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

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

t) of the conﬁguration d vector q(t). ˙ In the following we will often use the notation for velocities and other time derivatives d q(t) = q(t) and dtj = xj .. We attempt to evaluate t1 S[q(t) + δq(t)] = t0 ˙ ˙ dt L(q(t) + δq(t). δq(t)] = dt − δqj (t) + δqj (t) ∂qj dt ∂ qj ˙ ∂ qj ˙ t0 j=1 j=1 t1 .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. at the ends of the time interval of the trajectories the variations vanish. dt q(t). . · ) is a function diﬀerentiable in its three arguments.6) immediately. (1. · ] in conventional diﬀerential calculus does not correspond to a diﬀerentiated function. q(t).. but rather to a diﬀerential of the function which is simply the df diﬀerentiated function multiplied by the diﬀerential increment of the variable.e. It is also important to appreciate that δS[ · . It holds M M ∂L t1 d ∂L ∂L δS[q(t).7) through Taylor expansion and identiﬁcation of terms linear in δqj (t).5) where L( · .g. t) = L(q(t). ˙ Theorem: Let S[q(t)] = t0 dx dt t1 ˙ dt L(q(t).2 Lagrangian Mechanics Note: δq(t) describes small variations around the trajectory q(t). q(t) + δ q(t). t) + j=1 ∂L ∂L δqj + δ qj ˙ ∂qj ∂ qj ˙ + O( 2 ) (1. the velocity vector dt q(t) and time t. df = dx dx or. t1 ] where the integrand is a function L(q(t). We will later often assume that only variations of a trajectory q(t) are permitted for which δq(t0 ) = 0 and δq(t1 ) = 0 holds. e. j=1 We will now consider a particular class of functionals S[ ] which are expressed through an integral d over the the interval [t0 . t) (1. ∂f in case of a function of M variables.e. q(t) + δ q(t). · .6) For a proof we can use conventional diﬀerential calculus since the functional (1. We focus on such functionals because they play a central role in the so-called action integrals of Classical Mechanics.6) is expressed in terms of ‘normal’ functions. equating these terms with δS[q(t). q(t) + δq(t) is a ‘slightly’ diﬀerent trajectory than q(t). q(t). t) (1.10) From this follows (1.8) 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 . i. i. df = M ∂xj dxj . t0 (1. δq(t)]. For this purpose we consider M ˙ ˙ ˙ L(q(t) + δq(t).

5).1. the above theorem. Theorem: Euler–Lagrange Condition For the functional deﬁned through (1. The proof of the above theorem starts from (1.6) which reads in the present case M ∂L t1 d ∂L δS[q(t). y(x1 ) = y1 .15) We will not provide a proof for this Lemma. (1. . According to the Lemma above follows then (1.13) dt f (t)h(t) = 0 t0 (1. (1. On the other side. y2 ).12) for j = 1. . . For this purpose we deﬁne Deﬁnition: An extremal of a diﬀerentiable functional S[ ] is a function qe (t) with the property δS[qe (t). 2. from (1. 2. t1 ]. . ∂qj dt ∂ qj ˙ t0 j=1 (1. hence.17) . M . dx (1.16) the property δS[qe (t). y(x2 ) = y2 . y1 ) and (x2 .1: Variational Calculus 3 We now consider the question for which functions the functionals of the type (1.12) for j = 1. This condition is stated in the following theorem.5) assume extreme values. · ] ≡ 0 and.16) This property holds for any δqj with δq(t) ∈ F . . M ) d dt ∂L ∂ qj ˙ − ∂L = 0 ∂qj (1. . y(x + dx)) is ds = (dx)2 + ( dy dx)2 = dx dx 1+( dy 2 ) .11) The extremals qe (t) can be identiﬁed through a condition which provides a suitable diﬀerential equation for this purpose. δq(t)] = dt − δqj (t) . The length of such path can be determined starting from the fact that the incremental length ds in going from point (x. . . . 2. δq(t)] = 0 for all δq(t) ∈ F . M and δqj (t0 ) = δqj (t1 ) = 0 follows according to (1.14) for any continuous function h(t) ∈ F with h(t0 ) = h(t1 ) = 0. t1 ] ⊂ R → R holds t1 (1. then f (t) ≡ 0 on [t0 .12) The proof of this theorem is based on the property Lemma: If for a continuous function f(t) f : [t0 . . y(x)) to (x + dx. Let us assume that the two points are connected by the path y(x). if and only if it satisﬁes the conditions (j = 1. it holds in case δq(t0 ) = δq(t1 ) = 0 that qe (t) is an extremal. An Example As an application of the above rules of the variational calculus we like to prove the well-known result that a straight line in R is the shortest connection (geodesics) between two points (x1 .

hence.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. 1. t) (1. dx (1. Threorem: Hamiltonian Principle of Least Action The trajectories q(t) of systems of particles described through the Newtonian equations of motion d ∂U (mj qj ) + ˙ = 0 dt ∂qj . j = 1. . q(t). t1 S[q(t)] = t0 ˙ dt L(q(t). Velocity–dependent potentials which describe particles moving in electromagnetic ﬁelds will be considered below.1.5) with L(y(x).18) dy s is a functional of y(x) of the type (1. according to the theorems above.e. M (1. . The constants a and b are readily identiﬁed through the conditons y(x1 ) = y1 and y(x2 ) = y2 . q(t). (1. . obey the Euler–Lagrange dy condition..23) Presently we consider only velocity–independent potentials.1: Show that the shortest path between two points on a sphere are great circles. q(t). . ˙2 2 (1. The shortest path is an extremal of s[y(x)] which must. t) = j=1 1 mj qj − U (q1 . q2 . Using y = dx the condition reads d dx ∂L ∂y = d dx y 1 + (y )2 = 0. This in turn yields y(x) = ax + b. t) is the so-called Lagrangian M ˙ L(q(t).22) ˙ where L(q(t). the so-called action integral. y = const. . qM ) . (1. i.19) From this follows y / 1 + (y )2 = const and. .4 The total path length is then given by the integral x1 Lagrangian Mechanics s = x0 dx 1+( dy 2 ) .20) x1 − x2 Exercise 1.21) are extremals of the functional. 2. . dx ) = 1 + (dy/dx)2 . . circles whose centers lie at the center of the sphere. One obtains y 1 − y2 y(x) = (x − x2 ) + y2 .

t) as follows B E = ×A 1 ∂A . t) + v × B(r. respectively. x2 (t). t) c ∂t (1.25) = 0 = 0 4π J c = 4πρ = (1.30.27) (1.31) allows one to transform the potentials without aﬀecting the ﬁelds and without aﬀecting the equations of motion (1. Equations (1.30) (1.25) of a particle moving in the ﬁeld.23).27) and (1. t) − K(r.31) = − V − Gauge Symmetry of the Electromagnetic Field It is well known that the relationship between ﬁelds and potentials (1. t) dt c where dr = v and where F (r.32) (1. t) describes the charge density present in the ﬁeld and J(r. t) and a vector potential A(r. t) . t) and B(r.22. t) describes the charge current density.24) ∂qj dt ∂ qj ˙ ∂qj dt which are obviously equivalent to the Newtonian equations of motion. dt The ﬁelds E(r. t).33) . t) obey the Maxwell equations ×E + ·B ×B − ·E 1 ∂ c ∂t E 1 ∂ c ∂t B (1. t) and B(r. t) 1∂ K(r. 1. t) is the Lorentz force. c ∂t (1. t) = q E(r.2: Lagrangian 5 For a proof of the Hamiltonian Principle of Least Action we inspect the Euler–Lagrange conditions associated with the action integral deﬁned through (1. The transformation which leaves the ﬁelds invariant is A (r. t) + V (r. 1.28) (1. The equations of motion for such a particle are d q ˙ (mr) = F (r. F (r. t) = V (r.29) where ρ(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). These conditions read in the present case ∂L d ∂L ∂U d − = 0 → − − (mj qj ) = 0 ˙ (1.28) can be satisﬁed implicitly if one represents the ﬁelds through a scalar potential V (r.1.26) (1. x3 (t)) moving in an electromagnetic ﬁeld described through the electrical and magnetic ﬁeld components E(r. t) = A(r.

equivalent. 1. if one assumes for a particle the Lagrangian ˙ L(r.37) The results (1.35) shows that the Newtonian equations of motion and the Euler–Lagrange conditions are. r. namely. t) + A(r. (1.41) which is identical to the term (1. d ∂V q (mv1 ) = F1 = −q + [v × B]1 .g.40.39) ∂A1 ∂A2 ∂A3 x1 + ˙ x2 + ˙ x3 ˙ ∂x1 ∂x1 ∂x1 . t) · v . t) = 1 q mv 2 − q V (r.30).38. 1. B3 = ∂A2 ∂x1 (1.39) together yield ∂V q d (mx1 ) = −q ˙ + O dt ∂x1 c where O = ∂A1 ∂A2 ∂A3 ∂A1 ∂A1 ∂A1 x1 + ˙ x2 + ˙ x3 − ˙ x1 − ˙ x2 − ˙ x3 ˙ ∂x1 ∂x1 ∂x1 ∂x1 ∂x2 ∂x3 ∂A2 ∂A1 ∂A1 ∂A3 = x2 ˙ − − x3 ˙ − ∂x1 ∂x2 ∂x3 ∂x1 (1.35) − ∂A1 ∂x2 [v × B]1 = x2 B3 − x3 B2 = x2 ˙ ˙ ˙ ∂A2 ∂A1 − ∂x1 ∂x2 − x3 ˙ ∂A1 ∂A3 − ∂x3 ∂x1 . Comparing then (1.37) is d dt ∂L ∂ x1 ˙ = d q dA1 d q (mx1 ) + ˙ = (mx1 ) + ˙ dt c dt dt c ∂A1 ∂A1 ∂A1 x1 + ˙ x2 + ˙ x3 ˙ ∂x1 ∂x2 ∂x3 . .6 Lagrangian of Particle Moving in Electromagnetic Field Lagrangian Mechanics We want to show now that the equation of motion (1. 2 c (1.35) is equivalent to the Euler–Lagrange condition d dt The second term in (1.37) is ∂L ∂V q = −q + ∂x1 ∂x1 c The ﬁrst term in (1.40) (1. dt ∂x1 c We notice using (1. e. (1.36) This expression allows us to show that (1.38) ∂L ∂ x1 ˙ − ∂L = 0.25). ∂x1 (1.. (1.41) with (1.25) follows from the Hamiltonian Principle of Least Action.34) For this purpose we consider only one component of the equation of motion (1. in fact.36) in the Newtonian equation of motion.

They are the subject of the following theorem. a term not aﬀected by the variations allowed. t) the Lagrangian.33) ˙ L (r.32. This term is d ∂K ∂K ∂K ∂K ∂K K(r.. For this purpose we consider the transformation of the single particle Lagrangian (1. t) = mv 2 − q V (r. q. (1. q. 1. t) . 2 c = A(r. t) − + 2 c ∂t c 1 q = mv 2 − q V (r. Note that one adds the total time derivative of a function K(r.33) of electromagnetic potentials. and since these properties allow one often to simplify mathematical descriptions.45) and reordering terms yields using (1. gauge symmetry and symmetries with respect to spatial transformations. 1. t) 1 1 ∂K q mv 2 − q V (r.1.3 Symmetry Properties in Lagrangian Mechanics Symmetry properties play an eminent role in Quantum Mechanics since they reﬂect the properties of the elementary constituents of physical systems. i.32. t) = x1 + ˙ x2 + ˙ x3 + ˙ = ( K) · v + .42) This transformation is termed gauge transformation. appear in a diﬀerent.32. equivalent to the gauge transformation (1. t) + K ·v (1. t) = q K(q. We will consider in the following two symmetries. t) = L(q. t) + A (r. q. 1. t) · v + K(r. The gauge symmetry.44) implies that the gauge transformation amounts to adding a constant term to the action integral. q. 1.44) Since the condition δq(t0 ) = δq(t1 ) = 0 holds for the variational functions of Lagrangian Mechanics.43) To prove this theorem we determine the action integral corresponding to the transformed Lagrangian t1 S [q(t)] = t0 ˙ dtL (q.e. t) · v . t) L (q. dt ∂x1 ∂x2 ∂x3 ∂t ∂t (1.34) q 1 q d ˙ L (r.46) (1. The factor q has been introduced to make this c transformation equivalent to the gauge transformation (1. 2 c c dt Inserting (1. We want to demonstrate now that the transformation (1. t) c t1 t0 = S[q(t)] + (1. t) + A(r. t) + q K(q. Theorem: Gauge Transformation of Lagrangian The equation of motion (Euler–Lagrange conditions) of a classical mechanical system are unaﬀected by the following transformation of its Lagrangian d q ˙ ˙ K(q. t) + dt c (1.43) into (1. surprisingly simple fashion in Lagrangian Mechanics. encountered above in connection with the transformations (1.33) of electromagnetic potentials.3: Symmetry Properties 7 1.42) is. r. in fact. r. t) c t1 t0 t1 t0 ˙ dtL(q. One can conclude then immediately that any extremal of S [q(t)] is also an extremal of S[q(t)].45) .33) of electyromagnetic potentials. Eq.32.

r ∈ R . t) by gauge transformed potentials V (r.47) (1. the reason being that our interest lies in achieving familiarity with the principles (just mentioned above ) of symmetry properties rather than in providing a general tool in the context of Classical Mechanics. for the connection between invariance properties and constants of motion. However.8 Lagrangian Mechanics Obviously. ) . where the origin of is chosen such that r (r. which has an important analogy in Quantum Mechanics. in the case of central force ﬁelds which are invariant with respect to rotations of coordinates around the center. 0) = r . for example. we like to ˆ provide the transformation here anyway for later reference r = r + e. t). Such invariance properties are very familiar. We consider now invariance properties connected with coordinate transformations. A(r..42) corresponds to replacing in the Lagrangian potentials V (r. A (r. The transformations we consider are the most simple kind. We have proven.e. t).33). ˆ ˆ ˆ Examples of Inﬁnitesimal Transformations The beauty of inﬁnitesimal transformations is that they can be stated in a very simple manner. R(r) = through ∂r ∂ (1. ˆ (1. r. ˆ (1. and for the introduction of inﬁnitesimal transformations which will provide a crucial method for the study of symmetry in Quantum Mechanics. therefore. 1. Deﬁnition: Inﬁnitesimal One-Parameter Coordinate Transformations A one-parameter coordinate transformation is decribed through r = r (r. In this case the transformation can be written in matrix form cos −sin 0 x1 x1 x2 = sin cos 0 x2 (1. The corresponding inﬁnitesimal transformation is deﬁned for small r (r.51) We would like to derive this transformation in a somewhat complicated. In case of a translation transformation in the direction e nothing new is gained. for such vectors holds a · a = 1. t). The transformations considered are speciﬁed in the following deﬁnition.48) In the following we will denote unit vectors as a. ) = r + R(r) + O( 2 ) .52) x3 0 0 1 x3 .49) =0 ∈ R (1. The following description of spatial symmetry is important in two respects.50) A non-trivial example is furnished by the inﬁnitesimal rotation around axis e ˆ r = r + e×r .32. i.42) and (1. the transformation (1. but nevertheless instructive way considering rotations around the x3 –axis. the equivalence of (1.

expansion. by Noether. i. namely for ﬁelds rather than only for particle motion. q..53) x3 x3 0 0 0 x3 One can readily verify that in case e = e3 (ˆj denoting the unit vector in the direction of the ˆ ˆ e xj –axis) (1. (1. Before the embark on this theorem we will comment on what is meant by the statement that a classical mechanical system is invariant under a coordinate transformation. ˙ We have generalized in this theorem the deﬁnition of inﬁnitesimal coordinate transformation to M –dimensional vectors q. . t) is invariant with respect to an inﬁnitesimal transformation q = q + Q(q).. t) + j=1 ∂L Qj + ∂qj M j. q.3: Symmetry Properties 9 In case of small this transformation can be written neglecting terms O( 2 ) using cos = 1 + O( 2 ). Invariance implies L = L.k=1 ∂L ∂Qj qk ˙ ∂ qj ∂qk ˙ (1. i. In the context of Lagrangian Mechanics this implies that such transformation leaves the Lagrangian of the system unchanged. a classical mechanical system is invariant with respect to a coordinate transformation ˙ a constant of motion exists. neglecting terms of order O( M ˙ ˙ L (q. q. the second and third term ˙ k=1 in (1.54) which is equivalent to (1. We have used here the notation corresponding to single particle motion. Using the fact.e. the property holds for any system.58) From this follows the statement of the theorem. t) yields after Taylor ˙ 2 ). a quantity C(r. In order to prove Noether’s theorem we note qj qj ˙ = qj + = qj + ˙ k=1 Qj (q) M (1. The property has been shown to hold in a more general context. ˙ ∂qk ˙ Inserting these inﬁnitesimal changes of qj and qj into the Lagrangian L(q.53).55) (1.57) M Qj j=1 d dt ∂L ∂ qj ˙ + ∂L d Qj ∂ qj dt ˙ = d dt M Qj j=1 ∂L = 0 ∂ qj ˙ (1. We consider here only the ‘particle version’ of the theorem. then M ∂L j=1 Qj ∂ xj is a constant of motion. 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 (1.57) ∂ d where we used dt Qj = M ( ∂qk Qj )qk .57) must cancel each other or both vanish. however. r) which is constant along the classical path of the system.1.e. t) = L(q. Theorem: Noether’s Theorem ˙ If L(q.56) ∂Qj qk . sin = + O( 2 ) 0 − 0 x1 x1 x1 x2 = x2 + 0 0 x2 + O( 2 ) .51) reads r = r − x2 e1 + x1 e2 ˆ ˆ (1. Anytime. q.

e. 2 We ﬁrst determine the constant of motion in case of invariance with respect to translations as deﬁned in (1. v) = 1 mv 2 − U (r). A calculation similar to that in (1. Noether’s theorem yields ˆ ˆ the constant of motion (qj = xj . for which translational invariance holds. 3 and. hence. In this case we have Qj = ej · e. j = 1. is conserved. 2. not inﬁnitesimal transformations. It was. to be expected that this is the constant of motion. i.50). except using now Qj = ej · (ˆ × r). 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. j = 1. of course. . ˆ ˙ ˆ j=1 (1.10 Application of Noether’s Theorem Lagrangian Mechanics We consider brieﬂy two examples of invariances with respect to coordinate transformations for the Lagrangian L(r.59) yields the constant of motion ˆ e (ˆ × r) · mv. In this case we will use the same notation as in (1. We will now investigate the consequence of rotational invariance as described according to the inﬁnitesimal transformation (1.59) We obtain the well known result that in this case the momentum in the direction. 2. It is not necessary to investigate the consequences of global.59). 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. 3) 3 Qj j=1 ∂L = e· ˆ ∂ xj ˙ 3 ej mxj = e · mv .51).

t) where z is the conjugate complex of z. 2. The particle is described by a wave function ψ(r.1 The Double Slit Experiment Will be supplied at later date 2. t is |ψ(r. t1 ] .3) where Ω is the space volume in which the particle can exist. t)|2 = 1 Ω ∀t. The wave function ψ(r. At present one may think of f (r) as a sum over δ–functions which represent a multi–slit screen. t) is normalized d3 r |ψ(r. t).2 Axioms for Quantum Mechanical Description of Single Particle ˙ Let us consider a particle which is described by a Lagrangian L(r. The probability to detect the particle with a detector of sensitivity f (r) is d3 r f (r) |ψ(r. 3. We provide now a set of formal rules which state how the probability to observe such a particle at some space–time point r. t is described in Quantum Mechanics. 1. r. t ∈ [t0 . t)ψ(r. is unity.1) (2. (2. t)|2 = ψ(r. t) ψ : R3 ⊗ R → C. 4.2) (2.Chapter 2 Quantum Mechanical Path Integral 2. placed into the space at some particular time and with a detector behind each slit. t1 ] in which the particle is known to exist.4) a condition which enforces that the probability of ﬁnding the particle somewhere in Ω at any particular time t in an interval [t0 . The probability that the particle is detected at space–time point r. t)|2 Ω (2. 11 .

t|r . t)|2 = (2. t| rN −1 . t0 ) = Ωd 3r N −1 Ωd 3r N −2 ··· Ωd 3r 1 φ(r. t ) which requires d3 r φ(r. The following so-called completeness relationship holds for the propagator (t > t [t0 . the path integral r(tN )=rN d[r(t)] · · · r(t0 )=r0 (2. t|r . (2. t|r0 . tN −1 ) φ(rN −1 . t )φ(r. t|r . t ) = δ(r − r ) Ω (2.12 Quantum Mechanical Path Integral 5. This is diﬀerent from the classical situation where the particle follows a discrete path and. t|r0 .7) t. at any intermediate time the particle needs only be known at one space point. tN −2 ) · · · φ(r1 . t ) ψ(r . . t) = Ω d3 r φ(r.. t ) ψ(r . (2. > t1 > t0 is φ(r. This symbol will be deﬁned further below. t1 ] yields an expression of the form r(tN )=rN φ(r. t ) φ(r. t0 ) Ω This relationship has the following interpretation: Assume that at time t0 a particle is generated by a source at one point r0 in space. t1 |r0 . t|r .10) We have introduced here a new symbol. t) is described by a linear map of the type ψ(r. t ) Ωd 3 r |ψ(r. i. namely the point on the classical path at time t . t ∈ (2. t1 ]) d3 r φ(r.5) 6. requires then according to (2. the propagator is unitary.9) Employing a continuum of intermediate times t ∈ [t0 . t ) φ(r . . The time evolution of ψ(r.e. The state of a system at time t.. hence. t0 ) = r(t0 )=r0 d[r(t)] Φ[r(t)] .4) holds for all times. t0 ) = δ(r − r0 ). t0 ) = φ(r. t|r .8) a knowledge of the state at all space points r ∈ Ω at some intermediate time t . t ) t > t .11) which denotes an integral over all paths r(t) with end points r(t0 ) = r0 and r(tN ) = rN . t ) ψ(r . t = )|2 = 1 . t1 ] (2. The deﬁnition will actually assume an inﬁnite number of intermediate times and express the path integral through integrals of the type (2. t1 ]) Ωd Ωd 3r Ωd 3r Ωd 3r 3 r |ψ(r. t0 ) . ψ(r0 . t|r0 . tN −1 |rN −2 .9) for N → ∞. t. (t > t . t.e. Since (2. t ∈ [t0 . This must hold for any ψ(r .8) 7. The generalization of the completeness property to N − 1 intermediate points t > tN −1 > tN −2 > . described by ψ(r. t|r .6) φ(r. i. t). 8. t ∈ [t0 . . t |r0 .

The situation is very diﬀerent for microscopic particles. (2. This implies that small variations of the path near the classical path alter the value of the action integral by very little. such that destructive interference of the contributions of such paths does not occur. The value of S[r(t)] is then Scl = 1 1 m v 2 t = erg s . 2. r.12) S[r(t)] = t0 ˙ dt L(r.14) ˙ In (2. a very large number.0545 · 10−27 erg s . Since this number is multiplied by ‘i’.14) has the same dimension as the action integral S[r(t)].13).e. However.13) are built on action integrals for all possible paths. i.2.12.12) is then Scl / ≈ 0.. contribution to the path integral (2. . t) (2. for the electron many paths contribute and the action integrals for non-classical paths need to be known. it is comparable to action integrals as they arise for microscopic particles under typical circumstances. Its value is extremely small in comparision with typical values for action integrals of macroscopic particles. i. namely due to the property of the classical path to be an extremal of the action integral. The functional Φ[r(t)] in (2. t) is the Lagrangian of the classical particle. in complete distinction from Classical Mechanics. The constant given in (2.. r.11) is Φ[r(t)] = exp where S[r(t)] is the classical action integral tN 13 i S[r(t)] (2. However. However. interferences should be much less dramatic than in case of the macroscopic particle.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.13) L(r. To show this we consider the value of the action integral for a particle of mass m = 1 g moving over a distance of 1 cm/s in a time period of 1 s. In case of a proton with mass m = 1. Scl / ≈ 8. for microscopic particles like the electron this is by no means the case.5 · 1027 .2: Axioms 9.15) The exponent of (2. 2 2 (2. expressions (2. the exponent is a very large imaginary number. However. This number is much smaller than the one for the macroscopic particle considered above and one expects that variations of the exponent of Φ[r(t)] are of the order of unity for protons. Any variations of Scl would then lead to strong oscillations of the contributions exp( i S) to the path integral and one can expect destructive interference betwen these contributions. not only for the classical path. One would still expect signiﬁcant descructive interference between contributions of diﬀerent paths since the value calculated is comparable to 2π.e.12. Only for paths close to the classical path is such interference ruled out. 2. Situations which are well described classically will be distinguished through the property that the classical path gives the dominant. actually often essentially exclusive.13) and = 1. accordingly.

21) N 2.9). . (2. . as in (2. etc. . namely 1 m(xj+1 − xj )2 / 2 and U (xj ). tN −1 ). x. ˙ 2 (2. namely tj+1 − tj = (tN − t0 )/N = N . t) = ˙ 1 mx2 − U (x) . (xN −1 .14 Quantum Mechanical Path Integral 2. 2. 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. a series of times tN > tN −1 > tN −2 > . respectively. (xN . Its value is CN = m 2πi N 2 (2. They can be anywhere in the allowed volume which we will choose to be the interval ] − ∞. > t1 > t0 letting N go to inﬁnity later. 2 2m N (2. . This allows us to write the evolution operator using (2. the xj values are not. In passing from one space–time instance (xj . tj ) to the next (xj+1 . (x1 .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. For the sake of simplicity we will consider the case of particles moving in one dimension labelled by the position coordinate x.20) can be properly taken.13) and. ∞[. (2.12.16) In order to deﬁne the path integral we assume. tN |x0 . 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 . .3 How to Evaluate the Path Integral In this section we will provide an explicit algorithm which deﬁnes the path integral (2. . −∞ dxN −1 (xj+1 −xj )2 1 − U (xj ) . tN )} .19) The main idea is that one can replace the path integral now by a multiple integral over x1 . t1 ). The particles have associated with them a Lagrangian L(x. however. . .10) and (2. at the same time. t0 ).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 .18) The time instances are ﬁxed. tj+1 ) we assume that kinetic energy and potential energy are constant.12) φ(xN .20) Here. t0 ) = limN →∞ CN exp i N N −1 j=0 +∞ +∞ +∞ −∞ dx1 −∞ dx2 . x2 . provides an avenue to evaluate path integrals. The spacings between the times tj+1 and tj will all be identical. (2.

.23). since x(t0 ) = xcl (t0 ) = x0 and x(tN ) = xcl (tN ) = xN .22) where xcl (t) is the classical path which connects the space–time points (x0 . t0 ) (2. tN dt y = y(tN ) − y(t0 ) = 0 . The resulting expression for S[x(t)|x(t0 ) = x0 .s. namely. which can also be written φ(xN . it is more suitable to deﬁne new integration variables yj . tN ).2.27) The third term can be written in the notation introduced tN 1 dt my 2 = S[x(t)|x(t0 ) = 0. using (2.h.24) implies for the second term on the r. ˙ t0 (2.s.30) d[x(t)] exp 2 tN − t0 x(t0 )=0 a result. tN dt t0 1 1 (xN − x0 )2 m x2 = m ˙ cl . = (2.4: Propagator for a Free Particle 15 Rather then using the integration variables xj . The ˙ 1 S[x(t)|x(t ) = x . ˙ ˙ ˙ 2 + y2) = ˙ (2.e.13). (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. t0 ) = exp 1 im (xN − x0 )2 2 tN − t0 φ(0.24) Also we use the fact that the velocity of the classical path xcl = (xN −x0 )/(tn −t0 ) is constant.25) is. x(tN ) = 0. 2 2 tN − t0 (2. ˙ 2 t0 (2.25) + The condition (2.12) yields x(tN )=0 im (xN − x0 )2 i φ(xN .h.23) tN − t0 It is essential for the following to note that.29) This expression corresponds to the action integral in (2. t0 ) = exp S[x(t)] (2. 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 . tN |x0 .. t0 ) and (xN . x(tN ) = xN ] 1 (xN − x0 )2 m + 0 + 2 tN − t 0 + S[x(t)|x(t0 ) = 0. it holds y(t0 ) = y(tN ) = 0 . x(tN ) = xN ] is S[x(t)|x(t0 ) = x0 . t0 ) and (xN . (2. tN ) is to be evaluated.26) The ﬁrst term on the r.28) i.31) We have denoted explicitly that the action integral for a path connecting the space–time points (x0 . 2. xN − x0 xcl (t) = x0 + ( t − t0 ) . due to (2. can be expressed through a path integral with endpoints x(t0 ) = 0.10. tN |0. Inserting the result into (2. One obtains S[x(t)|x(t0 ) = x0 . x(tN ) = 0] . x(tN ) = 0] . To see the beniﬁt of such approach we deﬁne a path y(t) as follows x(t) = xcl (t) + y(t) (2.24). of (2. tN |x0 . the origin of which coincides with the classical path of the particle.

.32) we split oﬀ the factor 2 mN in the deﬁnition (2. In the appendix we prove +∞ +∞ d j. 2 −1 −1 2 (2.38) .k=1 dy1 · · · −∞ −∞ dyN −1 exp i yj bjk yk = (iπ)d det(bjk ) 1 2 .36) which holds for a d-dimensional. .33) where ajk are the elements of the following symmetric (N 2 −1 0 −1 2 −1 2 m 0 −1 ajk = .37) det(Ajk ) . . .. . . . tN |0. symmetric matrix (bjk ) and det(bjk ) = 0. . 0 0 . . noting y0 = yN = 0. 0 0 .k=1 yj ajk yk − 1) × (N − 1) matrix .16 Evaluation of the necessary path integral Quantum Mechanical Path Integral To determine the propagator (2. .. t0 ) = limN →∞ × +∞ +∞ −∞ dy1 · · · −∞ dyN −1 exp i N m 2πi N 2 N × (2. In order to complete the evaluation of (2.. (2..k=1 yj ajk yk (2.. .34) of (ajk ) deﬁning a new matrix (Ajk ) through ajk = Using det(ajk ) = m 2 N N −1 m 2 N Ajk .32) (yj+1 − yj )2 N −1 1 2 j=0 2 m N The exponent E can be written. . . . real.35) must be determined... 0 0 . 2 N ...31) for a free particle one needs to evaluate the following path integral φ(0. (2. 0 0 0 0 0 0 The following integral +∞ +∞ (2. as the quadratic form E = im 2 2 ( 2y1 − y1 y2 − y2 y1 + 2y2 − y2 y3 − y3 y2 2 N 2 2 + 2y3 − · · · − yN −2 yN −1 − yN −1 yN −2 + 2yN −1 ) N −1 = i j. .34) dy1 · · · −∞ −∞ dyN −1 exp i N −1 j. (2.

variable. .43) We like to note here for further use below that one might as well employ the ‘artiﬁcial’ starting values D0 = 1. which follows from (2. . 0 0 Dn = .45) Expressions for Free Particle Propagator We have now collected all pieces for the ﬁnal expression of the propagator (2. Using D1 = |(2)| = 2 .40) .2. x = xN φ(x.18) we obtain φ(0. . tN |0. . n = 1.. det(Ajk ) (2.. . deﬁning t = tN . We seek then to evaluate the determinant of the n × n matrix 2 −1 0 . t0 ) = m 2πi (tN − t0 ) 1 2 .4: Propagator for a Free Particle a property which follows from det(cB) = cn detB for any n × n matrix B. (2. tN |0. Our derivation has provided us with the value det(Ajk ) = N . . . one can readily verify Dn = n + 1 .40) in terms of subdeterminants along the last column. One can readily verify that this procedure leads to the following recursion equation for the determinants Dn = 2 Dn−1 − Dn−2 .. 0 0 0 .42) m 2πi NN 1 2 (2. D3 .39) yields φ(0.. 2. D1 = 2 and obtain from (2.31) and obtain.39) In order to determine det(Ajk ) we consider the dimension n of (Ajk ).. (2. To solve this three term recursion relationship one needs two starting values. . 0 0 0 −1 2 . (2. . .41) D2 = 2 −1 −1 2 = 3 (2. . 0 0 −1 2 −1 . we obtain φ(0. . t0 ) = limN →∞ m 2πi N 2 17 N 2πi m N N −1 2 1 . .. Inserting this into (2. . . 2 −1 0 0 0 −1 2 For this purpose we expand (2.41) the same result for D2 . . presently N − 1. t0 ) = limN →∞ and with NN (2. . .. tN |0. let say n. (2..44) = tN − t0 . .. t|x0 . . . . t0 ) = m 2πi (t − t0 ) 1 2 exp im (x − x0 )2 2 t − t0 .46) . . .

18 Quantum Mechanical Path Integral This propagator.52) and applying (2. according to (2.247). (2.46). We want to apply axiom (2.50) One refers to such states as wave packets.53) . x) + E(x) in (2.51) as follows Eo (xo . We devide the contributions to the exponent Eo (xo .49) is Gaussian of width δ. Below we will apply this to a particle at rest and a particle forming a so-called wave packet. t|r0 .48) as the initial state using the propagator (2. t0 )| = 2 1 πδ 2 1 2 exp − x2 0 δ2 (2. centered around x0 = 0. t0 ) = m 2πi (t − t0 ) 3 2 exp im (r − r0 )2 2 t − t0 . 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 . (2. One obtains then for the propagator (2. the associated probability distribution |ψ(x0 . We will obtain.51) For this evaluation we adopt the strategy of combining in the exponential the terms quadratic (∼ x2 ) and linear (∼ x0 ) in the integration variable to a complete square 0 ax2 + 2bx0 = a 0 x0 + b a 2 − b2 a (2.5) allows us to predict the time evolution of any state function ψ(x. We need to evaluate for this purpose the integral ψ(x. The result (2. the wave function of the particle at later times. x) = im 2 t x2 1 + i o t mδ 2 − 2xo x − po t m + f (x) (2.5) to (2. t) of a free particle. x) + E(x) (2.46) can be generalized to three dimensions in a rather obvious way. t0 ) = 1 πδ 2 1 4 exp − x2 po 0 +i x 2 2δ (2.10) φ(r.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 .48) Obviously. thereby. and describes a single particle since 1 πδ 2 1 2 +∞ dx0 exp −∞ − x2 0 δ2 = 1.

x) (2.54) f (x) = This yields Eo (xo .51) ψ(x.2. according to (2. We proceed as follows.4: Propagator for a Free Particle im x2 − f (x) .61) .53) E(x) = x− po mt 19 (2. We consider a transformation to a new integration variable ρ deﬁned through i 1−i t mδ 2 ρ = x0 − x− 1+ po m t t i mδ2 . that the new path does not lead to additional contributions to the integral. 2 (2. making certain. (2.55) im xo 2 t 1+i t − mδ 2 x− 1+ po m t t i mδ2 . z2 = +∞ × i 1−i t mδ 2 (2.57) and needs to determine the integral +∞ I = −∞ +∞ dx0 eEo (xo .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 . (2.58) The integrand is an analytical function everywhere in the complex plane and we can alter the integration path. t) = 1 πδ 2 1 4 2 t 1 + i mδ2 . x) = One can write then (2. (2.60) is then represented by real ρ values.52). 2 t One chooses then f (x) to complete. however. The beginning and the end of such path are the points z1 = −∞ × i 1−i t mδ 2 . the square in (2.56) m 2πi t 1 2 +∞ eE(x) −∞ dx0 eEo (xo .59) An integration path in the complex x0 –plane along the direction i 1−i t mδ 2 (2.

(2.62) If one can show that an integration of (2.68) and.58) has the end points z1 = −∞ . the paths between z1 → z1 and z2 → z2 have a real part of x0 of ±∞. In fact.20 whereas the original path in (2. Quantum Mechanical Path Integral (2.55). I = Equation (2. Since the exponent in (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 .58) has a leading contribution in x0 of −x2 /δ 2 the integrand of (2. one can show that z1 lies at −∞ − i × ∞ and z2 at +∞ + i × ∞.58) along the path z1 → z1 and along the path z2 → z2 gives only vanishing contributions one can replace (2.54) using (2.69) t)2 t 2δ 2 (1 + i mδ2 ) + i po x − i p2 o t 2m (2. (2. 1+b 1+b ﬁnally E(x) = − (x − po m (2.63) which can be readily evaluated. accordingly. t) = Seperating the phase factor 1 t 1 + i mδ2 1 4 1 2 exp [ E(x) ] .64) ψ(x. using a ab = a − . t m(1 + i mδ2 ) (2.58) vanishes 0 for Re x0 → ±∞.63) holds and.58) by t mδ 2 +∞ I = i 1−i dρ exp − −∞ m 2 t 1+ t mδ 2 2 ρ2 (2. (2.66) yields ψ(x.70) .67) We need to determine ﬁnally (2. We can conclude then that (2. Hence. z2 = +∞ .57) reads then 1 πδ 2 1 4 2πi t . 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) ] .

74) which corresponds to (2. (2. The center of the distribution (2.71).72) 2 t2 δ 1+ 2 4 (2.48) and for the ﬁnal state (2. However.71. t0 ) = exp i po xo m .74) remains invariant in time. The conservation of this factor is particularly striking for the (unnormalized) initial wave function ψ(x0 .75) i.71) conserves the phase factor exp[i(po / )x] of the original wave function (2. (2. t) = exp i po i p2 o x − t m 2m . This ‘spreading’ of the wave function is a genuine quantum phenomenon. the spatial dependence of the initial state (2. Another interesting observation is that the wave function (2. respectively. (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 (2.49).73) m δ increases with time. the case of arbitrary to is recovered iby replacing t → to in (2.72) for the initial state (2. (2. The width of the distribution (2.72).77) .49). (2.71) × exp − (x − po m t)2 ) 2δ 2 (1 + 2 t2 m2 δ 4 (1 − i .48)] to = 0.72) moves in the direction of the positive x-axis with velocity vo = po /m which identiﬁes po as the momentum of the particle. 2.75) ψ(x. This yields. 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 . instead of (2.72) Comparision of Moving Wave Packet with Classical Motion It is revealing to compare the probability distributions (2. coinciding at t = 0 with the width of the initial distribution (2. The corresponding probability distribution is |ψ(x. t0 ) = exp i po i p2 o xo − to m 2m .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 . In this case holds ψ(x. (2.67) provides the complete expression of the wave function at all times t ψ(x.48) and that the respective phase factor is related with the velocity of the classical particle and of the center of the distribution (2. (2.48) for δ → ∞. t) = exp i i p2 po o x − (t − to ) m 2m .4: Propagator for a Free Particle which inserted in (2.2.. We had assumed above [c.f.e.72).

xcl + y(t). tN |x0 .85) We want to show now that the contribution of δL to the action integral (2.10.22 becomes at t > to ψ(x. y(t)] which vanishes according to the Hamiltonian principle as discussed in Sect. the end points of y(t) are y(t0 ) = 0 Inserting (2.79) for a quadratic Lagrangian L(x. 2. For this purpose we use d xcl y = ˙ ˙ (xcl y) − xcl y ˙ ¨ (2. Obviously.78) i. t) = ˙ 1 1 mx2 − c(t)x2 − e(t)x . ˙ ˙ = (2. (2. xcl .80) For this purpose we need to determine the action integral S[x(t)] = t0 dt 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. 2 . ˙ 2 2 tN (2. 2. y(t).12. the spatial as well as the temporal dependence of the wave function remains invariant in this case.83) (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. t) ˙ L (y. t0 ) = x(t0 )=x0 d[x(t)] exp i S[x(t)] (2. 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 . y(t).82) one obtains L(xcl + y.84) .81) vanishes2 .e. x. 2. (2.80) into (2. t) = L(xcl .5 Propagator for a Quadratic Lagrangian x(tN )=xN We will now determine the propagator (2. 1.13) φ(xN . t) ˙ (2. t) + δL ˙ ˙ ˙ ˙ where L(xcl . y(tN ) = 0 . One refers to the respective states as stationary states. x.82) which describes the deviation from the classical path xcl (t) with end points xcl (t0 ) = x0 and xcl (tN ) = xN .. Such states play a cardinal role in quantum mechanics. xcl . t) = exp i Quantum Mechanical Path Integral po i p2 o x − t m 2m .

. 2 N . (2. 0 0 0 c2 0 . . .88) and. 0 0 N 0 − .87) According to (2.79) i ˜ φ(xN .s.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..93) .h.83) the ﬁrst term on the r. . . . of (2. t) ˙ t0 . . y. x t0 (2.. Applying the Euler–Lagrange conditions (1. 0 0 m ajk = . vanishes. tN |0. .s. .. . One can express the exponent E in (2.. . t0 ) = exp S[xcl (t)] φ(0. . cN −2 0 0 0 0 0 cN −1 (2. One can then express the propagator (2. . . .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. .31) for the free particle propagator. . hence. . . . The discretization scheme adopted above yields in the present case ˜ φ(0. also the second contribution on the r.. .. . . . tj = t0 + form j. . tN |0. t0 ) (2. . 0 0 0 c3 . . . .92) where ajk are the elements of the following (N − 1) × (N − 1) matrix 2 −1 0 . tN |0. .h.24) to the Lagrangian (2. For the evaluation of φ(0. 2 −1 0 0 0 −1 2 c1 0 0 . 0 0 −1 2 −1 .k=1 yj ajk yk (2.32). 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) .88) vanishes.91) through the quadratic N −1 E = i j.89) where ˜ φ(0. 0 0 0 . . tN |x0 . 2 . t0 ) = y(tN )=0 d[y(t)] exp y(t0 )=0 i tN dt L (y.91) (yj+1 − yj )2 1 2 2m N where cj = c(tj ). 0 0 0 . . . 0 0 0 −1 2 . . . t0 ) we will adopt a strategy which is similar to that used for the evaluation of (2. . .80) yields for the classical path m¨cl + c(t) xcl + e(t) = 0 x (2. ..

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

In contrast to W (x. of (2. t) = exp[−iω(n + 1 ) t] φn (x) as invariants under the 2 action of the propagator φ(x. t − to ) 1 ˜ φn (xo ) n! (2.. but not of t. t) +∞ −∞ dto 1 ˜ exp[−iω(m + 1 ) to ] φm (xo ) 2 m! (2.187) or 1 ˜ exp[−iω(n + 2 ) t] φn (x) +∞ = −∞ ˜ dxo φ(x. results in 1 ˜ exp[−iω(n + 1 ) t] φn (x) 2 n! +∞ = −∞ dxo Φ(x. .147) in (2. n = 1.188) ˜ ˜ Equation (2.. 2.172) in the form ∞ n=0 (2.185) Replacing t → t + to yields ∞ n=0 1 ˜ exp[−iω(n + 1 ) (t + to )] φn (x) 2 n! ∞ +∞ = m=0 −∞ dxo Φ(x. t). t) = n=0 1 ˜ exp[ − iω(n + 1 ) t ] φn (x) 2 n! (2. xo .171) concludes the proof of (2. which also exhibits such invariance.171) can be expanded in terms of e−inωt .184) 1 ˜ exp[−iω(n + 1 ) t] φn (x) 2 n! ∞ +∞ = m=0 −∞ dxo Φ(x. t|xo . t|xo . Accoordingly. i.h. t|xo . to ) exp[−iω(n + 1 ) to ] φn (xo ) . xo . .172) I = exp 2 mω x e−iωt − e−2iωt − Quantum Mechanical Path Integral mω 2 1 x − iωt 2 2 .36 Altogether. (2.182) Comparision with (2.172). We note that the factor exp(2 mω/ xe−iωt − e−2iωt ) in (2. xo . 2 (2. t − to ) 1 ˜ exp[−iω(m + 1 ) to ] φm (xo ) 2 m! (2. xo . one obtains for the r. t − to ) = φ(x. to ). Noting that the propagator (2. to ) we express (2.188) identiﬁes the functions ψn (x.171) ∞ W (x. 2.172). .186) Fourier transform.s. one can expand (2.172) is a function of t − to and deﬁning accordingly Φ(x. We want to inspect the consequences of the invariance property (2.171. 1 exp[iω(n + 2 ) to ] · · · . .183) where the expansion coeﬃcients are functions of x.e.

z) z 2 e−y or w(y. This follows from ∂n w(y. t)|2 = Nn +∞ −∞ n = 0. (2. z) (2. .171) W (x.183). t) such that +∞ −∞ 2 dx |ψ(x. The Hermite Polynomials ˜ The function (2. 2. t) provide all stationary states of the quantum mechanical harmonic oscillator. (2. 1.196) where Hn (y) = ey 2 /2 (2.183). in a ‘nutshell’ all information on the Hermite polynomials. t) are associated with a time-independent probablity density |ψn (x. t) = exp[−iω(n + 1 ) t] Nn φn (x) .192) w(y. ˜ |φ stationary wave functions ψn (x.197) The expansion coeﬃcients Hn (y) in (2. z) = Hn (y) (2. t) of the quantum mechanical harmonic oscillator ˜ ψn (x.194) plays a central role for the Hermite polynomials since it contains. according to (2. In the following we will characterize the functions φn (x) and determine the normalization constants Nn .197) are called Hermite polynomials which are polynomials of degree n which will be evaluated below. through expansion (2. For this purpose we introduce ﬁrst closed expressions for φ the new variables y z and write (2.2. t) = z 2 e−y where w(y. characterizes the wave functions φn (x).195) zn Hn (y) n! ˜ φn (y) . we have identiﬁed then.5: Propagator for a Quadratic Lagrangian 37 ˜ ˜ the functions ψn (x.171).194) = z2 n=0 ∞ 1 ∞ zn ˜ φn (y) n! (2. Actually. z) = n=0 1 2 /2 1 2 /2 = = mω e−iωt x (2. Expansion (2.189) ˜ dx φ2 (x) = 1 n (2. 2 Here Nn are constants which normalize ψn (x.190) ˜ is obeyed. . We will also argue that the functions ψn (x.194). . To obtain ˜n (x) we simplify the expansion (2.193) (2. Expression (2.191) (2. t)|2 = ˜n (x)|2 . through the expansion coeﬃcients φn (x) in (2.183) reads then w(y.183). z) = exp(2yz − z 2 ) .198) ∂z n z=0 .

2. ν H2 (y) = 4y 2 − 2.200) One can deduce from this expression the polynomial character of Hn (y). z) the generating function for the Hermite polynomials. (2.38 Quantum Mechanical Path Integral which is a direct consequence of (2. One calls w(y.. µ lattice (left diagram) corresponds to a summation over only every other point in an n.202) generatingfunctionology by H. The identity (2. among other properties. .196) ∂n w(y.E.199) Comparision with (2.1: Schematic representation of change of summation variables ν and µ to n = ν + µ and m = ν −µ. 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. and recursion equations for the eﬃcient evaluation of Hn (y). .S.196) to ˜ derive. .L. . closed expressions for Hn (y).Graham. m lattice (right diagram). H1 (y) = 2y.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). For this purpose we expand the generating function (2. We will employ (2.194. that Hn (y) is a polynomial of degree n.201) ν−µ µ µ ν−µ Figure 2. and O. 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. The diagrams illustrate that a summation over all points in a ν.196) results in the so-called Rodrigues formula for the Hermite polynomials Hn (y) = (−1)n ey 2 ∂ n −y2 e .200) yields for the ﬁrst Hermite polynomials H0 (y) = 1.. As will become evident in the present case generating functions provide an extremely elegant access to the special functions of Mathematical Physics3 .Patashnik (AddisonWesley.Knuth. Reading. Massachusetts.196).198) for the Hermite polynomials can be expressed in a more convenient form employing deﬁnition (2. normalization factors for φ(y). D. ν+µ ν H3 (y) = 8y 3 − 12y. ν+µ (2.e. i. 1990) is a useful introduction to this tool as is a chapter in the eminently useful Concrete Mathematics by R. The diagrams also identify the areas over which the summation is to be carried out. Boston. ∂y n (2. We want to derive now explicit expressions for the Hermite polynomials. Inc. 1989).Wilf (Academic Press.

209). The lattices representing the summation terms are shown in Fig.207) This expression yields for the ﬁrst four Hermite polynomials H0 (y) = 1.194) holds w(−y. . m = ν − µ 0 ≤ n < ∞. i. . . −z) and.1. in fact. z) = w(y.201). Hence.210) from which one can conclude the property (2. (2.205) Since (n − m)/2 is an integer we can introduce now the summation variable k = (n − m)/2 . according to 2. otherwise.2. for odd n. From (2. m must be restricted such that either both n and m are even or both n and m are odd. µ are integers the summation over n. (2. 2 µ = n−m . k! (n − 2k)! = Hn (y) (2. H1 (y) = 2y. m are ν = n+m . In case of even n .e. . is a polynomial of degree n. it holds Hn (−y) = (−1)n Hn (y) . 2. 0 ≤ m ≤ n . µ expressend in terms of n. the sum in (2.5: Propagator for a Quadratic Lagrangian and introduce now new summation variables n = ν + µ.206) From this expansion we can identify Hn (y) [n/2] Hn (y) = k=0 (−1)k n! (2y)n−2k .197) ∞ n=0 zn Hn (−y) = n! ∞ n=0 (−z)n Hn (y) = n! ∞ n=0 zn (−1)n Hn (y) n! (2. k! (n − 2k)! (2. 39 (2. hence. 2 (2. 0 ≤ k ≤ [n/2] where [x] denotes the largest integer p.204) Since ν.207) one can deduce that Hn (y). H2 (y) = 4y 2 − 2.207) contains only even powers. it contains only odd powers. (2. According to (2.208) which agrees with the expressions in (2. With this restriction in mind one can express (2.203) The old summation variables ν..209) This property follows also from the generating function.202) e 2yz − z 2 ∞ = n=0 zn n! ≤n m≥0 n! (−1) n−m 2 n−m 2 ! m! (2y)m . H3 (y) = 8y 3 − 12y. p ≤ x. 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 .

2.194). (2. . a property which is consistent with (2. z). Hn+1 (y) − 2y Hn (y) + 2n Hn−1 (y) = 0 . 2. (2. . The reader may verify that w(y. . For Hermite polynomials holds.197) above and given by (2.208). have a node at y = 0. namely. z) ∂ w(y. n = 0. n! (2. Hn+1 (y) − 2y Hn (y) + 2n Hn−1 (y) = 0. Another relationship is d Hn (y) = 2n Hn−1 (y). 2. .218) The reader should recognize the connection between the pattern of the diﬀerential equation (??) and the pattern of the recursion equation (??): a diﬀerential operator d/dz increases the order n of Hn by one. z) should be equivalent to the relationship (??).194). . indeed satisﬁes the latter relationship. dy n = 1. a factor z reduces the order of Hn by one and introduces also a factor n. H2n (0) = (−1)n Recursion Relationships A useful set of properties for special functions are the so-called recursion relationships. dw/dy − 2zw = 0. for example. z) = 0 (2. as deﬁned through (2. . Substituting expansion (2. .217) gives H1 (y) − 2y H0 (y) = 0.215) yields ∞ n=1 z n−1 Hn (y) − 2y (n − 1)! ∞ n=0 zn Hn (y) + 2 n! ∞ n=0 z n+1 Hn (y) = 0 .215) ∂z which can be readily veriﬁed using (2. For this purpose one considers w(0. as given in (2. One can then readily state which diﬀerential equation of w(y.209) since odd functions have a node at the origin.213) which allow one to evaluate Hn (y) from H0 (y) and H1 (y) given by (2.211) m! n! m=0 n=0 Comparing terms on both sides of the equation yields (2n)! . . Starting point of the derivation is the property of w(y. 2.188.216) Combining the sums and collecting terms with identical powers of z ∞ n=1 zn Hn+1 (y) − 2y Hn (y) + 2n Hn−1 (y) n! + H1 (y) − 2yH0 (y) = 0 (2. (2. H2n+1 (0) = 0 . (2.196) into the diﬀerential equation (2. 2. . .233) below. n = 1.40 Quantum Mechanical Path Integral The generating function allows one to determine the values of Hn (y) at y = 0.212) n! This implies that stationary states of the harmonic oscillator φ2n+1 (x). n = 1.214) We want to derive (??) using the generating function. . (2. z) = exp(−z 2 ) and carries out the Taylor expansion on both sides of this expression resulting in ∞ ∞ (−1)m z 2m zn = Hn (0) . . z) − (2y − 2z) w(y. 1. .

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! (2.194.s. .200) yields Hn (y) = The identity (−1)n y2 √ e π +∞ dt −∞ dn 2iyt − t2 e . 2 (2.h.226) π −∞ 2n z n z n n=0 n! Expressing the l. of the Rodrigues formula (2. ﬁnally.196) the orthogonality properties of the Hermite polynomials. 2.220) Using (2. 1.223) dn 2iy t − t2 2 e = (2 i t)n e2iy t − t n dy results.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 .247) for a = i one obtains I(y) = and.2. .224) +∞ −∞ dt tn e2iy t − t . 2 (2. For this purpose we consider the integral +∞ −∞ dy w(y. e−y 2 π e−y 2 (2.196) yields ∞ n.228) .219) which can be written I(y) = e−y 2 +∞ −∞ dte−(t−iy) = e−y √ 2 2 +∞ −∞ dze−z . 2.226a).s.h.194. through a double series over Hermite polynomials using (2. z) w(y. (2. 2 n = 0. 2. 2 (2.221) 1 = √ π +∞ −∞ dt e2iyt − t .225) We want to derive from the generating function (2.222) Employing this expression now on the r. dy n (2. 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 √ π e2 z z (2. (2. . acording to the deﬁnition (2.227) Comparing the terms of the expansions allows one to conclude the orthonormality conditions +∞ −∞ dy Hn (y) Hn (y) e−y 2 √ = 2n n! π δn n .

229) Hn (y) . The normalized states are [c. 3. According to (2. 2. that the wave functions are even for n = 0. in harmony with (??). 1. Normalized Stationary States The orthonormality conditions (2. 4. in agreement with our above discussions.233) . (2. 3. 1. 4 in Fig.190)] φn (y) = Nn e−y 2 /2 and for the normalization constants Nn follows from (2. One can recognize.228) allow us to construct normalized stationary states of the harmonic oscillator. ¡ ¢ ¡ -4 -2 0 2 4 y (2.233) are presented for n = 0. (2.189. Furthermore. 2.2. 4.228) +∞ 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: Stationary states φn (y) of the harmonic oscillator for n = 0. the value of the wave function at y = 0 is positive for n = 0. 2.f. One can also recognize that n is equal to the number of nodes of the wave function. negative for n = 2 and vanishes for n = 1.197) holds 2 ˜ φn (y) = e−y /2 Hn (y) .232) Hn (y) . (2.231) √ π (2. 2. 4 and odd for n = 1. 3. 2.42 Quantum Mechanical Path Integral E 9/2 h ω ¥ ¥ ¥ ¥ ¥ 7/2 h ω 5/2 h ω ¤ 3/2 h ω £ 1/2 h ω ¦ Figure 2.230) (2. 3.

using (2. 2. (2. Ω (2. Other examples of orthogonal polynomials are the Legendre and Jacobi polynomials which arise in solving three-dimensional stationary Schr¨dinger equations.e.179] holds Ω = [−1. 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.228).e. i.237) for Ω = R. +∞[.10 and eq.233) and (2.189) φn (x) = √ 1 2n n! mω π 1 4 e− mωx2 2 Hn ( mω x) . (5. (10.231) of the wave functions diﬀers from that postulated in (2.236) where w(x) is a so-called weight function with the property w(x) ≥ 0. 1].2. In = Γ(n+α+1)/n! where Γ(z) is the so-called Gamma function. of functions which obey dx f 2 (x) w(x) = < ∞ . (2. w(x) = 0 only at a discrete set of points xk ∈ Ω (2. ??? and eq. w(x) = √ exp(−x2 ). Comparision with (2. w(x) ≡ 1... The latter are written for polynomials pn (x) in the general form dx pn (x) pm (x) w(x) = In δnm Ω (2. g ∈ F .189) by the Jacobian dx/dy. denoted by P (x) and introduced in Sect.459] hydrogen atom.238) where the space is endowed with the scalar product (f |g) = Ω dx f (x) g(x) w(x) = < ∞ .234) The explicit form of the stationary states of the harmonic oscillator in terms of the position variable x is then. 5 below [c. f. w(x) = xα e−x .228) shows that the orthonogality condition of the Hermite polynomials is in complience with (2. ???] and relativistic [see Sect. 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. The orthogonal polynomials pn mentioned above have the important property that they form a complete basis in the space F of normalizable functions. 5. holds Ω = [0. and I = 2/(2 + 1).150 . The various orthonogal polynomials diﬀer in the spaces Ω ⊂ R over which they are deﬁned and diﬀer in a weight function w(y) which enter in their orthonogality conditions. 5. by dx = dy mω 1 4 . (2. In case of the Legendre polynomials.237) and where In denotes some constants. 10. and In = 2n n! π.f.156.151. 5. In case of the associated Laguerre (α) polynomials. The Hermite polynomials are so-called orthonogal polynomials since they obey the conditions (2.239) . i.5: Propagator for a Quadratic Lagrangian 43 The normalization condition (2.236 .

..e. (2.240) where cn = 1 In dx w(x) f (x) pn (x) . The only possibility for this to be true is dn = 0. t) = n=0 cn (t) e−y 2 /2 Hn (y) .197) it must hold cn (t) = dn exp[−iω(n + 1 )t] and.239) becomes the conventional scalar product of quantum mechanics f |g = Ω dx f (x) g(x) . hence.244) For the state to be stationary |ψ(x. n 2 (2. Appendix: Exponential Integral We want to prove +∞ +∞ I = dy1 . (2.e. det(a) (2.44 As a result holds for any f ∈ F f (x) = n Quantum Mechanical Path Integral cn pn (x) (2. a (2.242) Let us assume now the case of a function space governed by the norm (2. . symmetric a.233) exhaust all stationary states of the harmonic oscillator. pn (x) → w(x) pn (x) the scalar product (2. 2 (2. ψ(y. Since the Hermite polynomials form a complete basis for such states we can expand ∞ ψ(y.. a state for which (2.242) and the existence of a normalizable state ψ(y.247) . If one replaces for all f ∈ F: f (x) → w(x) f (x) and. except for a single n = no . In case of n = 1 this reads +∞ −∞ dx ei a x = 2 iπ . i.245) must be time-independent. in particular.172) holds. t) = n=0 1 2 dn exp[−iω(n + )t] e−y /2 Hn (y) . 2.m=0 d∗ dm exp[iω(m − n)t] e−y Hn (y)Hm (y) . Therefore.246) for det(a) = 0 and real. t) which is stationary under the action of the harmonic oscillator propagator (2. the 2 stationary state ψ(y. Ω (2. ∞ n. t) is ∞ ψ(y.233).147).243) To be consistent with(2.k yj ajk yk = (iπ)n . . −∞ dyn ei −∞ n j. t)|2 .236). the states (2.e. t) must be identical to one of the stationary states (2. aT = a.188.241) The latter identity follows from (2. i. i.e. i.

255) . . yk = k Skj yj . .k yj ajk yk n = j.k n n yj = ˜ k (S −1 )jk yk . . .251) where. This property allows one to simplify the bilinear ˜ form n yj ajk yk by introducing new integration variables j. .249) The akk are the eigenvalues of a and are real. 0 ˜ ˜ S−1 a S = a = .e.246) exploits that for any real.k n m = = j. The proof of (2. according to (2.k n m n y Sj ajk Skm ym ˜ ˜ y (S T ) j ajk Skm ym ˜ ˜ j. 0 ˜ 0 a22 . One can conclude det(a) = j=1 n (2.253) as well as det(˜) = det(S−1 aS) = det(S−1 ) det(a) .2. 0 0 .7: Appendix / Exponential Integral 45 which holds for a ∈ C as long as a = 0. (2. ˜ (2.. . . . 2.248.254) ajj . i.249) n ajk = ˜ l. (2.246) reads then in terms of yj ˜ n n j..250) The bilinear form in (2. ˜ (2. . . . .k yj ajk yk ˜˜ ˜ (2.248) . det(S) a = (det(S))−1 det(a) det(S) = det(a) . ST S = 1 1 or S = S−1 .252) ˜ For the determinant of a holds det(˜) = a n ajj ˜ j=1 (2. . . symmetric matrix exists a similarity transformation such that a11 0 . . ann ˜ where S can be chosen as an orthonormal transformation. . (2.m (S T )jl alm Smk .

. .258) +∞ +∞ ∂(y1 . . . . holds n Quantum Mechanical Path Integral ajj = 0 ˜ j=1 (2.266) . yn ) ∂(˜1 .250) holds J = S and. . .. .264) I = d˜1 . (2. i. n (2. . Accordingly. . .249) follows 1 = det ST S such that one can conclude det S = ±1 One can right then (2. ˜ for j = 1.250) +∞ +∞ I = d˜1 .259) (2.257) Substitution of the integration variables (2. yn ) ) = det(S) . . yn ) y ˜ = ( det S )2 (2. ∂ ys ˜ (2. .263) (2.46 We have assumed det(a) = 0.261) ∂(y1 . . . . hence. .265) k=1 −∞ which leaves us to determine integrals of the type +∞ dx eicx −∞ 2 (2. .256) such that none of the eigenvalues of a vanishes. . . . d˜n e y −∞ = (2. det( From (2.. . . yn ) y ˜ ei n k akk yk ˜ ˜2 . y −∞ d˜n det y −∞ ∂(y1 .260) (2. yn ) ∂(˜1 . . 2.250) allows one to express (2.e.262) (2. .. . ajj = 0 . . 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 −∞ . . . .258) where we introduced the Jacobian matrix J = with elements Jjs = According to (2. . . yn ) y ˜ ∂yj . ∂(˜1 .

268) vanishes. for x. 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 . We consider ﬁrst the case c > 0 and discuss the case c < 0 further below. c (2.2.266) to the well-known Gaussian integral +∞ dx e−cx −∞ 2 = π . p ∈ IR. One can relate integral (2. the integrand does not exhibit any singularities anywhere in C. according to (2.7: Appendix / Exponential Integral 47 where.c > 0 .268) can be written as the sum of the following path integrals J = J1 + J2 + J3 + J4 .. i.e.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. 2 since eicz is a holomorphic function.257) holds c = 0. The contour integral (2. Jk = γk dz eicz 2 (2.271) .267) by considering the contour integral J = γ dz eicz = 0 2 (2. (2.270) dx e−cx 2 2 − 2p √ γ4 : z = ix − p J4 = −p i dx eic(ix−p) . The contour intergral (2.268) along the path γ = γ1 + γ2 + γ3 + γ4 displayed in Figure 2.3.

c ¡ -p γ1 p Re(z) 2 | |e−2cxp | .267) one has shown then ∞ dx eicx −∞ 2 = iπ .273) p→+∞ lim J2 and J4 do not contribute then to integral (2.48 Quantum Mechanical Path Integral Im(z) ip γ2 ¡ γ4 ¡ γ3 ¡ -i p Figure 2.275) .268) for p = +∞.272) = 0 . This follows from the following calculation p p→+∞ lim |J2 or 4 | = p→+∞ lim | 0 p i dx eic(ix+p) | ≤ It holds |e ic(p2 −x2 ) p→+∞ 0 lim |i| dx |eic(p 2 −x2 ) | = 1 since the exponent of e is purely imaginary. (2. (2. p p→+∞ lim |J2 or 4 | ≤ = p→+∞ 0 lim dx |e−2cxp | 1 − e−2cp 2cp = 0.274) (2. One can state accordingly ∞ J = dx e −∞ icx2 − √ ∞ i dx e−cx −∞ 2 Using 2.3: Contour path γ in the complex plain. We will now show that the two integrals J2 and J4 vanish for p → +∞. (2. Hence.

7: Appendix / Exponential Integral 49 One can derive the same result for c < 0. This leads to ∞ icx2 J = dx e −∞ + √ −∞ −i ∞ dx ecx = 0 2 (2.255) this result can be expressed in terms of the matrix a I = which concludes our proof.2.276) and (c < 0) ∞ dx eicx −∞ 2 = −iπ = −|c| iπ . (iπ)n det(a) (2. c (2.279) .275.277) We apply the above results (2.268).277) to (2. It holds n I = k=1 iπ = akk ˜ (iπ)n . if one chooses the same contour integral as (2. but with a path γ that is reﬂected at the real axis.265). n ˜ j=1 ajj (2.278) Noting (2. 2.

50 Quantum Mechanical Path Integral .

t) satisﬁes a partial diﬀerential equation. the Schr¨dinger o o equation provides the simpler avenue towards describing quantum systems than the path ingral formulation of Section 2. t) . t) = From this follows ψ(r. This yields ψ(r.1) We will expand the l.20. In the limit of very small it is suﬃcient to employ a single discretization interval in (2. t + ) = Ω d3 r0 φ(r. . and the r.s.21). We will denote the components of s by (x1 .1) can be expressed through the discretization scheme (2. (3.Chapter 3 The Schr¨dinger Equation o 3. It holds then according to (2. Generalizing (2. t) by an inﬁnitesimal time step . t) ψ(r0 .20) to R one obtains then for small φ(r. x3 51 . t + |r0 .2) d 3 r0 m 2πi 3 2 exp i m (r − r0 )2 − 2 U (r. namely the Schr¨dinger equation. t) . t) ψ(r + s. of this equation in terms of powers of and we will demonstrate that the terms of order require that ψ(r. For many situations. but by no means all. even in x1 . t) .20) to evaluate the propagator. Notable exceptions are non-stationary systems involving time-dependent linear and quadratic Lagrangians. t) ψ(r0 . t + ) = × exp i 2 +∞ +∞ +∞ −∞ dx1 −∞ dx2 −∞ dx3 m x2 + x2 + x2 1 2 3 m 2πi 3 2 × (3. (3.1 Derivation of the Schr¨dinger Equation o We will consider now the propagation of a wave function ψ(r. t) .s. x2 .5) ψ(r.3) In order to carry out the integration we set r0 = r + s and use s as the new integration variable. 2.h. x2 . (3. t + |r0 .h. x3 )T .4) − U (r. t + ) = Ω m 2πi 3 2 exp i m (r − r0 )2 − 2 U (r. The propagator in (3.

According to (3.8) (3. n = 0.10) It is now important to note that in case of integral (3. It is then suﬃcient to consider the terms ψ(r. 1.6) make contributions of the order O( 3 2 ) . 2 (3. t) . we need then to evaluate integrals of the type +∞ In (a) = −∞ dx x2n exp i a x2 .4) is even in all three coordinates x1 .11) holds 3 m 3 iπ 2 2 × = 1 (3. only terms of the expansion (3. a supposition which will be examined below..36) holds I0 (a) = Inspection of (3. O( 5 2 ) ..4) only for small x2 + x2 + x2 values 1 2 3 we expand 3 ψ(r + s..13) 2πi a . x3 . a (3. t) and the kinetic energy term. t) = ψ(r.g. t) + . Obviously. t) j k . The integration involves only the wave function ψ(r +s. (3. but over s = (x1 . O( 7 2 ).6) 3 4 ∂4 j=1 xj ∂x4 ψ(r. the terms collected in (3. t) + ∂xj 2 3 xj xk j. O( 7 2 ) .k=1 ∂2 ψ(r. a I2 (a) = − 3 4a2 iπ . x3 )T . t) + j=1 xj ∂ 1 ψ(r.. of the expansion of ψ(r + s. 1 2 3 2 ∂2 j=1 xj ∂x2 ψ(r.7) shows 1 ∂ In (a). .52 Schr¨dinger Equation o It is important to note that the integration is not over r.5) assuming that only the leading terms contribute. Since the latter contributes to (3. 1 4 3 2 2 ∂4 j. t) j 1 12 .7) According to (2. U (r.4) one identiﬁes 2 1 = a m = O( ) (3. t) j . x2 .9) iπ a (3. t) is a constant with respect to this integration. ∂xj ∂xk (3. . It holds In+1 (a) = I1 (a) = i 2a iπ ..5) which are even separately in all three coordinates yield non-vanishing contributions. e. consequently..11) and. x2 . i ∂a Starting from (3.k=1 xj xk ∂x2 ∂x2 ψ(r.12) Here one needs to note that we are actually dealing with a three-fold integral. t).8) one can evaluate recursively all integrals In (a). (3. Since the kinetic energy contribution in (3.

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

however. D. In (3. J.21) Ω This leads one to consider the integral f | H |g Ω = Ω ˆ d3 rf ∗ (r) H g(r) (3.22) and (3. g ∈ H is deﬁned as follows f |g Ω = d3 rf ∗ (r)g(r) (3.23). Jackson.. The hermitian property.54 Schr¨dinger Equation o linear vector spaces of functions f (r) in which a scalar product between two elements f. g|r) is called the ˆ concomitant of H and is P (f ∗ . (3. Calssical Electrodynamics. Since energy is a real ˆ ˆ quantity one needs to require that the eigenvalues of the operator H are real and. g|r) (3. Chapter 1. in principle. New York.26) the vector–valued function P (f ∗ . 1975). g|r) = − 2 2m ( f ∗ (r) g(r) − g(r) f ∗ (r) ) (3. Interchanging f ∗ (r) and g(r) results in g| H |f Ω = Ω ˆ d3 r g(r) Hf ∗ (r) .27) ˆ We will postulate below that H is an operator in H which represents energy.28) See. 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.25) ˆ f | H |g Ω ˆ = g| H |f Ω + ∂Ω da · P (f ∗ . the surface elements da pointing out of the surface in a direction normal to the surface and the vector–valued function A(r) is taken at points r ∈ ∂Ω. hence. that H is 2 . 2nd Ed. t) g(r) − Ω d3 rg(r) U (r. implies hermitian ˆ f | H |g 1 2 Ω ˆ = g| H |f Ω (3. f ∗ (r) d3 rg(r) Ω 2 = − 2m d3 rf ∗ (r) Ω 2 g(r) − f ∗ (r) (3.22) ˆ where H is deﬁned in (3.24) Using Green’s theorem1 Ωd 3r f ∗ (r) 2 g(r) − g(r) 2 f ∗ (r) = one obtains the identity ∂Ω da · ( f ∗ (r) g(r) − g(r) f ∗ (r) ) (3.23) ˆ Since H is a diﬀerential operator the expressions (3.20). for example.26) where ∂Ω da · A(r) denotes an integral over the surface ∂Ω of the volume Ω. The reader is advised to consult a reference text on ‘Linear Algebra’ to follow this argument. . t). diﬀer from each other. (John Wiley.

29. t)∂t ψ ∗ (r. t)∂t ψ(r. e. obey (3. A most common boundary condition is encountered for the volume Ω = R3 in which case one postulates lim f (r) = 0 “natural boundary condition” . The probability to observe the particle anywhere in the subvolume ω ⊂ Ω is p(ω. (3. t)|2 . In either case does f (r) and all of its derivatives vanish asymptotically. . namely. g|r) vanish on ∂Ω. (3.. κ > 0 or like r−α . to avoid discontinuities in ψ(r. i. in this case also all derivatives of f (r) vanish at inﬁnity. r ∈ ∂Ω (3. if f and g satisfy these conditions than also does any linear combination αf + βg.29) In fact. 3. t) is ∂t p(ω. |ψ(r. (3. α > 2.3 Particle Flux and Schr¨dinger Equation o The solution of the Schr¨dinger equation is the wave function ψ(r.32) the wave function must decay for |r| → ∞ rapidly enough to be square integrable.30) Note that these boundary conditions are linear in f . i. . t) = ω d3 r |ψ(r. t) ] .32). Often the closed surface of a volume ∂Ω is the union of disconnected surfaces3 ∂Ωj . t) on a single connected surface ∂Ωj only either one of the conditions (3. r ∈ ∂Ω or da · f (r) = 0 ∀r.31) usually arises when a particle existing in a bound state is described. in which case ∂Ω1 is the inner sphere and ∂Ω2 is the outer sphere. t)|2 . ∂Ω = ∂Ω1 ∩ ∂Ω2 ∩ ∂Ω3 ∩ .e. . 3. The latter property stems from the fact that the boundary condition (3.e..34) An example is a volume between two concentric spheres..33) The time derivative of p(ω. like exp(−κr). t) which describes the state of o a particle moving in the potential U (r. t) + ψ(r. i. and that the density decays rapidly when one moves away from that area. In this case one can postulate both conditions (3. Since the total probability of ﬁnding the particle anywhere in space is d3 r|f (r)|2 = 1 (3.31) |r|→∞ (3. It must hold then for all f ∈ H f (r) = 0 ∀r.3: Particle Flux 55 and.. t).30) can be postulated.3.g. we can only allow functions which make the diﬀerential da · P (f ∗ . t) = ω 3 d3 r [ ψ ∗ (r.29. 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.e.30) each condition holding on an entire surface ∂Ωj . therefore. The observable directly linked to the wave function is the probability to ﬁnd the particle at position r at time t. However.

Accordingly the probability to observe the particle anywhere in the total volume Ω is constant.56 Using (3. (3.37) If one applies this identity to ω = Ω one obtains according to (3.41) where j(r. t) = H ψ ∗ (r.37) can be expressed through a volume integral according to da · A(r) = ∂ω ω (3. Note that for a normalized wave function the quantity ρ(r. t) = ∂ω da · P (ψ ∗ (r. Using (3. (3. t) H ψ ∗ (r. 3.27) one can express this j(r. t) .41) holds for any volume ω ⊂ Ω one can conclude ∂t ρ(r. t)r. t) + 4 · j(r. t) = 0 . (3.27) this can be written i ∂t p(ω. t) − ψ(r.18) by any complex number and accordingly one can deﬁne ψ(r.30) ∂t p(Ω. t) = 0. ψ(r.43) (3. A natural choice for this constant is 1. t) + ω · j(r. t) = |ψ(r. t) (3. t) . t) = P (ψ ∗ .19) and its conjugate complex4 −i yields i ∂t p(ω.35) d3 r .44) ˆ Note that the Hamiltonian H involves only real terms.29. t) ψ ∗ (r. ψ|r. t)|2 is a probability density with units 1/volume. . t) = 0 (3.26. t) − ψ ∗ (r. t) such that d3 r|ψ(r. t) ] . One can multiply the solution of (3. t).38) holds. t) = ω Schr¨dinger Equation o ∂ ∗ ˆ ψ (r. The surface integral (3. (3.37) d3 r ∂t ρ(r.39) d3 r · A(r) (3. t) H ψ(r. t)|2 = 1 Ω (3. One refers to such solution as normalized. t) = [ ψ(r. We will assume in the remainder of this Section that the solutions discussed are normalized.36) According to (3.40) One can rewrite then (3. t) ψ(r.42) 2mi Since (3. t) ∂t ˆ ˆ ψ ∗ (r. 3.

18) in Ω∞ = R in the case o U (r.148. Note that j(r. One often encounters wave functions of the type φ(r) = f (r) ei k·r The corresponding ﬂux is j(r) = k 2 f (r) . t) is that of density ﬂux. 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. It is of interest to note from (3.71)]. t) gives rise to a descrease of the total probability in volume ω due to the disappearence of probability density at the surface ∂ω.f.46) i. i. (3. 2m (3. m (3.49) ˜ Obviously. (3. One can readily show by insertion into (3. t) (3.45) Obviously.167)] potential. the initial condition at ψ(r. equal to the velocity of the particle. (3. (2.48) that the general solution is of the form ψ(r.105.e. t) has vanishing ﬂux anywhere. t) = − ψ(r.43) that any real wave function ψ(r. This ﬁnding is consistent with the present evaluation of the particle ﬂux: a factor exp(ipo xo / ) gives rise to a ﬂux po /m.47) .f.e. 3.51) .4: Free Particle 57 The interpretation of j(r. arises solely from the complex factor exp(i k · r). 2 for a free particle [c. (2. for f (r) ∈ R. In Sect. (3.3. (2. t0 ) = [2π]− 2 Ω∞ 3 ˜ d3 k φ(k) exp i(k · r − ωt0 ) .48. t) = − ∂ω da · j(r.41) written in the form ∂t ω d3 r ρ(r. and for particles moving in a linear [c. t) = 0 2 ∂ 2 i ψ(r. j(r.48) ∂t 2m which describes the motion of free particles.50) 3 ˜ d3 k φ(k) exp i(k · r − ωt) Ω∞ (3. This follows directly from an inspection of Eq. t0 ) determines φ(k).f. t) points in the direction to the outside of volume ω. t) .125)] and in a quadratic [c. 2. Equation (3. Such case arose in Sect.. t) = [2π]− 2 where the dispersion relationship holds ω = k2 . 2. 2 we had demonstrated that a factor exp(ipo xo / ) induces a motion of 1-dimensional wave packets such that po /m corresponds to the initial velocity..49) reads at t = t0 ψ(r. 2.4 Solution of the Free Particle Schr¨dinger Equation o We want to consider now solutions of the Schr¨dinger equation (3.

56) and using (2. (3.31) applies. t0 ) ψ(r0 .47). i.54) yields (2.53) where φ(r. In case of the harmonoc oscillator the expansion can be stated in a closed form given in (4. i.51. t|r0 .247). In fact. t|r0 . that the “natural boundary condition” (3.49). t) as given by (3. We have identiﬁed then with (3.54) This expression obviously has the same form as postulated in the path integral formulation of Quantum Mechanics introduced above.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. If one chooses the initial state f (r) deﬁned in (3. The ensuing solutions are called wave packets.54) an alternative representation of the propagator (2. To show this one needs to note 1 2π = +∞ dk1 exp −∞ ik1 (x − x0 ) − 1 2 i 2 k2 1 2m .e. 3. t0 ) .49. The general form for this propagator involves an expansion in terms of a complete set of eigenfunctions as in (3.70) derived below for a particle in a box and the harmonic oscillator..54) valid for this case. respectively.52) above ψ(r.49) is square integrable at all subsequent times and.54) to the case of non-vanishing potentials U (r). 3.58 The inverse Fourier transform yields 3 ˜ φ(k) = [2π]− 2 Schr¨dinger Equation o d3 r0 exp(−ik · r0 ) ψ(r0 .. t0 ) (3. t0 ) = 1 2π 3 d3 k exp Ω∞ ik · (r − r0 ) − i 2 k2 2m (t − t0 ) . derive an expression similar to (3. hence.51) to be square integrable it follows according to the properties of the Fourier–transform that ψ(r.e.81) .5). (t − t0 ) (3. Below we will generalize the propagator (3.52) We have not speciﬁed the spatial boundary condition in case of (3. Ω∞ (3. t) = Ω∞ d3 r0 φ(r.114) and (4. in (2. Comparision with Path Integral Formulation One can write solution (3.47). The solution as stated is deﬁned in the inﬁnte space Ω∞ = R . evaluating the integral in (3.

3.29. The remaining integration over k leads to the integral +∞ −∞ dk exp ikx − 1 2 (k − po 2 2 ) δ − i k2 m (t − t0 ) (3.55.e. (3. (3.53. the functions obey boundary conditions of the type (3. The 1-dimensional version of (3.19.61) × exp − (x − po m t)2 2δ 2 (1 + 2 t2 ) m2 δ 4 (1 − i .60) =???? Combining (3..20) which are of o the form ψ(r.62) We will restrict the space of allowed solutions to a volume Ω such that the functions also make the concomitant (3.3. t) is given by (??). t0 ) ψ(x0 . t0 ) = 1 2π +∞ dk exp −∞ ik (x − x0 ) − i 2m (t − t0 ) .52) to the case that the initial state of a 1-dimensional free particle ψ(x0 . We have demonstrated then in this case that the Schr¨dinger formulation of Quantum o Mechanics is equivalent to the Feynman path integral formulation.57) where φ(x.57–3. the boundary conditions are φ(r) = 0 ∀r. a result which is identical to the expression (??) obtained by means of the path integral propagator (2. r ∈ ∂Ω or da · φ(r) = 0 ∀r.59) which is solved through completion of the square in the exponent [c. t|x0 .64) (3. t|x0 . 3. 3.56)].f. 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. t) 2 k2 (3. t) = −∞ dx0 φ(x.49.27) vanish on the surface ∂Ω of Ω. (3.46). t) = f (t) φ(r) . Stationary States We consider now solutions of the time-dependent Schr¨dinger equation (3. 3. Accordingly.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 (3.60) yields with t0 = 0 ψ(x.30). i.63) . (3. r ∈ ∂Ω (3.54) is +∞ ψ(x.4: Free Particle Free Particle at Rest 59 We want to apply solution (3.

65) is constant. hence. n = 1. . one factor depending only on the time variable and the other only on the space variables are called separable in space and time..69) If these two equations can be solved simultaneously a solution of the type (3.62) is |ψ(r.19).e. It turns out that a solution for f (t) can be found for any E. . It follows then from (3. namely f (t) = f (0) exp − i Et . the eigenvalues of the operator H. We will encounter many such problems in the subsequent Sections. i.69) exist. t) = Ω Ω d3 r |ψ(r.68) ˆ where g1 (t) = i ∂t f (t). 2. a common case is Ω = Ω∞ and the ‘natural boundary condition’ (3. by no means all solutions are of this type. We must postulate therefore ∂t f (t) ˆ H φ(r) = E f (t) = E φ(r) .66) i.45) that the total probability d3 r ρ(r. We have then shown that ψ(r.42) the ﬂux j(r.31). if f (t) = eiα .62) have the particular property that the associated probability distributions are independent of time. and h2 (r) = H φ(r)..64).. We will demonstrate that solutions of the type (3. One can conclude that the probability density for the state (3. α ∈ R .60 Schr¨dinger Equation o and aﬀect only the spatial wave function φ(r). (3. As pointed out above. In fact.27) holds and. This yields an expression g1 (t) h1 (r) = g2 (t) h2 (r) (3. often only for a discrete set of values En .62) which consist of two factors. t) vanishes on the surface of ∂Ω.62) exists. g2 (t) = f (t). We may note in passing that solutions of the type (3.70) The task of ﬁnding solutions φ(r) which solve (3. (3.e.e. in general. (3.67) (3. solutions exist only for a set of discrete E ˆ values. We want to demonstrate this property now. We denote the corresponding solution by φE (r). i. . t) = f (0) exp − i E φE (r) where ˆ H φE (r) = E φE (r) . according to (3. At this point we will accept that solutions φ(r) of the type (3. The identity (3.71) . h1 (r) = φ(r). however.63. for the eigenvalue problems in the conﬁned function space.68) can hold only for all t and all r if g1 (t) = E g2 (t) and E h1 (r) = h2 (r) for some E ∈ C. the solutions o (3. for functions required to obey boundary conditions (3. is time-independent. One calls such states stationary states.62) we insert it into (3. It is important to realize that the separable solutions (3. t)|2 = |f (t)|2 Ω d3 r |φ(r)|2 (3. (3. At this point we state without proof that. t)|2 = |φ(r)|2 . It follows from the observation that for the solution space considered (3. 3.62) are special solutions of the timedependent Schr¨dinger equation.69) is called an eigenvalue problem.62) do exist and we will characterize the two factors of the solution f (t) and φ(r). This can hold only if |f (t)| is time-independent. In order to further characterize the solution (3.

− 2 2m φ(r) = E φ(r) .75) 2m One can ascertain this statement by inserting the expression for φk (r) into the eigenvalue problem posed in (3. We want to prove now that the eigenvalues E which arise in the eigenvalue problem (3. t) = exp − i 2 k2 2m t exp i k · r (3. Stationary State of a Free Particle We consider now the stationary state of a free particle described by i ψ(r.76) has kinetic energy 2 k 2 /2m. We start our proof using the property (3..66) E must be real. The corresponding stationary solution ψ(r. k ∈ R 2 k2 φk (r) = N exp i k · r . of course. e. in fact. one can interpret then k as the magnitude of the momentum of the particle.20). As a result we cannot postulate in the present case that wave functions are localized and normalizable.71) this yields E Ω d3 r φ∗ (r) φE (r) = E ∗ E Ω d3 r φE (r) φ∗ (r) . The ﬂux corresponding to (3. real.73) from which follows E = E ∗ and. .3. = E. o According to (3. The resulting total energy values E are positive. m (3.43) is j(r.74) using exp i k · r = ik exp i k · r . a property which is to be expected since the energy is purely kinetic energy which.72) According to (3. particles scattered of a potential. (3.g. E ∈ R.28) for the special case that f and g in (3. (3.74) The classical free particle with constant energy E > 0 moves without bounds in the space Ω∞ .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 . E (3. The solution φE (r) corresponding to the eigenvalue problem posed by (3.74) is actually best labelled by an index k. t) = exp − E t 2 φE (r) . We will wave this assumption as we always need to do later whenever we deal with unbound particles. i.e. We will show in Section 5 that E can be interprerted as the total energy of a stationary state. should be positive. ˆ d3 r φ∗ (r) H φE (r) = E Ω Ω ˆ d3 r φE (r) H φ∗ (r) . 3.4: Free Particle 61 is a solution of the time-dependent Schr¨dinger equation (3. E (3.71) are.19.76) according to (3. hence. t) = |N |2 k .28) both represent the state φE (r). Obviously.

