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

In case of N classical particles holds M = 3N . namely. i. · ] exists with the properties (i) S[q(t) + δq(t)] = S[q(t)] + δS[q(t). if for any q(t) ∈ F and δq(t) ∈ F where F = {δq(t). there are 3N conﬁgurational coordinates.e. An important concept is that the equations of motion of Classical Mechanics can be based on a variational principle. namely. . δq(t)] + O( 2 ) (1. For this purpose we will review the relevant concepts of Classical Mechanics. δq(t)] is linear in δq(t). q2 (t). 1. q(t) is called the trajectory of a system of M degrees of freedom described by the conﬁgurational coordinates q(t) = (q1 (t). Deﬁnition: A functional S[ ] is diﬀerentiable. often in the Cartesian coordinate system. t ∈ [t0 . |δq(t)| < . ∀t. α1 δq1 (t) + α2 δq2 (t)] = α1 δS[q(t). δq2 (t)] .4) d δq(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).1) from a space F of vector-valued functions q(t) onto the real numbers. · ] is called the diﬀerential of S[ ]. . q : [t0 .Chapter 1 Lagrangian Mechanics Our introduction to Quantum Mechanics will be based on its correspondence to Classical Mechanics. . 1 (1. that along a path describing classical motion the action integral assumes a minimal value (Hamiltonian Principle of Least Action). The linearity property above implies δS[q(t). | a functional δS[ · . δq(t) ∈ F. δq1 (t)] + α2 δS[q(t).. q(t) diﬀerentiable} (1. t1 ] ⊂ R → RM .2) . qM (t)). the position coordinates of the particles in any kind of coordianate system. Deﬁnition: A functional S[ ] is a map S[ ] : F → R .1 Basics of Variational Calculus The derivation of the Principle of Least Action requires the tools of the calculus of variation which we will provide now. F = {q(t). δS[ · . t1 ] ⊂ R} dt (1.3) (ii) δS[q(t). The latter is the velocity vector of the system.

t) of the conﬁguration d vector q(t). (1. · . We attempt to evaluate t1 S[q(t) + δq(t)] = t0 ˙ ˙ dt L(q(t) + δq(t). · ] in conventional diﬀerential calculus does not correspond to a diﬀerentiated function.. e.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 ) (1.6) For a proof we can use conventional diﬀerential calculus since the functional (1. · ) is a function diﬀerentiable in its three arguments. dt q(t). q(t) + δ q(t). ˙ Theorem: Let S[q(t)] = t0 dx dt t1 ˙ dt L(q(t). t0 (1. at the ends of the time interval of the trajectories the variations vanish. q(t) + δq(t) is a ‘slightly’ diﬀerent trajectory than q(t). q(t). t) + j=1 ∂L ∂L δqj + δ qj ˙ ∂qj ∂ qj ˙ + O( 2 ) (1.. We focus on such functionals because they play a central role in the so-called action integrals of Classical Mechanics. i. equating these terms with δS[q(t). the velocity vector dt q(t) and time t.7) through Taylor expansion and identiﬁcation of terms linear in δqj (t). ∂f in case of a function of M variables.e. .6) is expressed in terms of ‘normal’ functions. j=1 We will now consider a particular class of functionals S[ ] which are expressed through an integral d over the the interval [t0 . but rather to a diﬀerential of the function which is simply the df diﬀerentiated function multiplied by the diﬀerential increment of the variable.10) From this follows (1. 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. t) = L(q(t). δq(t)]. t) (1. q(t). df = M ∂xj dxj . It holds M M ∂L t1 d ∂L ∂L δS[q(t).e. ˙ In the following we will often use the notation for velocities and other time derivatives d q(t) = q(t) and dtj = xj . It is also important to appreciate that δS[ · . t) (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 .5) where L( · . t1 ] where the integrand is a function L(q(t). i.g. δq(t)] = dt − δqj (t) + δqj (t) ∂qj dt ∂ qj ˙ ∂ qj ˙ t0 j=1 j=1 t1 .2 Lagrangian Mechanics Note: δq(t) describes small variations around the trajectory q(t). q(t) + δ q(t).6) immediately. df = dx dx or. For this purpose we consider M ˙ ˙ ˙ L(q(t) + δq(t).

. if and only if it satisﬁes the conditions (j = 1. .14) for any continuous function h(t) ∈ F with h(t0 ) = h(t1 ) = 0. M ) d dt ∂L ∂ qj ˙ − ∂L = 0 ∂qj (1. y(x)) to (x + dx. . δq(t)] = dt − δqj (t) . t1 ] ⊂ R → R holds t1 (1.11) The extremals qe (t) can be identiﬁed through a condition which provides a suitable diﬀerential equation for this purpose.16) the property δS[qe (t). . The length of such path can be determined starting from the fact that the incremental length ds in going from point (x. . 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 .17) .1. . The proof of the above theorem starts from (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).12) The proof of this theorem is based on the property Lemma: If for a continuous function f(t) f : [t0 . then f (t) ≡ 0 on [t0 . y2 ).12) for j = 1. t1 ]. . ∂qj dt ∂ qj ˙ t0 j=1 (1.13) dt f (t)h(t) = 0 t0 (1. the above theorem. (1. 2. y1 ) and (x2 . Theorem: Euler–Lagrange Condition For the functional deﬁned through (1. y(x + dx)) is ds = (dx)2 + ( dy dx)2 = dx dx 1+( dy 2 ) .6) which reads in the present case M ∂L t1 d ∂L δS[q(t).15) We will not provide a proof for this Lemma. . M .1: Variational Calculus 3 We now consider the question for which functions the functionals of the type (1. 2. According to the Lemma above follows then (1. (1.5).5) assume extreme values. dx (1. y(x2 ) = y2 . M and δqj (t0 ) = δqj (t1 ) = 0 follows according to (1. . On the other side. δq(t)] = 0 for all δq(t) ∈ F . hence. . Let us assume that the two points are connected by the path y(x).12) for j = 1. y(x1 ) = y1 . · ] ≡ 0 and. it holds in case δq(t0 ) = δq(t1 ) = 0 that qe (t) is an extremal. This condition is stated in the following theorem. from (1.16) This property holds for any δqj with δq(t) ∈ F . 2.

20) x1 − x2 Exercise 1. q2 . the so-called action integral. t1 S[q(t)] = t0 ˙ dt L(q(t). q(t). Velocity–dependent potentials which describe particles moving in electromagnetic ﬁelds will be considered below.21) are extremals of the functional.23) Presently we consider only velocity–independent potentials. i. .22) ˙ where L(q(t). y = const. according to the theorems above.e.19) From this follows y / 1 + (y )2 = const and.1: Show that the shortest path between two points on a sphere are great circles. One obtains y 1 − y2 y(x) = (x − x2 ) + y2 .. . obey the Euler–Lagrange dy condition.4 The total path length is then given by the integral x1 Lagrangian Mechanics s = x0 dx 1+( dy 2 ) . (1. The constants a and b are readily identiﬁed through the conditons y(x1 ) = y1 and y(x2 ) = y2 . hence. The shortest path is an extremal of s[y(x)] which must. t) = j=1 1 mj qj − U (q1 . q(t). q(t). (1.1. 2. . dx (1. . t) (1. . . 1. M (1. j = 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 . circles whose centers lie at the center of the sphere. Using y = dx the condition reads d dx ∂L ∂y = d dx y 1 + (y )2 = 0. dx ) = 1 + (dy/dx)2 . . ˙2 2 (1.5) with L(y(x).18) dy s is a functional of y(x) of the type (1. qM ) . t) is the so-called Lagrangian M ˙ L(q(t).2 Lagrangian Mechanics The results of variational calculus derived above allow us now to formulate the Hamiltonian Principle of Least Action of Classical Mechanics and study its equivalence to the Newtonian equations of motion. . This in turn yields y(x) = ax + b.

t) describes the charge density present in the ﬁeld and J(r. t) = V (r. 1.26) (1. t) obey the Maxwell equations ×E + ·B ×B − ·E 1 ∂ c ∂t E 1 ∂ c ∂t B (1. t) = q E(r. respectively. t) and a vector potential A(r. These conditions read in the present case ∂L d ∂L ∂U d − = 0 → − − (mj qj ) = 0 ˙ (1. t) + V (r.31) allows one to transform the potentials without aﬀecting the ﬁelds and without aﬀecting the equations of motion (1.29) where ρ(r.22.31) = − V − Gauge Symmetry of the Electromagnetic Field It is well known that the relationship between ﬁelds and potentials (1. x3 (t)) moving in an electromagnetic ﬁeld described through the electrical and magnetic ﬁeld components E(r. t) = A(r.28) (1.27) (1. The equations of motion for such a particle are d q ˙ (mr) = F (r. c ∂t (1. x2 (t).27) and (1. t) dt c where dr = v and where F (r.30. t) + v × B(r. t) and B(r. t) − K(r.32) (1. t) 1∂ K(r. t).28) can be satisﬁed implicitly if one represents the ﬁelds through a scalar potential V (r. t) . t) as follows B E = ×A 1 ∂A . t) and B(r. t) c ∂t (1.25) of a particle moving in the ﬁeld.23). Equations (1. t) describes the charge current density. 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). dt The ﬁelds E(r. t) is the Lorentz force.25) = 0 = 0 4π J c = 4πρ = (1.24) ∂qj dt ∂ qj ˙ ∂qj dt which are obviously equivalent to the Newtonian equations of motion. F (r. 1.1.33) .30) (1.2: Lagrangian 5 For a proof of the Hamiltonian Principle of Least Action we inspect the Euler–Lagrange conditions associated with the action integral deﬁned through (1. The transformation which leaves the ﬁelds invariant is A (r.

38.39) together yield ∂V q d (mx1 ) = −q ˙ + O dt ∂x1 c where O = ∂A1 ∂A2 ∂A3 ∂A1 ∂A1 ∂A1 x1 + ˙ x2 + ˙ x3 − ˙ x1 − ˙ x2 − ˙ x3 ˙ ∂x1 ∂x1 ∂x1 ∂x1 ∂x2 ∂x3 ∂A2 ∂A1 ∂A1 ∂A3 = x2 ˙ − − x3 ˙ − ∂x1 ∂x2 ∂x3 ∂x1 (1. 1.40) (1.g. namely. 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 . r.35) is equivalent to the Euler–Lagrange condition d dt The second term in (1. t) = 1 q mv 2 − q V (r.38) ∂L ∂ x1 ˙ − ∂L = 0. Comparing then (1. (1. t) + A(r. (1. ∂x1 (1.37) The results (1. (1. t) · v .40. if one assumes for a particle the Lagrangian ˙ L(r. dt ∂x1 c We notice using (1.35) − ∂A1 ∂x2 [v × B]1 = x2 B3 − x3 B2 = x2 ˙ ˙ ˙ ∂A2 ∂A1 − ∂x1 ∂x2 − x3 ˙ ∂A1 ∂A3 − ∂x3 ∂x1 . d ∂V q (mv1 ) = F1 = −q + [v × B]1 . 2 c (1.34) For this purpose we consider only one component of the equation of motion (1.37) is ∂L ∂V q = −q + ∂x1 ∂x1 c The ﬁrst term in (1.36) in the Newtonian equation of motion.30). B3 = ∂A2 ∂x1 (1.39) ∂A1 ∂A2 ∂A3 x1 + ˙ x2 + ˙ x3 ˙ ∂x1 ∂x1 ∂x1 .6 Lagrangian of Particle Moving in Electromagnetic Field Lagrangian Mechanics We want to show now that the equation of motion (1.35) shows that the Newtonian equations of motion and the Euler–Lagrange conditions are.25) follows from the Hamiltonian Principle of Least Action.. . in fact.25). equivalent.36) This expression allows us to show that (1. e.41) which is identical to the term (1.41) with (1.

t) 1 1 ∂K q mv 2 − q V (r. r.43) To prove this theorem we determine the action integral corresponding to the transformed Lagrangian t1 S [q(t)] = t0 ˙ dtL (q. t) = q K(q.32. 1. r.44) Since the condition δq(t0 ) = δq(t1 ) = 0 holds for the variational functions of Lagrangian Mechanics. 2 c c dt Inserting (1. q. The factor q has been introduced to make this c transformation equivalent to the gauge transformation (1.42) This transformation is termed gauge transformation. i.33) of electromagnetic potentials. t) + A (r. q. equivalent to the gauge transformation (1. Note that one adds the total time derivative of a function K(r.32. This term is d ∂K ∂K ∂K ∂K ∂K K(r. t) = mv 2 − q V (r. t) = L(q.33) of electromagnetic potentials. t) + A(r.. t) L (q. encountered above in connection with the transformations (1. They are the subject of the following theorem.32. 1. t) · v . The gauge symmetry.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. We will consider in the following two symmetries. One can conclude then immediately that any extremal of S [q(t)] is also an extremal of S[q(t)]. t) + dt c (1.44) implies that the gauge transformation amounts to adding a constant term to the action integral. q. t) . appear in a diﬀerent. surprisingly simple fashion in Lagrangian Mechanics. 2 c = A(r.43) into (1. in fact. 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. We want to demonstrate now that the transformation (1. t) c t1 t0 t1 t0 ˙ dtL(q. (1.33) of electyromagnetic potentials.e.46) (1. t) = x1 + ˙ x2 + ˙ x3 + ˙ = ( K) · v + . t) · v + K(r.45) and reordering terms yields using (1. q.32.45) . 1.33) ˙ L (r. gauge symmetry and symmetries with respect to spatial transformations. 1. t) − + 2 c ∂t c 1 q = mv 2 − q V (r.42) is. Eq.3: Symmetry Properties 7 1.34) q 1 q d ˙ L (r.1. t) c t1 t0 = S[q(t)] + (1. t) the Lagrangian. dt ∂x1 ∂x2 ∂x3 ∂t ∂t (1. and since these properties allow one often to simplify mathematical descriptions. t) + K ·v (1. a term not aﬀected by the variations allowed. For this purpose we consider the transformation of the single particle Lagrangian (1. t) + q K(q.

i.32. the equivalence of (1. Such invariance properties are very familiar. t). A (r.48) In the following we will denote unit vectors as a.51) We would like to derive this transformation in a somewhat complicated. ) . We have proven. A(r. in the case of central force ﬁelds which are invariant with respect to rotations of coordinates around the center. the transformation (1. However. but nevertheless instructive way considering rotations around the x3 –axis. r ∈ R . which has an important analogy in Quantum Mechanics. Deﬁnition: Inﬁnitesimal One-Parameter Coordinate Transformations A one-parameter coordinate transformation is decribed through r = r (r. We consider now invariance properties connected with coordinate transformations. The corresponding inﬁnitesimal transformation is deﬁned for small r (r. we like to ˆ provide the transformation here anyway for later reference r = r + e. The transformations considered are speciﬁed in the following deﬁnition. ˆ ˆ ˆ Examples of Inﬁnitesimal Transformations The beauty of inﬁnitesimal transformations is that they can be stated in a very simple manner. for example.42) corresponds to replacing in the Lagrangian potentials V (r. therefore. and for the introduction of inﬁnitesimal transformations which will provide a crucial method for the study of symmetry in Quantum Mechanics..49) =0 ∈ R (1. 1. In this case the transformation can be written in matrix form cos −sin 0 x1 x1 x2 = sin cos 0 x2 (1. ˆ (1.52) x3 0 0 1 x3 .e.33). r. t). In case of a translation transformation in the direction e nothing new is gained. t) by gauge transformed potentials V (r. The transformations we consider are the most simple kind.42) and (1. The following description of spatial symmetry is important in two respects. 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. ) = r + R(r) + O( 2 ) . for the connection between invariance properties and constants of motion. 0) = r . where the origin of is chosen such that r (r. R(r) = through ∂r ∂ (1.8 Lagrangian Mechanics Obviously. for such vectors holds a · a = 1. t). ˆ (1.47) (1.50) A non-trivial example is furnished by the inﬁnitesimal rotation around axis e ˆ r = r + e×r .

Theorem: Noether’s Theorem ˙ If L(q. however. r) which is constant along the classical path of the system. a quantity C(r. t) + j=1 ∂L Qj + ∂qj M j.e. the second and third term ˙ k=1 in (1. a classical mechanical system is invariant with respect to a coordinate transformation ˙ a constant of motion exists. t) = L(q. We consider here only the ‘particle version’ of the theorem.56) ∂Qj qk . . the property holds for any system. neglecting terms of order O( M ˙ ˙ L (q. ˙ We have generalized in this theorem the deﬁnition of inﬁnitesimal coordinate transformation to M –dimensional vectors q. that along the classical path holds d ∂L ∂L the Euler-Lagrange condition ∂qj = dt ( ∂ qj ) one can rewrite the sum of the second and third term ˙ in (1. q.e. In order to prove Noether’s theorem we note qj qj ˙ = qj + = qj + ˙ k=1 Qj (q) M (1. Anytime. t) is invariant with respect to an inﬁnitesimal transformation q = q + Q(q).57) ∂ d where we used dt Qj = M ( ∂qk Qj )qk .57) M Qj j=1 d dt ∂L ∂ qj ˙ + ∂L d Qj ∂ qj dt ˙ = d dt M Qj j=1 ∂L = 0 ∂ qj ˙ (1. expansion.1. We have used here the notation corresponding to single particle motion.53) x3 x3 0 0 0 x3 One can readily verify that in case e = e3 (ˆj denoting the unit vector in the direction of the ˆ ˆ e xj –axis) (1. In the context of Lagrangian Mechanics this implies that such transformation leaves the Lagrangian of the system unchanged.3: Symmetry Properties 9 In case of small this transformation can be written neglecting terms O( 2 ) using cos = 1 + O( 2 ).53).55) (1.51) reads r = r − x2 e1 + x1 e2 ˆ ˆ (1. q.54) which is equivalent to (1.k=1 ∂L ∂Qj qk ˙ ∂ qj ∂qk ˙ (1.. Before the embark on this theorem we will comment on what is meant by the statement that a classical mechanical system is invariant under a coordinate transformation. t) yields after Taylor ˙ 2 ). i. by Noether.58) From this follows the statement of the theorem. Using the fact. sin = + O( 2 ) 0 − 0 x1 x1 x1 x2 = x2 + 0 0 x2 + O( 2 ) . q. (1. i. The property has been shown to hold in a more general context. namely for ﬁelds rather than only for particle motion. q.57) must cancel each other or both vanish.. Invariance implies L = L. ˙ ∂qk ˙ Inserting these inﬁnitesimal changes of qj and qj into the Lagrangian L(q. then M ∂L j=1 Qj ∂ xj is a constant of motion.

3) 3 Qj j=1 ∂L = e· ˆ ∂ xj ˙ 3 ej mxj = e · mv . 3 and. ˆ ˙ ˆ j=1 (1. i. 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. is conserved. It was.50). 2. 2. Using the cyclic property (a × b) · c = (b × c) · a = (c × a) · b allows one to rewrite the e constant of motion e · (r × mv) which can be identiﬁed as the component of the angular momentum ˆ mr × v in the e direction. v) = 1 mv 2 − U (r).10 Application of Noether’s Theorem Lagrangian Mechanics We consider brieﬂy two examples of invariances with respect to coordinate transformations for the Lagrangian L(r. not inﬁnitesimal transformations. In this case we have Qj = ej · e. 2 We ﬁrst determine the constant of motion in case of invariance with respect to translations as deﬁned in (1. except using now Qj = ej · (ˆ × r).59) yields the constant of motion ˆ e (ˆ × r) · mv. j = 1. j = 1. A calculation similar to that in (1. . It is not necessary to investigate the consequences of global.51).59). Noether’s theorem yields ˆ ˆ the constant of motion (qj = xj . In this case we will use the same notation as in (1.e. 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. hence. of course. to be expected that this is the constant of motion. for which translational invariance holds.

t1 ] . The probability that the particle is detected at space–time point r. t)|2 = ψ(r. t1 ] in which the particle is known to exist.3) where Ω is the space volume in which the particle can exist.1) (2.2 Axioms for Quantum Mechanical Description of Single Particle ˙ Let us consider a particle which is described by a Lagrangian L(r. t) where z is the conjugate complex of z. 2. t is |ψ(r. t)|2 Ω (2. t)|2 = 1 Ω ∀t. The wave function ψ(r. t ∈ [t0 . At present one may think of f (r) as a sum over δ–functions which represent a multi–slit screen. is unity. (2.1 The Double Slit Experiment Will be supplied at later date 2. r.2) (2. The probability to detect the particle with a detector of sensitivity f (r) is d3 r f (r) |ψ(r. The particle is described by a wave function ψ(r. 3. 11 . 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. t) is normalized d3 r |ψ(r. 4.4) a condition which enforces that the probability of ﬁnding the particle somewhere in Ω at any particular time t in an interval [t0 . 1. t) ψ : R3 ⊗ R → C.Chapter 2 Quantum Mechanical Path Integral 2. placed into the space at some particular time and with a detector behind each slit. t)ψ(r. t).

e. t. tN −1 ) φ(rN −1 . the path integral r(tN )=rN d[r(t)] · · · r(t0 )=r0 (2. t. t0 ) .. t1 ] yields an expression of the form r(tN )=rN φ(r.8) 7. i. (2. t ) t > t . hence. t ∈ [t0 . t1 |r0 . t0 ) = Ωd 3r N −1 Ωd 3r N −2 ··· Ωd 3r 1 φ(r. namely the point on the classical path at time t . The time evolution of ψ(r. t0 ) = φ(r.7) t. t ) which requires d3 r φ(r. This must hold for any ψ(r . t ) ψ(r . t1 ]) d3 r φ(r.10) We have introduced here a new symbol. t|r . t) is described by a linear map of the type ψ(r.9) for N → ∞. t |r0 . t|r0 . t0 ) = r(t0 )=r0 d[r(t)] Φ[r(t)] . The state of a system at time t.5) 6.. 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.9) Employing a continuum of intermediate times t ∈ [t0 . t = )|2 = 1 . The generalization of the completeness property to N − 1 intermediate points t > tN −1 > tN −2 > . tN −1 |rN −2 . t ∈ [t0 . t0 ) = δ(r − r0 ).6) φ(r. ψ(r0 . The deﬁnition will actually assume an inﬁnite number of intermediate times and express the path integral through integrals of the type (2. . t ) ψ(r . at any intermediate time the particle needs only be known at one space point. t1 ] (2. t ∈ (2. t|r . requires then according to (2. t ) φ(r . t|r0 . t) = Ω d3 r φ(r. i. This symbol will be deﬁned further below. the propagator is unitary.8) a knowledge of the state at all space points r ∈ Ω at some intermediate time t . t|r . t|r . t| rN −1 . t)|2 = (2. The following so-called completeness relationship holds for the propagator (t > t [t0 . 8.11) which denotes an integral over all paths r(t) with end points r(t0 ) = r0 and r(tN ) = rN . described by ψ(r. (t > t . t ) Ωd 3 r |ψ(r. This is diﬀerent from the classical situation where the particle follows a discrete path and. tN −2 ) · · · φ(r1 . . t )φ(r. t|r .4) holds for all times. t|r0 . > t1 > t0 is φ(r. t ) = δ(r − r ) Ω (2.e. t ) φ(r. Since (2. t|r . (2. t1 ]) Ωd Ωd 3r Ωd 3r Ωd 3r 3 r |ψ(r.12 Quantum Mechanical Path Integral 5. t ) ψ(r . . t).

. 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. The value of S[r(t)] is then Scl = 1 1 m v 2 t = erg s ..13) L(r. the exponent is a very large imaginary number. i. Since this number is multiplied by ‘i’. Its value is extremely small in comparision with typical values for action integrals of macroscopic particles.e. t) (2. Any variations of Scl would then lead to strong oscillations of the contributions exp( i S) to the path integral and one can expect destructive interference betwen these contributions. 2 2 (2. The situation is very diﬀerent for microscopic particles.12. for the electron many paths contribute and the action integrals for non-classical paths need to be known.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. However. 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. Scl / ≈ 8.2. for microscopic particles like the electron this is by no means the case.e. such that destructive interference of the contributions of such paths does not occur. i. This implies that small variations of the path near the classical path alter the value of the action integral by very little. accordingly. Only for paths close to the classical path is such interference ruled out. not only for the classical path. a very large number.15) The exponent of (2.13) and = 1. The functional Φ[r(t)] in (2. Situations which are well described classically will be distinguished through the property that the classical path gives the dominant. However. 2.0545 · 10−27 erg s . However. (2.2: Axioms 9. One would still expect signiﬁcant descructive interference between contributions of diﬀerent paths since the value calculated is comparable to 2π. r.14) has the same dimension as the action integral S[r(t)]. 2. In case of a proton with mass m = 1.11) is Φ[r(t)] = exp where S[r(t)] is the classical action integral tN 13 i S[r(t)] (2.12) is then Scl / ≈ 0.12. .14) ˙ In (2. expressions (2. contribution to the path integral (2. r.13).12) S[r(t)] = t0 ˙ dt L(r. However. actually often essentially exclusive. The constant given in (2. it is comparable to action integrals as they arise for microscopic particles under typical circumstances.5 · 1027 . interferences should be much less dramatic than in case of the macroscopic particle. t) is the Lagrangian of the classical particle. in complete distinction from Classical Mechanics. namely due to the property of the classical path to be an extremal of the action integral.13) are built on action integrals for all possible paths.

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 . tN )} . .3 How to Evaluate the Path Integral In this section we will provide an explicit algorithm which deﬁnes the path integral (2. t0 ) = limN →∞ CN exp i N N −1 j=0 +∞ +∞ +∞ −∞ dx1 −∞ dx2 . The spacings between the times tj+1 and tj will all be identical. however. provides an avenue to evaluate path integrals.9). namely 1 m(xj+1 − xj )2 / 2 and U (xj ). x.20) can be properly taken.13) and. tj ) to the next (xj+1 . etc. 2 2m N (2. tj+1 ) we assume that kinetic energy and potential energy are constant. They can be anywhere in the allowed volume which we will choose to be the interval ] − ∞. a series of times tN > tN −1 > tN −2 > .18) The time instances are ﬁxed. .20) Here. at the same time. ∞[. tN |x0 . Its value is CN = m 2πi N 2 (2. 2. x2 .14 Quantum Mechanical Path Integral 2. t0 ). tN −1 ). respectively. t) = ˙ 1 mx2 − U (x) . These assumptions lead then to the following N 2 Riemann form for the action integral N −1 S[x(t)] = lim N →∞ N j=0 1 (xj+1 − xj )2 m − U (xj ) 2 2 N . as in (2.12. (x1 .21) N 2. the xj values are not. . (xN .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. . . namely tj+1 − tj = (tN − t0 )/N = N . (2. (2. ˙ 2 (2. The particles have associated with them a Lagrangian L(x. (2. t1 ). . In passing from one space–time instance (xj . This allows us to write the evolution operator using (2. . > t1 > t0 letting N go to inﬁnity later. −∞ dxN −1 (xj+1 −xj )2 1 − U (xj ) .19) The main idea is that one can replace the path integral now by a multiple integral over x1 .12) φ(xN . 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. (xN −1 .10) and (2.16) In order to deﬁne the path integral we assume. .

. (2.22) where xcl (t) is the classical path which connects the space–time points (x0 . x(tN ) = 0. tN dt y = y(tN ) − y(t0 ) = 0 .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. of (2. tN ). which can also be written φ(xN . The resulting expression for S[x(t)|x(t0 ) = x0 . tN |0. One obtains S[x(t)|x(t0 ) = x0 .s. due to (2. can be expressed through a path integral with endpoints x(t0 ) = 0. tN |x0 .25) is. ˙ t0 (2.h.31) We have denoted explicitly that the action integral for a path connecting the space–time points (x0 .s.23) tN − t0 It is essential for the following to note that.4: Propagator for a Free Particle 15 Rather then using the integration variables xj . The ˙ 1 S[x(t)|x(t ) = x . namely. tN dt t0 1 1 (xN − x0 )2 m x2 = m ˙ cl . tN |x0 .29) This expression corresponds to the action integral in (2.24) Also we use the fact that the velocity of the classical path xcl = (xN −x0 )/(tn −t0 ) is constant. tN ) is to be evaluated. t0 ) = exp 1 im (xN − x0 )2 2 tN − t0 φ(0.h. since x(t0 ) = xcl (t0 ) = x0 and x(tN ) = xcl (tN ) = xN . t0 ) and (xN . the origin of which coincides with the classical path of the particle. it is more suitable to deﬁne new integration variables yj . t0 ) = exp S[x(t)] (2. x(tN ) = 0] . xN − x0 xcl (t) = x0 + ( t − t0 ) . x(tN ) = xN ] 1 (xN − x0 )2 m + 0 + 2 tN − t 0 + S[x(t)|x(t0 ) = 0. using (2. t0 ) (2. = (2.10.24). ˙ 2 t0 (2. (2. Inserting the result into (2.23)..26) The ﬁrst term on the r. t0 ) and (xN .24) implies for the second term on the r.12) yields x(tN )=0 im (xN − x0 )2 i φ(xN . ˙ ˙ ˙ 2 + y2) = ˙ (2. 2. 2 2 tN − t0 (2. x(tN ) = xN ] is S[x(t)|x(t0 ) = x0 .e. To see the beniﬁt of such approach we deﬁne a path y(t) as follows x(t) = xcl (t) + y(t) (2. x(tN ) = 0] . it holds y(t0 ) = y(tN ) = 0 .28) i.13).27) The third term can be written in the notation introduced tN 1 dt my 2 = S[x(t)|x(t0 ) = 0. 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 .25) + The condition (2.30) d[x(t)] exp 2 tN − t0 x(t0 )=0 a result.

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

. . det(Ajk ) (2. .46) . Our derivation has provided us with the value det(Ajk ) = N . . n = 1.42) m 2πi NN 1 2 (2. . 2 −1 0 0 0 −1 2 For this purpose we expand (2. . which follows from (2. 0 0 Dn = . t|x0 .. . Using D1 = |(2)| = 2 . D1 = 2 and obtain from (2. (2. t0 ) = m 2πi (t − t0 ) 1 2 exp im (x − x0 )2 2 t − t0 . . . tN |0.. We seek then to evaluate the determinant of the n × n matrix 2 −1 0 . one can readily verify Dn = n + 1 .. . . One can readily verify that this procedure leads to the following recursion equation for the determinants Dn = 2 Dn−1 − Dn−2 . 0 0 0 .. .45) Expressions for Free Particle Propagator We have now collected all pieces for the ﬁnal expression of the propagator (2. .39) yields φ(0. t0 ) = m 2πi (tN − t0 ) 1 2 . x = xN φ(x.40) . (2. presently N − 1. tN |0. t0 ) = limN →∞ and with NN (2. To solve this three term recursion relationship one needs two starting values.44) = tN − t0 .2.. 0 0 0 −1 2 . .40) in terms of subdeterminants along the last column. .. 2. . let say n.. .18) we obtain φ(0.. Inserting this into (2.31) and obtain.. . . tN |0.39) In order to determine det(Ajk ) we consider the dimension n of (Ajk ).41) the same result for D2 . t0 ) = limN →∞ m 2πi N 2 17 N 2πi m N N −1 2 1 . (2. . . we obtain φ(0. (2. variable. 0 0 −1 2 −1 . .41) D2 = 2 −1 −1 2 = 3 (2.4: Propagator for a Free Particle a property which follows from det(cB) = cn detB for any n × n matrix B. D3 . .43) We like to note here for further use below that one might as well employ the ‘artiﬁcial’ starting values D0 = 1. deﬁning t = tN . .

We will obtain. t0 ) = m 2πi (t − t0 ) 3 2 exp im (r − r0 )2 2 t − t0 . x) + E(x) in (2. the wave function of the particle at later times. t0 )| = 2 1 πδ 2 1 2 exp − x2 0 δ2 (2.46).53) . thereby. t0 ) = 1 πδ 2 1 4 exp − x2 po 0 +i x 2 2δ (2. and describes a single particle since 1 πδ 2 1 2 +∞ dx0 exp −∞ − x2 0 δ2 = 1.48) Obviously. One obtains then for the propagator (2. x) = im 2 t x2 1 + i o t mδ 2 − 2xo x − po t m + f (x) (2. t|r0 . (2.51) as follows Eo (xo .46) can be generalized to three dimensions in a rather obvious way.50) One refers to such states as wave packets.5) allows us to predict the time evolution of any state function ψ(x. We devide the contributions to the exponent Eo (xo . We need to evaluate for this purpose the integral ψ(x.10) φ(r. x) + E(x) (2.49) is Gaussian of width δ. centered around x0 = 0. (2.48) as the initial state using the propagator (2.18 Quantum Mechanical Path Integral This propagator. We want to apply axiom (2. t) = 1 πδ 2 1 4 m 2πi t 1 2 +∞ dx0 exp −∞ im (x − x0 )2 x2 po − 02 + i xo 2 t 2δ Eo (xo . the associated probability distribution |ψ(x0 . The result (2.52) and applying (2. t) of a free particle. according to (2. Below we will apply this to a particle at rest and a particle forming a so-called wave packet.247).5) to (2.47) One-Dimensional Free Particle Described by Wave Packet We assume a particle at time t = to = 0 is described by the wave function ψ(x0 .51) For this evaluation we adopt the strategy of combining in the exponential the terms quadratic (∼ x2 ) and linear (∼ x0 ) in the integration variable to a complete square 0 ax2 + 2bx0 = a 0 x0 + b a 2 − b2 a (2.

t) = 1 πδ 2 1 4 2 t 1 + i mδ2 .2. (2. however.56) m 2πi t 1 2 +∞ eE(x) −∞ dx0 eEo (xo . x) = One can write then (2.52). 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 .x) (2. (2. according to (2.x) im xo 2 t t − mδ 2 x− 1+ x− 1+ po m = −∞ +∞ dx0 exp 1+i t 2 2 t i mδ2 = −∞ im dx0 exp 2 t t 1+i 2 mδ xo − po m t t i mδ2 .55) im xo 2 t 1+i t − mδ 2 x− 1+ po m t t i mδ2 .60) is then represented by real ρ values. z2 = +∞ × i 1−i t mδ 2 (2. the square in (2. that the new path does not lead to additional contributions to the integral.51) ψ(x. We proceed as follows.61) .54) f (x) = This yields Eo (xo . The beginning and the end of such path are the points z1 = −∞ × i 1−i t mδ 2 .4: Propagator for a Free Particle im x2 − f (x) . making certain. 2 (2.59) An integration path in the complex x0 –plane along the direction i 1−i t mδ 2 (2.53) E(x) = x− po mt 19 (2.57) and needs to determine the integral +∞ I = −∞ +∞ dx0 eEo (xo . (2.58) The integrand is an analytical function everywhere in the complex plane and we can alter the integration path. 2 t One chooses then f (x) to complete.

(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 (2.70) . (2.58) vanishes 0 for Re x0 → ±∞.58) along the path z1 → z1 and along the path z2 → z2 gives only vanishing contributions one can replace (2.66) yields ψ(x. using a ab = a − . We can conclude then that (2. Quantum Mechanical Path Integral (2. t m(1 + i mδ2 ) (2. z2 = +∞ .54) using (2.20 whereas the original path in (2. one can show that z1 lies at −∞ − i × ∞ and z2 at +∞ + i × ∞. In fact.68) and.63) holds and. accordingly. t) = Seperating the phase factor 1 t 1 + i mδ2 1 4 1 2 exp [ E(x) ] .55). I = Equation (2.63) which can be readily evaluated. 1+b 1+b ﬁnally E(x) = − (x − po m (2. (2.57) reads then 1 πδ 2 1 4 2πi t . Since the exponent in (2.58) has a leading contribution in x0 of −x2 /δ 2 the integrand of (2. the paths between z1 → z1 and z2 → z2 have a real part of x0 of ±∞.62) If one can show that an integration of (2. 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) ] .58) has the end points z1 = −∞ .64) ψ(x.65) t 1 − i mδ2 t 1 + i mδ2 .58) by t mδ 2 +∞ I = i 1−i dρ exp − −∞ m 2 t 1+ t mδ 2 2 ρ2 (2.69) t)2 t 2δ 2 (1 + i mδ2 ) + i po x − i p2 o t 2m (2. Hence.67) We need to determine ﬁnally (2.

The width of the distribution (2.72) for the initial state (2.48) and for the ﬁnal state (2. This ‘spreading’ of the wave function is a genuine quantum phenomenon. However.49).. t) = t 1 − i mδ2 t 1 + i mδ2 1 4 21 1 πδ 2 (1 + 2 t2 ) m2 δ 4 1 4 × po i p2 t o ) + i x − t mδ 2 2m (2. respectively.73) m δ increases with time. Another interesting observation is that the wave function (2.71) × exp − (x − po m t)2 ) 2δ 2 (1 + 2 t2 m2 δ 4 (1 − i . The corresponding probability distribution is |ψ(x.77) . (2.49). (2. t0 ) = exp i po xo m . This yields.74) remains invariant in time. (2.72).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 → ∞. the case of arbitrary to is recovered iby replacing t → to in (2. t) = exp i po i p2 o x − t m 2m .72) moves in the direction of the positive x-axis with velocity vo = po /m which identiﬁes po as the momentum of the particle. The center of the distribution (2.74) which corresponds to (2.72) Comparision of Moving Wave Packet with Classical Motion It is revealing to compare the probability distributions (2. t)| = 2 1 πδ 2 (1 + 2 t2 m2 δ 4 1 2 ) exp − (x − δ 2 (1 + po m t)2 ) 2 t2 m2 δ 4 . 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 .71) conserves the phase factor exp[i(po / )x] of the original wave function (2. (2. In this case holds ψ(x.76) From this we conclude that an initial wave function ψ(xo . instead of (2.71.67) provides the complete expression of the wave function at all times t ψ(x.f.75) i.72) 2 t2 δ 1+ 2 4 (2.75) ψ(x. t) = exp i i p2 po o x − (t − to ) m 2m . the spatial dependence of the initial state (2.71). We had assumed above [c.e. t0 ) = exp i po i p2 o xo − to m 2m . (2.48)] to = 0.2. coinciding at t = 0 with the width of the initial distribution (2. The conservation of this factor is particularly striking for the (unnormalized) initial wave function ψ(x0 . 2. (2.4: Propagator for a Free Particle which inserted in (2.72).

. 2. One refers to the respective states as stationary states. y(tN ) = 0 . y(t).82) which describes the deviation from the classical path xcl (t) with end points xcl (t0 ) = x0 and xcl (tN ) = xN . t) = L(xcl .78) i.82) one obtains L(xcl + y.79) for a quadratic Lagrangian L(x. ˙ ˙ = (2.85) We want to show now that the contribution of δL to the action integral (2.83) (2. t) ˙ L (y. Such states play a cardinal role in quantum mechanics. xcl . x.80) For this purpose we need to determine the action integral S[x(t)] = t0 dt L(x. In order to simplify this task we deﬁne again a new path y(t) x(t) = xcl (t) + y(t) (2.81) vanishes2 .12. 2. t) = exp i Quantum Mechanical Path Integral po i p2 o x − t m 2m . For this purpose we use d xcl y = ˙ ˙ (xcl y) − xcl y ˙ ¨ (2. the spatial as well as the temporal dependence of the wave function remains invariant in this case.81) for an arbitrary path x(t) with end points x(t0 ) = x0 and x(tN ) = xN . y(t).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 . (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 . xcl + y(t).e.5 Propagator for a Quadratic Lagrangian x(tN )=xN We will now determine the propagator (2.10. t) = ˙ 1 1 mx2 − c(t)x2 − e(t)x . t0 ) = x(t0 )=x0 d[x(t)] exp i S[x(t)] (2.80) into (2. Obviously. the end points of y(t) are y(t0 ) = 0 Inserting (2. xcl . t) + L (y. t) ˙ (2. t) + δL ˙ ˙ ˙ ˙ where L(xcl . 2 . 1.84) . ˙ 2 2 tN (2. tN |x0 . y(t)] which vanishes according to the Hamiltonian principle as discussed in Sect. 2.22 becomes at t > to ψ(x. (2. x.13) φ(xN .

tN |x0 .87) According to (2. . . One can then express the propagator (2. also the second contribution on the r. . 0 0 0 c2 0 . .91) (yj+1 − yj )2 1 2 2m N where cj = c(tj ). vanishes. 2 −1 0 0 0 −1 2 c1 0 0 . cN −2 0 0 0 0 0 cN −1 (2.91) through the quadratic N −1 E = i j. .. The discretization scheme adopted above yields in the present case ˜ φ(0. . y.83) the ﬁrst term on the r.s. 2 N . 0 0 0 .. t0 ) = y(tN )=0 d[y(t)] exp y(t0 )=0 i tN dt L (y. . . . . . . . Applying the Euler–Lagrange conditions (1. . tN |0. . 2 . . tN |0.80) yields for the classical path m¨cl + c(t) xcl + e(t) = 0 x (2. 0 0 N 0 − . . . of (2. ..h.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. . . One can express the exponent E in (2.79) i ˜ φ(xN .88) vanishes. . .24) to the Lagrangian (2. t0 ) = exp S[xcl (t)] φ(0. tj = t0 + form j. . . . . . .92) where ajk are the elements of the following (N − 1) × (N − 1) matrix 2 −1 0 . . t0 ) = limN →∞ × +∞ +∞ −∞ dy1 · · · −∞ dyN −1 exp N i N N −1 j=0 m 2πi N 2 N × − 1 2 cj 2 yj (2. t) ˙ t0 .88) and. . 0 0 0 . t0 ) (2. . . 0 0 m ajk = . tN |0.32).s. t0 ) we will adopt a strategy which is similar to that used for the evaluation of (2. (2.2. . x t0 (2.. 0 0 −1 2 −1 . .h..89) where ˜ φ(0. . 0 0 0 −1 2 . .. .. For the evaluation of φ(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) . tN |0. 0 0 0 c3 . .31) for the free particle propagator. hence. .93) .k=1 yj ajk yk (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

1 exp[iω(n + 2 ) to ] · · · .e. i. t). to ).36 Altogether.186) Fourier transform. . t|xo . Accoordingly. In contrast to W (x.171) concludes the proof of (2.h.188) identiﬁes the functions ψn (x. to ) we express (2. one obtains for the r. 2 (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. We note that the factor exp(2 mω/ xe−iωt − e−2iωt ) in (2.184) 1 ˜ exp[−iω(n + 1 ) t] φn (x) 2 n! ∞ +∞ = m=0 −∞ dxo Φ(x. t|xo .172) I = exp 2 mω x e−iωt − e−2iωt − Quantum Mechanical Path Integral mω 2 1 x − iωt 2 2 .171.171) ∞ W (x. which also exhibits such invariance.188) ˜ ˜ Equation (2. but not of t. .147) in (2.182) Comparision with (2.172). Noting that the propagator (2. t) +∞ −∞ dto 1 ˜ exp[−iω(m + 1 ) to ] φm (xo ) 2 m! (2.. n = 1. We want to inspect the consequences of the invariance property (2. (2. t) = exp[−iω(n + 1 ) t] φn (x) as invariants under the 2 action of the propagator φ(x. of (2. results in 1 ˜ exp[−iω(n + 1 ) t] φn (x) 2 n! +∞ = −∞ dxo Φ(x. xo . 2. xo . 2. .. to ) exp[−iω(n + 1 ) to ] φn (xo ) .183) where the expansion coeﬃcients are functions of x. xo . one can expand (2. t − to ) = φ(x.187) or 1 ˜ exp[−iω(n + 2 ) t] φn (x) +∞ = −∞ ˜ dxo φ(x.172) in the form ∞ n=0 (2.172).172) is a function of t − to and deﬁning accordingly Φ(x. xo . t|xo . t) = n=0 1 ˜ exp[ − iω(n + 1 ) t ] φn (x) 2 n! (2. t − to ) 1 ˜ φn (xo ) n! (2.171) can be expanded in terms of e−inωt . t − to ) 1 ˜ exp[−iω(m + 1 ) to ] φm (xo ) 2 m! (2.s. .

through expansion (2. 2 Here Nn are constants which normalize ψn (x.190) ˜ is obeyed. we have identiﬁed then.191) (2.183). z) = n=0 1 2 /2 1 2 /2 = = mω e−iωt x (2.197) are called Hermite polynomials which are polynomials of degree n which will be evaluated below. t)|2 = ˜n (x)|2 . t) = z 2 e−y where w(y. t) are associated with a time-independent probablity density |ψn (x. z) = Hn (y) (2. Expression (2.193) (2. For this purpose we introduce ﬁrst closed expressions for φ the new variables y z and write (2. . In the following we will characterize the functions φn (x) and determine the normalization constants Nn . ˜ |φ stationary wave functions ψn (x.194) = z2 n=0 ∞ 1 ∞ zn ˜ φn (y) n! (2. This follows from ∂n w(y.192) w(y. characterizes the wave functions φn (x). The Hermite Polynomials ˜ The function (2. t) provide all stationary states of the quantum mechanical harmonic oscillator.194) plays a central role for the Hermite polynomials since it contains. according to (2. z) (2.171).171) W (x.183). To obtain ˜n (x) we simplify the expansion (2. .2. 2. .197) The expansion coeﬃcients Hn (y) in (2. Actually. t) such that +∞ −∞ 2 dx |ψ(x. (2.183) reads then w(y. 1.198) ∂z n z=0 . through the expansion coeﬃcients φn (x) in (2. z) = exp(2yz − z 2 ) .194). t) = exp[−iω(n + 1 ) t] Nn φn (x) . in a ‘nutshell’ all information on the Hermite polynomials.195) zn Hn (y) n! ˜ φn (y) . z) z 2 e−y or w(y.5: Propagator for a Quadratic Lagrangian 37 ˜ ˜ the functions ψn (x. Expansion (2.196) where Hn (y) = ey 2 /2 (2.189) ˜ dx φ2 (x) = 1 n (2. t) of the quantum mechanical harmonic oscillator ˜ ψn (x. (2. t)|2 = Nn +∞ −∞ n = 0.183). We will also argue that the functions ψn (x.

196) results in the so-called Rodrigues formula for the Hermite polynomials Hn (y) = (−1)n ey 2 ∂ n −y2 e . Massachusetts. 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 (2.38 Quantum Mechanical Path Integral which is a direct consequence of (2. We start from 2 e2yz − z ∞ ν = ν=0 µ=0 ∞ ν 1 ν! 1 ν ν µ ν µ z 2µ (−1)µ (2y)ν−µ z ν−µ (−1)µ (2y)ν−µ z ν+µ = ν=0 µ=0 3 (2.200) yields for the ﬁrst Hermite polynomials H0 (y) = 1. D. µ lattice (left diagram) corresponds to a summation over only every other point in an n. 2.194. m lattice (right diagram).201) ν−µ µ µ ν−µ Figure 2. z) the generating function for the Hermite polynomials. normalization factors for φ(y).202) generatingfunctionology by H. H1 (y) = 2y.L. The identity (2..196) ∂n w(y. One calls w(y. (2. Inc. We want to derive now explicit expressions for the Hermite polynomials. ν H2 (y) = 4y 2 − 2. . For this purpose we expand the generating function (2. . that Hn (y) is a polynomial of degree n. Boston. 1990) is a useful introduction to this tool as is a chapter in the eminently useful Concrete Mathematics by R. . 1989). ν+µ (2.Graham. i. We will employ (2. and recursion equations for the eﬃcient evaluation of Hn (y). The diagrams illustrate that a summation over all points in a ν.196). and O. . The diagrams also identify the areas over which the summation is to be carried out.Knuth. among other properties.e.Wilf (Academic Press.198) for the Hermite polynomials can be expressed in a more convenient form employing deﬁnition (2.194) in a Taylor series in terms of y p z q and identify the corresponding coeﬃcient cpq with the coeﬃcient of the p–th power of y in Hq (y). ν+µ ν H3 (y) = 8y 3 − 12y. Reading.Patashnik (AddisonWesley. As will become evident in the present case generating functions provide an extremely elegant access to the special functions of Mathematical Physics3 .E. closed expressions for Hn (y).200) One can deduce from this expression the polynomial character of Hn (y).199) Comparision with (2.S. ∂y n (2.196) to ˜ derive.1: Schematic representation of change of summation variables ν and µ to n = ν + µ and m = ν −µ..

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. . µ expressend in terms of n.207) This expression yields for the ﬁrst four Hermite polynomials H0 (y) = 1.209). m = ν − µ 0 ≤ n < ∞. m are ν = n+m . .203) The old summation variables ν. H1 (y) = 2y. With this restriction in mind one can express (2.209) This property follows also from the generating function. 0 ≤ m ≤ n . for odd n. hence. µ are integers the summation over n. H3 (y) = 8y 3 − 12y.204) Since ν. i. H2 (y) = 4y 2 − 2. the sum in (2. it contains only odd powers. Hence. According to (2. is a polynomial of degree n. 2 (2. m must be restricted such that either both n and m are even or both n and m are odd. . (2. .194) holds w(−y.208) which agrees with the expressions in (2. in fact. In case of even n .202) e 2yz − z 2 ∞ = n=0 zn n! ≤n m≥0 n! (−1) n−m 2 n−m 2 ! m! (2y)m . The lattices representing the summation terms are shown in Fig. (2.207) contains only even powers. −z) and. p ≤ x. k! (n − 2k)! (2.1. according to 2.. 2 µ = n−m .207) one can deduce that Hn (y).5: Propagator for a Quadratic Lagrangian and introduce now new summation variables n = ν + µ.e. 2. otherwise.206) From this expansion we can identify Hn (y) [n/2] Hn (y) = k=0 (−1)k n! (2y)n−2k . k! (n − 2k)! = Hn (y) (2. it holds Hn (−y) = (−1)n Hn (y) . z) = w(y.2.205) Since (n − m)/2 is an integer we can introduce now the summation variable k = (n − m)/2 .210) from which one can conclude the property (2.201). 39 (2. 0 ≤ k ≤ [n/2] where [x] denotes the largest integer p.197) ∞ n=0 zn Hn (−y) = n! ∞ n=0 (−z)n Hn (y) = n! ∞ n=0 zn (−1)n Hn (y) n! (2. From (2.

. H2n+1 (0) = 0 . z) ∂ w(y. 2.216) 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. Hn+1 (y) − 2y Hn (y) + 2n Hn−1 (y) = 0. . (2. 2. 2. z) = 0 (2.196) into the diﬀerential equation (2. indeed satisﬁes the latter relationship. . . have a node at y = 0. namely. For this purpose one considers w(0. for example.215) yields ∞ n=1 z n−1 Hn (y) − 2y (n − 1)! ∞ n=0 zn Hn (y) + 2 n! ∞ n=0 z n+1 Hn (y) = 0 . z) = exp(−z 2 ) and carries out the Taylor expansion on both sides of this expression resulting in ∞ ∞ (−1)m z 2m zn = Hn (0) . Hn+1 (y) − 2y Hn (y) + 2n Hn−1 (y) = 0 .188. .215) ∂z which can be readily veriﬁed using (2. 1. a factor z reduces the order of Hn by one and introduces also a factor n. .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. H2n (0) = (−1)n Recursion Relationships A useful set of properties for special functions are the so-called recursion relationships. For Hermite polynomials holds.194). 2.208). as deﬁned through (2. . (2. dy n = 1. as given in (2. Substituting expansion (2.213) which allow one to evaluate Hn (y) from H0 (y) and H1 (y) given by (2. One can then readily state which diﬀerential equation of w(y.197) above and given by (2. .211) m! n! m=0 n=0 Comparing terms on both sides of the equation yields (2n)! .233) below. The reader may verify that w(y. . n! (2. n = 0.218) The reader should recognize the connection between the pattern of the diﬀerential equation (??) and the pattern of the recursion equation (??): a diﬀerential operator d/dz increases the order n of Hn by one. . .40 Quantum Mechanical Path Integral The generating function allows one to determine the values of Hn (y) at y = 0. a property which is consistent with (2. z) should be equivalent to the relationship (??). . 2. z). Another relationship is d Hn (y) = 2n Hn−1 (y). z) − (2y − 2z) w(y.194). (2. n = 1. dw/dy − 2zw = 0.217) gives H1 (y) − 2y H0 (y) = 0. . (2.209) since odd functions have a node at the origin. Starting point of the derivation is the property of w(y. (2.

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 .221) 1 = √ π +∞ −∞ dt e2iyt − t .194. 2 (2. in the integral representation of the Hermite polynomials 2n (−i)n ey √ Hn (y) = π Orthonormality Properties 2 (2.200) yields Hn (y) = The identity (−1)n y2 √ e π +∞ dt −∞ dn 2iyt − t2 e . e−y 2 π e−y 2 (2. of the Rodrigues formula (2.226a). dy n (2.227) Comparing the terms of the expansions allows one to conclude the orthonormality conditions +∞ −∞ dy Hn (y) Hn (y) e−y 2 √ = 2n n! π δn n .219) which can be written I(y) = e−y 2 +∞ −∞ dte−(t−iy) = e−y √ 2 2 +∞ −∞ dze−z . (2. 2 (2.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. . through a double series over Hermite polynomials using (2.s. 2 (2.s.228) . 2.222) Employing this expression now on the r.223) dn 2iy t − t2 2 e = (2 i t)n e2iy t − t n dy results.194. (2. ﬁnally. acording to the deﬁnition (2.247) for a = i one obtains I(y) = and. 2 √ π e2 z z (2.h. 2. .220) Using (2.2. 2.225) We want to derive from the generating function (2. For this purpose we consider the integral +∞ −∞ dy w(y. z) w(y.h.196) yields ∞ n. 2 n = 0.224) +∞ −∞ dt tn e2iy t − t .226) π −∞ 2n z n z n n=0 n! Expressing the l. 1. z ) e−y 2 +∞ = e2 z z = √ ∞ dy e−(y−z−z ) = .196) the orthogonality properties of the Hermite polynomials. .

f. 2. 3. 3.231) √ π (2. Normalized Stationary States The orthonormality conditions (2. (2.232) Hn (y) .233) are presented for n = 0. 2. One can recognize. 1. 3. Furthermore.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. 4. the value of the wave function at y = 0 is positive for n = 0. 4 in Fig. ¡ ¢ ¡ -4 -2 0 2 4 y (2. (2. 2. 2.230) (2. in harmony with (??).228) allow us to construct normalized stationary states of the harmonic oscillator. that the wave functions are even for n = 0. 4 and odd for n = 1.229) Hn (y) . 1. (2.190)] φn (y) = Nn e−y 2 /2 and for the normalization constants Nn follows from (2.42 Quantum Mechanical Path Integral E 9/2 h ω ¥ ¥ ¥ ¥ ¥ 7/2 h ω 5/2 h ω ¤ 3/2 h ω £ 1/2 h ω ¦ Figure 2. 2. negative for n = 2 and vanishes for n = 1. in agreement with our above discussions.2.189. 3.233) . One can also recognize that n is equal to the number of nodes of the wave function. According to (2. 4.197) holds 2 ˜ φn (y) = e−y /2 Hn (y) . The normalized states are [c.2: Stationary states φn (y) of the harmonic oscillator for n = 0.

.234) The explicit form of the stationary states of the harmonic oscillator in terms of the position variable x is then. In case of the associated Laguerre (α) polynomials. 5.. Ω (2.e.f.150 . 1].151. w(x) = √ exp(−x2 ). 5.236 . ??? and eq.236) where w(x) is a so-called weight function with the property w(x) ≥ 0. (10. g ∈ F .459] hydrogen atom.239) . Other examples of orthogonal polynomials are the Legendre and Jacobi polynomials which arise in solving three-dimensional stationary Schr¨dinger equations. 5.189) by the Jacobian dx/dy.156. The latter are written for polynomials pn (x) in the general form dx pn (x) pm (x) w(x) = In δnm Ω (2. using (2.238) where the space is endowed with the scalar product (f |g) = Ω dx f (x) g(x) w(x) = < ∞ . holds Ω = [0.235) Completeness of the Hermite Polynomials The Hermite polynomials are the ﬁrst members of a large class of special functions which one encounters in the course of describing stationary quantum states for various potentials and in spaces of diﬀerent dimensions. In case of the Legendre polynomials. 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. ???] and relativistic [see Sect. 2. (2.10 and eq. 10.179] holds Ω = [−1.228) shows that the orthonogality condition of the Hermite polynomials is in complience with (2. of functions which obey dx f 2 (x) w(x) = < ∞ . and I = 2/(2 + 1). i.2. w(x) = xα e−x . denoted by Ln (x) and encountered in case of the stationary states of the nonrelativistic [see Sect.e. f.237) for Ω = R. In = Γ(n+α+1)/n! where Γ(z) is the so-called Gamma function. 5 below [c. w(x) = 0 only at a discrete set of points xk ∈ Ω (2. denoted by P (x) and introduced in Sect. Comparision with (2. The orthogonal polynomials pn mentioned above have the important property that they form a complete basis in the space F of normalizable functions. (5. (2. i.189) φn (x) = √ 1 2n n! mω π 1 4 e− mωx2 2 Hn ( mω x) .5: Propagator for a Quadratic Lagrangian 43 The normalization condition (2.233) and (2. The Hermite polynomials are so-called orthonogal polynomials since they obey the conditions (2. w(x) ≡ 1. (2.237) and where In denotes some constants. and In = 2n n! π. by dx = dy mω 1 4 .231) of the wave functions diﬀers from that postulated in (2.228). +∞[. 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.

233) exhaust all stationary states of the harmonic oscillator. 2 (2. pn (x) → w(x) pn (x) the scalar product (2.m=0 d∗ dm exp[iω(m − n)t] e−y Hn (y)Hm (y) .241) The latter identity follows from (2.236). ∞ n. the 2 stationary state ψ(y.242) Let us assume now the case of a function space governed by the norm (2. Therefore.e.e.233). Appendix: Exponential Integral We want to prove +∞ +∞ I = dy1 .. i. t) is ∞ ψ(y. If one replaces for all f ∈ F: f (x) → w(x) f (x) and.e.239) becomes the conventional scalar product of quantum mechanics f |g = Ω dx f (x) g(x) . . aT = a. in particular. t) = n=0 cn (t) e−y 2 /2 Hn (y) .44 As a result holds for any f ∈ F f (x) = n Quantum Mechanical Path Integral cn pn (x) (2.245) must be time-independent.244) For the state to be stationary |ψ(x. i. det(a) (2. The only possibility for this to be true is dn = 0. t) = n=0 1 2 dn exp[−iω(n + )t] e−y /2 Hn (y) .247) .243) To be consistent with(2.. ψ(y. n 2 (2. Since the Hermite polynomials form a complete basis for such states we can expand ∞ ψ(y.172) holds.147).e. except for a single n = no .242) and the existence of a normalizable state ψ(y. i. . t) must be identical to one of the stationary states (2.. a (2. hence. i. (2. a state for which (2. symmetric a. In case of n = 1 this reads +∞ −∞ dx ei a x = 2 iπ . Ω (2.k yj ajk yk = (iπ)n . t)|2 . the states (2.246) for det(a) = 0 and real. (2.240) where cn = 1 In dx w(x) f (x) pn (x) . 2. −∞ dyn ei −∞ n j. t) which is stationary under the action of the harmonic oscillator propagator (2.197) it must hold cn (t) = dn exp[−iω(n + 1 )t] and.188.

249) The akk are the eigenvalues of a and are real. This property allows one to simplify the bilinear ˜ form n yj ajk yk by introducing new integration variables j. yk = k Skj yj . (2. The proof of (2. .k yj ajk yk n = j.m (S T )jl alm Smk . One can conclude det(a) = j=1 n (2. .248. i.7: Appendix / Exponential Integral 45 which holds for a ∈ C as long as a = 0. 0 ˜ 0 a22 ..249) n ajk = ˜ l. ˜ (2.k n n yj = ˜ k (S −1 )jk yk .k n m = = j. . .252) ˜ For the determinant of a holds det(˜) = a n ajj ˜ j=1 (2.251) where. .. according to (2.2. . 0 ˜ ˜ S−1 a S = a = . ST S = 1 1 or S = S−1 . .246) reads then in terms of yj ˜ n n j.254) ajj . . . 2.e. . . .253) as well as det(˜) = det(S−1 aS) = det(S−1 ) det(a) . . symmetric matrix exists a similarity transformation such that a11 0 . .246) exploits that for any real.248) . det(S) a = (det(S))−1 det(a) det(S) = det(a) . .k n m n y Sj ajk Skm ym ˜ ˜ y (S T ) j ajk Skm ym ˜ ˜ j.255) . . (2. (2. ˜ (2.k yj ajk yk ˜˜ ˜ (2. 0 0 . ann ˜ where S can be chosen as an orthonormal transformation.250) The bilinear form in (2.

.250) allows one to express (2. . yn ) y ˜ ∂yj . . . . y −∞ d˜n det y −∞ ∂(y1 . . yn ) ∂(˜1 . n (2. .266) . ∂(˜1 . . .e. 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 −∞ . ∂ ys ˜ (2. Accordingly. holds n Quantum Mechanical Path Integral ajj = 0 ˜ j=1 (2.262) (2. i. .249) follows 1 = det ST S such that one can conclude det S = ±1 One can right then (2. ˜ for j = 1. . . d˜n e y −∞ = (2.256) such that none of the eigenvalues of a vanishes. .258) +∞ +∞ ∂(y1 . .46 We have assumed det(a) = 0.265) k=1 −∞ which leaves us to determine integrals of the type +∞ dx eicx −∞ 2 (2. . . yn ) y ˜ ei n k akk yk ˜ ˜2 . . . . 2. .. . . .250) +∞ +∞ I = d˜1 . yn ) y ˜ = ( det S )2 (2. ajj = 0 .260) (2.250) holds J = S and. det( From (2.263) (2.261) ∂(y1 . hence. . . . yn ) ∂(˜1 .258) where we introduced the Jacobian matrix J = with elements Jjs = According to (2. .264) I = d˜1 .. . yn ) ) = det(S) . . .257) Substitution of the integration variables (2.259) (2.. (2. . .

268) along the path γ = γ1 + γ2 + γ3 + γ4 displayed in Figure 2. the integrand does not exhibit any singularities anywhere in C. i.e. 2 since eicz is a holomorphic function.2.268) vanishes. (2. p ∈ IR. We consider ﬁrst the case c > 0 and discuss the case c < 0 further below.7: Appendix / Exponential Integral 47 where.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 √ (2.3.c > 0 .257) holds c = 0. according to (2.268) can be written as the sum of the following path integrals J = J1 + J2 + J3 + J4 .. The contour intergral (2.271) .267) by considering the contour integral J = γ dz eicz = 0 2 (2.266) to the well-known Gaussian integral +∞ dx e−cx −∞ 2 = π . for x.270) dx e−cx 2 2 − 2p √ γ4 : z = ix − p J4 = −p i dx eic(ix−p) . The contour integral (2. Jk = γk dz eicz 2 (2. One can relate integral (2. Substituting −x for x into integral J4 one obtains 0 J4 = p p (−i) dx eic(−ix+p) 2 = 0 i dx eic(ix−p) 2 = J2 . c (2.

(2.267) one has shown then ∞ dx eicx −∞ 2 = iπ . We will now show that the two integrals J2 and J4 vanish for p → +∞. (2.3: Contour path γ in the complex plain. c ¡ -p γ1 p Re(z) 2 | |e−2cxp | . p p→+∞ lim |J2 or 4 | ≤ = p→+∞ 0 lim dx |e−2cxp | 1 − e−2cp 2cp = 0. One can state accordingly ∞ J = dx e −∞ icx2 − √ ∞ i dx e−cx −∞ 2 Using 2.275) . (2.272) = 0 .48 Quantum Mechanical Path Integral Im(z) ip γ2 ¡ γ4 ¡ γ3 ¡ -i p Figure 2. This follows from the following calculation p p→+∞ lim |J2 or 4 | = p→+∞ lim | 0 p i dx eic(ix+p) | ≤ It holds |e ic(p2 −x2 ) p→+∞ 0 lim |i| dx |eic(p 2 −x2 ) | = 1 since the exponent of e is purely imaginary.274) (2. Hence.273) p→+∞ lim J2 and J4 do not contribute then to integral (2.268) for p = +∞.

2. (iπ)n det(a) (2.278) Noting (2.275.277) to (2.277) We apply the above results (2. n ˜ j=1 ajj (2.279) . It holds n I = k=1 iπ = akk ˜ (iπ)n .7: Appendix / Exponential Integral 49 One can derive the same result for c < 0.268). 2.276) and (c < 0) ∞ dx eicx −∞ 2 = −iπ = −|c| iπ . This leads to ∞ icx2 J = dx e −∞ + √ −∞ −i ∞ dx ecx = 0 2 (2.265). c (2. if one chooses the same contour integral as (2. but with a path γ that is reﬂected at the real axis.255) this result can be expressed in terms of the matrix a I = which concludes our proof.

50 Quantum Mechanical Path Integral .

t + ) = Ω d3 r0 φ(r.h.5) ψ(r.h. but by no means all.s. t) . x3 )T . t + ) = Ω m 2πi 3 2 exp i m (r − r0 )2 − 2 U (r. t + ) = × exp i 2 +∞ +∞ +∞ −∞ dx1 −∞ dx2 −∞ dx3 m x2 + x2 + x2 1 2 3 m 2πi 3 2 × (3.20) to evaluate the propagator. This yields ψ(r. (3. of this equation in terms of powers of and we will demonstrate that the terms of order require that ψ(r. Notable exceptions are non-stationary systems involving time-dependent linear and quadratic Lagrangians. t) .4) − U (r. x2 . Generalizing (2. t) . The propagator in (3.20) to R one obtains then for small φ(r.1) can be expressed through the discretization scheme (2.2) d 3 r0 m 2πi 3 2 exp i m (r − r0 )2 − 2 U (r.21). We will denote the components of s by (x1 . t) ψ(r0 . t) by an inﬁnitesimal time step . namely the Schr¨dinger equation. t + |r0 . x3 51 . . t) = From this follows ψ(r. (3. and the r.1 Derivation of the Schr¨dinger Equation o We will consider now the propagation of a wave function ψ(r. t + |r0 . even in x1 .3) In order to carry out the integration we set r0 = r + s and use s as the new integration variable. t) ψ(r + s.s. 2. In the limit of very small it is suﬃcient to employ a single discretization interval in (2.Chapter 3 The Schr¨dinger Equation o 3. the Schr¨dinger o o equation provides the simpler avenue towards describing quantum systems than the path ingral formulation of Section 2.1) We will expand the l.20. (3. For many situations. It holds then according to (2. t) . x2 . t) satisﬁes a partial diﬀerential equation. t) ψ(r0 .

O( 7 2 ) .10) It is now important to note that in case of integral (3.. (3.. e. x3 . Obviously. of the expansion of ψ(r + s.5) which are even separately in all three coordinates yield non-vanishing contributions. the terms collected in (3. 1 4 3 2 2 ∂4 j. a I2 (a) = − 3 4a2 iπ ..6) make contributions of the order O( 3 2 ) . O( 5 2 ) . Since the latter contributes to (3. we need then to evaluate integrals of the type +∞ In (a) = −∞ dx x2n exp i a x2 . The integration involves only the wave function ψ(r +s.g. t) is a constant with respect to this integration. t). t) + . O( 7 2 ). .4) one identiﬁes 2 1 = a m = O( ) (3. a (3. .k=1 ∂2 ψ(r.11) and. i ∂a Starting from (3. t) and the kinetic energy term.13) 2πi a .36) holds I0 (a) = Inspection of (3. 1 2 3 2 ∂2 j=1 xj ∂x2 ψ(r. (3.9) iπ a (3. consequently. x2 .6) 3 4 ∂4 j=1 xj ∂x4 ψ(r.11) holds 3 m 3 iπ 2 2 × = 1 (3. only terms of the expansion (3. t) . but over s = (x1 . t) j 1 12 . ∂xj ∂xk (3. t) j k .7) According to (2.5) assuming that only the leading terms contribute. 1.4) only for small x2 + x2 + x2 values 1 2 3 we expand 3 ψ(r + s. x2 . U (r.4) is even in all three coordinates x1 .7) shows 1 ∂ In (a). It holds In+1 (a) = I1 (a) = i 2a iπ . According to (3. 2 (3. t) + ∂xj 2 3 xj xk j. t) j .. n = 0. x3 )T .. Since the kinetic energy contribution in (3.12) Here one needs to note that we are actually dealing with a three-fold integral.. t) + j=1 xj ∂ 1 ψ(r. a supposition which will be examined below.8) one can evaluate recursively all integrals In (a).52 Schr¨dinger Equation o It is important to note that the integration is not over r.k=1 xj xk ∂x2 ∂x2 ψ(r.8) (3. t) = ψ(r. It is then suﬃcient to consider the terms ψ(r.

This equation is often written in the o form ∂ ˆ i ψ(r. t) + O( 2 ) . t) + O( 2 ) . t) − + i 2 2 i U (r.e. t) = 1 − (3.15) ψ(r. t) . Due to its linear character any linear combination of solutions of the time-dependent Schr¨dinger o equation is also a solution. using (3.2 Boundary Conditions The time-dependent Schr¨dinger equation is a partial diﬀerential equation.2: Boundary Conditions and one can conclude.17) 2m ψ(r.14) This expansion in terms of powers of suggests that we also expand ∂ ψ(r. We will derive brieﬂy the type of boundary conditions encountered. 2nd o order in the spatial variables and linear in the solution ψ(r. i. The following general remarks can be made about the solution.3. ψ(r. (3. t) + ∂ ∂t ψ(r.14) results in ψ(r. Obviously. one speciﬁes the so-called initial condition. 1st order in time. The 1st order time derivative requires that for any solution a single temporal condition needs to be speciﬁed. (3. t + ) = exp − i U (r.g. t) = H ψ(r. this equation is trivially satisﬁed to order O( 0 ). Usually. t) + O( 2 ) . t) + and exp − i U (r. a solution is thought for t ≥ t0 and the solution is speciﬁed at the intial time t0 .19) ∂t where 2 2 ˆ H = − + U (r.16) Inserting this into (3. t) ψ(r.18) This is the celebrated time-dependent Schr¨dinger equation. t) (3.. In order O( ) the equation reads i ∂ ψ(r. The 2nd order spatial derivatives require that one speciﬁes also properties of the solution on a closed boundary ∂Ω surrounding the volume Ω in which a solution is to be determined.20) 2m 3. t) . t) + O( 2 ) ∂t i U (r. (3.. t) ψ(r. t) = ∂t 2 − 2 2m + U (r. t). (3. t)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) (3. ψ(r. t) + 1 2i 4 m 2 53 ψ(r. e. t1 ) = f (r). t) = ψ(r. t + ) = ψ(r.10).

t).22) ˆ where H is deﬁned in (3. Calssical Electrodynamics. g|r) (3. The hermitian property. Chapter 1. t) g(r) − Ω d3 rg(r) U (r.. In (3. .23). g ∈ H is deﬁned as follows f |g Ω = d3 rf ∗ (r)g(r) (3. 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.27) ˆ We will postulate below that H is an operator in H which represents energy. Jackson. Interchanging f ∗ (r) and g(r) results in g| H |f Ω = Ω ˆ d3 r g(r) Hf ∗ (r) . that H is 2 . g|r) is called the ˆ concomitant of H and is P (f ∗ . hence.20). implies hermitian ˆ f | H |g 1 2 Ω ˆ = g| H |f Ω (3. The diﬀerence between the integrals is g| H |f = Ω Ω = − g| H |f d3 rf ∗ (r) − Ω 2 2 2m g(r) − Ω d3 rg(r) 2 − 2 2m f ∗ (r) + Ω 2 d3 rf ∗ (r) U (r. however. (John Wiley. g|r) = − 2 2m ( f ∗ (r) g(r) − g(r) f ∗ (r) ) (3. diﬀer from each other. 1975).23) ˆ Since H is a diﬀerential operator the expressions (3. D. for example.54 Schr¨dinger Equation o linear vector spaces of functions f (r) in which a scalar product between two elements f. Since energy is a real ˆ ˆ quantity one needs to require that the eigenvalues of the operator H are real and. J.26) the vector–valued function P (f ∗ . 2nd Ed. (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) ) (3.28) See. New York.25) ˆ f | H |g Ω ˆ = g| H |f Ω + ∂Ω da · P (f ∗ . The reader is advised to consult a reference text on ‘Linear Algebra’ to follow this argument.22) and (3. f ∗ (r) d3 rg(r) Ω 2 = − 2m d3 rf ∗ (r) Ω 2 g(r) − f ∗ (r) (3. in principle.21) Ω This leads one to consider the integral f | H |g Ω = Ω ˆ d3 rf ∗ (r) H g(r) (3.26) where ∂Ω da · A(r) denotes an integral over the surface ∂Ω of the volume Ω.

if f and g satisfy these conditions than also does any linear combination αf + βg. t) which describes the state of o a particle moving in the potential U (r.30) Note that these boundary conditions are linear in f . i. t) = ω d3 r |ψ(r. .29) In fact. In this case one can postulate both conditions (3. However. α > 2. t)∂t ψ(r. e.3: Particle Flux 55 and. 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.32) the wave function must decay for |r| → ∞ rapidly enough to be square integrable. like exp(−κr). t)|2 . (3. we can only allow functions which make the diﬀerential da · P (f ∗ . 3. i. in this case also all derivatives of f (r) vanish at inﬁnity.. |ψ(r. and that the density decays rapidly when one moves away from that area. . t)∂t ψ ∗ (r.31) usually arises when a particle existing in a bound state is described. t) ] . (3. κ > 0 or like r−α .32). (3. The latter property stems from the fact that the boundary condition (3. r ∈ ∂Ω (3. 3.33) The time derivative of p(ω.. i. 3.e.30) can be postulated.34) An example is a volume between two concentric spheres.g. A most common boundary condition is encountered for the volume Ω = R3 in which case one postulates lim f (r) = 0 “natural boundary condition” .29. g|r) vanish on ∂Ω. t) is ∂t p(ω.e. The observable directly linked to the wave function is the probability to ﬁnd the particle at position r at time t. t) + ψ(r. t) on a single connected surface ∂Ωj only either one of the conditions (3.30) each condition holding on an entire surface ∂Ωj . t) = ω 3 d3 r [ ψ ∗ (r. r ∈ ∂Ω or da · f (r) = 0 ∀r. The probability to observe the particle anywhere in the subvolume ω ⊂ Ω is p(ω. namely. Since the total probability of ﬁnding the particle anywhere in space is d3 r|f (r)|2 = 1 (3. In either case does f (r) and all of its derivatives vanish asymptotically.3. to avoid discontinuities in ψ(r. therefore. Often the closed surface of a volume ∂Ω is the union of disconnected surfaces3 ∂Ωj . ∂Ω = ∂Ω1 ∩ ∂Ω2 ∩ ∂Ω3 ∩ ..31) |r|→∞ (3. t). obey (3. It must hold then for all f ∈ H f (r) = 0 ∀r.3 Particle Flux and Schr¨dinger Equation o The solution of the Schr¨dinger equation is the wave function ψ(r.. t)|2 . . in which case ∂Ω1 is the inner sphere and ∂Ω2 is the outer sphere.e.29.

36) According to (3. t) = |ψ(r. t) H ψ ∗ (r. t) ψ ∗ (r. t) + 4 · j(r. t) − ψ ∗ (r. (3. t) . t) = 0.38) holds. t) (3.43) (3. t) − ψ(r. t) such that d3 r|ψ(r. Using (3. The surface integral (3. t) = ∂ω da · P (ψ ∗ (r. ψ(r.18) by any complex number and accordingly one can deﬁne ψ(r. One can multiply the solution of (3. Accordingly the probability to observe the particle anywhere in the total volume Ω is constant. t) + ω · j(r. 3. 3.19) and its conjugate complex4 −i yields i ∂t p(ω. . t) ∂t ˆ ˆ ψ ∗ (r. t) H ψ(r. A natural choice for this constant is 1. t) ] . (3. t) = P (ψ ∗ .41) holds for any volume ω ⊂ Ω one can conclude ∂t ρ(r.37) d3 r ∂t ρ(r.35) d3 r . t) ψ(r.26. t)|2 is a probability density with units 1/volume. One refers to such solution as normalized.56 Using (3.27) this can be written i ∂t p(ω. We will assume in the remainder of this Section that the solutions discussed are normalized. t)r. t) = H ψ ∗ (r. ψ|r. t) = ω Schr¨dinger Equation o ∂ ∗ ˆ ψ (r. Note that for a normalized wave function the quantity ρ(r.41) where j(r.29.39) d3 r · A(r) (3.44) ˆ Note that the Hamiltonian H involves only real terms. t)|2 = 1 Ω (3.40) One can rewrite then (3.42) 2mi Since (3. (3.37) If one applies this identity to ω = Ω one obtains according to (3. t) . t) = 0 . t) = 0 (3.37) can be expressed through a volume integral according to da · A(r) = ∂ω ω (3.27) one can express this j(r. t). (3.30) ∂t p(Ω. t) = [ ψ(r.

t) is that of density ﬂux. (3. t) = [2π]− 2 where the dispersion relationship holds ω = k2 .48) ∂t 2m which describes the motion of free particles. (2. Note that j(r. (2. t) . This ﬁnding is consistent with the present evaluation of the particle ﬂux: a factor exp(ipo xo / ) gives rise to a ﬂux po /m.148.18) in Ω∞ = R in the case o U (r. t) points in the direction to the outside of volume ω.3.f. 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. for f (r) ∈ R. 2. The generalization to three dimenisons implies then that the factor exp(i k · r) corresponds to an intial velocit k/m and a ﬂux of the same magnitude. the initial condition at ψ(r.50) 3 ˜ d3 k φ(k) exp i(k · r − ωt) Ω∞ (3. arises solely from the complex factor exp(i k · r).125)] and in a quadratic [c. (3.71)].45) Obviously. t) has vanishing ﬂux anywhere. One often encounters wave functions of the type φ(r) = f (r) ei k·r The corresponding ﬂux is j(r) = k 2 f (r) .49) reads at t = t0 ψ(r. t0 ) = [2π]− 2 Ω∞ 3 ˜ d3 k φ(k) exp i(k · r − ωt0 ) .51) .105.43) that any real wave function ψ(r. j(r. It is of interest to note from (3.48) that the general solution is of the form ψ(r.4 Solution of the Free Particle Schr¨dinger Equation o We want to consider now solutions of the Schr¨dinger equation (3.. In Sect. Such case arose in Sect..41) written in the form ∂t ω d3 r ρ(r. t) = 0 2 ∂ 2 i ψ(r.e.167)] potential.f.46) i. equal to the velocity of the particle. and for particles moving in a linear [c. One can readily show by insertion into (3. This follows directly from an inspection of Eq. 2 for a free particle [c. 2. t) = − ∂ω da · j(r. (2.47) . t) gives rise to a descrease of the total probability in volume ω due to the disappearence of probability density at the surface ∂ω.48.49) ˜ Obviously.f. (3. (3. 3. 2.4: Free Particle 57 The interpretation of j(r. t0 ) determines φ(k). 2m (3. m (3. t) (3. Equation (3. i. t) = − ψ(r.e.

54) an alternative representation of the propagator (2.52) above ψ(r.247). i.54) This expression obviously has the same form as postulated in the path integral formulation of Quantum Mechanics introduced above. The ensuing solutions are called wave packets.49). that the “natural boundary condition” (3. To show this one needs to note 1 2π = +∞ dk1 exp −∞ ik1 (x − x0 ) − 1 2 i 2 k2 1 2m .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. t0 ) ψ(r0 . In fact. i.51) to be square integrable it follows according to the properties of the Fourier–transform that ψ(r.. Comparision with Path Integral Formulation One can write solution (3. (3.70) derived below for a particle in a box and the harmonic oscillator.31) applies.49) is square integrable at all subsequent times and.49.e.54) to the case of non-vanishing potentials U (r).54) valid for this case.81) . The general form for this propagator involves an expansion in terms of a complete set of eigenfunctions as in (3. t|r0 .56) and using (2. We have identiﬁed then with (3. 3. Below we will generalize the propagator (3. respectively. If one chooses the initial state f (r) deﬁned in (3.. t|r0 . t0 ) = 1 2π 3 d3 k exp Ω∞ ik · (r − r0 ) − i 2 k2 2m (t − t0 ) . derive an expression similar to (3.52) We have not speciﬁed the spatial boundary condition in case of (3.47).47). t0 ) (3. in (2.54) yields (2.114) and (4. The solution as stated is deﬁned in the inﬁnte space Ω∞ = R . Ω∞ (3. In case of the harmonoc oscillator the expansion can be stated in a closed form given in (4.e.58 The inverse Fourier transform yields 3 ˜ φ(k) = [2π]− 2 Schr¨dinger Equation o d3 r0 exp(−ik · r0 ) ψ(r0 .5).53) where φ(r. t) = Ω∞ d3 r0 φ(r. t) as given by (3. t0 ) . evaluating the integral in (3. hence. (t − t0 ) (3.51. 3.

53.52) to the case that the initial state of a 1-dimensional free particle ψ(x0 .60) yields with t0 = 0 ψ(x.55.57–3.49. (3.64) (3.62) We will restrict the space of allowed solutions to a volume Ω such that the functions also make the concomitant (3.56)].57) where φ(x.19. t0 ) ψ(x0 . 3.29. Accordingly.3.e. (3. t) = f (t) φ(r) .4: Free Particle Free Particle at Rest 59 We want to apply solution (3.20) which are of o the form ψ(r. t|x0 . 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. (3. 3.60) =???? Combining (3. We have demonstrated then in this case that the Schr¨dinger formulation of Quantum o Mechanics is equivalent to the Feynman path integral formulation. Stationary States We consider now solutions of the time-dependent Schr¨dinger equation (3. the functions obey boundary conditions of the type (3. The 1-dimensional version of (3.. the boundary conditions are φ(r) = 0 ∀r. (3.30). t) 2 k2 (3. 3. t) = t 1 − i mδ2 t 1 + i mδ2 1 4 1 πδ 2 (1 + 2 t2 m2 δ 4 1 4 ) × t po i p2 o ) + i x − t mδ 2 2m (3. 3.59) which is solved through completion of the square in the exponent [c.63) . t) is given by (??).46).27) vanish on the surface ∂Ω of Ω. i. r ∈ ∂Ω (3.58) 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. r ∈ ∂Ω or da · φ(r) = 0 ∀r. t0 ) = 1 2π +∞ dk exp −∞ ik (x − x0 ) − i 2m (t − t0 ) .54) is +∞ ψ(x. a result which is identical to the expression (??) obtained by means of the path integral propagator (2. t) = −∞ dx0 φ(x.f. t|x0 .61) × exp − (x − po m t)2 2δ 2 (1 + 2 t2 ) m2 δ 4 (1 − i .

