7 the dynamics of a particle moving in three dimensions under the influence of a spherically symmetric potential. l. m r (12. . . l and the corresponding energy levels are independent of m.7) 2 n where the orbital angular momentum.. Therefore. (12.12). 2m r (12. . leaves the energy of the state invariant H exp(iαk Sk ) ψE.8) ˆ [H. In order to illustrate this. .39). (7.λn = E exp(iαk Sk ) ψE.m (r) = vE. We will investigate this shortly in more detail in the case of systems with spherical symmetry..18). Jk .4) The stationary Schr¨dinger equation. The energy levels are mk 2 En = − 2 2 . (12.λ1 .. Presence of additional symmetries may further increase the degeneracy of the system.3) If the newly obtained state is linearly independent of the original one. on an energy eigenstate. Lie groups play an essential role in the discussion of mass degeneracies in particle physics. has been discussed. Sk .λ1 . 3p and 3d all have the same energy.m (r) Y m (θ.5) with m = −l.λn .374 Spinor Formulation The action of any symmetry generator. (7.6) From elementary quantum mechanics we know the spectrum of the hydrogen atom.8) . is allowed to take values in 0. generates a state which has the same energy as the original one. 3 . . 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. n − 1. can then be reduced to a one-dimensional radial o equation. where the symmetry generators.. As an example of this we will consider the Coulomb problem with the Hamiltonian H = p2 k − . (7... (12. The spherical symmetry implies the commutation of the Hamiltonian with angular momentum operators (7. .

16) Through some algebra [4] the following commutation relations can be verified [Ki .11) (12. which have a negative energy E. Jj ] = i [Ki . 2 which can be shown to satisfy [Ai . in the subspace of the Hilbert space corresponding to a certain energy we can replace H by E. 2m r 1 x3 (J2 p1 − J1 p2 + i p3 ) − k .9) (12. 2m r 1 x1 (J3 p2 − J2 p3 + i p1 ) − k . Kj ] = i ijk Kk .21) . (12. Aj ] = i ijk Ak . (12. We will also need the following identity [4] 2 = 2H J2 + m 2 + k2 . ijk Jk .12) (12.3.20) (12. which is valid even when a and b do not commute. The corresponding quantum mechanical hermitian operator can be defined by 1 1 2 3 = = = = x1 1 (p2 J3 − p3 J2 + J3 p2 − J2 p3 ) − k . we will compute the hydrogen spectrum using the additional symmetry.19) (12. The vector in (12.17) (12. For this purpose we first note that J · = 0. 2 1 B = (J − K). We introduce the following new operators 1 A = (J + K). (12. 2m r (12. In order to understand the aforementioned extra degeneracy.11. 2E (12. (12.14) This follows from a · (a × b) = (a × b) · a = 0.18) which complement the familiar angular momentum algebra of section 5.15) which can be proved after very considerable algebra. 2m r 1 x2 (J1 p3 − J3 p1 + i p2 ) − k .8) is not a hermitian operator.10) (12.13) It can be verified explicitly that its components commute with the Hamiltonian.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. In the following we consider the bound states. Now we scale the eccentricity vector as follows K= − m . Therefore.

Chaos is described classically as exponential . the energy eigenvalues are found to be E=− mk 2 1 . can be seen to follow from the triangle inequality as applied to J = A + B. . Bj ] = i [Ai . Noticing that A2 and B 2 have the same eigenvalues because of (12. .7) tells us that (2a + 1) = n. which is identical to the the direct sum of the Lie algebra of of two SO(3) (or SU (2)) algebras. . .24) (12. . 2 (2a + 1)2 2 2 (12. b = 0. This illustrates the effect of additional symmetries to the degeneracy structure of a quantum mechanical system. . l. By comparing to the rotation algebra introduced in chapter 5. In contrast to the discussion above about extra symmetries. .376 [Bi . . 1.28) . . The most extreme case of this is the quantum analogue of a classically chaotic system.23) (12.25) So far we have shown that the symmetry generators form an algebra. . namely that J J > < A − B =0 A + B =2 A (12. A. H = 0.31) where we have used (12. we can read off the eigenvalues of A2 and B 2 from (12.29) (12. 1. = 0.28). Furthermore the bound on the orbital angular momentum.15). . Bj ] = 0. .22) (12.32) A comparison with (12. a = 0. 1.33) (12.30) 2 2 2 1 . . 2 (12.22): A2 = a(a + 1) B 2 = b(b + 1) Following (12. Using this equation the energy eigenvalues can be written in terms of the eigenvalues of A2 and B 2 operators. 2 1 . ijk Bk .14) we note that A2 − B 2 = J · = 0.34) It follows that l has to have values in {0 = |a − b|. This implies that a = b. . Spinor Formulation (12. .27) (12. (12. .26) (12. . a = 0. a lack of symmetry implies a lack of degeneracy in the energy levels of a quantum mechanical system. H B. a + b = n − 1}.21) and (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. . . 1.

