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

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

. · ] in conventional diﬀerential calculus does not correspond to a diﬀerentiated function. .e.5) where L( · . t) (1. t) = L(q(t). ˙ In the following we will often use the notation for velocities and other time derivatives d q(t) = q(t) and dtj = xj . equating these terms with δS[q(t). i. t1 ] where the integrand is a function L(q(t).. j=1 We will now consider a particular class of functionals S[ ] which are expressed through an integral d over the the interval [t0 . e.8) We note then using d dt f (t)g(t) ˙ = f˙(t)g(t) + f (t)g(t) ∂L δqj ∂ qj ˙ − d ∂L dt ∂ qj ˙ δqj . t) of the conﬁguration d vector q(t).6) immediately. ∂f in case of a function of M variables. at the ends of the time interval of the trajectories the variations vanish. It holds M M ∂L t1 d ∂L ∂L δS[q(t). It is also important to appreciate that δS[ · .10) From this follows (1. q(t).6) For a proof we can use conventional diﬀerential calculus since the functional (1. t) + j=1 ∂L ∂L δqj + δ qj ˙ ∂qj ∂ qj ˙ + O( 2 ) (1.e. q(t). t) (1. but rather to a diﬀerential of the function which is simply the df diﬀerentiated function multiplied by the diﬀerential increment of the variable.7) through Taylor expansion and identiﬁcation of terms linear in δqj (t). the velocity vector dt q(t) and time t. q(t) + δ q(t). δq(t)]. We attempt to evaluate t1 S[q(t) + δq(t)] = t0 ˙ ˙ dt L(q(t) + δq(t).g. · ) is a function diﬀerentiable in its three arguments. dt q(t). For this purpose we consider M ˙ ˙ ˙ L(q(t) + δq(t). We will later often assume that only variations of a trajectory q(t) are permitted for which δq(t0 ) = 0 and δq(t1 ) = 0 holds. (1.2 Lagrangian Mechanics Note: δq(t) describes small variations around the trajectory q(t). q(t) + δ q(t). q(t) + δq(t) is a ‘slightly’ diﬀerent trajectory than q(t). · . df = M ∂xj dxj . t0 (1. df = dx dx or.6) is expressed in terms of ‘normal’ functions.9) d ∂L δ qj = ˙ ∂ qj ˙ dt This yields for S[q(t) + δq(t)] t1 M S[q(t)] + t0 dt j=1 d ∂L − ∂qj dt ∂L ∂ qj ˙ t1 M δqj + t0 dt j=1 d dt ∂L δqj ∂ qj ˙ + O( 2 ) (1. δq(t)] = dt − δqj (t) + δqj (t) ∂qj dt ∂ qj ˙ ∂ qj ˙ t0 j=1 j=1 t1 . ˙ Theorem: Let S[q(t)] = t0 dx dt t1 ˙ dt L(q(t). We focus on such functionals because they play a central role in the so-called action integrals of Classical Mechanics. i.

y(x1 ) = y1 . This condition is stated in the following theorem.12) for j = 1. it holds in case δq(t0 ) = δq(t1 ) = 0 that qe (t) is an extremal. . .12) for j = 1. then f (t) ≡ 0 on [t0 . Let us assume that the two points are connected by the path y(x). dx (1. M ) d dt ∂L ∂ qj ˙ − ∂L = 0 ∂qj (1.5) assume extreme values. y(x + dx)) is ds = (dx)2 + ( dy dx)2 = dx dx 1+( dy 2 ) .1. The proof of the above theorem starts from (1.13) dt f (t)h(t) = 0 t0 (1. from (1.17) .14) for any continuous function h(t) ∈ F with h(t0 ) = h(t1 ) = 0. 2. y(x)) to (x + dx. According to the Lemma above follows then (1. . M and δqj (t0 ) = δqj (t1 ) = 0 follows according to (1.6) which reads in the present case M ∂L t1 d ∂L δS[q(t). (1. . . M . Theorem: Euler–Lagrange Condition For the functional deﬁned through (1. (1.11) The extremals qe (t) can be identiﬁed through a condition which provides a suitable diﬀerential equation for this purpose. · ] ≡ 0 and.5). ∂qj dt ∂ qj ˙ t0 j=1 (1. y(x2 ) = y2 . .16) the property δS[qe (t). δq(t)] = dt − δqj (t) . t1 ]. 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 . . On the other side.12) The proof of this theorem is based on the property Lemma: If for a continuous function f(t) f : [t0 .16) This property holds for any δqj with δq(t) ∈ F . . 2. . δq(t)] = 0 for all δq(t) ∈ F . 2. hence.1: Variational Calculus 3 We now consider the question for which functions the functionals of the type (1. the above theorem. y1 ) and (x2 . if and only if it satisﬁes the conditions (j = 1. y2 ). The length of such path can be determined starting from the fact that the incremental length ds in going from point (x. t1 ] ⊂ R → R holds t1 (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).15) We will not provide a proof for this Lemma. .

j = 1.. M (1. This in turn yields y(x) = ax + b. t) is the so-called Lagrangian M ˙ L(q(t). .20) x1 − x2 Exercise 1. .1: Show that the shortest path between two points on a sphere are great circles. The shortest path is an extremal of s[y(x)] which must. ˙2 2 (1. q2 .22) ˙ where L(q(t).18) dy s is a functional of y(x) of the type (1. .19) From this follows y / 1 + (y )2 = const and. t) (1. t1 S[q(t)] = t0 ˙ dt L(q(t). hence. .e. circles whose centers lie at the center of the sphere. 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 .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. 2.5) with L(y(x). . q(t). One obtains y 1 − y2 y(x) = (x − x2 ) + y2 . (1. the so-called action integral. dx ) = 1 + (dy/dx)2 . Using y = dx the condition reads d dx ∂L ∂y = d dx y 1 + (y )2 = 0. (1. dx (1. i. . y = const. qM ) . q(t).1. The constants a and b are readily identiﬁed through the conditons y(x1 ) = y1 and y(x2 ) = y2 .23) Presently we consider only velocity–independent potentials.21) are extremals of the functional.4 The total path length is then given by the integral x1 Lagrangian Mechanics s = x0 dx 1+( dy 2 ) . obey the Euler–Lagrange dy condition. according to the theorems above. Velocity–dependent potentials which describe particles moving in electromagnetic ﬁelds will be considered below. 1. q(t). . . t) = j=1 1 mj qj − U (q1 .

t) describes the charge density present in the ﬁeld and J(r. Particle Moving in an Electromagnetic Field We will now consider the Newtonian equations of motion for a single particle of charge q with a trajectory r(t) = (x1 (t). t).24) ∂qj dt ∂ qj ˙ ∂qj dt which are obviously equivalent to the Newtonian equations of motion. The transformation which leaves the ﬁelds invariant is A (r.1.31) allows one to transform the potentials without aﬀecting the ﬁelds and without aﬀecting the equations of motion (1. t) = V (r. t) dt c where dr = v and where F (r. Equations (1.23). 1.27) (1. t) and B(r.28) (1.29) where ρ(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. t) − K(r. dt The ﬁelds E(r. t) 1∂ K(r. t) describes the charge current density.22.25) of a particle moving in the ﬁeld.30) (1. x2 (t). t) + V (r. 1.28) can be satisﬁed implicitly if one represents the ﬁelds through a scalar potential V (r.25) = 0 = 0 4π J c = 4πρ = (1.33) . t) is the Lorentz force. t) + v × B(r. t) obey the Maxwell equations ×E + ·B ×B − ·E 1 ∂ c ∂t E 1 ∂ c ∂t B (1. t) = A(r.27) and (1. t) = q E(r. F (r.30. The equations of motion for such a particle are d q ˙ (mr) = F (r. t) and a vector potential A(r. These conditions read in the present case ∂L d ∂L ∂U d − = 0 → − − (mj qj ) = 0 ˙ (1. t) as follows B E = ×A 1 ∂A . t) and B(r.31) = − V − Gauge Symmetry of the Electromagnetic Field It is well known that the relationship between ﬁelds and potentials (1. t) c ∂t (1.26) (1. x3 (t)) moving in an electromagnetic ﬁeld described through the electrical and magnetic ﬁeld components E(r.32) (1. t) . respectively. c ∂t (1.

1. Comparing then (1. namely. r. B3 = ∂A2 ∂x1 (1.41) with (1.30).6 Lagrangian of Particle Moving in Electromagnetic Field Lagrangian Mechanics We want to show now that the equation of motion (1. 2 c (1.41) which is identical to the term (1. (1. t) · v .38) ∂L ∂ x1 ˙ − ∂L = 0.37) is ∂L ∂V q = −q + ∂x1 ∂x1 c The ﬁrst term in (1.40.37) 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 . d ∂V q (mv1 ) = F1 = −q + [v × B]1 .35) − ∂A1 ∂x2 [v × B]1 = x2 B3 − x3 B2 = x2 ˙ ˙ ˙ ∂A2 ∂A1 − ∂x1 ∂x2 − x3 ˙ ∂A1 ∂A3 − ∂x3 ∂x1 . t) = 1 q mv 2 − q V (r.37) The results (1.25) follows from the Hamiltonian Principle of Least Action. equivalent.35) is equivalent to the Euler–Lagrange condition d dt The second term in (1.39) ∂A1 ∂A2 ∂A3 x1 + ˙ x2 + ˙ x3 ˙ ∂x1 ∂x1 ∂x1 .34) For this purpose we consider only one component of the equation of motion (1.25).40) (1. if one assumes for a particle the Lagrangian ˙ L(r. 1. e. ∂x1 (1. t) + A(r..38.36) This expression allows us to show that (1.35) shows that the Newtonian equations of motion and the Euler–Lagrange conditions are. in fact.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 (1.36) in the Newtonian equation of motion. dt ∂x1 c We notice using (1. (1. (1.g. .

surprisingly simple fashion in Lagrangian Mechanics.32.32. t) = mv 2 − q V (r. q. For this purpose we consider the transformation of the single particle Lagrangian (1.. t) + A (r. 2 c c dt Inserting (1. i. encountered above in connection with the transformations (1.46) (1.42) This transformation is termed gauge transformation. 1. gauge symmetry and symmetries with respect to spatial transformations. 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. The factor q has been introduced to make this c transformation equivalent to the gauge transformation (1.33) of electromagnetic potentials. q. This term is d ∂K ∂K ∂K ∂K ∂K K(r. We want to demonstrate now that the transformation (1. 1.33) ˙ L (r.33) of electromagnetic potentials. t) · v + K(r. t) − + 2 c ∂t c 1 q = mv 2 − q V (r.43) into (1. t) + A(r. We will consider in the following two symmetries. 1. t) the Lagrangian. dt ∂x1 ∂x2 ∂x3 ∂t ∂t (1. t) + q K(q. They are the subject of the following theorem. 2 c = A(r.42) is. Note that one adds the total time derivative of a function K(r. r.44) Since the condition δq(t0 ) = δq(t1 ) = 0 holds for the variational functions of Lagrangian Mechanics. t) 1 1 ∂K q mv 2 − q V (r. appear in a diﬀerent.44) implies that the gauge transformation amounts to adding a constant term to the action integral.33) of electyromagnetic potentials.32. in fact. t) = L(q. and since these properties allow one often to simplify mathematical descriptions. t) + K ·v (1.1. t) = x1 + ˙ x2 + ˙ x3 + ˙ = ( K) · v + . One can conclude then immediately that any extremal of S [q(t)] is also an extremal of S[q(t)]. q. t) + dt c (1. 1. r. t) · v . t) c t1 t0 = S[q(t)] + (1. q.34) q 1 q d ˙ L (r. t) c t1 t0 t1 t0 ˙ dtL(q. t) .3: Symmetry Properties 7 1. (1.e.32.45) .43) To prove this theorem we determine the action integral corresponding to the transformed Lagrangian t1 S [q(t)] = t0 ˙ dtL (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. t) L (q. a term not aﬀected by the variations allowed. t) = q K(q.45) and reordering terms yields using (1. Eq. The gauge symmetry. equivalent to the gauge transformation (1.

t) by gauge transformed potentials V (r. In this case the transformation can be written in matrix form cos −sin 0 x1 x1 x2 = sin cos 0 x2 (1.47) (1. the transformation (1.8 Lagrangian Mechanics Obviously. ˆ (1.. The corresponding inﬁnitesimal transformation is deﬁned for small r (r. ˆ (1. and for the introduction of inﬁnitesimal transformations which will provide a crucial method for the study of symmetry in Quantum Mechanics. we like to ˆ provide the transformation here anyway for later reference r = r + e. for the connection between invariance properties and constants of motion. 0) = r . In case of a translation transformation in the direction e nothing new is gained.33). 1.49) =0 ∈ R (1. therefore.50) A non-trivial example is furnished by the inﬁnitesimal rotation around axis e ˆ r = r + e×r . A(r. for example. Deﬁnition: Inﬁnitesimal One-Parameter Coordinate Transformations A one-parameter coordinate transformation is decribed through r = r (r.42) corresponds to replacing in the Lagrangian potentials V (r. for such vectors holds a · a = 1.e. The transformations considered are speciﬁed in the following deﬁnition. Such invariance properties are very familiar. but nevertheless instructive way considering rotations around the x3 –axis. t). in the case of central force ﬁelds which are invariant with respect to rotations of coordinates around the center. The transformations we consider are the most simple kind.42) and (1.51) We would like to derive this transformation in a somewhat complicated. i. R(r) = through ∂r ∂ (1. The following description of spatial symmetry is important in two respects.52) x3 0 0 1 x3 . However. r ∈ R . where the origin of is chosen such that r (r. which has an important analogy in Quantum Mechanics. A (r. ) = r + R(r) + O( 2 ) . We consider now invariance properties connected with coordinate transformations. We have proven. r. the reason being that our interest lies in achieving familiarity with the principles (just mentioned above ) of symmetry properties rather than in providing a general tool in the context of Classical Mechanics.32. ) .48) In the following we will denote unit vectors as a. the equivalence of (1. t). t). ˆ ˆ ˆ Examples of Inﬁnitesimal Transformations The beauty of inﬁnitesimal transformations is that they can be stated in a very simple manner.

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

3) 3 Qj j=1 ∂L = e· ˆ ∂ xj ˙ 3 ej mxj = e · mv .51). It is not necessary to investigate the consequences of global.50). 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. is conserved. It was. for which translational invariance holds.e. In this case we have Qj = ej · e. 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. to be expected that this is the constant of motion. . hence. 3 and.59) yields the constant of motion ˆ e (ˆ × r) · mv.59). 2. of course. We will now investigate the consequence of rotational invariance as described according to the inﬁnitesimal transformation (1. 2 We ﬁrst determine the constant of motion in case of invariance with respect to translations as deﬁned in (1. In this case we will use the same notation as in (1. 2.10 Application of Noether’s Theorem Lagrangian Mechanics We consider brieﬂy two examples of invariances with respect to coordinate transformations for the Lagrangian L(r. i. A calculation similar to that in (1. not inﬁnitesimal transformations. ˆ ˙ ˆ j=1 (1.59) We obtain the well known result that in this case the momentum in the direction. except using now Qj = ej · (ˆ × r). j = 1. j = 1. v) = 1 mv 2 − U (r). Noether’s theorem yields ˆ ˆ the constant of motion (qj = xj .

t1 ] . t)|2 Ω (2.1 The Double Slit Experiment Will be supplied at later date 2. 1. (2. 2. is unity. The probability that the particle is detected at space–time point r. 4. At present one may think of f (r) as a sum over δ–functions which represent a multi–slit screen. t) is normalized d3 r |ψ(r. 11 . t is |ψ(r. r.3) where Ω is the space volume in which the particle can exist. t1 ] in which the particle is known to exist. t)|2 = ψ(r. The wave function ψ(r. The particle is described by a wave function ψ(r. We provide now a set of formal rules which state how the probability to observe such a particle at some space–time point r. The probability to detect the particle with a detector of sensitivity f (r) is d3 r f (r) |ψ(r. t)ψ(r. placed into the space at some particular time and with a detector behind each slit. t) ψ : R3 ⊗ R → C. t) where z is the conjugate complex of z. t).Chapter 2 Quantum Mechanical Path Integral 2.4) a condition which enforces that the probability of ﬁnding the particle somewhere in Ω at any particular time t in an interval [t0 . t)|2 = 1 Ω ∀t. 3.2 Axioms for Quantum Mechanical Description of Single Particle ˙ Let us consider a particle which is described by a Lagrangian L(r.1) (2. t is described in Quantum Mechanics. t ∈ [t0 .2) (2.

The following so-called completeness relationship holds for the propagator (t > t [t0 . t|r . t|r0 . t1 ]) Ωd Ωd 3r Ωd 3r Ωd 3r 3 r |ψ(r. t ) = δ(r − r ) Ω (2.8) 7. t )φ(r. t ) ψ(r . at any intermediate time the particle needs only be known at one space point.9) Employing a continuum of intermediate times t ∈ [t0 . t|r0 . namely the point on the classical path at time t . t0 ) = φ(r. t1 ] yields an expression of the form r(tN )=rN φ(r. t0 ) . Since (2. t0 ) Ω This relationship has the following interpretation: Assume that at time t0 a particle is generated by a source at one point r0 in space. t|r . t = )|2 = 1 . . the propagator is unitary. the path integral r(tN )=rN d[r(t)] · · · r(t0 )=r0 (2. t|r . t) = Ω d3 r φ(r. (t > t . t ) ψ(r . 8. t ) ψ(r .e. t0 ) = δ(r − r0 ).e. t ∈ [t0 . t ) φ(r. t ) which requires d3 r φ(r. hence. t ∈ (2. t |r0 . t)|2 = (2. t) is described by a linear map of the type ψ(r. . The deﬁnition will actually assume an inﬁnite number of intermediate times and express the path integral through integrals of the type (2.6) φ(r. t|r . t.7) t.9) for N → ∞. This is diﬀerent from the classical situation where the particle follows a discrete path and. > t1 > t0 is φ(r... (2. t).4) holds for all times. This symbol will be deﬁned further below. tN −1 ) φ(rN −1 . The state of a system at time t.12 Quantum Mechanical Path Integral 5. i. t. The generalization of the completeness property to N − 1 intermediate points t > tN −1 > tN −2 > . t ) t > t . t1 ]) d3 r φ(r. The time evolution of ψ(r.11) which denotes an integral over all paths r(t) with end points r(t0 ) = r0 and r(tN ) = rN . t0 ) = Ωd 3r N −1 Ωd 3r N −2 ··· Ωd 3r 1 φ(r. . t ∈ [t0 . t0 ) = r(t0 )=r0 d[r(t)] Φ[r(t)] .5) 6. (2. tN −1 |rN −2 . t|r . tN −2 ) · · · φ(r1 . i. t| rN −1 . t|r . t|r0 . requires then according to (2. This must hold for any ψ(r . t1 ] (2.10) We have introduced here a new symbol. t1 |r0 . t ) Ωd 3 r |ψ(r. ψ(r0 . described by ψ(r.8) a knowledge of the state at all space points r ∈ Ω at some intermediate time t . t ) φ(r .

One would still expect signiﬁcant descructive interference between contributions of diﬀerent paths since the value calculated is comparable to 2π. r. However. the exponent is a very large imaginary number. The functional Φ[r(t)] in (2. 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. actually often essentially exclusive. such that destructive interference of the contributions of such paths does not occur. namely due to the property of the classical path to be an extremal of the action integral. 2 2 (2.13).12) S[r(t)] = t0 ˙ dt L(r. Only for paths close to the classical path is such interference ruled out.14) ˙ In (2. The constant given in (2. it is comparable to action integrals as they arise for microscopic particles under typical circumstances.12) is then Scl / ≈ 0. The value of S[r(t)] is then Scl = 1 1 m v 2 t = erg s . i. 2. However. a very large number.13) and = 1.2. t) (2. Situations which are well described classically will be distinguished through the property that the classical path gives the dominant. However. However.13) L(r. for microscopic particles like the electron this is by no means the case. in complete distinction from Classical Mechanics.e. t) is the Lagrangian of the classical particle.2: Axioms 9. (2. accordingly.e. In case of a proton with mass m = 1. i.0545 · 10−27 erg s . 2.6725 · 10−24 g moving over a distance of 1 ˚ in a time period of 10−14 s the value of S[r(t)] is A Scl ≈ 10−26 erg s and.14) has the same dimension as the action integral S[r(t)]. Its value is extremely small in comparision with typical values for action integrals of macroscopic particles. To show this we consider the value of the action integral for a particle of mass m = 1 g moving over a distance of 1 cm/s in a time period of 1 s. .. This implies that small variations of the path near the classical path alter the value of the action integral by very little. Scl / ≈ 8. interferences should be much less dramatic than in case of the macroscopic particle..11) is Φ[r(t)] = exp where S[r(t)] is the classical action integral tN 13 i S[r(t)] (2. for the electron many paths contribute and the action integrals for non-classical paths need to be known.12. contribution to the path integral (2. Since this number is multiplied by ‘i’. The situation is very diﬀerent for microscopic particles. r.12. not only for the classical path. 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.15) The exponent of (2.13) are built on action integrals for all possible paths. expressions (2.5 · 1027 .

namely tj+1 − tj = (tN − t0 )/N = N . t1 ). 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 .19) The main idea is that one can replace the path integral now by a multiple integral over x1 . a series of times tN > tN −1 > tN −2 > . as in (2. tN )} . (2. respectively. namely 1 m(xj+1 − xj )2 / 2 and U (xj ). −∞ dxN −1 (xj+1 −xj )2 1 − U (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. The particles have associated with them a Lagrangian L(x.12) φ(xN . For the sake of simplicity we will consider the case of particles moving in one dimension labelled by the position coordinate x. Its value is CN = m 2πi N 2 (2. .4 Propagator for a Free Particle As a ﬁrst example we will evaluate the path integral for a free particle following the algorithm introduced above. .9).20) can be properly taken. In passing from one space–time instance (xj . They can be anywhere in the allowed volume which we will choose to be the interval ] − ∞. provides an avenue to evaluate path integrals.10) and (2. 2. x2 .12. . however.20) Here. tN −1 ). 2 2m N (2. the xj values are not. t) = ˙ 1 mx2 − U (x) .14 Quantum Mechanical Path Integral 2. (xN . at the same time.3 How to Evaluate the Path Integral In this section we will provide an explicit algorithm which deﬁnes the path integral (2. .18) The time instances are ﬁxed. . x.21) N 2.13) and.17) The discretization in time leads to a discretization of the paths x(t) which will be represented through the series of space–time points {(x0 . t0 ) = limN →∞ CN exp i N N −1 j=0 +∞ +∞ +∞ −∞ dx1 −∞ dx2 . . This allows us to write the evolution operator using (2. (2. ˙ 2 (2. etc. (2. tj+1 ) we assume that kinetic energy and potential energy are constant.16) In order to deﬁne the path integral we assume. The spacings between the times tj+1 and tj will all be identical. . (x1 . > t1 > t0 letting N go to inﬁnity later. t0 ). tN |x0 . ∞[. tj ) to the next (xj+1 . . (xN −1 .

it is more suitable to deﬁne new integration variables yj .24) Also we use the fact that the velocity of the classical path xcl = (xN −x0 )/(tn −t0 ) is constant. can be expressed through a path integral with endpoints x(t0 ) = 0.27) The third term can be written in the notation introduced tN 1 dt my 2 = S[x(t)|x(t0 ) = 0. tN dt t0 1 1 (xN − x0 )2 m x2 = m ˙ cl . x(tN ) = 0] .31) We have denoted explicitly that the action integral for a path connecting the space–time points (x0 . Inserting the result into (2. t0 ) = exp S[x(t)] (2. x(tN ) = xN ] 1 (xN − x0 )2 m + 0 + 2 tN − t 0 + S[x(t)|x(t0 ) = 0.23) tN − t0 It is essential for the following to note that. due to (2. which can also be written φ(xN . it holds y(t0 ) = y(tN ) = 0 . t0 ) (2.25) + The condition (2. 2. ˙ 2 t0 (2. namely.e.h.2.. tN ) is to be evaluated. tN |x0 . .24). (2. tN |0. ˙ t0 (2.s.13).30) d[x(t)] exp 2 tN − t0 x(t0 )=0 a result.24) implies for the second term on the r. the origin of which coincides with the classical path of the particle. x(tN ) = 0] .s.29) This expression corresponds to the action integral in (2. tN dt y = y(tN ) − y(t0 ) = 0 .10.28) i. t0 ) and (xN .26) The ﬁrst term on the r. t0 ) = exp 1 im (xN − x0 )2 2 tN − t0 φ(0. using (2.22) where xcl (t) is the classical path which connects the space–time points (x0 . since x(t0 ) = xcl (t0 ) = x0 and x(tN ) = xcl (tN ) = xN . The ˙ 1 S[x(t)|x(t ) = x . t0 ) and (xN .h. tN ). 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 . To see the beniﬁt of such approach we deﬁne a path y(t) as follows x(t) = xcl (t) + y(t) (2. x(t ) = x ] for any path x(t) can then be expressed through action integral 0 0 N N an action integral over the path y(t) relative to the classical path. x(tN ) = 0.23). ˙ ˙ ˙ 2 + y2) = ˙ (2. xN − x0 xcl (t) = x0 + ( t − t0 ) . (2. One obtains S[x(t)|x(t0 ) = x0 . The resulting expression for S[x(t)|x(t0 ) = x0 . 2 2 tN − t0 (2. tN |x0 .12) yields x(tN )=0 im (xN − x0 )2 i φ(xN .4: Propagator for a Free Particle 15 Rather then using the integration variables xj . x(tN ) = xN ] is S[x(t)|x(t0 ) = x0 . = (2. of (2.25) is.

.35) must be determined.. 0 0 0 0 0 0 The following integral +∞ +∞ (2... noting y0 = yN = 0. . .31) for a free particle one needs to evaluate the following path integral φ(0.. 0 0 . . (2.34) dy1 · · · −∞ −∞ dyN −1 exp i N −1 j.32) we split oﬀ the factor 2 mN in the deﬁnition (2. 2 −1 −1 2 (2.. t0 ) = limN →∞ × +∞ +∞ −∞ dy1 · · · −∞ dyN −1 exp i N m 2πi N 2 N × (2.37) det(Ajk ) . real. In order to complete the evaluation of (2. symmetric matrix (bjk ) and det(bjk ) = 0. .k=1 dy1 · · · −∞ −∞ dyN −1 exp i yj bjk yk = (iπ)d det(bjk ) 1 2 .34) of (ajk ) deﬁning a new matrix (Ajk ) through ajk = Using det(ajk ) = m 2 N N −1 m 2 N Ajk . . . . . 0 0 . (2. .33) where ajk are the elements of the following symmetric (N 2 −1 0 −1 2 −1 2 m 0 −1 ajk = . In the appendix we prove +∞ +∞ d j.. . .36) which holds for a d-dimensional.k=1 yj ajk yk (2. 0 0 . .16 Evaluation of the necessary path integral Quantum Mechanical Path Integral To determine the propagator (2. (2.38) ..k=1 yj ajk yk − 1) × (N − 1) matrix . .32) (yj+1 − yj )2 N −1 1 2 j=0 2 m N The exponent E can be written. .. as the quadratic form E = im 2 2 ( 2y1 − y1 y2 − y2 y1 + 2y2 − y2 y3 − y3 y2 2 N 2 2 + 2y3 − · · · − yN −2 yN −1 − yN −1 yN −2 + 2yN −1 ) N −1 = i j. . 2 N . tN |0.

.44) = tN − t0 .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 (2.43) We like to note here for further use below that one might as well employ the ‘artiﬁcial’ starting values D0 = 1. We seek then to evaluate the determinant of the n × n matrix 2 −1 0 . presently N − 1.39) yields φ(0.4: Propagator for a Free Particle a property which follows from det(cB) = cn detB for any n × n matrix B. . 2 −1 0 0 0 −1 2 For this purpose we expand (2. (2. (2. det(Ajk ) (2. let say n. .46) . . . one can readily verify Dn = n + 1 . we obtain φ(0. t|x0 . One can readily verify that this procedure leads to the following recursion equation for the determinants Dn = 2 Dn−1 − Dn−2 . . . Inserting this into (2. .. D3 . .31) and obtain. (2. tN |0. Using D1 = |(2)| = 2 ... 2. t0 ) = m 2πi (tN − t0 ) 1 2 . . Our derivation has provided us with the value det(Ajk ) = N . D1 = 2 and obtain from (2. To solve this three term recursion relationship one needs two starting values. x = xN φ(x.. ..40) . tN |0. t0 ) = m 2πi (t − t0 ) 1 2 exp im (x − x0 )2 2 t − t0 . . n = 1. 0 0 −1 2 −1 . . . t0 ) = limN →∞ m 2πi N 2 17 N 2πi m N N −1 2 1 . . ..41) the same result for D2 .2. tN |0.18) we obtain φ(0. . t0 ) = limN →∞ and with NN (2. . 0 0 0 −1 2 . .40) in terms of subdeterminants along the last column.. 0 0 0 . . variable. deﬁning t = tN .42) m 2πi NN 1 2 (2.. . 0 0 Dn = . . (2.39) In order to determine det(Ajk ) we consider the dimension n of (Ajk ). which follows from (2. .

(2. t|r0 . the wave function of the particle at later times. and describes a single particle since 1 πδ 2 1 2 +∞ dx0 exp −∞ − x2 0 δ2 = 1.46) can be generalized to three dimensions in a rather obvious way. t0 ) = m 2πi (t − t0 ) 3 2 exp im (r − r0 )2 2 t − t0 . according to (2. We will obtain. The result (2. t0 ) = 1 πδ 2 1 4 exp − x2 po 0 +i x 2 2δ (2.48) as the initial state using the propagator (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 . t0 )| = 2 1 πδ 2 1 2 exp − x2 0 δ2 (2. x) + E(x) (2. the associated probability distribution |ψ(x0 .5) allows us to predict the time evolution of any state function ψ(x.46).50) One refers to such states as wave packets. (2. x) = im 2 t x2 1 + i o t mδ 2 − 2xo x − po t m + f (x) (2. We devide the contributions to the exponent Eo (xo . Below we will apply this to a particle at rest and a particle forming a so-called wave packet.47) One-Dimensional Free Particle Described by Wave Packet We assume a particle at time t = to = 0 is described by the wave function ψ(x0 . centered around x0 = 0.18 Quantum Mechanical Path Integral This propagator.10) φ(r.49) is Gaussian of width δ. x) + E(x) in (2.51) as follows Eo (xo .51) For this evaluation we adopt the strategy of combining in the exponential the terms quadratic (∼ x2 ) and linear (∼ x0 ) in the integration variable to a complete square 0 ax2 + 2bx0 = a 0 x0 + b a 2 − b2 a (2.52) and applying (2. We want to apply axiom (2. One obtains then for the propagator (2.5) to (2. thereby. t) of a free particle.247).53) .48) Obviously. We need to evaluate for this purpose the integral ψ(x.

(2. however. z2 = +∞ × i 1−i t mδ 2 (2. making certain. t) = 1 πδ 2 1 4 2 t 1 + i mδ2 . The beginning and the end of such path are the points z1 = −∞ × i 1−i t mδ 2 .x) (2.58) The integrand is an analytical function everywhere in the complex plane and we can alter the integration path.60) is then represented by real ρ values.53) E(x) = x− po mt 19 (2.59) An integration path in the complex x0 –plane along the direction i 1−i t mδ 2 (2.x) im xo 2 t t − mδ 2 x− 1+ x− 1+ po m = −∞ +∞ dx0 exp 1+i t 2 2 t i mδ2 = −∞ im dx0 exp 2 t t 1+i 2 mδ xo − po m t t i mδ2 .55) im xo 2 t 1+i t − mδ 2 x− 1+ po m t t i mδ2 .54) f (x) = This yields Eo (xo .61) . 2 (2. (2. We proceed as follows.52).4: Propagator for a Free Particle im x2 − f (x) .57) and needs to determine the integral +∞ I = −∞ +∞ dx0 eEo (xo . We consider a transformation to a new integration variable ρ deﬁned through i 1−i t mδ 2 ρ = x0 − x− 1+ po m t t i mδ2 . the square in (2.2. x) = One can write then (2. (2. that the new path does not lead to additional contributions to the integral.56) m 2πi t 1 2 +∞ eE(x) −∞ dx0 eEo (xo . 2 t One chooses then f (x) to complete.51) ψ(x. according to (2.

63) which can be readily evaluated.66) yields ψ(x. Quantum Mechanical Path Integral (2.63) holds and. Since the exponent in (2.58) vanishes 0 for Re x0 → ±∞. Hence. accordingly. We can conclude then that (2. One obtains i po i po x x2 2m t E(x) = − 2 + − t t t 2δ (1 + i mδ2 ) 1 + i mδ2 1 + i mδ2 2 (2.65) t 1 − i mδ2 t 1 + i mδ2 . t m(1 + i mδ2 ) (2. (2. (2.57) reads then 1 πδ 2 1 4 2πi t .58) has the end points z1 = −∞ .58) along the path z1 → z1 and along the path z2 → z2 gives only vanishing contributions one can replace (2. z2 = +∞ . t) = Seperating the phase factor 1 t 1 + i mδ2 1 4 1 2 exp [ E(x) ] .68) and.55). t) = t 1 − i mδ2 t 1 + i mδ2 1 4 1 πδ 2 (1 + 2 t2 m2 δ 4 1 4 ) exp [ E(x) ] .54) using (2.67) We need to determine ﬁnally (2.70) . In fact. the paths between z1 → z1 and z2 → z2 have a real part of x0 of ±∞.64) ψ(x.62) If one can show that an integration of (2. I = Equation (2.69) t)2 t 2δ 2 (1 + i mδ2 ) + i po x − i p2 o t 2m (2. (2. one can show that z1 lies at −∞ − i × ∞ and z2 at +∞ + i × ∞.58) by t mδ 2 +∞ I = i 1−i dρ exp − −∞ m 2 t 1+ t mδ 2 2 ρ2 (2.20 whereas the original path in (2. 1+b 1+b ﬁnally E(x) = − (x − po m (2. using a ab = a − .58) has a leading contribution in x0 of −x2 /δ 2 the integrand of (2.

respectively.67) provides the complete expression of the wave function at all times t ψ(x. (2. t) = t 1 − i mδ2 t 1 + i mδ2 1 4 21 1 πδ 2 (1 + 2 t2 ) m2 δ 4 1 4 × po i p2 t o ) + i x − t mδ 2 2m (2.77) .49). t) = exp i po i p2 o x − t m 2m .74) remains invariant in time.71). (2.75) ψ(x.71) × exp − (x − po m t)2 ) 2δ 2 (1 + 2 t2 m2 δ 4 (1 − i .75) i. the spatial dependence of the initial state (2.71. instead of (2.72). (2.2.f. Another interesting observation is that the wave function (2.72). the case of arbitrary to is recovered iby replacing t → to in (2.49).76) From this we conclude that an initial wave function ψ(xo . 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 .73) m δ increases with time. (2.72) for the initial state (2.72) Comparision of Moving Wave Packet with Classical Motion It is revealing to compare the probability distributions (2. The corresponding probability distribution is |ψ(x. This yields. (2.71) conserves the phase factor exp[i(po / )x] of the original wave function (2. t0 ) = exp i po i p2 o xo − to m 2m .4: Propagator for a Free Particle which inserted in (2.72) moves in the direction of the positive x-axis with velocity vo = po /m which identiﬁes po as the momentum of the particle. t) = exp i i p2 po o x − (t − to ) m 2m . The conservation of this factor is particularly striking for the (unnormalized) initial wave function ψ(x0 . However. (2. The center of the distribution (2. The width of the distribution (2. (2.e.48) and that the respective phase factor is related with the velocity of the classical particle and of the center of the distribution (2. t0 ) = exp i po xo m .48) and for the ﬁnal state (2.48)] to = 0.72) 2 t2 δ 1+ 2 4 (2. 2. coinciding at t = 0 with the width of the initial distribution (2. a timedependent phase factor exp[− i (p2 /2m) t] arises which is related to the energy = p2 /2m of a o o particle with momentum po .74) which corresponds to (2. This ‘spreading’ of the wave function is a genuine quantum phenomenon..48) for δ → ∞. We had assumed above [c. In this case holds ψ(x.

xcl . For this purpose we use d xcl y = ˙ ˙ (xcl y) − xcl y ˙ ¨ (2. x.e.81) vanishes2 . 2 . y(t)] which vanishes according to the Hamiltonian principle as discussed in Sect. t0 ) = x(t0 )=x0 d[x(t)] exp i S[x(t)] (2. xcl . x. the spatial as well as the temporal dependence of the wave function remains invariant in this case. 2. One refers to the respective states as stationary states.13) φ(xN . y(t).78) i. ˙ ˙ = (2. t) = L(xcl . 1. t) = exp i Quantum Mechanical Path Integral po i p2 o x − t m 2m . t) ˙ (2. (2. t) = ˙ 1 1 mx2 − c(t)x2 − e(t)x .79) for a quadratic Lagrangian L(x.81) for an arbitrary path x(t) with end points x(t0 ) = x0 and x(tN ) = xN . In order to simplify this task we deﬁne again a new path y(t) x(t) = xcl (t) + y(t) (2. t) + δL ˙ ˙ ˙ ˙ where L(xcl .80) into (2.10. t) ˙ δL 1 1 mx2 − c(t)x2 − e(t)xcl ˙ cl cl 2 2 1 1 = my 2 − c(t)y 2 ˙ 2 2 = mxcl y(t) − c(t)xcl y − e(t)y . xcl + y(t).82) one obtains L(xcl + y. 2. ˙ 2 2 tN (2. Obviously.5 Propagator for a Quadratic Lagrangian x(tN )=xN We will now determine the propagator (2.12. 2.85) We want to show now that the contribution of δL to the action integral (2. t) + L (y.83) (2. y(t). the end points of y(t) are y(t0 ) = 0 Inserting (2.86) 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) ˙ L (y. Such states play a cardinal role in quantum mechanics.22 becomes at t > to ψ(x.84) .80) For this purpose we need to determine the action integral S[x(t)] = t0 dt L(x..82) which describes the deviation from the classical path xcl (t) with end points xcl (t0 ) = x0 and xcl (tN ) = xN . (2. tN |x0 . y(tN ) = 0 .

..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) . x t0 (2. . . .93) . 0 0 −1 2 −1 . . . .s. (2. . 0 0 0 −1 2 . tN |0.91) (yj+1 − yj )2 1 2 2m N where cj = c(tj ).91) through the quadratic N −1 E = i j. t0 ) = y(tN )=0 d[y(t)] exp y(t0 )=0 i tN dt L (y. 0 0 0 c3 .. .. .32). ..s.. t0 ) we will adopt a strategy which is similar to that used for the evaluation of (2. . . . . . tN |x0 . For the evaluation of φ(0. tj = t0 + form j. tN |0.83) the ﬁrst term on the r. 0 0 m ajk = .88) and. Applying the Euler–Lagrange conditions (1. . One can express the exponent E in (2. . 2 N ..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 (2.87) According to (2. . t0 ) = exp S[xcl (t)] φ(0. . hence. . . cN −2 0 0 0 0 0 cN −1 (2. . .89) where ˜ φ(0. One can then express the propagator (2. also the second contribution on the r.h. 0 0 N 0 − . 0 0 0 c2 0 . The discretization scheme adopted above yields in the present case ˜ φ(0. .88) vanishes. t) ˙ t0 . . . vanishes. . tN |0.k=1 yj ajk yk (2.h. . .31) for the free particle propagator. 2 .2.79) i ˜ φ(xN . 2 −1 0 0 0 −1 2 c1 0 0 . . 0 0 0 . tN |0. . of (2. t0 ) (2. . 0 0 0 . y. .80) yields for the classical path m¨cl + c(t) xcl + e(t) = 0 x (2. .. ..24) to the Lagrangian (2. . . . t0 ) = limN →∞ × +∞ +∞ −∞ dy1 · · · −∞ dyN −1 exp N i N N −1 j=0 m 2πi N 2 N × − 1 2 cj 2 yj (2.92) where ajk are the elements of the following (N − 1) × (N − 1) matrix 2 −1 0 . .

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

xo .184) 1 ˜ exp[−iω(n + 1 ) t] φn (x) 2 n! ∞ +∞ = m=0 −∞ dxo Φ(x. to ) exp[−iω(n + 1 ) to ] φn (xo ) . xo . to ).188) identiﬁes the functions ψn (x.172) I = exp 2 mω x e−iωt − e−2iωt − Quantum Mechanical Path Integral mω 2 1 x − iωt 2 2 . (2. to ) we express (2.187) or 1 ˜ exp[−iω(n + 2 ) t] φn (x) +∞ = −∞ ˜ dxo φ(x. t) = exp[−iω(n + 1 ) t] φn (x) as invariants under the 2 action of the propagator φ(x. We want to inspect the consequences of the invariance property (2.172). t) +∞ −∞ dto 1 ˜ exp[−iω(m + 1 ) to ] φm (xo ) 2 m! (2.183) where the expansion coeﬃcients are functions of x.147) in (2. but not of t. t − to ) 1 ˜ exp[−iω(m + 1 ) to ] φm (xo ) 2 m! (2.171.185) Replacing t → t + to yields ∞ n=0 1 ˜ exp[−iω(n + 1 ) (t + to )] φn (x) 2 n! ∞ +∞ = m=0 −∞ dxo Φ(x. Noting that the propagator (2. which also exhibits such invariance.182) Comparision with (2.171) can be expanded in terms of e−inωt . t). . . xo .171) concludes the proof of (2.. 2.e. t|xo . i. t) = n=0 1 ˜ exp[ − iω(n + 1 ) t ] φn (x) 2 n! (2. .. of (2. t|xo .36 Altogether. xo . 1 exp[iω(n + 2 ) to ] · · · .171) ∞ W (x. t − to ) 1 ˜ φn (xo ) n! (2.s.186) Fourier transform. 2. t − to ) = φ(x. t|xo . one obtains for the r. .172) in the form ∞ n=0 (2. n = 1. Accoordingly.172). In contrast to W (x.h. 2 (2.172) is a function of t − to and deﬁning accordingly Φ(x. results in 1 ˜ exp[−iω(n + 1 ) t] φn (x) 2 n! +∞ = −∞ dxo Φ(x.188) ˜ ˜ Equation (2. one can expand (2. We note that the factor exp(2 mω/ xe−iωt − e−2iωt ) in (2.

t) = z 2 e−y where w(y. Actually. t) = exp[−iω(n + 1 ) t] Nn φn (x) . To obtain ˜n (x) we simplify the expansion (2. in a ‘nutshell’ all information on the Hermite polynomials. z) = exp(2yz − z 2 ) .195) zn Hn (y) n! ˜ φn (y) . z) = Hn (y) (2.171). . through the expansion coeﬃcients φn (x) in (2.196) where Hn (y) = ey 2 /2 (2. Expansion (2. t)|2 = ˜n (x)|2 .171) W (x.194). This follows from ∂n w(y.197) are called Hermite polynomials which are polynomials of degree n which will be evaluated below. through expansion (2.194) plays a central role for the Hermite polynomials since it contains.183) reads then w(y.194) = z2 n=0 ∞ 1 ∞ zn ˜ φn (y) n! (2.190) ˜ is obeyed. t) of the quantum mechanical harmonic oscillator ˜ ψn (x.191) (2. We will also argue that the functions ψn (x. 1. z) (2. z) z 2 e−y or w(y. 2 Here Nn are constants which normalize ψn (x. In the following we will characterize the functions φn (x) and determine the normalization constants Nn .2. 2. t) are associated with a time-independent probablity density |ψn (x. t) such that +∞ −∞ 2 dx |ψ(x.183).183).183). . For this purpose we introduce ﬁrst closed expressions for φ the new variables y z and write (2. Expression (2. . t) provide all stationary states of the quantum mechanical harmonic oscillator.197) The expansion coeﬃcients Hn (y) in (2.189) ˜ dx φ2 (x) = 1 n (2. (2. z) = n=0 1 2 /2 1 2 /2 = = mω e−iωt x (2. characterizes the wave functions φn (x).198) ∂z n z=0 . ˜ |φ stationary wave functions ψn (x.193) (2. we have identiﬁed then. according to (2. t)|2 = Nn +∞ −∞ n = 0.192) w(y.5: Propagator for a Quadratic Lagrangian 37 ˜ ˜ the functions ψn (x. The Hermite Polynomials ˜ The function (2. (2.

198) for the Hermite polynomials can be expressed in a more convenient form employing deﬁnition (2. that Hn (y) is a polynomial of degree n.196) results in the so-called Rodrigues formula for the Hermite polynomials Hn (y) = (−1)n ey 2 ∂ n −y2 e . and recursion equations for the eﬃcient evaluation of Hn (y).200) One can deduce from this expression the polynomial character of Hn (y). . 2. 1990) is a useful introduction to this tool as is a chapter in the eminently useful Concrete Mathematics by R. (2.Graham. among other properties.E. ν+µ ν H3 (y) = 8y 3 − 12y.38 Quantum Mechanical Path Integral which is a direct consequence of (2.202) generatingfunctionology by H. 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.. 1989). As will become evident in the present case generating functions provide an extremely elegant access to the special functions of Mathematical Physics3 . We will employ (2. We start from 2 e2yz − z ∞ ν = ν=0 µ=0 ∞ ν 1 ν! 1 ν ν µ ν µ z 2µ (−1)µ (2y)ν−µ z ν−µ (−1)µ (2y)ν−µ z ν+µ = ν=0 µ=0 3 (2.Knuth.Patashnik (AddisonWesley. Inc.194.201) ν−µ µ µ ν−µ Figure 2.e. The diagrams illustrate that a summation over all points in a ν. .200) yields for the ﬁrst Hermite polynomials H0 (y) = 1. For this purpose we expand the generating function (2. Reading..S.L.Wilf (Academic Press.199) Comparision with (2. The diagrams also identify the areas over which the summation is to be carried out. Massachusetts. ∂y n (2. and O.1: Schematic representation of change of summation variables ν and µ to n = ν + µ and m = ν −µ.196) ∂n w(y. ν H2 (y) = 4y 2 − 2.194) in a Taylor series in terms of y p z q and identify the corresponding coeﬃcient cpq with the coeﬃcient of the p–th power of y in Hq (y). closed expressions for Hn (y). D. i. One calls w(y.196). m lattice (right diagram). Boston. . . normalization factors for φ(y). The identity (2. µ lattice (left diagram) corresponds to a summation over only every other point in an n. We want to derive now explicit expressions for the Hermite polynomials. ν+µ (2.196) to ˜ derive. z) the generating function for the Hermite polynomials. H1 (y) = 2y.

m must be restricted such that either both n and m are even or both n and m are odd. it contains only odd powers.e.5: Propagator for a Quadratic Lagrangian and introduce now new summation variables n = ν + µ. m = ν − µ 0 ≤ n < ∞.194) holds w(−y. k! (n − 2k)! = Hn (y) (2. (2. With this restriction in mind one can express (2.206) From this expansion we can identify Hn (y) [n/2] Hn (y) = k=0 (−1)k n! (2y)n−2k . The lattices representing the summation terms are shown in Fig. H3 (y) = 8y 3 − 12y.205) Since (n − m)/2 is an integer we can introduce now the summation variable k = (n − m)/2 . . Hence. 39 (2. hence.209) This property follows also from the generating function. From (2.207) contains only even powers.210) from which one can conclude the property (2. .1.208) which agrees with the expressions in (2.207) This expression yields for the ﬁrst four Hermite polynomials H0 (y) = 1.201). i.204) Since ν.2. (2.209). In case of even n . otherwise. it holds Hn (−y) = (−1)n Hn (y) . 2 µ = n−m . µ expressend in terms of n. k! (n − 2k)! (2. H2 (y) = 4y 2 − 2. 0 ≤ k ≤ [n/2] where [x] denotes the largest integer p. according to 2. 0 ≤ m ≤ n . 2 (2. p ≤ x. According to (2. (2. the sum in (2. m are ν = n+m ..202) e 2yz − z 2 ∞ = n=0 zn n! ≤n m≥0 n! (−1) n−m 2 n−m 2 ! m! (2y)m . H1 (y) = 2y. −z) and.207) one can deduce that Hn (y). is a polynomial of degree 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 . µ are integers the summation over n. 2. .197) ∞ n=0 zn Hn (−y) = n! ∞ n=0 (−z)n Hn (y) = n! ∞ n=0 zn (−1)n Hn (y) n! (2.203) The old summation variables ν. . z) = w(y. for odd n. in fact.

2. as deﬁned through (2. Another relationship is d Hn (y) = 2n Hn−1 (y). z) = 0 (2. . n = 0. (2.217) gives H1 (y) − 2y H0 (y) = 0.216) Combining the sums and collecting terms with identical powers of z ∞ n=1 zn Hn+1 (y) − 2y Hn (y) + 2n Hn−1 (y) n! + H1 (y) − 2yH0 (y) = 0 (2. H2n+1 (0) = 0 . 2.209) since odd functions have a node at the origin. Substituting expansion (2. Hn+1 (y) − 2y Hn (y) + 2n Hn−1 (y) = 0 . dw/dy − 2zw = 0. 2. Starting point of the derivation is the property of w(y.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. dy n = 1.213) which allow one to evaluate Hn (y) from H0 (y) and H1 (y) given by (2. . a factor z reduces the order of Hn by one and introduces also a factor n. n = 1. . namely. The reader may verify that w(y. z). n! (2. . (2. have a node at y = 0. . (2. One can then readily state which diﬀerential equation of w(y.215) ∂z which can be readily veriﬁed using (2. (2. (2. For Hermite polynomials holds.208). z) − (2y − 2z) w(y. H2n (0) = (−1)n Recursion Relationships A useful set of properties for special functions are the so-called recursion relationships. . as given in (2. z) ∂ w(y.233) below. . . 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) .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 . 1. For this purpose one considers w(0.196) into the diﬀerential equation (2. . indeed satisﬁes the latter relationship.194).197) above and given by (2. 2. .211) m! n! m=0 n=0 Comparing terms on both sides of the equation yields (2n)! . . Hn+1 (y) − 2y Hn (y) + 2n Hn−1 (y) = 0.188. z) should be equivalent to the relationship (??).194).214) We want to derive (??) using the generating function. . . n = 1. for example. a property which is consistent with (2.212) n! This implies that stationary states of the harmonic oscillator φ2n+1 (x). 2.40 Quantum Mechanical Path Integral The generating function allows one to determine the values of Hn (y) at y = 0.

194.196) yields ∞ n.247) for a = i one obtains I(y) = and. z) w(y. 2 n = 0.228) .225) We want to derive from the generating function (2.220) Using (2.223) dn 2iy t − t2 2 e = (2 i t)n e2iy t − t n dy results.h. For this purpose we consider the integral +∞ −∞ dy w(y. 2 (2.s. .n =0 +∞ −∞ dy Hn (y) Hn (y) e −y 2 zn z n = n! n ! ∞ n=0 √ zn z n 2n n! π n! n! (2. 2.226) π −∞ 2n z n z n n=0 n! Expressing the l.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 . z ) e−y 2 +∞ = e2 z z = √ ∞ dy e−(y−z−z ) = . in the integral representation of the Hermite polynomials 2n (−i)n ey √ Hn (y) = π Orthonormality Properties 2 (2. 2. of the Rodrigues formula (2.224) +∞ −∞ dt tn e2iy t − t .200) yields Hn (y) = The identity (−1)n y2 √ e π +∞ dt −∞ dn 2iyt − t2 e .s.2. e−y 2 π e−y 2 (2.219) which can be written I(y) = e−y 2 +∞ −∞ dte−(t−iy) = e−y √ 2 2 +∞ −∞ dze−z . (2.h. dy n (2. . 2 (2. 2.194.5: 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 . 1.222) Employing this expression now on the r. 2 √ π e2 z z (2.196) the orthogonality properties of the Hermite polynomials. (2. ﬁnally. . 2 (2.226a). acording to the deﬁnition (2. through a double series over Hermite polynomials using (2.221) 1 = √ π +∞ −∞ dt e2iyt − t .

1. 4.231) √ π (2. One can recognize. in agreement with our above discussions.190)] φn (y) = Nn e−y 2 /2 and for the normalization constants Nn follows from (2.230) (2. 3.229) Hn (y) . in harmony with (??).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 2.232) Hn (y) . 4. that the wave functions are even for n = 0.2. 4 and odd for n = 1. (2.228) +∞ 2 Nn −∞ √ 2 2 2 dy e−y Hn (y) = Nn 2n n! π = 1 1 2n n! 1 2n n! We conclude Nn = and can ﬁnally state the explicit form of the normalized stationary states φn (y) = −y √ e π 2 /2 The stationary states (2. (2. Furthermore. the value of the wave function at y = 0 is positive for n = 0. 3. According to (2.233) . Normalized Stationary States The orthonormality conditions (2. 4 in Fig.197) holds 2 ˜ φn (y) = e−y /2 Hn (y) .233) are presented for n = 0. The normalized states are [c.f. 3. (2. ¡ ¢ ¡ -4 -2 0 2 4 y (2. 2.189. One can also recognize that n is equal to the number of nodes of the wave function. 2. 1. negative for n = 2 and vanishes for n = 1. 3. 2. 2. 2.2: Stationary states φn (y) of the harmonic oscillator for n = 0.228) allow us to construct normalized stationary states of the harmonic oscillator.

Comparision with (2. The Hermite polynomials are so-called orthonogal polynomials since they obey the conditions (2. denoted by Ln (x) and encountered in case of the stationary states of the nonrelativistic [see Sect.235) Completeness of the Hermite Polynomials The Hermite polynomials are the ﬁrst members of a large class of special functions which one encounters in the course of describing stationary quantum states for various potentials and in spaces of diﬀerent dimensions.231) of the wave functions diﬀers from that postulated in (2.189) by the Jacobian dx/dy. Ω (2. i. ??? and eq. w(x) = √ exp(−x2 ).e. 5.. +∞[. w(x) = xα e−x . f.179] holds Ω = [−1.e. i.237) and where In denotes some constants. ???] and relativistic [see Sect. (2.156. 5.236 . In = Γ(n+α+1)/n! where Γ(z) is the so-called Gamma function. g ∈ F . 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.234) The explicit form of the stationary states of the harmonic oscillator in terms of the position variable x is then. of functions which obey dx f 2 (x) w(x) = < ∞ .151. using (2.233) and (2. In case of the associated Laguerre (α) polynomials. (2.237) for Ω = R. denoted by P (x) and introduced in Sect. In case of the Legendre polynomials.. and In = 2n n! π.239) . holds Ω = [0. by dx = dy mω 1 4 . and I = 2/(2 + 1). (10. w(x) = 0 only at a discrete set of points xk ∈ Ω (2. 2.5: Propagator for a Quadratic Lagrangian 43 The normalization condition (2. 5. Other examples of orthogonal polynomials are the Legendre and Jacobi polynomials which arise in solving three-dimensional stationary Schr¨dinger equations. 1].f.10 and eq. 5 below [c.228). The latter are written for polynomials pn (x) in the general form dx pn (x) pm (x) w(x) = In δnm Ω (2. The various orthonogal polynomials diﬀer in the spaces Ω ⊂ R over which they are deﬁned and diﬀer in a weight function w(y) which enter in their orthonogality conditions.459] hydrogen atom. The orthogonal polynomials pn mentioned above have the important property that they form a complete basis in the space F of normalizable functions. (2.2.189) φn (x) = √ 1 2n n! mω π 1 4 e− mωx2 2 Hn ( mω x) . w(x) ≡ 1.238) where the space is endowed with the scalar product (f |g) = Ω dx f (x) g(x) w(x) = < ∞ .150 . (5. 10.228) shows that the orthonogality condition of the Hermite polynomials is in complience with (2.236) where w(x) is a so-called weight function with the property w(x) ≥ 0.

(2.236). Therefore.242) and the existence of a normalizable state ψ(y. t)|2 . aT = a. . t) which is stationary under the action of the harmonic oscillator propagator (2. Ω (2.241) The latter identity follows from (2. a (2.. t) is ∞ ψ(y.247) . i. Appendix: Exponential Integral We want to prove +∞ +∞ I = dy1 . n 2 (2.e.e.e. (2. i. t) = n=0 1 2 dn exp[−iω(n + )t] e−y /2 Hn (y) .147). i. a state for which (2.240) where cn = 1 In dx w(x) f (x) pn (x) . 2.233).e. If one replaces for all f ∈ F: f (x) → w(x) f (x) and. −∞ dyn ei −∞ n j. hence.172) holds. except for a single n = no .m=0 d∗ dm exp[iω(m − n)t] e−y Hn (y)Hm (y) . In case of n = 1 this reads +∞ −∞ dx ei a x = 2 iπ . the 2 stationary state ψ(y. Since the Hermite polynomials form a complete basis for such states we can expand ∞ ψ(y. 2 (2. pn (x) → w(x) pn (x) the scalar product (2.188. ψ(y. the states (2.239) becomes the conventional scalar product of quantum mechanics f |g = Ω dx f (x) g(x) . .233) exhaust all stationary states of the harmonic oscillator.. ∞ n. i. det(a) (2.246) for det(a) = 0 and real.197) it must hold cn (t) = dn exp[−iω(n + 1 )t] and.44 As a result holds for any f ∈ F f (x) = n Quantum Mechanical Path Integral cn pn (x) (2. symmetric a. t) must be identical to one of the stationary states (2..242) Let us assume now the case of a function space governed by the norm (2.244) For the state to be stationary |ψ(x.243) To be consistent with(2. t) = n=0 cn (t) e−y 2 /2 Hn (y) . The only possibility for this to be true is dn = 0.k yj ajk yk = (iπ)n . in particular.245) must be time-independent.

.249) n ajk = ˜ l. . 0 ˜ ˜ S−1 a S = a = . The proof of (2. ˜ (2.254) ajj . ann ˜ where S can be chosen as an orthonormal transformation.246) exploits that for any real.k yj ajk yk n = j.249) The akk are the eigenvalues of a and are real. .k n n yj = ˜ k (S −1 )jk yk . ˜ (2. 0 ˜ 0 a22 . (2. . . yk = k Skj yj . . . This property allows one to simplify the bilinear ˜ form n yj ajk yk by introducing new integration variables j. 2. symmetric matrix exists a similarity transformation such that a11 0 . det(S) a = (det(S))−1 det(a) det(S) = det(a) . (2.e.246) reads then in terms of yj ˜ n n j.251) where. One can conclude det(a) = j=1 n (2..k n m n y Sj ajk Skm ym ˜ ˜ y (S T ) j ajk Skm ym ˜ ˜ j. . .7: Appendix / Exponential Integral 45 which holds for a ∈ C as long as a = 0. . i. . . .255) .k n m = = j.253) as well as det(˜) = det(S−1 aS) = det(S−1 ) det(a) .k yj ajk yk ˜˜ ˜ (2.2.248.250) The bilinear form in (2. (2.. . 0 0 .248) . ST S = 1 1 or S = S−1 . . .252) ˜ For the determinant of a holds det(˜) = a n ajj ˜ j=1 (2.m (S T )jl alm Smk . according to (2.

yn ) ) = det(S) . . .258) +∞ +∞ ∂(y1 .257) Substitution of the integration variables (2.262) (2. .46 We have assumed det(a) = 0. .266) .. . . . .258) where we introduced the Jacobian matrix J = with elements Jjs = According to (2.261) ∂(y1 .250) holds J = S and. n (2. . . ∂(˜1 .. . d˜n e y −∞ = (2. .256) such that none of the eigenvalues of a vanishes.263) (2. .264) I = d˜1 . . yn ) ∂(˜1 .250) allows one to express (2. .250) +∞ +∞ I = d˜1 . . . . Accordingly. . 2. ajj = 0 . . y −∞ d˜n det y −∞ ∂(y1 . det( From (2.260) (2.259) (2.e. . .249) follows 1 = det ST S such that one can conclude det S = ±1 One can right then (2. yn ) y ˜ = ( det S )2 (2. . yn ) y ˜ ∂yj . . . (2.265) k=1 −∞ which leaves us to determine integrals of the type +∞ dx eicx −∞ 2 (2. holds n Quantum Mechanical Path Integral ajj = 0 ˜ j=1 (2. ∂ ys ˜ (2. . yn ) y ˜ ei n k akk yk ˜ ˜2 . . ˜ for j = 1. i.. . . . yn ) ∂(˜1 . hence. . . 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 −∞ .

for x.268) along the path γ = γ1 + γ2 + γ3 + γ4 displayed in Figure 2.268) can be written as the sum of the following path integrals J = J1 + J2 + J3 + J4 .267) by considering the contour integral J = γ dz eicz = 0 2 (2.c > 0 . 2 since eicz is a holomorphic function. We consider ﬁrst the case c > 0 and discuss the case c < 0 further below. 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 .266) to the well-known Gaussian integral +∞ dx e−cx −∞ 2 = π .3.268) vanishes. The contour integral (2. p ∈ IR.257) holds c = 0. the integrand does not exhibit any singularities anywhere in C.e.7: Appendix / Exponential Integral 47 where.271) . according to (2. i.270) dx e−cx 2 2 − 2p √ γ4 : z = ix − p J4 = −p i dx eic(ix−p) .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.. c (2. Jk = γk dz eicz 2 (2.2. One can relate integral (2. (2. The contour intergral (2.

274) (2. Hence.268) for p = +∞.267) one has shown then ∞ dx eicx −∞ 2 = iπ . (2. (2. This follows from the following calculation p p→+∞ lim |J2 or 4 | = p→+∞ lim | 0 p i dx eic(ix+p) | ≤ It holds |e ic(p2 −x2 ) p→+∞ 0 lim |i| dx |eic(p 2 −x2 ) | = 1 since the exponent of e is purely imaginary. c ¡ -p γ1 p Re(z) 2 | |e−2cxp | . p p→+∞ lim |J2 or 4 | ≤ = p→+∞ 0 lim dx |e−2cxp | 1 − e−2cp 2cp = 0.273) p→+∞ lim J2 and J4 do not contribute then to integral (2.272) = 0 .3: Contour path γ in the complex plain. (2. One can state accordingly ∞ J = dx e −∞ icx2 − √ ∞ i dx e−cx −∞ 2 Using 2. We will now show that the two integrals J2 and J4 vanish for p → +∞.275) .48 Quantum Mechanical Path Integral Im(z) ip γ2 ¡ γ4 ¡ γ3 ¡ -i p Figure 2.

265). (iπ)n det(a) (2. It holds n I = k=1 iπ = akk ˜ (iπ)n .278) Noting (2.277) We apply the above results (2.277) to (2.268). n ˜ j=1 ajj (2. c (2.255) this result can be expressed in terms of the matrix a I = which concludes our proof.2. if one chooses the same contour integral as (2. 2.275.279) .7: Appendix / Exponential Integral 49 One can derive the same result for c < 0. but with a path γ that is reﬂected at the real axis.276) and (c < 0) ∞ dx eicx −∞ 2 = −iπ = −|c| iπ . This leads to ∞ icx2 J = dx e −∞ + √ −∞ −i ∞ dx ecx = 0 2 (2.

50 Quantum Mechanical Path Integral .

Generalizing (2.s. x2 . t) . t + |r0 . t + ) = Ω m 2πi 3 2 exp i m (r − r0 )2 − 2 U (r. This yields ψ(r.3) In order to carry out the integration we set r0 = r + s and use s as the new integration variable.20.21). (3. The propagator in (3. t) ψ(r0 . x3 )T .s. For many situations. x3 51 . We will denote the components of s by (x1 . It holds then according to (2. t + |r0 .h. .1 Derivation of the Schr¨dinger Equation o We will consider now the propagation of a wave function ψ(r.2) d 3 r0 m 2πi 3 2 exp i m (r − r0 )2 − 2 U (r.5) ψ(r. t) . t) = From this follows ψ(r. Notable exceptions are non-stationary systems involving time-dependent linear and quadratic Lagrangians. of this equation in terms of powers of and we will demonstrate that the terms of order require that ψ(r.h.20) to R one obtains then for small φ(r. even in x1 .20) to evaluate the propagator. In the limit of very small it is suﬃcient to employ a single discretization interval in (2. 2. t) ψ(r0 . t + ) = Ω d3 r0 φ(r. t) . x2 . t) satisﬁes a partial diﬀerential equation. but by no means all. the Schr¨dinger o o equation provides the simpler avenue towards describing quantum systems than the path ingral formulation of Section 2. (3. t) ψ(r + s. namely the Schr¨dinger equation.Chapter 3 The Schr¨dinger Equation o 3. t + ) = × exp i 2 +∞ +∞ +∞ −∞ dx1 −∞ dx2 −∞ dx3 m x2 + x2 + x2 1 2 3 m 2πi 3 2 × (3. t) by an inﬁnitesimal time step . and the r.4) − U (r.1) We will expand the l.1) can be expressed through the discretization scheme (2. (3. t) .

x2 . (3. of the expansion of ψ(r + s. t) and the kinetic energy term. 1. Since the latter contributes to (3. 1 2 3 2 ∂2 j=1 xj ∂x2 ψ(r. . x2 . O( 7 2 ). only terms of the expansion (3. U (r. consequently. e. t) + .6) make contributions of the order O( 3 2 ) . ∂xj ∂xk (3. t) j 1 12 .. t). It holds In+1 (a) = I1 (a) = i 2a iπ . 2 (3. O( 5 2 ) .4) only for small x2 + x2 + x2 values 1 2 3 we expand 3 ψ(r + s. t) j .11) and. According to (3. a I2 (a) = − 3 4a2 iπ . O( 7 2 ) . i ∂a Starting from (3. t) + ∂xj 2 3 xj xk j.g.k=1 xj xk ∂x2 ∂x2 ψ(r. a supposition which will be examined below.8) (3.8) one can evaluate recursively all integrals In (a). t) j k .5) assuming that only the leading terms contribute.7) shows 1 ∂ In (a).4) one identiﬁes 2 1 = a m = O( ) (3. t) + j=1 xj ∂ 1 ψ(r. n = 0.52 Schr¨dinger Equation o It is important to note that the integration is not over r.4) is even in all three coordinates x1 .10) It is now important to note that in case of integral (3.36) holds I0 (a) = Inspection of (3. Obviously.11) holds 3 m 3 iπ 2 2 × = 1 (3.k=1 ∂2 ψ(r.. The integration involves only the wave function ψ(r +s.7) According to (2..9) iπ a (3. 1 4 3 2 2 ∂4 j. x3 )T .. (3. .6) 3 4 ∂4 j=1 xj ∂x4 ψ(r.12) Here one needs to note that we are actually dealing with a three-fold integral. t) = ψ(r. we need then to evaluate integrals of the type +∞ In (a) = −∞ dx x2n exp i a x2 . t) is a constant with respect to this integration.. Since the kinetic energy contribution in (3. t) . but over s = (x1 . the terms collected in (3..5) which are even separately in all three coordinates yield non-vanishing contributions.13) 2πi a . x3 . a (3. It is then suﬃcient to consider the terms ψ(r.

t) + and exp − i U (r.. This equation is often written in the o form ∂ ˆ i ψ(r. ψ(r. (3. t)ψ(r. this equation is trivially satisﬁed to order O( 0 ). The 2nd order spatial derivatives require that one speciﬁes also properties of the solution on a closed boundary ∂Ω surrounding the volume Ω in which a solution is to be determined. t) (3.2: Boundary Conditions and one can conclude.16) Inserting this into (3. t) + 1 2i 4 m 2 53 ψ(r. t + ) = ψ(r. t) ψ(r. The 1st order time derivative requires that for any solution a single temporal condition needs to be speciﬁed.20) 2m 3. t1 ) = f (r). (3. t + ) = exp − i U (r.14) results in ψ(r.2 Boundary Conditions The time-dependent Schr¨dinger equation is a partial diﬀerential equation. Due to its linear character any linear combination of solutions of the time-dependent Schr¨dinger o equation is also a solution. In order O( ) the equation reads i ∂ ψ(r. (3. t) = ψ(r. 2nd o order in the spatial variables and linear in the solution ψ(r.g. e. (3. t).10). t) + ∂ ∂t ψ(r. Obviously. 1st order in time. t) = H ψ(r.3. t) + O( 2 ) . i. t) (3. using (3. t) + O( 2 ) .14) This expansion in terms of powers of suggests that we also expand ∂ ψ(r. t) . t) = 1 − (3. The following general remarks can be made about the solution. t) + O( 2 ) ..18) This is the celebrated time-dependent Schr¨dinger equation. t) . Usually. a solution is thought for t ≥ t0 and the solution is speciﬁed at the intial time t0 . t) ψ(r. one speciﬁes the so-called initial condition. t) = ∂t 2 − 2 2m + U (r.19) ∂t where 2 2 ˆ H = − + U (r. t) − + i 2 2 i U (r.15) ψ(r. t) + O( 2 ) ∂t i U (r. We will derive brieﬂy the type of boundary conditions encountered.e. ψ(r. As we will discuss in Chapter 5 below the solutions of the Schr¨dinger equation are restricted to particular Hilbert spaces H which are o .17) 2m ψ(r.

g ∈ H is deﬁned as follows f |g Ω = d3 rf ∗ (r)g(r) (3. implies hermitian ˆ f | H |g 1 2 Ω ˆ = g| H |f Ω (3. Calssical Electrodynamics.20). g|r) is called the ˆ concomitant of H and is P (f ∗ .23) ˆ Since H is a diﬀerential operator the expressions (3. for example. hence. 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.22) ˆ where H is deﬁned in (3. diﬀer from each other. The diﬀerence between the integrals is g| H |f = Ω Ω = − g| H |f d3 rf ∗ (r) − Ω 2 2 2m g(r) − Ω d3 rg(r) 2 − 2 2m f ∗ (r) + Ω 2 d3 rf ∗ (r) U (r. t). . t) g(r) − Ω d3 rg(r) U (r. The hermitian property. In (3.28) See. Chapter 1.26) where ∂Ω da · A(r) denotes an integral over the surface ∂Ω of the volume Ω. Interchanging f ∗ (r) and g(r) results in g| H |f Ω = Ω ˆ d3 r g(r) Hf ∗ (r) . The reader is advised to consult a reference text on ‘Linear Algebra’ to follow this argument.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|r) (3. in principle.25) ˆ f | H |g Ω ˆ = g| H |f Ω + ∂Ω da · P (f ∗ . J.27) ˆ We will postulate below that H is an operator in H which represents energy. Jackson.26) the vector–valued function P (f ∗ . New York. D. that H is 2 .21) Ω This leads one to consider the integral f | H |g Ω = Ω ˆ d3 rf ∗ (r) H g(r) (3.54 Schr¨dinger Equation o linear vector spaces of functions f (r) in which a scalar product between two elements f. 1975).23). g|r) = − 2 2m ( f ∗ (r) g(r) − g(r) f ∗ (r) ) (3. (3.. f ∗ (r) d3 rg(r) Ω 2 = − 2m d3 rf ∗ (r) Ω 2 g(r) − f ∗ (r) (3. Since energy is a real ˆ ˆ quantity one needs to require that the eigenvalues of the operator H are real and. however. (John Wiley.22) and (3. 2nd Ed.

g|r) vanish on ∂Ω. Often the closed surface of a volume ∂Ω is the union of disconnected surfaces3 ∂Ωj . (3. we can only allow functions which make the diﬀerential da · P (f ∗ . in this case also all derivatives of f (r) vanish at inﬁnity. The probability to observe the particle anywhere in the subvolume ω ⊂ Ω is p(ω. t). α > 2. The latter property stems from the fact that the boundary condition (3.29) In fact.g.31) |r|→∞ (3. in which case ∂Ω1 is the inner sphere and ∂Ω2 is the outer sphere.30) Note that these boundary conditions are linear in f . 3. In this case one can postulate both conditions (3. and that the density decays rapidly when one moves away from that area. to avoid discontinuities in ψ(r. i.3 Particle Flux and Schr¨dinger Equation o The solution of the Schr¨dinger equation is the wave function ψ(r. Since the total probability of ﬁnding the particle anywhere in space is d3 r|f (r)|2 = 1 (3.e.33) The time derivative of p(ω. However. i. It must hold then for all f ∈ H f (r) = 0 ∀r. (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.31) usually arises when a particle existing in a bound state is described. ∂Ω = ∂Ω1 ∩ ∂Ω2 ∩ ∂Ω3 ∩ . . . t)|2 . t)∂t ψ(r. t) = ω 3 d3 r [ ψ ∗ (r.29.30) each condition holding on an entire surface ∂Ωj . obey (3. i. (3.34) An example is a volume between two concentric spheres. e. t) + ψ(r. 3. like exp(−κr).30) can be postulated. |ψ(r.32). A most common boundary condition is encountered for the volume Ω = R3 in which case one postulates lim f (r) = 0 “natural boundary condition” . r ∈ ∂Ω or da · f (r) = 0 ∀r.e. r ∈ ∂Ω (3..3: Particle Flux 55 and. t)|2 . The observable directly linked to the wave function is the probability to ﬁnd the particle at position r at time t. t) = ω d3 r |ψ(r.. t) on a single connected surface ∂Ωj only either one of the conditions (3. κ > 0 or like r−α .32) the wave function must decay for |r| → ∞ rapidly enough to be square integrable. namely.3. t) is ∂t p(ω. 3. t) which describes the state of o a particle moving in the potential U (r.e..29.. therefore. if f and g satisfy these conditions than also does any linear combination αf + βg. t) ] . t)∂t ψ ∗ (r. . In either case does f (r) and all of its derivatives vanish asymptotically.

41) holds for any volume ω ⊂ Ω one can conclude ∂t ρ(r.35) d3 r . t) = |ψ(r. t) = H ψ ∗ (r.40) One can rewrite then (3.37) d3 r ∂t ρ(r. 3. t) = 0 . t) = ∂ω da · P (ψ ∗ (r. t) ∂t ˆ ˆ ψ ∗ (r. Note that for a normalized wave function the quantity ρ(r. t) − ψ ∗ (r. t) H ψ(r. t) ψ(r.44) ˆ Note that the Hamiltonian H involves only real terms.39) d3 r · A(r) (3. t) = P (ψ ∗ . .18) by any complex number and accordingly one can deﬁne ψ(r. t) = 0. t) + 4 · j(r. One can multiply the solution of (3. We will assume in the remainder of this Section that the solutions discussed are normalized.30) ∂t p(Ω.26. t) + ω · j(r. t) − ψ(r.27) this can be written i ∂t p(ω.37) can be expressed through a volume integral according to da · A(r) = ∂ω ω (3. One refers to such solution as normalized. ψ|r. t) ψ ∗ (r. t) H ψ ∗ (r.19) and its conjugate complex4 −i yields i ∂t p(ω.38) holds. t) = 0 (3. t) (3. t)r. t) ] . t). t) = [ ψ(r. ψ(r. t)|2 = 1 Ω (3. (3. (3. The surface integral (3. t) = ω Schr¨dinger Equation o ∂ ∗ ˆ ψ (r.42) 2mi Since (3. t) .27) one can express this j(r. (3. Using (3. A natural choice for this constant is 1. t) such that d3 r|ψ(r.41) where j(r. Accordingly the probability to observe the particle anywhere in the total volume Ω is constant.36) According to (3.37) If one applies this identity to ω = Ω one obtains according to (3.43) (3. (3.29.56 Using (3. t) . 3. t)|2 is a probability density with units 1/volume.

(2.50) 3 ˜ d3 k φ(k) exp i(k · r − ωt) Ω∞ (3. (3. (3. the initial condition at ψ(r. t) = [2π]− 2 where the dispersion relationship holds ω = k2 .e.e.48) ∂t 2m which describes the motion of free particles. 2 we had demonstrated that a factor exp(ipo xo / ) induces a motion of 1-dimensional wave packets such that po /m corresponds to the initial velocity.. t) is that of density ﬂux. arises solely from the complex factor exp(i k · r). 2 for a free particle [c.47) .49) ˜ Obviously. 2. The generalization to three dimenisons implies then that the factor exp(i k · r) corresponds to an intial velocit k/m and a ﬂux of the same magnitude. and for particles moving in a linear [c.48) that the general solution is of the form ψ(r.41) written in the form ∂t ω d3 r ρ(r. t0 ) = [2π]− 2 Ω∞ 3 ˜ d3 k φ(k) exp i(k · r − ωt0 ) . t) (3. In Sect.18) in Ω∞ = R in the case o U (r.f. This ﬁnding is consistent with the present evaluation of the particle ﬂux: a factor exp(ipo xo / ) gives rise to a ﬂux po /m. (2.167)] potential. t0 ) determines φ(k).105.3.148. t) gives rise to a descrease of the total probability in volume ω due to the disappearence of probability density at the surface ∂ω. This follows directly from an inspection of Eq.49) reads at t = t0 ψ(r. 3. 2. j(r. 2m (3. It is of interest to note from (3. (3.43) that any real wave function ψ(r. (3. Such case arose in Sect. m (3. t) = − ∂ω da · j(r. One can readily show by insertion into (3.4: Free Particle 57 The interpretation of j(r. Note that j(r. t) points in the direction to the outside of volume ω. t) .48. 2.51) .f.71)]..46) i. t) has vanishing ﬂux anywhere.45) Obviously. Equation (3. i. equal to the velocity of the particle. 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) .4 Solution of the Free Particle Schr¨dinger Equation o We want to consider now solutions of the Schr¨dinger equation (3. t) = − ψ(r. for f (r) ∈ R. t) = 0 2 ∂ 2 i ψ(r.f. (2.125)] and in a quadratic [c.

53) where φ(r.5).47). The general form for this propagator involves an expansion in terms of a complete set of eigenfunctions as in (3.52) above ψ(r. t0 ) = 1 2π 3 d3 k exp Ω∞ ik · (r − r0 ) − i 2 k2 2m (t − t0 ) . Ω∞ (3.49).54) to the case of non-vanishing potentials U (r).e.70) derived below for a particle in a box and the harmonic oscillator. respectively. (3.e. In case of the harmonoc oscillator the expansion can be stated in a closed form given in (4.51) to be square integrable it follows according to the properties of the Fourier–transform that ψ(r.. i. t|r0 . i. Comparision with Path Integral Formulation One can write solution (3. that the “natural boundary condition” (3.56) and using (2.54) yields (2.54) valid for this case..114) and (4. hence.49) is square integrable at all subsequent times and. in (2.58 The inverse Fourier transform yields 3 ˜ φ(k) = [2π]− 2 Schr¨dinger Equation o d3 r0 exp(−ik · r0 ) ψ(r0 . (t − t0 ) (3. If one chooses the initial state f (r) deﬁned in (3.31) applies. t|r0 . derive an expression similar to (3. t) as given by (3. 3. Below we will generalize the propagator (3. In fact.47). t0 ) . To show this one needs to note 1 2π = +∞ dk1 exp −∞ ik1 (x − x0 ) − 1 2 i 2 k2 1 2m .49.51. The ensuing solutions are called wave packets. 3. The solution as stated is deﬁned in the inﬁnte space Ω∞ = R .247). t0 ) (3.54) 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 (3.55) 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. We have identiﬁed then with (3. evaluating the integral in (3.81) . t0 ) ψ(r0 .54) an alternative representation of the propagator (2. t) = Ω∞ d3 r0 φ(r.

3.27) vanish on the surface ∂Ω of Ω.29. We have demonstrated then in this case that the Schr¨dinger formulation of Quantum o Mechanics is equivalent to the Feynman path integral formulation. 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 . t|x0 . Accordingly.19.52) to the case that the initial state of a 1-dimensional free particle ψ(x0 . (3. (3. t) is given by (??). (3.62) We will restrict the space of allowed solutions to a volume Ω such that the functions also make the concomitant (3. r ∈ ∂Ω (3.30).60) yields with t0 = 0 ψ(x.49.53. Stationary States We consider now solutions of the time-dependent Schr¨dinger equation (3. The 1-dimensional version of (3. i. 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. the functions obey boundary conditions of the type (3.56)]. t) = −∞ dx0 φ(x. t0 ) = 1 2π +∞ dk exp −∞ ik (x − x0 ) − i 2m (t − t0 ) . t|x0 . 3. (3. t) 2 k2 (3. 3.20) which are of o the form ψ(r. 3. t) = f (t) φ(r) . the boundary conditions are φ(r) = 0 ∀r.57–3. r ∈ ∂Ω or da · φ(r) = 0 ∀r.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 (3.59) which is solved through completion of the square in the exponent [c.60) =???? Combining (3. t0 ) ψ(x0 . 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 (3.63) .e.f.64) (3..4: Free Particle Free Particle at Rest 59 We want to apply solution (3.57) where φ(x. 3.46).55.54) is +∞ ψ(x.

one factor depending only on the time variable and the other only on the space variables are called separable in space and time. however.70) The task of ﬁnding solutions φ(r) which solve (3. (3.60 Schr¨dinger Equation o and aﬀect only the spatial wave function φ(r). We want to demonstrate this property now.62) do exist and we will characterize the two factors of the solution f (t) and φ(r). a common case is Ω = Ω∞ and the ‘natural boundary condition’ (3. .64). in general.65) is constant.63.66) i.69) is called an eigenvalue problem. It follows from the observation that for the solution space considered (3.69) If these two equations can be solved simultaneously a solution of the type (3.62) which consist of two factors.62) we insert it into (3. t)|2 = |f (t)|2 Ω d3 r |φ(r)|2 (3. the solutions o (3.31). At this point we state without proof that. by no means all solutions are of this type. . 2. In fact. for functions required to obey boundary conditions (3.67) (3. i. We denote the corresponding solution by φE (r).42) the ﬂux j(r. g2 (t) = f (t). n = 1.e. t) = Ω Ω d3 r |ψ(r. (3..71) .62) are special solutions of the timedependent Schr¨dinger equation. At this point we will accept that solutions φ(r) of the type (3. t) vanishes on the surface of ∂Ω. The identity (3.e. It is important to realize that the separable solutions (3. often only for a discrete set of values En .. α ∈ R . t) = f (0) exp − i E φE (r) where ˆ H φE (r) = E φE (r) . As pointed out above. and h2 (r) = H φ(r).62) have the particular property that the associated probability distributions are independent of time.19). (3.62) is |ψ(r. i. We will demonstrate that solutions of the type (3. One can conclude that the probability density for the state (3. hence. 3. . t)|2 = |φ(r)|2 .45) that the total probability d3 r ρ(r. is time-independent. One calls such states stationary states. This can hold only if |f (t)| is time-independent. We have then shown that ψ(r.62) exists. (3. solutions exist only for a set of discrete E ˆ values. the eigenvalues of the operator H.27) holds and. We will encounter many such problems in the subsequent Sections.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. if f (t) = eiα . This yields an expression g1 (t) h1 (r) = g2 (t) h2 (r) (3.e.68) ˆ where g1 (t) = i ∂t f (t). It follows then from (3. according to (3. We may note in passing that solutions of the type (3. namely f (t) = f (0) exp − i Et . It turns out that a solution for f (t) can be found for any E. h1 (r) = φ(r). In order to further characterize the solution (3. We must postulate therefore ∂t f (t) ˆ H φ(r) = E f (t) = E φ(r) ..69) exist. for the eigenvalue problems in the conﬁned function space.