Solutions of the Schr¨dinger Equation in F1 o The time–dependent solutions satisfy i ∂t ψ(x.84) . We assume.82) satisfy the diﬀerential equation in (3. therefore. two types of solutions.82) have the property that either both boundary conditions are satisﬁed or none. 3. In case of the even solutions (3.81) 2m = E. .e. The boundary conditions which according to (3.5 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. a] ⊂ R assuming that the potential outside of this interval is inﬁnite.81. 3. so-called even solutions obeying φ(x) = φ(−x) φE (x) = A coskx . φ(±a) = 0 .78) (3. a] ⊂ R which vanish on the surface ∂Ω1 = {−a.o) (−a) = 0 (3. and so-called odd solutions obeying φ(x) = −φ(−x) φE (x) = A sinkx. Setting up the Space F1 of Proper Spatial Functions The presence of the inﬁnite energy wall is accounted for by restricting the spatial dependence of the solutions to functions f (x) deﬁned in the domain Ω1 = [−a. (o) 2 k2 (e) 2 k2 2m = E (3. 5 . let say the one at x = a. 2m dx2 2 (± ) = } (3.o) (a) = 0 and φE (e. hence. t) = exp(−iEt/ )φE (x) where φE (x) is determined by 2 d2 − φE (x) = E φE (x) . f continuous. t) . . It turns out that this boundary condition can only be satisﬁed for a discrete set of k–values kn . Hence. i. we have to consider only one boundary condition. We will refer to this as a particle in a one-dimensional ‘box’.80) 2m dx2 We note that the box is symmetric with respect to the origin. f ∈ F1 = {f : [−a. (3. a] ⊂ R → R. 3. (3.80) need to be satisﬁed are φE (e..62 Schr¨dinger Equation o 3. (3. We can expect.80).81) they are kn = nπ . t) = − d2 ψ(x. 2a n = 1. . a}. that the solutions obey this symmetry as well.82) One can readily verify that (3.83) The solutions (3. n ∈ N.81.

φ(e) (a. 4.85) (3.87) (Note that. . 2a 1 . (3. .5: Particle in One-Dimensional Box since for such kn 63 nπa nπ = cos 2a 2 In case of the odd solutions (3.89) nπx . 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. . 2a n = 1.90) 2a The wave functions represent stationary states of the particle in a one-dimensional box.o) (a. (3. 2. = 0. The wave functions for the ﬁve lowest energies En are presented in Fig.1).82).94) .3.86). . (3.93) The normalization constants are then An = (3. . . By counting the number of their nodes one can determine the energy ordering of the wave functions.82). are En = 2π2 8ma2 n2 . respectively.81. . 3. (3. 4.92) and for the odd states +a |An |2 −a dx sin2 nπx = |An |2 a = 1 . It is desirable to normalize the wave functions such that +a −a dx |φ(e. x) = An sin n nπx . 3. n = 2.) The Energy Spectrum and Stationary State Wave Functions The energy values corresponding to the kn –values in (3.88) where the energies for odd (even) n–values correspond to the even (odd) solutions given in (3.e. . 2a n = 2. This condition implies for the even states +a |An |2 −a dx cos2 nπx = |An |2 a = 1 . . 2a n = 1. x)|2 = 1 n (3. 3 . . 5 . (3.84. . (3. 6 . .82) only the kn –values cos(kn a) = cos kn = nπ . 5 . x) = An cos n and φ(o) (a.. according to (3.91) holds. 6 .86). . 6 . n = 1.86) satisfy the boundary condition since for such kn nπa sin(kn a) = sin 2a = sin nπ 2 = 0. (3. (3.81) and (3. according to the dispersion relationships given in (3. a n = 2. 4. i. n is assumed to be even.

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

4. g ∈ F1 (3. this integral vanishes.. 3.100) +a −a dx cos (n−m)πx + cos (n+m)πx 2a 2a The periods of the two cos-functions in the interval [−a. 3.. The latter case leads to integrals φn |φm Ω1 = 1 a +a dx cos −a nπx mπx sin .89. In fact.e. n = 1. . 3.98) the elements of B1 must be linearly independent.98) ∞ f |f 5 Ω1 = n=1 d2 . the integrals vanish. 2a 2a (3.94) B1 = {φn (a. . i.98) The latter property is obviously true for n = m. (ii) n. (3. 3.3. x) . m both odd. . In case of n = m we have to consider three cases. the integrand is odd. Similarly. x) = together with the scalar product5 +a 65 (3. and (iii) the mixed case. N ≥ 1.e. m even φn |φm 1 a Ω1 = 1 a +a nπx −a dx sin 2a sin mπx = 2a (3. (3. i.97) form an orthonormal basis set. In case of n.90. the integral arises φn |φm 1 a Ω1 = 1 a +a nπx −a dx cos 2a cos mπx = 2a (3. 2. too. (i) n. . Obviously.95) 1 a cos nπx 2a sin nπx 2a for n = 1. the integral vanishes. .5: Particle in One-Dimensional Box The Stationary States form a Complete Orthonormal Basis of F1 We want to demonstrate now that the set of solutions (3. n = m. it holds φn |φm Ω1 = δn m . 6 . it holds: f |f Ω1 = 0 → f (x) ≡ 0.103) We will show in Section 5 that the property of a scalar product do indeed apply. . one obtains for n. 5 .} where φn (a. n (3. .96) f |g Ω1 = −a dxf (x) g(x) . In particular. Hence we need to consider only the ﬁrst two cases.99) Since in this case the integrand is a product of an even and of an odd function. m odd. x) (3. for ∞ f (x) = n=1 dn φn (a. f. Because of the property (3. for n = 2. m both even. a] are N. . hence.102) holds according to (3.101) +a −a dx cos (n−m)πx − cos (n+m)πx 2a 2a and.

Hence. Evaluating the Propagator We can now use the expansion of any initial wave function ψ(x. . they can be expanded in terms of a Fourier series.} such that ∞ ∞ nπx nπx ˜ = + an sin (3. .105) must vanish. x0 ) ψ(x0 . t0 ) = +a ∞ i n=1 − En φn (a. . x0 ) ψ(x0 . .104) ˜ where [y]a = y mod2a. a] yields then also an expansion for any element in F1 and B1 is a complete basis for F1 . .96). In the present case. x0 ) ψ(x0 . . however. Restricting the expansion (3.. 3.109) Insertion of (3. x) . 4.108) into the Schr¨dinger equation yields o +a ∞ n=1 φn (a. t0 ) = dm . and bn . t0 ) (3. We like to show ﬁnally that the basis (3. t) at times t > t0 .105) to the interval [−a. 2. We have then shown that any f corresponding to elements of F1 can be expanded in terms of elements in B1 . (3. i.66 Schr¨dinger Equation o f (x) ≡ 0 implies f |f Ω1 = 0 which in turn implies dn = 0 since d2 ≥ 0. x) cn (t) −a dx0 φn (a. . . ( ) = ([ + ] − ) (3. i. m = 1.98) one obtains +a dx0 φm (a. t0 ) = n=1 dn φn (a. t0 ) .108) where the functions cn (t) are to be determined from the Schr¨dinger equation (3. in (3.110) .. .e. any element of the function space F1 deﬁned in (3. . (3.106) Using the orthonormality property (3. .106) and generalizing to t ≥ t0 one can write ∞ +a ψ(x.} and {bn . 3. x) ∂t cn (t) −a dx0 φn (a. n = 1. (3. Accordingly. For this purpose we expand ∞ ψ(x.e. there exist real constants {an . 2. 1.95) is also complete.107) Inserting this into (3.78) can be expressed as a linear combination of the elements of B1 deﬁned in (3.95. such demonstration can be based on the theory of Fourier series. . −a (3.95) is an orthonormal basis. This implies that only the trigonometric functions which are elements of B1 enter into the Fourier ˜ series. Demonstration of completeness is a formidable task. n = 2.79) requiring the o initial condition cn (t0 ) = 1 . t) = n=1 φn (a. For this purpose we extend the deﬁnition of the elements of F1 to the whole real axis through ˜ f : R → R. n = 0. The functions f are periodic with period 2a. x0 ) ψ(x0 . x) cn (t) −a dx0 φn (a. It follows that B1 n deﬁned in (3. x) to obtain an expression for ψ(x. 3. 5 . n = 1. . t0 ) in terms of eigenfunctions φn (a. .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. the coeﬃcients an .

a] yields.3. t0 ). 67 (3.54) as in the case of the free particle system which does not exhibit any bound states. x) exp n=1 i − En (t − t0 ) φn (a. The last frame shows the wave front reaching again the right wall and the onset of new interference.108. t0 ) = 1 2πσ 2 1 4 exp − x2 0 + i k0 x0 2σ 2 .5). σ = a 15 . t0 |x0 . rather than by an integral (3.5: Particle in One-Dimensional Box Multiplying both sides by φm (a.114) This expression has the same form as postulated in the path integral formulation of Quantum Mechanics introduced above. t0 ) itself is a solution of the time-dependent Schr¨dinger o equation (3. x) and integrating over [−a. t|x0 . i ∂t cm (t) = − Em cm (t) . Often the propagator φ(x..2 presents the probability distribution of the particle at subsequent times.113). 4 a (3. t|x0 . We have identiﬁed then with (3. but the collision with the left wall leads to new interference eﬀects. x0 ) . Figure 3. t|x0 .109) are cm (t) = exp i − Em (t − t0 ) . In the present system which is composed solely of bound states the propagator is given by a sum. (3. k0 = .e. The particle moves then to the left. i. according to (3. the elements of which satisfy the appropriate boundary conditions. t) = −a dx0 φ(x. being reﬂected at the right wall. cm (t0 ) = 1 . t|x0 .115) This initial state corresponds to the particle being localized initially near x0 = 0 with a velocity v0 = 15 /m a in the direction of the positive x-axis. 3. The interference pattern begins to ‘smear out’ ﬁrst.78). t0 ) = ∞ φn (a. Example of a Non-Stationary State As an illustration of a non-stationary state we consider a particle in an initial state ψ(x0 .111) The solutions of these equations which satisfy (3. It is of interest to note that φ(x.112) Equations (3. t0 ) = δ(x − x0 ) as can be readily veriﬁed using (3. t0 ) (3. . Note that the propagator has been evaluated in terms of elements of a particular function space F1 . t0 ) ψ(x0 . This solution can be written +a ψ(x.114) the representation of the propagator for a particle in a box with inﬁnite walls.98).112) determine now ψ(x.113) where φ(x. t0 ) is also referred to as a Greens function. t) for any initial condition ψ(x. in (2.79) which lies in the proper function space (3. One can recognize that the particle moves ﬁrst to the left and that near the right wall of the box interference eﬀects develop. In case that diﬀerent boundary conditions hold the propagator will be diﬀerent as well. The respective initial condition is φ(x. (3.

t)|2 for a particle in a box starting in a Gaussian distribution with momentum 15 /a.00 m a h2 |Ψ|2 2 t = 0.48 m a h2 |Ψ|2 2 t = 0.68 m a h2 1 πσ 1 πσ -a a -a a |Ψ|2 2 t = 1.72 m a h2 1 πσ 1 πσ -a a -a a |Ψ|2 2 t = 0.68 Schr¨dinger Equation o |Ψ|2 2 t = 0.24 m a h2 1 πσ k = 15 a 1 πσ -a a -a a |Ψ|2 2 t = 0.20 m a h2 1 πσ 1 πσ -a a -a a |Ψ|2 2 t = 1.96 m a h2 |Ψ|2 2 t = 1.92 m a h2 |Ψ|2 2 t = 2.16 m a h2 1 πσ 1 πσ -a a -a a Figure 3. .2: Stroboscopic views of the probability distribution |ψ(x.44 m a h2 |Ψ|2 2 t = 1.

6 Particle in Three-Dimensional Box We consider now a particle moving in a three-dimensional rectangular box with side lengths 2a1 . x3 .117) 2m 2 2 2 ∂1 + ∂2 + ∂3 (3.118) which are stationary states.95). t) = H ψ(x1 .3. x2 . x3 ) is an element of the function space F3 deﬁned in (3. 3 (3.114) in terms of which the solutions for all initial conditions can be expressed.116) and obeys the partial diﬀerential equation ˆ H φE (x1 . φ(j) (±aj ) = 0 .97) with respect to which the eigenfunctions are orthonormal. x2 . a2 ] ⊗ [−a3 . 2a2 . hence. Placing the origin at the center and aligning the x1 . 2. The description employed a space of functions F1 deﬁned in (3. We have deﬁned in the space F1 a scalar product (3. 2a3 . A complete basis of F1 is given by the inﬁnite set B1 (3.6: Particle in Three-Dimensional Box Summary: Particle in One-Dimensional Box 69 We like to summarize our description of the particle in the one-dimensional box from a point of view which will be elaborated further in Section 5. An important property of this basis and.78). (3. x2 .. t) = exp − E t φE (x1 . i. The corresponding solutions have the form i ψ(x1 .120) ˆ Since the Hamiltonian H is a sum of operators O(xj ) each dependent only on a single variable. a3 ] ⊂ R = {(x1 . . x3 = ±a3 } . of the space F1 is that the elements of the basis set can be enumerated by integer numbers. x1 = ±a1 } ∪ {(x1 . This property allowed us to evaluate the propagator (3. ˆ = O(x1 ) + O(x2 ) + O(x3 ). x3 ) (3. x2 .e. ( . ˆ H = − 2 (3. can be counted. x3 ) = j=1 φ(j) (xj ) (3. ) = ∀( .122) . x2 . one can express H 3 φ(x1 .121) where − 2 2m 2 ∂j φ(j) (xj ) = Ej φ(j) (xj ) . x3 ) = E φE (x1 . f continuous. x2 . x2 = ±a2 } ∪ {(x1 . . 3. )T ∈ ∂ } . We seek then solutions of the time-dependent Schr¨dinger equation o ˆ i ∂t ψ(x. t) . i. a1 ] ⊗ [−a2 . x2 . x2 . x3 )T ∈ Ω. x3 )T ∈ Ω. (3.. x3 . x3 ) .116) where Ω is the interior of the box and ∂Ω its surface Ω ∂Ω = [−a1 . x2 . x3 –axes with the edges of the box yields spatial boundary conditions which are obeyed by the elements of the function space F3 = {f : Ω → R.119) where φE (x1 . x2 .e. j = 1. x3 )T ∈ Ω. x2 .

a2 . n 3 and E(n1 . 2. t|r0 . . 2. (3. a2 . 2. .n3 ) (a1 . 3. n2 .} ∞ (3. x1 ) φn2 (a2 .n2 .n3 ) = 2π2 = φn1 (a1 . a2 . a3 .122) are given by (3. a2 .n2 .n2 . if all three lengths a1 . For example.n2 . x3 ) n1 ..125) is a complete orthonormal basis of F3 and that the propagator for the three-dimensional box is φ(r. hence.n3 ) (a1 .n3 ) (a1 . a3 .126) Symmetries The three-dimensional box conﬁning a particle allows three symmetry operations that leave the box unchanged. . The energies and corresponding degeneracies of the particle in the three-dimensional box with all side lengths equal to 2a are given in the following Table: n1 1 1 1 1 2 1 2 1 2 3 1 n2 1 1 2 1 2 2 2 1 3 3 1 n3 1 2 2 3 2 3 3 4 3 3 5 E/[ 2 π 2 /8ma2 ] 3 6 9 11 12 14 17 18 22 27 27 degeneracy single three-fold three-fold three-fold single six-fold three-fold three-fold three-fold single three-fold One can readily verify that the symmetry of the box leads to three-fold and six-fold degeneracies. x3 –axes by π/2 also leaves the system unaltered. x1 . . x2 . i.n2 . a3 .124) The same considerations as in the one-dimensional case allow one to show that B3 = {φ(n1 . x3 –axes. x1 . Such degeneracies are always a signature of an underlying symmetry.e. 3. x2 . (3. a3 are identical.96) and. . in the present case . n1 . n3 = 1. . r0 ) . .70 Schr¨dinger Equation o and E1 + E2 + E3 = E. (3.n3 =1 φ(n1 . r) exp i − E(n1 n2 n3 ) (t − t0 ) φ(n1 . namely rotation by π around the x1 . n 2 .n3 ) (a1 .n2 . . a2 . This symmetry has been exploited in deriving the stationary states. If two or all three orthogonal sides of the box have the same length further symmetry operations leave the system unaltered. x2 ) φn3 (a3 . x3 ) .80) shows that the solutions of (3. t0 ) = n1 . the solutions of (3. a3 . x2 . n2 . x2 . Actually. n1 .123) 8m n2 n2 n2 1 2 3 + 2 + 2 a2 a2 a3 1 . a1 = a2 = a3 = a then rotation around the x1 . n3 = 1.122) with (3. This additional symmetry is also reﬂected by degeneracies in the energy levels. Comparing (3. x3 ) = 1.120) can be written φ(n1 . 3. .

a. However.3. . However.1) (a.. r) + γ φ(2. this linear combination o is not necessarily orthogonal to other degeneraste states. Hence. however. φ(1. to construct n orthogonal stationary states6 . 3.g.2) (a.1.127) ˜ ˆ ˜ obeys the stationary Schr¨dinger equation H φ(r) = E122 φ(r). in case of an n–fold degeneracy it is always possible.2. r) + β φ(2. r). a. 6 This is a result of linear algebra which the reader may ﬁnd in a respective textbook. One particular aspect of the degeneracies illustrated in the Table above is worth mentioning.6: Particle in Three-Dimensional Box 71 ‘accidental’ degeneracies also occur. in case of degenerate states one cannot necessarily expect that stationary states are orthogonal. the identity 32 + 32 + 32 = 12 + 12 + 52 . e.2) (a. a. n2 . due to the hermitian character of ˆ H. Any linear combination of wave functions ˜ φ(r) = α φ(1. r) (3. 1. The origin of this degeneracy is. 5) as shown in the Table above. a. a.2. a. a. for (n1 .2.2) (a. We consider the degeneracy of the energy E122 which is due to the identity E122 = E212 = E221 . a. for example. n3 ) = (3. 3) (1.

72 Schr¨dinger Equation o .

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

however. We will point out explicitly where assumptions are made which built on this restriction. In the following algebraic solution of (4.5) →±∞ where C∞ denotes the set of functions which together with all of their derivatives are continuous.3) Due to the nature of the harmonic potential which does not allow a particle with ﬁnite energy to move to arbitrary large distances.5) ˆ H = 1 1 2 p + m ω 2 x2 ˆ ˆ 2m 2 (4. only for a discrete set of E values can the boundary conditions (4.2) can be expressed in terms of the two operators p = ˆ d . is the commutation property 1 1 (4. The cardinal property exploited below. This property states that p and x obey an ˆ ˆ algebra in which the two do not commute. beside the representation (4. ∈ C∞ . The stationary states of the harmonic oscillator have been considered already in Chapter 2 where the corresponding wave functions (2.7) of the Hamiltonian.235) had been determined. The operators act ˆ on the space of functions N1 deﬁned in (4.7) which is why these operators are of interest to us. the spectrum of ˆ H in (4.1) of the form ψ(x.8) i which holds for any position and momentum operator.1 Creation and Annihilation Operators The Hamiltonian operator (4.5).74 Linear Harmonic Oscillator In the following we consider ﬁrst the stationary states of the linear harmonic oscillator and later consider the propagator which describes the time evolution of any initial state. the natural boundary conditions apply x→±∞ lim φE (x) = 0 . If this restriction would not apply and all functions f : R → R would be admitted. i dx x = x ˆ (4. t) = exp(−iEt/ )φE (x. the commutator has a simple form.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. (4. in which the operators used below. therefore.3) would be continuous and the eigenfunctions φE (x) would not be normalizable. however. In order [ˆ.4) Equation (4. act. It is important to keep in mind this restriction of the space. 4. The Hamiltonian H can be expressed in terms of the operators acting on the space (4. x] = p ˆ . all stationary states of the harmonic oscillator must be bound states and.3) can be solved for any E ∈ R. (4.4) be satisﬁed. t) where ˆ H φE (x) = E φE (x) .6) the ﬁrst being a diﬀerential operator and the second a multiplicative operator. lim ( ) = } (4. In the framework of the Schr¨dinger equation the stationary o states are solutions of (4.

10) we deﬁne a+ = ˆ mω i x − √ ˆ p. ˆ 2 2m ω a− = ˆ mω i x + √ ˆ p.15) 1 ˆ 1 H + 1 . p ˆ 2 2m ω 2 (4. a ˆ 1 (4.7) assuming that the factors are easier to handle than the Hamiltonian in yielding spectrum and eigenstates. Since a+ and a− are ˆ ˆ + and a− since operators and not scalars we cannot simply expect that the identy (4. 1 ω 2 (4. in fact.1: Creation and Annihilation Operators 75 to derive (4.15) together lead to the commutation property (4.9) From this follows (4. the action of [ˆ. ˆ ˆ . hence.12) To prove this commutation property we determine using (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 .14) 1 i mω 2 x + ˆ p2 + ˆ [ˆ. x] .11) ˆ The reader may note that we have attempted. a+ ] = a− a+ − a+ a− does not necessarily vanish.8) we recall that all operators act on functions in N1 and.12). 4. 2mω dx (4. 1 2 (4.17) d dx We also express a+ and a− directly in terms of the coordinate x and the diﬀerential operator ˆ ˆ a+ = ˆ mω x − 2 d .14) and (4. 1 i dx i dx i i (4. a+ ] = 1 .11) a− a+ = ˆ ˆ (4.13) (4. Before we continue we like to write (4. One obtains [ˆ. In fact.4. x] f (x) = p ˆ d d x f (x) − x f (x) = f (x) = 1 f (x) .16) (4. 2mω dx a− = ˆ mω x + ˆ 2 d .14.7) and (4. the commutator property (4.8) a ˆ ˆ ˆ ˆ ˆ implies [ˆ− . ˆ 2 2m ω (4. to factor H/ ω. Being guided by the identity for scalar numbers (b − ic)(b + ic) = b2 − i(cb − bc) + c2 = b2 + c2 (4.18) It is of interest to note that the operators a+ and a− are real diﬀerential operators.8). x] p ˆ on such functions must be considered.8) yield a− a+ = ˆ ˆ Similarly one can show a+ a− = ˆ ˆ 1 ˆ 1 H − 1 .10) holds for a ˆ ˆ [ˆ− . Our next step is an attempt to factorize the Hamiltonian (4. 1 ω 2 (4.

76 Relationship between a+ and a− Linear Harmonic Oscillator The operators a+ and a− are related to each other by the following property which holds for all ˆ ˆ functions f. one obtains using (4.23) Together.19). 4.17.17.16.12. 4. it holds that for a stationary state φE (x) of energy E deﬁned through (4. (4. 4. ˆ (4.16. 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. (4.20) ˆ In the following we will determine the spectrum of H and its eigenstates. 4.16.17. 4. The derivation will be based solely on the properties (4. The property ˆ ˆ follows directly from (??). Using (??) we like to state (4.19) in the form1 f |ˆ+ g a Ω∞ = g|ˆ− f a Ω∞ .24) − m This property states that the operators in the function space N1 are the hermitian conjugate of each other. This property of operators is investigated more systematically in Section 5. (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. 4. In fact.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) .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. ˆ The results (4.21) ˆˆ H a− φE (x) ω 1 ) a− φE (x) 1 ˆ 2 ω = a− ( ω a+ a− + ˆ ˆ ˆ 1 − ω 1 ) φE (x) 1 1 2 ˆ = a− ( H − ω ) φE (x) ˆ = ( ω a− a+ − ˆ ˆ = (E − ω ) a− φE (x) .22. .21) is available. g ∈ N1 +∞ +∞ dx f (x) a+ g(x) = −∞ −∞ dx g(x) a− f (x) .22) Similarly one can show using again (4.19) This property states that the operators a+ and a− are the adjoints of each other.

(4.4. Let φE (x) and φE (x) be two normalized stationary states corresponding to two diﬀerent energies E. on ˆ the other side such a state is itself an eigenstate of H as can be shown using (4.29) where y = mω x. (4. +∞[ one can write (4.2 Ground State of the Harmonic Oscillator d + y dy A suitable solution of (4. 2 2 (4.) ˆ ˆ 0 = φE |HφE Ω∞ − φE |HφE Ω∞ = ( E − E ) φE |φE Ω∞ ..29) d 1 d φ0 (y) = = lnφ0 (y) = − y . φ0 (y) dy dy the solution of which is 1 lnφ0 (y) = − y 2 + c0 2 φ0 (y) = c0 exp 1 − y2 2 . There arise. For obvious reasons one refers to a+ and a− as ladder ˆ ˆ operators. (4. a state φ0 (x) which obeys the property a− φ0 (x) = 0 ˆ (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.25) on one side would lead to termination of the sequence E0 + m.e. two diﬃculties: (i) one needs to know at least one stationary state to start the construction.32) for some constant c0 or (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 (4.25) can. m ∈ Z when m is decreased.31) (4.5). (4.27) Since E = E one can conclude φE |φE Ω∞ = 0. It turns out that both diﬃculties can be resolved together.33) .25) needs to be normalizable in order to represent a bound state of the harmonic oscillator. in fact. in fact. Using (4.25) φ0 (y) = 0 (4. In fact. E .6. The property (??) has an important consequence for the stationary states φE (x).28) 4. however. 4. For f (x) = φE (x) and g(x) = φE (x) in (??) follows (Note that according to (??) the eigenvalue E is real. be found. E = E . i.17) ˆ H φ0 (x) = ( ω a+ a− + ˆ ˆ 1 ω 1 ) φ0 (x) = 1 ω φ0 (x) .26) Of course. (ii) the construction appears to yield energy eigenvalues without lower bounds when.11) one can rewrite (4. φ0 (x) should be an element of the function space N1 deﬁned in (4. one expects that E = 0 should be a lower bound.30) Assuming that φ0 (y) does not vanish anywhere in its domain ] − ∞.

. .39) We notice that the ground state wave function φ0 (x) as well as the operators (ˆ+ )n are real. 2. 2.e.36) n times. which for a narrowly localized state yields a large positive value. (4. . it holds ˆ H φn (x) = ω(n + 1 ) φn (x) .24) can then be written as follows En = ω(n + 1 ). The conventional normalization condition φ0 |φ0 reads +∞ +∞ Ω∞ = 1 (4. .. would yield a vanishing contribution.26) the energy value associated with this state is 1 ω. The set of allowed energies of the oscillator according to (4. n = 0. a state conﬁned to the minimum corresponding to the classical state of lowest energy. .2) there are two competing contributions to the energy. We recall that according to (4. mω (4. 2 n = 0. 2.36) Since the ﬁrst order diﬀerential equation (4.38) These states correspond to the energy eigenvalues (4..78 Linear Harmonic Oscillator This solution is obviously normalizable. (4. 4.3 Excited States of the Harmonic Oscillator Having obtained a suitable stationary state with lowest energy...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 n = 0.39). the so-called excited states.36) assumes a functional form such that both terms together assume a minimum value. i. We will consider this point more systematically in Section 5. 1.25) admits only one solution there is only one set of states with energy E + m ω. we construct the states for n = 1.35) The appropriate ground state is φ0 (x) = mω π 1 4 exp − mωx2 2 . For this purpose we apply the operator a+ to the ˆ ground state (4. . . . . and the kinetic energy contribution. 1. we can now construct the stationary states corresponding to energies (4. The ground state (4.e. the potential energy contribution which for a state δ(x). 2 This state of lowest energy is called the ground state of the oscillator. m ∈ Z which properly terminate at some minimum value E + m0 ω ≥ 0. Such states need to be suitably normalized for which purpose we introduce a normalization constant cn φn (x) = cn a+ ˆ n φ0 (x) .e. The reason is that in the Hamiltonian (4.34) dx |φ0 (x)|2 = |c0 |2 −∞ −∞ dx exp − mωx2 2 = |c0 |2 π . . 2. .37) above the ground state energy. (4. . i. therefore. choose the normalization constants cn and the functions φn (x) real as well. 1. (4. . We a can. i.

18)] and since the ˆ φn (x) are real functions. To determine these constants we adopt a recursion scheme.45) form an orthonormal set.45) We like to note that these functions according to (4. 1.40) φn−1 (x) = βn a− φn (x) ˆ (4..e. according to (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∞ (4.14) 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. . 4.41) Employing the adjoint property (4. 2.40–4. (4. (4. (4.4. they obey φn |φn = δn n . √ a− φn (x) = n φn−1 (x) . To determine βn we note using (4.3: Excited States 79 We need to determine now the normalization constants cn . For n = 0 holds c0 = 1.42) ˆ Using (4.48) (4. n = 0.39) yield √ From this follows βn = 1/ n and. ˆ (4. n = 0.15 . βn must be real as well.49) (4.40) = 1.50) φn (x) .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) .43) (4.20) 1 = φn−1 |φn−1 Equations (4. n.47). .46) According to (4. 1. . i. Since a− is a real diﬀerential operator [see (4. (4. ˆ Repeated application of this relationship yields φn−s (x) = (n − s)! − a ˆ n! s 2 1 = βn n φn |φn Ω∞ 2 = βn n . 2.51) . . according to (4. √ a+ φn (x) = n + 1 φn+1 (x) . We consider then the situation that we have obtained a properly normalized state φn (x).47) for some suitable constants βn .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∞ . (4. .40).28) and the construction (4. . (4. .24) holds in analogy to (4.

28) of the ground state and the deﬁnition (4.40–4.e. 2.54) The eigenstates of the Hamiltonian are then given by φn (y) = − √ e n n! π 2 1 y2 2 e y2 2 y − d dy n e− y2 2 . (4. however. by dx = dy mω 1 4 .35) rather than (4.53) We will later re-introduce this factor to account for the proper normalization (4. (4.45). We will demonstrate below the identity Hn (y) = e y2 2 y − d dy n e− y2 2 (4. (4.7.39). For this purpose we start from expression (4.57) 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) .55) 1 y2 2 (4. 2. (4. namely the square root of the Jacobian dx/dy. by introducing the variable y deﬁned in (4. ..233) derived in Sect.200).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. i. n–independent factor.46) yields a set of normalized states.7 and given. It should be noted that √ the normalization (4. (4. for example. by the Rodrigues formula (2.56) Relationship to Hermite Polynomials We want to demonstrate now that the expression (4.52) This normalization of the wave functions diﬀers from that postulated in (4.52).56) can be expressed in terms of Hermite polynomials Hn (y) introduced in Sect. 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 .58) This result agrees with the expression (2.35) by a constant.57) which includes the factor 1/ n! according to Eqs.30) and employing the normalization +∞ −∞ dy φ2 (y) = 1 n (4. simplifying the calculation.

63) one obtains g(y) = y − d dy f (y) (4. (4. i.61) One can factor g(y) and employ (4. 4.59) holds for n = 0.59) hold.30) and return later to the coordinate x.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 . 1.4: Propagator 81 To verify the realtionship between the deﬁnition (4. therefore.39) ∞ ψ(y0 . (4.1. g(y) = (−1)n+1 e y2 2 dn+1 −y2 e = dy n+1 y − d dy n+1 e− y2 2 .e. the Rodrigues formula (2.57) of the hermite polynomials and the deﬁnition given by (2. 4. Our derivation will follow closely the procedure adopted for the case of a ‘particle in a box’ [see Eqs.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 .64) which implies that (4.62) Denoting f (y) = (y − d/dy)exp(−y 2 /2) and employing y2 y2 d d − y2 e 2 f (y) = −y e− 2 f (y) + e− 2 f (y) dy dy (4. dy dy (4.61) and. Starting point of our derivation is the assumption that the initial condition can be expanded in terms of the eigenstates φn (y) (4. t0 ).114)].4 Propagator for the Harmonic Oscillator We consider now the solution of the time-dependent Schr¨dinger equation of the harmonic oscillator o (4.65) .4.2) for an arbitrary initial wave function ψ(x0 . For the sake of notational simplicity we employ initially the coordinate y as deﬁned in (4.200).60) and.59) by induction noting ﬁrst that (4. t0 ) = n=0 dn φn (y0 ) .59) dn −y2 e dy n (4. hence. dy n (4. and showing then that the property also holds for n + 1 in case it holds for n. (4..106–3. (3. We prove (4.

pp. Employing orthogonality condition (4. y0 .39) holds cn (t1 ) = exp −i ω ( n + 1 ) ( t1 − t0 ) 2 . t0 ) φn (y0 ) . t0 ) which is an element of N1 deﬁned in (4.66) One can extend expansion (4. y0 . t0 ) ψ(y0 . For this purpose we start from the integral representation (2.7.67) for which according to (3.147) for the harmonic iscillator determined in Sect.65) to times t ≥ t0 through insertion of time-dependent coeﬃcients cn (t1 ) ∞ ψ(y.15.70) and (2. 1 t = e−iω(t1 −t0 ) (4. 2 A demonstration of this property can be found in Special Functions and their Applications by N. We want to demonstrate now that this propagator is identical to the propagator (2. t0 ) = n=0 φn (y) φn (y0 ) tn+ 2 . t1 ) = n=0 dn φn (y) cn (t) (4. t|y0 . t) = π −3/2 exp 2 2 y 2 y0 + 2 2 2 +∞ +∞ ∞ du −∞ −∞ dv n=0 1 ( −2tuv )n × n! (4.70). t) (4. t1 |y0 .68) Altogether one can express then the solution +∞ ψ(y. Inc.82 Linear Harmonic Oscillator Such expansion is possible for any f (y0 ) = ψ(y0 .J.46) the expansion coeﬃcients dn are +∞ dn = −∞ dy0 ψ(y0 . We consider for this purpose the following expression for |t| < 1 ∞ w(y. sec:harm. 1965) Sect.72) × exp( −u − v + 2iyu + 2iy0 v ) .225) which allows one to derive a generating function for products of Hermite polynomials which can be applied to the r.Lebedev (Prentice Hall. In order to prove the equivalence of (4.N.. N. t) = n=0 Hn (y)e−y /2 Hn (y0 )e−y0 /2 n √ t . a supposition which is not proven here2 . . 2n n! π 2 2 (4.112) and (4. 2. Englewood Cliﬀs.70) is the propagator of the linear harmonic oscillator. 68.5).147) we employ the technique of generating functions as in Sect. this is an excellent textbook from which we have borrowed heavily in this Section. (4.108–3.. of (4. t1 ) = −∞ dy0 φ(y. 4.69) where ∞ φ(y.h.s. (4.71) Applying (??) to express Hn (y) and Hn (y0 ) yields exp(−2tuv) w(y.

i. t1 |y0 .h. t) = π −3/2 exp +∞ 83 y2 y2 + 0 2 2 +∞ dv exp( −v 2 + 2iy0 v ) × −∞ × −∞ du exp( −u2 − 2u(tv − iy) ) . indeed. t0 ) = where F (t1 . t) = π −1 exp − 2 y 2 y0 + 2 2 +∞ dv exp(−(1 − t2 ) v 2 − 2i (yt − y0 ) v) .78) in terms of the coordinates x and x0 .s.81) imω 2 sinω(t1 − t0 ) (x2 + x2 ) cosω(t1 − t0 ) − 2 x x0 0 This result agrees with the propagator (2. The resulting propagator is G(t1 . one obtains w(y. t0 ) = and 1 2 i π F (t1 .74) a second time yields ﬁnally together with the deﬁnition (4. by a factor 4 mω/ for both φn (y) and φn (y0 ).78) 1 − e−2iω(t1 −t0 ) = sinω(t1 − t0 ) 2 i e−iω(t1 −t0 ) (4...77).e.30) to replace y and y0 and that we multiply the propagator by mω/ . t) √ 1 π(1−t2 ) 2 exp − 1 (y 2 + y0 ) 1 + t2 + 2 y y0 2 1−t Hn (y)e−y /2 Hn (y0 )e−y0 /2 n ∞ √ t n=0 2n n! π 2 2 2 t 1 − t2 = .4. identical to the generating function (4. y0 .73) The incomplete Gaussian integral +∞ −∞ dx e−a 2 x2 √ − 2bx = π b2 /a2 e . t0 ) = i φ(x. a Re a2 > 0 (4.74) applied once results in w(y. t0 ) = exp mω 2iπ sinω(t1 − t0 ) 1 2 × .75) Applying (4.77) The sum in (4. (4.147) derived by means of the path integral description. t0 ) (4. i. t|x0 .70) is. √ πexp( t2 v 2 − 2iytv − y 2 ) (4.80) −2iω(t1 −t0 ) sinω(t1 − t0 ) 1−e We can ﬁnally express the propagator (4. it holds φ(y.4: Propagator Carrying out the sum on the r. y0 . y0 . t0 ) − i y y0 2 F (t1 .79) 1 + e−2iω(t1 −t0 ) cosω(t1 − t0 ) = .76) One can express this in terms of the stationary states (4. .58) of the harmonic oscillator ∞ φn (y) φn (y0 ) tn = n=0 2 1 t 2 1 + t exp − (y 2 + y0 ) + 2 y y0 2 2) 2 1−t 1 − t2 π(1 − t 1 . −∞ (4. (4. t0 ) exp i 2 1 2 (y + y0 ) G(t1 . (4. This requires that we employ (4.e.71) of w(y. (4.

d/dy and a+ . 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! .e. the approaches a ˆ a ˆ using y. 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.82) a+ = √ ˆ dy dy 2 2 √ √ Obviously. Here f (ν) (y0 ) denotes the ν-th derivative of f (y) taken at y = y0 . 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 − . 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! (4.84) = 1 1 = 0 √ = n + 1 φn+1 (y) √ = n φn−1 (y) = f |ˆ+ g a Ω∞ . We want to introduce now techniques based on the ladder operators a+ and ˆ a− .g. 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. a (4. each corresponding to a single harmonic oscillator of the type studied presently by us. f (ˆ+ ). ˆ ˆ Calculus of Creation and Annihilation Operators We summarize ﬁrst the key properties of the operators a+ and a− ˆ ˆ [ˆ− . 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. and of the modes of the quantized electromagnetic ﬁeld. both quantum systems are endowed with a large number of modes. e. (4.86) f (ˆ+ ) = a a ˆ ν! ν=0 . the use of generating functions.. i. a− must be equivalent. one can express y = 2(ˆ− + a+ ) and d/dy = 2(ˆ− − a+ ) and. (4.85) is convergent everywhere in R. a− = √ ˆ y + .84 Linear Harmonic Oscillator 4.83) We will encounter below functions of a+ and a− . hence. 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.. phonons.

ˆ a a a ˆ In particular. The following important property holds a a− f (ˆ+ ) = f (1) (ˆ+ ) + f (ˆ+ ) a− . f1 (ˆ+ )] = [ˆ− . (ˆ+ )n a+ ] = (ˆ+ )n [ˆ− . The convergence of the Taylor expansion ascertains then that (4.90) holds for any function fn (ˆ+ ) = (ˆ+ )n which is a power of a a a+ .88) we need to show actually [ˆ− . a a a 85 (4. a+ ] + [ˆ− .88) (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.95) (4.87) (4.92) f (y) = n=0 dn φn (y) = n=0 (ˆ+ )n a dn √ φ0 (y) n! (4. f (ˆ+ )] = f (1) (ˆ+ ) .4.90) holds for f0 and for f1 . i. property (4.94) This identity follows from (4.89) (4.96) . f (ˆ+ )] + f (ˆ+ ) a− ˆ a a a a ˆ which implies that in order to prove (4. a a (4. The ﬁrst case is trivial.e.91) (4.. For fn+1 follows then [ˆ− .93) one can reduce all operators acting on some proper state function by operators acting on the state φ0 (y).88) is (ˆ− )n f (ˆ+ ) φo (y) = f (n) (ˆ+ ) φ0 (y) . Note that ˆ a an immediate consequence of (4.90) holds for f (ˆ+ ). We note ˆ a− f (ˆ+ ) = [ˆ− . As long as the operators act on φ0 (y) one can state then that a− behaves like a diﬀerential operator with respect to functions f (ˆ+ ). An example is the so-called shift operator exp(uˆ− ).90) To prove (4. a+ ] = 1 = f1 (ˆ+ ) .87. a a a ˆ 1 a Let us assume that (4. Hence. 4. the second case follows from [ˆ− .90) holds for fn . ˆ a We proceed by induction noticing ﬁrst that (4.83) and can be proven for all powers fn by induction and then inferred for all proper functions f (ˆ± ).5: Working with Creation and Annihilation Operators and similarly for f (ˆ− ). a− f (ˆ+ ) φ0 (y) = f (1) (ˆ+ ) φ0 (y) ˆ a a which follows from a− φo (y) = 0. 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 An important operator function is the exponential function. a a − (4. It holds a a euˆ f (ˆ+ ) φ0 (y) = f (ˆ+ + u) φ0 (y) . we will only assume operator functions acting on φ0 (y). fn+1 (ˆ+ )] a a = [ˆ− .90) we show that (4.

that the operators considered act on φ0 (y).98) The related operators exp[vˆ+ ± v ∗ a− ] play also an important role. (4. + (4.106) yields a eαˆ + ∗ a− ˆ a a eαˆ φ0 (y) = e−αα eαˆ φ0 (y) . ∗ = exp(uzˆ a ˆ and determine its derivative d du ˆ C(u) a = zˆ+ euzˆ euz a + ∗ a− ˆ a + euzˆ z ∗ a− euz ˆ + ∗ a− ˆ = + zˆ+ + z ∗ a− a ˆ Using (4.101) ˆ ˆ ˆ C(u) obviously obeys C(0) = 1 This results in D(0) = 1 and.100) ˆ ˆ To solve this diﬀerential equation we deﬁne C(u) = D(u) exp(−u2 /2) which leads to d ˆ D(u) = du zˆ+ + z ∗ a− a ˆ ˆ D(u) (4. z ∈ C.98) for u = −α∗ and v = α yields e−α This together with (4. hence. Below we will use the operator identity which follows for the choice of v = α in (4.101) 1. + ∗ + (4. + ∗ a− ˆ a − u euzˆ euz + ∗ a− ˆ = (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.106) Applying (4. 1 + + uz ∗ a− ). the solution of (4.102) Similarly.108) . euzˆ ˆ euz ∗ a− ˆ (4. + (4.103) (4. We assume in the following a ˆ derivation.99) ˆ C(u) = zˆ a zˆ+ + z ∗ a− a ˆ + +z a ˆ ∗ − a euzˆ euz ˆ − u C(u) .107) − α∗ a− ˆ φ0 (y) = e− αα∗ 2 a eαˆ φ0 (y) . a a ν! (4.90) and zz ∗ = 1 we can write this d du e uzˆ+ a e uz ∗ a− ˆ a − z ∗ a− .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) . one obtains a evˆ + − v ∗ a− ˆ a φ0 (y) = e− 2 vv evˆ e− v 1 ∗ + ∗ a− ˆ φ0 (y) (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) .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) .96) is applied is a a a euˆ evˆ φ0 (y) = ev (ˆ − + + + u) a φ0 (y) = euv evˆ φ0 (y) . without explicitly stating this.105) a eαˆ + − α∗ a− ˆ φ0 (y) = e αα∗ 2 e−α ∗ a− ˆ a eαˆ φ0 (y) .104) (4. + (4.97) An example in which (4.

is the equivalent of the generating function (??).h.5: Working with Creation and Annihilation Operators Generating Function in Terms of a+ ˆ 87 We want to demonstrate now that generating functions are as useful a tool in the calculus of the ladder operators as they are in the calculus of diﬀerential operators.84.95) one obtains a a a a euˆ φ0 |evˆ φ0 = φ0 |euv evˆ φ0 = euv evˆ φ0 |φ0 = euv − + + + − (4.114) where we have employed (4.. + (4.112) Expanding both sides of this equation and using (4. i.115) a where the latter step follows after expansion of evˆ and using a− φ0 (y) = 0.h.s. provide the same values for Hn (0) as provided in Eq. the expected from which follows by comparision of each term on both sides φµ |φν orthonormality property.110) √ y2 u2 + 2uy − 2 2 a = euˆ φ0 (y) . For this purpose we consider a a a a euˆ φ0 |evˆ φ0 = φ0 |euˆ evˆ φ0 + + − + (4. in the ladder operator calculus. we can reproduce by means of the generating function (4.116) = δµ ν . 4.h. (??). and ˆ l. Using (4.111) yields π − 4 exp − 1 u2 2 a = euˆ φ0 (0) . 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).98) and again (4.s.115) and using (4. We want to derive now the values Hn (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.s. Setting y = 0 in (4.111) This expression.95).s.84) yields ∞ ν. Similarly. and on the r. Expanding r. We start from the generating function (??) and replace according to (4.e. of (4.h. .58) the Hermite polynomials Hn (y) by the eigenstates φn (y) √ ∞ 1 ( 2t)n y2 /2 2yt − t2 √ e φn (y) (4. (4.µ=0 uµ v ν √ φµ |φν µ!ν! ∞ = µ=0 uµ v µ µ! (4.113) Comparision of all terms on the l.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! + (4.111) the orthonormality properties of the wave functions φn (y).4.109) e = π4 n! n=0 √ Using (4.

dp (4. For this purpose we employ the Schr¨dinger equation in the momentum o representation. i. the eigenvalues are identical to those determined in the position representation. ˜ (4.88 Linear Harmonic Oscillator 4.2).122). according to (4.121).122) For a solution of the Schr¨dinger equation in the momentum representation. .2. 1. The momentum representation eigenfunctions are. . i. (4.118) d2 1 2 + p dp2 2m ˜ ˜ φE (p) = E φE (p) . are ˜ En = or.125) .123) (4.121) ω (n + ˜ 1 ) 2 .7) is given by (4..e. 2. using (4. n = 0. (4. .235). (−i)n ˜ φn (p) = √ 2n n! m˜ ω π 1 4 exp − m˜ p2 ω 2 Hn ( m˜ ω p) . is given by the ˜ function ψ(p).7) is 1 2 m 2 ω 2 d2 p − . In the position representation the wave function is a function of x.6) and the Hamiltonian (4. .e. As to be expected. for an oscillator in a stationary state φn (x) as given by (2.37). 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. En = ω (n + 1 ) 2 d2 1 + m˜ 2 p2 ω 2 2m dp 2 2 ˜ ˜˜ φE (p) = E φE (p) (4..120) Multiplying this equation by 1/m2 ω 2 yields − where ˜ E = E / m2 ω 2 . i. n = 0. 2. the momentum and position operators are p = p.3). the time-independent Schr¨dinger equation for the oscillator can be written o ˆ H = − mω 2 2 2 x = i ˆ d .235). The eigenvalues. . For example. ˆ and the Hamiltonian (4. 1. (4.117) In this section we want to provide a representation for the harmonic oscillator which is most natural ˜ if one wishes to determine for a stationary state of the system the probability density P (p) of ﬁnding the system at momentum p.119) 2m 2 dp2 Accordingly. 4. is given by a function ψ(x).. ω = 1 / m2 ω . The solutions can be stated readily exploiting the earlier results.124) ). .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.e. of (4. the momentum and position operators are as stated in (4. using (2. the probability density is P (x) = |φn (x)|2 . (4. (4. In the momentum representation the wave function is a function of p.

m ω (4.127) exp − p2 m ω 2 Hn ( 1 p) .128) We may also express the eigenstates (4.122) (−i)n ˜ φn (p) = √ 2n n! 1 πm ω 1 4 89 exp − p2 2m ω Hn ( 1 p) .1: Demonstrate the validity of (4.134).129) m ω one obtains.. i. ˜ We can now state the probability density Pn (p) for an oscillator with energy En to assume momentum p.126) in terms of dimensionless units following the procedure adopted for the position representation where we employed the substitution (4.130) dx exp(−ipx/ ) φn (x) −∞ (4. 4. . the correctness of the phase factors (−i)n . instead of (4.235)] absorbs the replacement dx → dy.133) where we note that a change of the normalization of φn (y) [c.58).. this allowed us to introduce in (4. (4.4. hence.e. deﬁned only up to an arbitrary phase factor eiα . m ω (4. Deﬁning k = m˜ / p or ω 1 k = p (4.125.30) and (2. 1 ˜ φn (k) = √ 2π +∞ (4.134) Exercise 4.129) follows ky = px/ and.6. 2n n! π (4. of course. φn (k) = We want to ﬁnally apply the relationship 1 ˜ φn (p) = √ 2π +∞ 2 (−i)n −k √ e 2 Hn (k) . 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. Proceed as follows.126). This implies the property of Hermite polynomials (−i)n e− k2 2 +∞ Hn (k) = −∞ dy exp(−iky) e− y2 2 Hn (y) (4. From (4.132) dy exp(−iky) φn (y) −∞ (4. i. α ∈ R.126) a phase factor (−i)n .f.131) in the present case employing dimensionless units. through direct integration.6: Momentum Representation or with (4.30) to obtain (4. the Jacobian.126) Normalized eigenfunctions are.e.30) and (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 . except that α changes periodically in time. +∞ Linear Harmonic Oscillator dy exp(−iky) g(z. For this purpose we assume such state exists and expand ∞ ∞ φ(α) (y) = n=0 d(α) φn (y) . z) = √ 2π where f (k.Grynberg (John Wiley & Sons. Inc.134). We want to construct and characterize such states.137) where y(t) describes the center of the particle and where the mass of the particle remains narrowly distributed around y(t) for an arbitrary period of time. The question arises if similar states exist also for the quantum oscillator.194. 4.139) where we have added the condition that the states be normalized. Such states are referred to as quasi-classical states. (4. z) = exp −2ikz + z 2 − k 2 /2 . ˆ α ∈ R. a pendulum swinging at small amplitudes. and G.4.C.7 Quasi-Classical States of the Harmonic Oscillator A classical harmonic oscillator. φ(α) | φ(α) = .135) g(k. We ﬁrst construct the solution of (4.Dupont-Roc. The answer is ‘yes’. 2.90 (a) Prove ﬁrst the property 1 f (k. namely through the eigenvalue problem for normalized states a− φ(α) (y) = α φ(α) (y) . 192 3 . pp.196) of Hermite polynomials and equate coeﬃcients of equal powers of z to prove (4.Chen-Tannoudji. Inserting this expansion into √ (4. The quasi-classical states can be obtained by generalizing the construction of the ground state of the oscillator. z) = exp 2yz − z 2 − y 2 /2 (4. coherent states. J. n n=0 d(α) n 2 = 1 (4. or Glauber states of the harmonic oscillator and they play. 1989) which discusses quasi-classical states of the free electromagnetic ﬁeld in Sect.136) (b) Employ the generating function (2.138) We will show that a harmonic oscillator prepared in such a state will remain forever in such a state. y) −∞ (4.140) An excellent textbook is Photons and Atoms: Introduction to Quantum Electrodynamics by C. The motion of the probability distribution of such state is presented in Fig. for example. for example. We will ﬁnd that α can be characterized as a displacement form the minimum of the oscillator and that the state φ(α) (y) displays the same spatial probability distribution as the ground state. carries out a periodic motion described by y(t) = Re y0 eiωt (4. New York.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..138).1 showing in its top row the attributes of such state just discussed.— III.

145) Using the generating function in the form (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) . (4.139)] 1 = n=0 d(α) n 2 = |c|2 n=0 α2 2 = |c|2 eα .147) allows one to interprete T (α) as an operator √ which shifts the ground state by 2α.149) with (4. n n+1 (4.146) √ 1 − (y − 2 α)2 2 (4. the quasi-classical states are φ (α) (y) = exp α2 − 2 ∞ n=0 αn √ φn (y) . i.147) √ which identiﬁes this state as a ground state of the oscillator displaced by 2α. One can also write (4. n! (4. hence. n! (4.144) Normalization requires c = exp(−α2 /2) and. .150) for α presently real.149) − α∗ a− ˆ (4.143) for a constant c which is determined through the normalization condition [see (4..e.138). n! (4.109).84) φ(α) (y) = exp − α2 2 ∞ n=0 αα∗ (αˆ+ ) a a+ φ0 (y) = e− 2 eαˆ φ0 (y) .148) The identity (4. However.7: Quasi-Classical States or ∞ n=0 91 √ n + 1 dn+1 − α d(α) n (α) φn (y) = 0 .4. This interpretation justiﬁes the choice of α ∈ R in (4.141) Because of the linear independence of the states φn (y) all coeﬃcients multiplying φn (y) must vanish and on obtains the recursion relationship (α) dn+1 = √ α d(α) . we will see below that α ∈ C are also admissible and will provide a corresponding interpretation.108) allows one to write φ(α) (y) = T + (α) φ0 (y) = T + (α) [π]− 4 e−y where a T + (α) = eαˆ + 1 2 /2 (4.142) The solution is αn d(α) = √ c n n! ∞ ∞ (4. n! n (4.145) using (4. Comparing (4.

s. .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 (4.151) which follows from a generalization of (4. In particular.152) Since the operator function on the r.h. 1.158) Comparision of this expression with (4. the shift operator leaves the ground state normalized.67.150) for complex α obeys the unitary property T + (α) T (α) = 1 1 (4. α ∈ C (4. t1 ) = u exp 1 2 α2 − 2 (α u)n √ φn (y) .e.145) is replaced by a complex number αu which at times t1 = t0 + 2πn/ω. 4.h. a a ˆ ˆ (4.s.92 Linear Harmonic Oscillator The shift operator as deﬁned in (4.109) serves again to simplify this sum √ y2 1 1 α2 (u2 + 1) φ(α) (y.157) (y.150) and one can conclude that T + (α) is a so-called unitary operator with the property T + (α) T (α) = T (α) T + (α) = 1 1 or applying this to a scalar product like (4. (4.156) (4. Following the procedure adopted in determining the propagator of the harmonic oscillator [see (4. it holds T + (α) φ0 |T + (α) φ0 = f |T (α) T + (α) g = φ0 |φ0 .160) .153) This is obviously the inverse of T + (α) as deﬁned in (4. n = 0. becomes real.152) T + (α) f |T + (α) g = f |T (α) T + (α) g = f |g . i. t1 ) = π − 4 u 2 exp − + 2yαu − 2 2 E (4.. (4. . n! u = e−iω(t1 − t0 ) . is the adjoint of the operator function on the l.145) shows that α 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! (4.154) .145) as the initial state φ(α) (y.155) (4.68)] one can write the time-dependent solution with (4. Relationship (4. one can write in case of T + (α) T + (α) + = T (α) = eα ∗ a− ˆ − αˆ+ a a = e − (αˆ + − α∗ a− ) ˆ .95) to linear combinations αˆ+ + βˆ− and reads then a a f (αˆ+ + βˆ− ) φ|ψ = φ|f (α∗ a− + β ∗ a+ ) ψ .

165) .147).138): a real α corresponds to a displacement such that initially the oscillator is at rest. (4. The diagram illustrates that the wave function retains its Gaussian shape with constant width for all times.163) also through the shift operator.164) where according to (4. Comparing (4. 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. The oscillations of the mean position y0 (t1 ) and of the mean momentum k0 (t1 ) are out of phase by π just as in the case of the classical oscillator.161) are time-dependent √ y0 (t1 ) = 2 α cosω(t1 − t0 ) √ k0 (t1 ) = − 2 α sinω(t1 − t0 ) 1 φ0 (t1 ) = − α2 sin2ω(t1 − t0 ) (4. being displaced around the center with period 2π/ω and.159) is then 1 1 i φ(α) (y. 4. One can express (4. moving solely its center of mass in an oscillator fashion around the minimum of the harmonic potential.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. t1 ) = π − 4 exp − [y − y0 (t1 )]2 + iy k0 (t1 ) − iφ0 (t1 ) − ω(t1 − t0 ) 2 2 .150) with (4. t1 ) = u 2 T + (α u) φ0 (y) (4.148. 4. in general. Our interpretation 2 explains also the meaning of a complex α in (4.163) This solution corresponds to the initial state given by (4. (4. In Fig. This shows clearly that the Glauber state is a close analogue to the classical oscillator. a complex α corresponds to a displaced oscillator which has initially a non-vanishing velocity. 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. this characterization corresponds closely to that of the possible initial states of a classical oscillator. a momentum factor and a phase factor.4.1 (top row) we present the probability distribution of the Glauber state for various instances in time. except that it is not pointlike. It is of interest to note that the displacement y0 of the Gaussian as well as the momentum k0 and phase φ0 associated with (4.138.158) allows one to write 1 φ(α) (y. experiences a change of its width (relative to . 4.162) 2 with a periodic change. The time-dependent wave function (4.161) and is found to describe a displaced Gaussian.

1: Time dependence of Gaussian wave packets in harmonic oscillator potential. t0 ) = [πσ 2 ]− 4 exp we expand ∞ 1 − (y − ao )2 2σ 2 (4. that of the Glauber states) with a period π/ω. 4.1 (second and third row).169) . 2σ 2 q = a .167) One can derive for the expansion coeﬃcients +∞ dn = −∞ dy φn (y) φ(ao . σ|y. σ|y. t0 ) +∞ = √ 1 n n!πσ 2 e−ao q/2 = √ 2n n!πσ dy exp −∞ +∞ −∞ − (y − ao )2 y2 − 2σ 2 2 2 Hn (y) (4.94 E E E E Linear Harmonic Oscillator E 2h ω 2h ω 2h ω 2h ω 2h ω π t= 0ω σ2 = 1 ¥ ¦ £ y ¨ ¡ ¢ ¡ π t = 1/4 ω σ2 = 1 ¥ ¦ £ ¡ y ¨ ¢ ¡ π t = 1/2 ω σ2 = 1 ¥ ¦ £ ¡ y ¨ ¢ ¡ π t = 3/4 ω σ2 = 1 ¥ ¦ £ ¡ y ¨ ¢ ¡ π t= 1 ω σ2 = 1 ¥ ¦ £ ¤ ¤ ¤ ¤ ¤ y ¡ ¢ ¡ ¨ -4 -2 0 2 4 -4 -2 0 2 4 -4 -2 0 2 4 -4 -2 0 2 4 -4 -2 0 2 4 E E E E E 2h ω π t = 0ω σ2 = 1/3 ¥ ¤ § £ ¢ 2h ω π t = 1/4 ω σ2 = 1/3 ¥ § £ 2h ω π t = 1/2 ω σ2 = 1/3 ¥ § £ 2h ω π t = 3/4 ω σ2 = 1/3 ¥ § £ 2h ω π t= 1 ω σ2 = 1/3 ¥ § £ ¤ ¤ ¤ y ¨ y ¨ ¢ y ¨ ¢ y ¨ ¢ ¤ y 0 ¢ -4 -2 0 2 4 -4 -2 0 2 4 -4 -2 0 2 4 -4 -2 0 2 4 -4 -2 2 4 © © © © E E E E E 2h ω π t = 0ω σ2 = 2 ¥ £ 2h ω π t = 1/4 ω σ2 = 2 ¥ £ 2h ω π t = 1/2 ω σ2 = 2 ¥ £ 2h ω π t = 3/4 ω σ2 = 2 ¥ £ © 2h ω π t= 1 ω σ2 = 2 ¥ ¤ ¤ ¤ ¤ y ¨ ¡ ¢ ¡ y ¨ ¢ ¡ y ¨ ¢ ¡ y ¨ ¢ ¡ ¤ y ¡ ¢ ¡ ¨ ¡ ¡ ¡ £ ¨ -4 -2 0 2 4 -4 -2 0 2 4 -4 -2 0 2 4 -4 -2 0 2 4 -4 -2 0 2 4 Figure 4. (4.168) dy e−p(y−q) Hn (y) where p = 1 + σ2 . t0 ) = n=0 dn φn (y) . This behaviour is illustrated in Fig. 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. σ|y. To describe such state which satisﬁes the initial condition φ(ao .166) φ(ao . 1 + σ2 (4.

For this purpose we note that according to the explicit expresssion (??) for Hn (y) the leading power of the Hermite polynomial is Hn (y) = (2 y)n + and. t1 ) = √ × where yo = q p a = √ .76) permits us to write this φ(ao .174) Hn (q p p−1 ) φn (y) . 2 by A. this 3 volume integral table is likely the most complete today.172) One can then conclude e−a /4 n lim dn = √ a σ→1 2n n! (4. a ground state shifted by ao = 2α. Prudnikov. Following the strategy adopted previously one can express the time-dependent state corresponding to the initial condition (4. t1 ) = e−aq/2 √ pσ ∞ n=0 e −iω(n+ 1 )(t1 −to ) √ 2 1 2n n! p−1 p n/2 (4. σ→1 p−1 2 (4..e.4. 1990). vol. 1 (4. 1 a2 2 1 + σ2 + 1 2 2 y o + 2 y yo This integral can be found in Integrals and Series. i.I. unfortunately very expensive.177) − 1 ω(t1 − to ) 2 . in fact. New York.170) In order to check the resulting coeﬃcients we consider the limit σ → 1. σ→1 O(y n−2 ) for n ≥ 2 1 for n = 0. Yu. Brychkov. t1 ) = √ 2 × exp − 1 (y 2 + yo ) 1+t2 − 2 1−t 2 1 1 √ √ pσ π 1 − t2 × t 1 − t2 (4. σ|y.143) of the Glauber state.175) ∞ tn n=0 2n n! Hn (yo ) e−yo /2 Hn (y) e−y 2 2 /2 The generating function (4. in agreement with the expected result (4.157) as follows φ(ao . and O. In this limit the coeﬃcients dn should√ become identical to the coeﬃcients (4. a worthy successor of the famous Gradshteyn.168) is4 e−aq/2 (p − 1)n/2 dn = √ Hn (q 2n n!σ p(n+1)/2 p ).P.171) lim p−1 p n 2 Hn (q p ) = lim (2q)n = an . σ|y. This expansion can be written φ(ao .7: Quasi-Classical States The solution of integral (4.166) in analogy to (4.173) which is.143). p−1 95 (4. σ|y. namely $750 4 . therefore. Marichev (Wiley.176) 1 √ pσ π a 2 exp − 1 1+σ2 + 1 yo − 1 ω(t1 − to ) 2 2 2 2 × (4. p−1 1 − σ4 t = 1 − σ 2 −iω(t1 −to ) e 1 + σ2 (4.

96 Linear Harmonic Oscillator .

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

(5. the matrix represents a proper rotation. in Cartesian coordinates r = (x1 . ϕ)T ) = ± sinϕ cosϕ 0 (5.1) In Cartesian coordinates this reads 3 xk = j=1 [R(ϑ)]kj xj k = 1. We assume the convention that rotations are right-handed. in case of a prefactor −1 a rotation and an inversion at the origin.2) Rotations conserve the scalar product between any pair of vectors a and b.e. then the ﬁngers of your ﬁst point in the direction of the rotation..e. (5. can be represented by 3 × 3 matrices R(ϑ).6) follows immediatety for the determinant of R(ϑ) using detA B = detA detB and detAT = detA detR(ϑ) = ±1 . (5. It holds then j=1 3 3 a ·b = j. .7) Let us consider brieﬂy an example to illustrate rotational transformations and to interpret the sign of detR(ϑ). 3 . In the latter case the determinant of R(ϑ = (0.3) Since this holds for any a and b. i. A rotation around the x3 -axis by an angle ϕ is described by the matrix cosϕ − sinϕ 0 R(ϑ = (0.k. (5. where ϑ denotes the rotation. 0.e. they conserve a · b = 3 aj bj .7) implies that a rotation is associated with an inversion. (5. In the following we want to exclude inversions and consider only rotation matrices with detR = 1. it follows 3 [R(ϑ)]jk [R(ϑ)]j = δk j=1 . i.(5. Let us deﬁne the rotated vector as r = R(ϑ) r . if the thumb in your right hand ﬁst points in the direction of ϑ. x3 )T .4) With the deﬁnition of the transposed matrix RT [RT ]jk ≡ Rkj this property can be written R(ϑ) RT (ϑ) = RT (ϑ) R(ϑ) = 1 1 . ϕ)T ) is negative. From (5.98 Properties of Rotations in R Theory of Angular Momentum and Spin Rotational transformations of vectors r ∈ R .8) 0 0 1 In case of a prefactor +1.5) This equation states the key characteristic of rotation matrices. 2. are linear and. (5.. x2 . =1 [R(ϑ)]jk [R(ϑ)]j ak b = j=1 aj bj . namely around an axis given by the direction of ϑ and by an angle |ϑ|. therefore. ϑ parametrizes the rotations uniquely as long as |ϑ| < 2π.6) (5. 0. i. a minus sign in Eq.

99 (5. In the following we will assume R1 .5. b ∈ G the product c = a ◦ b is also in G.11) (iii) From detR1 = 0 follows that R1 is non-singular and. (iii) ∀a. a ∈ G → ∃a−1 ∈ G with a ◦ a−1 = a−1 ◦ a = e.1: Matrix Representation of SO(3) Deﬁnition of the Group SO(3) We will consider then the following set SO(3) = { real 3 × 3 matrices R.9) This set of matrices is closed under matrix multiplication and. hence.13) . We need to demonstrate that this inverse belongs also to −1 T SO(3). T R1 R1 −1 = 1 −1 = 1 1 1 (5. (5. (5. (iv) For products of three elements holds the associative law a ◦ (b ◦ c) = (a ◦ b) ◦ c. We want to prove now that SO(3) as deﬁned in (5. there exists a real 3 × 3–matrix −1 R1 which is the inverse of R1 .12) (iv) Since the associative law holds for multiplication of any square matrices this property holds for the elements of SO(3). R2 ∈ SO(3) and do not write out “◦” explicitly since it represents the matrix product. For this purpose we must demonstrate that the group properties (i–iv) hold. R RT = RT R = 1 detR = 1 } 1. it holds 1. We have shown altogether that the elements of SO(3) form a group. one can show R3 R3 = 1 Furthermore. 3 × 3–matrix. 1 T Similarly. Since (R1 )T = (R1 )−1 it follows −1 −1 −1 T (R1 )T R1 = (R1 )−1 R1 = −1 which implies R1 ∈ SO(3).10) detR3 = det ( R1 R2 ) = detR1 detR2 = 1 . (ii) The group element ‘e’ is played by the 3 × 3 identity matrix 1 1. in fact. R3 = R1 R2 is a real.9) forms a group. It follows that R3 is also an element of SO(3). (ii) There exists an element e in G with the property ∀a ∈ G → e ◦ a = a ◦ e = a. forms a group G satisfying the axioms (i) For any pair a. (i) Obviously. We would like to point out the obvious property that for all elements R of SO(3) holds R−1 = RT (5. It holds T T T T R3 R3 = ( R1 R2 )T R1 R2 = R2 R1 R1 R2 = R2 R2 = 1 .

Lie Group. mappings between the sets which conserve the group properties: (a) the set of real and imaginary numbers {1. 0 −1 1 0 (5. i. establish if 1-1 homomorphic mappings exist.100 Theory of Angular Momentum and Spin Exercise 5. −i} together with multiplication as the binary operation ‘◦’. compact representation of R(ϑ) which will be of general use and. Generators.16) must be antisymmetric.1: Test if the following sets together with the stated binary operation ‘◦’ form groups. we arrive at the property exp ϑk LT k from which follows LT = −Lk . x2 –plane {rotation by Oo . Lie Algebra We want to consider now a most convenient. we want to assume that they exist. k = exp ( −ϑk Lk ) (5. (c) the set of four rotations in the x1 . will play a central role in the representation of many other symmetry transformations in Physics. M3 . We want to show that for the property R RT = 1 to hold. −1 0 0 −1 .18) (5. identify subgroups.e.17) . 0 1 −1 0 .20) (5. rotation by 90o . due to the uniqueness of the inverse. M2 .13) R(ϑk ek )−1 = R(ϑk ek )T = [exp ( ϑk Lk )]T = exp ϑk LT ˆ ˆ k and. in fact.19) (5. This representation expresses rotation matrices in terms of three 3 × 3– matrices J1 .1.15) Below we will construct appropriate matrices Jk . i. −1. M4 } = 1 0 0 1 . rotation by 180o . the matrices i Lk = − Jk (5. presently. (b) the set of matrices {M1 . rotation by 270o } and consecutive execution of rotation as the binary operation ‘◦’. To demonstrate this property consider rotations with ϑ = ϑk ek where ek ˆ ˆ is a unit vector in the direction of the Cartesian k–axis. J3 as follows i R(ϑ) = exp − ϑ · J (5.14) together with matrix multiplication as the binary operation ‘◦’. J2 . 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.

5. 0)T ) − 1 1 1 ϑ1 →0 (5. 0 0 0 0 0 0 (5.e. ϑ3 )T such that for ϑ = (0. (5. Using (5. 0)T the rotation is the identity. One can determine then the matrices Lk deﬁned in (5.27) . lim R[(0. ϑ3 and three matrices L1 . Using the deﬁnition of partial derivatives we can state L1 = lim ϑ−1 R(ϑ = (ϑ1 . L3 which are independent of the rotation angles.22). 0)T . L2 .23) (5. 0. We assume now that we parameterize rotations through a three-dimensional rotation vector ϑ = (ϑ1 .24) (5.22) and similar for L2 and L3 .1: Matrix Representation of SO(3) 101 We can conclude then that if elements of SO(3) can be expressed as R = eA . ϑ2 .2: Determine the generator L1 according to Eq.1. 0. the real matrix A is anti-symmetric. Exercise 5..8) we can state cosϑ3 −sinϑ3 0 1 −ϑ3 0 1 0 .25) 0 −ϑ3 0 0 −1 0 ϑ3 0 0 = 1 0 0 . (5. 0. 0. This property gives A then only three independent matrix elements.26) (5. i. A must have the form 0 a b A = −a 0 c . Let us use this deﬁnition to evaluate L3 .15) with three real parameters ϑ1 .(5. ϑ2 . ϑ3 )T ] = lim sinϑ3 cosϑ3 0 = ϑ3 ϑ3 →0 ϑ3 →0 0 0 1 0 0 1 Insertion into (5.21) −b −c 0 It is this the reason why one can expect that three-dimensional rotation matrices can be expressed through (5.22) 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).15) as the derivatives of R(ϑ) with respect to the Cartesian components ϑk taken at ϑ = (0.

30) are called Lie groups.. = 2.15) will be closed under matrix multiplication only in the case that commutators of Lk can be expressed in terms of linear combinations of generators.. holds the algebra [Jk . m = 3 for any even permutation of k = 1. m = 3 . 0)T ) = exp( 2 L2 ) and R(ϑ = (0.15) obey the group property. m = 3 for any odd permutation of k = 1. 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. 2 .15). as we see later. L ] = k m Lm (5. however. only the latter example and . [A. that the operators (5. In case of the group SO(3) the structure constants are particularly simple. the matrix (5. = 2.g. it must hold 3 [Lk . i. e. the elements of if any two indices are identical for k = 1.. ϕ)T . R(ϑ = (0. x2 -. (5.15) and their generators obey the property (5. etc. B]] + [[A.8) in case of ϑ = (0. Z2 = 2 [A.28) The construction of the generators through the limit taken in Eq. An answer can be found considering the Baker-Campbell-Hausdorf expansion ∞ exp( λA ) exp( λB ) = exp( n=1 λn Zn ) (5.e. that the generators represent inﬁnitesimal rotations with 1 >> |ϑ| around the x1 -.33) We want to demonstrate now that exp(ϑ · L) yields the known rotational transformations. however. in fact.e. 0)T ) = exp( 1 L1 ). B] ).30) Groups with the property that their elements can be expressed like (5. Lie groups can have any number of generators. Of course. (5. We want to consider. B]. and x3 -axes. 0. and the constants fk m are called the structure constants. B]. (5. From this result one can conclude that expressions of the type (5. commutators of commutators of A and B. i. = 2.g. the Lie algebra of SO(3) can be written [Lk .22) implies.29) where Z1 = A + B and where the remaining operators Zn are commutators of A and B. In fact.31) where k m are the elements which are 0 1 k m = 1 −1 of the totally antisymmetric 3-dimensional tensor. Z3 = 12 ( [A. 3 )T ) = exp( 3 L3 ). The expression for Zn are actually rather complicated.30) is called the Lie algebra. are related to angular momentum operators. three being a special case.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. if the resulting products of exponential operators can be expressed in the form (5. The question is then. 0. 1 1 e. Finite rotations can be obtained by applications of many inﬁnitesimal rotations of the type R(ϑ = ( 1 . 0.32) For the matrices Jk = i Lk which. L ] = m=1 fk m Lm . J ] = i k m Jm . (5. the property (5.

1: Matrix Representation of SO(3) determine exp(ϕL3 ). that the matrix ϕL3 .27). eϕL3 which agrees with (5.41) cos ϕ − sin ϕ sin ϕ cos ϕ (5.40) . (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 .3: Show that the generators Lk of SO(3) obey the commutation relationship (5. We note. accordingly.37) = (−1)n 1 0 0 1 . (5.36) The property 0 −1 1 0 allows one to write 0 −1 1 0 and. ∞ 2n = − 1 0 0 1 2n+1 (5. using (5.31). (5.38) eA = n=0 (−1)n 2n ϕ (2n)! 1 0 0 1 ∞ + n=0 (−1)n ϕ2n+1 (2n + 1)! 0 −1 1 0 .5.39) Recognizing the Taylor expansions of cos ϕ and sin ϕ one obtains eA = and. (2n + 1)! (5. cos ϕ − sin ϕ 0 = sin ϕ cos ϕ 0 0 0 1 (5.8).35). 5. Exercise 5. can be written ϕL3 = 0 103 A 0 0 0 0 . 0 −1 1 0 = (−1)n 0 −1 1 0 (5. according to (5.1. A = ϕ 0 −1 1 0 .34.35) 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 .

e.2.44) holds.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).104 Theory of Angular Momentum and Spin 5.e. Since this holds for any ψ(r ) one can conclude ρ R(ϑ1 ) ρ R(ϑ2 ) group property (5. complex-valued functions over the 3-dimensional space R which can be diﬀerentiated inﬁnitely often. Deﬁnition We ﬁrst deﬁne how a rotational transformation acts on a wavefunction ψ(r ). hence. a relationship which we like to express as follows R(ϑ) = ρ R(ϑ) .47) = ρ R(ϑ1 ) R(ϑ2 ) . one isomorphic to SO(3) and.43) ρ( ) conserves the group property of SO(3). Exercise 5. its elements can be represented through an exponential R(ϑ) = exp i − ϑ·J (5.e. the R(ϑ) ψ(r) = ψ(R(ϑ) r ). i.42) Obviously. In particular. the transformation R(ϑ): C∞ ( ) → C∞ ( ) is related to the transformation R(ϑ): R → R .45) (5. 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. B ∈ SO(3) holds ρ(A B) = ρ(A) ρ(B) . for A.1: Assume the deﬁnition ρ ρ (A B) = ρ (B) ρ (A).. in fact. we deﬁne rotations R(ϑ) as linear maps C∞ ( ) → C∞ ( ). (5. i. (5. i. In analogy to R(ϑ) being a linear map R → R . namely R(ϑ)ψ(r ) = ψ(R−1 (ϑ) r ) .46) (5. = ψ([ R(ϑ1 ) R(ϑ2 ) ]−1 r) = ρ R(ϑ2 ) ψ([R(ϑ1 )]−1 r) (5.48) . For this purpose we require stringent continuity properties of the wavefunction: the wavefunctions under consideration must be elements of the set C∞ ( ). (5.

J3 ] and [J3 .2: (a) Derive the generators Jk . −ϑ3 ) r ) − f (r) ) . it holds [Jk .55) and f (r) = f [J1 .53) (5. 3 1 ϑ3 0 x1 x1 + ϑ3 x2 R(0.49) (note the minus sign of −ϑ1 which originates from the inverse of the rotation in Eq. 0.24) one can expand ϑ3 →0 ϑ3 →0 lim ϑ−1 R(ϑ = (0.(5.53–5.50) lim ϑ−1 ( f ( R(0.56) holds for [J2 . (5.56) We like to verify this property for [J1 . − 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) .23.2.e. 0)T ) − 1 1 ϑ1 →0 105 (5.52) ( x2 ∂1 − x1 ∂2 ) f (r) .57) Exercise 5. The generators Jk can be determined by applying (5. show that (5. 0. J2 ] f (r) = − = − 2[ 2 [− i J . . One obtains using (5.. 0 0 1 x3 x3 (5. 0. but also the Lie algebra of the generators.54) (5.2: Function Space Representation of SO(3) where the generators can be determined in analogy to Eq. (5. in case of J3 . 2 by means of limits as suggested in Eq. −ϑ3 ) r ≈ −ϑ3 1 0 x2 = −ϑ3 x1 + x2 . (5.49). i − J3 f (r) = = Using (5.49) to a function f (r).55) Not only is the group property of SO(3) conserved by ρ( ). ϑ3 )T ) f (r) − f (r) 3 (5. k = 1. 0. i. 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 . J2 ].22) according to i − J1 = lim ϑ−1 R(ϑ = (−ϑ1 . J1 ].5. In fact.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. (5. (5. J ] = i k m Jm .51) Inserting this into (5. 5.42)) and similarly for J2 and J3 .

φ) = = 2 ( + 1) Y m (θ. the operator 2 2 2 J 2 = J1 + J2 + J3 (5. J3 ] (5. k = 1. 2. [J 2 . J3 ] J2 −i J1 J2 − i J2 J1 + i J2 J1 + i J1 J2 = 0 According to (5.61) simultaneous eigenstates of J 2 and of one of the Jk . φ) m (θ. However. C] = A[B. The commutation relationships imply that no simultaneous eigenstates of all three Jk exist.60) commutes with all three Jk .58) i = r ×p ˆ (5. J3 ] + [J1 . 5.59) which. Equation (5. Show that for J3 as deﬁned above holds exp( iα J3 ) ψ(φ) = ψ(φ + α).56) one obtains [J 2 . can be found. Jk ] = 0 . φ).106 Theory of Angular Momentum and Spin Exercise 5. φ) as well as the eﬀect of the so-called raising and lowering operators J± = J1 ± i J2 (5. 3. 3 .53–5. according to the correspondence principle between classical and quantum mechanics. (5. and x3 = r cosθ. usually chosen as J3 . identiﬁes J as the quantum mechanical angular momentum operator.2. Using [AB. φ).62) J1 [J1 . . We will show now that the properties of spherical harmonics J2Y J3 Y m (θ. J3 ] + [J2 . 2.3: Consider a wave function ψ(φ) which depends only on the azimutal angle φ of the spherical coordinate system (r. using the momentum operator p = ˆ J = r × i .63) (5.55) can be written as a vector product of r and i − J = −r × From this follows. x2 = r sinθ sinφ. i.61) We demonstrate this property for k = 3.3 Angular Momentum Operators The generators Jk can be readily recognized as the angular momentum operators of quantum mechanics.64) mY m (θ. k = 1.56) are the famous commutation relationships of the quantium mechanical angular momentum operators..e. the system related to the Cartesian coordinates through x1 = r sinθ cosφ. To show this we ﬁrst note that (5. θ. These eigenstates are the well-known spherical harmonics Y m (θ.65) .e. i. C]B and (5. φ) (5. (5. C] + [A. J3 ] J1 + J2 [J2 . J3 ] = = = 2 2 [J1 + J2 .

φ) m+1 (θ.66) (5. Continuing our proof through induction we assume now that property (i) holds for m. and since we have already shown | − 1 = (iβ −1 )−1 L− | we can state L3 | − 1 = −i( − 1)| − 1 . φ) (5.5. L1 − iL2 ] = L2 + iL1 = i (L1 − iL2 ) = iL− and L3 | = −i | we obtain L3 L− | = ( iL− + L− L3 )| = i(− + 1) L− | . φ) 107 = = 0 = = 0 ( + m + 1)( − m) Y ( + m + 1)( − m) Y m+1 (θ. L− ] = [L3 . φ) are a consequence of the Lie algebra (5. 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 (5.1 Let Lk . The deﬁnitions of the coeﬃcients α −1 and β −1 yield L+ L− | = L+ ( iβ −1 | − 1 = − α −1 β −1 | .72) = 0 = −i | (5.68) (5. L ] = k m Lm and let L± = L1 ± i L2 . For this purpose we prove the following theorem. An Important Theorem Theorem 1. Using [L3 . k = 1. φ) (θ. L1 − iL2 ] = −2iL3 and L+ | = 0 one obtains L+ L− | = −2iL3 | = −2 | . L+ raises the m-value of | − 1 from m = − 1 to m = . The arguments are very similar to the ones used above and we can be brief. We will show that this property holds then also for m − 1.56). L− ] + L− L+ )| .76) We now show that property (ii) also holds for m = − 1 proceeding in a similar way. L2 = L2 + L2 + L2 .69) J+ Y (θ. in fact. L− ] + L− L3 ) | . L− ] = [L1 + iL2 . From this follows that L− | is an eigenstate of L3 . We note L3 L− | = ( [L3 . α −1 β −1 = 2 . namely. Noting [L+ .73) (5.. i.71) To prove this theorem we ﬁrst show by induction that (i) holds. J+ Y J− Y J− Y m (θ. 3 be operators acting on a Hilbert space H which obey the algebra [Lk .74) (5. It holds L+ L− | = (L+ L− − L− L+ + L− L+ )| = ([L+ .75) (5. We consider L+ L− | m = ( [L+ . i.67) (5.e. L− ] + L− L+ )| m = ( −2iL3 + .3: Angular Momentum Operators on the spherical harmonics. φ) − m (θ.70) (5. (∗) (5.e. For this purpose we ﬁrst demonstrate that the property holds for m = − 1 by considering (i) for the state L− | ∼ | − 1 . 2.