This enables us to utilize what we already know about the SU (2) symmetry group from the study of angular momentum.57M eV /c2 . for example. Originally it was believed that isospin was an exact symmetry of strong interactions and that it was violated by electromagnetic and weak interactions. that the proton and the neutron behave identically under the so-called strong interactions and that their difference is solely in their charge content. n= 0 1 . after the discovery of the neutron in 1932. which is a particle in box problem in two dimensions with a boundary which can be chosen arbitrarily.) If the proton and the neutron are to be viewed as two linearly independent states of the same particle.11. This (approximate) degeneracy led into the idea of the existence of an (approximate) symmetry obeyed by the underlying nuclear interactions. In the next section we will proceed with the discussion of a symmetry.5: Lorentz Invariant Field Equations 377 sensitivity to initial conditions. in the sense that nearby trajectories in the phase space diverge from each other over time. suggested that the proton and the neutron can be regarded as two states of a single particle. namely. the classical trajectories will diverge from each other after successive bounces from the boundary. This is known as level repulsion. (Strong interactions bind the atomic nucleus together. (12. 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. A typical example of this so called quantum chaos is the quantum billiard problem. it is natural to represent them in terms of a two component vector. analogous to the 1 spin-up and spin-down states of a spin. For a more detailed discussion of quantum chaos in billiard systems we refer the reader to [7] and the references therein.36) into each other in abstract isospin space. For example. It is important to place the isospin concept in its proper historical context.35) this mass equivalence can be viewed as an energy degeneracy of the underlying interactions. 12. which was discovered by observing degeneracies in the particle spectrum.36) In analogy to the concept of spin regarding the rotations in 3-space as discussed in chapter 5.2 Isospin and the SU (2) flavor symmetry The concept of isospin goes back to Heisenberg. the isospin symmetry is also governed by an SU (2) group ‘rotating’ components in (12. (Weak interactions are responsible.2 system p= 1 0 . However. Following the mass-energy equivalence of special relativity E = mc2 . another manifestation of chaos is the lack of independent operators commuting with the Hamiltonian. for the . This was motivated by the observation that their masses are approximately equal: mp = 938. (12. mp = 939. 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 6.28M eV /c2 . If the chosen boundary is ‘irregular’ in a suitably defined sense. who.

1 π 0 = |1. 2 2 1 2 2 2 (12. as good quantum numbers. and its third component.36) as a multiplet with I = 2 1 1 p = I = . Isospin symmetry is not an exact symmetry of strong interactions. They u form an isotriplet: π + = |1. We can describe isospin multiplets the same way we have described the angular momentum and spin multiplets. 2.− 2 2 1 2 2 2 1 1 1 1 . I3 . as we shall see below. (mπ± = 139. 2 2 1 2 2 2 1 1 1 1 1 √ .6M eV /c2 . . 2 2 . For example. we refer the reader to [1]. 0 π − = |1. respectively. In the framework of the quark of model. . I3 = 2 2 1 1 . albeit it is a good approximate one. −1 . 2 2 1 1 1 .− 2 2 . sec. If it were otherwise the proton would be unstable by decaying into the neutron.40) + 2 1 1 . −1 = = = 1 1 1 1 .39) All other isomultiplets. it can be seen as part of a larger (and more approximate) symmetry which is of great utility to classify observed particle families.− . π 0 = |1. the proton and the neutron can be written as totally symmetric uud and udd states. 1 .0M eV /c2 ). Shortly we 1 will see how to describe both the pion and nucleon states as composites of more fundamental I = 2 states. If the mass difference (or the energy difference) were to be to be purely electrostatic in nature. including the proton and the neutron. . n = I = . spelling disaster for the stability of matter. u¯ and d¯ states. Denoting the total isospin. (12. mπ0 = 135.− 2 2 . d= 1 1 . Further than that. (12. 1 we can re-write (12. Therefore it remains a useful concept. .378 Spinor Formulation beta decay). They can be constructed with the same rules that have been used for angular momentum addition ¯ u in chapter 6.43) Similarly.38) are ud. 0 . The mass difference between the neutron and the proton could then be attributed to the charge content of the latter. the fundamental representation of the isospin symmetry corresponds to the doublet that contains the so-called up and down quarks u= 1 1 . are made up of these two quarks.11. the proton is the lighter of the two. the proton had to be heavier. 2 (12. I.37) As an example of a multiplet with I = 1 we have the three pions or π-mesons π + = |1. including the spin and color quantum numbers of their constituent quarks. the three pions in (12.41) (12.− . π − = |1.38) which have all nearly identical masses. (12.42) (12. However. For a precise description of the two nucleons as composite states. I3 = − 2 2 .

