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

K. Schulten

Department of Physics and Beckman Institute University of Illinois at Urbana–Champaign 405 N. Mathews Street, Urbana, IL 61801 USA

(April 18, 2000)

Preface Preface

i

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

ii

Preface

Contents

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

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

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

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

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

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

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

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

iii

iv 5.4 5.5 5.6 5.7 5.8 5.9 5.10 5.11 5.12

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

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

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

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

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

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

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

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

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

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

vi

Contents

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

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

y(x)) to (x + dx. The length of such path can be determined starting from the fact that the incremental length ds in going from point (x.6) which reads in the present case M ∂L t1 d ∂L δS[q(t). y2 ). it holds in case δq(t0 ) = δq(t1 ) = 0 that qe (t) is an extremal. An Example As an application of the above rules of the variational calculus we like to prove the well-known result that a straight line in R is the shortest connection (geodesics) between two points (x1 . . M ) d dt ∂L ∂ qj ˙ − ∂L = 0 ∂qj (1. δq(t)] = dt − δqj (t) . This condition is stated in the following theorem. y(x1 ) = y1 .5). y(x2 ) = y2 . . .12) The proof of this theorem is based on the property Lemma: If for a continuous function f(t) f : [t0 . t1 ]. (1. M and δqj (t0 ) = δqj (t1 ) = 0 follows according to (1. (1.1: Variational Calculus 3 We now consider the question for which functions the functionals of the type (1. y1 ) and (x2 . hence.14) for any continuous function h(t) ∈ F with h(t0 ) = h(t1 ) = 0. δq(t)] = 0 for all δq(t) ∈ F . from (1.11) The extremals qe (t) can be identiﬁed through a condition which provides a suitable diﬀerential equation for this purpose.12) for j = 1.13) dt f (t)h(t) = 0 t0 (1. the above theorem. . 2. if and only if it satisﬁes the conditions (j = 1.16) This property holds for any δqj with δq(t) ∈ F . . y(x + dx)) is ds = (dx)2 + ( dy dx)2 = dx dx 1+( dy 2 ) . t1 ] ⊂ R → R holds t1 (1. Theorem: Euler–Lagrange Condition For the functional deﬁned through (1. ∂qj dt ∂ qj ˙ t0 j=1 (1. On the other side.17) . .15) We will not provide a proof for this Lemma. 2.16) the property δS[qe (t). then f (t) ≡ 0 on [t0 . dx (1. . .12) for j = 1. For this purpose we deﬁne Deﬁnition: An extremal of a diﬀerentiable functional S[ ] is a function qe (t) with the property δS[qe (t). · ] ≡ 0 and.5) assume extreme values.1. Let us assume that the two points are connected by the path y(x). 2. The proof of the above theorem starts from (1. According to the Lemma above follows then (1. . M . .

.21) are extremals of the functional.23) Presently we consider only velocity–independent potentials. . The shortest path is an extremal of s[y(x)] which must. .18) dy s is a functional of y(x) of the type (1. . y = const. dx (1. hence.19) From this follows y / 1 + (y )2 = const and. t1 S[q(t)] = t0 ˙ dt L(q(t). q(t). One obtains y 1 − y2 y(x) = (x − x2 ) + y2 . The constants a and b are readily identiﬁed through the conditons y(x1 ) = y1 and y(x2 ) = y2 . t) is the so-called Lagrangian M ˙ L(q(t). t) = j=1 1 mj qj − U (q1 . 1. This in turn yields y(x) = ax + b.1.2 Lagrangian Mechanics The results of variational calculus derived above allow us now to formulate the Hamiltonian Principle of Least Action of Classical Mechanics and study its equivalence to the Newtonian equations of motion. Velocity–dependent potentials which describe particles moving in electromagnetic ﬁelds will be considered below. according to the theorems above. qM ) . circles whose centers lie at the center of the sphere.20) x1 − x2 Exercise 1. obey the Euler–Lagrange dy condition. (1. ˙2 2 (1. . t) (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 . the so-called action integral. .1: Show that the shortest path between two points on a sphere are great circles. i. . (1. 2.5) with L(y(x). dx ) = 1 + (dy/dx)2 . Using y = dx the condition reads d dx ∂L ∂y = d dx y 1 + (y )2 = 0.22) ˙ where L(q(t). q(t).e.4 The total path length is then given by the integral x1 Lagrangian Mechanics s = x0 dx 1+( dy 2 ) . q2 . M (1. q(t). . j = 1..

t) = A(r. t) + V (r. respectively.1. t) describes the charge density present in the ﬁeld and J(r. t) dt c where dr = v and where F (r. dt The ﬁelds E(r. t) and a vector potential A(r. Equations (1. t) obey the Maxwell equations ×E + ·B ×B − ·E 1 ∂ c ∂t E 1 ∂ c ∂t B (1. t) c ∂t (1.31) allows one to transform the potentials without aﬀecting the ﬁelds and without aﬀecting the equations of motion (1. t) .25) of a particle moving in the ﬁeld.31) = − V − Gauge Symmetry of the Electromagnetic Field It is well known that the relationship between ﬁelds and potentials (1. t) = V (r. t) is the Lorentz force.32) (1. These conditions read in the present case ∂L d ∂L ∂U d − = 0 → − − (mj qj ) = 0 ˙ (1. t). The equations of motion for such a particle are d q ˙ (mr) = F (r.25) = 0 = 0 4π J c = 4πρ = (1.27) and (1. t) as follows B E = ×A 1 ∂A .23). t) = q E(r.26) (1.28) can be satisﬁed implicitly if one represents the ﬁelds through a scalar potential V (r. t) and B(r.2: Lagrangian 5 For a proof of the Hamiltonian Principle of Least Action we inspect the Euler–Lagrange conditions associated with the action integral deﬁned through (1. The transformation which leaves the ﬁelds invariant is A (r. t) − K(r. x3 (t)) moving in an electromagnetic ﬁeld described through the electrical and magnetic ﬁeld components E(r. F (r.22. t) describes the charge current density. x2 (t). 1. 1. Particle Moving in an Electromagnetic Field We will now consider the Newtonian equations of motion for a single particle of charge q with a trajectory r(t) = (x1 (t).33) .30) (1.27) (1.24) ∂qj dt ∂ qj ˙ ∂qj dt which are obviously equivalent to the Newtonian equations of motion.29) where ρ(r. t) + v × B(r.30. t) 1∂ K(r.28) (1. c ∂t (1. t) and B(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. (1. in fact. (1.41) which is identical to the term (1.35) is equivalent to the Euler–Lagrange condition d dt The second term in (1.41) with (1. dt ∂x1 c We notice using (1.36) in the Newtonian equation of motion.37) The results (1. e. t) · v . t) = 1 q mv 2 − q V (r.39) ∂A1 ∂A2 ∂A3 x1 + ˙ x2 + ˙ x3 ˙ ∂x1 ∂x1 ∂x1 .35) − ∂A1 ∂x2 [v × B]1 = x2 B3 − x3 B2 = x2 ˙ ˙ ˙ ∂A2 ∂A1 − ∂x1 ∂x2 − x3 ˙ ∂A1 ∂A3 − ∂x3 ∂x1 .40) (1.36) This expression allows us to show that (1.25).6 Lagrangian of Particle Moving in Electromagnetic Field Lagrangian Mechanics We want to show now that the equation of motion (1. Comparing then (1.25) follows from the Hamiltonian Principle of Least Action.37) is ∂L ∂V q = −q + ∂x1 ∂x1 c The ﬁrst term in (1. 2 c (1. 1.. B3 = ∂A2 ∂x1 (1.g.38.38) ∂L ∂ x1 ˙ − ∂L = 0. t) + A(r. ∂x1 (1. 1. equivalent.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 . (1. . namely. r. if one assumes for a particle the Lagrangian ˙ L(r.35) shows that the Newtonian equations of motion and the Euler–Lagrange conditions are.40.34) For this purpose we consider only one component of the equation of motion (1.30). d ∂V q (mv1 ) = F1 = −q + [v × B]1 .

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

we like to ˆ provide the transformation here anyway for later reference r = r + e.51) We would like to derive this transformation in a somewhat complicated. t) by gauge transformed potentials V (r. the transformation (1. t). i. However. 1.47) (1. in the case of central force ﬁelds which are invariant with respect to rotations of coordinates around the center. A (r.42) corresponds to replacing in the Lagrangian potentials V (r. R(r) = through ∂r ∂ (1. but nevertheless instructive way considering rotations around the x3 –axis. r ∈ R . where the origin of is chosen such that r (r. We consider now invariance properties connected with coordinate transformations. ˆ ˆ ˆ Examples of Inﬁnitesimal Transformations The beauty of inﬁnitesimal transformations is that they can be stated in a very simple manner.32.. t).50) A non-trivial example is furnished by the inﬁnitesimal rotation around axis e ˆ r = r + e×r . the equivalence of (1. r.e. t). for example. The corresponding inﬁnitesimal transformation is deﬁned for small r (r. The transformations considered are speciﬁed in the following deﬁnition. for the connection between invariance properties and constants of motion. The transformations we consider are the most simple kind. which has an important analogy in Quantum Mechanics. ) . We have proven.49) =0 ∈ R (1. ) = r + R(r) + O( 2 ) . ˆ (1.33). The following description of spatial symmetry is important in two respects. Such invariance properties are very familiar. and for the introduction of inﬁnitesimal transformations which will provide a crucial method for the study of symmetry in Quantum Mechanics. the reason being that our interest lies in achieving familiarity with the principles (just mentioned above ) of symmetry properties rather than in providing a general tool in the context of Classical Mechanics.8 Lagrangian Mechanics Obviously. therefore. In case of a translation transformation in the direction e nothing new is gained. Deﬁnition: Inﬁnitesimal One-Parameter Coordinate Transformations A one-parameter coordinate transformation is decribed through r = r (r.42) and (1.48) In the following we will denote unit vectors as a. In this case the transformation can be written in matrix form cos −sin 0 x1 x1 x2 = sin cos 0 x2 (1. A(r. 0) = r .52) x3 0 0 1 x3 . ˆ (1. for such vectors holds a · a = 1.

t) = L(q. q. (1. The property has been shown to hold in a more general context. Invariance implies L = L. r) which is constant along the classical path of the system. Using the fact. neglecting terms of order O( M ˙ ˙ L (q. namely for ﬁelds rather than only for particle motion. i. 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. the property holds for any system.k=1 ∂L ∂Qj qk ˙ ∂ qj ∂qk ˙ (1. a classical mechanical system is invariant with respect to a coordinate transformation ˙ a constant of motion exists... We have used here the notation corresponding to single particle motion.53).1. We consider here only the ‘particle version’ of the theorem. by Noether.55) (1. expansion. then M ∂L j=1 Qj ∂ xj is a constant of motion. t) + j=1 ∂L Qj + ∂qj M j. Theorem: Noether’s Theorem ˙ If L(q. ˙ We have generalized in this theorem the deﬁnition of inﬁnitesimal coordinate transformation to M –dimensional vectors q.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) (1.51) reads r = r − x2 e1 + x1 e2 ˆ ˆ (1. ˙ ∂qk ˙ Inserting these inﬁnitesimal changes of qj and qj into the Lagrangian L(q. In order to prove Noether’s theorem we note qj qj ˙ = qj + = qj + ˙ k=1 Qj (q) M (1. q. the second and third term ˙ k=1 in (1. .56) ∂Qj qk . a quantity C(r. t) is invariant with respect to an inﬁnitesimal transformation q = q + Q(q).3: Symmetry Properties 9 In case of small this transformation can be written neglecting terms O( 2 ) using cos = 1 + O( 2 ).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. Anytime. however. sin = + O( 2 ) 0 − 0 x1 x1 x1 x2 = x2 + 0 0 x2 + O( 2 ) . i. In the context of Lagrangian Mechanics this implies that such transformation leaves the Lagrangian of the system unchanged.57) ∂ d where we used dt Qj = M ( ∂qk Qj )qk .54) which is equivalent to (1. t) yields after Taylor ˙ 2 ).58) From this follows the statement of the theorem. q.57) must cancel each other or both vanish.e. 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.

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

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

t|r . the propagator is unitary.9) Employing a continuum of intermediate times t ∈ [t0 . t|r . This is diﬀerent from the classical situation where the particle follows a discrete path and. t).e.e. t|r0 . t1 ]) d3 r φ(r. This symbol will be deﬁned further below. t ∈ (2. requires then according to (2.9) for N → ∞. . (t > t . tN −1 ) φ(rN −1 . t ∈ [t0 . . > t1 > t0 is φ(r. at any intermediate time the particle needs only be known at one space point. (2. t |r0 .4) holds for all times. t0 ) = δ(r − r0 ). Since (2. The state of a system at time t. t )φ(r. t| rN −1 .7) t. t ) ψ(r . t ) which requires d3 r φ(r. tN −1 |rN −2 . The deﬁnition will actually assume an inﬁnite number of intermediate times and express the path integral through integrals of the type (2. t1 ]) Ωd Ωd 3r Ωd 3r Ωd 3r 3 r |ψ(r. This must hold for any ψ(r . t) is described by a linear map of the type ψ(r. t1 ] (2. described by ψ(r. t. t|r0 . namely the point on the classical path at time t . t|r . The time evolution of ψ(r. t = )|2 = 1 . t ) t > t . t0 ) = φ(r. t|r .12 Quantum Mechanical Path Integral 5. t ) Ωd 3 r |ψ(r. t.8) 7. the path integral r(tN )=rN d[r(t)] · · · r(t0 )=r0 (2.5) 6. ψ(r0 . 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.11) which denotes an integral over all paths r(t) with end points r(t0 ) = r0 and r(tN ) = rN . i. t0 ) = r(t0 )=r0 d[r(t)] Φ[r(t)] . t|r0 . t ) φ(r . t0 ) . t ) φ(r. The following so-called completeness relationship holds for the propagator (t > t [t0 .. t)|2 = (2. t ) ψ(r .6) φ(r. t ∈ [t0 . hence. i.8) a knowledge of the state at all space points r ∈ Ω at some intermediate time t .. t1 ] yields an expression of the form r(tN )=rN φ(r. t ) = δ(r − r ) Ω (2.10) We have introduced here a new symbol. 8. . The generalization of the completeness property to N − 1 intermediate points t > tN −1 > tN −2 > . t|r . (2. t ) ψ(r . tN −2 ) · · · φ(r1 . t1 |r0 . t) = Ω d3 r φ(r. t0 ) = Ωd 3r N −1 Ωd 3r N −2 ··· Ωd 3r 1 φ(r. t|r .

This implies that small variations of the path near the classical path alter the value of the action integral by very little.13). Its value is extremely small in comparision with typical values for action integrals of macroscopic particles. namely due to the property of the classical path to be an extremal of the action integral. r.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. a very large number.13) L(r. i. 2. 2.14) ˙ In (2. the exponent is a very large imaginary number. Only for paths close to the classical path is such interference ruled out. contribution to the path integral (2. However. In case of a proton with mass m = 1. for the electron many paths contribute and the action integrals for non-classical paths need to be known. Situations which are well described classically will be distinguished through the property that the classical path gives the dominant. The value of S[r(t)] is then Scl = 1 1 m v 2 t = erg s .11) is Φ[r(t)] = exp where S[r(t)] is the classical action integral tN 13 i S[r(t)] (2. Scl / ≈ 8. interferences should be much less dramatic than in case of the macroscopic particle.15) The exponent of (2. in complete distinction from Classical Mechanics.12. However. it is comparable to action integrals as they arise for microscopic particles under typical circumstances. i.. such that destructive interference of the contributions of such paths does not occur. The situation is very diﬀerent for microscopic particles. actually often essentially exclusive.5 · 1027 .e.2: Axioms 9.e.12) S[r(t)] = t0 ˙ dt L(r.12.0545 · 10−27 erg s .14) has the same dimension as the action integral S[r(t)]. r.. (2. Any variations of Scl would then lead to strong oscillations of the contributions exp( i S) to the path integral and one can expect destructive interference betwen these contributions. not only for the classical path. t) is the Lagrangian of the classical particle. t) (2. To show this we consider the value of the action integral for a particle of mass m = 1 g moving over a distance of 1 cm/s in a time period of 1 s. The constant given in (2. However. 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. However.13) and = 1. expressions (2.13) are built on action integrals for all possible paths. .2. The functional Φ[r(t)] in (2. for microscopic particles like the electron this is by no means the case. Since this number is multiplied by ‘i’.12) is then Scl / ≈ 0. 2 2 (2. One would still expect signiﬁcant descructive interference between contributions of diﬀerent paths since the value calculated is comparable to 2π. accordingly.

The particles have associated with them a Lagrangian L(x. tj ) to the next (xj+1 . . (2.18) The time instances are ﬁxed. For the sake of simplicity we will consider the case of particles moving in one dimension labelled by the position coordinate x.19) The main idea is that one can replace the path integral now by a multiple integral over x1 . the xj values are not.13) and. (xN −1 . (2. .3 How to Evaluate the Path Integral In this section we will provide an explicit algorithm which deﬁnes the path integral (2. They can be anywhere in the allowed volume which we will choose to be the interval ] − ∞. t) = ˙ 1 mx2 − U (x) . . at the same time.16) In order to deﬁne the path integral we assume. x.20) can be properly taken.12) φ(xN . (x1 . −∞ dxN −1 (xj+1 −xj )2 1 − U (xj ) . ∞[. t1 ). a series of times tN > tN −1 > tN −2 > . tN )} . Its value is CN = m 2πi N 2 (2. > t1 > t0 letting N go to inﬁnity later. t0 ). ˙ 2 (2.12.10) and (2. etc. however. t0 ) = limN →∞ CN exp i N N −1 j=0 +∞ +∞ +∞ −∞ dx1 −∞ dx2 . 2. In passing from one space–time instance (xj . 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. .21) N 2. tN −1 ). 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. .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 . x2 . (xN . tN |x0 . .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. as in (2. . tj+1 ) we assume that kinetic energy and potential energy are constant. namely 1 m(xj+1 − xj )2 / 2 and U (xj ). .20) Here. provides an avenue to evaluate path integrals. 2 2m N (2. namely tj+1 − tj = (tN − t0 )/N = N . respectively. (2.9). The spacings between the times tj+1 and tj will all be identical. This allows us to write the evolution operator using (2.

t0 ) (2.h.24) Also we use the fact that the velocity of the classical path xcl = (xN −x0 )/(tn −t0 ) is constant.e. t0 ) = exp S[x(t)] (2.24) implies for the second term on the r. The resulting expression for S[x(t)|x(t0 ) = x0 . One obtains S[x(t)|x(t0 ) = x0 . of (2. tN dt y = y(tN ) − y(t0 ) = 0 .25) is. 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. 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 ) = 0.29) This expression corresponds to the action integral in (2. t0 ) and (xN . tN |0. t0 ) and (xN . Inserting the result into (2.25) + The condition (2. x(tN ) = 0] . the origin of which coincides with the classical path of the particle.12) yields x(tN )=0 im (xN − x0 )2 i φ(xN . x(tN ) = 0] . which can also be written φ(xN .22) where xcl (t) is the classical path which connects the space–time points (x0 . .23) tN − t0 It is essential for the following to note that. (2.10. namely.23).4: Propagator for a Free Particle 15 Rather then using the integration variables xj . x(tN ) = xN ] is S[x(t)|x(t0 ) = x0 . it holds y(t0 ) = y(tN ) = 0 .s. ˙ t0 (2.. ˙ 2 t0 (2. ˙ ˙ ˙ 2 + y2) = ˙ (2. tN dt t0 1 1 (xN − x0 )2 m x2 = m ˙ cl . The ˙ 1 S[x(t)|x(t ) = x . due to (2. since x(t0 ) = xcl (t0 ) = x0 and x(tN ) = xcl (tN ) = xN .h.2. tN ).s.28) i.24). it is more suitable to deﬁne new integration variables yj . t0 ) = exp 1 im (xN − x0 )2 2 tN − t0 φ(0. 2.26) The ﬁrst term on the r.13). xN − x0 xcl (t) = x0 + ( t − t0 ) . = (2. To see the beniﬁt of such approach we deﬁne a path y(t) as follows x(t) = xcl (t) + y(t) (2.27) The third term can be written in the notation introduced tN 1 dt my 2 = S[x(t)|x(t0 ) = 0. using (2. tN |x0 . tN |x0 . tN ) is to be evaluated. 2 2 tN − t0 (2. can be expressed through a path integral with endpoints x(t0 ) = 0.31) We have denoted explicitly that the action integral for a path connecting the space–time points (x0 . (2. x(tN ) = xN ] 1 (xN − x0 )2 m + 0 + 2 tN − t 0 + S[x(t)|x(t0 ) = 0.30) d[x(t)] exp 2 tN − t0 x(t0 )=0 a result.

. . . 2 N .k=1 dy1 · · · −∞ −∞ dyN −1 exp i yj bjk yk = (iπ)d det(bjk ) 1 2 . real. In the appendix we prove +∞ +∞ d j. (2. (2. 2 −1 −1 2 (2. .34) of (ajk ) deﬁning a new matrix (Ajk ) through ajk = Using det(ajk ) = m 2 N N −1 m 2 N Ajk . tN |0.36) which holds for a d-dimensional. (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... . symmetric matrix (bjk ) and det(bjk ) = 0. ..32) (yj+1 − yj )2 N −1 1 2 j=0 2 m N The exponent E can be written..34) dy1 · · · −∞ −∞ dyN −1 exp i N −1 j.33) where ajk are the elements of the following symmetric (N 2 −1 0 −1 2 −1 2 m 0 −1 ajk = .. .. . 0 0 . . . . noting y0 = yN = 0. In order to complete the evaluation of (2.k=1 yj ajk yk (2. . 0 0 . .k=1 yj ajk yk − 1) × (N − 1) matrix .32) we split oﬀ the factor 2 mN in the deﬁnition (2.38) .37) det(Ajk ) . 0 0 0 0 0 0 The following integral +∞ +∞ (2.16 Evaluation of the necessary path integral Quantum Mechanical Path Integral To determine the propagator (2. t0 ) = limN →∞ × +∞ +∞ −∞ dy1 · · · −∞ dyN −1 exp i N m 2πi N 2 N × (2. ...31) for a free particle one needs to evaluate the following path integral φ(0.35) must be determined. 0 0 . .

which follows from (2. 2. .. 0 0 Dn = . 0 0 −1 2 −1 . We seek then to evaluate the determinant of the n × n matrix 2 −1 0 .. variable. n = 1. .. t0 ) = limN →∞ and with NN (2. . t0 ) = m 2πi (t − t0 ) 1 2 exp im (x − x0 )2 2 t − t0 .18) we obtain φ(0. . . . To solve this three term recursion relationship one needs two starting values. deﬁning t = tN . . 2 −1 0 0 0 −1 2 For this purpose we expand (2. presently N − 1.39) yields φ(0.45) Expressions for Free Particle Propagator We have now collected all pieces for the ﬁnal expression of the propagator (2.41) D2 = 2 −1 −1 2 = 3 (2.46) . Our derivation has provided us with the value det(Ajk ) = N . 0 0 0 −1 2 .40) in terms of subdeterminants along the last column.4: Propagator for a Free Particle a property which follows from det(cB) = cn detB for any n × n matrix B. . Using D1 = |(2)| = 2 . . Inserting this into (2. det(Ajk ) (2.43) We like to note here for further use below that one might as well employ the ‘artiﬁcial’ starting values D0 = 1. one can readily verify Dn = n + 1 . x = xN φ(x. tN |0. .31) and obtain. .. One can readily verify that this procedure leads to the following recursion equation for the determinants Dn = 2 Dn−1 − Dn−2 .2.42) m 2πi NN 1 2 (2. tN |0. . t|x0 . . .. . (2. . tN |0. .41) the same result for D2 .40) ..39) In order to determine det(Ajk ) we consider the dimension n of (Ajk ). we obtain φ(0. . . (2. .. let say n. D1 = 2 and obtain from (2. (2. D3 . t0 ) = limN →∞ m 2πi N 2 17 N 2πi m N N −1 2 1 . . (2. . .44) = tN − t0 . t0 ) = m 2πi (tN − t0 ) 1 2 ... 0 0 0 . .

53) .247). thereby. x) + E(x) in (2. according to (2.48) as the initial state using the propagator (2.51) as follows Eo (xo . Below we will apply this to a particle at rest and a particle forming a so-called wave packet. t|r0 . t0 ) = 1 πδ 2 1 4 exp − x2 po 0 +i x 2 2δ (2. We need to evaluate for this purpose the integral ψ(x.5) to (2. t) = 1 πδ 2 1 4 m 2πi t 1 2 +∞ dx0 exp −∞ im (x − x0 )2 x2 po − 02 + i xo 2 t 2δ Eo (xo . We devide the contributions to the exponent Eo (xo .50) One refers to such states as wave packets. (2.47) One-Dimensional Free Particle Described by Wave Packet We assume a particle at time t = to = 0 is described by the wave function ψ(x0 .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 (2.52) and applying (2.46) can be generalized to three dimensions in a rather obvious way. the associated probability distribution |ψ(x0 . t0 )| = 2 1 πδ 2 1 2 exp − x2 0 δ2 (2. x) = im 2 t x2 1 + i o t mδ 2 − 2xo x − po t m + f (x) (2. The result (2.46). (2. the wave function of the particle at later times.10) φ(r. One obtains then for the propagator (2. t) of a free particle. and describes a single particle since 1 πδ 2 1 2 +∞ dx0 exp −∞ − x2 0 δ2 = 1. centered around x0 = 0.5) allows us to predict the time evolution of any state function ψ(x.48) Obviously. We want to apply axiom (2. x) + E(x) (2.18 Quantum Mechanical Path Integral This propagator. t0 ) = m 2πi (t − t0 ) 3 2 exp im (r − r0 )2 2 t − t0 . We will obtain.49) is Gaussian of width δ.

4: Propagator for a Free Particle im x2 − f (x) . (2. making certain. We proceed as follows.2.56) m 2πi t 1 2 +∞ eE(x) −∞ dx0 eEo (xo . The beginning and the end of such path are the points z1 = −∞ × i 1−i t mδ 2 . x) = One can write then (2. t) = 1 πδ 2 1 4 2 t 1 + i mδ2 . 2 t One chooses then f (x) to complete.55) im xo 2 t 1+i t − mδ 2 x− 1+ po m t t i mδ2 .52).x) (2. the square in (2.61) .53) E(x) = x− po mt 19 (2.60) is then represented by real ρ values.54) f (x) = This yields Eo (xo . z2 = +∞ × i 1−i t mδ 2 (2. We consider a transformation to a new integration variable ρ deﬁned through i 1−i t mδ 2 ρ = x0 − x− 1+ po m t t i mδ2 . that the new path does not lead to additional contributions to the integral.51) ψ(x. according to (2. (2. however.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 .58) The integrand is an analytical function everywhere in the complex plane and we can alter the integration path. (2.59) An integration path in the complex x0 –plane along the direction i 1−i t mδ 2 (2.57) and needs to determine the integral +∞ I = −∞ +∞ dx0 eEo (xo . 2 (2.

Quantum Mechanical Path Integral (2.20 whereas the original path in (2.58) along the path z1 → z1 and along the path z2 → z2 gives only vanishing contributions one can replace (2. (2. We can conclude then that (2. t m(1 + i mδ2 ) (2. Since the exponent in (2. one can show that z1 lies at −∞ − i × ∞ and z2 at +∞ + i × ∞.62) If one can show that an integration of (2.58) vanishes 0 for Re x0 → ±∞.65) t 1 − i mδ2 t 1 + i mδ2 .54) using (2.58) has a leading contribution in x0 of −x2 /δ 2 the integrand of (2. t) = Seperating the phase factor 1 t 1 + i mδ2 1 4 1 2 exp [ E(x) ] .67) We need to determine ﬁnally (2. (2. I = Equation (2.68) and. the paths between z1 → z1 and z2 → z2 have a real part of x0 of ±∞.63) holds and. One obtains i po i po x x2 2m t E(x) = − 2 + − t t t 2δ (1 + i mδ2 ) 1 + i mδ2 1 + i mδ2 2 (2.57) reads then 1 πδ 2 1 4 2πi t . using a ab = a − .55). z2 = +∞ .70) .58) by t mδ 2 +∞ I = i 1−i dρ exp − −∞ m 2 t 1+ t mδ 2 2 ρ2 (2. (2.63) which can be readily evaluated. Hence. 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) ] .66) yields ψ(x. In fact.64) ψ(x.69) t)2 t 2δ 2 (1 + i mδ2 ) + i po x − i p2 o t 2m (2. 1+b 1+b ﬁnally E(x) = − (x − po m (2. accordingly.58) has the end points z1 = −∞ .

However. (2. The conservation of this factor is particularly striking for the (unnormalized) initial wave function ψ(x0 . respectively.71) conserves the phase factor exp[i(po / )x] of the original wave function (2. 2. In this case holds 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. 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. coinciding at t = 0 with the width of the initial distribution (2.72) 2 t2 δ 1+ 2 4 (2. (2.72). 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 .f.77) .48) and that the respective phase factor is related with the velocity of the classical particle and of the center of the distribution (2.73) m δ increases with time..49).48) for δ → ∞. (2.71.72) Comparision of Moving Wave Packet with Classical Motion It is revealing to compare the probability distributions (2.67) provides the complete expression of the wave function at all times t ψ(x. (2.76) From this we conclude that an initial wave function ψ(xo .72). t) = exp i po i p2 o x − t m 2m . Another interesting observation is that the wave function (2.49). We had assumed above [c. This ‘spreading’ of the wave function is a genuine quantum phenomenon. 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 .48) and for the ﬁnal state (2.74) which corresponds to (2. the case of arbitrary to is recovered iby replacing t → to in (2. t0 ) = exp i po i p2 o xo − to m 2m . The center of the distribution (2. This yields.72) for the initial state (2. the spatial dependence of the initial state (2.2. The corresponding probability distribution is |ψ(x. (2.75) i.75) ψ(x.71) × exp − (x − po m t)2 ) 2δ 2 (1 + 2 t2 m2 δ 4 (1 − i .71).48)] to = 0. The width of the distribution (2.e. t) = exp i i p2 po o x − (t − to ) m 2m . t0 ) = exp i po xo m . (2. instead of (2.74) remains invariant in time. (2.4: Propagator for a Free Particle which inserted in (2.

t) ˙ L (y. t) + L (y. y(t).82) which describes the deviation from the classical path xcl (t) with end points xcl (t0 ) = x0 and xcl (tN ) = xN . y(tN ) = 0 .85) We want to show now that the contribution of δL to the action integral (2.12. y(t).13) φ(xN . xcl + y(t). (2. 2.5 Propagator for a Quadratic Lagrangian x(tN )=xN We will now determine the propagator (2.10.80) into (2. In order to simplify this task we deﬁne again a new path y(t) x(t) = xcl (t) + y(t) (2.82) one obtains L(xcl + y. 2. y(t)] which vanishes according to the Hamiltonian principle as discussed in Sect. ˙ ˙ = (2. tN |x0 . xcl . For this purpose we use d xcl y = ˙ ˙ (xcl y) − xcl y ˙ ¨ (2. x. 2.79) for a quadratic Lagrangian L(x. t) ˙ (2.22 becomes at t > to ψ(x. t) = L(xcl . t0 ) = x(t0 )=x0 d[x(t)] exp i S[x(t)] (2. ˙ 2 2 tN (2.80) For this purpose we need to determine the action integral S[x(t)] = t0 dt L(x.84) .. 2 . Obviously. Such states play a cardinal role in quantum mechanics. One refers to the respective states as stationary states.86) dt The reader may want to verify that the contribution of δL to the action integral is actually equal to the diﬀerential δS[xcl . t) ˙ δL 1 1 mx2 − c(t)x2 − e(t)xcl ˙ cl cl 2 2 1 1 = my 2 − c(t)y 2 ˙ 2 2 = mxcl y(t) − c(t)xcl y − e(t)y . x.83) (2. t) = exp i Quantum Mechanical Path Integral po i p2 o x − t m 2m . the spatial as well as the temporal dependence of the wave function remains invariant in this case. t) + δL ˙ ˙ ˙ ˙ where L(xcl .e. t) = ˙ 1 1 mx2 − c(t)x2 − e(t)x .81) vanishes2 .81) for an arbitrary path x(t) with end points x(t0 ) = x0 and x(tN ) = xN .78) i. (2. xcl . 1. the end points of y(t) are y(t0 ) = 0 Inserting (2.

. . tN |x0 . also the second contribution on the r. 0 0 −1 2 −1 . 0 0 0 c2 0 . . .. t0 ) = y(tN )=0 d[y(t)] exp y(t0 )=0 i tN dt L (y. tj = t0 + form j..91) (yj+1 − yj )2 1 2 2m N where cj = c(tj ). . . . . 0 0 0 .. tN |0. One can then express the propagator (2.5: Propagator for a Quadratic Lagrangian and obtain tN t0 23 dt δL = m [xcl y]|t0 − ˙ t N tN dt [ m¨cl (t) + c(t) xcl (t) + e(t) ] y(t) . . . t0 ) (2. .93) . t0 ) we will adopt a strategy which is similar to that used for the evaluation of (2. . t) ˙ t0 . . tN |0. . . . . 0 0 m ajk = . . . .88) and.89) where ˜ φ(0. For the evaluation of φ(0.24) to the Lagrangian (2. 0 0 0 c3 . 2 N . . . 0 0 0 . .. .s.88) vanishes. tN |0. . 0 0 0 −1 2 . x t0 (2.80) yields for the classical path m¨cl + c(t) xcl + e(t) = 0 x (2.79) i ˜ φ(xN . . y.91) through the quadratic N −1 E = i j.32).2. vanishes.s. tN |0.90) Evaluation of the Necessary Path Integral We have achieved for the quadratic Lagrangian a separation in terms of a classical action integral and a propagator connecting the end points y(t0 ) = 0 and y(tN ) = 0 which is analogue to the ˜ result (2. (2.92) where ajk are the elements of the following (N − 1) × (N − 1) matrix 2 −1 0 .. The discretization scheme adopted above yields in the present case ˜ φ(0. of (2. t0 ) = limN →∞ × +∞ +∞ −∞ dy1 · · · −∞ dyN −1 exp N i N N −1 j=0 m 2πi N 2 N × − 1 2 cj 2 yj (2.. . .87) According to (2. 0 0 N 0 − . . . . . cN −2 0 0 0 0 0 cN −1 (2.h. . 2 −1 0 0 0 −1 2 c1 0 0 . . .83) the ﬁrst term on the r. .h. t0 ) = exp S[xcl (t)] φ(0. One can express the exponent E in (2. . .k=1 yj ajk yk (2. 2 .31) for the free particle propagator. ... . Applying the Euler–Lagrange conditions (1. hence.

24

Quantum Mechanical Path Integral

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

N 2

N

(iπ)N −1 det(a) 1

1 2

= limN →∞

1

2

N

2

N

N −1

m

det(a)

.

(2.94)

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

m c1

2 N 2 N

2

N

N

N −1

m

det(a)

.

(2.95)

def

2

N

N −1

m

det(a) 0 0 0 . . .

m cN −2

2 N

(2.96) 0

−1 −1 2 − . . . 0 0

2 N

0 ... −1 . . . ... . .. . . .

m c3

−1 2 − 0 . . . 0 0

m c2

0 ... 2 − 0

−1 2 −

m cN −1

2 N

0 0 . . . −1

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

2 N

m

cn

Dn−1 − Dn−2

(2.97)

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

2 N

m

c1

(2.98)

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

2 N

= −

cn+1 Dn . m

(2.99)

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

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

df (t0 ,t) dt t=t0

25

= =

N N

D0 = 0 ; D1 − D0

N

= 2−

2 N

m

c1 − 1 = 1 .

(2.101)

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

1 2

exp

i

S[xcl (t)]

(2.102)

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

2.6

Wave Packet Moving in Homogeneous Force Field

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

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

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

1 2

exp

i

S[xcl (τ )] .

(2.106)

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

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

Quantum Mechanical Path Integral

(2.108)

x − xo 1 f − (t − to ) t − to 2m

τ +

1 f 2 τ 2m

(2.109)

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

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

t 2

=

+ 2f

x − xo 1 f − (t − to ) t − to 2m

τ (2.111)

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

to

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

2

=

1 m 2 +f +

(t − to ) (t − to )2

1 f2 (t − to )3 + xo f (t − to ) 3 m

=

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

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

1 2

×

(2.113)

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

2.5: Propagator for a Quadratic Lagrangian

27

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

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

1

(2.114)

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

E(x) =

f t2 x − f (x) m

1 f 2 t3 . 24 m

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

1 4 2 t2 ) m2 δ 4

(2.117)

po m

f t2 2m

2

.

(2.118)

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

t 1 − i mδ2 t 1 + i mδ2

1 4

1 πδ 2 (1 +

exp [ E(x) ]

(2.119)

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

2

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

2 2

(2.120)

1+i t mδ 2 −

− x−

x−

x− x

f t2 po t− m 2m

po f t2 t− m 2m

2

f t2 + 2x m

po f t2 t+ m 2m

f t2 po t+ m 2m

1+i

t mδ 2 (2.121)

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

po m

Quantum Mechanical Path Integral

E(x)

=

− +

t−

2δ 2 1 + i

f t2 2m t mδ 2

2

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

(po + f t) x −

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

t

dτ (po + f τ )2 .

0

(2.123)

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

t 1 − i mδ2 t 1 + i mδ2

1 4

ψ(x, t)

=

1 πδ 2 (1 + x− 2δ 2

po m

2 t2 ) m2 δ 4

1 4

×

× exp − × exp i

t−

f t2 2m

2

1+

2 t2 m2 δ 4

1−i

t

t mδ 2

× . (2.124)

(po + f t) x −

i

0

dτ

(po + f τ )2 2m

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

2 t2 m2 δ 4 1 2

)

x− t− exp − 2 2 δ 2 1 + m2tδ4

po m

f t2 2m

2

.

(2.125)

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

2.5: Propagator for a Quadratic Lagrangian

29

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

**Propagator of a Harmonic Oscillator
**

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

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

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

t

(2.129) (2.130)

(2.131)

A =

x − x0 cosω(t − to ) , sinω(t − t0 )

(2.132)

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

(2.133)

S[xcl (τ )] =

t0

dτ

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

(2.135)

30

Quantum Mechanical Path Integral

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

(2.137)

(2.138)

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

t 0 dτ g(τ )

(2.139) (2.140) (2.141)

(2.142)

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

t

requires the integrals (2.143) (2.144)

dτ cos2ωτ

0 t

= =

dτ sin2ωτ

0

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

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

(2.145)

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

31

(2.146)

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

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

1 2

× . (2.147)

(x2 + x2 ) cosω(t − t0 ) − 2x0 x 0

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

1 4

exp

−

mω(x0 − bo )2 i + po xo 2

.

(2.148)

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

o

(2.149)

=

energy

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

∞

ψ(x, t) =

−∞

dx0 φ(x, t|x0 , t0 ) ψ(x0 , t0 ) .

(2.150)

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

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

1 4

m 2πiω sinω(t − t0 )

1 2

∞

dx0 exp [ E0 + E ]

−∞

(2.152)

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

Quantum Mechanical Path Integral

x2 c − 2xo x + isx2 − 2isxo bo + o o

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

E(x)

=

x2 c + isb2 − f (x) o

.

c = cos ω(t − to ) ,

s = sin ω(t − to ) .

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

1 4

m 2πiω sinω(t − t0 )

1 2

eE(x)

∞

dx0 exp [ E0 (xo , x)]

−∞

(2.156)

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

2

(2.157)

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

1 2

(2.158)

.

(2.159)

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

+∞

dx0 e

−∞

E0 (x0 )

=

2πi sinω(t − t0 ) mω exp[iω(t − t0 )]

(2.160)

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

imω 2 s

(2.161)

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

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

s2 c + i mpo 2 s3 2ω

2

=

− mω 2

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

p2 o m2 ω 2 po s2 − 2i mω xc

2

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

=

− mω (x − cbo − 2 − i

2

po mω

s)2 +

i

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

po ( 2mω − 1 mωb2 )sc + o 2

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

t

33

dτ

to

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

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

o (t

− to ) −

1 2

p2 mωb2 o o − 2mω 2

sc +

1 bo po s2 2

(2.164)

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

t to

˙ dτ L[b(τ ), b(τ )] =

p2 mωb2 o o − 2mω 2

sc − bo po s

(2.165)

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

t to

˙ dτ L[b(τ ), b(τ )]

(2.166)

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

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

o

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

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

1 2

exp −

mω

[x − b(t)]2

(2.168)

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

34

Quantum Mechanical Path Integral

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

1 4

exp

−

mωx2 iω 0 − to 2 2

.

(2.169)

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

1 2

exp

−

mωx2 0

,

(2.170)

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

2.7

Stationary States of the Harmonic Oscillator

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

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

+∞

W (x, t) =

−∞

dxo φ(x, t|xo , to ) W (xo , to ) .

(2.172)

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

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

− 1 iωT 2

(2.173)

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

− 1 iωt 2

35

mω π (1 − e−2iωT )

1 2

+∞

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

−∞

(2.174)

(2.175)

+ 2xo E(x) = − mω 2 x2

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

2

(2.176)

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

2

2xe−iωT + 2 1 − e−2iωT

mω

e

−iωto

.

(2.177)

=

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

+∞

mω

e

−iωto

(2.178)

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

−∞ Eo (xo ,x)

=

π (1 − e−2iωT mω

1 2

(2.179)

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

1

(2.180)

mω

=

−

x2 − 4

mω mω

x e−iωt +

2 e−2iωt mω (2.181)

=

−

mω 2 x + 2 2

x e−iωt − e−2iωt

171) can be expanded in terms of e−inωt . t − to ) = φ(x.184) 1 ˜ exp[−iω(n + 1 ) t] φn (x) 2 n! ∞ +∞ = m=0 −∞ dxo Φ(x. results in 1 ˜ exp[−iω(n + 1 ) t] φn (x) 2 n! +∞ = −∞ dxo Φ(x.188) ˜ ˜ Equation (2. Accoordingly. xo .171) concludes the proof of (2.147) in (2. t|xo .186) Fourier transform. t|xo . to ) exp[−iω(n + 1 ) to ] φn (xo ) . t) +∞ −∞ dto 1 ˜ exp[−iω(m + 1 ) to ] φm (xo ) 2 m! (2. t|xo . We want to inspect the consequences of the invariance property (2. .s.172) in the form ∞ n=0 (2.171.172) I = exp 2 mω x e−iωt − e−2iωt − Quantum Mechanical Path Integral mω 2 1 x − iωt 2 2 . xo .172). t − to ) 1 ˜ exp[−iω(m + 1 ) to ] φm (xo ) 2 m! (2. n = 1.188) identiﬁes the functions ψn (x. t) = exp[−iω(n + 1 ) t] φn (x) as invariants under the 2 action of the propagator φ(x.. 2. to ) we express (2. but not of t. which also exhibits such invariance.h.185) Replacing t → t + to yields ∞ n=0 1 ˜ exp[−iω(n + 1 ) (t + to )] φn (x) 2 n! ∞ +∞ = m=0 −∞ dxo Φ(x. of (2.36 Altogether. t). .183) where the expansion coeﬃcients are functions of x.172) is a function of t − to and deﬁning accordingly Φ(x. one obtains for the r. 2. Noting that the propagator (2. t) = n=0 1 ˜ exp[ − iω(n + 1 ) t ] φn (x) 2 n! (2. i. xo .. to ).172).182) Comparision with (2. one can expand (2.187) or 1 ˜ exp[−iω(n + 2 ) t] φn (x) +∞ = −∞ ˜ dxo φ(x. We note that the factor exp(2 mω/ xe−iωt − e−2iωt ) in (2.e. t − to ) 1 ˜ φn (xo ) n! (2. .171) ∞ W (x. 2 (2. (2. In contrast to W (x. . 1 exp[iω(n + 2 ) to ] · · · . xo .

For this purpose we introduce ﬁrst closed expressions for φ the new variables y z and write (2. z) = Hn (y) (2.194). according to (2. Actually. t) = z 2 e−y where w(y.189) ˜ dx φ2 (x) = 1 n (2.194) plays a central role for the Hermite polynomials since it contains. (2.171) W (x. Expression (2.5: Propagator for a Quadratic Lagrangian 37 ˜ ˜ the functions ψn (x. t) provide all stationary states of the quantum mechanical harmonic oscillator.183). t)|2 = Nn +∞ −∞ n = 0. z) = exp(2yz − z 2 ) .194) = z2 n=0 ∞ 1 ∞ zn ˜ φn (y) n! (2. In the following we will characterize the functions φn (x) and determine the normalization constants Nn .197) are called Hermite polynomials which are polynomials of degree n which will be evaluated below. . z) = n=0 1 2 /2 1 2 /2 = = mω e−iωt x (2.171). ˜ |φ stationary wave functions ψn (x.183).192) w(y.198) ∂z n z=0 . Expansion (2. t) = exp[−iω(n + 1 ) t] Nn φn (x) . z) (2. characterizes the wave functions φn (x).190) ˜ is obeyed. The Hermite Polynomials ˜ The function (2. t)|2 = ˜n (x)|2 .191) (2.195) zn Hn (y) n! ˜ φn (y) . This follows from ∂n w(y. t) such that +∞ −∞ 2 dx |ψ(x. 2. t) of the quantum mechanical harmonic oscillator ˜ ψn (x.197) The expansion coeﬃcients Hn (y) in (2.183). through expansion (2. we have identiﬁed then. To obtain ˜n (x) we simplify the expansion (2. We will also argue that the functions ψn (x. . (2. 2 Here Nn are constants which normalize ψn (x. through the expansion coeﬃcients φn (x) in (2.2.183) reads then w(y. 1. t) are associated with a time-independent probablity density |ψn (x.193) (2. .196) where Hn (y) = ey 2 /2 (2. z) z 2 e−y or w(y. in a ‘nutshell’ all information on the Hermite polynomials.

The diagrams also identify the areas over which the summation is to be carried out. and O. among other properties.199) Comparision with (2. .S. ν+µ ν H3 (y) = 8y 3 − 12y. normalization factors for φ(y). and recursion equations for the eﬃcient evaluation of Hn (y). closed expressions for Hn (y). ∂y n (2. .196).202) generatingfunctionology by H.1: Schematic representation of change of summation variables ν and µ to n = ν + µ and m = ν −µ.38 Quantum Mechanical Path Integral which is a direct consequence of (2.200) yields for the ﬁrst Hermite polynomials H0 (y) = 1. ν+µ (2. z) the generating function for the Hermite polynomials.194.200) One can deduce from this expression the polynomial character of Hn (y).Graham. Massachusetts..e.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). The diagrams illustrate that a summation over all points in a ν. (2.Patashnik (AddisonWesley. that Hn (y) is a polynomial of degree n. m lattice (right diagram). 2.201) ν−µ µ µ ν−µ Figure 2. Reading. Boston.196) results in the so-called Rodrigues formula for the Hermite polynomials Hn (y) = (−1)n ey 2 ∂ n −y2 e . i. One calls w(y. We want to derive now explicit expressions for the Hermite polynomials.Knuth. . z) ∂z n = = z=0 2 ∂ n 2 y z − z2 e ∂z n e−(y−z) 2 ey z=0 2 ∂ n −(y−z)2 e ∂z n 2 z=0 2 (−1)n ey ∂n ∂y n = (−1)n ey z=0 ∂n ∂y n e−y (2.E.198) for the Hermite polynomials can be expressed in a more convenient form employing deﬁnition (2. The identity (2. As will become evident in the present case generating functions provide an extremely elegant access to the special functions of Mathematical Physics3 .196) ∂n w(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. We will employ (2. H1 (y) = 2y. 1990) is a useful introduction to this tool as is a chapter in the eminently useful Concrete Mathematics by R.Wilf (Academic Press. 1989). ν H2 (y) = 4y 2 − 2. Inc. For this purpose we expand the generating function (2.196) to ˜ derive. D.L. µ lattice (left diagram) corresponds to a summation over only every other point in an n.. .

m must be restricted such that either both n and m are even or both n and m are odd. 2 (2. In case of even n .207) one can deduce that Hn (y). From (2. (2. . it contains only odd powers. p ≤ x.207) contains only even powers.201).209) This property follows also from the generating function. 2 µ = n−m . i..209).205) Since (n − m)/2 is an integer we can introduce now the summation variable k = (n − m)/2 . H3 (y) = 8y 3 − 12y. is a polynomial of degree n. in fact. 0 ≤ k ≤ [n/2] where [x] denotes the largest integer p. the sum in (2. H2 (y) = 4y 2 − 2. 2. 0 ≤ m ≤ n .203) The old summation variables ν. Hence. for odd n. 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 .197) ∞ n=0 zn Hn (−y) = n! ∞ n=0 (−z)n Hn (y) = n! ∞ n=0 zn (−1)n Hn (y) n! (2. it holds Hn (−y) = (−1)n Hn (y) . (2. 39 (2. The lattices representing the summation terms are shown in Fig. m = ν − µ 0 ≤ n < ∞.204) Since ν.5: Propagator for a Quadratic Lagrangian and introduce now new summation variables n = ν + µ. otherwise.208) which agrees with the expressions in (2. −z) and. µ are integers the summation over n. (2.206) From this expansion we can identify Hn (y) [n/2] Hn (y) = k=0 (−1)k n! (2y)n−2k . H1 (y) = 2y.e.207) This expression yields for the ﬁrst four Hermite polynomials H0 (y) = 1. according to 2.210) from which one can conclude the property (2. µ expressend in terms of n.202) e 2yz − z 2 ∞ = n=0 zn n! ≤n m≥0 n! (−1) n−m 2 n−m 2 ! m! (2y)m .1.194) holds w(−y. With this restriction in mind one can express (2. hence. According to (2. m are ν = n+m . k! (n − 2k)! (2. .2. k! (n − 2k)! = Hn (y) (2. z) = w(y. . .

194). . H2n+1 (0) = 0 .188. Hn+1 (y) − 2y Hn (y) + 2n Hn−1 (y) = 0 . have a node at y = 0. 1. (2. .211) m! n! m=0 n=0 Comparing terms on both sides of the equation yields (2n)! .217) gives H1 (y) − 2y H0 (y) = 0. . . a property which is consistent with (2. . One can then readily state which diﬀerential equation of w(y.213) which allow one to evaluate Hn (y) from H0 (y) and H1 (y) given by (2. n! (2. . . z) = exp(−z 2 ) and carries out the Taylor expansion on both sides of this expression resulting in ∞ ∞ (−1)m z 2m zn = Hn (0) . dw/dy − 2zw = 0.218) The reader should recognize the connection between the pattern of the diﬀerential equation (??) and the pattern of the recursion equation (??): a diﬀerential operator d/dz increases the order n of Hn by one. 2.212) n! This implies that stationary states of the harmonic oscillator φ2n+1 (x).214) We want to derive (??) using the generating function.194). 2. n = 1. For this purpose one considers w(0.215) ∂z which can be readily veriﬁed using (2. n = 0.209) since odd functions have a node at the origin. . as deﬁned through (2. for example. . indeed satisﬁes the latter relationship. 2. n = 1. Another relationship is d Hn (y) = 2n Hn−1 (y). For Hermite polynomials holds. The reader may verify that w(y. (2. Starting point of the derivation is the property of w(y. (2. . z) − (2y − 2z) w(y.208).197) above and given by (2. as given in (2. (2.40 Quantum Mechanical Path Integral The generating function allows one to determine the values of Hn (y) at y = 0. z). dy n = 1. (2. 2. 2. . H2n (0) = (−1)n Recursion Relationships A useful set of properties for special functions are the so-called recursion relationships. . Hn+1 (y) − 2y Hn (y) + 2n Hn−1 (y) = 0.196) into the diﬀerential equation (2. Substituting expansion (2. a factor z reduces the order of Hn by one and introduces also a factor n.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. namely.233) below. z) = 0 (2. . z) ∂ w(y.215) yields ∞ n=1 z n−1 Hn (y) − 2y (n − 1)! ∞ n=0 zn Hn (y) + 2 n! ∞ n=0 z n+1 Hn (y) = 0 . z) should be equivalent to the relationship (??).

221) 1 = √ π +∞ −∞ dt e2iyt − t .219) which can be written I(y) = e−y 2 +∞ −∞ dte−(t−iy) = e−y √ 2 2 +∞ −∞ dze−z .247) for a = i one obtains I(y) = and. dy n (2. 2 (2. (2.194. 2 n = 0.h. . z ) e−y 2 +∞ = e2 z z = √ ∞ dy e−(y−z−z ) = . 1. 2. 2. 2.h.224) +∞ −∞ dt tn e2iy t − t .5: Propagator for a Quadratic Lagrangian Integral Representation of Hermite Polynomials An integral representation of the Hermite polynomials can be derived starting from the integral +∞ 41 I(y) = −∞ dt e2iy t − t . z) w(y.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! (2.227) Comparing the terms of the expansions allows one to conclude the orthonormality conditions +∞ −∞ dy Hn (y) Hn (y) e−y 2 √ = 2n n! π δn n .226a).225) We want to derive from the generating function (2.223) dn 2iy t − t2 2 e = (2 i t)n e2iy t − t n dy results.s. of the Rodrigues formula (2.196) yields ∞ n.220) Using (2.194.228) . 2 (2. through a double series over Hermite polynomials using (2.2.196) the orthogonality properties of the Hermite polynomials. 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. 2 (2. acording to the deﬁnition (2. .226) π −∞ 2n z n z n n=0 n! Expressing the l.s. ﬁnally.200) yields Hn (y) = The identity (−1)n y2 √ e π +∞ dt −∞ dn 2iyt − t2 e . . 2 √ π e2 z z (2.222) Employing this expression now on the r. For this purpose we consider the integral +∞ −∞ dy w(y.

233) .190)] φn (y) = Nn e−y 2 /2 and for the normalization constants Nn follows from (2.42 Quantum Mechanical Path Integral E 9/2 h ω ¥ ¥ ¥ ¥ ¥ 7/2 h ω 5/2 h ω ¤ 3/2 h ω £ 1/2 h ω ¦ Figure 2. (2. 4 and odd for n = 1.189.228) +∞ 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 (2.229) Hn (y) . 4.232) Hn (y) . 3. The normalized states are [c.233) are presented for n = 0. (2. 2.230) (2.197) holds 2 ˜ φn (y) = e−y /2 Hn (y) . the value of the wave function at y = 0 is positive for n = 0. 3. Furthermore. in harmony with (??). According to (2. 4.f. 1. ¡ ¢ ¡ -4 -2 0 2 4 y (2. that the wave functions are even for n = 0. One can also recognize that n is equal to the number of nodes of the wave function. 2. in agreement with our above discussions.2. Normalized Stationary States The orthonormality conditions (2. 3.231) √ π (2.228) allow us to construct normalized stationary states of the harmonic oscillator. One can recognize. 2. 2. 1. (2. 2. 4 in Fig.2: Stationary states φn (y) of the harmonic oscillator for n = 0. negative for n = 2 and vanishes for n = 1. 3.

In case of the Legendre polynomials.. (2. and In = 2n n! π. of functions which obey dx f 2 (x) w(x) = < ∞ . 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. g ∈ F . 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. using (2. f.236 . ??? and eq.189) by the Jacobian dx/dy. w(x) = 0 only at a discrete set of points xk ∈ Ω (2. Other examples of orthogonal polynomials are the Legendre and Jacobi polynomials which arise in solving three-dimensional stationary Schr¨dinger equations. The latter are written for polynomials pn (x) in the general form dx pn (x) pm (x) w(x) = In δnm Ω (2.. i.10 and eq. holds Ω = [0.228) shows that the orthonogality condition of the Hermite polynomials is in complience with (2. w(x) ≡ 1. 2.2. (10. w(x) = xα e−x .233) and (2.150 . +∞[.234) The explicit form of the stationary states of the harmonic oscillator in terms of the position variable x is then.238) where the space is endowed with the scalar product (f |g) = Ω dx f (x) g(x) w(x) = < ∞ .228).239) . 5. 5. 5.151.235) 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. 10.236) where w(x) is a so-called weight function with the property w(x) ≥ 0.237) for Ω = R.189) φn (x) = √ 1 2n n! mω π 1 4 e− mωx2 2 Hn ( mω x) .e. Ω (2. Comparision with (2. The Hermite polynomials are so-called orthonogal polynomials since they obey the conditions (2. and I = 2/(2 + 1). ???] and relativistic [see Sect. denoted by P (x) and introduced in Sect. 5 below [c.f.237) and where In denotes some constants. (2.e. by dx = dy mω 1 4 .459] hydrogen atom. i. (5. In = Γ(n+α+1)/n! where Γ(z) is the so-called Gamma function.179] holds Ω = [−1. denoted by Ln (x) and encountered in case of the stationary states of the nonrelativistic [see Sect. In case of the associated Laguerre (α) polynomials.156.5: Propagator for a Quadratic Lagrangian 43 The normalization condition (2. The orthogonal polynomials pn mentioned above have the important property that they form a complete basis in the space F of normalizable functions. 1]. w(x) = √ exp(−x2 ). (2.231) of the wave functions diﬀers from that postulated in (2.

e. (2. t) = n=0 1 2 dn exp[−iω(n + )t] e−y /2 Hn (y) .. t) = n=0 cn (t) e−y 2 /2 Hn (y) . det(a) (2. .e. t) is ∞ ψ(y. Therefore. hence.246) for det(a) = 0 and real. a (2.247) . 2. a state for which (2. (2. i.244) For the state to be stationary |ψ(x.241) The latter identity follows from (2.. Since the Hermite polynomials form a complete basis for such states we can expand ∞ ψ(y.k yj ajk yk = (iπ)n .245) must be time-independent.242) and the existence of a normalizable state ψ(y. aT = a.197) it must hold cn (t) = dn exp[−iω(n + 1 )t] and. in particular. The only possibility for this to be true is dn = 0.188.239) becomes the conventional scalar product of quantum mechanics f |g = Ω dx f (x) g(x) .172) holds. symmetric a.233).147).e. t) must be identical to one of the stationary states (2. . i. Ω (2. ψ(y.243) To be consistent with(2. t)|2 . i. ∞ n. i.233) exhaust all stationary states of the harmonic oscillator.m=0 d∗ dm exp[iω(m − n)t] e−y Hn (y)Hm (y) . the 2 stationary state ψ(y. If one replaces for all f ∈ F: f (x) → w(x) f (x) and.236).240) where cn = 1 In dx w(x) f (x) pn (x) .e. the states (2.. 2 (2. −∞ dyn ei −∞ n j. pn (x) → w(x) pn (x) the scalar product (2. except for a single n = no . In case of n = 1 this reads +∞ −∞ dx ei a x = 2 iπ . t) which is stationary under the action of the harmonic oscillator propagator (2. n 2 (2. Appendix: Exponential Integral We want to prove +∞ +∞ I = dy1 .44 As a result holds for any f ∈ F f (x) = n Quantum Mechanical Path Integral cn pn (x) (2.242) Let us assume now the case of a function space governed by the norm (2.

. 0 ˜ ˜ S−1 a S = a = . (2. This property allows one to simplify the bilinear ˜ form n yj ajk yk by introducing new integration variables j.7: Appendix / Exponential Integral 45 which holds for a ∈ C as long as a = 0. .e.249) n ajk = ˜ l.248.k n n yj = ˜ k (S −1 )jk yk . . .249) The akk are the eigenvalues of a and are real. (2.k yj ajk yk ˜˜ ˜ (2. .248) . ˜ (2.254) ajj .246) exploits that for any real.252) ˜ For the determinant of a holds det(˜) = a n ajj ˜ j=1 (2. .k n m n y Sj ajk Skm ym ˜ ˜ y (S T ) j ajk Skm ym ˜ ˜ j. (2. One can conclude det(a) = j=1 n (2. symmetric matrix exists a similarity transformation such that a11 0 . . i.255) . . ST S = 1 1 or S = S−1 . ann ˜ where S can be chosen as an orthonormal transformation.246) reads then in terms of yj ˜ n n j.. 2. . .k n m = = j. . yk = k Skj yj . . det(S) a = (det(S))−1 det(a) det(S) = det(a) .k yj ajk yk n = j. . . 0 ˜ 0 a22 . ˜ (2. .2. The proof of (2. according to (2..250) The bilinear form in (2. 0 0 .253) as well as det(˜) = det(S−1 aS) = det(S−1 ) det(a) . .m (S T )jl alm Smk .251) where.

. . .256) such that none of the eigenvalues of a vanishes. .261) ∂(y1 . holds n Quantum Mechanical Path Integral ajj = 0 ˜ j=1 (2. . .262) (2. . .. . yn ) y ˜ ∂yj . ∂ ys ˜ (2. .. . .250) holds J = S and.263) (2. d˜n e y −∞ = (2.257) Substitution of the integration variables (2. . . . det( From (2. ˜ for j = 1. yn ) y ˜ ei n k akk yk ˜ ˜2 . y −∞ d˜n det y −∞ ∂(y1 . . y −∞ +∞ d˜n ei y −∞ n k akk yk ˜ ˜2 +∞ i˜11 y1 a ˜2 n i˜nn yn a ˜2 +∞ a ˜ d˜k ei˜kk yk y 2 = d˜1 e y −∞ .249) follows 1 = det ST S such that one can conclude det S = ±1 One can right then (2. ∂(˜1 . . . (2. .259) (2. yn ) y ˜ = ( det S )2 (2. .258) where we introduced the Jacobian matrix J = with elements Jjs = According to (2. .250) allows one to express (2. 2.. yn ) ∂(˜1 .258) +∞ +∞ ∂(y1 .250) +∞ +∞ I = d˜1 . yn ) ) = det(S) .260) (2. . n (2. . . .266) . . i. ajj = 0 . . .e.46 We have assumed det(a) = 0.265) k=1 −∞ which leaves us to determine integrals of the type +∞ dx eicx −∞ 2 (2. . hence. Accordingly. . yn ) ∂(˜1 .264) I = d˜1 . . .

p ∈ IR.268) vanishes.2. the integrand does not exhibit any singularities anywhere in C.c > 0 .266) to the well-known Gaussian integral +∞ dx e−cx −∞ 2 = π . for x. i.268) can be written as the sum of the following path integrals J = J1 + J2 + J3 + J4 .. (2.270) dx e−cx 2 2 − 2p √ γ4 : z = ix − p J4 = −p i dx eic(ix−p) . We consider ﬁrst the case c > 0 and discuss the case c < 0 further below.7: Appendix / Exponential Integral 47 where.267) by considering the contour integral J = γ dz eicz = 0 2 (2.268) along the path γ = γ1 + γ2 + γ3 + γ4 displayed in Figure 2. One can relate integral (2. 2 since eicz is a holomorphic function. The contour intergral (2.3.257) holds c = 0.269) 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. according to (2.271) .e. Jk = γk dz eicz 2 (2. The contour integral (2. c (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 .

275) . (2. Hence.272) = 0 .3: Contour path γ in the complex plain. (2.274) (2.268) for p = +∞.273) p→+∞ lim J2 and J4 do not contribute then to integral (2. p p→+∞ lim |J2 or 4 | ≤ = p→+∞ 0 lim dx |e−2cxp | 1 − e−2cp 2cp = 0. We will now show that the two integrals J2 and J4 vanish for p → +∞. c ¡ -p γ1 p Re(z) 2 | |e−2cxp | . This follows from the following calculation p p→+∞ lim |J2 or 4 | = p→+∞ lim | 0 p i dx eic(ix+p) | ≤ It holds |e ic(p2 −x2 ) p→+∞ 0 lim |i| dx |eic(p 2 −x2 ) | = 1 since the exponent of e is purely imaginary. (2.267) one has shown then ∞ dx eicx −∞ 2 = iπ . One can state accordingly ∞ J = dx e −∞ icx2 − √ ∞ i dx e−cx −∞ 2 Using 2.48 Quantum Mechanical Path Integral Im(z) ip γ2 ¡ γ4 ¡ γ3 ¡ -i p Figure 2.

It holds n I = k=1 iπ = akk ˜ (iπ)n .275.277) We apply the above results (2.279) .2. if one chooses the same contour integral as (2.278) Noting (2. 2.255) this result can be expressed in terms of the matrix a I = which concludes our proof.7: Appendix / Exponential Integral 49 One can derive the same result for c < 0.276) and (c < 0) ∞ dx eicx −∞ 2 = −iπ = −|c| iπ .265). n ˜ j=1 ajj (2.277) to (2. but with a path γ that is reﬂected at the real axis. c (2.268). (iπ)n det(a) (2. This leads to ∞ icx2 J = dx e −∞ + √ −∞ −i ∞ dx ecx = 0 2 (2.

50 Quantum Mechanical Path Integral .

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

1 2 3 2 ∂2 j=1 xj ∂x2 ψ(r. only terms of the expansion (3.k=1 ∂2 ψ(r. t). t) j 1 12 . 2 (3.4) one identiﬁes 2 1 = a m = O( ) (3. t) + j=1 xj ∂ 1 ψ(r. n = 0.6) make contributions of the order O( 3 2 ) . Since the kinetic energy contribution in (3.4) is even in all three coordinates x1 .g. t) and the kinetic energy term. x3 . the terms collected in (3. consequently.. a (3.7) According to (2. O( 7 2 ). t) j . (3.7) shows 1 ∂ In (a).52 Schr¨dinger Equation o It is important to note that the integration is not over r.6) 3 4 ∂4 j=1 xj ∂x4 ψ(r. . (3. According to (3. a supposition which will be examined below. x2 . t) + ∂xj 2 3 xj xk j. The integration involves only the wave function ψ(r +s.12) Here one needs to note that we are actually dealing with a three-fold integral.5) assuming that only the leading terms contribute. Obviously. ∂xj ∂xk (3. t) . x2 .k=1 xj xk ∂x2 ∂x2 ψ(r. t) + .10) It is now important to note that in case of integral (3.4) only for small x2 + x2 + x2 values 1 2 3 we expand 3 ψ(r + s.9) iπ a (3.11) holds 3 m 3 iπ 2 2 × = 1 (3. of the expansion of ψ(r + s.. 1 4 3 2 2 ∂4 j. . O( 5 2 ) .. i ∂a Starting from (3. O( 7 2 ) . x3 )T .8) (3. U (r.13) 2πi a . we need then to evaluate integrals of the type +∞ In (a) = −∞ dx x2n exp i a x2 . t) is a constant with respect to this integration.. t) = ψ(r. It is then suﬃcient to consider the terms ψ(r.5) which are even separately in all three coordinates yield non-vanishing contributions. t) j k .11) and. but over s = (x1 .. 1.36) holds I0 (a) = Inspection of (3. a I2 (a) = − 3 4a2 iπ . It holds In+1 (a) = I1 (a) = i 2a iπ ..8) one can evaluate recursively all integrals In (a). Since the latter contributes to (3. e.

t) ψ(r. a solution is thought for t ≥ t0 and the solution is speciﬁed at the intial time t0 . Usually. t + ) = ψ(r. t)ψ(r.14) This expansion in terms of powers of suggests that we also expand ∂ ψ(r. t) + O( 2 ) . Obviously. t) = H ψ(r.3. using (3.14) results in ψ(r. t) + 1 2i 4 m 2 53 ψ(r.15) ψ(r.18) This is the celebrated time-dependent Schr¨dinger equation. t1 ) = f (r). t) + and exp − i U (r.19) ∂t where 2 2 ˆ H = − + U (r.2 Boundary Conditions The time-dependent Schr¨dinger equation is a partial diﬀerential equation. 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 . 2nd o order in the spatial variables and linear in the solution ψ(r. ψ(r. t) + O( 2 ) . t) ψ(r. We will derive brieﬂy the type of boundary conditions encountered.16) Inserting this into (3. t) + ∂ ∂t ψ(r.2: Boundary Conditions and one can conclude. This equation is often written in the o form ∂ ˆ i ψ(r. The 2nd order spatial derivatives require that one speciﬁes also properties of the solution on a closed boundary ∂Ω surrounding the volume Ω in which a solution is to be determined.e. Due to its linear character any linear combination of solutions of the time-dependent Schr¨dinger o equation is also a solution. t) (3. t).20) 2m 3. this equation is trivially satisﬁed to order O( 0 ).. In order O( ) the equation reads i ∂ ψ(r. t) . t) + O( 2 ) ∂t i U (r. e.. t + ) = exp − i U (r. The following general remarks can be made about the solution.17) 2m ψ(r. t) . 1st order in time. t) = 1 − (3. (3. The 1st order time derivative requires that for any solution a single temporal condition needs to be speciﬁed. t) = ∂t 2 − 2 2m + U (r. (3. t) + O( 2 ) . ψ(r. t) (3.10). t) = ψ(r.g. (3. i. t) − + i 2 2 i U (r. one speciﬁes the so-called initial condition. (3.

The diﬀerence between the integrals is g| H |f = Ω Ω = − g| H |f d3 rf ∗ (r) − Ω 2 2 2m g(r) − Ω d3 rg(r) 2 − 2 2m f ∗ (r) + Ω 2 d3 rf ∗ (r) U (r.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.. g ∈ H is deﬁned as follows f |g Ω = d3 rf ∗ (r)g(r) (3.28) See. 2nd Ed. (3. g|r) (3.26) the vector–valued function P (f ∗ . .23) ˆ Since H is a diﬀerential operator the expressions (3. g|r) = − 2 2m ( f ∗ (r) g(r) − g(r) f ∗ (r) ) (3.23). t) g(r) − Ω d3 rg(r) U (r. Chapter 1. g|r) is called the ˆ concomitant of H and is P (f ∗ . J. (John Wiley. in principle. Since energy is a real ˆ ˆ quantity one needs to require that the eigenvalues of the operator H are real and. implies hermitian ˆ f | H |g 1 2 Ω ˆ = g| H |f Ω (3.20). however. The reader is advised to consult a reference text on ‘Linear Algebra’ to follow this argument.22) ˆ where H is deﬁned in (3.22) and (3. for example. Interchanging f ∗ (r) and g(r) results in g| H |f Ω = Ω ˆ d3 r g(r) Hf ∗ (r) . The hermitian property. f ∗ (r) d3 rg(r) Ω 2 = − 2m d3 rf ∗ (r) Ω 2 g(r) − f ∗ (r) (3. Jackson. that H is 2 .27) ˆ We will postulate below that H is an operator in H which represents energy. Calssical Electrodynamics. t). New York.21) Ω This leads one to consider the integral f | H |g Ω = Ω ˆ d3 rf ∗ (r) H g(r) (3. hence. In (3. 1975).26) where ∂Ω da · A(r) denotes an integral over the surface ∂Ω of the volume Ω. diﬀer from each other.54 Schr¨dinger Equation o linear vector spaces of functions f (r) in which a scalar product between two elements f. the surface elements da pointing out of the surface in a direction normal to the surface and the vector–valued function A(r) is taken at points r ∈ ∂Ω. D.25) ˆ f | H |g Ω ˆ = g| H |f Ω + ∂Ω da · P (f ∗ .

29. A most common boundary condition is encountered for the volume Ω = R3 in which case one postulates lim f (r) = 0 “natural boundary condition” . in this case also all derivatives of f (r) vanish at inﬁnity. (3..e. (3.32) the wave function must decay for |r| → ∞ rapidly enough to be square integrable.30) can be postulated. t)∂t ψ ∗ (r. t)|2 . . we can only allow functions which make the diﬀerential da · P (f ∗ .e. In either case does f (r) and all of its derivatives vanish asymptotically. namely. Since the total probability of ﬁnding the particle anywhere in space is d3 r|f (r)|2 = 1 (3. to avoid discontinuities in ψ(r. The probability to observe the particle anywhere in the subvolume ω ⊂ Ω is p(ω. t) which describes the state of o a particle moving in the potential U (r.30) each condition holding on an entire surface ∂Ωj . ∂Ω = ∂Ω1 ∩ ∂Ω2 ∩ ∂Ω3 ∩ .g. κ > 0 or like r−α . Often the closed surface of a volume ∂Ω is the union of disconnected surfaces3 ∂Ωj . and that the density decays rapidly when one moves away from that area. |ψ(r. (3.32). t) on a single connected surface ∂Ωj only either one of the conditions (3. . in which case ∂Ω1 is the inner sphere and ∂Ω2 is the outer sphere. 3..33) The time derivative of p(ω.3 Particle Flux and Schr¨dinger Equation o The solution of the Schr¨dinger equation is the wave function ψ(r.31) usually arises when a particle existing in a bound state is described.. g|r) vanish on ∂Ω. . r ∈ ∂Ω or da · f (r) = 0 ∀r.30) Note that these boundary conditions are linear in f .3: Particle Flux 55 and. therefore. like exp(−κr). α > 2. i.29. t)∂t ψ(r. The latter property stems from the fact that the boundary condition (3. In this case one can expect that the particle density is localized in the area where the energy of the particle exceeds the potential eenrgy. 3. However. 3. e. if f and g satisfy these conditions than also does any linear combination αf + βg. t) ] . t)|2 ..3.29) In fact. It must hold then for all f ∈ H f (r) = 0 ∀r. i. t) is ∂t p(ω. t) = ω 3 d3 r [ ψ ∗ (r. t) = ω d3 r |ψ(r. t) + ψ(r.e.31) |r|→∞ (3. r ∈ ∂Ω (3. t). obey (3. In this case one can postulate both conditions (3.34) An example is a volume between two concentric spheres. The observable directly linked to the wave function is the probability to ﬁnd the particle at position r at time t. i.

44) ˆ Note that the Hamiltonian H involves only real terms. t) ∂t ˆ ˆ ψ ∗ (r.37) can be expressed through a volume integral according to da · A(r) = ∂ω ω (3. t) = |ψ(r. t) = ∂ω da · P (ψ ∗ (r. t) ψ(r. t) such that d3 r|ψ(r. 3. t) = H ψ ∗ (r. t) H ψ ∗ (r. t) − ψ(r. t) = ω Schr¨dinger Equation o ∂ ∗ ˆ ψ (r. t) = 0.30) ∂t p(Ω.18) by any complex number and accordingly one can deﬁne ψ(r. (3. t) ψ ∗ (r. t) (3.26. 3. Using (3. Accordingly the probability to observe the particle anywhere in the total volume Ω is constant. t)r. The surface integral (3. t) + 4 · j(r. t)|2 is a probability density with units 1/volume. t) H ψ(r. . t) . t) = [ ψ(r. t) . We will assume in the remainder of this Section that the solutions discussed are normalized.35) d3 r .37) d3 r ∂t ρ(r. One refers to such solution as normalized. Note that for a normalized wave function the quantity ρ(r. t) = P (ψ ∗ . (3. t) − ψ ∗ (r.43) (3.40) One can rewrite then (3. t) ] .41) holds for any volume ω ⊂ Ω one can conclude ∂t ρ(r. ψ|r. t) = 0 (3.19) and its conjugate complex4 −i yields i ∂t p(ω.27) this can be written i ∂t p(ω.29. t) = 0 .42) 2mi Since (3.36) According to (3.39) d3 r · A(r) (3. ψ(r. (3.56 Using (3.41) where j(r.37) If one applies this identity to ω = Ω one obtains according to (3. (3. A natural choice for this constant is 1.38) holds. t)|2 = 1 Ω (3.27) one can express this j(r. t). t) + ω · j(r. One can multiply the solution of (3.

(2. t) = − ψ(r. j(r. equal to the velocity of the particle.3. t) = 0 2 ∂ 2 i ψ(r. (2. 2. (3. Such case arose in Sect. t) .48) that the general solution is of the form ψ(r.45) Obviously.167)] potential. 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) .105. It is of interest to note from (3.f. m (3.e. t) points in the direction to the outside of volume ω. This follows directly from an inspection of Eq. (3.125)] and in a quadratic [c.48) ∂t 2m which describes the motion of free particles. 2. arises solely from the complex factor exp(i k · r). the initial condition at ψ(r.f. t) has vanishing ﬂux anywhere.148. t) is that of density ﬂux..49) ˜ Obviously. 2 for a free particle [c.71)].48. (3.e.47) .41) written in the form ∂t ω d3 r ρ(r. (3. One can readily show by insertion into (3. t) gives rise to a descrease of the total probability in volume ω due to the disappearence of probability density at the surface ∂ω. t0 ) = [2π]− 2 Ω∞ 3 ˜ d3 k φ(k) exp i(k · r − ωt0 ) .49) reads at t = t0 ψ(r. t) = − ∂ω da · j(r. This ﬁnding is consistent with the present evaluation of the particle ﬂux: a factor exp(ipo xo / ) gives rise to a ﬂux po /m. 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.43) that any real wave function ψ(r.f.. 2m (3. (2. t0 ) determines φ(k). t) = [2π]− 2 where the dispersion relationship holds ω = k2 .4 Solution of the Free Particle Schr¨dinger Equation o We want to consider now solutions of the Schr¨dinger equation (3. 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. and for particles moving in a linear [c. for f (r) ∈ R.4: Free Particle 57 The interpretation of j(r. 3. 2. In Sect. Equation (3. i.18) in Ω∞ = R in the case o U (r.51) . Note that j(r. t) (3.50) 3 ˜ d3 k φ(k) exp i(k · r − ωt) Ω∞ (3.46) i.

that the “natural boundary condition” (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 (3.5). 3.81) . hence. The solution as stated is deﬁned in the inﬁnte space Ω∞ = R . Comparision with Path Integral Formulation One can write solution (3. derive an expression similar to (3. t0 ) (3. If one chooses the initial state f (r) deﬁned in (3. The general form for this propagator involves an expansion in terms of a complete set of eigenfunctions as in (3.31) applies. in (2. respectively. t|r0 . t0 ) = 1 2π 3 d3 k exp Ω∞ ik · (r − r0 ) − i 2 k2 2m (t − t0 ) . i.51.. t) = Ω∞ d3 r0 φ(r. We have identiﬁed then with (3.54) yields (2. The ensuing solutions are called wave packets. t|r0 . evaluating the integral in (3. t0 ) ψ(r0 . t) as given by (3.49) is square integrable at all subsequent times and.47).47).e. Ω∞ (3.51) to be square integrable it follows according to the properties of the Fourier–transform that ψ(r.247).70) derived below for a particle in a box and the harmonic oscillator. i.114) and (4..54) an alternative representation of the propagator (2.56) and using (2. In case of the harmonoc oscillator the expansion can be stated in a closed form given in (4. Below we will generalize the propagator (3.49.53) where φ(r.49).54) valid for this case. To show this one needs to note 1 2π = +∞ dk1 exp −∞ ik1 (x − x0 ) − 1 2 i 2 k2 1 2m . 3.58 The inverse Fourier transform yields 3 ˜ φ(k) = [2π]− 2 Schr¨dinger Equation o d3 r0 exp(−ik · r0 ) ψ(r0 . (3.e. (t − t0 ) (3. t0 ) .54) This expression obviously has the same form as postulated in the path integral formulation of Quantum Mechanics introduced above.52) We have not speciﬁed the spatial boundary condition in case of (3. In fact.52) above ψ(r.54) to the case of non-vanishing potentials U (r).

(3. r ∈ ∂Ω or da · φ(r) = 0 ∀r. t) = −∞ dx0 φ(x. (3.57–3.4: Free Particle Free Particle at Rest 59 We want to apply solution (3. i. We have demonstrated then in this case that the Schr¨dinger formulation of Quantum o Mechanics is equivalent to the Feynman path integral formulation.29. r ∈ ∂Ω (3.60) =???? Combining (3. Stationary States We consider now solutions of the time-dependent Schr¨dinger equation (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. Accordingly.3.57) where φ(x. 3.59) which is solved through completion of the square in the exponent [c.64) (3..54) is +∞ ψ(x.f.27) vanish on the surface ∂Ω of Ω. a result which is identical to the expression (??) obtained by means of the path integral propagator (2.61) × exp − (x − po m t)2 2δ 2 (1 + 2 t2 ) m2 δ 4 (1 − i . t0 ) = 1 2π +∞ dk exp −∞ ik (x − x0 ) − i 2m (t − t0 ) . (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 (3.55.19. t) is given by (??).49. (3. t|x0 . 3. 3.60) yields with t0 = 0 ψ(x.63) .58) Integration over x0 leads to the integral +∞ dx0 exp −ikx0 + −∞ i po x0 − x2 0 2δ 2 = √ 2πδ 2 exp − (k − po 2 2 ) δ 2 (3. t|x0 .52) to the case that the initial state of a 1-dimensional free particle ψ(x0 . the boundary conditions are φ(r) = 0 ∀r.56)]. The 1-dimensional version of (3.e. the functions obey boundary conditions of the type (3. t) = f (t) φ(r) . t0 ) ψ(x0 . t) 2 k2 (3. 3.20) which are of o the form ψ(r.62) We will restrict the space of allowed solutions to a volume Ω such that the functions also make the concomitant (3.53.30).46).