53–5. Again the argumemts are similar to the ones used above. yield properties (5. We note L3 L− | m = ( [L3 . i. One can normalize the states | m such that αm = βm . 5.72. φ. in particular. We note that we can write L2 = 1 1 1 1 2 2 2 2 L+ L− + 2 L− L+ + L3 . Angular Momentum Operators in Spherical Coordinates We want to express now the generators Jk .69). We will have various opportunities to employ the Theorem just derived.70. 5.e. 5.73. (5. given in (5.84) . We can deduce from (∗) and (∗∗) α α α −1 β −1 −2 β −2 −3 β −3 = 2 = 2( − 1 + ) = 2( − 2 + · · −1 + ) Obviously.81) i. it holds αm βm = 2 one obtains αm βm = = ( + 1) k=m+1 k . Before we can ﬁnally show that property (iii) holds as well for all − ≤ m ≤ we need to determine the coeﬃcients αm and βm .79) We can now show that (iii) holds for all proper m–values. 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 .79) read L+ | m L− | m + 1 = −i = −i ( + m + 1)( − m) | m + 1 ( + m + 1)( − m) | m . (∗∗) (5.66. in terms of spherical coordinates deﬁned through x1 x2 x3 = = = r sin θ cos φ r sin θ sin φ r cos θ (5.e. it also holds for m − 1. For this purpose we need to express the angular momentum operators in terms of spherical coordinates (r. This completes the proof of the theorem.108 Theory of Angular Momentum and Spin L− L+ )| m = (−2m − αm βm )| m .63. ﬁnally αm = βm = (5. 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 (5.83) (5. 5. (5. θ...77) We will show now that if (ii) holds for m.55).80) (5. This implies that (i) holds for m − 1. 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 . A ﬁrst application will involve the construction of the eigenstates of the angular momentum operators as deﬁned in (5. 5.64).82) (5. 2 We note that Eqs.78) ( + m + 1)( − m) ( + m + 1)( − m) .

using ∂/∂tanφ = cos2 φ ∂/∂φ and ∂/∂cosθ = −(1/sinθ) ∂/∂θ.87) To derive these properties we consider.86).82–5. in terms of spherical coordinates. L1 f (θ. Using (5. the latter deﬁned in (5.86) (5.87) holds J3 = To determine J 2 we note. (5.84 ) one obtains. φ) = (5. using (5. L1 f (θ.88) (5. ∂φ = sin φ (5.60).85). φ) (5.5. actually.91) which agrees with expression (5. applying repeatedly the chain rule.85) (5. cos θ ∂ cos2 φ + sin θ cos φ ∂φ cos2 θ sin θ sin φ ∂ sin θ sin φ ∂ + + sin2 θ f (θ. φ) ∂x2 /x1 ∂ ∂x3 /r ∂ = x3 + x3 ∂x2 ∂ tan φ ∂x2 ∂ cos θ ∂x2 /x1 /r ∂ ∂ − x2 − x2 ∂x3 3 ∂ cos θ f (θ.89) = ∂ x2 x2 ∂ x2 + x2 ∂ x3 − 33 − x2 1 3 2 x1 ∂ tan φ r ∂ cos θ r ∂ cos θ f (θ. φ). using (5.92) (5. φ) ∂x ∂x3 ∂ tan φ =0 (5. φ) sin θ ∂θ sin θ ∂θ L1 f (θ. for example. φ) = (x3 ∂2 − x2 ∂3 )f (θ.93) and similarly. We like to express now the operators J3 and J 2 .90) This yields.85). 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 ∂ = − . According to (5.3: Angular Momentum Operators 109 We ﬁrst like to demonstrate that the generators Jk . involve only the angular variables θ and φ and not the radius r. (L2 )2 = − cos φ ∂ ∂ + cot θ sin φ ∂θ ∂φ 2 . φ) = ( x3 ∂2 − x2 ∂3 ) f (θ. (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 . In fact.

95) sinθ ∂ ∂θ = ∂2 ∂ + cotθ . 5.101) 1 ∂2 r r ∂r2 and comparision with (5. 5. 5.94) It follows (L1 )2 + (L2 )2 + (L3 )2 = With cot2 θ + 1 = 1/sin2 θ.99) = = ∂2 2 ∂ 1 + + 2 2 2 ∂r r ∂r r sin θ ∂2 2 ∂ + 2 ∂r r ∂r sinθ ∂ ∂θ 2 + ∂2 ∂φ2 (5. 1 sin2 θ sinθ ∂ ∂θ (5. The functions in C∞ (S ) have real variables .4 Angular Momentum Eigenstates According to the theorem on page 107 above the eigenfunctions (5. 2 ∂θ ∂θ (5.87). sphere S2 .98) where Jclass = r × mv is the angular momentum and pr = mr is the radial momentum.63.87).97).100) (5. i. (5.85. C∞ (S ). 5. 2m r ∂r2 2m r2 2 (5.e. ∂θ ∂φ ∂2 ∂ ∂2 + cotθ + ( cot2 θ + 1 ) 2 . 5.110 = cos2 φ Theory of Angular Momentum and Spin ∂2 1 ∂ ∂2 + sinφ cosφ − 2 cotθ sinφ cosφ ∂θ2 ∂φ ∂θ∂φ sin2 θ 2 ∂ ∂ + cotθ sin2 φ + cot2 θ sin2 φ 2 . These generators act on a subspace of C∞ ( ). 5.64) can be constructed as follows. 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. ∂θ2 ∂θ ∂φ (5.86.86. The operators in the present case are Lk = − i Jk where the Jk are diﬀerential operators in θ and φ given in (5.96) and the relationship between Lk and Jk as given in (5. one can conclude J2 = − 2 sin θ 2 sinθ ∂ ∂θ 2 + ∂2 ∂φ2 .85. 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 + . namely on the space of complex-valued functions on the unit 3-dim.

64) can be constructed then by seeking ﬁrst functions Y (θ. φ) L3 Y (θ. (5. − 2. φ) = −i Y (θ. . m) Constructing Y → Y →→ ··· Y ( + m)! (2 )! ( − m)! −1 − We like to characterize the resulting eigenfunctions Y m (θ. φ) = −i ( + m + 1)( − m)Y m (θ.63. . 0 ≤ φ < 2π. and to be admissable for description of quantum states. m) = 1 −m J− 1 2 Y (θ.109) using (5. } applying L− Y m+1 (θ.105) ∆( . φ) (5. φ)f (θ. φ) = 0 as well as L3 Y (θ. . One obtains in this way Y m (θ. 5.87). The norm in H is deﬁned through |f |2 = S2 dΩ f ∗ (θ.105) explicitly. can be expressed L± = Accordingly. by carrying out the construction according to (5. (5. The raising and lowering operators L± = L1 ± i L2 . and then determining the family {Y m (θ. . the so-called spherical harmonics.102) exists.107) (5.111) . 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. .104) (5. φ) in H which satisfy L+ Y (θ. (5.106) The latter property is obviously satisﬁed for the chosen φ-dependence.5. φ). m = − .102) where dΩ = sinθ dθ dφ is the volume element of S2 . must be cyclic in φ with period 2π. φ) = = 0 −i Y (θ.103) iteratively for m = − 1.104. φ. is indeed a Hilbert space. 0 ≤ θ < π. .4: Angular Momentum Eigenstates 111 θ. φ) (5. (5.108) m (θ. φ). The eigenfunctions (5. The function space endowed with this norm and including only functions for which the integral (5. φ) . . − . . normalizing these functions. φ) = ∆( . − + 1.110) = 0 (5. (5. φ) = 2 + 1 ( − m)! P 4π ( + m)! m (cosθ) eimφ . φ) (5. 5.107) reads ∂ ∂ + i cotθ ∂θ ∂φ P (cosθ) ei φ i e±iφ ∂ ∂ ± i cotθ ∂θ ∂φ .85–5.

π +1 dθ sin2 0 +1 θ = −1 +1 −1 dx (1 − x2 ) +1 = = . We note that (5. 5.107).106.120) See. The normalization factor N is chosen such that π (5. = = x (1 − x2 ) 2 x3 3 + 2 −1 +1 −1 −1 dx x2 (1 − x2 ) + 2 · (2 − 2) 1·3 −1 +1 (1 − x2 ) dx x4 (1 − x2 ) −1 −2 +1 2 · (2 − 2) · · · 2 dx x2 1 · 3 · 5 · · · (2 − 1) −1 2 · (2 − 2) · · · 2 2 (2 )! 2 = 2 2 +1 1 · 3 · 5 · · · (2 − 1) 2 + 1 [1 · 3 · 5 · · · (2 − 1)] (5. One obtains. New York.119) provide the following expression for Y Y (θ. (5. 2nd Ed. .113) 0 is ﬁnite.e.114) dθ sinθ |Y (cosθ)|2 = 1 0 (5.112) where we employed the deﬁnition of P (cosθ) in (5.117) Accordingly.115) holds.114.112 or Theory of Angular Momentum and Spin ∂ − ∂θ cotθ P (cosθ) = 0 (5. 5.106.D.119) where the factor (−1) has been included to agree with convention1 . .116) The integral appearing in the last expression can be evaluated by repeated integration by parts. 5. (5. φ) = (−1) 1 1 = 1 [1 · 3 · 5 · · · (2 − 1)]2 (5. i.116) implies |N |2 from which we conclude N = (−1) 1 · 3 · 5 · · · (2 − 1) (5..114) yield |N |2 2 +1 1 2π 4π (2 )! π dθ sin2 0 +1 θ = 1. 1975) . for example.” by J. one for which π dθ sinθ |P (cosθ)|2 (5. Jackson (John Wiley. (5. (5. A proper solution of this equation.118) 2 +1 1 1 sin θ ei 4π (2 )! 2 ! φ . ”Classical Electrodynamics. using the new integration variable x = cos θ. is P (cos θ) = N sin θ as one can readily verify.

124) reads P (x) = = (2 )! (1 − x2 )− 2 (x2 − 1) 2 ! (−1) 1 · 3 · 5 · · · (2 − 1) sin θ (5. − deﬁned in (5. . We want to demonstrate now that the recursion equation (5.121) states then for the functions P m (cos θ) ∂ + (m + 1) cotθ ∂θ P m+1 (cos θ) = − ( + m + 1) ( − m) P m (cos θ) . For m = (5. therefore. .124) we proceed by induction. according to (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 . namely Y (θ.126) Then holds.110). ∂x −m−1 . (5. Let us then assume that (5.127) ∂ −m−1 (x2 − 1) . 5. and can obtain now.103. φ). (5. indeed.123) P m+1 (x) ∂ x + (m + 1) 1 ∂x (1 − x2 ) 2 . we seek to determine the functions P m (cos θ) and.e. use (5.4: Angular Momentum Eigenstates 113 We have constructed a normalized solution of (5.103. m = − 1. describes the φ-dependence of Y m (θ.119). as speciﬁed in (5.63.122) The latter identity can be written.110) together with (5. 5. To prove (5. The reader should note that the associated Legendre polynomials.64). P m+1 (x) = −m−1 m+1 ∂ 1 ( + m + 1)! (1 − x2 )− 2 (x2 − 1) . the eigenfunctions Y m (θ.125) which agrees with (5. are real.124) holds for m + 1. φ). P m (x) = −1 × ( + m + 1) ( − m) × −(1 − x2 ) 2 1 (5. m (cos θ) e (5. P m (x) = −1 ( + m)! 2 ! ( − m)! × (1 − x2 )− m+1 2 −(1 − x2 ) 2 1 ∂ x + (m + 1) 1 ∂x (1 − x2 ) 2 × (5.124) called the associated Legendre polynomials. 5. every application of L− reduces the power of eiφ by one.124). 2 ! ( − m − 1)! ∂x −m−1 (5. through repeated application of (5. − 2..5. employing again the variable x = cos θ.123) leads to the expression P m (x) = −m m ∂ 1 ( + m)! (1 − x2 )− 2 (x2 − 1) 2 ! ( − m)! ∂x −m (5. φ).123).106).121) This expressions shows that the factor eimφ . .114. Actually. 5. i.

(5. therefore. φ) = Y − (θ. We want to test if the recursion (5. using x = cos θ.106. φ) choosing the proper sign.129) holds.130) (5. In that case (1 − x2 ) is a polynomial of degree 2 and. if L− Y (θ.128) one can show that (5. therefore. it holds ∂2 (x2 − 1) = (2 )! ∂x2 and. It holds.132) which reduces the number of spherical harmonics which need to be evaluated independently roughly by half . φ) An alternative route to construct the eigenfunctions Y of L− f (θ. from such construction emerges an important symmetry property of Y m (θ.132).131) identiﬁes f (θ. φ) as given in (5. P − (x) = 1 ∂2 (1 − x2 ) 2 (x2 − 1) . 2 ! (2 )! ∂x2 1 (5. and constructs then the eigenfunctions Y − +1 . Constructing Y − → Y − +1 → ···Y m (θ. we embark on this construction in order to prove (5.124) holds for m if it holds for m + 1.133) Since the term with the highest power of (x2 − 1) is x2 . φ) = 0 (5.123) terminates for m = − . i.124) holds for m = . However.124). φ) using (5. yields a vanishing expression for m = − − 1 as long as is a non-negative integer. (5.124) and. Such construction reproduces the eigenfunctions Y m (θ. appears not very interesting. expression (5. that expression (5. φ) = (−1)m Y m −m (θ. Y ∗ (θ. P − (5. the derivative ∂ 2 +1 /∂x2 +1 of this expression vanishes. 5. Y − +2 . Since (5. therefore. hence. by repeated application of the operator L+ . which is equivalent to recursive application of L− . namely. φ) (5. 5. it holds then for all m.134) (x) = 1 2 ! (1 − x2 ) 2 .127) reproduces (5. φ) . In fact.e. φ) L3 f (θ. etc.124).. φ).124) and. therefore.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 (5.135) .106. We ﬁrst determine Y − (θ. φ) = = determines ﬁrst a normalized solution 0 i f (θ.

φ) = 2 +1 1 1 sin θ e−i 4π (2 )! 2 ! φ 115 − .5.110) applies the = ( + m + 1)( − m) Y m+1 (θ. i.124).142) yields P − (x) = (−1) 1 (1 − x2 )− 2 (x2 − 1) = (1 − x2 ) 2 2 ! 2 ! (5.106).124) one arrives at Y (θ.142) For this purpose we proceed by induction. Obviously.136) to construct all other Y recursion eiφ ∂ ∂ + i cotθ ∂θ ∂φ Y m (θ. a sign consistent with the family of functions Y m constructed above. the expression (5. This follows from the identity ∂ ∂ − i cotθ ∂θ ∂φ One can use (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 (5.80) and (5.136) We note in passing that this expression and the expression (5.139) = ∂ − m cotθ ∂θ P m (cos θ) .138) Employing (5.123)] P m+1 (x) = −(1 − x2 ) 2 1 ∂ x − m 1 ∂x (1 − x2 ) 2 P m (x) .120) for Y obey the postulated relationship (5.143) . φ) sin θ e−i φ = 0.137) m (θ..140) Introducing again the variable x = cos θ leads to the recursion equation [c. We ﬁrst note that for m = − (5. (5. (5. (5.136) is normalized and has the proper sign. (5.129). (5. (5. (5. φ). 5. following closely the proof of eq.e.106. 5.122. According to (5.f.4: Angular Momentum Eigenstates Due to (5.132). 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. φ) . One can readily verify that this expression provides a solution of (5.

135).s. equivalently. P (x) = P 0 (x) are called Legendre polynomials.106) this implies for Y The Legendre Polynomials The functions the identity (5. etc. i. of which agrees with the r. Y −2 .124).144) reads P m+1 (x) = +m+1 m+1 ∂ (−1)m+1 (1 − x2 ) 2 (x2 − 1) . are identical.132).142) one can state P (x) = ∂ (x2 − 1) .e.149) (5. 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 . 1 (1 − x ) ∂x +m (1 − x2 ) 2 1 (5. one can conclude P m (x) = (−1)m m (θ.h. According to both (5. φ) ( − m)! P ( + m)! −m (x) . The construction beginning with Y − and continuing with Y − +1 . yields the same eigenfunctions as the previous construction beginning with Y and stepping down the series of functions Y −1 .124) yields (−1)m ( − m)! P ( + m)! −m (x) = +m m ∂ (−1)m (1 − x2 ) 2 (x2 − 1) .. Hence.. Since (5. Y − +2 .128) m + 1 → −m or. etc. one notes that application of (5.144) ∂ x − m 1 ∂x (1 − x2 ) 2 +m m ∂ (−1)m (1 − x2 ) 2 (x2 − 1) .147) the r. 2 ! ∂x +m+1 (5. 2 ! ∂x 1 (5. Accordingly. However.116 Theory of Angular Momentum and Spin wich agrees with (5. 2 ! ∂x +m −(1 − x2 ) 2 Replacing in (5.146) i. (5.150) .142) holds for m + 1 if it holds for m.142).148) According to (5.141) reads then P m+1 (x) = 1 (5. also the associated Legendre polynomials determined this way. of (5.142) holds for m.e. 2 ! ∂x +m (5.145) Hence.h.142) holds for m = − we veriﬁed that it holds for all m.s. (5. (5.142) and by (5.124) and (5. (5. We then assume that (5. given by (5.

pp. (5. pp.158) see..155) P (x) t . (5. in case of x = 1 holds w(1. m ≥ 0 ∂xm (5. for example.154) can be written 1 = w(x. =0 (5. 1965) which is an excellent compendium on special functions employed in physics. 2nd Ed.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) . 92 To prove this property see. The spherical harmonics for m < 0 can be obtained using (5.N. 1. 1975). 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. . t) = √ Comparision with (5. and r2 denotes the location of the charge. 45 of “Special Functions and their Applications” by N. The generating function allows one to derive useful properties of the Legendre polynomials. . 2. e. Lebedev (Prentice-Hall..J.4: Angular Momentum Eigenstates The ﬁrst few polynomials are P0 (x) = 1 . (5. P3 (x) = 1 2 117 P2 (x) = − 3x) 1 2 2 (3x − 1) . t) = √ 1 − 2xt + t2 ∞ 1 − 2 x t + t2 .153) P (cos θ) eimφ . r1 is the point where the potential is measured. t = r2 /r1 .g.” by J. In case q = 1 and |r2 | < |r1 | holds the identity2 ∞ 1 r2 = P (cos γ) (5. N. Englewood Cliﬀs. Jackson (John Wiley. For example. φ) = 2 + 1 ( − m)! (−1)m sinm θ × 4π ( + m)! × ∂ ∂ cos θ m (5.150) and (5.151) (5x3 Comparision of (5. ”Classical Electrodynamics. . Using x = cos γ.132).156) w(x.152) and. t) is called a generating function of the Legendre polynomials3 .5.D. . and |r1 − r2 | = r1 (5. the spherical harmonics for m ≥ 0 Y m (θ.156) yields P (1) = 1 . New York. =0 (5.154) |r1 − r2 | r +1 =0 1 where γ is the angle between r1 and r2 . P1 (x) = x . accordingly. 2 3 1 1 = = 2 1−t 1 − 2t + t ∞ t .157) = 0.

t) x−t = . φ) (5. 2.. The spherical harmonics are eigenfunctions of the angular momentum operators J2Y J3 Y m (θ. . If the spherical coordinates of r are (r. π + φ). φ) m (π − θ.162) − (2 + 1)x P (x) + P −1 (x) ] t + =1 [( + 1) P +1 (x) The coeﬃcients for all powers t must vanish individually.163) (5. t) (1 − 2xt + t2 ) Employing (5. Accordingly. π + φ).166) We want to summarize the properties of the spherical harmonics Y derived above.159) Using ∂lnw/∂t = (1/w) ∂w/∂t and multiplying (5.118 For w(x. ∂t P (x)t −1 −1 (5. . Inspection of (5.159) by w(1 − 2xt + t2 ) leads to the diﬀerential equation obeyed by w(x.164) which. holds P1 (x) ( + 1) P +1 (x) = = x P0 (x) (2 + 1) x P (x) − P −1 (x) (5. . (5. 1. under inversion Y m (θ. Accordingly. π − θ. ∂t 1 − 2xt + t2 (5.156) and (∂/∂t) to P (x)t = ∞ ∂w + (t − x) w = 0 . t) holds Theory of Angular Momentum and Spin ∂lnw(x. Noting exp[im(π + φ)] = (−1)m exp(imφ) we determine Y Properties of Y m (θ. φ) = = 2 ( + 1) Y m (θ. φ) .151)] allows one to determine P (x) for = 1.142) allows one to conclude P m (−x) = (−1) +m P m (x) (5.167) (5. φ) Under inversion at the origin vectors r are replaced by − r. φ) . then the coordinates of − r are (r. φ) goes over to Y m (π − θ.168) mY m (θ.160) this diﬀerential equation is equivalent ∞ (1 − 2xt + t2 ) =0 P (x) t + (t − x) =0 P (x) t = 0 (5. θ.165) since ∂ n /∂(−x)n = (−1)n ∂ n /∂xn . (5. φ) m (θ.161) Collecting coeﬃcients with equal powers t yields 0 = P1 (x) − x P0 (x) ∞ (5. using P0 (x) = 1 [c.f. φ) m (θ. π + φ) = (−1) Y m (θ. . Due to cos(π − θ) = − cos θ the replacement r → −r alters P m (x) into P m (−x). Inversion Symmetry of Y m (θ. φ).

177) Note that the spherical harmonics are real. i. (5. φ)f (θ. φ) = 2 +1 P (cos θ) . φ) C = =0 m=− π C sin θdθ 0 m Y m (θ. for any f (θ.176) = 1 +1 [ (2 + 1) x P (x) − are the Legendre polynomials. 5. . except for the factor exp(imφ).4: Angular Momentum Eigenstates 119 2. .173) m+1 (θ.169) 3.e.) P0 (x) = 1 . a complete basis of C∞ (S2 ).174) m ≥ 0 P (cos θ) eimφ .170) (5.178) 7. φ) where J± = J1 ± iJ2 .172) (5. φ) .175) (5. φ) 2π 0 (5. m 4. 6. For the Legendre polynomials holds the orthogonality property +1 dx P (x) P ( (x) = −1 2 δ 2 +1 . P +1 (x) P1 (x) = x . φ) m (θ. φ) = (−1)m Y ∗ (θ. φ) = 2 + 1 ( − m)! (−1)m sinm θ × 4π ( + m)! 2 ! × ∂ ∂ cos θ m (5. 4π (5.5. φ) (5. The spherical harmonics are given by the formula Y m (θ. The spherical harmonics form an orthonormal basis of the space C∞ (S2 ) of normalizable. The spherical harmonics obey the recursion relationships J+ Y J− Y m (θ. where ( = 1. P −1 (x) ] (5. φ) = = ( + m + 1)( − m) Y ( + m + 1)( − m) Y m+1 (θ. m (5. (5.171) m = dφ Y ∗ (θ. The spherical harmonics for m < 0 are given by Y −m (θ. in fact. φ) ∈ C∞ (S2 ) holds ∞ f (θ. 2. φ) . . uniquely deﬁned functions over the unit sphere S2 which are inﬁnitely often diﬀerentiable π 2π sin θdθ 0 0 dφY ∗m (θ. φ) = δ δm m . φ) Y m (θ.. The spherical harmonics Y 0 are Y 0 (θ. The spherical harmonics form.179) .

The spherical harmonics obey the inversion symmetry Y 10. The spherical harmonics for m (θ.4.s.h.86). (5.187) h(r) Y m (θ. φ) = 1 ∂2 r − r ∂r2 ( + 1) r2 h(r) Y m (θ.4. (5. For this purpose start from the identity +1 ∞ +1 dx w(x. For the Laplacian holds 2 = = = (5. φ) = (−1) Y m (π − θ. 1.189) evaluate the integral on the l. 4π (5..s.184) Y22 Y21 Y20 together with (5.177). t)2 = −1 .87).4.188) Exercise 5.185) (5. Exercise 5.181) = 0. φ) . Exercise 5. φ) = δm0 2 +1 .1: Derive expressions (5. 11. φ).3: Construct all spherical harmonics Y3m (θ.2: Derive the orthogonality property for the Legendre polynomials (5.189).179) using the generating function w(x. 2 are given by Y00 = 1 √ 4π 3 sin θ eiφ 8π 3 cos θ 4π 1 4 − 15 sin2 θ e2iφ 2π 15 sin θ cos θ eiφ 8π 5 4π 3 1 cos2 θ − 2 2 (5. .183) (5. (5. =0 t + −1 dxP (x) P (x) . (5.182) Y11 Y10 = = − (5.180) 9. expand the result in powers of t and equate the resulting powers to those arising on the r.h.120 8. The spherical harmonics for θ = 0 are Y m (θ Theory of Angular Momentum and Spin = 0. π + φ) . t) stated in (5.156).186) (5. of (5.

1. on functions f (θ.. − + 1.e. = .195) i. if one speciﬁes the functional form of the coeﬃcients [D(ϑ)] m .190) provide a complete. φ) ˜ Y m (θ.194) Y m (θ. exp − i ϑ · J . φ) ˜ under such rotational transformation by f (θ. φ) (5.. . are characterized completely if one speciﬁes the transformation of any Y m (θ. = . φ ) exp m One can also represent C∞ (S ) C∞ (S ) = [ {Y . called the irreducible representation. assumes a particularly simple form. (5. (5. ∞. with elements given by (5.. ˜ f (θ.196) C∞ (S ) = = X . m = − .− + . = − . The transformations exp − i ϑ · J . i. ... 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 . therefore.. and (ii) they form the lowest dimensional sets X obeying C∞ (S ) = X which have this invariance property.. φ) D(ϑ) = . For this purpose we consider the subspaces of C∞ (S ) X = [ {Y . . . }] (5. . ∞. m = S2 i dΩ Y ∗m (θ.. Comparision with (5. }] .193) where [ { } ] denotes closure of a set by taking all possible linear combinations of the elements of the set. . φ) c = . .. φ) ∈ C∞ (S ). . .. . }. φ) exp − ϑ · J .e.42). . φ) m.− + . (5. orthonormal basis for the function c m Y m (θ.195). φ) i − ϑ·J f (θ. φ). φ) = exp Since the spherical harmonics Y space C∞ (S ) one can expand ˜ f (θ. i. φ ) (5. . = 0. ..192) = S2 dΩ Y ∗ (θ .m m m (θ. exp − i ϑ · J (X ) = X .193) shows ∞ = − .e.197) The subspaces X have the important property that (i) they are invariant under rotations exp − i ϑ · J . m Y m (θ. φ) .191) i − ϑ·J f (θ .m D(ϑ) m.5: Irreducible Representations 121 5.5. m . . (5.5 Irreducible Representations We will consider now the eﬀect of the rotational transformations introduced in (5.. . φ) (5. We denote the image of a function f (θ. We like to argue that the matrix representing the rotational transformations of the function space C∞ (S ). .

. Since any additive term n n n cn1 .198) 2 2 where ϑ± = ϑ1 ( 1 ϑ− J+ 2 + 1 2 ϑ+ J− ϑ2 . 3. and J3 1 1 ϑ · J = ϑ− J+ + ϑ+ J− + ϑ3 J3 (5.194) only spherical harmonics with or. (5.200) in (5... one can conclude that J+ leaves X invariant. .197) of C∞ (S ) the matrix representation of D(ϑ) assumes a block-diagonal form. Since the Y m are n3 eigenfunctions of J3 the action of J3 reproduces the basis functions of X . which follow readily from (5. in case m = − . equivalently. 5. . Application of (5. one needs to consider the action of J+ 1 J− 2 J3 3 on the subspaces X .195) [D(ϑ)] m.e. the transformation exp − i ϑ · J leaves X invariant as well. with blocks of dimension 1.56.. produces 0. J− ] = 2 J3 . (5. . . .122 Theory of Angular Momentum and Spin The invariance of the subspaces X under rotations follows from a Taylor expansion of the exponential operator exp − i ϑ · J .n2 .200) of the rotational transformations one can also conclude that the X are the smallest invariant subspaces of C∞ (S ).e. can be expanded in powers One employs then the commutation properties [J3 .200) leaves X invariant. J− . 1×1 3×3 .201) From expression (5. leaves X invariant.e. J− 2 also leaves n1 X invariant.n3 J+ 1 J− 2 J3 3 . i. One = contribute can conclude then that in the expansion (5..n3 n n n Accordingly. i. m = δ S2 dΩ Y ∗ (θ. The exponential operator exp − i ϑ · J + ϑ3 J3 )n .199) [J+ . 5.202) D(ϑ) = (2 + 1) × (2 + 1) . φ) . φ) exp m i − ϑ·J Y m (θ.n2 . (5. Similarly. n The action of J− on Y m produces Y m−1 or. J± ] = ± J± (5. one concludes for the matrix elements (5..65). i.. If one orders the basis according to the partitoning (5.199) allows one to express exp i − ϑ·J = n1 . One proceeds by expanding ﬁrst ϑ · J in terms of J+ .

(d) Show that the similarity transformation T (ϕ) deﬁned in the space of non-singular.6 Wigner Rotation Matrices We will now take an important ﬁrst step to determine the functional form of the matrix elements of D(ϑ). does not provide the simplest mathematical description of rotations. ∈ N} invariant. 5.5. (b) Prove that the elements of SO(2) can be completely characterized through a single parameter. this parametrization. A more suitable parametrization had been suggested by Euler: every rotation exp − i ϑ · J can be represented also uniquely by three consecutive rotations: (i) a ﬁrst rotation around the original x3 –axis by an angle α. . Determine the generator of ρ(ϕ) in analogy to (5.e. (c) Show that the map R(ϕ) x y = x y = cosϕ −sinϕ sinϕ cosϕ x y = R(ϕ) x y deﬁnes a representation of SO(2). (a) Show that SO(2) with the group operation ◦ deﬁned as matrix multiplication. Give the corresponding expression of ρ(ϕ). i.1: Representations of the group SO(2) SO(2) is the set of all 2×2 matrices R which are orthogonal. (iii) a third rotation around the new x3 –axis by an angle γ.5. 2. (ii) a second rotation around the new x2 –axis by an angle β. (e) In the space C∞ ( ) of inﬁnitely often diﬀerentiable. This step reconsiders the parametrization of rotations by the vector ϑ assumed sofar. and for which detR = 1. f (α + 2π) = f (α).e. 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). 3 certainly allow one to describe any rotation around the origin. and periodic functions f (α). for which holds RT R = R RT = 1 1. The representation is referred to as the irreducible representation. k = 1. though seemingly natural. The three components ϑk . Exercise 5. However.6: Wigner Rotation Matrices 123 This representation of D(ϑ) is the one wich has blocks of the lowest dimensions possible. (f) The transformation ρ(ϕ) as deﬁned in (e) leaves the following subspace of functions considered in (e) Xm = { f (α) = A e−imα . other representations are termed reducible representations. is a group. 5.49).22. A ∈ C. real 2 × 2– matrices O through O = T (ϕ)O = R(ϕ)OR−1 (ϕ) with R(ϕ) as in (c) is also a representation of SO(2). i.

γ)] m.s.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 (5.205) Y m (θ.210) The third rotation in (5. The Euler rotation replaces exp − i ϑ · J by e− i γJ3 − i βJ2 − i αJ3 e e .203) has the disadvantage that it employs rotations deﬁned with respect to three diﬀerent frames of reference. m [D(α. (i) and (ii).207) which replaces J2 by the inverse transformation (i). we can write e− i −1 = S eA S −1 . i (5. 5.209) The ﬁrst two rotations in (5.207) are equivalent.203). the transformation from the original frame to the rotated frame. followed by transformation (i). one can replace (5.e.h. can then be written e− i βJ2 − i αJ3 e = e− i αJ3 e− i βJ2 (5.203) For any ϑ ∈ R one can ﬁnd Euler angles α. γ ∈ R such that exp i − ϑ·J = e− i γJ3 − i βJ2 − i αJ3 e e (5. φ) = S2 dΩ Y ∗m (θ.194. the axis x3 is deﬁned in the coordinate frame which is related to the original frame by the consecutive rotations (i) and (ii).. Obviously. (5. and the r.h. of (5.206) The expression (5. φ) (5.203). We will demonstrate that (5. γ will be referred to as Euler angles. β. (5. φ) e− γJ3 − i βJ2 − i αJ3 e e . For any similarity transformation involving operators A and S holds eS A S Accordingly. the l. in analogy to (5.212) .211) Using (5. Accordingly.203) can be expressed. followed by J2 in the original frame.m m. i. β. m i Y m (θ.209). For this purpose we notice that J2 can be expressed through the similarity transformation J2 = e− i αJ3 J2 e i αJ3 (5.s.e.208) αJ3 βJ2 = e− αJ3 e− i βJ2 e i (5. φ) [D(α. in terms of rotations deﬁned with respect to the original frame.. i. is e− i γJ3 = e− i βJ2 e− i αJ3 e− i γJ3 e i αJ3 e i βJ2 (5.124 Theory of Angular Momentum and Spin The angles α. however.e. The axis x2 is deﬁned in the coordinate frame which is related to the original frame by rotation (i). β.. i.195) by ˜ Y m (θ. γ)] = . β.204) is satisﬁed. the transformation from the rotated frame to the original frame.

Examples are the other ﬁve members of the lepton family to which the electron belongs.215) The evaluation of the matrix elements (5. but many other elementary components of matter are endowed with spin-like properties. φ) .220) We will derive below [see Eqs. (5. φ) i i (5.7 Spin 1 2 and the group SU(2) The spin describes a basic and fascinating property of matter.64) yields dΩ f (θ. φ) [D(α. β.219). the muon µ and its neutrino νµ of the second . φ) e− m S2 i αJ3 f (θ. Best known is the spin of the electron. φ) Y m (θ. 5.203) with respect to the original (unprimed) frame one simply needs to reverse the order of the rotations. (5. γ)] m.5.215) beneﬁts from the choice of Euler angles for the parametrization of rotational transformations. φ) = e−im α S2 dΩ Y ∗ (θ. γ)] m.218) Deﬁning the so-called Wigner rotation matrix dmm (β) = one can express the rotation matrices [D(α. φ) e−imγ .. (5. φ) e− αJ3 e− βJ2 Y m (θ. one can write [D(α. φ) e− m i βJ2 Y m (θ.210) yields the simple result e− i γJ3 − i βJ2 − i αJ3 e e = e− i αJ3 e− i βJ2 e− i γJ3 . m (5. in addition.309. This allows one to express the rotational trasnformation (5.m m. φ) f (θ. β. φ) (5.216) Using.214) e− γJ3 = S2 dΩ Y ∗m (θ.213) i. m = e−iαm S2 dΩ Y ∗ (θ.217) Accordingly.206) by ˜ Y m (θ. β. The eigenvalue property (5.219) = e−iαm dm m (β) e−iγm δ ( ) .310)] an explicit expression for the Wigner rotation matrix (5. the electron e and the electron neutrino νe of the ﬁrst generation. φ) .7: Spin 1 2 and the group SU(2) 125 Multiplication from the left with (5. 5.205. γ)] = . γ)] m. 5. m [D(α. φ) e− m i βJ2 Y m (θ.e. β. to redeﬁne the three rotations in (5. (5. the self-adjointness of the operator J3 one can state dΩ Y ∗ (θ. (5. φ) = S2 dΩ f (θ. φ) e− S2 i γJ3 Y m (θ. m ( ) S2 dΩ Y ∗ (θ. m i Y m (θ. φ) e−iγm δ (5.

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

the simplest choice being S1 = 1 σ1 . z.7: Spin 1 2 and the group SU(2) 127 exist three such linear independent matrices. (σ · a) (σ · b) = a · b 1 + iσ · (a × b) 1 (5. hermitian 2 × 2–matrices A can be expandend. (5.e. σk ]+ = 2δjk 1 .229) and avoids commuting the components aj and bk . σj 2 (5. Any such A can be expressed in terms of three real parameters x.230) where a.231) . z A = In fact. they are associated with the important fermion property of matter.225) σ (5. In fact.228) (5. According to this property the Pauli matrices generate a 3dimensional Cliﬀord algebra C3 . (5. k = 1. We will show below that a 2-1–homomorphic mapping exists between SU(2) and SO(3) which establishes the close relationship between the two groups. it might hold aj bk − bk aj = 0. One can readily verify the commutation property σj σk − σk σj = [σj . e. it holds. but their comonents must not necessarily commute with each other.g.223) where σk . Cliﬀord algebras play an important role in the mathematical structure of physics. At this point we will state a useful property.227) which is essentially identical to the Lie algebra (5. The Pauli matrices obey the following anti-commutation properties σj σk + σk σj = [σj .e. 2 S2 = 1 σ2 .k=1 σ σ 3 j k j. 1 i. 3 are the Pauli matrices σ1 = 0 1 1 0 . σk ]− = 2i jk z x − iy x + iy −z . x. i.224) provide a basis in terms of which all traceless. b are vectors commuting with σ.k=1 j<k + σ j σ k aj bk .226) The Pauli matrices satisfy very special commutation and anti-commutation relationships. one obtains (σ · a) (σ · b) = 3 j 2 j=1 (σ ) aj bj 3 j k j. (5. 5.. 2.5.k=1 σ σ aj bk + j>k aj bk = 3 j. σ3 = 1 0 0 −1 . 1 σj σk = − σk σj for j = k which can also be readily veriﬁed.224) Algebraic Properties of the Pauli Matrices The Pauli matrices (5. namely.229) = 1 .227. (5. y. The proof of this relation rests on the commutation relationship and the anti-commutation relationship (5. A = x σ 1 + y σ 2 + z σ3 . y.31) of the group SO(3). 2 S3 = 1 σ3 2 (5. σ2 = 0 −i i 0 . ∈ R .

236) The matrix elements of this 2 × 2 operator can be labelled by the basis states χ+ and χ− . 1 1 1 ( + 1) χ± 2 2 . In the special case a = b ∈ R holds (σ · a)2 = a 2 1 . (5.237) + 1)1 and.234) 5. to rotations in space. 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.k=1 j>k i σl (aj bk − ak bj ) = i =1 σ a×b . This relationship emerges. hence. The two remaining terms yield using σ j σ k = −σ k σ j for j = k and altering ‘dummy’ summation indices 3 j. Well almost like it. We like to derive this result now by evaluating the transformations of SU(2) explicitly. let say around the x2 –axis by 360o .233) This result together with (5. i.231) proves (5. S3 χ± = ± 1 χ± 2 . (5. yields a · b.230). S ] = i k m Sm (5. For this purpose we choose to replace (5.235) which is identical to that of the generators of SO(3).222) by the Euler form exp(−iγS3 ) exp(−iβS2 ) exp(−iαS3 ) .222) as a rotation vector and you rotate once. spin changes sign. The commutation property (5.k=1 j>k 1 2 3 j.h.8 Generators and Rotation Matrices of SU(2) Sofar there is nothing which relates the transformations U .227) leads to 3 3 jk j. however.222).232) − σ k σ j ) (aj bk − ak bj ) .k=1 (σ σ j>k σ k σ j (aj bk − ak bj ) = (5. (5. (5.s. through the algebra obeyed by the operators Sk [Sk . however. 1 (5. we like to draw in this respect also on a close analogy to angular momentum states which developes 2 2 2 if one considers the operators S 2 = S1 + S2 + S3 and S3 . 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. since there is a slight diﬀerence: when you interprete the 3-component ϑ in (5.e. only a 720o rotation leaves the spin invariant.k=1 j>k σ j σ k (aj bk − ak bj ) − 1 2 3 j k j.238) S 2 χ± = .128 Theory of Angular Momentum and Spin Using (σ j )2 = 1 the ﬁrst term on the r.k=1 j>k σ j σ k (aj bk − ak bj ) = 1 2 3 j.

243) We note the property of this rotation matrix 2 dmm (2π) (1) = −1 1 .. 1 .239) where we made use of the notation (5. . following Jourdan and Schwinger. − + 1.e.5.5. i. γ)] m. that states | m for higher quantum numbers = 0. . .240) .242) The expressions in brackets [.236) 1 2m | exp(−iγS3 ) exp(−iβS2 ) exp(−iαS3 ) | 1 m 2 1 2m = e−iγm | exp(−iβS2 ) | 1 m e−iαm 2 (1) 2 = e−iγm dmm (β)e−iαm (5. 1.e. (5.241) 2n+1 Taylor expansion of the exponential operator yields then ∞ exp(−iβS2 ) = n=0 (−1)n β 2 2n ∞ 1 + 1 n=0 (−1)n β 2 0 1 −1 0 (5. The complete matrix elements (5. .244) i. (5. rotation by 360o changes the sign of the spin state. 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 .239 ) one expands the exponential operator exp(−iβS2 ). 2 2 We obtain with this notation for the transformations (5.9 Constructing Spin States with Larger Quantum Numbers Through Spinor Operators In this section we like to demonstrate. no two such particles can exist in the same state.220) are [D(α. One cannot consider the 1 entities carrying the spin– 2 to be particles in the ordinary sense for it can be shown that spin– 1 2 particles have fermion character. therefore. (5. β. . 2.245) 5. use the label 2 2 | 1 ± 1 for the states χ± . . (5. m = − . The idempotence property of Sk yields particularly simple expressions for these powers.e. We may.219). For this purpose one needs to determine the powers of −iβS2 .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 . identical particles any number of which can exist in the same state χ+ and χ− .239) in the notation (5. 3 . . ( − β S2 )2n = (−1)n ( − β S2 )2n+1 = (−1)n β 2 β 2 2n 1 1 0 1 −1 0 2n+1 (5.] can be identiﬁed with the Taylor expansion of the cos. namely. can be constructed formally from spin 2 2 states χ± if one considers the two properties χ± to be carried by two kinds of bosons.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). i. . .219. In order to determine the Wigner matrix in (5.

249) In the following will refer to b† = (b† . We consider.e. b† |Ψ0 = χ− − . . b ζ The States |Ψ(j. 3 . . b † ζ ζ =0 (5. + − For example. ζ = +. also operators of the type x b† = x+ b† + x− b† + − .248) that bζ relationships † † b ζ. b− )T is expressed through the commutation relationships + − (ζ. m = −j. b† )T and b = (b+ . . b± |Ψ0 = 0 (5. −j + 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± . b ζ = 0 b ζ. b† |Ψ0 = χ+ + .247) (5. m) † and bζ obey the commutation (5.248) bζ . 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. 2 2 .247. . . The boson character of the operators b† = (b† .254) We will show below that these states are orthonormal and form spin states with higher quantum numbers.252) (5. b ζ = b ζ. 1 . . j. x∗ b = x∗ b+ + x∗ b− + − (5. These operators are associated with a given spatial reference system. using (5.243) the creation operators in a coordinate system rotated by an angle β around the y–axis are b+ † = cos β β † b + sin b† 2 + 2 − . b− † = − sin β † β b + cos b† . b † ζ For the vacuum state |Ψ0 holds = δζζ .5. i.250) which for x∗ x = x∗ x+ + x∗ x− = 1 represent spinor operators in an arbitrary reference system.e. −) bζ .253) = δζζ . b† )T and b = (b+ . b ζ = b† . 2 + 2 − † (5. b− )T as spinor creation and annihilation + − operators. 1. .246) The corresponding adjoint operators are denoted by b+ and b− .251) One can show using this property and (5. m) = b† − j−m (j + m)! (j − m)! |Ψ0 . therefore. j = 0. (5. (5. i.

256) and (5. we need to evaluate b+ . b† + n b† + b† + b† + + n n n b+ . If it is true for n then using (5. f b† Obviously. that the annihilation operators play a role similar to the diﬀerential operator in calculus. 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. b† + n+1 = = b+ .248) we obtain b+ . (5. f (b† ) |Ψ0 . (5. We will derive the result.255) is b+ .254). b† + = n b† + n−1 . An example + − is the monomial b† which generates the states (5. possibly inﬁnite power series. f (b† ) |state + f (b† ) b+ |state .248) and from (5. For this purpose we consider the following polynomial of creation operators N j+m j−m f (b† ) = + n=0 cn b† + n .10 Algebraic Properties of Spinor Operators We want to establish ﬁrst a few important and useful algebraic properties which result from the commutation relationships (5.255) In order to determine how this operator is modiﬁed by multiplication from the left by b+ we note b+ f (b† ) |state = + b+ . (5.249).5.247.257) which for (5.258) Equations (5. b† + n . 5. b† + b+ .10: Algebraic Properties of Spinor Operators 131 5.259) The property is obviously true for n = 0. We will show that for this commutator holds n b+ . b† + − well-known for the quantum mechanical harmonic oscillator. + + (5. .260) .249) b+ f (b† ) |Ψ0 = + b+ . b† + n = n cn b+ .257) motivate us to determine the commutator (5.256) In the special case |state = |Ψ0 this reads using (5. in b† and b† . f (b† ) + (5. + b+ . b † + = n b† + n−1 b† + b † + + n = (n + 1) b† + .

f (x)] = ∂k f (x) .. Equation (5. the property holds also for n + 1. m) .262) Because of the commutation relationships (5. f (x)] g(x) = (∂k f (x))g(x) or [∂k .268) . b† ) in b† and b† holds (ζ = +. m) φjm (x) = j−m xj+m x− + (j + m)! (j − m)! .265) Generating Function of the States |Ψ(j. 2 .255) b+ . (5.e.267) exp xb† |Ψ0 is called a generating function of |Ψ(j.266) where xb† has been deﬁned in (5. m) We want to prove now the property ∞ j exp xb † |Ψ0 = 1 j=0. b† ) .1.264) This demonstrates the equivalence of the spinor operators bζ and the derivative operation. b† ) + − = ∂ ∂b† ζ f (b† . φjm (x) Ψ (j.261) Similarly..266) we compare the terms x+ b† + j+m φjm (x) Ψ (j.247) the more general property for polynomials f (b† . f (b† . = f (b† ) + (5. (5. m) = x− b† − j−m (j + m)! (j − m)! |Ψ0 .250) and where φjm (x) represents the function of the two variables x+ and x− closely related to |Ψ(j. In order to derive (5.263) According to (5. b† ) |Ψ0 = + − ∂ ∂b† ζ f (b† . + − (5. + − (5.. f (b† ) + where f (x) = df dx .132 Theory of Angular Momentum and Spin i. (5. One can ﬁnally conclude for polynomials (5. one can prove b− .263) corresponds to the product rule of calculus ∂k f (x)g(x) = (∂k f (x))g(x) + f (x)(∂k g(x)) which can be written ( ∂k f (x) − f (x)∂k ) g(x) = [∂k . f (b† ) − = f (b† ) − (5. b† ) |Ψ0 . m) m=−j (5. By induction we can conclude that (5. −) + − + − bζ .257) we can conclude in particular bζ f (b† .259) holds for all n ∈ N.

. .1.. 2.5. m=−j = j=0.271) 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. m) 1 j=0. 1 . (5.1. 1.275) = exp xb† |Ψ0 . one obtains φjm (x) Ψ (j..270) N 2j j → s=0 j−m=0 → m=−j ..10: Algebraic Properties of Spinor Operators with the s–th term in the binomial expansion of (a + b)n n! an−s bs .. 2 2 leads to ∞ j ∞ φjm (x) |Ψ (j. 2 . . 1. (5. 2 ∞ 1 xb† 2j! u 2j |Ψ0 = u=0.274) Summing this expression over 2j = 0. 1 . 1 xb† u! |Ψ0 (5. s! (n − s)! Deﬁning j+m = N −s j−m = s.1.273) Change of the summation indices allows one to conclude j φjm (x) Ψ (j....272) The summation over s can be written in terms of j and m using (5. 2. 3 .270) s x+ b† + N −s x− b† − s |Ψ0 . chosing the summation index j = 0. (2j)! 2j |Ψ0 (5. m) m=−j = = 1 x+ b† + x− b† + − (2j)! 2j 1 xb† |Ψ0 . i... ... .269) (5. . m) = = |Ψ0 (N − s)! s! N −s 1 N! x+ b† x− b† + − N ! (N − s)!s! N −s 133 (5.e.

m φjm (x∗ ) φj m (y) Ψ (j. m) |Ψ j . hence. m) and will be used for this purpose below.283) (5. The generating function allows one to derive various properties of the states |Ψ (j.277) To evaluate this expression we ﬁrst notice † e(xb ) † = e(x ∗ b) .284) .249) for any non-vanishing integer s holds + − Ψ0 |bs Ψ0 = 0 and. For this purpose we consider the inner product † † e(xb ) Ψ0 |e(yb ) Ψ0 = δjj δmm (5.280) The commutation properties [b+ .h. (5. one can write † † e(xb ) Ψ0 |e(yb ) Ψ0 † ∗ Ψ |e(x y) e(yb ) Ψ = 0 0 = e (x∗ y) † Ψ0 |e(yb ) Ψ0 (5.282) where we deﬁned x∗ y = x∗ y+ + x∗ y− . b† ] = 0 and [b− .277) by Ψ0 |e(x b) e(yb ) Ψ0 . y− )e(yb ) |Ψ0 (5. m) |Ψ j .s. Orthonormality of the States |Ψ (j.264) implies † † bζ e(yb ) |Ψ0 = yζ e(yb ) |Ψ0 . m holds.281) and. that Ψ (j.m.279) One can generalize this to † † s (bζ )s e(yb ) |Ψ0 = yζ e(yb ) |Ψ0 . In order to evaluate the ∗ b) (yb† ) operator e(x e we notice that the derivative property (5. According to (5.134 Theory of Angular Momentum and Spin which concludes our derivation. (5. (5.j .276) = j.278) † ∗ which allows us to replace the l. [b† . m . we can conclude † † ∗ e(xb ) Ψ0 |e(yb ) Ψ0 = e(x y) (5. therefore. b† ] = 0 allow one to state that for + − + any polynomial f (b+ . m) We want to show now that the states (5.254) are orthonormal. b− ] = 0. b− )e(yb ) |Ψ0 = f (y+ . of (5.e. b− ) holds † † f (b+ . (5. i.

k = 1. . 1.n .286) from which follows immediately the orthonormality property (5..m. using < + |σ1 | + > = < − |σ1 | − > = 0. One deﬁnes corresponding operators Jk through 1 Jk = b† < ζ |σk | ζ > bζ .m ijk Jk . J1 = J2 = J3 = 1 † b b− + b† b+ − 2 + 1 † b b− − b† b+ − 2i + 1 † b b+ − b† b− .m.277) φjm (x∗ ) φj m (y) Ψ (j.m. b† bm n n n.n . 3 .5.287) ζ 2 ζ. (5. The 2 operators are explicitly.m 135 = e(x ∗ y) . < + |σ1 | − > = < − |σ1 | + > = 1.224).m. 2. (5.275) one can show e(x ∗ y) = j. bm bm } n n = 1 4 < n |σi | m >< n |σj | m > {b† bm δmn − b† bm δm n } n n n. (5. − 2 + (5.285) Following steps similar to those in Eqs.j . States 2 We want to demonstrate now that the states |Ψ (j. 5. m) |Ψ j .m +b† b† . m) as given in (5.224).n . etc.ζ where σk . Jj ] = = 1 4 1 4 < n |σi | m >< n |σj | m > b† bm .223. 3 denote the Pauli spin matrices (5. (5.10: Algebraic Properties of Spinor Operators and comparing (5.276). . m j.289) < n |σi | m >< n |σj | m > {b† bm . Jj ] = i This property is derived as follows [Ji .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. 2 .m . In our present notation the matrix elements < ζ |σk | ζ > for ζ = ± correspond to the matrix elements < ζ |σk | ζ > for σ = ± 1 . (5.268–5. .282) and (5.288) The three operators obey the Lie algebra of SU(2) [Ji . 1 New Representation of Spin 0.m φjm (x∗ ) φjm (y) . b† bm n n n.

293) . 2 J 2 = J+ J− + J3 − J3 = J3 (J3 − 1) + J+ J− (5.297) .291) Here we have deﬁned J+ J− = J1 + iJ2 = J1 − iJ2 = b† b− + = b† b+ .296) ( b † + b + b † + b+ + b † + b + b † − b − − 2 b † + b + − b † − b − b † + b + + b † − b − b † − b − + 2 b † − b − + 4 b † + b − b† − b + ) (5. (5.m < n |2i n. J1 ] + i [J1 . − (5.292) The commutation relationships (5. J1 + iJ2 ] = −i [J2 .k ijk ijk σk | m > b † bm n = i = i 1 2 < n |σk | m > b† bm n n.m < n |[σi .m. (5.m ijk Jk . J+ ] = [J1 − iJ2 . To this end we note 2 2 2 J 2 = J1 + J2 + J3 = 1 1 2 J+ J− + J− J+ + J3 2 2 . (5.292).290) |Ψ (j.289) yield [J− . σj ]| m > b† bm n n. (5. m) are eigenstates of J 2 and J3 .m.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 .136 = 1 4 − = = 1 4 1 4 Theory of Angular Momentum and Spin < n |σi | m >< m |σj | m > b† bm n n.m 1 4 < n |σj | m >< m |σi | m > b† bm n n . (5.294) . m) as Eigenstates of J 2 and J3 We wish to show now that the states |Ψ (j.m. J2 ] = −2 J3 from which we can conclude the property 1 1 J− J+ = J+ J− − J3 2 2 Hence.295) The last result together with (5. (5.

. 2 2 137 (5. m − 1) . m . . . m + 1) (j + m) (j − m + 1)|Ψ (j.304) (5. −λ + 1. the number of states to the possible energy eigenvalues. Construct.303) Exercise 5. . . 1 2. m) J− |Ψ (j. Ik ] = i jk I . and eigenstates of k with eigenvalues j. I3 ||λ. (5. One can then conclude J 2 |Ψ (j. b+ ] = δjk .10. m = −λ. j = 1. Show λ = n1 +n2 n1 +n2 2 ( 2 + 1) and m = 1 2 (n1 − n2 ). n} eigenstates of I 2 = I1 + I2 + I3 and of I3 I 2 ||λ.300) (5. .302) (5.1: The system investigated in this section is formally identical to a 2-dimensional isotropic harmonic oscillator governed by the Hamiltonian H = ω(b+ b1 + b+ b2 + 1) . m) m |Ψ (j.s. 1 2 k (a) Show that the eigenstates are given by 1 |n1 . 2 .e. can be written b† + b − b † − b + = = One can then state J2 = 1 (b† + b+ )2 + b† + b+ b† − b− + b† − b− b† + b+ 4 + (b† − b− )2 + 2 b† + b+ + 2 b† − b− Deﬁning the operator 1 ˆ k = ( b† + b+ + b† − b− ) 2 one can write ﬁnally ˆ ˆ J2 = k ( k + 1 ) b† . m) One can furthermore derive readily J+ |Ψ (j. I3 = (b+ b1 + b+ b2 ) 2 2 2 2 2i 2 1 satisfy the Lie algebra of SU(2). [bj . m) . ˆ respectively.254) are eigenstates of + b+ and − b− with eigenvalues j + m and j − m.305) = = j ( j + 1 ) |Ψ (j. n2 .298) (5.10: Algebraic Properties of Spinor Operators The last term on the r. b† . m) = = (j + m + 1) (j − m) |Ψ (j. b † + b + + b † + b+ b † − b − 1 1 b† + b + + b † + b + b † − b− + b † − b − b † + b + . m) J3 |Ψ (j. m = λ||λ. (5. ..299) (5. . i. [Ij . λ. n1 = 0.301) Obviously.h. 2. n = n1 + n2 ﬁxed. .e. using operators I± = I1 ± iI2 2 2 2 and the subspace {|n1 . (b) Show that the three operators 1 + 1 1 + (b1 b2 + b+ b1 ) . I2 = (b1 b2 − b+ b1 ) . . i. the states (5. . 1. 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. m = m||λ.5. . m I1 = where λ = 0. .

m ) are related to the + − states |Ψ (j. σ values which yields (b† )j+m (b† )j−m is + − then σ = j + m − σ.308) The latter sum involves terms (b† )j+m (b† )j−m for σ + σ = j + m and 2j − σ − σ = j − m.254) are |Ψ j.310) it reduces to (5. On the other side the states |Ψ (j. for example.5.307) Comparision of (5.306.251). However. The prefactor of (b† )j+m (b† )j−m which according to (5. In case j = 1 this expression yields.243).306) and (5.306) where b † and b † are given by (5. m = b† + j+m b† − j−m (j + m )! (j − m )! |Ψ0 (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† . 1 1 √ sinβ 2 (1 + cosβ) 2 (1) 1 cosβ (dm m ) = − √2 sinβ 1 1 √ 2 (1 − cosβ) − 2 sinβ and in case j = 1 2 1 2 (1 − cosβ) 1 √ sinβ 2 1 (1 + cosβ) 2 (5. For this purpose we 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) .309) .307) can be + − (j) identiﬁed with the elements dm m (β) of the rotation matrix can then be written (j) dm m (β) = (j + m)!(j − m)! (j + m )!(j − m )! β 2 j−m × (−1)j−m −σ sin σ=0 2j−m−m −2σ j+m j+m−σ cos β 2 j−m σ m+m +2σ × (5. both conditions are satisﬁed for σ = j + m − σ. + − One may expect that these two conditions restrict both σ and σ. The combination of σ. (j) (5. For this purpose we note that the states |Ψ(jm) in a rotated coordinate system according m to (5. .138 Theory of Angular Momentum and Spin 5.11 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 = m=−j dm m (β) |Ψ (j. m) in the original coordinate system by j |Ψ j.

The SU(2) transformation entered in (5.e. . however. is the expected element of SO(3). namely that the component of the vector on the l.309) does not distinguish between SU(2) transformations (5. i. Hence. Evaluation of A yields √ −1 0 1 1 √ cosβ + 2 sinβ 1√ cosβ − 1 ˜ 0 A = 4 i √ i − 2 sinβ 2 cosβ 2 sinβ × √ 0 2 0 1 − cosβ − 2 sinβ 1 + cosβ √ −1 −i √ 0 −2 cosβ −2 2 sinβ 2 cosβ 2i 2i × × 0 0 2 = 1 4 √0 1 −i 0 −2 sinβ 2 2 cosβ 2 sinβ −1 −i √ 0 cosβ 0 −sinβ 1 0 (5. e. b+ † = −cos β † β b+ − sin b† 2 2 − . One can.h. that the the representation (5.251) and (5. 2 2 − (5.251).5.. = U x2 .311) (5.313) × 0 0 2 = 0 sinβ 0 cosβ 1 −i 0 which. i.309). Replacing the latter transformation by its negative form.310) establishes a mapping of SU(2) onto SO(3).312) transforms like an angular momentum state ˜ |1 m . therefore. This can be shown by (1) applying to the matrix A = (dm m ) in (5. It is of interest to trace the mapping from SU(2) onto SO(3) as represented by (5.s. in case of j = 1. b− † = sin β † β b+ − cos b† .g. of (5. conclude that the mapping from SU(2) onto SO(3) is a 2–1 mapping. x3 ) ∈ R .310) to the SU(2) transformation assumed in deriving the general result (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. the matrix A should represent a rotation around the x2 –axis in the space of vectors ˜ (x1 .314) leaves (5. x2 . the orthogonal 3 × 3 matrix which describes a rotation around the x2 –axis.314) in case of integer j–values.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 .12 Mapping of SU(2) onto SO(3) The representation (5.e. in fact.309) and the particular matrix (5. This factor implies.312) The choice of this transformation derives from a property shown further below..308) and had the form (5. U = 0 1 1 i √ ( x1 − ix2 ) √ x3 − √2 0 2 2 (5.

140 Theory of Angular Momentum and Spin .

γ are needed to specify a general rotational transformation 1 141 .. and particle C impinges on the compound AB coming from a large distance. We like to illustrate this for an example involving particle motion. three Eulerian angles α. and to express the Hamiltonian in terms of |rAB |. give rise to good quantum numbers. each carrying orbital angular momentum decribed by spherical harmonics Y m (θ. rB . The rotational part of the scattering motion is described then in terms of the unit vectors rAB and ˆ We remind the reader that.Chapter 6 Quantum Mechanical Addition of Angular Momenta and Spin In this section we consider composite systems made up of several particles. and (rAB .. Often the socalled total angular momentum. e. e. How should they be chosen? Actually there is no unique choice. |ρC |. 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 . Further below we will consider composite systems involving spin states. by a rotational vector ϑ. β. In this case a proper choice for a description of interactions would be to consider the vectors rAB = rA − rB and ρC = (mA rA + mB rB )/(mA + mB ) − rC . rC of the particles. six parameters must suﬃce to describe the interaction of the system. The overall orientation of any three particle conﬁguration can be speciﬁed by three parameters1 . B. is the quantity of interest. C governed by a Hamiltonian H which depends only on the internal coordinates of the system. sums of orbital angular momentum and of spin operators of the particles. on the distances between the three particles. classically speaking the sum of all angular momenta and spins of the composite system. We like to consider a choice which is physically most reasonable in a situation that the scattering proceeds such that particles A and B are bound. φ) as eigenfunctions and/or spin.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 ). ρC ). since related operators. Example: Three Particle Scattering Consider the scattering of three particles A. for example.g. This eleminates three further parameters from the dependency of the Hamiltonian on the three particle conﬁguration and one is left with three parameters. 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. commute with the Hamiltonian of the composite system and. hence.

. .max = 10 one obtains d(B) = 14 641. . (6.m2 Y 1 m1 (ˆAB ) Y 2 m2 (ˆC ). 2} 1. have not even be considered. (6. 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 ). mC . i.2) The dimension d(B) of B is 1. Such large number of dynamically coupled states would constitute a serious problem in any detailed description of the scattering process. = B such..max and 2. . and by the moment of inertia IA−B of the diatomic molecule A–B approximately as follows 1. 1. that only states within the subspaces Bk are coupled in the scattering process. The rotational symmetry of the interaction between the particles allows one. n = 1. 2 . mA mB + mB mC + mA mC 2. B1 ⊕ B2 ⊕ . 2. as we will demonstrate below.max + 1)2 ( 2. albeit puzzling. 14 641} . to separate the 14 641 dimensional space of rotational states Y 1 m1 (ˆAB ) Y 2 m2 (ˆC ) into r ρ subspaces Bk . . − 2 ≤ m1 ≤ (6. the values of which are determined by the size of the interaction domain ∆V . In fact.1) where 1.e.m1 .max d(B) = 1 =0 (2 1 + 1) 2 =0 (2 2 + 1) = ( 1.max denote the largest orbital and rotational angular momentum values. in fact. Obviously. .142 Addition of Angular Momenta and Spin ρC . a very large number. 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. in particular. composite systems behaving like elementary objects are common.max 2. . rather than by Classical Mechanics. will make their appearence a natural consequence of the symmetry of the building blocks of matter.m1 2 . by the total energy E.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. − 1 ≤ m1 ≤ 1.max + 1)2 (6. . ρ 1 = 0.3) For rather moderate values 1. 1. 2.m2 c (n) r ρ 1. the dimensions d(Bk ) of these subspaces does not exceed 100. = 0. One may consider then to describe the motion in terms ˆ of products of spherical harmonics Y 1 m1 (ˆAB ) Y 2 m2 (ˆC ) describing rotation of the compound AB r ρ and the orbital angular momentum of C around AB.max . and the following mathematical description will shed light on their ubiquitous appearence in physics.max = 2. each of which stands for two angles. . The latter often arrives at the physical properties of composite systems by adding the . . .max = 1 2 IA−B E . vibrations and rearrangement of the particles in reactions like AB + C → A + BC.max = ∆ 2 mA mB mC E . since further important degrees of freedom. . Such extremely useful transformation of the problem can be achieved through the choice of a new basis set B = { 1 . mB . . however. by the masses mA .max .

It holds then H R(ϑ) ψ = R(ϑ) H ψ.10) . Hence. 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). In this sense. (6. ρC ) = ψ(R−1 (ϑ) rAB .7) i R(ϑ) = exp − ϑ · J (6. One can equivalently express therefore (6.8) Obviously. For ˆ ˆ example. assuming presently that H might. but solely under simultaneous and identical rotations of rAB or ρC . in fact. To specify this property mathematically let us denote by H the Hamiltonian in the rotated frame. 3 (6. Jq(2) = 0 for p. q = 1. Since this is true for any ψ(rAB .53.6) For the following it is essential to note that H is not invariant under rotations of only rAB or ρC . 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. presumably a facility the reader would like to acquire with great eagerness. the existence of a basis (6.0: Addition of Angular Momenta and Spin 143 corresponding physical properties of the elementary components. Rotational Symmetry of the Hamiltonian As pointed out already. ρC ) deﬁned through R(ϑ) ψ(rAB . i. Following our description of rotations of single particle wave functions we express (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. The invariance of the Hamiltonian under overall rotations of the three particle system implies then H = R(ϑ) H R−1 (ϑ) .4) which decouples rotational states is connected with the rotational symmetry of the Hamiltonian of the three particle system considered.48) i i R(ϑ) = exp − ϑ · J (1) exp − ϑ · J (2) (6. rotations R(ϑ) of the wave functions ψ(rAB .. 2. connected with the fact that the Hamiltonian H does not depend on the overall orientation of the three interacting particles. R−1 (ϑ) ρC ) (6. be diﬀerent from H.e. according to (5. ∂yAB ∂zAB i (2) ∂ ∂ − J3 = ρy − ρx . ∂ρx ∂ρy (6.5) do not aﬀect the Hamiltonian. the reader will learn in the following section how to add and subtract in the microscopic world of Quantum Physics.55) holds i (1) − J1 = zAB ∂ ∂ − yAB .5) according to (5.6. the commutation relationships (1) Jp . ρC ) it follows H R(ϑ) = R(ϑ) H.7) where the generators J (k) are diﬀerential operators acting on rAB (k = 1) and on ρC (k = 2). 5.9) hold since the components of J (k) are diﬀerential operators with respect to diﬀerent variables.

3 cannot have simultaneous eigenstates among each other. a supposition which can be tested through the commutation properties of these operators.71–5. B + C] = [A. For a proof one uses (6.14) We will refer in the following to Jk .12) each individually can have simultaneous eigenstates with the Hamiltonian. Jk ] = 0 . 2. however. r ρ Deﬁnition of Total Angular Momentum States The commutation property (6.71–5. J ] = i k m Jm for n = 1. of course. Before we apply the procedure (5. ˆ Ω 2 = ρC .12) We consider this equation for inﬁnitesimal rotations. (6.9). the important fact that the Jk obey the Lie algebra of SO(3).81) we want to consider the relationship between YJM and Y 1 m1 (ˆAB ) Y 2 m2 (ˆC ). C]. the (n) (n) (n) properties [Jk .13) Hϑ · J − Since this holds for any ϑ it follows HJ − J H = 0 or. the operators Jk . To order O(|ϑ|) one obtains i ϑ · JH .81)]. 2. The property (6.f. that the components Jk .11) By means of (6. k = 1.14) implies that the components of the total angular momentum operator (6. According to the theorem above this property implies that one can construct eigenstates YJM of J3 and of J2 = J2 + J2 + J2 1 2 3 (6.15) are satisﬁed. i..e. ˆ (6. we will ﬁnd that the states yield the basis B with the desired property of a maximal uncoupling of rotational states.. [H. Eqs. k = 1. 3 as the three components of the total angular momentum operator. J ] = i k m Jm (6. In fact. and the eigenvalues of J2 and J3 are good quantum numbers. 2 together with the property [A.17) . i. (5. 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 ). 2.11) we can write the condition (6. 3 do not commute. We recognize.144 where J = J (1) + J (2) Addition of Angular Momenta and Spin . B] + [A. componentwise.14) implies that the total angular momentum is conserved during the scattering process. (6.e.6) for rotational invariance of the Hamiltonian in the form i i H = exp − ϑ · J H exp + ϑ · J . 3 .16) following the procedure stated in the theorem above [c. (6. We suspect. (6. for |ϑ| H ≈ 1 − 1 i ϑ·J H 1 + 1 i ϑ·J ≈ H + i 1. k = 1. 2.e. k = 1. One can show readily that the commutation relationships [Jk . i. In the following we will use the notation r ρ Ω1 = rAB . that energy.

22) We will not adopt such explicit summation since it leads to cumbersum notation.18) where the states YJM ( 1 .1: Show that J2 .21) hold.20) This equation can be satisﬁed only if the Clebsch-Gordan coeﬃcients satisfy (J M | 1 m1 2 m2 ) = 0 for m1 + m2 = M . The two missing quantum 2 2 numbers are 1 and 2 corresponding to the eigenvalues of J (1) and J (2) .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) .. 2 |Ω1 . Ω2 ) (1) (2) = 1 m1 2 m2 ) Y 1 m1 (Ω1 ) Y 2 m2 (Ω2 ) (m1 + m2 ) (J M | . (6.. i. the eigenvalue being speciﬁed by the quantum number M . 2 |Ω1 . the reader should always keep in mind that conditions equivalent to (6. and three quarks. (6. baryons. 2 M YJM ( 1 . We. hence.18) to avoid summation of vanishing terms YJM ( 1 . k = 1. J3 .m2 (J M | 1 m1 2 m2 ) Y 1 m1 (Ω1 ) Y 2 m2 (Ω2 ) (6. 2 |Ω1 . ρC ) are normalized. J (2) . These coeﬃcients. or the closely related Wigner 3j-coeﬃcients introduced further below. Ω2 ) = m1 . J3 YJM = M YJM .1.18) yields using the property (k) J3 Y k mk (Ωk ) = mk Y k mk (Ωk ) . mesons. However. Exercise 6. an important example being the symmetry groups SU(N) governing elementary particles made up of two quarks. m1 . play a cardinal role in the mathematical description of microscopic physical systems. J (2) . assume the expansion YJM ( 1 . and J3 .e. (6. Why can states YJM then not be speciﬁed by 5 quantum numbers? 2 2 Properties of Clebsch-Gordan Coeﬃcients A few important properties of Clebsch-Gordan coeﬃcients can be derived rather easily. The expansion coeﬃcients (J M | 1 m1 2 m2 ) ˆ are called the Clebsch-Gordan coeﬃcients which we seek to determine now. i. of (6. Equivalent coeﬃcients exist for other symmetry properties of multi–component systems.21) One can.18) is an eigenfunction of J3 . (6.1: Clebsch-Gordan Coeﬃcients 145 6.e. two further quantum numbers need to be speciﬁed for a complete characterization of the total angular momentum states.h. Ω2 ) = m1 (J M | 1 m1 2 M − m1 ) Y 1 m1 (Ω1 ) Y 2 m2 (Ω2 ) . restrict the sum in (6. . therefore.s. J (1) .e. i. Since YJM sofar speciﬁes solely two quantum numbers.m2 2 |Ω1 . J3 . and J (1) · J (2) commute. We ﬁrst notice that YJM in (6.6.19) Noting J3 = J3 + J3 and applying this to the l.

. . the basis B ( 1 .23) corresponding to the r. (6. This obviously corresponds quantum mechanically to a range of J–values J = | 1 − 2 |. J = | 1 − 2 |. − + 1.18) applied to Y∗ and to YJ JM (J M | m1 . 2 ) and B ( 1 . JM 2 |Ω1 .e. The basis B ( 1 . m1 = − 1 . . 2 ).24) The basis B( 1 . .27) together with (6. it is not immediately obvious what the J– values are. . = {YJM ( 1 .h. 2 ) has (2 1 + 1)(2 2 + 1) elements.24) yields 1 m1 2 m2 ) ∗ 1 m1 2 m2 ) (J M| = δJJ δM M . to a new basis B ( 1 . Ω2 ) = δJJ δM M (6. 2 ) is also orthonormal2 and must have the same number of elements.26) We will show below in an explicit construction of the Clebsch-Gordan coeﬃcients that.h. . Our derivation below will also yield real values for the Clebsch-Gordan coeﬃcients. the range of values assumed for J is correct.. In fact. in fact. 1 + 2 . m2 = − 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 .25) i. The property dΩ1 dΩ2 Y∗ ( 1 . −J + 1. 2) = {Y 1 m1 (Ω1 ) Y 2 2 m2 (Ω2 ). 2 |Ω1 . 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. . (6. . − 2} 1 + 1. property implies dΩ1 dΩ2 Y 1 m1 corresponding to the l.m2 2 M and with (6.s. In this case the range of |J|–values would be the interval [ |J (1) | − |J (2) | . . (6.1.28) This property follows from the fact that the basis elements are eigenstates of hermitian operators with diﬀerent eigenvalues.25) We want to state now two summation conditions which follow from the orthonormality of the two basis sets B( 1 . |J (1) | + |J (2) |]. .. Ω2 ). respectively. it holds 1+ 2 ( 2J + 1 ) = (2 J=| 1 − 2 | 1 + 1) (2 2 + 1) . . J} . B2 2) should be 2 |Ω1 . (6. J (1) and J (2) .2: Prove Eq. . However. . Ω2 ) YJ M ( 1.18) constitutes a change of an orthonormal basis B( 1 . .s. J.146 Theory of Angular Momentum and Spin The expansion (6. 2) . −J + 1. . . . . and that the states can be normalized. | 1 − 2 | + 1. 1. M = −J. . | 1 − 2| + 1. (6. Exercise 6. 1 + 2. (6. For each quantum number J there should be 2J + 1 elements YJM with M = −J.

28) from (6. the group SU(3) in case of meson multiplets involving two quarks. 2 |Ω1 . The result of this construction will include all the properties previewed above. or J > 1 + 2 (6. We can. (6. also apply the results. it is possible to expand Y 1 m1 (Ω1 )Y 2 m2 (Ω2 ) in terms of YJM ( 1 .29) where the expansion coeﬃcients are given by the respective scalar products in function space cJ M = dΩ1 dΩ2 Y∗ J M ( 1. e. i.31) are identical to Comparision with (6.. rather the convention (J M | has been assumed. 1+ 2 J Y 1 m1 (Ω1 )Y 2 m2 (Ω2 ) = J =| 1 − 2 | M =−J cJ M YJ M ( 1. 2 ) and B ( 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..2 Construction of Clebsch-Gordan Coeﬃcients We will now construct the Clebsch-Gordan coeﬃcients. The (J M | 1 m1 2 m2 ) cJ M . (6. .18).e. Ω2 ) ..28) allows one to conclude that the coeﬃcients cJ (J M | 1 m1 2 m2 )∗ . Ω2 ) .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 . 2 |Ω1 . to composite systems involving spin.81) stated above.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. (6. 2 |Ω1 . i.34) 6.6.e. Ω2 ). and actually also the properties of Clebsch-Gordan coeﬃcients stated above. Hence. 2 |Ω1 .e. i. (6.18) by Y∗ J orthogonality property (6. (6.1 states. One can show readily using the same reasoning as applied in the derivation of (6..m2 M ( 1. Ω2 ) and integrating. or baryons multiplets involving three quarks. 1 m1 2 m2 ) = 0 if J < | 1 − 2| .71–5.32) which complements (6. Ω2 )Y 1 m1 (Ω1 )Y 2 m2 (Ω2 ) .g.2: Construction of Clebsch-Gordan Coeﬃcients 147 The second summation condition starts from the fact that the basis sets B( 1 . A similar 2 construction will also be applied to composite systems governed by other symmetry groups. will be based solely on the commutation properties of the operators J and J (k) . 2 ) span the same function space. therefore.33) JM The latter summation has not been restricted explicitly to allowed J–values. 2 |Ω1 .27) yields δJJ δM M = m1 .

therefore. Ω2 ) = (J + M + 1)(J − M )YJM ( 1 .39) into (6. The construction assumes a particular choice of J ∈ {| such J–value seeks an expansion (6. J − 2.148 For the construction of YJM we will need the operators J± = J1 + iJ2 . . 2 |Ω1 . 2 |Ω1 . . (6.35) 1 − 2 |.11) J+ by J+ + J+ . 1+ 2 In fact. .41) 2) Y (Ω1 ) Y (Ω2 ) .66. (6.25) using (6. Ω2 ) (Ω1 ) dΩ2 Y 2 2 (Ω2 ) = 1. .37) and replace (1) (2) according to (6.43) . 1+ 2 ( 1.39) which we will try for a solution of (6. We obtain using (5.40) (Ω1 ) J+ Y (2) 2 2 J+ Y 1 1 (Ω1 ) Y (Ω2 ) + Y (Ω2 ) = 0 . . Ω2 ) .37). Ω2 ) = Y 1 1 (Ω1 ) Y 2 2 (Ω2 ) (6. Ω2 ) = Y 1 1 (Ω1 ) Y 2 2 (Ω2 ) . have shown Y 1+ 2. 2 |Ω1 .42) ( 1. 1+ 2 ( 1. −J. Ω2 ) . . Similarly. 2 |Ω1 . The corresponding classical total angular momentum vector Jclass (1) (2) would be obtained by aligning both Jclass and Jclass also along the z–axis and adding these two vectors.5. 2 |Ω1 . 2 |Ω1 . Ω2 ) Addition of Angular Momenta and Spin (6. M = −J. Ω2 ) 2 |Ω1 . 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.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. Ω2 ). (6. Quantum mechanically this corresponds to a state G 1+ 2. J} by applying repeatedly J− YJM +1 ( 1 . 1+ 2 ( 1.11) and (5.37) The solution needs to be normalized. We embark on the suggested construction for the choice J = 1 + 2 . .18) which satisﬁes J+ YJJ ( 1 . 1+ 2. we can demonstrate condition (6.36) = 0 = J YJJ ( 1 .38) for M = J − 1. For this purpose we insert (6. Ω2 ) is normalized dΩ2 G dΩ1 Y 1 1 1+ 2. J3 YJJ ( 1 . −J + 1. 2 |Ω1 . 2 |Ω1 . we can also demonstrate using (??) that G dΩ1 = We. Having determined such YJJ we then construct the whole family of functions XJ = {YJM ( 1 . 2 |Ω1 . . 1 + 2} and for (6.68) J+ = (1) + J+ (1) (2) Y 1 1 (Ω1 ) Y 2 2 2 2 (Ω2 ) 1 1 (6. | 1 − 2| + 1. . We ﬁrst seek an unnormalized solution GJJ and later normalize. (6. .

. We demonstrate the procedure ex(1) (1) plicitly only for M = 1 + 2 − 1. all the Clebsch-Gordan coeﬃcients ( 1 + 2 M | 1 m1 2 m2 ). Having constructed this family we have determined the Clebsch-Gordan coeﬃcients ( 1 + 2 − 1M | 1 m1 2 m2 ). 2 |Ω1 . Ω2 ) 2 1+ 2 = (Ω1 )Y 2 2 −1 (6. . The r. . with the symmetry pattern and with the terms Y 1 m1 Y 2 m2 contributing to each YJ M . According to the condition (6.2: Construction of Clebsch-Gordan Coeﬃcients 149 We now employ property (6. To demonstrate (6. The reader should familiarize himself with the entries of the Table. ( + )}. .38) to construct the family of functions B + = {Y + M ( . Ω2) for M = 1 + 2 − 2 and J = 1 + 2 . . We seek for this purpose ﬁrst an unnormalized solution G 1 + 2 −1 1 + 2 −1 of (6. of (6. Y 1 1 −1 (Ω1 )Y (Ω2 ) + Y 1 1 This construction can be continued to obtain all 2( 1 + 2 ) + 1 elements of B + and. One can obviously continue the construction outlined to determine Y 1 + 2 −2 1 + 2 −2 . 1 + 2 − 3. One obtains then Y 1 1+ 2 1 + 2 1 + 2 −1 2 2 ( 1.s. .44) (Ω2 ) .2. 1 + 2 − 2.46) 2 1 +c Y (Ω1 ) Y 2 2 (Ω2 ) = 0 . This expression is orthogonal to (6. . . The l.s. One can readily show that (6. . . 2 = 1 in Table 1. | .27).38) yields pression 2( 1 + 2 )Y 1 + 2 1 + 2 −1 ( 1 . We have provided in Table 1 the explicit form of YJ M ( 1 2 |Ω1 Ω2 ) for 1 = 2 and 2 = 1 to illustrate the construction. Ω2 ). The result is illustrated for the case 1 = 2. Expression (6.47) can serve now to obtain recursively the elements of the family B + − for M = 1 + 2 − 2. 2 |Ω1 . −( 1 + 2 − 1).37). To satisfy this equation one needs to choose c = − 1 / 2 . M = −( + ). M = −( + − ).1: Construct following the procedure above the three functions YJM ( 1 . in particular.h. | . 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 (6.45) for some unknown constant c. . Exercise 6. etc. 2 |Ω1 .37) is satisﬁed.47) (Ω2 ) − 1 1 2 + Y 2 1 1 −1 (Ω1 )Y 2 2 + Y 2 1 1 (Ω1 )Y 2 2 −1 (Ω2 ) . thereby. We like to construct now the family of total angular momentum functions B + − = {Y + − M ( . Normalization yields furthermore Y = 1 1 + 2 −1 1 + 2 −1 We have thereby determined ( 1 2 |Ω1 Ω2 ) (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) 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 ).36. . of (6.h. ).39) as required by (6. and all total angular momentum functions for a given choice of 1 and 2 .45).6. . ). Show that the resulting functions are orthonormal. 1 + 2 − 1.

1|Ω1 . Ω2) Y20 (2.632456 0. 1|Ω1 . . 1|Ω1 .547723 0. 1|Ω1 .150 Addition of Angular Momenta and Spin Y33 (2. 1|Ω1 . 1|Ω1 . Ω2) Y3−3 (2. Ω2) Y1−1 (2. 1|Ω1 . Ω2) Y30 (2.57735 0.774597 − − − − 0. Ω2) Y3−1 (2.447214 −0.258199 −0.816497 0. 1|Ω1 .57735 0. 1|Ω1 . Ω2) Y32 (2. Ω2) Y2−1 (2.774597 0 2 5 8 15 1 6 3 10 1 3 0.632456 0.547723 0. Ω2) Y2−2 (2.408248 −0.730297 0.1: Some explicit analytical and numerical values of Clebsch-Gordan coeﬃcients and their relationship to the total angular momentum wave functions and single particle angular momentum wave functions.547723 0. Ω2) Y3−2 (2. Ω2) Y11 (2. 1|Ω1 .316228 0.258199 0.816497 Y2−2 (Ω1 )Y11 (Ω2 ) Y2−1 (Ω1 )Y10 (Ω2 ) Y20 (Ω1 )Y1−1 (Ω2 ) Y2−2 (Ω1 )Y10 (Ω2 ) − 2 3 Y2−1 (Ω1 )Y1−1 (Ω2 ) Y2−2 (Ω1 )Y1−1 (Ω2 ) 1 Table 6.816497 −0. Ω2) Y31 (2.816497 Y22 (Ω1 )Y1−1 (Ω2 ) 1 15 1 3 3 5 1 5 1 2 3 10 2 5 1 2 1 10 2 3 1 3 Y20 (Ω1 )Y11 (Ω2 ) Y21 (Ω1 )Y10 (Ω2 ) 0.408248 −0. Ω2) Y22 (2. 1|Ω1 .57735 3 10 3 5 Y2−1 (Ω1 )Y11 (Ω2 ) Y20 (Ω1 )Y10 (Ω2 ) Y21 (Ω1 )Y1−1 (Ω2 ) −0.57735 0.707107 0.447214 0. 1|Ω1 . 1|Ω1 .632456 −0.730297 −0.707107 0.707107 0.774597 0.316228 0. Ω2) − − − − Y22 (Ω1 )Y11 (Ω2 ) 1 Y21 (Ω1 )Y11 (Ω2 ) 2 3 1 3 2 5 1 2 1 10 1 5 1 2 3 10 1 15 1 3 3 5 Y22 (Ω1 )Y10 (Ω2 ) 1 3 2 3 8 15 1 6 0.57735 0. Ω2) Y10 (2.707107 0. 1|Ω1 .57735 −0. 1|Ω1 . Ω2) Y21 (2.547723 0.

48) (6.53) (6. ζ = ± and the matrix elements < ζ |σk | ζ > are as deﬁned in Section 5.254)] 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 . commute with each other ζ ζ aζ . 5. b† ζ ζ = aζ . aζ .52) and | 2 m2 2 According to Section 5.ζ ζ.49) where ζ. a† and bζ .3: Explicit Expressions of CG–Coeﬃcients 151 Exercise 6. m − 2 .9. (6. 5. hence. − 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. (5. b† ζ ζ = 0 = 0 . Deﬁnition of Spinor Operators for Two Particles In contrast to Sections 5.54) 2 + = √ b† ( 2 +m2 )! √− ( 2 −m2 )! .9. 1 . (6. + 2 − J−M +1 for = J + 1 2J+2 2 J−M for = J − 1 2J 2 = .10 [cf. bζ = = a† . aζ bζ . we are dealing now with two particles carrying angular momentum or spin. a† ζ .2. 6. a† ζ ζ b† . These expressions will be derived now. b † ζ = δζζ = δζζ . bζ . m + 1 .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 ).50) (6. 1 (6. The creation and annihilation operators are again of the boson type with commutation properties aζ . J. Accordingly.287) to two particles (1) J k k = = (2) J 1 2 1 2 ζ.6. For this purpsose we will employ the spinor operators introduced in Sections 5.51) The operators aζ . b† refer to diﬀerent particles and.ζ a† < ζ |σk | ζ > aζ ζ b† < ζ |σk | ζ > bζ ζ (6.10.10. b† ζ = 0.10 where we studied single particle angular momentum and spin. we extent deﬁnition (5. M | . It is actually possible to state explicit expressions for any single coeﬃcient (JM | 1 m1 2 m2 ). 2 . M | . bζ = a† .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.

m2 |J. 2) We will also require below the raising and lowering operators associated with the total angular momentum operator (6. m1 1 | 2 . 5. 2 )) . (6. J3 |J. m2 a† + ( 1 2 = a† − ( 1 1 −m1 (6.69) − 2| ≤ J ≤ + 2 . can be expressed in terms of the number operators 1 † 1 ˆ ˆ a† a+ + a† a− .55) . ˆ kj |J. (1) J 2 | (2) J 2 | 1 m1 1 2 m2 2 Addition of Angular Momentum and Spin = = 1( 1 2( 2 + 1) | 1 m1 + 1) | 2 m2 1 2 . (2) J which. k = 1. 2) 1.54). (6.e. (6. M ( 1 . we like to recall for future reference that the operators (5. M ( 1 . (2) J 2 . j = 1. = M |J. m2 |J. m1 . 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 . 2) 2) can be expressed in terms of Clebsch-Gordan coeﬃcients (6. 2 .302.60) (6. M ( 1 .58) Jk + (2) Jk . 2) + 1) |J. = = 1( 1 2( 2 2) 2) (6. 2) 2) (6. |J. . The operator of the total angular momentum/spin of the two particle system is J = with Cartesian components Jk = (1) J + J (6. 2) . (j) 2 ˆ ˆ J = kj ( kj + 1 ) .m2 | 1 . J3 . 2 .152 It holds.68) The states |J. The aim of the present Section is to determine closed expressions for the Clebsch–Gordan coeﬃcents ( 1 . 1 . in analogy to Eqs. (5. i. m2 1 2 ( 1 . (1) J .59) J± = J1 ± i J2 . M ( 1 . M ( 2) 2) = J(J + 1) |J. M.65) (6. (1) J (2) J 3 | 1 m1 1 3 | 2 m2 2 = m1 | 1 m1 = m2 | 2 m2 1 2 (6. M ( 1 . (1) J 2 |J. 2 )). M ( 1 . M ( (2) J 2 |J. (6.. m1 1 | 2 . M ( 1 . + 1) |J.64) (6.61) (6. m2 2 .18) as follows = | 1 m1 . 1. are | 1 . hence.67) = j |J. 2 .62) .300). M ( 1 . M ( 1 .66) (6. 3 .63) according to At this point. mj j and.59) We seek to determine states |J. m1 . 6. for such states should hold J2 |J. (6. M ( 1 .56) The states | 1 . (1) J 2 . −J ≤ M ≤ J . 2 ) which are simultaneous eigenstates of the operators 2 2 J2 . as usual are denoted by their respective quantum numbers J.303). m1 1 | 2 . M ( 1 . . 2 . M ( 1 . which describe a two particle system according to (6. M ( 1 . mj j = j | j . 2.53. ˆ For the operators kj holds ˆ kj | j .