ˆ d3 r φ∗ (r) H φE (r) = E Ω Ω ˆ d3 r φE (r) H φ∗ (r) . We start our proof using the property (3. k ∈ R 2 k2 φk (r) = N exp i k · r . o According to (3. a property which is to be expected since the energy is purely kinetic energy which. real.28) for the special case that f and g in (3. As a result we cannot postulate in the present case that wave functions are localized and normalizable. = E. We will wave this assumption as we always need to do later whenever we deal with unbound particles. The solution φE (r) corresponding to the eigenvalue problem posed by (3. (3. E (3. We want to prove now that the eigenvalues E which arise in the eigenvalue problem (3. e.4: Free Particle 61 is a solution of the time-dependent Schr¨dinger equation (3. The corresponding stationary solution ψ(r.74) using exp i k · r = ik exp i k · r .3.76) has kinetic energy 2 k 2 /2m. E (3.72) According to (3. E ∈ R.e. The resulting total energy values E are positive. particles scattered of a potential.28) both represent the state φE (r). m (3. Obviously.76) according to (3.75) 2m One can ascertain this statement by inserting the expression for φk (r) into the eigenvalue problem posed in (3. We will show in Section 5 that E can be interprerted as the total energy of a stationary state. hence. in fact. The ﬂux corresponding to (3.74) is actually best labelled by an index k.43) is j(r. one can interpret then k as the magnitude of the momentum of the particle. 3. of course.19. t) = exp − i 2 k2 2m t exp i k · r (3.20).74) The classical free particle with constant energy E > 0 moves without bounds in the space Ω∞ . t) = |N |2 k .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 . (3.66) E must be real. t) = exp − E t 2 φE (r) .71) this yields E Ω d3 r φ∗ (r) φE (r) = E ∗ E Ω d3 r φE (r) φ∗ (r) . should be positive. Stationary State of a Free Particle We consider now the stationary state of a free particle described by i ψ(r.g.73) from which follows E = E ∗ and. .. i.71) are. − 2 2m φ(r) = E φ(r) .

83) The solutions (3. a] ⊂ R → R.84) . The boundary conditions which according to (3. We will refer to this as a particle in a one-dimensional ‘box’. ..81. Solutions of the Schr¨dinger Equation in F1 o The time–dependent solutions satisfy i ∂t ψ(x. hence. 3. It turns out that this boundary condition can only be satisﬁed for a discrete set of k–values kn . .78) (3.80) need to be satisﬁed are φE (e. therefore. (3. In case of the even solutions (3.62 Schr¨dinger Equation o 3. (o) 2 k2 (e) 2 k2 2m = E (3. We can expect.81. 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. i. We assume.o) (−a) = 0 (3. 2a n = 1.5 Particle in One-Dimensional Box As an example of a situation in which only bound states exist in a quantum system we consider the stationary states of a particle conﬁned to a one-dimensional interval [−a.79) The stationary solutions have the form ψ(x. t) . a] ⊂ R which vanish on the surface ∂Ω1 = {−a. Hence. (3. and so-called odd solutions obeying φ(x) = −φ(−x) φE (x) = A sinkx.o) (a) = 0 and φE (e. (3. two types of solutions. n ∈ N. so-called even solutions obeying φ(x) = φ(−x) φE (x) = A coskx .80) 2m dx2 We note that the box is symmetric with respect to the origin.82) satisfy the diﬀerential equation in (3. a}. t) = − d2 ψ(x. φ(±a) = 0 . 3. 5 . f continuous. 3.81) 2m = E. f ∈ F1 = {f : [−a.e. 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 have to consider only one boundary condition. a] ⊂ R assuming that the potential outside of this interval is inﬁnite. . that the solutions obey this symmetry as well.82) One can readily verify that (3.81) they are kn = nπ . 2m dx2 2 (± ) = } (3.80).82) have the property that either both boundary conditions are satisﬁed or none.

3. x)|2 = 1 n (3.1). 6 . 2a n = 1. (3.o) (a. 6 . according to (3. a n = 2. 5 . are En = 2π2 8ma2 n2 . . 3. 2a n = 2.82) only the kn –values cos(kn a) = cos kn = nπ . . i. . 3. according to the dispersion relationships given in (3.85) (3.86).93) The normalization constants are then An = (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 . (3. (3. (3. ..91) holds. φ(e) (a. . . The wave functions for the ﬁve lowest energies En are presented in Fig. 4.3.87) (Note that.82). n = 1. 4. x) = An sin n nπx .89) nπx . respectively. . n = 2. .92) and for the odd states +a |An |2 −a dx sin2 nπx = |An |2 a = 1 . By counting the number of their nodes one can determine the energy ordering of the wave functions.81) and (3. = 0.94) .84.) The Energy Spectrum and Stationary State Wave Functions The energy values corresponding to the kn –values in (3. (3. . 2. This condition implies for the even states +a |An |2 −a dx cos2 nπx = |An |2 a = 1 .5: Particle in One-Dimensional Box since for such kn 63 nπa nπ = cos 2a 2 In case of the odd solutions (3. (3.90) 2a The wave functions represent stationary states of the particle in a one-dimensional box. 5 . 3 . (3. 4. 6 .86).86) satisfy the boundary condition since for such kn nπa sin(kn a) = sin 2a = sin nπ 2 = 0. 2a 1 .88) where the energies for odd (even) n–values correspond to the even (odd) solutions given in (3. 3.e. . x) = An cos n and φ(o) (a.82). . .81. n is assumed to be even. It is desirable to normalize the wave functions such that +a −a dx |φ(e. . (3. 2a n = 1.

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

it holds φn |φm Ω1 = δn m . 3.98) ∞ f |f 5 Ω1 = n=1 d2 . In case of n = m we have to consider three cases. 2. 2a 2a (3. this integral vanishes. the integrand is odd. (3.e. m both odd. 6 . n = 1. hence.102) holds according to (3.101) +a −a dx cos (n−m)πx − cos (n+m)πx 2a 2a and. the integral vanishes. m both even. . .103) We will show in Section 5 that the property of a scalar product do indeed apply. .99) Since in this case the integrand is a product of an even and of an odd function. x) .96) f |g Ω1 = −a dxf (x) g(x) . the integrals vanish.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.} where φn (a. (i) n. for ∞ f (x) = n=1 dn φn (a. 3. one obtains for n. 4. Obviously. In fact. a] are N.95) 1 a cos nπx 2a sin nπx 2a for n = 1. it holds: f |f Ω1 = 0 → f (x) ≡ 0. 3. n = m. f.e. x) = together with the scalar product5 +a 65 (3. the integral arises φn |φm 1 a Ω1 = 1 a +a nπx −a dx cos 2a cos mπx = 2a (3. . Because of the property (3. In case of n. . . for n = 2. In particular. (3. .98) the elements of B1 must be linearly independent. Similarly.90. i. Hence we need to consider only the ﬁrst two cases. N ≥ 1.97) form an orthonormal basis set. too. x) (3. m odd.98) The latter property is obviously true for n = m.89. . The latter case leads to integrals φn |φm Ω1 = 1 a +a dx cos −a nπx mπx sin . i.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.3. and (iii) the mixed case. (ii) n. n (3.. m even φn |φm 1 a Ω1 = 1 a +a nπx −a dx sin 2a sin mπx = 2a (3.. g ∈ F1 (3. 5 . 3.94) B1 = {φn (a.

66 Schr¨dinger Equation o f (x) ≡ 0 implies f |f Ω1 = 0 which in turn implies dn = 0 since d2 ≥ 0. . x) ∂t cn (t) −a dx0 φn (a. Evaluating the Propagator We can now use the expansion of any initial wave function ψ(x.98) one obtains +a dx0 φm (a.. (3.105) f an cos 2a 2a n=0 n=1 ˜ The functions f corresponding to the functions in the space F1 have zeros at x = ±m a.105) to the interval [−a. t) at times t > t0 . t) = n=1 φn (a. however. a] yields then also an expansion for any element in F1 and B1 is a complete basis for F1 . Demonstration of completeness is a formidable task. For this purpose we expand ∞ ψ(x. n = 0. . n = 1.109) Insertion of (3. We have then shown that any f corresponding to elements of F1 can be expanded in terms of elements in B1 . such demonstration can be based on the theory of Fourier series. 3. .95) is an orthonormal basis. t0 ) = n=1 dn φn (a. −a (3.108) into the Schr¨dinger equation yields o +a ∞ n=1 φn (a. . Restricting the expansion (3. . In the present case. (3.108) where the functions cn (t) are to be determined from the Schr¨dinger equation (3.95) is also complete. . For this purpose we extend the deﬁnition of the elements of F1 to the whole real axis through ˜ f : R → R. . (3.} such that ∞ ∞ nπx nπx ˜ = + an sin (3. i. . 3.107) Inserting this into (3. 2. x) cn (t) −a dx0 φn (a.} and {bn . The functions f are periodic with period 2a.e.95. and bn . i.79) requiring the o initial condition cn (t0 ) = 1 . t0 ) in terms of eigenfunctions φn (a. n = 2. m = 1. . 1. n = 1.106) Using the orthonormality property (3. x) to obtain an expression for ψ(x. This implies that only the trigonometric functions which are elements of B1 enter into the Fourier ˜ series.96). there exist real constants {an . t0 ) = dm . 3. x0 ) ψ(x0 . . . 4. x0 ) ψ(x0 . ( ) = ([ + ] − ) (3. x) cn (t) −a dx0 φn (a. 5 . 2. x) . .e. t0 ) (3. .110) . t0 ) .104) ˜ where [y]a = y mod2a.. in (3. the coeﬃcients an . x0 ) ψ(x0 . It follows that B1 n deﬁned in (3. We like to show ﬁnally that the basis (3. t0 ) = +a ∞ i n=1 − En φn (a.78) can be expressed as a linear combination of the elements of B1 deﬁned in (3. any element of the function space F1 deﬁned in (3. Hence. Accordingly. . x0 ) ψ(x0 .106) and generalizing to t ≥ t0 one can write ∞ +a ψ(x.105) must vanish. they can be expanded in terms of a Fourier series.

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.2 presents the probability distribution of the particle at subsequent times. In case that diﬀerent boundary conditions hold the propagator will be diﬀerent as well. The last frame shows the wave front reaching again the right wall and the onset of new interference. Figure 3.79) which lies in the proper function space (3. t|x0 . (3.. t0 ) = ∞ φn (a. 3.113). The respective initial condition is φ(x. t0 |x0 . a] yields. x) and integrating over [−a. i. One can recognize that the particle moves ﬁrst to the left and that near the right wall of the box interference eﬀects develop. 4 a (3. t|x0 . being reﬂected at the right wall. In the present system which is composed solely of bound states the propagator is given by a sum.108. t) = −a dx0 φ(x. . It is of interest to note that φ(x.5). t0 ) ψ(x0 .78). in (2. cm (t0 ) = 1 .98). t0 ) (3.5: Particle in One-Dimensional Box Multiplying both sides by φm (a. according to (3. x) exp n=1 i − En (t − t0 ) φn (a. t0 ) is also referred to as a Greens function. Often the propagator φ(x. The interference pattern begins to ‘smear out’ ﬁrst. rather than by an integral (3. t0 ) = δ(x − x0 ) as can be readily veriﬁed using (3. Note that the propagator has been evaluated in terms of elements of a particular function space F1 .3.114) the representation of the propagator for a particle in a box with inﬁnite walls. We have identiﬁed then with (3. i ∂t cm (t) = − Em cm (t) . t0 ) itself is a solution of the time-dependent Schr¨dinger o equation (3.109) are cm (t) = exp i − Em (t − t0 ) . 67 (3. x0 ) . σ = a 15 . Example of a Non-Stationary State As an illustration of a non-stationary state we consider a particle in an initial state ψ(x0 . t|x0 . k0 = . t0 ) = 1 2πσ 2 1 4 exp − x2 0 + i k0 x0 2σ 2 .111) The solutions of these equations which satisfy (3. t|x0 . t) for any initial condition ψ(x.113) where φ(x.114) This expression has the same form as postulated in the path integral formulation of Quantum Mechanics introduced above. but the collision with the left wall leads to new interference eﬀects.54) as in the case of the free particle system which does not exhibit any bound states. The particle moves then to the left.112) Equations (3. t0 ).112) determine now ψ(x. the elements of which satisfy the appropriate boundary conditions. This solution can be written +a ψ(x. (3.e.

68 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 = 0.16 m a h2 1 πσ 1 πσ -a a -a a Figure 3.48 m a h2 |Ψ|2 2 t = 0.20 m a h2 1 πσ 1 πσ -a a -a a |Ψ|2 2 t = 1.96 m a h2 |Ψ|2 2 t = 1.72 m a h2 1 πσ 1 πσ -a a -a a |Ψ|2 2 t = 0.2: Stroboscopic views of the probability distribution |ψ(x.00 m a h2 |Ψ|2 2 t = 0. t)|2 for a particle in a box starting in a Gaussian distribution with momentum 15 /a.68 Schr¨dinger Equation o |Ψ|2 2 t = 0.44 m a h2 |Ψ|2 2 t = 1. .92 m a h2 |Ψ|2 2 t = 2.

An important property of this basis and. t) . x3 = ±a3 } . x2 . Placing the origin at the center and aligning the x1 . The description employed a space of functions F1 deﬁned in (3. ˆ = O(x1 ) + O(x2 ) + O(x3 ). x2 = ±a2 } ∪ {(x1 . j = 1. x2 .97) with respect to which the eigenfunctions are orthonormal..117) 2m 2 2 2 ∂1 + ∂2 + ∂3 (3. can be counted.e. hence. x2 . x3 ) . ˆ H = − 2 (3. x1 = ±a1 } ∪ {(x1 .116) and obeys the partial diﬀerential equation ˆ H φE (x1 . φ(j) (±aj ) = 0 . x3 . t) = exp − E t φE (x1 . a3 ] ⊂ R = {(x1 . x3 )T ∈ Ω. ( . 3 (3. x3 ) is an element of the function space F3 deﬁned in (3. We seek then solutions of the time-dependent Schr¨dinger equation o ˆ i ∂t ψ(x. a2 ] ⊗ [−a3 . one can express H 3 φ(x1 . (3.122) . x3 )T ∈ Ω. )T ∈ ∂ } . This property allowed us to evaluate the propagator (3.78). x2 . We have deﬁned in the space F1 a scalar product (3.119) where φE (x1 . The corresponding solutions have the form i ψ(x1 .118) which are stationary states. x2 .116) where Ω is the interior of the box and ∂Ω its surface Ω ∂Ω = [−a1 . x3 –axes with the edges of the box yields spatial boundary conditions which are obeyed by the elements of the function space F3 = {f : Ω → R. x3 ) (3. x3 )T ∈ Ω. x3 . x3 ) = E φE (x1 .e.6 Particle in Three-Dimensional Box We consider now a particle moving in a three-dimensional rectangular box with side lengths 2a1 . x2 . x2 . 3. x2 .6: Particle in Three-Dimensional Box Summary: Particle in One-Dimensional Box 69 We like to summarize our description of the particle in the one-dimensional box from a point of view which will be elaborated further in Section 5. . ) = ∀( . A complete basis of F1 is given by the inﬁnite set B1 (3.3.. 2a3 . i. x3 ) = j=1 φ(j) (xj ) (3. i. t) = H ψ(x1 . 2a2 .114) in terms of which the solutions for all initial conditions can be expressed. (3.120) ˆ Since the Hamiltonian H is a sum of operators O(xj ) each dependent only on a single variable. of the space F1 is that the elements of the basis set can be enumerated by integer numbers. x2 . x2 . x2 . 2. .121) where − 2 2m 2 ∂j φ(j) (xj ) = Ej φ(j) (xj ) . f continuous.95). a1 ] ⊗ [−a2 .

a3 . x2 . This symmetry has been exploited in deriving the stationary states. a2 . . Such degeneracies are always a signature of an underlying symmetry. n3 = 1. . t0 ) = n1 . .122) are given by (3. .n2 . a2 . (3.120) can be written φ(n1 . 3. a3 . (3.n3 ) (a1 . x2 . a1 = a2 = a3 = a then rotation around the x1 .n3 ) (a1 .} ∞ (3.122) with (3.126) Symmetries The three-dimensional box conﬁning a particle allows three symmetry operations that leave the box unchanged. 3. n1 .n3 ) = 2π2 = φn1 (a1 .124) The same considerations as in the one-dimensional case allow one to show that B3 = {φ(n1 . x1 .n2 . . 3. 2.125) is a complete orthonormal basis of F3 and that the propagator for the three-dimensional box is φ(r. . x2 . . a2 .n2 . x3 ) . For example. This additional symmetry is also reﬂected by degeneracies in the energy levels. r0 ) .e. Comparing (3. x3 –axes by π/2 also leaves the system unaltered.70 Schr¨dinger Equation o and E1 + E2 + E3 = E. a3 are identical. If two or all three orthogonal sides of the box have the same length further symmetry operations leave the system unaltered. x1 ) φn2 (a2 . if all three lengths a1 . 2.n3 =1 φ(n1 .n3 ) (a1 .n2 . x3 ) = 1. the solutions of (3. 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. n2 .80) shows that the solutions of (3. n1 .n2 . x1 . (3. . 2. r) exp i − E(n1 n2 n3 ) (t − t0 ) φ(n1 . x2 ) φn3 (a3 .123) 8m n2 n2 n2 1 2 3 + 2 + 2 a2 a2 a3 1 . x2 . i. namely rotation by π around the x1 . x3 ) n1 . x3 –axes. . in the present case . hence.96) and. a3 . n2 . a2 . a3 . a2 . t|r0 .n3 ) (a1 . n 3 and E(n1 .. Actually. n 2 . n3 = 1.n2 .

Hence. r) (3. for example. n2 . a.2.2) (a. a. due to the hermitian character of ˆ H. n3 ) = (3. r) + γ φ(2.1) (a. 3) (1.2) (a. One particular aspect of the degeneracies illustrated in the Table above is worth mentioning. in case of an n–fold degeneracy it is always possible. a.2. a. 5) as shown in the Table above. this linear combination o is not necessarily orthogonal to other degeneraste states. for (n1 . φ(1. However.g. 3.2.127) ˜ ˆ ˜ obeys the stationary Schr¨dinger equation H φ(r) = E122 φ(r). ..1.6: Particle in Three-Dimensional Box 71 ‘accidental’ degeneracies also occur. the identity 32 + 32 + 32 = 12 + 12 + 52 . The origin of this degeneracy is. to construct n orthogonal stationary states6 . However. a.2) (a. 6 This is a result of linear algebra which the reader may ﬁnd in a respective textbook. We consider the degeneracy of the energy E122 which is due to the identity E122 = E212 = E221 . e. however. a. 1. a.3. Any linear combination of wave functions ˜ φ(r) = α φ(1. r). in case of degenerate states one cannot necessarily expect that stationary states are orthogonal. r) + β φ(2. a.

72 Schr¨dinger Equation o .

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

4.1) of the form ψ(x.1 Creation and Annihilation Operators The Hamiltonian operator (4. however.5) ˆ H = 1 1 2 p + m ω 2 x2 ˆ ˆ 2m 2 (4. is the commutation property 1 1 (4. ∈ C∞ . the commutator has a simple form. in which the operators used below. the spectrum of ˆ H in (4.2) can be expressed in terms of the two operators p = ˆ d . The operators act ˆ on the space of functions N1 deﬁned in (4. x] = p ˆ . act. This property states that p and x obey an ˆ ˆ algebra in which the two do not commute.4) be satisﬁed. We will point out explicitly where assumptions are made which built on this restriction. The cardinal property exploited below. The stationary states of the harmonic oscillator have been considered already in Chapter 2 where the corresponding wave functions (2. (4. all stationary states of the harmonic oscillator must be bound states and. If this restriction would not apply and all functions f : R → R would be admitted.3) we restrict the ˆ Hamiltonian H and the operators appearing in the Hamiltonian from the outset to the space of functions N1 = {f : R → R.3) Due to the nature of the harmonic potential which does not allow a particle with ﬁnite energy to move to arbitrary large distances.8) i which holds for any position and momentum operator. In the following algebraic solution of (4. t) = exp(−iEt/ )φE (x. i dx x = x ˆ (4. t) where ˆ H φE (x) = E φE (x) .5). only for a discrete set of E values can the boundary conditions (4. In order [ˆ. (4. beside the representation (4.6) the ﬁrst being a diﬀerential operator and the second a multiplicative operator.235) had been determined.7) of the Hamiltonian. however. In the framework of the Schr¨dinger equation the stationary o states are solutions of (4. It is important to keep in mind this restriction of the space.5) →±∞ where C∞ denotes the set of functions which together with all of their derivatives are continuous.4) Equation (4.7) which is why these operators are of interest to us.3) can be solved for any E ∈ R.3) would be continuous and the eigenfunctions φE (x) would not be normalizable. the natural boundary conditions apply x→±∞ lim φE (x) = 0 . The Hamiltonian H can be expressed in terms of the operators acting on the space (4. lim ( ) = } (4. therefore.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.

ˆ 2 2m ω (4.15) together lead to the commutation property (4.4. p ˆ 2 2m ω 2 (4. 1 i dx i dx i i (4. 4.14.10) holds for a ˆ ˆ [ˆ− . to factor H/ ω.9) From this follows (4. Since a+ and a− are ˆ ˆ + and a− since operators and not scalars we cannot simply expect that the identy (4.7) and (4.15) in a form which will be useful below ˆ H = ˆ H = ω a− a+ − ˆ ˆ ω a+ a− + ˆ ˆ ω 1 1 2 ω 1 . x] p ˆ on such functions must be considered.14) 1 i mω 2 x + ˆ p2 + ˆ [ˆ. a+ ] = a− a+ − a+ a− does not necessarily vanish.11) ˆ The reader may note that we have attempted.18) It is of interest to note that the operators a+ and a− are real diﬀerential operators. 2mω dx (4. 2mω dx a− = ˆ mω x + ˆ 2 d . Our next step is an attempt to factorize the Hamiltonian (4. Before we continue we like to write (4. 1 ω 2 (4. ˆ 2 2m ω a− = ˆ mω i x + √ ˆ p. 1 2 (4.1: Creation and Annihilation Operators 75 to derive (4. in fact.13) (4. the action of [ˆ.11) a− a+ = ˆ ˆ (4. 1 ω 2 (4.8) we recall that all operators act on functions in N1 and. x] f (x) = p ˆ d d x f (x) − x f (x) = f (x) = 1 f (x) .17) d dx We also express a+ and a− directly in terms of the coordinate x and the diﬀerential operator ˆ ˆ a+ = ˆ mω x − 2 d .8) yield a− a+ = ˆ ˆ Similarly one can show a+ a− = ˆ ˆ 1 ˆ 1 H − 1 .7) assuming that the factors are easier to handle than the Hamiltonian in yielding spectrum and eigenstates. One obtains [ˆ.12).8).12) To prove this commutation property we determine using (4. Being guided by the identity for scalar numbers (b − ic)(b + ic) = b2 − i(cb − bc) + c2 = b2 + c2 (4.15) 1 ˆ 1 H + 1 .14) and (4. x] . In fact. hence. a+ ] = 1 . ˆ ˆ . the commutator property (4.16) (4. a ˆ 1 (4.8) a ˆ ˆ ˆ ˆ ˆ implies [ˆ− .10) we deﬁne a+ = ˆ mω i x − √ ˆ p.

16.22) Similarly one can show using again (4. g ∈ N1 +∞ +∞ dx f (x) a+ g(x) = −∞ −∞ dx g(x) a− f (x) . it holds that for a stationary state φE (x) of energy E deﬁned through (4.23) Together.20) ˆ In the following we will determine the spectrum of H and its eigenstates. one obtains using (4.16.21) 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) .12. 4. (4.21) is available. a+ and a− as Ladder Operators ˆ ˆ The operators a+ and a− allow one to generate all stationary states of a harmonic oscillator once ˆ ˆ one such state φE (x) ˆ H φE (x) = E φE (x) (4. 4.23) 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) . ˆ (4. This property of operators is investigated more systematically in Section 5. 4. ˆ The results (4.21) a− φE (x) is ˆ a stationary state of energy E − ω and a+ φE (x) is a stationary state of energy E + ω. The property ˆ ˆ follows directly from (??).22.17. 4. 4. Using (??) we like to state (4. In fact.76 Relationship between a+ and a− Linear Harmonic Oscillator The operators a+ and a− are related to each other by the following property which holds for all ˆ ˆ functions f. 4. The derivation will be based solely on the properties (4. 4.19) in the form1 f |ˆ+ g a Ω∞ = g|ˆ− f a Ω∞ .17.24) − m This property states that the operators in the function space N1 are the hermitian conjugate of each other. ˆ (4. (4. (4.19) This property states that the operators a+ and a− are the adjoints of each other. 4. .17.16.19).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) .

It turns out that both diﬃculties can be resolved together. (4.27) Since E = E one can conclude φE |φE Ω∞ = 0. (4. i..31) (4.11) one can rewrite (4.29) where y = mω x. For f (x) = φE (x) and g(x) = φE (x) in (??) follows (Note that according to (??) the eigenvalue E is real.30) Assuming that φ0 (y) does not vanish anywhere in its domain ] − ∞. E . (4.32) for some constant c0 or (4. (4. Another name is creation (ˆ+ ) and annihilation (ˆ− ) operators since these operators a a create and annihilate vibrational quanta ω. the solution φ0 (x) of (4.6. The property (??) has an important consequence for the stationary states φE (x).29) d 1 d φ0 (y) = = lnφ0 (y) = − y .e. 2 2 (4. Using (4. a state φ0 (x) which obeys the property a− φ0 (x) = 0 ˆ (4. In fact. 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 .33) . however.25) on one side would lead to termination of the sequence E0 + m. be found. in fact.25) can. on ˆ the other side such a state is itself an eigenstate of H as can be shown using (4. φ0 (x) should be an element of the function space N1 deﬁned in (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. m ∈ Z when m is decreased.28) 4.17) ˆ H φ0 (x) = ( ω a+ a− + ˆ ˆ 1 ω 1 ) φ0 (x) = 1 ω φ0 (x) .25) φ0 (y) = 0 (4. in fact.26) Of course.4.5). For obvious reasons one refers to a+ and a− as ladder ˆ ˆ operators. two diﬃculties: (i) one needs to know at least one stationary state to start the construction. +∞[ one can write (4.25) needs to be normalizable in order to represent a bound state of the harmonic oscillator. 4.) ˆ ˆ 0 = φE |HφE Ω∞ − φE |HφE Ω∞ = ( E − E ) φE |φE Ω∞ . E = E . (ii) the construction appears to yield energy eigenvalues without lower bounds when.2 Ground State of the Harmonic Oscillator d + y dy A suitable solution of (4. Let φE (x) and φE (x) be two normalized stationary states corresponding to two diﬀerent energies E. one expects that E = 0 should be a lower bound.

.36) assumes a functional form such that both terms together assume a minimum value.37) above the ground state energy. which for a narrowly localized state yields a large positive value. 2.2) there are two competing contributions to the energy. (4. . 2 n = 0. . mω (4.39) We notice that the ground state wave function φ0 (x) as well as the operators (ˆ+ )n are real.34) dx |φ0 (x)|2 = |c0 |2 −∞ −∞ dx exp − mωx2 2 = |c0 |2 π . and the kinetic energy contribution. . 2 n = 0.38) These states correspond to the energy eigenvalues (4.24) can then be written as follows En = ω(n + 1 ). the so-called excited states. we can now construct the stationary states corresponding to energies (4. We recall that according to (4. would yield a vanishing contribution. . .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. 2 This state of lowest energy is called the ground state of the oscillator. a state conﬁned to the minimum corresponding to the classical state of lowest energy.36) Since the ﬁrst order diﬀerential equation (4.26) the energy value associated with this state is 1 ω. (4. 1. 2. For this purpose we apply the operator a+ to the ˆ ground state (4.. 2..e. choose the normalization constants cn and the functions φn (x) real as well. 4. 1. i. 1. .35) The appropriate ground state is φ0 (x) = mω π 1 4 exp − mωx2 2 . The set of allowed energies of the oscillator according to (4. .e. . it holds ˆ H φn (x) = ω(n + 1 ) φn (x) . The reason is that in the Hamiltonian (4. 2.. i. . therefore. We will consider this point more systematically in Section 5.25) admits only one solution there is only one set of states with energy E + m ω.3 Excited States of the Harmonic Oscillator Having obtained a suitable stationary state with lowest energy. . we construct the states for n = 1. . The ground state (4. the potential energy contribution which for a state δ(x).36) n times. (4. (4.39). Such states need to be suitably normalized for which purpose we introduce a normalization constant cn φn (x) = cn a+ ˆ n φ0 (x) .78 Linear Harmonic Oscillator This solution is obviously normalizable. m ∈ Z which properly terminate at some minimum value E + m0 ω ≥ 0. We a can. i. . The conventional normalization condition φ0 |φ0 reads +∞ +∞ Ω∞ = 1 (4.e.. n = 0.

41) Employing the adjoint property (4.40) = 1.15 . n = 0.43) (4. (4. n = 0.40–4. A properly normalized state φn+1 (x) is then of the form φn+1 (x) = αn a+ φn (x) ˆ for some real constant αn which is chosen to satisfy φn+1 |φn+1 Ω∞ 2 = αn a+ φn |ˆ+ φn ˆ a Ω∞ (4. For n = 0 holds c0 = 1..3: Excited States 79 We need to determine now the normalization constants cn . i.47).e. .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) . (4.48) (4. Since a− is a real diﬀerential operator [see (4.20) 1 = φn−1 |φn−1 Equations (4.51) . To determine βn we note using (4.14) together with H φn (x) = ω(n+ 1 ) φn (x) leads to the condition (note that we assumed 2 φn (x) to be normalized) 2 2 αn ( n + 1 ) φn |ˆ− a+ φn Ω∞ = αn ( n + 1 ) = 1 a ˆ √ From this follows αn = 1/ n + 1 and.45) We like to note that these functions according to (4. . .40) φn−1 (x) = βn a− φn (x) ˆ (4. 1.40).46) According to (4. ˆ Repeated application of this relationship yields φn−s (x) = (n − s)! − a ˆ n! s 2 1 = βn n φn |φn Ω∞ 2 = βn n . . . To determine these constants we adopt a recursion scheme. √ a− φn (x) = n φn−1 (x) . 2.24) holds in analogy to (4.28) and the construction (4. (4.49) (4. βn must be real as well. according to (4. 4. according to (4. (4.18)] and since the ˆ φn (x) are real functions.39) yield √ From this follows βn = 1/ n and. . (4. (4. We consider then the situation that we have obtained a properly normalized state φn (x).4.45) form an orthonormal set.47) for some suitable constants βn .42) ˆ Using (4.50) φn (x) . they obey φn |φn = δn n .20) yields 2 αn φn |ˆ− a+ φn a ˆ Ω∞ = 1. Ω∞ 2 = βn a− φn |ˆ− φn ˆ a Ω∞ 2 = βn φn |ˆ+ a− φn a ˆ Ω∞ . √ a+ φn (x) = n + 1 φn+1 (x) . n. ˆ (4. 2. 1. .

For this purpose we start from expression (4. by introducing the variable y deﬁned in (4. (4.57) which includes the factor 1/ n! according to Eqs.56) Relationship to Hermite Polynomials We want to demonstrate now that the expression (4.35) by a constant. .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) .52).55) 1 y2 2 (4. however. n–independent factor.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 .56) can be expressed in terms of Hermite polynomials Hn (y) introduced in Sect. i.45). (4. by dx = dy mω 1 4 .52) This normalization of the wave functions diﬀers from that postulated in (4.7.35) rather than (4.46) yields a set of normalized states.53) We will later re-introduce this factor to account for the proper normalization (4. (4. for example. We will demonstrate below the identity Hn (y) = e y2 2 y − d dy n e− y2 2 (4. by the Rodrigues formula (2.233) derived in Sect.28) of the ground state and the deﬁnition (4.30) and employing the normalization +∞ −∞ dy φ2 (y) = 1 n (4. 2.39).e. namely the square root of the Jacobian dx/dy. simplifying the calculation.. 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 .7 and given. It should be noted that √ the normalization (4.200). (4. 2. (4.40–4.80 Evaluating the Stationary States Linear Harmonic Oscillator We want to derive now an analytical expression for the stationary state wave functions φn (x) deﬁned through (4.58) This result agrees with the expression (2.

(3.4: Propagator 81 To verify the realtionship between the deﬁnition (4. (4.65) .39) ∞ ψ(y0 .4. hence. dy dy (4. the Rodrigues formula (2. i.61) One can factor g(y) and employ (4.30) and return later to the coordinate x. 4.60) and.. For the sake of notational simplicity we employ initially the coordinate y as deﬁned in (4.64) which implies that (4. We prove (4. (4. dy n (4.59) hold.61) and.200).62) Denoting f (y) = (y − d/dy)exp(−y 2 /2) and employing y2 y2 d d − y2 e 2 f (y) = −y e− 2 f (y) + e− 2 f (y) dy dy (4.2) for an arbitrary initial wave function ψ(x0 .63) one obtains g(y) = y − d dy f (y) (4.106–3. t0 ) = n=0 dn φn (y0 ) .4 Propagator for the Harmonic Oscillator We consider now the solution of the time-dependent Schr¨dinger equation of the harmonic oscillator o (4.e.59) by induction noting ﬁrst that (4.59) holds for n = 0. therefore. t0 ). and showing then that the property also holds for n + 1 in case it holds for n. 1.59) dn −y2 e dy n (4. (4. Our derivation will follow closely the procedure adopted for the case of a ‘particle in a box’ [see Eqs.114)]. 4. Starting point of our derivation is the assumption that the initial condition can be expanded in terms of the eigenstates φn (y) (4.59) as follows g(y) = −e = −e y2 2 y2 2 n y2 d d − y2 2 e 2 (−1)n e 2 e−y n dy dy d − y2 d n − y2 e 2 y − e 2 .57) of the hermite polynomials and the deﬁnition given by (2.200) we need to verify y − which implies Hn (y) = (−1)n ey 2 d dy n e− y2 2 = (−1)n e y2 2 dn −y2 e . g(y) = (−1)n+1 e y2 2 dn+1 −y2 e = dy n+1 y − d dy n+1 e− y2 2 .1.

y0 . 68. pp.67) for which according to (3.82 Linear Harmonic Oscillator Such expansion is possible for any f (y0 ) = ψ(y0 . t1 |y0 . For this purpose we start from the integral representation (2.70).7. N.h.46) the expansion coeﬃcients dn are +∞ dn = −∞ dy0 ψ(y0 . Englewood Cliﬀs. We want to demonstrate now that this propagator is identical to the propagator (2. (4.70) is the propagator of the linear harmonic oscillator.72) × exp( −u − v + 2iyu + 2iy0 v ) . (4. sec:harm.15..71) Applying (??) to express Hn (y) and Hn (y0 ) yields exp(−2tuv) w(y. this is an excellent textbook from which we have borrowed heavily in this Section. Inc. In order to prove the equivalence of (4. Employing orthogonality condition (4.112) and (4.J. a supposition which is not proven here2 . t) = π −3/2 exp 2 2 y 2 y0 + 2 2 2 +∞ +∞ ∞ du −∞ −∞ dv n=0 1 ( −2tuv )n × n! (4.225) which allows one to derive a generating function for products of Hermite polynomials which can be applied to the r.65) to times t ≥ t0 through insertion of time-dependent coeﬃcients cn (t1 ) ∞ ψ(y. We consider for this purpose the following expression for |t| < 1 ∞ w(y. 2.147) for the harmonic iscillator determined in Sect.39) holds cn (t1 ) = exp −i ω ( n + 1 ) ( t1 − t0 ) 2 . t0 ) φn (y0 ) .Lebedev (Prentice Hall.N.. t) (4.108–3. t0 ) which is an element of N1 deﬁned in (4.69) where ∞ φ(y. t0 ) ψ(y0 . 1965) Sect. t|y0 .5).68) Altogether one can express then the solution +∞ ψ(y.147) we employ the technique of generating functions as in Sect. t1 ) = n=0 dn φn (y) cn (t) (4. 2 A demonstration of this property can be found in Special Functions and their Applications by N. . 4.s. 1 t = e−iω(t1 −t0 ) (4.66) One can extend expansion (4. 2n n! π 2 2 (4. t1 ) = −∞ dy0 φ(y. y0 . of (4.70) and (2. t0 ) = n=0 φn (y) φn (y0 ) tn+ 2 . t) = n=0 Hn (y)e−y /2 Hn (y0 )e−y0 /2 n √ t .

t0 ) (4.77) The sum in (4.30) to replace y and y0 and that we multiply the propagator by mω/ .147) derived by means of the path integral description.81) imω 2 sinω(t1 − t0 ) (x2 + x2 ) cosω(t1 − t0 ) − 2 x x0 0 This result agrees with the propagator (2. t0 ) = and 1 2 i π F (t1 .77). t) = π −1 exp − 2 y 2 y0 + 2 2 +∞ dv exp(−(1 − t2 ) v 2 − 2i (yt − y0 ) v) . t0 ) exp i 2 1 2 (y + y0 ) G(t1 . i.80) −2iω(t1 −t0 ) sinω(t1 − t0 ) 1−e We can ﬁnally express the propagator (4. indeed.70) is. .h.. y0 .e. by a factor 4 mω/ for both φn (y) and φn (y0 ).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 .74) a second time yields ﬁnally together with the deﬁnition (4. y0 .s.78) 1 − e−2iω(t1 −t0 ) = sinω(t1 − t0 ) 2 i e−iω(t1 −t0 ) (4. (4. −∞ (4.71) of w(y. √ πexp( t2 v 2 − 2iytv − y 2 ) (4. t0 ) = exp mω 2iπ sinω(t1 − t0 ) 1 2 × . y0 . t0 ) = where F (t1 . The resulting propagator is G(t1 . t|x0 . t0 ) − i y y0 2 F (t1 . i. (4. identical to the generating function (4. (4. This requires that we employ (4.4: Propagator Carrying out the sum on the r.74) applied once results in w(y.76) One can express this in terms of the stationary states (4. (4. t1 |y0 .79) 1 + e−2iω(t1 −t0 ) cosω(t1 − t0 ) = .78) in terms of the coordinates x and x0 .4.e.73) The incomplete Gaussian integral +∞ −∞ dx e−a 2 x2 √ − 2bx = π b2 /a2 e . it holds φ(y. one obtains w(y.75) Applying (4. a Re a2 > 0 (4.. t) = π −3/2 exp +∞ 83 y2 y2 + 0 2 2 +∞ dv exp( −v 2 + 2iy0 v ) × −∞ × −∞ du exp( −u2 − 2u(tv − iy) ) . 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 ) = i φ(x.

83) We will encounter below functions of a+ and a− . the use of generating functions. (4. both quantum systems are endowed with a large number of modes. 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. (4. and of the modes of the quantized electromagnetic ﬁeld.84) = 1 1 = 0 √ = n + 1 φn+1 (y) √ = n φn−1 (y) = f |ˆ+ g a Ω∞ . 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! .. We want to introduce now techniques based on the ladder operators a+ and ˆ a− . Here f (ν) (y0 ) denotes the ν-th derivative of f (y) taken at y = y0 .g. 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. the approaches a ˆ a ˆ using y.84 Linear Harmonic Oscillator 4. ˆ ˆ To put the following material in the proper modest perspective we may phrase it as an approach which rather than employing the coordinate y and the diﬀerential operator d/dy uses the operators 1 d 1 d y − .. a− must be equivalent. e. In this case we deﬁne ∞ f (ν) (0) + ν (4.5 Working with Ladder Operators In the last section we have demonstrated the use of diﬀerential equation techniques. 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. d/dy and a+ .82) a+ = √ ˆ dy dy 2 2 √ √ Obviously. i. f (ˆ+ ). a− = √ ˆ y + . one can express y = 2(ˆ− + a+ ) and d/dy = 2(ˆ− − a+ ) and. each corresponding to a single harmonic oscillator of the type studied presently by us. hence. a (4. The reason for introducing the calculus of the operators a+ and a− is that ˆ ˆ this calculus proves to be useful for the description of vibrations in crystals. phonons. ˆ ˆ Calculus of Creation and Annihilation Operators We summarize ﬁrst the key properties of the operators a+ and a− ˆ ˆ [ˆ− .86) f (ˆ+ ) = a a ˆ ν! ν=0 .85) is convergent everywhere in R.e.

87) (4.88) we need to show actually [ˆ− . For fn+1 follows then [ˆ− . a+ ] + [ˆ− .94) This identity follows from (4.4. ˆ a a a ˆ In particular. ˆ a We proceed by induction noticing ﬁrst 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) (4.88) is a fundamental one and will be used in the following.92) f (y) = n=0 dn φn (y) = n=0 (ˆ+ )n a dn √ φ0 (y) n! (4. The convergence of the Taylor expansion ascertains then that (4.89) (4. we will only assume operator functions acting on φ0 (y).96) . f1 (ˆ+ )] = [ˆ− .90) holds for f0 and for f1 .90) holds for fn .90) we show that (4. An example is the so-called shift operator exp(uˆ− ). a An important operator function is the exponential function.95) (4. a a a We like to state the following property of the functions f (ˆ± ) a f (ˆ− ) φ|ψ = φ|f (ˆ+ ) ψ . a+ ] = 1 = f1 (ˆ+ ) . a a a 85 (4. The ﬁrst case is trivial.. 4.5: Working with Creation and Annihilation Operators and similarly for f (ˆ− ).87. Hence. property (4. a a (4. As long as the operators act on φ0 (y) one can state then that a− behaves like a diﬀerential operator with respect to functions f (ˆ+ ).83) and can be proven for all powers fn by induction and then inferred for all proper functions f (ˆ± ).91) (4.90) To prove (4. Note that ˆ a an immediate consequence of (4. (ˆ+ )n a+ ] = (ˆ+ )n [ˆ− . f (ˆ+ )] + f (ˆ+ ) a− ˆ a a a a ˆ which implies that in order to prove (4. The following important property holds a a− f (ˆ+ ) = f (1) (ˆ+ ) + f (ˆ+ ) a− . a− f (ˆ+ ) φ0 (y) = f (1) (ˆ+ ) φ0 (y) ˆ a a which follows from a− φo (y) = 0. f (ˆ+ )] = f (1) (ˆ+ ) .93) one can reduce all operators acting on some proper state function by operators acting on the state φ0 (y). i.e. the second case follows from [ˆ− .90) holds for f (ˆ+ ). We note ˆ a− f (ˆ+ ) = [ˆ− .88) is (ˆ− )n f (ˆ+ ) φo (y) = f (n) (ˆ+ ) φ0 (y) .90) holds for any function fn (ˆ+ ) = (ˆ+ )n which is a power of a a a+ . fn+1 (ˆ+ )] a a = [ˆ− . It holds a a euˆ f (ˆ+ ) φ0 (y) = f (ˆ+ + u) φ0 (y) . a a − (4. a a a ˆ 1 a Let us assume that (4.88) (4.

100) ˆ ˆ To solve this diﬀerential equation we deﬁne C(u) = D(u) exp(−u2 /2) which leads to d ˆ D(u) = du zˆ+ + z ∗ a− a ˆ ˆ D(u) (4.98) for u = −α∗ and v = α yields e−α This together with (4. hence. without explicitly stating this.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 (4.102) Similarly.99) ˆ C(u) = zˆ a zˆ+ + z ∗ a− a ˆ + +z a ˆ ∗ − a euzˆ euz ˆ − u C(u) .108) . + ∗ a− ˆ a − u euzˆ euz + ∗ a− ˆ = (4.105) a eαˆ + − α∗ a− ˆ φ0 (y) = e αα∗ 2 e−α ∗ a− ˆ a eαˆ φ0 (y) .107) − α∗ a− ˆ φ0 (y) = e− αα∗ 2 a eαˆ φ0 (y) . the solution of (4. 1 + + uz ∗ a− ). + (4. that the operators considered act on 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) .97) An example in which (4.104) (4.106) Applying (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) . + (4.103) (4. euzˆ ˆ euz ∗ a− ˆ (4. One can conclude then using the deﬁnition of C(u) and deﬁning ˆ ˆ is D(u) = exp(uzˆ a ˆ v = uz 1 ∗ ˆ− ∗ ∗ ˆ− a+ a+ evˆ + v a φ0 (y) = e 2 vv evˆ ev a φ0 (y) .106) yields a eαˆ + ∗ a− ˆ a a eαˆ φ0 (y) = e−αα eαˆ φ0 (y) . We assume in the following a ˆ derivation.90) and zz ∗ = 1 we can write this d du e uzˆ+ a e uz ∗ a− ˆ a − z ∗ a− . + (4.98) The related operators exp[vˆ+ ± v ∗ a− ] play also an important role. (4. one obtains a evˆ + − v ∗ a− ˆ a φ0 (y) = e− 2 vv evˆ e− v 1 ∗ + ∗ a− ˆ φ0 (y) (4. + ∗ + (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. To express ˆ these operators as products of operators exp(vˆ+ ) and exp(v ∗ a− ) we consider the operator C(u) = a ˆ + ) exp(uz ∗ a− ) where u ∈ R.101) ˆ ˆ ˆ C(u) obviously obeys C(0) = 1 This results in D(0) = 1 and. a a ν! (4.101) 1. z ∈ C.

.115) a where the latter step follows after expansion of evˆ and using a− φ0 (y) = 0.58) 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. Similarly.58) the Hermite polynomials Hn (y) by the eigenstates φn (y) √ ∞ 1 ( 2t)n y2 /2 2yt − t2 √ e φn (y) (4.e.µ=0 uµ v ν √ φµ |φν µ!ν! ∞ = µ=0 uµ v µ µ! (4. and ˆ l.h.114) where we have employed (4.95).s.111) This expression.h.s. Setting y = 0 in (4. is the equivalent of the generating function (??). the expected from which follows by comparision of each term on both sides φµ |φν orthonormality property. Expanding r. i.111) the orthonormality properties of the wave functions φn (y). (4.4.84) yields ∞ ν.s.84) and deﬁning 2t = u one can write then π − 4 exp − or π − 4 exp − 1 1 √ y2 u2 + 2uy − 2 2 ∞ = n=0 un + n (ˆ ) φ0 (y) a n! + (4. in the ladder operator calculus.98) and again (4. 4. provide the same values for Hn (0) as provided in Eq. + (4.h.95) one obtains a a a a euˆ φ0 |evˆ φ0 = φ0 |euv evˆ φ0 = euv evˆ φ0 |φ0 = euv − + + + − (4.h. For this purpose we consider a a a a euˆ φ0 |evˆ φ0 = φ0 |euˆ evˆ φ0 + + − + (4.115) and using (4. We start from the generating function (??) and replace according to (4. We want to derive now the values Hn (0). Using (4.116) = δµ ν . of (4.84.113) Comparision of all terms on the l. . 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).110) √ y2 u2 + 2uy − 2 2 a = euˆ φ0 (y) .109) e = π4 n! n=0 √ Using (4.111) yields π − 4 exp − 1 u2 2 a = euˆ φ0 (0) . we can reproduce by means of the generating function (4.s.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. and on the r. (??).112) Expanding both sides of this equation and using (4.

2). using (4.124) ).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. according to (4. the momentum and position operators are as stated in (4.125) .119) 2m 2 dp2 Accordingly. For this purpose we employ the Schr¨dinger equation in the momentum o representation. The eigenvalues. the momentum and position operators are p = p. 1. of (4.e.235). n = 0. the time-independent Schr¨dinger equation for the oscillator can be written o ˆ H = − mω 2 2 2 x = i ˆ d . is given by the ˜ function ψ(p). 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. i. 2.120) Multiplying this equation by 1/m2 ω 2 yields − where ˜ E = E / m2 ω 2 . are ˜ En = or. n = 0. (4. dp (4. .235). .122) For a solution of the Schr¨dinger equation in the momentum representation.121).. 4. The momentum representation eigenfunctions are. ˜ (4. . (4.6) and the Hamiltonian (4.3).7) is 1 2 m 2 ω 2 d2 p − . ω = 1 / m2 ω .. is given by a function ψ(x). . The solutions can be stated readily exploiting the earlier results.2. (−i)n ˜ φn (p) = √ 2n n! m˜ ω π 1 4 exp − m˜ p2 ω 2 Hn ( m˜ ω p) . In the momentum representation the wave function is a function of p. (4. the eigenvalues are identical to those determined in the position representation.118) d2 1 2 + p dp2 2m ˜ ˜ φE (p) = E φE (p) . using (2. For example.e. the probability density is P (x) = |φn (x)|2 .7) is given by (4.121) ω (n + ˜ 1 ) 2 . In the position representation the wave function is a function of x. i. i.123) (4. (4.37). (4.e. En = ω (n + 1 ) 2 d2 1 + m˜ 2 p2 ω 2 2m dp 2 2 ˜ ˜˜ φE (p) = E φE (p) (4.122). As to be expected.88 Linear Harmonic Oscillator 4. for an oscillator in a stationary state φn (x) as given by (2. 1. . ˆ and the Hamiltonian (4. . 2.6 Momentum Representation for the Harmonic Oscillator The description of the harmonic oscillator provided so far allows one to determine the probability density P (x) of ﬁnding an oscillator at position x..

of course. m ω (4.129) m ω one obtains.30) and (4.129) follows ky = px/ and.235)] absorbs the replacement dx → dy. m ω (4. i.4.f.. instead of (4. i. the correctness of the phase factors (−i)n . hence.. the Jacobian.127) exp − p2 m ω 2 Hn ( 1 p) .e. φn (k) = We want to ﬁnally apply the relationship 1 ˜ φn (p) = √ 2π +∞ 2 (−i)n −k √ e 2 Hn (k) . this allowed us to introduce in (4.130) dx exp(−ipx/ ) φn (x) −∞ (4. Deﬁning k = m˜ / p or ω 1 k = p (4.133) where we note that a change of the normalization of φn (y) [c.125. 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 (4.134). 4.131) in the present case employing dimensionless units.126) a phase factor (−i)n . (4. ˜ We can now state the probability density Pn (p) for an oscillator with energy En to assume momentum p.e. deﬁned only up to an arbitrary phase factor eiα .58).1: Demonstrate the validity of (4.122) (−i)n ˜ φn (p) = √ 2n n! 1 πm ω 1 4 89 exp − p2 2m ω Hn ( 1 p) .132) dy exp(−iky) φn (y) −∞ (4. Proceed as follows.126) in terms of dimensionless units following the procedure adopted for the position representation where we employed the substitution (4.6. 1 ˜ φn (k) = √ 2π +∞ (4. . 2n n! π (4.6: Momentum Representation or with (4.30) to obtain (4. From (4. α ∈ R.126) Normalized eigenfunctions are.128) We may also express the eigenstates (4. This implies the property of Hermite polynomials (−i)n e− k2 2 +∞ Hn (k) = −∞ dy exp(−iky) e− y2 2 Hn (y) (4.126). through direct integration.30) and (2.134) Exercise 4.

The question arises if similar states exist also for the quantum oscillator. z) = exp −2ikz + z 2 − k 2 /2 .138) yields using a− φn (y) = n φn−1 (y) ˆ ∞ n=0 √ n d(α) φn−1 (y) − α d(α) φn (y) n n = 0 (4. 2. φ(α) | φ(α) = . y) −∞ (4. Inc. The quasi-classical states can be obtained by generalizing the construction of the ground state of the oscillator. for example.Grynberg (John Wiley & Sons. We ﬁrst construct the solution of (4. +∞ Linear Harmonic Oscillator dy exp(−iky) g(z. pp. except that α changes periodically in time. The answer is ‘yes’. n n=0 d(α) n 2 = 1 (4.. Inserting this expansion into √ (4. and G. 4. We will ﬁnd that α can be characterized as a displacement form the minimum of the oscillator and that the state φ(α) (y) displays the same spatial probability distribution as the ground state. New York.— III.C.90 (a) Prove ﬁrst the property 1 f (k. 4. (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 . carries out a periodic motion described by y(t) = Re y0 eiωt (4.137) where y(t) describes the center of the particle and where the mass of the particle remains narrowly distributed around y(t) for an arbitrary period of time. 192 3 .4. a pendulum swinging at small amplitudes. Such states are referred to as quasi-classical states. ˆ α ∈ R. for example.136) (b) Employ the generating function (2.194. z) = √ 2π where f (k.139) where we have added the condition that the states be normalized.135) g(k.7 Quasi-Classical States of the Harmonic Oscillator A classical harmonic oscillator. 1989) which discusses quasi-classical states of the free electromagnetic ﬁeld in Sect. For this purpose we assume such state exists and expand ∞ ∞ φ(α) (y) = n=0 d(α) φn (y) . The motion of the probability distribution of such state is presented in Fig.Chen-Tannoudji.138) We will show that a harmonic oscillator prepared in such a state will remain forever in such a state.134). namely through the eigenvalue problem for normalized states a− φ(α) (y) = α φ(α) (y) . z) = exp 2yz − z 2 − y 2 /2 (4. We want to construct and characterize such states.1 showing in its top row the attributes of such state just discussed. or Glauber states of the harmonic oscillator and they play.Dupont-Roc.140) An excellent textbook is Photons and Atoms: Introduction to Quantum Electrodynamics by C. coherent states.196) of Hermite polynomials and equate coeﬃcients of equal powers of z to prove (4. J.138).

149) with (4.148) The identity (4. n n+1 (4.84) φ(α) (y) = exp − α2 2 ∞ n=0 αα∗ (αˆ+ ) a a+ φ0 (y) = e− 2 eαˆ φ0 (y) .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 (4.142) The solution is αn d(α) = √ c n n! ∞ ∞ (4. n! (4.139)] 1 = n=0 d(α) n 2 = |c|2 n=0 α2 2 = |c|2 eα . n! n (4.138).. (4. This interpretation justiﬁes the choice of α ∈ R in (4.146) √ 1 − (y − 2 α)2 2 (4. .e. n! (4. Comparing (4. However.147) allows one to interprete T (α) as an operator √ which shifts the ground state by 2α.145) using (4. [π]− 4 e− one can write φ(α) (y) = [π]− 4 exp 1 1 y2 2 + √ 2 α y − α2 = exp − α2 2 ∞ n=0 αn √ φn (y) .144) Normalization requires c = exp(−α2 /2) and. n! (4.7: Quasi-Classical States or ∞ n=0 91 √ n + 1 dn+1 − α d(α) n (α) φn (y) = 0 . i. the quasi-classical states are φ (α) (y) = exp α2 − 2 ∞ n=0 αn √ φn (y) .4.147) √ which identiﬁes this state as a ground state of the oscillator displaced by 2α.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) .109).150) for α presently real.143) for a constant c which is determined through the normalization condition [see (4.145) Using the generating function in the form (4. we will see below that α ∈ C are also admissible and will provide a corresponding interpretation. One can also write (4. hence.149) − α∗ a− ˆ (4.

154) . n! u = e−iω(t1 − t0 ) . n = 0. 1.150) 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 (4.158) Comparision of this expression with (4. .151) which follows from a generalization of (4. Relationship (4.160) .95) to linear combinations αˆ+ + βˆ− and reads then a a f (αˆ+ + βˆ− ) φ|ψ = φ|f (α∗ a− + β ∗ a+ ) ψ .145) shows that α in (4.e. one can write in case of T + (α) T + (α) + = T (α) = eα ∗ a− ˆ − αˆ+ a a = e − (αˆ + − α∗ a− ) ˆ .h.h.150) for complex α obeys the unitary property T + (α) T (α) = 1 1 (4.157) (y. i. (4. α ∈ C (4..156) (4.153) This is obviously the inverse of T + (α) as deﬁned in (4. .152) Since the operator function on the r.68)] one can write the time-dependent solution with (4. it holds T + (α) φ0 |T + (α) φ0 = f |T (α) T + (α) g = φ0 |φ0 .67.145) is replaced by a complex number αu which at times t1 = t0 + 2πn/ω. the shift operator leaves the ground state normalized.145) as the initial state φ(α) (y. is the adjoint of the operator function on the l. a a ˆ ˆ (4.92 Linear Harmonic Oscillator The shift operator as deﬁned in (4. t1 ) = exp or φ (α) − α2 2 ∞ exp n=0 −i ω ( n + ∞ n=0 1 ) ( t1 − t0 ) 2 αn √ φn (y) n! (4. In particular.109) serves again to simplify this sum √ y2 1 1 α2 (u2 + 1) φ(α) (y. t1 ) = π − 4 u 2 exp − + 2yαu − 2 2 E (4. t1 ) = u exp 1 2 α2 − 2 (α u)n √ φn (y) . (4.s.s. becomes real.155) (4.152) T + (α) f |T + (α) g = f |T (α) T + (α) g = f |g .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. 4. Following the procedure adopted in determining the propagator of the harmonic oscillator [see (4. .

t1 ) = π − 4 exp − [y − y0 (t1 )]2 + iy k0 (t1 ) − iφ0 (t1 ) − ω(t1 − t0 ) 2 2 .1 (top row) we present the probability distribution of the Glauber state for various instances in time.165) . The time-dependent wave function (4.138): a real α corresponds to a displacement such that initially the oscillator is at rest.161) are time-dependent √ y0 (t1 ) = 2 α cosω(t1 − t0 ) √ k0 (t1 ) = − 2 α sinω(t1 − t0 ) 1 φ0 (t1 ) = − α2 sin2ω(t1 − t0 ) (4. The diagram illustrates that the wave function retains its Gaussian shape with constant width for all times. 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.148. This shows clearly that the Glauber state is a close analogue to the classical oscillator.162) 2 with a periodic change. experiences a change of its width (relative to . 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. 4. t1 ) = u 2 T + (α u) φ0 (y) (4. In Fig. Our interpretation 2 explains also the meaning of a complex α in (4. moving solely its center of mass in an oscillator fashion around the minimum of the harmonic potential.147).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 (4.150) with (4.158) allows one to write 1 φ(α) (y.164) where according to (4.161) and is found to describe a displaced Gaussian.4. a momentum factor and a phase factor. (4. being displaced around the center with period 2π/ω and. 4.138. a complex α corresponds to a displaced oscillator which has initially a non-vanishing velocity. except that it is not pointlike. Comparing (4.163) This solution corresponds to the initial state given by (4. (4. this characterization corresponds closely to that of the possible initial states of a classical oscillator. obviously.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. 4.159) is then 1 1 i φ(α) (y. One can express (4. in general.163) also through the shift operator. It is of interest to note that the displacement y0 of the Gaussian as well as the momentum k0 and phase φ0 associated with (4.

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. t0 ) = n=0 dn φn (y) . (4. t0 ) = [πσ 2 ]− 4 exp we expand ∞ 1 − (y − ao )2 2σ 2 (4. t0 ) +∞ = √ 1 n n!πσ 2 e−ao q/2 = √ 2n n!πσ dy exp −∞ +∞ −∞ − (y − ao )2 y2 − 2σ 2 2 2 Hn (y) (4.169) .1: Time dependence of Gaussian wave packets in harmonic oscillator potential. One can recognize the cyclic displacement of the wave packet and the cyclic change of its width with a period of twice that of the oscillator.1 (second and third row). 2σ 2 q = a . σ|y.166) φ(ao . This behaviour is illustrated in Fig.167) One can derive for the expansion coeﬃcients +∞ dn = −∞ dy φn (y) φ(ao . 4.168) dy e−p(y−q) Hn (y) where p = 1 + σ2 . σ|y. σ|y. 1 + σ2 (4. To describe such state which satisﬁes the initial condition φ(ao . that of the Glauber states) with a period π/ω.

143) of the Glauber state.. σ|y. t1 ) = e−aq/2 √ pσ ∞ n=0 e −iω(n+ 1 )(t1 −to ) √ 2 1 2n n! p−1 p n/2 (4.175) ∞ tn n=0 2n n! Hn (yo ) e−yo /2 Hn (y) e−y 2 2 /2 The generating function (4.e. New York. σ→1 O(y n−2 ) for n ≥ 2 1 for n = 0. a ground state shifted by ao = 2α. Following the strategy adopted previously one can express the time-dependent state corresponding to the initial condition (4. Marichev (Wiley. t1 ) = √ 2 × exp − 1 (y 2 + yo ) 1+t2 − 2 1−t 2 1 1 √ √ pσ π 1 − t2 × t 1 − t2 (4.166) in analogy to (4. σ|y. namely $750 4 . in agreement with the expected result (4. t1 ) = √ × where yo = q p a = √ .7: Quasi-Classical States The solution of integral (4. unfortunately very expensive.157) as follows φ(ao .I. in fact. vol. 1990).143). This expansion can be written φ(ao .P.176) 1 √ pσ π a 2 exp − 1 1+σ2 + 1 yo − 1 ω(t1 − to ) 2 2 2 2 × (4.172) One can then conclude e−a /4 n lim dn = √ a σ→1 2n n! (4. therefore. and O. 2 by A. σ|y. Yu.171) lim p−1 p n 2 Hn (q p ) = lim (2q)n = an . i. a worthy successor of the famous Gradshteyn. this 3 volume integral table is likely the most complete today.173) which is.174) Hn (q p p−1 ) φn (y) . σ→1 p−1 2 (4. In this limit the coeﬃcients dn should√ become identical to the coeﬃcients (4. 1 (4.170) In order to check the resulting coeﬃcients we consider the limit σ → 1. Brychkov. p−1 1 − σ4 t = 1 − σ 2 −iω(t1 −to ) e 1 + σ2 (4.4. p−1 95 (4.76) permits us to write this φ(ao . Prudnikov.168) is4 e−aq/2 (p − 1)n/2 dn = √ Hn (q 2n n!σ p(n+1)/2 p ).177) − 1 ω(t1 − to ) 2 . 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. 1 a2 2 1 + σ2 + 1 2 2 y o + 2 y yo This integral can be found in Integrals and Series.

96 Linear Harmonic Oscillator .

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

A rotation around the x3 -axis by an angle ϕ is described by the matrix cosϕ − sinϕ 0 R(ϑ = (0.6) (5. 2.e.e.7) Let us consider brieﬂy an example to illustrate rotational transformations and to interpret the sign of detR(ϑ). therefore. In the latter case the determinant of R(ϑ = (0. then the ﬁngers of your ﬁst point in the direction of the rotation. 3 . ϕ)T ) is negative. From (5. i. where ϑ denotes the rotation.e.98 Properties of Rotations in R Theory of Angular Momentum and Spin Rotational transformations of vectors r ∈ R .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. in case of a prefactor −1 a rotation and an inversion at the origin. i. (5. 0. x3 )T .8) 0 0 1 In case of a prefactor +1. We assume the convention that rotations are right-handed. (5. x2 .k.3) Since this holds for any a and b.2) Rotations conserve the scalar product between any pair of vectors a and b. the matrix represents a proper rotation. ϑ parametrizes the rotations uniquely as long as |ϑ| < 2π. (5..6) follows immediatety for the determinant of R(ϑ) using detA B = detA detB and detAT = detA detR(ϑ) = ±1 . 0. a minus sign in Eq.. namely around an axis given by the direction of ϑ and by an angle |ϑ|. are linear and. if the thumb in your right hand ﬁst points in the direction of ϑ. it follows 3 [R(ϑ)]jk [R(ϑ)]j = δk j=1 . It holds then j=1 3 3 a ·b = j. ϕ)T ) = ± sinϕ cosϕ 0 (5. In the following we want to exclude inversions and consider only rotation matrices with detR = 1. (5. i. can be represented by 3 × 3 matrices R(ϑ).7) implies that a rotation is associated with an inversion. they conserve a · b = 3 aj bj . Let us deﬁne the rotated vector as r = R(ϑ) r . . in Cartesian coordinates r = (x1 . =1 [R(ϑ)]jk [R(ϑ)]j ak b = j=1 aj bj .1) In Cartesian coordinates this reads 3 xk = j=1 [R(ϑ)]kj xj k = 1.5) This equation states the key characteristic of rotation matrices.(5. (5.

R3 = R1 R2 is a real. For this purpose we must demonstrate that the group properties (i–iv) hold.5. (ii) The group element ‘e’ is played by the 3 × 3 identity matrix 1 1. R2 ∈ SO(3) and do not write out “◦” explicitly since it represents the matrix product. (ii) There exists an element e in G with the property ∀a ∈ G → e ◦ a = a ◦ e = a. hence. (5. In the following we will assume R1 .11) (iii) From detR1 = 0 follows that R1 is non-singular and. We would like to point out the obvious property that for all elements R of SO(3) holds R−1 = RT (5.12) (iv) Since the associative law holds for multiplication of any square matrices this property holds for the elements of SO(3). one can show R3 R3 = 1 Furthermore. (i) Obviously.13) . b ∈ G the product c = a ◦ b is also in G. We need to demonstrate that this inverse belongs also to −1 T SO(3).9) forms a group. We want to prove now that SO(3) as deﬁned in (5. 99 (5. T R1 R1 −1 = 1 −1 = 1 1 1 (5. it holds 1. It holds T T T T R3 R3 = ( R1 R2 )T R1 R2 = R2 R1 R1 R2 = R2 R2 = 1 . there exists a real 3 × 3–matrix −1 R1 which is the inverse of R1 . 1 T Similarly. a ∈ G → ∃a−1 ∈ G with a ◦ a−1 = a−1 ◦ a = e. (iii) ∀a. (5. 3 × 3–matrix. It follows that R3 is also an element of SO(3).9) This set of matrices is closed under matrix multiplication and. Since (R1 )T = (R1 )−1 it follows −1 −1 −1 T (R1 )T R1 = (R1 )−1 R1 = −1 which implies R1 ∈ SO(3). (iv) For products of three elements holds the associative law a ◦ (b ◦ c) = (a ◦ b) ◦ c. R RT = RT R = 1 detR = 1 } 1. in fact.10) detR3 = det ( R1 R2 ) = detR1 detR2 = 1 .1: Matrix Representation of SO(3) Deﬁnition of the Group SO(3) We will consider then the following set SO(3) = { real 3 × 3 matrices R. We have shown altogether that the elements of SO(3) form a group. forms a group G satisfying the axioms (i) For any pair a.

x2 –plane {rotation by Oo . J2 . Generators.16) must be antisymmetric. presently. establish if 1-1 homomorphic mappings exist. k = exp ( −ϑk Lk ) (5. M4 } = 1 0 0 1 .e. Lie Algebra We want to consider now a most convenient.13) R(ϑk ek )−1 = R(ϑk ek )T = [exp ( ϑk Lk )]T = exp ϑk LT ˆ ˆ k and. 0 −1 1 0 (5.20) (5. Geometric intuition tells us R(ϑk ek )−1 = R(−ϑk ek ) = exp ( −ϑk Lk ) ˆ ˆ Using the property which holds for matrix functions [f (R)]T = f (RT ) we can employ (5. rotation by 90o . i. M3 .19) (5. we want to assume that they exist.100 Theory of Angular Momentum and Spin Exercise 5. −1. the matrices i Lk = − Jk (5. will play a central role in the representation of many other symmetry transformations in Physics.15) Below we will construct appropriate matrices Jk . in fact. 0 1 −1 0 . (b) the set of matrices {M1 . identify subgroups. M2 . we arrive at the property exp ϑk LT k from which follows LT = −Lk . due to the uniqueness of the inverse. rotation by 180o .1: Test if the following sets together with the stated binary operation ‘◦’ form groups. 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.14) together with matrix multiplication as the binary operation ‘◦’.1. We want to show that for the property R RT = 1 to hold. Lie Group.17) .18) (5. J3 as follows i R(ϑ) = exp − ϑ · J (5. i. −i} together with multiplication as the binary operation ‘◦’. rotation by 270o } and consecutive execution of rotation as the binary operation ‘◦’. (c) the set of four rotations in the x1 . −1 0 0 −1 . This representation expresses rotation matrices in terms of three 3 × 3– matrices J1 . mappings between the sets which conserve the group properties: (a) the set of real and imaginary numbers {1. compact representation of R(ϑ) which will be of general use and.

1. Using the deﬁnition of partial derivatives we can state L1 = lim ϑ−1 R(ϑ = (ϑ1 . We assume now that we parameterize rotations through a three-dimensional rotation vector ϑ = (ϑ1 .23) (5.8) we can state cosϑ3 −sinϑ3 0 1 −ϑ3 0 1 0 . A must have the form 0 a b A = −a 0 c . ϑ2 .27) .21) −b −c 0 It is this the reason why one can expect that three-dimensional rotation matrices can be expressed through (5. (5. L3 which are independent of the rotation angles. 0)T ) − 1 1 1 ϑ1 →0 (5. Let us use this deﬁnition to evaluate L3 .25) 0 −ϑ3 0 0 −1 0 ϑ3 0 0 = 1 0 0 . This property gives A then only three independent matrix elements. L2 . ϑ3 )T such that for ϑ = (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).e. One can determine then the matrices Lk deﬁned in (5. 0.1: Matrix Representation of SO(3) 101 We can conclude then that if elements of SO(3) can be expressed as R = eA . (5.26) (5. Exercise 5.22) and similar for L2 and L3 .24) (5. the real matrix A is anti-symmetric. 0)T . 0. ϑ3 )T ] = lim sinϑ3 cosϑ3 0 = ϑ3 ϑ3 →0 ϑ3 →0 0 0 1 0 0 1 Insertion into (5. 0.5.22). ϑ3 and three matrices L1 .15) as the derivatives of R(ϑ) with respect to the Cartesian components ϑk taken at ϑ = (0. Using (5.. 0. 0)T the rotation is the identity. i. lim R[(0.15) with three real parameters ϑ1 .2: Determine the generator L1 according to Eq. ϑ2 .(5. 0 0 0 0 0 0 (5.

the Lie algebra of SO(3) can be written [Lk .. holds the algebra [Jk . the matrix (5. if the resulting products of exponential operators can be expressed in the form (5. L ] = m=1 fk m Lm . (5. (5.32) For the matrices Jk = i Lk which. the elements of if any two indices are identical for k = 1. B]. only the latter example and . 2 .g. In case of the group SO(3) the structure constants are particularly simple. From this result one can conclude that expressions of the type (5.33) We want to demonstrate now that exp(ϑ · L) yields the known rotational transformations. commutators of commutators of A and B. The expression for Zn are actually rather complicated. 0)T ) = exp( 2 L2 ) and R(ϑ = (0. etc. x2 -.22) implies. i. 0.8) in case of ϑ = (0. however. as we see later..30) Groups with the property that their elements can be expressed like (5. it must hold 3 [Lk .31) where k m are the elements which are 0 1 k m = 1 −1 of the totally antisymmetric 3-dimensional tensor.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.30) are called Lie groups. = 2. that the operators (5. [A. m = 3 for any odd permutation of k = 1.29) where Z1 = A + B and where the remaining operators Zn are commutators of A and B. (5. e. J ] = i k m Jm . 0)T ) = exp( 1 L1 )..e. R(ϑ = (0. B]] + [[A. and the constants fk m are called the structure constants. Lie groups can have any number of generators. 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. Z2 = 2 [A. 3 )T ) = exp( 3 L3 ). 0. The question is then.g. In fact. B] ). = 2. We want to consider. ϕ)T . that the generators represent inﬁnitesimal rotations with 1 >> |ϑ| around the x1 -. Finite rotations can be obtained by applications of many inﬁnitesimal rotations of the type R(ϑ = ( 1 . L ] = k m Lm (5.15) and their generators obey the property (5. m = 3 for any even permutation of k = 1. in fact.30) is called the Lie algebra. Z3 = 12 ( [A. i.e. the property (5. 0. = 2. however. 1 1 e. (5. m = 3 .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. are related to angular momentum operators.28) The construction of the generators through the limit taken in Eq. three being a special case.15).15) obey the group property. and x3 -axes. An answer can be found considering the Baker-Campbell-Hausdorf expansion ∞ exp( λA ) exp( λB ) = exp( n=1 λn Zn ) (5. Of course. B].

(5. 0 −1 1 0 = (−1)n 0 −1 1 0 (5. (5. using (5.34) One obtains then for the rotational transformation exp A 0 0 0 0 0 1 0 0 1 0 0 0 1 0 0 1 0 0 0 1 0 0 1 ∞ ν=1 ∞ ν=1 0 0 1 = = = + + 1 ν! 1 ν! A 0 0 0 0 0 ν Aν 0 0 0 0 0 0 0 1 ∞ 1 ν ν=0 ν! A 0 0 = eA 0 0 .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 .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 .3: Show that the generators Lk of SO(3) obey the commutation relationship (5.36) The property 0 −1 1 0 allows one to write 0 −1 1 0 and. cos ϕ − sin ϕ 0 = sin ϕ cos ϕ 0 0 0 1 (5. that the matrix ϕL3 .35). accordingly.41) cos ϕ − sin ϕ sin ϕ cos ϕ (5.5. according to (5.34.40) . (2n + 1)! (5.27).37) = (−1)n 1 0 0 1 .39) Recognizing the Taylor expansions of cos ϕ and sin ϕ one obtains eA = and. A = ϕ 0 −1 1 0 .1: Matrix Representation of SO(3) determine exp(ϕL3 ). ∞ 2n = − 1 0 0 1 2n+1 (5.8).1. can be written ϕL3 = 0 103 A 0 0 0 0 .31). eϕL3 which agrees with (5. 5. We note. (5. Exercise 5.

Deﬁnition We ﬁrst deﬁne how a rotational transformation acts on a wavefunction ψ(r ).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).48) . the R(ϑ) ψ(r) = ψ(R(ϑ) r ).44) holds. i. In particular. In analogy to R(ϑ) being a linear map R → R . (5. Exercise 5. its elements can be represented through an exponential R(ϑ) = exp i − ϑ·J (5.46) (5. a relationship which we like to express as follows R(ϑ) = ρ R(ϑ) .45) (5.e. the transformation R(ϑ): C∞ ( ) → C∞ ( ) is related to the transformation R(ϑ): R → R .43) ρ( ) conserves the group property of SO(3). hence. in fact. we deﬁne rotations R(ϑ) as linear maps C∞ ( ) → C∞ ( ). Since this holds for any ψ(r ) one can conclude ρ R(ϑ1 ) ρ R(ϑ2 ) group property (5. B ∈ SO(3) holds ρ(A B) = ρ(A) ρ(B) . one isomorphic to SO(3) and. = ψ([ R(ϑ1 ) R(ϑ2 ) ]−1 r) = ρ R(ϑ2 ) ψ([R(ϑ1 )]−1 r) (5. For this purpose we require stringent continuity properties of the wavefunction: the wavefunctions under consideration must be elements of the set C∞ ( ). i.e. namely R(ϑ)ψ(r ) = ψ(R−1 (ϑ) r ) . complex-valued functions over the 3-dimensional space R which can be diﬀerentiated inﬁnitely often.2.47) = ρ R(ϑ1 ) R(ϑ2 ) .104 Theory of Angular Momentum and Spin 5. for A.. i. (5.e. (5.1: Assume the deﬁnition ρ ρ (A B) = ρ (B) ρ (A).42) Obviously. Show that in this case holds Generators One can assume then that R(ϑ) should also form a Lie group.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) . we will learn that transformation patterns and invariances are connected with the angular momentum states of quantum mechanics.

3 1 ϑ3 0 x1 x1 + ϑ3 x2 R(0. 0 0 1 x3 x3 (5. it holds [Jk . (5. .2: Function Space Representation of SO(3) where the generators can be determined in analogy to Eq. show that (5. (5.22) according to i − J1 = lim ϑ−1 R(ϑ = (−ϑ1 .. In fact. (5. 2 by means of limits as suggested in Eq.50) lim ϑ−1 ( f ( R(0. The generators Jk can be determined by applying (5.23.57) Exercise 5. 0.24) one can expand ϑ3 →0 ϑ3 →0 lim ϑ−1 R(ϑ = (0. Carrying out similar calculations for J1 and J2 one can derive i − J1 i − J2 i − J3 = = = x3 ∂2 − x2 ∂3 x1 ∂3 − x3 ∂1 x2 ∂1 − x1 ∂2 . 0. One obtains using (5.2: (a) Derive the generators Jk . but also the Lie algebra of the generators. J3 ] and [J3 . − i J ] f (r) 1 2 =− 2× [ (x3 ∂2 − x2 ∂3 ) (x1 ∂3 f − x3 ∂1 f ) − (x1 ∂3 − x3 ∂1 ) (x3 ∂2 f − x2 ∂3 f ) ] 2 +x1 x3 ∂2 ∂3 f − x1 x2 ∂3 f − x2 ∂1 ∂2 f + x2 x3 ∂1 ∂3 f + x2 ∂1 f 3 2(x ∂ 2 1 2 −x1 ∂2 f − x1 x3 ∂2 ∂3 f + x2 ∂1 ∂2 f + x1 x2 ∂3 f − x2 x3 ∂1 ∂3 f ] 3 = − − x1 ∂2 ) f (r) = i J3 f (r) . (5.56) We like to verify this property for [J1 .5. −ϑ3 ) r ) − f (r) ) .51) Inserting this into (5. ϑ3 )T ) f (r) − f (r) 3 (5. (5.56) holds for [J2 . in case of J3 .42)) and similarly for J2 and J3 . 0.49). −ϑ3 ) r ≈ −ϑ3 1 0 x2 = −ϑ3 x1 + x2 .53) (5.52) ( x2 ∂1 − x1 ∂2 ) f (r) .55) Not only is the group property of SO(3) conserved by ρ( ). J ] = i k m Jm .49) to a function f (r).53–5. J2 ] f (r) = − = − 2[ 2 [− i J . 0. i.(5.55) and f (r) = f [J1 . k = 1. i − J3 f (r) = = Using (5.2. 0)T ) − 1 1 ϑ1 →0 105 (5. J1 ]. 5.50) 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.e.49) (note the minus sign of −ϑ1 which originates from the inverse of the rotation in Eq.54) (5. J2 ].

J3 ] J1 + J2 [J2 .61) simultaneous eigenstates of J 2 and of one of the Jk . x2 = r sinθ sinφ. J3 ] (5. However. J3 ] + [J1 .53–5.63) (5. (5. 2. Equation (5. Show that for J3 as deﬁned above holds exp( iα J3 ) ψ(φ) = ψ(φ + α). k = 1. θ. φ). C]B and (5.59) which. 3. φ) (5. [J 2 . and x3 = r cosθ.3 Angular Momentum Operators The generators Jk can be readily recognized as the angular momentum operators of quantum mechanics. Using [AB.56) one obtains [J 2 . using the momentum operator p = ˆ J = r × i .56) are the famous commutation relationships of the quantium mechanical angular momentum operators. . identiﬁes J as the quantum mechanical angular momentum operator.3: Consider a wave function ψ(φ) which depends only on the azimutal angle φ of the spherical coordinate system (r. C] = A[B.e. φ) = = 2 ( + 1) Y m (θ.65) .2. usually chosen as J3 . The commutation relationships imply that no simultaneous eigenstates of all three Jk exist. (5. 2. i.61) We demonstrate this property for k = 3.e. φ) m (θ. We will show now that the properties of spherical harmonics J2Y J3 Y m (θ. can be found. J3 ] = = = 2 2 [J1 + J2 . J3 ] J2 −i J1 J2 − i J2 J1 + i J2 J1 + i J1 J2 = 0 According to (5. To show this we ﬁrst note that (5. C] + [A.62) J1 [J1 . φ) as well as the eﬀect of the so-called raising and lowering operators J± = J1 ± i J2 (5. 3 . φ). the operator 2 2 2 J 2 = J1 + J2 + J3 (5. 5. J3 ] + [J2 .64) mY m (θ. the system related to the Cartesian coordinates through x1 = r sinθ cosφ.106 Theory of Angular Momentum and Spin Exercise 5. k = 1.60) commutes with all three Jk . i.58) i = r ×p ˆ (5.55) can be written as a vector product of r and i − J = −r × From this follows. Jk ] = 0 .. These eigenstates are the well-known spherical harmonics Y m (θ. according to the correspondence principle between classical and quantum mechanics.

