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

· ] is called the diﬀerential of S[ ]. For this purpose we will review the relevant concepts of Classical Mechanics. Deﬁnition: A functional S[ ] is a map S[ ] : F → R . F = {q(t).1) from a space F of vector-valued functions q(t) onto the real numbers. there are 3N conﬁgurational coordinates. The linearity property above implies δS[q(t). 1.1 Basics of Variational Calculus The derivation of the Principle of Least Action requires the tools of the calculus of variation which we will provide now.. δq(t)] is linear in δq(t). δq(t) ∈ F. 1 (1. It is important to note at the outset that for the description of a d classical system it will be necessary to provide information q(t) as well as dt q(t). |δq(t)| < . δS[ · . often in the Cartesian coordinate system. The latter is the velocity vector of the system. · ] exists with the properties (i) S[q(t) + δq(t)] = S[q(t)] + δS[q(t). that along a path describing classical motion the action integral assumes a minimal value (Hamiltonian Principle of Least Action). . namely. t ∈ [t0 . qM (t)). q : [t0 . | a functional δS[ · . the position coordinates of the particles in any kind of coordianate system. Deﬁnition: A functional S[ ] is diﬀerentiable. t1 ] ⊂ R → RM . . ∀t. In case of N classical particles holds M = 3N . An important concept is that the equations of motion of Classical Mechanics can be based on a variational principle. δq2 (t)] .2) . δq(t)] + O( 2 ) (1. α1 δq1 (t) + α2 δq2 (t)] = α1 δS[q(t).e. namely. if for any q(t) ∈ F and δq(t) ∈ F where F = {δq(t). q2 (t). q(t) is called the trajectory of a system of M degrees of freedom described by the conﬁgurational coordinates q(t) = (q1 (t). q(t) diﬀerentiable} (1. t1 ] ⊂ R} dt (1.3) (ii) δS[q(t). i. δq1 (t)] + α2 δS[q(t). .Chapter 1 Lagrangian Mechanics Our introduction to Quantum Mechanics will be based on its correspondence to Classical Mechanics.4) d δq(t)| < .

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

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. ∂qj dt ∂ qj ˙ t0 j=1 (1. The length of such path can be determined starting from the fact that the incremental length ds in going from point (x. (1. M . y1 ) and (x2 . δq(t)] = dt − δqj (t) . δq(t)] = 0 for all δq(t) ∈ F . For this purpose we deﬁne Deﬁnition: An extremal of a diﬀerentiable functional S[ ] is a function qe (t) with the property δS[qe (t). dx (1. . Theorem: Euler–Lagrange Condition For the functional deﬁned through (1. 2. · ] ≡ 0 and. the above theorem. . . from (1. Let us assume that the two points are connected by the path y(x).13) dt f (t)h(t) = 0 t0 (1. The proof of the above theorem starts from (1. .16) This property holds for any δqj with δq(t) ∈ F .12) for j = 1.14) for any continuous function h(t) ∈ F with h(t0 ) = h(t1 ) = 0. 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 . t1 ]. hence. y(x)) to (x + dx. . it holds in case δq(t0 ) = δq(t1 ) = 0 that qe (t) is an extremal. . .12) for j = 1. .11) The extremals qe (t) can be identiﬁed through a condition which provides a suitable diﬀerential equation for this purpose. if and only if it satisﬁes the conditions (j = 1. y(x1 ) = y1 . t1 ] ⊂ R → R holds t1 (1. y(x2 ) = y2 . 2. (1.5) assume extreme values. 2.1: Variational Calculus 3 We now consider the question for which functions the functionals of the type (1. y(x + dx)) is ds = (dx)2 + ( dy dx)2 = dx dx 1+( dy 2 ) .1.16) the property δS[qe (t).5). then f (t) ≡ 0 on [t0 . y2 ).12) The proof of this theorem is based on the property Lemma: If for a continuous function f(t) f : [t0 . On the other side. . M ) d dt ∂L ∂ qj ˙ − ∂L = 0 ∂qj (1. M and δqj (t0 ) = δqj (t1 ) = 0 follows according to (1.17) . . This condition is stated in the following theorem. According to the Lemma above follows then (1.

t) = j=1 1 mj qj − U (q1 . The shortest path is an extremal of s[y(x)] which must. . 2. q(t). j = 1. circles whose centers lie at the center of the sphere. q(t). q(t). obey the Euler–Lagrange dy condition. t) (1.2 Lagrangian Mechanics The results of variational calculus derived above allow us now to formulate the Hamiltonian Principle of Least Action of Classical Mechanics and study its equivalence to the Newtonian equations of motion.19) From this follows y / 1 + (y )2 = const and. q2 . The constants a and b are readily identiﬁed through the conditons y(x1 ) = y1 and y(x2 ) = y2 .18) dy s is a functional of y(x) of the type (1. This in turn yields y(x) = ax + b. dx ) = 1 + (dy/dx)2 . . according to the theorems above. . t1 S[q(t)] = t0 ˙ dt L(q(t). One obtains y 1 − y2 y(x) = (x − x2 ) + y2 . dx (1. .5) with L(y(x). . . the so-called action integral. hence. . y = const.4 The total path length is then given by the integral x1 Lagrangian Mechanics s = x0 dx 1+( dy 2 ) . i.22) ˙ where L(q(t). 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 .23) Presently we consider only velocity–independent potentials. Using y = dx the condition reads d dx ∂L ∂y = d dx y 1 + (y )2 = 0. M (1.. t) is the so-called Lagrangian M ˙ L(q(t). .21) are extremals of the functional.e. ˙2 2 (1. (1. Velocity–dependent potentials which describe particles moving in electromagnetic ﬁelds will be considered below. (1.20) x1 − x2 Exercise 1. qM ) . 1.1.1: Show that the shortest path between two points on a sphere are great circles.

t) and B(r. These conditions read in the present case ∂L d ∂L ∂U d − = 0 → − − (mj qj ) = 0 ˙ (1.28) can be satisﬁed implicitly if one represents the ﬁelds through a scalar potential V (r.32) (1. t) obey the Maxwell equations ×E + ·B ×B − ·E 1 ∂ c ∂t E 1 ∂ c ∂t B (1. dt The ﬁelds E(r.31) = − V − Gauge Symmetry of the Electromagnetic Field It is well known that the relationship between ﬁelds and potentials (1. t) as follows B E = ×A 1 ∂A .30. The transformation which leaves the ﬁelds invariant is A (r. t). t) − K(r. t) . 1.27) (1.24) ∂qj dt ∂ qj ˙ ∂qj dt which are obviously equivalent to the Newtonian equations of motion. 1.30) (1. t) = V (r.22. t) + v × B(r. t) and a vector potential A(r. Equations (1. The equations of motion for such a particle are d q ˙ (mr) = F (r.23).26) (1. x3 (t)) moving in an electromagnetic ﬁeld described through the electrical and magnetic ﬁeld components E(r. t) 1∂ K(r.33) .1. t) + V (r.2: Lagrangian 5 For a proof of the Hamiltonian Principle of Least Action we inspect the Euler–Lagrange conditions associated with the action integral deﬁned through (1.27) and (1.31) allows one to transform the potentials without aﬀecting the ﬁelds and without aﬀecting the equations of motion (1.25) = 0 = 0 4π J c = 4πρ = (1. t) is the Lorentz force. F (r.25) of a particle moving in the ﬁeld.28) (1. Particle Moving in an Electromagnetic Field We will now consider the Newtonian equations of motion for a single particle of charge q with a trajectory r(t) = (x1 (t). x2 (t). t) = q E(r. c ∂t (1. t) describes the charge current density. respectively. t) describes the charge density present in the ﬁeld and J(r. t) = A(r. t) c ∂t (1.29) where ρ(r. t) and B(r. t) dt c where dr = v and where F (r.

r.37) is ∂L ∂V q = −q + ∂x1 ∂x1 c The ﬁrst term in (1.35) shows that the Newtonian equations of motion and the Euler–Lagrange conditions are.. 1. (1. 1. t) + A(r.25) follows from the Hamiltonian Principle of Least Action.35) is equivalent to the Euler–Lagrange condition d dt The second term in (1. t) = 1 q mv 2 − q V (r.40. B3 = ∂A2 ∂x1 (1.6 Lagrangian of Particle Moving in Electromagnetic Field Lagrangian Mechanics We want to show now that the equation of motion (1.25). in fact.30).41) with (1.41) which is identical to the term (1. (1. .37) The results (1. (1. dt ∂x1 c We notice using (1.40) (1. equivalent.35) − ∂A1 ∂x2 [v × B]1 = x2 B3 − x3 B2 = x2 ˙ ˙ ˙ ∂A2 ∂A1 − ∂x1 ∂x2 − x3 ˙ ∂A1 ∂A3 − ∂x3 ∂x1 .39) ∂A1 ∂A2 ∂A3 x1 + ˙ x2 + ˙ x3 ˙ ∂x1 ∂x1 ∂x1 . d ∂V q (mv1 ) = F1 = −q + [v × B]1 . Comparing then (1. namely. if one assumes for a particle the Lagrangian ˙ L(r.38) ∂L ∂ x1 ˙ − ∂L = 0.39) 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 (1.34) For this purpose we consider only one component of the equation of motion (1.g.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 . e.38. 2 c (1. ∂x1 (1.36) in the Newtonian equation of motion. t) · v .36) This expression allows us to show that (1.

33) of electromagnetic potentials. t) = mv 2 − q V (r. t) 1 1 ∂K q mv 2 − q V (r. t) c t1 t0 = S[q(t)] + (1.33) ˙ L (r.44) Since the condition δq(t0 ) = δq(t1 ) = 0 holds for the variational functions of Lagrangian Mechanics. 1. t) + dt c (1. t) − + 2 c ∂t c 1 q = mv 2 − q V (r.42) This transformation is termed gauge transformation. t) + A(r. r.42) is. equivalent to the gauge transformation (1. This term is d ∂K ∂K ∂K ∂K ∂K K(r. t) L (q. r. t) · v .34) q 1 q d ˙ L (r.43) into (1. q. i.44) implies that the gauge transformation amounts to adding a constant term to the action integral. 1. in fact.33) of electyromagnetic potentials. For this purpose we consider the transformation of the single particle Lagrangian (1. We will consider in the following two symmetries.43) To prove this theorem we determine the action integral corresponding to the transformed Lagrangian t1 S [q(t)] = t0 ˙ dtL (q. t) = x1 + ˙ x2 + ˙ x3 + ˙ = ( K) · v + . appear in a diﬀerent. t) . t) · v + K(r. They are the subject of the following theorem. Note that one adds the total time derivative of a function K(r. dt ∂x1 ∂x2 ∂x3 ∂t ∂t (1.32. q.45) and reordering terms yields using (1. Theorem: Gauge Transformation of Lagrangian The equation of motion (Euler–Lagrange conditions) of a classical mechanical system are unaﬀected by the following transformation of its Lagrangian d q ˙ ˙ K(q. q. t) = L(q. (1. 1. t) = q K(q. t) + A (r. Eq.3: Symmetry Properties 7 1.1.e. q. encountered above in connection with the transformations (1.32. The factor q has been introduced to make this c transformation equivalent to the gauge transformation (1. 1. t) + q K(q.33) of electromagnetic potentials.46) (1. We want to demonstrate now that the transformation (1. 2 c c dt Inserting (1. The gauge symmetry. t) + K ·v (1. t) c t1 t0 t1 t0 ˙ dtL(q.3 Symmetry Properties in Lagrangian Mechanics Symmetry properties play an eminent role in Quantum Mechanics since they reﬂect the properties of the elementary constituents of physical systems. 2 c = A(r.32. t) the Lagrangian. gauge symmetry and symmetries with respect to spatial transformations. and since these properties allow one often to simplify mathematical descriptions.45) .32. surprisingly simple fashion in Lagrangian Mechanics. a term not aﬀected by the variations allowed.. One can conclude then immediately that any extremal of S [q(t)] is also an extremal of S[q(t)].

for the connection between invariance properties and constants of motion. for example. The corresponding inﬁnitesimal transformation is deﬁned for small r (r. therefore. ˆ ˆ ˆ Examples of Inﬁnitesimal Transformations The beauty of inﬁnitesimal transformations is that they can be stated in a very simple manner. R(r) = through ∂r ∂ (1.. r ∈ R . ) = r + R(r) + O( 2 ) . which has an important analogy in Quantum Mechanics. in the case of central force ﬁelds which are invariant with respect to rotations of coordinates around the center.49) =0 ∈ R (1. i. In case of a translation transformation in the direction e nothing new is gained. r.42) corresponds to replacing in the Lagrangian potentials V (r. Such invariance properties are very familiar. t) by gauge transformed potentials V (r. the reason being that our interest lies in achieving familiarity with the principles (just mentioned above ) of symmetry properties rather than in providing a general tool in the context of Classical Mechanics. 0) = r .33). Deﬁnition: Inﬁnitesimal One-Parameter Coordinate Transformations A one-parameter coordinate transformation is decribed through r = r (r. However. t). we like to ˆ provide the transformation here anyway for later reference r = r + e.50) A non-trivial example is furnished by the inﬁnitesimal rotation around axis e ˆ r = r + e×r . the transformation (1.8 Lagrangian Mechanics Obviously. but nevertheless instructive way considering rotations around the x3 –axis.47) (1. We consider now invariance properties connected with coordinate transformations. t).51) We would like to derive this transformation in a somewhat complicated.e. for such vectors holds a · a = 1. ) . A (r.32. ˆ (1. The following description of spatial symmetry is important in two respects. A(r. 1.42) and (1. t). The transformations we consider are the most simple kind. We have proven. and for the introduction of inﬁnitesimal transformations which will provide a crucial method for the study of symmetry in Quantum Mechanics. where the origin of is chosen such that r (r. the equivalence of (1. ˆ (1. In this case the transformation can be written in matrix form cos −sin 0 x1 x1 x2 = sin cos 0 x2 (1. The transformations considered are speciﬁed in the following deﬁnition.48) In the following we will denote unit vectors as a.52) x3 0 0 1 x3 .

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

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

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

> t1 > t0 is φ(r.e. This is diﬀerent from the classical situation where the particle follows a discrete path and. . t ) = δ(r − r ) Ω (2. t0 ) = Ωd 3r N −1 Ωd 3r N −2 ··· Ωd 3r 1 φ(r. t ∈ [t0 .9) for N → ∞. t = )|2 = 1 . . t) = Ω d3 r φ(r.8) 7. This must hold for any ψ(r .8) a knowledge of the state at all space points r ∈ Ω at some intermediate time t . 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. requires then according to (2. t ) ψ(r . t|r . t ∈ [t0 . (2.e. i.11) which denotes an integral over all paths r(t) with end points r(t0 ) = r0 and r(tN ) = rN . t1 ] yields an expression of the form r(tN )=rN φ(r.5) 6.6) φ(r. t0 ) . t ) t > t . t0 ) = δ(r − r0 ). ψ(r0 . t1 ] (2. t|r . t |r0 . t|r0 . t1 ]) Ωd Ωd 3r Ωd 3r Ωd 3r 3 r |ψ(r. The deﬁnition will actually assume an inﬁnite number of intermediate times and express the path integral through integrals of the type (2. This symbol will be deﬁned further below. namely the point on the classical path at time t . t1 ]) d3 r φ(r. t|r . i. The generalization of the completeness property to N − 1 intermediate points t > tN −1 > tN −2 > . t). t ) ψ(r . t|r . Since (2. t ) Ωd 3 r |ψ(r. t ∈ (2. The state of a system at time t.4) holds for all times. the propagator is unitary.. t) is described by a linear map of the type ψ(r. t| rN −1 .. t|r0 . hence.7) t. the path integral r(tN )=rN d[r(t)] · · · r(t0 )=r0 (2. tN −1 |rN −2 . t ) φ(r . t|r0 . at any intermediate time the particle needs only be known at one space point.10) We have introduced here a new symbol. t1 |r0 . t|r . t)|2 = (2. (2. 8. t. t )φ(r. t ) which requires d3 r φ(r. described by ψ(r. tN −1 ) φ(rN −1 . t0 ) = φ(r. t.12 Quantum Mechanical Path Integral 5. t ) φ(r. tN −2 ) · · · φ(r1 .9) Employing a continuum of intermediate times t ∈ [t0 . The time evolution of ψ(r. . t|r . t0 ) = r(t0 )=r0 d[r(t)] Φ[r(t)] . t ) ψ(r . The following so-called completeness relationship holds for the propagator (t > t [t0 . (t > t .

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. in complete distinction from Classical Mechanics.13) and = 1. (2. t) (2.2: Axioms 9.12) S[r(t)] = t0 ˙ dt L(r.e. t) is the Lagrangian of the classical particle. 2. To show this we consider the value of the action integral for a particle of mass m = 1 g moving over a distance of 1 cm/s in a time period of 1 s. . 2 2 (2. 2. Since this number is multiplied by ‘i’. However. Situations which are well described classically will be distinguished through the property that the classical path gives the dominant. a very large number. r.14) has the same dimension as the action integral S[r(t)].e. This implies that small variations of the path near the classical path alter the value of the action integral by very little. such that destructive interference of the contributions of such paths does not occur. interferences should be much less dramatic than in case of the macroscopic particle. for the electron many paths contribute and the action integrals for non-classical paths need to be known. The functional Φ[r(t)] in (2.2. i. the exponent is a very large imaginary number.13). i. Only for paths close to the classical path is such interference ruled out. r. actually often essentially exclusive. 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. namely due to the property of the classical path to be an extremal of the action integral..12. Its value is extremely small in comparision with typical values for action integrals of macroscopic particles. The situation is very diﬀerent for microscopic particles.11) is Φ[r(t)] = exp where S[r(t)] is the classical action integral tN 13 i S[r(t)] (2. accordingly. However.5 · 1027 .0545 · 10−27 erg s . Scl / ≈ 8. for microscopic particles like the electron this is by no means the case. not only for the classical path.13) L(r.15) The exponent of (2.13) are built on action integrals for all possible paths. The constant given in (2. One would still expect signiﬁcant descructive interference between contributions of diﬀerent paths since the value calculated is comparable to 2π.12. it is comparable to action integrals as they arise for microscopic particles under typical circumstances. The value of S[r(t)] is then Scl = 1 1 m v 2 t = erg s . expressions (2.14) ˙ In (2. However. In case of a proton with mass m = 1. However..12) is then Scl / ≈ 0. 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. contribution to the path integral (2.

t0 ). provides an avenue to evaluate path integrals. The spacings between the times tj+1 and tj will all be identical.20) can be properly taken. (2.3 How to Evaluate the Path Integral In this section we will provide an explicit algorithm which deﬁnes the path integral (2. x. namely 1 m(xj+1 − xj )2 / 2 and U (xj ). They can be anywhere in the allowed volume which we will choose to be the interval ] − ∞. respectively.12. t0 ) = limN →∞ CN exp i N N −1 j=0 +∞ +∞ +∞ −∞ dx1 −∞ dx2 . however. For the sake of simplicity we will consider the case of particles moving in one dimension labelled by the position coordinate x.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 −1 ). the xj values are not. (xN −1 . tj ) to the next (xj+1 . a series of times tN > tN −1 > tN −2 > .20) Here. . tN |x0 .21) N 2. etc.19) The main idea is that one can replace the path integral now by a multiple integral over x1 .13) and.14 Quantum Mechanical Path Integral 2.18) The time instances are ﬁxed. In passing from one space–time instance (xj . namely tj+1 − tj = (tN − t0 )/N = N . . t) = ˙ 1 mx2 − U (x) . (2.16) In order to deﬁne the path integral we assume. 2 2m N (2.10) and (2. −∞ dxN −1 (xj+1 −xj )2 1 − U (xj ) . . t1 ). The particles have associated with them a Lagrangian L(x. (xN . ˙ 2 (2.4 Propagator for a Free Particle As a ﬁrst example we will evaluate the path integral for a free particle following the algorithm introduced above. This allows us to write the evolution operator using (2. as in (2. tj+1 ) we assume that kinetic energy and potential energy are constant. (x1 . > t1 > t0 letting N go to inﬁnity later. . . Its value is CN = m 2πi N 2 (2. 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 . 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. . x2 . . at the same time. tN )} . 2. ∞[. (2.12) φ(xN . .9).

10. x(tN ) = xN ] 1 (xN − x0 )2 m + 0 + 2 tN − t 0 + S[x(t)|x(t0 ) = 0. One obtains S[x(t)|x(t0 ) = x0 . t0 ) and (xN . can be expressed through a path integral with endpoints x(t0 ) = 0. tN |x0 . tN ) is to be evaluated. x(tN ) = 0] .22) where xcl (t) is the classical path which connects the space–time points (x0 .31) We have denoted explicitly that the action integral for a path connecting the space–time points (x0 . .12) yields x(tN )=0 im (xN − x0 )2 i φ(xN . 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.30) d[x(t)] exp 2 tN − t0 x(t0 )=0 a result. it is more suitable to deﬁne new integration variables yj .h. namely. tN |x0 . x(tN ) = xN ] is S[x(t)|x(t0 ) = x0 . which can also be written φ(xN . tN ). t0 ) (2. x(tN ) = 0. To see the beniﬁt of such approach we deﬁne a path y(t) as follows x(t) = xcl (t) + y(t) (2. of (2.. tN |0.23) tN − t0 It is essential for the following to note that.29) This expression corresponds to the action integral in (2.e.13). due to (2. Inserting the result into (2.27) The third term can be written in the notation introduced tN 1 dt my 2 = S[x(t)|x(t0 ) = 0. tN dt y = y(tN ) − y(t0 ) = 0 . ˙ 2 t0 (2. xN − x0 xcl (t) = x0 + ( t − t0 ) . The ˙ 1 S[x(t)|x(t ) = x . x(tN ) = 0] .24) implies for the second term on the r. t0 ) = exp S[x(t)] (2.25) is.26) The ﬁrst term on the r.23). the origin of which coincides with the classical path of the particle.2. 2 2 tN − t0 (2.25) + The condition (2.28) i. ˙ t0 (2. The resulting expression for S[x(t)|x(t0 ) = x0 . = (2.s. 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 .4: Propagator for a Free Particle 15 Rather then using the integration variables xj . (2. (2. t0 ) = exp 1 im (xN − x0 )2 2 tN − t0 φ(0.24). t0 ) and (xN .24) Also we use the fact that the velocity of the classical path xcl = (xN −x0 )/(tn −t0 ) is constant. since x(t0 ) = xcl (t0 ) = x0 and x(tN ) = xcl (tN ) = xN . ˙ ˙ ˙ 2 + y2) = ˙ (2. it holds y(t0 ) = y(tN ) = 0 . using (2. tN dt t0 1 1 (xN − x0 )2 m x2 = m ˙ cl . 2.h.s.

31) for a free particle one needs to evaluate the following path integral φ(0.34) of (ajk ) deﬁning a new matrix (Ajk ) through ajk = Using det(ajk ) = m 2 N N −1 m 2 N Ajk .. . 0 0 0 0 0 0 The following integral +∞ +∞ (2.. . 0 0 .32) we split oﬀ the factor 2 mN in the deﬁnition (2. . .k=1 yj ajk yk − 1) × (N − 1) matrix . 2 −1 −1 2 (2. In order to complete the evaluation of (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.. In the appendix we prove +∞ +∞ d j.16 Evaluation of the necessary path integral Quantum Mechanical Path Integral To determine the propagator (2.37) det(Ajk ) .k=1 yj ajk yk (2. t0 ) = limN →∞ × +∞ +∞ −∞ dy1 · · · −∞ dyN −1 exp i N m 2πi N 2 N × (2.33) where ajk are the elements of the following symmetric (N 2 −1 0 −1 2 −1 2 m 0 −1 ajk = .. . symmetric matrix (bjk ) and det(bjk ) = 0. 0 0 ... .32) (yj+1 − yj )2 N −1 1 2 j=0 2 m N The exponent E can be written.38) ..34) dy1 · · · −∞ −∞ dyN −1 exp i N −1 j. (2. . . . (2.. 2 N . . . real. .36) which holds for a d-dimensional.35) must be determined. noting y0 = yN = 0. tN |0. . 0 0 .. .k=1 dy1 · · · −∞ −∞ dyN −1 exp i yj bjk yk = (iπ)d det(bjk ) 1 2 . (2.

Using D1 = |(2)| = 2 . To solve this three term recursion relationship one needs two starting values.18) we obtain φ(0.. . deﬁning t = tN . .31) and obtain... (2.45) Expressions for Free Particle Propagator We have now collected all pieces for the ﬁnal expression of the propagator (2. .4: Propagator for a Free Particle a property which follows from det(cB) = cn detB for any n × n matrix B. . tN |0.41) the same result for D2 .42) m 2πi NN 1 2 (2. D3 . n = 1. .. (2. . . t|x0 .41) D2 = 2 −1 −1 2 = 3 (2. . det(Ajk ) (2. . 0 0 0 −1 2 . . Inserting this into (2. let say n.40) . (2. . we obtain φ(0. . D1 = 2 and obtain from (2.44) = tN − t0 . tN |0. .. 0 0 0 . one can readily verify Dn = n + 1 . Our derivation has provided us with the value det(Ajk ) = N . (2. .2.. tN |0. t0 ) = limN →∞ m 2πi N 2 17 N 2πi m N N −1 2 1 ..46) . . presently N − 1. . t0 ) = m 2πi (t − t0 ) 1 2 exp im (x − x0 )2 2 t − t0 .40) in terms of subdeterminants along the last column. . 0 0 −1 2 −1 .. x = xN φ(x. t0 ) = m 2πi (tN − t0 ) 1 2 . .39) In order to determine det(Ajk ) we consider the dimension n of (Ajk ). . ..39) yields φ(0. 2. . 2 −1 0 0 0 −1 2 For this purpose we expand (2. . . One can readily verify that this procedure leads to the following recursion equation for the determinants Dn = 2 Dn−1 − Dn−2 . variable. 0 0 Dn = . . which follows from (2. t0 ) = limN →∞ and with NN (2.43) We like to note here for further use below that one might as well employ the ‘artiﬁcial’ starting values D0 = 1. We seek then to evaluate the determinant of the n × n matrix 2 −1 0 .

51) as follows Eo (xo . centered around x0 = 0. t) of a free particle. t0 )| = 2 1 πδ 2 1 2 exp − x2 0 δ2 (2. One obtains then for the propagator (2.10) φ(r. Below we will apply this to a particle at rest and a particle forming a so-called wave packet.247).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 . 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 .18 Quantum Mechanical Path Integral This propagator. x) + E(x) in (2. The result (2. We need to evaluate for this purpose the integral ψ(x. t0 ) = 1 πδ 2 1 4 exp − x2 po 0 +i x 2 2δ (2. the associated probability distribution |ψ(x0 . thereby.48) Obviously. and describes a single particle since 1 πδ 2 1 2 +∞ dx0 exp −∞ − x2 0 δ2 = 1. x) + E(x) (2.46) can be generalized to three dimensions in a rather obvious way.5) to (2.50) One refers to such states as wave packets. (2. t|r0 .53) . (2.46).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 (2.49) is Gaussian of width δ. We will obtain. We want to apply axiom (2. the wave function of the particle at later times. t0 ) = m 2πi (t − t0 ) 3 2 exp im (r − r0 )2 2 t − t0 . x) = im 2 t x2 1 + i o t mδ 2 − 2xo x − po t m + f (x) (2.52) and applying (2.48) as the initial state using the propagator (2. We devide the contributions to the exponent Eo (xo . according to (2.5) allows us to predict the time evolution of any state function ψ(x.