Employing (6. j = 1. expressing (k) J 3 through the creation and annihilation operators according to (5. (6. a† b† − a† b† − + − − + 2 + 1 1 = a† a+ .73) can be readily proven. K † 1 a† a+ − a† a− + b† b+ − b† b− . K † ] = (j) J 2 .74) + 1 J− J+ + J2 .70) will play a crucial role in the evaluation of the Clebsch-Gordan-Coeﬃcients.51) one obtains ˆ k1 . 4 Using J3 = (1) J 3 + (2) J 3 .75) The relationships (6.64. + − + − 2 2 − + = . K † 2 1ˆ ˆ 1 ˆ ˆ kj kj + 1.73) yields J3 . ˆ A similar calculation yields [k2 .73. K † 1 † a a+ + a† a− . a† b† + a† a− .288). + − − + 2 2 = 1 † 2K . the relationships (6.73) (6.71) (6. K † J± . 6. For example.6. 2 4 = 0 = = 0. This operator obeys the following commutation relationships with the other pertinent angular momentum / spin operators ˆ kj . b† = 0 .50.72) (6. K † (kj + 1) 2 2 1ˆ † 1 ˆ kj K + K † (kj + 1) 2 2 1 ˆ 1 ˆ K † kj + kj . 6. We note that. a† b† + + − − + 2 2 − 1 1 = a† b† − a† b† = K† . + − − + 153 (6. (6. K† J3 . a† b† − a† b† + − + − + − − + 2 1 1 = a† a+ .71–6. K † + K † 2 2 3 † †ˆ K kj + K .3: Explicit Expressions of CG–Coeﬃcients The Operator K † The following operator K † = a† b† − a† b† .74) imply 3 2 J2 . a† b† − a† a− . K † + kj . a† b† + − − + 2 + 2 − 1 1 − b† a† b+ . 6. K † (j) J 2 . using (6.71–6.71) one can show K† = = = = = 1 ˆ ˆ kj (kj + 1).65) and (6. K † = 0. 2 2 3 ˆ = K † kj + K † . j = 1. due to J2 = K† 1 2 J+ J− 1 † K . b† − b† a† b− . and applying the relationships (6.

j1 + ) K † 2 2 (6. 2 ) remain eigenstates of J2 and J3 with the same quantum numbers. j2 + ) 2 2 n n = N (n.63. 2) 2) ˆ K † kj + = K† = j 3 † K + K † j ( j + 1) |J. j2 ) = |j1 . M ( 1 .75) ascertain that under the action of K † the states |J.67) (j) J 2 K † |J. according to (6. a† b− − a† b† + + + − + = 0. 2) = = (j) 2 J . i. 2) = N |J. M ( 1 . One can generalize property (6. M ( 1 . 6.292) one can derive similarly J+ . j2 ) . 2 2 + n ) 2 (6. j1 + j2 (j1 + n n .2. M ( 1 + . 2 ) j + 4 1 3 + K † |J. 2) is a state with quantum numbers 1 1 + 1 2 and = N |J. b† + − + + + − The property [J− . M ( 1. 2 ) produces again total angular momentum eigenstates to the same J and M quantum numbers of J2 and J3 . 2 2 1 + ) . but for diﬀerent 1 and 2 quantum numbers of the operators (1) J 2 and (2) J 2 .78) a state which has been used already in the construction of Clebsch-Gordan coeﬃcients in Section 6. 6. it holds K † |J. 2 (6. 1 2 + 2 . j1 + j2 (j1 .62. j1 1 + .. M ( 1 + n . |j1 + j2 .77). K † ] = 0 is demonstrated in an analoguous way.154 Starting from (5. M ( 1 . j1 + j2 (j1 . j1 1 |j2 .76) Here N is an unknown normalization constant. 2 ) . = − a† a− . M ( 1 . (6. K † = Addition of Angular Momentum and Spin a† a− + b† b− .79) n |j1 + j2 . M ( 1 . 2) We want to demonstrate now that the action of K † on the states |J.77) where N is another normalization constant.76) and state K† n |J. on the state |j1 + j2 . Action of K † on the states |J. M ( 1 .73. j + 2 2 2) However. The commutation properties (6. M ( 1 . namely.72) and (6. K† + K † (j) J 2 |J. M ( 1 . We consider now the case that (K † )n acts on the simplest total angular momentum / spin state. To demonstrate that the resulting states are eigenstates of (1) J 2 and (2) J 2 we exploit (6. Application of (K † )n to this state yields. this result implies that K † |J. 6. M ( 1 . a† b† + b† a† b− .e. 4 3 + j ( j + 1) |J. j2 2 .

57) and exploiting the fact that J− and K † commute [c. Strategy for Generating the States |J. 2) . 1 . 2 ) can be expressed through the r. M ( 1 . 5. j1 + n .6.82. 6. M ) . 2) = N( 1 + 2 − J. Combining (6. 2 = j2 + n 2 (6. − 1. 2 ) exploits the expression (6. 1. j1 .83) The derivation of this expression will be provided further below (see page 158 ﬀ). j2 + n ). M ( 1 . 2 ) for −J ≤ M ≤ J as follows |J. allow one to obtain the states |J. M ( 1 .82) 1 × | (J + 2 1 × | (J + 2 1 (J + 2 1 − 1 ) . (6. 2 ) ∆(J. The latter states. J( 1 .86) b† + J+ 2 − 1 |Ψo . K† 1 2 1 + 2 −J × × .81) = N( 1 + 2 1 2 − J. (6. For this purpose one needs to choose in (6. = (J + M )! (2J)! (J − M )! 1 2 2) (6.84) (6. j2 as follows J = j1 + j2 which is equivalent to n j1 j2 Accodingly. J( 1 . M ( 1 . M ( 1 . M ) (J + × K† 1 − 2 )! (J + 2 − 1 )! J+ 1 − 1 + 2 −J (J− )J−M a† + × 2 (6. of (6. 1 = j1 + n 2 .104.82) for |J. J( 1 . 2) Our construction of the states |J. M ) (J− )J−M |J. 2 2 It is now important to notice that any state of the type |J.105) of the spherical harmonics. (J + 2 − − 2) 1 1) 2 The normalization constant appearing here is actually N( + − J. in analogy to the construction (5.79).80) = = = 1 1 2 2 − J 1 n − = (J + 2 2 n 1 − = (J + 2 2 + 1 1 − − 2) 2) .h. 2) 2) .3: Explicit Expressions of CG–Coeﬃcients 155 where we denoted the associated normalization constant by N (n.84) with (6. J( 1 .79) n.f. 2) 1 2 = ( 1 + 2 (2J + 1)! − J)! ( 1 + 2 + J + 1)! 1 2 . 2 ) .85) ∆(J. 2) = ∆(J.s. holds |J. (6.74)] yields |J. (6.

90) (6.e.264). st The desired expansion of |Gr can be obtained from an alternate evaluation of I(λ. + + Taylor expansion of the two exponential operators yields immediately I(λ. in terms of | 1 . x. x.e.88) in terms of monomials (6.t (6. (6. For this purpose we introduce the generating function I(λ.57)]. m2 Gordan-coeﬃcients.. m1 1 | 2 .87) + − + − i. (6. r!s!t! r.91) which.86) in terms of monomials 1 +m1 1 −m1 2 +m2 2 −m2 a† a† b† b† |Ψo .87). One obtains then using J− = a† a+ + b† b+ (6.50–6. y) = exp (λJ− ) exp x a† exp x b† |Ψo . follows from the commutation properties (6.93) = f (a† + λ a† ) g(b† + λ b† ) |Ψo .89) i. y) is a generating function for the states |Gr deﬁned in (6.s.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. x. + − + − . (6. a† ] = [b+ . b† ] = 0 [cf. Comparision with (6. in analogy to (5.156 Addition of Angular Momentum and Spin Our strategy for the evaluation of the Clebsch-Gordan-coeﬃcients is to expand (6. I(λ.52). x. 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 (6.92) − − and noting [a+ .86) |Gr = Jr a† st − + s b† + t |Ψo (6. 2 [cf. y) = λr xs y t r |Gst ..69) yields then the Clebsch- Expansion of an Intermediate State We ﬁrst consider the expansion of the following factor appearing in (6.88).

94) yields I(λ. y) = s. 2) = s. M ( 1 . y) = exp a† + λ a† + − exp b† + λ b† + − |Ψo . using the commutation property (6.98) a† − s a† + b† + s .90) allows one to infer |Gr s. With K † given by (6.88).94) One can infer from this result the desired expressions for |Gr .97).99) .t × a† + s−q s q b† + t r−q t−r+q a† − q b† − |Ψo (6. 2 ) in terms of states | 1 . to obtain the desired expansion of |J.96) and.3: Explicit Expressions of CG–Coeﬃcients We conclude I(λ.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. m1 1 | 2 . Operation of this operator on (6. m2 2 .s. Expanding the exponentials in st (6. x.70) holds K† 1 + 2 −J = s 1 + 2 −J b† − s 1 + 2 −J−s (−1)s 1 + 2 −J−s (6. one can write the right factor in (6.86) (J− )J−M a† + = q J+ 1 − 2 b† + J+ 2 − 1 |Ψo 1 (6. M ( 1 .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. |J.t = q r! × a† + s q s−q t r−q a† − q × b† + t−r+q b† − r−q |Ψo (6.95) Comparision with (6. x.t × λ v b† − × r−q v × = t s−u v λu a† − r! q |Ψo λr xs y t r! s! t! r. 157 (6.50).97) yields.q (−1)s (6. using the deﬁnition (6.

Using (6. The summation over q corresponds then to the summation over m1 .82) and (6. M − +q+s One can conclude that this expression reproduces (6.104) .101).82). n) 2 )|J.103) = 1. m2 in (6. For this purpose we introduce j1 = 1 (J + 2 1 2 − J.q (−1)s × (6. J( 1 . √ 2J + 1 1 2 .69) since.69) if one identiﬁes m1 = 1 −q − s .83) of the normalization constant N ( 1 + deﬁned through (6. 2 ) (6. J( 1 .103) this can be written 1 = N 2 ψ(j1 . q = 1 − m1 − s and m2 = M − m1 .102) We want to determine now the expression (6. The Clebsch-Gordan coeﬃcents are then ﬁnally ( 1 .158 Addition of Angular Momentum and Spin ( 1 + 2 − J)!(J − M )!(J + 1 − 2 )!(J + 2 − 1 )! s!( 1 + 2 − J − s)!q!(J + 1 − 2 − q)!(J − M − q)!(M + 2 − a† + 2 1 −q−s 1 + q)! a† − q+s b† + M + 2 − 1 +q+s b† − 1 + 2 −M −q−s |Ψo The relationships (6. j2 . (6. n = 1 + 2 −J. n)|ψ(j1 . M ) (J + 2 )! (J + 2 − 1 )! s.53. 1 .6. M ( 1 .100) ( 1 + 2 − J)!(J − M )!(J + 1 − 2 )!(J + 2 − 1 )! s!( 1 + 2 − J − s)!q!(J + 1 − 2 − q)!(J − M − q)!(M + 2 − 1 1 + q)! × (2 ×| 1 .101) Note that m1 + m2 = M holds. j2 = 1 (J + 2 2 − 1) . − q − s)!(q + s)! 2 × (M + 1 − 1 + q + s)!( 1 1 + 2 − M − q − s)! 2 − q − s 1 | 2. 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 . m2 |J.54) between creation operator monomials and angular momentum states allow one to write this |J. j2 . − 1 . m1 . 2 ) we consider the scalar product J. To determine N = N ( 1 + 2 −J. (6. × 2) = N( 1 + 2 1 − J. according to (6. 2 ) − 2) . 2 ) ∆(J. m2 = M − 1 + q + s.

m2 n1 + m1 n1 n2 + m2 n2 λm1 +m2 (6. j2 .112) s . (6. One obtains then.106) The orthonormality of the states occurring in the last expression allows one to write (6. m2 2 .107) s = (n!)2 s 2j1 + n − s 2j1 The latter sum can be evaluated using 1 1−λ a property which follows from ∂ν ∂λν 1 1−λ n1 +1 n1 +1 = m1 n1 + m1 n1 λm1 (6.108).108) yields the identity n1 + r − s n1 n2 + s n2 = n1 + n2 + r + 1 n1 + n2 + 1 . 1 1−λ n1 +1 1 1−λ n2 +1 = m1 .111) Comparision with (6. (6.110) which can be written 1 1−λ n1 +n2 +2 = r s n1 + r − s n1 n2 + s n2 λr (6. applying (6. We employ the expression (6.109) and Taylor expansion of the left hand side of (6.70) for the operator K † . n) = K† n 159 |j1 . j2 . n) = 1 (2j1 )!(2j2 )! a† + n! (2j1 )!(2j2 )! a† + 2j1 −n−s s 2j1 s n s 2j2 a† b† + − |Ψo = n−s (−1)s a† b† − + s b† + (−1)s a† − s (2j1 + n − s)!s!(2j2 + s)!(n − s)! s!(n − s)! 2j2 +s b† + b† − n−s (2j1 + n − s)!s!(2j2 + s)!(n − s)! |Ψo .57) for these states and the expression (6. m1 1 |j2 . j2 2 .3: Explicit Expressions of CG–Coeﬃcients where |ψ(j1 .108) = λ=0 (n1 + ν)! n1 ! (6.6. we obtain |ψ(j1 .104) 1 = N2 (n!)2 (2j1 )!(2j2 )! (2j1 + n − s)!(2j2 + s)! s!(n − s)! 2j2 + s 2j2 (6. (6. j1 1 |j2 . Accordingly.105) The ﬁrst step of our calculation is the expansion of ψ(j1 . n) in terms of states |j1 . j2 .108) twice.

11) J = J (1) + J (2) .. and J (2) are vectors in R .114) as long as J. 2 . 2 . (6. 1 . m2 |J.103) one obtains the desired result (6. (6. m2 |J. J. e. −m2 .118).h. m1 ). namely. −m2 |J.115) This relationship is a trivial consequence of (6.102) by replacing ( 1 . of (6. m1 . (2j1 + 2j2 + 1)! (6. 6.s. M ) corresponding to (6. interchanging the operators J (1) and J (2) results in J = J (2) + J (1) . m2 ) and. −m1 2|J. and . m2 |J.113) Using the identities (6. −m1 . (6. of this equation J (1) = J (2) − J . (6.120) The symmetry properties (6. m1 . M | 1 .114) by. m1 . m1 .s. 2 . m2 . M ) .114) one obtains − J = − J (1) − J (2) The respective Clebsch-Gordan coeﬃcients ( 1 . For the quantum mechanical addition of angular momenta the Clebsch Gordan coeﬃcients ( 1 . −M ) = (−1) 1 + 2 −J ( 1 . 1 .83). (6. J (1) on the r. −m1 . J (1) . m1 |J. 2 . m2 |J. of (6. M ) ( 1 .116). M ) = (−1) 2 +m2 2J + 1 ( 2 .g. 2 .119) The corresponding symmetry property of the Clebsch-Gordan coeﬃcients is ( 1 . m2 ) by ( 1 . one can interchange also the operator J on the l.116) one expresses the Clebsch-Gordan coeﬃcient on the r. manner to the coeﬃcients ( 1 . M ) .s.160 Applying this to (6. (6.118) Finally. m2 |J.102) of the Clebsch-Gordan coeﬃcients.116) through formula (6.114) show a simple relationship to the Clebsch Gordan coeﬃcients ( 2 . m1 . m1 |J.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 .h.116) If one takes the negatives of the operators in (6. M ) = (−1) 1 + 2 −J ( 2 . 2 1+1 (6.4 Symmetries of the Clebsch-Gordan Coeﬃcients The Clebsch-Gordan coeﬃcients obey symmetry properties which reﬂect geometrical aspects of the operator relationship (6. m1 ) by ( 2 . We will demonstrate this now. (6. and (6.120) can be readily derived from the expression (6.115). m2 . ( 2 . ( 1 . 2 . M ) corresponding to (6.h. m1 ) .117) are again related in a simple 2 . To derive relationship (6.114) For example. vice versa. −M ) .

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

f. (6. 1 |ˆ) + f (r) σ · p Yjm (j ± 1 . hence.174–5.186) Here (( · · · )) denotes again conﬁnement of the diﬀusion operator ∂r to within the double bracket. for θ = 0 (6. ˆ (6. 6. The result can also be stated in the compact form σ · r Yjm (j ± 1 . 1 |ˆ) r r r 2 2 2 2 2 2 (6. φ) 2 2 2 = − j+ 1 2 4π (6. Since f (r) is independent of θ and φ follows ˆ ((σ · p f (r))) = −i (( ∂r f (r) )) σ · r . Noting that p = −i r is a ﬁrst order 2 2 diﬀerential operator it holds ˆ ˆ ˆ σ · p f (r) Yjm (j ± 1 .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 |θ.182) σ · r Yj 1 (j − 1 .182) 0 Since σ · r.180) and (6. ˆ (6. 1 |ˆ). φ) . φ) 2 2 2 . 1 |ˆ) = ((σ · p f (r)))Yjm (j ± 1 . For this purpose we consider the application of σ · r in the case of θ = 0.180).184) 0 1 space χ 1±1 2 2 one can conclude from (6. identiﬁed the sign of (6.181.1 states χ 1 ± 1 2 2 2 [c.180) where the sign could depend on j. 1 |ˆ) at θ = 0 are r r 2 2 2 2 Yj 1 (j − 1 .187) 1 2 .168 Addition of Angular Momentum and Spin 1 Yjm (j ± 2 .148) the particular ˆ 1 1 states Yj 1 (j − 2 .185) We have. ˆ 2 2 2 2 2 2 for θ = 0. 2 |ˆ) and Yj 1 (j + 1 . One can conclude then ˆ σ·r = ± ˆ 0 1 1 0 (6. We want to determine ˆ its action on the wave functions f (r) Yjm (j ± 1 . 1 |θ = 0. φ) and (5. (5.147.160). 1 |θ = 0.224)] 1 0 σ·r = ˆ . 1 |ˆ) r 2 (6. therefore. According to (5.180) and.183) in case θ = 0 becomes in the standard representation with respect to the spin.181) 0 = j+ 1 2 4π Yj 1 (j + 1 . have proven (6. given by ˆ σ · r = σ1 sin θ cos φ + σ2 sin θ sin φ + σ3 cos θ . 6.188) . φ) = − Yj 1 (j + 1 . (6. 1 |θ = 0. 1 |θ = 0. We want to demonstrate ﬁnally that the “−”-sign holds in (6. 1 |ˆ) = −Yjm (j ˆ r 2 2 ˆ The Operator σ · p ˆ The operator σ · p plays an important role in relativistic quantum mechanics.

1 |ˆ) r 2 (6.190) can be related to the angular momentum operator. e. hence. r 2 (6. ˆ For the test function h(r) holds ˆ ˆ p·rh = = Using (1/r) = −r/r3 and r · ˆ −i −i · r h r = −i h r ·r − i r· h. 1 |ˆ) r 2 (6. we conclude the intuitively expected identity 1 ˆ ˆ p×r = − J . 1 |ˆ) r 2 1 2 (6.191) (6.6. 1 |ˆ) = r 2 2 ˆ i ∂r f (r) − f (r) σ · p σ · r Yjm (j ˆ 1 2 169 . r r r Using ∂2 (1/r) = −x2 /r3 . 1 |ˆ) . 1 |ˆ) = 0 and σ = 2S/ .g. h r (6.155. 1 |ˆ) = i [ ∂r + r + ] f (r) Yjm (j 2 r r Using (6. 1 |ˆ) r 2 (6. 6.. r 2 2 1 J ·S 2 1 ˆ σ · p f (r) Yjm (j ± 2 .195) .187) one obtains ˆ σ · p f (r) Yjm (j ± 1 .197) 1 ˆ σ · p g(r) Yjm (j − 2 .156) this yields ﬁnally 1 ˆ σ · p f (r) Yjm (j + 2 .198) .186) for both terms in (6. ∂3 (1/r) = −x3 /r3 and ∂2 x3 = x3 ∂2 . x3 x2 ˆ ˆ ( p × r )1 h = −i (∂2 − ∂3 ) h (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 (6.193) r r 1 1 1 = −i (∂2 x3 − ∂3 x2 ) h − i h (x3 ∂2 − x2 ∂3 ) .230) allows one to express ˆ ˆ ˆ ˆ ˆ σ · p σ · r = p · r + i σ · (p × r) .196) = i j+ ∂r + r 1 2 3 2 1 f (r) Yjm (j − 2 . r Altogether we obtain.189) The celebrated property of the Pauli matrices (5. Corresponding results are obtained for the other components of p × r and. ∂3 x2 = x2 ∂3 we obtain ˆ ˆ ( p × r )1 h = −i 1 1 (x3 ∂2 − x2 ∂3 ) h = − J1 h r r (6.53). using ∂r Yjm (j ± 1 .194) ˆ ˆ where J1 is deﬁned in (5.192) ˆ ˆ The operator p × r in (6. 1 |ˆ) r 2 = i ∂r + −j r 1 g(r) Yjm (j + 2 .5: Spin–Orbital Angular Momentum States Using (6. To demonstrate this we consider one of its cartision components.

r (6. 6. 5.198).200) = (j + 1 )(j + 3 ).203) (6. We begin with the coeﬃcients (6. For m = j − 1 one can state. 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 (6.170 Addition of Angular Momentum and Spin To demonstrate the validity of this key result we note that according to (5.204) (6. 1 |ˆ) r r 2 2 2 2 2 2 2 yields ˆ (σ · p)2 f (r) Yjm (j + 1 . (6. 1 |ˆ) = r 2 2 which agrees with (6. 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 . j 2 + 2j + 2 3 4 (6. r ∂r2 r 2 (6. .143–6.197. 6. 6.2. Yjj−1 (j − 1 . 6.199).143. 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 . j + . 1 |ˆ) r 2 2 r r2 r2 j 2 + 2j + 3 2 2 2 4 = [ −∂r − ∂r + ] f (r) Yjm (j + 1 .202) Yjj (j − 1 .206) . according to (6. 1 |ˆ) r 2 2 r 2 3 i i 1 2 1 r = [ i∂r + ( − j) ] [ i∂r + (j + ) ] f (r) Yjm (j + 2 .99) holds ˆ (σ · p)2 = − 2 2 = ∂2 J2 r + 2 . adopting the method in Sec. In this case holds. as well as (6. according to (6.146). in fact.151). 1 |ˆ) r 2 2 The Clebsch-Gordan coeﬃcients are then (j − 1 .199) We want to show that eqs. 1 |ˆ) = J 2 Yjm (j + 1 . 1 |ˆ) r 2 2 r r2 and.43).205) which agrees with the expressions (6.201) 1 3 (j + )(j + ) Yjm (j + 1 .143. − |j. i.144) and. 2 2 (6.. We note ˆ (σ · p)2 f (r) Yjm (j + 1 .101). j) 1 2 1 2 1 2 1 2 = = 1 0 (6. j) 2 2 2 2 (j − .2. j − 1 . using (5. are consistent with this identity.e. consider ﬁrst the case of the largest m-value m = j. For this purpose we use the construction method introduced in Sec. 1 |j.144) for m = j. Evaluation of Relevant Clebsch-Gordan Coeﬃcients We want to determine now the Clebsch-Gordan coeﬃcients (6. 6. r 2 2 2 2 2 2 − r 2 ∂r r + J2 r2 f (r) Yjm (j + 1 .45).230.

137)] J− (tot) = J− + S− (6.208) (6. j − 1 .5: Spin–Orbital Angular Momentum States The corresponding Clebsch-Gordan coeﬃcients are then 1 (j − 1 .s. (· · · |jj − 3).210).144) for m = j − 1. The further Clebsch-Gordan coeﬃcients (· · · |jj − 2).s. 1 |j. [c. 2 |j.143.210)] yields (j + m)(j − m + 1) Yjm−1 (j − 1 .f. (6. 6. 1 |ˆ) = r 2 2 j+m 2j + (j + m − 1)(j − m + 1) Yj− 1 m− 3 χ 1 1 2 2 2 2 j+m Yj− 1 m− 1 χ 1 − 1 2 2 2 2 2j j−m 2j (j + m)(j − m) Yj− 1 m− 1 χ 1 − 1 2 2 2 2 + or Yjm−1 (j − 1 . j − 1) 2 2 2 171 = = 2j − 1 2j 1 2j (6. 1 |ˆ) = r 2 2 Applying J− (tot) j+m Yj− 1 m− 1 χ 1 1 + 2 2 2 2 2j j−m Yj− 1 m+ 1 χ 1 − 1 .210) to (6. (6. m − 3 . 2 2 2 2 2j (6. etc. m − 1) 2 2 2 2 = j +m−1 2j (6. 1 .143. Expression (6.6.2. j − 3 .206).207) (6. 2 2 2 2 2j . 1 . j − 1) 2 2 2 2 which again agrees with the expressions (6.143.h. and J− + S− to the r.211) to the l. 6. 6. 1 .209) (j − 1 .h.144) for the Clebsch-Gordan coeﬃcients by induction. as described in Sec. Let us verify then the expression (6. is obtained by applying the operator [c. 6.144) implies for j = m Yjm (j − 1 . − 1 |j.f.212) 1 + 2j(j − m + 1) 1 2j(j − m + 1) Yj− 1 m− 1 χ 1 − 1 2 2 2 2 j+m−1 Yj− 1 m− 3 χ 1 1 + 2 2 2 2 2j j−m+1 Yj− 1 m− 1 χ 1 − 1 . 1 |ˆ) = r 2 2 j+m−1 Yj− 1 m− 3 χ 1 1 + 2 2 22 2j (j − m) = This implies (j − 1 .202). are obtained by iterating the application of (6. (6..

172 Addition of Angular Momentum and Spin j −m+1 2j (j − 1 .144) by induction.218) where we have replaced the quantum numbers J. In this context Jclass and Jclass play the same role. j2 .143). Jclass and Jclass play equivalent roles. 2 . 2 . 2j + 2 (6. m) 2 2 2 2 2j + 1 (6. 6.144) for j = m − 1.6 The 3j–Coeﬃcients The Clebsch-Gordan coeﬃcients describe the quantum mechanical equivalent of the addition of two (1) (2) classical angular momentum vectors Jclass and Jclass to obtain the total angular momentum vector (tot) (1) (2) (1) (2) Jclass = Jclass + Jclass . 1 |j + 1 .215). m. m − 1) 2 2 2 2 = (6.144).145) from (6. (1) (2) (−tot) Obviously. M.214) yields (j. proven (6. all three vectors Jclass .143) by applying the symmetry relationships (6. m1 ) (6. m2 | 3 . m3 .116. 2j + 1 (6. 1+1 3 For 1 (j. one can obtain expression (6. m3 . The Clebsch-Gordan coeﬃcients (6. The latter relationships applied together read ( 1 . m + 1 ) = 2 2 2 2 and. (1) (2) (−tot) The coeﬃcients which are the quantum mechanical equivalent to Jclass + Jclass + Jclass = 0 are the 3j–coeﬃcients introduced by Wigner. j3 .146) can be obtained from (6. m1 .215) which follows from (6. 1 .143. − 1 |j. We have. 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 . hence. to reﬂect in the notation the symmetry of these quantities. m1 . m1 . − 1 |j. m2 1 (6. m2 . 1 . m.216) 6. j+m+1 . m + 1 . j2 m2 ) 2 . m3 ) = (−1) × 2 + 3 − 1 + 2 +m2 × 2 . 1 |j + 1 . 6.120). one obtains (j + 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. 6. m) = 2 2 2 2 Similarly.217) 2j + 2 (j + 1 .213) which is in agreement with (6. leading quantum mechanically to a symmetry of the Clebsch-Gordan coeﬃcients (JM | 1 m1 2 m2 ) with respect to exchange of 1 m1 and 2 m2 . m + 1 . 1 . by the set j1 . 1 .143. m − 1 .214) 2 2 +1 ( 3 . − 1 |j. the relationship (6. However. −m2 | 1 . 1 . using (6. m + 1 ) = 2 2 2 j+m+1 .

173 (6. However. The two integer entries J − 1 − m2 and J − 2 + m1 in (6.218) relating 3jcoeﬃcients and Clebsch-Gordan coeﬃcients.102) of the Clebsch-Gordan coeﬃcients. j2 . −j1 + j2 + j3 . To represent the full symmetry one expresses the 3j–coeﬃcients through a 3 × 3–matrix. for which no known classical analogue exists. The Regge symbol reﬂects a remarkable degree of symmetry of the related 3j–coeﬃcients: One can exchange rows.g.6: The 3j–Coeﬃcients We ﬁrst like to point out that conditions (6.34) imply j1 j2 j3 m1 m2 m3 j1 j2 j3 m1 m2 m3 = 0 if not m1 + m2 + m3 = 0 = 0 if not |j1 − j2 | ≤ j3 ≤ j1 + j2 . j3 form the sides of a triangle and the condition is symmetric in the three quantum numbers. m2 . are identical to the integer arguments which enter the analytical expression (6.219) (6. m3 . discovered by Regge. According to the deﬁnition of the 3j–coeﬃcients one would expect symmetry properties with respect to exchange of j1 .102) are obtained each through the diﬀerence of two entries of the Regge-symbol. For this purpose one can use the symmetry transformations to place the smallest element into the upper left corner of the Regge symbol. e. Because of its high degree of symmetry the Regge symbol is very suited for numerical evaluations of the 3j–coeﬃcents. there exist even further symmetry properties. the Regge-symbol. 6. states that j1 . The reader may note that the entries of the Regge-symbol. j2 . In case of a non-cyclic exchange of rows and columns the 3j–coeﬃcent assumes a prefactor (−1)Σ .222) m1 m2 m3 j1 + m1 j2 + m2 j3 + m3 The Regge symbol vanishes. m1 . These symmetry operations relate altogether 72 3j–coeﬃcients. one can exchange columns and one can reﬂect at the diagonal (transposition). m2 and j3 . i.223) (Σ + 1)!n11 !n22 !n33 !n23 !n32 !)) n=0 n31 n32 n33 .220) The latter condition |j1 − j2 | ≤ j3 ≤ j1 + j2 ..6. m3 and with respect to sign reversals of all three values m1 . anti-cyclic exchange and of a sign reversal are stated. In this way the values of 12 3j-coeﬃcients are related. as follows −j1 + j2 + j3 j1 − j2 + j3 j1 + j2 − j3 j1 j2 j3 j1 − m1 j2 − m2 j3 − m3 . √ safe for the prefactor 2j3 + 1 which is cancelled according to the deﬁnition (6. the so-called triangle condition.221) j1 j2 j3 −m1 −m2 −m3 where the the results of a cyclic. with respect to alltogether 12 symmetry operations. except when all elements are non-negative integers and each row and column has the same integer value Σ = j1 + j2 + j3 . 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.e.21. The Regge symbol also vanishes in case that two rows or columns are identical and Σ is an odd integer. = (6. These symmetries follow the equations j1 j2 j3 m1 m2 m3 = (−1)j1 +j2 +j3 = j2 j3 j1 m2 m3 m1 = (−1)j1 +j2 +j3 j2 j1 j3 m2 m1 m3 (6.

224) (6. m1 + m2 ) (2 − m1 )! (2 + m1 )! √ 10 (1 − m)! (1 + m)! (1 − m2 )! (1 + m2 )! 1 ≥ |m| ∧ 1 ≥ |m2 | ∧ 2 ≥ |m1 | (6. m1 + m2 ) (3 − m)! (3 + m)! (3m|2m1 1m2 ) = √ 105 (2 − m1 )! (2 + m1 )! (1 − m2 )! (1 + m2 )! 1 ≥ |m2 |) ∧ 2 ≥ |m1 | ∧ 3 ≥ |m| (6.228) √ (−1)2m1 −2m2 7 δ(m. (6.1.229) 1 1 (0m| m1 m2 ) = 2 2 i(−1)1−m2 δ(0. m2 ) √ 2 1 ≥ |m1 |) 2 (6.223-6.231) Here is an explicit value of a Clebsch-Gordan coeﬃcient: 1 1 (70 2 −15 1 |120 −10 60 2 −5 1 ) = 2 2 √ √ √ 4793185293503147294940209340 127 142 35834261990081573635135027068718971996984731222241334046198355 0. 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. 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. m) δ(−m1 . (6.227) (2m|2m1 1m2 ) = (−1)m+m1 (m + 2m2 ) δ(m.226) by means of a symbolic manipulation package (Mathematica) (1m|2m1 1m2 ) = (−1)1+m+m1 δ(m.225) sn−1 .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) (6.226) We like to state ﬁnally a few explicit analytical expressions for Clebsch-Gordan coeﬃcients which were actually obtained using (6.232) .10752786393409395427444450130056540562826159542886 We also illustrate the numerical values of a sequence of 3j-coeﬃcients in Figure 6.

233) . . 2 = 1.6: The 3j–Coeﬃcients 10 -3 ¥ ¢ £ 175 3j coefficient 20 120 18 16 14 12 10 8 6 4 2 0 -2 -4 -6 -8 -10 -12 -14 -16 ¡ ¢ -60 ¡ Figure 6. m2 = − 2 . . 1 . .5). . provide the irreducible representation for D(ϑ). . the 2–particle total angular momentum wave functions YJM ( 1 ..1: Oscillatory behavior of 3j-coeﬃcients. . . 2.+ 1 . . then for a basis {Y 1 m1 (Ω1 )Y 2 m2 (Ω2 ).. i. (2 (2 1 2 + 1) (2 + 1) × (2 1 2 + 1) + 1) .e. eigenfunctions of the single particle angular momentum operators J 2 and J3 . (6. If we deﬁne the matrix representation of R(ϑ) by D(ϑ). .+ 2 } the matrix has the blockdiagonal form D(ϑ) = 1·1×1·1 ¤ -10 60 700 m 10-m -40 -20 0 20 40 60 m 1·3×1·3 . . the rotations in single particle function space. . rotations in 2–particle function space. . m1 = − 1 . Similarly. Irreducible Representation We had stated above that the spherical harmonics Y m (Ω)).6.e. . i. Ω2 ) provide the irreducible representation for the rotation R(ϑ) deﬁned in (6. 2 |Ω1 .

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

. −k + 1. Examples of Tensor Operators The multiplicative operators C∞ ( ) → C∞ ( ) Y ( ) = def Y ( ) (6. the operator (5.7 Tensor Operators and Wigner-Eckart Theorem In this Section we want to discuss operators which have the property that they impart angular momentum and spin properties onto a quantum state. −k + 1. e.222) in case (ii) of single particle spin operators. k} such that k Tkq = q =−k (k) Dq q (ϑ) Tkq . that T belongs to a family of operators {Tkq .236) In this equation Dq q (ϑ) denotes the rotation matrix Dq q (ϑ) = kq |R(ϑ)|kq . that in case (i) T is an operator C∞ ( ) → C∞ ( ) acting on single particle wave functions. . e.223). The operator T may also act on multi-particle states like Y 1 m1 (ˆ1 )Y 2 m2 (ˆ2 ). S 2 = S1 + S2 + S3 or any other polynomial of Sk . In fact.e. Such operators T can be characterized through their behaviour under rotational transformations.238) . 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. The latter implies that T transforms like an angular momentum or spin state | m . i. .g. .42) in case (i) of the position representation of single particle wave functions or the operator (5.236.7: Tensor Operators 177 6. . Rotations transform |ψ as |ψ = R(ϑ)|ψ and T |ψ as R(ϑ)T |ψ . (k) (6.237) The operators T ∈ {Tkq . k} with the transformation property (6.g. This implies. . (6. 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) .g. for example. q = −k.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 .237) are called tensor operators of rank k. Note that in the examples mentioned the operator T would be deﬁned within the same representation as R(ϑ). q = −k. e.6. (k) (6. some of the r r examples considered below involve tensor operators T of this type. ∂x2 3 1 2 2 2 2 case (ii) the operator T could be a spin operator Sk deﬁned in (5. a mul∂2 ∂2 tiplicative operator T ψ(r) = f (r) ψ(r) or a diﬀerential operator T ψ(r) = ( ∂x2 + ∂x2 + ∂2 ) ψ(r).

7. we will assume presently that the rotation is chosen as follows R(ϑ) = exp ( ϑ+ L+ + ϑ− L− + ϑ3 L3 ) (6.237) do not require that the rotation R−1 (ϑ) is expressed in terms of Euler angles according to (5. but rather any rotation and any parametrization can be assumed. but are far easier to ascertain for any particular operator. Examples are Y00 Y1±1 Y10 Y2±2 Y2±1 Y20 1 = √ 4π Theory of Angular Momentum and Spin 3 3 r sinθ e±iφ = (x1 ± ix2 ) 8π 8π 3 3 = r cosθ = x3 4π 4π 1 15 2 2 ±2iφ 1 15 = r sin θ e = (x1 ± ix2 )2 4 2π 4 2π 15 2 15 = r sinθ e±iφ = (x1 ± ix2 ) x3 8π 8π 5 2 3 2 1 5 3 2 r2 = r cos θ − = x3 − 4π 2 2 4π 2 2 = (6. 2.2: Express the transformed versions of the following operators (a) T = x2 − x2 and 1 2 (b) S2 S3 in terms of Wigner rotation matrices and untransformed operators.240) 1 where we have deﬁned ϑ± = 2 (ϑ1 iϑ2 ) and L± = L1 ± iL2 . The fact that these operators form tensor operators of rank 1.3. x2 . In fact.7. x3 .235 . 1. For the following it is important to note that the rotation matrix elements (6. Exercise 6. Exercise 6. 3 follows from the transformation properties of the spherical harmonics derived in Section 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 0.239) These operators can be expressed in terms of the coordinates x1 . 2.6.178 are tensor operators of rank k.237). This choice of parametrization allows us to derive conditions which are equivalent to the property (6. .203).

(6. The property (6.80) follows kq |L+ |kq = −i (k + q + 1)(k − q)δq and.246. (6. (k + q + 1)(k − q) (k + q)(k − q + 1) (6. (6. For this 2 2 purpose state ﬁrst the proper rotation operator R(ϑ) and note then that S (1) · S (2) commutes with 1 the generators of the rotation of | 2 m1 (1) | 1 m2 (2) .235 . q =−k q+1 (6.237) yields ﬁrst k R(ϑ) Tkq R −1 (ϑ) = q =−k kq |R(ϑ)|kq Tkq . 0)T in case of small ϑ+ . 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) .7.6. J3 yields [J+ .237) for transformations characterized by inﬁnitesimal values of ϑ+ . ϑ3 )T . Tkq ] [J− .3: Derive Eqs. Exercise 6. hence.235 .243) From (5. Similar equations can be derived for inﬁnitesimal rotations of the form ϑ = (0. 6. J− . Expressing the results in terms of the angular momentum operators J+ .6: Form tensor operators of rank 1 in terms of the three components of acting on the space of 1-particle wave functions. Tkq ] [J3 . 2 Exercise 6. Tkq ] = = δqq and by subtracting Tkq on both sides of the k ϑ+ kq |L+ |kq Tkq .245) (6.244) [L+ . Tkq ] = −iTk q+1 (k + q + 1)(k − q) . Tkq ] = = = Tkq+1 Tkq−1 q Tkq . 0)T . (0.6.6.4: Is the 1-particle Hamiltonian 2 H = − 2 2m + V (|r|) (6. .235 .1 particles for which the ﬁrst spin is described by 2 the operator S (1) and the second spin by the operator S (2) .7. ϑ− . 0. 1 (6.241) Using R(ϑ) = 1 + ϑ+ L+ + O(ϑ2 ) this equation can be rewritten neglecting terms of order O(ϑ2 ) 1 + + k (1 + ϑ+ L+ ) Tkq (1 − ϑ+ L+ ) = 1 1 q =−k kq |1 + ϑ+ L+ |kq Tkq .247) These properties often can be readily demonstrated for operators and the transformation properties (6. Exercise 6. 0.6.237) be assumed then. ϑ− .248) a tensor operator? Exercise 6.247).7.7. We ﬁrst consider a rotation with ϑ = (ϑ+ .242) from which follows by means of kq |1 1|kq equation ϑ+ [L+ . ϑ3 .246) (6.5: Consider a system of two spin.7: Tensor Operators 179 The conditions can be derived if we consider the property (6.

behave like |kq | 1 m1 .237) is that their matrix elements 1 m1 γ1 |Tkq | 2 m2 γ2 obey simple relationships expressed in terms of Clebsch-Gordan coeﬃcients. i.235 .180 Theory of Angular Momentum and Spin 6. hence.m2 ( 1 m1 |kq m1 q. in fact.249) which demonstrates. J2 .8 Wigner-Eckart Theorem A second important property of tensor operators Tkq beside (6. Before we proceed we like to point out that the states | 2 m2 γ2 are also eigenstates of J 2 . J3 | 2 m2 γ2 = m2 | 2 m2 γ2 . m1 . These states according to (6. J 2 | 2 m2 γ2 = 2 2( 2 + 1) | 2 m2 γ2 . .m2 2 m2 ) (q + m2 ) Tkq | 2 m2 γ2 | 2 m2 γ2 (6.252) ∼δm1 q+m2 = ( 1 m1 |kq 2 m2 ) Tkq Similarly. i.1: Show that Φ 1 m1 (k.6. One can. J3 | 2 m2 γ2 = m2 γ2 | 2 m2 and the property (6.m2 ( 1 m1 |kq 2 m2 ) (J3 Tkq − Tkq J3 + Tkq J3 ) | 2 m2 γ2 = qTkq = q.250) 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. 2 |γ2 ) is an eigenstate of J 2 with eigenvalue 2 1 ( 1 + 1).247).253) Exercise 6. 2 |γ2 ) is an eigenstate of J 2 with eigenvalue + 1). (6.e. 2 |γ2 ) = ( 1 m1 |kq 2 m2 ) Tkq | 2 m2 γ2 .8.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 . one can show that Φ 1 m1 (k. possibly the total angular momentum– spin state of a compositie system.18) are deﬁned through Φ 1 m1 (k. Starting point of the derivation is the fact that the states Tkq | 2 m2 γ2 behave like angular momentum states of a composite system of two particles each carrying angular momentum or spin.251) The corresponding property for Φ 1 m1 (k. construct states Φ 1 m1 (k.21) of Clebsch-Gordan coeﬃcients J3 Φ 1 m1 (k. 2 |γ2 ) are orthogonal to | 1 m1 in case of diﬀerent quantum numbers 1 . (6. 1 m1 |Φ 1 m1 (k.e. 2 |γ2 ) which correspond to total angular momentum states.e. J3 . 2 |γ2 ) can be shown readily as follows using (6. The hermitian property of J3 and J 2 implies that states Φ 1 m1 (k. | 1 m1 γ1 denotes an angular momentum (spin) state. which is characterized also by a set of other quantum numbers γ1 which are not aﬀected by the rotational transformation R(ϑ). i. 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. 2 |γ2 ) = Cδ 1 1 δm1 m1 2 1( 1 (6. J3 are the generators of the rotation R(ϑ). the stated property. 2 |γ2 ) = q.

γ1 |Tkq | 2 m2 .8: Wigner-Eckart Theorem 181 In order to evaluate the matrix elements (6. γ1 ||Tk || 2 .258) We can then ﬁnally express the matrix elements of the tensor operators Tkq as follows 1 m1 .e.s. (6. (6. Using | 1 m1 + 1γ1 = 1 ( 1 + m1 + 1)( 1 − m1 ) J+ | 1 m1 γ1 (6. To prove this property we consider 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) for a diﬀerent m1 value. can be evaluated as easily as possible. γ2 = 2 1 +1 1 m1 .259) to a combination of magnetic quantum numbers m1 . e. One can then evaluate also the corresponding Clebsch-Gordan coeﬃcient ( 1 m1 |kq 2 m2 ) and determine 1 . say m1 + 1. γ2 ( 1 m1 |kq (6. m2 .257) which establishes the m1 –independence of 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) .259) The socalled reduced matrix element 1 . γ1 ||Tk || 2 . γ2 (−1)k− 2 + 1 ( 1 m1 |kq 2 m2 ) (6.2: Determine the matrix elements of the gradient operator d3 rF (r) G(r) of the type (6.8. 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. for which the l.261) .255) At this point the important property can be proven that 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) is independent of m1 . the matrix elements 1 m1 γ1 |Tkq | 2 m2 γ2 can be reduced to an m1 –independent factor. γ1 ||Tk || 2 . γ2 k− 2 + 1 √ 1 2 m2 ) (−1) 2 1 +1 = 1 .6.h.254) = 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) ( 1 m1 |kq 2 m2 ) .256) and noting that the operator adjoint to J+ is J− . γ1 ||Tk || 2 . i. its m1 –dependence being expressed solely through a Clebsch-Gordan coeﬃcient. q .32) Tkq | 2 m2 γ2 = and orthogonality property (6. m1 = q = m2 = 0. γ2 is determined by applying (6.g.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. γ2 . 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 . γ1 |Tkq | 2 m2 .260) Exercise 6.

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

(7. 1993) 1 183 .e. i. Poststrasse 22. and crystals. The potential (7. It leads to the stationary electronic states of the hydrogen atom (Z = 1). 7.Chapter 7 Motion in Spherically Symmetric Potentials We describe in this section the stationary bound states of quantum mechanical particles in spherically symmetric potentials V (r). φ. in case of the so-called bag model of hadronic matter.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 .1. The fourth potential V (r) = − Z e2 r (7.2) serve as schematic descriptions of quantum particles. Potential (7.2) V (r) = 0 r > R The third potential V (r) = 1 mω 2 r2 2 (7. See. Talmi (Harwood Academic Publishers. The potentials (7.4) is by far the most relevant of the four choices. The ﬁrst potential V (r) = 0 0 ≤ r ≤ R ∞ r > R (7. in potentials which are solely a function of r and are independent of the angles θ. Switzerland.1) conﬁnes a freely moving particle to a spherical box of radius R.3) describes an isotropic harmonic oscillator . for example. The corresponding wave functions serve basis functions for multi-electron systems in atoms. Four examples will be studied. Simple Models of Complex Nuclei / The Shell Model and the Interacting Boson Model by I. 7000 Chur.4) governs the motion of electrons in hydrogen-type atoms. The second potential is of the square well type −Vo 0 ≤ r ≤ R . for example. molecules..

3.12) .5) In the case of an angular independent potential angular momentum J = m r × v is a constant of ˙ motion.13) . 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 . For this purpose one concludes ﬁrst that the conservation of angular momentum J implies a motion of the particle conﬁned to a plane. . 2. (7. J3 ψE. k = 1. d ˙ J = v × mv + r × mv = 0 + r × er (−∂r V (r)) = 0 . (7. J ψE.99)] 2 1 2 J2 2 2 = − ∂r r + .61) of J 2 and Jk .f. ˆ A classical particle moving in a potential V (r) is governed by the Newtonian equation of motion (7.85– 5.7) ˆ The correspondence principle dictates then for the quantum mechanical Hamiltonian operator H and angular momentum operators Jk ˆ [H. using (7.15) (7.14) (7. Jk } = 0 .11) (7.9) − 2m 2m r 2mr2 the expressions (5..8) This property can also be proven readily employing the expression of the kinetic energy operator [c. Employing in this plane the coordinates r. i. one can state ˙ J = m r2 θ . ( + 1) ψE. Jk } where H is the Hamiltonian. 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. ˆ H ψE. the time variation of J can be written.m (r) m ψE. Accordingly.1 Radial Schr¨dinger Equation o ˙ mv = − er ∂r V (r) .m (r) can be chosen as simultaneous eigenstates of the Hamiltonian ˆ operator H as well as of J 2 and J3 . 2 . ˆ dt (7.e. 2.m (r) (7. one can conclude ˙ Jk = {H.87) for Jk as well as the commutation property (5.10) (7. (7.m (r) = = = E ψE. 2 .184 Spherically Symmetric Potentials 7. 2 2mr2 (7. In fact. Jk ] = 0 . (5.6) ˙ Since Jk is also equal to the poisson bracket {H. . stationary states ψE. 3 .5) and r = v.m (r) .m (r) . k = 1.m (r) .

. .24) . for the wave function of a quantum mechanical particle a diﬀerential equation which governs solely the r-dependence.9) one can write the stationary Schr¨dinger o equation (7. (7. (7.10) − 1 2 ∂ r + 2m r r 2 2 ( + 1) + V (r) − E 2mr2 . .20) by −2mr/ 2 yields the so-called radial Schr¨dinger equation o 2 ∂r − ( + 1) − U (r) − κ2 r2 r vκ.20) in the form of the one-dimensional Schr¨dinger equation o 2 − where 2m 2 ∂r + Veﬀ (r) − E φE (r) = 0 . i.19) where Y m (θ.m ψE. using (7.7.13). but can also trap particles in the latter space giving rise to strong scattering resonances (see Section ??). This barrier.m .18) = vE. in the present case.m vE. . One can write (7.m (r) Y m (θ.20) Since this equation is independent of the quantum number m we drop the index m on the radial wave function vE.m (r) . (7.m (r) 2 .4. (7. can exclude particles from the space with small r values. (r) = 0 . It can be solved by integration of 1 2 2 J2 r = ± ˙ E − V (r) − . Multiplying (7. .e.22) (7. φ) (7. r(t) is determined the angular motion follows from (7. − Adopting for ψE.21) Veﬀ (r) φE (r) = = V (r) + r vE.12) are obeyed and one obtains for (7.17) In analogy to the classical description one can derivem. 2 (7. together with the original potential. (7. ∞[ governed by the eﬀective potential (7..m (r) and E . Employing the kinetic energy operator in the form (7. ( + 1) 2mr2 . mr2 (t) (7.11.10). . .m (r) = 0. 7.m (r) describes the radial motion as a one-dimensional motion in the interval [0.m (r) = 0.m (r) 1 2 J2 ∂r r + + V (r) − E 2m r 2mr2 the functional form ψE.16) m 2mr2 Once. by integration of ˙ θ = J .1: Radial Schr¨dinger Equation o 185 This is a diﬀerential equation which governs solely the radial coordinate.22) which is the original potential 2 ( +1) V (r) with an added rotational barrier potential 2mr2 .23) This demonstrates that the function r vE. φ) are the angular momentum eigenstates deﬁned in Section 5. equations (7.11).

The radial part of this integral is ∞ 0 2 dr r2 vκ. the term Br−( +1) would contribute |B|2 0 dr r−2 +1) (7.28) (7. (r) ∼ A r + B r− −1 +1 + B r− (7. 2 2 2m 2 ψE.186 where we deﬁned U (r) κ2 = = − − 2m 2 Spherically Symmetric Potentials V (r) E (7.m (r) ∼ √ 4π B δ(x1 )δ(x2 )δ(x3 ) . for > 0 is not integrable. according to a well-known result in Classical Electromagnetism3 .182) for Y00 . Galindo and P. hence.30) and. (7. We replaced in (7. For r → 0 one may assume that the term ( + 1)/r2 becomes larger than the potential U (r). 2nd Ed. (r) (7. ”Classical Electrodynamics. In this case the solution is governed my ( + 1) 2 r vκ. κ assumes real values.m (r) ∼ √ B . New York.” by J. see. assumes the general form r vκ. . ψE. For = 0 the contribution of Br−( function is. This follows for > 0 from consideration of the integral which measures the total particle density. at r = 0 is found in the excellent monographs Quantum Mechanics I. using the expression (5. 1975).20) one needs to specify proper boundary conditions2 . in particular.25) (7. (r) = 0 . II by A. Pascual (Springer. for example.32) The total kinetic energy resulting from this contribution is. Berlin. Boundary Conditions In order to solve (7. Jackson (John Wiley.33) A detailed discussion of the proper boundary conditions.20) the index E by the equivalent index κ. r → 0 (7. . 4π|r| (7.29) Only the ﬁrst term is admissable. accordingly. .27) ∂r − r2 and.26) 2m 2 In case E < 0.31) to the complete wave which. 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).D. (r) ∼ A r or vκ.

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).

(7. 7. φ) for non-vanishing m has a non-trivial φ-dependence as described by (5. can be written exp(ikr cos θ). the wave function does not depend on φ.99) +1 d cos θ eikr cos θ P (cos θ) = b j (kr) −1 2 . one given o by (7. −1 .98) by ∞ eik3 x3 = =0 b j (kr) P (cos θ) .s. For example.44.e. Since the spherical harmonics Y 0 (θ.102) −1 d −1 +1 −1 dx −1 (x2 − 1) .s.50). This holds for −1 and one can conclude +1 consecutive integrations by part d izx e dx (7. z = kr.h. in case of a free particle moving along the x3 -axis one expands eik3 x3 = .178) are given in terms of Legendre polynomials P (cos θ) one can replace the expansion in (7. (7. One can expand the former solution in terms of solutions (7.98) does not involve any non-vanishing m-values since Y m (θ.50). 2 +1 (7.103) and. One can show d −1 2 (x − 1) ∼ (x2 − 1) × polynomial in x dx −1 (7. φ). The orthogonality properties (5. and using the Rodrigues formula for Legendre polynomials (5. i.150) one obtains +1 +1 1 ∂ dx eizx P (x) = dx eizx (x2 − 1) .43) has two solutions.h.98) The l. φ) . In this case the expansion on the r. namely.179) yield from (7. hence.100) Deﬁning x = cos θ.104) dx eizx P (x) −1 = (−1) 2 ! (iz) = 2 ! +1 dx (x2 − 1) −1 +1 dx (1 − x2 ) eizx .194 Generating Function of Spherical Bessel Functions Spherically Symmetric Potentials The stationary Schr¨dinger equation of free particles (7.101) 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. of (7.106).. the surface term ∼ [· · ·]+1 vanishes. according to (5.45) and one given by (7.99) We want to determine the expansion coeﬃcients b .m a mj (kr) Y m (θ. (7.

109) results in the integral representation of j (z) j (z) = (z) 2 +1 ! +1 dx (1 − x2 ) eizx .111) Employing (7.106) yields ﬁnally b = i (2 + 1) or. after insertion into (7.105) This expression allows one to determine the expansion coeﬃcients b .s.96) one can express this.108) where the last factor is equal to unity. Comparision with the l.114) . −1 (7. −1 (7. 1 (7.107) 1 2 (2 )! 1 · 3 · 5 · · · (2 + 1) 2 + 1 1 · 2 · 3 · 4 · · · (2 − 1) · 2 (7. Integral Representation of Bessel Functions Combining (7.f.h.110) One refers to the l.117) one can write the r. z i or i (2 + 1) z (2 )! 2 2 2 +1 2 ! [1 · 3 · 5 · · · (2 − 1)] 1 (7.s. in particular. The identity (7.h. ∞ (7.s. (7.69)] b z 2 (iz) = 1 · 3 · 5 · · · (2 + 1) 2 + 1 2 ! +1 dx (1 − x2 ) . (7. 1 (7.h. for the leading power x [c. using ν = Jν (z) = √ 1 πΓ(ν + 1 ) 2 z 2 ν + 1. of (7. 2 +1 −1 dx (1 − x2 )ν− 2 eizx .2: Free Particle Comparision with (7.109) eikr cos θ = =0 i (2 + 1) j (kr) P (cos θ) .100) gives b j (kr) 2 (iz) = 2 +1 2 ! +1 195 dx (1 − x2 ) eizx . as the generating function of the spherical Bessel functions.112) We want to consider the expression Gν (z) = aν z ν fν (z) where we deﬁne fν (z) = C (7. −1 (7.105) and (7.7.113) dt (1 − t2 )ν− 2 eizt .99).106) Employing (5.105) must hold for all powers of z.

t2 of the integration paths C. it holds for the derivatives of Gν (z) Gν (z) G (z) = = ν aν z ν fν (z) + aν z ν fν (z) z 2ν ν(ν − 1) aν z ν fν (z) + aν z ν fν (z) + aν z ν fν (z) 2 z z (7.113) satisﬁes (7.116) fν (z) = − or fν (z) = − From this we can conclude fν (z) + 2ν + 1 fν (z) + fν (z) = 0 . 1 (7.122) (7. We consider now the functions u(j) (z) = √ 1 πΓ(ν + 1 ) 2 z 2 ν Cj dx (1 − x2 )ν− 2 eizx .4: Prove (7.121). Gν (z). 4.117) dt (1 − t2 )ν− 2 t2 eizt 1 (7.93) as long as the integration path in (7.117).93) for arbitrary ν.116) In case that the endpoints of the integration path C satisfy (1 − t2 )ν+ 2 eizt one can write (7. 7. t2 . z (7.93). (7.123) Insertion of these identities into Bessel’s equation leads to a diﬀerential equation for fν (z) which is identical to (7.114.119) We note that equations (7.115) Integration by part yields fν (z) = − i 2ν + 1 (1 − t2 )ν+ 2 eizt 1 t2 t1 − z 2ν + 1 dt (1 − t2 )ν+ 2 eizt . 2. 3. We can now demonstrate that Gν (z) deﬁned in (7. 1 j = 1. Exercise 7.118) z 2ν + 1 fν (z) + fν (z) .120) z 2ν + 1 dt (1 − t2 )ν− 2 eizt + C C 1 1 t2 t1 = 0 (7.113) obeys the Bessel equation (7.124) .121) (7. C 1 (7. for properly chosen endpoints t1 . To prove this we note fν (z) = i C dt (1 − t2 )ν− 2 t eizt .2.120) such that we can conclude that Gν (z) for proper integration paths is a solution of (7. (7.120) imply also the property (2ν + 1) fν (z) + z fν+1 (z) = 0 . In fact. obeys Bessel’s equation (7.196 Spherically Symmetric Potentials Here C is an integration path in the complex plane with endpoints t1 .

112) shows u(1) (z) = Jν (z). we introduce the so-called Hankel functions (1) Hν (z) = −2 u(2) (z) . dt = i ds. (7.124) is analytical in the part of the complex plane surrounded by the path C we conclude 4 u(j) (z) ≡ 0 j=1 (7. Comparision with (7. one can state Jν (z) = − u(2) (z) + u(4) (z) . u(3) (z) ≡ 0. one can derive (2) Hν (z) = e −i(z−πν/2−π/4) ∞ 0 √ ds [s(1 − is/2)]ν− 2 e−zs .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 . 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 . Since the integrand in (7. 1 (7. using t = 1 + is.128) C = C1 ∪ C2 ∪ C3 ∪ C4 is a closed path.131) According to (7. therefore.125) 0 ≤ s < ∞ (7.7.124) vanishes along the whole path C3 and. for Rez > 0 and ν ∈ R obey (7. (7.117) and. 1 (7.135) (7.126) −1 ≤ s ≤ 1 (7. Following convention. (2) Hν (z) = −2 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. hence. Accordingly.93).130) holds Jν (z) = (1) 1 2 1 (1) (2) Hν (z) + Hν (z) . u(2) (z) and u(2) (z) are both solutions of Bessel’s equation (7.130) We note that the endpoints of the integration paths C2 and C4 . .132) For Hν (z) one derives.127) 0 ≤ s < ∞ (7. the integral expression (1) Hν (z) 1 1 1 1 (7.134) and (7. (7.129) The integrand in (7.134) Similarly.135) are known as the Poisson integrals of the Bessel functions.

(7.141) is o W (z) = − In case of g(z) = H (1) (z) +1 2 c . z (7. h(z) to be linearly independent.2) (z) +1 2 in terms of z −1 such that (z) = e±i[z − ( ∞ +1) π ] 2 2z z ( 1 ) f 2 (z) π ! (7.93) only as ν 2 . h) must be a non-vanishing function.136) (1) 2 (z) = 0 ds [s(s ± is/2)] 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 (7.93) holds the identity 1 W = − W z (7. converge fast for |z| → ∞. πz (7.135) which read for ν = + 1 2 H where f The binomial formula yields f (1) 2 (1) 2 1 +2 (1.138) The formula for the Laplace transform of sn leads to f and.140) Bessel Functions with Negative Index Since ν enters the Bessel equation (7.137) (z) = r=0 r ± i 2 r 0 ∞ ds s +r e−zs . the Wronskian W (g.141) (1) (1) the derivation of which follows the derivation on page 192 for the Wronskian of the radial Schr¨dinger equation. For such solutions g(z).. For the Wronskian connected with the Bessel equation (7. The general solution of (7. One obtains W (z) = − 4i . hence. i. We employ for this purpose the Poisson integrals (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.140). As a second order diﬀerential equation the Bessel equation has two linearly independent solutions.139) (z) = 2 ±i[z − ( e πz +1) π ] 2 r=0 ( + r)! r!( − r)! i ± 2z r (7.143) . Hν (z) as well as H−ν (z) are solutions of this equation. we obtain H (1) 2 1 +2 (1) 2 (z) = r=0 ( + r)! ! r!( − r)! ± i 2 r 1 z +r+1 (7.134) and (7.e. (7.

one can show H−( Spherical Hankel Functions In analogy to equations (7.140). (7.144) The expansion coeﬃcients A. 7. From (7. We conclude 2 H−( Simarly.2) H + 1 (z) .149) We want to establish also the relationship between h (z) and the irregular spherical Bessel function n (z) deﬁned in (7. According to (7. 7. 7. H (1.2) (z) +1 2 199 are.146) (z) = −i (−1) H (2) 1 (z) +2 . (7.2: Free Particle i. 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| → ∞ .e.90.152) (7.147) this can be written n (z) = 1 2i h (1) (z) − h (2) (z) . 7.96.149. B can be obtained from the asymptotic expansion (7. (7.97) one deﬁnes the spherical Hankel functions h (1.. 2z 2 (1.132. 7. 2 (1.7. (7. H−ν (z) = A H (1) (1) (z) +1 2 One can then expand + BH (2) (z) +1 2 . (7.132) follows n (z) = 1 2 π (−1) 2z +1 H−( (1) 1 +2) (z) + H−( (2) +1) 2 (z) . 7.146. linearly independent.153) .97.148) holds for the o regular spherical Bessel function j (z) = 1 (1) (2) [ h (z) + h (z) ] .145) This equation can hold only for B = 0 and A = exp[i( + 1 )π].151) are equivalent to h h (1) (2) (z) (z) = = j (z) + i n (z) j (z) − i n (z) . in fact.97).148) Following the arguments provided above (see page 193) the functions h (z) are solutions of the radial Schr¨dinger equation of free particles (7.151) Equations (7.2) (7. (7.2) (7.53).2) (2) +1) 2 (1) +1) 2 (z) = i (−1) H (1) (z) +1 2 .147) (z) = π (1.150) According to (7. (7.

it follows z ∈ R : Using p (z) = Re r=0 j (z) = Re[h (1) (z)] .163) (7. are complex conjugates.162) The leading terms in these expansions.140) and (7.159) (7. 7. From (7.140. Hence.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). 2z 2 z sin(z − π/2) z cos(z − π/2) − . = −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) (z). at large |z|. z (7.154) The leading term in this expansion. j (z) (z) = ( i) z +1 e±iz r=0 ( + r)! r!( − r)! ± i 2z r (7. 2 2z z (7. (1) (7. as given by (7.148) one obtains readily h (1) 2 (1.161) (7. at large |z|.160) Employing cos[z − ( + 1) π ] .164) . is h (1) 2 (z) = ( i) z +1 e±iz . ne ll(z) = Im[h (1) (z)] .148). = sin(z − π/2) and sin[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 (7.154) 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) .158) one can derive then from (7. are j (z) n (z) = = (7. (7.156) ( + r)! r!( − r)! ( + r)! r!( − r)! i 2z r [ /2] = r=0 ( + 2r)! (2r)!( − 2r)! −1 4z 2 r (7.155) (z) and To determine j (z) and n (z) we note that for z ∈ R the spherical Hankel functions h (2) h (z).

2) g (z) − g (z) z 2 +1 g (z) − g z (7. j (z) and n (z).124. One can combine (7.171–7.4 dx (1 − x2 )ν− 2 eizx .168) (7.149. 2 j0 (z) n0 (z) j1 (z) n1 (z) j2 (z) n2 (z) = = = = = = sin z z cos z − z sin z cos z − z2 z cos z sin z − 2 − z z 3 1 3 − sin z − 2 cos z 3 z z z 3 1 3 − 3 + cos z − 2 sin z z z z (7.174) and from (7. 7.2: Free Particle Expressions for the Spherical bessel Functions j (z) and n (z) 201 The identities (7.172) where g (z) is either of the functions h obtain the recursion relationship g g (z).167) (7.131.173) A (z) = z ( +1) z2 +2 z (7. 7.2) (z) = −2 π 1 √ 2z πΓ(ν + 1 ) 2 z 2 ν C2.162) allow one to provide explicit expressions for j (z) and n (z).171) holds for g (z) = h (z) we employ (7.174) We want to prove these relationships.169) (7. 1. 7. 1 (7. 7.170) Recursion Formulas of Spherical Bessel Functions The spherical Bessel functions obey the recursion relationships g g +1 (z) +1 (z) = = (1.175) Using Γ( + 1) = !.7.172) to +1 (z) +1 (z) = A (z) 1− g (z) g (z) −1 (7. For this purpose we need to demonstrate only that the (1.161.2) To demonstrate that (7. (7. 7. (1.171) −1 (z) (7. deﬁning a = −1 and employing fν (z) as deﬁned in (7.171.151) follows then that the relatiosnhips hold also for g (z) = j (z) and g (z) = n (z).114) we can write g (z) = a z f 2 ! + 1 (z) 2 .2) relationships hold for g (z) = h (z).148) and express g (z) = h (1. One obtains for = 0.176) .165) (7. From the linearity of the relationships (7.166) (7.

together with (7.181) Using this identity to replace the second derivative in (7.2. and (b) n1 (z).6: Employ the recursion relationship (7. f yields.180) yields g +1 (z) = g (z) − ( + 2) g (z) + z2 +2 g (z) . 7.171).174) we start from (7. in fact.179) leads to (7.. . The ﬁrst component of (7.176) g (z) = z g (z) − g +1 (z) +1 2 Spherically Symmetric Potentials z g (z) + a z f 2 ! +1 2 (z) .173.172).174) to determine (a) j1 (z). j2 (z) from j0 (z).177) (z) = − z f 2 +2 + 3 (z) 2 . (7. 7.173. i.173) is equivalent to (7.180) Since g (z) is a solution of the radial Schr¨dinger equation (7.182).5: Provide a detailed derivation of (7.e.121).171). z (7.173). In order to prove (7. is equivalent to (7.171). n0 (z) using (7.2.179) g (z) = − z2 g (z) + g (z) − g +1 (z) z . z2 (7. Exercise 7. (7. (7. Exercise 7. j0 (z) using (7. n2 (z) from n0 (z). The second component of (7.166).182) Replacing all ﬁrst derivatives employing (7.179) from which follows (7.178) (7.58) it holds o 2 g (z) = − g (z) + z ( + 1) g (z) − g (z) . To prove (7.202 The derivative of this expression is g (z) = Employing (7.172) we diﬀerentiate (7.165).172).

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

2) In addition.1) (8. The following substitutions. by J. D. t)] and. the relationship between ﬁelds and potentials is not unique. t) Interaction of Radiation with Matter = 4 π ρ(r. t) is a scalar function. t) − ∂t E(r. t) .204 Maxwell equations1 · E(r. called gauge transformations. t] (8. (8. From this follows E(r. in turn. 1975) for a discussion of these and other conventional units.1) the term qδ(r − ro (t)) and to the current density ˙ J(r. accordingly. t) = − V (r. The reader is referred to the well-known textbook ”Classical Electrodynamics”. t] + v × B[ro (t). called the vector potential. (8. We assume so-called Gaussian units. t) − ∂t χ(r. t) which is solved by E(r. obeys the equation of motion d p = q dt E[ro (t). called the scalar potential. t) = − V (r.11) V (r.4). In the non-relativistic limit holds p ≈ mr and (8. t) + v × B(r. t) where V (r. Jackson (John Wiley & Sons. t) + ∂t A(r. t) (8.6) for some vector-valued function A(r. t). t) + ∂t A(r.4) Lorentz Force A classical particle with charge q moving in the electromagnetic ﬁeld experiences the so-called Lorentz force q[E(r.10) (8. alter the potentials.9) (8. t) −→ V (r.5) above agrees with the equation of motion as given in (1. 2nd Edition. t) · B(r. . t) = 0 = 0. Equation (8. the two homogeneous Maxwell equations hold × E(r. t) + ∂t B(r. New York.25).5) where p is the momentum of the particle and ro (t) it’s position at time t. t) (8. but leave the ﬁelds unaltered: A(r.2) the term q r˙o δ(r − ro (t)). t) and A(r. The particle.8) = 0 (8. t) through the scalar and vector potentials V (r. (8. t) in (8. t) . t) . t) = × A(r. t) −→ 1 A(r. t) + χ(r.3) reads then × E(r. t) − ∂t A(r. contributes to the charge density ρ(r. t). t) × B(r. t) = 4 π J(r. t) in (8. t) and B(r. As is well known. Scalar and Vector Potential Setting B(r. solves implicitly (8.7) Gauge Transformations We have expressed now the electric and magnetic ﬁelds E(r.3) (8.

20) .19) couples the vector potential A(r. t) = − 4 π J(r.12) leads us to 2 2 A(r.12) The corresponding gauge is referred to as the Coulomb gauge. t) .17) A(r.16) and the vector potential are uncoupled. One would prefer a description in which the Coulomb potential (8. t) . (8.18) together with condition (8.8. in fact. t) = ∂t = 4 π ρ(r. t) − 2 × A(r.2). t) and V (r. the second inhomogeneous Maxwell equation. a name which is due to the form of the resulting scalar potential V (r. t) . t) + ∂t V (r. Using the expressions (8. t) − ∂t A(r. t) − ∂t V (r. 4π (8. t) and the current due to moving net charges. t) in (8. t) = − 4 π ρ(r. t) (8. Such description can. (8.9) into (8. and a component due to the remaining currents. t) |r − r | (8.13) · A(r. t) = Ω∞ for r ∈ ∂Ω∞ (8. and the former the Coulomb interactions in the unperturbed bound particle system.9) for the ﬁelds results in × The identity × × A(r. t) .14) In case of the boundary condition V (r. the gauge freedom allows us to impose on the vector potential A(r. t) = 0 the solution is given by the Coulomb integral V (r. The vector potential A(r. t) − ∂t A(r. For this purpose we examine the oﬀending term ∂t V (r. (8.16) This is the potential commonly employed in quantum mechanical calculations for the description of Coulomb interactions between charged particles. t) Using · ∂t A(r. t) can be obtained employing (8. this potential results from inserting (8. (8. (8. t = 4 π J(r. t) the condition · A(r. t) . be achieved. t) = 1 ∂t V (r. equation (8. In fact. In fact.6) and (8. t) together with (8. such that the latter describes the electromagnetic radiation.19) Unfortunately. t) + ∂t A(r.19) and deﬁne J (r. t) = · A(r.12) yields then the Poisson equation 2 V (r. t) = 0 . t). 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.15) d3 r ρ(r .1) · − V (r.

we want to assume that no ring-type currents. t) = or = 2 Interaction of Radiation with Matter (8. = − ∂t A(r. yet a current.21) ∂t and the Poisson equation (8. t) = 0).. (8.e. 8. t) These ﬁelds are obviously divergence -free (e.26) and (8. in a ring-shaped antenna which exhibits no net charge. t) = 0 and from this follows · Jt (r. t) − ∂t A(r.26) turns into the well-known wave equation 2 2 A(r. using ∂t · J (r. (8. t) as the current due to the time-dependence of the charge distribution ρ(r.12) and the Coulomb potential (8.25) Because of properties (8. V (r. t) = − 4 π Jt (r. hence. t) = 0 (8. hence. t) . t) = 0 . the name longitudinal electric ﬁeld.27) which is obviously curl-free ( × E (r. the name transverse ﬁelds. t) .16) yield ﬁnally the electric and magnetic ﬁelds. (8. exist in the space considered. t).2 Planar Electromagnetic Waves The current density Jt describes ring-type currents in the space under consideration. t) + ∂t ρ(r.28) (8. t) = 0 . t) − ∂t A(r. t) holds. The vector potential determined through (8. For J also holds the continuity equation · J(r. (8. for example.21) and (8. respectively. t) (8.14).23) This continuity equation identiﬁes J (r.22) 1 ∂t 4π · J (r.g.30) . t) Bt (r.25) one refers to J and Jt as the longitudinal and the transverse currents. t) = − ∂t ρ(r. V (r. t) = − V (r. t) be the total current of the system under investigation and let Jt = J − J . Presently. A(r. The deﬁnitions of J and Jt applied to (8. Let J(r. For the divergence of J (r. In this case (8.29) · Et (r.26) This equation does not couple anymore scalar and vector potentials. i.206 For the curl of J holds × J (r. no divergence-free currents. t) = 0). t) contributes solely an electric ﬁeld component E (r. t) contributes an electrical ﬁeld component as well as the total magnetic ﬁeld Et (r.. t) = × A(r.24) (8.19) yield 2 2 A(r. such current densities exist. t) = 0 . t) + ∂t ρ(r. t) (8.

A complete set of solutions is given by the so-called plane waves A(r.28. ˆ (8.35). t) = ± ω2 ˆ |Ao |2 k sin2 ( k · r − ωt ) 4π (8. t) = Ao u exp i(k · r ˆ where the dispersion relationship |k| = ω (8. at each point r and moment t.12) yields u·k = 0.32) holds. 8. hence. t) = Re Et (r.33) ωt) (8. t) . t) . In applying the potential and ﬁelds to physical observables and processes we will only employ the real parts.29). the constant k is referred to as the wave vector.34. are orthogonal to each other and are both orthogonal to the wave vector k.38. is orthogonal to k. The Coulomb gauge condition (8. t) = ± ω2 ˆ |Ao |2 k . 8.31) and the resulting ﬁelds (8.37) ˆ ˆ where k is the unit vector k = k/|k|.31).2: Planar Electromagnetic Waves 207 which describes electromagnetic ﬁelds in vacuum.34.35) one obtains S(r. Et (r. 8. accordingly.31. t) in (8. see also the comment below Eqs.35) are complex-valued quantities. t) × Re B(r. (8. t) = ±i ω A(r.34. This ﬂux is given by 1 S(r.34) (8.36) 4π Using the identity a × (b × c) = b (a · c) − c (a · b) and (8.31) u is a unit vector (|ˆ| = 1) which.38) ˆ In this expression for the energy ﬂux one can interprete k as the propagation velocity (note c = 1) and.32. Here the “-” sign corresponds to so-called incoming waves and the “+” sign to outgoing waves2 . 8. ˆ We want to characterize now the radiation ﬁeld connected with the plane wave solutions (8. are Et (r. 8.158). obviously. Note that in the units chosen the velocity of light is c = 1. Obviously. The latter vector describes the direction of propagation of the energy ﬂux connected with the plane wave electromagnetic radiation. 8. there exist two linearly ˆ u independent orientations for u corresponding to two independent planes of polarization. t) = i k × A(r. Time average over one period 2π/ω yields S(r.39) 8π The deﬁnition incoming waves and outgoing waves is rationalized below in the discussion following Eq. t) Bt (r. ω2 = |Ao |2 (8.8.39). (8. The corresponding electric and magnetic ﬁelds. 8π (8. t) and Bt (r. 8.35) The vector potential in (8. according to (8. (8. (8. 2 .

44) .43) Here the ﬁelds are deﬁned through Eqs. hence.31. (8. deﬁned below Eqs. t) = H Ψ(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.39) and (8.29) together with the potentials (8. |k| = ω .27. We assume that the energy content of the ﬁelds is not altered signiﬁcantly in the processes described and. Note that E (r.31). 8. 8. A ‘poor man’s’ quantization of the ﬁeld is possible at this point by expressing the energy density (8.40) allows one to express then the ﬁeld ˆ amplitudes 8πNω Ao = . These photons each carry the energy ω. (8. t) = 8πNω u exp i(k · r − ωt) ˆ ωV . (8. a ˆ speciﬁc k and a speciﬁc u. t) are complex leading to the diﬀerence of a factor 2 in the energy densities of the lontitudinal and transverse components of the ﬁelds. t).31). We are left with a classical Hamiltonian function which has an obvious quantum mechanical analogue ˆ H = ˆ p − q A(r. 8. The sign in (8. ˆ (8. respectively. we will neglect the respective terms in the Hamiltonian (8. u · k = 0 .38) implies that for incoming waves.8.28.32). The integrals express the integration over the energy density of the ﬁelds. Bt (r. (8.42) 8. t) . t) 2m + 1 8π Ω∞ 2 H = + qV (r) 1 16π d3 r |Et |2 + |Bt |2 Ω∞ d3 r E 2 + . t) 2m 2 + qV (r) . i (8.31) allows one ﬁnally to state for the planar wave vector potential A(r. A correct description of the electromagnetic ﬁeld requires that the ﬁeld be quantized.40) = V It should be pointed out that Nω represents the number of photons for a speciﬁc frequency ω.39) through the density of photons connected with the planar waves (8.43).208 Interaction of Radiation with Matter as the energy density. The wave function Ψ(r. If we consider a volume V with a number of photons Nω the energy density is obviously Nω ω . t). is p − q A(r. (8. whereas in the case of outgoing waves the energy is transported in the direction of k. Comparision of (8. (8.3 Hamilton Operator The classical Hamiltonian for a particle of charge q in a scalar and vector potential V (r) and A(r. t) is real 1 and that Et (r. the energy of the plane wave is transported in the direction of −k.16.41) ωV Inserting this into (8.45) .

t)/ Ψ(r.45) leads to the Schr¨dinger equation o 2 ˆ p − q A − q(( χ)) + qV − q((∂t χ)) Ψ(r. t) ˆ p eiqχ(r..t)/ .46). t) = eiqχ(r.10. hence. t) ˆ = eiqχ(r. (8.49) (8. t) are necessary to compensate terms which arise through the phase factor.t)/ Ψ(r.11) to (8. Obviously. t) . (8. It should be noted.g. accordingly. t) = + qV Ψ(r. not to phenomenological interactions like the molecular van der Waals interaction.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.. (8. t) and A(r. Applying the gauge transformations (8. t) + q r × B(r.50) The equivalence of (8. t) .t)/ constitutes a unitary transformation of a scalar quantity.t)/ [ i ∂t − q((∂t χ)) ] Ψ(r.11) aﬀect the quantum mechanical description. e. have no eﬀect on the motion of the particle described by (8. t) = + qV eiqχ(r.11) of the potentials is equivalent to multiplying the wave function Ψ(r. 8. does not change expectation values which contain the probability densities3 .e.48) implies that the gauge transformation (8.47) i ∂t Ψ(r. t) enter whereas in the classical equations of motion ˙ m¨ = q E(r.51) 2m the potentials A(r. 8.48) 2m For this purpose one notes i ∂t eiqχ(r.46) the ﬁelds enter.47. such phase factor does not change the probability density |Ψ(r. One can show that (8. t) .47) is equivalent to 2 ˆ p − qA i ∂t eiqχ(r. 8.8. t) r (8.t)/ p + q(( χ)) Ψ(r. The idea just stated can be generalized by noting that multiplication by a phase factor eiqχ(r.3: Hamilton Operator 209 Gauge Transformations It is interesting to note that in the quantum mechanical description of a charged particle the potentials V (r. t) by a local and time-dependent phase factor eiqχ(r.t)/ Ψ(r. hence. in the Schr¨dinger equation o 2 ˆ p − qA i ∂t Ψ(r. by the group 3 The eﬀect on other expectation values is not discussed here. t) (8. however. Elementary constituents of matter which are governed by other symmetry groups. In the classical case such question is mute since the gauge transformations do not alter the ﬁelds and. 8. that this principle applies only to fundamental interactions.10. i. .t)/ should not aﬀect a system and that.44. t) = 2m where ((· · ·)) denotes derivatives in (( χ)) and ((∂t χ)) which are conﬁned to the function χ(r. t)|2 and. an element of the group U(1). t) inside the double brackets. t) and V (r. This leads to the question in how far the gauge transformations (8.10. 8.

Using Cartesian coordinates this yields 2 H = − 2me 2 2 2 ∂1 + ∂2 + ∂3 + 2 eBo e2 Bo 2 x1 ∂2 + x i me 2me 1 (8. t) ≡ 0. H = 2m m 2m (8. (8.57) satisﬁes · A = 0 and. the so-called Landau gauge ˆ associates the vector potential AL (r) = Bo x1 e2 ˆ (8.57) with a homogeneous magnetic ﬁeld Bo . x1 . 3.53) ˆ ˆ p·A + A·p + q2 2 A + qV 2m (8. (8. likewise can demand the existence of ﬁelds which compensate local transformations described by eiσ·χ(r. (8. aﬀects the complexity of the mathematical derivation of the wave functions. holds (8.55) 8. x3 ) = exp(ik2 x2 + ik3 x3 ) φE (x1 ) . therefore. · A = 0 [cf.56) due to the gauge freedom. e. The choice of a vector potential aﬀects the form of the wave functions describing the eigenstates and. k3 . the generators of SU(2).45) with Hamiltonian (8. can be described by various vector potentials. Accordingly. one can employ the Hamiltonian (8. The resulting ﬁelds are called Yang-Mills ﬁelds. k2 . For this purpose we use the wave function in the form Ψ(E.. consequently.12)]. (8. thereby.55). 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. The vector potential (8.44) can be expanded ˆ2 p q H = − 2m 2m For any function f (r) holds ˆ ˆ p · A − A · p f (r) = i A· f + f ·A − A· f = i f ·A.59) .54) ˆ2 p q ˆ q2 2 − p·A + A + qV . 2.g.56). t) = Bo . We want to describe the stationary states corresponding to the Hamiltonian (8. The stationary homogeneous magnetic ﬁeld B(r.58) where ∂j = (∂/∂xj ). for Bo = Bo e3 in (8.55). t). x2 . In case of a homogeneous potential pointing in the x3 -direction. The Hamiltonian (8.58).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 (8.210 Interaction of Radiation with Matter SU(2).t) where σ is the vector of Pauli matrices. j = 1.52) This expression vanishes in the present case since since ˆ ˆ p · Af = A · pf and.