56). Using [L3 . L− ] + L− L+ )| m = ( −2iL3 + .1 Let Lk .75) (5.67) (5. Continuing our proof through induction we assume now that property (i) holds for m.5. i. For this purpose we ﬁrst demonstrate that the property holds for m = − 1 by considering (i) for the state L− | ∼ | − 1 . k = 1. The deﬁnitions of the coeﬃcients α −1 and β −1 yield L+ L− | = L+ ( iβ −1 | − 1 = − α −1 β −1 | .68) (5. L− ] = [L3 . and since we have already shown | − 1 = (iβ −1 )−1 L− | we can state L3 | − 1 = −i( − 1)| − 1 .. From this follows that L− | is an eigenstate of L3 .71) To prove this theorem we ﬁrst show by induction that (i) holds. L2 = L2 + L2 + L2 . φ) 107 = = 0 = = 0 ( + m + 1)( − m) Y ( + m + 1)( − m) Y m+1 (θ. 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. L− ] + L− L3 ) | . We will show that this property holds then also for m − 1. φ) are a consequence of the Lie algebra (5.74) (5. We note L3 L− | = ( [L3 . 2.e. J+ Y J− Y J− Y m (θ. φ) (5. An Important Theorem Theorem 1. i. 3 be operators acting on a Hilbert space H which obey the algebra [Lk . in fact.76) We now show that property (ii) also holds for m = − 1 proceeding in a similar way. We consider L+ L− | m = ( [L+ .69) J+ Y (θ. L ] = k m Lm and let L± = L1 ± i L2 .70) (5. Noting [L+ . L1 − iL2 ] = −2iL3 and L+ | = 0 one obtains L+ L− | = −2iL3 | = −2 | . φ) (θ. φ) − m (θ.72) = 0 = −i | (5. For this purpose we prove the following theorem. (∗) (5.73) (5. The arguments are very similar to the ones used above and we can be brief.3: Angular Momentum Operators on the spherical harmonics. L− ] = [L1 + iL2 .e. It holds L+ L− | = (L+ L− − L− L+ + L− L+ )| = ([L+ . α −1 β −1 = 2 . L+ raises the m-value of | − 1 from m = − 1 to m = . φ) m+1 (θ.66) (5. L− ] + L− L+ )| . namely. L1 − iL2 ] = L2 + iL1 = i (L1 − iL2 ) = iL− and L3 | = −i | we obtain L3 L− | = ( iL− + L− L3 )| = i(− + 1) L− | .

it also holds for m − 1. (∗∗) (5. Angular Momentum Operators in Spherical Coordinates We want to express now the generators Jk .53–5. θ.78) ( + m + 1)( − m) ( + m + 1)( − m) . This completes the proof of the theorem. 5. φ.63.82) (5. ﬁnally αm = βm = (5.84) . in terms of spherical coordinates deﬁned through x1 x2 x3 = = = r sin θ cos φ r sin θ sin φ r cos θ (5.81) i. in particular.66. yield properties (5. given in (5. Again the argumemts are similar to the ones used above.108 Theory of Angular Momentum and Spin L− L+ )| m = (−2m − αm βm )| m ..69). One can normalize the states | m such that αm = βm .79) We can now show that (iii) holds for all proper m–values. For this purpose we need to express the angular momentum operators in terms of spherical coordinates (r. We can deduce from (∗) and (∗∗) α α α −1 β −1 −2 β −2 −3 β −3 = 2 = 2( − 1 + ) = 2( − 2 + · · −1 + ) Obviously.80) (5. We note L3 L− | m = ( [L3 .70.72. i. it holds αm βm = 2 one obtains αm βm = = ( + 1) k=m+1 k . 5.73. We will have various opportunities to employ the Theorem just derived.. This implies that (i) holds for m − 1.e. (5.79) read L+ | m L− | m + 1 = −i = −i ( + m + 1)( − m) | m + 1 ( + m + 1)( − m) | m . 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 . 5.55). 2 We note that Eqs. A ﬁrst application will involve the construction of the eigenstates of the angular momentum operators as deﬁned in (5.77) We will show now that if (ii) holds for m.64). 5. Before we can ﬁnally show that property (iii) holds as well for all − ≤ m ≤ we need to determine the coeﬃcients αm and βm . from L+ L− | m = − αm−1 βm−1 | m follows the recursion relationship αm−1 βm−1 = 2m + αm βm . Using the familiar formula + m − n k=0 k = n (n + 1)/2 − (m + 1) m = ( + 1) m − (m + 1) m (5.e. 5.83) (5. L− ] + L− L3 )| m = ( iL− − i m L− )| m = −i (m − 1) L− | m from which we can deduce L3 | m − 1 = −i(m − 1)| m − 1 . (5. We note that we can write L2 = 1 1 1 1 2 2 2 2 L+ L− + 2 L− L+ + L3 .

using (5. L1 f (θ. φ) ∂x2 /x1 ∂ ∂x3 /r ∂ = x3 + x3 ∂x2 ∂ tan φ ∂x2 ∂ cos θ ∂x2 /x1 /r ∂ ∂ − x2 − x2 ∂x3 3 ∂ cos θ f (θ. for example. applying repeatedly the chain rule. 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 ∂ = − . In fact.89) = ∂ x2 x2 ∂ x2 + x2 ∂ x3 − 33 − x2 1 3 2 x1 ∂ tan φ r ∂ cos θ r ∂ cos θ f (θ. φ) = (5.88) (5. the latter deﬁned in (5.85). using (5.3: Angular Momentum Operators 109 We ﬁrst like to demonstrate that the generators Jk . φ) = (x3 ∂2 − x2 ∂3 )f (θ.93) and similarly. Using (5.82–5. According to (5.92) (5. φ) sin θ ∂θ sin θ ∂θ L1 f (θ. (L2 )2 = − cos φ ∂ ∂ + cot θ sin φ ∂θ ∂φ 2 . φ) = ( x3 ∂2 − x2 ∂3 ) f (θ.91) which agrees with expression (5.87) holds J3 = To determine J 2 we note. cos θ ∂ cos2 φ + sin θ cos φ ∂φ cos2 θ sin θ sin φ ∂ sin θ sin φ ∂ + + sin2 θ f (θ.85) (5. φ) ∂x ∂x3 ∂ tan φ =0 (5. (L1 )2 = = ∂ 2 ∂ + cot θ cos φ ∂θ ∂φ 2 ∂ ∂ 1 ∂2 sinφ cosφ sin2 φ 2 − + 2 cotθ sinφ cosφ ∂θ ∂φ ∂θ∂φ sin2 θ ∂2 ∂ + cot2 θ cos2 φ 2 + cotθ cos2 φ ∂θ ∂φ sin φ i ∂φ .90) This yields. φ) (5.84 ) one obtains.85). actually.5. in terms of spherical coordinates. using ∂/∂tanφ = cos2 φ ∂/∂φ and ∂/∂cosθ = −(1/sinθ) ∂/∂θ. L1 f (θ. involve only the angular variables θ and φ and not the radius r.87) To derive these properties we consider.86) (5. φ). (5. We like to express now the operators J3 and J 2 . ∂φ = sin φ (5.86).60).

5. namely on the space of complex-valued functions on the unit 3-dim.e. (5.87). 5. sphere S2 .64) can be constructed as follows. ∂θ2 ∂θ ∂φ (5. 5.94) It follows (L1 )2 + (L2 )2 + (L3 )2 = With cot2 θ + 1 = 1/sin2 θ.87).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 . 5. one can conclude J2 = − 2 sin θ 2 sinθ ∂ ∂θ 2 + ∂2 ∂φ2 .95) sinθ ∂ ∂θ = ∂2 ∂ + cotθ .97).99) = = ∂2 2 ∂ 1 + + 2 2 2 ∂r r ∂r r sin θ ∂2 2 ∂ + 2 ∂r r ∂r sinθ ∂ ∂θ 2 + ∂2 ∂φ2 (5.63.96) and the relationship between Lk and Jk as given in (5. 2m r ∂r2 2m r2 2 (5. The ˙ corresponding expression for the quantum mechanical kinetic energy operator is ˆ T = − which follows from the identity 2 2 2 2m = − 1 ∂2 J2 r + . The functions in C∞ (S ) have real variables .85. 5. i. 5. The operators in the present case are Lk = − i Jk where the Jk are diﬀerential operators in θ and φ given in (5.85.4 Angular Momentum Eigenstates According to the theorem on page 107 above the eigenfunctions (5.86. 1 sin2 θ sinθ ∂ ∂θ (5. ∂θ ∂φ ∂2 ∂ ∂2 + cotθ + ( cot2 θ + 1 ) 2 .98) where Jclass = r × mv is the angular momentum and pr = mr is the radial momentum.101) 1 ∂2 r r ∂r2 and comparision with (5. One ﬁrst identiﬁes the Hilbert space H on which the generators Lk operate.97) Kinetic Energy Operator The kinetic energy of a classical particle can be expressed J2 p2 p2 = r + class 2m 2m 2m r2 (5. C∞ (S ).86.100) (5. 2 ∂θ ∂θ (5. These generators act on a subspace of C∞ ( ).

φ) in H which satisfy L+ Y (θ.103) iteratively for m = − 1. φ)f (θ. φ) = = 0 −i Y (θ.107) reads ∂ ∂ + i cotθ ∂θ ∂φ P (cosθ) ei φ i e±iφ ∂ ∂ ± i cotθ ∂θ ∂φ .108) m (θ. .109) using (5. − . 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. φ. One obtains in this way Y m (θ. φ). The norm in H is deﬁned through |f |2 = S2 dΩ f ∗ (θ. − 2.102) where dΩ = sinθ dθ dφ is the volume element of S2 . (5. .104. φ) . is indeed a Hilbert space. by carrying out the construction according to (5.5. φ) (5. m = − . (5.102) exists. and then determining the family {Y m (θ.106) The latter property is obviously satisﬁed for the chosen φ-dependence. φ). the so-called spherical harmonics. φ) = −i ( + m + 1)( − m)Y m (θ. normalizing these functions.87).105) ∆( . 5.107) (5.104) (5. − + 1.64) can be constructed then by seeking ﬁrst functions Y (θ. (5. . . φ) = 0 as well as L3 Y (θ. and to be admissable for description of quantum states. φ) = −i Y (θ. must be cyclic in φ with period 2π. φ) = 2 + 1 ( − m)! P 4π ( + m)! m (cosθ) eimφ . The function space endowed with this norm and including only functions for which the integral (5.105) explicitly. m) Constructing Y → Y →→ ··· Y ( + m)! (2 )! ( − m)! −1 − We like to characterize the resulting eigenfunctions Y m (θ. The eigenfunctions (5. 0 ≤ φ < 2π. (5. can be expressed L± = Accordingly.63.111) . } applying L− Y m+1 (θ. 5. . The raising and lowering operators L± = L1 ± i L2 . . 0 ≤ θ < π. m) = 1 −m J− 1 2 Y (θ. . φ) = ∆( . φ) L3 Y (θ. φ) (5.85–5. (5. φ) (5.4: Angular Momentum Eigenstates 111 θ.110) = 0 (5.

114) dθ sinθ |Y (cosθ)|2 = 1 0 (5. one for which π dθ sinθ |P (cosθ)|2 (5. is P (cos θ) = N sin θ as one can readily verify. . using the new integration variable x = cos θ. for example.116) implies |N |2 from which we conclude N = (−1) 1 · 3 · 5 · · · (2 − 1) (5.107). We note that (5. (5. 2nd Ed. i. 5.120) See.112 or Theory of Angular Momentum and Spin ∂ − ∂θ cotθ P (cosθ) = 0 (5. φ) = (−1) 1 1 = 1 [1 · 3 · 5 · · · (2 − 1)]2 (5.116) The integral appearing in the last expression can be evaluated by repeated integration by parts. One obtains. (5. (5. (5.113) 0 is ﬁnite. 5. New York.” by J.112) where we employed the deﬁnition of P (cosθ) in (5. The normalization factor N is chosen such that π (5.118) 2 +1 1 1 sin θ ei 4π (2 )! 2 ! φ . 5. π +1 dθ sin2 0 +1 θ = −1 +1 −1 dx (1 − x2 ) +1 = = .D..119) where the factor (−1) has been included to agree with convention1 . Jackson (John Wiley.114) yield |N |2 2 +1 1 2π 4π (2 )! π dθ sin2 0 +1 θ = 1.117) Accordingly.119) provide the following expression for Y Y (θ.106. A proper solution of this equation. .106. ”Classical Electrodynamics. 1975) . = = 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.e.115) holds.114.

5. (5.63. m = − 1.114. − deﬁned in (5. For m = (5. (5. ∂x −m−1 .123) P m+1 (x) ∂ x + (m + 1) 1 ∂x (1 − x2 ) 2 .103. namely Y (θ. To prove (5. φ). and can obtain now.126) Then holds.5.123). 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.e.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) . − 2. . according to (5.103. indeed.121) This expressions shows that the factor eimφ .123) leads to the expression P m (x) = −m m ∂ 1 ( + m)! (1 − x2 )− 2 (x2 − 1) 2 ! ( − m)! ∂x −m (5. use (5.124).124) we proceed by induction.127) ∂ −m−1 (x2 − 1) . m (cos θ) e (5. P m (x) = −1 × ( + m + 1) ( − m) × −(1 − x2 ) 2 1 (5.110) together with (5. 5. . Let us then assume that (5. 2 ! ( − m − 1)! ∂x −m−1 (5. φ). every application of L− reduces the power of eiφ by one.122) The latter identity can be written. the eigenfunctions Y m (θ. therefore. .64).125) which agrees with (5.119). The reader should note that the associated Legendre polynomials.124) reads P (x) = = (2 )! (1 − x2 )− 2 (x2 − 1) 2 ! (−1) 1 · 3 · 5 · · · (2 − 1) sin θ (5.110). we seek to determine the functions P m (cos θ) and. 5. through repeated application of (5. φ). Actually. employing again the variable x = cos θ. describes the φ-dependence of Y m (θ. 5.106) i e−iφ ∂ ∂ − i cotθ ∂θ ∂φ 2 + 1 ( − m − 1)! P 4π ( + m + 1)! 2 + 1 ( − m)! P 4π ( + m)! i(m+1)φ m+1 (cos θ) e = −i ( + m + 1)( − m) imφ .124) holds for m + 1. are real.124) called the associated Legendre polynomials. We want to demonstrate now that the recursion equation (5. i. as speciﬁed in (5.106). P m+1 (x) = −m−1 m+1 ∂ 1 ( + m + 1)! (1 − x2 )− 2 (x2 − 1) ..4: Angular Momentum Eigenstates 113 We have constructed a normalized solution of (5.

φ) = = determines ﬁrst a normalized solution 0 i f (θ. 5. φ) (5. In fact. it holds ∂2 (x2 − 1) = (2 )! ∂x2 and.132) which reduces the number of spherical harmonics which need to be evaluated independently roughly by half . 2 ! (2 )! ∂x2 1 (5.131) identiﬁes f (θ. that expression (5. appears not very interesting. therefore.134) (x) = 1 2 ! (1 − x2 ) 2 . using x = cos θ. φ) L3 f (θ. φ) as given in (5. hence. It holds. namely. yields a vanishing expression for m = − − 1 as long as is a non-negative integer. Y − +2 . we embark on this construction in order to prove (5.128) one can show that (5. which is equivalent to recursive application of L− . We ﬁrst determine Y − (θ.124) holds for m = . by repeated application of the operator L+ .127) reproduces (5. if L− Y (θ. Since (5. 5. In that case (1 − x2 ) is a polynomial of degree 2 and. the derivative ∂ 2 +1 /∂x2 +1 of this expression vanishes.106. φ) using (5. We want to test if the recursion (5.133) Since the term with the highest power of (x2 − 1) is x2 . (5. P − (5. Such construction reproduces the eigenfunctions Y m (θ. φ) choosing the proper sign. from such construction emerges an important symmetry property of Y m (θ. and constructs then the eigenfunctions Y − +1 . φ) .124).124). Constructing Y − → Y − +1 → ···Y m (θ. i. Y ∗ (θ. φ).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. expression (5. therefore.124) holds for m if it holds for m + 1. φ) An alternative route to construct the eigenfunctions Y of L− f (θ. therefore. φ) = (−1)m Y m −m (θ. it holds then for all m.130) (5.106.129) holds.124) and. P − (x) = 1 ∂2 (1 − x2 ) 2 (x2 − 1) .123) terminates for m = − .124) and. therefore.132).135) . φ) = Y − (θ. However.e. φ) = 0 (5.. etc. (5.

122.120) for Y obey the postulated relationship (5. (5. i.136) We note in passing that this expression and the expression (5. This follows from the identity ∂ ∂ − i cotθ ∂θ ∂φ One can use (5.106.106).110) applies the = ( + m + 1)( − m) Y m+1 (θ.132).124). φ) .139) = ∂ − m cotθ ∂θ P m (cos θ) . a sign consistent with the family of functions Y m constructed above. Obviously.129).5.136) is normalized and has the proper sign. 5.80) and (5.4: Angular Momentum Eigenstates Due to (5. (5. following closely the proof of eq. (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 (5.143) . We ﬁrst note that for m = − (5. One can readily verify that this expression provides a solution of (5. one can conclude for the associated Legendre polynomials ( − m)! ( + m)! = or P m+1 (cos θ) ∂ − m cotθ ∂θ P m (cos θ) ( + m + 1)( − m) ( − m − 1)! P ( + m + 1)! m+1 (cos θ) (5. (5.. φ) sin θ e−i φ = 0. According to (5.124) one arrives at Y (θ.136) to construct all other Y recursion eiφ ∂ ∂ + i cotθ ∂θ ∂φ Y m (θ. φ) = 2 +1 1 1 sin θ e−i 4π (2 )! 2 ! φ 115 − .137) m (θ. φ).e.123)] P m+1 (x) = −(1 − x2 ) 2 1 ∂ x − m 1 ∂x (1 − x2 ) 2 P m (x) .142) yields P − (x) = (−1) 1 (1 − x2 )− 2 (x2 − 1) = (1 − x2 ) 2 2 ! 2 ! (5. 5. (5. the expression (5. (5.138) Employing (5.f. (5.142) For this purpose we proceed by induction.140) Introducing again the variable x = cos θ leads to the recursion equation [c.

124) yields (−1)m ( − m)! P ( + m)! −m (x) = +m m ∂ (−1)m (1 − x2 ) 2 (x2 − 1) .142) one can state P (x) = ∂ (x2 − 1) . The construction beginning with Y − and continuing with Y − +1 .148) According to (5. Hence. given by (5. one notes that application of (5.144) ∂ x − m 1 ∂x (1 − x2 ) 2 +m m ∂ (−1)m (1 − x2 ) 2 (x2 − 1) . i.144) reads P m+1 (x) = +m+1 m+1 ∂ (−1)m+1 (1 − x2 ) 2 (x2 − 1) .e.147) the r. 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 .h. equivalently. are identical. (5. Accordingly.. (5.142) holds for m = − we veriﬁed that it holds for all m.150) .124) and (5.142) holds for m.s.. However.141) reads then P m+1 (x) = 1 (5. 2 ! ∂x +m (5.124).135).e. (5. Y − +2 .128) m + 1 → −m or. one can conclude P m (x) = (−1)m m (θ. P (x) = P 0 (x) are called Legendre polynomials. etc.116 Theory of Angular Momentum and Spin wich agrees with (5. 2 ! ∂x +m −(1 − x2 ) 2 Replacing in (5.142).h. also the associated Legendre polynomials determined this way. We then assume that (5.142) holds for m + 1 if it holds for m. 2 ! ∂x 1 (5. of which agrees with the r. etc.s. (5.145) Hence. According to both (5. Y −2 .142) and by (5.132). Since (5. φ) ( − m)! P ( + m)! −m (x) . of (5. 1 (1 − x ) ∂x +m (1 − x2 ) 2 1 (5.149) (5. yields the same eigenfunctions as the previous construction beginning with Y and stepping down the series of functions Y −1 . 2 ! ∂x +m+1 (5.146) i.106) this implies for Y The Legendre Polynomials The functions the identity (5.

t) = √ Comparision with (5. e. Jackson (John Wiley.157) = 0.154) can be written 1 = w(x. t = r2 /r1 . 92 To prove this property see. 45 of “Special Functions and their Applications” by N. 2nd Ed.N. pp. (5. Lebedev (Prentice-Hall.158) see.156) yields P (1) = 1 . the spherical harmonics for m ≥ 0 Y m (θ.150) and (5.D. 1975).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) . for example. ”Classical Electrodynamics. (5. 1. Englewood Cliﬀs. Using x = cos γ. In case q = 1 and |r2 | < |r1 | holds the identity2 ∞ 1 r2 = P (cos γ) (5. and r2 denotes the location of the charge. t) = √ 1 − 2xt + t2 ∞ 1 − 2 x t + t2 . and |r1 − r2 | = r1 (5. . =0 (5.155) P (x) t .4: Angular Momentum Eigenstates The ﬁrst few polynomials are P0 (x) = 1 . P1 (x) = x . t) is called a generating function of the Legendre polynomials3 . =0 (5. in case of x = 1 holds w(1. φ) = 2 + 1 ( − m)! (−1)m sinm θ × 4π ( + m)! × ∂ ∂ cos θ m (5.154) |r1 − r2 | r +1 =0 1 where γ is the angle between r1 and r2 . 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. N. 1965) which is an excellent compendium on special functions employed in physics.151) (5x3 Comparision of (5. For example.5. 2 3 1 1 = = 2 1−t 1 − 2t + t ∞ t . P3 (x) = 1 2 117 P2 (x) = − 3x) 1 2 2 (3x − 1) .152) and. The generating function allows one to derive useful properties of the Legendre polynomials.g. The spherical harmonics for m < 0 can be obtained using (5.132). . . accordingly. r1 is the point where the potential is measured.J. m ≥ 0 ∂xm (5. . (5..” by J. 2.. pp. New York.156) w(x.153) P (cos θ) eimφ .

. Due to cos(π − θ) = − cos θ the replacement r → −r alters P m (x) into P m (−x). Inspection of (5. 2.156) and (∂/∂t) to P (x)t = ∞ ∂w + (t − x) w = 0 . φ) (5. t) (1 − 2xt + t2 ) Employing (5. . φ) . holds P1 (x) ( + 1) P +1 (x) = = x P0 (x) (2 + 1) x P (x) − P −1 (x) (5.159) Using ∂lnw/∂t = (1/w) ∂w/∂t and multiplying (5. φ) = = 2 ( + 1) Y m (θ.165) since ∂ n /∂(−x)n = (−1)n ∂ n /∂xn .f. .118 For w(x. under inversion Y m (θ.163) (5. Inversion Symmetry of Y m (θ. Accordingly. If the spherical coordinates of r are (r. .167) (5. The spherical harmonics are eigenfunctions of the angular momentum operators J2Y J3 Y m (θ.160) this diﬀerential equation is equivalent ∞ (1 − 2xt + t2 ) =0 P (x) t + (t − x) =0 P (x) t = 0 (5. t) x−t = . Accordingly. π + φ) = (−1) Y m (θ. using P0 (x) = 1 [c.161) Collecting coeﬃcients with equal powers t yields 0 = P1 (x) − x P0 (x) ∞ (5. φ) m (π − θ.151)] allows one to determine P (x) for = 1. (5. Noting exp[im(π + φ)] = (−1)m exp(imφ) we determine Y Properties of Y m (θ. π + φ). ∂t 1 − 2xt + t2 (5.162) − (2 + 1)x P (x) + P −1 (x) ] t + =1 [( + 1) P +1 (x) The coeﬃcients for all powers t must vanish individually. φ) . φ) Under inversion at the origin vectors r are replaced by − r. t) holds Theory of Angular Momentum and Spin ∂lnw(x. φ) m (θ. φ) goes over to Y m (π − θ.164) which.159) by w(1 − 2xt + t2 ) leads to the diﬀerential equation obeyed by w(x. 1. θ. π − θ. ∂t P (x)t −1 −1 (5.168) mY m (θ. φ) m (θ. . π + φ).166) We want to summarize the properties of the spherical harmonics Y derived above.142) allows one to conclude P m (−x) = (−1) +m P m (x) (5. then the coordinates of − r are (r. φ). (5.

φ) (5. The spherical harmonics for m < 0 are given by Y −m (θ. P +1 (x) P1 (x) = x . i. φ)f (θ. φ) ∈ C∞ (S2 ) holds ∞ f (θ. where ( = 1.172) (5. The spherical harmonics form. φ) where J± = J1 ± iJ2 . 2. φ) = (−1)m Y ∗ (θ. φ) = 2 + 1 ( − m)! (−1)m sinm θ × 4π ( + m)! 2 ! × ∂ ∂ cos θ m (5. φ) = δ δm m .169) 3. The spherical harmonics Y 0 are Y 0 (θ. .178) 7. φ) Y m (θ. (5. uniquely deﬁned functions over the unit sphere S2 which are inﬁnitely often diﬀerentiable π 2π sin θdθ 0 0 dφY ∗m (θ.4: Angular Momentum Eigenstates 119 2.175) (5.5.) P0 (x) = 1 .176) = 1 +1 [ (2 + 1) x P (x) − are the Legendre polynomials. The spherical harmonics form an orthonormal basis of the space C∞ (S2 ) of normalizable. 4π (5. For the Legendre polynomials holds the orthogonality property +1 dx P (x) P ( (x) = −1 2 δ 2 +1 . (5. φ) .. m (5.179) . φ) . in fact. φ) C = =0 m=− π C sin θdθ 0 m Y m (θ. φ) = 2 +1 P (cos θ) .e.173) m+1 (θ. 5. The spherical harmonics are given by the formula Y m (θ.171) m = dφ Y ∗ (θ. m 4.170) (5.174) m ≥ 0 P (cos θ) eimφ . P −1 (x) ] (5. . 6. φ) 2π 0 (5. for any f (θ. The spherical harmonics obey the recursion relationships J+ Y J− Y m (θ. a complete basis of C∞ (S2 ). except for the factor exp(imφ). .177) Note that the spherical harmonics are real. φ) = = ( + m + 1)( − m) Y ( + m + 1)( − m) Y m+1 (θ. φ) m (θ.

189).180) 9. t)2 = −1 .184) Y22 Y21 Y20 together with (5. expand the result in powers of t and equate the resulting powers to those arising on the r.179) using the generating function w(x. φ) = δm0 2 +1 .187) h(r) Y m (θ.186) (5.156).188) Exercise 5.86). The spherical harmonics for m (θ.120 8.177). φ).183) (5.185) (5. 1.189) evaluate the integral on the l.4. (5. π + φ) .4.s.87).s. (5. 4π (5. The spherical harmonics for θ = 0 are Y m (θ Theory of Angular Momentum and Spin = 0. t) stated in (5. Exercise 5..h. Exercise 5. The spherical harmonics obey the inversion symmetry Y 10. (5. (5. φ) = 1 ∂2 r − r ∂r2 ( + 1) r2 h(r) Y m (θ.3: Construct all spherical harmonics Y3m (θ. For the Laplacian holds 2 = = = (5.181) = 0. φ) . of (5. 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. .1: Derive expressions (5. =0 t + −1 dxP (x) P (x) .4. For this purpose start from the identity +1 ∞ +1 dx w(x.2: Derive the orthogonality property for the Legendre polynomials (5. φ) = (−1) Y m (π − θ. 11.h.182) Y11 Y10 = = − (5.

= .196) C∞ (S ) = = X . . We denote the image of a function f (θ. Comparision with (5. − + 1. = . .. are characterized completely if one speciﬁes the transformation of any Y m (θ. }. For this purpose we consider the subspaces of C∞ (S ) X = [ {Y ..m D(ϑ) m.e. called the irreducible representation. ...5 Irreducible Representations We will consider now the eﬀect of the rotational transformations introduced in (5. (5. φ) exp − ϑ · J . = 0. orthonormal basis for the function c m Y m (θ. . φ) (5.. m = − .. φ) m.193) shows ∞ = − ..195) i. m Y m (θ. φ ) exp m One can also represent C∞ (S ) C∞ (S ) = [ {Y .197) The subspaces X have the important property that (i) they are invariant under rotations exp − i ϑ · J .5: Irreducible Representations 121 5. .191) i − ϑ·J f (θ .195). }] (5.. φ) i − ϑ·J f (θ. .. The transformations exp − i ϑ · J . . φ) (5. . (5.e. with elements given by (5. . These coeﬃcients can be considered the elements of an inﬁnite-dimensional matrix which provides the representation of exp − i ϑ · J in the basis {Y m . . ˜ f (θ.193) where [ { } ] denotes closure of a set by taking all possible linear combinations of the elements of the set. ∞. φ) = exp Since the spherical harmonics Y space C∞ (S ) one can expand ˜ f (θ.− + . φ).e. . 1. = − .− + . φ) ˜ Y m (θ. and (ii) they form the lowest dimensional sets X obeying C∞ (S ) = X which have this invariance property. φ) ∈ C∞ (S ). m = S2 i dΩ Y ∗m (θ.42). . i. . m . φ ) (5. exp − i ϑ · J (X ) = X .192) = S2 dΩ Y ∗ (θ . φ) c = . (5. (5.. on functions f (θ. φ) D(ϑ) = ... . φ) ˜ under such rotational transformation by f (θ. ∞. i. We like to argue that the matrix representing the rotational transformations of the function space C∞ (S ).190) provide a complete.5.194) Y m (θ. . }] .m m m (θ. assumes a particularly simple form. φ) . exp − i ϑ · J . . therefore. if one speciﬁes the functional form of the coeﬃcients [D(ϑ)] m .

201) From expression (5. 3..202) D(ϑ) = (2 + 1) × (2 + 1) . If one orders the basis according to the partitoning (5.n2 . n The action of J− on Y m produces Y m−1 or.56. Application of (5. φ) exp m i − ϑ·J Y m (θ.195) [D(ϑ)] m. (5. in case m = − .. J− ] = 2 J3 . J− . with blocks of dimension 1. 5.n2 . One proceeds by expanding ﬁrst ϑ · J in terms of J+ . which follow readily from (5.199) [J+ . . Similarly.n3 n n n Accordingly.e..197) of C∞ (S ) the matrix representation of D(ϑ) assumes a block-diagonal form. The exponential operator exp − i ϑ · J + ϑ3 J3 )n . One = contribute can conclude then that in the expansion (5. i. . φ) . 5. 1×1 3×3 . i. i. produces 0. Since any additive term n n n cn1 . one needs to consider the action of J+ 1 J− 2 J3 3 on the subspaces X .200) leaves X invariant. (5. (5. J± ] = ± J± (5.198) 2 2 where ϑ± = ϑ1 ( 1 ϑ− J+ 2 + 1 2 ϑ+ J− ϑ2 .e.194) only spherical harmonics with or. Since the Y m are n3 eigenfunctions of J3 the action of J3 reproduces the basis functions of X .199) allows one to express exp i − ϑ·J = n1 . one concludes for the matrix elements (5. and J3 1 1 ϑ · J = ϑ− J+ + ϑ+ J− + ϑ3 J3 (5. .e... .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 . leaves X invariant. equivalently..200) in (5.n3 J+ 1 J− 2 J3 3 . one can conclude that J+ leaves X invariant.200) of the rotational transformations one can also conclude that the X are the smallest invariant subspaces of C∞ (S ). . can be expanded in powers One employs then the commutation properties [J3 . J− 2 also leaves n1 X invariant. the transformation exp − i ϑ · J leaves X invariant as well.65). . m = δ S2 dΩ Y ∗ (θ.

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

φ) = S2 dΩ Y ∗m (θ. followed by transformation (i). followed by J2 in the original frame. (5.e. the axis x3 is deﬁned in the coordinate frame which is related to the original frame by the consecutive rotations (i) and (ii).195) by ˜ Y m (θ.h. (i) and (ii).e. i (5. we can write e− i −1 = S eA S −1 .208) αJ3 βJ2 = e− αJ3 e− i βJ2 e i (5. β. m [D(α.207) are equivalent. of (5. Accordingly. Obviously.207) which replaces J2 by the inverse transformation (i). β.s. the transformation from the rotated frame to the original frame. one can replace (5. γ)] = . γ will be referred to as Euler angles. For this purpose we notice that J2 can be expressed through the similarity transformation J2 = e− i αJ3 J2 e i αJ3 (5. is e− i γJ3 = e− i βJ2 e− i αJ3 e− i γJ3 e i αJ3 e i βJ2 (5. The axis x2 is deﬁned in the coordinate frame which is related to the original frame by rotation (i). β. i. i. For any similarity transformation involving operators A and S holds eS A S Accordingly.. the l.203) has the disadvantage that it employs rotations deﬁned with respect to three diﬀerent frames of reference.212) .203) For any ϑ ∈ R one can ﬁnd Euler angles α. in analogy to (5.211) Using (5. in terms of rotations deﬁned with respect to the original frame.m m. φ) e− γJ3 − i βJ2 − i αJ3 e e .209). can then be written e− i βJ2 − i αJ3 e = e− i αJ3 e− i βJ2 (5.209) The ﬁrst two rotations in (5..203). γ)] m. φ) (5. The Euler rotation replaces exp − i ϑ · J by e− i γJ3 − i βJ2 − i αJ3 e e .194. however.124 Theory of Angular Momentum and Spin The angles α. β. γ ∈ R such that exp i − ϑ·J = e− i γJ3 − i βJ2 − i αJ3 e e (5.203) can be expressed.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 (5.203).210) The third rotation in (5.205) Y m (θ. i.e. 5.s. and the r.206) The expression (5. m i Y m (θ.. φ) [D(α. the transformation from the original frame to the rotated frame.204) is satisﬁed. We will demonstrate that (5. (5.h.

This allows one to express the rotational trasnformation (5. Examples are the other ﬁve members of the lepton family to which the electron belongs. one can write [D(α. the electron e and the electron neutrino νe of the ﬁrst generation.218) Deﬁning the so-called Wigner rotation matrix dmm (β) = one can express the rotation matrices [D(α. 5. (5. in addition. (5.220) We will derive below [see Eqs.205.213) i. the self-adjointness of the operator J3 one can state dΩ Y ∗ (θ. φ) e− S2 i γJ3 Y m (θ.216) Using. m [D(α.206) by ˜ Y m (θ. φ) = S2 dΩ f (θ.219) = e−iαm dm m (β) e−iγm δ ( ) . to redeﬁne the three rotations in (5. m = e−iαm S2 dΩ Y ∗ (θ.215) beneﬁts from the choice of Euler angles for the parametrization of rotational transformations. φ) [D(α. but many other elementary components of matter are endowed with spin-like properties.64) yields dΩ f (θ.7: Spin 1 2 and the group SU(2) 125 Multiplication from the left with (5.210) yields the simple result e− i γJ3 − i βJ2 − i αJ3 e e = e− i αJ3 e− i βJ2 e− i γJ3 . γ)] = . m i Y m (θ.203) with respect to the original (unprimed) frame one simply needs to reverse the order of the rotations. φ) (5. φ) i i (5. γ)] m.214) e− γJ3 = S2 dΩ Y ∗m (θ. β.219). φ) e− m i βJ2 Y m (θ. φ) e− m i βJ2 Y m (θ. m ( ) S2 dΩ Y ∗ (θ. the muon µ and its neutrino νµ of the second . (5.m m. Best known is the spin of the electron. γ)] m.e. β. φ) e− m S2 i αJ3 f (θ. The eigenvalue property (5. φ) = e−im α S2 dΩ Y ∗ (θ. (5.215) The evaluation of the matrix elements (5. β.217) Accordingly. φ) . φ) e−imγ . γ)] m. β. (5. φ) e− αJ3 e− βJ2 Y m (θ..5. φ) .309. φ) f (θ. 5. φ) Y m (θ.310)] an explicit expression for the Wigner rotation matrix (5. m (5. φ) e−iγm δ (5. 5.7 Spin 1 2 and the group SU(2) The spin describes a basic and fascinating property of matter.

particles which mediate the strong and the electro-weak interactions. If we identify the states χ+ and χ− with the basis of a Hilbert space.221) One can show readily that this set forms a group with the groups binary operation being matrix multiplication. atoms and molecules. U U † = U † U = 1 detU = 1 } 1. 3 representing a 2 × 2 matrix. as long as |c+ |2 + |c− |2 = 1. (5. Spin. It appears to be rather impossible for a Physicist not to be enamored with the spin property. k = 1. 2. the gluon. The important feature is that all U ∈ SU(2) can be parametrized by an exponential operator U = exp − i ϑ · S (5. and the condition detU = 1 requires the Sk to have vanishing trace. two of which (in certain linear combinations) make up the vector mesons which carry spin 1. Because of the unitarity condition U † U = 1 which are really four equations in terms of real 1 quantities. then allowed + − symmetry transformations of spin states are described by 2 × 2-matrices U with complex-valued matrix elements Ujk . Conservation of the scalar product under symmetry transformations requires ∗ the property U U † = U † U = 1 where U † denotes the adjoint matrix with elements U † jk = Ukj . How can the elements of SU(2) be parametrized. i. So do the three generations of six quarks.126 Theory of Angular Momentum and Spin generation. We may ﬁnally mention that the spin is at the heart of many properties of condensed matter systems. 2.1 and spin. There are also the mediators. there are together ﬁve conditions in terms of real quantities to be met by the matrix elements and. there is nothing more elementary to matter than the spin property. carry other spin-like properties which together. j. like superconductivity and magnetism. 1 We will specify for the transformations considered detU = 1. which might be the reason why consideration of rotational symmetry is so often fruitful in the study of matter. [Sk ]mn = ([Sk ]nm )∗ . c± ∈ C. In any case. a relationship. We will ﬁnd that the spin in its transformation behaviour is closely related to angular momentum states.1 systems 2 2 can be related to two states which we denote by χ+ and χ− . however. eight real numbers to specify the matrix elements. The transformation matrices are then elements of the set SU (2) = { complex 2 × 2 matrices U . There . As complex 2 × 2 matrices one needs. the photon.2 .222) where the vector S has three components. the quarks. have properties beyond those of single spins. The particles mentioned. four real and four imaginary parts of Ujk . the tau τ and its neutrino ντ of the third generation and their antiparticles carry a 1 spin. leaving its imprint on the properties of nuclei. hence. in principle. the spin of the electron is likely the most important property in Chemistry. This speciﬁcation implies that we consider transformations save for overall factors eiφ since such factors are known not to aﬀect any observable properties. We will consider ﬁrst only the so-called spin– 1 .e. in fact. and three of which make up the baryons which carry spin. the two W± and the 2 2 Zo . one for each matrix element of U † U . e. The presence of this property permeates matter also at larger scales than those of the elementary particles mentioned. and which carry spin 1. in which we deﬁne the scalar product between any state |1 = a+ χ+ + a− χ− and |2 = b+ χ+ + b− χ− as 1|2 = a∗ b+ + a∗ b− . k = 1. the degrees of freedom of the matrices U are three real quantities. each component Sk . Such systems can also assume any linear combination c+ χ+ + c− χ− .g. generalizing then further below. and because of detU = 1.3 . One can show that the unitarity condition requires these matrices to be hermitian.

k=1 σ σ aj bk + j>k aj bk = 3 j. i. (σ · a) (σ · b) = a · b 1 + iσ · (a × b) 1 (5.223) where σk .e. but their comonents must not necessarily commute with each other.k=1 j<k + σ j σ k aj bk . it holds.7: Spin 1 2 and the group SU(2) 127 exist three such linear independent matrices. 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. (5. 1 σj σk = − σk σj for j = k which can also be readily veriﬁed. hermitian 2 × 2–matrices A can be expandend. 3 are the Pauli matrices σ1 = 0 1 1 0 . A = x σ 1 + y σ 2 + z σ3 .5. In fact. (5. According to this property the Pauli matrices generate a 3dimensional Cliﬀord algebra C3 . 1 i. z A = In fact. it might hold aj bk − bk aj = 0.224) provide a basis in terms of which all traceless.31) of the group SO(3). ∈ R . y. The proof of this relation rests on the commutation relationship and the anti-commutation relationship (5. 5.225) σ (5.226) The Pauli matrices satisfy very special commutation and anti-commutation relationships.227) which is essentially identical to the Lie algebra (5. σk ]− = 2i jk z x − iy x + iy −z . One can readily verify the commutation property σj σk − σk σj = [σj . namely. y. Cliﬀord algebras play an important role in the mathematical structure of physics. x. 2 S3 = 1 σ3 2 (5.. k = 1. the simplest choice being S1 = 1 σ1 . one obtains (σ · a) (σ · b) = 3 j 2 j=1 (σ ) aj bj 3 j k j. σk ]+ = 2δjk 1 .227. σj 2 (5.231) . At this point we will state a useful property.229) = 1 . z. b are vectors commuting with σ.e.229) and avoids commuting the components aj and bk . (5. e.k=1 σ σ 3 j k j.g. 2 S2 = 1 σ2 . σ3 = 1 0 0 −1 . they are associated with the important fermion property of matter. Any such A can be expressed in terms of three real parameters x. σ2 = 0 −i i 0 .230) where a. The Pauli matrices obey the following anti-commutation properties σj σk + σk σj = [σj .228) (5. 2. (5.224) Algebraic Properties of the Pauli Matrices The Pauli matrices (5.

1 1 1 ( + 1) χ± 2 2 .233) This result together with (5.k=1 j>k 1 2 3 j. however. we like to draw in this respect also on a close analogy to angular momentum states which developes 2 2 2 if one considers the operators S 2 = S1 + S2 + S3 and S3 .232) − σ k σ j ) (aj bk − ak bj ) . The commutation property (5.k=1 j>k i σl (aj bk − ak bj ) = i =1 σ a×b . S3 χ± = ± 1 χ± 2 . The two remaining terms yield using σ j σ k = −σ k σ j for j = k and altering ‘dummy’ summation indices 3 j. let say around the x2 –axis by 360o . In the special case a = b ∈ R holds (σ · a)2 = a 2 1 . spin changes sign.236) The matrix elements of this 2 × 2 operator can be labelled by the basis states χ+ and χ− .238) S 2 χ± = .222).k=1 (σ σ j>k σ k σ j (aj bk − ak bj ) = (5. (5. yields a · b. however.235) which is identical to that of the generators of SO(3). i. Well almost like it. This relationship emerges.230).222) as a rotation vector and you rotate once. S ] = i k m Sm (5. 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.8 Generators and Rotation Matrices of SU(2) Sofar there is nothing which relates the transformations U . (5.237) + 1)1 and. For this purpose we choose to replace (5. 1 (5. Noticing the idempotence of the Sk ’s 2 Sk = 3 one obtains S 2 = 4 1 = 1 1 1 2(2 1 4 1 0 0 1 (5.s. hence. through the algebra obeyed by the operators Sk [Sk .h. to rotations in space. (5.k=1 j>k σ j σ k (aj bk − ak bj ) = 1 2 3 j. (5.e.227) leads to 3 3 jk j.234) 5.231) proves (5.k=1 j>k σ j σ k (aj bk − ak bj ) − 1 2 3 j k j. We like to derive this result now by evaluating the transformations of SU(2) explicitly.128 Theory of Angular Momentum and Spin Using (σ j )2 = 1 the ﬁrst term on the r. since there is a slight diﬀerence: when you interprete the 3-component ϑ in (5.222) by the Euler form exp(−iγS3 ) exp(−iβS2 ) exp(−iαS3 ) . only a 720o rotation leaves the spin invariant.

9: Constructing Spin States with S > 1 2 129 This result implies that the states χ± behave in relation to the generators of SU(2) like an angular momentum state | 1 ± 1 in relation to the generators of SO(3).243) We note the property of this rotation matrix 2 dmm (2π) (1) = −1 1 .242) The expressions in brackets [.5. 1. identical particles any number of which can exist in the same state χ+ and χ− .] can be identiﬁed with the Taylor expansion of the cos.239) where we made use of the notation (5. − + 1. 3 . namely. .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. (5. In order to determine the Wigner matrix in (5. 2. (5.219. . 1 .244) i. 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.239) in the notation (5. . 2 2 We obtain with this notation for the transformations (5.220) are [D(α. ( − β S2 )2n = (−1)n ( − β S2 )2n+1 = (−1)n β 2 β 2 2n 1 1 0 1 −1 0 2n+1 (5.240) . i.9 Constructing Spin States with Larger Quantum Numbers Through Spinor Operators In this section we like to demonstrate. use the label 2 2 | 1 ± 1 for the states χ± . rotation by 360o changes the sign of the spin state. following Jourdan and Schwinger. i. .245) 5.e.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 . . We may. (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. no two such particles can exist in the same state.219). therefore. The idempotence property of Sk yields particularly simple expressions for these powers. m = − . .5. m = cos β e−i( 2 + 2 2 γ α sin β ei(− 2 + 2 2 α γ − sin β ei( 2 − 2 2 α γ cos β ei( 2 + 2 2 α γ .e. . For this purpose one needs to determine the powers of −iβS2 .e. The complete matrix elements (5. . that states | m for higher quantum numbers = 0.239 ) one expands the exponential operator exp(−iβS2 ).. (5. . One cannot consider the 1 entities carrying the spin– 2 to be particles in the ordinary sense for it can be shown that spin– 1 2 particles have fermion character. γ)] m. β.

b † ζ ζ =0 (5.247) (5. The boson character of the operators b† = (b† . + − For example. .243) the creation operators in a coordinate system rotated by an angle β around the y–axis are b+ † = cos β β † b + sin b† 2 + 2 − . b− )T is expressed through the commutation relationships + − (ζ. b ζ = b† . therefore. b ζ = 0 b ζ. b± |Ψ0 = 0 (5. .251) One can show using this property and (5.254) We will show below that these states are orthonormal and form spin states with higher quantum numbers. also operators of the type x b† = x+ b† + x− b† + − . b− † = − sin β † β b + cos b† .e.247. i. −j + 1.248) that bζ relationships † † b ζ. (5. (5. m = −j. m) † and bζ obey the commutation (5. 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.e. .246) The corresponding adjoint operators are denoted by b+ and b− . . .253) = δζζ . j. 3 .252) (5.249) In the following will refer to b† = (b† . ζ = +. We consider. x∗ b = x∗ b+ + x∗ b− + − (5. b− )T as spinor creation and annihilation + − operators. b ζ The States |Ψ(j. j = 0. 1.130 Theory of Angular Momentum and Spin Deﬁnition of Spinor Creation and Annihilation Operators We present the states χ± through creation operators b† which when applied to a formal vacuum ± state |Ψ0 generate χ± . 1 . 2 2 . b † ζ For the vacuum state |Ψ0 holds = δζζ . 2 + 2 − † (5.248) bζ . . m) = b† − j−m (j + m)! (j − m)! |Ψ0 .250) which for x∗ x = x∗ x+ + x∗ x− = 1 represent spinor operators in an arbitrary reference system. using (5. b ζ = b ζ. b† |Ψ0 = χ− − . . These operators are associated with a given spatial reference system. b† )T and b = (b+ . −) bζ . i. b† |Ψ0 = χ+ + .5. b† )T and b = (b+ .

For this purpose we consider the following polynomial of creation operators N j+m j−m f (b† ) = + n=0 cn b† + n . in b† and b† .10: Algebraic Properties of Spinor Operators 131 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.5.256) In the special case |state = |Ψ0 this reads using (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 (5.257) motivate us to determine the commutator (5. . 5. f b† Obviously.260) . f (b† ) + (5.257) which for (5.258) Equations (5. we need to evaluate b+ . + b+ . An example + − is the monomial b† which generates the states (5. possibly inﬁnite power series.256) and (5. b† + b+ .255) is b+ .259) The property is obviously true for n = 0. (5. that the annihilation operators play a role similar to the diﬀerential operator in calculus. b† + n = n cn b+ . We will derive the result. b† + n . (5. b† + n b† + b† + b† + + n n n b+ .247.249) b+ f (b† ) |Ψ0 = + b+ . f (b† ) |state + f (b† ) b+ |state .248) we obtain b+ . b† + − well-known for the quantum mechanical harmonic oscillator. + + (5. If it is true for n then using (5.248) and from (5. We will show that for this commutator holds n b+ .254). (5. f (b† ) |Ψ0 .255) In order to determine how this operator is modiﬁed by multiplication from the left by b+ we note b+ f (b† ) |state = + b+ . b † + = n b† + n−1 b† + b † + + n = (n + 1) b† + .249). b† + n+1 = = b+ . b† + = n b† + n−1 .

266) we compare the terms x+ b† + j+m φjm (x) Ψ (j. One can ﬁnally conclude for polynomials (5.247) the more general property for polynomials f (b† . f (x)] g(x) = (∂k f (x))g(x) or [∂k . φjm (x) Ψ (j. f (b† .267) exp xb† |Ψ0 is called a generating function of |Ψ(j.. + − (5. b† ) + − = ∂ ∂b† ζ f (b† .. (5. f (x)] = ∂k f (x) .265) Generating Function of the States |Ψ(j. (5.132 Theory of Angular Momentum and Spin i. Equation (5. b† ) |Ψ0 = + − ∂ ∂b† ζ f (b† .262) Because of the commutation relationships (5.257) we can conclude in particular bζ f (b† . + − (5. (5.e.268) .259) holds for all n ∈ N. b† ) in b† and b† holds (ζ = +. b† ) |Ψ0 .263) According to (5. = f (b† ) + (5. f (b† ) + where f (x) = df dx . m) φjm (x) = j−m xj+m x− + (j + m)! (j − m)! . one can prove b− . In order to derive (5. m) We want to prove now the property ∞ j exp xb † |Ψ0 = 1 j=0.263) corresponds to the product rule of calculus ∂k f (x)g(x) = (∂k f (x))g(x) + f (x)(∂k g(x)) which can be written ( ∂k f (x) − f (x)∂k ) g(x) = [∂k .. By induction we can conclude that (5. m) = x− b† − j−m (j + m)! (j − m)! |Ψ0 .266) where xb† has been deﬁned in (5. −) + − + − bζ . 2 .264) This demonstrates the equivalence of the spinor operators bζ and the derivative operation. f (b† ) − = f (b† ) − (5. the property holds also for n + 1.250) and where φjm (x) represents the function of the two variables x+ and x− closely related to |Ψ(j.1. m) .261) Similarly.255) b+ . b† ) . m) m=−j (5.

270) N 2j j → s=0 j−m=0 → m=−j . 2 2 leads to ∞ j ∞ φjm (x) |Ψ (j.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.. 1.269) (5..272) The summation over s can be written in terms of j and m using (5. . (5. .5.10: Algebraic Properties of Spinor Operators with the s–th term in the binomial expansion of (a + b)n n! an−s bs .1. m) m=−j = = 1 x+ b† + x− b† + − (2j)! 2j 1 xb† |Ψ0 . .270) s x+ b† + N −s x− b† − s |Ψ0 .. (2j)! 2j |Ψ0 (5..1. 2. 1 xb† u! |Ψ0 (5. m) = = |Ψ0 (N − s)! s! N −s 1 N! x+ b† x− b† + − N ! (N − s)!s! N −s 133 (5.275) = exp xb† |Ψ0 . 3 ..e.. .. ... (5. 1 .. 2 ∞ 1 xb† 2j! u 2j |Ψ0 = u=0. chosing the summation index j = 0. 2. s! (n − s)! Deﬁning j+m = N −s j−m = s. 1 . .274) Summing this expression over 2j = 0. m) 1 j=0.. i. m=−j = j=0.1. 2 .273) Change of the summation indices allows one to conclude j φjm (x) Ψ (j. one obtains φjm (x) Ψ (j. 1.

e.s. Orthonormality of the States |Ψ (j.283) (5.m φjm (x∗ ) φj m (y) Ψ (j. (5. m holds. b† ] = 0 allow one to state that for + − + any polynomial f (b+ . [b† . hence.281) and.j . According to (5. b† ] = 0 and [b− .278) † ∗ which allows us to replace the l. 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. (5. m) |Ψ j . m) We want to show now that the states (5. b− )e(yb ) |Ψ0 = f (y+ .280) The commutation properties [b+ . that Ψ (j.h. (5. The generating function allows one to derive various properties of the states |Ψ (j. m) and will be used for this purpose below. b− ) holds † † f (b+ .264) implies † † bζ e(yb ) |Ψ0 = yζ e(yb ) |Ψ0 .279) One can generalize this to † † s (bζ )s e(yb ) |Ψ0 = yζ e(yb ) |Ψ0 .254) are orthonormal. y− )e(yb ) |Ψ0 (5. For this purpose we consider the inner product † † e(xb ) Ψ0 |e(yb ) Ψ0 = δjj δmm (5.134 Theory of Angular Momentum and Spin which concludes our derivation.282) where we deﬁned x∗ y = x∗ y+ + x∗ y− . In order to evaluate the ∗ b) (yb† ) operator e(x e we notice that the derivative property (5. we can conclude † † ∗ e(xb ) Ψ0 |e(yb ) Ψ0 = e(x y) (5. (5. of (5.277) by Ψ0 |e(x b) e(yb ) Ψ0 .277) To evaluate this expression we ﬁrst notice † e(xb ) † = e(x ∗ b) . m .276) = j.m. i. b− ] = 0. therefore.249) for any non-vanishing integer s holds + − Ψ0 |bs Ψ0 = 0 and. m) |Ψ j .284) .

j . 2 . b† bm n n n. etc.m 135 = e(x ∗ y) .m .5. In our present notation the matrix elements < ζ |σk | ζ > for ζ = ± correspond to the matrix elements < ζ |σk | ζ > for σ = ± 1 .224).287) ζ 2 ζ.n . States 2 We want to demonstrate now that the states |Ψ (j. m) as given in (5.m. m j. . One deﬁnes corresponding operators Jk through 1 Jk = b† < ζ |σk | ζ > bζ . b† bm n n n.288) The three operators obey the Lie algebra of SU(2) [Ji .277) φjm (x∗ ) φj m (y) Ψ (j.m.224). 5. < + |σ1 | − > = < − |σ1 | + > = 1. using < + |σ1 | + > = < − |σ1 | − > = 0. Jj ] = i This property is derived as follows [Ji . (5.m ijk Jk . 1.276).n .285) Following steps similar to those in Eqs. (5.268–5. 3 denote the Pauli spin matrices (5.275) one can show e(x ∗ y) = j. The 2 operators are explicitly. m) |Ψ j .ζ where σk . bm bm } n n = 1 4 < n |σi | m >< n |σj | m > {b† bm δmn − b† bm δm n } n n n.254) are eigenstates of operators J 2 and J3 with eigenvalues j(j + 1) and m where J 2 and J3 are representations of the spin 2 2 2 operators S 2 = S1 + S2 + S3 and S3 deﬁned in 5.m.m +b† b† .223.282) and (5.m φjm (x∗ ) φjm (y) . 2.289) < n |σi | m >< n |σj | m > {b† bm .286) from which follows immediately the orthonormality property (5. J1 = J2 = J3 = 1 † b b− + b† b+ − 2 + 1 † b b− − b† b+ − 2i + 1 † b b+ − b† b− .. (5.n .10: Algebraic Properties of Spinor Operators and comparing (5. . 3 . Jj ] = = 1 4 1 4 < n |σi | m >< n |σj | m > b† bm . (5. k = 1. .m. 1 New Representation of Spin 0. (5. − 2 + (5.

(5.m < n |2i n. J1 + iJ2 ] = −i [J2 .136 = 1 4 − = = 1 4 1 4 Theory of Angular Momentum and Spin < n |σi | m >< m |σj | m > b† bm n n.m < n |[σi . 2 J 2 = J+ J− + J3 − J3 = J3 (J3 − 1) + J+ J− (5. To this end we note 2 2 2 J 2 = J1 + J2 + J3 = 1 1 2 J+ J− + J− J+ + J3 2 2 . (5. − (5. (5.292). (5. J1 ] + i [J1 .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 . (5. J+ ] = [J1 − iJ2 .m 1 4 < n |σj | m >< m |σi | 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.m.294) .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. m) are eigenstates of J 2 and J3 .m.293) .297) . (5.291) Here we have deﬁned J+ J− = J1 + iJ2 = J1 − iJ2 = b† b− + = b† b+ .295) The last result together with (5.290) |Ψ (j.m.292) The commutation relationships (5.m ijk Jk .289) yield [J− .k ijk ijk σk | m > b † bm n = i = i 1 2 < n |σk | m > b† bm n n. m) as Eigenstates of J 2 and J3 We wish to show now that the states |Ψ (j. σj ]| m > b† bm n n.

5.301) Obviously. [Ij . .e. 1. m) J− |Ψ (j. m) = = (j + m + 1) (j − m) |Ψ (j.10. . . (5. I3 = (b+ b1 + b+ b2 ) 2 2 2 2 2i 2 1 satisfy the Lie algebra of SU(2). n2 . m) . m + 1) (j + m) (j − m + 1)|Ψ (j. the number of states to the possible energy eigenvalues. and eigenstates of k with eigenvalues j. m) m |Ψ (j. . n1 = 0.. .303) Exercise 5.298) (5. 1 2 k (a) Show that the eigenstates are given by 1 |n1 . .e. . i. (5. m . λ. i. . .304) (5. .305) = = j ( j + 1 ) |Ψ (j. Ik ] = i jk I . 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. State the corresponding eigenvalues and the degree of degeneracy. using operators I± = I1 ± iI2 2 2 2 and the subspace {|n1 .302) (5. ˆ respectively. n = n1 + n2 ﬁxed. 2. One can then conclude J 2 |Ψ (j. can be written b† + b − b † − b + = = One can then state J2 = 1 (b† + b+ )2 + b† + b+ b† − b− + b† − b− b† + b+ 4 + (b† − b− )2 + 2 b† + b+ + 2 b† − b− Deﬁning the operator 1 ˆ k = ( b† + b+ + b† − b− ) 2 one can write ﬁnally ˆ ˆ J2 = k ( k + 1 ) b† . Construct. m I1 = where λ = 0. I3 ||λ. . b † + b + + b † + b+ b † − b − 1 1 b† + b + + b † + b + b † − b− + b † − b − b † + b + . m = λ||λ.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 − 1) . .300) (5. [bj . m = −λ. m) One can furthermore derive readily J+ |Ψ (j. 1 2. . I2 = (b1 b2 − b+ b1 ) . n} eigenstates of I 2 = I1 + I2 + I3 and of I3 I 2 ||λ. 2 2 137 (5.299) (5. Show λ = n1 +n2 n1 +n2 2 ( 2 + 1) and m = 1 2 (n1 − n2 ).10: Algebraic Properties of Spinor Operators The last term on the r. b+ ] = δjk .h. (b) Show that the three operators 1 + 1 1 + (b1 b2 + b+ b1 ) .254) are eigenstates of + b+ and − b− with eigenvalues j + m and j − m. −λ + 1.s. j = 1. m) J3 |Ψ (j. 2 . m = m||λ. the states (5. b† .

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). m) in the original coordinate system by j |Ψ j. For this purpose we note that the states |Ψ(jm) in a rotated coordinate system according m to (5.307) shows that the elements of the rotation matrix can be obtained by binomial expansion of b † and b † in terms of b† and b† . (j) (5. The combination of σ. m ) are related to the + − states |Ψ (j. for example.309) . 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) .306) and (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 (5.307) Comparision of (5. m = m=−j dm m (β) |Ψ (j.243). + − One may expect that these two conditions restrict both σ and σ.5.307) 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. In case j = 1 this expression yields. m = b† + j+m b† − j−m (j + m )! (j − m )! |Ψ0 (5. σ values which yields (b† )j+m (b† )j−m is + − then σ = j + m − σ. However. On the other side the states |Ψ (j.306) where b † and b † are given by (5.251).138 Theory of Angular Momentum and Spin 5. . both conditions are satisﬁed for σ = j + m − σ. The prefactor of (b† )j+m (b† )j−m which according to (5.310) it reduces to (5.254) are |Ψ j.308) The latter sum involves terms (b† )j+m (b† )j−m for σ + σ = j + m and 2j − σ − σ = j − m.306.

314) leaves (5.314) in case of integer j–values. i. Hence. of (5. b+ † = −cos β † β b+ − sin b† 2 2 − . that the the representation (5. 2 2 − (5. is the expected element of SO(3).308) and had the form (5.312) The choice of this transformation derives from a property shown further below. x3 ) ∈ R .310) the similarity transformation ˜ A = U† A U where U is the 3 × 3 unitary matrix which establishes the transformation 1 1 i √ ( −x1 − ix2 ) − √2 − √2 0 x1 2 x3 0 1 . in case of j = 1. the orthogonal 3 × 3 matrix which describes a rotation around the x2 –axis. e..251). = U x2 . .s. however. 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 (5.310) to the SU(2) transformation assumed in deriving the general result (5. in fact.. It is of interest to trace the mapping from SU(2) onto SO(3) as represented by (5.251) and (5. therefore.e.5. One can. The SU(2) transformation entered in (5.309) does not distinguish between SU(2) transformations (5.g.312) transforms like an angular momentum state ˜ |1 m . the matrix A should represent a rotation around the x2 –axis in the space of vectors ˜ (x1 .309). Replacing the latter transformation by its negative form.e. namely that the component of the vector on the l. x2 . U = 0 1 1 i √ ( x1 − ix2 ) √ x3 − √2 0 2 2 (5.310).12: Mapping of SU(2) onto SO(3) 139 5.308) unaltered except for a factor (−1)2j which multiplies then also the ﬁnal result (5.h.311) (5. b− † = sin β † β b+ − cos b† .310) establishes a mapping of SU(2) onto SO(3).309) and the particular matrix (5. This can be shown by (1) applying to the matrix A = (dm m ) in (5. conclude that the mapping from SU(2) onto SO(3) is a 2–1 mapping.12 Mapping of SU(2) onto SO(3) The representation (5. i. This factor implies.313) × 0 0 2 = 0 sinβ 0 cosβ 1 −i 0 which.

140 Theory of Angular Momentum and Spin .

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

1) where 1.max = 1 2 IA−B E . in fact. mC . . In fact. will make their appearence a natural consequence of the symmetry of the building blocks of matter. i. . = 0. by the total energy E. the dimensions d(Bk ) of these subspaces does not exceed 100. Obviously. mA mB + mB mC + mA mC 2. . − 2 ≤ m1 ≤ (6. composite systems behaving like elementary objects are common. to separate the 14 641 dimensional space of rotational states Y 1 m1 (ˆAB ) Y 2 m2 (ˆC ) into r ρ subspaces Bk . 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. . 14 641} . = B such. . however. Such extremely useful transformation of the problem can be achieved through the choice of a new basis set B = { 1 .m1 . . vibrations and rearrangement of the particles in reactions like AB + C → A + BC.3) For rather moderate values 1. n = 1. have not even be considered. albeit puzzling. Such large number of dynamically coupled states would constitute a serious problem in any detailed description of the scattering process.2) The dimension d(B) of B is 1. a very large number.e.max + 1)2 (6. . 2. since further important degrees of freedom.max + 1)2 ( 2.max . rather than by Classical Mechanics. The rotational symmetry of the interaction between the particles allows one. . and the following mathematical description will shed light on their ubiquitous appearence in physics.max denote the largest orbital and rotational angular momentum values. 1. (6.max .max = 2.. The latter often arrives at the physical properties of composite systems by adding the .m1 2 . 2 . that only states within the subspaces Bk are coupled in the scattering process.max = ∆ 2 mA mB mC E . 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 and 2. B1 ⊕ B2 ⊕ . as we will demonstrate below.max = 10 one obtains d(B) = 14 641.max 2. 1. . 2} 1. (6. each of which stands for two angles.max d(B) = 1 =0 (2 1 + 1) 2 =0 (2 2 + 1) = ( 1. in particular.4) The basis set which provides a maximum degree of decoupling between rotational states is of great principle interest since the new states behave in many respects like states with the attributes of a single angular momentum state: to an observer the three particle system prepared in such states my look like a two particle system governed by a single angular momentum state.142 Addition of Angular Momenta and Spin ρC . and by the moment of inertia IA−B of the diatomic molecule A–B approximately as follows 1. by the masses mA . .m2 c (n) r ρ 1. − 1 ≤ m1 ≤ 1. 2. 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 values of which are determined by the size of the interaction domain ∆V . . mB . . ρ 1 = 0.m2 Y 1 m1 (ˆAB ) Y 2 m2 (ˆC ).

3 (6.7) where the generators J (k) are diﬀerential operators acting on rAB (k = 1) and on ρC (k = 2).53. 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). For ˆ ˆ example. R−1 (ϑ) ρC ) (6.6) For the following it is essential to note that H is not invariant under rotations of only rAB or ρC . Since this is true for any ψ(rAB .9) hold since the components of J (k) are diﬀerential operators with respect to diﬀerent variables. One can equivalently express therefore (6. according to (5. Jq(2) = 0 for p. (6. the existence of a basis (6. 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. rotations R(ϑ) of the wave functions ψ(rAB . The invariance of the Hamiltonian under overall rotations of the three particle system implies then H = R(ϑ) H R−1 (ϑ) . ρC ) it follows H R(ϑ) = R(ϑ) H. ∂yAB ∂zAB i (2) ∂ ∂ − J3 = ρy − ρx . ρC ) = ψ(R−1 (ϑ) rAB . 5.48) i i R(ϑ) = exp − ϑ · J (1) exp − ϑ · J (2) (6. 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. In this sense. be diﬀerent from H. ∂ρx ∂ρy (6.6.55) holds i (1) − J1 = zAB ∂ ∂ − yAB . To specify this property mathematically let us denote by H the Hamiltonian in the rotated frame. Rotational Symmetry of the Hamiltonian As pointed out already.5) according to (5. but solely under simultaneous and identical rotations of rAB or ρC .5) do not aﬀect the Hamiltonian. in fact. the reader will learn in the following section how to add and subtract in the microscopic world of Quantum Physics. ρC ) deﬁned through R(ϑ) ψ(rAB . It holds then H R(ϑ) ψ = R(ϑ) H ψ. connected with the fact that the Hamiltonian H does not depend on the overall orientation of the three interacting particles. assuming presently that H might. 2. i. Hence. q = 1. presumably a facility the reader would like to acquire with great eagerness.8) Obviously..10) . Following our description of rotations of single particle wave functions we express (6.e.4) which decouples rotational states is connected with the rotational symmetry of the Hamiltonian of the three particle system considered.7) i R(ϑ) = exp − ϑ · J (6.0: Addition of Angular Momenta and Spin 143 corresponding physical properties of the elementary components.

2 together with the property [A.9).e.16) following the procedure stated in the theorem above [c. To order O(|ϑ|) one obtains i ϑ · JH . B] + [A. a supposition which can be tested through the commutation properties of these operators. J ] = i k m Jm (6. 2. In fact. Jk ] = 0 . k = 1.12) We consider this equation for inﬁnitesimal rotations.71–5.11) we can write the condition (6. B + C] = [A. (6. k = 1. however.81)].12) each individually can have simultaneous eigenstates with the Hamiltonian. [H.f.13) Hϑ · J − Since this holds for any ϑ it follows HJ − J H = 0 or.17) . 3 . k = 1. 3 as the three components of the total angular momentum operator. We suspect.14) We will refer in the following to Jk . the important fact that the Jk obey the Lie algebra of SO(3). i. The property (6. the operators Jk . (6. we will ﬁnd that the states yield the basis B with the desired property of a maximal uncoupling of rotational states.. Eqs.. 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. For a proof one uses (6. 2. 3 cannot have simultaneous eigenstates among each other. 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. J ] = i k m Jm for n = 1. i. We recognize. (5.14) implies that the components of the total angular momentum operator (6. that energy. C].11) By means of (6. 2.e. for |ϑ| H ≈ 1 − 1 i ϑ·J H 1 + 1 i ϑ·J ≈ H + i 1.e. 2. that the components Jk . 3 do not commute. of course. (6. One can show readily that the commutation relationships [Jk . and the eigenvalues of J2 and J3 are good quantum numbers.144 where J = J (1) + J (2) Addition of Angular Momenta and Spin . In the following we will use the notation r ρ Ω1 = rAB . ˆ Ω 2 = ρC .81) we want to consider the relationship between YJM and Y 1 m1 (ˆAB ) Y 2 m2 (ˆC ).6) for rotational invariance of the Hamiltonian in the form i i H = exp − ϑ · J H exp + ϑ · J . ˆ (6. Before we apply the procedure (5. componentwise. i.14) implies that the total angular momentum is conserved during the scattering process. k = 1. the (n) (n) (n) properties [Jk .71–5. (6. r ρ Deﬁnition of Total Angular Momentum States The commutation property (6.

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

In fact. . . it is not immediately obvious what the J– values are. 2 ) and B ( 1 .. M = −J.s. 2) . . Ω2 ) YJ M ( 1.h. For each quantum number J there should be 2J + 1 elements YJM with M = −J. Our derivation below will also yield real values for the Clebsch-Gordan coeﬃcients. .s.. This obviously corresponds quantum mechanically to a range of J–values J = | 1 − 2 |. 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 ). In this case the range of |J|–values would be the interval [ |J (1) | − |J (2) | . (6. m2 = − 2 . . J. −J + 1. . . .146 Theory of Angular Momentum and Spin The expansion (6. −J + 1. Exercise 6. to a new basis B ( 1 . . 1 + 2. The orthonormality (Ω1 ) Y 2 m2 (Ω2 )Y 1 m1 Ω1 ) Y 2 m2 (Ω2 ) = δ 1 1 δm1 m1 δ 2 2 δm2 m2 . .28) This property follows from the fact that the basis elements are eigenstates of hermitian operators with diﬀerent eigenvalues. Ω2 ).1. 1 + 2 . respectively. property implies dΩ1 dΩ2 Y 1 m1 corresponding to the l.24) yields 1 m1 2 m2 ) ∗ 1 m1 2 m2 ) (J M| = δJJ δM M . The property dΩ1 dΩ2 Y∗ ( 1 . The basis B ( 1 . . . J = | 1 − 2 |. 2) = {Y 1 m1 (Ω1 ) Y 2 2 m2 (Ω2 ).e. (6. (6. (6.2: Prove Eq. B2 2) should be 2 |Ω1 . JM 2 |Ω1 .18) constitutes a change of an orthonormal basis B( 1 .25) i. and that the states can be normalized.18) applied to Y∗ and to YJ JM (J M | m1 . 2 ) has (2 1 + 1)(2 2 + 1) elements. (6.25) We want to state now two summation conditions which follow from the orthonormality of the two basis sets B( 1 .h. − 2} 1 + 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. . |J (1) | + |J (2) |]. | 1 − 2 | + 1. m1 = − 1 . | 1 − 2| + 1. the range of values assumed for J is correct.m2 2 M and with (6. . . − + 1.27) together with (6. . . . 2 |Ω1 .24) The basis B( 1 . . However. (6. J} . .23) corresponding to the r. the basis B ( 1 . J (1) and J (2) .26) We will show below in an explicit construction of the Clebsch-Gordan coeﬃcients that. Ω2 ) = δJJ δM M (6. in fact. = {YJM ( 1 .

Y = J =| 1 − 2 | M =−J 1 m1 M (Ω1 )Y 1+ 2 2 m2 (Ω2 ) (J M | ∗ 1 m1 2 m2 ) YJ M J ( 1. i. The result of this construction will include all the properties previewed above.28) from (6. 2 |Ω1 .27) yields δJJ δM M = m1 .. it is possible to expand Y 1 m1 (Ω1 )Y 2 m2 (Ω2 ) in terms of YJM ( 1 ..e. will be based solely on the commutation properties of the operators J and J (k) .30) The latter property follows from multiplying (6. At this point we like to stress that the construction will be based on the theorems (5. 1 m1 2 m2 ) = 0 if J < | 1 − 2| . (6.18) by Y∗ J orthogonality property (6. The (J M | 1 m1 2 m2 ) cJ M . 2 ) and B ( 1 .33) JM The latter summation has not been restricted explicitly to allowed J–values.2: Construction of Clebsch-Gordan Coeﬃcients 147 The second summation condition starts from the fact that the basis sets B( 1 . also apply the results.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 . One can show readily using the same reasoning as applied in the derivation of (6. 2 |Ω1 . 2 |Ω1 . Ω2 ) . Ω2 ) .32) which complements (6.28) allows one to conclude that the coeﬃcients cJ (J M | 1 m1 2 m2 )∗ .34) 6.m2 M ( 1.e. to composite systems involving spin. rather the convention (J M | has been assumed. (6. Ω2 )Y 1 m1 (Ω1 )Y 2 m2 (Ω2 ) .6. 1+ 2 J Y 1 m1 (Ω1 )Y 2 m2 (Ω2 ) = J =| 1 − 2 | M =−J cJ M YJ M ( 1. 2 |Ω1 .71–5. (6. therefore. i.81) stated above. and actually also the properties of Clebsch-Gordan coeﬃcients stated above.1 states. 2 |Ω1 . e..29) where the expansion coeﬃcients are given by the respective scalar products in function space cJ M = dΩ1 dΩ2 Y∗ J M ( 1. 2 ) span the same function space. A similar 2 construction will also be applied to composite systems governed by other symmetry groups. or J > 1 + 2 (6.31) are identical to Comparision with (6. or baryons multiplets involving three quarks.e. . Ω2 ) and integrating..18).2 Construction of Clebsch-Gordan Coeﬃcients We will now construct the Clebsch-Gordan coeﬃcients.g. (6. Ω2 ). the group SU(3) in case of meson multiplets involving two quarks. We can. Hence. (6. i.

therefore.43) .35) 1 − 2 |. 2 |Ω1 .37) and replace (1) (2) according to (6. Ω2 ) = Y 1 1 (Ω1 ) Y 2 2 (Ω2 ) (6.39) which we will try for a solution of (6. For this purpose we insert (6. The construction assumes a particular choice of J ∈ {| such J–value seeks an expansion (6. have shown Y 1+ 2.18) which satisﬁes J+ YJJ ( 1 . Ω2 ) . . .40) (Ω1 ) J+ Y (2) 2 2 J+ Y 1 1 (Ω1 ) Y (Ω2 ) + Y (Ω2 ) = 0 . 1+ 2 ( 1.25) using (6. .148 For the construction of YJM we will need the operators J± = J1 + iJ2 . We embark on the suggested construction for the choice J = 1 + 2 . we can demonstrate condition (6. 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 . (6. we can also demonstrate using (??) that G dΩ1 = We. We ﬁrst seek an unnormalized solution GJJ and later normalize. −J + 1. 1+ 2 ( 1. . −J. J3 YJJ ( 1 . Similarly. (6. . . Ω2 ) 2 |Ω1 . Ω2 ) = Y 1 1 (Ω1 ) Y 2 2 (Ω2 ) . 2 |Ω1 .41) 2) Y (Ω1 ) Y (Ω2 ) . (6. The corresponding classical total angular momentum vector Jclass (1) (2) would be obtained by aligning both Jclass and Jclass also along the z–axis and adding these two vectors. . Ω2 ).64) J3 = = (1) + J3 (1) 1 (2) Y 1 1 (Ω1 ) Y 2 2 2 2 (Ω2 ) 1 1 J3 Y ( + 1 1 (Ω1 ) Y 1 1 (Ω2 ) + Y 2 2 (Ω1 ) J3 Y (2) 2 2 (Ω2 ) (6.37) The solution needs to be normalized. 2 |Ω1 . Ω2 ) is normalized dΩ2 G dΩ1 Y 1 1 1+ 2.66. 1+ 2 ( 1.11) J+ by J+ + J+ . (6.38) for M = J − 1.5. Ω2 ) . | 1 − 2| + 1.37). Quantum mechanically this corresponds to a state G 1+ 2. 2 |Ω1 .42) ( 1.36) = 0 = J YJJ ( 1 . . 2 |Ω1 . 2 |Ω1 .68) J+ = (1) + J+ (1) (2) Y 1 1 (Ω1 ) Y 2 2 2 2 (Ω2 ) 1 1 (6. J − 2.11) and (5. J} by applying repeatedly J− YJM +1 ( 1 .39) into (6. We obtain using (5. 2 |Ω1 . Having determined such YJJ we then construct the whole family of functions XJ = {YJM ( 1 . Ω2 ) = (J + M + 1)(J − M )YJM ( 1 . 2 |Ω1 . 1 + 2} and for (6. 1+ 2 In fact. . 1+ 2. Ω2 ) (Ω1 ) dΩ2 Y 2 2 (Ω2 ) = 1. Ω2 ) Addition of Angular Momenta and Spin (6. M = −J.

Y 1 1 −1 (Ω1 )Y (Ω2 ) + Y 1 1 This construction can be continued to obtain all 2( 1 + 2 ) + 1 elements of B + and. The l. We seek for this purpose ﬁrst an unnormalized solution G 1 + 2 −1 1 + 2 −1 of (6.2: Construction of Clebsch-Gordan Coeﬃcients 149 We now employ property (6. .36) we proceed as above and obtain J+ Y = (1) 1 1 −1 (Ω1 ) Y 2 2 2 2 (Ω2 ) + c Y 1 1 1 1 (Ω1 ) J+ Y (2) 2 2 −1 (Ω2 ) (6. . G 1 + 2 −1 1 + 2 −1 in (6. . .h. of (6. M = −( + ). 1 + 2 − 2. The result is illustrated for the case 1 = 2. According to the condition (6.21) we set G 1 + 2 −1 1 + 2 −1 ( 1 2 |Ω1 Ω2 ) = Y 1 1 −1 (Ω1 )Y 2 2 (Ω2 ) + c Y 1 1 (Ω1 )Y 2 2 −1 (Ω2 ) (6.38) to construct the family of functions B + = {Y + M ( . ). 2 |Ω1 . To demonstrate (6.45).27).39) as required by (6. . 1 + 2 − 3.1: Construct following the procedure above the three functions YJM ( 1 .45) for some unknown constant c.s. .47) can serve now to obtain recursively the elements of the family B + − for M = 1 + 2 − 2. The r. . Ω2) for M = 1 + 2 − 2 and J = 1 + 2 . We have provided in Table 1 the explicit form of YJ M ( 1 2 |Ω1 Ω2 ) for 1 = 2 and 2 = 1 to illustrate the construction. Expression (6. with the symmetry pattern and with the terms Y 1 m1 Y 2 m2 contributing to each YJ M . Ω2 ). M = −( + − ). of (6.6.2. ( + )}.37) is satisﬁed. and all total angular momentum functions for a given choice of 1 and 2 . | . 1 + 2 − 1. 2 |Ω1 . Show that the resulting functions are orthonormal.47) (Ω2 ) − 1 1 2 + Y 2 1 1 −1 (Ω1 )Y 2 2 + Y 2 1 1 (Ω1 )Y 2 2 −1 (Ω2 ) . . ). To satisfy this equation one needs to choose c = − 1 / 2 . One can readily show that (6. . etc. Ω2 ) 2 1+ 2 = (Ω1 )Y 2 2 −1 (6. Having constructed this family we have determined the Clebsch-Gordan coeﬃcients ( 1 + 2 − 1M | 1 m1 2 m2 ). | . all the Clebsch-Gordan coeﬃcients ( 1 + 2 M | 1 m1 2 m2 ).s.38) yields pression 2( 1 + 2 )Y 1 + 2 1 + 2 −1 ( 1 . Normalization yields furthermore Y = 1 1 + 2 −1 1 + 2 −1 We have thereby determined ( 1 2 |Ω1 Ω2 ) (6. One can obviously continue the construction outlined to determine Y 1 + 2 −2 1 + 2 −2 . −( 1 + 2 − 1). . thereby. .h. One obtains then Y 1 1+ 2 1 + 2 1 + 2 −1 2 2 ( 1. We like to construct now the family of total angular momentum functions B + − = {Y + − M ( .38) yields with J− = J− + J− the ex√ √ 2 1 Y 1 1 −1 (Ω1 )Y 2 2 (Ω2 ) + 2 2 Y 1 1 −1 (Ω1 )Y 2 2 −1 (Ω2 ). 2 |Ω1 .37). 2 = 1 in Table 1. 6. We demonstrate the procedure ex(1) (1) plicitly only for M = 1 + 2 − 1. This expression is orthogonal to (6. in particular. ( + − )}.46) 2 1 +c Y (Ω1 ) Y 2 2 (Ω2 ) = 0 . . Exercise 6.36.44) (Ω2 ) . . The reader should familiarize himself with the entries of the Table.

. Ω2) Y3−2 (2.816497 Y2−2 (Ω1 )Y11 (Ω2 ) Y2−1 (Ω1 )Y10 (Ω2 ) Y20 (Ω1 )Y1−1 (Ω2 ) Y2−2 (Ω1 )Y10 (Ω2 ) − 2 3 Y2−1 (Ω1 )Y1−1 (Ω2 ) Y2−2 (Ω1 )Y1−1 (Ω2 ) 1 Table 6. Ω2) Y10 (2. 1|Ω1 .547723 0. Ω2) Y2−1 (2.447214 0.774597 0. 1|Ω1 . 1|Ω1 .816497 Y22 (Ω1 )Y1−1 (Ω2 ) 1 15 1 3 3 5 1 5 1 2 3 10 2 5 1 2 1 10 2 3 1 3 Y20 (Ω1 )Y11 (Ω2 ) Y21 (Ω1 )Y10 (Ω2 ) 0.1: 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. Ω2) Y30 (2.408248 −0.57735 0. 1|Ω1 . 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) Y1−1 (2.447214 −0.258199 0.258199 −0. 1|Ω1 .774597 0 2 5 8 15 1 6 3 10 1 3 0. 1|Ω1 . Ω2) Y21 (2. 1|Ω1 . Ω2) Y11 (2.316228 0.547723 0. Ω2) Y31 (2.632456 0.816497 0.408248 −0. Ω2) Y20 (2. 1|Ω1 .57735 −0.707107 0.57735 3 10 3 5 Y2−1 (Ω1 )Y11 (Ω2 ) Y20 (Ω1 )Y10 (Ω2 ) Y21 (Ω1 )Y1−1 (Ω2 ) −0.57735 0.632456 0.316228 0.57735 0.57735 0.730297 −0. Ω2) Y3−1 (2.816497 −0.547723 0. 1|Ω1 . 1|Ω1 . 1|Ω1 . 1|Ω1 . 1|Ω1 . Ω2) Y32 (2. Ω2) Y3−3 (2.774597 − − − − 0. 1|Ω1 .707107 0.707107 0.632456 −0.547723 0. Ω2) Y22 (2.707107 0.730297 0.150 Addition of Angular Momenta and Spin Y33 (2. 1|Ω1 . Ω2) Y2−2 (2.

53) (6. + 2 − J−M +1 for = J + 1 2J+2 2 J−M for = J − 1 2J 2 = . commute with each other ζ ζ aζ . (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. 1 .ζ ζ. m − 2 . bζ . J. For this purpsose we will employ the spinor operators introduced in Sections 5.3: Explicit Expressions of CG–Coeﬃcients 151 Exercise 6.ζ a† < ζ |σk | ζ > aζ ζ b† < ζ |σk | ζ > bζ ζ (6.254)] 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 .10 where we studied single particle angular momentum and spin. aζ . (5. ζ = ± and the matrix elements < ζ |σk | ζ > are as deﬁned in Section 5.6. (6. 1 (6.2: Use the construction for Clebsch-Gordan coeﬃcients above to prove the following formulas J+M 1 for = J − 2 2J 1 1 1 = J. bζ = a† . b† refer to diﬀerent particles and. b† ζ = 0.9.10.287) to two particles (1) J k k = = (2) J 1 2 1 2 ζ. 2 . 5.48) (6. Accordingly. hence.52) and | 2 m2 2 According to Section 5.10. bζ = = a† . aζ bζ . b† ζ ζ = 0 = 0 . M | . we extent deﬁnition (5. a† ζ . m + 1 . a† and bζ . These expressions will be derived now.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 are dealing now with two particles carrying angular momentum or spin. It is actually possible to state explicit expressions for any single coeﬃcient (JM | 1 m1 2 m2 ).54) 2 + = √ b† ( 2 +m2 )! √− ( 2 −m2 )! . a† ζ ζ b† .49) where ζ.51) The operators aζ .10 [cf. The creation and annihilation operators are again of the boson type with commutation properties aζ . 5.9. b † ζ = δζζ = δζζ . Deﬁnition of Spinor Operators for Two Particles In contrast to Sections 5. b† ζ ζ = aζ .2. M | . 6.50) (6.

for such states should hold J2 |J.e. M ( 1 . 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) J 3 | 1 m1 1 3 | 2 m2 2 = m1 | 1 m1 = m2 | 2 m2 1 2 (6. m1 . 2 . i. M ( 1 . (6.66) (6. The aim of the present Section is to determine closed expressions for the Clebsch–Gordan coeﬃcents ( 1 . 2) + 1) |J. (1) J 2 . M ( 1 . 2. mj j = j | j . ˆ For the operators kj holds ˆ kj | j . m2 |J.59) J± = J1 ± i J2 ..69) − 2| ≤ J ≤ + 2 . 2 )). m2 |J. which describe a two particle system according to (6.54). 2) 2) (6. 3 . = = 1( 1 2( 2 2) 2) (6. M ( 1 . m1 1 | 2 . (5.68) The states |J. (1) J 2 |J. 5. M ( 2) 2) = J(J + 1) |J. |J.62) . −J ≤ M ≤ J .56) The states | 1 . M ( (2) J 2 |J. J3 |J.64) (6. M. . = M |J. 2) 1. + 1) |J.55) . (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 . 6. m2 2 . 2) 2) can be expressed in terms of Clebsch-Gordan coeﬃcients (6.303). mj j and. M ( 1 . 2 . . ˆ kj |J.61) (6. M ( 1 . M ( 1 . m1 1 | 2 .65) (6. (2) J 2 . m2 1 2 ( 1 .53.63) according to At this point.m2 | 1 . can be expressed in terms of the number operators 1 † 1 ˆ ˆ a† a+ + a† a− . are | 1 . m1 1 | 2 . we like to recall for future reference that the operators (5. M ( 1 . 2) We will also require below the raising and lowering operators associated with the total angular momentum operator (6. 2 . 2) .18) as follows = | 1 m1 . (6. k2 = b b+ + b† b− k1 = + − − 2 2 + namely.57) b† + ( (2) 2 2 +m2 1 +m1 b† − ( 2 2 −m2 + m1 )! − m1 )! (1) + m2 )! − m2 )! |Ψo .152 It holds. (1) J . (6. m1 . M ( 1 . as usual are denoted by their respective quantum numbers J. 2 . (6. 1 . M ( 1 .302. J3 . 2 ) which are simultaneous eigenstates of the operators 2 2 J2 . 1. k = 1. 2 )) . (6. in analogy to Eqs.58) Jk + (2) Jk . (j) 2 ˆ ˆ J = kj ( kj + 1 ) . M ( 1 .59) We seek to determine states |J.60) (6. j = 1. hence. (2) J which. M ( 1 . M ( 1 . m2 a† + ( 1 2 = a† − ( 1 1 −m1 (6.300).67) = j |J.