71) . It is important to realize that the separable solutions (3. t) vanishes on the surface of ∂Ω. We will encounter many such problems in the subsequent Sections. . (3.63.60 Schr¨dinger Equation o and aﬀect only the spatial wave function φ(r).62) which consist of two factors. t)|2 = |φ(r)|2 . a common case is Ω = Ω∞ and the ‘natural boundary condition’ (3. however. according to (3. It follows then from (3. (3.e. We denote the corresponding solution by φE (r).62) is |ψ(r. for the eigenvalue problems in the conﬁned function space.19).27) holds and.62) do exist and we will characterize the two factors of the solution f (t) and φ(r). if f (t) = eiα . one factor depending only on the time variable and the other only on the space variables are called separable in space and time.45) that the total probability d3 r ρ(r. often only for a discrete set of values En .62) exists. We will demonstrate that solutions of the type (3. is time-independent.. i. We want to demonstrate this property now.68) ˆ where g1 (t) = i ∂t f (t). t)|2 = |f (t)|2 Ω d3 r |φ(r)|2 (3. One calls such states stationary states.62) have the particular property that the associated probability distributions are independent of time. . . It turns out that a solution for f (t) can be found for any E. n = 1. We may note in passing that solutions of the type (3. i.e.. (3. g2 (t) = f (t). We have then shown that ψ(r. This can hold only if |f (t)| is time-independent.62) we insert it into (3. h1 (r) = φ(r). the eigenvalues of the operator H..69) If these two equations can be solved simultaneously a solution of the type (3.31).65) is constant.69) exist. t) = f (0) exp − i E φE (r) where ˆ H φE (r) = E φE (r) . namely f (t) = f (0) exp − i Et .66) i. The identity (3. (3. This yields an expression g1 (t) h1 (r) = g2 (t) h2 (r) (3. 2. As pointed out above. 3.e. by no means all solutions are of this type. t) = Ω Ω d3 r |ψ(r. the solutions o (3.42) the ﬂux j(r. It follows from the observation that for the solution space considered (3. and h2 (r) = H φ(r). We must postulate therefore ∂t f (t) ˆ H φ(r) = E f (t) = E φ(r) . in general.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.62) are special solutions of the timedependent Schr¨dinger equation. In fact.69) is called an eigenvalue problem.67) (3. One can conclude that the probability density for the state (3. α ∈ R . In order to further characterize the solution (3. for functions required to obey boundary conditions (3. At this point we will accept that solutions φ(r) of the type (3. At this point we state without proof that. solutions exist only for a set of discrete E ˆ values.70) The task of ﬁnding solutions φ(r) which solve (3.64). hence.

The resulting total energy values E are positive.72) According to (3.74) using exp i k · r = ik exp i k · r . = E. The ﬂux corresponding to (3. .74) The classical free particle with constant energy E > 0 moves without bounds in the space Ω∞ .76) according to (3. should be positive.66) E must be real.71) this yields E Ω d3 r φ∗ (r) φE (r) = E ∗ E Ω d3 r φE (r) φ∗ (r) .28) both represent the state φE (r). a property which is to be expected since the energy is purely kinetic energy which.19. Obviously.e. real.3.75) 2m One can ascertain this statement by inserting the expression for φk (r) into the eigenvalue problem posed in (3.77) Noting that k can be interpreted as the magnitude of the momentum of the particle the ﬂux is equal to the velocity of the particle v = k/m multiplied by |N |2 . 3. As a result we cannot postulate in the present case that wave functions are localized and normalizable. of course. We will show in Section 5 that E can be interprerted as the total energy of a stationary state. o According to (3. t) = |N |2 k . k ∈ R 2 k2 φk (r) = N exp i k · r . t) = exp − i 2 k2 2m t exp i k · r (3.76) has kinetic energy 2 k 2 /2m. particles scattered of a potential. − 2 2m φ(r) = E φ(r) . The corresponding stationary solution ψ(r. one can interpret then k as the magnitude of the momentum of the particle. (3. m (3. (3. We start our proof using the property (3. i. We want to prove now that the eigenvalues E which arise in the eigenvalue problem (3.20).4: Free Particle 61 is a solution of the time-dependent Schr¨dinger equation (3.74) is actually best labelled by an index k.43) is j(r. t) = exp − E t 2 φE (r) .g. ˆ d3 r φ∗ (r) H φE (r) = E Ω Ω ˆ d3 r φE (r) H φ∗ (r) . e. Stationary State of a Free Particle We consider now the stationary state of a free particle described by i ψ(r. We will wave this assumption as we always need to do later whenever we deal with unbound particles. The solution φE (r) corresponding to the eigenvalue problem posed by (3.71) are. E ∈ R.28) for the special case that f and g in (3. E (3.73) from which follows E = E ∗ and.. E (3. in fact. hence.

The boundary conditions which according to (3.81. We assume. In case of the even solutions (3. 5 . It turns out that this boundary condition can only be satisﬁed for a discrete set of k–values kn . a] ⊂ R → R. 2a n = 1. (3. We can expect. 3. that the solutions obey this symmetry as well. 3. we have to consider only one boundary condition. a}. Solutions of the Schr¨dinger Equation in F1 o The time–dependent solutions satisfy i ∂t ψ(x.81) 2m = E. 3. Hence. t) = exp(−iEt/ )φE (x) where φE (x) is determined by 2 d2 − φE (x) = E φE (x) .81.80) need to be satisﬁed are φE (e. t) . and so-called odd solutions obeying φ(x) = −φ(−x) φE (x) = A sinkx..81) they are kn = nπ .o) (−a) = 0 (3. (3.82) have the property that either both boundary conditions are satisﬁed or none. 2m dx2 2 (± ) = } (3.82) One can readily verify that (3. a] ⊂ R assuming that the potential outside of this interval is inﬁnite.82) satisfy the diﬀerential equation in (3.o) (a) = 0 and φE (e.83) The solutions (3.79) The stationary solutions have the form ψ(x.84) . i. We will refer to this as a particle in a one-dimensional ‘box’.78) (3. let say the one at x = a. two types of solutions.e.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. (3. f continuous. so-called even solutions obeying φ(x) = φ(−x) φE (x) = A coskx .62 Schr¨dinger Equation o 3. (o) 2 k2 (e) 2 k2 2m = E (3. hence. .80). .80) 2m dx2 We note that the box is symmetric with respect to the origin. therefore. n ∈ N. . 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. a] ⊂ R which vanish on the surface ∂Ω1 = {−a. φ(±a) = 0 . t) = − d2 ψ(x. f ∈ F1 = {f : [−a.

82). (3. 3 . 6 . = 0.e. . a n = 2. . The wave functions for the ﬁve lowest energies En are presented in Fig.) The Energy Spectrum and Stationary State Wave Functions The energy values corresponding to the kn –values in (3. . n = 1. (3.84. 2. φ(e) (a. 2a n = 1.87) (Note that. This condition implies for the even states +a |An |2 −a dx cos2 nπx = |An |2 a = 1 . respectively. i. according to (3. 5 . 4.89) nπx . .81.o) (a. . 3. It is desirable to normalize the wave functions such that +a −a dx |φ(e.3.92) and for the odd states +a |An |2 −a dx sin2 nπx = |An |2 a = 1 . . (3. (3. . . n is assumed to be even. 6 . 2a n = 1. 4.93) The normalization constants are then An = (3. are En = 2π2 8ma2 n2 . n = 2. x)|2 = 1 n (3. 2a n = 2. (3.5: Particle in One-Dimensional Box since for such kn 63 nπa nπ = cos 2a 2 In case of the odd solutions (3. (3.82) only the kn –values cos(kn a) = cos kn = nπ .81) and (3. 6 . 3. 5 .86). . .91) holds. 3. 4..1).82).86) satisfy the boundary condition since for such kn nπa sin(kn a) = sin 2a = sin nπ 2 = 0. 3. 2a 1 . x) = An sin n nπx . . (3. By counting the number of their nodes one can determine the energy ordering of the wave functions. .94) .88) where the energies for odd (even) n–values correspond to the even (odd) solutions given in (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 .86). . x) = An cos n and φ(o) (a.85) (3. (3.90) 2a The wave functions represent stationary states of the particle in a one-dimensional box. according to the dispersion relationships given in (3.

3.o) (a. ¡ 1 -a a x . 4. 2. 5 of particle in a box.1: Eigenvalues En and eigenfunctions φn (e. x) for n = 1.64 Schr¨dinger Equation o E h2 ] [ 8 m a2 25 ¢ 16 9 4 Figure 3.

In case of n.102) holds according to (3. the integrand is odd. the integral vanishes.103) We will show in Section 5 that the property of a scalar product do indeed apply.96) f |g Ω1 = −a dxf (x) g(x) . this integral vanishes. n = 1. n (3.3. Because of the property (3. x) . 3..98) ∞ f |f 5 Ω1 = n=1 d2 . hence.99) Since in this case the integrand is a product of an even and of an odd function. it holds: f |f Ω1 = 0 → f (x) ≡ 0.98) The latter property is obviously true for n = m.94) B1 = {φn (a. 2a 2a (3. 3.5: Particle in One-Dimensional Box The Stationary States form a Complete Orthonormal Basis of F1 We want to demonstrate now that the set of solutions (3. m even φn |φm 1 a Ω1 = 1 a +a nπx −a dx sin 2a sin mπx = 2a (3. m both even. (3. . 2. (i) n. (3.100) +a −a dx cos (n−m)πx + cos (n+m)πx 2a 2a The periods of the two cos-functions in the interval [−a. Similarly.e. for ∞ f (x) = n=1 dn φn (a. m both odd. 6 . g ∈ F1 (3. In case of n = m we have to consider three cases. 3. In fact. .101) +a −a dx cos (n−m)πx − cos (n+m)πx 2a 2a and.90. . it holds φn |φm Ω1 = δn m . too. i. 4. and (iii) the mixed case. m odd. Hence we need to consider only the ﬁrst two cases. (ii) n.95) 1 a cos nπx 2a sin nπx 2a for n = 1. for n = 2.. .} where φn (a. The latter case leads to integrals φn |φm Ω1 = 1 a +a dx cos −a nπx mπx sin . . 3. 5 . the integrals vanish.98) the elements of B1 must be linearly independent. n = m.e. .97) form an orthonormal basis set. In particular. . N ≥ 1.89. a] are N. . f. x) (3. Obviously. the integral arises φn |φm 1 a Ω1 = 1 a +a nπx −a dx cos 2a cos mπx = 2a (3. i. one obtains for n. x) = together with the scalar product5 +a 65 (3.

m = 1. t) = n=1 φn (a. .95) is also complete. (3. . 1.107) Inserting this into (3. For this purpose we expand ∞ ψ(x. t0 ) = dm .78) can be expressed as a linear combination of the elements of B1 deﬁned in (3. in (3.110) .95) is an orthonormal basis. x0 ) ψ(x0 . . x0 ) ψ(x0 .e. 3. . t0 ) = n=1 dn φn (a. The functions f are periodic with period 2a. Accordingly. 2. the coeﬃcients an . however. t0 ) = +a ∞ i n=1 − En φn (a.} and {bn .104) ˜ where [y]a = y mod2a. We like to show ﬁnally that the basis (3. x) . x) ∂t cn (t) −a dx0 φn (a. . n = 1.106) Using the orthonormality property (3.79) requiring the o initial condition cn (t0 ) = 1 .105) to the interval [−a. (3. there exist real constants {an . ( ) = ([ + ] − ) (3. . It follows that B1 n deﬁned in (3.108) into the Schr¨dinger equation yields o +a ∞ n=1 φn (a. x0 ) ψ(x0 . n = 2.. Restricting the expansion (3. any element of the function space F1 deﬁned in (3.108) where the functions cn (t) are to be determined from the Schr¨dinger equation (3. . such demonstration can be based on the theory of Fourier series. 3. . t) at times t > t0 . and bn . . (3. i. a] yields then also an expansion for any element in F1 and B1 is a complete basis for F1 . This implies that only the trigonometric functions which are elements of B1 enter into the Fourier ˜ series. Demonstration of completeness is a formidable task. t0 ) . x) cn (t) −a dx0 φn (a. x) to obtain an expression for ψ(x.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. 5 . x) cn (t) −a dx0 φn (a. n = 1. t0 ) (3. 4. .109) Insertion of (3. they can be expanded in terms of a Fourier series. 3. .96).106) and generalizing to t ≥ t0 one can write ∞ +a ψ(x.} such that ∞ ∞ nπx nπx ˜ = + an sin (3. n = 0. t0 ) in terms of eigenfunctions φn (a. .. i. x0 ) ψ(x0 . Hence.e. We have then shown that any f corresponding to elements of F1 can be expanded in terms of elements in B1 .95. . For this purpose we extend the deﬁnition of the elements of F1 to the whole real axis through ˜ f : R → R. −a (3.66 Schr¨dinger Equation o f (x) ≡ 0 implies f |f Ω1 = 0 which in turn implies dn = 0 since d2 ≥ 0. Evaluating the Propagator We can now use the expansion of any initial wave function ψ(x.105) must vanish. .98) one obtains +a dx0 φm (a. 2. In the present case.

113). σ = a 15 . t0 ) (3.3. The last frame shows the wave front reaching again the right wall and the onset of new interference. t|x0 . The particle moves then to the left.2 presents the probability distribution of the particle at subsequent times. Example of a Non-Stationary State As an illustration of a non-stationary state we consider a particle in an initial state ψ(x0 . t|x0 . but the collision with the left wall leads to new interference eﬀects.e.98).109) are cm (t) = exp i − Em (t − t0 ) . Figure 3. t0 ) itself is a solution of the time-dependent Schr¨dinger o equation (3. i ∂t cm (t) = − Em cm (t) . t0 ) ψ(x0 . t|x0 .5: Particle in One-Dimensional Box Multiplying both sides by φm (a. (3. x0 ) . t0 ). The interference pattern begins to ‘smear out’ ﬁrst. 67 (3. In the present system which is composed solely of bound states the propagator is given by a sum. t0 |x0 .79) which lies in the proper function space (3. t0 ) = δ(x − x0 ) as can be readily veriﬁed using (3.111) The solutions of these equations which satisfy (3.114) This expression has the same form as postulated in the path integral formulation of Quantum Mechanics introduced above.54) as in the case of the free particle system which does not exhibit any bound states. t0 ) = ∞ φn (a.112) determine now ψ(x. This solution can be written +a ψ(x. 4 a (3.115) This initial state corresponds to the particle being localized initially near x0 = 0 with a velocity v0 = 15 /m a in the direction of the positive x-axis. The respective initial condition is φ(x. in (2. the elements of which satisfy the appropriate boundary conditions. .114) the representation of the propagator for a particle in a box with inﬁnite walls.5).78). One can recognize that the particle moves ﬁrst to the left and that near the right wall of the box interference eﬀects develop.108. t) for any initial condition ψ(x. k0 = . t0 ) = 1 2πσ 2 1 4 exp − x2 0 + i k0 x0 2σ 2 . We have identiﬁed then with (3.112) Equations (3. Often the propagator φ(x. cm (t0 ) = 1 . being reﬂected at the right wall.113) where φ(x. t|x0 . according to (3. x) and integrating over [−a. 3.. t0 ) is also referred to as a Greens function. rather than by an integral (3. (3. x) exp n=1 i − En (t − t0 ) φn (a. a] yields. It is of interest to note that φ(x. Note that the propagator has been evaluated in terms of elements of a particular function space F1 . In case that diﬀerent boundary conditions hold the propagator will be diﬀerent as well. i. t) = −a dx0 φ(x.

24 m a h2 1 πσ k = 15 a 1 πσ -a a -a a |Ψ|2 2 t = 0.96 m a h2 |Ψ|2 2 t = 1. .00 m a h2 |Ψ|2 2 t = 0.44 m a h2 |Ψ|2 2 t = 1.72 m a h2 1 πσ 1 πσ -a a -a a |Ψ|2 2 t = 0. t)|2 for a particle in a box starting in a Gaussian distribution with momentum 15 /a.16 m a h2 1 πσ 1 πσ -a a -a a Figure 3.68 Schr¨dinger Equation o |Ψ|2 2 t = 0.92 m a h2 |Ψ|2 2 t = 2.48 m a h2 |Ψ|2 2 t = 0.2: Stroboscopic views of the probability distribution |ψ(x.68 m a h2 1 πσ 1 πσ -a a -a a |Ψ|2 2 t = 1.20 m a h2 1 πσ 1 πσ -a a -a a |Ψ|2 2 t = 1.

t) . x2 .116) where Ω is the interior of the box and ∂Ω its surface Ω ∂Ω = [−a1 .3. x3 )T ∈ Ω.78). hence. x3 )T ∈ Ω. a2 ] ⊗ [−a3 . The corresponding solutions have the form i ψ(x1 .e. i. x2 . x2 . (3. x2 . Placing the origin at the center and aligning the x1 . x3 ) .6 Particle in Three-Dimensional Box We consider now a particle moving in a three-dimensional rectangular box with side lengths 2a1 .116) and obeys the partial diﬀerential equation ˆ H φE (x1 .119) where φE (x1 . ..120) ˆ Since the Hamiltonian H is a sum of operators O(xj ) each dependent only on a single variable. We seek then solutions of the time-dependent Schr¨dinger equation o ˆ i ∂t ψ(x. )T ∈ ∂ } . x2 . t) = exp − E t φE (x1 . 2a2 . f continuous. j = 1. x3 . x3 ) = j=1 φ(j) (xj ) (3. ˆ H = − 2 (3. 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.97) with respect to which the eigenfunctions are orthonormal. The description employed a space of functions F1 deﬁned in (3. x2 . one can express H 3 φ(x1 .. a1 ] ⊗ [−a2 .95). φ(j) (±aj ) = 0 .118) which are stationary states. 3. x1 = ±a1 } ∪ {(x1 . This property allowed us to evaluate the propagator (3.117) 2m 2 2 2 ∂1 + ∂2 + ∂3 (3.e. 2a3 . a3 ] ⊂ R = {(x1 . 3 (3. ( . x2 . A complete basis of F1 is given by the inﬁnite set B1 (3. x2 = ±a2 } ∪ {(x1 . We have deﬁned in the space F1 a scalar product (3. can be counted. t) = H ψ(x1 . x2 . x3 ) = E φE (x1 . (3. x2 . x3 = ±a3 } . ˆ = O(x1 ) + O(x2 ) + O(x3 ).114) in terms of which the solutions for all initial conditions can be expressed. An important property of this basis and. x3 ) (3. . of the space F1 is that the elements of the basis set can be enumerated by integer numbers. x3 .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. x2 . i. x3 )T ∈ Ω. ) = ∀( . x2 .121) where − 2 2m 2 ∂j φ(j) (xj ) = Ej φ(j) (xj ) . x3 ) is an element of the function space F3 deﬁned in (3.122) . 2.

.96) and. 3. x3 ) .n2 . a3 are identical.n3 ) (a1 . n3 = 1.n2 . n2 . . .. x2 ) φn3 (a3 . Actually. 3. r0 ) . 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. r) exp i − E(n1 n2 n3 ) (t − t0 ) φ(n1 . (3. hence. This symmetry has been exploited in deriving the stationary states. t|r0 .n3 ) (a1 . 2. .n3 =1 φ(n1 . a2 . a3 .n3 ) (a1 . Comparing (3. . namely rotation by π around the x1 . a1 = a2 = a3 = a then rotation around the x1 . 3. a2 .122) are given by (3. 2. a3 . x3 –axes. i. . For example.} ∞ (3.126) Symmetries The three-dimensional box conﬁning a particle allows three symmetry operations that leave the box unchanged.123) 8m n2 n2 n2 1 2 3 + 2 + 2 a2 a2 a3 1 . x2 . n3 = 1. x3 ) = 1.n2 .e. t0 ) = n1 .n2 . n 2 . a2 . a3 .n3 ) (a1 . This additional symmetry is also reﬂected by degeneracies in the energy levels. x1 ) φn2 (a2 .124) The same considerations as in the one-dimensional case allow one to show that B3 = {φ(n1 . x2 .n2 . a2 .120) can be written φ(n1 . Such degeneracies are always a signature of an underlying symmetry. (3. in the present case .122) with (3. x1 . 2. If two or all three orthogonal sides of the box have the same length further symmetry operations leave the system unaltered. . n1 . x3 –axes by π/2 also leaves the system unaltered. if all three lengths a1 . . n 3 and E(n1 . x2 . n1 . x1 .n2 . the solutions of (3. a2 .70 Schr¨dinger Equation o and E1 + E2 + E3 = E. n2 .n3 ) = 2π2 = φn1 (a1 . a3 .125) is a complete orthonormal basis of F3 and that the propagator for the three-dimensional box is φ(r.80) shows that the solutions of (3. x2 . . (3. x3 ) n1 .

2) (a. a. However.2) (a. 3) (1.1) (a.2. a.2.6: Particle in Three-Dimensional Box 71 ‘accidental’ degeneracies also occur. however. to construct n orthogonal stationary states6 . in case of degenerate states one cannot necessarily expect that stationary states are orthogonal. Hence.. n2 . a. n3 ) = (3. r) + γ φ(2. The origin of this degeneracy is.g. One particular aspect of the degeneracies illustrated in the Table above is worth mentioning. . 6 This is a result of linear algebra which the reader may ﬁnd in a respective textbook. a. e. the identity 32 + 32 + 32 = 12 + 12 + 52 .2) (a. r) (3. a. due to the hermitian character of ˆ H. in case of an n–fold degeneracy it is always possible. a.1. We consider the degeneracy of the energy E122 which is due to the identity E122 = E212 = E221 . Any linear combination of wave functions ˜ φ(r) = α φ(1. 1. this linear combination o is not necessarily orthogonal to other degeneraste states.127) ˜ ˆ ˜ obeys the stationary Schr¨dinger equation H φ(r) = E122 φ(r). a. r). 5) as shown in the Table above. 3. φ(1. for (n1 . r) + β φ(2.3. However. for example.2. a.

72 Schr¨dinger Equation o .

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

the spectrum of ˆ H in (4. 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.3) Due to the nature of the harmonic potential which does not allow a particle with ﬁnite energy to move to arbitrary large distances. t) = exp(−iEt/ )φE (x. In the following algebraic solution of (4. in which the operators used below. the natural boundary conditions apply x→±∞ lim φE (x) = 0 . however. therefore. beside the representation (4. The cardinal property exploited below.5) →±∞ where C∞ denotes the set of functions which together with all of their derivatives are continuous. is the commutation property 1 1 (4. all stationary states of the harmonic oscillator must be bound states and. The Hamiltonian H can be expressed in terms of the operators acting on the space (4. It is important to keep in mind this restriction of the space.235) had been determined.6) the ﬁrst being a diﬀerential operator and the second a multiplicative operator.4) Equation (4.5). t) where ˆ H φE (x) = E φE (x) .3) would be continuous and the eigenfunctions φE (x) would not be normalizable. The operators act ˆ on the space of functions N1 deﬁned in (4.2) can be expressed in terms of the two operators p = ˆ d . (4. i dx x = x ˆ (4.5) ˆ H = 1 1 2 p + m ω 2 x2 ˆ ˆ 2m 2 (4.3) can be solved for any E ∈ R. (4.7) of the Hamiltonian. We will point out explicitly where assumptions are made which built on this restriction.1) of the form ψ(x. In the framework of the Schr¨dinger equation the stationary o states are solutions of (4. If this restriction would not apply and all functions f : R → R would be admitted. 4. only for a discrete set of E values can the boundary conditions (4. however.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.1 Creation and Annihilation Operators The Hamiltonian operator (4. The stationary states of the harmonic oscillator have been considered already in Chapter 2 where the corresponding wave functions (2.7) which is why these operators are of interest to us. lim ( ) = } (4. In order [ˆ. ∈ C∞ . x] = p ˆ . This property states that p and x obey an ˆ ˆ algebra in which the two do not commute.8) i which holds for any position and momentum operator.4) be satisﬁed. act.

Before we continue we like to write (4.12) To prove this commutation property we determine using (4. to factor H/ ω.10) holds for a ˆ ˆ [ˆ− . 1 i dx i dx i i (4. a+ ] = 1 .8) yield a− a+ = ˆ ˆ Similarly one can show a+ a− = ˆ ˆ 1 ˆ 1 H − 1 . a ˆ 1 (4.7) and (4.14) 1 i mω 2 x + ˆ p2 + ˆ [ˆ.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 . ˆ 2 2m ω a− = ˆ mω i x + √ ˆ p.18) It is of interest to note that the operators a+ and a− are real diﬀerential operators. One obtains [ˆ. Being guided by the identity for scalar numbers (b − ic)(b + ic) = b2 − i(cb − bc) + c2 = b2 + c2 (4.13) (4.15) together lead to the commutation property (4. the action of [ˆ. Our next step is an attempt to factorize the Hamiltonian (4.16) (4.14. In fact. p ˆ 2 2m ω 2 (4.4.11) a− a+ = ˆ ˆ (4. ˆ ˆ . in fact. hence. x] p ˆ on such functions must be considered.15) 1 ˆ 1 H + 1 .1: Creation and Annihilation Operators 75 to derive (4. a+ ] = a− a+ − a+ a− does not necessarily vanish. 1 2 (4. 2mω dx (4. 2mω dx a− = ˆ mω x + ˆ 2 d . 4.8) a ˆ ˆ ˆ ˆ ˆ implies [ˆ− . the commutator property (4. Since a+ and a− are ˆ ˆ + and a− since operators and not scalars we cannot simply expect that the identy (4.10) we deﬁne a+ = ˆ mω i x − √ ˆ p. x] f (x) = p ˆ d d x f (x) − x f (x) = f (x) = 1 f (x) .8).11) ˆ The reader may note that we have attempted.7) assuming that the factors are easier to handle than the Hamiltonian in yielding spectrum and eigenstates. 1 ω 2 (4.15) in a form which will be useful below ˆ H = ˆ H = ω a− a+ − ˆ ˆ ω a+ a− + ˆ ˆ ω 1 1 2 ω 1 .14) and (4.12).8) we recall that all operators act on functions in N1 and. ˆ 2 2m ω (4. 1 ω 2 (4. x] .9) From this follows (4.

(4.23) Together. g ∈ N1 +∞ +∞ dx f (x) a+ g(x) = −∞ −∞ dx g(x) a− f (x) . 4.17. Using (??) we like to state (4.16.76 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.21) ˆˆ H a+ φE (x) ω 1 ) a+ φE (x) 1 ˆ 2 ω = a+ ( ω a− a+ − ˆ ˆ ˆ 1 + ω 1 ) φE (x) 1 1 2 ˆ = a+ ( H + ω ) φE (x) ˆ = ( ω a+ a− + ˆ ˆ = (E + ω ) a+ φE (x) . (4.16.16. it holds that for a stationary state φE (x) of energy E deﬁned through (4. 4.24) − m This property states that the operators in the function space N1 are the hermitian conjugate of each other. The derivation will be based solely on the properties (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) . . 4. 4. ˆ The results (4. one obtains using (4. 4.17.19).22.17. In fact. ˆ (4. This property of operators is investigated more systematically in Section 5.12.21) is available.19) This property states that the operators a+ and a− are the adjoints of each other. ˆ (4. The property ˆ ˆ follows directly from (??).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. (4.20) ˆ In the following we will determine the spectrum of H and its eigenstates. 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) (4.22) Similarly one can show using again (4. 4.19) in the form1 f |ˆ+ g a Ω∞ = g|ˆ− f a Ω∞ . 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) .

2 Ground State of the Harmonic Oscillator d + y dy A suitable solution of (4. a state φ0 (x) which obeys the property a− φ0 (x) = 0 ˆ (4. Another name is creation (ˆ+ ) and annihilation (ˆ− ) operators since these operators a a create and annihilate vibrational quanta ω. (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.30) Assuming that φ0 (y) does not vanish anywhere in its domain ] − ∞. E . 4. Using (4. (ii) the construction appears to yield energy eigenvalues without lower bounds when. (4.25) φ0 (y) = 0 (4.25) can. in fact. E = E .27) Since E = E one can conclude φE |φE Ω∞ = 0.28) 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. m ∈ Z when m is decreased. (4. be found. In fact.6.5). φ0 (y) dy dy the solution of which is 1 lnφ0 (y) = − y 2 + c0 2 φ0 (y) = c0 exp 1 − y2 2 .33) . It turns out that both diﬃculties can be resolved together.25) needs to be normalizable in order to represent a bound state of the harmonic oscillator. 2 2 (4. φ0 (x) should be an element of the function space N1 deﬁned in (4.29) d 1 d φ0 (y) = = lnφ0 (y) = − y .29) where y = mω x. one expects that E = 0 should be a lower bound. two diﬃculties: (i) one needs to know at least one stationary state to start the construction. For f (x) = φE (x) and g(x) = φE (x) in (??) follows (Note that according to (??) the eigenvalue E is real.e.11) one can rewrite (4. There arise. i. (4.32) for some constant c0 or (4.26) Of course. however.) ˆ ˆ 0 = φE |HφE Ω∞ − φE |HφE Ω∞ = ( E − E ) φE |φE Ω∞ . The property (??) has an important consequence for the stationary states φE (x).31) (4..17) ˆ H φ0 (x) = ( ω a+ a− + ˆ ˆ 1 ω 1 ) φ0 (x) = 1 ω φ0 (x) . the solution φ0 (x) of (4. +∞[ one can write (4.4. For obvious reasons one refers to a+ and a− as ladder ˆ ˆ operators.25) on one side would lead to termination of the sequence E0 + m. in fact.

therefore.36) Since the ﬁrst order diﬀerential equation (4.36) n times.e. we construct the states for n = 1. m ∈ Z which properly terminate at some minimum value E + m0 ω ≥ 0. i. which for a narrowly localized state yields a large positive value.2) there are two competing contributions to the energy.24) can then be written as follows En = ω(n + 1 ).35) The appropriate ground state is φ0 (x) = mω π 1 4 exp − mωx2 2 .39). (4. the potential energy contribution which for a state δ(x). we can now construct the stationary states corresponding to energies (4. 1. 2. 2.39) We notice that the ground state wave function φ0 (x) as well as the operators (ˆ+ )n are real. 2 n = 0.34) dx |φ0 (x)|2 = |c0 |2 −∞ −∞ dx exp − mωx2 2 = |c0 |2 π . The ground state (4. i. . .3 Excited States of the Harmonic Oscillator Having obtained a suitable stationary state with lowest energy. . The reason is that in the Hamiltonian (4..37) above the ground state energy. Such states need to be suitably normalized for which purpose we introduce a normalization constant cn φn (x) = cn a+ ˆ n φ0 (x) . (4. it holds ˆ H φn (x) = ω(n + 1 ) φn (x) .. . We recall that according to (4. n = 0. We a can.. 1. 4. 1. 2..78 Linear Harmonic Oscillator This solution is obviously normalizable.e. would yield a vanishing contribution.37) It is most remarkable that the energy 1 ω of the ground state is larger than the lowest classically 2 allowed energy E = 0. We will consider this point more systematically in Section 5. choose the normalization constants cn and the functions φn (x) real as well. For this purpose we apply the operator a+ to the ˆ ground state (4. 2 This state of lowest energy is called the ground state of the oscillator.38) These states correspond to the energy eigenvalues (4.e. . and the kinetic energy contribution. . mω (4. . . the so-called excited states. .36) assumes a functional form such that both terms together assume a minimum value.25) admits only one solution there is only one set of states with energy E + m ω. (4.26) the energy value associated with this state is 1 ω. . . 2 n = 0. The set of allowed energies of the oscillator according to (4. . i. 2. The conventional normalization condition φ0 |φ0 reads +∞ +∞ Ω∞ = 1 (4. a state conﬁned to the minimum corresponding to the classical state of lowest energy. . (4.

41) Employing the adjoint property (4.40) φn−1 (x) = βn a− φn (x) ˆ (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 .51) .40) = 1. Ω∞ 2 = βn a− φn |ˆ− φn ˆ a Ω∞ 2 = βn φn |ˆ+ a− φn a ˆ Ω∞ . We consider then the situation that we have obtained a properly normalized state φn (x). (4. 1.50) φn (x) . according to (4.47).20) 1 = φn−1 |φn−1 Equations (4.18)] and since the ˆ φn (x) are real functions. n. 1.20) yields 2 αn φn |ˆ− a+ φn a ˆ Ω∞ = 1.42) ˆ Using (4.45) We like to note that these functions according to (4. √ a− φn (x) = n φn−1 (x) . . (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. For n = 0 holds c0 = 1.24) holds in analogy to (4.28) and the construction (4.44) 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) .40–4. (4. A properly normalized state φn+1 (x) is then of the form φn+1 (x) = αn a+ φn (x) ˆ for some real constant αn which is chosen to satisfy φn+1 |φn+1 Ω∞ 2 = αn a+ φn |ˆ+ φn ˆ a Ω∞ (4. . . ˆ (4. n = 0. 2.43) (4. . 4. they obey φn |φn = δn n . .45) form an orthonormal set.4. .39) yield √ From this follows βn = 1/ n and.e. To determine βn we note using (4. according to (4. 2. n = 0. √ a+ φn (x) = n + 1 φn+1 (x) .. (4.40). i. (4. βn must be real as well. . (4.46) According to (4.15 . To determine these constants we adopt a recursion scheme.49) (4. Since a− is a real diﬀerential operator [see (4.48) (4.3: Excited States 79 We need to determine now the normalization constants cn .47) for some suitable constants βn .

52).. It should be noted that √ the normalization (4. by introducing the variable y deﬁned in (4.57) which includes the factor 1/ n! according to Eqs.35) rather than (4.54) The eigenstates of the Hamiltonian are then given by φn (y) = − √ e n n! π 2 1 y2 2 e y2 2 y − d dy n e− y2 2 .53) We will later re-introduce this factor to account for the proper normalization (4. i.35) by a constant. (4. namely the square root of the Jacobian dx/dy. (4. by the Rodrigues formula (2.52) This normalization of the wave functions diﬀers from that postulated in (4.46) yields a set of normalized states.40–4.e. 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 . n–independent factor.200).7 and given. by dx = dy mω 1 4 . for example. . (4. (4.55) 1 y2 2 (4. however. (4. 2.56) Relationship to Hermite Polynomials We want to demonstrate now that the expression (4. simplifying the calculation.39). 2. For this purpose we start from expression (4.28) of the ground state and the deﬁnition (4.30) and employing the normalization +∞ −∞ dy φ2 (y) = 1 n (4.45). We will demonstrate below the identity Hn (y) = e y2 2 y − d dy n e− y2 2 (4.56) can be expressed in terms of Hermite polynomials Hn (y) introduced in Sect.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) .58) This result agrees with the expression (2.7.80 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 (4.233) derived in Sect.

dy dy (4. dy n (4.106–3.30) and return later to the coordinate x.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 (4.59) holds for n = 0.4 Propagator for the Harmonic Oscillator We consider now the solution of the time-dependent Schr¨dinger equation of the harmonic oscillator o (4.59) by induction noting ﬁrst that (4. (4.59) dn −y2 e dy n (4.59) hold.39) ∞ ψ(y0 . 1. the Rodrigues formula (2. hence. For the sake of notational simplicity we employ initially the coordinate y as deﬁned in (4. (4.114)].60) and. and showing then that the property also holds for n + 1 in case it holds for n.e.4.61) and.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 . We prove (4. i. Our derivation will follow closely the procedure adopted for the case of a ‘particle in a box’ [see Eqs.. (4. (3. t0 ).2) for an arbitrary initial wave function ψ(x0 . 4.200) we need to verify y − which implies Hn (y) = (−1)n ey 2 d dy n e− y2 2 = (−1)n e y2 2 dn −y2 e . Starting point of our derivation is the assumption that the initial condition can be expanded in terms of the eigenstates φn (y) (4.65) . t0 ) = n=0 dn φn (y0 ) .200).1.4: Propagator 81 To verify the realtionship between the deﬁnition (4. g(y) = (−1)n+1 e y2 2 dn+1 −y2 e = dy n+1 y − d dy n+1 e− y2 2 .57) of the hermite polynomials and the deﬁnition given by (2.64) which implies that (4.61) One can factor g(y) and employ (4. therefore.63) one obtains g(y) = y − d dy f (y) (4. 4.

(4. t0 ) = n=0 φn (y) φn (y0 ) tn+ 2 . Englewood Cliﬀs.70) and (2. this is an excellent textbook from which we have borrowed heavily in this Section. t|y0 . t0 ) which is an element of N1 deﬁned in (4.112) and (4. 68. 2 A demonstration of this property can be found in Special Functions and their Applications by N. t) = π −3/2 exp 2 2 y 2 y0 + 2 2 2 +∞ +∞ ∞ du −∞ −∞ dv n=0 1 ( −2tuv )n × n! (4. t1 ) = n=0 dn φn (y) cn (t) (4.15. y0 .J.Lebedev (Prentice Hall. 2. 1965) Sect. t0 ) φn (y0 ) .5).225) which allows one to derive a generating function for products of Hermite polynomials which can be applied to the r. . 1 t = e−iω(t1 −t0 ) (4.147) we employ the technique of generating functions as in Sect. t) = n=0 Hn (y)e−y /2 Hn (y0 )e−y0 /2 n √ t .s.108–3. sec:harm. t) (4. t1 |y0 .46) the expansion coeﬃcients dn are +∞ dn = −∞ dy0 ψ(y0 . N.39) holds cn (t1 ) = exp −i ω ( n + 1 ) ( t1 − t0 ) 2 .65) to times t ≥ t0 through insertion of time-dependent coeﬃcients cn (t1 ) ∞ ψ(y.N. 4. We consider for this purpose the following expression for |t| < 1 ∞ w(y. 2n n! π 2 2 (4.71) Applying (??) to express Hn (y) and Hn (y0 ) yields exp(−2tuv) w(y.67) for which according to (3. pp.. t0 ) ψ(y0 . Employing orthogonality condition (4. of (4. We want to demonstrate now that this propagator is identical to the propagator (2.72) × exp( −u − v + 2iyu + 2iy0 v ) . In order to prove the equivalence of (4.66) One can extend expansion (4.. y0 .70) is the propagator of the linear harmonic oscillator.147) for the harmonic iscillator determined in Sect.h. (4. Inc.82 Linear Harmonic Oscillator Such expansion is possible for any f (y0 ) = ψ(y0 .70).7. a supposition which is not proven here2 . For this purpose we start from the integral representation (2. t1 ) = −∞ dy0 φ(y.68) Altogether one can express then the solution +∞ ψ(y.69) where ∞ φ(y.

it holds φ(y.4: Propagator Carrying out the sum on the r. by a factor 4 mω/ for both φn (y) and φn (y0 ). t0 ) = where F (t1 .77) The sum in (4. t) = π −1 exp − 2 y 2 y0 + 2 2 +∞ dv exp(−(1 − t2 ) v 2 − 2i (yt − y0 ) v) .. t|x0 . The resulting propagator is G(t1 .e.73) The incomplete Gaussian integral +∞ −∞ dx e−a 2 x2 √ − 2bx = π b2 /a2 e . t) = π −3/2 exp +∞ 83 y2 y2 + 0 2 2 +∞ dv exp( −v 2 + 2iy0 v ) × −∞ × −∞ du exp( −u2 − 2u(tv − iy) ) . This requires that we employ (4. y0 .74) a second time yields ﬁnally together with the deﬁnition (4. (4. t0 ) exp i 2 1 2 (y + y0 ) G(t1 .74) applied once results in w(y.79) 1 + e−2iω(t1 −t0 ) cosω(t1 − t0 ) = . . t1 |y0 . 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. t0 ) − i y y0 2 F (t1 .77). one obtains w(y.70) is.4.e. t0 ) = exp mω 2iπ sinω(t1 − t0 ) 1 2 × .80) −2iω(t1 −t0 ) sinω(t1 − t0 ) 1−e We can ﬁnally express the propagator (4..147) derived by means of the path integral description. indeed. y0 . t0 ) = and 1 2 i π F (t1 .s. (4.78) in terms of the coordinates x and x0 .71) of w(y.30) to replace y and y0 and that we multiply the propagator by mω/ .76) One can express this in terms of the stationary states (4.81) imω 2 sinω(t1 − t0 ) (x2 + x2 ) cosω(t1 − t0 ) − 2 x x0 0 This result agrees with the propagator (2. √ πexp( t2 v 2 − 2iytv − y 2 ) (4. a Re a2 > 0 (4. y0 .78) 1 − e−2iω(t1 −t0 ) = sinω(t1 − t0 ) 2 i e−iω(t1 −t0 ) (4. i. t0 ) (4.h. (4. t0 ) = i φ(x.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. i.75) Applying (4. identical to the generating function (4.

It is with quantum electrodynamics and solid state physics in mind that we cease the opportunity of the single quantum mechanical harmonic oscillator to develop a working knowledge for a+ and a− in the most simple setting. d/dy and a+ .82) a+ = √ ˆ dy dy 2 2 √ √ Obviously. (4.g.85) is convergent everywhere in R. each corresponding to a single harmonic oscillator of the type studied presently by us. e. both quantum systems are endowed with a large number of modes. one can express y = 2(ˆ− + a+ ) and d/dy = 2(ˆ− − a+ ) and. Here f (ν) (y0 ) denotes the ν-th derivative of f (y) taken at y = y0 . hence.84) = 1 1 = 0 √ = n + 1 φn+1 (y) √ = n φn−1 (y) = f |ˆ+ g a Ω∞ . and of the modes of the quantized electromagnetic ﬁeld.83) We will encounter below functions of a+ and a− . a (4.86) f (ˆ+ ) = a a ˆ ν! ν=0 .84 Linear Harmonic Oscillator 4. 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.. a− = √ ˆ y + . In this case we deﬁne ∞ f (ν) (0) + ν (4.. We want to introduce now techniques based on the ladder operators a+ and ˆ a− .e. 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. the use of generating functions. (4. i. a− must be equivalent. 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. 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! . the approaches a ˆ a ˆ using y. 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 − .5 Working with Ladder Operators In the last section we have demonstrated the use of diﬀerential equation techniques. phonons. f (ˆ+ ).

88) (4.88) is (ˆ− )n f (ˆ+ ) φo (y) = f (n) (ˆ+ ) φ0 (y) .90) we show that (4. It holds a a euˆ f (ˆ+ ) φ0 (y) = f (ˆ+ + u) φ0 (y) . (ˆ+ )n a+ ] = (ˆ+ )n [ˆ− . f (ˆ+ )] = f (1) (ˆ+ ) . We note ˆ a− f (ˆ+ ) = [ˆ− .91) (4.92) f (y) = n=0 dn φn (y) = n=0 (ˆ+ )n a dn √ φ0 (y) n! (4.90) holds for f0 and for f1 .. 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+ ). 4.90) holds for fn . a a (4. property (4.95) (4. The ﬁrst case is trivial.87) (4. a a a 85 (4. a− f (ˆ+ ) φ0 (y) = f (1) (ˆ+ ) φ0 (y) ˆ a a which follows from a− φo (y) = 0. the second case follows from [ˆ− .90) holds for f (ˆ+ ).93) one can reduce all operators acting on some proper state function by operators acting on the state φ0 (y).88) we need to show actually [ˆ− .90) holds for any function fn (ˆ+ ) = (ˆ+ )n which is a power of a a a+ .4. The following important property holds a a− f (ˆ+ ) = f (1) (ˆ+ ) + f (ˆ+ ) a− .94) This identity follows from (4.e. Hence. f1 (ˆ+ )] = [ˆ− . fn+1 (ˆ+ )] a a = [ˆ− .89) (4.90) To prove (4. ˆ a a a ˆ In particular. For fn+1 follows then [ˆ− .87. i.88) is a fundamental one and will be used in the following. f (ˆ+ )] + f (ˆ+ ) a− ˆ a a a a ˆ which implies that in order to prove (4.5: Working with Creation and Annihilation Operators and similarly for f (ˆ− ). a a a We like to state the following property of the functions f (ˆ± ) a f (ˆ− ) φ|ψ = φ|f (ˆ+ ) ψ . we will only assume operator functions acting on φ0 (y). 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± ).96) . An example is the so-called shift operator exp(uˆ− ). a a − (4. a An important operator function is the exponential function. a a a ˆ 1 a Let us assume that (4. a+ ] + [ˆ− . (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. The convergence of the Taylor expansion ascertains then that (4. a+ ] = 1 = f1 (ˆ+ ) . Note that ˆ a an immediate consequence of (4.

euzˆ ˆ euz ∗ a− ˆ (4. without explicitly stating this. + (4. (4.101) 1. + (4. z ∈ C. a a ν! (4. 1 + + uz ∗ a− ).105) a eαˆ + − α∗ a− ˆ φ0 (y) = e αα∗ 2 e−α ∗ a− ˆ a eαˆ φ0 (y) . We assume in the following a ˆ derivation.107) − α∗ a− ˆ φ0 (y) = e− αα∗ 2 a eαˆ φ0 (y) . To express ˆ these operators as products of operators exp(vˆ+ ) and exp(v ∗ a− ) we consider the operator C(u) = a ˆ + ) exp(uz ∗ a− ) where u ∈ R. Below we will use the operator identity which follows for the choice of v = α in (4.98) The related operators exp[vˆ+ ± v ∗ a− ] play also an important role.99) ˆ C(u) = zˆ a zˆ+ + z ∗ a− a ˆ + +z a ˆ ∗ − a euzˆ euz ˆ − u C(u) .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) .106) yields a eαˆ + ∗ a− ˆ a a eαˆ φ0 (y) = e−αα eαˆ φ0 (y) . one obtains a evˆ + − v ∗ a− ˆ a φ0 (y) = e− 2 vv evˆ e− v 1 ∗ + ∗ a− ˆ φ0 (y) (4.104) (4.105) e e vˆ+ a + v ∗ a− ˆ φ0 (y) = e − 1 vv ∗ 2 e v ∗ a− ˆ e vˆ+ a φ0 (y) vˆ+ a − v ∗ a− ˆ φ0 (y) = e 1 vv ∗ 2 e −v ∗ a− ˆ e vˆ+ a φ0 (y) .101) ˆ ˆ ˆ C(u) obviously obeys C(0) = 1 This results in D(0) = 1 and.102) Similarly. the solution of (4. + ∗ + (4.97) An example in which (4.100) ˆ ˆ To solve this diﬀerential equation we deﬁne C(u) = D(u) exp(−u2 /2) which leads to d ˆ D(u) = du zˆ+ + z ∗ a− a ˆ ˆ D(u) (4. that the operators considered act on φ0 (y).108) . + (4.98) for u = −α∗ and v = α yields e−α This together with (4. One can conclude then using the deﬁnition of C(u) and deﬁning ˆ ˆ is D(u) = exp(uzˆ a ˆ v = uz 1 ∗ ˆ− ∗ ∗ ˆ− a+ a+ evˆ + v a φ0 (y) = e 2 vv evˆ ev a φ0 (y) .106) Applying (4.90) and zz ∗ = 1 we can write this d du e uzˆ+ a e uz ∗ a− ˆ a − z ∗ a− . hence. ∗ = 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. + ∗ a− ˆ a − u euzˆ euz + ∗ a− ˆ = (4.103) (4.

s.58) the Hermite polynomials Hn (y) by the eigenstates φn (y) √ ∞ 1 ( 2t)n y2 /2 2yt − t2 √ e φn (y) (4.115) and using (4. is the equivalent of the generating function (??).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.h.111) This expression. Using (4. we can reproduce by means of the generating function (4.4.h. and on the r. provide the same values for Hn (0) as provided in Eq.84) yields ∞ ν.h.111) the orthonormality properties of the wave functions φn (y).114) where we have employed (4.115) a where the latter step follows after expansion of evˆ and using a− φ0 (y) = 0. We want to derive now the values Hn (0).h.s.s. Setting y = 0 in (4. + (4.95) one obtains a a a a euˆ φ0 |evˆ φ0 = φ0 |euv evˆ φ0 = euv evˆ φ0 |φ0 = euv − + + + − (4. of (4. 4. in the ladder operator calculus.110) √ y2 u2 + 2uy − 2 2 a = euˆ φ0 (y) .98) and again (4. Expanding r. the expected from which follows by comparision of each term on both sides φµ |φν orthonormality property.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! + (4.109) e = π4 n! n=0 √ Using (4.84. We start from the generating function (??) and replace according to (4.58) 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.µ=0 uµ v ν √ φµ |φν µ!ν! ∞ = µ=0 uµ v µ µ! (4. and ˆ l.s.. 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). Similarly.112) Expanding both sides of this equation and using (4.111) yields π − 4 exp − 1 u2 2 a = euˆ φ0 (0) .116) = δµ ν . .95). i. (??).e.113) Comparision of all terms on the l. For this purpose we consider a a a a euˆ φ0 |evˆ φ0 = φ0 |euˆ evˆ φ0 + + − + (4. (4.

the momentum and position operators are as stated in (4. for an oscillator in a stationary state φn (x) as given by (2.7) is 1 2 m 2 ω 2 d2 p − . ˆ and the Hamiltonian (4.122).e.123) (4.125) .124) ). . (4.122) For a solution of the Schr¨dinger equation in the momentum representation. 1. . . i. (−i)n ˜ φn (p) = √ 2n n! m˜ ω π 1 4 exp − m˜ p2 ω 2 Hn ( m˜ ω p) .. The solutions can be stated readily exploiting the earlier results. As to be expected.. The eigenvalues. are ˜ En = or. In the momentum representation the wave function is a function of p. using (4. En = ω (n + 1 ) 2 d2 1 + m˜ 2 p2 ω 2 2m dp 2 2 ˜ ˜˜ φE (p) = E φE (p) (4.235). . .3).2).6) and the Hamiltonian (4. ω = 1 / m2 ω . 4. the momentum and position operators are p = p.235). In the position representation the wave function is a function of x. n = 0.e. n = 0.. the time-independent Schr¨dinger equation for the oscillator can be written o ˆ H = − mω 2 2 2 x = i ˆ d .119) 2m 2 dp2 Accordingly.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. ˜ (4. according to (4.37). is given by a function ψ(x). is given by the ˜ function ψ(p). i. 2. . using (2. we o note that the posed eigenvalue problem is formally identical to the Schr¨dinger equation in the o position representation as given by (4.88 Linear Harmonic Oscillator 4. 2. dp (4.e. (4. 1.117) 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.120) Multiplying this equation by 1/m2 ω 2 yields − where ˜ E = E / m2 ω 2 . (4.118) d2 1 2 + p dp2 2m ˜ ˜ φE (p) = E φE (p) .121) ω (n + ˜ 1 ) 2 . the eigenvalues are identical to those determined in the position representation. For this purpose we employ the Schr¨dinger equation in the momentum o representation. For example.121). i.7) is given by (4. of (4.2. The momentum representation eigenfunctions are. the probability density is P (x) = |φn (x)|2 . (4. (4.

128) We may also express the eigenstates (4. (4.30) to obtain (4. 2n n! π (4. α ∈ R. From (4.126) in terms of dimensionless units following the procedure adopted for the position representation where we employed the substitution (4.f.e. the correctness of the phase factors (−i)n .134) Exercise 4. m ω (4. φn (k) = We want to ﬁnally apply the relationship 1 ˜ φn (p) = √ 2π +∞ 2 (−i)n −k √ e 2 Hn (k) .30) and (2.6: Momentum Representation or with (4.6.134). the Jacobian. deﬁned only up to an arbitrary phase factor eiα . m ω (4. .131) in the present case employing dimensionless units.235)] absorbs the replacement dx → dy.1: Demonstrate the validity of (4.58). through direct integration. i.129) m ω one obtains.130) dx exp(−ipx/ ) φn (x) −∞ (4.30) and (4. ˜ We can now state the probability density Pn (p) for an oscillator with energy En to assume momentum p..126) a phase factor (−i)n .4. 1 ˜ φn (k) = √ 2π +∞ (4. This implies the property of Hermite polynomials (−i)n e− k2 2 +∞ Hn (k) = −∞ dy exp(−iky) e− y2 2 Hn (y) (4.129) follows ky = px/ and. 4.127) exp − p2 m ω 2 Hn ( 1 p) . 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 (4.122) (−i)n ˜ φn (p) = √ 2n n! 1 πm ω 1 4 89 exp − p2 2m ω Hn ( 1 p) .. Deﬁning k = m˜ / p or ω 1 k = p (4.133) where we note that a change of the normalization of φn (y) [c. instead of (4.e. this allowed us to introduce in (4.126) Normalized eigenfunctions are. hence. Proceed as follows.126). i. of course.132) dy exp(−iky) φn (y) −∞ (4.125.

134). Inserting this expansion into √ (4. z) = √ 2π where f (k. 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..7 Quasi-Classical States of the Harmonic Oscillator A classical harmonic oscillator. We ﬁrst construct the solution of (4. For this purpose we assume such state exists and expand ∞ ∞ φ(α) (y) = n=0 d(α) φn (y) . for example. The quasi-classical states can be obtained by generalizing the construction of the ground state of the oscillator. except that α changes periodically in time. n n=0 d(α) n 2 = 1 (4.1 showing in its top row the attributes of such state just discussed. or Glauber states of the harmonic oscillator and they play. pp. J.Chen-Tannoudji.139) where we have added the condition that the states be normalized.4.90 (a) Prove ﬁrst the property 1 f (k. New York. φ(α) | φ(α) = . y) −∞ (4. Inc. namely through the eigenvalue problem for normalized states a− φ(α) (y) = α φ(α) (y) . z) = exp 2yz − z 2 − y 2 /2 (4.137) 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.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 (4.196) of Hermite polynomials and equate coeﬃcients of equal powers of z to prove (4. The motion of the probability distribution of such state is presented in Fig. coherent states.135) g(k. for example.Dupont-Roc. 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 . and G. We want to construct and characterize such states. 4.138). Such states are referred to as quasi-classical states. z) = exp −2ikz + z 2 − k 2 /2 .— III. a pendulum swinging at small amplitudes. 1989) which discusses quasi-classical states of the free electromagnetic ﬁeld in Sect.194. The question arises if similar states exist also for the quantum oscillator. 2.138) We will show that a harmonic oscillator prepared in such a state will remain forever in such a state. 4. (4.C. 192 3 .136) (b) Employ the generating function (2. The answer is ‘yes’. carries out a periodic motion described by y(t) = Re y0 eiωt (4. +∞ Linear Harmonic Oscillator dy exp(−iky) g(z.Grynberg (John Wiley & Sons. ˆ α ∈ R.140) An excellent textbook is Photons and Atoms: Introduction to Quantum Electrodynamics by C.

149) with (4. hence.149) − α∗ a− ˆ (4.146) √ 1 − (y − 2 α)2 2 (4.144) Normalization requires c = exp(−α2 /2) and.145) Using the generating function in the form (4. n! (4. n! (4.108) allows one to write φ(α) (y) = T + (α) φ0 (y) = T + (α) [π]− 4 e−y where a T + (α) = eαˆ + 1 2 /2 (4.109).145) using (4. i.84) φ(α) (y) = exp − α2 2 ∞ n=0 αα∗ (αˆ+ ) a a+ φ0 (y) = e− 2 eαˆ φ0 (y) .147) allows one to interprete T (α) as an operator √ which shifts the ground state by 2α.e..142) The solution is αn d(α) = √ c n n! ∞ ∞ (4. Comparing (4.4. One can also write (4.139)] 1 = n=0 d(α) n 2 = |c|2 n=0 α2 2 = |c|2 eα .143) for a constant c which is determined through the normalization condition [see (4. n! (4. .148) The identity (4. However. (4.138).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) .147) √ which identiﬁes this state as a ground state of the oscillator displaced by 2α. we will see below that α ∈ C are also admissible and will provide a corresponding interpretation. n! n (4. This interpretation justiﬁes the choice of α ∈ R in (4. the quasi-classical states are φ (α) (y) = exp α2 − 2 ∞ n=0 αn √ φn (y) . n n+1 (4.150) for α presently real. [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) .7: Quasi-Classical States or ∞ n=0 91 √ n + 1 dn+1 − α d(α) n (α) φn (y) = 0 .

.h.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 (4.145) shows that α in (4. becomes real.154) . t1 ) = exp or φ (α) − α2 2 ∞ exp n=0 −i ω ( n + ∞ n=0 1 ) ( t1 − t0 ) 2 αn √ φn (y) n! (4. .160) .67.145) is replaced by a complex number αu which at times t1 = t0 + 2πn/ω.152) Since the operator function on the r.92 Linear Harmonic Oscillator The shift operator as deﬁned in (4. a a ˆ ˆ (4. In particular. (4.156) (4. 4. (4. .109) serves again to simplify this sum √ y2 1 1 α2 (u2 + 1) φ(α) (y.95) to linear combinations αˆ+ + βˆ− and reads then a a f (αˆ+ + βˆ− ) φ|ψ = φ|f (α∗ a− + β ∗ a+ ) ψ .151) which follows from a generalization of (4.68)] one can write the time-dependent solution with (4. n! u = e−iω(t1 − t0 ) .h.155) (4.153) This is obviously the inverse of T + (α) as deﬁned in (4. Relationship (4.s. the shift operator leaves the ground state normalized. one can write in case of T + (α) T + (α) + = T (α) = eα ∗ a− ˆ − αˆ+ a a = e − (αˆ + − α∗ a− ) ˆ .150) for complex α obeys the unitary property T + (α) T (α) = 1 1 (4. i. n = 0. 1. it holds T + (α) φ0 |T + (α) φ0 = f |T (α) T + (α) g = φ0 |φ0 .s.e. t1 ) = u exp 1 2 α2 − 2 (α u)n √ φn (y) . α ∈ C (4.158) Comparision of this expression with (4.152) T + (α) f |T + (α) g = f |T (α) T + (α) g = f |g .. Following the procedure adopted in determining the propagator of the harmonic oscillator [see (4. t1 ) = π − 4 u 2 exp − + 2yαu − 2 2 E (4.159) Noticing the identity which holds for |u|2 = 1 u2 + 1 = (Re u)2 − (Im u)2 + 1 + 2 i Re u Im u = 2 (Re u)2 + 2 i Re u Im u (4.145) as the initial state φ(α) (y. is the adjoint of the operator function on the l.

4.159) is then 1 1 i φ(α) (y.147).1 (top row) we present the probability distribution of the Glauber state for various instances in time.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 ) (4.150) with (4. Arbitrary Gaussian Wave Packet Moving in a Harmonic Potential We want to demonstrate now that any initial state described by a Gaussian shows a time dependence very similar to that of the Glauber states in that such state remains Gaussian. 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. this characterization corresponds closely to that of the possible initial states of a classical oscillator. 4. experiences a change of its width (relative to .163) also through the shift operator. a momentum factor and a phase factor. t1 ) = u 2 T + (α u) φ0 (y) (4. moving solely its center of mass in an oscillator fashion around the minimum of the harmonic potential. obviously. (4.163) This solution corresponds to the initial state given by (4. (4.162) 2 with a periodic change. In Fig.161) and is found to describe a displaced Gaussian. a complex α corresponds to a displaced oscillator which has initially a non-vanishing velocity. One can express (4. in general. being displaced around the center with period 2π/ω and.138): a real α corresponds to a displacement such that initially the oscillator is at rest. 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 (4.158) allows one to write 1 φ(α) (y.138. The time-dependent wave function (4. Comparing (4. The diagram illustrates that the wave function retains its Gaussian shape with constant width for all times. t1 ) = π − 4 exp − [y − y0 (t1 )]2 + iy k0 (t1 ) − iφ0 (t1 ) − ω(t1 − t0 ) 2 2 . 4.4.164) where according to (4.148.165) . This shows clearly that the Glauber state is a close analogue to the classical oscillator.7: Quasi-Classical States the exponent E can be written E = − √ √ y2 + 2 αy Reu − α2 (Re u)2 + i 2 αy Imu − i α2 Reu Imu 2 √ √ 2 1 = − y − 2αReu + i 2yα Imu − i α2 Reu Imu 2 phase term momentum displaced Gaussian 93 (4.150) 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. Our interpretation 2 explains also the meaning of a complex α in (4.

169) . To describe such state which satisﬁes the initial condition φ(ao . 2σ 2 q = a . σ|y. This behaviour is illustrated in Fig. (4. t0 ) = n=0 dn φn (y) .1 (second and third row).1: Time dependence of Gaussian wave packets in harmonic oscillator potential. 4. 1 + σ2 (4. that of the Glauber states) with a period π/ω. 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.166) φ(ao .168) dy e−p(y−q) Hn (y) where p = 1 + σ2 .167) One can derive for the expansion coeﬃcients +∞ dn = −∞ dy φn (y) φ(ao . 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. σ|y. t0 ) = [πσ 2 ]− 4 exp we expand ∞ 1 − (y − ao )2 2σ 2 (4. σ|y.

Yu. unfortunately very expensive.177) − 1 ω(t1 − to ) 2 .143) of the Glauber state. 2 by A. New York. a worthy successor of the famous Gradshteyn. Marichev (Wiley.171) lim p−1 p n 2 Hn (q p ) = lim (2q)n = an . i.e. Following the strategy adopted previously one can express the time-dependent state corresponding to the initial condition (4. and O. σ→1 O(y n−2 ) for n ≥ 2 1 for n = 0.176) 1 √ pσ π a 2 exp − 1 1+σ2 + 1 yo − 1 ω(t1 − to ) 2 2 2 2 × (4. Brychkov. a ground state shifted by ao = 2α.7: Quasi-Classical States The solution of integral (4. in fact. in agreement with the expected result (4.157) as follows φ(ao . σ|y. 1 a2 2 1 + σ2 + 1 2 2 y o + 2 y yo This integral can be found in Integrals and Series. p−1 1 − σ4 t = 1 − σ 2 −iω(t1 −to ) e 1 + σ2 (4. σ|y. t1 ) = e−aq/2 √ pσ ∞ n=0 e −iω(n+ 1 )(t1 −to ) √ 2 1 2n n! p−1 p n/2 (4. this 3 volume integral table is likely the most complete today.173) which is. 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.I.166) in analogy to (4. In this limit the coeﬃcients dn should√ become identical to the coeﬃcients (4. t1 ) = √ × where yo = q p a = √ . t1 ) = √ 2 × exp − 1 (y 2 + yo ) 1+t2 − 2 1−t 2 1 1 √ √ pσ π 1 − t2 × t 1 − t2 (4. p−1 95 (4.175) ∞ tn n=0 2n n! Hn (yo ) e−yo /2 Hn (y) e−y 2 2 /2 The generating function (4. This expansion can be written φ(ao ..143).76) permits us to write this φ(ao .4.P. 1990). therefore.170) In order to check the resulting coeﬃcients we consider the limit σ → 1. σ|y. σ→1 p−1 2 (4. 1 (4. Prudnikov. namely $750 4 .168) is4 e−aq/2 (p − 1)n/2 dn = √ Hn (q 2n n!σ p(n+1)/2 p ). vol.172) One can then conclude e−a /4 n lim dn = √ a σ→1 2n n! (4.174) Hn (q p p−1 ) φn (y) .

96 Linear Harmonic Oscillator .

To develop a systematic description of rotational properties of composite quantum systems the consideration of rotational transformations is the best starting point.g. i. the group SO(3) of the associated rotation matrices and the corresponding transformation matrices of spin– 1 states forming the group SU(2) occupy a very 2 important position in physics. we will 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. 5. argon. we will then consider transformations of wavefunctions ψ(r) of single particles in R . The reason is that these transformations and groups are closely tied to the properties of elementary particles. or the proton and the neutron made up of three quarks.. e. all composite systems which appear spherical as far as their charge distribution is concerned. the building blocks of matter. 97 .e. We will also review below the well-known fact that spin states under rotations behave essentially identical to angular momentum states. neon. and ﬁnally transformations of products of wavefunctions like N ψj (rj ) which represent a system of N (spinj=1 zero) particles in R . r ∈ R . Examples of the latter with particularly simple transformation properties are closed shell atoms. the magic number nuclei like carbon.1 Matrix Representation of the group SO(3) In the following we provide a brief introduction to the group of three-dimensional rotation matrices.. helium.e. In this section we want to investigate how elementary and composite systems are described.Chapter 5 Theory of Angular Momentum and Spin Rotational symmetry transformations. The mathematical techniques presented in this section will be used throughout the remainder of these notes. As an illustration we will consider ﬁrst rotational transformations acting on vectors r in 3-dimensional space. We will also introduce the generators of this group and their algebra as well as the representation of rotations through exponential operators. but also to the properties of composite systems. i..

(5. A rotation around the x3 -axis by an angle ϕ is described by the matrix cosϕ − sinϕ 0 R(ϑ = (0. are linear and. =1 [R(ϑ)]jk [R(ϑ)]j ak b = j=1 aj bj ..2) Rotations conserve the scalar product between any pair of vectors a and b.5) This equation states the key characteristic of rotation matrices. From (5. in case of a prefactor −1 a rotation and an inversion at the origin. then the ﬁngers of your ﬁst point in the direction of the rotation.. 2. 0. i. (5.6) follows immediatety for the determinant of R(ϑ) using detA B = detA detB and detAT = detA detR(ϑ) = ±1 . the matrix represents a proper rotation. a minus sign in Eq. i. ϑ parametrizes the rotations uniquely as long as |ϑ| < 2π. x2 . 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.7) implies that a rotation is associated with an inversion. 0.e. In the following we want to exclude inversions and consider only rotation matrices with detR = 1.4) With the deﬁnition of the transposed matrix RT [RT ]jk ≡ Rkj this property can be written R(ϑ) RT (ϑ) = RT (ϑ) R(ϑ) = 1 1 . (5. (5. 3 . In the latter case the determinant of R(ϑ = (0.1) In Cartesian coordinates this reads 3 xk = j=1 [R(ϑ)]kj xj k = 1. ϕ)T ) is negative. therefore. x3 )T . in Cartesian coordinates r = (x1 .e. (5. they conserve a · b = 3 aj bj . i. namely around an axis given by the direction of ϑ and by an angle |ϑ|.8) 0 0 1 In case of a prefactor +1. can be represented by 3 × 3 matrices R(ϑ).7) Let us consider brieﬂy an example to illustrate rotational transformations and to interpret the sign of detR(ϑ). (5. where ϑ denotes the rotation.3) Since this holds for any a and b. . (5. it follows 3 [R(ϑ)]jk [R(ϑ)]j = δk j=1 .98 Properties of Rotations in R Theory of Angular Momentum and Spin Rotational transformations of vectors r ∈ R . It holds then j=1 3 3 a ·b = j.6) (5.k.e. ϕ)T ) = ± sinϕ cosϕ 0 (5. We assume the convention that rotations are right-handed.

forms a group G satisfying the axioms (i) For any pair a. (5. Since (R1 )T = (R1 )−1 it follows −1 −1 −1 T (R1 )T R1 = (R1 )−1 R1 = −1 which implies R1 ∈ SO(3). it holds 1. We need to demonstrate that this inverse belongs also to −1 T SO(3). 99 (5. In the following we will assume R1 . (iii) ∀a. b ∈ G the product c = a ◦ b is also in G. in fact.10) detR3 = det ( R1 R2 ) = detR1 detR2 = 1 . (i) Obviously. We have shown altogether that the elements of SO(3) form a group. (iv) For products of three elements holds the associative law a ◦ (b ◦ c) = (a ◦ b) ◦ c.9) This set of matrices is closed under matrix multiplication and. R2 ∈ SO(3) and do not write out “◦” explicitly since it represents the matrix product. We would like to point out the obvious property that for all elements R of SO(3) holds R−1 = RT (5. 1 T Similarly. a ∈ G → ∃a−1 ∈ G with a ◦ a−1 = a−1 ◦ a = e. T R1 R1 −1 = 1 −1 = 1 1 1 (5. hence. It holds T T T T R3 R3 = ( R1 R2 )T R1 R2 = R2 R1 R1 R2 = R2 R2 = 1 . one can show R3 R3 = 1 Furthermore.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. It follows that R3 is also an element of SO(3). R3 = R1 R2 is a real. there exists a real 3 × 3–matrix −1 R1 which is the inverse of R1 . (ii) There exists an element e in G with the property ∀a ∈ G → e ◦ a = a ◦ e = a.12) (iv) Since the associative law holds for multiplication of any square matrices this property holds for the elements of SO(3).5.11) (iii) From detR1 = 0 follows that R1 is non-singular and.9) forms a group. For this purpose we must demonstrate that the group properties (i–iv) hold. (5. We want to prove now that SO(3) as deﬁned in (5.13) . 3 × 3–matrix. R RT = RT R = 1 detR = 1 } 1. (ii) The group element ‘e’ is played by the 3 × 3 identity matrix 1 1.

0 −1 1 0 (5. J2 . due to the uniqueness of the inverse. Generators. presently. Lie Group. rotation by 90o . −1 0 0 −1 . 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.16) must be antisymmetric. 0 1 −1 0 . 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 (5.18) (5. i.17) . identify subgroups. We want to show that for the property R RT = 1 to hold. (c) the set of four rotations in the x1 .e. we arrive at the property exp ϑk LT k from which follows LT = −Lk . −1.15) Below we will construct appropriate matrices Jk . the matrices i Lk = − Jk (5.13) R(ϑk ek )−1 = R(ϑk ek )T = [exp ( ϑk Lk )]T = exp ϑk LT ˆ ˆ k and. in fact. M3 .1: Test if the following sets together with the stated binary operation ‘◦’ form groups. will play a central role in the representation of many other symmetry transformations in Physics. Lie Algebra We want to consider now a most convenient.100 Theory of Angular Momentum and Spin Exercise 5. establish if 1-1 homomorphic mappings exist. rotation by 180o . compact representation of R(ϑ) which will be of general use and. M4 } = 1 0 0 1 . x2 –plane {rotation by Oo . −i} together with multiplication as the binary operation ‘◦’. i. k = exp ( −ϑk Lk ) (5. J3 as follows i R(ϑ) = exp − ϑ · J (5. (b) the set of matrices {M1 . we want to assume that they exist. mappings between the sets which conserve the group properties: (a) the set of real and imaginary numbers {1.1. M2 .20) (5.19) (5. This representation expresses rotation matrices in terms of three 3 × 3– matrices J1 .14) together with matrix multiplication as the binary operation ‘◦’.

(5.1: Matrix Representation of SO(3) 101 We can conclude then that if elements of SO(3) can be expressed as R = eA .21) −b −c 0 It is this the reason why one can expect that three-dimensional rotation matrices can be expressed through (5. ϑ2 .2: Determine the generator L1 according to Eq. 0.e.(5.1.22) for k = 3 yields L3 = ϑ−1 3 One obtains in this way 0 0 0 i − J1 = L1 = 0 0 −1 0 1 0 0 0 1 i − J2 = L2 = 0 0 0 −1 0 0 0 −1 0 i 1 0 0 − J3 = L3 = 0 0 0 The matrices Lk are called the generators of the group SO(3). 0 0 0 0 0 0 (5.25) 0 −ϑ3 0 0 −1 0 ϑ3 0 0 = 1 0 0 .15) as the derivatives of R(ϑ) with respect to the Cartesian components ϑk taken at ϑ = (0. One can determine then the matrices Lk deﬁned in (5.22) and similar for L2 and L3 . 0)T . ϑ3 and three matrices L1 . 0)T ) − 1 1 1 ϑ1 →0 (5. Let us use this deﬁnition to evaluate L3 . This property gives A then only three independent matrix elements.. 0. 0. Using (5.22).5. A must have the form 0 a b A = −a 0 c .26) (5.8) we can state cosϑ3 −sinϑ3 0 1 −ϑ3 0 1 0 . 0)T the rotation is the identity. ϑ2 . lim R[(0.24) (5. ϑ3 )T ] = lim sinϑ3 cosϑ3 0 = ϑ3 ϑ3 →0 ϑ3 →0 0 0 1 0 0 1 Insertion into (5. i. the real matrix A is anti-symmetric.15) with three real parameters ϑ1 .23) (5.27) . L3 which are independent of the rotation angles. L2 . Exercise 5. Using the deﬁnition of partial derivatives we can state L1 = lim ϑ−1 R(ϑ = (ϑ1 . 0. ϑ3 )T such that for ϑ = (0. (5. We assume now that we parameterize rotations through a three-dimensional rotation vector ϑ = (ϑ1 .

29) where Z1 = A + B and where the remaining operators Zn are commutators of A and B.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. 0. 0. (5. L ] = m=1 fk m Lm . x2 -. 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. the Lie algebra of SO(3) can be written [Lk . commutators of commutators of A and B. only the latter example and .. We want to consider. The question is then. the property (5. m = 3 for any odd permutation of k = 1. An answer can be found considering the Baker-Campbell-Hausdorf expansion ∞ exp( λA ) exp( λB ) = exp( n=1 λn Zn ) (5. B] ). holds the algebra [Jk ..e. if the resulting products of exponential operators can be expressed in the form (5. B]] + [[A. the elements of if any two indices are identical for k = 1. it must hold 3 [Lk .30) are called Lie groups. In fact.15) and their generators obey the property (5. three being a special case. m = 3 . [A.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.. however. and x3 -axes. etc.8) in case of ϑ = (0. 3 )T ) = exp( 3 L3 ).g. as we see later. From this result one can conclude that expressions of the type (5. R(ϑ = (0.31) where k m are the elements which are 0 1 k m = 1 −1 of the totally antisymmetric 3-dimensional tensor. are related to angular momentum operators. i.e. Lie groups can have any number of generators.28) The construction of the generators through the limit taken in Eq.g.30) Groups with the property that their elements can be expressed like (5. however. 0. Of course. 1 1 e. = 2. that the generators represent inﬁnitesimal rotations with 1 >> |ϑ| around the x1 -.32) For the matrices Jk = i Lk which. The expression for Zn are actually rather complicated.15) obey the group property. In case of the group SO(3) the structure constants are particularly simple. (5. in fact. ϕ)T . (5. (5. B]. Finite rotations can be obtained by applications of many inﬁnitesimal rotations of the type R(ϑ = ( 1 . 0)T ) = exp( 1 L1 ). J ] = i k m Jm .15). that the operators (5. e. i. = 2. m = 3 for any even permutation of k = 1. Z2 = 2 [A.30) is called the Lie algebra. = 2. and the constants fk m are called the structure constants.33) We want to demonstrate now that exp(ϑ · L) yields the known rotational transformations.22) implies. 2 . L ] = k m Lm (5. the matrix (5. Z3 = 12 ( [A. 0)T ) = exp( 2 L2 ) and R(ϑ = (0. B].

41) cos ϕ − sin ϕ sin ϕ cos ϕ (5.1: Matrix Representation of SO(3) determine exp(ϕL3 ). can be written ϕL3 = 0 103 A 0 0 0 0 .35).39) Recognizing the Taylor expansions of cos ϕ and sin ϕ one obtains eA = and.31).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 . (2n + 1)! (5.40) .1. eϕL3 which agrees with (5.37) = (−1)n 1 0 0 1 . that the matrix ϕL3 . according to (5.27).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. (5. Exercise 5. using (5. cos ϕ − sin ϕ 0 = sin ϕ cos ϕ 0 0 0 1 (5.34) One obtains then for the rotational transformation exp A 0 0 0 0 0 1 0 0 1 0 0 0 1 0 0 1 0 0 0 1 0 0 1 ∞ ν=1 ∞ ν=1 0 0 1 = = = + + 1 ν! 1 ν! A 0 0 0 0 0 ν Aν 0 0 0 0 0 0 0 1 ∞ 1 ν ν=0 ν! A 0 0 = eA 0 0 .36) The property 0 −1 1 0 allows one to write 0 −1 1 0 and. We note.8).34.5. 5. ∞ 2n = − 1 0 0 1 2n+1 (5.3: Show that the generators Lk of SO(3) obey the commutation relationship (5. 0 −1 1 0 = (−1)n 0 −1 1 0 (5. (5. accordingly. A = ϕ 0 −1 1 0 .

(5. i. i. in fact.2. we deﬁne rotations R(ϑ) as linear maps C∞ ( ) → C∞ ( ). For this purpose we require stringent continuity properties of the wavefunction: the wavefunctions under consideration must be elements of the set C∞ ( ). complex-valued functions over the 3-dimensional space R which can be diﬀerentiated inﬁnitely often. B ∈ SO(3) holds ρ(A B) = ρ(A) ρ(B) ..2 Function space representation of the group SO(3) In this section we will investigate how rotational transformations act on single particle wavefunctions ψ(r).44) holds. its elements can be represented through an exponential R(ϑ) = exp i − ϑ·J (5. In analogy to R(ϑ) being a linear map R → R . we will learn that transformation patterns and invariances are connected with the angular momentum states of quantum mechanics. for A.47) = ρ R(ϑ1 ) R(ϑ2 ) . Deﬁnition We ﬁrst deﬁne how a rotational transformation acts on a wavefunction ψ(r ).45) (5. (5.48) . i.e. = ψ([ R(ϑ1 ) R(ϑ2 ) ]−1 r) = ρ R(ϑ2 ) ψ([R(ϑ1 )]−1 r) (5.42) Obviously.e.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) . (5.1: Assume the deﬁnition ρ ρ (A B) = ρ (B) ρ (A). Show that in this case holds Generators One can assume then that R(ϑ) should also form a Lie group. a relationship which we like to express as follows R(ϑ) = ρ R(ϑ) . Exercise 5.43) ρ( ) conserves the group property of SO(3). the transformation R(ϑ): C∞ ( ) → C∞ ( ) is related to the transformation R(ϑ): R → R . namely R(ϑ)ψ(r ) = ψ(R−1 (ϑ) r ) . Since this holds for any ψ(r ) one can conclude ρ R(ϑ1 ) ρ R(ϑ2 ) group property (5.e. the R(ϑ) ψ(r) = ψ(R(ϑ) r ).104 Theory of Angular Momentum and Spin 5. In particular. hence.46) (5. one isomorphic to SO(3) and.

(5. (5.2: (a) Derive the generators Jk .51) Inserting this into (5.57) Exercise 5. − i J ] f (r) 1 2 =− 2× [ (x3 ∂2 − x2 ∂3 ) (x1 ∂3 f − x3 ∂1 f ) − (x1 ∂3 − x3 ∂1 ) (x3 ∂2 f − x2 ∂3 f ) ] 2 +x1 x3 ∂2 ∂3 f − x1 x2 ∂3 f − x2 ∂1 ∂2 f + x2 x3 ∂1 ∂3 f + x2 ∂1 f 3 2(x ∂ 2 1 2 −x1 ∂2 f − x1 x3 ∂2 ∂3 f + x2 ∂1 ∂2 f + x1 x2 ∂3 f − x2 x3 ∂1 ∂3 f ] 3 = − − x1 ∂2 ) f (r) = i J3 f (r) .56) We like to verify this property for [J1 .50) and Taylor expansion yields i − J3 f (r) = = ϑ3 →0 lim ϑ−1 ( f (r) + ϑ3 x2 ∂1 f (r) − ϑ3 x1 ∂2 f (r) − f (r) ) 3 (5.52) ( x2 ∂1 − x1 ∂2 ) f (r) . 0)T ) − 1 1 ϑ1 →0 105 (5. J2 ]. i − J3 f (r) = = Using (5.2: Function Space Representation of SO(3) where the generators can be determined in analogy to Eq. i. −ϑ3 ) r ≈ −ϑ3 1 0 x2 = −ϑ3 x1 + x2 . One obtains using (5.55) and f (r) = f [J1 .24) one can expand ϑ3 →0 ϑ3 →0 lim ϑ−1 R(ϑ = (0.23. J1 ]..53–5. show that (5. 5.49) (note the minus sign of −ϑ1 which originates from the inverse of the rotation in Eq. J2 ] f (r) = − = − 2[ 2 [− i J . but also the Lie algebra of the generators. (5. J ] = i k m Jm . 3 1 ϑ3 0 x1 x1 + ϑ3 x2 R(0. in case of J3 . 0.54) (5.2. −ϑ3 ) r ) − f (r) ) . (5. .42)) and similarly for J2 and J3 .49) to a function f (r).49). 2 by means of limits as suggested in Eq. 0.50) lim ϑ−1 ( f ( R(0.(5.22) according to i − J1 = lim ϑ−1 R(ϑ = (−ϑ1 . it holds [Jk . 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 . The generators Jk can be determined by applying (5. In fact. 0 0 1 x3 x3 (5. 0.e.55) Not only is the group property of SO(3) conserved by ρ( ). k = 1. J3 ] and [J3 . (5.56) holds for [J2 . ϑ3 )T ) f (r) − f (r) 3 (5. 0.53) (5.5.

3 .60) commutes with all three Jk . the operator 2 2 2 J 2 = J1 + J2 + J3 (5.3 Angular Momentum Operators The generators Jk can be readily recognized as the angular momentum operators of quantum mechanics.53–5. i.63) (5. However.59) which. k = 1.62) J1 [J1 . identiﬁes J as the quantum mechanical angular momentum operator. J3 ] J1 + J2 [J2 . . 2.106 Theory of Angular Momentum and Spin Exercise 5.65) . [J 2 . using the momentum operator p = ˆ J = r × i . x2 = r sinθ sinφ. and x3 = r cosθ. The commutation relationships imply that no simultaneous eigenstates of all three Jk exist.3: Consider a wave function ψ(φ) which depends only on the azimutal angle φ of the spherical coordinate system (r. φ).58) i = r ×p ˆ (5.56) one obtains [J 2 .. k = 1. J3 ] + [J2 . C] = A[B. usually chosen as J3 . We will show now that the properties of spherical harmonics J2Y J3 Y m (θ.e. θ.e. according to the correspondence principle between classical and quantum mechanics. 2. the system related to the Cartesian coordinates through x1 = r sinθ cosφ.61) We demonstrate this property for k = 3. (5.55) can be written as a vector product of r and i − J = −r × From this follows. 5.64) mY m (θ. i. φ) = = 2 ( + 1) Y m (θ. Show that for J3 as deﬁned above holds exp( iα J3 ) ψ(φ) = ψ(φ + α). φ) m (θ. Using [AB. J3 ] = = = 2 2 [J1 + J2 . Equation (5.2. φ) (5.61) simultaneous eigenstates of J 2 and of one of the Jk . These eigenstates are the well-known spherical harmonics Y m (θ. (5. J3 ] J2 −i J1 J2 − i J2 J1 + i J2 J1 + i J1 J2 = 0 According to (5.56) are the famous commutation relationships of the quantium mechanical angular momentum operators. J3 ] + [J1 . φ). J3 ] (5. To show this we ﬁrst note that (5. can be found. Jk ] = 0 . C]B and (5. φ) as well as the eﬀect of the so-called raising and lowering operators J± = J1 ± i J2 (5. C] + [A. 3.

76) We now show that property (ii) also holds for m = − 1 proceeding in a similar way.56)..74) (5. φ) (5. L1 − iL2 ] = L2 + iL1 = i (L1 − iL2 ) = iL− and L3 | = −i | we obtain L3 L− | = ( iL− + L− L3 )| = i(− + 1) L− | . and since we have already shown | − 1 = (iβ −1 )−1 L− | we can state L3 | − 1 = −i( − 1)| − 1 . L− ] = [L3 . 3 be operators acting on a Hilbert space H which obey the algebra [Lk . L+ raises the m-value of | − 1 from m = − 1 to m = . 2. J+ Y J− Y J− Y m (θ. It holds L+ L− | = (L+ L− − L− L+ + L− L+ )| = ([L+ . L− ] + L− L+ )| m = ( −2iL3 + . L2 = L2 + L2 + L2 . For this purpose we prove the following theorem. Using [L3 .70) (5.e.71) To prove this theorem we ﬁrst show by induction that (i) holds. L− ] + L− L3 ) | . Noting [L+ . α −1 β −1 = 2 . φ) (θ.72) = 0 = −i | (5.67) (5. For this purpose we ﬁrst demonstrate that the property holds for m = − 1 by considering (i) for the state L− | ∼ | − 1 .1 Let Lk .75) (5.69) J+ Y (θ.e. An Important Theorem Theorem 1. φ) m+1 (θ. i. L− ] + L− L+ )| . We will show that this property holds then also for m − 1. φ) − m (θ. in fact. The deﬁnitions of the coeﬃcients α −1 and β −1 yield L+ L− | = L+ ( iβ −1 | − 1 = − α −1 β −1 | . The arguments are very similar to the ones used above and we can be brief. k = 1. From this follows that L− | is an eigenstate of L3 . L− ] = [L1 + iL2 . namely. L ] = k m Lm and let L± = L1 ± i L2 . (∗) (5.73) (5. L1 − iL2 ] = −2iL3 and L+ | = 0 one obtains L+ L− | = −2iL3 | = −2 | . φ) 107 = = 0 = = 0 ( + m + 1)( − m) Y ( + m + 1)( − m) Y m+1 (θ. i.68) (5. Continuing our proof through induction we assume now that property (i) holds for m. We note L3 L− | = ( [L3 . We consider L+ L− | m = ( [L+ .66) (5.5. φ) are a consequence of the Lie algebra (5.3: Angular Momentum Operators on the spherical harmonics. 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.