66) Obviously.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 (8.64) is the classical Larmor frequency (c = 1). according to (8. However. 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 ) (8.65) where we replaced the parameter E by the integer n. is Ψ(n. (8. x1 .65).62) is that of a displaced (by x1o ) harmonic oscillator with o 2 k 2 /2m ) energies. Without aﬀecting the energy one can form wave packets in terms of the solutions (8. x2 .and x2 -coordinates. k3 .63) (8.67) . k2 .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 ) . like the corresponding gauge (8. k3 ) = ω(n + 2 k2 1 3 ) + . (8. The energies corresponding to these states are E(n. 2 (8. is not symmetric in the x1 . Unfortunately. 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 (8.62) x1o = and where ω = k2 eBo eBo me (8. Solution for Symmetric Gauge The solution obtained above has the advantage that the derivation is comparatively simple. the so-called symmetric gauge which expresses the homogeneous potential (8.56) through the vector potential A(r) = 1 Bo × r .59). the familiar harmonic oscillator quantum number. 2 2me (8.63) this induces a spread of the wave function in the x1 direction.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 ) .59). k2 .65) which localize the electrons. From this observation one can immediately conclude that the wave shifted (by e 3 function of the system. the states are degenerate in the quantum number k2 describing displacement along the x2 coordinate. the wave function (8. The stationary Schr¨dinger equation (8. according to (8. We want to employ .57). therefore.

73. 8. in the present case of constant Bo .73) where µclass is the magnetic moment connected with a current density j µclass = 1 2 r × j(r) dr (8.68) × u yields. For the vector potential (8. ˆ Identifying r × p with the angular momentum operator L.12) for the Coulomb gauge. (8.74).70) ˆ ˆ p · A f = − Bo · p × r f .212 Interaction of Radiation with Matter One can readily verify that this vector potential satisﬁes the condition (8. (8. Vmag = e 2me class = r me v e3 of the electron ˆ (8.76) e 2me class class · Bo . for any (8. The latter can be rewritten. (8. accordingly. namely. y-plane with constant velocity v on a ring of radius r.75) 2 The latter can be related to the angular momentum µclass = − and.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 .74) We consider a simple case to relate (8. Vmag = −µclass · Bo (8.67) one can write ˆ p·A = Using · (u × v) = − u · function f (r) ×v + v· 2i · Bo × r . the magnetic moment (8. namely. Accordingly. In this case the current density measures −e v oriented tangentially to the ring. ˆ (8. (8. using × (uf ) = −u × f +f ˆ ˆ Bo · p × r f = Bo · r × p f .71). The theory of classical electromagnetism predicts an analogue energy contribution .71) (8.69) × u and × r = 0. the Hamiltonian (8.72) to Hamiltonian (8.78) .77) Comparision with (8. an electron moving in the x.72) and (8.72) allows one to interpret µ = − e L 2me (8.74) is in the present case 1 µclass = − e r v e3 .

i. We will demonstrate in Sect. Accordingly.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 . for Bo = Bo e3 . we separate the variable x1 . x2 ) = E ψ(x1 . x2 . x2 . 2 .88) me ω xj . E = E − (8.87) .85) The Hamiltonian (8. x2 ) obey then ˆ Ho ψ(x1 .71) is ˆ H = 2 ˆ2 p e2 Bo + 2me 8me x2 + x2 + 1 2 eBo L3 .72) with an associated magnetic moment − g 2me S where g ≈ 2. (8.84) .67) yields a more symmetric solution which decays to zero along both the ±x1 .and the ±x2 -direction.79) To obtain the stationary states. x3 ) = E ΨE (x1 . For a magnetic ﬁeld (8. the Hamiltonian (8.4: Electron in Homogenous Magnetic Field 213 as the quantum mechanical magnetic moment operator for the electron (charge −e). 10 that the spin of the electron.e. The functions ψ(x1 .. 2me We have used here the expression for the angular momentum operator ˆ L3 = ( /i)(x1 ∂2 − x2 ∂1 ) . (8.e.82) 1 ω (x1 ∂2 − x2 ∂1 ) i 2me + 1 me ω 2 x2 + x2 1 2 2 2 k2 3 + (8. x2 ) . requires a Lorentz-invariant quantum mechanical description as provided in Sect. x3 ) = exp(ik3 x3 ) ψ(x1 . likewise. A derivation of his property and the value of g. x2 from x3 setting ΨE (x1 .8. we seek stationary states which are simultaneous eigenstates of the Hamiltonian of the two-dimensional isotropic harmonic oscillator 2 1 2 ˆ Hosc = − ∂ 2 + ∂2 + me ω 2 x2 + x2 (8.86) 1 2 2me 1 2 ˆ as well as of the angular momentum operator L3 .56) pointing in the x3 -direction the symmetric gauge (8. ˆ i. the solutions of ˆ H ΨE (x1 .83) (8. 2me (8. 10. x2 ) where ˆ Ho = − 2 2 2 ∂1 + ∂2 (8.83) describes two identical oscillators along the x1 -and x2 -directions which ˆ are coupled through the angular momentum operator L3 . the so-called gyromagnetic ratio of the electron. x3 ) . j = 1. To obtain these eigenstates we introduce the customary dimensionless variables of the harmonic oscillator Xj = (8. x2 . e gives rise to an energy contribution (8. described by the operator S. In this case.81) (8.80) (8.

x1 . 2 2 . . x2 ) . m .92) where Ψ(0. If this property is. x1 . 1. attempt to express eigenstates in terms of vibrational wave functions a† 1 j+m Ψ(j.92) represents a state with j + m quanta in the x1 -oscillator and j − m quanta in the x2 -oscillator. In order to cover all posible vibrational quantum numbers one needs to choose j. (8.214 Employing the creation and annihilation operators 1 a† = √ j 2 and the identity ˆ ω L3 = which can readily be proven. x2 ) = ( 2j + 1 ) Ψ (j. m .95) (j) ˆ We want to choose the coeﬃcients αmm such that (8.89) . such that 1 i a† a2 − a† a1 1 2 Ψ (j. x1 . (8. x1 . We.90) a† a2 − a† a1 1 2 . m .93) ˆ The states (8.e. m = −j. x1 . one obtains 1 ˆ 1 H = a† a1 + a† a2 + 1 + 1 1 2 ω i 1 i a† a2 − a† a1 1 2 Xj − ∂ ∂Xj 1 . x2 ) = a† 1 j−m (j + m)! (j − m)! Ψ(0. j = 1.94) is an eigenstate of the vibrational Hamiltonian. 2 (8. m .92) are not eigenstates of L3 . in fact. (8. x1 .. x2 ) (8. x2 ) . x1 .as well as for the x2 -oscillator. x1 . . x1 .94) is an eigenstate of Ho ˆ Ho Ψ (j. . (8. the total vibrational energy being ω(2j + 1).94) is also an eigenstate of L3 . (8. through a combination of states j Ψ (j. x1 . +j . .. x2 ) = m=−j αmm Ψ(j. (8. x2 ) (8. m. x1 . (j) (8. x2 ) = 2m Ψ (j. 1 2 since one phonon is annihilated and one created. x2 ) is the wave function for the state with zero vibrational quanta for the x1 .94) Since this state is a linear combination of states which all have vibrational energy (2j + 1) ω.e. −j + 1. x2 ) = ω 2j + 2m + 1 Ψ (j. m .91) We note that the operator a† a2 − a† a1 leaves the total number of vibrational quanta invariant. . however. .97) . i. 0. i. . 1 3 . . (8. m. Such eigenstates can be expressed. obeyed. it holds a† a1 + a† a2 + 1 1 1 2 Ψ (j. therefore. m . aj = √ 2 Interaction of Radiation with Matter Xj + ∂ ∂Xj . 0. (8.96) ˆ holds. x2 ) . m . m as follows: j = 0.

m . 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. cf. According to the derivation given there. b− + − notation in Sects..55). x2 ). x1 . a2 1 2 present notation ←→ b† . which corresponds there to the angular momentum in the x2 –direction.84) above. If we identify 2 a† . According to Sect.55) conﬁnes the particle to stationary bound states.102) We have identiﬁed. 5. x2 ) = Dmm ( . except for a factor 2 . 5. a transformation moving the x3 –axis into the x2 – 2 axis. π . We assume that the scalar potential V in (8.9.309) yields ﬁnally Ψ (j. m. the states are eigenstates of the operator [we use for the operator the notation of Sect. the second rotation a factor dmm ( π ).83) and conﬁrmed the eigenvalues stated in (8. 5. a1 .92) correspond to the eigenstates |Ψ(j.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 (8.99) with eigenvalue m. an example is the Coulomb potential V (r. b+ .98) then the states Ψ(j. 0) Ψ(j. x1 . i.e.288)] 1 ˆ J3 = 2 a† a1 − a† a2 1 2 (8. is identical to the operator L3 introduced in (8.11. m.101) 2 2 where Dmm ( π .5. x2 ) (8.(5. x1 . m. 5. 5.97). x2 ) deﬁned in (8. x1 . Eq. The external radiation ﬁeld is accounted for by the vector potential A(r.8.10.11 such transformation is provided through π π (j) Ψ (j. is in the notation of the present section 1 ˆ a† a2 − a† a1 .. m) in Sect. thus.11 (8. the eigenstates of (8.5.9. The ﬁrst rotation contributes a factor exp(−im π ).10. . m .5. (8. x2 ) . b† .10. the 2 2 latter representing the Wigner rotation matrix of Sect. 5. Other radiation ﬁelds can .5.10. The required rotation must transform the x3 –axis into the x2 –axis. This implies that ˆ we can obtain eigenstates of L3 by rotation of the states Ψ(j. x1 . m.e. 5. x1 . In the simplest case the radiation ﬁeld consists of a single planar electromagnetic wave described through the potential (8. Using the explicit form of the Wigner rotation matrix as given in (5. The connection with the present problem arises due to the fact that the operator J2 in Sect.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. t) introduced above. 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 .9. a† .100) J2 = 1 2 2i 1 ˆ i.11.31). (8. t) = 1/4πr conﬁning an electron with energy E < 0 to move in the well known orbitals of the hydrogen atom. 8.

42). t). using (8.103) (8. as given in (8. we assume n|m = δnm and for the identity ∞ (8.216 Interaction of Radiation with Matter be expanded through Fourier analysis in terms of such plane waves. t). The states |n are thought to form a complete. 2 .e. and that the ﬁrst term in (8. This result will allow us to neglect the larger than the second term. the term ∼ A second term in (8. is much 2 (r. this assumption can be waved as our results below will not depend on it. However. (8.106) where we adopted the Dirac notation for the states of the quantum system. i.105) for the case of radiationinduced transitions in atomic systems. one can estimate that the perturbation. i. i.. can be considered a weak perturbation.109) The reader should note that the estimates are very crude since we are establishing an order of magnitude estimate only.. i. . the term ∼ p · A(r. . .e.105). Here the so-called unperturbed system is governed by the Hamiltonian Ho with stationary states deﬁned through the eigenvalue problem Ho |n = n |n .. orthonormal basis. We ﬁrst note. (8. 1.105) as follows4 . Estimate of the Magnitude of VS We want to demonstrate now that the interaction VS (t).. is much smaller than the eigenvalue diﬀerences near ˆ typical atomic bound states. 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.e.105) in further calculations and to expand the wave function in terms of powers of VS (t) in a perturbation calculation.108) We assume for the sake of simplicity that the eigenstates of Ho can be labeled through integers.105) where A(r. We will see below that the perturbation resulting from a ‘pure’ plane wave radiation ﬁeld will serve us to describe also the perturbation resulting from a radiation ﬁeld made up of a superposition of many planar waves.107) 1 = 1 n=0 |n n| .41) e ˆ e p2 p·A ∼ 2me me me 2me 4 1 2 8πNω . In fact. t) − m 2m (8. n = 0. in this case. t) + A (r. ωV (8.e. For an electron charge q = −e and an electron mass m = me one can provide the estimate for the ﬁrst term of (8. we discount the possibility of a continuum of eigenstates.104) (8. t) is given by (8.

and me c2 ≈ A A 500 keV 2 ao ω ≈ 10−8 (8. one can neglect this term as long as the ﬁrst term gives non-vanishing contributions. u) p · u exp i(k · r − ωt) ˆ ˆ ˆ kV . altogether. replace the perturbation (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 .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. We want to estimate now the second term in (8.e. 2me (8.8.105) due to a radiation ﬁeld by VS = − q ˆ p · A(r. Consequently.115) (8.118) . We can. (8. e ˆ p · A ∼ 10 eV · 10−4 = 10−3 eV .5 ˚.42) it holds VS ≈ e m k. Assuming a single photon..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. Using again (8.e. one obtains for (8. Nω = 1. m (8. t) . me (8..114) Employing for the second factor the estimate as stated in (8. and as long as the photon densities Nω /V are small. Hence.117) In case that such perturbation acts on an electron and is due to superpositions of planar waves described through the vector potential (8.112) yields e2 A2 ∼ 10 eV · 10−8 = 10−7 eV .105).112) π λ me c2 and with e2 /ao ≈ 27 eV. i. i. hence. with ao ≈ 0. (8. λ = 2πc/λ. a volume V = λ3 where λ is the wave length corresponding to a plane wave with frequency ω.105) for radiation ﬁelds can be considered a small perturbation.116) This term is obviously much smaller than the ﬁrst term.111) For ω = 3 eV and a corresponding λ = 4000 ˚ one obtains.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.ˆ u 4πNk α(k. the ﬁrst term in (8.110) where ao is the Bohr radius.

The sum runs over all possible k vectors and might actually be an integral.122) where |ΨD (to ) = |0 . The probability to ﬁnd the particle in the state |n at time t is then p0→n (t) = | n|ΨS (t) |2 . Perturbation Expansion The generic situation we attempt to describe entails a particle at time t = to in a state |0 and a radiation ﬁeld beginning to act at t = to on the particle promoting it into some of the other states |n . For this purpose one needs to determine the state |ΨS (t) of the particle..120) (8.108) as the eigenstates of the unperturbed Hamiltonian Ho . A factor α(k.42) through k = |k| = ω. One seeks to predict the probability to observe the particle in one of the states |n . (8. the sum over u involves the two possible polarizations of planar ˆ electromagnetic waves.125) (8. describes the eﬀect of a radiation ﬁeld on an electron system and which will be assumed below to describe radiative transitions. Using i i ∂t exp − Ho (t − to ) and (8. n = 1.127) i ∂t |ΨD (t) = VD (t) |ΨD (t) .121) (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) . u) has been added to describe eliptically or circularly polarized ˆ waves. |Ψ(to ) = |0 (8. 2.218 Interaction of Radiation with Matter where we have replaced ω in (8. |n are deﬁned in (8. .126) Ho (t − to ) yields ﬁnally . This state obeys the Schr¨dinger equation o i ∂t |ΨS (t) = [ Ho + VS (t) ] |ΨS (t) subject to the initial condition |ΨS (to ) = |0 .123) i = Ho exp − Ho (t − to ) (8. . n = 0 at some later time t ≥ to .118) is the form of the perturbation which.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. Multiplying the latter equation by the operator exp i (8.106–8. . under ordinary circumstances. The states |0 . Equation (8.

128) We note that the transition probability (8. We will consider here only the two leading contributions to |ΨD (t) .137) are consistent with (8.131) p0→n (t) = | n |ΨD (t) |2 . consequently.129) n| and. using |exp[− i n (t n (8. (8.137) follows |ΨD (t) = |0 . Equations (8.130) = exp − i n (t − to ) − to )]| = 1.127) we assume the expansion ∞ |ΨD (t) = n=0 (n) |ΨD (t) (n) (8.121) expressed in terms of ΨD (t) is i p0→n (t) = | n|exp − Ho (t − to ) |ΨD (t) |2 .128). . 8.5: Time-Dependent Perturbation Theory where VD (t) = exp 219 i i Ho (t − to ) VS (t) exp − Ho (t − to ) . Employing this result one obtains for (8. . we deﬁne |ΨD (t) through the evolution equations i ∂t |ΨD (t) i ∂t |ΨD (t) i (2) ∂t |ΨD (t) (1) (0) = = = . 0 VD (t) |ΨD (t) (1) VD (t) |ΨD (t) (0) (8.139) .133. to (8.133–8. (8.137) One can readily verify that (8. From (8.134) (8.134. 8. . .138) 1 i t dt VD (t ) |0 .127.137) can be solved recursively. (8.137) |ΨD (t) = (1) (0) (8.136) together with the initial conditions |ψD (to ) = |0 0 for n = 0 for n = 1. Accord(n) ingly. In order to determine |ΨD (t) described through (8. = . . 8. Due to the Hermitean property of the Hamiltonian Ho holds n|Ho = i n|exp − Ho (t − to ) from which we conclude.132–8. .133) (8. 2.135) i ∂t |ΨD (t) (n) VD (t) |ΨD (n−1) (t) (8. .8. n| (8.132) where |ΨD (t) accounts for the contribution due to n-fold products of VD (t) to |ΨD (t) .

We want to apply now the perturbation expansion derived to such perturbations. t) = + A1 u1 exp i (k1 · r − ω1 t) ˆ A2 u2 exp i (k2 · r ˆ ω2 t) incoming wave incoming or outgoing wave (8. the direction of ˆ the propagation being determined by k = k/|k|. 8.e.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. . such that the combined radiation ﬁeld is decribed by the vector potential A(r. This factor will serve mainly the purpose of keeping in the following derivation all mathematical quantities properly behaved. λ → 0+ has been introduced which describes that at time to → −∞ the perturbation is turned on gradually.41). non-singular. note that uj . λ → 0+ . yields for (8.141) + . The coeﬃcients A1 .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). to → −∞ (8.6 Perturbations due to Electromagnetic Radiation We had identiﬁed in Eq..140) Altogether we have provided the formal expansion for the transition amplitude n|ΨD (t) = n|0 + ∞ 1 i 1 i t dt n|VD (t ) |0 to 2 t t (8. For the sake of including the eﬀect of superpositions of plane waves we will assume.142) combining an incoming and an incoming or outgoing wave. II III (8.142). The factor III in (8.34.. i. + m=0 dt to to dt n|VD (t )|m m|VD (t ) |0 8. is VS = ˆ ˆ V1 exp(−iω1 t) + V2 exp( iω2 t) eλt . according to (8. A2 are deﬁned through (8. 8. We will demonstrate below that the the sign of iωt determines if the energy of the planar wave is absorbed (“-” sign) or emitted (“+” sign) by ˆ the quantum system.118). A factor exp(λt). according to (8.137) |ΨD (t) = (2) Interaction of Radiation with Matter 1 i 2 t t dt to to dt VD (t ) VD (t ) |0 . however. deﬁnes the direction of the E-ﬁeld of the ˆ radiation.135. (8. in turn.144) r is the position of the electron and p = ( /i) is the momentum operator of the electron. (8. In (8.. that two planar waves simulataneously interact with an electronic system.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 .144) describes the propagation of the planar wave. The resulting perturbation on an electron system.220 This result.

145) n|ΨD (t) (1) m (t − to ) |m .143). According to (8.143) to the expansion (8.149) Using the deﬁnition of VD stated in (8. using (8. employing the perturbation (8. For the 1st order contribution to the transition amplitude n|ΨD (t) = (1) 1 i t dt n|VD (t ) |0 to (8.145) we obtain then.128) one obtains k|VD (t) | = k|VS (t)| exp i ( k − ) (8.150) and. (8.141).146) = 1 λ→0+ t→−∞ i lim lim × t dt exp to i ( n − o − i λ) t iω2 t × . 8.8. (8. (8.6: Perturbations due to Electromagnetic Radiation 1st Order Processes 221 We employ now the perturbation (8.147) ˆ ˆ n|V1 |0 e−iω1 t + n|V2 |0 e Carrying out the time integration and taking the limit limt→−∞ yields (1) n|ΨD (t) = λ→0+ lim e λt exp ˆ n|V1 |0 o − o − ω1 ) t + + ω1 − n + iλ ( n i i ˆ + n|V2 |0 2nd Order Processes exp o − o ω2 ) t ± ω2 − n + iλ ( n .130) and (for m = 0) i i exp − Ho (t − to ) |m = exp − for (8.140. 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. (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 .128). (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 .

131).. We also assume for the ﬁnal state n = 0 such that n|0 = 0 and p0→n (t) = | n|ΨD (t) |2 holds.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. respectively. the time derivative of p0→n (t).155) λ→0+ ( o We are actually interested in the transition rate. We assume ﬁrst that the ﬁrst order transi(1) tion amplitude n|ΨD (t) is non-zero. in which case one can expect that it is larger than the 2nd (2) order contribution n|ΨD (t) which we will neglect.148) and (8.154) yields p0→n (t) = lim e2λt + + 2 Re ˆ n|V1 |0 |2 ( o + ω1 − n )2 + (λ )2 ˆ n|V2 |0 |2 ( o ± ω1 − n )2 + (λ )2 ˆ ˆ n|V1 |0 0|V2 |n exp i (± ω2 − ω1 ) t + ω1 − n + iλ ) ( o ± ω2 − n − iλ ) (8. Using (8.153) (8. provide now the transition probability p0→n (t) according to Eq. For this rate holds .148) and ∗ |z1 + z2 |2 = |z1 |2 + |z2 |2 + 2Re(z1 z2 ) (1) (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 (8.152). i.e. (8.

Using lim x2 + 2 δ→0+ = π δ(x) (8.160) leads to ﬁnal states |n with energy n = o ± ω2 .8.144). describing either an incoming or an outgoing plane wave due to the vector potential A2 u2 exp i (k1 · r ˆ ω2 t) .158).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. is much shorter than could be resolved in any observation. The second contribution to (8.. The time average will be denoted by · · · t . the matrix elements can vary strongly for diﬀerent states |n of the system. ˆ (8. (8. In case the matrix elements are non-zero.158) ˆ + | n|V2 |0 |2 δ( − ω2 ) Obviously.31) and (8. in fact.e. i. The δ-functions appearing in this expression reﬂect energy conservation: the incoming plane wave contribution of (8. to cancel the third term in (8. any attempt to do so.143. 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.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+ (8.158) play an essential role for the transition rates of radiative transitions. The result of such average is.144). due to the vector potential A1 u1 exp i (k1 · r − ω1 t) .156) such that the 1st order contributions of the two planar waves of the perturbation simply add.142) ˆ ˆ The matrix elements n|V1 |0 and n|V2 |0 in (8. a property. the two terms apearing on the rhs. which is observed through the so-called spectral intensities of transitions |0 → |n .159) leads to ﬁnal states |n with energy n = o + ω1 . of this expression describe the individual eﬀects of the two planar wave contributions of the perturbation (8.156) ( o The period of electromagnetic radiation absorbed by electronic systems in atoms is of the order 10−17 s. however. one should average the rate over many periods of the radiation. Hence. . 8. The result supports our deﬁnition of incoming and outgoing waves in (8. For the resulting expression the limit limλ→0+ can be taken. due to the uncertainty relationship would introduce a considerable perturbation to the system. First.142–8.

k=1 j=k 2 2 2 2 ∗ zj zk (8.167) Taking now the limit limλ→0+ and using (8. similarly. (8.158).166) over many periods of the radiation yields exp[ i (±ω2 − ω1 )] t = 0 and.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.224 2nd Order Radiative Transitions Interaction of Radiation with Matter We now consider situations where the ﬁrst order transition amplitude in (8.162) × and. it holds p0→n (t) = | n|ΨD (t) |2 . in analogy to (8. as we did in the the case of 1st order transitions. We deﬁne ∞ z1 = m=0 ˆ ˆ n|V1 |m m|V1 |0 m − o − ω1 − i λ exp i × (8. ∞ − 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 × (8.152).161) To determine the transition rate we proceed again.. i.163) × (8.164) (8.153) vanishes such that the leading contribution to the transition probability p0→n (t) arises from the 2nd order amplitude (8.e.153–8.158)..168) absorption of 2 photons ω1 .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. 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 ) (8.157) yields. hence. (2) (8. e. i.. |z1 + z2 + z3 |2 t = |z1 |2 + |z2 |2 + |z3 |2 (8. Time average · · · t of expression (8. in Eqs.g.e. exp[ i (±ω2 − ω1 )].

e. this state is only virtually excited.e. the system is perturbed through absorption of a photon with energy ω1 from the initial state |0 into a state |m . In case of the ﬁrst term in (8. describe the absorption of two photons.7: One-Photon Emission and Absorption 2π ∞ m=0 225 2 + ˆ ˆ n|V2 |m m|V2 |0 ω2 − i λ m− o δ( m − o 2 ω2 ) absorption/emission of 2 photons ω2 + 2π ∞ m=0 ˆ ˆ n|V2 |m m|V1 |0 + m − o − ω1 − i λ + ˆ ˆ n|V2 |m m|V1 |0 ω2 − i λ m− o 2 δ( n − o − ω1 ω2 ) absorption of a photon ω1 and absorption/emission of a photon ω2 This transition rate is to be interpreted as follows. the third term in (8. describes processes which lead to ﬁnal states |n with energy n = o + 2 ω1 and. Similarly. i. m = o + ω1 ) and the evolution of state |m is described by a factor 1/( m − o − ω1 −i λ). Similarly. |n ← |m .168) describe the time sequence of the two photon absorption/ emission processes. it must hold n = o + 2 ω1 . accordingly. The factors | · · · |2 in (8.. (8.158) which accounts for such transitions. ﬁrst absorption of photon ω1 and then absorption/ emission of photon ω2 . through absorption of a photon.168) describe similar histories of the excitation process. The ﬁrst term. the third term describes processes in which a photon of energy ω1 is absorbed and a second photon of energy ω2 is absorbed/emitted. namely absorption/ emission ﬁrst of photon ω2 and then absorption of photon ω1 and.. each of energy ω1 . according to its δ-function factor. i. a second perturbation. promotes the system then to the state |n .169) pert. i. 8. i. which is stationary and energy is conserved. Most remarkably. |m ← |0 (8. each of energy ω2 . accordingly.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. describe the absorption/emission of two photons.e. The expression sums over all possible virtually occupied states |m and takes the absolute value of this sum. The remaining two contributions in (8. . vice versa.e.7 One-Photon Absorption and Emission in Atoms We ﬁnally can apply the results derived to describe transition processes which involve the absorption or emission of a single photon. interference between the contributions from all intermediate states |m can arise.168) describes two intermediate histories. the second term describes the processes leading to ﬁnal states |n with energy n = o ± 2 ω2 and..8.. there is no energy conservation necessary (in general. For this purpose we will employ the transition rate as given in Eq.

158). φ) = 2 − 1 2 1 −r/ao e Y00 (θ. ˆ (8.170) We will employ only the real part of this potential. . the wave functions are (n. in this range one can expand eik·r ≈ 1 + ik · r + . θ. (8. m denote the relevant quantum numbers) ψn=1.144) VS = where e 2πN ˆ ˆ p · u e±ik·r . i. θ. 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 . in case of a radiative transition from the 1s state of hydrogen to one of its three 2p states. The absorption rate.226 Absorption of a Plane Polarized Wave Interaction of Radiation with Matter We consider ﬁrst the case of absorption of a monochromatic. .143. φ) × × eik·r e−r/ao Y00 (θ. t) = 8πN ωV u exp ˆ ı (k · r − ωt) . the vector potential actually assumed is A(r.174) We seek to evaluate the matrix element ˆ M = n| p eik·r |0 ..175) The matrix element involves a spatial integral over the electronic wave functions associated with states |n and |0 . For example.m (r. =0. 8.179) . φ) (8. φ) = 6 r −r/2ao e Y1m (θ.176) ψn=2.143) will contribute to the absorption process.173) V1. λ → 0+ . to → −∞ (8. However.172) δ( n − o − ω) (8.e. (8. (8.171) The perturbation on an atomic electron system is then according to (8.177) √ M = 6 0 ∞ 1 2π ia4 o r2 dr −1 dcosθ 0 ∗ dφ r e−r/2ao Y1m (θ. t) = 2πN ωV u exp ˆ ı (k · r − ωt) + 2πN ωV u exp ˆ ı (−k · r + ωt) .178) These wave functions make signiﬁcant contributions to this integral only for r-values in the range r < 10 ao . plane polarized wave described through the complex vector potential A(r. . φ) a3 ao o 2p (8. according to (8. the second term can be discounted in case of absorption. and the integral is =1.m=0 (r. φ) a3 o 1s (8. (8.2 = m ωV Only the ﬁrst term of (8.

175). Ho ] |0 = i i n) n|r |0 . ] + [ r.180) ∆E2p−1s and ao = 0.182) one ﬁnds [ r.8. (8. This simpliﬁcation results from the identity m ˆ p = [ r.7: One-Photon Emission and Absorption 227 One can estimate that the absolute magnitude of the second term in (8.179) and other terms are never larger than 20π ao /λ. Using (8.186) δ( n − o − ω) (8.k=1 i m 3 pk ek = ˆ j.k=1 i ˆ p m (8.185) from which follows (8.183) (8. Transition Dipole Moment A further simpliﬁcation of the matrix element (8. V (r) ] 2me =0 = Using r = 3 ˆ j=1 xj ej 1 2me 3 pk [ r. Ho ] = i m 3 pk ej δjk = ˆ j.175) can then be ˆ achieved and the diﬀerential operator p = i replaced by by the simpler multiplicative operator r. Ho ] = ˆ2 p [ r.187) . We are now in a position to obtain an alternative expression for the matrix element (8. Using |k| = 2π/λ. the value of the wave length for the 1s → 2p transition 2π c λ = = 1216 ˚ A (8. Ho ] i where Ho is the Hamiltonian given by (8. V (r) = − .182). pj ] = i δkj one obtains ˆ [ r. is Ho = ˆ (p)2 e2 + V (r) .174) yields kabs = 4π 2 e2 N ω V u · n| ˆ |0 ˆ r 2 m( o − m n| [r. in case of the hydrogen atom.181) For the commutator in (8.529 ˚ one concludes that in the signiﬁcant integration range in (8.181) and (8. pk ] pk ˆ k=1 (8.182) (8. One refers to this approximation as the dipole approximation.182) one obtains M ≈ Insertion into (8. pk ] + ˆ ˆ k=1 1 2me 3 [ r. 2me r (8.184) and the commutation property [xk .104) and.178) holds eik·r ≈ A 1 1 + O( 50 ) such that one can approximate eik·r ≈ 1 .

x2 .191) (8.190) (8. k frame is described by the angles ϑ.188) ˆ Here dk is the integral over all orientations of k. Instead of adding the components of all possible k values we integrate over all k using the following rule +∞ =⇒ k u ˆ V −∞ k 2 dk (2π)3 ˆ dk u ˆ (8. We have demonstrated in our derivation of the rate of one-photon processes (8. in practical situations.192) and obtain ˆ dk u ˆ u · n| ˆ |0 ˆ r 2 2π 1 = |ρ|2 0 −1 dcosϑ (1 − cos2 ϑ) = 8π 3 (8. u2 and k ˆ ˆ coordinate system. The unit ˆ ˆ ˆ ˆ can be chosen to point along the x1 . ˆ In order to carry out the integral dk we note that u describes the possible polarizations of the ˆ ˆ and u. a very hot oven. will actually be replaced by a distribution function which reﬂects (1) the ﬁnite life time of the states |n . obtain the transition rate in the present case by adding the individual transition rates of all planar waves present in the oven. k ˆ As a result.35). hence. We can. The direction of this vector in the u1 .174) one can replace n − o by ω. such that the thermal radiation present supplies a continuum of frequencies. i. Absorption of Thermal Radiation We want to assume now that the hydrogen atom is placed in an evironment which is suﬃciently hot. ˆ u ˆ (8. ϕ. For the two angles α = ∠(ˆ1 . u2 . there are ony two linearly independent directions of u possible. Integrating and summing accordingly over all contributions as given by (8. say u1 and u2 . according to (8.158) above that in ﬁrst order the contributions of all components of the radiation ﬁeld add. ρ) and ˆ u β = ∠(ˆ2 .. ϕ2 = π/2.228 Interaction of Radiation with Matter where we used the fact that due to the δ-function factor in (8.189) where the factor 1/ arose from the integral over the δ-function. x3 -axes of a right-handed cartesian vectors u1 . Accordingly. |0 . The δ-function appearing in this expression.193) . planar waves as deﬁned in (8.31–8. ϕ1 = 0 and of u2 by ϑ2 = π/2.33) are orthogonal to each other. directions. three-dimensional vector. one can express | u · n| r |0 |2 = |ρ|2 ( cos2 α + cos2 β ) = sin2 θ . 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. and all polarizations of the radiation. the direction of u1 is described by the ˆ ˆ ˆ ˆ angles ϑ1 = π/2.e. ρ) holds then u cosα = cosϑ1 cosϑ + sinϑ1 sinϑ cos(ϕ1 − ϕ) = sinϑcosϕ and cosβ = cosϑ2 cosϑ + sinϑ2 sinϑ cos(ϕ2 − ϕ) = sinϑsinϕ . and (2) the fact that strictly monochromatic radiation cannot be prepared such that any radiation source provides radiation with a frequency distribution.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.

194) combined are typically somewhat smaller than (1 ˚/1000 ˚)3 = A A 10−9 .189) to yield the total absorption rate kabs (tot) = Nω 4 e2 ω 3 | n|r|0 |2 3 c3 . for Nω = 0.e. Transition Dipole Moment The expression (8. 2 . in particular. . For this purpose we use ω/c = 2π/λ and obtain 4 e2 ω 3 32π 3 e2 ao 1 ao = = 1. Otherwise. ˆ In case of emission only the second term V2 exp(+iωt) in (8. however. However.172. .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 . in agreement with observations.. Accordingly. as for the absorption process.. (8.e.37 × 1019 1 ao | n|r|0 |2 × s λ λ2 .196) The last two factors in (8. by planar waves with vector potential (8. The value of | n| r |0 | determines the strength of an optical transition.7: One-Photon Emission and Absorption 229 This geometrical average. However. The quantum nature of radiation manifests itself in that the number of photons Nω msut be an integer. a problem in case of emission by quantum systems in complete darkness.. This poses. This dependence predicts. Nω = 0.37 × 1019 × 3 3 c3 3 ao λ3 s λ and kabs where λ = 2πc n − (8. The most intensely absorbing molecules are long. 1. can be inserted into (8. The radiation ﬁeld is described. This diﬀerence between emission and absorption is due to the quantum nature of the radiation ﬁeld.173). (8.194) for the absorption rate shows that the essential property of a molecule which determines the absorption rate is the so-called transition dipole moment | n| r |0 |.173) contributes. the absorption rate is of the order of 109 s−1 or 1/nanosecond. there is also a contribution to the emission rate which is proportional to Nω . the calculation of the emission rate proceeds as in the case of absorption. In case of a classical radiation ﬁeld one would expect that emission cannot occur.171) and perturbation (8. The corresponding process is termed spontaneous emission. 8. a quantum mechanical treatment of the radiation ﬁeld leads to a total emission rate which is proportional to Nω + 1 where Nω is the number of photons before emission. the resulting total rate of emission bears a diﬀerent dependence on the number of photons present in the environment. . that emission occurs even if no photon is present in the environment. The transition dipole moment can vanish for many transitions between stationary states of a quantum system.195) = Nω 1. linear molecules. Emission of Radiation We now consider the rate of emission of a photon. However. ﬁnally. i.8. Nω photons before absorption. for atoms or symmetric molecules. i. For practical evaluations we provide an expression which eliminates the physical constants and allows one to determine numerical values readily.197) o (tot) (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 . exp[− ω/kB T ] = This equation yields Nω = 1 . respectively.8 Two-Photon Processes In many important processes induced by interactions between radiation and matter two or more photons participate... it must hold Nω ksp N0 = (Nω + 1) ksp Nn (8.201) ksp = 3 c3 the rates of absorption and emission are Nω ksp and (Nω + 1)ksp . Let No and Nn denote the number of atoms in state |0 and |n . The number of atoms undergoing absorption in unit time are Nω ksp No and undergoing emission are (Nω +1)ksp Nn .200) where kB is the Boltzmann constant. Nω + 1 (8.195) and (8. . 8.199) Planck’s Radiation Law The postulate of the Nω + 1 dependence of the rate of emission as given in (8.198) (8. For these numbers holds Nn / No = exp[−( n − o )/kB T ] (8. In the following we discuss several examples. 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.203) i.198) is consistent with Planck’s radiation law which reﬂects the (boson) quantum nature of the radiation ﬁeld. in lasers. = (8. accordingly.200).g. using (8.204) Nω . Deﬁning the rate of spontaneous emission 4 e2 ω 3 | n|r|0 |2 (8. e.230 Interaction of Radiation with Matter which is termed induced emission since it can be induced through radiation provided.198) to determine the stationary distribution of photons ω in an oven of temperature T .202) It follows. the well-known Planck radiation formula. 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. exp[ ω/kB T ] − 1 (8. respectively. Hence. We assume n − o = ω.e. To demonstrate this property we apply the transition rates (8. The total rate of emission.

If the sum in (8. ﬁnally. The transition requires that the transition dipole moment n| r |0 does not vanish for two states |0 and |n .208) has been prepared. 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 (8.106). dropping the index 1 characterizing the radiation. 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. but then requires the absorption of two photons.206) − o − 2 ω) . In applying (8. We wish to describe an observation in which a detector is placed at a solid angle element dΩ2 = sinθ2 dθ2 dφ2 with respect to the origin of the coordinate system in which the electron is described.168) where the ﬁrst term describes the relevant contribution. however.207) Expression (8.207) does not provide a suitable avenue of computing the rates of 2-photon transitions. faced with the dilemma of having to sum over all intermediate states |m of the system. however. One obtains.187). as described through the vector potential ˆ A(r. (8. (8. under ordinary circumstances.207) one is. 8. then expression (8.173). .207) does not converge rapidly. We assume that a planar wave with wave vector k1 and polarization u1 . (8. The resulting rate of the transition depends on 2 Nω . a transition between the states |0 and |n may be possible.181) and using (8. ˆ (8. which is not necessarily the case.104) with stationary states |n deﬁned through (8. In this case one needs to choose the energy of the photons to obey n = o + 2 ω.8. Employing the dipole approximation (8.172. even if n| r |0 vanishes. in case ˆ of absorption only V1 contributes. 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 (8.207) 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. induce single photon absorption processes as described by the transition rate Eq. The electron absorbs the radiation and emits immediately a second photon.205) The respective radiative transition is of 2nd order as described by the transition rate (8.182) yields. However. The intense radiation ﬁelds of lasers allow one to increase transition rates to levels which can readily be observed in the laboratory.8: Two-Photon Processes Two-Photon Absorption 231 The interaction of electrons with radiation. 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) . t) = Ao1 u1 cos(k1 · r − ω1 t) .

105). t) = A+ u1 exp[ i (k1 · r − ω1 t) ] + A− u2 exp[ i (k2 · r − ω2 t) ] . i. 8. Let us assume for the present that the emitted photon has a wave vector ˆ k2 with cartesian components sinθ2 cosφ2 k2 = k2 sinθ2 sinφ2 (8. This expansion yields.210) is accounted for by the following contributions of (8.105) using the vector potantial (8.209) cosθ2 where the value of k2 has been ﬁxed.210) A(r.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 . In the present case we consider only scattering processes which absorb radiation corresponding to the vector potential (8.208) and the emission of the radiation in (8. u2 . ˆ The perturbation which arises due to the vector potential (8. o1 ˆ o2 ˆ (8. The relevant terms of the perturbation (8.6 above that the absorption of the radiation in (8.213) where N2 /V is the density of photons characterized through k2 . t) = Ao2 u2 cos(k2 · r − ω2 t) .211) is o2 A− = o2 8π (N2 + 1) ω1 V (8. the amplititude A− deﬁned in (8.e.211) The ﬁrst term describes the absorption of a photon and.208). ˆ (8.212) where N1 /V is the density of photons for the wave described by (8. 8. later we will allow the quantum system to select appropriate values. the amplitude A+ is given by o1 A+ = o1 8πN1 ω1 V (8.. the wave characterized through k1 .211) are given by VS (t) = e ˆ p · A+ u1 exp[i(k1 · r − ω1 t)] + A− u2 exp[−i(k2 · r − ω2 t)] o1 ˆ o2 ˆ 2me contributes in 2nd order + e2 4me A+ A− u1 · u2 exp{i[(k1 − k2 ) · r − (ω1 − ω2 ) t]} ˆ o1 o2 ˆ contributes in 1st order The eﬀect of the perturbation on the state of the electronic system is as stated in the perturbation expansion (8.210) The vector potential which describes both incoming wave and outgoing wave is a superposition of the potentials in (8. u1 .214) .210). The second term in (8.208.208) and emit radiation corresponding to the vector potential (8. We know already from our description in Section 8. according to the ˆ description of emission processes on page 229. for the components of the wave (8. The vector potential describing the emitted plane wave is then A(r.210). in the present case. (8.141). hence. however.211) accounts for the emitted wave and.208.211) is stated in Eq.

Accordingly.e..212.8. Following closely the procedures adopted in evaluating the rates of 1st order and 2nd order radiative transitions on page 222–225. only these two terms contribute to the scattering amplitude. evaluating the time integrals in (8. one needs to carry out the integration 2 V(2π)−3 k2 dk2 · · ·.216) over all available modes of the ﬁeld. i.8: Two-Photon Processes function accounting for absorption and re-emission of a photon n|ΨD (t) = + + m=0 233 n|0 + 1 e2 A+ A− u1 · u2 n|0 ˆ ˆ i 4me o1 o2 ∞ (8.216) 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 .181) in stating this result. i..217) 2 Here ro denotes the classical electron radius ro = e2 = 2. Inserting also the values (8.215) t dt ei( to n− o− ω1 + ω2 +i λ)t 1 i 2 e2 4m2 e A+ A− × o1 o2 × ˆ ˆ u1 · n| p |m u2 · m| p |0 × ˆ ˆ t t × to dt to dt ei( n− m− ω1 +i λ)t i( e m− o+ ω2 +i λ)t ˆ ˆ + u2 · n| p |m u1 · m| p |0 × ˆ ˆ t t × to dt to dt ei( n− m+ ω2 +i λ)t i( e m− o− ω1 +i λ)t We have adopted the dipole approximation (8.218) .8 · 10−15 m . one needs to integrate the rate as given by (8.e. 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 (8.215) correspond to scattering processes in which the radiation ﬁeld ‘looses’ a photon ω1 and ‘gains’ a photon ω2 . Only the second (1st order) and the third (2nd order) terms in (8. Hence. 8.215) and taking the limits limto →−∞ and limλ→0+ yields the transition rate k − m = 2π δ( n − o − ω1 + ω2 ) e2 A+ A− u1 · u2 n|0 ˆ ˆ 4m2 o1 o2 e 2 (8.

Rayleigh scattering is deﬁned as the limit in which the wave length of the scattered radiation is so long that none of the quantum states of the electronic system can be excited. .. one can relate (8.224) = m 1 − o ( m ω ( ω)2 + + .221) suggests to expand S( ω) S( ω) = S(0) + S (0) ω + Using 1 − o ± ω one can readily determine m (8. one assumes the even stronger condition ω1 << | o − m| .228) S (0) = m S (0) = 2 m . .226) u ˆ 0|ˆ2 · p |m m|ˆ1 · p |0 u ˆ u ˆ 0|ˆ1 · p |m m|ˆ2 · p |0 u ˆ − me ( m − o )2 me ( m − o )2 (8. Rayleigh Scattering We turn ﬁrst to an example of so-called elastic scattering.. consequently. in fact.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 + . it follows 2 dσ = ro (N2 + 1) dΩ2 |ˆ1 · u2 − S( ω) |2 u ˆ (8. − o )2 ( m − o )3 (8.221) Using |n = |0 and.223) m Condition (8.220) n|ˆ2 · p |m m|ˆ1 · p |0 u ˆ u ˆ + m − o − ω1 2 − − o + ω2 In the following we want to consider various applications of this formula.219) (8. for all states |m of the electronic system (8.e.. Accordingly.234 Interaction of Radiation with Matter The factor N1 c/V can be interpreted as the ﬂux of incoming photons.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. i.227) u ˆ 0|ˆ1 · p |m m|ˆ2 · p |0 u ˆ u ˆ 0|ˆ2 · p |m m|ˆ1 · p |0 u ˆ + me ( m − o )3 me ( m − o )3 (8. ω1 = ω2 . 2 . a process in which the electronic state remains unaltered after the scattering. (8.225) S(0) = m u ˆ 0|ˆ1 · p |m m|ˆ2 · p |0 u ˆ u ˆ 0|ˆ2 · p |m m|ˆ1 · p |0 u ˆ + me ( m − o ) me ( m − o ) (8.

108) yields S (0) = me 2 0| [ˆ2 ·r. this term cancels the u1 · u2 term in (8.108) this is 1 0|ˆ1 ·r u2 · p − u2 · p u1 ·r |0 .235) and obtain the leading contribution to the expression (8. 8. ˆ ˆ S (0) given in (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 ) . 8.232) Obviously. ˆ ˆ We want to prove now that expression (8.230) According to (8.232.231) (ˆ1 )j (ˆ2 )k 0|[xj .227) vanishes. For this purpose we replace p using (8.226). honey bees.229) both for the u1 · p and the to u2 · p terms in (8. . It explains. pk ] = i δjk yields ﬁnally ˆ S(0) = 1 S(0) = i 3 3 (8.234) where we used for the second identity the fact that u1 ·r and u2 ·r commute. u ˆ ˆ ˆ ˆˆ i The commutator property [xj .236) is of great practical importance. pk ]|0 = u u ˆ j.226) into S(0) = 1 i ( 0|ˆ1 ·r |m m|ˆ2 · p |0 − 0|ˆ2 · p |m m|ˆ1 ·r |0 ) u u ˆ u ˆ u m .233) m Employing again (8. (8.182) 0|ˆ1 · p |m u ˆ 1 = 0|ˆ1 ·r |m u me ( m − o ) i This transforms (8.224.108) for the identity operator.235) We can now combine eqs. Using again (8.. u u u u (8.k=1 (ˆ1 )j (ˆ2 )k δjk = u1 · u2 u u ˆ ˆ (8. ˆ We want to simplify ﬁrst (8.236) 2 × m 0|ˆ∗ ·r |m m|ˆ2 ·r |0 u1 u 0|ˆ∗ ·r |m m|ˆ1 ·r |0 u2 u + m − o m − o We have applied here a modiﬁcation which arises in case of complex polarization vectors u which ˆ describe circular and elliptical polarizaed light. for example.229) (8. the blue color of the sky and the polarization pattern in the sky which serves many animals. For this purpose we apply (8.222).234. i.e. Expression (8.182) for p and the expression (8.8.k=1 j. as a compass.228) provides then the ﬁrst non-vanishing contribution to the scattering cross section (8.8: Two-Photon Processes 235 ˆ These three expressions can be simpliﬁed using the expression (8.222). 8. u1 ·r ] |0 = 0 u ˆ (8. m|ˆ1 · p |0 u ˆ 1 = − m|ˆ1 ·r |0 u me ( m − o ) i (8.222) of the cross section for Rayleigh scattering dσ = 2 ro m2 ω 4 (N2 + 1) dΩ2 × e (8.228) we obtain ˆ ˆ ˆ ˆ S (0) = 2me 2 m 0|ˆ2 ·r |m m|ˆ1 ·r |0 u u 0|ˆ1 ·r |m m|ˆ2 ·r |0 u u + m − o m − o (8.

(8.e. |n = |0 and ω1 = ω2 = ω in (8. assume now that the scattered radiation has very short wave length such that ω >> | o − m| . To evaluate |ˆ1 · u2 |2 we assume that k1 is oriented along the x3 -axis and. however.and the x2 -axes 1 0 (1) (2) u1 = 0 . u1 = 1 ˆ ˆ (8. We want to assume.240) cosθ2 We choose for the polarization of the incoming radiation the directions along the x1 . ao Bohr radius .238) (8.241) 0 0 Similarly.. for all states |m of the electronic system . u2 = u2 ˆ ˆ ˆ ˆ for u1 = u1 .244) . hence. The ﬁrst choice is ˆ sinφ2 k2 × k1 (1) u2 = ˆ = −cosφ2 (8. u ˆ 1 ao .239) We will show below that this expression decribes the non-relativistic limit of Compton scattering.236 Thomson Scattering Interaction of Radiation with Matter We consider again elastic scattering. the emitted u ˆ radiation is decribed by the wave vector sinθ2 cosφ2 k2 = k1 sinθ2 sinφ2 (8. we choose for the polarization of the emitted radiation two perpendicular directions u2 ˆ (2) and u2 which are also orthogonal to the direction of k2 . i. u2 = u2 ˆ ˆ ˆ ˆ (8..243) cos2 θ2 cos2 φ2 for u1 = u1 . that the dipole approximation is still valid which restricts the applicability of the following derivation to k1 >> One obtains immediately from (8. (8.220). though. u2 = u2 ˆ ˆ ˆ ˆ (1) (2) (2) (1) (1) (2) (1) (2) (2) (1) needs to be (8. Since u2 ˆ ˆ (1) orthogonal to k2 as well as to u2 the sole choice is ˆ cosθ2 cosφ2 (1) k2 × u2 ˆ (2) u2 = ˆ = cosθ2 sinφ2 (1) |k2 × u2 | ˆ −sinθ2 The resulting scattering cross sections 2 dσ = ro (N2 + 1) dΩ2 × for the various choices of polarizations are sin2 φ2 for u1 = u1 .240). u2 = u2 ˆ ˆ ˆ ˆ cos2 φ2 cos2 θ2 sin2 φ2 for u1 = u1 .242) |k2 × k1 | 0 where the second identity follows readily from k1 = e3 and from (8.220) 2 dσ = ro (N2 + 1) dΩ2 |ˆ1 · u2 |2 .237) The resulting scattering is called Thomson scattering.

one speaks of anti-Stokes scattering. In the case of such scattering an electron system absorbs and re-emits radiation without ending up in the initial state. Raman Scattering and Brillouin Scattering We now consider ineleastic scattering described by the Kramers-Heisenberg formula. The excitation can be electronic.g. as in the case of excitations of accustical modes of crystals. φ(vibr.247) ( 1 − cosθ2 ) (8.8. i.248) me c2 One can readily show that in the non-relativistic limit.)n . the resulting scattering cross section is 2 ro (N2 + 1) ( 1 + cos2 θ2 ) dΩ2 . (8.247) becomes identical with the Thomson scattering cross section (8. One obtains then for the scattering cross section of ˆ ˆ unpolarized radiation (1) 1 for u2 = u2 ˆ ˆ 2 2 dσ = ro (N2 + 1) dΩ2 × (8. but most often involves other degrees of freedom.247.8: Two-Photon Processes 237 In case that the incident radiation is not polarized the cross section needs to be averaged over (1) (2) the two polarization directions u1 and u1 . The radiation scattered at right angles is even completely polarized (1) along the u2 -direction. for c → ∞ the Compton scattering cross section (8. Such scattering is called Raman scattering.)n . or in case of translational motion of molecules in liquids. The Compton scattering cross section which is derived from a model which treats photons and electrons as colliding relativistic particles is dσ tot = dσ tot where (rel) = 2 ro (N2 + 1) 2 ω2 ω1 2 ω1 ω2 + − sin2 θ2 ω2 ω1 dΩ2 (8. if energy is released. the states |n etc..220) vanishes and one obtains in case of Raman scattering ω2 2 dΩ2 | u2 · R · u1 |2 ˆ ˆ (8. molecular vibrations or crystal vibrational modes.. e. In case that the nuclear degrees of freedom excited absorb very little energy. in case of a diatomic molecule |n = |φ(elect.e. 8. If energy is absorbed by the system. The energy deﬁcit is used to excite the system.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. the ﬁrst term in (8. one speaks of Stokes scattering. In the case that the scattering excites other than electronic degrees of freedom. deﬁned in (8.e.246) 2 This expression is the non-relativistic limit of the cross section of Compton scattering. i. the scattering is termed Brillouin scattering.220) represent actually electronic as well as nuclear motions.246).249) dσ = ro (N2 + 1) ω1 −1 −1 ω2 − ω1 = . the scattered radiation is polarized to some degree. For electronic systems in molecules or crystals the degrees of freedom excited are nuclear motions.. ˆ In case that one measures the scattered radiation irrespective of its polarization. Since the scattering is inelastic.

the Raman scattering cross section will be much enhanced.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. ω1 ≈ m − o for a particular state |m .250) (8. In case that the incoming photon energy ω1 is chosen to match one of the electronic excitations. e.. Nevertheless. no singlularity developes in such case due to the ﬁnite life time of the state |m . .g. a property which can be exploited to selectively probe suitable molecules of low concentration in bulk matter.k xj Rjk yk .251) ω1 − ( We deﬁne x · R · y = j. a case called resonant Raman scattering. the cross section for resonant Raman scattering can be several orders of magnitude larger than that of ordinary Raman scattering. Of course.

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

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

k=1 (9. 1 together with multiplication. for the scalar product holds j|k = δj k .. the second group to fermions.1: Bosons and Fermions Permutation Symmetry 241 A second consequence of the identical character of the particles under consideration is that the Hamiltonian must reﬂect the permutation symmetry. i. {1} (trivial) and {1. ∀P ∈ SN . S where S denotes the number of available single particle states. particles with integer spin values. i.. The single particle states are assumed to be orthonormal. it is essential that the functions F ( ) and G( .11) . In fact. (9. (9. The terms of the Hamiltonian will be discussed further below. 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 .10) We will construct now wave functions which exhibit the appropriate properties. the Hamiltonian must be of the type N H = j=1 1 F (xj ) + 2 N G(xj . Obviously. P ] = 0 . usually a very large number. 2. in turn.9.9) = 1 for P even −1 for P odd (9. . Presently.e. i.5) implies that a permutation eﬀectively changes the indices of the particles) and. both groups. particles with half–integer spin values. all particles being governed by the same interactions.e. the Hamiltonian for permuted indices P HP −1 is identical to H. −1. (9. The ﬁrst case applies to bosons. Fock-Space Our derivation will start by placing each particle in a set of S orthonormal single particle states x|k = φk (x). In fact..7) This property.-1}. From this property follows that H is independent of any speciﬁc assignment of indices to the particles (note that (9. In fact.8) |Ψ (9. ) are the same for all particles and for all pairs of particles. or the corresponding group of two elements 1. . the representation is either isomorphic to the group composed of only the ‘1’. xk ) j. it is a postulate of quantum mechanics that for any description of many–particle systems the permutation operators must be diagonal. a diagonal representation can only berealized in a simpler group.e. hence. xk ) is symmetric in its two arguments. In the latter case all even permutations are represented by ‘1’ and all odd permutations by ‘-1’.e. implies that the permutation operators can be chosen in a diagonal representation. i. k = 1.6) which describes one–particle interactions and two–particle interactions of the system. provide a representation of the group structure represented by the multiplication table of even and odd permutations given above. . i.. From the latter results [H. Since permutations do not necessarily commute.e. G(xj .

. the nj specify how often a single particle state |j is occupied. xN |ΨB (n1 . ...e.15) (nj !) N! 1 2 P ∈Λ(λ1 . ..13) These states form an orthonormal basis of N -particle states. i. the states |ΨFock (λ1 . . .8) is x1 . We ﬁrst consider orthonormal many–particle wave functions. .. . . (9. λN ) only if λj = λk . . hence. .15) with λk = j.8. λj ∈ {1. . . . λN ) = j=1 φλj (xj ) . . . nS ). . n2 . . . It holds N ΨFock (λ1 . . k. .14) Obviously. The integers nj are equal to the number of λk in (9. . This detail is connected with the fact that a two–particle state of the type ΨFock (λ. λN ) = j=1 δλj λj . . . . . . . λ) in which two particles occupy the same single– particle state |λ does not allow transposition of particles. with λj = λk = . .15) completely and. x2 . . . . . λ2 .15) is that the sum over permutations does not involve permutations among indices j. N x1 . 9. xN |ΨFock (λ1 . The reader is well advised to pause and grasp the deﬁnition of the nj . 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.. |λN . . . . (9. referred to as the occupation numbers. appear severalfold in (9. . . . 2.11. |λ2 . λ2 . 9. nS ) = S j=1 (9. xN |Ψ Fock (λ1 .15). each particle j into a speciﬁc state |λj . The numbers nj . i.9). . λ2 .9). . ..12) where the second factor represents the scalar product between spin states. A wave function which obeys the boson symmetry (9. . . In a second step we form linear combinations of Fock states with the desired symmetries. . . λ2 . . λN )|ΨFock (λ1 . λ2 . .λN ) on the rhs. S} of the type (9. An important detail of the deﬁnition (9. λ2 . therefore.12). characterize the wave function (9. . . e. . is a particular choice of placing N particles consistent with the occupation numbers (n1 . . .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. T (j. λN ) do not exhibit the symmetries dictated by (9. The Fock states represent N particles which are placed into S states |λ1 . n2 . .λN ) Here the states (λ1 . x2 . which do not yet obey the symmetries (9.λN ) P x1 . . are essential.8. the so-called Fock-states. . . x2 . .e. . the reason is that such transposition duplicates the state which would.g. . k) ∈ Λ(λ1 . Λ(λ1 . 9. . . . λN ) ⊂ SN is the set of all permutations involving only particles j with diﬀerent λj .

n2 . .16). . φλN (x1 ) φλN (x2 ) .. . . . These states form an orthonormal basis. correspond to a particular choice of placing N particles consistent with the occupation numbers (n1 . nS ) = φλ1 (x1 ) φλ1 (x2 ) . nS ) = √1 N! P ∈SN P P x1 . n2 .17) Here the states (λ1 .19) φλ1 (xN ) φλ2 (xN ) . ..1. . for two sets of occupation numbers N = (n1 . . nS ) and N = (n1 .. .17) describe. n2 . . . Using the identity governing the determinant of N × N –matrices A N det(A) = P ∈SN P j=1 Aj P (j) (9. . . electrons in an atom. for example. . The fermion states (9. xN |ΨF (n1 . .16) Exercise 9. . nS ). . (9.1: (a) Show that the states (9. . i.. . . λ2 . .e. for two sets of occupation numbers N = (n1 . √1 N! (9. x2 . . . the photons of the electromagnetic ﬁeld or the phonons in a crystal. the integers nj denote the occupancy of the single–particle state |j . . . . Wave Functions with Fermion Symmetry in the Occupation Number Representation One can construct similarly a wave function which obeys the fermion symmetry property (9. nS ) holds S ΨF (N )|ΨF (N ) = j=1 δnj nj . a molecule or a crystal. . . φλN (xN ) Here. . (b) Show that for the states (9. .1: Bosons and Fermions 243 The wave functions (9.λN ) (9.9). . φλ2 (x1 ) φλ2 (x2 ) .e. . . . x2 . n2 . . . The important property holds that nj can only assume the two values nj = 0 or nj = 1. nS ) and N = (n1 . for example. x2 . (9. .λN ) on the rhs. . n2 . . . .9. n2 . i. Such wave function is x1 .8).15) describe. . xN |ΨF (n1 . . n2 . . λ2 . xN |ΨFock (λ1 . These states form an orthonormal basis. . . . . .18) this wave function can also be expressed through the so-called Slater determinant x1 . . .15) holds (9. . .20) ..15) obey the boson symmetry (9. . . For any value nj > 1 two or more of the columns of the Slater matrix are identical and the wave function vanishes. . . nS ) holds S Ψ (N )|Ψ (N ) = j=1 B B δnj nj . ..

244 Many–Particle Systems The representation of wave functions (9. .17) obey the orthonormality property (9.15) and (9. nS ) = nj ΨB (n1 . (b) Show that the states (9. Creation and Annihilation Operators for Bosons We consider the operator aj deﬁned through aj ΨB (n1 . For the adjoint operator a† holds j a† ΨB (n1 . (9. n − 1. . n ) are normalΨ 1 j 1 j S S ized according to (9. [a† . . . nj . . .22) The operators aj and a† obey the commutation properties j [aj . . . . To prove that (9. . . .17) obey the fermion symmetry (9. 9. . . . a† ] = δj k . . . . . . . . nj . 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 .2: (a) Show that the states (9. .15). .25) . guaranteeing the correct permutation properties of the many–particle states. . . .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 . n ) and the N − 1-particle wave function ΨB (n .1. ak ] = 0 . .23) (9.20). are the so-called operators of 2nd quantization. . . . Employing the superindices N and N introduced above and using the orthonormality property (9. . . . .21) √ The factor nj appears here on the rhs. since both the N -particle wave function B (n . Exercise 9. . . a† ] = 0 j k (9. . . nS ) = j nj + 1 ΨB (n1 .24) [aj . . .15). .22) follows from (9. n . These operators allow one to construct and manipulate many–particle wave functions while preserving permutation symmetry. (9. . nS ).16) one obtains (Aj )N = ΨB (N )| aj |ΨB (N ) = √ S N nj δnj nj −1 k=1 k=j δnk nk . . . n2 . nj − 1. (9. nj + 1.21) we consider the matrix Aj corresponding to aj in the basis of many–particle states (9. . . nS ) nj ≥ 1 0 nj = 0 √ . . nS ) .9). . .2 Operators of 2nd Quantization An important tool to describe mathematically systems of many bosons or fermions. . .

. . . (9. nS = 0) as follows S (9. a† ] = 1 For j = k follows similarly 1. .32) ˆ ˆ i. nS ) . . It holds for j = k (aj a† − a† aj )|Ψ(N ) = (nj + 1 − nj )|Ψ(N ) = |Ψ(N ) . the eigenvalues of Nj are the occupation numbers nj . 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 . . One can readily show using (9. . n2 . the properties (9. .21. i. . . . .e. . . j (9.27) Equations (9. . 9.9. n2 . (9.33) is called the particle number operator since.22) follows from (9.. (9.29) |ΨB (n1 .e. nS ) = j=1 a† j nj nj ! |0 . . .2: Operators of 2nd Quantization Let A† be the matrix adjoint to Aj . j j Since this holds for any |Ψ(N ) it follows [aj .31) This operator is diagonal in the occupation number representation.22) ˆ Nj |ΨB (N ) = nj |ΨB (N ) .28) (9. . . n2 = 0. . (9. . nS ) N |ΨB (n1 . nj − 1 . n2 . .24).21) and vice versa. nS ) = 0 . nS ) = |ΨB (N ) . . 9. . j (aj a† − a† aj )|Ψ(N ) = k k ( nj (nk + 1) − (nk + 1)nj )|Ψ(n1 .34) = . The creation operators a† allow one to construct boson states ΨB (N ) from the vacuum state j |0 = |ΨB (n1 = 0. We will prove the commutation properties (9. .26) From this result one can immediately conclude that (9. S ˆ N |ΨB (n1 . .. . .27. . n2 .29) yield the commutation relationship (9. obviously. . The related operator S ˆ N = j=1 ˆ Nj (9. One refers to Nj as the occupation number operator. nS ) = j=1 nj |ΨB (n1 .30) Of particular interest is the operator ˆ Nj = a† aj . n2 . nk + 1 . for basis states |ΨB (n1 . (9.23) follow in an analogous way.24). .

. in case that the single particle states |j and |j + 1 are both occupied.e. one does not need to worry ever about proper symmetries of wave functions.29.. nj+1 = 1. 9.9). 9.38) . i. . ... φj+1 (x1 ) φj+1 (x2 ) . since they are induced through the algebra of a† and aj .38) corresponds to x1 . . .. the fermion wave function changes sign when one exchanges the order of the occupancy. nS ) .. 9. 9. e. in case it holds for nj . . x2 . φj (x1 ) φj (x2 ) . nj (9. . . . xN |ΨF (n1 . . (9. operators which can alter the occupancy of the wave function (9. .246 Many–Particle Systems In using the boson creation and annihilation operators.. (n − 1)! (9. .35) and where the derivative operation is deﬁned through ∂f (A) = ∂A ∞ n=1 1 f (n) (0) An−1 . . . Obviously.. .2. nS ) = − x1 . . . it also holds for nj + 1.s. . nj+1 .2: Show that the boson operators a† and aj satisfy j [aj . . j Exercise 9. nj .23) hold for the state deﬁned through (9. xN |ΨF (n 1 . . . one needs to apply only the commutation propertes (9.23.30). . 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 .39) .e. in general.22. .37) (9. x2 . of (9. . Prove this by induction. f (a† )] j [a† . √1 N! . . 9.36) Creation and Annihilation Operators for Fermions We want to consider now creation and annihilation operators for fermions. x2 . showing that the property holds for nj = 0 and. i.19) without aﬀecting the fermion symmetry (9.. xN |ΨF (n1 . As long as one starts from a wave function with the proper boson symmetry. nj+1 . from the vacuum state |0 or states |ΨB (N ) .24) and aj |0 = 0 imply that the the properties (9. .h. . ..1: The commutation relationships (9.23. nj+1 . Exercise 9. . . . f (aj )] j = ∂f (a† ) j ∂a† j ∂f (aj ) ∂aj = − where the operator function is assumed to have a convergent Taylor expansion ∞ f (A) = n=0 1 (n) f (0) An n! (9. . .. To prove this property we notice that the l.2. . . . 9. . .g. φj (xN ) φj+1 (xN ) .21. . nS ) = .17. . . x1 . .. .24) and the property aj |0 = 0.

A proper example is the two particle fermion wave function x1 . . nS ) qj states occupied In case that the single particle state |j is not occupied. . We will be adopting below the notational convention that qj corresponds to occupancies N = (n1 . . . nS ) = nj (−1)qj |ΨF ( n1 . vanishes.. . nS ). . n4 = 0. Obviously.e. . . nj → 0. . . . i. . . nS = 0) φ2 (x1 ) φ5 (x1 ) 1 1 = √2 = √2 [φ2 (x1 )φ5 (x2 ) − φ5 (x1 )φ2 (x2 )] . nj+1 .42) i. . . . ..43) .41) Before we consider the deﬁnition of fermion creation and annihilation operators we need to take notice of the fact that one also needs to deﬁne at which position in the wave function. .e. (9.40) are identical. . .e. nS ) = . that occupancies are always ordered according to a monotonous increase of the single–particle state index. nj+1 . at the ﬁrst column of the Slater determinant (9. φj (x1 ) φj (x2 ) . x2 . . . . in order to be consistent with this convention. One adopts the convention that particles are created and annihilated by these operators always at the ﬁrst position of the wave function. of (9. particles are being created or annihilated... . φj+1 (x1 ) φj+1 (x2 ) . the rhs... x2 |ΨF (n1 = 0. n2 . . it is necessary to specify for a fermion wave function the order in which the single– particle states |λj are occupied. . n2 . n2 . . . n3 = 0. . . .9. n6 = n7 = . xN |ΨF (n1 . . nj . n2 . We are now ready to deﬁne the annihilation operator cj for fermions in the single–particle state |j as follows cj |ΨF (n1 .19). nj−1 .e. 247 − √1 ! N .19).40) .. . nS ) as follows: j−1 qj = k=1 nk . . .. 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). . . i. n2 = 1. (9. (9. however. . This requires one.. . the expressions (9.39) and (9..h. 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. Because of the property of the determinant to change sign when two columns are interchanged. at which column of the Slater determinant (9. φ2 (x2 ) φ5 (x2 ) (9.. This change of position is connected with a possible sign change (−1)qj where qj is deﬁned for a given N = (n1 . nj = 0. . For this purpose one must adhere to a strict convention: the labelling of single–particle states by indices j = 1. . φj+1 (xN ) φj (xN ) .38) reads − x1 . n3 . . nj . 2. . i. .2: Operators of 2nd Quantization The r. qj is the number of states |k with k < j which are occupied in a fermion wave function. .s.. . n5 = 1.

It holds c† cj |ΨF (N ) = c† (−1)qj nj |ΨF (n1 . nj−1 .46). 9. nj .17. . nS ) = j Many–Particle Systems (−1)qj (1 − nj ) |ΨF ( n1 .48) From this follows (9. . . ck ]+ = 0 . Similarly one obtains cj c† |ΨF (N ) = cj (−1)qj (1 − nj ) |ΨF (n1 . . (9. the last factor on the rhs. [c† .43). (9. nS ) qj states occupied The operators thus deﬁned obey the commutation properties [cj .46) ﬁrst for the case j = k.50) .19). . nS ) = j (−1)qj (nj + 1))(−1)qj (1 − nj ) |ΨF (N ) = (1 − nj ) |ΨF (N ) (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 .44). nj → 0 . c† ]+ = δj k k where we have used the deﬁnition of the so-called anti-commutators [A. B]+ = AB + BA. . Let Cj be the matrix corresponding to the operator cj in the basis of fermion states (9. One obtains for its matrix elements j C† j NN = ΨF (N )| c† |ΨF (N ) = (Cj )N j S N = nj (−1)qj δnj nj −1 k=1 k=j δnk nk S = (−1) (1 − nj ) δnj nj +1 k=1 k=j qj δnk nk . otherwise.44) (9. nS ) = j j (−1)qj (1 − nj + 1)(−1)qj nj |ΨF (N ) = nj ΨF (N ) (9.48) vanishes. c† ]+ = 0 j k (9. We have used here the deﬁnition qj = k<j nk along with qj = qj . of (9. nj → 1 . The anti-commutation properties (9. n2 .45) follow in an analogous way and will not be derived here. . n2 .49) The derivation involves realization of the fact that a non-zero result is obtained only in case nj = 1. .47) Let C† be the matrix adjoint to Cj . . . We will prove now the anti-commutation property (9.45) (9. We consider (9. n3 .46) [cj . . . We ﬁrst show that (9. .44) follows from the deﬁnition (9. . . nj+1 . nj → 1.248 The operator c† adjoint to cj exhibits the following property j c† |ΨF (n1 . . .

29)) |ΨF (.s. We j will often separate the states |j .h.e. nk = 1 . x|j → φj (r) | 1 σ .58) . 2 (9. j j For j < k one obtains cj c† |ΨF (N ) = (−1)qj +qk nj (1 − nk )|ΨF (n1 . . i.51.56) Here we have made use of cj cj = 0 and c† c† = 0 which follows from nj ∈ {0. coordinates x and index j into a space part and a spin part. On the r. of this equation we meant to indicate only those N occupation numbers nλj which are non-vanishing. . . 1 In the following we will apply fermion operators c† and cj only to electrons which carry spin 2 .. . .46). the creation operators must operate in the proper order.54) ˆ N = j=1 ˆ Nj (9. i.52) hold for any |ΨF (N ) one can conclude (9.. The stated property ˆ follows from the idempotence of Nj which can be derived as follows † ˆ2 ˆ Nj = c† cj c† cj = c† (1 − c† cj )cj = c† cj − c† c† cj cj = cj cj = Nj . j j j j j j j (9.9. . 1}. . . for which purpose one needs to envoke only the algebraic (anti-commutation) properties (9. with eigenvalues equal to the occupation numbers nj ˆ Nj |ΨF (N ) = nj |ΨF (N ) ˆ Nj . nλN . . is referred to as the occupation number operator. . in the basis |ΨF (N ) . nk = 1 . ˆ It is an interesting exercise to demonstrate that Nj only has eigenvalues 0 or 1. S (9.50) yield ( cj c† + c† cj ) |ΨF (N ) = |ΨF (N ) . 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. .e.52) One can obtain similarly the same relationship in the case j > k.55) is called the particle number operator. . . . nλ2 . j → j.53) is diagonal in the occupation number representation. . hence. nj = 0 .s. . Since (9. . λ λ λ (9.) = c† 1 c† 2 · · · c† N |0 . σ 1 (σ = ± 2 ) . . hence. Equation (9.49) and (9.49) shows that the operator ˆ Nj = c† cj j (9.57) On the l. ( cj c† + c† cj ) |ΨF (N ) = 0 . an operator cj must be on the left of an operator ck for j < k.51) (9. Correspondingly.h. nj = 0 . .46). k k 249 (9. 9.e. nλ1 . i.2: Operators of 2nd Quantization (9. nS ) k c† cj |ΨF (N ) = (−1)qj +qk −1 nj (1 − nk )|ΨF (n1 . nS ) k and. ..

59) and (9. x2 . namely. The operators g (x.250 Many–Particle Systems 9. The essential aspect of deﬁnition (9.F (N )|F |ΨB.17). ˆ ΨB.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. 2. Operators of the type 1 ˆ G = 2 N g (xj . . We adopt a similar expression to distinguish two–particle operators G and g . .60) involves generic operators – acting on xj and on xj . xk . (9. S of single–particle spinorbital states (see page 241) and the operators are speciﬁed through the associated generic operators ˆ f and g . . 9. 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 ) . .F (N ) .15. 9.59) and over pairs of particles in (9.61) is not a one particle operator as long as the N operators rj (x).F (N )|G|ΨB. xk ) ˆ j. respectively – which are all identical. 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. i.F (N ) .15. N are not all identical. . . . x ) which constitute G are ˆ called the associated generic two–particle operators. . r = 1. j = 1. ˆ r|f |s = 1 ∗ ˆ d3 r ψr (r. The operators f (x) which constitute F are called the associated generic single–particle operators1 . 9. (9. respectively.59.63) ˆ ˆ This expression has been speciﬁed for the purpose of these notes to distinguish F and f and is not common ˆ terminology.17) and integrating over all particle space-spin coordinates x1 .60) ˆ correspondingly are called two–particle operators.6) above. ˆ . 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.e. These operators had been introduced already in Eq. xN ..62) These matrix elements can be obtained by chosing the operators and many-particle states in the position representation. The action of the operators boson and many–fermion states |Ψ is obviously described through the many–particle state matrix elements ˆ ΨB. σr ) f (x) ψs (r. . (9. . An operator N ˆ R = j=1 rj (xj ) ˆ (9.60) is that the sum over particles in (9.59) ˆ ˆ are called single–particle operators. σs ) (9. 2.59.k=1 j=k (9. 9. Since the many–particle states are built-up from a basis |r .

∂r 2 (9. u = s.68) Another single–particle operator is the one–particle density operator N ρ(r ) = ˆ˜ j=1 δ(r − rj ) . that the symmetry implies that one can switch simultaneously the pairs of indices r. ˜ (9.59).64) r. σt )ψu (r . according to (9. ∂rj2 (9. s|ˆ|t.69).64) We like to note an important symmetry of the matrix element r.64). Notice.. x ) ψt (r. is t(x) = (− 2 /2m)(∂ 2 /∂r 2 ). u = g d3 r ∗ ∗ d3 r ψr (r.71) . ∂r 2 ∂x2 ∂x2 ∂x2 1 2 3 (9. We will show that the boson and fermion creation and annihilation operators serve this purpose. σs ) g (x. σu ) .66) Here we adopt the notation for the Laplacian ∂2 ∂2 ∂2 ∂2 = + + .3: One– and Two–Particle Operators r. ˆ 251 (9.66) and (9.9. 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. t .62) in terms of the matrix elements of single–particle states (9. u which follows directly g from the deﬁnition (9. s|ˆ|t. 9.67) ˆ In this case the generic single–particle operator. An example for a two–particle operator is the Coulomb repulsion operator 1 ˆ V = 2 N j. g g (9. s|ˆ|t. one cannot switch the indices of one pair individually. 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 .k=1 j=k q2 |rj − rk | (9. i. σr )ψs (r . u.e. r|ˆ|u. The matrix elements of the generic operator ˜ in the one–particle basis are ˆ rσ|f |sσ = δσσ d3 r φ∗ (r) δ(r − r) φs (r) = δσσ φ∗ (r) φs (r) δσσ ˜ ˜ r r ˜ (9. (9. s and t.70) Both operators.69) ˆ In this case the generic operator is f (x) = δ(r − r).63. are spin-independent as is evident from the factor δσσ . 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) .65) This symmetry will be exploited repeatedly below.

The generic operator is s12 = S1 · S2 = ˆ 1 1 S1. β|ˆ3 |s. i. α|ˆ+ |s. 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. Here.k=1 (9.3 S2. σ . The generic operator is given through the Pauli matrices. β s r.77) are orbital-independent as evidenced by the factors δrs and δrt δsu in the respective formulas.+ + S1. σ .. s = 1 σ. α s r.3 2 2 (9.74) and (9. The operator N S = j=1 Sj (9. however. The non-vanishing matrix elements of the spherical components of s are (α = + 1 .− . β = ˆ 2 2 1 −2) r.− + S1.75) is a two–particle operator.77) δσ σ ( δσσ − δ−σσ ) The operators (9. α s r. The ˆ matrix elements of the generic operator in terms of single–particle states are ˆ r. β s The operator S 2 = δrs = δrs = = + − 1 2 1 2 δrs δrs (9.e.+ .+ S2. s. u. (9. However. σ | t |t. u. Sj.74) = N j=1 Sj = 2 N Sj · Sk j.76) the single–particle state matrix elements of which are r.252 Many–Particle Systems which is also spin-independent.3 . s. one can obviously extract the term j Sj to be left with a two–particle operator in the strict sense of our deﬁnition. x ) = q 2 /|r1 − r2 |. Sj is the spin operator for particle j with components (in the spherical representation) Sj. β|ˆ− |s. . σ.72) As a ﬁnal example of one– and two–particle operators we consider operators of total spin. In this case the generic operator is v (x. σ.73) is a one-particle operator. α|ˆ3 |s. 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 | s12 |t.− S2. not in the strict sense of our deﬁnition above since the restriction 2 j = k does not apply in the summation.

78.79) is opposite to that in the matrix element r.9. 9. In order to determine these matrix elements one needs to evaluate ﬁrst the matrix elements ˆ of the generic operators r|f |s r. s|ˆ|t. (9.80) (9.. g and not ct cu . u cr cs cu ct g fermions (9.45.82) ΨF (N )|c† cs |ΨF (N ) = r cr 2 (9. r S Ψ (N In case of r = s one obtains F )|c† cr |ΨF (N ) r = nr r=1 δnr nr . s|ˆ|t. s|ˆ|t. 9.t.62).k=1 j=k S r.64).59) and G in (9.171 and pp.176 where the matrix elements of fermion states can be found. To show that the resulting values of the matrix elements (9.u=1 (9.u=1 S r.81) For the latter matrix elements simple rules can be derived from the algebraic properties of the boson and fermion operators (9.s=1 S r. respectively. u as deﬁned in (9.62) through single–particle state matrix ˆ ˆ elements (9. Expressions (9.63) and (9. pp.64) one replaces the operators F in (9.F (N ) have the same values as the respective matrix elements (9.23. and then in g a second step the matrix elements ΨB (N )|a† as |ΨB (N ) .79) It is crucial to notice in the expression for the fermion two–particle operator that the order of the annihilation operators in (9. These rules will be provided below only for the case of fermions. u a† a† at au bosons g r s † † r. r ΨB (N )|a† a† at au |ΨB (N ) r s ΨF (N )|c† c† ct cu |ΨF (N ) r s bosons fermions (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’.e. r ΨF (N )|c† cs |ΨF (N ) . The Matrix Elements ΨF (N )|c† cs |ΨF (N ) r We consider ﬁrst the case r = s.63.83) t=1 t=r. 9.24) and (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 . . s|ˆ|t. i. The matrix elements are then actually those of the number operator c† cr which is diagonal in the occupation number representation. u .s.79) have the following implication: The operators (9.3: One– and Two–Particle Operators Deﬁnition in Terms of Creation and Annihilation Operators 253 In order to express many–particle state matrix elements (9. xk ) → ˆ j.79) in the basis of many–particle states |ΨB. namely cu ct .s.60) as follows N ˆ f (xj ) → j=1 N 1 2 1 2 S r. 9. respectively.t.46).78) and (9.62) are correct one needs to compare the result derived with conventional derivations of the matrix elements2 .

u) are included in the summation for which all indices are diﬀerent. where each contribution corresponds to one of the three cases mentioned. i. for which holds r = s.s=1 a.i. c2 = 0 r (9. Comparision with the results in Condon&Shortley (pp. t. those for which N and N diﬀer in more than two occupation numbers. t=u 2 = 0.i.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 (9.84) The oﬀ-diagonal elements with nr = nr except for r = s.d.d.i.79) and using c† r we can write S 1 ˆ G = rs|ˆ|tu c† c† cu ct g r s 2 r.’ (all indices diﬀerent) that only tuples (r.s=1 a. t.e. s. u need to be considered.d. 9.i. ˆ ˆ ˆ ˆ Accordingly.85) All other matrix elements vanish. t.d. t). i. t.s=1 a. s. The anti-commutation property (9. in which case ns = ns + 1 and nt = nt − 1 holds.53) 1 ˆ G0 = 2 S r s|ˆ|r s − r s|ˆ|s r g g r. ˆ ˆ Nr Ns . Case 2 all four indices are diﬀerent.78) does yield the same results as the operator deﬁned in (9. (9.86) For the combination of indices r. u in this sum three possibilities exist Case 0 two of the four indices are diﬀerent.s..86) into three contributions. (r. 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. are ˆ ˆ ΨF (N )|F |ΨF (N ) = (1 − ns )(−1)qs nt (−1)qt s|f |t (9.254 From (9. namely. 1 2 S r s|ˆ|s r c† c† cr cs .. G = G0 + G1 + G2 . Case 1 three of the four indices are diﬀerent. r = t etc. (r.87) Here and in the following we denote by ‘a. (9.45) yields together with the deﬁnition of the occupation number operator (9. s). g r s r.88) . ˆ The ﬁrst contribution G0 is 1 ˆ G0 = 2 S r s|ˆ|r s c† c† cs cr + g r s r.59) when the matrix elements are determined between Slater determinant wave functions. s. Starting from the deﬁnition (9.t.e. 171) shows that the operator deﬁned in (9.82. s. we split the sum in (9.u r=s.

t=1 a. Only G0 .t=1 a. s|ˆ|r. (9.t=1 a.s.u r>s. ns .d.d.92) ˆ This contribution to G has non-vanishing matrix elements only when N diﬀers from N in four ˆ occupation numbers nr .u r<s.i.u r>s. F (N )|G|ΨF (N ) . r g g ..i.t>u S r.t.u r<s.i. (9.91) ˆ G2 = r.t.s. r s (9. As we will ﬁnd. One obtains 1 ˆ ΨF (N )|G|ΨF (N ) = 2 S nr ns r.s.84) ˆ of the one–particle operator F .s.t<u S r.93) . r s|ˆ|r t − r s|ˆ|t r g g ˆ c† ct Nr . 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. this part accounts for all diagonal contributions to G. 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.t.s. this contribution obviously behaves similarly to the oﬀ-diagonal part (9. 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 .t=1 a. s|ˆ|s.s r. employing the symmetry property (9. g r s (9.i. Commutation of ct cu according to (9. ˆ A similar series of transformations allows one to express G2 which can be written ˆ G2 = + + + as follows S 1 2 1 2 1 2 1 2 r. s (9.s. given in (9.t. ˆ Case 0 This case covers diagonal matrix elements.86).65).3: One– and Two–Particle Operators 255 Obviously.89) S r.t.45).e. S r.t>u r s|ˆ|t u − r s|ˆ|u t g g c† c† cu ct .9.90) ˆ Since Nr is diagonal. this contribution to the two–particle operator is diagonal in the occupation number ˆ representation.s.t=1 a. contributes in this case.t>u S r.d.d.s.s. s − r.88). nu . N = N or nr = nr for all r.t<u S r s|ˆ|t u c† c† cu ct g r s r s|ˆ|t u c† c† cu ct g r s r s|ˆ|t u c† c† cu ct g r s r s|ˆ|t u c† c† cu ct g r s (9.d. S r.u r>s. A similar contribution does not arise for F . ˆ The second contribution G1 is ˆ G1 = + + + 1 2 1 2 1 2 1 2 S r.s. nt .i.

one–particle and two–particle operators.73) and S 2 given in (9.94) Case 2 This case covers oﬀ-diagonal elements with nr = nr except for r = s. the ‘−’-sign applies for s < r. Only G2 . ˆ ˆ Show that the matrix elements of F and of G conserve particle number.90) from (9. particle number is conserved. In particular.92). u. Derive the non-vanishing matrix elements (9. nu = nu − 1.92). u < v. Derive (9. t − r.3. One obtains for the three components of S ˆ S3 ˆ S+ ˆ S− = 1 2 S ˆ ˆ Nrα − Nrβ = = r=1 S c† crβ rα r=1 S c† crα rβ r=1 (9.2: 9.62) for bosons.95) In the latter formula the ‘+’-sign applies for s < r. We like to express these operators through ˆ fermion creation and annihilation operators.3.92) from (9. s|ˆ|u. r g g (9.91). u < v or s > r. u > v or s > r.74) and (9. s|ˆ|r.96) .3: 9.3. One obtains ˆ ΨF (N )|G|ΨF (N ) = ± (−1)qs +qr +qu +qv (1 − ns )(1 − nr )nu nv r. nv = nv − 1. Spin Operators ˆ ˆ In this section we will brieﬂy consider the spin operators S given in (9. given in (9.4: 9.3. contributes in this case. respectively. u g g (9.e. it holds S nr = S nr .256 Many–Particle Systems Case 1 This case covers oﬀ-diagonal elements with nr = nr except for r = s. v − r. t.t nr r.89).75) which are. Only G1 . r=1 r=1 Exercise Exercise Exercise Exercise Exercise 9. given in (9. u > v. nt = nt + 1. s|ˆ|t.. Derive (9. One obtains ˆ ΨF (N )|G|ΨF (N ) = (−1)qs +qt (1 − ns )nt r r=s. v for which holds ˆ ns = ns + 1. for ˆ non-vanishing matrix elements ΨF (N )|G|ΨF (N ) at most four occupation numbers nr and nr can diﬀer. contributes in this case.90). s|ˆ|v.5: Derive (9. t for which holds ˆ ns = ns + 1 and nt = nt − 1. The matrix elements of the corresponding generic operators are (9.95) from (9.3.76). i. All matrix elements which are not covered by the three cases above vanish. Furthermore.1: 9.