3: Explicit Expressions of CG–Coeﬃcients The Operator K † The following operator K † = a† b† − a† b† . b† = 0 .70) will play a crucial role in the evaluation of the Clebsch-Gordan-Coeﬃcients. a† b† − a† b† − + − − + 2 + 1 1 = a† a+ . For example.288).73. Employing (6.73) can be readily proven. K † + kj . ˆ A similar calculation yields [k2 . due to J2 = K† 1 2 J+ J− 1 † K . + − + − 2 2 − + = .72) (6. 2 2 3 ˆ = K † kj + K † . a† b† − a† b† + − + − + − − + 2 1 1 = a† a+ . the relationships (6. K † 2 1ˆ ˆ 1 ˆ ˆ kj kj + 1. using (6. We note that.73) yields J3 .64. (6. a† b† + a† a− . a† b† + + − − + 2 2 − 1 1 = a† b† − a† b† = K† . K† J3 . 4 Using J3 = (1) J 3 + (2) J 3 . 6.74) + 1 J− J+ + J2 . a† b† + − − + 2 + 2 − 1 1 − b† a† b+ . j = 1. and applying the relationships (6. K † J± .65) and (6. + − − + 2 2 = 1 † 2K . (6. This operator obeys the following commutation relationships with the other pertinent angular momentum / spin operators ˆ kj . K † (j) J 2 . K † 1 a† a+ − a† a− + b† b+ − b† b− . K † = 0.75) The relationships (6. K † (kj + 1) 2 2 1ˆ † 1 ˆ kj K + K † (kj + 1) 2 2 1 ˆ 1 ˆ K † kj + kj . j = 1.71–6.6. 2 4 = 0 = = 0.51) one obtains ˆ k1 . b† − b† a† b− . K † 1 † a a+ + a† a− . 6. a† b† − a† a− . 6.71) (6.74) imply 3 2 J2 .50. + − − + 153 (6.71–6.71) one can show K† = = = = = 1 ˆ ˆ kj (kj + 1).73) (6. K † + K † 2 2 3 † †ˆ K kj + K . K † ] = (j) J 2 . expressing (k) J 3 through the creation and annihilation operators according to (5.

One can generalize property (6.e.72) and (6. but for diﬀerent 1 and 2 quantum numbers of the operators (1) J 2 and (2) J 2 . j1 + j2 (j1 + n n . K † = Addition of Angular Momentum and Spin a† a− + b† b− .2. 2) = N |J. M ( 1 . j2 + ) 2 2 n n = N (n. 6.63. We consider now the case that (K † )n acts on the simplest total angular momentum / spin state. 2 ) remain eigenstates of J2 and J3 with the same quantum numbers. The commutation properties (6. j1 + j2 (j1 . j1 + j2 (j1 . |j1 + j2 . it holds K † |J. j1 1 + .76) Here N is an unknown normalization constant. 6. M ( 1 .78) a state which has been used already in the construction of Clebsch-Gordan coeﬃcients in Section 6. M ( 1 .62. M ( 1 . b† + − + + + − The property [J− . 2) = = (j) 2 J . = − a† a− . M ( 1 + . M ( 1 . To demonstrate that the resulting states are eigenstates of (1) J 2 and (2) J 2 we exploit (6. M ( 1 . K † ] = 0 is demonstrated in an analoguous way. 2 ) . j + 2 2 2) However.. on the state |j1 + j2 . (6. M ( 1 + n . i. 1 2 + 2 .77) where N is another normalization constant. j2 2 . 2 2 + n ) 2 (6.75) ascertain that under the action of K † the states |J.67) (j) J 2 K † |J. K† + K † (j) J 2 |J. 2 ) j + 4 1 3 + K † |J.73. M ( 1 . this result implies that K † |J. Action of K † on the states |J.77). j2 ) . M ( 1 . M ( 1 . 2 (6. 4 3 + j ( j + 1) |J. 2) We want to demonstrate now that the action of K † on the states |J.79) n |j1 + j2 .76) and state K† n |J. 6.292) one can derive similarly J+ . namely. 2 2 1 + ) . 2) 2) ˆ K † kj + = K† = j 3 † K + K † j ( j + 1) |J. M ( 1 . a† b− − a† b† + + + − + = 0. j1 + ) K † 2 2 (6. according to (6. M ( 1. 2) is a state with quantum numbers 1 1 + 1 2 and = N |J. j2 ) = |j1 . 2 ) produces again total angular momentum eigenstates to the same J and M quantum numbers of J2 and J3 . j1 1 |j2 .154 Starting from (5. a† b† + b† a† b− . Application of (K † )n to this state yields.

2 = j2 + n 2 (6. (6. M ( 1 . 1 = j1 + n 2 .84) (6. in analogy to the construction (5. 2 2 It is now important to notice that any state of the type |J.80) = = = 1 1 2 2 − J 1 n − = (J + 2 2 n 1 − = (J + 2 2 + 1 1 − − 2) 2) . allow one to obtain the states |J. − 1. j2 + n ). holds |J. (J + 2 − − 2) 1 1) 2 The normalization constant appearing here is actually N( + − J. M ) (J + × K† 1 − 2 )! (J + 2 − 1 )! J+ 1 − 1 + 2 −J (J− )J−M a† + × 2 (6. 2) . 2 ) ∆(J. J( 1 .105) of the spherical harmonics. 2 ) for −J ≤ M ≤ J as follows |J.82.81) = N( 1 + 2 1 2 − J. 2) Our construction of the states |J. K† 1 2 1 + 2 −J × × . 6. 2 ) .86) b† + J+ 2 − 1 |Ψo . M ( 1 . = (J + M )! (2J)! (J − M )! 1 2 2) (6. M ) (J− )J−M |J. 1.74)] yields |J. of (6.6. Combining (6. M ( 1 . (6. Strategy for Generating the States |J. J( 1 .79) n. j1 . j1 + n . The latter states. M ) . 2) = N( 1 + 2 − J.85) ∆(J.3: Explicit Expressions of CG–Coeﬃcients 155 where we denoted the associated normalization constant by N (n. J( 1 .104. J( 1 .h. M ( 1 . (6.s. For this purpose one needs to choose in (6.82) 1 × | (J + 2 1 × | (J + 2 1 (J + 2 1 − 1 ) . j2 as follows J = j1 + j2 which is equivalent to n j1 j2 Accodingly. 2) 2) .79).84) with (6.82) for |J. (6. 2 ) can be expressed through the r. 5. 1 .f. 2) = ∆(J. 2 ) exploits the expression (6. M ( 1 .57) and exploiting the fact that J− and K † commute [c.83) The derivation of this expression will be provided further below (see page 158 ﬀ). 2) 1 2 = ( 1 + 2 (2J + 1)! − J)! ( 1 + 2 + J + 1)! 1 2 .

y) = exp (λJ− ) exp x a† exp x b† |Ψo .88) in terms of monomials (6. y) is a generating function for the states |Gr deﬁned in (6. follows from the commutation properties (6. in terms of | 1 . I(λ.e. + − + − .89) i. x.57)]. y) = λr xs y t r |Gst . + + Taylor expansion of the two exponential operators yields immediately I(λ. in analogy to (5.86) in terms of monomials 1 +m1 1 −m1 2 +m2 2 −m2 a† a† b† b† |Ψo .. a† ] = [b+ .93) = f (a† + λ a† ) g(b† + λ b† ) |Ψo . x. (6. r!s!t! r. b† ] = 0 [cf.86) |Gr = Jr a† st − + s b† + t |Ψo (6.264). One obtains then using J− = a† a+ + b† b+ (6. m1 1 | 2 .91) which. For this purpose we introduce the generating function I(λ.156 Addition of Angular Momentum and Spin Our strategy for the evaluation of the Clebsch-Gordan-coeﬃcients is to expand (6. Comparision with (6.88).87). x. m2 Gordan-coeﬃcients.69) yields then the Clebsch- Expansion of an Intermediate State We ﬁrst consider the expansion of the following factor appearing in (6.92) − − and noting [a+ .52).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..t (6.s.90) (6.50–6. x. (6. (6. st The desired expansion of |Gr can be obtained from an alternate evaluation of I(λ.87) + − + − i. 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.e. 2 [cf.

x.s. m2 2 .t = q r! × a† + s q s−q t r−q a† − q × b† + t−r+q b† − r−q |Ψo (6.86) (J− )J−M a† + = q J+ 1 − 2 b† + J+ 2 − 1 |Ψo 1 (6. 157 (6.95) Comparision with (6.70) holds K† 1 + 2 −J = s 1 + 2 −J b† − s 1 + 2 −J−s (−1)s 1 + 2 −J−s (6.t × a† + s−q s q b† + t r−q t−r+q a† − q b† − |Ψo (6.50). using the commutation property (6.97) + q)! |Ψo × (J − M )!(J + 1 − 2 )!(J + 2 − 1 )! q!(J + 1 − 2 − q)!(J − M − q)!(M + 2 − J+ 1 − 2 −q × Final Result a† + a† − q b† + M + 2 − 1 +q b† − J−M −q Our last step is to apply the operator (K † ) 1 + 2 −J to expression (6. Expanding the exponentials in st (6. m1 1 | 2 .t xs y t s! t! xs y t s! t! a† + a† + λ a† + − s u u s b† + λ b† + − t v b† + t−v t |Ψo = s. With K † given by (6. Operation of this operator on (6. y) = s. one can write the right factor in (6.98) a† − s a† + b† + s .3: Explicit Expressions of CG–Coeﬃcients We conclude I(λ.q (−1)s (6.90) allows one to infer |Gr s. M ( 1 .96) and. 2 ) in terms of states | 1 .99) .88).94) One can infer from this result the desired expressions for |Gr .t × λ v b† − × r−q v × = t s−u v λu a† − r! q |Ψo λr xs y t r! s! t! r. |J. using the deﬁnition (6.94) yields I(λ.97) yields. 2) = s. to obtain the desired expansion of |J.6. x. y) = exp a† + λ a† + − exp b† + λ b† + − |Ψo .97). M ( 1 .

j2 .104) . × 2) = N( 1 + 2 1 − J.103) this can be written 1 = N 2 ψ(j1 .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 (6.53. The summation over q corresponds then to the summation over m1 .103) = 1. n) 2 )|J. M − +q+s One can conclude that this expression reproduces (6.6.69) since.102) We want to determine now the expression (6.54) between creation operator monomials and angular momentum states allow one to write this |J. j2 = 1 (J + 2 2 − 1) . M ( 1 .82) and (6. n = 1 + 2 −J.83) of the normalization constant N ( 1 + deﬁned through (6. M ) 1 = + J)!(− 1 + 2 + J + 1)! 2 2 + 2 − J)!( 1 − ( 1+ + J)! 1 2 × [( × s + m1 )!( − m1 )!( 2 + m2 )!( 2 − m2 )!(J + M )!(J − M )!] 2 (−1)s s!( 1 − m1 − s)!( 2 + m2 − s)! 1 2 1 × The Normalization 1 ( 1 + 2 − J − s)!(J − 1 − m2 + s)!(J − + m1 + s)! (6. − 1 . √ 2J + 1 1 2 . − q − s)!(q + s)! 2 × (M + 1 − 1 + q + s)!( 1 1 + 2 − M − q − s)! 2 − q − s 1 | 2.69) if one identiﬁes m1 = 1 −q − s . For this purpose we introduce j1 = 1 (J + 2 1 2 − J. (6.101). (6. 2 ) we consider the scalar product J. To determine N = N ( 1 + 2 −J. J( 1 . M ) (J + 2 )! (J + 2 − 1 )! s.100) ( 1 + 2 − J)!(J − M )!(J + 1 − 2 )!(J + 2 − 1 )! s!( 1 + 2 − J − s)!q!(J + 1 − 2 − q)!(J − M − q)!(M + 2 − 1 1 + q)! × (2 ×| 1 . m2 |J. according to (6. q = 1 − m1 − s and m2 = M − m1 . 2 ) ∆(J. 2 ) (6. j2 . J( 1 .82).q (−1)s × (6. 2 ) − 2) .101) Note that m1 + m2 = M holds. The Clebsch-Gordan coeﬃcents are then ﬁnally ( 1 . Using (6. 1 . m2 = M − 1 + q + s. 1 . m2 in (6. n)|ψ(j1 . m1 .

111) Comparision with (6.108) = λ=0 (n1 + ν)! n1 ! (6. we obtain |ψ(j1 .109) and Taylor expansion of the left hand side of (6.104) 1 = N2 (n!)2 (2j1 )!(2j2 )! (2j1 + n − s)!(2j2 + s)! s!(n − s)! 2j2 + s 2j2 (6. n) in terms of states |j1 . (6.6. We employ the expression (6. j2 .108).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 (6.112) s .70) for the operator K † . (6.m2 n1 + m1 n1 n2 + m2 n2 λm1 +m2 (6.3: Explicit Expressions of CG–Coeﬃcients where |ψ(j1 . m2 2 . j2 2 .105) The ﬁrst step of our calculation is the expansion of ψ(j1 . (6. n) = K† n 159 |j1 . applying (6. j2 .108) yields the identity n1 + r − s n1 n2 + s n2 = n1 + n2 + r + 1 n1 + n2 + 1 .108) twice. 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 . j1 1 |j2 . j2 . One obtains then. m1 1 |j2 .110) which can be written 1 1−λ n1 +n2 +2 = r s n1 + r − s n1 n2 + s n2 λr (6.57) for these states and the expression (6. 1 1−λ n1 +1 1 1−λ n2 +1 = m1 .106) The orthonormality of the states occurring in the last expression allows one to write (6. Accordingly.

2 .115) This relationship is a trivial consequence of (6. m1 |J.116) one expresses the Clebsch-Gordan coeﬃcient on the r. m2 ) and.102) by replacing ( 1 . (2j1 + 2j2 + 1)! (6.h. m1 .107) yields 1 = N 2 (n!)2 Theory of Angular Momentum and Spin 2j1 + 2j2 + n + 1 2j1 + 2j2 + 1 n!(2j1 + 2j2 + n + 1)! = N2 . m1 ) by ( 2 . −m1 2|J. m1 ) . m1 |J. (6. −m2 . 2 . of (6. of this equation J (1) = J (2) − J .118). 2 1+1 (6.g. and J (2) are vectors in R .120) The symmetry properties (6. e. −m1 . M ) = (−1) 1 + 2 −J ( 2 . 2 . namely.114) as long as J. (6. −m1 . ( 2 . (6. −m2 |J. interchanging the operators J (1) and J (2) results in J = J (2) + J (1) .114) one obtains − J = − J (1) − J (2) The respective Clebsch-Gordan coeﬃcients ( 1 . 2 . −M ) = (−1) 1 + 2 −J ( 1 . 6. 2 . We will demonstrate this now. M ) . m2 |J. M ) ( 1 . To derive relationship (6.119) The corresponding symmetry property of the Clebsch-Gordan coeﬃcients is ( 1 .h. of (6.. −M ) .11) J = J (1) + J (2) .114) by.s. m1 . m1 ).102) of the Clebsch-Gordan coeﬃcients. manner to the coeﬃcients ( 1 . For the quantum mechanical addition of angular momenta the Clebsch Gordan coeﬃcients ( 1 .s. m2 |J. m2 |J.114) show a simple relationship to the Clebsch Gordan coeﬃcients ( 2 . M ) = (−1) 2 +m2 2J + 1 ( 2 . J (1) on the r. and (6.116). J. 1 . ( 1 . 2 . m2 |J.118) Finally. m2 . M ) corresponding to (6. 1 . m2 . M | 1 .120) can be readily derived from the expression (6. m2 |J.4 Symmetries of the Clebsch-Gordan Coeﬃcients The Clebsch-Gordan coeﬃcients obey symmetry properties which reﬂect geometrical aspects of the operator relationship (6.113) Using the identities (6. (6. one can interchange also the operator J on the l.h. (6.103) one obtains the desired result (6.116) If one takes the negatives of the operators in (6.160 Applying this to (6. (6.114) For example. M ) corresponding to (6. J (1) .115). m1 . m1 . vice versa. m1 .116) through formula (6. (6. M ) . and .s.83). m2 ) by ( 1 .117) are again related in a simple 2 .

6.4: Symmetries of the Clebsch-Gordan Coeﬃcients

161

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

2 , m2 |J, M )

=

s

(−1)s s!( 1 − m1 − s)!( 2 + m2 − s)!

2

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

+ m1 + s)!

(6.121)

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

2 , m2 |J, M )

= (−1)

1 + 2 −J

S( 2 , m2 ,

1 , m1 |J, M )

.

(6.122)

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

1 , m1 |J, M )

=

s

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

1

×

(

1+

2 − J − s)!(J −

+ m2 + s)!

.

(6.123)

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

1 1

+

2

−J −s

(6.124)

+

2

− J − s ,

−s = J − −

1

−

2

+ s

(6.125)

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

1 , m1 |J, M )

1

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

2 2

+

2

−J s

+ m1 + s )! (6.126)

×

s !(

1 − m1 − s )!( 1

+ m2 − s )!

.

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

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

162

Theory of Angular Momentum and Spin

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

2 , m2 |J, M )

= (−1)

1 + 2 −J

S( 1 , −m1 ,

2 , −m1 2|J, −M )

.

(6.128)

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

2 , −m1 2|J, −M ) = s

(−1)s s!( 1 + m1 − s)!( 2 − m2 − s)!

2

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

− m1 + s)!

.

(6.129)

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

2 , −m2 |J, −M )

=

1

1 + 2 −J

s

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

1

− m2 + s )! (6.130)

×

1 s !( 2 + m2 − s )!(

− m1 − s )!

.

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

s 2 , m2 |J, M )

= (−1)

2 +m2

S( 2 , −m2 , J, M | 1 , m1 ) .

(6.131)

(−1)s s!( 2 + m2 − s)!(J + M − s)!

1

×

1 ( 2 + J − 1 − s)!( 1 − 2 − M + s)!(

− J − m2 + s)!

.

(6.132)

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

2 2

+ m2 − s

(6.133)

+ m2 − s ,

−s = −

2

− m2 + s

(6.134)

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

2 , −m1 2|J, −M )

163

=

2

2 +m2

s

(−1)−s ( 2 + m2 − s )!s !(J −

+ m1 + s )!

1

×

1 (J − 1 − m2 + s )!( 1 − m1 − s )!(

+

2

− J − s )!

.

(6.135)

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

6.5

Example: Spin–Orbital Angular Momentum States

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

2

m ,σ

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

2 2

(tot)

2

σ

(6.138)

space C χ1 1 =

2 2

1 0

,

χ1−1 =

2 2

0 1

.

(6.139)

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

1 σ = ±2

(6.140)

+ ( , m + 1 , − 1 , 1 |j, m) Y 2 2 2

r m+ 1 (ˆ) 2

164 or Yjm ( , |ˆ) = r

1 2

Addition of Angular Momentum and Spin

( , m − 1 , 1 , 1 |j, m) Y 2 2 2

r m− 1 (ˆ) 2 r m+ 1 (ˆ) 2

1 ( , m + 1 , − 1 , 2 |j, m) Y 2 2

.

(6.142)

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

1 (j − 2 , m − 1 , 1 , 1 |j, m) 2 2 2

=

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

(6.143)

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

=

(6.144)

=

(6.145)

=

(6.146)

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

2 2

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

2

1 2 1 2 1 2

, 1 |ˆ) r 2 , 1 |ˆ) r 2 , |ˆ) r

1 2

= = =

2

j(j + 1) Yjm (j

1 2

1 2

, 1 |ˆ) r 2

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

1 2

m Yjm (j

2

, 1 |ˆ) r 2

1 2

(j

2

1 2

)(j

1 2

+ 1) Yjm (j

, |ˆ) r

1 2

1 2

, 1 |ˆ) r 2

=

3 4

Yjm (j

, 1 |ˆ) r 2

(6.152)

6.5: Spin–Orbital Angular Momentum States

165

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

2 2 2

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

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

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

2

(6.156)

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

+π 2π

sin θdθ

−π 0

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

(6.157)

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

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

j = 1, 2 assumes the block-diagonal form

3 2

. . . ; m = −j, −j + 1, . . . + j }

(6.159)

0 −1 −1 0 0 −1 σ·r = ˆ −1 0

(6.160)

..

.

166

Addition of Angular Momentum and Spin

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

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

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

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

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

(tot)

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

(6.167)

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

2

1 σ · r Yjm (j ± 2 , 1 |ˆ) ˆ r 2

(tot)

, σ · r] = 0 , ˆ

k = 1, 2 3 .

(6.168)

σ · r J (tot) ˆ

2

2

1 Yjm (j ± 2 , 1 |ˆ) r 2

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

(6.169)

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

(6.170)

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

6.5: Spin–Orbital Angular Momentum States

167

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

(tot)

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

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

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

2 2 2 2 1

(6.174)

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

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

(6.175)

(6.176)

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

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

(6.178)

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

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

1 |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. φ) and (5.224)] 1 0 σ·r = ˆ . 1 |ˆ) at θ = 0 are r r 2 2 2 2 Yj 1 (j − 1 . φ) 2 2 2 = − j+ 1 2 4π (6. 2 |ˆ) and Yj 1 (j + 1 . hence. The result can also be stated in the compact form σ · r Yjm (j ± 1 .147.182) 0 Since σ · r. (5.188) .185) We have. φ) 2 2 2 . We want to determine ˆ its action on the wave functions f (r) Yjm (j ± 1 .148) the particular ˆ 1 1 states Yj 1 (j − 2 . given by ˆ σ · r = σ1 sin θ cos φ + σ2 sin θ sin φ + σ3 cos θ .174–5.184) 0 1 space χ 1±1 2 2 one can conclude from (6.180) and. For this purpose we consider the application of σ · r in the case of θ = 0. 1 |θ. ˆ (6.168 Addition of Angular Momentum and Spin 1 Yjm (j ± 2 . 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 . 6.180) and (6. 1 |ˆ) + f (r) σ · p Yjm (j ± 1 .182) σ · r Yj 1 (j − 1 .160). ˆ 2 2 2 2 2 2 for θ = 0. 1 |ˆ) r 2 (6. One can conclude then ˆ σ·r = ± ˆ 0 1 1 0 (6.181) 0 = j+ 1 2 4π Yj 1 (j + 1 .1 states χ 1 ± 1 2 2 2 [c. for θ = 0 (6.183) in case θ = 0 becomes in the standard representation with respect to the spin.187) 1 2 . (6. (6. 1 |ˆ) = ((σ · p f (r)))Yjm (j ± 1 . Since f (r) is independent of θ and φ follows ˆ ((σ · p f (r))) = −i (( ∂r f (r) )) σ · r . We want to demonstrate ﬁnally that the “−”-sign holds in (6. φ) = − Yj 1 (j + 1 . φ) . identiﬁed the sign of (6.186) Here (( · · · )) denotes again conﬁnement of the diﬀusion operator ∂r to within the double bracket.f. ˆ (6.180). 1 |θ = 0. have proven (6. 1 |θ = 0.181. 6. therefore. 1 |ˆ) r r r 2 2 2 2 2 2 (6.180) where the sign could depend on j. 1 |θ = 0. According to (5.177) one notes that for φ = 0 the spin-angular momentum states 2 are entirely real such that σ · r must be real as well. 1 |ˆ).

1 |ˆ) = i [ ∂r + r + ] f (r) Yjm (j 2 r r Using (6. ∂3 (1/r) = −x3 /r3 and ∂2 x3 = x3 ∂2 .g. 1 |ˆ) = r 2 2 ˆ i ∂r f (r) − f (r) σ · p σ · r Yjm (j ˆ 1 2 169 . Corresponding results are obtained for the other components of p × r and. 1 |ˆ) = 0 and σ = 2S/ . e. 1 |ˆ) r 2 1 2 (6. r r r Using ∂2 (1/r) = −x2 /r3 . x3 x2 ˆ ˆ ( p × r )1 h = −i (∂2 − ∂3 ) h (6. r 2 (6. using ∂r Yjm (j ± 1 . r Altogether we obtain.230) allows one to express ˆ ˆ ˆ ˆ ˆ σ · p σ · r = p · r + i σ · (p × r) . 1 |ˆ) . 1 |ˆ) r 2 = i ∂r + −j r 1 g(r) Yjm (j + 2 .6.195) .5: Spin–Orbital Angular Momentum States Using (6.190) can be related to the angular momentum operator. 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. h r (6.191) (6.194) ˆ ˆ where J1 is deﬁned in (5.189) The celebrated property of the Pauli matrices (5. 6..192) ˆ ˆ The operator p × r in (6. 1 |ˆ) r 2 (6.186) for both terms in (6. 1 |ˆ) r 2 (6.190) 3 h − i hr · r 1 − i r· ˆ r h = ∂r h one can conclude ˆ ˆ p · r h = −i 2 + ∂r r h (6. r 2 2 1 J ·S 2 1 ˆ σ · p f (r) Yjm (j ± 2 . To demonstrate this we consider one of its cartision components.156) this yields ﬁnally 1 ˆ σ · p f (r) Yjm (j + 2 . hence. ∂3 x2 = x2 ∂3 we obtain ˆ ˆ ( p × r )1 h = −i 1 1 (x3 ∂2 − x2 ∂3 ) h = − J1 h r r (6.197) 1 ˆ σ · p g(r) Yjm (j − 2 .53).155. we conclude the intuitively expected identity 1 ˆ ˆ p×r = − J .187) one obtains ˆ σ · p f (r) Yjm (j ± 1 .196) = i j+ ∂r + r 1 2 3 2 1 f (r) Yjm (j − 2 .193) r r 1 1 1 = −i (∂2 x3 − ∂3 x2 ) h − i h (x3 ∂2 − x2 ∂3 ) .198) . 1 |ˆ) r 2 (6.

in fact.144) for m = j. according to (6. 2 2 (6.197. consider ﬁrst the case of the largest m-value m = j.199). For m = j − 1 one can state. 1 |ˆ) = J 2 Yjm (j + 1 .143.170 Addition of Angular Momentum and Spin To demonstrate the validity of this key result we note that according to (5. .199) We want to show that eqs.99) holds ˆ (σ · p)2 = − 2 2 = ∂2 J2 r + 2 . 1 |ˆ) r 2 2 r r2 r2 j 2 + 2j + 3 2 2 2 4 = [ −∂r − ∂r + ] f (r) Yjm (j + 1 . Evaluation of Relevant Clebsch-Gordan Coeﬃcients We want to determine now the Clebsch-Gordan coeﬃcients (6. 5. 1 |ˆ) = Yj− 1 j− 1 (ˆ) χ 1 1 . 1 |ˆ) r 2 2 i 3 ˆ = σ · p [ i∂r + (j + ) ] f (r) Yjm (j − 1 .2. j) 1 2 1 2 1 2 1 2 = = 1 0 (6.e.205) which agrees with the expressions (6. as well as (6. 6. 6. 1 |j. j + . using (5.206) . 6.200) = (j + 1 )(j + 3 ).202) Yjj (j − 1 .2. Yjj−1 (j − 1 . j − 1 . For this purpose we use the construction method introduced in Sec. 1 |ˆ) r r 2 2 2 2 2 2 2 yields ˆ (σ · p)2 f (r) Yjm (j + 1 . 6.101).43). 1 . 1 |ˆ) = r 2 2 2j − 1 Yj− 1 j− 3 χ 1 1 + 2 2 2 2 2j 1 Y 1 1 χ1 1 2j j− 2 j− 2 2 − 2 (6. r (6. 1 |ˆ) = r 2 2 which agrees with (6.203) (6. − |j. i. according to (6. j 2 + 2j + 2 3 4 (6..201) 1 3 (j + )(j + ) Yjm (j + 1 .151). r ∂r2 r 2 (6. 1 |ˆ) r 2 2 r r2 and. 1 |ˆ) r 2 2 The Clebsch-Gordan coeﬃcients are then (j − 1 .198). 1 |ˆ) 2 r 2 r 2 j+3 ( 1 − j)(j + 3 ) 2 2 2 2 2 = [ −∂r − ∂r + − 2 ] f (r) Yjm (j + 1 . r 2 2 2 2 2 2 − r 2 ∂r r + J2 r2 f (r) Yjm (j + 1 . are consistent with this identity. We note ˆ (σ · p)2 f (r) Yjm (j + 1 . We begin with the coeﬃcients (6.45).230. adopting the method in Sec. (6. j) 2 2 2 2 (j − . 6.143–6. In this case holds. 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 .144) and.146).204) (6.143.

6.207) (6. 6.210)] yields (j + m)(j − m + 1) Yjm−1 (j − 1 . is obtained by applying the operator [c. (· · · |jj − 3).6. 1 .h. 6. 2 2 2 2 2j (6. (6. etc. j − 1) 2 2 2 2 which again agrees with the expressions (6.s. m − 3 . The further Clebsch-Gordan coeﬃcients (· · · |jj − 2). and J− + S− to the r. 6. Let us verify then the expression (6.f. 2 |j.202). j − 1 .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 .5: Spin–Orbital Angular Momentum States The corresponding Clebsch-Gordan coeﬃcients are then 1 (j − 1 .143. are obtained by iterating the application of (6.s.211) to the l. 1 |ˆ) = r 2 2 Applying J− (tot) j+m Yj− 1 m− 1 χ 1 1 + 2 2 2 2 2j j−m Yj− 1 m+ 1 χ 1 − 1 .208) (6. − 1 |j.143.f.. 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 . [c.210). 1 |j.144) for the Clebsch-Gordan coeﬃcients by induction. m − 1) 2 2 2 2 = j +m−1 2j (6. (6.h. 1 .206). j − 1) 2 2 2 171 = = 2j − 1 2j 1 2j (6.144) implies for j = m Yjm (j − 1 .2. j − 3 . 1 . 1 |ˆ) = r 2 2 j+m−1 Yj− 1 m− 3 χ 1 1 + 2 2 22 2j (j − m) = This implies (j − 1 .209) (j − 1 . Expression (6. (6.143. as described in Sec. 2 2 2 2 2j .137)] J− (tot) = J− + S− (6.144) for m = j − 1.210) to (6.

218) where we have replaced the quantum numbers J. all three vectors Jclass . m + 1 ) = 2 2 2 j+m+1 . − 1 |j. 1 . 2j + 1 (6.146) can be obtained from (6. the relationship (6. 1 . 6. proven (6.143). 2 .214) yields (j. In this context Jclass and Jclass play the same role. m − 1 . m. j2 . 6.144) by induction. 1 |j + 1 . −m2 | 1 . 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 = 0.116.144) for j = m − 1.145) from (6. using (6. j+m+1 . leading quantum mechanically to a symmetry of the Clebsch-Gordan coeﬃcients (JM | 1 m1 2 m2 ) with respect to exchange of 1 m1 and 2 m2 . one obtains (j + 1 . m2 .143.213) which is in agreement with (6.216) 6. The latter relationships applied together read ( 1 . 2 .217) 2j + 2 (j + 1 . 1+1 3 For 1 (j.143. m1 .214) 2 2 +1 ( 3 . m + 1 ) = 2 2 2 2 and.143) by applying the symmetry relationships (6.215). They are related in a simple manner to the ClebschGordan coeﬃcients j1 j2 j3 m1 m2 m3 = (−1)j1 −j2 +m3 (2j3 + 1)− 2 (j3 − m3 |j1 m1 . M. m3 ) = (−1) × 2 + 3 − 1 + 2 +m2 × 2 . m1 . We have. 1 . m2 | 3 .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 . j2 m2 ) 2 . m. hence. 2j + 2 (6. 1 . m1 ) (6. m − 1) 2 2 2 2 = (6. (1) (2) (−tot) Obviously. m + 1 .144).172 Addition of Angular Momentum and Spin j −m+1 2j (j − 1 . However. m + 1 . m3 . m) = 2 2 2 2 Similarly. − 1 |j. by the set j1 . one can obtain expression (6. m1 .215) which follows from (6. The Clebsch-Gordan coeﬃcients (6. 1 . 1 |j + 1 . − 1 |j. (1) (2) (−tot) The coeﬃcients which are the quantum mechanical equivalent to Jclass + Jclass + Jclass = 0 are the 3j–coeﬃcients introduced by Wigner. Jclass and Jclass play equivalent roles. m2 1 (6. j3 . to reﬂect in the notation the symmetry of these quantities. m3 . m) 2 2 2 2 2j + 1 (6. 6.120).

The reader may note that the entries of the Regge-symbol.218) relating 3jcoeﬃcients and Clebsch-Gordan coeﬃcients.e. These symmetry operations relate altogether 72 3j–coeﬃcients. i. states that j1 . one can exchange columns and one can reﬂect at the diagonal (transposition). discovered by Regge. According to the deﬁnition of the 3j–coeﬃcients one would expect symmetry properties with respect to exchange of j1 . j2 . To represent the full symmetry one expresses the 3j–coeﬃcients through a 3 × 3–matrix. In case of a non-cyclic exchange of rows and columns the 3j–coeﬃcent assumes a prefactor (−1)Σ . there exist even further symmetry properties.102) of the Clebsch-Gordan coeﬃcients. the Regge-symbol. as follows −j1 + j2 + j3 j1 − j2 + j3 j1 + j2 − j3 j1 j2 j3 j1 − m1 j2 − m2 j3 − m3 . anti-cyclic exchange and of a sign reversal are stated. for which no known classical analogue exists.g. The Regge symbol reﬂects a remarkable degree of symmetry of the related 3j–coeﬃcients: One can exchange rows. = (6.21. 6. For this purpose one can use the symmetry transformations to place the smallest element into the upper left corner of the Regge symbol.6: The 3j–Coeﬃcients We ﬁrst like to point out that conditions (6.102) are obtained each through the diﬀerence of two entries of the Regge-symbol. However. except when all elements are non-negative integers and each row and column has the same integer value Σ = j1 + j2 + j3 . −j1 + j2 + j3 . j3 form the sides of a triangle and the condition is symmetric in the three quantum numbers.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 .. m2 . The two integer entries J − 1 − m2 and J − 2 + m1 in (6. with respect to alltogether 12 symmetry operations. 173 (6. m3 . 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 (6.219) (6. j2 .221) j1 j2 j3 −m1 −m2 −m3 where the the results of a cyclic. Because of its high degree of symmetry the Regge symbol is very suited for numerical evaluations of the 3j–coeﬃcents. Assuming this placement the Regge symbol can be determined as follows (n11 is the smallest element!) n11 n11 n12 n13 n12 !n13 !n21 !n31 ! n21 n22 n23 = (−1)n23 +n32 sn (6. m3 and with respect to sign reversals of all three values m1 . The Regge symbol also vanishes in case that two rows or columns are identical and Σ is an odd integer. m2 and j3 .223) (Σ + 1)!n11 !n22 !n33 !n23 !n32 !)) n=0 n31 n32 n33 . √ safe for the prefactor 2j3 + 1 which is cancelled according to the deﬁnition (6.6. the so-called triangle condition. are identical to the integer arguments which enter the analytical expression (6. m1 .222) m1 m2 m3 j1 + m1 j2 + m2 j3 + m3 The Regge symbol vanishes.220) The latter condition |j1 − j2 | ≤ j3 ≤ j1 + j2 . In this way the values of 12 3j-coeﬃcients are related. e.

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. (6.232) . (6.223-6. m1 + m2 ) (2 − m1 )! (2 + m1 )! √ 6 (2 − m)! (2 + m)! (1 − m2 )! (1 + m2 )! 1 ≥ |m2 | ∧ 2 ≥ |m| ∧ 2 ≥ |m1 | (6.230) 1 1 (1m| m1 m2 ) = 2 2 √ (−1)2m1 −2m2 3 δ(m.228) √ (−1)2m1 −2m2 7 δ(m. m) δ(−m1 .229) 1 1 (0m| m1 m2 ) = 2 2 i(−1)1−m2 δ(0.226) by means of a symbolic manipulation package (Mathematica) (1m|2m1 1m2 ) = (−1)1+m+m1 δ(m. 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. m2 ) √ 2 1 ≥ |m1 |) 2 (6.226) We like to state ﬁnally a few explicit analytical expressions for Clebsch-Gordan coeﬃcients which were actually obtained using (6.224) (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 (6.1.225) sn−1 .227) (2m|2m1 1m2 ) = (−1)m+m1 (m + 2m2 ) δ(m. m1 + m2 ) (2 − m1 )! (2 + m1 )! √ 10 (1 − m)! (1 + m)! (1 − m2 )! (1 + m2 )! 1 ≥ |m| ∧ 1 ≥ |m2 | ∧ 2 ≥ |m1 | (6.231) Here is an explicit value of a Clebsch-Gordan coeﬃcient: 1 1 (70 2 −15 1 |120 −10 60 2 −5 1 ) = 2 2 √ √ √ 4793185293503147294940209340 127 142 35834261990081573635135027068718971996984731222241334046198355 0.

233) .+ 2 } the matrix has the blockdiagonal form D(ϑ) = 1·1×1·1 ¤ -10 60 700 m 10-m -40 -20 0 20 40 60 m 1·3×1·3 . . 2 |Ω1 . . . Similarly. . Irreducible Representation We had stated above that the spherical harmonics Y m (Ω)). 2. i. .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. eigenfunctions of the single particle angular momentum operators J 2 and J3 . If we deﬁne the matrix representation of R(ϑ) by D(ϑ).6. . 2 = 1. (2 (2 1 2 + 1) (2 + 1) × (2 1 2 + 1) + 1) . (6. .e. m2 = − 2 . . i. m1 = − 1 . .e.1: Oscillatory behavior of 3j-coeﬃcients. Ω2 ) provide the irreducible representation for the rotation R(ϑ) deﬁned in (6.5). provide the irreducible representation for D(ϑ). .. rotations in 2–particle function space.. . the 2–particle total angular momentum wave functions YJM ( 1 .+ 1 . 1 . . then for a basis {Y 1 m1 (Ω1 )Y 2 m2 (Ω2 ). . the rotations in single particle function space.

6. (2( 1 + 2 ) + 1) ×(2( 1 + 2 ) + 1) Partitioning in smaller blocks is not possible. .233) is further block-diagonalized as follows + 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. Ω2 ). M = −J. . 2 |Ω1 . 1).234) Exercise 6.3: Two triplet states |1m1 (1) |1m2 (2) are coupled to an overall singlet state Y00 (1. . Show that the probability of detecting a triplet substate |1m1 (1) for arbitrary polarization (m2 – 1 value) of the other triplet is 3 . . 1 .234) . . (6.. .6. 2 = 1. .1: Prove Eqs. each of the blocks in (6. .233. .176 Addition of Angular Momentum and Spin 1 For the basis {YJM ( 1 . Exercise 6. . Exercise 6. 2.6. J = | 1 − 2 |. .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. .

g.7: Tensor Operators 177 6. −k + 1. . e. q = −k. (k) (6. that in case (i) T is an operator C∞ ( ) → C∞ ( ) acting on single particle wave functions. This implies. k} such that k Tkq = q =−k (k) Dq q (ϑ) Tkq . for example.g. . S 2 = S1 + S2 + S3 or any other polynomial of Sk . some of the r r examples considered below involve tensor operators T of this type. 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) .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. the operator (5. e. that T belongs to a family of operators {Tkq . (6. The operator T may also act on multi-particle states like Y 1 m1 (ˆ1 )Y 2 m2 (ˆ2 ). . (k) (6. e. The latter implies that T transforms like an angular momentum or spin state | m .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 . 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. 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).237) are called tensor operators of rank k. .222) in case (ii) of single particle spin operators. −k + 1.223). . Rotations transform |ψ as |ψ = R(ϑ)|ψ and T |ψ as R(ϑ)T |ψ . k} with the transformation property (6.42) in case (i) of the position representation of single particle wave functions or the operator (5.238) . In fact. . Note that in the examples mentioned the operator T would be deﬁned within the same representation as R(ϑ).236) In this equation Dq q (ϑ) denotes the rotation matrix Dq q (ϑ) = kq |R(ϑ)|kq . i.g. q = −k.e. Examples of Tensor Operators The multiplicative operators C∞ ( ) → C∞ ( ) Y ( ) = def Y ( ) (6. ∂x2 3 1 2 2 2 2 case (ii) the operator T could be a spin operator Sk deﬁned in (5.6. Such operators T can be characterized through their behaviour under rotational transformations.237) The operators T ∈ {Tkq .236.

239) These operators can be expressed in terms of the coordinates x1 .237). For the following it is important to note that the rotation matrix elements (6. This choice of parametrization allows us to derive conditions which are equivalent to the property (6. we will assume presently that the rotation is chosen as follows R(ϑ) = exp ( ϑ+ L+ + ϑ− L− + ϑ3 L3 ) (6. . Exercise 6.235 . but are far easier to ascertain for any particular operator. In fact.6. 1.1: Show that the following set of spin operators T00 T1±1 T10 T2±2 T2±1 T20 = 1 = 1 √ S± 2 S± 2 = S3 = = = (S3 S ± + S ± S3 ) 2 2 (3S3 − S 2 ) 3 are tensor operators of rank 0.7. but rather any rotation and any parametrization can be assumed. Exercise 6.203).178 are tensor operators of rank k.240) 1 where we have deﬁned ϑ± = 2 (ϑ1 iϑ2 ) and L± = L1 ± iL2 .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. The fact that these operators form tensor operators of rank 1.3.237) do not require that the rotation R−1 (ϑ) is expressed in terms of Euler angles according to (5. x3 . 3 follows from the transformation properties of the spherical harmonics derived in Section 1. Examples are Y00 Y1±1 Y10 Y2±2 Y2±1 Y20 1 = √ 4π Theory of Angular Momentum and Spin 3 3 r sinθ e±iφ = (x1 ± ix2 ) 8π 8π 3 3 = r cosθ = x3 4π 4π 1 15 2 2 ±2iφ 1 15 = r sin θ e = (x1 ± ix2 )2 4 2π 4 2π 15 2 15 = r sinθ e±iφ = (x1 ± ix2 ) x3 8π 8π 5 2 3 2 1 5 3 2 r2 = r cos θ − = x3 − 4π 2 2 4π 2 2 = (6. 2. 2. x2 .7.

Exercise 6. .6. Show that S (1) · S (2) is a tensor operator of rank 0 in the space of the products of the corresponding spin states | 1 m1 (1) | 1 m2 (2) . (6.6.235 .7.246) (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) .248) a tensor operator? Exercise 6. Tkq ] = = = Tkq+1 Tkq−1 q Tkq .7.6: Form tensor operators of rank 1 in terms of the three components of acting on the space of 1-particle wave functions. Tkq ] [J− . The property (6.7.246.3: Derive Eqs. (6.235 . Expressing the results in terms of the angular momentum operators J+ .6.7.247) These properties often can be readily demonstrated for operators and the transformation properties (6.5: Consider a system of two spin. 6. (k + q + 1)(k − q) (k + q)(k − q + 1) (6.244) [L+ . ϑ3 .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 . ϑ− . For this 2 2 purpose state ﬁrst the proper rotation operator R(ϑ) and note then that S (1) · S (2) commutes with 1 the generators of the rotation of | 2 m1 (1) | 1 m2 (2) . (6. 1 (6. Tkq ] = = δqq and by subtracting Tkq on both sides of the k ϑ+ kq |L+ |kq Tkq . (0. 0)T . Tkq ] = −iTk q+1 (k + q + 1)(k − q) .242) from which follows by means of kq |1 1|kq equation ϑ+ [L+ . 0. q =−k q+1 (6.237) yields ﬁrst k R(ϑ) Tkq R −1 (ϑ) = q =−k kq |R(ϑ)|kq Tkq . We ﬁrst consider a rotation with ϑ = (ϑ+ .80) follows kq |L+ |kq = −i (k + q + 1)(k − q)δq and. 0. 2 Exercise 6. ϑ3 )T . Exercise 6. 0)T in case of small ϑ+ . J3 yields [J+ .237) be assumed then.243) From (5.7: Tensor Operators 179 The conditions can be derived if we consider the property (6. ϑ− .237) for transformations characterized by inﬁnitesimal values of ϑ+ . hence. Tkq ] [J3 .235 . J− .247).4: Is the 1-particle Hamiltonian 2 H = − 2 2m + V (|r|) (6.6. Similar equations can be derived for inﬁnitesimal rotations of the form ϑ = (0.245) (6.

one can show that Φ 1 m1 (k. i. J2 . i. 2 |γ2 ) = ( 1 m1 |kq 2 m2 ) Tkq | 2 m2 γ2 . (6.247).180 Theory of Angular Momentum and Spin 6. J3 are the generators of the rotation R(ϑ).m2 2 m2 ) (q + m2 ) Tkq | 2 m2 γ2 | 2 m2 γ2 (6. 2 |γ2 ) are orthogonal to | 1 m1 in case of diﬀerent quantum numbers 1 . in fact. 2 |γ2 ) is an eigenstate of J 2 with eigenvalue 2 1 ( 1 + 1).m2 ( 1 m1 |kq m1 q. 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. J3 | 2 m2 γ2 = m2 | 2 m2 γ2 .252) ∼δm1 q+m2 = ( 1 m1 |kq 2 m2 ) Tkq Similarly. construct states Φ 1 m1 (k.e. which is characterized also by a set of other quantum numbers γ1 which are not aﬀected by the rotational transformation R(ϑ).249) which demonstrates. possibly the total angular momentum– spin state of a compositie system.8 Wigner-Eckart Theorem A second important property of tensor operators Tkq beside (6.8. .1: Show that Φ 1 m1 (k. | 1 m1 γ1 denotes an angular momentum (spin) state.18) are deﬁned through Φ 1 m1 (k. J3 . the stated property.250) q. m1 .6.m2 ( 1 m1 |kq 2 m2 ) (J3 Tkq − Tkq J3 + Tkq J3 ) | 2 m2 γ2 = qTkq = q.e. One can. 2 |γ2 ) which correspond to total angular momentum states. 1 m1 |Φ 1 m1 (k.237) is that their matrix elements 1 m1 γ1 |Tkq | 2 m2 γ2 obey simple relationships expressed in terms of Clebsch-Gordan coeﬃcients. 2 |γ2 ) = Cδ 1 1 δm1 m1 2 1( 1 (6. (6.21) of Clebsch-Gordan coeﬃcients J3 Φ 1 m1 (k. The relationships among the matrix elements 1 m1 γ1 |Tkq | 2 m2 γ2 are stated by the Wigner–Eckart theorem which we will derive now. J3 | 2 m2 γ2 = m2 γ2 | 2 m2 and the property (6. Before we proceed we like to point out that the states | 2 m2 γ2 are also eigenstates of J 2 . 2 |γ2 ) is an eigenstate of J 2 with eigenvalue + 1). 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 ) = q. hence.253) Exercise 6. J 2 | 2 m2 γ2 = 2 2( 2 + 1) | 2 m2 γ2 . i.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 .e.251) The corresponding property for Φ 1 m1 (k. The hermitian property of J3 and J 2 implies that states Φ 1 m1 (k. 2 |γ2 ) can be shown readily as follows using (6.235 . These states according to (6. behave like |kq | 1 m1 .

q .32) Tkq | 2 m2 γ2 = and orthogonality property (6. One can then evaluate also the corresponding Clebsch-Gordan coeﬃcient ( 1 m1 |kq 2 m2 ) and determine 1 .256) and noting that the operator adjoint to J+ is J− . 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 ) = (6.254) = 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) ( 1 m1 |kq 2 m2 ) . To prove this property we consider 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) for a diﬀerent m1 value. say m1 + 1. γ2 . e. γ2 ( 1 m1 |kq (6.260) Exercise 6. γ2 (−1)k− 2 + 1 ( 1 m1 |kq 2 m2 ) (6. can be evaluated as easily as possible.g.261) . γ2 = 2 1 +1 1 m1 . m2 . γ2 is determined by applying (6. γ1 ||Tk || 2 . (6. γ1 ||Tk || 2 . m1 = q = m2 = 0. γ1 |Tkq | 2 m2 .6. γ1 ||Tk || 2 .h.8: Wigner-Eckart Theorem 181 In order to evaluate the matrix elements (6. its m1 –dependence being expressed solely through a Clebsch-Gordan coeﬃcient.253) 1 m1 γ1 |Tkq | 2 m2 γ2 1 m1 γ1 |Tkq | 2 m2 γ2 we express using the equivalent of 1 m1 ( 1 m1 |kq 2 m2 ) Φ 1 m1 (k 2 |γ2 ) (6. γ1 |Tkq | 2 m2 . In order to express the m1 – independence explicitly we adopt the following notation 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) = (−1)k− 2+ 1 √ 1 2 1 +1 1 .e. i. γ1 ||Tk || 2 . (6.s.258) We can then ﬁnally express the matrix elements of the tensor operators Tkq as follows 1 m1 . for which the l. γ2 k− 2 + 1 √ 1 2 m2 ) (−1) 2 1 +1 = 1 .255) At this point the important property can be proven that 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) is independent of m1 .2: Determine the matrix elements of the gradient operator d3 rF (r) G(r) of the type (6.259) The socalled reduced matrix element 1 .8.257) which establishes the m1 –independence of 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) .259) to a combination of magnetic quantum numbers m1 . Using | 1 m1 + 1γ1 = 1 ( 1 + m1 + 1)( 1 − m1 ) J+ | 1 m1 γ1 (6. the matrix elements 1 m1 γ1 |Tkq | 2 m2 γ2 can be reduced to an m1 –independent factor.

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

3) describes the motion of a charge in a uniformely charged sphere and can be employed to describe the motion of protons and neutrons in atomic nuclei1 .4) is by far the most relevant of the four choices. for example. The fourth potential V (r) = − Z e2 r (7. in case of the so-called bag model of hadronic matter. The ﬁrst potential V (r) = 0 0 ≤ r ≤ R ∞ r > R (7.e. The second potential is of the square well type −Vo 0 ≤ r ≤ R . Talmi (Harwood Academic Publishers. (7.1) conﬁnes a freely moving particle to a spherical box of radius R. Potential (7. molecules. 1993) 1 183 .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). Switzerland. The potentials (7. 7. in potentials which are solely a function of r and are independent of the angles θ. Poststrasse 22. See.1.2) V (r) = 0 r > R The third potential V (r) = 1 mω 2 r2 2 (7. 7000 Chur. Simple Models of Complex Nuclei / The Shell Model and the Interacting Boson Model by I.4) governs the motion of electrons in hydrogen-type atoms. The corresponding wave functions serve basis functions for multi-electron systems in atoms. and crystals.2) serve as schematic descriptions of quantum particles.. It leads to the stationary electronic states of the hydrogen atom (Z = 1). i. Four examples will be studied. φ. for example. The potential (7.3) describes an isotropic harmonic oscillator .

( + 1) ψE.1 Radial Schr¨dinger Equation o ˙ mv = − er ∂r V (r) . (7. Accordingly. In fact.e.5) In the case of an angular independent potential angular momentum J = m r × v is a constant of ˙ motion.6) ˙ Since Jk is also equal to the poisson bracket {H. using (7. Jk } where H is the Hamiltonian.15) (7.m (r) can be chosen as simultaneous eigenstates of the Hamiltonian ˆ operator H as well as of J 2 and J3 . J3 ψE. (7. Jk } = 0 ..184 Spherically Symmetric Potentials 7.m (r) . one can state ˙ J = m r2 θ . . 2 .87) for Jk as well as the commutation property (5.5) and r = v. one can conclude ˙ Jk = {H. 2 2mr2 (7.9) − 2m 2m r 2mr2 the expressions (5. 2. the time variation of J can be written. θ for the distance from the origin and for the angular position. 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 .99)] 2 1 2 J2 2 2 = − ∂r r + . (5.m (r) = = = E ψE. ˆ dt (7. 2.13) .12) . k = 1.m (r) .14) (7. For this purpose one concludes ﬁrst that the conservation of angular momentum J implies a motion of the particle conﬁned to a plane. Jk ] = 0 .85– 5.11) (7.7) ˆ The correspondence principle dictates then for the quantum mechanical Hamiltonian operator H and angular momentum operators Jk ˆ [H. 2 . In classical mechanics one can exploit the conservation of angular momentum to reduce the equation of motion to an equation governing solely the radial coordinate of the particle.m (r) (7. d ˙ J = v × mv + r × mv = 0 + r × er (−∂r V (r)) = 0 . Employing in this plane the coordinates r. i. .10) (7. stationary states ψE. 3.m (r) m ψE. k = 1.f. J ψE.61) of J 2 and Jk . 3 . ˆ H ψE.m (r) .8) This property can also be proven readily employing the expression of the kinetic energy operator [c. ˆ A classical particle moving in a potential V (r) is governed by the Newtonian equation of motion (7. (7.

. (7. for the wave function of a quantum mechanical particle a diﬀerential equation which governs solely the r-dependence.m (r) Y m (θ.20) in the form of the one-dimensional Schr¨dinger equation o 2 − where 2m 2 ∂r + Veﬀ (r) − E φE (r) = 0 . It can be solved by integration of 1 2 2 J2 r = ± ˙ E − V (r) − .4.10). together with the original potential.20) by −2mr/ 2 yields the so-called radial Schr¨dinger equation o 2 ∂r − ( + 1) − U (r) − κ2 r2 r vκ.. using (7.16) m 2mr2 Once.m (r) 1 2 J2 ∂r r + + V (r) − E 2m r 2mr2 the functional form ψE. .13). (7.7.21) Veﬀ (r) φE (r) = = V (r) + r vE. .22) which is the original potential 2 ( +1) V (r) with an added rotational barrier potential 2mr2 .m (r) = 0. (7.23) This demonstrates that the function r vE. equations (7. in the present case. but can also trap particles in the latter space giving rise to strong scattering resonances (see Section ??). ( + 1) 2mr2 .20) Since this equation is independent of the quantum number m we drop the index m on the radial wave function vE. 7.11. can exclude particles from the space with small r values.11).24) . (r) = 0 . ∞[ governed by the eﬀective potential (7.18) = vE. 2 (7. φ) are the angular momentum eigenstates deﬁned in Section 5.1: Radial Schr¨dinger Equation o 185 This is a diﬀerential equation which governs solely the radial coordinate. − Adopting for ψE. Multiplying (7. One can write (7. .m (r) describes the radial motion as a one-dimensional motion in the interval [0.m .m (r) and E . by integration of ˙ θ = J .m vE. (7. i.17) In analogy to the classical description one can derivem. . mr2 (t) (7.12) are obeyed and one obtains for (7. . This barrier.m (r) = 0. φ) (7. . r(t) is determined the angular motion follows from (7.10) − 1 2 ∂ r + 2m r r 2 2 ( + 1) + V (r) − E 2mr2 .9) one can write the stationary Schr¨dinger o equation (7.22) (7.m (r) .e.m (r) 2 . Employing the kinetic energy operator in the form (7.m ψE. (7.19) where Y m (θ.

In this case the solution is governed my ( + 1) 2 r vκ. accordingly.182) for Y00 . (r) (7. the term Br−( +1) would contribute |B|2 0 dr r−2 +1) (7.28) (7. Pascual (Springer. for > 0 is not integrable. 2nd Ed. at r = 0 is found in the excellent monographs Quantum Mechanics I. according to a well-known result in Classical Electromagnetism3 . This follows for > 0 from consideration of the integral which measures the total particle density. (r) ∼ A r or vκ.20) one needs to specify proper boundary conditions2 . (r) = 0 . hence. Berlin. . Galindo and P. (7. κ assumes real values. ψE. (r) ∼ A r + B r− −1 +1 + B r− (7. 1975). Boundary Conditions In order to solve (7. .25) (7. New York.31) to the complete wave which.30) and. 2 2 2m 2 ψE.D.20) the index E by the equivalent index κ. The radial part of this integral is ∞ 0 2 dr r2 vκ.32) The total kinetic energy resulting from this contribution is.” by J. We replaced in (7. for example. .m (r) ∼ √ B .186 where we deﬁned U (r) κ2 = = − − 2m 2 Spherically Symmetric Potentials V (r) E (7. 4π|r| (7. assumes the general form r vκ. using the expression (5.26) 2m 2 In case E < 0.29) Only the ﬁrst term is admissable. r → 0 (7.27) ∂r − r2 and.m (r) ∼ √ 4π B δ(x1 )δ(x2 )δ(x3 ) . in particular. see. Jackson (John Wiley. ”Classical Electrodynamics. 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).33) A detailed discussion of the proper boundary conditions. For r → 0 one may assume that the term ( + 1)/r2 becomes larger than the potential U (r). II by A. For = 0 the contribution of Br−( function is.

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).

50). This holds for −1 and one can conclude +1 consecutive integrations by part d izx e dx (7. (7. One can expand the former solution in terms of solutions (7. the wave function does not depend on φ. and using the Rodrigues formula for Legendre polynomials (5. namely.99) We want to determine the expansion coeﬃcients b .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. one given o by (7. the surface term ∼ [· · ·]+1 vanishes. −1 .106).194 Generating Function of Spherical Bessel Functions Spherically Symmetric Potentials The stationary Schr¨dinger equation of free particles (7. can be written exp(ikr cos θ). φ).50). i.43) has two solutions. One can show d −1 2 (x − 1) ∼ (x2 − 1) × polynomial in x dx −1 (7.98) by ∞ eik3 x3 = =0 b j (kr) P (cos θ) .e. In this case the expansion on the r. in case of a free particle moving along the x3 -axis one expands eik3 x3 = .104) dx eizx P (x) −1 = (−1) 2 ! (iz) = 2 ! +1 dx (x2 − 1) −1 +1 dx (1 − x2 ) eizx . (7.98) The l.100) Deﬁning x = cos θ.h. z = kr.103) and.102) −1 d −1 +1 −1 dx −1 (x2 − 1) .h. For example.s.. φ) .178) are given in terms of Legendre polynomials P (cos θ) one can replace the expansion in (7. (7.179) yield from (7. 2 +1 (7. of (7.99) +1 d cos θ eikr cos θ P (cos θ) = b j (kr) −1 2 . according to (5.45) and one given by (7. φ) for non-vanishing m has a non-trivial φ-dependence as described by (5.44. hence.98) does not involve any non-vanishing m-values since Y m (θ.m a mj (kr) Y m (θ.s. 7. The orthogonality properties (5.150) one obtains +1 +1 1 ∂ dx eizx P (x) = dx eizx (x2 − 1) . Since the spherical harmonics Y 0 (θ.

h.2: Free Particle Comparision with (7. The identity (7. 2 +1 −1 dx (1 − x2 )ν− 2 eizx .105) This expression allows one to determine the expansion coeﬃcients b .107) 1 2 (2 )! 1 · 3 · 5 · · · (2 + 1) 2 + 1 1 · 2 · 3 · 4 · · · (2 − 1) · 2 (7.f.113) dt (1 − t2 )ν− 2 eizt .h.99). as the generating function of the spherical Bessel functions. Comparision with the l. in particular.69)] b z 2 (iz) = 1 · 3 · 5 · · · (2 + 1) 2 + 1 2 ! +1 dx (1 − x2 ) .h. using ν = Jν (z) = √ 1 πΓ(ν + 1 ) 2 z 2 ν + 1. −1 (7.117) one can write the r.96) one can express this. 1 (7.100) gives b j (kr) 2 (iz) = 2 +1 2 ! +1 195 dx (1 − x2 ) eizx .108) where the last factor is equal to unity.110) One refers to the l. −1 (7.106) yields ﬁnally b = i (2 + 1) or.114) .111) Employing (7. (7. Integral Representation of Bessel Functions Combining (7.7.105) and (7.112) We want to consider the expression Gν (z) = aν z ν fν (z) where we deﬁne fν (z) = C (7. z i or i (2 + 1) z (2 )! 2 2 2 +1 2 ! [1 · 3 · 5 · · · (2 − 1)] 1 (7. 1 (7.109) eikr cos θ = =0 i (2 + 1) j (kr) P (cos θ) .105) must hold for all powers of z.s. (7. −1 (7.109) results in the integral representation of j (z) j (z) = (z) 2 +1 ! +1 dx (1 − x2 ) eizx . after insertion into (7. for the leading power x [c.s. of (7.s. ∞ (7.106) Employing (5.

114. t2 . In fact.124) .116) fν (z) = − or fν (z) = − From this we can conclude fν (z) + 2ν + 1 fν (z) + fν (z) = 0 .196 Spherically Symmetric Potentials Here C is an integration path in the complex plane with endpoints t1 .118) z 2ν + 1 fν (z) + fν (z) . 1 j = 1.121) (7.93).93) as long as the integration path in (7.120) imply also the property (2ν + 1) fν (z) + z fν+1 (z) = 0 .116) In case that the endpoints of the integration path C satisfy (1 − t2 )ν+ 2 eizt one can write (7. To prove this we note fν (z) = i C dt (1 − t2 )ν− 2 t eizt .120) such that we can conclude that Gν (z) for proper integration paths is a solution of (7.117) dt (1 − t2 )ν− 2 t2 eizt 1 (7. Exercise 7.113) obeys the Bessel equation (7. obeys Bessel’s equation (7.115) Integration by part yields fν (z) = − i 2ν + 1 (1 − t2 )ν+ 2 eizt 1 t2 t1 − z 2ν + 1 dt (1 − t2 )ν+ 2 eizt .117). C 1 (7.120) z 2ν + 1 dt (1 − t2 )ν− 2 eizt + C C 1 1 t2 t1 = 0 (7. 1 (7.119) We note that equations (7. for properly chosen endpoints t1 . (7. We consider now the functions u(j) (z) = √ 1 πΓ(ν + 1 ) 2 z 2 ν Cj dx (1 − x2 )ν− 2 eizx .122) (7. z (7.121). t2 of the integration paths C.2. (7. Gν (z). We can now demonstrate that Gν (z) deﬁned in (7. 7.113) satisﬁes (7. 2.123) Insertion of these identities into Bessel’s equation leads to a diﬀerential equation for fν (z) which is identical to (7. 3.93) for arbitrary ν. 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. 4.4: Prove (7.

one can derive (2) Hν (z) = e −i(z−πν/2−π/4) ∞ 0 √ ds [s(1 − is/2)]ν− 2 e−zs . therefore. (2) Hν (z) = −2 u(4) (z) .130) We note that the endpoints of the integration paths C2 and C4 .131) According to (7.117) and. u(3) (z) ≡ 0.134) Similarly.126) −1 ≤ s ≤ 1 (7.130) holds Jν (z) = (1) 1 2 1 (1) (2) Hν (z) + Hν (z) .124) vanishes along the whole path C3 and.132) For Hν (z) one derives.93).124) is analytical in the part of the complex plane surrounded by the path C we conclude 4 u(j) (z) ≡ 0 j=1 (7.112) shows u(1) (z) = Jν (z). Since the integrand in (7.133) = e i(z−πν/2−π/4) √ 2 zν π Γ(ν + 1 ) 2 2 zν π Γ(ν + 1 ) 2 √ √ ∞ 0 ds [s(1 + is/2)]ν− 2 e−zs . hence. Comparision with (7. u(2) (z) and u(2) (z) are both solutions of Bessel’s equation (7. using t = 1 + is.128) C = C1 ∪ C2 ∪ C3 ∪ C4 is a closed path. for Rez > 0 and ν ∈ R obey (7. we introduce the so-called Hankel functions (1) Hν (z) = −2 u(2) (z) . dt = i ds. .127) 0 ≤ s < ∞ (7. 1 (7. 1 (7.7.135) are known as the Poisson integrals of the Bessel functions.129) The integrand in (7. the integral expression (1) Hν (z) 1 1 1 1 (7.125) 0 ≤ s < ∞ (7. (7. Following convention. (7. Accordingly. (7.134) and (7. one can state Jν (z) = − u(2) (z) + u(4) (z) .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. 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) (7.

e.135) which read for ν = + 1 2 H where f The binomial formula yields f (1) 2 (1) 2 1 +2 (1. we obtain H (1) 2 1 +2 (1) 2 (z) = r=0 ( + r)! ! r!( − r)! ± i 2 r 1 z +r+1 (7.2) (z) +1 2 in terms of z −1 such that (z) = e±i[z − ( ∞ +1) π ] 2 2z z ( 1 ) f 2 (z) π ! (7. For the Wronskian connected with the Bessel equation (7.93) holds the identity 1 W = − W z (7. i. (7. h) must be a non-vanishing function. We employ for this purpose the Poisson integrals (7.140).134) and (7. (7.143) . πz (7. Hν (z) as well as H−ν (z) are solutions of this equation. The general solution of (7.139) (z) = 2 ±i[z − ( e πz +1) π ] 2 r=0 ( + r)! r!( − r)! i ± 2z r (7.137) (z) = r=0 r ± i 2 r 0 ∞ ds s +r e−zs .142) and h(z) = H (2) (z) +1 2 one can identify the constant c by using the leading terms in the expansions (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.93) only as ν 2 . z (7. For such solutions g(z).141) (1) (1) the derivation of which follows the derivation on page 192 for the Wronskian of the radial Schr¨dinger equation.140) Bessel Functions with Negative Index Since ν enters the Bessel equation (7.136) (1) 2 (z) = 0 ds [s(s ± is/2)] e−zs . hence. the Wronskian W (g. As a second order diﬀerential equation the Bessel equation has two linearly independent solutions.138) The formula for the Laplace transform of sn leads to f and. converge fast for |z| → ∞. One obtains W (z) = − 4i ..141) is o W (z) = − In case of g(z) = H (1) (z) +1 2 c . h(z) to be linearly independent.

2 (1. 7..132) follows n (z) = 1 2 π (−1) 2z +1 H−( (1) 1 +2) (z) + H−( (2) +1) 2 (z) .147) this can be written n (z) = 1 2i h (1) (z) − h (2) (z) . in fact.148) Following the arguments provided above (see page 193) the functions h (z) are solutions of the radial Schr¨dinger equation of free particles (7.153) .152) (7. (7. 7.148) holds for the o regular spherical Bessel function j (z) = 1 (1) (2) [ h (z) + h (z) ] . (7. H (1.140). 7. (7. (7. 7.97.145) This equation can hold only for B = 0 and A = exp[i( + 1 )π]. 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| → ∞ .2) (2) +1) 2 (1) +1) 2 (z) = i (−1) H (1) (z) +1 2 .147) (z) = π (1.146.2) H + 1 (z) . We conclude 2 H−( Simarly. (7. 7.2) (7. (7.2: Free Particle i. 7.2) (z) +1 2 199 are. linearly independent. According to (7.132.7.149.97) one deﬁnes the spherical Hankel functions h (1. From (7.96.151) Equations (7.90.151) are equivalent to h h (1) (2) (z) (z) = = j (z) + i n (z) j (z) − i n (z) .97).2) (7.146) (z) = −i (−1) H (2) 1 (z) +2 . B can be obtained from the asymptotic expansion (7. H−ν (z) = A H (1) (1) (z) +1 2 One can then expand + BH (2) (z) +1 2 .53).e. 2z 2 (1. one can show H−( Spherical Hankel Functions In analogy to equations (7. (7.144) The expansion coeﬃcients A.149) We want to establish also the relationship between h (z) and the irregular spherical Bessel function n (z) deﬁned in (7.150) According to (7.

2z 2 z sin(z − π/2) z cos(z − π/2) − .2) (z).163) (7.159) (7. 7. From (7. are j (z) n (z) = = (7. are complex conjugates.148) one obtains readily h (1) 2 (1.164) .155) (z) and To determine j (z) and n (z) we note that for z ∈ R the spherical Hankel functions h (2) h (z). it follows z ∈ R : Using p (z) = Re r=0 j (z) = Re[h (1) (z)] . Hence. = sin(z − π/2) and sin[z − ( + 1) π ] . (1) (7.160) Employing cos[z − ( + 1) π ] .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. j (z) (z) = ( i) z +1 e±iz r=0 ( + r)! r!( − r)! ± i 2z r (7. at large |z|.154) the identities j (z) n (z) = = sin[z − ( + 1) π ] cos[z − ( + 1) π ] 2 2 p (z) − q (z) z 2z 2 cos[z − ( + 1) π ] cos(z − π/2) 2 q (z) + p (z) . ne ll(z) = Im[h (1) (z)] . = −cos(z − π/2) results in 2 2 the alternative expressions j (z) n (z) = = sin(z − π/2) cos(z − π/2) p (z) + q (z) z 2z 2 cos(z − π/2) sin(z − π/2) q (z) − p (z) . 2 2z z (7.200 Asymptotic Behaviour of Spherical Bessel Functions Spherically Symmetric Potentials We want to derive now the asymptotic behaviour of the spherical Bessel functions h and n (z).154) The leading term in this expansion.161) (7.140) and (7.148). is h (1) 2 (z) = ( i) z +1 e±iz . as given by (7.158) one can derive then from (7. at large |z|.156) ( + r)! r!( − r)! ( + r)! r!( − r)! i 2z r [ /2] = r=0 ( + 2r)! (2r)!( − 2r)! −1 4z 2 r (7. (7.140. z (7.162) The leading terms in these expansions.

1 (7.131. (1.170) Recursion Formulas of Spherical Bessel Functions The spherical Bessel functions obey the recursion relationships g g +1 (z) +1 (z) = = (1.2) relationships hold for g (z) = h (z).165) (7.148) and express g (z) = h (1.171.171) −1 (z) (7.175) Using Γ( + 1) = !. 7. 7. 7.151) follows then that the relatiosnhips hold also for g (z) = j (z) and g (z) = n (z).2) (z) = −2 π 1 √ 2z πΓ(ν + 1 ) 2 z 2 ν C2. 7.169) (7.114) we can write g (z) = a z f 2 ! + 1 (z) 2 . One obtains for = 0. 7.172) to +1 (z) +1 (z) = A (z) 1− g (z) g (z) −1 (7. j (z) and n (z).4 dx (1 − x2 )ν− 2 eizx .171–7.2) To demonstrate that (7.162) allow one to provide explicit expressions for j (z) and n (z).167) (7.166) (7.2: Free Particle Expressions for the Spherical bessel Functions j (z) and n (z) 201 The identities (7. 2 j0 (z) n0 (z) j1 (z) n1 (z) j2 (z) n2 (z) = = = = = = sin z z cos z − z sin z cos z − z2 z cos z sin z − 2 − z z 3 1 3 − sin z − 2 cos z 3 z z z 3 1 3 − 3 + cos z − 2 sin z z z z (7.171) holds for g (z) = h (z) we employ (7.2) g (z) − g (z) z 2 +1 g (z) − g z (7.176) .172) where g (z) is either of the functions h obtain the recursion relationship g g (z). From the linearity of the relationships (7.149. (7.124.7. 1.168) (7.161. One can combine (7. For this purpose we need to demonstrate only that the (1.173) A (z) = z ( +1) z2 +2 z (7.174) and from (7. deﬁning a = −1 and employing fν (z) as deﬁned in (7.174) We want to prove these relationships.

173.174) we start from (7.166).172). in fact. Exercise 7.178) (7. 7.6: Employ the recursion relationship (7. together with (7. 7. . (7.174) to determine (a) j1 (z).171).179) from which follows (7.173) is equivalent to (7. To prove (7.182) Replacing all ﬁrst derivatives employing (7.121).2.182).202 The derivative of this expression is g (z) = Employing (7. is equivalent to (7.179) leads to (7.5: Provide a detailed derivation of (7. n0 (z) using (7.58) it holds o 2 g (z) = − g (z) + z ( + 1) g (z) − g (z) .180) Since g (z) is a solution of the radial Schr¨dinger equation (7.172). (7. j2 (z) from j0 (z). and (b) n1 (z).2.173. z2 (7. The ﬁrst component of (7.179) g (z) = − z2 g (z) + g (z) − g +1 (z) z .181) Using this identity to replace the second derivative in (7. The second component of (7. n2 (z) from n0 (z).177) (z) = − z f 2 +2 + 3 (z) 2 . i. f yields.180) yields g +1 (z) = g (z) − ( + 2) g (z) + z2 +2 g (z) .171).165).172) we diﬀerentiate (7.. j0 (z) using (7.171).176) g (z) = z g (z) − g +1 (z) +1 2 Spherically Symmetric Potentials z g (z) + a z f 2 ! +1 2 (z) . Exercise 7. In order to prove (7.173).e. z (7. (7.

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

We assume so-called Gaussian units. t) which is solved by E(r. The particle. As is well known. t) (8.5) where p is the momentum of the particle and ro (t) it’s position at time t. t) − ∂t A(r. t) in (8. The following substitutions. t] (8. t) = 4 π J(r. called the scalar potential. but leave the ﬁelds unaltered: A(r. In the non-relativistic limit holds p ≈ mr and (8. called the vector potential.2) the term q r˙o δ(r − ro (t)). (8. t) .5) above agrees with the equation of motion as given in (1. t). t)] and.4). t) + ∂t A(r. called gauge transformations. Jackson (John Wiley & Sons. t) and A(r. t) · B(r. t) through the scalar and vector potentials V (r.10) (8.4) Lorentz Force A classical particle with charge q moving in the electromagnetic ﬁeld experiences the so-called Lorentz force q[E(r.6) for some vector-valued function A(r. The reader is referred to the well-known textbook ”Classical Electrodynamics”. t) = 0 = 0. the relationship between ﬁelds and potentials is not unique. 2nd Edition.3) (8. (8. t) = − V (r. in turn.11) V (r. alter the potentials. accordingly. t) = × A(r.204 Maxwell equations1 · E(r. t). t) + v × B(r.1) (8.2) In addition. t) where V (r. t) + χ(r. t) −→ 1 A(r. t) × B(r.3) reads then × E(r. New York. obeys the equation of motion d p = q dt E[ro (t). . From this follows E(r. t) in (8. the two homogeneous Maxwell equations hold × E(r. contributes to the charge density ρ(r. t) + ∂t B(r. t) −→ V (r. Equation (8.25).1) the term qδ(r − ro (t)) and to the current density ˙ J(r. t) (8.7) Gauge Transformations We have expressed now the electric and magnetic ﬁelds E(r. t) = − V (r. (8. t) + ∂t A(r. t) and B(r.9) (8. t) − ∂t E(r. t) . 1975) for a discussion of these and other conventional units. Scalar and Vector Potential Setting B(r. t) is a scalar function. D. t) Interaction of Radiation with Matter = 4 π ρ(r. t) − ∂t χ(r. solves implicitly (8.8) = 0 (8. t) . by J. t] + v × B[ro (t).

(8. t) + ∂t A(r.16) and the vector potential are uncoupled.13) · A(r. t) |r − r | (8. t) .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.19) Unfortunately. and the former the Coulomb interactions in the unperturbed bound particle system. t) − ∂t V (r.15) d3 r ρ(r . t) the condition · A(r. such that the latter describes the electromagnetic radiation.6) and (8. In fact.16) This is the potential commonly employed in quantum mechanical calculations for the description of Coulomb interactions between charged particles. (8. t) in (8. (8.19) and deﬁne J (r. t) . the second inhomogeneous Maxwell equation. t) . t) − 2 × A(r.12) yields then the Poisson equation 2 V (r. t) = 0 the solution is given by the Coulomb integral V (r. equation (8. t) and the current due to moving net charges.1) · − V (r. t) (8. t) Using · ∂t A(r. be achieved. t) = − 4 π ρ(r.20) .9) into (8.12) The corresponding gauge is referred to as the Coulomb gauge. and a component due to the remaining currents. (8. t) = 0 . 4π (8. t) and V (r. t). In fact.12) leads us to 2 2 A(r.14) In case of the boundary condition V (r.19) couples the vector potential A(r. a name which is due to the form of the resulting scalar potential V (r. Using the expressions (8. t). t) . t = 4 π J(r. t) = − 4 π J(r. t) together with (8. t) . t) can be obtained employing (8.17) A(r. in fact. Such description can. this potential results from inserting (8.18) together with condition (8. One would prefer a description in which the Coulomb potential (8. t) + ∂t V (r. t) − ∂t A(r.2). t) = ∂t = 4 π ρ(r. t) − ∂t A(r.9) for the ﬁelds results in × The identity × × A(r. For this purpose we examine the oﬀending term ∂t V (r. (8. t) = Ω∞ for r ∈ ∂Ω∞ (8.8. t) = · A(r. t) = 1 ∂t V (r. the gauge freedom allows us to impose on the vector potential A(r. The vector potential A(r.

22) 1 ∂t 4π · J (r.30) .24) (8.21) and (8. t) = 0 (8. t). t) − ∂t A(r. Presently.2 Planar Electromagnetic Waves The current density Jt describes ring-type currents in the space under consideration.29) · Et (r. t) = − 4 π Jt (r. t) = 0 and from this follows · Jt (r. in a ring-shaped antenna which exhibits no net charge. t) (8. t) − ∂t A(r. hence.21) ∂t and the Poisson equation (8.19) yield 2 2 A(r. t) = 0).27) which is obviously curl-free ( × E (r. t) = 0 . for example. For J also holds the continuity equation · J(r. yet a current.16) yield ﬁnally the electric and magnetic ﬁelds. t) (8. t) + ∂t ρ(r.28) (8. using ∂t · J (r. hence.. t) holds.e.. the name longitudinal electric ﬁeld.26) This equation does not couple anymore scalar and vector potentials. 8. exist in the space considered. (8. t) = 0). we want to assume that no ring-type currents.26) turns into the well-known wave equation 2 2 A(r.206 For the curl of J holds × J (r. the name transverse ﬁelds. t) contributes solely an electric ﬁeld component E (r. t) = − ∂t ρ(r. t) = − V (r. = − ∂t A(r.14). t) be the total current of the system under investigation and let Jt = J − J . (8. (8.25) Because of properties (8. Let J(r. In this case (8. V (r. respectively.25) one refers to J and Jt as the longitudinal and the transverse currents. t) These ﬁelds are obviously divergence -free (e. t) = × A(r. (8. A(r. t) .23) This continuity equation identiﬁes J (r. t) + ∂t ρ(r.g. t) = 0 . The vector potential determined through (8. t) Bt (r. i. t) = 0 . For the divergence of J (r. t) contributes an electrical ﬁeld component as well as the total magnetic ﬁeld Et (r.26) and (8. t) .12) and the Coulomb potential (8. such current densities exist. no divergence-free currents. t) as the current due to the time-dependence of the charge distribution ρ(r. t) = or = 2 Interaction of Radiation with Matter (8. V (r. The deﬁnitions of J and Jt applied to (8.

ˆ We want to characterize now the radiation ﬁeld connected with the plane wave solutions (8. (8. t) = ± ω2 ˆ |Ao |2 k .31) u is a unit vector (|ˆ| = 1) which.34.35) one obtains S(r.31) and the resulting ﬁelds (8. t) = ±i ω A(r.28. The Coulomb gauge condition (8.38) ˆ In this expression for the energy ﬂux one can interprete k as the propagation velocity (note c = 1) and. t) . (8. Here the “-” sign corresponds to so-called incoming waves and the “+” sign to outgoing waves2 . Time average over one period 2π/ω yields S(r. Obviously. 8. ˆ (8. is orthogonal to k. t) × Re B(r.33) ωt) (8. see also the comment below Eqs. (8.158). t) = i k × A(r. accordingly. The corresponding electric and magnetic ﬁelds. 8. at each point r and moment t. ω2 = |Ao |2 (8. This ﬂux is given by 1 S(r.35) are complex-valued quantities. are orthogonal to each other and are both orthogonal to the wave vector k. 8. In applying the potential and ﬁelds to physical observables and processes we will only employ the real parts. t) = ± ω2 ˆ |Ao |2 k sin2 ( k · r − ωt ) 4π (8. A complete set of solutions is given by the so-called plane waves A(r.29).36) 4π Using the identity a × (b × c) = b (a · c) − c (a · b) and (8.8.34. according to (8.2: Planar Electromagnetic Waves 207 which describes electromagnetic ﬁelds in vacuum. t) = Re Et (r.31). t) .35) The vector potential in (8. 8. 2 . 8. (8. t) Bt (r.32) holds. 8.34) (8. hence.31.34.35). obviously. Note that in the units chosen the velocity of light is c = 1.39) 8π The deﬁnition incoming waves and outgoing waves is rationalized below in the discussion following Eq.12) yields u·k = 0. are Et (r.37) ˆ ˆ where k is the unit vector k = k/|k|. The latter vector describes the direction of propagation of the energy ﬂux connected with the plane wave electromagnetic radiation.38. 8π (8. there exist two linearly ˆ u independent orientations for u corresponding to two independent planes of polarization. t) = Ao u exp i(k · r ˆ where the dispersion relationship |k| = ω (8.32. the constant k is referred to as the wave vector. t) in (8. Et (r. t) and Bt (r. 8.39).

we will neglect the respective terms in the Hamiltonian (8.8. i (8.31. |k| = ω .208 Interaction of Radiation with Matter as the energy density.43). respectively.42) 8. deﬁned below Eqs. A correct description of the electromagnetic ﬁeld requires that the ﬁeld be quantized. t). (8.28. Note that E (r. (8. We assume that the energy content of the ﬁelds is not altered signiﬁcantly in the processes described and. If we consider a volume V with a number of photons Nω the energy density is obviously Nω ω . 8. We are left with a classical Hamiltonian function which has an obvious quantum mechanical analogue ˆ H = ˆ p − q A(r.31) allows one ﬁnally to state for the planar wave vector potential A(r.3 Hamilton Operator The classical Hamiltonian for a particle of charge q in a scalar and vector potential V (r) and A(r. These photons each carry the energy ω. whereas in the case of outgoing waves the energy is transported in the direction of k.31). u · k = 0 . t) . A ‘poor man’s’ quantization of the ﬁeld is possible at this point by expressing the energy density (8. The integrals express the integration over the energy density of the ﬁelds. t) = H Ψ(r.27.40) allows one to express then the ﬁeld ˆ amplitudes 8πNω Ao = . (8.40) = V It should be pointed out that Nω represents the number of photons for a speciﬁc frequency ω. t) are complex leading to the diﬀerence of a factor 2 in the energy densities of the lontitudinal and transverse components of the ﬁelds. is p − q A(r. Bt (r.38) implies that for incoming waves.32). t) 2m 2 + qV (r) .39) through the density of photons connected with the planar waves (8. the energy of the plane wave is transported in the direction of −k. t).29) together with the potentials (8.31). ˆ (8. Comparision of (8. 8.39) and (8. t) = 8πNω u exp i(k · r − ωt) ˆ ωV . 8. The wave function Ψ(r. (8. (8.41) ωV Inserting this into (8.43) Here the ﬁelds are deﬁned through Eqs.44) . a ˆ speciﬁc k and a speciﬁc u. t) 2m + 1 8π Ω∞ 2 H = + qV (r) 1 16π d3 r |Et |2 + |Bt |2 Ω∞ d3 r E 2 + . (8. t) is real 1 and that Et (r. 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.16.45) . The sign in (8. hence.