5. yield properties (5.66.77) We will show now that if (ii) holds for m.55).82) (5. (5.79) read L+ | m L− | m + 1 = −i = −i ( + m + 1)( − m) | m + 1 ( + m + 1)( − m) | m . given in (5.72.80) (5. We note that we can write L2 = 1 1 1 1 2 2 2 2 L+ L− + 2 L− L+ + L3 .69).83) (5. (∗∗) (5. This implies that (i) holds for m − 1. in terms of spherical coordinates deﬁned through x1 x2 x3 = = = r sin θ cos φ r sin θ sin φ r cos θ (5. it holds αm βm = 2 one obtains αm βm = = ( + 1) k=m+1 k . ﬁnally αm = βm = (5. We note L3 L− | m = ( [L3 .53–5. 5. (5. We can deduce from (∗) and (∗∗) α α α −1 β −1 −2 β −2 −3 β −3 = 2 = 2( − 1 + ) = 2( − 2 + · · −1 + ) Obviously. from L+ L− | m = − αm−1 βm−1 | m follows the recursion relationship αm−1 βm−1 = 2m + αm βm . Before we can ﬁnally show that property (iii) holds as well for all − ≤ m ≤ we need to determine the coeﬃcients αm and βm . in particular. 5.108 Theory of Angular Momentum and Spin L− L+ )| m = (−2m − αm βm )| m .63. We will have various opportunities to employ the Theorem just derived. i.79) We can now show that (iii) holds for all proper m–values. Angular Momentum Operators in Spherical Coordinates We want to express now the generators Jk . This completes the proof of the theorem. For this purpose we need to express the angular momentum operators in terms of spherical coordinates (r..e.e. θ.64). One can normalize the states | m such that αm = βm . It follows then L | m = − 2 αm−1 βm−1 + 2 αm βm + m | m = 1 − 2 ( + m + 1)( − m) + 1 ( + m)( − m + 1) + m2 = − ( + 1)| m . 5.84) . 2 We note that Eqs. Again the argumemts are similar to the ones used above. A ﬁrst application will involve the construction of the eigenstates of the angular momentum operators as deﬁned in (5. Using the familiar formula + m − n k=0 k = n (n + 1)/2 − (m + 1) m = ( + 1) m − (m + 1) m (5. 5. L− ] + L− L3 )| m = ( iL− − i m L− )| m = −i (m − 1) L− | m from which we can deduce L3 | m − 1 = −i(m − 1)| m − 1 .78) ( + m + 1)( − m) ( + m + 1)( − m) .73.81) i.. φ. it also holds for m − 1.70.

84 ) one obtains. According to (5. using (5. in terms of spherical coordinates.91) which agrees with expression (5. Using (5. φ) ∂x2 /x1 ∂ ∂x3 /r ∂ = x3 + x3 ∂x2 ∂ tan φ ∂x2 ∂ cos θ ∂x2 /x1 /r ∂ ∂ − x2 − x2 ∂x3 3 ∂ cos θ f (θ.3: Angular Momentum Operators 109 We ﬁrst like to demonstrate that the generators Jk . using ∂/∂tanφ = cos2 φ ∂/∂φ and ∂/∂cosθ = −(1/sinθ) ∂/∂θ. φ) (5. ∂φ = sin φ (5.85). In fact.86) (5. involve only the angular variables θ and φ and not the radius r. actually. cos θ ∂ cos2 φ + sin θ cos φ ∂φ cos2 θ sin θ sin φ ∂ sin θ sin φ ∂ + + sin2 θ 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 . φ) sin θ ∂θ sin θ ∂θ L1 f (θ. φ) = ( x3 ∂2 − x2 ∂3 ) f (θ.82–5. (5.92) (5. applying repeatedly the chain rule.87) holds J3 = To determine J 2 we note.89) = ∂ x2 x2 ∂ x2 + x2 ∂ x3 − 33 − x2 1 3 2 x1 ∂ tan φ r ∂ cos θ r ∂ cos θ f (θ.85). We like to express now the operators J3 and J 2 .86).88) (5.85) (5. the latter deﬁned in (5. using (5. we will prove − − − i i i J1 J2 J3 = L1 = L2 = L3 ∂ ∂ + cot θ cos φ ∂θ ∂φ ∂ ∂ = − cos φ + cot θ sin φ ∂θ ∂φ ∂ = − .60). L1 f (θ.87) To derive these properties we consider. φ) ∂x ∂x3 ∂ tan φ =0 (5.90) This yields. φ) = (5. φ) = (x3 ∂2 − x2 ∂3 )f (θ. for example.93) and similarly. (L2 )2 = − cos φ ∂ ∂ + cot θ sin φ ∂θ ∂φ 2 .5. φ). L1 f (θ.

99) = = ∂2 2 ∂ 1 + + 2 2 2 ∂r r ∂r r sin θ ∂2 2 ∂ + 2 ∂r r ∂r sinθ ∂ ∂θ 2 + ∂2 ∂φ2 (5. 5. The operators in the present case are Lk = − i Jk where the Jk are diﬀerential operators in θ and φ given in (5.86. one can conclude J2 = − 2 sin θ 2 sinθ ∂ ∂θ 2 + ∂2 ∂φ2 .101) 1 ∂2 r r ∂r2 and comparision with (5. One ﬁrst identiﬁes the Hilbert space H on which the generators Lk operate. ∂θ2 ∂θ ∂φ (5. i. 2m r ∂r2 2m r2 2 (5.97). 5.95) sinθ ∂ ∂θ = ∂2 ∂ + cotθ . (5.85. 5. sphere S2 .86.97) Kinetic Energy Operator The kinetic energy of a classical particle can be expressed J2 p2 p2 = r + class 2m 2m 2m r2 (5. 5. namely on the space of complex-valued functions on the unit 3-dim.64) can be constructed as follows. These generators act on a subspace of C∞ ( ).98) where Jclass = r × mv is the angular momentum and pr = mr is the radial momentum.96) and the relationship between Lk and Jk as given in (5.87).63. 2 ∂θ ∂θ (5. The ˙ corresponding expression for the quantum mechanical kinetic energy operator is ˆ T = − which follows from the identity 2 2 2 2m = − 1 ∂2 J2 r + . The functions in C∞ (S ) have real variables .87). ∂θ ∂φ ∂2 ∂ ∂2 + cotθ + ( cot2 θ + 1 ) 2 . C∞ (S ).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 .85. 5.e. 1 sin2 θ sinθ ∂ ∂θ (5. 5.4 Angular Momentum Eigenstates According to the theorem on page 107 above the eigenfunctions (5.94) It follows (L1 )2 + (L2 )2 + (L3 )2 = With cot2 θ + 1 = 1/sin2 θ.100) (5.

m) Constructing Y → Y →→ ··· Y ( + m)! (2 )! ( − m)! −1 − We like to characterize the resulting eigenfunctions Y m (θ. φ) . One obtains in this way Y m (θ. φ) = 0 as well as L3 Y (θ.111) . (5. is indeed a Hilbert space. . φ) = −i Y (θ. φ. 0 ≤ φ < 2π. the so-called spherical harmonics. − 2. .108) m (θ. φ). . 0 ≤ θ < π.107) (5. φ) = −i ( + m + 1)( − m)Y m (θ. .102) exists.107) reads ∂ ∂ + i cotθ ∂θ ∂φ P (cosθ) ei φ i e±iφ ∂ ∂ ± i cotθ ∂θ ∂φ . The function space endowed with this norm and including only functions for which the integral (5.64) can be constructed then by seeking ﬁrst functions Y (θ. must be cyclic in φ with period 2π.106) The latter property is obviously satisﬁed for the chosen φ-dependence. (5. (5.87).104. 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. } applying L− Y m+1 (θ. φ) (5.103) iteratively for m = − 1. − + 1. .110) = 0 (5. by carrying out the construction according to (5. and to be admissable for description of quantum states. normalizing these functions.104) (5. φ)f (θ. − . 5. and then determining the family {Y m (θ.5. φ) (5. The eigenfunctions (5. The norm in H is deﬁned through |f |2 = S2 dΩ f ∗ (θ. m) = 1 −m J− 1 2 Y (θ. φ) = ∆( .105) ∆( . m = − . φ) in H which satisfy L+ Y (θ. φ) = 2 + 1 ( − m)! P 4π ( + m)! m (cosθ) eimφ . . φ) = = 0 −i Y (θ.102) where dΩ = sinθ dθ dφ is the volume element of S2 . φ) L3 Y (θ. . 5.4: Angular Momentum Eigenstates 111 θ. The raising and lowering operators L± = L1 ± i L2 . (5.63. can be expressed L± = Accordingly. φ). . φ) (5. φ). (5.85–5.105) explicitly.109) using (5.

is P (cos θ) = N sin θ as one can readily verify.114) dθ sinθ |Y (cosθ)|2 = 1 0 (5. φ) = (−1) 1 1 = 1 [1 · 3 · 5 · · · (2 − 1)]2 (5. Jackson (John Wiley.e. New York. We note that (5.114.” by J.113) 0 is ﬁnite. (5. . using the new integration variable x = cos θ. (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. for example. (5.116) implies |N |2 from which we conclude N = (−1) 1 · 3 · 5 · · · (2 − 1) (5..118) 2 +1 1 1 sin θ ei 4π (2 )! 2 ! φ . One obtains.112) where we employed the deﬁnition of P (cosθ) in (5. 5. i.112 or Theory of Angular Momentum and Spin ∂ − ∂θ cotθ P (cosθ) = 0 (5. one for which π dθ sinθ |P (cosθ)|2 (5.120) See.106.117) Accordingly.114) yield |N |2 2 +1 1 2π 4π (2 )! π dθ sin2 0 +1 θ = 1.106. The normalization factor N is chosen such that π (5. 1975) . 2nd Ed. A proper solution of this equation.116) The integral appearing in the last expression can be evaluated by repeated integration by parts. π +1 dθ sin2 0 +1 θ = −1 +1 −1 dx (1 − x2 ) +1 = = . ”Classical Electrodynamics.107).D.119) provide the following expression for Y Y (θ.115) holds. 5. 5.119) where the factor (−1) has been included to agree with convention1 . . (5.

. namely Y (θ.110) together with (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 . according to (5. . (5.124) holds for m + 1. We want to demonstrate now that the recursion equation (5.106). .119). through repeated application of (5. 5.122) The latter identity can be written.114. employing again the variable x = cos θ. φ).103. and can obtain now.124) called the associated Legendre polynomials. 5.124) reads P (x) = = (2 )! (1 − x2 )− 2 (x2 − 1) 2 ! (−1) 1 · 3 · 5 · · · (2 − 1) sin θ (5.126) Then holds.103. P m (x) = −1 ( + m)! 2 ! ( − m)! × (1 − x2 )− m+1 2 −(1 − x2 ) 2 1 ∂ x + (m + 1) 1 ∂x (1 − x2 ) 2 × (5. φ).123).5. φ). every application of L− reduces the power of eiφ by one. − 2.e. use (5. (5.4: Angular Momentum Eigenstates 113 We have constructed a normalized solution of (5. For m = (5. we seek to determine the functions P m (cos θ) and.64). m (cos θ) e (5.110).124) we proceed by induction.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) . are real. the eigenfunctions Y m (θ. 5. 5. P m (x) = −1 × ( + m + 1) ( − m) × −(1 − x2 ) 2 1 (5.123) P m+1 (x) ∂ x + (m + 1) 1 ∂x (1 − x2 ) 2 . P m+1 (x) = −m−1 m+1 ∂ 1 ( + m + 1)! (1 − x2 )− 2 (x2 − 1) .124). Actually. indeed. m = − 1. . Let us then assume that (5. as speciﬁed in (5.121) This expressions shows that the factor eimφ . therefore. ∂x −m−1 .123) leads to the expression P m (x) = −m m ∂ 1 ( + m)! (1 − x2 )− 2 (x2 − 1) 2 ! ( − m)! ∂x −m (5. The reader should note that the associated Legendre polynomials.63. 2 ! ( − m − 1)! ∂x −m−1 (5. describes the φ-dependence of Y m (θ.125) which agrees with (5.127) ∂ −m−1 (x2 − 1) . − deﬁned in (5. To prove (5. i.

hence. φ) using (5. φ).124) and.129) holds. by repeated application of the operator L+ .106. and constructs then the eigenfunctions Y − +1 .106. φ) (5.133) Since the term with the highest power of (x2 − 1) is x2 .. Y − +2 . it holds then for all m. 2 ! (2 )! ∂x2 1 (5. the derivative ∂ 2 +1 /∂x2 +1 of this expression vanishes.132). appears not very interesting.134) (x) = 1 2 ! (1 − x2 ) 2 . P − (5.135) . it holds ∂2 (x2 − 1) = (2 )! ∂x2 and.127) reproduces (5. if L− Y (θ. φ) = 0 (5. from such construction emerges an important symmetry property of Y m (θ.123) terminates for m = − .124) and.132) which reduces the number of spherical harmonics which need to be evaluated independently roughly by half . (5. φ) An alternative route to construct the eigenfunctions Y of L− f (θ.124) holds for m = .124). φ) = Y − (θ. yields a vanishing expression for m = − − 1 as long as is a non-negative integer. namely.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. we embark on this construction in order to prove (5. i. We ﬁrst determine Y − (θ.131) identiﬁes f (θ. therefore. P − (x) = 1 ∂2 (1 − x2 ) 2 (x2 − 1) . Since (5. Y ∗ (θ. Such construction reproduces the eigenfunctions Y m (θ. In fact.e. φ) L3 f (θ. It holds.128) one can show that (5. φ) = (−1)m Y m −m (θ. therefore. expression (5. Constructing Y − → Y − +1 → ···Y m (θ. (5. 5.124).130) (5. φ) choosing the proper sign. that expression (5. φ) as given in (5. However. therefore. φ) . etc. In that case (1 − x2 ) is a polynomial of degree 2 and. using x = cos θ. 5. which is equivalent to recursive application of L− . therefore. We want to test if the recursion (5.124) holds for m if it holds for m + 1. φ) = = determines ﬁrst a normalized solution 0 i f (θ.

5.136) to construct all other Y recursion eiφ ∂ ∂ + i cotθ ∂θ ∂φ Y m (θ.4: Angular Momentum Eigenstates Due to (5.106). (5. one can conclude for the associated Legendre polynomials ( − m)! ( + m)! = or P m+1 (cos θ) ∂ − m cotθ ∂θ P m (cos θ) ( + m + 1)( − m) ( − m − 1)! P ( + m + 1)! m+1 (cos θ) (5.143) . the expression (5.123)] P m+1 (x) = −(1 − x2 ) 2 1 ∂ x − m 1 ∂x (1 − x2 ) 2 P m (x) .142) yields P − (x) = (−1) 1 (1 − x2 )− 2 (x2 − 1) = (1 − x2 ) 2 2 ! 2 ! (5.122.124).110) applies the = ( + m + 1)( − m) Y m+1 (θ.136) We note in passing that this expression and the expression (5.120) for Y obey the postulated relationship (5. (5.138) Employing (5.129).139) = ∂ − m cotθ ∂θ P m (cos θ) . i. φ) . (5.e. φ).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. According to (5..142) For this purpose we proceed by induction. We ﬁrst note that for m = − (5.137) m (θ. a sign consistent with the family of functions Y m constructed above. (5.f.136) is normalized and has the proper sign. φ) = 2 +1 1 1 sin θ e−i 4π (2 )! 2 ! φ 115 − . following closely the proof of eq. (5. This follows from the identity ∂ ∂ − i cotθ ∂θ ∂φ One can use (5.80) and (5.140) Introducing again the variable x = cos θ leads to the recursion equation [c. φ) sin θ e−i φ = 0.106.5.124) one arrives at Y (θ. (5. Obviously.132). 5. One can readily verify that this expression provides a solution of (5. (5.

h.116 Theory of Angular Momentum and Spin wich agrees with (5.142) one can state P (x) = ∂ (x2 − 1) . Since (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 .147) the r. Hence.. P (x) = P 0 (x) are called Legendre polynomials.124).135).. 2 ! ∂x 1 (5.142) holds for m = − we veriﬁed that it holds for all m. 1 (1 − x ) ∂x +m (1 − x2 ) 2 1 (5.142). 2 ! ∂x +m (5.142) holds for m + 1 if it holds for m.s.124) and (5.s.144) ∂ x − m 1 ∂x (1 − x2 ) 2 +m m ∂ (−1)m (1 − x2 ) 2 (x2 − 1) .149) (5. Y −2 . According to both (5.e. one notes that application of (5. (5.141) reads then P m+1 (x) = 1 (5. φ) ( − m)! P ( + m)! −m (x) . (5.145) Hence. also the associated Legendre polynomials determined this way. 2 ! ∂x +m+1 (5. are identical.132). 2 ! ∂x +m −(1 − x2 ) 2 Replacing in (5. We then assume that (5.148) According to (5. Accordingly. (5.h. given by (5. of which agrees with the r. (5.142) and by (5. one can conclude P m (x) = (−1)m m (θ. etc. equivalently. However.144) reads P m+1 (x) = +m+1 m+1 ∂ (−1)m+1 (1 − x2 ) 2 (x2 − 1) . yields the same eigenfunctions as the previous construction beginning with Y and stepping down the series of functions Y −1 .106) this implies for Y The Legendre Polynomials The functions the identity (5.e.150) . of (5. etc.124) yields (−1)m ( − m)! P ( + m)! −m (x) = +m m ∂ (−1)m (1 − x2 ) 2 (x2 − 1) .128) m + 1 → −m or.142) holds for m. i. The construction beginning with Y − and continuing with Y − +1 . Y − +2 .146) i.

The spherical harmonics for m < 0 can be obtained using (5.” by J. . t) is called a generating function of the Legendre polynomials3 . m ≥ 0 ∂xm (5. t = r2 /r1 . 45 of “Special Functions and their Applications” by N.153) P (cos θ) eimφ . In case q = 1 and |r2 | < |r1 | holds the identity2 ∞ 1 r2 = P (cos γ) (5. New York.152) and. 2.. r1 is the point where the potential is measured.158) see. Englewood Cliﬀs.150) and (5.N. .154) |r1 − r2 | r +1 =0 1 where γ is the angle between r1 and r2 .154) can be written 1 = w(x. and |r1 − r2 | = r1 (5. t) = √ 1 − 2xt + t2 ∞ 1 − 2 x t + t2 . pp. t) = √ Comparision with (5. 1. Using x = cos γ.156) yields P (1) = 1 . P1 (x) = x .. For example.157) = 0. N.156) w(x. . 92 To prove this property see. in case of x = 1 holds w(1. for example.D.g. (5. =0 (5. φ) = 2 + 1 ( − m)! (−1)m sinm θ × 4π ( + m)! × ∂ ∂ cos θ m (5. 2 3 1 1 = = 2 1−t 1 − 2t + t ∞ t . the spherical harmonics for m ≥ 0 Y m (θ. Jackson (John Wiley. 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. 1965) which is an excellent compendium on special functions employed in physics. accordingly. e.132). (5.151) (5x3 Comparision of (5. and r2 denotes the location of the charge. The generating function allows one to derive useful properties of the Legendre polynomials. Lebedev (Prentice-Hall. (5. pp.4: Angular Momentum Eigenstates The ﬁrst few polynomials are P0 (x) = 1 . . ”Classical Electrodynamics. 2nd Ed.5.142) 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) .J. =0 (5.155) P (x) t . P3 (x) = 1 2 117 P2 (x) = − 3x) 1 2 2 (3x − 1) . 1975).

If the spherical coordinates of r are (r.118 For w(x. φ) Under inversion at the origin vectors r are replaced by − r.167) (5. θ. φ) m (θ. 1. φ) . then the coordinates of − r are (r. Due to cos(π − θ) = − cos θ the replacement r → −r alters P m (x) into P m (−x). Noting exp[im(π + φ)] = (−1)m exp(imφ) we determine Y Properties of Y m (θ.168) mY m (θ. φ) . π + φ).151)] allows one to determine P (x) for = 1.164) which.142) allows one to conclude P m (−x) = (−1) +m P m (x) (5. t) (1 − 2xt + t2 ) Employing (5.f. 2. ∂t 1 − 2xt + t2 (5. φ). holds P1 (x) ( + 1) P +1 (x) = = x P0 (x) (2 + 1) x P (x) − P −1 (x) (5. ∂t P (x)t −1 −1 (5. φ) goes over to Y m (π − θ. Accordingly.166) We want to summarize the properties of the spherical harmonics Y derived above. Inversion Symmetry of Y m (θ. Accordingly. (5. t) x−t = . ..163) (5. φ) = = 2 ( + 1) Y m (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. under inversion Y m (θ.161) Collecting coeﬃcients with equal powers t yields 0 = P1 (x) − x P0 (x) ∞ (5. t) holds Theory of Angular Momentum and Spin ∂lnw(x.165) since ∂ n /∂(−x)n = (−1)n ∂ n /∂xn .159) Using ∂lnw/∂t = (1/w) ∂w/∂t and multiplying (5. π + φ) = (−1) Y m (θ. π − θ.156) and (∂/∂t) to P (x)t = ∞ ∂w + (t − x) w = 0 . using P0 (x) = 1 [c. The spherical harmonics are eigenfunctions of the angular momentum operators J2Y J3 Y m (θ. φ) m (π − θ.160) this diﬀerential equation is equivalent ∞ (1 − 2xt + t2 ) =0 P (x) t + (t − x) =0 P (x) t = 0 (5. φ) m (θ. π + φ). . . (5. . Inspection of (5. φ) (5.159) by w(1 − 2xt + t2 ) leads to the diﬀerential equation obeyed by w(x.

176) = 1 +1 [ (2 + 1) x P (x) − are the Legendre polynomials. m (5.174) m ≥ 0 P (cos θ) eimφ . except for the factor exp(imφ). The spherical harmonics Y 0 are Y 0 (θ. φ) where J± = J1 ± iJ2 . The spherical harmonics are given by the formula Y m (θ. The spherical harmonics form.170) (5. The spherical harmonics for m < 0 are given by Y −m (θ.172) (5.e. For the Legendre polynomials holds the orthogonality property +1 dx P (x) P ( (x) = −1 2 δ 2 +1 . where ( = 1.171) m = dφ Y ∗ (θ. P +1 (x) P1 (x) = x . 5. φ) Y m (θ. φ) . in fact. φ) = δ δm m . φ) = 2 +1 P (cos θ) . The spherical harmonics form an orthonormal basis of the space C∞ (S2 ) of normalizable. φ) ∈ C∞ (S2 ) holds ∞ f (θ. φ)f (θ.4: Angular Momentum Eigenstates 119 2. The spherical harmonics obey the recursion relationships J+ Y J− Y m (θ. 4π (5. φ) . (5. φ) 2π 0 (5. φ) (5. 2. P −1 (x) ] (5. . 6. m 4.5.175) (5. uniquely deﬁned functions over the unit sphere S2 which are inﬁnitely often diﬀerentiable π 2π sin θdθ 0 0 dφY ∗m (θ. φ) = = ( + m + 1)( − m) Y ( + m + 1)( − m) Y m+1 (θ. a complete basis of C∞ (S2 ). . i.173) m+1 (θ.178) 7.. φ) C = =0 m=− π C sin θdθ 0 m Y m (θ.179) . (5.) P0 (x) = 1 . for any f (θ.177) Note that the spherical harmonics are real.169) 3. φ) = (−1)m Y ∗ (θ. φ) m (θ. φ) = 2 + 1 ( − m)! (−1)m sinm θ × 4π ( + m)! 2 ! × ∂ ∂ cos θ m (5. .

181) = 0.h.4. The spherical harmonics for m (θ. The spherical harmonics for θ = 0 are Y m (θ Theory of Angular Momentum and Spin = 0.87). (5.187) h(r) Y m (θ.86). For the Laplacian holds 2 = = = (5. (5. (5.189) evaluate the integral on the l. 1. For this purpose start from the identity +1 ∞ +1 dx w(x. expand the result in powers of t and equate the resulting powers to those arising on the r. 2 are given by Y00 = 1 √ 4π 3 sin θ eiφ 8π 3 cos θ 4π 1 4 − 15 sin2 θ e2iφ 2π 15 sin θ cos θ eiφ 8π 5 4π 3 1 cos2 θ − 2 2 (5. .1: Derive expressions (5.h. φ).4.189). 4π (5. (5. t) stated in (5. t)2 = −1 .182) Y11 Y10 = = − (5. 11.177). The spherical harmonics obey the inversion symmetry Y 10. φ) . φ) = (−1) Y m (π − θ. of (5. φ) = δm0 2 +1 .183) (5.3: Construct all spherical harmonics Y3m (θ.180) 9. π + φ) .2: Derive the orthogonality property for the Legendre polynomials (5.184) Y22 Y21 Y20 together with (5.185) (5. Exercise 5.188) Exercise 5.s.s..179) using the generating function w(x. Exercise 5.186) (5.120 8.4. φ) = 1 ∂2 r − r ∂r2 ( + 1) r2 h(r) Y m (θ. =0 t + −1 dxP (x) P (x) .156).

are characterized completely if one speciﬁes the transformation of any Y m (θ.5. = . φ ) exp m One can also represent C∞ (S ) C∞ (S ) = [ {Y . .. exp − i ϑ · J . φ) m. These coeﬃcients can be considered the elements of an inﬁnite-dimensional matrix which provides the representation of exp − i ϑ · J in the basis {Y m .e.190) provide a complete. We denote the image of a function f (θ.5 Irreducible Representations We will consider now the eﬀect of the rotational transformations introduced in (5. . φ) (5. φ) i − ϑ·J f (θ. . φ) c = . with elements given by (5. m = S2 i dΩ Y ∗m (θ. i. ∞.e. φ) . φ) (5.193) shows ∞ = − .. .... exp − i ϑ · J (X ) = X . φ) D(ϑ) = .195).5: Irreducible Representations 121 5. assumes a particularly simple form. The transformations exp − i ϑ · J . . 1. . φ) ∈ C∞ (S ). }. = − .42).. therefore.m m m (θ. = 0. . φ) exp − ϑ · J ...− + .196) C∞ (S ) = = X .194) Y m (θ.192) = S2 dΩ Y ∗ (θ . i. on functions f (θ. . . = .193) where [ { } ] denotes closure of a set by taking all possible linear combinations of the elements of the set.. m . m = − . φ). φ) ˜ under such rotational transformation by f (θ. (5. (5. and (ii) they form the lowest dimensional sets X obeying C∞ (S ) = X which have this invariance property. − + 1. . φ) ˜ Y m (θ. . }] (5. ˜ f (θ..m D(ϑ) m.197) The subspaces X have the important property that (i) they are invariant under rotations exp − i ϑ · J .191) i − ϑ·J f (θ . . (5. We like to argue that the matrix representing the rotational transformations of the function space C∞ (S ). }] .. called the irreducible representation. . ∞. . Comparision with (5. (5. φ ) (5.e. if one speciﬁes the functional form of the coeﬃcients [D(ϑ)] m ..− + . m Y m (θ.195) i. For this purpose we consider the subspaces of C∞ (S ) X = [ {Y . . φ) = exp Since the spherical harmonics Y space C∞ (S ) one can expand ˜ f (θ. orthonormal basis for the function c m Y m (θ. .

n2 . . One = contribute can conclude then that in the expansion (5.200) in (5.65). one concludes for the matrix elements (5. φ) exp m i − ϑ·J Y m (θ. one can conclude that J+ leaves X invariant.. J± ] = ± J± (5. Similarly. in case m = − .194) only spherical harmonics with or.n3 J+ 1 J− 2 J3 3 . i.e.e. . 5.198) 2 2 where ϑ± = ϑ1 ( 1 ϑ− J+ 2 + 1 2 ϑ+ J− ϑ2 . (5.199) allows one to express exp i − ϑ·J = n1 . . and J3 1 1 ϑ · J = ϑ− J+ + ϑ+ J− + ϑ3 J3 (5. Application of (5.200) leaves X invariant. which follow readily from (5. .n2 .199) [J+ .197) of C∞ (S ) the matrix representation of D(ϑ) assumes a block-diagonal form. The exponential operator exp − i ϑ · J + ϑ3 J3 )n .. m = δ S2 dΩ Y ∗ (θ. leaves X invariant. 3.. φ) . with blocks of dimension 1. 5. J− ] = 2 J3 .. i. .n3 n n n Accordingly. . (5.e. If one orders the basis according to the partitoning (5. One proceeds by expanding ﬁrst ϑ · J in terms of J+ . i. Since any additive term n n n cn1 .200) of the rotational transformations one can also conclude that the X are the smallest invariant subspaces of C∞ (S ). one needs to consider the action of J+ 1 J− 2 J3 3 on the subspaces X .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 .56.202) D(ϑ) = (2 + 1) × (2 + 1) . can be expanded in powers One employs then the commutation properties [J3 . J− 2 also leaves n1 X invariant. n The action of J− on Y m produces Y m−1 or. equivalently. Since the Y m are n3 eigenfunctions of J3 the action of J3 reproduces the basis functions of X .. 1×1 3×3 .195) [D(ϑ)] m. produces 0. (5.. the transformation exp − i ϑ · J leaves X invariant as well.201) From expression (5. J− .

(d) Show that the similarity transformation T (ϕ) deﬁned in the space of non-singular. this parametrization. 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). 2.5. k = 1.6 Wigner Rotation Matrices We will now take an important ﬁrst step to determine the functional form of the matrix elements of D(ϑ). other representations are termed reducible representations. (b) Prove that the elements of SO(2) can be completely characterized through a single parameter. is a group. (iii) a third rotation around the new x3 –axis by an angle γ. The representation is referred to as the irreducible representation. i. Exercise 5. 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. 3 certainly allow one to describe any rotation around the origin. for which holds RT R = R RT = 1 1.6: Wigner Rotation Matrices 123 This representation of D(ϑ) is the one wich has blocks of the lowest dimensions possible.1: Representations of the group SO(2) SO(2) is the set of all 2×2 matrices R which are orthogonal. does not provide the simplest mathematical description of rotations. Determine the generator of ρ(ϕ) in analogy to (5. (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).e. (f) The transformation ρ(ϕ) as deﬁned in (e) leaves the following subspace of functions considered in (e) Xm = { f (α) = A e−imα .5. though seemingly natural. (e) In the space C∞ ( ) of inﬁnitely often diﬀerentiable. A ∈ C. This step reconsiders the parametrization of rotations by the vector ϑ assumed sofar.22. (ii) a second rotation around the new x2 –axis by an angle β. and for which detR = 1. 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). . Give the corresponding expression of ρ(ϕ). However. 5. i. 5.e. and periodic functions f (α). f (α + 2π) = f (α). The three components ϑk . ∈ N} invariant. (a) Show that SO(2) with the group operation ◦ deﬁned as matrix multiplication.49).

φ) (5. φ) [D(α.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. we can write e− i −1 = S eA S −1 . β.207) are equivalent.205) Y m (θ. in terms of rotations deﬁned with respect to the original frame. m [D(α.211) Using (5. i (5. the axis x3 is deﬁned in the coordinate frame which is related to the original frame by the consecutive rotations (i) and (ii).194. γ)] m. the l. the transformation from the rotated frame to the original frame.210) The third rotation in (5.195) by ˜ Y m (θ. (5.203) can be expressed. in analogy to (5.208) αJ3 βJ2 = e− αJ3 e− i βJ2 e i (5. β. β. φ) = S2 dΩ Y ∗m (θ. We will demonstrate that (5. φ) e− γJ3 − i βJ2 − i αJ3 e e . m i Y m (θ. and the r. γ)] = . γ ∈ R such that exp i − ϑ·J = e− i γJ3 − i βJ2 − i αJ3 e e (5.209) The ﬁrst two rotations in (5. Accordingly.209).203) For any ϑ ∈ R one can ﬁnd Euler angles α.h. i.124 Theory of Angular Momentum and Spin The angles α. followed by J2 in the original frame. followed by transformation (i).e.m m. one can replace (5. For this purpose we notice that J2 can be expressed through the similarity transformation J2 = e− i αJ3 J2 e i αJ3 (5.e.s. can then be written e− i βJ2 − i αJ3 e = e− i αJ3 e− i βJ2 (5.206) The expression (5. γ will be referred to as Euler angles. (i) and (ii).212) .. The axis x2 is deﬁned in the coordinate frame which is related to the original frame by rotation (i). i. i. For any similarity transformation involving operators A and S holds eS A S Accordingly.203). 5. Obviously. is e− i γJ3 = e− i βJ2 e− i αJ3 e− i γJ3 e i αJ3 e i βJ2 (5.h. the transformation from the original frame to the rotated frame. (5. The Euler rotation replaces exp − i ϑ · J by e− i γJ3 − i βJ2 − i αJ3 e e . however.207) which replaces J2 by the inverse transformation (i).e. of (5.204) is satisﬁed.203) has the disadvantage that it employs rotations deﬁned with respect to three diﬀerent frames of reference. β.203)..s..

φ) e− αJ3 e− βJ2 Y m (θ. 5. φ) (5.309. m ( ) S2 dΩ Y ∗ (θ.7: Spin 1 2 and the group SU(2) 125 Multiplication from the left with (5. to redeﬁne the three rotations in (5. γ)] = .7 Spin 1 2 and the group SU(2) The spin describes a basic and fascinating property of matter. φ) e− m i βJ2 Y m (θ.205.210) yields the simple result e− i γJ3 − i βJ2 − i αJ3 e e = e− i αJ3 e− i βJ2 e− i γJ3 . Examples are the other ﬁve members of the lepton family to which the electron belongs.203) with respect to the original (unprimed) frame one simply needs to reverse the order of the rotations.m m.220) We will derive below [see Eqs.219). φ) e−imγ . the electron e and the electron neutrino νe of the ﬁrst generation. m = e−iαm S2 dΩ Y ∗ (θ. φ) i i (5.213) i. β. m [D(α. in addition. φ) = S2 dΩ f (θ..214) e− γJ3 = S2 dΩ Y ∗m (θ. γ)] m.216) Using. φ) = e−im α S2 dΩ Y ∗ (θ. γ)] m. φ) [D(α. the self-adjointness of the operator J3 one can state dΩ Y ∗ (θ.5. φ) e− S2 i γJ3 Y m (θ.215) The evaluation of the matrix elements (5. φ) e− m S2 i αJ3 f (θ.219) = e−iαm dm m (β) e−iγm δ ( ) .64) yields dΩ f (θ.218) Deﬁning the so-called Wigner rotation matrix dmm (β) = one can express the rotation matrices [D(α. (5. The eigenvalue property (5. φ) e− m i βJ2 Y m (θ.217) Accordingly. Best known is the spin of the electron. m i Y m (θ. one can write [D(α. β.e. β. (5. 5. (5. φ) .206) by ˜ Y m (θ. φ) e−iγm δ (5. This allows one to express the rotational trasnformation (5.310)] an explicit expression for the Wigner rotation matrix (5. φ) f (θ. γ)] m. 5. β. m (5.215) beneﬁts from the choice of Euler angles for the parametrization of rotational transformations. but many other elementary components of matter are endowed with spin-like properties. φ) . the muon µ and its neutrino νµ of the second . (5. φ) Y m (θ. (5.

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

it might hold aj bk − bk aj = 0. According to this property the Pauli matrices generate a 3dimensional Cliﬀord algebra C3 .5.31) of the group SO(3).229) and avoids commuting the components aj and bk . At this point we will state a useful property.231) . 2 S2 = 1 σ2 . 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.227. One can readily verify the commutation property σj σk − σk σj = [σj . they are associated with the important fermion property of matter.k=1 σ σ 3 j k j.e.. (5. ∈ R . (5.k=1 j<k + σ j σ k aj bk .k=1 σ σ aj bk + j>k aj bk = 3 j. y. Cliﬀord algebras play an important role in the mathematical structure of physics. k = 1. 2 S3 = 1 σ3 2 (5.223) where σk .g. 1 σj σk = − σk σj for j = k which can also be readily veriﬁed.229) = 1 . b are vectors commuting with σ. namely. σ2 = 0 −i i 0 .227) which is essentially identical to the Lie algebra (5.226) The Pauli matrices satisfy very special commutation and anti-commutation relationships. the simplest choice being S1 = 1 σ1 . The proof of this relation rests on the commutation relationship and the anti-commutation relationship (5. (σ · a) (σ · b) = a · b 1 + iσ · (a × b) 1 (5. (5. σ3 = 1 0 0 −1 . 2.228) (5. x. 1 i. hermitian 2 × 2–matrices A can be expandend. but their comonents must not necessarily commute with each other. it holds.224) provide a basis in terms of which all traceless.7: Spin 1 2 and the group SU(2) 127 exist three such linear independent matrices. Any such A can be expressed in terms of three real parameters x. i. In fact. one obtains (σ · a) (σ · b) = 3 j 2 j=1 (σ ) aj bj 3 j k j. A = x σ 1 + y σ 2 + z σ3 .230) where a. σk ]− = 2i jk z x − iy x + iy −z . 5. σk ]+ = 2δjk 1 .225) σ (5. σj 2 (5. 3 are the Pauli matrices σ1 = 0 1 1 0 .e. e. z. z A = In fact. (5.224) Algebraic Properties of the Pauli Matrices The Pauli matrices (5. y. The Pauli matrices obey the following anti-commutation properties σj σk + σk σj = [σj .

The two remaining terms yield using σ j σ k = −σ k σ j for j = k and altering ‘dummy’ summation indices 3 j. 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 . For this purpose we choose to replace (5. spin changes sign. only a 720o rotation leaves the spin invariant. let say around the x2 –axis by 360o . 1 (5.k=1 (σ σ j>k σ k σ j (aj bk − ak bj ) = (5. through the algebra obeyed by the operators Sk [Sk . (5. S ] = i k m Sm (5. i.227) leads to 3 3 jk j.k=1 j>k σ j σ k (aj bk − ak bj ) − 1 2 3 j k j.8 Generators and Rotation Matrices of SU(2) Sofar there is nothing which relates the transformations U . This relationship emerges.h.230).238) S 2 χ± = . In the special case a = b ∈ R holds (σ · a)2 = a 2 1 .222) by the Euler form exp(−iγS3 ) exp(−iβS2 ) exp(−iαS3 ) .222) as a rotation vector and you rotate once.233) This result together with (5. hence. (5. The commutation property (5. Well almost like it. however.237) + 1)1 and.231) proves (5. (5.234) 5. however.k=1 j>k 1 2 3 j.e.s.k=1 j>k σ j σ k (aj bk − ak bj ) = 1 2 3 j. yields a · b. since there is a slight diﬀerence: when you interprete the 3-component ϑ in (5. (5.128 Theory of Angular Momentum and Spin Using (σ j )2 = 1 the ﬁrst term on the r. Noticing the idempotence of the Sk ’s 2 Sk = 3 one obtains S 2 = 4 1 = 1 1 1 2(2 1 4 1 0 0 1 (5. 1 1 1 ( + 1) χ± 2 2 .222).k=1 j>k i σl (aj bk − ak bj ) = i =1 σ a×b .236) The matrix elements of this 2 × 2 operator can be labelled by the basis states χ+ and χ− .232) − σ k σ j ) (aj bk − ak bj ) . 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. We like to derive this result now by evaluating the transformations of SU(2) explicitly.235) which is identical to that of the generators of SO(3). S3 χ± = ± 1 χ± 2 . to rotations in space.

i. . One cannot consider the 1 entities carrying the spin– 2 to be particles in the ordinary sense for it can be shown that spin– 1 2 particles have fermion character. i. ..242) The expressions in brackets [.5. m = − . For this purpose one needs to determine the powers of −iβS2 . (5. rotation by 360o changes the sign of the spin state.220) are [D(α. . namely. that states | m for higher quantum numbers = 0. The complete matrix elements (5.245) 5. following Jourdan and Schwinger. − + 1.] can be identiﬁed with the Taylor expansion of the cos. In order to determine the Wigner matrix in (5. use the label 2 2 | 1 ± 1 for the states χ± . 1 . .9 Constructing Spin States with Larger Quantum Numbers Through Spinor Operators In this section we like to demonstrate.239) in the notation (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 (5.244) i. 1.240) . (5.236) 1 2m | exp(−iγS3 ) exp(−iβS2 ) exp(−iαS3 ) | 1 m 2 1 2m = e−iγm | exp(−iβS2 ) | 1 m e−iαm 2 (1) 2 = e−iγm dmm (β)e−iαm (5.e. no two such particles can exist in the same state. (5. 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.e.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). 3 . . . (5. We may.239) where we made use of the notation (5. identical particles any number of which can exist in the same state χ+ and χ− . 2 2 We obtain with this notation for the transformations (5.239 ) one expands the exponential operator exp(−iβS2 ). .219). γ)] m.219. m = cos β e−i( 2 + 2 2 γ α sin β ei(− 2 + 2 2 α γ − sin β ei( 2 − 2 2 α γ cos β ei( 2 + 2 2 α γ .e. 2. . The idempotence property of Sk yields particularly simple expressions for these powers. therefore.243) We note the property of this rotation matrix 2 dmm (2π) (1) = −1 1 . β.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 . ( − β S2 )2n = (−1)n ( − β S2 )2n+1 = (−1)n β 2 β 2 2n 1 1 0 1 −1 0 2n+1 (5. .5.

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. b± |Ψ0 = 0 (5. b ζ The States |Ψ(j. also operators of the type x b† = x+ b† + x− b† + − . b ζ = b† . . b† |Ψ0 = χ− − .e. i. i. (5. b ζ = 0 b ζ.248) bζ . 3 .250) which for x∗ x = x∗ x+ + x∗ x− = 1 represent spinor operators in an arbitrary reference system.e.248) that bζ relationships † † b ζ. m) † and bζ obey the commutation (5. . 1 . + − For example. b− † = − sin β † β b + cos b† .247. b− )T as spinor creation and annihilation + − operators. j = 0. (5. b ζ = b ζ. 2 + 2 − † (5. .253) = δζζ . . These operators are associated with a given spatial reference system.252) (5. b † ζ ζ =0 (5. 1. We consider. therefore. −) bζ .246) The corresponding adjoint operators are denoted by b+ and b− . ζ = +. 2 2 . −j + 1. The boson character of the operators b† = (b† .247) (5.251) One can show using this property and (5. x∗ b = x∗ b+ + x∗ b− + − (5.254) We will show below that these states are orthonormal and form spin states with higher quantum numbers. b † ζ For the vacuum state |Ψ0 holds = δζζ . b− )T is expressed through the commutation relationships + − (ζ. m) = b† − j−m (j + m)! (j − m)! |Ψ0 . .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± .243) the creation operators in a coordinate system rotated by an angle β around the y–axis are b+ † = cos β β † b + sin b† 2 + 2 − .249) In the following will refer to b† = (b† . . . b† )T and b = (b+ . b† |Ψ0 = χ+ + .5. b† )T and b = (b+ . j. using (5. m = −j.

10 Algebraic Properties of Spinor Operators We want to establish ﬁrst a few important and useful algebraic properties which result from the commutation relationships (5.259) The property is obviously true for n = 0. + b+ . . in b† and b† . We will show that for this commutator holds n b+ . b† + − well-known for the quantum mechanical harmonic oscillator. (5.248) we obtain b+ .256) and (5. We will derive the result.10: Algebraic Properties of Spinor Operators 131 5. (5.257) motivate us to determine the commutator (5. (5. If it is true for n then using (5.258) Equations (5.247. b† + n = n cn b+ .255) In order to determine how this operator is modiﬁed by multiplication from the left by b+ we note b+ f (b† ) |state = + b+ .248) and from (5.254). b† + n . + + (5.249). f (b† ) + (5.249) b+ f (b† ) |Ψ0 = + b+ .257) which for (5. b† + b+ . b † + = n b† + n−1 b† + b † + + n = (n + 1) b† + . that the annihilation operators play a role similar to the diﬀerential operator in calculus. b† + = n b† + n−1 . f b† Obviously. 5. f (b† ) |state + f (b† ) b+ |state . 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+ . b† + n b† + b† + b† + + n n n b+ . For this purpose we consider the following polynomial of creation operators N j+m j−m f (b† ) = + n=0 cn b† + n .256) In the special case |state = |Ψ0 this reads using (5.260) .5.255) is b+ . An example + − is the monomial b† which generates the states (5. possibly inﬁnite power series. we need to evaluate b+ . f (b† ) |Ψ0 .

m) = x− b† − j−m (j + m)! (j − m)! |Ψ0 . f (b† ) + where f (x) = df dx . (5. In order to derive (5. (5. f (x)] = ∂k f (x) . m) We want to prove now the property ∞ j exp xb † |Ψ0 = 1 j=0.266) where xb† has been deﬁned in (5.257) we can conclude in particular bζ f (b† . m) φjm (x) = j−m xj+m x− + (j + m)! (j − m)! . b† ) |Ψ0 = + − ∂ ∂b† ζ f (b† .. f (b† . one can prove b− . 2 . f (b† ) − = f (b† ) − (5. By induction we can conclude that (5. b† ) + − = ∂ ∂b† ζ f (b† . b† ) in b† and b† holds (ζ = +. m) ..267) exp xb† |Ψ0 is called a generating function of |Ψ(j. the property holds also for n + 1.263) 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 . One can ﬁnally conclude for polynomials (5.255) b+ .268) . m) m=−j (5. φjm (x) Ψ (j.132 Theory of Angular Momentum and Spin i. (5.250) and where φjm (x) represents the function of the two variables x+ and x− closely related to |Ψ(j.265) Generating Function of the States |Ψ(j.263) According to (5.262) Because of the commutation relationships (5.259) holds for all n ∈ N. b† ) . b† ) |Ψ0 . −) + − + − bζ .247) the more general property for polynomials f (b† . Equation (5.. + − (5.e. = f (b† ) + (5.266) we compare the terms x+ b† + j+m φjm (x) Ψ (j.264) This demonstrates the equivalence of the spinor operators bζ and the derivative operation.261) Similarly. f (x)] g(x) = (∂k f (x))g(x) or [∂k . + − (5.1.

. 2 ∞ 1 xb† 2j! u 2j |Ψ0 = u=0.275) = exp xb† |Ψ0 ..270) s x+ b† + N −s x− b† − s |Ψ0 . 1 .273) Change of the summation indices allows one to conclude j φjm (x) Ψ (j..269) (5. m) 1 j=0. 2. 1 . chosing the summation index j = 0. .. 1.1.5.10: Algebraic Properties of Spinor Operators with the s–th term in the binomial expansion of (a + b)n n! an−s bs . 3 ..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. one obtains φjm (x) Ψ (j.274) Summing this expression over 2j = 0. m) = = |Ψ0 (N − s)! s! N −s 1 N! x+ b† x− b† + − N ! (N − s)!s! N −s 133 (5. i.e. 1 xb† u! |Ψ0 (5.272) The summation over s can be written in terms of j and m using (5. . (2j)! 2j |Ψ0 (5... 2 .. .1. m=−j = j=0.. (5. . 2 2 leads to ∞ j ∞ φjm (x) |Ψ (j. (5.270) N 2j j → s=0 j−m=0 → m=−j .1.. m) m=−j = = 1 x+ b† + x− b† + − (2j)! 2j 1 xb† |Ψ0 . 2.. s! (n − s)! Deﬁning j+m = N −s j−m = s. 1. . .

134 Theory of Angular Momentum and Spin which concludes our derivation.s. (5.j .284) . that Ψ (j. m holds.277) To evaluate this expression we ﬁrst notice † e(xb ) † = e(x ∗ b) . b− ) holds † † f (b+ .264) implies † † bζ e(yb ) |Ψ0 = yζ e(yb ) |Ψ0 .281) and. m) |Ψ j . y− )e(yb ) |Ψ0 (5.282) where we deﬁned x∗ y = x∗ y+ + x∗ y− . b† ] = 0 allow one to state that for + − + any polynomial f (b+ .249) for any non-vanishing integer s holds + − Ψ0 |bs Ψ0 = 0 and.e. m) |Ψ j .254) are orthonormal. m) and will be used for this purpose below.h.279) One can generalize this to † † s (bζ )s e(yb ) |Ψ0 = yζ e(yb ) |Ψ0 .276) = j. m . one can write † † e(xb ) Ψ0 |e(yb ) Ψ0 † ∗ Ψ |e(x y) e(yb ) Ψ = 0 0 = e (x∗ y) † Ψ0 |e(yb ) Ψ0 (5.m. b− ] = 0. hence.278) † ∗ which allows us to replace the l. Orthonormality of the States |Ψ (j. m) We want to show now that the states (5. we can conclude † † ∗ e(xb ) Ψ0 |e(yb ) Ψ0 = e(x y) (5.280) The commutation properties [b+ .283) (5. In order to evaluate the ∗ b) (yb† ) operator e(x e we notice that the derivative property (5. For this purpose we consider the inner product † † e(xb ) Ψ0 |e(yb ) Ψ0 = δjj δmm (5. therefore. b† ] = 0 and [b− . i.277) by Ψ0 |e(x b) e(yb ) Ψ0 . According to (5. (5. (5. [b† . (5. The generating function allows one to derive various properties of the states |Ψ (j. b− )e(yb ) |Ψ0 = f (y+ . of (5.m φjm (x∗ ) φj m (y) Ψ (j.

m ijk Jk . 2.m .5. 1.m. (5.n . Jj ] = = 1 4 1 4 < n |σi | m >< n |σj | m > b† bm .m. − 2 + (5. 2 . m j. J1 = J2 = J3 = 1 † b b− + b† b+ − 2 + 1 † b b− − b† b+ − 2i + 1 † b b+ − b† b− . 3 denote the Pauli spin matrices (5.n .m +b† b† .285) Following steps similar to those in Eqs. .277) φjm (x∗ ) φj m (y) Ψ (j. m) as given in (5. (5. The 2 operators are explicitly. 3 .223. 5. Jj ] = i This property is derived as follows [Ji . etc..m φjm (x∗ ) φjm (y) . (5. (5. States 2 We want to demonstrate now that the states |Ψ (j.n .224).282) and (5. bm bm } n n = 1 4 < n |σi | m >< n |σj | m > {b† bm δmn − b† bm δm n } n n n.j .m.287) ζ 2 ζ.m 135 = e(x ∗ y) . In our present notation the matrix elements < ζ |σk | ζ > for ζ = ± correspond to the matrix elements < ζ |σk | ζ > for σ = ± 1 .275) one can show e(x ∗ y) = j. b† bm n n n.276).268–5. . m) |Ψ j . 1 New Representation of Spin 0.286) from which follows immediately the orthonormality property (5.254) are eigenstates of operators J 2 and J3 with eigenvalues j(j + 1) and m where J 2 and J3 are representations of the spin 2 2 2 operators S 2 = S1 + S2 + S3 and S3 deﬁned in 5. .m.ζ where σk . One deﬁnes corresponding operators Jk through 1 Jk = b† < ζ |σk | ζ > bζ .10: Algebraic Properties of Spinor Operators and comparing (5. k = 1.224). using < + |σ1 | + > = < − |σ1 | − > = 0. < + |σ1 | − > = < − |σ1 | + > = 1.289) < n |σi | m >< n |σj | m > {b† bm .288) The three operators obey the Lie algebra of SU(2) [Ji . b† bm n n n. (5.

289) yield [J− .m. (5. To this end we note 2 2 2 J 2 = J1 + J2 + J3 = 1 1 2 J+ J− + J− J+ + J3 2 2 . m) as Eigenstates of J 2 and J3 We wish to show now that the states |Ψ (j. (5. σj ]| m > b† bm n n.290) |Ψ (j.m.136 = 1 4 − = = 1 4 1 4 Theory of Angular Momentum and Spin < n |σi | m >< m |σj | m > b† bm n n.m.k ijk ijk σk | m > b † bm n = i = i 1 2 < n |σk | m > b† bm n n.295) The last result together with (5.292).m < n |[σi . (5. J1 ] + i [J1 .293) .m < n |2i n.294) . J+ ] = [J1 − iJ2 . m) are eigenstates of J 2 and J3 .m ijk Jk .292) The commutation relationships (5. (5.288) yields ﬁrst the expression J2 = which results in J2 = 1 4 1 † ( b + b + − b † − b− ) ( b † + b+ − b† − b − − 2 ) + b † + b − b † − b + 4 . (5. J2 ] = −2 J3 from which we can conclude the property 1 1 J− J+ = J+ J− − J3 2 2 Hence.291) Here we have deﬁned J+ J− = J1 + iJ2 = J1 − iJ2 = b† b− + = b† b+ . (5. J1 + iJ2 ] = −i [J2 . − (5. 2 J 2 = J+ J− + J3 − J3 = J3 (J3 − 1) + J+ J− (5.m 1 4 < n |σj | m >< m |σi | m > b† bm n n .297) .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.

302) (5. Construct. 2. j = 1. m) . m) J− |Ψ (j. −λ + 1. .300) (5.303) Exercise 5. . m) m |Ψ (j. m) J3 |Ψ (j. . λ. . Show λ = n1 +n2 n1 +n2 2 ( 2 + 1) and m = 1 2 (n1 − n2 ). One can then conclude J 2 |Ψ (j. 2 2 137 (5. . I3 = (b+ b1 + b+ b2 ) 2 2 2 2 2i 2 1 satisfy the Lie algebra of SU(2).e. . [Ij . m .10: Algebraic Properties of Spinor Operators The last term on the r. m = −λ. m = λ||λ. and eigenstates of k with eigenvalues j. ˆ respectively. .254) are eigenstates of + b+ and − b− with eigenvalues j + m and j − m. 1.5. n} eigenstates of I 2 = I1 + I2 + I3 and of I3 I 2 ||λ. 2 . n2 . m + 1) (j + m) (j − m + 1)|Ψ (j. m) = = (j + m + 1) (j − m) |Ψ (j. I3 ||λ.10.1: 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) . i. Ik ] = i jk I . 1 2 k (a) Show that the eigenstates are given by 1 |n1 . .301) Obviously. b † + b + + b † + b+ b † − b − 1 1 b† + b + + b † + b + b † − b− + b † − b − b † + b + . the states (5. b+ ] = δjk . 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. .s. n = n1 + n2 ﬁxed.h. . (5. . I2 = (b1 b2 − b+ b1 ) . b† . using operators I± = I1 ± iI2 2 2 2 and the subspace {|n1 . m − 1) . State the corresponding eigenvalues and the degree of degeneracy.e. (5..305) = = j ( j + 1 ) |Ψ (j. m I1 = where λ = 0. the number of states to the possible energy eigenvalues.298) (5.304) (5. m) One can furthermore derive readily J+ |Ψ (j.299) (5. n1 = 0. [bj . m = m||λ. 1 2. i. . . can be written b† + b − b † − b + = = One can then state J2 = 1 (b† + b+ )2 + b† + b+ b† − b− + b† − b− b† + b+ 4 + (b† − b− )2 + 2 b† + b+ + 2 b† − b− Deﬁning the operator 1 ˆ k = ( b† + b+ + b† − b− ) 2 one can write ﬁnally ˆ ˆ J2 = k ( k + 1 ) b† . (b) Show that the three operators 1 + 1 1 + (b1 b2 + b+ b1 ) .

m) . On the other side the states |Ψ (j. The combination of σ.307) shows that the elements of the rotation matrix can be obtained by binomial expansion of b † and b † in terms of b† and b† . σ values which yields (b† )j+m (b† )j−m is + − then σ = j + m − σ.309) .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).254) are |Ψ j.310) it reduces to (5. + − One may expect that these two conditions restrict both σ and σ. The prefactor of (b† )j+m (b† )j−m which according to (5.243).307) Comparision of (5.306. .138 Theory of Angular Momentum and Spin 5.306) where b † and b † are given by (5. (j) (5.306) and (5. In case j = 1 this expression yields.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. m = b† + j+m b† − j−m (j + m )! (j − m )! |Ψ0 (5.308) The latter sum involves terms (b† )j+m (b† )j−m for σ + σ = j + m and 2j − σ − σ = j − m. m) in the original coordinate system by j |Ψ j. for example. 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. 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. m ) are related to the + − states |Ψ (j. However.5. m = m=−j dm m (β) |Ψ (j. For this purpose we note that the states |Ψ(jm) in a rotated coordinate system according m to (5.251). both conditions are satisﬁed for σ = j + m − σ.

5.12 Mapping of SU(2) onto SO(3) The representation (5. is the expected element of SO(3). . = U x2 ..s.309) and the particular matrix (5.309). b− † = sin β † β b+ − cos b† .309) does not distinguish between SU(2) transformations (5. that the the representation (5. x2 . in case of j = 1.312) The choice of this transformation derives from a property shown further below.313) × 0 0 2 = 0 sinβ 0 cosβ 1 −i 0 which. however. It is of interest to trace the mapping from SU(2) onto SO(3) as represented by (5. i. conclude that the mapping from SU(2) onto SO(3) is a 2–1 mapping.314) leaves (5.g. The SU(2) transformation entered in (5. i.311) (5. of (5.310) to the SU(2) transformation assumed in deriving the general result (5. x3 ) ∈ R .h.308) and had the form (5.314) in case of integer j–values. U = 0 1 1 i √ ( x1 − ix2 ) √ x3 − √2 0 2 2 (5.251) and (5.12: Mapping of SU(2) onto SO(3) 139 5. in fact.. Replacing the latter transformation by its negative form. 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. the orthogonal 3 × 3 matrix which describes a rotation around the x2 –axis. Hence. This factor implies. One can. therefore.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 .e. 2 2 − (5.312) transforms like an angular momentum state ˜ |1 m .e.310).310) establishes a mapping of SU(2) onto SO(3).308) unaltered except for a factor (−1)2j which multiplies then also the ﬁnal result (5. b+ † = −cos β † β b+ − sin b† 2 2 − .251). This can be shown by (1) applying to the matrix A = (dm m ) in (5. the matrix A should represent a rotation around the x2 –axis in the space of vectors ˜ (x1 . namely that the component of the vector on the l. e.

140 Theory of Angular Momentum and Spin .

and to express the Hamiltonian in terms of |rAB |. 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. |ρC |.g. Often the socalled total angular momentum. six parameters must suﬃce to describe the interaction of the system. Example: Three Particle Scattering Consider the scattering of three particles A.Chapter 6 Quantum Mechanical Addition of Angular Momenta and Spin In this section we consider composite systems made up of several particles. The overall orientation of any three particle conﬁguration can be speciﬁed by three parameters1 . Further below we will consider composite systems involving spin states. 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 . hence. Since the Hamiltonian does not depend on the position of the center of mass R = (mA rA + mB rB + mC rC )/(mA +mB +mC ). 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 . and (rAB . β. We like to illustrate this for an example involving particle motion. The rotational part of the scattering motion is described then in terms of the unit vectors rAB and ˆ We remind the reader that. each carrying orbital angular momentum decribed by spherical harmonics Y m (θ. classically speaking the sum of all angular momenta and spins of the composite system. rB . ρC ). γ are needed to specify a general rotational transformation 1 141 . C governed by a Hamiltonian H which depends only on the internal coordinates of the system. three Eulerian angles α.g. B. is the quantity of interest. for example. φ) as eigenfunctions and/or spin. give rise to good quantum numbers. How should they be chosen? Actually there is no unique choice. sums of orbital angular momentum and of spin operators of the particles. e. commute with the Hamiltonian of the composite system and. but neither on the position of the center of mass of the particles nor on the overall orientation of the three particle system with respect to a laboratory–ﬁxed coordinate frame. e. since related operators... rC of the particles. by a rotational vector ϑ. This eleminates three further parameters from the dependency of the Hamiltonian on the three particle conﬁguration and one is left with three parameters. on the distances between the three particles. and particle C impinges on the compound AB coming from a large distance.

. the dimensions d(Bk ) of these subspaces does not exceed 100. . − 2 ≤ m1 ≤ (6. . − 1 ≤ m1 ≤ 1. 2 . mA mB + mB mC + mA mC 2. 1. n = 1. by the total energy E. to separate the 14 641 dimensional space of rotational states Y 1 m1 (ˆAB ) Y 2 m2 (ˆC ) into r ρ subspaces Bk . = B such. . .max d(B) = 1 =0 (2 1 + 1) 2 =0 (2 2 + 1) = ( 1. and by the moment of inertia IA−B of the diatomic molecule A–B approximately as follows 1. (6.max = 2.1) where 1. . i. in fact. 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. 2.4) The basis set which provides a maximum degree of decoupling between rotational states is of great principle interest since the new states behave in many respects like states with the attributes of a single angular momentum state: to an observer the three particle system prepared in such states my look like a two particle system governed by a single angular momentum state.max = 1 2 IA−B E . 2} 1. ρ 1 = 0. 14 641} . and the following mathematical description will shed light on their ubiquitous appearence in physics. . as we will demonstrate below.max . 2. . The rotational symmetry of the interaction between the particles allows one. in particular.m1 .142 Addition of Angular Momenta and Spin ρC .max and 2.max . 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. composite systems behaving like elementary objects are common. B1 ⊕ B2 ⊕ . . In fact. albeit puzzling. rather than by Classical Mechanics.max = 10 one obtains d(B) = 14 641.max = ∆ 2 mA mB mC E . each of which stands for two angles. . (6.max denote the largest orbital and rotational angular momentum values.max + 1)2 ( 2. however.m1 2 .m2 c (n) r ρ 1.max + 1)2 (6. 1. mB . since further important degrees of freedom. by the masses mA . the values of which are determined by the size of the interaction domain ∆V .3) For rather moderate values 1. . . Obviously. The latter often arrives at the physical properties of composite systems by adding the . = 0.2) The dimension d(B) of B is 1. have not even be considered. Such large number of dynamically coupled states would constitute a serious problem in any detailed description of the scattering process. 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 ).. Such extremely useful transformation of the problem can be achieved through the choice of a new basis set B = { 1 .max 2. . a very large number.m2 Y 1 m1 (ˆAB ) Y 2 m2 (ˆC ).e. that only states within the subspaces Bk are coupled in the scattering process. will make their appearence a natural consequence of the symmetry of the building blocks of matter. mC . vibrations and rearrangement of the particles in reactions like AB + C → A + BC.

the reader will learn in the following section how to add and subtract in the microscopic world of Quantum Physics.6) For the following it is essential to note that H is not invariant under rotations of only rAB or ρC . To specify this property mathematically let us denote by H the Hamiltonian in the rotated frame. It holds then H R(ϑ) ψ = R(ϑ) H ψ. The invariance of the Hamiltonian under overall rotations of the three particle system implies then H = R(ϑ) H R−1 (ϑ) . the commutation relationships (1) Jp . presumably a facility the reader would like to acquire with great eagerness. the existence of a basis (6.7) where the generators J (k) are diﬀerential operators acting on rAB (k = 1) and on ρC (k = 2). i. 3 (6.10) . rotations R(ϑ) of the wave functions ψ(rAB . ∂ρx ∂ρy (6. Following our description of rotations of single particle wave functions we express (6. in fact. In this sense. (6. ρC ) deﬁned through R(ϑ) ψ(rAB . ρC ) = ψ(R−1 (ϑ) rAB . For ˆ ˆ example. One can equivalently express therefore (6. Since this is true for any ψ(rAB .6.5) do not aﬀect the Hamiltonian.5) according to (5. but solely under simultaneous and identical rotations of rAB or ρC . 2.7) i R(ϑ) = exp − ϑ · J (6. ρC ) it follows H R(ϑ) = R(ϑ) H. assuming presently that H might.4) which decouples rotational states is connected with the rotational symmetry of the Hamiltonian of the three particle system considered.8) Obviously. be diﬀerent from H.48) i i R(ϑ) = exp − ϑ · J (1) exp − ϑ · J (2) (6. q = 1. R−1 (ϑ) ρC ) (6.0: Addition of Angular Momenta and Spin 143 corresponding physical properties of the elementary components. 5. Jq(2) = 0 for p.9) hold since the components of J (k) are diﬀerential operators with respect to diﬀerent variables. 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. ∂yAB ∂zAB i (2) ∂ ∂ − J3 = ρy − ρx . Hence.e..53. 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.55) holds i (1) − J1 = zAB ∂ ∂ − yAB . 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). according to (5. Rotational Symmetry of the Hamiltonian As pointed out already. connected with the fact that the Hamiltonian H does not depend on the overall orientation of the three interacting particles.

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. B + C] = [A. k = 1. i. k = 1.17) .6) for rotational invariance of the Hamiltonian in the form i i H = exp − ϑ · J H exp + ϑ · J .71–5.13) Hϑ · J − Since this holds for any ϑ it follows HJ − J H = 0 or.f.14) We will refer in the following to Jk . We recognize. the (n) (n) (n) properties [Jk .11) we can write the condition (6.16) following the procedure stated in the theorem above [c. In fact. however.71–5. that the components Jk . 3 as the three components of the total angular momentum operator.14) implies that the components of the total angular momentum operator (6. ˆ (6. 2. 3 do not commute. the important fact that the Jk obey the Lie algebra of SO(3). we will ﬁnd that the states yield the basis B with the desired property of a maximal uncoupling of rotational states.144 where J = J (1) + J (2) Addition of Angular Momenta and Spin . J ] = i k m Jm for n = 1..e.14) implies that the total angular momentum is conserved during the scattering process. (5. i. (6. that energy. 3 cannot have simultaneous eigenstates among each other. To order O(|ϑ|) one obtains i ϑ · JH . (6. The property (6. the operators Jk . k = 1. of course. Jk ] = 0 . 2. k = 1. C].11) By means of (6..e. One can show readily that the commutation relationships [Jk .15) are satisﬁed. i. 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 ). 2.9). For a proof one uses (6. [H. (6. a supposition which can be tested through the commutation properties of these operators.e. We suspect. In the following we will use the notation r ρ Ω1 = rAB .12) each individually can have simultaneous eigenstates with the Hamiltonian. for |ϑ| H ≈ 1 − 1 i ϑ·J H 1 + 1 i ϑ·J ≈ H + i 1. 3 . componentwise. ˆ Ω 2 = ρC . Before we apply the procedure (5. and the eigenvalues of J2 and J3 are good quantum numbers. 2.81)]. J ] = i k m Jm (6. Eqs. r ρ Deﬁnition of Total Angular Momentum States The commutation property (6. 2 together with the property [A. B] + [A. (6.81) we want to consider the relationship between YJM and Y 1 m1 (ˆAB ) Y 2 m2 (ˆC ).12) We consider this equation for inﬁnitesimal rotations.

1: Show that J2 . We. an important example being the symmetry groups SU(N) governing elementary particles made up of two quarks. J (2) . assume the expansion YJM ( 1 . hence. i. 2 M YJM ( 1 . Ω2 ) (1) (2) = 1 m1 2 m2 ) Y 1 m1 (Ω1 ) Y 2 m2 (Ω2 ) (m1 + m2 ) (J M | . Exercise 6.20) This equation can be satisﬁed only if the Clebsch-Gordan coeﬃcients satisfy (J M | 1 m1 2 m2 ) = 0 for m1 + m2 = M . Ω2 ) = m1 (J M | 1 m1 2 M − m1 ) Y 1 m1 (Ω1 ) Y 2 m2 (Ω2 ) . therefore.e. i. 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. play a cardinal role in the mathematical description of microscopic physical systems. mesons.. Ω2 ) = m1 . and J3 .e. J3 .h. ρC ) are normalized. J3 YJM = M YJM .22) We will not adopt such explicit summation since it leads to cumbersum notation.1. We ﬁrst notice that YJM in (6. .m2 (J M | 1 m1 2 m2 ) Y 1 m1 (Ω1 ) Y 2 m2 (Ω2 ) (6. two further quantum numbers need to be speciﬁed for a complete characterization of the total angular momentum states. of (6. However. The two missing quantum 2 2 numbers are 1 and 2 corresponding to the eigenvalues of J (1) and J (2) . and three quarks. restrict the sum in (6.1: Clebsch-Gordan Coeﬃcients 145 6. J3 .s.18) is an eigenfunction of J3 .21) One can. or the closely related Wigner 3j-coeﬃcients introduced further below. J (1) .m2 2 |Ω1 . i.6.18) to avoid summation of vanishing terms YJM ( 1 . the eigenvalue being speciﬁed by the quantum number M .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) .e. the reader should always keep in mind that conditions equivalent to (6. 2 |Ω1 . Equivalent coeﬃcients exist for other symmetry properties of multi–component systems. (6.19) Noting J3 = J3 + J3 and applying this to the l. 2 |Ω1 . J (2) .18) yields using the property (k) J3 Y k mk (Ωk ) = mk Y k mk (Ωk ) .. (6. baryons. 2 |Ω1 . m1 . The expansion coeﬃcients (J M | 1 m1 2 m2 ) ˆ are called the Clebsch-Gordan coeﬃcients which we seek to determine now.21) hold. These coeﬃcients.18) where the states YJM ( 1 . (6. k = 1. and J (1) · J (2) commute. (6. Since YJM sofar speciﬁes solely two quantum numbers.

= {YJM ( 1 .146 Theory of Angular Momentum and Spin The expansion (6.h. (6. 2 ) has (2 1 + 1)(2 2 + 1) elements. Exercise 6. . .28) This property follows from the fact that the basis elements are eigenstates of hermitian operators with diﬀerent eigenvalues. (6. property implies dΩ1 dΩ2 Y 1 m1 corresponding to the l. B2 2) should be 2 |Ω1 . Our derivation below will also yield real values for the Clebsch-Gordan coeﬃcients. .25) We want to state now two summation conditions which follow from the orthonormality of the two basis sets B( 1 . it holds 1+ 2 ( 2J + 1 ) = (2 J=| 1 − 2 | 1 + 1) (2 2 + 1) . . the basis B ( 1 .2: Prove Eq.m2 2 M and with (6. −J + 1.e. 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.s. 1. (6. 2) = {Y 1 m1 (Ω1 ) Y 2 2 m2 (Ω2 ). JM 2 |Ω1 . 2 ) and B ( 1 .18) constitutes a change of an orthonormal basis B( 1 . Ω2 ) YJ M ( 1. . .27) together with (6. M = −J. . (6. − + 1. .1. For each quantum number J there should be 2J + 1 elements YJM with M = −J. Ω2 ). . . (6.24) The basis B( 1 . ..25) i. . . J = | 1 − 2 |. Ω2 ) = δJJ δM M (6. . . In this case the range of |J|–values would be the interval [ |J (1) | − |J (2) | . In fact.s. and that the states can be normalized. | 1 − 2| + 1. 1 + 2. −J + 1.24) yields 1 m1 2 m2 ) ∗ 1 m1 2 m2 ) (J M| = δJJ δM M . | 1 − 2 | + 1.h. . The basis B ( 1 . J (1) and J (2) . . it is not immediately obvious what the J– values are.23) corresponding to the r. 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 . − 2} 1 + 1.. m1 = − 1 .26) We will show below in an explicit construction of the Clebsch-Gordan coeﬃcients that. |J (1) | + |J (2) |]. 2) . in fact. m2 = − 2 . 2 ). The property dΩ1 dΩ2 Y∗ ( 1 . J.18) applied to Y∗ and to YJ JM (J M | m1 . (6. . . J} . to a new basis B ( 1 . 1 + 2 . However. respectively. 2 |Ω1 . This obviously corresponds quantum mechanically to a range of J–values J = | 1 − 2 |. 2 ) is also orthonormal2 and must have the same number of elements. the range of values assumed for J is correct. .

30) The latter property follows from multiplying (6. to composite systems involving spin.2 Construction of Clebsch-Gordan Coeﬃcients We will now construct the Clebsch-Gordan coeﬃcients.18) that the Clebsch-Gordan coeﬃcients obey the second summation condition (J M | 1 m1 2 m2 )∗ (J M | 1 m1 2 m2 ) = δm1 m1 δm2 m2 . Ω2 )Y 1 m1 (Ω1 )Y 2 m2 (Ω2 ) . Ω2 ) . Ω2 ) and integrating. 2 |Ω1 . .. or baryons multiplets involving three quarks.31) are identical to Comparision with (6. therefore. i. At this point we like to stress that the construction will be based on the theorems (5.18). We can.81) stated above. (6. Ω2 ). (6.34) 6.2: Construction of Clebsch-Gordan Coeﬃcients 147 The second summation condition starts from the fact that the basis sets B( 1 . The (J M | 1 m1 2 m2 ) cJ M .29) where the expansion coeﬃcients are given by the respective scalar products in function space cJ M = dΩ1 dΩ2 Y∗ J M ( 1. (6. A similar 2 construction will also be applied to composite systems governed by other symmetry groups. 2 |Ω1 . Ω2 ) . and actually also the properties of Clebsch-Gordan coeﬃcients stated above.e. rather the convention (J M | has been assumed.. 1 m1 2 m2 ) = 0 if J < | 1 − 2| .m2 M ( 1.. or J > 1 + 2 (6.33) JM The latter summation has not been restricted explicitly to allowed J–values.18) by Y∗ J orthogonality property (6. will be based solely on the commutation properties of the operators J and J (k) . Hence. 2 ) and B ( 1 .28) from (6.6. i. The result of this construction will include all the properties previewed above.27) yields δJJ δM M = m1 .g. 2 |Ω1 . 2 |Ω1 . 2 |Ω1 . also apply the results. i.e. (6. Y = J =| 1 − 2 | M =−J 1 m1 M (Ω1 )Y 1+ 2 2 m2 (Ω2 ) (J M | ∗ 1 m1 2 m2 ) YJ M J ( 1. the group SU(3) in case of meson multiplets involving two quarks. 2 ) span the same function space. 1+ 2 J Y 1 m1 (Ω1 )Y 2 m2 (Ω2 ) = J =| 1 − 2 | M =−J cJ M YJ M ( 1. it is possible to expand Y 1 m1 (Ω1 )Y 2 m2 (Ω2 ) in terms of YJM ( 1 .28) allows one to conclude that the coeﬃcients cJ (J M | 1 m1 2 m2 )∗ .32) which complements (6.. One can show readily using the same reasoning as applied in the derivation of (6.1 states.71–5.e. e. (6.

2 |Ω1 . (6.36) = 0 = J YJJ ( 1 .11) J+ by J+ + J+ . . . .43) . Having determined such YJJ we then construct the whole family of functions XJ = {YJM ( 1 . we can also demonstrate using (??) that G dΩ1 = We. Ω2 ) 2 |Ω1 . 1+ 2 ( 1. We obtain using (5. 1+ 2 In fact. . Ω2 ). M = −J.66. 1+ 2.18) which satisﬁes J+ YJJ ( 1 .35) 1 − 2 |.40) (Ω1 ) J+ Y (2) 2 2 J+ Y 1 1 (Ω1 ) Y (Ω2 ) + Y (Ω2 ) = 0 .37) and replace (1) (2) according to (6. 2 |Ω1 . (6.38) for M = J − 1. 2 |Ω1 . .42) ( 1.148 For the construction of YJM we will need the operators J± = J1 + iJ2 .25) using (6. The corresponding classical total angular momentum vector Jclass (1) (2) would be obtained by aligning both Jclass and Jclass also along the z–axis and adding these two vectors. We ﬁrst seek an unnormalized solution GJJ and later normalize. 2 |Ω1 . therefore. −J.39) which we will try for a solution of (6. 2 |Ω1 . Ω2 ) . . (6. Ω2 ) is normalized dΩ2 G dΩ1 Y 1 1 1+ 2. 2 |Ω1 . Ω2 ) = Y 1 1 (Ω1 ) Y 2 2 (Ω2 ) (6. Ω2 ) (Ω1 ) dΩ2 Y 2 2 (Ω2 ) = 1. . 2 |Ω1 . Ω2 ) = (J + M + 1)(J − M )YJM ( 1 .68) J+ = (1) + J+ (1) (2) Y 1 1 (Ω1 ) Y 2 2 2 2 (Ω2 ) 1 1 (6.11) and (5. have shown Y 1+ 2. Ω2 ) .64) J3 = = (1) + J3 (1) 1 (2) Y 1 1 (Ω1 ) Y 2 2 2 2 (Ω2 ) 1 1 J3 Y ( + 1 1 (Ω1 ) Y 1 1 (Ω2 ) + Y 2 2 (Ω1 ) J3 Y (2) 2 2 (Ω2 ) (6. 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. J3 YJJ ( 1 . | 1 − 2| + 1. Quantum mechanically this corresponds to a state G 1+ 2. 2 |Ω1 . For this purpose we insert (6. We embark on the suggested construction for the choice J = 1 + 2 . J − 2. 1+ 2 ( 1. Ω2 ) Addition of Angular Momenta and Spin (6. J} by applying repeatedly J− YJM +1 ( 1 .37). . we can demonstrate condition (6.37) The solution needs to be normalized.39) into (6. . Ω2 ) = Y 1 1 (Ω1 ) Y 2 2 (Ω2 ) . 1 + 2} and for (6. (6.41) 2) Y (Ω1 ) Y (Ω2 ) . . 2 |Ω1 . 1+ 2 ( 1. The construction assumes a particular choice of J ∈ {| such J–value seeks an expansion (6. Similarly.5. −J + 1.

38) yields pression 2( 1 + 2 )Y 1 + 2 1 + 2 −1 ( 1 . . Exercise 6. One can readily show that (6.45). 1 + 2 − 2.2. G 1 + 2 −1 1 + 2 −1 in (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. all the Clebsch-Gordan coeﬃcients ( 1 + 2 M | 1 m1 2 m2 ). ).46) 2 1 +c Y (Ω1 ) Y 2 2 (Ω2 ) = 0 .47) (Ω2 ) − 1 1 2 + Y 2 1 1 −1 (Ω1 )Y 2 2 + Y 2 1 1 (Ω1 )Y 2 2 −1 (Ω2 ) . . We seek for this purpose ﬁrst an unnormalized solution G 1 + 2 −1 1 + 2 −1 of (6. | . etc. Expression (6.2: Construction of Clebsch-Gordan Coeﬃcients 149 We now employ property (6. . This expression is orthogonal to (6.6. | . of (6.47) can serve now to obtain recursively the elements of the family B + − for M = 1 + 2 − 2. . . The result is illustrated for the case 1 = 2. 2 |Ω1 . We like to construct now the family of total angular momentum functions B + − = {Y + − M ( . The r.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 ). 1 + 2 − 3.38) to construct the family of functions B + = {Y + M ( .s. One obtains then Y 1 1+ 2 1 + 2 1 + 2 −1 2 2 ( 1.37) is satisﬁed.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. To demonstrate (6. 2 |Ω1 . ( + − )}. According to the condition (6. . Show that the resulting functions are orthonormal. −( 1 + 2 − 1). . Ω2 ). Normalization yields furthermore Y = 1 1 + 2 −1 1 + 2 −1 We have thereby determined ( 1 2 |Ω1 Ω2 ) (6.h. The reader should familiarize himself with the entries of the Table. . One can obviously continue the construction outlined to determine Y 1 + 2 −2 1 + 2 −2 . . . thereby. 2 |Ω1 .45) for some unknown constant c. and all total angular momentum functions for a given choice of 1 and 2 . with the symmetry pattern and with the terms Y 1 m1 Y 2 m2 contributing to each YJ M . M = −( + ). Having constructed this family we have determined the Clebsch-Gordan coeﬃcients ( 1 + 2 − 1M | 1 m1 2 m2 ). We demonstrate the procedure ex(1) (1) plicitly only for M = 1 + 2 − 1.s. To satisfy this equation one needs to choose c = − 1 / 2 . M = −( + − ). 1 + 2 − 1.37). . .44) (Ω2 ) .39) as required by (6. 2 = 1 in Table 1. 6. in particular. 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.27).36.h. Ω2 ) 2 1+ 2 = (Ω1 )Y 2 2 −1 (6. ). of (6. Ω2) for M = 1 + 2 − 2 and J = 1 + 2 . We have provided in Table 1 the explicit form of YJ M ( 1 2 |Ω1 Ω2 ) for 1 = 2 and 2 = 1 to illustrate the construction.1: Construct following the procedure above the three functions YJM ( 1 . . ( + )}. The l.

730297 0. 1|Ω1 . Ω2) Y11 (2. 1|Ω1 .816497 −0.774597 0.258199 −0.632456 0. Ω2) Y21 (2.707107 0. 1|Ω1 .816497 0.547723 0.632456 −0. Ω2) Y10 (2.57735 3 10 3 5 Y2−1 (Ω1 )Y11 (Ω2 ) Y20 (Ω1 )Y10 (Ω2 ) Y21 (Ω1 )Y1−1 (Ω2 ) −0. Ω2) Y32 (2. .547723 0.57735 0. Ω2) Y3−3 (2. 1|Ω1 . 1|Ω1 . Ω2) Y31 (2. 1|Ω1 . Ω2) Y3−1 (2. 1|Ω1 . Ω2) Y30 (2. 1|Ω1 .774597 − − − − 0.707107 0. Ω2) Y1−1 (2.316228 0.816497 Y22 (Ω1 )Y1−1 (Ω2 ) 1 15 1 3 3 5 1 5 1 2 3 10 2 5 1 2 1 10 2 3 1 3 Y20 (Ω1 )Y11 (Ω2 ) Y21 (Ω1 )Y10 (Ω2 ) 0. Ω2) − − − − 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. 1|Ω1 . Ω2) Y2−2 (2. 1|Ω1 . 1|Ω1 .57735 0.258199 0. 1|Ω1 .447214 −0. 1|Ω1 .57735 −0.316228 0.707107 0.447214 0. Ω2) Y22 (2.547723 0.707107 0. Ω2) Y2−1 (2.57735 0. 1|Ω1 .730297 −0. Ω2) Y3−2 (2.57735 0. 1|Ω1 . Ω2) Y20 (2.408248 −0.150 Addition of Angular Momenta and Spin Y33 (2.816497 Y2−2 (Ω1 )Y11 (Ω2 ) Y2−1 (Ω1 )Y10 (Ω2 ) Y20 (Ω1 )Y1−1 (Ω2 ) Y2−2 (Ω1 )Y10 (Ω2 ) − 2 3 Y2−1 (Ω1 )Y1−1 (Ω2 ) Y2−2 (Ω1 )Y1−1 (Ω2 ) 1 Table 6.1: 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 0.408248 −0.774597 0 2 5 8 15 1 6 3 10 1 3 0.632456 0.