98) In many cases the Hamiltonian is spin-independent and the equivalent notation S H = r. u refer only to the orbital part of the single–particle basis. the study of problems governed by Hamiltonians of the type (9. atoms and nuclei.s. 2 2 Exercise 9.9. the spin part of the single–particle states is represented by α = 1 and β = − 1 . In all these systems the Hamiltonian is a sum of one– and two–particle operators S H = r=1 1 ˆ r r|t|s c† cs + r. which represents the interactions between particles. but also ordinary phenomena like freezing and evaporation.97) is over the orbital states..s=1 r.g. e. In the latter case the indices r.t.s=1 σ 1 ˆ rσ r|t|s c† csσ + 2 S r. If it were not for the two–particle contribution to the Hamiltonian. rα sβ (9. the concerted behaviour of interacting systems can deviate qualitatively from that of systems made up of independent components.3.u=1 σ. Examples is the laser action.99) has preoccupied an important part of all intellectual eﬀorts in Theoretical Physics during the past ﬁfty years.97) The summation in (9. e. v r s 2 r. u c† c† cuσ ctσ v rσ sσ r. however.96. evaluation of their stationary states and excitation energies. molecules.4: Independent-Particle Models ˆ For S 2 one obtains 1ˆ 1 ˆ S2 = N + 2 4 S S 257 ˆ ˆ ˆ ˆ (Nrα − Nrβ )(Nsα − Nsβ ) − r. would be a simple exercise in linear algebra.99) will be used. 9. 9. v c† c† cu ct . Unfortunately. s|ˆ|t. it is one of the main intellectual achievements of Physics among the Sciences to have addressed systematically the study of systems composed of many strongly interacting components. in crystals. 9.t. Often. s|ˆ|u. the description of many–particle systems. The fortunate side of this is. t.96) and (9. the two–particle operator makes the description of many–particle systems a very hard problem..98.s=1 c† c† crβ csα . The great generality of the concepts developed in the context of many–particle systems can be judged from the fact . superconductivity. s.6: Derive (9.u S (9.97).g.σ (9. In fact. The outcome of these studies is that even when interactions between components (particles) are simple. that the two–particle operator representing interactions between particles induces a variety of interesting phenomena. Actually.4 Independent-Particle Models In the remaining part of this chapter we will apply the technique of operators of 2nd quantization to the description of many–fermion systems as they arise. the systems investigated involve a macroscopic number of components such that statistical mechanical approaches are envoked.s.

1: Imagine that in a closed. would not exist.s=1 σ ˆ rσ r|t|s c† csσ (9.4. namely of neurons3 . a Hamiltonian without a two–particle operator contribution accounting for interactions among the particles. ultimately depend in a crucial way on the fermion character of its constituent building blocks.100) ˆ ˆ We will denote the matrix elements of t as r|t|s = trs . r=1 3 Urm U ∗sm = δrs m=1 (9.101) The unitary property of Urn U† U = U U† = 1 1 or S S (9.. all material systems would condensate into states which would depend little on particle number. m = 1.e. If it were’nt for the fermion nature. Before continuing one may ﬁnally point out that the material world as we know it and as it establishes the varieties of natural substances. electrons and nucleons. The electronic properties of atoms with diﬀerent numbers of electrons would diﬀer little. i. What might happen? Independent-Particle Hamiltonian A many–fermion system governed by an independent-particle Hamiltonian. We will restrict our description in the following to systems composed of an even number (2N ) of particles and to spin-independent interactions..100) in a form in which the factors c† csσ become rσ occupation number operators. according to which nuclei and atoms change their qualitative properties when they grow larger. . 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). . ranging from nuclei to the molecules of living systems . Amit (Cambridge University Press.J. New York.102) U ∗rm Urn = δmn . The Hamiltonian of such system is S Ho = r. S}.103) An account of some of these eﬀorts can be found in Modelling Brain Function – The World of Attractor Neural Networks by D.e. another prototypical systems of many strongly interacting components. can be described in a rather straightforward way. Chemistry essentially would know only one element and Life would not exist. The Aufbau principle. i. Exercise 9. Transformation to a New Set of Creation and Annihilation Operators Our aim is to represent the Hamiltonian (9. 1989) . 2.258 Many–Particle Systems that today these concepts are providing insight into the functioning of biological brains. {|m). water-ﬁlled jar all electrons of water turn their fermion nature into a boson nature. . (9. where S |m) = r=1 Urm |r . In such a representation Ho is diagonal and the stationary states can be stated readily.

behave exactly like the operators c† and crσ behave for states |ΨF (N ) .101) of states |r . (9.105) where ˜ ˆ tmn = (m|t|n) = S ∗ trs Urm Usn = r.108) These operators describe the creation and annihilation operators of fermions in states |n) which are linear combinations (9. The unitary property (9.108) and (9. [d† . (9.4: Independent-Particle Models allows one to express conversely |r in terms of |m) S 259 |r = m=1 ∗ Urm |m) ..108) obey the same anti-commutation properties (9. 9.109) The operators (9.104)] S S ∗ Urm d† mσ m=1 c† rσ = . rσ namely. hence.and two-particle operators hold in an analogous way for d† and dnσ . of (9. (9.9.101) are occupied. using (9.46) as c† and crσ . The l.s. ∗ ∗ Urm Urn .e. they are Fermion creation and annihilation operators. d† ]+ = sσ nσ r.s r The unitarity property r ∗ Urm Urn = δmn yields immediately identity (??).106) which together with (9. accordingly. (9.111).100) yields ˜ Ho = tmn d† dnσ mσ where d† nσ = r=1 S S (9.46). the expressions drived above for the matrix elements of one. .f.110) can be demonstrated similarly. the operators d† and dnσ .103) allows one to express [c. c† ]+ = δσσ [dmσ . dnσ = r=1 ∗ Urn crσ (9.111) and. d† ]+ = δmn δσσ nσ (9. expand S trs = mn ∗ ˜ tmn Urm Usn (9. In a basis of states N d† j σj |0 n j=1 (9. nσ We demonstrate here the anti-commutation property (9. dnσ ]+ = 0 .104) One can.h.111) can be written.107) Urn c† rσ . csσ = n=1 Usn dnσ . the anti-commutation properties (9. for nσ rσ example. of N -Fermion states in which single particle states |nj ) as deﬁned in (9.113) Urm Usn [crσ . [dmσ .s=1 ˆ U † tU mn (9. i.112) .45. d† ]+ = 0 (9.110) mσ nσ [dmσ .

e. 2. σj ) = (nk . S (9.106) diagonal. Before proceeding we like to address the issue how the new representation. We apply for this purpose (9.114) together with (9. Vn = (U1n . .e. ) = j=1 nj . i. The 2N fermion state 2N d† j σj |0 n j=1 where (nj . . .117) are eigenstates of (9. Eigenstates of (9. σ2N . . the so-called ground state.. . A non-degenerate orbital state which is occupied by two fermions. . the associated ( properly normalized) eigenvectors Vn are the columns of (Urn ). a simple matter to nσ state many–particle states in the new representation. .. is obtained. i. USn ).118) Ground State In case of an ordering m < n for m < n the state of lowest energy. U2n .106) is equivalent to S trs Usn = s=1 n Urn ∀r. . σ1 . 2.e. . The condition (9. − 1 . 1 and |β = | 1 .115) can be stated immediately since any many–particle wave function in the occupation number representation is suitable. σ2 . It is.. S are T real.106) and the unitary property of Urn . r = 1. The corresponding eigenvalues n . − 1 is called a closed 2 2 2 2 . This equation poses the eigenvalue problem for the hermitian matrix (trs ). . 2 2 2 2 i.116) which follows from (9. transformation matrices Urn and energy values n . ..e. (9.260 Diagonal Representation Many–Particle Systems The transformation (9. i.115) ˜ and involves solely occupation number operators N nσ = d† dnσ .57) to the case of 2N particles. hence. ˜ tmn = In this case Ho has the desired form S n δmn . σk ) for j = k (9. 1 as well as a fermion in a spin state | 1 .119) In this state the N lowest orbital eigenstates of trs are occupied each by an electron with spin |α = | 1 . .114) Ho = n=1 † n dnσ dnσ (9. is N |Φo = j=1 d† α d† β |0 . a fermion in a spin state | 1 . j j (9. (9. n = 1. n2N .115) with eigenvalues 2N E(n1 . n2 .101) gives us the freedom to choose the matrix (9.

α = 2 |β. We will demonstrate now that this property endows the ground state with total spin zero. In case of m = n it gives a contribution ‘1’ for each closed shell which cancels the contribution of the ﬁrst term. α N +1 − N .120) where the occupation number operators refer to the single-particle states |m). Similarly. β = 2 |α. ˆ one can conclude that |Φo is an eigenstate of S 2 with eigenvalue zero.e. the states diﬀer in their spin character. β |β. i. The action of the operator (9. However. Alltogether.120).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. 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. 1 S 4 m. namely.n=1 m. There are four such states.n=1 † † if either |m) or |n) are closed shells. both with eigenvalues 2 ˆ S 2 |α.97). (9. i. the third term − S m. energy above the ground state. ˆ S 2 ||β.n=1 d† d† dmβ dnα mα nβ (9.. β ˆ are eigenstates of S 2 given in (9. β = d† +1α dN α |Φo N = d† +1α dN β |Φo N = d† +1β dN α |Φo N = d† +1β dN β |Φo N (9. Excited States We want to construct now excited states for the system governed by the independent-particle Hamiltonian (9. 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).4: Independent-Particle Models 261 shell.n=1 d† d† dmβ dnα mα nβ (9.e. β .100). 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. is a singlet state. The ground state has only closed shells. α |α. The two states |α. |α. Obviously. α |β. of ∆E = and |β.n=1 m.n=N (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 particularly simple for closed shells. α .. ˆ In order to investigate the total spin of |Φo we apply to this state the total spin operator S 2 as given by (9.9.124) .122) This property can be derived as follows: The two states are obviously diagonal with respect to the ˆ following contribution to S 2 1ˆ 1 ˆ Σ1 = N + 2 4 N −1 N −1 ˆ ˆ ˆ ˆ (Nmα − Nmβ )(Nnα − Nnβ ) − m.n=1 cmα cnβ cmβ cnα does not give a contribution if m = n and either |m) or |n) are closed shells.121) All four states have the same excitation energy.97). namely.

. α ± |β. one with eigenvalue zero (singlet) and one with eigenvalue ‘2’ (triplet). conclude that (9. c† +1 σ = c† σ 1 2N (9.2: Construct four operators O ˆ O ˆ such that O ˆ S3 . . 2.129) The symmetry of the Hamiltonian suggests to choose a new representation deﬁned through the operators (r = 1.σ = (9. We identify the states |2N + 1 = |1 to avoid cumbersome summation limits. .σ dN σ N σ. β both with eigenvalues 1. .e. 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 (9.130) . do not need to be considered since they ˆ give vanishing contributions. The system is described by the Hamiltonian 2N ˆ Ho = − t n=1 σ c† σ cn σ + c† σ cn+1 σ n n+1 . β . The system is assumed to posses a 2N –fold symmetry axis and interactions connect only orbital states |r → |r ± 1 . . ˆ An eigenvalue ‘2’ of the spin operator S 2 identiﬁes the states |α. ˆ Exercise 9. . β are not eigenstates of S 2 . hence.128) The cyclic property of the system is expressed through c2N +1 σ = c1 σ . |N + 3 . α .100) outlined above we consider a system of 2N electrons which move in a set of atomic orbitals |r . The action of Σ N +1 ˆ Σ3 |α. 2. i. β qualify as eigenstates. The same result holds for |β.127) ˆ are triplet and singlet states. α as triplet states. also eigenstates of Σ2 . however.262 and ˆ Σ3 = − m. One can. vanish.n=N d† d† dmβ dnα d† +1α dN β |Φo = 0 mα nβ N (9. mα nβ (9. . β =− m. 2N which are located on a ring.122) is correct. . m |Φo m m m γσ. 2N ) drσ = √ 1 2N 2N e−irnπ/N cnσ n=1 (9. for example. for |α. etc. β and |β. the linear combinations 1/2(|α. (9.. .125) ˆ Contributions to S 2 acting on states |N + 2 . |β.126) which follows from the occurence of squares of fermion operators which. ˆ 3 on the two states produces . α and |β.4. of course. r = 1. ˆ The remaining states |α.n=N Many–Particle Systems N +1 d† d† dmβ dnα .σ d† +1. α are. of course. The two states |α.

4N. one obtains the new representation of Ho 2N ˆ Ho = −t r. . (9.139) −2t cos r=1 σ rπ N d† drσ .133) Application to the l.134) which.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 . One can then express[.109)] ¸ cnσ = √ 1 2N 2N eirnπ/N drσ r=1 2N (9. rσ (9.132) an identity which can be derived employing the well-known expression for a ﬁnite geometric series m as = a s=1 1 − am .136) ˆ and. 9. . . 2N. eisπ/N 1 − e2N isπ/N 1 − eisπ/N (9.131) The (9. (9.9.s=1 σ ˜ rσ trs d† dsσ (9.132) gives for r = 0. 1−a (9. n=1 (9. of (9.130. thereby. indeed.138) ˜ trs = from which follows ˆ Ho = eirπ/N + e−irπ/N 2N δrs (9.h. vanishes for integer s.135) c† nσ 1 = √ 2N e−irnπ/N d† rσ r=1 (9.f.132) one can conclude 2N ei(s−r)nπ/N + eisπ/N n=1 1 2N 2N ei(r−s)nπ/N . 2N (9.137) where 1 ˜ trs = e−irπ/N 2N Using (9. . .s.140) The Hamiltonian is now diagonal and the construction outlined above can be applied.

e. if one can choose the singleparticle states such that the residual perturbations are small. i. A second reason why independent-particle models can be successful is that the pair interaction for electrons is the slowly decaying Coulomb interaction. These states do not altogether neglect the pair interactions. i. in states which are constructed according to the ‘Aufbauprinzip’ in a manner which neglects the mutual repulsion between electrons. The properties of molecules can be understood largely in terms of models which place electrons into so-called molecular orbitals in which the electrons move as if they were not interacting with other electrons..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. Since independent-particle behaviour is so universal one may wonder why the presence of particleparticle interactions. We will present in this and the following section a method which constructs such optimal single-fermion states. .4. . assuming that electrons move independently from each other in so-called bands. does not spoil it. 2. The regularities of the elements can be rationalized in terms of electrons moving in hydrogen atom-type orbitals. e.e. i. Of course. A similar principle accounts approximately for properties of nuclei as described by the nuclear shell model...264 Many–Particle Systems Exercise 9. those with energy above that of the (energetically) highest occupied single-particle state. valence bands or conduction bands.. if one places fermions into an independent-particle ground state of the type (9. . n ± 1) orbitals and Hamiltonian 2N −1 ˆ Ho = − t n=1 σ c† σ cn σ + c† σ cn+1 σ n n+1 . i..3: Construct the ground state and the lowest energy excited state(s) for a system of 2N electrons moving in a linear chain of single-electron orbitals |n . in particular. the everpresent Coulomb repulsion between electrons. the fermion nature restricts the possibility for perturbations on the system. n = 1. 2N with interactions between neighbouring (n. Many electronic properties of solids can be understood in a similar way.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. There are two reasons why this is so: ﬁrst.e. properties of many-fermion properties deviant from such simple description is then of as much interest as independent fermion descriptions are useful. . and that independent-particle behaviour surfaces mainly because it applies well to the ground state.g. The most famous example is the periodic table of the elements.e. The reason is mainly connected with the fact that most many-fermion systems are found in their ground state or removed from it through low energy excitations. Hence. The slow decay of the interaction makes the motion of any one electron rather independent of the exact position of most other electrons. one expects that mean-ﬁeld descriptions should be rather suﬃcient. rather they assume that each particle experiences the average interaction due to the fermions frozen into the ground state.e. 9.. (9.141) State the excitation energy ∆E. i.

. If one restricts m = 1. u c† c† cuσ ctσ v (9. . . r = 1. . t.144) deﬁned as in (9. naturally. ˜ (9. m = 1.t. . In our formulas below we will adopt strictly the convention that indices r.5: Self-Consistent Field Theory 265 In this section we will consider systems of 2N fermions described by the spin-independent Hamiltonian S S 1 ˆ rσ H = r.u=1 σ.108). 2. S is a unitary matrix.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. . n. .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.t.145). . s. m = 1. . 2. We will adopt a second set of single-particle states {|m . 2. S} the elements of which will be labelled by indices ˜ m.142) over this reference state to turn the contribution into an eﬀective one-particle operator. .146) . N . The states |m are connected with the states |r through (m = 1. Mean-Field Potential The mean ﬁeld for the reference state |Φo (U) is deﬁned in a straightforward way by averaging the two-particle contribution to (9. |r = m=1 ∗ Urm |m .σ and derive a single-particle operator which approximates this Hamiltonian. S) ˜ S S |m = ˜ r=1 Urm |r . j j (9. 2. . . . s|ˆ|t. q. S on which (9. u refer to the initial basis of single-particle states |r . p. .142) is based.u=1 σ.143) with r. 2. is not unitary.s=1 σ r. U = ( Urm ) as deﬁned in (9. .9. u v r. This is done as follows: ˆ Vmf (|Φo (U) ) = 1 2 S r.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. s|ˆ|t.142) r|t|s c† csσ + rσ sσ 2 r. as in in (9. The states |m serve to deﬁne a reference state of the system ˜ N |Φo (U) = j=1 d† α d† β |0 . . the corresponding U = ( Urm ) forms an S × N matrix and.

146) accordingly.σ .σ S ctσ + m.266 Here ··· denotes the average O = Φo (U)| O |Φo (U) .σ S dmσ dnσ + r.150) r.σ S d† dnσ cuσ mσ − r.149) The remaining matrix elements appearing in (9. u c† d† dnσ cuσ ˜ vn rσ mσ r. s|ˆ|t.s.σ S d† dnσ mσ ctσ (9. n c† ˜ v ˜ sσ m.u=1 σ.t. n c† c† v ˜ ˜ rσ sσ r. u c† ˜ vn sσ m. Many–Particle Systems (9.s.s. s|ˆ|˜ . m. m|ˆ|t.n=1 σ. 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. d† d† mσ nσ dmσ dnσ d† dnσ mσ = = = = and. for example. m| v |t.148) are obtained in a similar way by transforming only two of the four single-particle states into the new representation. m|ˆ|t.t=1 σ.u=1 σ.148) one obtains. n|ˆ|t.t=1 σ. u ˜ ˜v m. n c† ctσ δσ σ ˜ v ˜ rσ m≤N r.t. m|ˆ|˜ .u ∗ r. s|ˆ|m.u=1 σ. s| v t.147) Since the reference state |Φo (U) is deﬁned in terms of the single-particle states |m it is preferable ˜ to switch the representation of (9.σ S 1 − 2 where. u Usm Uun . n c† d† dnσ ˜ v ˜ rσ mσ r.148) r.n.m. by means of the rules (9.m.σ S d† d† mσ nσ cuσ ctσ + r. For the averages ··· in (9. n = ˜ ˆ ˜ s.n.n=1 σ.n.m. ˆ (9.84).n. ˆ Vmf (|Φo (U) ) = 1 2 S Φo (U)|d† d† |Φo (U) = 0 mσ nσ Φo (U)|dmσ dnσ |Φo (U) = 0 Φo (U)|d† dnσ mσ δmn δσσ 0 |Φo (U) m ≤ N m > N (9. hence.

s|ˆ|t.s σ where r| vmf (|Φo (U) ) |s = ˆ t.s=1 σ ˆ r|t|s + r| vmf (|Φo (U) ) |s ˆ c† csσ rσ (9. u = s. We will address now a proper choice of the reference state. r| v |u.. s ) ˆ ˆ m≤N ∗ Utm Uum (9.. u − r. m=1. Exploiting the symmetry r. s| v |t.156) which deﬁnes the reference state |Φo (U) .t=1 σ (9.2.153) rσ r. u c† cuσ ˜ vn rσ m≤N r. t| v |s. u c† cuσ ˜ vn rσ (9. m|ˆ|t.149) for the matrix elements in the mixed representation one can state the right hand side explicitly in terms of Urm . ˜ v ˜ sσ m≤N s.u=1 σ S − m. as introduced here. 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. m|ˆ|˜ ..u ( 2 r. The one-particle operator (9.2. t| v |u.152) can then be written in the form ˆ Vmf (|Φo (U) ) = r| vmf (|Φo (U) ) |s c† csσ ˆ (9.S.N (9.142) by Hmf (U) deﬁned through S ˆ Hmf (U) = r. u c† cuσ δσσ ˜ vn sσ m≤N s.151) Carrying out the sum σ δσσ = 2 leads to a factor 2 for the ﬁrst two terms..m.152) By means of expression (9. leaves open the question how the reference state should be chosen.154) ˆ The mean ﬁeld approximation replaces then the Hamiltonian (9..9..m. At this point this state is completely arbitrary.5: Self-Consistent Field Theory 1 2 1 2 1 2 S 267 m.σ r.σ S r.u=1 σ. n c† ctσ .σ S + − r.t=1 σ.u=1 σ. s|ˆ|˜ . n c† ctσ δσ σ ˜ v ˜ rσ − r. m|ˆ|˜ . The mean ﬁeld approximation. .

.155) following the method outlined in Section 9. i..119. 2. 2.e.145). U(1) is the S ×N –matrix (1) of the ﬁrst N column vectors of Urm . however. the independent-particle nature stemming from the fact that the functional form of the ground state.4.. . . . The state thus determined is referred to as the self-consistent independent-particle ground state. . SCF-Algorithm.142).4 and deﬁning this as the new reference state. .158) where the labels m are ordered to obey the condition (1) m < (1) n for m < n . i.157) and determines the associated diagonal representation deﬁned through S (1) ˆ r|t|s Usm = s=1 (1) (1) m Urm . S . . SCF-Algorithm. determining in a step (2) the corresponding mean ﬁeld Hamiltonian (9. (9. m = 1. . . .e. assuming in an intial step (1) a properly chosen reference state. 9. r = 1. i. steps (2) and (3) are being repeated M times until the procedure converges. M = 1.: Determine Mean Field Hamiltonian In this step the mean ﬁeld Hamiltonian S ˆ (M ) Hmf = r.e. . (9.160) . Let us state now the construction of the self-consistent ﬁeld ground state in more detail.268 Many–Particle Systems 9.155). r = 1. can achieve this goal only iteratively. Step 2M. . 2. is the lowest energy independent-particle ground state one can construct.159) One deﬁnes the reference state |Φo (U(1) ) using the deﬁnition (9. often also referred to as the Hartree-Fock approximation. the reference state resulting from step 2M + 1 (within numerical error) is equal to the reference state resulting from step 2M − 1.s=1 σ ˆ r|hmf |s c† csσ rσ (M ) (9. (9.145). The procedure determines the reference state as the ground state of the independent-particle Hamiltonian (9.6 Self-Consistent Field Algorithm The Self-Consistent Field (SCF) approximation. .s=1 σ S ˆ rσ r|t|s c† csσ (9. S . 2. under conditions which often are realized. This procedure.s=1 σ S ˆ r|t|s + r| vmf |Φo (U(M ) ) ˆ |s c† csσ rσ = r. and obtaining in a step (3) the ground state of this one-particle Hamiltonian according to the construction in Section 9. We will argue below that the self-consistent ﬁeld ground state. 2. is exact only for an independent-particle Hamiltonian. m = 1. S . S. is based on the mean ﬁeld approach and provides an algorithm to construct a reference state |Φo (U) for a 2N fermion system described by a spin-independent Hamiltonian (9. Step 1: Choosing an Initial Reference State One deﬁnes the Hamiltonian ˆ (1) Hmf = r. . .

166) ˆ (d) Deﬁne a similar two-particle operator Prstu .e. 2. u − r. . (9. . . 2.162) is solved adopting an ordering of the labels m which obeys the condition (M +1) m < (M +1) n for m < n .161) is calculated. 2. S do not exceed a preset threshold. .. (c) Show that any spin-independent one-particle operator S ˆ F = r. .163) The result yields the reference state |Φo (U(M +1) ) using the deﬁnition (9.6: Self-Consistent Field Algorithm is evaluated. . r = 1. S (9. Step 2M + 1. etc.164) ˆ (a) Determine the expectation value of Prs for the state as deﬁned in (9.: Diagonalize Mean Field Hamiltonian In this step the eigenvalue problem S (M r|hmf |s Usm +1) = s=1 (M ) (M +1) (M +1) Urm m . 2. U(M +1) is the (M +1) S × N –matrix of the ﬁrst N column vectors of Urm . i. The procedure is continued until it ∗(M ) (M ) ∗(M +1) (M +1) Uum − N Urm Uum | is detected that the one-particle density diﬀerences | N Urm m=1 m=1 for r. (9. SCF-Algorithm.. m = 1. SCF-Algorithm: Continuation and Convergence Condition When step 2M + 1 is completed. S. 2.e. M = 1. ˆ (b) How can the expectation values of the operator Prr be interpreted.s=1 σ ˆ r|f |s c† csσ rσ (9.167) . S . (e) Consider the creation operators N † gqσ = m=1 Vmq d† mσ (9. . t| v |s. indicating that the state converged.9. .s=1 (9. t| v |u.145). r = 1. σ p = δtr δsu δσσ . σ|ˆrs |u.165) can be written ˆ F = S ˆ ˆ r|f |s Prs r.u=1 ( 2 r. .6. m = 1. ˆ Exercise 9. . i. 2 .1: Let Prs be the one-particle operator with the generic operator prs = |r s|. i. the S × S–matrix S (M ) r|hmf |s N 269 ˆ = r|t|s + t. . . s ) ˆ ˆ m=1 Utm ∗(M ) (M Uum ) (9. . . . . . step 2M + 2 is carried out.e. ˆ t.145). s = 1. S .

. i.2: Determine the SCF ground state and its energy expectation value for a system of two particles described through the Hamiltonian (S = 2) H = σ ( c† c1σ + c† c2σ ) − t ( c† c2σ + c† c1σ ) 1σ 2σ 1σ 2σ ( c† c1σ c† c2σ 1σ 2σ σ.e.172) is obeyed. S denotes the initial single-particle states and Urm is the unitary S × S– transformation matrix obtained in (9.170) (SCF ) Here |r .173) . We begin by summarizing the result of the SCF algorithm. .163).169) 9.163). .145) and (9. Equation (9.171) implies that for the mean ﬁeld Hamiltonian holds S ˆ Hmf (U(SCF ) ) = m=1 σ (SCF ) † dmσ dmσ m .145).171) where the convention (9.168) ˆ ˆ has the same expectation values for Prs and Prstu as the reference state (9. We denote the representation which results from the SCF algorithm after its convergence as follows: S |m = ˜ r=1 (SCF Urm ) |r .. 2.162. (9. (9. r = 1.168) through a physical observation? Exercise 9.270 Many–Particle Systems which are connected with d† through a unitary N × N –matrix (Vqm ). (f) Can one distinguish the states (9. (9.σ + 2v ( c† c1α c† c1β + c† c2α c† c2β ) + v 1α 2α 1β 2β + c† c2σ c† 2σ 1σ c1σ ) .7 Properties of the SCF Ground State We want to investigate now the properties of the SCF ground state. For this representation holds N m| t |˜ + ˜ ˆn m 2 mm | v |˜ m ˜ ˜ ˆ n˜ − mm | v |m n ˜˜ ˆ ˜ ˜ = (SCF ) δmn m (9. (SCF ) m < (SCF ) n for m < n (9. Note that this is not an mσ S × S–matrix! Show that the reference state deﬁned through N † † gqα gqβ |0 q=1 (9.9.6.

173) holds N ˆ SCF Hmf (U(SCF ) ) SCF = 2 r=1 (SCF ) m .177) 2 mm | v |mm ˜˜ ˆ ˜˜ − mm | v |m m ˜˜ ˆ ˜ ˜ .171) in terms of matrix elements r|t|s .9. 2 m=1 m|t|m + ˜ ˆ˜ m.142).93) for the expectation values (diagonal elements) of the one-particle and two-particle part of the Hamiltonian (9. i. Employing expressions (9.142). exploiting the spinindependence of (9.171) yields ˆ SCF H SCF N = N (9. v (SCF ) Show that the resulting eigenvalue problem of the type (9.7.177).142). Finally.1: Reformulate (9.175) However. Exercise 9.4 that SCF is a total singlet state. . the mean-ﬁeld Hamiltonian is diagonal in the SCF representation.176) 2 mm | v |mm ˜˜ ˆ ˜˜ − mm | v |m m ˜˜ ˆ ˜ ˜ .146) counts a particle twice. one obtains ˆ SCF H SCF N = N (9.84) and (9. r.e. Within the mean ﬁeld approximation as described by (9. s|ˆ|t.174) The ﬁrst property we consider is the total spin character of SCF .m =1 The second term originates from the fact that in the mean ﬁeld approximation one assumes that each of the fermions is subject to an average pair interaction involving a 2N -particle reference state..2: Derive (9. mα mβ (9. Since the system under consideration has only 2N particles altogether.177). We like to determine next the energy expectation value of SCF .m =1 Comparision with (9.e. (SCF ) ˆ Exercise 9. 2 m=1 (SCF ) m − m. u and Urm . and once as a probe particle being subject to the mean pair interaction. ˆ SCF H SCF .. (9. Since this state is composed only of closed shells one can conclude following Section 9.7: Properties of the SCF Ground State 271 i. one can also determine the expectation value of SCF for the Hamiltonian (9. This over-counting leads to the correction term in (9.7. the SCF ground state is N SCF = m=1 d† d† |0 .162) is non-linear in Urm . once as a member of the ground state. the mean ﬁeld potential ˆ Vmf (|Φo (U) ) deﬁned in (9.

σ crσ + crσ cr+1. .σ crσ + crσ cr+1. is characterized through the parameters t. v represents the ‘on-site’ Coulomb repulsion. .128). approximation schemes like the self-consistent ﬁeld approximation play an important role. † ˆ Ho = −t c† .e.192) derived below is ﬁlled half. The operator c† (crσ ) creates (destroys) an electron with spin σ (σ = α. β for ‘up’ and ‘down’ rσ spins.178) and through the so-called ﬁlling factor n n= 2N N = .e. v (9. s|ˆ|u. describes the sites of a linear lattice. in the usual two–particle operator form. 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. rσ rσ r σσ (9. The ﬁrst term on the r.. the second term on ther.. referring to the fact that in this case the band of single particle energies (9. s|ˆ|t.178).8 Mean Field Theory for Macroscopic Systems The SCF algorithm has been developed in Sections 9. labeled r = −N0 + 1. r ∈ Z.h. in the two-dimensional copper oxide lattices of high temperature superconductors. N0 which are populated by 2N electrons (we choose an even number of electrons for convenience).178). i. t describes the coupling of state |r to states |r ± 1 at the two neighboring lattice sites. .s. Both N0 and N are macroscopically large numbers. In the present Section we want to adapt the algorithm to systems containing a macroscopically large number of fermions.178) Here the index r. respectively) in state |r at lattice site r.h.s.178) v contributes only in case that two electrons occupy the same lattice site.. of (9. v (9.σ rσ The potential energy term. real number.181) The Hubbard model. (9. u c† c† cuσ ctσ . S N0 (9. of (9.272 Many–Particle Systems 9.179) r+1.4–9.t. The case n = 1 is termed the half ﬁlled band case. t = v δru δst δrs .s. except in case of one-dimensional systems. We assume that there are altogether S = 2N0 lattice sites.e. Due to the lack of an exact solution in dimensions higher than one. as stated through Hamiltonian (9.180) V = rσ sσ 2 r. . t is assumed to be a positive. i. cannot be solved exactly.178). for example. We consider as an example the Hubbard model of an inﬁnite linear lattice and apply the model to describe magnetic instabilities in metals.σ + v 2 c† c† crσ crσ . The one-dimensional Hubbard model is described by the Hamiltonian ˆ H = −t rσ † c† r+1. reads 1 ˆ r. is identical to that of the independent–particle Hamiltonian H0 (9.u σσ where r.7 for ﬁnite systems. In spite of the simplicity of its Hamiltonian the Hubbard model. v of the Hubbard Hamiltonian (9.182) n can assume values 0 ≤ n ≤ 2. According to (9. i. .

Accordingly. For this purpose we apply the SCF theory presented in Section 9.185) ˆ According to (9. 9.181.184) and using (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. i.154). .178) can be determined by applying (9. This task has ˆ been solved already on page 262. 2N0 (9. rσ rσ (9. rσ (9. N0 (9. The self-consistent ﬁeld Hamiltonian Hmf which corresponds to the Hubbard Hamiltonian (9. the operator H0 can be re-written in the diagonal form N0 ˆ H0 = m=−N0 +1 σ m d† dmσ mσ (9. in fact.178) within the framework of the self-consistent ﬁeld theory. boundary eﬀects are assumed to be negligible.185) into (9.188) .6.174). we assume that the SCF ground state is given by (9. Accordingly.184) ˆ We are now ready to apply step 1 of the SCF algorithm and diagonalize term H0 .187) The energy levels are labeled such that 0 < ±1 < . by N −1 ||SCF = m=0 d† d† |0 . . < ±(N0 −1) < N0 holds as can be readily veriﬁed.179).8: Macroscopic Systems 273 Since we are dealing presently with a macroscopic system.140).186) where m = −2t cos mπ . SCF Ground State for the Hubbard Model We seek to determine the ground state of the Hubbard model stated by (9. we identify |r + N0 = |r and later let N0 go to inﬁnity.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 .. 9.9.e.144). We can now embark on step 2 of the SCF algorithm. Inserting (9. is Urm = √ 1 exp (irmπ/N0 ) .183) where the operators d† are related to the original creation operators c† through the unitary mσ rσ ˆ transformation U deﬁned in (9.155. one obtains for the mean ﬁeld Hamiltonian N −1 ˆ ˆ Hmf (U) = H0 + v r. Using (9. The unitary transformation which diagonalizes H0 .σ m=0 ∗ Urm Urm c† crσ . mα mβ (9.

. . as given in (9.274 Many–Particle Systems Expressing c† . . through the so-called dispersion law (k) = −2t cos k .191) where m is given in (9. dnσ according to (9.e. mσ (9.183) is not the only candidate for the mean ﬁeld ground state of the Hubbard model.187) form a quasi1 2 continuous energy band. The ground state of the system.190). denoted by F (see Fig.189) and. The energy Emf of this state can be calculated readily by using (9. The bottom of the energy band corresponds to (0) = −2t and the width of the energy band is 4t. The energy m can be expressed as a function of a continuous variable k ∈] − π. In other words. Indeed.192) This dispersion law is presented in Fig. When N0 becomes macroscopically large the single particle discrete energy levels (9. Indeed. for km = mπ/N0 holds km ∈ −π + N0 .1). N0 ˆ (2) ˆ Hmf = Hmf = m=−N0 +1 σ m + vn † dmσ dmσ 2 (9. is already diagonal.178).190) We have used in the latter expression the deﬁnition (9. . namely.1. Usually. the results of diﬀerent formulations agree qualitatively.182)].187). . −π + N0 . To demonstrate this we consider an alternative formulation of the mean ﬁeld approximation for Hamiltonian (9. called the Fermi energy.185) one can prove rσ mσ N0 c† crσ = rσ rσ m=−N0 +1 σ d† dmσ . there exist several ways of formulating mean ﬁeld approximations. in the limit N0 → ∞. the self–consistency condition is exactly met and the SCF algorithm converged after two steps. (9.e. holds k ≡ km ∈] − π.. but may diﬀer quantitatively. −π ≤k ≤π . can be obtained. hence. One can see that −2t ≤ (k) ≤ 2t holds. crσ through d† . (9. Alternative Description of the Mean Field Approximation The state (9. the ground state of the Hubbard model in the self-consistent ﬁeld approximation can be determined exactly.f.177) and mm |ˆ|mm = mm |ˆ|m m = ˜˜ v ˜˜ ˜˜ v ˜ ˜ which holds in the present case. 9. according to the Pauli principle. π]. π]. the electrons in the present description behave like independent particles with energies ¯m = m + vn . π and. even for one and the same Hamiltonian. i. 9. i. Apparently. 4 So far we have not exploited the fact that N0 and N are macroscopically large quantities. the state with the lowest possible energy.109) and using (9. Emf is then N −1 v 2N0 Emf = 2 m=0 vn vn −N = 2 m+ 2 2 N −1 m m=0 + vn 4 (9. ˆ The Hamiltonian Hmf . by ﬁlling up this energy band with electrons from the bottom of the band (which corresponds to k = 0) to a maximum energy.182) of the ﬁlling factor n [c.

e. (9.56). Nrσ = SCF ||Nrσ ||SCF . F = (±kF ) denotes the Fermi energy.1: The dispersion law (k) for independent electrons on an inﬁnite one dimensional lattice.178) in terms of the occupation number operaˆ tors Nrσ = c† crσ .. the operator Nrσ can be written ˆ Nrσ = Nrσ + δ (Nrσ ) .178) and performing the summation over the spin indices results in ˆ ˆ H = H0 + v r ˆ ˆ Nrα Nrβ . we have employed (9.e. the deviation of Nrσ from its mean value Nrσ for the corresponding reference state.45.46) 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 .195) ˆ ˆ ˆ and where δ (Nrσ ) = Nrσ − Nrσ describes the ﬂuctuations of Nrσ . i.194) ˆ For a given reference state. i. where Nrσ is the mean occupation number of the one particle state |rσ .8: Macroscopic Systems 275 Figure 9.. Inserting (9.193) where. let us express the interaction term in (9. rσ rσ rσ (9.183). 9. The anti-commutation properties (9.196) (9. The mean ﬁeld approximation neglects the . First. say the one given in (9. in the last term on the right hand side.9.193) into (9. (9.

197) and the corresponding mean ﬁeld Hamiltonian becomes ˆ ˆ Hmf = H0 + v r ˆ ˆ Nrα Nrβ + Nrβ Nrα − v r Nrα Nrβ . Therefore. non-vanishing magnetization.276 Many–Particle Systems ﬂuctuations to second order and the mean ﬁeld Hamiltonian is obtained from (9.191) for the ground state energy. i. (9. To determine a magnetic ground state we note ˆ Nrσ = r mm r ∗ Urm Urm d† dm σ = mσ mm ˆ δmm Nmσ = m ˆ Nmσ .146).198) Let us assume that the ground state of the Hubbard model. is given by (9. Note that the procedure employed in (9. 2N0 2 holds ˆ Hmf = = mσ mv ˆ H0 + 2 vn2 2 rσ nv † vn dmσ dmσ − N .183) .. For such state the mean occupation number nrσ is uniform at all lattice sites. Spin-Polarized Mean Field Ground State ˆ We want to consider now a ground state for Hmf deﬁned through Nrσ = nσ = const (9.198) as the one obtained by using the mean ﬁeld potential (9.194) by dropping all the terms which contain the ﬂuctuations to second order.200) allowing. In case nα = nβ the following construction will lead to a ground state which coincides with the non-magnetic ground state (9.195).199) which is identical to expression (9. In this approximation one can express ˆ ˆ Nrα Nrβ = ≈ = ( Nrα + δ (Nrα )) ( Nrβ + δ (Nrβ )) Nrα Nrβ + Nrα δ (Nrβ ) + Nrβ δ (Nrα ) ˆ ˆ Nrα Nrβ + Nrβ Nrα − Nrα Nrβ (9. but the mean number of particles with spin α can be diﬀerent from the mean number of particles with spin β. Because of Nrσ = c† crσ rσ N −1 = m. . separating the occupation number operator into a mean value plus ﬂuctuation and neglecting the ﬂuctuations in 2nd order. that nα and nβ assume diﬀerent values.m ∗ Urm Urm d† dm σ mσ = m=0 ∗ Urm Urm = n N = . however. consequently. in the mean ﬁeld approximation.183). m+ 2 2 c† crσ − 2N0 rσ (9.e. the ground state of the system can have a non-zero local spin and. yields essentially the same mean ﬁeld Hamiltonian (9.

205) Such minimization is realized through the method of Lagrangian multipliers. if nα > nβ then β (k) > α (k). Therefore. hence.e. According to this method one minimizes ˜ Emf = Emf + µ (n − nα − nβ ) . Since β (k) − α (k) = v(nα − nβ ). one can expect the system to lower its energy by assuming nα = nβ and. N0 ∼ 1023 . In this limit the discrete energy spectrum (9. nβ and µ. (9.2a. the smaller is σ (k) for a given k.202) This Hamiltonian describes a system of non-interacting electrons with a spin-dependent dispersion law (9. subject to the constraint n = nα + nβ . (9. For this purpose the sum in (9. The ground state (corresponding to the lowest possible energy) of the many electron system is obtained by ﬁlling these two energy sub–bands with electrons up to the same maximum energy value.201) ( m ˆ + vnβ ) Nmα + ( m ˆ + vnα ) Nmβ − 2N0 vnα nβ . a spin–polarized ground state.9. (9.8: Macroscopic Systems from which we obtain nσ = The mean ﬁeld Hamiltonian is then ˆ Hmf = m 277 1 2N0 Nmσ . The values of nα and nβ are determined by minimizing the energy density (9.206) ascertains that condition (9.203) mσ = m + vn−σ .208) . 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 ). 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. At this point we exploit the fact that our system is macroscopically large. The actual values of nσ (σ = α. 2π (9. (9. e.192). i.e. 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) . In the limit N0 → ∞ this sum can be expressed as an integral.. β) are determined by minimizing the energy density of the ground state Hmf 1 = [ mσ Nmσ ] − vnα nβ (9.207) where (k) is given by (9. E with respect to nα . m (9. The additional term in (9..σ with respect to nα and nβ .204) Emf ≡ 2N0 2N0 m..206) needs to be evaluated. In fact. as is shown schematically in Figure 9. the so-called Fermi energy F = µ (see below). such that the larger nσ .g.206) ˜mf is minimized where µ is the Lagrangian multiplier considered an independent variable.206). i.205) is met.203) is provided by the continuous function σ (k) = (k) + vn−σ .

Since the equation f (k) ≡ 2t cos k + = 0 has two simple roots. unit energy and a given spin orientation of the particle.212) where xi are the simple roots of the function f (x). gives the the number of available one particle states per site. 2π (9. For a given dispersion law (k) the density of states ρ( ) can be calculated by employing equation (9. In our case (k) is given by (9. δ(k + π − arccos( /2t)) + δ(k − π + arccos( /2t)) (2t)2 − 2 .211) and using π dkδ(k ± (π − arccos( /2t))) = θ(2t − | |) . If the function f depends explicitly only on (k) one can state π −π dk f ( (k)) = 2π π ∞ ρ( )d f ( ) −∞ (9.210) Here δ(x) is the Dirac–delta function and ρ( ).2: (a) Relative position of the two energy sub–bands α (k) and β (k) for ∆ = nα − nβ > 0.278 Many–Particle Systems (a) (b) Figure 9. holds π ρ( ) = −π dk δ( + 2t cos k) .211) In order to calculate the integral we recall the following property of the Dirac–delta function δ [f (x)] = i δ(x − xi ) . namely k1.210).192) and. Inserting this last result into (9. |f (xi )| (9.209) where ρ( ) = −π dk δ ( − (k)) . and noting |f (k)| = |2t sin k| = 2t one obtains δ( − (k)) = 1 − cos2 k = (2t)2 − 2 . which is usually called the density of states. The shaded areas represent the ﬁlled portions of these bands by electrons.2 = ±(π − arccos( /2t)). 2π (9. −π . (b) Graphical deﬁnition of the limiting energies α and β . therefore.

in principle.218) read ˜ ∂ Emf = ∂ α ˜ ∂ Emf = ∂ β and α + vnβ − µ = 0 .216) allow one.216) Using (9.f.213) can simply be replaced by 1.213) and carrying out the resulting integral yields nσ = 1 arcsin( σ /2t) .f. (9. (9. if we assume that is restricted to the interval ] − 2t. the necessary ˜ Since Emf [c.221) ∂µ Equations (9. of t. (9.. + vnα − µ = 0 . F = α + vnβ = β + vnα .217) and µ. 2t[ then the θ function in (9. v and n. One obtains ˜ Emf = α β ρ( )d + −2t −2t ρ( )d + vnα nβ + µ(n − nα − nβ ) .219. namely.215) nα and nβ in (9.8: Macroscopic Systems 279 where θ(x) is the step function. ∂ β ˜ ∂ Emf =0. π α. 9.e. i. Fig.2]. ∂ α ˜ ∂ Emf =0. ∂µ (9.218) The derivatives can be readily determined and the conditions (9.206)] depends only on continuous quantities. (9. to determine the unknown quantities α . µ.219–9.e.222) .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.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. θ(x) = 1 if x > 0 and θ(x) = 0 if x < 0.201). We can employ the results obtained to express the ground state energy density (9. i. β . 9.. on conditions for a ground state of minimum energy are ˜ ∂ Emf =0. 2 (9. outside the energy band (k).e.204) through an integral expression. (9. Replacing again the sum by an integral over energy one can write σ nσ = −2t ρ( )d (9.220) µ can be readily obtained µ= 1 ( 2 α + β) + vn .221) and (9.214) are given by the sum (9. nα and nβ as a function of the parameters of the Hubbard model. From (9.219) (9. (9.9. β (9. i. one arrives at the following expression of the density of states ρ( ) = θ(2t − | |) π (2t)2 − 2 .. namely.

(9. First. f (∆) converges to a maximum value as ∆ → ∆max = n.229) This cindition is known as the Stoner criterion and gives the critical value of v which determines the onset of the ferromagnetic long range order. from (9. if. We assume in the following discussion that the implicit ∆ dependence of β in f (∆) can be neglected..222) and (9.f. the slope of f (∆) at the origin is larger than 1 [c. (9. (9. One can express ∆ as a function of α and β in two diﬀerent ways.e. One still needs to determine α and β .217) and (9.215) reveals that µ is indeed equal to the Fermi energy F (see also Figure 9. through the condition [c.226) one can obtain ∆ by solving (cf. as reﬂected by the expression d[f (∆)] = vρ( d∆ β + ∆) = π (2t)2 v − (v∆ + 2 β) >0. non trivial.223 one obtains ∆= and second. (9.223) follows α α − v β .227) can be solved graphically.227). and only if. From the deﬁnition (9. If the slope of f (∆) at the origin is less than 1.3.f. For this purpose one plots f (∆) versus ∆ as is shown schematically in Figure 9. so called ferromagnetic solution. from (9.229) has a wider range of validity than one can infer from the present analysis of the Hubbard model.228) Since the total number of states per site is ﬁnite.. (9.219–9.225) Deﬁning the function v∆ f (∆) = 0 ρ( β + )d .227) will have a second. Accordingly.3]. (9.183) without magnetization. The slope of f (∆) at the origin depends on the Hubbard model parameter v. the Hubbard model has a ground state with ∆ = 0. For this purpose we consider the magnetization of the mean ﬁeld ground state ∆ ≡ nα − nβ . Equation (9.224) v∆ ∆= β ρ( )d = 0 ρ( β + )d .e. Indeed. i. For v above a critical value vc . condition (9.227) Since f (0) = 0 one can infer that ∆ = 0 is always a solution (the so called paramagnetic solution) of condition (9.280 Many–Particle Systems Comparision of (9.228)] vc ρ ( β ) = 1 . The Stoner criterion (9. the slope of f (∆) is positive for all ∆ ≥ 0. i.223) ∆ = 0 corresponds to the ground state (9. .226) follows that f (∆) is a monotonically increasing function of ∆.2b).227) has only the trivial solution ∆ = 0. (9. condition (9. Figure 9. a magnetic ground state. (9.226) ∆ = f (∆) . One can determine vc by equating the slope of f (∆) at the origin to 1.220) and (9. (9.

n).205) the following set of equations determining x. The corresponding 2t analytical expression provides the phase diagram of the ground state: when v approaches vc from above. (9. 2t As already mentioned. v (9. We like to determine ﬁnally the dependence of vc on the ﬁlling factor n. the plot of vc vs. (9. the ground state is characterized by the quantities (v. is presented in Figure 9.231) (9. i. β = 2t sin y .234) 2t 2 The resulting phase diagram of the ground state. From (9.233) Employing the approximation sin(y + δ) − sin y ≈ δ cos y for small δ.e.219. By parameterizing α and β α = 2t sin x . α. Accordingly.232) For a given 2t the values of x and y can be obtained numerically and other relevant quantities v mentioned above can be calculated.233) yield the desired expression vc π = π cos (n − 1) .220.8: Macroscopic Systems 281 Figure 9. β x+y x−y = = π(n − 1) 2t π (sin x − sin y) = π∆ .230) one obtains from (9.. n. the magnetization ∆ of the ground state vanishes and remains zero for values v < vc .232) states that x − y is proportional to ∆. 2 2 2 (9. Since ∆ vanishes near vc we set x = y + δ where δ is a small quantity. Equation (9.9.231) follows y= δ π π (n − 1) − ≈ (n − 1) .4. t. 9.3: Graphical solution of the mean ﬁeld equation ∆ = f (∆). y and. 9. hence. any .232) together with (9. (9.

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

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

284 Many–Particle Systems .

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

1). but in a setting of representation theory methods as applied in Secti.7) equation.4 to cover ﬁelds. This space is called the Minkowski space.e. 10. 0 1 2 3 (10. Rather than starting from (10. Note that the components of xµ all have the same dimension.1).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.5.1 will be extended in Sect. x 1 . namely the Klein–Gordon (Sect.. In such a space Lorentz transformations correspond to 4-dimensional rotations. The reason for this choice is that the transformations (10. The Group of Lorentz Transformations L = O(3. x3 ) = r describes the space coordinates. 10. the group of Lorentz transformations (10. namely that of length. x 3 ) holds (x0 )2 − (x1 )2 − (x2 )2 − (x3 )2 = (x )2 − (x )2 − (x )2 − (x )2 . x 2 . i.1 Natural Representation of the Lorentz Group In this Section we consider the natural representation of the Lorentz group L. assume new units for time such that the velocity of light c becomes c = 1. we will provide a more basic deﬁnition of the transformations. x2 . 10. We will.. Denoting by xµ the . however. i.e. Before introduco ing these general Lorentz–invariant ﬁeld equations we will provide in Sects. for the transformed space-time–cordinates x µ = (x 0 . 10. This choice implies dim(time) = dim(length). 10.286 Relativistic Quantum Mechanics in Sect. 10. t) and vectors with functions ψ(r. wave functions ψ(r. 10. i. 5 to the groups SO(3) and SU(2) of rotation transformations of space coordinates and of spin. 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. x3 ) (10. t) as components. Rather than following this avenue we will introduce Lorentz transformations within a setting which does not require real and imaginary coordinates.7 a heuristic derivation of the two most widely used and best known Lorentz–invariant ﬁeld equations.e. This will provide us with a general set of Lorentz–invariant equations which for various particles take the place of the Schr¨dinger equation. henceforth. We will ﬁnd that this deﬁnition will lead us back to the transformation law (10.4) invariant.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) leave the quantity s2 = (x0 )2 − (x1 )2 − (x2 )2 − (x3 )2 (10. x2 .1). The elements L ∈ L act on 4–dimensional vectors of position– and time–coordinates. Minkowski Space Historically.3) where x0 = ct describes the time coordinate and (x1 .5) and the Dirac (Sect. We will denote these vectors as follows def xµ = (x0 . x1 .

The second reason is that upper and lower positions allow us to accomodate the expression (10. 3: 4-vector: xµ . Aµ ν . First. Aµ ν B ν ρ = ν=0 Aµ ν B ν ρ . This will be explained further below. R). the latter notation will be used for diﬀerent quantities further below. 1. The Lorentz transformations form the subgroup of all matrices which leave the expression (10. Aµν .8) new old old The summation convention allows us to write (10. This condition can be written xµ gµν xν = x gµν x where µ ν (10.6) x µ = Lµ ν xν . xµ .9) 1 0 0 0 −1 0 0 ( gµν ) = 0 0 −1 0 = g . however. 4 × 4 tensor: Aµ ν . the Lorentz transformations constitute a linear transformation which we denote by 3 x µ = ν=0 Lµ ν xν . We will exploit this property below to determine the general form of the Lorentz transformations. (10. i. The Lorentz transformations are non-singular 4 × 4–matrices with real coeﬃcients.e.9) and (10. the latter set constituting a group. (10.10.6) yields Lµ ρ gµν Lν σ xρ xσ = gρσ xρ xσ . ν = 0. Since this holds for any xµ it must be true Lµ ρ gµν Lν σ = gρσ . (10.10) Combining condition (10. Aµν . the .7) The reason for the notation is two-fold. L ∈ GL(4..11) This condition speciﬁes the key property of Lorentz transformations. The subset of GL(4. summation over that index is assumed without explicitly noting it.e. i.6) Here Lµ ν are the elements of a 4 × 4–matrix representing the Lorentz transformation.5) into scalar products. 3 3 3 yµ xµ = new yµ xµ . ν=0 (10..1: Natural Representation of the Lorentz Group 287 coordinates in one reference frame and by x µ the coordinates in the other reference frame.6) allows us to introduce the so-called summation conventon: any time the same index appears in an upper and a lower position.12) (10. [ A more conventional notation would be Lµν . The upper index closer to ‘L’ denotes the ﬁrst index of the matrix and the lower index ν further away from ‘L’ denotes the second index.] The following possibilities exist for the positioning of the indices µ. Aµ ν xν = µ=0 new old Aµ ν xν . R). the notation in (10. 0 0 0 −1 (10. 2.5) invariant.

For L3 = L1 L2 holds LT g L3 = LT LT g L1 L2 = LT (LT gL1 ) L2 = LT g L2 = g . of O(3.. 10..18) (10. R). 1) = { L. 3 2 1 2 1 2 (10. L2 ∈ O(3.13) L = ( Lµ ν ) = 2 0 L 0 L2 1 L2 2 L3 3 . the inverse of L 1 L−1 (10.e. L−1 ∈ O(3.1). One can also show that if L ∈ O(3.14).19) i. R) is a group.20) Multiplying (10.20) results in the desired property (L−1 )T g L−1 = g .e.288 Relativistic Quantum Mechanics elements of which satisfy this condition. From this one can obtain using g2 = 1 1 (gLT g)L = 1 and. 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.1).1). .e. L ∈ GL(4.. a group.1) since it satisﬁes (10.15) (10. of (10.h.e. is called O(3. the associated inverse obeys (10.17) we note ﬁrst that the identity matrix 1 is an element of O(3.22) (10. LT gL = g } .1). i. L3 0 L3 1 L3 2 L3 3 Using the deﬁnition (10. In fact. = g LT g = 0 1 2 −L 2 L 2 L 2 L3 2 −L0 3 L1 3 L2 3 L3 3 (10.21) i. We consider 1 then L1 . This set is identical with the set of all Lorentz transformations L. (10. employing expressions (10. property (10. This stipulates that O(3. To demonstrate the group property of O(3.10) of g one can rewrite the invariance property (10.1) ⊂ GL(4.14) L0 0 −L1 0 −L2 0 −L3 0 −L0 1 L1 1 L2 1 L3 1 ..1).1). (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 (10.1) is.14) from the right by gLT and using (10. (10.14) holds for the inverse of L.17) one obtains (L−1 )T g L−1 = gLgggLT g = gLgLT g . i.12) LT gL = g .s. L3 ∈ O(3.16.16) The corresponding expression for (LT )−1 is obviously (LT )−1 = (L−1 )T = g L g . hence. in fact.14).

10. L0 0 ≤ −1} . L0 0 ≥ 1} .24) or since (L1 0 )2 + (L2 0 )2 + (L3 0 )2 ≥ 0 it holds (L0 0 )2 ≥ 1. Schwartz’s inequality yields j=1 2 3 3 3 j=1 A jB 0 ≤ 0 j A 0 2 j Bj 0 2 . L ∈ O(3.28) (10.12) the case ρ = 0. M1 = a M2 }. 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 . Using det AB = det A det B and det AT = det A yields (det L)2 = 1 or det L = ±1 . Considering in (10. σ = 0 yields L0 0 2 − L1 0 2 − L2 0 2 − L3 0 2 = 1. L ∈ O(3.14) which we employ in the formulation (10.34) j=1 j=1 . L↑ contains 1 1. det L = 1. From this we can conclude L0 0 ≥ 1 or L0 0 ≤ −1 . L0 0 ≥ 1} . L↑ . 1). = { L.23) and (10. Lorentz transformations in L↑ .14). A statement on this value can be made on account of property (10. B ∈ L+ holds C = AB ∈ L+ . Properties (10.26) (10. One can also 1 verify that one can write L↑ − L↓ + L↓ − = gL↑ + = = −L↑ . + + ↑ ↑ We can also demonstrate that for A.23) One can classify Lorentz transformations according to the value of the determinant into two distinct classes. For this purpose we consider ﬁrst the value of det L.33) where we used the deﬁnition aM = {M1 . L↓ are disjunct sets deﬁned as follows + + − − L↑ + L↓ + L↑ − L↓ − = { L. L0 0 (10.27) (10. hence. + = − L↑ g + (10.30) ≤ −1} . (10. + We start our investigation by demonstrating that L↑ forms a group.25) can be stated as follows: The set of all Lorentz transformations L is given as the union L = L↑ ∪ L↓ ∪ L↑ ∪ L↓ + + − − where L↑ . + except for the trivial factors g and −1 It is. The above shows that the set of proper Lorentz transformations L↑ allows one to generate all Lorentz transformations.31) (10. = { L.12). det L = −1. L ∈ O(3. ∃M2 . L↓ . 1). M2 ∈ M.14).1). there exist two other distinct classes.32) (10. Obviously. L ∈ O(3.e. = { L. entirely suitable to investigate ﬁrst only 1. 1). det L = −1.29) (10. A second class property follows from (10.1: Natural Representation of the Lorentz Group Classiﬁcation of Lorentz Transformations 289 We like to classify now the elements of L = O(3. 1).. (10. (10. + − gL↑ + = L↑ g . det L = 1. (10. It holds g ∈ L and −1 ∈ L as one can readily verify testing for property (10.25) i.

37) (10. For the inverse of + ↑ any L ∈ L+ holds (10.e. The next group property of L↑ to be demonstrated is the existence of the inverse. Using the above expression for C 0 0 one can state C 0 0 > 0. 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. + transformations in a small neighborhood of 1 which we parametrize by inﬁnitesimal parameters.1). In the following we + will consider solely this subgroup of SO(3. In fact. (10.35) One can conclude. we consider transformations Lµ ν = δ µ ν + µ ν ..38) For these transformations. . | 3 A0 j B j 0 | < A0 0 B 0 0 .39) where we have employed the matrix form T 1 deﬁned as in (10. + i.14) actually L0 0 ≥ 1 and det L = 1 must hold. since the group property of O(3. obviously.1) ascertains CT gC = g it must hold C 0 0 ≥ 1.40) and. 1 We will then employ the Lie group properties to generate all transformations in L↑ . This relationship shows (L−1 )0 0 = L0 0 . if we enforce (10.13). accordingly.290 Relativistic Quantum Mechanics From (10. µ ν small . Since A0 0 ≥ 1 and B 0 0 ≥ 1. 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 (10.16). i.34) provides then the estimate conclude from (10. + Accordingly.12) follows (B 0 0 )2 − 3 (B j 0 )2 = 1 or 3 (B j 0 )2 = (B 0 0 )2 − 1.. Similarly. Since the associative property holds for matrix multipli1 1 + ↑ cation we have veriﬁed that L+ is indeed a subgroup of SO(3. obviously j=1 A0 0 B 0 0 ≥ 1. Inﬁnitesimal Lorentz transformations The transformations in L↑ have the property that they are continously connected to the identity 1 1. 1 0 0 = ≥ 1 and. ρ ν (10. from which one can conclude L−1 ∈ L↑ .36) It holds. (10. orthochronous Lorentz transformations. Property (10.21) j=1 (A j 0 3 j=1 A0 j B j 0 ≤ 2 (A0 0 )2 − 1 (B 0 0 )2 − 1 < (A0 0 )2 (B 0 0 )2 . holds L0 0 > 0 and the value of the determinant is close to unity. one can j=1 j=1 3 0 )2 = (A0 )2 − 1.14) implies 1 + T g (1 + ) = g 1 1 (10. 1 ∈ L↑ . To order O( 2 ) holds g + g = 0. hence. L↑ is called the subgroup of proper.e. δ00 = 1 and It should be noted that according to our present deﬁnition holds δµν = gµρ δ δ11 = δ22 = δ33 = −1. therefore. (10.1).

k = 1. other ﬁve parameters zero) Kk = (wk = 1. 2. K ] The operators also obey J · K = J1 J1 + J2 J2 + J3 J3 = 0 = = k m Jm k m Jm 0 −1 0 0 0 0 0 0 (10. ϑ3 .44) This result allows us now to deﬁne six generators for the Lorentz transformations(k = 1.50) . 3) Jk = (ϑk = 1. other ﬁve parameters zero) . j. 2. J2 = 0 0 0 −1 0 0 0 0 0 0 0 0 0 0 0 −1 0 1 . J = 0 3 0 1 0 0 0 0 0 0 0 0 0 0 0 (10. w1 . k j j = 1. K ] [ Jk .10. 2. = − k m Km (10. w2 . ϑ2 .42) This relationship implies µ 0 j µ j = 0 = j 0 .47) 0 −1 0 0 0 0 −1 0 0 −1 0 0 0 0 0 0 0 0 K1 = .46) 0 0 0 0 .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 .1: Natural Representation of the Lorentz Group Using (10.49) . 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 . (10.45) (10.48) (10.15) one can conclude T 291 = −g g (10. J ] [ Kk . w3 ) = −w2 ϑ3 0 −ϑ1 −w3 −ϑ2 ϑ1 0 (10. K2 = . The generators are explicitly 0 0 0 0 0 0 0 0 J1 = 0 0 0 −1 0 0 1 0 (10. 3 (10. 3 .43) k = − Inspection shows that the matrix has 6 independent elements and can be written 0 −w1 −w2 −w3 −w1 0 −ϑ3 ϑ2 . (ϑ1 .

k = 1.49. In analogy to equation (5. however.292 as can be readily veriﬁed.50). that the transformations (10. w) = exp ϑ · J + w · K . A boost in the x1 –direction is L = exp(w1 K1 ).52) where the 3 × 3–matrices Lk are the generators of SO(3) deﬁned in Chapter 5. Kk . 0 1 n (10.49) deﬁne the Lie algebra associated with the Lie group L↑ .. 2. 3 (10.54) which. (10.e. w = 0) = 0 0 0 R(ϑ) . constitutes an overcomplete parametrization of the rotations (see Chapter 5). ϑ. 10. 2. We consider now the Lorentz transformations for ϑ = 0 which are referred to as ‘boosts’. k = 1. 3 where we have deﬁned ϑ · J = k=1 ϑk Jk and w · K = following the algebra in Chapter 5. and using the relationship Jk = 0 0 0 Lk (10. L(ϑ. To determine the explicit form of this transformation we evaluate the exponential operator by Taylor expansion.56) = 1 0 0 1 = 1 1 (10.1: Demonstrate the commutation relationships (10. (10.57) . For the parameters ϑk of the Lorentz transformations holds obviously ϑk ∈ [0.55) L 0 0 0 0 exp (w1 K1 ) = exp Using the idempotence property 0 −1 −1 0 2 0 0 1 0 (10. Exercise 7. Relativistic Quantum Mechanics The commutation relationships (10. Following the treatment of the rotation group SO(3) one can express the elements of L↑ through + the exponential operators L(ϑ. i. + The commutation relationships imply that the algebra of the generators Jk .51) for w = 0 correspond to rotations of the spatial coordinates.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 .53) Here R(ϑ) are the 3 × 3–rotation matrices constructed in Chapter 5. w ∈ R3 3 k=1 wk Kk . 2π[ .51) One can readily show. 3 is closed.

10.6.59) v1 = These two equations yield cosh w1 = 1/ and (10. Correspondingly. 10.3. we can state vk ∈ ] − 1.1: 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 .63) which agrees with (10.59) cosh w1 Using cosh2 w1 − sinh2 w1 = 1 one can identify sinhw1 = sinh2 w1 + 1.51) the explicit transformation for space–time–coordinates is then x1 = x1 − v1 t 1 − 2 v1 . however.1) of the Lorentz transformations is obtained through the parameter change sinh w1 v1 = = tanh w1 (10.1). Accordingly.59) follows.59) can be written 1 2 1 − v1 √−v1 2 1 − v1 2 1 − v1 .61) √ √−v1 √ exp (w1 K1 ) = 2 1 − v1 1 2 1 − v1 0 0 0 0 0 0 0 0 1 0 0 1 (10.62) According to (10. in fact. cosh2 w1 − 1 and coshw1 = cosh2 w1 − 1 = cosh w1 sinh w1 sinh2 w1 + 1 . From (10. This property is. for wk wk ∈ ] − ∞. x3 = x3 (10.10.59) for the case k = 1 corresponds to the relative velocity of two frames of reference. (10.58) = cosh w1 1 + sinh w1 1 0 −1 −1 0 The conventional form (10. . t = t − v1 x1 2 1 − v1 . vk deﬁned in (10. ∞[ .60) sinh w1 = v1 / 2 1 − v1 . (10. x2 = x2 . 1[. consistent with (10. We expect that vk can only assume values less than the velocity of light c which in the present units is c = 1. (10. one obtains from (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. The range of the parameters wk can now be speciﬁed. (10.56. 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.59).

p = √ 2 1−v 1 − v2 (10. 2 |ˆ1 . x3 )T . Another example is the potential 4-vector Aµ = (V. we have not yet deﬁned representations of Lorentz transformations beyond the ‘natural’ representation acting in the 4–dimensional space of position– and time–coordinates. t)|2 or the scalar product r1 · r2 of two particle positions. not in the so-called natural representation associated with the 3–dimensional Euclidean space E3 .294 Relativistic Quantum Mechanics 10. spherical harmonics Y m (ˆ). x2 . 4-Vectors The quantities with the transformation behaviour like that of the position–time vector xµ deﬁned in (10.69) where V (r. (10. namely. These quantites always come as four components (a0 . A) (10.4). The 4-vector character of J µ and of Aµ will be demonstrated further below. Nevertheless. scalars like r = x2 + x2 + x2 . of a system of charges. ﬁnally. Some of these r ˆ quantities were actually deﬁned with respect to representations of the rotation group in function spaces. not any physical property f ∈ R is a scalar.68) where ρ(r. r2 ) and. 4–Vectors and Tensors In this Section we deﬁne quantities according to their behaviour under Lorentz transformations. t) and A(r. However.67) the transformation behaviour of which we will demonstrate further below. J) (10.2 Scalars. Counterexamples are the energy. We have encountered examples in connection with rotational transformations. t) are the electrical and the vector potential of an electromagnetic ﬁeld. . t) are the charge density and the current density. vectors like r = (x1 . total angular momentum r 1 2 3 states of composite systems like Y m ( 1 . respectively. a3 )T and transform according to a µ = Lµ ν aν . Such quantities appear in the description of physical systems and statements about transformation properties are often extremely helpful and usually provide important physical insight. i. We will see below how true scalars under Lorentz transformations can be constructed. a1 . the square of the electric ﬁeld |E(r. p) . the charge density.66) Examples of 4-vectors beside xµ are the momentum 4-vector pµ = (E. t) and J(r. Scalars The quantities with the simplest transformation behaviour are so-called scalars f ∈ R which are invariant under transformations. E = √ m mv . tensor operators Tkm .e. another example is the rest mass m of a particle.. a2 . f = f.65) An example is s2 deﬁned in (10. (10. we will encounter an impressive example of physical properties. A third 4-vector is the so-called current vector J µ = (ρ.3) are the so-called 4–vectors aµ . Hence. the z–component x3 of a particle. Presently. our deﬁnition of quantities with special properties under Lorentz transformations presently is conﬁned to the natural representation.

Covariant 4-vectors transform like a µ = gµν Lν ρ g ρσ aσ where we deﬁned g µν = gµν . gσν xν = Lρ σ gρµ x µ and g µσ gσν = δ µ ν .e.2: Scalars.78) To prove the 4-vector property of ∂µ we will show that g µν ∂ν transforms like a contravariant 4vector.17) Lµ σ is the transformation inverse to Lσ µ . aµ (10. The 4-Vector ∂µ (10. The respective vectors aµ are associated with the 4-vectors aµ . i. Multiplication (and summation) of x µ = Lµ ν xν by Lρ σ gρµ yields.12) a gµν b µ ν = Lµ ρ gµν Lν σ aρ bσ = aρ gρσ bσ (10. one can express [(L−1 )T ]µ ν = (L−1 )ν µ and.66). −a3 ) (10. If aµ and bµ are 4-vectors aµ gµν bν (10.73) aµ = Lµ σ aσ . lower. This property follows from (10. 4–Vectors and Tensors Scalar Product then 295 4-vectors allow one to construct scalar quantities. the relationship being aµ = gµν aν = (a0 .70) is a scalar. We like to point out that from deﬁnition (10.17) can be written (L−1 )ν µ = Lµ ν .72) of the covariant 4-vector follows = In µν and g µ as well.73) Note that according to (10. xν = g νσ Lρ σ gρµ x µ .72) where aν is a vector with transformation behaviour as stated in (10.. In fact. one can employ the tensors g µν to raise and lower indices of L ν establish here the consistency of the ensuing notation. We start from x µ = Lµ ν xν . ∂x µ (10. (10.79) This is the inverse Lorentz transformation consistent with (10. −a2 .76) An important example of a covariant 4-vector is the diﬀerential operator ∂µ = ∂ = ∂xµ ∂ . i. (10. We do not fact.16). ∂t (10.e.. One calls 4-vectors aµ covariant and 4-vectors aµ contravariant.75) (10. (10.10. using (10.74) g µν aν . We have duplicated the expression for the inverse of Lµ ν to obtain the correct notation in terms of covariant. In any case one can express (10.66) together with (10.71) Contravariant and Covariant 4-Vectors It is convenient to deﬁne a second class of 4-vectors.77) The transformed diﬀerential operator will be denoted by ∂µ = def ∂ .12). accordingly. −a1 . g µν ∂ν = Lµ ρ g ρσ ∂σ . and .

83) is a scalar under Lorentz transformations. In fact.. 2 2 dt 1 − v 1 − v (10.e.67) transforms like (10.80) Multiplication by g λµ (and summation over µ) together with g λµ gρµ = δ λ ρ yields g λµ ∂µ = Lλ σ g σν ∂ν . i.89) . 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. (10. p1 = √ m v1 m dx1 = √ . d’Alembert Operator We want to construct now a scalar diﬀerential operator.− ∂t .85) 1 d d = √ .79) allows us to determine the connection between ∂µ and ∂µ .84) = 1 − (v)2 (10. this operator is equal to the d’Alembert operator which is known to be Lorentz-invariant. Using the chain rule of diﬀerential calculus we obtain 3 ∂µ = ν=0 ∂xν ∂ = g νσ Lρ σ gρµ ∂ν = Lµ ν ∂ν ∂x µ ∂xν (10. dτ 1 − v 2 dt p0 = E = √ m m dt = √ ..e.86) One can write (10. 1 − v2 1 − v 2 dt (10. 2 dt dτ 1 − v (10.296 Relativistic Quantum Mechanics contravariant.g. indices. Proof that pµ is a 4-vector We will demonstrate now that the momentum 4-vector pµ deﬁned in (10. (10.. e. i. For this purpose we deﬁne ﬁrst the contravariant diﬀerential operator ∂ µ = g µν ∂ν = Then the operator 2 ∂µ ∂ µ = ∂t − 2 (10.88) One can express then the momentum vector pµ = √ d m dxµ = m xµ .87) The remaining components of pµ can be written. ∂µ does indeed transform like a covariant vector.82) (10.66).81) ∂ . upper.

We like to rewrite the last result E 2 = p 2 + m2 or E = ± p 2 + m2 .h. hence. equivalently.12). namely pµ pµ = pµ gµν pν = E 2 − p 2 Evaluation of the r. hence.93) (10.96) is a scalar and.67) E2 − p 2 = or pµ pµ = m2 which.. t) + · A(r. a contravariant 4-vector then the l.95) 2 p This obviously describes the energy of a free particle with rest energy ±m.s. and. 10.h.96) in the form ∂µ Aµ = 0 . Since xµ transforms like a contravariant 4-vector. m2 m2 v 2 − = m2 1 − v2 1 − v2 (10. in fact.3: Relativistic Electrodynamics 297 d The operator m dτ transforms like a scalar. and had chosen the so-called Coulomb gauge (8. t) = 0 . Lorentz Gauge In our previous description of the electrodynamic ﬁeld we had introduced the scalar and vector potential V (r.90) (10.69) for the electrodynamic potential and the 4-vector derivative (10.89) must be a 4-vector.s. the r. We will demonstrate now the 4-vector property of Aµ .h. kinetic energy ± 2m and relativistic corrections. (10.77) which allow one to express (10.96) The Lorentz-invariance of this gauge.69) is.92) (10. The momentum 4-vector allows us to construct a scalar quantity. i. ∂t V (r. Expansion in should then be rapidly convergent. in (10. If we can show that Aµ deﬁned in (10. in (10. t).10. for these potentials. of (10. 8.94) 1 m In the non-relativistic limit the rest energy m is the dominant contribution to E.e.97) and. namely.s. can be demonstrated readily using the 4-vector notation (10. pµ on the l. respectively.s.91) (10. t) and A(r.3 Relativistic Electrodynamics In the following we summarize the Lorentz-invariant formulation of electrodynamics and demonstrate its connection to the conventional formulation as provided in Sect. of (10. Lorentz-invariant. (10. One obtains E = ±m ± p2 2m (p 2 )2 + O 4m3 (p 2 )3 4m5 . yields according to (10.h. . (10. This gauge is not Lorentz-invariant and we will adopt here another gauge. is a scalar. the so-called Lorentz gauge.97) We have proven already that ∂µ is a contravariant 4-vector. · A = 0. in fact.89) alltogether transforms like a contravariant 4-vector.

96). The respective equation for A0 = V can be obtained from Eq. t) = 4πρ(r.77) and (10. This principle should hold in any frame of reference. 10. · A(r.99) Since this equation must be true in any frame of reference the right hand side must vanish in all frames. must transform likewise. has to transform like a contravariant 4-vector. (10.h. of (10. t) = −∂t V (r.102) In this equation the r. must be a scalar. t) + · J(r. (8. yields ∂µ ∂ µ Aν (xσ ) = 4 π J ν (xσ ) . in fact.101) Combining equations (10. it follows that J µ . t) = ∂t · A(r. using (10. Since ∂µ transforms like a covariant 4-vector. (10.298 Transformation Properties of J µ and Aµ Relativistic Quantum Mechanics The charge density ρ(r. ∂µ J µ (xµ ) = 0 .9). t) and current density J(r.100.18) and. transforms like a 4-vector. . also the l. according to (10.68). (10..h. t) − 2 A(r. Since ∂µ ∂ µ transforms like a scalar one can conclude that Aν (xσ ) must transform like a 4-vector.69).h. t) from (8. that F µν can be expressed through the potential Aµ in (10. thereby. One can readily verify. t) .99) must be a scalar.103) B . The l. t) − 2 V (r. in fact. i. Consequently.17) using the identity (8. t) = 0 (10. t) together with (10. t) 2 ∂t A(r. Equation (10. i. prove that Aµ is. t) .32).82) as follows F µν = ∂ µ Aν − ∂ ν Aµ .83) and (10. a 4-vector. The Field Tensor The electric and magnetic ﬁelds can be collected into an anti-symmetric 4×4 tensor 0 −Ex −Ey −Ez E 0 −Bz By .69) and ∂ µ in (10. t) are known to obey the continuity property ∂t ρ(r.98) can be written. t) = 4 π J(r. using (10. We want to derive now the diﬀerential equations which determine the 4-potential Aµ in the Lorentz gauge (10. k.97) and. F mn = − mn (10.s. using (8. Using · ∂t A(r. t) = −∂t V (r. one obtains for A(r.104) where mn = mn is the totally anti-symmetric three-dimensional tensor deﬁned in (5.98) which reﬂects the principle of charge conservation..101). m.96).13). one obtains 2 ∂t V (r. 2.6) and (8.e. t). n = 1.s. · A(r.100) Similarly. 3 (10.s. (10.e. this can be stated F k0 = −F 0k = E k .105) . F µν = x Ey B z 0 −Bx Ez −By Bx 0 Alternatively. (10.

108) yields then the expressions for the transformed ﬁelds E and B .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. B .106) In case that the Lorentz transformation Lµ ν is given by (10. 8.112) (10.3: Relativistic Electrodynamics 299 The relationships (10. E⊥ = B⊥ E⊥ + v × B √ 1 − v2 B⊥ − v × E = √ 1 − v2 (10. 8.103. B and E⊥ . t) and B(r. The results can be put into the more general form E B = = E . t). 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. Noting ∂µ F µν = ∂µ ∂ µ Aν − ∂µ ∂ ν Aµ = ∂µ ∂ µ Aν − ∂ ν ∂µ Aµ = ∂µ ∂ µ Aν .109) (10. by (10. equivalently. one can conclude from (10.111) One can readily prove that this equation is equivalent to the two inhomogeneous Maxwell equations (8.105) and (10. In a new frame of reference holds F µν = Lµ α Lν β F αβ (10.1.110) where E . .2).62) or.63). Maxwell Equations in Lorentz-Invariant Form One can express the Maxwell equations in terms of the tensor F µν in Lorentz-invariant form. These equations show that under Lorentz transformations electric and magnetic ﬁelds convert into one another.104) establishe the transformation behaviour of E(r. (10. B⊥ are. respectively.3.4). where we used (10.113) which can be shown to be equivalent to the two homogeneous Maxwell equations (8.105) of the tensor F µν one can conclude the property ∂ σ F µν + ∂ µ F νσ + ∂ ν F σµ = 0 (10. From the deﬁnition (10.97).102) ∂µ F µν = 4π J ν . the components of the ﬁelds parallel and perpendicular to the velocity v which determines the Lorentz transformation.10. 10.

5).114) reads then.e. also from the equation of motion (8.103) dpx q = ( EEx + py Bz − pz By ) .87)] and retain the deﬁnition E for the electric ﬁeld.114) where d/dτ is given by (10. dτ m √ Employing again (10..86) dE q = p·E .h. equivalent to (8.116) (10.5) where p = mv/ 1 − v 2 . provides an alternative description of the action of the Lorentz force. We want to √ demonstrate now that this equation is equivalent to the equation of motion (8.118) p· dt √ where we exploited the fact that according to p = mv/ 1 − v 2 holds p v. 1 dE 2 1 dp 2 = = qp·E 2 dt 2 dt (10. We have. (10. E = m/ 1 − v 2 . e. hence. dt E From this one can conclude. For the spatial components. yields dpx = q dt Ex + (v × B)x (10.5) taking the scalar product with p dp = qp · E (10. hence.93). The µ = 0 component of (10.. Equation (10.117) (10. . We want to express this force through the tensor F µν .67) and charge q dpµ q = pν F µν dτ m (10.5).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. in fact.120) is referred to as the Lorentz force.119) which is the x-component of the equation of motion (8.114) is. It holds for a particle with 4-momentum pµ as given by (10.114). employing (10.120) (10. To avoid √ confusion we will employ in the following for the energy of the particle the notation E = m/ 1 − v 2 [see (10.114) reads using (10. however. using (10.g.86) and (10.86).104).115) This equation follows. i. for µ = 1. The term on the r. dE q = p·E dτ m or with (10. of (10. demonstrated that (10.67).s.