(8.. 8. in the Schr¨dinger equation o 2 ˆ p − qA i ∂t Ψ(r. An important conceptual step of modern physics has been to turn the derivation given around and to state that introduction of a local phase factor eiqχ(r.t)/ Ψ(r. hence. t) .47. an element of the group U(1).51) 2m the potentials A(r.11) of the potentials is equivalent to multiplying the wave function Ψ(r. Applying the gauge transformations (8. The idea just stated can be generalized by noting that multiplication by a phase factor eiqχ(r.48) 2m For this purpose one notes i ∂t eiqχ(r.t)/ constitutes a unitary transformation of a scalar quantity. Elementary constituents of matter which are governed by other symmetry groups.10. by the group 3 The eﬀect on other expectation values is not discussed here.45) leads to the Schr¨dinger equation o 2 ˆ p − q A − q(( χ)) + qV − q((∂t χ)) Ψ(r.10.g. hence.t)/ Ψ(r. t) = + qV Ψ(r. t) r (8. t) inside the double brackets. t) ˆ = eiqχ(r. accordingly.t)/ p + q(( χ)) Ψ(r. t) . t) . that this principle applies only to fundamental interactions.44.11) aﬀect the quantum mechanical description.47) is equivalent to 2 ˆ p − qA i ∂t eiqχ(r.t)/ should not aﬀect a system and that.t)/ . have no eﬀect on the motion of the particle described by (8. however.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.46). 8.48) implies that the gauge transformation (8.t)/ [ i ∂t − q((∂t χ)) ] Ψ(r. t) ˆ p eiqχ(r.11) to (8. t) = + qV eiqχ(r. t) (8. not to phenomenological interactions like the molecular van der Waals interaction.8. In the classical case such question is mute since the gauge transformations do not alter the ﬁelds and. t) = 2m where ((· · ·)) denotes derivatives in (( χ)) and ((∂t χ)) which are conﬁned to the function χ(r. t) enter whereas in the classical equations of motion ˙ m¨ = q E(r. One can show that (8. t) are necessary to compensate terms which arise through the phase factor. 8. . (8. t) by a local and time-dependent phase factor eiqχ(r.46) the ﬁelds enter. This leads to the question in how far the gauge transformations (8. 8. t) = eiqχ(r. t) and V (r. t)|2 and. t) + q r × B(r. i.10.t)/ Ψ(r.47) i ∂t Ψ(r.49) (8. does not change expectation values which contain the probability densities3 . such phase factor does not change the probability density |Ψ(r. t) and A(r.e. It should be noted. 8. (8. Obviously.50) The equivalence of (8.. e.t)/ Ψ(r.

54) ˆ2 p q ˆ q2 2 − p·A + A + qV . x1 .55). In case of a homogeneous potential pointing in the x3 -direction. 3. likewise can demand the existence of ﬁelds which compensate local transformations described by eiσ·χ(r.53) ˆ ˆ p·A + A·p + q2 2 A + qV 2m (8. j = 1.59) . k3 .58). Accordingly.12)].57) satisﬁes · A = 0 and.52) This expression vanishes in the present case since since ˆ ˆ p · Af = A · pf and. Using Cartesian coordinates this yields 2 H = − 2me 2 2 2 ∂1 + ∂2 + ∂3 + 2 eBo e2 Bo 2 x1 ∂2 + x i me 2me 1 (8. The resulting ﬁelds are called Yang-Mills ﬁelds. t) ≡ 0.. The Hamiltonian (8. the so-called Landau gauge ˆ associates the vector potential AL (r) = Bo x1 e2 ˆ (8. 2.56). 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. t) = Bo . (8. · A = 0 [cf. The vector potential (8. x2 . We want to describe the stationary states corresponding to the Hamiltonian (8. one can employ the Hamiltonian (8.55) 8. for Bo = Bo e3 in (8. The choice of a vector potential aﬀects the form of the wave functions describing the eigenstates and.g.55). (8. x3 ) = exp(ik2 x2 + ik3 x3 ) φE (x1 ) . In this o case holds V (r. 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.57) with a homogeneous magnetic ﬁeld Bo . holds (8.56) due to the gauge freedom. thereby. H = 2m m 2m (8. k2 . e. the generators of SU(2). can be described by various vector potentials.45) with Hamiltonian (8. therefore.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.210 Interaction of Radiation with Matter SU(2). For this purpose we use the wave function in the form Ψ(E. t). (8. The stationary homogeneous magnetic ﬁeld B(r. (8. aﬀects the complexity of the mathematical derivation of the wave functions.t) where σ is the vector of Pauli matrices.58) where ∂j = (∂/∂xj ).

the wave function (8. We want to employ . 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. Unfortunately. 2 2me (8.4: Electron in Homogenous Magnetic Field This results in a stationary Schr¨dinger equation o 2 211 − 2me 2 ∂1 + 2 k2 2 2me + 2 k2 3 2me + 2 eBo k2 e2 Bo 2 x1 + x φE (x1 ) me 2me 1 = E φE (x1 ) . according to (8. according to (8.63) (8. k3 ) = ω(n + 2 k2 1 3 ) + .61) leads to − where 2 2me 2 ∂1 + 2 k2 1 3 me ω 2 ( x1 + x1o )2 + 2 2me φE (x1 ) = E φE (x1 ) . x1 .57). k2 . is Ψ(n. (8. From this observation one can immediately conclude that the wave shifted (by e 3 function of the system.59).62) x1o = and where ω = k2 eBo eBo me (8.66) Obviously.63) this induces a spread of the wave function in the x1 direction.56) through the vector potential A(r) = 1 Bo × r . 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. However.59). (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 . the so-called symmetric gauge which expresses the homogeneous potential (8.65).64) is the classical Larmor frequency (c = 1). The stationary Schr¨dinger equation (8. The energies corresponding to these states are E(n.and x2 -coordinates. therefore. 2 (8. is not symmetric in the x1 . k3 .8. like the corresponding gauge (8.65) which localize the electrons. Solution for Symmetric Gauge The solution obtained above has the advantage that the derivation is comparatively simple.65) where we replaced the parameter E by the integer n. the familiar harmonic oscillator quantum number.62) is that of a displaced (by x1o ) harmonic oscillator with o 2 k 2 /2m ) energies.67) . x2 . Without aﬀecting the energy one can form wave packets in terms of the solutions (8. the states are degenerate in the quantum number k2 describing displacement along the x2 coordinate.

for any (8.71). The latter can be rewritten. In this case the current density measures −e v oriented tangentially to the ring. the Hamiltonian (8.74) is in the present case 1 µclass = − e r v e3 . The theory of classical electromagnetism predicts an analogue energy contribution .69) × u and × r = 0.73) where µclass is the magnetic moment connected with a current density j µclass = 1 2 r × j(r) dr (8. ˆ (8.67) one can write ˆ p·A = Using · (u × v) = − u · function f (r) ×v + v· 2i · Bo × r .78) . Accordingly. in the present case of constant Bo . the magnetic moment (8.74) We consider a simple case to relate (8. namely. y-plane with constant velocity v on a ring of radius r. namely. For the vector potential (8.74). (8. ˆ Identifying r × p with the angular momentum operator L.71) (8. (8.77) Comparision with (8.72) allows one to interpret µ = − e L 2me (8. Vmag = −µclass · Bo (8.75) 2 The latter can be related to the angular momentum µclass = − and. (8. accordingly. (8.73.76) e 2me class class · Bo .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 .68) × u yields.70) ˆ ˆ p · A f = − Bo · p × r f . Vmag = e 2me class = r me v e3 of the electron ˆ (8.72) to Hamiltonian (8.12) for the Coulomb gauge. an electron moving in the x.72) and (8. using × (uf ) = −u × f +f ˆ ˆ Bo · p × r f = Bo · r × p f .212 Interaction of Radiation with Matter One can readily verify that this vector potential satisﬁes the condition (8. 8.

we separate the variable x1 .83) describes two identical oscillators along the x1 -and x2 -directions which ˆ are coupled through the angular momentum operator L3 . x2 from x3 setting ΨE (x1 . 10 that the spin of the electron. 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. i. described by the operator S. Accordingly. ˆ i. x2 .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 . (8.84) . x2 .4: Electron in Homogenous Magnetic Field 213 as the quantum mechanical magnetic moment operator for the electron (charge −e)..80) (8.67) yields a more symmetric solution which decays to zero along both the ±x1 . the so-called gyromagnetic ratio of the electron. e gives rise to an energy contribution (8. x3 ) .87) . We will demonstrate in Sect. the Hamiltonian (8. For a magnetic ﬁeld (8.86) 1 2 2me 1 2 ˆ as well as of the angular momentum operator L3 .e.8.and the ±x2 -direction.71) is ˆ H = 2 ˆ2 p e2 Bo + 2me 8me x2 + x2 + 1 2 eBo L3 .56) pointing in the x3 -direction the symmetric gauge (8.88) me ω xj .72) with an associated magnetic moment − g 2me S where g ≈ 2. In this case. E = E − (8. x3 ) = exp(ik3 x3 ) ψ(x1 .85) The Hamiltonian (8. To obtain these eigenstates we introduce the customary dimensionless variables of the harmonic oscillator Xj = (8. x2 ) = E ψ(x1 . likewise. x2 .79) To obtain the stationary states. x3 ) = E ΨE (x1 . The functions ψ(x1 . the solutions of ˆ H ΨE (x1 . requires a Lorentz-invariant quantum mechanical description as provided in Sect. x2 ) obey then ˆ Ho ψ(x1 . 2me We have used here the expression for the angular momentum operator ˆ L3 = ( /i)(x1 ∂2 − x2 ∂1 ) .82) 1 ω (x1 ∂2 − x2 ∂1 ) i 2me + 1 me ω 2 x2 + x2 1 2 2 2 k2 3 + (8. (8. for Bo = Bo e3 . A derivation of his property and the value of g.83) (8. x2 ) where ˆ Ho = − 2 2 2 ∂1 + ∂2 (8. j = 1. 2 . x2 ) .e. 10. 2me (8.81) (8.

0. . In order to cover all posible vibrational quantum numbers one needs to choose j. x1 . (8.97) . x2 ) = 2m Ψ (j.94) Since this state is a linear combination of states which all have vibrational energy (2j + 1) ω. 2 (8. such that 1 i a† a2 − a† a1 1 2 Ψ (j.92) where Ψ(0. (8. Such eigenstates can be expressed. aj = √ 2 Interaction of Radiation with Matter Xj + ∂ ∂Xj . x1 .93) ˆ The states (8. . in fact. through a combination of states j Ψ (j. . the total vibrational energy being ω(2j + 1).91) We note that the operator a† a2 − a† a1 leaves the total number of vibrational quanta invariant. m . however. +j . obeyed. We. x1 . m . x1 . m. x2 ) is the wave function for the state with zero vibrational quanta for the x1 . x1 .e. 1 3 . (8. 1. 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 . m . (8. x2 ) = m=−j αmm Ψ(j. attempt to express eigenstates in terms of vibrational wave functions a† 1 j+m Ψ(j. . m as follows: j = 0. therefore. x2 ) = ω 2j + 2m + 1 Ψ (j. m .. it holds a† a1 + a† a2 + 1 1 1 2 Ψ (j. m .. x1 .89) . x2 ) (8.92) represents a state with j + m quanta in the x1 -oscillator and j − m quanta in the x2 -oscillator. (8. . x1 .94) is an eigenstate of the vibrational Hamiltonian. x1 . . 2 2 . x2 ) (8. x2 ) . x2 ) . 1 2 since one phonon is annihilated and one created. If this property is.96) ˆ holds. m .94) is an eigenstate of Ho ˆ Ho Ψ (j. x1 . (8. i.as well as for the x2 -oscillator. j = 1.90) a† a2 − a† a1 1 2 . m = −j. x2 ) = ( 2j + 1 ) Ψ (j.214 Employing the creation and annihilation operators 1 a† = √ j 2 and the identity ˆ ω L3 = which can readily be proven. (j) (8. m . x2 ) . x1 . i. −j + 1. .92) are not eigenstates of L3 . 0. x2 ) = a† 1 j−m (j + m)! (j − m)! Ψ(0. (8.e. (8.95) (j) ˆ We want to choose the coeﬃcients αmm such that (8. m. x1 . .94) is also an eigenstate of L3 .

Using the explicit form of the Wigner rotation matrix as given in (5. 5.84) above.309) yields ﬁnally Ψ (j. x1 . an example is the Coulomb potential V (r. x2 ) = Dmm ( .102) We have identiﬁed. the second rotation a factor dmm ( π ). The external radiation ﬁeld is accounted for by the vector potential A(r..9.11 such transformation is provided through π π (j) Ψ (j. 5. According to the derivation given there.92) correspond to the eigenstates |Ψ(j.9. i. x1 . m .10. the 2 2 latter representing the Wigner rotation matrix of Sect.e. thus.5.5. 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. except for a factor 2 . the states are eigenstates of the operator [we use for the operator the notation of Sect. We assume that the scalar potential V in (8. Other radiation ﬁelds can . The ﬁrst rotation contributes a factor exp(−im π ).5.100) J2 = 1 2 2i 1 ˆ i. According to Sect.8.55). . a† .83) and conﬁrmed the eigenvalues stated in (8. the eigenstates of (8. b− + − notation in Sects. Eq. x1 . (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. m. x1 . 0) Ψ(j.99) with eigenvalue m.10. 8.5. 5.10.5: 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.10. a2 1 2 present notation ←→ b† .. m . x2 ). 5.101) 2 2 where Dmm ( π .11. The required rotation must transform the x3 –axis into the x2 –axis. x2 ) (8.11 (8.11. π .97). m. x1 .55) conﬁnes the particle to stationary bound states. x2 ) . 5. t) introduced above.31). is identical to the operator L3 introduced in (8. If we identify 2 a† . m. a transformation moving the x3 –axis into the x2 – 2 axis. m) in Sect. b+ . In the simplest case the radiation ﬁeld consists of a single planar electromagnetic wave described through the potential (8. b† . a1 .288)] 1 ˆ J3 = 2 a† a1 − a† a2 1 2 (8.9. 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 .98) then the states Ψ(j. (8. cf. The connection with the present problem arises due to the fact that the operator J2 in Sect. which corresponds there to the angular momentum in the x2 –direction.(5. This implies that ˆ we can obtain eigenstates of L3 by rotation of the states Ψ(j. is in the notation of the present section 1 ˆ a† a2 − a† a1 .e. m. x2 ) deﬁned in (8. t) = 1/4πr conﬁning an electron with energy E < 0 to move in the well known orbitals of the hydrogen atom. x1 . 5.

. i. orthonormal basis. . we discount the possibility of a continuum of eigenstates. i. t) − m 2m (8. is much 2 (r. 2 .41) e ˆ e p2 p·A ∼ 2me me me 2me 4 1 2 8πNω .. 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.105) in further calculations and to expand the wave function in terms of powers of VS (t) in a perturbation calculation. This result will allow us to neglect the larger than the second term. 1. this assumption can be waved as our results below will not depend on it. is much smaller than the eigenvalue diﬀerences near ˆ typical atomic bound states. (8. Here the so-called unperturbed system is governed by the Hamiltonian Ho with stationary states deﬁned through the eigenvalue problem Ho |n = n |n . in this case.105). the term ∼ A second term in (8.216 Interaction of Radiation with Matter be expanded through Fourier analysis in terms of such plane waves.e. one can estimate that the perturbation.e.108) We assume for the sake of simplicity that the eigenstates of Ho can be labeled through integers.105) for the case of radiationinduced transitions in atomic systems.42).103) (8. t)...106) where we adopted the Dirac notation for the states of the quantum system. n = 0. the term ∼ p · A(r. i. . t). as given in (8. using (8. t) + A (r. we assume n|m = δnm and for the identity ∞ (8. The states |n are thought to form a complete.. (8.e.104) (8. Estimate of the Magnitude of VS We want to demonstrate now that the interaction VS (t). We ﬁrst note. and that the ﬁrst term in (8. can be considered a weak perturbation.107) 1 = 1 n=0 |n n| . For an electron charge q = −e and an electron mass m = me one can provide the estimate for the ﬁrst term of (8.105) as follows4 . i.109) The reader should note that the estimates are very crude since we are establishing an order of magnitude estimate only.105) where A(r. ωV (8. However. In fact. t) is given by (8.e. 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.

2me (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 . t) .5 ˚. with ao ≈ 0.e.112) π λ me c2 and with e2 /ao ≈ 27 eV.111) For ω = 3 eV and a corresponding λ = 4000 ˚ one obtains. u) p · u exp i(k · r − ωt) ˆ ˆ ˆ kV .114) Employing for the second factor the estimate as stated in (8. Assuming a single photon. and me c2 ≈ A A 500 keV 2 ao ω ≈ 10−8 (8. (8. (8.115) (8.116) This term is obviously much smaller than the ﬁrst term. and as long as the photon densities Nω /V are small.8. λ = 2πc/λ. Using again (8. altogether. one can neglect this term as long as the ﬁrst term gives non-vanishing contributions. i. a volume V = λ3 where λ is the wave length corresponding to a plane wave with frequency ω.e.105) due to a radiation ﬁeld by VS = − q ˆ p · A(r. m (8.42) it holds VS ≈ e m k. one obtains for (8.110) where ao is the Bohr radius. me (8. e ˆ p · A ∼ 10 eV · 10−4 = 10−3 eV .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..105).117) In case that such perturbation acts on an electron and is due to superpositions of planar waves described through the vector potential (8. replace the perturbation (8. We can.5: Time-Dependent Perturbation Theory 217 The virial theorem for the Coulomb problem provides the estimate for the case of a hydrogen atom p2 2me ∼ 1 e2 2 ao (8.112) yields e2 A2 ∼ 10 eV · 10−8 = 10−7 eV . i.118) . Nω = 1..105) for radiation ﬁelds can be considered a small perturbation. Hence. Consequently. We want to estimate now the second term in (8. hence. the ﬁrst term in (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.ˆ u 4πNk α(k.

125) (8.106–8. The states |0 . 2.122) where |ΨD (to ) = |0 . Equation (8.42) through k = |k| = ω. Using i i ∂t exp − Ho (t − to ) and (8. This state obeys the Schr¨dinger equation o i ∂t |ΨS (t) = [ Ho + VS (t) ] |ΨS (t) subject to the initial condition |ΨS (to ) = |0 .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) (8.123) i = Ho exp − Ho (t − to ) (8.218 Interaction of Radiation with Matter where we have replaced ω in (8.118) is the form of the perturbation which. |n are deﬁned in (8.126) Ho (t − to ) yields ﬁnally . 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 . A factor α(k.108) as the eigenstates of the unperturbed Hamiltonian Ho .121) (8.. u) has been added to describe eliptically or circularly polarized ˆ waves. For this purpose one needs to determine the state |ΨS (t) of the particle. under ordinary circumstances. (8. .124) (8.119) one obtains i exp − Ho (t − to ) ( Ho + i ∂t ) |ΨD (t) i = [ Ho + VS (t) ] exp − Ho (t − to ) |ΨD (t) from which follows i i exp − Ho (t − to ) i ∂t |ΨD (t) = VS (t) exp − Ho (t − to ) |ΨD (t) . n = 0 at some later time t ≥ to . The sum runs over all possible k vectors and might actually be an integral. . One seeks to predict the probability to observe the particle in one of the states |n . the sum over u involves the two possible polarizations of planar ˆ electromagnetic waves. Multiplying the latter equation by the operator exp i (8.120) (8. describes the eﬀect of a radiation ﬁeld on an electron system and which will be assumed below to describe radiative transitions. |Ψ(to ) = |0 (8. . The probability to ﬁnd the particle in the state |n at time t is then p0→n (t) = | n|ΨS (t) |2 .127) i ∂t |ΨD (t) = VD (t) |ΨD (t) . n = 1.

121) expressed in terms of ΨD (t) is i p0→n (t) = | n|exp − Ho (t − to ) |ΨD (t) |2 .5: Time-Dependent Perturbation Theory where VD (t) = exp 219 i i Ho (t − to ) VS (t) exp − Ho (t − to ) . (8.136) together with the initial conditions |ψD (to ) = |0 0 for n = 0 for n = 1.134) (8. we deﬁne |ΨD (t) through the evolution equations i ∂t |ΨD (t) i ∂t |ΨD (t) i (2) ∂t |ΨD (t) (1) (0) = = = .137) can be solved recursively. . . Due to the Hermitean property of the Hamiltonian Ho holds n|Ho = i n|exp − Ho (t − to ) from which we conclude.134.132–8.131) p0→n (t) = | n |ΨD (t) |2 .137) follows |ΨD (t) = |0 . .138) 1 i t dt VD (t ) |0 . 8.8.137) are consistent with (8.128).127. In order to determine |ΨD (t) described through (8. . We will consider here only the two leading contributions to |ΨD (t) . .127) we assume the expansion ∞ |ΨD (t) = n=0 (n) |ΨD (t) (n) (8. 0 VD (t) |ΨD (t) (1) VD (t) |ΨD (t) (0) (8.133. Equations (8. . 8. Employing this result one obtains for (8.137) One can readily verify that (8. (8. 8. 2. (8. Accord(n) ingly.133) (8.130) = exp − i n (t − to ) − to )]| = 1. to (8. = .132) where |ΨD (t) accounts for the contribution due to n-fold products of VD (t) to |ΨD (t) .137) |ΨD (t) = (1) (0) (8. .133–8.139) . n| (8. From (8.129) n| and. using |exp[− i n (t n (8. consequently.135) i ∂t |ΨD (t) (n) VD (t) |ΨD (n−1) (t) (8.128) We note that the transition probability (8.

135. λ → 0+ .e. We want to apply now the perturbation expansion derived to such perturbations.137) |ΨD (t) = (2) Interaction of Radiation with Matter 1 i 2 t t dt to to dt VD (t ) VD (t ) |0 . to → −∞ (8. note that uj . + m=0 dt to to dt n|VD (t )|m m|VD (t ) |0 8. 8.144) r is the position of the electron and p = ( /i) is the momentum operator of the electron. non-singular. deﬁnes the direction of the E-ﬁeld of the ˆ radiation.34. (8. II III (8. in turn.144) Here the factor I describes the strength of the radiation ﬁeld (for the speciﬁed planar wave) as determined through the photon density Nj /V and the factor II describes the polarization of the planar wave. The factor III in (8.141) + . according to (8.. 8. λ → 0+ has been introduced which describes that at time to → −∞ the perturbation is turned on gradually. The resulting perturbation on an electron system. This factor will serve mainly the purpose of keeping in the following derivation all mathematical quantities properly behaved. however. such that the combined radiation ﬁeld is decribed by the vector potential A(r.118).118) above that the eﬀect of a radiation ﬁeld on an electronic system is accounted for by perturbations with a so-called harmonic time dependence ∼ exp(−iωt)..140) Altogether we have provided the formal expansion for the transition amplitude n|ΨD (t) = n|0 + ∞ 1 i 1 i t dt n|VD (t ) |0 to 2 t t (8.41). i.220 This result.. 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.6 Perturbations due to Electromagnetic Radiation We had identiﬁed in Eq.144) describes the propagation of the planar wave. The coeﬃcients A1 . is VS = ˆ ˆ V1 exp(−iω1 t) + V2 exp( iω2 t) eλt . yields for (8.143) ˆ ˆ where V1 and V2 are time-independent operators deﬁned as e ˆ Vj = m 8πNj ωj V I ˆ ˆ p · uj eik·r . In (8. t) = + A1 u1 exp i (k1 · r − ω1 t) ˆ A2 u2 exp i (k2 · r ˆ ω2 t) incoming wave incoming or outgoing wave (8. . For the sake of including the eﬀect of superpositions of plane waves we will assume. A2 are deﬁned through (8.142). (8. according to (8.142) combining an incoming and an incoming or outgoing wave. that two planar waves simulataneously interact with an electronic system. A factor exp(λt). the direction of ˆ the propagation being determined by k = k/|k|.

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 .140.128). For the 1st order contribution to the transition amplitude n|ΨD (t) = (1) 1 i t dt n|VD (t ) |0 to (8.146) = 1 λ→0+ t→−∞ i lim lim × t dt exp to i ( n − o − i λ) t iω2 t × . (8.143) to the expansion (8.150) and.6: Perturbations due to Electromagnetic Radiation 1st Order Processes 221 We employ now the perturbation (8. (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 (8.148) We consider now the 2nd order contribution to the transition amplitude.130) and (for m = 0) i i exp − Ho (t − to ) |m = exp − for (8. 8. According to (8.143).145) n|ΨD (t) (1) m (t − to ) |m . (8.8.128) one obtains k|VD (t) | = k|VS (t)| exp i ( k − ) (8. (8.151) ˆ ˆ n|V1 |m m|V1 |0 exp ˆ ˆ + n|V2 |m m|V2 |0 exp i i n − − m − ω1 − i λ)t exp ω2 − i λ)t exp m o n m m o .145) we obtain then. using (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 . (8.141).149) Using the deﬁnition of VD stated in (8. employing the perturbation (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.e. i.152). We assume ﬁrst that the ﬁrst order transi(1) tion amplitude n|ΨD (t) is non-zero. 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. in which case one can expect that it is larger than the 2nd (2) order contribution n|ΨD (t) which we will neglect.222 ˆ ˆ + n|V1 |m m|V2 |0 exp ˆ ˆ + n|V2 |m m|V1 |0 exp i i ( ( − − Interaction of Radiation with Matter − ω1 − i λ)t exp ω2 − i λ)t exp i i ( ( − − ω2 − i λ)t − ω1 − i λ)t n m m o n m m o Carrying out the integrations and the limit limt→−∞ provides the result (2) n|ΨD (t) = − 1 ∞ 2 λ→0+ lim (8.148) and ∗ |z1 + z2 |2 = |z1 |2 + |z2 |2 + 2Re(z1 z2 ) (1) (8. respectively.131).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. Using (8. (8. the time derivative of p0→n (t). provide now the transition probability p0→n (t) according to Eq. For this rate holds .153) (8.148) and (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 (8.8. of this expression describe the individual eﬀects of the two planar wave contributions of the perturbation (8. The result of such average is.158). i.160) leads to ﬁnal states |n with energy n = o ± ω2 . is much shorter than could be resolved in any observation. a property. In case the matrix elements are non-zero.158) play an essential role for the transition rates of radiative transitions.158) ˆ + | n|V2 |0 |2 δ( − ω2 ) Obviously.6: Perturbations due to Electromagnetic Radiation d p0→n (t) dt lim e2λt + × 2 Re ˆ 2 λ n|V1 |0 |2 ( o + ω1 − n )2 + (λ )2 ˆ 2 λ n|V2 |0 |2 d + 2λ + × ( o ± ω1 − n )2 + (λ )2 dt ˆ ˆ n|V1 |0 0|V2 |n exp i (± ω2 − ω1 ) t + ω1 − n + iλ ) ( o ± ω2 − n − iλ ) 223 = λ→0+ (8. The second contribution to (8. due to the uncertainty relationship would introduce a considerable perturbation to the system.144). to cancel the third term in (8. For the resulting expression the limit limλ→0+ can be taken.156) ( o The period of electromagnetic radiation absorbed by electronic systems in atoms is of the order 10−17 s. Hence.143. due to the vector potential A1 u1 exp i (k1 · r − ω1 t) . The time average will be denoted by · · · t . 8.142–8. 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. The δ-functions appearing in this expression reﬂect energy conservation: the incoming plane wave contribution of (8. in fact. . one should average the rate over many periods of the radiation.142) ˆ ˆ The matrix elements n|V1 |0 and n|V2 |0 in (8. Using lim x2 + 2 δ→0+ = π δ(x) (8. any attempt to do so. The result supports our deﬁnition of incoming and outgoing waves in (8.159) leads to ﬁnal states |n with energy n = o + ω1 . however.31) and (8.. the two terms apearing on the rhs.156) such that the 1st order contributions of the two planar waves of the perturbation simply add. (8. describing either an incoming or an outgoing plane wave due to the vector potential A2 u2 exp i (k1 · r ˆ ω2 t) .e. which is observed through the so-called spectral intensities of transitions |0 → |n . the matrix elements can vary strongly for diﬀerent states |n of the system.144). First. ˆ (8.

153) vanishes such that the leading contribution to the transition probability p0→n (t) arises from the 2nd order amplitude (8.157) yields.g. i. We deﬁne ∞ z1 = m=0 ˆ ˆ n|V1 |m m|V1 |0 m − o − ω1 − i λ exp i × (8. in Eqs. as we did in the the case of 1st order transitions.168) absorption of 2 photons ω1 . |z1 + z2 + z3 |2 t = |z1 |2 + |z2 |2 + |z3 |2 (8.158)..e.158). (8.165) × ∞ z3 = m=0 2 ω2 − 2i λ)t 2 ω2 − 2i λ n− o ˆ ˆ ˆ ˆ n|V2 |m m|V1 |0 n|V2 |m m|V1 |0 + ω2 − i λ m − o − ω1 − i λ m− o ( n o i − × It holds exp − ω1 ω2 − 2i λ)t ω2 − 2i λ n − o − ω1 ( n o 3 − |z1 + z2 + z3 | = |z1 | + |z2 | + |z3 | + j.e..167) Taking now the limit limλ→0+ and using (8. exp[ i (±ω2 − ω1 )]. 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 ) (8.166) over many periods of the radiation yields exp[ i (±ω2 − ω1 )] t = 0 and. (2) (8.153–8. e.k=1 j=k 2 2 2 2 ∗ zj zk (8.163) × (8.224 2nd Order Radiative Transitions Interaction of Radiation with Matter We now consider situations where the ﬁrst order transition amplitude in (8. in analogy to (8.162) × and. ∞ − 2 ω1 − 2i λ)t n − o − 2 ω1 − 2i λ ( n o − z2 = m=0 ˆ ˆ n|V2 |m m|V2 |0 ω2 − i λ m− o exp i × (8.166) 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. similarly.164) (8. hence.. i.152). it holds p0→n (t) = | n|ΨD (t) |2 .161) To determine the transition rate we proceed again. Time average · · · t of expression (8.

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

144) VS = where e 2πN ˆ ˆ p · u e±ik·r . m denote the relevant quantum numbers) ψn=1.m=0 (r.158). .171) The perturbation on an atomic electron system is then according to (8. (8. φ) a3 ao o 2p (8. the wave functions are (n.177) √ M = 6 0 ∞ 1 2π ia4 o r2 dr −1 dcosθ 0 ∗ dφ r e−r/2ao Y1m (θ.143) will contribute to the absorption process.178) These wave functions make signiﬁcant contributions to this integral only for r-values in the range r < 10 ao .179) . However. plane polarized wave described through the complex vector potential A(r. in this range one can expand eik·r ≈ 1 + ik · r + . to → −∞ (8. is then kabs = Dipole Approximation 2π e2 2πN m2 ωV e ˆ u · n| p eik·r |0 ˆ 2 ˆ ˆ V1 exp(−iωt) + V2 exp(+iωt) eλt . φ) = 2 − 1 2 1 −r/ao e Y00 (θ.. (8.2 = m ωV Only the ﬁrst term of (8.174) We seek to evaluate the matrix element ˆ M = n| p eik·r |0 . λ → 0+ . 8. θ.170) We will employ only the real part of this potential.176) ψn=2.226 Absorption of a Plane Polarized Wave Interaction of Radiation with Matter We consider ﬁrst the case of absorption of a monochromatic. the vector potential actually assumed is A(r. φ) = 6 r −r/2ao e Y1m (θ. θ.m (r.143. and the integral is =1. φ) (8.173) V1. in case of a radiative transition from the 1s state of hydrogen to one of its three 2p states.172) δ( n − o − ω) (8. ˆ (8. t) = 8πN ωV u exp ˆ ı (k · r − ωt) .175) The matrix element involves a spatial integral over the electronic wave functions associated with states |n and |0 . t) = 2πN ωV u exp ˆ ı (k · r − ωt) + 2πN ωV u exp ˆ ı (−k · r + ωt) . according to (8. =0. For example. (8. . The absorption rate. the second term can be discounted in case of absorption. i.e. (8. . φ) × × eik·r e−r/ao Y00 (θ. φ) a3 o 1s (8.

175). We are now in a position to obtain an alternative expression for the matrix element (8.186) δ( n − o − ω) (8. the value of the wave length for the 1s → 2p transition 2π c λ = = 1216 ˚ A (8.k=1 i m 3 pk ek = ˆ j.182).180) ∆E2p−1s and ao = 0. (8. Ho ] |0 = i i n) n|r |0 . One refers to this approximation as the dipole approximation.179) and other terms are never larger than 20π ao /λ. pk ] pk ˆ k=1 (8.178) holds eik·r ≈ A 1 1 + O( 50 ) such that one can approximate eik·r ≈ 1 .182) (8.183) (8.175) can then be ˆ achieved and the diﬀerential operator p = i replaced by by the simpler multiplicative operator r. V (r) ] 2me =0 = Using r = 3 ˆ j=1 xj ej 1 2me 3 pk [ r.182) one obtains M ≈ Insertion into (8. Using (8.7: One-Photon Emission and Absorption 227 One can estimate that the absolute magnitude of the second term in (8.k=1 i ˆ p m (8.182) one ﬁnds [ r.174) yields kabs = 4π 2 e2 N ω V u · n| ˆ |0 ˆ r 2 m( o − m n| [r. is Ho = ˆ (p)2 e2 + V (r) . Ho ] = i m 3 pk ej δjk = ˆ j. in case of the hydrogen atom. Transition Dipole Moment A further simpliﬁcation of the matrix element (8. pk ] + ˆ ˆ k=1 1 2me 3 [ r.8. pj ] = i δkj one obtains ˆ [ r.187) .184) and the commutation property [xk .181) For the commutator in (8.181) and (8.529 ˚ one concludes that in the signiﬁcant integration range in (8. This simpliﬁcation results from the identity m ˆ p = [ r. Ho ] = ˆ2 p [ r.185) from which follows (8.104) and. ] + [ r. V (r) = − . Using |k| = 2π/λ. 2me r (8. Ho ] i where Ho is the Hamiltonian given by (8.

Integrating and summing accordingly over all contributions as given by (8. 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. one can express | u · n| r |0 |2 = |ρ|2 ( cos2 α + cos2 β ) = sin2 θ .31–8. and (2) the fact that strictly monochromatic radiation cannot be prepared such that any radiation source provides radiation with a frequency distribution. The δ-function appearing in this expression.193) . directions.189) where the factor 1/ arose from the integral over the δ-function. say u1 and u2 .228 Interaction of Radiation with Matter where we used the fact that due to the δ-function factor in (8.191) (8. a very hot oven. For the two angles α = ∠(ˆ1 . the direction of u1 is described by the ˆ ˆ ˆ ˆ angles ϑ1 = π/2. ρ) holds then u cosα = cosϑ1 cosϑ + sinϑ1 sinϑ cos(ϕ1 − ϕ) = sinϑcosϕ and cosβ = cosϑ2 cosϑ + sinϑ2 sinϑ cos(ϕ2 − ϕ) = sinϑsinϕ . Accordingly. ϕ2 = π/2. k ˆ As a result. |0 . The unit ˆ ˆ ˆ ˆ can be chosen to point along the x1 . three-dimensional vector.158) above that in ﬁrst order the contributions of all components of the radiation ﬁeld add. planar waves as deﬁned in (8. We can. u2 . ˆ In order to carry out the integral dk we note that u describes the possible polarizations of the ˆ ˆ and u. hence.. i. ˆ u ˆ (8.190) (8. and all polarizations of the radiation.35). 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. k frame is described by the angles ϑ. The direction of this vector in the u1 . will actually be replaced by a distribution function which reﬂects (1) the ﬁnite life time of the states |n . x2 . such that the thermal radiation present supplies a continuum of frequencies. ϕ. We have demonstrated in our derivation of the rate of one-photon processes (8.174) one can replace n − o by ω. in practical situations.e. according to (8. ρ) and ˆ u β = ∠(ˆ2 .188) ˆ Here dk is the integral over all orientations of k. obtain the transition rate in the present case by adding the individual transition rates of all planar waves present in the oven. x3 -axes of a right-handed cartesian vectors u1 .33) are orthogonal to each other. there are ony two linearly independent directions of u possible. Absorption of Thermal Radiation We want to assume now that the hydrogen atom is placed in an evironment which is suﬃciently hot.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. ϕ1 = 0 and of u2 by ϑ2 = π/2. u2 and k ˆ ˆ coordinate system.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.

a problem in case of emission by quantum systems in complete darkness. 8.172.171) and perturbation (8. This dependence predicts. In case of a classical radiation ﬁeld one would expect that emission cannot occur.37 × 1019 1 ao | n|r|0 |2 × s λ λ2 . The corresponding process is termed spontaneous emission. ﬁnally.8. For practical evaluations we provide an expression which eliminates the physical constants and allows one to determine numerical values readily. However.7: One-Photon Emission and Absorption 229 This geometrical average.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 |. can be inserted into (8. (8.37 × 1019 × 3 3 c3 3 ao λ3 s λ and kabs where λ = 2πc n − (8. the absorption rate is of the order of 109 s−1 or 1/nanosecond. 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. for Nω = 0.. linear molecules. This poses. for atoms or symmetric molecules. as for the absorption process. in agreement with observations. i. Emission of Radiation We now consider the rate of emission of a photon. . Nω = 0.196) The last two factors in (8. Transition Dipole Moment The expression (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 . the resulting total rate of emission bears a diﬀerent dependence on the number of photons present in the environment.e. the calculation of the emission rate proceeds as in the case of absorption. However. 2 .e.173) contributes. by planar waves with vector potential (8. Nω photons before absorption. The radiation ﬁeld is described.. there is also a contribution to the emission rate which is proportional to Nω . . The most intensely absorbing molecules are long.173). that emission occurs even if no photon is present in the environment. Accordingly. in particular. 1. This diﬀerence between emission and absorption is due to the quantum nature of the radiation ﬁeld. The value of | n| r |0 | determines the strength of an optical transition.189) to yield the total absorption rate kabs (tot) = Nω 4 e2 ω 3 | n|r|0 |2 3 c3 .. The quantum nature of radiation manifests itself in that the number of photons Nω msut be an integer. i. However. (8. Otherwise. ˆ In case of emission only the second term V2 exp(+iωt) in (8. For this purpose we use ω/c = 2π/λ and obtain 4 e2 ω 3 32π 3 e2 ao 1 ao = = 1.194) combined are typically somewhat smaller than (1 ˚/1000 ˚)3 = A A 10−9 .197) o (tot) (8.195) = Nω 1. The transition dipole moment can vanish for many transitions between stationary states of a quantum system. however. .

the well-known Planck radiation formula. exp[− ω/kB T ] = This equation yields Nω = 1 .198) to determine the stationary distribution of photons ω in an oven of temperature T . Hence.. using (8. 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.204) Nω . in lasers. The total rate of emission.202) It follows. To demonstrate this property we apply the transition rates (8. = (8. exp[ ω/kB T ] − 1 (8. In the following we discuss several examples. 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.198) is consistent with Planck’s radiation law which reﬂects the (boson) quantum nature of the radiation ﬁeld.8 Two-Photon Processes In many important processes induced by interactions between radiation and matter two or more photons participate. Let No and Nn denote the number of atoms in state |0 and |n . Deﬁning the rate of spontaneous emission 4 e2 ω 3 | n|r|0 |2 (8.199) Planck’s Radiation Law The postulate of the Nω + 1 dependence of the rate of emission as given in (8. The number of atoms undergoing absorption in unit time are Nω ksp No and undergoing emission are (Nω +1)ksp Nn ..201) ksp = 3 c3 the rates of absorption and emission are Nω ksp and (Nω + 1)ksp . respectively.e.195) and (8. 8. We assume n − o = ω.203) i.200) where kB is the Boltzmann constant. .200). accordingly. e. respectively.g. it must hold Nω ksp N0 = (Nω + 1) ksp Nn (8. For these numbers holds Nn / No = exp[−( n − o )/kB T ] (8.198) (8.230 Interaction of Radiation with Matter which is termed induced emission since it can be induced through radiation provided. Nω + 1 (8. Under stationary conditions the number of hydrogen atoms undergoing an absorption process |0 → |n must be the same as the number of atoms undergoing an emission process |n → |0 .

8.207) for the rate of 2-photon transitions shows that the transition |0 → |n becomes possible through intermediate states |m which become virtually excited through absorption of a single photon. We wish to describe an observation in which a detector is placed at a solid angle element dΩ2 = sinθ2 dθ2 dφ2 with respect to the origin of the coordinate system in which the electron is described. even if n| r |0 vanishes.106). The electron absorbs the radiation and emits immediately a second photon.207) Expression (8. ˆ (8. which is not necessarily the case. In applying (8. then expression (8. a transition between the states |0 and |n may be possible.173).181) and using (8. (8.187).182) yields.207) does not converge rapidly. but then requires the absorption of two photons.172. (8. t) = Ao1 u1 cos(k1 · r − ω1 t) . The resulting rate of the transition depends on 2 Nω . In this case one needs to choose the energy of the photons to obey n = o + 2 ω. We assume that a planar wave with wave vector k1 and polarization u1 . Employing the dipole approximation (8. in case ˆ of absorption only V1 contributes. however. dropping the index 1 characterizing the radiation. 8. The perturbation which accounts for the coupling of the electronic system and the radiation ﬁeld is the same as in case of 1st order absorption processes and given by (8. If the sum in (8. .205) The respective radiative transition is of 2nd order as described by the transition rate (8.207) one is. The intense radiation ﬁelds of lasers allow one to increase transition rates to levels which can readily be observed in the laboratory. Scattering of Photons at Electrons – Kramers-Heisenberg Cross Section We consider in the following the scattering of a photon at an electron governed by the Hamiltonian Ho as given in (8. faced with the dilemma of having to sum over all intermediate states |m of the system.104) with stationary states |n deﬁned through (8.207) does not provide a suitable avenue of computing the rates of 2-photon transitions.206) − o − 2 ω) . ﬁnally.168) where the ﬁrst term describes the relevant contribution. 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 × (8.208) has been prepared. The transition requires that the transition dipole moment n| r |0 does not vanish for two states |0 and |n . 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) . as described through the vector potential ˆ A(r. However. induce single photon absorption processes as described by the transition rate Eq. under ordinary circumstances. (8.8: Two-Photon Processes Two-Photon Absorption 231 The interaction of electrons with radiation. One obtains. however.

ˆ The perturbation which arises due to the vector potential (8. for the components of the wave (8. u2 . t) = A+ u1 exp[ i (k1 · r − ω1 t) ] + A− u2 exp[ i (k2 · r − ω2 t) ] .208).213) where N2 /V is the density of photons characterized through k2 .210).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 (8.208) and emit radiation corresponding to the vector potential (8. however. in the present case. later we will allow the quantum system to select appropriate values. 8. We know already from our description in Section 8.208) and the emission of the radiation in (8.141).210) A(r. o1 ˆ o2 ˆ (8.211) accounts for the emitted wave and..211) is stated in Eq.6 above that the absorption of the radiation in (8. according to the ˆ description of emission processes on page 229. the wave characterized through k1 . i.105). The relevant terms of the perturbation (8. In the present case we consider only scattering processes which absorb radiation corresponding to the vector potential (8. The second term in (8. 8.211) is o2 A− = o2 8π (N2 + 1) ω1 V (8. u1 . The vector potential describing the emitted plane wave is then A(r. (8.214) .210).209) cosθ2 where the value of k2 has been ﬁxed. Let us assume for the present that the emitted photon has a wave vector ˆ k2 with cartesian components sinθ2 cosφ2 k2 = k2 sinθ2 sinφ2 (8. t) = Ao2 u2 cos(k2 · r − ω2 t) . the amplititude A− deﬁned in (8.232 Interaction of Radiation with Matter We assume that the experimental set-up also includes a polarizer which selects only radiation with a certain polarization u2 .210) The vector potential which describes both incoming wave and outgoing wave is a superposition of the potentials in (8.208. the amplitude A+ is given by o1 A+ = o1 8πN1 ω1 V (8.211) The ﬁrst term describes the absorption of a photon and.212) where N1 /V is the density of photons for the wave described by (8.105) using the vector potantial (8. ˆ (8. This expansion yields.210) is accounted for by the following contributions of (8.208. hence.e.

Hence.217) 2 Here ro denotes the classical electron radius ro = e2 = 2. evaluating the time integrals in (8. one needs to integrate the rate as given by (8. i. Accordingly.e. one needs to carry out the integration 2 V(2π)−3 k2 dk2 · · ·..212.8 · 10−15 m .181) in stating this result. only these two terms contribute to the scattering amplitude.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. me c2 (8.213) 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.. Following closely the procedures adopted in evaluating the rates of 1st order and 2nd order radiative transitions on page 222–225.215) 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 (8.218) . Only the second (1st order) and the third (2nd order) terms in (8. i.216) over all available modes of the ﬁeld.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 (8.216) 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 .e.215) correspond to scattering processes in which the radiation ﬁeld ‘looses’ a photon ω1 and ‘gains’ a photon ω2 . 8.

. for all states |m of the electronic system (8.219) (8. one can relate (8.221) suggests to expand S( ω) S( ω) = S(0) + S (0) ω + Using 1 − o ± ω one can readily determine m (8. a process in which the electronic state remains unaltered after the scattering. 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. (8. it follows 2 dσ = ro (N2 + 1) dΩ2 |ˆ1 · u2 − S( ω) |2 u ˆ (8.217) to the scattering cross section deﬁned through dσ = It holds then dσ = 2 ro rate of photons arriving in the the solid angle element dΩ2 ﬂux of incoming photons ω2 (N2 + 1) dΩ2 u1 · u2 n|0 ˆ ˆ ω1 1 n|ˆ1 · p |m m|ˆ2 · p |0 u ˆ u ˆ me m m (8. one assumes the even stronger condition ω1 << | o − m| .e. ω1 = ω2 . − o )2 ( m − o )3 (8.223) m Condition (8. consequently.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 (8.222) 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 + . in fact. 2 .234 Interaction of Radiation with Matter The factor N1 c/V can be interpreted as the ﬂux of incoming photons.224) = m 1 − o ( m ω ( ω)2 + + . i. Accordingly..228) S (0) = m S (0) = 2 m .227) u ˆ 0|ˆ1 · p |m m|ˆ2 · p |0 u ˆ u ˆ 0|ˆ2 · p |m m|ˆ1 · p |0 u ˆ + me ( m − o )3 me ( m − o )3 (8.221) Using |n = |0 and. Rayleigh Scattering We turn ﬁrst to an example of so-called elastic scattering...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.

this term cancels the u1 · u2 term in (8. m|ˆ1 · p |0 u ˆ 1 = − m|ˆ1 ·r |0 u me ( m − o ) i (8. .235) and obtain the leading contribution to the expression (8.224. 8. u u u u (8.232) Obviously.236) is of great practical importance.234) where we used for the second identity the fact that u1 ·r and u2 ·r commute. For this purpose we replace p using (8. ˆ ˆ S (0) given in (8. ˆ We want to simplify ﬁrst (8.229) both to u1 · p and to u2 · p which results in ˆ ˆ ˆ ˆ me S (0) = 2 ( 0|ˆ2 ·r |m m|ˆ1 ·r |0 − 0|ˆ1 ·r |m m|ˆ2 ·r |0 ) .236) 2 × m 0|ˆ∗ ·r |m m|ˆ2 ·r |0 u1 u 0|ˆ∗ ·r |m m|ˆ1 ·r |0 u2 u + m − o m − o We have applied here a modiﬁcation which arises in case of complex polarization vectors u which ˆ describe circular and elliptical polarizaed light.222).8: Two-Photon Processes 235 ˆ These three expressions can be simpliﬁed using the expression (8.108) for the identity operator.229) (8..235) We can now combine eqs.228) provides then the ﬁrst non-vanishing contribution to the scattering cross section (8.108) yields S (0) = me 2 0| [ˆ2 ·r.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 (8.108) this is 1 0|ˆ1 ·r u2 · p − u2 · p u1 ·r |0 . It explains. (8.182) 0|ˆ1 · p |m u ˆ 1 = 0|ˆ1 ·r |m u me ( m − o ) i This transforms (8. For this purpose we apply (8.226). Using again (8. the blue color of the sky and the polarization pattern in the sky which serves many animals. Expression (8.233) m Employing again (8.230) According to (8. 8.182) for p and the expression (8. u ˆ ˆ ˆ ˆˆ i The commutator property [xj .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 . u1 ·r ] |0 = 0 u ˆ (8. pk ] = i δjk yields ﬁnally ˆ S(0) = 1 S(0) = i 3 3 (8. i.222).e.234. pk ]|0 = u u ˆ j. 8.231) (ˆ1 )j (ˆ2 )k 0|[xj . for example.k=1 j. ˆ ˆ We want to prove now that expression (8.222) of the cross section for Rayleigh scattering dσ = 2 ro m2 ω 4 (N2 + 1) dΩ2 × e (8.232.229) both for the u1 · p and the to u2 · p terms in (8. as a compass.8.k=1 (ˆ1 )j (ˆ2 )k δjk = u1 · u2 u u ˆ ˆ (8. honey bees.227) vanishes.

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 .236 Thomson Scattering Interaction of Radiation with Matter We consider again elastic scattering. however.220) 2 dσ = ro (N2 + 1) dΩ2 |ˆ1 · u2 |2 . The ﬁrst choice is ˆ sinφ2 k2 × k1 (1) u2 = ˆ = −cosφ2 (8. u ˆ 1 ao . (8. To evaluate |ˆ1 · u2 |2 we assume that k1 is oriented along the x3 -axis and.240) cosθ2 We choose for the polarization of the incoming radiation the directions along the x1 ..241) 0 0 Similarly. |n = |0 and ω1 = ω2 = ω in (8. i.238) (8. the emitted u ˆ radiation is decribed by the wave vector sinθ2 cosφ2 k2 = k1 sinθ2 sinφ2 (8. though.220). that the dipole approximation is still valid which restricts the applicability of the following derivation to k1 >> One obtains immediately from (8. We want to assume. u2 = u2 ˆ ˆ ˆ ˆ cos2 φ2 cos2 θ2 sin2 φ2 for u1 = u1 . u2 = u2 ˆ ˆ ˆ ˆ (8.237) The resulting scattering is called Thomson 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 .e. ao Bohr radius .243) cos2 θ2 cos2 φ2 for u1 = u1 .240).244) . for all states |m of the electronic system .242) |k2 × k1 | 0 where the second identity follows readily from k1 = e3 and from (8. u1 = 1 ˆ ˆ (8. u2 = u2 ˆ ˆ ˆ ˆ for u1 = u1 ..and the x2 -axes 1 0 (1) (2) u1 = 0 . hence. assume now that the scattered radiation has very short wave length such that ω >> | o − m| . (8. u2 = u2 ˆ ˆ ˆ ˆ (1) (2) (2) (1) (1) (2) (1) (2) (2) (1) needs to be (8.239) We will show below that this expression decribes the non-relativistic limit of Compton scattering.

The energy deﬁcit is used to excite the system. If energy is absorbed by the system. but most often involves other degrees of freedom. deﬁned in (8. The Compton scattering cross section which is derived from a model which treats photons and electrons as colliding relativistic particles is dσ tot = dσ tot where (rel) = 2 ro (N2 + 1) 2 ω2 ω1 2 ω1 ω2 + − sin2 θ2 ω2 ω1 dΩ2 (8.. the resulting scattering cross section is 2 ro (N2 + 1) ( 1 + cos2 θ2 ) dΩ2 .247. i.247) becomes identical with the Thomson scattering cross section (8. In the case that the scattering excites other than electronic degrees of freedom.)n .220) represent actually electronic as well as nuclear motions.)n . one speaks of Stokes scattering.8.8: 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 . Raman Scattering and Brillouin Scattering We now consider ineleastic scattering described by the Kramers-Heisenberg formula. The radiation scattered at right angles is even completely polarized (1) along the u2 -direction. Such scattering is called Raman scattering. the scattered radiation is polarized to some degree. Since the scattering is inelastic.249) dσ = ro (N2 + 1) ω1 −1 −1 ω2 − ω1 = .220) vanishes and one obtains in case of Raman scattering ω2 2 dΩ2 | u2 · R · u1 |2 ˆ ˆ (8. 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.g. the scattering is termed Brillouin scattering. ˆ In case that one measures the scattered radiation irrespective of its polarization.e. In the case of such scattering an electron system absorbs and re-emits radiation without ending up in the initial state. For electronic systems in molecules or crystals the degrees of freedom excited are nuclear motions. one speaks of anti-Stokes scattering. if energy is released. in case of a diatomic molecule |n = |φ(elect. as in the case of excitations of accustical modes of crystals.e. i. the states |n etc.247) ( 1 − cosθ2 ) (8.246) 2 This expression is the non-relativistic limit of the cross section of Compton scattering. the ﬁrst term in (8.246). In case that the nuclear degrees of freedom excited absorb very little energy.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.. molecular vibrations or crystal vibrational modes. (8. φ(vibr. e.248) me c2 One can readily show that in the non-relativistic limit. The excitation can be electronic. for c → ∞ the Compton scattering cross section (8.. or in case of translational motion of molecules in liquids.

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

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

(j. For the following we will require only two properties of SN . number k under the entry j of the ﬁrst row indicating that particle j is switched with particle k.3) the j-th index in the second row. However. i. xP (j) . x2 . namely implicitly the group property. . . xN ) = Ψ(xP (1) . . We will not discuss here at any depth the properties of the permutation groups SN . particles 4 and 5. xP (2) . 5) (9. (9. 8) . the representations in the space of N –particle wave functions are extremely useful in dealing with N –particle systems. 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 = .240 Many–Particle Systems repeatedly like degenerate states. the set of all permutations of N objects. 3) T (7. 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. For the permutations P1 and P2 given above holds P1 = T (3. Denoting by P (j) the image of j.e. 6) T (6. k) are characterized by the two indices (j.. P2 = . and the second row showing the numbers 1 to N again. xj . The latter factorization has the essential property that the number of factors is either even or odd.. . not any state can be chosen as to represent the many– particle system.5) . in notation (9.3) P2 which aﬀects only two particles. one can state P Ψ(x1 . Transpositions denoted by T (j. . .. . . . The permutations we need to consider for a system of N particles are the elements of the group SN . k) of the two particles transposed. One calls permutations of the ﬁrst type even permutations and permutations of the second type odd permutations. This property will be established now. . in particular. A reader interested in the quantum mechanical description of N particle systems is strongly encouraged to study these representations. the top row representing the numbers 1 to N labelling the N particles under consideration. 5) T (5. leaving all other particles unchanged. . We can then state P2 = T (4. . (9. 5).e. i. .e. i.2). k) = (4. P2 = T (4. 5) in case of P2 . . . 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. . The elements P ∈ SN can be represented conveniently as 2 × N matrices. i. a given P can either only be presented by even numbers of transpositions or by odd numbers of transpositions. k).. . .4) both permutations being obviously odd. 4) T (4. . and explicitly the fact that any P ∈ SN can be given as a product of transpositions T (j. . The group SN is then composed of two disjunct classes of even and of odd permutations.e. even though these properties. rather only states which obey a certain transformation property under permutations are acceptable. but in a diﬀerent order. xP (N ) ) . is called a transposition.

.e. . . Presently. (9. i. In the latter case all even permutations are represented by ‘1’ and all odd permutations by ‘-1’. i. (9.7) This property. {1} (trivial) and {1. a diagonal representation can only berealized in a simpler group. The ﬁrst case applies to bosons. In fact. the Hamiltonian must be of the type N H = j=1 1 F (xj ) + 2 N G(xj .e.-1}. From this property follows that H is independent of any speciﬁc assignment of indices to the particles (note that (9.k=1 (9.11) . The single particle states are assumed to be orthonormal. or the corresponding group of two elements 1.9. 1 together with multiplication.8) |Ψ (9. P ] = 0 . The terms of the Hamiltonian will be discussed further below.. G(xj . the second group to fermions.e. 2. implies that the permutation operators can be chosen in a diagonal representation. usually a very large number. provide a representation of the group structure represented by the multiplication table of even and odd permutations given above. ) are the same for all particles and for all pairs of particles. ∀P ∈ SN . both groups.10) We will construct now wave functions which exhibit the appropriate properties.6) which describes one–particle interactions and two–particle interactions of the system. it is essential that the functions F ( ) and G( .5) implies that a permutation eﬀectively changes the indices of the particles) and. the representation is either isomorphic to the group composed of only the ‘1’. The boson and fermion property stated implies that for a system of N bosons holds P |Ψ = |Ψ For fermions holds P |Ψ = where P P ∀P ∈ SN . i. particles with half–integer spin values. xk ) is symmetric in its two arguments. i. In fact.. S where S denotes the number of available single particle states.9) = 1 for P even −1 for P odd (9. xk ) j. k = 1. hence. it is a postulate of quantum mechanics that for any description of many–particle systems the permutation operators must be diagonal.. −1. . Since permutations do not necessarily commute.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. for the scalar product holds j|k = δj k . i. all particles being governed by the same interactions. particles with integer spin values. In fact.e. Fock-Space Our derivation will start by placing each particle in a set of S orthonormal single particle states x|k = φk (x). in turn. the Hamiltonian for permuted indices P HP −1 is identical to H. Obviously.e. (9. From the latter results [H.

.12). λ2 . 2.g. . . . xN |Ψ Fock (λ1 . . Λ(λ1 . . . k) ∈ Λ(λ1 .. 9. each particle j into a speciﬁc state |λj . . therefore.9). . An important detail of the deﬁnition (9.15) with λk = j. .12) where the second factor represents the scalar product between spin states. . . λN ) ⊂ SN is the set of all permutations involving only particles j with diﬀerent λj . .. The Fock states represent N particles which are placed into S states |λ1 . .. The numbers nj . . . .8. e. .λN ) P x1 . . with λj = λk = . . . nS ). . . λN ) = j=1 φλj (xj ) .11. N x1 . . |λ2 . . . . λ2 . . are essential. .15) (nj !) N! 1 2 P ∈Λ(λ1 . . λ2 . λN )|ΨFock (λ1 . characterize the wave function (9. We ﬁrst consider orthonormal many–particle wave functions.e.. λj ∈ {1.. |λN . the so-called Fock-states.14) Obviously. x2 . A wave function which obeys the boson symmetry (9. . .λN ) on the rhs. n2 . i. (9. . . . x2 . . the states |ΨFock (λ1 . λ2 . . . is a particular choice of placing N particles consistent with the occupation numbers (n1 . λN ) do not exhibit the symmetries dictated by (9. λN ) only if λj = λk . S} of the type (9. . λN ) = j=1 δλj λj . . which do not yet obey the symmetries (9. .13) These states form an orthonormal basis of N -particle states. . . . 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.. . It holds N ΨFock (λ1 .15) completely and.λN ) Here the states (λ1 . referred to as the occupation numbers. i. λ2 . .9). k. n2 . appear severalfold in (9. T (j. 9. In a second step we form linear combinations of Fock states with the desired symmetries.8) is x1 . λ) in which two particles occupy the same single– particle state |λ does not allow transposition of particles.e. λ2 . . . hence. .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. xN |ΨB (n1 . This detail is connected with the fact that a two–particle state of the type ΨFock (λ. . . . nS ) = S j=1 (9. . the nj specify how often a single particle state |j is occupied. (9.8. . xN |ΨFock (λ1 .15). . . the reason is that such transposition duplicates the state which would. . . The integers nj are equal to the number of λk in (9. x2 . 9. . The reader is well advised to pause and grasp the deﬁnition of the nj . . .15) is that the sum over permutations does not involve permutations among indices j. .

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

. . (9. . [a† .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 . These operators allow one to construct and manipulate many–particle wave functions while preserving permutation symmetry.20). . k The operators a† and aj are referred to as the creation and annihilation operators of bosons for the j orthonormal single–particle states |j . Exercise 9. a† ] = δj k . nS ) . (9. . . . . Employing the superindices N and N introduced above and using the orthonormality property (9. (b) Show that the states (9. . . . Creation and Annihilation Operators for Bosons We consider the operator aj deﬁned through aj ΨB (n1 . .23) (9.9). .244 Many–Particle Systems The representation of wave functions (9. . n ) are normalΨ 1 j 1 j S S ized according to (9. . .16) one obtains (Aj )N = ΨB (N )| aj |ΨB (N ) = √ S N nj δnj nj −1 k=1 k=j δnk nk . nS ) = nj ΨB (n1 . . . . . .17) obey the orthonormality property (9. . since both the N -particle wave function B (n . nj .22) The operators aj and a† obey the commutation properties j [aj . . n2 . . . . For the adjoint operator a† holds j a† ΨB (n1 . .17) obey the fermion symmetry (9. . n ) and the N − 1-particle wave function ΨB (n .15).2 Operators of 2nd Quantization An important tool to describe mathematically systems of many bosons or fermions. nj + 1. nS ) nj ≥ 1 0 nj = 0 √ . . .25) .21) we consider the matrix Aj corresponding to aj in the basis of many–particle states (9. . nj . . . . 9. . .15) and (9. . guaranteeing the correct permutation properties of the many–particle states. ak ] = 0 . . n .2: (a) Show that the states (9. . . . . a† ] = 0 j k (9. .1. . (9.22) follows from (9. .15). . nj − 1. n − 1. . . nS ) = j nj + 1 ΨB (n1 . To prove that (9.21) √ The factor nj appears here on the rhs. are the so-called operators of 2nd quantization. . nS ). .24) [aj . . .

j (9. (9. (9.22) ˆ Nj |ΨB (N ) = nj |ΨB (N ) . . . a† ] = 1 For j = k follows similarly 1. j (aj a† − a† aj )|Ψ(N ) = k k ( nj (nk + 1) − (nk + 1)nj )|Ψ(n1 . . 9. .e. . The related operator S ˆ N = j=1 ˆ Nj (9.26) From this result one can immediately conclude that (9.27. .29) yield the commutation relationship (9. nS ) = 0 . obviously.2: Operators of 2nd Quantization Let A† be the matrix adjoint to Aj . . n2 . . . S ˆ N |ΨB (n1 . 9.9. . . .31) This operator is diagonal in the occupation number representation. .21) and vice versa. . .21. . .32) ˆ ˆ i. n2 .34) = . The creation operators a† allow one to construct boson states ΨB (N ) from the vacuum state j |0 = |ΨB (n1 = 0.27) Equations (9.30) Of particular interest is the operator ˆ Nj = a† aj .. . . . .. nS ) = |ΨB (N ) .24). nj − 1 . nS ) = j=1 nj |ΨB (n1 . n2 . the eigenvalues of Nj are the occupation numbers nj .23) follow in an analogous way.22) follows from (9. n2 . . It holds for j = k (aj a† − a† aj )|Ψ(N ) = (nj + 1 − nj )|Ψ(N ) = |Ψ(N ) . nS ) = j=1 a† j nj nj ! |0 . . (9.29) |ΨB (n1 . (9. . (9. . . . One refers to Nj as the occupation number operator. . for basis states |ΨB (n1 . nS = 0) as follows S (9.e. nS ) N |ΨB (n1 . nS ) . We will prove the commutation properties (9. . . One obtains for its matrix elements j A† j N N 245 = ΨB (N )| a† |ΨB (N ) = (Aj )N N j S k=1 k=j nj δnj nj −1 δnk nk = nj + 1 δnj nj +1 S k=1 k=j δnk nk . the properties (9. i. j j Since this holds for any |Ψ(N ) it follows [aj .24). nk + 1 . n2 = 0. n2 . . One can readily show using (9.28) (9.33) is called the particle number operator since.

. .19) without aﬀecting the fermion symmetry (9. 9. xN |ΨF (n 1 . .2. nj . .e. nj+1 .17. .35) and where the derivative operation is deﬁned through ∂f (A) = ∂A ∞ n=1 1 f (n) (0) An−1 . . . . nj+1 . x2 . . 9. .21. since they are induced through the algebra of a† and aj . . x1 .24) and the property aj |0 = 0. . one does not need to worry ever about proper symmetries of wave functions.22. nj (9. . . . . .e.9). nj+1 = 1. .. the fermion wave function changes sign when one exchanges the order of the occupancy. .38) . .2. φj+1 (x1 ) φj+1 (x2 ) . . 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. xN |ΨF (n1 . 9. in case it holds for nj .23.2: Show that the boson operators a† and aj satisfy j [aj .38) corresponds to x1 .. Exercise 9. φj (x1 ) φj (x2 ) . φj (xN ) φj+1 (xN ) . .g..30). .s.29. (9. nS ) = − x1 . . in case that the single particle states |j and |j + 1 are both occupied. nS ) = . nS ) . . .23.h. ..36) Creation and Annihilation Operators for Fermions We want to consider now creation and annihilation operators for fermions. . 9. . (n − 1)! (9. √1 N! . . .246 Many–Particle Systems In using the boson creation and annihilation operators. . .37) (9.. it also holds for nj + 1. nj . from the vacuum state |0 or states |ΨB (N ) .23) hold for the state deﬁned through (9. . Prove this by induction. . ..39) . . j Exercise 9. . Before proceeding in this respect we need to account for the following property of the fermion wave function which applies in the case nj . nj+1 . x2 .1: The commutation relationships (9. of (9.. . e. As long as one starts from a wave function with the proper boson symmetry. . x2 . i. xN |ΨF (n1 . i. f (a† )] j [a† . 9. To prove this property we notice that the l. one needs to apply only the commutation propertes (9. . showing that the property holds for nj = 0 and.24) and aj |0 = 0 imply that the the properties (9... .. . Obviously. in general. 9. operators which can alter the occupancy of the wave function (9. .

. . .e. . This change of position is connected with a possible sign change (−1)qj where qj is deﬁned for a given N = (n1 . i. . .. φ2 (x2 ) φ5 (x2 ) (9. φj (x1 ) φj (x2 ) . . nj = 0. φj+1 (xN ) φj (xN ) ..38) reads − x1 .. .19). x2 |ΨF (n1 = 0. . n2 . nS = 0) φ2 (x1 ) φ5 (x1 ) 1 1 = √2 = √2 [φ2 (x1 )φ5 (x2 ) − φ5 (x1 )φ2 (x2 )] . at the ﬁrst column of the Slater determinant (9. . 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. φj+1 (x1 ) φj+1 (x2 ) .42) i.s.39) and (9. . the rhs. . .. . (9.e. . Because of the property of the determinant to change sign when two columns are interchanged. . . the expressions (9. x2 . in order to be consistent with this convention. nj+1 . (9. . . it is necessary to specify for a fermion wave function the order in which the single– particle states |λj are occupied.. n2 . . n5 = 1. n2 = 1. . . . . nj → 0. n6 = n7 = . n3 . . xN |ΨF (n1 .40) are identical.19). i. . . . 2. .e. nS ). that occupancies are always ordered according to a monotonous increase of the single–particle state index.h. . nS ) qj states occupied In case that the single particle state |j is not occupied. .. nj . n2 .. vanishes. Obviously. nj . n4 = 0. . (9. n2 . i. . 247 − √1 ! N .. .2: Operators of 2nd Quantization The r. nS ) = nj (−1)qj |ΨF ( n1 . nS ) as follows: j−1 qj = k=1 nk . A proper example is the two particle fermion wave function x1 . This requires one. of (9. . . .. n3 = 0. . nS ) = . We are now ready to deﬁne the annihilation operator cj for fermions in the single–particle state |j as follows cj |ΨF (n1 . .43) . . One adopts the convention that particles are created and annihilated by these operators always at the ﬁrst position of the wave function. .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. at which column of the Slater determinant (9. We will be adopting below the notational convention that qj corresponds to occupancies N = (n1 . 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). . however. .e. qj is the number of states |k with k < j which are occupied in a fermion wave function. particles are being created or annihilated. For this purpose one must adhere to a strict convention: the labelling of single–particle states by indices j = 1.. .. . nj+1 . nj−1 .. .40) .9.

. n2 .46). (9.248 The operator c† adjoint to cj exhibits the following property j c† |ΨF (n1 . ck ]+ = 0 . nj → 1 .49) The derivation involves realization of the fact that a non-zero result is obtained only in case nj = 1. 9. nS ) = j (−1)qj (nj + 1))(−1)qj (1 − nj ) |ΨF (N ) = (1 − nj ) |ΨF (N ) (9. Let Cj be the matrix corresponding to the operator cj in the basis of fermion states (9. . The elements of Cj are then ( note that nj only assumes values 0 or 1) S (Cj )N N = Ψ (N )| cj |Ψ (N ) = nj (−1) δnj nj −1 k=1 k=j F F qj δnk nk . nj → 0 . of (9. . .47) Let C† be the matrix adjoint to Cj . c† ]+ = 0 j k (9. . n2 . n3 . otherwise. We ﬁrst show that (9. We will prove now the anti-commutation property (9. . .17.19).45) (9.44). It holds c† cj |ΨF (N ) = c† (−1)qj nj |ΨF (n1 .43).50) .46) [cj . nj . We consider (9. nj → 1. One obtains for its matrix elements j C† j NN = ΨF (N )| c† |ΨF (N ) = (Cj )N j S N = nj (−1)qj δnj nj −1 k=1 k=j δnk nk S = (−1) (1 − nj ) δnj nj +1 k=1 k=j qj δnk nk . nS ) = j Many–Particle Systems (−1)qj (1 − nj ) |ΨF ( n1 .44) follows from the deﬁnition (9. nS ) qj states occupied The operators thus deﬁned obey the commutation properties [cj . nS ) = j j (−1)qj (1 − nj + 1)(−1)qj nj |ΨF (N ) = nj ΨF (N ) (9.48) vanishes. . The anti-commutation properties (9. . c† ]+ = δj k k where we have used the deﬁnition of the so-called anti-commutators [A. We have used here the deﬁnition qj = k<j nk along with qj = qj . . Similarly one obtains cj c† |ΨF (N ) = cj (−1)qj (1 − nj ) |ΨF (n1 .48) From this follows (9.44) (9. . [c† . .46) ﬁrst for the case j = k. B]+ = AB + BA. . (9. nj+1 . the last factor on the rhs. . . . nj−1 .45) follow in an analogous way and will not be derived here. .

.50) yield ( cj c† + c† cj ) |ΨF (N ) = |ΨF (N ) .) = c† 1 c† 2 · · · c† N |0 . . k k 249 (9. Correspondingly. in the basis |ΨF (N ) . coordinates x and index j into a space part and a spin part. Since (9. . ˆ It is an interesting exercise to demonstrate that Nj only has eigenvalues 0 or 1. nS ) k and.h. . σ 1 (σ = ± 2 ) .51.55) is called the particle number operator. .. i. . i.s.2: Operators of 2nd Quantization (9. .58) .29)) |ΨF (. nS ) k c† cj |ΨF (N ) = (−1)qj +qk −1 nj (1 − nk )|ΨF (n1 . of this equation we meant to indicate only those N occupation numbers nλj which are non-vanishing. λ λ λ (9. . nk = 1 .. . nj = 0 . hence.57) On the l. . an operator cj must be on the left of an operator ck for j < k.51) (9. is referred to as the occupation number operator.53) is diagonal in the occupation number representation.46). 2 (9. . j j We ﬁnally note that the creation operators c† allow one to construct any fermion state |ΨF (N ) j from the vacuum state |0 deﬁned as above (see (9.h.54) ˆ N = j=1 ˆ Nj (9. . with eigenvalues equal to the occupation numbers nj ˆ Nj |ΨF (N ) = nj |ΨF (N ) ˆ Nj .49) and (9.e. j j For j < k one obtains cj c† |ΨF (N ) = (−1)qj +qk nj (1 − nk )|ΨF (n1 . We j will often separate the states |j .s. for which purpose one needs to envoke only the algebraic (anti-commutation) properties (9. 9. . 1}.9. nk = 1 . . nλ1 . j → j. Equation (9. i. 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 . 1 In the following we will apply fermion operators c† and cj only to electrons which carry spin 2 .. hence. .e. j j j j j j j (9. .56) Here we have made use of cj cj = 0 and c† c† = 0 which follows from nj ∈ {0.52) hold for any |ΨF (N ) one can conclude (9. S (9. the creation operators must operate in the proper order. x|j → φj (r) | 1 σ . . . ( cj c† + c† cj ) |ΨF (N ) = 0 .49) shows that the operator ˆ Nj = c† cj j (9.46). nλN . nλ2 . nj = 0 .e. .52) One can obtain similarly the same relationship in the case j > k. On the r. .

ˆ r|f |s = 1 ∗ ˆ d3 r ψr (r.F (N ) .F (N ) . . xN .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. We adopt a similar expression to distinguish two–particle operators G and g . 9.60) is that the sum over particles in (9. N are not all identical.15.62) These matrix elements can be obtained by chosing the operators and many-particle states in the position representation. j = 1.17).250 Many–Particle Systems 9. respectively – which are all identical. 2. Operators of the type 1 ˆ G = 2 N g (xj .. . 9.F (N )|F |ΨB.k=1 j=k (9. 9.15. ˆ ˆ ˆ We seek to determine now how one–particle operators F and two–particle operators G act on many– B (N ) and |ΨF (N ) . 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. (9. (9. . σs ) (9. . .59.F (N )|G|ΨB. xk ) ˆ j. 9. S of single–particle spinorbital states (see page 241) and the operators are speciﬁed through the associated generic operators ˆ f and g .61) is not a one particle operator as long as the N operators rj (x).60) ˆ correspondingly are called two–particle operators. . An operator N ˆ R = j=1 rj (xj ) ˆ (9. . .63) ˆ ˆ This expression has been speciﬁed for the purpose of these notes to distinguish F and f and is not common ˆ terminology. The essential aspect of deﬁnition (9. respectively.17) and integrating over all particle space-spin coordinates x1 .6) above. i. Since the many–particle states are built-up from a basis |r .59) and (9. . σr ) f (x) ψs (r. x2 . The action of the operators boson and many–fermion states |Ψ is obviously described through the many–particle state matrix elements ˆ ΨB. ˆ ΨB. . These operators had been introduced already in Eq. 2.60) involves generic operators – acting on xj and on xj . (9.59) and over pairs of particles in (9.e. namely. r = 1. one expects that the matrix elements can actually be expressed in terms of matrix elements ˆ involving solely the generic operators and the single–particle states. x ) which constitute G are ˆ called the associated generic two–particle operators.59. ˆ .59) ˆ ˆ are called single–particle operators. xk . The operators f (x) which constitute F are called the associated generic single–particle operators1 . The operators g (x.

i. We will show that the boson and fermion creation and annihilation operators serve this purpose. ∂rj2 (9. s and t. g g (9.65) This symmetry will be exploited repeatedly below.68) Another single–particle operator is the one–particle density operator N ρ(r ) = ˆ˜ j=1 δ(r − rj ) . σs ) g (x. An example for a two–particle operator is the Coulomb repulsion operator 1 ˆ V = 2 N j. σt )ψu (r . according to (9. 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. one cannot switch the indices of one pair individually.64).e..k=1 j=k q2 |rj − rk | (9. 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. x ) ψt (r.9. s|ˆ|t. u which follows directly g from the deﬁnition (9. s|ˆ|t. σr )ψs (r . are spin-independent as is evident from the factor δσσ . ∂r 2 ∂x2 ∂x2 ∂x2 1 2 3 (9. s|ˆ|t.63.64) We like to note an important symmetry of the matrix element r.71) . 9.62) in terms of the matrix elements of single–particle states (9.67) ˆ In this case the generic single–particle operator. u = s. t . The matrix elements of this operator in a single–particle basis are ˆ rσ|t|sσ = δσσ d3 r φ∗ (r) r 2 − 2m ∂2 φs (r) .69). ˜ (9. ˆ 251 (9. that the symmetry implies that one can switch simultaneously the pairs of indices r. is t(x) = (− 2 /2m)(∂ 2 /∂r 2 ).66) Here we adopt the notation for the Laplacian ∂2 ∂2 ∂2 ∂2 = + + .64) r.59).3: One– and Two–Particle Operators r. σu ) . (9. ∂r 2 (9. u = g d3 r ∗ ∗ d3 r ψr (r. r|ˆ|u. u. Notice.66) and (9. Examples of One– and Two–Particle Operators An example for a single–particle operator is the kinetic energy operator N ˆ T = j=1 2 − 2m ∂2 .69) ˆ In this case the generic operator is f (x) = δ(r − r).70) Both operators.

3 . The ˆ matrix elements of the generic operator in terms of single–particle states are ˆ r. β|ˆ3 |s.− S2. not in the strict sense of our deﬁnition above since the restriction 2 j = k does not apply in the summation. σ. u. Sj is the spin operator for particle j with components (in the spherical representation) Sj. The generic operator is s12 = S1 · S2 = ˆ 1 1 S1. 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 (9.75) is a two–particle operator.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.+ .e. σ . σ .+ S2. The operator N S = j=1 Sj (9. β s r. In this case the generic operator is v (x. . α s r. β = ˆ 2 2 1 −2) r. Sj.74) = N j=1 Sj = 2 N Sj · Sk j. The generic operator is given through the Pauli matrices. α|ˆ+ |s.76) the single–particle state matrix elements of which are r. β s The operator S 2 = δrs = δrs = = + − 1 2 1 2 δrs δrs (9. σ | t |t. σ | s12 |t. β|ˆ− |s.252 Many–Particle Systems which is also spin-independent. 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 + . σ.3 S2. u.− .77) δσ σ ( δσσ − δ−σσ ) The operators (9. However.3 2 2 (9. Here. s. α s r. s = 1 σ.− + S1. The non-vanishing matrix elements of the spherical components of s are (α = + 1 . i. x ) = q 2 /|r1 − r2 |. α|ˆ3 |s.72) As a ﬁnal example of one– and two–particle operators we consider operators of total spin.77) are orbital-independent as evidenced by the factors δrs and δrt δsu in the respective formulas.+ + S1. s.k=1 (9. however. (9.73) is a one-particle operator..

59) and G in (9.63) and (9. s|ˆ|t.176 where the matrix elements of fermion states can be found.79) It is crucial to notice in the expression for the fermion two–particle operator that the order of the annihilation operators in (9.82) ΨF (N )|c† cs |ΨF (N ) = r cr 2 (9.t. namely cu ct . and then in g a second step the matrix elements ΨB (N )|a† as |ΨB (N ) .62). 9.s. In order to determine these matrix elements one needs to evaluate ﬁrst the matrix elements ˆ of the generic operators r|f |s r. g and not ct cu . xk ) → ˆ j. s|ˆ|t.s=1 S r. 9.k=1 j=k S r. 9.s ΨF (N )|cs ΨF (N ) = nr (−1)qr ns (−1)qs δnr −1 nr δns ns −1 δnt nt We refer the reader to Condon and Shortley’s ‘Theory of Atomic Spectra’. u as deﬁned in (9.46). respectively.62) are correct one needs to compare the result derived with conventional derivations of the matrix elements2 . (9.9. u cr cs cu ct g fermions (9. To show that the resulting values of the matrix elements (9. 9.F (N ) have the same values as the respective matrix elements (9. u ..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. u a† a† at au bosons g r s † † r.64) one replaces the operators F in (9. r S Ψ (N In case of r = s one obtains F )|c† cr |ΨF (N ) r = nr r=1 δnr nr . r ΨB (N )|a† a† at au |ΨB (N ) r s ΨF (N )|c† c† ct cu |ΨF (N ) r s bosons fermions (9.79) in the basis of many–particle states |ΨB.u=1 (9.81) For the latter matrix elements simple rules can be derived from the algebraic properties of the boson and fermion operators (9.80) (9.s=1 ˆ r|f |s a† as bosons r ˆ r|f |s c† cs fermions r r.78) 1 2 g (xj .45.78. respectively.60) as follows N ˆ f (xj ) → j=1 N 1 2 1 2 S r.62) through single–particle state matrix ˆ ˆ elements (9.s.79) is opposite to that in the matrix element r.e. s|ˆ|t. These rules will be provided below only for the case of fermions.79) have the following implication: The operators (9. r ΨF (N )|c† cs |ΨF (N ) . s|ˆ|t.171 and pp.64). Expressions (9. The matrix elements are then actually those of the number operator c† cr which is diagonal in the occupation number representation.24) and (9.63.u=1 S r.78) and (9.23.t. .83) t=1 t=r. The Matrix Elements ΨF (N )|c† cs |ΨF (N ) r We consider ﬁrst the case r = s. i. pp.