ζ = ± and the matrix elements < ζ |σk | ζ > are as deﬁned in Section 5. M | .48) (6.53) (6. a† and bζ . − 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. J.50) (6. aζ bζ . (5.54) 2 + = √ b† ( 2 +m2 )! √− ( 2 −m2 )! . 1 .10.254)] the angular momentum / spin eigenstates | 1 m1 of the two particles are | 1 m1 | 2 m2 1 + = √ a† 1 +m1 ( 1 +m1 )! 2 +m2 √− b† a† 1 −m1 ( 1 −m1 )! 2 −m2 |Ψ0 |Ψ0 .10. b † ζ = δζζ = δζζ . we extent deﬁnition (5.51) The operators aζ . (6.ζ a† < ζ |σk | ζ > aζ ζ b† < ζ |σk | ζ > bζ ζ (6.287) to two particles (1) J k k = = (2) J 1 2 1 2 ζ.2.49) where ζ. bζ . M | . m + 1 .6. For this purpsose we will employ the spinor operators introduced in Sections 5. 5. commute with each other ζ ζ aζ .10 where we studied single particle angular momentum and spin. a† ζ ζ b† .9.ζ ζ. These expressions will be derived now. aζ . b† ζ = 0. 1 (6. It is actually possible to state explicit expressions for any single coeﬃcient (JM | 1 m1 2 m2 ). 5. we are dealing now with two particles carrying angular momentum or spin. hence. bζ = = a† . Accordingly.10 [cf. bζ = a† . Deﬁnition of Spinor Operators for Two Particles In contrast to Sections 5. m − 2 . b† ζ ζ = aζ .52) and | 2 m2 2 According to Section 5. b† refer to diﬀerent particles and. 2 . (6.9. The creation and annihilation operators are again of the boson type with commutation properties aζ . a† ζ . 6. + 2 − J−M +1 for = J + 1 2J+2 2 J−M for = J − 1 2J 2 = .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. b† ζ ζ = 0 = 0 .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 ).3: Explicit Expressions of CG–Coeﬃcients 151 Exercise 6.

(2) J 2 . 2 )) . (1) J 2 . 5. hence. we like to recall for future reference that the operators (5. M ( 1 . M ( 2) 2) = J(J + 1) |J.68) The states |J.e. The aim of the present Section is to determine closed expressions for the Clebsch–Gordan coeﬃcents ( 1 . m2 |J. 1.61) (6. 2) 1. M ( (2) J 2 |J. (6. 2 )). J3 |J. m1 1 | 2 . 2) 2) (6. m1 1 | 2 . M ( 1 . are | 1 . (1) J 2 | (2) J 2 | 1 m1 1 2 m2 2 Addition of Angular Momentum and Spin = = 1( 1 2( 2 + 1) | 1 m1 + 1) | 2 m2 1 2 . as usual are denoted by their respective quantum numbers J.303). 2) 2) can be expressed in terms of Clebsch-Gordan coeﬃcients (6. 2) + 1) |J. + 1) |J.67) = j |J.54). m2 2 . 3 . k2 = b b+ + b† b− k1 = + − − 2 2 + namely. (1) J (2) J 3 | 1 m1 1 3 | 2 m2 2 = m1 | 1 m1 = m2 | 2 m2 1 2 (6. M ( 1 .69) − 2| ≤ J ≤ + 2 . M ( 1 . M ( 1 . 6..53. M ( 1 . (2) J which. . = = 1( 1 2( 2 2) 2) (6.300).302. M ( 1 . (6. M ( 1 . 2 . m1 1 | 2 .66) (6. ˆ For the operators kj holds ˆ kj | j . mj j = j | j . |J. (j) 2 ˆ ˆ J = kj ( kj + 1 ) . m2 a† + ( 1 2 = a† − ( 1 1 −m1 (6. M ( 1 . 2) We will also require below the raising and lowering operators associated with the total angular momentum operator (6.62) . m1 . 2) . which describe a two particle system according to (6. (1) J . 2 ) which are simultaneous eigenstates of the operators 2 2 J2 . (6. m2 1 2 ( 1 . M ( 1 .56) The states | 1 . can be expressed in terms of the number operators 1 † 1 ˆ ˆ a† a+ + a† a− . j = 1.58) Jk + (2) Jk .59) We seek to determine states |J.m2 | 1 . The operator of the total angular momentum/spin of the two particle system is J = with Cartesian components Jk = (1) J + J (6.59) J± = J1 ± i J2 . (6. M ( 1 . ˆ kj |J. i. m2 |J.65) (6. 2 . in analogy to Eqs. 1 .55) . k = 1. (6.60) (6. 2.57) b† + ( (2) 2 2 +m2 1 +m1 b† − ( 2 2 −m2 + m1 )! − m1 )! (1) + m2 )! − m2 )! |Ψo .152 It holds. m1 . M. −J ≤ M ≤ J . for such states should hold J2 |J. . J3 . (5.63) according to At this point. = M |J.18) as follows = | 1 m1 . (1) J 2 |J. 2 . M ( 1 . mj j and. M ( 1 . 2 .64) (6.

j = 1. a† b† − a† a− . 6.74) + 1 J− J+ + J2 .50. Employing (6. K † 1 † a a+ + a† a− . K † = 0.73) can be readily proven. K † 1 a† a+ − a† a− + b† b+ − b† b− . For example.71) (6.75) The relationships (6. K† J3 . due to J2 = K† 1 2 J+ J− 1 † K . + − − + 153 (6. a† b† + + − − + 2 2 − 1 1 = a† b† − a† b† = K† . + − + − 2 2 − + = .73. 2 4 = 0 = = 0. a† b† − a† b† + − + − + − − + 2 1 1 = a† a+ . the relationships (6.65) and (6. We note that. and applying the relationships (6. 4 Using J3 = (1) J 3 + (2) J 3 . j = 1. + − − + 2 2 = 1 † 2K . expressing (k) J 3 through the creation and annihilation operators according to (5. 6. K † J± . K † (kj + 1) 2 2 1ˆ † 1 ˆ kj K + K † (kj + 1) 2 2 1 ˆ 1 ˆ K † kj + kj . K † (j) J 2 .71–6. 2 2 3 ˆ = K † kj + K † . K † 2 1ˆ ˆ 1 ˆ ˆ kj kj + 1.3: Explicit Expressions of CG–Coeﬃcients The Operator K † The following operator K † = a† b† − a† b† .6. using (6.288). b† = 0 .70) will play a crucial role in the evaluation of the Clebsch-Gordan-Coeﬃcients. This operator obeys the following commutation relationships with the other pertinent angular momentum / spin operators ˆ kj . a† b† + a† a− .72) (6.64. 6.71–6. ˆ A similar calculation yields [k2 .74) imply 3 2 J2 .71) one can show K† = = = = = 1 ˆ ˆ kj (kj + 1). (6.51) one obtains ˆ k1 .73) (6. K † ] = (j) J 2 . b† − b† a† b− . a† b† + − − + 2 + 2 − 1 1 − b† a† b+ . K † + K † 2 2 3 † †ˆ K kj + K . a† b† − a† b† − + − − + 2 + 1 1 = a† a+ . (6. K † + kj .73) yields J3 .

it holds K † |J. but for diﬀerent 1 and 2 quantum numbers of the operators (1) J 2 and (2) J 2 . namely. M ( 1 .62. 6.67) (j) J 2 K † |J. j2 ) . M ( 1 . j2 + ) 2 2 n n = N (n. j1 + j2 (j1 + n n .2. according to (6. M ( 1 + .. To demonstrate that the resulting states are eigenstates of (1) J 2 and (2) J 2 we exploit (6. j1 + ) K † 2 2 (6. 2 ) j + 4 1 3 + K † |J. K† + K † (j) J 2 |J. M ( 1 . 2) = = (j) 2 J . this result implies that K † |J. M ( 1 . a† b† + b† a† b− . |j1 + j2 . 1 2 + 2 . i. 2) is a state with quantum numbers 1 1 + 1 2 and = N |J. M ( 1 . M ( 1 + n . M ( 1 .154 Starting from (5. j1 + j2 (j1 .77). a† b− − a† b† + + + − + = 0.63. j + 2 2 2) However. b† + − + + + − The property [J− . 2) = N |J.76) Here N is an unknown normalization constant. j2 2 . M ( 1 .77) where N is another normalization constant.72) and (6. 2) 2) ˆ K † kj + = K† = j 3 † K + K † j ( j + 1) |J. The commutation properties (6. K † = Addition of Angular Momentum and Spin a† a− + b† b− . 2 ) . j1 + j2 (j1 . 2 ) produces again total angular momentum eigenstates to the same J and M quantum numbers of J2 and J3 . = − a† a− .78) a state which has been used already in the construction of Clebsch-Gordan coeﬃcients in Section 6. 2 2 1 + ) . j1 1 + . 2 (6. Application of (K † )n to this state yields.79) n |j1 + j2 . Action of K † on the states |J. M ( 1 .75) ascertain that under the action of K † the states |J. 2 2 + n ) 2 (6. M ( 1. 6. 2 ) remain eigenstates of J2 and J3 with the same quantum numbers. 6. M ( 1 .76) and state K† n |J. 2) We want to demonstrate now that the action of K † on the states |J. (6. K † ] = 0 is demonstrated in an analoguous way.e. j1 1 |j2 . M ( 1 .292) one can derive similarly J+ .73. One can generalize property (6. We consider now the case that (K † )n acts on the simplest total angular momentum / spin state. on the state |j1 + j2 . j2 ) = |j1 . 4 3 + j ( j + 1) |J.

2) . in analogy to the construction (5. (6.74)] yields |J. M ( 1 . − 1. 2 ) ∆(J. (J + 2 − − 2) 1 1) 2 The normalization constant appearing here is actually N( + − J. (6.6. 1 .82. holds |J. 6. 2 ) can be expressed through the r.84) with (6. K† 1 2 1 + 2 −J × × . 2 ) exploits the expression (6. M ( 1 .3: Explicit Expressions of CG–Coeﬃcients 155 where we denoted the associated normalization constant by N (n. 2) Our construction of the states |J.83) The derivation of this expression will be provided further below (see page 158 ﬀ). 2) = ∆(J.80) = = = 1 1 2 2 − J 1 n − = (J + 2 2 n 1 − = (J + 2 2 + 1 1 − − 2) 2) .82) 1 × | (J + 2 1 × | (J + 2 1 (J + 2 1 − 1 ) .57) and exploiting the fact that J− and K † commute [c. 2 ) for −J ≤ M ≤ J as follows |J. 2) = N( 1 + 2 − J.104. j1 + n . The latter states.h. of (6. 2 = j2 + n 2 (6. = (J + M )! (2J)! (J − M )! 1 2 2) (6. 2) 1 2 = ( 1 + 2 (2J + 1)! − J)! ( 1 + 2 + J + 1)! 1 2 . 1. Combining (6. J( 1 . 2) 2) . 2 ) .f. j2 + n ). M ( 1 . j1 . 2 2 It is now important to notice that any state of the type |J.84) (6. J( 1 .85) ∆(J. j2 as follows J = j1 + j2 which is equivalent to n j1 j2 Accodingly. 1 = j1 + n 2 . Strategy for Generating the States |J.79). M ( 1 . allow one to obtain the states |J. J( 1 . (6.s. For this purpose one needs to choose in (6.82) for |J.86) b† + J+ 2 − 1 |Ψo . M ( 1 .81) = N( 1 + 2 1 2 − J. (6.79) n. M ) (J− )J−M |J. 5. M ) (J + × K† 1 − 2 )! (J + 2 − 1 )! J+ 1 − 1 + 2 −J (J− )J−M a† + × 2 (6. J( 1 .105) of the spherical harmonics. M ) .

For this purpose we introduce the generating function I(λ. x. y) = λr xs y t r |Gst . I(λ. + − + − . a† ] = [b+ .90) (6.88)..91) which. 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..264).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 + (6.69) yields then the Clebsch- Expansion of an Intermediate State We ﬁrst consider the expansion of the following factor appearing in (6. (6.86) in terms of monomials 1 +m1 1 −m1 2 +m2 2 −m2 a† a† b† b† |Ψo . st The desired expansion of |Gr can be obtained from an alternate evaluation of I(λ.e. in terms of | 1 .52).s.87) + − + − i.88) in terms of monomials (6.50–6. r!s!t! r. + + Taylor expansion of the two exponential operators yields immediately I(λ. One obtains then using J− = a† a+ + b† b+ (6. 2 [cf.e.t (6. 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. b† ] = 0 [cf. x.89) i.86) |Gr = Jr a† st − + s b† + t |Ψo (6. y) is a generating function for the states |Gr deﬁned in (6. (6.92) − − and noting [a+ .93) = f (a† + λ a† ) g(b† + λ b† ) |Ψo . (6. x.87).57)]. follows from the commutation properties (6. m2 Gordan-coeﬃcients. Comparision with (6. m1 1 | 2 . y) = exp (λJ− ) exp x a† exp x b† |Ψo .

y) = s.99) .90) allows one to infer |Gr s.3: Explicit Expressions of CG–Coeﬃcients We conclude I(λ. 157 (6. m2 2 .95) Comparision with (6.88).86) (J− )J−M a† + = q J+ 1 − 2 b† + J+ 2 − 1 |Ψo 1 (6.70) holds K† 1 + 2 −J = s 1 + 2 −J b† − s 1 + 2 −J−s (−1)s 1 + 2 −J−s (6. x. to obtain the desired expansion of |J.96) and.t × a† + s−q s q b† + t r−q t−r+q a† − q b† − |Ψo (6. 2) = s.94) yields I(λ. With K † given by (6.t × λ v b† − × r−q v × = t s−u v λu a† − r! q |Ψo λr xs y t r! s! t! r.97).97) yields. M ( 1 . |J.6.94) One can infer from this result the desired expressions for |Gr .98) a† − s a† + b† + s . x. using the deﬁnition (6. Expanding the exponentials in st (6.50). using the commutation property (6. 2 ) in terms of states | 1 . one can write the right factor in (6.t = q r! × a† + s q s−q t r−q a† − q × b† + t−r+q b† − r−q |Ψo (6. m1 1 | 2 .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.s.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. y) = exp a† + λ a† + − exp b† + λ b† + − |Ψo .q (−1)s (6. Operation of this operator on (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 . q = 1 − m1 − s and m2 = M − m1 . n)|ψ(j1 . m2 in (6. j2 . according to (6. The summation over q corresponds then to the summation over m1 .82) and (6. 2 ) − 2) . m1 .101) Note that m1 + m2 = M holds. n) 2 )|J. J( 1 .103) this can be written 1 = N 2 ψ(j1 . j2 = 1 (J + 2 2 − 1) . The Clebsch-Gordan coeﬃcents are then ﬁnally ( 1 . To determine N = N ( 1 + 2 −J.69) since. × 2) = N( 1 + 2 1 − J. (6. m2 = M − 1 + q + s.69) if one identiﬁes m1 = 1 −q − s . M − +q+s One can conclude that this expression reproduces (6. M ) 1 = + J)!(− 1 + 2 + J + 1)! 2 2 + 2 − J)!( 1 − ( 1+ + J)! 1 2 × [( × s + m1 )!( − m1 )!( 2 + m2 )!( 2 − m2 )!(J + M )!(J − M )!] 2 (−1)s s!( 1 − m1 − s)!( 2 + m2 − s)! 1 2 1 × The Normalization 1 ( 1 + 2 − J − s)!(J − 1 − m2 + s)!(J − + m1 + s)! (6. 2 ) we consider the scalar product J. (6. 2 ) ∆(J. m2 |J.53. 1 . j2 .103) = 1.104) .83) of the normalization constant N ( 1 + deﬁned through (6. M ) (J + 2 )! (J + 2 − 1 )! s.82). − q − s)!(q + s)! 2 × (M + 1 − 1 + q + s)!( 1 1 + 2 − M − q − s)! 2 − q − s 1 | 2. n = 1 + 2 −J. 2 ) (6. 1 .101).6.54) between creation operator monomials and angular momentum states allow one to write this |J.102) We want to determine now the expression (6. √ 2J + 1 1 2 . − 1 . For this purpose we introduce j1 = 1 (J + 2 1 2 − J.158 Addition of Angular Momentum and Spin ( 1 + 2 − J)!(J − M )!(J + 1 − 2 )!(J + 2 − 1 )! s!( 1 + 2 − J − s)!q!(J + 1 − 2 − q)!(J − M − q)!(M + 2 − a† + 2 1 −q−s 1 + q)! a† − q+s b† + M + 2 − 1 +q+s b† − 1 + 2 −M −q−s |Ψo The relationships (6. Using (6. M ( 1 .q (−1)s × (6. J( 1 .

m2 2 . 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 . Accordingly. (6. we obtain |ψ(j1 .108) = λ=0 (n1 + ν)! n1 ! (6. We employ the expression (6.3: Explicit Expressions of CG–Coeﬃcients where |ψ(j1 . m1 1 |j2 .110) which can be written 1 1−λ n1 +n2 +2 = r s n1 + r − s n1 n2 + s n2 λr (6. n) = K† n 159 |j1 . j2 2 . (6.70) for the operator K † .107) s = (n!)2 s 2j1 + n − s 2j1 The latter sum can be evaluated using 1 1−λ a property which follows from ∂ν ∂λν 1 1−λ n1 +1 n1 +1 = m1 n1 + m1 n1 λm1 (6. (6.104) 1 = N2 (n!)2 (2j1 )!(2j2 )! (2j1 + n − s)!(2j2 + s)! s!(n − s)! 2j2 + s 2j2 (6.109) and Taylor expansion of the left hand side of (6. j2 .m2 n1 + m1 n1 n2 + m2 n2 λm1 +m2 (6.105) The ﬁrst step of our calculation is the expansion of ψ(j1 .108) yields the identity n1 + r − s n1 n2 + s n2 = n1 + n2 + r + 1 n1 + n2 + 1 .106) The orthonormality of the states occurring in the last expression allows one to write (6.57) for these states and the expression (6. applying (6. 1 1−λ n1 +1 1 1−λ n2 +1 = m1 .108).6.108) twice.112) s . n) in terms of states |j1 . One obtains then. j2 . j1 1 |j2 .111) Comparision with (6.

and (6. m2 ) by ( 1 . m2 . m2 |J. −m2 . (6. 2 .102) of the Clebsch-Gordan coeﬃcients. 2 1+1 (6. J (1) .116) If one takes the negatives of the operators in (6.116). m1 . (6. (2j1 + 2j2 + 1)! (6.s.h. m1 ).120) The symmetry properties (6.114) by. M | 1 . 1 .83).160 Applying this to (6. e. M ) corresponding to (6.113) Using the identities (6. 2 . one can interchange also the operator J on the l.h.116) one expresses the Clebsch-Gordan coeﬃcient on the r.114) For example. −m1 2|J. J. −M ) = (−1) 1 + 2 −J ( 1 .114) one obtains − J = − J (1) − J (2) The respective Clebsch-Gordan coeﬃcients ( 1 . M ) ( 1 . −m1 . manner to the coeﬃcients ( 1 .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 . 2 . ( 1 . (6. 2 . m2 |J.118) Finally. (6. and J (2) are vectors in R .102) by replacing ( 1 . ( 2 . m1 .120) can be readily derived from the expression (6.114) as long as J. of this equation J (1) = J (2) − J . 6. and . of (6.s.103) one obtains the desired result (6. We will demonstrate this now. To derive relationship (6. (6. 1 . −m1 .. m1 |J.115) This relationship is a trivial consequence of (6. m1 . m2 . M ) = (−1) 2 +m2 2J + 1 ( 2 .116) through formula (6. vice versa.g. m1 ) by ( 2 . 2 . namely. −m2 |J. (6. M ) corresponding to (6. M ) = (−1) 1 + 2 −J ( 2 . 2 .119) The corresponding symmetry property of the Clebsch-Gordan coeﬃcients is ( 1 .114) show a simple relationship to the Clebsch Gordan coeﬃcients ( 2 . m1 . M ) .117) are again related in a simple 2 .h. J (1) on the r. m1 .11) J = J (1) + J (2) . interchanging the operators J (1) and J (2) results in J = J (2) + J (1) .4 Symmetries of the Clebsch-Gordan Coeﬃcients The Clebsch-Gordan coeﬃcients obey symmetry properties which reﬂect geometrical aspects of the operator relationship (6. of (6. m1 |J. m2 |J. −M ) . m2 |J.118).115). (6. For the quantum mechanical addition of angular momenta the Clebsch Gordan coeﬃcients ( 1 . m2 ) and. m1 ) . m2 |J. M ) .s.

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

147. given by ˆ σ · r = σ1 sin θ cos φ + σ2 sin θ sin φ + σ3 cos θ .1 states χ 1 ± 1 2 2 2 [c.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.180) where the sign could depend on j.f. 1 |ˆ) = ((σ · p f (r)))Yjm (j ± 1 .160).181) 0 = j+ 1 2 4π Yj 1 (j + 1 . have proven (6.187) 1 2 . 1 |θ = 0.180) and. 1 |θ = 0. We want to demonstrate ﬁnally that the “−”-sign holds in (6. therefore. hence. 1 |ˆ) + f (r) σ · p Yjm (j ± 1 . (6. φ) . (5. One can conclude then ˆ σ·r = ± ˆ 0 1 1 0 (6. The result can also be stated in the compact form σ · r Yjm (j ± 1 . 1 |ˆ) r 2 (6.148) the particular ˆ 1 1 states Yj 1 (j − 2 .183) in case θ = 0 becomes in the standard representation with respect to the spin.224)] 1 0 σ·r = ˆ . 1 |θ = 0. Noting that p = −i r is a ﬁrst order 2 2 diﬀerential operator it holds ˆ ˆ ˆ σ · p f (r) Yjm (j ± 1 . According to (5. φ) = − Yj 1 (j + 1 . 1 |ˆ). ˆ (6.185) We have. ˆ 2 2 2 2 2 2 for θ = 0.182) 0 Since σ · r. 6.184) 0 1 space χ 1±1 2 2 one can conclude from (6. 1 |θ = 0. 2 |ˆ) and Yj 1 (j + 1 .188) .180). φ) 2 2 2 .180) and (6. ˆ (6. for θ = 0 (6. identiﬁed the sign of (6. 6. We want to determine ˆ its action on the wave functions f (r) Yjm (j ± 1 . 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.168 Addition of Angular Momentum and Spin 1 Yjm (j ± 2 . φ) and (5.181.182) σ · r Yj 1 (j − 1 . φ) 2 2 2 = − j+ 1 2 4π (6. (6.186) Here (( · · · )) denotes again conﬁnement of the diﬀusion operator ∂r to within the double bracket.174–5. 1 |ˆ) at θ = 0 are r r 2 2 2 2 Yj 1 (j − 1 . For this purpose we consider the application of σ · r in the case of θ = 0. 1 |ˆ) r r r 2 2 2 2 2 2 (6. Since f (r) is independent of θ and φ follows ˆ ((σ · p f (r))) = −i (( ∂r f (r) )) σ · r .

1 |ˆ) = 0 and σ = 2S/ . 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.6. 1 |ˆ) r 2 1 2 (6.189) The celebrated property of the Pauli matrices (5.197) 1 ˆ σ · p g(r) Yjm (j − 2 . r r r Using ∂2 (1/r) = −x2 /r3 .g. r 2 (6. we conclude the intuitively expected identity 1 ˆ ˆ p×r = − J . 1 |ˆ) r 2 (6.190) can be related to the angular momentum operator.53). h r (6. r Altogether we obtain.196) = i j+ ∂r + r 1 2 3 2 1 f (r) Yjm (j − 2 .191) (6. Corresponding results are obtained for the other components of p × r and.186) for both terms in (6.192) ˆ ˆ The operator p × r in (6. ∂3 x2 = x2 ∂3 we obtain ˆ ˆ ( p × r )1 h = −i 1 1 (x3 ∂2 − x2 ∂3 ) h = − J1 h r r (6.. To demonstrate this we consider one of its cartision components.156) this yields ﬁnally 1 ˆ σ · p f (r) Yjm (j + 2 . 1 |ˆ) r 2 (6. x3 x2 ˆ ˆ ( p × r )1 h = −i (∂2 − ∂3 ) h (6.193) r r 1 1 1 = −i (∂2 x3 − ∂3 x2 ) h − i h (x3 ∂2 − x2 ∂3 ) .194) ˆ ˆ where J1 is deﬁned in (5. e.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 (6.230) allows one to express ˆ ˆ ˆ ˆ ˆ σ · p σ · r = p · r + i σ · (p × r) . 1 |ˆ) . hence. ∂3 (1/r) = −x3 /r3 and ∂2 x3 = x3 ∂2 . using ∂r Yjm (j ± 1 . 1 |ˆ) r 2 (6.198) . 1 |ˆ) r 2 = i ∂r + −j r 1 g(r) Yjm (j + 2 .5: Spin–Orbital Angular Momentum States Using (6.195) . r 2 2 1 J ·S 2 1 ˆ σ · p f (r) Yjm (j ± 2 .155. 1 |ˆ) = r 2 2 ˆ i ∂r f (r) − f (r) σ · p σ · r Yjm (j ˆ 1 2 169 . 6.187) one obtains ˆ σ · p f (r) Yjm (j ± 1 . 1 |ˆ) = i [ ∂r + r + ] f (r) Yjm (j 2 r r Using (6.

199) We want to show that eqs. j 2 + 2j + 2 3 4 (6.204) (6.200) = (j + 1 )(j + 3 ).45). r ∂r2 r 2 (6. 1 |ˆ) r 2 2 r r2 r2 j 2 + 2j + 3 2 2 2 4 = [ −∂r − ∂r + ] f (r) Yjm (j + 1 . according to (6.146).201) 1 3 (j + )(j + ) Yjm (j + 1 .99) holds ˆ (σ · p)2 = − 2 2 = ∂2 J2 r + 2 .e. consider ﬁrst the case of the largest m-value m = j. 1 |ˆ) = J 2 Yjm (j + 1 . j − 1 . We note ˆ (σ · p)2 f (r) Yjm (j + 1 .2. r (6. 1 |ˆ) = Yj− 1 j− 1 (ˆ) χ 1 1 . according to (6. j) 2 2 2 2 (j − .197. Yjj−1 (j − 1 . − |j. 2 2 (6. Evaluation of Relevant Clebsch-Gordan Coeﬃcients We want to determine now the Clebsch-Gordan coeﬃcients (6. We begin with the coeﬃcients (6. i.199). 6. 1 |ˆ) = r 2 2 2j − 1 Yj− 1 j− 3 χ 1 1 + 2 2 2 2 2j 1 Y 1 1 χ1 1 2j j− 2 j− 2 2 − 2 (6. 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 . For m = j − 1 one can state. 1 |ˆ) r r 2 2 2 2 2 2 2 yields ˆ (σ · p)2 f (r) Yjm (j + 1 .144) and. as well as (6.170 Addition of Angular Momentum and Spin To demonstrate the validity of this key result we note that according to (5. 6. In this case holds. j + . For this purpose we use the construction method introduced in Sec. 1 |ˆ) r 2 2 i 3 ˆ = σ · p [ i∂r + (j + ) ] f (r) Yjm (j − 1 . 6. adopting the method in Sec.143. 6. 1 |ˆ) r 2 2 The Clebsch-Gordan coeﬃcients are then (j − 1 . . using (5.198). 6.101). r 2 2 2 2 2 2 − r 2 ∂r r + J2 r2 f (r) Yjm (j + 1 .144) for m = j. 1 |ˆ) = r 2 2 which agrees with (6. 1 |j.2.. 1 .151).143–6. j) 1 2 1 2 1 2 1 2 = = 1 0 (6. (6.205) which agrees with the expressions (6.202) Yjj (j − 1 .230.206) .203) (6.143. are consistent with this identity. 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 .43). 5. in fact. 1 |ˆ) r 2 2 r r2 and.

6.209) (j − 1 .202). 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.s. is obtained by applying the operator [c. j − 3 . 2 2 2 2 2j .h. 1 |j.144) for the Clebsch-Gordan coeﬃcients by induction.h. etc. as described in Sec. 6. The further Clebsch-Gordan coeﬃcients (· · · |jj − 2). 1 . − 1 |j.. Let us verify then the expression (6. 2 |j.137)] J− (tot) = J− + S− (6. Expression (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 . (6. and J− + S− to the r. 1 . 6. 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 . (· · · |jj − 3). are obtained by iterating the application of (6. 1 .143.s.210)] yields (j + m)(j − m + 1) Yjm−1 (j − 1 . j − 1) 2 2 2 2 which again agrees with the expressions (6.208) (6.211) to the l.207) (6.210).143. j − 1 . (6. (6.6.144) implies for j = m Yjm (j − 1 .212) 1 + 2j(j − m + 1) 1 2j(j − m + 1) Yj− 1 m− 1 χ 1 − 1 2 2 2 2 j+m−1 Yj− 1 m− 3 χ 1 1 + 2 2 2 2 2j j−m+1 Yj− 1 m− 1 χ 1 − 1 .f. m − 1) 2 2 2 2 = j +m−1 2j (6.143.5: Spin–Orbital Angular Momentum States The corresponding Clebsch-Gordan coeﬃcients are then 1 (j − 1 . 6.f.210) to (6. 2 2 2 2 2j (6.144) for m = j − 1.206).2. m − 3 . [c.

m. 2j + 2 (6.143). 1 |j + 1 . the relationship (6. proven (6. m1 . m1 . 1 . The Clebsch-Gordan coeﬃcients (6.218) where we have replaced the quantum numbers J. However. j2 m2 ) 2 .143) by applying the symmetry relationships (6. 2 . They are related in a simple manner to the ClebschGordan coeﬃcients j1 j2 j3 m1 m2 m3 = (−1)j1 −j2 +m3 (2j3 + 1)− 2 (j3 − m3 |j1 m1 . 2j + 1 (6. m + 1 . 1 . 2 . m − 1 . m.143. one can obtain expression (6. m2 1 (6.6 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 .172 Addition of Angular Momentum and Spin j −m+1 2j (j − 1 .215) which follows from (6. by the set j1 . − 1 |j. In this context Jclass and Jclass play the same role.214) yields (j.116. m3 ) = (−1) × 2 + 3 − 1 + 2 +m2 × 2 .214) 2 2 +1 ( 3 . to reﬂect in the notation the symmetry of these quantities. m) = 2 2 2 2 Similarly.120). j3 . one obtains (j + 1 . m2 | 3 . m) 2 2 2 2 2j + 1 (6. all three vectors Jclass . j2 . − 1 |j. m3 . The latter relationships applied together read ( 1 . 6.215). 1 . 1+1 3 For 1 (j. M.217) 2j + 2 (j + 1 .146) can be obtained from (6. m − 1) 2 2 2 2 = (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 . using (6. m1 ) (6. (1) (2) (−tot) Obviously. m + 1 .144) for j = m − 1.145) from (6. Jclass and Jclass play equivalent roles. j+m+1 . We have. m3 . 6. a higher degree of symmetry is obtained if one rather (−tot) (1) (2) (−tot) considers classically to obtain a vector Jclass with the property Jclass + Jclass + Jclass = 0. 1 . −m2 | 1 . m + 1 ) = 2 2 2 2 and. m2 . m + 1 ) = 2 2 2 j+m+1 .213) which is in agreement with (6.216) 6. 6.144) by induction. m1 . − 1 |j. 1 |j + 1 .143.144). hence. (1) (2) (−tot) The coeﬃcients which are the quantum mechanical equivalent to Jclass + Jclass + Jclass = 0 are the 3j–coeﬃcients introduced by Wigner.

−j1 + j2 + j3 . The two integer entries J − 1 − m2 and J − 2 + m1 in (6.21. 173 (6. m2 and j3 .219) (6. there exist even further symmetry properties.223) (Σ + 1)!n11 !n22 !n33 !n23 !n32 !)) n=0 n31 n32 n33 . m3 and with respect to sign reversals of all three values m1 .102) of the Clebsch-Gordan coeﬃcients. The Regge symbol also vanishes in case that two rows or columns are identical and Σ is an odd integer. 6. i. are identical to the integer arguments which enter the analytical expression (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 . 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.6: The 3j–Coeﬃcients We ﬁrst like to point out that conditions (6.222) m1 m2 m3 j1 + m1 j2 + m2 j3 + m3 The Regge symbol vanishes. as follows −j1 + j2 + j3 j1 − j2 + j3 j1 + j2 − j3 j1 j2 j3 j1 − m1 j2 − m2 j3 − m3 . These symmetries follow the equations j1 j2 j3 m1 m2 m3 = (−1)j1 +j2 +j3 = j2 j3 j1 m2 m3 m1 = (−1)j1 +j2 +j3 j2 j1 j3 m2 m1 m3 (6. However.102) are obtained each through the diﬀerence of two entries of the Regge-symbol. In case of a non-cyclic exchange of rows and columns the 3j–coeﬃcent assumes a prefactor (−1)Σ . m2 . To represent the full symmetry one expresses the 3j–coeﬃcients through a 3 × 3–matrix. j2 . one can exchange columns and one can reﬂect at the diagonal (transposition)..6. anti-cyclic exchange and of a sign reversal are stated. For this purpose one can use the symmetry transformations to place the smallest element into the upper left corner of the Regge symbol. j3 form the sides of a triangle and the condition is symmetric in the three quantum numbers. the Regge-symbol.g. with respect to alltogether 12 symmetry operations. states that j1 .221) j1 j2 j3 −m1 −m2 −m3 where the the results of a cyclic.218) relating 3jcoeﬃcients and Clebsch-Gordan coeﬃcients. The Regge symbol reﬂects a remarkable degree of symmetry of the related 3j–coeﬃcients: One can exchange rows. Because of its high degree of symmetry the Regge symbol is very suited for numerical evaluations of the 3j–coeﬃcents. m1 . = (6. The reader may note that the entries of the Regge-symbol. According to the deﬁnition of the 3j–coeﬃcients one would expect symmetry properties with respect to exchange of j1 . √ safe for the prefactor 2j3 + 1 which is cancelled according to the deﬁnition (6. for which no known classical analogue exists. the so-called triangle condition.220) The latter condition |j1 − j2 | ≤ j3 ≤ j1 + j2 . except when all elements are non-negative integers and each row and column has the same integer value Σ = j1 + j2 + j3 . In this way the values of 12 3j-coeﬃcients are related. discovered by Regge. m3 .e. j2 . These symmetry operations relate altogether 72 3j–coeﬃcients.

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 )! √ 10 (1 − m)! (1 + m)! (1 − m2 )! (1 + m2 )! 1 ≥ |m| ∧ 1 ≥ |m2 | ∧ 2 ≥ |m1 | (6.225) sn−1 . (6. m) δ(−m1 .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) by means of a symbolic manipulation package (Mathematica) (1m|2m1 1m2 ) = (−1)1+m+m1 δ(m. m2 ) √ 2 1 ≥ |m1 |) 2 (6. m1 + m2 ) (2 − m1 )! (2 + m1 )! √ 6 (2 − m)! (2 + m)! (1 − m2 )! (1 + m2 )! 1 ≥ |m2 | ∧ 2 ≥ |m| ∧ 2 ≥ |m1 | (6.1.230) 1 1 (1m| m1 m2 ) = 2 2 √ (−1)2m1 −2m2 3 δ(m.228) √ (−1)2m1 −2m2 7 δ(m.229) 1 1 (0m| m1 m2 ) = 2 2 i(−1)1−m2 δ(0.226) We like to state ﬁnally a few explicit analytical expressions for Clebsch-Gordan coeﬃcients which were actually obtained using (6.224) (6.227) (2m|2m1 1m2 ) = (−1)m+m1 (m + 2m2 ) δ(m.10752786393409395427444450130056540562826159542886 We also illustrate the numerical values of a sequence of 3j-coeﬃcients in Figure 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| (6.223-6.232) .231) Here is an explicit value of a Clebsch-Gordan coeﬃcient: 1 1 (70 2 −15 1 |120 −10 60 2 −5 1 ) = 2 2 √ √ √ 4793185293503147294940209340 127 142 35834261990081573635135027068718971996984731222241334046198355 0. (6.

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

. . . .6.6. M = −J.1: Prove Eqs. Exercise 6. (6.3: Two triplet states |1m1 (1) |1m2 (2) are coupled to an overall singlet state Y00 (1. .176 Addition of Angular Momentum and Spin 1 For the basis {YJM ( 1 . . Show that the probability of detecting a triplet substate |1m1 (1) for arbitrary polarization (m2 – 1 value) of the other triplet is 3 . . 1 .234) Exercise 6.6. (2( 1 + 2 ) + 1) ×(2( 1 + 2 ) + 1) Partitioning in smaller blocks is not possible. each of the blocks in (6. .. .6.234) . Exercise 6. . . J} (2 (2 1 2 + 1) (2 + 1) × (2 1 2 + 1) + 1) = (2| 1 − 2 | + 1) ×(2| 1 − 2 | + 1) (2| 1 − 2 | + 3) ×(2| 1 − 2 | + 3) (6.2: 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. 2 = 1. 2 |Ω1 . J = | 1 − 2 |. 1). . . 2.233.233) is further block-diagonalized as follows + 2. Ω2 ).

g. i. . . Note that in the examples mentioned the operator T would be deﬁned within the same representation as R(ϑ). . ∂x2 3 1 2 2 2 2 case (ii) the operator T could be a spin operator Sk deﬁned in (5. −k + 1. (k) (6.237) The operators T ∈ {Tkq . . (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) . This implies. Examples of Tensor Operators The multiplicative operators C∞ ( ) → C∞ ( ) Y ( ) = def Y ( ) (6. The operator T may also act on multi-particle states like Y 1 m1 (ˆ1 )Y 2 m2 (ˆ2 ). In fact. . 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 . for example.e. k} with the transformation property (6.223). that in case (i) T is an operator C∞ ( ) → C∞ ( ) acting on single particle wave functions. k} such that k Tkq = q =−k (k) Dq q (ϑ) Tkq . −k + 1.g. 6. q = −k. some of the r r examples considered below involve tensor operators T of this type. 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). S 2 = S1 + S2 + S3 or any other polynomial of Sk . .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 . (k) (6.237) are called tensor operators of rank k.6.236. Such operators T can be characterized through their behaviour under rotational transformations. e.7: Tensor Operators 177 6. e.238) .222) in case (ii) of single particle spin operators.42) in case (i) of the position representation of single particle wave functions or the operator (5. Rotations transform |ψ as |ψ = R(ϑ)|ψ and T |ψ as R(ϑ)T |ψ .g. the operator (5. e.236) In this equation Dq q (ϑ) denotes the rotation matrix Dq q (ϑ) = kq |R(ϑ)|kq .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. that T belongs to a family of operators {Tkq . q = −k.

Exercise 6.240) 1 where we have deﬁned ϑ± = 2 (ϑ1 iϑ2 ) and L± = L1 ± iL2 .7. but rather any rotation and any parametrization can be assumed. Exercise 6.1: Show that the following set of spin operators T00 T1±1 T10 T2±2 T2±1 T20 = 1 = 1 √ S± 2 S± 2 = S3 = = = (S3 S ± + S ± S3 ) 2 2 (3S3 − S 2 ) 3 are tensor operators of rank 0. 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 = (6.3. x3 . . The fact that these operators form tensor operators of rank 1.2: 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.7. 2. but are far easier to ascertain for any particular operator. 2.203). In fact.178 are tensor operators of rank k. 1.237) do not require that the rotation R−1 (ϑ) is expressed in terms of Euler angles according to (5. 3 follows from the transformation properties of the spherical harmonics derived in Section 1. x2 . For the following it is important to note that the rotation matrix elements (6. we will assume presently that the rotation is chosen as follows R(ϑ) = exp ( ϑ+ L+ + ϑ− L− + ϑ3 L3 ) (6. This choice of parametrization allows us to derive conditions which are equivalent to the property (6.6.239) These operators can be expressed in terms of the coordinates x1 .235 .237).

1 (6. 0)T in case of small ϑ+ . Tkq ] [J3 . The property (6. (6. hence. ϑ3 . (6.7: Tensor Operators 179 The conditions can be derived if we consider the property (6.245) (6.5: Consider a system of two spin.4: Is the 1-particle Hamiltonian 2 H = − 2 2m + V (|r|) (6.247) These properties often can be readily demonstrated for operators and the transformation properties (6. ϑ− .7. .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 .1 particles for which the ﬁrst spin is described by 2 the operator S (1) and the second spin by the operator S (2) . Similar equations can be derived for inﬁnitesimal rotations of the form ϑ = (0.246. 2 Exercise 6. Tkq ] = = δqq and by subtracting Tkq on both sides of the k ϑ+ kq |L+ |kq Tkq . Exercise 6.237) for transformations characterized by inﬁnitesimal values of ϑ+ .237) be assumed then.6.6: Form tensor operators of rank 1 in terms of the three components of acting on the space of 1-particle wave functions. We ﬁrst consider a rotation with ϑ = (ϑ+ .80) follows kq |L+ |kq = −i (k + q + 1)(k − q)δq and. 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) .242) from which follows by means of kq |1 1|kq equation ϑ+ [L+ . Tkq ] = = = Tkq+1 Tkq−1 q Tkq . (0.6.235 . q =−k q+1 (6. (k + q + 1)(k − q) (k + q)(k − q + 1) (6. Expressing the results in terms of the angular momentum operators J+ .7.7.243) From (5. 0)T . ϑ− .6.235 . 0. Exercise 6.248) a tensor operator? Exercise 6. J3 yields [J+ .247). 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 .6.246) (6.3: Derive Eqs. Tkq ] [J− . Tkq ] = −iTk q+1 (k + q + 1)(k − q) .244) [L+ . (6. 6. 0. ϑ3 )T .237) yields ﬁrst k R(ϑ) Tkq R −1 (ϑ) = q =−k kq |R(ϑ)|kq Tkq .7. J− .

hence. (6. Starting point of the derivation is the fact that the states Tkq | 2 m2 γ2 behave like angular momentum states of a composite system of two particles each carrying angular momentum or spin. (6. i.8 Wigner-Eckart Theorem A second important property of tensor operators Tkq beside (6.251) The corresponding property for Φ 1 m1 (k. m1 . .m2 2 m2 ) (q + m2 ) Tkq | 2 m2 γ2 | 2 m2 γ2 (6.e.247). J2 . in fact. | 1 m1 γ1 denotes an angular momentum (spin) state. i. 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.m2 ( 1 m1 |kq m1 q. construct states Φ 1 m1 (k. 1 m1 |Φ 1 m1 (k. 2 |γ2 ) is an eigenstate of J 2 with eigenvalue 2 1 ( 1 + 1). 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.e.180 Theory of Angular Momentum and Spin 6.253) Exercise 6.21) of Clebsch-Gordan coeﬃcients J3 Φ 1 m1 (k. To prove this we consider the transformation of Tkq | 2 m2 γ2 R(ϑ) Tkq | 2 m2 γ2 = R(ϑ) Tkq R−1 (ϑ) R(ϑ) | 2 m2 γ2 = q m2 Dq q Tkq Dm2 m2 | 2 m2 γ2 2 (k) ( ) (6. 2 |γ2 ) is an eigenstate of J 2 with eigenvalue + 1).252) ∼δm1 q+m2 = ( 1 m1 |kq 2 m2 ) Tkq Similarly. 2 |γ2 ) = Cδ 1 1 δm1 m1 2 1( 1 (6. J3 .e. which is characterized also by a set of other quantum numbers γ1 which are not aﬀected by the rotational transformation R(ϑ).18) are deﬁned through Φ 1 m1 (k. 2 |γ2 ) = ( 1 m1 |kq 2 m2 ) Tkq | 2 m2 γ2 . i.237) is that their matrix elements 1 m1 γ1 |Tkq | 2 m2 γ2 obey simple relationships expressed in terms of Clebsch-Gordan coeﬃcients. The hermitian property of J3 and J 2 implies that states Φ 1 m1 (k. J 2 | 2 m2 γ2 = 2 2( 2 + 1) | 2 m2 γ2 . 2 |γ2 ) = q. one can show that Φ 1 m1 (k. J3 are the generators of the rotation R(ϑ). One can. the stated property. J3 | 2 m2 γ2 = m2 | 2 m2 γ2 . 2 |γ2 ) which correspond to total angular momentum states.235 .8. Before we proceed we like to point out that the states | 2 m2 γ2 are also eigenstates of J 2 . 2 |γ2 ) are orthogonal to | 1 m1 in case of diﬀerent quantum numbers 1 .6.249) which demonstrates.1: Show that Φ 1 m1 (k. possibly the total angular momentum– spin state of a compositie system. behave like |kq | 1 m1 .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 . 2 |γ2 ) can be shown readily as follows using (6.250) q. These states according to (6.

γ1 ||Tk || 2 .258) We can then ﬁnally express the matrix elements of the tensor operators Tkq as follows 1 m1 .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 ) (6.6.257) which establishes the m1 –independence of 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) . γ2 is determined by applying (6.261) .254) = 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) ( 1 m1 |kq 2 m2 ) .259) The socalled reduced matrix element 1 . its m1 –dependence being expressed solely through a Clebsch-Gordan coeﬃcient.260) Exercise 6. γ2 (−1)k− 2 + 1 ( 1 m1 |kq 2 m2 ) (6. can be evaluated as easily as possible. γ1 ||Tk || 2 . q .32) Tkq | 2 m2 γ2 = and orthogonality property (6.256) and noting that the operator adjoint to J+ is J− .e. Using | 1 m1 + 1γ1 = 1 ( 1 + m1 + 1)( 1 − m1 ) J+ | 1 m1 γ1 (6. γ2 = 2 1 +1 1 m1 . m2 . γ1 |Tkq | 2 m2 .h. (6.8: Wigner-Eckart Theorem 181 In order to evaluate the matrix elements (6. m1 = q = m2 = 0. In order to express the m1 – independence explicitly we adopt the following notation 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) = (−1)k− 2+ 1 √ 1 2 1 +1 1 .g. One can then evaluate also the corresponding Clebsch-Gordan coeﬃcient ( 1 m1 |kq 2 m2 ) and determine 1 . say m1 + 1.s. To prove this property we consider 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) for a diﬀerent m1 value.2: Determine the matrix elements of the gradient operator d3 rF (r) G(r) of the type (6. γ1 ||Tk || 2 . γ1 ||Tk || 2 . 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.8. γ2 .259) to a combination of magnetic quantum numbers m1 . i.255) At this point the important property can be proven that 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) is independent of m1 . γ1 |Tkq | 2 m2 . (6. γ2 k− 2 + 1 √ 1 2 m2 ) (−1) 2 1 +1 = 1 . e. γ2 ( 1 m1 |kq (6. the matrix elements 1 m1 γ1 |Tkq | 2 m2 γ2 can be reduced to an m1 –independent factor.

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. φ) = Y −1m (θ. φ) +1m (θ. φ) = ( −m)( +m) (2 −1)(2 +1) Y +1m (θ. evaluate the matrix for m1 = q = m2 = 0 using cosθ Y ( +1−m)( +1+m) (2 +1)(2 +3) m (θ. φ) + Y −1m (θ. but a very useful exercise!) . (The necessary evaluations are cumbersome. For this purpose relate r operator T1q .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).

Switzerland. (7. and crystals. See. The potentials (7. The corresponding wave functions serve basis functions for multi-electron systems in atoms. The ﬁrst potential V (r) = 0 0 ≤ r ≤ R ∞ r > R (7.2) V (r) = 0 r > R The third potential V (r) = 1 mω 2 r2 2 (7.4) governs the motion of electrons in hydrogen-type atoms. Potential (7.2) serve as schematic descriptions of quantum particles. It leads to the stationary electronic states of the hydrogen atom (Z = 1). The potential (7. The second potential is of the square well type −Vo 0 ≤ r ≤ R . for example. molecules. 7. i. The fourth potential V (r) = − Z e2 r (7.1. in case of the so-called bag model of hadronic matter.3) describes an isotropic harmonic oscillator ..e. 7000 Chur. 1993) 1 183 .Chapter 7 Motion in Spherically Symmetric Potentials We describe in this section the stationary bound states of quantum mechanical particles in spherically symmetric potentials V (r). Talmi (Harwood Academic Publishers. φ.1) conﬁnes a freely moving particle to a spherical box of radius R.4) is by far the most relevant of the four choices. Poststrasse 22. Simple Models of Complex Nuclei / The Shell Model and the Interacting Boson Model by I. in potentials which are solely a function of r and are independent of the angles θ.3) describes the motion of a charge in a uniformely charged sphere and can be employed to describe the motion of protons and neutrons in atomic nuclei1 . Four examples will be studied. for example.

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. The expression of the kinetic energy 1 1 1 J2 p2 ˙ = m r2 + m r2 θ2 = m r2 + ˙ ˙ 2m 2 2 2 2mr2 and conservation of energy yield J2 1 m r2 + ˙ + V (r) = E .99)] 2 1 2 J2 2 2 = − ∂r r + .12) . i.11) (7. 2 .5) In the case of an angular independent potential angular momentum J = m r × v is a constant of ˙ motion. k = 1. d ˙ J = v × mv + r × mv = 0 + r × er (−∂r V (r)) = 0 .m (r) can be chosen as simultaneous eigenstates of the Hamiltonian ˆ operator H as well as of J 2 and J3 . 2.87) for Jk as well as the commutation property (5.184 Spherically Symmetric Potentials 7. .1 Radial Schr¨dinger Equation o ˙ mv = − er ∂r V (r) . (7. one can state ˙ J = m r2 θ .8) This property can also be proven readily employing the expression of the kinetic energy operator [c.m (r) (7. 2 . stationary states ψE.7) ˆ The correspondence principle dictates then for the quantum mechanical Hamiltonian operator H and angular momentum operators Jk ˆ [H. ˆ dt (7.10) (7. k = 1.e. Employing in this plane the coordinates r. Accordingly. ( + 1) ψE. (7. Jk ] = 0 .m (r) = = = E ψE. one can conclude ˙ Jk = {H.85– 5. J3 ψE. For this purpose one concludes ﬁrst that the conservation of angular momentum J implies a motion of the particle conﬁned to a plane..14) (7.13) .6) ˙ Since Jk is also equal to the poisson bracket {H.61) of J 2 and Jk . J ψE.m (r) .f.5) and r = v.9) − 2m 2m r 2mr2 the expressions (5.15) (7. 3 . In fact. 2 2mr2 (7. Jk } = 0 . . ˆ H ψE. 3. θ for the distance from the origin and for the angular position.m (r) m ψE. Jk } where H is the Hamiltonian. using (7. 2.m (r) . (5.m (r) . ˆ A classical particle moving in a potential V (r) is governed by the Newtonian equation of motion (7. the time variation of J can be written. (7.

One can write (7.18) = vE. (7. i. r(t) is determined the angular motion follows from (7. .m (r) .m (r) = 0.21) Veﬀ (r) φE (r) = = V (r) + r vE. for the wave function of a quantum mechanical particle a diﬀerential equation which governs solely the r-dependence. .10) − 1 2 ∂ r + 2m r r 2 2 ( + 1) + V (r) − E 2mr2 . ( + 1) 2mr2 .9) one can write the stationary Schr¨dinger o equation (7.. equations (7.m (r) = 0. (r) = 0 . 2 (7. This barrier. (7. Employing the kinetic energy operator in the form (7.24) .13).20) in the form of the one-dimensional Schr¨dinger equation o 2 − where 2m 2 ∂r + Veﬀ (r) − E φE (r) = 0 .m (r) Y m (θ.20) by −2mr/ 2 yields the so-called radial Schr¨dinger equation o 2 ∂r − ( + 1) − U (r) − κ2 r2 r vκ. in the present case. φ) are the angular momentum eigenstates deﬁned in Section 5. .16) m 2mr2 Once. .e. (7.20) Since this equation is independent of the quantum number m we drop the index m on the radial wave function vE. using (7. (7.11. mr2 (t) (7.22) which is the original potential 2 ( +1) V (r) with an added rotational barrier potential 2mr2 .17) In analogy to the classical description one can derivem.m (r) and E .m ψE.22) (7. .m vE.10). Multiplying (7.12) are obeyed and one obtains for (7. together with the original potential. by integration of ˙ θ = J .m (r) describes the radial motion as a one-dimensional motion in the interval [0.11). . . ∞[ governed by the eﬀective potential (7. but can also trap particles in the latter space giving rise to strong scattering resonances (see Section ??). φ) (7.m (r) 2 .m (r) 1 2 J2 ∂r r + + V (r) − E 2m r 2mr2 the functional form ψE.m .4.19) where Y m (θ.23) This demonstrates that the function r vE.1: Radial Schr¨dinger Equation o 185 This is a diﬀerential equation which governs solely the radial coordinate. can exclude particles from the space with small r values.7. It can be solved by integration of 1 2 2 J2 r = ± ˙ E − V (r) − . (7. 7. − Adopting for ψE.

m (r) ∼ √ 4π B δ(x1 )δ(x2 )δ(x3 ) . the term Br−( +1) would contribute |B|2 0 dr r−2 +1) (7.182) for Y00 . . ”Classical Electrodynamics.30) and.20) the index E by the equivalent index κ. For r → 0 one may assume that the term ( + 1)/r2 becomes larger than the potential U (r). (r) = 0 . in particular.D. 2nd Ed. at r = 0 is found in the excellent monographs Quantum Mechanics I.26) 2m 2 In case E < 0. hence. 4π|r| (7. The radial part of this integral is ∞ 0 2 dr r2 vκ. In this case the solution is governed my ( + 1) 2 r vκ. New York. . accordingly.27) ∂r − r2 and. II by A. Boundary Conditions In order to solve (7.29) Only the ﬁrst term is admissable. assumes the general form r vκ. Jackson (John Wiley.186 where we deﬁned U (r) κ2 = = − − 2m 2 Spherically Symmetric Potentials V (r) E (7. 2 2 2m 2 ψE. ψE. for example.31) to the complete wave which. Galindo and P. r → 0 (7. 1975). (r) ∼ A r + B r− −1 +1 + B r− (7.33) A detailed discussion of the proper boundary conditions. Berlin. using the expression (5. (r) (7.32) The total kinetic energy resulting from this contribution is.20) one needs to specify proper boundary conditions2 . We replaced in (7.25) (7. 1990) 3 We refer here to the fact that the function Φ(r) = 1/r is the solution of the Poisson equation 2 = −4πδ(x)δ(y)δ(z). For = 0 the contribution of Br−( function is.m (r) ∼ √ B . . (7. for > 0 is not integrable. see.” by J. This follows for > 0 from consideration of the integral which measures the total particle density. Pascual (Springer. according to a well-known result in Classical Electromagnetism3 . κ assumes real values.28) (7. (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).

h.98) by ∞ eik3 x3 = =0 b j (kr) P (cos θ) ..50).102) −1 d −1 +1 −1 dx −1 (x2 − 1) .45) and one given by (7. z = kr.194 Generating Function of Spherical Bessel Functions Spherically Symmetric Potentials The stationary Schr¨dinger equation of free particles (7. The orthogonality properties (5.s. can be written exp(ikr cos θ). and using the Rodrigues formula for Legendre polynomials (5. hence. −1 .43) has two solutions. One can show d −1 2 (x − 1) ∼ (x2 − 1) × polynomial in x dx −1 (7.50). Since the spherical harmonics Y 0 (θ. 7. φ) for non-vanishing m has a non-trivial φ-dependence as described by (5.103) and. in case of a free particle moving along the x3 -axis one expands eik3 x3 = .99) +1 d cos θ eikr cos θ P (cos θ) = b j (kr) −1 2 . φ).179) yield from (7. φ) . This holds for −1 and one can conclude +1 consecutive integrations by part d izx e dx (7.99) We want to determine the expansion coeﬃcients b .e.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.s.h. One can expand the former solution in terms of solutions (7. according to (5. the surface term ∼ [· · ·]+1 vanishes.178) are given in terms of Legendre polynomials P (cos θ) one can replace the expansion in (7. of (7. namely.100) Deﬁning x = cos θ. 2 +1 (7.150) one obtains +1 +1 1 ∂ dx eizx P (x) = dx eizx (x2 − 1) . i.98) The l. (7. (7. In this case the expansion on the r. the wave function does not depend on φ.44.m a mj (kr) Y m (θ.98) does not involve any non-vanishing m-values since Y m (θ.106). one given o by (7. For example. (7.104) dx eizx P (x) −1 = (−1) 2 ! (iz) = 2 ! +1 dx (x2 − 1) −1 +1 dx (1 − x2 ) eizx .

−1 (7. 2 +1 −1 dx (1 − x2 )ν− 2 eizx . in particular.105) and (7. 1 (7.s.106) Employing (5. (7.100) gives b j (kr) 2 (iz) = 2 +1 2 ! +1 195 dx (1 − x2 ) eizx . as the generating function of the spherical Bessel functions.114) .117) one can write the r. after insertion into (7. of (7.105) This expression allows one to determine the expansion coeﬃcients b .110) One refers to the l. Comparision with the l.2: Free Particle Comparision with (7.107) 1 2 (2 )! 1 · 3 · 5 · · · (2 + 1) 2 + 1 1 · 2 · 3 · 4 · · · (2 − 1) · 2 (7. 1 (7.7.111) Employing (7.105) must hold for all powers of z. Integral Representation of Bessel Functions Combining (7.108) where the last factor is equal to unity. for the leading power x [c.69)] b z 2 (iz) = 1 · 3 · 5 · · · (2 + 1) 2 + 1 2 ! +1 dx (1 − x2 ) .109) results in the integral representation of j (z) j (z) = (z) 2 +1 ! +1 dx (1 − x2 ) eizx .h. z i or i (2 + 1) z (2 )! 2 2 2 +1 2 ! [1 · 3 · 5 · · · (2 − 1)] 1 (7. ∞ (7.h.s. The identity (7.112) We want to consider the expression Gν (z) = aν z ν fν (z) where we deﬁne fν (z) = C (7. −1 (7. (7.109) eikr cos θ = =0 i (2 + 1) j (kr) P (cos θ) .106) yields ﬁnally b = i (2 + 1) or.s.113) dt (1 − t2 )ν− 2 eizt .99).96) one can express this. using ν = Jν (z) = √ 1 πΓ(ν + 1 ) 2 z 2 ν + 1.f. −1 (7.h.

1 j = 1. 7.121) (7. Gν (z).116) fν (z) = − or fν (z) = − From this we can conclude fν (z) + 2ν + 1 fν (z) + fν (z) = 0 . t2 of the integration paths C. t2 .4: Prove (7. 2.120) imply also the property (2ν + 1) fν (z) + z fν+1 (z) = 0 . Exercise 7.122) (7. 1 (7.118) z 2ν + 1 fν (z) + fν (z) . C 1 (7.2.123) Insertion of these identities into Bessel’s equation leads to a diﬀerential equation for fν (z) which is identical to (7.120) such that we can conclude that Gν (z) for proper integration paths is a solution of (7. 3.93) as long as the integration path in (7. 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. To prove this we note fν (z) = i C dt (1 − t2 )ν− 2 t eizt .119) We note that equations (7. 4. (7.113) obeys the Bessel equation (7.116) In case that the endpoints of the integration path C satisfy (1 − t2 )ν+ 2 eizt one can write (7. for properly chosen endpoints t1 .114.120) z 2ν + 1 dt (1 − t2 )ν− 2 eizt + C C 1 1 t2 t1 = 0 (7. We consider now the functions u(j) (z) = √ 1 πΓ(ν + 1 ) 2 z 2 ν Cj dx (1 − x2 )ν− 2 eizx . In fact.93) for arbitrary ν.93). We can now demonstrate that Gν (z) deﬁned in (7. z (7.121). obeys Bessel’s equation (7.115) Integration by part yields fν (z) = − i 2ν + 1 (1 − t2 )ν+ 2 eizt 1 t2 t1 − z 2ν + 1 dt (1 − t2 )ν+ 2 eizt .117).196 Spherically Symmetric Potentials Here C is an integration path in the complex plane with endpoints t1 . (7.124) .117) dt (1 − t2 )ν− 2 t2 eizt 1 (7.113) satisﬁes (7.

therefore. Following convention.7.130) holds Jν (z) = (1) 1 2 1 (1) (2) Hν (z) + Hν (z) .125) 0 ≤ s < ∞ (7. using t = 1 + is. for Rez > 0 and ν ∈ R obey (7. Accordingly.93).135) (7. 1 (7. (7.133) = e i(z−πν/2−π/4) √ 2 zν π Γ(ν + 1 ) 2 2 zν π Γ(ν + 1 ) 2 √ √ ∞ 0 ds [s(1 + is/2)]ν− 2 e−zs . one can state Jν (z) = − u(2) (z) + u(4) (z) . Comparision with (7.132) For Hν (z) one derives.127) 0 ≤ s < ∞ (7.134) Similarly. hence.135) are known as the Poisson integrals of the Bessel functions.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. (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 (7.124) vanishes along the whole path C3 and. u(3) (z) ≡ 0.117) and. 1 (7. the integral expression (1) Hν (z) 1 1 1 1 (7.130) We note that the endpoints of the integration paths C2 and C4 .131) According to (7. . (7. (7. dt = i ds. u(2) (z) and u(2) (z) are both solutions of Bessel’s equation (7. one can derive (2) Hν (z) = e −i(z−πν/2−π/4) ∞ 0 √ ds [s(1 − is/2)]ν− 2 e−zs .126) −1 ≤ s ≤ 1 (7. and (1 − t2 )ν− 2 eizt = [(1 − t)(1 + t)]ν− 2 eizt = [−is(2 + is)]ν− 2 eiz e−zs = ei(z−πν/2+π/4) 2ν− 2 [s(1 + is/2)]ν− 2 e−zs . we introduce the so-called Hankel functions (1) Hν (z) = −2 u(2) (z) .112) shows u(1) (z) = Jν (z).129) The integrand in (7.128) C = C1 ∪ C2 ∪ C3 ∪ C4 is a closed path.134) and (7. Since the integrand in (7.

136) (1) 2 (z) = 0 ds [s(s ± is/2)] e−zs .198 Asymptotic Behaviour of Bessel Functions We want to obtain now expansions of the Hankel functions H Spherically Symmetric Potentials the expansions converge fast asymptotically. we obtain H (1) 2 1 +2 (1) 2 (z) = r=0 ( + r)! ! r!( − r)! ± i 2 r 1 z +r+1 (7.142) and h(z) = H (2) (z) +1 2 one can identify the constant c by using the leading terms in the expansions (7. (7.. For such solutions g(z).140) Bessel Functions with Negative Index Since ν enters the Bessel equation (7. Hν (z) as well as H−ν (z) are solutions of this equation. (7. We employ for this purpose the Poisson integrals (7. hence. h) must be a non-vanishing function.140). πz (7. The general solution of (7.139) (z) = 2 ±i[z − ( e πz +1) π ] 2 r=0 ( + r)! r!( − r)! i ± 2z r (7.2) (z) +1 2 in terms of z −1 such that (z) = e±i[z − ( ∞ +1) π ] 2 2z z ( 1 ) f 2 (z) π ! (7. converge fast for |z| → ∞. As a second order diﬀerential equation the Bessel equation has two linearly independent solutions.138) The formula for the Laplace transform of sn leads to f and. i. the Wronskian W (g. One obtains W (z) = − 4i .137) (z) = r=0 r ± i 2 r 0 ∞ ds s +r e−zs .93) holds the identity 1 W = − W z (7.141) is o W (z) = − In case of g(z) = H (1) (z) +1 2 c .135) which read for ν = + 1 2 H where f The binomial formula yields f (1) 2 (1) 2 1 +2 (1.143) . h(z) to be linearly independent. For the Wronskian connected with the Bessel equation (7.134) and (7.e.141) (1) (1) the derivation of which follows the derivation on page 192 for the Wronskian of the radial Schr¨dinger equation. z (7.93) only as ν 2 .

140). From (7. 7. (7.152) (7. linearly independent. H−ν (z) = A H (1) (1) (z) +1 2 One can then expand + BH (2) (z) +1 2 .90.2) (2) +1) 2 (1) +1) 2 (z) = i (−1) H (1) (z) +1 2 . 7.147) this can be written n (z) = 1 2i h (1) (z) − h (2) (z) .153) . (7. 7.148) holds for the o regular spherical Bessel function j (z) = 1 (1) (2) [ h (z) + h (z) ] .151) Equations (7.2) (z) +1 2 199 are. 7.146) (z) = −i (−1) H (2) 1 (z) +2 .149. 2z 2 (1.145) This equation can hold only for B = 0 and A = exp[i( + 1 )π].147) (z) = π (1.2) (7.148) Following the arguments provided above (see page 193) the functions h (z) are solutions of the radial Schr¨dinger equation of free particles (7.7. 7. H (1. 7.146.e. 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| → ∞ .149) We want to establish also the relationship between h (z) and the irregular spherical Bessel function n (z) deﬁned in (7.2: Free Particle i.97. (7. 2 (1.151) are equivalent to h h (1) (2) (z) (z) = = j (z) + i n (z) j (z) − i n (z) . (7. B can be obtained from the asymptotic expansion (7. We conclude 2 H−( Simarly. one can show H−( Spherical Hankel Functions In analogy to equations (7.2) (7.144) The expansion coeﬃcients A.97).97) one deﬁnes the spherical Hankel functions h (1.132) follows n (z) = 1 2 π (−1) 2z +1 H−( (1) 1 +2) (z) + H−( (2) +1) 2 (z) .2) H + 1 (z) .150) According to (7. (7. According to (7.53).96.132.. (7. (7. in fact.

155) (z) and To determine j (z) and n (z) we note that for z ∈ R the spherical Hankel functions h (2) h (z). as given by (7. (1) (7.164) .148). at large |z|.163) (7.154) The leading term in this expansion.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).148) one obtains readily h (1) 2 (1. = −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) . Hence.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. 7.2) (z). = sin(z − π/2) and sin[z − ( + 1) π ] . are j (z) n (z) = = (7.140.159) (7.162) The leading terms in these expansions. 2 2z z (7. From (7. ne ll(z) = Im[h (1) (z)] . z (7. at large |z|.161) (7.160) Employing cos[z − ( + 1) π ] .154) the identities j (z) n (z) = = sin[z − ( + 1) π ] cos[z − ( + 1) π ] 2 2 p (z) − q (z) z 2z 2 cos[z − ( + 1) π ] cos(z − π/2) 2 q (z) + p (z) . (7. are complex conjugates. j (z) (z) = ( i) z +1 e±iz r=0 ( + r)! r!( − r)! ± i 2z r (7.156) ( + r)! r!( − r)! ( + r)! r!( − r)! i 2z r [ /2] = r=0 ( + 2r)! (2r)!( − 2r)! −1 4z 2 r (7. is h (1) 2 (z) = ( i) z +1 e±iz .140) and (7. it follows z ∈ R : Using p (z) = Re r=0 j (z) = Re[h (1) (z)] . 2z 2 z sin(z − π/2) z cos(z − π/2) − .158) one can derive then from (7.

124.169) (7. (1. 7. For this purpose we need to demonstrate only that the (1.161.2) To demonstrate that (7. One can combine (7.171. 7.171) holds for g (z) = h (z) we employ (7.166) (7.171) −1 (z) (7.175) Using Γ( + 1) = !.7.2) (z) = −2 π 1 √ 2z πΓ(ν + 1 ) 2 z 2 ν C2.2) relationships hold for g (z) = h (z).149.174) We want to prove these relationships. 1 (7. (7.4 dx (1 − x2 )ν− 2 eizx .148) and express g (z) = h (1. 1.170) Recursion Formulas of Spherical Bessel Functions The spherical Bessel functions obey the recursion relationships g g +1 (z) +1 (z) = = (1. 7.167) (7.151) follows then that the relatiosnhips hold also for g (z) = j (z) and g (z) = n (z).165) (7. One obtains for = 0. From the linearity of the relationships (7. 2 j0 (z) n0 (z) j1 (z) n1 (z) j2 (z) n2 (z) = = = = = = sin z z cos z − z sin z cos z − z2 z cos z sin z − 2 − z z 3 1 3 − sin z − 2 cos z 3 z z z 3 1 3 − 3 + cos z − 2 sin z z z z (7. deﬁning a = −1 and employing fν (z) as deﬁned in (7.171–7.131. 7.173) A (z) = z ( +1) z2 +2 z (7.168) (7.114) we can write g (z) = a z f 2 ! + 1 (z) 2 .162) allow one to provide explicit expressions for j (z) and n (z). j (z) and n (z). 7.2: Free Particle Expressions for the Spherical bessel Functions j (z) and n (z) 201 The identities (7.2) g (z) − g (z) z 2 +1 g (z) − g z (7.176) .172) where g (z) is either of the functions h obtain the recursion relationship g g (z).174) and from (7.172) to +1 (z) +1 (z) = A (z) 1− g (z) g (z) −1 (7.

7.e. The second component of (7.180) yields g +1 (z) = g (z) − ( + 2) g (z) + z2 +2 g (z) .179) from which follows (7.182).6: Employ the recursion relationship (7.181) Using this identity to replace the second derivative in (7.173).121). is equivalent to (7.179) g (z) = − z2 g (z) + g (z) − g +1 (z) z .202 The derivative of this expression is g (z) = Employing (7.174) we start from (7.165).171). (7.173) is equivalent to (7.172). Exercise 7. (7. z (7.5: Provide a detailed derivation of (7.2. in fact. j0 (z) using (7. (7. 7. and (b) n1 (z).172).179) leads to (7.178) (7.180) Since g (z) is a solution of the radial Schr¨dinger equation (7.182) Replacing all ﬁrst derivatives employing (7.173. z2 (7. j2 (z) from j0 (z).171).174) to determine (a) j1 (z). 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) . n0 (z) using (7.172) we diﬀerentiate (7. Exercise 7.171). together with (7. f yields.58) it holds o 2 g (z) = − g (z) + z ( + 1) g (z) − g (z) . .. n2 (z) from n0 (z). To prove (7. i.166).2.173.177) (z) = − z f 2 +2 + 3 (z) 2 . In order to prove (7.

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

alter the potentials.1) the term qδ(r − ro (t)) and to the current density ˙ J(r.4) Lorentz Force A classical particle with charge q moving in the electromagnetic ﬁeld experiences the so-called Lorentz force q[E(r. In the non-relativistic limit holds p ≈ mr and (8. t) and A(r. solves implicitly (8. called the vector potential.9) (8. Jackson (John Wiley & Sons. t) Interaction of Radiation with Matter = 4 π ρ(r.25). called the scalar potential. (8. t) − ∂t E(r. As is well known. t) · B(r. t) = × A(r. in turn. t) −→ 1 A(r. The particle. The following substitutions.204 Maxwell equations1 · E(r.3) (8. contributes to the charge density ρ(r. t) − ∂t χ(r. Scalar and Vector Potential Setting B(r. New York. From this follows E(r. the relationship between ﬁelds and potentials is not unique. t) −→ V (r. called gauge transformations.5) above agrees with the equation of motion as given in (1. t) + ∂t A(r. t) . t) is a scalar function. t) + ∂t A(r. (8.3) reads then × E(r. t)] and. by J. t) = 4 π J(r. t) which is solved by E(r.10) (8. t) = 0 = 0. D. .6) for some vector-valued function A(r.4). t] + v × B[ro (t). t) + ∂t B(r.2) the term q r˙o δ(r − ro (t)). (8. t) in (8. 2nd Edition.5) where p is the momentum of the particle and ro (t) it’s position at time t. t) in (8. The reader is referred to the well-known textbook ”Classical Electrodynamics”. t) = − V (r. obeys the equation of motion d p = q dt E[ro (t).7) Gauge Transformations We have expressed now the electric and magnetic ﬁelds E(r. t) through the scalar and vector potentials V (r. accordingly. t). but leave the ﬁelds unaltered: A(r. t) . Equation (8.8) = 0 (8. t) (8. t) − ∂t A(r.11) V (r. 1975) for a discussion of these and other conventional units. t) = − V (r. t) . t) + χ(r. We assume so-called Gaussian units. t) and B(r. t) + v × B(r.1) (8. the two homogeneous Maxwell equations hold × E(r. t) (8. t) × B(r. t) where V (r. t). t] (8.2) In addition.

t) = ∂t = 4 π ρ(r. t). in fact. t) . t) = 0 .20) . In fact.1) · − V (r. For this purpose we examine the oﬀending term ∂t V (r. In fact. the second inhomogeneous Maxwell equation. t) . (8.12) yields then the Poisson equation 2 V (r. t = 4 π J(r. t) together with (8. and a component due to the remaining currents.17) A(r.15) d3 r ρ(r . t) + ∂t V (r.13) · A(r. t) and the current due to moving net charges.12) The corresponding gauge is referred to as the Coulomb gauge. t) = 0 the solution is given by the Coulomb integral V (r.9) for the ﬁelds results in × The identity × × A(r. t) can be obtained employing (8. t) Using · ∂t A(r. The vector potential A(r. t) . t) the condition · A(r. 4π (8. t) = − 4 π J(r. t) (8. t) and V (r.8. this potential results from inserting (8. t) − ∂t V (r.16) and the vector potential are uncoupled. t) − ∂t A(r. the gauge freedom allows us to impose on the vector potential A(r. t) − ∂t A(r.6) and (8. (8. Such description can. (8. t) + ∂t A(r. (8. t) = · A(r.9) into (8.19) Unfortunately.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. be achieved. Using the expressions (8.14) In case of the boundary condition V (r. t) .19) couples the vector potential A(r.2). and the former the Coulomb interactions in the unperturbed bound particle system. t) |r − r | (8. such that the latter describes the electromagnetic radiation. equation (8.16) This is the potential commonly employed in quantum mechanical calculations for the description of Coulomb interactions between charged particles. t) = 1 ∂t V (r. t).12) leads us to 2 2 A(r. (8. t) in (8. t) . t) − 2 × A(r. t) = − 4 π ρ(r. a name which is due to the form of the resulting scalar potential V (r. t) = Ω∞ for r ∈ ∂Ω∞ (8.19) and deﬁne J (r. One would prefer a description in which the Coulomb potential (8.18) together with condition (8.

t) + ∂t ρ(r. The deﬁnitions of J and Jt applied to (8. t) = 0 (8.16) yield ﬁnally the electric and magnetic ﬁelds. t) + ∂t ρ(r.26) turns into the well-known wave equation 2 2 A(r.25) one refers to J and Jt as the longitudinal and the transverse currents. t) = − ∂t ρ(r. exist in the space considered. t) as the current due to the time-dependence of the charge distribution ρ(r. t) contributes solely an electric ﬁeld component E (r. t) = 0). Let J(r. such current densities exist.26) and (8. t) contributes an electrical ﬁeld component as well as the total magnetic ﬁeld Et (r. V (r.28) (8. t) Bt (r. t). t) . t) = or = 2 Interaction of Radiation with Matter (8. Presently. t) − ∂t A(r.. In this case (8. t) These ﬁelds are obviously divergence -free (e. t) (8. t) . 8.23) This continuity equation identiﬁes J (r. i.12) and the Coulomb potential (8. no divergence-free currents. The vector potential determined through (8.22) 1 ∂t 4π · J (r. in a ring-shaped antenna which exhibits no net charge. t) (8. (8. t) = 0 . (8. t) = 0 and from this follows · Jt (r.24) (8. t) holds.2 Planar Electromagnetic Waves The current density Jt describes ring-type currents in the space under consideration.21) and (8.21) ∂t and the Poisson equation (8. respectively. t) be the total current of the system under investigation and let Jt = J − J . t) = − V (r. For J also holds the continuity equation · J(r.. t) = × A(r. the name transverse ﬁelds.29) · Et (r.30) . (8. t) = 0 .e. (8. for example.206 For the curl of J holds × J (r.19) yield 2 2 A(r.g. = − ∂t A(r. V (r. t) = − 4 π Jt (r.26) This equation does not couple anymore scalar and vector potentials. t) − ∂t A(r. we want to assume that no ring-type currents. yet a current. hence. A(r. For the divergence of J (r. hence.27) which is obviously curl-free ( × E (r. the name longitudinal electric ﬁeld.25) Because of properties (8.14). t) = 0 . t) = 0). using ∂t · J (r.

(8.8. t) = i k × A(r. Time average over one period 2π/ω yields S(r.32. t) = ± ω2 ˆ |Ao |2 k . ω2 = |Ao |2 (8.29). This ﬂux is given by 1 S(r. ˆ We want to characterize now the radiation ﬁeld connected with the plane wave solutions (8.34. t) = Re Et (r. according to (8.34) (8. Et (r.33) ωt) (8. 8. accordingly.2: Planar Electromagnetic Waves 207 which describes electromagnetic ﬁelds in vacuum. The corresponding electric and magnetic ﬁelds.35).31) and the resulting ﬁelds (8.35) one obtains S(r.32) holds. 8.158). A complete set of solutions is given by the so-called plane waves A(r. (8. t) = ±i ω A(r. The Coulomb gauge condition (8.39) 8π The deﬁnition incoming waves and outgoing waves is rationalized below in the discussion following Eq.37) ˆ ˆ where k is the unit vector k = k/|k|. t) = Ao u exp i(k · r ˆ where the dispersion relationship |k| = ω (8.12) yields u·k = 0. ˆ (8.34. The latter vector describes the direction of propagation of the energy ﬂux connected with the plane wave electromagnetic radiation. t) × Re B(r. 8. 2 . t) . (8. t) and Bt (r. see also the comment below Eqs. is orthogonal to k.35) The vector potential in (8. Note that in the units chosen the velocity of light is c = 1. t) in (8.38) ˆ In this expression for the energy ﬂux one can interprete k as the propagation velocity (note c = 1) and. 8π (8. Obviously.31. are Et (r. (8. Here the “-” sign corresponds to so-called incoming waves and the “+” sign to outgoing waves2 . 8. t) . t) Bt (r. the constant k is referred to as the wave vector.31) u is a unit vector (|ˆ| = 1) which.36) 4π Using the identity a × (b × c) = b (a · c) − c (a · b) and (8. 8.39). obviously.35) are complex-valued quantities.34. there exist two linearly ˆ u independent orientations for u corresponding to two independent planes of polarization.28. hence.31). 8.38. t) = ± ω2 ˆ |Ao |2 k sin2 ( k · r − ωt ) 4π (8. at each point r and moment t. are orthogonal to each other and are both orthogonal to the wave vector k. In applying the potential and ﬁelds to physical observables and processes we will only employ the real parts. 8.

is p − q A(r. |k| = ω . u · k = 0 . ˆ (8.38) implies that for incoming waves.44) . t) is real 1 and that Et (r. deﬁned below Eqs. (8. t) . Bt (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. whereas in the case of outgoing waves the energy is transported in the direction of k. (8.16.40) allows one to express then the ﬁeld ˆ amplitudes 8πNω Ao = . We are left with a classical Hamiltonian function which has an obvious quantum mechanical analogue ˆ H = ˆ p − q A(r. i (8.29) together with the potentials (8. hence. respectively.28. we will neglect the respective terms in the Hamiltonian (8. t) 2m 2 + qV (r) . (8. t). Comparision of (8. A ‘poor man’s’ quantization of the ﬁeld is possible at this point by expressing the energy density (8. The wave function Ψ(r. 8.41) ωV Inserting this into (8.39) through the density of photons connected with the planar waves (8. 8.40) = V It should be pointed out that Nω represents the number of photons for a speciﬁc frequency ω.208 Interaction of Radiation with Matter as the energy density. The sign in (8.31) allows one ﬁnally to state for the planar wave vector potential A(r.3 Hamilton Operator The classical Hamiltonian for a particle of charge q in a scalar and vector potential V (r) and A(r. These photons each carry the energy ω. t). (8.39) and (8. We assume that the energy content of the ﬁelds is not altered signiﬁcantly in the processes described and. If we consider a volume V with a number of photons Nω the energy density is obviously Nω ω .42) 8.8. A correct description of the electromagnetic ﬁeld requires that the ﬁeld be quantized.45) .31). the energy of the plane wave is transported in the direction of −k. t) = 8πNω u exp i(k · r − ωt) ˆ ωV . Note that E (r.43) Here the ﬁelds are deﬁned through Eqs. 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) = H Ψ(r. The integrals express the integration over the energy density of the ﬁelds.31.43).31). t) 2m + 1 8π Ω∞ 2 H = + qV (r) 1 16π d3 r |Et |2 + |Bt |2 Ω∞ d3 r E 2 + .32). (8. 8.27. (8.

(8.47) is equivalent to 2 ˆ p − qA i ∂t eiqχ(r. (8. t) ˆ p eiqχ(r.48) implies that the gauge transformation (8.47.t)/ [ i ∂t − q((∂t χ)) ] Ψ(r. The idea just stated can be generalized by noting that multiplication by a phase factor eiqχ(r.46) the ﬁelds enter. t) = + qV eiqχ(r.10.t)/ Ψ(r. by the group 3 The eﬀect on other expectation values is not discussed here. t) = eiqχ(r. It should be noted.t)/ should not aﬀect a system and that. t) = + qV Ψ(r.44.t)/ constitutes a unitary transformation of a scalar quantity. . have no eﬀect on the motion of the particle described by (8.49) (8.47) i ∂t Ψ(r. i. t) .t)/ . does not change expectation values which contain the probability densities3 .51) 2m the potentials A(r. in the Schr¨dinger equation o 2 ˆ p − qA i ∂t Ψ(r. hence. t) enter whereas in the classical equations of motion ˙ m¨ = q E(r. t) by a local and time-dependent phase factor eiqχ(r.11) aﬀect the quantum mechanical description.48) 2m For this purpose one notes i ∂t eiqχ(r. An important conceptual step of modern physics has been to turn the derivation given around and to state that introduction of a local phase factor eiqχ(r.t)/ Ψ(r.8. Elementary constituents of matter which are governed by other symmetry groups. t) .e. 8. such phase factor does not change the probability density |Ψ(r. hence. not to phenomenological interactions like the molecular van der Waals interaction.t)/ Ψ(r. Obviously.46).g. t)|2 and.10. One can show that (8. 8.3: Hamilton Operator 209 Gauge Transformations It is interesting to note that in the quantum mechanical description of a charged particle the potentials V (r.t)/ Ψ(r. t) and A(r. This leads to the question in how far the gauge transformations (8. t) and V (r. (8. t) + q r × B(r.. 8.10. accordingly. an element of the group U(1). that this principle applies only to fundamental interactions. t) ˆ = eiqχ(r. t) = 2m where ((· · ·)) denotes derivatives in (( χ)) and ((∂t χ)) which are conﬁned to the function χ(r. In the classical case such question is mute since the gauge transformations do not alter the ﬁelds and. t) inside the double brackets. t) are necessary to compensate terms which arise through the phase factor.11) to (8.. Applying the gauge transformations (8. e. t) (8.t)/ p + q(( χ)) Ψ(r.45) leads to the Schr¨dinger equation o 2 ˆ p − q A − q(( χ)) + qV − q((∂t χ)) Ψ(r. 8. however.11) of the potentials is equivalent to multiplying the wave function Ψ(r. t) r (8. 8. t) .50) The equivalence of (8.