Therefore the deuteron has to be an isosinglet state (|0. 0 ). The value of the strangeness quantum number is taken. Following (12. baryons generally are heavy. By convention ¯ baryons have baryon number 1. but such states do not exist.37) and in analogy to (12. The names. All other composite states have their strangeness given by the sum of the strangeness content of their constituents.) are light. The possibilities are that of an isosinglet 1 |0. 1 |1. Naturally. as some mesons discovered later are heavier than some baryons and so on. and quarks have baryon number 1 . Now let us consider another example of combining the isospins of two particles. Therefore it has to have isospin. mesons have intermediate mass ranges. the pions being examples thereof. . the neutrinos etc. strongly interacting particles) which is about a GeV /c2 . (12.or flavors as they are commonly called. the up and down quarks are not the only quark ‘species’ . 0 = √ (|p > |n > −|n > |p >) 2 and that of an isotriplet |1. a hydrogen isotope.44) 2 where B is the baryon number and S is the strangeness. this will be mathematically identical to adding two spins. All antiparticles have their 3 quantum numbers reversed. The reader may know that the deuteron. whereas mesons are q q states. a number strange particles have been found which presumably contained a third quark species: the strange quark. 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. 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). This is why isospin is such a good symmetry and why isomultiplets have nearly identical masses. by accidental convention. In the late 1940’s and early 1950’s. such as proton and the neutron. originally refer to the relative weight of particles.48) (12. The up and down quarks have strangeness zero. 1 = √ (|p > |n > +|n > |p >). −1 = |p > |p >.11. muon. where leptons (electron. but at the time the empirical concepts like isospin and strangeness quantum numbers were in use. As it later turned out. I3 = 0. 2 = |n > |n > . we shall setup some terminology: baryons are qqq states. consists of a proton and a neutron. this remains only an inaccurate naming convention today.45) Is the deuteron an isosinglet state or an isotriplet? If it were an isotriplet (|1.46) (12. 0 ) we should have seen nn and pp bound states of comparable energy in nature (because of isospin symmetry). It shall be noted here that the quark model was not invented until 1960’s.43). mesons have baryon number 0. Before proceeding further. to be −1 for the strange quark. We will now try to describe its wave function in terms of its constituent nucleons.47) (12. (12. 0 |1.5: 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.e. baryon and meson. If taken literally.