62) do exist and we will characterize the two factors of the solution f (t) and φ(r). t)|2 = |φ(r)|2 .70) The task of ﬁnding solutions φ(r) which solve (3. At this point we state without proof that.. This can hold only if |f (t)| is time-independent. however.62) we insert it into (3.. It turns out that a solution for f (t) can be found for any E.62) exists. t)|2 = |f (t)|2 Ω d3 r |φ(r)|2 (3.e.27) holds and. (3. We will encounter many such problems in the subsequent Sections. (3. 3. We want to demonstrate this property now. .65) is constant. As pointed out above. . for functions required to obey boundary conditions (3..66) i. in general. One calls such states stationary states. We will demonstrate that solutions of the type (3. often only for a discrete set of values En .69) If these two equations can be solved simultaneously a solution of the type (3. t) vanishes on the surface of ∂Ω.71) . according to (3. At this point we will accept that solutions φ(r) of the type (3. the solutions o (3. h1 (r) = φ(r). In fact.31).60 Schr¨dinger Equation o and aﬀect only the spatial wave function φ(r). the eigenvalues of the operator H.67) (3. for the eigenvalue problems in the conﬁned function space. In order to further characterize the solution (3. We may note in passing that solutions of the type (3. solutions exist only for a set of discrete E ˆ values.64). i.62) is |ψ(r. We must postulate therefore ∂t f (t) ˆ H φ(r) = E f (t) = E φ(r) . We have then shown that ψ(r.e.62) have the particular property that the associated probability distributions are independent of time. i.69) exist. a common case is Ω = Ω∞ and the ‘natural boundary condition’ (3. t) = Ω Ω d3 r |ψ(r. (3. We denote the corresponding solution by φE (r). hence.68) ˆ where g1 (t) = i ∂t f (t). if f (t) = eiα .62) are special solutions of the timedependent Schr¨dinger equation.19). 2.62) which consist of two factors.63.69) is called an eigenvalue problem.45) that the total probability d3 r ρ(r. is time-independent. by no means all solutions are of this type. This yields an expression g1 (t) h1 (r) = g2 (t) h2 (r) (3. namely f (t) = f (0) exp − i Et . n = 1. .e. The identity (3. It follows then from (3. α ∈ R . (3. It follows from the observation that for the solution space considered (3. It is important to realize that the separable solutions (3. g2 (t) = f (t). and h2 (r) = H φ(r). One can conclude that the probability density for the state (3. one factor depending only on the time variable and the other only on the space variables are called separable in space and time. t) = f (0) exp − i E φE (r) where ˆ H φE (r) = E φE (r) .42) the ﬂux j(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.

The corresponding stationary solution ψ(r. = E. E (3.43) is j(r.73) from which follows E = E ∗ and.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 .19. one can interpret then k as the magnitude of the momentum of the particle. We will wave this assumption as we always need to do later whenever we deal with unbound particles.74) The classical free particle with constant energy E > 0 moves without bounds in the space Ω∞ . real.76) according to (3. o According to (3. We will show in Section 5 that E can be interprerted as the total energy of a stationary state.71) are.66) E must be real. The solution φE (r) corresponding to the eigenvalue problem posed by (3. The resulting total energy values E are positive. . − 2 2m φ(r) = E φ(r) . We start our proof using the property (3. Obviously. e. t) = |N |2 k . The ﬂux corresponding to (3.72) According to (3.74) is actually best labelled by an index k. As a result we cannot postulate in the present case that wave functions are localized and normalizable.28) for the special case that f and g in (3. of course.28) both represent the state φE (r). t) = exp − E t 2 φE (r) . m (3.3. (3. particles scattered of a potential. Stationary State of a Free Particle We consider now the stationary state of a free particle described by i ψ(r. ˆ d3 r φ∗ (r) H φE (r) = E Ω Ω ˆ d3 r φE (r) H φ∗ (r) .g. E (3. should be positive. hence.e. (3.76) has kinetic energy 2 k 2 /2m. in fact.4: Free Particle 61 is a solution of the time-dependent Schr¨dinger equation (3.74) using exp i k · r = ik exp i k · r .71) this yields E Ω d3 r φ∗ (r) φE (r) = E ∗ E Ω d3 r φE (r) φ∗ (r) .. E ∈ R.20). t) = exp − i 2 k2 2m t exp i k · r (3. We want to prove now that the eigenvalues E which arise in the eigenvalue problem (3. i. 3. a property which is to be expected since the energy is purely kinetic energy which. k ∈ R 2 k2 φk (r) = N exp i k · r .75) 2m One can ascertain this statement by inserting the expression for φk (r) into the eigenvalue problem posed in (3.

a] ⊂ R which vanish on the surface ∂Ω1 = {−a. t) = − d2 ψ(x. We can expect.82) satisfy the diﬀerential equation in (3. we have to consider only one boundary condition. hence.5 Particle in One-Dimensional Box As an example of a situation in which only bound states exist in a quantum system we consider the stationary states of a particle conﬁned to a one-dimensional interval [−a. 2m dx2 2 (± ) = } (3.82) One can readily verify that (3. t) = exp(−iEt/ )φE (x) where φE (x) is determined by 2 d2 − φE (x) = E φE (x) . let say the one at x = a. We will refer to this as a particle in a one-dimensional ‘box’. We assume.81.80). 2a n = 1.78) (3.83) The solutions (3. t) . and so-called odd solutions obeying φ(x) = −φ(−x) φE (x) = A sinkx.81. Hence. (3. a}. . n ∈ N. 3.80) need to be satisﬁed are φE (e. Solutions of the Schr¨dinger Equation in F1 o The time–dependent solutions satisfy i ∂t ψ(x. f continuous.. 3.80) 2m dx2 We note that the box is symmetric with respect to the origin.82) have the property that either both boundary conditions are satisﬁed or none.84) . i.62 Schr¨dinger Equation o 3.81) they are kn = nπ . a] ⊂ R → R.e. (3. so-called even solutions obeying φ(x) = φ(−x) φE (x) = A coskx . that the solutions obey this symmetry as well.o) (a) = 0 and φE (e. 5 . In case of the even solutions (3. therefore. 3. The boundary conditions which according to (3. . (3. f ∈ F1 = {f : [−a.o) (−a) = 0 (3.79) The stationary solutions have the form ψ(x. . a] ⊂ R assuming that the potential outside of this interval is inﬁnite. It turns out that this boundary condition can only be satisﬁed for a discrete set of k–values kn . two types of solutions. (o) 2 k2 (e) 2 k2 2m = E (3. φ(±a) = 0 .81) 2m = E. 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.

o) (a. (3. . 3 . .93) The normalization constants are then An = (3. (3. 5 . . By counting the number of their nodes one can determine the energy ordering of the wave functions..1). 6 . 6 . .86). n = 2.86) satisfy the boundary condition since for such kn nπa sin(kn a) = sin 2a = sin nπ 2 = 0.5: Particle in One-Dimensional Box since for such kn 63 nπa nπ = cos 2a 2 In case of the odd solutions (3.87) (Note that.82). are En = 2π2 8ma2 n2 . .92) and for the odd states +a |An |2 −a dx sin2 nπx = |An |2 a = 1 . The wave functions for the ﬁve lowest energies En are presented in Fig. 3. . 3.82) only the kn –values cos(kn a) = cos kn = nπ . (3. 3.94) . x)|2 = 1 n (3.81) and (3. (3. . 6 . 4. . n is assumed to be even.) The Energy Spectrum and Stationary State Wave Functions The energy values corresponding to the kn –values in (3. n = 1. (3.88) where the energies for odd (even) n–values correspond to the even (odd) solutions given in (3.81. 3. . 2a n = 1.91) holds. φ(e) (a.84. 4. .e. 4. . . 2. This condition implies for the even states +a |An |2 −a dx cos2 nπx = |An |2 a = 1 .89) nπx . It is desirable to normalize the wave functions such that +a −a dx |φ(e.90) 2a The wave functions represent stationary states of the particle in a one-dimensional box. respectively.86).3. a n = 2. x) = An cos n and φ(o) (a. 2a 1 . (3. (3. 2a n = 2. 2a n = 1.82). i. according to the dispersion relationships given in (3. . (3. = 0. 5 . according to (3. x) = An sin n nπx .85) (3. Notice that the number of nodes of the wave functions increase by one in going from one state to the state with the next higher energy En .

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

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. Similarly. In particular. it holds φn |φm Ω1 = δn m . 2a 2a (3. f. (3.102) holds according to (3.96) f |g Ω1 = −a dxf (x) g(x) . the integrals vanish. .95) 1 a cos nπx 2a sin nπx 2a for n = 1.98) the elements of B1 must be linearly independent. 2. In fact.99) Since in this case the integrand is a product of an even and of an odd function. .89. this integral vanishes. 6 . for ∞ f (x) = n=1 dn φn (a. n (3.5: Particle in One-Dimensional Box The Stationary States form a Complete Orthonormal Basis of F1 We want to demonstrate now that the set of solutions (3. Hence we need to consider only the ﬁrst two cases. . m both even. . the integral vanishes. the integral arises φn |φm 1 a Ω1 = 1 a +a nπx −a dx cos 2a cos mπx = 2a (3.e. 3. (3.101) +a −a dx cos (n−m)πx − cos (n+m)πx 2a 2a and.} where φn (a. Because of the property (3. n = 1. (i) n. and (iii) the mixed case. i. g ∈ F1 (3.90. Obviously. x) = together with the scalar product5 +a 65 (3. In case of n. m both odd. N ≥ 1.. a] are N.3.97) form an orthonormal basis set. m even φn |φm 1 a Ω1 = 1 a +a nπx −a dx sin 2a sin mπx = 2a (3. . . too. hence. for n = 2.94) B1 = {φn (a. 3. 4. 3. (ii) n.e.103) We will show in Section 5 that the property of a scalar product do indeed apply. the integrand is odd. one obtains for n.. . 3. m odd. x) (3. x) . i. . it holds: f |f Ω1 = 0 → f (x) ≡ 0. n = m.98) ∞ f |f 5 Ω1 = n=1 d2 . 5 . In case of n = m we have to consider three cases. The latter case leads to integrals φn |φm Ω1 = 1 a +a dx cos −a nπx mπx sin .98) The latter property is obviously true for n = m.

t0 ) in terms of eigenfunctions φn (a.. x0 ) ψ(x0 . 5 . . Accordingly. . n = 2. t0 ) (3.108) where the functions cn (t) are to be determined from the Schr¨dinger equation (3.105) must vanish. n = 0.105) 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.78) can be expressed as a linear combination of the elements of B1 deﬁned in (3. . t0 ) . x) .96). For this purpose we expand ∞ ψ(x. Evaluating the Propagator We can now use the expansion of any initial wave function ψ(x. a] yields then also an expansion for any element in F1 and B1 is a complete basis for F1 . x0 ) ψ(x0 . . x) cn (t) −a dx0 φn (a. . ( ) = ([ + ] − ) (3.e. 2. It follows that B1 n deﬁned in (3. .} such that ∞ ∞ nπx nπx ˜ = + an sin (3. 3. In the present case. t) at times t > t0 . any element of the function space F1 deﬁned in (3. 4.e. m = 1. t) = n=1 φn (a. t0 ) = +a ∞ i n=1 − En φn (a.95. . 2. The functions f are periodic with period 2a. t0 ) = dm . x) ∂t cn (t) −a dx0 φn (a.109) Insertion of (3.108) into the Schr¨dinger equation yields o +a ∞ n=1 φn (a.. n = 1. 1. (3. . x0 ) ψ(x0 .104) ˜ where [y]a = y mod2a. x) to obtain an expression for ψ(x. t0 ) = n=1 dn φn (a. they can be expanded in terms of a Fourier series.106) and generalizing to t ≥ t0 one can write ∞ +a ψ(x. n = 1. there exist real constants {an . such demonstration can be based on the theory of Fourier series.106) Using the orthonormality property (3. .79) requiring the o initial condition cn (t0 ) = 1 . We like to show ﬁnally that the basis (3. (3. x) cn (t) −a dx0 φn (a. 3.110) .} and {bn . in (3. Restricting the expansion (3. i.98) one obtains +a dx0 φm (a. Hence.66 Schr¨dinger Equation o f (x) ≡ 0 implies f |f Ω1 = 0 which in turn implies dn = 0 since d2 ≥ 0. the coeﬃcients an .105) to the interval [−a. Demonstration of completeness is a formidable task. This implies that only the trigonometric functions which are elements of B1 enter into the Fourier ˜ series. 3. however. . −a (3. . For this purpose we extend the deﬁnition of the elements of F1 to the whole real axis through ˜ f : R → R. (3.95) is an orthonormal basis. and bn . We have then shown that any f corresponding to elements of F1 can be expanded in terms of elements in B1 . .107) Inserting this into (3.95) is also complete. . . i. x0 ) ψ(x0 .

i ∂t cm (t) = − Em cm (t) .3. t|x0 .79) which lies in the proper function space (3.98). x0 ) . We have identiﬁed then with (3. . The last frame shows the wave front reaching again the right wall and the onset of new interference.108. t) for any initial condition ψ(x. The particle moves then to the left. t0 ) = 1 2πσ 2 1 4 exp − x2 0 + i k0 x0 2σ 2 . Example of a Non-Stationary State As an illustration of a non-stationary state we consider a particle in an initial state ψ(x0 . t0 ) = ∞ φn (a.112) determine now ψ(x. t0 ) ψ(x0 . One can recognize that the particle moves ﬁrst to the left and that near the right wall of the box interference eﬀects develop.115) This initial state corresponds to the particle being localized initially near x0 = 0 with a velocity v0 = 15 /m a in the direction of the positive x-axis. t0 ). In case that diﬀerent boundary conditions hold the propagator will be diﬀerent as well.2 presents the probability distribution of the particle at subsequent times.. t0 |x0 .112) Equations (3.54) as in the case of the free particle system which does not exhibit any bound states. k0 = . t0 ) is also referred to as a Greens function. (3. t|x0 . x) and integrating over [−a. t0 ) itself is a solution of the time-dependent Schr¨dinger o equation (3. x) exp n=1 i − En (t − t0 ) φn (a. but the collision with the left wall leads to new interference eﬀects.5). (3. according to (3.113). 3.114) This expression has the same form as postulated in the path integral formulation of Quantum Mechanics introduced above. The respective initial condition is φ(x. t0 ) (3. In the present system which is composed solely of bound states the propagator is given by a sum.e.5: Particle in One-Dimensional Box Multiplying both sides by φm (a. σ = a 15 . being reﬂected at the right wall. t) = −a dx0 φ(x. t0 ) = δ(x − x0 ) as can be readily veriﬁed using (3. cm (t0 ) = 1 .113) where φ(x. i. This solution can be written +a ψ(x. 4 a (3. 67 (3.114) the representation of the propagator for a particle in a box with inﬁnite walls. The interference pattern begins to ‘smear out’ ﬁrst. a] yields. in (2.111) The solutions of these equations which satisfy (3. Note that the propagator has been evaluated in terms of elements of a particular function space F1 . Often the propagator φ(x. Figure 3.109) are cm (t) = exp i − Em (t − t0 ) . t|x0 . It is of interest to note that φ(x.78). the elements of which satisfy the appropriate boundary conditions. rather than by an integral (3. t|x0 .

68 Schr¨dinger Equation o |Ψ|2 2 t = 0.48 m a h2 |Ψ|2 2 t = 0.92 m a h2 |Ψ|2 2 t = 2.2: Stroboscopic views of the probability distribution |ψ(x.68 m a h2 1 πσ 1 πσ -a a -a a |Ψ|2 2 t = 1.72 m a h2 1 πσ 1 πσ -a a -a a |Ψ|2 2 t = 0. 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 = 0.20 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 = 0.96 m a h2 |Ψ|2 2 t = 1.16 m a h2 1 πσ 1 πσ -a a -a a Figure 3.

122) . x3 )T ∈ Ω. x1 = ±a1 } ∪ {(x1 . This property allowed us to evaluate the propagator (3. )T ∈ ∂ } . a1 ] ⊗ [−a2 .3. ( .119) where φE (x1 . x3 –axes with the edges of the box yields spatial boundary conditions which are obeyed by the elements of the function space F3 = {f : Ω → R. i. 3.97) with respect to which the eigenfunctions are orthonormal. x2 = ±a2 } ∪ {(x1 . x3 ) . (3. The description employed a space of functions F1 deﬁned in (3.. 2a3 .6: Particle in Three-Dimensional Box Summary: Particle in One-Dimensional Box 69 We like to summarize our description of the particle in the one-dimensional box from a point of view which will be elaborated further in Section 5. x3 ) (3. ˆ H = − 2 (3. x2 .116) and obeys the partial diﬀerential equation ˆ H φE (x1 .117) 2m 2 2 2 ∂1 + ∂2 + ∂3 (3. x3 = ±a3 } .e. of the space F1 is that the elements of the basis set can be enumerated by integer numbers.120) ˆ Since the Hamiltonian H is a sum of operators O(xj ) each dependent only on a single variable. x3 ) = j=1 φ(j) (xj ) (3. j = 1. x2 . t) = exp − E t φE (x1 .116) where Ω is the interior of the box and ∂Ω its surface Ω ∂Ω = [−a1 . .6 Particle in Three-Dimensional Box We consider now a particle moving in a three-dimensional rectangular box with side lengths 2a1 . φ(j) (±aj ) = 0 . a2 ] ⊗ [−a3 . x2 .95). Placing the origin at the center and aligning the x1 .118) which are stationary states. x2 . t) . x3 . A complete basis of F1 is given by the inﬁnite set B1 (3. (3. t) = H ψ(x1 . . ˆ = O(x1 ) + O(x2 ) + O(x3 ). x2 .. We have deﬁned in the space F1 a scalar product (3. x2 . x2 . An important property of this basis and. 3 (3.114) in terms of which the solutions for all initial conditions can be expressed. one can express H 3 φ(x1 . The corresponding solutions have the form i ψ(x1 . x2 . 2a2 . x2 .78). ) = ∀( .e. i.121) where − 2 2m 2 ∂j φ(j) (xj ) = Ej φ(j) (xj ) . x3 ) is an element of the function space F3 deﬁned in (3. x3 )T ∈ Ω. 2. hence. x3 )T ∈ Ω. a3 ] ⊂ R = {(x1 . We seek then solutions of the time-dependent Schr¨dinger equation o ˆ i ∂t ψ(x. x3 ) = E φE (x1 . f continuous. x2 . x3 . can be counted. x2 .

i. If two or all three orthogonal sides of the box have the same length further symmetry operations leave the system unaltered. r0 ) . (3. This symmetry has been exploited in deriving the stationary states. . n 3 and E(n1 . 3.125) is a complete orthonormal basis of F3 and that the propagator for the three-dimensional box is φ(r.122) are given by (3.n3 ) (a1 . a3 . Comparing (3. 3. x1 .70 Schr¨dinger Equation o and E1 + E2 + E3 = E.124) The same considerations as in the one-dimensional case allow one to show that B3 = {φ(n1 . hence. a2 .120) can be written φ(n1 . n3 = 1.. x3 ) = 1. .n3 =1 φ(n1 . .n2 . x3 ) n1 . . . a1 = a2 = a3 = a then rotation around the x1 . x2 . x2 . the solutions of (3. Such degeneracies are always a signature of an underlying symmetry. t|r0 .n2 . n 2 . a3 are identical. namely rotation by π around the x1 . a2 . x1 ) φn2 (a2 . n1 . 2. n2 . r) exp i − E(n1 n2 n3 ) (t − t0 ) φ(n1 .n2 .122) with (3. x3 –axes.e. .96) and.80) shows that the solutions of (3. This additional symmetry is also reﬂected by degeneracies in the energy levels. a3 . n1 . a2 . . x2 . For example. n2 . (3.n2 . a2 . Actually.n3 ) = 2π2 = φn1 (a1 . 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. 2. t0 ) = n1 . in the present case .126) Symmetries The three-dimensional box conﬁning a particle allows three symmetry operations that leave the box unchanged. x2 ) φn3 (a3 .n2 . a3 . x1 .n3 ) (a1 . x2 . if all three lengths a1 . x3 ) . (3. 3. n3 = 1. .} ∞ (3.n3 ) (a1 . x3 –axes by π/2 also leaves the system unaltered.n2 . . a2 .123) 8m n2 n2 n2 1 2 3 + 2 + 2 a2 a2 a3 1 . a3 .n3 ) (a1 . 2.

e. this linear combination o is not necessarily orthogonal to other degeneraste states.2) (a. r) + β φ(2.127) ˜ ˆ ˜ obeys the stationary Schr¨dinger equation H φ(r) = E122 φ(r). The origin of this degeneracy is. However. in case of degenerate states one cannot necessarily expect that stationary states are orthogonal. a. a.2) (a. for (n1 . r).2) (a. a. a.6: Particle in Three-Dimensional Box 71 ‘accidental’ degeneracies also occur. r) (3. φ(1. n3 ) = (3. due to the hermitian character of ˆ H. n2 . Any linear combination of wave functions ˜ φ(r) = α φ(1.1. a.g. One particular aspect of the degeneracies illustrated in the Table above is worth mentioning. .2. We consider the degeneracy of the energy E122 which is due to the identity E122 = E212 = E221 . to construct n orthogonal stationary states6 . 5) as shown in the Table above. a.3. 3) (1. however. 3. 6 This is a result of linear algebra which the reader may ﬁnd in a respective textbook. for example. r) + γ φ(2. Hence. the identity 32 + 32 + 32 = 12 + 12 + 52 . 1. in case of an n–fold degeneracy it is always possible. However..2.2.1) (a. a. a.

72 Schr¨dinger Equation o .

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

2) can be expressed in terms of the two operators p = ˆ d . In the framework of the Schr¨dinger equation the stationary o states are solutions of (4. however. (4. in which the operators used below.1) of the form ψ(x. The Hamiltonian H can be expressed in terms of the operators acting on the space (4. t) where ˆ H φE (x) = E φE (x) .3) Due to the nature of the harmonic potential which does not allow a particle with ﬁnite energy to move to arbitrary large distances. The stationary states of the harmonic oscillator have been considered already in Chapter 2 where the corresponding wave functions (2. lim ( ) = } (4. The cardinal property exploited below.6) the ﬁrst being a diﬀerential operator and the second a multiplicative operator. is the commutation property 1 1 (4. This property states that p and x obey an ˆ ˆ algebra in which the two do not commute.4) Equation (4.3) can be solved for any E ∈ R. It is important to keep in mind this restriction of the space. all stationary states of the harmonic oscillator must be bound states and. The operators act ˆ on the space of functions N1 deﬁned in (4.74 Linear Harmonic Oscillator In the following we consider ﬁrst the stationary states of the linear harmonic oscillator and later consider the propagator which describes the time evolution of any initial state.3) we restrict the ˆ Hamiltonian H and the operators appearing in the Hamiltonian from the outset to the space of functions N1 = {f : R → R. the spectrum of ˆ H in (4. 4.3) would be continuous and the eigenfunctions φE (x) would not be normalizable. act.1 Creation and Annihilation Operators The Hamiltonian operator (4. t) = exp(−iEt/ )φE (x. In the following algebraic solution of (4. however.5) ˆ H = 1 1 2 p + m ω 2 x2 ˆ ˆ 2m 2 (4.8) i which holds for any position and momentum operator. i dx x = x ˆ (4. the natural boundary conditions apply x→±∞ lim φE (x) = 0 . We will point out explicitly where assumptions are made which built on this restriction. beside the representation (4.5) →±∞ where C∞ denotes the set of functions which together with all of their derivatives are continuous. the commutator has a simple form.7) of the Hamiltonian.7) which is why these operators are of interest to us.4) be satisﬁed. only for a discrete set of E values can the boundary conditions (4.5). If this restriction would not apply and all functions f : R → R would be admitted. x] = p ˆ . therefore. ∈ C∞ . In order [ˆ. (4.235) had been determined.

7) and (4.17) d dx We also express a+ and a− directly in terms of the coordinate x and the diﬀerential operator ˆ ˆ a+ = ˆ mω x − 2 d .10) holds for a ˆ ˆ [ˆ− . the action of [ˆ. x] .10) we deﬁne a+ = ˆ mω i x − √ ˆ p. in fact.9) From this follows (4.4.12) To prove this commutation property we determine using (4.15) together lead to the commutation property (4. 2mω dx (4. Before we continue we like to write (4.15) 1 ˆ 1 H + 1 .14) and (4. a+ ] = 1 . 1 ω 2 (4.13) (4. ˆ 2 2m ω (4.1: Creation and Annihilation Operators 75 to derive (4. ˆ 2 2m ω a− = ˆ mω i x + √ ˆ p. One obtains [ˆ. 4.8) we recall that all operators act on functions in N1 and.14) 1 i mω 2 x + ˆ p2 + ˆ [ˆ.14. p ˆ 2 2m ω 2 (4. Since a+ and a− are ˆ ˆ + and a− since operators and not scalars we cannot simply expect that the identy (4.16) (4.18) It is of interest to note that the operators a+ and a− are real diﬀerential operators. a+ ] = a− a+ − a+ a− does not necessarily vanish.7) assuming that the factors are easier to handle than the Hamiltonian in yielding spectrum and eigenstates.8) a ˆ ˆ ˆ ˆ ˆ implies [ˆ− .12). Our next step is an attempt to factorize the Hamiltonian (4. 1 ω 2 (4. x] f (x) = p ˆ d d x f (x) − x f (x) = f (x) = 1 f (x) . to factor H/ ω. In fact.15) in a form which will be useful below ˆ H = ˆ H = ω a− a+ − ˆ ˆ ω a+ a− + ˆ ˆ ω 1 1 2 ω 1 . ˆ ˆ . the commutator property (4.11) ˆ The reader may note that we have attempted.8) yield a− a+ = ˆ ˆ Similarly one can show a+ a− = ˆ ˆ 1 ˆ 1 H − 1 . hence. x] p ˆ on such functions must be considered. 2mω dx a− = ˆ mω x + ˆ 2 d .11) a− a+ = ˆ ˆ (4. 1 2 (4. a ˆ 1 (4.8). 1 i dx i dx i i (4. Being guided by the identity for scalar numbers (b − ic)(b + ic) = b2 − i(cb − bc) + c2 = b2 + c2 (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) .24) − m This property states that the operators in the function space N1 are the hermitian conjugate of each other.19).16.17.21) a− φE (x) is ˆ a stationary state of energy E − ω and a+ φE (x) is a stationary state of energy E + ω.21) ˆˆ H a+ φE (x) ω 1 ) a+ φE (x) 1 ˆ 2 ω = a+ ( ω a− a+ − ˆ ˆ ˆ 1 + ω 1 ) φE (x) 1 1 2 ˆ = a+ ( H + ω ) φE (x) ˆ = ( ω a+ a− + ˆ ˆ = (E + ω ) a+ φE (x) . 4. 4. In fact. one obtains using (4. 4.19) This property states that the operators a+ and a− are the adjoints of each other.20) ˆ In the following we will determine the spectrum of H and its eigenstates. The property ˆ ˆ follows directly from (??). This property of operators is investigated more systematically in Section 5.19) in the form1 f |ˆ+ g a Ω∞ = g|ˆ− f a Ω∞ .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) .16. ˆ (4.21) is available. it holds that for a stationary state φE (x) of energy E deﬁned through (4.22) Similarly one can show using again (4.12. The derivation will be based solely on the properties (4.17. (4. 4.16. 4. (4. ˆ (4. 4. 4. Using (??) we like to state (4. .17. 4. a+ and a− as Ladder Operators ˆ ˆ The operators a+ and a− allow one to generate all stationary states of a harmonic oscillator once ˆ ˆ one such state φE (x) ˆ H φE (x) = E φE (x) (4. ˆ The results (4. g ∈ N1 +∞ +∞ dx f (x) a+ g(x) = −∞ −∞ dx g(x) a− f (x) .23) Together. (4.22.76 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.

+∞[ one can write (4. one expects that E = 0 should be a lower bound.5). i. be found. Using (4. Let φE (x) and φE (x) be two normalized stationary states corresponding to two diﬀerent energies E.17) ˆ H φ0 (x) = ( ω a+ a− + ˆ ˆ 1 ω 1 ) φ0 (x) = 1 ω φ0 (x) . (4. however. Another name is creation (ˆ+ ) and annihilation (ˆ− ) operators since these operators a a create and annihilate vibrational quanta ω.26) Of course. in fact.29) d 1 d φ0 (y) = = lnφ0 (y) = − y . the solution φ0 (x) of (4. in fact. E = E . φ0 (x) should be an element of the function space N1 deﬁned in (4. (ii) the construction appears to yield energy eigenvalues without lower bounds when.25) φ0 (y) = 0 (4. (4.4.33) .25) can.28) 4.) ˆ ˆ 0 = φE |HφE Ω∞ − φE |HφE Ω∞ = ( E − E ) φE |φE Ω∞ .29) where y = mω x. on ˆ the other side such a state is itself an eigenstate of H as can be shown using (4.30) Assuming that φ0 (y) does not vanish anywhere in its domain ] − ∞.6. (4. m ∈ Z when m is decreased.25) needs to be normalizable in order to represent a bound state of the harmonic oscillator.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.11) one can rewrite (4.e. The property (??) has an important consequence for the stationary states φE (x). It turns out that both diﬃculties can be resolved together. (4.31) (4. For f (x) = φE (x) and g(x) = φE (x) in (??) follows (Note that according to (??) the eigenvalue E is real.27) Since E = E one can conclude φE |φE Ω∞ = 0. a state φ0 (x) which obeys the property a− φ0 (x) = 0 ˆ (4. E . 4.. For obvious reasons one refers to a+ and a− as ladder ˆ ˆ operators.25) on one side would lead to termination of the sequence E0 + m. In fact. two diﬃculties: (i) one needs to know at least one stationary state to start the construction. 2 2 (4. There arise. φ0 (y) dy dy the solution of which is 1 lnφ0 (y) = − y 2 + c0 2 φ0 (y) = c0 exp 1 − y2 2 .2 Ground State of the Harmonic Oscillator d + y dy A suitable solution of (4.

(4. a state conﬁned to the minimum corresponding to the classical state of lowest energy.26) the energy value associated with this state is 1 ω. . 1.36) n times. (4. . .35) The appropriate ground state is φ0 (x) = mω π 1 4 exp − mωx2 2 .37) above the ground state energy. i.. which for a narrowly localized state yields a large positive value.36) Since the ﬁrst order diﬀerential equation (4. n = 0. .24) can then be written as follows En = ω(n + 1 ).2) there are two competing contributions to the energy.34) dx |φ0 (x)|2 = |c0 |2 −∞ −∞ dx exp − mωx2 2 = |c0 |2 π . .. The reason is that in the Hamiltonian (4. 2 n = 0. mω (4. the potential energy contribution which for a state δ(x). .e. 2. 2 This state of lowest energy is called the ground state of the oscillator. i. . The conventional normalization condition φ0 |φ0 reads +∞ +∞ Ω∞ = 1 (4. i. we can now construct the stationary states corresponding to energies (4.39) We notice that the ground state wave function φ0 (x) as well as the operators (ˆ+ )n are real. m ∈ Z which properly terminate at some minimum value E + m0 ω ≥ 0.25) admits only one solution there is only one set of states with energy E + m ω. . and the kinetic energy contribution. would yield a vanishing contribution. 1. 2.36) assumes a functional form such that both terms together assume a minimum value. (4. therefore. The set of allowed energies of the oscillator according to (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. the so-called excited states. 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. . 2 n = 0. we construct the states for n = 1.. 2. We will consider this point more systematically in Section 5. .38) These states correspond to the energy eigenvalues (4. . We a can. 4. . (4.e. We recall that according to (4. For this purpose we apply the operator a+ to the ˆ ground state (4.e. The ground state (4. 1.3 Excited States of the Harmonic Oscillator Having obtained a suitable stationary state with lowest energy. choose the normalization constants cn and the functions φn (x) real as well.39).. . it holds ˆ H φn (x) = ω(n + 1 ) φn (x) .78 Linear Harmonic Oscillator This solution is obviously normalizable.

they obey φn |φn = δn n .47) for some suitable constants βn . .50) φn (x) . √ a− φn (x) = n φn−1 (x) .41) Employing the adjoint property (4. 2.51) .43) (4.42) ˆ Using (4. n. according to (4.40) = 1. i.39) yield √ From this follows βn = 1/ n and.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) .24) holds in analogy to (4. .46) According to (4. (4. βn must be real as well. 4.47). . (4. .15 . We consider then the situation that we have obtained a properly normalized state φn (x). Ω∞ 2 = βn a− φn |ˆ− φn ˆ a Ω∞ 2 = βn φn |ˆ+ a− φn a ˆ Ω∞ .40) φn−1 (x) = βn a− φn (x) ˆ (4.48) (4. ˆ (4.18)] and since the ˆ φn (x) are real functions. 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. √ a+ φn (x) = n + 1 φn+1 (x) . 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 .40).40–4.49) (4. To determine βn we note using (4. 1. according to (4.3: Excited States 79 We need to determine now the normalization constants cn . (4. 2.28) and the construction (4.. (4.4. Since a− is a real diﬀerential operator [see (4. . . For n = 0 holds c0 = 1.45) We like to note that these functions according to (4.20) 1 = φn−1 |φn−1 Equations (4.20) yields 2 αn φn |ˆ− a+ φn a ˆ Ω∞ = 1.14) 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. . 1. n = 0. To determine these constants we adopt a recursion scheme.45) form an orthonormal set. (4. n = 0.e. (4.

56) can be expressed in terms of Hermite polynomials Hn (y) introduced in Sect.28) of the ground state and the deﬁnition (4. n–independent factor. simplifying the calculation.. (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 . (4.52).35) by a constant. for example.e.30) and employing the normalization +∞ −∞ dy φ2 (y) = 1 n (4. (4.7 and given.53) We will later re-introduce this factor to account for the proper normalization (4.200). .7.57) 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) . by the Rodrigues formula (2. i.55) 1 y2 2 (4.57) which includes the factor 1/ n! according to Eqs. We will demonstrate below the identity Hn (y) = e y2 2 y − d dy n e− y2 2 (4. however. 2. by introducing the variable y deﬁned in (4. (4. For this purpose we start from expression (4. 2.56) Relationship to Hermite Polynomials We want to demonstrate now that the expression (4. It should be noted that √ the normalization (4.46) yields a set of normalized states.58) This result agrees with the expression (2.233) derived in Sect. by dx = dy mω 1 4 .39).45). In terms of y the ground state wave function is φ0 (y) = π − 4 e− and a+ is ˆ 1 a+ = √ ˆ 2 y − d dy .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. (4.40–4.52) This normalization of the wave functions diﬀers from that postulated in (4. namely the square root of the Jacobian dx/dy.35) rather than (4.

hence.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 . (3. and showing then that the property also holds for n + 1 in case it holds for n.106–3. dy dy (4. We prove (4.59) holds for n = 0. t0 ).114)].61) and.200). For the sake of notational simplicity we employ initially the coordinate y as deﬁned in (4.e. 1. (4.60) and. Our derivation will follow closely the procedure adopted for the case of a ‘particle in a box’ [see Eqs. t0 ) = n=0 dn φn (y0 ) . 4.57) of the hermite polynomials and the deﬁnition given by (2.62) Denoting f (y) = (y − d/dy)exp(−y 2 /2) and employing y2 y2 d d − y2 e 2 f (y) = −y e− 2 f (y) + e− 2 f (y) dy dy (4.4: Propagator 81 To verify the realtionship between the deﬁnition (4.30) and return later to the coordinate x.2) for an arbitrary initial wave function ψ(x0 . (4.39) ∞ ψ(y0 .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 . g(y) = (−1)n+1 e y2 2 dn+1 −y2 e = dy n+1 y − d dy n+1 e− y2 2 .63) one obtains g(y) = y − d dy f (y) (4.61) One can factor g(y) and employ (4. therefore..59) dn −y2 e dy n (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. Starting point of our derivation is the assumption that the initial condition can be expanded in terms of the eigenstates φn (y) (4.4. 4. dy n (4.59) by induction noting ﬁrst that (4.65) . i. (4.59) hold.64) which implies that (4. the Rodrigues formula (2.1.

h. 2n n! π 2 2 (4. Inc. Englewood Cliﬀs. 1 t = e−iω(t1 −t0 ) (4.46) the expansion coeﬃcients dn are +∞ dn = −∞ dy0 ψ(y0 .39) holds cn (t1 ) = exp −i ω ( n + 1 ) ( t1 − t0 ) 2 . t0 ) ψ(y0 . this is an excellent textbook from which we have borrowed heavily in this Section. y0 . 1965) Sect. We want to demonstrate now that this propagator is identical to the propagator (2.147) for the harmonic iscillator determined in Sect. (4.225) which allows one to derive a generating function for products of Hermite polynomials which can be applied to the r.J. Employing orthogonality condition (4.65) to times t ≥ t0 through insertion of time-dependent coeﬃcients cn (t1 ) ∞ ψ(y.70) and (2. .N.72) × exp( −u − v + 2iyu + 2iy0 v ) .70).147) we employ the technique of generating functions as in Sect. N. 2.69) where ∞ φ(y.68) Altogether one can express then the solution +∞ ψ(y.82 Linear Harmonic Oscillator Such expansion is possible for any f (y0 ) = ψ(y0 . y0 . t) = π −3/2 exp 2 2 y 2 y0 + 2 2 2 +∞ +∞ ∞ du −∞ −∞ dv n=0 1 ( −2tuv )n × n! (4.7. t) (4.67) for which according to (3.108–3.. (4.5).15. sec:harm. t1 |y0 . 68.s. t1 ) = n=0 dn φn (y) cn (t) (4. t0 ) which is an element of N1 deﬁned in (4. 2 A demonstration of this property can be found in Special Functions and their Applications by N. For this purpose we start from the integral representation (2. of (4. pp.71) Applying (??) to express Hn (y) and Hn (y0 ) yields exp(−2tuv) w(y. t1 ) = −∞ dy0 φ(y. t0 ) = n=0 φn (y) φn (y0 ) tn+ 2 .70) is the propagator of the linear harmonic oscillator. 4. We consider for this purpose the following expression for |t| < 1 ∞ w(y.112) and (4.Lebedev (Prentice Hall. In order to prove the equivalence of (4.. a supposition which is not proven here2 . t|y0 .66) One can extend expansion (4. t) = n=0 Hn (y)e−y /2 Hn (y0 )e−y0 /2 n √ t . t0 ) φn (y0 ) .

4: Propagator Carrying out the sum on the r.147) derived by means of the path integral description. This requires that we employ (4. t0 ) (4. t0 ) exp i 2 1 2 (y + y0 ) G(t1 .s.58) of the harmonic oscillator ∞ φn (y) φn (y0 ) tn = n=0 2 1 t 2 1 + t exp − (y 2 + y0 ) + 2 y y0 2 2) 2 1−t 1 − t2 π(1 − t 1 . (4.e.. by a factor 4 mω/ for both φn (y) and φn (y0 ).74) a second time yields ﬁnally together with the deﬁnition (4.73) The incomplete Gaussian integral +∞ −∞ dx e−a 2 x2 √ − 2bx = π b2 /a2 e .e. y0 .30) to replace y and y0 and that we multiply the propagator by mω/ . t0 ) = and 1 2 i π F (t1 . t0 ) − i y y0 2 F (t1 . √ πexp( t2 v 2 − 2iytv − y 2 ) (4.79) 1 + e−2iω(t1 −t0 ) cosω(t1 − t0 ) = . t) = π −1 exp − 2 y 2 y0 + 2 2 +∞ dv exp(−(1 − t2 ) v 2 − 2i (yt − y0 ) v) . t) = π −3/2 exp +∞ 83 y2 y2 + 0 2 2 +∞ dv exp( −v 2 + 2iy0 v ) × −∞ × −∞ du exp( −u2 − 2u(tv − iy) ) .77) The sum in (4. it holds φ(y. 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 = . t0 ) = exp mω 2iπ sinω(t1 − t0 ) 1 2 × . t1 |y0 . y0 .h.70) is.4. a Re a2 > 0 (4.81) imω 2 sinω(t1 − t0 ) (x2 + x2 ) cosω(t1 − t0 ) − 2 x x0 0 This result agrees with the propagator (2.74) applied once results in w(y. i.76) One can express this in terms of the stationary states (4. (4.77). The resulting propagator is G(t1 . indeed. (4. (4.78) 1 − e−2iω(t1 −t0 ) = sinω(t1 − t0 ) 2 i e−iω(t1 −t0 ) (4. y0 . t0 ) = i φ(x.80) −2iω(t1 −t0 ) sinω(t1 − t0 ) 1−e We can ﬁnally express the propagator (4. i. −∞ (4. one obtains w(y. .75) Applying (4. t0 ) = where F (t1 .78) in terms of the coordinates x and x0 .. identical to the generating function (4.71) of w(y. t|x0 .

the use of generating functions. a (4. 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. d/dy and a+ . Here f (ν) (y0 ) denotes the ν-th derivative of f (y) taken at y = y0 . hence.g.86) f (ˆ+ ) = a a ˆ ν! ν=0 . 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.5 Working with Ladder Operators In the last section we have demonstrated the use of diﬀerential equation techniques. the approaches a ˆ a ˆ using y.85) is convergent everywhere in R. ˆ ˆ Calculus of Creation and Annihilation Operators We summarize ﬁrst the key properties of the operators a+ and a− ˆ ˆ [ˆ− . phonons. In this case we deﬁne ∞ f (ν) (0) + ν (4. e.84) = 1 1 = 0 √ = n + 1 φn+1 (y) √ = n φn−1 (y) = f |ˆ+ g a Ω∞ . 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. one can express y = 2(ˆ− + a+ ) and d/dy = 2(ˆ− − a+ ) and. 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 − .82) a+ = √ ˆ dy dy 2 2 √ √ Obviously..83) We will encounter below functions of a+ and a− .e. and of the modes of the quantized electromagnetic ﬁeld. each corresponding to a single harmonic oscillator of the type studied presently by us.. i.84 Linear Harmonic Oscillator 4. a− = √ ˆ y + . f (ˆ+ ). a− must be equivalent. (4. both quantum systems are endowed with a large number of modes. (4. a+ ] a ˆ a− φ0 (y) ˆ a+ φn (y) ˆ a− φn (y) ˆ a− f |g Ω∞ ˆ We note that these properties imply √ φn (y) = (ˆ+ )n φ0 (y)/ n! . 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. We want to introduce now techniques based on the ladder operators a+ and ˆ a− .

88) we need to show actually [ˆ− . a− f (ˆ+ ) φ0 (y) = f (1) (ˆ+ ) φ0 (y) ˆ a a which follows from a− φo (y) = 0. a An important operator function is the exponential function. f (ˆ+ )] = f (1) (ˆ+ ) . ˆ a We proceed by induction noticing ﬁrst that (4. f1 (ˆ+ )] = [ˆ− . a a a 85 (4.5: Working with Creation and Annihilation Operators and similarly for f (ˆ− ). (ˆ+ )n a+ ] = (ˆ+ )n [ˆ− . fn+1 (ˆ+ )] a a = [ˆ− . Hence. We note ˆ a− f (ˆ+ ) = [ˆ− .90) holds for f (ˆ+ ). The ﬁrst case is trivial.94) This identity follows from (4. For fn+1 follows then [ˆ− .88) is (ˆ− )n f (ˆ+ ) φo (y) = f (n) (ˆ+ ) φ0 (y) . ˆ a a a ˆ In particular. a+ ] = 1 = f1 (ˆ+ ) .88) (4. a+ ] + [ˆ− . The convergence of the Taylor expansion ascertains then that (4. (ˆ+ )n ] a+ a a ˆ a a ˆ a a ˆ = (ˆ+ )n 1 + n (ˆ+ )n−1 a+ = (n + 1) (ˆ+ )n a 1 a ˆ a Since any function f (y) in the proper function space N1 can be expanded ∞ ∞ (1) (4.90) holds for f0 and for f1 . Note that ˆ a an immediate consequence of (4.87) (4. f (ˆ+ )] + f (ˆ+ ) a− ˆ a a a a ˆ which implies that in order to prove (4. An example is the so-called shift operator exp(uˆ− ). a a a ˆ 1 a Let us assume that (4. i.96) . 4.95) (4.87.90) To prove (4.83) and can be proven for all powers fn by induction and then inferred for all proper functions f (ˆ± ). As long as the operators act on φ0 (y) one can state then that a− behaves like a diﬀerential operator with respect to functions f (ˆ+ ).90) holds for fn . the second case follows from [ˆ− . It holds a a euˆ f (ˆ+ ) φ0 (y) = f (ˆ+ + u) φ0 (y) . a a a We like to state the following property of the functions f (ˆ± ) a f (ˆ− ) φ|ψ = φ|f (ˆ+ ) ψ .90) we show that (4. a a (4..91) (4. we will only assume operator functions acting on φ0 (y). property (4.4. The following important property holds a a− f (ˆ+ ) = f (1) (ˆ+ ) + f (ˆ+ ) a− . a a − (4.90) holds for any function fn (ˆ+ ) = (ˆ+ )n which is a power of a a a+ .89) (4.88) is a fundamental one and will be used in the following.e.92) f (y) = n=0 dn φn (y) = n=0 (ˆ+ )n a dn √ φ0 (y) n! (4.93) one can reduce all operators acting on some proper state function by operators acting on the state φ0 (y).

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.104) (4. + ∗ + (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) yields a eαˆ + ∗ a− ˆ a a eαˆ φ0 (y) = e−αα eαˆ φ0 (y) .86 To prove this property we expand exp(uˆ− ) a ∞ ν=0 Linear Harmonic Oscillator uν − ν (ˆ ) f (ˆ+ ) φ0 (y) = a a ν! ∞ ν=0 uν (ν) + f (ˆ ) φ0 (y) = f (ˆ+ + u) φ0 (y) .107) − α∗ a− ˆ φ0 (y) = e− αα∗ 2 a eαˆ φ0 (y) .90) and zz ∗ = 1 we can write this d du e uzˆ+ a e uz ∗ a− ˆ a − z ∗ a− .98) for u = −α∗ and v = α yields e−α This together with (4. hence.101) 1. a a ν! (4.106) Applying (4. euzˆ ˆ euz ∗ a− ˆ (4. We assume in the following a ˆ derivation. One can conclude then using the deﬁnition of C(u) and deﬁning ˆ ˆ is D(u) = exp(uzˆ a ˆ v = uz 1 ∗ ˆ− ∗ ∗ ˆ− a+ a+ evˆ + v a φ0 (y) = e 2 vv evˆ ev a φ0 (y) . without explicitly stating this. 1 + + uz ∗ a− ).102) Similarly. that the operators considered act on φ0 (y). z ∈ C. one obtains a evˆ + − v ∗ a− ˆ a φ0 (y) = e− 2 vv evˆ e− v 1 ∗ + ∗ a− ˆ φ0 (y) (4. ∗ = exp(uzˆ a ˆ and determine its derivative d du ˆ C(u) a = zˆ+ euzˆ euz a + ∗ a− ˆ a + euzˆ z ∗ a− euz ˆ + ∗ a− ˆ = + zˆ+ + z ∗ a− a ˆ Using (4.101) ˆ ˆ ˆ C(u) obviously obeys C(0) = 1 This results in D(0) = 1 and.100) ˆ ˆ To solve this diﬀerential equation we deﬁne C(u) = D(u) exp(−u2 /2) which leads to d ˆ D(u) = du zˆ+ + z ∗ a− a ˆ ˆ D(u) (4. + (4. + (4. the solution of (4.105) a eαˆ + − α∗ a− ˆ φ0 (y) = e αα∗ 2 e−α ∗ a− ˆ a eαˆ φ0 (y) .97) An example in which (4.98) The related operators exp[vˆ+ ± v ∗ a− ] play also an important role. (4.99) ˆ C(u) = zˆ a zˆ+ + z ∗ a− a ˆ + +z a ˆ ∗ − a euzˆ euz ˆ − u C(u) .103) (4. + ∗ a− ˆ a − u euzˆ euz + ∗ a− ˆ = (4. + (4.108) .96) is applied is a a a euˆ evˆ φ0 (y) = ev (ˆ − + + + u) a φ0 (y) = euv evˆ φ0 (y) . Below we will use the operator identity which follows for the choice of v = α in (4.

Similarly.h.84) yields ∞ ν.58) one obtains π− 4 1 ∞ ν=0 (−1)ν u2ν = 2ν ν! ∞ ν=0 uν √ φν (0) = ν! ∞ ν=0 uν Hν (0) √ 2ν π ν! (4. is the equivalent of the generating function (??).84.112) Expanding both sides of this equation and using (4. We want to derive now the values Hn (0).109) e = π4 n! n=0 √ Using (4.111) This expression.s.113) Comparision of all terms on the l.116) = δµ ν .98) and again (4. the expected from which follows by comparision of each term on both sides φµ |φν orthonormality property.95).h.110) √ y2 u2 + 2uy − 2 2 a = euˆ φ0 (y) .114) where we have employed (4.95) one obtains a a a a euˆ φ0 |evˆ φ0 = φ0 |euv evˆ φ0 = euv evˆ φ0 |φ0 = euv − + + + − (4.h. (??). .84) and deﬁning 2t = u one can write then π − 4 exp − or π − 4 exp − 1 1 √ y2 u2 + 2uy − 2 2 ∞ = n=0 un + n (ˆ ) φ0 (y) a n! + (4. Expanding r.5: Working with Creation and Annihilation Operators Generating Function in Terms of a+ ˆ 87 We want to demonstrate now that generating functions are as useful a tool in the calculus of the ladder operators as they are in the calculus of diﬀerential operators. We will derive the equivalent of a generating function and use it to rederive the values Hn (0) and the orthonormality properties of φn (y). in the ladder operator calculus. of (4. we can reproduce by means of the generating function (4.s. and on the r. Using (4.. provide the same values for Hn (0) as provided in Eq.111) yields π − 4 exp − 1 u2 2 a = euˆ φ0 (0) .58) the Hermite polynomials Hn (y) by the eigenstates φn (y) √ ∞ 1 ( 2t)n y2 /2 2yt − t2 √ e φn (y) (4.h.s.115) a where the latter step follows after expansion of evˆ and using a− φ0 (y) = 0. For this purpose we consider a a a a euˆ φ0 |evˆ φ0 = φ0 |euˆ evˆ φ0 + + − + (4. 4. Setting y = 0 in (4.s.4.111) the orthonormality properties of the wave functions φn (y). (4.µ=0 uµ v ν √ φµ |φν µ!ν! ∞ = µ=0 uµ v µ µ! (4. + (4. We start from the generating function (??) and replace according to (4.115) and using (4.e. i. and ˆ l.

4.7) is given by (4. i. ˆ and the Hamiltonian (4.124) ).119) 2m 2 dp2 Accordingly.. (−i)n ˜ φn (p) = √ 2n n! m˜ ω π 1 4 exp − m˜ p2 ω 2 Hn ( m˜ ω p) . the momentum and position operators are p = p. i.122). . 1. the momentum and position operators are as stated in (4. n = 0.123) (4. As to be expected. The solutions can be stated readily exploiting the earlier results. (4. En = ω (n + 1 ) 2 d2 1 + m˜ 2 p2 ω 2 2m dp 2 2 ˜ ˜˜ φE (p) = E φE (p) (4. For this purpose we employ the Schr¨dinger equation in the momentum o representation.37). (4. ˜ (4.7) is 1 2 m 2 ω 2 d2 p − . . 2.235).122) For a solution of the Schr¨dinger equation in the momentum representation. is given by the ˜ function ψ(p). dp (4.e.3). is given by a function ψ(x).118) d2 1 2 + p dp2 2m ˜ ˜ φE (p) = E φE (p) . In the position representation the wave function is a function of x. are ˜ En = or.120) Multiplying this equation by 1/m2 ω 2 yields − where ˜ E = E / m2 ω 2 .2.235). The eigenvalues. The momentum representation eigenfunctions are.e.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. (4. using (2. i. . .e. the probability density is P (x) = |φn (x)|2 . according to (4. .6) and the Hamiltonian (4.88 Linear Harmonic Oscillator 4. 1.125) . using (4. (4. In the momentum representation the wave function is a function of p. the time-independent Schr¨dinger equation for the oscillator can be written o ˆ H = − mω 2 2 2 x = i ˆ d .121) ω (n + ˜ 1 ) 2 .2). For example. of (4. ω = 1 / m2 ω .. (4. . 2..121). for an oscillator in a stationary state φn (x) as given by (2. we o note that the posed eigenvalue problem is formally identical to the Schr¨dinger equation in the o position representation as given by (4. n = 0.6 Momentum Representation for the Harmonic Oscillator The description of the harmonic oscillator provided so far allows one to determine the probability density P (x) of ﬁnding an oscillator at position x. the eigenvalues are identical to those determined in the position representation.

instead of (4.6: Momentum Representation or with (4. This implies the property of Hermite polynomials (−i)n e− k2 2 +∞ Hn (k) = −∞ dy exp(−iky) e− y2 2 Hn (y) (4. (4.126) in terms of dimensionless units following the procedure adopted for the position representation where we employed the substitution (4.f.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) to obtain (4. hence.128) We may also express the eigenstates (4. . Deﬁning k = m˜ / p or ω 1 k = p (4.6.30) and (2..134) Exercise 4.134). 1 ˜ φn (k) = √ 2π +∞ (4. α ∈ R. the correctness of the phase factors (−i)n . ˜ We can now state the probability density Pn (p) for an oscillator with energy En to assume momentum p. deﬁned only up to an arbitrary phase factor eiα .131) in the present case employing dimensionless units.129) follows ky = px/ and.125.126) Normalized eigenfunctions are. of course. through direct integration. i. the Jacobian.129) m ω one obtains. From (4.e. 2n n! π (4.e. m ω (4.30) and (4.127) exp − p2 m ω 2 Hn ( 1 p) . this allowed us to introduce in (4.132) dy exp(−iky) φn (y) −∞ (4..126).122) (−i)n ˜ φn (p) = √ 2n n! 1 πm ω 1 4 89 exp − p2 2m ω Hn ( 1 p) .126) a phase factor (−i)n . i.133) where we note that a change of the normalization of φn (y) [c.1: Demonstrate the validity of (4. 4.130) dx exp(−ipx/ ) φn (x) −∞ (4. m ω (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.58). Proceed as follows.

— III. 1989) which discusses quasi-classical states of the free electromagnetic ﬁeld in Sect. 4.Grynberg (John Wiley & Sons.134). z) = √ 2π where f (k. for example. n n=0 d(α) n 2 = 1 (4.138) We will show that a harmonic oscillator prepared in such a state will remain forever in such a state. The motion of the probability distribution of such state is presented in Fig.C.90 (a) Prove ﬁrst the property 1 f (k. y) −∞ (4. and G. 2. z) = exp −2ikz + z 2 − k 2 /2 . a pendulum swinging at small amplitudes.Chen-Tannoudji. ˆ α ∈ R. For this purpose we assume such state exists and expand ∞ ∞ φ(α) (y) = n=0 d(α) φn (y) .135) g(k..1 showing in its top row the attributes of such state just discussed. Inserting this expansion into √ (4.138). pp.136) (b) Employ the generating function (2.140) An excellent textbook is Photons and Atoms: Introduction to Quantum Electrodynamics by C. z) = exp 2yz − z 2 − y 2 /2 (4. J.139) where we have added the condition that the states be normalized. 4.Dupont-Roc. New York. except that α changes periodically in time. The quasi-classical states can be obtained by generalizing the construction of the ground state of the oscillator.196) of Hermite polynomials and equate coeﬃcients of equal powers of z to prove (4. Inc. +∞ Linear Harmonic Oscillator dy exp(−iky) g(z. We ﬁrst construct the solution of (4. We want to construct and characterize such states.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. (4. or Glauber states of the harmonic oscillator and they play.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. 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 . coherent states. namely through the eigenvalue problem for normalized states a− φ(α) (y) = α φ(α) (y) . The question arises if similar states exist also for the quantum oscillator.7 Quasi-Classical States of the Harmonic Oscillator A classical harmonic oscillator. Such states are referred to as quasi-classical states. 192 3 .194. carries out a periodic motion described by y(t) = Re y0 eiωt (4.4. φ(α) | φ(α) = . The answer is ‘yes’. for example. 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.

hence. One can also write (4.149) with (4.138). n! (4.142) The solution is αn d(α) = √ c n n! ∞ ∞ (4.143) for a constant c which is determined through the normalization condition [see (4.4. [π]− 4 e− one can write φ(α) (y) = [π]− 4 exp 1 1 y2 2 + √ 2 α y − α2 = exp − α2 2 ∞ n=0 αn √ φn (y) .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) . However.7: Quasi-Classical States or ∞ n=0 91 √ n + 1 dn+1 − α d(α) n (α) φn (y) = 0 . (4.84) φ(α) (y) = exp − α2 2 ∞ n=0 αα∗ (αˆ+ ) a a+ φ0 (y) = e− 2 eαˆ φ0 (y) . the quasi-classical states are φ (α) (y) = exp α2 − 2 ∞ n=0 αn √ φn (y) .149) − α∗ a− ˆ (4.146) √ 1 − (y − 2 α)2 2 (4.e. n! (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 (4.147) √ which identiﬁes this state as a ground state of the oscillator displaced by 2α. .145) Using the generating function in the form (4.. n n+1 (4.147) allows one to interprete T (α) as an operator √ which shifts the ground state by 2α. we will see below that α ∈ C are also admissible and will provide a corresponding interpretation.145) using (4.144) Normalization requires c = exp(−α2 /2) and. i.148) The identity (4.139)] 1 = n=0 d(α) n 2 = |c|2 n=0 α2 2 = |c|2 eα .109). Comparing (4. n! (4.150) for α presently real. n! n (4. This interpretation justiﬁes the choice of α ∈ R in (4.

4. Following the procedure adopted in determining the propagator of the harmonic oscillator [see (4.109) serves again to simplify this sum √ y2 1 1 α2 (u2 + 1) φ(α) (y.68)] one can write the time-dependent solution with (4.s. t1 ) = exp or φ (α) − α2 2 ∞ exp n=0 −i ω ( n + ∞ n=0 1 ) ( t1 − t0 ) 2 αn √ φn (y) n! (4.h.157) (y. .152) T + (α) f |T + (α) g = f |T (α) T + (α) g = f |g .155) (4.145) as the initial state φ(α) (y.145) shows that α in (4.153) This is obviously the inverse of T + (α) as deﬁned in (4.95) to linear combinations αˆ+ + βˆ− and reads then a a f (αˆ+ + βˆ− ) φ|ψ = φ|f (α∗ a− + β ∗ a+ ) ψ .154) . is the adjoint of the operator function on the l. n = 0. .156) (4. the shift operator leaves the ground state normalized.145) is replaced by a complex number αu which at times t1 = t0 + 2πn/ω. t1 ) = π − 4 u 2 exp − + 2yαu − 2 2 E (4. Relationship (4. i..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. t1 ) = u exp 1 2 α2 − 2 (α u)n √ φn (y) .158) Comparision of this expression with (4. a a ˆ ˆ (4. α ∈ C (4. one can write in case of T + (α) T + (α) + = T (α) = eα ∗ a− ˆ − αˆ+ a a = e − (αˆ + − α∗ a− ) ˆ .67. becomes real.s. n! u = e−iω(t1 − t0 ) .e.151) which follows from a generalization of (4. 1.h.150) for complex α obeys the unitary property T + (α) T (α) = 1 1 (4.92 Linear Harmonic Oscillator The shift operator as deﬁned in (4.152) Since the operator function on the r. . it holds T + (α) φ0 |T + (α) φ0 = f |T (α) T + (α) g = φ0 |φ0 .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 (4.160) . (4. (4. In particular.

162) 2 with a periodic change. this characterization corresponds closely to that of the possible initial states of a classical oscillator.147). (4. The time-dependent wave function (4. Arbitrary Gaussian Wave Packet Moving in a Harmonic Potential We want to demonstrate now that any initial state described by a Gaussian shows a time dependence very similar to that of the Glauber states in that such state remains Gaussian. obviously. t1 ) = π − 4 exp − [y − y0 (t1 )]2 + iy k0 (t1 ) − iφ0 (t1 ) − ω(t1 − t0 ) 2 2 . in general.158) allows one to write 1 φ(α) (y. Our interpretation 2 explains also the meaning of a complex α in (4. The diagram illustrates that the wave function retains its Gaussian shape with constant width for all times. t1 ) = u 2 T + (α u) φ0 (y) (4.161) are time-dependent √ y0 (t1 ) = 2 α cosω(t1 − t0 ) √ k0 (t1 ) = − 2 α sinω(t1 − t0 ) 1 φ0 (t1 ) = − α2 sin2ω(t1 − t0 ) (4. except that it is not pointlike. One can express (4.165) .150) with (4.161) and is found to describe a displaced Gaussian.1 (top row) we present the probability distribution of the Glauber state for various instances in time.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.138): a real α corresponds to a displacement such that initially the oscillator is at rest. moving solely its center of mass in an oscillator fashion around the minimum of the harmonic potential. 4. This shows clearly that the Glauber state is a close analogue to the classical oscillator. It is of interest to note that the displacement y0 of the Gaussian as well as the momentum k0 and phase φ0 associated with (4. (4. experiences a change of its width (relative to .7: Quasi-Classical States the exponent E can be written E = − √ √ y2 + 2 αy Reu − α2 (Re u)2 + i 2 αy Imu − i α2 Reu Imu 2 √ √ 2 1 = − y − 2αReu + i 2yα Imu − i α2 Reu Imu 2 phase term momentum displaced Gaussian 93 (4. 4.163) This solution corresponds to the initial state given by (4. Comparing (4. In Fig. a complex α corresponds to a displaced oscillator which has initially a non-vanishing velocity.138.164) where according to (4. 4. being displaced around the center with period 2π/ω and.159) is then 1 1 i φ(α) (y. a momentum factor and a phase factor.148.163) also through the shift operator. The oscillations of the mean position y0 (t1 ) and of the mean momentum k0 (t1 ) are out of phase by π just as in the case of the classical oscillator.

(4. σ|y.1 (second and third row). This behaviour is illustrated in Fig.167) One can derive for the expansion coeﬃcients +∞ dn = −∞ dy φn (y) φ(ao .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 4.166) φ(ao .168) dy e−p(y−q) Hn (y) where p = 1 + σ2 . 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. t0 ) = [πσ 2 ]− 4 exp we expand ∞ 1 − (y − ao )2 2σ 2 (4. To describe such state which satisﬁes the initial condition φ(ao . that of the Glauber states) with a period π/ω. σ|y. σ|y. 2σ 2 q = a . 4. t0 ) = n=0 dn φn (y) . 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 + σ2 (4.169) .1: Time dependence of Gaussian wave packets in harmonic oscillator potential.

P. namely $750 4 .7: Quasi-Classical States The solution of integral (4. p−1 1 − σ4 t = 1 − σ 2 −iω(t1 −to ) e 1 + σ2 (4. a ground state shifted by ao = 2α. i. vol. New York. 2 by A. Marichev (Wiley. t1 ) = √ 2 × exp − 1 (y 2 + yo ) 1+t2 − 2 1−t 2 1 1 √ √ pσ π 1 − t2 × t 1 − t2 (4. This expansion can be written φ(ao . Following the strategy adopted previously one can express the time-dependent state corresponding to the initial condition (4. Prudnikov. Yu.157) as follows φ(ao .143) of the Glauber state.173) which is. 1 a2 2 1 + σ2 + 1 2 2 y o + 2 y yo This integral can be found in Integrals and Series. σ→1 p−1 2 (4.143). σ→1 O(y n−2 ) for n ≥ 2 1 for n = 0.174) Hn (q p p−1 ) φn (y) .176) 1 √ pσ π a 2 exp − 1 1+σ2 + 1 yo − 1 ω(t1 − to ) 2 2 2 2 × (4. In this limit the coeﬃcients dn should√ become identical to the coeﬃcients (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. Brychkov.172) One can then conclude e−a /4 n lim dn = √ a σ→1 2n n! (4.4.76) permits us to write this φ(ao . this 3 volume integral table is likely the most complete today.177) − 1 ω(t1 − to ) 2 . therefore.. σ|y.166) in analogy to (4. t1 ) = e−aq/2 √ pσ ∞ n=0 e −iω(n+ 1 )(t1 −to ) √ 2 1 2n n! p−1 p n/2 (4. in fact.I.e. and O.175) ∞ tn n=0 2n n! Hn (yo ) e−yo /2 Hn (y) e−y 2 2 /2 The generating function (4. 1990). in agreement with the expected result (4. σ|y.168) is4 e−aq/2 (p − 1)n/2 dn = √ Hn (q 2n n!σ p(n+1)/2 p ). a worthy successor of the famous Gradshteyn.170) In order to check the resulting coeﬃcients we consider the limit σ → 1. t1 ) = √ × where yo = q p a = √ . 1 (4. σ|y. p−1 95 (4.171) lim p−1 p n 2 Hn (q p ) = lim (2q)n = an . unfortunately very expensive.

96 Linear Harmonic Oscillator .

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

ϑ parametrizes the rotations uniquely as long as |ϑ| < 2π.(5. i. In the latter case the determinant of R(ϑ = (0.7) implies that a rotation is associated with an inversion.6) (5.e.8) 0 0 1 In case of a prefactor +1. 0.k.7) Let us consider brieﬂy an example to illustrate rotational transformations and to interpret the sign of detR(ϑ). i. are linear and. then the ﬁngers of your ﬁst point in the direction of the rotation. 3 .e. in Cartesian coordinates r = (x1 . (5. (5.e. . We assume the convention that rotations are right-handed. namely around an axis given by the direction of ϑ and by an angle |ϑ|. In the following we want to exclude inversions and consider only rotation matrices with detR = 1. if the thumb in your right hand ﬁst points in the direction of ϑ. 0.1) In Cartesian coordinates this reads 3 xk = j=1 [R(ϑ)]kj xj k = 1. the matrix represents a proper rotation. ϕ)T ) is negative. in case of a prefactor −1 a rotation and an inversion at the origin. they conserve a · b = 3 aj bj .. therefore. (5. It holds then j=1 3 3 a ·b = j. x3 )T . A rotation around the x3 -axis by an angle ϕ is described by the matrix cosϕ − sinϕ 0 R(ϑ = (0.5) This equation states the key characteristic of rotation matrices. a minus sign in Eq.. it follows 3 [R(ϑ)]jk [R(ϑ)]j = δk j=1 .2) Rotations conserve the scalar product between any pair of vectors a and b.3) Since this holds for any a and b. =1 [R(ϑ)]jk [R(ϑ)]j ak b = j=1 aj bj . ϕ)T ) = ± sinϕ cosϕ 0 (5. i. (5. From (5.4) With the deﬁnition of the transposed matrix RT [RT ]jk ≡ Rkj this property can be written R(ϑ) RT (ϑ) = RT (ϑ) R(ϑ) = 1 1 . (5.6) follows immediatety for the determinant of R(ϑ) using detA B = detA detB and detAT = detA detR(ϑ) = ±1 . where ϑ denotes the rotation. can be represented by 3 × 3 matrices R(ϑ). 2. (5.98 Properties of Rotations in R Theory of Angular Momentum and Spin Rotational transformations of vectors r ∈ R . Let us deﬁne the rotated vector as r = R(ϑ) r . x2 .

99 (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 need to demonstrate that this inverse belongs also to −1 T SO(3). (5. one can show R3 R3 = 1 Furthermore. R RT = RT R = 1 detR = 1 } 1. T R1 R1 −1 = 1 −1 = 1 1 1 (5. In the following we will assume R1 .12) (iv) Since the associative law holds for multiplication of any square matrices this property holds for the elements of SO(3). It holds T T T T R3 R3 = ( R1 R2 )T R1 R2 = R2 R1 R1 R2 = R2 R2 = 1 . (ii) The group element ‘e’ is played by the 3 × 3 identity matrix 1 1. in fact. it holds 1. R2 ∈ SO(3) and do not write out “◦” explicitly since it represents the matrix product. (iv) For products of three elements holds the associative law a ◦ (b ◦ c) = (a ◦ b) ◦ c. We want to prove now that SO(3) as deﬁned in (5. For this purpose we must demonstrate that the group properties (i–iv) hold. (ii) There exists an element e in G with the property ∀a ∈ G → e ◦ a = a ◦ e = a.9) forms a group.11) (iii) From detR1 = 0 follows that R1 is non-singular and. a ∈ G → ∃a−1 ∈ G with a ◦ a−1 = a−1 ◦ a = e. there exists a real 3 × 3–matrix −1 R1 which is the inverse of R1 .5.13) . We would like to point out the obvious property that for all elements R of SO(3) holds R−1 = RT (5. It follows that R3 is also an element of SO(3). (iii) ∀a. We have shown altogether that the elements of SO(3) form a group. (i) Obviously. Since (R1 )T = (R1 )−1 it follows −1 −1 −1 T (R1 )T R1 = (R1 )−1 R1 = −1 which implies R1 ∈ SO(3). R3 = R1 R2 is a real. b ∈ G the product c = a ◦ b is also in G. (5. 1 T Similarly. 3 × 3–matrix.9) This set of matrices is closed under matrix multiplication and. forms a group G satisfying the axioms (i) For any pair a.10) detR3 = det ( R1 R2 ) = detR1 detR2 = 1 .

due to the uniqueness of the inverse.15) Below we will construct appropriate matrices Jk .100 Theory of Angular Momentum and Spin Exercise 5. −1 0 0 −1 . 0 −1 1 0 (5. rotation by 270o } and consecutive execution of rotation as the binary operation ‘◦’. −i} together with multiplication as the binary operation ‘◦’. M3 .18) (5. identify subgroups. (b) the set of matrices {M1 . J3 as follows i R(ϑ) = exp − ϑ · J (5. To demonstrate this property consider rotations with ϑ = ϑk ek where ek ˆ ˆ is a unit vector in the direction of the Cartesian k–axis. establish if 1-1 homomorphic mappings exist. Lie Algebra We want to consider now a most convenient. rotation by 90o . i. in fact. compact representation of R(ϑ) which will be of general use and.17) . mappings between the sets which conserve the group properties: (a) the set of real and imaginary numbers {1.1: Test if the following sets together with the stated binary operation ‘◦’ form groups.19) (5.e. Lie Group. we arrive at the property exp ϑk LT k from which follows LT = −Lk .14) together with matrix multiplication as the binary operation ‘◦’. rotation by 180o . 0 1 −1 0 .1.16) must be antisymmetric. will play a central role in the representation of many other symmetry transformations in Physics. the matrices i Lk = − Jk (5.13) R(ϑk ek )−1 = R(ϑk ek )T = [exp ( ϑk Lk )]T = exp ϑk LT ˆ ˆ k and. (c) the set of four rotations in the x1 .20) (5. Generators. we want to assume that they exist. We want to show that for the property R RT = 1 to hold. M4 } = 1 0 0 1 . M2 . x2 –plane {rotation by Oo . J2 . i. −1. This representation expresses rotation matrices in terms of three 3 × 3– matrices J1 . 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. k = exp ( −ϑk Lk ) (5. presently.

lim R[(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).8) we can state cosϑ3 −sinϑ3 0 1 −ϑ3 0 1 0 . We assume now that we parameterize rotations through a three-dimensional rotation vector ϑ = (ϑ1 .21) −b −c 0 It is this the reason why one can expect that three-dimensional rotation matrices can be expressed through (5.(5.1: Matrix Representation of SO(3) 101 We can conclude then that if elements of SO(3) can be expressed as R = eA . Exercise 5. Let us use this deﬁnition to evaluate L3 . ϑ3 )T ] = lim sinϑ3 cosϑ3 0 = ϑ3 ϑ3 →0 ϑ3 →0 0 0 1 0 0 1 Insertion into (5. 0)T the rotation is the identity. ϑ2 . One can determine then the matrices Lk deﬁned in (5..1. 0)T ) − 1 1 1 ϑ1 →0 (5. i.5.26) (5. ϑ2 . L2 .e.23) (5. ϑ3 )T such that for ϑ = (0.2: Determine the generator L1 according to Eq. 0. Using the deﬁnition of partial derivatives we can state L1 = lim ϑ−1 R(ϑ = (ϑ1 . 0. This property gives A then only three independent matrix elements. 0.22) and similar for L2 and L3 .15) with three real parameters ϑ1 . A must have the form 0 a b A = −a 0 c . L3 which are independent of the rotation angles. 0. the real matrix A is anti-symmetric.15) as the derivatives of R(ϑ) with respect to the Cartesian components ϑk taken at ϑ = (0. 0)T . 0 0 0 0 0 0 (5.25) 0 −ϑ3 0 0 −1 0 ϑ3 0 0 = 1 0 0 .24) (5.27) . (5.22). ϑ3 and three matrices L1 . (5. Using (5.

B] ).15) obey the group property. that the operators (5. commutators of commutators of A and B. i.22) implies..32) For the matrices Jk = i Lk which. 0. The expression for Zn are actually rather complicated.30) is called the Lie algebra. In fact. that the generators represent inﬁnitesimal rotations with 1 >> |ϑ| around the x1 -.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. R(ϑ = (0. and x3 -axes.g.31) where k m are the elements which are 0 1 k m = 1 −1 of the totally antisymmetric 3-dimensional tensor. only the latter example and . 1 1 e. x2 -.g. Of course. L ] = m=1 fk m Lm .28) The construction of the generators through the limit taken in Eq. (5. the Lie algebra of SO(3) can be written [Lk .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. = 2. however.. B]. if the resulting products of exponential operators can be expressed in the form (5.30) are called Lie groups. three being a special case. it must hold 3 [Lk .30) Groups with the property that their elements can be expressed like (5. [A. 3 )T ) = exp( 3 L3 ).15) and their generators obey the property (5. (5. e. in fact. (5. Z2 = 2 [A. 0)T ) = exp( 1 L1 ). (5. the property (5. as we see later. J ] = i k m Jm . We want to consider. etc. m = 3 . = 2.e.33) We want to demonstrate now that exp(ϑ · L) yields the known rotational transformations. 2 . m = 3 for any even permutation of k = 1. the matrix (5. however. L ] = k m Lm (5. In case of the group SO(3) the structure constants are particularly simple. B]] + [[A.8) in case of ϑ = (0.. Lie groups can have any number of generators. 0. The question is then. 0)T ) = exp( 2 L2 ) and R(ϑ = (0.29) where Z1 = A + B and where the remaining operators Zn are commutators of A and B. ϕ)T . Z3 = 12 ( [A. An answer can be found considering the Baker-Campbell-Hausdorf expansion ∞ exp( λA ) exp( λB ) = exp( n=1 λn Zn ) (5.15). 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. and the constants fk m are called the structure constants. the elements of if any two indices are identical for k = 1. m = 3 for any odd permutation of k = 1. = 2. i. From this result one can conclude that expressions of the type (5. holds the algebra [Jk . Finite rotations can be obtained by applications of many inﬁnitesimal rotations of the type R(ϑ = ( 1 .e. 0. are related to angular momentum operators. B].

(2n + 1)! (5.37) = (−1)n 1 0 0 1 .8).31).1: Matrix Representation of SO(3) determine exp(ϕL3 ). 5.40) .35). cos ϕ − sin ϕ 0 = sin ϕ cos ϕ 0 0 0 1 (5.39) Recognizing the Taylor expansions of cos ϕ and sin ϕ one obtains eA = and.3: Show that the generators Lk of SO(3) obey the commutation relationship (5. (5. A = ϕ 0 −1 1 0 .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 . (5. Exercise 5. (5. accordingly. We note. eϕL3 which agrees with (5.27).36) The property 0 −1 1 0 allows one to write 0 −1 1 0 and.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 .34. 0 −1 1 0 = (−1)n 0 −1 1 0 (5. ∞ 2n = − 1 0 0 1 2n+1 (5. using (5.41) cos ϕ − sin ϕ sin ϕ cos ϕ (5.35) To determine expA = ∞ ν ν=1 A /ν! we split its Taylor expansion into even and odd powers 1 A2n + (2n)! 2 ∞ n=0 ∞ eA = n=0 1 A2n+1 .1.5. according to (5. that the matrix ϕL3 . can be written ϕL3 = 0 103 A 0 0 0 0 .

1: Assume the deﬁnition ρ ρ (A B) = ρ (B) ρ (A). i. For this purpose we require stringent continuity properties of the wavefunction: the wavefunctions under consideration must be elements of the set C∞ ( ).e. one isomorphic to SO(3) and. i. a relationship which we like to express as follows R(ϑ) = ρ R(ϑ) . its elements can be represented through an exponential R(ϑ) = exp i − ϑ·J (5.48) . for A. in fact.46) (5.42) Obviously. we deﬁne rotations R(ϑ) as linear maps C∞ ( ) → C∞ ( ).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).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) . Show that in this case holds Generators One can assume then that R(ϑ) should also form a Lie group.104 Theory of Angular Momentum and Spin 5.47) = ρ R(ϑ1 ) R(ϑ2 ) . we will learn that transformation patterns and invariances are connected with the angular momentum states of quantum mechanics. = ψ([ R(ϑ1 ) R(ϑ2 ) ]−1 r) = ρ R(ϑ2 ) ψ([R(ϑ1 )]−1 r) (5..43) ρ( ) conserves the group property of SO(3).e.45) (5. the R(ϑ) ψ(r) = ψ(R(ϑ) r ). complex-valued functions over the 3-dimensional space R which can be diﬀerentiated inﬁnitely often. In particular. B ∈ SO(3) holds ρ(A B) = ρ(A) ρ(B) . (5. Since this holds for any ψ(r ) one can conclude ρ R(ϑ1 ) ρ R(ϑ2 ) group property (5. i. (5.2. Exercise 5. the transformation R(ϑ): C∞ ( ) → C∞ ( ) is related to the transformation R(ϑ): R → R . hence. (5.e. In analogy to R(ϑ) being a linear map R → R . namely R(ϑ)ψ(r ) = ψ(R−1 (ϑ) r ) . Deﬁnition We ﬁrst deﬁne how a rotational transformation acts on a wavefunction ψ(r ).44) holds.

(5.2: Function Space Representation of SO(3) where the generators can be determined in analogy to Eq.55) and f (r) = f [J1 . 0. J1 ].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. In fact.57) Exercise 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) . but also the Lie algebra of the generators. 5. in case of J3 .2: (a) Derive the generators Jk . i.54) (5.42)) and similarly for J2 and J3 .50) lim ϑ−1 ( f ( R(0. 0)T ) − 1 1 ϑ1 →0 105 (5. The generators Jk can be determined by applying (5. J ] = i k m Jm . J2 ]. (5. it holds [Jk . ϑ3 )T ) f (r) − f (r) 3 (5.56) holds for [J2 . 3 1 ϑ3 0 x1 x1 + ϑ3 x2 R(0. (5. 2 by means of limits as suggested in Eq. 0.. show that (5. −ϑ3 ) r ) − f (r) ) . k = 1.51) Inserting this into (5. −ϑ3 ) r ≈ −ϑ3 1 0 x2 = −ϑ3 x1 + x2 . (5.56) We like to verify this property for [J1 . One obtains using (5.55) Not only is the group property of SO(3) conserved by ρ( ). J3 ] and [J3 .23. J2 ] f (r) = − = − 2[ 2 [− i J .49) to a function f (r).49) (note the minus sign of −ϑ1 which originates from the inverse of the rotation in Eq.(5.52) ( x2 ∂1 − x1 ∂2 ) f (r) . 0.22) according to i − J1 = lim ϑ−1 R(ϑ = (−ϑ1 .2. 0. 0 0 1 x3 x3 (5.53–5. (5.53) (5.e. Carrying out similar calculations for J1 and J2 one can derive i − J1 i − J2 i − J3 = = = x3 ∂2 − x2 ∂3 x1 ∂3 − x3 ∂1 x2 ∂1 − x1 ∂2 .49).5. i − J3 f (r) = = Using (5.24) one can expand ϑ3 →0 ϑ3 →0 lim ϑ−1 R(ϑ = (0.

3 Angular Momentum Operators The generators Jk can be readily recognized as the angular momentum operators of quantum mechanics. φ) (5. φ). J3 ] J1 + J2 [J2 . x2 = r sinθ sinφ. the system related to the Cartesian coordinates through x1 = r sinθ cosφ. i. J3 ] + [J2 . i.59) which. φ) as well as the eﬀect of the so-called raising and lowering operators J± = J1 ± i J2 (5. and x3 = r cosθ.62) J1 [J1 .e.. identiﬁes J as the quantum mechanical angular momentum operator. C] + [A.56) are the famous commutation relationships of the quantium mechanical angular momentum operators. J3 ] + [J1 . φ) = = 2 ( + 1) Y m (θ. J3 ] = = = 2 2 [J1 + J2 . These eigenstates are the well-known spherical harmonics Y m (θ. Using [AB. Jk ] = 0 . C]B and (5. can be found. 3.60) commutes with all three Jk .64) mY m (θ. J3 ] (5. usually chosen as J3 . C] = A[B. 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. 2. k = 1. We will show now that the properties of spherical harmonics J2Y J3 Y m (θ.61) We demonstrate this property for k = 3. 5. The commutation relationships imply that no simultaneous eigenstates of all three Jk exist. φ). (5. k = 1. the operator 2 2 2 J 2 = J1 + J2 + J3 (5. φ) m (θ.63) (5. (5. Show that for J3 as deﬁned above holds exp( iα J3 ) ψ(φ) = ψ(φ + α). . 3 . according to the correspondence principle between classical and quantum mechanics.56) one obtains [J 2 . using the momentum operator p = ˆ J = r × i . Equation (5.55) can be written as a vector product of r and i − J = −r × From this follows.61) simultaneous eigenstates of J 2 and of one of the Jk . [J 2 .106 Theory of Angular Momentum and Spin Exercise 5.2. However.65) .58) i = r ×p ˆ (5.3: Consider a wave function ψ(φ) which depends only on the azimutal angle φ of the spherical coordinate system (r. θ. 2.e.53–5.