. i.e. Comparision with the results in Condon&Shortley (pp.s=1 a. u) are included in the summation for which all indices are diﬀerent.84) we can conclude that for diagonal elements holds S Many–Particle Systems ˆ ΨF (N )|F |ΨF (N ) = r=1 ˆ nr r|f |r (9. t=u 2 = 0.53) 1 ˆ G0 = 2 S r s|ˆ|r s − r s|ˆ|s r g g r. ˆ ˆ Nr Ns . (r.s=1 a. s. t. c2 = 0 r (9.45) yields together with the deﬁnition of the occupation number operator (9.86) into three contributions.59) when the matrix elements are determined between Slater determinant wave functions. ˆ ˆ ˆ ˆ Accordingly. 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. in which case ns = ns + 1 and nt = nt − 1 holds. ˆ The ﬁrst contribution G0 is 1 ˆ G0 = 2 S r s|ˆ|r s c† c† cs cr + g r s r. where each contribution corresponds to one of the three cases mentioned. (r.s.u r=s.t. we split the sum in (9. t.i.’ (all indices diﬀerent) that only tuples (r.s=1 a.86) For the combination of indices r. g r s r.254 From (9. 9. 1 2 S r s|ˆ|s r c† c† cr cs . The anti-commutation property (9.i. 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.78) does yield the same results as the operator deﬁned in (9.i.d..88) . s.85) All other matrix elements vanish. Starting from the deﬁnition (9. r = t etc. s).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.87) Here and in the following we denote by ‘a. G = G0 + G1 + G2 . 171) shows that the operator deﬁned in (9. t. u need to be considered. s. namely.d. u in this sum three possibilities exist Case 0 two of the four indices are diﬀerent. Case 1 three of the four indices are diﬀerent.d. t). i. (9.82.e.i. t. those for which N and N diﬀer in more than two occupation numbers. s. (9. for which holds r = s.d.

s r. 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. As we will ﬁnd. N = N or nr = nr for all r.s. this contribution to the two–particle operator is diagonal in the occupation number ˆ representation. this part accounts for all diagonal contributions to G.45). F (N )|G|ΨF (N ) .d.d. g r s (9.t. s − r.u r>s. contributes in this case. r g g .s. ˆ Case 0 This case covers diagonal matrix elements.t=1 a. nt .3: One– and Two–Particle Operators 255 Obviously. s (9.t=1 a.t. i.t=1 a. (9. r s|ˆ|r t − r s|ˆ|t r g g ˆ c† ct Nr .u r>s.i.t.t..i.s. S r.u r>s.92) ˆ This contribution to G has non-vanishing matrix elements only when N diﬀers from N in four ˆ occupation numbers nr .t<u S r.t>u r s|ˆ|t u − r s|ˆ|u t g g c† c† cu ct .s.89) S r.65).i. this contribution obviously behaves similarly to the oﬀ-diagonal part (9.i.u r<s. ns .9. 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. 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. nu .t=1 a.93) .s.s.90) ˆ Since Nr is diagonal.84) ˆ of the one–particle operator F . employing the symmetry property (9. A similar contribution does not arise for F .s.d.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. s|ˆ|s. r s (9. Commutation of ct cu according to (9.86).t.t>u S r.i. 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 . (9. s|ˆ|r.s.u r<s. One obtains 1 ˆ ΨF (N )|G|ΨF (N ) = 2 S nr ns r.e.t=1 a. ˆ The second contribution G1 is ˆ G1 = + + + 1 2 1 2 1 2 1 2 S r. Only G0 .d. given in (9.d.91) ˆ G2 = r.s.88).s. S r.t>u S r.

t for which holds ˆ ns = ns + 1 and nt = nt − 1. nv = nv − 1. u < v or s > r. r=1 r=1 Exercise Exercise Exercise Exercise Exercise 9. The matrix elements of the corresponding generic operators are (9. Only G1 . Only G2 .62) for bosons. s|ˆ|r. u g g (9.89).90). u < v. Derive (9.91). s|ˆ|v.3.92). Spin Operators ˆ ˆ In this section we will brieﬂy consider the spin operators S given in (9.e.94) Case 2 This case covers oﬀ-diagonal elements with nr = nr except for r = s. One obtains ˆ ΨF (N )|G|ΨF (N ) = (−1)qs +qt (1 − ns )nt r r=s.73) and S 2 given in (9.3. r g g (9. one–particle and two–particle operators. t − r.95) from (9. v − r.92) from (9. respectively. We like to express these operators through ˆ fermion creation and annihilation operators.t nr r. the ‘−’-sign applies for s < r.. contributes in this case.96) . v for which holds ˆ ns = ns + 1. given in (9. Derive (9.75) which are.90) from (9.3: 9.92). i.95) In the latter formula the ‘+’-sign applies for s < r.4: 9. given in (9.2: 9.1: 9.76). u. contributes in this case. In particular. u > v or s > r. particle number is conserved. it holds S nr = S nr . s|ˆ|t.3. nt = nt + 1. s|ˆ|u.256 Many–Particle Systems Case 1 This case covers oﬀ-diagonal elements with nr = nr except for r = s. ˆ ˆ Show that the matrix elements of F and of G conserve particle number. Derive the non-vanishing matrix elements (9.3.5: Derive (9.3. for ˆ non-vanishing matrix elements ΨF (N )|G|ΨF (N ) at most four occupation numbers nr and nr can diﬀer. 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 (9.74) and (9. One obtains ˆ ΨF (N )|G|ΨF (N ) = ± (−1)qs +qr +qu +qv (1 − ns )(1 − nr )nu nv r. All matrix elements which are not covered by the three cases above vanish. t. nu = nu − 1. u > v. Furthermore.

98. 9. e. In fact. The great generality of the concepts developed in the context of many–particle systems can be judged from the fact . Examples is the laser action. The outcome of these studies is that even when interactions between components (particles) are simple.99) will be used. the description of many–particle systems.6: Derive (9.s=1 r. the systems investigated involve a macroscopic number of components such that statistical mechanical approaches are envoked. s|ˆ|u. Often.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. Actually.97) is over the orbital states. would be a simple exercise in linear algebra. v c† c† cu ct . in crystals.t.9.t.99) has preoccupied an important part of all intellectual eﬀorts in Theoretical Physics during the past ﬁfty years. If it were not for the two–particle contribution to the Hamiltonian. the spin part of the single–particle states is represented by α = 1 and β = − 1 . the study of problems governed by Hamiltonians of the type (9.. it is one of the main intellectual achievements of Physics among the Sciences to have addressed systematically the study of systems composed of many strongly interacting components. 9.u S (9.u=1 σ. s|ˆ|t. u c† c† cuσ ctσ v rσ sσ r.. molecules. the concerted behaviour of interacting systems can deviate qualitatively from that of systems made up of independent components. v r s 2 r. however.σ (9. atoms and nuclei. s. which represents the interactions between particles.s=1 c† c† crβ csα .g. 9. the two–particle operator makes the description of many–particle systems a very hard problem. e.s.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.98) In many cases the Hamiltonian is spin-independent and the equivalent notation S H = r. that the two–particle operator representing interactions between particles induces a variety of interesting phenomena.97).3. superconductivity. In the latter case the indices r. The fortunate side of this is.96.96) and (9.s.g. rα sβ (9.97) The summation in (9. 2 2 Exercise 9.s=1 σ 1 ˆ rσ r|t|s c† csσ + 2 S r. evaluation of their stationary states and excitation energies. but also ordinary phenomena like freezing and evaporation. t. Unfortunately. 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. u refer only to the orbital part of the single–particle basis.

The Aufbau principle. Exercise 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. namely of neurons3 . .J. S}. We will restrict our description in the following to systems composed of an even number (2N ) of particles and to spin-independent interactions.4.258 Many–Particle Systems that today these concepts are providing insight into the functioning of biological brains. Amit (Cambridge University Press. 2.s=1 σ ˆ rσ r|t|s c† csσ (9. a Hamiltonian without a two–particle operator contribution accounting for interactions among the particles. . all material systems would condensate into states which would depend little on particle number.. 1989) . can be described in a rather straightforward way. ranging from nuclei to the molecules of living systems . i. Chemistry essentially would know only one element and Life would not exist.100) ˆ ˆ We will denote the matrix elements of t as r|t|s = trs . water-ﬁlled jar all electrons of water turn their fermion nature into a boson nature.102) U ∗rm Urn = δmn . another prototypical systems of many strongly interacting components. The Hamiltonian of such system is S Ho = r. (9. The electronic properties of atoms with diﬀerent numbers of electrons would diﬀer little. m = 1. r=1 3 Urm U ∗sm = δrs m=1 (9. 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). What might happen? Independent-Particle Hamiltonian A many–fermion system governed by an independent-particle Hamiltonian.. i. In such a representation Ho is diagonal and the stationary states can be stated readily.e.101) The unitary property of Urn U† U = U U† = 1 1 or S S (9. electrons and nucleons. {|m). according to which nuclei and atoms change their qualitative properties when they grow larger. 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. where S |m) = r=1 Urm |r .e. . ultimately depend in a crucial way on the fermion character of its constituent building blocks. Transformation to a New Set of Creation and Annihilation Operators Our aim is to represent the Hamiltonian (9. New York.100) in a form in which the factors c† csσ become rσ occupation number operators.1: Imagine that in a closed. If it were’nt for the fermion nature.

d† ]+ = sσ nσ r. the expressions drived above for the matrix elements of one. csσ = n=1 Usn dnσ . of N -Fermion states in which single particle states |nj ) as deﬁned in (9.f. hence. rσ namely. for nσ rσ example.e. .. [d† . [dmσ . ∗ ∗ Urm Urn .104)] S S ∗ Urm d† mσ m=1 c† rσ = . dnσ = r=1 ∗ Urn crσ (9. using (9.s=1 ˆ U † tU mn (9. accordingly. d† ]+ = 0 (9. the operators d† and dnσ . the anti-commutation properties (9.108) These operators describe the creation and annihilation operators of fermions in states |n) which are linear combinations (9.4: Independent-Particle Models allows one to express conversely |r in terms of |m) S 259 |r = m=1 ∗ Urm |m) . In a basis of states N d† j σj |0 n j=1 (9.108) and (9.s.105) where ˜ ˆ tmn = (m|t|n) = S ∗ trs Urm Usn = r. d† ]+ = δmn δσσ nσ (9. c† ]+ = δσσ [dmσ .9.106) which together with (9.110) mσ nσ [dmσ . dnσ ]+ = 0 .109) The operators (9.108) obey the same anti-commutation properties (9. (9. (9.110) can be demonstrated similarly. nσ We demonstrate here the anti-commutation property (9.112) . The unitary property (9. of (9.103) allows one to express [c. (9.s r The unitarity property r ∗ Urm Urn = δmn yields immediately identity (??).107) Urn c† rσ .101) are occupied.104) One can. 9. behave exactly like the operators c† and crσ behave for states |ΨF (N ) .46).111).45. expand S trs = mn ∗ ˜ tmn Urm Usn (9. they are Fermion creation and annihilation operators.113) Urm Usn [crσ .101) of states |r .111) can be written. (9.46) as c† and crσ . The l.111) and. i.100) yields ˜ Ho = tmn d† dnσ mσ where d† nσ = r=1 S S (9.and two-particle operators hold in an analogous way for d† and dnσ .h.

. is N |Φo = j=1 d† α d† β |0 .116) which follows from (9. σ1 . a fermion in a spin state | 1 . − 1 is called a closed 2 2 2 2 . This equation poses the eigenvalue problem for the hermitian matrix (trs ). n = 1.115) ˜ and involves solely occupation number operators N nσ = d† dnσ . ) = j=1 nj .115) with eigenvalues 2N E(n1 .101) gives us the freedom to choose the matrix (9. . i. 2. S are T real. − 1 . σj ) = (nk . It is.e. σ2N . j j (9. 2 2 2 2 i. σ2 . . . 2. . (9. is obtained. . . transformation matrices Urn and energy values n . the associated ( properly normalized) eigenvectors Vn are the columns of (Urn ). S (9.. A non-degenerate orbital state which is occupied by two fermions.106) is equivalent to S trs Usn = s=1 n Urn ∀r.260 Diagonal Representation Many–Particle Systems The transformation (9.e. ˜ tmn = In this case Ho has the desired form S n δmn . r = 1. . 1 and |β = | 1 .. (9.. . i. Eigenstates of (9.114) together with (9.e.57) to the case of 2N particles.114) Ho = n=1 † n dnσ dnσ (9. We apply for this purpose (9. n2 . . . U2n . a simple matter to nσ state many–particle states in the new representation.115) can be stated immediately since any many–particle wave function in the occupation number representation is suitable. The condition (9. n2N .106) and the unitary property of Urn . The 2N fermion state 2N d† j σj |0 n j=1 where (nj . Before proceeding we like to address the issue how the new representation. σk ) for j = k (9. hence..119) In this state the N lowest orbital eigenstates of trs are occupied each by an electron with spin |α = | 1 .e. 1 as well as a fermion in a spin state | 1 .117) are eigenstates of (9. the so-called ground state.118) Ground State In case of an ordering m < n for m < n the state of lowest energy. i. USn ). Vn = (U1n . . The corresponding eigenvalues n .106) diagonal.

β ˆ are eigenstates of S 2 given in (9.n=1 † † if either |m) or |n) are closed shells.n=1 m.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. the states diﬀer in their spin character. β |β. α |β. ˆ In order to investigate the total spin of |Φo we apply to this state the total spin operator S 2 as given by (9.124) . Similarly.n=1 d† d† dmβ dnα mα nβ (9.. α = 2 |β. Excited States We want to construct now excited states for the system governed by the independent-particle Hamiltonian (9. ˆ one can conclude that |Φo is an eigenstate of S 2 with eigenvalue zero.123) ˆ which acts only on the N − 1 closed shells (except for N ) of the two states and has an eigenvalue ˆ N + 0 − (N − 1) = 1 in both cases.n=1 d† d† dmβ dnα mα nβ (9. (9. both with eigenvalues 2 ˆ S 2 |α.9. One can conclude immediately that the ˜ ˜ ˜ ˜ (N mα − N mβ )(N nα − N nβ ) does not give any contribution second term in (9.120). is a singlet state.n=1 cmα cnβ cmβ cnα does not give a contribution if m = n and either |m) or |n) are closed shells. Obviously. namely. |α. α N +1 − N . α |α.97).120) where the occupation number operators refer to the single-particle states |m). the third term − S m. In case of m = n it gives a contribution ‘1’ for each closed shell which cancels the contribution of the ﬁrst term.. 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. We will demonstrate now that this property endows the ground state with total spin zero. i.e. The action of the operator (9. The ground state has only closed shells.4: Independent-Particle Models 261 shell. β = 2 |α. the states with energy closest to the ground state are those in which a particle is promoted from the highest occupied state |N ) to the lowest unoccupied state |N + 1). ˆ S 2 ||β. However. i. The two states |α. Alltogether. β . namely.97).n=N (9. α .120) is particularly simple for closed shells. 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.n=1 m.e. There are four such states. of ∆E = and |β. 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.121) All four states have the same excitation energy.100). energy above the ground state.

|N + 3 . however. The system is described by the Hamiltonian 2N ˆ Ho = − t n=1 σ c† σ cn σ + c† σ cn+1 σ n n+1 . The action of Σ N +1 ˆ Σ3 |α.127) ˆ are triplet and singlet states. i.4. 2. .e. .129) The symmetry of the Hamiltonian suggests to choose a new representation deﬁned through the operators (r = 1. ˆ Exercise 9. hence. α and |β. . α ± |β. β are not eigenstates of S 2 .σ d† +1.σ dN σ N σ. β qualify as eigenstates. ˆ 3 on the two states produces . α as triplet states.100) outlined above we consider a system of 2N electrons which move in a set of atomic orbitals |r . the linear combinations 1/2(|α. β and |β. β =− m. One can. apropriate eigenstates of the operators S 2 and Example: 2N Independent Electrons on a Ring In order to illustrate the procedure to obtain eigenstates of one-particle Hamiltonians (9. . . α . . vanish. of course.125) ˆ Contributions to S 2 acting on states |N + 2 . of course. The same result holds for |β.n=N Many–Particle Systems N +1 d† d† dmβ dnα .. 2N which are located on a ring. do not need to be considered since they ˆ give vanishing contributions. etc. The system is assumed to posses a 2N –fold symmetry axis and interactions connect only orbital states |r → |r ± 1 . conclude that (9. We identify the states |2N + 1 = |1 to avoid cumbersome summation limits. |β. also eigenstates of Σ2 . mα nβ (9. one with eigenvalue zero (singlet) and one with eigenvalue ‘2’ (triplet).122) is correct. 2. for example. ˆ An eigenvalue ‘2’ of the spin operator S 2 identiﬁes the states |α. α are. r = 1. 2N ) drσ = √ 1 2N 2N e−irnπ/N cnσ n=1 (9. The two states |α.2: Construct four operators O ˆ O ˆ such that O ˆ S3 . for |α.σ = (9.n=N d† d† dmβ dnα d† +1α dN β |Φo = 0 mα nβ N (9.130) . c† +1 σ = c† σ 1 2N (9. ˆ The remaining states |α. β both with eigenvalues 1. β .126) which follows from the occurence of squares of fermion operators which. . m |Φo m m m γσ.128) The cyclic property of the system is expressed through c2N +1 σ = c1 σ .262 and ˆ Σ3 = − m. . . (9.

h. 2N (9.139) −2t cos r=1 σ rπ N d† drσ .132) an identity which can be derived employing the well-known expression for a ﬁnite geometric series m as = a s=1 1 − am . One can then express[. one obtains the new representation of Ho 2N ˆ Ho = −t r.138) ˜ trs = from which follows ˆ Ho = eirπ/N + e−irπ/N 2N δrs (9.9. eisπ/N 1 − e2N isπ/N 1 − eisπ/N (9.130.134) which.140) The Hamiltonian is now diagonal and the construction outlined above can be applied. n=1 (9.137) where 1 ˜ trs = e−irπ/N 2N Using (9. vanishes for integer s. of (9. 4N. . (9. 9.131) The (9. .136) ˆ and. (9. thereby. .132) gives for r = 0. indeed.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 .131) constitute a unitary transformation U deﬁned through Urn = unitarity property follows from 2N eirsπ/N = 0 r=1 for 0 < s < 2N .132) one can conclude 2N ei(s−r)nπ/N + eisπ/N n=1 1 2N 2N ei(r−s)nπ/N .133) Application to the l.135) c† nσ 1 = √ 2N e−irnπ/N d† rσ r=1 (9. rσ (9. 2N. . 1−a (9.s=1 σ ˜ rσ trs d† dsσ (9.f.s. .

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

145) which will play a pivotal role: it is this the state in which the fermions are assumed to be moving and relative to which the mean interactions acting on individual fermions is determined.t. t.s. . The states |m serve to deﬁne a reference state of the system ˜ N |Φo (U) = j=1 d† α d† β |0 . .σ c† c† rσ sσ cuσ ctσ + c† c† rσ sσ c† ctσ cuσ rσ c† cuσ rσ ctσ cuσ ctσ + c† c† cuσ rσ sσ ctσ + c† sσ − c† c† ctσ cuσ − c† rσ sσ sσ (9. S) ˜ S S |m = ˜ r=1 Urm |r . n. N .t. 2. naturally. s. 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. 2. 2. |r = m=1 ∗ Urm |m .9. m = 1. .s=1 σ r.s. S on which (9. We will adopt a second set of single-particle states {|m . . . .145). . 2. In our formulas below we will adopt strictly the convention that indices r. m = 1. s|ˆ|t. . . . ˜ (9.108). S is a unitary matrix.142) r|t|s c† csσ + rσ sσ 2 r.146) . . . the corresponding U = ( Urm ) forms an S × N matrix and.144) deﬁned as in (9.143) These S states are associated with the creation and annihilation operators S d† mσ = r=1 S Urm c† rσ ∗ Urm crσ r=1 dmσ = (9. u v r. .142) over this reference state to turn the contribution into an eﬀective one-particle operator.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. 2. is not unitary. j j (9. . as in in (9. u refer to the initial basis of single-particle states |r .142) is based. r = 1. . q. p. s|ˆ|t. u c† c† cuσ ctσ v (9. This is done as follows: ˆ Vmf (|Φo (U) ) = 1 2 S r.u=1 σ. If one restricts m = 1. U = ( Urm ) as deﬁned in (9. .u=1 σ. S} the elements of which will be labelled by indices ˜ m.σ and derive a single-particle operator which approximates this Hamiltonian. The states |m are connected with the states |r through (m = 1.143) with r.

t=1 σ.u ∗ r.t.s.n.σ . s|ˆ|t.266 Here ··· denotes the average O = Φo (U)| O |Φo (U) . u c† d† dnσ cuσ ˜ vn rσ mσ r.n.u=1 σ. n c† d† dnσ ˜ v ˜ rσ mσ r. u ˜ ˜v m.t.σ S ctσ + m. u Usm Uun .m. u c† ˜ vn sσ m.n=1 σ. m|ˆ|˜ .σ S d† dnσ cuσ mσ − r.m.s.σ S 1 − 2 where. by means of the rules (9. n = ˜ ˆ ˜ s.m.s. hence. 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 (9.148) r. Many–Particle Systems (9.n. n c† ctσ δσ σ ˜ v ˜ rσ m≤N r.150) r. m. m|ˆ|t.149) The remaining matrix elements appearing in (9. Since the averages aﬀect only two creationannihilation operators actually a mixed representation is most suitable ˆ Vmf (|Φo (U) ) 1 = 2 1 2 1 2 1 2 1 2 S m. for example.148) are obtained in a similar way by transforming only two of the four single-particle states into the new representation.σ S dmσ dnσ + r.146) accordingly. d† d† mσ nσ dmσ dnσ d† dnσ mσ = = = = and. m| v |t. s|ˆ|˜ .σ S d† d† mσ nσ cuσ ctσ + r. ˆ (9.u=1 σ.n=1 σ.t=1 σ. n|ˆ|t. s| v t.n. n c† c† v ˜ ˜ rσ sσ r.148) one obtains.u=1 σ.84).σ S d† dnσ mσ ctσ (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. m|ˆ|t. s|ˆ|m. n c† ˜ v ˜ sσ m. For the averages ··· in (9.

m...σ S r.. The one-particle operator (9. n c† ctσ .152) By means of expression (9.t=1 σ. u = s.142) by Hmf (U) deﬁned through S ˆ Hmf (U) = r.5: Self-Consistent Field Theory 1 2 1 2 1 2 S 267 m. At this point this state is completely arbitrary. m=1. n c† ctσ δσ σ ˜ v ˜ rσ − r. m|ˆ|˜ .σ r.153) rσ r. s ) ˆ ˆ m≤N ∗ Utm Uum (9...9. u c† cuσ ˜ vn rσ (9. The mean ﬁeld approximation. leaves open the question how the reference state should be chosen. r| v |u. t| v |s.2. as introduced here. We will address now a proper choice of the reference state. m|ˆ|˜ . t and renaming dummy summation indices one can demonstrate ˆ ˆ that term one and term two as well as term three and four are identical and one obtains S ˆ Vmf (|Φo (U) ) = 2 r.155) which is a function of the S × N –matrix U = ( Urm )r=1.154) ˆ The mean ﬁeld approximation replaces then the Hamiltonian (9. s|ˆ|t. ..u=1 σ S − m.s=1 σ ˆ r|t|s + r| vmf (|Φo (U) ) |s ˆ c† csσ rσ (9. t| v |u.u=1 σ. u c† cuσ δσσ ˜ vn sσ m≤N s.156) which deﬁnes the reference state |Φo (U) . u c† cuσ ˜ vn rσ m≤N r. s| v |t. s|ˆ|˜ .m.s σ where r| vmf (|Φo (U) ) |s = ˆ t.N (9. ˜ v ˜ sσ m≤N s.152) can then be written in the form ˆ Vmf (|Φo (U) ) = r| vmf (|Φo (U) ) |s c† csσ ˆ (9. Exploiting the symmetry r. m|ˆ|t.151) Carrying out the sum σ δσσ = 2 leads to a factor 2 for the ﬁrst two terms.u ( 2 r.2.S.149) for the matrix elements in the mixed representation one can state the right hand side explicitly in terms of Urm .σ S + − r.u=1 σ. u − r.t=1 σ (9.

S . 2. Let us state now the construction of the self-consistent ﬁeld ground state in more detail.e. . Step 2M.119. m = 1. SCF-Algorithm. We will argue below that the self-consistent ﬁeld ground state.: Determine Mean Field Hamiltonian In this step the mean ﬁeld Hamiltonian S ˆ (M ) Hmf = r. . i. The state thus determined is referred to as the self-consistent independent-particle ground state. (9..6 Self-Consistent Field Algorithm The Self-Consistent Field (SCF) approximation.4 and deﬁning this as the new reference state. . U(1) is the S ×N –matrix (1) of the ﬁrst N column vectors of Urm ..155). 2. steps (2) and (3) are being repeated M times until the procedure converges. the independent-particle nature stemming from the fact that the functional form of the ground state. . 9.160) . under conditions which often are realized. . . S. The procedure determines the reference state as the ground state of the independent-particle Hamiltonian (9.159) One deﬁnes the reference state |Φo (U(1) ) using the deﬁnition (9.145). 2. . i. m = 1. assuming in an intial step (1) a properly chosen reference state. (9. 2.s=1 σ S ˆ rσ r|t|s c† csσ (9. 2.155) following the method outlined in Section 9. the reference state resulting from step 2M + 1 (within numerical error) is equal to the reference state resulting from step 2M − 1. . SCF-Algorithm. .268 Many–Particle Systems 9. and obtaining in a step (3) the ground state of this one-particle Hamiltonian according to the construction in Section 9.145). Step 1: Choosing an Initial Reference State One deﬁnes the Hamiltonian ˆ (1) Hmf = r.e. M = 1. This procedure. however. is exact only for an independent-particle Hamiltonian. (9. .142).e. r = 1. i. . r = 1. .158) where the labels m are ordered to obey the condition (1) m < (1) n for m < n . determining in a step (2) the corresponding mean ﬁeld Hamiltonian (9. .157) and determines the associated diagonal representation deﬁned through S (1) ˆ r|t|s Usm = s=1 (1) (1) m Urm .s=1 σ ˆ r|hmf |s c† csσ rσ (M ) (9. S .4. often also referred to as the Hartree-Fock approximation. S . is the lowest energy independent-particle ground state one can construct. 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.s=1 σ S ˆ r|t|s + r| vmf |Φo (U(M ) ) ˆ |s c† csσ rσ = r. . . can achieve this goal only iteratively.

145).u=1 ( 2 r. 2.163) The result yields the reference state |Φo (U(M +1) ) using the deﬁnition (9.s=1 σ ˆ r|f |s c† csσ rσ (9. (c) Show that any spin-independent one-particle operator S ˆ F = r. .145). . u − r. σ p = δtr δsu δσσ .164) ˆ (a) Determine the expectation value of Prs for the state as deﬁned in (9. SCF-Algorithm: Continuation and Convergence Condition When step 2M + 1 is completed. S do not exceed a preset threshold. . ˆ t. σ|ˆrs |u. (9. S . 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. . s ) ˆ ˆ m=1 Utm ∗(M ) (M Uum ) (9.s=1 (9. . i. . . .166) ˆ (d) Deﬁne a similar two-particle operator Prstu . U(M +1) is the (M +1) S × N –matrix of the ﬁrst N column vectors of Urm . i.e. . s = 1. indicating that the state converged.. (9. ˆ Exercise 9.e.9. . . etc. 2. . S (9. (e) Consider the creation operators N † gqσ = m=1 Vmq d† mσ (9. m = 1. 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 .1: Let Prs be the one-particle operator with the generic operator prs = |r s|. t| v |s.e.167) . 2. the S × S–matrix S (M ) r|hmf |s N 269 ˆ = r|t|s + t. .161) is calculated. . M = 1. 2.165) can be written ˆ F = S ˆ ˆ r|f |s Prs r. m = 1.6: Self-Consistent Field Algorithm is evaluated. r = 1. SCF-Algorithm. Step 2M + 1. . t| v |u. . i. r = 1.. S.162) is solved adopting an ordering of the labels m which obeys the condition (M +1) m < (M +1) n for m < n .6. 2 . 2. step 2M + 2 is carried out. . ˆ (b) How can the expectation values of the operator Prr be interpreted.

. For this representation holds N m| t |˜ + ˜ ˆn m 2 mm | v |˜ m ˜ ˜ ˆ n˜ − mm | v |m n ˜˜ ˆ ˜ ˜ = (SCF ) δmn m (9. S denotes the initial single-particle states and Urm is the unitary S × S– transformation matrix obtained in (9.270 Many–Particle Systems which are connected with d† through a unitary N × N –matrix (Vqm ).169) 9.170) (SCF ) Here |r .7 Properties of the SCF Ground State We want to investigate now the properties of the SCF ground state. .171) where the convention (9. .168) ˆ ˆ has the same expectation values for Prs and Prstu as the reference state (9.σ + 2v ( c† c1α c† c1β + c† c2α c† c2β ) + v 1α 2α 1β 2β + c† c2σ c† 2σ 1σ c1σ ) .173) .172) is obeyed.e. Note that this is not an mσ S × S–matrix! Show that the reference state deﬁned through N † † gqα gqβ |0 q=1 (9. (SCF ) m < (SCF ) n for m < n (9.168) through a physical observation? Exercise 9. .6.145) and (9. We begin by summarizing the result of the SCF algorithm.9. Equation (9.145). 2. (f) Can one distinguish the states (9. (9. i.171) implies that for the mean ﬁeld Hamiltonian holds S ˆ Hmf (U(SCF ) ) = m=1 σ (SCF ) † dmσ dmσ m .163). We denote the representation which results from the SCF algorithm after its convergence as follows: S |m = ˜ r=1 (SCF Urm ) |r .162. (9. r = 1.2: Determine the SCF ground state and its energy expectation value for a system of two particles described through the Hamiltonian (S = 2) H = σ ( c† c1σ + c† c2σ ) − t ( c† c2σ + c† c1σ ) 1σ 2σ 1σ 2σ ( c† c1σ c† c2σ 1σ 2σ σ. (9.163).

one can also determine the expectation value of SCF for the Hamiltonian (9. Within the mean ﬁeld approximation as described by (9. ˆ SCF H SCF .e..1: Reformulate (9. the mean ﬁeld potential ˆ Vmf (|Φo (U) ) deﬁned in (9.174) The ﬁrst property we consider is the total spin character of SCF .162) is non-linear in Urm . s|ˆ|t.7: Properties of the SCF Ground State 271 i. once as a member of the ground state.177).7. We like to determine next the energy expectation value of SCF . (SCF ) ˆ Exercise 9. i. (9.4 that SCF is a total singlet state. the mean-ﬁeld Hamiltonian is diagonal in the SCF representation.e. Since the system under consideration has only 2N particles altogether.142).173) holds N ˆ SCF Hmf (U(SCF ) ) SCF = 2 r=1 (SCF ) m . and once as a probe particle being subject to the mean pair interaction.175) However. Exercise 9.177) 2 mm | v |mm ˜˜ ˆ ˜˜ − mm | v |m m ˜˜ ˆ ˜ ˜ .171) in terms of matrix elements r|t|s . exploiting the spinindependence of (9. Employing expressions (9. Since this state is composed only of closed shells one can conclude following Section 9. v (SCF ) Show that the resulting eigenvalue problem of the type (9. 2 m=1 (SCF ) m − m. the SCF ground state is N SCF = m=1 d† d† |0 .142).m =1 Comparision with (9. one obtains ˆ SCF H SCF N = N (9.142). r.. .9. 2 m=1 m|t|m + ˜ ˆ˜ m.176) 2 mm | v |mm ˜˜ ˆ ˜˜ − mm | v |m m ˜˜ ˆ ˜ ˜ .171) yields ˆ SCF H SCF N = N (9.2: Derive (9. Finally. mα mβ (9.146) counts a particle twice. u and Urm .m =1 The second term originates from the fact that in the mean ﬁeld approximation one assumes that each of the fermions is subject to an average pair interaction involving a 2N -particle reference state. This over-counting leads to the correction term in (9.93) for the expectation values (diagonal elements) of the one-particle and two-particle part of the Hamiltonian (9.84) and (9.7.177).

in the usual two–particle operator form.σ crσ + crσ cr+1. In the present Section we want to adapt the algorithm to systems containing a macroscopically large number of fermions. i. for example.180) V = rσ sσ 2 r. .σ rσ The potential energy term.. except in case of one-dimensional systems. . as stated through Hamiltonian (9.s. approximation schemes like the self-consistent ﬁeld approximation play an important role. . in the two-dimensional copper oxide lattices of high temperature superconductors. v (9.178) v contributes only in case that two electrons occupy the same lattice site. 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. real number.181) The Hubbard model. of (9. the second term on ther.h. u c† c† cuσ ctσ . .e. We consider as an example the Hubbard model of an inﬁnite linear lattice and apply the model to describe magnetic instabilities in metals. s|ˆ|t.e. v (9.192) derived below is ﬁlled half. (9.179) r+1. t = v δru δst δrs . i. In spite of the simplicity of its Hamiltonian the Hubbard model. referring to the fact that in this case the band of single particle energies (9. We assume that there are altogether S = 2N0 lattice sites.182) n can assume values 0 ≤ n ≤ 2. Due to the lack of an exact solution in dimensions higher than one. t describes the coupling of state |r to states |r ± 1 at the two neighboring lattice sites.178) Here the index r. β for ‘up’ and ‘down’ rσ spins. According to (9.178).4–9.σ + v 2 c† c† crσ crσ . is identical to that of the independent–particle Hamiltonian H0 (9.178). The one-dimensional Hubbard model is described by the Hamiltonian ˆ H = −t rσ † c† r+1.128).178).8 Mean Field Theory for Macroscopic Systems The SCF algorithm has been developed in Sections 9. The operator c† (crσ ) creates (destroys) an electron with spin σ (σ = α. † ˆ Ho = −t c† . reads 1 ˆ r. v of the Hubbard Hamiltonian (9. respectively) in state |r at lattice site r.. of (9.7 for ﬁnite systems. The ﬁrst term on the r. N0 which are populated by 2N electrons (we choose an even number of electrons for convenience).h.σ crσ + crσ cr+1. Both N0 and N are macroscopically large numbers. labeled r = −N0 + 1. r ∈ Z. The case n = 1 is termed the half ﬁlled band case. cannot be solved exactly. s|ˆ|u.s.e.s.272 Many–Particle Systems 9. S N0 (9. i. . describes the sites of a linear lattice.t. t is assumed to be a positive.. v represents the ‘on-site’ Coulomb repulsion.178) and through the so-called ﬁlling factor n n= 2N N = . rσ rσ r σσ (9. is characterized through the parameters t.u σσ where r.

185) into (9. SCF Ground State for the Hubbard Model We seek to determine the ground state of the Hubbard model stated by (9.185) ˆ According to (9. in fact. rσ (9.e. 9.186) where m = −2t cos mπ .9.178) within the framework of the self-consistent ﬁeld theory. i.154)..140).144). . Accordingly. we assume that the SCF ground state is given by (9. by N −1 ||SCF = m=0 d† d† |0 . 2N0 (9.6.181. For this purpose we apply the SCF theory presented in Section 9.188) .σ m=0 ∗ Urm Urm c† crσ .179).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 . This task has ˆ been solved already on page 262. mα mβ (9.155. is Urm = √ 1 exp (irmπ/N0 ) . The unitary transformation which diagonalizes H0 . rσ rσ (9.174). boundary eﬀects are assumed to be negligible. one obtains for the mean ﬁeld Hamiltonian N −1 ˆ ˆ Hmf (U) = H0 + v r.178) can be determined by applying (9.183) where the operators d† are related to the original creation operators c† through the unitary mσ rσ ˆ transformation U deﬁned in (9.187) The energy levels are labeled such that 0 < ±1 < .184) and using (9. The self-consistent ﬁeld Hamiltonian Hmf which corresponds to the Hubbard Hamiltonian (9. we identify |r + N0 = |r and later let N0 go to inﬁnity. the operator H0 can be re-written in the diagonal form N0 ˆ H0 = m=−N0 +1 σ m d† dmσ mσ (9. We can now embark on step 2 of the SCF algorithm. Inserting (9.184) ˆ We are now ready to apply step 1 of the SCF algorithm and diagonalize term H0 . 9. N0 (9. a freedom which we employ to adopt so-called periodic boundary conditions like those for the example “2N independent electrons on a ring” on page 262.8: Macroscopic Systems 273 Since we are dealing presently with a macroscopic system. Accordingly. . Using (9. < ±(N0 −1) < N0 holds as can be readily veriﬁed.

Usually.274 Many–Particle Systems Expressing c† .177) and mm |ˆ|mm = mm |ˆ|m m = ˜˜ v ˜˜ ˜˜ v ˜ ˜ which holds in the present case. N0 ˆ (2) ˆ Hmf = Hmf = m=−N0 +1 σ m + vn † dmσ dmσ 2 (9. for km = mπ/N0 holds km ∈ −π + N0 . . even for one and the same Hamiltonian.1. crσ through d† . the self–consistency condition is exactly met and the SCF algorithm converged after two steps.190).189) and. can be obtained. Indeed. One can see that −2t ≤ (k) ≤ 2t holds. The energy m can be expressed as a function of a continuous variable k ∈] − π. −π ≤k ≤π .187). 4 So far we have not exploited the fact that N0 and N are macroscopically large quantities. (9. π and. but may diﬀer quantitatively. i. . −π + N0 .e. Apparently. 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. in the limit N0 → ∞.109) and using (9.185) one can prove rσ mσ N0 c† crσ = rσ rσ m=−N0 +1 σ d† dmσ . The energy Emf of this state can be calculated readily by using (9. there exist several ways of formulating mean ﬁeld approximations. the ground state of the Hubbard model in the self-consistent ﬁeld approximation can be determined exactly. In other words. according to the Pauli principle. . The bottom of the energy band corresponds to (0) = −2t and the width of the energy band is 4t. as given in (9. π]. (9.187) form a quasi1 2 continuous energy band.182) of the ﬁlling factor n [c. 9. the state with the lowest possible energy. through the so-called dispersion law (k) = −2t cos k .191) where m is given in (9.1).. the electrons in the present description behave like independent particles with energies ¯m = m + vn . π]. . mσ (9. holds k ≡ km ∈] − π.182)]. dnσ according to (9.183) is not the only candidate for the mean ﬁeld ground state of the Hubbard model. Indeed. hence. the results of diﬀerent formulations agree qualitatively. When N0 becomes macroscopically large the single particle discrete energy levels (9.e. is already diagonal.178). ˆ The Hamiltonian Hmf .192) This dispersion law is presented in Fig. namely. i.f. denoted by F (see Fig. To demonstrate this we consider an alternative formulation of the mean ﬁeld approximation for Hamiltonian (9. 9. The ground state of the system. called the Fermi energy. Alternative Description of the Mean Field Approximation The state (9.190) We have used in the latter expression the deﬁnition (9. by ﬁlling up this energy band with electrons from the bottom of the band (which corresponds to k = 0) to a maximum energy..

F = (±kF ) denotes the Fermi energy. Nrσ = SCF ||Nrσ ||SCF . (9.178) and performing the summation over the spin indices results in ˆ ˆ H = H0 + v r ˆ ˆ Nrα Nrβ ..193) into (9.. the operator Nrσ can be written ˆ Nrσ = Nrσ + δ (Nrσ ) .194) ˆ For a given reference state.46) of the fermionic creation and rσ annihilation operators yield c† c† crσ crσ = −c† c† crσ crσ = rσ rσ rσ rσ ˆ ˆ ˆ2 c† crσ c† crσ − c† crσ δσσ = Nrσ Nrσ − Nrσ δσσ .8: Macroscopic Systems 275 Figure 9. The anti-commutation properties (9. we have employed (9.9. where Nrσ is the mean occupation number of the one particle state |rσ . 9.196) (9. the deviation of Nrσ from its mean value Nrσ for the corresponding reference state. The mean ﬁeld approximation neglects the .1: The dispersion law (k) for independent electrons on an inﬁnite one dimensional lattice.193) where. let us express the interaction term in (9.178) in terms of the occupation number operaˆ tors Nrσ = c† crσ . in the last term on the right hand side.183). say the one given in (9.e.195) ˆ ˆ ˆ and where δ (Nrσ ) = Nrσ − Nrσ describes the ﬂuctuations of Nrσ .56). (9.e. rσ rσ rσ (9. First.45. Inserting (9. i. i.

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. Note that the procedure employed in (9.198) Let us assume that the ground state of the Hubbard model. but the mean number of particles with spin α can be diﬀerent from the mean number of particles with spin β. in the mean ﬁeld approximation.e. 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 . .198) as the one obtained by using the mean ﬁeld potential (9. i.183). separating the occupation number operator into a mean value plus ﬂuctuation and neglecting the ﬂuctuations in 2nd order. non-vanishing magnetization.191) for the ground state energy. Therefore..146).197) and the corresponding mean ﬁeld Hamiltonian becomes ˆ ˆ Hmf = H0 + v r ˆ ˆ Nrα Nrβ + Nrβ Nrα − v r Nrα Nrβ .195). however. is given by (9.199) which is identical to expression (9.276 Many–Particle Systems ﬂuctuations to second order and the mean ﬁeld Hamiltonian is obtained from (9. Spin-Polarized Mean Field Ground State ˆ We want to consider now a ground state for Hmf deﬁned through Nrσ = nσ = const (9. Because of Nrσ = c† crσ rσ N −1 = m. 2N0 2 holds ˆ Hmf = = mσ mv ˆ H0 + 2 vn2 2 rσ nv † vn dmσ dmσ − N . consequently. In case nα = nβ the following construction will lead to a ground state which coincides with the non-magnetic ground state (9. that nα and nβ assume diﬀerent values.194) by dropping all the terms which contain the ﬂuctuations to second order.183) .m ∗ Urm Urm d† dm σ mσ = m=0 ∗ Urm Urm = n N = .200) allowing. For such state the mean occupation number nrσ is uniform at all lattice sites. the ground state of the system can have a non-zero local spin and. yields essentially the same mean ﬁeld Hamiltonian (9. m+ 2 2 c† crσ − 2N0 rσ (9. (9.

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

unit energy and a given spin orientation of the particle.212) where xi are the simple roots of the function f (x).209) where ρ( ) = −π dk δ ( − (k)) . namely k1.210).211) and using π dkδ(k ± (π − arccos( /2t))) = θ(2t − | |) . and noting |f (k)| = |2t sin k| = 2t one obtains δ( − (k)) = 1 − cos2 k = (2t)2 − 2 . gives the the number of available one particle states per site. 2π (9. Since the equation f (k) ≡ 2t cos k + = 0 has two simple roots.2 = ±(π − arccos( /2t)). Inserting this last result into (9. |f (xi )| (9.2: (a) Relative position of the two energy sub–bands α (k) and β (k) for ∆ = nα − nβ > 0. −π . In our case (k) is given by (9.192) and. 2π (9. (b) Graphical deﬁnition of the limiting energies α and β . If the function f depends explicitly only on (k) one can state π −π dk f ( (k)) = 2π π ∞ ρ( )d f ( ) −∞ (9. therefore. which is usually called the density of states. For a given dispersion law (k) the density of states ρ( ) can be calculated by employing equation (9.210) Here δ(x) is the Dirac–delta function and ρ( ).278 Many–Particle Systems (a) (b) Figure 9. The shaded areas represent the ﬁlled portions of these bands by electrons.211) In order to calculate the integral we recall the following property of the Dirac–delta function δ [f (x)] = i δ(x − xi ) . δ(k + π − arccos( /2t)) + δ(k − π + arccos( /2t)) (2t)2 − 2 . holds π ρ( ) = −π dk δ( + 2t cos k) .

π α. 2 (9.201).218) read ˜ ∂ Emf = ∂ α ˜ ∂ Emf = ∂ β and α + vnβ − µ = 0 . ∂ α ˜ ∂ Emf =0. ∂ β ˜ ∂ Emf =0.. (9.219) (9.213) can simply be replaced by 1. Fig.. We can employ the results obtained to express the ground state energy density (9. F = α + vnβ = β + vnα . (9. 9.219. namely.f. of t. if we assume that is restricted to the interval ] − 2t. One obtains ˜ Emf = α β ρ( )d + −2t −2t ρ( )d + vnα nβ + µ(n − nα − nβ ) .204) through an integral expression.214) are given by the sum (9. in principle.215) nα and nβ in (9. to determine the unknown quantities α .8: Macroscopic Systems 279 where θ(x) is the step function.e.. one arrives at the following expression of the density of states ρ( ) = θ(2t − | |) π (2t)2 − 2 . (9. outside the energy band (k). β (9.216) allow one. 2t[ then the θ function in (9.206)] depends only on continuous quantities.220) β ˜ ∂ Emf = n − nα − nβ = 0 .213) The presence of the θ function in the above formula guaranties that the density of states vanishes for | | > 2t. From (9. + vnα − µ = 0 . 9.222) . on conditions for a ground state of minimum energy are ˜ ∂ Emf =0.221) ∂µ Equations (9.220) µ can be readily obtained µ= 1 ( 2 α + β) + vn .219–9.e.e.221) and (9.213) and carrying out the resulting integral yields nσ = 1 arcsin( σ /2t) . µ.218) The derivatives can be readily determined and the conditions (9. nα and nβ as a function of the parameters of the Hubbard model. the necessary ˜ Since Emf [c. i. Replacing again the sum by an integral over energy one can write σ nσ = −2t ρ( )d (9.216) Using (9. i. θ(x) = 1 if x > 0 and θ(x) = 0 if x < 0. ∂µ (9. namely.f.9. (9. (9. β .2]. (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. i. v and n.217) and µ.

3. condition (9. from (9. We assume in the following discussion that the implicit ∆ dependence of β in f (∆) can be neglected.. The slope of f (∆) at the origin depends on the Hubbard model parameter v.228)] vc ρ ( β ) = 1 .226) ∆ = f (∆) . First. (9. (9.f. non trivial. (9.183) without magnetization. so called ferromagnetic solution. For v above a critical value vc ..227). Accordingly.e.226) follows that f (∆) is a monotonically increasing function of ∆. (9.226) one can obtain ∆ by solving (cf. (9. Indeed. One can express ∆ as a function of α and β in two diﬀerent ways.224) v∆ ∆= β ρ( )d = 0 ρ( β + )d . One can determine vc by equating the slope of f (∆) at the origin to 1. i. .280 Many–Particle Systems Comparision of (9.229) has a wider range of validity than one can infer from the present analysis of the Hubbard model. i. through the condition [c. (9.227) Since f (0) = 0 one can infer that ∆ = 0 is always a solution (the so called paramagnetic solution) of condition (9.223) ∆ = 0 corresponds to the ground state (9. the slope of f (∆) is positive for all ∆ ≥ 0.223 one obtains ∆= and second. (9. from (9.f.e.3].227) can be solved graphically.217) and (9.223) follows α α − v β . the Hubbard model has a ground state with ∆ = 0. and only if. if.219–9. the slope of f (∆) at the origin is larger than 1 [c. One still needs to determine α and β .228) Since the total number of states per site is ﬁnite.227) has only the trivial solution ∆ = 0. If the slope of f (∆) at the origin is less than 1. The Stoner criterion (9. Equation (9.222) and (9. a magnetic ground state.215) reveals that µ is indeed equal to the Fermi energy F (see also Figure 9. Figure 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. For this purpose one plots f (∆) versus ∆ as is shown schematically in Figure 9.2b).220) and (9.227) will have a second. From the deﬁnition (9. as reﬂected by the expression d[f (∆)] = vρ( d∆ β + ∆) = π (2t)2 v − (v∆ + 2 β) >0.225) Deﬁning the function v∆ f (∆) = 0 ρ( β + )d . condition (9. (9. For this purpose we consider the magnetization of the mean ﬁeld ground state ∆ ≡ nα − nβ . (9. f (∆) converges to a maximum value as ∆ → ∆max = n.

2 2 2 (9. Since ∆ vanishes near vc we set x = y + δ where δ is a small quantity. n.231) follows y= δ π π (n − 1) − ≈ (n − 1) . is presented in Figure 9. β x+y x−y = = π(n − 1) 2t π (sin x − sin y) = π∆ .233) yield the desired expression vc π = π cos (n − 1) .219. (9. hence. (9.232) together with (9.220. the plot of vc vs.205) the following set of equations determining x.233) Employing the approximation sin(y + δ) − sin y ≈ δ cos y for small δ. The corresponding 2t analytical expression provides the phase diagram of the ground state: when v approaches vc from above.e. β = 2t sin y .9. From (9. Accordingly. n).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.231) (9.3: Graphical solution of the mean ﬁeld equation ∆ = f (∆).230) one obtains from (9.8: Macroscopic Systems 281 Figure 9. (9. v (9. 2t As already mentioned. t. 9. the ground state is characterized by the quantities (v.232) states that x − y is proportional to ∆. 9.. y and. any . By parameterizing α and β α = 2t sin x .4. α. the magnetization ∆ of the ground state vanishes and remains zero for values v < vc .234) 2t 2 The resulting phase diagram of the ground state. Equation (9. i. We like to determine ﬁnally the dependence of vc on the ﬁlling factor n.

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

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

284 Many–Particle Systems .

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

286 Relativistic Quantum Mechanics in Sect.1). i. x3 ) = r describes the space coordinates. namely the Klein–Gordon (Sect. This will provide us with a general set of Lorentz–invariant equations which for various particles take the place of the Schr¨dinger equation.3) where x0 = ct describes the time coordinate and (x1 . 0 1 2 3 (10.e. 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.5) √ One can interprete the quantity −s2 as a distance in a 4–dimensional Euclidean space if one chooses the time component purely imaginary.1 Natural Representation of the Lorentz Group In this Section we consider the natural representation of the Lorentz group L. Rather than following this avenue we will introduce Lorentz transformations within a setting which does not require real and imaginary coordinates.1). x3 ) (10. the group of Lorentz transformations (10. 5 to the groups SO(3) and SU(2) of rotation transformations of space coordinates and of spin. Note that the components of xµ all have the same dimension. 10. Minkowski Space Historically. The elements L ∈ L act on 4–dimensional vectors of position– and time–coordinates. wave functions ψ(r. we will provide a more basic deﬁnition of the transformations. x2 .5) and the Dirac (Sect. This choice implies dim(time) = dim(length). We will ﬁnd that this deﬁnition will lead us back to the transformation law (10.7) equation. t) and vectors with functions ψ(r. This space is called the Minkowski space. i..e. for the transformed space-time–cordinates x µ = (x 0 . but in a setting of representation theory methods as applied in Secti.1). The reason for this choice is that the transformations (10. henceforth. 10. x1 . 10. The Group of Lorentz Transformations L = O(3.1) leave the quantity s2 = (x0 )2 − (x1 )2 − (x2 )2 − (x3 )2 (10. x 2 . assume new units for time such that the velocity of light c becomes c = 1. however. We will.4) invariant. i. namely that of length.7 a heuristic derivation of the two most widely used and best known Lorentz–invariant ﬁeld equations. Denoting by xµ the . 10.1) The Lorentz transformations L describe the relationship between space-time coordinates xµ of two reference frames which move relative to each other with uniform ﬁxed velocity v and which might be reoriented relative to each other by a rotation around a common origin. x 3 ) holds (x0 )2 − (x1 )2 − (x2 )2 − (x3 )2 = (x )2 − (x )2 − (x )2 − (x )2 .1 will be extended in Sect.. 10. We will denote these vectors as follows def xµ = (x0 .e. Before introduco ing these general Lorentz–invariant ﬁeld equations we will provide in Sects. x 1 . 10. In such a space Lorentz transformations correspond to 4-dimensional rotations. Rather than starting from (10.5.4 to cover ﬁelds. x2 . t) as components. 10.

6) x µ = Lµ ν xν . Aµν ..9) 1 0 0 0 −1 0 0 ( gµν ) = 0 0 −1 0 = g . This will be explained further below. ν = 0.. R). (10. 1. 3 3 3 yµ xµ = new yµ xµ . The Lorentz transformations form the subgroup of all matrices which leave the expression (10.9) and (10. R). Aµ ν xν = µ=0 new old Aµ ν xν . the Lorentz transformations constitute a linear transformation which we denote by 3 x µ = ν=0 Lµ ν xν .1: Natural Representation of the Lorentz Group 287 coordinates in one reference frame and by x µ the coordinates in the other reference frame. First. xµ . i. Aµν . Since this holds for any xµ it must be true Lµ ρ gµν Lν σ = gρσ . the latter notation will be used for diﬀerent quantities further below. (10. The upper index closer to ‘L’ denotes the ﬁrst index of the matrix and the lower index ν further away from ‘L’ denotes the second index. however. Aµ ν .6) Here Lµ ν are the elements of a 4 × 4–matrix representing the Lorentz transformation.6) allows us to introduce the so-called summation conventon: any time the same index appears in an upper and a lower position. ν=0 (10. The second reason is that upper and lower positions allow us to accomodate the expression (10.10) Combining condition (10. This condition can be written xµ gµν xν = x gµν x where µ ν (10.5) into scalar products. [ A more conventional notation would be Lµν . Aµ ν B ν ρ = ν=0 Aµ ν B ν ρ .] The following possibilities exist for the positioning of the indices µ.5) invariant. The Lorentz transformations are non-singular 4 × 4–matrices with real coeﬃcients. i.7) The reason for the notation is two-fold.e.6) yields Lµ ρ gµν Lν σ xρ xσ = gρσ xρ xσ . L ∈ GL(4. 4 × 4 tensor: Aµ ν . the latter set constituting a group.8) new old old The summation convention allows us to write (10.e.12) (10. summation over that index is assumed without explicitly noting it. the notation in (10. 3: 4-vector: xµ . the .10. (10. 0 0 0 −1 (10.11) This condition speciﬁes the key property of Lorentz transformations. The subset of GL(4. We will exploit this property below to determine the general form of the Lorentz transformations. 2.

1).. i. L−1 ∈ O(3.16..18) (10. From this one can obtain using g2 = 1 1 (gLT g)L = 1 and. the associated inverse obeys (10. L2 ∈ O(3. We want to show now L = O(3.15) one can derive LT gLgLT = LT and multiplying this from the left by by g(LT )−1 yields L g LT = g Using this result to simplify the r.14) L0 0 −L1 0 −L2 0 −L3 0 −L0 1 L1 1 L2 1 L3 1 .13) L = ( Lµ ν ) = 2 0 L 0 L2 1 L2 2 L3 3 .288 Relativistic Quantum Mechanics elements of which satisfy this condition. .1).10) of g one can rewrite the invariance property (10.1) ⊂ GL(4. hence.1) is. LT gL = g } . (10. 10. 3 2 1 2 1 2 (10. of O(3. R) is a group.1).22) (10.14) holds for the inverse of L.20) results in the desired property (L−1 )T g L−1 = g . (10.e. This stipulates that O(3. L ∈ GL(4.14). One can also show that if L ∈ O(3..14) from the right by gLT and using (10.1) since it satisﬁes (10. To demonstrate the group property of O(3.e. R).12) LT gL = g .14). property (10..21) i. of (10. In fact.h. is called O(3.20) Multiplying (10.17) we note ﬁrst that the identity matrix 1 is an element of O(3. employing expressions (10. 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. For L3 = L1 L2 holds LT g L3 = LT LT g L1 L2 = LT (LT gL1 ) L2 = LT g L2 = g .16) The corresponding expression for (LT )−1 is obviously (LT )−1 = (L−1 )T = g L g . 1) = { L. L3 0 L3 1 L3 2 L3 3 Using the deﬁnition (10. in fact. i. = g LT g = 0 1 2 −L 2 L 2 L 2 L3 2 −L0 3 L1 3 L2 3 L3 3 (10. L3 ∈ O(3. (10. This set is identical with the set of all Lorentz transformations L.e.19) i.e. a group.15) (10.1).1). We consider 1 then L1 .s.1). the inverse of L 1 L−1 (10.17) one obtains (L−1 )T g L−1 = gLgggLT g = gLgLT g .

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 ∈ O(3. det L = −1. From this we can conclude L0 0 ≥ 1 or L0 0 ≤ −1 .12). Lorentz transformations in L↑ . A second class property follows from (10. (10. (10.32) (10. = { L. A statement on this value can be made on account of property (10. Considering in (10. For this purpose we consider ﬁrst the value of det L. entirely suitable to investigate ﬁrst only 1. (10. 1).27) (10. L ∈ O(3.34) j=1 j=1 . + − gL↑ + = L↑ g . det L = 1. L0 0 ≥ 1} . L0 0 ≤ −1} .28) (10. = { L.10.14). + We start our investigation by demonstrating that L↑ forms a group. hence. L↓ are disjunct sets deﬁned as follows + + − − L↑ + L↓ + L↑ − L↓ − = { L.12) the case ρ = 0. = { L. 1).14) which we employ in the formulation (10.25) can be stated as follows: The set of all Lorentz transformations L is given as the union L = L↑ ∪ L↓ ∪ L↑ ∪ L↓ + + − − where L↑ . One can also 1 verify that one can write L↑ − L↓ + L↓ − = gL↑ + = = −L↑ .1). 1). Properties (10. + except for the trivial factors g and −1 It is. M1 = a M2 }.30) ≤ −1} . 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↑ contains 1 1.29) (10. det L = −1. + = − L↑ g + (10.26) (10. L↓ .14).e. M2 ∈ M.31) (10. ∃M2 . det L = 1. L0 0 (10. L ∈ O(3.1: Natural Representation of the Lorentz Group Classiﬁcation of Lorentz Transformations 289 We like to classify now the elements of L = O(3. 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 . B ∈ L+ holds C = AB ∈ L+ . The above shows that the set of proper Lorentz transformations L↑ allows one to generate all Lorentz transformations. σ = 0 yields L0 0 2 − L1 0 2 − L2 0 2 − L3 0 2 = 1. L ∈ O(3.23) and (10. L↑ . (10. + + ↑ ↑ We can also demonstrate that for A. Using det AB = det A det B and det AT = det A yields (det L)2 = 1 or det L = ±1 .33) where we used the deﬁnition aM = {M1 .24) or since (L1 0 )2 + (L2 0 )2 + (L3 0 )2 ≥ 0 it holds (L0 0 )2 ≥ 1.25) i. Obviously.

(10. 1 ∈ L↑ . δ00 = 1 and It should be noted that according to our present deﬁnition holds δµν = gµρ δ δ11 = δ22 = δ33 = −1. (10.34) provides then the estimate conclude from (10. ρ ν (10. + Accordingly.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 . therefore. + i.1). from which one can conclude L−1 ∈ L↑ .1). This relationship shows (L−1 )0 0 = L0 0 .1) ascertains CT gC = g it must hold C 0 0 ≥ 1. obviously j=1 A0 0 B 0 0 ≥ 1. .e. (10. 1 We will then employ the Lie group properties to generate all transformations in L↑ .. Property (10. Inﬁnitesimal Lorentz transformations The transformations in L↑ have the property that they are continously connected to the identity 1 1. if we enforce (10. | 3 A0 j B j 0 | < A0 0 B 0 0 .14) implies 1 + T g (1 + ) = g 1 1 (10.290 Relativistic Quantum Mechanics From (10.13).16). 1 0 0 = ≥ 1 and.. + transformations in a small neighborhood of 1 which we parametrize by inﬁnitesimal parameters. orthochronous Lorentz transformations. To order O( 2 ) holds g + g = 0. since the group property of O(3.14) actually L0 0 ≥ 1 and det L = 1 must hold. hence. obviously. one can j=1 j=1 3 0 )2 = (A0 )2 − 1. we consider transformations Lµ ν = δ µ ν + µ ν . Using the above expression for C 0 0 one can state C 0 0 > 0. accordingly.12) follows (B 0 0 )2 − 3 (B j 0 )2 = 1 or 3 (B j 0 )2 = (B 0 0 )2 − 1. The next group property of L↑ to be demonstrated is the existence of the inverse. L↑ is called the subgroup of proper. In fact.38) For these transformations. Since A0 0 ≥ 1 and B 0 0 ≥ 1. µ ν small . Similarly. 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 (10.36) It holds.37) (10. For the inverse of + ↑ any L ∈ L+ holds (10. these transformations can be parametrized such that a continuous variation of the parameters connects any element of L↑ with 1 This property will be exploited now in that we consider ﬁrst 1. holds L0 0 > 0 and the value of the determinant is close to unity.e. i. Since the associative property holds for matrix multipli1 1 + ↑ cation we have veriﬁed that L+ is indeed a subgroup of SO(3. In the following we + will consider solely this subgroup of SO(3.39) where we have employed the matrix form T 1 deﬁned as in (10.35) One can conclude.40) and.

j. k = 1.42) This relationship implies µ 0 j µ j = 0 = j 0 . w1 . J2 = 0 0 0 −1 0 0 0 0 0 0 0 0 0 0 0 −1 0 1 . = − k m Km (10.15) one can conclude T 291 = −g g (10. (ϑ1 . K2 = . ϑ2 .46) 0 0 0 0 . J = 0 3 0 1 0 0 0 0 0 0 0 0 0 0 0 (10. 2.50) .48) (10. 3 . 3 (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 . The generators are explicitly 0 0 0 0 0 0 0 0 J1 = 0 0 0 −1 0 0 1 0 (10. K ] The operators also obey J · K = J1 J1 + J2 J2 + J3 J3 = 0 = = k m Jm k m Jm 0 −1 0 0 0 0 0 0 (10.49) . ϑ3 .10. w2 . K ] [ Jk .47) 0 −1 0 0 0 0 −1 0 0 −1 0 0 0 0 0 0 0 0 K1 = .45) (10.44) This result allows us now to deﬁne six generators for the Lorentz transformations(k = 1. other ﬁve parameters zero) Kk = (wk = 1. 2. J ] [ Kk . w3 ) = −w2 ϑ3 0 −ϑ1 −w3 −ϑ2 ϑ1 0 (10. 3) Jk = (ϑk = 1. k j j = 1.43) k = − Inspection shows that the matrix has 6 independent elements and can be written 0 −w1 −w2 −w3 −w1 0 −ϑ3 ϑ2 . (10.1: Natural Representation of the Lorentz Group Using (10. other ﬁve parameters zero) . 2.41) which reads explicitly 0 0 0 0 0 1 2 3 1 1 1 1 0 1 2 3 2 2 2 2 0 1 2 3 3 3 3 3 0 1 2 3 = − 0 1 2 3 0 0 0 0 0 1 1 1 1 0 2 2 2 2 0 3 3 3 3 − − − 1 2 3 − − − 1 2 3 − − − 1 2 3 .

52) where the 3 × 3–matrices Lk are the generators of SO(3) deﬁned in Chapter 5. 2. that the transformations (10. + The commutation relationships imply that the algebra of the generators Jk . 2. 2π[ . 3 (10. 3 is closed.49) deﬁne the Lie algebra associated with the Lie group L↑ . Relativistic Quantum Mechanics The commutation relationships (10. For the parameters ϑk of the Lorentz transformations holds obviously ϑk ∈ [0.49.54) which.292 as can be readily veriﬁed.51) One can readily show. k = 1. We consider now the Lorentz transformations for ϑ = 0 which are referred to as ‘boosts’. constitutes an overcomplete parametrization of the rotations (see Chapter 5). 0 1 n (10. In analogy to equation (5. i. ϑ. A boost in the x1 –direction is L = exp(w1 K1 ). w = 0) = 0 0 0 R(ϑ) . Exercise 7.e.53) Here R(ϑ) are the 3 × 3–rotation matrices constructed in Chapter 5. Following the treatment of the rotation group SO(3) one can express the elements of L↑ through + the exponential operators L(ϑ.55) L 0 0 0 0 exp (w1 K1 ) = exp Using the idempotence property 0 −1 −1 0 2 0 0 1 0 (10. (10. however. L(ϑ.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 .57) . 10. k = 1.51) for w = 0 correspond to rotations of the spatial coordinates.1: Demonstrate the commutation relationships (10. and using the relationship Jk = 0 0 0 Lk (10.. 3 where we have deﬁned ϑ · J = k=1 ϑk Jk and w · K = following the algebra in Chapter 5.56) = 1 0 0 1 = 1 1 (10.50). To determine the explicit form of this transformation we evaluate the exponential operator by Taylor expansion. w) = exp ϑ · J + w · K . w ∈ R3 3 k=1 wk Kk . (10. Kk .

59) for the case k = 1 corresponds to the relative velocity of two frames of reference.58) = cosh w1 1 + sinh w1 1 0 −1 −1 0 The conventional form (10. for wk wk ∈ ] − ∞.59).63) which agrees with (10. in fact. The range of the parameters wk can now be speciﬁed. x3 = x3 (10. however.10.1).51) the explicit transformation for space–time–coordinates is then x1 = x1 − v1 t 1 − 2 v1 . .59) can be written 1 2 1 − v1 √−v1 2 1 − v1 2 1 − v1 . ∞[ . (10.1) of the Lorentz transformations is obtained through the parameter change sinh w1 v1 = = tanh w1 (10. We expect that vk can only assume values less than the velocity of light c which in the present units is c = 1.3.6. one obtains from (10.1: Natural Representation of the Lorentz Group one can carry out the Taylor expansion above: ∞ 293 L = n=0 2n w1 1 + 1 (2n)! ∞ n=0 2n+1 w1 (2n + 1)! 0 −1 −1 0 = cosh w1 −sinh w1 −sinh w1 cosh w1 . x2 = x2 . 1[.62) According to (10. 10.61) √ √−v1 √ exp (w1 K1 ) = 2 1 − v1 1 2 1 − v1 0 0 0 0 0 0 0 0 1 0 0 1 (10. From (10.60) sinh w1 = v1 / 2 1 − v1 .59) v1 = These two equations yield cosh w1 = 1/ and (10.59) cosh w1 Using cosh2 w1 − sinh2 w1 = 1 one can identify sinhw1 = sinh2 w1 + 1. 10. we can state vk ∈ ] − 1. cosh2 w1 − 1 and coshw1 = cosh2 w1 − 1 = cosh w1 sinh w1 sinh2 w1 + 1 . consistent with (10. (10. 10.64) We note that the range of wk -values is not a compact set even though the range of vk -values is compact.56. (10. vk deﬁned in (10. Accordingly. Correspondingly. t = t − v1 x1 2 1 − v1 .59) follows. 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. (10. This property is.

a3 )T and transform according to a µ = Lµ ν aν . . scalars like r = x2 + x2 + x2 . Another example is the potential 4-vector Aµ = (V.66) Examples of 4-vectors beside xµ are the momentum 4-vector pµ = (E. t) and A(r. tensor operators Tkm . r2 ) and. A third 4-vector is the so-called current vector J µ = (ρ. the charge density.69) where V (r. t) are the charge density and the current density.3) are the so-called 4–vectors aµ . we will encounter an impressive example of physical properties. f = f. A) (10. t) and J(r. Some of these r ˆ quantities were actually deﬁned with respect to representations of the rotation group in function spaces. x3 )T . a1 . the square of the electric ﬁeld |E(r. x2 . p) . not any physical property f ∈ R is a scalar. p = √ 2 1−v 1 − v2 (10.4). a2 .e. spherical harmonics Y m (ˆ). Counterexamples are the energy.294 Relativistic Quantum Mechanics 10. of a system of charges. These quantites always come as four components (a0 . total angular momentum r 1 2 3 states of composite systems like Y m ( 1 . The 4-vector character of J µ and of Aµ will be demonstrated further below. 4–Vectors and Tensors In this Section we deﬁne quantities according to their behaviour under Lorentz transformations. i. the z–component x3 of a particle. Hence. We have encountered examples in connection with rotational transformations. respectively.. J) (10. 4-Vectors The quantities with the transformation behaviour like that of the position–time vector xµ deﬁned in (10. another example is the rest mass m of a particle. Such quantities appear in the description of physical systems and statements about transformation properties are often extremely helpful and usually provide important physical insight.67) the transformation behaviour of which we will demonstrate further below. t) are the electrical and the vector potential of an electromagnetic ﬁeld. However. 2 |ˆ1 . t)|2 or the scalar product r1 · r2 of two particle positions. Presently. namely. vectors like r = (x1 . not in the so-called natural representation associated with the 3–dimensional Euclidean space E3 . (10. our deﬁnition of quantities with special properties under Lorentz transformations presently is conﬁned to the natural representation. Scalars The quantities with the simplest transformation behaviour are so-called scalars f ∈ R which are invariant under transformations. we have not yet deﬁned representations of Lorentz transformations beyond the ‘natural’ representation acting in the 4–dimensional space of position– and time–coordinates.68) where ρ(r. (10. Nevertheless. E = √ m mv . ﬁnally.65) An example is s2 deﬁned in (10.2 Scalars. We will see below how true scalars under Lorentz transformations can be constructed.

one can express [(L−1 )T ]µ ν = (L−1 )ν µ and.78) To prove the 4-vector property of ∂µ we will show that g µν ∂ν transforms like a contravariant 4vector. −a3 ) (10.12) a gµν b µ ν = Lµ ρ gµν Lν σ aρ bσ = aρ gρσ bσ (10. Covariant 4-vectors transform like a µ = gµν Lν ρ g ρσ aσ where we deﬁned g µν = gµν .66) together with (10.12). xν = g νσ Lρ σ gρµ x µ . This property follows from (10. aµ (10.73) Note that according to (10. In fact. −a2 .71) Contravariant and Covariant 4-Vectors It is convenient to deﬁne a second class of 4-vectors. using (10.70) is a scalar. ∂t (10. If aµ and bµ are 4-vectors aµ gµν bν (10. The 4-Vector ∂µ (10.17) can be written (L−1 )ν µ = Lµ ν . In any case one can express (10. lower.73) aµ = Lµ σ aσ . We do not fact.79) This is the inverse Lorentz transformation consistent with (10. 4–Vectors and Tensors Scalar Product then 295 4-vectors allow one to construct scalar quantities. The respective vectors aµ are associated with the 4-vectors aµ .. One calls 4-vectors aµ covariant and 4-vectors aµ contravariant.75) (10. ∂x µ (10. (10.76) An important example of a covariant 4-vector is the diﬀerential operator ∂µ = ∂ = ∂xµ ∂ .e. We like to point out that from deﬁnition (10. i.16).e. We start from x µ = Lµ ν xν . and . one can employ the tensors g µν to raise and lower indices of L ν establish here the consistency of the ensuing notation.2: Scalars. g µν ∂ν = Lµ ρ g ρσ ∂σ .77) The transformed diﬀerential operator will be denoted by ∂µ = def ∂ . (10. Multiplication (and summation) of x µ = Lµ ν xν by Lρ σ gρµ yields.72) where aν is a vector with transformation behaviour as stated in (10. i. (10. We have duplicated the expression for the inverse of Lµ ν to obtain the correct notation in terms of covariant. −a1 . the relationship being aµ = gµν aν = (a0 .66).72) of the covariant 4-vector follows = In µν and g µ as well. accordingly.17) Lµ σ is the transformation inverse to Lσ µ . gσν xν = Lρ σ gρµ x µ and g µσ gσν = δ µ ν ..10.74) g µν aν .

82) (10. i. Proof that pµ is a 4-vector We will demonstrate now that the momentum 4-vector pµ deﬁned in (10.89) .84) = 1 − (v)2 (10. this operator is equal to the d’Alembert operator which is known to be Lorentz-invariant.79) allows us to determine the connection between ∂µ and ∂µ .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 .e. dτ 1 − v 2 dt p0 = E = √ m m dt = √ .86) One can write (10.87) The remaining components of pµ can be written.67) transforms like (10. 2 dt dτ 1 − v (10. p1 = √ m v1 m dx1 = √ ..g. indices.85) 1 d d = √ . 1 − v2 1 − v 2 dt (10.83) is a scalar under Lorentz transformations. (10.81) ∂ . d’Alembert Operator We want to construct now a scalar diﬀerential operator.88) One can express then the momentum vector pµ = √ d m dxµ = m xµ .296 Relativistic Quantum Mechanics contravariant.. Using the chain rule of diﬀerential calculus we obtain 3 ∂µ = ν=0 ∂xν ∂ = g νσ Lρ σ gρµ ∂ν = Lµ ν ∂ν ∂x µ ∂xν (10. For this purpose we deﬁne ﬁrst the contravariant diﬀerential operator ∂ µ = g µν ∂ν = Then the operator 2 ∂µ ∂ µ = ∂t − 2 (10. (10. upper. In fact. 2 2 dt 1 − v 1 − v (10. For this purpose we consider the scalar diﬀerential (dτ )2 = dxµ dxµ = (dt)2 − (dr)2 It holds dτ dt from which follows 2 (10. i.66).− ∂t .e. ∂µ does indeed transform like a covariant vector.. e.

We will demonstrate now the 4-vector property of Aµ . the r. i.67) E2 − p 2 = or pµ pµ = m2 which. t). (10.s.h.91) (10. pµ on the l.94) 1 m In the non-relativistic limit the rest energy m is the dominant contribution to E. Expansion in should then be rapidly convergent. kinetic energy ± 2m and relativistic corrections.97) We have proven already that ∂µ is a contravariant 4-vector.h. in fact.89) must be a 4-vector. for these potentials. 8. 10. The momentum 4-vector allows us to construct a scalar quantity. of (10. t) = 0 .92) (10. (10. and had chosen the so-called Coulomb gauge (8. equivalently. If we can show that Aµ deﬁned in (10. hence. the so-called Lorentz gauge.69) is.10. in fact. respectively. namely. t) and A(r. (10. is a scalar. and.93) (10.69) for the electrodynamic potential and the 4-vector derivative (10. m2 m2 v 2 − = m2 1 − v2 1 − v2 (10. Lorentz-invariant. Lorentz Gauge In our previous description of the electrodynamic ﬁeld we had introduced the scalar and vector potential V (r.s. a contravariant 4-vector then the l. of (10. We like to rewrite the last result E 2 = p 2 + m2 or E = ± p 2 + m2 .77) which allow one to express (10.96) is a scalar and. in (10. One obtains E = ±m ± p2 2m (p 2 )2 + O 4m3 (p 2 )3 4m5 . This gauge is not Lorentz-invariant and we will adopt here another gauge.e.96) in the form ∂µ Aµ = 0 . Since xµ transforms like a contravariant 4-vector.96) The Lorentz-invariance of this gauge.h.89) alltogether transforms like a contravariant 4-vector. · A = 0.s.3: Relativistic Electrodynamics 297 d The operator m dτ transforms like a scalar.h.12).97) and. can be demonstrated readily using the 4-vector notation (10.s.95) 2 p This obviously describes the energy of a free particle with rest energy ±m. ∂t V (r.90) (10. t) + · A(r.. . yields according to (10. hence. in (10. namely pµ pµ = pµ gµν pν = E 2 − p 2 Evaluation of the r.3 Relativistic Electrodynamics In the following we summarize the Lorentz-invariant formulation of electrodynamics and demonstrate its connection to the conventional formulation as provided in Sect.

(10. according to (10. of (10.13). Equation (10. m. .e. yields ∂µ ∂ µ Aν (xσ ) = 4 π J ν (xσ ) . The l. · A(r. using (8. t) . ∂µ J µ (xµ ) = 0 . Since ∂µ ∂ µ transforms like a scalar one can conclude that Aν (xσ ) must transform like a 4-vector.105) . using (10.77) and (10. t) = −∂t V (r.s.69) and ∂ µ in (10. it follows that J µ . 3 (10. t) are known to obey the continuity property ∂t ρ(r. Consequently.s. F mn = − mn (10. k.98) can be written.s.9).100. this can be stated F k0 = −F 0k = E k .h. (10.102) In this equation the r. t) − 2 A(r.100) Similarly.298 Transformation Properties of J µ and Aµ Relativistic Quantum Mechanics The charge density ρ(r. t) = 0 (10. t) = 4 π J(r. 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 . must be a scalar.h. i.17) using the identity (8. 10. (10. t) = 4πρ(r. t) 2 ∂t A(r.99) Since this equation must be true in any frame of reference the right hand side must vanish in all frames. t) together with (10.68). One can readily verify. (10. t) − 2 V (r. F µν = x Ey B z 0 −Bx Ez −By Bx 0 Alternatively.101).6) and (8..82) as follows F µν = ∂ µ Aν − ∂ ν Aµ .98) which reﬂects the principle of charge conservation. t).83) and (10. n = 1.96). t) + · J(r. one obtains 2 ∂t V (r. using (10.69).18) and.99) must be a scalar. 2.97) and. t) . This principle should hold in any frame of reference. t) from (8.32). also the l. thereby. one obtains for A(r.e.103) B . (10. We want to derive now the diﬀerential equations which determine the 4-potential Aµ in the Lorentz gauge (10.96). (8. prove that Aµ is. i. a 4-vector. Using · ∂t A(r. The respective equation for A0 = V can be obtained from Eq. Since ∂µ transforms like a covariant 4-vector.h. t) and current density J(r. that F µν can be expressed through the potential Aµ in (10. transforms like a 4-vector. · A(r.101) Combining equations (10. has to transform like a contravariant 4-vector. t) = −∂t V (r. must transform likewise. t) = ∂t · A(r.. in fact. in fact.104) where mn = mn is the totally anti-symmetric three-dimensional tensor deﬁned in (5.

108) yields then the expressions for the transformed ﬁelds E and B .63). 10.102) ∂µ F µν = 4π J ν . by (10. t).111) One can readily prove that this equation is equivalent to the two inhomogeneous Maxwell equations (8. (10. respectively. E⊥ = B⊥ E⊥ + v × B √ 1 − v2 B⊥ − v × E = √ 1 − v2 (10.109) (10.2). From the deﬁnition (10. t) and B(r. equivalently.10.3.97).110) where E .112) (10. the components of the ﬁelds parallel and perpendicular to the velocity v which determines the Lorentz transformation.62) or. 8.106) In case that the Lorentz transformation Lµ ν is given by (10. where we used (10. one can conclude from (10.113) which can be shown to be equivalent to the two homogeneous Maxwell equations (8.107) y 1 z Bz −v1 Ey √ 2 √ 2 0 −Bx 1−v1 1−v1 Ez +v1 By By +v1 Ez √ 2 − √ 2 Bx 0 1−v1 1−v1 Comparision with F µν 0 −Ex −Ey −Ez E 0 −Bz By = x Ey B z 0 −Bx Ez −By Bx 0 (10. 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.105) and (10.1. These equations show that under Lorentz transformations electric and magnetic ﬁelds convert into one another.105) of the tensor F µν one can conclude the property ∂ σ F µν + ∂ µ F νσ + ∂ ν F σµ = 0 (10. B⊥ are. In a new frame of reference holds F µν = Lµ α Lν β F αβ (10. B .4). The results can be put into the more general form E B = = E .3: Relativistic Electrodynamics 299 The relationships (10. 8. Noting ∂µ F µν = ∂µ ∂ µ Aν − ∂µ ∂ ν Aµ = ∂µ ∂ µ Aν − ∂ ν ∂µ Aµ = ∂µ ∂ µ Aν .103.104) establishe the transformation behaviour of E(r. B and E⊥ .

93).g.5). We want to express this force through the tensor F µν . however. It holds for a particle with 4-momentum pµ as given by (10.114) reads using (10. hence.118) p· dt √ where we exploited the fact that according to p = mv/ 1 − v 2 holds p v.114) where d/dτ is given by (10.5). in fact. To avoid √ confusion we will employ in the following for the energy of the particle the notation E = m/ 1 − v 2 [see (10.115) This equation follows.s.67) and charge q dpµ q = pν F µν dτ m (10. For the spatial components.e. We have. i. of (10. Equation (10.86) and (10. yields dpx = q dt Ex + (v × B)x (10.87)] and retain the deﬁnition E for the electric ﬁeld.120) is referred to as the Lorentz force.116) (10.h. We want to √ demonstrate now that this equation is equivalent to the equation of motion (8. dτ m √ Employing again (10. (10. The term on the r. dt E From this one can conclude. also from the equation of motion (8. using (10.86) dE q = p·E .. provides an alternative description of the action of the Lorentz force. hence.114).114) is. e. demonstrated that (10. dE q = p·E dτ m or with (10. equivalent to (8.117) (10.5) where p = mv/ 1 − v 2 .5) taking the scalar product with p dp = qp · E (10.119) which is the x-component of the equation of motion (8.104). employing (10..114) reads then. 1 dE 2 1 dp 2 = = qp·E 2 dt 2 dt (10. The µ = 0 component of (10. for µ = 1.120) (10. E = m/ 1 − v 2 .103) dpx q = ( EEx + py Bz − pz By ) .86). .300 Lorentz Force Relativistic Quantum Mechanics One important property of the electromagnetic ﬁeld is the Lorentz force acting on charged particles moving through the ﬁeld.67).

The resulting expression should be a generalization of the function space representation (5. i. (10.124) In this expression J = (J1 .121) by (L−1 )µ ν xν and obtains ψ (xµ ) = ψ((L−1 )µ ν xν ) . We will denote the intended representation by L(ϑ. (10.4: Function Space Representation of Lorentz Group 301 10. (5. In this section we will investigate the transformation properties of scalar functions ψ(xµ ).e.e.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 . 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 (10.7. w. w) which correspond to the generators Jk and Kk in (10.127) .51) for the natural representation of the Lorentz group. For such functions holds in the transformed frame ψ (Lµ ν xν ) = ψ(xµ ) (10. xµ ). We like to express now ψ (x µ ) in terms of the old coordinates xµ . K2 . w)) = ρ eϑ·J + w·K which we present in the form L(ϑ.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 ) . K3 ) are the generators of L(ϑ.125) def (10. in as far as SO(3. (10.125)].10. ψ ∈ C∞ (4).123) This deﬁnition corresponds closely to the function space representation (5. For this purpose one replaces xµ in (10. However. obey a diﬀerent transformation law. def (10. w) = ρ(Lµ ν (ϑ.4 Function Space Representation of Lorentz Group In the following it will be required to decribe the transformation of wave functions under Lorentz transformations. J3 ) and K = (K1 ... taken at the pairs of points (x µ = Lµ ν xν .126) ρ ewk Kk − 1 1 .48) and (10. e.g. Functions which represent 4-vectorsor other non-scalar entities. i.48) of SO(3). we seek an expression which corresponds to (10. w) = exp ϑ · J + w · K . and which can be constructed following the procedure adopted for the function space representation of SO(3).3 or the bi-spinor wave function of electron-positron pairs in Sect. 10. in the present case we exclude the factor ‘−i’ [cf. We need to emphasize that (10. 10. Accordingly. 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. J2 .1) is a generalization (rotation + boosts) of the group SO(3).42) of SO(3). the charge-current density in case of Sect..121) covers solely the transformation behaviour of scalar functions.47).

obviously. [ Jk . i. x3 ) 2 = ψ(xµ ) + w1 (x1 ∂0 + x0 ∂1 ) ψ(xµ ) + O(w1 ) (10. In order to evaluate (10.134) . and [J1 . 1 This result. sinhw1 x0 + coshw1 x1 . K2 ] = K3 : [ J1 . which we like to demonstrate for J1 and K1 . K obey the same Lie algebra (10. [K1 .130) ew1 K1 −1 = e−w1 K1 (10. x1 . x2 . x1 ∂3 ] − [x2 ∂3 .129) which yields for small ϑ1 ρ eϑ1 J1 ψ(xµ ) = ψ(x0 . 0 1 (10. J2 ] = J3 . obviously. J ] [ Kk . The generators J . −sinϑ1 x2 + cosϑ1 x3 ) = ψ(xµ ) + ϑ1 (x3 ∂2 − x2 ∂3 ) ψ(xµ ) + O(ϑ2 ) . (10.133) We demonstrate this for three cases. the expression for K1 in (10. One can determine similarly K1 starting from coshw1 sinhw1 sinhw1 coshw1 = 0 0 0 0 0 0 1 0 0 0 . x1 ∂3 − x3 ∂1 ] = [x3 ∂2 . (10.131) This yields for small w1 ρ ew1 K1 ψ(xµ ) = ψ(coshw1 x0 + sinhw1 x1 .e. J3 = x2 ∂1 − x1 ∂2 . cosϑ1 x2 + sinϑ1 x3 .132) and. K ] = = k m Jm k m Jm = − k m Km .49) as the generators of the natural representation. namely [J1 . reproduces the expression for J1 in (10. K2 ] = −J3 . J2 = x1 ∂3 − x3 ∂1 . x3 ∂1 ] = −x1 ∂2 + x2 ∂1 = J3 .128) eϑ1 J1 = e−ϑ1 J1 1 0 = 0 0 0 0 0 1 0 0 0 cosϑ1 sinϑ1 0 −sinϑ1 cosϑ1 (10. K ] [ Jk .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. J2 ] = [x3 ∂2 − x2 ∂3 .128).126).302 One obtains J1 = x3 ∂2 − x2 ∂3 .