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 ) . 10. (10. w) which correspond to the generators Jk and Kk in (10. K3 ) are the generators of L(ϑ.123) This deﬁnition corresponds closely to the function space representation (5. The resulting expression should be a generalization of the function space representation (5. 10.48) of SO(3).e. Functions which represent 4-vectorsor other non-scalar entities.. we seek an expression which corresponds to (10. obey a diﬀerent transformation law. For such functions holds in the transformed frame ψ (Lµ ν xν ) = ψ(xµ ) (10. K2 . (10. We like to express now ψ (x µ ) in terms of the old coordinates xµ .125) def (10.48) and (10.7. the charge-current density in case of Sect.125)]. in the present case we exclude the factor ‘−i’ [cf.. In this section we will investigate the transformation properties of scalar functions ψ(xµ ).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 .127) .4 Function Space Representation of Lorentz Group In the following it will be required to decribe the transformation of wave functions under Lorentz transformations.. i. For this purpose one replaces xµ in (10.1) is a generalization (rotation + boosts) of the group SO(3). ψ ∈ C∞ (4). xµ ).4: Function Space Representation of Lorentz Group 301 10. (5. def (10. J3 ) and K = (K1 .g.121) covers solely the transformation behaviour of scalar functions. We will denote the intended representation by L(ϑ. e.124) In this expression J = (J1 . i. 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 . w)) = ρ eϑ·J + w·K which we present in the form L(ϑ. (10. taken at the pairs of points (x µ = Lµ ν xν .3 or the bi-spinor wave function of electron-positron pairs in Sect. w. We need to emphasize that (10.51) for the natural representation of the Lorentz group. 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. w) = ρ(Lµ ν (ϑ.126) ρ ewk Kk − 1 1 .e. Accordingly. However.42) of SO(3). in as far as SO(3.121) by (L−1 )µ ν xν and obtains ψ (xµ ) = ψ((L−1 )µ ν xν ) .10.47). w) = exp ϑ · J + w · K . and which can be constructed following the procedure adopted for the function space representation of SO(3).

the expression for K1 in (10.126). One can determine similarly K1 starting from coshw1 sinhw1 sinhw1 coshw1 = 0 0 0 0 0 0 1 0 0 0 . cosϑ1 x2 + sinϑ1 x3 .131) This yields for small w1 ρ ew1 K1 ψ(xµ ) = ψ(coshw1 x0 + sinhw1 x1 . namely [J1 .130) ew1 K1 −1 = e−w1 K1 (10. K obey the same Lie algebra (10.e. (10.302 One obtains J1 = x3 ∂2 − x2 ∂3 . 1 This result. J2 ] = J3 . x1 . x3 ) 2 = ψ(xµ ) + w1 (x1 ∂0 + x0 ∂1 ) ψ(xµ ) + O(w1 ) (10. 0 1 (10. K ] = = k m Jm k m Jm = − k m Km . (10. x1 ∂3 ] − [x2 ∂3 .134) . x3 ∂1 ] = −x1 ∂2 + x2 ∂1 = J3 . sinhw1 x0 + coshw1 x1 . K2 ] = −J3 . which we like to demonstrate for J1 and K1 .132) and. i. x2 .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. and [J1 . obviously. obviously. J2 ] = [x3 ∂2 − x2 ∂3 . K ] [ Jk .49) as the generators of the natural representation.133) We demonstrate this for three cases. −sinϑ1 x2 + cosϑ1 x3 ) = ψ(xµ ) + ϑ1 (x3 ∂2 − x2 ∂3 ) ψ(xµ ) + O(ϑ2 ) . J ] [ Kk . [ Jk .129) which yields for small ϑ1 ρ eϑ1 J1 ψ(xµ ) = ψ(x0 . J3 = x2 ∂1 − x1 ∂2 . reproduces the expression for J1 in (10.128) eϑ1 J1 = e−ϑ1 J1 1 0 = 0 0 0 0 0 1 0 0 0 cosϑ1 sinϑ1 0 −sinϑ1 cosϑ1 (10. The generators J . [K1 . In order to evaluate (10.128). x1 ∂3 − x3 ∂1 ] = [x3 ∂2 . J2 = x1 ∂3 − x3 ∂1 . K2 ] = K3 : [ J1 .

x0 ∂2 + x2 ∂0 ] = [x0 ∂1 . x0 ∂2 ] = −x2 ∂1 + x1 ∂2 = − J3 . In the following we consider solely the case x0 ≥ 0.143) . 10. (10. The relationships (10.125) in the case of a one-dimensional transformation of the type L(w3 ) = exp w3 K3 .140) allow one to ∂ ∂ express the derivatives ∂0 . can be simpliﬁed considerably.137). 303 (10. cothΩ = 3 . Ω which are connected with x0 . ∂3 in terms of ∂R .136) (10. In both cases the radial coordinate is the quantity conserved under the transformations. x0 ∂2 + x2 ∂0 ] = [x3 ∂2 . i.139) x3 = R sinhΩ (10. We note tanhΩ = ∂R x0 = .4: Function Space Representation of Lorentz Group [ K1 . x0 ∂2 ] = x3 ∂0 + x0 ∂3 = K3 .135) (10..10. K2 ] = [x0 ∂1 + x1 ∂0 . 0 ∂cothΩ ∂x x = ∂R x0 = − ∂x3 R (10.138) x3 x0 .141) (10. x2 ∂0 ] − [x2 ∂3 . Eqs.87)]. (5.137) where K3 is given in (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. ∂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 . x2 + x2 + x2 1 2 3 in the case of SO(3) and (x0 )2 − (x3 )2 in case of transformation (10.142) (10.85-5.128). 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 . One-Dimensional Function Space Representation The exponential operator (10. x2 ∂0 ] − [x1 ∂0 . x3 as follows x0 = R coshΩ . K2 ] = [x3 ∂2 − x2 ∂3 . For this purpose one expresses K3 in terms of hyperbolic coordinates R. ∂Ω .e.139. [ J1 . a relationship which can also be stated R = and + − (x0 )2 − (x3 )2 (x0 )2 − (x3 )2 if x0 ≥ 0 if x0 < 0 (10.

137) 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 ) . which allows one to replace classical observables by quantum mechanical operators acting on wave functions.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. Free Particle Case A quantum mechanical description of a relativistic free particle results from applying the correspondence principle. representation theoretic treatment. The historic route to these two equations. In these units the quantities energy.121.147) (10. complex-valued function. ˆ pµ =⇒ pµ = i(∂t .149) where ψ(xµ ) is a scalar. however. mass. The partial diﬀerential equation (10. is not very insightful. In the position representation the correspondence principle states ˆ E =⇒ E = − ∂t i ˆ p =⇒ p = i which. In natural units the Klein–Gordon equation (10. ∂Ω Relativistic Quantum Mechanics (10. (10. Further below we will provide a more systematic. extremely useful.148) ∂ µ ∂µ + m2 (10.150) .122).146) Applying the correspondence principle to (10.151) is called the Klein-Gordon equation.92) one obtains the wave equation − or 2 2 µ ∂ ∂µ ψ(xν ) = m2 ψ(xν ) ψ(xν ) = 0 . ˆ (10.128) yields K3 = x0 ∂3 + x3 ∂0 = ∂ .151) reads ∂µ ∂ µ + m2 ψ(xµ ) = 0 (10. in 4-vector notation reads pµ =⇒ pµ = i (∂t .144) The action of the exponential operator (10. and (time)−1 all have the same dimension. but certainly is short and. namely the Klein–Gordon and the Dirac equations.304 Inserting these results into the deﬁnition of K3 in (10. momentum. In the following we will employ so-called natural units = c = 1. We will provide a ‘derivation’ of these two equations which stem from the historical development of relativistic quantum mechanics. The latter property implies that upon change of reference frame ψ(xµ ) transforms according to (10. (10.145) ∂Ω 10. 10. ) = i ∂µ . therefore. (length)−1 . − ) = i ∂ µ .

t) = · jKG (r. t) − ψ(r. t) ) . t)∂t ψ(r. t) + where ρS (r. t)∂t ψ ∗ (r. t) − ψ(r.161) . t) 1 [ ψ ∗ (r.150) becomes ∂µ ∂ µ + m2 ψ (x µ ) = 0 (10. Under Lorentz transformations the free particle Klein–Gordon equation (10.154). In order to obtain the conservation law connected with the Klein–Gordon equation (10. t) ψ(r.155) describes the positive deﬁnite probability to detect a particle at position r at time t and where jS (r.156) · jS (r. t) = 0 (10. and that (as postulated) ψ(xµ ) is a scalar.150) is invariant under Lorentz transformations.152) which has the same form as the Klein–Gordon equation in the original frame. t) ψ ∗ (r. This follows from the fact that ∂µ ∂ µ and m2 are scalars.10. To 1 derive this conservation law one rewrites the Schr¨dinger equation in the form (i∂t − 2m 2 )ψ = 0 o and considers 1 2 Im ψ ∗ i∂t − ψ = 0 (10. 2mi (10.159) (10. t) (10.157) 2m which is equivalent to (10. t) ψ(r. t) ψ ∗ (r.160) ∂t (ψ ∗ ∂t ψ which corresponds to − ψ∂t ψ∗) + · (ψ ψ∗ ∂t ρKG (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) = (10. t) ψ(r.150) one considers Im ψ ∗ ∂µ ∂ µ + m2 ψ = 0 (10.158) 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) + where ρKG (r. t) = 0 i ( ψ ∗ (r. t) ) 2m 1 = ( ψ ∗ (r.153) is associated with a conservation law for particle probability ∂t ρS (r. t) = − 1 2m 2 ψ(r.154) describes the current density connected with motion of the particle probability distribution. t) = ψ ∗ (r.5: Klein–Gordon Equation or 2 ∂t − 2 305 + m2 ψ(xµ ) = 0 . t) ] 2mi (10. t) − ψ(r. (10. t) jKG (r.

the Klein-Gordon equation is considered as a suitable equation to describe spin–0 particles. (10. Generating a Solution Through Lorentz Transformation A particle at rest. λ = ± . We denote this by summarizing the solutions as follows ∂µ ∂ µ + m2 ψo (p. λ) associated with the corresponding wave functions ψo (p.e.164) The corresponding energy is Eo (po .166) is Eo (p) ∗ ρKG (p. λ = ± (10. p0 . corresponding to λ = +.168) .162) Inserting this into the Klein–Gordon equation (10. λ) actually describes charge density rather than probability. Solution of the Free Particle Klein–Gordon Equation Solutions of the free particle Klein–Gordon equation are ψ(xµ ) = N e−ipµ x µ = N ei(p0 ·r − Eo t) . λ) = λ ψo (p. and the density ρKG (p. according to (??) is decribed by the r–independent wave function ψo (p = 0.165) This result together with (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. λ|xµ )ψo (p. The density ρKG (p.161) and (10.166) ψo (p. λ|xµ ) = Nλ. The proper interpretation of ρKG (r. ±) = ± 2 m2 + po (10.163) which results in the expected [see (10. corresponding to λ = −.154). t) = 0 (10. o namely.162) shows that the solutions of the free particle Klein-Gordon equation (10.. not of particle probability. Today. with p = 0. is actually that of a charge density.150) is a continuum of positive energies E ≥ m. it had been realized later. for example pions. λ|xµ ) = N e−iλmt .p ei(p·r − λEo (p)t) Eo (p) = The spectrum of the Klein–Gordon equation (10. in that the expression for ρKG is not positive deﬁnite.306 Relativistic Quantum Mechanics This conservation law diﬀers in one important aspect from that of the Schr¨dinger equation (10. λ|xµ ) according to (10. and of negative energies E ≤ −m. (10.151) yields 2 2 Eo − p0 − m2 ψ(r.150) are actually determined by po and by the choice of sign ±. i. λ|xµ ) = 0 2 m2 + po . m 2 2 E0 = m2 + po . (10. t). λ|xµ ) (10.93] dispersion relationship connecting E0 . The proper interpretation of the two cases is that the Klein–Gordon equation describes particles as well as anti-particles.167) m which is positive for λ = + and negative for λ = −. the anti-particles carry a charge opposite to that of the associated particles.

in fact. for negative energy solutions.168).170) The coordinate transformation (10. (10. This yields.168). λ = −|xµ ) = N ei(px 3 + Eo (p)x0 .173) (10. t)) .166)] of the particle.172) as the energy [c. −A(r.177) .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. (10.e.175) as the momentum of the particle and one obtains L(w3 )ψo (p = 0. for the wave function (10..174) which agrees with the expression given in (10.f.166).145) choosing m = sinhw3 . i.61) yield for this expression −iλ √ m mv x0 − iλ √ x3 .169) The addition theorem of hyperbolic functions cosh(Ω + w3 ) = coshΩ coshw3 + sinhΩ sinhw3 allows us to rewrite the exponent on the r. t)) (10. In case of λ = + one obtains ﬁnally L(w3 )ψo (p = 0. one has to interprete p = mv/ 1 − v 2 (10. In case of λ = −. Ω. 2 2 1−v 1−v (10. using (10.138) to replace t = x0 by hyperbolic coordinates R. λ = +|xµ ) = N ei(px 3 − Eo (p)x0 (10. For this purpose we consider the wave function for a particle moving with momentum velocity v in the direction of the x3 –axis.e.. for positive energy solutions.h. L(w3 )ψo (p = 0. Aµ (xµ ) = (V (r. (10.s.10. A(r.138) and the relationships (10. p = − mv/ 1 − v 2 as the momentum of the particle relative to the moving frame and √ m = 1 − v2 m2 = 1 − v2 m2 + m2 v 2 = 1 − v2 m2 + p2 = Eo (p) (10.171) One can interpret then for λ = +. t). Such wave function should be generated by applying the Lorentz transformation in the function space representation p (10. i. of (10. (10.6: Klein–Gordon Equation with Electromagnetic Field 307 We want to demonstrate now that the wave functions for p = 0 in (??) can be obtained through appropriate Lorentz transformation of (10.176) 10. t). λ|xµ ) = = exp w3 ∂ ∂Ω N e−iλmRcoshΩ 3) N e−iλmRcosh(Ω+w .169) −iλ ( m coshw3 ) ( R coshΩ ) − iλ ( m sinhw3 ) ( R sinhΩ ) .

We will also assume.147). (10.179) (10. i.93). and focus on the remaining part of the wave function. is much larger than all other energy terms. i.69)] one replaces the quantum mechanical operators. In terms of space-time derivatives this reads (i∂t − qV (r. m2 ) ψ(xµ ) = 0 . t) . For this purpose one writes the Klein-Gordon equation (10. (10.150) −g µν (i∂µ )(i∂ν ) + m2 ψ(xµ ) = 0 .178) According to the replacements in Table 10. t) 2 (10. in keeping withnthe non-relativistic limit. according to the rules shown in Table 10. t) = e−imt Ψ(r. that the mass m of the particle. A) free quantum particle quantum 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 − .. (10. albeit in a form.182) and seek an equation for Ψ(r.1. See also Eq. we deﬁne ψ(r. momenta for classical particles and the respective operators for quantum mechanical particles in the manner shown.150). t))2 ψ(r. According to the principle of minimal coupling [see (10.e. For this purpose we split-oﬀ a factor exp(−imt) which describes the oscillations of the wave function due to the rest energy.308 free classical particle classical particle in ﬁeld (V. t). i. 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. t) = −i − q A(r. t) . in . −A).e. 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...e. According to the so-called minimum coupling principle the presence of the ﬁeld is accounted for by altering energy.1: Coupling of a particle of charge q to an electromagnetic ﬁeld described by the 4-vector potential Aµ = (V.181) Non-Relativistic Limit of Free Particle Klein–Gordon Equation In order to consider further the interpretation of the positive and negative energy solutions of the Klein–Gordon equation one can consider the non-relativistic limit. (10. A) or Aµ = (V. i∂t and −i in (10. it’s rest energy.180) + m2 ψ(r. which couples a particle of charge q to an electromagnetic ﬁeld described through the potential Aµ (xν ).

The approximation is justiﬁed on the ground of the inequalities (10. (10. The Klein-Gordon equation (10.6: Klein–Gordon Equation with Electromagnetic Field particular. t) 2m Ψ(r. larger than |i∂t Ψ/Ψ| and alrger than qV .183) The term on the l. t) = − Ze and A(r. t) ≡ 0 r + Ze2 r 2 + 2 − m2 π ϕ(r ) = 0 .183). e.. Pionic Atoms To apply the Klein–Gordon equation (10.10. identical to the Schr¨dinger equation (10. i.181) yields (we assume now that the Klein–Gordon equation describes a 2 particle with mass mπ and charge −e) for qV (r. (10. 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− . (10.181) reads then i ∂t Ψ(r.188) Inserting this into (10.187) We want to investigate in the following a description of a stationary state of a pionic atom involving a nucleus with charge +Ze and a π − meson. t) (10.2) of a non-relativistic spin-0 particle o moving in an electromagnetic ﬁeld. 309 (10. | i∂t Ψ | << m .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. of (10..e. (10.189) . t) = ˆ [p − q A(r. i. 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. however. Ψ |q V | << m .g. atoms in which one or more electrons are replaced by π − mesons.s.h.184) where we neglected all terms which did not contain factors m. π − mesons are generated through inelastic proton–proton scattering p + p −→ p + p + π − + π + . To ‘manufacture’ pionic atoms.185) This is. A stationary state of the Klein–Gordon equation is described by a wave function ψ(xµ ) = ϕ(r ) e−i t . This application demonstrates that the Klein–Gordon equation describes spin zero particles. The process of π − meson capture involves the so-called Auger eﬀect. t)]2 + qV (r.. spin-0 mesons.e.181) to a physical system we consider pionic atoms.

. . (10.. n = 1. (10. φ.192) leads then to the ordinary diﬀerential equation d2 − dr2 ( + 1) − Z 2 e4 2 Ze2 + + r2 r 2 − m2 π R (r) = 0 . . .e. 1.199) ( + 1) − Z 2 e4 . i.195) The latter problem leads to the well-known spectrum En = − mπ (Ze2 )2 . the eigenfunctions of the operator L2 ˆ L2 Y m (θ.197) counts the number of nodes of the wave function.194) and (10.192) ˆ are spherical harmonics. n − 1 . .194) Bound state solutions can be obtained readily noticing that this equation is essentially identical to 2 that posed by the Coulomb problem (potential − Ze ) for the Schr¨dinger equation o r d2 − dr2 ( + 1) 2mπ Ze2 + + 2mπ E r2 r R (r) = 0 (10. φ) (10. r r r r sin θ r sin θ Ze2 2 r ) (10. The similarity of (10. 2. (10.195) can be made complete if one determines λ such that λ( ) (λ( ) + 1) = The suitable choice is λ( ) = − 1 + 2 ( + 1 2 ) − Z 2 e4 2 (10.198) .310 Relativistic Quantum Mechanics Because of the radial symmetry of the Coulomb potential we express this equation in terms of spherical coordinates r. φ) R (r) Y r m (θ.e. θ. i. this quantity deﬁnitely must be an integer. . (10.196) In this expression the number n deﬁned through n = n− −1 (10. 2n2 = 0.191) ˆ The operator L2 in this equation suggests to choose a solution of the type ϕ(r ) = where the functions Y in (10.. φ) = ( + 1) Y m (θ.191) m (θ. The Laplacian is 2 = ˆ 1 2 1 1 1 2 L2 2 ∂r r + 2 2 ∂θ sinθ∂θ + 2 2 ∂φ = ∂r r − 2 . (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 .193) (10. φ) . .

1. . . EKG = Using (10. . which corresponds to a bound = 0.201) mπ Z 2 e4 m2 − m2 π π = − .194) d2 λ( ) ( λ( ) + 1 ) 2 Ze2 − + + dr2 r2 r 2 311 − m2 π R (r) = 0 . 2mπ 2 −→ λ( ) . . .6: Klein–Gordon Equation with Electromagnetic Field and one can write (10. .. 2mπ 2 (n + λ( ) + 1)2 (10. . 1.197. . 2. − + 1. Introducing α = Z 2 e4 and 1 β = + 2 (10. .203) spectrum . m) = mπ 1 + Z 2 e4 (n− − 1 2 + ( + 1 )2 − Z 2 e4 )2 2 n = 1.202) Solving this for state) yields (choosing the root which renders = 1 + mπ Z 2 e4 ( n + λ( )+1 )2 ≤ mπ . . .199) and deﬁning EKG (n.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 ). .204) reads 1 (10. + (10.205) α √ 2 2 1+ (n − β + β −α) and one obtains the series of approximations 1 1+ ≈ ≈ ≈ 1+ ≈ 1+ (n − β + α √ β 2 −α)2 1 1+ (n − α 2β α + O(α2 ))2 1 1+ n2 − α n + O(α2 ) β α 1 α n2 + α2 βn3 + O(α3 ) − α2 8n4 1 α 2n2 + α2 2βn3 + O(α3 ) . 1. 10.10. . . . . . . . i. n − 1 m = − . 2 e2 −→ e2 mπ . (10. n = 0. results in the (10.e. = 0.196) if one makes the replacement 2 values which can be obtained E −→ One obtains − m2 π . .200) The bound state solutions of this equation should correspond to from (10. (10.

209) which.209). Finally. . is identical to the two equations (10. a splitting of the six n = 2.7 The Dirac Equation Historically.e.208)..206) 1 2 3 + O(Z 6 e12 ) . but not an operator iγ · as suggested by the principle of relativity (equivalence of space and time). a feature which is connected with the 2nd order time derivative in this equation. t) (10. His solution . but for a long time to no avail. Equations (10. would have only a ﬁrst order time derivative. are unsatisfactory since expansion of the square root operator involves all powers of the Laplace operator. 4n (10. This derivative. m) ≈ m − mZ 2 e4 mZ 4 e8 − 2n2 2n3 1 + − 1− Relativistic Quantum Mechanics α α2 α2 α2 − + + + O(α3 ) . The latter term agrees with observations of pionic atoms.312 ≈ From this results for (10.e..146) leads to the wave equation i∂t Ψ(r. in turn. Application of the correspondence principle (10. i.208) These two equation can be combined i∂t + m2 − 2 i∂t − m2 − 2 Ψ(r. 10. 10. It should feature then a diﬀerential operator of the type D = iγ µ ∂µ . through the correspondence principle. i. but in the present case rather the negative of the energy. the Dirac equation introduced below.207) Here the ﬁrst term represents the rest energy. (10. 10. = 1 states into groups of two and four degenerate states. The latter spectrum shows. it does not agree with observations of the hydrogen spectrum. However. 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 . however.2 particles. arises because the Klein–Gordon equation.. however. In order to describe electron spectra one must employ the Lorentz-invariant 1 wave equation for spin. the π − meson is a pseudoscalar particle. t) = ± m2 − 2 Ψ(r. the second term the non-relativistic energy. t) . Also. in fact.204) EKG (n. 2n2 2βn3 8n4 4n4 (10.208. Dirac succeeded. one with a positive-deﬁnite density. Many attempts were made by theoretical physicists to ‘linearize’ the square root operator in (10. the wave function changes sign under reﬂection. 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). and the third term gives the leading relativistic correction.e.208. for example. i. the Klein–Gordon equation had been rejected since it did not yield a positive-deﬁnite probability density. is related to the equation E 2 = m2 + p 2 of the classical theory which involves a term E 2 . A more satisfactory Lorentz–invariant wave equation. It must be pointed out here that does not denote energy.

66). The discovery by Dirac.213) We want to determine now the simplest algebraic realization of γ µ .213) reads (γ µ )2 = 1 µ = 0 −1 µ = 1. Instead. according to (10. Accordingly.209).215) . In fact. yet to be discovered. Properties of the Dirac Matrices Let us now trace Dirac’s steps in achieving the linearization of the ‘square root operator’ in (10. 2. γ 2 . It turns out that no 4-vector of real or complex coeﬃcients can satisfy these conditions. 3 . therefore. these descriptions ultimately merging with the true theory of matter. γ j . Inserting (10. namely. γ ν ]+ = 2 g µν (10. ψ2 (xµ ).7: Dirac Equation 313 to the problem involved an ingenious step. j = 1. unspeciﬁed algebraic objects γ µ and γ ν .1 2 and anti-particles. the 4-dimensionsional representation discovered by Dirac involved new physical properties. quantities with the transformation law (10. one route to important discoveries in physics is the creation of new mathematical descriptions of nature.210) where P = iγ µ ∂µ . . are associated with a positive-deﬁnite particle density. this was not so. 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 ). We seek to identify the coeﬃcients γ µ .208). beautiful mathematical theory and that. .211) into (10. ψd (xµ ). one can impose the condition γ0 is hermitian . but did not commute the. 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. γ 1 . . Initially. spin.. (10.212) where we have changed ‘dummy’ summation indices. it was assumed that the 4-dimensional space introduced by Dirac could be linked to 4vectors. 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. strengthens the believe of many theoretical physicists today that the properties of physical matter ultimately derive from a. 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 . For µ = ν condition (10. However.211) Obviously. so far. the quantities γ 0 .214) From this follows that γ 0 has real eigenvalues ±1 and γ j . the operator of the Klein–Gordon equation ∂µ ∂ µ + m2 = −( P + m ) ( P − m ) (10. i. 3 are anti-hermitian . (10. exploited ∂µ ∂ν = ∂ν ∂µ . j = 1. the realization that the linearization can be carried out only if one assumes a 4-dimensional representation of the coeﬃcients γ µ . achieved through a beautiful mathematical theory. Starting point is to boldly factorize. 3 has imaginary eigenvalues ±i.10. 2. (10. therefore.e.

(10..219) Using property (10. The Adjoint Dirac Equation The adjoint equation is Ψ† (xµ ) i(γ µ )† ∂µ + m ← ← (10. A proper choice is γ0 = 1 1 0 0 −1 1 . For d = 2 there exist only three anti-commuting matrices.218) The Pauli matrices allow one. One can readily show using the hermitian property of the Pauli matrices (γ 0 )† = γ 0 and (γ j )† = −γ j for j = 1.222) and multiplication from the right by γ 0 yields the adjoint Dirac equation Ψ† (xµ ) γ 0 iγ µ ∂µ + m ← =0. (10. Relativistic Quantum Mechanics (10. ψ3 . in fact. is unique.221) =0 (10. is implied by (10.e.222) ∗ ∗ ∗ ∗ where we have deﬁned Ψ† = (ψ1 .220) where Ψ(xµ ) represents a 4-dimensional wave function. rather than to the right side.215). namely the Pauli matrices σ 1 . (10. We will argue further below that the choice f γ µ . rather than a scalar wave function. 3 which. ψ2 . the γ µ are anti-commuting. in fact. From this one can conclude for the determinants of γ µ det(γ µ γ ν ) = det(−γ ν γ µ ) = (−1)d det(γ ν γ µ ) = (−1)d det(γ µ γ ν ) .213) is satisﬁed. to construct four matrices γ µ for the next possible dimension d = 4. 2.216) i. This property can also be written (γ µ )† = γ 0 γ µ γ 0 .213) reads γ µ γ ν = −γ ν γ µ . σ j σ k = − σ k σ j for j = k . γj = 0 −σ j σj 0 .224) . 1 (10.314 For µ = ν (10. From this equation one can conclude that also the following should hold ( iγ µ ∂µ − m ) Ψ(xµ ) = 0 which is the celebrated Dirac equation. σ 2 . as long as det(γ µ ) = 0 the dimension d of the square matrices γ µ must be even so that (−1)d = 1. σ 3 for which. however. except for similarity trasnformations. ψ4 ) and where ∂µ denotes the diﬀerential operator ∂ µ operating to the left side. (10.218) of the Pauli matrices one can readily prove that condition (10. holds 2 σj = 1 . 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 (10.223) Inserting this into (10.217) Obviously.

γj = ˜ 0 σj 1 S = √ 2 −σ j 0 1 1 1 1 1 −1 1 1 (10. (10.1: Derive (refeq:Dirac-intro20a) from (10. 3 and ˆ β = 1 1 0 0 −1 1 .10. 10. (10.229) The similarity transformation (10..225) .213) still holds. and γ0 = ˜ 0 1 1 1 0 1 . γ ν ]+ = 2 g µν .227) leaves the algebra of the Dirac matrices unaﬀected and commutation property (10. 2. 2.227) A representation often adopted beside the one given by (10. 3 .7.e.Chiral Representation 315 The Dirac equation can be subject to any similarity transformation deﬁned through a non-singular ˜ 4 × 4–matrix S. j = 1.227). j = 1. γ ˜ Exercise 10. t) = 0 (10.233) The eigenstates and eigenvalues of H correspond to the stationary states and energies of the particles described by the Dirac equation. (10.231) ˆ whereα has the three components αj . i.219) is the socalled chiral representation deﬁned through ˜ Ψ(xµ ) = S Ψ(xµ ) .222.232) 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 ) . j = 1. 3 . Deﬁning a new representation of the wave function Ψ(xµ ) ˜ Ψ(xµ ) = S Ψ(xµ ) leads to the ‘new’ Dirac equation ˜ ( i˜ µ ∂µ − m ) Ψ(xµ ) = 0 γ where γ µ = S γ µ S −1 ˜ (10. (10. αj = ˆ 0 σj σj 0 . (10.221) from the left by γ 0 ˆ ˆ i ∂t + i α · − β m Ψ(r.7: Dirac Equation Similarity Transformations of the Dirac Equation .228) (10. . ˆ ˆ ˆ Ho = α · p + β m . 2.230) Schr¨dinger Form of the Dirac Equation o Another form in which the Dirac equation is used often results from multiplying (10. [˜ µ .226) (10.213).

235) and.Good. 27.316 Cliﬀord Algebra and Dirac Matrices The matrices deﬁned through dj = iγ j . d3 . page 187. the matrix elements of C = A ⊗ B are Cjk. 3 . The reader may .Mod. d2 . form a Cliﬀord algebra C3 . m = Aj Bkm . j = 1.H.e. . i. Γ±8 = ±d1 d4 . For example. hence. Obviously. Γ±3 = ±d2 . satisfy the anti-commutation property dj dk + dk dj = 2 for j = k 0 for j = k Relativistic Quantum Mechanics d4 = γ 0 (10. d4 = 1 0 0 −1 1 0 0 −1 0 1 1 0 0 i −i 0 (10. . Since the Γj are hermitian the similarity transformations are actually given in terms of unitary transformations. to Dirac matrices.. The Cliﬀord algebra C4 entails a subgroup G4 of elements ± dj1 dj2 · · · djs . 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 . The three Pauli matrices also obey the property (10. Γ±4 = ±d3 .237) which are the ordered products of the operators ±1 and d1 . The representations of this group are given by a set of 32 4 × 4–matrices which are equivalent with respect to similarity transformations.238) These elements form a group since obviously any product of two Γr ’s can be expressed in terms of a third Γr . d4 is called a Cliﬀord algebra C4 . The associative algebra generated by d1 .237) by means of the property (10.236) where ‘⊗’ denotes the Kronecker product between matrices.213). any product 1 of the dj ’s can be brought to the form (10. There are (including the diﬀerent signs) 32 elements in G4 which we deﬁne as follows Γ±1 = ±1 1 Γ±2 = ±d1 . To complete the proof in this section the reader may consult Miller ‘Symmetry Groups and their Application’ Chapter 9. d2 = .e. Γ±9 = ±d2 d3 Γ±10 = ±d2 d4 . (1955). Γ±7 = ±d1 d3 .234) (10. j1 < j2 < · · · < js s ≤ 4 (10. Rev. d4 .. 2. in case of C4 .235). i. Γ±15 = ±d2 d3 d4 Γ±16 = ±d1 d2 d3 d4 (10. Γ±5 = ±d4 Γ±6 = ±d1 d2 .6 and R. Γ±11 = ±d3 d4 Γ±12 = ±d1 d2 d3 .235) can diﬀer only with respect to unitary similarity transformations.235) as can be readily veriﬁed from (10. This property extends then to 4 × 4–matrices which obey (10. The representations of Cliﬀord algebras Cm are well established. One can conclude then that also any set of 4 × 4–matrices obeying (10.213). Γ±13 = ±d1 d2 d4 . Γ±14 = ±d1 d3 d4 .Phys.

e. exist.76) yields ∂ρ = Lν ρ ∂ ν .e.8: Lorentz Invariance of the Dirac Equation 317 also want to establish the unitary transformation which connects the Dirac matrices in the form (10. 10.e.236) with the Dirac representation (10.8 Lorentz Invariance of the Dirac Equation We want to show now that the Dirac equation is invariant under Lorentz transformations. namely. a result one could have also obtained by applying the chain rule to (10.3: Demonstrate the anti-commutation relationships (10..6). 3 has imaginary eigenvalues ±i and can be represented by an anti-hermitian matrix. 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. i. i. in fact. Exercise 7. Lorentz Transformation of the Bispinor State Actually. However. i. i. The latter transformations imply that coordinates transform according to (10. the form of the Dirac equation is identical in equivalent frames of reference. these properties determine the matrices γ µ uniquely.239) Form invariance implies that the matrices γ µ should have the same properties as γ µ . 10.6).218) of the Pauli matrices σ j .80).. and derivatives according to (10.213. transform coordinates and derivatives of the Dirac equation.219).. Except for a similarity transformation. in a moving frame holds iγ µ ∂µ − m Ψ (x ) = 0 . in frames connected by Lorentz transformations. Multiplication and summation of (10.e.e..4: Show that from (10.2: Demonstrate the anti-commutation relationships (10.10. Exercise 7.240) .214) follows that γ 0 has real eigenvalues ±1 and can be represented by a hermitian matrix. it must hold γ µ = γ µ . x µ = Lµ ν xν (10. in a moving frame should hold iγ ∂µ − m Ψ (x ) = 0 .80) by Lµ ρ and using (10. therefore.215). x = Lµ ν xν . j = 1. Exercise 7.241) .218) of the Dirac matrices γ µ . Hence. We can. i. Inﬁnitesimal Bispinor State Transformation We want to show now that a suitable transformation of Ψ(xµ ) does. and γ j .. we do not know yet how to transform the 4-dimensional wave function Ψ and the Dirac matrices γ µ . 2. The transformation is assumed to be linear and of the form Ψ (x ) = S(Lµ ν ) Ψ(xµ ) µ µ (10. (10. µ µ (10.

249) from which we can conclude σµν = −σµν νµ = −σνµ i. S( µν ) should not change its value if one replaces in its argument µν by − νµ .e.247) the inﬁnitesimal matrix µν is anti-symmetric. Ob1 viously. of the equation. One can ﬁnally conclude multiplying both sides by g ρµ S(Lη ξ )γ µ S −1 (Lη ξ ) = Lν µ γν .245) The construction of S(Lη ξ ) will proceed using the avenue of inﬁnitesimal transformations. It holds then S( µν i ) = 1 − σµν 1 4 µν µν = S(− νµ i ) = 1 + σµν 1 4 µν . hence.250) σµν = −σνµ . the transformation (10.h. i.241) implies a similarity transformation Sγ µ S −1 .221) upon Lorentz transformation yields iS(Lη ξ )γ µ S −1 (Lη ξ ) Lν µ ∂ν − m Ψ (x ) = 0 .12) in the form Lν µ gνσ Lσ ρ = gµρ = gρµ . the coeﬃcients of which depend on the matrix Lµ ν deﬁning the Lorentz transformation in such a way that S(Lµ ν ) = 1 for Lµ ν = δ µ ν holds.243) in a form in which the Lorentz transformation in the form Lµν is on the r. We had introduced in (10.243) We want to determine now the 4 × 4–matrix S(Lη ξ ) which satisﬁes this condition.. (10..243) from the left by Lσ ρ gσν yields S(Lη ξ )γ µ S −1 (Lη ξ ) gρµ = Lσ ρ gσν γσ . follows S(Lη ξ )γρ S −1 (Lη ξ ) = Lσ ρ γσ .s. (10. One can. using gρµ γ µ = γρ . νµ (10. in the expression of the inﬁnitesimal transformation ρ Lµν = g µν + µν (10. . µ g ρν = µν and. µ (10. g is an anti-symmetric matrix. hence.244) from which.318 Relativistic Quantum Mechanics where S(Lµ ν ) is a non-singular 4 × 4–matrix. Multiplication of (10.248) Here σµν denote 4 × 4–matrices operating in the 4-dimensional space of the wave functions Ψ. The elements of g are. it must hold (10.247) can be expanded S( µν i ) = 1 − σµν 1 4 µν (10. (10. The proper starting point for a constructiuon of S(Lη ξ ) is actually (10. The inﬁnitesimal transformation S(Lρ σ ) which corresponds to (10. however. Multiplication of this property by g from the right yields ( g)T = − g. For this purpose we exploit (10.e.246) (10.242) The form invariance of the Dirac equation under this transformation implies then the condition S(Lη ξ )γ µ S −1 (Lη ξ ) Lν µ = γ ν .38) the inﬁnitesimal Lorentz transformations in the form Lµ ν = δ µ ν + µ ν where the inﬁnitesimal operator µ ν obeyed T = − g g. state that the Dirac equation (10.

254).255) satisfy condition (10.228. (10. β [ σαβ . In fact. the expression on the r.h.. a solution of condition (10.256) One can readily verify that the non-vanishing generators σµν are given by (note σµν = −˜νµ .251) into (10.s. For this purpose we inspect the properties of the generators in a particular representation.248. (10. (10. of (10. γβ ]− 2 (10. this condition can actually be expressed through six independent conditions.229).h.. i.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. 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 (10.257) . σjk = [˜j .h. γj = ˜ 0 −σ j σj 0 .254) The proper σαβ must be anti-symmetric in the indices α.e.251) S −1 ( µν ) = 1 + σµν µν 1 4 Inserting (10. must be symmetric with respect to interchange of the indices α and β. Algebra of Generators of Bispinor Transformation We want to construct the bispinor Lorentz transformation by exponentiating the generators σµν . γ µ ]− = 2i ( δ µ β γα − δ µ α γβ ) .s.255) which can be demonstrated using the properties (10. 10. of (10. In this representation the Dirac matrices γµ = (˜ 0 . 2.252) results in the condition for each α. Furthermore. Exercise 7.246) results then in a condition for the generators σµν − i ( σαβ γ µ − γ µ σαβ ) 4 αβ = νµ γν .216) of the Dirac matrices. γk ] = ˜ γ ˜ jk σ 0 0 σ . 10.s.253) 1 2 αβ ( δ µ β γα − δ µ α γβ ) . β and operate in the same space as the Dirac matrices. j = 1. (10.252) also as a sum over both indices of αβ .10. 3 .5: Show that the σαβ deﬁned through (10. γj ] = γ ˜ 2 −iσ j 0 0 iσ j . (10. 10.254) is σαβ = i [ γα . Comparing this with the l. the chiral representation introduced above in Eqs. For this purpose we need to verify that the algebra of the generators involving addition and multiplication is closed.h. For this purpose one needs to express formally the r.252) Since six of the coeﬃcients αβ can be chosen independently. namely. −γ ) are ˜ γ ˜ γ0 = ˜ 0 1 1 1 0 1 . ˜ ˜ σ only six generators need to be determined) σ0j = ˜ i [˜0 .s.213. like the expression on the l.

259) yields the desired connection between the Lorentz transformation (10. in particular.259) We had stated before that the transformation S is actually determined through the Lorentz transformation Lµ ν . therefore. i.51) and S. ϑ) = e 2 (w − iϑ)·σ 0 − 1 (w + iϑ)·σ 0 e 2 1 (10.e. since both operations convert block-diagonal operators A 0 0 B (10. We can ﬁnally note that the closedness of the algebra of the generators σµν is not aﬀected by similarity transformations and that. Since such operator does not change its block-diagonal form upon exponentiation the bispinor transformation (10.259) becomes in the chiral representation ˜ S(w.258) again into block-diagonal operators. .44) 0 −w1 −w2 −w3 w 0 ϑ3 −ϑ2 = 1 w2 −ϑ3 0 ϑ1 w3 ϑ2 −ϑ1 0 µν (10.248) allows us to write the transformation S for any. therefore. not necessarily inﬁnitesimal. In fact.51). 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 = . one can express the tensor µν through w and ϑ using µν = µ ρ g ρν and the expression (10. In order to construct an explicit expression of S in terms of w and ϑ we employ again the chiral representation. and since the algebra of the Pauli matrices is closed. (10. the representation (10. Finite Bispinor Transformation The closedness of the algebra of the generators σµν deﬁned through (10.241) bispinor wave functions from one frame of reference into another frame of reference. be able to state S in terms of the same parameters w and ϑ as the Lorentz transformation in (10. the algebra of these generators is closed under addition and multiplication.255) yields a closed algebra..320 Relativistic Quantum Mechanics Obviously.260) Inserting this into (10. µν in the exponential form i S = exp − σµν 4 µν .262) This expression allows one to transform according to (10. 2 0 −(w + iϑ) · σ 13 + σ23 ˜ 23 ) (10.261) We note that this operator is block-diagonal. One should. any representation of the generators.

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 . j µ must transform like a contravariant 4-vector.264) (10.221) from the left by Ψ† (xµ )γ 0 . for j µ (xµ ) = (ρ. For this purpose we prove ﬁrst the relationship S −1 = γ 0 S † γ 0 .263) obviously is a scalar under Lorentz transformations and that the left hand side must then also transform like a scalar. (10. Since ∂µ transforms like a covariant 4-vector.268) holds for the bispinor Lorentz transformation.224). This transformation behaviour can also be deduced from the transformation properties of the bispinor wave function Ψ(xµ ). (10. For our proof we note ﬁrst ˜ ˜ S −1 (w.8: Lorentz Invariance of the Dirac Equation Current 4-Vector Associated with Dirac Equation 321 We like to derive now an expression for the current 4-vector j µ associated with the Dirac equation which satisﬁes the conservation law ∂µ j µ = 0 . i.269) One can readily show that the same operator is obtained evaluating γ S (w. (10.263) Starting point are the Dirac equation in the form (10. The reason is that the r. We will now determine the relationship between the ﬂux j µ = Ψ (x )γ 0 γ µ Ψ (x ) † µ (10. j) = Ψ† (xµ )γ 0 γ µ Ψ(xµ ) .265) (10.221) and the adjoint Dirac equation (10.s.268) We will prove this property in the chiral representation. −ϑ) = e− 2 (w − iϑ)·σ 0 1 (w + iϑ)·σ 0 e2 1 (10.h.263) allows one to conclude that j µ must transform like a contravariant 4-vector as the notation implies.263) does hold. the property applies then in any representation of S. (10..224) from the right by Ψ(xµ ). Multiplying (10. in fact. the conservation law (10.e. of (10. ϑ) = S(−w. (10. The time-like component ρ of j µ 4 iγ µ ∂µ + iγ µ ∂µ ← Ψ(xµ ) = 0 .266) ρ(xµ ) = Ψ† (xµ )Ψ(xµ ) = s=1 |ψs (xµ )|2 (10.10.270) We conclude that (10. The conservation law (10.267) has the desired property of being positive deﬁnite.271) . and addition yields Ψ† (xµ )γ 0 The last result can be written ∂µ Ψ† (xν )γ 0 γ µ Ψ(xν ) = 0 . Obviously.

as just mentioned. The solutions provide also a complete. In the following we want to provide this characterization.272) With S −1 (Lη ξ ) = S((L−1 )η ξ ) one can restate (10.268) µ j = Ψ† (xµ )S † γ 0 γ µ SΨ(xµ ) = Ψ† (xµ )γ 0 S −1 γ µ SΨ(xµ ) . 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.271) that the Dirac matrices are independent of the frame of reference. µ (10. bispinor wave functions and we need to determine corresponding components of the wave function. Note that we have assumed in (10. t) and is determined through the momentum p ∈ R3 .246) S −1 (Lη ξ )γ µ S ( Lη ξ ) = (L−1 )νµ γν = (L−1 )ν γ ν = Lµ ν γν . not only as the most simple demonstration of the theory.282. Like in non-relativistic quantum mechanics the free particle wave function plays a central role. Combining this with (10. the Dirac particles are described by 4-dimensional. In case of non-relativistic quantum mechanics the free particle wave function has a single component ψ(r. 10. only two degrees of freedom of the bispinor four degrees of freedom are independent [c.76).274) 10. The operators commute also with Ho in (10.273) where we have employed (10. We will start from the Dirac equation in the Schr¨dinger form (10. One obtains using (10. but also as providing a basis in which the wave functions of interacting particle systems can be expanded and characterized. 10. In relativistic quantum mechanics a Dirac particle can also be characterized through a momentum.e.f. commute with each other. the wave function has four components which invite further characterization of the free particle state. Like for the free particle wave functions of the non-relativistic Schr¨dinger and the Klein–Gordon o equations one expects that the space–time dependence is governed by a factor exp[i(p · r − t)]..283)]. however. speciﬁc for the Dirac free particle. of which the wave functions are eigenfunctions as well. As pointed out. (10. orthonormal basis and allows one to quantize the Dirac ﬁeld Ψ(xµ ) much like the classical electromagnetic ﬁeld is quantized through creation and annihilation operators representing free electromagnetic waves of ﬁxed momentum and frequency. These operaˆ tors. a property which leads to the energy– momentum (dispersion) relationship of the Dirac particle.275) The free particle wave function is an eigenfunction of Ho .233) o i∂t Ψ(xµ ) = Ho Ψ(xµ ) . as is required for the mentioned property.232. additional characterizations. As it turns out. (10. (10.231.322 Relativistic Quantum Mechanics in a moving frame of reference and the ﬂux j µ in a frame at rest. The additional degrees of freedom described by the four components of the bispinor wave function require. i.272) results in the expected transformation behaviour µ j = Lµ ν j ν . For this purpose we consider .233). (10.9 Solutions of the Free Particle Dirac Equation We want to determine now the wave functions of free particles described by the Dirac equation. the identiﬁcation of observables and their quantum mechanical operators. 10.

χo each represent a constant. (10.278) provides us with information about the components of the bispinor wave function (10.282. (10. 1 (10. These two relationships are consistent with each other. one ﬁnds using (5. 10. and φo . the Dirac equation.276) where p and together represent four real constants.280) σ·p (σ · p)2 φ0 = χ0 = +m ( + m) ( + m) p2 χ = χ0 . like the Klein–Gordon equation. namely φo and χo are related as follows φ0 χo = = σ·p χ0 −m σ·p φ0 .94) Equation (10. two-dimensional spinor state. reproduce the classical relativistic energy–momentum relationships (10.276).234) and (10.282) (10. In fact.10.283) imply that the bispinor part of the wave function allows only two degrees of freedom to be chosen independently.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 .277) To solve this problem we write (10. We want to show now that these degrees of freedom correspond to a spin-like property.281) Obviously.277) explicitly ( − m) 1 φo − σ · p χo 1 − σ · p φo + ( + m) 1 χo 1 = 0 = 0. 10. conclude 1. Inserting (10.283) where is deﬁned in (10. hence.234) of Pauli matrices holds (σ · p)2 = p 2 1 One can.279) the well-known relativistic dispersion relationship 2 = m2 + p 2 (10. later to be identiﬁed with momentum and energy. 10.231.281). the socalled helicity of the particle.93.9: 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. E(p) = m2 + p 2 . +m (10. 2 − m2 0 (10. (10.232) leads to the 4-dimensional eigenvalue problem m σ·p σ · p −m φo χo = φo χo .279) According to the property (5. .276) into (10.284) The relationships (10.280) which has a positive and a negative solution = ± E(p) . from (10.

±).290) should diﬀer.268). We want to demonstrate now that the signs on the r.259). +) = −1 . (10..s.s..e.241). +) γ 0 Ψ† (p. using γ 0 as given in (10.. ±) Ψ† (0. (10. S † = γ 0 S −1 γ 0 . Employing (10. −|xµ ) = N− (p) −σ·p E(p) + m u u ei(p·r − t) . where S is given by (10. we present the negative energy solution. ±) = S Ψ(0. Ψ† (0.281)] separately. ±) γ 0 Ψ(p. First. The form of this condition and of (10. One can see this as follows: Let Ψ(p. i. of (10. (10.287) the form of which will be justiﬁed further below. ±).219) and u† u = 1 [c.285)].287. Hence. ±) = = = Ψ† (0.288) corresponding to the wave function Ψ(p. hence. ±) denote the solution of a free particle moving with momentum p in the laboratory frame.289) Here N− (p) is a constant which will be chosen to satisfy the normalization condition Ψ† (p. the solution for = +E(p). and let Ψ(0.290). 0) γ 0 u 0 = |N+ (0)|2 . ±) S † γ 0 S Ψ(0.286) Here N+ (p) is a constant which will be chosen to satisfy the normalization condition Ψ† (p. u† u = |u1 |2 + |u2 |2 = 1 . the l. u† u = |u1 |2 + |u2 |2 = 1 . +) γ 0 Ψ(0. The connection between the solutions. (10.324 Relativistic Quantum Mechanics For our further characterization we will deal with the positive and negative energy solutions [cf. is Ψ(p. +) = 1 . For this purpose we consider ﬁrst the positive energy solution. 10.e.h. we demonstrate that the product Ψ† (p. +) γ 0 Ψ† (p.h. = −E(p) .f.287) in the minus sign on the r. according to (10. +) = |N+ (0)|2 (u† . (10. (10. ±) γ 0 Ψ(0. we present φo through the normalized vector φo = u1 u2 = u ∈ C2 . For the positive energy solution.291) Note that we have used that. = +E(p) . through χo given by χo = u1 u2 = u ∈ C2 . (10. Similarly.287) will be justiﬁed now. of (10.s. S −1 = γ 0 S † γ 0 and.290) which diﬀers from the normalization condition (10. is invariant under Lorentz transformations. ±) . ±) denote the corresponding solution of a particle in its rest frame. ±) γ 0 S −1 γ 0 γ 0 S Ψ(0. Ψ† (p. (10.286) for p = 0 yields.e. (10.287.292) . ±)γ 0 Ψ(p. ±) Ψ† (0. (10. i. the solution for = −E(p).285) The corresponding free Dirac particle is then described through the wave function Ψ(p. 10.h. +|xµ ) = N+ (p) u σ·p E(p) + m u ei(p·r − t) . according to (10. i.

286). this requires the choice of a negative side on the r.s.219) yields 2 N+ (p) u† u − u† (σ · p)2 u (E(p) + m)2 = 1.s. σ·p u∗ E(p) + m T γo u σ·p E(p) + m u = 1 (10. ( m + E(p) )2 − p 2 (10. We can also conclude from our derivation N± (0) = 1 .f.290) to assign a positive value to |N− (0)|2 .10. (10. 2 E(p) (10.h. (10. (5.286) is 2 N+ (p) (u∗ )T . We consider ﬁrst the positive energy solution.285) results in 2 N+ (p) 1 − p2 (E(p) + m)2 = 1 (10.302) . −) = |N− (0)|2 (0.301) yields the normalization coeﬃcient N+ (p) = m + E(p) .9: Solutions of the Free Particle Dirac Equation The same calculation for the negative energy wave function as given in (10. Exercise 7. 325 (10.300) (10.298) This result completes the expression for the wave function (10. using γ 0 as given in (10.295) Evaluating the l.296) Replacing (σ · p)2 by p 2 [c.289) yields Ψ† (0. −) γ 0 Ψ(0. 2m (10.6: Show that the normalization condition 2 N+ (p) (u∗ )T .287) written explicitly using (10.293) Obviously. u† ) γ 0 0 u = − |N− (0)|2 .234)] and using the normalization of u in (10. of (10.297) from which follows N+ (p) = Noting ( m + E(p) )2 − p 2 = m2 − p 2 + 2mE(p) + E 2 (p) = 2(m + E(p)) m the normalization coeﬃcient (10.294) We want to determine now N± (p) for arbitrary p.298) becomes N+ (p) = m + E(p) . Condition (10. σ·p u∗ E(p) + m T u σ·p E(p) + m u = 1 (10.h.299) ( m + E(p) )2 .

This attribute is the so-called helicity. yields 2 N− (p) u† (σ · p)2 u − u† u (E(p) + m)2 = −1 .289. exp[i(p · r − t)] = exp(ipµ xµ ). thereby. not a constant vector. identical to the condition (10.s.286. 10.305) and. i.306) i.309) .275). λ|xµ ) in 4-vector notation. The commutation property [σ · p. We can. λ|xµ ) . (u∗ )T γ o = −1 (10. 10.. Rather than considering the ˆ observable (10.e.289) are not completely speciﬁed. 2. the wave functions constructed represent eigenstates of Ho . pj ] = 0 . i∂µ Ψ(p.296) for the normalization constant N+ (p) of the positive energy solution. −p). 3 is fairly obvious. λ|xµ ) = λ E(p) Ψ(p.e.e.308) 2 |p| ˆ Note that p represents here an operator.303) E(p) + m u Evaluating the l. λ|xµ ) = pµ Ψ(p. (10. i. have completed the determination for the wave function (10. Helicity The free Dirac particle wave functions (10. 10. λ|xµ ) (10. This can be veriﬁed expressing the space–time factor of Ψ(p. This degree of freedom 1 describes a spin– 2 attribute. The corresponding operator which measures this observable is 1 p ˆ Λ = σ· . Condition (10. deﬁned as the component of the particle spin along the direction of motion.304) This condition is..233) and ˆ ˆ ˆ from the two identities 1 1 0 0 −1 1 σ 0 0 σ ·p − ˆ σ 0 0 σ ·p ˆ 1 1 0 0 −1 1 = 0 (10. the two components of u indicate another degree of freedom which needs to be deﬁned.307) where pµ = ( .326 Relativistic Quantum Mechanics We consider now the negative energy solution.275) yields Ho Ψ(p.. We ˆ want to show that this operator commutes with Ho and p to ascertain that the free particle wave ˆ function can be simultaneously an eigenvector of all three operators. The wave functions (10. Inserting (10.h.289) is T −σ·p −σ · p 2 E(p) + m u N− (p) u∗ .286) into (10. (10.307) we investigate ﬁrst the observable due to the simpler operator σ · p. The property [σ · p. however.289).286. Ho ] = 0 follows from (10. The wave functions are also eigenstates of the momentum operator i∂µ . j = 1. (10.290) written explicitly using (10. conclude N− (p) = m + E(p) 2m (10. hence.300) have been constructed to satisfy the free particle Dirac equation (10.

− 1 |r. Ho and σ·p commute with each other and. i. +. t) = Np e (10. We assume ﬁrst particles with positive energy. i.e. −. in principle. ˆ The condition that the wave functions (10. 0. States which are simultaneously eigenvectors of these three operators are also simulteneously eigenvectors of the three operators p. Since helicity is deﬁned relative to the direction of motion of a particle the characterization of u as an eigenvector of the helicity operator.313) 2 0 1 where Ep and Np are deﬁned in (10.224) 1 of σ3 the two u vectors (1. We consider ﬁrst the simplest case that particles move along the x3 –direction. + 1 |r. Ho and Λ deﬁned in (10.10. −. The wave functions which are eigenfunctions of the helicity operator are in this case 1 −p m + Ep i(px3 + E t) 0 1 p e Ψ(pˆ3 . 1 p = (0. = −E(p). hence. 2m (10. 327 0 (σ · p)2 ˆ 2 (σ · p) ˆ 0 0 (σ · p)2 ˆ 2 (σ · p) ˆ 0 (10. p3 ).. +. Np = m + Ep .308) above.e.312)..311) 2 −p 0 m + Ep 1 where e3 denotes the unit vector in the x3 -direction and where ˆ Ep = m2 + p2 . . In this case Λ = 2 σ 3 . 0)T and (0. t) = Np e 2 1 p m + Ep 0 0 i(px3 − E t) 1 p e Ψ(pˆ3 . are 1 i(px3 − E t) 0 p e Ψ(pˆ3 . is independent of the direction of motion of the particle.. can ˆ ˆ be simultaneously diagonal. 2 the wave functions which are eigenstates of the helicity operator.286) are eigenfunctions of Λ as well will specify now the vectors u.310) We have shown altogether that the operators p. − 1 |r. = +E(p).e. Therefore. According to the deﬁnition (5. 1)T are eigenstates of 2 σ 3 with eigenvalues ± 1 . + 2 |r. t) = Np e 1 0 0 p m + Ep i(px3 + E t) 1 p e Ψ(pˆ3 .312) We assume now particles with negative energy. i. t) = Np e (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.

316) describes a rotation which aligns the x3 –axis with the direction of p.328 Relativistic Quantum Mechanics To obtain free particle wave functions for arbitrary directions of p one can employ the wave functions (10.229). for an r–independent wave function. − 2 |r. +. − ) 2 where ϑ(p) = = = exp exp i − ϑ(p) · σ 2 i − ϑ(p) · σ 2 1 0 0 1 . + 2 ) ei(p·r − Ep t) (10.223) as follows 1 u(p. + 2 ) p 1 m + Ep u(p. The corresponding wave functions are determined through ˜ i˜ 0 ∂t − m Ψ(t) = 0 . 1 |t) = √2 e−imt .262) for the bispinor wave function and the transformation (10. +. 1)T have to be replaced by eigenstates u± (p) of the spin operator along the direction of p.] The corresponding free particle wave functions are then Ψ(p.313) except that the states (1. 5.312) can be obtained also by means of the Lorentz transformation (10. − 1 ) 2 ei(p·r − Ep t) (10. 10.123). − 1 |r. i.311.311.e. + 1 ) 2 p 1 m + Ep u(p.. t) 2 = Np u(p. [One can also express the rotation through Euler angles α. + 1 |r. These eigenstates are obtained through a rotational transformation (5. − 2 ) −p 1 m + Ep u(p. in which case the transformation is given by (5.317) Ψ(p.321) . (10. denoted by ˜.226. Generating Solutions Through Lorentz Transformation The solutions (10.314) (10. γ and are 1 0 1 ˜ Ψ(p = 0. t) 2 = Np 1 u(p.318) 1 Ψ(p. + ) 2 1 u(p. γ . a wave function which represents free particles at rest. − 1 ) 2 −p 1 m + Ep u(p. −.222. 0)T and (0.320) where Ep and N are again given by (10. − 2 ) u(p.312).315) e3 × p ˆ ∠(ˆ3 .319) 1 Ψ(p. −. For this purpose one starts from the solutions of the Dirac equation in the chiral representation (10. +. 2 1 0 (10. t) = Np ei(p·r + Ep t) (10. + 2 ) u(p. 10.220). + 2 |r. p) e |p| (10. 10. t) = N√ ei(p·r + Ep t) (10. β.

10. −.324. +.322) can be written in spinor form 1 √ 2 φo χo e imt . yields for the exponential space–time dependence according to (10.322). φo . (10. 2 |t) = √2 . χo are eigenstates of σ 3 with eigenvalues ±1. 2 0 1 1 0 +imt 1 1 ˜ Ψ(p = 0. 0. For the resulting wave functions in the chiral representation one can use then a notation corresponding to that adopted in (10.9: Solutions of the Free Particle Dirac Equation 0 1 1 ˜ Ψ(p = 0.228) yields the corresponding solutions (10. 2 0 e −1 329 (10. − 1 |r.. 2 The solutions (10.324) 0 e − 1 w3 σ 3 2 φo χo = e 2 w3 σ φo 1 3 e− 2 w3 σ χo 1 3 . +.311) e 1 w 2 3 ˜ Ψ(p(w3 )ˆ3 .323) should yield the solutions for non-vanishing momentum p in the x3 –direction.311. − 1 |t) = √2 e−imt .262) φo χo → e 2 w3 σ 0 1 3 Et ) (10.313) in the p → 0 limit. for w = (0.10.e. 0 1 (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 . χo ∈ 1 0 .322) The reader can readily verify that transformation of these solutions to the Dirac representationsas deﬁned in (10. −1 e 0 0 1 +imt 1 ˜ Ψ(p = 0. 10. ± 1 of the wave functions stated in (10.323) Transformation (10. Applying (10. + 1 |r. t) e 2 e− 2 w3 1 = √ 1 2 e 2 w3 i(px3 − E t) p e .176) imt → i ( p3 x3 and for the bispinor part according to (10.325) One should note that φo .262) for a boost in the x3 –direction. − 1 |t) = √2 .174. +. i. w3 ). This correspondence justiﬁes the characterization ±. 10.325) to (10. −.

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

333) . This transformation. Invariance of Dirac Equation Revisited At this point we like to provide a variation of the derivation of (10. is suﬃcient since (1) it must hold then for any ﬁnite transformation.9: Invariance of Dirac Equation Revisited and similarly sinh w3 = 2 Ep − m = 2m p 2m (Ep + m) 331 (10.313) for −p.331) is a solution as well. in general. 10. ρ(Lη ξ ) has been deﬁned in (10. Such transformation had been applied by us. the proper transformation S(Lη ξ ). reproduces the positive energy wave functions (10. refered to as the active transformation.329. and since (2) the calculations following (10. that if Ψ(xµ ) is a solution of the Dirac equation that Ψ (xµ ) as given in (10.331) where S(Lη ξ ) denotes again the transformation acting on the bispinor character of the wave function Ψ(xµ ) and where ρ(Lη ξ ) denotes the transformation acting on the space-time character of the wave function Ψ(xµ ). = γν .243). λ. (10. indeed.123) above and characterized there.331) i S(Lη ξ )γ µ S −1 (Lη ξ ) ρ(Lη ξ )∂µ ρ−1 (Lη ξ ) − m Ψ (xµ ) = 0 (10. Making this expectation a postulate allows one to derive the condition (10. To show this we rewrite the Dirac equation (10. The reason why we provide another derivation of (10.311) as well as the negative energy solutions (10. The transformation is Ψ (xµ ) = S(Lη ξ ) ρ(Lη ξ ) Ψ(xµ ) (10.221) using (10. which however. Λ|xµ ) from the solutions describing particles at rest Ψ(p = 0. when we generated the solutions Ψ(p.243) considered solely the limit of inﬁnitesimal transformations anyway. 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. 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 . In the new derivation we consider the particle described by the wave function transformed. Λ|t). λ. expresses the system in the old coordinates.334) (10.10. Such description will be essential for the formal description of Lorentz invariant wave equations for arbitray spin further below. of course.330) Inserting expressions (10.327).332) Here we have made use of the fact that S(Lη ξ ) commutes with ∂µ and ρ(Lη ξ ) commutes with γ µ . We expect.330) into (10. Actually. but not the observer. the essential property stating the Lorentz–invariance of the Dirac equation.176). thereby.243) and.168–10. 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.

341) [K3 . 0. 0).335) shows that we need to demonstrate [J . . w = (0. of course. that any solution Ψ(xµ ) transformed according to (10. 0). . it is met by S(Lη ξ ) as given in the chiral representation by (10. ∂µ ] = [x2 ∂1 − x1 ∂2 . ∂µ ] = [x3 ∂2 − x2 ∂3 .332 Relativistic Quantum Mechanics We will demonstrate now that the ﬁrst condition is satisﬁed by ρ(Lη ξ ).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. ∂µ ] = (10.336).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 (10.243) and. . 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 (10.342) One can readily convince oneself that these results are consistent with (10. The second condition is identical to (10. We have demonstrated. w = (0. ∂µ ] = (10. 1. ∂µ ] = (10.338) [J3 .262). ∂µ ] = [x1 ∂3 − x3 ∂1 .336) We will proceed with this task considering all six cases: [J1 . As mentioned already we will show condition (10.340) [K2 . . w = (0. 0. 0. ∂µ ] = (10. ∂µ ] = [x0 ∂3 + x3 ∂0 .. [K .331) is again a solution of the Dirac equation.337) [J2 . We will proceed by employing the generators (10. 0). ∂µ ] = [x0 ∂2 + x2 ∂0 . 0. ∂µ ] = [x0 ∂1 + x1 ∂0 . i. ϑ = (0. the Dirac equation is invariant under active Lorentz transformations.e. 0. 0). therefore. 0). ∂µ ] = (K )ν µ ∂ν . ϑ = (0.334) for inﬁnitesimal Lorentz transformations Lη ξ . . 1) . Inspection of (10.339) [K1 . ∂µ ] = (10. ∂µ ] = (J )ν µ ∂ν . ∂µ ] = (10.

(10.343) One may also include the electromagnetic ﬁeld in the Dirac equation given in the Schr¨dinger form o ˆ (10.6 above). σ3 )T one obtains then i∂t φ i∂t χ = = ˆ σ · π χ + qV φ + mφ ˆ σ · π φ + qV χ − mχ .350) (10. σ2 .. For this purpose we choose the decomposition Ψ(xµ ) = φ(xµ ) χ(xµ ) .345) holds |qV | << m . The Dirac equation in the Schr¨dinger form reads then o i∂t Ψ(xµ ) = ˆ ˆ where α and β are deﬁned in (10. Equivalently.344) Using the notation σ = (σ1 .6 we assume that the ﬁeld is described through the 4-vector potential Aµ and.347) (10.10 Dirac Particles in Electromagnetic Field We like to provide now a description for particles governed by the Dirac equation which includes the coupling to an electromagnetic ﬁeld in the minimum coupling description. The Dirac equation (10.348) (10.1) i∂t − qV and p by ˆ ˆ π = p − qA . This solution exhibits a timedependence exp[−i(m + )t] where for holds in the non-relativistic limit | | << m.1 in Sect 10.349) We want to focus on the stationary positive energy solution. accordingly. Non-Relativistic Limit We want to consider now the Dirac equation (10.345) in the so-called non-relativistic limit in which all energies are much smaller than m.233) by replacing i∂t by (see Table 10.10.351) X (x ) µ . we deﬁne φ(xµ ) = e−imt Φ(xµ ) χ(x ) = e µ −imt (10. Following the respective procedure developed for the Klein-Gordon equation in Sect.232).g. (10. 10. for the scalar ﬁeld V in (10.345) (10.346) ˆ ˆ ˆ α · π + qV + β m Ψ(xµ ) (10. Accordingly. e.10: Dirac Particles in Electromagnetic Field 333 10.221) reads then [ iγ µ (∂µ + iqAµ ) − m ] Ψ(xν ) = 0 (10. we replace in the Dirac equation the momentum operator pµ = i∂µ by i∂µ − qAµ where q is the charge of the respective particles ˆ (see Table 10. we replace the operator ∂µ by ∂µ + iqAµ .

(10. henceforth. which in the present case states ˆ ˆ2 ˆ ˆ (σ · π)2 = π + i σ · (π × π) . 10.360) For an arbitrary function f (r) holds ( [Aj .358) ( πj πk − πk πj ) = jk [πj . π ] .361) . Φ Using (10.353) (10. i i (10.351) in (10.354) ˆ 0 ≈ σ · π Φ − 2m X .352) allow one to approximate (10. and. ∂k ] + [∂j . ∂k ] + [∂j .357) Equation (10.359) We want to evaluate the latter commutator. accordingly. 10.356). Ak ] i i i i =0 =0 = q q [Aj . Ak ] + q 2 [Aj .7. one can replace X in (10. 5.356) i∂t Φ ≈ Φ + qV Φ . (10.355) ˆ σ·π Φ 2m (10. ˆ Equation (10. ∂k ] + q [Aj . ∂k ] + q [ ∂j .352) ˆ σ · π X + qV Φ ˆ σ · π Φ + qV X − 2mX . (10.349) yields i∂t Φ i∂t X = = ∂t X X Relativistic Quantum Mechanics << m .346.357) for Φ can be reformulated by expansion of (σ · π)2 . One obtains [πj .350. does not need to be considered anymore.353) by X ≈ to obtain a closed equation for Φ ˆ σ·π 2m 2 (10. 10. πk ] = = 1 1 [ ∂j + qAj .348. For this purpose we employ the identity (5. ∂k + qAk ] i i 1 1 1 1 [ ∂j .230). identiﬁes X as the small component of the bi-spinor wave function which. derived in Sect.334 and assume that for the time-derivative of Φ and X holds ∂t Φ << m . due to the m−1 factor. Ak ] ) f = ( ∂j Ak − Ak ∂j + Aj ∂k − ∂k Aj ) f .354) The properties (10. (10. (10. ˆ ˆ For the components of π × π holds ˆ ˆ π×π = jk (10. Ak ] .

r (10.1 which emerged in the Lorentz-invariant theory as an algebraic 2 necessity.2) governing a one-dimensional wave function ψ ∈ C. Dirac Particle in Coulomb Field . πk ] = q q (( ∂j Ak − ∂k Aj )) = − i i ×A jk q = − B i jk (10.365) Comparision of this expression with (5.Spectrum We want to describe now the spectrum of a relativistic electron (q = −e) in the Coulomb ﬁeld of a nucleus with charge Ze.367) .1 particle with a magnetic 2 ﬁeld B. 5.344.e.359. t) + q V (r.10. 10. but rather remains as a steady “guest” of non-relativistic physics with the proper interaction term. i. the interaction qσ · B induces a 1 precession of the spin. (10.363) where we employed B(r. The respective bispinor wave function Ψ(xµ ) ∈ C4 is described as the stationary solution of the Dirac equation (10. 10. t) (10.. t) ≈ ˆ [p − q A(r.343) for the vector potential Aµ = (− Ze2 .6)]. 10. ∂k ] + [∂j . the spin. [πj .357) in the ﬁnal form i∂t Φ(r. t) [see (8.363) allow us to write (10. 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) .366) can be interpreted as a rotation around the ﬁeld B by an angle qtB. Comparision of (10. t) 2m 2m Φ(r. −A).2 around the magnetic ﬁeld. 0.10: Dirac Particles in Electromagnetic Field Using ∂j Ak f = ((∂j Ak )) f + Ak ∂j f ∂k Aj f = ((∂k Aj )) f + Aj ∂k f 335 where ((∂j · · ·)) denotes conﬁnement of the diﬀerential operator to within the brackets ((· · ·)).364) governing a two-dimensional wave function Φ ∈ C2 with the corresponding non-relativistic Schr¨dinger equation (10. In other words. t) = × A(r.223) shows that the propagator in (10.1 with the magnetic ﬁeld. 0. 0) . The formal solution of this equation is Φ(t) = e−iqtB·σ Φ(0) . Ak ] ) f = [ ((∂j Ak )) − ((∂k Aj )) ] f (10.2) it introduces the o extra term qσ · B Φ which describes the well-known interaction of a spin.364) which is referred to as the Pauli equation. using (10.366) (10.362) or. Let us consider brieﬂy the consequences of the interaction of a spin.222. Equations (10. one obtains ( [Aj . o reveals a stunning feature: the Pauli equation does justice to its two-dimensional character. does not leave the theory again when one takes the non-relativistic limit. while agreeing in all other respects with the non-relativistic Schr¨dinger equation (10. t)]2 q − σ · B(r.358.360) and Aµ = (V.

368) ˜ where Ψ(xµ ) and γ µ are deﬁned in (10.370) ˜ [ i˜ µ (∂µ + iqAµ ) − m ] [ i˜ µ (∂µ + iqAµ ) + m ] Ψ(xµ ) = 0 γ γ (10.370) Any solution of (10. ˜ However. 5.369) from the left by γ ν πν + m. µ = ν and altering ‘dummy’ summation ˜ ˜ ˜ ˜ .7 one can write the second term in (10.231). multiplying (10.6. (5. 3 µ=0 ˆ2 (˜ µ )2 πµ = π0 − π .370) for which holds ˜ ˆ ˜ ˆ 3 γ πµ γ πν = ˜ ˆ ˜ ˆ µ=0 µ ν (˜ µ )2 πµ + γ ˆ2 µ.368) is also a solution of (10.368) ˜ ( γ µ πµ − m ) Ψ(xµ ) = 0 . (˜ 0 )2 = 1 and (˜ j )2 = γ 1 γ −1 j = 1. but the converse is not necessarily true.232) in Sect. For this purpose we ‘square’ the Dirac equation.373) (10. γ ˆ ˜ ˆ (10.229). can be rewritten using.370) resembles closely the Klein-Gordon equation (10.372) (10. we solve ˜ [ i˜ µ (∂µ + iqAµ ) − m ] Ψ(xµ ) = 0 γ (10. Employing πµ as deﬁned ˜ in Table 10. This follows from [ i˜ µ (∂µ + iqAµ ) + m ] [ i˜ µ (∂µ + iqAµ ) − m ] γ γ = [ i˜ µ (∂µ + iqAµ ) − m ] [ i˜ µ (∂µ + iqAµ ) + m ] γ γ according to which follows from (10. but diﬀers from it in an essential way. Equation (10.s.369).371).228) and in (10.371) such that we can conclude that (10.374) The ﬁrst term on the r. according to (10. 2. noting from (10.e.336 Relativistic Quantum Mechanics For the purpose of the solution we assume the chiral representation.h. is a solution of (10.370).370) is obtained then ˜ [ i˜ µ (∂µ + iqAµ ) + m ] Ψ(xµ ) γ is a solution of (10.369).ν=1 µ =ν γ µ γ ν πµ πν .374). The diﬀerence arises from the term γ µ πµ γ ν πν in (10. γ ˆ2 ˆ2 (10. ˜ (10. (5. once a solution Ψ(xµ ) of (10. i.1 one can write (10. 3. 1. indeed. 10.230).375) Following the algebra that connected Eqs. respectively.180).369) For our solution we will adopt presently a strategy which follows closely that for the spectrum of pionic atoms in Sect. This yields ˜ [ i˜ µ (∂µ + iqAµ ) + m ] [ i˜ µ (∂µ + iqAµ ) − m ] Ψ(xµ ) γ γ ˜ = (˜ µ πµ γ ν πν − m2 ) Ψ(xµ ) = 0 . ˜ ˜ ˆ ˆ (10.230) γ µ γ ν = − γ ν γ µ .

πj ] in (10. γ j ] [ˆ0 .379) The commutators [ˆ0 .381) According to (10.378) According to the deﬁnition (10. −i∂j ] = − [ (∂t + iqA0 ) ∂j − ∂j (∂t + iqA0 ) ] f i ((∂j qA0 )) f (10.ν=1 µ =ν = [˜ µ .ν=1 µ =ν σ 0 0 −σ · (( qA0 )) (10.376) This expression can be simpliﬁed due to the special form (10. 2.10: Dirac Particles in Electromagnetic Field indices. it holds 1 4 [˜ µ .367) of Aµ .ν=1 µ =ν (10.377) which follows readily from the deﬁnition (10.380) γ µ γ ν πµ πν = i ˜ ˜ ˆ ˆ µ.229). γ ν ] [ˆµ . Since [ˆµ .e. πν ] γ ˜ π ˆ µ. one can summarize (10. t) is a suitable test function and where ((· · ·)) denotes the range to which the derivative is limited.ν=1 µ =ν 3 j=1 3 3 = 1 4 1 2 1 [˜ . πj ] . γ j ] γ ˜ = = 2 0 1 1 1 0 1 σj 0 0 −σ j 0 σj −σ j 0 − 0 σj −σ j 0 0 1 1 1 0 1 (10. γ µ γ ν πµ πν = ˜ ˜ ˆ ˆ µ. i. γ ˜ π ˆ j=1 (10. due to A = 0.376–10.344). r2 (10. Altogether. γ ] [ˆ0 .382) .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.367) holds qA0 = r ˆ Ze2 .367) and the deﬁnition (10. π0 ] γ ˜ π ˆ j=1 = [˜ 0 .344) π ˆ [ˆ0 .ν=1 µ =ν 337 1 2 ( γ µ γ ν πµ πν + γ ν γ µ πν πµ ) ˜ ˜ ˆ ˆ ˜ ˜ ˆ ˆ µ.380) where f = f (r.378) can be evaluated using (10. πj ] + γ ˜ π ˆ 4 0 j [˜ j . πν ] = 0 π ˆ for µ. the commutators [˜ 0 . πν ] γ ˜ π ˆ µ. πj ] π ˆ = = (−i∂t + qA0 . γ j ] are γ ˜ [˜ 0 . γ 0 ] [ˆj .. γ ν ] [ˆµ . 3 (10. ν = 1.10.

couples the orbital angular momentum of the electron to its spin. .387) The expression (10.5 the operator iσ · r is block-diagonal in this basis such that only ˆ 1 the states for identical j. m values are coupled. The term iσ · r.383) yields the purely spatial four-dimensional diﬀerential equation + Ze2 r 2 + 2 + i σ·r ˆ 0 0 −σ · r ˆ Ze2 − m2 r2 Φ(r) = 0 . 2 |ˆ)} as given in (6. .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. .189) for the Laplacian and expansion of the term (· · ·)2 result in the twodimensional equation 2 ∂r − ˆ L2 − Z 2 e4 iσ · r Ze2 ˆ 2Ze2 + r2 r + 2 − m2 r φ± (r) = 0 . 1 |ˆ). 1 |ˆ). 1 |ˆ) ) . it is this quantity that we want to determine. 10. r r 2 2 2 j = 1. (10.384) into (10. 6. we express 2 the solution of (10. In the latter case. Such solutions are of the form ˜ ˜ Ψ(xµ ) = Φ(r) e−i t . 6. only the two states {Yjm (j − 2 . Accordingly.338 Relativistic Quantum Mechanics where r = r/|r| is a unit vector.383) We seek stationary solutions of this equation..375) the ˆ ‘squared’ Dirac equation (10.389) According to the results in Sect.1 . m = −j.381). (10. + j } (10. Yjm (j + r 2 1 1 . a solution of the form ∼ Y m (ˆ) can be obtained. Yjm (j + 1 .368) reads − Ze2 ∂t − i r 2 + 2 + i σ·r ˆ 0 0 −σ · r ˆ Ze2 − m2 r2 Ψ(xµ ) = 0 (10. 6. −j + 1. We note that these states are also eigenstates of the angular r 2 . (10.384) can be interpreted as the energy of the stationary state and.374). (10. .6. (10. .385) 1 We split the wave function into two spin.2 components ˜ Ψ(r) = and obtain for the separate components φ± (r) Ze2 + r 2 ˜ φ+ (r) ˜ φ− (r) (10.e.388) in terms of states introduced in Sect. i.5 which describe the coupling of orbital angular momentum and spin 1 {( Yjm (j − 2 .147.148).386) + 2 ± iσ · r ˆ Ze2 − m2 r2 φ± (r) = 0 . in r ˆ fact. Combining this result with (10.191) solved in Sect. however. (10. . is genuinely two-dimensional and. hence. 2 3 2 . Insertion of (10.

(6. and except for the fact that 2 2 we have two sets of values for λ1. We select.396) and (10. the missing additive term − 1 .2 (j) + 1) 2 Ze2 + + r2 r 2 − m2 f1. namely.397) This equation is identical to the Klein-Gordon equation for pionic atoms written in the form (10.5 [c.2 (r) = 0 (10. we want to determine solely the spectrum.186)].392) reads then in the diagonal representation 2 ∂r − λ1. in the present treatment. property (6. (6. rather than = 0. 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. 10. 1. as in the case of pionic atoms. together with the orthonormality of these two states leads to the coupled diﬀerential equation 2 ∂r + 2Ze2 + r 2 −m 1 0 0 1 ±iZe2 1 2 3 2 (10. which states that the states Yjm (j ± 1 .200). hence. conclude that the . m and expand φ± (r) = Using σ · r Yjm (j ± 1 .390) (10.391) derived in Sect.10: Dirac Particles in Electromagnetic Field 339 ˆ momentum operator L2 [cf. . λ1 (j) and λ2 (j). 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 ..392) into diagonal form.395. 1 |ˆ) = − Yjm (j ˆ r 2 2 1 2 h± (r) g± (r) Yjm (j − 1 .2 (j). unaltered.392) h± (r) g± (r) = 0. namely. 6.10. 1 |ˆ) + r Yjm (j + 1 . 1 |ˆ) r 2 (10. .2 (j) being 2 j = 1 .393) but do not explicitly consider further the wavefunctions. . Obviously. . 3 . However. therefore. Any similarity transformation leaves the ﬁrst term in (10.2 (j) as given by (10. a speciﬁc pair of total spin-orbital angular momentum quantum numbers j. involving the 2 × 2 unit matrix.151). Since.394) Equation (10. . We can.f. 1 |ˆ) r 2 2 2 2 r r . such transformation can be chosen as to diagonalize the second term. (10. the values of the argument of λ1. not the wave functions. − 1 r2 (j − 1 )(j + 1 ) − Z 2 e4 2 2 ±i Ze2 (j + )(j + ) − Z 2 e4 We seek to bring (10. 1 |ˆ) are r 2 2 ˆ eigenfunctions of L2 .151].392). the eigenvalues are independent of m. except for the slight diﬀerence in the expression of λ1.396) and λ2 (j) [λ2 (j) + 1] (10. .199).2 (j)[λ1.395) (10.

1 = 1 + m Z 2 e4 (n + 1 + (j+ 1 )2 − Z 2 e4 )2 2 Z 2 e4 (n + (j+ 1 )2 − Z 2 e4 )2 2 .340 Relativistic Quantum Mechanics spectrum of (10.401) corresponds to a non-relativistic state 2 with quantum numbers n. . (10. of these equations describe the binding energy. .400) corresponds to a non-relativistic state with quantum numbers n.403) √ Z 2 e4 2 ED (n. . . . . and spin-orbital angular momentum j = + 1 .400) (10. One may also state 2 this in the following way: (10. It is given by n = n + + 1 where is the orbital angular momentum quantum number and n = 0.399) 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 = 0.398) 2 = 1 + m . . 2. j = 1 . the stationary states have binding energies E = − mZ 2 e4 n = 1. n − 1 . accordingly. One can equate (10. (10. m = −j. . . . 1. In case of non-relativistic hydrogen-type 1 atoms. m = − . Obviously. m) 2 n = 1. ( +1) − Z 2 e4 )2 (10. j . . . Using (10. . 10. .396). . 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.397) is again given by eq. (10. . the second term on the r. j = + 1 . j .401) if one attributes to the respective 1 states the angular momentum quantum numbers = j + 2 and = j − 1 . . = 1 + m Z 2 e4 √ ( n − − 1 + 2 − Z 2 e4 )2 . One obtains in case of (10. n − 1 . . . 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 . . . .s. 3. 1 ms = ± 2 (10. .396) we obtain. 2n2 = 0. . . . . . 2 2 1 ≈ m − (10. (10. . −j + 1. (10. . 10.402) In this expression n is the so-called main quantum number. . . j where 1 corresponds to λ1 (j) as given in (10. . . 2.395) and 2 corresponds to λ2 (j) as given in (10. . . . . 2 2 m = −j. .401) These expressions can be equated with the non-relativistic spectrum.402) with (10. . − + 1. and spin-orbital angular momentum j = − 1 . . j = 1. 1. . m) = 1 + m (n− + . .203).395. . . counts the nodes of the wave function. These considerations 2 are summarized in the following equations 1 ED (n. 3 .404) = 0. −j + 1. . m = −j. including spin. j = − 2 . .400) and (10.h. . . The magnitude of the relativistic eﬀect 2 2 e4 .399) n = 0. .2 . . 1. albeit with some modiﬁcations. . 1. 1. 2.398. −j + 1. 2. .

notation 1s 1 2 non-rel. ± 3 . The energies were evaluated with m = 511. 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. j 1 2 1 2 1 2 3 2 1 2 1 2 3 2 3 2 5 2 341 rel. in fact. n − 1 j = 1 2 1 2 ± m = −j. This is clearly demonstrated by the comparision of non-relativistic and relativistic spectra of a hydrogen-type atom with Z = 100 in Table 10.0041 keV and e2 = 1/137.405) In order to demonstrate relativistic eﬀects in the spectrum of the hydrogen atom we compare in Table 10. the 2p 3 states with r 2 2 2 2 1 involve four degenerate states corresponding to Y 3 m (1. .. relate closely to the corresponding nonrelativistic states.036 by means of Eqs. However.10.e.60601 -3. is in the range of 10 eV. Degeneracies are denoted by ⇑. binding energy / eV Eq.402) -13. the nonrelativistic degeneracy for the six 2p states of the hydrogen atom: in the present. j (10. . . for example. i. (10. 10. The reason is that the mean kinetic energy of the electron in the hydrogen atom. Of particular interest is the eﬀect of spin-orbit coupling which removes.40146 ⇑ ⇑ -1. 1 |ˆ) for m = ± 1 . 2.3.2: Binding energies for the hydrogen (Z = 1) atom.405) in terms of the non-relativistic quantum numbers n. r 2 2 2 .40147 -1. . = 0.405)] spectrum of the hydrogen atom.404) ﬁnally into the single formula ED (n. (10. (10. the non-relativistic and relativistic energies are hardly discernible. much less than the rest mass of the electron (511 keV). 2 |ˆ) for m = ± 1 . 1.51176 ⇑ ⇑ ⇑ ⇑ 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. .40151 ⇑ .60583 -3.1. binding energy / eV Eq.e. 0 0 1 1 0 1 1 2 2 spinorbital ang. i. (10.551177 ⇑ . 10. m) = m 1 + Z 2 e4 (n−j −1 + 2 (j+ 1 )2 − Z 2 e4 )2 2 n = 1.403.1. −j + 1. 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 table entries demonstrate that the energies as given by the expression (10. .402.405 ) -13. . .402)] and the relativistic [cf.405). mom.551178 ⇑ .3. case these six states are split into energetically diﬀerent 2p 1 and 2p 3 states. j. One can combine the expressions (10. ..2 the non-relativistic [cf. relativistic. . . (10.10: Dirac Particles in Electromagnetic Field main quantum number n 1 2 2 2 3 3 3 3 3 orbital angular mom.551176 spectr. .

(10.407) 1 2 This expression allows one. 0. Introducing α = Z 2 e4 1 and β = j + 2 (10. 0) (10.15.8 ⇑ . .0041 keV and e2 = 1/137.405) to order O(Z 4 e8 ). mom. for example. 0 0 1 1 0 1 1 2 2 spinorbital ang.402) -136. 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 10.405) reads 1 (10.9 ⇑ .3: Binding energies for the hydrogen-type (Z = 100) atom. 32 (10. 10.10.1 -34. Degeneracies are denoted by ⇑. 0. to estimate the diﬀerence between the energies of the states 2p 3 and 2p 1 (cf. An example for such potential is the Coulomb potential V (r) = −Ze2 /r considered further below. The energies were evaluated with m = 511. (10.15.408) Radial Dirac Equation We want to determine now the wave functions for the stationary states of a Dirac particle in a 4-vector potential Aµ = (V (r).409) where V (r) is spherically symmetric.6 -42. j.1 ⇑ .3 spectr.206) provides in the present case ED (n. binding energy / keV Eq. binding energy / keV Eq.402. It holds for n = 2 and j = 3 . In order to investigate further the deviation between relativistic and non-relativistic spectra of hydrogen-type atoms we expand the expression (10. (10.406) α √ 1+ (n − β + β 2 −α)2 The expansion (10.2.405).342 main quantum number n 1 2 2 2 3 3 3 3 3 orbital angular mom.405 ) -161.2 -17. 1 2 2 2 2 E (2p 3 ) − E (2p 1 ) ≈ − 2 2 mZ 4 e8 2 · 23 1 −1 2 = mZ 4 e8 .036 by means of Eqs. m) ≈ m − mZ 2 e4 mZ 4 e8 − 2n2 2n3 1 j+ − 3 4n + O(Z 6 e12 ) .3).0 ⇑ ⇑ -15. Tables 10.35. (10. We assume for the wave function the stationary state .1 ⇑ ⇑ ⇑ ⇑ rel. j 1 2 1 2 1 2 3 2 1 2 1 2 3 2 3 2 5 2 Relativistic Quantum Mechanics non-rel.