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. can be described by various vector potentials.t) where σ is the vector of Pauli matrices. The resulting ﬁelds are called Yang-Mills ﬁelds. In this o case holds V (r. (8. x1 .53) ˆ ˆ p·A + A·p + q2 2 A + qV 2m (8. In case of a homogeneous potential pointing in the x3 -direction. therefore. (8.56) due to the gauge freedom. x2 .54) ˆ2 p q ˆ q2 2 − p·A + A + qV . j = 1. the generators of SU(2).55).g.58) where ∂j = (∂/∂xj ). H = 2m m 2m (8. Accordingly. 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.45) with Hamiltonian (8.. k2 . We want to describe the stationary states corresponding to the Hamiltonian (8. t). · A = 0 [cf. 3.44) 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. one can employ the Hamiltonian (8. (8. the so-called Landau gauge ˆ associates the vector potential AL (r) = Bo x1 e2 ˆ (8.59) . 2. t) = Bo . The stationary homogeneous magnetic ﬁeld B(r. consequently.210 Interaction of Radiation with Matter SU(2). The vector potential (8. For this purpose we use the wave function in the form Ψ(E. (8. aﬀects the complexity of the mathematical derivation of the wave functions. e. The Hamiltonian (8. k3 .56).12)]. 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. holds (8.55).57) with a homogeneous magnetic ﬁeld Bo . x3 ) = exp(ik2 x2 + ik3 x3 ) φE (x1 ) .52) This expression vanishes in the present case since since ˆ ˆ p · Af = A · pf and.55) 8. thereby.57) satisﬁes · A = 0 and. likewise can demand the existence of ﬁelds which compensate local transformations described by eiσ·χ(r.58). for Bo = Bo e3 in (8. The choice of a vector potential aﬀects the form of the wave functions describing the eigenstates and. t) ≡ 0.

and x2 -coordinates.65).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. From this observation one can immediately conclude that the wave shifted (by e 3 function of the system. We want to employ .56) through the vector potential A(r) = 1 Bo × r .62) x1o = and where ω = k2 eBo eBo me (8.64) is the classical Larmor frequency (c = 1).57).67) .63) (8.59). 2 2me (8. x2 . k2 . The energies corresponding to these states are E(n. The stationary Schr¨dinger equation (8.66) Obviously. However. the familiar harmonic oscillator quantum number.65) where we replaced the parameter E by the integer n. according to (8. like the corresponding gauge (8. the states are degenerate in the quantum number k2 describing displacement along the x2 coordinate. Without aﬀecting the energy one can form wave packets in terms of the solutions (8. 2 (8. k3 . Solution for Symmetric Gauge The solution obtained above has the advantage that the derivation is comparatively simple. k3 ) = ω(n + 2 k2 1 3 ) + . 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. the wave function (8. 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. the so-called symmetric gauge which expresses the homogeneous potential (8.59). according to (8. (8.61) leads to − where 2 2me 2 ∂1 + 2 k2 1 3 me ω 2 ( x1 + x1o )2 + 2 2me φE (x1 ) = E φE (x1 ) .65) which localize the electrons. Unfortunately. (8. is Ψ(n. is not symmetric in the x1 .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 ) .63) this induces a spread of the wave function in the x1 direction.8. therefore. x1 . k2 .62) is that of a displaced (by x1o ) harmonic oscillator with o 2 k 2 /2m ) energies.

(8. for any (8.74) We consider a simple case to relate (8.68) × u yields. (8. using × (uf ) = −u × f +f ˆ ˆ Bo · p × r f = Bo · r × p f . (8.71). an electron moving in the x. 8.74). In this case the current density measures −e v oriented tangentially to the ring. (8.73.76) e 2me class class · Bo . For the vector potential (8. namely.71) (8.212 Interaction of Radiation with Matter One can readily verify that this vector potential satisﬁes the condition (8.70) ˆ ˆ p · A f = − Bo · p × r f .75) 2 The latter can be related to the angular momentum µclass = − and. Vmag = −µclass · Bo (8.69) × u and × r = 0. ˆ (8.72) allows one to interpret µ = − e L 2me (8. in the present case of constant Bo .77) Comparision with (8.12) for the Coulomb gauge.72) and (8. Accordingly.72) to Hamiltonian (8.78) .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 . the Hamiltonian (8.74) is in the present case 1 µclass = − e r v e3 . Vmag = e 2me class = r me v e3 of the electron ˆ (8. namely. ˆ Identifying r × p with the angular momentum operator L. The theory of classical electromagnetism predicts an analogue energy contribution .73) where µclass is the magnetic moment connected with a current density j µclass = 1 2 r × j(r) dr (8. the magnetic moment (8. y-plane with constant velocity v on a ring of radius r. accordingly. The latter can be rewritten.67) one can write ˆ p·A = Using · (u × v) = − u · function f (r) ×v + v· 2i · Bo × r .

the Hamiltonian (8. For a magnetic ﬁeld (8.83) can then be expressed 1 ˆ 1 Ho = − ω 2 ∂2 ∂2 + 2 2 ∂X1 ∂X2 + 1 2 2 X1 + X2 2 + 1 i X1 ∂ ∂ − X2 ∂X2 ∂X1 . We will demonstrate in Sect. 2me (8.88) me ω xj . x3 ) = E ΨE (x1 .4: Electron in Homogenous Magnetic Field 213 as the quantum mechanical magnetic moment operator for the electron (charge −e). (8. 10 that the spin of the electron. (8.72) with an associated magnetic moment − g 2me S where g ≈ 2. x2 from x3 setting ΨE (x1 .80) (8.67) yields a more symmetric solution which decays to zero along both the ±x1 . 2 . e gives rise to an energy contribution (8. Accordingly. i. x2 ) where ˆ Ho = − 2 2 2 ∂1 + ∂2 (8. In this case. j = 1.and the ±x2 -direction.84) . likewise. x2 ) = E ψ(x1 .81) (8. To obtain these eigenstates we introduce the customary dimensionless variables of the harmonic oscillator Xj = (8. x3 ) . requires a Lorentz-invariant quantum mechanical description as provided in Sect. we separate the variable x1 . x3 ) = exp(ik3 x3 ) ψ(x1 .71) is ˆ H = 2 ˆ2 p e2 Bo + 2me 8me x2 + x2 + 1 2 eBo L3 . The functions ψ(x1 . the solutions of ˆ H ΨE (x1 . 10.8.83) (8.e.85) The Hamiltonian (8. described by the operator S. x2 .82) 1 ω (x1 ∂2 − x2 ∂1 ) i 2me + 1 me ω 2 x2 + x2 1 2 2 2 k2 3 + (8. the so-called gyromagnetic ratio of the electron. A derivation of his property and the value of g.79) To obtain the stationary states.86) 1 2 2me 1 2 ˆ as well as of the angular momentum operator L3 . 2me We have used here the expression for the angular momentum operator ˆ L3 = ( /i)(x1 ∂2 − x2 ∂1 ) . x2 ) .83) describes two identical oscillators along the x1 -and x2 -directions which ˆ are coupled through the angular momentum operator L3 . x2 . ˆ i.e.. we seek stationary states which are simultaneous eigenstates of the Hamiltonian of the two-dimensional isotropic harmonic oscillator 2 1 2 ˆ Hosc = − ∂ 2 + ∂2 + me ω 2 x2 + x2 (8. x2 ) obey then ˆ Ho ψ(x1 .56) pointing in the x3 -direction the symmetric gauge (8. x2 . for Bo = Bo e3 .87) . E = E − (8.

m. x1 . . aj = √ 2 Interaction of Radiation with Matter Xj + ∂ ∂Xj . (8.e. j = 1. (8. (8. it holds a† a1 + a† a2 + 1 1 1 2 Ψ (j. 1 2 since one phonon is annihilated and one created. obeyed.92) are not eigenstates of L3 .94) is an eigenstate of Ho ˆ Ho Ψ (j. . i.90) a† a2 − a† a1 1 2 . m . m . . . m .e. however.92) where Ψ(0.95) (j) ˆ We want to choose the coeﬃcients αmm such that (8. i. x2 ) = m=−j αmm Ψ(j.94) Since this state is a linear combination of states which all have vibrational energy (2j + 1) ω. 0. (8. 2 (8. m . x2 ) is the wave function for the state with zero vibrational quanta for the x1 . x1 . Such eigenstates can be expressed. x1 . 2 2 . (8.92) represents a state with j + m quanta in the x1 -oscillator and j − m quanta in the x2 -oscillator. −j + 1. x2 ) . .as well as for the x2 -oscillator. In order to cover all posible vibrational quantum numbers one needs to choose j. in fact. x1 . (8. If this property is. (j) (8. the total vibrational energy being ω(2j + 1).89) . . therefore. m . We. through a combination of states j Ψ (j. x1 . +j .94) is an eigenstate of the vibrational Hamiltonian.97) . (8. x1 . x1 .. x1 . 0. (8. x2 ) . x2 ) = 2m Ψ (j.96) ˆ holds. m = −j. x2 ) (8. such that 1 i a† a2 − a† a1 1 2 Ψ (j. x2 ) .91) We note that the operator a† a2 − a† a1 leaves the total number of vibrational quanta invariant. x1 .214 Employing the creation and annihilation operators 1 a† = √ j 2 and the identity ˆ ω L3 = which can readily be proven. m . m . x1 . . x2 ) (8. 1 3 .94) is also an eigenstate of L3 . . x2 ) = a† 1 j−m (j + m)! (j − m)! Ψ(0. x2 ) = ω 2j + 2m + 1 Ψ (j.. x1 . attempt to express eigenstates in terms of vibrational wave functions a† 1 j+m Ψ(j. 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 .93) ˆ The states (8. 1. x2 ) = ( 2j + 1 ) Ψ (j. m as follows: j = 0. m.

9. m) in Sect. The ﬁrst rotation contributes a factor exp(−im π ).5. except for a factor 2 . t) introduced above.55). the states are eigenstates of the operator [we use for the operator the notation of Sect. π . In the simplest case the radiation ﬁeld consists of a single planar electromagnetic wave described through the potential (8. . thus. 5. x1 . i.10.e. According to the derivation given there.98) then the states Ψ(j. 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.100) J2 = 1 2 2i 1 ˆ i.101) 2 2 where Dmm ( π . x1 .. x2 ) .5 Time-Dependent Perturbation Theory We want to consider now a quantum system involving a charged particle in a bound state perturbed by an external radiation ﬁeld described through the Hamiltonian (8. m.102) We have identiﬁed.84) above. m . the eigenstates of (8.11.309) yields ﬁnally Ψ (j. 5.(5.5. 5. m. a† . a2 1 2 present notation ←→ b† . the second rotation a factor dmm ( π ).31). This implies that ˆ we can obtain eigenstates of L3 by rotation of the states Ψ(j. 0) is a rotation matrix which describes the rotation around the x3 –axis by π 2 2 2 and then around the new x2 –axis by π . m. 5. x1 . Eq. t) = 1/4πr conﬁning an electron with energy E < 0 to move in the well known orbitals of the hydrogen atom.10. b+ .9. x1 . If we identify 2 a† .288)] 1 ˆ J3 = 2 a† a1 − a† a2 1 2 (8. Using the explicit form of the Wigner rotation matrix as given in (5. a transformation moving the x3 –axis into the x2 – 2 axis. The external radiation ﬁeld is accounted for by the vector potential A(r. 5. m. 5.11 such transformation is provided through π π (j) Ψ (j.92) correspond to the eigenstates |Ψ(j. (8.5.11 (8. Other radiation ﬁelds can . b− + − notation in Sects. cf.5. x2 ) (8. x2 ) = Dmm ( .55) conﬁnes the particle to stationary bound states.97). The required rotation must transform the x3 –axis into the x2 –axis. is in the notation of the present section 1 ˆ a† a2 − a† a1 .10.99) with eigenvalue m. 0) Ψ(j. x2 ).9. an example is the Coulomb potential V (r. the 2 2 latter representing the Wigner rotation matrix of Sect.. b† . m . x2 ) deﬁned in (8.8.10.e.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. According to Sect. is identical to the operator L3 introduced in (8. x1 . 5. which corresponds there to the angular momentum in the x2 –direction. 8. The connection with the present problem arises due to the fact that the operator J2 in Sect.83) and conﬁrmed the eigenvalues stated in (8.11. (8. a1 . x1 . We assume that the scalar potential V in (8.

one can estimate that the perturbation.41) e ˆ e p2 p·A ∼ 2me me me 2me 4 1 2 8πNω .e. is much smaller than the eigenvalue diﬀerences near ˆ typical atomic bound states. In fact. this assumption can be waved as our results below will not depend on it. t) is given by (8. we discount the possibility of a continuum of eigenstates.216 Interaction of Radiation with Matter be expanded through Fourier analysis in terms of such plane waves..108) We assume for the sake of simplicity that the eigenstates of Ho can be labeled through integers. ωV (8. is much 2 (r.107) 1 = 1 n=0 |n n| . using (8. as given in (8.105) in further calculations and to expand the wave function in terms of powers of VS (t) in a perturbation calculation. the term ∼ A second term in (8.105) as follows4 . t) − m 2m (8.104) (8. We ﬁrst note. in this case. we assume n|m = δnm and for the identity ∞ (8.e. t). . t) + A (r. This result will allow us to neglect the larger than the second term. can be considered a weak perturbation. orthonormal basis. However.103) (8. . the term ∼ p · A(r.e. (8.... Estimate of the Magnitude of VS We want to demonstrate now that the interaction VS (t).105). and that the ﬁrst term in (8. 1. i. 2 .42). i. .105) for the case of radiationinduced transitions in atomic systems. 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. The states |n are thought to form a complete. i. Here the so-called unperturbed system is governed by the Hamiltonian Ho with stationary states deﬁned through the eigenvalue problem Ho |n = n |n . i. t). 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.109) The reader should note that the estimates are very crude since we are establishing an order of magnitude estimate only. n = 0.e.106) where we adopted the Dirac notation for the states of the quantum system.105) where A(r. For an electron charge q = −e and an electron mass m = me one can provide the estimate for the ﬁrst term of (8. (8.

Hence.5 ˚.115) (8. Using again (8. one can neglect this term as long as the ﬁrst term gives non-vanishing contributions. λ = 2πc/λ. and me c2 ≈ A A 500 keV 2 ao ω ≈ 10−8 (8. with ao ≈ 0. e ˆ p · A ∼ 10 eV · 10−4 = 10−3 eV .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. and as long as the photon densities Nω /V are small. 2me (8. me (8.e.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..118) .8. (8.105) due to a radiation ﬁeld by VS = − q ˆ p · A(r. t) .105) for radiation ﬁelds can be considered a small perturbation. hence.110) where ao is the Bohr radius. Consequently. one obtains for (8.ˆ u 4πNk α(k.111) For ω = 3 eV and a corresponding λ = 4000 ˚ one obtains.117) In case that such perturbation acts on an electron and is due to superpositions of planar waves described through the vector potential (8. i.105).e. u) p · u exp i(k · r − ωt) ˆ ˆ ˆ kV . We want to estimate now the second term in (8. Nω = 1.113) 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.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 .116) This term is obviously much smaller than the ﬁrst term. We can. replace the perturbation (8. altogether. m (8..114) Employing for the second factor the estimate as stated in (8. a volume V = λ3 where λ is the wave length corresponding to a plane wave with frequency ω. the ﬁrst term in (8. i. Assuming a single photon. (8.42) it holds VS ≈ e m k.112) yields e2 A2 ∼ 10 eV · 10−8 = 10−7 eV .112) π λ me c2 and with e2 /ao ≈ 27 eV.

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 . .124) (8. . |Ψ(to ) = |0 (8. The sum runs over all possible k vectors and might actually be an integral. Using i i ∂t exp − Ho (t − to ) and (8.127) i ∂t |ΨD (t) = VD (t) |ΨD (t) . The states |0 . n = 0 at some later time t ≥ to . (8.118) is the form of the perturbation which. Multiplying the latter equation by the operator exp i (8. n = 1. describes the eﬀect of a radiation ﬁeld on an electron system and which will be assumed below to describe radiative transitions.123) i = Ho exp − Ho (t − to ) (8. 2. This state obeys the Schr¨dinger equation o i ∂t |ΨS (t) = [ Ho + VS (t) ] |ΨS (t) subject to the initial condition |ΨS (to ) = |0 . u) has been added to describe eliptically or circularly polarized ˆ waves. The probability to ﬁnd the particle in the state |n at time t is then p0→n (t) = | n|ΨS (t) |2 .108) as the eigenstates of the unperturbed Hamiltonian Ho .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) .126) Ho (t − to ) yields ﬁnally . under ordinary circumstances.125) (8. Equation (8. .120) (8. For this purpose one needs to determine the state |ΨS (t) of the particle.106–8. A factor α(k. |n are deﬁned in (8.. One seeks to predict the probability to observe the particle in one of the states |n . the sum over u involves the two possible polarizations of planar ˆ electromagnetic waves.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) (8.42) through k = |k| = ω.122) where |ΨD (to ) = |0 .218 Interaction of Radiation with Matter where we have replaced ω in (8.121) (8.

Due to the Hermitean property of the Hamiltonian Ho holds n|Ho = i n|exp − Ho (t − to ) from which we conclude. Employing this result one obtains for (8. . Accord(n) ingly. .129) n| and.127) we assume the expansion ∞ |ΨD (t) = n=0 (n) |ΨD (t) (n) (8. = . we deﬁne |ΨD (t) through the evolution equations i ∂t |ΨD (t) i ∂t |ΨD (t) i (2) ∂t |ΨD (t) (1) (0) = = = .137) follows |ΨD (t) = |0 .131) p0→n (t) = | n |ΨD (t) |2 .136) together with the initial conditions |ψD (to ) = |0 0 for n = 0 for n = 1. 8. to (8.134) (8.127. 8.133–8.128).5: Time-Dependent Perturbation Theory where VD (t) = exp 219 i i Ho (t − to ) VS (t) exp − Ho (t − to ) . From (8. 2.137) One can readily verify that (8.139) .8. n| (8. (8.121) expressed in terms of ΨD (t) is i p0→n (t) = | n|exp − Ho (t − to ) |ΨD (t) |2 . .137) are consistent with (8.137) can be solved recursively. using |exp[− i n (t n (8. Equations (8. In order to determine |ΨD (t) described through (8. consequently.132–8.130) = exp − i n (t − to ) − to )]| = 1.138) 1 i t dt VD (t ) |0 . We will consider here only the two leading contributions to |ΨD (t) .134.133.135) i ∂t |ΨD (t) (n) VD (t) |ΨD (n−1) (t) (8.137) |ΨD (t) = (1) (0) (8.133) (8. (8. 0 VD (t) |ΨD (t) (1) VD (t) |ΨD (t) (0) (8. 8.132) where |ΨD (t) accounts for the contribution due to n-fold products of VD (t) to |ΨD (t) .128) We note that the transition probability (8. . (8. . . .

note that uj . 8. 8..141) + .. This factor will serve mainly the purpose of keeping in the following derivation all mathematical quantities properly behaved.137) |ΨD (t) = (2) Interaction of Radiation with Matter 1 i 2 t t dt to to dt VD (t ) VD (t ) |0 . A2 are deﬁned through (8. in turn.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 .6 Perturbations due to Electromagnetic Radiation We had identiﬁed in Eq.144) r is the position of the electron and p = ( /i) is the momentum operator of the electron. 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. according to (8. . λ → 0+ . The factor III in (8. that two planar waves simulataneously interact with an electronic system. The coeﬃcients A1 . For the sake of including the eﬀect of superpositions of plane waves we will assume.e. The resulting perturbation on an electron system.220 This result. λ → 0+ has been introduced which describes that at time to → −∞ the perturbation is turned on gradually. + m=0 dt to to dt n|VD (t )|m m|VD (t ) |0 8.142) combining an incoming and an incoming or outgoing wave. (8. We want to apply now the perturbation expansion derived to such perturbations. deﬁnes the direction of the E-ﬁeld of the ˆ radiation.41). the direction of ˆ the propagation being determined by k = k/|k|. 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. In (8. yields for (8. A factor exp(λt). to → −∞ (8. is VS = ˆ ˆ V1 exp(−iω1 t) + V2 exp( iω2 t) eλt .118) above that the eﬀect of a radiation ﬁeld on an electronic system is accounted for by perturbations with a so-called harmonic time dependence ∼ exp(−iωt).140) Altogether we have provided the formal expansion for the transition amplitude n|ΨD (t) = n|0 + ∞ 1 i 1 i t dt n|VD (t ) |0 to 2 t t (8. non-singular. such that the combined radiation ﬁeld is decribed by the vector potential A(r.118)..34. however. (8.135.144) describes the propagation of the planar wave.142). i. according to (8.144) Here the factor I describes the strength of the radiation ﬁeld (for the speciﬁed planar wave) as determined through the photon density Nj /V and the factor II describes the polarization of the planar wave. II III (8.

(8. (8.145) we obtain then.128) one obtains k|VD (t) | = k|VS (t)| exp i ( k − ) (8. (8. (8. (8.151) ˆ ˆ n|V1 |m m|V1 |0 exp ˆ ˆ + n|V2 |m m|V2 |0 exp i i n − − m − ω1 − i λ)t exp ω2 − i λ)t exp m o n m m o .141).6: Perturbations due to Electromagnetic Radiation 1st Order Processes 221 We employ now the perturbation (8.143) to the expansion (8.140.149) Using the deﬁnition of VD stated in (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 . For the 1st order contribution to the transition amplitude n|ΨD (t) = (1) 1 i t dt n|VD (t ) |0 to (8. using (8. yields n|ΨD (t) = − (2) 1 ∞ t t 2 λ→0+ t→−∞ lim lim ( ( dt m=0 to to dt i i ( ( − − − ω1 − i λ)t ω2 − i λ)t (8.143).146) = 1 λ→0+ t→−∞ i lim lim × t dt exp to i ( n − o − i λ) t iω2 t × .8.130) and (for m = 0) i i exp − Ho (t − to ) |m = exp − for (8. employing 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 .145) n|ΨD (t) (1) m (t − to ) |m . 8. According to (8.150) and.148) We consider now the 2nd order contribution to the transition amplitude.128).

e.153) (8. i.131). For this rate holds . Using (8..222 ˆ ˆ + n|V1 |m m|V2 |0 exp ˆ ˆ + n|V2 |m m|V1 |0 exp i i ( ( − − Interaction of Radiation with Matter − ω1 − i λ)t exp ω2 − i λ)t exp i i ( ( − − ω2 − i λ)t − ω1 − i λ)t n m m o n m m o Carrying out the integrations and the limit limt→−∞ provides the result (2) n|ΨD (t) = − 1 ∞ 2 λ→0+ lim (8. 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. We also assume for the ﬁnal state n = 0 such that n|0 = 0 and p0→n (t) = | n|ΨD (t) |2 holds.148) and ∗ |z1 + z2 |2 = |z1 |2 + |z2 |2 + 2Re(z1 z2 ) (1) (8.155) λ→0+ ( o We are actually interested in the transition rate.152). the time derivative of p0→n (t). respectively. (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 (8.154) 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 ) (8.148) and (8. provide now the transition probability p0→n (t) according to Eq.

The δ-functions appearing in this expression reﬂect energy conservation: the incoming plane wave contribution of (8. the two terms apearing on the rhs. In case the matrix elements are non-zero. due to the uncertainty relationship would introduce a considerable perturbation to the system. one should average the rate over many periods of the radiation. For the resulting expression the limit limλ→0+ can be taken.159) leads to ﬁnal states |n with energy n = o + ω1 . The result supports our deﬁnition of incoming and outgoing waves in (8.158) ˆ + | n|V2 |0 |2 δ( − ω2 ) Obviously.142) ˆ ˆ The matrix elements n|V1 |0 and n|V2 |0 in (8.144). the matrix elements can vary strongly for diﬀerent states |n of the system.158) play an essential role for the transition rates of radiative transitions. Using lim x2 + 2 δ→0+ = π δ(x) (8. a property.8. . The second contribution to (8.31) and (8.144).157) one can conclude for the average transition rate k = d p0→n (t) dt t = 2π ˆ | n|V1 |0 |2 δ( n − n o − ω1 ) o (8. however.143..156) such that the 1st order contributions of the two planar waves of the perturbation simply add. of this expression describe the individual eﬀects of the two planar wave contributions of the perturbation (8. ˆ (8.e. to cancel the third term in (8. is much shorter than could be resolved in any observation. due to the vector potential A1 u1 exp i (k1 · r − ω1 t) . 8. First. describing either an incoming or an outgoing plane wave due to the vector potential A2 u2 exp i (k1 · r ˆ ω2 t) . which is observed through the so-called spectral intensities of transitions |0 → |n .160) leads to ﬁnal states |n with energy n = o ± ω2 . any attempt to do so. The result of such average is. The time average will be denoted by · · · t . in fact.142–8.158). i. these matrix elements determine the so-called selection rules for the transition: the matrix elements vanish for many states |n and |0 on the ground of symmetry and geometrical properties. (8.156) ( o The period of electromagnetic radiation absorbed by electronic systems in atoms is of the order 10−17 s. Hence.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.

. as we did in the the case of 1st order transitions. (2) (8.157) yields. hence. ∞ − 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.e.153–8. i.163) × (8. (8. e.165) × ∞ z3 = m=0 2 ω2 − 2i λ)t 2 ω2 − 2i λ n− o ˆ ˆ ˆ ˆ n|V2 |m m|V1 |0 n|V2 |m m|V1 |0 + ω2 − i λ m − o − ω1 − i λ m− o ( n o i − × It holds exp − ω1 ω2 − 2i λ)t ω2 − 2i λ n − o − ω1 ( n o 3 − |z1 + z2 + z3 | = |z1 | + |z2 | + |z3 | + j.166) In this expression the terms |zj |2 exhibit only a time dependence through a factor e2λt whereas the terms zj z ∗ k for j = k have also time-dependent phase factors.153) vanishes such that the leading contribution to the transition probability p0→n (t) arises from the 2nd order amplitude (8.161) To determine the transition rate we proceed again.e.167) Taking now the limit limλ→0+ and using (8.158).224 2nd Order Radiative Transitions Interaction of Radiation with Matter We now consider situations where the ﬁrst order transition amplitude in (8.g. |z1 + z2 + z3 |2 t = |z1 |2 + |z2 |2 + |z3 |2 (8. in Eqs. exp[ i (±ω2 − ω1 )].162) × and.k=1 j=k 2 2 2 2 ∗ zj zk (8.166) over many periods of the radiation yields exp[ i (±ω2 − ω1 )] t = 0 and.158).152). i. it holds p0→n (t) = | n|ΨD (t) |2 . We deﬁne ∞ z1 = m=0 ˆ ˆ n|V1 |m m|V1 |0 m − o − ω1 − i λ exp i × (8. Time average · · · t of expression (8. similarly. in analogy to (8.164) (8. k = = d p0→n (t) dt 2π ∞ m=0 t 2 ˆ ˆ n|V1 |m m|V1 |0 m − o − ω1 − i λ δ( m − o − 2 ω1 ) (8.168) absorption of 2 photons ω1 ...

. accordingly. ﬁrst absorption of photon ω1 and then absorption/ emission of photon ω2 . i. describe the absorption of two photons. a second perturbation.e. the second term describes the processes leading to ﬁnal states |n with energy n = o ± 2 ω2 and. 8. The ﬁrst term. interference between the contributions from all intermediate states |m can arise. |n ← |m .168) describe similar histories of the excitation process. this state is only virtually excited. namely absorption/ emission ﬁrst of photon ω2 and then absorption of photon ω1 and.. each of energy ω2 .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.158) which accounts for such transitions.e. through absorption of a photon. i. |m ← |0 (8. each of energy ω1 .169) pert. it must hold n = o + 2 ω1 . Similarly. (8. describe the absorption/emission of two photons. which is stationary and energy is conserved. there is no energy conservation necessary (in general.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 factors | · · · |2 in (8. the third term describes processes in which a photon of energy ω1 is absorbed and a second photon of energy ω2 is absorbed/emitted. The expression sums over all possible virtually occupied states |m and takes the absolute value of this sum. The remaining two contributions in (8. the system is perturbed through absorption of a photon with energy ω1 from the initial state |0 into a state |m .168) describe the time sequence of the two photon absorption/ emission processes. m = o + ω1 ) and the evolution of state |m is described by a factor 1/( m − o − ω1 −i λ). i.e. according to its δ-function factor. Most remarkably.168) describes two intermediate histories. accordingly. In case of the ﬁrst term in (8.8.e. For this purpose we will employ the transition rate as given in Eq. i.. Similarly. vice versa. promotes the system then to the state |n . describes processes which lead to ﬁnal states |n with energy n = o + 2 ω1 and. the third term in (8.. .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.

plane polarized wave described through the complex vector potential A(r. θ. in case of a radiative transition from the 1s state of hydrogen to one of its three 2p states. 8. φ) = 6 r −r/2ao e Y1m (θ. =0. φ) = 2 − 1 2 1 −r/ao e Y00 (θ. .179) . m denote the relevant quantum numbers) ψn=1. (8. t) = 2πN ωV u exp ˆ ı (k · r − ωt) + 2πN ωV u exp ˆ ı (−k · r + ωt) . However.226 Absorption of a Plane Polarized Wave Interaction of Radiation with Matter We consider ﬁrst the case of absorption of a monochromatic. the second term can be discounted in case of absorption. . (8. the vector potential actually assumed is A(r.176) ψn=2. φ) a3 ao o 2p (8. (8. is then kabs = Dipole Approximation 2π e2 2πN m2 ωV e ˆ u · n| p eik·r |0 ˆ 2 ˆ ˆ V1 exp(−iωt) + V2 exp(+iωt) eλt ..170) We will employ only the real part of this potential.172) δ( n − o − ω) (8. the wave functions are (n.143) will contribute to the absorption process. θ. to → −∞ (8. The absorption rate. ˆ (8.174) We seek to evaluate the matrix element ˆ M = n| p eik·r |0 . λ → 0+ .e. For example. i.2 = m ωV Only the ﬁrst term of (8. (8.144) VS = where e 2πN ˆ ˆ p · u e±ik·r .175) The matrix element involves a spatial integral over the electronic wave functions associated with states |n and |0 . φ) (8.158). φ) a3 o 1s (8.m=0 (r.177) √ M = 6 0 ∞ 1 2π ia4 o r2 dr −1 dcosθ 0 ∗ dφ r e−r/2ao Y1m (θ.173) V1.171) The perturbation on an atomic electron system is then according to (8.143. and the integral is =1.m (r. t) = 8πN 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 . . in this range one can expand eik·r ≈ 1 + ik · r + . φ) × × eik·r e−r/ao Y00 (θ. according to (8.

] + [ r.181) For the commutator in (8.182).182) one ﬁnds [ r. is Ho = ˆ (p)2 e2 + V (r) . One refers to this approximation as the dipole approximation.175) can then be ˆ achieved and the diﬀerential operator p = i replaced by by the simpler multiplicative operator r.181) and (8.185) from which follows (8. Ho ] |0 = i i n) n|r |0 .7: One-Photon Emission and Absorption 227 One can estimate that the absolute magnitude of the second term in (8.182) one obtains M ≈ Insertion into (8.179) and other terms are never larger than 20π ao /λ.178) holds eik·r ≈ A 1 1 + O( 50 ) such that one can approximate eik·r ≈ 1 .104) and. Ho ] i where Ho is the Hamiltonian given by (8. V (r) ] 2me =0 = Using r = 3 ˆ j=1 xj ej 1 2me 3 pk [ r.184) and the commutation property [xk .k=1 i m 3 pk ek = ˆ j. Ho ] = ˆ2 p [ r.175).186) δ( n − o − ω) (8.180) ∆E2p−1s and ao = 0. V (r) = − . pk ] + ˆ ˆ k=1 1 2me 3 [ r. 2me r (8. Using |k| = 2π/λ. This simpliﬁcation results from the identity m ˆ p = [ r.183) (8. pj ] = i δkj one obtains ˆ [ r.8. (8. the value of the wave length for the 1s → 2p transition 2π c λ = = 1216 ˚ A (8.174) yields kabs = 4π 2 e2 N ω V u · n| ˆ |0 ˆ r 2 m( o − m n| [r.187) . Transition Dipole Moment A further simpliﬁcation of the matrix element (8. Using (8. in case of the hydrogen atom.k=1 i ˆ p m (8. pk ] pk ˆ k=1 (8.529 ˚ one concludes that in the signiﬁcant integration range in (8.182) (8. We are now in a position to obtain an alternative expression for the matrix element (8. Ho ] = i m 3 pk ej δjk = ˆ j.

three-dimensional vector. ϕ.158) above that in ﬁrst order the contributions of all components of the radiation ﬁeld add. ϕ2 = π/2. according to (8. 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. and all polarizations of the radiation. The unit ˆ ˆ ˆ ˆ can be chosen to point along the x1 .189) where the factor 1/ arose from the integral over the δ-function. 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. such that the thermal radiation present supplies a continuum of frequencies.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. We have demonstrated in our derivation of the rate of one-photon processes (8. The direction of this vector in the u1 . The δ-function appearing in this expression.193) . hence. ρ) and ˆ u β = ∠(ˆ2 .191) (8.33) are orthogonal to each other. there are ony two linearly independent directions of u possible.e. the direction of u1 is described by the ˆ ˆ ˆ ˆ angles ϑ1 = π/2. say u1 and u2 . x3 -axes of a right-handed cartesian vectors u1 . u2 . We can.35).190) (8. i. ˆ u ˆ (8. x2 . in practical situations. obtain the transition rate in the present case by adding the individual transition rates of all planar waves present in the oven. Integrating and summing accordingly over all contributions as given by (8. ϕ1 = 0 and of u2 by ϑ2 = π/2. u2 and k ˆ ˆ coordinate system. |0 .228 Interaction of Radiation with Matter where we used the fact that due to the δ-function factor in (8. ˆ In order to carry out the integral dk we note that u describes the possible polarizations of the ˆ ˆ and u.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 (8.31–8. directions. and (2) the fact that strictly monochromatic radiation cannot be prepared such that any radiation source provides radiation with a frequency distribution. a very hot oven. Accordingly. For the two angles α = ∠(ˆ1 .174) one can replace n − o by ω. ρ) holds then u cosα = cosϑ1 cosϑ + sinϑ1 sinϑ cos(ϕ1 − ϕ) = sinϑcosϕ and cosβ = cosϑ2 cosϑ + sinϑ2 sinϑ cos(ϕ2 − ϕ) = sinϑsinϕ . one can express | u · n| r |0 |2 = |ρ|2 ( cos2 α + cos2 β ) = sin2 θ . will actually be replaced by a distribution function which reﬂects (1) the ﬁnite life time of the states |n . planar waves as deﬁned in (8. k frame is described by the angles ϑ. k ˆ As a result.. Absorption of Thermal Radiation We want to assume now that the hydrogen atom is placed in an evironment which is suﬃciently hot.188) ˆ Here dk is the integral over all orientations of k.

in agreement with observations. The most intensely absorbing molecules are long. In case of a classical radiation ﬁeld one would expect that emission cannot occur. The radiation ﬁeld is described. the absorption rate is of the order of 109 s−1 or 1/nanosecond.173). in particular. 2 . The corresponding process is termed spontaneous emission. linear molecules.37 × 1019 1 ao | n|r|0 |2 × s λ λ2 .172. However. The value of | n| r |0 | determines the strength of an optical transition. a problem in case of emission by quantum systems in complete darkness. However.. Nω = 0. 1. can be inserted into (8. i.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 |. However.7: One-Photon Emission and Absorption 229 This geometrical average.. (8. . (8. ˆ In case of emission only the second term V2 exp(+iωt) in (8. . This dependence predicts. i. Transition Dipole Moment The expression (8.194) combined are typically somewhat smaller than (1 ˚/1000 ˚)3 = A A 10−9 .e. This diﬀerence between emission and absorption is due to the quantum nature of the radiation ﬁeld. however. that emission occurs even if no photon is present in the environment.197) o (tot) (8. Emission of Radiation We now consider the rate of emission of a photon. For this purpose we use ω/c = 2π/λ and obtain 4 e2 ω 3 32π 3 e2 ao 1 ao = = 1.37 × 1019 × 3 3 c3 3 ao λ3 s λ and kabs where λ = 2πc n − (8.. The transition dipole moment can vanish for many transitions between stationary states of a quantum system. Otherwise. This poses. The quantum nature of radiation manifests itself in that the number of photons Nω msut be an integer. for atoms or symmetric molecules.8.189) to yield the total absorption rate kabs (tot) = Nω 4 e2 ω 3 | n|r|0 |2 3 c3 . Accordingly.173) contributes.e.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 . ﬁnally. Nω photons before absorption.195) = Nω 1. the resulting total rate of emission bears a diﬀerent dependence on the number of photons present in the environment. as for the absorption process. 8.196) The last two factors in (8. the calculation of the emission rate proceeds as in the case of absorption. there is also a contribution to the emission rate which is proportional to Nω . for Nω = 0. by planar waves with vector potential (8. .171) and perturbation (8. For practical evaluations we provide an expression which eliminates the physical constants and allows one to determine numerical values readily. 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.

Deﬁning the rate of spontaneous emission 4 e2 ω 3 | n|r|0 |2 (8.202) It follows. exp[ ω/kB T ] − 1 (8. e. 8. . exp[− ω/kB T ] = This equation yields Nω = 1 .. 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. For these numbers holds Nn / No = exp[−( n − o )/kB T ] (8.201) ksp = 3 c3 the rates of absorption and emission are Nω ksp and (Nω + 1)ksp . 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.198) to determine the stationary distribution of photons ω in an oven of temperature T . The number of atoms undergoing absorption in unit time are Nω ksp No and undergoing emission are (Nω +1)ksp Nn .195) and (8. Nω + 1 (8. accordingly.199) Planck’s Radiation Law The postulate of the Nω + 1 dependence of the rate of emission as given in (8..e. The total rate of emission. To demonstrate this property we apply the transition rates (8.200) where kB is the Boltzmann constant. it must hold Nω ksp N0 = (Nω + 1) ksp Nn (8. Hence. the well-known Planck radiation formula. In the following we discuss several examples.8 Two-Photon Processes In many important processes induced by interactions between radiation and matter two or more photons participate. We assume n − o = ω. Let No and Nn denote the number of atoms in state |0 and |n . respectively. using (8. in lasers.198) is consistent with Planck’s radiation law which reﬂects the (boson) quantum nature of the radiation ﬁeld.204) Nω . Under stationary conditions the number of hydrogen atoms undergoing an absorption process |0 → |n must be the same as the number of atoms undergoing an emission process |n → |0 .230 Interaction of Radiation with Matter which is termed induced emission since it can be induced through radiation provided.198) (8. respectively.200). = (8.g.203) i.

We assume that a planar wave with wave vector k1 and polarization u1 .207) does not converge rapidly. We wish to describe an observation in which a detector is placed at a solid angle element dΩ2 = sinθ2 dθ2 dφ2 with respect to the origin of the coordinate system in which the electron is described. t) = Ao1 u1 cos(k1 · r − ω1 t) .173).207) one is. induce single photon absorption processes as described by the transition rate Eq. The resulting rate of the transition depends on 2 Nω . . (8. Employing the dipole approximation (8.207) Expression (8.207) does not provide a suitable avenue of computing the rates of 2-photon transitions.168) where the ﬁrst term describes the relevant contribution. faced with the dilemma of having to sum over all intermediate states |m of the system. dropping the index 1 characterizing the radiation.187). as described through the vector potential ˆ A(r. but then requires the absorption of two photons. k = 2π e2 2πNω m2 ωV e × δ( m 2 ∞ m=0 n|ˆ · p eik·r |m m|ˆ · p eik·r |0 u ˆ u ˆ m − o − ω1 − i λ 2 × (8.208) has been prepared.206) − o − 2 ω) .205) The respective radiative transition is of 2nd order as described by the transition rate (8. 8. The transition requires that the transition dipole moment n| r |0 does not vanish for two states |0 and |n . (8. even if n| r |0 vanishes. If the sum in (8.106).181) and using (8. In applying (8. ˆ (8. The electron absorbs the radiation and emits immediately a second photon. The intense radiation ﬁelds of lasers allow one to increase transition rates to levels which can readily be observed in the laboratory.182) yields.8. which is not necessarily the case. however. 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) . under ordinary circumstances.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. (8. Scattering of Photons at Electrons – Kramers-Heisenberg Cross Section We consider in the following the scattering of a photon at an electron governed by the Hamiltonian Ho as given in (8.104) with stationary states |n deﬁned through (8.8: Two-Photon Processes Two-Photon Absorption 231 The interaction of electrons with radiation. then expression (8. ﬁnally. a transition between the states |0 and |n may be possible. however. in case ˆ of absorption only V1 contributes.172. In this case one needs to choose the energy of the photons to obey n = o + 2 ω. The perturbation which accounts for the coupling of the electronic system and the radiation ﬁeld is the same as in case of 1st order absorption processes and given by (8. One obtains. However.

208) and the emission of the radiation in (8.105).210).210).212) where N1 /V is the density of photons for the wave described by (8. t) = A+ u1 exp[ i (k1 · r − ω1 t) ] + A− u2 exp[ i (k2 · r − ω2 t) ] .211) is stated in Eq. however. i. for the components of the wave (8.213) where N2 /V is the density of photons characterized through k2 . hence. We know already from our description in Section 8.208) and emit radiation corresponding to the vector potential (8. 8.141). o1 ˆ o2 ˆ (8.210) A(r. In the present case we consider only scattering processes which absorb radiation corresponding to the vector potential (8.211) are given by VS (t) = e ˆ p · A+ u1 exp[i(k1 · r − ω1 t)] + A− u2 exp[−i(k2 · r − ω2 t)] o1 ˆ o2 ˆ 2me contributes in 2nd order + e2 4me A+ A− u1 · u2 exp{i[(k1 − k2 ) · r − (ω1 − ω2 ) t]} ˆ o1 o2 ˆ contributes in 1st order The eﬀect of the perturbation on the state of the electronic system is as stated in the perturbation expansion (8. (8. according to the ˆ description of emission processes on page 229.211) is o2 A− = o2 8π (N2 + 1) ω1 V (8.105) using the vector potantial (8.208. later we will allow the quantum system to select appropriate values. The vector potential describing the emitted plane wave is then A(r. u1 . ˆ (8. in the present case.210) The vector potential which describes both incoming wave and outgoing wave is a superposition of the potentials in (8. The second term in (8.211) The ﬁrst term describes the absorption of a photon and.214) .. the amplitude A+ is given by o1 A+ = o1 8πN1 ω1 V (8.208). 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 (8.208.210) is accounted for by the following contributions of (8. the wave characterized through k1 . u2 .211) accounts for the emitted wave and. ˆ The perturbation which arises due to the vector potential (8.209) cosθ2 where the value of k2 has been ﬁxed.6 above that the absorption of the radiation in (8. 8. t) = Ao2 u2 cos(k2 · r − ω2 t) .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 . This expansion yields. the amplititude A− deﬁned in (8. The relevant terms of the perturbation (8.e.

215) t dt ei( to n− o− ω1 + ω2 +i λ)t 1 i 2 e2 4m2 e A+ A− × o1 o2 × ˆ ˆ u1 · n| p |m u2 · m| p |0 × ˆ ˆ t t × to dt to dt ei( n− m− ω1 +i λ)t i( e m− o+ ω2 +i λ)t ˆ ˆ + u2 · n| p |m u1 · m| p |0 × ˆ ˆ t t × to dt to dt ei( n− m+ ω2 +i λ)t i( e m− o− ω1 +i λ)t We have adopted the dipole approximation (8. Following closely the procedures adopted in evaluating the rates of 1st order and 2nd order radiative transitions on page 222–225.218) . only these two terms contribute to the scattering amplitude. Hence. evaluating the time integrals in (8.e.8 · 10−15 m . Accordingly.215) correspond to scattering processes in which the radiation ﬁeld ‘looses’ a photon ω1 and ‘gains’ a photon ω2 .e. Only the second (1st order) and the third (2nd order) terms in (8.8. Inserting also the values (8.216) over all available modes of the ﬁeld. one needs to integrate the rate as given by (8. i.217) 2 Here ro denotes the classical electron radius ro = e2 = 2.8: 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. i. 8.181) in stating this result. one needs to carry out the integration 2 V(2π)−3 k2 dk2 · · ·.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 .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.. me c2 (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 (8.212..

2 . in fact.. consequently. 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 assumes the even stronger condition ω1 << | o − m| . for all states |m of the electronic system (8. .234 Interaction of Radiation with Matter The factor N1 c/V can be interpreted as the ﬂux of incoming photons.228) S (0) = m S (0) = 2 m .221) suggests to expand S( ω) S( ω) = S(0) + S (0) ω + Using 1 − o ± ω one can readily determine m (8. a process in which the electronic state remains unaltered after the scattering.224) = m 1 − o ( m ω ( ω)2 + + . Rayleigh Scattering We turn ﬁrst to an example of so-called elastic scattering. (8.227) 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. i.. ω1 = ω2 . . Accordingly. − o )2 ( m − o )3 (8.221) Using |n = |0 and.225) S(0) = m u ˆ 0|ˆ1 · p |m m|ˆ2 · p |0 u ˆ u ˆ 0|ˆ2 · p |m m|ˆ1 · p |0 u ˆ + me ( m − o ) me ( m − o ) (8.223) m Condition (8.226) 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.e.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 + .. one can relate (8.219) (8.217) to the scattering cross section deﬁned through dσ = It holds then dσ = 2 ro rate of photons arriving in the the solid angle element dΩ2 ﬂux of incoming photons ω2 (N2 + 1) dΩ2 u1 · u2 n|0 ˆ ˆ ω1 1 n|ˆ1 · p |m m|ˆ2 · p |0 u ˆ u ˆ me m m (8. it follows 2 dσ = ro (N2 + 1) dΩ2 |ˆ1 · u2 − S( ω) |2 u ˆ (8.220) n|ˆ2 · p |m m|ˆ1 · p |0 u ˆ u ˆ + m − o − ω1 2 − − o + ω2 In the following we want to consider various applications of this formula.

k=1 (ˆ1 )j (ˆ2 )k δjk = u1 · u2 u u ˆ ˆ (8. u1 ·r ] |0 = 0 u ˆ (8.. For this purpose we apply (8.222) of the cross section for Rayleigh scattering dσ = 2 ro m2 ω 4 (N2 + 1) dΩ2 × e (8. this term cancels the u1 · u2 term in (8. 8. (8. as a compass. Expression (8. pk ] = i δjk yields ﬁnally ˆ S(0) = 1 S(0) = i 3 3 (8.235) We can now combine eqs.182) for p and the expression (8.224.8: Two-Photon Processes 235 ˆ These three expressions can be simpliﬁed using the expression (8.234) where we used for the second identity the fact that u1 ·r and u2 ·r commute. honey bees. 8. u ˆ ˆ ˆ ˆˆ i The commutator property [xj .226) into S(0) = 1 i ( 0|ˆ1 ·r |m m|ˆ2 · p |0 − 0|ˆ2 · p |m m|ˆ1 ·r |0 ) u u ˆ u ˆ u m .108) this is 1 0|ˆ1 ·r u2 · p − u2 · p u1 ·r |0 .108) for the identity operator.227) vanishes.229) (8. m|ˆ1 · p |0 u ˆ 1 = − m|ˆ1 ·r |0 u me ( m − o ) i (8. Using again (8.108) yields S (0) = me 2 0| [ˆ2 ·r.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.236) is of great practical importance.234.222). the blue color of the sky and the polarization pattern in the sky which serves many animals.k=1 j.182) 0|ˆ1 · p |m u ˆ 1 = 0|ˆ1 ·r |m u me ( m − o ) i This transforms (8.222).226).231) (ˆ1 )j (ˆ2 )k 0|[xj . 8. i.228) provides then the ﬁrst non-vanishing contribution to the scattering cross section (8.230) According to (8. for example. pk ]|0 = u u ˆ j.229) both for the u1 · p and the to u2 · p terms in (8.233) m Employing again (8.235) and obtain the leading contribution to the expression (8. ˆ ˆ We want to prove now that expression (8.232.228) 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. It explains. ˆ We want to simplify ﬁrst (8.8.229) 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 ) .232) Obviously.e. u u u u (8. . For this purpose we replace p using (8. ˆ ˆ S (0) given in (8.

|n = |0 and ω1 = ω2 = ω in (8. To evaluate |ˆ1 · u2 |2 we assume that k1 is oriented along the x3 -axis and. however.240). (8.e.. u1 = 1 ˆ ˆ (8.240) cosθ2 We choose for the polarization of the incoming radiation the directions along the x1 . ao Bohr radius . The ﬁrst choice is ˆ sinφ2 k2 × k1 (1) u2 = ˆ = −cosφ2 (8. hence. assume now that the scattered radiation has very short wave length such that ω >> | o − m| .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 (8.220) 2 dσ = ro (N2 + 1) dΩ2 |ˆ1 · u2 |2 . i.and the x2 -axes 1 0 (1) (2) u1 = 0 .241) 0 0 Similarly.244) .220). 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 . (8. u2 = u2 ˆ ˆ ˆ ˆ for u1 = u1 . u2 = u2 ˆ ˆ ˆ ˆ cos2 φ2 cos2 θ2 sin2 φ2 for u1 = u1 . we choose for the polarization of the emitted radiation two perpendicular directions u2 ˆ (2) and u2 which are also orthogonal to the direction of k2 .237) The resulting scattering is called Thomson scattering.242) |k2 × k1 | 0 where the second identity follows readily from k1 = e3 and from (8.239) We will show below that this expression decribes the non-relativistic limit of Compton scattering. though. for all states |m of the electronic system .243) cos2 θ2 cos2 φ2 for u1 = u1 . u2 = u2 ˆ ˆ ˆ ˆ (8.. u2 = u2 ˆ ˆ ˆ ˆ (1) (2) (2) (1) (1) (2) (1) (2) (2) (1) needs to be (8. We want to assume.238) (8. that the dipole approximation is still valid which restricts the applicability of the following derivation to k1 >> One obtains immediately from (8. u ˆ 1 ao .

deﬁned in (8. but most often involves other degrees of freedom. 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. ˆ In case that one measures the scattered radiation irrespective of its polarization. If energy is absorbed by the system. In case that the nuclear degrees of freedom excited absorb very little energy. or in case of translational motion of molecules in liquids. In the case of such scattering an electron system absorbs and re-emits radiation without ending up in the initial state. one speaks of Stokes scattering.)n . Such scattering is called Raman scattering.)n . The radiation scattered at right angles is even completely polarized (1) along the u2 -direction.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.e. Raman Scattering and Brillouin Scattering We now consider ineleastic scattering described by the Kramers-Heisenberg formula. The energy deﬁcit is used to excite the system. one speaks of anti-Stokes scattering. Since the scattering is inelastic. as in the case of excitations of accustical modes of crystals. 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. the states |n etc. for c → ∞ the Compton scattering cross section (8. In the case that the scattering excites other than electronic degrees of freedom. i.249) dσ = ro (N2 + 1) ω1 −1 −1 ω2 − ω1 = .247) becomes identical with the Thomson scattering cross section (8. 8. i.g. φ(vibr. (8. the resulting scattering cross section is 2 ro (N2 + 1) ( 1 + cos2 θ2 ) dΩ2 .220) represent actually electronic as well as nuclear motions. if energy is released.. The excitation can be electronic. in case of a diatomic molecule |n = |φ(elect.220) vanishes and one obtains in case of Raman scattering ω2 2 dΩ2 | u2 · R · u1 |2 ˆ ˆ (8.246)... the scattered radiation is polarized to some degree.247) ( 1 − cosθ2 ) (8.8: Two-Photon Processes 237 In case that the incident radiation is not polarized the cross section needs to be averaged over (1) (2) the two polarization directions u1 and u1 . molecular vibrations or crystal vibrational modes. e.8.e.247.246) 2 This expression is the non-relativistic limit of the cross section of Compton scattering.248) me c2 One can readily show that in the non-relativistic limit. For electronic systems in molecules or crystals the degrees of freedom excited are nuclear motions. the scattering is termed Brillouin scattering.

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

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

. We can then state P2 = T (4. However. in notation (9.5) . i. not any state can be chosen as to represent the many– particle system.3) the j-th index in the second row. P2 = . . (9. .4) both permutations being obviously odd.. . 5) (9.240 Many–Particle Systems repeatedly like degenerate states. k) = (4. particles 4 and 5. We will not discuss here at any depth the properties of the permutation groups SN . .2). Denoting by P (j) the image of j.. rather only states which obey a certain transformation property under permutations are acceptable. xj . . 5). . 5) T (5. number k under the entry j of the ﬁrst row indicating that particle j is switched with particle k.e. in particular. The product P1 P2 of even and odd permutations P1 and P2 obey the following multiplication table: P2 \ P1 even odd even even odd odd odd even We ﬁnally determine the action of permutations P on the wave functions (9. the top row representing the numbers 1 to N labelling the N particles under consideration. Transpositions denoted by T (j. xP (N ) ) . . k) of the two particles transposed. is called a transposition. . xP (2) . and the second row showing the numbers 1 to N again.e. For the permutations P1 and P2 given above holds P1 = T (3. i. 3) T (7. i. xN ) = Ψ(xP (1) .3) P2 which aﬀects only two particles.e. even though these properties. . In the following sections we will then introduce the operators of second quantization which provide a means to ascertain that manipulation of wave functions always leave the transformation property under permutations uncompromised. For the following we will require only two properties of SN . leaving all other particles unchanged. (9. k). This property will be established now. . (j. . The elements P ∈ SN can be represented conveniently as 2 × N matrices. but in a diﬀerent order.e. The group SN is then composed of two disjunct classes of even and of odd permutations. The latter factorization has the essential property that the number of factors is either even or odd. . 6) T (6. the set of all permutations of N objects. The permutations we need to consider for a system of N particles are the elements of the group SN .. . A reader interested in the quantum mechanical description of N particle systems is strongly encouraged to study these representations. i. namely implicitly the group property. x2 . . . one can state P Ψ(x1 . P2 = T (4. . . xP (j) . the representations in the space of N –particle wave functions are extremely useful in dealing with N –particle systems. 5) in case of P2 . . and explicitly the fact that any P ∈ SN can be given as a product of transpositions T (j. a given P can either only be presented by even numbers of transpositions or by odd numbers of transpositions. k) are characterized by the two indices (j.. 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 = . . One calls permutations of the ﬁrst type even permutations and permutations of the second type odd permutations. 8) . 4) T (4.

e. {1} (trivial) and {1.e.. 1 together with multiplication. Obviously. xk ) j. In fact. usually a very large number. hence. In fact. all particles being governed by the same interactions. i. i. . From the latter results [H.e. it is a postulate of quantum mechanics that for any description of many–particle systems the permutation operators must be diagonal.8) |Ψ (9.1: 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. xk ) is symmetric in its two arguments.e. the Hamiltonian must be of the type N H = j=1 1 F (xj ) + 2 N G(xj . for the scalar product holds j|k = δj k . the Hamiltonian for permuted indices P HP −1 is identical to H. a diagonal representation can only berealized in a simpler group.11) . Presently.6) which describes one–particle interactions and two–particle interactions of the system.9. i. S where S denotes the number of available single particle states. In the latter case all even permutations are represented by ‘1’ and all odd permutations by ‘-1’. i. From this property follows that H is independent of any speciﬁc assignment of indices to the particles (note that (9. particles with integer spin values. 2. Since permutations do not necessarily commute. ) are the same for all particles and for all pairs of particles.9) = 1 for P even −1 for P odd (9. .7) This property. provide a representation of the group structure represented by the multiplication table of even and odd permutations given above. both groups. (9. The ﬁrst case applies to bosons. 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 . In fact. in turn.. i. ∀P ∈ SN .-1}. G(xj .k=1 (9. P ] = 0 . The terms of the Hamiltonian will be discussed further below. The single particle states are assumed to be orthonormal.5) implies that a permutation eﬀectively changes the indices of the particles) and. the second group to fermions. the representation is either isomorphic to the group composed of only the ‘1’. or the corresponding group of two elements 1. Fock-Space Our derivation will start by placing each particle in a set of S orthonormal single particle states x|k = φk (x).10) We will construct now wave functions which exhibit the appropriate properties. it is essential that the functions F ( ) and G( . (9. implies that the permutation operators can be chosen in a diagonal representation.. particles with half–integer spin values. k = 1. (9.e. . −1..

. . . . nS ). λ2 . λN )|ΨFock (λ1 . .15). . . . The reader is well advised to pause and grasp the deﬁnition of the nj . . . λ2 . . with λj = λk = . (9. x2 . . This detail is connected with the fact that a two–particle state of the type ΨFock (λ. is a particular choice of placing N particles consistent with the occupation numbers (n1 . each particle j into a speciﬁc state |λj . . |λ2 . The Fock states represent N particles which are placed into S states |λ1 . k) ∈ Λ(λ1 . . 9. . S} of the type (9.12) where the second factor represents the scalar product between spin states.. appear severalfold in (9. .15) is that the sum over permutations does not involve permutations among indices j. . 9. |λN . . . . λ2 . n2 . . λ2 . 2. . .15) completely and.15) (nj !) N! 1 2 P ∈Λ(λ1 . . .11. It holds N ΨFock (λ1 .. . λN ) = j=1 δλj λj . . We ﬁrst consider orthonormal many–particle wave functions.. Λ(λ1 . which do not yet obey the symmetries (9. .8. the so-called Fock-states. . . 9. are essential.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. . An important detail of the deﬁnition (9.8) is x1 .14) Obviously.9). k. . . . .λN ) Here the states (λ1 . .g. In a second step we form linear combinations of Fock states with the desired symmetries. . The numbers nj .12). . λN ) = j=1 φλj (xj ) . hence.λN ) P x1 . xN |ΨB (n1 . . x2 ..8. . λN ) ⊂ SN is the set of all permutations involving only particles j with diﬀerent λj . . . . λ2 . characterize the wave function (9. . . xN |Ψ Fock (λ1 . N x1 . 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. . . i. λN ) only if λj = λk . the nj specify how often a single particle state |j is occupied. λ2 . . λ) in which two particles occupy the same single– particle state |λ does not allow transposition of particles. .9). i.e. A wave function which obeys the boson symmetry (9. . The integers nj are equal to the number of λk in (9. . .e. . x2 . . . . nS ) = S j=1 (9. referred to as the occupation numbers. .13) These states form an orthonormal basis of N -particle states. . . T (j.15) with λk = j. therefore. (9.λN ) on the rhs. λN ) do not exhibit the symmetries dictated by (9. e. . λj ∈ {1. xN |ΨFock (λ1 . .. .. the reason is that such transposition duplicates the state which would. n2 . the states |ΨFock (λ1 .

. for example.e.. . . . . . the photons of the electromagnetic ﬁeld or the phonons in a crystal.. ..9). . . xN |ΨFock (λ1 .9. . . (b) Show that for the states (9. . for example.1. . .e.. correspond to a particular choice of placing N particles consistent with the occupation numbers (n1 . electrons in an atom.1: (a) Show that the states (9. .λN ) (9. Such wave function is x1 . for two sets of occupation numbers N = (n1 .20) . 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 ) and N = (n1 . . . . n2 . . . . n2 . . (9.. . a molecule or a crystal. the integers nj denote the occupancy of the single–particle state |j . i.17) describe. φλ2 (x1 ) φλ2 (x2 ) . .17) Here the states (λ1 . . xN |ΨF (n1 . x2 . n2 . . λ2 . . . .15) holds (9.8). .15) describe. . . . . . . . nS ) holds S Ψ (N )|Ψ (N ) = j=1 B B δnj nj .15) obey the boson symmetry (9. . Using the identity governing the determinant of N × N –matrices A N det(A) = P ∈SN P j=1 Aj P (j) (9. i. . .1: Bosons and Fermions 243 The wave functions (9. . x2 .λN ) on the rhs. .16). n2 . .. φλN (xN ) Here. The fermion states (9. . n2 . . . . These states form an orthonormal basis. φλN (x1 ) φλN (x2 ) .19) φλ1 (xN ) φλ2 (xN ) . λ2 . . The important property holds that nj can only assume the two values nj = 0 or nj = 1. x2 . For any value nj > 1 two or more of the columns of the Slater matrix are identical and the wave function vanishes. nS ) = φλ1 (x1 ) φλ1 (x2 ) . . xN |ΨF (n1 .16) Exercise 9. . (9. nS ) and N = (n1 . √1 N! (9. . n2 . for two sets of occupation numbers N = (n1 . These states form an orthonormal basis. . .18) this wave function can also be expressed through the so-called Slater determinant x1 . nS ). . . n2 . nS ) holds S ΨF (N )|ΨF (N ) = j=1 δnj nj . nS ) = √1 N! P ∈SN P P x1 . . . .

. . . nj + 1.15) and (9. . ak ] = 0 .22) The operators aj and a† obey the commutation properties j [aj . . n − 1.17) obey the fermion symmetry (9. . . guaranteeing the correct permutation properties of the many–particle states. . n ) are normalΨ 1 j 1 j S S ized according to (9. . nj . nS ) = nj ΨB (n1 . n . nj . . To prove that (9. . nS ) = j nj + 1 ΨB (n1 . since both the N -particle wave function B (n . [a† . . . nS ) .20). (9. nS ). nj − 1. . .21) √ The factor nj appears here on the rhs.9). . . (b) Show that the states (9.15). n2 . . a† ] = δj k . . . . n ) and the N − 1-particle wave function ΨB (n .24) [aj .21) we consider the matrix Aj corresponding to aj in the basis of many–particle states (9.25) .23) (9. Creation and Annihilation Operators for Bosons We consider the operator aj deﬁned through aj ΨB (n1 . Employing the superindices N and N introduced above and using the orthonormality property (9. . . . . . a† ] = 0 j k (9. . . . . (9. . are the so-called operators of 2nd quantization. . Exercise 9.2 Operators of 2nd Quantization An important tool to describe mathematically systems of many bosons or fermions. .22) follows from (9. nS ) nj ≥ 1 0 nj = 0 √ .16) one obtains (Aj )N = ΨB (N )| aj |ΨB (N ) = √ S N nj δnj nj −1 k=1 k=j δnk nk . . . . . . . . . . . 9.244 Many–Particle Systems The representation of wave functions (9. .15). . These operators allow one to construct and manipulate many–particle wave functions while preserving permutation symmetry. 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 (9.17) obey the orthonormality property (9. For the adjoint operator a† holds j a† ΨB (n1 .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 .1. . . . . . . (9.

. S ˆ N |ΨB (n1 .29) |ΨB (n1 .30) Of particular interest is the operator ˆ Nj = a† aj .27) Equations (9. n2 = 0. (9.31) This operator is diagonal in the occupation number representation. a† ] = 1 For j = k follows similarly 1. . . . . j (aj a† − a† aj )|Ψ(N ) = k k ( nj (nk + 1) − (nk + 1)nj )|Ψ(n1 . . . n2 .27. It holds for j = k (aj a† − a† aj )|Ψ(N ) = (nj + 1 − nj )|Ψ(N ) = |Ψ(N ) .28) (9. nS ) N |ΨB (n1 . .33) is called the particle number operator since. . nS ) = |ΨB (N ) .24). nk + 1 .24).34) = . (9. . (9. The creation operators a† allow one to construct boson states ΨB (N ) from the vacuum state j |0 = |ΨB (n1 = 0.26) From this result one can immediately conclude that (9.. .22) ˆ Nj |ΨB (N ) = nj |ΨB (N ) . .2: Operators of 2nd Quantization Let A† be the matrix adjoint to Aj . . .21. . . . . nS = 0) as follows S (9. .21) and vice versa. . 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 . . i.e. One refers to Nj as the occupation number operator. obviously. . 9. (9. the eigenvalues of Nj are the occupation numbers nj . 9. nj − 1 .e. nS ) = j=1 nj |ΨB (n1 . . for basis states |ΨB (n1 . j (9. n2 . (9. . nS ) = j=1 a† j nj nj ! |0 .22) follows from (9. the properties (9. . n2 . We will prove the commutation properties (9.9. nS ) = 0 . n2 .. . One can readily show using (9. . . .32) ˆ ˆ i. nS ) . . . n2 . The related operator S ˆ N = j=1 ˆ Nj (9. j j Since this holds for any |Ψ(N ) it follows [aj .23) follow in an analogous way.29) yield the commutation relationship (9.

nS ) .29.9). φj (x1 ) φj (x2 ) . .2.24) and the property aj |0 = 0. . in case that the single particle states |j and |j + 1 are both occupied. in case it holds for nj .. xN |ΨF (n 1 . .24) and aj |0 = 0 imply that the the properties (9.38) . .. i. . xN |ΨF (n1 . . . j Exercise 9.1: The commutation relationships (9. .. nj . .. nj+1 . x2 .23) hold for the state deﬁned through (9.36) Creation and Annihilation Operators for Fermions We want to consider now creation and annihilation operators for fermions. x1 . from the vacuum state |0 or states |ΨB (N ) . . 9. .246 Many–Particle Systems In using the boson creation and annihilation operators. .22. . nj+1 = 1. x2 .. i. . ..e. e. the fermion wave function changes sign when one exchanges the order of the occupancy. ... .. .23. one needs to apply only the commutation propertes (9. xN |ΨF (n1 .. showing that the property holds for nj = 0 and. . it also holds for nj + 1. . . (n − 1)! (9. x2 . nj+1 . . of (9.21. 9.37) (9. . . nj+1 . Before proceeding in this respect we need to account for the following property of the fermion wave function which applies in the case nj . . nS ) = .s.h. Prove this by induction. φj+1 (x1 ) φj+1 (x2 ) . . in general. 9. As long as one starts from a wave function with the proper boson symmetry.19) without aﬀecting the fermion symmetry (9. Obviously.38) corresponds to x1 . . .2.17. . . . . . To prove this property we notice that the l. 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. . . (9.39) . √1 N! . 9. . . . 9. . nj (9.e.2: Show that the boson operators a† and aj satisfy j [aj . nS ) = − x1 . . .30). since they are induced through the algebra of a† and aj . . nj .g. Exercise 9. one does not need to worry ever about proper symmetries of wave functions. f (a† )] j [a† ..23. 9. operators which can alter the occupancy of the wave function (9. φj (xN ) φj+1 (xN ) .35) and where the derivative operation is deﬁned through ∂f (A) = ∂A ∞ n=1 1 f (n) (0) An−1 . .

x2 . n2 .e. . Obviously. nj → 0. .2: Operators of 2nd Quantization The r. in order to be consistent with this convention.. . . . φj+1 (x1 ) φj+1 (x2 ) . (9. qj is the number of states |k with k < j which are occupied in a fermion wave function. . 247 − √1 ! N . . n2 . . i. n3 . . to move the particle to the ﬁrst position (to be annihilated there) or to move it from the ﬁrst position to its canonical position (after it had been created at the ﬁrst position).e. .40) . . .9.19). x2 |ΨF (n1 = 0. of (9. . n6 = n7 = . . particles are being created or annihilated. n4 = 0.39) and (9. φj (x1 ) φj (x2 ) . . For this purpose one must adhere to a strict convention: the labelling of single–particle states by indices j = 1. i.. it is necessary to specify for a fermion wave function the order in which the single– particle states |λj are occupied..e.41) Before we consider the deﬁnition of fermion creation and annihilation operators we need to take notice of the fact that one also needs to deﬁne at which position in the wave function. . nj+1 .. One adopts the convention that particles are created and annihilated by these operators always at the ﬁrst position of the wave function. n5 = 1. . nS ) as follows: j−1 qj = k=1 nk . . however. n2 = 1.19). . the rhs. . ..42) i. . 2.h. . φ2 (x2 ) φ5 (x2 ) (9.. . . at the ﬁrst column of the Slater determinant (9. . . .40) are identical. . . nj = 0. at which column of the Slater determinant (9. nS ). the expressions (9. n3 = 0.. . . .s. 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. We are now ready to deﬁne the annihilation operator cj for fermions in the single–particle state |j as follows cj |ΨF (n1 . 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. . nj−1 . φj+1 (xN ) φj (xN ) . . nS = 0) φ2 (x1 ) φ5 (x1 ) 1 1 = √2 = √2 [φ2 (x1 )φ5 (x2 ) − φ5 (x1 )φ2 (x2 )] .. This requires one. vanishes. We will be adopting below the notational convention that qj corresponds to occupancies N = (n1 . n2 . . . nj .. that occupancies are always ordered according to a monotonous increase of the single–particle state index. i.. nS ) = . . . (9.. .e. . nj . n2 . nj+1 . nS ) = nj (−1)qj |ΨF ( n1 . Because of the property of the determinant to change sign when two columns are interchanged. (9. . This change of position is connected with a possible sign change (−1)qj where qj is deﬁned for a given N = (n1 .38) reads − x1 . xN |ΨF (n1 . .. . .43) .

of (9. . (9. . One obtains for its matrix elements j C† j NN = ΨF (N )| c† |ΨF (N ) = (Cj )N j S N = nj (−1)qj δnj nj −1 k=1 k=j δnk nk S = (−1) (1 − nj ) δnj nj +1 k=1 k=j qj δnk nk . nj → 1 . We consider (9. We have used here the deﬁnition qj = k<j nk along with qj = qj . n2 . .50) . 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 .48) vanishes. .19). the last factor on the rhs. . . ck ]+ = 0 .48) From this follows (9. n2 . . . (9. nj−1 . c† ]+ = δj k k where we have used the deﬁnition of the so-called anti-commutators [A. We will prove now the anti-commutation property (9.17.46) [cj .45) (9. We ﬁrst show that (9. nj → 0 .46).44) follows from the deﬁnition (9. .46) ﬁrst for the case j = k. nS ) = j (−1)qj (nj + 1))(−1)qj (1 − nj ) |ΨF (N ) = (1 − nj ) |ΨF (N ) (9. The anti-commutation properties (9. . nS ) = j j (−1)qj (1 − nj + 1)(−1)qj nj |ΨF (N ) = nj ΨF (N ) (9.44). . . . c† ]+ = 0 j k (9. nj .248 The operator c† adjoint to cj exhibits the following property j c† |ΨF (n1 . [c† .45) follow in an analogous way and will not be derived here. B]+ = AB + BA.49) The derivation involves realization of the fact that a non-zero result is obtained only in case nj = 1. nj+1 .47) Let C† be the matrix adjoint to Cj . Similarly one obtains cj c† |ΨF (N ) = cj (−1)qj (1 − nj ) |ΨF (n1 . nS ) qj states occupied The operators thus deﬁned obey the commutation properties [cj . otherwise.43). . It holds c† cj |ΨF (N ) = c† (−1)qj nj |ΨF (n1 . .44) (9. nS ) = j Many–Particle Systems (−1)qj (1 − nj ) |ΨF ( n1 . . n3 . Let Cj be the matrix corresponding to the operator cj in the basis of fermion states (9. . nj → 1.

h.52) One can obtain similarly the same relationship in the case j > k. is referred to as the occupation number operator. . . Equation (9.e. .57) On the l. Correspondingly. .49) and (9. . nλ2 .9. . 1 In the following we will apply fermion operators c† and cj only to electrons which carry spin 2 . Since (9.53) is diagonal in the occupation number representation. nS ) k and. k k 249 (9. j j j j j j j (9. . for which purpose one needs to envoke only the algebraic (anti-commutation) properties (9.58) . .) = c† 1 c† 2 · · · c† N |0 . nk = 1 . nj = 0 .54) ˆ N = j=1 ˆ Nj (9. with eigenvalues equal to the occupation numbers nj ˆ Nj |ΨF (N ) = nj |ΨF (N ) ˆ Nj . j → j.51) (9.2: Operators of 2nd Quantization (9. nk = 1 . . nj = 0 .50) yield ( cj c† + c† cj ) |ΨF (N ) = |ΨF (N ) . nS ) k c† cj |ΨF (N ) = (−1)qj +qk −1 nj (1 − nk )|ΨF (n1 .29)) |ΨF (. . . 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 . coordinates x and index j into a space part and a spin part. . 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.. . i. We j will often separate the states |j . 2 (9. . ˆ It is an interesting exercise to demonstrate that Nj only has eigenvalues 0 or 1.. i. . hence.46). λ λ λ (9.51.46).55) is called the particle number operator. nλN .52) hold for any |ΨF (N ) one can conclude (9.h. On the r. in the basis |ΨF (N ) . . . σ 1 (σ = ± 2 ) . i. . of this equation we meant to indicate only those N occupation numbers nλj which are non-vanishing. the creation operators must operate in the proper order. 1}. x|j → φj (r) | 1 σ . j j For j < k one obtains cj c† |ΨF (N ) = (−1)qj +qk nj (1 − nk )|ΨF (n1 . an operator cj must be on the left of an operator ck for j < k.56) Here we have made use of cj cj = 0 and c† c† = 0 which follows from nj ∈ {0. 9.49) shows that the operator ˆ Nj = c† cj j (9. . ( cj c† + c† cj ) |ΨF (N ) = 0 .e. S (9.. . hence.s.s. nλ1 .

σr ) f (x) ψs (r.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. ˆ . The essential aspect of deﬁnition (9. The operators g (x.F (N ) . . We adopt a similar expression to distinguish two–particle operators G and g . 9.59. x ) which constitute G are ˆ called the associated generic two–particle operators.60) involves generic operators – acting on xj and on xj . xN .15. ˆ ˆ ˆ We seek to determine now how one–particle operators F and two–particle operators G act on many– B (N ) and |ΨF (N ) . (9.60) is that the sum over particles in (9. .e. An operator N ˆ R = j=1 rj (xj ) ˆ (9. xk . r = 1. . 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. i. 9. x2 . namely. respectively – which are all identical. .F (N ) .61) is not a one particle operator as long as the N operators rj (x). . 9. respectively.59. . Since the many–particle states are built-up from a basis |r . . xk ) ˆ j.60) ˆ correspondingly are called two–particle operators. The operators f (x) which constitute F are called the associated generic single–particle operators1 .6) above. Operators of the type 1 ˆ G = 2 N g (xj . (9. one expects that the matrix elements can actually be expressed in terms of matrix elements ˆ involving solely the generic operators and the single–particle states. The action of the operators boson and many–fermion states |Ψ is obviously described through the many–particle state matrix elements ˆ ΨB. 9.63) ˆ ˆ This expression has been speciﬁed for the purpose of these notes to distinguish F and f and is not common ˆ terminology..F (N )|F |ΨB.59) and over pairs of particles in (9. S of single–particle spinorbital states (see page 241) and the operators are speciﬁed through the associated generic operators ˆ f and g . These operators had been introduced already in Eq.62) These matrix elements can be obtained by chosing the operators and many-particle states in the position representation.17).17) and integrating over all particle space-spin coordinates x1 . 2. . σs ) (9. 2.59) and (9.250 Many–Particle Systems 9. (9. . j = 1. ˆ r|f |s = 1 ∗ ˆ d3 r ψr (r.15.59) ˆ ˆ are called single–particle operators. ˆ ΨB.k=1 j=k (9.F (N )|G|ΨB. . N are not all identical.

9. Examples of One– and Two–Particle Operators An example for a single–particle operator is the kinetic energy operator N ˆ T = j=1 2 − 2m ∂2 .63. g g (9. x ) ψt (r.59).64). σu ) .69). σt )ψu (r . i.64) r. (9. We will show that the boson and fermion creation and annihilation operators serve this purpose. Notice. s|ˆ|t.65) This symmetry will be exploited repeatedly below.68) Another single–particle operator is the one–particle density operator N ρ(r ) = ˆ˜ j=1 δ(r − rj ) .k=1 j=k q2 |rj − rk | (9. ∂r 2 ∂x2 ∂x2 ∂x2 1 2 3 (9. according to (9. ∂rj2 (9. ˆ 251 (9.70) Both operators.66) and (9. r|ˆ|u.e. 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..64) We like to note an important symmetry of the matrix element r.66) Here we adopt the notation for the Laplacian ∂2 ∂2 ∂2 ∂2 = + + . one cannot switch the indices of one pair individually. 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. ˜ (9. are spin-independent as is evident from the factor δσσ .67) ˆ In this case the generic single–particle operator. u = g d3 r ∗ ∗ d3 r ψr (r.69) ˆ In this case the generic operator is f (x) = δ(r − r). σs ) g (x. s and t. s|ˆ|t. 9. u = s.71) . u which follows directly g from the deﬁnition (9. ∂r 2 (9. σr )ψs (r . An example for a two–particle operator is the Coulomb repulsion operator 1 ˆ V = 2 N j. u. is t(x) = (− 2 /2m)(∂ 2 /∂r 2 ). t .62) in terms of the matrix elements of single–particle states (9.3: One– and Two–Particle Operators r. s|ˆ|t. that the symmetry implies that one can switch simultaneously the pairs of indices r. 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) .

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

82) ΨF (N )|c† cs |ΨF (N ) = r cr 2 (9.45.79) is opposite to that in the matrix element r.46). The Matrix Elements ΨF (N )|c† cs |ΨF (N ) r We consider ﬁrst the case r = s.83) t=1 t=r.62) through single–particle state matrix ˆ ˆ elements (9.s.78. s|ˆ|t. Expressions (9.s. . i. s|ˆ|t.81) For the latter matrix elements simple rules can be derived from the algebraic properties of the boson and fermion operators (9. u .k=1 j=k S r.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.23. xk ) → ˆ j. 9. and then in g a second step the matrix elements ΨB (N )|a† as |ΨB (N ) . These rules will be provided below only for the case of fermions. g and not ct cu .24) and (9. 9.64).62). The matrix elements are then actually those of the number operator c† cr which is diagonal in the occupation number representation. pp.171 and pp. 9. u a† a† at au bosons g r s † † r.63) and (9. r ΨB (N )|a† a† at au |ΨB (N ) r s ΨF (N )|c† c† ct cu |ΨF (N ) r s bosons fermions (9.59) and G in (9. r S Ψ (N In case of r = s one obtains F )|c† cr |ΨF (N ) r = nr r=1 δnr nr .e. s|ˆ|t.176 where the matrix elements of fermion states can be found. u cr cs cu ct g fermions (9.79) have the following implication: The operators (9. To show that the resulting values of the matrix elements (9.78) and (9.3: One– and Two–Particle Operators Deﬁnition in Terms of Creation and Annihilation Operators 253 In order to express many–particle state matrix elements (9. s|ˆ|t..62) are correct one needs to compare the result derived with conventional derivations of the matrix elements2 .t.79) in the basis of many–particle states |ΨB. respectively.F (N ) have the same values as the respective matrix elements (9.64) one replaces the operators F in (9.63.u=1 (9. r ΨF (N )|c† cs |ΨF (N ) .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’. namely cu ct .u=1 S r. 9.80) (9. respectively. (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 .60) as follows N ˆ f (xj ) → j=1 N 1 2 1 2 S r.t. u as deﬁned in (9.s=1 S r.79) It is crucial to notice in the expression for the fermion two–particle operator that the order of the annihilation operators in (9.

53) 1 ˆ G0 = 2 S r s|ˆ|r s − r s|ˆ|s r g g r. (r.s=1 a. i. t).86) into three contributions. Case 1 three of the four indices are diﬀerent. Case 2 all four indices are diﬀerent.d. s.i.86) For the combination of indices r.’ (all indices diﬀerent) that only tuples (r.84) we can conclude that for diagonal elements holds S Many–Particle Systems ˆ ΨF (N )|F |ΨF (N ) = r=1 ˆ nr r|f |r (9. where each contribution corresponds to one of the three cases mentioned.i.s=1 a. u in this sum three possibilities exist Case 0 two of the four indices are diﬀerent.79) and using c† r we can write S 1 ˆ G = rs|ˆ|tu c† c† cu ct g r s 2 r. we split the sum in (9.d. t=u 2 = 0. G = G0 + G1 + G2 . for which holds r = s. s. ˆ ˆ Nr Ns .e... t.85) All other matrix elements vanish.d. s). those for which N and N diﬀer in more than two occupation numbers. r = t etc.59) when the matrix elements are determined between Slater determinant wave functions. (9. 171) shows that the operator deﬁned in (9. u) are included in the summation for which all indices are diﬀerent. in which case ns = ns + 1 and nt = nt − 1 holds. s.82. ˆ ˆ ˆ ˆ Accordingly. The Matrix Elements ΨF (N )|c† c† ct cu |ΨF (N ) r s Before we determine these matrix elements we will investigate which possible combination of indices r.87) Here and in the following we denote by ‘a.i.s=1 a. namely.t. 9. (9. The anti-commutation property (9. (r. 1 2 S r s|ˆ|s r c† c† cr cs .84) The oﬀ-diagonal elements with nr = nr except for r = s.e.d.u r=s. Comparision with the results in Condon&Shortley (pp. i. u need to be considered. are ˆ ˆ ΨF (N )|F |ΨF (N ) = (1 − ns )(−1)qs nt (−1)qt s|f |t (9.78) does yield the same results as the operator deﬁned in (9. c2 = 0 r (9.254 From (9. ˆ The ﬁrst contribution G0 is 1 ˆ G0 = 2 S r s|ˆ|r s c† c† cs cr + g r s r.s.i. t. t. Starting from the deﬁnition (9. t. s. g r s r.88) .45) yields together with the deﬁnition of the occupation number operator (9.

t.i. (9.d. s|ˆ|r. r s (9.i. S r. employing the symmetry property (9.s. One obtains 1 ˆ ΨF (N )|G|ΨF (N ) = 2 S nr ns r.91) ˆ G2 = r.9.s.i.u r<s.s.3: One– and Two–Particle Operators 255 Obviously.s.t=1 a. s (9.t.s.t=1 a. ˆ Case 0 This case covers diagonal matrix elements.90) ˆ Since Nr is diagonal. r g g .t<u S r s|ˆ|t u c† c† cu ct g r s r s|ˆ|t u c† c† cu ct g r s r s|ˆ|t u c† c† cu ct g r s r s|ˆ|t u c† c† cu ct g r s (9.s.t=1 a. (9.84) ˆ of the one–particle operator F .s. 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. contributes in this case. ns .t=1 a. S r.s. As we will ﬁnd.86).89) S r.s.d.45).i.e. A similar contribution does not arise for F .t>u S r.93) . F (N )|G|ΨF (N ) . For this purpose we consider three ˆ We can state now the matrix elements Ψ cases which actually correspond to the three cases considered below Eq.92) ˆ This contribution to G has non-vanishing matrix elements only when N diﬀers from N in four ˆ occupation numbers nr . 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. nt .d.u r>s. Only G0 .65).t.d.u r>s.u r<s. given in (9.t.t<u S r. ˆ The second contribution G1 is ˆ G1 = + + + 1 2 1 2 1 2 1 2 S r.d.i. this contribution to the two–particle operator is diagonal in the occupation number ˆ representation.t.s r.. g r s (9.t>u r s|ˆ|t u − r s|ˆ|u t g g c† c† cu ct .88). i. N = N or nr = nr for all r.t>u S r. Commutation of ct cu according to (9. r s|ˆ|r t − r s|ˆ|t r g g ˆ c† ct Nr . 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 . this contribution obviously behaves similarly to the oﬀ-diagonal part (9.u r>s. this part accounts for all diagonal contributions to G. s|ˆ|s.s.t=1 a. nu . s − r.

Derive (9. 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. it holds S nr = S nr .76). The matrix elements of the corresponding generic operators are (9. Only G2 . given in (9.4: 9. u < v or s > r. respectively.3. All matrix elements which are not covered by the three cases above vanish.e. u > v or s > r.95) In the latter formula the ‘+’-sign applies for s < r.5: Derive (9.3: 9. given in (9. i. r g g (9. s|ˆ|t.73) and S 2 given in (9..1: 9.92). contributes in this case. v − r.74) and (9.3.3. nt = nt + 1.3. ˆ ˆ Show that the matrix elements of F and of G conserve particle number. Derive (9.91).3. Only G1 . one–particle and two–particle operators. In particular. s|ˆ|r.2: 9.90) from (9. u > v. nu = nu − 1. t for which holds ˆ ns = ns + 1 and nt = nt − 1. We like to express these operators through ˆ fermion creation and annihilation operators.92) from (9. contributes in this case. s|ˆ|v. One obtains ˆ ΨF (N )|G|ΨF (N ) = ± (−1)qs +qr +qu +qv (1 − ns )(1 − nr )nu nv r.62) for bosons. Derive the non-vanishing matrix elements (9. Spin Operators ˆ ˆ In this section we will brieﬂy consider the spin operators S given in (9.94) Case 2 This case covers oﬀ-diagonal elements with nr = nr except for r = s.90).t nr r. s|ˆ|u. u g g (9. u.96) . for ˆ non-vanishing matrix elements ΨF (N )|G|ΨF (N ) at most four occupation numbers nr and nr can diﬀer.89). t.75) which are. One obtains ˆ ΨF (N )|G|ΨF (N ) = (−1)qs +qt (1 − ns )nt r r=s. u < v. Furthermore. particle number is conserved. nv = nv − 1. t − r.256 Many–Particle Systems Case 1 This case covers oﬀ-diagonal elements with nr = nr except for r = s.92). v for which holds ˆ ns = ns + 1. the ‘−’-sign applies for s < r.95) from (9. r=1 r=1 Exercise Exercise Exercise Exercise Exercise 9.