8. Show that I(1) ·I(2) is an ‘isotensor’ of rank zero.48) the initial states in (I) and (II) have isospin values |1. (f ) (i) (I (f ) I3 |00I (i) I3 ) is a Clebsch-Gordon coefficient and I (f ) .380 Spinor Formulation As an exercise on the implications of isospin symmetry we will consider nucleon-nucleon scattering.45) and (12. The dot product here refers to the abstract isospin space.7 and 6.38) and the fact that the deuteron is an isosinglet we know that the isospins of the final states in (I) and (II) are |1. 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. . Refer to exercise 6. γ (i) I (f ) .7. 0 + |0.5 for the spin-analogue of the same problem. 1 and |1. γ (i) = (f ) (i) (I (f ) I3 |00I (i) I3 ) (f ) (i) (−1)I −I (f ) (i) (12. I3 = 1.7 to the case of isospin. 0 ).53) 0 √ 1 2I (i) +1 I (f ) . such as the spatial dependence of the wave function and spin. γ (i) is a reduced matrix element defined in the same sense as in section 6. σ. γ (i) (i) (12. (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 (f ) |T00 |I (i) I3 . is proportional to |M|2 . For completeness let us start by restating the Wigner-Eckart theorem (6. 1 4 in an isotriplet state and − 3 in an isosinglet 4 Using (12. γ (f ) T00 I (i) = 1. γ (f ) ||T0 ||I (i) .259) discussed in detail in sections 6.50) . Show that the expectation of I(1) · I(2) is state. We will eventually be able to compute ratios of scattering cross-sections between different processes. γ (f ) ||T ||I (i) . The cross-section.52) where γ (i) and γ (f ) denote degrees of freedom other than isospin. which is an isoscalar. We will now employ the (isospin analogue) of the Wigner-Eckart theorem (6. γ (i) . respectively. as discussed in the exercise above. γ (f ) (f ) (i) (12. consider (I) p + p → d + π + (II) p + n → d + π 0 (12. 1 and (1/ 2)(|1. where M is the scattering amplitude given by M = final | α I(1) · I(2) | initial .8 to compute the ratio of the scattering amplitudes MI and MII .51) (12. I3 .55) . Exercise. √ According to (12. 0 .54) Here T00 ≡ V = α I(1) · I(2) . Consider the generalization of tensor operators discussed in section 6. Exercise. (12.259) in the present context: I (f ) I3 . I3 (f ) = 1. I3 . For example. The initial and final states can be denoted in more detail as | initial | final = = I (i) .

γ (i) 2 1 = (10|0010) √ I (f ) = 1. −2 . −2 1 1 2. for example. γ (f ) T00 I (i) = 1. which are obtained by a standard Clebsch-Gordan expansion π + + p : |1.62) We can now write the ratio of the scattering amplitudes: MI 1 √ . σ( π − + p → anything ) The possibilities are (a) π + + p → π + + p . The relevant Clebsch-Gordon coefficients are easily evaluated: (11|0011) = (10|0010) = 1. γ (f ) ||T00 || I (i) = 1. (12. (b) π − + p → π − + p . involving. 1 π − + p : |1. I3 = 0. γ (f ) T00 I (i) = 0.66) = 2 3 1 3 2.60) (12. I3 = 0. [2] As a further example. we will consider pion-nucleon scattering.65) (12. γ (f ) ||T00 || I (i) = 1. γ (i) 3 1 (f ) (i) I (f ) = 1.61) MII Note that the second term in MII vanishes due to the isospin conservation. I3 = 0. We want to compute the ratio of total cross-sections assuming a similar interaction as in the previous example σ( π + + p → anything ) . (c) π − + p → π 0 + n . 2 = 3 2. 2 . γ (i) = √ 2 1 (f ) (i) +√ I (f ) = 1. = MII (1/ 2) (12. which is also manifested by a vanishing Clebsch-Gordon prefactor. −2 2 1 1 3 2. −1 2 3 2 − + . (12.59) (12.56) (12. −2 1 1 √ . γ (i) 3 +0. −1 π0 + n : |1.11. It follows σI = 2. . 1 1 1 1 3 √ 2. but these are not dominant at relatively low energies. σII which is in approximate agreement with the observed ratio.64) There are more exotic possibilities.63) where common dynamical factors (which would not be as easy to compute) have dropped out thanks to the Wigner-Eckart theorem. 0 1 1 3 3 2. I3 = 0.5: Lorentz Invariant Field Equations 1 (11|0011) √ I (f ) = 1. Once again we need the isospins for the initial and final states. 2 = 2.58) (12. particles with strangeness.57) (12. = 381 (12. (12.