58) The integrand is an analytical function everywhere in the complex plane and we can alter the integration path. that the new path does not lead to additional contributions to the integral. according to (2. 2 (2. (2.53) E(x) = x− po mt 19 (2.52).56) m 2πi t 1 2 +∞ eE(x) −∞ dx0 eEo (xo .2. t) = 1 πδ 2 1 4 2 t 1 + i mδ2 . We proceed as follows.51) ψ(x.55) im xo 2 t 1+i t − mδ 2 x− 1+ po m t t i mδ2 .x) (2. x) = One can write then (2.59) An integration path in the complex x0 –plane along the direction i 1−i t mδ 2 (2.4: Propagator for a Free Particle im x2 − f (x) .54) f (x) = This yields Eo (xo . We consider a transformation to a new integration variable ρ deﬁned through i 1−i t mδ 2 ρ = x0 − x− 1+ po m t t i mδ2 . 2 t One chooses then f (x) to complete. z2 = +∞ × i 1−i t mδ 2 (2.61) . the square in (2. making certain. (2. (2.x) im xo 2 t t − mδ 2 x− 1+ x− 1+ po m = −∞ +∞ dx0 exp 1+i t 2 2 t i mδ2 = −∞ im dx0 exp 2 t t 1+i 2 mδ xo − po m t t i mδ2 .57) and needs to determine the integral +∞ I = −∞ +∞ dx0 eEo (xo . however.60) is then represented by real ρ values. The beginning and the end of such path are the points z1 = −∞ × i 1−i t mδ 2 .

In fact.55). t) = Seperating the phase factor 1 t 1 + i mδ2 1 4 1 2 exp [ E(x) ] . We can conclude then that (2. (2.57) reads then 1 πδ 2 1 4 2πi t . Since the exponent in (2.64) ψ(x. using a ab = a − . z2 = +∞ . Hence.58) along the path z1 → z1 and along the path z2 → z2 gives only vanishing contributions one can replace (2. the paths between z1 → z1 and z2 → z2 have a real part of x0 of ±∞.58) vanishes 0 for Re x0 → ±∞.70) .68) and. One obtains i po i po x x2 2m t E(x) = − 2 + − t t t 2δ (1 + i mδ2 ) 1 + i mδ2 1 + i mδ2 2 (2.54) using (2.20 whereas the original path in (2.67) We need to determine ﬁnally (2.63) which can be readily evaluated.62) If one can show that an integration of (2.66) yields ψ(x. (2. I = Equation (2.58) has a leading contribution in x0 of −x2 /δ 2 the integrand of (2.58) 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. t m(1 + i mδ2 ) (2. 1+b 1+b ﬁnally E(x) = − (x − po m (2.65) t 1 − i mδ2 t 1 + i mδ2 . 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) ] . Quantum Mechanical Path Integral (2. one can show that z1 lies at −∞ − i × ∞ and z2 at +∞ + i × ∞. accordingly. (2.63) holds and.58) has the end points z1 = −∞ .

48) and for the ﬁnal state (2.71) × exp − (x − po m t)2 ) 2δ 2 (1 + 2 t2 m2 δ 4 (1 − i . (2.71).f.4: Propagator for a Free Particle which inserted in (2.72) Comparision of Moving Wave Packet with Classical Motion It is revealing to compare the probability distributions (2. t) = exp i i p2 po o x − (t − to ) m 2m .76) From this we conclude that an initial wave function ψ(xo . (2.48)] to = 0. (2. In this case holds ψ(x. However.72).71) conserves the phase factor exp[i(po / )x] of the original wave function (2. The center of the distribution (2.75) i. the spatial dependence of the initial state (2. Another interesting observation is that the wave function (2.49). (2. t) = exp i po i p2 o x − t m 2m . t0 ) = exp i po i p2 o xo − to m 2m . We had assumed above [c. 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 . (2..74) which corresponds to (2. respectively. The conservation of this factor is particularly striking for the (unnormalized) initial wave function ψ(x0 .73) m δ increases with time. 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 (2.48) and that the respective phase factor is related with the velocity of the classical particle and of the center of the distribution (2. 2.75) ψ(x.77) . This yields. instead of (2.e.72) 2 t2 δ 1+ 2 4 (2.2. the case of arbitrary to is recovered iby replacing t → to in (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 . t0 ) = exp i po xo m . (2. This ‘spreading’ of the wave function is a genuine quantum phenomenon.72) for the initial state (2. coinciding at t = 0 with the width of the initial distribution (2.71. (2. The width of the distribution (2.72) moves in the direction of the positive x-axis with velocity vo = po /m which identiﬁes po as the momentum of the particle.67) provides the complete expression of the wave function at all times t ψ(x.49). The corresponding probability distribution is |ψ(x.48) for δ → ∞.74) remains invariant in time.72).

12.82) one obtains L(xcl + y.81) for an arbitrary path x(t) with end points x(t0 ) = x0 and x(tN ) = xN .84) .82) which describes the deviation from the classical path xcl (t) with end points xcl (t0 ) = x0 and xcl (tN ) = xN . ˙ ˙ = (2.22 becomes at t > to ψ(x.78) i. t) ˙ (2. 2. xcl + y(t). Such states play a cardinal role in quantum mechanics. 2 . 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 . t) = ˙ 1 1 mx2 − c(t)x2 − e(t)x . For this purpose we use d xcl y = ˙ ˙ (xcl y) − xcl y ˙ ¨ (2.. y(t)] which vanishes according to the Hamiltonian principle as discussed in Sect. t) = exp i Quantum Mechanical Path Integral po i p2 o x − t m 2m . y(t). (2.81) vanishes2 . the end points of y(t) are y(t0 ) = 0 Inserting (2. y(tN ) = 0 .80) For this purpose we need to determine the action integral S[x(t)] = t0 dt L(x. tN |x0 .85) We want to show now that the contribution of δL to the action integral (2. y(t). t) = L(xcl . 1. xcl . (2. xcl . t) ˙ L (y. t) + L (y.5 Propagator for a Quadratic Lagrangian x(tN )=xN We will now determine the propagator (2. In order to simplify this task we deﬁne again a new path y(t) x(t) = xcl (t) + y(t) (2. t) + δL ˙ ˙ ˙ ˙ where L(xcl . Obviously. 2.83) (2. x.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 .13) φ(xN . the spatial as well as the temporal dependence of the wave function remains invariant in this case. One refers to the respective states as stationary states.10. x. t0 ) = x(t0 )=x0 d[x(t)] exp i S[x(t)] (2. ˙ 2 2 tN (2.79) for a quadratic Lagrangian L(x.80) into (2.e.

. 2 N .s. . 2 . tN |0.91) (yj+1 − yj )2 1 2 2m N where cj = c(tj ). . 0 0 0 c3 . vanishes. . (2. . tN |0. . .80) yields for the classical path m¨cl + c(t) xcl + e(t) = 0 x (2. . . t0 ) = limN →∞ × +∞ +∞ −∞ dy1 · · · −∞ dyN −1 exp N i N N −1 j=0 m 2πi N 2 N × − 1 2 cj 2 yj (2. . . .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) . 0 0 0 c2 0 . tN |x0 . One can then express the propagator (2.87) According to (2. . cN −2 0 0 0 0 0 cN −1 (2.. y. tj = t0 + form j.79) i ˜ φ(xN . The discretization scheme adopted above yields in the present case ˜ φ(0. .92) where ajk are the elements of the following (N − 1) × (N − 1) matrix 2 −1 0 .83) the ﬁrst term on the r. 0 0 0 . . . .. 0 0 m ajk = ... . 0 0 0 . 2 −1 0 0 0 −1 2 c1 0 0 . tN |0. Applying the Euler–Lagrange conditions (1.32). .2. of (2. . ..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. 0 0 N 0 − . . t0 ) (2.s. .h. . .31) for the free particle propagator.. For the evaluation of φ(0. . t0 ) we will adopt a strategy which is similar to that used for the evaluation of (2. One can express the exponent E in (2. .89) where ˜ φ(0. . . also the second contribution on the r. . ..93) .k=1 yj ajk yk (2. . . x t0 (2. 0 0 −1 2 −1 .. . hence. tN |0. . . t0 ) = y(tN )=0 d[y(t)] exp y(t0 )=0 i tN dt L (y. .88) vanishes.88) and. .91) through the quadratic N −1 E = i j. . t0 ) = exp S[xcl (t)] φ(0.h. 0 0 0 −1 2 . .24) to the Lagrangian (2. t) ˙ t0 .

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

t) = n=0 1 ˜ exp[ − iω(n + 1 ) t ] φn (x) 2 n! (2. xo .188) identiﬁes the functions ψn (x.171) can be expanded in terms of e−inωt . t − to ) 1 ˜ exp[−iω(m + 1 ) to ] φm (xo ) 2 m! (2.183) where the expansion coeﬃcients are functions of x.172) is a function of t − to and deﬁning accordingly Φ(x. . In contrast to W (x.172).186) Fourier transform.182) Comparision with (2. of (2. to ) exp[−iω(n + 1 ) to ] φn (xo ) .171) ∞ W (x. . t|xo . 2 (2. t − to ) = φ(x. We want to inspect the consequences of the invariance property (2. 2. .171. one obtains for the r.184) 1 ˜ exp[−iω(n + 1 ) t] φn (x) 2 n! ∞ +∞ = m=0 −∞ dxo Φ(x.s. (2.172) I = exp 2 mω x e−iωt − e−2iωt − Quantum Mechanical Path Integral mω 2 1 x − iωt 2 2 . to ). Noting that the propagator (2.. t|xo .h. one can expand (2. t − to ) 1 ˜ φn (xo ) n! (2. t) +∞ −∞ dto 1 ˜ exp[−iω(m + 1 ) to ] φm (xo ) 2 m! (2. .172) in the form ∞ n=0 (2.185) Replacing t → t + to yields ∞ n=0 1 ˜ exp[−iω(n + 1 ) (t + to )] φn (x) 2 n! ∞ +∞ = m=0 −∞ dxo Φ(x.36 Altogether. but not of t. 2. We note that the factor exp(2 mω/ xe−iωt − e−2iωt ) in (2. t) = exp[−iω(n + 1 ) t] φn (x) as invariants under the 2 action of the propagator φ(x..172).147) in (2.187) or 1 ˜ exp[−iω(n + 2 ) t] φn (x) +∞ = −∞ ˜ dxo φ(x. to ) we express (2. xo . results in 1 ˜ exp[−iω(n + 1 ) t] φn (x) 2 n! +∞ = −∞ dxo Φ(x. xo .171) concludes the proof of (2. xo . which also exhibits such invariance.188) ˜ ˜ Equation (2. t|xo . n = 1. i.e. Accoordingly. 1 exp[iω(n + 2 ) to ] · · · . t).

The Hermite Polynomials ˜ The function (2. t) = exp[−iω(n + 1 ) t] Nn φn (x) .192) w(y. . (2. according to (2.2. characterizes the wave functions φn (x).195) zn Hn (y) n! ˜ φn (y) .196) where Hn (y) = ey 2 /2 (2. (2.183). This follows from ∂n w(y. 1.5: Propagator for a Quadratic Lagrangian 37 ˜ ˜ the functions ψn (x. To obtain ˜n (x) we simplify the expansion (2.198) ∂z n z=0 . . We will also argue that the functions ψn (x.183) reads then w(y.194). through expansion (2.191) (2. t) of the quantum mechanical harmonic oscillator ˜ ψn (x. we have identiﬁed then.194) = z2 n=0 ∞ 1 ∞ zn ˜ φn (y) n! (2. t)|2 = ˜n (x)|2 . For this purpose we introduce ﬁrst closed expressions for φ the new variables y z and write (2. t)|2 = Nn +∞ −∞ n = 0. z) z 2 e−y or w(y. ˜ |φ stationary wave functions ψn (x. Expression (2. Actually.197) The expansion coeﬃcients Hn (y) in (2. t) are associated with a time-independent probablity density |ψn (x. Expansion (2.189) ˜ dx φ2 (x) = 1 n (2.171). z) (2.197) are called Hermite polynomials which are polynomials of degree n which will be evaluated below.171) W (x. t) such that +∞ −∞ 2 dx |ψ(x.183). z) = Hn (y) (2. through the expansion coeﬃcients φn (x) in (2. z) = exp(2yz − z 2 ) .194) plays a central role for the Hermite polynomials since it contains. 2 Here Nn are constants which normalize ψn (x. z) = n=0 1 2 /2 1 2 /2 = = mω e−iωt x (2.183). in a ‘nutshell’ all information on the Hermite polynomials. . t) provide all stationary states of the quantum mechanical harmonic oscillator. 2. t) = z 2 e−y where w(y.190) ˜ is obeyed. In the following we will characterize the functions φn (x) and determine the normalization constants Nn .193) (2.

H1 (y) = 2y.Wilf (Academic Press. Reading.1: Schematic representation of change of summation variables ν and µ to n = ν + µ and m = ν −µ.196) ∂n w(y.38 Quantum Mechanical Path Integral which is a direct consequence of (2..202) generatingfunctionology by H.Graham. i.198) for the Hermite polynomials can be expressed in a more convenient form employing deﬁnition (2.196) to ˜ derive.. ν+µ ν H3 (y) = 8y 3 − 12y.200) yields for the ﬁrst Hermite polynomials H0 (y) = 1.Knuth. As will become evident in the present case generating functions provide an extremely elegant access to the special functions of Mathematical Physics3 . ∂y n (2. among other properties. 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. Massachusetts. One calls w(y. (2. 1990) is a useful introduction to this tool as is a chapter in the eminently useful Concrete Mathematics by R.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). m lattice (right diagram). .e. The diagrams illustrate that a summation over all points in a ν. z) the generating function for the Hermite polynomials.201) ν−µ µ µ ν−µ Figure 2. The diagrams also identify the areas over which the summation is to be carried out. We will employ (2. D.196) results in the so-called Rodrigues formula for the Hermite polynomials Hn (y) = (−1)n ey 2 ∂ n −y2 e . We want to derive now explicit expressions for the Hermite polynomials. z) ∂z n = = z=0 2 ∂ n 2 y z − z2 e ∂z n e−(y−z) 2 ey z=0 2 ∂ n −(y−z)2 e ∂z n 2 z=0 2 (−1)n ey ∂n ∂y n = (−1)n ey z=0 ∂n ∂y n e−y (2. The identity (2. Boston.E. ν+µ (2. 2. that Hn (y) is a polynomial of degree n.S. For this purpose we expand the generating function (2. Inc.196). normalization factors for φ(y).194. and recursion equations for the eﬃcient evaluation of Hn (y).200) One can deduce from this expression the polynomial character of Hn (y). closed expressions for Hn (y).L.199) Comparision with (2. and O. .Patashnik (AddisonWesley. . 1989). µ lattice (left diagram) corresponds to a summation over only every other point in an n. ν H2 (y) = 4y 2 − 2. .

207) contains only even powers.206) From this expansion we can identify Hn (y) [n/2] Hn (y) = k=0 (−1)k n! (2y)n−2k . (2. 39 (2. −z) and.2. m must be restricted such that either both n and m are even or both n and m are odd.209) This property follows also from the generating function. According to (2.205) Since (n − m)/2 is an integer we can introduce now the summation variable k = (n − m)/2 . 2 (2. otherwise. m = ν − µ 0 ≤ n < ∞. p ≤ x. 2 µ = n−m . µ expressend in terms of n. according to 2.203) The old summation variables ν. is a polynomial of degree n.201). for odd n. µ are integers the summation over n. Hence.208) which agrees with the expressions in (2. . hence. in fact.209). the sum in (2. k! (n − 2k)! = Hn (y) (2. From (2.e. 0 ≤ k ≤ [n/2] where [x] denotes the largest integer p. i. The lattices representing the summation terms are shown in Fig. 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 . With this restriction in mind one can express (2. k! (n − 2k)! (2.197) ∞ n=0 zn Hn (−y) = n! ∞ n=0 (−z)n Hn (y) = n! ∞ n=0 zn (−1)n Hn (y) n! (2. (2.204) Since ν. . (2. .194) holds w(−y.207) This expression yields for the ﬁrst four Hermite polynomials H0 (y) = 1. m are ν = n+m . . it contains only odd powers. H1 (y) = 2y.1.5: Propagator for a Quadratic Lagrangian and introduce now new summation variables n = ν + µ. 2.207) one can deduce that Hn (y). it holds Hn (−y) = (−1)n Hn (y) . H2 (y) = 4y 2 − 2. z) = w(y..210) from which one can conclude the property (2. H3 (y) = 8y 3 − 12y.202) e 2yz − z 2 ∞ = n=0 zn n! ≤n m≥0 n! (−1) n−m 2 n−m 2 ! m! (2y)m . In case of even n . 0 ≤ m ≤ n .

40 Quantum Mechanical Path Integral The generating function allows one to determine the values of Hn (y) at y = 0. (2.197) above and given by (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. 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) . One can then readily state which diﬀerential equation of w(y. indeed satisﬁes the latter relationship. . have a node at y = 0.196) into the diﬀerential equation (2. z) ∂ w(y. 2. a property which is consistent with (2. The reader may verify that w(y. 1. 2.213) which allow one to evaluate Hn (y) from H0 (y) and H1 (y) given by (2. dw/dy − 2zw = 0. 2.211) m! n! m=0 n=0 Comparing terms on both sides of the equation yields (2n)! .218) The reader should recognize the connection between the pattern of the diﬀerential equation (??) and the pattern of the recursion equation (??): a diﬀerential operator d/dz increases the order n of Hn by one. (2.208). namely.194). . a factor z reduces the order of Hn by one and introduces also a factor n. n = 0. z) = 0 (2.209) since odd functions have a node at the origin. z). n = 1. (2. Hn+1 (y) − 2y Hn (y) + 2n Hn−1 (y) = 0 . n = 1. . . For Hermite polynomials holds. .233) below. (2. H2n (0) = (−1)n Recursion Relationships A useful set of properties for special functions are the so-called recursion relationships. n! (2. 2. . . Hn+1 (y) − 2y Hn (y) + 2n Hn−1 (y) = 0. For this purpose one considers w(0. 2. .217) gives H1 (y) − 2y H0 (y) = 0. for example.188.194). dy n = 1. as given in (2.212) n! This implies that stationary states of the harmonic oscillator φ2n+1 (x). Another relationship is d Hn (y) = 2n Hn−1 (y). Substituting expansion (2. H2n+1 (0) = 0 . as deﬁned through (2.214) We want to derive (??) using the generating function. z) should be equivalent to the relationship (??). . (2.215) yields ∞ n=1 z n−1 Hn (y) − 2y (n − 1)! ∞ n=0 zn Hn (y) + 2 n! ∞ n=0 z n+1 Hn (y) = 0 . Starting point of the derivation is the property of w(y.215) ∂z which can be readily veriﬁed using (2. z) − (2y − 2z) w(y. .

For this purpose we consider the integral +∞ −∞ dy w(y. acording to the deﬁnition (2. .2. .s.219) which can be written I(y) = e−y 2 +∞ −∞ dte−(t−iy) = e−y √ 2 2 +∞ −∞ dze−z . 2 (2.228) .222) Employing this expression now on the r. 2 (2.200) yields Hn (y) = The identity (−1)n y2 √ e π +∞ dt −∞ dn 2iyt − t2 e .194. z ) e−y 2 +∞ = e2 z z = √ ∞ dy e−(y−z−z ) = .225) We want to derive from the generating function (2. 1. in the integral representation of the Hermite polynomials 2n (−i)n ey √ Hn (y) = π Orthonormality Properties 2 (2. 2 (2.h.226) π −∞ 2n z n z n n=0 n! Expressing the l.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 . 2. dy n (2.224) +∞ −∞ dt tn e2iy t − t . 2 n = 0. through a double series over Hermite polynomials using (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 .194. 2. (2.223) dn 2iy t − t2 2 e = (2 i t)n e2iy t − t n dy results.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! (2.221) 1 = √ π +∞ −∞ dt e2iyt − t . (2. 2.196) the orthogonality properties of the Hermite polynomials.s. 2 √ π e2 z z (2. of the Rodrigues formula (2.h.226a).247) for a = i one obtains I(y) = and.220) Using (2. ﬁnally. e−y 2 π e−y 2 (2.196) yields ∞ n. . z) w(y.

233) are presented for n = 0.190)] φn (y) = Nn e−y 2 /2 and for the normalization constants Nn follows from (2. 3.42 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 2.189.2. 4 in Fig. 2. in harmony with (??). 3. According to (2.231) √ π (2. (2.233) . 1. 3. 2. 3. ¡ ¢ ¡ -4 -2 0 2 4 y (2. 4. in agreement with our above discussions. Normalized Stationary States The orthonormality conditions (2.229) Hn (y) . (2. 2. 2.197) holds 2 ˜ φn (y) = e−y /2 Hn (y) . 4. One can also recognize that n is equal to the number of nodes of the wave function. (2. The normalized states are [c. One can recognize. the value of the wave function at y = 0 is positive for n = 0.230) (2.2: Stationary states φn (y) of the harmonic oscillator for n = 0. negative for n = 2 and vanishes for n = 1. 4 and odd for n = 1.228) allow us to construct normalized stationary states of the harmonic oscillator. that the wave functions are even for n = 0.228) +∞ 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. 1. Furthermore. 2.232) Hn (y) .f.

459] hydrogen atom.237) and where In denotes some constants.156. In case of the associated Laguerre (α) polynomials. denoted by P (x) and introduced in Sect. Other examples of orthogonal polynomials are the Legendre and Jacobi polynomials which arise in solving three-dimensional stationary Schr¨dinger equations. (10.. (2.236) where w(x) is a so-called weight function with the property w(x) ≥ 0. 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.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. w(x) = xα e−x . 1]. +∞[.238) where the space is endowed with the scalar product (f |g) = Ω dx f (x) g(x) w(x) = < ∞ .234) The explicit form of the stationary states of the harmonic oscillator in terms of the position variable x is then. w(x) = 0 only at a discrete set of points xk ∈ Ω (2. The Hermite polynomials are so-called orthonogal polynomials since they obey the conditions (2. and I = 2/(2 + 1). (2. Comparision with (2.228) shows that the orthonogality condition of the Hermite polynomials is in complience with (2. 5. ??? and eq.2.179] holds Ω = [−1. i.236 . 10. 2.e.233) and (2. 5 below [c. 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.231) of the wave functions diﬀers from that postulated in (2. 5. and In = 2n n! π.5: Propagator for a Quadratic Lagrangian 43 The normalization condition (2.189) φn (x) = √ 1 2n n! mω π 1 4 e− mωx2 2 Hn ( mω x) . Ω (2. 5. using (2.150 .. f.e. The orthogonal polynomials pn mentioned above have the important property that they form a complete basis in the space F of normalizable functions.151. In case of the Legendre polynomials. of functions which obey dx f 2 (x) w(x) = < ∞ . g ∈ F . w(x) ≡ 1.10 and eq. denoted by Ln (x) and encountered in case of the stationary states of the nonrelativistic [see Sect. holds Ω = [0.189) by the Jacobian dx/dy. i. w(x) = √ exp(−x2 ). In = Γ(n+α+1)/n! where Γ(z) is the so-called Gamma function. ???] and relativistic [see Sect.f.228). The latter are written for polynomials pn (x) in the general form dx pn (x) pm (x) w(x) = In δnm Ω (2. (2.237) for Ω = R. by dx = dy mω 1 4 . (5.239) .

233). a (2. ψ(y. t) which is stationary under the action of the harmonic oscillator propagator (2. (2. aT = a.233) exhaust all stationary states of the harmonic oscillator. symmetric a. 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. the states (2. . Therefore.239) becomes the conventional scalar product of quantum mechanics f |g = Ω dx f (x) g(x) . the 2 stationary state ψ(y. 2. hence.244) For the state to be stationary |ψ(x. n 2 (2.e.172) holds. t) is ∞ ψ(y. i. i.243) To be consistent with(2. t) = n=0 1 2 dn exp[−iω(n + )t] e−y /2 Hn (y) .e. . in particular..240) where cn = 1 In dx w(x) f (x) pn (x) . Ω (2.242) and the existence of a normalizable state ψ(y. det(a) (2.245) must be time-independent.e. ∞ n.147). Appendix: Exponential Integral We want to prove +∞ +∞ I = dy1 . The only possibility for this to be true is dn = 0. t) must be identical to one of the stationary states (2. except for a single n = no .246) for det(a) = 0 and real. In case of n = 1 this reads +∞ −∞ dx ei a x = 2 iπ . If one replaces for all f ∈ F: f (x) → w(x) f (x) and. t)|2 . Since the Hermite polynomials form a complete basis for such states we can expand ∞ ψ(y. 2 (2.197) it must hold cn (t) = dn exp[−iω(n + 1 )t] and.e. i.241) The latter identity follows from (2.188.m=0 d∗ dm exp[iω(m − n)t] e−y Hn (y)Hm (y) . −∞ dyn ei −∞ n j.k yj ajk yk = (iπ)n . a state for which (2.236). i.247) . pn (x) → w(x) pn (x) the scalar product (2... (2.242) Let us assume now the case of a function space governed by the norm (2.

248) .249) The akk are the eigenvalues of a and are real. det(S) a = (det(S))−1 det(a) det(S) = det(a) . . .252) ˜ For the determinant of a holds det(˜) = a n ajj ˜ j=1 (2. . . yk = k Skj yj . .m (S T )jl alm Smk . ˜ (2. i.k n m n y Sj ajk Skm ym ˜ ˜ y (S T ) j ajk Skm ym ˜ ˜ j..246) exploits that for any real.255) . . 0 ˜ 0 a22 . 2.246) reads then in terms of yj ˜ n n j.2. (2.k yj ajk yk n = j. One can conclude det(a) = j=1 n (2. 0 0 . (2. . ST S = 1 1 or S = S−1 . 0 ˜ ˜ S−1 a S = a = . ˜ (2. ann ˜ where S can be chosen as an orthonormal transformation. symmetric matrix exists a similarity transformation such that a11 0 .251) where. . .253) as well as det(˜) = det(S−1 aS) = det(S−1 ) det(a) .k yj ajk yk ˜˜ ˜ (2. . . . .7: Appendix / Exponential Integral 45 which holds for a ∈ C as long as a = 0. according to (2.k n m = = j.254) ajj . .e.248. (2. This property allows one to simplify the bilinear ˜ form n yj ajk yk by introducing new integration variables j. .k n n yj = ˜ k (S −1 )jk yk .249) n ajk = ˜ l. .250) The bilinear form in (2. The proof of (2..

263) (2. .266) . Accordingly. . yn ) y ˜ ei n k akk yk ˜ ˜2 . . .46 We have assumed det(a) = 0. . d˜n e y −∞ = (2. . hence. . yn ) y ˜ = ( det S )2 (2. . . . yn ) ) = det(S) .258) where we introduced the Jacobian matrix J = with elements Jjs = According to (2. (2. yn ) ∂(˜1 . .258) +∞ +∞ ∂(y1 .250) holds J = S and. . ∂ ys ˜ (2. . . i.261) ∂(y1 . .249) follows 1 = det ST S such that one can conclude det S = ±1 One can right then (2.256) such that none of the eigenvalues of a vanishes. 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 −∞ .e.264) I = d˜1 .260) (2. holds n Quantum Mechanical Path Integral ajj = 0 ˜ j=1 (2. . 2.. . .265) k=1 −∞ which leaves us to determine integrals of the type +∞ dx eicx −∞ 2 (2. .259) (2. ajj = 0 . yn ) ∂(˜1 . .. . ∂(˜1 . yn ) y ˜ ∂yj . . . .. det( From (2. . . . . . .257) Substitution of the integration variables (2. n (2. y −∞ d˜n det y −∞ ∂(y1 .250) +∞ +∞ I = d˜1 . .262) (2. ˜ for j = 1. .250) allows one to express (2.