4: Independent-Particle Models ˆ For S 2 one obtains 1ˆ 1 ˆ S2 = N + 2 4 S S 257 ˆ ˆ ˆ ˆ (Nrα − Nrβ )(Nsα − Nsβ ) − r. s.97).. the description of many–particle systems. Unfortunately.s=1 r. would be a simple exercise in linear algebra.97) The summation in (9. The great generality of the concepts developed in the context of many–particle systems can be judged from the fact . the spin part of the single–particle states is represented by α = 1 and β = − 1 . in crystals. molecules.s.3. If it were not for the two–particle contribution to the Hamiltonian.97) is over the orbital states. atoms and nuclei. rα sβ (9.g. In fact. u c† c† cuσ ctσ v rσ sσ r. that the two–particle operator representing interactions between particles induces a variety of interesting phenomena. 9.96) and (9. In the latter case the indices r.u S (9.. 9. In all these systems the Hamiltonian is a sum of one– and two–particle operators S H = r=1 1 ˆ r r|t|s c† cs + r. the concerted behaviour of interacting systems can deviate qualitatively from that of systems made up of independent components. e. the systems investigated involve a macroscopic number of components such that statistical mechanical approaches are envoked. u refer only to the orbital part of the single–particle basis. Actually.96.9.99) has preoccupied an important part of all intellectual eﬀorts in Theoretical Physics during the past ﬁfty years. v r s 2 r. evaluation of their stationary states and excitation energies.u=1 σ.6: Derive (9. the study of problems governed by Hamiltonians of the type (9.99) will be used. superconductivity. The outcome of these studies is that even when interactions between components (particles) are simple. which represents the interactions between particles.98. but also ordinary phenomena like freezing and evaporation. Examples is the laser action. s|ˆ|u.s=1 c† c† crβ csα .s=1 σ 1 ˆ rσ r|t|s c† csσ + 2 S r. Often.98) In many cases the Hamiltonian is spin-independent and the equivalent notation S H = r.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. however. t. v c† c† cu ct .t. 2 2 Exercise 9. it is one of the main intellectual achievements of Physics among the Sciences to have addressed systematically the study of systems composed of many strongly interacting components. The fortunate side of this is.t.s. the two–particle operator makes the description of many–particle systems a very hard problem.g.σ (9. s|ˆ|t. e. 9.

according to which nuclei and atoms change their qualitative properties when they grow larger. r=1 3 Urm U ∗sm = δrs m=1 (9.1: Imagine that in a closed.101) The unitary property of Urn U† U = U U† = 1 1 or S S (9. (9.100) ˆ ˆ We will denote the matrix elements of t as r|t|s = trs .e.s=1 σ ˆ rσ r|t|s c† csσ (9. Chemistry essentially would know only one element and Life would not exist. a Hamiltonian without a two–particle operator contribution accounting for interactions among the particles. 1989) . New York.4. What might happen? Independent-Particle Hamiltonian A many–fermion system governed by an independent-particle Hamiltonian. ranging from nuclei to the molecules of living systems . i. If it were’nt for the fermion nature. can be described in a rather straightforward way. m = 1. The electronic properties of atoms with diﬀerent numbers of electrons would diﬀer little. In such a representation Ho is diagonal and the stationary states can be stated readily. i. The Aufbau principle.100) in a form in which the factors c† csσ become rσ occupation number operators.258 Many–Particle Systems that today these concepts are providing insight into the functioning of biological brains. We will restrict our description in the following to systems composed of an even number (2N ) of particles and to spin-independent interactions. Amit (Cambridge University Press. 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. The Hamiltonian of such system is S Ho = r. Transformation to a New Set of Creation and Annihilation Operators Our aim is to represent the Hamiltonian (9. water-ﬁlled jar all electrons of water turn their fermion nature into a boson 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.J. another prototypical systems of many strongly interacting components.. all material systems would condensate into states which would depend little on particle number. ultimately depend in a crucial way on the fermion character of its constituent building blocks. namely of neurons3 . would not exist. electrons and nucleons. Exercise 9. . {|m). 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).102) U ∗rm Urn = δmn . . 2. .. where S |m) = r=1 Urm |r .

of N -Fermion states in which single particle states |nj ) as deﬁned in (9. accordingly.101) are occupied.109) The operators (9. behave exactly like the operators c† and crσ behave for states |ΨF (N ) . csσ = n=1 Usn dnσ . the anti-commutation properties (9.103) allows one to express [c. dnσ ]+ = 0 .112) . i. (9.h. [dmσ .s. d† ]+ = sσ nσ r. [d† .9.110) mσ nσ [dmσ . for nσ rσ example. 9. The l.46) as c† and crσ .. In a basis of states N d† j σj |0 n j=1 (9. using (9. c† ]+ = δσσ [dmσ .104) One can. hence. dnσ = r=1 ∗ Urn crσ (9.113) Urm Usn [crσ . the expressions drived above for the matrix elements of one. The unitary property (9.f.100) yields ˜ Ho = tmn d† dnσ mσ where d† nσ = r=1 S S (9. rσ namely. nσ We demonstrate here the anti-commutation property (9. the operators d† and dnσ . .46). (9.108) These operators describe the creation and annihilation operators of fermions in states |n) which are linear combinations (9.110) can be demonstrated similarly.e.45.111) can be written.105) where ˜ ˆ tmn = (m|t|n) = S ∗ trs Urm Usn = r. d† ]+ = 0 (9.106) which together with (9.s r The unitarity property r ∗ Urm Urn = δmn yields immediately identity (??).107) Urn c† rσ . (9.104)] S S ∗ Urm d† mσ m=1 c† rσ = .and two-particle operators hold in an analogous way for d† and dnσ . expand S trs = mn ∗ ˜ tmn Urm Usn (9. d† ]+ = δmn δσσ nσ (9.111).4: Independent-Particle Models allows one to express conversely |r in terms of |m) S 259 |r = m=1 ∗ Urm |m) .101) of states |r . they are Fermion creation and annihilation operators.108) obey the same anti-commutation properties (9.s=1 ˆ U † tU mn (9. ∗ ∗ Urm Urn .108) and (9.111) and. of (9. (9.

119) In this state the N lowest orbital eigenstates of trs are occupied each by an electron with spin |α = | 1 ..101) gives us the freedom to choose the matrix (9. .115) ˜ and involves solely occupation number operators N nσ = d† dnσ . σ1 . .115) can be stated immediately since any many–particle wave function in the occupation number representation is suitable. . U2n . transformation matrices Urn and energy values n . (9.. the so-called ground state. The 2N fermion state 2N d† j σj |0 n j=1 where (nj .e. i. σ2 . .e.114) together with (9.114) Ho = n=1 † n dnσ dnσ (9. (9. S (9.. is obtained.260 Diagonal Representation Many–Particle Systems The transformation (9. Vn = (U1n . .115) with eigenvalues 2N E(n1 . USn ). . ) = j=1 nj . ˜ tmn = In this case Ho has the desired form S n δmn . − 1 .106) diagonal. j j (9.106) and the unitary property of Urn . is N |Φo = j=1 d† α d† β |0 . A non-degenerate orbital state which is occupied by two fermions. 2.e. a fermion in a spin state | 1 . a simple matter to nσ state many–particle states in the new representation. n2N . σ2N . 2 2 2 2 i.117) are eigenstates of (9. i. The condition (9.e.57) to the case of 2N particles. r = 1. 1 as well as a fermion in a spin state | 1 . The corresponding eigenvalues n . Eigenstates of (9. . hence. .118) Ground State In case of an ordering m < n for m < n the state of lowest energy. 2. S are T real.. 1 and |β = | 1 . the associated ( properly normalized) eigenvectors Vn are the columns of (Urn ). This equation poses the eigenvalue problem for the hermitian matrix (trs ). .106) is equivalent to S trs Usn = s=1 n Urn ∀r. . σk ) for j = k (9. . n2 . It is.116) which follows from (9. . σj ) = (nk . i. Before proceeding we like to address the issue how the new representation. We apply for this purpose (9. − 1 is called a closed 2 2 2 2 . n = 1.

α = 2 |β. ˆ In order to investigate the total spin of |Φo we apply to this state the total spin operator S 2 as given by (9. namely. namely. Alltogether. β = 2 |α. Similarly. The ground state has only closed shells. The two states |α. One can conclude immediately that the ˜ ˜ ˜ ˜ (N mα − N mβ )(N nα − N nβ ) does not give any contribution second term in (9.e.n=1 d† d† dmβ dnα mα nβ (9. ˆ one can conclude that |Φo is an eigenstate of S 2 with eigenvalue zero. α |α. Obviously. 1 S 4 m..n=1 d† d† dmβ dnα mα nβ (9. The action of the operator (9. |α.9..97).n=1 m. β . 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).e. 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. α .100). 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. However.n=1 † † if either |m) or |n) are closed shells. 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. ˆ S 2 ||β. of ∆E = and |β. the states diﬀer in their spin character.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. Excited States We want to construct now excited states for the system governed by the independent-particle Hamiltonian (9. i. is a singlet state. (9.120). In case of m = n it gives a contribution ‘1’ for each closed shell which cancels the contribution of the ﬁrst term.n=N (9.n=1 m.120) where the occupation number operators refer to the single-particle states |m). β |β.124) . β ˆ are eigenstates of S 2 given in (9. α |β. energy above the ground state.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. α N +1 − N . i. We will demonstrate now that this property endows the ground state with total spin zero.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.4: Independent-Particle Models 261 shell. There are four such states.120) is particularly simple for closed shells.97). both with eigenvalues 2 ˆ S 2 |α. the third term − S m.

β are not eigenstates of S 2 . α and |β. 2.σ d† +1. ˆ 3 on the two states produces . The action of Σ N +1 ˆ Σ3 |α.2: Construct four operators O ˆ O ˆ such that O ˆ S3 . mα nβ (9.σ dN σ N σ. β both with eigenvalues 1.130) . etc.4.262 and ˆ Σ3 = − m. The two states |α. The system is described by the Hamiltonian 2N ˆ Ho = − t n=1 σ c† σ cn σ + c† σ cn+1 σ n n+1 .e. .n=N Many–Particle Systems N +1 d† d† dmβ dnα . α . .. (9.126) which follows from the occurence of squares of fermion operators which.100) outlined above we consider a system of 2N electrons which move in a set of atomic orbitals |r . The system is assumed to posses a 2N –fold symmetry axis and interactions connect only orbital states |r → |r ± 1 . α ± |β. i. vanish. ˆ Exercise 9. . one with eigenvalue zero (singlet) and one with eigenvalue ‘2’ (triplet). conclude that (9. also eigenstates of Σ2 . apropriate eigenstates of the operators S 2 and Example: 2N Independent Electrons on a Ring In order to illustrate the procedure to obtain eigenstates of one-particle Hamiltonians (9. β =− m. |β. . We identify the states |2N + 1 = |1 to avoid cumbersome summation limits. hence.122) is correct. for example. 2. ˆ The remaining states |α. ˆ An eigenvalue ‘2’ of the spin operator S 2 identiﬁes the states |α.128) The cyclic property of the system is expressed through c2N +1 σ = c1 σ . c† +1 σ = c† σ 1 2N (9. α are. . β . the linear combinations 1/2(|α. α as triplet states. m |Φo m m m γσ. .129) The symmetry of the Hamiltonian suggests to choose a new representation deﬁned through the operators (r = 1.125) ˆ Contributions to S 2 acting on states |N + 2 . β and |β.σ = (9. of course. |N + 3 . The same result holds for |β. 2N which are located on a ring.n=N d† d† dmβ dnα d† +1α dN β |Φo = 0 mα nβ N (9. for |α. One can. . β qualify as eigenstates. 2N ) drσ = √ 1 2N 2N e−irnπ/N cnσ n=1 (9. do not need to be considered since they ˆ give vanishing contributions.127) ˆ are triplet and singlet states. . . r = 1. of course. however.

136) ˆ and.139) −2t cos r=1 σ rπ N d† drσ . of (9. indeed. One can then express[.137) where 1 ˜ trs = e−irπ/N 2N Using (9. 2N (9. 4N.132) gives for r = 0. . n=1 (9. rσ (9.131) The (9.134) which. (9. vanishes for integer s. . eisπ/N 1 − e2N isπ/N 1 − eisπ/N (9.133) Application to the l.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 .109)] ¸ cnσ = √ 1 2N 2N eirnπ/N drσ r=1 2N (9. thereby.s=1 σ ˜ rσ trs d† dsσ (9. (9.138) ˜ trs = from which follows ˆ Ho = eirπ/N + e−irπ/N 2N δrs (9.130.132) an identity which can be derived employing the well-known expression for a ﬁnite geometric series m as = a s=1 1 − am .h.140) The Hamiltonian is now diagonal and the construction outlined above can be applied. .s.9. . one obtains the new representation of Ho 2N ˆ Ho = −t r. .132) one can conclude 2N ei(s−r)nπ/N + eisπ/N n=1 1 2N 2N ei(r−s)nπ/N . 1−a (9.135) c† nσ 1 = √ 2N e−irnπ/N d† rσ r=1 (9. 9.f. 2N.

n = 1. Many electronic properties of solids can be understood in a similar way. (9. valence bands or conduction bands. if one places fermions into an independent-particle ground state of the type (9. one expects that mean-ﬁeld descriptions should be rather suﬃcient. We will present in this and the following section a method which constructs such optimal single-fermion states.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. The regularities of the elements can be rationalized in terms of electrons moving in hydrogen atom-type orbitals. . 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. if one can choose the singleparticle states such that the residual perturbations are small. the everpresent Coulomb repulsion between electrons. A similar principle accounts approximately for properties of nuclei as described by the nuclear shell model. i. and that independent-particle behaviour surfaces mainly because it applies well to the ground state. e. The most famous example is the periodic table of the elements. These states do not altogether neglect the pair interactions. i.. properties of many-fermion properties deviant from such simple description is then of as much interest as independent fermion descriptions are useful. A second reason why independent-particle models can be successful is that the pair interaction for electrons is the slowly decaying Coulomb interaction. 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 . in states which are constructed according to the ‘Aufbauprinzip’ in a manner which neglects the mutual repulsion between electrons. Of course. assuming that electrons move independently from each other in so-called bands. 9..141) State the excitation energy ∆E.e.. 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.e. does not spoil it. rather they assume that each particle experiences the average interaction due to the fermions frozen into the ground state. those with energy above that of the (energetically) highest occupied single-particle state. i.g.e.e..264 Many–Particle Systems Exercise 9. 2N with interactions between neighbouring (n.4. Since independent-particle behaviour is so universal one may wonder why the presence of particleparticle interactions. The slow decay of the interaction makes the motion of any one electron rather independent of the exact position of most other electrons.119) then perturbations due to pair interactions according to the Pauli exclusion principle can only involve independent particle states not occupied in the ground state. in particular. There are two reasons why this is so: ﬁrst. i. ...e. Hence. 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 . 2. . . the fermion nature restricts the possibility for perturbations on the system.

2. |r = m=1 ∗ Urm |m .s. .142) is based. m = 1. The states |m are connected with the states |r through (m = 1.143) These S states are associated with the creation and annihilation operators S d† mσ = r=1 S Urm c† rσ ∗ Urm crσ r=1 dmσ = (9. S on which (9. N . ˜ (9. p. .143) with r.u=1 σ. .u=1 σ. . u v r. q. . This is done as follows: ˆ Vmf (|Φo (U) ) = 1 2 S r. . . If one restricts m = 1. S is a unitary matrix. . . s|ˆ|t. In our formulas below we will adopt strictly the convention that indices r. . n. . . r = 1. 2.144) deﬁned as in (9.5: Self-Consistent Field Theory 265 In this section we will consider systems of 2N fermions described by the spin-independent Hamiltonian S S 1 ˆ rσ H = r. .σ and derive a single-particle operator which approximates this Hamiltonian. as in in (9. naturally. 2.146) .142) over this reference state to turn the contribution into an eﬀective one-particle operator.s. U = ( Urm ) as deﬁned in (9. the corresponding U = ( Urm ) forms an S × N matrix and. s. . j j (9.σ c† c† rσ sσ cuσ ctσ + c† c† rσ sσ c† ctσ cuσ rσ c† cuσ rσ ctσ cuσ ctσ + c† c† cuσ rσ sσ ctσ + c† sσ − c† c† ctσ cuσ − c† rσ sσ sσ (9. 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. s|ˆ|t.142) r|t|s c† csσ + rσ sσ 2 r.9. We will adopt a second set of single-particle states {|m . u refer to the initial basis of single-particle states |r . m = 1.t.145) which will play a pivotal role: it is this the state in which the fermions are assumed to be moving and relative to which the mean interactions acting on individual fermions is determined. t.108). 2. S} the elements of which will be labelled by indices ˜ m.145). S) ˜ S S |m = ˜ r=1 Urm |r . is not unitary.s=1 σ r. . u c† c† cuσ ctσ v (9.t. 2. The states |m serve to deﬁne a reference state of the system ˜ N |Φo (U) = j=1 d† α d† β |0 . .

σ S ctσ + m.s.m.n=1 σ.148) r. For the averages ··· in (9.t=1 σ.σ S d† dnσ cuσ mσ − r. n c† c† v ˜ ˜ rσ sσ r. u c† ˜ vn sσ m.u=1 σ.m. s|ˆ|˜ .σ S d† dnσ mσ ctσ (9. for example. ˆ Vmf (|Φo (U) ) = 1 2 S Φo (U)|d† d† |Φo (U) = 0 mσ nσ Φo (U)|dmσ dnσ |Φo (U) = 0 Φo (U)|d† dnσ mσ δmn δσσ 0 |Φo (U) m ≤ N m > N (9.n.150) r. n c† d† dnσ ˜ v ˜ rσ mσ r.148) are obtained in a similar way by transforming only two of the four single-particle states into the new representation.u=1 σ.n.t=1 σ. n = ˜ ˆ ˜ s. n c† ˜ v ˜ sσ m.u=1 σ.σ .t. m|ˆ|t. m|ˆ|˜ . u c† d† dnσ cuσ ˜ vn rσ mσ r. d† d† mσ nσ dmσ dnσ d† dnσ mσ = = = = and. n|ˆ|t. Since the averages aﬀect only two creationannihilation operators actually a mixed representation is most suitable ˆ Vmf (|Φo (U) ) 1 = 2 1 2 1 2 1 2 1 2 S m. m|ˆ|t.m.149) The remaining matrix elements appearing in (9.148) one obtains.u ∗ r.n. hence. s|ˆ|t.266 Here ··· denotes the average O = Φo (U)| O |Φo (U) .t. by means of the rules (9. m| v |t. m. ˆ (9.147) Since the reference state |Φo (U) is deﬁned in terms of the single-particle states |m it is preferable ˜ to switch the representation of (9. s|ˆ|m. u ˜ ˜v m. u Usm Uun . s| v t.σ S d† d† mσ nσ cuσ ctσ + r.s.84).σ S 1 − 2 where.σ S dmσ dnσ + r.146) accordingly.n.n=1 σ. n c† ctσ δσ σ ˜ v ˜ rσ m≤N r.s. Many–Particle Systems (9.

σ S + − r.156) which deﬁnes the reference state |Φo (U) .152) can then be written in the form ˆ Vmf (|Φo (U) ) = r| vmf (|Φo (U) ) |s c† csσ ˆ (9.142) by Hmf (U) deﬁned through S ˆ Hmf (U) = r. t and renaming dummy summation indices one can demonstrate ˆ ˆ that term one and term two as well as term three and four are identical and one obtains S ˆ Vmf (|Φo (U) ) = 2 r. s| v |t. s|ˆ|˜ . u c† cuσ δσσ ˜ vn sσ m≤N s. s|ˆ|t.u=1 σ...9. leaves open the question how the reference state should be chosen.u=1 σ S − m. u c† cuσ ˜ vn rσ (9.. The one-particle operator (9.t=1 σ (9.. u c† cuσ ˜ vn rσ m≤N r.2.σ r.m.151) Carrying out the sum σ δσσ = 2 leads to a factor 2 for the ﬁrst two terms. r| v |u.s=1 σ ˆ r|t|s + r| vmf (|Φo (U) ) |s ˆ c† csσ rσ (9.N (9. t| v |s.2. The mean ﬁeld approximation. m|ˆ|˜ .152) By means of expression (9.S. Exploiting the symmetry r. u − r.153) rσ r.m. s ) ˆ ˆ m≤N ∗ Utm Uum (9.t=1 σ.σ S r. .149) for the matrix elements in the mixed representation one can state the right hand side explicitly in terms of Urm . u = s.u=1 σ. m|ˆ|t. as introduced here.. At this point this state is completely arbitrary. n c† ctσ . m=1. n c† ctσ δσ σ ˜ v ˜ rσ − r.s σ where r| vmf (|Φo (U) ) |s = ˆ t.u ( 2 r. t| v |u.155) which is a function of the S × N –matrix U = ( Urm )r=1. ˜ v ˜ sσ m≤N s..154) ˆ The mean ﬁeld approximation replaces then the Hamiltonian (9. m|ˆ|˜ . We will address now a proper choice of the reference state.5: Self-Consistent Field Theory 1 2 1 2 1 2 S 267 m.

S . i. The state thus determined is referred to as the self-consistent independent-particle ground state. SCF-Algorithm. i. Let us state now the construction of the self-consistent ﬁeld ground state in more detail. S . . m = 1. U(1) is the S ×N –matrix (1) of the ﬁrst N column vectors of Urm . however.4.268 Many–Particle Systems 9..155). under conditions which often are realized. (9. .145). S . m = 1. i. . 2. 2. .158) where the labels m are ordered to obey the condition (1) m < (1) n for m < n . . 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.e.: Determine Mean Field Hamiltonian In this step the mean ﬁeld Hamiltonian S ˆ (M ) Hmf = r.159) One deﬁnes the reference state |Φo (U(1) ) using the deﬁnition (9. . the reference state resulting from step 2M + 1 (within numerical error) is equal to the reference state resulting from step 2M − 1. . .155) following the method outlined in Section 9. (9. is exact only for an independent-particle Hamiltonian. . 2. r = 1. the independent-particle nature stemming from the fact that the functional form of the ground state..s=1 σ S ˆ r|t|s + r| vmf |Φo (U(M ) ) ˆ |s c† csσ rσ = r. determining in a step (2) the corresponding mean ﬁeld Hamiltonian (9.145). can achieve this goal only iteratively. . This procedure. SCF-Algorithm. S. .4 and deﬁning this as the new reference state.s=1 σ ˆ r|hmf |s c† csσ rσ (M ) (9. .142).e. . M = 1. Step 1: Choosing an Initial Reference State One deﬁnes the Hamiltonian ˆ (1) Hmf = r.160) .157) and determines the associated diagonal representation deﬁned through S (1) ˆ r|t|s Usm = s=1 (1) (1) m Urm . steps (2) and (3) are being repeated M times until the procedure converges.e. (9. We will argue below that the self-consistent ﬁeld ground state. assuming in an intial step (1) a properly chosen reference state. often also referred to as the Hartree-Fock approximation. 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 2M. 2. 9. 2. . .119. r = 1. is the lowest energy independent-particle ground state one can construct. and obtaining in a step (3) the ground state of this one-particle Hamiltonian according to the construction in Section 9.

u − r. 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. the S × S–matrix S (M ) r|hmf |s N 269 ˆ = r|t|s + t. .e.165) can be written ˆ F = S ˆ ˆ r|f |s Prs r. m = 1. . .u=1 ( 2 r. . r = 1. S . (e) Consider the creation operators N † gqσ = m=1 Vmq d† mσ (9. . σ p = δtr δsu δσσ . 2. U(M +1) is the (M +1) S × N –matrix of the ﬁrst N column vectors of Urm . ˆ (b) How can the expectation values of the operator Prr be interpreted. m = 1. S. . r = 1. ˆ t.9. 2.e. Step 2M + 1. SCF-Algorithm: Continuation and Convergence Condition When step 2M + 1 is completed.162) is solved adopting an ordering of the labels m which obeys the condition (M +1) m < (M +1) n for m < n . .1: Let Prs be the one-particle operator with the generic operator prs = |r s|.145).. 2..145).161) is calculated.e. S . . .167) .163) The result yields the reference state |Φo (U(M +1) ) using the deﬁnition (9. s = 1.164) ˆ (a) Determine the expectation value of Prs for the state as deﬁned in (9. . . 2.s=1 σ ˆ r|f |s c† csσ rσ (9. (9. σ|ˆrs |u.166) ˆ (d) Deﬁne a similar two-particle operator Prstu . i.: Diagonalize Mean Field Hamiltonian In this step the eigenvalue problem S (M r|hmf |s Usm +1) = s=1 (M ) (M +1) (M +1) Urm m . . . t| v |s. . i. .6: Self-Consistent Field Algorithm is evaluated. 2 . M = 1. indicating that the state converged. s ) ˆ ˆ m=1 Utm ∗(M ) (M Uum ) (9. ˆ Exercise 9. i. SCF-Algorithm. t| v |u. (9. .6. 2.s=1 (9. step 2M + 2 is carried out. S do not exceed a preset threshold. etc. (c) Show that any spin-independent one-particle operator S ˆ F = r. S (9. .

173) . (9. We denote the representation which results from the SCF algorithm after its convergence as follows: S |m = ˜ r=1 (SCF Urm ) |r .7 Properties of the SCF Ground State We want to investigate now the properties of the SCF ground state.172) is obeyed.168) ˆ ˆ has the same expectation values for Prs and Prstu as the reference state (9.170) (SCF ) Here |r .. (9. i. Note that this is not an mσ S × S–matrix! Show that the reference state deﬁned through N † † gqα gqβ |0 q=1 (9.163). For this representation holds N m| t |˜ + ˜ ˆn m 2 mm | v |˜ m ˜ ˜ ˆ n˜ − mm | v |m n ˜˜ ˆ ˜ ˜ = (SCF ) δmn m (9.145).270 Many–Particle Systems which are connected with d† through a unitary N × N –matrix (Vqm ). S denotes the initial single-particle states and Urm is the unitary S × S– transformation matrix obtained in (9.171) implies that for the mean ﬁeld Hamiltonian holds S ˆ Hmf (U(SCF ) ) = m=1 σ (SCF ) † dmσ dmσ m .2: Determine the SCF ground state and its energy expectation value for a system of two particles described through the Hamiltonian (S = 2) H = σ ( c† c1σ + c† c2σ ) − t ( c† c2σ + c† c1σ ) 1σ 2σ 1σ 2σ ( c† c1σ c† c2σ 1σ 2σ σ. r = 1.162.6.168) through a physical observation? Exercise 9. .145) and (9. . (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 ) . Equation (9. (SCF ) m < (SCF ) n for m < n (9. 2. (9. .163).e. We begin by summarizing the result of the SCF algorithm.171) where the convention (9.169) 9.

93) for the expectation values (diagonal elements) of the one-particle and two-particle part of the Hamiltonian (9. i. and once as a probe particle being subject to the mean pair interaction.177). exploiting the spinindependence of (9. Since the system under consideration has only 2N particles altogether. the mean-ﬁeld Hamiltonian is diagonal in the SCF representation. (9. This over-counting leads to the correction term in (9.146) counts a particle twice.7.7: Properties of the SCF Ground State 271 i. once as a member of the ground state.2: Derive (9. r.171) yields ˆ SCF H SCF N = N (9.173) holds N ˆ SCF Hmf (U(SCF ) ) SCF = 2 r=1 (SCF ) m . 2 m=1 (SCF ) m − m. Within the mean ﬁeld approximation as described by (9.142).177). Employing expressions (9.171) in terms of matrix elements r|t|s .162) is non-linear in Urm .. u and Urm .m =1 The second term originates from the fact that in the mean ﬁeld approximation one assumes that each of the fermions is subject to an average pair interaction involving a 2N -particle reference state.142). Since this state is composed only of closed shells one can conclude following Section 9. Finally.142). v (SCF ) Show that the resulting eigenvalue problem of the type (9. the SCF ground state is N SCF = m=1 d† d† |0 .177) 2 mm | v |mm ˜˜ ˆ ˜˜ − mm | v |m m ˜˜ ˆ ˜ ˜ . mα mβ (9. ˆ SCF H SCF . (SCF ) ˆ Exercise 9. one can also determine the expectation value of SCF for the Hamiltonian (9.e. 2 m=1 m|t|m + ˜ ˆ˜ m.84) and (9.175) However.174) The ﬁrst property we consider is the total spin character of SCF . . the mean ﬁeld potential ˆ Vmf (|Φo (U) ) deﬁned in (9.1: Reformulate (9. s|ˆ|t.. Exercise 9. We like to determine next the energy expectation value of SCF .e.m =1 Comparision with (9.4 that SCF is a total singlet state. one obtains ˆ SCF H SCF N = N (9.9.176) 2 mm | v |mm ˜˜ ˆ ˜˜ − mm | v |m m ˜˜ ˆ ˜ ˜ .7.

except in case of one-dimensional systems. S N0 (9.192) derived below is ﬁlled half. Due to the lack of an exact solution in dimensions higher than one. is characterized through the parameters t. v (9. v represents the ‘on-site’ Coulomb repulsion. Both N0 and N are macroscopically large numbers. in the usual two–particle operator form. We assume that there are altogether S = 2N0 lattice sites.σ crσ + crσ cr+1. s|ˆ|u. N0 which are populated by 2N electrons (we choose an even number of electrons for convenience).σ + v 2 c† c† crσ crσ . u c† c† cuσ ctσ .t.178) v contributes only in case that two electrons occupy the same lattice site.182) n can assume values 0 ≤ n ≤ 2. In spite of the simplicity of its Hamiltonian the Hubbard model. as stated through Hamiltonian (9.179) r+1.. . The operator c† (crσ ) creates (destroys) an electron with spin σ (σ = α.178) Here the index r.181) The Hubbard model. v (9..h. t is assumed to be a positive. for example. i.e.180) V = rσ sσ 2 r. respectively) in state |r at lattice site r. The one-dimensional Hubbard model is described by the Hamiltonian ˆ H = −t rσ † c† r+1.s.178). The ﬁrst term on the r.. We consider as an example the Hubbard model of an inﬁnite linear lattice and apply the model to describe magnetic instabilities in metals. . .7 for ﬁnite systems.128).178) and through the so-called ﬁlling factor n n= 2N N = . (9. labeled r = −N0 + 1.178).σ crσ + crσ cr+1.8 Mean Field Theory for Macroscopic Systems The SCF algorithm has been developed in Sections 9.272 Many–Particle Systems 9. describes the sites of a linear lattice.s.σ rσ The potential energy term. rσ rσ r σσ (9. v of the Hubbard Hamiltonian (9. . cannot be solved exactly. the second term on ther. r ∈ Z. of (9. of (9. t = v δru δst δrs .u σσ where r. β for ‘up’ and ‘down’ rσ spins. i. The case n = 1 is termed the half ﬁlled band case. is identical to that of the independent–particle Hamiltonian H0 (9.s. 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. † ˆ Ho = −t c† . approximation schemes like the self-consistent ﬁeld approximation play an important role. According to (9. real number.e.4–9. s|ˆ|t. in the two-dimensional copper oxide lattices of high temperature superconductors. reads 1 ˆ r. t describes the coupling of state |r to states |r ± 1 at the two neighboring lattice sites.178).h. referring to the fact that in this case the band of single particle energies (9. In the present Section we want to adapt the algorithm to systems containing a macroscopically large number of fermions.e. . i.

We can now embark on step 2 of the SCF algorithm. the operator H0 can be re-written in the diagonal form N0 ˆ H0 = m=−N0 +1 σ m d† dmσ mσ (9.σ m=0 ∗ Urm Urm c† crσ . a freedom which we employ to adopt so-called periodic boundary conditions like those for the example “2N independent electrons on a ring” on page 262. This task has ˆ been solved already on page 262.8: Macroscopic Systems 273 Since we are dealing presently with a macroscopic system.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. rσ rσ (9. The unitary transformation which diagonalizes H0 . < ±(N0 −1) < N0 holds as can be readily veriﬁed. rσ (9. Inserting (9. For this purpose we apply the SCF theory presented in Section 9.e.184) ˆ We are now ready to apply step 1 of the SCF algorithm and diagonalize term H0 .155. is Urm = √ 1 exp (irmπ/N0 ) .184) and using (9.188) .183) where the operators d† are related to the original creation operators c† through the unitary mσ rσ ˆ transformation U deﬁned in (9..154).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 . we identify |r + N0 = |r and later let N0 go to inﬁnity.6.185) ˆ According to (9.181. The self-consistent ﬁeld Hamiltonian Hmf which corresponds to the Hubbard Hamiltonian (9.174). we assume that the SCF ground state is given by (9. by N −1 ||SCF = m=0 d† d† |0 .187) The energy levels are labeled such that 0 < ±1 < . 9. 2N0 (9. boundary eﬀects are assumed to be negligible. Accordingly.185) into (9.179). 9. Accordingly. in fact. N0 (9.9. Using (9. i. mα mβ (9.178) can be determined by applying (9.178) within the framework of the self-consistent ﬁeld theory.144). .140). one obtains for the mean ﬁeld Hamiltonian N −1 ˆ ˆ Hmf (U) = H0 + v r.

274 Many–Particle Systems Expressing c† . by ﬁlling up this energy band with electrons from the bottom of the band (which corresponds to k = 0) to a maximum energy. π]. Apparently. N0 ˆ (2) ˆ Hmf = Hmf = m=−N0 +1 σ m + vn † dmσ dmσ 2 (9. but may diﬀer quantitatively. . for km = mπ/N0 holds km ∈ −π + N0 .1. Indeed. hence. as given in (9.192) This dispersion law is presented in Fig. The bottom of the energy band corresponds to (0) = −2t and the width of the energy band is 4t. . To demonstrate this we consider an alternative formulation of the mean ﬁeld approximation for Hamiltonian (9. the ground state of the Hubbard model in the self-consistent ﬁeld approximation can be determined exactly.190).109) and using (9.187) form a quasi1 2 continuous energy band. is already diagonal. dnσ according to (9. the state with the lowest possible energy. called the Fermi energy. 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. the electrons in the present description behave like independent particles with energies ¯m = m + vn . i. according to the Pauli principle. through the so-called dispersion law (k) = −2t cos k .191) where m is given in (9.185) one can prove rσ mσ N0 c† crσ = rσ rσ m=−N0 +1 σ d† dmσ . Usually. the results of diﬀerent formulations agree qualitatively. the self–consistency condition is exactly met and the SCF algorithm converged after two steps. denoted by F (see Fig. crσ through d† . in the limit N0 → ∞.182) of the ﬁlling factor n [c. .. When N0 becomes macroscopically large the single particle discrete energy levels (9.. Alternative Description of the Mean Field Approximation The state (9. even for one and the same Hamiltonian. One can see that −2t ≤ (k) ≤ 2t holds. The energy Emf of this state can be calculated readily by using (9.e.189) and.183) is not the only candidate for the mean ﬁeld ground state of the Hubbard model.e. In other words. ˆ The Hamiltonian Hmf . can be obtained.182)]. The ground state of the system. −π + N0 . 9. −π ≤k ≤π . (9.190) We have used in the latter expression the deﬁnition (9.177) and mm |ˆ|mm = mm |ˆ|m m = ˜˜ v ˜˜ ˜˜ v ˜ ˜ which holds in the present case. . Indeed. holds k ≡ km ∈] − π. there exist several ways of formulating mean ﬁeld approximations. (9. mσ (9. π].178). i. π and. 4 So far we have not exploited the fact that N0 and N are macroscopically large quantities. namely. 9.187). The energy m can be expressed as a function of a continuous variable k ∈] − π.1).f.

say the one given in (9. rσ rσ rσ (9.193) where.8: Macroscopic Systems 275 Figure 9.196) (9.56). Nrσ = SCF ||Nrσ ||SCF . i.e.. the deviation of Nrσ from its mean value Nrσ for the corresponding reference state.193) into (9. (9. The anti-commutation properties (9.46) 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 . we have employed (9.195) ˆ ˆ ˆ and where δ (Nrσ ) = Nrσ − Nrσ describes the ﬂuctuations of Nrσ .45..1: The dispersion law (k) for independent electrons on an inﬁnite one dimensional lattice. let us express the interaction term in (9.e.9.194) ˆ For a given reference state. 9. F = (±kF ) denotes the Fermi energy. Inserting (9. (9. The mean ﬁeld approximation neglects the . where Nrσ is the mean occupation number of the one particle state |rσ . in the last term on the right hand side. i. the operator Nrσ can be written ˆ Nrσ = Nrσ + δ (Nrσ ) .183).178) and performing the summation over the spin indices results in ˆ ˆ H = H0 + v r ˆ ˆ Nrα Nrβ . First.178) in terms of the occupation number operaˆ tors Nrσ = c† crσ .

For such state the mean occupation number nrσ is uniform at all lattice sites. Because of Nrσ = c† crσ rσ N −1 = m. non-vanishing magnetization. . consequently.195). however. (9. yields essentially the same mean ﬁeld Hamiltonian (9. the ground state of the system can have a non-zero local spin and.200) allowing. Therefore.m ∗ Urm Urm d† dm σ mσ = m=0 ∗ Urm Urm = n N = . 2N0 2 holds ˆ Hmf = = mσ mv ˆ H0 + 2 vn2 2 rσ nv † vn dmσ dmσ − N . that nα and nβ assume diﬀerent values.194) by dropping all the terms which contain the ﬂuctuations to second order. i.199) which is identical to expression (9. separating the occupation number operator into a mean value plus ﬂuctuation and neglecting the ﬂuctuations in 2nd order.191) for the ground state energy.. is given by (9.146). In case nα = nβ the following construction will lead to a ground state which coincides with the non-magnetic ground state (9. in the mean ﬁeld approximation.e.198) Let us assume that the ground state of the Hubbard model. but the mean number of particles with spin α can be diﬀerent from the mean number of particles with spin β.276 Many–Particle Systems ﬂuctuations to second order and the mean ﬁeld Hamiltonian is obtained from (9. m+ 2 2 c† crσ − 2N0 rσ (9.183). In this approximation one can express ˆ ˆ Nrα Nrβ = ≈ = ( Nrα + δ (Nrα )) ( Nrβ + δ (Nrβ )) Nrα Nrβ + Nrα δ (Nrβ ) + Nrβ δ (Nrα ) ˆ ˆ Nrα Nrβ + Nrβ Nrα − Nrα Nrβ (9.197) and the corresponding mean ﬁeld Hamiltonian becomes ˆ ˆ Hmf = H0 + v r ˆ ˆ Nrα Nrβ + Nrβ Nrα − v r Nrα Nrβ .198) as the one obtained by using the mean ﬁeld potential (9. Note that the procedure employed in (9.183) . 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 . Spin-Polarized Mean Field Ground State ˆ We want to consider now a ground state for Hmf deﬁned through Nrσ = nσ = const (9.

one can expect the system to lower its energy by assuming nα = nβ and. Therefore. (9.g.9. i. In this limit the discrete energy spectrum (9.206) ascertains that condition (9. The actual values of nσ (σ = α. 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) .206) ˜mf is minimized where µ is the Lagrangian multiplier considered an independent variable.8: Macroscopic Systems from which we obtain nσ = The mean ﬁeld Hamiltonian is then ˆ Hmf = m 277 1 2N0 Nmσ . 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. (9..201) ( m ˆ + vnβ ) Nmα + ( m ˆ + vnα ) Nmβ − 2N0 vnα nβ . This last equation tells us that the electrons with spin α (β) are accommodated by an energy sub–band α (k) ( β (k)) which is obtained from the dispersion law of the non-interactive electrons (k) by an overall shift of vnβ (vnα ). According to this method one minimizes ˜ Emf = Emf + µ (n − nα − nβ ) . the so-called Fermi energy F = µ (see below). In fact.203) mσ = m + vn−σ . e. The additional term in (9. as is shown schematically in Figure 9. Since β (k) − α (k) = v(nα − nβ ).. N0 ∼ 1023 .σ with respect to nα and nβ .206) needs to be evaluated. The values of nα and nβ are determined by minimizing the energy density (9.202) This Hamiltonian describes a system of non-interacting electrons with a spin-dependent dispersion law (9.203) is provided by the continuous function σ (k) = (k) + vn−σ . (9. subject to the constraint n = nα + nβ . In the limit N0 → ∞ this sum can be expressed as an integral.204) Emf ≡ 2N0 2N0 m. hence. such that the larger nσ .205) Such minimization is realized through the method of Lagrangian multipliers. i. nβ and µ.2a.206). β) are determined by minimizing the energy density of the ground state Hmf 1 = [ mσ Nmσ ] − vnα nβ (9. m (9. At this point we exploit the fact that our system is macroscopically large. E with respect to nα . 2π (9. a spin–polarized ground state.e. if nα > nβ then β (k) > α (k). the smaller is σ (k) for a given k.205) is met.207) where (k) is given by (9. For this purpose the sum in (9.e.. (9.192). 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.208) .

278 Many–Particle Systems (a) (b) Figure 9. unit energy and a given spin orientation of the particle.211) In order to calculate the integral we recall the following property of the Dirac–delta function δ [f (x)] = i δ(x − xi ) . (b) Graphical deﬁnition of the limiting energies α and β . gives the the number of available one particle states per site. −π . which is usually called the density of states. 2π (9. |f (xi )| (9.212) where xi are the simple roots of the function f (x). therefore. Inserting this last result into (9. In our case (k) is given by (9.210) Here δ(x) is the Dirac–delta function and ρ( ).209) where ρ( ) = −π dk δ ( − (k)) . Since the equation f (k) ≡ 2t cos k + = 0 has two simple roots.192) and. namely k1. For a given dispersion law (k) the density of states ρ( ) can be calculated by employing equation (9. 2π (9. and noting |f (k)| = |2t sin k| = 2t one obtains δ( − (k)) = 1 − cos2 k = (2t)2 − 2 .2: (a) Relative position of the two energy sub–bands α (k) and β (k) for ∆ = nα − nβ > 0.2 = ±(π − arccos( /2t)).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). holds π ρ( ) = −π dk δ( + 2t cos k) . The shaded areas represent the ﬁlled portions of these bands by electrons. δ(k + π − arccos( /2t)) + δ(k − π + arccos( /2t)) (2t)2 − 2 .

in principle.218) The derivatives can be readily determined and the conditions (9.e.e. µ.222) .9.219) (9. From (9. One obtains ˜ Emf = α β ρ( )d + −2t −2t ρ( )d + vnα nβ + µ(n − nα − nβ ) .220) µ can be readily obtained µ= 1 ( 2 α + β) + vn . nα and nβ as a function of the parameters of the Hubbard model.213) can simply be replaced by 1.2].219.217) and µ.214) where σ denotes the value of (k) which corresponds to the top of the ﬁlled portion of the energy sub–band σ (k) [c. (9. (9. F = α + vnβ = β + vnα .. namely.219–9.215) nα and nβ in (9. i. Replacing again the sum by an integral over energy one can write σ nσ = −2t ρ( )d (9. the necessary ˜ Since Emf [c. (9. 2t[ then the θ function in (9. of t.220) β ˜ ∂ Emf = n − nα − nβ = 0 . namely.214) are given by the sum (9.204) through an integral expression. We can employ the results obtained to express the ground state energy density (9.8: Macroscopic Systems 279 where θ(x) is the step function. ∂µ (9. β (9. 9.221) ∂µ Equations (9.213) and carrying out the resulting integral yields nσ = 1 arcsin( σ /2t) . β . 9.216) allow one. ∂ β ˜ ∂ Emf =0. + vnα − µ = 0 . i. if we assume that is restricted to the interval ] − 2t. (9.201). π α. (9.218) read ˜ ∂ Emf = ∂ α ˜ ∂ Emf = ∂ β and α + vnβ − µ = 0 .221) and (9. v and n. θ(x) = 1 if x > 0 and θ(x) = 0 if x < 0. 2 (9..206)] depends only on continuous quantities. outside the energy band (k).f. Fig. i..f. on conditions for a ground state of minimum energy are ˜ ∂ Emf =0. ∂ α ˜ ∂ Emf =0. one arrives at the following expression of the density of states ρ( ) = θ(2t − | |) π (2t)2 − 2 . to determine the unknown quantities α . (9.213) The presence of the θ function in the above formula guaranties that the density of states vanishes for | | > 2t.216) Using (9.e.

2b). (9.227) has only the trivial solution ∆ = 0. For this purpose one plots f (∆) versus ∆ as is shown schematically in Figure 9. Accordingly. i. (9.217) and (9.183) without magnetization. Indeed. First. (9. Equation (9. from (9. a magnetic ground state.. (9. f (∆) converges to a maximum value as ∆ → ∆max = n. From the deﬁnition (9. We assume in the following discussion that the implicit ∆ dependence of β in f (∆) can be neglected.219–9.222) and (9. The slope of f (∆) at the origin depends on the Hubbard model parameter v.e. (9.3]..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. Figure 9. the slope of f (∆) is positive for all ∆ ≥ 0. and only if. If the slope of f (∆) at the origin is less than 1. as reﬂected by the expression d[f (∆)] = vρ( d∆ β + ∆) = π (2t)2 v − (v∆ + 2 β) >0.229) has a wider range of validity than one can infer from the present analysis of the Hubbard model. For this purpose we consider the magnetization of the mean ﬁeld ground state ∆ ≡ nα − nβ .225) Deﬁning the function v∆ f (∆) = 0 ρ( β + )d . i. so called ferromagnetic solution. through the condition [c.f.223 one obtains ∆= and second. non trivial. the slope of f (∆) at the origin is larger than 1 [c. The Stoner criterion (9.227) can be solved graphically.228)] vc ρ ( β ) = 1 .228) Since the total number of states per site is ﬁnite.226) one can obtain ∆ by solving (cf.220) and (9.227). the Hubbard model has a ground state with ∆ = 0.3. condition (9.227) will have a second. from (9.280 Many–Particle Systems Comparision of (9.223) ∆ = 0 corresponds to the ground state (9. (9. (9. For v above a critical value vc .e.224) v∆ ∆= β ρ( )d = 0 ρ( β + )d . (9. if. One can express ∆ as a function of α and β in two diﬀerent ways.226) ∆ = f (∆) . One can determine vc by equating the slope of f (∆) at the origin to 1. condition (9. .226) follows that f (∆) is a monotonically increasing function of ∆.215) reveals that µ is indeed equal to the Fermi energy F (see also Figure 9.227) Since f (0) = 0 one can infer that ∆ = 0 is always a solution (the so called paramagnetic solution) of condition (9. (9. One still needs to determine α and β .f.223) follows α α − v β .

220. The corresponding 2t analytical expression provides the phase diagram of the ground state: when v approaches vc from above.234) 2t 2 The resulting phase diagram of the ground state..231) (9.9.205) the following set of equations determining x. is presented in Figure 9. the plot of vc vs. From (9. 9.232) states that x − y is proportional to ∆. t. (9.8: Macroscopic Systems 281 Figure 9.230) one obtains from (9. (9. Accordingly.219.4.e. 2 2 2 (9. hence.233) Employing the approximation sin(y + δ) − sin y ≈ δ cos y for small δ. β x+y x−y = = π(n − 1) 2t π (sin x − sin y) = π∆ . the magnetization ∆ of the ground state vanishes and remains zero for values v < vc . n. We like to determine ﬁnally the dependence of vc on the ﬁlling factor n. y and. n). Since ∆ vanishes near vc we set x = y + δ where δ is a small quantity. By parameterizing α and β α = 2t sin x . i. v (9. any . the ground state is characterized by the quantities (v.231) follows y= δ π π (n − 1) − ≈ (n − 1) . 9. β = 2t sin y .232) together with (9. (9. 2t As already mentioned.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. α.233) yield the desired expression vc π = π cos (n − 1) .3: Graphical solution of the mean ﬁeld equation ∆ = f (∆). Equation (9.

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

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

284 Many–Particle Systems .

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

Denoting by xµ the .5. Rather than starting from (10. x1 . We will. We will denote these vectors as follows def xµ = (x0 .e. We will ﬁnd that this deﬁnition will lead us back to the transformation law (10. namely the Klein–Gordon (Sect. 5 to the groups SO(3) and SU(2) of rotation transformations of space coordinates and of spin. 10. This choice implies dim(time) = dim(length). however. i. henceforth.7 a heuristic derivation of the two most widely used and best known Lorentz–invariant ﬁeld equations. 10.1 will be extended in Sect. x 3 ) holds (x0 )2 − (x1 )2 − (x2 )2 − (x3 )2 = (x )2 − (x )2 − (x )2 − (x )2 . In such a space Lorentz transformations correspond to 4-dimensional rotations.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)..1 Natural Representation of the Lorentz Group In this Section we consider the natural representation of the Lorentz group L. i.7) equation. 0 1 2 3 (10.1). the Lorentz transformations were formulated in a space in which the time component of xµ was chosen as a purely imaginary number and the space components real.e. Before introduco ing these general Lorentz–invariant ﬁeld equations we will provide in Sects. The elements L ∈ L act on 4–dimensional vectors of position– and time–coordinates. wave functions ψ(r.3) where x0 = ct describes the time coordinate and (x1 . 10.1) leave the quantity s2 = (x0 )2 − (x1 )2 − (x2 )2 − (x3 )2 (10. Minkowski Space Historically.1).e. x2 . i.4) invariant. x 2 . t) and vectors with functions ψ(r. x3 ) (10. 10. Note that the components of xµ all have the same dimension.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.286 Relativistic Quantum Mechanics in Sect. 10. namely that of length. 10. for the transformed space-time–cordinates x µ = (x 0 . The reason for this choice is that the transformations (10.4 to cover ﬁelds. t) as components.. we will provide a more basic deﬁnition of the transformations. but in a setting of representation theory methods as applied in Secti. x 1 . 10. The Group of Lorentz Transformations L = O(3. x2 . assume new units for time such that the velocity of light c becomes c = 1. This space is called the Minkowski space. This will provide us with a general set of Lorentz–invariant equations which for various particles take the place of the Schr¨dinger equation.5) and the Dirac (Sect. Rather than following this avenue we will introduce Lorentz transformations within a setting which does not require real and imaginary coordinates. the group of Lorentz transformations (10. x3 ) = r describes the space coordinates.

The Lorentz transformations are non-singular 4 × 4–matrices with real coeﬃcients. i. ν = 0.6) Here Lµ ν are the elements of a 4 × 4–matrix representing the Lorentz transformation.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) x µ = Lµ ν xν . ν=0 (10. 3 3 3 yµ xµ = new yµ xµ .11) This condition speciﬁes the key property of Lorentz transformations... The Lorentz transformations form the subgroup of all matrices which leave the expression (10. Aµν .e. however. the . We will exploit this property below to determine the general form of the Lorentz transformations. Aµ ν xν = µ=0 new old Aµ ν xν . xµ . The subset of GL(4. R).12) (10. 4 × 4 tensor: Aµ ν .5) into scalar products. the notation in (10. (10. The upper index closer to ‘L’ denotes the ﬁrst index of the matrix and the lower index ν further away from ‘L’ denotes the second index. 2. Aµ ν B ν ρ = ν=0 Aµ ν B ν ρ .5) invariant. (10. 3: 4-vector: xµ .6) yields Lµ ρ gµν Lν σ xρ xσ = gρσ xρ xσ . The second reason is that upper and lower positions allow us to accomodate the expression (10. 1. First.7) The reason for the notation is two-fold.8) new old old The summation convention allows us to write (10. Since this holds for any xµ it must be true Lµ ρ gµν Lν σ = gρσ .10) Combining condition (10. [ A more conventional notation would be Lµν . 0 0 0 −1 (10. the latter set constituting a group.9) 1 0 0 0 −1 0 0 ( gµν ) = 0 0 −1 0 = g . Aµ ν . Aµν .e.6) allows us to introduce the so-called summation conventon: any time the same index appears in an upper and a lower position. L ∈ GL(4. (10.] The following possibilities exist for the positioning of the indices µ. This will be explained further below. i. This condition can be written xµ gµν xν = x gµν x where µ ν (10. R). the Lorentz transformations constitute a linear transformation which we denote by 3 x µ = ν=0 Lµ ν xν .10. the latter notation will be used for diﬀerent quantities further below. summation over that index is assumed without explicitly noting it.9) and (10.

property (10. i.14) from the right by gLT and using (10. of O(3. i.288 Relativistic Quantum Mechanics elements of which satisfy this condition. L3 ∈ O(3. From this one can obtain using g2 = 1 1 (gLT g)L = 1 and.1). = g LT g = 0 1 2 −L 2 L 2 L 2 L3 2 −L0 3 L1 3 L2 3 L3 3 (10.15) one can derive LT gLgLT = LT and multiplying this from the left by by g(LT )−1 yields L g LT = g Using this result to simplify the r. L2 ∈ O(3.22) (10. of (10.h. 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) since it satisﬁes (10.14) holds for the inverse of L. hence.e. L ∈ GL(4. We want to show now L = O(3. This set is identical with the set of all Lorentz transformations L. 10.15) (10. (10.s. . LT gL = g } . R) is a group.20) results in the desired property (L−1 )T g L−1 = g .19) i. (10.12) LT gL = g . the associated inverse obeys (10. a group. is called O(3. the inverse of L 1 L−1 (10..13) L = ( Lµ ν ) = 2 0 L 0 L2 1 L2 2 L3 3 .e. 3 2 1 2 1 2 (10.10) of g one can rewrite the invariance property (10.1). We consider 1 then L1 .1) is.17) we note ﬁrst that the identity matrix 1 is an element of O(3.1).14)..14) L0 0 −L1 0 −L2 0 −L3 0 −L0 1 L1 1 L2 1 L3 1 ..e. For L3 = L1 L2 holds LT g L3 = LT LT g L1 L2 = LT (LT gL1 ) L2 = LT g L2 = g .1).1).16) The corresponding expression for (LT )−1 is obviously (LT )−1 = (L−1 )T = g L g . One can also show that if L ∈ O(3.21) i. in fact.1) ⊂ GL(4. L3 0 L3 1 L3 2 L3 3 Using the deﬁnition (10. R). (10. 1) = { L.e. This stipulates that O(3. L−1 ∈ O(3.14). In fact. employing expressions (10.1).17) one obtains (L−1 )T g L−1 = gLgggLT g = gLgLT g .18) (10.16.20) Multiplying (10.. To demonstrate the group property of O(3.

L0 0 ≥ 1} .14) which we employ in the formulation (10. 1). L0 0 ≤ −1} . M1 = a M2 }. (10.1). L0 0 ≥ 1} . L↓ .32) (10. 1). For this purpose we consider the value of C 0 0 = A0 µ B µ 0 = 3 A0 j B j 0 + A0 0 B 0 0 . there exist two other distinct classes.29) (10. L ∈ O(3.. The above shows that the set of proper Lorentz transformations L↑ allows one to generate all Lorentz transformations.10.24) or since (L1 0 )2 + (L2 0 )2 + (L3 0 )2 ≥ 0 it holds (L0 0 )2 ≥ 1. σ = 0 yields L0 0 2 − L1 0 2 − L2 0 2 − L3 0 2 = 1. 1). L ∈ O(3. Properties (10. + = − L↑ g + (10.23) One can classify Lorentz transformations according to the value of the determinant into two distinct classes. L ∈ O(3. ∃M2 . (10. + + ↑ ↑ We can also demonstrate that for A. det L = −1.28) (10. L ∈ O(3. From this we can conclude L0 0 ≥ 1 or L0 0 ≤ −1 . Obviously.12) the case ρ = 0. L↓ are disjunct sets deﬁned as follows + + − − L↑ + L↓ + L↑ − L↓ − = { L. A second class property follows from (10.25) can be stated as follows: The set of all Lorentz transformations L is given as the union L = L↑ ∪ L↓ ∪ L↑ ∪ L↓ + + − − where L↑ . (10.14). = { L. = { L.12). det L = −1. Considering in (10. It holds g ∈ L and −1 ∈ L as one can readily verify testing for property (10.31) (10. L0 0 (10. 1).25) i.33) where we used the deﬁnition aM = {M1 . L↑ . Schwartz’s inequality yields j=1 2 3 3 3 j=1 A jB 0 ≤ 0 j A 0 2 j Bj 0 2 . M2 ∈ M. For this purpose we consider ﬁrst the value of det L. One can also 1 verify that one can write L↑ − L↓ + L↓ − = gL↑ + = = −L↑ . (10. + except for the trivial factors g and −1 It is.34) j=1 j=1 . A statement on this value can be made on account of property (10.14).30) ≤ −1} . det L = 1. Using det AB = det A det B and det AT = det A yields (det L)2 = 1 or det L = ±1 .e. B ∈ L+ holds C = AB ∈ L+ . L↑ contains 1 1. + We start our investigation by demonstrating that L↑ forms a group.1: Natural Representation of the Lorentz Group Classiﬁcation of Lorentz Transformations 289 We like to classify now the elements of L = O(3.27) (10. hence. = { L.26) (10.23) and (10. + − gL↑ + = L↑ g . Lorentz transformations in L↑ . entirely suitable to investigate ﬁrst only 1. det L = 1.

since the group property of O(3.12) follows (B 0 0 )2 − 3 (B j 0 )2 = 1 or 3 (B j 0 )2 = (B 0 0 )2 − 1. L↑ is called the subgroup of proper. obviously.40) and.14) implies 1 + T g (1 + ) = g 1 1 (10. orthochronous Lorentz transformations. hence. In the following we + will consider solely this subgroup of SO(3.1).. Since the associative property holds for matrix multipli1 1 + ↑ cation we have veriﬁed that L+ is indeed a subgroup of SO(3.. | 3 A0 j B j 0 | < A0 0 B 0 0 . holds L0 0 > 0 and the value of the determinant is close to unity.1) ascertains CT gC = g it must hold C 0 0 ≥ 1. Similarly.290 Relativistic Quantum Mechanics From (10. The next group property of L↑ to be demonstrated is the existence of the inverse. if we enforce (10. obviously j=1 A0 0 B 0 0 ≥ 1. (10. Using the above expression for C 0 0 one can state C 0 0 > 0. ρ ν (10. (10. For the inverse of + ↑ any L ∈ L+ holds (10.14) actually L0 0 ≥ 1 and det L = 1 must hold. Property (10. we consider transformations Lµ ν = δ µ ν + µ ν .13). i. + i. + Accordingly.e. + transformations in a small neighborhood of 1 which we parametrize by inﬁnitesimal parameters. 1 We will then employ the Lie group properties to generate all transformations in L↑ .16). accordingly.34) provides then the estimate conclude from (10. We also note that the identity operator 1 has elements 1 + 1 µν = δµν 1 where we deﬁned1 δµν = 1 for µ = ν 0 for µ = ν (10.36) It holds. one can j=1 j=1 3 0 )2 = (A0 )2 − 1.39) where we have employed the matrix form T 1 deﬁned as in (10. 1 ∈ L↑ . Inﬁnitesimal Lorentz transformations The transformations in L↑ have the property that they are continously connected to the identity 1 1.1). To order O( 2 ) holds g + g = 0. Since A0 0 ≥ 1 and B 0 0 ≥ 1. µ ν small .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 . δ00 = 1 and It should be noted that according to our present deﬁnition holds δµν = gµρ δ δ11 = δ22 = δ33 = −1. In fact. (10.35) One can conclude. 1 0 0 = ≥ 1 and. therefore. This relationship shows (L−1 )0 0 = L0 0 .e. . these transformations can be parametrized such that a continuous variation of the parameters connects any element of L↑ with 1 This property will be exploited now in that we consider ﬁrst 1.38) For these transformations. from which one can conclude L−1 ∈ L↑ .37) (10.

10.47) 0 −1 0 0 0 0 −1 0 0 −1 0 0 0 0 0 0 0 0 K1 = .48) (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 . J2 = 0 0 0 −1 0 0 0 0 0 0 0 0 0 0 0 −1 0 1 . k j j = 1. 2.41) which reads explicitly 0 0 0 0 0 1 2 3 1 1 1 1 0 1 2 3 2 2 2 2 0 1 2 3 3 3 3 3 0 1 2 3 = − 0 1 2 3 0 0 0 0 0 1 1 1 1 0 2 2 2 2 0 3 3 3 3 − − − 1 2 3 − − − 1 2 3 − − − 1 2 3 . 3 (10. = − k m Km (10.50) . (10. K2 = .1: Natural Representation of the Lorentz Group Using (10. other ﬁve parameters zero) . The generators are explicitly 0 0 0 0 0 0 0 0 J1 = 0 0 0 −1 0 0 1 0 (10.45) (10. w3 ) = −w2 ϑ3 0 −ϑ1 −w3 −ϑ2 ϑ1 0 (10. J ] [ Kk . 3) Jk = (ϑk = 1.44) This result allows us now to deﬁne six generators for the Lorentz transformations(k = 1.42) This relationship implies µ 0 j µ j = 0 = j 0 . (ϑ1 . other ﬁve parameters zero) Kk = (wk = 1. 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 . J = 0 3 0 1 0 0 0 0 0 0 0 0 0 0 0 (10. ϑ3 . w2 . 2. w1 . K ] [ Jk .46) 0 0 0 0 . 2. k = 1.49) .15) one can conclude T 291 = −g g (10. ϑ2 . j. 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.

56) = 1 0 0 1 = 1 1 (10.49. To determine the explicit form of this transformation we evaluate the exponential operator by Taylor expansion. constitutes an overcomplete parametrization of the rotations (see Chapter 5). 2π[ . and using the relationship Jk = 0 0 0 Lk (10. w) = exp ϑ · J + w · K .54) which.1: Demonstrate the commutation relationships (10. 2. L(ϑ.. 3 where we have deﬁned ϑ · J = k=1 ϑk Jk and w · K = following the algebra in Chapter 5. Relativistic Quantum Mechanics The commutation relationships (10.49) deﬁne the Lie algebra associated with the Lie group L↑ . 3 (10.35) it issuﬃcient to consider in the present case the 2 × 2–matrix L = exp since w1 0 −1 −1 0 ∞ = n=0 n w1 n! 0 −1 −1 0 0 0 .53) Here R(ϑ) are the 3 × 3–rotation matrices constructed in Chapter 5. k = 1. 3 is closed. 0 1 n (10. k = 1. Following the treatment of the rotation group SO(3) one can express the elements of L↑ through + the exponential operators L(ϑ.50).e. however. Exercise 7. In analogy to equation (5. (10.52) where the 3 × 3–matrices Lk are the generators of SO(3) deﬁned in Chapter 5.57) . w = 0) = 0 0 0 R(ϑ) .51) for w = 0 correspond to rotations of the spatial coordinates. w ∈ R3 3 k=1 wk Kk . 10.51) One can readily show. + The commutation relationships imply that the algebra of the generators Jk . ϑ. (10. A boost in the x1 –direction is L = exp(w1 K1 ).55) L 0 0 0 0 exp (w1 K1 ) = exp Using the idempotence property 0 −1 −1 0 2 0 0 1 0 (10. 2.292 as can be readily veriﬁed. i. that the transformations (10. For the parameters ϑk of the Lorentz transformations holds obviously ϑk ∈ [0. We consider now the Lorentz transformations for ϑ = 0 which are referred to as ‘boosts’. Kk .

t = t − v1 x1 2 1 − v1 .59) can be written 1 2 1 − v1 √−v1 2 1 − v1 2 1 − v1 . 10. however.1) of the Lorentz transformations is obtained through the parameter change sinh w1 v1 = = tanh w1 (10. This property of the wk -values contrasts with the property of the parameters ϑk specifying rotational angles which assume only values in a compact range.6. vk deﬁned in (10.3.63) which agrees with (10. for wk wk ∈ ] − ∞.51) the explicit transformation for space–time–coordinates is then x1 = x1 − v1 t 1 − 2 v1 .58) = cosh w1 1 + sinh w1 1 0 −1 −1 0 The conventional form (10. (10.59) v1 = These two equations yield cosh w1 = 1/ and (10.64) We note that the range of wk -values is not a compact set even though the range of vk -values is compact. (10.60) sinh w1 = v1 / 2 1 − v1 . ∞[ . The range of the parameters wk can now be speciﬁed. 10. From (10. consistent with (10. This property is. 1[.62) According to (10.59) for the case k = 1 corresponds to the relative velocity of two frames of reference. Correspondingly.1). one obtains from (10. x2 = x2 . Accordingly.56.10.1: Natural Representation of the Lorentz Group one can carry out the Taylor expansion above: ∞ 293 L = n=0 2n w1 1 + 1 (2n)! ∞ n=0 2n+1 w1 (2n + 1)! 0 −1 −1 0 = cosh w1 −sinh w1 −sinh w1 cosh w1 . 10. x3 = x3 (10.59) follows. We expect that vk can only assume values less than the velocity of light c which in the present units is c = 1. . we can state vk ∈ ] − 1.61) √ √−v1 √ exp (w1 K1 ) = 2 1 − v1 1 2 1 − v1 0 0 0 0 0 0 0 0 1 0 0 1 (10.59). in fact.59) cosh w1 Using cosh2 w1 − sinh2 w1 = 1 one can identify sinhw1 = sinh2 w1 + 1. (10. (10. cosh2 w1 − 1 and coshw1 = cosh2 w1 − 1 = cosh w1 sinh w1 sinh2 w1 + 1 .

A) (10. not any physical property f ∈ R is a scalar. Nevertheless.e.68) where ρ(r. respectively. f = f.65) An example is s2 deﬁned in (10. of a system of charges.2 Scalars. Counterexamples are the energy. 2 |ˆ1 . we have not yet deﬁned representations of Lorentz transformations beyond the ‘natural’ representation acting in the 4–dimensional space of position– and time–coordinates. x2 . the z–component x3 of a particle. Scalars The quantities with the simplest transformation behaviour are so-called scalars f ∈ R which are invariant under transformations. Another example is the potential 4-vector Aµ = (V. p) . tensor operators Tkm . another example is the rest mass m of a particle. 4-Vectors The quantities with the transformation behaviour like that of the position–time vector xµ deﬁned in (10. J) (10. the square of the electric ﬁeld |E(r. 4–Vectors and Tensors In this Section we deﬁne quantities according to their behaviour under Lorentz transformations. the charge density.67) the transformation behaviour of which we will demonstrate further below. spherical harmonics Y m (ˆ). t) are the charge density and the current density. t) and J(r. (10. t) and A(r.3) are the so-called 4–vectors aµ .. A third 4-vector is the so-called current vector J µ = (ρ. Presently.294 Relativistic Quantum Mechanics 10. t) are the electrical and the vector potential of an electromagnetic ﬁeld.69) where V (r. Hence.4). i. x3 )T . However. scalars like r = x2 + x2 + x2 . a1 . We will see below how true scalars under Lorentz transformations can be constructed. Such quantities appear in the description of physical systems and statements about transformation properties are often extremely helpful and usually provide important physical insight. our deﬁnition of quantities with special properties under Lorentz transformations presently is conﬁned to the natural representation. r2 ) and. p = √ 2 1−v 1 − v2 (10. ﬁnally. namely. vectors like r = (x1 . a2 . we will encounter an impressive example of physical properties. t)|2 or the scalar product r1 · r2 of two particle positions. These quantites always come as four components (a0 . not in the so-called natural representation associated with the 3–dimensional Euclidean space E3 . The 4-vector character of J µ and of Aµ will be demonstrated further below. Some of these r ˆ quantities were actually deﬁned with respect to representations of the rotation group in function spaces. total angular momentum r 1 2 3 states of composite systems like Y m ( 1 . . E = √ m mv . (10.66) Examples of 4-vectors beside xµ are the momentum 4-vector pµ = (E. We have encountered examples in connection with rotational transformations. a3 )T and transform according to a µ = Lµ ν aν .

2: Scalars.. 4–Vectors and Tensors Scalar Product then 295 4-vectors allow one to construct scalar quantities.17) Lµ σ is the transformation inverse to Lσ µ .73) aµ = Lµ σ aσ .77) The transformed diﬀerential operator will be denoted by ∂µ = def ∂ . −a3 ) (10.74) g µν aν .e. (10. ∂x µ (10.66).17) can be written (L−1 )ν µ = Lµ ν . The respective vectors aµ are associated with the 4-vectors aµ . If aµ and bµ are 4-vectors aµ gµν bν (10. Multiplication (and summation) of x µ = Lµ ν xν by Lρ σ gρµ yields.79) This is the inverse Lorentz transformation consistent with (10. aµ (10. and . accordingly. We do not fact. One calls 4-vectors aµ covariant and 4-vectors aµ contravariant.12) a gµν b µ ν = Lµ ρ gµν Lν σ aρ bσ = aρ gρσ bσ (10.70) is a scalar. using (10.76) An important example of a covariant 4-vector is the diﬀerential operator ∂µ = ∂ = ∂xµ ∂ . We have duplicated the expression for the inverse of Lµ ν to obtain the correct notation in terms of covariant. (10.73) Note that according to (10. We start from x µ = Lµ ν xν . This property follows from (10.10.75) (10.e. Covariant 4-vectors transform like a µ = gµν Lν ρ g ρσ aσ where we deﬁned g µν = gµν . lower. gσν xν = Lρ σ gρµ x µ and g µσ gσν = δ µ ν . (10.12). i. The 4-Vector ∂µ (10. −a2 . We like to point out that from deﬁnition (10.16). the relationship being aµ = gµν aν = (a0 . i.72) of the covariant 4-vector follows = In µν and g µ as well.78) To prove the 4-vector property of ∂µ we will show that g µν ∂ν transforms like a contravariant 4vector. one can employ the tensors g µν to raise and lower indices of L ν establish here the consistency of the ensuing notation. In any case one can express (10.71) Contravariant and Covariant 4-Vectors It is convenient to deﬁne a second class of 4-vectors. −a1 . xν = g νσ Lρ σ gρµ x µ . In fact. ∂t (10.66) together with (10..72) where aν is a vector with transformation behaviour as stated in (10. g µν ∂ν = Lµ ρ g ρσ ∂σ . one can express [(L−1 )T ]µ ν = (L−1 )ν µ and.

87) The remaining components of pµ can be written. 1 − v2 1 − v 2 dt (10. In fact.85) 1 d d = √ .e.296 Relativistic Quantum Mechanics contravariant. (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 ... this operator is equal to the d’Alembert operator which is known to be Lorentz-invariant. 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. indices.− ∂t . (10. i. 2 dt dτ 1 − v (10. d’Alembert Operator We want to construct now a scalar diﬀerential operator.86) One can write (10. ∂µ does indeed transform like a covariant vector.g.81) ∂ .82) (10.66).89) .84) = 1 − (v)2 (10. 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. Proof that pµ is a 4-vector We will demonstrate now that the momentum 4-vector pµ deﬁned in (10. upper. i. For this purpose we deﬁne ﬁrst the contravariant diﬀerential operator ∂ µ = g µν ∂ν = Then the operator 2 ∂µ ∂ µ = ∂t − 2 (10. dτ 1 − v 2 dt p0 = E = √ m m dt = √ ..83) is a scalar under Lorentz transformations. 2 2 dt 1 − v 1 − v (10.88) One can express then the momentum vector pµ = √ d m dxµ = m xµ .e.79) allows us to determine the connection between ∂µ and ∂µ . p1 = √ m v1 m dx1 = √ .

of (10.h. (10.s. . can be demonstrated readily using the 4-vector notation (10.89) alltogether transforms like a contravariant 4-vector. t) = 0 . namely pµ pµ = pµ gµν pν = E 2 − p 2 Evaluation of the r. This gauge is not Lorentz-invariant and we will adopt here another gauge. is a scalar. If we can show that Aµ deﬁned in (10.h. in fact. Expansion in should then be rapidly convergent. kinetic energy ± 2m and relativistic corrections.67) E2 − p 2 = or pµ pµ = m2 which. equivalently. · A = 0.69) is.96) The Lorentz-invariance of this gauge. 8.95) 2 p This obviously describes the energy of a free particle with rest energy ±m. respectively. the r. pµ on the l. Since xµ transforms like a contravariant 4-vector.93) (10.3: Relativistic Electrodynamics 297 d The operator m dτ transforms like a scalar. i. (10.77) which allow one to express (10.. the so-called Lorentz gauge.97) and. and.s. for these potentials.e. Lorentz-invariant.96) is a scalar and. in (10. t) and A(r. a contravariant 4-vector then the l. Lorentz Gauge In our previous description of the electrodynamic ﬁeld we had introduced the scalar and vector potential V (r.10.96) in the form ∂µ Aµ = 0 .89) must be a 4-vector.s.h. in fact.90) (10. in (10. namely.12). One obtains E = ±m ± p2 2m (p 2 )2 + O 4m3 (p 2 )3 4m5 .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.s. We will demonstrate now the 4-vector property of Aµ . m2 m2 v 2 − = m2 1 − v2 1 − v2 (10. t) + · A(r.91) (10. 10. yields according to (10.94) 1 m In the non-relativistic limit the rest energy m is the dominant contribution to E. hence.69) for the electrodynamic potential and the 4-vector derivative (10. The momentum 4-vector allows us to construct a scalar quantity. (10.97) We have proven already that ∂µ is a contravariant 4-vector. We like to rewrite the last result E 2 = p 2 + m2 or E = ± p 2 + m2 .92) (10.h. ∂t V (r. hence. and had chosen the so-called Coulomb gauge (8. t).

This principle should hold in any frame of reference. t) from (8.s.h. F µν = x Ey B z 0 −Bx Ez −By Bx 0 Alternatively. Using · ∂t A(r. must transform likewise. t) 2 ∂t A(r.6) and (8. using (10.97) and.32). of (10.96). . 2. (10. according to (10. (10. We want to derive now the diﬀerential equations which determine the 4-potential Aµ in the Lorentz gauge (10. ∂µ J µ (xµ ) = 0 . F mn = − mn (10. this can be stated F k0 = −F 0k = E k . t) . t) . t) − 2 V (r.298 Transformation Properties of J µ and Aµ Relativistic Quantum Mechanics The charge density ρ(r. has to transform like a contravariant 4-vector.102) In this equation the r. also the l. t) are known to obey the continuity property ∂t ρ(r.h. · A(r.18) and.98) which reﬂects the principle of charge conservation.9). t) = 4πρ(r. using (10. transforms like a 4-vector.103) B . yields ∂µ ∂ µ Aν (xσ ) = 4 π J ν (xσ ) .99) must be a scalar. Since ∂µ ∂ µ transforms like a scalar one can conclude that Aν (xσ ) must transform like a 4-vector.100) Similarly. t) = 4 π J(r.17) using the identity (8.83) and (10. t) = −∂t V (r.82) as follows F µν = ∂ µ Aν − ∂ ν Aµ . i. n = 1. (10. it follows that J µ . using (8.104) where mn = mn is the totally anti-symmetric three-dimensional tensor deﬁned in (5.101). a 4-vector. in fact. 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 .96).. · A(r. i. t) − 2 A(r. Consequently. t) = 0 (10.105) . The respective equation for A0 = V can be obtained from Eq. one obtains for A(r. One can readily verify. that F µν can be expressed through the potential Aµ in (10. thereby. one obtains 2 ∂t V (r.69) and ∂ µ in (10.101) Combining equations (10. (8. k. in fact. The l.h. t) and current density J(r.e.s..e. (10.s. t) = ∂t · A(r. m.69). (10. prove that Aµ is. t). 10.99) Since this equation must be true in any frame of reference the right hand side must vanish in all frames.68). must be a scalar. t) + · J(r.98) can be written.100.77) and (10. t) = −∂t V (r.13). 3 (10. t) together with (10. Since ∂µ transforms like a covariant 4-vector. Equation (10.

equivalently.111) One can readily prove that this equation is equivalent to the two inhomogeneous Maxwell equations (8.2). B and E⊥ . E⊥ = B⊥ E⊥ + v × B √ 1 − v2 B⊥ − v × E = √ 1 − v2 (10.102) ∂µ F µν = 4π J ν .4).62) or.106) In case that the Lorentz transformation Lµ ν is given by (10. 10. where we used (10. t) and B(r. by (10.108) yields then the expressions for the transformed ﬁelds E and B . t).63).105) and (10.105) of the tensor F µν one can conclude the property ∂ σ F µν + ∂ µ F νσ + ∂ ν F σµ = 0 (10. B⊥ are.10.104) establishe the transformation behaviour of E(r.109) (10. From the deﬁnition (10.3: Relativistic Electrodynamics 299 The relationships (10. Noting ∂µ F µν = ∂µ ∂ µ Aν − ∂µ ∂ ν Aµ = ∂µ ∂ µ Aν − ∂ ν ∂µ Aµ = ∂µ ∂ µ Aν . 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 (10.112) (10. (10. .113) which can be shown to be equivalent to the two homogeneous Maxwell equations (8.103. the components of the ﬁelds parallel and perpendicular to the velocity v which determines the Lorentz transformation. B . 8.110) where E .97). These equations show that under Lorentz transformations electric and magnetic ﬁelds convert into one another.107) y 1 z Bz −v1 Ey √ 2 √ 2 0 −Bx 1−v1 1−v1 Ez +v1 By By +v1 Ez √ 2 − √ 2 Bx 0 1−v1 1−v1 Comparision with F µν 0 −Ex −Ey −Ez E 0 −Bz By = x Ey B z 0 −Bx Ez −By Bx 0 (10. 8.3.1. In a new frame of reference holds F µν = Lµ α Lν β F αβ (10. Maxwell Equations in Lorentz-Invariant Form One can express the Maxwell equations in terms of the tensor F µν in Lorentz-invariant form. The results can be put into the more general form E B = = E . one can conclude from (10. respectively.

demonstrated that (10. also from the equation of motion (8. For the spatial components.5).5) taking the scalar product with p dp = qp · E (10. e.s.67).114). i.120) is referred to as the Lorentz force.117) (10.5). (10. yields dpx = q dt Ex + (v × B)x (10.h. provides an alternative description of the action of the Lorentz force.67) and charge q dpµ q = pν F µν dτ m (10. hence. We have. dt E From this one can conclude.87)] and retain the deﬁnition E for the electric ﬁeld. We want to express this force through the tensor F µν .120) (10.5) where p = mv/ 1 − v 2 . The term on the r.g.116) (10. of (10. equivalent to (8. . Equation (10.119) which is the x-component of the equation of motion (8.86).86) dE q = p·E .118) p· dt √ where we exploited the fact that according to p = mv/ 1 − v 2 holds p v.103) dpx q = ( EEx + py Bz − pz By ) . employing (10. We want to √ demonstrate now that this equation is equivalent to the equation of motion (8. using (10. E = m/ 1 − v 2 .114) where d/dτ is given by (10. hence.114) reads using (10.300 Lorentz Force Relativistic Quantum Mechanics One important property of the electromagnetic ﬁeld is the Lorentz force acting on charged particles moving through the ﬁeld.93). dτ m √ Employing again (10. It holds for a particle with 4-momentum pµ as given by (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. dE q = p·E dτ m or with (10.86) and (10.104). in fact. The µ = 0 component of (10.115) This equation follows.114) is.. 1 dE 2 1 dp 2 = = qp·E 2 dt 2 dt (10.114) reads then.. for µ = 1.e. however.

(10. (5. In this section we will investigate the transformation properties of scalar functions ψ(xµ ).125)]. (10.127) . (10.51) for the natural representation of the Lorentz group. 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.42) of SO(3).48) and (10.g.1) is a generalization (rotation + boosts) of the group SO(3). ψ ∈ C∞ (4). Functions which represent 4-vectorsor other non-scalar entities.121) by (L−1 )µ ν xν and obtains ψ (xµ ) = ψ((L−1 )µ ν xν ) . and which can be constructed following the procedure adopted for the function space representation of SO(3). K3 ) are the generators of L(ϑ.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 . i. w.. We need to emphasize that (10. obey a diﬀerent transformation law.48) of SO(3).3 or the bi-spinor wave function of electron-positron pairs in Sect. J3 ) and K = (K1 . 10..124) In this expression J = (J1 .125) def (10.4 Function Space Representation of Lorentz Group In the following it will be required to decribe the transformation of wave functions under Lorentz transformations. in the present case we exclude the factor ‘−i’ [cf.e. def (10. The resulting expression should be a generalization of the function space representation (5.7. For such functions holds in the transformed frame ψ (Lµ ν xν ) = ψ(xµ ) (10.e. w) = ρ(Lµ ν (ϑ. xµ ). We will denote the intended representation by L(ϑ.. w) which correspond to the generators Jk and Kk in (10. taken at the pairs of points (x µ = Lµ ν xν . in as far as SO(3. e. 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. For this purpose one replaces xµ in (10. w) = exp ϑ · J + w · K . We like to express now ψ (x µ ) in terms of the old coordinates xµ .123) This deﬁnition corresponds closely to the function space representation (5. J2 .121) covers solely the transformation behaviour of scalar functions. i.10.4: Function Space Representation of Lorentz Group 301 10.122) This result gives rise to the deﬁnition of the function space representation ρ(Lµ ν ) of the Lorentz group (ρ(Lµ ν )ψ)(xµ ) = ψ((L−1 )µ ν xν ) . However.47). the charge-current density in case of Sect. w)) = ρ eϑ·J + w·K which we present in the form L(ϑ. 10. we seek an expression which corresponds to (10. K2 . Accordingly.126) ρ ewk Kk − 1 1 .