are up to 40% heavier. . 0 0 1 . for example. S. while weak interactions did not. m . S(Σ− ) = −1 and S(π) = S(N ) = 0. [2] (12. Hence the strangeness changing strong decays of K + (or Σ− ) were forbidden. such that S(K + ) = 1. however. m 1 2.s =  0  . called strangeness.67) which are independent of m. mentioned in the previous section. giving it a higher than usual lifetime.71) u =  0  . which leads to the following 2 2 ratios for the cross-sections σa : σb : σc = 9 : 1 : 2.382 Spinor Formulation As in the example of nucleon-meson scattering we define the relevant matrix elements M3 = 2 M1 = 2 3 2. A computation similar to the previous example of nucleon-nucleon scattering yields (apart from common prefactors) the following amplitudes for the reactions in (12. was motivated by the existence of particles that are produced strongly but decay only weakly.65) Ma = M 3 2 Mb = 1 M 3 + 2 M 1 3 3 √ 2 Mc = 2 3 M3 2 √ 2 (12.3 The Eightfold Way and the flavor SU (3) symmetry The discovery of the concept of strangeness. etc. K + . which can be produced by π − + p → K + + Σ− . m V V 3 2.69) 12. Hence Gell-Mann and independently Nishijima postulated the existence of a separate quantum number. has a lifetime which is comparable to that of π + albeit being more than three times heavier. For instance.d =  1  .68) − 2 3 M1 2 Guided by empirical data we will further assume that M 3 >> M 1 . (12. The classification of the newly found particles as members of some higher multiplet structure was less obvious then the case of isospin. making an identification of the underlying symmetry and the multiplet structure less straightforward. Strange partners of the familiar nucleons. It was assumed that strong interactions conserved S (at least approximately). In the light of the quark model.70) (12. m 1 2. 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. it appears an obvious generalization to add another component for an extra quark to the isospin vector space       1 0 0 (12.

In this perspective the flavor SU (3) symmetry may appear to be mainly of historical interest. for a period they were considered as useful bookkeeping devices without physical content. have to be elements of the group SU (3) (which we will investigate closely very soon). (12. while preserving the norm.3 for a brief discussion of gauge transformations). In many respects it will resemble the more familiar group SU (2) discussed in some detail in chapter 5. Flavor SU (Nf ) symmetry on the other hand. This is believed to be an exact symmetry of strong interactions.73) To verify that all such matrices form a group. we will establish some mathematical preliminaries about the group SU (3). as in (12. which again form representations of an SU (3) group. we want to find those transformations ak → Ukl al that preserve the norm. of a. However. Lie algebras and related concepts. It is seen that such a matrix U has to satisfy the following k U † = U −1 . but we will stick to N = 3 in the following. Since arbitrary phase factors are of no physical interest.5: Lorentz Invariant Field Equations 383 In this case the transformations that ‘rotate’ the components of (12. unitary relation k (12.71) into each other. First particle multiplets were identified as representations of the SU (3) group. where Nf is the number quark flavors. should correspond to. still leaves the norm invariant. it is the group . As quarks were never directly observed. (The reader is referred to section 8.71).74) for any two unitary matrices U and V . 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. Then came the question of what the fundamental representation. of three dimensions. Multiplying U by a phase. However SU (3) symmetry appears in another and much more fundamental context in strong interaction physics. but there are a number of subtle differences. Given a complex vector. bottom (beauty) and top (truth) are significantly heavier than the hadronic energy scale. giving rise to the quark model. we observe that (U V )† = V † U † = V −1 U −1 = (U V )−1 . in reference to Lie groups. history followed the reverse of this path. eiφ . becomes increasingly inaccurate for Nf > 3. Hence the group U (3) has 9 dimensions. called quantum chromodynamics. The quarks posses another quantum number. the same way nucleons and pions were identified as representations of the isospin SU (2) symmetry. Because of this additional constraint SU (3) has 8 dimensions.11. which set the hadronic energy scale. (The ‘bare’ mass of the charm quark is already heavier than the two nucleons. (12.1 whenever necessary.72) a∗ ak . The reason is that the other known quarks. in fact modern theory of strong interactions is a ‘gauge theory’ of this color group. called color. The reader shall note that most of what is being said trivially generalizes to other unitary groups. [2] ) Before discussing the significance and the physical implications of the quark model. namely charm. The reader is also invited to revisit section 5. ak . This group of 3 × 3 unitary matrices is denoted by U (3). The unitarity relation imposes 9 constraints on the total of 18 real degrees of freedom of a 3 × 3 complex matrix. The bottom and top are even heavier. SU (N ).