Jk = γk dz eicz 2 (2. The contour 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 .3. (2.268) along the path γ = γ1 + γ2 + γ3 + γ4 displayed in Figure 2. c (2.7: Appendix / Exponential Integral 47 where.267) by considering the contour integral J = γ dz eicz = 0 2 (2. The contour intergral (2. One can relate integral (2.268) vanishes. i. the integrand does not exhibit any singularities anywhere in C..2.257) holds c = 0.271) .268) can be written as the sum of the following path integrals J = J1 + J2 + J3 + J4 . 2 since eicz is a holomorphic function.c > 0 . according to (2. We consider ﬁrst the case c > 0 and discuss the case c < 0 further below.266) to the well-known Gaussian integral +∞ dx e−cx −∞ 2 = π .e. for x. p ∈ IR.270) dx e−cx 2 2 − 2p √ γ4 : z = ix − p J4 = −p i dx eic(ix−p) .269) The contributions Jk can be expressed through integrals along a real coordinate axis by realizing that the paths γk can be parametrized by real coordinates x p γ1 : z = x γ2 : z = ix + p γ3 : z = √ ix J1 = −p p dx eicx 2 J2 = J3 = = i dx eic(ix+p) 0√ − 2p √ √ 2p 2 i dx eic( 2p √ ix)2 √ − i 0 √ (2.

267) one has shown then ∞ dx eicx −∞ 2 = iπ . p p→+∞ lim |J2 or 4 | ≤ = p→+∞ 0 lim dx |e−2cxp | 1 − e−2cp 2cp = 0.273) p→+∞ lim J2 and J4 do not contribute then to integral (2. (2.275) .3: Contour path γ in the complex plain. (2.272) = 0 . This follows from the following calculation p p→+∞ lim |J2 or 4 | = p→+∞ lim | 0 p i dx eic(ix+p) | ≤ It holds |e ic(p2 −x2 ) p→+∞ 0 lim |i| dx |eic(p 2 −x2 ) | = 1 since the exponent of e is purely imaginary. We will now show that the two integrals J2 and J4 vanish for p → +∞.274) (2. c ¡ -p γ1 p Re(z) 2 | |e−2cxp | .268) for p = +∞. Hence. (2.48 Quantum Mechanical Path Integral Im(z) ip γ2 ¡ γ4 ¡ γ3 ¡ -i p Figure 2. One can state accordingly ∞ J = dx e −∞ icx2 − √ ∞ i dx e−cx −∞ 2 Using 2.

but with a path γ that is reﬂected at the real axis.7: Appendix / Exponential Integral 49 One can derive the same result for c < 0.277) We apply the above results (2. 2. (iπ)n det(a) (2.275.276) and (c < 0) ∞ dx eicx −∞ 2 = −iπ = −|c| iπ .278) Noting (2.277) to (2. if one chooses the same contour integral as (2. This leads to ∞ icx2 J = dx e −∞ + √ −∞ −i ∞ dx ecx = 0 2 (2.255) this result can be expressed in terms of the matrix a I = which concludes our proof.2.268). c (2.265).279) . It holds n I = k=1 iπ = akk ˜ (iπ)n . n ˜ j=1 ajj (2.

50 Quantum Mechanical Path Integral .

x2 .5) ψ(r.1) can be expressed through the discretization scheme (2. (3.21). It holds then according to (2. of this equation in terms of powers of and we will demonstrate that the terms of order require that ψ(r. t) . namely the Schr¨dinger equation. The propagator in (3. (3. t) . t) ψ(r0 . Notable exceptions are non-stationary systems involving time-dependent linear and quadratic Lagrangians.h.s. t) = From this follows ψ(r. even in x1 . t) satisﬁes a partial diﬀerential equation.20. t) .2) d 3 r0 m 2πi 3 2 exp i m (r − r0 )2 − 2 U (r. t + ) = Ω d3 r0 φ(r.Chapter 3 The Schr¨dinger Equation o 3.20) to R one obtains then for small φ(r. t) ψ(r0 . the Schr¨dinger o o equation provides the simpler avenue towards describing quantum systems than the path ingral formulation of Section 2. and the r. We will denote the components of s by (x1 . t + |r0 .3) In order to carry out the integration we set r0 = r + s and use s as the new integration variable. 2. This yields ψ(r. t + ) = × exp i 2 +∞ +∞ +∞ −∞ dx1 −∞ dx2 −∞ dx3 m x2 + x2 + x2 1 2 3 m 2πi 3 2 × (3. (3.1) We will expand the l. x3 51 . For many situations. x3 )T .h. t + |r0 . t) by an inﬁnitesimal time step .20) to evaluate the propagator. x2 .1 Derivation of the Schr¨dinger Equation o We will consider now the propagation of a wave function ψ(r.s. t) . but by no means all. t + ) = Ω m 2πi 3 2 exp i m (r − r0 )2 − 2 U (r. In the limit of very small it is suﬃcient to employ a single discretization interval in (2. Generalizing (2. . t) ψ(r + s.4) − U (r.

5) assuming that only the leading terms contribute. t) + .10) It is now important to note that in case of integral (3.6) 3 4 ∂4 j=1 xj ∂x4 ψ(r.36) holds I0 (a) = Inspection of (3.4) only for small x2 + x2 + x2 values 1 2 3 we expand 3 ψ(r + s.k=1 ∂2 ψ(r.5) which are even separately in all three coordinates yield non-vanishing contributions. but over s = (x1 . . a I2 (a) = − 3 4a2 iπ . Since the kinetic energy contribution in (3.8) (3. 1 4 3 2 2 ∂4 j. 1 2 3 2 ∂2 j=1 xj ∂x2 ψ(r.11) and. of the expansion of ψ(r + s. t) is a constant with respect to this integration. a (3.. It is then suﬃcient to consider the terms ψ(r. 1. n = 0. i ∂a Starting from (3. (3. t) + j=1 xj ∂ 1 ψ(r.6) make contributions of the order O( 3 2 ) ..4) is even in all three coordinates x1 . t) j 1 12 .8) one can evaluate recursively all integrals In (a).k=1 xj xk ∂x2 ∂x2 ψ(r. t). 2 (3.11) holds 3 m 3 iπ 2 2 × = 1 (3. only terms of the expansion (3. x3 . (3. t) j k . t) j . x2 . The integration involves only the wave function ψ(r +s..7) shows 1 ∂ In (a).12) Here one needs to note that we are actually dealing with a three-fold integral. t) .13) 2πi a . O( 7 2 ) . t) + ∂xj 2 3 xj xk j. the terms collected in (3. x3 )T . O( 5 2 ) . t) = ψ(r. ∂xj ∂xk (3. t) and the kinetic energy term. Since the latter contributes to (3. we need then to evaluate integrals of the type +∞ In (a) = −∞ dx x2n exp i a x2 .7) According to (2. x2 .52 Schr¨dinger Equation o It is important to note that the integration is not over r.9) iπ a (3. e. O( 7 2 ).g. a supposition which will be examined below.. consequently. Obviously. . According to (3.. It holds In+1 (a) = I1 (a) = i 2a iπ .. U (r.4) one identiﬁes 2 1 = a m = O( ) (3.

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

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. The reader is advised to consult a reference text on ‘Linear Algebra’ to follow this argument. Jackson. Calssical Electrodynamics. Since energy is a real ˆ ˆ quantity one needs to require that the eigenvalues of the operator H are real and. that H is 2 . Chapter 1. 2nd Ed.22) and (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. D.28) See.54 Schr¨dinger Equation o linear vector spaces of functions f (r) in which a scalar product between two elements f. diﬀer from each other.20). g|r) (3. g ∈ H is deﬁned as follows f |g Ω = d3 rf ∗ (r)g(r) (3.23)..25) ˆ f | H |g Ω ˆ = g| H |f Ω + ∂Ω da · P (f ∗ . . New York. f ∗ (r) d3 rg(r) Ω 2 = − 2m d3 rf ∗ (r) Ω 2 g(r) − f ∗ (r) (3. (3.21) Ω This leads one to consider the integral f | H |g Ω = Ω ˆ d3 rf ∗ (r) H g(r) (3. Interchanging f ∗ (r) and g(r) results in g| H |f Ω = Ω ˆ d3 r g(r) Hf ∗ (r) . for example.22) ˆ where H is deﬁned in (3. J. t). 1975). implies hermitian ˆ f | H |g 1 2 Ω ˆ = g| H |f Ω (3. In (3. in principle. g|r) is called the ˆ concomitant of H and is P (f ∗ . hence. The hermitian property. (John Wiley.23) ˆ Since H is a diﬀerential operator the expressions (3. however.27) ˆ We will postulate below that H is an operator in H which represents energy.26) the vector–valued function P (f ∗ . t) g(r) − Ω d3 rg(r) U (r. 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.26) where ∂Ω da · A(r) denotes an integral over the surface ∂Ω of the volume Ω. g|r) = − 2 2m ( f ∗ (r) g(r) − g(r) f ∗ (r) ) (3.

i. In this case one can expect that the particle density is localized in the area where the energy of the particle exceeds the potential eenrgy.3 Particle Flux and Schr¨dinger Equation o The solution of the Schr¨dinger equation is the wave function ψ(r. It must hold then for all f ∈ H f (r) = 0 ∀r. i. we can only allow functions which make the diﬀerential da · P (f ∗ . in which case ∂Ω1 is the inner sphere and ∂Ω2 is the outer sphere. ..32) the wave function must decay for |r| → ∞ rapidly enough to be square integrable.30) Note that these boundary conditions are linear in f . t) which describes the state of o a particle moving in the potential U (r. The probability to observe the particle anywhere in the subvolume ω ⊂ Ω is p(ω. Often the closed surface of a volume ∂Ω is the union of disconnected surfaces3 ∂Ωj . (3.33) The time derivative of p(ω. A most common boundary condition is encountered for the volume Ω = R3 in which case one postulates lim f (r) = 0 “natural boundary condition” . and that the density decays rapidly when one moves away from that area.34) An example is a volume between two concentric spheres.e.29) In fact.29. Since the total probability of ﬁnding the particle anywhere in space is d3 r|f (r)|2 = 1 (3. ∂Ω = ∂Ω1 ∩ ∂Ω2 ∩ ∂Ω3 ∩ . in this case also all derivatives of f (r) vanish at inﬁnity.31) usually arises when a particle existing in a bound state is described. t)|2 . t) + ψ(r. e. t) = ω 3 d3 r [ ψ ∗ (r. The observable directly linked to the wave function is the probability to ﬁnd the particle at position r at time t.29.3: Particle Flux 55 and. to avoid discontinuities in ψ(r.32). 3. In this case one can postulate both conditions (3. t) on a single connected surface ∂Ωj only either one of the conditions (3. r ∈ ∂Ω or da · f (r) = 0 ∀r. t) ] . r ∈ ∂Ω (3. i.30) each condition holding on an entire surface ∂Ωj . . 3. t) is ∂t p(ω. namely. However. α > 2.. t). t) = ω d3 r |ψ(r. t)∂t ψ ∗ (r. (3. obey (3.g. t)|2 . (3.30) can be postulated. therefore. 3.e. In either case does f (r) and all of its derivatives vanish asymptotically. . t)∂t ψ(r. The latter property stems from the fact that the boundary condition (3.. g|r) vanish on ∂Ω.e. |ψ(r.3..31) |r|→∞ (3. like exp(−κr). if f and g satisfy these conditions than also does any linear combination αf + βg. κ > 0 or like r−α .

40) One can rewrite then (3. One can multiply the solution of (3.30) ∂t p(Ω. t) .26.42) 2mi Since (3. (3. t).41) where j(r. A natural choice for this constant is 1. t) − ψ(r. One refers to such solution as normalized.35) d3 r . Accordingly the probability to observe the particle anywhere in the total volume Ω is constant. t) = 0.56 Using (3. Using (3. t) + ω · j(r. t) such that d3 r|ψ(r. 3. t) = 0 (3. t) = [ ψ(r. t) ψ ∗ (r.37) d3 r ∂t ρ(r. t) . t)|2 is a probability density with units 1/volume. t) = |ψ(r. (3. (3. t) = ∂ω da · P (ψ ∗ (r.27) this can be written i ∂t p(ω. Note that for a normalized wave function the quantity ρ(r. (3. t) ] . t) ψ(r. t) = 0 . t) H ψ ∗ (r. t) H ψ(r.41) holds for any volume ω ⊂ Ω one can conclude ∂t ρ(r. We will assume in the remainder of this Section that the solutions discussed are normalized. t) (3.27) one can express this j(r. t) = P (ψ ∗ . t) + 4 · j(r. t)|2 = 1 Ω (3.18) by any complex number and accordingly one can deﬁne ψ(r. t) = ω Schr¨dinger Equation o ∂ ∗ ˆ ψ (r. t)r.44) ˆ Note that the Hamiltonian H involves only real terms. t) = H ψ ∗ (r.36) According to (3.29. t) − ψ ∗ (r. . The surface integral (3. ψ|r.37) can be expressed through a volume integral according to da · A(r) = ∂ω ω (3. 3.43) (3. ψ(r. t) ∂t ˆ ˆ ψ ∗ (r.38) holds.19) and its conjugate complex4 −i yields i ∂t p(ω.37) If one applies this identity to ω = Ω one obtains according to (3.39) d3 r · A(r) (3.

and for particles moving in a linear [c.. This ﬁnding is consistent with the present evaluation of the particle ﬂux: a factor exp(ipo xo / ) gives rise to a ﬂux po /m. 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) . equal to the velocity of the particle. Equation (3.4: Free Particle 57 The interpretation of j(r.18) in Ω∞ = R in the case o U (r.48. 2. Note that j(r. (2. t) = [2π]− 2 where the dispersion relationship holds ω = k2 .43) that any real wave function ψ(r.51) . i.45) Obviously. (3. t) = − ψ(r.e.71)]. In Sect.105. (3. j(r.47) .f.49) reads at t = t0 ψ(r.50) 3 ˜ d3 k φ(k) exp i(k · r − ωt) Ω∞ (3.3.48) ∂t 2m which describes the motion of free particles.167)] potential. for f (r) ∈ R. t) = 0 2 ∂ 2 i ψ(r. 2 for a free particle [c. (2. One can readily show by insertion into (3. m (3.e.f. t) (3. t) is that of density ﬂux. The generalization to three dimenisons implies then that the factor exp(i k · r) corresponds to an intial velocit k/m and a ﬂux of the same magnitude.46) i.. t0 ) determines φ(k). This follows directly from an inspection of Eq.49) ˜ Obviously.f.48) that the general solution is of the form ψ(r. 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.125)] and in a quadratic [c. (2. t) . t) has vanishing ﬂux anywhere.4 Solution of the Free Particle Schr¨dinger Equation o We want to consider now solutions of the Schr¨dinger equation (3. t) = − ∂ω da · j(r. Such case arose in Sect. t) gives rise to a descrease of the total probability in volume ω due to the disappearence of probability density at the surface ∂ω. t) points in the direction to the outside of volume ω. the initial condition at ψ(r. 2.41) written in the form ∂t ω d3 r ρ(r. t0 ) = [2π]− 2 Ω∞ 3 ˜ d3 k φ(k) exp i(k · r − ωt0 ) . (3. It is of interest to note from (3. 2. (3.148. 3. 2m (3. arises solely from the complex factor exp(i k · r).

52) above ψ(r. t|r0 .54) valid for this case.53) where φ(r. In case of the harmonoc oscillator the expansion can be stated in a closed form given in (4.e. i. t) as given by (3. Ω∞ (3. in (2.47).5).49.51) to be square integrable it follows according to the properties of the Fourier–transform that ψ(r. In fact.54) to the case of non-vanishing potentials U (r). t0 ) (3.114) and (4.31) applies.54) an alternative representation of the propagator (2.56) and using (2.e. t0 ) = 1 2π 3 d3 k exp Ω∞ ik · (r − r0 ) − i 2 k2 2m (t − t0 ) . (3. 3. The solution as stated is deﬁned in the inﬁnte space Ω∞ = R .52) We have not speciﬁed the spatial boundary condition in case of (3.55) m 2πi (t − t0 ) exp im (x − x0 )2 2 t − t0 This follows from completion of the square ik1 (x − x0 ) − (t − t0 ) 2m (t − to ) m x − xo = −i k1 − 2m t − to i 2 k2 1 2 + i m (x − xo )2 2 t − to (3.70) derived below for a particle in a box and the harmonic oscillator. i.54) yields (2. t0 ) . t) = Ω∞ d3 r0 φ(r. t|r0 .51.49).58 The inverse Fourier transform yields 3 ˜ φ(k) = [2π]− 2 Schr¨dinger Equation o d3 r0 exp(−ik · r0 ) ψ(r0 . t0 ) ψ(r0 .. (t − t0 ) (3. respectively. Comparision with Path Integral Formulation One can write solution (3.247). If one chooses the initial state f (r) deﬁned in (3.81) .49) is square integrable at all subsequent times and. derive an expression similar to (3. evaluating the integral in (3.. that the “natural boundary condition” (3. The ensuing solutions are called wave packets.47). Below we will generalize the propagator (3. The general form for this propagator involves an expansion in terms of a complete set of eigenfunctions as in (3. We have identiﬁed then with (3. 3. To show this one needs to note 1 2π = +∞ dk1 exp −∞ ik1 (x − x0 ) − 1 2 i 2 k2 1 2m . hence.54) This expression obviously has the same form as postulated in the path integral formulation of Quantum Mechanics introduced above.

3. 3. The remaining integration over k leads to the integral +∞ −∞ dk exp ikx − 1 2 (k − po 2 2 ) δ − i k2 m (t − t0 ) (3. (3. t|x0 . (3. i. r ∈ ∂Ω (3.62) We will restrict the space of allowed solutions to a volume Ω such that the functions also make the concomitant (3. the boundary conditions are φ(r) = 0 ∀r. (3..57) where φ(x.53.58) 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. Stationary States We consider now solutions of the time-dependent Schr¨dinger equation (3. t) 2 k2 (3. Accordingly. t|x0 . t0 ) ψ(x0 . t) is given by (??). 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 (3.30). 3.54) is +∞ ψ(x.56)]. a result which is identical to the expression (??) obtained by means of the path integral propagator (2.3. the functions obey boundary conditions of the type (3.20) which are of o the form ψ(r.e.61) × exp − (x − po m t)2 2δ 2 (1 + 2 t2 ) m2 δ 4 (1 − i .4: Free Particle Free Particle at Rest 59 We want to apply solution (3.63) . t) = f (t) φ(r) .64) (3. t) = −∞ dx0 φ(x.60) yields with t0 = 0 ψ(x.49. We have demonstrated then in this case that the Schr¨dinger formulation of Quantum o Mechanics is equivalent to the Feynman path integral formulation. (3.60) =???? Combining (3. 3.f.19. r ∈ ∂Ω or da · φ(r) = 0 ∀r.55.59) which is solved through completion of the square in the exponent [c.29. The 1-dimensional version of (3.52) to the case that the initial state of a 1-dimensional free particle ψ(x0 .57–3.27) vanish on the surface ∂Ω of Ω.46). t0 ) = 1 2π +∞ dk exp −∞ ik (x − x0 ) − i 2m (t − t0 ) .

67) (3. h1 (r) = φ(r). It turns out that a solution for f (t) can be found for any E. We must postulate therefore ∂t f (t) ˆ H φ(r) = E f (t) = E φ(r) .e. One calls such states stationary states.31). according to (3. In order to further characterize the solution (3. This can hold only if |f (t)| is time-independent. We have then shown that ψ(r.e. t) = Ω Ω d3 r |ψ(r. one factor depending only on the time variable and the other only on the space variables are called separable in space and time.66) i. i. It follows then from (3. (3.65) is constant.62) exists.69) is called an eigenvalue problem. (3. At this point we will accept that solutions φ(r) of the type (3. t)|2 = |f (t)|2 Ω d3 r |φ(r)|2 (3. We will demonstrate that solutions of the type (3. . the eigenvalues of the operator H. 3. solutions exist only for a set of discrete E ˆ values. We denote the corresponding solution by φE (r). We will encounter many such problems in the subsequent Sections.62) are special solutions of the timedependent Schr¨dinger equation.63.62) is |ψ(r.68) can hold only for all t and all r if g1 (t) = E g2 (t) and E h1 (r) = h2 (r) for some E ∈ C. . if f (t) = eiα . t) vanishes on the surface of ∂Ω.70) The task of ﬁnding solutions φ(r) which solve (3. One can conclude that the probability density for the state (3.62) do exist and we will characterize the two factors of the solution f (t) and φ(r).62) have the particular property that the associated probability distributions are independent of time. This yields an expression g1 (t) h1 (r) = g2 (t) h2 (r) (3.69) If these two equations can be solved simultaneously a solution of the type (3. t) = f (0) exp − i E φE (r) where ˆ H φE (r) = E φE (r) .19).27) holds and.64).42) the ﬂux j(r. (3. by no means all solutions are of this type. It follows from the observation that for the solution space considered (3. n = 1. however.68) ˆ where g1 (t) = i ∂t f (t).62) we insert it into (3. for functions required to obey boundary conditions (3. We want to demonstrate this property now.. We may note in passing that solutions of the type (3. The identity (3. is time-independent. for the eigenvalue problems in the conﬁned function space. α ∈ R .62) which consist of two factors.60 Schr¨dinger Equation o and aﬀect only the spatial wave function φ(r).. t)|2 = |φ(r)|2 . In fact. i. (3.. .e. g2 (t) = f (t). often only for a discrete set of values En . a common case is Ω = Ω∞ and the ‘natural boundary condition’ (3.45) that the total probability d3 r ρ(r. the solutions o (3. namely f (t) = f (0) exp − i Et . As pointed out above.71) . hence.69) exist. At this point we state without proof that. in general. and h2 (r) = H φ(r). 2. It is important to realize that the separable solutions (3.

We want to prove now that the eigenvalues E which arise in the eigenvalue problem (3. in fact.28) for the special case that f and g in (3. E (3.e. t) = exp − E t 2 φE (r) . t) = exp − i 2 k2 2m t exp i k · r (3. The resulting total energy values E are positive.20). m (3.66) E must be real. E ∈ R. ˆ d3 r φ∗ (r) H φE (r) = E Ω Ω ˆ d3 r φE (r) H φ∗ (r) . (3.71) this yields E Ω d3 r φ∗ (r) φE (r) = E ∗ E Ω d3 r φE (r) φ∗ (r) . one can interpret then k as the magnitude of the momentum of the particle. E (3.3. o According to (3. Stationary State of a Free Particle We consider now the stationary state of a free particle described by i ψ(r.g. i. Obviously.28) both represent the state φE (r). We start our proof using the property (3. We will show in Section 5 that E can be interprerted as the total energy of a stationary state. particles scattered of a potential. should be positive.74) is actually best labelled by an index k.4: Free Particle 61 is a solution of the time-dependent Schr¨dinger equation (3. a property which is to be expected since the energy is purely kinetic energy which.76) has kinetic energy 2 k 2 /2m.73) from which follows E = E ∗ and. − 2 2m φ(r) = E φ(r) . .74) using exp i k · r = ik exp i k · r . The corresponding stationary solution ψ(r. t) = |N |2 k .. We will wave this assumption as we always need to do later whenever we deal with unbound particles.43) is j(r. = E. e.71) are. The ﬂux corresponding to (3.75) 2m One can ascertain this statement by inserting the expression for φk (r) into the eigenvalue problem posed in (3. 3. As a result we cannot postulate in the present case that wave functions are localized and normalizable. real.72) According to (3.76) according to (3. (3.19.74) The classical free particle with constant energy E > 0 moves without bounds in the space Ω∞ . of course. The solution φE (r) corresponding to the eigenvalue problem posed by (3. hence.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 . k ∈ R 2 k2 φk (r) = N exp i k · r .

e.o) (−a) = 0 (3. (3.82) satisfy the diﬀerential equation in (3.84) .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. φ(±a) = 0 . Hence. a] ⊂ R → R. therefore.79) The stationary solutions have the form ψ(x. The boundary conditions which according to (3. Solutions of the Schr¨dinger Equation in F1 o The time–dependent solutions satisfy i ∂t ψ(x. i.82) have the property that either both boundary conditions are satisﬁed or none. 2a n = 1. t) . It turns out that this boundary condition can only be satisﬁed for a discrete set of k–values kn . and so-called odd solutions obeying φ(x) = −φ(−x) φE (x) = A sinkx.81) 2m = E.62 Schr¨dinger Equation o 3. hence. 5 . a] ⊂ R which vanish on the surface ∂Ω1 = {−a.81.. In case of the even solutions (3. .83) The solutions (3.80) need to be satisﬁed are φE (e. . 3. t) = exp(−iEt/ )φE (x) where φE (x) is determined by 2 d2 − φE (x) = E φE (x) . a] ⊂ R assuming that the potential outside of this interval is inﬁnite. We can expect. t) = − d2 ψ(x.80) 2m dx2 We note that the box is symmetric with respect to the origin. 3. We assume.80). that the solutions obey this symmetry as well. so-called even solutions obeying φ(x) = φ(−x) φE (x) = A coskx . (3.o) (a) = 0 and φE (e. f ∈ F1 = {f : [−a. . 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.82) One can readily verify that (3. (o) 2 k2 (e) 2 k2 2m = E (3. let say the one at x = a. two types of solutions. a}. we have to consider only one boundary condition.81) they are kn = nπ .78) (3. We will refer to this as a particle in a one-dimensional ‘box’. n ∈ N. 2m dx2 2 (± ) = } (3. f continuous. (3.81.

81) and (3.88) where the energies for odd (even) n–values correspond to the even (odd) solutions given in (3.94) . n = 1.) The Energy Spectrum and Stationary State Wave Functions The energy values corresponding to the kn –values in (3. 3.89) nπx . . (3. . 2. 6 .e.5: Particle in One-Dimensional Box since for such kn 63 nπa nπ = cos 2a 2 In case of the odd solutions (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 . 2a 1 .93) The normalization constants are then An = (3. 6 . . 4. It is desirable to normalize the wave functions such that +a −a dx |φ(e. . (3. . 3. (3. 4. x) = An sin n nπx . (3. x) = An cos n and φ(o) (a.o) (a. 4. respectively. 5 .84. . 3 .1). 3. .82).81. .3.87) (Note that.90) 2a The wave functions represent stationary states of the particle in a one-dimensional box. (3. . (3. are En = 2π2 8ma2 n2 . .86). . 6 . .86) satisfy the boundary condition since for such kn nπa sin(kn a) = sin 2a = sin nπ 2 = 0.91) holds. 2a n = 1. This condition implies for the even states +a |An |2 −a dx cos2 nπx = |An |2 a = 1 . 3. n = 2. 2a n = 2. (3. 5 . according to (3. (3. n is assumed to be even. φ(e) (a. x)|2 = 1 n (3. The wave functions for the ﬁve lowest energies En are presented in Fig. a n = 2.85) (3. 2a n = 1.82) only the kn –values cos(kn a) = cos kn = nπ . i.86).. . according to the dispersion relationships given in (3.82). = 0.92) and for the odd states +a |An |2 −a dx sin2 nπx = |An |2 a = 1 . By counting the number of their nodes one can determine the energy ordering of the wave functions.

64 Schr¨dinger Equation o E h2 ] [ 8 m a2 25 ¢ 16 9 4 Figure 3. x) for n = 1. 3. 2.1: Eigenvalues En and eigenfunctions φn (e. ¡ 1 -a a x .o) (a. 4. 5 of particle in a box.

x) .} where φn (a. it holds: f |f Ω1 = 0 → f (x) ≡ 0. for n = 2. (3. N ≥ 1. (ii) n. In case of n = m we have to consider three cases. . 4. Hence we need to consider only the ﬁrst two cases. The latter case leads to integrals φn |φm Ω1 = 1 a +a dx cos −a nπx mπx sin . 3. 3. one obtains for n. the integrand is odd. x) = together with the scalar product5 +a 65 (3. x) (3.95) 1 a cos nπx 2a sin nπx 2a for n = 1. and (iii) the mixed case. In fact. In case of n. the integral vanishes. for ∞ f (x) = n=1 dn φn (a.e.97) form an orthonormal basis set.98) the elements of B1 must be linearly independent. it holds φn |φm Ω1 = δn m . too. 3. . Similarly. . m even φn |φm 1 a Ω1 = 1 a +a nπx −a dx sin 2a sin mπx = 2a (3. m odd. . hence. m both odd.102) holds according to (3. . the integrals vanish. this integral vanishes. n = 1. n = m.98) The latter property is obviously true for n = m. i. Obviously..e.101) +a −a dx cos (n−m)πx − cos (n+m)πx 2a 2a and. (3. 2. n (3. the integral arises φn |φm 1 a Ω1 = 1 a +a nπx −a dx cos 2a cos mπx = 2a (3. g ∈ F1 (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 (3.96) f |g Ω1 = −a dxf (x) g(x) . .90.94) B1 = {φn (a..103) We will show in Section 5 that the property of a scalar product do indeed apply.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.89. 3.99) Since in this case the integrand is a product of an even and of an odd function.98) ∞ f |f 5 Ω1 = n=1 d2 . f. In particular. Because of the property (3. (i) n. a] are N. m both even. 5 . 6 . . i. 2a 2a (3. .3.