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

135) (10.128).141) (10. One-Dimensional Function Space Representation The exponential operator (10. x2 + x2 + x2 1 2 3 in the case of SO(3) and (x0 )2 − (x3 )2 in case of transformation (10. ∂Ω . a relationship which can also be stated R = and + − (x0 )2 − (x3 )2 (x0 )2 − (x3 )2 if x0 ≥ 0 if x0 < 0 (10. x3 as follows x0 = R coshΩ . x0 ∂2 + x2 ∂0 ] = [x3 ∂2 .87)]. [ J1 . Ω which are connected with x0 . x2 ∂0 ] − [x2 ∂3 . K2 ] = [x0 ∂1 + x1 ∂0 . 10.136) (10.85-5. x0 ∂2 ] = −x2 ∂1 + x1 ∂2 = − J3 . x0 ∂2 + x2 ∂0 ] = [x0 ∂1 .10.. For this purpose one expresses K3 in terms of hyperbolic coordinates R. K2 ] = [x3 ∂2 − x2 ∂3 .139) x3 = R sinhΩ (10.140) allow one to ∂ ∂ express the derivatives ∂0 .4: Function Space Representation of Lorentz Group [ K1 . x0 ∂2 ] = x3 ∂0 + x0 ∂3 = K3 . 0 ∂cothΩ ∂x x = ∂R x0 = − ∂x3 R (10. ∂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 .139. (5. i. Eqs.137).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. We note tanhΩ = ∂R x0 = .143) .142) (10. 3 ∂R 3 ∂Ω ∂x ∂x R ∂R x ∂Ω = ∂Ω ∂tanhΩ 1 = cosh2 Ω 0 3 ∂tanhΩ ∂x x ∂Ω ∂cothΩ 1 = = − sinh2 Ω 3 . (10.138) x3 x0 . can be simpliﬁed considerably. x2 ∂0 ] − [x1 ∂0 .125) in the case of a one-dimensional transformation of the type L(w3 ) = exp w3 K3 . ∂3 in terms of ∂R .137) where K3 is given in (10. In both cases the radial coordinate is the quantity conserved under the transformations. cothΩ = 3 . 303 (10.e. In the following we consider solely the case x0 ≥ 0. The relationships (10.

extremely useful. namely the Klein–Gordon and the Dirac equations.128) yields K3 = x0 ∂3 + x3 ∂0 = ∂ . which allows one to replace classical observables by quantum mechanical operators acting on wave functions. (10.146) Applying the correspondence principle to (10. Further below we will provide a more systematic. ˆ (10.151) reads ∂µ ∂ µ + m2 ψ(xµ ) = 0 (10. − ) = i ∂ µ . In the position representation the correspondence principle states ˆ E =⇒ E = − ∂t i ˆ p =⇒ p = i which. (length)−1 . however.92) one obtains the wave equation − or 2 2 µ ∂ ∂µ ψ(xν ) = m2 ψ(xν ) ψ(xν ) = 0 . ) = i ∂µ .121. Free Particle Case A quantum mechanical description of a relativistic free particle results from applying the correspondence principle.122).149) where ψ(xµ ) is a scalar. complex-valued function. The latter property implies that upon change of reference frame ψ(xµ ) transforms according to (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 ) . mass.150) . representation theoretic treatment. (10.151) is called the Klein-Gordon equation. momentum. The partial diﬀerential equation (10. The historic route to these two equations. and (time)−1 all have the same dimension. We will provide a ‘derivation’ of these two equations which stem from the historical development of relativistic quantum mechanics. therefore. In natural units the Klein–Gordon equation (10. ˆ pµ =⇒ pµ = i(∂t . In these units the quantities energy.148) ∂ µ ∂µ + m2 (10.304 Inserting these results into the deﬁnition of K3 in (10. 10. in 4-vector notation reads pµ =⇒ pµ = i (∂t . is not very insightful.145) ∂Ω 10. In the following we will employ so-called natural units = c = 1.5 Klein–Gordon Equation In the following Sections we will provide a heuristic derivation of the two most widely used quantum mechanical descriptions in the relativistic regime. ∂Ω Relativistic Quantum Mechanics (10. but certainly is short and.147) (10.144) The action of the exponential operator (10.

t) − ψ(r. t) ] 2mi (10.151) One can notice immeadiately that (10. t) − ψ(r.154). t) jKG (r.159) (10.161) .5: Klein–Gordon Equation or 2 ∂t − 2 305 + m2 ψ(xµ ) = 0 . t) + where 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. 2mi (10. t) = · jKG (r. In order to obtain the conservation law connected with the Klein–Gordon equation (10. t) = 0 (10. t) ) 2m 1 = ( ψ ∗ (r. t) = ψ ∗ (r.157) 2m which is equivalent to (10.160) ∂t (ψ ∗ ∂t ψ which corresponds to − ψ∂t ψ∗) + · (ψ ψ∗ ∂t ρKG (r. t) − ψ(r. t) ψ ∗ (r. t) ψ(r.10. t) ψ ∗ (r. t) 1 [ ψ ∗ (r. t) = (10.150) is invariant under Lorentz transformations. t) = − 1 2m 2 ψ(r. and that (as postulated) ψ(xµ ) is a scalar. Under Lorentz transformations the free particle Klein–Gordon equation (10. t) + where ρS (r. (10. t)∂t ψ ∗ (r. t) ) . t) = 0 i ( ψ ∗ (r. To 1 derive this conservation law one rewrites the Schr¨dinger equation in the form (i∂t − 2m 2 )ψ = 0 o and considers 1 2 Im ψ ∗ i∂t − ψ = 0 (10. t) ψ(r. t) ψ(r. This follows from the fact that ∂µ ∂ µ and m2 are scalars.156) · jS (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 (10.152) which has the same form as the Klein–Gordon equation in the original frame.155) describes the positive deﬁnite probability to detect a particle at position r at time t and where jS (r.153) is associated with a conservation law for particle probability ∂t ρS (r.150) becomes ∂µ ∂ µ + m2 ψ (x µ ) = 0 (10. t)∂t ψ(r.154) describes the current density connected with motion of the particle probability distribution. t) (10.

i.e. Today. λ) = λ ψo (p. λ|xµ ) (10. not of particle probability. with p = 0.166) is Eo (p) ∗ ρKG (p.167) m which is positive for λ = + and negative for λ = −. λ|xµ ) according to (10. We denote this by summarizing the solutions as follows ∂µ ∂ µ + m2 ψo (p. λ = ± (10. is actually that of a charge density. o namely. (10. and of negative energies E ≤ −m. λ|xµ ) = Nλ. t). (10. λ) associated with the corresponding wave functions ψo (p.306 Relativistic Quantum Mechanics This conservation law diﬀers in one important aspect from that of the Schr¨dinger equation (10.162) Inserting this into the Klein–Gordon equation (10. the anti-particles carry a charge opposite to that of the associated particles.166) ψo (p. 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) .168) . it had been realized later.162) shows that the solutions of the free particle Klein-Gordon equation (10. corresponding to λ = +. p0 . λ) actually describes charge density rather than probability. The density ρKG (p.150) are actually determined by po and by the choice of sign ±. t) = 0 (10. m 2 2 E0 = m2 + po .93] dispersion relationship connecting E0 .154).150) is a continuum of positive energies E ≥ m. and the density ρKG (p.. The proper interpretation of the two cases is that the Klein–Gordon equation describes particles as well as anti-particles.161) and (10. corresponding to λ = −. ±) = ± 2 m2 + po (10. Generating a Solution Through Lorentz Transformation A particle at rest. 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 ) = N e−iλmt . according to (??) is decribed by the r–independent wave function ψo (p = 0. in that the expression for ρKG is not positive deﬁnite.164) The corresponding energy is Eo (po .p ei(p·r − λEo (p)t) Eo (p) = The spectrum of the Klein–Gordon equation (10.165) This result together with (10. The proper interpretation of ρKG (r. λ|xµ )ψo (p.151) yields 2 2 Eo − p0 − m2 ψ(r. (10. for example pions. λ|xµ ) = 0 2 m2 + po .163) which results in the expected [see (10. λ = ± . the Klein-Gordon equation is considered as a suitable equation to describe spin–0 particles.

λ|xµ ) = = exp w3 ∂ ∂Ω N e−iλmRcoshΩ 3) N e−iλmRcosh(Ω+w .f.168). using (10. of (10.177) . In case of λ = + one obtains ﬁnally L(w3 )ψo (p = 0.175) as the momentum of the particle and one obtains L(w3 )ψo (p = 0.173) (10.170) The coordinate transformation (10.166)] of the particle. t)) . i.166). for the wave function (10. t)) (10. for positive energy solutions.s..e. This yields. λ = −|xµ ) = N ei(px 3 + Eo (p)x0 .171) One can interpret then for λ = +. 2 2 1−v 1−v (10. In case of λ = −..176) 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) (10.61) yield for this expression −iλ √ m mv x0 − iλ √ x3 . one has to interprete p = mv/ 1 − v 2 (10. −A(r. t).138) and the relationships (10.145) choosing m = sinhw3 .e.169) The addition theorem of hyperbolic functions cosh(Ω + w3 ) = coshΩ coshw3 + sinhΩ sinhw3 allows us to rewrite the exponent on the r. i. λ = +|xµ ) = N ei(px 3 − Eo (p)x0 (10.169) −iλ ( m coshw3 ) ( R coshΩ ) − iλ ( m sinhw3 ) ( R sinhΩ ) .138) to replace t = x0 by hyperbolic coordinates R. (10.h.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.6: Klein–Gordon Equation with Electromagnetic Field 307 We want to demonstrate now that the wave functions for p = 0 in (??) can be obtained through appropriate Lorentz transformation of (10. For this purpose we consider the wave function for a particle moving with momentum velocity v in the direction of the x3 –axis. for negative energy solutions. A(r. (10.174) which agrees with the expression given in (10. t). (10. Ω. L(w3 )ψo (p = 0.168). (10. in fact.10.172) as the energy [c. Aµ (xµ ) = (V (r. Such wave function should be generated by applying the Lorentz transformation in the function space representation p (10.

1: Coupling of a particle of charge q to an electromagnetic ﬁeld described by the 4-vector potential Aµ = (V.308 free classical particle classical particle in ﬁeld (V. −A). 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. 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. A) or Aµ = (V.178) According to the replacements in Table 10. (10. See also Eq. which couples a particle of charge q to an electromagnetic ﬁeld described through the potential Aµ (xν )..1. we deﬁne ψ(r.e.179) (10.69)] one replaces the quantum mechanical operators. that the mass m of the particle. According to the so-called minimum coupling principle the presence of the ﬁeld is accounted for by altering energy.93). according to the rules shown in Table 10.150) −g µν (i∂µ )(i∂ν ) + m2 ψ(xµ ) = 0 .147). (10. t))2 ψ(r. m2 ) ψ(xµ ) = 0 . and focus on the remaining part of the wave function. t). i∂t and −i in (10. in .e. For this purpose one writes the Klein-Gordon equation (10. (10.182) and seek an equation for Ψ(r. A) free quantum particle quantum particle in ﬁeld (V. t) . t) = −i − q A(r. i.150). is much larger than all other energy terms.180) + m2 ψ(r. In terms of space-time derivatives this reads (i∂t − qV (r. albeit in a form.e. t) 2 (10.. (10. According to the principle of minimal coupling [see (10. For this purpose we split-oﬀ a factor exp(−imt) which describes the oscillations of the wave function due to the rest energy.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 − . We will also assume. momenta for classical particles and the respective operators for quantum mechanical particles in the manner shown.. i. t) = e−imt Ψ(r. t) . it’s rest energy.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. in keeping withnthe non-relativistic limit.

. Ψ |q V | << m ..e.s. t) = − Ze and A(r. ﬁltered out of the beam and ﬁnally fall as slow pions onto elements for which a pionic variant is to be studied. The process of π − meson capture involves the so-called Auger eﬀect. t) ≡ 0 r + Ze2 r 2 + 2 − m2 π ϕ(r ) = 0 . The Klein-Gordon equation (10.e. atoms in which one or more electrons are replaced by π − mesons. spin-0 mesons.185) This is. larger than |i∂t Ψ/Ψ| and alrger than qV . e.6: Klein–Gordon Equation with Electromagnetic Field particular.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. 309 (10. (10. i. of (10. (10. t) = ˆ [p − q A(r.184) where we neglected all terms which did not contain factors m.183). (10.183) The term on the l. A stationary state of the Klein–Gordon equation is described by a wave function ψ(xµ ) = ϕ(r ) e−i t .181) reads then i ∂t Ψ(r. π − mesons are generated through inelastic proton–proton scattering p + p −→ p + p + π − + π + . t)]2 + qV (r.2) of a non-relativistic spin-0 particle o moving in an electromagnetic ﬁeld.g. (10.. t) 2m Ψ(r.188) Inserting this into (10. i.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. Pionic Atoms To apply the Klein–Gordon equation (10. The approximation is justiﬁed on the ground of the inequalities (10. identical to the Schr¨dinger equation (10.h. | i∂t Ψ | << m .10. t) (10. To ‘manufacture’ pionic atoms. however. 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− .189) . This application demonstrates that the Klein–Gordon equation describes spin zero particles.181) yields (we assume now that the Klein–Gordon equation describes a 2 particle with mass mπ and charge −e) for qV (r.186) then are slowed down.181) to a physical system we consider pionic atoms.

199) ( + 1) − Z 2 e4 .. φ) .195) The latter problem leads to the well-known spectrum En = − mπ (Ze2 )2 . 1.196) In this expression the number n deﬁned through n = n− −1 (10. φ.. . r r r r sin θ r sin θ Ze2 2 r ) (10. (10. .193) (10. φ) R (r) Y r m (θ.191) ˆ The operator L2 in this equation suggests to choose a solution of the type ϕ(r ) = where the functions Y in (10.192) leads then to the ordinary diﬀerential equation d2 − dr2 ( + 1) − Z 2 e4 2 Ze2 + + r2 r 2 − m2 π R (r) = 0 . . n − 1 .194) and (10. . 2n2 = 0. The similarity of (10. . (10. the eigenfunctions of the operator L2 ˆ L2 Y m (θ. this quantity deﬁnitely must be an integer.198) . 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 .e.191) m (θ. n = 1.e.310 Relativistic Quantum Mechanics Because of the radial symmetry of the Coulomb potential we express this equation in terms of spherical coordinates r. (10. (10. i.192) ˆ are spherical harmonics. (10. φ) = ( + 1) Y m (θ. . 2. . φ) (10.194) Bound state solutions can be obtained readily noticing that this equation is essentially identical to 2 that posed by the Coulomb problem (potential − Ze ) for the Schr¨dinger equation o r d2 − dr2 ( + 1) 2mπ Ze2 + + 2mπ E r2 r R (r) = 0 (10.195) 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. i.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 .197) counts the number of nodes of the wave function.

204) reads 1 (10. . Introducing α = Z 2 e4 and 1 β = + 2 (10.203) spectrum .194) d2 λ( ) ( λ( ) + 1 ) 2 Ze2 − + + dr2 r2 r 2 311 − m2 π R (r) = 0 . . 2 e2 −→ e2 mπ . 1. . 10.204) 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 ). 1. − + 1. . .202) Solving this for state) yields (choosing the root which renders = 1 + mπ Z 2 e4 ( n + λ( )+1 )2 ≤ mπ . 2mπ 2 −→ λ( ) .197. . (10.e.201) mπ Z 2 e4 m2 − m2 π π = − . ..196) if one makes the replacement 2 values which can be obtained E −→ One obtains − m2 π . . + (10. .205) α √ 2 2 1+ (n − β + β −α) and one obtains the series of approximations 1 1+ ≈ ≈ ≈ 1+ ≈ 1+ (n − β + α √ β 2 −α)2 1 1+ (n − α 2β α + O(α2 ))2 1 1+ n2 − α n + O(α2 ) β α 1 α n2 + α2 βn3 + O(α3 ) − α2 8n4 1 α 2n2 + α2 2βn3 + O(α3 ) . . = 0.200) The bound state solutions of this equation should correspond to from (10. .10. n − 1 m = − . . . m) = mπ 1 + Z 2 e4 (n− − 1 2 + ( + 1 )2 − Z 2 e4 )2 2 n = 1. . (10. .199) and deﬁning EKG (n. EKG = Using (10. . n = 0. . results in the (10. 1. . 2mπ 2 (n + λ( ) + 1)2 (10. . which corresponds to a bound = 0. i. 2.6: Klein–Gordon Equation with Electromagnetic Field and one can write (10. .

the π − meson is a pseudoscalar particle. It should feature then a diﬀerential operator of the type D = iγ µ ∂µ . Also.. i.208. His solution . is related to the equation E 2 = m2 + p 2 of the classical theory which involves a term E 2 . but not an operator iγ · as suggested by the principle of relativity (equivalence of space and time). through the correspondence principle. Dirac succeeded. the Dirac equation introduced below.207) Here the ﬁrst term represents the rest energy. the Klein–Gordon equation had been rejected since it did not yield a positive-deﬁnite probability density. (10.e.208). The latter term agrees with observations of pionic atoms.146) leads to the wave equation i∂t Ψ(r.7 The Dirac Equation Historically. and the third term gives the leading relativistic correction. for example. i. However. one with a positive-deﬁnite density. in fact. however. t) = ± m2 − 2 Ψ(r. however. are unsatisfactory since expansion of the square root operator involves all powers of the Laplace operator. 10.209). a splitting of the six n = 2. would have only a ﬁrst order time derivative.209). 2n2 2βn3 8n4 4n4 (10. 10.209) which. the wave function changes sign under reﬂection.206) 1 2 3 + O(Z 6 e12 ) . 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. A more satisfactory Lorentz–invariant wave equation. Application of the correspondence principle (10. It must be pointed out here that does not denote energy. it does not agree with observations of the hydrogen spectrum. The latter spectrum shows.204) EKG (n.e. Many attempts were made by theoretical physicists to ‘linearize’ the square root operator in (10. 10. Equations (10.2 particles. t) . 4n (10.208) These two equation can be combined i∂t + m2 − 2 i∂t − m2 − 2 Ψ(r. In order to describe electron spectra one must employ the Lorentz-invariant 1 wave equation for spin. in turn.. a feature which is connected with the 2nd order time derivative in this equation. t) (10. m) ≈ m − mZ 2 e4 mZ 4 e8 − 2n2 2n3 1 + − 1− Relativistic Quantum Mechanics α α2 α2 α2 − + + + O(α3 ) .312 ≈ From this results for (10. . Heuristic Derivation Starting from the Klein-Gordon Equation An obvious starting point for a Lorentz-invariant wave equation with only a ﬁrst order time derivative is E = ± m2 + p 2 . but in the present case rather the negative of the energy. arises because the Klein–Gordon equation. is identical to the two equations (10.208.. i. = 1 states into groups of two and four degenerate states. This derivative. Finally.e. but for a long time to no avail. the second term the non-relativistic energy.

γ j . It turns out that no 4-vector of real or complex coeﬃcients can satisfy these conditions.215) . ψ2 (xµ ). 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 . (10. . therefore. j = 1. For µ = ν condition (10. one route to important discoveries in physics is the creation of new mathematical descriptions of nature.212) where we have changed ‘dummy’ summation indices.209).211) Obviously.1 2 and anti-particles.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. 3 has imaginary eigenvalues ±i. 2. the quantities γ 0 . However. Accordingly. beautiful mathematical theory and that. are associated with a positive-deﬁnite particle density. γ 1 . i.213) reads (γ µ )2 = 1 µ = 0 −1 µ = 1. Instead. ψd (xµ ). quantities with the transformation law (10.7: Dirac Equation 313 to the problem involved an ingenious step. 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 ). strengthens the believe of many theoretical physicists today that the properties of physical matter ultimately derive from a. these descriptions ultimately merging with the true theory of matter. Properties of the Dirac Matrices Let us now trace Dirac’s steps in achieving the linearization of the ‘square root operator’ in (10. Starting point is to boldly factorize.e. this would lead to the two wave equations (P − m)Ψ = 0 and (P + m)Ψ = 0 which have a ﬁrst order time derivative and. it was assumed that the 4-dimensional space introduced by Dirac could be linked to 4vectors. We seek to identify the coeﬃcients γ µ . this was not so. The discovery by Dirac.208). . the realization that the linearization can be carried out only if one assumes a 4-dimensional representation of the coeﬃcients γ µ . therefore. γ 2 . γ ν ]+ = 2 g µν (10. j = 1.213) We want to determine now the simplest algebraic realization of γ µ . the operator of the Klein–Gordon equation ∂µ ∂ µ + m2 = −( P + m ) ( P − m ) (10. (10. In fact. unspeciﬁed algebraic objects γ µ and γ ν . yet to be discovered.211) into (10. so far.. 2. 2. the 4-dimensionsional representation discovered by Dirac involved new physical properties. Initially. according to (10. Inserting (10.66). . but did not commute the. exploited ∂µ ∂ν = ∂ν ∂µ .214) From this follows that γ 0 has real eigenvalues ±1 and γ j .10.210) where P = iγ µ ∂µ . achieved through a beautiful mathematical theory. (10. one can impose the condition γ0 is hermitian . 3 . 3 are anti-hermitian . spin. namely.

220) where Ψ(xµ ) represents a 4-dimensional wave function.217) Obviously.213) reads γ µ γ ν = −γ ν γ µ . except for similarity trasnformations. holds 2 σj = 1 . namely the Pauli matrices σ 1 . (10. rather than to the right side. The Adjoint Dirac Equation The adjoint equation is Ψ† (xµ ) i(γ µ )† ∂µ + m ← ← (10.219) Using property (10. in fact. the γ µ are anti-commuting. (10.215). ψ4 ) and where ∂µ denotes the diﬀerential operator ∂ µ operating to the left side. Relativistic Quantum Mechanics (10. σ 3 for which. γj = 0 −σ j σj 0 . 3 which. (10. is implied by (10.223) Inserting this into (10. From this one can conclude for the determinants of γ µ det(γ µ γ ν ) = det(−γ ν γ µ ) = (−1)d det(γ ν γ µ ) = (−1)d det(γ µ γ ν ) .224) . 2. in fact.221) =0 (10. σ 2 . (10. For d = 2 there exist only three anti-commuting matrices.218) of the Pauli matrices one can readily prove that condition (10. This property can also be written (γ µ )† = γ 0 γ µ γ 0 .216) i.222) and multiplication from the right by γ 0 yields the adjoint Dirac equation Ψ† (xµ ) γ 0 iγ µ ∂µ + m ← =0. One can readily show using the hermitian property of the Pauli matrices (γ 0 )† = γ 0 and (γ j )† = −γ j for j = 1. to construct four matrices γ µ for the next possible dimension d = 4. σ j σ k = − σ k σ j for j = k ..218) The Pauli matrices allow one. is unique.222) ∗ ∗ ∗ ∗ where we have deﬁned Ψ† = (ψ1 .e. 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. We will argue further below that the choice f γ µ . 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. ψ2 . however.213) is satisﬁed. ψ3 . 1 (10. rather than a scalar wave function.314 For µ = ν (10. as long as det(γ µ ) = 0 the dimension d of the square matrices γ µ must be even so that (−1)d = 1. A proper choice is γ0 = 1 1 0 0 −1 1 .

229) The similarity transformation (10.233) The eigenstates and eigenvalues of H correspond to the stationary states and energies of the particles described by the Dirac equation. 3 and ˆ β = 1 1 0 0 −1 1 . αj = ˆ 0 σj σj 0 . t) = 0 (10. (10.219) is the socalled chiral representation deﬁned through ˜ Ψ(xµ ) = S Ψ(xµ ) . γj = ˜ 0 σj 1 S = √ 2 −σ j 0 1 1 1 1 1 −1 1 1 (10.7. i. j = 1.230) Schr¨dinger Form of the Dirac Equation o Another form in which the Dirac equation is used often results from multiplying (10. 2.227) A representation often adopted beside the one given by (10. 2. γ ˜ Exercise 10. (10. (10.e. 3 .10. .Chiral Representation 315 The Dirac equation can be subject to any similarity transformation deﬁned through a non-singular ˜ 4 × 4–matrix S. (10.231) ˆ whereα has the three components αj .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 ) . [˜ µ . 2. 10.227) leaves the algebra of the Dirac matrices unaﬀected and commutation property (10.227).225) .213). 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. j = 1. γ ν ]+ = 2 g µν . ˆ ˆ ˆ Ho = α · p + β m .222.226) (10. 3 ..1: Derive (refeq:Dirac-intro20a) from (10.213) still holds.228) (10.7: Dirac Equation Similarity Transformations of the Dirac Equation . (10. j = 1. and γ0 = ˜ 0 1 1 1 0 1 .221) from the left by γ 0 ˆ ˆ i ∂t + i α · − β m Ψ(r.

235) as can be readily veriﬁed from (10.235). . in case of C4 .e. 3 . d4 = 1 0 0 −1 1 0 0 −1 0 1 1 0 0 i −i 0 (10. Γ±11 = ±d3 d4 Γ±12 = ±d1 d2 d3 . Obviously. to Dirac matrices. Rev.Good.213). satisfy the anti-commutation property dj dk + dk dj = 2 for j = k 0 for j = k Relativistic Quantum Mechanics d4 = γ 0 (10. a representation of the dj ’s is d1 = d3 = 0 1 1 0 0 i −i 0 1 0 0 1 1 0 0 1 . d2 . d4 is called a Cliﬀord algebra C4 . Γ±15 = ±d2 d3 d4 Γ±16 = ±d1 d2 d3 d4 (10. Γ±7 = ±d1 d3 . d4 . This property extends then to 4 × 4–matrices which obey (10. d3 .237) which are the ordered products of the operators ±1 and d1 . i. Γ±3 = ±d2 . To complete the proof in this section the reader may consult Miller ‘Symmetry Groups and their Application’ Chapter 9. 27.Phys.H.6 and R. The associative algebra generated by d1 . the matrix elements of C = A ⊗ B are Cjk. The three Pauli matrices also obey the property (10. page 187.. Γ±9 = ±d2 d3 Γ±10 = ±d2 d4 . form a Cliﬀord algebra C3 .237) by means of the property (10. Γ±4 = ±d3 . The Cliﬀord algebra C4 entails a subgroup G4 of elements ± dj1 dj2 · · · djs . The representations of this group are given by a set of 32 4 × 4–matrices which are equivalent with respect to similarity transformations. The reader may .213).235) and. (1955). Since the Γj are hermitian the similarity transformations are actually given in terms of unitary transformations. For example. . d2 = . j = 1.236) where ‘⊗’ denotes the Kronecker product between matrices.Mod.. Γ±5 = ±d4 Γ±6 = ±d1 d2 .e. m = Aj Bkm .316 Cliﬀord Algebra and Dirac Matrices The matrices deﬁned through dj = iγ j . j1 < j2 < · · · < js s ≤ 4 (10. hence. Γ±13 = ±d1 d2 d4 . 2. any product 1 of the dj ’s can be brought to the form (10. i. The representations of Cliﬀord algebras Cm are well established.234) (10. There are (including the diﬀerent signs) 32 elements in G4 which we deﬁne as follows Γ±1 = ±1 1 Γ±2 = ±d1 .238) These elements form a group since obviously any product of two Γr ’s can be expressed in terms of a third Γr . Γ±8 = ±d1 d4 .235) can diﬀer only with respect to unitary similarity transformations. Γ±14 = ±d1 d3 d4 . One can conclude then that also any set of 4 × 4–matrices obeying (10.

transform coordinates and derivatives of the Dirac equation.218) of the Dirac matrices γ µ .e. x = Lµ ν xν . Except for a similarity transformation. i. the form of the Dirac equation is identical in equivalent frames of reference.. Exercise 7. Hence.. it must hold γ µ = γ µ .215). a result one could have also obtained by applying the chain rule to (10. in fact. 2.241) .10.80) by Lµ ρ and using (10. i.. µ µ (10. We can.239) Form invariance implies that the matrices γ µ should have the same properties as γ µ . therefore. Lorentz Transformation of the Bispinor State Actually.e. namely. The latter transformations imply that coordinates transform according to (10. j = 1. Exercise 7. and derivatives according to (10. exist. i. i. (10. we do not know yet how to transform the 4-dimensional wave function Ψ and the Dirac matrices γ µ . in a moving frame should hold iγ ∂µ − m Ψ (x ) = 0 .219).e.3: Demonstrate the anti-commutation relationships (10.. 10.2: Demonstrate the anti-commutation relationships (10.e.240) . these properties determine the matrices γ µ uniquely. 3 has imaginary eigenvalues ±i and can be represented by an anti-hermitian matrix.4: Show that from (10.8: Lorentz Invariance of the Dirac Equation 317 also want to establish the unitary transformation which connects the Dirac matrices in the form (10. we 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.213.218) of the Pauli matrices σ j .236) with the Dirac representation (10. Exercise 7. i.. in a moving frame holds iγ µ ∂µ − m Ψ (x ) = 0 . x µ = Lµ ν xν (10.214) follows that γ 0 has real eigenvalues ±1 and can be represented by a hermitian matrix. Multiplication and summation of (10. 10. Inﬁnitesimal Bispinor State Transformation We want to show now that a suitable transformation of Ψ(xµ ) does.76) yields ∂ρ = Lν ρ ∂ ν .6). However.80). in frames connected by Lorentz transformations.e.6). The transformation is assumed to be linear and of the form Ψ (x ) = S(Lµ ν ) Ψ(xµ ) µ µ (10.8 Lorentz Invariance of the Dirac Equation We want to show now that the Dirac equation is invariant under Lorentz transformations. and γ j .

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. . (10. Ob1 viously. using gρµ γ µ = γρ . (10.249) from which we can conclude σµν = −σµν νµ = −σνµ i. Multiplication of this property by g from the right yields ( g)T = − g. hence. For this purpose we exploit (10. of the equation.241) implies a similarity transformation Sγ µ S −1 .243) in a form in which the Lorentz transformation in the form Lµν is on the r. however. µ (10.247) can be expanded S( µν i ) = 1 − σµν 1 4 µν (10.243) We want to determine now the 4 × 4–matrix S(Lη ξ ) which satisﬁes this condition.243) from the left by Lσ ρ gσν yields S(Lη ξ )γ µ S −1 (Lη ξ ) gρµ = Lσ ρ gσν γσ . We had introduced in (10. hence. νµ (10.12) in the form Lν µ gνσ Lσ ρ = gµρ = gρµ . the transformation (10. The inﬁnitesimal transformation S(Lρ σ ) which corresponds to (10. in the expression of the inﬁnitesimal transformation ρ Lµν = g µν + µν (10. follows S(Lη ξ )γρ S −1 (Lη ξ ) = Lσ ρ γσ .247) the inﬁnitesimal matrix µν is anti-symmetric.221) upon Lorentz transformation yields iS(Lη ξ )γ µ S −1 (Lη ξ ) Lν µ ∂ν − m Ψ (x ) = 0 . state that the Dirac equation (10.e.248) Here σµν denote 4 × 4–matrices operating in the 4-dimensional space of the wave functions Ψ. Multiplication of (10..246) (10. One can ﬁnally conclude multiplying both sides by g ρµ S(Lη ξ )γ µ S −1 (Lη ξ ) = Lν µ γν .. 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. g is an anti-symmetric matrix.242) The form invariance of the Dirac equation under this transformation implies then the condition S(Lη ξ )γ µ S −1 (Lη ξ ) Lν µ = γ ν . One can.245) The construction of S(Lη ξ ) will proceed using the avenue of inﬁnitesimal transformations. i.250) σµν = −σνµ . µ g ρν = µν and.s. it must hold (10.244) from which. The elements of g are. S( µν ) should not change its value if one replaces in its argument µν by − νµ .e.318 Relativistic Quantum Mechanics where S(Lµ ν ) is a non-singular 4 × 4–matrix. The proper starting point for a constructiuon of S(Lη ξ ) is actually (10.h. (10. It holds then S( µν i ) = 1 − σµν 1 4 µν µν = S(− νµ i ) = 1 + σµν 1 4 µν .

213. For this purpose one needs to express formally the r. γj = ˜ 0 −σ j σj 0 . (10.255) satisfy condition (10.h. the chiral representation introduced above in Eqs.e.248. (10.229). γj ] = γ ˜ 2 −iσ j 0 0 iσ j . must be symmetric with respect to interchange of the indices α and β. (10. In fact.246) results then in a condition for the generators σµν − i ( σαβ γ µ − γ µ σαβ ) 4 αβ = νµ γν . like the expression on the l. In this representation the Dirac matrices γµ = (˜ 0 .251) S −1 ( µν ) = 1 + σµν µν 1 4 Inserting (10.8: 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. β [ σαβ .252) Since six of the coeﬃcients αβ can be chosen independently.s. of (10. 2.253) 1 2 αβ ( δ µ β γα − δ µ α γβ ) . Algebra of Generators of Bispinor Transformation We want to construct the bispinor Lorentz transformation by exponentiating the generators σµν .252) also as a sum over both indices of αβ .h. σjk = [˜j .228.5: Show that the σαβ deﬁned through (10. j = 1. i. (10.s.254) is σαβ = i [ γα . For this purpose we express νµ γν = 1 2 αµ γα + 1 2 βµ γβ = = 1 2 αβ δ µ β γα + 1 2 αβ δ µ α γβ (10.h. 3 . namely. (10. Comparing this with the l. 10.s.254). γk ] = ˜ γ ˜ jk σ 0 0 σ . β and operate in the same space as the Dirac matrices. 10.255) which can be demonstrated using the properties (10. the expression on the r.s. ˜ ˜ σ only six generators need to be determined) σ0j = ˜ i [˜0 .257) . Furthermore. this condition can actually be expressed through six independent conditions. γ µ ]− = 2i ( δ µ β γα − δ µ α γβ ) . −γ ) are ˜ γ ˜ γ0 = ˜ 0 1 1 1 0 1 .254) The proper σαβ must be anti-symmetric in the indices α. Exercise 7.251) into (10.216) of the Dirac matrices. of (10. For this purpose we inspect the properties of the generators in a particular representation.10.256) One can readily verify that the non-vanishing generators σµν are given by (note σµν = −˜νµ ... a solution of condition (10.h. 10. γβ ]− 2 (10. For this purpose we need to verify that the algebra of the generators involving addition and multiplication is closed.252) results in the condition for each α.

51). Finite Bispinor Transformation The closedness of the algebra of the generators σµν deﬁned through (10. one can express the tensor µν through w and ϑ using µν = µ ρ g ρν and the expression (10. .259) yields the desired connection between the Lorentz transformation (10. We can ﬁnally note that the closedness of the algebra of the generators σµν is not aﬀected by similarity transformations and that. and since the algebra of the Pauli matrices is closed. Since such operator does not change its block-diagonal form upon exponentiation the bispinor transformation (10.320 Relativistic Quantum Mechanics Obviously..e. not necessarily inﬁnitesimal. µν in the exponential form i S = exp − σµν 4 µν .241) bispinor wave functions from one frame of reference into another frame of reference. the representation (10.259) becomes in the chiral representation ˜ S(w. In order to construct an explicit expression of S in terms of w and ϑ we employ again the chiral representation.261) We note that this operator is block-diagonal. ϑ) = e 2 (w − iϑ)·σ 0 − 1 (w + iϑ)·σ 0 e 2 1 (10. 2 0 −(w + iϑ) · σ 13 + σ23 ˜ 23 ) (10. therefore. any representation of the generators. 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 = .258) again into block-diagonal operators.255) yields a closed algebra.51) and S.248) allows us to write the transformation S for any. since both operations convert block-diagonal operators A 0 0 B (10. (10. In fact. One should.260) Inserting this into (10. i. in particular.262) This expression allows one to transform according to (10. be able to state S in terms of the same parameters w and ϑ as the Lorentz transformation in (10.44) 0 −w1 −w2 −w3 w 0 ϑ3 −ϑ2 = 1 w2 −ϑ3 0 ϑ1 w3 ϑ2 −ϑ1 0 µν (10.259) We had stated before that the transformation S is actually determined through the Lorentz transformation Lµ ν . the algebra of these generators is closed under addition and multiplication. therefore.

268) holds for the bispinor Lorentz transformation. for j µ (xµ ) = (ρ.270) We conclude that (10.263) Starting point are the Dirac equation in the form (10.s. the conservation law (10. We will now determine the relationship between the ﬂux j µ = Ψ (x )γ 0 γ µ Ψ (x ) † µ (10. (10.8: Lorentz Invariance of the Dirac Equation Current 4-Vector Associated with Dirac Equation 321 We like to derive now an expression for the current 4-vector j µ associated with the Dirac equation which satisﬁes the conservation law ∂µ j µ = 0 .263) does hold. (10.221) and the adjoint Dirac equation (10. Obviously.266) ρ(xµ ) = Ψ† (xµ )Ψ(xµ ) = s=1 |ψs (xµ )|2 (10.271) . j µ must transform like a contravariant 4-vector. 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 .265) (10.224). This transformation behaviour can also be deduced from the transformation properties of the bispinor wave function Ψ(xµ ). (10. Multiplying (10. of (10. The time-like component ρ of j µ 4 iγ µ ∂µ + iγ µ ∂µ ← Ψ(xµ ) = 0 . Since ∂µ transforms like a covariant 4-vector. and addition yields Ψ† (xµ )γ 0 The last result can be written ∂µ Ψ† (xν )γ 0 γ µ Ψ(xν ) = 0 . The reason is that the r.264) (10. in fact.10.224) from the right by Ψ(xµ ).h. (10. For this purpose we prove ﬁrst the relationship S −1 = γ 0 S † γ 0 .267) has the desired property of being positive deﬁnite. −ϑ) = e− 2 (w − iϑ)·σ 0 1 (w + iϑ)·σ 0 e2 1 (10.263) allows one to conclude that j µ must transform like a contravariant 4-vector as the notation implies.. (10.269) One can readily show that the same operator is obtained evaluating γ S (w. For our proof we note ﬁrst ˜ ˜ S −1 (w. ϑ) = S(−w. The conservation law (10. i.e.221) from the left by Ψ† (xµ )γ 0 . j) = Ψ† (xµ )γ 0 γ µ Ψ(xµ ) .263) obviously is a scalar under Lorentz transformations and that the left hand side must then also transform like a scalar.268) We will prove this property in the chiral representation. the property applies then in any representation of S.

bispinor wave functions and we need to determine corresponding components of the wave function. One obtains using (10. Note that we have assumed in (10. In the following we want to provide this characterization. however.283)].268) µ j = Ψ† (xµ )S † γ 0 γ µ SΨ(xµ ) = Ψ† (xµ )γ 0 S −1 γ µ SΨ(xµ ) . not only as the most simple demonstration of the theory. of which the wave functions are eigenfunctions as well. Combining this with (10. only two degrees of freedom of the bispinor four degrees of freedom are independent [c.322 Relativistic Quantum Mechanics in a moving frame of reference and the ﬂux j µ in a frame at rest. (10.272) With S −1 (Lη ξ ) = S((L−1 )η ξ ) one can restate (10. The independent degrees of freedom allow one to choose the states of the free Dirac particle as eigenstates of the ˆ ˆ 4-momentum operator pµ and of the helicity operator Γ ∼ σ · p/|p| introduced below.. commute with each other.275) The free particle wave function is an eigenfunction of Ho . (10. These operaˆ tors. In case of non-relativistic quantum mechanics the free particle wave function has a single component ψ(r. 10. We will start from the Dirac equation in the Schr¨dinger form (10.273) where we have employed (10.274) 10. but also as providing a basis in which the wave functions of interacting particle systems can be expanded and characterized. As it turns out. The additional degrees of freedom described by the four components of the bispinor wave function require. i. a property which leads to the energy– momentum (dispersion) relationship of the Dirac particle. as just mentioned. speciﬁc for the Dirac free particle. Like in non-relativistic quantum mechanics the free particle wave function plays a central role. the identiﬁcation of observables and their quantum mechanical operators. (10.246) S −1 (Lη ξ )γ µ S ( Lη ξ ) = (L−1 )νµ γν = (L−1 )ν γ ν = Lµ ν γν . additional characterizations. The solutions provide also a complete.282. µ (10. The operators commute also with Ho in (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. 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)]. In relativistic quantum mechanics a Dirac particle can also be characterized through a momentum. the wave function has four components which invite further characterization of the free particle state. as is required for the mentioned property.233).f.271) that the Dirac matrices are independent of the frame of reference.76). (10. For this purpose we consider . the Dirac particles are described by 4-dimensional. 10.272) results in the expected transformation behaviour µ j = Lµ ν j ν . As pointed out.231.9 Solutions of the Free Particle Dirac Equation We want to determine now the wave functions of free particles described by the Dirac equation. 10. t) and is determined through the momentum p ∈ R3 .233) o i∂t Ψ(xµ ) = Ho Ψ(xµ ) .e.232.

+m (10. E(p) = m2 + p 2 . hence.234) and (10.280) which has a positive and a negative solution = ± E(p) . (10.283) imply that the bispinor part of the wave function allows only two degrees of freedom to be chosen independently.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 (10.277) To solve this problem we write (10. from (10. the Dirac equation. (10. the socalled helicity of the particle. In fact. two-dimensional spinor state.234) of Pauli matrices holds (σ · p)2 = p 2 1 One can. one ﬁnds using (5.277) explicitly ( − m) 1 φo − σ · p χo 1 − σ · p φo + ( + m) 1 χo 1 = 0 = 0. 10. .278) Multiplication of the 1st equation by ( +m)1 and of the second equation by −σ · p and subtraction 1 of the results yields the 2-dimensional equation ( 2 − m2 ) 1 − (σ · p)2 φo = 0 . later to be identiﬁed with momentum and energy. 10.283) where is deﬁned in (10.93.278) provides us with information about the components of the bispinor wave function (10.279) According to the property (5.232) leads to the 4-dimensional eigenvalue problem m σ·p σ · p −m φo χo = φo χo . 10.280) σ·p (σ · p)2 φ0 = χ0 = +m ( + m) ( + m) p2 χ = χ0 .284) The relationships (10.279) the well-known relativistic dispersion relationship 2 = m2 + p 2 (10. 2 − m2 0 (10.94) Equation (10. conclude 1. We want to show now that these degrees of freedom correspond to a spin-like property. and φo . 1 (10.282. like the Klein–Gordon equation.276) where p and together represent four real constants.276).10.276) into (10.281) Obviously.282) (10. These two relationships are consistent with each other. (10. reproduce the classical relativistic energy–momentum relationships (10. Inserting (10. χo each represent a constant.231. namely φo and χo are related as follows φ0 χo = = σ·p χ0 −m σ·p φ0 .281).

S † = γ 0 S −1 γ 0 .287. Ψ† (0.287) will be justiﬁed now. (10. 0) γ 0 u 0 = |N+ (0)|2 . of (10.e.290).e.. (10. the l. is invariant under Lorentz transformations. hence. The connection between the solutions. +) = 1 . = +E(p) . +) γ 0 Ψ(0. +) γ 0 Ψ† (p. i. i. the solution for = +E(p). +) γ 0 Ψ† (p. (10. We want to demonstrate now that the signs on the r.s.285) The corresponding free Dirac particle is then described through the wave function Ψ(p.286) for p = 0 yields.290) should diﬀer. For the positive energy solution. ±) γ 0 Ψ(p.h. is Ψ(p. we present the negative energy solution.287..288) corresponding to the wave function Ψ(p.281)] separately. (10. ±) = S Ψ(0. ±). −|xµ ) = N− (p) −σ·p E(p) + m u u ei(p·r − t) . ±) . ±) Ψ† (0.292) . ±) denote the solution of a free particle moving with momentum p in the laboratory frame. = −E(p) . S −1 = γ 0 S † γ 0 and.219) and u† u = 1 [c. ±) γ 0 S −1 γ 0 γ 0 S Ψ(0. ±) S † γ 0 S Ψ(0. ±) = = = Ψ† (0. ±).287) in the minus sign on the r. +|xµ ) = N+ (p) u σ·p E(p) + m u ei(p·r − t) . of (10. Ψ† (p.259). One can see this as follows: Let Ψ(p. ±) Ψ† (0.289) Here N− (p) is a constant which will be chosen to satisfy the normalization condition Ψ† (p.287) the form of which will be justiﬁed further below.s. (10.324 Relativistic Quantum Mechanics For our further characterization we will deal with the positive and negative energy solutions [cf. First. according to (10. we present φo through the normalized vector φo = u1 u2 = u ∈ C2 . (10.h. Employing (10. u† u = |u1 |2 + |u2 |2 = 1 .285)]. according to (10.241). Similarly. +) = |N+ (0)|2 (u† . using γ 0 as given in (10. For this purpose we consider ﬁrst the positive energy solution. 10.268).290) which diﬀers from the normalization condition (10.291) Note that we have used that.f.h. (10.s. i. 10. ±) γ 0 Ψ(0. we demonstrate that the product Ψ† (p. where S is given by (10. and let Ψ(0. u† u = |u1 |2 + |u2 |2 = 1 . Hence. +) = −1 . (10.286) Here N+ (p) is a constant which will be chosen to satisfy the normalization condition Ψ† (p. (10. ±)γ 0 Ψ(p. (10. the solution for = −E(p). The form of this condition and of (10. through χo given by χo = u1 u2 = u ∈ C2 .. ±) denote the corresponding solution of a particle in its rest frame.e.

(10. σ·p u∗ E(p) + m T γo u σ·p E(p) + m u = 1 (10.234)] and using the normalization of u in (10. We can also conclude from our derivation N± (0) = 1 .285) results in 2 N+ (p) 1 − p2 (E(p) + m)2 = 1 (10.286) is 2 N+ (p) (u∗ )T . (10. of (10.295) Evaluating the l. We consider ﬁrst the positive energy solution. −) γ 0 Ψ(0.h.293) Obviously.s. this requires the choice of a negative side on the r. ( m + E(p) )2 − p 2 (10.287) written explicitly using (10. 2 E(p) (10.h.300) (10.6: Show that the normalization condition 2 N+ (p) (u∗ )T . (5.9: Solutions of the Free Particle Dirac Equation The same calculation for the negative energy wave function as given in (10.298) This result completes the expression for the wave function (10.s. σ·p u∗ E(p) + m T u σ·p E(p) + m u = 1 (10. Condition (10.286).301) yields the normalization coeﬃcient N+ (p) = m + E(p) .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. −) = |N− (0)|2 (0. 2m (10. 325 (10.219) yields 2 N+ (p) u† u − u† (σ · p)2 u (E(p) + m)2 = 1.10. Exercise 7.f.296) Replacing (σ · p)2 by p 2 [c.294) We want to determine now N± (p) for arbitrary p.302) .289) yields Ψ† (0. using γ 0 as given in (10.299) ( m + E(p) )2 . u† ) γ 0 0 u = − |N− (0)|2 .290) to assign a positive value to |N− (0)|2 .298) becomes N+ (p) = m + E(p) .

λ|xµ ) = pµ Ψ(p. The wave functions (10. 10. This attribute is the so-called helicity. 10.307) we investigate ﬁrst the observable due to the simpler operator σ · p. λ|xµ ) . Ho ] = 0 follows from (10. The commutation property [σ · p.286) into (10.289) is T −σ·p −σ · p 2 E(p) + m u N− (p) u∗ . 10. λ|xµ ) = λ E(p) Ψ(p.303) E(p) + m u Evaluating the l.e.h. We can. We ˆ want to show that this operator commutes with Ho and p to ascertain that the free particle wave ˆ function can be simultaneously an eigenvector of all three operators. 2. yields 2 N− (p) u† (σ · p)2 u − u† u (E(p) + m)2 = −1 .. −p). identical to the condition (10. hence. (10.289) are not completely speciﬁed.296) for the normalization constant N+ (p) of the positive energy solution.. pj ] = 0 . exp[i(p · r − t)] = exp(ipµ xµ ). i∂µ Ψ(p.e.326 Relativistic Quantum Mechanics We consider now the negative energy solution.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 (10.305) and.e. (u∗ )T γ o = −1 (10.309) . This degree of freedom 1 describes a spin– 2 attribute. i.275). 3 is fairly obvious. λ|xµ ) (10.286. The wave functions are also eigenstates of the momentum operator i∂µ . Rather than considering the ˆ observable (10. conclude N− (p) = m + E(p) 2m (10.306) i. have completed the determination for the wave function (10. λ|xµ ) in 4-vector notation. i. (10.290) written explicitly using (10. however. Inserting (10.308) 2 |p| ˆ Note that p represents here an operator. (10. deﬁned as the component of the particle spin along the direction of motion.300) have been constructed to satisfy the free particle Dirac equation (10. not a constant vector.275) yields Ho Ψ(p.s. j = 1. Helicity The free Dirac particle wave functions (10. The property [σ · p.286. The corresponding operator which measures this observable is 1 p ˆ Λ = σ· .. thereby.304) This condition is. Condition (10. the wave functions constructed represent eigenstates of Ho .307) where pµ = ( . This can be veriﬁed expressing the space–time factor of Ψ(p. the two components of u indicate another degree of freedom which needs to be deﬁned.289).289.

2m (10. ˆ The condition that the wave functions (10. In this case Λ = 2 σ 3 . are 1 i(px3 − E t) 0 p e Ψ(pˆ3 .224) 1 of σ3 the two u vectors (1. According to the deﬁnition (5. We assume ﬁrst particles with positive energy. t) = Np e (10.312) We assume now particles with negative energy. − 1 |r.313) 2 0 1 where Ep and Np are deﬁned in (10.. i. p3 ). States which are simultaneously eigenvectors of these three operators are also simulteneously eigenvectors of the three operators p.. 1 p = (0. can ˆ ˆ be simultaneously diagonal.. −.308) above. Ho and σ·p commute with each other and.310) We have shown altogether that the operators p. is independent of the direction of motion of the particle. 0)T and (0. Np = m + Ep . 2 the wave functions which are eigenstates of the helicity operator. i. + 2 |r. We consider ﬁrst the simplest case that particles move along the x3 –direction. 1)T are eigenstates of 2 σ 3 with eigenvalues ± 1 . +. in principle.e. +. t) = Np e 2 1 p m + Ep 0 0 i(px3 − E t) 1 p e Ψ(pˆ3 . The wave functions which are eigenfunctions of the helicity operator are in this case 1 −p m + Ep i(px3 + E t) 0 1 p e Ψ(pˆ3 . t) = Np e 1 0 0 p m + Ep i(px3 + E t) 1 p e Ψ(pˆ3 . + 1 |r. −. Since helicity is deﬁned relative to the direction of motion of a particle the characterization of u as an eigenvector of the helicity operator.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.e. 0.286) are eigenfunctions of Λ as well will specify now the vectors u. . = −E(p). t) = Np e (10.312). − 1 |r.e. Therefore. i.311) 2 −p 0 m + Ep 1 where e3 denotes the unit vector in the x3 -direction and where ˆ Ep = m2 + p2 . = +E(p).10. hence. 327 0 (σ · p)2 ˆ 2 (σ · p) ˆ 0 0 (σ · p)2 ˆ 2 (σ · p) ˆ 0 (10. Ho and Λ deﬁned in (10.

p) e |p| (10.312).312) can be obtained also by means of the Lorentz transformation (10. −. − ) 2 where ϑ(p) = = = exp exp i − ϑ(p) · σ 2 i − ϑ(p) · σ 2 1 0 0 1 . + 2 |r. −.319) 1 Ψ(p.311. + 1 ) 2 p 1 m + Ep u(p.315) e3 × p ˆ ∠(ˆ3 . − 2 |r. + 2 ) p 1 m + Ep u(p. for an r–independent wave function.223) as follows 1 u(p. γ . 0)T and (0. [One can also express the rotation through Euler angles α. t) 2 = Np u(p. 5.320) where Ep and N are again given by (10.229).222.262) for the bispinor wave function and the transformation (10. − 1 |r. + 2 ) u(p. 2 1 0 (10. γ and are 1 0 1 ˜ Ψ(p = 0. − 2 ) −p 1 m + Ep u(p. − 1 ) 2 −p 1 m + Ep u(p.316) describes a rotation which aligns the x3 –axis with the direction of p. (10. a wave function which represents free particles at rest.e.123). 1 |t) = √2 e−imt .220). + 1 |r. Generating Solutions Through Lorentz Transformation The solutions (10. 10. t) = N√ ei(p·r + Ep t) (10. − 2 ) u(p. +.. β.314) (10. +.313) except that the states (1. denoted by ˜. t) 2 = Np 1 u(p. 10. For this purpose one starts from the solutions of the Dirac equation in the chiral representation (10. 1)T have to be replaced by eigenstates u± (p) of the spin operator along the direction of p.318) 1 Ψ(p. +.321) . − 1 ) 2 ei(p·r − Ep t) (10. in which case the transformation is given by (5.311. i.226. + ) 2 1 u(p. + 2 ) ei(p·r − Ep t) (10. These eigenstates are obtained through a rotational transformation (5. 10.328 Relativistic Quantum Mechanics To obtain free particle wave functions for arbitrary directions of p one can employ the wave functions (10.] The corresponding free particle wave functions are then Ψ(p. The corresponding wave functions are determined through ˜ i˜ 0 ∂t − m Ψ(t) = 0 . t) = Np ei(p·r + Ep t) (10.317) Ψ(p.

2 0 e −1 329 (10.. χo are eigenstates of σ 3 with eigenvalues ±1. 10.325) to (10. 2 |t) = √2 .322) The reader can readily verify that transformation of these solutions to the Dirac representationsas deﬁned in (10.9: Solutions of the Free Particle Dirac Equation 0 1 1 ˜ Ψ(p = 0. +.323) should yield the solutions for non-vanishing momentum p in the x3 –direction. t) e 2 e− 2 w3 1 = √ 1 2 e 2 w3 i(px3 − E t) p e . − 1 |t) = √2 . +.176) imt → i ( p3 x3 and for the bispinor part according to (10. 0. −1 e 0 0 1 +imt 1 ˜ Ψ(p = 0.262) for a boost in the x3 –direction. (10. i. φo .323) Transformation (10. 2 0 1 1 0 +imt 1 1 ˜ Ψ(p = 0.322). yields for the exponential space–time dependence according to (10.262) φo χo → e 2 w3 σ 0 1 3 Et ) (10. Applying (10. for w = (0. For the resulting wave functions in the chiral representation one can use then a notation corresponding to that adopted in (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 . +. 0 1 (10. − 1 |t) = √2 e−imt .228) yields the corresponding solutions (10.324. 10.10.e.313) in the p → 0 limit. w3 ). −. This correspondence justiﬁes the characterization ±.322) can be written in spinor form 1 √ 2 φo χo e imt . 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 . + 1 |r.311. 10.325) One should note that φo . 2 The solutions (10. ± 1 of the wave functions stated in (10. −.311) e 1 w 2 3 ˜ Ψ(p(w3 )ˆ3 . χo ∈ 1 0 . − 1 |r.174.

61) p(w3 ) = m sinhw3 .328) the relationship (10.327) Employing the hyperbolic function properties cosh x = 2 coshx + 1 . t) e = i(px3 + E t) p e (10. t) e 1 = √ 2 −e− 2 w3 − 1 w3 2 1 i(px3 + E t) p e ˜ Ψ(p(w3 )ˆ3 . + 1 |r. + 2 |r. 2 sinh x = 2 coshx − 1 . t) e 1 0 1 0 0 1 0 1 1 0 1 0 0 1 0 1 i(px3 − E t) p e 1 Ψ(p(w3 )ˆ3 .329) . t) e 2 e 1 = √ 1 2 − e 2 w3 i(px3 + E t) p e (10. − 2 |r. +. Transformation to the Dirac representation by means of (10. −. − 2 |r. −.61) between the parameter w3 and boost velocity v3 . 2 (10.228) yields w3 cosh 2 = sinh w3 2 1 Ψ(p(w3 )ˆ3 . − 1 |r.326) where according to (10.311) for Ep one obtains 1 = √ 2 1 1 √ 2 + 1 = 1 − v3 2 m2 + = 2 m2 v3 2 1 − v3 cosh w3 2 1+ 2 v3 2 + 1 1 − v3 +m = 2m Ep + m 2m (10. and the expression (10. 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 . t) e 2 = i(px3 + E t) p e 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. −. −.

334) (10. Making this expectation a postulate allows one to derive the condition (10. To show this we rewrite the Dirac equation (10.330) Inserting expressions (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 .176). and since (2) the calculations following (10.333) .331) i S(Lη ξ )γ µ S −1 (Lη ξ ) ρ(Lη ξ )∂µ ρ−1 (Lη ξ ) − m Ψ (xµ ) = 0 (10.313) for −p. indeed. reproduces the positive energy wave functions (10. of course.243) and.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 ). expresses the system in the old coordinates. 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. This transformation. which however.330) into (10.10. that if Ψ(xµ ) is a solution of the Dirac equation that Ψ (xµ ) as given in (10.332) Here we have made use of the fact that S(Lη ξ ) commutes with ∂µ and ρ(Lη ξ ) commutes with γ µ . the essential property stating the Lorentz–invariance of the Dirac equation. λ. we will derive this equation only for inﬁnitesimal transformations. but not the observer. the proper transformation S(Lη ξ ). Invariance of Dirac Equation Revisited At this point we like to provide a variation of the derivation of (10. ρ(Lη ξ ) has been deﬁned in (10. We expect.123) above and characterized there. when we generated the solutions Ψ(p. is suﬃcient since (1) it must hold then for any ﬁnite transformation.243) considered solely the limit of inﬁnitesimal transformations anyway. thereby.9: Invariance of Dirac Equation Revisited and similarly sinh w3 = 2 Ep − m = 2m p 2m (Ep + m) 331 (10. In the new derivation we consider the particle described by the wave function transformed. 10. Such transformation had been applied by us.327).311) as well as the negative energy solutions (10. refered to as the active transformation. The reason why we provide another derivation of (10.221) using (10.243).168–10. (10. Λ|xµ ) from the solutions describing particles at rest Ψ(p = 0. in general.329.331) is a solution as well. The transformation is Ψ (xµ ) = S(Lη ξ ) ρ(Lη ξ ) Ψ(xµ ) (10. Such description will be essential for the formal description of Lorentz invariant wave equations for arbitray spin further below.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. = γν . Actually. Λ|t). λ.

0). w = (0. ∂µ ] = [x0 ∂2 + x2 ∂0 . w = (0.e. 0. it is met by S(Lη ξ ) as given in the chiral representation by (10. ∂µ ] = (10.335) shows that we need to demonstrate [J . ∂µ ] = (10. 1) .340) [K2 .. ∂µ ] = [x1 ∂3 − x3 ∂1 . As mentioned already we will show condition (10.334) for inﬁnitesimal Lorentz transformations Lη ξ . that any solution Ψ(xµ ) transformed according to (10.331) is again a solution of the Dirac equation. ϑ = (0. ∂µ ] = (10. . ∂µ ] = (10. 0). .337) [J2 . therefore.336). of course.338) [J3 . w = (0. ϑ = (0. 0. 0). ∂µ ] = [x0 ∂1 + x1 ∂0 . 0. 0.243) and. 0.339) [K1 . We have demonstrated. 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. . the Dirac equation is invariant under active Lorentz transformations. The second condition is identical to (10. ∂µ ] = (10.262). . 0). [K . .341) [K3 . ∂µ ] = (J )ν µ ∂ν . i. We will proceed by employing the generators (10. 0). ∂µ ] = [x3 ∂2 − x2 ∂3 . ∂µ ] = (10.336) We will proceed with this task considering all six cases: [J1 .332 Relativistic Quantum Mechanics We will demonstrate now that the ﬁrst condition is satisﬁed by ρ(Lη ξ ). ∂µ ] = [x2 ∂1 − x1 ∂2 .342) One can readily convince oneself that these results are consistent with (10.335) = ∂µ + M ν µ ∂ν + O( 2 ) The result will show that the matrix M ν µ is identical to the generators of Lν µ for the six choices ϑ = (1.128) 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. 1. ∂µ ] = (K )ν µ ∂ν . Inspection of (10. ∂µ ] = [x0 ∂3 + x3 ∂0 .

232).1 in Sect 10.345) holds |qV | << m . for the scalar ﬁeld V in (10.350) (10..343) One may also include the electromagnetic ﬁeld in the Dirac equation given in the Schr¨dinger form o ˆ (10. Non-Relativistic Limit We want to consider now the Dirac equation (10.6 we assume that the ﬁeld is described through the 4-vector potential Aµ and. This solution exhibits a timedependence exp[−i(m + )t] where for holds in the non-relativistic limit | | << m.346) ˆ ˆ ˆ α · π + qV + β m Ψ(xµ ) (10. 10.233) by replacing i∂t by (see Table 10. we deﬁne φ(xµ ) = e−imt Φ(xµ ) χ(x ) = e µ −imt (10.348) (10. For this purpose we choose the decomposition Ψ(xµ ) = φ(xµ ) χ(xµ ) .1) i∂t − qV and p by ˆ ˆ π = p − qA .349) We want to focus on the stationary positive energy solution. 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 . (10. The Dirac equation in the Schr¨dinger form reads then o i∂t Ψ(xµ ) = ˆ ˆ where α and β are deﬁned in (10.344) Using the notation σ = (σ1 . e. accordingly. Accordingly. σ2 .6 above).10: Dirac Particles in Electromagnetic Field 333 10.347) (10.10 Dirac Particles in Electromagnetic Field We like to provide now a description for particles governed by the Dirac equation which includes the coupling to an electromagnetic ﬁeld in the minimum coupling description. The Dirac equation (10. Following the respective procedure developed for the Klein-Gordon equation in Sect.g. Equivalently.345) (10. we replace the operator ∂µ by ∂µ + iqAµ .10. (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 10.345) in the so-called non-relativistic limit in which all energies are much smaller than m.351) X (x ) µ .

5. (10. which in the present case states ˆ ˆ2 ˆ ˆ (σ · π)2 = π + i σ · (π × π) . πk ] = = 1 1 [ ∂j + qAj .358) ( πj πk − πk πj ) = jk [πj . ∂k + qAk ] i i 1 1 1 1 [ ∂j .356). (10.353) (10. Φ Using (10.352) allow one to approximate (10. π ] . one can replace X in (10. 10.356) i∂t Φ ≈ Φ + qV Φ . accordingly. (10.357) Equation (10. ∂k ] + [∂j .334 and assume that for the time-derivative of Φ and X holds ∂t Φ << m . henceforth. due to the m−1 factor. ˆ Equation (10. i i (10. Ak ] + q 2 [Aj .349) yields i∂t Φ i∂t X = = ∂t X X Relativistic Quantum Mechanics << m .354) ˆ 0 ≈ σ · π Φ − 2m X . (10. For this purpose we employ the identity (5.7.360) For an arbitrary function f (r) holds ( [Aj . (10.230). ˆ ˆ For the components of π × π holds ˆ ˆ π×π = jk (10. identiﬁes X as the small component of the bi-spinor wave function which. Ak ] ) f = ( ∂j Ak − Ak ∂j + Aj ∂k − ∂k Aj ) f . ∂k ] + q [ ∂j .354) The properties (10. One obtains [πj .351) in (10. does not need to be considered anymore.350.359) We want to evaluate the latter commutator.346.361) . 10. 10.348. ∂k ] + q [Aj . and.357) for Φ can be reformulated by expansion of (σ · π)2 .355) ˆ σ·π Φ 2m (10. derived in Sect. Ak ] i i i i =0 =0 = q q [Aj .353) by X ≈ to obtain a closed equation for Φ ˆ σ·π 2m 2 (10. Ak ] .352) ˆ σ · π X + qV Φ ˆ σ · π Φ + qV X − 2mX . ∂k ] + [∂j .

222. using (10.1 with the magnetic ﬁeld. 0.2) it introduces the o extra term qσ · B Φ which describes the well-known interaction of a spin.10: Dirac Particles in Electromagnetic Field Using ∂j Ak f = ((∂j Ak )) f + Ak ∂j f ∂k Aj f = ((∂k Aj )) f + Aj ∂k f 335 where ((∂j · · ·)) denotes conﬁnement of the diﬀerential operator to within the brackets ((· · ·)). 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) .365) Comparision of this expression with (5. 0.2 around the magnetic ﬁeld. Equations (10. t) + q V (r.359.e. The formal solution of this equation is Φ(t) = e−iqtB·σ Φ(0) . Comparision of (10.363) allow us to write (10. t)]2 q − σ · B(r. while agreeing in all other respects with the non-relativistic Schr¨dinger equation (10.343) for the vector potential Aµ = (− Ze2 . Let us consider brieﬂy the consequences of the interaction of a spin.1 particle with a magnetic 2 ﬁeld B.367) . one obtains ( [Aj . t) (10. 10.10. −A). 5. Dirac Particle in Coulomb Field . In other words.362) or. t) ≈ ˆ [p − q A(r.364) which is referred to as the Pauli equation. 10. does not leave the theory again when one takes the non-relativistic limit. o reveals a stunning feature: the Pauli equation does justice to its two-dimensional character. 0) . but rather remains as a steady “guest” of non-relativistic physics with the proper interaction term.364) governing a two-dimensional wave function Φ ∈ C2 with the corresponding non-relativistic Schr¨dinger equation (10.. Ak ] ) f = [ ((∂j Ak )) − ((∂k Aj )) ] f (10. r (10.6)].357) in the ﬁnal form i∂t Φ(r. t) = × A(r. The respective bispinor wave function Ψ(xµ ) ∈ C4 is described as the stationary solution of the Dirac equation (10. t) 2m 2m Φ(r.1 which emerged in the Lorentz-invariant theory as an algebraic 2 necessity.2) governing a one-dimensional wave function ψ ∈ C.344. (10. i. ∂k ] + [∂j . 10. the interaction qσ · B induces a 1 precession of the spin.223) shows that the propagator in (10. [πj .358.366) can be interpreted as a rotation around the ﬁeld B by an angle qtB. πk ] = q q (( ∂j Ak − ∂k Aj )) = − i i ×A jk q = − B i jk (10.366) (10.Spectrum We want to describe now the spectrum of a relativistic electron (q = −e) in the Coulomb ﬁeld of a nucleus with charge Ze. the spin.360) and Aµ = (V. t) [see (8.363) where we employed B(r.

374).374) The ﬁrst term on the r. 3 µ=0 ˆ2 (˜ µ )2 πµ = π0 − π .373) (10. but the converse is not necessarily true.7 one can write the second term in (10. Employing πµ as deﬁned ˜ in Table 10. but diﬀers from it in an essential way. 3.370) Any solution of (10.370) for which holds ˜ ˆ ˜ ˆ 3 γ πµ γ πν = ˜ ˆ ˜ ˆ µ=0 µ ν (˜ µ )2 πµ + γ ˆ2 µ.ν=1 µ =ν γ µ γ ν πµ πν . γ ˆ ˜ ˆ (10. 1.230). 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 (10.6.368) ˜ where Ψ(xµ ) and γ µ are deﬁned in (10.232) in Sect. γ ˆ2 ˆ2 (10. is a solution of (10.1 one can write (10.229). according to (10.371). ˜ ˜ ˆ ˆ (10. µ = ν and altering ‘dummy’ summation ˜ ˜ ˜ ˜ .369) from the left by γ ν πν + m.180). noting from (10. (˜ 0 )2 = 1 and (˜ j )2 = γ 1 γ −1 j = 1.372) (10.369) For our solution we will adopt presently a strategy which follows closely that for the spectrum of pionic atoms in Sect. For this purpose we ‘square’ the Dirac equation. 5. 10. ˜ (10. we solve ˜ [ i˜ µ (∂µ + iqAµ ) − m ] Ψ(xµ ) = 0 γ (10.369). once a solution Ψ(xµ ) of (10. i. indeed.370).228) and in (10.369).370) is obtained then ˜ [ i˜ µ (∂µ + iqAµ ) + m ] Ψ(xµ ) γ is a solution of (10.230) γ µ γ ν = − γ ν γ µ . multiplying (10. respectively. ˜ However. can be rewritten using.370) ˜ [ i˜ µ (∂µ + iqAµ ) − m ] [ i˜ µ (∂µ + iqAµ ) + m ] Ψ(xµ ) = 0 γ γ (10. 2. (5.368) is also a solution of (10.375) Following the algebra that connected Eqs.h.231).e. (5.336 Relativistic Quantum Mechanics For the purpose of the solution we assume the chiral representation.370) resembles closely the Klein-Gordon equation (10.s.368) ˜ ( γ µ πµ − m ) Ψ(xµ ) = 0 . The diﬀerence arises from the term γ µ πµ γ ν πν in (10. Equation (10. This yields ˜ [ i˜ µ (∂µ + iqAµ ) + m ] [ i˜ µ (∂µ + iqAµ ) − m ] Ψ(xµ ) γ γ ˜ = (˜ µ πµ γ ν πν − m2 ) Ψ(xµ ) = 0 .371) such that we can conclude that (10.

π0 ] γ ˜ π ˆ j=1 = [˜ 0 . it holds 1 4 [˜ µ . r2 (10. ν = 1.367) holds qA0 = r ˆ Ze2 . πν ] γ ˜ π ˆ µ. πν ] γ ˜ π ˆ µ.367) of Aµ . πj ] . t) is a suitable test function and where ((· · ·)) denotes the range to which the derivative is limited.380) where f = f (r.378) According to the deﬁnition (10.367) and the deﬁnition (10. πν ] = 0 π ˆ for µ.ν=1 µ =ν σ 0 0 −σ · (( qA0 )) (10. γ µ γ ν πµ πν = ˜ ˜ ˆ ˆ µ.229). one can summarize (10. γ ] [ˆ0 . γ j ] are γ ˜ [˜ 0 .344) π ˆ [ˆ0 . 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. i.379) The commutators [ˆ0 . due to A = 0.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. Since [ˆµ .378) can be evaluated using (10.382) . γ ˜ π ˆ j=1 (10..381) According to (10.ν=1 µ =ν 3 j=1 3 3 = 1 4 1 2 1 [˜ .ν=1 µ =ν (10.376–10. the commutators [˜ 0 . πj ] in (10. γ ν ] [ˆµ .377) which follows readily from the deﬁnition (10.376) This expression can be simpliﬁed due to the special form (10.380) γ µ γ ν πµ πν = i ˜ ˜ ˆ ˆ µ.10: Dirac Particles in Electromagnetic Field indices.10. γ ν ] [ˆµ . πj ] + γ ˜ π ˆ 4 0 j [˜ j .ν=1 µ =ν = [˜ µ . 2.344).e. 3 (10. πj ] π ˆ = = (−i∂t + qA0 . γ 0 ] [ˆj . γ j ] [ˆ0 . −i∂j ] = − [ (∂t + iqA0 ) ∂j − ∂j (∂t + iqA0 ) ] f i ((∂j qA0 )) f (10. Altogether.ν=1 µ =ν 337 1 2 ( γ µ γ ν πµ πν + γ ν γ µ πν πµ ) ˜ ˜ ˆ ˆ ˜ ˜ ˆ ˆ µ.

Such solutions are of the form ˜ ˜ Ψ(xµ ) = Φ(r) e−i t .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. (10. 6. (10. Yjm (j + r 2 1 1 . 1 |ˆ) ) .387) The expression (10.388) in terms of states introduced in Sect.384) into (10.389) According to the results in Sect. (10. i. 10.338 Relativistic Quantum Mechanics where r = r/|r| is a unit vector.191) solved in Sect. .384) can be interpreted as the energy of the stationary state and. The term iσ · r.e.2 components ˜ Ψ(r) = and obtain for the separate components φ± (r) Ze2 + r 2 ˜ φ+ (r) ˜ φ− (r) (10. −j + 1. 6. . Yjm (j + 1 . (10. .1 . In the latter case.383) yields the purely spatial four-dimensional diﬀerential equation + Ze2 r 2 + 2 + i σ·r ˆ 0 0 −σ · r ˆ Ze2 − m2 r2 Φ(r) = 0 . + j } (10.147.5 which describe the coupling of orbital angular momentum and spin 1 {( Yjm (j − 2 . (10.5 the operator iσ · r is block-diagonal in this basis such that only ˆ 1 the states for identical j.6. a solution of the form ∼ Y m (ˆ) can be obtained.148). it is this quantity that we want to determine.375) the ˆ ‘squared’ Dirac equation (10. Insertion of (10.368) reads − Ze2 ∂t − i r 2 + 2 + i σ·r ˆ 0 0 −σ · r ˆ Ze2 − m2 r2 Ψ(xµ ) = 0 (10. . . we express 2 the solution of (10. in r ˆ fact.386) + 2 ± iσ · r ˆ Ze2 − m2 r2 φ± (r) = 0 . 2 |ˆ)} as given in (6. m = −j. however. couples the orbital angular momentum of the electron to its spin. Combining this result with (10.381). is genuinely two-dimensional and.383) We seek stationary solutions of this equation. (10. We note that these states are also eigenstates of the angular r 2 . Accordingly. hence.385) 1 We split the wave function into two spin. 1 |ˆ). 1 |ˆ). .. r r 2 2 2 j = 1.374). only the two states {Yjm (j − 2 . m values are coupled. 6. 2 3 2 .189) for the Laplacian and expansion of the term (· · ·)2 result in the twodimensional equation 2 ∂r − ˆ L2 − Z 2 e4 iσ · r Ze2 ˆ 2Ze2 + r2 r + 2 − m2 r φ± (r) = 0 .

151]. the missing additive term − 1 . Obviously.2 (j). property (6. rather than = 0. 10. The two eigenvalues of both matrices are identical and can be written in the form λ1 (j) [λ1 (j) + 1] where λ1 (j) λ2 (j) = = (j + 1 )2 − Z 2 e4 2 (j + 1 )2 − Z 2 e4 − 1 2 (10. 1 |ˆ) are r 2 2 ˆ eigenfunctions of L2 . Any similarity transformation leaves the ﬁrst term in (10. the eigenvalues are independent of m. namely. 1 |ˆ) = − Yjm (j ˆ r 2 2 1 2 h± (r) g± (r) Yjm (j − 1 . hence. conclude that the . λ1 (j) and λ2 (j).. . (6.390) (10. the values of the argument of λ1.10. .392) reads then in the diagonal representation 2 ∂r − λ1. 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 . Since.5 [c.2 (j)[λ1. not the wave functions. a speciﬁc pair of total spin-orbital angular momentum quantum numbers j. (6. in the present treatment. therefore. We select. 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. 6. 1 |ˆ) r 2 2 2 2 r r .396) and λ2 (j) [λ2 (j) + 1] (10.200). (10. unaltered.396) and (10. namely. 1.397) This equation is identical to the Klein-Gordon equation for pionic atoms written in the form (10. We can. . m and expand φ± (r) = Using σ · r Yjm (j ± 1 .392) h± (r) g± (r) = 0. which states that the states Yjm (j ± 1 .f.151). .395.391) derived in Sect. . such transformation can be chosen as to diagonalize the second term. .2 (j) as given by (10. − 1 r2 (j − 1 )(j + 1 ) − Z 2 e4 2 2 ±i Ze2 (j + )(j + ) − Z 2 e4 We seek to bring (10. except for the slight diﬀerence in the expression of λ1. 1 |ˆ) + r Yjm (j + 1 .10: Dirac Particles in Electromagnetic Field 339 ˆ momentum operator L2 [cf.395) (10.199). as in the case of pionic atoms.393) but do not explicitly consider further the wavefunctions.392).186)].2 (r) = 0 (10. However. 3 .2 (j) + 1) 2 Ze2 + + r2 r 2 − m2 f1. 1 |ˆ) r 2 (10.2 (j) being 2 j = 1 .394) Equation (10. we want to determine solely the spectrum. involving the 2 × 2 unit matrix.392) into diagonal form. and except for the fact that 2 2 we have two sets of values for λ1.

the stationary states have binding energies E = − mZ 2 e4 n = 1. . . 1.401) corresponds to a non-relativistic state 2 with quantum numbers n. .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. .340 Relativistic Quantum Mechanics spectrum of (10. −j + 1.404) = 0. j = − 2 . . . 3 . 2 2 1 ≈ m − (10. 2. .402) with (10. Obviously. 2n2 = 0. m) = 1 + m (n− + . . 1. . . (10. . accordingly. the second term on the r.400) corresponds to a non-relativistic state with quantum numbers n. 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 . −j + 1. m = − . n − 1 . and spin-orbital angular momentum j = + 1 . 2 2 m = −j. . . j where 1 corresponds to λ1 (j) as given in (10. 1. 1 ms = ± 2 (10. . 10. 2. . 1. −j + 1. (10. .396) we obtain.397) is again given by eq. (10. . 2. . counts the nodes of the wave function. .203). One can equate (10. . 2. . In case of non-relativistic hydrogen-type 1 atoms. It is given by n = n + + 1 where is the orbital angular momentum quantum number and n = 0. j = 1 . j = 1.396). Using (10. n − 1 .399) n = 0. . The magnitude of the relativistic eﬀect 2 2 e4 . . j = + 1 .402) In this expression n is the so-called main quantum number. . (10. .2 . 3. and spin-orbital angular momentum j = − 1 . . (10. . . m = −j. .h. including spin. . . .395) and 2 corresponds to λ2 (j) as given in (10. . m) 2 n = 1. 10. . of these equations describe the binding energy. . 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 . . . . 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.395.398.s. One may also state 2 this in the following way: (10. ( +1) − Z 2 e4 )2 (10. These considerations 2 are summarized in the following equations 1 ED (n. j . − + 1.401) if one attributes to the respective 1 states the angular momentum quantum numbers = j + 2 and = j − 1 . One obtains in case of (10. . . .400) (10. . . . . . albeit with some modiﬁcations. m = −j. . .398) 2 = 1 + m . . .400) and (10. j . .401) These expressions can be equated with the non-relativistic spectrum. 1. = 1 + m Z 2 e4 √ ( n − − 1 + 2 − Z 2 e4 )2 . . .403) √ Z 2 e4 2 ED (n.

60601 -3. . One can combine the expressions (10. binding energy / eV Eq. . (10.404) ﬁnally into the single formula ED (n.0041 keV and e2 = 1/137. The 2p 1 states with j = j = 1 2 3 2 2 2 2 for = 0 otherwise involve two degenerate states corresponding to Y 1 m (1.036 by means of Eqs. (10. case these six states are split into energetically diﬀerent 2p 1 and 2p 3 states. i. n − 1 j = 1 2 1 2 ± m = −j.60583 -3.3. ± 3 . in fact. The energies were evaluated with m = 511.2 the non-relativistic [cf. . . (10. 2 |ˆ) for m = ± 1 . .2: Binding energies for the hydrogen (Z = 1) atom. 10.402)] and the relativistic [cf.40151 ⇑ . the 2p 3 states with r 2 2 2 2 1 involve four degenerate states corresponding to Y 3 m (1. is in the range of 10 eV. 1 |ˆ) for m = ± 1 .405) In order to demonstrate relativistic eﬀects in the spectrum of the hydrogen atom we compare in Table 10.40146 ⇑ ⇑ -1.403. The table entries demonstrate that the energies as given by the expression (10.405)] spectrum of the hydrogen atom.402) -13. . The reason is that the mean kinetic energy of the electron in the hydrogen atom.1. the nonrelativistic degeneracy for the six 2p states of the hydrogen atom: in the present.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.. . r 2 2 2 .405). for example.551176 spectr. much less than the rest mass of the electron (511 keV). 10. . However. (10.1.e. . 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.. relate closely to the corresponding nonrelativistic states. (10. . −j + 1.40147 -1.10. i. the non-relativistic and relativistic energies are hardly discernible.402. 2. .405) in terms of the non-relativistic quantum numbers n. 0 0 1 1 0 1 1 2 2 spinorbital ang. j (10. Degeneracies are denoted by ⇑.e.551178 ⇑ .405 ) -13.10: Dirac Particles in Electromagnetic Field main quantum number n 1 2 2 2 3 3 3 3 3 orbital angular mom. m) = m 1 + Z 2 e4 (n−j −1 + 2 (j+ 1 )2 − Z 2 e4 )2 2 n = 1. . notation 1s 1 2 non-rel. j. = 0. Of particular interest is the eﬀect of spin-orbit coupling which removes. relativistic. mom. 1. This is clearly demonstrated by the comparision of non-relativistic and relativistic spectra of a hydrogen-type atom with Z = 100 in Table 10.3. binding energy / eV Eq.551177 ⇑ . j 1 2 1 2 1 2 3 2 1 2 1 2 3 2 3 2 5 2 341 rel.

206) provides in the present case ED (n. Introducing α = Z 2 e4 1 and β = j + 2 (10.35. (10.405).3 spectr. In order to investigate further the deviation between relativistic and non-relativistic spectra of hydrogen-type atoms we expand the expression (10.036 by means of Eqs. . to estimate the diﬀerence between the energies of the states 2p 3 and 2p 1 (cf.15. An example for such potential is the Coulomb potential V (r) = −Ze2 /r considered further below.407) 1 2 This expression allows one.3: Binding energies for the hydrogen-type (Z = 100) atom. binding energy / keV Eq. for example. (10.406) α √ 1+ (n − β + β 2 −α)2 The expansion (10.405) reads 1 (10.402) -136. (10. 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 10.8 ⇑ . 10.408) 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).2 -17.3).409) where V (r) is spherically symmetric.9 ⇑ .0 ⇑ ⇑ -15. Degeneracies are denoted by ⇑. 0) (10. 32 (10. It holds for n = 2 and j = 3 . binding energy / keV Eq.1 ⇑ . m) ≈ m − mZ 2 e4 mZ 4 e8 − 2n2 2n3 1 j+ − 3 4n + O(Z 6 e12 ) .10.1 ⇑ ⇑ ⇑ ⇑ rel.405) to order O(Z 4 e8 ).2. 0. 0.0041 keV and e2 = 1/137.6 -42. The energies were evaluated with m = 511. 0 0 1 1 0 1 1 2 2 spinorbital ang.342 main quantum number n 1 2 2 2 3 3 3 3 3 orbital angular mom. (10. Tables 10. We assume for the wave function the stationary state .402. j 1 2 1 2 1 2 3 2 1 2 1 2 3 2 3 2 5 2 Relativistic Quantum Mechanics non-rel.405 ) -161. 1 2 2 2 2 E (2p 3 ) − E (2p 1 ) ≈ − 2 2 mZ 4 e8 2 · 23 1 −1 2 = mZ 4 e8 . mom.1 -34. j.15.

5 [see. but are timeindependent. 1 |ˆ) .414) These equations can be brought into a more symmetric form using ∂r + which allows one to write (10. r 2 (10.e. 1 |ˆ) r 2 = i ∂r − 1 2 r g(r) Yjm (j + 1 . 6.411. σ · p ˆ σ · p f (r) Yjm (j + 1 .198)] not couple states with diﬀerent j. r r 2 2 2 a(r) b(r) 1 1 1 1 Yjm (j + 2 .e. 2 |ˆ) r r r .10. This term has been discussed in detail in Sect.5. (6. 1 |ˆ) r 2 2 = i ∂r + −j r 1 g(r) Yjm (j + 2 . The Dirac equation reads then. 1 |ˆ) r 2 2 = i ∂r + j+ r 1 2 3 2 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 .10: Dirac Particles in Electromagnetic Field form Ψ(xµ ) = e−i t 343 Φ(r) X (r) . 10. the term is block-diagonal in the space spanned by the states Yjm (j ± 1 . 6. X (r) ∈ C2 describe the spatial and spin. 1 |ˆ) and Yjm (j − 1 ..416) The diﬀerential equations (10. according to (10.413) ˆ σ · p g(r) Yjm (j − 1 . r 2 2 (10. i.415) 1 ˆ σ · p rg(r) Yjm (j − 2 . 1 |ˆ) r 2 2 (10. 1 |ˆ). ˆ σ · p X + m Φ + V (r) Φ ˆ σ · p Φ − mX + V (r) X = = Φ X (10. 10. 168]: the term is a scalar (rank zero tensor) in the space of the spin-angular momentum states Yjm (j ± 1 . 2 |ˆ) + r Yjm (j − 2 .2 degrees of freedom. (10. 10. 2 |ˆ) and does r 2 ˆ has odd parity and it holds [c.f. m-values.232. = r (10. 1 |ˆ) introduced in r 2 2 1 Sect. The arguments above allow one to eliminate the angular dependence by expanding Φ(r) and X (r) 1 in terms of Yjm (j + 2 .413. 1 |ˆ) . in particular. 2 |ˆ) + r Yjm (j − 2 . 2 |ˆ) r Φ(r) r ..414) ˆ σ · p rf (r) Yjm (j + 1 .411) (10.412) ˆ In this equation a coupling between the wave functions Φ(r) and X (r) arises due to the term σ · p.197. i. 6.410) 1 where Φ(r).345). pp.417) c(r) d(r) X (r) 1 1 1 1 Yjm (j + 2 .412) are four-dimensional with r-dependent wave functions. 1 |ˆ) r 2 1 1 = ∂r r r r (10.