.) The unit determinant condition requires that all λk are traceless.79). any unitary matrix. δjkl .79) (12. Therefore we will express elements of SU (3) as U = ei k (12.76) where λk are 8 linearly independent matrices forming the basis of the Lie algebra of SU (3). −2 λ4 λ6 λ8 The generators.77)  λ1  . 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  . As in the case of SU (2) higher dimensional representations obeying the same structure can be found. U . can be written in the form U = eiH where H is a hermitian matrix.78) where fjkl are the antisymmetric structure constants similar to the jkl of SU (2) given in (5. λk . λ7 =   0 0  . since det(eA ) = etrA .75) αk λk (12. λ3 =  0 −1 0  . We can also introduce the constants. As discussed in section 5. The fundamental relation to be preserved is (12. via the anticommutator relation 4 [λj . 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. regardless of the dimension of the representation.7.1.32). λk ]+ = δjk + 2δjkl λl . (12. λk ] = 2ifjkl λl . 0 0 0   . which can be verified explicitly as matrix identities. (12. (We shall at times refer to the Lie algebra with the name of the group. the meaning being apparent from the context.80) This is the fundamental or defining representation of SU (3).   . The Lie algebra structure is given by the commutators of λk [λj . λ2 =     . λ5 =   . 3 (12. obey the following relation tr(λj λk ) = 2δjk .

U± ] = U± .87) Exercise.44).85) (12. (Recall that the strange quark has S = −1 by convention). J− } proved useful. T0 } and [Y.87) the commutation relations between the generators can be expressed in a succinct manner. Using the convention in (12. [Y. (12. U± ] = ±U± . which satisfied an ‘eigenvalue equation’ [J0 . V± ] = ±V± . namely spherical harmonics.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 . (12. Y is called the hypercharge.89) (12.77) is not the most useful basis in the study of SU (3).86) (12. (12. 2 Y = √ F8 .92) (12. starting with the observation that Y = B + S. 3 (12. We first introduce the F-spin operators 1 Fi = λ i . 1 [T0 .91) [T0 .84) (12. In the case of SU (2) the identification of ‘ladder’ operators {J+ .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. [Y. The generators of SU (3) can be arranged into a very similar form to that of SU (2). T 0 = F3 . we have [Y.90) (12.5: Lorentz Invariant Field Equations 385 As it turns out the set of generators in (12. 2 1 [T0 . 2 (12.93) (12.94) . Exercise. T0 ] = 0. J± ] = ± J± . I3 .88) which defines (not uniquely) a maximal set of mutually commuting operators {Y. Derive the Gell-Mann–Nishijima relation (12. T± ] = 0.11. U± = F6 ± iF7 .81) In chapter 5. In the basis (12. 2 (12. V± = F4 ± iF5 . First. V± ] = ± V± .83) (12. T± ] = ±T± .

In the case of SU (2) the representations lay on a line on the J0 axis. 2 3 Y + T0 = 2V0 . Casimir operators can be used to label irreducible representations of the Lie algebra. . [U+ . V± . This form is essential for easy labeling of the representations of the group. jkl (12. U± } to this maximal set by ‘eigenvalue equations’ and [T+ . T0 }. [T+ . V− ] = −U− . We can construct two such independent Casimir operators for the group SU (3). similar to the way it was done in section 5. When the basis of a Lie algebra is expressed in such a way to satisfy the form of the eigenvalue relations as given above. [U+ .103) Any remaining commutators follow from hermiticity. as the relation between the states in a given representation can be conveniently expressed in terms of ladder operators. U− ] = 0. Note that.102) (12.99) (12.386 Spinor Formulation which relate the remaining generators {T± .105) C2 = jkl djkl λj λk λl In general the number of independent Casimir operators of a Lie group is equal to its rank. J 2 ) which commutes with all of the generators. V+ ] = 0. C1 = k λ2 = − k 2i 3 fjkl λj λk λl . Finally.97) which relate commutators of generators with opposite eigenvalues to the maximal set {Y. the representations will now lie in a T0 − Y -plane.96) (12. we have [T+ . U− ] = Y − T0 = 2U0 . The interested reader is instead referred to a very readable account given in chapter 12 of [4]. A formal definition and a detailed discussion of the Cartan-Weyl form is beyond the scope of this chapter. However. while SU (3) has rank 2.100) (12. SU (2) has rank 1. As in the case of J 2 and SU (2). [T+ . we will use these commutation relations to construct representations of SU (3). U0 and V0 are linear combinations of T0 and Y . Thus.88). 3 [U+ . V− ] = T− .101) (12. The same way the angular momentum ladder operators have been used to construct the representations of SU (2). The maximum number of mutually commuting generators of a Lie algebra is called its rank. V− ] = 2 (12. (12. V+ ] = 0. it is said to be in Cartan-Weyl form. since there are two mutually commuting generators in SU (3) as given in (12. An operator which commutes with all generators of a Lie group is called a Casimir operator.104) (12. T− ] = 2T0 .98) (12. [V+ . Another important property of SU (2) is the existence of an operator (namely the total angular momentum. U+ ] = V+ .5. [T+ .95) (12.