79) requiring the o initial condition cn (t0 ) = 1 .108) where the functions cn (t) are to be determined from the Schr¨dinger equation (3. .106) Using the orthonormality property (3. 2.105) must vanish. however.78) can be expressed as a linear combination of the elements of B1 deﬁned in (3. x) cn (t) −a dx0 φn (a. Accordingly. x0 ) ψ(x0 . 4. . . 3.98) one obtains +a dx0 φm (a. . Demonstration of completeness is a formidable task. For this purpose we expand ∞ ψ(x. .108) into the Schr¨dinger equation yields o +a ∞ n=1 φn (a. .e. Restricting the expansion (3. 3. i. t0 ) = n=1 dn φn (a. x0 ) ψ(x0 . The functions f are periodic with period 2a. x) cn (t) −a dx0 φn (a. 1. In the present case. in (3. (3. . such demonstration can be based on the theory of Fourier series. . It follows that B1 n deﬁned in (3. For this purpose we extend the deﬁnition of the elements of F1 to the whole real axis through ˜ f : R → R.96). x0 ) ψ(x0 . 5 .110) . Evaluating the Propagator We can now use the expansion of any initial wave function ψ(x. Hence. there exist real constants {an .} and {bn . This implies that only the trigonometric functions which are elements of B1 enter into the Fourier ˜ series.95. We like to show ﬁnally that the basis (3.105) to the interval [−a. t) = n=1 φn (a. t0 ) . 2. t0 ) (3.109) Insertion of (3. t0 ) = +a ∞ i n=1 − En φn (a.95) is also complete..} such that ∞ ∞ nπx nπx ˜ = + an sin (3. n = 1.104) ˜ where [y]a = y mod2a. m = 1.107) Inserting this into (3. We have then shown that any f corresponding to elements of F1 can be expanded in terms of elements in B1 . a] yields then also an expansion for any element in F1 and B1 is a complete basis for F1 . t0 ) in terms of eigenfunctions φn (a.. t0 ) = dm . (3. . . x) to obtain an expression for ψ(x.106) and generalizing to t ≥ t0 one can write ∞ +a ψ(x. x) ∂t cn (t) −a dx0 φn (a. .95) is an orthonormal basis. n = 2.105) 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. and bn . . x) . . . 3. t) at times t > t0 . n = 0. i. (3. any element of the function space F1 deﬁned in (3.66 Schr¨dinger Equation o f (x) ≡ 0 implies f |f Ω1 = 0 which in turn implies dn = 0 since d2 ≥ 0. the coeﬃcients an . −a (3. they can be expanded in terms of a Fourier series.e. n = 1. ( ) = ([ + ] − ) (3. x0 ) ψ(x0 .

114) the representation of the propagator for a particle in a box with inﬁnite walls. t0 ) ψ(x0 . in (2. (3. according to (3. The interference pattern begins to ‘smear out’ ﬁrst.5: Particle in One-Dimensional Box Multiplying both sides by φm (a. One can recognize that the particle moves ﬁrst to the left and that near the right wall of the box interference eﬀects develop. i ∂t cm (t) = − Em cm (t) . t0 ) is also referred to as a Greens function.113) where φ(x. t0 ) (3. t|x0 . t0 ) = δ(x − x0 ) as can be readily veriﬁed using (3. t0 ) itself is a solution of the time-dependent Schr¨dinger o equation (3. t) for any initial condition ψ(x. but the collision with the left wall leads to new interference eﬀects. the elements of which satisfy the appropriate boundary conditions. being reﬂected at the right wall.5).54) as in the case of the free particle system which does not exhibit any bound states.e. 3. The particle moves then to the left.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. k0 = .108. Example of a Non-Stationary State As an illustration of a non-stationary state we consider a particle in an initial state ψ(x0 .2 presents the probability distribution of the particle at subsequent times. cm (t0 ) = 1 . Figure 3. We have identiﬁed then with (3. t|x0 . x) and integrating over [−a. Note that the propagator has been evaluated in terms of elements of a particular function space F1 . t) = −a dx0 φ(x. t|x0 . The respective initial condition is φ(x.78). 4 a (3. In the present system which is composed solely of bound states the propagator is given by a sum. (3.109) are cm (t) = exp i − Em (t − t0 ) .112) Equations (3. 67 (3.. rather than by an integral (3. i. The last frame shows the wave front reaching again the right wall and the onset of new interference. σ = a 15 .79) which lies in the proper function space (3.114) This expression has the same form as postulated in the path integral formulation of Quantum Mechanics introduced above. Often the propagator φ(x. a] yields. t0 ) = 1 2πσ 2 1 4 exp − x2 0 + i k0 x0 2σ 2 . . This solution can be written +a ψ(x.111) The solutions of these equations which satisfy (3. It is of interest to note that φ(x. t0 ) = ∞ φn (a. t0 |x0 .98). In case that diﬀerent boundary conditions hold the propagator will be diﬀerent as well. t|x0 .3. t0 ).113).112) determine now ψ(x. x0 ) . x) exp n=1 i − En (t − t0 ) φn (a.

.2: Stroboscopic views of the probability distribution |ψ(x.16 m a h2 1 πσ 1 πσ -a a -a a Figure 3.72 m a h2 1 πσ 1 πσ -a a -a a |Ψ|2 2 t = 0.68 Schr¨dinger Equation o |Ψ|2 2 t = 0.24 m a h2 1 πσ k = 15 a 1 πσ -a a -a a |Ψ|2 2 t = 0.00 m a h2 |Ψ|2 2 t = 0. t)|2 for a particle in a box starting in a Gaussian distribution with momentum 15 /a.96 m a h2 |Ψ|2 2 t = 1.48 m a h2 |Ψ|2 2 t = 0.92 m a h2 |Ψ|2 2 t = 2.20 m a h2 1 πσ 1 πσ -a a -a a |Ψ|2 2 t = 1.44 m a h2 |Ψ|2 2 t = 1.68 m a h2 1 πσ 1 πσ -a a -a a |Ψ|2 2 t = 1.

95). of the space F1 is that the elements of the basis set can be enumerated by integer numbers. 2a2 .118) which are stationary states.119) where φE (x1 . x2 . x3 –axes with the edges of the box yields spatial boundary conditions which are obeyed by the elements of the function space F3 = {f : Ω → R. x2 . x3 ) (3. x2 . 3. x2 . An important property of this basis and. a3 ] ⊂ R = {(x1 . x3 ) = j=1 φ(j) (xj ) (3. The description employed a space of functions F1 deﬁned in (3.117) 2m 2 2 2 ∂1 + ∂2 + ∂3 (3. f continuous.. x3 ) . (3. . x2 . Placing the origin at the center and aligning the x1 .e. x3 ) = E φE (x1 . x1 = ±a1 } ∪ {(x1 . can be counted. A complete basis of F1 is given by the inﬁnite set B1 (3. ) = ∀( . x2 .3. x2 . ˆ H = − 2 (3. 2. This property allowed us to evaluate the propagator (3. i. x2 . j = 1.78).97) with respect to which the eigenfunctions are orthonormal. 2a3 . x3 ) is an element of the function space F3 deﬁned in (3. x3 = ±a3 } .. x3 . hence. a1 ] ⊗ [−a2 . (3.120) ˆ Since the Hamiltonian H is a sum of operators O(xj ) each dependent only on a single variable. one can express H 3 φ(x1 .114) in terms of which the solutions for all initial conditions can be expressed. )T ∈ ∂ } . . x2 . x3 . ( . a2 ] ⊗ [−a3 . i.e. We have deﬁned in the space F1 a scalar product (3. t) = H ψ(x1 . x3 )T ∈ Ω. t) = exp − E t φE (x1 . 3 (3. We seek then solutions of the time-dependent Schr¨dinger equation o ˆ i ∂t ψ(x. x3 )T ∈ Ω. x3 )T ∈ Ω. t) . x2 .116) where Ω is the interior of the box and ∂Ω its surface Ω ∂Ω = [−a1 . x2 . The corresponding solutions have the form i ψ(x1 .121) where − 2 2m 2 ∂j φ(j) (xj ) = Ej φ(j) (xj ) . φ(j) (±aj ) = 0 .122) .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. ˆ = O(x1 ) + O(x2 ) + O(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 . x2 = ±a2 } ∪ {(x1 .

x2 .n3 ) (a1 .120) can be written φ(n1 . x3 –axes.70 Schr¨dinger Equation o and E1 + E2 + E3 = E.n2 .124) The same considerations as in the one-dimensional case allow one to show that B3 = {φ(n1 . a2 .. a3 are identical.n3 ) = 2π2 = φn1 (a1 . 2.n2 . a2 .125) is a complete orthonormal basis of F3 and that the propagator for the three-dimensional box is φ(r.n3 =1 φ(n1 .e. a3 . a2 . .123) 8m n2 n2 n2 1 2 3 + 2 + 2 a2 a2 a3 1 . a1 = a2 = a3 = a then rotation around the x1 . (3.122) are given by (3. x1 . .n2 . 3. x3 –axes by π/2 also leaves the system unaltered. x2 ) φn3 (a3 . 2. x3 ) . .n2 . a3 . r0 ) . This symmetry has been exploited in deriving the stationary states. n 3 and E(n1 . (3. For example. t0 ) = n1 . x2 . n2 .n2 . (3. a3 . the solutions of (3. n1 . if all three lengths a1 .} ∞ (3. 3.126) Symmetries The three-dimensional box conﬁning a particle allows three symmetry operations that leave the box unchanged. a3 . .n2 . . t|r0 . . n2 .n3 ) (a1 . a2 . This additional symmetry is also reﬂected by degeneracies in the energy levels. . hence. 3. n3 = 1. x1 . Comparing (3. . The energies and corresponding degeneracies of the particle in the three-dimensional box with all side lengths equal to 2a are given in the following Table: n1 1 1 1 1 2 1 2 1 2 3 1 n2 1 1 2 1 2 2 2 1 3 3 1 n3 1 2 2 3 2 3 3 4 3 3 5 E/[ 2 π 2 /8ma2 ] 3 6 9 11 12 14 17 18 22 27 27 degeneracy single three-fold three-fold three-fold single six-fold three-fold three-fold three-fold single three-fold One can readily verify that the symmetry of the box leads to three-fold and six-fold degeneracies. i. n 2 . in the present case .n3 ) (a1 . x3 ) = 1. r) exp i − E(n1 n2 n3 ) (t − t0 ) φ(n1 . Such degeneracies are always a signature of an underlying symmetry. x2 . 2.122) with (3. a2 . Actually. namely rotation by π around the x1 . n1 . x1 ) φn2 (a2 . x2 . If two or all three orthogonal sides of the box have the same length further symmetry operations leave the system unaltered.n3 ) (a1 . x3 ) n1 .80) shows that the solutions of (3. n3 = 1. .96) and.

r) + γ φ(2. the identity 32 + 32 + 32 = 12 + 12 + 52 . a. r) + β φ(2. The origin of this degeneracy is. However. We consider the degeneracy of the energy E122 which is due to the identity E122 = E212 = E221 . 1. However. to construct n orthogonal stationary states6 . for (n1 . n3 ) = (3. 3) (1. φ(1. this linear combination o is not necessarily orthogonal to other degeneraste states. n2 . r). Any linear combination of wave functions ˜ φ(r) = α φ(1. for example. 3.127) ˜ ˆ ˜ obeys the stationary Schr¨dinger equation H φ(r) = E122 φ(r). .6: Particle in Three-Dimensional Box 71 ‘accidental’ degeneracies also occur. a. Hence.3. a.2) (a. in case of an n–fold degeneracy it is always possible. 6 This is a result of linear algebra which the reader may ﬁnd in a respective textbook.2. a. due to the hermitian character of ˆ H. a.2) (a. a. r) (3.2.1. One particular aspect of the degeneracies illustrated in the Table above is worth mentioning.2. 5) as shown in the Table above..g. a. a. however. in case of degenerate states one cannot necessarily expect that stationary states are orthogonal. e.2) (a.1) (a.

72 Schr¨dinger Equation o .

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

all stationary states of the harmonic oscillator must be bound states and. the natural boundary conditions apply x→±∞ lim φE (x) = 0 . In the following algebraic solution of (4. We will point out explicitly where assumptions are made which built on this restriction. 4.235) had been determined.5) ˆ H = 1 1 2 p + m ω 2 x2 ˆ ˆ 2m 2 (4.6) the ﬁrst being a diﬀerential operator and the second a multiplicative operator.74 Linear Harmonic Oscillator In the following we consider ﬁrst the stationary states of the linear harmonic oscillator and later consider the propagator which describes the time evolution of any initial state.5). the commutator has a simple form. ∈ C∞ .2) can be expressed in terms of the two operators p = ˆ d . however. If this restriction would not apply and all functions f : R → R would be admitted. t) = exp(−iEt/ )φE (x.3) would be continuous and the eigenfunctions φE (x) would not be normalizable. This property states that p and x obey an ˆ ˆ algebra in which the two do not commute. is the commutation property 1 1 (4. The Hamiltonian H can be expressed in terms of the operators acting on the space (4.3) Due to the nature of the harmonic potential which does not allow a particle with ﬁnite energy to move to arbitrary large distances. t) where ˆ H φE (x) = E φE (x) . In order [ˆ. The operators act ˆ on the space of functions N1 deﬁned in (4. In the framework of the Schr¨dinger equation the stationary o states are solutions of (4. i dx x = x ˆ (4. lim ( ) = } (4. only for a discrete set of E values can the boundary conditions (4. beside the representation (4. therefore. the spectrum of ˆ H in (4.5) →±∞ where C∞ denotes the set of functions which together with all of their derivatives are continuous.1) of the form ψ(x. (4. The cardinal property exploited below.7) of the Hamiltonian. 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. in which the operators used below.4) be satisﬁed.3) can be solved for any E ∈ R. act. It is important to keep in mind this restriction of the space.8) i which holds for any position and momentum operator. however.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.1 Creation and Annihilation Operators The Hamiltonian operator (4.4) Equation (4. (4. x] = p ˆ .

4. 1 i dx i dx i i (4.7) and (4. 1 ω 2 (4.8) we recall that all operators act on functions in N1 and.10) holds for a ˆ ˆ [ˆ− . 1 2 (4.17) d dx We also express a+ and a− directly in terms of the coordinate x and the diﬀerential operator ˆ ˆ a+ = ˆ mω x − 2 d . x] p ˆ on such functions must be considered. x] f (x) = p ˆ d d x f (x) − x f (x) = f (x) = 1 f (x) .7) assuming that the factors are easier to handle than the Hamiltonian in yielding spectrum and eigenstates. ˆ 2 2m ω (4. Our next step is an attempt to factorize the Hamiltonian (4. In fact. the action of [ˆ. to factor H/ ω.12) To prove this commutation property we determine using (4.15) 1 ˆ 1 H + 1 .8) yield a− a+ = ˆ ˆ Similarly one can show a+ a− = ˆ ˆ 1 ˆ 1 H − 1 .11) ˆ The reader may note that we have attempted.16) (4. ˆ ˆ .13) (4. ˆ 2 2m ω a− = ˆ mω i x + √ ˆ p. a+ ] = 1 . 1 ω 2 (4.18) It is of interest to note that the operators a+ and a− are real diﬀerential operators.12). Since a+ and a− are ˆ ˆ + and a− since operators and not scalars we cannot simply expect that the identy (4. x] . a+ ] = a− a+ − a+ a− does not necessarily vanish. the commutator property (4.11) a− a+ = ˆ ˆ (4. 2mω dx a− = ˆ mω x + ˆ 2 d .9) From this follows (4. One obtains [ˆ. p ˆ 2 2m ω 2 (4.15) together lead to the commutation property (4. Before we continue we like to write (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 .8) a ˆ ˆ ˆ ˆ ˆ implies [ˆ− . hence. a ˆ 1 (4. in fact.14) 1 i mω 2 x + ˆ p2 + ˆ [ˆ.14.10) we deﬁne a+ = ˆ mω i x − √ ˆ p.8).1: Creation and Annihilation Operators 75 to derive (4. 2mω dx (4.14) and (4. Being guided by the identity for scalar numbers (b − ic)(b + ic) = b2 − i(cb − bc) + c2 = b2 + c2 (4. 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) .76 Relationship between a+ and a− Linear Harmonic Oscillator The operators a+ and a− are related to each other by the following property which holds for all ˆ ˆ functions f. one obtains using (4. 4.16.17.16. 4. ˆ The results (4.19) This property states that the operators a+ and a− are the adjoints of each other.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) . g ∈ N1 +∞ +∞ dx f (x) a+ g(x) = −∞ −∞ dx g(x) a− f (x) .17. it holds that for a stationary state φE (x) of energy E deﬁned through (4. 4. (4. This property of operators is investigated more systematically in Section 5.20) ˆ In the following we will determine the spectrum of H and its eigenstates. 4.24) − m This property states that the operators in the function space N1 are the hermitian conjugate of each other. 4.19). ˆ (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) . 4.17. The derivation will be based solely on the properties (4. The property ˆ ˆ follows directly from (??). . ˆ (4.19) in the form1 f |ˆ+ g a Ω∞ = g|ˆ− f a Ω∞ . (4. a+ and a− as Ladder Operators ˆ ˆ The operators a+ and a− allow one to generate all stationary states of a harmonic oscillator once ˆ ˆ one such state φE (x) ˆ H φE (x) = E φE (x) (4. (4. Using (??) we like to state (4. 4. 4.22.21) is available.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. In fact.22) Similarly one can show using again (4.23) Together.16.12.

a state φ0 (x) which obeys the property a− φ0 (x) = 0 ˆ (4. 2 2 (4.33) . (4.30) Assuming that φ0 (y) does not vanish anywhere in its domain ] − ∞.28) 4. Another name is creation (ˆ+ ) and annihilation (ˆ− ) operators since these operators a a create and annihilate vibrational quanta ω. one expects that E = 0 should be a lower bound.2: Ground State 77 One can use these relationships to construct an inﬁnite number of stationary states by stepping up and down the spectrum in steps of ± ω.11) one can rewrite (4.27) Since E = E one can conclude φE |φE Ω∞ = 0.2 Ground State of the Harmonic Oscillator d + y dy A suitable solution of (4. m ∈ Z when m is decreased.e.32) for some constant c0 or (4. (4. in fact.. E = E . i.) ˆ ˆ 0 = φE |HφE Ω∞ − φE |HφE Ω∞ = ( E − E ) φE |φE Ω∞ . the solution φ0 (x) of (4.17) ˆ H φ0 (x) = ( ω a+ a− + ˆ ˆ 1 ω 1 ) φ0 (x) = 1 ω φ0 (x) . 4. φ0 (y) dy dy the solution of which is 1 lnφ0 (y) = − y 2 + c0 2 φ0 (y) = c0 exp 1 − y2 2 .25) on one side would lead to termination of the sequence E0 + m. For f (x) = φE (x) and g(x) = φE (x) in (??) follows (Note that according to (??) the eigenvalue E is real.25) can. however. There arise.29) d 1 d φ0 (y) = = lnφ0 (y) = − y .25) φ0 (y) = 0 (4.26) Of course. The property (??) has an important consequence for the stationary states φE (x). two diﬃculties: (i) one needs to know at least one stationary state to start the construction. (ii) the construction appears to yield energy eigenvalues without lower bounds when. be found.6.5).29) where y = mω x. In fact. in fact. For obvious reasons one refers to a+ and a− as ladder ˆ ˆ operators. Let φE (x) and φE (x) be two normalized stationary states corresponding to two diﬀerent energies E. φ0 (x) should be an element of the function space N1 deﬁned in (4.31) (4.4. E . Using (4. on ˆ the other side such a state is itself an eigenstate of H as can be shown using (4. (4. +∞[ one can write (4.25) needs to be normalizable in order to represent a bound state of the harmonic oscillator. It turns out that both diﬃculties can be resolved together. (4.

. i. (4. 2. it holds ˆ H φn (x) = ω(n + 1 ) φn (x) .. For this purpose we apply the operator a+ to the ˆ ground state (4. .38) These states correspond to the energy eigenvalues (4. which for a narrowly localized state yields a large positive value. 2 n = 0. . (4. 2 n = 0.2) there are two competing contributions to the energy.. would yield a vanishing contribution. .37) above the ground state energy.34) dx |φ0 (x)|2 = |c0 |2 −∞ −∞ dx exp − mωx2 2 = |c0 |2 π .e. therefore. Such states need to be suitably normalized for which purpose we introduce a normalization constant cn φn (x) = cn a+ ˆ n φ0 (x) .25) admits only one solution there is only one set of states with energy E + m ω. 2.36) Since the ﬁrst order diﬀerential equation (4.39) We notice that the ground state wave function φ0 (x) as well as the operators (ˆ+ )n are real. m ∈ Z which properly terminate at some minimum value E + m0 ω ≥ 0. 4. the so-called excited states. 1. . 1.36) assumes a functional form such that both terms together assume a minimum value. i. 2.37) It is most remarkable that the energy 1 ω of the ground state is larger than the lowest classically 2 allowed energy E = 0. i. mω (4.3 Excited States of the Harmonic Oscillator Having obtained a suitable stationary state with lowest energy.78 Linear Harmonic Oscillator This solution is obviously normalizable. the potential energy contribution which for a state δ(x). a state conﬁned to the minimum corresponding to the classical state of lowest energy. we can now construct the stationary states corresponding to energies (4. 2. . n = 0.24) can then be written as follows En = ω(n + 1 ). . We will consider this point more systematically in Section 5.e. The conventional normalization condition φ0 |φ0 reads +∞ +∞ Ω∞ = 1 (4. .26) the energy value associated with this state is 1 ω.. and the kinetic energy contribution. The ground state (4.e. . (4.35) The appropriate ground state is φ0 (x) = mω π 1 4 exp − mωx2 2 .39). . 2 This state of lowest energy is called the ground state of the oscillator. We recall that according to (4. The reason is that in the Hamiltonian (4. . . We a can. 1. we construct the states for n = 1.36) n times. choose the normalization constants cn and the functions φn (x) real as well. (4. The set of allowed energies of the oscillator according to (4.. .

.42) ˆ Using (4.40) φn−1 (x) = βn a− φn (x) ˆ (4. according to (4. 4. . . according to (4. √ a− φn (x) = n φn−1 (x) . n = 0. 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∞ (4.51) . We consider then the situation that we have obtained a properly normalized state φn (x).40). 1. n = 0. To determine these constants we adopt a recursion scheme.46) According to (4.24) holds in analogy to (4.50) φn (x) .3: Excited States 79 We need to determine now the normalization constants cn .47) for some suitable constants βn .41) Employing the adjoint property (4.4. . (4.39) yield √ From this follows βn = 1/ n and. .45) We like to note that these functions according to (4. . ˆ Repeated application of this relationship yields φn−s (x) = (n − s)! − a ˆ n! s 2 1 = βn n φn |φn Ω∞ 2 = βn n .45) form an orthonormal set. . To determine βn we note using (4.47). 2.28) and the construction (4.20) 1 = φn−1 |φn−1 Equations (4. For n = 0 holds c0 = 1.40) = 1. ˆ (4.e. (4. (4. 1. (4.40–4. √ a+ φn (x) = n + 1 φn+1 (x) . they obey φn |φn = δn n .20) yields 2 αn φn |ˆ− a+ φn a ˆ Ω∞ = 1. (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. βn must be real as well. 2.48) (4.49) (4. n. i.43) (4.18)] and since the ˆ φn (x) are real functions.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) .15 . Ω∞ 2 = βn a− φn |ˆ− φn ˆ a Ω∞ 2 = βn φn |ˆ+ a− φn a ˆ Ω∞ . Since a− is a real diﬀerential operator [see (4. (4.

35) rather than (4. by dx = dy mω 1 4 . . by introducing the variable y deﬁned in (4. (4.7 and given. 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.233) derived in Sect.. (4. for example.56) Relationship to Hermite Polynomials We want to demonstrate now that the expression (4. i.45). simplifying the calculation. It should be noted that √ the normalization (4.7.46) yields a set of normalized states.80 Evaluating the Stationary States Linear Harmonic Oscillator We want to derive now an analytical expression for the stationary state wave functions φn (x) deﬁned through (4. For this purpose we start from expression (4. however. namely the square root of the Jacobian dx/dy.30) and employing the normalization +∞ −∞ dy φ2 (y) = 1 n (4.52).35) by a constant.57) such that one can write the stationary state wave functions of the harmonic oscillator φn (y) = − √ e 2n n! π 1 y2 2 Hn (y) .200).39). (4. (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 . 2. (4.57) which includes the factor 1/ n! according to Eqs.e.40–4.58) This result agrees with the expression (2. 2. In terms of y the ground state wave function is φ0 (y) = π − 4 e− and a+ is ˆ 1 a+ = √ ˆ 2 y − d dy .53) We will later re-introduce this factor to account for the proper normalization (4. by the Rodrigues formula (2.55) 1 y2 2 (4.52) This normalization of the wave functions diﬀers from that postulated in (4.28) of the ground state and the deﬁnition (4. n–independent factor.

61) One can factor g(y) and employ (4.2) for an arbitrary initial wave function ψ(x0 . For the sake of notational simplicity we employ initially the coordinate y as deﬁned in (4. (4.114)]. Starting point of our derivation is the assumption that the initial condition can be expanded in terms of the eigenstates φn (y) (4. therefore.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 .106–3.60) and.59) holds for n = 0. g(y) = (−1)n+1 e y2 2 dn+1 −y2 e = dy n+1 y − d dy n+1 e− y2 2 . dy dy (4. (4.4: Propagator 81 To verify the realtionship between the deﬁnition (4. t0 ) = n=0 dn φn (y0 ) .e. 1. (4. t0 ). hence.39) ∞ ψ(y0 .59) by induction noting ﬁrst that (4.65) .59) dn −y2 e dy n (4.4. the Rodrigues formula (2. i.61) and.1.30) and return later to the coordinate x.4 Propagator for the Harmonic Oscillator We consider now the solution of the time-dependent Schr¨dinger equation of the harmonic oscillator o (4. (3.57) of the hermite polynomials and the deﬁnition given by (2.62) Denoting f (y) = (y − d/dy)exp(−y 2 /2) and employing y2 y2 d d − y2 e 2 f (y) = −y e− 2 f (y) + e− 2 f (y) dy dy (4.200) we need to verify y − which implies Hn (y) = (−1)n ey 2 d dy n e− y2 2 = (−1)n e y2 2 dn −y2 e .64) which implies that (4.59) hold.63) one obtains g(y) = y − d dy f (y) (4. We prove (4. 4.. 4. dy n (4. and showing then that the property also holds for n + 1 in case it holds for n. Our derivation will follow closely the procedure adopted for the case of a ‘particle in a box’ [see Eqs.200).