We consider L+ L− | m = ( [L+ . L− ] + L− L+ )| . The arguments are very similar to the ones used above and we can be brief. α −1 β −1 = 2 .e.73) (5. φ) − m (θ. φ) m+1 (θ.66) (5. (∗) (5.1 Let Lk .56). The deﬁnitions of the coeﬃcients α −1 and β −1 yield L+ L− | = L+ ( iβ −1 | − 1 = − α −1 β −1 | . and since we have already shown | − 1 = (iβ −1 )−1 L− | we can state L3 | − 1 = −i( − 1)| − 1 .67) (5. We will show that this property holds then also for m − 1.70) (5. 3 be operators acting on a Hilbert space H which obey the algebra [Lk . k = 1.75) (5.74) (5. in fact. We note L3 L− | = ( [L3 . L1 − iL2 ] = −2iL3 and L+ | = 0 one obtains L+ L− | = −2iL3 | = −2 | . φ) 107 = = 0 = = 0 ( + m + 1)( − m) Y ( + m + 1)( − m) Y m+1 (θ. φ) are a consequence of the Lie algebra (5. For this purpose we ﬁrst demonstrate that the property holds for m = − 1 by considering (i) for the state L− | ∼ | − 1 . i. L ] = k m Lm and let L± = L1 ± i L2 . Noting [L+ . Using [L3 . It holds L+ L− | = (L+ L− − L− L+ + L− L+ )| = ([L+ .72) = 0 = −i | (5. L− ] + L− L+ )| m = ( −2iL3 + . L2 = L2 + L2 + L2 . For this purpose we prove the following theorem.71) To prove this theorem we ﬁrst show by induction that (i) holds. 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. Continuing our proof through induction we assume now that property (i) holds for m.5. i.69) J+ Y (θ. L− ] = [L3 .. φ) (5. namely. φ) (θ. L− ] = [L1 + iL2 . L+ raises the m-value of | − 1 from m = − 1 to m = . J+ Y J− Y J− Y m (θ.e. An Important Theorem Theorem 1.76) We now show that property (ii) also holds for m = − 1 proceeding in a similar way.3: Angular Momentum Operators on the spherical harmonics. 2. From this follows that L− | is an eigenstate of L3 .68) (5. L− ] + L− L3 ) | . L1 − iL2 ] = L2 + iL1 = i (L1 − iL2 ) = iL− and L3 | = −i | we obtain L3 L− | = ( iL− + L− L3 )| = i(− + 1) L− | .

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 . A ﬁrst application will involve the construction of the eigenstates of the angular momentum operators as deﬁned in (5. 5. We note L3 L− | m = ( [L3 .81) i. 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 . in particular. it also holds for m − 1. (5.70. Again the argumemts are similar to the ones used above.72.53–5. Angular Momentum Operators in Spherical Coordinates We want to express now the generators Jk . 5.80) (5. One can normalize the states | m such that αm = βm . 5. This completes the proof of the theorem. it holds αm βm = 2 one obtains αm βm = = ( + 1) k=m+1 k . yield properties (5.78) ( + m + 1)( − m) ( + m + 1)( − m) . We can deduce from (∗) and (∗∗) α α α −1 β −1 −2 β −2 −3 β −3 = 2 = 2( − 1 + ) = 2( − 2 + · · −1 + ) Obviously.69).63..e.83) (5. We note that we can write L2 = 1 1 1 1 2 2 2 2 L+ L− + 2 L− L+ + L3 .84) .79) read L+ | m L− | m + 1 = −i = −i ( + m + 1)( − m) | m + 1 ( + m + 1)( − m) | m . given in (5.64).55). Before we can ﬁnally show that property (iii) holds as well for all − ≤ m ≤ we need to determine the coeﬃcients αm and βm . in terms of spherical coordinates deﬁned through x1 x2 x3 = = = r sin θ cos φ r sin θ sin φ r cos θ (5.108 Theory of Angular Momentum and Spin L− L+ )| m = (−2m − αm βm )| m . We will have various opportunities to employ the Theorem just derived. 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 . θ. Using the familiar formula + m − n k=0 k = n (n + 1)/2 − (m + 1) m = ( + 1) m − (m + 1) m (5.66. 2 We note that Eqs.77) We will show now that if (ii) holds for m. φ.73. 5. (5.79) We can now show that (iii) holds for all proper m–values. i.82) (5.. ﬁnally αm = βm = (5. 5.e. (∗∗) (5. For this purpose we need to express the angular momentum operators in terms of spherical coordinates (r.

92) (5.85).82–5.5. ∂φ = sin φ (5. in terms of spherical coordinates. φ) ∂x ∂x3 ∂ tan φ =0 (5. cos θ ∂ cos2 φ + sin θ cos φ ∂φ cos2 θ sin θ sin φ ∂ sin θ sin φ ∂ + + sin2 θ f (θ.89) = ∂ x2 x2 ∂ x2 + x2 ∂ x3 − 33 − x2 1 3 2 x1 ∂ tan φ r ∂ cos θ r ∂ cos θ f (θ. In fact. Using (5. According to (5.85). actually.3: Angular Momentum Operators 109 We ﬁrst like to demonstrate that the generators Jk .60). the latter deﬁned in (5.86) (5. (L2 )2 = − cos φ ∂ ∂ + cot θ sin φ ∂θ ∂φ 2 . using ∂/∂tanφ = cos2 φ ∂/∂φ and ∂/∂cosθ = −(1/sinθ) ∂/∂θ. φ) = (x3 ∂2 − x2 ∂3 )f (θ. (5. for example. (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 .87) To derive these properties we consider.87) holds J3 = To determine J 2 we note.85) (5.88) (5. using (5.84 ) one obtains. using (5.86). φ) (5. φ) = (5. involve only the angular variables θ and φ and not the radius r. L1 f (θ. We like to express now the operators J3 and J 2 . 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 ∂ = − . φ) ∂x2 /x1 ∂ ∂x3 /r ∂ = x3 + x3 ∂x2 ∂ tan φ ∂x2 ∂ cos θ ∂x2 /x1 /r ∂ ∂ − x2 − x2 ∂x3 3 ∂ cos θ f (θ. applying repeatedly the chain rule. L1 f (θ.93) and similarly. φ) sin θ ∂θ sin θ ∂θ L1 f (θ.91) which agrees with expression (5.90) This yields. φ). φ) = ( x3 ∂2 − x2 ∂3 ) f (θ.

99) = = ∂2 2 ∂ 1 + + 2 2 2 ∂r r ∂r r sin θ ∂2 2 ∂ + 2 ∂r r ∂r sinθ ∂ ∂θ 2 + ∂2 ∂φ2 (5. 5. 5.87). namely on the space of complex-valued functions on the unit 3-dim. 2m r ∂r2 2m r2 2 (5.86.95) sinθ ∂ ∂θ = ∂2 ∂ + cotθ . 5.63.86. 1 sin2 θ sinθ ∂ ∂θ (5.e.64) can be constructed as follows.4 Angular Momentum Eigenstates According to the theorem on page 107 above the eigenfunctions (5. (5.85.94) It follows (L1 )2 + (L2 )2 + (L3 )2 = With cot2 θ + 1 = 1/sin2 θ. These generators act on a subspace of C∞ ( ).85. 2 ∂θ ∂θ (5. ∂θ2 ∂θ ∂φ (5. One ﬁrst identiﬁes the Hilbert space H on which the generators Lk operate. 5. one can conclude J2 = − 2 sin θ 2 sinθ ∂ ∂θ 2 + ∂2 ∂φ2 .98) where Jclass = r × mv is the angular momentum and pr = mr is the radial momentum. ∂θ ∂φ ∂2 ∂ ∂2 + cotθ + ( cot2 θ + 1 ) 2 . The ˙ corresponding expression for the quantum mechanical kinetic energy operator is ˆ T = − which follows from the identity 2 2 2 2m = − 1 ∂2 J2 r + .96) and the relationship between Lk and Jk as given in (5.97).87).101) 1 ∂2 r r ∂r2 and comparision with (5. 5. The functions in C∞ (S ) have real variables . i.100) (5. sphere S2 .97) Kinetic Energy Operator The kinetic energy of a classical particle can be expressed J2 p2 p2 = r + class 2m 2m 2m r2 (5. 5. C∞ (S ). The operators in the present case are Lk = − i Jk where the Jk are diﬀerential operators in θ and φ given in (5.110 = cos2 φ Theory of Angular Momentum and Spin ∂2 1 ∂ ∂2 + sinφ cosφ − 2 cotθ sinφ cosφ ∂θ2 ∂φ ∂θ∂φ sin2 θ 2 ∂ ∂ + cotθ sin2 φ + cot2 θ sin2 φ 2 .

The norm in H is deﬁned through |f |2 = S2 dΩ f ∗ (θ. 0 ≤ φ < 2π. φ). .104. φ)f (θ. 0 ≤ θ < π.5. φ) in H which satisfy L+ Y (θ. φ) = 0 as well as L3 Y (θ. − . The function space endowed with this norm and including only functions for which the integral (5.106) The latter property is obviously satisﬁed for the chosen φ-dependence. (5. can be expressed L± = Accordingly. φ) = 2 + 1 ( − m)! P 4π ( + m)! m (cosθ) eimφ . − + 1. } applying L− Y m+1 (θ. φ. m = − . The raising and lowering operators L± = L1 ± i L2 . m) = 1 −m J− 1 2 Y (θ. φ).102) exists. The eigenfunctions (5.4: Angular Momentum Eigenstates 111 θ.87).104) (5. φ) = = 0 −i Y (θ.64) can be constructed then by seeking ﬁrst functions Y (θ. (5. One obtains in this way Y m (θ. φ) (5. must be cyclic in φ with period 2π.111) . and then determining the family {Y m (θ. . (5. φ) L3 Y (θ. φ) = −i Y (θ. (5.107) reads ∂ ∂ + i cotθ ∂θ ∂φ P (cosθ) ei φ i e±iφ ∂ ∂ ± i cotθ ∂θ ∂φ . φ) = ∆( . .102) where dΩ = sinθ dθ dφ is the volume element of S2 . φ). . φ) . 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. is indeed a Hilbert space. m) Constructing Y → Y →→ ··· Y ( + m)! (2 )! ( − m)! −1 − We like to characterize the resulting eigenfunctions Y m (θ.63. φ) (5.107) (5. 5. the so-called spherical harmonics.105) ∆( .108) m (θ. normalizing these functions. φ) = −i ( + m + 1)( − m)Y m (θ. φ) (5.109) using (5. 5. . . and to be admissable for description of quantum states.105) explicitly. .110) = 0 (5.103) iteratively for m = − 1. by carrying out the construction according to (5. . (5. − 2.85–5.

112 or Theory of Angular Momentum and Spin ∂ − ∂θ cotθ P (cosθ) = 0 (5.106.114. 5. Jackson (John Wiley. (5. for example. π +1 dθ sin2 0 +1 θ = −1 +1 −1 dx (1 − x2 ) +1 = = . i.119) provide the following expression for Y Y (θ. One obtains. (5.106.117) Accordingly..113) 0 is ﬁnite. φ) = (−1) 1 1 = 1 [1 · 3 · 5 · · · (2 − 1)]2 (5.112) where we employed the deﬁnition of P (cosθ) in (5.120) See. (5.118) 2 +1 1 1 sin θ ei 4π (2 )! 2 ! φ . 5. 5. 1975) . ”Classical Electrodynamics. . using the new integration variable x = cos θ. one for which π dθ sinθ |P (cosθ)|2 (5.119) where the factor (−1) has been included to agree with convention1 . is P (cos θ) = N sin θ as one can readily verify.” by J. The normalization factor N is chosen such that π (5.116) The integral appearing in the last expression can be evaluated by repeated integration by parts. A proper solution of this equation. = = 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. 2nd Ed. We note that (5.e. New York. (5.115) holds. .114) yield |N |2 2 +1 1 2π 4π (2 )! π dθ sin2 0 +1 θ = 1.107).D.114) dθ sinθ |Y (cosθ)|2 = 1 0 (5.116) implies |N |2 from which we conclude N = (−1) 1 · 3 · 5 · · · (2 − 1) (5.

123). therefore.64). the eigenfunctions Y m (θ.121) This expressions shows that the factor eimφ . we seek to determine the functions P m (cos θ) and. Let us then assume that (5.4: Angular Momentum Eigenstates 113 We have constructed a normalized solution of (5. We want to demonstrate now that the recursion equation (5. The reader should note that the associated Legendre polynomials.123) P m+1 (x) ∂ x + (m + 1) 1 ∂x (1 − x2 ) 2 . according to (5.124) reads P (x) = = (2 )! (1 − x2 )− 2 (x2 − 1) 2 ! (−1) 1 · 3 · 5 · · · (2 − 1) sin θ (5.103.124) we proceed by induction.114.e. 5.123) leads to the expression P m (x) = −m m ∂ 1 ( + m)! (1 − x2 )− 2 (x2 − 1) 2 ! ( − m)! ∂x −m (5.5. Actually. ∂x −m−1 .124).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 . − deﬁned in (5. describes the φ-dependence of Y m (θ.110). (5.122) The latter identity can be written. indeed.124) called the associated Legendre polynomials.106).63. (5.126) Then holds. namely Y (θ.124) holds for m + 1.. φ). as speciﬁed in (5.119). 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. − 2. For m = (5. i. and can obtain now.125) which agrees with (5. P m+1 (x) = −m−1 m+1 ∂ 1 ( + m + 1)! (1 − x2 )− 2 (x2 − 1) . use (5. . employing again the variable x = cos θ. . are real. To prove (5. 5. 5. 2 ! ( − m − 1)! ∂x −m−1 (5. φ). through repeated application of (5. 5. P m (x) = −1 × ( + m + 1) ( − m) × −(1 − x2 ) 2 1 (5.103.127) ∂ −m−1 (x2 − 1) . m = − 1. m (cos θ) e (5. every application of L− reduces the power of eiφ by one.110) together with (5. φ). .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) .

However. using x = cos θ. φ) as given in (5. In fact.123) terminates for m = − . by repeated application of the operator L+ .128) one can show that (5. therefore. we embark on this construction in order to prove (5. We want to test if the recursion (5. 5. φ). it holds then for all m.124).124) and. φ) L3 f (θ. 2 ! (2 )! ∂x2 1 (5. φ) = Y − (θ..134) (x) = 1 2 ! (1 − x2 ) 2 .129) holds.127) reproduces (5.106.124). φ) choosing the proper sign. and constructs then the eigenfunctions Y − +1 .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. φ) = = determines ﬁrst a normalized solution 0 i f (θ. i. which is equivalent to recursive application of L− . Y − +2 . Since (5. therefore. from such construction emerges an important symmetry property of Y m (θ.124) holds for m if it holds for m + 1. P − (x) = 1 ∂2 (1 − x2 ) 2 (x2 − 1) . it holds ∂2 (x2 − 1) = (2 )! ∂x2 and. (5. (5. φ) = (−1)m Y m −m (θ. if L− Y (θ. yields a vanishing expression for m = − − 1 as long as is a non-negative integer.124) holds for m = . φ) using (5. In that case (1 − x2 ) is a polynomial of degree 2 and.133) Since the term with the highest power of (x2 − 1) is x2 . φ) An alternative route to construct the eigenfunctions Y of L− f (θ. the derivative ∂ 2 +1 /∂x2 +1 of this expression vanishes. therefore. φ) = 0 (5.130) (5. that expression (5. expression (5.132). We ﬁrst determine Y − (θ. P − (5. therefore. It holds.132) which reduces the number of spherical harmonics which need to be evaluated independently roughly by half . hence.106. φ) (5. Y ∗ (θ. appears not very interesting. Constructing Y − → Y − +1 → ···Y m (θ. etc. namely.124) and. Such construction reproduces the eigenfunctions Y m (θ. 5.131) identiﬁes f (θ.135) . φ) .e.

106.143) . (5.110) applies the = ( + m + 1)( − m) Y m+1 (θ. φ). (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. (5. 5. φ) . This follows from the identity ∂ ∂ − i cotθ ∂θ ∂φ One can use (5.132).e.120) for Y obey the postulated relationship (5.136) is normalized and has the proper sign.. We ﬁrst note that for m = − (5. φ) = 2 +1 1 1 sin θ e−i 4π (2 )! 2 ! φ 115 − . According to (5.f.142) yields P − (x) = (−1) 1 (1 − x2 )− 2 (x2 − 1) = (1 − x2 ) 2 2 ! 2 ! (5. (5.122. (5. One can readily verify that this expression provides a solution of (5.138) Employing (5.124) one arrives at Y (θ. 5.136) We note in passing that this expression and the expression (5.123)] P m+1 (x) = −(1 − x2 ) 2 1 ∂ x − m 1 ∂x (1 − x2 ) 2 P m (x) .4: Angular Momentum Eigenstates Due to (5.124).136) to construct all other Y recursion eiφ ∂ ∂ + i cotθ ∂θ ∂φ Y m (θ. Obviously. 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) (5.80) and (5. following closely the proof of eq.139) = ∂ − m cotθ ∂θ P m (cos θ) .129).106). a sign consistent with the family of functions Y m constructed above.142) For this purpose we proceed by induction. (5.137) m (θ. the expression (5. (5. i.140) Introducing again the variable x = cos θ leads to the recursion equation [c. φ) sin θ e−i φ = 0.

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 . one notes that application of (5. Since (5.146) i.149) (5. Y −2 .s. The construction beginning with Y − and continuing with Y − +1 .145) Hence.e. φ) ( − m)! P ( + m)! −m (x) . 2 ! ∂x +m (5.141) reads then P m+1 (x) = 1 (5. yields the same eigenfunctions as the previous construction beginning with Y and stepping down the series of functions Y −1 . etc.. Accordingly. However. equivalently.. of which agrees with the r.142) one can state P (x) = ∂ (x2 − 1) .128) m + 1 → −m or. 1 (1 − x ) ∂x +m (1 − x2 ) 2 1 (5. 2 ! ∂x +m+1 (5.142) holds for m + 1 if it holds for m. (5.106) this implies for Y The Legendre Polynomials The functions the identity (5.142) holds for m.124).142).148) According to (5.144) ∂ x − m 1 ∂x (1 − x2 ) 2 +m m ∂ (−1)m (1 − x2 ) 2 (x2 − 1) . i.h.116 Theory of Angular Momentum and Spin wich agrees with (5. of (5.124) yields (−1)m ( − m)! P ( + m)! −m (x) = +m m ∂ (−1)m (1 − x2 ) 2 (x2 − 1) .144) reads P m+1 (x) = +m+1 m+1 ∂ (−1)m+1 (1 − x2 ) 2 (x2 − 1) . (5.132).147) the r.135). (5. one can conclude P m (x) = (−1)m m (θ.142) holds for m = − we veriﬁed that it holds for all m. According to both (5. Y − +2 . We then assume that (5.e.142) and by (5.124) and (5. Hence. also the associated Legendre polynomials determined this way.150) . 2 ! ∂x 1 (5. (5.h. given by (5. P (x) = P 0 (x) are called Legendre polynomials. 2 ! ∂x +m −(1 − x2 ) 2 Replacing in (5.s. etc. are identical.

J. 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. The generating function allows one to derive useful properties of the Legendre polynomials.156) yields P (1) = 1 . t) = √ 1 − 2xt + t2 ∞ 1 − 2 x t + t2 . (5.4: Angular Momentum Eigenstates The ﬁrst few polynomials are P0 (x) = 1 . . .154) |r1 − r2 | r +1 =0 1 where γ is the angle between r1 and r2 . 2 3 1 1 = = 2 1−t 1 − 2t + t ∞ t . pp. The spherical harmonics for m < 0 can be obtained using (5. the spherical harmonics for m ≥ 0 Y m (θ.156) w(x.151) (5x3 Comparision of (5. 1.g. (5.158) see. and r2 denotes the location of the charge. m ≥ 0 ∂xm (5.150) and (5.155) P (x) t .N. pp..D.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) . Lebedev (Prentice-Hall. P3 (x) = 1 2 117 P2 (x) = − 3x) 1 2 2 (3x − 1) . t) is called a generating function of the Legendre polynomials3 . t) = √ Comparision with (5.153) P (cos θ) eimφ . N. for example.132). Englewood Cliﬀs. (5. ”Classical Electrodynamics. For example.5.154) can be written 1 = w(x. P1 (x) = x . Jackson (John Wiley. =0 (5. r1 is the point where the potential is measured.. φ) = 2 + 1 ( − m)! (−1)m sinm θ × 4π ( + m)! × ∂ ∂ cos θ m (5. .157) = 0. 1975). in case of x = 1 holds w(1. e. 45 of “Special Functions and their Applications” by N. 2nd Ed. 2. New York. . Using x = cos γ. t = r2 /r1 .152) and. =0 (5.” by J. 92 To prove this property see. 1965) which is an excellent compendium on special functions employed in physics. and |r1 − r2 | = r1 (5. accordingly. In case q = 1 and |r2 | < |r1 | holds the identity2 ∞ 1 r2 = P (cos γ) (5.

∂t P (x)t −1 −1 (5. φ) . Inspection of (5. Inversion Symmetry of Y m (θ.163) (5.167) (5.159) Using ∂lnw/∂t = (1/w) ∂w/∂t and multiplying (5. Noting exp[im(π + φ)] = (−1)m exp(imφ) we determine Y Properties of Y m (θ. π + φ) = (−1) Y m (θ. φ) .162) − (2 + 1)x P (x) + P −1 (x) ] t + =1 [( + 1) P +1 (x) The coeﬃcients for all powers t must vanish individually. Accordingly. φ) (5. φ).151)] allows one to determine P (x) for = 1. (5. holds P1 (x) ( + 1) P +1 (x) = = x P0 (x) (2 + 1) x P (x) − P −1 (x) (5. t) holds Theory of Angular Momentum and Spin ∂lnw(x. ∂t 1 − 2xt + t2 (5. . π − θ.160) this diﬀerential equation is equivalent ∞ (1 − 2xt + t2 ) =0 P (x) t + (t − x) =0 P (x) t = 0 (5.166) We want to summarize the properties of the spherical harmonics Y derived above. φ) goes over to Y m (π − θ. π + φ).142) allows one to conclude P m (−x) = (−1) +m P m (x) (5. The spherical harmonics are eigenfunctions of the angular momentum operators J2Y J3 Y m (θ..168) mY m (θ. t) x−t = . . 1. under inversion Y m (θ. Accordingly. 2.164) which. φ) Under inversion at the origin vectors r are replaced by − r. then the coordinates of − r are (r. (5. π + φ).165) since ∂ n /∂(−x)n = (−1)n ∂ n /∂xn .159) by w(1 − 2xt + t2 ) leads to the diﬀerential equation obeyed by w(x. . Due to cos(π − θ) = − cos θ the replacement r → −r alters P m (x) into P m (−x). If the spherical coordinates of r are (r.f. .156) and (∂/∂t) to P (x)t = ∞ ∂w + (t − x) w = 0 . φ) m (θ. φ) = = 2 ( + 1) Y m (θ. t) (1 − 2xt + t2 ) Employing (5. φ) m (θ.118 For w(x. θ. φ) m (π − θ. using P0 (x) = 1 [c.161) Collecting coeﬃcients with equal powers t yields 0 = P1 (x) − x P0 (x) ∞ (5.

φ) (5.173) m+1 (θ. m (5. P +1 (x) P1 (x) = x . 2. The spherical harmonics form. The spherical harmonics form an orthonormal basis of the space C∞ (S2 ) of normalizable. φ) = = ( + m + 1)( − m) Y ( + m + 1)( − m) Y m+1 (θ. (5. φ) 2π 0 (5. . 4π (5.176) = 1 +1 [ (2 + 1) x P (x) − are the Legendre polynomials. The spherical harmonics obey the recursion relationships J+ Y J− Y m (θ. The spherical harmonics Y 0 are Y 0 (θ. uniquely deﬁned functions over the unit sphere S2 which are inﬁnitely often diﬀerentiable π 2π sin θdθ 0 0 dφY ∗m (θ.174) m ≥ 0 P (cos θ) eimφ .) P0 (x) = 1 . φ) m (θ. .170) (5. .179) . φ) ∈ C∞ (S2 ) holds ∞ f (θ. φ) C = =0 m=− π C sin θdθ 0 m Y m (θ.175) (5. 5. a complete basis of C∞ (S2 ). 6. in fact. except for the factor exp(imφ). (5.5.. m 4. φ) = 2 + 1 ( − m)! (−1)m sinm θ × 4π ( + m)! 2 ! × ∂ ∂ cos θ m (5. For the Legendre polynomials holds the orthogonality property +1 dx P (x) P ( (x) = −1 2 δ 2 +1 . φ) . i.171) m = dφ Y ∗ (θ. φ) . φ)f (θ. φ) = 2 +1 P (cos θ) . P −1 (x) ] (5. The spherical harmonics for m < 0 are given by Y −m (θ. φ) where J± = J1 ± iJ2 . where ( = 1.178) 7.4: Angular Momentum Eigenstates 119 2. The spherical harmonics are given by the formula Y m (θ. φ) = (−1)m Y ∗ (θ.177) Note that the spherical harmonics are real. for any f (θ. φ) Y m (θ.172) (5.169) 3. φ) = δ δm m .e.

181) = 0. For this purpose start from the identity +1 ∞ +1 dx w(x. (5. of (5. The spherical harmonics for m (θ.2: Derive the orthogonality property for the Legendre polynomials (5.184) Y22 Y21 Y20 together with (5. (5. φ) = 1 ∂2 r − r ∂r2 ( + 1) r2 h(r) Y m (θ.s.86). 1. (5. =0 t + −1 dxP (x) P (x) .183) (5.1: Derive expressions (5.187) h(r) Y m (θ. Exercise 5.189). t) stated in (5.179) using the generating function w(x.87). 11. The spherical harmonics obey the inversion symmetry Y 10. The spherical harmonics for θ = 0 are Y m (θ Theory of Angular Momentum and Spin = 0.4.4. For the Laplacian holds 2 = = = (5.177).182) Y11 Y10 = = − (5.156).180) 9.188) Exercise 5. π + φ) . (5.189) evaluate the integral on the l.120 8. φ).. t)2 = −1 . .4. 4π (5.185) (5. expand the result in powers of t and equate the resulting powers to those arising on the r.3: Construct all spherical harmonics Y3m (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. Exercise 5.186) (5.s. φ) . φ) = δm0 2 +1 . φ) = (−1) Y m (π − θ.h.h.

= − . . (5. . = . i. }..m m m (θ.5 Irreducible Representations We will consider now the eﬀect of the rotational transformations introduced in (5.... φ) . m = − . φ) ˜ under such rotational transformation by f (θ. = . − + 1. orthonormal basis for the function c m Y m (θ.. φ) (5.5: Irreducible Representations 121 5. 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. on functions f (θ. exp − i ϑ · J . }] .e. called the irreducible representation. φ).5. assumes a particularly simple form. The transformations exp − i ϑ · J . φ) c = . m Y m (θ..e.e. φ) m.196) C∞ (S ) = = X . therefore. φ) (5. exp − i ϑ · J (X ) = X . We like to argue that the matrix representing the rotational transformations of the function space C∞ (S ). φ) ˜ Y m (θ.192) = S2 dΩ Y ∗ (θ . . φ) = exp Since the spherical harmonics Y space C∞ (S ) one can expand ˜ f (θ. φ) exp − ϑ · J .. (5.194) Y m (θ.. i. φ) D(ϑ) = . .191) i − ϑ·J f (θ . . m . = 0..42). ∞. .m D(ϑ) m.195). m = S2 i dΩ Y ∗m (θ. . ˜ f (θ. Comparision with (5. (5. We denote the image of a function f (θ. . . with elements given by (5. .193) shows ∞ = − . 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 . 1.190) provide a complete.− + . . φ ) exp m One can also represent C∞ (S ) C∞ (S ) = [ {Y . (5. For this purpose we consider the subspaces of C∞ (S ) X = [ {Y . φ) i − ϑ·J f (θ. .197) The subspaces X have the important property that (i) they are invariant under rotations exp − i ϑ · J ..195) i. . ∞. . . }] (5. φ ) (5. are characterized completely if one speciﬁes the transformation of any Y m (θ. .193) where [ { } ] denotes closure of a set by taking all possible linear combinations of the elements of the set. φ) ∈ C∞ (S )..− + .

201) From expression (5.n2 .199) allows one to express exp i − ϑ·J = n1 . . 3. If one orders the basis according to the partitoning (5. .e. .199) [J+ ..65). . One = contribute can conclude then that in the expansion (5. i. (5.. n The action of J− on Y m produces Y m−1 or. Similarly. J− ] = 2 J3 . . with blocks of dimension 1.n3 n n n Accordingly.e. in case m = − . i.197) of C∞ (S ) the matrix representation of D(ϑ) assumes a block-diagonal form.e.n3 J+ 1 J− 2 J3 3 . J− .195) [D(ϑ)] m. 5.202) D(ϑ) = (2 + 1) × (2 + 1) . The exponential operator exp − i ϑ · J + ϑ3 J3 )n .194) only spherical harmonics with or. m = δ S2 dΩ Y ∗ (θ..198) 2 2 where ϑ± = ϑ1 ( 1 ϑ− J+ 2 + 1 2 ϑ+ J− ϑ2 . φ) . and J3 1 1 ϑ · J = ϑ− J+ + ϑ+ J− + ϑ3 J3 (5. . one concludes for the matrix elements (5. one can conclude that J+ leaves X invariant.200) leaves X invariant.. one needs to consider the action of J+ 1 J− 2 J3 3 on the subspaces X . (5. can be expanded in powers One employs then the commutation properties [J3 . 5. the transformation exp − i ϑ · J leaves X invariant as well. φ) exp m i − ϑ·J Y m (θ.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 .200) of the rotational transformations one can also conclude that the X are the smallest invariant subspaces of C∞ (S ).200) in (5. One proceeds by expanding ﬁrst ϑ · J in terms of J+ . produces 0. Since the Y m are n3 eigenfunctions of J3 the action of J3 reproduces the basis functions of X .. J± ] = ± J± (5. Since any additive term n n n cn1 . (5. Application of (5. leaves X invariant. J− 2 also leaves n1 X invariant. equivalently. which follow readily from (5.56.n2 . 1×1 3×3 . i..

The representation is referred to as the irreducible representation. k = 1.22.e. Give the corresponding expression of ρ(ϕ).5. the map ρ(ϕ) deﬁned through ρ f (α) → g(ϕ) = f (α − ϕ) = ρ(ϕ)f (α) also deﬁnes a representation of SO(2). (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). (f) The transformation ρ(ϕ) as deﬁned in (e) leaves the following subspace of functions considered in (e) Xm = { f (α) = A e−imα . However. i. (a) Show that SO(2) with the group operation ◦ deﬁned as matrix multiplication. . for which holds RT R = R RT = 1 1. ∈ N} invariant.49). (ii) a second rotation around the new x2 –axis by an angle β. though seemingly natural. 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. and periodic functions f (α). Exercise 5. this parametrization. and for which detR = 1. (d) Show that the similarity transformation T (ϕ) deﬁned in the space of non-singular. f (α + 2π) = f (α). (b) Prove that the elements of SO(2) can be completely characterized through a single parameter.5. A ∈ C. does not provide the simplest mathematical description of rotations. 2.1: Representations of the group SO(2) SO(2) is the set of all 2×2 matrices R which are orthogonal.e. i. 5. real 2 × 2– matrices O through O = T (ϕ)O = R(ϕ)OR−1 (ϕ) with R(ϕ) as in (c) is also a representation of SO(2). Determine the generator of ρ(ϕ) in analogy to (5. (e) In the space C∞ ( ) of inﬁnitely often diﬀerentiable. The three components ϑk . This step reconsiders the parametrization of rotations by the vector ϑ assumed sofar. (iii) a third rotation around the new x3 –axis by an angle γ. is a group. other representations are termed reducible representations. 5. 3 certainly allow one to describe any rotation around the origin.6 Wigner Rotation Matrices We will now take an important ﬁrst step to determine the functional form of the matrix elements of D(ϑ).6: Wigner Rotation Matrices 123 This representation of D(ϑ) is the one wich has blocks of the lowest dimensions possible.

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 (5.207) are equivalent.124 Theory of Angular Momentum and Spin The angles α.209). the l. 5..205) Y m (θ. γ)] = .195) by ˜ Y m (θ. i.e. (5. one can replace (5.203). φ) (5. i (5.203) can be expressed. We will demonstrate that (5. the axis x3 is deﬁned in the coordinate frame which is related to the original frame by the consecutive rotations (i) and (ii).m m. γ will be referred to as Euler angles.s. The axis x2 is deﬁned in the coordinate frame which is related to the original frame by rotation (i).207) which replaces J2 by the inverse transformation (i).211) Using (5. β.s.210) The third rotation in (5. followed by transformation (i). the transformation from the original frame to the rotated frame.203) has the disadvantage that it employs rotations deﬁned with respect to three diﬀerent frames of reference.209) The ﬁrst two rotations in (5. followed by J2 in the original frame.203). m [D(α. γ)] m.. in terms of rotations deﬁned with respect to the original frame. Obviously.208) αJ3 βJ2 = e− αJ3 e− i βJ2 e i (5. (5. is e− i γJ3 = e− i βJ2 e− i αJ3 e− i γJ3 e i αJ3 e i βJ2 (5.h. For any similarity transformation involving operators A and S holds eS A S Accordingly.203) For any ϑ ∈ R one can ﬁnd Euler angles α. Accordingly. can then be written e− i βJ2 − i αJ3 e = e− i αJ3 e− i βJ2 (5. φ) e− γJ3 − i βJ2 − i αJ3 e e . i.e.194. β. φ) [D(α. the transformation from the rotated frame to the original frame. The Euler rotation replaces exp − i ϑ · J by e− i γJ3 − i βJ2 − i αJ3 e e .212) . however. of (5. φ) = S2 dΩ Y ∗m (θ. and the r. m i Y m (θ. (i) and (ii).204) is satisﬁed. in analogy to (5. β. β.. γ ∈ R such that exp i − ϑ·J = e− i γJ3 − i βJ2 − i αJ3 e e (5.h.206) The expression (5.e. we can write e− i −1 = S eA S −1 . i. For this purpose we notice that J2 can be expressed through the similarity transformation J2 = e− i αJ3 J2 e i αJ3 (5.

Best known is the spin of the electron. φ) e−imγ .210) yields the simple result e− i γJ3 − i βJ2 − i αJ3 e e = e− i αJ3 e− i βJ2 e− i γJ3 . in addition.m m. γ)] m. β. one can write [D(α. φ) e− m i βJ2 Y m (θ. 5.217) Accordingly. (5.203) with respect to the original (unprimed) frame one simply needs to reverse the order of the rotations. (5. 5. φ) i i (5. β.. the self-adjointness of the operator J3 one can state dΩ Y ∗ (θ. to redeﬁne the three rotations in (5. The eigenvalue property (5. (5. m (5.206) by ˜ Y m (θ. φ) [D(α. m i Y m (θ. φ) e− αJ3 e− βJ2 Y m (θ.213) i. β.216) Using. φ) e− S2 i γJ3 Y m (θ. (5. φ) = S2 dΩ f (θ.e. γ)] m. 5. φ) Y m (θ. Examples are the other ﬁve members of the lepton family to which the electron belongs. m = e−iαm S2 dΩ Y ∗ (θ.64) yields dΩ f (θ. φ) (5.218) Deﬁning the so-called Wigner rotation matrix dmm (β) = one can express the rotation matrices [D(α.215) The evaluation of the matrix elements (5.310)] an explicit expression for the Wigner rotation matrix (5. the muon µ and its neutrino νµ of the second .219) = e−iαm dm m (β) e−iγm δ ( ) . φ) e− m i βJ2 Y m (θ. φ) .219). m [D(α. φ) . This allows one to express the rotational trasnformation (5. but many other elementary components of matter are endowed with spin-like properties. γ)] = .309.7: Spin 1 2 and the group SU(2) 125 Multiplication from the left with (5. (5. β. φ) e− m S2 i αJ3 f (θ. m ( ) S2 dΩ Y ∗ (θ.215) beneﬁts from the choice of Euler angles for the parametrization of rotational transformations. φ) e−iγm δ (5.7 Spin 1 2 and the group SU(2) The spin describes a basic and fascinating property of matter.5. γ)] m. φ) f (θ.205. the electron e and the electron neutrino νe of the ﬁrst generation.220) We will derive below [see Eqs.214) e− γJ3 = S2 dΩ Y ∗m (θ. φ) = e−im α S2 dΩ Y ∗ (θ.

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

2 S2 = 1 σ2 .k=1 σ σ 3 j k j. Any such A can be expressed in terms of three real parameters x.e.231) .7: Spin 1 2 and the group SU(2) 127 exist three such linear independent matrices. 1 i.229) and avoids commuting the components aj and bk . σ2 = 0 −i i 0 .224) provide a basis in terms of which all traceless. x.31) of the group SO(3). (5.g. z A = In fact. k = 1. it holds. 3 are the Pauli matrices σ1 = 0 1 1 0 .227) which is essentially identical to the Lie algebra (5. 5. the simplest choice being S1 = 1 σ1 . but their comonents must not necessarily commute with each other. y. 1 σj σk = − σk σj for j = k which can also be readily veriﬁed.228) (5. ∈ R . hermitian 2 × 2–matrices A can be expandend.226) The Pauli matrices satisfy very special commutation and anti-commutation relationships.230) where a. One can readily verify the commutation property σj σk − σk σj = [σj . they are associated with the important fermion property of matter.227. b are vectors commuting with σ. (5. z.. σ3 = 1 0 0 −1 .224) Algebraic Properties of the Pauli Matrices The Pauli matrices (5. (5. A = x σ 1 + y σ 2 + z σ3 . e.5.225) σ (5. In fact. 2. σk ]+ = 2δjk 1 . one obtains (σ · a) (σ · b) = 3 j 2 j=1 (σ ) aj bj 3 j k j. it might hold aj bk − bk aj = 0.229) = 1 . According to this property the Pauli matrices generate a 3dimensional Cliﬀord algebra C3 . 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. y. σj 2 (5. The Pauli matrices obey the following anti-commutation properties σj σk + σk σj = [σj . namely. The proof of this relation rests on the commutation relationship and the anti-commutation relationship (5. (5.223) where σk .k=1 j<k + σ j σ k aj bk . At this point we will state a useful property. σk ]− = 2i jk z x − iy x + iy −z .e. 2 S3 = 1 σ3 2 (5. Cliﬀord algebras play an important role in the mathematical structure of physics.k=1 σ σ aj bk + j>k aj bk = 3 j. (σ · a) (σ · b) = a · b 1 + iσ · (a × b) 1 (5.

1 1 1 ( + 1) χ± 2 2 . hence.8 Generators and Rotation Matrices of SU(2) Sofar there is nothing which relates the transformations U .k=1 j>k σ j σ k (aj bk − ak bj ) − 1 2 3 j k j. The two remaining terms yield using σ j σ k = −σ k σ j for j = k and altering ‘dummy’ summation indices 3 j.237) + 1)1 and. 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. The commutation property (5. spin changes sign. only a 720o rotation leaves the spin invariant.222). to rotations in space.k=1 j>k i σl (aj bk − ak bj ) = i =1 σ a×b .234) 5. S3 χ± = ± 1 χ± 2 . For this purpose we choose to replace (5.e. Well almost like it. through the algebra obeyed by the operators Sk [Sk . let say around the x2 –axis by 360o .k=1 (σ σ j>k σ k σ j (aj bk − ak bj ) = (5.232) − σ k σ j ) (aj bk − ak bj ) . yields a · b. (5. In the special case a = b ∈ R holds (σ · a)2 = a 2 1 .231) proves (5.238) S 2 χ± = .236) The matrix elements of this 2 × 2 operator can be labelled by the basis states χ+ and χ− . (5. however. We like to derive this result now by evaluating the transformations of SU(2) explicitly.s. 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 . however.h.230).222) as a rotation vector and you rotate once. The algebra of the generators of a transformation is such a basic property that the ‘accident’ that the generators of spin transformations behave in this respect like the generators of 3-dimensional rotations makes spins appear in their physical behaviour like rotations.128 Theory of Angular Momentum and Spin Using (σ j )2 = 1 the ﬁrst term on the r.222) by the Euler form exp(−iγS3 ) exp(−iβS2 ) exp(−iαS3 ) . since there is a slight diﬀerence: when you interprete the 3-component ϑ in (5.233) This result together with (5. i. (5. 1 (5.227) leads to 3 3 jk j.k=1 j>k 1 2 3 j.k=1 j>k σ j σ k (aj bk − ak bj ) = 1 2 3 j. S ] = i k m Sm (5. This relationship emerges.235) which is identical to that of the generators of SO(3). (5.

243) We note the property of this rotation matrix 2 dmm (2π) (1) = −1 1 . can be constructed formally from spin 2 2 states χ± if one considers the two properties χ± to be carried by two kinds of bosons. i.e. We may.239 ) one expands the exponential operator exp(−iβS2 ). .e.239) in the notation (5.e.219. .220) are [D(α. 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. use the label 2 2 | 1 ± 1 for the states χ± . 1 . . 2 2 We obtain with this notation for the transformations (5. For this purpose one needs to determine the powers of −iβS2 .. (5. (5.236) 1 2m | exp(−iγS3 ) exp(−iβS2 ) exp(−iαS3 ) | 1 m 2 1 2m = e−iγm | exp(−iβS2 ) | 1 m e−iαm 2 (1) 2 = e−iγm dmm (β)e−iαm (5. i. 3 . no two such particles can exist in the same state.244) i. .and sinfunctions and one obtains for the rotation matrices 2 dmm (β) (1) = cos β 2 sin β 2 − sin β 2 cos β 2 . m = − .] can be identiﬁed with the Taylor expansion of the cos. ( − 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. identical particles any number of which can exist in the same state χ+ and χ− . therefore. 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 .219). (5.240) . γ)] m. . namely. following Jourdan and Schwinger. . − + 1. In order to determine the Wigner matrix in (5. .242) The expressions in brackets [. 1. 2.5. β. that states | m for higher quantum numbers = 0.239) where we made use of the notation (5.9 Constructing Spin States with Larger Quantum Numbers Through Spinor Operators In this section we like to demonstrate.5.9: Constructing Spin States with S > 1 2 129 This result implies that the states χ± behave in relation to the generators of SU(2) like an angular momentum state | 1 ± 1 in relation to the generators of SO(3). . (5.245) 5. The complete matrix elements (5.241) 2n+1 Taylor expansion of the exponential operator yields then ∞ exp(−iβS2 ) = n=0 (−1)n β 2 2n ∞ 1 + 1 n=0 (−1)n β 2 0 1 −1 0 (5. . The idempotence property of Sk yields particularly simple expressions for these powers. rotation by 360o changes the sign of the spin state.

. b† )T and b = (b+ . j. . These operators are associated with a given spatial reference system.247) (5. . b ζ = 0 b ζ. 1 . x∗ b = x∗ b+ + x∗ b− + − (5. b ζ = b† .251) One can show using this property and (5. + − For example.254) We will show below that these states are orthonormal and form spin states with higher quantum numbers. −j + 1. b− )T as spinor creation and annihilation + − operators. 1. using (5.e. (5. m) = b† − j−m (j + m)! (j − m)! |Ψ0 . (5.246) The corresponding adjoint operators are denoted by b+ and b− . 2 2 . The operators b† allow one to construct a set of states which represent j + m–fold and j − m–fold ± χ+ and χ− states as follows b† + j+m |Ψ (j.248) that bζ relationships † † b ζ.248) bζ . m) † and bζ obey the commutation (5. . b† )T and b = (b+ . b− † = − sin β † β b + cos b† . i. therefore.5. b † ζ ζ =0 (5. b ζ The States |Ψ(j. The boson character of the operators b† = (b† . i.250) which for x∗ x = x∗ x+ + x∗ x− = 1 represent spinor operators in an arbitrary reference system. j = 0.130 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 = b ζ.253) = δζζ .249) In the following will refer to b† = (b† . b † ζ For the vacuum state |Ψ0 holds = δζζ . 2 + 2 − † (5. m = −j. . . −) bζ . We consider. 3 .252) (5. b− )T is expressed through the commutation relationships + − (ζ.e. also operators of the type x b† = x+ b† + x− b† + − . ζ = +.243) the creation operators in a coordinate system rotated by an angle β around the y–axis are b+ † = cos β β † b + sin b† 2 + 2 − . b† |Ψ0 = χ− − . b† |Ψ0 = χ+ + . b± |Ψ0 = 0 (5. .247.

.10: Algebraic Properties of Spinor Operators 131 5.256) and (5. The Spinor Derivative Operator A most important operation is the action of the annihilation operators b+ and b− on operators which can be expressed as polynomials.248) and from (5. + b+ . possibly inﬁnite power series. in b† and b† . that the annihilation operators play a role similar to the diﬀerential operator in calculus. f b† Obviously. b † + = n b† + n−1 b† + b † + + n = (n + 1) b† + . 5. (5. b† + n = n cn b+ . f (b† ) |state + f (b† ) b+ |state .259) The property is obviously true for n = 0. f (b† ) |Ψ0 .247. If it is true for n then using (5.258) Equations (5. For this purpose we consider the following polynomial of creation operators N j+m j−m f (b† ) = + n=0 cn b† + n . An example + − is the monomial b† which generates the states (5.260) . (5.257) which for (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. f (b† ) + (5.256) In the special case |state = |Ψ0 this reads using (5. (5.254). b† + n b† + b† + b† + + n n n b+ . b† + − well-known for the quantum mechanical harmonic oscillator.255) In order to determine how this operator is modiﬁed by multiplication from the left by b+ we note b+ f (b† ) |state = + b+ .249). + + (5.248) we obtain b+ . we need to evaluate b+ . b† + n . b† + = n b† + n−1 .249) b+ f (b† ) |Ψ0 = + b+ . We will show that for this commutator holds n b+ .255) is b+ . We will derive the result.5. b† + b+ .257) motivate us to determine the commutator (5. b† + n+1 = = b+ .

In order to derive (5. (5. f (b† ) + where f (x) = df dx . b† ) .. f (b† ) − = f (b† ) − (5.261) Similarly. + − (5. (5.1. b† ) in b† and b† holds (ζ = +. −) + − + − bζ . m) φjm (x) = j−m xj+m x− + (j + m)! (j − m)! . By induction we can conclude that (5. m) We want to prove now the property ∞ j exp xb † |Ψ0 = 1 j=0. One can ﬁnally conclude for polynomials (5.250) and where φjm (x) represents the function of the two variables x+ and x− closely related to |Ψ(j.266) where xb† has been deﬁned in (5.268) . b† ) |Ψ0 .265) Generating Function of the States |Ψ(j.263) According to (5. (5..255) b+ . φjm (x) Ψ (j.e. the property holds also for n + 1. b† ) + − = ∂ ∂b† ζ f (b† . m) = x− b† − j−m (j + m)! (j − m)! |Ψ0 .257) we can conclude in particular bζ f (b† . f (x)] = ∂k f (x) .266) we compare the terms x+ b† + j+m φjm (x) Ψ (j.264) This demonstrates the equivalence of the spinor operators bζ and the derivative operation. 2 . f (x)] g(x) = (∂k f (x))g(x) or [∂k . f (b† .259) holds for all n ∈ N.267) exp xb† |Ψ0 is called a generating function of |Ψ(j.263) corresponds to the product rule of calculus ∂k f (x)g(x) = (∂k f (x))g(x) + f (x)(∂k g(x)) which can be written ( ∂k f (x) − f (x)∂k ) g(x) = [∂k . Equation (5..132 Theory of Angular Momentum and Spin i. m) . + − (5. = f (b† ) + (5. m) m=−j (5. one can prove b− . b† ) |Ψ0 = + − ∂ ∂b† ζ f (b† .262) Because of the commutation relationships (5.247) the more general property for polynomials f (b† .

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. ...1.275) = exp xb† |Ψ0 . m) = = |Ψ0 (N − s)! s! N −s 1 N! x+ b† x− b† + − N ! (N − s)!s! N −s 133 (5. (5.. .269) (5.1. 2 2 leads to ∞ j ∞ φjm (x) |Ψ (j. 2 ∞ 1 xb† 2j! u 2j |Ψ0 = u=0.. 1. m) 1 j=0. s! (n − s)! Deﬁning j+m = N −s j−m = s. . (5. .. 1 .e. 1 .272) The summation over s can be written in terms of j and m using (5. i. one obtains φjm (x) Ψ (j.5.274) Summing this expression over 2j = 0.270) N 2j j → s=0 j−m=0 → m=−j ... 2.1. 3 . m) m=−j = = 1 x+ b† + x− b† + − (2j)! 2j 1 xb† |Ψ0 ...10: Algebraic Properties of Spinor Operators with the s–th term in the binomial expansion of (a + b)n n! an−s bs . 1 xb† u! |Ψ0 (5. (2j)! 2j |Ψ0 (5. .. chosing the summation index j = 0. 2. 1. 2 .270) s x+ b† + N −s x− b† − s |Ψ0 . .273) Change of the summation indices allows one to conclude j φjm (x) Ψ (j.. m=−j = j=0.

277) by Ψ0 |e(x b) e(yb ) Ψ0 . b− )e(yb ) |Ψ0 = f (y+ . hence.249) for any non-vanishing integer s holds + − Ψ0 |bs Ψ0 = 0 and.279) One can generalize this to † † s (bζ )s e(yb ) |Ψ0 = yζ e(yb ) |Ψ0 .282) where we deﬁned x∗ y = x∗ y+ + x∗ y− . b− ) holds † † f (b+ .m φjm (x∗ ) φj m (y) Ψ (j. m holds.264) implies † † bζ e(yb ) |Ψ0 = yζ e(yb ) |Ψ0 . The generating function allows one to derive various properties of the states |Ψ (j.276) = j. m) and will be used for this purpose below. [b† .h. (5. one can write † † e(xb ) Ψ0 |e(yb ) Ψ0 † ∗ Ψ |e(x y) e(yb ) Ψ = 0 0 = e (x∗ y) † Ψ0 |e(yb ) Ψ0 (5. m .283) (5. m) We want to show now that the states (5. we can conclude † † ∗ e(xb ) Ψ0 |e(yb ) Ψ0 = e(x y) (5. According to (5.m. m) |Ψ j . therefore.134 Theory of Angular Momentum and Spin which concludes our derivation.s. b− ] = 0.j . For this purpose we consider the inner product † † e(xb ) Ψ0 |e(yb ) Ψ0 = δjj δmm (5. that Ψ (j.281) and. b† ] = 0 and [b− .277) To evaluate this expression we ﬁrst notice † e(xb ) † = e(x ∗ b) .280) The commutation properties [b+ . (5. b† ] = 0 allow one to state that for + − + any polynomial f (b+ . y− )e(yb ) |Ψ0 (5.278) † ∗ which allows us to replace the l. (5.e. of (5.284) . Orthonormality of the States |Ψ (j. i. m) |Ψ j .254) are orthonormal. In order to evaluate the ∗ b) (yb† ) operator e(x e we notice that the derivative property (5. (5.

k = 1. Jj ] = i This property is derived as follows [Ji . 2 . 1.223.10: Algebraic Properties of Spinor Operators and comparing (5.m. (5. 1 New Representation of Spin 0. Jj ] = = 1 4 1 4 < n |σi | m >< n |σj | m > b† bm . .277) φjm (x∗ ) φj m (y) Ψ (j.n .n .288) The three operators obey the Lie algebra of SU(2) [Ji .275) one can show e(x ∗ y) = j. b† bm n n n.m. J1 = J2 = J3 = 1 † b b− + b† b+ − 2 + 1 † b b− − b† b+ − 2i + 1 † b b+ − b† b− . In our present notation the matrix elements < ζ |σk | ζ > for ζ = ± correspond to the matrix elements < ζ |σk | ζ > for σ = ± 1 . (5.224). (5.m φjm (x∗ ) φjm (y) . b† bm n n n.5. m j.m +b† b† . m) as given in (5. States 2 We want to demonstrate now that the states |Ψ (j. bm bm } n n = 1 4 < n |σi | m >< n |σj | m > {b† bm δmn − b† bm δm n } n n n. One deﬁnes corresponding operators Jk through 1 Jk = b† < ζ |σk | ζ > bζ . (5. .224).m. 3 . 5.268–5.285) Following steps similar to those in Eqs. m) |Ψ j .j .276).282) and (5. 2. etc. < + |σ1 | − > = < − |σ1 | + > = 1. − 2 + (5. using < + |σ1 | + > = < − |σ1 | − > = 0. .287) ζ 2 ζ..286) from which follows immediately the orthonormality property (5.m 135 = e(x ∗ y) .254) are eigenstates of operators J 2 and J3 with eigenvalues j(j + 1) and m where J 2 and J3 are representations of the spin 2 2 2 operators S 2 = S1 + S2 + S3 and S3 deﬁned in 5.m ijk Jk .m . The 2 operators are explicitly. 3 denote the Pauli spin matrices (5.ζ where σk .m.289) < n |σi | m >< n |σj | m > {b† bm .n . (5.

(5. (5. 2 J 2 = J+ J− + J3 − J3 = J3 (J3 − 1) + J+ J− (5.290) |Ψ (j. J1 ] + i [J1 .m. To this end we note 2 2 2 J 2 = J1 + J2 + J3 = 1 1 2 J+ J− + J− J+ + J3 2 2 .297) . J+ ] = [J1 − iJ2 .291) Here we have deﬁned J+ J− = J1 + iJ2 = J1 − iJ2 = b† b− + = b† b+ . m) as Eigenstates of J 2 and J3 We wish to show now that the states |Ψ (j.294) .292).295) The last result together with (5.292) The commutation relationships (5. − (5.m.m.136 = 1 4 − = = 1 4 1 4 Theory of Angular Momentum and Spin < n |σi | m >< m |σj | m > b† bm n n. J2 ] = −2 J3 from which we can conclude the property 1 1 J− J+ = J+ J− − J3 2 2 Hence. (5. m) are eigenstates of J 2 and J3 .m < n |2i n.m < n |[σi .k ijk ijk σk | m > b † bm n = i = i 1 2 < n |σk | m > b† bm n n.m 1 4 < n |σj | m >< m |σi | m > b† bm n n . σj ]| m > b† bm n n.289) yield [J− .m ijk Jk . (5. J1 + iJ2 ] = −i [J2 .293) . (5. (5.296) ( b † + b + b † + b+ + b † + b + b † − b − − 2 b † + b + − b † − b − b † + b + + b † − b − b † − b − + 2 b † − b − + 4 b † + b − b† − b + ) (5.288) 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 .

305) = = j ( j + 1 ) |Ψ (j. n2 . m = m||λ. . 1. m I1 = where λ = 0. [Ij .301) Obviously. m) m |Ψ (j. n1 = 0. 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. Construct. . 2 2 137 (5. m = λ||λ.254) are eigenstates of + b+ and − b− with eigenvalues j + m and j − m.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) . m) . m − 1) .303) Exercise 5.s. .e. I3 ||λ. b+ ] = δjk . −λ + 1. . Ik ] = i jk I . b† . b † + b + + b † + b+ b † − b − 1 1 b† + b + + b † + b + b † − b− + b † − b − b † + b + . (5. I3 = (b+ b1 + b+ b2 ) 2 2 2 2 2i 2 1 satisfy the Lie algebra of SU(2). m . . i.10. n} eigenstates of I 2 = I1 + I2 + I3 and of I3 I 2 ||λ. n = n1 + n2 ﬁxed. [bj . 2 .299) (5. j = 1. . .298) (5.. m) J− |Ψ (j. m) = = (j + m + 1) (j − m) |Ψ (j. m + 1) (j + m) (j − m + 1)|Ψ (j. the number of states to the possible energy eigenvalues. the states (5. 1 2.h.300) (5. I2 = (b1 b2 − b+ b1 ) . (5. . 2. State the corresponding eigenvalues and the degree of degeneracy. . using operators I± = I1 ± iI2 2 2 2 and the subspace {|n1 . . m = −λ. . . ˆ respectively. Show λ = n1 +n2 n1 +n2 2 ( 2 + 1) and m = 1 2 (n1 − n2 ). can be written b† + b − b † − b + = = One can then state J2 = 1 (b† + b+ )2 + b† + b+ b† − b− + b† − b− b† + b+ 4 + (b† − b− )2 + 2 b† + b+ + 2 b† − b− Deﬁning the operator 1 ˆ k = ( b† + b+ + b† − b− ) 2 one can write ﬁnally ˆ ˆ J2 = k ( k + 1 ) b† .10: Algebraic Properties of Spinor Operators The last term on the r.5. One can then conclude J 2 |Ψ (j. and eigenstates of k with eigenvalues j. m) One can furthermore derive readily J+ |Ψ (j.e. .304) (5. (b) Show that the three operators 1 + 1 1 + (b1 b2 + b+ b1 ) . 1 2 k (a) Show that the eigenstates are given by 1 |n1 .302) (5. i. m) J3 |Ψ (j. λ.

The combination of σ. m) in the original coordinate system by j |Ψ j. for example.306) and (5.251). The prefactor of (b† )j+m (b† )j−m which according to (5. (j) (5.5.306.243).306) where b † and b † are given by (5. In case j = 1 this expression yields.309) .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† . σ values which yields (b† )j+m (b† )j−m is + − then σ = j + m − σ. However. m ) are related to the + − states |Ψ (j.310) it reduces to (5.307) Comparision of (5. For this purpose we note that the states |Ψ(jm) in a rotated coordinate system according m to (5.138 Theory of Angular Momentum and Spin 5. + − One may expect that these two conditions restrict both σ and σ.308) The latter sum involves terms (b† )j+m (b† )j−m for σ + σ = j + m and 2j − σ − σ = j − m. . On the other side the states |Ψ (j. m = m=−j dm m (β) |Ψ (j.307) can be + − (j) identiﬁed with the elements dm m (β) of the rotation matrix can then be written (j) dm m (β) = (j + m)!(j − m)! (j + m )!(j − m )! β 2 j−m × (−1)j−m −σ sin σ=0 2j−m−m −2σ j+m j+m−σ cos β 2 j−m σ m+m +2σ × (5. 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 (5.254) are |Ψ j. m = b† + j+m b† − j−m (j + m )! (j − m )! |Ψ0 (5. both conditions are satisﬁed for σ = j + m − σ.11 Evaluation of the Elements dj m (β) of the Wigner Rotation m Matrix The spinor algorithm allows one to derive expressions for the Wigner rotation matrix elements dj m (β). For this purpose we expand + − + − b† + cos β b† + sin β b† 2 + 2 − j+m σ =0 j−m σ=0 j+m j+m b† − j−m = j−m −sin β b† + cos β b† 2 + 2 − cos β 2 j−m −σ σ = j+m σ σ sin β 2 σ +σ j+m −σ j−m σ −sin β 2 cos β 2 b† + b† − 2j−σ −σ (5. m) .

is the expected element of SO(3). e.251).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 .309) does not distinguish between SU(2) transformations (5.312) transforms like an angular momentum state ˜ |1 m .312) The choice of this transformation derives from a property shown further below. This factor implies.308) unaltered except for a factor (−1)2j which multiplies then also the ﬁnal result (5. conclude that the mapping from SU(2) onto SO(3) is a 2–1 mapping.314) in case of integer j–values.308) and had the form (5.251) and (5. . the orthogonal 3 × 3 matrix which describes a rotation around the x2 –axis. 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.e.12 Mapping of SU(2) onto SO(3) The representation (5. 2 2 − (5. of (5. b− † = sin β † β b+ − cos b† . namely that the component of the vector on the l. = U x2 . The SU(2) transformation entered in (5. Replacing the latter transformation by its negative form.309). the matrix A should represent a rotation around the x2 –axis in the space of vectors ˜ (x1 .309) and the particular matrix (5. This can be shown by (1) applying to the matrix A = (dm m ) in (5.310).h.311) (5.310) to the SU(2) transformation assumed in deriving the general result (5.s. i.g.314) leaves (5. therefore.313) × 0 0 2 = 0 sinβ 0 cosβ 1 −i 0 which. b+ † = −cos β † β b+ − sin b† 2 2 − . in case of j = 1.. One can. It is of interest to trace the mapping from SU(2) onto SO(3) as represented by (5.310) establishes a mapping of SU(2) onto SO(3). x2 . x3 ) ∈ R ..e. however. in fact. that the the representation (5.12: Mapping of SU(2) onto SO(3) 139 5. U = 0 1 1 i √ ( x1 − ix2 ) √ x3 − √2 0 2 2 (5. i. Hence.5.

140 Theory of Angular Momentum and Spin .

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

n = 1. mA mB + mB mC + mA mC 2.max . . 2 . composite systems behaving like elementary objects are common. . a very large number. .max = 10 one obtains d(B) = 14 641. (6.max 2. . ρ 1 = 0. each of which stands for two angles. 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.max d(B) = 1 =0 (2 1 + 1) 2 =0 (2 2 + 1) = ( 1. .max = ∆ 2 mA mB mC E .max . 2. to separate the 14 641 dimensional space of rotational states Y 1 m1 (ˆAB ) Y 2 m2 (ˆC ) into r ρ subspaces Bk . − 2 ≤ m1 ≤ (6. . by the total energy E. . vibrations and rearrangement of the particles in reactions like AB + C → A + BC. . . i.max + 1)2 (6.max + 1)2 ( 2. 2} 1. albeit puzzling.max denote the largest orbital and rotational angular momentum values. . Obviously.m2 Y 1 m1 (ˆAB ) Y 2 m2 (ˆC ). 1. the values of which are determined by the size of the interaction domain ∆V . that only states within the subspaces Bk are coupled in the scattering process. 1.4) 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.max and 2. since further important degrees of freedom.1) where 1. and the following mathematical description will shed light on their ubiquitous appearence in physics. mC . as we will demonstrate below. In fact.e. have not even be considered. in fact.. rather than by Classical Mechanics. 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 ). the dimensions d(Bk ) of these subspaces does not exceed 100. by the masses mA . and by the moment of inertia IA−B of the diatomic molecule A–B approximately as follows 1. in particular.142 Addition of Angular Momenta and Spin ρC . . 2.3) For rather moderate values 1.2) The dimension d(B) of B is 1. Such large number of dynamically coupled states would constitute a serious problem in any detailed description of the scattering process. 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.m1 . The rotational symmetry of the interaction between the particles allows one.max = 2. however. . 14 641} .m1 2 . (6. = 0. The latter often arrives at the physical properties of composite systems by adding the . mB . . will make their appearence a natural consequence of the symmetry of the building blocks of matter. − 1 ≤ m1 ≤ 1. = B such. Such extremely useful transformation of the problem can be achieved through the choice of a new basis set B = { 1 .m2 c (n) r ρ 1. B1 ⊕ B2 ⊕ .max = 1 2 IA−B E .

5) according to (5. ∂yAB ∂zAB i (2) ∂ ∂ − J3 = ρy − ρx . the commutation relationships (1) Jp . Describing quantum mechanically a property of a composite object as a whole and relating this property to the properties of the elementary building blocks is then the quantum mechanical equivalent of the important operation of addittion. 3 (6. the existence of a basis (6.6) For the following it is essential to note that H is not invariant under rotations of only rAB or ρC . the reader will learn in the following section how to add and subtract in the microscopic world of Quantum Physics. connected with the fact that the Hamiltonian H does not depend on the overall orientation of the three interacting particles. but solely under simultaneous and identical rotations of rAB or ρC .. In this sense. i. R−1 (ϑ) ρC ) (6.7) i R(ϑ) = exp − ϑ · J (6. For ˆ ˆ example. Hence. The invariance of the Hamiltonian under overall rotations of the three particle system implies then H = R(ϑ) H R−1 (ϑ) .0: Addition of Angular Momenta and Spin 143 corresponding physical properties of the elementary components. q = 1.6.e. It holds then H R(ϑ) ψ = R(ϑ) H ψ. rotations R(ϑ) of the wave functions ψ(rAB . be diﬀerent from H. according to (5. in fact. Since this is true for any ψ(rAB . 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. presumably a facility the reader would like to acquire with great eagerness. 2. (6.53. Following our description of rotations of single particle wave functions we express (6. To specify this property mathematically let us denote by H the Hamiltonian in the rotated frame.7) where the generators J (k) are diﬀerential operators acting on rAB (k = 1) and on ρC (k = 2).8) Obviously.48) i i R(ϑ) = exp − ϑ · J (1) exp − ϑ · J (2) (6. 5. Jq(2) = 0 for p. One can equivalently express therefore (6. assuming presently that H might. ∂ρx ∂ρy (6. ρC ) = ψ(R−1 (ϑ) rAB . Rotational Symmetry of the Hamiltonian As pointed out already. ρC ) deﬁned through R(ϑ) ψ(rAB .5) do not aﬀect the Hamiltonian. ρC ) it follows H R(ϑ) = R(ϑ) H.10) .9) hold since the components of J (k) are diﬀerential operators with respect to diﬀerent variables.55) holds i (1) − J1 = zAB ∂ ∂ − yAB .4) which decouples rotational states is connected with the rotational symmetry of the Hamiltonian of the three particle system considered. 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).

14) implies that the total angular momentum is conserved during the scattering process. i. In fact.14) We will refer in the following to Jk . 3 do not commute. k = 1. 2. 2 together with the property [A. however.16) following the procedure stated in the theorem above [c.11) we can write the condition (6. and the eigenvalues of J2 and J3 are good quantum numbers. the important fact that the Jk obey the Lie algebra of SO(3). 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. The property (6. 2. B] + [A. for |ϑ| H ≈ 1 − 1 i ϑ·J H 1 + 1 i ϑ·J ≈ H + i 1. (6.14) implies that the components of the total angular momentum operator (6. 3 . B + C] = [A. [H.6) for rotational invariance of the Hamiltonian in the form i i H = exp − ϑ · J H exp + ϑ · J . the operators Jk .13) Hϑ · J − Since this holds for any ϑ it follows HJ − J H = 0 or. of course. For a proof one uses (6.81)].f. 3 cannot have simultaneous eigenstates among each other. k = 1. that energy.144 where J = J (1) + J (2) Addition of Angular Momenta and Spin . (6. ˆ Ω 2 = ρC . we will ﬁnd that the states yield the basis B with the desired property of a maximal uncoupling of rotational states. 2. C].71–5. (6. (6. Jk ] = 0 . To order O(|ϑ|) one obtains i ϑ · JH .81) we want to consider the relationship between YJM and Y 1 m1 (ˆAB ) Y 2 m2 (ˆC ). Before we apply the procedure (5. We recognize. J ] = i k m Jm (6.. i.11) By means of (6.12) We consider this equation for inﬁnitesimal rotations.9). k = 1.e. 2. Eqs. componentwise. i. the (n) (n) (n) properties [Jk .17) . r ρ Deﬁnition of Total Angular Momentum States The commutation property (6..e.12) each individually can have simultaneous eigenstates with the Hamiltonian. 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 ).15) are satisﬁed. (5. One can show readily that the commutation relationships [Jk . ˆ (6.e. that the components Jk . We suspect. In the following we will use the notation r ρ Ω1 = rAB . J ] = i k m Jm for n = 1. k = 1.71–5. 3 as the three components of the total angular momentum operator.

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

. . 2 ) and B ( 1 . J. 1 + 2 . . . J = | 1 − 2 |. J} . Ω2 ) YJ M ( 1. −J + 1. (6. . respectively. J (1) and J (2) . and that the states can be normalized. it is not immediately obvious what the J– values are. (6. . Exercise 6. . The property dΩ1 dΩ2 Y∗ ( 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.h. | 1 − 2| + 1. The basis B ( 1 . For each quantum number J there should be 2J + 1 elements YJM with M = −J. | 1 − 2 | + 1. . . |J (1) | + |J (2) |]. Ω2 ) = δJJ δM M (6.18) constitutes a change of an orthonormal basis B( 1 . . This obviously corresponds quantum mechanically to a range of J–values J = | 1 − 2 |.e.23) corresponding to the r. JM 2 |Ω1 . .1. M = −J. in fact. property implies dΩ1 dΩ2 Y 1 m1 corresponding to the l. . (6.24) The basis B( 1 . (6. − 2} 1 + 1.m2 2 M and with (6.28) This property follows from the fact that the basis elements are eigenstates of hermitian operators with diﬀerent eigenvalues.26) We will show below in an explicit construction of the Clebsch-Gordan coeﬃcients that. B2 2) should be 2 |Ω1 . − + 1. the range of values assumed for J is correct. .2: Prove Eq.h. 2) = {Y 1 m1 (Ω1 ) Y 2 2 m2 (Ω2 ). . 1 + 2.s. 2 ). = {YJM ( 1 . However. . . (6. (6. In fact. 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 . Ω2 ). −J + 1. 2 ) has (2 1 + 1)(2 2 + 1) elements.25) i.25) We want to state now two summation conditions which follow from the orthonormality of the two basis sets B( 1 . to a new basis B ( 1 .18) applied to Y∗ and to YJ JM (J M | m1 . . m1 = − 1 . the basis B ( 1 .s.146 Theory of Angular Momentum and Spin The expansion (6.24) yields 1 m1 2 m2 ) ∗ 1 m1 2 m2 ) (J M| = δJJ δM M . m2 = − 2 .27) together with (6.. In this case the range of |J|–values would be the interval [ |J (1) | − |J (2) | . . Our derivation below will also yield real values for the Clebsch-Gordan coeﬃcients. 1. . it holds 1+ 2 ( 2J + 1 ) = (2 J=| 1 − 2 | 1 + 1) (2 2 + 1) . 2 ) is also orthonormal2 and must have the same number of elements. . 2) . . 2 |Ω1 .

30) The latter property follows from multiplying (6. e. A similar 2 construction will also be applied to composite systems governed by other symmetry groups. Ω2 ) and integrating. 1 m1 2 m2 ) = 0 if J < | 1 − 2| . 2 ) span the same function space.18). 2 |Ω1 .33) JM The latter summation has not been restricted explicitly to allowed J–values. We can.71–5. rather the convention (J M | has been assumed. Ω2 )Y 1 m1 (Ω1 )Y 2 m2 (Ω2 ) .29) where the expansion coeﬃcients are given by the respective scalar products in function space cJ M = dΩ1 dΩ2 Y∗ J M ( 1..e.e.2 Construction of Clebsch-Gordan Coeﬃcients We will now construct the Clebsch-Gordan coeﬃcients.g.m2 M ( 1.. Ω2 ) .31) are identical to Comparision with (6. (6.34) 6. At this point we like to stress that the construction will be based on the theorems (5.81) stated above. 2 |Ω1 . also apply the results. therefore.2: Construction of Clebsch-Gordan Coeﬃcients 147 The second summation condition starts from the fact that the basis sets B( 1 .. and actually also the properties of Clebsch-Gordan coeﬃcients stated above. One can show readily using the same reasoning as applied in the derivation of (6.6. Ω2 ).. The (J M | 1 m1 2 m2 ) cJ M . i. (6. it is possible to expand Y 1 m1 (Ω1 )Y 2 m2 (Ω2 ) in terms of YJM ( 1 .18) by Y∗ J orthogonality property (6. will be based solely on the commutation properties of the operators J and J (k) . 1+ 2 J Y 1 m1 (Ω1 )Y 2 m2 (Ω2 ) = J =| 1 − 2 | M =−J cJ M YJ M ( 1.27) yields δJJ δM M = m1 . 2 ) and B ( 1 . 2 |Ω1 . (6. i. the group SU(3) in case of meson multiplets involving two quarks.1 states.18) that the Clebsch-Gordan coeﬃcients obey the second summation condition (J M | 1 m1 2 m2 )∗ (J M | 1 m1 2 m2 ) = δm1 m1 δm2 m2 . i. The result of this construction will include all the properties previewed above. (6.28) from (6. Y = J =| 1 − 2 | M =−J 1 m1 M (Ω1 )Y 1+ 2 2 m2 (Ω2 ) (J M | ∗ 1 m1 2 m2 ) YJ M J ( 1. (6.32) which complements (6. 2 |Ω1 . Hence.e. . Ω2 ) .28) allows one to conclude that the coeﬃcients cJ (J M | 1 m1 2 m2 )∗ . or J > 1 + 2 (6. or baryons multiplets involving three quarks. 2 |Ω1 . to composite systems involving spin.

. J} by applying repeatedly J− YJM +1 ( 1 .37). Ω2 ) = Y 1 1 (Ω1 ) Y 2 2 (Ω2 ) (6.5. We ﬁrst seek an unnormalized solution GJJ and later normalize. 1 + 2} and for (6.18) which satisﬁes J+ YJJ ( 1 .11) J+ by J+ + J+ . . have shown Y 1+ 2. . Ω2 ) 2 |Ω1 . Ω2 ) .38) for M = J − 1.39) into (6. Ω2 ) (Ω1 ) dΩ2 Y 2 2 (Ω2 ) = 1. . J − 2. (6. −J. 2 |Ω1 . | 1 − 2| + 1. we can demonstrate condition (6. therefore. For this purpose we insert (6. M = −J. −J + 1.37) The solution needs to be normalized. Ω2 ) = Y 1 1 (Ω1 ) Y 2 2 (Ω2 ) .42) ( 1.39) which we will try for a solution of (6. we can also demonstrate using (??) that G dΩ1 = We.36) = 0 = J YJJ ( 1 . Ω2 ) = (J + M + 1)(J − M )YJM ( 1 . 2 |Ω1 .43) . . 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. 2 |Ω1 . Similarly. Ω2 ) is normalized dΩ2 G dΩ1 Y 1 1 1+ 2. Ω2 ) .35) 1 − 2 |.37) and replace (1) (2) according to (6. 2 |Ω1 . 1+ 2 In fact. 2 |Ω1 . (6. Ω2 ) Addition of Angular Momenta and Spin (6. . . (6. 1+ 2 ( 1. The construction assumes a particular choice of J ∈ {| such J–value seeks an expansion (6. 1+ 2 ( 1. 2 |Ω1 . The corresponding classical total angular momentum vector Jclass (1) (2) would be obtained by aligning both Jclass and Jclass also along the z–axis and adding these two vectors.41) 2) Y (Ω1 ) Y (Ω2 ) . J3 YJJ ( 1 . Quantum mechanically this corresponds to a state G 1+ 2.148 For the construction of YJM we will need the operators J± = J1 + iJ2 . 2 |Ω1 . 2 |Ω1 . Ω2 ). . 1+ 2 ( 1. Having determined such YJJ we then construct the whole family of functions XJ = {YJM ( 1 . . 2 |Ω1 .40) (Ω1 ) J+ Y (2) 2 2 J+ Y 1 1 (Ω1 ) Y (Ω2 ) + Y (Ω2 ) = 0 .64) J3 = = (1) + J3 (1) 1 (2) Y 1 1 (Ω1 ) Y 2 2 2 2 (Ω2 ) 1 1 J3 Y ( + 1 1 (Ω1 ) Y 1 1 (Ω2 ) + Y 2 2 (Ω1 ) J3 Y (2) 2 2 (Ω2 ) (6. 1+ 2. (6. .25) using (6.68) J+ = (1) + J+ (1) (2) Y 1 1 (Ω1 ) Y 2 2 2 2 (Ω2 ) 1 1 (6. We obtain using (5. We embark on the suggested construction for the choice J = 1 + 2 .11) and (5.66.

46) 2 1 +c Y (Ω1 ) Y 2 2 (Ω2 ) = 0 .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 ). The l. ). To demonstrate (6. ( + )}. We like to construct now the family of total angular momentum functions B + − = {Y + − M ( . 2 |Ω1 . We seek for this purpose ﬁrst an unnormalized solution G 1 + 2 −1 1 + 2 −1 of (6. . . The reader should familiarize himself with the entries of the Table. . ).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 ) (6.47) can serve now to obtain recursively the elements of the family B + − for M = 1 + 2 − 2.38) yields pression 2( 1 + 2 )Y 1 + 2 1 + 2 −1 ( 1 . One can obviously continue the construction outlined to determine Y 1 + 2 −2 1 + 2 −2 . of (6. .45).2: Construction of Clebsch-Gordan Coeﬃcients 149 We now employ property (6. −( 1 + 2 − 1).h. 1 + 2 − 1.44) (Ω2 ) .37). . According to the condition (6.39) as required by (6. | . | . Ω2 ) 2 1+ 2 = (Ω1 )Y 2 2 −1 (6. The result is illustrated for the case 1 = 2. Ω2) for M = 1 + 2 − 2 and J = 1 + 2 . 6. Exercise 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 ).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. in particular.37) is satisﬁed. all the Clebsch-Gordan coeﬃcients ( 1 + 2 M | 1 m1 2 m2 ). Normalization yields furthermore Y = 1 1 + 2 −1 1 + 2 −1 We have thereby determined ( 1 2 |Ω1 Ω2 ) (6. . of (6. . etc. . The r. thereby. .47) (Ω2 ) − 1 1 2 + Y 2 1 1 −1 (Ω1 )Y 2 2 + Y 2 1 1 (Ω1 )Y 2 2 −1 (Ω2 ) . 2 |Ω1 .2. G 1 + 2 −1 1 + 2 −1 in (6.1: Construct following the procedure above the three functions YJM ( 1 . . M = −( + ). . 1 + 2 − 3. .38) to construct the family of functions B + = {Y + M ( . To satisfy this equation one needs to choose c = − 1 / 2 . 2 = 1 in Table 1. We have provided in Table 1 the explicit form of YJ M ( 1 2 |Ω1 Ω2 ) for 1 = 2 and 2 = 1 to illustrate the construction. 1 + 2 − 2. Show that the resulting functions are orthonormal. with the symmetry pattern and with the terms Y 1 m1 Y 2 m2 contributing to each YJ M . . M = −( + − ). 2 |Ω1 . 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.36.27). One obtains then Y 1 1+ 2 1 + 2 1 + 2 −1 2 2 ( 1. One can readily show that (6.h.s. Expression (6.45) for some unknown constant c.6. Ω2 ). and all total angular momentum functions for a given choice of 1 and 2 . This expression is orthogonal to (6.s.

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 ) 0.730297 −0.547723 0.57735 0. 1|Ω1 . 1|Ω1 . 1|Ω1 .547723 0. Ω2) Y32 (2. Ω2) Y31 (2. 1|Ω1 . Ω2) Y1−1 (2.447214 0.408248 −0. Ω2) Y3−3 (2. Ω2) Y3−2 (2. Ω2) Y21 (2.547723 0. 1|Ω1 .316228 0.632456 0. Ω2) Y2−1 (2. Ω2) Y22 (2. Ω2) Y2−2 (2. 1|Ω1 .707107 0. Ω2) Y3−1 (2.57735 3 10 3 5 Y2−1 (Ω1 )Y11 (Ω2 ) Y20 (Ω1 )Y10 (Ω2 ) Y21 (Ω1 )Y1−1 (Ω2 ) −0.632456 −0. 1|Ω1 . Ω2) Y30 (2.1: Some explicit analytical and numerical values of Clebsch-Gordan coeﬃcients and their relationship to the total angular momentum wave functions and single particle angular momentum wave functions. 1|Ω1 .316228 0.632456 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) Y20 (2.57735 −0.707107 0. 1|Ω1 . 1|Ω1 .816497 −0.816497 0.408248 −0.57735 0.774597 − − − − 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 6.57735 0.258199 0.774597 0.258199 −0.57735 0. 1|Ω1 . . 1|Ω1 . 1|Ω1 . Ω2) Y11 (2.707107 0.707107 0.447214 −0.150 Addition of Angular Momenta and Spin Y33 (2.547723 0. 1|Ω1 . 1|Ω1 .774597 0 2 5 8 15 1 6 3 10 1 3 0.730297 0. Ω2) Y10 (2.

5. ζ = ± and the matrix elements < ζ |σk | ζ > are as deﬁned in Section 5. aζ .6.52) and | 2 m2 2 According to Section 5. a† and bζ . hence. 6. aζ bζ . m − 2 .ζ a† < ζ |σk | ζ > aζ ζ b† < ζ |σk | ζ > bζ ζ (6.10.9. Accordingly.9. 2 . M | . It is actually possible to state explicit expressions for any single coeﬃcient (JM | 1 m1 2 m2 ).54) 2 + = √ b† ( 2 +m2 )! √− ( 2 −m2 )! .3 Explicit Expression for the Clebsch–Gordan Coeﬃcients We want to establish in this Section an explicit expression for the Clebsch–Gordan coeﬃcients (JM | 1 m1 2 m2 ). we extent deﬁnition (5.51) The operators aζ . b† ζ = 0.49) where ζ. bζ = a† . a† ζ ζ b† . 1 (6. − 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.50) (6.10 [cf.10. Deﬁnition of Spinor Operators for Two Particles In contrast to Sections 5. For this purpsose we will employ the spinor operators introduced in Sections 5. bζ = = a† . 1 . bζ . b† ζ ζ = 0 = 0 . we are dealing now with two particles carrying angular momentum or spin. b† refer to diﬀerent particles and. b † ζ = δζζ = δζζ .3: Explicit Expressions of CG–Coeﬃcients 151 Exercise 6. a† ζ . 5. b† ζ ζ = aζ . (6. The creation and annihilation operators are again of the boson type with commutation properties aζ . m + 1 .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 . These expressions will be derived now. (5.2.287) to two particles (1) J k k = = (2) J 1 2 1 2 ζ. + 2 − J−M +1 for = J + 1 2J+2 2 J−M for = J − 1 2J 2 = .48) (6.53) (6.10 where we studied single particle angular momentum and spin.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. J. commute with each other ζ ζ aζ .ζ ζ. M | . (6.