2 |ˆ) r r f2 (r) 1 1 r Φ(r) i r Yjm (j − 2 .411.157). d(r) are coupled and b(r). using (10. g2 (r) ∂r + 1 2 f1 (r) − [ + m − V (r) ] g1 (r) = 0 (10. m pair contribute.421) g2 (r) X (r) 1 1 − Yjm (j + 2 .420) g1 (r) X (r) 1 1 − Yjm (j − 2 . 10. Accordingly.423) .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. there exist two independent solutions (10.418) holds for f1 (r). Presently. g1 (r) ∂r − j+ r j+ r 1 2 g1 (r) + [ − m − V (r) ] f1 (r) = 0 ∂r + and for f2 (r). According to (10.422) j+ r j+ r 1 2 g2 (r) + [ − m − V (r) ] f2 (r) = 0 ∂r − 1 2 f2 (r) − [ + m − V (r) ] g2 (r) = 0 (10. such expansion must include states with all possible j. only a(r). 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 (10.415. 2 |ˆ) = (10.419) Obviously. 10. m values.344 Relativistic Quantum Mechanics In general. Inserting (10. 2 |ˆ) r r where the factors i and −1 have been introduced for convenience. we consider the case that only states for one speciﬁc j.417) into (10. 1 |ˆ) r Φ(r) 2 2 r = (10.416).417) of the form f1 (r) i Yjm (j + 1 . (10. the orthonormality property (6.412). c(r) are coupled.

except for the opposite sign of the term (j + 1 ).422) follows in this case from the same procedure as that adopted for the radial wave function of the non-relativistic hydrogen-type atom. to states with angular momentum = j + 2 . enforcing the polynomial to be of ﬁnite order leads to discrete eigenvalues . Similarly. (10. 10. .422).422) and (10.420). 1. and obtains ﬁnally a polynomial function p(r) such that rγ exp(−µr)p(r) solves (10. the radial wave functions for Dirac particles in the potential (10. . determining wave functions of the type (10. (10. for small r. The solution of (10. i.420) is the large component and X is the small component.422).422) near r = 0.367) which is spherically symmetric such that equations (10. (10.405). 2s 1 .e. . . 3d 5 . correspodingly the states We consider ﬁrst the solution of (10. In the non-relativistic limit. Equation (10. together with the appropriate boundary conditions at r = 0 and r → ∞.422) corresponds to states f1 (r) Yjm (j + 1 . . We assume the 4-vector potential of pure Coulomb type (10. etc. = 0.423) apply for V (r) = −Ze2 /r. . 2 2 2 (10. 2p 3 .422) determines solutions of the form (10. 3s 1 .423) are identical.405).421). Behaviour at r → 0 We consider ﬁrst the behaviour of the solutions f1 (r) and g1 (r) of (10. the ones given in (10.424) Ψ(xµ ) ≈ r 0 1 i. According to the discussion of the spectrum (10. namely.426) . 3p 1 .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 ﬁrst that the wave function at r → 0 behaves as rγ for some suitable γ. Hence.409).422) describes the states 2p 1 . n − 1.10. 2 2 2 2 2 2 = j − 1 2 and.10: Dirac Particles in Electromagnetic Field 345 Equations (10.423).425) r 0 covers states with angular momentum 1s 1 .e. (10. one demonstrates then that the wave functions for r → ∞ behaves as exp(−µr) for some suitable µ.. .422). (10. 3d 3 . Hence. 1 |ˆ) r i 2 2 . Dirac Particle in Coulomb Field . 3p 3 . the 2 equations determine. We note that (10. in the non-relativistic limit wave functions f2 (r) 1 1 i Yjm (j − 2 . 2. can be written ∂r − j+ r j+ r 1 2 g1 (r) + Ze2 f1 (r) r Ze2 g1 (r) r = 0 ∂r + 1 2 f1 (r) − = 0.422. etc.405) of the relativistic hydrogen atom the corresponding states have quantum numbers n = 1. . Φ in (10.. 2 |ˆ) r Ψ(xµ ) ≈ . According to this procedure.

becomes imaginary. in case j + 1 )2 < Ze2 . One obtains γ = ± (j + 1 )2 − Z 2 e4 .427) makes only the positive solution 2 possible.266. (10. ) f1 (r) ( + m ) g1 (r) (10.435) . Only the exponentially decaying solution is admissable and. r2 (10.431) inﬁnite since.420).427) = = 0 0. 10. 10. according to (10.432) = = − ( − m. assume the r-dependence in (10.428) = 0. Such r-dependence 2 would make the expectation value of the potential r2 dr ρ(r) 1 r (10.267. determined that the solutions f1 (r) and g1 (r). µ = m2 − 2 (10. for the particle density holds then ρ(r) ∼ |rγ−1 |2 = Behaviour at r → ∞ For very large r values (10.429) This equation poses an eigenvalue problem (eigenvalue −γ) for proper γ values. for small r.422) becomes ∂r g1 (r) ∂r f1 (r) Iterating this equation once yields 2 ∂r g1 (r) 2 ∂r f1 (r) 1 . hence.427). (10.427) with 1 γ = (j + )2 − Z 2 e4 . hence.434) √ The solutions of these equations are f1 . we conclude f1 (r) →∞ ∼ e−µr . g1 ∼ exp(± m2 − 2 r).346 Setting f1 (r) yields γ b rγ−1 − (j + 1 ) b rγ−1 + Ze2 a rγ−1 2 γ a rγ−1 + (j + 1 ) a rγ−1 − Ze2 b rγ−1 2 or γ + (j + 1 ) −Ze2 2 2 Ze γ − (j + 1 ) 2 a b →0 Relativistic Quantum Mechanics ∼ a rγ . The assumed r-dependence in (10.433) = = ( m2 − 2 ) g1 (r) ( m2 − 2 ) f1 (r) (10. g1 (r) →∞ ∼ e−µr . We have. (10. g1 (r) →0 ∼ b rγ (10.430) 2 Note that the exponent in (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. (10.442) which leads to a partial cancellation of the asymptotically dominant terms. without loss of generality we can choose ˜ f1 (r) = φ1 (r) + φ2 (r) .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 . µ is real. However.438) (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. ρ m+ ρ [∂ρ − (10.436) f1 (r) = Insertion into (10. we set f1 (r) g1 (r) = = √ √ ˜ m + e−µr f1 (r) −µr = = 0 0 (10. (10.436).445) . We also introduce the new variable ρ = 2µ r .10: Dirac Particles in Electromagnetic Field For bound states holds 347 < m and.435 ).441) The last two terms on the l. g1 (r) = φ1 (r) − φ2 (r) ˜ (10.437) where they ˜ ˜ cancelled in case f1 = g1 = a. Let us consider then for the solution of (10.10. of both (10. √ g1 (r) = − m − a e−µr .441) correspond to (10. hence.s.434) √ m + a e−µ r .439) − m− e g1 (r) ˜ where µ is given in (10.h.434) results in √ √ (m − ) m + a − (m − ) m + a √ √ (m + ) m − a − (m + ) m − a which is obviously correct.444) (10. Equation (10. Accordingly. Solution of the Radial Dirac Equation for a Coulomb Potential To solve (10.440) (10.422 ) leads to √ √ j+1 Ze2 ˜ 2 − m − [∂r − ] g1 + m + ˜ f1 + r √r √ ˜ (m − ) m + g1 − (m − ) m + f1 = 0 ˜ 1 2 √ √ j+2 ˜ Ze m + [∂r + ] f1 + m − g1 − ˜ r r √ √ ˜ (m + ) m − f1 + (m + ) m − g1 = 0 ˜ (10.437) (10.440) and (10.

453) (s + γ)2 + Z 2 e4 − (j + 1 )2 2 2 √ Ze m2 − 2 βs . 10.452) From (10.430) which conform to the proper r → 0 behaviour determined above [c.448) φ2 (ρ) = ρ (10.454) .447) We seek solutions of (10.430)]. 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. (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 (10.447) leads to (s + γ) αs + (j + 1 ) βs − √ 2 s Ze2 m2 − 2 αs (10.449) for γ given in (10.446 .448.451) 1 √mZe − (j + 2 ) 2 2 αs .447) of the form n φ1 (ρ) = ργ s=0 n γ s=0 αs ρ2 βs ρ2 (10.450) follows Ze2 m2 − 2 βs − βs−1 =0 (10. = m− Ze2 βs s + γ − √m2 − 2 2 (10.449) into (10.f.446) (10.426–10. Using (10.446.430) one can write this βs = s+γ − 2 √ Ze m2 − 2 s (s + 2γ) βs−1 . Inserting (10. 10. (10.450) −√ mZe2 m2 − 2 βs ρs+γ−1 = 0 mZe2 m2 − αs (s + γ) β2 + (j + 1 ) αs + √ 2 s 2 +√ From (10.

10. ρ) . in fact. . We can.405) and the ensuing so values ( 10.430.405) allows one to identify n = n − j + 2 which.457) for βs and αs imply that so must then be an integer. = From (10. .449) must be of ﬁnite order. 2.449) and (10.456) − so 2 √mZe m2 − 2 (−so )(1 − so )(2 − so ) . 10. hence. . i.460) (10. 3d 3 holds n = 1. ρ) so β0 ργ F (1 − so . x) . n = 0.. . equivalently. Comparision with (10.448. (s − so ) β0 s! (2γ + 1)(2γ + 2) · · · (2γ + s) (10. is an integer.456) and (10.457) with the conﬂuent hypergeometric functions F (a. β0 j+ 1 2 (10. (10. namely.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 ) . with the associated Laguerre polynomials L(α) = F (−n.10: Dirac Particles in Electromagnetic Field Deﬁning so = √ one obtains βs = = .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 (10.461 for values given by (10. 3p 1 .457) One can relate the polynomials φ1 (ρ) and φ2 (ρ) deﬁned through (10. 1. . 2γ + 1. c c(c + 1) 2! (10. x) = 1 + a a(a + 1) x2 x + + . . conclude 2 2 2 that the polynomials in (10.456. 10. 2.. According to the deﬁnitions (10. .455) this conﬁnement of so implies discrete values for .461) In order that the wave functions remain normalizable the power series (10. m (n ) = . 2γ + 1.448. so = n .452) follows j+ αs = 1 2 349 Ze2 m2 − 2 −γ (10.. For example. 10. for the states 2p 1 .455) are ﬁnite. . . 1.405). .458) or. α + 1. c.e. n It holds φ1 (ρ) φ2 (ρ) = = ργ F (−so . This requires that all coeﬃcients αs and βs must vanish for s ≥ n for some n ∈ N. The expressions (10. (s − so ) β0 s! (2γ + 1)(2γ + 2) · · · (2γ + s) (10.10.459) − 2 √mZe m2 − 2 (10.

They are described through quantum numbers n.463) and yield [note the 1/r factor in (10.460.465) (10. 9. m.438. are two-dimensional vectors determined through the explicit form of the spin-orbital angular momentum states Yjm (j ± 1 . The coeﬃcients β0 in (10. 6.461). (Springer.421) with radial wave functions f1 (r). 10.350 Relativistic Quantum Mechanics Altogether we have determined the stationary states of the type (10. 2γ + 1.m (j − 1 .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 .6. 10. 2γ + 1. and (10. can follow the procedure adopted for the wave functions of the non-relativistic hydrogen atom and will not be carried out here. 10. Sect. The complete wave function is given by the following set of formulas 1 Ψ(n.464) 0 The evaluation of the integrals. 2µr) (2µ) 2 Γ(2γ + 1) 3 N = and m 4m )Γ(2γ + n + 1) (n +γ)m (n +γ)m (10. Berlin. 6.460. 2 . = j + 2 .461) are to be chosen to satisfy a normalization condition and to assign an overall phase. given by expression (10. 2µr) (10. Due to the form (10. (10. The normalization integral is then ∞ 0 π 2π r2 dr 0 sin θdθ 0 dφ (|Φ(r)|2 + |X (r)|2 ) = 1 (10. is time-independent. 1 |ˆ) in (6. 10.267).421) correspond to non-relativistic states with orbital angular momentum 1 1 = j + 2 .466) F1 (r) = F− (κ|r) .460.157.m (j + 1 .467) ± n F (1 − n . where2 F± (κ|r) = G1 (r) = F+ (κ|r) .421)] ∞ dr (|f1 (r)|2 + |g1 (r)|2 ) = 1 (10. The wave functions (10.147. as in (10.463) where Φ and X .442). 1 |ˆ) 2 2 r G1 (r) Yj. m|xµ ) = e−i t iF1 (r) Yj.421). 1 |ˆ) 2 2 r κ = j+ 1 2 (10.461).158) of the latter states absorb the angular integral in (10.148). = j + 2 . The orthonormality 2 2 r properties (6. j. (n + γ)m N (2µr)γ−1 e−µr − κ F (−n . j. which involve the conﬂuent hypergeometric functions in (10. g1 (r) determined by (10.439). 1990). Greiner.410) of the stationary state wave function the density ρ(xµ ) of the states under consideration. (10.469) This formula has been adapted from ”Relativistic Quantum Mechanics” by W.

The radial wave functions f2 (r) and g2 (r) in (10. 3p 3 . . this wavefunction has an orbital angular momentum quantum number = j − 1 and.470) r 0 Obviously.469). etc. in the non-relativistic limit.469). where F± (κ|r) are as given in (10. (10.421) which. describes the complementary set of states 1s 1 . become f2 (r) 1 1 i Yjm (j − 2 . 2s 1 . 3d 5 .467–10. not not 2 2 2 2 2 2 covered by the wave functions given by (10.422) to (10.421) are governed by the radial Dirac equation (10. hence.10: Dirac Particles in Electromagnetic Field 351 We want to consider now the stationary states of the type (10.472) F2 (r) = F− (κ|r) . 1 |ˆ) 2 2 r G2 (r) Yj.469) that the following wave functions result Ψ(n. m|x ) = e −i t iF2 (r) Yj.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. 3s 1 . tracing all steps which lead from (10.10.471) (10. 2 accordingly. = j − µ 1 2 . We have.465–10. 2 |ˆ) r Ψ(xµ ) ≈ e−1 t .m (j − 1 . j.m (j + 1 . 1 |ˆ) 2 2 r κ = −j − 1 2 (10. G2 (r) = F+ (κ|r) . obtained closed expressions for the wave functions of all the stationary bound states of relativistic hydrogen-type atoms. One can verify. 2p 3 .

352 Relativistic Quantum Mechanics .

in particular. 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. indeed.. may be exploited to express the Lorentz-invariant equations of relativistic quantum mechanics. ϑ). ϑ)µ ν correspond to diﬀerent a(z) and b(z).3) (11. isomorphic to the natural representation.e. This fact is remarkable since it implies that the representations provided through a(z) and b(z) are of lower dimension then the four-dimensional natural representation1 L(w. Starting point is the Lorentz transformation S(w. a complex variable z. ˜ We have altered here slightly our notation of S(w. diﬀerent L(w.4) = exp = w − iϑ . a(z) and b(z). expressing its dependence on w. i.225–10.2) (11. We will characterize the space on which the transformations a(z) We will see below that the representations a(z) and b(z) are. 1 353 . ϑ)µ ν .e. in fact. ϑ) for the ˜ in the chiral representation as given by (10. This is.262).Chapter 11 Spinor Formulation of Relativistic Quantum Mechanics 11. i. must be two-dimensional irreducible representations of the Lorentz group. what will be achieved in the following.1) (11. The lower dimensionality of a(z) and b(z) implies. that the corresponding representation of the Lorentz group is more basic than the natural representation and may serve as a building block for all representations. We will proceed by building as much as possible on the results obtained sofar in the chiral representation of the Dirac equation..1 The Lorentz Transformation of the Dirac Bispinor We will provide in the following a new formulation of the Dirac equation in the chiral representation ˜ deﬁned through (10. in a sense.229). ϑ through 3. z ∈ C ˜ Because of its block-diagonal form each of the diagonal components of S(z).

b(z).1 states as described by Dm m (ϑ). +1 2 2 |1. i. β. 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 . usually expressed 2 as product of rotations and of functions of Euler angles α. ± 1 represents the familiar spin– 1 states. the operator ∂µ and the Pauli matrices σ in the new representation and. +1 2 |1. (11. −1 2 2 |1. a distinct diﬀerence in the transformation behaviour of ˜ ˜ ˜ ˜ Ψ1 (xµ ). for z = iϑ. Ψ4 (xµ ) in case z = w + iϑ for w = 0. (11.354 Spinor Formulation and b(z) act. This characterization allows one to draw conclusions regarding the state space in which a(z) and b(z) operate. In this case holds a(z) = b(z) ˜ 1 (xµ ). θ ∈ R3 .7) where “∼” stands for “transforms like”. A First Characterization of the Bispinor States We note that in case w = 0 the Dirac transformations are pure rotations. − 1 2 2 z = iϑ. in 2 fact.e. and since b(z) ˜ ˜ acts on the third and fourth component of Ψ one can characterize Ψ ˜ Ψ1 (xµ ) ˜ Ψ2 (xµ ) Ψ3 (xµ ) ˜ ˜ Ψ4 (xµ ) ∼ ∼ ∼ ∼ 1 |2. β. will establish the map between L(w. formulate the Dirac equation. Ψ2 (xµ ) and Ψ3 (xµ ). to 1 ∗ ∗ z ·σ (11.8) We like to stress that there exists. In this case a(z) and b(z) are identical and read a(i ϑ) = b(i ϑ) = exp 1 − ϑ·σ 2 . Here | 1 . namely.e. express 4-vectors Aµ .. For ϑ = (0. 2 2 2 ˜ Since a(z) acts on the ﬁrst two components of the solution Ψ of the Dirac equation. ϑ ∈ R3 . a space of vectors state1 for which holds state2 state 1 ∼ | 1 . to the rotational transformations of spin.9) a∗ (z) = exp 2 . ϑ ∈ R3 . ϑ)µ ν and a(z). Relationship Between a(z) and b(z) The transformation b(z) can be related to the conjugate complex of the transformation a(z). ﬁnally. Ψ2 (xµ ) transform according to a(z) whereas Ψ3 (xµ ). + 1 2 2 state 2 ∼ | 1 . and the Klein–Gordon equation in the spinor representation.6) as given by (5. Aν . −1 2 2 z = iϑ..5) The transformations in this case. are elements of SU(2) and correspond. Ψ4 (xµ ) transform according to ˜ ˜ ˜ and Ψ b(z). the so-called spinor space.243). however. ϑ ∈ R3 (11. i. γ (see Chapter 5). (11. neutrino equation.

4) and .12) To prove (11. hold −1 = 1 One notices then. 11.2. −1 given by (11. therefore. (11. 11.10. in fact. Ψ (xµ ) as well as Ψ (xµ ). 11. (11.14) . (11. a result to be used further below. −1 = ∗ −σ3 = −σ ∗ .1: Lorentz Transformation of Dirac Bispinor ∗ ∗ ∗ where σ ∗ = (σ1 . that they alter w into −w. (11. however.12) does. σ2 . one can express a∗ (z) −1 = exp 1 − z ∗· σ 2 = b(z) .224)] ∗ σ1 = σ1 .12) yields σ1 σ3 or in short σ Similarly. Spatial Inversion ˜ One may question from the form of S(z) why the Dirac equation needs to be four-dimensional. one can show −1 −1 −1 = −σ1 = σ2 = −σ3 ∗ = −σ1 ∗ = −σ2 σ2 −1 (11. (5.16) σ = −σ ∗ .1) can be written in the form ˜ S(z) = a(z) 0 0 a∗ (z) −1 (11.f. Hence. µ ).11) 0 −1 1 0 = . The eﬀect of inversion on Lorentz transformations is. σ3 ). ∗ σ 2 = − σ2 . This demonstrates that a(z) is the ˜ transformation which characterizes both components of S(z).10) From this one can derive b(z) = where = 0 1 −1 0 .11. a∗ (z) −1 = exp 1 ∗ z σ∗ 2 −1 . but leave rotation angles ϑ unaltered.224.13) Explicit matrix multiplication using (5. Ψ (xµ ) even though these pairs of ˜ ˜3 ˜4 featuring the components Ψ1 (x ˜ 2 components transform independently of each other.11) one ﬁrst demonstrates that for and −1 as given in (11. that the transformation (11. The answer lies in the necessity that application of spatial inversion should transform a solution of the Dirac equation into another possible solution of the Dirac equation. 1. using f (a) −1 = f ( a −1 ).17) We conclude. ∗ σ 3 = σ3 . . −1 a∗ (z) −1 (11.12). One can readily verify [c. 355 (11.15) (11.18) with a(z) given by (11.

f.21) together with matrix multiplication as the binary operation forms a group. M is a complex 2 × 2–matrix. de Kerf Lie a Algebras.23) Exercise 11. is an element of SU(2).e.1: Show that SL(2.e. B¨uerle and E. ˜ ˜ P S(w + iϑ) P = S(−w + iϑ) = b(z) 0 0 a(z) ..21) = etr( 2 z·σ) = 1 1 (11. the space spanned by only two of the components of Ψ is not invariant under spatial inversion and. 2 = 1 i.2 Relationship Between the Lie Groups SL(2.19) i. (11.2. 2 . 3 . 11. can be found in G. Amsterdam. This implies ˜ Ψ1 ˜ Ψ2 P ˜ Ψ3 ˜ Ψ4 ˜ Ψ3 ˜ Ψ4 = Ψ1 ˜ ˜ Ψ2 . C) deﬁned in (11. ϑ ∈ R3 . the Lorentz invariant equation for neutrinos is only 2–dimensional. In fact. However.. P −1 = P.A.10. does not suﬃce for particles like the electron which obey inversion symmetry.1) We have pointed out that a(iϑ). based on the Jordan–Chevalley theorem.20) ˜ Obviously. hence. (11. which describes pure rotations.24) = etr(M ) (11. j = 1.22) which follows from the fact that for any complex.G. for particles like the neutrinos which do not obey inversion symmetry two components of the wave function are suﬃcient.224)] tr( σj ) = 0 . non-singular matrix M holds2 det eM and from [c. a(w + iϑ) for w = 0 is an element of SL(2. det(M ) = 1 } . (11. However. 1990). Exercise 1.A. C) = { M. the transformations a(z) and b(z) become interchanged. The transformation S(z) in the transformed space is then ˜ P 1. Obviously. One can verify this by evaluating the determinant of a(z) det( a(z) ) = det e 2 z·σ 1 (11.356 Spinor Formulation ˜ Let P denote the representation of spatial inversion in the space of the wave functions Ψ. (5. Part (Elsevier.C) and SO(3. 2.3. For the general case the proof. The proof of this important property is straightforward in case of hermitian M (see Chapter 5).

Noting that for covariant vectors according to (10.31) holds. in a transformed frame should hold Lρ ν σ µ Aµ σµ A µ .25) of the matrix M (Aµ ) is independent of the frame of reference.26) (11.29) where we used (10.2: Lie Groups SL(2. 0 −i i 0 .30) Since the components of Aµ are real. −→ µ Consistency of (11. 1 0 0 −1 . We will prove below [cf.11.2: Show that σ0 .28) σµ = 1 0 0 1 .31) Exercise 11. (11.2. C) and the group L↑ of proper.30). one wishes that the deﬁnition (11.25) can also be written M (Aµ ) = A0 + A3 A1 − iA2 A1 + iA2 A0 − A3 ..32) . σ2 . one can provide a simple expression for the inverse of M (Aµ ) M = M (Aµ ) ↔ Aµ = 1 tr M σµ 2 .27) requires then σ µ Aµ Lν µ σµ = σν (11.1) Mapping Aµ onto matrices M (Aµ ) 357 We want to establish now the relationship between SL(2. i. (11. σ2 . σ3 provide a linear–independent basis for the space of hermitian 2 × 2–matrices. (11. M (Aµ ) according to (11. Demonstrate that (11. Argue why M (Aµ ) = σµ Aµ provides a bijective map. The following important property holds for M (Aµ ) det ( M (Aµ ) ) = Aµ Aµ which follows directly from (11.27) −→ Straightforward transformation into another frame of reference would replace σµ by σµ . 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 . in fact. (11. (11. or+ thochronous Lorentz transformations. The function M (Aµ ) is bijective.26) or from the fact that the matrices σ0 .28) and (11.76).135)] this transformation behaviour. the matrix M (Aµ ) is hermitian as can be seen from inspection of (11.C) and SO(3.e. 0 1 1 0 . σ1 . σ1 . σ3 are hermitian. (11. In fact. Starting point is a bijective map between R4 and the set of two-dimensional hermitian matrices deﬁned through M (Aµ ) = σµ Aµ where (11.25) σ0 σ1 σ2 σ3 The quantity σµ thus deﬁned does not transform like a covariant 4–vector.75) holds aν = Lν µ aµ one realizes that σµ transforms inversely to covariant 4–vectors.

32. any a ∈ SL(2.31)] 1 a Aµ −→ A µ = tr a M (Aµ ) a† σµ 2 Because of (11.2.40) 1 tr a σν a† σµ 2 Aν (11.35) holds for this transformation A µ Aµ = Aµ Aµ (11. L(a)µ ν = 1 tr a σν a† σµ 2 . Due to det(a) = 1 the transformation (11. (11. + .358 Transforming the matrices M (Aµ ) Spinor Formulation We deﬁne now a transformation of the matrix M (Aµ ) in the space of hermitian 2 × 2–matrices a M −→ M = a M a† . a ∈ SL(2.37) using (11. (11. 11. (11.35) We now apply the transformation (11.3: Show that L(a)µ ν deﬁned in (11.31).40) is an element of L↑ .39) Exercise 11. C) and for any Aµ there exists an A µ such that M (A µ ) = a M (Aµ ) a† . C) deﬁnes the transformation [c. for any a ∈ SL(2. C) .34) where we used the properties M † = M and (a† )† = a. (11.36) The suitable A µ can readily be constructed using (11.33) This transformation conserves the hermitian property of M since (M )† = ( a M a† )† = (a† )† M † a† = a M a† = M (11. (11.33) conserves the determinant of M . Accordingly.37) deﬁnes actually a Lorentz transformation.f.33) det(M ) = = det( a M a† ) = det(a) det(a† ) det(M ) [det(a)]2 det(M ) = det(M ) .38) . (11. In fact.33) to M (Aµ ) describing the action of the transformation in terms of transformations of Aµ .25) Aµ = which allows us to express ﬁnally A µ = L(a)µ ν Aν . it holds for the matrix M deﬁned through (11. The linear character of the transformation becomes apparent expressing A µ as given in (11. In fact.37) which implies that (11.

α. C) and of SO(3.1) deﬁned through L(a)µ ν [cf.δ 1 4 Aαβγδ Γβα Γγδ α.C) and SO(3. i.1) For this purpose one writes using tr(AB) = tr(BA) L(a1 )µ ν L(a2 )µ ρ = ν a1 a2 product in SL(2. β 2.47) This completes the proof of the homomorphic property of L(a)µ ν .40)] respects the group property of SL(2.11.β γ.1) L(a)µ ν provides a homomorphism 359 We want to demonstrate now that the map between SL(2.46) which yields ¯ Lµ ρ = = = 1 1 Γ11 Γ11 + Γ22 Γ22 + Γ12 Γ21 + Γ21 Γ12 = tr ΓΓ 2 2 1 1 tr a† σµ a1 a2 σρ a† = tr σµ a1 a2 σρ a† a† 1 2 2 1 2 2 1 tr a1 a2 σρ (a1 a2 )† σµ = L(a1 a2 )µ ρ ..1).δ (11. (11. (11.2: Lie Groups SL(2.44) (σν )αβ (σν )δγ .43) (σν )αβ Γβα Γγδ (σν )δγ = ν. 2 (11. ¯ Lµ ρ = L(a1 )µ ν L(a2 )µ ρ = L( product in SO(3.41) 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 . (11.42) = ν Deﬁning Γ = a† σµ a1 . β δ = δ = = δ = δ 1 2 = 2 = 1 Aαβγδ (11. 1 ¯ and using the deﬁnition of Lµ ρ in (11.C) )µ ρ (11. .45) One can demonstrate through direct evaluation 2 α = 2 α = 2 α = = 2 α = 0 else β β γ γ = = = = γ = γ = 1. C) and SO(3.β γ.41) results in 1 ¯ Lµ ρ = 4 where Aαβγδ = ν Γ = a2 σρ a† 2 (11.e.

0. To obtain the generator of a(z) corresponding to the generator K1 . J1 .55) (11. 0). 1 (11.49). Aµ = (0.50) (11. (11. J Spinor Formulation The transformations a ∈ SL(2. K2 . 10.49) we obtain using (11. denoted below as κ1 . −A0 . we write (11. of (11.47. 0) ) σ0 A 1 − σ1 A 0 (11.h. 0.25) in the last step. Because of the condition det(a) = 1 only six independent real numbers actually suﬃce for the deﬁnition of a.s.48).53) Equating (11. 0. of (11..52) and (11.h. 0. Aµ = (0.e.57) . i.51) Insertion of (K1 )µ ν as given in (10. C) which correspond to K. K3 . 0. One expects then that six generators Gj and six real coordinates fj can be deﬁned which allow one to represent a in the form 6 a = exp j=1 fj Gj . J3 of the Lorentz transformations Lµ ν in the natural representations.48) We want to determine now the generators of the transformation a(z) as deﬁned in (11. 1 (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 . (11. Aµ = (0. correspond to the generators given by (10. 0). 1.59) = = = = 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.s. 0. 1. To this end we consider inﬁnitesimal transformations and keep only terms of zero order and ﬁrst order in the small variables.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 ) . correspondingly. C) as complex 2 × 2–matrices are deﬁned through four complex or.56) (11. 1) − σ1 σ0 0 0 One can verify readily that 1 κ1 = − σ 1 2 (11. eight real numbers. (11.36) M (Lµ ν Aν ) = a M (Aµ ) a† (11.360 Generators for SL(2.58) . For the r.48) yields for the l. J2 . 0).53) results in the condition σ 0 A1 − σ 1 A0 = ( κ 1 σ µ + σ µ κ † ) Aµ .2) which correspond to the generators K1 . noting the linearity of M (Aµ ).52) where we employed (11.54) This reads for the four cases Aµ = (1. M (Aµ + (K1 )µ ν Aν ) = = M (Aµ ) + M (Aµ ) − M ( (−A1 .

4: Show that the generators (11.1) acts on 4–vectors Aµ . C) acts on spinors φα ∈ C2 (characterized below) 1 (11. K3 and λ1 . J3 are given by 1 κ = − σ. state that the following two transformations are equivalent L(w. one can show that the generators κ2 .60) We can. − 1 . def α = 1. Here a(z) = ( aα β ) = describes the matrix a(z).61) This identiﬁes the transformations a(z = w − iϑ) as representations of Lorentz transformations. a(w − iϑ) = e− 2 (w − iϑ)·σ ∈ SL(2. a1 1 a1 2 a2 1 a2 2 (11. λ2 . C) correspond to the generators K and J of Lµ ν . λ3 corresponding to J1 . ϑ) = ew·K + ϑ·J ∈ SO(3. 2 2 φ1 φ2 ∈ C2 . the so-called contravariant spinors. 11. Exercise 11.11. 2 (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 .3 Spinors Deﬁnition of contravariant spinors We will now further characterize the states on which the transformation a(z) and its conjugate complex a∗ (z) act.64) . 2 (11. Similarly. w describing boosts and ϑ describing rotations. 2 λ = i σ. κ3 of a(z) corresponding to K2 .2. J2 . + 1 2 2 1 transforms under rotations (z = iϑ) like a spin– 2 state | 1 . hence. (11.3: Spinors 361 obeys these conditions. 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 .60) of a ∈ SL(2.

i. 1 . In the notation developed in Chapter 5 holds for the singlet state | .67) We will refer to φα . 0| 1 . as contravariant spinors and to φα . 0. (11. χβ . 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. 1 ) = 2 2 2 2 2 2 2 2 √ ±1/ 2 and equating the spin states of “particle 1” with the spinor φα . 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. −m) | . . should remain invariant under transformations a(iϑ).e. The corresponding states are deﬁned through φ1 χ2 − φ2 χ1 = φ1 χ1 + φ2 χ2 It obviously holds χα = χ1 χ2 = χ2 −χ1 . invariant and as such has the necessary properties of a scalar3 product. Deﬁnition of covariant spinors Expression (11.65) where | · · · 1 describes the spin state of “particle 1” and | · · · 2 describes the spin state of “particle 2” and (0. 0| 1 . as covariant spinors.69) (11.66) should constitute a singlet spin state. m 1 | .. is the singlet state. . In fact. In this case spinors φα transform like spin– 1 states and an invariant.68) (11..e. Using (0. Such a scalar product does. . The existence of a scalar product gives rise to the deﬁnition of a dual representation of the states φα denoted by φα . . ± 1 . 1 . However. To arrive at a suitable scalar product we consider ﬁrst only rotational transformations a(iϑ). The relationship between the two can be expressed φ1 φ2 = = 3 φ1 φ2 0 1 −1 0 (11.. 0| . those of “particle 2” with the spinor χβ one can state that the quantity 1 Σ = √ 2 φ1 χ2 − φ2 χ1 (11. . . . −m 2 2 2 2 1 1 1 1 2 (11. ± 1 . . . 1 ) stands for the Clebsch–Gordon coeﬃcient. i. χβ . 0 = 2 2 m=± 1 2 1 1 (0. i.e. exchange of φα and χβ alters the sign of the expression. etc. which 2 can be constructed from products φ1 χ2 . indeed.70) ‘Scalar’ implies invariance under rotations and is conventionally generalized to invariance under other symmetry trasnformations. this scalar product is anti-symmetric..66) is a bilinear form. invariant under transformations a(z). as we will demonstrate below such states are invariant under general Lorentz transformations a(z).

70) is φ1 φ2 −1 363 = = −1 φ1 φ2 .71. g µν of the Minkowski space [cf.10. 11.79) = a −1 T a κβ .3: Spinors as can be veriﬁed from (11.80) One can write in matrix notation aα β γ δκ αγ a δ αβ (11.69. 10.77) (11. Proof that φα χα is Lorentz invariant We want to verify now that the scalar product (11. 11.75) is given by φα = as can be readily veriﬁed.63.74)]. 11. (11. (11.72) 0 −1 1 0 Exercise 11.81) .67) will be expressed as φα χα = φ1 χ1 + φ2 χ2 = φ1 χ2 − φ2 χ1 .70) and (11.74) = = 0 1 −1 0 0 −1 1 0 (11.76) αβ αβ The scalar product (11. we will express (11. (10.3.11.1: Show that for any non-singular complex 2 × 2–matrix M holds M −1 = det(M ) M −1 T The matrices .71) (11. −1 connecting contravariant and covariant spinors play the role of the metric tensors gµν . 11.78) aβ γ γδ φδ (11. For this scalar product holds φα χα = −χα φα . (11. The inverse of (11. The transformation behaviour of φα according to (11. In the transformed frame holds φ α χα = aα β αγ aγ δ δκ φβ χκ .73) (11.73. 11. Accordingly.72)] φα φ α = = = = αβ αβ φβ φβ 11 21 11 21 12 22 12 22 (11.77) is Lorentz invariant.75) (11.68).72) in a notation analogous to that chosen for contravariant and covariant 4–vectors [cf. (10.69.

a∗ (z) provides a representation of the Lorentz group which is distinct from that provided by a(z).89) such that (11.82) and with f (A)T = f (AT ) for polynomial f (A) a −1 T = e− 2 z·(σ 1 ∗ )T = e− 2 z·σ = a−1 1 (11.83) Here we have employed the hermitian property of σ.83) into (11.85) φ1 φ2 ∗ ∗ .84) (11. k = 1. Insertion of (11. We denote ˙ (aα β )∗ = aα β ˙ (11. (11. from (11.80) φ α χα = φβ χβ . ˙ (11. Hence.91) . ˙ We also deﬁne covariant versions of φα φα = (φα )∗ . Obviously. σ † = (σ ∗ )T = σ. The complex conjugate spinors We consider now the conjugate complex spinors (φ ) = α ∗ γ δκ αγ a δ = a −1 T a κβ = a−1 a κβ = δκβ (11.e.87) which we will employ from now on.86) A concise notation of the conjugate complex spinors is provided by ˙ (φα )∗ = φα = φ1 ˙ φ2 ˙ ˙ (11.364 Using (11.14) one can write a −1 Spinor Formulation = e 2 z·σ 1 −1 = e 2 z· 1 σ −1 = e− 2 z·σ 1 ∗ (11.81) yields aα β and. As discussed above.63)..88) which one veriﬁes taking the conjugate complex of (11. it holds φk = (φk )∗ . i. 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 .90) extending the summation convention to ‘dotted’ indices.2) and (11. hence. 2.88) reads ˙ ˙ φ α = aα β φβ ˙ ˙ (11.

92) (11. 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. A comparision of (11.18) of the bispinor wave function Ψ.96) is Lorentz invariant. For the spinors φα and χα thus ˙ deﬁned holds that the scalar product ˙ φα χα = φ1 χ1 + φ2 χ2 ˙ ˙ ˙ ˙ ˙ (11.73.95) ˙ where αβ and αβ are the real matrices deﬁned in (11.97) implies then that φα transforms like Ψ ˙ φ1 ˙ φ2 ˙ transforms under rotations (z = iϑ) like a spin– 1 state | 1 . − 1 . one can state ˜ and (11.75. 11.e. (11.98) ˜ ∗ (z) −1 Ψ3 a ˜ Ψ 4 obviously implies that the solution of the Dirac equation in the chiral representation can be written in spinor form ˜ 1 µ Ψ1 (xµ ) φ (x ) ˜ Ψ2 (xµ ) φ2 (xµ ) φα (xµ ) = = .4: Spinor Tensors 365 The relationship between contravariant and covariant conjugate complex spinors can be expressed in analogy to (11.99) Ψ3 (xµ ) χ ˙ (xµ ) ˜ χβ (xµ ) ˙ 1 ˜ χ2 (xµ ) Ψ4 (xµ ) ˙ 11.18) ˜ 3 .93) (11. Ψ4 .100) .94) (11. 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 (11.75).97) The transformation.4 Spinor Tensors ˙ ˙ ˙ We generalize now our deﬁnition of spinors φα to tensors. A tensor tα1 α2 ···αk β1 β2 ···β (11.11..76) φα ˙ ˙ φα αβ ˙ ˙ αβ ˙ ˙ = = = = αβ ˙ ˙ αβ ˙ ˙ αβ αβ φβ φβ ˙ ˙ (11. 11. The transformation behaviour of φα is ˙ φα = ˙ ˙ β γδ ˙˙ φδ ˙ ˙ αβ a γ ˙ ˙ . in matrix notation. + 1 2 2 2 1 transforms under rotations (z = iϑ) like a spin– 2 state | 1 . (11. a property which is rather evident. i.

104) Indices on tensors can also be lowered employing the formula tα β = and generalizations thereof. we want to express Aµ through a spinor tensor.101) An example is the tensor tαβ which will play an important role in the spinor presentation of the Dirac equation.m aj akm t m (11.366 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 .109) .106) Exercise 11. tjk = a t aT jk = .102.103) Similarly.107) which reads in spinor notation [cf.102) = . ˙ ˙ ˙ ˙ (11. The 4–vector Aµ in spinor form We want to provide now the spinor form of the 4-vector Aµ ..106). i. i. An obvious candidate is [cf. (11.m aj a∗ t km m . the transformation bevaviour of a tensor tαβ reads in spinor and matrix notation t αβ = aαγ aβδ tγδ .e. 11. We had demonstrated that M (Aµ ) transforms according to M = M (Lµ ν Aν ) = a M (Aµ ) a† (11. This tensor transforms according to t αβ = aα γ aβ δ tγ δ ˙ This reads in matrix notation.108) Obviously.105) This tensor is actually invariant under Lorentz transforma= αβ . (11. An example of a tensor is αβ and tions. . this transformation behaviour is in harmony with the tensor notation adopted. αβ = αβ (11.25] M (Aµ ) = σµ Aµ . using conventional matrix indices j. it holds αβ . ˙ ˙ ˙ ˙ (11.e.1: Prove equation (11.e. (11. αβ ˙ αγ t ˙ γβ (11.. with ˙ contravariant indices αβ. the elements of which are linear functions of Aµ . k. (11. tjk = a t a† jk ˙ ˙ ˙ ˙ (11.. m.25) the tensor is explicitly Aαβ = ˙ A1 1 A1 2 ˙ ˙ A2 1 A2 2 ˙ ˙ = A0 + A3 A1 − iA2 A1 + iA2 A0 − A3 . i. This task implies that we seek a tensor.4. According to (11.103) A αβ = aα γ aβ δ Aγ δ .

∂µ in spinor notation The relationship between 4–vectors Aµ . (11.110) one obtains Aα β = ˙ A0 − A3 −A1 − iA2 −A1 + iA2 A0 + A3 A0 + A3 A1 + iA2 A1 − iA2 A0 − A3 .112) A22 −A21 ˙ ˙ −A12 A1 1 . Using (11.113) Aα β = ˙ (11.115) is correct.116) . 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 .114) We ﬁnally note that the 4–vector scalar product Aµ Bµ reads in spinor notation Aµ B µ = 1 αβ ˙ A B αβ . employing (11. 11.109.110) The 4-vectors Aµ .114) yields ∂αβ = ˙ ∂0 + ∂3 ∂1 + i∂2 ∂1 − i∂2 ∂0 − ∂3 .4: Spinor Tensors One can express Aαβ also through Aµ Aα β = ˙ ˙ 367 A0 − A3 −A1 + iA2 −A1 − iA2 A0 + A3 . ˙ 2 (11.110) one can state ∂ αβ = Similarly.111) A11 A12 ˙ ˙ A21 A22 ˙ ˙ ˙ 0 −1 1 0 (11.11.4. Aµ and tensors tαβ can be applied to the partial diﬀerential operator ∂µ .117) ˙ ˙ ∂0 − ∂3 −∂1 + i∂2 −∂1 − i∂2 ∂0 + ∂3 . (11. (11. (11. Hence.2: Prove that (11. (11.115) Exercise 11. (11.

one can conclude aσ µ ∗ −1 Lν µ σν 0 = .. Equation (10. (11. Note that (10. To obtain the transformation properties of σ µ we employ then (10.119) (11. (11. (11.s. Hence.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.125) is equivalent to µ ∗ −1 −1 a∗ (σ µ )∗ −1 −1 a Lν µ = (σ ν )∗ (11.123) (a∗ )−1 Lν µ a L ν µ = = σν (σ ) ν ∗ −1 (11. hence. (11.26) and its contravariant analogue σ µ σµ = σµ = 1 0 0 1 1 0 0 1 . holds also in the chiral representation.124). .125) is equivalent to (11.229) and (11. 0 1 1 0 .15) one can write σµ = and. . 0 −i i 0 .120) σµ 0 . (10..118) 0 −1 −1 0 0 i −i 0 For this purpose we consider ﬁrst the transformation behaviour of σ µ and σµ .125) a (σ ) Equation (11.243) in the chiral representation expressing S(Lη ξ ) by (11.122) aσ µ −1 a−1 −1 (a∗ )−1 0 Lν µ (11.e. σµ building on the known transformation behaviour of γ µ .e. We will obtain the transformation behaviour of σ µ .124) . 1 0 0 −1 .f.h.243) holds independently of the representation chosen.126) which is the complex conjugate of (11. γµ = ˜ 0 µ )∗ (σ −1 0 σµ σµ 0 (σ µ )∗ −1 .243)]. Comparision of (10. of this equation is 0 ∗ (σ µ )∗ a and. This is ˜ possible since γ µ can be expressed through σ µ .368 σµ in Tensor Notation Spinor Formulation We want to develop now the tensor notation for σµ (11. i. (11. −1 0 0 1 (11. σµ .121).124). . (11. hence.18) and γ µ by (11.118) yields ˜ γµ = ˜ Using (11.124) constitutes the essential transformation property of σ µ and will be considered further. i.121) One expects then that the transformation properties of σ µ should follow from the transformation properties established already for γ µ [c.

2) −1 369 (a∗ )−1 = −1 − 1 z ∗ ·σ ∗ 2 e = e− 2 z 1 ∗ · −1 σ ∗ = e2z 1 ∗ ·σ .. (σ ∗ )T = σ yields using (11. 0 −i i 0 µ=2 .4). −1 0 0 1 µ=3 . transformations (11. 0 −1 −1 0 µ=1 . i.2. (11.11. 11.2) −1 (a∗ )−1 = e2z 1 ∗ ·(σ ∗ )T = e2z 1 ∗ ·σ ∗ T = [a∗ ]T . 11. (11. under rotations the Pauli matrices transform like (j = 1.118) (σ µ )11 (σ µ )12 ˙ ˙ (σ µ )21 (σ µ )22 and (σµ )11 (σµ )12 (σ µ )21 (σ µ )22 ˙ ˙ 1 0 0 1 µ=0 ˙ ˙ ˙ ˙ = (11. One can. equation (11. hence.128) One can express. (11. = .132) 1 0 0 1 µ=0 ..124) a σ µ [a∗ ]T Lν µ = σ ν . .e. 3) σj −→ a(iϑ) σj a(iϑ) † = a(iϑ) σj a∗ (iϑ) T . therefore.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 .133) Combining (11.g. (11. according to (11. 11. This transformation behaviour. 11.124) using (11. In fact. e. (11.134) . 11.4) with w = 0.131) where a is given by (11. 0 1 1 0 µ=1 .4: Spinor Tensors One can rewrite (11.103) ˙ identiﬁes σ µ as a tensor of type tαβ . 0 i −i 0 µ=2 .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.108) that the same transformation behaviour applies then for general Lorentz transformations.2.129. 2. state that σ µ in a new reference frame is σ µ = a σ µ a† (11.102.107) to (11. (11.127) Exploiting the hermitian property of σ.16. It holds according to (11. 1 0 0 −1 µ=3 .

i. We will now consider the representation of σ µ . account for the 4–vector index µ.29)].132) and (11. 1.141) σ αβ = σ0 − σ3 σ1 + iσ2 −σ1 − iσ2 σ0 + σ3 = 2 0 0 0 1 0 −1 0 0 .12) for of σµ yields together with (5.133) is characterized through a 4-vector ˙ index µ.139) g −f g and.137) σαβ ˙ = δαγ δβ δ .110) to express σ αβ in terms 0 −1 1 0 0 0 0 1 (11. in fact.224) ˙ Using (11. of (11. We want to express now σ µ and σµ also with respect to the 4–vector index µ in spinor form employing (11. the transformation is inverse to that of ordinary 4–vectors. Each of the 4 × 4 = 16 matrix elements in (11. We can.136) σαβ = = 2 ˙ σ1 − iσ2 σ0 − σ3 0 0 0 0 1 0 0 1 where on the rhs. This yields 1 0 0 1 0 0 0 0 σ0 + σ3 σ1 + iσ2 (11. however.138) the ‘inner’ covariant spinor indices. ˙˙ (11.. γ. state ˙ ˙ ˙ ˙ (σ11 )11 (σ11 )12 (σ12 )11 (σ12 )12 ˙ ˙ ˙ ˙ ˙ ˙ ˙ ˙ ˙ γδ (σ11 )21 (σ11 )22 (σ12 )21 (σ12 )22 ˙ ˙ ˙ ˙ .e. ˙ ˙ 0 −1 1 0 ˙ (11. 3 as well as through two spinor indices αβ. δ. We can summarize that σ µ and σµ transform like a 4–vector. σµ in which the contravariant indices are moved ‘inside’. i. corresponds to the transformation (σµ )11 (σµ )12 ˙ ˙ (σµ )21 (σµ )22 ˙ ˙ = 0 1 −1 0 (σµ )11 (σµ )12 ˙ ˙ (σµ )21 (σµ )22 and −1 .136) results in the succinct expression 1 2 ˙ γδ ˙ (11. i. µ = 0. hence. matrices. (11.370 Inverting contravariant and covariant indices one can also state Lν µ σµ = σν . β. = σαβ ˙ ˙ ˙ ˙ ˙ (σ21 )11 (σ21 )12 (σ22 )11 (σ22 )12 ˙ ˙ ˙ ˙ ˙ ˙ ˙ ˙ 21 (σ )22 21 (σ )22 (σ21 ) (σ22 ) ˙ ˙ ˙ ˙ 21 22 Equating this with the r.e. 2. account for the 4-vector µ. Spinor Formulation (11. and the covariant indices are moved outside.. the submatrices correspond to tαβ spinors. whereas ˙ account for the elements of the individual Pauli the ‘outer’ contravariant spinor indices..e. α. The ˙ desired change of representation(σµ )αβ −→ (σµ )αβ corresponds to a transformation of the basis of ˙ spin states f g f −→ = (11.s.135) This is the property surmised already above [cf.138) Note that all elements of σαβ are real and that there are only four non-vanishing elements.114).140) where we employed the expressions (11. ˙ ˙ In (11.h.

121) of γ ˜ ˜ ˙ basis |αβ whereas the element of γ µ given by σ µ is in the basis |αβ .145) Let Aµ be a covariant 4–vector. One can write then the scalar product using (11. For this purpose we start from the ˜ µ .140) to transform each of the four submatrices. (11.144) results in the simple relationship γ µ Aµ = ˜ 0 Aα β ˙ Aα β 0 ˙ . Accordingly.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 . (11. employing then transformation (11.138) one can conclude that the following property holds 1 2 ˙ γδ σ αβ ˙ = 1 2 σ αβ ˙ ˙ γδ = δαγ δβ δ . ˙˙ (11. which in ˙ (11. ˙ ˙ ˙ ˙ (11.147) .144) The Dirac Matrices γ µ in spinor notation We want to express now the Dirac matrices γ µ in spinor form.11. (11.143) Combined with (11. This expression implies that the element of γ µ given by σ µ −1 is in the expression (11.4: Spinor Tensors 371 and.146) Exploiting the property (11.137) are in a basis (· · ·)αβ to a basis (· · ·)αβ results in ˙ σ αβ ˙ ˙ γδ = 2 (σ 11 )11 (σ 11 )12 ˙ ˙ ˙ ˙ (σ 11 )21 (σ 11 )22 ˙ ˙ (σ 21 )11 (σ 21 )12 ˙ ˙ ˙ ˙ 21 ) 21 ) (σ 21 (σ 22 ˙ ˙ 1 0 0 1 0 0 0 0 0 0 0 0 1 0 0 1 ˙ ˙ ˙ ˙ (σ 12 )11 (σ 12 )12 ˙ ˙ ˙ ˙ (σ 12 )21 (σ 12 )22 ˙ ˙ (σ 22 )11 (σ 22 )12 ˙ ˙ ˙ ˙ 22 ) 22 ) (σ 21 (σ 22 ˙ ˙ .142) = This can be expressed 1 2 σ αβ ˙ ˙ γδ = δαγ δβ δ . ˙˙ (11. we write ˜ ˙ γµ = ˜ 0 (σ µ )αβ µ ) )∗ ((σ αβ 0 ˙ ˙ .

226) ˜ i γ µ ∂µ Ψ(xµ ) = 0 i ∂ αβ ˙ i ∂ αβ 0 ˙ ˜ ˜ Ψ(xµ ) = m Ψ(xµ ) .150) .5 Lorentz Invariant Field Equations in Spinor Form Dirac Equation (11. α ˙ = = m φα m χβ .148) ˜ Employing Ψ(xµ ) in the form (11.147) allows us to rewrite the Dirac equation in the chiral representation (10. ˙ (11. (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 11.149) (11.

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

each energy level is (2l + 1)-fold degenerate. The energy levels are totally independent of l.8) ˆ [H.. Lie groups play an essential role in the discussion of mass degeneracies in particle physics. (12.. As an example of this we will consider the Coulomb problem with the Hamiltonian H = p2 k − . Sk . This degeneracy follows from the fact that any rotation. (7. l and the corresponding energy levels are independent of m. . Jk . (12. namely motion in a spherically symmetric potential described by the group SO(3) (or its double covering SU (2) as discussed in section 5.. is allowed to take values in 0. (7.λ1 . .. For example.. Jk ] = 0 . (7. The energy levels are mk 2 En = − 2 2 . (12. generates a state which has the same energy as the original one.. 2m r (12. . 3p and 3d all have the same energy.λ1 . this implies a degeneracy in the spectrum. 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.m (r) Y m (θ. V (r).3) If the newly obtained state is linearly independent of the original one. the states 3s.λn = E exp(iαk Sk ) ψE. which yields a set of eigenstates of the form ψE.5) with m = −l. .39). . 3 . l. We will investigate this shortly in more detail in the case of systems with spherical symmetry.18). . Classically.. 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 . m r (12. can then be reduced to a one-dimensional radial o equation.4) The stationary Schr¨dinger equation. as studied in chapter 7. Presence of additional symmetries may further increase the degeneracy of the system. . φ). as represented by an element of SO(3). (12. leaves the energy of the state invariant H exp(iαk Sk ) ψE.. Sk . where the symmetry generators. on an energy eigenstate. n − 1. In order to illustrate this. we ﬁrst consider a particular example of the implications of symmetry.. can be identiﬁed with the angular momentum operators.λn . Therefore.8) ... 2.λ1 .24). k = 1.12). In chapter 7 the dynamics of a particle moving in three dimensions under the inﬂuence of a spherically symmetric potential.7) 2 n where the orbital angular momentum. has been discussed.374 Spinor Formulation The action of any symmetry generator.. · · · . The spherical symmetry implies the commutation of the Hamiltonian with angular momentum operators (7.6) From elementary quantum mechanics we know the spectrum of the hydrogen atom.m (r) = vE. ψE.λn .

Aj ] = i ijk Ak .10) (12.3. Jj ] = i [Ki . ijk Jk . which is valid even when a and b do not commute. (12.21) . Kj ] = i ijk Kk .11) (12. which have a negative energy E.18) which complement the familiar angular momentum algebra of section 5. we will compute the hydrogen spectrum using the additional symmetry. 2 which can be shown to satisfy [Ai .17) (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. In order to understand the aforementioned extra degeneracy.15) which can be proved after very considerable algebra.16) Through some algebra [4] the following commutation relations can be veriﬁed [Ki . 2m r 1 x1 (J3 p2 − J2 p3 + i p1 ) − k .14) This follows from a · (a × b) = (a × b) · a = 0. Therefore. 2m r 1 x2 (J1 p3 − J3 p1 + i p2 ) − k . (12. 2E (12.19) (12. (12.20) (12. The vector in (12.13) It can be veriﬁed explicitly that its components commute with the Hamiltonian.12) (12. For this purpose we ﬁrst note that J · = 0. In the following we consider the bound states. (12. The corresponding quantum mechanical hermitian operator can be deﬁned by 1 1 2 3 = = = = x1 1 (p2 J3 − p3 J2 + J3 p2 − J2 p3 ) − k . We will also need the following identity [4] 2 = 2H J2 + m 2 + k2 . in the subspace of the Hilbert space corresponding to a certain energy we can replace H by E.11.9) (12. 2m r 1 x3 (J2 p1 − J1 p2 + i p3 ) − k . 2 1 B = (J − K). We introduce the following new operators 1 A = (J + K).8) is not a hermitian operator. 2m r (12. Now we scale the eccentricity vector as follows K= − m .

a = 0. Bj ] = 0.23) (12. .26) (12. In contrast to the discussion above about extra symmetries. A. This implies that a = b.27) (12. .15). 2 (12. b = 0. .22) (12. .32) A comparison with (12. 1. . . 1. ijk Bk . H B. 1.33) (12. = 0. a + b = n − 1}.31) where we have used (12. 2 (2a + 1)2 2 2 (12. Spinor Formulation (12.14) we note that A2 − B 2 = J · = 0. . 1. can be seen to follow from the triangle inequality as applied to J = A + B. the energy eigenvalues are found to be E=− mk 2 1 .22): A2 = a(a + 1) B 2 = b(b + 1) Following (12. . . .25) So far we have shown that the symmetry generators form an algebra. By comparing to the rotation algebra introduced in chapter 5. namely that J J > < A − B =0 A + B =2 A (12.21) and (12. .28) .24) (12. Using this equation the energy eigenvalues can be written in terms of the eigenvalues of A2 and B 2 operators. 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. . . which is identical to the the direct sum of the Lie algebra of of two SO(3) (or SU (2)) algebras.29) (12. . Noticing that A2 and B 2 have the same eigenvalues because of (12. Furthermore the bound on the orbital angular momentum. The most extreme case of this is the quantum analogue of a classically chaotic system. we can read oﬀ the eigenvalues of A2 and B 2 from (12. .7) tells us that (2a + 1) = n.30) 2 2 2 1 . . . Bj ] = i [Ai . a = 0. H = 0. . .34) It follows that l has to have values in {0 = |a − b|.376 [Bi .28). This illustrates the eﬀect of additional symmetries to the degeneracy structure of a quantum mechanical system. (12. a lack of symmetry implies a lack of degeneracy in the energy levels of a quantum mechanical system. Chaos is described classically as exponential . l. 2 1 .

If the chosen boundary is ‘irregular’ in a suitably deﬁned sense. However. after the discovery of the neutron in 1932. it is natural to represent them in terms of a two component vector. This is known as level repulsion. This enables us to utilize what we already know about the SU (2) symmetry group from the study of angular momentum. (12. another manifestation of chaos is the lack of independent operators commuting with the Hamiltonian. mp = 939. 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.) 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. For a more detailed discussion of quantum chaos in billiard systems we refer the reader to [7] and the references therein. which was discovered by observing degeneracies in the particle spectrum. which is a particle in box problem in two dimensions with a boundary which can be chosen arbitrarily.35) this mass equivalence can be viewed as an energy degeneracy of the underlying interactions. (Weak interactions are responsible. namely.57M eV /c2 .11. for the . (12.28M eV /c2 . Following the mass-energy equivalence of special relativity E = mc2 . It is important to place the isospin concept in its proper historical context.2 system p= 1 0 . in the sense that nearby trajectories in the phase space diverge from each other over time. the classical trajectories will diverge from each other after successive bounces from the boundary. Originally it was believed that isospin was an exact symmetry of strong interactions and that it was violated by electromagnetic and weak interactions. 12.2 Isospin and the SU (2) ﬂavor symmetry The concept of isospin goes back to Heisenberg. This (approximate) degeneracy led into the idea of the existence of an (approximate) symmetry obeyed by the underlying nuclear interactions. In the next section we will proceed with the discussion of a symmetry. For example. suggested that the proton and the neutron can be regarded as two states of a single particle.36) In analogy to the concept of spin regarding the rotations in 3-space as discussed in chapter 5. who. A typical example of this so called quantum chaos is the quantum billiard problem. This was motivated by the observation that their masses are approximately equal: mp = 938. 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. for example.5: Lorentz Invariant Field Equations 377 sensitivity to initial conditions. analogous to the 1 spin-up and spin-down states of a spin.36) into each other in abstract isospin space. the isospin symmetry is also governed by an SU (2) group ‘rotating’ components in (12. n= 0 1 . that the proton and the neutron behave identically under the so-called strong interactions and that their diﬀerence is solely in their charge content.

are made up of these two quarks.43) Similarly. In the framework of the quark of model. the proton had to be heavier.− 2 2 . They u form an isotriplet: π + = |1. Further than that. (12.38) which have all nearly identical masses. . If the mass diﬀerence (or the energy diﬀerence) were to be to be purely electrostatic in nature. 2 2 1 2 2 2 (12.6M eV /c2 .− 2 2 . and its third component. I. 1 π 0 = |1. Isospin symmetry is not an exact symmetry of strong interactions. albeit it is a good approximate one. 2 2 . 2 2 1 2 2 2 1 1 1 1 1 √ . respectively. 2 (12.− 2 2 1 2 2 2 1 1 1 1 . The mass diﬀerence between the neutron and the proton could then be attributed to the charge content of the latter. . the three pions in (12. However. I3 = 2 2 1 1 . (12. π − = |1. 2 2 1 1 1 . π 0 = |1. .39) All other isomultiplets. sec. 0 . the proton and the neutron can be written as totally symmetric uud and udd states. −1 . (12. it can be seen as part of a larger (and more approximate) symmetry which is of great utility to classify observed particle families. u¯ and d¯ states. 1 . −1 = = = 1 1 1 1 . d= 1 1 . For example.11. including the proton and the neutron. 2.41) (12.38) are ud. We can describe isospin multiplets the same way we have described the angular momentum and spin multiplets.37) As an example of a multiplet with I = 1 we have the three pions or π-mesons π + = |1.42) (12. spelling disaster for the stability of matter. They can be constructed with the same rules that have been used for angular momentum addition ¯ u in chapter 6. (mπ± = 139.36) as a multiplet with I = 2 1 1 p = I = . If it were otherwise the proton would be unstable by decaying into the neutron. Denoting the total isospin. as we shall see below. I3 .− . as good quantum numbers. the fundamental representation of the isospin symmetry corresponds to the doublet that contains the so-called up and down quarks u= 1 1 . mπ0 = 135. Shortly we 1 will see how to describe both the pion and nucleon states as composites of more fundamental I = 2 states.40) + 2 1 1 . Therefore it remains a useful concept.− . I3 = − 2 2 . 0 π − = |1. 1 we can re-write (12. For a precise description of the two nucleons as composite states. we refer the reader to [1]. including the spin and color quantum numbers of their constituent quarks. .0M eV /c2 ). the proton is the lighter of the two.378 Spinor Formulation beta decay). n = I = .

The names. 0 |1. (12. where leptons (electron. The relation between electric charge and isospin are given by the Gell-Mann–Nishijima relation which was ﬁrst derived empirically 1 Q = I3 + (B + S). such as proton and the neutron. 1 = √ (|p > |n > +|n > |p >). 2 = |n > |n > . Naturally.44) 2 where B is the baryon number and S is the strangeness. a number strange particles have been found which presumably contained a third quark species: the strange quark. In the late 1940’s and early 1950’s. a hydrogen isotope. 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.) are light. The possibilities are that of an isosinglet 1 |0.46) (12. but at the time the empirical concepts like isospin and strangeness quantum numbers were in use. It shall be noted here that the quark model was not invented until 1960’s.11. mesons have baryon number 0. Therefore the deuteron has to be an isosinglet state (|0. the neutrinos etc. 1 |1. whereas mesons are q q states.5: Lorentz Invariant Field Equations 379 The mass of the up and down quarks are not identical but they are both of the order of a few M eV /c2 ’s which is minuscule compared to the typical energy scale of hadrons (i. baryon and meson. this remains only an inaccurate naming convention today. By convention ¯ baryons have baryon number 1. If taken literally. (12. All other composite states have their strangeness given by the sum of the strangeness content of their constituents. −1 = |p > |p >. mesons have intermediate mass ranges. This is why isospin is such a good symmetry and why isomultiplets have nearly identical masses. 0 ) we should have seen nn and pp bound states of comparable energy in nature (because of isospin symmetry). muon. Now let us consider another example of combining the isospins of two particles. the up and down quarks are not the only quark ‘species’ .45) Is the deuteron an isosinglet state or an isotriplet? If it were an isotriplet (|1. 0 = √ (|p > |n > −|n > |p >) 2 and that of an isotriplet |1. I3 = 0.43). 0 ).47) (12. The value of the strangeness quantum number is taken. by accidental convention.48) (12. Following (12. the pions being examples thereof. originally refer to the relative weight of particles. strongly interacting particles) which is about a GeV /c2 . this will be mathematically identical to adding two spins. baryons generally are heavy. and quarks have baryon number 1 . Before proceeding further. All antiparticles have their 3 quantum numbers reversed. we shall setup some terminology: baryons are qqq states. to be −1 for the strange quark. but such states do not exist. As it later turned out. as some mesons discovered later are heavier than some baryons and so on. The reader may know that the deuteron. We will now try to describe its wave function in terms of its constituent nucleons.or ﬂavors as they are commonly called.37) and in analogy to (12. . consists of a proton and a neutron.e. The up and down quarks have strangeness zero. Therefore it has to have isospin.

49) The only assumption that we put in will be that the interaction is of the form V = α I(1) · I(2) . Refer to exercise 6.380 Spinor Formulation As an exercise on the implications of isospin symmetry we will consider nucleon-nucleon scattering. 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. 0 + |0. which is an isoscalar. We will eventually be able to compute ratios of scattering cross-sections between diﬀerent processes. 1 and (1/ 2)(|1. (f ) (i) (I (f ) I3 |00I (i) I3 ) is a Clebsch-Gordon coeﬃcient and I (f ) .48) the initial states in (I) and (II) have isospin values |1. Show that the expectation of I(1) · I(2) is state. γ (f ) ||T0 ||I (i) . Consider the generalization of tensor operators discussed in section 6. Exercise.55) . consider (I) p + p → d + π + (II) p + n → d + π 0 (12. γ (i) I (f ) . as discussed in the exercise above. γ (f ) |T00 |I (i) I3 .38) and the fact that the deuteron is an isosinglet we know that the isospins of the ﬁnal states in (I) and (II) are |1.51) (12.7 and 6. respectively. γ (i) = (f ) (i) (I (f ) I3 |00I (i) I3 ) (f ) (i) (−1)I −I (f ) (i) (12. where M is the scattering amplitude given by M = ﬁnal | α I(1) · I(2) | initial .5 for the spin-analogue of the same problem.259) in the present context: I (f ) I3 . (12. √ According to (12.8.53) 0 √ 1 2I (i) +1 I (f ) . We will now employ the (isospin analogue) of the Wigner-Eckart theorem (6. I3 .52) where γ (i) and γ (f ) denote degrees of freedom other than isospin.54) Here T00 ≡ V = α I(1) · I(2) . I3 (f ) = 1. γ (f ) T00 I (i) = 1. such as the spatial dependence of the wave function and spin. . 1 and |1.45) and (12. I3 = 1. 0 . σ. For completeness let us start by restating the Wigner-Eckart theorem (6. The initial and ﬁnal states can be denoted in more detail as | initial | ﬁnal = = I (i) . (12. γ (i) . γ (i) is a reduced matrix element deﬁned in the same sense as in section 6.259) discussed in detail in sections 6. I3 .7 to the case of isospin.8 to compute the ratio of the scattering amplitudes MI and MII .7. For example. γ (f ) (f ) (i) (12. γ (i) (i) (12. γ (f ) ||T ||I (i) . is proportional to |M|2 .50) . 1 4 in an isotriplet state and − 3 in an isosinglet 4 Using (12. Show that I(1) ·I(2) is an ‘isotensor’ of rank zero. 0 ). The dot product here refers to the abstract isospin space. The cross-section. Exercise.

(12.59) (12. 0 1 1 3 3 2. we will consider pion-nucleon scattering. I3 = 0. We want to compute the ratio of total cross-sections assuming a similar interaction as in the previous example σ( π + + p → anything ) . γ (f ) ||T00 || I (i) = 1. involving. [2] As a further example. −2 1 1 √ . −2 1 1 2. . γ (i) 2 1 = (10|0010) √ I (f ) = 1.58) (12. Once again we need the isospins for the initial and ﬁnal states.61) MII Note that the second term in MII vanishes due to the isospin conservation. γ (i) 3 +0. 2 . which is also manifested by a vanishing Clebsch-Gordon prefactor. (12. −2 2 1 1 3 2. but these are not dominant at relatively low energies.57) (12.62) We can now write the ratio of the scattering amplitudes: MI 1 √ . γ (i) = √ 2 1 (f ) (i) +√ I (f ) = 1. γ (i) 3 1 (f ) (i) I (f ) = 1. It follows σI = 2. for example. γ (f ) T00 I (i) = 1. = 381 (12. (12.11. σII which is in approximate agreement with the observed ratio. I3 = 0. The relevant Clebsch-Gordon coeﬃcients are easily evaluated: (11|0011) = (10|0010) = 1. 1 π − + p : |1. γ (f ) ||T00 || I (i) = 1. 2 = 3 2.5: Lorentz Invariant Field Equations 1 (11|0011) √ I (f ) = 1. −1 2 3 2 − + .65) (12.63) where common dynamical factors (which would not be as easy to compute) have dropped out thanks to the Wigner-Eckart theorem. which are obtained by a standard Clebsch-Gordan expansion π + + p : |1. (c) π − + p → π 0 + n .64) There are more exotic possibilities. σ( π − + p → anything ) The possibilities are (a) π + + p → π + + p . −2 . (b) π − + p → π − + p . −1 π0 + n : |1. = MII (1/ 2) (12.56) (12. I3 = 0. 1 1 1 1 3 √ 2. γ (f ) T00 I (i) = 0.60) (12.66) = 2 3 1 3 2. 2 = 2. particles with strangeness. I3 = 0.

etc. The classiﬁcation of the newly found particles as members of some higher multiplet structure was less obvious then the case of isospin.69) 12. 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. m 1 2.d = 1 .67) which are independent of m. Hence Gell-Mann and independently Nishijima postulated the existence of a separate quantum number. m . are up to 40% heavier. giving it a higher than usual lifetime.71) u = 0 . K + . however. In the light of the quark model. has a lifetime which is comparable to that of π + albeit being more than three times heavier. it appears an obvious generalization to add another component for an extra quark to the isospin vector space 1 0 0 (12.68) − 2 3 M1 2 Guided by empirical data we will further assume that M 3 >> M 1 . [2] (12. S(Σ− ) = −1 and S(π) = S(N ) = 0.65) Ma = M 3 2 Mb = 1 M 3 + 2 M 1 3 3 √ 2 Mc = 2 3 M3 2 √ 2 (12. m V V 3 2. For instance. for example. (12. called strangeness. while weak interactions did not.382 Spinor Formulation As in the example of nucleon-meson scattering we deﬁne the relevant matrix elements M3 = 2 M1 = 2 3 2. A computation similar to the previous example of nucleon-nucleon scattering yields (apart from common prefactors) the following amplitudes for the reactions in (12. Hence the strangeness changing strong decays of K + (or Σ− ) were forbidden. was motivated by the existence of particles that are produced strongly but decay only weakly. making an identiﬁcation of the underlying symmetry and the multiplet structure less straightforward. Strange partners of the familiar nucleons. mentioned in the previous section. such that S(K + ) = 1. 0 0 1 . . m 1 2. S. It was assumed that strong interactions conserved S (at least approximately).70) (12. which can be produced by π − + p → K + + Σ− .3 The Eightfold Way and the ﬂavor SU (3) symmetry The discovery of the concept of strangeness.s = 0 . which leads to the following 2 2 ratios for the cross-sections σa : σb : σc = 9 : 1 : 2.

in fact modern theory of strong interactions is a ‘gauge theory’ of this color group. This is believed to be an exact symmetry of strong interactions. history followed the reverse of this path. Lie algebras and related concepts. but we will stick to N = 3 in the following. called color. while preserving the norm. The reason is that the other known quarks. [2] ) Before discussing the signiﬁcance and the physical implications of the quark model. still leaves the norm invariant. As quarks were never directly observed. eiφ . (12. unitary relation k (12. we observe that (U V )† = V † U † = V −1 U −1 = (U V )−1 .5: Lorentz Invariant Field Equations 383 In this case the transformations that ‘rotate’ the components of (12. Multiplying U by a phase. namely charm. in reference to Lie groups. However SU (3) symmetry appears in another and much more fundamental context in strong interaction physics. becomes increasingly inaccurate for Nf > 3. which set the hadronic energy scale.3 for a brief discussion of gauge transformations).72) a∗ ak . Flavor SU (Nf ) symmetry on the other hand. However. The reader is also invited to revisit section 5. as in (12. This group of 3 × 3 unitary matrices is denoted by U (3). (The reader is referred to section 8.73) To verify that all such matrices form a group. 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. In this perspective the ﬂavor SU (3) symmetry may appear to be mainly of historical interest. which again form representations of an SU (3) group. of three dimensions. It is seen that such a matrix U has to satisfy the following k U † = U −1 . called quantum chromodynamics. but there are a number of subtle diﬀerences. The quarks posses another quantum number. The reader shall note that most of what is being said trivially generalizes to other unitary groups.71) into each other. have to be elements of the group SU (3) (which we will investigate closely very soon). First particle multiplets were identiﬁed as representations of the SU (3) group. we want to ﬁnd those transformations ak → Ukl al that preserve the norm. should correspond to. ak . SU (N ). (The ‘bare’ mass of the charm quark is already heavier than the two nucleons. In many respects it will resemble the more familiar group SU (2) discussed in some detail in chapter 5. (12. we will establish some mathematical preliminaries about the group SU (3). of a. it is the group .71). Since arbitrary phase factors are of no physical interest. for a period they were considered as useful bookkeeping devices without physical content. the same way nucleons and pions were identiﬁed as representations of the isospin SU (2) symmetry. giving rise to the quark model.74) for any two unitary matrices U and V . The bottom and top are even heavier. Hence the group U (3) has 9 dimensions. The unitarity relation imposes 9 constraints on the total of 18 real degrees of freedom of a 3 × 3 complex matrix. Given a complex vector. Then came the question of what the fundamental representation. Because of this additional constraint SU (3) has 8 dimensions.11.1 whenever necessary. bottom (beauty) and top (truth) are signiﬁcantly heavier than the hadronic energy scale. where Nf is the number quark ﬂavors.

7.79). U . λ7 = 0 0 . via the anticommutator relation 4 [λj .) The unit determinant condition requires that all λk are traceless. λ2 = . obey the following relation tr(λj λk ) = 2δjk . . 0 0 0 .78) where fjkl are the antisymmetric structure constants similar to the jkl of SU (2) given in (5. λk ] = 2ifjkl λl . −2 λ4 λ6 λ8 The generators. Therefore we will express elements of SU (3) as U = ei k (12. any unitary matrix.32). 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 . can be written in the form U = eiH where H is a hermitian matrix. The fundamental relation to be preserved is (12. The Lie algebra structure is given by the commutators of λk [λj . We can also introduce the constants. As in the case of SU (2) higher dimensional representations obeying the same structure can be found. the meaning being apparent from the context. λ3 = 0 −1 0 . since det(eA ) = etrA . δjkl . λ5 = . 3 (12. As discussed in section 5. regardless of the dimension of the representation. (12. The reader is invited to compare the structure of SU (3) to that of SU (2) discussed in section 5.80) This is the fundamental or deﬁning representation of SU (3). (12.79) (12.1.76) where λk are 8 linearly independent matrices forming the basis of the Lie algebra of SU (3).384 Spinor Formulation SU (3) and not U (3) that is of main interest. λk . .77) λ1 . λk ]+ = δjk + 2δjkl λl . (We shall at times refer to the Lie algebra with the name of the group.75) αk λk (12. which can be veriﬁed explicitly as matrix identities.

I3 . 2 1 [T0 .86) (12. 3 (12. Y is called the hypercharge. First. [Y.11.88) which deﬁnes (not uniquely) a maximal set of mutually commuting operators {Y.81) In chapter 5.89) (12.77) is not the most useful basis in the study of SU (3).94) . V± ] = ±V± . U± = F6 ± iF7 . we have [Y.44). Derive the Gell-Mann–Nishijima relation (12. these relations have been used to construct the angular momentum spectrum as well as the function space representation of the rotation algebra. The generators of SU (3) can be arranged into a very similar form to that of SU (2). J− } proved useful. T 0 = F3 .71) show that T0 is the isospin operator.93) (12.5: Lorentz Invariant Field Equations 385 As it turns out the set of generators in (12. We ﬁrst introduce the F-spin operators 1 Fi = λ i . (12. which satisﬁed an ‘eigenvalue equation’ [J0 . V± ] = ± V± . 1 [T0 .91) [T0 . Exercise.90) (12. Using the convention in (12. 2 (12.83) (12.85) (12. In the basis (12. In the case of SU (2) the identiﬁcation of ‘ladder’ operators {J+ . namely spherical harmonics.82) With another change of basis we arrive at the ‘standard’ form of the generators of the Lie algebra of SU (3) T± = F1 ± iF2 . V± = F4 ± iF5 . U± ] = U± . 2 Y = √ F8 . T± ] = 0. T0 ] = 0. J± ] = ± J± . [Y.87) Exercise. U± ] = ±U± . (Recall that the strange quark has S = −1 by convention). starting with the observation that Y = B + S.92) (12. T± ] = ±T± . T0 } and [Y. 2 (12.84) (12. (12.87) the commutation relations between the generators can be expressed in a succinct manner. (12.

The maximum number of mutually commuting generators of a Lie algebra is called its rank. we will use these commutation relations to construct representations of SU (3).104) (12. Thus. V± . When the basis of a Lie algebra is expressed in such a way to satisfy the form of the eigenvalue relations as given above. An operator which commutes with all generators of a Lie group is called a Casimir operator. since there are two mutually commuting generators in SU (3) as given in (12. [V+ . The interested reader is instead referred to a very readable account given in chapter 12 of [4]. However. [T+ .95) (12.103) Any remaining commutators follow from hermiticity. U± } to this maximal set by ‘eigenvalue equations’ and [T+ . as the relation between the states in a given representation can be conveniently expressed in terms of ladder operators.386 Spinor Formulation which relate the remaining generators {T± . we have [T+ .102) (12. As in the case of J 2 and SU (2). V+ ] = 0. it is said to be in Cartan-Weyl form. jkl (12. V+ ] = 0. U− ] = 0. similar to the way it was done in section 5. J 2 ) which commutes with all of the generators. the representations will now lie in a T0 − Y -plane. [U+ . U− ] = Y − T0 = 2U0 . Another important property of SU (2) is the existence of an operator (namely the total angular momentum. We can construct two such independent Casimir operators for the group SU (3). . Finally. U+ ] = V+ .5. Note that.99) (12.101) (12. while SU (3) has rank 2. Casimir operators can be used to label irreducible representations of the Lie algebra. (12. T0 }. [T+ . U0 and V0 are linear combinations of T0 and Y . C1 = k λ2 = − k 2i 3 fjkl λj λk λl . SU (2) has rank 1. V− ] = 2 (12. In the case of SU (2) the representations lay on a line on the J0 axis. 3 [U+ .96) (12.98) (12. This form is essential for easy labeling of the representations of the group. V− ] = −U− . T− ] = 2T0 . 2 3 Y + T0 = 2V0 .97) which relate commutators of generators with opposite eigenvalues to the maximal set {Y.100) (12.88). 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+ .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− . [T+ .

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.) All other representations can be constructed by combining these two conjugate representations. The representations for hexagons with sides of length p and q in the T0 − Y -plane. y .1).antiquark pair. 1) as an example. The representations D(1. y U± |t3 . = y |t3 . 2 1 = γ t3 . For example. 2 (12.108) (12.110) (12. y V0 |t3 . y = t3 |t3 . 2 (12.y ± 1 .3). y . respectively. 2 1 t3 ) |t3 .109) The same way that J± |m is proportional to |m ± 1 in the case of the angular momentum algebra. Such a representation is labeled as 1 D(p.106) (12. The representations of SU (3) are constructed analogous to those of SU (2) by identifying the ‘boundary’ states annihilated by raising (or lowering) operators. y .112) The eﬀect of these operators to the states in the y − t3 plane have been outlined in Fig.111) (12. A notation suggestive of the dimensionality of the representation can be used to identify 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. The details of this procedure is beyond the scope of this chapter. Now we will construct explicit representations of SU (3). All other states of the representation are then constructed by successive application of ladder operators T± .11. 1) correspond to the triplets of quarks and antiquarks. For example. it was shown in section 5. Figure (12. 1) representation. y . we have U0 |t3 . 2 1 = β t3 ± . (See Fig. y Y |t3 .2) shows the representation D(2. V± . |t3 . (12. (12. the pion family forms part of an octet corresponding to the D(1.107) 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 . For example. especially chapters 7 and 8. y 3 = ( y− 4 3 = ( y+ 4 1 t3 ) |t3 . 0) and D(0. The interested reader is referred to [4]. This property can be used to label representations in terms of the values of the Casimir operators. we have T± |t3 . y 1 = α t3 ± . As another example for the representations of SU (3). q) and it has a dimensionality of 2 (p + 1)(q + 1)(p + q + 2). y : T0 |t3 . For example the pion octet (or any other meson octet) is therefore realized as states of a quark . . U± . y ± 1 .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. y . y V± |t3 .88) we can label states by the eigenvalues of T0 and Y operators. y . (12. Because of (12.

¡ ¢ ¡ ¢ 8 6 2 '745301) ¥¦¤ ¤ #" ! (& % '$ © ¨ ¦ §¥ ©§ ¨ . £ Figure 12.2: D(2. 1) representation of SU (3).388 £ Spinor Formulation Figure 12.1: The eﬀect of SU (3) ladder operators on the y − t3 plane. The states in the inner triangle are doubly degenerate.

3 (12. This expansion can be compared. 1) are denoted by [3] and [¯ respectively.116) (12.3: D(1. where we would write [3] ⊗ [3] = [1] ⊕ [3] ⊕ [5].115) (12. 3]. . 0) and D(0. 1) is written as [8] etc. 0) or the fundamental representation of ﬂavor SU (3) symmetry. in the case of SU (2). . £ (12. to the case of adding two spin triplet states. This way the quark-antiquark states can be represented as follows [3] ⊗ [¯ = [8] ⊕ [1] 3] (12.118) ¡ ¢ ¨ ©§ ¨¥¦¤ " ! . 2 3 2 0.5: Lorentz Invariant Field Equations 389 D(1. The octet D(1.113) The additional singlet state corresponds to the η meson.117) The Gell-Mann–Nishijima relation can then be succinctly expressed as 1 Q = y + t3 . for example. − . 2 3 1 1 − .11. 2 (12. The three quarks in the fundamental representation can now be written as u = d = s = 1 1 .114) Figure 12.

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

- Mathematical Methods Quantum Mechanicsby mojicap
- (Major American Universities Ph. D. Qualifying Questions and )Yung-Kuo Lim, Chung-Kuo K'O Hsueh Chi Shu Ta Hsueh Physics Coaching Class-Problems and Solutions on Quantum Mechanics -World Scientific Puby Ccny Cuny
- Solution Manual String Theoryby 邱泰尹
- Wick Contractions in Latexby sthitadhi91

- Annihilation and Creation Operators.Chapter 8
- Relativistic Quantum Mechanics
- 51891292 Quantum Mechanics I
- Quantum Mechanics Lecture
- Advanced Quantum Mechanics
- Merzbacher Quantum Mechanics
- Sakurai - Modern Quantum Mechanics Rev Ed- Solutions Manual
- Quanten_physics
- Quantum mechanics
- 611 Electromagnetic Theory II
- Relativistic Quantum Mechanics
- Classical Electrodynamics 3rd Ed J.D. Jackson - Solutions - 214 Pg
- Physics - Quantum Field Theory
- Kittel Charles - Introduction to Solid State Physics 8Th Edition - Solution Manual
- Quantum Mechanics for Engineers
- Sakurai quantum mechanics solutions 4
- 1222405266 Quantum Mechanics
- Cambridge.quantum.mechanics.apr.2009.eBook ELOHiM
- Quantum Mechanics Review
- Book
- Lectures on Quantum Mechanics - Gordon Baym
- Problems and Solutions Quantum Mechanics - Y. K. Lim, World Scientific (1998, 751p)
- Relativistic Quantum Mechanics-Bjorken j.d., Drell s.d.
- Peskin Chapter 4
- Mathematical Methods Quantum Mechanics
- (Major American Universities Ph. D. Qualifying Questions and )Yung-Kuo Lim, Chung-Kuo K'O Hsueh Chi Shu Ta Hsueh Physics Coaching Class-Problems and Solutions on Quantum Mechanics -World Scientific Pu
- Solution Manual String Theory
- Wick Contractions in Latex
- What is String Theory - J. Pol Chin Ski
- [Ashcroft & Mermin]Solid State Physics Solution
- Notes on Quantum Mechanics

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