∂x0 R and ∂Ω ∂x3 ∂Ω ∂x0 The chain rule yields then ∂0 ∂3 ∂R ∂ ∂Ω ∂ x0 ∂ 1 ∂ + = − sinh2 Ω 3 0 ∂R 0 ∂Ω ∂x ∂x R ∂R x ∂Ω ∂R ∂ ∂Ω ∂ x3 ∂ 1 ∂ = + = − + cosh2 Ω 0 . [ J1 .140) allow one to ∂ ∂ express the derivatives ∂0 . 0 ∂cothΩ ∂x x = ∂R x0 = − ∂x3 R (10. i.87)].143) . Ω which are connected with x0 . ∂Ω .138) x3 x0 . The relationships (10. In both cases the radial coordinate is the quantity conserved under the transformations. x2 + x2 + x2 1 2 3 in the case of SO(3) and (x0 )2 − (x3 )2 in case of transformation (10.139. (5. x2 ∂0 ] − [x1 ∂0 . (10.137)..e.139) x3 = R sinhΩ (10. K2 ] = [x3 ∂2 − x2 ∂3 .135) (10. K2 ] = [x0 ∂1 + x1 ∂0 . can be simpliﬁed considerably. x3 as follows x0 = R coshΩ . For this purpose one expresses K3 in terms of hyperbolic coordinates R.125) in the case of a one-dimensional transformation of the type L(w3 ) = exp w3 K3 . 3 ∂R 3 ∂Ω ∂x ∂x R ∂R x ∂Ω = ∂Ω ∂tanhΩ 1 = cosh2 Ω 0 3 ∂tanhΩ ∂x x ∂Ω ∂cothΩ 1 = = − sinh2 Ω 3 . cothΩ = 3 . 10.10. 303 (10.140) x0 x The transformation to hyperbolic coordinates closely resembles the transformation to radial coordinates for the generators of SO(3) in the function space representation [cf.136) (10.141) (10. x0 ∂2 ] = −x2 ∂1 + x1 ∂2 = − J3 .4: Function Space Representation of Lorentz Group [ K1 . In the following we consider solely the case x0 ≥ 0.85-5. Eqs. x0 ∂2 ] = x3 ∂0 + x0 ∂3 = K3 . x0 ∂2 + x2 ∂0 ] = [x0 ∂1 . ∂3 in terms of ∂R .137) where K3 is given in (10. We note tanhΩ = ∂R x0 = . x0 ∂2 + x2 ∂0 ] = [x3 ∂2 .142) (10. One-Dimensional Function Space Representation The exponential operator (10. x2 ∂0 ] − [x2 ∂3 . a relationship which can also be stated R = and + − (x0 )2 − (x3 )2 (x0 )2 − (x3 )2 if x0 ≥ 0 if x0 < 0 (10.128).

144) The action of the exponential operator (10.304 Inserting these results into the deﬁnition of K3 in (10. therefore. complex-valued function. (10.5 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.151) is called the Klein-Gordon equation. In natural units the Klein–Gordon equation (10. momentum.92) one obtains the wave equation − or 2 2 µ ∂ ∂µ ψ(xν ) = m2 ψ(xν ) ψ(xν ) = 0 .147) (10. The partial diﬀerential equation (10.146) Applying the correspondence principle to (10. (10. in 4-vector notation reads pµ =⇒ pµ = i (∂t . mass. namely the Klein–Gordon and the Dirac equations.151) reads ∂µ ∂ µ + m2 ψ(xµ ) = 0 (10.122). representation theoretic treatment. The historic route to these two equations.148) ∂ µ ∂µ + m2 (10. however. ˆ (10.150) .145) ∂Ω 10. 10. and (time)−1 all have the same dimension. extremely useful. The latter property implies that upon change of reference frame ψ(xµ ) transforms according to (10. − ) = i ∂ µ . In the following we will employ so-called natural units = c = 1. Further below we will provide a more systematic. Free Particle Case A quantum mechanical description of a relativistic free particle results from applying the correspondence principle.128) yields K3 = x0 ∂3 + x3 ∂0 = ∂ . (length)−1 . is not very insightful. ˆ pµ =⇒ pµ = i(∂t .137) on a function f (Ω) ∈ C∞ (1) is then that of a shift operator ∂ L(w3 ) f (Ω) = exp w3 f (Ω) = f (Ω + w3 ) . In these units the quantities energy. We will provide a ‘derivation’ of these two equations which stem from the historical development of relativistic quantum mechanics.121. In the position representation the correspondence principle states ˆ E =⇒ E = − ∂t i ˆ p =⇒ p = i which. ∂Ω Relativistic Quantum Mechanics (10. ) = i ∂µ . but certainly is short and. which allows one to replace classical observables by quantum mechanical operators acting on wave functions.149) where ψ(xµ ) is a scalar.

t) = 0 (10. t) ) 2m 1 = ( ψ ∗ (r. t) ψ(r.150) is invariant under Lorentz transformations. t) = ψ ∗ (r. t)∂t ψ ∗ (r.150) one considers Im ψ ∗ ∂µ ∂ µ + m2 ψ = 0 (10.152) which has the same form as the Klein–Gordon equation in the original frame. t) ψ ∗ (r. t) ψ(r. 2mi (10. Under Lorentz transformations the free particle Klein–Gordon equation (10.5: Klein–Gordon Equation or 2 ∂t − 2 305 + m2 ψ(xµ ) = 0 . t) = (10. In order to obtain the conservation law connected with the Klein–Gordon equation (10. and that (as postulated) ψ(xµ ) is a scalar. t) − ψ(r. t) = 0 i ( ψ ∗ (r.154) describes the current density connected with motion of the particle probability distribution. t) − ψ(r. t) ] 2mi (10. t)∂t ψ(r. t) = − 1 2m 2 ψ(r. t) 1 [ ψ ∗ (r. Current 4-Vector Associated with the Klein-Gordon Equation As is well-known the Schr¨dinger equation of a free particle o i∂t ψ(r. t) ψ(r. t) = · jKG (r. t) + where ρKG (r.155) describes the positive deﬁnite probability to detect a particle at position r at time t and where jS (r.153) is associated with a conservation law for particle probability ∂t ρS (r. (10.159) (10.156) · jS (r.160) ∂t (ψ ∗ ∂t ψ which corresponds to − ψ∂t ψ∗) + · (ψ ψ∗ ∂t ρKG (r.10.150) becomes ∂µ ∂ µ + m2 ψ (x µ ) = 0 (10.161) . t) + where ρS (r. t) − ψ(r.157) 2m which is equivalent to (10. t) (10. t) jKG (r.158) which yields 2 2 ψ ∗ ∂t ψ − ψ∂t ψ ∗ − ψ ∗ 2ψ +ψ ψ∗ − 2ψ∗ = ψ) = 0 (10.151) One can notice immeadiately that (10. t) ) . This follows from the fact that ∂µ ∂ µ and m2 are scalars.154). t) ψ ∗ (r. To 1 derive this conservation law one rewrites the Schr¨dinger equation in the form (i∂t − 2m 2 )ψ = 0 o and considers 1 2 Im ψ ∗ i∂t − ψ = 0 (10.

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.154). ±) = ± 2 m2 + po (10. it had been realized later.150) are actually determined by po and by the choice of sign ±. m 2 2 E0 = m2 + po .. the Klein-Gordon equation is considered as a suitable equation to describe spin–0 particles. The density ρKG (p. λ|xµ ) = 0 2 m2 + po .306 Relativistic Quantum Mechanics This conservation law diﬀers in one important aspect from that of the Schr¨dinger equation (10. and of negative energies E ≤ −m. with p = 0.166) ψo (p. for example pions.162) Inserting this into the Klein–Gordon equation (10.164) The corresponding energy is Eo (po .165) This result together with (10.e. The proper interpretation of the two cases is that the Klein–Gordon equation describes particles as well as anti-particles. The proper interpretation of ρKG (r. in that the expression for ρKG is not positive deﬁnite.p ei(p·r − λEo (p)t) Eo (p) = The spectrum of the Klein–Gordon equation (10. λ) = λ ψo (p. (10.167) m which is positive for λ = + and negative for λ = −. λ|xµ )ψo (p. λ) associated with the corresponding wave functions ψo (p. λ|xµ ) = Nλ. o namely. is actually that of a charge density. (10. corresponding to λ = +.168) . i. λ|xµ ) (10. λ|xµ ) = N e−iλmt .161) and (10. Today. λ|xµ ) according to (10. t).93] dispersion relationship connecting E0 . λ) actually describes charge density rather than probability. according to (??) is decribed by the r–independent wave function ψo (p = 0.163) which results in the expected [see (10. λ = ± (10. Solution of the Free Particle Klein–Gordon Equation Solutions of the free particle Klein–Gordon equation are ψ(xµ ) = N e−ipµ x µ = N ei(p0 ·r − Eo t) . and the density ρKG (p. p0 . t) = 0 (10. Generating a Solution Through Lorentz Transformation A particle at rest. We denote this by summarizing the solutions as follows ∂µ ∂ µ + m2 ψo (p.166) is Eo (p) ∗ ρKG (p. (10.162) shows that the solutions of the free particle Klein-Gordon equation (10. λ = ± . not of particle probability.150) is a continuum of positive energies E ≥ m. the anti-particles carry a charge opposite to that of the associated particles.151) yields 2 2 Eo − p0 − m2 ψ(r. corresponding to λ = −.

for the wave function (10. t).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. for negative energy solutions.174) which agrees with the expression given in (10.168). For this purpose we consider the wave function for a particle moving with momentum velocity v in the direction of the x3 –axis. In case of λ = + one obtains ﬁnally L(w3 )ψo (p = 0. i.166)] of the particle. 2 2 1−v 1−v (10.10. λ|xµ ) = = exp w3 ∂ ∂Ω N e−iλmRcoshΩ 3) N e−iλmRcosh(Ω+w . i. Aµ (xµ ) = (V (r. t).f. L(w3 )ψo (p = 0.166). (10.h.169) The addition theorem of hyperbolic functions cosh(Ω + w3 ) = coshΩ coshw3 + sinhΩ sinhw3 allows us to rewrite the exponent on the r.169) −iλ ( m coshw3 ) ( R coshΩ ) − iλ ( m sinhw3 ) ( R sinhΩ ) . This yields.145) choosing m = sinhw3 .138) and the relationships (10. using (10.177) . −A(r.61) yield for this expression −iλ √ m mv x0 − iλ √ x3 .e. t)) . in fact. of (10.. Such wave function should be generated by applying the Lorentz transformation in the function space representation p (10.. A(r.168). 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) (10.172) as the energy [c. for positive energy solutions.s. one has to interprete p = mv/ 1 − v 2 (10. (10. λ = +|xµ ) = N ei(px 3 − Eo (p)x0 (10. λ = −|xµ ) = N ei(px 3 + Eo (p)x0 .176) 10.6 Klein–Gordon Equation for Particles in an Electromagnetic Field We consider now the quantum mechanical wave equation for a spin–0 particle moving in an electromagnetic ﬁeld described by the 4-vector potential Aµ (xµ ) = (V (r. In case of λ = −.171) One can interpret then for λ = +.173) (10. t)) (10. Ω.170) The coordinate transformation (10.138) to replace t = x0 by hyperbolic coordinates R. (10. (10.175) as the momentum of the particle and one obtains L(w3 )ψo (p = 0.e.

We will also assume. For this purpose we split-oﬀ a factor exp(−imt) which describes the oscillations of the wave function due to the rest energy.e.150). (10.179) (10. According to the so-called minimum coupling principle the presence of the ﬁeld is accounted for by altering energy. (10. t) . A) or Aµ = (V. t) 2 (10..178) According to the replacements in Table 10. according to the rules shown in Table 10. i. 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. i. t) . See also Eq. t) = e−imt Ψ(r. A) free quantum particle quantum particle in ﬁeld (V. it’s rest energy. momenta for classical particles and the respective operators for quantum mechanical particles in the manner shown.1. m2 ) ψ(xµ ) = 0 .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.180) + m2 ψ(r. is much larger than all other energy terms.69)] one replaces the quantum mechanical operators. which couples a particle of charge q to an electromagnetic ﬁeld described through the potential Aµ (xν ).150) −g µν (i∂µ )(i∂ν ) + m2 ψ(xµ ) = 0 . t) = −i − q A(r. i∂t and −i in (10. t). we deﬁne ψ(r. that the mass m of the particle.1: Coupling of a particle of charge q to an electromagnetic ﬁeld described by the 4-vector potential Aµ = (V. In terms of space-time derivatives this reads (i∂t − qV (r. (10. in keeping withnthe non-relativistic limit. 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 10.147).182) and seek an equation for Ψ(r. For this purpose one writes the Klein-Gordon equation (10. in . According to the principle of minimal coupling [see (10. −A).e.308 free classical particle classical particle in ﬁeld (V.1 this becomes g µν (i∂µ − qAµ )(i∂ν − Aν ) ψ(xµ ) = m2 ψ(xµ ) which can also be written g µν πµ πν − .. t))2 ψ(r.. i. albeit in a form.93). and focus on the remaining part of the wave function. (10.e.

A stationary state of the Klein–Gordon equation is described by a wave function ψ(xµ ) = ϕ(r ) e−i t . identical to the Schr¨dinger equation (10. 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. i. larger than |i∂t Ψ/Ψ| and alrger than qV .. Ψ |q V | << m . (10.2) of a non-relativistic spin-0 particle o moving in an electromagnetic ﬁeld. t) 2m Ψ(r.183) The term on the l. | i∂t Ψ | << m . t) ≡ 0 r + Ze2 r 2 + 2 − m2 π ϕ(r ) = 0 .185) This is. To ‘manufacture’ pionic atoms.181) to a physical system we consider pionic atoms. however. (10.189) .181) yields (we assume now that the Klein–Gordon equation describes a 2 particle with mass mπ and charge −e) for qV (r.6: Klein–Gordon Equation with Electromagnetic Field particular.183).e. 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− . (10. The approximation is justiﬁed on the ground of the inequalities (10. The Klein-Gordon equation (10. The process of π − meson capture involves the so-called Auger eﬀect. t)]2 + qV (r. This application demonstrates that the Klein–Gordon equation describes spin zero particles. t) = − Ze and A(r.g..10. ﬁltered out of the beam and ﬁnally fall as slow pions onto elements for which a pionic variant is to be studied.186) then are slowed down.188) Inserting this into (10.h.181) reads then i ∂t Ψ(r. of (10. e.e.184) where we neglected all terms which did not contain factors m. i.. π − mesons are generated through inelastic proton–proton scattering p + p −→ p + p + π − + π + . t) (10. (10.s. Pionic Atoms To apply the Klein–Gordon equation (10. t) = ˆ [p − q A(r. spin-0 mesons.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.

θ.196) In this expression the number n deﬁned through n = n− −1 (10.195) The latter problem leads to the well-known spectrum En = − mπ (Ze2 )2 . . 2. φ. φ) = ( + 1) Y m (θ. (10. (10. i. .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 (10.310 Relativistic Quantum Mechanics Because of the radial symmetry of the Coulomb potential we express this equation in terms of spherical coordinates r. 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 . 1. . (10.193) (10..191) m (θ.192) ˆ are spherical harmonics. . φ) (10. (10.197) counts the number of nodes of the wave function. . The similarity of (10.190) 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 . φ) .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 (10.191) 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. i.e. .192) leads then to the ordinary diﬀerential equation d2 − dr2 ( + 1) − Z 2 e4 2 Ze2 + + r2 r 2 − m2 π R (r) = 0 . n = 1. this quantity deﬁnitely must be an integer. 2n2 = 0.199) ( + 1) − Z 2 e4 . the eigenfunctions of the operator L2 ˆ L2 Y m (θ. φ) R (r) Y r m (θ.198) . n − 1 .e. . (10.. r r r r sin θ r sin θ Ze2 2 r ) (10.

. = 0. . n = 0.. − + 1. . (10.10. results in the (10. .201) mπ Z 2 e4 m2 − m2 π π = − . . .196) if one makes the replacement 2 values which can be obtained E −→ One obtains − m2 π . . i. . . 2. m) = mπ 1 + Z 2 e4 (n− − 1 2 + ( + 1 )2 − Z 2 e4 )2 2 n = 1.204) reads 1 (10. . . n − 1 m = − . .203) spectrum .e.202) Solving this for state) yields (choosing the root which renders = 1 + mπ Z 2 e4 ( n + λ( )+1 )2 ≤ mπ . . 2 e2 −→ e2 mπ .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 ).194) d2 λ( ) ( λ( ) + 1 ) 2 Ze2 − + + dr2 r2 r 2 311 − m2 π R (r) = 0 .205) α √ 2 2 1+ (n − β + β −α) and one obtains the series of approximations 1 1+ ≈ ≈ ≈ 1+ ≈ 1+ (n − β + α √ β 2 −α)2 1 1+ (n − α 2β α + O(α2 ))2 1 1+ n2 − α n + O(α2 ) β α 1 α n2 + α2 βn3 + O(α3 ) − α2 8n4 1 α 2n2 + α2 2βn3 + O(α3 ) . Introducing α = Z 2 e4 and 1 β = + 2 (10. (10. EKG = Using (10. . 2mπ 2 −→ λ( ) .197. 10. .6: Klein–Gordon Equation with Electromagnetic Field and one can write (10. 1.199) and deﬁning EKG (n. which corresponds to a bound = 0. . . 1.200) The bound state solutions of this equation should correspond to from (10. 2mπ 2 (n + λ( ) + 1)2 (10. . + (10. . . 1.

. i. the Dirac equation introduced below. A more satisfactory Lorentz–invariant wave equation. t) . In order to describe electron spectra one must employ the Lorentz-invariant 1 wave equation for spin.209). 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 . the second term the non-relativistic energy. However. It must be pointed out here that does not denote energy.. i.209). for example. in turn. the π − meson is a pseudoscalar particle.208. through the correspondence principle.204) EKG (n. 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.209) which. t) = ± m2 − 2 Ψ(r. The latter spectrum shows.207) Here the ﬁrst term represents the rest energy. however. the Klein–Gordon equation had been rejected since it did not yield a positive-deﬁnite probability density. a splitting of the six n = 2. It should feature then a diﬀerential operator of the type D = iγ µ ∂µ .2 particles. 2n2 2βn3 8n4 4n4 (10. and the third term gives the leading relativistic correction. the wave function changes sign under reﬂection. m) ≈ m − mZ 2 e4 mZ 4 e8 − 2n2 2n3 1 + − 1− Relativistic Quantum Mechanics α α2 α2 α2 − + + + O(α3 ) . it does not agree with observations of the hydrogen spectrum.206) 1 2 3 + O(Z 6 e12 ) . Equations (10. is related to the equation E 2 = m2 + p 2 of the classical theory which involves a term E 2 .e.312 ≈ From this results for (10. in fact..e. The latter term agrees with observations of pionic atoms. arises because the Klein–Gordon equation.208). 10. His solution . but for a long time to no avail. = 1 states into groups of two and four degenerate states. but not an operator iγ · as suggested by the principle of relativity (equivalence of space and time). Also.146) leads to the wave equation i∂t Ψ(r. however. a feature which is connected with the 2nd order time derivative in this equation. t) (10.7 The Dirac Equation Historically.208) These two equation can be combined i∂t + m2 − 2 i∂t − m2 − 2 Ψ(r. would have only a ﬁrst order time derivative.e. but in the present case rather the negative of the energy. This derivative. (10. Many attempts were made by theoretical physicists to ‘linearize’ the square root operator in (10.208. 10. i. . are unsatisfactory since expansion of the square root operator involves all powers of the Laplace operator. Finally. Dirac succeeded. Application of the correspondence principle (10. one with a positive-deﬁnite density. 4n (10. is identical to the two equations (10. 10.

3 .210) where P = iγ µ ∂µ . j = 1. (10. namely. . ψd (xµ ). Initially. strengthens the believe of many theoretical physicists today that the properties of physical matter ultimately derive from a.215) . 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. it was assumed that the 4-dimensional space introduced by Dirac could be linked to 4vectors. Properties of the Dirac Matrices Let us now trace Dirac’s steps in achieving the linearization of the ‘square root operator’ in (10.212) where we have changed ‘dummy’ summation indices.214) From this follows that γ 0 has real eigenvalues ±1 and γ j . one route to important discoveries in physics is the creation of new mathematical descriptions of nature. Starting point is to boldly factorize.211) into (10. 3 has imaginary eigenvalues ±i. (10. ψ2 (xµ ). are associated with a positive-deﬁnite particle density. according to (10. . therefore. beautiful mathematical theory and that.208). However.66). For µ = ν condition (10.7: Dirac Equation 313 to the problem involved an ingenious step. this was not so. the 4-dimensionsional representation discovered by Dirac involved new physical properties. exploited ∂µ ∂ν = ∂ν ∂µ .e. quantities with the transformation law (10. therefore. . Comparing the left-most and the right-most side of the equations above one can conclude the following property of γµ γ µ γ ν + γ ν γ µ = [ γ µ . j = 1. the operator of the Klein–Gordon equation ∂µ ∂ µ + m2 = −( P + m ) ( P − m ) (10. one can impose the condition γ0 is hermitian . Inserting (10. γ 3 can only be realized by d×d–matrices requiring that the wave function Ψ(xµ ) is actually a d–dimensional vector of functions ψ1 (xµ ). γ 2 . Accordingly. In fact. Instead. the quantities γ 0 . γ ν ]+ = 2 g µν (10.211) Obviously. γ 1 .10. i. γ j . but did not commute the. the realization that the linearization can be carried out only if one assumes a 4-dimensional representation of the coeﬃcients γ µ . 2. unspeciﬁed algebraic objects γ µ and γ ν . these descriptions ultimately merging with the true theory of matter.1 2 and anti-particles. The discovery by Dirac. spin. 3 are anti-hermitian . 2..213) reads (γ µ )2 = 1 µ = 0 −1 µ = 1.209). so far. achieved through a beautiful mathematical theory. 2.210) yields −g µν ∂µ ∂ν − m2 = (iγ µ ∂µ + m)(iγ µ ∂µ − m) 1 = −γ µ γ ν ∂µ ∂ν − m2 = − ( γ µ γ ν ∂µ ∂ν + γ ν γ µ ∂ν ∂µ ) − m2 2 1 µ ν ν µ = − ( γ γ + γ γ ) ∂µ ∂ν − m2 2 (10. (10. It turns out that no 4-vector of real or complex coeﬃcients can satisfy these conditions. We seek to identify the coeﬃcients γ µ .213) We want to determine now the simplest algebraic realization of γ µ . yet to be discovered.

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. 2.314 For µ = ν (10.218) The Pauli matrices allow one. is unique. We will argue further below that the choice f γ µ .222) ∗ ∗ ∗ ∗ where we have deﬁned Ψ† = (ψ1 . rather than to the right side.216) i.222) and multiplication from the right by γ 0 yields the adjoint Dirac equation Ψ† (xµ ) γ 0 iγ µ ∂µ + m ← =0. holds 2 σj = 1 . (10. One can readily show using the hermitian property of the Pauli matrices (γ 0 )† = γ 0 and (γ j )† = −γ j for j = 1. 1 (10. From this one can conclude for the determinants of γ µ det(γ µ γ ν ) = det(−γ ν γ µ ) = (−1)d det(γ ν γ µ ) = (−1)d det(γ µ γ ν ) . as long as det(γ µ ) = 0 the dimension d of the square matrices γ µ must be even so that (−1)d = 1. to construct four matrices γ µ for the next possible dimension d = 4. 3 which.221) =0 (10. ψ4 ) and where ∂µ denotes the diﬀerential operator ∂ µ operating to the left side. σ 3 for which. ψ3 . 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. rather than a scalar wave function..213) reads γ µ γ ν = −γ ν γ µ . is implied by (10. except for similarity trasnformations. ψ2 . the γ µ are anti-commuting.215).220) where Ψ(xµ ) represents a 4-dimensional wave function. For d = 2 there exist only three anti-commuting matrices. γj = 0 −σ j σj 0 . in fact.217) Obviously. (10. (10.219) Using property (10. σ j σ k = − σ k σ j for j = k .e. The Adjoint Dirac Equation The adjoint equation is Ψ† (xµ ) i(γ µ )† ∂µ + m ← ← (10. σ 2 .213) is satisﬁed.218) of the Pauli matrices one can readily prove that condition (10.224) . however. (10.223) Inserting this into (10. namely the Pauli matrices σ 1 . Relativistic Quantum Mechanics (10. A proper choice is γ0 = 1 1 0 0 −1 1 . This property can also be written (γ µ )† = γ 0 γ µ γ 0 . in fact.

2.1: Derive (refeq:Dirac-intro20a) from (10.. [˜ µ .7. (10.10.226) (10.229) The similarity transformation (10. and γ0 = ˜ 0 1 1 1 0 1 . t) = 0 (10.227) A representation often adopted beside the one given by (10.228) (10.230) Schr¨dinger Form of the Dirac Equation o Another form in which the Dirac equation is used often results from multiplying (10. 3 .227) leaves the algebra of the Dirac matrices unaﬀected and commutation property (10.213).233) The eigenstates and eigenvalues of H correspond to the stationary states and energies of the particles described by the Dirac equation.Chiral Representation 315 The Dirac equation can be subject to any similarity transformation deﬁned through a non-singular ˜ 4 × 4–matrix S.7: Dirac Equation Similarity Transformations of the Dirac Equation . j = 1. γ ˜ Exercise 10. γ ν ]+ = 2 g µν .219) is the socalled chiral representation deﬁned through ˜ Ψ(xµ ) = S Ψ(xµ ) . 2. αj = ˆ 0 σj σj 0 . (10. 3 and ˆ β = 1 1 0 0 −1 1 . j = 1. γj = ˜ 0 σj 1 S = √ 2 −σ j 0 1 1 1 1 1 −1 1 1 (10. 2. (10. (10. i.221) from the left by γ 0 ˆ ˆ i ∂t + i α · − β m Ψ(r.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 ) . .231) ˆ whereα has the three components αj . (10.213) still holds. j = 1.225) .222. 3 . 10.e.227). ˆ ˆ ˆ Ho = α · p + β m . 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.

d4 = 1 0 0 −1 1 0 0 −1 0 1 1 0 0 i −i 0 (10.Good. the matrix elements of C = A ⊗ B are Cjk.237) which are the ordered products of the operators ±1 and d1 . The Cliﬀord algebra C4 entails a subgroup G4 of elements ± dj1 dj2 · · · djs . d2 .. The representations of Cliﬀord algebras Cm are well established. m = Aj Bkm . There are (including the diﬀerent signs) 32 elements in G4 which we deﬁne as follows Γ±1 = ±1 1 Γ±2 = ±d1 .213).316 Cliﬀord Algebra and Dirac Matrices The matrices deﬁned through dj = iγ j .e.e. 3 . For example. in case of C4 .235) as can be readily veriﬁed from (10. The representations of this group are given by a set of 32 4 × 4–matrices which are equivalent with respect to similarity transformations. Rev. To complete the proof in this section the reader may consult Miller ‘Symmetry Groups and their Application’ Chapter 9.236) where ‘⊗’ denotes the Kronecker product between matrices. 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 .Phys.235) and. Γ±9 = ±d2 d3 Γ±10 = ±d2 d4 .235). satisfy the anti-commutation property dj dk + dk dj = 2 for j = k 0 for j = k Relativistic Quantum Mechanics d4 = γ 0 (10. The three Pauli matrices also obey the property (10.H. The associative algebra generated by d1 . j = 1. Γ±7 = ±d1 d3 . d3 . 27.213). to Dirac matrices. d4 . Obviously. Γ±15 = ±d2 d3 d4 Γ±16 = ±d1 d2 d3 d4 (10. .6 and R. Γ±4 = ±d3 . i.. i. any product 1 of the dj ’s can be brought to the form (10. d2 = . Γ±3 = ±d2 . hence. page 187. This property extends then to 4 × 4–matrices which obey (10. One can conclude then that also any set of 4 × 4–matrices obeying (10. j1 < j2 < · · · < js s ≤ 4 (10. . (1955).Mod. The reader may . 2. d4 is called a Cliﬀord algebra C4 . Γ±13 = ±d1 d2 d4 .238) These elements form a group since obviously any product of two Γr ’s can be expressed in terms of a third Γr . Γ±8 = ±d1 d4 . Γ±14 = ±d1 d3 d4 .234) (10.235) can diﬀer only with respect to unitary similarity transformations. Since the Γj are hermitian the similarity transformations are actually given in terms of unitary transformations. form a Cliﬀord algebra C3 . Γ±11 = ±d3 d4 Γ±12 = ±d1 d2 d3 .237) by means of the property (10. Γ±5 = ±d4 Γ±6 = ±d1 d2 .

µ µ (10. 10. Lorentz Transformation of the Bispinor State Actually. Hence. Except for a similarity transformation.4: Show that from (10. in frames connected by Lorentz transformations. 3 has imaginary eigenvalues ±i and can be represented by an anti-hermitian matrix.e. transform coordinates and derivatives of the Dirac equation.241) .239) Form invariance implies that the matrices γ µ should have the same properties as γ µ . (10. i.10.. j = 1. 10.e.. x = Lµ ν xν .e.. The latter transformations imply that coordinates transform according to (10.218) of the Pauli matrices σ j . Exercise 7.213.80).80) by Lµ ρ and using (10. Exercise 7. these properties determine the matrices γ µ uniquely.2: Demonstrate the anti-commutation relationships (10.6). in a moving frame holds iγ µ ∂µ − m Ψ (x ) = 0 .8: Lorentz Invariance of the Dirac Equation 317 also want to establish the unitary transformation which connects the Dirac matrices in the form (10. We can. namely. and derivatives according to (10.e. i.. we do not know yet how to transform the 4-dimensional wave function Ψ and the Dirac matrices γ µ .240) . and γ j . it must hold γ µ = γ µ .76) yields ∂ρ = Lν ρ ∂ ν . 2. exist.3: Demonstrate the anti-commutation relationships (10.236) with the Dirac representation (10. Inﬁnitesimal Bispinor State Transformation We want to show now that a suitable transformation of Ψ(xµ ) does.. i. the form of the Dirac equation is identical in equivalent frames of reference. The transformation is assumed to be linear and of the form Ψ (x ) = S(Lµ ν ) Ψ(xµ ) µ µ (10.6).219). Exercise 7. However.215). i. Multiplication and summation of (10. therefore.e. in fact. x µ = Lµ ν xν (10.214) follows that γ 0 has real eigenvalues ±1 and can be represented by a hermitian matrix. a result one could have also obtained by applying the chain rule to (10. 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.218) of the Dirac matrices γ µ . i. in a moving frame should hold iγ ∂µ − m Ψ (x ) = 0 .8 Lorentz Invariance of the Dirac Equation We want to show now that the Dirac equation is invariant under Lorentz transformations.

249) from which we can conclude σµν = −σµν νµ = −σνµ i.e. g is an anti-symmetric matrix. It holds then S( µν i ) = 1 − σµν 1 4 µν µν = S(− νµ i ) = 1 + σµν 1 4 µν . hence.h.245) The construction of S(Lη ξ ) will proceed using the avenue of inﬁnitesimal transformations. follows S(Lη ξ )γρ S −1 (Lη ξ ) = Lσ ρ γσ .318 Relativistic Quantum Mechanics where S(Lµ ν ) is a non-singular 4 × 4–matrix.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 . One can ﬁnally conclude multiplying both sides by g ρµ S(Lη ξ )γ µ S −1 (Lη ξ ) = Lν µ γν .s. i. state that the Dirac equation (10.248) Here σµν denote 4 × 4–matrices operating in the 4-dimensional space of the wave functions Ψ. µ g ρν = µν and. The inﬁnitesimal transformation S(Lρ σ ) which corresponds to (10. Multiplication of this property by g from the right yields ( g)T = − g.247) the inﬁnitesimal matrix µν is anti-symmetric.e.. One can.221) upon Lorentz transformation yields iS(Lη ξ )γ µ S −1 (Lη ξ ) Lν µ ∂ν − m Ψ (x ) = 0 . of the equation.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.243) We want to determine now the 4 × 4–matrix S(Lη ξ ) which satisﬁes this condition. using gρµ γ µ = γρ .244) from which.243) in a form in which the Lorentz transformation in the form Lµν is on the r. the transformation (10. S( µν ) should not change its value if one replaces in its argument µν by − νµ .246) (10. Multiplication of (10.247) can be expanded S( µν i ) = 1 − σµν 1 4 µν (10.250) σµν = −σνµ . (10.241) implies a similarity transformation Sγ µ S −1 . it must hold (10. however. The elements of g are. (10. µ (10. (10. νµ (10. the coeﬃcients of which depend on the matrix Lµ ν deﬁning the Lorentz transformation in such a way that S(Lµ ν ) = 1 for Lµ ν = δ µ ν holds. Ob1 viously. hence.. The proper starting point for a constructiuon of S(Lη ξ ) is actually (10. For this purpose we exploit (10.12) in the form Lν µ gνσ Lσ ρ = gµρ = gρµ . . We had introduced in (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.

254). (10.253) 1 2 αβ ( δ µ β γα − δ µ α γβ ) . (10.s.255) satisfy condition (10.252) Since six of the coeﬃcients αβ can be chosen independently. Exercise 7. 10. γk ] = ˜ γ ˜ jk σ 0 0 σ . a solution of condition (10.251) into (10. j = 1.s. β and operate in the same space as the Dirac matrices. ˜ ˜ σ only six generators need to be determined) σ0j = ˜ i [˜0 . of (10.. must be symmetric with respect to interchange of the indices α and β. −γ ) are ˜ γ ˜ γ0 = ˜ 0 1 1 1 0 1 . (10.216) 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 (10.5: Show that the σαβ deﬁned through (10. γβ ]− 2 (10. this condition can actually be expressed through six independent conditions.254) The proper σαβ must be anti-symmetric in the indices α. Comparing this with the l.252) also as a sum over both indices of αβ . γj ] = γ ˜ 2 −iσ j 0 0 iσ j .h.251) S −1 ( µν ) = 1 + σµν µν 1 4 Inserting (10..255) which can be demonstrated using the properties (10. 10. like the expression on the l. (10. 2.h.213.256) One can readily verify that the non-vanishing generators σµν are given by (note σµν = −˜νµ .h.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 (10.10.s.246) results then in a condition for the generators σµν − i ( σαβ γ µ − γ µ σαβ ) 4 αβ = νµ γν . In this representation the Dirac matrices γµ = (˜ 0 .254) is σαβ = i [ γα . (10. the expression on the r. In fact. 10. γj = ˜ 0 −σ j σj 0 .257) .228.252) results in the condition for each α. For this purpose we inspect the properties of the generators in a particular representation. Algebra of Generators of Bispinor Transformation We want to construct the bispinor Lorentz transformation by exponentiating the generators σµν .248. Furthermore. the chiral representation introduced above in Eqs. of (10. γ µ ]− = 2i ( δ µ β γα − δ µ α γβ ) . namely. For this purpose one needs to express formally the r.h.229).s. σjk = [˜j .e. i. β [ σαβ . 3 . For this purpose we need to verify that the algebra of the generators involving addition and multiplication is closed.

and since the algebra of the Pauli matrices is closed. the representation (10. not necessarily inﬁnitesimal.51). the algebra of these generators is closed under addition and multiplication.248) allows us to write the transformation S for any.255) yields a closed algebra.44) 0 −w1 −w2 −w3 w 0 ϑ3 −ϑ2 = 1 w2 −ϑ3 0 ϑ1 w3 ϑ2 −ϑ1 0 µν (10. (10. .258) again into block-diagonal operators. in particular.241) bispinor wave functions from one frame of reference into another frame of reference. 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.259) becomes in the chiral representation ˜ S(w.51) and S. In fact. ϑ) = e 2 (w − iϑ)·σ 0 − 1 (w + iϑ)·σ 0 e 2 1 (10.261) We note that this operator is block-diagonal. i. Since such operator does not change its block-diagonal form upon exponentiation the bispinor transformation (10. one can express the tensor µν through w and ϑ using µν = µ ρ g ρν and the expression (10. µν in the exponential form i S = exp − σµν 4 µν .e. since both operations convert block-diagonal operators A 0 0 B (10.260) Inserting this into (10.320 Relativistic Quantum Mechanics Obviously. One should.259) We had stated before that the transformation S is actually determined through the Lorentz transformation Lµ ν ..262) This expression allows one to transform according to (10. therefore. any representation of the generators. 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 (10.259) yields the desired connection between the Lorentz transformation (10. In order to construct an explicit expression of S in terms of w and ϑ we employ again the chiral representation. 2 0 −(w + iϑ) · σ 13 + σ23 ˜ 23 ) (10. be able to state S in terms of the same parameters w and ϑ as the Lorentz transformation in (10.

h. 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 . ϑ) = S(−w. For this purpose we prove ﬁrst the relationship S −1 = γ 0 S † γ 0 . The time-like component ρ of j µ 4 iγ µ ∂µ + iγ µ ∂µ ← Ψ(xµ ) = 0 .263) obviously is a scalar under Lorentz transformations and that the left hand side must then also transform like a scalar.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 . Obviously.263) allows one to conclude that j µ must transform like a contravariant 4-vector as the notation implies. The conservation law (10.268) holds for the bispinor Lorentz transformation. (10. the property applies then in any representation of S. the conservation law (10. j) = Ψ† (xµ )γ 0 γ µ Ψ(xµ ) . (10.270) We conclude that (10.264) (10. and addition yields Ψ† (xµ )γ 0 The last result can be written ∂µ Ψ† (xν )γ 0 γ µ Ψ(xν ) = 0 . in fact.224) from the right by Ψ(xµ ). i. The reason is that the r. (10. Since ∂µ transforms like a covariant 4-vector.e..224). for j µ (xµ ) = (ρ.268) We will prove this property in the chiral representation. (10.266) ρ(xµ ) = Ψ† (xµ )Ψ(xµ ) = s=1 |ψs (xµ )|2 (10. For our proof we note ﬁrst ˜ ˜ S −1 (w.263) does hold.267) has the desired property of being positive deﬁnite.265) (10.s. Multiplying (10. (10.269) One can readily show that the same operator is obtained evaluating γ S (w.263) Starting point are the Dirac equation in the form (10.271) . This transformation behaviour can also be deduced from the transformation properties of the bispinor wave function Ψ(xµ ).221) from the left by Ψ† (xµ )γ 0 . −ϑ) = e− 2 (w − iϑ)·σ 0 1 (w + iϑ)·σ 0 e2 1 (10. j µ must transform like a contravariant 4-vector. We will now determine the relationship between the ﬂux j µ = Ψ (x )γ 0 γ µ Ψ (x ) † µ (10. of (10.10.221) and the adjoint Dirac equation (10.

the Dirac particles are described by 4-dimensional.231. (10. Note that we have assumed in (10.272) results in the expected transformation behaviour µ j = Lµ ν j ν . speciﬁc for the Dirac free particle. Like in non-relativistic quantum mechanics the free particle wave function plays a central role. not only as the most simple demonstration of the theory.271) that the Dirac matrices are independent of the frame of reference. These operaˆ tors. bispinor wave functions and we need to determine corresponding components of the wave function.273) where we have employed (10. but also as providing a basis in which the wave functions of interacting particle systems can be expanded and characterized. One obtains using (10. 10. In the following we want to provide this characterization.282. additional characterizations. a property which leads to the energy– momentum (dispersion) relationship of the Dirac particle. 10.283)]. 10. µ (10. the identiﬁcation of observables and their quantum mechanical operators.e.f. only two degrees of freedom of the bispinor four degrees of freedom are independent [c.9 Solutions of the Free Particle Dirac Equation We want to determine now the wave functions of free particles described by the Dirac equation. 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. t) and is determined through the momentum p ∈ R3 .268) µ j = Ψ† (xµ )S † γ 0 γ µ SΨ(xµ ) = Ψ† (xµ )γ 0 S −1 γ µ SΨ(xµ ) . however. In relativistic quantum mechanics a Dirac particle can also be characterized through a momentum. We will start from the Dirac equation in the Schr¨dinger form (10.275) The free particle wave function is an eigenfunction of Ho . Like for the free particle wave functions of the non-relativistic Schr¨dinger and the Klein–Gordon o equations one expects that the space–time dependence is governed by a factor exp[i(p · r − t)]. (10. As pointed out. For this purpose we consider . commute with each other. The additional degrees of freedom described by the four components of the bispinor wave function require. as is required for the mentioned property. The solutions provide also a complete.232. as just mentioned. In case of non-relativistic quantum mechanics the free particle wave function has a single component ψ(r. of which the wave functions are eigenfunctions as well. the wave function has four components which invite further characterization of the free particle state.322 Relativistic Quantum Mechanics in a moving frame of reference and the ﬂux j µ in a frame at rest.272) With S −1 (Lη ξ ) = S((L−1 )η ξ ) one can restate (10.246) S −1 (Lη ξ )γ µ S ( Lη ξ ) = (L−1 )νµ γν = (L−1 )ν γ ν = Lµ ν γν .233). i.233) o i∂t Ψ(xµ ) = Ho Ψ(xµ ) . As it turns out. The operators commute also with Ho in (10. Combining this with (10. (10. 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.. (10.76).274) 10.

and φo .281) Obviously. E(p) = m2 + p 2 .234) and (10.282.276). 10.93. (10. We want to show now that these degrees of freedom correspond to a spin-like property. later to be identiﬁed with momentum and energy.234) of Pauli matrices holds (σ · p)2 = p 2 1 One can. the socalled helicity of the particle. 10. like the Klein–Gordon equation.231.276) where p and together represent four real constants. hence.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 .280) σ·p (σ · p)2 φ0 = χ0 = +m ( + m) ( + m) p2 χ = χ0 . These two relationships are consistent with each other. conclude 1. the Dirac equation. 10. namely φo and χo are related as follows φ0 χo = = σ·p χ0 −m σ·p φ0 . χo each represent a constant.9: Solutions of the Free Particle Dirac Equation the following form of the free Dirac particle wave function Ψ(xµ ) = φ(xµ ) χ(xµ ) = φo χo ei(p·r − t) 323 (10. reproduce the classical relativistic energy–momentum relationships (10.232) leads to the 4-dimensional eigenvalue problem m σ·p σ · p −m φo χo = φo χo .282) (10. +m (10.280) which has a positive and a negative solution = ± E(p) .277) To solve this problem we write (10.284) The relationships (10.283) imply that the bispinor part of the wave function allows only two degrees of freedom to be chosen independently. (10. (10. one ﬁnds using (5.281).10.276) into (10. . 1 (10.94) Equation (10.279) According to the property (5.278) provides us with information about the components of the bispinor wave function (10.277) explicitly ( − m) 1 φo − σ · p χo 1 − σ · p φo + ( + m) 1 χo 1 = 0 = 0.279) the well-known relativistic dispersion relationship 2 = m2 + p 2 (10. two-dimensional spinor state. 2 − m2 0 (10. Inserting (10.283) where is deﬁned in (10. In fact. from (10.

±) = = = Ψ† (0. where S is given by (10.281)] separately.e. using γ 0 as given in (10. (10.259). hence. the solution for = −E(p). of (10. Similarly. +|xµ ) = N+ (p) u σ·p E(p) + m u ei(p·r − t) .241).290) which diﬀers from the normalization condition (10.286) 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.290) should diﬀer. (10.h. we present φo through the normalized vector φo = u1 u2 = u ∈ C2 . S −1 = γ 0 S † γ 0 and.s. ±) Ψ† (0.219) and u† u = 1 [c. We want to demonstrate now that the signs on the r. Employing (10.e. Ψ† (0. (10. ±). ±) γ 0 Ψ(p. according to (10. ±) = S Ψ(0. the l.287. = −E(p) .s.290). The form of this condition and of (10. Hence. = +E(p) . 0) γ 0 u 0 = |N+ (0)|2 .. Ψ† (p.268)..286) for p = 0 yields. i. +) γ 0 Ψ† (p. is invariant under Lorentz transformations. One can see this as follows: Let Ψ(p.287) in the minus sign on the r. (10.287) the form of which will be justiﬁed further below. +) γ 0 Ψ(0. ±) γ 0 Ψ(0. ±) . ±). we demonstrate that the product Ψ† (p. ±) denote the corresponding solution of a particle in its rest frame. −|xµ ) = N− (p) −σ·p E(p) + m u u ei(p·r − t) .285) The corresponding free Dirac particle is then described through the wave function Ψ(p.287) will be justiﬁed now. according to (10. u† u = |u1 |2 + |u2 |2 = 1 .. 10. ±) γ 0 S −1 γ 0 γ 0 S Ψ(0. ±) denote the solution of a free particle moving with momentum p in the laboratory frame.288) corresponding to the wave function Ψ(p. ±)γ 0 Ψ(p. the solution for = +E(p). of (10. we present the negative energy solution.289) Here N− (p) is a constant which will be chosen to satisfy the normalization condition Ψ† (p.291) Note that we have used that. u† u = |u1 |2 + |u2 |2 = 1 . +) = |N+ (0)|2 (u† . +) = 1 . (10. ±) S † γ 0 S Ψ(0. 10. (10.h. (10.285)].287. is Ψ(p. through χo given by χo = u1 u2 = u ∈ C2 .s. S † = γ 0 S −1 γ 0 . +) γ 0 Ψ† (p. ±) Ψ† (0. and let Ψ(0.e.292) . (10. +) = −1 . (10. First. i. (10. i. For this purpose we consider ﬁrst the positive energy solution.f.h. The connection between the solutions. For the positive energy solution.

f. Exercise 7.h. 325 (10. σ·p u∗ E(p) + m T γo u σ·p E(p) + m u = 1 (10.9: Solutions of the Free Particle Dirac Equation The same calculation for the negative energy wave function as given in (10. −) = |N− (0)|2 (0.297) 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 (10.285) results in 2 N+ (p) 1 − p2 (E(p) + m)2 = 1 (10. (5.h. We consider ﬁrst the positive energy solution. We can also conclude from our derivation N± (0) = 1 . u† ) γ 0 0 u = − |N− (0)|2 .219) yields 2 N+ (p) u† u − u† (σ · p)2 u (E(p) + m)2 = 1.300) (10. 2 E(p) (10.299) ( m + E(p) )2 . using γ 0 as given in (10.289) yields Ψ† (0.s.290) to assign a positive value to |N− (0)|2 .302) .294) We want to determine now N± (p) for arbitrary p. Condition (10.287) written explicitly using (10.s.296) Replacing (σ · p)2 by p 2 [c. (10.298) This result completes the expression for the wave function (10. (10.293) Obviously.298) becomes N+ (p) = m + E(p) . −) γ 0 Ψ(0.301) yields the normalization coeﬃcient N+ (p) = m + E(p) .295) Evaluating the l. of (10. σ·p u∗ E(p) + m T u σ·p E(p) + m u = 1 (10. 2m (10.286) is 2 N+ (p) (u∗ )T .10.234)] and using the normalization of u in (10.286).6: Show that the normalization condition 2 N+ (p) (u∗ )T . ( m + E(p) )2 − p 2 (10. this requires the choice of a negative side on the r.

286) into (10.e. The wave functions are also eigenstates of the momentum operator i∂µ . (10.286. λ|xµ ) = λ E(p) Ψ(p.. Ho ] = 0 follows from (10. The corresponding operator which measures this observable is 1 p ˆ Λ = σ· . identical to the condition (10. Rather than considering the ˆ observable (10.289) are not completely speciﬁed. The property [σ · p. exp[i(p · r − t)] = exp(ipµ xµ ). Helicity The free Dirac particle wave functions (10. i. pj ] = 0 . j = 1.e.289. yields 2 N− (p) u† (σ · p)2 u − u† u (E(p) + m)2 = −1 . λ|xµ ) (10. 2.289) is T −σ·p −σ · p 2 E(p) + m u N− (p) u∗ . −p).289). thereby. (10.e. The commutation property [σ · p. 10. λ|xµ ) . i. hence.309) .275). λ|xµ ) in 4-vector notation.307) where pµ = ( .326 Relativistic Quantum Mechanics We consider now the negative energy solution. the wave functions constructed represent eigenstates of Ho . deﬁned as the component of the particle spin along the direction of motion. This can be veriﬁed expressing the space–time factor of Ψ(p. (u∗ )T γ o = −1 (10. We ˆ want to show that this operator commutes with Ho and p to ascertain that the free particle wave ˆ function can be simultaneously an eigenvector of all three operators. The wave functions (10. This degree of freedom 1 describes a spin– 2 attribute..275) yields Ho Ψ(p. 10.s.300) have been constructed to satisfy the free particle Dirac equation (10. 10.308) 2 |p| ˆ Note that p represents here an operator. This attribute is the so-called helicity. Inserting (10. (10.290) written explicitly using (10. however.. not a constant vector. conclude N− (p) = m + E(p) 2m (10.h. We can.303) E(p) + m u Evaluating the l.307) we investigate ﬁrst the observable due to the simpler operator σ · p. Condition (10.233) and ˆ ˆ ˆ from the two identities 1 1 0 0 −1 1 σ 0 0 σ ·p − ˆ σ 0 0 σ ·p ˆ 1 1 0 0 −1 1 = 0 (10.304) This condition is.286. 3 is fairly obvious. i∂µ Ψ(p.305) and. have completed the determination for the wave function (10.306) i.296) for the normalization constant N+ (p) of the positive energy solution. the two components of u indicate another degree of freedom which needs to be deﬁned. λ|xµ ) = pµ Ψ(p.

+ 2 |r. = +E(p). Therefore. 0. Ho and Λ deﬁned in (10. −.e. can ˆ ˆ be simultaneously diagonal.311) 2 −p 0 m + Ep 1 where e3 denotes the unit vector in the x3 -direction and where ˆ Ep = m2 + p2 .. t) = Np e (10. We consider ﬁrst the simplest case that particles move along the x3 –direction. +. 327 0 (σ · p)2 ˆ 2 (σ · p) ˆ 0 0 (σ · p)2 ˆ 2 (σ · p) ˆ 0 (10.9: Solutions of the Free Particle Dirac Equation and 0 σ σ 0 = ·p ˆ σ 0 0 σ ·p − ˆ − σ 0 0 σ ·p ˆ 0 σ σ 0 ·p ˆ = 0. ˆ The condition that the wave functions (10. are 1 i(px3 − E t) 0 p e Ψ(pˆ3 . 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 . Ho and σ·p commute with each other and..e. p3 ). Np = m + Ep .312). 1)T are eigenstates of 2 σ 3 with eigenvalues ± 1 .312) We assume now particles with negative energy. i.e. .10. In this case Λ = 2 σ 3 . − 1 |r. 1 p = (0. 0)T and (0.. in principle. Since helicity is deﬁned relative to the direction of motion of a particle the characterization of u as an eigenvector of the helicity operator. − 1 |r. t) = Np e (10. + 1 |r. We assume ﬁrst particles with positive energy. i. +. According to the deﬁnition (5. i. t) = Np e 2 1 p m + Ep 0 0 i(px3 − E t) 1 p e Ψ(pˆ3 .308) above. −. is independent of the direction of motion of the particle.224) 1 of σ3 the two u vectors (1.286) are eigenfunctions of Λ as well will specify now the vectors u. 2m (10. = −E(p).313) 2 0 1 where Ep and Np are deﬁned in (10. 2 the wave functions which are eigenstates of the helicity operator.310) We have shown altogether that the operators p. States which are simultaneously eigenvectors of these three operators are also simulteneously eigenvectors of the three operators p. t) = Np e 1 0 0 p m + Ep i(px3 + E t) 1 p e Ψ(pˆ3 . hence.

+. for an r–independent wave function. + ) 2 1 u(p. − 1 |r. 2 1 0 (10. Generating Solutions Through Lorentz Transformation The solutions (10. − 1 ) 2 ei(p·r − Ep t) (10.321) .311. These eigenstates are obtained through a rotational transformation (5.311. − 2 |r.319) 1 Ψ(p. 10. t) = Np ei(p·r + Ep t) (10.262) for the bispinor wave function and the transformation (10. in which case the transformation is given by (5.223) as follows 1 u(p. +. denoted by ˜. [One can also express the rotation through Euler angles α. + 1 |r. t) 2 = Np u(p. i. 10. − 2 ) u(p.220).226. 1)T have to be replaced by eigenstates u± (p) of the spin operator along the direction of p. p) e |p| (10.313) except that the states (1.312) can be obtained also by means of the Lorentz transformation (10.123).] The corresponding free particle wave functions are then Ψ(p.314) (10. −.318) 1 Ψ(p. +. + 2 ) ei(p·r − Ep t) (10. 10. t) 2 = Np 1 u(p. (10. + 2 ) p 1 m + Ep u(p. γ . − 1 ) 2 −p 1 m + Ep u(p.315) e3 × p ˆ ∠(ˆ3 . a wave function which represents free particles at rest. 1 |t) = √2 e−imt . 0)T and (0. For this purpose one starts from the solutions of the Dirac equation in the chiral representation (10.317) Ψ(p.e. − ) 2 where ϑ(p) = = = exp exp i − ϑ(p) · σ 2 i − ϑ(p) · σ 2 1 0 0 1 .328 Relativistic Quantum Mechanics To obtain free particle wave functions for arbitrary directions of p one can employ the wave functions (10. 5.312). γ and are 1 0 1 ˜ Ψ(p = 0.. + 1 ) 2 p 1 m + Ep u(p. − 2 ) −p 1 m + Ep u(p. t) = N√ ei(p·r + Ep t) (10. −. The corresponding wave functions are determined through ˜ i˜ 0 ∂t − m Ψ(t) = 0 . + 2 ) u(p.229). β.222. + 2 |r.316) describes a rotation which aligns the x3 –axis with the direction of p.320) where Ep and N are again given by (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 . (10.10. i.176) imt → i ( p3 x3 and for the bispinor part according to (10. This correspondence justiﬁes the characterization ±. 10. + 1 |r.323) should yield the solutions for non-vanishing momentum p in the x3 –direction.322) can be written in spinor form 1 √ 2 φo χo e imt .325) One should note that φo . t) e 2 e− 2 w3 1 = √ 1 2 e 2 w3 i(px3 − E t) p e . −.174. 10. χo are eigenstates of σ 3 with eigenvalues ±1.324. 0.325) to (10. − 1 |r. φo . χo ∈ 1 0 .322) The reader can readily verify that transformation of these solutions to the Dirac representationsas deﬁned in (10.e.. − 1 |t) = √2 .324) 0 e − 1 w3 σ 3 2 φo χo = e 2 w3 σ φo 1 3 e− 2 w3 σ χo 1 3 . +. 2 |t) = √2 . 2 The solutions (10. yields for the exponential space–time dependence according to (10.262) φo χo → e 2 w3 σ 0 1 3 Et ) (10.323) Transformation (10.322).311.9: Solutions of the Free Particle Dirac Equation 0 1 1 ˜ Ψ(p = 0. w3 ). −. − 1 |t) = √2 e−imt . ± 1 of the wave functions stated in (10.313) in the p → 0 limit. 10. 2 0 1 1 0 +imt 1 1 ˜ Ψ(p = 0.311) e 1 w 2 3 ˜ Ψ(p(w3 )ˆ3 . For the resulting wave functions in the chiral representation one can use then a notation corresponding to that adopted in (10. −1 e 0 0 1 +imt 1 ˜ Ψ(p = 0. 2 0 e −1 329 (10. for w = (0. +. Applying (10.262) for a boost in the x3 –direction. 0 1 (10.228) yields the corresponding solutions (10. +.

t) e 2 = i(px3 + E t) p e 1 Ψ(p(w3 )ˆ3 . +. t) e 1 0 1 0 0 1 0 1 1 0 1 0 0 1 0 1 i(px3 − E t) p e 1 Ψ(p(w3 )ˆ3 . − 1 |r.61) between the parameter w3 and boost velocity v3 . and the expression (10.328) the relationship (10. +. −.326) where according to (10. t) e 1 = √ 2 −e− 2 w3 − 1 w3 2 1 i(px3 + E t) p e ˜ Ψ(p(w3 )ˆ3 .330 e 1 w 2 3 Relativistic Quantum Mechanics 1 0 1 0 0 1 0 1 1 ˜ Ψ(p(w3 )ˆ3 . 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.329) . 2 (10. 2 sinh x = 2 coshx − 1 .327) Employing the hyperbolic function properties cosh x = 2 coshx + 1 . + 2 |r. −. −. + 2 |r.61) p(w3 ) = m sinhw3 .228) yields w3 cosh 2 = sinh w3 2 1 Ψ(p(w3 )ˆ3 . − 2 |r. t) e = i(px3 + E t) p e (10.311) for Ep one obtains 1 = √ 2 1 1 √ 2 + 1 = 1 − v3 2 m2 + = 2 m2 v3 2 1 − v3 cosh w3 2 1+ 2 v3 2 + 1 1 − v3 +m = 2m Ep + m 2m (10. − 2 |r. t) e 2 e 1 = √ 1 2 − e 2 w3 i(px3 + E t) p e (10. Transformation to the Dirac representation by means of (10.

in general.9: Invariance of Dirac Equation Revisited and similarly sinh w3 = 2 Ep − m = 2m p 2m (Ep + m) 331 (10.330) Inserting expressions (10.334) (10.329.243).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 ).243) and.330) into (10. the proper transformation S(Lη ξ ). ρ(Lη ξ ) has been deﬁned in (10. reproduces the positive energy wave functions (10. but not the observer. Λ|xµ ) from the solutions describing particles at rest Ψ(p = 0. Making this expectation a postulate allows one to derive the condition (10.311) as well as the negative energy solutions (10. 10.10. λ. indeed.331) i S(Lη ξ )γ µ S −1 (Lη ξ ) ρ(Lη ξ )∂µ ρ−1 (Lη ξ ) − m Ψ (xµ ) = 0 (10. To show this we rewrite the Dirac equation (10. Such transformation had been applied by us. the essential property stating the Lorentz–invariance of the Dirac equation. In the new derivation we consider the particle described by the wave function transformed. λ. (10.327). This transformation. Invariance of Dirac Equation Revisited At this point we like to provide a variation of the derivation of (10. when we generated the solutions Ψ(p. = γν . thereby. that if Ψ(xµ ) is a solution of the Dirac equation that Ψ (xµ ) as given in (10. The reason why we provide another derivation of (10. We expect. Actually. 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 (10.333) .176). expresses the system in the old coordinates. and since (2) the calculations following (10.221) using (10. refered to as the active transformation. is suﬃcient since (1) it must hold then for any ﬁnite transformation. of course.168–10.123) above and characterized there.331) is a solution as well. we will derive this equation only for inﬁnitesimal transformations.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. which however.313) for −p. 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 transformation is Ψ (xµ ) = S(Lη ξ ) ρ(Lη ξ ) Ψ(xµ ) (10. Λ|t).332) Here we have made use of the fact that S(Lη ξ ) commutes with ∂µ and ρ(Lη ξ ) commutes with γ µ .243) considered solely the limit of inﬁnitesimal transformations anyway.

0.341) [K3 . w = (0. i. ∂µ ] = [x0 ∂1 + x1 ∂0 . . 1. ∂µ ] = (J )ν µ ∂ν .337) [J2 . that any solution Ψ(xµ ) transformed according to (10. ∂µ ] = (10. ∂µ ] = [x2 ∂1 − x1 ∂2 . ϑ = (0. ∂µ ] = (K )ν µ ∂ν . of course.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. As mentioned already we will show condition (10.262). 0 0 ∂3 −∂2 0 −∂3 0 ∂1 0 ∂2 −∂1 0 −∂1 −∂0 0 0 −∂2 0 −∂0 0 −∂3 0 0 −∂0 µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ = = = = = = = = = = = = = = = = = = = = = = = = 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 (10. ∂µ ] = (10.342) One can readily convince oneself that these results are consistent with (10. w = (0. 0. Inspection of (10.334) for inﬁnitesimal Lorentz transformations Lη ξ . ϑ = (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. . [K . 0).338) [J3 . 0). . w = (0.e.339) [K1 . ∂µ ] = [x0 ∂2 + x2 ∂0 . therefore.336) We will proceed with this task considering all six cases: [J1 . 0). it is met by S(Lη ξ ) as given in the chiral representation by (10. 0). 0. We will proceed by employing the generators (10.340) [K2 . 1) . . ∂µ ] = (10. ∂µ ] = (10. We have demonstrated. The second condition is identical to (10.336).243) and. the Dirac equation is invariant under active Lorentz transformations. 0.335) shows that we need to demonstrate [J .. ∂µ ] = [x1 ∂3 − x3 ∂1 .331) is again a solution of the Dirac equation. 0). 0. ∂µ ] = [x3 ∂2 − x2 ∂3 .332 Relativistic Quantum Mechanics We will demonstrate now that the ﬁrst condition is satisﬁed by ρ(Lη ξ ). ∂µ ] = (10. ∂µ ] = (10. . ∂µ ] = [x0 ∂3 + x3 ∂0 .

221) reads then [ iγ µ (∂µ + iqAµ ) − m ] Ψ(xν ) = 0 (10.345) holds |qV | << m . The Dirac equation (10. we deﬁne φ(xµ ) = e−imt Φ(xµ ) χ(x ) = e µ −imt (10. This solution exhibits a timedependence exp[−i(m + )t] where for holds in the non-relativistic limit | | << m.233) by replacing i∂t by (see Table 10. Following the respective procedure developed for the Klein-Gordon equation in Sect.347) (10.g. 10. for the scalar ﬁeld V in (10.10. Non-Relativistic Limit We want to consider now the Dirac equation (10.1 in Sect 10. For this purpose we choose the decomposition Ψ(xµ ) = φ(xµ ) χ(xµ ) .6 we assume that the ﬁeld is described through the 4-vector potential Aµ and.6 above).1) i∂t − qV and p by ˆ ˆ π = p − qA . 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 10.349) We want to focus on the stationary positive energy solution. The Dirac equation in the Schr¨dinger form reads then o i∂t Ψ(xµ ) = ˆ ˆ where α and β are deﬁned in (10.344) Using the notation σ = (σ1 .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 ) (10.345) (10. Equivalently.345) in the so-called non-relativistic limit in which all energies are much smaller than m. we replace the operator ∂µ by ∂µ + iqAµ . (10. accordingly. Accordingly. (10.232). σ2 . 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 . e.348) (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.351) X (x ) µ .350) (10.10: Dirac Particles in Electromagnetic Field 333 10.

361) .360) For an arbitrary function f (r) holds ( [Aj . (10.357) Equation (10.354) The properties (10. Ak ] i i i i =0 =0 = q q [Aj . Ak ] .354) ˆ 0 ≈ σ · π Φ − 2m X . one can replace X in (10. πk ] = = 1 1 [ ∂j + qAj . For this purpose we employ the identity (5. and. ∂k ] + [∂j . Ak ] + q 2 [Aj .346. ∂k ] + q [Aj .352) ˆ σ · π X + qV Φ ˆ σ · π Φ + qV X − 2mX .350. ∂k ] + q [ ∂j .351) in (10.353) (10.230). (10. One obtains [πj .349) yields i∂t Φ i∂t X = = ∂t X X Relativistic Quantum Mechanics << m . henceforth. accordingly. Ak ] ) f = ( ∂j Ak − Ak ∂j + Aj ∂k − ∂k Aj ) f . (10. ˆ Equation (10. (10.352) allow one to approximate (10. which in the present case states ˆ ˆ2 ˆ ˆ (σ · π)2 = π + i σ · (π × π) .353) by X ≈ to obtain a closed equation for Φ ˆ σ·π 2m 2 (10. i i (10.356). ∂k + qAk ] i i 1 1 1 1 [ ∂j . 10. (10.355) ˆ σ·π Φ 2m (10.348. derived in Sect. 10. does not need to be considered anymore.359) We want to evaluate the latter commutator. due to the m−1 factor. 10.7. π ] . identiﬁes X as the small component of the bi-spinor wave function which. ∂k ] + [∂j . 5.334 and assume that for the time-derivative of Φ and X holds ∂t Φ << m .356) i∂t Φ ≈ Φ + qV Φ . Φ Using (10.358) ( πj πk − πk πj ) = jk [πj . ˆ ˆ For the components of π × π holds ˆ ˆ π×π = jk (10.357) for Φ can be reformulated by expansion of (σ · π)2 .

r (10. 10. the spin. but rather remains as a steady “guest” of non-relativistic physics with the proper interaction term. one obtains ( [Aj .2) it introduces the o extra term qσ · B Φ which describes the well-known interaction of a spin. t) ≈ ˆ [p − q A(r.2) governing a one-dimensional wave function ψ ∈ C. 0. Let us consider brieﬂy the consequences of the interaction of a spin.363) allow us to write (10.366) (10.e. t) (10.343) for the vector potential Aµ = (− Ze2 . i.344.362) or.358.Spectrum We want to describe now the spectrum of a relativistic electron (q = −e) in the Coulomb ﬁeld of a nucleus with charge Ze. the interaction qσ · B induces a 1 precession of the spin. 0. πk ] = q q (( ∂j Ak − ∂k Aj )) = − i i ×A jk q = − B i jk (10. 10.1 with the magnetic ﬁeld.1 particle with a magnetic 2 ﬁeld B. [πj .223) shows that the propagator in (10.364) governing a two-dimensional wave function Φ ∈ C2 with the corresponding non-relativistic Schr¨dinger equation (10. Ak ] ) f = [ ((∂j Ak )) − ((∂k Aj )) ] f (10. t) [see (8.366) can be interpreted as a rotation around the ﬁeld B by an angle qtB. 10.10: 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 ((· · ·)).365) Comparision of this expression with (5. while agreeing in all other respects with the non-relativistic Schr¨dinger equation (10. 5. t) = × A(r.363) where we employed B(r.364) which is referred to as the Pauli equation.357) in the ﬁnal form i∂t Φ(r.360) and Aµ = (V. In other words. using (10.367) .10.222. 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) . 0) .2 around the magnetic ﬁeld.. The formal solution of this equation is Φ(t) = e−iqtB·σ Φ(0) .359. The respective bispinor wave function Ψ(xµ ) ∈ C4 is described as the stationary solution of the Dirac equation (10. o reveals a stunning feature: the Pauli equation does justice to its two-dimensional character. (10. Equations (10. t)]2 q − σ · B(r. Comparision of (10. ∂k ] + [∂j .6)]. −A). t) + q V (r.1 which emerged in the Lorentz-invariant theory as an algebraic 2 necessity. Dirac Particle in Coulomb Field . does not leave the theory again when one takes the non-relativistic limit. t) 2m 2m Φ(r.

(5. noting from (10. 10.1 one can write (10.180). µ = ν and altering ‘dummy’ summation ˜ ˜ ˜ ˜ .368) ˜ where Ψ(xµ ) and γ µ are deﬁned in (10.h.371) such that we can conclude that (10.ν=1 µ =ν γ µ γ ν πµ πν .231).232) in Sect. γ ˆ ˜ ˆ (10.370) ˜ [ i˜ µ (∂µ + iqAµ ) − m ] [ i˜ µ (∂µ + iqAµ ) + m ] Ψ(xµ ) = 0 γ γ (10.s.e.370) Any solution of (10.336 Relativistic Quantum Mechanics For the purpose of the solution we assume the chiral representation. 2.371). (5. but the converse is not necessarily true.372) (10.6.368) is also a solution of (10.7 one can write the second term in (10.229). ˜ However.373) (10. The diﬀerence arises from the term γ µ πµ γ ν πν in (10.228) and in (10. once a solution Ψ(xµ ) of (10. we solve ˜ [ i˜ µ (∂µ + iqAµ ) − m ] Ψ(xµ ) = 0 γ (10.230) γ µ γ ν = − γ ν γ µ .369). 3 µ=0 ˆ2 (˜ µ )2 πµ = π0 − π . (˜ 0 )2 = 1 and (˜ j )2 = γ 1 γ −1 j = 1. 1. 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. is a solution of (10. ˜ (10. i. 5.370). 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 .375) Following the algebra that connected Eqs. according to (10. Equation (10.370) is obtained then ˜ [ i˜ µ (∂µ + iqAµ ) + m ] Ψ(xµ ) γ is a solution of (10.369) from the left by γ ν πν + m.374) The ﬁrst term on the r. indeed. γ ˆ2 ˆ2 (10.370) resembles closely the Klein-Gordon equation (10.368) ˜ ( γ µ πµ − m ) Ψ(xµ ) = 0 . ˜ ˜ ˆ ˆ (10. can be rewritten using.374). multiplying (10.369).230). 3. For this purpose we ‘square’ the Dirac equation.370) for which holds ˜ ˆ ˜ ˆ 3 γ πµ γ πν = ˜ ˆ ˜ ˆ µ=0 µ ν (˜ µ )2 πµ + γ ˆ2 µ. respectively. Employing πµ as deﬁned ˜ in Table 10. but diﬀers from it in an essential way.369) For our solution we will adopt presently a strategy which follows closely that for the spectrum of pionic atoms in Sect.

380) where f = f (r.378) can be evaluated using (10. γ µ γ ν πµ πν = ˜ ˜ ˆ ˆ µ.379) The commutators [ˆ0 .381) According to (10. i. it holds 1 4 [˜ µ . πj ] π ˆ = = (−i∂t + qA0 .378) According to the deﬁnition (10. Altogether.229). γ ] [ˆ0 .344). γ 0 ] [ˆj . πν ] γ ˜ π ˆ µ.367) and the deﬁnition (10. γ j ] [ˆ0 .377) which follows readily from the deﬁnition (10. Since [ˆµ . γ ν ] [ˆµ .ν=1 µ =ν 3 j=1 3 3 = 1 4 1 2 1 [˜ .367) of Aµ .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. 2.10: Dirac Particles in Electromagnetic Field indices. π0 ] γ ˜ π ˆ j=1 = [˜ 0 . πj ] in (10. πν ] = 0 π ˆ for µ.344) π ˆ [ˆ0 . t) is a suitable test function and where ((· · ·)) denotes the range to which the derivative is limited.376) This expression can be simpliﬁed due to the special form (10. γ ν ] [ˆµ . due to A = 0. πν ] γ ˜ π ˆ µ. −i∂j ] = − [ (∂t + iqA0 ) ∂j − ∂j (∂t + iqA0 ) ] f i ((∂j qA0 )) f (10. ν = 1.e. πj ] + γ ˜ π ˆ 4 0 j [˜ j . 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 (10.376–10. one can summarize (10.ν=1 µ =ν (10.367) holds qA0 = r ˆ Ze2 .380) γ µ γ ν πµ πν = i ˜ ˜ ˆ ˆ µ. πj ] ..ν=1 µ =ν = [˜ µ . the commutators [˜ 0 . 3 (10.382) . γ ˜ π ˆ j=1 (10.ν=1 µ =ν σ 0 0 −σ · (( qA0 )) (10. γ j ] are γ ˜ [˜ 0 . r2 (10.10.ν=1 µ =ν 337 1 2 ( γ µ γ ν πµ πν + γ ν γ µ πν πµ ) ˜ ˜ ˆ ˆ ˜ ˜ ˆ ˆ µ.

385) 1 We split the wave function into two spin. (10. 6.389) According to the results in Sect. r r 2 2 2 j = 1.2 components ˜ Ψ(r) = and obtain for the separate components φ± (r) Ze2 + r 2 ˜ φ+ (r) ˜ φ− (r) (10. . 6.375) the ˆ ‘squared’ Dirac equation (10.1 . m = −j.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 .5 the operator iσ · r is block-diagonal in this basis such that only ˆ 1 the states for identical j. Accordingly. . 2 |ˆ)} as given in (6.384) into (10. only the two states {Yjm (j − 2 . .5 which describe the coupling of orbital angular momentum and spin 1 {( Yjm (j − 2 . hence.381).338 Relativistic Quantum Mechanics where r = r/|r| is a unit vector. .191) solved in Sect. in r ˆ fact. couples the orbital angular momentum of the electron to its spin. Insertion of (10. (10.388) in terms of states introduced in Sect. −j + 1. m values are coupled. however. is genuinely two-dimensional and. (10.383) We seek stationary solutions of this equation. .148). Such solutions are of the form ˜ ˜ Ψ(xµ ) = Φ(r) e−i t . In the latter case. (10.384) can be interpreted as the energy of the stationary state and.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. i. 6. The term iσ · r. 10. We note that these states are also eigenstates of the angular r 2 .368) reads − Ze2 ∂t − i r 2 + 2 + i σ·r ˆ 0 0 −σ · r ˆ Ze2 − m2 r2 Ψ(xµ ) = 0 (10.147.e. . 2 3 2 . (10. 1 |ˆ) ) . 1 |ˆ).6. Yjm (j + r 2 1 1 . (10. it is this quantity that we want to determine. + j } (10. a solution of the form ∼ Y m (ˆ) can be obtained. Yjm (j + 1 . Combining this result with (10.386) + 2 ± iσ · r ˆ Ze2 − m2 r2 φ± (r) = 0 .374). we express 2 the solution of (10.. 1 |ˆ).387) The expression (10.383) yields the purely spatial four-dimensional diﬀerential equation + Ze2 r 2 + 2 + i σ·r ˆ 0 0 −σ · r ˆ Ze2 − m2 r2 Φ(r) = 0 .

except for the slight diﬀerence in the expression of λ1. a speciﬁc pair of total spin-orbital angular momentum quantum numbers j. unaltered. Any similarity transformation leaves the ﬁrst term in (10.186)]..199).f.151). such transformation can be chosen as to diagonalize the second term.200). (6.2 (r) = 0 (10.2 (j) being 2 j = 1 . hence.10. The two eigenvalues of both matrices are identical and can be written in the form λ1 (j) [λ1 (j) + 1] where λ1 (j) λ2 (j) = = (j + 1 )2 − Z 2 e4 2 (j + 1 )2 − Z 2 e4 − 1 2 (10. involving the 2 × 2 unit matrix. together with the orthonormality of these two states leads to the coupled diﬀerential equation 2 ∂r + 2Ze2 + r 2 −m 1 0 0 1 ±iZe2 1 2 3 2 (10. the values of the argument of λ1. 1 |ˆ) = − Yjm (j ˆ r 2 2 1 2 h± (r) g± (r) Yjm (j − 1 . − 1 r2 (j − 1 )(j + 1 ) − Z 2 e4 2 2 ±i Ze2 (j + )(j + ) − Z 2 e4 We seek to bring (10. as in the case of pionic atoms.390) (10. conclude that the .396) and λ2 (j) [λ2 (j) + 1] (10. therefore.2 (j) + 1) 2 Ze2 + + r2 r 2 − m2 f1. we want to determine solely the spectrum.393) but do not explicitly consider further the wavefunctions. However. the missing additive term − 1 .392) h± (r) g± (r) = 0. 3 . property (6. We can.2 (j). rather than = 0. 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 . m and expand φ± (r) = Using σ · r Yjm (j ± 1 . 1. 1 |ˆ) r 2 2 2 2 r r .395) (10. . . .151].395. Since. the eigenvalues are independent of m.392) into diagonal form. We select. not the wave functions. . 1 |ˆ) + r Yjm (j + 1 . λ1 (j) and λ2 (j). 6.392). and except for the fact that 2 2 we have two sets of values for λ1.394) Equation (10. (10.2 (j)[λ1. (6.392) reads then in the diagonal representation 2 ∂r − λ1.391) derived in Sect. . which states that the states Yjm (j ± 1 . 1 |ˆ) r 2 (10. .5 [c. 1 |ˆ) are r 2 2 ˆ eigenfunctions of L2 . Obviously.2 (j) as given by (10.397) This equation is identical to the Klein-Gordon equation for pionic atoms written in the form (10. namely.10: Dirac Particles in Electromagnetic Field 339 ˆ momentum operator L2 [cf. in the present treatment. 10. namely.396) and (10.

j . . −j + 1.401) corresponds to a non-relativistic state 2 with quantum numbers n. . . . These considerations 2 are summarized in the following equations 1 ED (n. 2 2 m = −j.402) In this expression n is the so-called main quantum number. albeit with some modiﬁcations. accordingly. of these equations describe the binding energy. . j = − 2 .398. and spin-orbital angular momentum j = − 1 . 10. .2 . . . . .397) is again given by eq. − + 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 . . . 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.396). 1. . . . . . (10. m) = 1 + m (n− + . 1. .s. j = 1.396) we obtain. . j = 1 . the stationary states have binding energies E = − mZ 2 e4 n = 1. .h. It is given by n = n + + 1 where is the orbital angular momentum quantum number and n = 0. −j + 1. 2. 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 . . .400) (10.399) 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. 2. −j + 1. n − 1 . . 2n2 = 0. j = + 1 .404) = 0.399) n = 0. . .402) with (10. m = −j. . . One may also state 2 this in the following way: (10. j where 1 corresponds to λ1 (j) as given in (10. 1. . 2. One can equate (10.340 Relativistic Quantum Mechanics spectrum of (10. the second term on the r.398) 2 = 1 + m . (10. . . . counts the nodes of the wave function. . 1 ms = ± 2 (10.400) and (10.203). . The magnitude of the relativistic eﬀect 2 2 e4 . m = −j. . 1. n − 1 . . 10. = 1 + m Z 2 e4 √ ( n − − 1 + 2 − Z 2 e4 )2 . (10. Using (10. One obtains in case of (10. 2.395. 3 . . . . . 3. . m = − .395) and 2 corresponds to λ2 (j) as given in (10. 2 2 1 ≈ m − (10. . m) 2 n = 1.403) √ Z 2 e4 2 ED (n. and spin-orbital angular momentum j = + 1 .401) if one attributes to the respective 1 states the angular momentum quantum numbers = j + 2 and = j − 1 . . . . Obviously. . . . . . ( +1) − Z 2 e4 )2 (10. (10. 1. .400) corresponds to a non-relativistic state with quantum numbers n. . j . (10. including spin. In case of non-relativistic hydrogen-type 1 atoms.401) These expressions can be equated with the non-relativistic spectrum. . .

. j 1 2 1 2 1 2 3 2 1 2 1 2 3 2 3 2 5 2 341 rel. Degeneracies are denoted by ⇑.405 ) -13.551176 spectr. 0 0 1 1 0 1 1 2 2 spinorbital ang.0041 keV and e2 = 1/137. 10.40147 -1.40151 ⇑ . relate closely to the corresponding nonrelativistic states. (10. 2. i. (10.10. binding energy / eV Eq.402.403. r 2 2 2 . (10. for example.402) -13.. mom. Of particular interest is the eﬀect of spin-orbit coupling which removes. is in the range of 10 eV.10: Dirac Particles in Electromagnetic Field main quantum number n 1 2 2 2 3 3 3 3 3 orbital angular mom. . i.. (10.405) In order to demonstrate relativistic eﬀects in the spectrum of the hydrogen atom we compare in Table 10.e.60583 -3. j (10.402)] and the relativistic [cf. 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. the non-relativistic and relativistic energies are hardly discernible. .405)] spectrum of the hydrogen atom. 10. This is clearly demonstrated by the comparision of non-relativistic and relativistic spectra of a hydrogen-type atom with Z = 100 in Table 10. j. m) = m 1 + Z 2 e4 (n−j −1 + 2 (j+ 1 )2 − Z 2 e4 )2 2 n = 1. ± 3 .404) ﬁnally into the single formula ED (n. The table entries demonstrate that the energies as given by the expression (10. −j + 1.3. (10. .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. binding energy / eV Eq. the nonrelativistic degeneracy for the six 2p states of the hydrogen atom: in the present. case these six states are split into energetically diﬀerent 2p 1 and 2p 3 states. The energies were evaluated with m = 511. The reason is that the mean kinetic energy of the electron in the hydrogen atom. . 2 |ˆ) for m = ± 1 . 1 |ˆ) for m = ± 1 . . . .2: Binding energies for the hydrogen (Z = 1) atom. = 0.405).60601 -3. 1.3. much less than the rest mass of the electron (511 keV).551178 ⇑ . n − 1 j = 1 2 1 2 ± m = −j.551177 ⇑ . notation 1s 1 2 non-rel. in fact.2 the non-relativistic [cf. However.1. .1. One can combine the expressions (10. . 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 (1.036 by means of Eqs.40146 ⇑ ⇑ -1. .e. . relativistic.405) in terms of the non-relativistic quantum numbers n. the 2p 3 states with r 2 2 2 2 1 involve four degenerate states corresponding to Y 3 m (1.

. The energies were evaluated with m = 511. j.1 ⇑ ⇑ ⇑ ⇑ rel. m) ≈ m − mZ 2 e4 mZ 4 e8 − 2n2 2n3 1 j+ − 3 4n + O(Z 6 e12 ) .1 -34. j 1 2 1 2 1 2 3 2 1 2 1 2 3 2 3 2 5 2 Relativistic Quantum Mechanics non-rel. to estimate the diﬀerence between the energies of the states 2p 3 and 2p 1 (cf. mom.0041 keV and e2 = 1/137. 0) (10.15. 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. An example for such potential is the Coulomb potential V (r) = −Ze2 /r considered further below.2.406) α √ 1+ (n − β + β 2 −α)2 The expansion (10. 0.036 by means of Eqs.405).9 ⇑ . Introducing α = Z 2 e4 1 and β = j + 2 (10.3).8 ⇑ . 0 0 1 1 0 1 1 2 2 spinorbital ang.407) 1 2 This expression allows one.409) where V (r) is spherically symmetric. We assume for the wave function the stationary state .402. binding energy / keV Eq.6 -42.2 -17.405 ) -161.0 ⇑ ⇑ -15.206) provides in the present case ED (n. In order to investigate further the deviation between relativistic and non-relativistic spectra of hydrogen-type atoms we expand the expression (10. 0. (10. 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) to order O(Z 4 e8 ).15. 32 (10. It holds for n = 2 and j = 3 .3: Binding energies for the hydrogen-type (Z = 100) atom. for example. Tables 10. (10. 10.405) reads 1 (10. binding energy / keV Eq.10.35.1 ⇑ .402) -136. (10. Degeneracies are denoted by ⇑. (10.342 main quantum number n 1 2 2 2 3 3 3 3 3 orbital angular mom.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).3 spectr.