+ 1) |J. (1) J 2 . M.67) = j |J.e. (6. (1) J 2 | (2) J 2 | 1 m1 1 2 m2 2 Addition of Angular Momentum and Spin = = 1( 1 2( 2 + 1) | 1 m1 + 1) | 2 m2 1 2 . (2) J 2 .61) (6. 6. (2) J which. M ( 2) 2) = J(J + 1) |J. 2) 1. −J ≤ M ≤ J . m1 1 | 2 . J3 |J. (5. = M |J. 3 .60) (6. (1) J (2) J 3 | 1 m1 1 3 | 2 m2 2 = m1 | 1 m1 = m2 | 2 m2 1 2 (6. (6. = = 1( 1 2( 2 2) 2) (6. (1) J 2 |J. k = 1. |J. we like to recall for future reference that the operators (5. (6. m2 |J. j = 1. 2) 2) (6. M ( 1 . 2 )). for such states should hold J2 |J. are | 1 . (1) J .18) as follows = | 1 m1 . M ( 1 . 1. (6. in analogy to Eqs.68) The states |J. 2 .65) (6. 2 ) which are simultaneous eigenstates of the operators 2 2 J2 . M ( 1 .56) The states | 1 . m1 1 | 2 . 2) We will also require below the raising and lowering operators associated with the total angular momentum operator (6. ˆ kj |J.54). i.59) J± = J1 ± i J2 . mj j and. 2) 2) can be expressed in terms of Clebsch-Gordan coeﬃcients (6. hence. m1 . (6. The operator of the total angular momentum/spin of the two particle system is J = with Cartesian components Jk = (1) J + J (6. 1 . (j) 2 ˆ ˆ J = kj ( kj + 1 ) . M ( 1 . M ( 1 . 2 . m1 . 2) + 1) |J.63) according to At this point. m2 |J. M ( 1 . m2 a† + ( 1 2 = a† − ( 1 1 −m1 (6.69) − 2| ≤ J ≤ + 2 . m1 1 | 2 .m2 | 1 . J3 . 5.64) (6. M ( 1 .53. as usual are denoted by their respective quantum numbers J. mj j = j | j . The aim of the present Section is to determine closed expressions for the Clebsch–Gordan coeﬃcents ( 1 .57) b† + ( (2) 2 2 +m2 1 +m1 b† − ( 2 2 −m2 + m1 )! − m1 )! (1) + m2 )! − m2 )! |Ψo . m2 2 . 2 . 2) . M ( 1 .58) Jk + (2) Jk . k2 = b b+ + b† b− k1 = + − − 2 2 + namely.. . M ( 1 . 2 )) . ˆ For the operators kj holds ˆ kj | j .55) .302.300). 2 .152 It holds. . M ( 1 . M ( 1 . which describe a two particle system according to (6.66) (6. can be expressed in terms of the number operators 1 † 1 ˆ ˆ a† a+ + a† a− . M ( 1 . m2 1 2 ( 1 . M ( 1 .303).62) .59) We seek to determine states |J. M ( (2) J 2 |J. 2.

j = 1.74) imply 3 2 J2 . a† b† − a† b† − + − − + 2 + 1 1 = a† a+ . using (6.65) and (6. For example.74) + 1 J− J+ + J2 . 4 Using J3 = (1) J 3 + (2) J 3 .6. a† b† + − − + 2 + 2 − 1 1 − b† a† b+ . the relationships (6. K † = 0. ˆ A similar calculation yields [k2 .73) can be readily proven. and applying the relationships (6.288).75) The relationships (6. expressing (k) J 3 through the creation and annihilation operators according to (5. K † 1 † a a+ + a† a− . a† b† + + − − + 2 2 − 1 1 = a† b† − a† b† = K† . K † 1 a† a+ − a† a− + b† b+ − b† b− . K † (j) J 2 . 6. (6. + − − + 153 (6. K † + K † 2 2 3 † †ˆ K kj + K .73) yields J3 .71–6.64. (6. a† b† + a† a− . due to J2 = K† 1 2 J+ J− 1 † K . a† b† − a† b† + − + − + − − + 2 1 1 = a† a+ . K † (kj + 1) 2 2 1ˆ † 1 ˆ kj K + K † (kj + 1) 2 2 1 ˆ 1 ˆ K † kj + kj . 2 4 = 0 = = 0.71) one can show K† = = = = = 1 ˆ ˆ kj (kj + 1). K † + kj . 6. We note that.73) (6. a† b† − a† a− .70) will play a crucial role in the evaluation of the Clebsch-Gordan-Coeﬃcients.71–6. b† = 0 . + − + − 2 2 − + = .71) (6.50. K † ] = (j) J 2 .72) (6. K† J3 .3: Explicit Expressions of CG–Coeﬃcients The Operator K † The following operator K † = a† b† − a† b† . j = 1.73. K † 2 1ˆ ˆ 1 ˆ ˆ kj kj + 1. 6. Employing (6. b† − b† a† b− . K † J± . + − − + 2 2 = 1 † 2K . This operator obeys the following commutation relationships with the other pertinent angular momentum / spin operators ˆ kj . 2 2 3 ˆ = K † kj + K † .51) one obtains ˆ k1 .

2 ) j + 4 1 3 + K † |J.. M ( 1 . 4 3 + j ( j + 1) |J. 2 ) produces again total angular momentum eigenstates to the same J and M quantum numbers of J2 and J3 . a† b− − a† b† + + + − + = 0.76) and state K† n |J. j2 + ) 2 2 n n = N (n. M ( 1 .79) n |j1 + j2 . To demonstrate that the resulting states are eigenstates of (1) J 2 and (2) J 2 we exploit (6. on the state |j1 + j2 . according to (6. j1 1 |j2 .62. j2 2 . The commutation properties (6. 2) is a state with quantum numbers 1 1 + 1 2 and = N |J.67) (j) J 2 K † |J. 2) = N |J. 6. j1 + j2 (j1 . 1 2 + 2 . (6. M ( 1 . it holds K † |J. M ( 1 + . i.76) Here N is an unknown normalization constant.73. j1 + ) K † 2 2 (6.77). M ( 1 . j2 ) . 6.78) a state which has been used already in the construction of Clebsch-Gordan coeﬃcients in Section 6. One can generalize property (6. M ( 1 . 2) 2) ˆ K † kj + = K† = j 3 † K + K † j ( j + 1) |J. M ( 1 . j1 + j2 (j1 + n n . Action of K † on the states |J.154 Starting from (5. = − a† a− . 2 (6. j1 1 + . We consider now the case that (K † )n acts on the simplest total angular momentum / spin state.63. 2) = = (j) 2 J . |j1 + j2 .292) one can derive similarly J+ . j1 + j2 (j1 .72) and (6.77) where N is another normalization constant. 2 2 + n ) 2 (6. a† b† + b† a† b− . j + 2 2 2) However. M ( 1 . M ( 1 .2. K† + K † (j) J 2 |J. 2 ) . 2 ) remain eigenstates of J2 and J3 with the same quantum numbers. namely. M ( 1 . but for diﬀerent 1 and 2 quantum numbers of the operators (1) J 2 and (2) J 2 . j2 ) = |j1 . K † ] = 0 is demonstrated in an analoguous way. M ( 1. K † = Addition of Angular Momentum and Spin a† a− + b† b− . this result implies that K † |J. 2) We want to demonstrate now that the action of K † on the states |J. M ( 1 .e. 2 2 1 + ) . 6. b† + − + + + − The property [J− . Application of (K † )n to this state yields.75) ascertain that under the action of K † the states |J. M ( 1 + n .

j1 . 1 = j1 + n 2 . 2) 1 2 = ( 1 + 2 (2J + 1)! − J)! ( 1 + 2 + J + 1)! 1 2 . K† 1 2 1 + 2 −J × × . J( 1 . j2 as follows J = j1 + j2 which is equivalent to n j1 j2 Accodingly. holds |J. 1 .86) b† + J+ 2 − 1 |Ψo . 2 ) can be expressed through the r. of (6. 2) Our construction of the states |J. (J + 2 − − 2) 1 1) 2 The normalization constant appearing here is actually N( + − J.82.s. The latter states. j2 + n ).84) (6. M ( 1 .79). 2) 2) .82) for |J. − 1.84) with (6.57) and exploiting the fact that J− and K † commute [c. M ( 1 .74)] yields |J. J( 1 .104.80) = = = 1 1 2 2 − J 1 n − = (J + 2 2 n 1 − = (J + 2 2 + 1 1 − − 2) 2) .6. j1 + n . M ( 1 . 2 ) ∆(J. 5.85) ∆(J. M ( 1 . M ) . (6.h.83) The derivation of this expression will be provided further below (see page 158 ﬀ). J( 1 . Combining (6. 1. 2 = j2 + n 2 (6. 2 ) . M ( 1 . (6. 2) . 2 ) for −J ≤ M ≤ J as follows |J.105) of the spherical harmonics. 6. allow one to obtain the states |J. Strategy for Generating the States |J. (6. J( 1 .3: Explicit Expressions of CG–Coeﬃcients 155 where we denoted the associated normalization constant by N (n.81) = N( 1 + 2 1 2 − J.79) n. (6. in analogy to the construction (5. 2 2 It is now important to notice that any state of the type |J. = (J + M )! (2J)! (J − M )! 1 2 2) (6.82) 1 × | (J + 2 1 × | (J + 2 1 (J + 2 1 − 1 ) . 2) = N( 1 + 2 − J. M ) (J + × K† 1 − 2 )! (J + 2 − 1 )! J+ 1 − 1 + 2 −J (J− )J−M a† + × 2 (6. For this purpose one needs to choose in (6. 2) = ∆(J. 2 ) exploits the expression (6. M ) (J− )J−M |J.f.

For this purpose we introduce the generating function I(λ. b† ] = 0 [cf.. x.89) i.57)].e.264).. y) = exp (λJ− ) exp x a† exp x b† |Ψo . m1 1 | 2 . I(λ.e. (6.88). (6.50)] − − exp ( λJ− ) f (a† ) g(b† ) |Ψo + + = exp a† a+ − = u f (a† ) exp b† b+ + u g(b† ) |Ψo + λu u! a† a+ − λv v! u f (a† ) × + v × v b† b+ − ∂ g(b† ) |Ψo + u = u λ a† − u! × v λ b† − v! ∂a† + v ∂ ∂b† + f (a† ) × + v g(b† ) |Ψo + (6. in analogy to (5.86) |Gr = Jr a† st − + s b† + t |Ψo (6. + − + − . x. One obtains then using J− = a† a+ + b† b+ (6. Comparision with (6.88) in terms of monomials (6.91) which.50–6.87). a† ] = [b+ .93) = f (a† + λ a† ) g(b† + λ b† ) |Ψo .s.87) + − + − i. x.t (6.52). + + Taylor expansion of the two exponential operators yields immediately I(λ. y) is a generating function for the states |Gr deﬁned in (6. st The desired expansion of |Gr can be obtained from an alternate evaluation of I(λ. y) which is st based on the properties aζ f (a† ) |Ψo = ζ ∂ ∂a† ζ f (a† ) |Ψo ζ bζ f (b† ) |Ψo = ζ ∂ ∂b† ζ f (b† ) |Ψo ζ (6. 2 [cf. m2 Gordan-coeﬃcients. follows from the commutation properties (6.90) (6.156 Addition of Angular Momentum and Spin Our strategy for the evaluation of the Clebsch-Gordan-coeﬃcients is to expand (6. x.86) in terms of monomials 1 +m1 1 −m1 2 +m2 2 −m2 a† a† b† b† |Ψo . in terms of | 1 .92) − − and noting [a+ .69) yields then the Clebsch- Expansion of an Intermediate State We ﬁrst consider the expansion of the following factor appearing in (6. (6. r!s!t! r. y) = λr xs y t r |Gst .

s.98) a† − s a† + b† + 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. 157 (6.3: Explicit Expressions of CG–Coeﬃcients We conclude I(λ.70) holds K† 1 + 2 −J = s 1 + 2 −J b† − s 1 + 2 −J−s (−1)s 1 + 2 −J−s (6.50). 2) = s.97).94) One can infer from this result the desired expressions for |Gr . M ( 1 .96) and.97) yields.6.88). to obtain the desired expansion of |J.90) allows one to infer |Gr s. M ( 1 .t × a† + s−q s q b† + t r−q t−r+q a† − q b† − |Ψo (6. one can write the right factor in (6. y) = s. using the commutation property (6. x.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. With K † given by (6.99) . |J.t = q r! × a† + s q s−q t r−q a† − q × b† + t−r+q b† − r−q |Ψo (6. y) = exp a† + λ a† + − exp b† + λ b† + − |Ψo .95) Comparision with (6. Expanding the exponentials in st (6.q (−1)s (6. Operation of this operator on (6.94) yields I(λ. x. 2 ) in terms of states | 1 .86) (J− )J−M a† + = q J+ 1 − 2 b† + J+ 2 − 1 |Ψo 1 (6. using the deﬁnition (6.t × λ v b† − × r−q v × = t s−u v λu a† − r! q |Ψo λr xs y t r! s! t! r. m1 1 | 2 . m2 2 .

69) if one identiﬁes m1 = 1 −q − s . − q − s)!(q + s)! 2 × (M + 1 − 1 + q + s)!( 1 1 + 2 − M − q − s)! 2 − q − s 1 | 2. q = 1 − m1 − s and m2 = M − m1 . according to (6. The summation over q corresponds then to the summation over m1 . × 2) = N( 1 + 2 1 − J. j2 .53.103) = 1.82). j2 = 1 (J + 2 2 − 1) . M − +q+s One can conclude that this expression reproduces (6.q (−1)s × (6. (6. To determine N = N ( 1 + 2 −J. Using (6. M ) (J + 2 )! (J + 2 − 1 )! s. m1 . M ) 1 = + J)!(− 1 + 2 + J + 1)! 2 2 + 2 − J)!( 1 − ( 1+ + J)! 1 2 × [( × s + m1 )!( − m1 )!( 2 + m2 )!( 2 − m2 )!(J + M )!(J − M )!] 2 (−1)s s!( 1 − m1 − s)!( 2 + m2 − s)! 1 2 1 × The Normalization 1 ( 1 + 2 − J − s)!(J − 1 − m2 + s)!(J − + m1 + s)! (6. 1 .54) between creation operator monomials and angular momentum states allow one to write this |J. J( 1 . 1 .158 Addition of Angular Momentum and Spin ( 1 + 2 − J)!(J − M )!(J + 1 − 2 )!(J + 2 − 1 )! s!( 1 + 2 − J − s)!q!(J + 1 − 2 − q)!(J − M − q)!(M + 2 − a† + 2 1 −q−s 1 + q)! a† − q+s b† + M + 2 − 1 +q+s b† − 1 + 2 −M −q−s |Ψo The relationships (6.83) of the normalization constant N ( 1 + deﬁned through (6. For this purpose we introduce j1 = 1 (J + 2 1 2 − J. M ( 1 . m2 in (6. The Clebsch-Gordan coeﬃcents are then ﬁnally ( 1 . (6. 2 ) we consider the scalar product J. n)|ψ(j1 .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 .101) Note that m1 + m2 = M holds.103) this can be written 1 = N 2 ψ(j1 . n = 1 + 2 −J. − 1 .69) since. 2 ) (6. J( 1 . 2 ) − 2) .101). m2 = M − 1 + q + s. j2 . n) 2 )|J.102) We want to determine now the expression (6.6. √ 2J + 1 1 2 . 2 ) ∆(J. m2 |J.82) and (6.104) .

108) = λ=0 (n1 + ν)! n1 ! (6.57) for these states and the expression (6.108) yields the identity n1 + r − s n1 n2 + s n2 = n1 + n2 + r + 1 n1 + n2 + 1 .110) which can be written 1 1−λ n1 +n2 +2 = r s n1 + r − s n1 n2 + s n2 λr (6.105) The ﬁrst step of our calculation is the expansion of ψ(j1 .109) and Taylor expansion of the left hand side of (6. j1 1 |j2 .111) Comparision with (6. We employ the expression (6. applying (6.6. One obtains then.m2 n1 + m1 n1 n2 + m2 n2 λm1 +m2 (6.3: Explicit Expressions of CG–Coeﬃcients where |ψ(j1 .107) s = (n!)2 s 2j1 + n − s 2j1 The latter sum can be evaluated using 1 1−λ a property which follows from ∂ν ∂λν 1 1−λ n1 +1 n1 +1 = m1 n1 + m1 n1 λm1 (6. (6.108) twice. 1 1−λ n1 +1 1 1−λ n2 +1 = m1 . n) in terms of states |j1 . (6.70) for the operator K † .106) The orthonormality of the states occurring in the last expression allows one to write (6. j2 . Accordingly.104) 1 = N2 (n!)2 (2j1 )!(2j2 )! (2j1 + n − s)!(2j2 + s)! s!(n − s)! 2j2 + s 2j2 (6. n) = K† n 159 |j1 .108). m2 2 . j2 2 . we obtain |ψ(j1 . j2 . n) = 1 (2j1 )!(2j2 )! a† + n! (2j1 )!(2j2 )! a† + 2j1 −n−s s 2j1 s n s 2j2 a† b† + − |Ψo = n−s (−1)s a† b† − + s b† + (−1)s a† − s (2j1 + n − s)!s!(2j2 + s)!(n − s)! s!(n − s)! 2j2 +s b† + b† − n−s (2j1 + n − s)!s!(2j2 + s)!(n − s)! |Ψo .112) s . (6. j2 . m1 1 |j2 .

m2 . 6. namely.113) Using the identities (6.s.114) one obtains − J = − J (1) − J (2) The respective Clebsch-Gordan coeﬃcients ( 1 . J (1) .114) as long as J. 2 .103) one obtains the desired result (6.118). 1 .g.116) If one takes the negatives of the operators in (6. −m1 .160 Applying this to (6. manner to the coeﬃcients ( 1 . m2 |J. vice versa. m2 |J. m1 .119) The corresponding symmetry property of the Clebsch-Gordan coeﬃcients is ( 1 .h.114) by. and J (2) are vectors in R . m2 |J. 1 . M ) = (−1) 1 + 2 −J ( 2 . ( 2 . (2j1 + 2j2 + 1)! (6. of this equation J (1) = J (2) − J . 2 1+1 (6.114) show a simple relationship to the Clebsch Gordan coeﬃcients ( 2 . We will demonstrate this now. −m2 |J.120) can be readily derived from the expression (6.102) by replacing ( 1 .117) are again related in a simple 2 . ( 1 .115) This relationship is a trivial consequence of (6. (6. M | 1 .102) of the Clebsch-Gordan coeﬃcients. 2 .4 Symmetries of the Clebsch-Gordan Coeﬃcients The Clebsch-Gordan coeﬃcients obey symmetry properties which reﬂect geometrical aspects of the operator relationship (6.83). M ) = (−1) 2 +m2 2J + 1 ( 2 . 2 . (6.116). m2 |J. (6. −M ) = (−1) 1 + 2 −J ( 1 . For the quantum mechanical addition of angular momenta the Clebsch Gordan coeﬃcients ( 1 . and (6.116) one expresses the Clebsch-Gordan coeﬃcient on the r. (6.h. m2 . interchanging the operators J (1) and J (2) results in J = J (2) + J (1) . −m1 2|J. and . M ) corresponding to (6. m1 ) . (6. m2 ) by ( 1 . 2 . M ) corresponding to (6.120) The symmetry properties (6. m1 . 2 .s. m1 . M ) ( 1 . (6. m1 . (6. M ) . of (6. e. of (6. m1 . −M ) . 2 . m1 ) by ( 2 .116) through formula (6. −m2 . −m1 . M ) . m2 ) and..s. one can interchange also the operator J on the l. m1 |J. m1 ). To derive relationship (6.11) J = J (1) + J (2) .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 .118) Finally. J (1) on the r.115). J. m1 |J.114) For example. m2 |J.

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

180) and.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. ˆ (6. Since f (r) is independent of θ and φ follows ˆ ((σ · p f (r))) = −i (( ∂r f (r) )) σ · r . therefore.182) 0 Since σ · r. 1 |ˆ) = ((σ · p f (r)))Yjm (j ± 1 . We want to demonstrate ﬁnally that the “−”-sign holds in (6. identiﬁed the sign of (6.182) σ · r Yj 1 (j − 1 . ˆ 2 2 2 2 2 2 for θ = 0. have proven (6. φ) .160).168 Addition of Angular Momentum and Spin 1 Yjm (j ± 2 . The result can also be stated in the compact form σ · r Yjm (j ± 1 . (5. 1 |θ.180) where the sign could depend on j. 6.f. φ) 2 2 2 = − j+ 1 2 4π (6.180).183) in case θ = 0 becomes in the standard representation with respect to the spin. 2 |ˆ) and Yj 1 (j + 1 .180) and (6.186) Here (( · · · )) denotes again conﬁnement of the diﬀusion operator ∂r to within the double bracket. (6. 1 |θ = 0. 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 . We want to determine ˆ its action on the wave functions f (r) Yjm (j ± 1 . (6. 1 |ˆ) r r r 2 2 2 2 2 2 (6.174–5.188) .224)] 1 0 σ·r = ˆ . hence.181) 0 = j+ 1 2 4π Yj 1 (j + 1 .187) 1 2 .184) 0 1 space χ 1±1 2 2 one can conclude from (6.1 states χ 1 ± 1 2 2 2 [c. 1 |ˆ).147. For this purpose we consider the application of σ · r in the case of θ = 0. 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. 1 |ˆ) + f (r) σ · p Yjm (j ± 1 . φ) and (5. given by ˆ σ · r = σ1 sin θ cos φ + σ2 sin θ sin φ + σ3 cos θ . 1 |θ = 0. φ) = − Yj 1 (j + 1 . ˆ (6. 1 |θ = 0. 1 |θ = 0. φ) 2 2 2 . According to (5. 6.148) the particular ˆ 1 1 states Yj 1 (j − 2 . 1 |ˆ) at θ = 0 are r r 2 2 2 2 Yj 1 (j − 1 .181.185) We have. One can conclude then ˆ σ·r = ± ˆ 0 1 1 0 (6. 1 |ˆ) r 2 (6. for θ = 0 (6.

192) ˆ ˆ The operator p × r in (6.189) The celebrated property of the Pauli matrices (5. h r (6.196) = i j+ ∂r + r 1 2 3 2 1 f (r) Yjm (j − 2 . we conclude the intuitively expected identity 1 ˆ ˆ p×r = − J .187) one obtains ˆ σ · p f (r) Yjm (j ± 1 . r Altogether we obtain. Corresponding results are obtained for the other components of p × r and. hence. 1 |ˆ) = i [ ∂r + r + ] f (r) Yjm (j 2 r r Using (6.193) r r 1 1 1 = −i (∂2 x3 − ∂3 x2 ) h − i h (x3 ∂2 − x2 ∂3 ) . 1 |ˆ) r 2 (6.190) can be related to the angular momentum operator.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) .156) this yields ﬁnally 1 ˆ σ · p f (r) Yjm (j + 2 . r 2 2 1 J ·S 2 1 ˆ σ · p f (r) Yjm (j ± 2 . 1 |ˆ) r 2 (6. using ∂r Yjm (j ± 1 .197) 1 ˆ σ · p g(r) Yjm (j − 2 . 1 |ˆ) = 0 and σ = 2S/ .. e. 1 |ˆ) r 2 = i ∂r + −j r 1 g(r) Yjm (j + 2 .195) . 6.155. 1 |ˆ) r 2 (6. 1 |ˆ) . ∂3 x2 = x2 ∂3 we obtain ˆ ˆ ( p × r )1 h = −i 1 1 (x3 ∂2 − x2 ∂3 ) h = − J1 h r r (6.6. r r r Using ∂2 (1/r) = −x2 /r3 .186) for both terms in (6.g.198) .191) (6. ˆ For the test function h(r) holds ˆ ˆ p·rh = = Using (1/r) = −r/r3 and r · ˆ −i −i · r h r = −i h r ·r − i r· h. To demonstrate this we consider one of its cartision components.5: Spin–Orbital Angular Momentum States Using (6. x3 x2 ˆ ˆ ( p × r )1 h = −i (∂2 − ∂3 ) h (6. r 2 (6.190) 3 h − i hr · r 1 − i r· ˆ r h = ∂r h one can conclude ˆ ˆ p · r h = −i 2 + ∂r r h (6.53). ∂3 (1/r) = −x3 /r3 and ∂2 x3 = x3 ∂2 . 1 |ˆ) r 2 1 2 (6. 1 |ˆ) = r 2 2 ˆ i ∂r f (r) − f (r) σ · p σ · r Yjm (j ˆ 1 2 169 .

1 |ˆ) r 2 2 r r2 and. in fact. 1 |ˆ) = Yj− 1 j− 1 (ˆ) χ 1 1 . 1 |ˆ) r 2 2 The Clebsch-Gordan coeﬃcients are then (j − 1 .101). j) 1 2 1 2 1 2 1 2 = = 1 0 (6. We note ˆ (σ · p)2 f (r) Yjm (j + 1 .151). − |j. 6.170 Addition of Angular Momentum and Spin To demonstrate the validity of this key result we note that according to (5. j) 2 2 2 2 (j − . according to (6.230. are consistent with this identity. 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 . consider ﬁrst the case of the largest m-value m = j.199) We want to show that eqs. 1 |j. 5. (6. We begin with the coeﬃcients (6. In this case holds. j + .2.144) for m = j.143–6.204) (6. 6.144) and. j 2 + 2j + 2 3 4 (6. r ∂r2 r 2 (6. adopting the method in Sec. . Evaluation of Relevant Clebsch-Gordan Coeﬃcients We want to determine now the Clebsch-Gordan coeﬃcients (6. 6.206) . 1 .205) which agrees with the expressions (6. r 2 2 2 2 2 2 − r 2 ∂r r + J2 r2 f (r) Yjm (j + 1 . i. For m = j − 1 one can state.e.143.45). r (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 . as well as (6.202) Yjj (j − 1 . 2 2 (6. according to (6.146). 6.203) (6.99) holds ˆ (σ · p)2 = − 2 2 = ∂2 J2 r + 2 . For this purpose we use the construction method introduced in Sec.43).200) = (j + 1 )(j + 3 ). j − 1 .143. Yjj−1 (j − 1 .2. 1 |ˆ) r r 2 2 2 2 2 2 2 yields ˆ (σ · p)2 f (r) Yjm (j + 1 . 1 |ˆ) r 2 2 r r2 r2 j 2 + 2j + 3 2 2 2 4 = [ −∂r − ∂r + ] f (r) Yjm (j + 1 . 1 |ˆ) = r 2 2 which agrees with (6. using (5. 1 |ˆ) = J 2 Yjm (j + 1 .. 6.197. 1 |ˆ) = r 2 2 2j − 1 Yj− 1 j− 3 χ 1 1 + 2 2 2 2 2j 1 Y 1 1 χ1 1 2j j− 2 j− 2 2 − 2 (6. 1 |ˆ) r 2 2 i 3 ˆ = σ · p [ i∂r + (j + ) ] f (r) Yjm (j − 1 .199).198).201) 1 3 (j + )(j + ) Yjm (j + 1 .

2 |j. j − 3 .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 .6.h. (· · · |jj − 3). and J− + S− to the r.210)] yields (j + m)(j − m + 1) Yjm−1 (j − 1 . 1 |ˆ) = r 2 2 Applying J− (tot) j+m Yj− 1 m− 1 χ 1 1 + 2 2 2 2 2j j−m Yj− 1 m+ 1 χ 1 − 1 . Expression (6. as described in Sec.h. 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 . 1 . 6. j − 1 .144) for the Clebsch-Gordan coeﬃcients by induction. etc. j − 1) 2 2 2 2 which again agrees with the expressions (6. Let us verify then the expression (6. 6.207) (6. 6.143.208) (6. 2 2 2 2 2j (6.210). j − 1) 2 2 2 171 = = 2j − 1 2j 1 2j (6.f. The further Clebsch-Gordan coeﬃcients (· · · |jj − 2).211) to the l.137)] J− (tot) = J− + S− (6. − 1 |j. m − 1) 2 2 2 2 = j +m−1 2j (6. are obtained by iterating the application of (6. [c.209) (j − 1 .s.f.5: Spin–Orbital Angular Momentum States The corresponding Clebsch-Gordan coeﬃcients are then 1 (j − 1 .144) implies for j = m Yjm (j − 1 . (6. 1 . (6.2.206). m − 3 . 1 |j.143.210) to (6. 6.144) for m = j − 1.202).143. 1 . is obtained by applying the operator [c. 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 .s. 2 2 2 2 2j . (6..

−m2 | 1 . 2j + 2 (6. 6. In this context Jclass and Jclass play the same role. m1 . m2 . hence. 1 |j + 1 . − 1 |j. by the set j1 .143). to reﬂect in the notation the symmetry of these quantities. m + 1 . (1) (2) (−tot) Obviously. m. M.214) yields (j. the relationship (6. m + 1 ) = 2 2 2 2 and. m3 . The latter relationships applied together read ( 1 . 1+1 3 For 1 (j. 1 |j + 1 .143. j2 m2 ) 2 . m1 ) (6. 2 . one can obtain expression (6. m + 1 ) = 2 2 2 j+m+1 . j2 .6 The 3j–Coeﬃcients The Clebsch-Gordan coeﬃcients describe the quantum mechanical equivalent of the addition of two (1) (2) classical angular momentum vectors Jclass and Jclass to obtain the total angular momentum vector (tot) (1) (2) (1) (2) Jclass = Jclass + Jclass . The Clebsch-Gordan coeﬃcients (6. m1 . Jclass and Jclass play equivalent roles.146) can be obtained from (6. 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 .217) 2j + 2 (j + 1 . m. 2j + 1 (6.145) from (6. m2 | 3 . j3 . m − 1) 2 2 2 2 = (6. 1 . (1) (2) (−tot) The coeﬃcients which are the quantum mechanical equivalent to Jclass + Jclass + Jclass = 0 are the 3j–coeﬃcients introduced by Wigner.143) by applying the symmetry relationships (6. 1 . m2 1 (6.214) 2 2 +1 ( 3 . − 1 |j. However.172 Addition of Angular Momentum and Spin j −m+1 2j (j − 1 . 1 .144).213) which is in agreement with (6.144) by induction. 6.116.215). m3 . We have.216) 6. − 1 |j. m − 1 . 1 . j+m+1 . proven (6. m1 .144) for j = m − 1. m3 ) = (−1) × 2 + 3 − 1 + 2 +m2 × 2 .143. m + 1 . one obtains (j + 1 . m) 2 2 2 2 2j + 1 (6. 2 . all three vectors Jclass .215) which follows from (6. using (6. 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. 6. m) = 2 2 2 2 Similarly.218) where we have replaced the quantum numbers J. leading quantum mechanically to a symmetry of the Clebsch-Gordan coeﬃcients (JM | 1 m1 2 m2 ) with respect to exchange of 1 m1 and 2 m2 .120). 1 .

6. j3 form the sides of a triangle and the condition is symmetric in the three quantum numbers. √ safe for the prefactor 2j3 + 1 which is cancelled according to the deﬁnition (6. discovered by Regge.218) relating 3jcoeﬃcients and Clebsch-Gordan coeﬃcients.. According to the deﬁnition of the 3j–coeﬃcients one would expect symmetry properties with respect to exchange of j1 . the so-called triangle condition. except when all elements are non-negative integers and each row and column has the same integer value Σ = j1 + j2 + j3 . j2 . m3 and with respect to sign reversals of all three values m1 .219) (6.34) 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 .g.220) The latter condition |j1 − j2 | ≤ j3 ≤ j1 + j2 .6. These symmetry operations relate altogether 72 3j–coeﬃcients. 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 Regge symbol also vanishes in case that two rows or columns are identical and Σ is an odd integer.6: The 3j–Coeﬃcients We ﬁrst like to point out that conditions (6.e. e. However. are identical to the integer arguments which enter the analytical expression (6. For this purpose one can use the symmetry transformations to place the smallest element into the upper left corner of the Regge symbol. the Regge-symbol. = (6. as follows −j1 + j2 + j3 j1 − j2 + j3 j1 + j2 − j3 j1 j2 j3 j1 − m1 j2 − m2 j3 − m3 . i.222) m1 m2 m3 j1 + m1 j2 + m2 j3 + m3 The Regge symbol vanishes. states that j1 . 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 (6.102) of the Clebsch-Gordan coeﬃcients. m2 . The Regge symbol reﬂects a remarkable degree of symmetry of the related 3j–coeﬃcients: One can exchange rows. there exist even further symmetry properties. m3 . 173 (6.221) j1 j2 j3 −m1 −m2 −m3 where the the results of a cyclic. with respect to alltogether 12 symmetry operations. m1 .21. In this way the values of 12 3j-coeﬃcients are related. m2 and j3 . one can exchange columns and one can reﬂect at the diagonal (transposition). for which no known classical analogue exists. j2 . anti-cyclic exchange and of a sign reversal are stated. In case of a non-cyclic exchange of rows and columns the 3j–coeﬃcent assumes a prefactor (−1)Σ . To represent the full symmetry one expresses the 3j–coeﬃcients through a 3 × 3–matrix. The reader may note that the entries of the Regge-symbol.223) (Σ + 1)!n11 !n22 !n33 !n23 !n32 !)) n=0 n31 n32 n33 .102) are obtained each through the diﬀerence of two entries of the Regge-symbol. −j1 + j2 + j3 . The two integer entries J − 1 − m2 and J − 2 + m1 in (6. Because of its high degree of symmetry the Regge symbol is very suited for numerical evaluations of the 3j–coeﬃcents.

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. m) δ(−m1 .1.226) We like to state ﬁnally a few explicit analytical expressions for Clebsch-Gordan coeﬃcients which were actually obtained using (6.229) 1 1 (0m| m1 m2 ) = 2 2 i(−1)1−m2 δ(0.225) sn−1 .232) . m1 + m2 ) (3 − m)! (3 + m)! (3m|2m1 1m2 ) = √ 105 (2 − m1 )! (2 + m1 )! (1 − m2 )! (1 + m2 )! 1 ≥ |m2 |) ∧ 2 ≥ |m1 | ∧ 3 ≥ |m| (6.230) 1 1 (1m| m1 m2 ) = 2 2 √ (−1)2m1 −2m2 3 δ(m. m1 + m2 ) (2 − m1 )! (2 + m1 )! √ 10 (1 − m)! (1 + m)! (1 − m2 )! (1 + m2 )! 1 ≥ |m| ∧ 1 ≥ |m2 | ∧ 2 ≥ |m1 | (6.223-6.10752786393409395427444450130056540562826159542886 We also illustrate the numerical values of a sequence of 3j-coeﬃcients in Figure 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. (6.228) √ (−1)2m1 −2m2 7 δ(m. m1 + m2 ) (2 − m1 )! (2 + m1 )! √ 6 (2 − m)! (2 + m)! (1 − m2 )! (1 + m2 )! 1 ≥ |m2 | ∧ 2 ≥ |m| ∧ 2 ≥ |m1 | (6.227) (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) (6.224) (6. m2 ) √ 2 1 ≥ |m1 |) 2 (6.226) by means of a symbolic manipulation package (Mathematica) (1m|2m1 1m2 ) = (−1)1+m+m1 δ(m. (6.

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

6.1: Prove Eqs. Exercise 6.234) . . . M = −J. 2 |Ω1 . Show that the probability of detecting a triplet substate |1m1 (1) for arbitrary polarization (m2 – 1 value) of the other triplet is 3 . (6. . . J} (2 (2 1 2 + 1) (2 + 1) × (2 1 2 + 1) + 1) = (2| 1 − 2 | + 1) ×(2| 1 − 2 | + 1) (2| 1 − 2 | + 3) ×(2| 1 − 2 | + 3) (6.176 Addition of Angular Momentum and Spin 1 For the basis {YJM ( 1 . 1). . Exercise 6. .6..6. J = | 1 − 2 |.233) is further block-diagonalized as follows + 2.3: Two triplet states |1m1 (1) |1m2 (2) are coupled to an overall singlet state Y00 (1. 2 = 1. . 2. . .2: How many overall singlet states can be constructed from four spin– 1 states 2 | 1 m1 (1) | 1 m2 (2) | 1 m3 (3) | 1 m4 (4) ? Construct these singlet states in terms of the product wave 2 2 2 2 functions above.6.233. Ω2 ). . each of the blocks in (6. . . (2( 1 + 2 ) + 1) ×(2( 1 + 2 ) + 1) Partitioning in smaller blocks is not possible. 1 . .234) Exercise 6.

222) in case (ii) of single particle spin operators. q = −k. some of the r r examples considered below involve tensor operators T of this type. .g.223). The latter implies that T transforms like an angular momentum or spin state | m . S 2 = S1 + S2 + S3 or any other polynomial of Sk . Rotations transform |ψ as |ψ = R(ϑ)|ψ and T |ψ as R(ϑ)T |ψ . that in case (i) T is an operator C∞ ( ) → C∞ ( ) acting on single particle wave functions.236) In this equation Dq q (ϑ) denotes the rotation matrix Dq q (ϑ) = kq |R(ϑ)|kq . the operator (5.e. . ∂x2 3 1 2 2 2 2 case (ii) the operator T could be a spin operator Sk deﬁned in (5. e. −k + 1. (6. i. . . 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. k} with the transformation property (6. −k + 1. 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).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.6. . (k) (6. e. In fact. Such operators T can be characterized through their behaviour under rotational transformations. that T belongs to a family of operators {Tkq . q = −k.237) are called tensor operators of rank k. (k) (6. e.7: Tensor Operators 177 6. The latter can be written T |ψ where T denotes T in the rotated frame given by T = R(ϑ) T R−1 (ϑ) .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 .238) . for example.42) in case (i) of the position representation of single particle wave functions or the operator (5. Examples of Tensor Operators The multiplicative operators C∞ ( ) → C∞ ( ) Y ( ) = def Y ( ) (6. k} such that k Tkq = q =−k (k) Dq q (ϑ) Tkq . Note that in the examples mentioned the operator T would be deﬁned within the same representation as R(ϑ). .236. The operator T may also act on multi-particle states like Y 1 m1 (ˆ1 )Y 2 m2 (ˆ2 ). 6.237) The operators T ∈ {Tkq .g. This implies.g.

2. 3 follows from the transformation properties of the spherical harmonics derived in Section 1.7. . we will assume presently that the rotation is chosen as follows R(ϑ) = exp ( ϑ+ L+ + ϑ− L− + ϑ3 L3 ) (6. x2 .7.237) do not require that the rotation R−1 (ϑ) is expressed in terms of Euler angles according to (5. 1.6.2: Express the transformed versions of the following operators (a) T = x2 − x2 and 1 2 (b) S2 S3 in terms of Wigner rotation matrices and untransformed operators.237).3.203). The fact that these operators form tensor operators of rank 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. but rather any rotation and any parametrization can be assumed.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. In fact. This choice of parametrization allows us to derive conditions which are equivalent to the property (6.239) These operators can be expressed in terms of the coordinates x1 .240) 1 where we have deﬁned ϑ± = 2 (ϑ1 iϑ2 ) and L± = L1 ± iL2 . but are far easier to ascertain for any particular operator. 2. Exercise 6. For the following it is important to note that the rotation matrix elements (6.235 . x3 .178 are tensor operators of rank k. Exercise 6.

q =−k q+1 (6.246. Exercise 6. Tkq ] [J3 . Exercise 6.237) be assumed then.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 . (0. J− . Expressing the results in terms of the angular momentum operators J+ . Tkq ] = = δqq and by subtracting Tkq on both sides of the k ϑ+ kq |L+ |kq Tkq .245) (6.6.7. We ﬁrst consider a rotation with ϑ = (ϑ+ . . 0)T in case of small ϑ+ .235 .6.243) From (5. J3 yields [J+ . 0)T .6.1 particles for which the ﬁrst spin is described by 2 the operator S (1) and the second spin by the operator S (2) . The property (6. 2 Exercise 6. (k + q + 1)(k − q) (k + q)(k − q + 1) (6. For this 2 2 purpose state ﬁrst the proper rotation operator R(ϑ) and note then that S (1) · S (2) commutes with 1 the generators of the rotation of | 2 m1 (1) | 1 m2 (2) .4: Is the 1-particle Hamiltonian 2 H = − 2 2m + V (|r|) (6.242) from which follows by means of kq |1 1|kq equation ϑ+ [L+ . 0.5: Consider a system of two spin. (6. Tkq ] [J− .80) follows kq |L+ |kq = −i (k + q + 1)(k − q)δq and.235 . hence.7.6.3: Derive Eqs. 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) . Tkq ] = = = Tkq+1 Tkq−1 q Tkq .244) [L+ . ϑ− . ϑ− . (6.247) These properties often can be readily demonstrated for operators and the transformation properties (6. 1 (6.7.237) yields ﬁrst k R(ϑ) Tkq R −1 (ϑ) = q =−k kq |R(ϑ)|kq Tkq .248) a tensor operator? Exercise 6.247). 6.235 . Similar equations can be derived for inﬁnitesimal rotations of the form ϑ = (0. ϑ3 .246) (6.7: Tensor Operators 179 The conditions can be derived if we consider the property (6. (6. 0.237) for transformations characterized by inﬁnitesimal values of ϑ+ . ϑ3 )T .6: Form tensor operators of rank 1 in terms of the three components of acting on the space of 1-particle wave functions.7. Tkq ] = −iTk q+1 (k + q + 1)(k − q) .

| 1 m1 γ1 denotes an angular momentum (spin) state.e. m1 . J3 | 2 m2 γ2 = m2 γ2 | 2 m2 and the property (6. . possibly the total angular momentum– spin state of a compositie system. 2 |γ2 ) are orthogonal to | 1 m1 in case of diﬀerent quantum numbers 1 . i. 2 |γ2 ) = Cδ 1 1 δm1 m1 2 1( 1 (6.237) is that their matrix elements 1 m1 γ1 |Tkq | 2 m2 γ2 obey simple relationships expressed in terms of Clebsch-Gordan coeﬃcients. 2 |γ2 ) which correspond to total angular momentum states. J3 | 2 m2 γ2 = m2 | 2 m2 γ2 . which is characterized also by a set of other quantum numbers γ1 which are not aﬀected by the rotational transformation R(ϑ). One can. i. These states according to (6.252) ∼δm1 q+m2 = ( 1 m1 |kq 2 m2 ) Tkq Similarly. (6.8. 1 m1 |Φ 1 m1 (k. 2 |γ2 ) = ( 1 m1 |kq 2 m2 ) Tkq | 2 m2 γ2 .1: Show that Φ 1 m1 (k.249) which demonstrates.235 .251) The corresponding property for Φ 1 m1 (k. The hermitian property of J3 and J 2 implies that states Φ 1 m1 (k.m2 ( 1 m1 |kq m1 q. J2 . (6.6. 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. one can show that Φ 1 m1 (k. 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 ) is an eigenstate of J 2 with eigenvalue + 1). 2 |γ2 ) = q. 2 |γ2 ) can be shown readily as follows using (6. 2 |γ2 ) is an eigenstate of J 2 with eigenvalue 2 1 ( 1 + 1).m2 2 2 2 We want to show now that these states are eigenstates of J 2 = J1 + J2 + J3 and of J3 where J1 . J3 are the generators of the rotation R(ϑ). i.8 Wigner-Eckart Theorem A second important property of tensor operators Tkq beside (6. Before we proceed we like to point out that the states | 2 m2 γ2 are also eigenstates of J 2 .e.m2 ( 1 m1 |kq 2 m2 ) (J3 Tkq − Tkq J3 + Tkq J3 ) | 2 m2 γ2 = qTkq = q.21) of Clebsch-Gordan coeﬃcients J3 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) ( ) (6.180 Theory of Angular Momentum and Spin 6.247).253) Exercise 6. hence.e. J3 . behave like |kq | 1 m1 . J 2 | 2 m2 γ2 = 2 2( 2 + 1) | 2 m2 γ2 . in fact. construct states Φ 1 m1 (k.m2 2 m2 ) (q + m2 ) Tkq | 2 m2 γ2 | 2 m2 γ2 (6.18) are deﬁned through Φ 1 m1 (k.250) q. the stated property.

s. γ2 ( 1 m1 |kq (6. (6. 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 .g.259) to a combination of magnetic quantum numbers m1 .254) = 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) ( 1 m1 |kq 2 m2 ) . one obtains 1 m1 + 1γ1 |Φ 1 m1 +1 (k 2 |γ2 ) = 1 √ 1 m1 γ1 |J− Φ 1 m1 +1 (k 2 |γ2 ) ( 1 +m1 +1)( 1 −m1 ) 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) = (6. γ1 ||Tk || 2 . i.e.259) The socalled reduced matrix element 1 .258) We can then ﬁnally express the matrix elements of the tensor operators Tkq as follows 1 m1 . γ2 . m2 .32) Tkq | 2 m2 γ2 = and orthogonality property (6. for which the l. To prove this property we consider 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) for a diﬀerent m1 value.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. the matrix elements 1 m1 γ1 |Tkq | 2 m2 γ2 can be reduced to an m1 –independent factor.8.255) At this point the important property can be proven that 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) is independent of m1 . γ2 k− 2 + 1 √ 1 2 m2 ) (−1) 2 1 +1 = 1 . q . γ1 ||Tk || 2 . say m1 + 1. One can then evaluate also the corresponding Clebsch-Gordan coeﬃcient ( 1 m1 |kq 2 m2 ) and determine 1 . γ2 (−1)k− 2 + 1 ( 1 m1 |kq 2 m2 ) (6.2: Determine the matrix elements of the gradient operator d3 rF (r) G(r) of the type (6.256) and noting that the operator adjoint to J+ is J− .257) which establishes the m1 –independence of 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) .261) . Using | 1 m1 + 1γ1 = 1 ( 1 + m1 + 1)( 1 − m1 ) J+ | 1 m1 γ1 (6. γ1 ||Tk || 2 .h. γ1 ||Tk || 2 . e. γ2 = 2 1 +1 1 m1 . (6.6. can be evaluated as easily as possible.260) Exercise 6. γ1 |Tkq | 2 m2 . m1 = q = m2 = 0. γ2 is determined by applying (6. its m1 –dependence being expressed solely through a Clebsch-Gordan coeﬃcient.8: Wigner-Eckart Theorem 181 In order to evaluate the matrix elements (6. γ1 |Tkq | 2 m2 .

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

See. The potential (7. in case of the so-called bag model of hadronic matter. Switzerland. Poststrasse 22. The potentials (7. 7000 Chur.4) is by far the most relevant of the four choices. for example. φ.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).1.4) governs the motion of electrons in hydrogen-type atoms.2) V (r) = 0 r > R The third potential V (r) = 1 mω 2 r2 2 (7. The corresponding wave functions serve basis functions for multi-electron systems in atoms. It leads to the stationary electronic states of the hydrogen atom (Z = 1). 1993) 1 183 . Talmi (Harwood Academic Publishers.1) conﬁnes a freely moving particle to a spherical box of radius R. Potential (7.e. i. in potentials which are solely a function of r and are independent of the angles θ.3) describes an isotropic harmonic oscillator .3) describes the motion of a charge in a uniformely charged sphere and can be employed to describe the motion of protons and neutrons in atomic nuclei1 . The second potential is of the square well type −Vo 0 ≤ r ≤ R . and crystals. The fourth potential V (r) = − Z e2 r (7. Four examples will be studied. 7. (7. for example.2) serve as schematic descriptions of quantum particles. The ﬁrst potential V (r) = 0 0 ≤ r ≤ R ∞ r > R (7. molecules.. Simple Models of Complex Nuclei / The Shell Model and the Interacting Boson Model by I.

15) (7. one can state ˙ J = m r2 θ . stationary states ψE.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 . ˆ dt (7.99)] 2 1 2 J2 2 2 = − ∂r r + . In fact. 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. 3. Employing in this plane the coordinates r.14) (7.13) .f. 2 .12) .m (r) . k = 1. the time variation of J can be written.5) In the case of an angular independent potential angular momentum J = m r × v is a constant of ˙ motion. 2. For this purpose one concludes ﬁrst that the conservation of angular momentum J implies a motion of the particle conﬁned to a plane. J3 ψE. ˆ A classical particle moving in a potential V (r) is governed by the Newtonian equation of motion (7.7) ˆ The correspondence principle dictates then for the quantum mechanical Hamiltonian operator H and angular momentum operators Jk ˆ [H. 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 .10) (7. one can conclude ˙ Jk = {H.. d ˙ J = v × mv + r × mv = 0 + r × er (−∂r V (r)) = 0 .5) and r = v.e. Jk } where H is the Hamiltonian. Accordingly. (7.m (r) m ψE. 2 2mr2 (7.m (r) . k = 1.184 Spherically Symmetric Potentials 7. Jk ] = 0 . i. 2 .85– 5. (7. θ for the distance from the origin and for the angular position. ˆ H ψE. Jk } = 0 . .11) (7. 3 . using (7.8) This property can also be proven readily employing the expression of the kinetic energy operator [c. J ψE.87) for Jk as well as the commutation property (5.6) ˙ Since Jk is also equal to the poisson bracket {H. (7. ( + 1) ψE. .m (r) (7.9) − 2m 2m r 2mr2 the expressions (5.m (r) = = = E ψE.61) of J 2 and Jk . (5. 2.m (r) .

m ψE.m (r) and E .m (r) 1 2 J2 ∂r r + + V (r) − E 2m r 2mr2 the functional form ψE.10) − 1 2 ∂ r + 2m r r 2 2 ( + 1) + V (r) − E 2mr2 .m .12) are obeyed and one obtains for (7.1: Radial Schr¨dinger Equation o 185 This is a diﬀerential equation which governs solely the radial coordinate. Employing the kinetic energy operator in the form (7. by integration of ˙ θ = J . . 7. together with the original potential.7. but can also trap particles in the latter space giving rise to strong scattering resonances (see Section ??). φ) (7. 2 (7. (r) = 0 .m (r) .m (r) 2 . − Adopting for ψE.24) .20) in the form of the one-dimensional Schr¨dinger equation o 2 − where 2m 2 ∂r + Veﬀ (r) − E φE (r) = 0 . i.16) m 2mr2 Once. .11). ∞[ governed by the eﬀective potential (7.23) This demonstrates that the function r vE. equations (7.22) (7. It can be solved by integration of 1 2 2 J2 r = ± ˙ E − V (r) − . (7.m (r) describes the radial motion as a one-dimensional motion in the interval [0. r(t) is determined the angular motion follows from (7. in the present case.22) which is the original potential 2 ( +1) V (r) with an added rotational barrier potential 2mr2 .11. using (7. (7. .m (r) = 0. .20) Since this equation is independent of the quantum number m we drop the index m on the radial wave function vE.9) one can write the stationary Schr¨dinger o equation (7.18) = vE. for the wave function of a quantum mechanical particle a diﬀerential equation which governs solely the r-dependence.13). (7.19) where Y m (θ.. ( + 1) 2mr2 .4.21) Veﬀ (r) φE (r) = = V (r) + r vE. . φ) are the angular momentum eigenstates deﬁned in Section 5. Multiplying (7. mr2 (t) (7.e.17) In analogy to the classical description one can derivem.10).m vE. (7. One can write (7. . can exclude particles from the space with small r values.20) by −2mr/ 2 yields the so-called radial Schr¨dinger equation o 2 ∂r − ( + 1) − U (r) − κ2 r2 r vκ.m (r) Y m (θ. .m (r) = 0. This barrier. (7.

see. For r → 0 one may assume that the term ( + 1)/r2 becomes larger than the potential U (r). 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). Galindo and P. the term Br−( +1) would contribute |B|2 0 dr r−2 +1) (7. for > 0 is not integrable. ψE. for example. using the expression (5. (r) ∼ A r or vκ. In this case the solution is governed my ( + 1) 2 r vκ. We replaced in (7. Berlin. For = 0 the contribution of Br−( function is. .29) Only the ﬁrst term is admissable. II by A.27) ∂r − r2 and. (7. ”Classical Electrodynamics. (r) = 0 .20) the index E by the equivalent index κ.20) one needs to specify proper boundary conditions2 .25) (7.32) The total kinetic energy resulting from this contribution is.m (r) ∼ √ B . . 2nd Ed. 2 2 2m 2 ψE. The radial part of this integral is ∞ 0 2 dr r2 vκ. according to a well-known result in Classical Electromagnetism3 . accordingly. assumes the general form r vκ. This follows for > 0 from consideration of the integral which measures the total particle density. New York.186 where we deﬁned U (r) κ2 = = − − 2m 2 Spherically Symmetric Potentials V (r) E (7.182) for Y00 . r → 0 (7. hence. at r = 0 is found in the excellent monographs Quantum Mechanics I. 4π|r| (7.D. Jackson (John Wiley.” by J. 1975).30) and.m (r) ∼ √ 4π B δ(x1 )δ(x2 )δ(x3 ) . Boundary Conditions In order to solve (7.28) (7.33) A detailed discussion of the proper boundary conditions. in particular. . (r) ∼ A r + B r− −1 +1 + B r− (7.31) to the complete wave which.26) 2m 2 In case E < 0. κ assumes real values. (r) (7. Pascual (Springer.

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).

according to (5.178) are given in terms of Legendre polynomials P (cos θ) one can replace the expansion in (7.103) and. φ). This holds for −1 and one can conclude +1 consecutive integrations by part d izx e dx (7. One can show d −1 2 (x − 1) ∼ (x2 − 1) × polynomial in x dx −1 (7.s. one given o by (7.99) +1 d cos θ eikr cos θ P (cos θ) = b j (kr) −1 2 . hence.h. In this case the expansion on the r. 7.98) does not involve any non-vanishing m-values since Y m (θ.101) 2 ! ∂x −1 −1 Integration by parts yields +1 dx eizx P (x) −1 = 1 2 ! − 1 2 ! eizx d −1 2 (x − 1) dx −1 dx d izx e dx +1 (7.102) −1 d −1 +1 −1 dx −1 (x2 − 1) . One can expand the former solution in terms of solutions (7.194 Generating Function of Spherical Bessel Functions Spherically Symmetric Potentials The stationary Schr¨dinger equation of free particles (7.179) yield from (7.s. (7.44. For example.50). can be written exp(ikr cos θ).h. the surface term ∼ [· · ·]+1 vanishes. the wave function does not depend on φ.98) by ∞ eik3 x3 = =0 b j (kr) P (cos θ) .43) has two solutions. −1 .45) and one given by (7. 2 +1 (7.m a mj (kr) Y m (θ. z = kr. The orthogonality properties (5.98) The l. in case of a free particle moving along the x3 -axis one expands eik3 x3 = . and using the Rodrigues formula for Legendre polynomials (5. (7. Since the spherical harmonics Y 0 (θ. (7. i. namely.99) We want to determine the expansion coeﬃcients b . φ) .104) dx eizx P (x) −1 = (−1) 2 ! (iz) = 2 ! +1 dx (x2 − 1) −1 +1 dx (1 − x2 ) eizx .106)..100) Deﬁning x = cos θ.150) one obtains +1 +1 1 ∂ dx eizx P (x) = dx eizx (x2 − 1) . of (7.50). φ) for non-vanishing m has a non-trivial φ-dependence as described by (5.e.

113) dt (1 − t2 )ν− 2 eizt .106) Employing (5. ∞ (7.h. Comparision with the l.106) yields ﬁnally b = i (2 + 1) or.96) one can express this. 1 (7.100) gives b j (kr) 2 (iz) = 2 +1 2 ! +1 195 dx (1 − x2 ) eizx .111) Employing (7.109) eikr cos θ = =0 i (2 + 1) j (kr) P (cos θ) . −1 (7.114) .h. z i or i (2 + 1) z (2 )! 2 2 2 +1 2 ! [1 · 3 · 5 · · · (2 − 1)] 1 (7. The identity (7.s. after insertion into (7. −1 (7. for the leading power x [c.105) This expression allows one to determine the expansion coeﬃcients b . −1 (7. of (7. in particular.f.69)] b z 2 (iz) = 1 · 3 · 5 · · · (2 + 1) 2 + 1 2 ! +1 dx (1 − x2 ) . using ν = Jν (z) = √ 1 πΓ(ν + 1 ) 2 z 2 ν + 1. 1 (7.7.105) must hold for all powers of z.108) where the last factor is equal to unity.110) One refers to the l. 2 +1 −1 dx (1 − x2 )ν− 2 eizx .s.99). as the generating function of the spherical Bessel functions.112) We want to consider the expression Gν (z) = aν z ν fν (z) where we deﬁne fν (z) = C (7. (7. Integral Representation of Bessel Functions Combining (7.109) results in the integral representation of j (z) j (z) = (z) 2 +1 ! +1 dx (1 − x2 ) eizx .117) one can write the r.2: Free Particle Comparision with (7. (7.107) 1 2 (2 )! 1 · 3 · 5 · · · (2 + 1) 2 + 1 1 · 2 · 3 · 4 · · · (2 − 1) · 2 (7.s.105) and (7.h.

obeys Bessel’s equation (7. We can now demonstrate that Gν (z) deﬁned in (7.121) (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 . Gν (z). z (7. (7.116) fν (z) = − or fν (z) = − From this we can conclude fν (z) + 2ν + 1 fν (z) + fν (z) = 0 .122) (7.2. 4. Exercise 7. 2. for properly chosen endpoints t1 . We consider now the functions u(j) (z) = √ 1 πΓ(ν + 1 ) 2 z 2 ν Cj dx (1 − x2 )ν− 2 eizx .117) dt (1 − t2 )ν− 2 t2 eizt 1 (7. t2 .120) such that we can conclude that Gν (z) for proper integration paths is a solution of (7. To prove this we note fν (z) = i C dt (1 − t2 )ν− 2 t eizt .121).93). 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 (7.113) obeys the Bessel equation (7. (7.93) for arbitrary ν.124) .117). C 1 (7.116) In case that the endpoints of the integration path C satisfy (1 − t2 )ν+ 2 eizt one can write (7.123) Insertion of these identities into Bessel’s equation leads to a diﬀerential equation for fν (z) which is identical to (7.93) as long as the integration path in (7. In fact. t2 of the integration paths C.120) imply also the property (2ν + 1) fν (z) + z fν+1 (z) = 0 . 1 j = 1.4: Prove (7.118) z 2ν + 1 fν (z) + fν (z) . 1 (7.119) We note that equations (7.114.113) satisﬁes (7.120) z 2ν + 1 dt (1 − t2 )ν− 2 eizt + C C 1 1 t2 t1 = 0 (7. 3. 7.196 Spherically Symmetric Potentials Here C is an integration path in the complex plane with endpoints t1 .

124) vanishes along the whole path C3 and.128) C = C1 ∪ C2 ∪ C3 ∪ C4 is a closed path.133) = e i(z−πν/2−π/4) √ 2 zν π Γ(ν + 1 ) 2 2 zν π Γ(ν + 1 ) 2 √ √ ∞ 0 ds [s(1 + is/2)]ν− 2 e−zs .134) Similarly.117) and.112) shows u(1) (z) = Jν (z). (7. and (1 − t2 )ν− 2 eizt = [(1 − t)(1 + t)]ν− 2 eizt = [−is(2 + is)]ν− 2 eiz e−zs = ei(z−πν/2+π/4) 2ν− 2 [s(1 + is/2)]ν− 2 e−zs .135) are known as the Poisson integrals of the Bessel functions. for Rez > 0 and ν ∈ R obey (7.130) We note that the endpoints of the integration paths C2 and C4 . we introduce the so-called Hankel functions (1) Hν (z) = −2 u(2) (z) .7. 1 (7.125) 0 ≤ s < ∞ (7. Following convention.130) holds Jν (z) = (1) 1 2 1 (1) (2) Hν (z) + Hν (z) .132) For Hν (z) one derives. one can state Jν (z) = − u(2) (z) + u(4) (z) .134) and (7. u(3) (z) ≡ 0. (7. therefore.135) (7. dt = i ds.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. (2) Hν (z) = −2 u(4) (z) .93). the integral expression (1) Hν (z) 1 1 1 1 (7.129) The integrand in (7.127) 0 ≤ s < ∞ (7. one can derive (2) Hν (z) = e −i(z−πν/2−π/4) ∞ 0 √ ds [s(1 − is/2)]ν− 2 e−zs . Comparision with (7. using t = 1 + is.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. 1 (7. hence. Accordingly. .126) −1 ≤ s ≤ 1 (7. Since the integrand in (7.131) According to (7. u(2) (z) and u(2) (z) are both solutions of Bessel’s equation (7. (7.

135) which read for ν = + 1 2 H where f The binomial formula yields f (1) 2 (1) 2 1 +2 (1. the Wronskian W (g.134) and (7. (7. h(z) to be linearly independent. For such solutions g(z).143) .e. One obtains W (z) = − 4i .142) and h(z) = H (2) (z) +1 2 one can identify the constant c by using the leading terms in the expansions (7. For the Wronskian connected with the Bessel equation (7.138) The formula for the Laplace transform of sn leads to f and. (7. converge fast for |z| → ∞.140) Bessel Functions with Negative Index Since ν enters the Bessel equation (7. We employ for this purpose the Poisson integrals (7.93) only as ν 2 . As a second order diﬀerential equation the Bessel equation has two linearly independent solutions. hence. h) must be a non-vanishing function. z (7.198 Asymptotic Behaviour of Bessel Functions We want to obtain now expansions of the Hankel functions H Spherically Symmetric Potentials the expansions converge fast asymptotically.141) (1) (1) the derivation of which follows the derivation on page 192 for the Wronskian of the radial Schr¨dinger equation. Hν (z) as well as H−ν (z) are solutions of this equation.140).93) holds the identity 1 W = − W z (7. i. πz (7. The general solution of (7.137) (z) = r=0 r ± i 2 r 0 ∞ ds s +r e−zs .139) (z) = 2 ±i[z − ( e πz +1) π ] 2 r=0 ( + r)! r!( − r)! i ± 2z r (7.136) (1) 2 (z) = 0 ds [s(s ± is/2)] e−zs .141) is o W (z) = − In case of g(z) = H (1) (z) +1 2 c . we obtain H (1) 2 1 +2 (1) 2 (z) = r=0 ( + r)! ! r!( − r)! ± i 2 r 1 z +r+1 (7..2) (z) +1 2 in terms of z −1 such that (z) = e±i[z − ( ∞ +1) π ] 2 2z z ( 1 ) f 2 (z) π ! (7.

H−ν (z) = A H (1) (1) (z) +1 2 One can then expand + BH (2) (z) +1 2 .7. 7.150) According to (7. From (7. (7. (7. (7.97).144) The expansion coeﬃcients A.145) This equation can hold only for B = 0 and A = exp[i( + 1 )π].152) (7.151) Equations (7. 2z 2 (1.147) (z) = π (1.2) (2) +1) 2 (1) +1) 2 (z) = i (−1) H (1) (z) +1 2 .148) holds for the o regular spherical Bessel function j (z) = 1 (1) (2) [ h (z) + h (z) ] .97) one deﬁnes the spherical Hankel functions h (1. in fact.96.. 2 (1. According to (7.132) follows n (z) = 1 2 π (−1) 2z +1 H−( (1) 1 +2) (z) + H−( (2) +1) 2 (z) .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. 7.153) . 7. 7. (7. (7. B can be obtained from the asymptotic expansion (7.2) H + 1 (z) .146.97. 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| → ∞ .146) (z) = −i (−1) H (2) 1 (z) +2 . (7.2) (z) +1 2 199 are.2) (7.140).2) (7. 7.151) are equivalent to h h (1) (2) (z) (z) = = j (z) + i n (z) j (z) − i n (z) .e. We conclude 2 H−( Simarly. (7.53).149) We want to establish also the relationship between h (z) and the irregular spherical Bessel function n (z) deﬁned in (7.90.132. H (1.149.2: Free Particle i. 7. linearly independent.147) this can be written n (z) = 1 2i h (1) (z) − h (2) (z) . one can show H−( Spherical Hankel Functions In analogy to equations (7.

as given by (7.154) The leading term in this expansion. are j (z) n (z) = = (7. it follows z ∈ R : Using p (z) = Re r=0 j (z) = Re[h (1) (z)] .159) (7. (1) (7. = −cos(z − π/2) results in 2 2 the alternative expressions j (z) n (z) = = sin(z − π/2) cos(z − π/2) p (z) + q (z) z 2z 2 cos(z − π/2) sin(z − π/2) q (z) − p (z) . (7.161) (7.140. at large |z|. = sin(z − π/2) and sin[z − ( + 1) π ] .148). at large |z|.160) Employing cos[z − ( + 1) π ] .164) .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 (7.148) one obtains readily h (1) 2 (1.200 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).162) The leading terms in these expansions. 2z 2 z sin(z − π/2) z cos(z − π/2) − . are complex conjugates. 7. ne ll(z) = Im[h (1) (z)] . j (z) (z) = ( i) z +1 e±iz r=0 ( + r)! r!( − r)! ± i 2z r (7.163) (7. Hence. is h (1) 2 (z) = ( i) z +1 e±iz .156) ( + r)! r!( − r)! ( + r)! r!( − r)! i 2z r [ /2] = r=0 ( + 2r)! (2r)!( − 2r)! −1 4z 2 r (7. z (7.2) (z).158) one can derive then from (7. 2 2z z (7. From (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) .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.

167) (7. 7.151) follows then that the relatiosnhips hold also for g (z) = j (z) and g (z) = n (z). One can combine (7. For this purpose we need to demonstrate only that the (1. 2 j0 (z) n0 (z) j1 (z) n1 (z) j2 (z) n2 (z) = = = = = = sin z z cos z − z sin z cos z − z2 z cos z sin z − 2 − z z 3 1 3 − sin z − 2 cos z 3 z z z 3 1 3 − 3 + cos z − 2 sin z z z z (7.7. One obtains for = 0.174) and from (7.171) holds for g (z) = h (z) we employ (7.168) (7.165) (7. 7.161.124.166) (7.162) allow one to provide explicit expressions for j (z) and n (z).2) relationships hold for g (z) = h (z). 7.172) where g (z) is either of the functions h obtain the recursion relationship g g (z).176) . (1. From the linearity of the relationships (7.4 dx (1 − x2 )ν− 2 eizx .173) A (z) = z ( +1) z2 +2 z (7.2) To demonstrate that (7.2) g (z) − g (z) z 2 +1 g (z) − g z (7.148) and express g (z) = h (1.149.2) (z) = −2 π 1 √ 2z πΓ(ν + 1 ) 2 z 2 ν C2.172) to +1 (z) +1 (z) = A (z) 1− g (z) g (z) −1 (7. j (z) and n (z). 7. 1 (7.170) Recursion Formulas of Spherical Bessel Functions The spherical Bessel functions obey the recursion relationships g g +1 (z) +1 (z) = = (1.131.171) −1 (z) (7. deﬁning a = −1 and employing fν (z) as deﬁned in (7. 1.169) (7.114) we can write g (z) = a z f 2 ! + 1 (z) 2 . (7.171–7.175) Using Γ( + 1) = !.171.174) We want to prove these relationships.2: Free Particle Expressions for the Spherical bessel Functions j (z) and n (z) 201 The identities (7. 7.

is equivalent to (7. (7.171).5: Provide a detailed derivation of (7. j0 (z) using (7.178) (7.179) leads to (7. i.179) from which follows (7.172). n2 (z) from n0 (z). f yields.182).202 The derivative of this expression is g (z) = Employing (7. .58) it holds o 2 g (z) = − g (z) + z ( + 1) g (z) − g (z) .176) g (z) = z g (z) − g +1 (z) +1 2 Spherically Symmetric Potentials z g (z) + a z f 2 ! +1 2 (z) .121). 7. Exercise 7. Exercise 7. (7.6: Employ the recursion relationship (7.179) g (z) = − z2 g (z) + g (z) − g +1 (z) z .2.173. In order to prove (7.166). The second component of (7.172) we diﬀerentiate (7. To prove (7.173) is equivalent to (7. z2 (7. and (b) n1 (z). in fact.172).2. together with (7.174) we start from (7.174) to determine (a) j1 (z).180) yields g +1 (z) = g (z) − ( + 2) g (z) + z2 +2 g (z) . (7. n0 (z) using (7. z (7.177) (z) = − z f 2 +2 + 3 (z) 2 .. 7.171).173).182) Replacing all ﬁrst derivatives employing (7.180) Since g (z) is a solution of the radial Schr¨dinger equation (7.165).e. j2 (z) from j0 (z).171).181) Using this identity to replace the second derivative in (7. The ﬁrst component of (7.173.

is aﬀected by electromagnetic ﬁelds. t) are present. and are aﬀected by the external electromagnetic ﬁeld. for the existence of discrete photons.g. the attractive Coulomb force between proton and electron in case of the hydrogen atom. i. These sources enter the two inhomogeneous 203 . We will later introduce a simple rule which allows one to account to some limited degree for the quantum nature of the electromagnetic ﬁeld.27– 1.1 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. t) and J(r. 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. It turns out that the proper description of the electromagnetic ﬁeld requires a little bit of eﬀort. an electron in a hydrogen atom. The classical electromagnetic ﬁeld is governed by the Maxwell equations stated already in (1. The problem is that the latter electron. However.29).e.. hence.g. for situations in which the photons absorbed or emitted by the quantum system do not alter the energy content of the ﬁeld. are aﬀected by the Coulomb interaction V (r) which is part of the forces which produce the bound state. e. e. For this purpose we need to establish a suitable description of this ﬁeld. Such description should be suﬃcient for high quantum numbers.e.Chapter 8 Interaction of Charged Particles with Electromagnetic Radiation In this Section we want to describe how a quantum mechanical particle..g. We will describe the electromagnetic ﬁeld classically. i. 8. then state the Hamiltonian which describes the resulting interaction.. must be described consistently. both the Coulomb interaction due to charges contributing to binding the particle.. on an electron in a hydrogen atom. or other charged particles. e. This is achieved in the following derivation.

(8. t) + ∂t A(r. t) − ∂t E(r. t) = − V (r. t). t) (8. accordingly.6) for some vector-valued function A(r. t) + χ(r. t) . (8. t)] and. Equation (8. t) + ∂t A(r.8) = 0 (8.3) reads then × E(r. t) in (8.9) (8. 2nd Edition. t) is a scalar function.2) the term q r˙o δ(r − ro (t)). Jackson (John Wiley & Sons. t) = − V (r. t) − ∂t A(r. t) and B(r. the relationship between ﬁelds and potentials is not unique. t) Interaction of Radiation with Matter = 4 π ρ(r. . t) + ∂t B(r. t) = × A(r. From this follows E(r. contributes to the charge density ρ(r. t) = 4 π J(r.2) In addition. t) . t) through the scalar and vector potentials V (r. t) − ∂t χ(r. t) + v × B(r. t) · B(r. by J. D. t) .11) V (r. t).10) (8. t) where V (r. solves implicitly (8. 1975) for a discussion of these and other conventional units. called gauge transformations. The reader is referred to the well-known textbook ”Classical Electrodynamics”.4) Lorentz Force A classical particle with charge q moving in the electromagnetic ﬁeld experiences the so-called Lorentz force q[E(r. t) −→ V (r. t) which is solved by E(r. t) = 0 = 0. called the scalar potential. the two homogeneous Maxwell equations hold × E(r.7) Gauge Transformations We have expressed now the electric and magnetic ﬁelds E(r. t) −→ 1 A(r. We assume so-called Gaussian units. t] + v × B[ro (t). The particle. Scalar and Vector Potential Setting B(r.1) (8. but leave the ﬁelds unaltered: A(r.4). (8. New York. obeys the equation of motion d p = q dt E[ro (t). t) and A(r.5) where p is the momentum of the particle and ro (t) it’s position at time t.25). called the vector potential. alter the potentials. t) × B(r.1) the term qδ(r − ro (t)) and to the current density ˙ J(r. As is well known.3) (8.204 Maxwell equations1 · E(r. In the non-relativistic limit holds p ≈ mr and (8. in turn. t] (8.5) above agrees with the equation of motion as given in (1. The following substitutions. t) in (8. t) (8.

8. t) + ∂t A(r.9) into (8. t) can be obtained employing (8. (8.2).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) and V (r. t) |r − r | (8. the gauge freedom allows us to impose on the vector potential A(r. t) = 0 . (8. t) . t) (8. such that the latter describes the electromagnetic radiation. this potential results from inserting (8. be achieved. t) − ∂t A(r. t) = 0 the solution is given by the Coulomb integral V (r.9) for the ﬁelds results in × The identity × × A(r. 4π (8. t). (8. (8. t) .12) yields then the Poisson equation 2 V (r. in fact. and the former the Coulomb interactions in the unperturbed bound particle system. In fact. t) = Ω∞ for r ∈ ∂Ω∞ (8. t) . t) the condition · A(r.16) and the vector potential are uncoupled.19) Unfortunately. t) + ∂t V (r.1) · − V (r. t) − 2 × A(r. t) = · A(r. a name which is due to the form of the resulting scalar potential V (r. t) .15) d3 r ρ(r . One would prefer a description in which the Coulomb potential (8. t) .19) and deﬁne J (r.18) together with condition (8. t) − ∂t A(r.6) and (8. t) together with (8. Using the expressions (8. For this purpose we examine the oﬀending term ∂t V (r.13) · A(r.17) A(r. and a component due to the remaining currents. t) = 1 ∂t V (r. t).12) leads us to 2 2 A(r.12) The corresponding gauge is referred to as the Coulomb gauge. The vector potential A(r.19) couples the vector potential A(r. t) = ∂t = 4 π ρ(r. t) = − 4 π ρ(r. the second inhomogeneous Maxwell equation. t) Using · ∂t A(r.16) This is the potential commonly employed in quantum mechanical calculations for the description of Coulomb interactions between charged particles. t) in (8. In fact. t = 4 π J(r. t) = − 4 π J(r.14) In case of the boundary condition V (r. (8. t) and the current due to moving net charges. equation (8.20) . Such description can. t) − ∂t V (r.

t) = − 4 π Jt (r. t) = × A(r. V (r.29) · Et (r. A(r. t) = − ∂t ρ(r. t) These ﬁelds are obviously divergence -free (e.30) . t) contributes solely an electric ﬁeld component E (r. t) = 0).23) This continuity equation identiﬁes J (r. t) as the current due to the time-dependence of the charge distribution ρ(r. respectively. the name transverse ﬁelds..26) turns into the well-known wave equation 2 2 A(r. i. hence.12) and the Coulomb potential (8. For J also holds the continuity equation · J(r. no divergence-free currents. Presently. for example. Let J(r.g.25) Because of properties (8.e. t) = 0 . t). (8. t) (8. t) be the total current of the system under investigation and let Jt = J − J . t) − ∂t A(r. we want to assume that no ring-type currents. t) = 0 . t) holds. t) . (8. t) = 0 and from this follows · Jt (r.28) (8.25) one refers to J and Jt as the longitudinal and the transverse currents.14). t) = − V (r. using ∂t · J (r. (8. t) + ∂t ρ(r. t) = 0 .26) and (8. t) contributes an electrical ﬁeld component as well as the total magnetic ﬁeld Et (r. t) . such current densities exist.19) yield 2 2 A(r.206 For the curl of J holds × J (r. t) = 0). t) (8. hence.26) This equation does not couple anymore scalar and vector potentials. yet a current. = − ∂t A(r. exist in the space considered. In this case (8. t) + ∂t ρ(r. in a ring-shaped antenna which exhibits no net charge.24) (8.2 Planar Electromagnetic Waves The current density Jt describes ring-type currents in the space under consideration. V (r. (8. For the divergence of J (r. The vector potential determined through (8.21) and (8.21) ∂t and the Poisson equation (8. t) = 0 (8. t) − ∂t A(r. the name longitudinal electric ﬁeld.22) 1 ∂t 4π · J (r..16) yield ﬁnally the electric and magnetic ﬁelds. The deﬁnitions of J and Jt applied to (8. t) = or = 2 Interaction of Radiation with Matter (8.27) which is obviously curl-free ( × E (r. 8. t) Bt (r.

t) .32. Obviously. accordingly.33) ωt) (8.39). 8π (8. according to (8. The corresponding electric and magnetic ﬁelds.35) The vector potential in (8. see also the comment below Eqs. In applying the potential and ﬁelds to physical observables and processes we will only employ the real parts.35).12) yields u·k = 0.29).34. is orthogonal to k. ω2 = |Ao |2 (8. 2 . t) Bt (r. The Coulomb gauge condition (8. 8. (8.35) are complex-valued quantities.32) holds.34) (8.38) ˆ In this expression for the energy ﬂux one can interprete k as the propagation velocity (note c = 1) and. are Et (r. Et (r. t) in (8. (8. t) = Re Et (r. obviously.39) 8π The deﬁnition incoming waves and outgoing waves is rationalized below in the discussion following Eq.31) u is a unit vector (|ˆ| = 1) which.37) ˆ ˆ where k is the unit vector k = k/|k|. 8. (8. the constant k is referred to as the wave vector. Time average over one period 2π/ω yields S(r. hence. at each point r and moment t. (8. there exist two linearly ˆ u independent orientations for u corresponding to two independent planes of polarization. t) = ±i ω A(r.38. The latter vector describes the direction of propagation of the energy ﬂux connected with the plane wave electromagnetic radiation.36) 4π Using the identity a × (b × c) = b (a · c) − c (a · b) and (8.2: Planar Electromagnetic Waves 207 which describes electromagnetic ﬁelds in vacuum.158). t) = ± ω2 ˆ |Ao |2 k . t) .31. 8.31) and the resulting ﬁelds (8. 8. This ﬂux is given by 1 S(r. Note that in the units chosen the velocity of light is c = 1.35) one obtains S(r. 8. t) and Bt (r. 8. t) = Ao u exp i(k · r ˆ where the dispersion relationship |k| = ω (8. ˆ (8.31). A complete set of solutions is given by the so-called plane waves A(r. t) = i k × A(r.34. are orthogonal to each other and are both orthogonal to the wave vector k. t) × Re B(r.34.28. 8. ˆ We want to characterize now the radiation ﬁeld connected with the plane wave solutions (8. t) = ± ω2 ˆ |Ao |2 k sin2 ( k · r − ωt ) 4π (8. Here the “-” sign corresponds to so-called incoming waves and the “+” sign to outgoing waves2 .8.

t) 2m 2 + qV (r) . t). t) are complex leading to the diﬀerence of a factor 2 in the energy densities of the lontitudinal and transverse components of the ﬁelds. the energy of the plane wave is transported in the direction of −k.29) together with the potentials (8. The sign in (8.32).40) = V It should be pointed out that Nω represents the number of photons for a speciﬁc frequency ω. We assume that the energy content of the ﬁelds is not altered signiﬁcantly in the processes described and. whereas in the case of outgoing waves the energy is transported in the direction of k.41) ωV Inserting this into (8. These photons each carry the energy ω. t). If we consider a volume V with a number of photons Nω the energy density is obviously Nω ω .208 Interaction of Radiation with Matter as the energy density.31). (8. t) ˆ replacing the classical momentum p by the diﬀerential operator p = of the particle is then described by the Schr¨dinger equation o ˆ i ∂t Ψ(r. is p − q A(r. A correct description of the electromagnetic ﬁeld requires that the ﬁeld be quantized.3 Hamilton Operator The classical Hamiltonian for a particle of charge q in a scalar and vector potential V (r) and A(r.27.38) implies that for incoming waves. We are left with a classical Hamiltonian function which has an obvious quantum mechanical analogue ˆ H = ˆ p − q A(r. (8. 8. deﬁned below Eqs.28. hence. respectively.31) allows one ﬁnally to state for the planar wave vector potential A(r. Note that E (r. t) .42) 8. The wave function Ψ(r. A ‘poor man’s’ quantization of the ﬁeld is possible at this point by expressing the energy density (8.8. t) = 8πNω u exp i(k · r − ωt) ˆ ωV . 8. (8. t) = H Ψ(r.43) Here the ﬁelds are deﬁned through Eqs. The integrals express the integration over the energy density of the ﬁelds. (8.45) .43). Bt (r.31.39) through the density of photons connected with the planar waves (8. 8. a ˆ speciﬁc k and a speciﬁc u. (8. u · k = 0 .40) allows one to express then the ﬁeld ˆ amplitudes 8πNω Ao = .39) and (8. |k| = ω . (8. t) 2m + 1 8π Ω∞ 2 H = + qV (r) 1 16π d3 r |Et |2 + |Bt |2 Ω∞ d3 r E 2 + .31). ˆ (8. i (8.16. Comparision of (8. we will neglect the respective terms in the Hamiltonian (8. t) is real 1 and that Et (r.44) .

t)/ Ψ(r. an element of the group U(1).11) aﬀect the quantum mechanical description. e.t)/ [ i ∂t − q((∂t χ)) ] Ψ(r.45) leads to the Schr¨dinger equation o 2 ˆ p − q A − q(( χ)) + qV − q((∂t χ)) Ψ(r. does not change expectation values which contain the probability densities3 . by the group 3 The eﬀect on other expectation values is not discussed here.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. 8.t)/ should not aﬀect a system and that.47) i ∂t Ψ(r.t)/ Ψ(r. hence.47. 8. Elementary constituents of matter which are governed by other symmetry groups. t) (8.t)/ Ψ(r. t) enter whereas in the classical equations of motion ˙ m¨ = q E(r. such phase factor does not change the probability density |Ψ(r. t) = + qV eiqχ(r. t) r (8. One can show that (8.10. t) + q r × B(r.11) to (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.10.47) is equivalent to 2 ˆ p − qA i ∂t eiqχ(r. The idea just stated can be generalized by noting that multiplication by a phase factor eiqχ(r. in the Schr¨dinger equation o 2 ˆ p − qA i ∂t Ψ(r.10.. (8.46). .. t) are necessary to compensate terms which arise through the phase factor. In the classical case such question is mute since the gauge transformations do not alter the ﬁelds and. This leads to the question in how far the gauge transformations (8. hence.t)/ constitutes a unitary transformation of a scalar quantity. t) by a local and time-dependent phase factor eiqχ(r.g.48) implies that the gauge transformation (8.t)/ .11) of the potentials is equivalent to multiplying the wave function Ψ(r. 8. t) ˆ p eiqχ(r. Obviously. t) = eiqχ(r. however. It should be noted. Applying the gauge transformations (8.t)/ Ψ(r. t) ˆ = eiqχ(r. 8. t) . (8. t) = 2m where ((· · ·)) denotes derivatives in (( χ)) and ((∂t χ)) which are conﬁned to the function χ(r. accordingly.50) The equivalence of (8. not to phenomenological interactions like the molecular van der Waals interaction. t) and A(r.49) (8. t) inside the double brackets. have no eﬀect on the motion of the particle described by (8.51) 2m the potentials A(r. that this principle applies only to fundamental interactions. t) = + qV Ψ(r.48) 2m For this purpose one notes i ∂t eiqχ(r. t)|2 and. t) and V (r.44. (8.46) the ﬁelds enter. t) . 8. i.e. t) .8.t)/ p + q(( χ)) Ψ(r.