For this purpose we start from the integral representation (2.5). y0 . t) = n=0 Hn (y)e−y /2 Hn (y0 )e−y0 /2 n √ t . t) = π −3/2 exp 2 2 y 2 y0 + 2 2 2 +∞ +∞ ∞ du −∞ −∞ dv n=0 1 ( −2tuv )n × n! (4.Lebedev (Prentice Hall.70) is the propagator of the linear harmonic oscillator. 2. t1 |y0 . pp.15. t0 ) φn (y0 ) .70) and (2.112) and (4.108–3.66) One can extend expansion (4.7. t0 ) which is an element of N1 deﬁned in (4.70).67) for which according to (3. t1 ) = −∞ dy0 φ(y. (4. Employing orthogonality condition (4... We consider for this purpose the following expression for |t| < 1 ∞ w(y. (4. . t|y0 .39) holds cn (t1 ) = exp −i ω ( n + 1 ) ( t1 − t0 ) 2 . a supposition which is not proven here2 .46) the expansion coeﬃcients dn are +∞ dn = −∞ dy0 ψ(y0 . 2n n! π 2 2 (4. N.71) Applying (??) to express Hn (y) and Hn (y0 ) yields exp(−2tuv) w(y.N. 1 t = e−iω(t1 −t0 ) (4. this is an excellent textbook from which we have borrowed heavily in this Section.s. 4. Inc.J.69) where ∞ φ(y. 1965) Sect. 2 A demonstration of this property can be found in Special Functions and their Applications by N. t) (4. Englewood Cliﬀs. y0 . In order to prove the equivalence of (4. t1 ) = n=0 dn φn (y) cn (t) (4.82 Linear Harmonic Oscillator Such expansion is possible for any f (y0 ) = ψ(y0 .147) for the harmonic iscillator determined in Sect.147) we employ the technique of generating functions as in Sect.72) × exp( −u − v + 2iyu + 2iy0 v ) . sec:harm. 68. of (4. t0 ) ψ(y0 .68) Altogether one can express then the solution +∞ ψ(y. We want to demonstrate now that this propagator is identical to the propagator (2.225) which allows one to derive a generating function for products of Hermite polynomials which can be applied to the r. t0 ) = n=0 φn (y) φn (y0 ) tn+ 2 .h.65) to times t ≥ t0 through insertion of time-dependent coeﬃcients cn (t1 ) ∞ ψ(y.

t0 ) (4.4.h. t0 ) = exp mω 2iπ sinω(t1 − t0 ) 1 2 × . t0 ) = and 1 2 i π F (t1 . 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 = .. by a factor 4 mω/ for both φn (y) and φn (y0 ). (4. This requires that we employ (4. .74) a second time yields ﬁnally together with the deﬁnition (4.80) −2iω(t1 −t0 ) sinω(t1 − t0 ) 1−e We can ﬁnally express the propagator (4.4: Propagator Carrying out the sum on the r.70) is.s. y0 .75) Applying (4. t) = π −1 exp − 2 y 2 y0 + 2 2 +∞ dv exp(−(1 − t2 ) v 2 − 2i (yt − y0 ) v) . i. i. t0 ) = i φ(x.79) 1 + e−2iω(t1 −t0 ) cosω(t1 − t0 ) = . (4.147) derived by means of the path integral description. The resulting propagator is G(t1 .78) in terms of the coordinates x and x0 .e. one obtains w(y. y0 . √ πexp( t2 v 2 − 2iytv − y 2 ) (4..71) of w(y.77) The sum in (4. t) = π −3/2 exp +∞ 83 y2 y2 + 0 2 2 +∞ dv exp( −v 2 + 2iy0 v ) × −∞ × −∞ du exp( −u2 − 2u(tv − iy) ) . it holds φ(y. t1 |y0 .e.76) One can express this in terms of the stationary states (4.58) of the harmonic oscillator ∞ φn (y) φn (y0 ) tn = n=0 2 1 t 2 1 + t exp − (y 2 + y0 ) + 2 y y0 2 2) 2 1−t 1 − t2 π(1 − t 1 . indeed. (4. t0 ) − i y y0 2 F (t1 . −∞ (4. identical to the generating function (4. t0 ) = where F (t1 . t|x0 . y0 .74) applied once results in w(y. t0 ) exp i 2 1 2 (y + y0 ) G(t1 .81) imω 2 sinω(t1 − t0 ) (x2 + x2 ) cosω(t1 − t0 ) − 2 x x0 0 This result agrees with the propagator (2.77).73) The incomplete Gaussian integral +∞ −∞ dx e−a 2 x2 √ − 2bx = π b2 /a2 e . (4. a Re a2 > 0 (4.30) to replace y and y0 and that we multiply the propagator by mω/ .78) 1 − e−2iω(t1 −t0 ) = sinω(t1 − t0 ) 2 i e−iω(t1 −t0 ) (4.

84) = 1 1 = 0 √ = n + 1 φn+1 (y) √ = n φn−1 (y) = f |ˆ+ g a Ω∞ . phonons. 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. the approaches a ˆ a ˆ using y. each corresponding to a single harmonic oscillator of the type studied presently by us. (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..85) is convergent everywhere in R. a− must be equivalent. a (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 use of generating functions. d/dy and a+ . i. both quantum systems are endowed with a large number of modes. e.86) f (ˆ+ ) = a a ˆ ν! ν=0 .82) a+ = √ ˆ dy dy 2 2 √ √ Obviously. In this case we deﬁne ∞ f (ν) (0) + ν (4.84 Linear Harmonic Oscillator 4. Such functions are deﬁned for ˆ ˆ a f : R → R in case that the Taylor expansion ∞ f (y) = ν=0 f (ν) (0) ν y ν! (4.g. hence. We want to introduce now techniques based on the ladder operators a+ and ˆ a− .5 Working with Ladder Operators In the last section we have demonstrated the use of diﬀerential equation techniques. 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. one can express y = 2(ˆ− + a+ ) and d/dy = 2(ˆ− − a+ ) and. ˆ ˆ To put the following material in the proper modest perspective we may phrase it as an approach which rather than employing the coordinate y and the diﬀerential operator d/dy uses the operators 1 d 1 d y − .e. (4. f (ˆ+ ). a− = √ ˆ y + . and of the modes of the quantized electromagnetic ﬁeld.83) We will encounter below functions of a+ and a− . Here f (ν) (y0 ) denotes the ν-th derivative of f (y) taken at y = y0 .

94) This identity follows from (4. a a (4. The ﬁrst case is trivial. f (ˆ+ )] = f (1) (ˆ+ ) .89) (4.88) (4. a a a We like to state the following property of the functions f (ˆ± ) a f (ˆ− ) φ|ψ = φ|f (ˆ+ ) ψ .93) one can reduce all operators acting on some proper state function by operators acting on the state φ0 (y). (ˆ+ )n a+ ] = (ˆ+ )n [ˆ− .95) (4.91) (4. We note ˆ a− f (ˆ+ ) = [ˆ− . It holds a a euˆ f (ˆ+ ) φ0 (y) = f (ˆ+ + u) φ0 (y) . The following important property holds a a− f (ˆ+ ) = f (1) (ˆ+ ) + f (ˆ+ ) a− . The convergence of the Taylor expansion ascertains then that (4.90) holds for f (ˆ+ ). ˆ a We proceed by induction noticing ﬁrst that (4..4. a a a ˆ 1 a Let us assume that (4.87.90) we show that (4.87) (4. i. Note that ˆ a an immediate consequence of (4.88) we need to show actually [ˆ− . a− f (ˆ+ ) φ0 (y) = f (1) (ˆ+ ) φ0 (y) ˆ a a which follows from a− φo (y) = 0. Hence. a a − (4.92) f (y) = n=0 dn φn (y) = n=0 (ˆ+ )n a dn √ φ0 (y) n! (4.90) holds for fn . fn+1 (ˆ+ )] a a = [ˆ− . a+ ] + [ˆ− .5: Working with Creation and Annihilation Operators and similarly for f (ˆ− ).83) and can be proven for all powers fn by induction and then inferred for all proper functions f (ˆ± ). f1 (ˆ+ )] = [ˆ− . a a a 85 (4. An example is the so-called shift operator exp(uˆ− ). 4. f (ˆ+ )] + f (ˆ+ ) a− ˆ a a a a ˆ which implies that in order to prove (4. we will only assume operator functions acting on φ0 (y).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+ ). property (4. (ˆ+ )n ] a+ a a ˆ a a ˆ a a ˆ = (ˆ+ )n 1 + n (ˆ+ )n−1 a+ = (n + 1) (ˆ+ )n a 1 a ˆ a Since any function f (y) in the proper function space N1 can be expanded ∞ ∞ (1) (4. ˆ a a a ˆ In particular. For fn+1 follows then [ˆ− .90) holds for any function fn (ˆ+ ) = (ˆ+ )n which is a power of a a a+ .96) .90) To prove (4.88) is (ˆ− )n f (ˆ+ ) φo (y) = f (n) (ˆ+ ) φ0 (y) . a+ ] = 1 = f1 (ˆ+ ) .88) is a fundamental one and will be used in the following. a An important operator function is the exponential function. the second case follows from [ˆ− .e.

Below we will use the operator identity which follows for the choice of v = α in (4.96) is applied is a a a euˆ evˆ φ0 (y) = ev (ˆ − + + + u) a φ0 (y) = euv evˆ φ0 (y) . + ∗ a− ˆ a − u euzˆ euz + ∗ a− ˆ = (4.107) − 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. the solution of (4.105) e e vˆ+ a + v ∗ a− ˆ φ0 (y) = e − 1 vv ∗ 2 e v ∗ a− ˆ e vˆ+ a φ0 (y) vˆ+ a − v ∗ a− ˆ φ0 (y) = e 1 vv ∗ 2 e −v ∗ a− ˆ e vˆ+ a φ0 (y) .106) yields a eαˆ + ∗ a− ˆ a a eαˆ φ0 (y) = e−αα eαˆ φ0 (y) . (4. + (4.104) (4.100) ˆ ˆ To solve this diﬀerential equation we deﬁne C(u) = D(u) exp(−u2 /2) which leads to d ˆ D(u) = du zˆ+ + z ∗ a− a ˆ ˆ D(u) (4.98) The related operators exp[vˆ+ ± v ∗ a− ] play also an important role. one obtains a evˆ + − v ∗ a− ˆ a φ0 (y) = e− 2 vv evˆ e− v 1 ∗ + ∗ a− ˆ φ0 (y) (4. without explicitly stating this. 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) . + ∗ + (4. z ∈ C. + (4.106) Applying (4. euzˆ ˆ euz ∗ a− ˆ (4. We assume in the following a ˆ derivation.108) .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) .102) Similarly.98) for u = −α∗ and v = α yields e−α This together with (4. that the operators considered act on φ0 (y).101) ˆ ˆ ˆ C(u) obviously obeys C(0) = 1 This results in D(0) = 1 and.97) An example in which (4. 1 + + uz ∗ a− ).101) 1.105) a eαˆ + − α∗ a− ˆ φ0 (y) = e αα∗ 2 e−α ∗ a− ˆ a eαˆ φ0 (y) . + (4.99) ˆ C(u) = zˆ a zˆ+ + z ∗ a− a ˆ + +z a ˆ ∗ − a euzˆ euz ˆ − u C(u) . 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 a ν! (4.103) (4.90) and zz ∗ = 1 we can write this d du e uzˆ+ a e uz ∗ a− ˆ a − z ∗ a− .

provide the same values for Hn (0) as provided in Eq. 4. the expected from which follows by comparision of each term on both sides φµ |φν orthonormality property.113) Comparision of all terms on the l.95). + (4.4.s.111) This expression. Using (4.109) e = π4 n! n=0 √ Using (4. Setting y = 0 in (4. .µ=0 uµ v ν √ φµ |φν µ!ν! ∞ = µ=0 uµ v µ µ! (4. For this purpose we consider a a a a euˆ φ0 |evˆ φ0 = φ0 |euˆ evˆ φ0 + + − + (4. We start from the generating function (??) and replace according to (4.84. We want to derive now the values Hn (0).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! + (4.116) = δµ ν .112) Expanding both sides of this equation and using (4.h.s. and ˆ l. Similarly. of (4. we can reproduce by means of the generating function (4.111) the orthonormality properties of the wave functions φn (y). is the equivalent of the generating function (??). (??).111) yields π − 4 exp − 1 u2 2 a = euˆ φ0 (0) .84) yields ∞ ν.115) a where the latter step follows after expansion of evˆ and using a− φ0 (y) = 0.115) and using (4.114) where we have employed (4.h.s.s.110) √ y2 u2 + 2uy − 2 2 a = euˆ φ0 (y) . Expanding r. i.e.58) one obtains π− 4 1 ∞ ν=0 (−1)ν u2ν = 2ν ν! ∞ ν=0 uν √ φν (0) = ν! ∞ ν=0 uν Hν (0) √ 2ν π ν! (4.. We will derive the equivalent of a generating function and use it to rederive the values Hn (0) and the orthonormality properties of φn (y). and on the r.h. (4.95) one obtains a a a a euˆ φ0 |evˆ φ0 = φ0 |euv evˆ φ0 = euv evˆ φ0 |φ0 = euv − + + + − (4.98) and again (4.5: 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.58) the Hermite polynomials Hn (y) by the eigenstates φn (y) √ ∞ 1 ( 2t)n y2 /2 2yt − t2 √ e φn (y) (4.h. in the ladder operator calculus.

n = 0. . 1. of (4. (4.119) 2m 2 dp2 Accordingly. (4. i.88 Linear Harmonic Oscillator 4. is given by a function ψ(x). 2. .235).124) ).7) is given by (4. for an oscillator in a stationary state φn (x) as given by (2. For this purpose we employ the Schr¨dinger equation in the momentum o representation.7) is 1 2 m 2 ω 2 d2 p − .235). For example. using (2. 2. is given by the ˜ function ψ(p). we o note that the posed eigenvalue problem is formally identical to the Schr¨dinger equation in the o position representation as given by (4. 4. As to be expected.120) Multiplying this equation by 1/m2 ω 2 yields − where ˜ E = E / m2 ω 2 . 1.125) .6) and the Hamiltonian (4.123) (4.e.e. the momentum and position operators are p = p. ˆ and the Hamiltonian (4. .121).122) For a solution of the Schr¨dinger equation in the momentum representation.37). En = ω (n + 1 ) 2 d2 1 + m˜ 2 p2 ω 2 2m dp 2 2 ˜ ˜˜ φE (p) = E φE (p) (4..118) d2 1 2 + p dp2 2m ˜ ˜ φE (p) = E φE (p) . (4. ω = 1 / m2 ω . In the position representation the wave function is a function of x. i. (−i)n ˜ φn (p) = √ 2n n! m˜ ω π 1 4 exp − m˜ p2 ω 2 Hn ( m˜ ω p) .122). using (4. ˜ (4. .. the time-independent Schr¨dinger equation for the oscillator can be written o ˆ H = − mω 2 2 2 x = i ˆ d . according to (4. n = 0. . dp (4. . In the momentum representation the wave function is a function of p. (4. The momentum representation eigenfunctions are.2.121) ω (n + ˜ 1 ) 2 . (4.2). i.3). the probability density is P (x) = |φn (x)|2 .. The solutions can be stated readily exploiting the earlier results.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.e. the momentum and position operators are as stated in (4. the eigenvalues are identical to those determined in the position representation.6 Momentum Representation for the Harmonic Oscillator The description of the harmonic oscillator provided so far allows one to determine the probability density P (x) of ﬁnding an oscillator at position x. The eigenvalues. are ˜ En = or.

130) dx exp(−ipx/ ) φn (x) −∞ (4. i. of course..125.235)] absorbs the replacement dx → dy. hence. 1 ˜ φn (k) = √ 2π +∞ (4. ˜ We can now state the probability density Pn (p) for an oscillator with energy En to assume momentum p.134) Exercise 4.6: Momentum Representation or with (4.1: Demonstrate the validity of (4.128) We may also express the eigenstates (4.e. Deﬁning k = m˜ / p or ω 1 k = p (4.126) a phase factor (−i)n .132) dy exp(−iky) φn (y) −∞ (4.f.129) follows ky = px/ and. Proceed as follows.4. According to the theory of the momentum representation holds ˜ ˜ Pn (p) = |φn (p)|2 or 1 ˜ Pn (p) = n 2 n! 1 πm ω 1 2 (4. m ω (4. this allowed us to introduce in (4. φn (k) = We want to ﬁnally apply the relationship 1 ˜ φn (p) = √ 2π +∞ 2 (−i)n −k √ e 2 Hn (k) . . From (4. through direct integration. (4.126)..30) and (4.126) Normalized eigenfunctions are. 4.131) in the present case employing dimensionless units. instead of (4. deﬁned only up to an arbitrary phase factor eiα . i. the Jacobian. 2n n! π (4.126) in terms of dimensionless units following the procedure adopted for the position representation where we employed the substitution (4. m ω (4.30) and (2. This implies the property of Hermite polynomials (−i)n e− k2 2 +∞ Hn (k) = −∞ dy exp(−iky) e− y2 2 Hn (y) (4. the correctness of the phase factors (−i)n .133) where we note that a change of the normalization of φn (y) [c.127) exp − p2 m ω 2 Hn ( 1 p) .134).6.e.122) (−i)n ˜ φn (p) = √ 2n n! 1 πm ω 1 4 89 exp − p2 2m ω Hn ( 1 p) .58).30) to obtain (4.129) m ω one obtains. α ∈ R.

+∞ Linear Harmonic Oscillator dy exp(−iky) g(z. a pendulum swinging at small amplitudes. for example. z) = exp −2ikz + z 2 − k 2 /2 .— III.135) g(k. carries out a periodic motion described by y(t) = Re y0 eiωt (4.Grynberg (John Wiley & Sons. Inc.138) We will show that a harmonic oscillator prepared in such a state will remain forever in such a state.138).. 192 3 .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. coherent states. 4. The question arises if similar states exist also for the quantum oscillator. or Glauber states of the harmonic oscillator and they play.194. New York. z) = √ 2π where f (k. J. φ(α) | φ(α) = . 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. The motion of the probability distribution of such state is presented in Fig. 1989) which discusses quasi-classical states of the free electromagnetic ﬁeld in Sect.1 showing in its top row the attributes of such state just discussed. namely through the eigenvalue problem for normalized states a− φ(α) (y) = α φ(α) (y) . 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 . We ﬁrst construct the solution of (4.138) yields using a− φn (y) = n φn−1 (y) ˆ ∞ n=0 √ n d(α) φn−1 (y) − α d(α) φn (y) n n = 0 (4. The answer is ‘yes’. Inserting this expansion into √ (4. 2. for example.4.140) An excellent textbook is Photons and Atoms: Introduction to Quantum Electrodynamics by C. Such states are referred to as quasi-classical states. ˆ α ∈ R.Chen-Tannoudji. except that α changes periodically in time. (4.196) of Hermite polynomials and equate coeﬃcients of equal powers of z to prove (4. For this purpose we assume such state exists and expand ∞ ∞ φ(α) (y) = n=0 d(α) φn (y) . pp. n n=0 d(α) n 2 = 1 (4. The quasi-classical states can be obtained by generalizing the construction of the ground state of the oscillator.Dupont-Roc. z) = exp 2yz − z 2 − y 2 /2 (4. We want to construct and characterize such states. y) −∞ (4.139) where we have added the condition that the states be normalized. and G. 4.90 (a) Prove ﬁrst the property 1 f (k.7 Quasi-Classical States of the Harmonic Oscillator A classical harmonic oscillator.C.134).136) (b) Employ the generating function (2.

This interpretation justiﬁes the choice of α ∈ R in (4.145) Using the generating function in the form (4. n! (4.139)] 1 = n=0 d(α) n 2 = |c|2 n=0 α2 2 = |c|2 eα . hence. n! (4. n! (4. Comparing (4.148) The identity (4.150) for α presently real.149) − α∗ a− ˆ (4. n n+1 (4.4.84) φ(α) (y) = exp − α2 2 ∞ n=0 αα∗ (αˆ+ ) a a+ φ0 (y) = e− 2 eαˆ φ0 (y) .141) Because of the linear independence of the states φn (y) all coeﬃcients multiplying φn (y) must vanish and on obtains the recursion relationship (α) dn+1 = √ α d(α) .7: Quasi-Classical States or ∞ n=0 91 √ n + 1 dn+1 − α d(α) n (α) φn (y) = 0 .147) allows one to interprete T (α) as an operator √ which shifts the ground state by 2α.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. [π]− 4 e− one can write φ(α) (y) = [π]− 4 exp 1 1 y2 2 + √ 2 α y − α2 = exp − α2 2 ∞ n=0 αn √ φn (y) .144) Normalization requires c = exp(−α2 /2) and. (4. n! n (4. the quasi-classical states are φ (α) (y) = exp α2 − 2 ∞ n=0 αn √ φn (y) .146) √ 1 − (y − 2 α)2 2 (4. However.109).. i. One can also write (4. we will see below that α ∈ C are also admissible and will provide a corresponding interpretation.e. .138).143) for a constant c which is determined through the normalization condition [see (4.147) √ which identiﬁes this state as a ground state of the oscillator displaced by 2α.142) The solution is αn d(α) = √ c n n! ∞ ∞ (4.149) with (4.145) using (4.

is the adjoint of the operator function on the l.156) (4.95) to linear combinations αˆ+ + βˆ− and reads then a a f (αˆ+ + βˆ− ) φ|ψ = φ|f (α∗ a− + β ∗ a+ ) ψ .145) is replaced by a complex number αu which at times t1 = t0 + 2πn/ω.s. Following the procedure adopted in determining the propagator of the harmonic oscillator [see (4.145) shows that α in (4.158) Comparision of this expression with (4. t1 ) = π − 4 u 2 exp − + 2yαu − 2 2 E (4. t1 ) = u exp 1 2 α2 − 2 (α u)n √ φn (y) . In particular.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. . a a ˆ ˆ (4. n = 0. one can write in case of T + (α) T + (α) + = T (α) = eα ∗ a− ˆ − αˆ+ a a = e − (αˆ + − α∗ a− ) ˆ . 1.152) T + (α) f |T + (α) g = f |T (α) T + (α) g = f |g . α ∈ C (4.155) (4.67.153) This is obviously the inverse of T + (α) as deﬁned in (4. .e.109) serves again to simplify this sum √ y2 1 1 α2 (u2 + 1) φ(α) (y.152) Since the operator function on the r. it holds T + (α) φ0 |T + (α) φ0 = f |T (α) T + (α) g = φ0 |φ0 . 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. n! u = e−iω(t1 − t0 ) .145) as the initial state φ(α) (y.157) (y.s. becomes real.h. (4..68)] one can write the time-dependent solution with (4. Relationship (4.150) for complex α obeys the unitary property T + (α) T (α) = 1 1 (4. the shift operator leaves the ground state normalized. i.160) .h.154) . .92 Linear Harmonic Oscillator The shift operator as deﬁned in (4. t1 ) = exp or φ (α) − α2 2 ∞ exp n=0 −i ω ( n + ∞ n=0 1 ) ( t1 − t0 ) 2 αn √ φn (y) n! (4. (4.151) which follows from a generalization of (4.

in general.138): a real α corresponds to a displacement such that initially the oscillator is at rest. The diagram illustrates that the wave function retains its Gaussian shape with constant width for all times. being displaced around the center with period 2π/ω and.163) also through the shift operator.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. this characterization corresponds closely to that of the possible initial states of a classical oscillator.161) and is found to describe a displaced Gaussian. t1 ) = π − 4 exp − [y − y0 (t1 )]2 + iy k0 (t1 ) − iφ0 (t1 ) − ω(t1 − t0 ) 2 2 . In Fig. This shows clearly that the Glauber state is a close analogue to the classical oscillator.1 (top row) we present the probability distribution of the Glauber state for various instances in time. 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. Comparing (4. obviously. a momentum factor and a phase factor.164) where according to (4. 4. Our interpretation 2 explains also the meaning of a complex α in (4. 4.159) is then 1 1 i φ(α) (y.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. except that it is not pointlike.165) . The time-dependent wave function (4. One can express (4. (4.162) 2 with a periodic change.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 (4.138.148. experiences a change of its width (relative to . 4.147).150) with (4. t1 ) = u 2 T + (α u) φ0 (y) (4. (4. It is of interest to note that the displacement y0 of the Gaussian as well as the momentum k0 and phase φ0 associated with (4.163) This solution corresponds to the initial state given by (4. a complex α corresponds to a displaced oscillator which has initially a non-vanishing velocity. Arbitrary Gaussian Wave Packet Moving in a Harmonic Potential We want to demonstrate now that any initial state described by a Gaussian shows a time dependence very similar to that of the Glauber states in that such state remains Gaussian. moving solely its center of mass in an oscillator fashion around the minimum of the harmonic potential.158) allows one to write 1 φ(α) (y.4.

t0 ) = [πσ 2 ]− 4 exp we expand ∞ 1 − (y − ao )2 2σ 2 (4. that of the Glauber states) with a period π/ω. 2σ 2 q = a .169) .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. 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. 4. σ|y. To describe such state which satisﬁes the initial condition φ(ao . σ|y. This behaviour is illustrated in Fig.167) One can derive for the expansion coeﬃcients +∞ dn = −∞ dy φn (y) φ(ao .166) 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) (4.1: Time dependence of Gaussian wave packets in harmonic oscillator potential.1 (second and third row).168) dy e−p(y−q) Hn (y) where p = 1 + σ2 . t0 ) = n=0 dn φn (y) . 1 + σ2 (4. (4. σ|y.

unfortunately very expensive. in agreement with the expected result (4.173) which is. therefore.166) in analogy to (4.7: Quasi-Classical States The solution of integral (4. a ground state shifted by ao = 2α. p−1 1 − σ4 t = 1 − σ 2 −iω(t1 −to ) e 1 + σ2 (4. For this purpose we note that according to the explicit expresssion (??) for Hn (y) the leading power of the Hermite polynomial is Hn (y) = (2 y)n + and. this 3 volume integral table is likely the most complete today. This expansion can be written φ(ao .I. Marichev (Wiley. Following the strategy adopted previously one can express the time-dependent state corresponding to the initial condition (4. σ|y. New York. t1 ) = √ 2 × exp − 1 (y 2 + yo ) 1+t2 − 2 1−t 2 1 1 √ √ pσ π 1 − t2 × t 1 − t2 (4.157) as follows φ(ao .175) ∞ tn n=0 2n n! Hn (yo ) e−yo /2 Hn (y) e−y 2 2 /2 The generating function (4.171) lim p−1 p n 2 Hn (q p ) = lim (2q)n = an . i. 1 a2 2 1 + σ2 + 1 2 2 y o + 2 y yo This integral can be found in Integrals and Series. σ→1 O(y n−2 ) for n ≥ 2 1 for n = 0. σ|y. Brychkov.P.4. in fact.172) One can then conclude e−a /4 n lim dn = √ a σ→1 2n n! (4. p−1 95 (4.177) − 1 ω(t1 − to ) 2 . 2 by A. In this limit the coeﬃcients dn should√ become identical to the coeﬃcients (4. t1 ) = √ × where yo = q p a = √ .143). a worthy successor of the famous Gradshteyn. Prudnikov.168) is4 e−aq/2 (p − 1)n/2 dn = √ Hn (q 2n n!σ p(n+1)/2 p ). 1990).174) Hn (q p p−1 ) φn (y) .143) of the Glauber state. and O.. 1 (4. Yu. σ|y.e. t1 ) = e−aq/2 √ pσ ∞ n=0 e −iω(n+ 1 )(t1 −to ) √ 2 1 2n n! p−1 p n/2 (4. namely $750 4 . vol. σ→1 p−1 2 (4.170) In order to check the resulting coeﬃcients we consider the limit σ → 1.76) permits us to write this φ(ao .176) 1 √ pσ π a 2 exp − 1 1+σ2 + 1 yo − 1 ω(t1 − to ) 2 2 2 2 × (4.