y : T0 |t3 . y . 2 (12.y ± 1 . For example the pion octet (or any other meson octet) is therefore realized as states of a quark . 0) and D(0. y = t3 |t3 . 2 1 = β t3 ± .1). we have U0 |t3 . The representations for hexagons with sides of length p and q in the T0 − Y -plane. 1) as an example.88) we can label states by the eigenvalues of T0 and Y operators.112) The effect of these operators to the states in the y − t3 plane have been outlined in Fig. For example. 2 1 t3 ) |t3 . Because of (12. The representations D(1. U± .111) (12. For example. we have T± |t3 . y . (12. Figure (12.) All other representations can be constructed by combining these two conjugate representations. The interested reader is referred to [4]. 2 1 = γ t3 . The representations of SU (3) are constructed analogous to those of SU (2) by identifying the ‘boundary’ states annihilated by raising (or lowering) operators. y . (12. y . V± . (See Fig. especially chapters 7 and 8. The details of this procedure is beyond the scope of this chapter. y 1 = α 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. As another example for the representations of SU (3). Now we will construct explicit representations of SU (3). 2 (12. (12. y . |t3 .antiquark pair.5 how to label the irreducible representations of the angular momentum algebra SU (2) in terms of the value of the total angular momentum.108) (12.2) shows the representation D(2.11. 1) representation.106) (12. the pion family forms part of an octet corresponding to the D(1.110) (12. All other states of the representation are then constructed by successive application of ladder operators T± .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 . 1) correspond to the triplets of quarks and antiquarks. y . A notation suggestive of the dimensionality of the representation can be used to identify representations. y U± |t3 . This property can be used to label representations in terms of the values of the Casimir operators.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.3). y 3 = ( y− 4 3 = ( y+ 4 1 t3 ) |t3 . q) and it has a dimensionality of 2 (p + 1)(q + 1)(p + q + 2). y V± |t3 . respectively. Such a representation is labeled as 1 D(p. = y |t3 . . y V0 |t3 . y Y |t3 .109) The same way that J± |m is proportional to |m ± 1 in the case of the angular momentum algebra. For example. it was shown in section 5. y ± 1 .

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

− .11. This way the quark-antiquark states can be represented as follows [3] ⊗ [¯ = [8] ⊕ [1] 3] (12. 2 3 2 0. This expansion can be compared. 2 3 1 1 − . The three quarks in the fundamental representation can now be written as u = d = s = 1 1 . 3 (12. to the case of adding two spin triplet states. £ (12.118) ¡ ¢    ¨ ©§ ¨¥¦¤ "   ! . The octet D(1. where we would write [3] ⊗ [3] = [1] ⊕ [3] ⊕ [5]. .5: Lorentz Invariant Field Equations 389 D(1. 0) or the fundamental representation of flavor SU (3) symmetry. 1) is written as [8] etc. .113) The additional singlet state corresponds to the η meson. 0) and D(0.115) (12.116) (12. 3]. 2 (12.117) The Gell-Mann–Nishijima relation can then be succinctly expressed as 1 Q = y + t3 . 1) are denoted by [3] and [¯ respectively.3: D(1. in the case of SU (2). for example.114) Figure 12.

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