54) ˆ2 p q ˆ q2 2 − p·A + A + qV . one can employ the Hamiltonian (8. t) ≡ 0.53) ˆ ˆ p·A + A·p + q2 2 A + qV 2m (8. The choice of a vector potential aﬀects the form of the wave functions describing the eigenstates and. t) = Bo .. (8. j = 1. the generators of SU(2). e.55). Accordingly.210 Interaction of Radiation with Matter SU(2). The Hamiltonian (8. consequently.4 Electron in a Stationary Homogeneous Magnetic Field We consider now the motion of an electron with charge q = −e and mass m = me in a homogeneous magnetic ﬁeld as described by the Schr¨dinger equation (8. (8. therefore. 2. thereby. · A = 0 [cf. H = 2m m 2m (8. For this purpose we use the wave function in the form Ψ(E. can be described by various vector potentials. holds (8. t). The resulting ﬁelds are called Yang-Mills ﬁelds. The stationary homogeneous magnetic ﬁeld B(r. The vector potential (8. k3 .56). In this o case holds V (r. for Bo = Bo e3 in (8. x3 ) = exp(ik2 x2 + ik3 x3 ) φE (x1 ) . likewise can demand the existence of ﬁelds which compensate local transformations described by eiσ·χ(r. We want to describe the stationary states corresponding to the Hamiltonian (8. In case of a homogeneous potential pointing in the x3 -direction. (8.44) can be expanded ˆ2 p q H = − 2m 2m For any function f (r) holds ˆ ˆ p · A − A · p f (r) = i A· f + f ·A − A· f = i f ·A. 3.58).12)].45) with Hamiltonian (8. the so-called Landau gauge ˆ associates the vector potential AL (r) = Bo x1 e2 ˆ (8. aﬀects the complexity of the mathematical derivation of the wave functions. x2 .t) where σ is the vector of Pauli matrices.52) This expression vanishes in the present case since since ˆ ˆ p · Af = A · pf and.55).59) .58) where ∂j = (∂/∂xj ).g.57) satisﬁes · A = 0 and. (8.57) with a homogeneous magnetic ﬁeld Bo . 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. x1 . k2 .56) due to the gauge freedom. Solution for Landau Gauge A particularly convenient form for the Hamiltonian results for a choice of a so-called Landau gauge for the vector potential A(r.55) 8.

k3 . the wave function (8.66) Obviously. the familiar harmonic oscillator quantum number. therefore.67) . x2 . the states are degenerate in the quantum number k2 describing displacement along the x2 coordinate. k3 ) = ω(n + 2 k2 1 3 ) + .62) is that of a displaced (by x1o ) harmonic oscillator with o 2 k 2 /2m ) energies. is Ψ(n.59). like the corresponding gauge (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. 2 2me (8.65). Solution for Symmetric Gauge The solution obtained above has the advantage that the derivation is comparatively simple. according to (8.57). (8. From this observation one can immediately conclude that the wave shifted (by e 3 function of the system.59). The energies corresponding to these states are E(n. (8.60) Completing the square 2 2 e2 Bo 2 eBo k2 e2 Bo x1 + x1 = 2me me 2me x + k2 eBo 2 − 2 k2 2 2me (8. k2 .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 ) .56) through the vector potential A(r) = 1 Bo × r .8. the so-called symmetric gauge which expresses the homogeneous potential (8. Without aﬀecting the energy one can form wave packets in terms of the solutions (8.64) is the classical Larmor frequency (c = 1). 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. We want to employ .65) where we replaced the parameter E by the integer n. Unfortunately. The stationary Schr¨dinger equation (8. is not symmetric in the x1 . according to (8.63) this induces a spread of the wave function in the x1 direction.63) (8.62) x1o = and where ω = k2 eBo eBo me (8. However.and x2 -coordinates. x1 .65) which localize the electrons. 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 ) . k2 .

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 . In this case the current density measures −e v oriented tangentially to the ring. y-plane with constant velocity v on a ring of radius r.70) ˆ ˆ p · A f = − Bo · p × r f .71) (8.74). (8. accordingly. in the present case of constant Bo .73. namely.74) We consider a simple case to relate (8.69) × u and × r = 0. Vmag = −µclass · Bo (8. For the vector potential (8.78) . namely.77) Comparision with (8. The latter can be rewritten.12) for the Coulomb gauge.76) e 2me class class · Bo . ˆ (8.68) × u yields.73) where µclass is the magnetic moment connected with a current density j µclass = 1 2 r × j(r) dr (8. for any (8. the Hamiltonian (8.74) is in the present case 1 µclass = − e r v e3 . ˆ Identifying r × p with the angular momentum operator L. an electron moving in the x. the magnetic moment (8. using × (uf ) = −u × f +f ˆ ˆ Bo · p × r f = Bo · r × p f . 8.71). (8. (8.75) 2 The latter can be related to the angular momentum µclass = − and. (8. Accordingly.72) to Hamiltonian (8. The theory of classical electromagnetism predicts an analogue energy contribution .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 .212 Interaction of Radiation with Matter One can readily verify that this vector potential satisﬁes the condition (8.72) and (8. Vmag = e 2me class = r me v e3 of the electron ˆ (8.

j = 1. the Hamiltonian (8. we separate the variable x1 . Accordingly. i. 10 that the spin of the electron.88) me ω xj .67) yields a more symmetric solution which decays to zero along both the ±x1 .85) The Hamiltonian (8.and the ±x2 -direction.84) . To obtain these eigenstates we introduce the customary dimensionless variables of the harmonic oscillator Xj = (8.83) describes two identical oscillators along the x1 -and x2 -directions which ˆ are coupled through the angular momentum operator L3 . e gives rise to an energy contribution (8. x3 ) . ˆ i. In this case. x2 ) obey then ˆ Ho ψ(x1 . E = E − (8. requires a Lorentz-invariant quantum mechanical description as provided in Sect. 10.87) . likewise.80) (8. x3 ) = E ΨE (x1 .82) 1 ω (x1 ∂2 − x2 ∂1 ) i 2me + 1 me ω 2 x2 + x2 1 2 2 2 k2 3 + (8.83) (8..8. x2 from x3 setting ΨE (x1 .71) is ˆ H = 2 ˆ2 p e2 Bo + 2me 8me x2 + x2 + 1 2 eBo L3 . for Bo = Bo e3 . 2 .72) with an associated magnetic moment − g 2me S where g ≈ 2. the so-called gyromagnetic ratio of the electron. x2 .81) (8. For a magnetic ﬁeld (8. x2 ) = E ψ(x1 .79) To obtain the stationary states. (8. 2me We have used here the expression for the angular momentum operator ˆ L3 = ( /i)(x1 ∂2 − x2 ∂1 ) . (8.86) 1 2 2me 1 2 ˆ as well as of the angular momentum operator L3 . x2 . The functions ψ(x1 . described by the operator S.83) can then be expressed 1 ˆ 1 Ho = − ω 2 ∂2 ∂2 + 2 2 ∂X1 ∂X2 + 1 2 2 X1 + X2 2 + 1 i X1 ∂ ∂ − X2 ∂X2 ∂X1 .4: Electron in Homogenous Magnetic Field 213 as the quantum mechanical magnetic moment operator for the electron (charge −e). 2me (8. x2 ) .e. x2 . x2 ) where ˆ Ho = − 2 2 2 ∂1 + ∂2 (8.56) pointing in the x3 -direction the symmetric gauge (8.e. A derivation of his property and the value of g. the solutions of ˆ H ΨE (x1 . x3 ) = exp(ik3 x3 ) ψ(x1 . we seek stationary states which are simultaneous eigenstates of the Hamiltonian of the two-dimensional isotropic harmonic oscillator 2 1 2 ˆ Hosc = − ∂ 2 + ∂2 + me ω 2 x2 + x2 (8. We will demonstrate in Sect.

e. x1 .94) Since this state is a linear combination of states which all have vibrational energy (2j + 1) ω. 0. m . x1 . such that 1 i a† a2 − a† a1 1 2 Ψ (j. x2 ) (8. . . x2 ) = ω 2j + 2m + 1 Ψ (j. x2 ) is the wave function for the state with zero vibrational quanta for the x1 . m = −j.214 Employing the creation and annihilation operators 1 a† = √ j 2 and the identity ˆ ω L3 = which can readily be proven.97) .95) (j) ˆ We want to choose the coeﬃcients αmm such that (8. m as follows: j = 0. through a combination of states j Ψ (j. (8. We. m. x1 . x2 ) . x2 ) .94) is an eigenstate of the vibrational Hamiltonian.92) are not eigenstates of L3 . m . −j + 1. m . m . therefore. x2 ) . x2 ) = m=−j αmm Ψ(j. m . x2 ) = 2m Ψ (j. +j .94) is also an eigenstate of L3 . it holds a† a1 + a† a2 + 1 1 1 2 Ψ (j. .89) . 2 2 .. x2 ) (8. x2 ) = ( 2j + 1 ) Ψ (j.93) ˆ The states (8. m .92) represents a state with j + m quanta in the x1 -oscillator and j − m quanta in the x2 -oscillator. 1 2 since one phonon is annihilated and one created. . attempt to express eigenstates in terms of vibrational wave functions a† 1 j+m Ψ(j. m.92) where Ψ(0. If this property is. . (8. . j = 1. (8. .94) is an eigenstate of Ho ˆ Ho Ψ (j. (j) (8.96) ˆ holds.as well as for the x2 -oscillator. (8. x2 ) = a† 1 j−m (j + m)! (j − m)! Ψ(0. x1 . x1 . i. (8. Such eigenstates can be expressed. (8. in fact. 1 3 . 1. however. i.91) We note that the operator a† a2 − a† a1 leaves the total number of vibrational quanta invariant. m . x1 . x1 . obeyed. x1 . the total vibrational energy being ω(2j + 1). In order to cover all posible vibrational quantum numbers one needs to choose j. 2 (8.90) a† a2 − a† a1 1 2 .. x1 . 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. aj = √ 2 Interaction of Radiation with Matter Xj + ∂ ∂Xj . x1 . (8. x1 .e. . 0.

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. thus. the states are eigenstates of the operator [we use for the operator the notation of Sect. 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.9. x2 ) (8. x1 .11. is identical to the operator L3 introduced in (8.e.(5. According to Sect.5. .. x1 . The connection with the present problem arises due to the fact that the operator J2 in Sect. an example is the Coulomb potential V (r. b− + − notation in Sects. (8. If we identify 2 a† . x1 .84) above. m) in Sect. except for a factor 2 . Other radiation ﬁelds can . x2 ) deﬁned in (8. m. π . We assume that the scalar potential V in (8. This implies that ˆ we can obtain eigenstates of L3 by rotation of the states Ψ(j. m . In the simplest case the radiation ﬁeld consists of a single planar electromagnetic wave described through the potential (8.98) then the states Ψ(j. 8.92) correspond to the eigenstates |Ψ(j. t) introduced above. 0) Ψ(j.55) conﬁnes the particle to stationary bound states.11 (8.5.5 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 (8.31). 5. is in the notation of the present section 1 ˆ a† a2 − a† a1 . a2 1 2 present notation ←→ b† .9.101) 2 2 where Dmm ( π .5.11 such transformation is provided through π π (j) Ψ (j.8. i. The ﬁrst rotation contributes a factor exp(−im π ).e.10.10.309) yields ﬁnally Ψ (j. cf.11. m.97). The required rotation must transform the x3 –axis into the x2 –axis. Using the explicit form of the Wigner rotation matrix as given in (5. x1 .55). which corresponds there to the angular momentum in the x2 –direction.102) We have identiﬁed. a1 . x2 ) = Dmm ( . According to the derivation given there.99) with eigenvalue m. the eigenstates of (8.. x2 ).83) and conﬁrmed the eigenvalues stated in (8. the 2 2 latter representing the Wigner rotation matrix of Sect. 5. m . 5.10. t) = 1/4πr conﬁning an electron with energy E < 0 to move in the well known orbitals of the hydrogen atom. x1 .9. 5. 5.5. the second rotation a factor dmm ( π ). m.288)] 1 ˆ J3 = 2 a† a1 − a† a2 1 2 (8. Eq. 5. b† . 5. (8.10. x1 . The external radiation ﬁeld is accounted for by the vector potential A(r. x2 ) . 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. a† . b+ . a transformation moving the x3 –axis into the x2 – 2 axis.100) J2 = 1 2 2i 1 ˆ i.

105) for the case of radiationinduced transitions in atomic systems. In fact.105). However. 1. . the term ∼ A second term in (8.109) The reader should note that the estimates are very crude since we are establishing an order of magnitude estimate only. t). ωV (8. is much 2 (r.105) in further calculations and to expand the wave function in terms of powers of VS (t) in a perturbation calculation. i. we assume n|m = δnm and for the identity ∞ (8. i. t) − m 2m (8. 2 . t) + A (r. For an electron charge q = −e and an electron mass m = me one can provide the estimate for the ﬁrst term of (8.104) (8.42). 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.103) (8.105) as follows4 . We ﬁrst note. t). orthonormal basis.107) 1 = 1 n=0 |n n| . using (8. (8. Estimate of the Magnitude of VS We want to demonstrate now that the interaction VS (t).e. i. can be considered a weak perturbation.. one can estimate that the perturbation.41) e ˆ e p2 p·A ∼ 2me me me 2me 4 1 2 8πNω . and that the ﬁrst term in (8.216 Interaction of Radiation with Matter be expanded through Fourier analysis in terms of such plane waves.e. This result will allow us to neglect the larger than the second term. (8..e. . this assumption can be waved as our results below will not depend on it. t) is given by (8..e. The states |n are thought to form a complete. . we discount the possibility of a continuum of eigenstates.108) We assume for the sake of simplicity that the eigenstates of Ho can be labeled through integers. 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. i.105) where A(r. is much smaller than the eigenvalue diﬀerences near ˆ typical atomic bound states. n = 0. in this case.. the term ∼ p · A(r.106) where we adopted the Dirac notation for the states of the quantum system. Here the so-called unperturbed system is governed by the Hamiltonian Ho with stationary states deﬁned through the eigenvalue problem Ho |n = n |n . as given in (8.

117) In case that such perturbation acts on an electron and is due to superpositions of planar waves described through the vector potential (8. We want to estimate now the second term in (8.113) This magnitude is much less than the diﬀerences of the typical eigenvalues of the lowest states of the hydrogen atom which are of the order of 1 eV.115) (8.114) Employing for the second factor the estimate as stated in (8.42) it holds VS ≈ e m k.5 ˚. (8.112) π λ me c2 and with e2 /ao ≈ 27 eV. i. We can.e.105). e ˆ p · A ∼ 10 eV · 10−4 = 10−3 eV . one can neglect this term as long as the ﬁrst term gives non-vanishing contributions.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 (8.41) 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 ..105) due to a radiation ﬁeld by VS = − q ˆ p · A(r. and as long as the photon densities Nω /V are small. and me c2 ≈ A A 500 keV 2 ao ω ≈ 10−8 (8.8.118) . Nω = 1. λ = 2πc/λ. the ﬁrst term in (8. a volume V = λ3 where λ is the wave length corresponding to a plane wave with frequency ω. u) p · u exp i(k · r − ωt) ˆ ˆ ˆ kV .110) where ao is the Bohr radius. Consequently.112) yields e2 A2 ∼ 10 eV · 10−8 = 10−7 eV . hence. one obtains for (8. i. (8. replace the perturbation (8. altogether.e. t) . with ao ≈ 0. Assuming a single photon. 2me (8.116) This term is obviously much smaller than the ﬁrst term.105) for radiation ﬁelds can be considered a small perturbation. me (8.109) 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. Using again (8.. m (8. Hence.111) For ω = 3 eV and a corresponding λ = 4000 ˚ one obtains.ˆ u 4πNk α(k.

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 .123) i = Ho exp − Ho (t − to ) (8. under ordinary circumstances. The sum runs over all possible k vectors and might actually be an integral. Multiplying the latter equation by the operator exp i (8.118) is the form of the perturbation which.125) (8. One seeks to predict the probability to observe the particle in one of the states |n .. . n = 0 at some later time t ≥ to . 2. . . n = 1. |Ψ(to ) = |0 (8. describes the eﬀect of a radiation ﬁeld on an electron system and which will be assumed below to describe radiative transitions.127) i ∂t |ΨD (t) = VD (t) |ΨD (t) .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. 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 . For this purpose one needs to determine the state |ΨS (t) of the particle.126) Ho (t − to ) yields ﬁnally . Using i i ∂t exp − Ho (t − to ) and (8.120) (8. (8.122) where |ΨD (to ) = |0 .106–8.218 Interaction of Radiation with Matter where we have replaced ω in (8.42) through k = |k| = ω. Equation (8. The probability to ﬁnd the particle in the state |n at time t is then p0→n (t) = | n|ΨS (t) |2 .124) (8.108) as the eigenstates of the unperturbed Hamiltonian Ho . the sum over u involves the two possible polarizations of planar ˆ electromagnetic waves. A factor α(k.119) one obtains i exp − Ho (t − to ) ( Ho + i ∂t ) |ΨD (t) i = [ Ho + VS (t) ] exp − Ho (t − to ) |ΨD (t) from which follows i i exp − Ho (t − to ) i ∂t |ΨD (t) = VS (t) exp − Ho (t − to ) |ΨD (t) . |n are deﬁned in (8. u) has been added to describe eliptically or circularly polarized ˆ waves.121) (8. The states |0 .

139) . Employing this result one obtains for (8.128) We note that the transition probability (8.128). From (8. 8. (8.132–8. 8.136) together with the initial conditions |ψD (to ) = |0 0 for n = 0 for n = 1.137) follows |ΨD (t) = |0 . (8.131) p0→n (t) = | n |ΨD (t) |2 . Due to the Hermitean property of the Hamiltonian Ho holds n|Ho = i n|exp − Ho (t − to ) from which we conclude.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) = = = .134) (8. .137) are consistent with (8. .5: Time-Dependent Perturbation Theory where VD (t) = exp 219 i i Ho (t − to ) VS (t) exp − Ho (t − to ) . Equations (8.134. to (8.133–8.133.137) can be solved recursively.127. . Accord(n) ingly. . (8. n| (8. 8. consequently.133) (8. . 0 VD (t) |ΨD (t) (1) VD (t) |ΨD (t) (0) (8.8. 2. In order to determine |ΨD (t) described through (8.121) expressed in terms of ΨD (t) is i p0→n (t) = | n|exp − Ho (t − to ) |ΨD (t) |2 .132) where |ΨD (t) accounts for the contribution due to n-fold products of VD (t) to |ΨD (t) .138) 1 i t dt VD (t ) |0 .127) we assume the expansion ∞ |ΨD (t) = n=0 (n) |ΨD (t) (n) (8.129) n| and.130) = exp − i n (t − to ) − to )]| = 1. . We will consider here only the two leading contributions to |ΨD (t) . using |exp[− i n (t n (8. = . .137) One can readily verify that (8.135) i ∂t |ΨD (t) (n) VD (t) |ΨD (n−1) (t) (8.

in turn. A factor exp(λt).144) r is the position of the electron and p = ( /i) is the momentum operator of the electron.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. II III (8. 8. A2 are deﬁned through (8. deﬁnes the direction of the E-ﬁeld of the ˆ radiation. This factor will serve mainly the purpose of keeping in the following derivation all mathematical quantities properly behaved. (8. The resulting perturbation on an electron system.220 This result.e.142). note that uj . (8. The factor III in (8. We want to apply now the perturbation expansion derived to such perturbations. yields for (8.144) describes the propagation of the planar wave. is VS = ˆ ˆ V1 exp(−iω1 t) + V2 exp( iω2 t) eλt .137) |ΨD (t) = (2) Interaction of Radiation with Matter 1 i 2 t t dt to to dt VD (t ) VD (t ) |0 ...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 (8.118). that two planar waves simulataneously interact with an electronic system.135. + m=0 dt to to dt n|VD (t )|m m|VD (t ) |0 8. For the sake of including the eﬀect of superpositions of plane waves we will assume. such that the combined radiation ﬁeld is decribed by the vector potential A(r. according to (8..34. to → −∞ (8.142) combining an incoming and an incoming or outgoing wave. according to (8. In (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). λ → 0+ has been introduced which describes that at time to → −∞ the perturbation is turned on gradually. i. the direction of ˆ the propagation being determined by k = k/|k|. non-singular.41). 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. . however. 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 (8.6 Perturbations due to Electromagnetic Radiation We had identiﬁed in Eq.141) + . The coeﬃcients A1 .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 . λ → 0+ . 8.

149) Using the deﬁnition of VD stated in (8. (8.6: Perturbations due to Electromagnetic Radiation 1st Order Processes 221 We employ now the perturbation (8. yields n|ΨD (t) = − (2) 1 ∞ t t 2 λ→0+ t→−∞ lim lim ( ( dt m=0 to to dt i i ( ( − − − ω1 − i λ)t ω2 − i λ)t (8.146) = 1 λ→0+ t→−∞ i lim lim × t dt exp to i ( n − o − i λ) t iω2 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 .8.147) ˆ ˆ n|V1 |0 e−iω1 t + n|V2 |0 e Carrying out the time integration and taking the limit limt→−∞ yields (1) n|ΨD (t) = λ→0+ lim e λt exp ˆ n|V1 |0 o − o − ω1 ) t + + ω1 − n + iλ ( n i i ˆ + n|V2 |0 2nd Order Processes exp o − o ω2 ) t ± ω2 − n + iλ ( n .148) We consider now the 2nd order contribution to the transition amplitude.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 . 8. (8.128) one obtains k|VD (t) | = k|VS (t)| exp i ( k − ) (8. (8.130) and (for m = 0) i i exp − Ho (t − to ) |m = exp − for (8.128).150) and. For the 1st order contribution to the transition amplitude n|ΨD (t) = (1) 1 i t dt n|VD (t ) |0 to (8. (8.140. employing the perturbation (8.141). using (8.143). According to (8.145) n|ΨD (t) (1) m (t − to ) |m .145) we obtain then.143) to the expansion (8.

We also assume for the ﬁnal state n = 0 such that n|0 = 0 and p0→n (t) = | n|ΨD (t) |2 holds.155) λ→0+ ( o We are actually interested in the transition rate. the time derivative of p0→n (t). (8..153) (8.e. We assume ﬁrst that the ﬁrst order transi(1) tion amplitude n|ΨD (t) is non-zero.148) and ∗ |z1 + z2 |2 = |z1 |2 + |z2 |2 + 2Re(z1 z2 ) (1) (8. For this rate holds .131).152). i. Using (8.148) and (8.152) m=0 i ˆ ˆ n|V1 |m m|V1 |0 exp m − o − ω1 − i λ + − 2 ω1 − 2i λ)t n − o − 2 ω1 − 2i λ ( n o − ˆ ˆ n|V2 |m m|V2 |0 exp ω2 − i λ m− o ˆ1 |m m|V2 |0 exp ˆ n|V + ω2 − i λ m− o ˆ2 |m m|V1 |0 exp ˆ n|V + m − o − ω1 − i λ 1st Order Radiative Transitions i 2 ω2 − 2i λ)t 2 ω2 − 2i λ n− o i ( n − o − ω1 ω2 − 2i λ)t ω2 − 2i λ n − o − ω1 ( n o i − − ω1 ω2 − 2i λ)t ω2 − 2i λ n − o − ω1 ( n o − The 1st and 2nd order transition amplitudes (8. provide now the transition probability p0→n (t) according to Eq. in which case one can expect that it is larger than the 2nd (2) order contribution n|ΨD (t) which we will neglect.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. respectively.222 ˆ ˆ + n|V1 |m m|V2 |0 exp ˆ ˆ + n|V2 |m m|V1 |0 exp i i ( ( − − Interaction of Radiation with Matter − ω1 − i λ)t exp ω2 − i λ)t exp i i ( ( − − ω2 − i λ)t − ω1 − i λ)t n m m o n m m o Carrying out the integrations and the limit limt→−∞ provides the result (2) n|ΨD (t) = − 1 ∞ 2 λ→0+ lim (8.

. ˆ (8.159) leads to ﬁnal states |n with energy n = o + ω1 . which is observed through the so-called spectral intensities of transitions |0 → |n .158) play an essential role for the transition rates of radiative transitions. one should average the rate over many periods of the radiation. .142–8.157) one can conclude for the average transition rate k = d p0→n (t) dt t = 2π ˆ | n|V1 |0 |2 δ( n − n o − ω1 ) o (8. First. due to the uncertainty relationship would introduce a considerable perturbation to the system. 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.8. i. in fact. The δ-functions appearing in this expression reﬂect energy conservation: the incoming plane wave contribution of (8.31) and (8. In case the matrix elements are non-zero. of this expression describe the individual eﬀects of the two planar wave contributions of the perturbation (8. the matrix elements can vary strongly for diﬀerent states |n of the system. a property.144).e.158). The result supports our deﬁnition of incoming and outgoing waves in (8.156) ( o The period of electromagnetic radiation absorbed by electronic systems in atoms is of the order 10−17 s. (8.160) leads to ﬁnal states |n with energy n = o ± ω2 .156) such that the 1st order contributions of the two planar waves of the perturbation simply add. The time average will be denoted by · · · t . Hence.158) ˆ + | n|V2 |0 |2 δ( − ω2 ) Obviously. 8. For the resulting expression the limit limλ→0+ can be taken.143. describing either an incoming or an outgoing plane wave due to the vector potential A2 u2 exp i (k1 · r ˆ ω2 t) . Using lim x2 + 2 δ→0+ = π δ(x) (8. however. the two terms apearing on the rhs.142) ˆ ˆ The matrix elements n|V1 |0 and n|V2 |0 in (8. The result of such average is. due to the vector potential A1 u1 exp i (k1 · r − ω1 t) .6: Perturbations due to Electromagnetic Radiation d p0→n (t) dt lim e2λt + × 2 Re ˆ 2 λ n|V1 |0 |2 ( o + ω1 − n )2 + (λ )2 ˆ 2 λ n|V2 |0 |2 d + 2λ + × ( o ± ω1 − n )2 + (λ )2 dt ˆ ˆ n|V1 |0 0|V2 |n exp i (± ω2 − ω1 ) t + ω1 − n + iλ ) ( o ± ω2 − n − iλ ) 223 = λ→0+ (8. is much shorter than could be resolved in any observation.144). to cancel the third term in (8. any attempt to do so. The second contribution to (8.

157) yields.165) × ∞ z3 = m=0 2 ω2 − 2i λ)t 2 ω2 − 2i λ n− o ˆ ˆ ˆ ˆ n|V2 |m m|V1 |0 n|V2 |m m|V1 |0 + ω2 − i λ m − o − ω1 − i λ m− o ( n o i − × It holds exp − ω1 ω2 − 2i λ)t ω2 − 2i λ n − o − ω1 ( n o 3 − |z1 + z2 + z3 | = |z1 | + |z2 | + |z3 | + j.166) over many periods of the radiation yields exp[ i (±ω2 − ω1 )] t = 0 and. Time average · · · t of expression (8. 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..161) To determine the transition rate we proceed again. exp[ i (±ω2 − ω1 )].e. in analogy to (8.158).. ∞ − 2 ω1 − 2i λ)t n − o − 2 ω1 − 2i λ ( n o − z2 = m=0 ˆ ˆ n|V2 |m m|V2 |0 ω2 − i λ m− o exp i × (8. in Eqs. |z1 + z2 + z3 |2 t = |z1 |2 + |z2 |2 + |z3 |2 (8.k=1 j=k 2 2 2 2 ∗ zj zk (8.e.153) vanishes such that the leading contribution to the transition probability p0→n (t) arises from the 2nd order amplitude (8. similarly.168) absorption of 2 photons ω1 . We deﬁne ∞ z1 = m=0 ˆ ˆ n|V1 |m m|V1 |0 m − o − ω1 − i λ exp i × (8. it holds p0→n (t) = | n|ΨD (t) |2 .g.153–8.163) × (8.162) × and. i.158). hence. i.167) Taking now the limit limλ→0+ and using (8.224 2nd Order Radiative Transitions Interaction of Radiation with Matter We now consider situations where the ﬁrst order transition amplitude in (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. (8.164) (8. as we did in the the case of 1st order transitions. e.152). (2) (8.

168) describes two intermediate histories.168) describe the time sequence of the two photon absorption/ emission processes. 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. interference between the contributions from all intermediate states |m can arise. |m ← |0 (8. each of energy ω2 .168) describe similar histories of the excitation process. according to its δ-function factor. 8. i. a second perturbation. through absorption of a photon. |n ← |m . i. The expression sums over all possible virtually occupied states |m and takes the absolute value of this sum. Most remarkably.e.158) which accounts for such transitions. The factors | · · · |2 in (8. For this purpose we will employ the transition rate as given in Eq.e. there is no energy conservation necessary (in general. Similarly. the third term describes processes in which a photon of energy ω1 is absorbed and a second photon of energy ω2 is absorbed/emitted.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. this state is only virtually excited.7: One-Photon Emission and Absorption 2π ∞ m=0 225 2 + ˆ ˆ n|V2 |m m|V2 |0 ω2 − i λ m− o δ( m − o 2 ω2 ) absorption/emission of 2 photons ω2 + 2π ∞ m=0 ˆ ˆ n|V2 |m m|V1 |0 + m − o − ω1 − i λ + ˆ ˆ n|V2 |m m|V1 |0 ω2 − i λ m− o 2 δ( n − o − ω1 ω2 ) absorption of a photon ω1 and absorption/emission of a photon ω2 This transition rate is to be interpreted as follows. which is stationary and energy is conserved. i. describes processes which lead to ﬁnal states |n with energy n = o + 2 ω1 and. accordingly. namely absorption/ emission ﬁrst of photon ω2 and then absorption of photon ω1 and. vice versa. . the third term in (8. m = o + ω1 ) and the evolution of state |m is described by a factor 1/( m − o − ω1 −i λ). it must hold n = o + 2 ω1 . each of energy ω1 .e. describe the absorption/emission of two photons. ﬁrst absorption of photon ω1 and then absorption/ emission of photon ω2 .. The remaining two contributions in (8.169) pert. The ﬁrst term.. accordingly. In case of the ﬁrst term in (8. (8. promotes the system then to the state |n . i.8. Similarly.e.. describe the absorption of two photons.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..

179) . (8. λ → 0+ . φ) a3 o 1s (8.m (r. φ) a3 ao o 2p (8. m denote the relevant quantum numbers) ψn=1. and the integral is =1. The absorption rate. φ) × × eik·r e−r/ao Y00 (θ. φ) = 6 r −r/2ao e Y1m (θ.175) The matrix element involves a spatial integral over the electronic wave functions associated with states |n and |0 . (8. . θ.176) ψn=2. is then kabs = Dipole Approximation 2π e2 2πN m2 ωV e ˆ u · n| p eik·r |0 ˆ 2 ˆ ˆ V1 exp(−iωt) + V2 exp(+iωt) eλt . 8.171) The perturbation on an atomic electron system is then according to (8.e. plane polarized wave described through the complex vector potential A(r.226 Absorption of a Plane Polarized Wave Interaction of Radiation with Matter We consider ﬁrst the case of absorption of a monochromatic.178) These wave functions make signiﬁcant contributions to this integral only for r-values in the range r < 10 ao .m=0 (r. the second term can be discounted in case of absorption. For example. t) = 8πN ωV u exp ˆ ı (k · r − ωt) . φ) = 2 − 1 2 1 −r/ao e Y00 (θ.. θ.2 = m ωV Only the ﬁrst term of (8. in this range one can expand eik·r ≈ 1 + ik · r + . in case of a radiative transition from the 1s state of hydrogen to one of its three 2p states. (8. the wave functions are (n.144) VS = where e 2πN ˆ ˆ p · u e±ik·r . the vector potential actually assumed is A(r. =0.174) We seek to evaluate the matrix element ˆ M = n| p eik·r |0 .173) V1. . i.143) will contribute to the absorption process. to → −∞ (8.158).170) We will employ only the real part of this potential. t) = 2πN ωV u exp ˆ ı (k · r − ωt) + 2πN ωV u exp ˆ ı (−k · r + ωt) . ˆ (8. . However.177) √ M = 6 0 ∞ 1 2π ia4 o r2 dr −1 dcosθ 0 ∗ dφ r e−r/2ao Y1m (θ.143. (8. according to (8. φ) (8.172) δ( n − o − ω) (8.

(8.k=1 i m 3 pk ek = ˆ j. V (r) ] 2me =0 = Using r = 3 ˆ j=1 xj ej 1 2me 3 pk [ r. 2me r (8. Ho ] = i m 3 pk ej δjk = ˆ j.182).183) (8.174) yields kabs = 4π 2 e2 N ω V u · n| ˆ |0 ˆ r 2 m( o − m n| [r. V (r) = − .180) ∆E2p−1s and ao = 0. This simpliﬁcation results from the identity m ˆ p = [ r.178) holds eik·r ≈ A 1 1 + O( 50 ) such that one can approximate eik·r ≈ 1 .7: One-Photon Emission and Absorption 227 One can estimate that the absolute magnitude of the second term in (8. pk ] + ˆ ˆ k=1 1 2me 3 [ r. Using (8. the value of the wave length for the 1s → 2p transition 2π c λ = = 1216 ˚ A (8. in case of the hydrogen atom.181) and (8. One refers to this approximation as the dipole approximation. Transition Dipole Moment A further simpliﬁcation of the matrix element (8.182) one obtains M ≈ Insertion into (8.k=1 i ˆ p m (8.182) one ﬁnds [ r. is Ho = ˆ (p)2 e2 + V (r) .104) and.186) δ( n − o − ω) (8.181) For the commutator in (8.529 ˚ one concludes that in the signiﬁcant integration range in (8.184) and the commutation property [xk .187) .179) and other terms are never larger than 20π ao /λ.175). Ho ] |0 = i i n) n|r |0 . pk ] pk ˆ k=1 (8.8. Ho ] = ˆ2 p [ r.182) (8. ] + [ r. pj ] = i δkj one obtains ˆ [ r. Using |k| = 2π/λ.185) from which follows (8.175) can then be ˆ achieved and the diﬀerential operator p = i replaced by by the simpler multiplicative operator r. We are now in a position to obtain an alternative expression for the matrix element (8. Ho ] i where Ho is the Hamiltonian given by (8.

174) one can replace n − o by ω. hence. three-dimensional vector. will actually be replaced by a distribution function which reﬂects (1) the ﬁnite life time of the states |n .33) are orthogonal to each other. in practical situations.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. one can express | u · n| r |0 |2 = |ρ|2 ( cos2 α + cos2 β ) = sin2 θ . x2 . such that the thermal radiation present supplies a continuum of frequencies. k frame is described by the angles ϑ. ˆ u ˆ (8.193) . |0 . and (2) the fact that strictly monochromatic radiation cannot be prepared such that any radiation source provides radiation with a frequency distribution. Integrating and summing accordingly over all contributions as given by (8. the direction of u1 is described by the ˆ ˆ ˆ ˆ angles ϑ1 = π/2.. and all polarizations of the radiation. We can. there are ony two linearly independent directions of u possible. i.191) (8. planar waves as deﬁned in (8. ϕ1 = 0 and of u2 by ϑ2 = π/2. ϕ. u2 and k ˆ ˆ coordinate system. The direction of this vector in the u1 . ˆ In order to carry out the integral dk we note that u describes the possible polarizations of the ˆ ˆ and u. ρ) and ˆ u β = ∠(ˆ2 .188) ˆ Here dk is the integral over all orientations of k. 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. Accordingly. say u1 and u2 .158) above that in ﬁrst order the contributions of all components of the radiation ﬁeld add.31–8. directions. k ˆ As a result. according to (8.190) (8. For the two angles α = ∠(ˆ1 . 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. u2 .189) where the factor 1/ arose from the integral over the δ-function. x3 -axes of a right-handed cartesian vectors u1 . We have demonstrated in our derivation of the rate of one-photon processes (8. ϕ2 = π/2.e.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 (8.35). The δ-function appearing in this expression. obtain the transition rate in the present case by adding the individual transition rates of all planar waves present in the oven. The unit ˆ ˆ ˆ ˆ can be chosen to point along the x1 . ρ) holds then u cosα = cosϑ1 cosϑ + sinϑ1 sinϑ cos(ϕ1 − ϕ) = sinϑcosϕ and cosβ = cosϑ2 cosϑ + sinϑ2 sinϑ cos(ϕ2 − ϕ) = sinϑsinϕ . Absorption of Thermal Radiation We want to assume now that the hydrogen atom is placed in an evironment which is suﬃciently hot. a very hot oven.228 Interaction of Radiation with Matter where we used the fact that due to the δ-function factor in (8.

that emission occurs even if no photon is present in the environment.. This poses. in agreement with observations. In case of a classical radiation ﬁeld one would expect that emission cannot occur.8.e. ˆ In case of emission only the second term V2 exp(+iωt) in (8. Otherwise. The radiation ﬁeld is described. for atoms or symmetric molecules.195) = Nω 1.173) contributes.196) The last two factors in (8. there is also a contribution to the emission rate which is proportional to Nω . The quantum nature of radiation manifests itself in that the number of photons Nω msut be an integer. Emission of Radiation We now consider the rate of emission of a photon. Nω = 0. 2 . . Transition Dipole Moment The expression (8. The most intensely absorbing molecules are long.194) combined are typically somewhat smaller than (1 ˚/1000 ˚)3 = A A 10−9 . . by planar waves with vector potential (8. ﬁnally. the resulting total rate of emission bears a diﬀerent dependence on the number of photons present in the environment. the calculation of the emission rate proceeds as in the case of absorption. the absorption rate is of the order of 109 s−1 or 1/nanosecond. Nω photons before absorption.. For practical evaluations we provide an expression which eliminates the physical constants and allows one to determine numerical values readily. (8. For this purpose we use ω/c = 2π/λ and obtain 4 e2 ω 3 32π 3 e2 ao 1 ao = = 1.197) o (tot) (8.194) For absorption processes involving the electronic degrees of freedom of atoms and molecules this radiation rate is typicaly of the order of 109 s−1 . (8. 1.e.37 × 1019 × 3 3 c3 3 ao λ3 s λ and kabs where λ = 2πc n − (8. However. However.7: One-Photon Emission and Absorption 229 This geometrical average. The value of | n| r |0 | determines the strength of an optical transition. This diﬀerence between emission and absorption is due to the quantum nature of the radiation ﬁeld.173). Accordingly.172. can be inserted into (8. linear molecules. i. This dependence predicts. 8.171) and perturbation (8. a quantum mechanical treatment of the radiation ﬁeld leads to a total emission rate which is proportional to Nω + 1 where Nω is the number of photons before emission.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 |. in particular.189) to yield the total absorption rate kabs (tot) = Nω 4 e2 ω 3 | n|r|0 |2 3 c3 . as for the absorption process.. The corresponding process is termed spontaneous emission. i. However.37 × 1019 1 ao | n|r|0 |2 × s λ λ2 . The transition dipole moment can vanish for many transitions between stationary states of a quantum system. however. a problem in case of emission by quantum systems in complete darkness. for Nω = 0. .

respectively. it must hold Nω ksp N0 = (Nω + 1) ksp Nn (8.g. the well-known Planck radiation formula. in lasers.201) ksp = 3 c3 the rates of absorption and emission are Nω ksp and (Nω + 1)ksp .198) to determine the stationary distribution of photons ω in an oven of temperature T .203) i. Hence.198) is consistent with Planck’s radiation law which reﬂects the (boson) quantum nature of the radiation ﬁeld. Deﬁning the rate of spontaneous emission 4 e2 ω 3 | n|r|0 |2 (8.230 Interaction of Radiation with Matter which is termed induced emission since it can be induced through radiation provided. = (8. .e. 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. exp[− ω/kB T ] = This equation yields Nω = 1 .200) where kB is the Boltzmann constant. respectively. To demonstrate this property we apply the transition rates (8.202) It follows.. exp[ ω/kB T ] − 1 (8.199) Planck’s Radiation Law The postulate of the Nω + 1 dependence of the rate of emission as given in (8. For these numbers holds Nn / No = exp[−( n − o )/kB T ] (8. e. 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.8 Two-Photon Processes In many important processes induced by interactions between radiation and matter two or more photons participate.204) Nω . Let No and Nn denote the number of atoms in state |0 and |n ..200). Nω + 1 (8.198) (8. In the following we discuss several examples. using (8.195) and (8. 8. The number of atoms undergoing absorption in unit time are Nω ksp No and undergoing emission are (Nω +1)ksp Nn . accordingly. 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 . We assume n − o = ω. The total rate of emission.

207) one is.106). which is not necessarily the case. The transition requires that the transition dipole moment n| r |0 does not vanish for two states |0 and |n . even if n| r |0 vanishes.182) yields. however.187). faced with the dilemma of having to sum over all intermediate states |m of the system. in case ˆ of absorption only V1 contributes.207) Expression (8. Scattering of Photons at Electrons – Kramers-Heisenberg Cross Section We consider in the following the scattering of a photon at an electron governed by the Hamiltonian Ho as given in (8. 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) . In applying (8. but then requires the absorption of two photons.207) does not converge rapidly. . We wish to describe an observation in which a detector is placed at a solid angle element dΩ2 = sinθ2 dθ2 dφ2 with respect to the origin of the coordinate system in which the electron is described. (8. t) = Ao1 u1 cos(k1 · r − ω1 t) . (8. dropping the index 1 characterizing the radiation.208) has been prepared. However. We assume that a planar wave with wave vector k1 and polarization u1 .172.8. k = 2π e2 2πNω m2 ωV e × δ( m 2 ∞ m=0 n|ˆ · p eik·r |m m|ˆ · p eik·r |0 u ˆ u ˆ m − o − ω1 − i λ 2 × (8. Employing the dipole approximation (8.181) and using (8. 8. however. ˆ (8. under ordinary circumstances. The intense radiation ﬁelds of lasers allow one to increase transition rates to levels which can readily be observed in the laboratory. induce single photon absorption processes as described by the transition rate Eq. 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. In this case one needs to choose the energy of the photons to obey n = o + 2 ω. (8. The resulting rate of the transition depends on 2 Nω . If the sum in (8. One obtains.168) where the ﬁrst term describes the relevant contribution.207) does not provide a suitable avenue of computing the rates of 2-photon transitions. then expression (8.173).104) with stationary states |n deﬁned through (8.205) The respective radiative transition is of 2nd order as described by the transition rate (8. a transition between the states |0 and |n may be possible.206) − o − 2 ω) .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. The electron absorbs the radiation and emits immediately a second photon. as described through the vector potential ˆ A(r. ﬁnally.8: Two-Photon Processes Two-Photon Absorption 231 The interaction of electrons with radiation.

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 .213) where N2 /V is the density of photons characterized through k2 .. The second term in (8.211) is stated in Eq. u2 .210).141). The relevant terms of the perturbation (8. We know already from our description in Section 8. the wave characterized through k1 . u1 .209) cosθ2 where the value of k2 has been ﬁxed.211) is o2 A− = o2 8π (N2 + 1) ω1 V (8.210) is accounted for by the following contributions of (8.212) where N1 /V is the density of photons for the wave described by (8. This expansion yields.210).210) The vector potential which describes both incoming wave and outgoing wave is a superposition of the potentials in (8. (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. In the present case we consider only scattering processes which absorb radiation corresponding to the vector potential (8.211) are given by VS (t) = e ˆ p · A+ u1 exp[i(k1 · r − ω1 t)] + A− u2 exp[−i(k2 · r − ω2 t)] o1 ˆ o2 ˆ 2me contributes in 2nd order + e2 4me A+ A− u1 · u2 exp{i[(k1 − k2 ) · r − (ω1 − ω2 ) t]} ˆ o1 o2 ˆ contributes in 1st order The eﬀect of the perturbation on the state of the electronic system is as stated in the perturbation expansion (8. The vector potential describing the emitted plane wave is then A(r. 8.e. the amplititude A− deﬁned in (8.208. the amplitude A+ is given by o1 A+ = o1 8πN1 ω1 V (8. however. 8.208.6 above that the absorption of the radiation in (8. ˆ The perturbation which arises due to the vector potential (8.211) accounts for the emitted wave and. o1 ˆ o2 ˆ (8. in the present case.210) A(r. ˆ (8. according to the ˆ description of emission processes on page 229. for the components of the wave (8. hence. t) = Ao2 u2 cos(k2 · r − ω2 t) . i.214) . later we will allow the quantum system to select appropriate values.208) and the emission of the radiation in (8.105). t) = A+ u1 exp[ i (k1 · r − ω1 t) ] + A− u2 exp[ i (k2 · r − ω2 t) ] .208) and emit radiation corresponding to the vector potential (8.211) The ﬁrst term describes the absorption of a photon and.208).105) using the vector potantial (8.

.216) over all available modes of the ﬁeld.e. Hence. me c2 (8. evaluating the time integrals in (8. one needs to integrate the rate as given by (8.212. only these two terms contribute to the scattering amplitude.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: 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. Inserting also the values (8.215) and taking the limits limto →−∞ and limλ→0+ yields the transition rate k − m = 2π δ( n − o − ω1 + ω2 ) e2 A+ A− u1 · u2 n|0 ˆ ˆ 4m2 o1 o2 e 2 (8.216) 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 .217) 2 Here ro denotes the classical electron radius ro = e2 = 2. i. 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 222–225.218) . i. one needs to carry out the integration 2 V(2π)−3 k2 dk2 · · ·. Accordingly.8 · 10−15 m .8..181) in stating this result. Only the second (1st order) and the third (2nd order) terms in (8.215) correspond to scattering processes in which the radiation ﬁeld ‘looses’ a photon ω1 and ‘gains’ a photon ω2 .e.

221) suggests to expand S( ω) S( ω) = S(0) + S (0) ω + Using 1 − o ± ω one can readily determine m (8. consequently. . in fact. 2 ..227) u ˆ 0|ˆ1 · p |m m|ˆ2 · p |0 u ˆ u ˆ 0|ˆ2 · p |m m|ˆ1 · p |0 u ˆ + me ( m − o )3 me ( m − o )3 (8. it follows 2 dσ = ro (N2 + 1) dΩ2 |ˆ1 · u2 − S( ω) |2 u ˆ (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..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. − o )2 ( m − o )3 (8. ω1 = ω2 . a process in which the electronic state remains unaltered after the scattering.. Accordingly.226) u ˆ 0|ˆ2 · p |m m|ˆ1 · p |0 u ˆ u ˆ 0|ˆ1 · p |m m|ˆ2 · p |0 u ˆ − me ( m − o )2 me ( m − o )2 (8.228) S (0) = m S (0) = 2 m . i. (8.221) Using |n = |0 and.234 Interaction of Radiation with Matter The factor N1 c/V can be interpreted as the ﬂux of incoming photons. Rayleigh Scattering We turn ﬁrst to an example of so-called elastic scattering. one can relate (8. .220) n|ˆ2 · p |m m|ˆ1 · p |0 u ˆ u ˆ + m − o − ω1 2 − − o + ω2 In the following we want to consider various applications of this formula.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. for all states |m of the electronic system (8.223) m Condition (8.224) = m 1 − o ( m ω ( ω)2 + + . one assumes the even stronger condition ω1 << | o − m| .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 + .219) (8.e.

229) both for the u1 · p and the to u2 · p terms in (8.226). ˆ ˆ We want to prove now that expression (8.230) According to (8.234) where we used for the second identity the fact that u1 ·r and u2 ·r commute.108) this is 1 0|ˆ1 ·r u2 · p − u2 · p u1 ·r |0 .k=1 j.182) for p and the expression (8.235) We can now combine eqs. .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.234. Using again (8.222).232. Expression (8.182) 0|ˆ1 · p |m u ˆ 1 = 0|ˆ1 ·r |m u me ( m − o ) i This transforms (8. ˆ We want to simplify ﬁrst (8.235) and obtain the leading contribution to the expression (8.e.222) of the cross section for Rayleigh scattering dσ = 2 ro m2 ω 4 (N2 + 1) dΩ2 × e (8. For this purpose we replace p using (8. It explains. u1 ·r ] |0 = 0 u ˆ (8. for example. m|ˆ1 · p |0 u ˆ 1 = − m|ˆ1 ·r |0 u me ( m − o ) i (8.233) m Employing again (8.228) provides then the ﬁrst non-vanishing contribution to the scattering cross section (8.232) Obviously. For this purpose we apply (8. 8.8: Two-Photon Processes 235 ˆ These three expressions can be simpliﬁed using the expression (8.. pk ]|0 = u u ˆ j.226) into S(0) = 1 i ( 0|ˆ1 ·r |m m|ˆ2 · p |0 − 0|ˆ2 · p |m m|ˆ1 ·r |0 ) u u ˆ u ˆ u m . 8.228) 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.k=1 (ˆ1 )j (ˆ2 )k δjk = u1 · u2 u u ˆ ˆ (8. i. as a compass.222).108) for the identity operator.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 ) . pk ] = i δjk yields ﬁnally ˆ S(0) = 1 S(0) = i 3 3 (8. ˆ ˆ S (0) given in (8.108) yields S (0) = me 2 0| [ˆ2 ·r.224. this term cancels the u1 · u2 term in (8.236) is of great practical importance. u u u u (8. honey bees. u ˆ ˆ ˆ ˆˆ i The commutator property [xj . (8.231) (ˆ1 )j (ˆ2 )k 0|[xj . the blue color of the sky and the polarization pattern in the sky which serves many animals.229) (8.8.227) vanishes. 8.

. We want to assume. u ˆ 1 ao . hence. i.220). The ﬁrst choice is ˆ sinφ2 k2 × k1 (1) u2 = ˆ = −cosφ2 (8.e.236 Thomson Scattering Interaction of Radiation with Matter We consider again elastic scattering..243) cos2 θ2 cos2 φ2 for u1 = u1 . u2 = u2 ˆ ˆ ˆ ˆ (1) (2) (2) (1) (1) (2) (1) (2) (2) (1) needs to be (8.244) . the emitted u ˆ radiation is decribed by the wave vector sinθ2 cosφ2 k2 = k1 sinθ2 sinφ2 (8. u2 = u2 ˆ ˆ ˆ ˆ (8.240) cosθ2 We choose for the polarization of the incoming radiation the directions along the x1 . ao Bohr radius . u2 = u2 ˆ ˆ ˆ ˆ for u1 = u1 . |n = |0 and ω1 = ω2 = ω in (8. 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 .241) 0 0 Similarly.242) |k2 × k1 | 0 where the second identity follows readily from k1 = e3 and from (8.239) We will show below that this expression decribes the non-relativistic limit of Compton scattering. 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 . though. however. that the dipole approximation is still valid which restricts the applicability of the following derivation to k1 >> One obtains immediately from (8. To evaluate |ˆ1 · u2 |2 we assume that k1 is oriented along the x3 -axis and. assume now that the scattered radiation has very short wave length such that ω >> | o − m| . u1 = 1 ˆ ˆ (8.237) The resulting scattering is called Thomson scattering.220) 2 dσ = ro (N2 + 1) dΩ2 |ˆ1 · u2 |2 . for all states |m of the electronic system . u2 = u2 ˆ ˆ ˆ ˆ cos2 φ2 cos2 θ2 sin2 φ2 for u1 = u1 .240). (8. (8.and the x2 -axes 1 0 (1) (2) u1 = 0 .238) (8.

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

ω1 ≈ m − o for a particular state |m .251) ω1 − ( We deﬁne x · R · y = j.250) (8.k xj Rjk yk . Of course.g. In case that the incoming photon energy ω1 is chosen to match one of the electronic excitations. a case called resonant Raman scattering. a property which can be exploited to selectively probe suitable molecules of low concentration in bulk matter. the cross section for resonant Raman scattering can be several orders of magnitude larger than that of ordinary Raman scattering. Nevertheless. the Raman scattering cross section will be much enhanced. .. e.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 .

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

. x2 . We can then state P2 = T (4. A reader interested in the quantum mechanical description of N particle systems is strongly encouraged to study these representations. 4) T (4. but in a diﬀerent order. The latter factorization has the essential property that the number of factors is either even or odd. P2 = . is called a transposition. k) are characterized by the two indices (j.2).3) the j-th index in the second row. However. in notation (9. and explicitly the fact that any P ∈ SN can be given as a product of transpositions T (j. . the representations in the space of N –particle wave functions are extremely useful in dealing with N –particle systems. 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. Transpositions denoted by T (j. (9.5) . . . particles 4 and 5. rather only states which obey a certain transformation property under permutations are acceptable.. xP (N ) ) .240 Many–Particle Systems repeatedly like degenerate states. not any state can be chosen as to represent the many– particle system. . 3) T (7. i. . 5) in case of P2 . . xN ) = Ψ(xP (1) . 5) T (5. xj . one can state P Ψ(x1 . . . . P2 = T (4. and the second row showing the numbers 1 to N again. xP (2) . . 8) .. even though these properties. . One calls permutations of the ﬁrst type even permutations and permutations of the second type odd permutations. . Denoting by P (j) the image of j. the top row representing the numbers 1 to N labelling the N particles under consideration. i. 5) (9. This property will be established now. (j.3) P2 which aﬀects only two particles. (9. The elements P ∈ SN can be represented conveniently as 2 × N matrices. 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. . The permutations we need to consider for a system of N particles are the elements of the group SN .e.e. . k) of the two particles transposed. . 5). in particular.e. namely implicitly the group property. . xP (j) . For the permutations P1 and P2 given above holds P1 = T (3. i. . The group SN is then composed of two disjunct classes of even and of odd permutations. 6) T (6. k) = (4. a given P can either only be presented by even numbers of transpositions or by odd numbers of transpositions.. . 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 = . number k under the entry j of the ﬁrst row indicating that particle j is switched with particle k..4) both permutations being obviously odd. i. the set of all permutations of N objects.e. We will not discuss here at any depth the properties of the permutation groups SN . k). For the following we will require only two properties of SN . leaving all other particles unchanged. .

xk ) is symmetric in its two arguments. i. xk ) j. The ﬁrst case applies to bosons. −1. . for the scalar product holds j|k = δj k .k=1 (9.e.. The single particle states are assumed to be orthonormal. In fact. or the corresponding group of two elements 1. P ] = 0 . the Hamiltonian for permuted indices P HP −1 is identical to H. Fock-Space Our derivation will start by placing each particle in a set of S orthonormal single particle states x|k = φk (x).e. k = 1.9) = 1 for P even −1 for P odd (9.7) This property. {1} (trivial) and {1. G(xj . S where S denotes the number of available single particle states. 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 . 2.e. implies that the permutation operators can be chosen in a diagonal representation.e. In the latter case all even permutations are represented by ‘1’ and all odd permutations by ‘-1’. i.10) We will construct now wave functions which exhibit the appropriate properties. Obviously. the second group to fermions. all particles being governed by the same interactions.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. i. provide a representation of the group structure represented by the multiplication table of even and odd permutations given above.5) implies that a permutation eﬀectively changes the indices of the particles) and. From the latter results [H. From this property follows that H is independent of any speciﬁc assignment of indices to the particles (note that (9.6) which describes one–particle interactions and two–particle interactions of the system. (9. a diagonal representation can only berealized in a simpler group. Presently. it is essential that the functions F ( ) and G( . it is a postulate of quantum mechanics that for any description of many–particle systems the permutation operators must be diagonal. (9. ) are the same for all particles and for all pairs of particles.e. usually a very large number. in turn.11) . In fact.. Since permutations do not necessarily commute. the representation is either isomorphic to the group composed of only the ‘1’. the Hamiltonian must be of the type N H = j=1 1 F (xj ) + 2 N G(xj . both groups..8) |Ψ (9. In fact. . The terms of the Hamiltonian will be discussed further below. i. (9. 1 together with multiplication.. i.9. particles with integer spin values. particles with half–integer spin values. ∀P ∈ SN . hence. .-1}.

.. .12). . . . λ2 . The reader is well advised to pause and grasp the deﬁnition of the nj . . the reason is that such transposition duplicates the state which would. 2.13) These states form an orthonormal basis of N -particle states. n2 . λ2 . xN |Ψ Fock (λ1 . .. nS ) = S j=1 (9. i. k. 9. Λ(λ1 . λ2 . with λj = λk = . 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. . therefore. . .λN ) P x1 . . . . .15) is that the sum over permutations does not involve permutations among indices j. S} of the type (9.g. x2 . . . λN )|ΨFock (λ1 . This detail is connected with the fact that a two–particle state of the type ΨFock (λ. . λ2 . . . . are essential.11.9). . i. the states |ΨFock (λ1 . .λN ) on the rhs. .. xN |ΨFock (λ1 . . λN ) ⊂ SN is the set of all permutations involving only particles j with diﬀerent λj . In a second step we form linear combinations of Fock states with the desired symmetries. . . . . λN ) = j=1 φλj (xj ) . which do not yet obey the symmetries (9.15) with λk = j. An important detail of the deﬁnition (9. the nj specify how often a single particle state |j is occupied.. .15). . . A wave function which obeys the boson symmetry (9. e. . each particle j into a speciﬁc state |λj . . . appear severalfold in (9. x2 . . .8. . referred to as the occupation numbers. . . hence.e.. characterize the wave function (9. xN |ΨB (n1 . 9.12) where the second factor represents the scalar product between spin states. .λN ) Here the states (λ1 . . . . . . .9). k) ∈ Λ(λ1 . .15) (nj !) N! 1 2 P ∈Λ(λ1 . (9. The integers nj are equal to the number of λk in (9.14) Obviously. . nS ).e. . is a particular choice of placing N particles consistent with the occupation numbers (n1 . It holds N ΨFock (λ1 . N x1 . λN ) do not exhibit the symmetries dictated by (9.15) completely and. (9. λj ∈ {1. λ) in which two particles occupy the same single– particle state |λ does not allow transposition of particles. 9. T (j. x2 . The numbers nj .8. |λN . .8) is x1 . . The Fock states represent N particles which are placed into S states |λ1 . . λN ) = j=1 δλj λj .. .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. n2 . . . . . . the so-called Fock-states. λ2 . . |λ2 . λ2 . . . We ﬁrst consider orthonormal many–particle wave functions. . λN ) only if λj = λk .

These states form an orthonormal basis. . i. .. . . (9. φλN (x1 ) φλN (x2 ) . . . . n2 . . . nS ). a molecule or a crystal. the integers nj denote the occupancy of the single–particle state |j . .17) describe. For any value nj > 1 two or more of the columns of the Slater matrix are identical and the wave function vanishes. . . . nS ) = φλ1 (x1 ) φλ1 (x2 ) . Wave Functions with Fermion Symmetry in the Occupation Number Representation One can construct similarly a wave function which obeys the fermion symmetry property (9. . . λ2 . nS ) and N = (n1 . xN |ΨFock (λ1 .1: (a) Show that the states (9. . x2 .9. . . nS ) and N = (n1 . for example. . .19) φλ1 (xN ) φλ2 (xN ) . . . . . nS ) holds S Ψ (N )|Ψ (N ) = j=1 B B δnj nj . . . xN |ΨF (n1 . . . . . . .. n2 . .15) describe. n2 . . .15) obey the boson symmetry (9. for two sets of occupation numbers N = (n1 . . x2 .9). . . .λN ) on the rhs. nS ) = √1 N! P ∈SN P P x1 . for two sets of occupation numbers N = (n1 .17) Here the states (λ1 . .e.. φλN (xN ) Here. . . nS ) holds S ΨF (N )|ΨF (N ) = j=1 δnj nj .15) holds (9.16). .20) .. i. φλ2 (x1 ) φλ2 (x2 ) . xN |ΨF (n1 . (b) Show that for the states (9..1: Bosons and Fermions 243 The wave functions (9. . . . x2 .18) this wave function can also be expressed through the so-called Slater determinant x1 . . . . These states form an orthonormal basis.e. √1 N! (9. The fermion states (9. λ2 . . . . n2 . n2 . . for example. n2 . the photons of the electromagnetic ﬁeld or the phonons in a crystal. . electrons in an atom. . (9. .8). The important property holds that nj can only assume the two values nj = 0 or nj = 1. Using the identity governing the determinant of N × N –matrices A N det(A) = P ∈SN P j=1 Aj P (j) (9. correspond to a particular choice of placing N particles consistent with the occupation numbers (n1 . .1. n2 ..λN ) (9. . Such wave function is x1 .16) Exercise 9.

. nj + 1. guaranteeing the correct permutation properties of the many–particle states. n2 . .23) (9. .22) The operators aj and a† obey the commutation properties j [aj . . n − 1. are the so-called operators of 2nd quantization. . nS ) = j nj + 1 ΨB (n1 .21) we consider the matrix Aj corresponding to aj in the basis of many–particle states (9. . . .2 Operators of 2nd Quantization An important tool to describe mathematically systems of many bosons or fermions.25) . . . nS ) . . . . nS ) = nj ΨB (n1 . . (b) Show that the states (9.24) [aj . (9. . a† ] = δj k . . . . . . nS ) nj ≥ 1 0 nj = 0 √ .15).16) one obtains (Aj )N = ΨB (N )| aj |ΨB (N ) = √ S N nj δnj nj −1 k=1 k=j δnk nk . . . n ) and the N − 1-particle wave function ΨB (n . . . .1. . . . . . Creation and Annihilation Operators for Bosons We consider the operator aj deﬁned through aj ΨB (n1 . To prove that (9.22) follows from (9. . . [a† . . nj − 1.17) obey the orthonormality property (9.244 Many–Particle Systems The representation of wave functions (9. n ) are normalΨ 1 j 1 j S S ized according to (9. . .15) and (9.2: (a) Show that the states (9. These operators allow one to construct and manipulate many–particle wave functions while preserving permutation symmetry. .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 . . . For the adjoint operator a† holds j a† ΨB (n1 . . nS ). nj . nj . (9. Exercise 9. . since both the N -particle wave function B (n . . . . . a† ] = 0 j k (9.21) √ The factor nj appears here on the rhs. .15). .17) obey the fermion symmetry (9. . . (9. . . 9.20). . n . Employing the superindices N and N introduced above and using the orthonormality property (9. ak ] = 0 .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 . .

. . The related operator S ˆ N = j=1 ˆ Nj (9. obviously. The creation operators a† allow one to construct boson states ΨB (N ) from the vacuum state j |0 = |ΨB (n1 = 0. .e. S ˆ N |ΨB (n1 . (9. j (9.34) = . . .33) is called the particle number operator since.29) yield the commutation relationship (9. n2 .. . . One can readily show using (9. .21.21) and vice versa. i.27. nk + 1 . n2 .26) From this result one can immediately conclude that (9. . . j j Since this holds for any |Ψ(N ) it follows [aj . .23) follow in an analogous way. 9. n2 . nS ) = j=1 a† j nj nj ! |0 . .30) Of particular interest is the operator ˆ Nj = a† aj . (9.24). nS ) N |ΨB (n1 . .22) follows from (9.2: Operators of 2nd Quantization Let A† be the matrix adjoint to Aj . a† ] = 1 For j = k follows similarly 1.28) (9. .29) |ΨB (n1 .22) ˆ Nj |ΨB (N ) = nj |ΨB (N ) . . . the eigenvalues of Nj are the occupation numbers nj .9. n2 . .24). for basis states |ΨB (n1 . the properties (9. . It holds for j = k (aj a† − a† aj )|Ψ(N ) = (nj + 1 − nj )|Ψ(N ) = |Ψ(N ) . n2 = 0. . . . . (9. . We will prove the commutation properties (9.32) ˆ ˆ i. . (9. n2 . nS = 0) as follows S (9. 9. nS ) = 0 . One obtains for its matrix elements j A† j N N 245 = ΨB (N )| a† |ΨB (N ) = (Aj )N N j S k=1 k=j nj δnj nj −1 δnk nk = nj + 1 δnj nj +1 S k=1 k=j δnk nk . nS ) = |ΨB (N ) . nS ) . . . j (aj a† − a† aj )|Ψ(N ) = k k ( nj (nk + 1) − (nk + 1)nj )|Ψ(n1 .27) Equations (9. . . One refers to Nj as the occupation number operator.e. .. . nj − 1 .31) This operator is diagonal in the occupation number representation. nS ) = j=1 nj |ΨB (n1 . (9. .

21. . x1 . . . . . j Exercise 9. nS ) = . 9.. . . one does not need to worry ever about proper symmetries of wave functions.. xN |ΨF (n 1 . . . it also holds for nj + 1. .24) and aj |0 = 0 imply that the the properties (9. . .9). 9. . one needs to apply only the commutation propertes (9. 9. . nS ) . x2 .35) and where the derivative operation is deﬁned through ∂f (A) = ∂A ∞ n=1 1 f (n) (0) An−1 . nj+1 . . .e.e.2: Show that the boson operators a† and aj satisfy j [aj .17. i. nS ) = − x1 . nj+1 . e. . . nj+1 . 9.19) without aﬀecting the fermion symmetry (9. since they are induced through the algebra of a† and aj .30).23. xN |ΨF (n1 . 9. . from the vacuum state |0 or states |ΨB (N ) . x2 . operators which can alter the occupancy of the wave function (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 ..36) Creation and Annihilation Operators for Fermions We want to consider now creation and annihilation operators for fermions.22. .23) hold for the state deﬁned through (9. nj (9. i.23. (9. As long as one starts from a wave function with the proper boson symmetry. nj .2. √1 N! .. . To prove this property we notice that the l. φj (xN ) φj+1 (xN ) .1: The commutation relationships (9. . φj+1 (x1 ) φj+1 (x2 ) . .g. . 9. .. . . Prove this by induction. . in general. . nj .37) (9. .. . . in case that the single particle states |j and |j + 1 are both occupied. nj+1 = 1.h.. . .38) . Exercise 9. . .24) and the property aj |0 = 0.s. in case it holds for nj .. x2 .38) corresponds to x1 . of (9.246 Many–Particle Systems In using the boson creation and annihilation operators.. f (a† )] j [a† . . .39) . φj (x1 ) φj (x2 ) . . the fermion wave function changes sign when one exchanges the order of the occupancy. .2.. xN |ΨF (n1 . .29. 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. . . showing that the property holds for nj = 0 and. (n − 1)! (9.. Obviously.

.38) reads − x1 . .. . . n5 = 1. . We will be adopting below the notational convention that qj corresponds to occupancies N = (n1 . nS ) = . We are now ready to deﬁne the annihilation operator cj for fermions in the single–particle state |j as follows cj |ΨF (n1 . the expressions (9. . φj+1 (xN ) φj (xN ) .s. .h.. at the ﬁrst column of the Slater determinant (9. . .. it is necessary to specify for a fermion wave function the order in which the single– particle states |λj are occupied. . . One adopts the convention that particles are created and annihilated by these operators always at the ﬁrst position of the wave function.39) and (9. . that occupancies are always ordered according to a monotonous increase of the single–particle state index.2: Operators of 2nd Quantization The r. n2 . φj (x1 ) φj (x2 ) . . . This requires one.19). . particles are being created or annihilated.e.19). . .. nj+1 . (9. . Obviously. . 247 − √1 ! N .. at which column of the Slater determinant (9. . . . nj = 0.41) Before we consider the deﬁnition of fermion creation and annihilation operators we need to take notice of the fact that one also needs to deﬁne at which position in the wave function. nj → 0. nj . φj+1 (x1 ) φj+1 (x2 ) . . . (9. n2 . qj is the number of states |k with k < j which are occupied in a fermion wave function.. nj−1 . xN |ΨF (n1 . φ2 (x2 ) φ5 (x2 ) (9. . Because of the property of the determinant to change sign when two columns are interchanged.40) are identical. nj+1 . . in order to be consistent with this convention. . n2 = 1. . nj . A proper example is the two particle fermion wave function x1 . 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. n3 = 0. . . to move the particle to the ﬁrst position (to be annihilated there) or to move it from the ﬁrst position to its canonical position (after it had been created at the ﬁrst position). . . . x2 |ΨF (n1 = 0.e. . n2 . .e. vanishes. . n6 = n7 = . . . i.9. of (9. ....40) . .. .e. i. . nS ) = nj (−1)qj |ΨF ( n1 . n3 . . This change of position is connected with a possible sign change (−1)qj where qj is deﬁned for a given N = (n1 .. .. . however. n2 . nS = 0) φ2 (x1 ) φ5 (x1 ) 1 1 = √2 = √2 [φ2 (x1 )φ5 (x2 ) − φ5 (x1 )φ2 (x2 )] . For this purpose one must adhere to a strict convention: the labelling of single–particle states by indices j = 1.43) . nS ) as follows: j−1 qj = k=1 nk . (9. nS ) qj states occupied In case that the single particle state |j is not occupied. 2. nS ). the rhs.42) i. i. n4 = 0. x2 . .

. nS ) = j j (−1)qj (1 − nj + 1)(−1)qj nj |ΨF (N ) = nj ΨF (N ) (9. 9. It holds c† cj |ΨF (N ) = c† (−1)qj nj |ΨF (n1 . nj → 1.44). . (9.49) The derivation involves realization of the fact that a non-zero result is obtained only in case nj = 1. the last factor on the rhs. ck ]+ = 0 . otherwise. n3 . The anti-commutation properties (9. . We will prove now the anti-commutation property (9. nj+1 . nS ) = j (−1)qj (nj + 1))(−1)qj (1 − nj ) |ΨF (N ) = (1 − nj ) |ΨF (N ) (9. .17. nj−1 . . We consider (9. We have used here the deﬁnition qj = k<j nk along with qj = qj . n2 . . .43). nj . 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 . B]+ = AB + BA. nj → 0 . . . Similarly one obtains cj c† |ΨF (N ) = cj (−1)qj (1 − nj ) |ΨF (n1 .44) (9. nj → 1 . 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 . . n2 . nS ) = j Many–Particle Systems (−1)qj (1 − nj ) |ΨF ( n1 . .46) ﬁrst for the case j = k.47) Let C† be the matrix adjoint to Cj . c† ]+ = δj k k where we have used the deﬁnition of the so-called anti-commutators [A. .46).50) .48) From this follows (9.48) vanishes.44) follows from the deﬁnition (9. of (9. .46) [cj .45) follow in an analogous way and will not be derived here. . .248 The operator c† adjoint to cj exhibits the following property j c† |ΨF (n1 . We ﬁrst show that (9. c† ]+ = 0 j k (9. nS ) qj states occupied The operators thus deﬁned obey the commutation properties [cj . [c† . . Let Cj be the matrix corresponding to the operator cj in the basis of fermion states (9.19).45) (9. (9. .

.9.) = c† 1 c† 2 · · · c† N |0 .58) . .. hence. i. nk = 1 . On the r.s.56) Here we have made use of cj cj = 0 and c† c† = 0 which follows from nj ∈ {0. hence. coordinates x and index j into a space part and a spin part. nλ1 . j j j j j j j (9. .51) (9. of this equation we meant to indicate only those N occupation numbers nλj which are non-vanishing. k k 249 (9.h. ˆ It is an interesting exercise to demonstrate that Nj only has eigenvalues 0 or 1. nλ2 . the creation operators must operate in the proper order. . 2 (9. . i. . λ λ λ (9. . . The stated property ˆ follows from the idempotence of Nj which can be derived as follows † ˆ2 ˆ Nj = c† cj c† cj = c† (1 − c† cj )cj = c† cj − c† c† cj cj = cj cj = Nj . 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. . an operator cj must be on the left of an operator ck for j < k. nλN . . S (9.55) is called the particle number operator.e. We j will often separate the states |j .52) hold for any |ΨF (N ) one can conclude (9.50) yield ( cj c† + c† cj ) |ΨF (N ) = |ΨF (N ) . Since (9. j → j. 1 In the following we will apply fermion operators c† and cj only to electrons which carry spin 2 . σ 1 (σ = ± 2 ) .51.s.53) is diagonal in the occupation number representation. for which purpose one needs to envoke only the algebraic (anti-commutation) properties (9.52) One can obtain similarly the same relationship in the case j > k. Correspondingly. nS ) k and. Equation (9.46). .2: Operators of 2nd Quantization (9.e.. x|j → φj (r) | 1 σ . is referred to as the occupation number operator. . nk = 1 .. ( cj c† + c† cj ) |ΨF (N ) = 0 . nS ) k c† cj |ΨF (N ) = (−1)qj +qk −1 nj (1 − nk )|ΨF (n1 . nj = 0 .46). .h. .49) shows that the operator ˆ Nj = c† cj j (9. .57) On the l. 1}. .54) ˆ N = j=1 ˆ Nj (9.e. .29)) |ΨF (. in the basis |ΨF (N ) . with eigenvalues equal to the occupation numbers nj ˆ Nj |ΨF (N ) = nj |ΨF (N ) ˆ Nj . i. .49) and (9. 9. nj = 0 . . j j For j < k one obtains cj c† |ΨF (N ) = (−1)qj +qk nj (1 − nk )|ΨF (n1 . .

(9. 2. . . σs ) (9. xN . N are not all identical.61) is not a one particle operator as long as the N operators rj (x).F (N )|G|ΨB. . 9. (9.15.k=1 j=k (9. These operators had been introduced already in Eq. xk ) ˆ j. An operator N ˆ R = j=1 rj (xj ) ˆ (9. S of single–particle spinorbital states (see page 241) and the operators are speciﬁed through the associated generic operators ˆ f and g . The action of the operators boson and many–fermion states |Ψ is obviously described through the many–particle state matrix elements ˆ ΨB. xk . . (9. 2. 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 ) .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 ) . . .59. We adopt a similar expression to distinguish two–particle operators G and g . The essential aspect of deﬁnition (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. r = 1. . Since the many–particle states are built-up from a basis |r . 9. This procedure is most tedious and becomes essentially impossible in the general case that many-particle basis functions are linear combinations of wave functions of the type (9.59) and over pairs of particles in (9.17) and integrating over all particle space-spin coordinates x1 . i.59) and (9. 9.e. j = 1. .17). σr ) f (x) ψs (r.. respectively.15.60) ˆ correspondingly are called two–particle operators. ˆ .60) involves generic operators – acting on xj and on xj .59. x ) which constitute G are ˆ called the associated generic two–particle operators. ˆ r|f |s = 1 ∗ ˆ d3 r ψr (r.59) ˆ ˆ are called single–particle operators. 9.60) is that the sum over particles in (9. x2 . Operators of the type 1 ˆ G = 2 N g (xj . .62) These matrix elements can be obtained by chosing the operators and many-particle states in the position representation. The operators g (x.F (N ) . 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. namely.6) above. . The operators f (x) which constitute F are called the associated generic single–particle operators1 . respectively – which are all identical.250 Many–Particle Systems 9. ˆ ΨB.F (N )|F |ΨB.

65) This symmetry will be exploited repeatedly below. ∂r 2 ∂x2 ∂x2 ∂x2 1 2 3 (9. u which follows directly g from the deﬁnition (9. ∂rj2 (9.69) ˆ In this case the generic operator is f (x) = δ(r − r). An example for a two–particle operator is the Coulomb repulsion operator 1 ˆ V = 2 N j.k=1 j=k q2 |rj − rk | (9. We will show that the boson and fermion creation and annihilation operators serve this purpose.64) We like to note an important symmetry of the matrix element r.66) Here we adopt the notation for the Laplacian ∂2 ∂2 ∂2 ∂2 = + + . 9. that the symmetry implies that one can switch simultaneously the pairs of indices r.e.63. are spin-independent as is evident from the factor δσσ . 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. Notice. is t(x) = (− 2 /2m)(∂ 2 /∂r 2 ).62) in terms of the matrix elements of single–particle states (9.66) and (9.68) Another single–particle operator is the one–particle density operator N ρ(r ) = ˆ˜ j=1 δ(r − rj ) . s|ˆ|t.71) . u = g d3 r ∗ ∗ d3 r ψr (r. s|ˆ|t.. ˆ 251 (9. σt )ψu (r . g g (9. s and t. u = s. x ) ψt (r. σu ) . ˜ (9. ∂r 2 (9. 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) . (9. t . σr )ψs (r . u. 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. according to (9.3: One– and Two–Particle Operators r. r|ˆ|u.59). s|ˆ|t.70) Both operators.69). σs ) g (x.67) ˆ In this case the generic single–particle operator. one cannot switch the indices of one pair individually.9. i.64). 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 .64) r.

β = ˆ 2 2 1 −2) r. x ) = q 2 /|r1 − r2 |. . Sj. Sj. α|ˆ3 |s. σ | t |t. σ .72) As a ﬁnal example of one– and two–particle operators we consider operators of total spin. β s The operator S 2 = δrs = δrs = = + − 1 2 1 2 δrs δrs (9.77) are orbital-independent as evidenced by the factors δrs and δrt δsu in the respective formulas. s = 1 σ..76) the single–particle state matrix elements of which are r. s.e. σ ˆ + 1 δ− 1 σ δ 1 σ δ 1 σ δ− 1 σ 2 2 2 2 2 = δrt δsu + 1 4 δσσ 1 2 δ 1 σ δ− 1 σ 2 2 δ− 1 σ δ 1 σ 2 2 + . The ˆ matrix elements of the generic operator in terms of single–particle states are ˆ r. The generic operator is s12 = S1 · S2 = ˆ 1 1 S1.77) δσ σ ( δσσ − δ−σσ ) The operators (9. σ | s12 |t. however. The generic operator is given through the Pauli matrices.73) is a one-particle operator. β|ˆ3 |s. The operator N S = j=1 Sj (9.− + S1.74) = N j=1 Sj = 2 N Sj · Sk j. not in the strict sense of our deﬁnition above since the restriction 2 j = k does not apply in the summation. However. In this case the generic operator is v (x. u. α s r.75) is a two–particle operator. Sj is the spin operator for particle j with components (in the spherical representation) Sj.− . 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. s.k=1 (9.+ + S1. The non-vanishing matrix elements of the spherical components of s are (α = + 1 . σ .3 S2. σ.− S2. u. σ.252 Many–Particle Systems which is also spin-independent. β|ˆ− |s.74) and (9. one can obviously extract the term j Sj to be left with a two–particle operator in the strict sense of our deﬁnition. i. Here.3 . (9. α s r. α|ˆ+ |s.+ .3 2 2 (9. β s r.+ S2.

. g and not ct cu .64).t.63) and (9. xk ) → ˆ j. 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.e.78) and (9. Expressions (9.79) in the basis of many–particle states |ΨB. u a† a† at au bosons g r s † † r.60) as follows N ˆ f (xj ) → j=1 N 1 2 1 2 S r.64) one replaces the operators F in (9.s ΨF (N )|cs ΨF (N ) = nr (−1)qr ns (−1)qs δnr −1 nr δns ns −1 δnt nt We refer the reader to Condon and Shortley’s ‘Theory of Atomic Spectra’.79) have the following implication: The operators (9.78. To show that the resulting values of the matrix elements (9.80) (9. s|ˆ|t. respectively.63. s|ˆ|t.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. 9.s=1 S r.176 where the matrix elements of fermion states can be found.k=1 j=k S r.171 and pp. r S Ψ (N In case of r = s one obtains F )|c† cr |ΨF (N ) r = nr r=1 δnr nr . 9. r ΨF (N )|c† cs |ΨF (N ) .81) For the latter matrix elements simple rules can be derived from the algebraic properties of the boson and fermion operators (9.t. pp. and then in g a second step the matrix elements ΨB (N )|a† as |ΨB (N ) .9. (9.46).82) ΨF (N )|c† cs |ΨF (N ) = r cr 2 (9.. i.62).78) 1 2 g (xj .62) are correct one needs to compare the result derived with conventional derivations of the matrix elements2 .F (N ) have the same values as the respective matrix elements (9. 9.59) and G in (9. s|ˆ|t. u . 9.u=1 S r.79) is opposite to that in the matrix element r.45.62) through single–particle state matrix ˆ ˆ elements (9. u cr cs cu ct g fermions (9.24) and (9.s. These rules will be provided below only for the case of fermions. respectively.s=1 ˆ r|f |s a† as bosons r ˆ r|f |s c† cs fermions r r.u=1 (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. u as deﬁned in (9. s|ˆ|t.23. The matrix elements are then actually those of the number operator c† cr which is diagonal in the occupation number representation. r ΨB (N )|a† a† at au |ΨB (N ) r s ΨF (N )|c† c† ct cu |ΨF (N ) r s bosons fermions (9. namely cu ct .83) t=1 t=r.s.

78) does yield the same results as the operator deﬁned in (9. t.i.u r=s. (r. G = G0 + G1 + G2 .86) For the combination of indices r. Case 1 three of the four indices are diﬀerent. ˆ The ﬁrst contribution G0 is 1 ˆ G0 = 2 S r s|ˆ|r s c† c† cs cr + g r s r. i.e. u) are included in the summation for which all indices are diﬀerent.254 From (9. Starting from the deﬁnition (9.’ (all indices diﬀerent) that only tuples (r.84) we can conclude that for diagonal elements holds S Many–Particle Systems ˆ ΨF (N )|F |ΨF (N ) = r=1 ˆ nr r|f |r (9.d. s. s.s=1 a. g r s r.84) The oﬀ-diagonal elements with nr = nr except for r = s..79) and using c† r we can write S 1 ˆ G = rs|ˆ|tu c† c† cu ct g r s 2 r. s).59) when the matrix elements are determined between Slater determinant wave functions.s=1 a. c2 = 0 r (9..82. s. t=u 2 = 0. are ˆ ˆ ΨF (N )|F |ΨF (N ) = (1 − ns )(−1)qs nt (−1)qt s|f |t (9. Case 2 all four indices are diﬀerent. r = t etc. The anti-commutation property (9. t. ˆ ˆ Nr Ns .d.d.i. t). we split the sum in (9. those for which N and N diﬀer in more than two occupation numbers. s. Comparision with the results in Condon&Shortley (pp. t.88) . 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.86) into three contributions.d. t. u in this sum three possibilities exist Case 0 two of the four indices are diﬀerent.s=1 a. (r.e. where each contribution corresponds to one of the three cases mentioned. 171) shows that the operator deﬁned in (9.85) All other matrix elements vanish.87) Here and in the following we denote by ‘a.53) 1 ˆ G0 = 2 S r s|ˆ|r s − r s|ˆ|s r g g r.i. (9. namely. for which holds r = s. 1 2 S r s|ˆ|s r c† c† cr cs .45) yields together with the deﬁnition of the occupation number operator (9.t. in which case ns = ns + 1 and nt = nt − 1 holds. (9.i. 9. ˆ ˆ ˆ ˆ Accordingly. u need to be considered.s. i.