96 Linear Harmonic Oscillator .

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

therefore.6) (5. they conserve a · b = 3 aj bj ..e.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.k.5) This equation states the key characteristic of rotation matrices. then the ﬁngers of your ﬁst point in the direction of the rotation. i. In the latter case the determinant of R(ϑ = (0. i. We assume the convention that rotations are right-handed.7) Let us consider brieﬂy an example to illustrate rotational transformations and to interpret the sign of detR(ϑ). x2 . (5.1) In Cartesian coordinates this reads 3 xk = j=1 [R(ϑ)]kj xj k = 1. A rotation around the x3 -axis by an angle ϕ is described by the matrix cosϕ − sinϕ 0 R(ϑ = (0.6) follows immediatety for the determinant of R(ϑ) using detA B = detA detB and detAT = detA detR(ϑ) = ±1 . 2. can be represented by 3 × 3 matrices R(ϑ). are linear and. (5. in case of a prefactor −1 a rotation and an inversion at the origin. the matrix represents a proper rotation. From (5. ϑ parametrizes the rotations uniquely as long as |ϑ| < 2π.e. (5. i.e. it follows 3 [R(ϑ)]jk [R(ϑ)]j = δk j=1 .7) implies that a rotation is associated with an inversion. (5. in Cartesian coordinates r = (x1 . namely around an axis given by the direction of ϑ and by an angle |ϑ|.98 Properties of Rotations in R Theory of Angular Momentum and Spin Rotational transformations of vectors r ∈ R . (5.3) Since this holds for any a and b. 0. if the thumb in your right hand ﬁst points in the direction of ϑ. =1 [R(ϑ)]jk [R(ϑ)]j ak b = j=1 aj bj . a minus sign in Eq. 3 . x3 )T . In the following we want to exclude inversions and consider only rotation matrices with detR = 1. 0. (5. . It holds then j=1 3 3 a ·b = j. ϕ)T ) = ± sinϕ cosϕ 0 (5.2) Rotations conserve the scalar product between any pair of vectors a and b. where ϑ denotes the rotation. Let us deﬁne the rotated vector as r = R(ϑ) r .. ϕ)T ) is negative.8) 0 0 1 In case of a prefactor +1.

T R1 R1 −1 = 1 −1 = 1 1 1 (5.5. 3 × 3–matrix. (ii) The group element ‘e’ is played by the 3 × 3 identity matrix 1 1. b ∈ G the product c = a ◦ b is also in G.10) detR3 = det ( R1 R2 ) = detR1 detR2 = 1 . 99 (5. hence. We would like to point out the obvious property that for all elements R of SO(3) holds R−1 = RT (5. It follows that R3 is also an element of SO(3). R RT = RT R = 1 detR = 1 } 1. (iii) ∀a. (5.11) (iii) From detR1 = 0 follows that R1 is non-singular and. R2 ∈ SO(3) and do not write out “◦” explicitly since it represents the matrix product.9) This set of matrices is closed under matrix multiplication and. 1 T Similarly. it holds 1. We need to demonstrate that this inverse belongs also to −1 T SO(3). (i) Obviously. For this purpose we must demonstrate that the group properties (i–iv) hold.9) forms a group. there exists a real 3 × 3–matrix −1 R1 which is the inverse of R1 . a ∈ G → ∃a−1 ∈ G with a ◦ a−1 = a−1 ◦ a = e. R3 = R1 R2 is a real. one can show R3 R3 = 1 Furthermore. Since (R1 )T = (R1 )−1 it follows −1 −1 −1 T (R1 )T R1 = (R1 )−1 R1 = −1 which implies R1 ∈ SO(3). in fact.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 holds T T T T R3 R3 = ( R1 R2 )T R1 R2 = R2 R1 R1 R2 = R2 R2 = 1 .13) . In the following we will assume R1 . (ii) There exists an element e in G with the property ∀a ∈ G → e ◦ a = a ◦ e = a. (5. forms a group G satisfying the axioms (i) For any pair a. 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.12) (iv) Since the associative law holds for multiplication of any square matrices this property holds for the elements of SO(3). We want to prove now that SO(3) as deﬁned in (5.

14) together with matrix multiplication as the binary operation ‘◦’. compact representation of R(ϑ) which will be of general use and.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. (c) the set of four rotations in the x1 . k = exp ( −ϑk Lk ) (5. presently. M3 . J3 as follows i R(ϑ) = exp − ϑ · J (5.15) Below we will construct appropriate matrices Jk . x2 –plane {rotation by Oo .e. i. J2 . i.100 Theory of Angular Momentum and Spin Exercise 5. mappings between the sets which conserve the group properties: (a) the set of real and imaginary numbers {1.18) (5. we want to assume that they exist. rotation by 270o } and consecutive execution of rotation as the binary operation ‘◦’.13) R(ϑk ek )−1 = R(ϑk ek )T = [exp ( ϑk Lk )]T = exp ϑk LT ˆ ˆ k and. (b) the set of matrices {M1 . −1 0 0 −1 . rotation by 180o .20) (5. rotation by 90o . 0 −1 1 0 (5. due to the uniqueness of the inverse. identify subgroups. −i} together with multiplication as the binary operation ‘◦’.19) (5. we arrive at the property exp ϑk LT k from which follows LT = −Lk .16) must be antisymmetric. Lie Algebra We want to consider now a most convenient. 0 1 −1 0 . Generators.1. To demonstrate this property consider rotations with ϑ = ϑk ek where ek ˆ ˆ is a unit vector in the direction of the Cartesian k–axis. −1. We want to show that for the property R RT = 1 to hold.17) . 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. M4 } = 1 0 0 1 . in fact. This representation expresses rotation matrices in terms of three 3 × 3– matrices J1 . the matrices i Lk = − Jk (5. M2 . Lie Group. establish if 1-1 homomorphic mappings exist.

e. We assume now that we parameterize rotations through a three-dimensional rotation vector ϑ = (ϑ1 . 0. Let us use this deﬁnition to evaluate L3 .(5. This property gives A then only three independent matrix elements. 0. 0.27) .25) 0 −ϑ3 0 0 −1 0 ϑ3 0 0 = 1 0 0 .21) −b −c 0 It is this the reason why one can expect that three-dimensional rotation matrices can be expressed through (5. ϑ2 . i. ϑ3 )T such that for ϑ = (0. (5.8) we can state cosϑ3 −sinϑ3 0 1 −ϑ3 0 1 0 . ϑ3 )T ] = lim sinϑ3 cosϑ3 0 = ϑ3 ϑ3 →0 ϑ3 →0 0 0 1 0 0 1 Insertion into (5. Using the deﬁnition of partial derivatives we can state L1 = lim ϑ−1 R(ϑ = (ϑ1 . 0)T . (5. 0 0 0 0 0 0 (5. 0. ϑ3 and three matrices L1 .15) as the derivatives of R(ϑ) with respect to the Cartesian components ϑk taken at ϑ = (0.22) for k = 3 yields L3 = ϑ−1 3 One obtains in this way 0 0 0 i − J1 = L1 = 0 0 −1 0 1 0 0 0 1 i − J2 = L2 = 0 0 0 −1 0 0 0 −1 0 i 1 0 0 − J3 = L3 = 0 0 0 The matrices Lk are called the generators of the group SO(3).15) with three real parameters ϑ1 . A must have the form 0 a b A = −a 0 c . Using (5.1.1: Matrix Representation of SO(3) 101 We can conclude then that if elements of SO(3) can be expressed as R = eA .5.23) (5.22) and similar for L2 and L3 . lim R[(0. L2 .24) (5. the real matrix A is anti-symmetric. 0)T the rotation is the identity.2: Determine the generator L1 according to Eq.. Exercise 5.26) (5.22). ϑ2 . L3 which are independent of the rotation angles. 0)T ) − 1 1 1 ϑ1 →0 (5. One can determine then the matrices Lk deﬁned in (5.

and x3 -axes. B].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. B]. 0)T ) = exp( 1 L1 ). (5.8) in case of ϑ = (0. We want to consider. 0. commutators of commutators of A and B. Z2 = 2 [A.29) where Z1 = A + B and where the remaining operators Zn are commutators of A and B.32) For the matrices Jk = i Lk which.22) implies. that the operators (5. m = 3 .. it must hold 3 [Lk . e.g. = 2. the elements of if any two indices are identical for k = 1.30) is called the Lie algebra. the Lie algebra of SO(3) can be written [Lk .. The latter property implies that for any ϑ1 and ϑ2 there exists a ϑ3 such that exp i − ϑ1 · J exp i − ϑ2 · J = exp i − ϑ3 · J (5. (5. R(ϑ = (0. 0.102 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.e. [A. in fact. In case of the group SO(3) the structure constants are particularly simple. 2 . Lie groups can have any number of generators.e.28) The construction of the generators through the limit taken in Eq. An answer can be found considering the Baker-Campbell-Hausdorf expansion ∞ exp( λA ) exp( λB ) = exp( n=1 λn Zn ) (5. only the latter example and . i. Finite rotations can be obtained by applications of many inﬁnitesimal rotations of the type R(ϑ = ( 1 . 3 )T ) = exp( 3 L3 ). (5.15) and their generators obey the property (5.30) are called Lie groups. L ] = m=1 fk m Lm . In fact. ϕ)T . i. B] ). B]] + [[A. The question is then. are related to angular momentum operators. 0)T ) = exp( 2 L2 ) and R(ϑ = (0. J ] = i k m Jm . Z3 = 12 ( [A. (5. = 2. and the constants fk m are called the structure constants. however. From this result one can conclude that expressions of the type (5.g. holds the algebra [Jk . 0.15). the property (5. m = 3 for any even permutation of k = 1. The expression for Zn are actually rather complicated. 1 1 e. etc. if the resulting products of exponential operators can be expressed in the form (5. however.30) Groups with the property that their elements can be expressed like (5. that the generators represent inﬁnitesimal rotations with 1 >> |ϑ| around the x1 -.31) where k m are the elements which are 0 1 k m = 1 −1 of the totally antisymmetric 3-dimensional tensor.15) obey the group property. as we see later.33) We want to demonstrate now that exp(ϑ · L) yields the known rotational transformations. Of course. the matrix (5.. L ] = k m Lm (5. three being a special case. x2 -. m = 3 for any odd permutation of k = 1. = 2.

cos ϕ − sin ϕ 0 = sin ϕ cos ϕ 0 0 0 1 (5. accordingly. that the matrix ϕL3 .3: Show that the generators Lk of SO(3) obey the commutation relationship (5.35) To determine expA = ∞ ν ν=1 A /ν! we split its Taylor expansion into even and odd powers 1 A2n + (2n)! 2 ∞ n=0 ∞ eA = n=0 1 A2n+1 .34.39) Recognizing the Taylor expansions of cos ϕ and sin ϕ one obtains eA = and. 0 −1 1 0 = (−1)n 0 −1 1 0 (5. We note.5.41) cos ϕ − sin ϕ sin ϕ cos ϕ (5.31).1: Matrix Representation of SO(3) determine exp(ϕL3 ).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 . eϕL3 which agrees with (5. (5. 5.35). can be written ϕL3 = 0 103 A 0 0 0 0 . (2n + 1)! (5. according to (5. Exercise 5. (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 . (5.36) The property 0 −1 1 0 allows one to write 0 −1 1 0 and. A = ϕ 0 −1 1 0 .40) . ∞ 2n = − 1 0 0 1 2n+1 (5.37) = (−1)n 1 0 0 1 . using (5.8).1.

Show that in this case holds Generators One can assume then that R(ϑ) should also form a Lie group. (5. Since this holds for any ψ(r ) one can conclude ρ R(ϑ1 ) ρ R(ϑ2 ) group property (5.46) (5.e. In particular. namely R(ϑ)ψ(r ) = ψ(R−1 (ϑ) r ) .2 Function space representation of the group SO(3) In this section we will investigate how rotational transformations act on single particle wavefunctions ψ(r). Exercise 5.e. i. its elements can be represented through an exponential R(ϑ) = exp i − ϑ·J (5. we will learn that transformation patterns and invariances are connected with the angular momentum states of quantum mechanics. hence. the transformation R(ϑ): C∞ ( ) → C∞ ( ) is related to the transformation R(ϑ): R → R .. = ψ([ R(ϑ1 ) R(ϑ2 ) ]−1 r) = ρ R(ϑ2 ) ψ([R(ϑ1 )]−1 r) (5. complex-valued functions over the 3-dimensional space R which can be diﬀerentiated inﬁnitely often. i. i.47) = ρ R(ϑ1 ) R(ϑ2 ) . Deﬁnition We ﬁrst deﬁne how a rotational transformation acts on a wavefunction ψ(r ). a relationship which we like to express as follows R(ϑ) = ρ R(ϑ) .e.44) holds. for A. In analogy to R(ϑ) being a linear map R → R .1: Assume the deﬁnition ρ ρ (A B) = ρ (B) ρ (A).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) . the R(ϑ) ψ(r) = ψ(R(ϑ) r ).45) (5. in fact.48) .104 Theory of Angular Momentum and Spin 5.42) Obviously. (5. (5. B ∈ SO(3) holds ρ(A B) = ρ(A) ρ(B) . For this purpose we require stringent continuity properties of the wavefunction: the wavefunctions under consideration must be elements of the set C∞ ( ). we deﬁne rotations R(ϑ) as linear maps C∞ ( ) → C∞ ( ).43) ρ( ) conserves the group property of SO(3). one isomorphic to SO(3) and.2.

53–5.5. (5.51) Inserting this into (5.2: (a) Derive the generators Jk . The generators Jk can be determined by applying (5. i − J3 f (r) = = Using (5. J2 ].49). ϑ3 )T ) f (r) − f (r) 3 (5.53) (5. 0 0 1 x3 x3 (5. show that (5.56) We like to verify this property for [J1 . (5. 0. 5. −ϑ3 ) r ≈ −ϑ3 1 0 x2 = −ϑ3 x1 + x2 . but also the Lie algebra of the generators.57) Exercise 5. 0.e. One obtains using (5. .49) (note the minus sign of −ϑ1 which originates from the inverse of the rotation in Eq. J ] = i k m Jm . In fact.54) (5. in case of J3 . it holds [Jk . 0. i.56) holds for [J2 . J1 ].52) ( x2 ∂1 − x1 ∂2 ) f (r) .42)) and similarly for J2 and J3 .23.50) lim ϑ−1 ( f ( R(0. (5. (5. 3 1 ϑ3 0 x1 x1 + ϑ3 x2 R(0. J3 ] and [J3 .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. 0. 2 by means of limits as suggested in Eq. 0)T ) − 1 1 ϑ1 →0 105 (5.2: Function Space Representation of SO(3) where the generators can be determined in analogy to Eq.24) one can expand ϑ3 →0 ϑ3 →0 lim ϑ−1 R(ϑ = (0.49) to a function f (r).. −ϑ3 ) r ) − f (r) ) .(5.55) Not only is the group property of SO(3) conserved by ρ( ).22) according to i − J1 = lim ϑ−1 R(ϑ = (−ϑ1 . k = 1. (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) . J2 ] f (r) = − = − 2[ 2 [− i J .55) and f (r) = f [J1 .2. 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 .

φ) m (θ. The commutation relationships imply that no simultaneous eigenstates of all three Jk exist. Equation (5. 3 .2.65) .61) simultaneous eigenstates of J 2 and of one of the Jk . usually chosen as J3 . 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. (5.53–5. .58) i = r ×p ˆ (5.64) mY m (θ.56) one obtains [J 2 .56) are the famous commutation relationships of the quantium mechanical angular momentum operators. φ) as well as the eﬀect of the so-called raising and lowering operators J± = J1 ± i J2 (5. using the momentum operator p = ˆ J = r × i . φ) (5. 2. Jk ] = 0 . k = 1. C] + [A.62) J1 [J1 .63) (5. i. identiﬁes J as the quantum mechanical angular momentum operator. and x3 = r cosθ. J3 ] + [J2 . x2 = r sinθ sinφ. φ). These eigenstates are the well-known spherical harmonics Y m (θ.e. 5. J3 ] J2 −i J1 J2 − i J2 J1 + i J2 J1 + i J1 J2 = 0 According to (5. J3 ] = = = 2 2 [J1 + J2 . φ) = = 2 ( + 1) Y m (θ. according to the correspondence principle between classical and quantum mechanics.. [J 2 . C]B and (5.3: Consider a wave function ψ(φ) which depends only on the azimutal angle φ of the spherical coordinate system (r.59) which. (5. φ). k = 1. 2. J3 ] J1 + J2 [J2 . J3 ] + [J1 . θ. can be found. the system related to the Cartesian coordinates through x1 = r sinθ cosφ. We will show now that the properties of spherical harmonics J2Y J3 Y m (θ. J3 ] (5.61) We demonstrate this property for k = 3.55) can be written as a vector product of r and i − J = −r × From this follows. C] = A[B. Using [AB. However.106 Theory of Angular Momentum and Spin Exercise 5. Show that for J3 as deﬁned above holds exp( iα J3 ) ψ(φ) = ψ(φ + α). 3. i.e. To show this we ﬁrst note that (5.60) commutes with all three Jk .

(∗) (5. and since we have already shown | − 1 = (iβ −1 )−1 L− | we can state L3 | − 1 = −i( − 1)| − 1 . L− ] = [L3 .1 Let Lk . We note L3 L− | = ( [L3 .5. k = 1. An Important Theorem Theorem 1. φ) are a consequence of the Lie algebra (5. L1 − iL2 ] = −2iL3 and L+ | = 0 one obtains L+ L− | = −2iL3 | = −2 | . α −1 β −1 = 2 .68) (5. φ) − m (θ. in fact.67) (5. L ] = k m Lm and let L± = L1 ± i L2 . i. φ) 107 = = 0 = = 0 ( + m + 1)( − m) Y ( + m + 1)( − m) Y m+1 (θ. L− ] = [L1 + iL2 . 2. i. Using [L3 . From this follows that L− | is an eigenstate of L3 . For this purpose we ﬁrst demonstrate that the property holds for m = − 1 by considering (i) for the state L− | ∼ | − 1 .71) To prove this theorem we ﬁrst show by induction that (i) holds. φ) (θ. L+ raises the m-value of | − 1 from m = − 1 to m = .74) (5. φ) m+1 (θ.e.73) (5. We consider L+ L− | m = ( [L+ . We will show that this property holds then also for m − 1. φ) (5.70) (5. L2 = L2 + L2 + L2 . Let there exist states | in 1 2 3 H with the property L+ | L3 | then the states deﬁned through L− | m + 1 = −i βm | m have the following properties (i) L+ | m (ii) (iii) L3 | m L2 | m = −i αm | m + 1 = −i m | m = − ( + 1) | m (5.66) (5. L1 − iL2 ] = L2 + iL1 = i (L1 − iL2 ) = iL− and L3 | = −i | we obtain L3 L− | = ( iL− + L− L3 )| = i(− + 1) L− | .3: Angular Momentum Operators on the spherical harmonics. L− ] + L− L+ )| m = ( −2iL3 + . The deﬁnitions of the coeﬃcients α −1 and β −1 yield L+ L− | = L+ ( iβ −1 | − 1 = − α −1 β −1 | .56).76) We now show that property (ii) also holds for m = − 1 proceeding in a similar way. L− ] + L− L3 ) | . Noting [L+ . 3 be operators acting on a Hilbert space H which obey the algebra [Lk . J+ Y J− Y J− Y m (θ. It holds L+ L− | = (L+ L− − L− L+ + L− L+ )| = ([L+ . Continuing our proof through induction we assume now that property (i) holds for m. For this purpose we prove the following theorem. namely. L− ] + L− L+ )| .69) J+ Y (θ.e..75) (5. The arguments are very similar to the ones used above and we can be brief.72) = 0 = −i | (5.

This implies that (i) holds for m − 1. 2 We note that Eqs.. 5. from L+ L− | m = − αm−1 βm−1 | m follows the recursion relationship αm−1 βm−1 = 2m + αm βm . We note that we can write L2 = 1 1 1 1 2 2 2 2 L+ L− + 2 L− L+ + L3 . We will have various opportunities to employ the Theorem just derived.83) (5. (5.72. in terms of spherical coordinates deﬁned through x1 x2 x3 = = = r sin θ cos φ r sin θ sin φ r cos θ (5.e.55).69). it holds αm βm = 2 one obtains αm βm = = ( + 1) k=m+1 k . in particular. given in (5.78) ( + m + 1)( − m) ( + m + 1)( − m) . This completes the proof of the theorem.82) (5. A ﬁrst application will involve the construction of the eigenstates of the angular momentum operators as deﬁned in (5.77) We will show now that if (ii) holds for m. For this purpose we need to express the angular momentum operators in terms of spherical coordinates (r. (5.66.64). 5. Again the argumemts are similar to the ones used above. yield properties (5. 5.81) i. Angular Momentum Operators in Spherical Coordinates We want to express now the generators Jk .63.79) read L+ | m L− | m + 1 = −i = −i ( + m + 1)( − m) | m + 1 ( + m + 1)( − m) | m .79) We can now show that (iii) holds for all proper m–values.84) . 5.e.73. it also holds for m − 1. (∗∗) (5. θ. φ. 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 .. Before we can ﬁnally show that property (iii) holds as well for all − ≤ m ≤ we need to determine the coeﬃcients αm and βm . One can normalize the states | m such that αm = βm .80) (5.53–5.108 Theory of Angular Momentum and Spin L− L+ )| m = (−2m − αm βm )| m . ﬁnally αm = βm = (5.70. i. We note L3 L− | m = ( [L3 . Using the familiar formula + m − n k=0 k = n (n + 1)/2 − (m + 1) m = ( + 1) m − (m + 1) m (5. L− ] + L− L3 )| m = ( iL− − i m L− )| m = −i (m − 1) L− | m from which we can deduce L3 | m − 1 = −i(m − 1)| m − 1 . 5. We can deduce from (∗) and (∗∗) α α α −1 β −1 −2 β −2 −3 β −3 = 2 = 2( − 1 + ) = 2( − 2 + · · −1 + ) Obviously.

φ) = (5. (L2 )2 = − cos φ ∂ ∂ + cot θ sin φ ∂θ ∂φ 2 . applying repeatedly the chain rule. using (5.84 ) one obtains. using (5. We like to express now the operators J3 and J 2 .87) holds J3 = To determine J 2 we note. in terms of spherical coordinates.88) (5.90) This yields. (5.89) = ∂ x2 x2 ∂ x2 + x2 ∂ x3 − 33 − x2 1 3 2 x1 ∂ tan φ r ∂ cos θ r ∂ cos θ f (θ. φ) (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.91) which agrees with expression (5.5. φ).60). 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 ∂ = − . (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 . L1 f (θ. for example. Using (5. L1 f (θ. the latter deﬁned in (5. φ) = (x3 ∂2 − x2 ∂3 )f (θ. using ∂/∂tanφ = cos2 φ ∂/∂φ and ∂/∂cosθ = −(1/sinθ) ∂/∂θ.87) To derive these properties we consider.93) and similarly. According to (5.92) (5.85).86).85) (5.3: Angular Momentum Operators 109 We ﬁrst like to demonstrate that the generators Jk . actually. involve only the angular variables θ and φ and not the radius r. φ) = ( x3 ∂2 − x2 ∂3 ) f (θ. cos θ ∂ cos2 φ + sin θ cos φ ∂φ cos2 θ sin θ sin φ ∂ sin θ sin φ ∂ + + sin2 θ f (θ.85). φ) ∂x ∂x3 ∂ tan φ =0 (5.82–5. ∂φ = sin φ (5.86) (5. In fact. φ) sin θ ∂θ sin θ ∂θ L1 f (θ.

85. sphere S2 .97) Kinetic Energy Operator The kinetic energy of a classical particle can be expressed J2 p2 p2 = r + class 2m 2m 2m r2 (5. namely on the space of complex-valued functions on the unit 3-dim. 5.86. ∂θ ∂φ ∂2 ∂ ∂2 + cotθ + ( cot2 θ + 1 ) 2 . One ﬁrst identiﬁes the Hilbert space H on which the generators Lk operate.64) can be constructed as follows. i.85.87).4 Angular Momentum Eigenstates According to the theorem on page 107 above the eigenfunctions (5.98) where Jclass = r × mv is the angular momentum and pr = mr is the radial momentum.110 = cos2 φ Theory of Angular Momentum and Spin ∂2 1 ∂ ∂2 + sinφ cosφ − 2 cotθ sinφ cosφ ∂θ2 ∂φ ∂θ∂φ sin2 θ 2 ∂ ∂ + cotθ sin2 φ + cot2 θ sin2 φ 2 . 5. 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 + .97).96) and the relationship between Lk and Jk as given in (5. 1 sin2 θ sinθ ∂ ∂θ (5.87).95) sinθ ∂ ∂θ = ∂2 ∂ + cotθ . (5. 5.94) It follows (L1 )2 + (L2 )2 + (L3 )2 = With cot2 θ + 1 = 1/sin2 θ.86. The functions in C∞ (S ) have real variables .101) 1 ∂2 r r ∂r2 and comparision with (5. 2m r ∂r2 2m r2 2 (5. 5. C∞ (S ). 5. These generators act on a subspace of C∞ ( ).100) (5. ∂θ2 ∂θ ∂φ (5. 2 ∂θ ∂θ (5. The operators in the present case are Lk = − i Jk where the Jk are diﬀerential operators in θ and φ given in (5.99) = = ∂2 2 ∂ 1 + + 2 2 2 ∂r r ∂r r sin θ ∂2 2 ∂ + 2 ∂r r ∂r sinθ ∂ ∂θ 2 + ∂2 ∂φ2 (5. one can conclude J2 = − 2 sin θ 2 sinθ ∂ ∂θ 2 + ∂2 ∂φ2 .e.63. 5.

φ) . φ. φ) (5. 0 ≤ φ < 2π. φ). The function space endowed with this norm and including only functions for which the integral (5.4: Angular Momentum Eigenstates 111 θ. 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.108) m (θ. The norm in H is deﬁned through |f |2 = S2 dΩ f ∗ (θ. . 5.105) explicitly. φ) L3 Y (θ. φ) = −i Y (θ.110) = 0 (5. − . One obtains in this way Y m (θ. m) Constructing Y → Y →→ ··· Y ( + m)! (2 )! ( − m)! −1 − We like to characterize the resulting eigenfunctions Y m (θ.102) where dΩ = sinθ dθ dφ is the volume element of S2 . .104) (5. is indeed a Hilbert space. normalizing these functions. φ) = −i ( + m + 1)( − m)Y m (θ.64) can be constructed then by seeking ﬁrst functions Y (θ.109) using (5. .106) The latter property is obviously satisﬁed for the chosen φ-dependence.107) reads ∂ ∂ + i cotθ ∂θ ∂φ P (cosθ) ei φ i e±iφ ∂ ∂ ± i cotθ ∂θ ∂φ .103) iteratively for m = − 1. m = − . φ) = ∆( . the so-called spherical harmonics. . m) = 1 −m J− 1 2 Y (θ. (5. (5. φ)f (θ. (5.5. − 2. can be expressed L± = Accordingly.63. (5. and to be admissable for description of quantum states.102) exists.105) ∆( . − + 1. φ) = 0 as well as L3 Y (θ. .104. . φ) in H which satisfy L+ Y (θ. φ) (5. φ). and then determining the family {Y m (θ. φ) (5. . φ) = = 0 −i Y (θ. must be cyclic in φ with period 2π. φ) = 2 + 1 ( − m)! P 4π ( + m)! m (cosθ) eimφ .85–5. .87). The raising and lowering operators L± = L1 ± i L2 . by carrying out the construction according to (5. } applying L− Y m+1 (θ.111) . The eigenfunctions (5.107) (5. 0 ≤ θ < π. 5. (5. φ).