10.10: Dirac Particles in Electromagnetic Field form Ψ(xµ ) = e−i t 343 Φ(r) X (r) .5. in particular. 1 |ˆ) r 2 2 = i ∂r + −j r 1 g(r) Yjm (j + 2 .417) c(r) d(r) X (r) 1 1 1 1 Yjm (j + 2 . 2 |ˆ) r Φ(r) r .411) (10.413) ˆ σ · p g(r) Yjm (j − 1 . r r 2 2 2 a(r) b(r) 1 1 1 1 Yjm (j + 2 . 1 |ˆ) introduced in r 2 2 1 Sect. according to (10. 6. 2 |ˆ) r r r .413. 1 |ˆ) r 2 2 (10. The arguments above allow one to eliminate the angular dependence by expanding Φ(r) and X (r) 1 in terms of Yjm (j + 2 .414) ˆ σ · p rf (r) Yjm (j + 1 . 1 |ˆ) . 2 |ˆ) + r Yjm (j − 2 .412) ˆ In this equation a coupling between the wave functions Φ(r) and X (r) arises due to the term σ · p. r 2 2 (10.198)] not couple states with diﬀerent j. 1 |ˆ) r 2 = i ∂r − 1 2 r g(r) Yjm (j + 1 .414) These equations can be brought into a more symmetric form using ∂r + which allows one to write (10.197. ˆ σ · p X + m Φ + V (r) Φ ˆ σ · p Φ − mX + V (r) X = = Φ X (10. (10. 6. σ · p ˆ σ · p f (r) Yjm (j + 1 . = r (10. (6.415) 1 ˆ σ · p rg(r) Yjm (j − 2 . i. 6. but are timeindependent. 168]: the term is a scalar (rank zero tensor) in the space of the spin-angular momentum states Yjm (j ± 1 . r 2 (10.412) are four-dimensional with r-dependent wave functions. 10. 1 |ˆ).410) 1 where Φ(r). i. 10.411.416) The diﬀerential equations (10. 2 |ˆ) + r Yjm (j − 2 .5 [see.232.e. pp. 1 |ˆ) r 2 2 = i ∂r + j+ r 1 2 3 2 f (r) Yjm (j − 1 .e. X (r) ∈ C2 describe the spatial and spin.2 degrees of freedom. 1 |ˆ) r 2 1 1 = ∂r r r r (10.. 1 |ˆ) . This term has been discussed in detail in Sect. 1 |ˆ) r 2 2 = i j+ ∂r + r j+ r 1 2 1 r f (r) Yjm (j − 2 .345). 2 |ˆ) and does r 2 ˆ has odd parity and it holds [c.f.10. 1 |ˆ) and Yjm (j − 1 . m-values. The Dirac equation reads then.. the term is block-diagonal in the space spanned by the states Yjm (j ± 1 .

such expansion must include states with all possible j. there exist two independent solutions (10. 1 |ˆ) r Φ(r) 2 2 r = (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. only a(r).417) of the form f1 (r) i Yjm (j + 1 . d(r) are coupled and b(r).421) g2 (r) X (r) 1 1 − Yjm (j + 2 . According to (10. 2 |ˆ) r r f2 (r) 1 1 r Φ(r) i r Yjm (j − 2 .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. the orthonormality property (6. (10. we consider the case that only states for one speciﬁc j.157). 2 |ˆ) = (10. g1 (r) ∂r − j+ r j+ r 1 2 g1 (r) + [ − m − V (r) ] f1 (r) = 0 ∂r + and for f2 (r).417) into (10. Inserting (10. 10.423) .418) holds for f1 (r).419) Obviously. using (10. m values. 2 |ˆ) r r where the factors i and −1 have been introduced for convenience.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 (10.344 Relativistic Quantum Mechanics In general.411. m pair contribute.412). Presently. 10. c(r) are coupled.415.420) g1 (r) X (r) 1 1 − Yjm (j − 2 . g2 (r) ∂r + 1 2 f1 (r) − [ + m − V (r) ] g1 (r) = 0 (10.416). Accordingly.

409). Hence. except for the opposite sign of the term (j + 1 ). 2 2 2 2 2 2 = j − 1 2 and.421).. etc. According to the discussion of the spectrum (10. . 1 |ˆ) r i 2 2 .422.420) is the large component and X is the small component.422) corresponds to states f1 (r) Yjm (j + 1 . (10. the ones given in (10. determining wave functions of the type (10. 2. correspodingly the states We consider ﬁrst the solution of (10.Wave Functions We want to determine now the wave functions of the stationary states of hydrogen-type atoms which correspond to the energy levels (10. 10.422) describes the states 2p 1 . Dirac Particle in Coulomb Field . 2s 1 . 2p 3 . . 1.367) which is spherically symmetric such that equations (10. (10. .422) follows in this case from the same procedure as that adopted for the radial wave function of the non-relativistic hydrogen-type atom. in the non-relativistic limit wave functions f2 (r) 1 1 i Yjm (j − 2 . Equation (10. . and obtains ﬁnally a polynomial function p(r) such that rγ exp(−µr)p(r) solves (10. We assume the 4-vector potential of pure Coulomb type (10.405). 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. 3s 1 .425) r 0 covers states with angular momentum 1s 1 .10: Dirac Particles in Electromagnetic Field 345 Equations (10. . 2 |ˆ) r Ψ(xµ ) ≈ . Φ in (10. 3p 3 . . etc. = 0. . 2 2 2 (10.422).422).e.422) near r = 0. i. n − 1. the 2 equations determine. to states with angular momentum = j + 2 . In the non-relativistic limit.405). .e.423) are identical.423) apply for V (r) = −Ze2 /r. 3p 1 . Similarly. 3d 5 . one demonstrates then that the wave functions for r → ∞ behaves as exp(−µr) for some suitable µ. together with the appropriate boundary conditions at r = 0 and r → ∞.10. (10. The solution of (10.423). (10. for small r.424) Ψ(xµ ) ≈ r 0 1 i.405) of the relativistic hydrogen atom the corresponding states have quantum numbers n = 1. Behaviour at r → 0 We consider ﬁrst the behaviour of the solutions f1 (r) and g1 (r) of (10. According to this procedure.422) and (10. one demonstrates ﬁrst that the wave function at r → 0 behaves as rγ for some suitable γ. 3d 3 . the radial wave functions for Dirac particles in the potential (10. Hence. enforcing the polynomial to be of ﬁnite order leads to discrete eigenvalues . (10..422).420).422) determines solutions of the form (10.426) . We note that (10. namely.

we conclude f1 (r) →∞ ∼ e−µr . hence. becomes imaginary.431) inﬁnite since. hence. assume the r-dependence in (10. 10. (10.427) with 1 γ = (j + )2 − Z 2 e4 .427). µ = m2 − 2 (10.420).434) √ The solutions of these equations are f1 .427) = = 0 0.429) This equation poses an eigenvalue problem (eigenvalue −γ) for proper γ values.422) becomes ∂r g1 (r) ∂r f1 (r) Iterating this equation once yields 2 ∂r g1 (r) 2 ∂r f1 (r) 1 . in case j + 1 )2 < Ze2 .435) . Only the exponentially decaying solution is admissable and. for the particle density holds then ρ(r) ∼ |rγ−1 |2 = Behaviour at r → ∞ For very large r values (10.427) makes only the positive solution 2 possible. 10. We have. g1 (r) →∞ ∼ e−µr . determined that the solutions f1 (r) and g1 (r). ) f1 (r) ( + m ) g1 (r) (10.430) 2 Note that the exponent in (10. The assumed r-dependence in (10. for small r.267. Such r-dependence 2 would make the expectation value of the potential r2 dr ρ(r) 1 r (10. (10.433) = = ( m2 − 2 ) g1 (r) ( m2 − 2 ) f1 (r) (10. r2 (10. g1 ∼ exp(± m2 − 2 r). One obtains γ = ± (j + 1 )2 − Z 2 e4 .432) = = − ( − m. g1 (r) →0 ∼ b rγ (10. (10. according to (10.428) = 0.266.346 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 .

436) f1 (r) = Insertion into (10. without loss of generality we can choose ˜ f1 (r) = φ1 (r) + φ2 (r) .s.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 .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. Equation (10.437) where they ˜ ˜ cancelled in case f1 = g1 = a.444) (10. In the present case the functions f1 and g1 cannot be chosen ˜ ˜ identical due to the terms in the diﬀerential equations contributing for ﬁnite r.440) (10. hence. Accordingly.10. However. Solution of the Radial Dirac Equation for a Coulomb Potential To solve (10. ρ m+ ρ [∂ρ − (10. (10. we set f1 (r) g1 (r) = = √ √ ˜ m + e−µr f1 (r) −µr = = 0 0 (10.437) (10. g1 (r) = φ1 (r) − φ2 (r) ˜ (10.439) − m− e g1 (r) ˜ where µ is given in (10.440) and (10.441) The last two terms on the l.441) correspond to (10. √ g1 (r) = − m − a e−µr .422) for the Coulomb potential V (r) = −Ze2 /r We assume a form for the solution which is adopted to the asymptotic solution (10. (10.435 ). of both (10.h.438) (10.434) results in √ √ (m − ) m + a − (m − ) m + a √ √ (m + ) m − a − (m + ) m − a which is obviously correct.10: Dirac Particles in Electromagnetic Field For bound states holds 347 < m and.442) which leads to a partial cancellation of the asymptotically dominant terms.436). Let us consider then for the solution of (10. We also introduce the new variable ρ = 2µ r . µ is real.445) .434) √ m + a e−µ r .

452) From (10.447) We seek solutions of (10.449) into (10. 10.348 Relativistic Quantum Mechanics Addition and subtraction of these equations leads ﬁnally to the following two coupled diﬀerential equations for φ1 and φ2 ∂ρ φ1 + ∂ρ φ2 + j+ ρ 1 2 Ze2 mZe2 φ1 − √ φ2 = 0 m2 − 2 ρ m2 − 2 ρ mZe2 Ze2 φ1 + √ φ1 + √ φ2 − φ2 = 0 m2 − 2 ρ m2 − 2 ρ j+ ρ φ2 − √ 1 2 (10.448.446.451) follows βs−1 = = Ze2 s+γ + √ m2 − s+γ − βs + m2 Z 2 e4 m2 − 2 − (j + 1 )2 2 2 √ Ze m2 − 2 2 s+γ − βs (10. = m− Ze2 βs s + γ − √m2 − 2 2 (10.454) .430)].448) φ2 (ρ) = ρ (10. (10. (10.450) −√ mZe2 m2 − 2 βs ρs+γ−1 = 0 mZe2 m2 − αs (s + γ) β2 + (j + 1 ) αs + √ 2 s 2 +√ From (10.450) follows Ze2 m2 − 2 βs − βs−1 =0 (10. 10.449) for γ given in (10.447) leads to (s + γ) αs + (j + 1 ) βs − √ 2 s Ze2 m2 − 2 αs (10.447) of the form n φ1 (ρ) = ργ s=0 n γ s=0 αs ρ2 βs ρ2 (10.446 .430) which conform to the proper r → 0 behaviour determined above [c.f. Inserting (10.426–10.446) (10.451) 1 √mZe − (j + 2 ) 2 2 αs . 10.453) (s + γ)2 + Z 2 e4 − (j + 1 )2 2 2 √ Ze m2 − 2 βs .430) one can write this βs = s+γ − 2 √ Ze m2 − 2 s (s + 2γ) βs−1 . Using (10.

459) − 2 √mZe m2 − 2 (10. 2γ + 1. with the associated Laguerre polynomials L(α) = F (−n. 10. 10. 10.455) this conﬁnement of so implies discrete values for .10: Dirac Particles in Electromagnetic Field Deﬁning so = √ one obtains βs = = . for the states 2p 1 . 2. equivalently. hence. (10. 10. c. .e. α + 1. .457) with the conﬂuent hypergeometric functions F (a. . x) .461) In order that the wave functions remain normalizable the power series (10. 3p 1 . We can. 3d 3 holds n = 1. 1.458) or. The expressions (10. n It holds φ1 (ρ) φ2 (ρ) = = ργ F (−so .461 for values given by (10. 2. According to the deﬁnitions (10. (s − so ) β0 s! (2γ + 1)(2γ + 2) · · · (2γ + s) (10. m (n ) = .430. . (s − so ) β0 s! (2γ + 1)(2γ + 2) · · · (2γ + s) (10.. 2γ + 1.456) and (10. in fact. ρ) . β0 j+ 1 2 (10. c c(c + 1) 2! (10. so = n . conclude 2 2 2 that the polynomials in (10.448. This requires that all coeﬃcients αs and βs must vanish for s ≥ n for some n ∈ N. x) = 1 + a a(a + 1) x2 x + + .405) allows one to identify n = n − j + 2 which.449) must be of ﬁnite order.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. . i.452) follows j+ αs = 1 2 349 Ze2 m2 − 2 −γ (10. .457) One can relate the polynomials φ1 (ρ) and φ2 (ρ) deﬁned through (10. is an integer. .455) are ﬁnite. .. For example.460) (10. ρ) so β0 ργ F (1 − so . Comparision with (10.456) − so 2 √mZe m2 − 2 (−so )(1 − so )(2 − so ) . .405) and the ensuing so values ( 10. n = 0. . 1. = From (10. namely..405).449) and (10.10.456.448.455) 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 ) .457) for βs and αs imply that so must then be an integer.

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 .463) where Φ and X . 1990).421)] ∞ dr (|f1 (r)|2 + |g1 (r)|2 ) = 1 (10.465) (10.439).421) with radial wave functions f1 (r).157.463) and yield [note the 1/r factor in (10. j. 2γ + 1. The normalization integral is then ∞ 0 π 2π r2 dr 0 sin θdθ 0 dφ (|Φ(r)|2 + |X (r)|2 ) = 1 (10.438.350 Relativistic Quantum Mechanics Altogether we have determined the stationary states of the type (10.421) correspond to non-relativistic states with orbital angular momentum 1 1 = j + 2 . 6. 2γ + 1.469) This formula has been adapted from ”Relativistic Quantum Mechanics” by W.442).158) of the latter states absorb the angular integral in (10. The wave functions (10. as in (10.460. (10.410) of the stationary state wave function the density ρ(xµ ) of the states under consideration. 2µr) (2µ) 2 Γ(2γ + 1) 3 N = and m 4m )Γ(2γ + n + 1) (n +γ)m (n +γ)m (10. 1 |ˆ) 2 2 r G1 (r) Yj. Sect. 2µr) (10. 1 |ˆ) in (6.460. is time-independent.m (j + 1 . 10.461). can follow the procedure adopted for the wave functions of the non-relativistic hydrogen atom and will not be carried out here. 9. 10.464) 0 The evaluation of the integrals. given by expression (10. m. 10. 1 |ˆ) 2 2 r κ = j+ 1 2 (10. Greiner.147.460. (10.466) F1 (r) = F− (κ|r) . where2 F± (κ|r) = G1 (r) = F+ (κ|r) . (n + γ)m N (2µr)γ−1 e−µr − κ F (−n .m (j − 1 . The coeﬃcients β0 in (10. Due to the form (10. The orthonormality 2 2 r properties (6.467) ± n F (1 − n . 10. and (10.421).267). They are described through quantum numbers n. j. are two-dimensional vectors determined through the explicit form of the spin-orbital angular momentum states Yjm (j ± 1 . 2 . The complete wave function is given by the following set of formulas 1 Ψ(n.461).6. 6. which involve the conﬂuent hypergeometric functions in (10. m|xµ ) = e−i t iF1 (r) Yj. (Springer. = j + 2 . g1 (r) determined by (10. = j + 2 .148).461) are to be chosen to satisfy a normalization condition and to assign an overall phase. Berlin.

3s 1 . not not 2 2 2 2 2 2 covered by the wave functions given by (10. 3d 5 . We have. tracing all steps which lead from (10.m (j − 1 . become f2 (r) 1 1 i Yjm (j − 2 .465–10. G2 (r) = F+ (κ|r) . where F± (κ|r) are as given in (10.421) which. 2 |ˆ) r Ψ(xµ ) ≈ e−1 t .469) that the following wave functions result Ψ(n. obtained closed expressions for the wave functions of all the stationary bound states of relativistic hydrogen-type atoms.471) (10. describes the complementary set of states 1s 1 .469).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. 1 |ˆ) 2 2 r κ = −j − 1 2 (10.467–10. 2s 1 . 1 |ˆ) 2 2 r G2 (r) Yj. . 2 accordingly.421) are governed by the radial Dirac equation (10.10: Dirac Particles in Electromagnetic Field 351 We want to consider now the stationary states of the type (10. 2p 3 . this wavefunction has an orbital angular momentum quantum number = j − 1 and. 3p 3 .470) r 0 Obviously. = j − µ 1 2 .422) to (10. hence.m (j + 1 .10. j. m|x ) = e −i t iF2 (r) Yj. in the non-relativistic limit. etc. (10.472) F2 (r) = F− (κ|r) . The radial wave functions f2 (r) and g2 (r) in (10.469). One can verify.

352 Relativistic Quantum Mechanics .

may be exploited to express the Lorentz-invariant equations of relativistic quantum mechanics. ϑ). a(z) and b(z). Starting point is the Lorentz transformation S(w. We will proceed by building as much as possible on the results obtained sofar in the chiral representation of the Dirac equation.262). ˜ We have altered here slightly our notation of S(w. ϑ through 3. must be two-dimensional irreducible representations of the Lorentz group.Chapter 11 Spinor Formulation of Relativistic Quantum Mechanics 11.1) (11. This is. what will be achieved in the following.. in particular. 1 353 . We will characterize the space on which the transformations a(z) We will see below that the representations a(z) and b(z) are. in fact. The lower dimensionality of a(z) and b(z) implies. 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. i. expressing its dependence on w. ϑ)µ ν correspond to diﬀerent a(z) and b(z).3) (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.229). a complex variable z. z ∈ C ˜ Because of its block-diagonal form each of the diagonal components of S(z). diﬀerent L(w.2) (11.e. i.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. indeed.e.225–10.. in a sense. This transformation bispinor wave function Ψ can be written ˜ S(z) a(z) b(z) z = = a(z) 0 0 b(z) exp 1 z·σ 2 1 − z ∗· σ 2 (11. ϑ) for the ˜ in the chiral representation as given by (10. isomorphic to the natural representation. ϑ)µ ν .4) = exp = w − iϑ .

to the rotational transformations of spin.e. + 1 2 2 state 2 ∼ | 1 . − 1 2 2 z = iϑ. ϑ ∈ R3 (11.6) as given by (5. a distinct diﬀerence in the transformation behaviour of ˜ ˜ ˜ ˜ Ψ1 (xµ ).9) a∗ (z) = exp 2 . ϑ ∈ R3 . Relationship Between a(z) and b(z) The transformation b(z) can be related to the conjugate complex of the transformation a(z). 0)T the transformations are (1) a(i β e2 ) = b(i β e2 ) = ˆ ˆ 2 dmm (β) (1) = cos β 2 sin β 2 − sin β 2 cos β 2 . Ψ2 (xµ ) and Ψ3 (xµ ). a space of vectors state1 for which holds state2 state 1 ∼ | 1 . Aν . i. A First Characterization of the Bispinor States We note that in case w = 0 the Dirac transformations are pure rotations. namely. the operator ∂µ and the Pauli matrices σ in the new representation and.. usually expressed 2 as product of rotations and of functions of Euler angles α.5) The transformations in this case.354 Spinor Formulation and b(z) act. Here | 1 . will establish the map between L(w.1 states as described by Dm m (ϑ).e. b(z). 2 2 2 ˜ Since a(z) acts on the ﬁrst two components of the solution Ψ of the Dirac equation. neutrino equation. and since b(z) ˜ ˜ acts on the third and fourth component of Ψ one can characterize Ψ ˜ Ψ1 (xµ ) ˜ Ψ2 (xµ ) Ψ3 (xµ ) ˜ ˜ Ψ4 (xµ ) ∼ ∼ ∼ ∼ 1 |2. For ϑ = (0. i. express 4-vectors Aµ . +1 2 2 |1. γ (see Chapter 5). ϑ ∈ R3 .. are elements of SU(2) and correspond. −1 2 2 z = iϑ. in 2 fact. ϑ)µ ν and a(z). for z = iϑ. This characterization allows one to draw conclusions regarding the state space in which a(z) and b(z) operate. −1 2 2 |1. +1 2 |1. (11. (11. ﬁnally. and the Klein–Gordon equation in the spinor representation. the so-called spinor space. β. however.8) We like to stress that there exists. In this case a(z) and b(z) are identical and read a(i ϑ) = b(i ϑ) = exp 1 − ϑ·σ 2 . formulate the Dirac equation. In this case holds a(z) = b(z) ˜ 1 (xµ ). β. Ψ2 (xµ ) transform according to a(z) whereas Ψ3 (xµ ). ± 1 represents the familiar spin– 1 states. (11. to 1 ∗ ∗ z ·σ (11. Ψ4 (xµ ) in case z = w + iϑ for w = 0. θ ∈ R3 .243). Ψ4 (xµ ) transform according to ˜ ˜ ˜ and Ψ b(z).7) where “∼” stands for “transforms like”.

that the transformation (11. however. 355 (11. a result to be used further below. but leave rotation angles ϑ unaltered. 11.4) and . Ψ (xµ ) as well as Ψ (xµ ).12) To prove (11. µ ). −1 given by (11.11) one ﬁrst demonstrates that for and −1 as given in (11.f. (11. ∗ σ 3 = σ3 . (11. in fact. . 11. one can show −1 −1 −1 = −σ1 = σ2 = −σ3 ∗ = −σ1 ∗ = −σ2 σ2 −1 (11.12) yields σ1 σ3 or in short σ Similarly.14) .12).13) Explicit matrix multiplication using (5.17) We conclude.11) 0 −1 1 0 = . −1 a∗ (z) −1 (11. therefore. 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.16) σ = −σ ∗ . Ψ (xµ ) even though these pairs of ˜ ˜3 ˜4 featuring the components Ψ1 (x ˜ 2 components transform independently of each other.15) (11.12) does. −1 = ∗ −σ3 = −σ ∗ .224)] ∗ σ1 = σ1 . 1.11. This demonstrates that a(z) is the ˜ transformation which characterizes both components of S(z). (11.1: Lorentz Transformation of Dirac Bispinor ∗ ∗ ∗ where σ ∗ = (σ1 . Spatial Inversion ˜ One may question from the form of S(z) why the Dirac equation needs to be four-dimensional.1) can be written in the form ˜ S(z) = a(z) 0 0 a∗ (z) −1 (11.224. One can readily verify [c. hold −1 = 1 One notices then. 11. ∗ σ 2 = − σ2 . The eﬀect of inversion on Lorentz transformations is.2. that they alter w into −w.10.10) From this one can derive b(z) = where = 0 1 −1 0 . a∗ (z) −1 = exp 1 ∗ z σ∗ 2 −1 . Hence. σ3 ). one can express a∗ (z) −1 = exp 1 − z ∗· σ 2 = b(z) . (5.18) with a(z) given by (11. (11. σ2 . using f (a) −1 = f ( a −1 ).

2 Relationship Between the Lie Groups SL(2. Part (Elsevier.. the transformations a(z) and b(z) become interchanged. 3 . M is a complex 2 × 2–matrix. One can verify this by evaluating the determinant of a(z) det( a(z) ) = det e 2 z·σ 1 (11. (5.23) Exercise 11. does not suﬃce for particles like the electron which obey inversion symmetry. For the general case the proof.20) ˜ Obviously. ˜ ˜ P S(w + iϑ) P = S(−w + iϑ) = b(z) 0 0 a(z) .356 Spinor Formulation ˜ Let P denote the representation of spatial inversion in the space of the wave functions Ψ. hence. det(M ) = 1 } . (11. which describes pure rotations. the Lorentz invariant equation for neutrinos is only 2–dimensional. (11.24) = etr(M ) (11.3. non-singular matrix M holds2 det eM and from [c. a(w + iϑ) for w = 0 is an element of SL(2.e. for particles like the neutrinos which do not obey inversion symmetry two components of the wave function are suﬃcient.. ϑ ∈ R3 . j = 1.C) and SO(3.e.1: Show that SL(2.22) which follows from the fact that for any complex.19) i. de Kerf Lie a Algebras. (11.G.1) We have pointed out that a(iϑ). 2 = 1 i. C) = { M. The proof of this important property is straightforward in case of hermitian M (see Chapter 5).2.224)] tr( σj ) = 0 . However.10. is an element of SU(2). can be found in G. B¨uerle and E. the space spanned by only two of the components of Ψ is not invariant under spatial inversion and. The transformation S(z) in the transformed space is then ˜ P 1. However.A. Obviously. 2 . 1990).f. P −1 = P.21) = etr( 2 z·σ) = 1 1 (11. Amsterdam.21) together with matrix multiplication as the binary operation forms a group. Exercise 1. 11. C) deﬁned in (11. In fact.A. based on the Jordan–Chevalley theorem. 2. This implies ˜ Ψ1 ˜ Ψ2 P ˜ Ψ3 ˜ Ψ4 ˜ Ψ3 ˜ Ψ4 = Ψ1 ˜ ˜ Ψ2 .

11. 0 1 1 0 . M (Aµ ) according to (11.32) . We will prove below [cf. σ1 ..2: Lie Groups SL(2. σ1 .29) where we used (10.C) and SO(3. in a transformed frame should hold Lρ ν σ µ Aµ σµ A µ .28) σµ = 1 0 0 1 . In fact. 1 0 0 −1 . in fact.e. σ2 . σ3 are hermitian.25) of the matrix M (Aµ ) is independent of the frame of reference.76).31) Exercise 11.31) holds.25) σ0 σ1 σ2 σ3 The quantity σµ thus deﬁned does not transform like a covariant 4–vector. Demonstrate that (11.25) can also be written M (Aµ ) = A0 + A3 A1 − iA2 A1 + iA2 A0 − A3 .2.30) Since the components of Aµ are real. Argue why M (Aµ ) = σµ Aµ provides a bijective map. (11.30). (11. the matrix M (Aµ ) is hermitian as can be seen from inspection of (11.1) Mapping Aµ onto matrices M (Aµ ) 357 We want to establish now the relationship between SL(2. Starting point is a bijective map between R4 and the set of two-dimensional hermitian matrices deﬁned through M (Aµ ) = σµ Aµ where (11. (11.27) −→ Straightforward transformation into another frame of reference would replace σµ by σµ .26) (11.135)] this transformation behaviour. i. (11.75) holds aν = Lν µ aµ one realizes that σµ transforms inversely to covariant 4–vectors. one wishes that the deﬁnition (11. σ3 provide a linear–independent basis for the space of hermitian 2 × 2–matrices.26) or from the fact that the matrices σ0 . C) and the group L↑ of proper. Using Aµ = (L−1 )µ ν A ν one would expect in a transformed frame to hold Lρ ν σ (L−1 )µ ν A ν .28) and (11. The function M (Aµ ) is bijective. −→ µ Consistency of (11. (11.2: Show that σ0 . The following important property holds for M (Aµ ) det ( M (Aµ ) ) = Aµ Aµ which follows directly from (11.27) requires then σ µ Aµ Lν µ σµ = σν (11. or+ thochronous Lorentz transformations. one can provide a simple expression for the inverse of M (Aµ ) M = M (Aµ ) ↔ Aµ = 1 tr M σµ 2 . Noting that for covariant vectors according to (10. σ2 . (11. 0 −i i 0 .

31)] 1 a Aµ −→ A µ = tr a M (Aµ ) a† σµ 2 Because of (11.358 Transforming the matrices M (Aµ ) Spinor Formulation We deﬁne now a transformation of the matrix M (Aµ ) in the space of hermitian 2 × 2–matrices a M −→ M = a M a† . for any a ∈ SL(2.31).2. (11. C) deﬁnes the transformation [c. (11. L(a)µ ν = 1 tr a σν a† σµ 2 .39) Exercise 11. The linear character of the transformation becomes apparent expressing A µ as given in (11.33) This transformation conserves the hermitian property of M since (M )† = ( a M a† )† = (a† )† M † a† = a M a† = M (11.33) det(M ) = = det( a M a† ) = det(a) det(a† ) det(M ) [det(a)]2 det(M ) = det(M ) .40) is an element of L↑ . C) .36) The suitable A µ can readily be constructed using (11. (11.35) We now apply the transformation (11.35) holds for this transformation A µ Aµ = Aµ Aµ (11.37) using (11.37) deﬁnes actually a Lorentz transformation. any a ∈ SL(2. Due to det(a) = 1 the transformation (11. a ∈ SL(2.33) to M (Aµ ) describing the action of the transformation in terms of transformations of Aµ .40) 1 tr a σν a† σµ 2 Aν (11.25) Aµ = which allows us to express ﬁnally A µ = L(a)µ ν Aν .3: Show that L(a)µ ν deﬁned in (11.34) where we used the properties M † = M and (a† )† = a.33) conserves the determinant of M .38) . Accordingly. In fact. C) and for any Aµ there exists an A µ such that M (A µ ) = a M (Aµ ) a† . 11. it holds for the matrix M deﬁned through (11. (11.32.37) which implies that (11. (11.f. (11. + . In fact.

(11.41) 1 tr a1 σν a† σµ 1 2 1 tr σν a† σµ a1 1 2 1 tr a2 σρ a† σν 2 2 1 tr a2 σρ a† σν 2 2 . β 2.2: Lie Groups SL(2.1) L(a)µ ν provides a homomorphism 359 We want to demonstrate now that the map between SL(2. β δ = δ = = δ = δ 1 2 = 2 = 1 Aαβγδ (11. 1 ¯ and using the deﬁnition of Lµ ρ in (11.47) This completes the proof of the homomorphic property of L(a)µ ν .41) results in 1 ¯ Lµ ρ = 4 where Aαβγδ = ν Γ = a2 σρ a† 2 (11. .β γ.11.C) and SO(3.42) = ν Deﬁning Γ = a† σµ a1 . (11.C) )µ ρ (11.46) which yields ¯ Lµ ρ = = = 1 1 Γ11 Γ11 + Γ22 Γ22 + Γ12 Γ21 + Γ21 Γ12 = tr ΓΓ 2 2 1 1 tr a† σµ a1 a2 σρ a† = tr σµ a1 a2 σρ a† a† 1 2 2 1 2 2 1 tr a1 a2 σρ (a1 a2 )† σµ = L(a1 a2 )µ ρ . C) and SO(3..δ (11. 2 (11.α.δ 1 4 Aαβγδ Γβα Γγδ α.40)] respects the group property of SL(2. i. C) and of SO(3. ¯ Lµ ρ = L(a1 )µ ν L(a2 )µ ρ = L( product in SO(3.44) (σν )αβ (σν )δγ .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.β γ.1).1) deﬁned through L(a)µ ν [cf.e.43) (σν )αβ Γβα Γγδ (σν )δγ = ν. (11.45) One can demonstrate through direct evaluation 2 α = 2 α = 2 α = = 2 α = 0 else β β γ γ = = = = γ = γ = 1.

(11. One expects then that six generators Gj and six real coordinates fj can be deﬁned which allow one to represent a in the form 6 a = exp j=1 fj Gj . To this end we consider inﬁnitesimal transformations and keep only terms of zero order and ﬁrst order in the small variables. correspondingly. correspond to the generators given by (10. i. To obtain the generator of a(z) corresponding to the generator K1 . 1. 0. 0.50) (11.52) and (11. 10.51) ( 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) − σ1 σ0 0 0 One can verify readily that 1 κ1 = − σ 1 2 (11. Aµ = (0.49) assuming (note that g µ ν is just the familiar Kronecker δµν ) Lµ ν a = = gµν + 1 + 1 (K1 )µ ν κ1 . C) as complex 2 × 2–matrices are deﬁned through four complex or. (11.2) which correspond to the generators K1 .h.. Because of the condition det(a) = 1 only six independent real numbers actually suﬃce for the deﬁnition of a.h. M (Aµ + (K1 )µ ν Aν ) = = M (Aµ ) + M (Aµ ) − M ( (−A1 . we write (11.54) This reads for the four cases Aµ = (1. Aµ = (0.48). (11. J3 of the Lorentz transformations Lµ ν in the natural representations.57) .e.52) where we employed (11. 0. 0. eight real numbers. Aµ = (0. denoted below as κ1 . 1. K2 .49). 0).58) . 0) ) σ0 A 1 − σ1 A 0 (11.360 Generators for SL(2. 0).51) Insertion of (K1 )µ ν as given in (10. C) which correspond to K. of (11. 0).56) (11. For the r. 0. J Spinor Formulation The transformations a ∈ SL(2. 0.36) M (Lµ ν Aν ) = a M (Aµ ) a† (11.55) (11.s. of (11.25) in the last step. 0. 1 (11.s. J2 .53) results in the condition σ 0 A1 − σ 1 A0 = ( κ 1 σ µ + σ µ κ † ) Aµ .48) We want to determine now the generators of the transformation a(z) as deﬁned in (11.59) = = = = κ1 σ 0 + σ 0 κ† = κ 1 + κ† 1 1 κ1 σ 1 + κ1 σ 2 + κ1 σ 3 + σ 1 κ† 1 σ 2 κ† 1 σ 3 κ† 1 (11.53) Equating (11. K3 . 1 (11. noting the linearity of M (Aµ ). J1 .49) we obtain using (11. −A0 .48) yields for the l.47.

the so-called contravariant spinors. 2 λ = i σ. K3 and λ1 . J3 are given by 1 κ = − σ.61) This identiﬁes the transformations a(z = w − iϑ) as representations of Lorentz transformations. one can show that the generators κ2 . J2 . a(w − iϑ) = e− 2 (w − iϑ)·σ ∈ SL(2.62) We denote the general (z = w + iϑ) transformation by φ α = aα β φβ = aα 1 φ1 + aα 2 φ2 .64) . ϑ) = ew·K + ϑ·J ∈ SO(3. def α = 1.60) of a ∈ SL(2. 2 2 φ1 φ2 ∈ C2 . C) correspond to the generators K and J of Lµ ν . a1 1 a1 2 a2 1 a2 2 (11. Here a(z) = ( aα β ) = describes the matrix a(z). 2 (11. Exercise 11. C) acts on spinors φα ∈ C2 (characterized below) 1 (11. hence.11. 2 (11. κ3 of a(z) corresponding to K2 .3: Spinors 361 obeys these conditions. + 1 2 2 1 transforms under rotations (z = iϑ) like a spin– 2 state | 1 .2. 11. λ3 corresponding to J1 . 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 . Similarly.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. λ2 .1) acts on 4–vectors Aµ . (11. − 1 .4: Show that the generators (11.60) We can.63) where we extended the summation convention of 4-vectors to spinors. w describing boosts and ϑ describing rotations. state that the following two transformations are equivalent L(w.

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

73) (11.11.70) is φ1 φ2 −1 363 = = −1 φ1 φ2 .76) αβ αβ The scalar product (11.69.70) and (11.80) One can write in matrix notation aα β γ δκ αγ a δ αβ (11.74)]. (11.63.1: Show that for any non-singular complex 2 × 2–matrix M holds M −1 = det(M ) M −1 T The matrices .3: Spinors as can be veriﬁed from (11.71) (11.75) is given by φα = as can be readily veriﬁed.72) 0 −1 1 0 Exercise 11. 11.72) in a notation analogous to that chosen for contravariant and covariant 4–vectors [cf.10.68).73.78) aβ γ γδ φδ (11. (11. The inverse of (11.79) = a −1 T a κβ .77) is Lorentz invariant.75) (11. 10.74) = = 0 1 −1 0 0 −1 1 0 (11. (11. For this scalar product holds φα χα = −χα φα . In the transformed frame holds φ α χα = aα β αγ aγ δ δκ φβ χκ .77) (11.72)] φα φ α = = = = αβ αβ φβ φβ 11 21 11 21 12 22 12 22 (11. 11. −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. The transformation behaviour of φα according to (11. (10. g µν of the Minkowski space [cf.71.67) will be expressed as φα χα = φ1 χ1 + φ2 χ2 = φ1 χ2 − φ2 χ1 . (10. 11.69.81) .3. we will express (11. Accordingly. 11. 11.

82) and with f (A)T = f (AT ) for polynomial f (A) a −1 T = e− 2 z·(σ 1 ∗ )T = e− 2 z·σ = a−1 1 (11.86) A concise notation of the conjugate complex spinors is provided by ˙ (φα )∗ = φα = φ1 ˙ φ2 ˙ ˙ (11.81) yields aα β and.63).88) reads ˙ ˙ φ α = aα β φβ ˙ ˙ (11. (11.83) into (11. ˙ (11.84) (11. i.90) extending the summation convention to ‘dotted’ indices. We denote ˙ (aα β )∗ = aα β ˙ (11.80) φ α χα = φβ χβ . Hence..85) φ1 φ2 ∗ ∗ .364 Using (11. Obviously. σ † = (σ ∗ )T = σ. ˙ We also deﬁne covariant versions of φα φα = (φα )∗ .91) . from (11. hence. As discussed above. a∗ (z) provides a representation of the Lorentz group which is distinct from that provided by a(z). k = 1. Insertion of (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.e.87) which we will employ from now on.2) and (11. the ˙ conjugate complex spinors φα need to be considered separately from the spinors φα .83) Here we have employed the hermitian property of σ. 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. it holds φk = (φk )∗ .89) such that (11.88) which one veriﬁes taking the conjugate complex of (11. 2. The transformation ˙ behaviour of φα is ˙ φ α = (aα β )∗ φβ ˙ (11.

The transformation behaviour of φα is ˙ φα = ˙ ˙ β γδ ˙˙ φδ ˙ ˙ αβ a γ ˙ ˙ .75.75).97) The transformation.97) implies then that φα transforms like Ψ ˙ φ1 ˙ φ2 ˙ transforms under rotations (z = iϑ) like a spin– 1 state | 1 .4: Spinor Tensors 365 The relationship between contravariant and covariant conjugate complex spinors can be expressed in analogy to (11.e.100) . A comparision of (11. + 1 2 2 2 1 transforms under rotations (z = iϑ) like a spin– 2 state | 1 . 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. is governed by the operator a∗ (z) −1 which arises in ˜ the Lorentz transformation (11.94) (11.18) ˜ 3 .73. a∗ (z) −1 accounting for the ˜ transformation behaviour of the third and fourth spinor component of Ψ.4 Spinor Tensors ˙ ˙ ˙ We generalize now our deﬁnition of spinors φα to tensors.95) ˙ where αβ and αβ are the real matrices deﬁned in (11. i. in matrix notation. 11.18) of the bispinor wave function Ψ.96) is Lorentz invariant. For the spinors φα and χα thus ˙ deﬁned holds that the scalar product ˙ φα χα = φ1 χ1 + φ2 χ2 ˙ ˙ ˙ ˙ ˙ (11.92) (11. 11. (11.11. a property which is rather evident.93) (11.. − 1 . Ψ4 .76) φα ˙ ˙ φα αβ ˙ ˙ αβ ˙ ˙ = = = = αβ ˙ ˙ αβ ˙ ˙ αβ αβ φβ φβ ˙ ˙ (11. A tensor tα1 α2 ···αk β1 β2 ···β (11.98) ˜ ∗ (z) −1 Ψ3 a ˜ Ψ 4 obviously implies that the solution of the Dirac equation in the chiral representation can be written in spinor form ˜ 1 µ Ψ1 (xµ ) φ (x ) ˜ Ψ2 (xµ ) φ2 (xµ ) φα (xµ ) = = . (11.99) Ψ3 (xµ ) χ ˙ (xµ ) ˜ χβ (xµ ) ˙ 1 ˜ χ2 (xµ ) Ψ4 (xµ ) ˙ 11. one can state ˜ and (11.

with ˙ contravariant indices αβ. ˙ ˙ ˙ ˙ (11.106). it holds αβ .103) A αβ = aα γ aβ δ Aγ δ .. (11. using conventional matrix indices j. i. i. i.4. According to (11.m aj a∗ t km m .e. (11.103) Similarly. The 4–vector Aµ in spinor form We want to provide now the spinor form of the 4-vector Aµ .e.107) which reads in spinor notation [cf. An obvious candidate is [cf.m aj akm t m (11.104) Indices on tensors can also be lowered employing the formula tα β = and generalizations thereof. m.106) Exercise 11. (11. This tensor transforms according to t αβ = aα γ aβ δ tγ δ ˙ This reads in matrix notation.25] M (Aµ ) = σµ Aµ . (11.25) the tensor is explicitly Aαβ = ˙ A1 1 A1 2 ˙ ˙ A2 1 A2 2 ˙ ˙ = A0 + A3 A1 − iA2 A1 + iA2 A0 − A3 . αβ ˙ αγ t ˙ γβ (11. .1: Prove equation (11. This task implies that we seek a tensor. 11.e.101) An example is the tensor tαβ which will play an important role in the spinor presentation of the Dirac equation. this transformation behaviour is in harmony with the tensor notation adopted. ˙ ˙ ˙ ˙ (11.108) Obviously. k.. we want to express Aµ through a spinor tensor. tjk = a t a† jk ˙ ˙ ˙ ˙ (11. We had demonstrated that M (Aµ ) transforms according to M = M (Lµ ν Aν ) = a M (Aµ ) a† (11. An example of a tensor is αβ and tions.102. αβ = αβ (11.109) .366 is a quantity which under Lorentz transformations behaves as t ˙ ˙ ˙ α1 α2 ···αk β1 β2 ···β k Spinor Formulation = m=1 aαm γm n=1 aβn δn tγ1 ···γk δ1 ···δ .. the elements of which are linear functions of Aµ .102) = .105) This tensor is actually invariant under Lorentz transforma= αβ . tjk = a t aT jk = . the transformation bevaviour of a tensor tαβ reads in spinor and matrix notation t αβ = aαγ aβδ tγδ .

(11. employing (11. (11.111) A11 A12 ˙ ˙ A21 A22 ˙ ˙ ˙ 0 −1 1 0 (11.112) A22 −A21 ˙ ˙ −A12 A1 1 .110) one can state ∂ αβ = Similarly. ∂µ in spinor notation The relationship between 4–vectors Aµ .115) Exercise 11. Hence.110) The 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 .114) yields ∂αβ = ˙ ∂0 + ∂3 ∂1 + i∂2 ∂1 − i∂2 ∂0 − ∂3 .2: Prove that (11.115) is correct.113) Aα β = ˙ (11.11.116) .4: Spinor Tensors One can express Aαβ also through Aµ Aα β = ˙ ˙ 367 A0 − A3 −A1 + iA2 −A1 − iA2 A0 + A3 . (11. 11. Using (11.109. (11.110) one obtains Aα β = ˙ A0 − A3 −A1 − iA2 −A1 + iA2 A0 + A3 A0 + A3 A1 + iA2 A1 − iA2 A0 − A3 . 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 . Aµ and tensors tαβ can be applied to the partial diﬀerential operator ∂µ . (11. ˙ 2 (11.117) ˙ ˙ ∂0 − ∂3 −∂1 + i∂2 −∂1 − i∂2 ∂0 + ∂3 .4. (11.

243) holds independently of the representation chosen.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.122) aσ µ −1 a−1 −1 (a∗ )−1 0 Lν µ (11. Hence.124). one can conclude aσ µ ∗ −1 Lν µ σν 0 = . This is ˜ possible since γ µ can be expressed through σ µ .125) is equivalent to µ ∗ −1 −1 a∗ (σ µ )∗ −1 −1 a Lν µ = (σ ν )∗ (11.e. σµ building on the known transformation behaviour of γ µ . (11. of this equation is 0 ∗ (σ µ )∗ a and.124) constitutes the essential transformation property of σ µ and will be considered further. . (11.126) which is the complex conjugate of (11. To obtain the transformation properties of σ µ we employ then (10.f.26) and its contravariant analogue σ µ σµ = σµ = 1 0 0 1 1 0 0 1 . Equation (10. −1 0 0 1 (11.125) a (σ ) Equation (11. Comparision of (10.121). holds also in the chiral representation.e.18) and γ µ by (11. σµ . 0 1 1 0 .368 σµ in Tensor Notation Spinor Formulation We want to develop now the tensor notation for σµ (11. 0 −i i 0 .124) . 1 0 0 −1 . i.118) 0 −1 −1 0 0 i −i 0 For this purpose we consider ﬁrst the transformation behaviour of σ µ and σµ .123) (a∗ )−1 Lν µ a L ν µ = = σν (σ ) ν ∗ −1 (11. We will obtain the transformation behaviour of σ µ . .. hence. i. hence.15) one can write σµ = and..119) (11. (10.124).229) and (11.118) yields ˜ γµ = ˜ Using (11. (11.120) σµ 0 . Note that (10. γµ = ˜ 0 µ )∗ (σ −1 0 σµ σµ 0 (σ µ )∗ −1 .h. (11.243) in the chiral representation expressing S(Lη ξ ) by (11.125) is equivalent to (11.121) One expects then that the transformation properties of σ µ should follow from the transformation properties established already for γ µ [c.s. . (11.243)]. (11.

(11.131) one can express the transformation behaviour of σ µ in the succinct form Lν µ σ µ = σ ν . (11. under rotations the Pauli matrices transform like (j = 1. (11. 11.2) −1 369 (a∗ )−1 = −1 − 1 z ∗ ·σ ∗ 2 e = e− 2 z 1 ∗ · −1 σ ∗ = e2z 1 ∗ ·σ . One can.124) using (11. 0 −1 −1 0 µ=1 . 11. 0 −i i 0 µ=2 .102.129) 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. (11.131) where a is given by (11.2.129. 0 i −i 0 µ=2 .128) One can express.4: Spinor Tensors One can rewrite (11.107) to (11.127) Exploiting the hermitian property of σ. (11.133) Combining (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 ˙ ˙ ˙ ˙ = (11.108) that the same transformation behaviour applies then for general Lorentz transformations. (11. In fact. therefore. (σ ∗ )T = σ yields using (11. It holds according to (11..134) .. .11. This transformation behaviour. 1 0 0 −1 µ=3 . 11. equation (11.16. 11. state that σ µ in a new reference frame is σ µ = a σ µ a† (11.4) with w = 0. e. 2. = .g.e. 11.132) 1 0 0 1 µ=0 . according to (11.4). 3) σj −→ a(iϑ) σj a(iϑ) † = a(iϑ) σj a∗ (iϑ) T . 0 1 1 0 µ=1 .124) a σ µ [a∗ ]T Lν µ = σ ν . transformations (11.130) We argue in analogy to the logic applied in going from (11. −1 0 0 1 µ=3 . i.103) ˙ identiﬁes σ µ as a tensor of type tαβ . hence.2.2) −1 (a∗ )−1 = e2z 1 ∗ ·(σ ∗ )T = e2z 1 ∗ ·σ ∗ T = [a∗ ]T .

e. γ.h.141) σ αβ = σ0 − σ3 σ1 + iσ2 −σ1 − iσ2 σ0 + σ3 = 2 0 0 0 1 0 −1 0 0 .137) σαβ ˙ = δαγ δβ δ . i. β.136) results in the succinct expression 1 2 ˙ γδ ˙ (11. This yields 1 0 0 1 0 0 0 0 σ0 + σ3 σ1 + iσ2 (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. We want to express now σ µ and σµ also with respect to the 4–vector index µ in spinor form employing (11. 1.110) to express σ αβ in terms 0 −1 1 0 0 0 0 1 (11.133) is characterized through a 4-vector ˙ index µ.132) and (11. We can.136) σαβ = = 2 ˙ σ1 − iσ2 σ0 − σ3 0 0 0 0 1 0 0 1 where on the rhs.138) Note that all elements of σαβ are real and that there are only four non-vanishing elements. account for the 4–vector index µ. µ = 0. Each of the 4 × 4 = 16 matrix elements in (11.135) This is the property surmised already above [cf. σµ in which the contravariant indices are moved ‘inside’. the transformation is inverse to that of ordinary 4–vectors. ˙ ˙ In (11. 3 as well as through two spinor indices αβ.e. hence.12) for of σµ yields together with (5.. in fact. however.140) where we employed the expressions (11. The ˙ desired change of representation(σµ )αβ −→ (σµ )αβ corresponds to a transformation of the basis of ˙ spin states f g f −→ = (11. ˙ ˙ 0 −1 1 0 ˙ (11. δ. (11. i..e. the submatrices correspond to tαβ spinors.114). ˙˙ (11. 2. whereas ˙ account for the elements of the individual Pauli the ‘outer’ contravariant spinor indices. We will now consider the representation of σ µ . We can summarize that σ µ and σµ transform like a 4–vector. of (11.138) the ‘inner’ covariant spinor indices.370 Inverting contravariant and covariant indices one can also state Lν µ σµ = σν . α.. matrices. account for the 4-vector µ. state ˙ ˙ ˙ ˙ (σ11 )11 (σ11 )12 (σ12 )11 (σ12 )12 ˙ ˙ ˙ ˙ ˙ ˙ ˙ ˙ ˙ γδ (σ11 )21 (σ11 )22 (σ12 )21 (σ12 )22 ˙ ˙ ˙ ˙ .29)].s. i. = σαβ ˙ ˙ ˙ ˙ ˙ (σ21 )11 (σ21 )12 (σ22 )11 (σ22 )12 ˙ ˙ ˙ ˙ ˙ ˙ ˙ ˙ 21 (σ )22 21 (σ )22 (σ21 ) (σ22 ) ˙ ˙ ˙ ˙ 21 22 Equating this with the r.139) g −f g and. Spinor Formulation (11.224) ˙ Using (11.

Accordingly. For this purpose we start from the ˜ µ .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 . we write ˜ ˙ γµ = ˜ 0 (σ µ )αβ µ ) )∗ ((σ αβ 0 ˙ ˙ .121) of γ ˜ ˜ ˙ basis |αβ whereas the element of γ µ given by σ µ is in the basis |αβ .147) .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 ˙ ˙ . This expression implies that the element of γ µ given by σ µ −1 is in the expression (11. (11. employing then transformation (11.138) one can conclude that the following property holds 1 2 ˙ γδ σ αβ ˙ = 1 2 σ αβ ˙ ˙ γδ = δαγ δβ δ .143) Combined with (11.11.144) results in the simple relationship γ µ Aµ = ˜ 0 Aα β ˙ Aα β 0 ˙ .4: Spinor Tensors 371 and. (11.144) The Dirac Matrices γ µ in spinor notation We want to express now the Dirac matrices γ µ in spinor form. which in ˙ (11. ˙˙ (11.142) = This can be expressed 1 2 σ αβ ˙ ˙ γδ = δαγ δβ δ .146) Exploiting the property (11.145) Let Aµ be a covariant 4–vector. One can write then the scalar product using (11. ˙˙ (11.140) to transform each of the four submatrices. ˙ ˙ ˙ ˙ (11. (11.

˙ (11.149) (11.372 Spinor Formulation 11.99) yields the Dirac equation in spinor form i ∂ αβ χβ ˙ i ∂ αβ φ ˙ The simplicity of this equation is striking.226) ˜ i γ µ ∂µ Ψ(xµ ) = 0 i ∂ αβ ˙ i ∂ αβ 0 ˙ ˜ ˜ Ψ(xµ ) = m Ψ(xµ ) .5 Lorentz Invariant Field Equations in Spinor Form Dirac Equation (11.150) . (11.147) allows us to rewrite the Dirac equation in the chiral representation (10.148) ˜ Employing Ψ(xµ ) in the form (11. α ˙ = = m φα m χβ .

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

λn = E exp(iαk Sk ) ψE. 2m r (12. In chapter 7 the dynamics of a particle moving in three dimensions under the inﬂuence of a spherically symmetric potential. Therefore... 3p and 3d all have the same energy.λ1 . ... m r (12. The energy levels are mk 2 En = − 2 2 . as represented by an element of SO(3). is allowed to take values in 0.12). the states 3s. leaves the energy of the state invariant H exp(iαk Sk ) ψE.m (r) Y m (θ.18). · · · . .. ψE.λn .m (r) = vE. For example. 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 .7) 2 n where the orbital angular momentum. φ).. (7... 3 .3) If the newly obtained state is linearly independent of the original one.5) with m = −l. where the symmetry generators. . Sk . each energy level is (2l + 1)-fold degenerate. Classically. (12.24). (12. (12. n − 1. . 2. 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..λ1 . which yields a set of eigenstates of the form ψE. on an energy eigenstate.8) ˆ [H.λ1 . k = 1. The energy levels are totally independent of l. can then be reduced to a one-dimensional radial o equation. . generates a state which has the same energy as the original one.4) The stationary Schr¨dinger equation. .39). l. As an example of this we will consider the Coulomb problem with the Hamiltonian H = p2 k − . Jk . In order to illustrate this. V (r). as studied in chapter 7. this implies a degeneracy in the spectrum.6) From elementary quantum mechanics we know the spectrum of the hydrogen atom. The spherical symmetry implies the commutation of the Hamiltonian with angular momentum operators (7. (12.λn .. can be identiﬁed with the angular momentum operators. We will investigate this shortly in more detail in the case of systems with spherical symmetry. This degeneracy follows from the fact that any rotation. Jk ] = 0 . (7. namely motion in a spherically symmetric potential described by the group SO(3) (or its double covering SU (2) as discussed in section 5. Sk ..374 Spinor Formulation The action of any symmetry generator. (7.8) .. . Presence of additional symmetries may further increase the degeneracy of the system. we ﬁrst consider a particular example of the implications of symmetry. has been discussed. Lie groups play an essential role in the discussion of mass degeneracies in particle physics.

12) (12. (12. 2m r 1 x3 (J2 p1 − J1 p2 + i p3 ) − k .11.15) which can be proved after very considerable algebra. 2E (12.9) (12.21) . Now we scale the eccentricity vector as follows K= − m . which is valid even when a and b do not commute. For this purpose we ﬁrst note that J · = 0. we will compute the hydrogen spectrum using the additional symmetry.3. Aj ] = i ijk Ak .5: Lorentz Invariant Field Equations 375 The vector points along the symmetry axis of the elliptical orbit and its length equals the eccentricity of the orbit. In the following we consider the bound states. 2m r (12. The vector in (12.14) This follows from a · (a × b) = (a × b) · a = 0.20) (12. Therefore.10) (12. We introduce the following new operators 1 A = (J + K). which have a negative energy E. Kj ] = i ijk Kk . (12. 2 1 B = (J − K).8) is not a hermitian operator. (12. in the subspace of the Hilbert space corresponding to a certain energy we can replace H by E.18) which complement the familiar angular momentum algebra of section 5. (12. We will also need the following identity [4] 2 = 2H J2 + m 2 + k2 . Jj ] = i [Ki . 2m r 1 x1 (J3 p2 − J2 p3 + i p1 ) − 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 . In order to understand the aforementioned extra degeneracy. ijk Jk . 2m r 1 x2 (J1 p3 − J3 p1 + i p2 ) − k .17) (12. 2 which can be shown to satisfy [Ai .19) (12.13) It can be veriﬁed explicitly that its components commute with the Hamiltonian.16) Through some algebra [4] the following commutation relations can be veriﬁed [Ki .11) (12.

the energy eigenvalues are found to be E=− mk 2 1 . 1. which is identical to the the direct sum of the Lie algebra of of two SO(3) (or SU (2)) algebras. A. . . In contrast to the discussion above about extra symmetries. Spinor Formulation (12.33) (12. . = 0. . . . 2 1 . Chaos is described classically as exponential . Using this equation the energy eigenvalues can be written in terms of the eigenvalues of A2 and B 2 operators.29) (12. . .34) It follows that l has to have values in {0 = |a − b|.30) 2 2 2 1 .28).21) and (12.15). we can read oﬀ the eigenvalues of A2 and B 2 from (12.22): A2 = a(a + 1) B 2 = b(b + 1) Following (12. ijk Bk . . Bj ] = i [Ai . a = 0.28) .23) (12. namely that J J > < A − B =0 A + B =2 A (12.31) where we have used (12.27) (12. This illustrates the eﬀect of additional symmetries to the degeneracy structure of a quantum mechanical system.14) we note that A2 − B 2 = J · = 0. . (12. .25) So far we have shown that the symmetry generators form an algebra.24) (12. By comparing to the rotation algebra introduced in chapter 5.26) (12. The most extreme case of this is the quantum analogue of a classically chaotic system. 2 (12. 1. . . 1. . can be seen to follow from the triangle inequality as applied to J = A + B. Noticing that A2 and B 2 have the same eigenvalues because of (12. . . H B.7) tells us that (2a + 1) = n. a = 0. . 1. . . This implies that a = b.376 [Bi . a lack of symmetry implies a lack of degeneracy in the energy levels of a quantum mechanical system. l. 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 (12.32) A comparison with (12. b = 0. a + b = n − 1}. Bj ] = 0. H = 0.22) (12. Furthermore the bound on the orbital angular momentum. 2 (2a + 1)2 2 2 (12.

we will be able to use the familiar Clebsch-Gordan coeﬃcients to combine the isospin of two particles the same way we added spin in chapter 6.5: Lorentz Invariant Field Equations 377 sensitivity to initial conditions. analogous to the 1 spin-up and spin-down states of a spin. For example. for example. This enables us to utilize what we already know about the SU (2) symmetry group from the study of angular momentum. It is important to place the isospin concept in its proper historical context. If the chosen boundary is ‘irregular’ in a suitably deﬁned sense. that the proton and the neutron behave identically under the so-called strong interactions and that their diﬀerence is solely in their charge content. In the next section we will proceed with the discussion of a symmetry. namely. This (approximate) degeneracy led into the idea of the existence of an (approximate) symmetry obeyed by the underlying nuclear interactions. who. However.2 system p= 1 0 . which is a particle in box problem in two dimensions with a boundary which can be chosen arbitrarily. it is natural to represent them in terms of a two component vector. A typical example of this so called quantum chaos is the quantum billiard problem.35) this mass equivalence can be viewed as an energy degeneracy of the underlying interactions. 12.11. (12. Following the mass-energy equivalence of special relativity E = mc2 . in the sense that nearby trajectories in the phase space diverge from each other over time. the isospin symmetry is also governed by an SU (2) group ‘rotating’ components in (12.) If the proton and the neutron are to be viewed as two linearly independent states of the same particle. mp = 939. (Strong interactions bind the atomic nucleus together. the classical trajectories will diverge from each other after successive bounces from the boundary. n= 0 1 .2 Isospin and the SU (2) ﬂavor symmetry The concept of isospin goes back to Heisenberg. suggested that the proton and the neutron can be regarded as two states of a single particle. another manifestation of chaos is the lack of independent operators commuting with the Hamiltonian. This is known as level repulsion. 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. This was motivated by the observation that their masses are approximately equal: mp = 938. Originally it was believed that isospin was an exact symmetry of strong interactions and that it was violated by electromagnetic and weak interactions. (12. which was discovered by observing degeneracies in the particle spectrum. after the discovery of the neutron in 1932.36) In analogy to the concept of spin regarding the rotations in 3-space as discussed in chapter 5. (Weak interactions are responsible. For a more detailed discussion of quantum chaos in billiard systems we refer the reader to [7] and the references therein.28M eV /c2 . for the .36) into each other in abstract isospin space.57M eV /c2 .

−1 = = = 1 1 1 1 . (12. 0 . 2 2 1 2 2 2 1 1 1 1 1 √ . However.6M eV /c2 . 2 2 1 2 2 2 (12. −1 . π − = |1. (12. Shortly we 1 will see how to describe both the pion and nucleon states as composites of more fundamental I = 2 states. d= 1 1 . I3 = 2 2 1 1 . I3 = − 2 2 . .0M eV /c2 ). I. mπ0 = 135. Further than that. 2 (12. Therefore it remains a useful concept. including the proton and the neutron. we refer the reader to [1].− .− . For a precise description of the two nucleons as composite states. They u form an isotriplet: π + = |1. including the spin and color quantum numbers of their constituent quarks. albeit it is a good approximate one. as good quantum numbers. . u¯ and d¯ states. If the mass diﬀerence (or the energy diﬀerence) were to be to be purely electrostatic in nature.41) (12. the fundamental representation of the isospin symmetry corresponds to the doublet that contains the so-called up and down quarks u= 1 1 . 2. We can describe isospin multiplets the same way we have described the angular momentum and spin multiplets.38) which have all nearly identical masses. They can be constructed with the same rules that have been used for angular momentum addition ¯ u in chapter 6. 1 .11. (mπ± = 139.36) as a multiplet with I = 2 1 1 p = I = .40) + 2 1 1 . 2 2 1 1 1 . it can be seen as part of a larger (and more approximate) symmetry which is of great utility to classify observed particle families. sec. n = I = . The mass diﬀerence between the neutron and the proton could then be attributed to the charge content of the latter.42) (12.38) are ud. 2 2 . π 0 = |1. Denoting the total isospin.− 2 2 . and its third component. 0 π − = |1. Isospin symmetry is not an exact symmetry of strong interactions. For example. .− 2 2 1 2 2 2 1 1 1 1 . the proton is the lighter of the two. If it were otherwise the proton would be unstable by decaying into the neutron. I3 . 1 π 0 = |1. 1 we can re-write (12.39) All other isomultiplets. as we shall see below. the proton had to be heavier. In the framework of the quark of model.43) Similarly.37) As an example of a multiplet with I = 1 we have the three pions or π-mesons π + = |1. are made up of these two quarks. the three pions in (12.− 2 2 . the proton and the neutron can be written as totally symmetric uud and udd states. .378 Spinor Formulation beta decay). spelling disaster for the stability of matter. respectively. (12.

muon.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. by accidental convention. 0 ). The possibilities are that of an isosinglet 1 |0. as some mesons discovered later are heavier than some baryons and so on. consists of a proton and a neutron. I3 = 0. By convention ¯ baryons have baryon number 1. Naturally. this will be mathematically identical to adding two spins. In the late 1940’s and early 1950’s. The names. where leptons (electron. whereas mesons are q q states. 0 ) we should have seen nn and pp bound states of comparable energy in nature (because of isospin symmetry). a number strange particles have been found which presumably contained a third quark species: the strange quark. The up and down quarks have strangeness zero. a hydrogen isotope.44) 2 where B is the baryon number and S is the strangeness. strongly interacting particles) which is about a GeV /c2 . Before proceeding further. The reader may know that the deuteron.46) (12. to be −1 for the strange quark. but such states do not exist. originally refer to the relative weight of particles. It shall be noted here that the quark model was not invented until 1960’s.48) (12. such as proton and the neutron. If taken literally. Following (12. 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. (12.37) and in analogy to (12. mesons have intermediate mass ranges. baryon and meson. −1 = |p > |p >. The value of the strangeness quantum number is taken. Therefore the deuteron has to be an isosinglet state (|0. we shall setup some terminology: baryons are qqq states. Therefore it has to have isospin. the neutrinos etc. (12. All other composite states have their strangeness given by the sum of the strangeness content of their constituents. 2 = |n > |n > . All antiparticles have their 3 quantum numbers reversed. 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). 1 = √ (|p > |n > +|n > |p >).or ﬂavors as they are commonly called. Now let us consider another example of combining the isospins of two particles. but at the time the empirical concepts like isospin and strangeness quantum numbers were in use. and quarks have baryon number 1 . this remains only an inaccurate naming convention today. As it later turned out.43).11. the up and down quarks are not the only quark ‘species’ . mesons have baryon number 0.) are light. baryons generally are heavy.e.45) Is the deuteron an isosinglet state or an isotriplet? If it were an isotriplet (|1. 0 = √ (|p > |n > −|n > |p >) 2 and that of an isotriplet |1. . We will now try to describe its wave function in terms of its constituent nucleons. 0 |1. This is why isospin is such a good symmetry and why isomultiplets have nearly identical masses.47) (12. the pions being examples thereof. 1 |1.

49) The only assumption that we put in will be that the interaction is of the form V = α I(1) · I(2) .52) where γ (i) and γ (f ) denote degrees of freedom other than isospin.38) and the fact that the deuteron is an isosinglet we know that the isospins of the ﬁnal states in (I) and (II) are |1. γ (i) is a reduced matrix element deﬁned in the same sense as in section 6.8 to compute the ratio of the scattering amplitudes MI and MII . For completeness let us start by restating the Wigner-Eckart theorem (6. is proportional to |M|2 . which is an isoscalar. The cross-section. 1 and (1/ 2)(|1. γ (i) (i) (12. as discussed in the exercise above. 0 ).54) Here T00 ≡ V = α I(1) · I(2) . . (f ) (i) (I (f ) I3 |00I (i) I3 ) is a Clebsch-Gordon coeﬃcient and I (f ) . 1 4 in an isotriplet state and − 3 in an isosinglet 4 Using (12.55) .7. 0 . (12. √ According to (12. I3 . Exercise. 1 and |1. Exercise.45) and (12.259) discussed in detail in sections 6. γ (f ) |T00 |I (i) I3 . γ (f ) (f ) (i) (12. consider (I) p + p → d + π + (II) p + n → d + π 0 (12.380 Spinor Formulation As an exercise on the implications of isospin symmetry we will consider nucleon-nucleon scattering. I3 = 1. I3 . σ. 0 + |0. where M is the scattering amplitude given by M = ﬁnal | α I(1) · I(2) | initial . We will eventually be able to compute ratios of scattering cross-sections between diﬀerent processes. The initial and ﬁnal states can be denoted in more detail as | initial | ﬁnal = = I (i) . Refer to exercise 6. Now let us re-write more carefully the scattering amplitudes for the two processes in the light of what we have just learned MI = I (f ) = 1. γ (i) = (f ) (i) (I (f ) I3 |00I (i) I3 ) (f ) (i) (−1)I −I (f ) (i) (12.7 and 6. respectively. Consider the generalization of tensor operators discussed in section 6. γ (f ) ||T ||I (i) .7 to the case of isospin.50) . γ (i) I (f ) .48) the initial states in (I) and (II) have isospin values |1.5 for the spin-analogue of the same problem. Show that I(1) ·I(2) is an ‘isotensor’ of rank zero. I3 (f ) = 1. We will now employ the (isospin analogue) of the Wigner-Eckart theorem (6. The dot product here refers to the abstract isospin space. (12. γ (f ) T00 I (i) = 1. such as the spatial dependence of the wave function and spin.259) in the present context: I (f ) I3 . γ (i) .8. Show that the expectation of I(1) · I(2) is state. For example. γ (f ) ||T0 ||I (i) .53) 0 √ 1 2I (i) +1 I (f ) .51) (12.

which are obtained by a standard Clebsch-Gordan expansion π + + p : |1.5: Lorentz Invariant Field Equations 1 (11|0011) √ I (f ) = 1.59) (12.60) (12. σII which is in approximate agreement with the observed ratio.58) (12. I3 = 0. (c) π − + p → π 0 + n . = 381 (12. (12.56) (12. 2 . γ (i) 2 1 = (10|0010) √ I (f ) = 1. −1 2 3 2 − + . 2 = 2. . γ (f ) ||T00 || I (i) = 1.57) (12. which is also manifested by a vanishing Clebsch-Gordon prefactor.61) MII Note that the second term in MII vanishes due to the isospin conservation. Once again we need the isospins for the initial and ﬁnal states. σ( π − + p → anything ) The possibilities are (a) π + + p → π + + p . (12. 2 = 3 2. γ (i) 3 1 (f ) (i) I (f ) = 1. −2 . −2 1 1 2. we will consider pion-nucleon scattering. γ (f ) ||T00 || I (i) = 1. but these are not dominant at relatively low energies. I3 = 0. involving.63) where common dynamical factors (which would not be as easy to compute) have dropped out thanks to the Wigner-Eckart theorem. The relevant Clebsch-Gordon coeﬃcients are easily evaluated: (11|0011) = (10|0010) = 1. γ (f ) T00 I (i) = 1. 1 1 1 1 3 √ 2. I3 = 0. −2 2 1 1 3 2.66) = 2 3 1 3 2. I3 = 0. (12.65) (12. −1 π0 + n : |1. [2] As a further example.62) We can now write the ratio of the scattering amplitudes: MI 1 √ . 1 π − + p : |1. particles with strangeness. −2 1 1 √ .64) There are more exotic possibilities.11. γ (i) = √ 2 1 (f ) (i) +√ I (f ) = 1. γ (i) 3 +0. We want to compute the ratio of total cross-sections assuming a similar interaction as in the previous example σ( π + + p → anything ) . 0 1 1 3 3 2. for example. It follows σI = 2. γ (f ) T00 I (i) = 0. (b) π − + p → π − + p . = MII (1/ 2) (12.

For instance. giving it a higher than usual lifetime.71) u = 0 .69) 12. making an identiﬁcation of the underlying symmetry and the multiplet structure less straightforward. The classiﬁcation of the newly found particles as members of some higher multiplet structure was less obvious then the case of isospin. (12. m V V 3 2. it appears an obvious generalization to add another component for an extra quark to the isospin vector space 1 0 0 (12. mentioned in the previous section. Hence Gell-Mann and independently Nishijima postulated the existence of a separate quantum number. [2] (12. 0 0 1 . . was motivated by the existence of particles that are produced strongly but decay only weakly. etc.65) Ma = M 3 2 Mb = 1 M 3 + 2 M 1 3 3 √ 2 Mc = 2 3 M3 2 √ 2 (12. S.68) − 2 3 M1 2 Guided by empirical data we will further assume that M 3 >> M 1 . which leads to the following 2 2 ratios for the cross-sections σa : σb : σc = 9 : 1 : 2. S(Σ− ) = −1 and S(π) = S(N ) = 0. In the light of the quark model. As the total cross-section is the sum of individual processes we obtain σ( π + + p ) σa = =3 σ( π − + p ) σb + σc again in approximate agreement with the observed value. while weak interactions did not. are up to 40% heavier. m 1 2. A computation similar to the previous example of nucleon-nucleon scattering yields (apart from common prefactors) the following amplitudes for the reactions in (12. called strangeness. however.382 Spinor Formulation As in the example of nucleon-meson scattering we deﬁne the relevant matrix elements M3 = 2 M1 = 2 3 2. such that S(K + ) = 1. It was assumed that strong interactions conserved S (at least approximately). m 1 2. has a lifetime which is comparable to that of π + albeit being more than three times heavier. K + .3 The Eightfold Way and the ﬂavor SU (3) symmetry The discovery of the concept of strangeness.67) which are independent of m. which can be produced by π − + p → K + + Σ− . Strange partners of the familiar nucleons. Hence the strangeness changing strong decays of K + (or Σ− ) were forbidden.s = 0 .d = 1 . m .70) (12. for example.

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. This is believed to be an exact symmetry of strong interactions. history followed the reverse of this path. The reason is that the other known quarks. SU (N ). The reader is also invited to revisit section 5. Since arbitrary phase factors are of no physical interest. This group of 3 × 3 unitary matrices is denoted by U (3). where Nf is the number quark ﬂavors. of three dimensions. called color. It is seen that such a matrix U has to satisfy the following k U † = U −1 . In this perspective the ﬂavor SU (3) symmetry may appear to be mainly of historical interest. as in (12. should correspond to. Flavor SU (Nf ) symmetry on the other hand.71) into each other. Then came the question of what the fundamental representation.73) To verify that all such matrices form a group. Multiplying U by a phase.1 whenever necessary. The bottom and top are even heavier. have to be elements of the group SU (3) (which we will investigate closely very soon).74) for any two unitary matrices U and V . in fact modern theory of strong interactions is a ‘gauge theory’ of this color group. First particle multiplets were identiﬁed as representations of the SU (3) group. eiφ . called quantum chromodynamics. unitary relation k (12.5: Lorentz Invariant Field Equations 383 In this case the transformations that ‘rotate’ the components of (12. Given a complex vector. The unitarity relation imposes 9 constraints on the total of 18 real degrees of freedom of a 3 × 3 complex matrix.72) a∗ ak . The reader shall note that most of what is being said trivially generalizes to other unitary groups.71). in reference to Lie groups. which again form representations of an SU (3) group. while preserving the norm. (The reader is referred to section 8. The quarks posses another quantum number. it is the group . which set the hadronic energy scale. (The ‘bare’ mass of the charm quark is already heavier than the two nucleons. for a period they were considered as useful bookkeeping devices without physical content. we observe that (U V )† = V † U † = V −1 U −1 = (U V )−1 . of a. (12.3 for a brief discussion of gauge transformations). becomes increasingly inaccurate for Nf > 3.11. giving rise to the quark model. still leaves the norm invariant. However SU (3) symmetry appears in another and much more fundamental context in strong interaction physics. ak . namely charm. Because of this additional constraint SU (3) has 8 dimensions. the same way nucleons and pions were identiﬁed as representations of the isospin SU (2) symmetry. Lie algebras and related concepts. (12. we will establish some mathematical preliminaries about the group SU (3). [2] ) Before discussing the signiﬁcance and the physical implications of the quark model. we want to ﬁnd those transformations ak → Ukl al that preserve the norm. but we will stick to N = 3 in the following. but there are a number of subtle diﬀerences. bottom (beauty) and top (truth) are signiﬁcantly heavier than the hadronic energy scale. However. In many respects it will resemble the more familiar group SU (2) discussed in some detail in chapter 5. As quarks were never directly observed. Hence the group U (3) has 9 dimensions.

3 (12. (12. via the anticommutator relation 4 [λj .79). since det(eA ) = etrA . the meaning being apparent from the context. The reader is invited to compare the structure of SU (3) to that of SU (2) discussed in section 5. λ2 = . The fundamental relation to be preserved is (12. .) The unit determinant condition requires that all λk are traceless. δjkl . λk ]+ = δjk + 2δjkl λl . (12. We can also introduce the constants. The Lie algebra structure is given by the commutators of λk [λj .7. λk .78) where fjkl are the antisymmetric structure constants similar to the jkl of SU (2) given in (5. U .80) This is the fundamental or deﬁning representation of SU (3). can be written in the form U = eiH where H is a hermitian matrix. .76) where λk are 8 linearly independent matrices forming the basis of the Lie algebra of SU (3). λ5 = . λk ] = 2ifjkl λl . 0 0 0 . −2 λ4 λ6 λ8 The generators.75) αk λk (12.1. regardless of the dimension of the representation. 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 .384 Spinor Formulation SU (3) and not U (3) that is of main interest.77) λ1 . obey the following relation tr(λj λk ) = 2δjk .79) (12. (We shall at times refer to the Lie algebra with the name of the group.32). λ3 = 0 −1 0 . As discussed in section 5. Therefore we will express elements of SU (3) as U = ei k (12. which can be veriﬁed explicitly as matrix identities. λ7 = 0 0 . As in the case of SU (2) higher dimensional representations obeying the same structure can be found. any unitary matrix.

We ﬁrst introduce the F-spin operators 1 Fi = λ i .91) [T0 .86) (12.93) (12. First.84) (12. (12.5: Lorentz Invariant Field Equations 385 As it turns out the set of generators in (12. Exercise. U± = F6 ± iF7 . I3 . In the case of SU (2) the identiﬁcation of ‘ladder’ operators {J+ .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 . V± ] = ± V± . Y is called the hypercharge. V± ] = ±V± .85) (12. T± ] = ±T± . V± = F4 ± iF5 . (Recall that the strange quark has S = −1 by convention). 2 1 [T0 .87) Exercise. namely spherical harmonics. 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.92) (12. starting with the observation that Y = B + S.90) (12.94) .81) In chapter 5.83) (12.87) the commutation relations between the generators can be expressed in a succinct manner. The generators of SU (3) can be arranged into a very similar form to that of SU (2). In the basis (12. (12.11.89) (12. 2 (12. 2 (12. we have [Y. U± ] = U± . (12. T0 ] = 0. 1 [T0 . [Y. Derive the Gell-Mann–Nishijima relation (12. Using the convention in (12.88) which deﬁnes (not uniquely) a maximal set of mutually commuting operators {Y.77) is not the most useful basis in the study of SU (3). T0 } and [Y. 3 (12. J± ] = ± J± . T 0 = F3 .44). U± ] = ±U± . 2 Y = √ F8 . which satisﬁed an ‘eigenvalue equation’ [J0 . T± ] = 0. [Y.71) show that T0 is the isospin operator.

88). The same way the angular momentum ladder operators have been used to construct the representations of SU (2). [T+ .386 Spinor Formulation which relate the remaining generators {T± . This form is essential for easy labeling of the representations of the group. Finally. T0 }. U0 and V0 are linear combinations of T0 and Y . [T+ . the representations will now lie in a T0 − Y -plane. jkl (12. In the case of SU (2) the representations lay on a line on the J0 axis.101) (12.98) (12. C1 = k λ2 = − k 2i 3 fjkl λj λk λl .104) (12. [T+ . U− ] = 0. [U+ . V+ ] = 0. V− ] = 2 (12. U+ ] = V+ .99) (12.5. Casimir operators can be used to label irreducible representations of the Lie algebra. 3 [U+ . However. . since there are two mutually commuting generators in SU (3) as given in (12. we will use these commutation relations to construct representations of SU (3).103) Any remaining commutators follow from hermiticity. The interested reader is instead referred to a very readable account given in chapter 12 of [4]. V+ ] = 0. We can construct two such independent Casimir operators for the group SU (3). (12. similar to the way it was done in section 5. V± . J 2 ) which commutes with all of the generators. SU (2) has rank 1.100) (12. Note that. Another important property of SU (2) is the existence of an operator (namely the total angular momentum. U± } to this maximal set by ‘eigenvalue equations’ and [T+ .97) which relate commutators of generators with opposite eigenvalues to the maximal set {Y. Thus. [U+ . [V+ . The maximum number of mutually commuting generators of a Lie algebra is called its rank. while SU (3) has rank 2. V− ] = T− .105) C2 = jkl djkl λj λk λl In general the number of independent Casimir operators of a Lie group is equal to its rank. When the basis of a Lie algebra is expressed in such a way to satisfy the form of the eigenvalue relations as given above. T− ] = 2T0 . as the relation between the states in a given representation can be conveniently expressed in terms of ladder operators. U− ] = Y − T0 = 2U0 . it is said to be in Cartan-Weyl form. As in the case of J 2 and SU (2). 2 3 Y + T0 = 2V0 . V− ] = −U− .95) (12. we have [T+ . An operator which commutes with all generators of a Lie group is called a Casimir operator.96) (12.102) (12. A formal deﬁnition and a detailed discussion of the Cartan-Weyl form is beyond the scope of this chapter.

Because of (12. y U± |t3 . (See Fig. 0) and D(0.109) The same way that J± |m is proportional to |m ± 1 in the case of the angular momentum algebra. y .) All other representations can be constructed by combining these two conjugate representations.110) (12.y ± 1 . 2 1 = β t3 ± . respectively. 2 (12. For example.antiquark pair. (12.112) The eﬀect of these operators to the states in the y − t3 plane have been outlined in Fig. y .108) (12. the pion family forms part of an octet corresponding to the D(1. y : T0 |t3 . q) and it has a dimensionality of 2 (p + 1)(q + 1)(p + q + 2). U± . 2 1 = γ t3 .11. Such a representation is labeled as 1 D(p.106) (12. Figure (12.107) 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 . . y . The representations for hexagons with sides of length p and q in the T0 − Y -plane. (12. y . 1) as an example.5: Lorentz Invariant Field Equations 387 The utility of Casimir operators arises from the fact that all states in a given representation assume the same value for a Casimir operator. we have T± |t3 . (12. The representations of SU (3) are constructed analogous to those of SU (2) by identifying the ‘boundary’ states annihilated by raising (or lowering) operators. 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. y 1 = α t3 ± .5 how to label the irreducible representations of the angular momentum algebra SU (2) in terms of the value of the total angular momentum. 1) correspond to the triplets of quarks and antiquarks.1). V± . The representations D(1. y Y |t3 .88) we can label states by the eigenvalues of T0 and Y operators. y . 2 1 t3 ) |t3 . As another example for the representations of SU (3). For example the pion octet (or any other meson octet) is therefore realized as states of a quark . y V± |t3 . Now we will construct explicit representations of SU (3). All other states of the representation are then constructed by successive application of ladder operators T± .2) shows the representation D(2. 1) representation. This property can be used to label representations in terms of the values of the Casimir operators. especially chapters 7 and 8.111) (12. y . The details of this procedure is beyond the scope of this chapter. it was shown in section 5. A notation suggestive of the dimensionality of the representation can be used to identify representations. For example. y 3 = ( y− 4 3 = ( y+ 4 1 t3 ) |t3 .3). The interested reader is referred to [4]. y V0 |t3 . |t3 . = y |t3 . we have U0 |t3 . For example. y ± 1 . y = t3 |t3 . 2 (12.

£ Figure 12. 1) representation of SU (3).388 £ Spinor Formulation Figure 12.1: The eﬀect of SU (3) ladder operators on the y − t3 plane.2: D(2. ¡ ¢ ¡ ¢ 8 6 2 '745301) ¥¦¤ ¤ #" ! (& % '$ © ¨ ¦ §¥ ©§ ¨ . The states in the inner triangle are doubly degenerate.

for example.5: Lorentz Invariant Field Equations 389 D(1. . This expansion can be compared. to the case of adding two spin triplet states. 3 (12. 1) is written as [8] etc.117) The Gell-Mann–Nishijima relation can then be succinctly expressed as 1 Q = y + t3 .115) (12. This way the quark-antiquark states can be represented as follows [3] ⊗ [¯ = [8] ⊕ [1] 3] (12.113) The additional singlet state corresponds to the η meson. The three quarks in the fundamental representation can now be written as u = d = s = 1 1 . 0) or the fundamental representation of ﬂavor SU (3) symmetry. 3].114) Figure 12. 1) are denoted by [3] and [¯ respectively.116) (12. . − .11. where we would write [3] ⊗ [3] = [1] ⊕ [3] ⊕ [5]. 2 (12. in the case of SU (2).3: D(1. 0) and D(0.118) ¡ ¢ ¨ ©§ ¨¥¦¤ " ! . 2 3 1 1 − . £ (12. 2 3 2 0. The octet D(1.

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

Sign up to vote on this title

UsefulNot useful- Annihilation and Creation Operators.Chapter 8
- 51891292 Quantum Mechanics I
- Relativistic Quantum Mechanics
- Merzbacher Quantum Mechanics
- Advanced Quantum Mechanics
- Sakurai - Modern Quantum Mechanics Rev Ed- Solutions Manual
- 611 Electromagnetic Theory II
- 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
- Quantum Mechanics Review
- Cambridge.quantum.mechanics.apr.2009.eBook ELOHiM
- Problems and Solutions Quantum Mechanics - Y. K. Lim, World Scientific (1998, 751p)
- Quantum mechanics
- Advanced Quantum Mechanics
- Commutators, eigenvalues and quantum mechanics on surfaces...
- Lectures on Quantum Mechanics - Gordon Baym
- Relativistic Quantum Mechanics-Bjorken j.d., Drell s.d.
- Peskin Chapter 4
- Super Symmetry and Quantum Mechanics
- Quantum Mechanics
- Applied Quantum Mechanics
- Wick Contractions in Latex
- Quantum Mechanics for Engineers
- Solution Manual String Theory
- What is String Theory - J. Pol Chin Ski
- Relativistic Quantum Mechanics
- [Ashcroft & Mermin]Solid State Physics Solution
- Quantum Mechanics Lecture
- particle physics 3
- Notes on Quantum Mechanics