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

74) and (9. u g g (9. nu = nu − 1. We like to express these operators through ˆ fermion creation and annihilation operators. Derive the non-vanishing matrix elements (9.62) for bosons. s|ˆ|r. In particular. i. contributes in this case. All matrix elements which are not covered by the three cases above vanish. s|ˆ|u. One obtains ˆ ΨF (N )|G|ΨF (N ) = (−1)qs +qt (1 − ns )nt r r=s. contributes in this case. u > v. for ˆ non-vanishing matrix elements ΨF (N )|G|ΨF (N ) at most four occupation numbers nr and nr can diﬀer.3.3: 9. one–particle and two–particle operators.92).95) from (9. respectively. u > v or s > 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 (9.90) from (9.t nr r.89). particle number is conserved. Only G2 .4: 9. Derive (9. given in (9. nv = nv − 1..2: 9. it holds S nr = S nr . nt = nt + 1. u. Furthermore. t for which holds ˆ ns = ns + 1 and nt = nt − 1. given in (9. the ‘−’-sign applies for s < r. s|ˆ|v.3.3.73) and S 2 given in (9. ˆ ˆ Show that the matrix elements of F and of G conserve particle number.3. t − r.92).96) . s|ˆ|t. One obtains ˆ ΨF (N )|G|ΨF (N ) = ± (−1)qs +qr +qu +qv (1 − ns )(1 − nr )nu nv r. r g g (9.76). v − r. u < v. Spin Operators ˆ ˆ In this section we will brieﬂy consider the spin operators S given in (9.91).95) In the latter formula the ‘+’-sign applies for s < r. The matrix elements of the corresponding generic operators are (9.94) Case 2 This case covers oﬀ-diagonal elements with nr = nr except for r = s.e.5: Derive (9.75) which are. u < v or s > r.3.92) from (9.90). r=1 r=1 Exercise Exercise Exercise Exercise Exercise 9. v for which holds ˆ ns = ns + 1. Derive (9.1: 9.256 Many–Particle Systems Case 1 This case covers oﬀ-diagonal elements with nr = nr except for r = s. Only G1 . t.

the systems investigated involve a macroscopic number of components such that statistical mechanical approaches are envoked. that the two–particle operator representing interactions between particles induces a variety of interesting phenomena. e. the study of problems governed by Hamiltonians of the type (9.4: Independent-Particle Models ˆ For S 2 one obtains 1ˆ 1 ˆ S2 = N + 2 4 S S 257 ˆ ˆ ˆ ˆ (Nrα − Nrβ )(Nsα − Nsβ ) − r.s.g. In the latter case the indices r.u=1 σ.97).97) is over the orbital states. Actually.98) In many cases the Hamiltonian is spin-independent and the equivalent notation S H = r.σ (9. Examples is the laser action.. rα sβ (9.96) and (9. e.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. 9. t. it is one of the main intellectual achievements of Physics among the Sciences to have addressed systematically the study of systems composed of many strongly interacting components. atoms and nuclei.s=1 c† c† crβ csα . the concerted behaviour of interacting systems can deviate qualitatively from that of systems made up of independent components.6: Derive (9.. evaluation of their stationary states and excitation energies. which represents the interactions between particles. in crystals. the two–particle operator makes the description of many–particle systems a very hard problem. 2 2 Exercise 9.s. superconductivity. The fortunate side of this is.u S (9.3. the description of many–particle systems. s. molecules.99) will be used.s=1 r. the spin part of the single–particle states is represented by α = 1 and β = − 1 . v c† c† cu ct .g. Unfortunately. Often. In fact.97) The summation in (9. u refer only to the orbital part of the single–particle basis.t. The outcome of these studies is that even when interactions between components (particles) are simple. 9.9. would be a simple exercise in linear algebra. The great generality of the concepts developed in the context of many–particle systems can be judged from the fact .96. 9. u c† c† cuσ ctσ v rσ sσ r.s=1 σ 1 ˆ rσ r|t|s c† csσ + 2 S r. but also ordinary phenomena like freezing and evaporation. s|ˆ|u. 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. however.98. s|ˆ|t. If it were not for the two–particle contribution to the Hamiltonian.99) has preoccupied an important part of all intellectual eﬀorts in Theoretical Physics during the past ﬁfty years.t. v r s 2 r.

according to which nuclei and atoms change their qualitative properties when they grow larger.100) in a form in which the factors c† csσ become rσ occupation number operators.J. The Aufbau principle. r=1 3 Urm U ∗sm = δrs m=1 (9. electrons and nucleons. Exercise 9. {|m). . Chemistry essentially would know only one element and Life would not exist.102) U ∗rm Urn = δmn . where S |m) = r=1 Urm |r . ultimately depend in a crucial way on the fermion character of its constituent building blocks. The Hamiltonian of such system is S Ho = r.100) ˆ ˆ We will denote the matrix elements of t as r|t|s = trs .. In such a representation Ho is diagonal and the stationary states can be stated readily. water-ﬁlled jar all electrons of water turn their fermion nature into a boson nature. (9. Before continuing one may ﬁnally point out that the material world as we know it and as it establishes the varieties of natural substances. i. . all material systems would condensate into states which would depend little on particle number. New York. The electronic properties of atoms with diﬀerent numbers of electrons would diﬀer little. Transformation to a New Set of Creation and Annihilation Operators Our aim is to represent the Hamiltonian (9. 1989) .1: Imagine that in a closed. 2. another prototypical systems of many strongly interacting components. ranging from nuclei to the molecules of living systems . . namely of neurons3 . Amit (Cambridge University Press. m = 1.e.e. We will restrict our description in the following to systems composed of an even number (2N ) of particles and to spin-independent interactions. would not exist.103) An account of some of these eﬀorts can be found in Modelling Brain Function – The World of Attractor Neural Networks by D. can be described in a rather straightforward way. What might happen? Independent-Particle Hamiltonian A many–fermion system governed by an independent-particle Hamiltonian.4.. If it were’nt for the fermion nature.s=1 σ ˆ rσ r|t|s c† csσ (9.101) The unitary property of Urn U† U = U U† = 1 1 or S S (9. a Hamiltonian without a two–particle operator contribution accounting for interactions among the particles. S}.258 Many–Particle Systems that today these concepts are providing insight into the functioning of biological brains. To alter the representation of Ho we apply a unitary transformation U to the single–particle states |r to obtain a new basis of states |m). i.

s. for nσ rσ example.111) and. . the expressions drived above for the matrix elements of one.46).9. hence. [d† . they are Fermion creation and annihilation operators. using (9.4: Independent-Particle Models allows one to express conversely |r in terms of |m) S 259 |r = m=1 ∗ Urm |m) . csσ = n=1 Usn dnσ .46) as c† and crσ . (9. (9. [dmσ . accordingly.105) where ˜ ˆ tmn = (m|t|n) = S ∗ trs Urm Usn = r.103) allows one to express [c. of (9. ∗ ∗ Urm Urn . dnσ = r=1 ∗ Urn crσ (9.. d† ]+ = sσ nσ r.108) obey the same anti-commutation properties (9.108) and (9.101) are occupied. The unitary property (9. d† ]+ = δmn δσσ nσ (9.s r The unitarity property r ∗ Urm Urn = δmn yields immediately identity (??).112) .e.s=1 ˆ U † tU mn (9.100) yields ˜ Ho = tmn d† dnσ mσ where d† nσ = r=1 S S (9.h.106) which together with (9. dnσ ]+ = 0 . (9.111). of N -Fermion states in which single particle states |nj ) as deﬁned in (9. In a basis of states N d† j σj |0 n j=1 (9. The l. (9. behave exactly like the operators c† and crσ behave for states |ΨF (N ) .108) These operators describe the creation and annihilation operators of fermions in states |n) which are linear combinations (9. c† ]+ = δσσ [dmσ .and two-particle operators hold in an analogous way for d† and dnσ . nσ We demonstrate here the anti-commutation property (9. rσ namely. 9. the anti-commutation properties (9.f.110) mσ nσ [dmσ .113) Urm Usn [crσ . i. the operators d† and dnσ .109) The operators (9.111) can be written.45.110) can be demonstrated similarly.101) of states |r . expand S trs = mn ∗ ˜ tmn Urm Usn (9.104) One can.107) Urn c† rσ . d† ]+ = 0 (9.104)] S S ∗ Urm d† mσ m=1 c† rσ = .

2. (9. σk ) for j = k (9. . U2n . i. r = 1. The condition (9. . The corresponding eigenvalues n . S are T real.e. S (9. n2 . σj ) = (nk . ) = j=1 nj . ..260 Diagonal Representation Many–Particle Systems The transformation (9. transformation matrices Urn and energy values n . is obtained. σ2 . . USn ).117) are eigenstates of (9.. j j (9.101) gives us the freedom to choose the matrix (9.115) with eigenvalues 2N E(n1 .119) In this state the N lowest orbital eigenstates of trs are occupied each by an electron with spin |α = | 1 .106) is equivalent to S trs Usn = s=1 n Urn ∀r. 1 and |β = | 1 .. a fermion in a spin state | 1 . n = 1.106) diagonal. Eigenstates of (9. The 2N fermion state 2N d† j σj |0 n j=1 where (nj .e. . (9.114) Ho = n=1 † n dnσ dnσ (9..115) can be stated immediately since any many–particle wave function in the occupation number representation is suitable.e. 2.114) together with (9. n2N . It is. We apply for this purpose (9. . . ˜ tmn = In this case Ho has the desired form S n δmn . is N |Φo = j=1 d† α d† β |0 . . i. Vn = (U1n .57) to the case of 2N particles. . . − 1 is called a closed 2 2 2 2 . . σ1 . σ2N . Before proceeding we like to address the issue how the new representation. A non-degenerate orbital state which is occupied by two fermions. This equation poses the eigenvalue problem for the hermitian matrix (trs ). hence. the so-called ground state. i. the associated ( properly normalized) eigenvectors Vn are the columns of (Urn ).118) Ground State In case of an ordering m < n for m < n the state of lowest energy. . − 1 . a simple matter to nσ state many–particle states in the new representation. 2 2 2 2 i.106) and the unitary property of Urn .e.115) ˜ and involves solely occupation number operators N nσ = d† dnσ . 1 as well as a fermion in a spin state | 1 .116) which follows from (9.

Alltogether. namely. energy above the ground state. α |α. Excited States We want to construct now excited states for the system governed by the independent-particle Hamiltonian (9.100). |α.e.n=N (9. namely. i. In case of m = n it gives a contribution ‘1’ for each closed shell which cancels the contribution of the ﬁrst term.. α = 2 |β.121) All four states have the same excitation energy. ˆ one can conclude that |Φo is an eigenstate of S 2 with eigenvalue zero. The ground state has only closed shells. of ∆E = and |β.n=1 m. β . Obviously. However. β = 2 |α.4: Independent-Particle Models 261 shell.n=1 d† d† dmβ dnα mα nβ (9.120) where the occupation number operators refer to the single-particle states |m). β |β. One can conclude immediately that the ˜ ˜ ˜ ˜ (N mα − N mβ )(N nα − N nβ ) does not give any contribution second term in (9.n=1 cmα cnβ cmβ cnα does not give a contribution if m = n and either |m) or |n) are closed shells.124) .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. We will demonstrate now that this property endows the ground state with total spin zero. the states with energy closest to the ground state are those in which a particle is promoted from the highest occupied state |N ) to the lowest unoccupied state |N + 1). α .n=1 m. The two states |α. The action of the operator (9. 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..97). There are four such states. the third term − S m. Similarly.120).120) is particularly simple for closed shells.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. 1 S 4 m. 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.97). β ˆ are eigenstates of S 2 given in (9.n=1 † † if either |m) or |n) are closed shells. the states diﬀer in their spin character. The remaining contributions to S 2 which act on partially occupied orbitals |N and |N + 1 are 1 ˆ Σ2 = 4 N +1 ˆ ˆ ˆ ˆ (Nmα − Nmβ )(Nnα − Nnβ ) m. i. ˆ S 2 ||β. (9. α N +1 − N .e.9.n=1 d† d† dmβ dnα mα nβ (9. α |β. is a singlet state. both with eigenvalues 2 ˆ S 2 |α. β = d† +1α dN α |Φo N = d† +1α dN β |Φo N = d† +1β dN α |Φo N = d† +1β dN β |Φo N (9.

The same result holds for |β. β are not eigenstates of S 2 . . ˆ An eigenvalue ‘2’ of the spin operator S 2 identiﬁes the states |α. The two states |α.128) The cyclic property of the system is expressed through c2N +1 σ = c1 σ . .4. 2. vanish. One can.2: Construct four operators O ˆ O ˆ such that O ˆ S3 . 2. β =− m. β both with eigenvalues 1. for example. ˆ The remaining states |α. 2N ) drσ = √ 1 2N 2N e−irnπ/N cnσ n=1 (9. r = 1. mα nβ (9. The system is assumed to posses a 2N –fold symmetry axis and interactions connect only orbital states |r → |r ± 1 .n=N Many–Particle Systems N +1 d† d† dmβ dnα . β qualify as eigenstates. of course.262 and ˆ Σ3 = − m. hence. m |Φo m m m γσ. α ± |β. ˆ 3 on the two states produces . for |α. α . also eigenstates of Σ2 . The action of Σ N +1 ˆ Σ3 |α. etc. |β.127) ˆ are triplet and singlet states. β . .σ = (9. the linear combinations 1/2(|α. conclude that (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.. one with eigenvalue zero (singlet) and one with eigenvalue ‘2’ (triplet). of course. do not need to be considered since they ˆ give vanishing contributions. α are. α as triplet states. We identify the states |2N + 1 = |1 to avoid cumbersome summation limits. 2N which are located on a ring. . .σ d† +1. ˆ Exercise 9. i. (9. α and |β.129) The symmetry of the Hamiltonian suggests to choose a new representation deﬁned through the operators (r = 1.122) is correct.130) . .σ dN σ N σ. .e.n=N d† d† dmβ dnα d† +1α dN β |Φo = 0 mα nβ N (9. |N + 3 . c† +1 σ = c† σ 1 2N (9.125) ˆ Contributions to S 2 acting on states |N + 2 .100) outlined above we consider a system of 2N electrons which move in a set of atomic orbitals |r .126) which follows from the occurence of squares of fermion operators which. . . 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 . β and |β. however.

. (9.134) which. 9.131) The (9.133) Application to the l. of (9. (9. 4N.9.h. One can then express[.132) one can conclude 2N ei(s−r)nπ/N + eisπ/N n=1 1 2N 2N ei(r−s)nπ/N .f. vanishes for integer s. rσ (9.137) where 1 ˜ trs = e−irπ/N 2N Using (9.s.138) ˜ trs = from which follows ˆ Ho = eirπ/N + e−irπ/N 2N δrs (9. . 2N.131) constitute a unitary transformation U deﬁned through Urn = unitarity property follows from 2N eirsπ/N = 0 r=1 for 0 < s < 2N .s=1 σ ˜ rσ trs d† dsσ (9.140) The Hamiltonian is now diagonal and the construction outlined above can be applied. . eisπ/N 1 − e2N isπ/N 1 − eisπ/N (9.132) gives for r = 0. 1−a (9.132) an identity which can be derived employing the well-known expression for a ﬁnite geometric series m as = a s=1 1 − am . indeed.135) c† nσ 1 = √ 2N e−irnπ/N d† rσ r=1 (9. .139) −2t cos r=1 σ rπ N d† drσ .109)] ¸ cnσ = √ 1 2N 2N eirnπ/N drσ r=1 2N (9.4: Independent-Particle Models d† = √ rσ 1 2N 2N 263 eirnπ/N c† nσ n=1 √ 1 eirnπ/N . n=1 (9.136) ˆ and. 2N (9. one obtains the new representation of Ho 2N ˆ Ho = −t r. .130. thereby.

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

If one restricts m = 1. In our formulas below we will adopt strictly the convention that indices r. . . S) ˜ S S |m = ˜ r=1 Urm |r . .t. 2. s|ˆ|t. U = ( Urm ) as deﬁned in (9. The states |m are connected with the states |r through (m = 1. . . S} the elements of which will be labelled by indices ˜ m. 2. . The states |m serve to deﬁne a reference state of the system ˜ N |Φo (U) = j=1 d† α d† β |0 .142) is based. r = 1. 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. t.9.142) r|t|s c† csσ + rσ sσ 2 r. u refer to the initial basis of single-particle states |r . u v r. s.s. This is done as follows: ˆ Vmf (|Φo (U) ) = 1 2 S r. . . .145) which will play a pivotal role: it is this the state in which the fermions are assumed to be moving and relative to which the mean interactions acting on individual fermions is determined. .108). s|ˆ|t.σ and derive a single-particle operator which approximates this Hamiltonian. ˜ (9.u=1 σ. 2. .s. . . p.144) deﬁned as in (9. 2.146) . m = 1.143) 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. is not unitary.145).142) over this reference state to turn the contribution into an eﬀective one-particle operator.t. N . We will adopt a second set of single-particle states {|m .143) with r. u c† c† cuσ ctσ v (9.s=1 σ r. q. . 2.5: Self-Consistent Field Theory 265 In this section we will consider systems of 2N fermions described by the spin-independent Hamiltonian S S 1 ˆ rσ H = r. S is a unitary matrix. j j (9. as in in (9. m = 1.u=1 σ. . S on which (9.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. the corresponding U = ( Urm ) forms an S × N matrix and. naturally. n. |r = m=1 ∗ Urm |m . .

n c† ˜ v ˜ sσ m.266 Here ··· denotes the average O = Φo (U)| O |Φo (U) . s|ˆ|t.u=1 σ. hence. For the averages ··· in (9.σ S 1 − 2 where.n. u ˜ ˜v m. Many–Particle Systems (9.147) Since the reference state |Φo (U) is deﬁned in terms of the single-particle states |m it is preferable ˜ to switch the representation of (9.n=1 σ.150) r.84).σ S d† dnσ mσ ctσ (9. Since the averages aﬀect only two creationannihilation operators actually a mixed representation is most suitable ˆ Vmf (|Φo (U) ) 1 = 2 1 2 1 2 1 2 1 2 S m. s|ˆ|˜ . n|ˆ|t. m|ˆ|˜ . d† d† mσ nσ dmσ dnσ d† dnσ mσ = = = = and.n. u c† d† dnσ cuσ ˜ vn rσ mσ r.σ S dmσ dnσ + r.t.σ S ctσ + m. by means of the rules (9. m.146) accordingly.s. n = ˜ ˆ ˜ s.t.s.148) r. n c† ctσ δσ σ ˜ v ˜ rσ m≤N r.n. u c† ˜ vn sσ m. m| v |t. ˆ (9. for example.148) are obtained in a similar way by transforming only two of the four single-particle states into the new representation.u ∗ r.σ S d† dnσ cuσ mσ − r. s| v t.148) one obtains.149) The remaining matrix elements appearing in (9.n. ˆ Vmf (|Φo (U) ) = 1 2 S Φo (U)|d† d† |Φo (U) = 0 mσ nσ Φo (U)|dmσ dnσ |Φo (U) = 0 Φo (U)|d† dnσ mσ δmn δσσ 0 |Φo (U) m ≤ N m > N (9.m. s|ˆ|m.t=1 σ.n=1 σ.t=1 σ. n c† c† v ˜ ˜ rσ sσ r. m|ˆ|t. n c† d† dnσ ˜ v ˜ rσ mσ r. m|ˆ|t.m.σ .u=1 σ.s.σ S d† d† mσ nσ cuσ ctσ + r. u Usm Uun .u=1 σ.m.

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

The state thus determined is referred to as the self-consistent independent-particle ground state. Step 1: Choosing an Initial Reference State One deﬁnes the Hamiltonian ˆ (1) Hmf = r.155). . often also referred to as the Hartree-Fock approximation. i. Let us state now the construction of the self-consistent ﬁeld ground state in more detail. i. . SCF-Algorithm. S . . .157) and determines the associated diagonal representation deﬁned through S (1) ˆ r|t|s Usm = s=1 (1) (1) m Urm . . 2. steps (2) and (3) are being repeated M times until the procedure converges. is the lowest energy independent-particle ground state one can construct.4. . .s=1 σ ˆ r|hmf |s c† csσ rσ (M ) (9.158) where the labels m are ordered to obey the condition (1) m < (1) n for m < n .4 and deﬁning this as the new reference state. the independent-particle nature stemming from the fact that the functional form of the ground state. 2. .145). .6 Self-Consistent Field Algorithm The Self-Consistent Field (SCF) approximation. 2. . U(1) is the S ×N –matrix (1) of the ﬁrst N column vectors of Urm . .155) following the method outlined in Section 9. S . The procedure determines the reference state as the ground state of the independent-particle Hamiltonian (9. M = 1. 2. the reference state resulting from step 2M + 1 (within numerical error) is equal to the reference state resulting from step 2M − 1. and obtaining in a step (3) the ground state of this one-particle Hamiltonian according to the construction in Section 9.268 Many–Particle Systems 9.160) .159) One deﬁnes the reference state |Φo (U(1) ) using the deﬁnition (9. is exact only for an independent-particle Hamiltonian. 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. . This procedure. S. i. We will argue below that the self-consistent ﬁeld ground state. 2. under conditions which often are realized. SCF-Algorithm. ..e. assuming in an intial step (1) a properly chosen reference state. (9.145). however. can achieve this goal only iteratively.e. r = 1. . m = 1.. (9.: Determine Mean Field Hamiltonian In this step the mean ﬁeld Hamiltonian S ˆ (M ) Hmf = r. m = 1.e. r = 1. 9.142).s=1 σ S ˆ rσ r|t|s c† csσ (9. . determining in a step (2) the corresponding mean ﬁeld Hamiltonian (9. Step 2M. S .s=1 σ S ˆ r|t|s + r| vmf |Φo (U(M ) ) ˆ |s c† csσ rσ = r.119. (9.

m = 1. . etc. 2. t| v |s. . M = 1.162) is solved adopting an ordering of the labels m which obeys the condition (M +1) m < (M +1) n for m < n . m = 1. . . SCF-Algorithm: Continuation and Convergence Condition When step 2M + 1 is completed. 2. . σ|ˆrs |u.9.s=1 σ ˆ r|f |s c† csσ rσ (9. The procedure is continued until it ∗(M ) (M ) ∗(M +1) (M +1) Uum − N Urm Uum | is detected that the one-particle density diﬀerences | N Urm m=1 m=1 for r. . . U(M +1) is the (M +1) S × N –matrix of the ﬁrst N column vectors of Urm . u − r.1: Let Prs be the one-particle operator with the generic operator prs = |r s|.: 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 . r = 1. t| v |u.145).167) .6: Self-Consistent Field Algorithm is evaluated. . i.s=1 (9. (e) Consider the creation operators N † gqσ = m=1 Vmq d† mσ (9. S . ˆ t. .163) The result yields the reference state |Φo (U(M +1) ) using the deﬁnition (9.u=1 ( 2 r.6. SCF-Algorithm. . i.. step 2M + 2 is carried out. . . .161) is calculated. r = 1. (9. . s ) ˆ ˆ m=1 Utm ∗(M ) (M Uum ) (9. 2. S do not exceed a preset threshold. i. (c) Show that any spin-independent one-particle operator S ˆ F = r.e. . S . the S × S–matrix S (M ) r|hmf |s N 269 ˆ = r|t|s + t. S. . S (9.145).165) can be written ˆ F = S ˆ ˆ r|f |s Prs r. Step 2M + 1.164) ˆ (a) Determine the expectation value of Prs for the state as deﬁned in (9. 2 .e. (9. ˆ Exercise 9. 2. . σ p = δtr δsu δσσ .e. ˆ (b) How can the expectation values of the operator Prr be interpreted. s = 1.. indicating that the state converged. 2.166) ˆ (d) Deﬁne a similar two-particle operator Prstu .

171) where the convention (9. We denote the representation which results from the SCF algorithm after its convergence as follows: S |m = ˜ r=1 (SCF Urm ) |r .168) ˆ ˆ has the same expectation values for Prs and Prstu as the reference state (9. .170) (SCF ) Here |r .172) is obeyed.162. i.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.270 Many–Particle Systems which are connected with d† through a unitary N × N –matrix (Vqm )..e.9.σ + 2v ( c† c1α c† c1β + c† c2α c† c2β ) + v 1α 2α 1β 2β + c† c2σ c† 2σ 1σ c1σ ) .145). .168) through a physical observation? Exercise 9. (9.163).163). 2. (9.171) implies that for the mean ﬁeld Hamiltonian holds S ˆ Hmf (U(SCF ) ) = m=1 σ (SCF ) † dmσ dmσ m . For this representation holds N m| t |˜ + ˜ ˆn m 2 mm | v |˜ m ˜ ˜ ˆ n˜ − mm | v |m n ˜˜ ˆ ˜ ˜ = (SCF ) δmn m (9. (9.169) 9. S denotes the initial single-particle states and Urm is the unitary S × S– transformation matrix obtained in (9. Equation (9.145) and (9. . We begin by summarizing the result of the SCF algorithm. (SCF ) m < (SCF ) n for m < n (9.6. r = 1. (f) Can one distinguish the states (9. Note that this is not an mσ S × S–matrix! Show that the reference state deﬁned through N † † gqα gqβ |0 q=1 (9.7 Properties of the SCF Ground State We want to investigate now the properties of the SCF ground state.173) .

2 m=1 (SCF ) m − m.171) in terms of matrix elements r|t|s . v (SCF ) Show that the resulting eigenvalue problem of the type (9. Finally.4 that SCF is a total singlet state.. Since this state is composed only of closed shells one can conclude following Section 9.. 2 m=1 m|t|m + ˜ ˆ˜ m. mα mβ (9.177).174) The ﬁrst property we consider is the total spin character of SCF .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. exploiting the spinindependence of (9. Since the system under consideration has only 2N particles altogether.162) is non-linear in Urm .173) holds N ˆ SCF Hmf (U(SCF ) ) SCF = 2 r=1 (SCF ) m .146) counts a particle twice. one can also determine the expectation value of SCF for the Hamiltonian (9. and once as a probe particle being subject to the mean pair interaction.7.2: Derive (9.9.93) for the expectation values (diagonal elements) of the one-particle and two-particle part of the Hamiltonian (9.e.1: Reformulate (9.177). Exercise 9. the SCF ground state is N SCF = m=1 d† d† |0 . one obtains ˆ SCF H SCF N = N (9.7. s|ˆ|t.e.142). r.171) yields ˆ SCF H SCF N = N (9. Within the mean ﬁeld approximation as described by (9. once as a member of the ground state. . This over-counting leads to the correction term in (9.m =1 Comparision with (9.176) 2 mm | v |mm ˜˜ ˆ ˜˜ − mm | v |m m ˜˜ ˆ ˜ ˜ .142).142). the mean ﬁeld potential ˆ Vmf (|Φo (U) ) deﬁned in (9.175) However. ˆ SCF H SCF .7: Properties of the SCF Ground State 271 i. (SCF ) ˆ Exercise 9.84) and (9. i. (9. the mean-ﬁeld Hamiltonian is diagonal in the SCF representation. We like to determine next the energy expectation value of SCF .177) 2 mm | v |mm ˜˜ ˆ ˜˜ − mm | v |m m ˜˜ ˆ ˜ ˜ . u and Urm . Employing expressions (9.

. S N0 (9. The case n = 1 is termed the half ﬁlled band case.178)..s.s. s|ˆ|t. labeled r = −N0 + 1. v (9.4–9. We consider as an example the Hubbard model of an inﬁnite linear lattice and apply the model to describe magnetic instabilities in metals.u σσ where r. v of the Hubbard Hamiltonian (9. of (9. in the two-dimensional copper oxide lattices of high temperature superconductors.σ crσ + crσ cr+1. t describes the coupling of state |r to states |r ± 1 at the two neighboring lattice sites.s.178) and through the so-called ﬁlling factor n n= 2N N = .182) n can assume values 0 ≤ n ≤ 2.178) Here the index r.t..178).178) v contributes only in case that two electrons occupy the same lattice site. The one-dimensional Hubbard model is described by the Hamiltonian ˆ H = −t rσ † c† r+1. except in case of one-dimensional systems.h.e. in the usual two–particle operator form.180) V = rσ sσ 2 r. N0 which are populated by 2N electrons (we choose an even number of electrons for convenience). s|ˆ|u.σ rσ The potential energy term.σ + v 2 c† c† crσ crσ . We assume that there are altogether S = 2N0 lattice sites.h. t is assumed to be a positive.128).. 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.7 for ﬁnite systems.178). respectively) in state |r at lattice site r. i.σ crσ + crσ cr+1.181) The Hubbard model. v (9. for example. approximation schemes like the self-consistent ﬁeld approximation play an important role. v represents the ‘on-site’ Coulomb repulsion. describes the sites of a linear lattice. of (9. cannot be solved exactly. . In spite of the simplicity of its Hamiltonian the Hubbard model. i.179) r+1. Due to the lack of an exact solution in dimensions higher than one. † ˆ Ho = −t c† .8 Mean Field Theory for Macroscopic Systems The SCF algorithm has been developed in Sections 9.e.e. rσ rσ r σσ (9. In the present Section we want to adapt the algorithm to systems containing a macroscopically large number of fermions. referring to the fact that in this case the band of single particle energies (9. i. t = v δru δst δrs . as stated through Hamiltonian (9. is characterized through the parameters t. real number. . reads 1 ˆ r. The operator c† (crσ ) creates (destroys) an electron with spin σ (σ = α. Both N0 and N are macroscopically large numbers.192) derived below is ﬁlled half. is identical to that of the independent–particle Hamiltonian H0 (9. the second term on ther. According to (9. β for ‘up’ and ‘down’ rσ spins. The ﬁrst term on the r. (9. . r ∈ Z. . u c† c† cuσ ctσ .272 Many–Particle Systems 9.

178) can be determined by applying (9. < ±(N0 −1) < N0 holds as can be readily veriﬁed.184) ˆ We are now ready to apply step 1 of the SCF algorithm and diagonalize term H0 . the operator H0 can be re-written in the diagonal form N0 ˆ H0 = m=−N0 +1 σ m d† dmσ mσ (9.6. by N −1 ||SCF = m=0 d† d† |0 . Inserting (9. N0 (9. Accordingly.188) . The self-consistent ﬁeld Hamiltonian Hmf which corresponds to the Hubbard Hamiltonian (9. 9.186) one obtains for the mean ﬁeld Hamiltonian N0 ˆ (2) Hmf = m=−N0 +1 σ m d† dmσ + v mσ N 2N0 c† crσ .186) where m = −2t cos mπ . For this purpose we apply the SCF theory presented in Section 9.181.144).174).154). mα mβ (9. Using (9. We can now embark on step 2 of the SCF algorithm.185) ˆ According to (9.8: Macroscopic Systems 273 Since we are dealing presently with a macroscopic system. is Urm = √ 1 exp (irmπ/N0 ) .9. in fact. . The unitary transformation which diagonalizes H0 . 2N0 (9.σ m=0 ∗ Urm Urm c† crσ . rσ rσ (9. we assume that the SCF ground state is given by (9.140). boundary eﬀects are assumed to be negligible.. SCF Ground State for the Hubbard Model We seek to determine the ground state of the Hubbard model stated by (9.e.179). rσ (9.183) where the operators d† are related to the original creation operators c† through the unitary mσ rσ ˆ transformation U deﬁned in (9. 9.184) and using (9.178) within the framework of the self-consistent ﬁeld theory. Accordingly. a freedom which we employ to adopt so-called periodic boundary conditions like those for the example “2N independent electrons on a ring” on page 262. we identify |r + N0 = |r and later let N0 go to inﬁnity. i.155. This task has ˆ been solved already on page 262. .185) into (9.187) The energy levels are labeled such that 0 < ±1 < . one obtains for the mean ﬁeld Hamiltonian N −1 ˆ ˆ Hmf (U) = H0 + v r.

(9.177) and mm |ˆ|mm = mm |ˆ|m m = ˜˜ v ˜˜ ˜˜ v ˜ ˜ which holds in the present case. can be obtained.190) We have used in the latter expression the deﬁnition (9. called the Fermi energy. holds k ≡ km ∈] − π. the electrons in the present description behave like independent particles with energies ¯m = m + vn .183) is not the only candidate for the mean ﬁeld ground state of the Hubbard model. .e.f. as given in (9. there exist several ways of formulating mean ﬁeld approximations. The bottom of the energy band corresponds to (0) = −2t and the width of the energy band is 4t.. 4 So far we have not exploited the fact that N0 and N are macroscopically large quantities.185) one can prove rσ mσ N0 c† crσ = rσ rσ m=−N0 +1 σ d† dmσ . i.1). Indeed. Indeed.182)]. crσ through d† . by ﬁlling up this energy band with electrons from the bottom of the band (which corresponds to k = 0) to a maximum energy.189) and. for km = mπ/N0 holds km ∈ −π + N0 . in the limit N0 → ∞. mσ (9. the state with the lowest possible energy.191) where m is given in (9. −π ≤k ≤π .192) This dispersion law is presented in Fig.1. To demonstrate this we consider an alternative formulation of the mean ﬁeld approximation for Hamiltonian (9. even for one and the same Hamiltonian.. π and. 9. denoted by F (see Fig. In other words. namely. the results of diﬀerent formulations agree qualitatively. hence.109) and using (9. . dnσ according to (9. 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. is already diagonal.274 Many–Particle Systems Expressing c† .182) of the ﬁlling factor n [c. ˆ The Hamiltonian Hmf . i. (9. Emf is then N −1 v 2N0 Emf = 2 m=0 vn vn −N = 2 m+ 2 2 N −1 m m=0 + vn 4 (9. Apparently. but may diﬀer quantitatively. π]. Usually. . −π + N0 . The ground state of the system. the self–consistency condition is exactly met and the SCF algorithm converged after two steps.187) form a quasi1 2 continuous energy band.190). according to the Pauli principle. N0 ˆ (2) ˆ Hmf = Hmf = m=−N0 +1 σ m + vn † dmσ dmσ 2 (9.187). When N0 becomes macroscopically large the single particle discrete energy levels (9. through the so-called dispersion law (k) = −2t cos k . π].e. One can see that −2t ≤ (k) ≤ 2t holds. Alternative Description of the Mean Field Approximation The state (9. The energy Emf of this state can be calculated readily by using (9. 9. .178).

193) into (9.56). The mean ﬁeld approximation neglects the . First. let us express the interaction term in (9. the operator Nrσ can be written ˆ Nrσ = Nrσ + δ (Nrσ ) .196) (9. say the one given in (9.195) ˆ ˆ ˆ and where δ (Nrσ ) = Nrσ − Nrσ describes the ﬂuctuations of Nrσ .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 .45. F = (±kF ) denotes the Fermi energy. (9.e. the deviation of Nrσ from its mean value Nrσ for the corresponding reference state.8: Macroscopic Systems 275 Figure 9.1: The dispersion law (k) for independent electrons on an inﬁnite one dimensional lattice. we have employed (9. in the last term on the right hand side. where Nrσ is the mean occupation number of the one particle state |rσ .194) ˆ For a given reference state.193) where.e. i.183). (9..9.178) and performing the summation over the spin indices results in ˆ ˆ H = H0 + v r ˆ ˆ Nrα Nrβ .. i. 9. Nrσ = SCF ||Nrσ ||SCF . The anti-commutation properties (9.178) in terms of the occupation number operaˆ tors Nrσ = c† crσ . Inserting (9. rσ rσ rσ (9.

2N0 2 holds ˆ Hmf = = mσ mv ˆ H0 + 2 vn2 2 rσ nv † vn dmσ dmσ − N . separating the occupation number operator into a mean value plus ﬂuctuation and neglecting the ﬂuctuations in 2nd order. Spin-Polarized Mean Field Ground State ˆ We want to consider now a ground state for Hmf deﬁned through Nrσ = nσ = const (9. in the mean ﬁeld approximation.183) . non-vanishing magnetization. that nα and nβ assume diﬀerent values. yields essentially the same mean ﬁeld Hamiltonian (9. consequently.198) Let us assume that the ground state of the Hubbard model.194) by dropping all the terms which contain the ﬂuctuations to second order. In case nα = nβ the following construction will lead to a ground state which coincides with the non-magnetic ground state (9. m+ 2 2 c† crσ − 2N0 rσ (9.198) as the one obtained by using the mean ﬁeld potential (9. For such state the mean occupation number nrσ is uniform at all lattice sites.195).199) which is identical to expression (9.146). Note that the procedure employed in (9.183).m ∗ Urm Urm d† dm σ mσ = m=0 ∗ Urm Urm = n N = . 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 .e. the ground state of the system can have a non-zero local spin and. i. Therefore. (9. 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 (9.276 Many–Particle Systems ﬂuctuations to second order and the mean ﬁeld Hamiltonian is obtained from (9. Because of Nrσ = c† crσ rσ N −1 = m.200) allowing. but the mean number of particles with spin α can be diﬀerent from the mean number of particles with spin β. . is given by (9..191) for the ground state energy. however.197) and the corresponding mean ﬁeld Hamiltonian becomes ˆ ˆ Hmf = H0 + v r ˆ ˆ Nrα Nrβ + Nrβ Nrα − v r Nrα Nrβ .

a spin–polarized ground state. N0 ∼ 1023 .204) Emf ≡ 2N0 2N0 m.206). In this limit the discrete energy spectrum (9. hence. subject to the constraint n = nα + nβ . This last equation tells us that the electrons with spin α (β) are accommodated by an energy sub–band α (k) ( β (k)) which is obtained from the dispersion law of the non-interactive electrons (k) by an overall shift of vnβ (vnα ).e.e. such that the larger nσ . E with respect to nα . i. (9. The ground state (corresponding to the lowest possible energy) of the many electron system is obtained by ﬁlling these two energy sub–bands with electrons up to the same maximum energy value.203) mσ = m + vn−σ . Since β (k) − α (k) = v(nα − nβ ). (9.205) is met.201) ( m ˆ + vnβ ) Nmα + ( m ˆ + vnα ) Nmβ − 2N0 vnα nβ .206) ascertains that condition (9. In the limit N0 → ∞ this sum can be expressed as an integral.σ with respect to nα and nβ .206) ˜mf is minimized where µ is the Lagrangian multiplier considered an independent variable.203) is provided by the continuous function σ (k) = (k) + vn−σ ..206) needs to be evaluated. the so-called Fermi energy F = µ (see below). The actual values of nσ (σ = α.207) where (k) is given by (9. The values of nα and nβ are determined by minimizing the energy density (9.8: Macroscopic Systems from which we obtain nσ = The mean ﬁeld Hamiltonian is then ˆ Hmf = m 277 1 2N0 Nmσ . (9. nβ and µ.. one can expect the system to lower its energy by assuming nα = nβ and. (9. i. for any function f mπ holds (compare with the deﬁnition of a deﬁnite integral N0 as the limit of the corresponding Riemann sum) 1 N0 →∞ 2N0 lim N0 f m=−N0 +1 mπ N0 π = −π dk f (k) . At this point we exploit the fact that our system is macroscopically large. 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. For this purpose the sum in (9.2a. if nα > nβ then β (k) > α (k).205) Such minimization is realized through the method of Lagrangian multipliers.. e. as is shown schematically in Figure 9. The additional term in (9. Therefore. 2π (9. the smaller is σ (k) for a given k. In fact. m (9.g. 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 (9.192).208) .202) This Hamiltonian describes a system of non-interacting electrons with a spin-dependent dispersion law (9.9.

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

. ∂µ (9.217) and µ. θ(x) = 1 if x > 0 and θ(x) = 0 if x < 0. namely. v and n.215) nα and nβ in (9.221) and (9. ∂ β ˜ ∂ Emf =0. nα and nβ as a function of the parameters of the Hubbard model. in principle. 2t[ then the θ function in (9. 2 (9. (9.219) (9.2]. on conditions for a ground state of minimum energy are ˜ ∂ Emf =0. (9.f.213) The presence of the θ function in the above formula guaranties that the density of states vanishes for | | > 2t. β . i.213) and carrying out the resulting integral yields nσ = 1 arcsin( σ /2t) .220) µ can be readily obtained µ= 1 ( 2 α + β) + vn .216) Using (9. Replacing again the sum by an integral over energy one can write σ nσ = −2t ρ( )d (9.214) are given by the sum (9. β (9. F = α + vnβ = β + vnα .216) allow one.9.214) where σ denotes the value of (k) which corresponds to the top of the ﬁlled portion of the energy sub–band σ (k) [c. (9. π α. 9.219–9.201). ∂ α ˜ ∂ Emf =0. of t. One obtains ˜ Emf = α β ρ( )d + −2t −2t ρ( )d + vnα nβ + µ(n − nα − nβ ) . if we assume that is restricted to the interval ] − 2t. one arrives at the following expression of the density of states ρ( ) = θ(2t − | |) π (2t)2 − 2 . (9.. to determine the unknown quantities α . namely. Fig. + vnα − µ = 0 . µ.218) read ˜ ∂ Emf = ∂ α ˜ ∂ Emf = ∂ β and α + vnβ − µ = 0 . i.f. From (9.220) β ˜ ∂ Emf = n − nα − nβ = 0 .e. (9.204) through an integral expression. the necessary ˜ Since Emf [c.219. 9. outside the energy band (k).e..213) can simply be replaced by 1. i.e.206)] depends only on continuous quantities. (9.8: Macroscopic Systems 279 where θ(x) is the step function.221) ∂µ Equations (9.222) .218) The derivatives can be readily determined and the conditions (9. We can employ the results obtained to express the ground state energy density (9.

2b). One still needs to determine α and β . the Hubbard model has a ground state with ∆ = 0. (9.223) ∆ = 0 corresponds to the ground state (9.217) and (9. the slope of f (∆) at the origin is larger than 1 [c. Figure 9.223 one obtains ∆= and second.280 Many–Particle Systems Comparision of (9.227) will have a second. We assume in the following discussion that the implicit ∆ dependence of β in f (∆) can be neglected.222) and (9. (9. (9. If the slope of f (∆) at the origin is less than 1.e. as reﬂected by the expression d[f (∆)] = vρ( d∆ β + ∆) = π (2t)2 v − (v∆ + 2 β) >0.226) follows that f (∆) is a monotonically increasing function of ∆.f. f (∆) converges to a maximum value as ∆ → ∆max = n.228)] vc ρ ( β ) = 1 . (9. For this purpose one plots f (∆) versus ∆ as is shown schematically in Figure 9.. One can determine vc by equating the slope of f (∆) at the origin to 1. from (9. .229) has a wider range of validity than one can infer from the present analysis of the Hubbard model.215) reveals that µ is indeed equal to the Fermi energy F (see also Figure 9. From the deﬁnition (9. (9.227) Since f (0) = 0 one can infer that ∆ = 0 is always a solution (the so called paramagnetic solution) of condition (9.183) without magnetization.220) and (9. i. condition (9.3].f.227).219–9.224) v∆ ∆= β ρ( )d = 0 ρ( β + )d . (9. condition (9.229) This cindition is known as the Stoner criterion and gives the critical value of v which determines the onset of the ferromagnetic long range order. from (9.223) follows α α − v β .227) has only the trivial solution ∆ = 0. Accordingly. and only if. The slope of f (∆) at the origin depends on the Hubbard model parameter v. a magnetic ground state.227) can be solved graphically. For v above a critical value vc .e. through the condition [c. Indeed.228) Since the total number of states per site is ﬁnite.3. so called ferromagnetic solution. the slope of f (∆) is positive for all ∆ ≥ 0. (9. For this purpose we consider the magnetization of the mean ﬁeld ground state ∆ ≡ nα − nβ . The Stoner criterion (9.. Equation (9.226) one can obtain ∆ by solving (cf. (9. non trivial.226) ∆ = f (∆) . One can express ∆ as a function of α and β in two diﬀerent ways. if. i. (9. First.225) Deﬁning the function v∆ f (∆) = 0 ρ( β + )d .

9.9.4.234) 2t 2 The resulting phase diagram of the ground state.8: Macroscopic Systems 281 Figure 9. 9. (9.232) For a given 2t the values of x and y can be obtained numerically and other relevant quantities v mentioned above can be calculated. n. the plot of vc vs. i. We like to determine ﬁnally the dependence of vc on the ﬁlling factor n.231) follows y= δ π π (n − 1) − ≈ (n − 1) . α. 2 2 2 (9.232) together with (9.233) Employing the approximation sin(y + δ) − sin y ≈ δ cos y for small δ. Since ∆ vanishes near vc we set x = y + δ where δ is a small quantity. The corresponding 2t analytical expression provides the phase diagram of the ground state: when v approaches vc from above.3: Graphical solution of the mean ﬁeld equation ∆ = f (∆). the magnetization ∆ of the ground state vanishes and remains zero for values v < vc . y and. 2t As already mentioned.230) one obtains from (9..232) states that x − y is proportional to ∆. (9. the ground state is characterized by the quantities (v.233) yield the desired expression vc π = π cos (n − 1) . hence. β = 2t sin y .231) (9. t. n). From (9.205) the following set of equations determining x. any .220.219. v (9. (9. Accordingly.e. is presented in Figure 9. By parameterizing α and β α = 2t sin x . β x+y x−y = = π(n − 1) 2t π (sin x − sin y) = π∆ . Equation (9.

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

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

284 Many–Particle Systems .

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

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

summation over that index is assumed without explicitly noting it.10. The second reason is that upper and lower positions allow us to accomodate the expression (10. Aµν . The subset of GL(4.. 0 0 0 −1 (10.9) 1 0 0 0 −1 0 0 ( gµν ) = 0 0 −1 0 = g . (10. R). This will be explained further below. R).6) yields Lµ ρ gµν Lν σ xρ xσ = gρσ xρ xσ . the Lorentz transformations constitute a linear transformation which we denote by 3 x µ = ν=0 Lµ ν xν ..6) x µ = Lµ ν xν . ν=0 (10. however.1: Natural Representation of the Lorentz Group 287 coordinates in one reference frame and by x µ the coordinates in the other reference frame. 4 × 4 tensor: Aµ ν . We will exploit this property below to determine the general form of the Lorentz transformations. The Lorentz transformations form the subgroup of all matrices which leave the expression (10. 3 3 3 yµ xµ = new yµ xµ . Aµ ν B ν ρ = ν=0 Aµ ν B ν ρ .7) The reason for the notation is two-fold. 2.e. xµ . the notation in (10. the latter notation will be used for diﬀerent quantities further below. Aµν .e.6) allows us to introduce the so-called summation conventon: any time the same index appears in an upper and a lower position.9) and (10. Aµ ν xν = µ=0 new old Aµ ν xν . i.5) into scalar products. 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 latter set constituting a group. (10. L ∈ GL(4. (10. 1. i.10) Combining condition (10. First.6) Here Lµ ν are the elements of a 4 × 4–matrix representing the Lorentz transformation. This condition can be written xµ gµν xν = x gµν x where µ ν (10. The Lorentz transformations are non-singular 4 × 4–matrices with real coeﬃcients. Since this holds for any xµ it must be true Lµ ρ gµν Lν σ = gρσ .12) (10. Aµ ν . the . 3: 4-vector: xµ .8) new old old The summation convention allows us to write (10. [ A more conventional notation would be Lµν .] The following possibilities exist for the positioning of the indices µ.5) invariant.11) This condition speciﬁes the key property of Lorentz transformations. ν = 0.

1) is. i. .14). in fact.e. This set is identical with the set of all Lorentz transformations L.s.1) since it satisﬁes (10. R) is a group.14).1).18) (10. 10.1). of (10.20) Multiplying (10. (10. (10.1).21) i..h.14) L0 0 −L1 0 −L2 0 −L3 0 −L0 1 L1 1 L2 1 L3 1 . In fact.19) i.22) (10.20) results in the desired property (L−1 )T g L−1 = g .17) one obtains (L−1 )T g L−1 = gLgggLT g = gLgLT g .1). L ∈ GL(4. 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.e.10) of g one can rewrite the invariance property (10.1).15) (10. LT gL = g } . One can also show that if L ∈ O(3. the inverse of L 1 L−1 (10. property (10. L3 ∈ O(3. This stipulates that O(3.16. L−1 ∈ O(3.14) holds for the inverse of L.e.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.e.13) L = ( Lµ ν ) = 2 0 L 0 L2 1 L2 2 L3 3 .12) LT gL = g .16) The corresponding expression for (LT )−1 is obviously (LT )−1 = (L−1 )T = g L g .17) we note ﬁrst that the identity matrix 1 is an element of O(3. the associated inverse obeys (10. 3 2 1 2 1 2 (10.288 Relativistic Quantum Mechanics elements of which satisfy this condition. L3 0 L3 1 L3 2 L3 3 Using the deﬁnition (10. 1) = { L. of O(3.14) from the right by gLT and using (10.. For L3 = L1 L2 holds LT g L3 = LT LT g L1 L2 = LT (LT gL1 ) L2 = LT g L2 = g . R). We want to show now L = O(3. = g LT g = 0 1 2 −L 2 L 2 L 2 L3 2 −L0 3 L1 3 L2 3 L3 3 (10. We consider 1 then L1 .. employing expressions (10. To demonstrate the group property of O(3.. From this one can obtain using g2 = 1 1 (gLT g)L = 1 and. hence. a group.1) ⊂ GL(4.1). L2 ∈ O(3. (10. is called O(3. i.

(10. = { L.28) (10. + + ↑ ↑ We can also demonstrate that for A. The above shows that the set of proper Lorentz transformations L↑ allows one to generate all Lorentz transformations.27) (10. B ∈ L+ holds C = AB ∈ L+ . L0 0 ≥ 1} . + − gL↑ + = L↑ g . L0 0 ≤ −1} . det L = −1. M1 = a M2 }. L0 0 (10. M2 ∈ M.1: Natural Representation of the Lorentz Group Classiﬁcation of Lorentz Transformations 289 We like to classify now the elements of L = O(3. det L = 1.34) j=1 j=1 .30) ≤ −1} .12) the case ρ = 0.24) or since (L1 0 )2 + (L2 0 )2 + (L3 0 )2 ≥ 0 it holds (L0 0 )2 ≥ 1. A second class property follows from (10. One can also 1 verify that one can write L↑ − L↓ + L↓ − = gL↑ + = = −L↑ . L↓ . A statement on this value can be made on account of property (10. For this purpose we consider ﬁrst the value of det L. + We start our investigation by demonstrating that L↑ forms a group.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↑ . Using det AB = det A det B and det AT = det A yields (det L)2 = 1 or det L = ±1 .10. L↑ contains 1 1. (10. L↑ . Properties (10. 1).14). 1). = { L. hence. σ = 0 yields L0 0 2 − L1 0 2 − L2 0 2 − L3 0 2 = 1. L0 0 ≥ 1} . 1). = { L. ∃M2 .31) (10. From this we can conclude L0 0 ≥ 1 or L0 0 ≤ −1 .23) One can classify Lorentz transformations according to the value of the determinant into two distinct classes. It holds g ∈ L and −1 ∈ L as one can readily verify testing for property (10.. L ∈ O(3.32) (10. + except for the trivial factors g and −1 It is. entirely suitable to investigate ﬁrst only 1. L↓ are disjunct sets deﬁned as follows + + − − L↑ + L↓ + L↑ − L↓ − = { L. L ∈ O(3. det L = −1. Lorentz transformations in L↑ .1).14) which we employ in the formulation (10.33) where we used the deﬁnition aM = {M1 . L ∈ O(3.26) (10. Obviously. there exist two other distinct classes. Schwartz’s inequality yields j=1 2 3 3 3 j=1 A jB 0 ≤ 0 j A 0 2 j Bj 0 2 .23) and (10.12). Considering in (10.e. 1). For this purpose we consider the value of C 0 0 = A0 µ B µ 0 = 3 A0 j B j 0 + A0 0 B 0 0 . + = − L↑ g + (10. det L = 1. (10.14).29) (10. L ∈ O(3.25) i. (10.

hence. (10. i.16). orthochronous Lorentz transformations.. . In the following we + will consider solely this subgroup of SO(3. Similarly. For the inverse of + ↑ any L ∈ L+ holds (10. + transformations in a small neighborhood of 1 which we parametrize by inﬁnitesimal parameters.35) One can conclude. obviously.34) provides then the estimate conclude from (10. (10.39) where we have employed the matrix form T 1 deﬁned as in (10. accordingly. 1 0 0 = ≥ 1 and. one can j=1 j=1 3 0 )2 = (A0 )2 − 1.e. these transformations can be parametrized such that a continuous variation of the parameters connects any element of L↑ with 1 This property will be exploited now in that we consider ﬁrst 1. ρ ν (10. therefore. 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 (10.12) follows (B 0 0 )2 − 3 (B j 0 )2 = 1 or 3 (B j 0 )2 = (B 0 0 )2 − 1. obviously j=1 A0 0 B 0 0 ≥ 1. Using the above expression for C 0 0 one can state C 0 0 > 0. we consider transformations Lµ ν = δ µ ν + µ ν .40) and.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 . The next group property of L↑ to be demonstrated is the existence of the inverse.1). 1 We will then employ the Lie group properties to generate all transformations in L↑ . | 3 A0 j B j 0 | < A0 0 B 0 0 . from which one can conclude L−1 ∈ L↑ . Property (10. 1 ∈ L↑ . δ00 = 1 and It should be noted that according to our present deﬁnition holds δµν = gµρ δ δ11 = δ22 = δ33 = −1. + i.14) implies 1 + T g (1 + ) = g 1 1 (10. Since A0 0 ≥ 1 and B 0 0 ≥ 1..14) actually L0 0 ≥ 1 and det L = 1 must hold. since the group property of O(3. In fact.37) (10.1) ascertains CT gC = g it must hold C 0 0 ≥ 1.36) It holds.290 Relativistic Quantum Mechanics From (10.38) For these transformations. To order O( 2 ) holds g + g = 0. if we enforce (10. (10. This relationship shows (L−1 )0 0 = L0 0 . L↑ is called the subgroup of proper. Since the associative property holds for matrix multipli1 1 + ↑ cation we have veriﬁed that L+ is indeed a subgroup of SO(3. holds L0 0 > 0 and the value of the determinant is close to unity.13). + Accordingly.e. µ ν small . Inﬁnitesimal Lorentz transformations The transformations in L↑ have the property that they are continously connected to the identity 1 1.

k j j = 1. other ﬁve parameters zero) Kk = (wk = 1. w3 ) = −w2 ϑ3 0 −ϑ1 −w3 −ϑ2 ϑ1 0 (10.48) (10. ϑ2 .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 . 3 (10.10.46) 0 0 0 0 .42) This relationship implies µ 0 j µ j = 0 = j 0 . j.50) . K2 = . K ] The operators also obey J · K = J1 J1 + J2 J2 + J3 J3 = 0 = = k m Jm k m Jm 0 −1 0 0 0 0 0 0 (10. ϑ3 .49) . 3) Jk = (ϑk = 1.1: Natural Representation of the Lorentz Group Using (10. J2 = 0 0 0 −1 0 0 0 0 0 0 0 0 0 0 0 −1 0 1 .43) k = − Inspection shows that the matrix has 6 independent elements and can be written 0 −w1 −w2 −w3 −w1 0 −ϑ3 ϑ2 . 2. K ] [ Jk . 2. J = 0 3 0 1 0 0 0 0 0 0 0 0 0 0 0 (10. The generators are explicitly 0 0 0 0 0 0 0 0 J1 = 0 0 0 −1 0 0 1 0 (10.15) one can conclude T 291 = −g g (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 . other ﬁve parameters zero) .47) 0 −1 0 0 0 0 −1 0 0 −1 0 0 0 0 0 0 0 0 K1 = . = − k m Km (10. w2 . (10. k = 1. (ϑ1 . 2. 3 . w1 .44) This result allows us now to deﬁne six generators for the Lorentz transformations(k = 1. J ] [ Kk .45) (10.

2. and using the relationship Jk = 0 0 0 Lk (10.35) it issuﬃcient to consider in the present case the 2 × 2–matrix L = exp since w1 0 −1 −1 0 ∞ = n=0 n w1 n! 0 −1 −1 0 0 0 .1: Demonstrate the commutation relationships (10. k = 1.52) where the 3 × 3–matrices Lk are the generators of SO(3) deﬁned in Chapter 5.55) L 0 0 0 0 exp (w1 K1 ) = exp Using the idempotence property 0 −1 −1 0 2 0 0 1 0 (10. w) = exp ϑ · J + w · K . however. A boost in the x1 –direction is L = exp(w1 K1 ). that the transformations (10.54) which.49.49) deﬁne the Lie algebra associated with the Lie group L↑ .. constitutes an overcomplete parametrization of the rotations (see Chapter 5). We consider now the Lorentz transformations for ϑ = 0 which are referred to as ‘boosts’. For the parameters ϑk of the Lorentz transformations holds obviously ϑk ∈ [0. 3 (10. 0 1 n (10. In analogy to equation (5. i. Relativistic Quantum Mechanics The commutation relationships (10. 2. w = 0) = 0 0 0 R(ϑ) . Following the treatment of the rotation group SO(3) one can express the elements of L↑ through + the exponential operators L(ϑ.53) Here R(ϑ) are the 3 × 3–rotation matrices constructed in Chapter 5.50). + The commutation relationships imply that the algebra of the generators Jk .51) One can readily show.292 as can be readily veriﬁed. 10. 3 is closed.51) for w = 0 correspond to rotations of the spatial coordinates.56) = 1 0 0 1 = 1 1 (10.57) . L(ϑ.e. To determine the explicit form of this transformation we evaluate the exponential operator by Taylor expansion. 2π[ . Exercise 7. k = 1. (10. Kk . 3 where we have deﬁned ϑ · J = k=1 ϑk Jk and w · K = following the algebra in Chapter 5. w ∈ R3 3 k=1 wk Kk . ϑ. (10.

64) We note that the range of wk -values is not a compact set even though the range of vk -values is compact. 10. Correspondingly. 10.61) √ √−v1 √ exp (w1 K1 ) = 2 1 − v1 1 2 1 − v1 0 0 0 0 0 0 0 0 1 0 0 1 (10.51) the explicit transformation for space–time–coordinates is then x1 = x1 − v1 t 1 − 2 v1 . one obtains from (10.59).59) for the case k = 1 corresponds to the relative velocity of two frames of reference.1: Natural Representation of the Lorentz Group one can carry out the Taylor expansion above: ∞ 293 L = n=0 2n w1 1 + 1 (2n)! ∞ n=0 2n+1 w1 (2n + 1)! 0 −1 −1 0 = cosh w1 −sinh w1 −sinh w1 cosh w1 . Accordingly. This property is. however. (10. (10.59) cosh w1 Using cosh2 w1 − sinh2 w1 = 1 one can identify sinhw1 = sinh2 w1 + 1. cosh2 w1 − 1 and coshw1 = cosh2 w1 − 1 = cosh w1 sinh w1 sinh2 w1 + 1 .58) = cosh w1 1 + sinh w1 1 0 −1 −1 0 The conventional form (10. for wk wk ∈ ] − ∞. in fact. x3 = x3 (10. we can state vk ∈ ] − 1. 1[.59) follows. vk deﬁned in (10. .63) which agrees with (10.6. From (10. This property of the wk -values contrasts with the property of the parameters ϑk specifying rotational angles which assume only values in a compact range.56. x2 = x2 . (10.1) of the Lorentz transformations is obtained through the parameter change sinh w1 v1 = = tanh w1 (10. ∞[ . The range of the parameters wk can now be speciﬁed.3. 10. consistent with (10. We expect that vk can only assume values less than the velocity of light c which in the present units is c = 1.62) According to (10.1). t = t − v1 x1 2 1 − v1 .59) can be written 1 2 1 − v1 √−v1 2 1 − v1 2 1 − v1 .59) v1 = These two equations yield cosh w1 = 1/ and (10. (10.10.60) sinh w1 = v1 / 2 1 − v1 .

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

The respective vectors aµ are associated with the 4-vectors aµ ..e.66). −a2 .e. and . xν = g νσ Lρ σ gρµ x µ . gσν xν = Lρ σ gρµ x µ and g µσ gσν = δ µ ν .. We start from x µ = Lµ ν xν . We do not fact. This property follows from (10.2: Scalars. We have duplicated the expression for the inverse of Lµ ν to obtain the correct notation in terms of covariant. accordingly. Multiplication (and summation) of x µ = Lµ ν xν by Lρ σ gρµ yields. (10. one can employ the tensors g µν to raise and lower indices of L ν establish here the consistency of the ensuing notation.74) g µν aν .12).16).72) where aν is a vector with transformation behaviour as stated in (10. 4–Vectors and Tensors Scalar Product then 295 4-vectors allow one to construct scalar quantities. using (10. (10.12) a gµν b µ ν = Lµ ρ gµν Lν σ aρ bσ = aρ gρσ bσ (10. In fact. one can express [(L−1 )T ]µ ν = (L−1 )ν µ and.73) Note that according to (10. g µν ∂ν = Lµ ρ g ρσ ∂σ . aµ (10.17) can be written (L−1 )ν µ = Lµ ν . i. i. −a3 ) (10. −a1 .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 . The 4-Vector ∂µ (10. lower.70) is a scalar. ∂t (10. ∂x µ (10.17) Lµ σ is the transformation inverse to Lσ µ . We like to point out that from deﬁnition (10.76) An important example of a covariant 4-vector is the diﬀerential operator ∂µ = ∂ = ∂xµ ∂ .72) of the covariant 4-vector follows = In µν and g µ as well. In any case one can express (10.10.79) This is the inverse Lorentz transformation consistent with (10. (10. If aµ and bµ are 4-vectors aµ gµν bν (10.77) The transformed diﬀerential operator will be denoted by ∂µ = def ∂ . Covariant 4-vectors transform like a µ = gµν Lν ρ g ρσ aσ where we deﬁned g µν = gµν . One calls 4-vectors aµ covariant and 4-vectors aµ contravariant.75) (10.71) Contravariant and Covariant 4-Vectors It is convenient to deﬁne a second class of 4-vectors. the relationship being aµ = gµν aν = (a0 .66) together with (10.

2 dt dτ 1 − v (10. i.82) (10.e. 1 − v2 1 − v 2 dt (10. For this purpose we deﬁne ﬁrst the contravariant diﬀerential operator ∂ µ = g µν ∂ν = Then the operator 2 ∂µ ∂ µ = ∂t − 2 (10.85) 1 d d = √ .83) is a scalar under Lorentz transformations.66).84) = 1 − (v)2 (10. ∂µ does indeed transform like a covariant vector. indices.g.87) The remaining components of pµ can be written. 2 2 dt 1 − v 1 − v (10..86) One can write (10.e. d’Alembert Operator We want to construct now a scalar diﬀerential operator.89) .67) transforms like (10. e. In fact. (10. For this purpose we consider the scalar diﬀerential (dτ )2 = dxµ dxµ = (dt)2 − (dr)2 It holds dτ dt from which follows 2 (10. p1 = √ m v1 m dx1 = √ . Proof that pµ is a 4-vector We will demonstrate now that the momentum 4-vector pµ deﬁned in (10.296 Relativistic Quantum Mechanics contravariant. (10.80) Multiplication by g λµ (and summation over µ) together with g λµ gρµ = δ λ ρ yields g λµ ∂µ = Lλ σ g σν ∂ν .88) One can express then the momentum vector pµ = √ d m dxµ = m xµ .81) ∂ . dτ 1 − v 2 dt p0 = E = √ m m dt = √ .− ∂t . i. 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.. upper.79) allows us to determine the connection between ∂µ and ∂µ .. this operator is equal to the d’Alembert operator which is known to be Lorentz-invariant.

and. This gauge is not Lorentz-invariant and we will adopt here another gauge.. yields according to (10. Since xµ transforms like a contravariant 4-vector.96) is a scalar and.s. 8.97) and.12).94) 1 m In the non-relativistic limit the rest energy m is the dominant contribution to E. hence. in (10.10. is a scalar. We will demonstrate now the 4-vector property of Aµ . Lorentz Gauge In our previous description of the electrodynamic ﬁeld we had introduced the scalar and vector potential V (r. ∂t V (r. respectively.s. t) and A(r. of (10.77) which allow one to express (10. t). the so-called Lorentz gauge.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. One obtains E = ±m ± p2 2m (p 2 )2 + O 4m3 (p 2 )3 4m5 . (10. The momentum 4-vector allows us to construct a scalar quantity. . 10. in fact.91) (10.h.95) 2 p This obviously describes the energy of a free particle with rest energy ±m.69) is.h. kinetic energy ± 2m and relativistic corrections. a contravariant 4-vector then the l. (10. for these potentials.h. Lorentz-invariant.96) The Lorentz-invariance of this gauge. the r.67) E2 − p 2 = or pµ pµ = m2 which.3: Relativistic Electrodynamics 297 d The operator m dτ transforms like a scalar. i. t) = 0 .92) (10.93) (10.89) must be a 4-vector.96) in the form ∂µ Aµ = 0 . · A = 0. of (10. can be demonstrated readily using the 4-vector notation (10.s.69) for the electrodynamic potential and the 4-vector derivative (10. in fact. and had chosen the so-called Coulomb gauge (8. We like to rewrite the last result E 2 = p 2 + m2 or E = ± p 2 + m2 . hence. equivalently. Expansion in should then be rapidly convergent.e. namely pµ pµ = pµ gµν pν = E 2 − p 2 Evaluation of the r. t) + · A(r. namely. (10.97) We have proven already that ∂µ is a contravariant 4-vector. in (10. If we can show that Aµ deﬁned in (10.h. pµ on the l.89) alltogether transforms like a contravariant 4-vector.s.90) (10. m2 m2 v 2 − = m2 1 − v2 1 − v2 (10.

s. · A(r. (10. . We want to derive now the diﬀerential equations which determine the 4-potential Aµ in the Lorentz gauge (10.104) where mn = mn is the totally anti-symmetric three-dimensional tensor deﬁned in (5.96). must be a scalar. 2. Consequently. Since ∂µ transforms like a covariant 4-vector. 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 .101) Combining equations (10.6) and (8. (8. F µν = x Ey B z 0 −Bx Ez −By Bx 0 Alternatively. t) together with (10.s. using (10. t) .s.98) which reﬂects the principle of charge conservation. has to transform like a contravariant 4-vector. one obtains 2 ∂t V (r.96). t) from (8.101). n = 1. t) = 4 π J(r.77) and (10. t) = −∂t V (r. a 4-vector.h. t) − 2 A(r.97) and. i.98) can be written.103) B .102) In this equation the r. that F µν can be expressed through the potential Aµ in (10. t) and current density J(r.82) as follows F µν = ∂ µ Aν − ∂ ν Aµ . The l.99) Since this equation must be true in any frame of reference the right hand side must vanish in all frames. 3 (10. one obtains for A(r. t). The respective equation for A0 = V can be obtained from Eq. (10.298 Transformation Properties of J µ and Aµ Relativistic Quantum Mechanics The charge density ρ(r. t) = ∂t · A(r. (10. t) 2 ∂t A(r.100. t) − 2 V (r. t) = −∂t V (r. transforms like a 4-vector. t) . Since ∂µ ∂ µ transforms like a scalar one can conclude that Aν (xσ ) must transform like a 4-vector.9). Using · ∂t A(r. it follows that J µ .105) . thereby. must transform likewise. using (10..17) using the identity (8. · A(r. yields ∂µ ∂ µ Aν (xσ ) = 4 π J ν (xσ ) . of (10. 10.18) and.e. in fact.69). also the l. in fact.32). i. k. using (8..e. Equation (10. This principle should hold in any frame of reference.h. t) + · J(r. this can be stated F k0 = −F 0k = E k .100) Similarly. t) = 4πρ(r. m.83) and (10.99) must be a scalar.13). (10.68). ∂µ J µ (xµ ) = 0 . t) are known to obey the continuity property ∂t ρ(r. prove that Aµ is. (10. One can readily verify. t) = 0 (10.69) and ∂ µ in (10. F mn = − mn (10. according to (10.h.

97).108) yields then the expressions for the transformed ﬁelds E and B .113) which can be shown to be equivalent to the two homogeneous Maxwell equations (8. t) and B(r. B and E⊥ .104) establishe the transformation behaviour of E(r. 10. one obtains Ey −v B Ez +v B 0 −Ex − √ 1 2z − √ 1 2y 1−v1 1−v1 Bz −v1 Ey By +v1 Ez √ 2 Ex 0 − √ 2 1−v1 1−v1 F µν = E −v B (10. Maxwell Equations in Lorentz-Invariant Form One can express the Maxwell equations in terms of the tensor F µν in Lorentz-invariant form.3: Relativistic Electrodynamics 299 The relationships (10.109) (10.3.4). B⊥ are.110) where E .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. 8.2). where we used (10. one can conclude from (10. the components of the ﬁelds parallel and perpendicular to the velocity v which determines the Lorentz transformation. t). .62) or.105) of the tensor F µν one can conclude the property ∂ σ F µν + ∂ µ F νσ + ∂ ν F σµ = 0 (10.102) ∂µ F µν = 4π J ν .103. Noting ∂µ F µν = ∂µ ∂ µ Aν − ∂µ ∂ ν Aµ = ∂µ ∂ µ Aν − ∂ ν ∂µ Aµ = ∂µ ∂ µ Aν . by (10. B .63). respectively. These equations show that under Lorentz transformations electric and magnetic ﬁelds convert into one another. E⊥ = B⊥ E⊥ + v × B √ 1 − v2 B⊥ − v × E = √ 1 − v2 (10.105) and (10. (10. equivalently.10. In a new frame of reference holds F µν = Lµ α Lν β F αβ (10. From the deﬁnition (10. 8.1. The results can be put into the more general form E B = = E .112) (10.106) In case that the Lorentz transformation Lµ ν is given by (10.111) One can readily prove that this equation is equivalent to the two inhomogeneous Maxwell equations (8.

120) (10.104).e.. It holds for a particle with 4-momentum pµ as given by (10. hence. i. Equation (10.118) p· dt √ where we exploited the fact that according to p = mv/ 1 − v 2 holds p v.93)..5).120) is referred to as the Lorentz force.67) and charge q dpµ q = pν F µν dτ m (10.114) is. 1 dE 2 1 dp 2 = = qp·E 2 dt 2 dt (10.g. e. We want to √ demonstrate now that this equation is equivalent to the equation of motion (8. using (10.114). E = m/ 1 − v 2 .h.86). provides an alternative description of the action of the Lorentz force. The µ = 0 component of (10. employing (10. The term on the r.115) This equation follows. (10. To avoid √ confusion we will employ in the following for the energy of the particle the notation E = m/ 1 − v 2 [see (10.116) (10.s. yields dpx = q dt Ex + (v × B)x (10. dt E From this one can conclude.103) dpx q = ( EEx + py Bz − pz By ) . For the spatial components. dE q = p·E dτ m or with (10.119) which is the x-component of the equation of motion (8. also from the equation of motion (8.67).114) reads using (10. We have. . demonstrated that (10.86) and (10.86) dE q = p·E .5).114) reads then. hence. however. for µ = 1.5) where p = mv/ 1 − v 2 . in fact.117) (10.5) taking the scalar product with p dp = qp · E (10.114) where d/dτ is given by (10. of (10. equivalent to (8.87)] and retain the deﬁnition E for the electric ﬁeld. dτ m √ Employing again (10.300 Lorentz Force Relativistic Quantum Mechanics One important property of the electromagnetic ﬁeld is the Lorentz force acting on charged particles moving through the ﬁeld. We want to express this force through the tensor F µν .

and which can be constructed following the procedure adopted for the function space representation of SO(3). K3 ) are the generators of L(ϑ. In analogy to the situation for SO(3) we seek an expression for ρ(Lµ ν ) in terms of an exponential operator and transformation parameters ϑ. xµ ).4 Function Space Representation of Lorentz Group In the following it will be required to decribe the transformation of wave functions under Lorentz transformations. w)) = ρ eϑ·J + w·K which we present in the form L(ϑ. taken at the pairs of points (x µ = Lµ ν xν . J3 ) and K = (K1 .124) In this expression J = (J1 .e. obey a diﬀerent transformation law.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 .10. the charge-current density in case of Sect.121) covers solely the transformation behaviour of scalar functions. (10.127) .g. w.e. e. For such functions holds in the transformed frame ψ (Lµ ν xν ) = ψ(xµ ) (10. 10. i.122) This result gives rise to the deﬁnition of the function space representation ρ(Lµ ν ) of the Lorentz group (ρ(Lµ ν )ψ)(xµ ) = ψ((L−1 )µ ν xν ) . We will denote the intended representation by L(ϑ. Functions which represent 4-vectorsor other non-scalar entities. Accordingly. 10.. (10. w) = exp ϑ · J + w · K .123) This deﬁnition corresponds closely to the function space representation (5. in as far as SO(3. For this purpose one replaces xµ in (10. ψ ∈ C∞ (4). We need to emphasize that (10. in the present case we exclude the factor ‘−i’ [cf. (10. K2 . The resulting expression should be a generalization of the function space representation (5. We like to express now ψ (x µ ) in terms of the old coordinates xµ .48) and (10..48) of SO(3).4: Function Space Representation of Lorentz Group 301 10. i.121) by (L−1 )µ ν xν and obtains ψ (xµ ) = ψ((L−1 )µ ν xν ) . J2 .1) is a generalization (rotation + boosts) of the group SO(3). However.47).126) ρ ewk Kk − 1 1 .42) of SO(3). (5.125) def (10. one can evaluate Jk as follows Jk = lim and Kk Kk = lim wk →0 ϑk →0 1 ϑ1 1 w1 ρ eϑk Jk −1 1 (10.. In this section we will investigate the transformation properties of scalar functions ψ(xµ ).3 or the bi-spinor wave function of electron-positron pairs in Sect.7.125)].51) for the natural representation of the Lorentz group. w) = ρ(Lµ ν (ϑ. we seek an expression which corresponds to (10. def (10. w) which correspond to the generators Jk and Kk in (10.

i. (10.128) eϑ1 J1 = e−ϑ1 J1 1 0 = 0 0 0 0 0 1 0 0 0 cosϑ1 sinϑ1 0 −sinϑ1 cosϑ1 (10. 1 This result.131) This yields for small w1 ρ ew1 K1 ψ(xµ ) = ψ(coshw1 x0 + sinhw1 x1 .134) . namely [J1 . −sinϑ1 x2 + cosϑ1 x3 ) = ψ(xµ ) + ϑ1 (x3 ∂2 − x2 ∂3 ) ψ(xµ ) + O(ϑ2 ) . J2 ] = [x3 ∂2 − x2 ∂3 . x1 ∂3 ] − [x2 ∂3 . J2 = x1 ∂3 − x3 ∂1 . K ] = = k m Jm k m Jm = − k m Km . K2 ] = −J3 . (10. 0 1 (10. x3 ) 2 = ψ(xµ ) + w1 (x1 ∂0 + x0 ∂1 ) ψ(xµ ) + O(w1 ) (10.49) as the generators of the natural representation. J2 ] = J3 . x2 .126). In order to evaluate (10. sinhw1 x0 + coshw1 x1 . obviously. [ Jk . and [J1 . reproduces the expression for J1 in (10.129) which yields for small ϑ1 ρ eϑ1 J1 ψ(xµ ) = ψ(x0 . x1 ∂3 − x3 ∂1 ] = [x3 ∂2 . One can determine similarly K1 starting from coshw1 sinhw1 sinhw1 coshw1 = 0 0 0 0 0 0 1 0 0 0 . J ] [ Kk . cosϑ1 x2 + sinϑ1 x3 .e. which we like to demonstrate for J1 and K1 . K ] [ Jk .133) We demonstrate this for three cases. the expression for K1 in (10. [K1 . J3 = x2 ∂1 − x1 ∂2 . K obey the same Lie algebra (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 (10.130) ew1 K1 −1 = e−w1 K1 (10.132) and.302 One obtains J1 = x3 ∂2 − x2 ∂3 . x3 ∂1 ] = −x1 ∂2 + x2 ∂1 = J3 . x1 . The generators J . obviously.128). K2 ] = K3 : [ J1 .

303 (10. x2 + x2 + x2 1 2 3 in the case of SO(3) and (x0 )2 − (x3 )2 in case of transformation (10.139) x3 = R sinhΩ (10. 0 ∂cothΩ ∂x x = ∂R x0 = − ∂x3 R (10. cothΩ = 3 .e. [ J1 . ∂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 . Ω which are connected with x0 . Eqs.137) where K3 is given in (10.87)].10.140) allow one to ∂ ∂ express the derivatives ∂0 . (10. ∂3 in terms of ∂R . (5. K2 ] = [x0 ∂1 + x1 ∂0 .138) x3 x0 .141) (10. One-Dimensional Function Space Representation The exponential operator (10.125) in the case of a one-dimensional transformation of the type L(w3 ) = exp w3 K3 .136) (10. 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 .4: Function Space Representation of Lorentz Group [ K1 . For this purpose one expresses K3 in terms of hyperbolic coordinates R. can be simpliﬁed considerably.128). The relationships (10.139. 10. x2 ∂0 ] − [x1 ∂0 .140) x0 x The transformation to hyperbolic coordinates closely resembles the transformation to radial coordinates for the generators of SO(3) in the function space representation [cf.85-5. x0 ∂2 ] = x3 ∂0 + x0 ∂3 = K3 .137). x0 ∂2 + x2 ∂0 ] = [x0 ∂1 . K2 ] = [x3 ∂2 − x2 ∂3 . In the following we consider solely the case x0 ≥ 0. a relationship which can also be stated R = and + − (x0 )2 − (x3 )2 (x0 )2 − (x3 )2 if x0 ≥ 0 if x0 < 0 (10.135) (10. x3 as follows x0 = R coshΩ . x2 ∂0 ] − [x2 ∂3 . x0 ∂2 ] = −x2 ∂1 + x1 ∂2 = − J3 .143) . In both cases the radial coordinate is the quantity conserved under the transformations. ∂Ω . x0 ∂2 + x2 ∂0 ] = [x3 ∂2 . i.. We note tanhΩ = ∂R x0 = .142) (10.

150) . ˆ pµ =⇒ pµ = i(∂t . − ) = i ∂ µ . ) = i ∂µ . Further below we will provide a more systematic. is not very insightful.149) where ψ(xµ ) is a scalar. extremely useful. (10. Free Particle Case A quantum mechanical description of a relativistic free particle results from applying the correspondence principle. (length)−1 .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. 10. which allows one to replace classical observables by quantum mechanical operators acting on wave functions.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 ) . (10. momentum. The historic route to these two equations. ˆ (10. but certainly is short and. complex-valued function.122). The latter property implies that upon change of reference frame ψ(xµ ) transforms according to (10. in 4-vector notation reads pµ =⇒ pµ = i (∂t .151) reads ∂µ ∂ µ + m2 ψ(xµ ) = 0 (10. In natural units the Klein–Gordon equation (10.92) one obtains the wave equation − or 2 2 µ ∂ ∂µ ψ(xν ) = m2 ψ(xν ) ψ(xν ) = 0 .146) Applying the correspondence principle to (10. and (time)−1 all have the same dimension.304 Inserting these results into the deﬁnition of K3 in (10.147) (10. The partial diﬀerential equation (10. mass.145) ∂Ω 10.144) The action of the exponential operator (10. In these units the quantities energy.121. therefore. representation theoretic treatment.148) ∂ µ ∂µ + m2 (10.128) yields K3 = x0 ∂3 + x3 ∂0 = ∂ . We will provide a ‘derivation’ of these two equations which stem from the historical development of relativistic quantum mechanics.151) is called the Klein-Gordon equation. however. In the position representation the correspondence principle states ˆ E =⇒ E = − ∂t i ˆ p =⇒ p = i which. In the following we will employ so-called natural units = c = 1. ∂Ω Relativistic Quantum Mechanics (10. namely the Klein–Gordon and the Dirac equations.

t) (10. t) = ψ ∗ (r.153) is associated with a conservation law for particle probability ∂t ρS (r.160) ∂t (ψ ∗ ∂t ψ which corresponds to − ψ∂t ψ∗) + · (ψ ψ∗ ∂t ρKG (r. t) = − 1 2m 2 ψ(r.159) (10. and that (as postulated) ψ(xµ ) is a scalar. t) ] 2mi (10. t) − ψ(r. t) ) 2m 1 = ( ψ ∗ (r.151) One can notice immeadiately that (10. t) ψ(r. t) − ψ(r. t)∂t ψ(r. In order to obtain the conservation law connected with the Klein–Gordon equation (10. (10.154). t) − ψ(r. 2mi (10. t) ) . 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.155) describes the positive deﬁnite probability to detect a particle at position r at time t and where jS (r. Under Lorentz transformations the free particle Klein–Gordon equation (10.150) one considers Im ψ ∗ ∂µ ∂ µ + m2 ψ = 0 (10. t) ψ(r. t) 1 [ ψ ∗ (r. t) jKG (r. t) = · jKG (r. t) ψ ∗ (r.150) becomes ∂µ ∂ µ + m2 ψ (x µ ) = 0 (10. t)∂t ψ ∗ (r.156) · jS (r. t) = 0 i ( ψ ∗ (r. t) ψ ∗ (r.5: Klein–Gordon Equation or 2 ∂t − 2 305 + m2 ψ(xµ ) = 0 .150) is invariant under Lorentz transformations.161) . t) = 0 (10.158) which yields 2 2 ψ ∗ ∂t ψ − ψ∂t ψ ∗ − ψ ∗ 2ψ +ψ ψ∗ − 2ψ∗ = ψ) = 0 (10. t) = (10. t) + where ρKG (r.10.154) describes the current density connected with motion of the particle probability distribution. This follows from the fact that ∂µ ∂ µ and m2 are scalars. t) ψ(r.157) 2m which is equivalent to (10. 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) + where ρS (r.152) which has the same form as the Klein–Gordon equation in the original frame.