106.115) holds. (5.116) The integral appearing in the last expression can be evaluated by repeated integration by parts. using the new integration variable x = cos θ. ”Classical Electrodynamics. 5. one for which π dθ sinθ |P (cosθ)|2 (5.114) yield |N |2 2 +1 1 2π 4π (2 )! π dθ sin2 0 +1 θ = 1.119) provide the following expression for Y Y (θ.e.106.107). 5. 2nd Ed. One obtains.114. Jackson (John Wiley.112 or Theory of Angular Momentum and Spin ∂ − ∂θ cotθ P (cosθ) = 0 (5.” by J. We note that (5. The normalization factor N is chosen such that π (5. ..D. (5.118) 2 +1 1 1 sin θ ei 4π (2 )! 2 ! φ . i.113) 0 is ﬁnite. 5. 1975) . New York. (5.116) implies |N |2 from which we conclude N = (−1) 1 · 3 · 5 · · · (2 − 1) (5.120) See.112) where we employed the deﬁnition of P (cosθ) in (5. is P (cos θ) = N sin θ as one can readily verify. π +1 dθ sin2 0 +1 θ = −1 +1 −1 dx (1 − x2 ) +1 = = .119) where the factor (−1) has been included to agree with convention1 .117) Accordingly.114) dθ sinθ |Y (cosθ)|2 = 1 0 (5. . A proper solution of this equation. = = x (1 − x2 ) 2 x3 3 + 2 −1 +1 −1 −1 dx x2 (1 − x2 ) + 2 · (2 − 2) 1·3 −1 +1 (1 − x2 ) dx x4 (1 − x2 ) −1 −2 +1 2 · (2 − 2) · · · 2 dx x2 1 · 3 · 5 · · · (2 − 1) −1 2 · (2 − 2) · · · 2 2 (2 )! 2 = 2 2 +1 1 · 3 · 5 · · · (2 − 1) 2 + 1 [1 · 3 · 5 · · · (2 − 1)] (5. for example. (5. φ) = (−1) 1 1 = 1 [1 · 3 · 5 · · · (2 − 1)]2 (5.

123) leads to the expression P m (x) = −m m ∂ 1 ( + m)! (1 − x2 )− 2 (x2 − 1) 2 ! ( − m)! ∂x −m (5. indeed. the eigenfunctions Y m (θ.123) P m+1 (x) ∂ x + (m + 1) 1 ∂x (1 − x2 ) 2 .e.121) states then for the functions P m (cos θ) ∂ + (m + 1) cotθ ∂θ P m+1 (cos θ) = − ( + m + 1) ( − m) P m (cos θ) .5.114. 5. − 2. P m+1 (x) = −m−1 m+1 ∂ 1 ( + m + 1)! (1 − x2 )− 2 (x2 − 1) .4: Angular Momentum Eigenstates 113 We have constructed a normalized solution of (5.63. use (5.124) reads P (x) = = (2 )! (1 − x2 )− 2 (x2 − 1) 2 ! (−1) 1 · 3 · 5 · · · (2 − 1) sin θ (5. through repeated application of (5.124) holds for m + 1. We want to demonstrate now that the recursion equation (5. − deﬁned in (5. therefore. m (cos θ) e (5. we seek to determine the functions P m (cos θ) and.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 . Let us then assume that (5.. 2 ! ( − m − 1)! ∂x −m−1 (5. and can obtain now. (5.124) called the associated Legendre polynomials.123). according to (5.124).64).125) which agrees with (5. are real. ∂x −m−1 .110). For m = (5. 5. To prove (5. φ). . φ). The reader should note that the associated Legendre polynomials.122) The latter identity can be written.124) we proceed by induction. i. m = − 1. 5. .119). P m (x) = −1 × ( + m + 1) ( − m) × −(1 − x2 ) 2 1 (5. (5.121) This expressions shows that the factor eimφ . 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. Actually. every application of L− reduces the power of eiφ by one. . φ). namely Y (θ.103.126) Then holds.110) together with (5.106).127) ∂ −m−1 (x2 − 1) . 5. describes the φ-dependence of Y m (θ.103. as speciﬁed in (5. employing again the variable x = cos θ.

2 ! (2 )! ∂x2 1 (5.124). therefore. if L− Y (θ. φ) L3 f (θ. the derivative ∂ 2 +1 /∂x2 +1 of this expression vanishes.127) reproduces (5. Such construction reproduces the eigenfunctions Y m (θ.132) which reduces the number of spherical harmonics which need to be evaluated independently roughly by half . Since (5. i.124) and. that expression (5.124) holds for m if it holds for m + 1. In that case (1 − x2 ) is a polynomial of degree 2 and. 5. However. φ) = (−1)m Y m −m (θ. Constructing Y − → Y − +1 → ···Y m (θ. We want to test if the recursion (5. by repeated application of the operator L+ .114 By means of −(1 − x2 ) 2 = Theory of Angular Momentum and Spin −m−1 m+1 ∂ ∂ (1 − x2 )− 2 (x2 − 1) ∂x ∂x −m−1 −m m ∂ −(1 − x2 )− 2 (x2 − 1) ∂x −m −m−1 m+1 ∂ x (1 − x2 )− 2 − (m + 1) 1 ∂x −m−1 (1 − x2 ) 2 1 (5. φ) choosing the proper sign. yields a vanishing expression for m = − − 1 as long as is a non-negative integer. φ) . which is equivalent to recursive application of L− . φ) = Y − (θ. and constructs then the eigenfunctions Y − +1 . φ) using (5. Y − +2 . expression (5. therefore. φ) = 0 (5. It holds.124) and.e. Y ∗ (θ.132).. therefore. (5. hence.133) Since the term with the highest power of (x2 − 1) is x2 . using x = cos θ.130) (5. φ) (5. We ﬁrst determine Y − (θ. we embark on this construction in order to prove (5. 5. φ) as given in (5. therefore. etc. P − (x) = 1 ∂2 (1 − x2 ) 2 (x2 − 1) . from such construction emerges an important symmetry property of Y m (θ. appears not very interesting. it holds ∂2 (x2 − 1) = (2 )! ∂x2 and.131) identiﬁes f (θ.123) terminates for m = − . (5.124) holds for m = . φ) = = determines ﬁrst a normalized solution 0 i f (θ.129) holds. namely.128) one can show that (5. In fact.135) .106. φ) An alternative route to construct the eigenfunctions Y of L− f (θ. P − (5. φ).106.124). it holds then for all m.134) (x) = 1 2 ! (1 − x2 ) 2 .

(5.106..80) and (5.132).136) We note in passing that this expression and the expression (5.122. (5.5.138) Employing (5. According to (5. We ﬁrst note that for m = − (5.4: Angular Momentum Eigenstates Due to (5. φ) sin θ e−i φ = 0.137) m (θ. φ) .140) Introducing again the variable x = cos θ leads to the recursion equation [c.120) for Y obey the postulated relationship (5. (5.e. 5.129).142) yields P − (x) = (−1) 1 (1 − x2 )− 2 (x2 − 1) = (1 − x2 ) 2 2 ! 2 ! (5. One can readily verify that this expression provides a solution of (5. (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. φ). following closely the proof of eq.136) to construct all other Y recursion eiφ ∂ ∂ + i cotθ ∂θ ∂φ Y m (θ.139) = ∂ − m cotθ ∂θ P m (cos θ) . (5.142) For this purpose we proceed by induction.123)] P m+1 (x) = −(1 − x2 ) 2 1 ∂ x − m 1 ∂x (1 − x2 ) 2 P m (x) .124) one arrives at Y (θ.124). a sign consistent with the family of functions Y m constructed above.143) . This follows from the identity ∂ ∂ − i cotθ ∂θ ∂φ One can use (5.110) applies the = ( + m + 1)( − m) Y m+1 (θ. (5. Obviously. 5.f. the expression (5.136) is normalized and has the proper sign. i.106).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. φ) = 2 +1 1 1 sin θ e−i 4π (2 )! 2 ! φ 115 − . (5.

φ) ( − m)! P ( + m)! −m (x) . one notes that application of (5.124) yields (−1)m ( − m)! P ( + m)! −m (x) = +m m ∂ (−1)m (1 − x2 ) 2 (x2 − 1) .s. Hence. (5.145) Hence. 2 ! ∂x 1 (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 .124) and (5. P (x) = P 0 (x) are called Legendre polynomials. i. Y −2 .146) i.124).142) and by (5. Since (5.128) m + 1 → −m or. 2 ! ∂x +m (5. etc. (5.148) According to (5. one can conclude P m (x) = (−1)m m (θ.s.e.144) reads P m+1 (x) = +m+1 m+1 ∂ (−1)m+1 (1 − x2 ) 2 (x2 − 1) . also the associated Legendre polynomials determined this way.150) . Accordingly.e.135).106) this implies for Y The Legendre Polynomials The functions the identity (5. etc.142) holds for m + 1 if it holds for m. of which agrees with the r. According to both (5.142).144) ∂ x − m 1 ∂x (1 − x2 ) 2 +m m ∂ (−1)m (1 − x2 ) 2 (x2 − 1) . Y − +2 . yields the same eigenfunctions as the previous construction beginning with Y and stepping down the series of functions Y −1 . 1 (1 − x ) ∂x +m (1 − x2 ) 2 1 (5..142) holds for m.116 Theory of Angular Momentum and Spin wich agrees with (5. given by (5. 2 ! ∂x +m+1 (5. 2 ! ∂x +m −(1 − x2 ) 2 Replacing in (5.149) (5.. are identical.h. (5. However. We then assume that (5.132).142) holds for m = − we veriﬁed that it holds for all m.141) reads then P m+1 (x) = 1 (5. The construction beginning with Y − and continuing with Y − +1 .142) one can state P (x) = ∂ (x2 − 1) .h. equivalently. of (5. (5.147) the r.

.156) yields P (1) = 1 .152) and.” by J. 1975). 2. For example. m ≥ 0 ∂xm (5.D. . 92 To prove this property see. 2nd Ed.J. . The spherical harmonics for m < 0 can be obtained using (5. accordingly. 2 3 1 1 = = 2 1−t 1 − 2t + t ∞ t . New York.151) (5x3 Comparision of (5. P3 (x) = 1 2 117 P2 (x) = − 3x) 1 2 2 (3x − 1) .154) can be written 1 = w(x. φ) = 2 + 1 ( − m)! (−1)m sinm θ × 4π ( + m)! × ∂ ∂ cos θ m (5.155) P (x) t .N.154) |r1 − r2 | r +1 =0 1 where γ is the angle between r1 and r2 .g. t) = √ 1 − 2xt + t2 ∞ 1 − 2 x t + t2 . 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. r1 is the point where the potential is measured. . e. and r2 denotes the location of the charge. and |r1 − r2 | = r1 (5. In case q = 1 and |r2 | < |r1 | holds the identity2 ∞ 1 r2 = P (cos γ) (5. t) is called a generating function of the Legendre polynomials3 . Jackson (John Wiley.4: Angular Momentum Eigenstates The ﬁrst few polynomials are P0 (x) = 1 . t) = √ Comparision with (5. ”Classical Electrodynamics. (5.153) P (cos θ) eimφ . (5.158) see. Lebedev (Prentice-Hall. Englewood Cliﬀs. Using x = cos γ. (5. The generating function allows one to derive useful properties of the Legendre polynomials.132). 1. pp. 45 of “Special Functions and their Applications” by N. 1965) which is an excellent compendium on special functions employed in physics. =0 (5. . pp.156) w(x. the spherical harmonics for m ≥ 0 Y m (θ. P1 (x) = x .157) = 0. for example. N. t = r2 /r1 .142) allows one to express the associated Legendre polynomials in terms of Legendre polynomials P m (x) = (−1)m (1 − x2 ) 2 m ∂m P (x) . in case of x = 1 holds w(1..5. =0 (5.150) and (5.

φ) goes over to Y m (π − θ. (5. Accordingly.164) which.166) We want to summarize the properties of the spherical harmonics Y derived above. φ) (5. π + φ). Inversion Symmetry of Y m (θ.. If the spherical coordinates of r are (r. . φ) m (θ.161) Collecting coeﬃcients with equal powers t yields 0 = P1 (x) − x P0 (x) ∞ (5. φ) = = 2 ( + 1) Y m (θ.167) (5. . π − θ. 1.163) (5.f.159) Using ∂lnw/∂t = (1/w) ∂w/∂t and multiplying (5.142) allows one to conclude P m (−x) = (−1) +m P m (x) (5. φ) . under inversion Y m (θ.118 For w(x.156) and (∂/∂t) to P (x)t = ∞ ∂w + (t − x) w = 0 . Accordingly. t) (1 − 2xt + t2 ) Employing (5. Due to cos(π − θ) = − cos θ the replacement r → −r alters P m (x) into P m (−x). 2. φ) m (π − θ. π + φ) = (−1) Y m (θ. ∂t 1 − 2xt + t2 (5. π + φ). holds P1 (x) ( + 1) P +1 (x) = = x P0 (x) (2 + 1) x P (x) − P −1 (x) (5. then the coordinates of − r are (r. Inspection of (5.168) mY m (θ. θ. φ) . The spherical harmonics are eigenfunctions of the angular momentum operators J2Y J3 Y m (θ.160) this diﬀerential equation is equivalent ∞ (1 − 2xt + t2 ) =0 P (x) t + (t − x) =0 P (x) t = 0 (5. t) holds Theory of Angular Momentum and Spin ∂lnw(x.162) − (2 + 1)x P (x) + P −1 (x) ] t + =1 [( + 1) P +1 (x) The coeﬃcients for all powers t must vanish individually. . (5. φ). t) x−t = .159) by w(1 − 2xt + t2 ) leads to the diﬀerential equation obeyed by w(x. Noting exp[im(π + φ)] = (−1)m exp(imφ) we determine Y Properties of Y m (θ. . using P0 (x) = 1 [c.151)] allows one to determine P (x) for = 1. ∂t P (x)t −1 −1 (5.165) since ∂ n /∂(−x)n = (−1)n ∂ n /∂xn . φ) Under inversion at the origin vectors r are replaced by − r. φ) m (θ.

) P0 (x) = 1 . φ) = (−1)m Y ∗ (θ. φ) . φ) m (θ. φ) where J± = J1 ± iJ2 . .175) (5.e. where ( = 1. a complete basis of C∞ (S2 ). φ) ∈ C∞ (S2 ) holds ∞ f (θ. The spherical harmonics form an orthonormal basis of the space C∞ (S2 ) of normalizable. The spherical harmonics are given by the formula Y m (θ.170) (5. The spherical harmonics Y 0 are Y 0 (θ. P −1 (x) ] (5. 4π (5.169) 3. uniquely deﬁned functions over the unit sphere S2 which are inﬁnitely often diﬀerentiable π 2π sin θdθ 0 0 dφY ∗m (θ. . except for the factor exp(imφ). 2.174) m ≥ 0 P (cos θ) eimφ . .176) = 1 +1 [ (2 + 1) x P (x) − are the Legendre polynomials. φ) . φ) (5.4: Angular Momentum Eigenstates 119 2. 6. (5. φ) 2π 0 (5. i.171) m = dφ Y ∗ (θ. The spherical harmonics for m < 0 are given by Y −m (θ.177) Note that the spherical harmonics are real.173) m+1 (θ. φ) C = =0 m=− π C sin θdθ 0 m Y m (θ. φ)f (θ. 5.5. φ) = δ δm m . m 4.178) 7. The spherical harmonics obey the recursion relationships J+ Y J− Y m (θ. The spherical harmonics form. φ) = 2 + 1 ( − m)! (−1)m sinm θ × 4π ( + m)! 2 ! × ∂ ∂ cos θ m (5. For the Legendre polynomials holds the orthogonality property +1 dx P (x) P ( (x) = −1 2 δ 2 +1 .172) (5. m (5.. in fact. P +1 (x) P1 (x) = x . for any f (θ. φ) Y m (θ. (5. φ) = 2 +1 P (cos θ) .179) . φ) = = ( + m + 1)( − m) Y ( + m + 1)( − m) Y m+1 (θ.

86).189). t) stated in (5.4. The spherical harmonics obey the inversion symmetry Y 10.189) evaluate the integral on the l. (5. (5.s. (5. The spherical harmonics for m (θ.s. Exercise 5. For this purpose start from the identity +1 ∞ +1 dx w(x.186) (5.188) Exercise 5. φ) = δm0 2 +1 .183) (5. 1.1: Derive expressions (5. φ) = 1 ∂2 r − r ∂r2 ( + 1) r2 h(r) Y m (θ. For the Laplacian holds 2 = = = (5.. φ) = (−1) Y m (π − θ. 4π (5.187) h(r) Y m (θ.2: Derive the orthogonality property for the Legendre polynomials (5.87). expand the result in powers of t and equate the resulting powers to those arising on the r. φ) . of (5.179) using the generating function w(x.h.177). t)2 = −1 . π + φ) . =0 t + −1 dxP (x) P (x) . The spherical harmonics for θ = 0 are Y m (θ Theory of Angular Momentum and Spin = 0.120 8.3: Construct all spherical harmonics Y3m (θ. . 11. (5. 2 are given by Y00 = 1 √ 4π 3 sin θ eiφ 8π 3 cos θ 4π 1 4 − 15 sin2 θ e2iφ 2π 15 sin θ cos θ eiφ 8π 5 4π 3 1 cos2 θ − 2 2 (5.185) (5.4.4.180) 9. Exercise 5.181) = 0.h.182) Y11 Y10 = = − (5.156). φ).184) Y22 Y21 Y20 together with (5.

. We like to argue that the matrix representing the rotational transformations of the function space C∞ (S ).192) = S2 dΩ Y ∗ (θ .191) i − ϑ·J f (θ . = ... ˜ f (θ.. φ) (5. φ) ∈ C∞ (S ). φ) ..− + . called the irreducible representation.e... . (5. φ) exp − ϑ · J . . φ ) (5. }] . and (ii) they form the lowest dimensional sets X obeying C∞ (S ) = X which have this invariance property. For this purpose we consider the subspaces of C∞ (S ) X = [ {Y . on functions f (θ. φ) m. m . i..42). .194) Y m (θ. . . φ) (5. (5. φ) = exp Since the spherical harmonics Y space C∞ (S ) one can expand ˜ f (θ.m D(ϑ) m. orthonormal basis for the function c m Y m (θ.. Comparision with (5. We denote the image of a function f (θ. ∞. φ).5 Irreducible Representations We will consider now the eﬀect of the rotational transformations introduced in (5. . = − .5. . φ) i − ϑ·J f (θ. therefore..5: Irreducible Representations 121 5. φ) ˜ Y m (θ. . φ) ˜ under such rotational transformation by f (θ.− + . . φ) c = . ∞.197) The subspaces X have the important property that (i) they are invariant under rotations exp − i ϑ · J .m m m (θ. exp − i ϑ · J (X ) = X . exp − i ϑ · J . m Y m (θ. .195) i. }] (5.190) provide a complete. assumes a particularly simple form. 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 . . (5.193) shows ∞ = − .e. m = − ... .195).193) where [ { } ] denotes closure of a set by taking all possible linear combinations of the elements of the set. .. φ) D(ϑ) = . The transformations exp − i ϑ · J . }.e. . = . (5. = 0. m = S2 i dΩ Y ∗m (θ. i. − + 1. are characterized completely if one speciﬁes the transformation of any Y m (θ. . φ ) exp m One can also represent C∞ (S ) C∞ (S ) = [ {Y . 1. with elements given by (5.196) C∞ (S ) = = X . if one speciﬁes the functional form of the coeﬃcients [D(ϑ)] m .

65). equivalently.197) of C∞ (S ) the matrix representation of D(ϑ) assumes a block-diagonal form. one concludes for the matrix elements (5. .194) only spherical harmonics with or. i.. leaves X invariant. J− . (5.e.202) D(ϑ) = (2 + 1) × (2 + 1) .. one can conclude that J+ leaves X invariant. 1×1 3×3 . n The action of J− on Y m produces Y m−1 or.. can be expanded in powers One employs then the commutation properties [J3 . The exponential operator exp − i ϑ · J + ϑ3 J3 )n .201) From expression (5.200) of the rotational transformations one can also conclude that the X are the smallest invariant subspaces of C∞ (S ).56.. produces 0. 5. . J− ] = 2 J3 . one needs to consider the action of J+ 1 J− 2 J3 3 on the subspaces X .n3 n n n Accordingly. the transformation exp − i ϑ · J leaves X invariant as well.e.122 Theory of Angular Momentum and Spin The invariance of the subspaces X under rotations follows from a Taylor expansion of the exponential operator exp − i ϑ · J .200) leaves X invariant. which follow readily from (5. . (5. (5.n3 J+ 1 J− 2 J3 3 . φ) .n2 . Since any additive term n n n cn1 . If one orders the basis according to the partitoning (5. φ) exp m i − ϑ·J Y m (θ.195) [D(ϑ)] m.198) 2 2 where ϑ± = ϑ1 ( 1 ϑ− J+ 2 + 1 2 ϑ+ J− ϑ2 .. m = δ S2 dΩ Y ∗ (θ. and J3 1 1 ϑ · J = ϑ− J+ + ϑ+ J− + ϑ3 J3 (5. 3. .199) [J+ . .n2 . Similarly.199) allows one to express exp i − ϑ·J = n1 .. Since the Y m are n3 eigenfunctions of J3 the action of J3 reproduces the basis functions of X . i. One proceeds by expanding ﬁrst ϑ · J in terms of J+ . in case m = − . with blocks of dimension 1.200) in (5. J± ] = ± J± (5. Application of (5. .e. i. 5. One = contribute can conclude then that in the expansion (5. J− 2 also leaves n1 X invariant.

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

φ) = S2 dΩ Y ∗m (θ. m i Y m (θ.203) For any ϑ ∈ R one can ﬁnd Euler angles α.h.203).195) by ˜ Y m (θ. For this purpose we notice that J2 can be expressed through the similarity transformation J2 = e− i αJ3 J2 e i αJ3 (5.m m. the transformation from the rotated frame to the original frame. (5.h.211) Using (5.212) . Accordingly. and the r..e. we can write e− i −1 = S eA S −1 .207) which replaces J2 by the inverse transformation (i).s.205) Y m (θ. (i) and (ii). followed by J2 in the original frame. in terms of rotations deﬁned with respect to the original frame. (5.206) The expression (5. The axis x2 is deﬁned in the coordinate frame which is related to the original frame by rotation (i).203) can be expressed. the transformation from the original frame to the rotated frame. of (5.207) are equivalent.204) is satisﬁed. is e− i γJ3 = e− i βJ2 e− i αJ3 e− i γJ3 e i αJ3 e i βJ2 (5. β. For any similarity transformation involving operators A and S holds eS A S Accordingly. The Euler rotation replaces exp − i ϑ · J by e− i γJ3 − i βJ2 − i αJ3 e e .e. We will demonstrate that (5.e. the l.124 Theory of Angular Momentum and Spin The angles α.203) has the disadvantage that it employs rotations deﬁned with respect to three diﬀerent frames of reference. however. γ ∈ R such that exp i − ϑ·J = e− i γJ3 − i βJ2 − i αJ3 e e (5.209) in this expression one obtains e− i γJ3 = e− i αJ3 e− i βJ2 γJ3 e i αJ3 e− i i αJ3 e− i i γJ3 e i i αJ3 γJ3 e− e i i αJ3 e i i βJ2 e i αJ3 = e− i = e− αJ3 e− βJ2 e− βJ2 e αJ3 (5. i.s. φ) [D(α.203).210) The third rotation in (5.209). γ will be referred to as Euler angles. one can replace (5.194. β. β. can then be written e− i βJ2 − i αJ3 e = e− i αJ3 e− i βJ2 (5. in analogy to (5. i (5. φ) e− γJ3 − i βJ2 − i αJ3 e e . β. γ)] m..208) αJ3 βJ2 = e− αJ3 e− i βJ2 e i (5. Obviously. i. γ)] = .209) The ﬁrst two rotations in (5. φ) (5. the axis x3 is deﬁned in the coordinate frame which is related to the original frame by the consecutive rotations (i) and (ii). m [D(α. i.. 5. followed by transformation (i).

m (5. (5. β. 5. φ) . (5. The eigenvalue property (5. β. φ) e− m i βJ2 Y m (θ. m i Y m (θ. m ( ) S2 dΩ Y ∗ (θ. (5.210) yields the simple result e− i γJ3 − i βJ2 − i αJ3 e e = e− i αJ3 e− i βJ2 e− i γJ3 . φ) e−iγm δ (5.203) with respect to the original (unprimed) frame one simply needs to reverse the order of the rotations.7: Spin 1 2 and the group SU(2) 125 Multiplication from the left with (5. β. one can write [D(α. φ) e− m i βJ2 Y m (θ. 5. Examples are the other ﬁve members of the lepton family to which the electron belongs.e.218) Deﬁning the so-called Wigner rotation matrix dmm (β) = one can express the rotation matrices [D(α.220) We will derive below [see Eqs. (5. β. the self-adjointness of the operator J3 one can state dΩ Y ∗ (θ. m [D(α.214) e− γJ3 = S2 dΩ Y ∗m (θ. φ) e− m S2 i αJ3 f (θ. γ)] = . Best known is the spin of the electron. to redeﬁne the three rotations in (5.m m. in addition. φ) Y m (θ. φ) f (θ. m = e−iαm S2 dΩ Y ∗ (θ.215) The evaluation of the matrix elements (5.213) i.. the muon µ and its neutrino νµ of the second . γ)] m. the electron e and the electron neutrino νe of the ﬁrst generation. but many other elementary components of matter are endowed with spin-like properties. This allows one to express the rotational trasnformation (5. 5. (5.309. φ) i i (5. φ) e−imγ . γ)] m.219) = e−iαm dm m (β) e−iγm δ ( ) . φ) e− S2 i γJ3 Y m (θ.310)] an explicit expression for the Wigner rotation matrix (5.5.215) beneﬁts from the choice of Euler angles for the parametrization of rotational transformations. φ) = e−im α S2 dΩ Y ∗ (θ.205. φ) e− αJ3 e− βJ2 Y m (θ.206) by ˜ Y m (θ.216) Using. γ)] m. φ) [D(α.7 Spin 1 2 and the group SU(2) The spin describes a basic and fascinating property of matter.64) yields dΩ f (θ.219). φ) . φ) (5. φ) = S2 dΩ f (θ.217) Accordingly.