. 1 |ˆ) r 2 2 (10.410) 1 where Φ(r).412) are four-dimensional with r-dependent wave functions. m-values. pp.232.411) (10. 6.e. 6. 1 |ˆ) . The arguments above allow one to eliminate the angular dependence by expanding Φ(r) and X (r) 1 in terms of Yjm (j + 2 . ˆ σ · p X + m Φ + V (r) Φ ˆ σ · p Φ − mX + V (r) X = = Φ X (10.10. 1 |ˆ) and Yjm (j − 1 . This term has been discussed in detail in Sect. according to (10. the term is block-diagonal in the space spanned by the states Yjm (j ± 1 . 2 |ˆ) + r Yjm (j − 2 . in particular.. 2 |ˆ) and does r 2 ˆ has odd parity and it holds [c. 1 |ˆ) .417) c(r) d(r) X (r) 1 1 1 1 Yjm (j + 2 . 10. 2 |ˆ) + r Yjm (j − 2 .345). but are timeindependent.413) ˆ σ · p g(r) Yjm (j − 1 .5. 10.415) 1 ˆ σ · p rg(r) Yjm (j − 2 . = r (10. 1 |ˆ) r 2 2 = i ∂r + j+ r 1 2 3 2 f (r) Yjm (j − 1 .416) The diﬀerential equations (10. r 2 2 (10. 168]: the term is a scalar (rank zero tensor) in the space of the spin-angular momentum states Yjm (j ± 1 .414) ˆ σ · p rf (r) Yjm (j + 1 . (6.414) These equations can be brought into a more symmetric form using ∂r + which allows one to write (10. X (r) ∈ C2 describe the spatial and spin. i. (10. The Dirac equation reads then. 1 |ˆ) r 2 1 1 = ∂r r r r (10. 10.10: Dirac Particles in Electromagnetic Field form Ψ(xµ ) = e−i t 343 Φ(r) X (r) . 2 |ˆ) r Φ(r) r .198)] not couple states with diﬀerent j. r 2 (10. 6.411.e.2 degrees of freedom. 1 |ˆ) introduced in r 2 2 1 Sect. σ · p ˆ σ · p f (r) Yjm (j + 1 . 1 |ˆ) r 2 2 = i j+ ∂r + r j+ r 1 2 1 r f (r) Yjm (j − 2 . 1 |ˆ). 2 |ˆ) r r r . 1 |ˆ) r 2 = i ∂r − 1 2 r g(r) Yjm (j + 1 .5 [see. r r 2 2 2 a(r) b(r) 1 1 1 1 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. 1 |ˆ) r 2 2 = i ∂r + −j r 1 g(r) Yjm (j + 2 . i.197.f.413.

there exist two independent solutions (10.157).421) g2 (r) X (r) 1 1 − Yjm (j + 2 .411. c(r) are coupled. Presently. m values. 10.412).416). only a(r).417) of the form f1 (r) i Yjm (j + 1 .418) holds for f1 (r). 2 |ˆ) r r f2 (r) 1 1 r Φ(r) i r Yjm (j − 2 . m pair contribute. using (10. 1 |ˆ) r Φ(r) 2 2 r = (10.420) g1 (r) X (r) 1 1 − Yjm (j − 2 . and multiplying by r results in the following two independent pairs of coupled diﬀerential equations i ∂r − j+ r j+ r 1 2 d(r) + [ m + V (r) − ] a(r) = 0 i and i ∂r + 1 2 a(r) + [ −m + V (r) − ] d(r) = 0 (10. (10. According to (10. 2 |ˆ) r r where the factors i and −1 have been introduced for convenience. d(r) are coupled and b(r). g1 (r) ∂r − j+ r j+ r 1 2 g1 (r) + [ − m − V (r) ] f1 (r) = 0 ∂r + and for f2 (r).423) . Inserting (10. we consider the case that only states for one speciﬁc j. Accordingly.417) into (10.419) Obviously.415.418) ∂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. such expansion must include states with all possible j. 2 |ˆ) = (10. g2 (r) ∂r + 1 2 f1 (r) − [ + m − V (r) ] g1 (r) = 0 (10. the orthonormality property (6.422) 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. 10.344 Relativistic Quantum Mechanics In general.

the 2 equations determine.422).425) r 0 covers states with angular momentum 1s 1 .405). = 0. Equation (10. Hence. . 3p 1 .422) determines solutions of the form (10. According to this procedure.e.367) which is spherically symmetric such that equations (10. We note that (10. 3d 5 . 2 2 2 2 2 2 = j − 1 2 and. 1.422. except for the opposite sign of the term (j + 1 ). .. 10.423). . and obtains ﬁnally a polynomial function p(r) such that rγ exp(−µr)p(r) solves (10. enforcing the polynomial to be of ﬁnite order leads to discrete eigenvalues . .Wave Functions We want to determine now the wave functions of the stationary states of hydrogen-type atoms which correspond to the energy levels (10. one demonstrates then that the wave functions for r → ∞ behaves as exp(−µr) for some suitable µ.422) describes the states 2p 1 .423) are identical.420). 2 2 2 (10. in the non-relativistic limit wave functions f2 (r) 1 1 i Yjm (j − 2 . the ones given in (10. n − 1.422). Similarly.. Dirac Particle in Coulomb Field . (10. Φ in (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. one demonstrates ﬁrst that the wave function at r → 0 behaves as rγ for some suitable γ. to states with angular momentum = j + 2 . 2 |ˆ) r Ψ(xµ ) ≈ . (10. together with the appropriate boundary conditions at r = 0 and r → ∞. (10. 2p 3 . Behaviour at r → 0 We consider ﬁrst the behaviour of the solutions f1 (r) and g1 (r) of (10.409).422) corresponds to states f1 (r) Yjm (j + 1 . determining wave functions of the type (10.422) and (10. 3p 3 . (10.e. (10. In the non-relativistic limit. namely.405) of the relativistic hydrogen atom the corresponding states have quantum numbers n = 1. . 3d 3 . We assume the 4-vector potential of pure Coulomb type (10.10. 2s 1 .420) is the large component and X is the small component. etc.405). Hence. etc. . for small r. 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) − = 0.421). 2. correspodingly the states We consider ﬁrst the solution of (10. 3s 1 .10: Dirac Particles in Electromagnetic Field 345 Equations (10. . .422) near r = 0. The solution of (10.426) . i.422). 1 |ˆ) r i 2 2 .424) Ψ(xµ ) ≈ r 0 1 i. the radial wave functions for Dirac particles in the potential (10. According to the discussion of the spectrum (10.423) apply for V (r) = −Ze2 /r.

432) = = − ( − m.428) = 0. The assumed r-dependence in (10. Only the exponentially decaying solution is admissable and. g1 ∼ exp(± m2 − 2 r). ) f1 (r) ( + m ) g1 (r) (10. hence. for the particle density holds then ρ(r) ∼ |rγ−1 |2 = Behaviour at r → ∞ For very large r values (10.430) 2 Note that the exponent in (10. 10. µ = m2 − 2 (10. g1 (r) →∞ ∼ e−µr . One obtains γ = ± (j + 1 )2 − Z 2 e4 . in case j + 1 )2 < Ze2 . (10. according to (10. g1 (r) →0 ∼ b rγ (10.434) √ The solutions of these equations are f1 . Such r-dependence 2 would make the expectation value of the potential r2 dr ρ(r) 1 r (10. 10.427).266. we conclude f1 (r) →∞ ∼ e−µr . (10.422) becomes ∂r g1 (r) ∂r f1 (r) Iterating this equation once yields 2 ∂r g1 (r) 2 ∂r f1 (r) 1 . for small r.433) = = ( m2 − 2 ) g1 (r) ( m2 − 2 ) f1 (r) (10.346 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 .429) This equation poses an eigenvalue problem (eigenvalue −γ) for proper γ values.427) with 1 γ = (j + )2 − Z 2 e4 . becomes imaginary. determined that the solutions f1 (r) and g1 (r).427) makes only the positive solution 2 possible.267. assume the r-dependence in (10.427) = = 0 0. hence. (10. r2 (10.431) inﬁnite since.420). We have.435) .

Equation (10. √ g1 (r) = − m − a e−µr .440) and (10.10. (10. hence.434) results in √ √ (m − ) m + a − (m − ) m + a √ √ (m + ) m − a − (m + ) m − a which is obviously correct.435 ). Solution of the Radial Dirac Equation for a Coulomb Potential To solve (10. g1 (r) = φ1 (r) − φ2 (r) ˜ (10.422) for the Coulomb potential V (r) = −Ze2 /r We assume a form for the solution which is adopted to the asymptotic solution (10.442) which leads to a partial cancellation of the asymptotically dominant terms. (10. Accordingly.445) .439) − m− e g1 (r) ˜ where µ is given in (10.441) correspond to (10.10: Dirac Particles in Electromagnetic Field For bound states holds 347 < m and.441) The last two terms on the l.438) (10. without loss of generality we can choose ˜ f1 (r) = φ1 (r) + φ2 (r) .440) (10. we set f1 (r) g1 (r) = = √ √ ˜ m + e−µr f1 (r) −µr = = 0 0 (10. In the present case the functions f1 and g1 cannot be chosen ˜ ˜ identical due to the terms in the diﬀerential equations contributing for ﬁnite r.422 ) leads to √ √ j+1 Ze2 ˜ 2 − m − [∂r − ] g1 + m + ˜ f1 + r √r √ ˜ (m − ) m + g1 − (m − ) m + f1 = 0 ˜ 1 2 √ √ j+2 ˜ Ze m + [∂r + ] f1 + m − g1 − ˜ r r √ √ ˜ (m + ) m − f1 + (m + ) m − g1 = 0 ˜ (10. However.444) (10.s.h.437) (10.436). of both (10.436) f1 (r) = Insertion into (10.443) From this results after a little algebra j+1 m + Ze2 2 ] (φ1 − φ2 ) − (φ1 + φ2 ) + φ2 = 0 ρ m− ρ j+1 m − Ze2 2 [∂ρ + ] (φ1 + φ2 ) + (φ1 − φ2 ) − φ2 = 0 .434) √ m + a e−µ r . We also introduce the new variable ρ = 2µ r . µ is real. ρ m+ ρ [∂ρ − (10.437) where they ˜ ˜ cancelled in case f1 = g1 = a. Let us consider then for the solution of (10.

450) follows Ze2 m2 − 2 βs − βs−1 =0 (10.447) leads to (s + γ) αs + (j + 1 ) βs − √ 2 s Ze2 m2 − 2 αs (10. Inserting (10.426–10.451) follows βs−1 = = Ze2 s+γ + √ m2 − s+γ − βs + m2 Z 2 e4 m2 − 2 − (j + 1 )2 2 2 √ Ze m2 − 2 2 s+γ − βs (10. Using (10. (10.430) one can write this βs = s+γ − 2 √ Ze m2 − 2 s (s + 2γ) βs−1 .430)].430) which conform to the proper r → 0 behaviour determined above [c.449) into (10.449) for γ given in (10.454) .446.452) From (10. 10.448.453) (s + γ)2 + Z 2 e4 − (j + 1 )2 2 2 √ Ze m2 − 2 βs . 10.450) −√ mZe2 m2 − 2 βs ρs+γ−1 = 0 mZe2 m2 − αs (s + γ) β2 + (j + 1 ) αs + √ 2 s 2 +√ From (10.348 Relativistic Quantum Mechanics Addition and subtraction of these equations leads ﬁnally to the following two coupled diﬀerential equations for φ1 and φ2 ∂ρ φ1 + ∂ρ φ2 + j+ ρ 1 2 Ze2 mZe2 φ1 − √ φ2 = 0 m2 − 2 ρ m2 − 2 ρ mZe2 Ze2 φ1 + √ φ1 + √ φ2 − φ2 = 0 m2 − 2 ρ m2 − 2 ρ j+ ρ φ2 − √ 1 2 (10.446) (10. 10.448) φ2 (ρ) = ρ (10.f.447) of the form n φ1 (ρ) = ργ s=0 n γ s=0 αs ρ2 βs ρ2 (10.446 .447) We seek solutions of (10. (10. = m− Ze2 βs s + γ − √m2 − 2 2 (10.451) 1 √mZe − (j + 2 ) 2 2 αs .

10. x) = 1 + a a(a + 1) x2 x + + .460) (10. c. 3p 1 . 2γ + 1.405) allows one to identify n = n − j + 2 which.10: Dirac Particles in Electromagnetic Field Deﬁning so = √ one obtains βs = = . (s − so ) β0 s! (2γ + 1)(2γ + 2) · · · (2γ + s) (10.458) or. .461) In order that the wave functions remain normalizable the power series (10. The expressions (10.462) Z 2 e4 1 + 1 (n + (j+ 2 )2 − Z 2 e4 )2 This expression agrees with the spectrum of relativistic hydrogen-type atoms derived above and 1 given by (10.457) One can relate the polynomials φ1 (ρ) and φ2 (ρ) deﬁned through (10.448.456) and (10.455) s − so βs−1 s(s + 2γ) (s − 1 − so )(s − so ) βs−2 (s − 1)s (s − 1 + 2γ)(s + 2γ) (1 − so )(2 − so ) . i. equivalently.459) − 2 √mZe m2 − 2 (10.455) are ﬁnite. 10. 1. .461 for values given by (10.. .456.456) − so 2 √mZe m2 − 2 (−so )(1 − so )(2 − so ) .10. 2. α + 1. x) . with the associated Laguerre polynomials L(α) = F (−n. c c(c + 1) 2! (10. Comparision with (10.449) and (10. in fact. . 2. n It holds φ1 (ρ) φ2 (ρ) = = ργ F (−so . .405). so = n .457) for βs and αs imply that so must then be an integer. For example. ρ) .405) and the ensuing so values ( 10. 3d 3 holds n = 1..448. 1. We can. 10.. m (n ) = .430. 2γ + 1. ρ) so β0 ργ F (1 − so . According to the deﬁnitions (10. 10. for the states 2p 1 .457) with the conﬂuent hypergeometric functions F (a.e. conclude 2 2 2 that the polynomials in (10. hence. β0 j+ 1 2 (10. (10. = From (10.455) this conﬁnement of so implies discrete values for . (s − so ) β0 s! (2γ + 1)(2γ + 2) · · · (2γ + s) (10. . n = 0.452) follows j+ αs = 1 2 349 Ze2 m2 − 2 −γ (10. . . .449) must be of ﬁnite order. namely. is an integer. This requires that all coeﬃcients αs and βs must vanish for s ≥ n for some n ∈ N. .

2µr) (10. can follow the procedure adopted for the wave functions of the non-relativistic hydrogen atom and will not be carried out here.463) where Φ and X . m|xµ ) = e−i t iF1 (r) Yj.466) F1 (r) = F− (κ|r) .421) with radial wave functions f1 (r).m (j + 1 .469) This formula has been adapted from ”Relativistic Quantum Mechanics” by W.465) (10.158) of the latter states absorb the angular integral in (10. 2µr) (2µ) 2 Γ(2γ + 1) 3 N = and m 4m )Γ(2γ + n + 1) (n +γ)m (n +γ)m (10. Sect. 6.267). (n + γ)m N (2µr)γ−1 e−µr − κ F (−n . (Springer. Berlin. 10.461).m (j − 1 . The coeﬃcients β0 in (10. g1 (r) determined by (10. The complete wave function is given by the following set of formulas 1 Ψ(n.438.464) 0 The evaluation of the integrals. 10. is time-independent. j.147. and (10.421) correspond to non-relativistic states with orbital angular momentum 1 1 = j + 2 .461).442). m. 1 |ˆ) 2 2 r G1 (r) Yj.460. given by expression (10. 2γ + 1. = j + 2 .421).463) and yield [note the 1/r factor in (10. as in (10. They are described through quantum numbers n. 1 |ˆ) in (6.460. j. 9.468) −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 . are two-dimensional vectors determined through the explicit form of the spin-orbital angular momentum states Yjm (j ± 1 .460.148). Due to the form (10. = j + 2 . 10. 2γ + 1.157. Greiner.410) of the stationary state wave function the density ρ(xµ ) of the states under consideration.467) ± n F (1 − n .6. (10. where2 F± (κ|r) = G1 (r) = F+ (κ|r) . 10.439). (10. 1 |ˆ) 2 2 r κ = j+ 1 2 (10. The wave functions (10. which involve the conﬂuent hypergeometric functions in (10.461) are to be chosen to satisfy a normalization condition and to assign an overall phase.350 Relativistic Quantum Mechanics Altogether we have determined the stationary states of the type (10. 6. The orthonormality 2 2 r properties (6.421)] ∞ dr (|f1 (r)|2 + |g1 (r)|2 ) = 1 (10. 2 . The normalization integral is then ∞ 0 π 2π r2 dr 0 sin θdθ 0 dφ (|Φ(r)|2 + |X (r)|2 ) = 1 (10. 1990).

469). 3p 3 .421) are governed by the radial Dirac equation (10.m (j − 1 . 2 |ˆ) r Ψ(xµ ) ≈ e−1 t . The radial wave functions f2 (r) and g2 (r) in (10. hence. become f2 (r) 1 1 i Yjm (j − 2 . 2 accordingly. 2s 1 .471) (10. (10.469) that the following wave functions result Ψ(n. this wavefunction has an orbital angular momentum quantum number = j − 1 and. m|x ) = e −i t iF2 (r) Yj. 1 |ˆ) 2 2 r κ = −j − 1 2 (10. in the non-relativistic limit. G2 (r) = F+ (κ|r) .423) which diﬀers from the radial Dirac 1 equation for f1 (r) and g1 (r) solely by the sign of the terms (j + 2 )/r.10: Dirac Particles in Electromagnetic Field 351 We want to consider now the stationary states of the type (10. 3d 5 . tracing all steps which lead from (10. obtained closed expressions for the wave functions of all the stationary bound states of relativistic hydrogen-type atoms. describes the complementary set of states 1s 1 .421) which. One can verify. 1 |ˆ) 2 2 r G2 (r) Yj. where F± (κ|r) are as given in (10. .467–10.10.470) r 0 Obviously. We have. 2p 3 . not not 2 2 2 2 2 2 covered by the wave functions given by (10. j. 3s 1 .422) to (10.465–10.m (j + 1 .472) F2 (r) = F− (κ|r) .469). etc. = j − µ 1 2 .

352 Relativistic Quantum Mechanics .

i. ϑ)µ ν . expressing its dependence on w.229).1) (11. z ∈ C ˜ Because of its block-diagonal form each of the diagonal components of S(z)..3) (11.4) = exp = w − iϑ . We will proceed by building as much as possible on the results obtained sofar in the chiral representation of the Dirac equation. isomorphic to the natural representation. This is.262). We will characterize the space on which the transformations a(z) We will see below that the representations a(z) and b(z) are.225–10. ϑ). in particular. ϑ)µ ν correspond to diﬀerent a(z) and b(z).2) (11. may be exploited to express the Lorentz-invariant equations of relativistic quantum mechanics. in fact. must be two-dimensional irreducible representations of the Lorentz group. a complex variable z. The lower dimensionality of a(z) and b(z) implies. Starting point is the Lorentz transformation S(w.. what will be achieved in the following. diﬀerent L(w. indeed.e. i. 1 353 . ϑ through 3. ϑ) for the ˜ in the chiral representation as given by (10. 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. 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. in a sense. ˜ We have altered here slightly our notation of S(w.Chapter 11 Spinor Formulation of Relativistic Quantum Mechanics 11. a(z) and b(z). This transformation bispinor wave function Ψ can be written ˜ S(z) a(z) b(z) z = = a(z) 0 0 b(z) exp 1 z·σ 2 1 − z ∗· σ 2 (11.1 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.e.

9) a∗ (z) = exp 2 . neutrino equation. Aν . b(z). For ϑ = (0.e. ﬁnally. in 2 fact. θ ∈ R3 . (11. formulate the Dirac equation. Ψ2 (xµ ) transform according to a(z) whereas Ψ3 (xµ ). ± 1 represents the familiar spin– 1 states. − 1 2 2 z = iϑ. −1 2 2 |1. This characterization allows one to draw conclusions regarding the state space in which a(z) and b(z) operate. β. are elements of SU(2) and correspond. +1 2 |1. Relationship Between a(z) and b(z) The transformation b(z) can be related to the conjugate complex of the transformation a(z). +1 2 2 |1. will establish the map between L(w. In this case a(z) and b(z) are identical and read a(i ϑ) = b(i ϑ) = exp 1 − ϑ·σ 2 . Ψ4 (xµ ) in case z = w + iϑ for w = 0. to 1 ∗ ∗ z ·σ (11. (11.5) The transformations in this case.. namely. however. to the rotational transformations of spin. Ψ4 (xµ ) transform according to ˜ ˜ ˜ and Ψ b(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 . ϑ ∈ R3 . γ (see Chapter 5). (11. In this case holds a(z) = b(z) ˜ 1 (xµ ). a space of vectors state1 for which holds state2 state 1 ∼ | 1 .6) as given by (5. β.354 Spinor Formulation and b(z) act. ϑ ∈ R3 . ϑ)µ ν and a(z). express 4-vectors Aµ .8) We like to stress that there exists. a distinct diﬀerence in the transformation behaviour of ˜ ˜ ˜ ˜ Ψ1 (xµ ). −1 2 2 z = iϑ.7) where “∼” stands for “transforms like”. usually expressed 2 as product of rotations and of functions of Euler angles α. + 1 2 2 state 2 ∼ | 1 . for z = iϑ. A First Characterization of the Bispinor States We note that in case w = 0 the Dirac transformations are pure rotations.243).. and the Klein–Gordon equation in the spinor representation. 2 2 2 ˜ Since a(z) acts on the ﬁrst two components of the solution Ψ of the Dirac equation. i. Ψ2 (xµ ) and Ψ3 (xµ ). the so-called spinor space. the operator ∂µ and the Pauli matrices σ in the new representation and. 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. Here | 1 . ϑ ∈ R3 (11.1 states as described by Dm m (ϑ). i.e.

Ψ (xµ ) even though these pairs of ˜ ˜3 ˜4 featuring the components Ψ1 (x ˜ 2 components transform independently of each other. . This demonstrates that a(z) is the ˜ transformation which characterizes both components of S(z). 11. 11. (11. a∗ (z) −1 = exp 1 ∗ z σ∗ 2 −1 . 11. One can readily verify [c.18) with a(z) given by (11.14) .17) We conclude. (5.13) Explicit matrix multiplication using (5. Spatial Inversion ˜ One may question from the form of S(z) why the Dirac equation needs to be four-dimensional. that they alter w into −w. one can show −1 −1 −1 = −σ1 = σ2 = −σ3 ∗ = −σ1 ∗ = −σ2 σ2 −1 (11. using f (a) −1 = f ( a −1 ). 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. that the transformation (11. Hence.15) (11.12) To prove (11. The eﬀect of inversion on Lorentz transformations is.16) σ = −σ ∗ . σ3 ).2. (11. µ ).12) yields σ1 σ3 or in short σ Similarly.11. (11. a result to be used further below. however. σ2 .12) does. in fact.224)] ∗ σ1 = σ1 .f. (11. Ψ (xµ ) as well as Ψ (xµ ). −1 given by (11. −1 a∗ (z) −1 (11.10.224. ∗ σ 3 = σ3 . −1 = ∗ −σ3 = −σ ∗ .11) 0 −1 1 0 = . hold −1 = 1 One notices then.11) one ﬁrst demonstrates that for and −1 as given in (11. 1. 355 (11.1: Lorentz Transformation of Dirac Bispinor ∗ ∗ ∗ where σ ∗ = (σ1 .4) and .10) From this one can derive b(z) = where = 0 1 −1 0 . one can express a∗ (z) −1 = exp 1 − z ∗· σ 2 = b(z) . therefore. ∗ σ 2 = − σ2 .12).1) can be written in the form ˜ S(z) = a(z) 0 0 a∗ (z) −1 (11. but leave rotation angles ϑ unaltered.

2. is an element of SU(2).3. In fact.224)] tr( σj ) = 0 . One can verify this by evaluating the determinant of a(z) det( a(z) ) = det e 2 z·σ 1 (11. Exercise 1.10.C) and SO(3. P −1 = P. det(M ) = 1 } .f. 2. 1990). the transformations a(z) and b(z) become interchanged. does not suﬃce for particles like the electron which obey inversion symmetry.e. de Kerf Lie a Algebras. the space spanned by only two of the components of Ψ is not invariant under spatial inversion and.21) together with matrix multiplication as the binary operation forms a group.A..A. The transformation S(z) in the transformed space is then ˜ P 1.. Amsterdam. Obviously. which describes pure rotations. 2 . B¨uerle and E. (11.1: Show that SL(2. based on the Jordan–Chevalley theorem.356 Spinor Formulation ˜ Let P denote the representation of spatial inversion in the space of the wave functions Ψ. C) = { M. non-singular matrix M holds2 det eM and from [c. for particles like the neutrinos which do not obey inversion symmetry two components of the wave function are suﬃcient. This implies ˜ Ψ1 ˜ Ψ2 P ˜ Ψ3 ˜ Ψ4 ˜ Ψ3 ˜ Ψ4 = Ψ1 ˜ ˜ Ψ2 . a(w + iϑ) for w = 0 is an element of SL(2.21) = etr( 2 z·σ) = 1 1 (11. M is a complex 2 × 2–matrix.24) = etr(M ) (11.e. Part (Elsevier.19) i. 2 = 1 i. ϑ ∈ R3 .1) We have pointed out that a(iϑ). However. However. For the general case the proof.2 Relationship Between the Lie Groups SL(2. j = 1. 3 .22) which follows from the fact that for any complex. the Lorentz invariant equation for neutrinos is only 2–dimensional. can be found in G.G. (11. (11. (5. ˜ ˜ P S(w + iϑ) P = S(−w + iϑ) = b(z) 0 0 a(z) .20) ˜ Obviously. hence. The proof of this important property is straightforward in case of hermitian M (see Chapter 5). 11.23) Exercise 11. C) deﬁned in (11.

25) σ0 σ1 σ2 σ3 The quantity σµ thus deﬁned does not transform like a covariant 4–vector. (11.27) −→ Straightforward transformation into another frame of reference would replace σµ by σµ .25) of the matrix M (Aµ ) is independent of the frame of reference.135)] this transformation behaviour.2: Lie Groups SL(2. (11.C) and SO(3. σ1 . In fact. M (Aµ ) according to (11. (11.2. in a transformed frame should hold Lρ ν σ µ Aµ σµ A µ . Noting that for covariant vectors according to (10. σ3 provide a linear–independent basis for the space of hermitian 2 × 2–matrices. σ2 . 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 .e.11.27) requires then σ µ Aµ Lν µ σµ = σν (11. σ3 are hermitian. one can provide a simple expression for the inverse of M (Aµ ) M = M (Aµ ) ↔ Aµ = 1 tr M σµ 2 . −→ µ Consistency of (11. i.31) holds.30) Since the components of Aµ are real.32) .28) and (11. 1 0 0 −1 . Demonstrate that (11. 0 1 1 0 . σ2 . σ1 . C) and the group L↑ of proper. Starting point is a bijective map between R4 and the set of two-dimensional hermitian matrices deﬁned through M (Aµ ) = σµ Aµ where (11.29) where we used (10.26) or from the fact that the matrices σ0 .31) Exercise 11. The following important property holds for M (Aµ ) det ( M (Aµ ) ) = Aµ Aµ which follows directly from (11. (11.76). The function M (Aµ ) is bijective. one wishes that the deﬁnition (11.1) Mapping Aµ onto matrices M (Aµ ) 357 We want to establish now the relationship between SL(2. or+ thochronous Lorentz transformations.30).28) σµ = 1 0 0 1 .26) (11. Argue why M (Aµ ) = σµ Aµ provides a bijective map.75) holds aν = Lν µ aµ one realizes that σµ transforms inversely to covariant 4–vectors. 0 −i i 0 . in fact. the matrix M (Aµ ) is hermitian as can be seen from inspection of (11. We will prove below [cf.25) can also be written M (Aµ ) = A0 + A3 A1 − iA2 A1 + iA2 A0 − A3 .2: Show that σ0 . (11.. (11.

+ .33) det(M ) = = det( a M a† ) = det(a) det(a† ) det(M ) [det(a)]2 det(M ) = det(M ) .37) which implies that (11. 11. Accordingly.38) .32.36) The suitable A µ can readily be constructed using (11. L(a)µ ν = 1 tr a σν a† σµ 2 .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† . (11. (11.35) holds for this transformation A µ Aµ = Aµ Aµ (11.31).25) Aµ = which allows us to express ﬁnally A µ = L(a)µ ν Aν . C) deﬁnes the transformation [c.34) where we used the properties M † = M and (a† )† = a.2. (11.f. Due to det(a) = 1 the transformation (11. The linear character of the transformation becomes apparent expressing A µ as given in (11. In fact.37) deﬁnes actually a Lorentz transformation.3: Show that L(a)µ ν deﬁned in (11.40) is an element of L↑ .33) This transformation conserves the hermitian property of M since (M )† = ( a M a† )† = (a† )† M † a† = a M a† = M (11. (11.39) Exercise 11.40) 1 tr a σν a† σµ 2 Aν (11. a ∈ SL(2. it holds for the matrix M deﬁned through (11. C) and for any Aµ there exists an A µ such that M (A µ ) = a M (Aµ ) a† .33) conserves the determinant of M . for any a ∈ SL(2.33) to M (Aµ ) describing the action of the transformation in terms of transformations of Aµ . (11. (11.37) using (11.35) We now apply the transformation (11.31)] 1 a Aµ −→ A µ = tr a M (Aµ ) a† σµ 2 Because of (11. C) . any a ∈ SL(2. In fact.

C) and SO(3.45) One can demonstrate through direct evaluation 2 α = 2 α = 2 α = = 2 α = 0 else β β γ γ = = = = γ = γ = 1.11.41) results in 1 ¯ Lµ ρ = 4 where Aαβγδ = ν Γ = a2 σρ a† 2 (11.δ (11.. i. (11. (11.1) For this purpose one writes using tr(AB) = tr(BA) L(a1 )µ ν L(a2 )µ ρ = ν a1 a2 product in SL(2.C) )µ ρ (11.1) L(a)µ ν provides a homomorphism 359 We want to demonstrate now that the map between SL(2. 1 ¯ and using the deﬁnition of Lµ ρ in (11. 2 (11.44) (σν )αβ (σν )δγ . ¯ Lµ ρ = L(a1 )µ ν L(a2 )µ ρ = L( product in SO(3.40)] respects the group property of SL(2.42) = ν Deﬁning Γ = a† σµ a1 . C) and of SO(3.C) and SO(3.1) deﬁned through L(a)µ ν [cf.α.43) (σν )αβ Γβα Γγδ (σν )δγ = ν.e. (11. β δ = δ = = δ = δ 1 2 = 2 = 1 Aαβγδ (11.1).47) This completes the proof of the homomorphic property of L(a)µ ν .β γ.δ 1 4 Aαβγδ Γβα Γγδ α. β 2.41) 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 γ. .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 .2: Lie Groups SL(2.

denoted below as κ1 . (11. J3 of the Lorentz transformations Lµ ν in the natural representations.. For the r. K2 . 1 (11. Aµ = (0. of (11.h. J Spinor Formulation The transformations a ∈ SL(2.55) (11. correspond to the generators given by (10. 0). 0. noting the linearity of M (Aµ ).53) results in the condition σ 0 A1 − σ 1 A0 = ( κ 1 σ µ + σ µ κ † ) Aµ . 1) − σ1 σ0 0 0 One can verify readily that 1 κ1 = − σ 1 2 (11. C) as complex 2 × 2–matrices are deﬁned through four complex or. K3 .48).50) (11.36) M (Lµ ν Aν ) = a M (Aµ ) a† (11.49) we obtain using (11. (11. 0). 0).360 Generators for SL(2. Aµ = (0. −A0 . One expects then that six generators Gj and six real coordinates fj can be deﬁned which allow one to represent a in the form 6 a = exp j=1 fj Gj .51) Insertion of (K1 )µ ν as given in (10. 10. 0.56) (11.h. eight real numbers. 0.25) in the last step. 1. 0. To this end we consider inﬁnitesimal transformations and keep only terms of zero order and ﬁrst order in the small variables. J1 .57) .52) where we employed (11.54) This reads for the four cases Aµ = (1. correspondingly.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. we write (11. i.e.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 ) .49). Aµ = (0. (11. 0.49) assuming (note that g µ ν is just the familiar Kronecker δµν ) Lµ ν a = = gµν + 1 + 1 (K1 )µ ν κ1 . 1 (11.58) .48) yields for the l.52) and (11. To obtain the generator of a(z) corresponding to the generator K1 .47.s. 0. 1. J2 . M (Aµ + (K1 )µ ν Aν ) = = M (Aµ ) + M (Aµ ) − M ( (−A1 . Because of the condition det(a) = 1 only six independent real numbers actually suﬃce for the deﬁnition of a.2) which correspond to the generators K1 . of (11. 0) ) σ0 A 1 − σ1 A 0 (11. 0. C) which correspond to K.53) Equating (11.s.

hence.60) We can. w describing boosts and ϑ describing rotations. the so-called contravariant spinors. 2 (11. one can show that the generators κ2 . J3 are given by 1 κ = − σ.2. ϑ) = ew·K + ϑ·J ∈ SO(3. − 1 .4: Show that the generators (11. 2 (11. J2 . 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 . (11. C) correspond to the generators K and J of Lµ ν . + 1 2 2 1 transforms under rotations (z = iϑ) like a spin– 2 state | 1 .64) .3: Spinors 361 obeys these conditions. 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. a1 1 a1 2 a2 1 a2 2 (11. a(w − iϑ) = e− 2 (w − iϑ)·σ ∈ SL(2. K3 and λ1 .1) acts on 4–vectors Aµ . state that the following two transformations are equivalent L(w. κ3 of a(z) corresponding to K2 . λ2 . Here a(z) = ( aα β ) = describes the matrix a(z).60) of a ∈ SL(2. def α = 1. λ3 corresponding to J1 . C) acts on spinors φα ∈ C2 (characterized below) 1 (11.61) This identiﬁes the transformations a(z = w − iϑ) as representations of Lorentz transformations. 2 λ = i σ. Exercise 11.63) where we extended the summation convention of 4-vectors to spinors.62) We denote the general (z = w + iϑ) transformation by φ α = aα β φβ = aα 1 φ1 + aα 2 φ2 .11. 2 2 φ1 φ2 ∈ C2 . 11.

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

77) is Lorentz invariant. g µν of the Minkowski space [cf.70) and (11.11.74)]. (11. For this scalar product holds φα χα = −χα φα .81) . −1 connecting contravariant and covariant spinors play the role of the metric tensors gµν . 11.72) in a notation analogous to that chosen for contravariant and covariant 4–vectors [cf.74) = = 0 1 −1 0 0 −1 1 0 (11.73.72)] φα φ α = = = = αβ αβ φβ φβ 11 21 11 21 12 22 12 22 (11.67) will be expressed as φα χα = φ1 χ1 + φ2 χ2 = φ1 χ2 − φ2 χ1 . 11.3. The transformation behaviour of φα according to (11.73) (11. (11.80) One can write in matrix notation aα β γ δκ αγ a δ αβ (11. Accordingly.72) 0 −1 1 0 Exercise 11. (11. 11. The inverse of (11.68).78) aβ γ γδ φδ (11. we will express (11.3: Spinors as can be veriﬁed from (11.69. 10.76) αβ αβ The scalar product (11. 11.75) (11.79) = a −1 T a κβ .1: Show that for any non-singular complex 2 × 2–matrix M holds M −1 = det(M ) M −1 T The matrices .77) (11.10. 11.63. (10. (10. In the transformed frame holds φ α χα = aα β αγ aγ δ δκ φβ χκ .75) is given by φα = as can be readily veriﬁed.69.71.71) (11. Proof that φα χα is Lorentz invariant We want to verify now that the scalar product (11.70) is φ1 φ2 −1 363 = = −1 φ1 φ2 .

e.85) φ1 φ2 ∗ ∗ .82) and with f (A)T = f (AT ) for polynomial f (A) a −1 T = e− 2 z·(σ 1 ∗ )T = e− 2 z·σ = a−1 1 (11. from (11. 2. ˙ We also deﬁne covariant versions of φα φα = (φα )∗ . Insertion of (11. a∗ (z) provides a representation of the Lorentz group which is distinct from that provided by a(z).87) which we will employ from now on. it holds φk = (φk )∗ .89) such that (11.86) A concise notation of the conjugate complex spinors is provided by ˙ (φα )∗ = φα = φ1 ˙ φ2 ˙ ˙ (11.88) which one veriﬁes taking the conjugate complex of (11.81) yields aα β and.80) φ α χα = φβ χβ .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. We denote ˙ (aα β )∗ = aα β ˙ (11. The transformation ˙ behaviour of φα is ˙ φ α = (aα β )∗ φβ ˙ (11. the ˙ conjugate complex spinors φα need to be considered separately from the spinors φα . Obviously.63). ˙ (11. σ † = (σ ∗ )T = σ.91) . (11.83) into (11. As discussed above.84) (11.364 Using (11. The complex conjugate spinors We consider now the conjugate complex spinors (φ ) = α ∗ γ δκ αγ a δ = a −1 T a κβ = a−1 a κβ = δκβ (11.83) Here we have employed the hermitian property of σ. i.88) reads ˙ ˙ φ α = aα β φβ ˙ ˙ (11.2) and (11. k = 1..90) extending the summation convention to ‘dotted’ indices. hence. Hence.

in matrix notation. + 1 2 2 2 1 transforms under rotations (z = iϑ) like a spin– 2 state | 1 .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.. A tensor tα1 α2 ···αk β1 β2 ···β (11.75).75. i.93) (11. The transformation behaviour of φα is ˙ φα = ˙ ˙ β γδ ˙˙ φδ ˙ ˙ αβ a γ ˙ ˙ .4: Spinor Tensors 365 The relationship between contravariant and covariant conjugate complex spinors can be expressed in analogy to (11.76) φα ˙ ˙ φα αβ ˙ ˙ αβ ˙ ˙ = = = = αβ ˙ ˙ αβ ˙ ˙ αβ αβ φβ φβ ˙ ˙ (11. Ψ4 . a∗ (z) −1 accounting for the ˜ transformation behaviour of the third and fourth spinor component of Ψ. is governed by the operator a∗ (z) −1 which arises in ˜ the Lorentz transformation (11.92) (11. (11. 11.96) is Lorentz invariant.99) Ψ3 (xµ ) χ ˙ (xµ ) ˜ χβ (xµ ) ˙ 1 ˜ χ2 (xµ ) Ψ4 (xµ ) ˙ 11.18) of the bispinor wave function Ψ. − 1 . a property which is rather evident.97) The transformation.94) (11. For the spinors φα and χα thus ˙ deﬁned holds that the scalar product ˙ φα χα = φ1 χ1 + φ2 χ2 ˙ ˙ ˙ ˙ ˙ (11.e. one can state ˜ and (11.4 Spinor Tensors ˙ ˙ ˙ We generalize now our deﬁnition of spinors φα to tensors. (11.11.18) ˜ 3 .73. A comparision of (11.95) ˙ where αβ and αβ are the real matrices deﬁned in (11.100) . 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 = (11.97) implies then that φα transforms like Ψ ˙ φ1 ˙ φ2 ˙ transforms under rotations (z = iϑ) like a spin– 1 state | 1 .

the transformation bevaviour of a tensor tαβ reads in spinor and matrix notation t αβ = aαγ aβδ tγδ .104) Indices on tensors can also be lowered employing the formula tα β = and generalizations thereof.25] M (Aµ ) = σµ Aµ .105) This tensor is actually invariant under Lorentz transforma= αβ .106) Exercise 11.102.m aj akm t m (11.m aj a∗ t km m .108) Obviously. this transformation behaviour is in harmony with the tensor notation adopted.107) which reads in spinor notation [cf. i.e. We had demonstrated that M (Aµ ) transforms according to M = M (Lµ ν Aν ) = a M (Aµ ) a† (11.103) A αβ = aα γ aβ δ Aγ δ .1: Prove equation (11. .. the elements of which are linear functions of Aµ . αβ = αβ (11. it holds αβ .102) = . with ˙ contravariant indices αβ. According to (11. ˙ ˙ ˙ ˙ (11.e.. tjk = a t a† jk ˙ ˙ ˙ ˙ (11. i. ˙ ˙ ˙ ˙ (11. (11.101) An example is the tensor tαβ which will play an important role in the spinor presentation of the Dirac equation.103) Similarly. An obvious candidate is [cf.25) the tensor is explicitly Aαβ = ˙ A1 1 A1 2 ˙ ˙ A2 1 A2 2 ˙ ˙ = A0 + A3 A1 − iA2 A1 + iA2 A0 − A3 . An example of a tensor is αβ and tions.366 is a quantity which under Lorentz transformations behaves as t ˙ ˙ ˙ α1 α2 ···αk β1 β2 ···β k Spinor Formulation = m=1 aαm γm n=1 aβn δn tγ1 ···γk δ1 ···δ . i. we want to express Aµ through a spinor tensor.4. using conventional matrix indices j.109) . tjk = a t aT jk = . This task implies that we seek a tensor.106). This tensor transforms according to t αβ = aα γ aβ δ tγ δ ˙ This reads in matrix notation. (11. (11.e.. m. The 4–vector Aµ in spinor form We want to provide now the spinor form of the 4-vector Aµ . k. (11. αβ ˙ αγ t ˙ γβ (11. 11.

109. Hence. 11.2: Prove that (11.4: Spinor Tensors One can express Aαβ also through Aµ Aα β = ˙ ˙ 367 A0 − A3 −A1 + iA2 −A1 − iA2 A0 + A3 . ˙ 2 (11. (11.110) one obtains Aα β = ˙ A0 − A3 −A1 − iA2 −A1 + iA2 A0 + A3 A0 + A3 A1 + iA2 A1 − iA2 A0 − A3 . employing (11.114) yields ∂αβ = ˙ ∂0 + ∂3 ∂1 + i∂2 ∂1 − i∂2 ∂0 − ∂3 .112) A22 −A21 ˙ ˙ −A12 A1 1 . (11. Aµ can also be associated with tensors Aα β = ˙ This tensor reads in matrix notation A11 A12 ˙ ˙ A21 A22 ˙ ˙ = = 0 1 −1 0 ˙ ˙˙ αγ β δ A ˙ γδ .113) Aα β = ˙ (11. Using (11. (11. Aµ and tensors tαβ can be applied to the partial diﬀerential operator ∂µ .4.110) one can state ∂ αβ = Similarly.111) A11 A12 ˙ ˙ A21 A22 ˙ ˙ ˙ 0 −1 1 0 (11. (11.115) is correct.11. (11. (11.110) The 4-vectors Aµ .117) ˙ ˙ ∂0 − ∂3 −∂1 + i∂2 −∂1 − i∂2 ∂0 + ∂3 .114) We ﬁnally note that the 4–vector scalar product Aµ Bµ reads in spinor notation Aµ B µ = 1 αβ ˙ A B αβ . ∂µ in spinor notation The relationship between 4–vectors Aµ .116) .115) Exercise 11.

(11.121). hence..h. −1 0 0 1 (11.123) (a∗ )−1 Lν µ a L ν µ = = σν (σ ) ν ∗ −1 (11.229) and (11.125) a (σ ) Equation (11.. Equation (10.243) holds independently of the representation chosen.368 σµ in Tensor Notation Spinor Formulation We want to develop now the tensor notation for σµ (11.26) and its contravariant analogue σ µ σµ = σµ = 1 0 0 1 1 0 0 1 . . one can conclude aσ µ ∗ −1 Lν µ σν 0 = .124) constitutes the essential transformation property of σ µ and will be considered further. (11. i. We will obtain the transformation behaviour of σ µ .118) yields ˜ γµ = ˜ Using (11. holds also in the chiral representation. .118) 0 −1 −1 0 0 i −i 0 For this purpose we consider ﬁrst the transformation behaviour of σ µ and σµ . σµ . (11.124).119) (11. To obtain the transformation properties of σ µ we employ then (10. Note that (10. (11.126) which is the complex conjugate of (11.124) . (11. This is ˜ possible since γ µ can be expressed through σ µ .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.120) σµ 0 . .125) is equivalent to (11. of this equation is 0 ∗ (σ µ )∗ a and. γµ = ˜ 0 µ )∗ (σ −1 0 σµ σµ 0 (σ µ )∗ −1 .243) in the chiral representation expressing S(Lη ξ ) by (11. 0 1 1 0 . 1 0 0 −1 . (10.e. 0 −i i 0 .15) one can write σµ = and. σµ building on the known transformation behaviour of γ µ . hence.243)].18) and γ µ by (11. Comparision of (10.s. (11.e.124).f. i.121) One expects then that the transformation properties of σ µ should follow from the transformation properties established already for γ µ [c.125) is equivalent to µ ∗ −1 −1 a∗ (σ µ )∗ −1 −1 a Lν µ = (σ ν )∗ (11.122) aσ µ −1 a−1 −1 (a∗ )−1 0 Lν µ (11. Hence.

(11. state that σ µ in a new reference frame is σ µ = a σ µ a† (11.11. 11. = .. 0 1 1 0 µ=1 .16. It holds according to (11. (11. 0 −1 −1 0 µ=1 . 11. This transformation behaviour.e.108) that the same transformation behaviour applies then for general Lorentz transformations.. 11. equation (11.2) −1 (a∗ )−1 = e2z 1 ∗ ·(σ ∗ )T = e2z 1 ∗ ·σ ∗ T = [a∗ ]T . i. .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 ˙ ˙ ˙ ˙ = (11.2. (σ ∗ )T = σ yields using (11. (11. transformations (11. 3) σj −→ a(iϑ) σj a(iϑ) † = a(iϑ) σj a∗ (iϑ) T . under rotations the Pauli matrices transform like (j = 1. 0 i −i 0 µ=2 .124) using (11.128) One can express.127) Exploiting the hermitian property of σ.4).133) Combining (11. −1 0 0 1 µ=3 .134) . In fact. therefore.131) where a is given by (11.g.2. according to (11. (11.2) −1 369 (a∗ )−1 = −1 − 1 z ∗ ·σ ∗ 2 e = e− 2 z 1 ∗ · −1 σ ∗ = e2z 1 ∗ ·σ .124) a σ µ [a∗ ]T Lν µ = σ ν . 0 −i i 0 µ=2 . 2.132) 1 0 0 1 µ=0 .103) ˙ identiﬁes σ µ as a tensor of type tαβ . 11.130) We argue in analogy to the logic applied in going from (11.131) one can express the transformation behaviour of σ µ in the succinct form Lν µ σ µ = σ ν .4: Spinor Tensors One can rewrite (11. (11. (11. 11.102. One can.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.129.107) to (11. hence. e.4) with w = 0. 1 0 0 −1 µ=3 .

We can summarize that σ µ and σµ transform like a 4–vector.138) the ‘inner’ covariant spinor indices. 1.s.138) Note that all elements of σαβ are real and that there are only four non-vanishing elements. α.224) ˙ Using (11. and the covariant indices are moved outside. Spinor Formulation (11. of (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 . however. the transformation is inverse to that of ordinary 4–vectors. σµ in which the contravariant indices are moved ‘inside’.114).e. = σαβ ˙ ˙ ˙ ˙ ˙ (σ21 )11 (σ21 )12 (σ22 )11 (σ22 )12 ˙ ˙ ˙ ˙ ˙ ˙ ˙ ˙ 21 (σ )22 21 (σ )22 (σ21 ) (σ22 ) ˙ ˙ ˙ ˙ 21 22 Equating this with the r. We can. the submatrices correspond to tαβ spinors. 3 as well as through two spinor indices αβ. ˙ ˙ 0 −1 1 0 ˙ (11.e. hence.137) σαβ ˙ = δαγ δβ δ .. state ˙ ˙ ˙ ˙ (σ11 )11 (σ11 )12 (σ12 )11 (σ12 )12 ˙ ˙ ˙ ˙ ˙ ˙ ˙ ˙ ˙ γδ (σ11 )21 (σ11 )22 (σ12 )21 (σ12 )22 ˙ ˙ ˙ ˙ .110) to express σ αβ in terms 0 −1 1 0 0 0 0 1 (11. i.. ˙˙ (11. µ = 0.141) σ αβ = σ0 − σ3 σ1 + iσ2 −σ1 − iσ2 σ0 + σ3 = 2 0 0 0 1 0 −1 0 0 .133) is characterized through a 4-vector ˙ index µ.. The ˙ desired change of representation(σµ )αβ −→ (σµ )αβ corresponds to a transformation of the basis of ˙ spin states f g f −→ = (11. Each of the 4 × 4 = 16 matrix elements in (11.h. 2. β.136) results in the succinct expression 1 2 ˙ γδ ˙ (11.e. matrices. γ. We want to express now σ µ and σµ also with respect to the 4–vector index µ in spinor form employing (11. i.139) g −f g and. whereas ˙ account for the elements of the individual Pauli the ‘outer’ contravariant spinor indices.136) σαβ = = 2 ˙ σ1 − iσ2 σ0 − σ3 0 0 0 0 1 0 0 1 where on the rhs.140) where we employed the expressions (11. We will now consider the representation of σ µ .12) for of σµ yields together with (5.132) and (11. account for the 4–vector index µ.29)]. ˙ ˙ In (11.135) This is the property surmised already above [cf. account for the 4-vector µ. (11. δ. This yields 1 0 0 1 0 0 0 0 σ0 + σ3 σ1 + iσ2 (11. i. in fact.370 Inverting contravariant and covariant indices one can also state Lν µ σµ = σν .

138) one can conclude that the following property holds 1 2 ˙ γδ σ αβ ˙ = 1 2 σ αβ ˙ ˙ γδ = δαγ δβ δ . which in ˙ (11. we write ˜ ˙ γµ = ˜ 0 (σ µ )αβ µ ) )∗ ((σ αβ 0 ˙ ˙ .145) Let Aµ be a covariant 4–vector.146) Exploiting the property (11. ˙˙ (11. (11.4: Spinor Tensors 371 and.121) of γ ˜ ˜ ˙ basis |αβ whereas the element of γ µ given by σ µ is in the basis |αβ .11.115) γ µ Aµ ˜ = 0 ((σ µ )αβ )∗ Aµ ˙ 0 ˙ ∗ 1 γδ ˙ ˙ 2 ((σ )αβ ) Aγ δ (σ µ )αβ Aµ 0 ˙ ˙ αβ γ δ 1 ˙ 2 (σγ δ ) A ˙ = 0 .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 ˙ ˙ .143) Combined with (11.147) .144) results in the simple relationship γ µ Aµ = ˜ 0 Aα β ˙ Aα β 0 ˙ . This expression implies that the element of γ µ given by σ µ −1 is in the expression (11. Accordingly. (11. One can write then the scalar product using (11. employing then transformation (11. ˙˙ (11.142) = This can be expressed 1 2 σ αβ ˙ ˙ γδ = δαγ δβ δ . For this purpose we start from the ˜ µ . ˙ ˙ ˙ ˙ (11.140) to transform each of the four submatrices.144) The Dirac Matrices γ µ in spinor notation We want to express now the Dirac matrices γ µ in spinor form. (11.

˙ (11.226) ˜ i γ µ ∂µ Ψ(xµ ) = 0 i ∂ αβ ˙ i ∂ αβ 0 ˙ ˜ ˜ Ψ(xµ ) = m Ψ(xµ ) . (11.149) (11.148) ˜ Employing Ψ(xµ ) in the form (11.147) allows us to rewrite the Dirac equation in the chiral representation (10.150) .5 Lorentz Invariant Field Equations in Spinor Form Dirac Equation (11.99) yields the Dirac equation in spinor form i ∂ αβ χβ ˙ i ∂ αβ φ ˙ The simplicity of this equation is striking.372 Spinor Formulation 11. α ˙ = = m φα m χβ .

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

12). l. For example.. Presence of additional symmetries may further increase the degeneracy of the system. Therefore.. . each energy level is (2l + 1)-fold degenerate.. where the symmetry generators. 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. we ﬁrst consider a particular example of the implications of symmetry. V (r).m (r) = vE.. · · · . . Sk . 2. (7. this implies a degeneracy in the spectrum. . The spherical symmetry implies the commutation of the Hamiltonian with angular momentum operators (7. 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. The energy levels are mk 2 En = − 2 2 .18).5) with m = −l.λn . (7. n − 1.6) From elementary quantum mechanics we know the spectrum of the hydrogen atom. ψE.. leaves the energy of the state invariant H exp(iαk Sk ) ψE. l and the corresponding energy levels are independent of m. (7. .λ1 . .. .4) The stationary Schr¨dinger equation..λ1 . As an example of this we will consider the Coulomb problem with the Hamiltonian H = p2 k − . ..7) 2 n where the orbital angular momentum. 2m r (12. (12. which yields a set of eigenstates of the form ψE. (12.39). as represented by an element of SO(3). φ). Jk . can be identiﬁed with the angular momentum operators.λn = E exp(iαk Sk ) ψE. The energy levels are totally independent of l. has been discussed.m (r) Y m (θ. 3 . as studied in chapter 7.3) If the newly obtained state is linearly independent of the original one. is allowed to take values in 0.. (12.374 Spinor Formulation The action of any symmetry generator.24). can then be reduced to a one-dimensional radial o equation. Sk . generates a state which has the same energy as the original one. 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 .. (12.. Lie groups play an essential role in the discussion of mass degeneracies in particle physics. m r (12.λ1 . on an energy eigenstate. namely motion in a spherically symmetric potential described by the group SO(3) (or its double covering SU (2) as discussed in section 5. k = 1. 3p and 3d all have the same energy. Classically..8) .λn .8) ˆ [H. In order to illustrate this. In chapter 7 the dynamics of a particle moving in three dimensions under the inﬂuence of a spherically symmetric potential. Jk ] = 0 . the states 3s.

In the following we consider the bound states. which have a negative energy E. Therefore.3.13) It can be veriﬁed explicitly that its components commute with the Hamiltonian. We introduce the following new operators 1 A = (J + K). 2 which can be shown to satisfy [Ai . We will also need the following identity [4] 2 = 2H J2 + m 2 + k2 . 2 1 B = (J − K). Aj ] = i ijk Ak .17) (12.21) . 2m r 1 x1 (J3 p2 − J2 p3 + i p1 ) − k .16) Through some algebra [4] the following commutation relations can be veriﬁed [Ki .11) (12.11. 2E (12. The vector in (12. In order to understand the aforementioned extra degeneracy.14) This follows from a · (a × b) = (a × b) · a = 0. (12.20) (12. 2m r 1 x2 (J1 p3 − J3 p1 + i p2 ) − k . ijk Jk . Kj ] = i ijk Kk . (12. we will compute the hydrogen spectrum using the additional symmetry. in the subspace of the Hilbert space corresponding to a certain energy we can replace H by E. (12. (12. Now we scale the eccentricity vector as follows K= − m . For this purpose we ﬁrst note that J · = 0. 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 . 2m r 1 x3 (J2 p1 − J1 p2 + i p3 ) − k .18) which complement the familiar angular momentum algebra of section 5. Jj ] = i [Ki .15) which can be proved after very considerable algebra. which is valid even when a and b do not commute.9) (12.10) (12.5: Lorentz Invariant Field Equations 375 The vector points along the symmetry axis of the elliptical orbit and its length equals the eccentricity of the orbit. 2m r (12.8) is not a hermitian operator.12) (12.19) (12.

l. H = 0. A.376 [Bi .21) and (12. H B.31) where we have used (12.26) (12. Bj ] = i [Ai .32) A comparison with (12. a + b = n − 1}. . 1. .28).7) tells us that (2a + 1) = n. we can read oﬀ the eigenvalues of A2 and B 2 from (12.15). In contrast to the discussion above about extra symmetries. This implies that a = b. namely that J J > < A − B =0 A + B =2 A (12.29) (12. .14) we note that A2 − B 2 = J · = 0. = 0. 2 (12. . .24) (12.25) So far we have shown that the symmetry generators form an algebra.23) (12. 2 1 . Chaos is described classically as exponential . 1. Noticing that A2 and B 2 have the same eigenvalues because of (12. . a lack of symmetry implies a lack of degeneracy in the energy levels of a quantum mechanical system. Using this equation the energy eigenvalues can be written in terms of the eigenvalues of A2 and B 2 operators. . a = 0.30) 2 2 2 1 . Spinor Formulation (12. Bj ] = 0. .22): A2 = a(a + 1) B 2 = b(b + 1) Following (12. Furthermore the bound on the orbital angular momentum. ijk Bk . . .27) (12. a = 0. .33) (12. the energy eigenvalues are found to be E=− mk 2 1 . (12. can be seen to follow from the triangle inequality as applied to J = A + B. . . This illustrates the eﬀect of additional symmetries to the degeneracy structure of a quantum mechanical system. 2 (2a + 1)2 2 2 (12. . In order to arrive at the spectrum a ﬁnal bit algebra is needed A2 + B 2 = J 2 + K 2 m 2 = J2 − 2E mk 2 1 = − − 4E 2 (12. The most extreme case of this is the quantum analogue of a classically chaotic system. b = 0. . By comparing to the rotation algebra introduced in chapter 5.28) . . 1. . . 1. which is identical to the the direct sum of the Lie algebra of of two SO(3) (or SU (2)) algebras.22) (12.34) It follows that l has to have values in {0 = |a − b|. .

another manifestation of chaos is the lack of independent operators commuting with the Hamiltonian.2 system p= 1 0 . which is a particle in box problem in two dimensions with a boundary which can be chosen arbitrarily. For a more detailed discussion of quantum chaos in billiard systems we refer the reader to [7] and the references therein. namely. Originally it was believed that isospin was an exact symmetry of strong interactions and that it was violated by electromagnetic and weak interactions. after the discovery of the neutron in 1932. This (approximate) degeneracy led into the idea of the existence of an (approximate) symmetry obeyed by the underlying nuclear interactions. If the chosen boundary is ‘irregular’ in a suitably deﬁned sense.5: Lorentz Invariant Field Equations 377 sensitivity to initial conditions. which was discovered by observing degeneracies in the particle spectrum. mp = 939. This enables us to utilize what we already know about the SU (2) symmetry group from the study of angular momentum. (12.35) this mass equivalence can be viewed as an energy degeneracy of the underlying interactions. In the next section we will proceed with the discussion of a symmetry. the classical trajectories will diverge from each other after successive bounces from the boundary.28M eV /c2 . 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 sense that nearby trajectories in the phase space diverge from each other over time. This is known as level repulsion. (12.36) into each other in abstract isospin space. Following the mass-energy equivalence of special relativity E = mc2 . who. analogous to the 1 spin-up and spin-down states of a spin. for the . 12.2 Isospin and the SU (2) ﬂavor symmetry The concept of isospin goes back to Heisenberg. For example. we will be able to use the familiar Clebsch-Gordan coeﬃcients to combine the isospin of two particles the same way we added spin in chapter 6. In the case of billiards and other examples of quantum chaos one common observation is the almost nonexistence of degeneracies and the fact that the energy levels are more evenly spaced. n= 0 1 .36) In analogy to the concept of spin regarding the rotations in 3-space as discussed in chapter 5.) If the proton and the neutron are to be viewed as two linearly independent states of the same particle. (Strong interactions bind the atomic nucleus together. It is important to place the isospin concept in its proper historical context. the isospin symmetry is also governed by an SU (2) group ‘rotating’ components in (12. A typical example of this so called quantum chaos is the quantum billiard problem. suggested that the proton and the neutron can be regarded as two states of a single particle. it is natural to represent them in terms of a two component vector.11. (Weak interactions are responsible. This was motivated by the observation that their masses are approximately equal: mp = 938.57M eV /c2 . However. for example.

I. the proton had to be heavier. including the spin and color quantum numbers of their constituent quarks.6M eV /c2 . (12. π − = |1. respectively. 1 we can re-write (12. the three pions in (12. The mass diﬀerence between the neutron and the proton could then be attributed to the charge content of the latter. Therefore it remains a useful concept. . 2 2 1 2 2 2 1 1 1 1 1 √ . I3 = − 2 2 . 1 . . d= 1 1 .− 2 2 . 2 2 1 2 2 2 (12. However. For example. 0 . Further than that. If the mass diﬀerence (or the energy diﬀerence) were to be to be purely electrostatic in nature.40) + 2 1 1 . They can be constructed with the same rules that have been used for angular momentum addition ¯ u in chapter 6. u¯ and d¯ states. we refer the reader to [1]. For a precise description of the two nucleons as composite states.43) Similarly. . . −1 = = = 1 1 1 1 . albeit it is a good approximate one.38) are ud.− . Denoting the total isospin. spelling disaster for the stability of matter. 2. If it were otherwise the proton would be unstable by decaying into the neutron. it can be seen as part of a larger (and more approximate) symmetry which is of great utility to classify observed particle families.− 2 2 1 2 2 2 1 1 1 1 . −1 . Shortly we 1 will see how to describe both the pion and nucleon states as composites of more fundamental I = 2 states. In the framework of the quark of model. as good quantum numbers. (12.36) as a multiplet with I = 2 1 1 p = I = .37) As an example of a multiplet with I = 1 we have the three pions or π-mesons π + = |1.378 Spinor Formulation beta decay). and its third component. are made up of these two quarks. I3 . 2 2 1 1 1 .42) (12. π 0 = |1.0M eV /c2 ).− . sec.39) All other isomultiplets. the proton and the neutron can be written as totally symmetric uud and udd states. (12.38) which have all nearly identical masses. n = I = . I3 = 2 2 1 1 . as we shall see below. They u form an isotriplet: π + = |1.− 2 2 . mπ0 = 135. including the proton and the neutron. the fundamental representation of the isospin symmetry corresponds to the doublet that contains the so-called up and down quarks u= 1 1 . Isospin symmetry is not an exact symmetry of strong interactions.11. 2 2 . We can describe isospin multiplets the same way we have described the angular momentum and spin multiplets. 2 (12. the proton is the lighter of the two. 1 π 0 = |1. (mπ± = 139.41) (12. 0 π − = |1.

0 = √ (|p > |n > −|n > |p >) 2 and that of an isotriplet |1. originally refer to the relative weight of particles. We will now try to describe its wave function in terms of its constituent nucleons.11.37) and in analogy to (12. to be −1 for the strange quark. 1 = √ (|p > |n > +|n > |p >). but such states do not exist. The up and down quarks have strangeness zero. All other composite states have their strangeness given by the sum of the strangeness content of their constituents. mesons have intermediate mass ranges. The reader may know that the deuteron. a hydrogen isotope.47) (12. If taken literally. 2 = |n > |n > . 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).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. mesons have baryon number 0. we shall setup some terminology: baryons are qqq states. but at the time the empirical concepts like isospin and strangeness quantum numbers were in use. . this will be mathematically identical to adding two spins. baryons generally are heavy.48) (12. such as proton and the neutron. Therefore the deuteron has to be an isosinglet state (|0. Before proceeding further.e. Therefore it has to have isospin. The value of the strangeness quantum number is taken.or ﬂavors as they are commonly called. Now let us consider another example of combining the isospins of two particles. this remains only an inaccurate naming convention today. All antiparticles have their 3 quantum numbers reversed.43). where leptons (electron. muon.46) (12. The names. 0 ) we should have seen nn and pp bound states of comparable energy in nature (because of isospin symmetry). This is why isospin is such a good symmetry and why isomultiplets have nearly identical masses.44) 2 where B is the baryon number and S is the strangeness. the neutrinos etc. Following (12. 0 |1. 1 |1. and quarks have baryon number 1 . The possibilities are that of an isosinglet 1 |0. Naturally. the up and down quarks are not the only quark ‘species’ .) are light. I3 = 0. as some mesons discovered later are heavier than some baryons and so on. (12. As it later turned out. 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. baryon and meson.45) Is the deuteron an isosinglet state or an isotriplet? If it were an isotriplet (|1. a number strange particles have been found which presumably contained a third quark species: the strange quark. consists of a proton and a neutron. 0 ). (12. In the late 1940’s and early 1950’s. By convention ¯ baryons have baryon number 1. It shall be noted here that the quark model was not invented until 1960’s. −1 = |p > |p >. by accidental convention. whereas mesons are q q states. the pions being examples thereof. strongly interacting particles) which is about a GeV /c2 .

259) discussed in detail in sections 6. 1 4 in an isotriplet state and − 3 in an isosinglet 4 Using (12. We will now employ the (isospin analogue) of the Wigner-Eckart theorem (6.53) 0 √ 1 2I (i) +1 I (f ) . For completeness let us start by restating the Wigner-Eckart theorem (6.51) (12. γ (i) (i) (12. I3 = 1.7 to the case of isospin. (12. γ (f ) |T00 |I (i) I3 . such as the spatial dependence of the wave function and spin.49) The only assumption that we put in will be that the interaction is of the form V = α I(1) · I(2) . σ. γ (i) = (f ) (i) (I (f ) I3 |00I (i) I3 ) (f ) (i) (−1)I −I (f ) (i) (12.45) and (12. as discussed in the exercise above. 0 . Exercise. 1 and |1.7.7 and 6. γ (i) .54) Here T00 ≡ V = α I(1) · I(2) . 1 and (1/ 2)(|1. 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) is a reduced matrix element deﬁned in the same sense as in section 6.48) the initial states in (I) and (II) have isospin values |1. The initial and ﬁnal states can be denoted in more detail as | initial | ﬁnal = = I (i) . 0 ). γ (f ) ||T ||I (i) .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. Exercise. respectively. which is an isoscalar. consider (I) p + p → d + π + (II) p + n → d + π 0 (12.52) where γ (i) and γ (f ) denote degrees of freedom other than isospin.8. 0 + |0. γ (f ) (f ) (i) (12. is proportional to |M|2 . γ (i) I (f ) . We will eventually be able to compute ratios of scattering cross-sections between diﬀerent processes. Consider the generalization of tensor operators discussed in section 6. Show that I(1) ·I(2) is an ‘isotensor’ of rank zero.55) . √ According to (12. (f ) (i) (I (f ) I3 |00I (i) I3 ) is a Clebsch-Gordon coeﬃcient and I (f ) . γ (f ) T00 I (i) = 1. The dot product here refers to the abstract isospin space. The cross-section.380 Spinor Formulation As an exercise on the implications of isospin symmetry we will consider nucleon-nucleon scattering. For example. I3 . I3 . (12. Show that the expectation of I(1) · I(2) is state. γ (f ) ||T0 ||I (i) . where M is the scattering amplitude given by M = ﬁnal | α I(1) · I(2) | initial .8 to compute the ratio of the scattering amplitudes MI and MII .50) .259) in the present context: I (f ) I3 .5 for the spin-analogue of the same problem. I3 (f ) = 1.

It follows σI = 2. (b) π − + p → π − + p . (c) π − + p → π 0 + n .61) MII Note that the second term in MII vanishes due to the isospin conservation.62) We can now write the ratio of the scattering amplitudes: MI 1 √ . particles with strangeness. = 381 (12. γ (i) = √ 2 1 (f ) (i) +√ I (f ) = 1. which are obtained by a standard Clebsch-Gordan expansion π + + p : |1. γ (f ) T00 I (i) = 0.60) (12. We want to compute the ratio of total cross-sections assuming a similar interaction as in the previous example σ( π + + p → anything ) . 2 = 2. I3 = 0. we will consider pion-nucleon scattering.63) where common dynamical factors (which would not be as easy to compute) have dropped out thanks to the Wigner-Eckart theorem. −1 π0 + n : |1. −2 . −1 2 3 2 − + . (12. . I3 = 0.66) = 2 3 1 3 2. (12.57) (12.59) (12. 2 = 3 2. −2 2 1 1 3 2.64) There are more exotic possibilities.56) (12. 0 1 1 3 3 2. 1 1 1 1 3 √ 2. −2 1 1 2. γ (f ) T00 I (i) = 1. 1 π − + p : |1. γ (f ) ||T00 || I (i) = 1. Once again we need the isospins for the initial and ﬁnal states.11. γ (f ) ||T00 || I (i) = 1.5: Lorentz Invariant Field Equations 1 (11|0011) √ I (f ) = 1. γ (i) 2 1 = (10|0010) √ I (f ) = 1. involving. −2 1 1 √ . σII which is in approximate agreement with the observed ratio. [2] As a further example. I3 = 0. I3 = 0. γ (i) 3 +0. σ( π − + p → anything ) The possibilities are (a) π + + p → π + + p . but these are not dominant at relatively low energies. The relevant Clebsch-Gordon coeﬃcients are easily evaluated: (11|0011) = (10|0010) = 1. 2 . γ (i) 3 1 (f ) (i) I (f ) = 1. for example.58) (12. which is also manifested by a vanishing Clebsch-Gordon prefactor.65) (12. = MII (1/ 2) (12. (12.

For instance. m 1 2. It was assumed that strong interactions conserved S (at least approximately).68) − 2 3 M1 2 Guided by empirical data we will further assume that M 3 >> M 1 .69) 12. was motivated by the existence of particles that are produced strongly but decay only weakly. Strange partners of the familiar nucleons. Hence the strangeness changing strong decays of K + (or Σ− ) were forbidden. which can be produced by π − + p → K + + Σ− .d = 1 . 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. has a lifetime which is comparable to that of π + albeit being more than three times heavier. Hence Gell-Mann and independently Nishijima postulated the existence of a separate quantum number. giving it a higher than usual lifetime. which leads to the following 2 2 ratios for the cross-sections σa : σb : σc = 9 : 1 : 2.3 The Eightfold Way and the ﬂavor SU (3) symmetry The discovery of the concept of strangeness.71) u = 0 . making an identiﬁcation of the underlying symmetry and the multiplet structure less straightforward. m 1 2. m V V 3 2. The classiﬁcation of the newly found particles as members of some higher multiplet structure was less obvious then the case of isospin. K + .67) which are independent of m. (12. A computation similar to the previous example of nucleon-nucleon scattering yields (apart from common prefactors) the following amplitudes for the reactions in (12. are up to 40% heavier.70) (12. [2] (12. etc.382 Spinor Formulation As in the example of nucleon-meson scattering we deﬁne the relevant matrix elements M3 = 2 M1 = 2 3 2. for example. mentioned in the previous section. such that S(K + ) = 1. however. 0 0 1 . while weak interactions did not. m . S.65) Ma = M 3 2 Mb = 1 M 3 + 2 M 1 3 3 √ 2 Mc = 2 3 M3 2 √ 2 (12. S(Σ− ) = −1 and S(π) = S(N ) = 0. it appears an obvious generalization to add another component for an extra quark to the isospin vector space 1 0 0 (12. . In the light of the quark model. called strangeness.s = 0 .

in reference to Lie groups. (12. giving rise to the quark model. The reason is that the other known quarks. it is the group . called color. but there are a number of subtle diﬀerences. the same way nucleons and pions were identiﬁed as representations of the isospin SU (2) symmetry.72) a∗ ak . However SU (3) symmetry appears in another and much more fundamental context in strong interaction physics. for a period they were considered as useful bookkeeping devices without physical content. where Nf is the number quark ﬂavors. bottom (beauty) and top (truth) are signiﬁcantly heavier than the hadronic energy scale. history followed the reverse of this path. Then came the question of what the fundamental representation.73) To verify that all such matrices form a group. It is seen that such a matrix U has to satisfy the following k U † = U −1 . Lie algebras and related concepts. ak . Because of this additional constraint SU (3) has 8 dimensions.11.5: Lorentz Invariant Field Equations 383 In this case the transformations that ‘rotate’ the components of (12. which again form representations of an SU (3) group. The unitarity relation imposes 9 constraints on the total of 18 real degrees of freedom of a 3 × 3 complex matrix. The quarks posses another quantum number. but we will stick to N = 3 in the following. Since arbitrary phase factors are of no physical interest. However. As quarks were never directly observed. SU (N ). in fact modern theory of strong interactions is a ‘gauge theory’ of this color group. while preserving the norm. The reader is also invited to revisit section 5. In many respects it will resemble the more familiar group SU (2) discussed in some detail in chapter 5. should correspond to. The bottom and top are even heavier. eiφ . The reader shall note that most of what is being said trivially generalizes to other unitary groups. we want to ﬁnd those transformations ak → Ukl al that preserve the norm. (The ‘bare’ mass of the charm quark is already heavier than the two nucleons. This group of 3 × 3 unitary matrices is denoted by U (3). which set the hadronic energy scale. called quantum chromodynamics.1 whenever necessary.71). Given a complex vector. This is believed to be an exact symmetry of strong interactions. Hence the group U (3) has 9 dimensions. (12. Therefore U (3) can be decomposed into a direct product U (1) × SU (3) where SU (3) consists of 3 × 3 unitary matrices of unit determinant. of three dimensions. namely charm. In this perspective the ﬂavor SU (3) symmetry may appear to be mainly of historical interest. First particle multiplets were identiﬁed as representations of the SU (3) group. still leaves the norm invariant. have to be elements of the group SU (3) (which we will investigate closely very soon). as in (12. becomes increasingly inaccurate for Nf > 3. unitary relation k (12. [2] ) Before discussing the signiﬁcance and the physical implications of the quark model. (The reader is referred to section 8. Multiplying U by a phase. we observe that (U V )† = V † U † = V −1 U −1 = (U V )−1 . Flavor SU (Nf ) symmetry on the other hand.74) for any two unitary matrices U and V .3 for a brief discussion of gauge transformations).71) into each other. of a. we will establish some mathematical preliminaries about the group SU (3).

7.1.79) (12. .79). any unitary matrix. λk ] = 2ifjkl λl . λ2 = . (We shall at times refer to the Lie algebra with the name of the group. λk ]+ = δjk + 2δjkl λl . As in the case of SU (2) higher dimensional representations obeying the same structure can be found. (12.384 Spinor Formulation SU (3) and not U (3) that is of main interest. via the anticommutator relation 4 [λj . which can be veriﬁed explicitly as matrix identities. (12. −2 λ4 λ6 λ8 The generators. U . λ5 = . λk . The reader is invited to compare the structure of SU (3) to that of SU (2) discussed in section 5. regardless of the dimension of the representation. 0 0 0 . As discussed in section 5. λ3 = 0 −1 0 . obey the following relation tr(λj λk ) = 2δjk . the meaning being apparent from the context. Therefore we will express elements of SU (3) as U = ei k (12.75) αk λk (12. can be written in the form U = eiH where H is a hermitian matrix.78) where fjkl are the antisymmetric structure constants similar to the jkl of SU (2) given in (5.80) This is the fundamental or deﬁning representation of SU (3). The fundamental relation to be preserved is (12.) The unit determinant condition requires that all λk are traceless. We can also introduce the constants. 3 (12.32). δjkl . 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 .77) λ1 . since det(eA ) = etrA . The Lie algebra structure is given by the commutators of λk [λj .76) where λk are 8 linearly independent matrices forming the basis of the Lie algebra of SU (3). . λ7 = 0 0 .

J± ] = ± J± .81) In chapter 5.92) (12.82) With another change of basis we arrive at the ‘standard’ form of the generators of the Lie algebra of SU (3) T± = F1 ± iF2 . (12.88) which deﬁnes (not uniquely) a maximal set of mutually commuting operators {Y. these relations have been used to construct the angular momentum spectrum as well as the function space representation of the rotation algebra.44). T 0 = F3 . V± ] = ± V± . Y is called the hypercharge. U± = F6 ± iF7 . T± ] = 0. we have [Y. The generators of SU (3) can be arranged into a very similar form to that of SU (2). In the case of SU (2) the identiﬁcation of ‘ladder’ operators {J+ . V± ] = ±V± . T0 } and [Y. We ﬁrst introduce the F-spin operators 1 Fi = λ i . In the basis (12.87) Exercise. (12. 1 [T0 .77) is not the most useful basis in the study of SU (3).94) . which satisﬁed an ‘eigenvalue equation’ [J0 .93) (12.86) (12.89) (12. namely spherical harmonics.91) [T0 . 2 (12. Derive the Gell-Mann–Nishijima relation (12. First. (Recall that the strange quark has S = −1 by convention). J− } proved useful. I3 . [Y.83) (12. Exercise. [Y.11.85) (12. U± ] = U± .87) the commutation relations between the generators can be expressed in a succinct manner.90) (12. Using the convention in (12. U± ] = ±U± . T± ] = ±T± . starting with the observation that Y = B + S.84) (12.5: Lorentz Invariant Field Equations 385 As it turns out the set of generators in (12. 3 (12. V± = F4 ± iF5 . 2 Y = √ F8 . T0 ] = 0.71) show that T0 is the isospin operator. 2 (12. 2 1 [T0 . (12.

the representations will now lie in a T0 − Y -plane. U+ ] = V+ . [T+ . Finally. jkl (12. V+ ] = 0. As in the case of J 2 and SU (2). 3 [U+ . [T+ .95) (12.100) (12. [U+ .101) (12.99) (12. U− ] = 0.103) Any remaining commutators follow from hermiticity. This form is essential for easy labeling of the representations of the group. J 2 ) which commutes with all of the generators. similar to the way it was done in section 5. When the basis of a Lie algebra is expressed in such a way to satisfy the form of the eigenvalue relations as given above. Casimir operators can be used to label irreducible representations of the Lie algebra.102) (12. (12. T0 }. The interested reader is instead referred to a very readable account given in chapter 12 of [4]. V+ ] = 0. T− ] = 2T0 . . 2 3 Y + T0 = 2V0 . C1 = k λ2 = − k 2i 3 fjkl λj λk λl . The same way the angular momentum ladder operators have been used to construct the representations of SU (2). A formal deﬁnition and a detailed discussion of the Cartan-Weyl form is beyond the scope of this chapter. U± } to this maximal set by ‘eigenvalue equations’ and [T+ .88). we have [T+ . as the relation between the states in a given representation can be conveniently expressed in terms of ladder operators. while SU (3) has rank 2. [V+ . In the case of SU (2) the representations lay on a line on the J0 axis.96) (12. However. U0 and V0 are linear combinations of T0 and Y . The maximum number of mutually commuting generators of a Lie algebra is called its rank. V− ] = 2 (12. [U+ . since there are two mutually commuting generators in SU (3) as given in (12. V± . V− ] = −U− . We can construct two such independent Casimir operators for the group SU (3). An operator which commutes with all generators of a Lie group is called a Casimir operator.105) C2 = jkl djkl λj λk λl In general the number of independent Casimir operators of a Lie group is equal to its rank. V− ] = T− . U− ] = Y − T0 = 2U0 . it is said to be in Cartan-Weyl form.104) (12.386 Spinor Formulation which relate the remaining generators {T± . we will use these commutation relations to construct representations of SU (3). Another important property of SU (2) is the existence of an operator (namely the total angular momentum.5. Note that.98) (12. Thus.97) which relate commutators of generators with opposite eigenvalues to the maximal set {Y. [T+ . SU (2) has rank 1.

y 3 = ( y− 4 3 = ( y+ 4 1 t3 ) |t3 . All other states of the representation are then constructed by successive application of ladder operators T± . (See Fig. The representations for hexagons with sides of length p and q in the T0 − Y -plane.y ± 1 . For example. 1) representation.88) we can label states by the eigenvalues of T0 and Y operators. . we have U0 |t3 . y .112) The eﬀect of these operators to the states in the y − t3 plane have been outlined in Fig. y . For example the pion octet (or any other meson octet) is therefore realized as states of a quark . For example. 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. the pion family forms part of an octet corresponding to the D(1. 1) correspond to the triplets of quarks and antiquarks.110) (12. y V0 |t3 . The representations of SU (3) are constructed analogous to those of SU (2) by identifying the ‘boundary’ states annihilated by raising (or lowering) operators. y 1 = α t3 ± . For example.11. y ± 1 .109) The same way that J± |m is proportional to |m ± 1 in the case of the angular momentum algebra. y Y |t3 . The interested reader is referred to [4]. y .108) (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 . As another example for the representations of SU (3). especially chapters 7 and 8. 1) as an example. 2 (12.1). y U± |t3 . 2 1 = β t3 ± . Figure (12. Now we will construct explicit representations of SU (3). 2 (12. 2 1 = γ t3 . y .2) shows the representation D(2.antiquark pair.106) (12. (12. V± . y : T0 |t3 .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. 0) and D(0. (12. U± . respectively.3). 2 1 t3 ) |t3 . = y |t3 . we have T± |t3 . A notation suggestive of the dimensionality of the representation can be used to identify representations. The details of this procedure is beyond the scope of this chapter. This property can be used to label representations in terms of the values of the Casimir operators. (12. y V± |t3 . |t3 . y . The representations D(1. Such a representation is labeled as 1 D(p. y = t3 |t3 . Because of (12.111) (12.) All other representations can be constructed by combining these two conjugate representations.5 how to label the irreducible representations of the angular momentum algebra SU (2) in terms of the value of the total angular momentum. it was shown in section 5. y . q) and it has a dimensionality of 2 (p + 1)(q + 1)(p + q + 2).

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

3 (12. 0) and D(0.11. where we would write [3] ⊗ [3] = [1] ⊕ [3] ⊕ [5]. The octet D(1.115) (12. This expansion can be compared. − .117) The Gell-Mann–Nishijima relation can then be succinctly expressed as 1 Q = y + t3 . 2 3 1 1 − . 1) are denoted by [3] and [¯ respectively.113) The additional singlet state corresponds to the η meson. . in the case of SU (2). This way the quark-antiquark states can be represented as follows [3] ⊗ [¯ = [8] ⊕ [1] 3] (12. £ (12. 2 3 2 0.114) Figure 12. for example. The three quarks in the fundamental representation can now be written as u = d = s = 1 1 . to the case of adding two spin triplet states.3: D(1. 0) or the fundamental representation of ﬂavor SU (3) symmetry. 2 (12.5: Lorentz Invariant Field Equations 389 D(1. . 1) is written as [8] etc.118) ¡ ¢ ¨ ©§ ¨¥¦¤ " ! .116) (12. 3].

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

- Annihilation and Creation Operators.Chapter 8
- Relativistic Quantum Mechanics
- 51891292 Quantum Mechanics I
- Relativistic Quantum Mechanics
- Quantum Mechanics Lecture
- Advanced Quantum Mechanics
- Cambridge.quantum.mechanics.apr.2009.eBook ELOHiM
- Sakurai - Modern Quantum Mechanics Rev Ed- Solutions Manual
- Quanten_physics
- Quantum mechanics
- Commutators, eigenvalues and quantum mechanics on surfaces...
- Advanced Quantum Mechanics
- Quantum Mechanics
- 611 Electromagnetic Theory II
- Physics - Quantum Field Theory
- Classical Electrodynamics 3rd Ed J.D. Jackson - Solutions - 214 Pg
- Quantum Mechanics for Engineers
- Super Symmetry and Quantum Mechanics
- Sakurai quantum mechanics solutions 4
- 1222405266 Quantum Mechanics
- Merzbacher Quantum Mechanics
- Quantum Mechanics Review
- Kittel Charles - Introduction to Solid State Physics 8Th Edition - Solution Manual
- Book
- Lectures on Quantum Mechanics - Gordon Baym
- Relativistic Quantum Mechanics-Bjorken j.d., Drell s.d.
- Applied Quantum Mechanics
- Peskin Chapter 4
- Mathematical Methods Quantum Mechanics
- Solution Manual String Theory

Are you sure?

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

We've moved you to where you read on your other device.

Get the full title to continue

Get the full title to continue reading from where you left off, or restart the preview.

scribd