162) Inserting this into the Klein–Gordon equation (10. t) = 0 (10. The proper interpretation of ρKG (r. it had been realized later. (10. not of particle probability. λ|xµ ) = N e−iλmt . is actually that of a charge density.165) This result together with (10. and of negative energies E ≤ −m. the anti-particles carry a charge opposite to that of the associated particles. o namely. ±) = ± 2 m2 + po (10. The proper interpretation of the two cases is that the Klein–Gordon equation describes particles as well as anti-particles.306 Relativistic Quantum Mechanics This conservation law diﬀers in one important aspect from that of the Schr¨dinger equation (10.161) and (10.p ei(p·r − λEo (p)t) Eo (p) = The spectrum of the Klein–Gordon equation (10.150) is a continuum of positive energies E ≥ m. for example pions. in that the expression for ρKG is not positive deﬁnite.. λ = ± .166) ψo (p.163) which results in the expected [see (10. λ|xµ ) (10.162) shows that the solutions of the free particle Klein-Gordon equation (10. i. corresponding to λ = −.168) . λ) actually describes charge density rather than probability. m 2 2 E0 = m2 + po . λ = ± (10.154). Generating a Solution Through Lorentz Transformation A particle at rest. We denote this by summarizing the solutions as follows ∂µ ∂ µ + m2 ψo (p.150) are actually determined by po and by the choice of sign ±.167) m which is positive for λ = + and negative for λ = −.93] dispersion relationship connecting E0 . Today. 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. the Klein-Gordon equation is considered as a suitable equation to describe spin–0 particles. λ) associated with the corresponding wave functions ψo (p. (10.166) is Eo (p) ∗ ρKG (p. λ|xµ ) according to (10. corresponding to λ = +. and the density ρKG (p. according to (??) is decribed by the r–independent wave function ψo (p = 0. p0 . The density ρKG (p.151) yields 2 2 Eo − p0 − m2 ψ(r. λ|xµ ) = 0 2 m2 + po . t). λ|xµ ) = Nλ. with p = 0. Solution of the Free Particle Klein–Gordon Equation Solutions of the free particle Klein–Gordon equation are ψ(xµ ) = N e−ipµ x µ = N ei(p0 ·r − Eo t) .164) The corresponding energy is Eo (po .e. λ) = λ ψo (p. λ|xµ )ψo (p. (10.

of (10. t)) . In case of λ = + one obtains ﬁnally L(w3 )ψo (p = 0.166)] of the particle.174) which agrees with the expression given in (10.6: Klein–Gordon Equation with Electromagnetic Field 307 We want to demonstrate now that the wave functions for p = 0 in (??) can be obtained through appropriate Lorentz transformation of (10. one has to interprete p = mv/ 1 − v 2 (10. 2 2 1−v 1−v (10. for the wave function (10.168). Aµ (xµ ) = (V (r. λ = +|xµ ) = N ei(px 3 − Eo (p)x0 (10. (10.169) The addition theorem of hyperbolic functions cosh(Ω + w3 ) = coshΩ coshw3 + sinhΩ sinhw3 allows us to rewrite the exponent on the r. (10. λ = −|xµ ) = N ei(px 3 + Eo (p)x0 . for positive energy solutions.173) (10.s.145) choosing m = sinhw3 . λ|xµ ) = = exp w3 ∂ ∂Ω N e−iλmRcoshΩ 3) N e−iλmRcosh(Ω+w .176) 10.e. (10. For this purpose we consider the wave function for a particle moving with momentum velocity v in the direction of the x3 –axis.138) and the relationships (10.171) One can interpret then for λ = +.166).138) to replace t = x0 by hyperbolic coordinates R.172) as the energy [c.f. t).170) The coordinate transformation (10. L(w3 )ψo (p = 0. This yields.168).169) −iλ ( m coshw3 ) ( R coshΩ ) − iλ ( m sinhw3 ) ( R sinhΩ ) . −A(r. A(r.h. for negative energy solutions.. i. i.. t).10.175) as the momentum of the particle and one obtains L(w3 )ψo (p = 0.61) yield for this expression −iλ √ m mv x0 − iλ √ x3 . Ω. (10. t)) (10.e. using (10. p = − mv/ 1 − v 2 as the momentum of the particle relative to the moving frame and √ m = 1 − v2 m2 = 1 − v2 m2 + m2 v 2 = 1 − v2 m2 + p2 = Eo (p) (10. Such wave function should be generated by applying the Lorentz transformation in the function space representation p (10. in fact.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.177) . In case of λ = −.

308 free classical particle classical particle in ﬁeld (V. (10. A) or Aµ = (V. t) = −i − q A(r.e.150). t). See also Eq.1: Coupling of a particle of charge q to an electromagnetic ﬁeld described by the 4-vector potential Aµ = (V.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. i.69)] one replaces the quantum mechanical operators.1. t) = e−imt Ψ(r. For this purpose one writes the Klein-Gordon equation (10. t) .e. According to the so-called minimum coupling principle the presence of the ﬁeld is accounted for by altering energy. it’s rest energy. 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. momenta for classical particles and the respective operators for quantum mechanical particles in the manner shown. (10.1 this becomes g µν (i∂µ − qAµ )(i∂ν − Aν ) ψ(xµ ) = m2 ψ(xµ ) which can also be written g µν πµ πν − .150) −g µν (i∂µ )(i∂ν ) + m2 ψ(xµ ) = 0 . In terms of space-time derivatives this reads (i∂t − qV (r. We will also assume.180) + m2 ψ(r. i. For this purpose we split-oﬀ a factor exp(−imt) which describes the oscillations of the wave function due to the rest energy. in . (10.147). according to the rules shown in Table 10.182) and seek an equation for Ψ(r.. i∂t and −i in (10. According to the principle of minimal coupling [see (10.93). and focus on the remaining part of the wave function. m2 ) ψ(xµ ) = 0 . A) free quantum particle quantum particle in ﬁeld (V. t) . is much larger than all other energy terms. 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 10.179) (10. (10. t) 2 (10. t))2 ψ(r. that the mass m of the particle. i. −A). which couples a particle of charge q to an electromagnetic ﬁeld described through the potential Aµ (xν ). in keeping withnthe non-relativistic limit.. albeit in a form..e.178) According to the replacements in Table 10. we deﬁne ψ(r.

s. The approximation is justiﬁed on the ground of the inequalities (10.184) where we neglected all terms which did not contain factors m.183) The term on the l. The Klein-Gordon equation (10. larger than |i∂t Ψ/Ψ| and alrger than qV . (10. This application demonstrates that the Klein–Gordon equation describes spin zero particles.181) to a physical system we consider pionic atoms.. 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. i.6: Klein–Gordon Equation with Electromagnetic Field particular. i.10. Ψ |q V | << m . identical to the Schr¨dinger equation (10..181) can then be approximated as follows: (i∂t − qV )2 e−imt Ψ = = (i∂t − qV ) (me−imt Ψ + e−imt i∂t Ψ − qV e−imt Ψ) m2 e−imt Ψ + me−imt i∂t Ψ − qV e−imt Ψ +me−imt i∂t Ψ − e−imt ∂ 2 Ψ − qV e−imt i∂t Ψ −me−imt qV Ψ − e−imt i∂t qV Ψ + q 2 V 2 e−imt Ψ ≈ m2 e−imt Ψ − 2mqV e−imt Ψ − 2me−imt i∂t Ψ (10. The process of π − meson capture involves the so-called Auger eﬀect. (10.181) reads then i ∂t Ψ(r. t) ≡ 0 r + Ze2 r 2 + 2 − m2 π ϕ(r ) = 0 ..185) This is. To ‘manufacture’ pionic atoms. e.g. 309 (10. atoms in which one or more electrons are replaced by π − mesons.187) We want to investigate in the following a description of a stationary state of a pionic atom involving a nucleus with charge +Ze and a π − meson. t) = ˆ [p − q A(r. Pionic Atoms To apply the Klein–Gordon equation (10.183). t) = − Ze and A(r. t)]2 + qV (r.h. t) 2m Ψ(r. π − mesons are generated through inelastic proton–proton scattering p + p −→ p + p + π − + π + . (10. spin-0 mesons.188) Inserting this into (10.186) then are slowed down.189) . (10. | i∂t Ψ | << m . of (10. ﬁltered out of the beam and ﬁnally fall as slow pions onto elements for which a pionic variant is to be studied.181) yields (we assume now that the Klein–Gordon equation describes a 2 particle with mass mπ and charge −e) for qV (r. however. t) (10.e. A stationary state of the Klein–Gordon equation is described by a wave function ψ(xµ ) = ϕ(r ) e−i t .2) of a non-relativistic spin-0 particle o moving in an electromagnetic ﬁeld.

192) leads then to the ordinary diﬀerential equation d2 − dr2 ( + 1) − Z 2 e4 2 Ze2 + + r2 r 2 − m2 π R (r) = 0 . . 1.310 Relativistic Quantum Mechanics Because of the radial symmetry of the Coulomb potential we express this equation in terms of spherical coordinates r.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.. this quantity deﬁnitely must be an integer. (10.196) In this expression the number n deﬁned through n = n− −1 (10. 2n2 = 0. n − 1 . (10. φ) = ( + 1) Y m (θ. n = 1. φ) R (r) Y r m (θ. 2. φ) (10. φ) .198) .190) With this expression and after expanding ( + one obtains 2 ˆ d2 L2 − Z 2 e4 2 Ze2 − + + dr2 r2 r − m2 π r φ(r) = 0 . (10. (10.192) ˆ are spherical harmonics. . .193) (10.e. (10. θ. i. .195) The latter problem leads to the well-known spectrum En = − mπ (Ze2 )2 .194) Bound state solutions can be obtained readily noticing that this equation is essentially identical to 2 that posed by the Coulomb problem (potential − Ze ) for the Schr¨dinger equation o r d2 − dr2 ( + 1) 2mπ Ze2 + + 2mπ E r2 r R (r) = 0 (10. i. φ.199) ( + 1) − Z 2 e4 .191) m (θ. . The similarity of (10. .e. r r r r sin θ r sin θ Ze2 2 r ) (10. ..197) counts the number of nodes of the wave function.191) ˆ The operator L2 in this equation suggests to choose a solution of the type ϕ(r ) = where the functions Y in (10.194) and (10. 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 .

2 e2 −→ e2 mπ . results in the (10.. 2. 1. + (10. m) = mπ 1 + Z 2 e4 (n− − 1 2 + ( + 1 )2 − Z 2 e4 )2 2 n = 1.194) d2 λ( ) ( λ( ) + 1 ) 2 Ze2 − + + dr2 r2 r 2 311 − m2 π R (r) = 0 . .201) mπ Z 2 e4 m2 − m2 π π = − . = 0. 10. . .6: Klein–Gordon Equation with Electromagnetic Field and one can write (10.203) spectrum . . . . . .196) if one makes the replacement 2 values which can be obtained E −→ One obtains − m2 π . EKG = Using (10. . − + 1. . .10. . n = 0. . 2mπ 2 −→ λ( ) .204) reads 1 (10.197. 1.204) 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 ). . i. which corresponds to a bound = 0.e.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 ) . (10. 2mπ 2 (n + λ( ) + 1)2 (10.200) The bound state solutions of this equation should correspond to from (10. . . Introducing α = Z 2 e4 and 1 β = + 2 (10. . . 1. n − 1 m = − . .202) Solving this for state) yields (choosing the root which renders = 1 + mπ Z 2 e4 ( n + λ( )+1 )2 ≤ mπ . (10. .199) and deﬁning EKG (n.

through the correspondence principle. arises because the Klein–Gordon equation.e. 10. The latter spectrum shows.209). in fact. one with a positive-deﬁnite density.208. but in the present case rather the negative of the energy. the Klein–Gordon equation had been rejected since it did not yield a positive-deﬁnite probability density. 10.. . It must be pointed out here that does not denote energy. i. A more satisfactory Lorentz–invariant wave equation. Dirac succeeded.146) leads to the wave equation i∂t Ψ(r.7 The Dirac Equation Historically. (10. i. but not an operator iγ · as suggested by the principle of relativity (equivalence of space and time). the Dirac equation introduced below. His solution . Equations (10.206) 1 2 3 + O(Z 6 e12 ) . This derivative. It should feature then a diﬀerential operator of the type D = iγ µ ∂µ . is related to the equation E 2 = m2 + p 2 of the classical theory which involves a term E 2 .e.. a splitting of the six n = 2.208. Finally. however. Many attempts were made by theoretical physicists to ‘linearize’ the square root operator in (10. a feature which is connected with the 2nd order time derivative in this equation. Application of the correspondence principle (10. In order to describe electron spectra one must employ the Lorentz-invariant 1 wave equation for spin. however. m) ≈ m − mZ 2 e4 mZ 4 e8 − 2n2 2n3 1 + − 1− Relativistic Quantum Mechanics α α2 α2 α2 − + + + O(α3 ) . t) = ± m2 − 2 Ψ(r. t) (10. i. it does not agree with observations of the hydrogen spectrum.312 ≈ From this results for (10. Also.2 particles.209). 2n2 2βn3 8n4 4n4 (10. in turn. the second term the non-relativistic energy. and the third term gives the leading relativistic correction. but for a long time to no avail.e. However. the wave function changes sign under reﬂection. would have only a ﬁrst order time derivative. for example.209) which. The latter term agrees with observations of pionic atoms. = 1 states into groups of two and four degenerate states. is identical to the two equations (10.. Heuristic Derivation Starting from the Klein-Gordon Equation An obvious starting point for a Lorentz-invariant wave equation with only a ﬁrst order time derivative is E = ± m2 + p 2 . t) .208). are unsatisfactory since expansion of the square root operator involves all powers of the Laplace operator.207) Here the ﬁrst term represents the rest energy. 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. 10. the π − meson is a pseudoscalar particle.208) These two equation can be combined i∂t + m2 − 2 i∂t − m2 − 2 Ψ(r.204) EKG (n. 4n (10.

(10.213) We want to determine now the simplest algebraic realization of γ µ . one can impose the condition γ0 is hermitian .208). these descriptions ultimately merging with the true theory of matter. It turns out that no 4-vector of real or complex coeﬃcients can satisfy these conditions. γ 2 .212) where we have changed ‘dummy’ summation indices. ψ2 (xµ ). therefore. γ ν ]+ = 2 g µν (10. so far. 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 . . spin. 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 ). 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. but did not commute the. 3 are anti-hermitian .e.211) into (10. beautiful mathematical theory and that. However. 3 has imaginary eigenvalues ±i.210) where P = iγ µ ∂µ . Initially. the operator of the Klein–Gordon equation ∂µ ∂ µ + m2 = −( P + m ) ( P − m ) (10. strengthens the believe of many theoretical physicists today that the properties of physical matter ultimately derive from a. Inserting (10..215) . Starting point is to boldly factorize.211) Obviously. this was not so. In fact. the realization that the linearization can be carried out only if one assumes a 4-dimensional representation of the coeﬃcients γ µ . one route to important discoveries in physics is the creation of new mathematical descriptions of nature.214) From this follows that γ 0 has real eigenvalues ±1 and γ j . i.213) reads (γ µ )2 = 1 µ = 0 −1 µ = 1. unspeciﬁed algebraic objects γ µ and γ ν . ψd (xµ ). The discovery by Dirac. yet to be discovered. therefore.1 2 and anti-particles. (10. j = 1. γ j . according to (10.7: Dirac Equation 313 to the problem involved an ingenious step. For µ = ν condition (10. are associated with a positive-deﬁnite particle density. 2. 2.66). it was assumed that the 4-dimensional space introduced by Dirac could be linked to 4vectors. 2. We seek to identify the coeﬃcients γ µ .209).10. Accordingly. Properties of the Dirac Matrices Let us now trace Dirac’s steps in achieving the linearization of the ‘square root operator’ in (10. γ 1 . j = 1. Instead. 3 . quantities with the transformation law (10. (10. namely. the quantities γ 0 . . the 4-dimensionsional representation discovered by Dirac involved new physical properties.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. exploited ∂µ ∂ν = ∂ν ∂µ . achieved through a beautiful mathematical theory. .

213) is satisﬁed. The Adjoint Dirac Equation The adjoint equation is Ψ† (xµ ) i(γ µ )† ∂µ + m ← ← (10.222) and multiplication from the right by γ 0 yields the adjoint Dirac equation Ψ† (xµ ) γ 0 iγ µ ∂µ + m ← =0. as long as det(γ µ ) = 0 the dimension d of the square matrices γ µ must be even so that (−1)d = 1. For d = 2 there exist only three anti-commuting matrices. the γ µ are anti-commuting.218) The Pauli matrices allow one.e.213) reads γ µ γ ν = −γ ν γ µ .224) . (10.223) Inserting this into (10. We will argue further below that the choice f γ µ .219) Using property (10.. rather than a scalar wave function. Relativistic Quantum Mechanics (10. 1 (10. however. except for similarity trasnformations. to construct four matrices γ µ for the next possible dimension d = 4. is unique. in fact.216) i.220) where Ψ(xµ ) represents a 4-dimensional wave function. ψ4 ) and where ∂µ denotes the diﬀerential operator ∂ µ operating to the left side. ψ3 . σ 3 for which. γj = 0 −σ j σj 0 . A proper choice is γ0 = 1 1 0 0 −1 1 . (10. 2. is implied by (10.217) Obviously. σ 2 .221) =0 (10. 3 which. ψ2 . This property can also be written (γ µ )† = γ 0 γ µ γ 0 . 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. 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 ) . in fact. rather than to the right side. (10. namely the Pauli matrices σ 1 . σ j σ k = − σ k σ j for j = k .222) ∗ ∗ ∗ ∗ where we have deﬁned Ψ† = (ψ1 . One can readily show using the hermitian property of the Pauli matrices (γ 0 )† = γ 0 and (γ j )† = −γ j for j = 1. (10.218) of the Pauli matrices one can readily prove that condition (10. 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. holds 2 σj = 1 .215).314 For µ = ν (10.

and γ0 = ˜ 0 1 1 1 0 1 .228) (10.. (10. (10. j = 1. 3 and ˆ β = 1 1 0 0 −1 1 .7: Dirac Equation Similarity Transformations of the Dirac Equation .213) still holds. 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. γ ν ]+ = 2 g µν . γ ˜ Exercise 10. (10.227) A representation often adopted beside the one given by (10.229) The similarity transformation (10.227). γj = ˜ 0 σj 1 S = √ 2 −σ j 0 1 1 1 1 1 −1 1 1 (10. (10.10. [˜ µ . t) = 0 (10.213).221) from the left by γ 0 ˆ ˆ i ∂t + i α · − β m Ψ(r. 10.1: Derive (refeq:Dirac-intro20a) from (10. j = 1. i.230) Schr¨dinger Form of the Dirac Equation o Another form in which the Dirac equation is used often results from multiplying (10.233) The eigenstates and eigenvalues of H correspond to the stationary states and energies of the particles described by the Dirac equation. 2.226) (10. j = 1.231) ˆ whereα has the three components αj . (10. αj = ˆ 0 σj σj 0 .7.227) leaves the algebra of the Dirac matrices unaﬀected and commutation property (10.e. 2.222. ˆ ˆ ˆ Ho = α · p + β m . .219) is the socalled chiral representation deﬁned through ˜ Ψ(xµ ) = S Ψ(xµ ) .225) .Chiral Representation 315 The Dirac equation can be subject to any similarity transformation deﬁned through a non-singular ˜ 4 × 4–matrix S.232) This form of the Dirac equation is called the Schr¨dinger form since it can be written in analogy o to the time-dependent Schr¨dinger equation o i∂t Ψ(xµ ) = Ho Ψ(xµ ) . 3 . 2. 3 .

j1 < j2 < · · · < js s ≤ 4 (10.Good. d4 = 1 0 0 −1 1 0 0 −1 0 1 1 0 0 i −i 0 (10. 27. d4 .Phys. i. Rev. the matrix elements of C = A ⊗ B are Cjk.235) and. 3 .238) These elements form a group since obviously any product of two Γr ’s can be expressed in terms of a third Γr . The reader may .236) where ‘⊗’ denotes the Kronecker product between matrices.234) (10. d2 = . There are (including the diﬀerent signs) 32 elements in G4 which we deﬁne as follows Γ±1 = ±1 1 Γ±2 = ±d1 .235). i. Γ±4 = ±d3 .237) by means of the property (10.213). The three Pauli matrices also obey the property (10.237) which are the ordered products of the operators ±1 and d1 . 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 . d3 . The representations of this group are given by a set of 32 4 × 4–matrices which are equivalent with respect to similarity transformations. Γ±13 = ±d1 d2 d4 . in case of C4 . . m = Aj Bkm . d2 .Mod. 2. Γ±15 = ±d2 d3 d4 Γ±16 = ±d1 d2 d3 d4 (10. The associative algebra generated by d1 . Obviously.235) can diﬀer only with respect to unitary similarity transformations. d4 is called a Cliﬀord algebra C4 .. satisfy the anti-commutation property dj dk + dk dj = 2 for j = k 0 for j = k Relativistic Quantum Mechanics d4 = γ 0 (10. Γ±5 = ±d4 Γ±6 = ±d1 d2 . The representations of Cliﬀord algebras Cm are well established. page 187. Γ±11 = ±d3 d4 Γ±12 = ±d1 d2 d3 . to Dirac matrices.213). Γ±9 = ±d2 d3 Γ±10 = ±d2 d4 . Γ±14 = ±d1 d3 d4 .235) as can be readily veriﬁed from (10.. Γ±8 = ±d1 d4 . Since the Γj are hermitian the similarity transformations are actually given in terms of unitary transformations.H. (1955). any product 1 of the dj ’s can be brought to the form (10.316 Cliﬀord Algebra and Dirac Matrices The matrices deﬁned through dj = iγ j . This property extends then to 4 × 4–matrices which obey (10. Γ±7 = ±d1 d3 .6 and R. j = 1. One can conclude then that also any set of 4 × 4–matrices obeying (10. To complete the proof in this section the reader may consult Miller ‘Symmetry Groups and their Application’ Chapter 9. form a Cliﬀord algebra C3 . . Γ±3 = ±d2 . The Cliﬀord algebra C4 entails a subgroup G4 of elements ± dj1 dj2 · · · djs .e. hence.e. For example.

Lorentz Transformation of the Bispinor State Actually. µ µ (10.240) . Inﬁnitesimal Bispinor State Transformation We want to show now that a suitable transformation of Ψ(xµ ) does.4: Show that from (10. these properties determine the matrices γ µ uniquely. The transformation is assumed to be linear and of the form Ψ (x ) = S(Lµ ν ) Ψ(xµ ) µ µ (10. in a moving frame holds iγ µ ∂µ − m Ψ (x ) = 0 .e. The latter transformations imply that coordinates transform according to (10. 10. namely.8: Lorentz Invariance of the Dirac Equation 317 also want to establish the unitary transformation which connects the Dirac matrices in the form (10. 10. and derivatives according to (10.6). We can. Exercise 7.218) of the Pauli matrices σ j . it must hold γ µ = γ µ .213.2: Demonstrate the anti-commutation relationships (10.215)..10. j = 1. i. i.e.219). However. we will approach the proof of the Lorentz invariance of the Dirac equation by testing if there exists a transformation of the bispinor wave function Ψ and of the Dirac matrices γ µ which together with the transformations of coordinates and derivatives leaves the form of the Dirac equation invariant.e. Exercise 7. i. the form of the Dirac equation is identical in equivalent frames of reference. i. Except for a similarity transformation.8 Lorentz Invariance of the Dirac Equation We want to show now that the Dirac equation is invariant under Lorentz transformations.e. exist.. Hence. Exercise 7. in frames connected by Lorentz transformations. x = Lµ ν xν .241) .6)..214) follows that γ 0 has real eigenvalues ±1 and can be represented by a hermitian matrix.e.239) Form invariance implies that the matrices γ µ should have the same properties as γ µ ..76) yields ∂ρ = Lν ρ ∂ ν .3: Demonstrate the anti-commutation relationships (10. 2. i. a result one could have also obtained by applying the chain rule to (10. and γ j . 3 has imaginary eigenvalues ±i and can be represented by an anti-hermitian matrix. therefore.80) by Lµ ρ and using (10. x µ = Lµ ν xν (10. we do not know yet how to transform the 4-dimensional wave function Ψ and the Dirac matrices γ µ .236) with the Dirac representation (10. Multiplication and summation of (10. (10.. in a moving frame should hold iγ ∂µ − m Ψ (x ) = 0 . in fact.80).218) of the Dirac matrices γ µ . transform coordinates and derivatives of the Dirac equation.

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. hence. (10.247) the inﬁnitesimal matrix µν is anti-symmetric.246) (10. g is an anti-symmetric matrix..e. it must hold (10. of the equation.243) We want to determine now the 4 × 4–matrix S(Lη ξ ) which satisﬁes this condition.245) The construction of S(Lη ξ ) will proceed using the avenue of inﬁnitesimal transformations.318 Relativistic Quantum Mechanics where S(Lµ ν ) is a non-singular 4 × 4–matrix. Multiplication of (10.12) in the form Lν µ gνσ Lσ ρ = gµρ = gρµ .244) from which. The inﬁnitesimal transformation S(Lρ σ ) which corresponds to (10.243) in a form in which the Lorentz transformation in the form Lµν is on the r. follows S(Lη ξ )γρ S −1 (Lη ξ ) = Lσ ρ γσ . i. The proper starting point for a constructiuon of S(Lη ξ ) is actually (10. however. hence. using gρµ γ µ = γρ .221) upon Lorentz transformation yields iS(Lη ξ )γ µ S −1 (Lη ξ ) Lν µ ∂ν − m Ψ (x ) = 0 .. (10.248) Here σµν denote 4 × 4–matrices operating in the 4-dimensional space of the wave functions Ψ. Ob1 viously. µ g ρν = µν and.249) from which we can conclude σµν = −σµν νµ = −σνµ i. It holds then S( µν i ) = 1 − σµν 1 4 µν µν = S(− νµ i ) = 1 + σµν 1 4 µν .e. µ (10. One can.242) The form invariance of the Dirac equation under this transformation implies then the condition S(Lη ξ )γ µ S −1 (Lη ξ ) Lν µ = γ ν .38) the inﬁnitesimal Lorentz transformations in the form Lµ ν = δ µ ν + µ ν where the inﬁnitesimal operator µ ν obeyed T = − g g.250) σµν = −σνµ . Multiplication of this property by g from the right yields ( g)T = − g. νµ (10. (10.243) from the left by Lσ ρ gσν yields S(Lη ξ )γ µ S −1 (Lη ξ ) gρµ = Lσ ρ gσν γσ . in the expression of the inﬁnitesimal transformation ρ Lµν = g µν + µν (10. The elements of g are.h.247) can be expanded S( µν i ) = 1 − σµν 1 4 µν (10. For this purpose we exploit (10. We had introduced in (10.241) implies a similarity transformation Sγ µ S −1 . S( µν ) should not change its value if one replaces in its argument µν by − νµ .s. . state that the Dirac equation (10. the transformation (10. One can ﬁnally conclude multiplying both sides by g ρµ S(Lη ξ )γ µ S −1 (Lη ξ ) = Lν µ γν .

252) Since six of the coeﬃcients αβ can be chosen independently. 3 .s. Exercise 7.h. (10.256) One can readily verify that the non-vanishing generators σµν are given by (note σµν = −˜νµ .251) into (10. γj ] = γ ˜ 2 −iσ j 0 0 iσ j . like the expression on the l.s. this condition can actually be expressed through six independent conditions.229).255) satisfy condition (10. 10.246) results then in a condition for the generators σµν − i ( σαβ γ µ − γ µ σαβ ) 4 αβ = νµ γν . β and operate in the same space as the Dirac matrices.251) S −1 ( µν ) = 1 + σµν µν 1 4 Inserting (10.254) is σαβ = i [ γα . (10.h. of (10. β [ σαβ . 10. For this purpose one needs to express formally the r. In fact. (10. In this representation the Dirac matrices γµ = (˜ 0 . σjk = [˜j .213. j = 1.5: Show that the σαβ deﬁned through (10.h.252) also as a sum over both indices of αβ .10. Furthermore. γ µ ]− = 2i ( δ µ β γα − δ µ α γβ ) . ˜ ˜ σ only six generators need to be determined) σ0j = ˜ i [˜0 .254).228.255) which can be demonstrated using the properties (10.257) .216) of the Dirac matrices. 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.e. (10. γβ ]− 2 (10.254) The proper σαβ must be anti-symmetric in the indices α.. −γ ) are ˜ γ ˜ γ0 = ˜ 0 1 1 1 0 1 .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. γk ] = ˜ γ ˜ jk σ 0 0 σ ..253) 1 2 αβ ( δ µ β γα − δ µ α γβ ) . a solution of condition (10. the chiral representation introduced above in Eqs. of (10. 10. (10. γj = ˜ 0 −σ j σj 0 .248. For this purpose we inspect the properties of the generators in a particular representation. must be symmetric with respect to interchange of the indices α and β. i.252) results in the condition for each α.s. Comparing this with the l. Algebra of Generators of Bispinor Transformation We want to construct the bispinor Lorentz transformation by exponentiating the generators σµν . namely.s.h. For this purpose we need to verify that the algebra of the generators involving addition and multiplication is closed. the expression on the r. 2.

the representation (10.248) allows us to write the transformation S for any. be able to state S in terms of the same parameters w and ϑ as the Lorentz transformation in (10.51) and S.241) bispinor wave functions from one frame of reference into another frame of reference. In order to construct an explicit expression of S in terms of w and ϑ we employ again the chiral representation. Since such operator does not change its block-diagonal form upon exponentiation the bispinor transformation (10. µν in the exponential form i S = exp − σµν 4 µν .e.261) We note that this operator is block-diagonal. In fact. One should.255) yields a closed algebra. and since the algebra of the Pauli matrices is closed.259) yields the desired connection between the Lorentz transformation (10. 2 0 −(w + iϑ) · σ 13 + σ23 ˜ 23 ) (10.259) becomes in the chiral representation ˜ S(w.262) This expression allows one to transform according to (10.258) again into block-diagonal operators.259) We had stated before that the transformation S is actually determined through the Lorentz transformation Lµ ν . We can ﬁnally note that the closedness of the algebra of the generators σµν is not aﬀected by similarity transformations and that. Finite Bispinor Transformation The closedness of the algebra of the generators σµν deﬁned through (10. not necessarily inﬁnitesimal. since both operations convert block-diagonal operators A 0 0 B (10. ϑ) = e 2 (w − iϑ)·σ 0 − 1 (w + iϑ)·σ 0 e 2 1 (10. In this representation holds i ˜ − 4 σµν µν i = − (˜01 01 + σ02 02 + σ03 03 + σ12 12 + σ13 σ ˜ ˜ ˜ ˜ 2 1 (w − iϑ) · σ 0 = . therefore. . i. in particular. the algebra of these generators is closed under addition and multiplication.260) Inserting this into (10. any representation of the generators.320 Relativistic Quantum Mechanics Obviously.51).. one can express the tensor µν through w and ϑ using µν = µ ρ g ρν and the expression (10.44) 0 −w1 −w2 −w3 w 0 ϑ3 −ϑ2 = 1 w2 −ϑ3 0 ϑ1 w3 ϑ2 −ϑ1 0 µν (10. therefore. (10.

264) (10. (10. in fact. 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 .224). The conservation law (10. i.. the conservation law (10. for j µ (xµ ) = (ρ.h. The time-like component ρ of j µ 4 iγ µ ∂µ + iγ µ ∂µ ← Ψ(xµ ) = 0 . −ϑ) = e− 2 (w − iϑ)·σ 0 1 (w + iϑ)·σ 0 e2 1 (10. j) = Ψ† (xµ )γ 0 γ µ Ψ(xµ ) . This transformation behaviour can also be deduced from the transformation properties of the bispinor wave function Ψ(xµ ).270) We conclude that (10.265) (10. We will now determine the relationship between the ﬂux j µ = Ψ (x )γ 0 γ µ Ψ (x ) † µ (10. (10.268) holds for the bispinor Lorentz transformation.221) and the adjoint Dirac equation (10. ϑ) = S(−w.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 .e.263) allows one to conclude that j µ must transform like a contravariant 4-vector as the notation implies. the property applies then in any representation of S.268) We will prove this property in the chiral representation. Since ∂µ transforms like a covariant 4-vector.271) .269) One can readily show that the same operator is obtained evaluating γ S (w. (10. Multiplying (10. j µ must transform like a contravariant 4-vector.s. and addition yields Ψ† (xµ )γ 0 The last result can be written ∂µ Ψ† (xν )γ 0 γ µ Ψ(xν ) = 0 . of (10.221) from the left by Ψ† (xµ )γ 0 . (10.224) from the right by Ψ(xµ ).266) ρ(xµ ) = Ψ† (xµ )Ψ(xµ ) = s=1 |ψs (xµ )|2 (10.263) Starting point are the Dirac equation in the form (10.267) has the desired property of being positive deﬁnite. The reason is that the r. (10.10.263) does hold.263) obviously is a scalar under Lorentz transformations and that the left hand side must then also transform like a scalar. For this purpose we prove ﬁrst the relationship S −1 = γ 0 S † γ 0 . For our proof we note ﬁrst ˜ ˜ S −1 (w. Obviously.

only two degrees of freedom of the bispinor four degrees of freedom are independent [c. One obtains using (10. commute with each other. t) and is determined through the momentum p ∈ R3 . (10. The solutions provide also a complete.. 10.231.246) S −1 (Lη ξ )γ µ S ( Lη ξ ) = (L−1 )νµ γν = (L−1 )ν γ ν = Lµ ν γν .76). the identiﬁcation of observables and their quantum mechanical operators. (10.274) 10. In the following we want to provide this characterization.283)]. but also as providing a basis in which the wave functions of interacting particle systems can be expanded and characterized. For this purpose we consider .271) that the Dirac matrices are independent of the frame of reference. 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. (10.272) results in the expected transformation behaviour µ j = Lµ ν j ν . The operators commute also with Ho in (10. bispinor wave functions and we need to determine corresponding components of the wave function. Note that we have assumed in (10.322 Relativistic Quantum Mechanics in a moving frame of reference and the ﬂux j µ in a frame at rest.9 Solutions of the Free Particle Dirac Equation We want to determine now the wave functions of free particles described by the Dirac equation. speciﬁc for the Dirac free particle. however. the wave function has four components which invite further characterization of the free particle state.233) o i∂t Ψ(xµ ) = Ho Ψ(xµ ) .272) With S −1 (Lη ξ ) = S((L−1 )η ξ ) one can restate (10. These operaˆ tors.275) The free particle wave function is an eigenfunction of Ho .268) µ j = Ψ† (xµ )S † γ 0 γ µ SΨ(xµ ) = Ψ† (xµ )γ 0 S −1 γ µ SΨ(xµ ) . a property which leads to the energy– momentum (dispersion) relationship of the Dirac particle. 10.232. We will start from the Dirac equation in the Schr¨dinger form (10. i. the Dirac particles are described by 4-dimensional.f. 10. The additional degrees of freedom described by the four components of the bispinor wave function require.233). In relativistic quantum mechanics a Dirac particle can also be characterized through a momentum. as just mentioned. Combining this with (10. of which the wave functions are eigenfunctions as well.282. As it turns out.e. additional characterizations. not only as the most simple demonstration of the theory. Like in non-relativistic quantum mechanics the free particle wave function plays a central role.273) where we have employed (10. In case of non-relativistic quantum mechanics the free particle wave function has a single component ψ(r. Like for the free particle wave functions of the non-relativistic Schr¨dinger and the Klein–Gordon o equations one expects that the space–time dependence is governed by a factor exp[i(p · r − t)]. As pointed out. (10. µ (10. orthonormal basis and allows one to quantize the Dirac ﬁeld Ψ(xµ ) much like the classical electromagnetic ﬁeld is quantized through creation and annihilation operators representing free electromagnetic waves of ﬁxed momentum and frequency. as is required for the mentioned property.

276) into (10.282) (10.9: Solutions of the Free Particle Dirac Equation the following form of the free Dirac particle wave function Ψ(xµ ) = φ(xµ ) χ(xµ ) = φo χo ei(p·r − t) 323 (10. (10.232) leads to the 4-dimensional eigenvalue problem m σ·p σ · p −m φo χo = φo χo . like the Klein–Gordon equation. 10. (10.281) Obviously. namely φo and χo are related as follows φ0 χo = = σ·p χ0 −m σ·p φ0 . reproduce the classical relativistic energy–momentum relationships (10.234) of Pauli matrices holds (σ · p)2 = p 2 1 One can.234) and (10.276) where p and together represent four real constants.282.277) To solve this problem we write (10.283) imply that the bispinor part of the wave function allows only two degrees of freedom to be chosen independently. conclude 1. 1 (10. +m (10. E(p) = m2 + p 2 . χo each represent a constant.280) σ·p (σ · p)2 φ0 = χ0 = +m ( + m) ( + m) p2 χ = χ0 .10. hence.281).231.93.276).279) the well-known relativistic dispersion relationship 2 = m2 + p 2 (10. 10. 10.94) Equation (10.284) The relationships (10. and φo . from (10. the Dirac equation.278) provides us with information about the components of the bispinor wave function (10. We want to show now that these degrees of freedom correspond to a spin-like property. one ﬁnds using (5.280) which has a positive and a negative solution = ± E(p) .277) explicitly ( − m) 1 φo − σ · p χo 1 − σ · p φo + ( + m) 1 χo 1 = 0 = 0. In fact. later to be identiﬁed with momentum and energy. These two relationships are consistent with each other.279) According to the property (5. the socalled helicity of the particle. 2 − m2 0 (10. Inserting (10. . (10.283) where is deﬁned in (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 . two-dimensional spinor state.

285)]. is invariant under Lorentz transformations. and let Ψ(0. (10. +|xµ ) = N+ (p) u σ·p E(p) + m u ei(p·r − t) . (10. = +E(p) . ±). u† u = |u1 |2 + |u2 |2 = 1 . +) γ 0 Ψ† (p. ±) γ 0 Ψ(p. −|xµ ) = N− (p) −σ·p E(p) + m u u ei(p·r − t) . ±) S † γ 0 S Ψ(0. (10.h. (10.290).268).e. First..287) the form of which will be justiﬁed further below. u† u = |u1 |2 + |u2 |2 = 1 .s.281)] separately. One can see this as follows: Let Ψ(p.287) will be justiﬁed now. i. hence.290) should diﬀer. ±) γ 0 Ψ(0.s. ±) . S −1 = γ 0 S † γ 0 and.259).287. ±) = S Ψ(0. the solution for = −E(p). ±)γ 0 Ψ(p. +) = 1 . Employing (10. (10. +) γ 0 Ψ† (p. ±) Ψ† (0. ±) γ 0 S −1 γ 0 γ 0 S Ψ(0.286) Here N+ (p) is a constant which will be chosen to satisfy the normalization condition Ψ† (p.. 10. ±) denote the corresponding solution of a particle in its rest frame. (10. through χo given by χo = u1 u2 = u ∈ C2 . is Ψ(p. 0) γ 0 u 0 = |N+ (0)|2 .290) which diﬀers from the normalization condition (10. the solution for = +E(p).291) Note that we have used that. Ψ† (0. The form of this condition and of (10.f. S † = γ 0 S −1 γ 0 . (10.e. Hence. i. +) γ 0 Ψ(0.. (10.241). we present the negative energy solution. +) = −1 .292) . of (10. (10. 10. ±). Similarly.289) Here N− (p) is a constant which will be chosen to satisfy the normalization condition Ψ† (p.324 Relativistic Quantum Mechanics For our further characterization we will deal with the positive and negative energy solutions [cf. (10.s. according to (10. ±) denote the solution of a free particle moving with momentum p in the laboratory frame. we present φo through the normalized vector φo = u1 u2 = u ∈ C2 . ±) = = = Ψ† (0. +) = |N+ (0)|2 (u† . = −E(p) . The connection between the solutions. For the positive energy solution.285) The corresponding free Dirac particle is then described through the wave function Ψ(p. For this purpose we consider ﬁrst the positive energy solution.e.287) in the minus sign on the r.288) corresponding to the wave function Ψ(p.287. of (10. according to (10. we demonstrate that the product Ψ† (p.219) and u† u = 1 [c. the l.h.h. ±) Ψ† (0. Ψ† (p.286) for p = 0 yields. i. using γ 0 as given in (10. where S is given by (10. We want to demonstrate now that the signs on the r.

(10.s.s. this requires the choice of a negative side on the r. of (10.f.295) Evaluating the l. (5. σ·p u∗ E(p) + m T γo u σ·p E(p) + m u = 1 (10.298) becomes N+ (p) = m + E(p) .299) ( m + E(p) )2 .301) yields the normalization coeﬃcient N+ (p) = m + E(p) .290) to assign a positive value to |N− (0)|2 .6: Show that the normalization condition 2 N+ (p) (u∗ )T .9: Solutions of the Free Particle Dirac Equation The same calculation for the negative energy wave function as given in (10.294) We want to determine now N± (p) for arbitrary p. 2m (10. −) γ 0 Ψ(0. We can also conclude from our derivation N± (0) = 1 .287) written explicitly using (10.10.289) yields Ψ† (0.h. Exercise 7.298) This result completes the expression for the wave function (10.h. Condition (10. −) = |N− (0)|2 (0. 325 (10. σ·p u∗ E(p) + m T u σ·p E(p) + m u = 1 (10. u† ) γ 0 0 u = − |N− (0)|2 .219) yields 2 N+ (p) u† u − u† (σ · p)2 u (E(p) + m)2 = 1.286) is 2 N+ (p) (u∗ )T . (10.285) results in 2 N+ (p) 1 − p2 (E(p) + m)2 = 1 (10.302) .300) (10.293) 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 (10.234)] and using the normalization of u in (10. 2 E(p) (10.286). We consider ﬁrst the positive energy solution.296) Replacing (σ · p)2 by p 2 [c. using γ 0 as given in (10. ( m + E(p) )2 − p 2 (10.

identical to the condition (10. −p).289). deﬁned as the component of the particle spin along the direction of motion.300) have been constructed to satisfy the free particle Dirac equation (10. the wave functions constructed represent eigenstates of Ho . j = 1.286) into (10.286. We can. 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. The property [σ · p. hence.e. pj ] = 0 . yields 2 N− (p) u† (σ · p)2 u − u† u (E(p) + m)2 = −1 . Condition (10. (10. (u∗ )T γ o = −1 (10. λ|xµ ) (10. 10. i..e.308) 2 |p| ˆ Note that p represents here an operator. This can be veriﬁed expressing the space–time factor of Ψ(p.289. λ|xµ ) . 10. 3 is fairly obvious. 10.309) .296) for the normalization constant N+ (p) of the positive energy solution. λ|xµ ) = pµ Ψ(p. however.. λ|xµ ) = λ E(p) Ψ(p.305) and.e. i. The wave functions (10.275) yields Ho Ψ(p. This degree of freedom 1 describes a spin– 2 attribute. (10. conclude N− (p) = m + E(p) 2m (10.289) are not completely speciﬁed. The commutation property [σ · p. Ho ] = 0 follows from (10. thereby.307) we investigate ﬁrst the observable due to the simpler operator σ · p.307) where pµ = ( . not a constant vector.306) i.275).286. This attribute is the so-called helicity. λ|xµ ) in 4-vector notation. Helicity The free Dirac particle wave functions (10. The corresponding operator which measures this observable is 1 p ˆ Λ = σ· . Inserting (10.h.s. have completed the determination for the wave function (10. Rather than considering the ˆ observable (10. i∂µ Ψ(p.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. the two components of u indicate another degree of freedom which needs to be deﬁned.304) This condition is. The wave functions are also eigenstates of the momentum operator i∂µ . 2. (10.303) E(p) + m u Evaluating the l.. exp[i(p · r − t)] = exp(ipµ xµ ).326 Relativistic Quantum Mechanics We consider now the negative energy solution.289) is T −σ·p −σ · p 2 E(p) + m u N− (p) u∗ .290) written explicitly using (10.

States which are simultaneously eigenvectors of these three operators are also simulteneously eigenvectors of the three operators p. = +E(p). = −E(p). i. is independent of the direction of motion of the particle.312) We assume now particles with negative energy. 0)T and (0.310) We have shown altogether that the operators p. t) = Np e 2 1 p m + Ep 0 0 i(px3 − E t) 1 p e Ψ(pˆ3 . − 1 |r. can ˆ ˆ be simultaneously diagonal. . 2m (10. hence. +. We consider ﬁrst the simplest case that particles move along the x3 –direction. 2 the wave functions which are eigenstates of the helicity operator.10. Since helicity is deﬁned relative to the direction of motion of a particle the characterization of u as an eigenvector of the helicity operator.. 1 p = (0. +.313) 2 0 1 where Ep and Np are deﬁned in (10. −.e. 1)T are eigenstates of 2 σ 3 with eigenvalues ± 1 .. We assume ﬁrst particles with positive energy.224) 1 of σ3 the two u vectors (1.312).e.e. −. 327 0 (σ · p)2 ˆ 2 (σ · p) ˆ 0 0 (σ · p)2 ˆ 2 (σ · p) ˆ 0 (10. t) = Np e 1 0 0 p m + Ep i(px3 + E t) 1 p e Ψ(pˆ3 . ˆ The condition that the wave functions (10. − 1 |r.. Np = m + Ep . t) = Np e (10. are 1 i(px3 − E t) 0 p e Ψ(pˆ3 . 0.9: 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. i.308) above. p3 ). Ho and Λ deﬁned in (10. Therefore. According to the deﬁnition (5. in principle. 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 . t) = Np e (10.311) 2 −p 0 m + Ep 1 where e3 denotes the unit vector in the x3 -direction and where ˆ Ep = m2 + p2 . + 2 |r. Ho and σ·p commute with each other and. i. In this case Λ = 2 σ 3 .286) are eigenfunctions of Λ as well will specify now the vectors u. + 1 |r.

5.e. 10.123). p) e |p| (10.220). β. in which case the transformation is given by (5.226. [One can also express the rotation through Euler angles α. − 2 ) −p 1 m + Ep u(p. a wave function which represents free particles at rest. t) 2 = Np 1 u(p. − 1 ) 2 −p 1 m + Ep u(p.229). − ) 2 where ϑ(p) = = = exp exp i − ϑ(p) · σ 2 i − ϑ(p) · σ 2 1 0 0 1 . 1)T have to be replaced by eigenstates u± (p) of the spin operator along the direction of p. denoted by ˜. The corresponding wave functions are determined through ˜ i˜ 0 ∂t − m Ψ(t) = 0 . γ .320) where Ep and N are again given by (10.316) describes a rotation which aligns the x3 –axis with the direction of p. 10. 2 1 0 (10.315) e3 × p ˆ ∠(ˆ3 . + 1 ) 2 p 1 m + Ep u(p.319) 1 Ψ(p.311.262) for the bispinor wave function and the transformation (10. + 1 |r. 1 |t) = √2 e−imt . − 1 ) 2 ei(p·r − Ep t) (10. − 1 |r. + 2 ) u(p. +.318) 1 Ψ(p.314) (10.223) as follows 1 u(p. t) = Np ei(p·r + Ep t) (10.. for an r–independent wave function.311. +. + 2 |r. t) 2 = Np u(p. (10. 10. γ and are 1 0 1 ˜ Ψ(p = 0. +. t) = N√ ei(p·r + Ep t) (10.312). −.312) can be obtained also by means of the Lorentz transformation (10. These eigenstates are obtained through a rotational transformation (5.317) Ψ(p. + 2 ) p 1 m + Ep u(p. − 2 ) u(p. Generating Solutions Through Lorentz Transformation The solutions (10. For this purpose one starts from the solutions of the Dirac equation in the chiral representation (10. −.328 Relativistic Quantum Mechanics To obtain free particle wave functions for arbitrary directions of p one can employ the wave functions (10.] The corresponding free particle wave functions are then Ψ(p. + ) 2 1 u(p. i.222.321) . 0)T and (0.313) except that the states (1. + 2 ) ei(p·r − Ep t) (10. − 2 |r.

(10. −. yields for the exponential space–time dependence according to (10. 10. −. χo are eigenstates of σ 3 with eigenvalues ±1. 2 The solutions (10. +.325) to (10. 2 |t) = √2 .10. t) e 2 e− 2 w3 1 = √ 1 2 e 2 w3 i(px3 − E t) p e . + 1 |r. For the resulting wave functions in the chiral representation one can use then a notation corresponding to that adopted in (10. 10.325) One should note that φo .228) yields the corresponding solutions (10.174. − 1 |t) = √2 e−imt . φo . +.311) e 1 w 2 3 ˜ Ψ(p(w3 )ˆ3 . 2 0 e −1 329 (10. 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 .313) in the p → 0 limit. +.323) Transformation (10. χo ∈ 1 0 .324. −1 e 0 0 1 +imt 1 ˜ Ψ(p = 0. for w = (0.323) should yield the solutions for non-vanishing momentum p in the x3 –direction.322). i.. − 1 |t) = √2 .311. Applying (10.262) φo χo → e 2 w3 σ 0 1 3 Et ) (10.324) 0 e − 1 w3 σ 3 2 φo χo = e 2 w3 σ φo 1 3 e− 2 w3 σ χo 1 3 . w3 ). 0.e.9: Solutions of the Free Particle Dirac Equation 0 1 1 ˜ Ψ(p = 0. 10.176) imt → i ( p3 x3 and for the bispinor part according to (10. This correspondence justiﬁes the characterization ±.322) can be written in spinor form 1 √ 2 φo χo e imt .262) for a boost in the x3 –direction. − 1 |r. 2 0 1 1 0 +imt 1 1 ˜ Ψ(p = 0.322) The reader can readily verify that transformation of these solutions to the Dirac representationsas deﬁned in (10. ± 1 of the wave functions stated in (10. 0 1 (10.

61) between the parameter w3 and boost velocity v3 . −. + 1 |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. Transformation to the Dirac representation by means of (10.61) p(w3 ) = m sinhw3 . − 2 |r. −.326) where according to (10. t) e 2 = i(px3 + E t) p e 1 Ψ(p(w3 )ˆ3 . − 2 |r. −. t) e 1 = √ 2 −e− 2 w3 − 1 w3 2 1 i(px3 + E t) p e ˜ Ψ(p(w3 )ˆ3 . t) e = i(px3 + E t) p e (10.330 e 1 w 2 3 Relativistic Quantum Mechanics 1 0 1 0 0 1 0 1 1 ˜ Ψ(p(w3 )ˆ3 . +.228) yields w3 cosh 2 = sinh w3 2 1 Ψ(p(w3 )ˆ3 . t) e = cosh w3 2 − sinh w3 2 sinh w3 2 cosh w3 2 − sinh w3 2 cosh w3 2 i(px3 − E t) p e Ψ(p(w3 )ˆ3 . −. − 1 |r. + 2 |r.327) Employing the hyperbolic function properties cosh x = 2 coshx + 1 . +. t) e 1 0 1 0 0 1 0 1 1 0 1 0 0 1 0 1 i(px3 − E t) p e 1 Ψ(p(w3 )ˆ3 . + 2 |r. 2 (10.329) . 2 sinh x = 2 coshx − 1 .328) the relationship (10. and the expression (10. t) e 2 e 1 = √ 1 2 − e 2 w3 i(px3 + E t) p e (10.

243) considered solely the limit of inﬁnitesimal transformations anyway. ρ(Lη ξ ) has been deﬁned in (10. This transformation. and since (2) the calculations following (10. The fact that any such Ψ (xµ ) is a solution of the Dirac equation allows us to conclude S(Lη ξ )γ µ S −1 (Lη ξ ) ρ(Lη ξ )∂µ ρ−1 (Lη ξ ) = γ ν ∂ν which is satisﬁed in case that the following conditions are met ρ(Lη ξ )∂µ ρ−1 (Lη ξ ) S(Lη ξ )γ µ S −1 (Lη ξ ) Lν µ = Lν µ ∂ν . the proper transformation S(Lη ξ ). To show this we rewrite the Dirac equation (10. Such transformation had been applied by us.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 ). The reason why we provide another derivation of (10.9: Invariance of Dirac Equation Revisited and similarly sinh w3 = 2 Ep − m = 2m p 2m (Ep + m) 331 (10.329. that if Ψ(xµ ) is a solution of the Dirac equation that Ψ (xµ ) as given in (10. thereby. which however.10. reproduces the positive energy wave functions (10.168–10. the essential property stating the Lorentz–invariance of the Dirac equation. We expect. of course. is suﬃcient since (1) it must hold then for any ﬁnite transformation. The transformation is Ψ (xµ ) = S(Lη ξ ) ρ(Lη ξ ) Ψ(xµ ) (10.313) for −p. = γν .243).332) Here we have made use of the fact that S(Lη ξ ) commutes with ∂µ and ρ(Lη ξ ) commutes with γ µ .330) Inserting expressions (10. Making this expectation a postulate allows one to derive the condition (10.333) .176). Such description will be essential for the formal description of Lorentz invariant wave equations for arbitray spin further below. 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 (10.334) (10. (10. refered to as the active transformation.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) and. Invariance of Dirac Equation Revisited At this point we like to provide a variation of the derivation of (10. Λ|t).327). λ. Actually. indeed. but not the observer.331) i S(Lη ξ )γ µ S −1 (Lη ξ ) ρ(Lη ξ )∂µ ρ−1 (Lη ξ ) − m Ψ (xµ ) = 0 (10. λ. when we generated the solutions Ψ(p.330) into (10. In the new derivation we consider the particle described by the wave function transformed.331) is a solution as well. 10. in general.221) using (10. Λ|xµ ) from the solutions describing particles at rest Ψ(p = 0.123) above and characterized there.311) as well as the negative energy solutions (10. expresses the system in the old coordinates. we will derive this equation only for inﬁnitesimal transformations.

1) . ∂µ ] = [x3 ∂2 − x2 ∂3 . . 0.336) We will proceed with this task considering all six cases: [J1 . 0).339) [K1 . . 0. 0.. ∂µ ] = (10. ϑ = (0. the Dirac equation is invariant under active Lorentz transformations. We have demonstrated.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 = (1.331) is again a solution of the Dirac equation. 1.332 Relativistic Quantum Mechanics We will demonstrate now that the ﬁrst condition is satisﬁed by ρ(Lη ξ ). of course. w = (0. 0. .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. ∂µ ] = (10.262). [K .243) and. therefore. The second condition is identical to (10. 0). Inspection of (10.334) for inﬁnitesimal Lorentz transformations Lη ξ . ∂µ ] = (10. ∂µ ] = [x1 ∂3 − x3 ∂1 . ∂µ ] = [x0 ∂3 + x3 ∂0 . 0).337) [J2 . ∂µ ] = [x0 ∂1 + x1 ∂0 . i.340) [K2 . ϑ = (0. 0). ∂µ ] = (10. . As mentioned already we will show condition (10. w = (0. it is met by S(Lη ξ ) as given in the chiral representation by (10. ∂µ ] = [x0 ∂2 + x2 ∂0 . 0. ∂µ ] = (10. .338) [J3 .335) shows that we need to demonstrate [J . We will proceed by employing the generators (10. 0). ∂µ ] = (10.e.342) One can readily convince oneself that these results are consistent with (10. 0 0 ∂3 −∂2 0 −∂3 0 ∂1 0 ∂2 −∂1 0 −∂1 −∂0 0 0 −∂2 0 −∂0 0 −∂3 0 0 −∂0 µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ = = = = = = = = = = = = = = = = = = = = = = = = 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 (10. that any solution Ψ(xµ ) transformed according to (10. ∂µ ] = [x2 ∂1 − x1 ∂2 . w = (0. ∂µ ] = (K )ν µ ∂ν .341) [K3 .336). ∂µ ] = (J )ν µ ∂ν .

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 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. σ3 )T one obtains then i∂t φ i∂t χ = = ˆ σ · π χ + qV φ + mφ ˆ σ · π φ + qV χ − mχ .6 we assume that the ﬁeld is described through the 4-vector potential Aµ and.233) by replacing i∂t by (see Table 10.345) (10.345) holds |qV | << m . (10.351) X (x ) µ .345) in the so-called non-relativistic limit in which all energies are much smaller than m.g.347) (10. 10. (10.350) (10.344) Using the notation σ = (σ1 . Following the respective procedure developed for the Klein-Gordon equation in Sect. σ2 . Accordingly. Non-Relativistic Limit We want to consider now the Dirac equation (10. Equivalently.6 above).10: Dirac Particles in Electromagnetic Field 333 10.1) i∂t − qV and p by ˆ ˆ π = p − qA .10. for the scalar ﬁeld V in (10.10 Dirac Particles in Electromagnetic Field We like to provide now a description for particles governed by the Dirac equation which includes the coupling to an electromagnetic ﬁeld in the minimum coupling description. e. accordingly..1 in Sect 10. we deﬁne φ(xµ ) = e−imt Φ(xµ ) χ(x ) = e µ −imt (10.221) reads then [ iγ µ (∂µ + iqAµ ) − m ] Ψ(xν ) = 0 (10. This solution exhibits a timedependence exp[−i(m + )t] where for holds in the non-relativistic limit | | << m.348) (10. The Dirac equation (10. For this purpose we choose the decomposition Ψ(xµ ) = φ(xµ ) χ(xµ ) .343) One may also include the electromagnetic ﬁeld in the Dirac equation given in the Schr¨dinger form o ˆ (10.346) ˆ ˆ ˆ α · π + qV + β m Ψ(xµ ) (10. we replace the operator ∂µ by ∂µ + iqAµ .349) We want to focus on the stationary positive energy solution.232).

349) yields i∂t Φ i∂t X = = ∂t X X Relativistic Quantum Mechanics << m . due to the m−1 factor.346. (10.353) (10. henceforth. (10. identiﬁes X as the small component of the bi-spinor wave function which. (10.334 and assume that for the time-derivative of Φ and X holds ∂t Φ << m .351) in (10.354) ˆ 0 ≈ σ · π Φ − 2m X .361) .350.352) ˆ σ · π X + qV Φ ˆ σ · π Φ + qV X − 2mX .357) Equation (10. 10. derived in Sect.230). ˆ ˆ For the components of π × π holds ˆ ˆ π×π = jk (10. i i (10.357) for Φ can be reformulated by expansion of (σ · π)2 . (10. ˆ Equation (10. does not need to be considered anymore. Ak ] ) f = ( ∂j Ak − Ak ∂j + Aj ∂k − ∂k Aj ) f . For this purpose we employ the identity (5.348.354) The properties (10. π ] . Φ Using (10. Ak ] + q 2 [Aj . and. One obtains [πj . Ak ] i i i i =0 =0 = q q [Aj . accordingly. ∂k + qAk ] i i 1 1 1 1 [ ∂j .355) ˆ σ·π Φ 2m (10. one can replace X in (10.359) We want to evaluate the latter commutator.7.353) by X ≈ to obtain a closed equation for Φ ˆ σ·π 2m 2 (10.360) For an arbitrary function f (r) holds ( [Aj . ∂k ] + [∂j . 5. which in the present case states ˆ ˆ2 ˆ ˆ (σ · π)2 = π + i σ · (π × π) . 10. ∂k ] + [∂j .356).356) i∂t Φ ≈ Φ + qV Φ .352) allow one to approximate (10. (10. ∂k ] + q [Aj . ∂k ] + q [ ∂j .358) ( πj πk − πk πj ) = jk [πj . πk ] = = 1 1 [ ∂j + qAj . Ak ] . 10.

2 around the magnetic ﬁeld. ∂k ] + [∂j . r (10. (10.223) shows that the propagator in (10. 5.364) governing a two-dimensional wave function Φ ∈ C2 with the corresponding non-relativistic Schr¨dinger equation (10. The formal solution of this equation is Φ(t) = e−iqtB·σ Φ(0) . the spin. 10.366) (10.363) where we employed B(r.344. while agreeing in all other respects with the non-relativistic Schr¨dinger equation (10.360) and Aµ = (V.1 which emerged in the Lorentz-invariant theory as an algebraic 2 necessity. The respective bispinor wave function Ψ(xµ ) ∈ C4 is described as the stationary solution of the Dirac equation (10. t) = × A(r. t) (10. the interaction qσ · B induces a 1 precession of the spin. 0. In other words. t) [see (8.1 particle with a magnetic 2 ﬁeld B.343) for the vector potential Aµ = (− Ze2 .222. 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) .2) governing a one-dimensional wave function ψ ∈ C.10.359. one obtains ( [Aj . t) ≈ ˆ [p − q A(r.367) . using (10.363) allow us to write (10. o reveals a stunning feature: the Pauli equation does justice to its two-dimensional character.366) can be interpreted as a rotation around the ﬁeld B by an angle qtB. Let us consider brieﬂy the consequences of the interaction of a spin.10: Dirac Particles in Electromagnetic Field Using ∂j Ak f = ((∂j Ak )) f + Ak ∂j f ∂k Aj f = ((∂k Aj )) f + Aj ∂k f 335 where ((∂j · · ·)) denotes conﬁnement of the diﬀerential operator to within the brackets ((· · ·)).364) which is referred to as the Pauli equation.6)]. −A). Comparision of (10.358. Equations (10. does not leave the theory again when one takes the non-relativistic limit. 0.1 with the magnetic ﬁeld. 10. 0) . but rather remains as a steady “guest” of non-relativistic physics with the proper interaction term. 10.362) or.357) in the ﬁnal form i∂t Φ(r.Spectrum We want to describe now the spectrum of a relativistic electron (q = −e) in the Coulomb ﬁeld of a nucleus with charge Ze. Dirac Particle in Coulomb Field . t) 2m 2m Φ(r. i.365) Comparision of this expression with (5.. t)]2 q − σ · B(r. t) + q V (r.e. [πj .2) it introduces the o extra term qσ · B Φ which describes the well-known interaction of a spin. Ak ] ) f = [ ((∂j Ak )) − ((∂k Aj )) ] f (10. πk ] = q q (( ∂j Ak − ∂k Aj )) = − i i ×A jk q = − B i jk (10.

multiplying (10.373) (10. once a solution Ψ(xµ ) of (10.370) for which holds ˜ ˆ ˜ ˆ 3 γ πµ γ πν = ˜ ˆ ˜ ˆ µ=0 µ ν (˜ µ )2 πµ + γ ˆ2 µ. noting from (10. (˜ 0 )2 = 1 and (˜ j )2 = γ 1 γ −1 j = 1.370) ˜ [ i˜ µ (∂µ + iqAµ ) − m ] [ i˜ µ (∂µ + iqAµ ) + m ] Ψ(xµ ) = 0 γ γ (10. but diﬀers from it in an essential way.e.ν=1 µ =ν γ µ γ ν πµ πν . ˜ (10. 1.230) γ µ γ ν = − γ ν γ µ . µ = ν and altering ‘dummy’ summation ˜ ˜ ˜ ˜ .369) For our solution we will adopt presently a strategy which follows closely that for the spectrum of pionic atoms in Sect.7 one can write the second term in (10. but the converse is not necessarily true. 3. Equation (10.368) ˜ where Ψ(xµ ) and γ µ are deﬁned in (10.368) ˜ ( γ µ πµ − m ) Ψ(xµ ) = 0 . 10.230).369) from the left by γ ν πν + m.371).370) Any solution of (10.336 Relativistic Quantum Mechanics For the purpose of the solution we assume the chiral representation.371) such that we can conclude that (10.s.h. can be rewritten using.232) in Sect. 5.228) and in (10. indeed.370).370) resembles closely the Klein-Gordon equation (10. (5. 3 µ=0 ˆ2 (˜ µ )2 πµ = π0 − π .374). i.231). ˜ However.368) is also a solution of (10.369).1 one can write (10. γ ˆ2 ˆ2 (10. The diﬀerence arises from the term γ µ πµ γ ν πν in (10. For this purpose we ‘square’ the Dirac equation. γ ˆ ˜ ˆ (10. we solve ˜ [ i˜ µ (∂µ + iqAµ ) − m ] Ψ(xµ ) = 0 γ (10. is a solution of (10. ˜ ˜ ˆ ˆ (10. 2.374) The ﬁrst term on the r.6. respectively.370) is obtained then ˜ [ i˜ µ (∂µ + iqAµ ) + m ] Ψ(xµ ) γ is a solution of (10. This follows from [ i˜ µ (∂µ + iqAµ ) + m ] [ i˜ µ (∂µ + iqAµ ) − m ] γ γ = [ i˜ µ (∂µ + iqAµ ) − m ] [ i˜ µ (∂µ + iqAµ ) + m ] γ γ according to which follows from (10.229).369).180). according to (10. Employing πµ as deﬁned ˜ in Table 10. 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 . (5.372) (10.375) Following the algebra that connected Eqs.

one can summarize (10. γ ] [ˆ0 .367) of Aµ .ν=1 µ =ν = [˜ µ .ν=1 µ =ν σ 0 0 −σ · (( qA0 )) (10. πj ] + γ ˜ π ˆ 4 0 j [˜ j . t) is a suitable test function and where ((· · ·)) denotes the range to which the derivative is limited.378) can be evaluated using (10. γ j ] γ ˜ = = 2 0 1 1 1 0 1 σj 0 0 −σ j 0 σj −σ j 0 − 0 σj −σ j 0 0 1 1 1 0 1 (10. r2 (10. π0 ] γ ˜ π ˆ j=1 = [˜ 0 . γ µ γ ν πµ πν = ˜ ˜ ˆ ˆ µ.ν=1 µ =ν (10. πj ] π ˆ = = (−i∂t + qA0 .377) which follows readily from the deﬁnition (10.376–10.379) The commutators [ˆ0 .381) According to (10.367) and the deﬁnition (10. Since [ˆµ . the commutators [˜ 0 . 2. due to A = 0. 3 (10. πν ] = 0 π ˆ for µ.376) This expression can be simpliﬁed due to the special form (10.380) where f = f (r. −i∂j ] = − [ (∂t + iqA0 ) ∂j − ∂j (∂t + iqA0 ) ] f i ((∂j qA0 )) f (10.10. it holds 1 4 [˜ µ .344).e. i. γ ν ] [ˆµ .378) According to the deﬁnition (10.344) π ˆ [ˆ0 .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 ν ] [ˆµ . ν = 1.. πj ] . Altogether.367) holds qA0 = r ˆ Ze2 . πν ] γ ˜ π ˆ µ.10: Dirac Particles in Electromagnetic Field indices.ν=1 µ =ν 337 1 2 ( γ µ γ ν πµ πν + γ ν γ µ πν πµ ) ˜ ˜ ˆ ˆ ˜ ˜ ˆ ˆ µ.229).382) . γ ˜ π ˆ j=1 (10. γ j ] are γ ˜ [˜ 0 . γ 0 ] [ˆj . πν ] γ ˜ π ˆ µ.ν=1 µ =ν 3 j=1 3 3 = 1 4 1 2 1 [˜ .380) γ µ γ ν πµ πν = i ˜ ˜ ˆ ˆ µ. γ j ] [ˆ0 . πj ] in (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. In the latter case. 10. is genuinely two-dimensional and.387) The expression (10.e.368) reads − Ze2 ∂t − i r 2 + 2 + i σ·r ˆ 0 0 −σ · r ˆ Ze2 − m2 r2 Ψ(xµ ) = 0 (10. 1 |ˆ). The term iσ · r. it is this quantity that we want to determine. 6. .389) According to the results in Sect. + j } (10. (10.189) for the Laplacian and expansion of the term (· · ·)2 result in the twodimensional equation 2 ∂r − ˆ L2 − Z 2 e4 iσ · r Ze2 ˆ 2Ze2 + r2 r + 2 − m2 r φ± (r) = 0 . 1 |ˆ). i.374). −j + 1. .384) into (10.5 the operator iσ · r is block-diagonal in this basis such that only ˆ 1 the states for identical j. (10. r r 2 2 2 j = 1. Yjm (j + r 2 1 1 . Accordingly. in r ˆ fact. We note that these states are also eigenstates of the angular r 2 . (10. 6. 1 |ˆ) ) . Such solutions are of the form ˜ ˜ Ψ(xµ ) = Φ(r) e−i t .381). .338 Relativistic Quantum Mechanics where r = r/|r| is a unit vector. a solution of the form ∼ Y m (ˆ) can be obtained. . 6. 2 3 2 .384) can be interpreted as the energy of the stationary state and.385) 1 We split the wave function into two spin. (10. (10.383) We seek stationary solutions of this equation. however. . m = −j.386) + 2 ± iσ · r ˆ Ze2 − m2 r2 φ± (r) = 0 . we express 2 the solution of (10.191) solved in Sect. hence. (10. m values are coupled.6.375) the ˆ ‘squared’ Dirac equation (10..2 components ˜ Ψ(r) = and obtain for the separate components φ± (r) Ze2 + r 2 ˜ φ+ (r) ˜ φ− (r) (10. couples the orbital angular momentum of the electron to its spin. only the two states {Yjm (j − 2 .148). 2 |ˆ)} as given in (6.383) 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 . . Yjm (j + 1 .1 .147.5 which describe the coupling of orbital angular momentum and spin 1 {( Yjm (j − 2 .388) in terms of states introduced in Sect. Insertion of (10. Combining this result with (10.

396) and (10. involving the 2 × 2 unit matrix.2 (j).151). 10. which states that the states Yjm (j ± 1 . as in the case of pionic atoms.392) reads then in the diagonal representation 2 ∂r − λ1. such transformation can be chosen as to diagonalize the second term. We select. Any similarity transformation leaves the ﬁrst term in (10. in the present treatment.200). 1 |ˆ) r 2 (10.395) (10. . Since. namely.5 [c. namely. the values of the argument of λ1. 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 (10.391) 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 . rather than = 0. the missing additive term − 1 . (6.151]. except for the slight diﬀerence in the expression of λ1.392) into diagonal form.10.186)]. . . .199). (6. we want to determine solely the spectrum.10: Dirac Particles in Electromagnetic Field 339 ˆ momentum operator L2 [cf.395. 1 |ˆ) are r 2 2 ˆ eigenfunctions of L2 .2 (j) + 1) 2 Ze2 + + r2 r 2 − m2 f1. hence.2 (r) = 0 (10. m and expand φ± (r) = Using σ · r Yjm (j ± 1 . Obviously. 1 |ˆ) r 2 2 2 2 r r . λ1 (j) and λ2 (j). conclude that the . We can. the eigenvalues are independent of m. therefore.392) h± (r) g± (r) = 0. property (6. 6. However.2 (j) as given by (10. not the wave functions. (10..2 (j)[λ1.390) (10. 1. and except for the fact that 2 2 we have two sets of values for λ1.393) but do not explicitly consider further the wavefunctions. . unaltered. 1 |ˆ) + r Yjm (j + 1 .394) Equation (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. 1 |ˆ) = − Yjm (j ˆ r 2 2 1 2 h± (r) g± (r) Yjm (j − 1 .392).396) and λ2 (j) [λ2 (j) + 1] (10.f.397) This equation is identical to the Klein-Gordon equation for pionic atoms written in the form (10. − 1 r2 (j − 1 )(j + 1 ) − Z 2 e4 2 2 ±i Ze2 (j + )(j + ) − Z 2 e4 We seek to bring (10.2 (j) being 2 j = 1 . . 3 . a speciﬁc pair of total spin-orbital angular momentum quantum numbers j.

400) and (10. m = − . . (10. . 2. . . . counts the nodes of the wave function.399) n = 0. the second term on the r.398. 2n2 = 0. (10. = 1 + m Z 2 e4 √ ( n − − 1 + 2 − Z 2 e4 )2 . . . . . albeit with some modiﬁcations. . m) 2 n = 1. One may also state 2 this in the following way: (10.397) is again given by eq.404) = 0. .203). . . . 1 ms = ± 2 (10. . and spin-orbital angular momentum j = + 1 . .403) √ Z 2 e4 2 ED (n. . .340 Relativistic Quantum Mechanics spectrum of (10. (10. . − + 1. 1 = 1 + m Z 2 e4 (n + 1 + (j+ 1 )2 − Z 2 e4 )2 2 Z 2 e4 (n + (j+ 1 )2 − Z 2 e4 )2 2 . .396) we obtain. .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. j where 1 corresponds to λ1 (j) as given in (10. n − 1 . 1. 1. . 2. j = 1. 2. j . including spin. . . ( +1) − Z 2 e4 )2 (10. and spin-orbital angular momentum j = − 1 .h.401) These expressions can be equated with the non-relativistic spectrum. j = 1 . m = −j. 1. 2. . . . 10. Expanding the energies in terms of this parameter allows one to identify is determined by Z the relationship between the energies 1 and 2 and the non-relativistic spectrum. . the stationary states have binding energies E = − mZ 2 e4 n = 1.395) and 2 corresponds to λ2 (j) as given in (10. . . . . In case of non-relativistic hydrogen-type 1 atoms.s. . One obtains in case of (10.396). of these equations describe the binding energy. . 2 2 1 ≈ m − (10. (10. . . . .400) corresponds to a non-relativistic state with quantum numbers n. . 2 2 m = −j. 1. It is given by n = n + + 1 where is the orbital angular momentum quantum number and n = 0. . (10.401) if one attributes to the respective 1 states the angular momentum quantum numbers = j + 2 and = j − 1 . −j + 1. 10. . . n − 1 . 1. . .2 . . −j + 1. m = −j. . accordingly. The magnitude of the relativistic eﬀect 2 2 e4 . m) = 1 + m (n− + . −j + 1. j = + 1 . 3. . 3 .402) with (10. One can equate (10. j = − 2 .402) In this expression n is the so-called main quantum number.398) 2 = 1 + m . . Obviously.401) corresponds to a non-relativistic state 2 with quantum numbers n. Using (10. .400) (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 . . . These considerations 2 are summarized in the following equations 1 ED (n. .395. j .

0 0 1 1 0 1 1 2 2 spinorbital ang. relativistic.e.51176 ⇑ ⇑ ⇑ ⇑ 2s 1 2 2p 1 2 2p 3 2 3s 1 2 3p 1 2 3p 3 2 3d 3 2 3d 5 2 Table 10. 10.2: Binding energies for the hydrogen (Z = 1) atom.551178 ⇑ . binding energy / eV Eq. .405) in terms of the non-relativistic quantum numbers n. One can combine the expressions (10.402.0041 keV and e2 = 1/137. −j + 1. binding energy / eV Eq.405) In order to demonstrate relativistic eﬀects in the spectrum of the hydrogen atom we compare in Table 10. 2.3. = 0. r 2 2 2 .402) -13. .405). (10.036 by means of Eqs.402)] and the relativistic [cf. in fact.10: Dirac Particles in Electromagnetic Field main quantum number n 1 2 2 2 3 3 3 3 3 orbital angular mom. i.60601 -3. 1. Degeneracies are denoted by ⇑.40146 ⇑ ⇑ -1. n − 1 j = 1 2 1 2 ± m = −j. The energies were evaluated with m = 511.405)] spectrum of the hydrogen atom. . 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.60583 -3. . notation 1s 1 2 non-rel. m) = m 1 + Z 2 e4 (n−j −1 + 2 (j+ 1 )2 − Z 2 e4 )2 2 n = 1. j. The reason is that the mean kinetic energy of the electron in the hydrogen atom.1. 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 10.404) ﬁnally into the single formula ED (n.551177 ⇑ . 1 |ˆ) for m = ± 1 .3. .. (10. i. The 2p 1 states with j = j = 1 2 3 2 2 2 2 for = 0 otherwise involve two degenerate states corresponding to Y 1 m (1. The table entries demonstrate that the energies as given by the expression (10. the non-relativistic and relativistic energies are hardly discernible. for example. 10.40151 ⇑ . (10. (10. .551176 spectr.10. .1. Of particular interest is the eﬀect of spin-orbit coupling which removes. the nonrelativistic degeneracy for the six 2p states of the hydrogen atom: in the present.e.. However. . the 2p 3 states with r 2 2 2 2 1 involve four degenerate states corresponding to Y 3 m (1. ± 3 . j (10. relate closely to the corresponding nonrelativistic states. .405 ) -13. . is in the range of 10 eV.40147 -1. . mom. 2 |ˆ) for m = ± 1 .2 the non-relativistic [cf. case these six states are split into energetically diﬀerent 2p 1 and 2p 3 states. (10. .403.

15.8 ⇑ .3: Binding energies for the hydrogen-type (Z = 100) atom.402) -136.206) provides in the present case ED (n.10.1 ⇑ .402.2. j.405) to order O(Z 4 e8 ). 10. (10.1 -34. m) ≈ m − mZ 2 e4 mZ 4 e8 − 2n2 2n3 1 j+ − 3 4n + O(Z 6 e12 ) . (10. Introducing α = Z 2 e4 1 and β = j + 2 (10. binding energy / keV Eq.036 by means of Eqs.407) 1 2 This expression allows one.0 ⇑ ⇑ -15.409) where V (r) is spherically symmetric.405) reads 1 (10. 0) (10. 0 0 1 1 0 1 1 2 2 spinorbital ang.6 -42.2 -17.9 ⇑ . . (10. Degeneracies are denoted by ⇑. to estimate the diﬀerence between the energies of the states 2p 3 and 2p 1 (cf. for example. binding energy / keV Eq. It holds for n = 2 and j = 3 . In order to investigate further the deviation between relativistic and non-relativistic spectra of hydrogen-type atoms we expand the expression (10. 32 (10. Tables 10.1 ⇑ ⇑ ⇑ ⇑ rel.3 spectr. The energies were evaluated with m = 511.15. (10. 0.342 main quantum number n 1 2 2 2 3 3 3 3 3 orbital angular mom.406) α √ 1+ (n − β + β 2 −α)2 The expansion (10. An example for such potential is the Coulomb potential V (r) = −Ze2 /r considered further below.405 ) -161.0041 keV and e2 = 1/137. 0. 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. mom.35. We assume for the wave function the stationary state .408) Radial Dirac Equation We want to determine now the wave functions for the stationary states of a Dirac particle in a 4-vector potential Aµ = (V (r). j 1 2 1 2 1 2 3 2 1 2 1 2 3 2 3 2 5 2 Relativistic Quantum Mechanics non-rel.3). 1 2 2 2 2 E (2p 3 ) − E (2p 1 ) ≈ − 2 2 mZ 4 e8 2 · 23 1 −1 2 = mZ 4 e8 .405).

pp. 1 |ˆ) r 2 2 (10. i. i. (10. 10. (6.413) ˆ σ · p g(r) Yjm (j − 1 . in particular.e. 1 |ˆ) introduced in r 2 2 1 Sect. 1 |ˆ) and Yjm (j − 1 . 6. r r 2 2 2 a(r) b(r) 1 1 1 1 Yjm (j + 2 .410) 1 where Φ(r).414) ˆ σ · p rf (r) Yjm (j + 1 .416) The diﬀerential equations (10. 2 |ˆ) and does r 2 ˆ has odd parity and it holds [c.345). 6. r 2 (10.417) c(r) d(r) X (r) 1 1 1 1 Yjm (j + 2 .2 degrees of freedom.412) are four-dimensional with r-dependent wave functions.f.5 [see.415) 1 ˆ σ · p rg(r) Yjm (j − 2 . 10. The Dirac equation reads then. 6. = r (10.411) (10. ˆ σ · p X + m Φ + V (r) Φ ˆ σ · p Φ − mX + V (r) X = = Φ X (10. 1 |ˆ) . 1 |ˆ) r 2 1 1 = ∂r r r r (10. 1 |ˆ) r 2 2 = i ∂r + −j r 1 g(r) Yjm (j + 2 . 10.10: Dirac Particles in Electromagnetic Field form Ψ(xµ ) = e−i t 343 Φ(r) X (r) . The arguments above allow one to eliminate the angular dependence by expanding Φ(r) and X (r) 1 in terms of Yjm (j + 2 .414) These equations can be brought into a more symmetric form using ∂r + which allows one to write (10.. X (r) ∈ C2 describe the spatial and spin. This term has been discussed in detail in Sect.197. 1 |ˆ). σ · p ˆ σ · p f (r) Yjm (j + 1 . m-values. 1 |ˆ) r 2 2 = i ∂r + j+ r 1 2 3 2 f (r) Yjm (j − 1 .413.198)] not couple states with diﬀerent j. according to (10. r 2 2 (10. 168]: the term is a scalar (rank zero tensor) in the space of the spin-angular momentum states Yjm (j ± 1 . 2 |ˆ) + r Yjm (j − 2 .411. but are timeindependent. 2 |ˆ) r r r .5. 2 |ˆ) + r Yjm (j − 2 . 2 |ˆ) r Φ(r) r . 1 |ˆ) r 2 2 = i j+ ∂r + r j+ r 1 2 1 r f (r) Yjm (j − 2 .. the term is block-diagonal in the space spanned by the states Yjm (j ± 1 . 1 |ˆ) .232.e.10. 1 |ˆ) r 2 = i ∂r − 1 2 r g(r) Yjm (j + 1 .412) ˆ In this equation a coupling between the wave functions Φ(r) and X (r) arises due to the term σ · p.

2 |ˆ) r r where the factors i and −1 have been introduced for convenience.418) holds for f1 (r).411. d(r) are coupled and b(r). 2 |ˆ) = (10. we consider the case that only states for one speciﬁc j. the orthonormality property (6.344 Relativistic Quantum Mechanics In general.418) ∂r + j+ r 1 2 c(r) + [ m + V (r) − ] b(r) = 0 i j+ ∂r − r 1 2 b(r) + [ −m + V (r) − ] c(r) = 0. 10. m values.417) into (10.421) g2 (r) X (r) 1 1 − Yjm (j + 2 .423) . 2 |ˆ) r r f2 (r) 1 1 r Φ(r) i r Yjm (j − 2 . According to (10. 1 |ˆ) r Φ(r) 2 2 r = (10. 10. Presently. Accordingly. only a(r).157).417) of the form f1 (r) i Yjm (j + 1 .419) Obviously.416).420) g1 (r) X (r) 1 1 − Yjm (j − 2 . g2 (r) ∂r + 1 2 f1 (r) − [ + m − V (r) ] g1 (r) = 0 (10.415. (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. there exist two independent solutions (10. g1 (r) ∂r − j+ r j+ r 1 2 g1 (r) + [ − m − V (r) ] f1 (r) = 0 ∂r + and for f2 (r). such expansion must include states with all possible j.412). m pair contribute. Inserting (10. 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. c(r) are coupled. using (10.

Similarly. (10.422). (10.422).e. We assume the 4-vector potential of pure Coulomb type (10. According to this procedure. 2p 3 . 10. We note that (10. can be written ∂r − j+ r j+ r 1 2 g1 (r) + Ze2 f1 (r) r Ze2 g1 (r) r = 0 ∂r + 1 2 f1 (r) − = 0.422) follows in this case from the same procedure as that adopted for the radial wave function of the non-relativistic hydrogen-type atom. n − 1. 3s 1 . . one demonstrates ﬁrst that the wave function at r → 0 behaves as rγ for some suitable γ. Hence. one demonstrates then that the wave functions for r → ∞ behaves as exp(−µr) for some suitable µ. except for the opposite sign of the term (j + 1 ). etc. = 0. . . 2 |ˆ) r Ψ(xµ ) ≈ . the radial wave functions for Dirac particles in the potential (10.420) is the large component and X is the small component. etc.423) are identical.426) . (10.422) determines solutions of the form (10. . 3d 5 . i. (10. . 1. . Behaviour at r → 0 We consider ﬁrst the behaviour of the solutions f1 (r) and g1 (r) of (10. 1 |ˆ) r i 2 2 .422) and (10.422.425) r 0 covers states with angular momentum 1s 1 .. the ones given in (10. (10. to states with angular momentum = j + 2 . namely. 3d 3 . 3p 1 . The solution of (10.10: Dirac Particles in Electromagnetic Field 345 Equations (10.405) of the relativistic hydrogen atom the corresponding states have quantum numbers n = 1.424) Ψ(xµ ) ≈ r 0 1 i.422) near r = 0.423) apply for V (r) = −Ze2 /r. .422).422) corresponds to states f1 (r) Yjm (j + 1 . . enforcing the polynomial to be of ﬁnite order leads to discrete eigenvalues . Equation (10.405). correspodingly the states We consider ﬁrst the solution of (10. Φ in (10. In the non-relativistic limit. for small r.420).Wave Functions We want to determine now the wave functions of the stationary states of hydrogen-type atoms which correspond to the energy levels (10. in the non-relativistic limit wave functions f2 (r) 1 1 i Yjm (j − 2 . together with the appropriate boundary conditions at r = 0 and r → ∞. According to the discussion of the spectrum (10.10. and obtains ﬁnally a polynomial function p(r) such that rγ exp(−µr)p(r) solves (10.423).367) which is spherically symmetric such that equations (10. determining wave functions of the type (10. 2 2 2 2 2 2 = j − 1 2 and.e.421).405). 2s 1 . the 2 equations determine. Dirac Particle in Coulomb Field .409).. 2.422) describes the states 2p 1 . 3p 3 . Hence. 2 2 2 (10.

(10. for small r. determined that the solutions f1 (r) and g1 (r). g1 (r) →∞ ∼ e−µr .427). g1 ∼ exp(± m2 − 2 r). One obtains γ = ± (j + 1 )2 − Z 2 e4 . assume the r-dependence in (10. according to (10. µ = m2 − 2 (10. 10. ) f1 (r) ( + m ) g1 (r) (10. r2 (10. becomes imaginary.427) with 1 γ = (j + )2 − Z 2 e4 . (10.266.430) 2 Note that the exponent in (10.434) √ The solutions of these equations are f1 .267. hence. Such r-dependence 2 would make the expectation value of the potential r2 dr ρ(r) 1 r (10.432) = = − ( − m.427) makes only the positive solution 2 possible. Only the exponentially decaying solution is admissable and. we conclude f1 (r) →∞ ∼ e−µr .431) inﬁnite since. We have.346 Setting f1 (r) yields γ b rγ−1 − (j + 1 ) b rγ−1 + Ze2 a rγ−1 2 γ a rγ−1 + (j + 1 ) a rγ−1 − Ze2 b rγ−1 2 or γ + (j + 1 ) −Ze2 2 2 Ze γ − (j + 1 ) 2 a b →0 Relativistic Quantum Mechanics ∼ a rγ . hence. The assumed r-dependence in (10. 10.435) .428) = 0.433) = = ( m2 − 2 ) g1 (r) ( m2 − 2 ) f1 (r) (10. for the particle density holds then ρ(r) ∼ |rγ−1 |2 = Behaviour at r → ∞ For very large r values (10.427) = = 0 0.420). in case j + 1 )2 < Ze2 .429) This equation poses an eigenvalue problem (eigenvalue −γ) for proper γ values. (10.422) becomes ∂r g1 (r) ∂r f1 (r) Iterating this equation once yields 2 ∂r g1 (r) 2 ∂r f1 (r) 1 . g1 (r) →0 ∼ b rγ (10.

440) (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 .436).442) which leads to a partial cancellation of the asymptotically dominant terms. √ g1 (r) = − m − a e−µr .10.439) − m− e g1 (r) ˜ where µ is given in (10.441) correspond to (10. (10. (10.437) where they ˜ ˜ cancelled in case f1 = g1 = a.434) results in √ √ (m − ) m + a − (m − ) m + a √ √ (m + ) m − a − (m + ) m − a which is obviously correct.436) f1 (r) = Insertion into (10.435 ). g1 (r) = φ1 (r) − φ2 (r) ˜ (10. µ is real. we set f1 (r) g1 (r) = = √ √ ˜ m + e−µr f1 (r) −µr = = 0 0 (10.h.422) for the Coulomb potential V (r) = −Ze2 /r We assume a form for the solution which is adopted to the asymptotic solution (10.441) The last two terms on the l.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 ˜ (10.444) (10.438) (10. Accordingly. Solution of the Radial Dirac Equation for a Coulomb Potential To solve (10.s. In the present case the functions f1 and g1 cannot be chosen ˜ ˜ identical due to the terms in the diﬀerential equations contributing for ﬁnite r.437) (10. ρ m+ ρ [∂ρ − (10. However. without loss of generality we can choose ˜ f1 (r) = φ1 (r) + φ2 (r) .440) and (10. We also introduce the new variable ρ = 2µ r .445) . Let us consider then for the solution of (10.434) √ m + a e−µ r .10: Dirac Particles in Electromagnetic Field For bound states holds 347 < m and. Equation (10. hence. of both (10.

430) which conform to the proper r → 0 behaviour determined above [c.430) one can write this βs = s+γ − 2 √ Ze m2 − 2 s (s + 2γ) βs−1 .447) of the form n φ1 (ρ) = ργ s=0 n γ s=0 αs ρ2 βs ρ2 (10.449) for γ given in (10.453) (s + γ)2 + Z 2 e4 − (j + 1 )2 2 2 √ Ze m2 − 2 βs .450) −√ mZe2 m2 − 2 βs ρs+γ−1 = 0 mZe2 m2 − αs (s + γ) β2 + (j + 1 ) αs + √ 2 s 2 +√ From (10.348 Relativistic Quantum Mechanics Addition and subtraction of these equations leads ﬁnally to the following two coupled diﬀerential equations for φ1 and φ2 ∂ρ φ1 + ∂ρ φ2 + j+ ρ 1 2 Ze2 mZe2 φ1 − √ φ2 = 0 m2 − 2 ρ m2 − 2 ρ mZe2 Ze2 φ1 + √ φ1 + √ φ2 − φ2 = 0 m2 − 2 ρ m2 − 2 ρ j+ ρ φ2 − √ 1 2 (10. Using (10. (10.450) follows Ze2 m2 − 2 βs − βs−1 =0 (10.451) 1 √mZe − (j + 2 ) 2 2 αs .448) φ2 (ρ) = ρ (10. 10.448.452) From (10.f.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 (10.446) (10.447) leads to (s + γ) αs + (j + 1 ) βs − √ 2 s Ze2 m2 − 2 αs (10. (10.454) .447) We seek solutions of (10.446 . = m− Ze2 βs s + γ − √m2 − 2 2 (10. Inserting (10.446.426–10. 10. 10.430)].449) into (10.

405). so = n . . . The expressions (10.455) this conﬁnement of so implies discrete values for ...456) − so 2 √mZe m2 − 2 (−so )(1 − so )(2 − so ) . conclude 2 2 2 that the polynomials in (10..449) must be of ﬁnite order. c c(c + 1) 2! (10. This requires that all coeﬃcients αs and βs must vanish for s ≥ n for some n ∈ N.461 for values given by (10. 2. α + 1.459) − 2 √mZe m2 − 2 (10. . For example. 3p 1 . in fact. with the associated Laguerre polynomials L(α) = F (−n. (s − so ) β0 s! (2γ + 1)(2γ + 2) · · · (2γ + s) (10. 10. for the states 2p 1 . We can. ρ) .448.448. = From (10. .10: Dirac Particles in Electromagnetic Field Deﬁning so = √ one obtains βs = = . namely. x) = 1 + a a(a + 1) x2 x + + . 2. is an integer.10. According to the deﬁnitions (10. 10. 10.455) are ﬁnite.405) and the ensuing so values ( 10. . n It holds φ1 (ρ) φ2 (ρ) = = ργ F (−so .461) In order that the wave functions remain normalizable the power series (10.430. .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. β0 j+ 1 2 (10.449) and (10. 1. 1. equivalently. hence. (s − so ) β0 s! (2γ + 1)(2γ + 2) · · · (2γ + s) (10.457) with the conﬂuent hypergeometric functions F (a.457) One can relate the polynomials φ1 (ρ) and φ2 (ρ) deﬁned through (10. x) .405) allows one to identify n = n − j + 2 which.460) (10.e. 3d 3 holds n = 1. c.456) and (10. 2γ + 1.452) follows j+ αs = 1 2 349 Ze2 m2 − 2 −γ (10. .457) for βs and αs imply that so must then be an integer.455) s − so βs−1 s(s + 2γ) (s − 1 − so )(s − so ) βs−2 (s − 1)s (s − 1 + 2γ)(s + 2γ) (1 − so )(2 − so ) . Comparision with (10. ρ) so β0 ργ F (1 − so . . (10. 2γ + 1.458) or. i. n = 0. .456. . 10. m (n ) = .

and (10. 10. are two-dimensional vectors determined through the explicit form of the spin-orbital angular momentum states Yjm (j ± 1 . given by expression (10. j. The wave functions (10. The normalization integral is then ∞ 0 π 2π r2 dr 0 sin θdθ 0 dφ (|Φ(r)|2 + |X (r)|2 ) = 1 (10. Sect. 2µr) (10. Due to the form (10.158) of the latter states absorb the angular integral in (10.469) This formula has been adapted from ”Relativistic Quantum Mechanics” by W. 6. Greiner.267). 2γ + 1.m (j − 1 . j. is time-independent.438. (n + γ)m N (2µr)γ−1 e−µr − κ F (−n .466) F1 (r) = F− (κ|r) . The coeﬃcients β0 in (10.461). m|xµ ) = e−i t iF1 (r) Yj. 1 |ˆ) 2 2 r κ = j+ 1 2 (10. m.147.157. as in (10. 9. 10. 1 |ˆ) 2 2 r G1 (r) Yj. The complete wave function is given by the following set of formulas 1 Ψ(n.464) 0 The evaluation of the integrals.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 .442). (10. (Springer.410) of the stationary state wave function the density ρ(xµ ) of the states under consideration. 6.421)] ∞ dr (|f1 (r)|2 + |g1 (r)|2 ) = 1 (10. can follow the procedure adopted for the wave functions of the non-relativistic hydrogen atom and will not be carried out here. which involve the conﬂuent hypergeometric functions in (10.460.421).467) ± n F (1 − n . Berlin. where2 F± (κ|r) = G1 (r) = F+ (κ|r) .m (j + 1 .6. 1990).460. 10. 10. = j + 2 .439).350 Relativistic Quantum Mechanics Altogether we have determined the stationary states of the type (10. 2γ + 1. They are described through quantum numbers n.421) correspond to non-relativistic states with orbital angular momentum 1 1 = j + 2 .463) where Φ and X . (10. 1 |ˆ) in (6.421) with radial wave functions f1 (r). 2 . g1 (r) determined by (10.465) (10.463) and yield [note the 1/r factor in (10. = j + 2 . 2µr) (2µ) 2 Γ(2γ + 1) 3 N = and m 4m )Γ(2γ + n + 1) (n +γ)m (n +γ)m (10.148). The orthonormality 2 2 r properties (6.460.461) are to be chosen to satisfy a normalization condition and to assign an overall phase.461).

G2 (r) = F+ (κ|r) . 3s 1 .465–10.469) that the following wave functions result Ψ(n.m (j + 1 . m|x ) = e −i t iF2 (r) Yj. One can verify. The radial wave functions f2 (r) and g2 (r) in (10.422) to (10. etc.10: Dirac Particles in Electromagnetic Field 351 We want to consider now the stationary states of the type (10. We have. . obtained closed expressions for the wave functions of all the stationary bound states of relativistic hydrogen-type atoms. this wavefunction has an orbital angular momentum quantum number = j − 1 and.423) which diﬀers from the radial Dirac 1 equation for f1 (r) and g1 (r) solely by the sign of the terms (j + 2 )/r.471) (10.469). hence.469). 2 accordingly. not not 2 2 2 2 2 2 covered by the wave functions given by (10. 1 |ˆ) 2 2 r G2 (r) Yj. become f2 (r) 1 1 i Yjm (j − 2 . where F± (κ|r) are as given in (10. in the non-relativistic limit.467–10.421) are governed by the radial Dirac equation (10. tracing all steps which lead from (10. 3p 3 .472) F2 (r) = F− (κ|r) . j.470) r 0 Obviously. = j − µ 1 2 . 3d 5 . 1 |ˆ) 2 2 r κ = −j − 1 2 (10. 2s 1 . (10.421) which. 2p 3 .10. describes the complementary set of states 1s 1 .m (j − 1 . 2 |ˆ) r Ψ(xµ ) ≈ e−1 t .

352 Relativistic Quantum Mechanics .

.Chapter 11 Spinor Formulation of Relativistic Quantum Mechanics 11. isomorphic to the natural representation.e. in a sense. ˜ We have altered here slightly our notation of S(w. We will proceed by building as much as possible on the results obtained sofar in the chiral representation of the Dirac equation. a complex variable z.e.2) (11. The lower dimensionality of a(z) and b(z) implies. i. z ∈ C ˜ Because of its block-diagonal form each of the diagonal components of S(z). ϑ)µ ν .225–10.229). ϑ). Starting point is the Lorentz transformation S(w.3) (11. diﬀerent L(w. in fact. ϑ) for the ˜ in the chiral representation as given by (10. This transformation bispinor wave function Ψ can be written ˜ S(z) a(z) b(z) z = = a(z) 0 0 b(z) exp 1 z·σ 2 1 − z ∗· σ 2 (11. This fact is remarkable since it implies that the representations provided through a(z) and b(z) are of lower dimension then the four-dimensional natural representation1 L(w. indeed. We will characterize the space on which the transformations a(z) We will see below that the representations a(z) and b(z) are. may be exploited to express the Lorentz-invariant equations of relativistic quantum mechanics. that the corresponding representation of the Lorentz group is more basic than the natural representation and may serve as a building block for all representations.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. expressing its dependence on w. must be two-dimensional irreducible representations of the Lorentz group. ϑ through 3. what will be achieved in the following. a(z) and b(z). i. in particular.1) (11. 1 353 .. ϑ)µ ν correspond to diﬀerent a(z) and b(z).4) = exp = w − iϑ . This is.262).

In this case holds a(z) = b(z) ˜ 1 (xµ ). Aν . ϑ ∈ R3 . Ψ4 (xµ ) transform according to ˜ ˜ ˜ and Ψ b(z). express 4-vectors Aµ . +1 2 |1. ﬁnally. Ψ2 (xµ ) and Ψ3 (xµ ). ϑ ∈ R3 . β. γ (see Chapter 5). (11. for z = iϑ. ϑ)µ ν and a(z). (11. a distinct diﬀerence in the transformation behaviour of ˜ ˜ ˜ ˜ Ψ1 (xµ ). −1 2 2 |1. (11.. − 1 2 2 z = iϑ. formulate the Dirac equation. and since b(z) ˜ ˜ acts on the third and fourth component of Ψ one can characterize Ψ ˜ Ψ1 (xµ ) ˜ Ψ2 (xµ ) Ψ3 (xµ ) ˜ ˜ Ψ4 (xµ ) ∼ ∼ ∼ ∼ 1 |2. are elements of SU(2) and correspond.5) The transformations in this case. will establish the map between L(w.9) a∗ (z) = exp 2 . ϑ ∈ R3 (11. A First Characterization of the Bispinor States We note that in case w = 0 the Dirac transformations are pure rotations.243). and the Klein–Gordon equation in the spinor representation.6) as given by (5. usually expressed 2 as product of rotations and of functions of Euler angles α.. ± 1 represents the familiar spin– 1 states. the so-called spinor space.7) where “∼” stands for “transforms like”. b(z). β. Ψ4 (xµ ) in case z = w + iϑ for w = 0. This characterization allows one to draw conclusions regarding the state space in which a(z) and b(z) operate. to the rotational transformations of spin. the operator ∂µ and the Pauli matrices σ in the new representation and. −1 2 2 z = iϑ.e. a space of vectors state1 for which holds state2 state 1 ∼ | 1 . namely. Relationship Between a(z) and b(z) The transformation b(z) can be related to the conjugate complex of the transformation a(z). Here | 1 . 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 . + 1 2 2 state 2 ∼ | 1 . θ ∈ R3 .1 states as described by Dm m (ϑ).8) We like to stress that there exists.354 Spinor Formulation and b(z) act. 2 2 2 ˜ Since a(z) acts on the ﬁrst two components of the solution Ψ of the Dirac equation. however. +1 2 2 |1. in 2 fact. i. For ϑ = (0. Ψ2 (xµ ) transform according to a(z) whereas Ψ3 (xµ ).e. neutrino equation. i. In this case a(z) and b(z) are identical and read a(i ϑ) = b(i ϑ) = exp 1 − ϑ·σ 2 . to 1 ∗ ∗ z ·σ (11.

(11.224. (11. Ψ (xµ ) as well as Ψ (xµ ). The eﬀect of inversion on Lorentz transformations is. therefore.2.10) From this one can derive b(z) = where = 0 1 −1 0 .12) To prove (11. Hence. .1: Lorentz Transformation of Dirac Bispinor ∗ ∗ ∗ where σ ∗ = (σ1 .11) 0 −1 1 0 = .11.f.1) can be written in the form ˜ S(z) = a(z) 0 0 a∗ (z) −1 (11.4) and .13) Explicit matrix multiplication using (5. in fact. This demonstrates that a(z) is the ˜ transformation which characterizes both components of S(z). (5. one can express a∗ (z) −1 = exp 1 − z ∗· σ 2 = b(z) . 355 (11.10.12) does. (11. The answer lies in the necessity that application of spatial inversion should transform a solution of the Dirac equation into another possible solution of the Dirac equation. 11. 11.224)] ∗ σ1 = σ1 . a∗ (z) −1 = exp 1 ∗ z σ∗ 2 −1 .12). −1 a∗ (z) −1 (11. µ ). σ3 ). −1 = ∗ −σ3 = −σ ∗ .12) yields σ1 σ3 or in short σ Similarly. −1 given by (11.14) . that the transformation (11. but leave rotation angles ϑ unaltered. σ2 .16) σ = −σ ∗ . one can show −1 −1 −1 = −σ1 = σ2 = −σ3 ∗ = −σ1 ∗ = −σ2 σ2 −1 (11. ∗ σ 3 = σ3 . a result to be used further below. One can readily verify [c.11) one ﬁrst demonstrates that for and −1 as given in (11. hold −1 = 1 One notices then. using f (a) −1 = f ( a −1 ). Ψ (xµ ) even though these pairs of ˜ ˜3 ˜4 featuring the components Ψ1 (x ˜ 2 components transform independently of each other. (11. ∗ σ 2 = − σ2 . 11.17) We conclude.18) with a(z) given by (11. 1. Spatial Inversion ˜ One may question from the form of S(z) why the Dirac equation needs to be four-dimensional.15) (11. however. that they alter w into −w.

ϑ ∈ R3 .2.224)] tr( σj ) = 0 . 2.e. 11.21) = etr( 2 z·σ) = 1 1 (11. for particles like the neutrinos which do not obey inversion symmetry two components of the wave function are suﬃcient. Exercise 1. the space spanned by only two of the components of Ψ is not invariant under spatial inversion and.21) together with matrix multiplication as the binary operation forms a group.2 Relationship Between the Lie Groups SL(2. (11.356 Spinor Formulation ˜ Let P denote the representation of spatial inversion in the space of the wave functions Ψ. hence. (11. Amsterdam. which describes pure rotations.. B¨uerle and E. can be found in G. For the general case the proof. In fact. based on the Jordan–Chevalley theorem.19) i.. ˜ ˜ P S(w + iϑ) P = S(−w + iϑ) = b(z) 0 0 a(z) . C) deﬁned in (11. does not suﬃce for particles like the electron which obey inversion symmetry.C) and SO(3.f. Obviously.22) which follows from the fact that for any complex. The transformation S(z) in the transformed space is then ˜ P 1. the Lorentz invariant equation for neutrinos is only 2–dimensional. 3 . (11. M is a complex 2 × 2–matrix.A. det(M ) = 1 } . C) = { M.G. 1990).10. the transformations a(z) and b(z) become interchanged. (5. non-singular matrix M holds2 det eM and from [c.20) ˜ Obviously.3. However.e.A. de Kerf Lie a Algebras.1) We have pointed out that a(iϑ).1: Show that SL(2. One can verify this by evaluating the determinant of a(z) det( a(z) ) = det e 2 z·σ 1 (11.24) = etr(M ) (11. This implies ˜ Ψ1 ˜ Ψ2 P ˜ Ψ3 ˜ Ψ4 ˜ Ψ3 ˜ Ψ4 = Ψ1 ˜ ˜ Ψ2 . is an element of SU(2). P −1 = P. 2 . a(w + iϑ) for w = 0 is an element of SL(2. 2 = 1 i. j = 1. The proof of this important property is straightforward in case of hermitian M (see Chapter 5). However. Part (Elsevier.23) Exercise 11.

. Noting that for covariant vectors according to (10.75) holds aν = Lν µ aµ one realizes that σµ transforms inversely to covariant 4–vectors. Argue why M (Aµ ) = σµ Aµ provides a bijective map.25) of the matrix M (Aµ ) is independent of the frame of reference. or+ thochronous Lorentz transformations.26) or from the fact that the matrices σ0 .28) and (11.25) can also be written M (Aµ ) = A0 + A3 A1 − iA2 A1 + iA2 A0 − A3 . the matrix M (Aµ ) is hermitian as can be seen from inspection of (11.76). (11.29) where we used (10. The function M (Aµ ) is bijective. in a transformed frame should hold Lρ ν σ µ Aµ σµ A µ .32) . one wishes that the deﬁnition (11. Starting point is a bijective map between R4 and the set of two-dimensional hermitian matrices deﬁned through M (Aµ ) = σµ Aµ where (11.e.135)] this transformation behaviour. σ1 .25) σ0 σ1 σ2 σ3 The quantity σµ thus deﬁned does not transform like a covariant 4–vector.31) holds.26) (11. σ2 . −→ µ Consistency of (11.30).31) Exercise 11.30) Since the components of Aµ are real.C) and SO(3. We will prove below [cf. (11.2. 0 1 1 0 . M (Aµ ) according to (11. 0 −i i 0 . (11. in fact. 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 .11. (11. σ3 provide a linear–independent basis for the space of hermitian 2 × 2–matrices. C) and the group L↑ of proper. (11.27) requires then σ µ Aµ Lν µ σµ = σν (11.27) −→ Straightforward transformation into another frame of reference would replace σµ by σµ . i. σ2 . σ1 . 1 0 0 −1 . Demonstrate that (11.2: Show that σ0 .1) Mapping Aµ onto matrices M (Aµ ) 357 We want to establish now the relationship between SL(2. 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 . (11. σ3 are hermitian. In fact.28) σµ = 1 0 0 1 .2: Lie Groups SL(2.

33) This transformation conserves the hermitian property of M since (M )† = ( a M a† )† = (a† )† M † a† = a M a† = M (11.32.40) is an element of L↑ .33) to M (Aµ ) describing the action of the transformation in terms of transformations of Aµ . + .37) using (11. In fact.31).f. 11. In fact. (11.37) which implies that (11.35) We now apply the transformation (11.33) det(M ) = = det( a M a† ) = det(a) det(a† ) det(M ) [det(a)]2 det(M ) = det(M ) . C) .35) holds for this transformation A µ Aµ = Aµ Aµ (11.33) conserves the determinant of M . (11. C) and for any Aµ there exists an A µ such that M (A µ ) = a M (Aµ ) a† . L(a)µ ν = 1 tr a σν a† σµ 2 .25) Aµ = which allows us to express ﬁnally A µ = L(a)µ ν Aν .3: Show that L(a)µ ν deﬁned in (11. The linear character of the transformation becomes apparent expressing A µ as given in (11. C) deﬁnes the transformation [c.358 Transforming the matrices M (Aµ ) Spinor Formulation We deﬁne now a transformation of the matrix M (Aµ ) in the space of hermitian 2 × 2–matrices a M −→ M = a M a† .40) 1 tr a σν a† σµ 2 Aν (11.2. Due to det(a) = 1 the transformation (11. Accordingly. any a ∈ SL(2. a ∈ SL(2.34) where we used the properties M † = M and (a† )† = a. (11.31)] 1 a Aµ −→ A µ = tr a M (Aµ ) a† σµ 2 Because of (11.36) The suitable A µ can readily be constructed using (11. (11.39) Exercise 11. (11. for any a ∈ SL(2. (11.38) . it holds for the matrix M deﬁned through (11.37) deﬁnes actually a Lorentz transformation.

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 .e.45) One can demonstrate through direct evaluation 2 α = 2 α = 2 α = = 2 α = 0 else β β γ γ = = = = γ = γ = 1.46) 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 . C) and of SO(3. .1) deﬁned through L(a)µ ν [cf.42) = ν Deﬁning Γ = a† σµ a1 .11.43) (σν )αβ Γβα Γγδ (σν )δγ = ν.2: Lie Groups SL(2.40)] respects the group property of SL(2. ¯ Lµ ρ = L(a1 )µ ν L(a2 )µ ρ = L( product in SO(3.41) results in 1 ¯ Lµ ρ = 4 where Aαβγδ = ν Γ = a2 σρ a† 2 (11. (11.47) This completes the proof of the homomorphic property of L(a)µ ν .δ 1 4 Aαβγδ Γβα Γγδ α.δ (11.α. C) and SO(3. β 2.1) For this purpose one writes using tr(AB) = tr(BA) L(a1 )µ ν L(a2 )µ ρ = ν a1 a2 product in SL(2.C) and SO(3.β γ. i.β γ.C) )µ ρ (11..1) L(a)µ ν provides a homomorphism 359 We want to demonstrate now that the map between SL(2.1). (11. (11. 1 ¯ and using the deﬁnition of Lµ ρ in (11.44) (σν )αβ (σν )δγ . β δ = δ = = δ = δ 1 2 = 2 = 1 Aαβγδ (11. 2 (11.

J3 of the Lorentz transformations Lµ ν in the natural representations. To this end we consider inﬁnitesimal transformations and keep only terms of zero order and ﬁrst order in the small variables. 0. 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 . To obtain the generator of a(z) corresponding to the generator K1 . K2 . 0.54) This reads for the four cases Aµ = (1.55) (11.57) . Aµ = (0. denoted below as κ1 .49).58) .48) yields for the l.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 ) . 1. 0).48) We want to determine now the generators of the transformation a(z) as deﬁned in (11. 0.50) (11. Aµ = (0. C) as complex 2 × 2–matrices are deﬁned through four complex or. i. −A0 . J1 . of (11. (11. Aµ = (0. 0.360 Generators for SL(2. J Spinor Formulation The transformations a ∈ SL(2. 1 (11. we write (11. 10.s.s. eight real numbers.25) in the last step.h. K3 .56) (11. 0. (11. correspond to the generators given by (10.51) Insertion of (K1 )µ ν as given in (10. For the r. M (Aµ + (K1 )µ ν Aν ) = = M (Aµ ) + M (Aµ ) − M ( (−A1 .36) M (Lµ ν Aν ) = a M (Aµ ) a† (11.47. 1 (11..e.h. 1) − σ1 σ0 0 0 One can verify readily that 1 κ1 = − σ 1 2 (11. 0.48). noting the linearity of M (Aµ ). 1. 0). correspondingly. J2 . 0.49) assuming (note that g µ ν is just the familiar Kronecker δµν ) Lµ ν a = = gµν + 1 + 1 (K1 )µ ν κ1 .52) and (11. of (11. C) which correspond to K.52) where we employed (11.53) Equating (11.53) results in the condition σ 0 A1 − σ 1 A0 = ( κ 1 σ µ + σ µ κ † ) Aµ . 0). (11. Because of the condition det(a) = 1 only six independent real numbers actually suﬃce for the deﬁnition of a.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 (11.2) which correspond to the generators K1 . 0) ) σ0 A 1 − σ1 A 0 (11.49) we obtain using (11.

3 Spinors Deﬁnition of contravariant spinors We will now further characterize the states on which the transformation a(z) and its conjugate complex a∗ (z) act. 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 .4: Show that the generators (11. λ3 corresponding to J1 . 11. (11. 2 (11. + 1 2 2 1 transforms under rotations (z = iϑ) like a spin– 2 state | 1 . Similarly.60) of a ∈ SL(2. Here a(z) = ( aα β ) = describes the matrix a(z). def α = 1. w describing boosts and ϑ describing rotations. − 1 . 2 λ = i σ. J3 are given by 1 κ = − σ. 2 2 φ1 φ2 ∈ C2 .63) where we extended the summation convention of 4-vectors to spinors. hence. 2 (11. J2 .1) acts on 4–vectors Aµ . state that the following two transformations are equivalent L(w. C) acts on spinors φα ∈ C2 (characterized below) 1 (11. a(w − iϑ) = e− 2 (w − iϑ)·σ ∈ SL(2. κ3 of a(z) corresponding to K2 . C) correspond to the generators K and J of Lµ ν .61) This identiﬁes the transformations a(z = w − iϑ) as representations of Lorentz transformations.64) .2.60) We can. Exercise 11. 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 . λ2 . K3 and λ1 .3: Spinors 361 obeys these conditions. ϑ) = ew·K + ϑ·J ∈ SO(3.11. one can show that the generators κ2 . a1 1 a1 2 a2 1 a2 2 (11.

. ± 1 . To arrive at a suitable scalar product we consider ﬁrst only rotational transformations a(iϑ). 0| 1 .68) (11. . In this case spinors φα transform like spin– 1 states and an invariant. as covariant spinors. 1 . . 0| 1 . i. invariant under transformations a(z). as contravariant spinors and to φα . etc. The existence of a scalar product gives rise to the deﬁnition of a dual representation of the states φα denoted by φα . .. m 1 | . χβ . Deﬁnition of covariant spinors Expression (11.e. 1 . χβ .. 1 ) = 2 2 2 2 2 2 2 2 √ ±1/ 2 and equating the spin states of “particle 1” with the spinor φα . exchange of φα and χβ alters the sign of the expression. which 2 can be constructed from products φ1 χ2 . 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.e. is the singlet state. i. i. m.362 Deﬁnition of a scalar product Spinor Formulation The question arises if for the states φα there exists a scalar product which is invariant under Lorentz transformations. 1 ) stands for the Clebsch–Gordon coeﬃcient. this scalar product is anti-symmetric. . ..67) We will refer to φα . should remain invariant under transformations a(iϑ). ..70) ‘Scalar’ implies invariance under rotations and is conventionally generalized to invariance under other symmetry trasnformations. . (11.66) should constitute a singlet spin state. Using (0. invariant and as such has the necessary properties of a scalar3 product.66) is a bilinear form.e. 0| . those of “particle 2” with the spinor χβ one can state that the quantity 1 Σ = √ 2 φ1 χ2 − φ2 χ1 (11. In fact. as we will demonstrate below such states are invariant under general Lorentz transformations a(z).69) (11. indeed. The corresponding states are deﬁned through φ1 χ2 − φ2 χ1 = φ1 χ1 + φ2 χ2 It obviously holds χα = χ1 χ2 = χ2 −χ1 . The relationship between the two can be expressed φ1 φ2 = = 3 φ1 φ2 0 1 −1 0 (11. However. −m 2 2 2 2 1 1 1 1 2 (11. In the notation developed in Chapter 5 holds for the singlet state | . . −m) | .65) where | · · · 1 describes the spin state of “particle 1” and | · · · 2 describes the spin state of “particle 2” and (0. Such a scalar product does. 0. 0 = 2 2 m=± 1 2 1 1 (0. ± 1 .

11. (10.68). The transformation behaviour of φα according to (11.69.71. In the transformed frame holds φ α χα = aα β αγ aγ δ δκ φβ χκ . we will express (11.72)] φα φ α = = = = αβ αβ φβ φβ 11 21 11 21 12 22 12 22 (11.70) is φ1 φ2 −1 363 = = −1 φ1 φ2 .76) αβ αβ The scalar product (11. 11.74) = = 0 1 −1 0 0 −1 1 0 (11. 11.77) (11.72) in a notation analogous to that chosen for contravariant and covariant 4–vectors [cf.75) (11.73.77) is Lorentz invariant.3: Spinors as can be veriﬁed from (11.81) . 11. (11. (11.63.1: Show that for any non-singular complex 2 × 2–matrix M holds M −1 = det(M ) M −1 T The matrices .74)].3.79) = a −1 T a κβ .78) aβ γ γδ φδ (11. Accordingly. −1 connecting contravariant and covariant spinors play the role of the metric tensors gµν . Proof that φα χα is Lorentz invariant We want to verify now that the scalar product (11. 11. For this scalar product holds φα χα = −χα φα . (11.69.10.70) and (11.72) 0 −1 1 0 Exercise 11. The inverse of (11.80) One can write in matrix notation aα β γ δκ αγ a δ αβ (11. (10.73) (11.67) will be expressed as φα χα = φ1 χ1 + φ2 χ2 = φ1 χ2 − φ2 χ1 . g µν of the Minkowski space [cf. 10.71) (11.11.75) is given by φα = as can be readily veriﬁed.

84) (11.364 Using (11. As discussed above. k = 1. hence.89) such that (11.14) one can write a −1 Spinor Formulation = e 2 z·σ 1 −1 = e 2 z· 1 σ −1 = e− 2 z·σ 1 ∗ (11.83) Here we have employed the hermitian property of σ.88) which one veriﬁes taking the conjugate complex of (11.63). ˙ We also deﬁne covariant versions of φα φα = (φα )∗ . it holds φk = (φk )∗ . Obviously.83) into (11. ˙ (11.2) and (11.85) φ1 φ2 ∗ ∗ .81) yields aα β and.e.87) which we will employ from now on.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 .. The transformation ˙ behaviour of φα is ˙ φ α = (aα β )∗ φβ ˙ (11. from (11. a∗ (z) provides a representation of the Lorentz group which is distinct from that provided by a(z).90) extending the summation convention to ‘dotted’ indices. 2.88) reads ˙ ˙ φ α = aα β φβ ˙ ˙ (11. The complex conjugate spinors We consider now the conjugate complex spinors (φ ) = α ∗ γ δκ αγ a δ = a −1 T a κβ = a−1 a κβ = δκβ (11.91) . We denote ˙ (aα β )∗ = aα β ˙ (11. Hence. (11. Insertion of (11. i.80) φ α χα = φβ χβ . σ † = (σ ∗ )T = σ.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.

75.11. A comparision of (11.98) ˜ ∗ (z) −1 Ψ3 a ˜ Ψ 4 obviously implies that the solution of the Dirac equation in the chiral representation can be written in spinor form ˜ 1 µ Ψ1 (xµ ) φ (x ) ˜ Ψ2 (xµ ) φ2 (xµ ) φα (xµ ) = = .92) (11. a property which is rather evident.4: Spinor Tensors 365 The relationship between contravariant and covariant conjugate complex spinors can be expressed in analogy to (11. 11.96) is Lorentz invariant. (11.75).76) φα ˙ ˙ φα αβ ˙ ˙ αβ ˙ ˙ = = = = αβ ˙ ˙ αβ ˙ ˙ αβ αβ φβ φβ ˙ ˙ (11.18) ˜ 3 .18) of the bispinor wave function Ψ. 11. one can state ˜ and (11. in matrix notation. For the spinors φα and χα thus ˙ deﬁned holds that the scalar product ˙ φα χα = φ1 χ1 + φ2 χ2 ˙ ˙ ˙ ˙ ˙ (11..97) implies then that φα transforms like Ψ ˙ φ1 ˙ φ2 ˙ transforms under rotations (z = iϑ) like a spin– 1 state | 1 . Ψ4 . i.100) . A tensor tα1 α2 ···αk β1 β2 ···β (11. (11.e.93) (11.97) The transformation. 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. + 1 2 2 2 1 transforms under rotations (z = iϑ) like a spin– 2 state | 1 .94) (11. a∗ (z) −1 accounting for the ˜ transformation behaviour of the third and fourth spinor component of Ψ.95) ˙ where αβ and αβ are the real matrices deﬁned in (11.4 Spinor Tensors ˙ ˙ ˙ We generalize now our deﬁnition of spinors φα to tensors.73. is governed by the operator a∗ (z) −1 which arises in ˜ the Lorentz transformation (11. − 1 .99) Ψ3 (xµ ) χ ˙ (xµ ) ˜ χβ (xµ ) ˙ 1 ˜ χ2 (xµ ) Ψ4 (xµ ) ˙ 11. The transformation behaviour of φα is ˙ φα = ˙ ˙ β γδ ˙˙ φδ ˙ ˙ αβ a γ ˙ ˙ .

108) Obviously. i. using conventional matrix indices j. The 4–vector Aµ in spinor form We want to provide now the spinor form of the 4-vector Aµ . i. αβ ˙ αγ t ˙ γβ (11.104) Indices on tensors can also be lowered employing the formula tα β = and generalizations thereof.107) which reads in spinor notation [cf. ˙ ˙ ˙ ˙ (11..1: Prove equation (11. (11.102) = .e. tjk = a t a† jk ˙ ˙ ˙ ˙ (11.106) Exercise 11. the transformation bevaviour of a tensor tαβ reads in spinor and matrix notation t αβ = aαγ aβδ tγδ . i.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 .103) Similarly. with ˙ contravariant indices αβ.106). ˙ ˙ ˙ ˙ (11. According to (11. . An example of a tensor is αβ and tions. this transformation behaviour is in harmony with the tensor notation adopted. αβ = αβ (11. it holds αβ . m. This tensor transforms according to t αβ = aα γ aβ δ tγ δ ˙ This reads in matrix notation.25] M (Aµ ) = σµ Aµ .m aj akm t m (11. k.e.. the elements of which are linear functions of Aµ .105) This tensor is actually invariant under Lorentz transforma= αβ .. (11.109) .102. (11.e.m aj a∗ t km m .25) the tensor is explicitly Aαβ = ˙ A1 1 A1 2 ˙ ˙ A2 1 A2 2 ˙ ˙ = A0 + A3 A1 − iA2 A1 + iA2 A0 − A3 . we want to express Aµ through a spinor tensor. tjk = a t aT jk = .103) A αβ = aα γ aβ δ Aγ δ . 11.4.101) An example is the tensor tαβ which will play an important role in the spinor presentation of the Dirac equation. We had demonstrated that M (Aµ ) transforms according to M = M (Lµ ν Aν ) = a M (Aµ ) a† (11. This task implies that we seek a tensor. An obvious candidate is [cf. (11.

110) The 4-vectors Aµ .116) .114) yields ∂αβ = ˙ ∂0 + ∂3 ∂1 + i∂2 ∂1 − i∂2 ∂0 − ∂3 .115) is correct. (11. 11. (11.113) Aα β = ˙ (11. Using (11. ∂µ in spinor notation The relationship between 4–vectors Aµ .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. Aµ can also be associated with tensors Aα β = ˙ This tensor reads in matrix notation A11 A12 ˙ ˙ A21 A22 ˙ ˙ = = 0 1 −1 0 ˙ ˙˙ αγ β δ A ˙ γδ . (11.4.110) one obtains Aα β = ˙ A0 − A3 −A1 − iA2 −A1 + iA2 A0 + A3 A0 + A3 A1 + iA2 A1 − iA2 A0 − A3 .117) ˙ ˙ ∂0 − ∂3 −∂1 + i∂2 −∂1 − i∂2 ∂0 + ∂3 . (11.111) A11 A12 ˙ ˙ A21 A22 ˙ ˙ ˙ 0 −1 1 0 (11.4: Spinor Tensors One can express Aαβ also through Aµ Aα β = ˙ ˙ 367 A0 − A3 −A1 + iA2 −A1 − iA2 A0 + A3 . employing (11.11. ˙ 2 (11.109.110) one can state ∂ αβ = Similarly.115) Exercise 11. Hence. Aµ and tensors tαβ can be applied to the partial diﬀerential operator ∂µ .112) A22 −A21 ˙ ˙ −A12 A1 1 . (11.2: Prove that (11.

f. (11.e.120) σµ 0 .243)]. i.125) a (σ ) Equation (11.123) (a∗ )−1 Lν µ a L ν µ = = σν (σ ) ν ∗ −1 (11.26) and its contravariant analogue σ µ σµ = σµ = 1 0 0 1 1 0 0 1 .126) which is the complex conjugate of (11.243) holds independently of the representation chosen.125) is equivalent to (11.h. (11. (11. 0 1 1 0 .124) .118) yields ˜ γµ = ˜ Using (11.125) is equivalent to µ ∗ −1 −1 a∗ (σ µ )∗ −1 −1 a Lν µ = (σ ν )∗ (11.121). of this equation is 0 ∗ (σ µ )∗ a and. (11.368 σµ in Tensor Notation Spinor Formulation We want to develop now the tensor notation for σµ (11. σµ . Equation (10. . holds also in the chiral representation.124) constitutes the essential transformation property of σ µ and will be considered further. (10.243) in the chiral representation expressing S(Lη ξ ) by (11. Note that (10.122) aσ µ −1 a−1 −1 (a∗ )−1 0 Lν µ (11. (11. γµ = ˜ 0 µ )∗ (σ −1 0 σµ σµ 0 (σ µ )∗ −1 . i. 1 0 0 −1 . one can conclude aσ µ ∗ −1 Lν µ σν 0 = . 0 −i i 0 . Hence. σµ building on the known transformation behaviour of γ µ . This is ˜ possible since γ µ can be expressed through σ µ .. To obtain the transformation properties of σ µ we employ then (10. Comparision of (10.243) 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.. We will obtain the transformation behaviour of σ µ .124).s. hence. hence.121) One expects then that the transformation properties of σ µ should follow from the transformation properties established already for γ µ [c. (11.18) and γ µ by (11.15) one can write σµ = and.229) and (11.124).118) 0 −1 −1 0 0 i −i 0 For this purpose we consider ﬁrst the transformation behaviour of σ µ and σµ . −1 0 0 1 (11.119) (11.e. . .

0 i −i 0 µ=2 .130) We argue in analogy to the logic applied in going from (11..124) using (11.124) a σ µ [a∗ ]T Lν µ = σ ν .132) 1 0 0 1 µ=0 . 11. 11. e.133) Combining (11. . equation (11. i. 1 0 0 −1 µ=3 . In fact.. hence. 11.11. 0 −1 −1 0 µ=1 . according to (11.129) We want to demonstrate now that the expression aσ µ [a∗ ]T is to be interpreted as the transform of σ µ under Lorentz transformations.4).108) that the same transformation behaviour applies then for general Lorentz transformations.4) with w = 0.131) one can express the transformation behaviour of σ µ in the succinct form Lν µ σ µ = σ ν .102. under rotations the Pauli matrices transform like (j = 1. It holds according to (11. (11. (11.127) Exploiting the hermitian property of σ. transformations (11. therefore. (11.g. (σ ∗ )T = σ yields using (11.129. 11. (11.128) One can express.e.103) ˙ identiﬁes σ µ as a tensor of type tαβ . 2.2.4: Spinor Tensors One can rewrite (11. = . (11. One can. (11. −1 0 0 1 µ=3 . 0 1 1 0 µ=1 . state that σ µ in a new reference frame is σ µ = a σ µ a† (11.2. 3) σj −→ a(iϑ) σj a(iϑ) † = a(iϑ) σj a∗ (iϑ) T . 0 −i i 0 µ=2 . 11.131) where a is given by (11. This transformation behaviour.16.2) −1 369 (a∗ )−1 = −1 − 1 z ∗ ·σ ∗ 2 e = e− 2 z 1 ∗ · −1 σ ∗ = e2z 1 ∗ ·σ .134) .2) −1 (a∗ )−1 = e2z 1 ∗ ·(σ ∗ )T = e2z 1 ∗ ·σ ∗ T = [a∗ ]T .107) to (11.118) (σ µ )11 (σ µ )12 ˙ ˙ (σ µ )21 (σ µ )22 and (σµ )11 (σµ )12 (σ µ )21 (σ µ )22 ˙ ˙ 1 0 0 1 µ=0 ˙ ˙ ˙ ˙ = (11.

matrices.133) is characterized through a 4-vector ˙ index µ.137) σαβ ˙ = δαγ δβ δ . ˙˙ (11.139) g −f g and. 1. α. δ. state ˙ ˙ ˙ ˙ (σ11 )11 (σ11 )12 (σ12 )11 (σ12 )12 ˙ ˙ ˙ ˙ ˙ ˙ ˙ ˙ ˙ γδ (σ11 )21 (σ11 )22 (σ12 )21 (σ12 )22 ˙ ˙ ˙ ˙ . the transformation is inverse to that of ordinary 4–vectors. 3 as well as through two spinor indices αβ. γ.12) for of σµ yields together with (5. We can.s. i.136) σαβ = = 2 ˙ σ1 − iσ2 σ0 − σ3 0 0 0 0 1 0 0 1 where on the rhs.224) ˙ Using (11. (11. i. whereas ˙ account for the elements of the individual Pauli the ‘outer’ contravariant spinor indices.138) the ‘inner’ covariant spinor indices. ˙ ˙ In (11. ˙ ˙ 0 −1 1 0 ˙ (11. 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 . and the covariant indices are moved outside.. β.29)]. hence.h. Spinor Formulation (11.e.. Each of the 4 × 4 = 16 matrix elements in (11. We will now consider the representation of σ µ . 2. We want to express now σ µ and σµ also with respect to the 4–vector index µ in spinor form employing (11.e.e. This yields 1 0 0 1 0 0 0 0 σ0 + σ3 σ1 + iσ2 (11. σµ in which the contravariant indices are moved ‘inside’. account for the 4-vector µ. the submatrices correspond to tαβ spinors.132) and (11. The ˙ desired change of representation(σµ )αβ −→ (σµ )αβ corresponds to a transformation of the basis of ˙ spin states f g f −→ = (11. in fact.110) to express σ αβ in terms 0 −1 1 0 0 0 0 1 (11.. account for the 4–vector index µ. = σαβ ˙ ˙ ˙ ˙ ˙ (σ21 )11 (σ21 )12 (σ22 )11 (σ22 )12 ˙ ˙ ˙ ˙ ˙ ˙ ˙ ˙ 21 (σ )22 21 (σ )22 (σ21 ) (σ22 ) ˙ ˙ ˙ ˙ 21 22 Equating this with the r. i.141) σ αβ = σ0 − σ3 σ1 + iσ2 −σ1 − iσ2 σ0 + σ3 = 2 0 0 0 1 0 −1 0 0 . We can summarize that σ µ and σµ transform like a 4–vector.114).138) Note that all elements of σαβ are real and that there are only four non-vanishing elements.136) results in the succinct expression 1 2 ˙ γδ ˙ (11. µ = 0.135) This is the property surmised already above [cf.370 Inverting contravariant and covariant indices one can also state Lν µ σµ = σν . however. of (11.140) where we employed the expressions (11.

One can write then the scalar product using (11.115) γ µ Aµ ˜ = 0 ((σ µ )αβ )∗ Aµ ˙ 0 ˙ ∗ 1 γδ ˙ ˙ 2 ((σ )αβ ) Aγ δ (σ µ )αβ Aµ 0 ˙ ˙ αβ γ δ 1 ˙ 2 (σγ δ ) A ˙ = 0 .4: Spinor Tensors 371 and.121) of γ ˜ ˜ ˙ basis |αβ whereas the element of γ µ given by σ µ is in the basis |αβ .142) = This can be expressed 1 2 σ αβ ˙ ˙ γδ = δαγ δβ δ . which in ˙ (11.144) The Dirac Matrices γ µ in spinor notation We want to express now the Dirac matrices γ µ in spinor form. (11. This expression implies that the element of γ µ given by σ µ −1 is in the expression (11. ˙˙ (11.146) Exploiting the property (11. Accordingly.143) Combined with (11. employing then transformation (11.140) to transform each of the four submatrices.11. ˙ ˙ ˙ ˙ (11. (11. ˙˙ (11. we write ˜ ˙ γµ = ˜ 0 (σ µ )αβ µ ) )∗ ((σ αβ 0 ˙ ˙ .137) are in a basis (· · ·)αβ to a basis (· · ·)αβ results in ˙ σ αβ ˙ ˙ γδ = 2 (σ 11 )11 (σ 11 )12 ˙ ˙ ˙ ˙ (σ 11 )21 (σ 11 )22 ˙ ˙ (σ 21 )11 (σ 21 )12 ˙ ˙ ˙ ˙ 21 ) 21 ) (σ 21 (σ 22 ˙ ˙ 1 0 0 1 0 0 0 0 0 0 0 0 1 0 0 1 ˙ ˙ ˙ ˙ (σ 12 )11 (σ 12 )12 ˙ ˙ ˙ ˙ (σ 12 )21 (σ 12 )22 ˙ ˙ (σ 22 )11 (σ 22 )12 ˙ ˙ ˙ ˙ 22 ) 22 ) (σ 21 (σ 22 ˙ ˙ . (11.144) results in the simple relationship γ µ Aµ = ˜ 0 Aα β ˙ Aα β 0 ˙ .147) .145) Let Aµ be a covariant 4–vector. For this purpose we start from the ˜ µ .138) one can conclude that the following property holds 1 2 ˙ γδ σ αβ ˙ = 1 2 σ αβ ˙ ˙ γδ = δαγ δβ δ .

149) (11. (11. ˙ (11.5 Lorentz Invariant Field Equations in Spinor Form Dirac Equation (11.372 Spinor Formulation 11. α ˙ = = m φα m χβ .147) allows us to rewrite the Dirac equation in the chiral representation (10.99) 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 (11.150) .226) ˜ i γ µ ∂µ Ψ(xµ ) = 0 i ∂ αβ ˙ i ∂ αβ 0 ˙ ˜ ˜ Ψ(xµ ) = m Ψ(xµ ) .

In the ﬁnal section.2) . [H. Sk ] = 0. 12.1) Then the generators.e. In the following section we will introduce the concept of isospin as an approximate SU (2) symmetry. we will discuss the SU (3) symmetry of three quark ﬂavors. 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. Let us assume a continuous symmetry obeyed by a quantum mechanical system. i. 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. The action of the symmetry operations on quantum mechanical states are given by elements of a corresponding Lie group. O = exp k αk Sk . 373 (12. will commute with the Hamiltonian of the system.. (12. Sk .1 Symmetry and Degeneracies The degeneracies of energy levels of a quantum mechanical system are related to its symmetries.Chapter 12 Symmetries in Physics: Isospin and the Eightfold Way by Melih Sener and Klaus Schulten Symmetries and their consequences are central to physics. We will also introduce the quark model as a natural framework to represent the observed symmetries. We will apply these concepts to an analysis of nucleon-nucleon and nucleon-meson scattering. The organization of this chapter is as follows: In the next section we will discuss the relation between symmetries of a quantum mechanical system and the degeneracies between its energy levels. We will particularly use the example of spherically symmetric potentials. which identiﬁes the proton and the neutron as diﬀerent states of the same particle. the so-called Eightfold Way. These are the isospin symmetry of nuclear interactions and its natural extension.

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

which is valid even when a and b do not commute. Therefore.5: Lorentz Invariant Field Equations 375 The vector points along the symmetry axis of the elliptical orbit and its length equals the eccentricity of the orbit. (12. Jj ] = i [Ki . 2E (12.16) Through some algebra [4] the following commutation relations can be veriﬁed [Ki .18) which complement the familiar angular momentum algebra of section 5. In order to understand the aforementioned extra degeneracy. For this purpose we ﬁrst note that J · = 0. 2 1 B = (J − K).11. 2 which can be shown to satisfy [Ai .13) It can be veriﬁed explicitly that its components commute with the Hamiltonian. we will compute the hydrogen spectrum using the additional symmetry.9) (12. (12.20) (12.8) is not a hermitian operator. Now we scale the eccentricity vector as follows K= − m . which have a negative energy E. in the subspace of the Hilbert space corresponding to a certain energy we can replace H by E. We will also need the following identity [4] 2 = 2H J2 + m 2 + k2 .15) which can be proved after very considerable algebra. 2m r (12. 2m r 1 x2 (J1 p3 − J3 p1 + i p2 ) − k . 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 . Kj ] = i ijk Kk . 2m r 1 x3 (J2 p1 − J1 p2 + i p3 ) − k . ijk Jk .3. In the following we consider the bound states. Aj ] = i ijk Ak .21) .17) (12.12) (12. 2m r 1 x1 (J3 p2 − J2 p3 + i p1 ) − k .19) (12.14) This follows from a · (a × b) = (a × b) · a = 0.11) (12. The vector in (12. We introduce the following new operators 1 A = (J + K).10) (12. (12. (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 (12.28). . ijk Bk .26) (12. . .25) So far we have shown that the symmetry generators form an algebra. Noticing that A2 and B 2 have the same eigenvalues because of (12.30) 2 2 2 1 . A. we can read oﬀ the eigenvalues of A2 and B 2 from (12. 1.22) (12. . . This implies that a = b. . namely that J J > < A − B =0 A + B =2 A (12. . . a + b = n − 1}. a = 0.22): A2 = a(a + 1) B 2 = b(b + 1) Following (12.27) (12. Bj ] = i [Ai . Furthermore the bound on the orbital angular momentum. . 1. . In contrast to the discussion above about extra symmetries. Spinor Formulation (12.29) (12. . .31) where we have used (12. . This illustrates the eﬀect of additional symmetries to the degeneracy structure of a quantum mechanical system. 1.23) (12.15). (12. 2 (12.32) A comparison with (12. .7) tells us that (2a + 1) = n. Bj ] = 0.21) and (12. l. . Chaos is described classically as exponential . the energy eigenvalues are found to be E=− mk 2 1 .34) It follows that l has to have values in {0 = |a − b|. The most extreme case of this is the quantum analogue of a classically chaotic system. . = 0. a lack of symmetry implies a lack of degeneracy in the energy levels of a quantum mechanical system. By comparing to the rotation algebra introduced in chapter 5. .376 [Bi . b = 0. Using this equation the energy eigenvalues can be written in terms of the eigenvalues of A2 and B 2 operators.14) we note that A2 − B 2 = J · = 0. can be seen to follow from the triangle inequality as applied to J = A + B. H B. a = 0. . H = 0.28) . .24) (12. which is identical to the the direct sum of the Lie algebra of of two SO(3) (or SU (2)) algebras. 2 (2a + 1)2 2 2 (12. 1. 2 1 .33) (12.

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

the three pions in (12. π 0 = |1. I3 = − 2 2 . If the mass diﬀerence (or the energy diﬀerence) were to be to be purely electrostatic in nature. we refer the reader to [1]. d= 1 1 . and its third component. 2 2 .− 2 2 . including the proton and the neutron. We can describe isospin multiplets the same way we have described the angular momentum and spin multiplets. sec. 0 π − = |1. . the fundamental representation of the isospin symmetry corresponds to the doublet that contains the so-called up and down quarks u= 1 1 . .42) (12. (12. −1 = = = 1 1 1 1 . (mπ± = 139.41) (12. 1 we can re-write (12. the proton had to be heavier. . albeit it is a good approximate one.6M eV /c2 . mπ0 = 135. Further than that. However. In the framework of the quark of model. I.43) Similarly.0M eV /c2 ). respectively. (12. I3 .39) All other isomultiplets.11. 0 . the proton and the neutron can be written as totally symmetric uud and udd states. Isospin symmetry is not an exact symmetry of strong interactions. 2 2 1 2 2 2 1 1 1 1 1 √ . the proton is the lighter of the two. are made up of these two quarks. I3 = 2 2 1 1 . as we shall see below.38) which have all nearly identical masses. −1 . For example. For a precise description of the two nucleons as composite states. 1 .− . Shortly we 1 will see how to describe both the pion and nucleon states as composites of more fundamental I = 2 states. They u form an isotriplet: π + = |1.36) as a multiplet with I = 2 1 1 p = I = .− . including the spin and color quantum numbers of their constituent quarks. as good quantum numbers. π − = |1.378 Spinor Formulation beta decay). 2 (12. If it were otherwise the proton would be unstable by decaying into the neutron.− 2 2 .37) As an example of a multiplet with I = 1 we have the three pions or π-mesons π + = |1. 2. 2 2 1 2 2 2 (12. it can be seen as part of a larger (and more approximate) symmetry which is of great utility to classify observed particle families.− 2 2 1 2 2 2 1 1 1 1 . Therefore it remains a useful concept. Denoting the total isospin. u¯ and d¯ states. 1 π 0 = |1. The mass diﬀerence between the neutron and the proton could then be attributed to the charge content of the latter. spelling disaster for the stability of matter. (12. . n = I = .38) are ud. 2 2 1 1 1 . They can be constructed with the same rules that have been used for angular momentum addition ¯ u in chapter 6.40) + 2 1 1 .

The reader may know that the deuteron. In the late 1940’s and early 1950’s. By convention ¯ baryons have baryon number 1.37) and in analogy to (12.e. muon.) are light. mesons have baryon number 0. the up and down quarks are not the only quark ‘species’ .45) Is the deuteron an isosinglet state or an isotriplet? If it were an isotriplet (|1. whereas mesons are q q states. 2 = |n > |n > . As it later turned out. 0 ) we should have seen nn and pp bound states of comparable energy in nature (because of isospin symmetry). to be −1 for the strange quark.48) (12. The possibilities are that of an isosinglet 1 |0. I3 = 0. baryon and meson. this will be mathematically identical to adding two spins. as some mesons discovered later are heavier than some baryons and so on. The names. The up and down quarks have strangeness zero. We will now try to describe its wave function in terms of its constituent nucleons. 0 |1. 0 = √ (|p > |n > −|n > |p >) 2 and that of an isotriplet |1. It shall be noted here that the quark model was not invented until 1960’s. Therefore the deuteron has to be an isosinglet state (|0. the pions being examples thereof. All antiparticles have their 3 quantum numbers reversed. 1 = √ (|p > |n > +|n > |p >).43). 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. . consists of a proton and a neutron. this remains only an inaccurate naming convention today. we shall setup some terminology: baryons are qqq states. mesons have intermediate mass ranges. a number strange particles have been found which presumably contained a third quark species: the strange quark. Before proceeding further. the neutrinos etc.or ﬂavors as they are commonly called. This is why isospin is such a good symmetry and why isomultiplets have nearly identical masses. but such states do not exist. The value of the strangeness quantum number is taken. such as proton and the neutron.11. baryons generally are heavy. a hydrogen isotope. and quarks have baryon number 1 .46) (12. originally refer to the relative weight of particles. −1 = |p > |p >. If taken literally. 1 |1. Now let us consider another example of combining the isospins of two particles. by accidental convention.47) (12. strongly interacting particles) which is about a GeV /c2 . (12. but at the time the empirical concepts like isospin and strangeness quantum numbers were in use.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. All other composite states have their strangeness given by the sum of the strangeness content of their constituents. (12. Therefore it has to have isospin.44) 2 where B is the baryon number and S is the strangeness. where leptons (electron. The relation between electric charge and isospin are given by the Gell-Mann–Nishijima relation which was ﬁrst derived empirically 1 Q = I3 + (B + S). Naturally. Following (12. 0 ).

(f ) (i) (I (f ) I3 |00I (i) I3 ) is a Clebsch-Gordon coeﬃcient and I (f ) .5 for the spin-analogue of the same problem. I3 .50) . γ (f ) ||T0 ||I (i) . . The dot product here refers to the abstract isospin space.55) .52) where γ (i) and γ (f ) denote degrees of freedom other than isospin. consider (I) p + p → d + π + (II) p + n → d + π 0 (12. respectively. Consider the generalization of tensor operators discussed in section 6.259) discussed in detail in sections 6. Show that I(1) ·I(2) is an ‘isotensor’ of rank zero. We will now employ the (isospin analogue) of the Wigner-Eckart theorem (6. 1 4 in an isotriplet state and − 3 in an isosinglet 4 Using (12. γ (i) . γ (i) I (f ) . Exercise. Now let us re-write more carefully the scattering amplitudes for the two processes in the light of what we have just learned MI = I (f ) = 1. 1 and (1/ 2)(|1. 0 + |0. (12.45) and (12. γ (f ) (f ) (i) (12. We will eventually be able to compute ratios of scattering cross-sections between diﬀerent processes.8. I3 (f ) = 1.7 to the case of isospin. where M is the scattering amplitude given by M = ﬁnal | α I(1) · I(2) | initial . Show that the expectation of I(1) · I(2) is state. γ (i) = (f ) (i) (I (f ) I3 |00I (i) I3 ) (f ) (i) (−1)I −I (f ) (i) (12.38) and the fact that the deuteron is an isosinglet we know that the isospins of the ﬁnal states in (I) and (II) are |1.49) The only assumption that we put in will be that the interaction is of the form V = α I(1) · I(2) . 0 . γ (f ) |T00 |I (i) I3 . γ (i) (i) (12. (12. Refer to exercise 6. as discussed in the exercise above.51) (12. For completeness let us start by restating the Wigner-Eckart theorem (6. γ (f ) T00 I (i) = 1.7. √ According to (12. The initial and ﬁnal states can be denoted in more detail as | initial | ﬁnal = = I (i) .48) the initial states in (I) and (II) have isospin values |1. 1 and |1. I3 = 1.8 to compute the ratio of the scattering amplitudes MI and MII . γ (f ) ||T ||I (i) . σ. is proportional to |M|2 . γ (i) is a reduced matrix element deﬁned in the same sense as in section 6.7 and 6. which is an isoscalar. 0 ).380 Spinor Formulation As an exercise on the implications of isospin symmetry we will consider nucleon-nucleon scattering. For example. The cross-section.259) in the present context: I (f ) I3 . Exercise.54) Here T00 ≡ V = α I(1) · I(2) . such as the spatial dependence of the wave function and spin. I3 .53) 0 √ 1 2I (i) +1 I (f ) .

γ (f ) ||T00 || I (i) = 1. γ (i) 3 1 (f ) (i) I (f ) = 1. Once again we need the isospins for the initial and ﬁnal states. . (12. −1 π0 + n : |1. −1 2 3 2 − + . I3 = 0. 2 . which is also manifested by a vanishing Clebsch-Gordon prefactor. −2 . 2 = 3 2. γ (i) = √ 2 1 (f ) (i) +√ I (f ) = 1. The relevant Clebsch-Gordon coeﬃcients are easily evaluated: (11|0011) = (10|0010) = 1. = 381 (12. but these are not dominant at relatively low energies. γ (f ) T00 I (i) = 0.58) (12. −2 1 1 2.62) We can now write the ratio of the scattering amplitudes: MI 1 √ . 1 1 1 1 3 √ 2. σ( π − + p → anything ) The possibilities are (a) π + + p → π + + p . γ (i) 2 1 = (10|0010) √ I (f ) = 1.61) MII Note that the second term in MII vanishes due to the isospin conservation. which are obtained by a standard Clebsch-Gordan expansion π + + p : |1.56) (12. γ (f ) ||T00 || I (i) = 1. 0 1 1 3 3 2. I3 = 0.60) (12. involving. It follows σI = 2.57) (12. (b) π − + p → π − + p . = MII (1/ 2) (12. we will consider pion-nucleon scattering.66) = 2 3 1 3 2. (12.64) There are more exotic possibilities. 1 π − + p : |1. σII which is in approximate agreement with the observed ratio. γ (i) 3 +0. [2] As a further example. 2 = 2.11. −2 1 1 √ . (12.65) (12.63) where common dynamical factors (which would not be as easy to compute) have dropped out thanks to the Wigner-Eckart theorem. particles with strangeness.5: Lorentz Invariant Field Equations 1 (11|0011) √ I (f ) = 1.59) (12. We want to compute the ratio of total cross-sections assuming a similar interaction as in the previous example σ( π + + p → anything ) . −2 2 1 1 3 2. (c) π − + p → π 0 + n . I3 = 0. for example. I3 = 0. γ (f ) T00 I (i) = 1.

for example. called strangeness. (12. m 1 2. which leads to the following 2 2 ratios for the cross-sections σa : σb : σc = 9 : 1 : 2. mentioned in the previous section.s = 0 . while weak interactions did not. giving it a higher than usual lifetime. was motivated by the existence of particles that are produced strongly but decay only weakly.67) which are independent of m. Strange partners of the familiar nucleons. has a lifetime which is comparable to that of π + albeit being more than three times heavier. etc. such that S(K + ) = 1. 0 0 1 . m .3 The Eightfold Way and the ﬂavor SU (3) symmetry The discovery of the concept of strangeness. however.70) (12. m V V 3 2. are up to 40% heavier.68) − 2 3 M1 2 Guided by empirical data we will further assume that M 3 >> M 1 .d = 1 . [2] (12. Hence the strangeness changing strong decays of K + (or Σ− ) were forbidden. m 1 2. The classiﬁcation of the newly found particles as members of some higher multiplet structure was less obvious then the case of isospin. It was assumed that strong interactions conserved S (at least approximately). As the total cross-section is the sum of individual processes we obtain σ( π + + p ) σa = =3 σ( π − + p ) σb + σc again in approximate agreement with the observed value.69) 12. For instance. Hence Gell-Mann and independently Nishijima postulated the existence of a separate quantum number. making an identiﬁcation of the underlying symmetry and the multiplet structure less straightforward. it appears an obvious generalization to add another component for an extra quark to the isospin vector space 1 0 0 (12. S(Σ− ) = −1 and S(π) = S(N ) = 0. K + . In the light of the quark model. which can be produced by π − + p → K + + Σ− . S. A computation similar to the previous example of nucleon-nucleon scattering yields (apart from common prefactors) the following amplitudes for the reactions in (12.71) u = 0 .382 Spinor Formulation As in the example of nucleon-meson scattering we deﬁne the relevant matrix elements M3 = 2 M1 = 2 3 2.65) Ma = M 3 2 Mb = 1 M 3 + 2 M 1 3 3 √ 2 Mc = 2 3 M3 2 √ 2 (12. .

The reason is that the other known quarks. the same way nucleons and pions were identiﬁed as representations of the isospin SU (2) symmetry. we will establish some mathematical preliminaries about the group SU (3).11. eiφ . where Nf is the number quark ﬂavors. unitary relation k (12. in fact modern theory of strong interactions is a ‘gauge theory’ of this color group. (12. 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. Given a complex vector. The quarks posses another quantum number. The bottom and top are even heavier.71) into each other. in reference to Lie groups. [2] ) Before discussing the signiﬁcance and the physical implications of the quark model. while preserving the norm. First particle multiplets were identiﬁed as representations of the SU (3) group. (The reader is referred to section 8.72) a∗ ak . called quantum chromodynamics. Because of this additional constraint SU (3) has 8 dimensions. namely charm. history followed the reverse of this path.74) for any two unitary matrices U and V . called color. which again form representations of an SU (3) group. of a. Flavor SU (Nf ) symmetry on the other hand. ak .71). (The ‘bare’ mass of the charm quark is already heavier than the two nucleons. bottom (beauty) and top (truth) are signiﬁcantly heavier than the hadronic energy scale. Then came the question of what the fundamental representation. we want to ﬁnd those transformations ak → Ukl al that preserve the norm. have to be elements of the group SU (3) (which we will investigate closely very soon). Since arbitrary phase factors are of no physical interest. Hence the group U (3) has 9 dimensions.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. as in (12. should correspond to. This is believed to be an exact symmetry of strong interactions. giving rise to the quark model. The unitarity relation imposes 9 constraints on the total of 18 real degrees of freedom of a 3 × 3 complex matrix. As quarks were never directly observed.73) To verify that all such matrices form a group. still leaves the norm invariant. becomes increasingly inaccurate for Nf > 3.1 whenever necessary. Lie algebras and related concepts. Multiplying U by a phase. In this perspective the ﬂavor SU (3) symmetry may appear to be mainly of historical interest. which set the hadronic energy scale. SU (N ). The reader is also invited to revisit section 5. However. we observe that (U V )† = V † U † = V −1 U −1 = (U V )−1 . This group of 3 × 3 unitary matrices is denoted by U (3). It is seen that such a matrix U has to satisfy the following k U † = U −1 . for a period they were considered as useful bookkeeping devices without physical content. but we will stick to N = 3 in the following.5: Lorentz Invariant Field Equations 383 In this case the transformations that ‘rotate’ the components of (12. The reader shall note that most of what is being said trivially generalizes to other unitary groups. However SU (3) symmetry appears in another and much more fundamental context in strong interaction physics. but there are a number of subtle diﬀerences. of three dimensions. (12. it is the group .

λ5 = .78) where fjkl are the antisymmetric structure constants similar to the jkl of SU (2) given in (5.76) where λk are 8 linearly independent matrices forming the basis of the Lie algebra of SU (3). As discussed in section 5. The fundamental relation to be preserved is (12. −2 λ4 λ6 λ8 The generators.77) λ1 . can be written in the form U = eiH where H is a hermitian matrix. (12. 0 0 0 . As in the case of SU (2) higher dimensional representations obeying the same structure can be found. regardless of the dimension of the representation. 3 (12. The Lie algebra structure is given by the commutators of λk [λj . (We shall at times refer to the Lie algebra with the name of the group.79).7. The reader is invited to compare the structure of SU (3) to that of SU (2) discussed in section 5. . λk .32). δjkl . U . the meaning being apparent from the context. . 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 . We can also introduce the constants.) The unit determinant condition requires that all λk are traceless. λk ] = 2ifjkl λl .75) αk λk (12. λ7 = 0 0 .384 Spinor Formulation SU (3) and not U (3) that is of main interest. λ3 = 0 −1 0 . since det(eA ) = etrA .1.80) This is the fundamental or deﬁning representation of SU (3). λ2 = . which can be veriﬁed explicitly as matrix identities. Therefore we will express elements of SU (3) as U = ei k (12. via the anticommutator relation 4 [λj . λk ]+ = δjk + 2δjkl λl . obey the following relation tr(λj λk ) = 2δjk . (12. any unitary matrix.79) (12.

87) the commutation relations between the generators can be expressed in a succinct manner. Derive the Gell-Mann–Nishijima relation (12. [Y.81) In chapter 5.92) (12. T0 } and [Y. [Y. 3 (12. U± ] = U± . T0 ] = 0. 2 (12. Y is called the hypercharge.77) is not the most useful basis in the study of SU (3).86) (12.89) (12.91) [T0 .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 . we have [Y.11. First.71) show that T0 is the isospin operator.83) (12.87) Exercise.5: Lorentz Invariant Field Equations 385 As it turns out the set of generators in (12. (12. 1 [T0 . I3 . U± = F6 ± iF7 . (12. (12. We ﬁrst introduce the F-spin operators 1 Fi = λ i . 2 1 [T0 .85) (12. V± = F4 ± iF5 . The generators of SU (3) can be arranged into a very similar form to that of SU (2).88) which deﬁnes (not uniquely) a maximal set of mutually commuting operators {Y. 2 (12. T± ] = 0. which satisﬁed an ‘eigenvalue equation’ [J0 . Using the convention in (12. namely spherical harmonics. Exercise. 2 Y = √ F8 .44). In the case of SU (2) the identiﬁcation of ‘ladder’ operators {J+ .93) (12.94) .84) (12. starting with the observation that Y = B + S. V± ] = ±V± . T 0 = F3 . these relations have been used to construct the angular momentum spectrum as well as the function space representation of the rotation algebra. J− } proved useful. U± ] = ±U± . T± ] = ±T± .90) (12. J± ] = ± J± . In the basis (12. V± ] = ± V± . (Recall that the strange quark has S = −1 by convention).

U− ] = 0. U± } to this maximal set by ‘eigenvalue equations’ and [T+ . .97) which relate commutators of generators with opposite eigenvalues to the maximal set {Y.105) C2 = jkl djkl λj λk λl In general the number of independent Casimir operators of a Lie group is equal to its rank. 3 [U+ . U+ ] = V+ . The interested reader is instead referred to a very readable account given in chapter 12 of [4]. J 2 ) which commutes with all of the generators. Finally. T− ] = 2T0 . (12. jkl (12. V+ ] = 0. we will use these commutation relations to construct representations of SU (3). V+ ] = 0. [T+ . as the relation between the states in a given representation can be conveniently expressed in terms of ladder operators. V− ] = T− .5. 2 3 Y + T0 = 2V0 . When the basis of a Lie algebra is expressed in such a way to satisfy the form of the eigenvalue relations as given above.96) (12.104) (12. A formal deﬁnition and a detailed discussion of the Cartan-Weyl form is beyond the scope of this chapter. V− ] = 2 (12. However. it is said to be in Cartan-Weyl form. As in the case of J 2 and SU (2).88). This form is essential for easy labeling of the representations of the group. [U+ . V− ] = −U− . V± . T0 }. Another important property of SU (2) is the existence of an operator (namely the total angular momentum. In the case of SU (2) the representations lay on a line on the J0 axis.103) Any remaining commutators follow from hermiticity.99) (12. Thus. Note that. The maximum number of mutually commuting generators of a Lie algebra is called its rank.100) (12. An operator which commutes with all generators of a Lie group is called a Casimir operator. Casimir operators can be used to label irreducible representations of the Lie algebra. the representations will now lie in a T0 − Y -plane. C1 = k λ2 = − k 2i 3 fjkl λj λk λl . The same way the angular momentum ladder operators have been used to construct the representations of SU (2). since there are two mutually commuting generators in SU (3) as given in (12.95) (12.101) (12. [T+ .102) (12. we have [T+ . U− ] = Y − T0 = 2U0 . [V+ . similar to the way it was done in section 5.98) (12. [T+ . [U+ . while SU (3) has rank 2. U0 and V0 are linear combinations of T0 and Y . We can construct two such independent Casimir operators for the group SU (3).386 Spinor Formulation which relate the remaining generators {T± . SU (2) has rank 1.

y ± 1 . 2 (12. For example. U± . y = t3 |t3 .3). especially chapters 7 and 8. Now we will construct explicit representations of SU (3). 1) correspond to the triplets of quarks and antiquarks. we have U0 |t3 . 0) and D(0. 1) as an example. Because of (12. 2 1 t3 ) |t3 . y . The representations of SU (3) are constructed analogous to those of SU (2) by identifying the ‘boundary’ states annihilated by raising (or lowering) operators.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 .110) (12. The details of this procedure is beyond the scope of this chapter. Figure (12. y Y |t3 .antiquark pair. (12. For example the pion octet (or any other meson octet) is therefore realized as states of a quark . All other states of the representation are then constructed by successive application of ladder operators T± . Such a representation is labeled as 1 D(p. As another example for the representations of SU (3). y ± 1 . y : T0 |t3 . V± .1). y .5 how to label the irreducible representations of the angular momentum algebra SU (2) in terms of the value of the total angular momentum. A notation suggestive of the dimensionality of the representation can be used to identify representations. y . For example. y .2) shows the representation D(2. This property can be used to label representations in terms of the values of the Casimir operators. q) and it has a dimensionality of 2 (p + 1)(q + 1)(p + q + 2). 1) representation. (12. = y |t3 . 2 (12. For example. The interested reader is referred to [4].11.88) we can label states by the eigenvalues of T0 and Y operators. respectively. (See Fig. 2 1 = β t3 ± .111) (12.106) (12. .) All other representations can be constructed by combining these two conjugate representations.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. 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. |t3 . y V± |t3 .108) (12. y V0 |t3 . The representations D(1. it was shown in section 5. (12. we have T± |t3 .109) The same way that J± |m is proportional to |m ± 1 in the case of the angular momentum algebra. 2 1 = γ t3 . the pion family forms part of an octet corresponding to the D(1. y .112) The eﬀect of these operators to the states in the y − t3 plane have been outlined in Fig. y . y 1 = α t3 ± . y 3 = ( y− 4 3 = ( y+ 4 1 t3 ) |t3 . y U± |t3 . The representations for hexagons with sides of length p and q in the T0 − Y -plane.

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

to the case of adding two spin triplet states. This way the quark-antiquark states can be represented as follows [3] ⊗ [¯ = [8] ⊕ [1] 3] (12. − . .5: Lorentz Invariant Field Equations 389 D(1.114) Figure 12. 2 3 1 1 − . The octet D(1. 3].115) (12. This expansion can be compared. 1) are denoted by [3] and [¯ respectively. 2 (12. in the case of SU (2).113) The additional singlet state corresponds to the η meson. 2 3 2 0. 0) and D(0. The three quarks in the fundamental representation can now be written as u = d = s = 1 1 .116) (12. 0) or the fundamental representation of ﬂavor SU (3) symmetry. for example.3: D(1. 3 (12. .11. where we would write [3] ⊗ [3] = [1] ⊕ [3] ⊕ [5]. £ (12.117) The Gell-Mann–Nishijima relation can then be succinctly expressed as 1 Q = y + t3 .118) ¡ ¢ ¨ ©§ ¨¥¦¤ " ! . 1) is written as [8] etc.

W. (1999). ¡ ¢ S - S 0 ¤ ¥ S + . Griﬃths.120) (12. D. and spectroscopy. F. Wiley.121) L 0 X - X 0 Figure 12. Wiley. Harter. M. 1987.-F. Martin. M¨ller. 1) representation of SU (3). Li. Schulten. [2] Introduction to elementary particles. North holland. D. W. [6] Gauge theory of elementary particle physics. L. dynamics. T. Oxford. B. 1984. Kosztin. 1984. 1993. 133.-P. References [1] Quarks and leptons: an introductory course in particle physics. 67 (2). J. Eisenberg. [4] Quantum mechanics: symmetries. J.4: The baryon octet as a D(1.390 from which the quark charges follow 2 Qu = . G. L. D. [7] Expansion method for stationary states of quantum billiards. W. 3 £ Spinor Formulation (12. Springer-Verlag u [5] Principles of symmetry. 1989. Greiner. Halzen. Kaufman. p. [3] Microscopic theory of the nucleus. A. Am. K. I. Phys. Greiner.119) (12. 3 1 Qs = − . Wiley. 3 1 Qd = − . Cheng. 1972.

Sign up to vote on this title

UsefulNot useful- Kittel Charles - Introduction to Solid State Physics 8Th Edition - Solution Manualby nawarajbhandari
- Polchinski J. String Theory. Vol. 1. an Introduction to the Bosonic Stringby flipgimble
- Quantum Mechanics Reviewby 005235
- Problems and Solutions Quantum Mechanics - Y. K. Lim, World Scientific (1998, 751p)by imaksum

- Annihilation and Creation Operators.Chapter 8
- 51891292 Quantum Mechanics I
- Advanced Quantum Mechanics
- Relativistic Quantum Mechanics
- Merzbacher Quantum Mechanics
- Cambridge.quantum.mechanics.apr.2009.eBook ELOHiM
- 1222405266 Quantum Mechanics
- 611 Electromagnetic Theory II
- Sakurai - Modern Quantum Mechanics Rev Ed- Solutions Manual
- Applied Quantum Mechanics
- Classical Electrodynamics 3rd Ed J.D. Jackson - Solutions - 214 Pg
- Physics - Quantum Field Theory
- Kittel Charles - Introduction to Solid State Physics 8Th Edition - Solution Manual
- Polchinski J. String Theory. Vol. 1. an Introduction to the Bosonic String
- Quantum Mechanics Review
- Problems and Solutions Quantum Mechanics - Y. K. Lim, World Scientific (1998, 751p)
- Quantum mechanics
- Lectures on Quantum Mechanics - Gordon Baym
- Super Symmetry and Quantum Mechanics
- Peskin Chapter 4
- Relativistic Quantum Mechanics-Bjorken j.d., Drell s.d.
- Quantum Mechanics
- Solution Manual String Theory
- Wick Contractions in Latex
- Quantum Mechanics
- What is String Theory - J. Pol Chin Ski
- Quantum Mechanics Lecture
- [Ashcroft & Mermin]Solid State Physics Solution
- particle physics 3
- Quantum Mechanics
- Notes on Quantum Mechanics