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

z.5. 1 i. y. The proof of this relation rests on the commutation relationship and the anti-commutation relationship (5.229) = 1 .e.e. they are associated with the important fermion property of matter. x. namely. σk ]+ = 2δjk 1 . σ3 = 1 0 0 −1 . it might hold aj bk − bk aj = 0. 1 σj σk = − σk σj for j = k which can also be readily veriﬁed. σ2 = 0 −i i 0 .230) where a. b are vectors commuting with σ. (5. σk ]− = 2i jk z x − iy x + iy −z . 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. 2 S2 = 1 σ2 . k = 1. 3 are the Pauli matrices σ1 = 0 1 1 0 .k=1 σ σ 3 j k j. In fact.224) provide a basis in terms of which all traceless.224) Algebraic Properties of the Pauli Matrices The Pauli matrices (5.227) which is essentially identical to the Lie algebra (5.g. A = x σ 1 + y σ 2 + z σ3 . it holds.228) (5. (5.7: Spin 1 2 and the group SU(2) 127 exist three such linear independent matrices.31) of the group SO(3). hermitian 2 × 2–matrices A can be expandend. According to this property the Pauli matrices generate a 3dimensional Cliﬀord algebra C3 . ∈ R .225) σ (5. z A = In fact. σj 2 (5.k=1 σ σ aj bk + j>k aj bk = 3 j. e. At this point we will state a useful property.231) . 2 S3 = 1 σ3 2 (5.. (5. 2. but their comonents must not necessarily commute with each other.223) where σk . Cliﬀord algebras play an important role in the mathematical structure of physics. The Pauli matrices obey the following anti-commutation properties σj σk + σk σj = [σj . y. One can readily verify the commutation property σj σk − σk σj = [σj . (5. the simplest choice being S1 = 1 σ1 . (σ · a) (σ · b) = a · b 1 + iσ · (a × b) 1 (5. one obtains (σ · a) (σ · b) = 3 j 2 j=1 (σ ) aj bj 3 j k j.227. 5. i.226) The Pauli matrices satisfy very special commutation and anti-commutation relationships. Any such A can be expressed in terms of three real parameters x.k=1 j<k + σ j σ k aj bk .229) and avoids commuting the components aj and bk .

k=1 j>k σ j σ k (aj bk − ak bj ) = 1 2 3 j.222). 1 (5.k=1 j>k i σl (aj bk − ak bj ) = i =1 σ a×b .8 Generators and Rotation Matrices of SU(2) Sofar there is nothing which relates the transformations U . we like to draw in this respect also on a close analogy to angular momentum states which developes 2 2 2 if one considers the operators S 2 = S1 + S2 + S3 and S3 .235) which is identical to that of the generators of SO(3).k=1 j>k 1 2 3 j. S ] = i k m Sm (5. (5.s. i. through the algebra obeyed by the operators Sk [Sk . 1 1 1 ( + 1) χ± 2 2 . (5. (5. however.222) by the Euler form exp(−iγS3 ) exp(−iβS2 ) exp(−iαS3 ) . S3 χ± = ± 1 χ± 2 .232) − σ k σ j ) (aj bk − ak bj ) .237) + 1)1 and.222) as a rotation vector and you rotate once.233) This result together with (5. In the special case a = b ∈ R holds (σ · a)2 = a 2 1 . Well almost like it.236) The matrix elements of this 2 × 2 operator can be labelled by the basis states χ+ and χ− .238) S 2 χ± = . This relationship emerges. only a 720o rotation leaves the spin invariant. yields a · b.227) leads to 3 3 jk j. For this purpose we choose to replace (5. The two remaining terms yield using σ j σ k = −σ k σ j for j = k and altering ‘dummy’ summation indices 3 j.e. 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. The commutation property (5. Noticing the idempotence of the Sk ’s 2 Sk = 3 one obtains S 2 = 4 1 = 1 1 1 2(2 1 4 1 0 0 1 (5.128 Theory of Angular Momentum and Spin Using (σ j )2 = 1 the ﬁrst term on the r.h.k=1 j>k σ j σ k (aj bk − ak bj ) − 1 2 3 j k j.234) 5.k=1 (σ σ j>k σ k σ j (aj bk − ak bj ) = (5.231) proves (5. We like to derive this result now by evaluating the transformations of SU(2) explicitly. however. let say around the x2 –axis by 360o . to rotations in space. since there is a slight diﬀerence: when you interprete the 3-component ϑ in (5. hence.230). (5. spin changes sign.

The complete matrix elements (5.244) i.220) are [D(α. namely.219. 2. i. . . (5. (5. therefore. i.5. . . (5. One cannot consider the 1 entities carrying the spin– 2 to be particles in the ordinary sense for it can be shown that spin– 1 2 particles have fermion character. no two such particles can exist in the same state. .243) We note the property of this rotation matrix 2 dmm (2π) (1) = −1 1 . 1 . that states | m for higher quantum numbers = 0.e.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. (5.240) . ( − β S2 )2n = (−1)n ( − β S2 )2n+1 = (−1)n β 2 β 2 2n 1 1 0 1 −1 0 2n+1 (5.241) 2n+1 Taylor expansion of the exponential operator yields then ∞ exp(−iβS2 ) = n=0 (−1)n β 2 2n ∞ 1 + 1 n=0 (−1)n β 2 0 1 −1 0 (5. The idempotence property of Sk yields particularly simple expressions for these powers. . 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.242) The expressions in brackets [.219). m = − . β. − + 1. For this purpose one needs to determine the powers of −iβS2 .239) in the notation (5. 1. We may. In order to determine the Wigner matrix in (5. ..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 . γ)] m.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).245) 5. 3 . 2 2 We obtain with this notation for the transformations (5. rotation by 360o changes the sign of the spin state.e.239 ) one expands the exponential operator exp(−iβS2 ).] can be identiﬁed with the Taylor expansion of the cos. . 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 . identical particles any number of which can exist in the same state χ+ and χ− .e.5.239) where we made use of the notation (5. following Jourdan and Schwinger. . use the label 2 2 | 1 ± 1 for the states χ± .9 Constructing Spin States with Larger Quantum Numbers Through Spinor Operators In this section we like to demonstrate.

246) The corresponding adjoint operators are denoted by b+ and b− . −) bζ .e. b ζ = b ζ. i. b ζ = b† . b † ζ ζ =0 (5. 3 . .250) which for x∗ x = x∗ x+ + x∗ x− = 1 represent spinor operators in an arbitrary reference system. therefore.e.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 − .248) bζ . (5. b† )T and b = (b+ . b− )T as spinor creation and annihilation + − operators.247) (5. j. .253) = δζζ . also operators of the type x b† = x+ b† + x− b† + − . The boson character of the operators b† = (b† .251) One can show using this property and (5. m) = b† − j−m (j + m)! (j − m)! |Ψ0 . 1 . b− )T is expressed through the commutation relationships + − (ζ. b ζ = 0 b ζ. . ζ = +. 2 + 2 − † (5. j = 0. .252) (5. using (5. 2 2 . b† |Ψ0 = χ+ + . We consider. x∗ b = x∗ b+ + x∗ b− + − (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 ζ The States |Ψ(j. b− † = − sin β † β b + cos b† . 1. b† |Ψ0 = χ− − . .248) that bζ relationships † † b ζ.249) In the following will refer to b† = (b† . These operators are associated with a given spatial reference system.5. m = −j. . i.247. b † ζ For the vacuum state |Ψ0 holds = δζζ . b± |Ψ0 = 0 (5. −j + 1.254) We will show below that these states are orthonormal and form spin states with higher quantum numbers. + − For example. (5. b† )T and b = (b+ . m) † and bζ obey the commutation (5.

5. b† + n b† + b† + b† + + n n n b+ . (5. b † + = n b† + n−1 b† + b † + + n = (n + 1) b† + . If it is true for n then using (5. The Spinor Derivative Operator A most important operation is the action of the annihilation operators b+ and b− on operators which can be expressed as polynomials. f (b† ) |state + f (b† ) b+ |state .10: Algebraic Properties of Spinor Operators 131 5.247. in b† and b† . b† + n .5.256) In the special case |state = |Ψ0 this reads using (5. b† + = n b† + n−1 . For this purpose we consider the following polynomial of creation operators N j+m j−m f (b† ) = + n=0 cn b† + n .248) we obtain b+ . f (b† ) |Ψ0 . (5. .260) . possibly inﬁnite power series. we need to evaluate 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+ .249) b+ f (b† ) |Ψ0 = + b+ . + b+ . f b† Obviously.257) motivate us to determine the commutator (5.259) The property is obviously true for n = 0.258) Equations (5. We will show that for this commutator holds n b+ .249).256) and (5. that the annihilation operators play a role similar to the diﬀerential operator in calculus. We will derive the result. b† + n+1 = = b+ . b† + b+ .257) which for (5. + + (5.255) is b+ .10 Algebraic Properties of Spinor Operators We want to establish ﬁrst a few important and useful algebraic properties which result from the commutation relationships (5. f (b† ) + (5.248) and from (5. b† + n = n cn b+ .254). (5. An example + − is the monomial b† which generates the states (5. b† + − well-known for the quantum mechanical harmonic oscillator.

(5.. f (b† ) − = f (b† ) − (5. b† ) |Ψ0 = + − ∂ ∂b† ζ f (b† .259) holds for all n ∈ N. f (b† . By induction we can conclude that (5. In order to derive (5. the property holds also for n + 1.261) Similarly.255) b+ . b† ) in b† and b† holds (ζ = +. one can prove b− .1.264) This demonstrates the equivalence of the spinor operators bζ and the derivative operation. m) φjm (x) = j−m xj+m x− + (j + m)! (j − m)! . m) m=−j (5.268) . (5.266) we compare the terms x+ b† + j+m φjm (x) Ψ (j. (5. b† ) + − = ∂ ∂b† ζ f (b† . b† ) |Ψ0 . One can ﬁnally conclude for polynomials (5. m) = x− b† − j−m (j + m)! (j − m)! |Ψ0 .267) exp xb† |Ψ0 is called a generating function of |Ψ(j.247) the more general property for polynomials f (b† .262) Because of the commutation relationships (5.266) where xb† has been deﬁned in (5.257) we can conclude in particular bζ f (b† . 2 . φjm (x) Ψ (j.132 Theory of Angular Momentum and Spin i. = f (b† ) + (5. m) . f (x)] = ∂k f (x) ..250) and where φjm (x) represents the function of the two variables x+ and x− closely related to |Ψ(j. + − (5.263) 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 . b† ) . + − (5. Equation (5. f (b† ) + where f (x) = df dx .e.. f (x)] g(x) = (∂k f (x))g(x) or [∂k . −) + − + − bζ .265) Generating Function of the States |Ψ(j.263) According to (5. m) We want to prove now the property ∞ j exp xb † |Ψ0 = 1 j=0.

1 xb† u! |Ψ0 (5. 2 2 leads to ∞ j ∞ φjm (x) |Ψ (j. i. .1. one obtains φjm (x) Ψ (j. 1..e.275) = exp xb† |Ψ0 ..269) (5.. . .. 2 . .274) Summing this expression over 2j = 0..272) The summation over s can be written in terms of j and m using (5.1.. (5. (5. . s! (n − s)! Deﬁning j+m = N −s j−m = s. 1 .. m) 1 j=0. 3 ..270) s x+ b† + N −s x− b† − s |Ψ0 . chosing the summation index j = 0. m) m=−j = = 1 x+ b† + x− b† + − (2j)! 2j 1 xb† |Ψ0 . 1 .273) Change of the summation indices allows one to conclude j φjm (x) Ψ (j...10: Algebraic Properties of Spinor Operators with the s–th term in the binomial expansion of (a + b)n n! an−s bs . 1.271) 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. 2. m=−j = j=0. 2..270) N 2j j → s=0 j−m=0 → m=−j . . (2j)! 2j |Ψ0 (5.1. 2 ∞ 1 xb† 2j! u 2j |Ψ0 = u=0.5. m) = = |Ψ0 (N − s)! s! N −s 1 N! x+ b† x− b† + − N ! (N − s)!s! N −s 133 (5.

280) The commutation properties [b+ . one can write † † e(xb ) Ψ0 |e(yb ) Ψ0 † ∗ Ψ |e(x y) e(yb ) Ψ = 0 0 = e (x∗ y) † Ψ0 |e(yb ) Ψ0 (5. (5. that Ψ (j.264) implies † † bζ e(yb ) |Ψ0 = yζ e(yb ) |Ψ0 .j .277) by Ψ0 |e(x b) e(yb ) Ψ0 . According to (5. (5. hence.m.279) One can generalize this to † † s (bζ )s e(yb ) |Ψ0 = yζ e(yb ) |Ψ0 . For this purpose we consider the inner product † † e(xb ) Ψ0 |e(yb ) Ψ0 = δjj δmm (5. Orthonormality of the States |Ψ (j. of (5. m) We want to show now that the states (5. m) |Ψ j . m) |Ψ j .h. b− ) holds † † f (b+ .e.281) and. (5.249) for any non-vanishing integer s holds + − Ψ0 |bs Ψ0 = 0 and.m φjm (x∗ ) φj m (y) Ψ (j. In order to evaluate the ∗ b) (yb† ) operator e(x e we notice that the derivative property (5. b† ] = 0 allow one to state that for + − + any polynomial f (b+ . y− )e(yb ) |Ψ0 (5. i. [b† . b− ] = 0.282) where we deﬁned x∗ y = x∗ y+ + x∗ y− . b− )e(yb ) |Ψ0 = f (y+ . therefore. (5. m) and will be used for this purpose below.277) To evaluate this expression we ﬁrst notice † e(xb ) † = e(x ∗ b) .283) (5. we can conclude † † ∗ e(xb ) Ψ0 |e(yb ) Ψ0 = e(x y) (5. b† ] = 0 and [b− .134 Theory of Angular Momentum and Spin which concludes our derivation.s. m . m holds.278) † ∗ which allows us to replace the l.276) = j. The generating function allows one to derive various properties of the states |Ψ (j.284) .254) are orthonormal.

(5.275) one can show e(x ∗ y) = j. The 2 operators are explicitly. (5. 1 New Representation of Spin 0.5.224). using < + |σ1 | + > = < − |σ1 | − > = 0.m.m 135 = e(x ∗ y) .285) Following steps similar to those in Eqs.m φjm (x∗ ) φjm (y) . m) as given in (5.276).277) φjm (x∗ ) φj m (y) Ψ (j. bm bm } n n = 1 4 < n |σi | m >< n |σj | m > {b† bm δmn − b† bm δm n } n n n. − 2 + (5. (5.m. b† bm n n n.288) The three operators obey the Lie algebra of SU(2) [Ji .n . 5. 1. etc. (5.m. . Jj ] = = 1 4 1 4 < n |σi | m >< n |σj | m > b† bm .n . 2. < + |σ1 | − > = < − |σ1 | + > = 1.m +b† b† . b† bm n n n. k = 1.m .287) ζ 2 ζ.10: Algebraic Properties of Spinor Operators and comparing (5.223. m j.282) and (5. In our present notation the matrix elements < ζ |σk | ζ > for ζ = ± correspond to the matrix elements < ζ |σk | ζ > for σ = ± 1 . (5. One deﬁnes corresponding operators Jk through 1 Jk = b† < ζ |σk | ζ > bζ . 2 .268–5. .n . States 2 We want to demonstrate now that the states |Ψ (j..m.286) from which follows immediately the orthonormality property (5. 3 denote the Pauli spin matrices (5. J1 = J2 = J3 = 1 † b b− + b† b+ − 2 + 1 † b b− − b† b+ − 2i + 1 † b b+ − b† b− . m) |Ψ j . .ζ where σk .j . 3 . Jj ] = i This property is derived as follows [Ji .m ijk Jk .289) < n |σi | m >< n |σj | m > {b† bm .224).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.

J2 ] = −2 J3 from which we can conclude the property 1 1 J− J+ = J+ J− − J3 2 2 Hence.292).m < n |2i n.m.k ijk ijk σk | m > b † bm n = i = i 1 2 < n |σk | m > b† bm n n.292) The commutation relationships (5. J1 + iJ2 ] = −i [J2 .295) The last result together with (5. (5.m. 2 J 2 = J+ J− + J3 − J3 = J3 (J3 − 1) + J+ J− (5.m ijk Jk .293) .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. (5.291) Here we have deﬁned J+ J− = J1 + iJ2 = J1 − iJ2 = b† b− + = b† b+ . m) are eigenstates of J 2 and J3 . J+ ] = [J1 − iJ2 .294) .290) |Ψ (j.289) yield [J− . − (5.297) . (5. (5. m) as Eigenstates of J 2 and J3 We wish to show now that the states |Ψ (j.m < n |[σi .m.m 1 4 < n |σj | m >< m |σi | m > b† bm n n . To this end we note 2 2 2 J 2 = J1 + J2 + J3 = 1 1 2 J+ J− + J− J+ + J3 2 2 . (5.296) ( b † + b + b † + b+ + b † + b + b † − b − − 2 b † + b + − b † − b − b † + b + + b † − b − b † − b − + 2 b † − b − + 4 b † + b − b† − b + ) (5.136 = 1 4 − = = 1 4 1 4 Theory of Angular Momentum and Spin < n |σi | m >< m |σj | m > b† bm n n. σj ]| m > b† bm n n. J1 ] + i [J1 .

.302) (5.s. m) J3 |Ψ (j.304) (5. b † + b + + b † + b+ b † − b − 1 1 b† + b + + b † + b + b † − b− + b † − b − b † + b + . 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† . 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. m) One can furthermore derive readily J+ |Ψ (j.300) (5. . i. n} eigenstates of I 2 = I1 + I2 + I3 and of I3 I 2 ||λ. the states (5.299) (5.301) Obviously. . i. [Ij . . b† . .254) are eigenstates of + b+ and − b− with eigenvalues j + m and j − m. State the corresponding eigenvalues and the degree of degeneracy.305) = = j ( j + 1 ) |Ψ (j. m) = = (j + m + 1) (j − m) |Ψ (j. . m + 1) (j + m) (j − m + 1)|Ψ (j.298) (5. m) .303) Exercise 5. 1 2.e. .h. m − 1) . (b) Show that the three operators 1 + 1 1 + (b1 b2 + b+ b1 ) . . (5. b+ ] = δjk . n1 = 0. the number of states to the possible energy eigenvalues. . I2 = (b1 b2 − b+ b1 ) . Show λ = n1 +n2 n1 +n2 2 ( 2 + 1) and m = 1 2 (n1 − n2 ). Construct. m I1 = where λ = 0. m = m||λ. m . j = 1. 1.. One can then conclude J 2 |Ψ (j. 2 2 137 (5. 2. 1 2 k (a) Show that the eigenstates are given by 1 |n1 . 2 . I3 = (b+ b1 + b+ b2 ) 2 2 2 2 2i 2 1 satisfy the Lie algebra of SU(2). . [bj . (5.e. ˆ respectively. and eigenstates of k with eigenvalues j. m) J− |Ψ (j. .5. λ. −λ + 1. n = n1 + n2 ﬁxed. using operators I± = I1 ± iI2 2 2 2 and the subspace {|n1 . . . Ik ] = i jk I . I3 ||λ.10. m = λ||λ.1: The system investigated in this section is formally identical to a 2-dimensional isotropic harmonic oscillator governed by the Hamiltonian H = ω(b+ b1 + b+ b2 + 1) . m = −λ. n2 .10: Algebraic Properties of Spinor Operators The last term on the r. m) m |Ψ (j.

On the other side the states |Ψ (j.306) where b † and b † are given by (5. In case j = 1 this expression yields.307) Comparision of (5. (j) (5. σ values which yields (b† )j+m (b† )j−m is + − then σ = j + m − σ.306. m = m=−j dm m (β) |Ψ (j. 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 (5.309) . m) .306) and (5. m = b† + j+m b† − j−m (j + m )! (j − m )! |Ψ0 (5.243).310) it reduces to (5. for example. The combination of σ.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 × (5.5. The prefactor of (b† )j+m (b† )j−m which according to (5. both conditions are satisﬁed for σ = j + m − σ.251). However. 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. . m) in the original coordinate system by j |Ψ j. + − One may expect that these two conditions restrict both σ and σ.254) are |Ψ j.138 Theory of Angular Momentum and Spin 5.308) The latter sum involves terms (b† )j+m (b† )j−m for σ + σ = j + m and 2j − σ − σ = j − m. For this purpose we note that the states |Ψ(jm) in a rotated coordinate system according m to (5.307) shows that the elements of the rotation matrix can be obtained by binomial expansion of b † and b † in terms of b† and b† .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).

x3 ) ∈ R . the matrix A should represent a rotation around the x2 –axis in the space of vectors ˜ (x1 . It is of interest to trace the mapping from SU(2) onto SO(3) as represented by (5.311) (5. Hence.310) establishes a mapping of SU(2) onto SO(3).313) × 0 0 2 = 0 sinβ 0 cosβ 1 −i 0 which.. in fact. 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 (5.314) leaves (5. that the the representation (5. i. = U x2 ..g.308) and had the form (5. i. This can be shown by (1) applying to the matrix A = (dm m ) in (5.310) 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 .5. e. b− † = sin β † β b+ − cos b† .12: Mapping of SU(2) onto SO(3) 139 5.312) The choice of this transformation derives from a property shown further below.308) unaltered except for a factor (−1)2j which multiplies then also the ﬁnal result (5.e.e. in case of j = 1. therefore.309) and the particular matrix (5. however.309).310).s.12 Mapping of SU(2) onto SO(3) The representation (5. The SU(2) transformation entered in (5. One can.310) to the SU(2) transformation assumed in deriving the general result (5.251) and (5. This factor implies.h.314) in case of integer j–values. namely that the component of the vector on the l. conclude that the mapping from SU(2) onto SO(3) is a 2–1 mapping. x2 . b+ † = −cos β † β b+ − sin b† 2 2 − .312) transforms like an angular momentum state ˜ |1 m . 2 2 − (5. is the expected element of SO(3). U = 0 1 1 i √ ( x1 − ix2 ) √ x3 − √2 0 2 2 (5.309) does not distinguish between SU(2) transformations (5. . of (5. Replacing the latter transformation by its negative form.251). the orthogonal 3 × 3 matrix which describes a rotation around the x2 –axis.

140 Theory of Angular Momentum and Spin .

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

. albeit puzzling. . mC . . the values of which are determined by the size of the interaction domain ∆V . .1) where 1. In fact. to separate the 14 641 dimensional space of rotational states Y 1 m1 (ˆAB ) Y 2 m2 (ˆC ) into r ρ subspaces Bk . a very large number.max and 2.max d(B) = 1 =0 (2 1 + 1) 2 =0 (2 2 + 1) = ( 1. that only states within the subspaces Bk are coupled in the scattering process. vibrations and rearrangement of the particles in reactions like AB + C → A + BC. n = 1.max = 1 2 IA−B E . − 1 ≤ m1 ≤ 1. and by the moment of inertia IA−B of the diatomic molecule A–B approximately as follows 1.. since further important degrees of freedom. have not even be considered.max + 1)2 (6. B1 ⊕ B2 ⊕ . each of which stands for two angles.m1 2 .m1 . = 0. − 2 ≤ m1 ≤ (6. The rotational symmetry of the interaction between the particles allows one. 14 641} .max = ∆ 2 mA mB mC E . 2} 1. 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.3) For rather moderate values 1. as we will demonstrate below.max = 10 one obtains d(B) = 14 641. by the total energy E. .m2 Y 1 m1 (ˆAB ) Y 2 m2 (ˆC ). mA mB + mB mC + mA mC 2. mB . 2 . .142 Addition of Angular Momenta and Spin ρC . will make their appearence a natural consequence of the symmetry of the building blocks of matter. (6. The latter often arrives at the physical properties of composite systems by adding the .max + 1)2 ( 2. i.max .max = 2.e.max denote the largest orbital and rotational angular momentum values.max . however. Such extremely useful transformation of the problem can be achieved through the choice of a new basis set B = { 1 .max 2. in particular. composite systems behaving like elementary objects are common. 1. Such large number of dynamically coupled states would constitute a serious problem in any detailed description of the scattering process.m2 c (n) r ρ 1. the dimensions d(Bk ) of these subspaces does not exceed 100. . rather than by Classical Mechanics. = B such. . (6. . 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. by the masses mA . in fact. ρ 1 = 0. 2. 1.2) The dimension d(B) of B is 1.4) The basis set which provides a maximum degree of decoupling between rotational states is of great principle interest since the new states behave in many respects like states with the attributes of a single angular momentum state: to an observer the three particle system prepared in such states my look like a two particle system governed by a single angular momentum state. . 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 ). . Obviously. . . and the following mathematical description will shed light on their ubiquitous appearence in physics.

8) Obviously. (6. 3 (6. Since this is true for any ψ(rAB .5) do not aﬀect the Hamiltonian.7) i R(ϑ) = exp − ϑ · J (6. To specify this property mathematically let us denote by H the Hamiltonian in the rotated frame. ∂yAB ∂zAB i (2) ∂ ∂ − J3 = ρy − ρx . the commutation relationships (1) Jp .e.. Hence. according to (5. rotations R(ϑ) of the wave functions ψ(rAB . For ˆ ˆ example. One can equivalently express therefore (6. Following our description of rotations of single particle wave functions we express (6. It holds then H R(ϑ) ψ = R(ϑ) H ψ. be diﬀerent from H. ∂ρx ∂ρy (6.5) according to (5.48) i i R(ϑ) = exp − ϑ · J (1) exp − ϑ · J (2) (6. the existence of a basis (6. Rotational Symmetry of the Hamiltonian As pointed out already. R−1 (ϑ) ρC ) (6.6.9) hold since the components of J (k) are diﬀerential operators with respect to diﬀerent variables. in fact.6) For the following it is essential to note that H is not invariant under rotations of only rAB or ρC . In this sense. ρC ) = ψ(R−1 (ϑ) rAB . Describing quantum mechanically a property of a composite object as a whole and relating this property to the properties of the elementary building blocks is then the quantum mechanical equivalent of the important operation of addittion. presumably a facility the reader would like to acquire with great eagerness.0: Addition of Angular Momenta and Spin 143 corresponding physical properties of the elementary components. The invariance of the Hamiltonian under overall rotations of the three particle system implies then H = R(ϑ) H R−1 (ϑ) .4) which decouples rotational states is connected with the rotational symmetry of the Hamiltonian of the three particle system considered. ρC ) it follows H R(ϑ) = R(ϑ) H.55) holds i (1) − J1 = zAB ∂ ∂ − yAB . i. ρC ) deﬁned through R(ϑ) ψ(rAB . 5.7) where the generators J (k) are diﬀerential operators acting on rAB (k = 1) and on ρC (k = 2). 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. connected with the fact that the Hamiltonian H does not depend on the overall orientation of the three interacting particles. 2.53. assuming presently that H might. Jq(2) = 0 for p. q = 1. but solely under simultaneous and identical rotations of rAB or ρC . the reader will learn in the following section how to add and subtract in the microscopic world of Quantum Physics. 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).10) .

f. To order O(|ϑ|) one obtains i ϑ · JH . We recognize. k = 1. of course. 2. k = 1. (6. [H. a supposition which can be tested through the commutation properties of these operators. C]. In fact. One can show readily that the commutation relationships [Jk .16) following the procedure stated in the theorem above [c. that the components Jk .6) for rotational invariance of the Hamiltonian in the form i i H = exp − ϑ · J H exp + ϑ · J .12) each individually can have simultaneous eigenstates with the Hamiltonian. 2. (6. (6.14) implies that the total angular momentum is conserved during the scattering process. Eqs.14) implies that the components of the total angular momentum operator (6. k = 1.12) We consider this equation for inﬁnitesimal rotations. the (n) (n) (n) properties [Jk .9). (5. 2. for |ϑ| H ≈ 1 − 1 i ϑ·J H 1 + 1 i ϑ·J ≈ H + i 1. Jk ] = 0 .. that energy. i.e. ˆ Ω 2 = ρC . B] + [A.81) we want to consider the relationship between YJM and Y 1 m1 (ˆAB ) Y 2 m2 (ˆC ). the operators Jk . (6. 3 as the three components of the total angular momentum operator. 2 together with the property [A.11) we can write the condition (6. componentwise. r ρ Deﬁnition of Total Angular Momentum States The commutation property (6.15) are satisﬁed.71–5. 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. 2.. J ] = i k m Jm (6. 3 cannot have simultaneous eigenstates among each other.81)]. 3 . the important fact that the Jk obey the Lie algebra of SO(3). k = 1. i. and the eigenvalues of J2 and J3 are good quantum numbers. For a proof one uses (6. J ] = i k m Jm for n = 1.144 where J = J (1) + J (2) Addition of Angular Momenta and Spin .11) By means of (6.e. 3 do not commute. ˆ (6. i.14) We will refer in the following to Jk . The property (6. B + C] = [A. we will ﬁnd that the states yield the basis B with the desired property of a maximal uncoupling of rotational states. however. Before we apply the procedure (5.71–5.13) Hϑ · J − Since this holds for any ϑ it follows HJ − J H = 0 or.17) . In the following we will use the notation r ρ Ω1 = rAB .e. We suspect. 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 ).

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

In this case the range of |J|–values would be the interval [ |J (1) | − |J (2) | .24) yields 1 m1 2 m2 ) ∗ 1 m1 2 m2 ) (J M| = δJJ δM M . it is not immediately obvious what the J– values are. .h. | 1 − 2| + 1. Ω2 ) YJ M ( 1. Ω2 ) = δJJ δM M (6. 1 + 2 .. (6. . 1. J (1) and J (2) . 1 + 2. The property dΩ1 dΩ2 Y∗ ( 1 .25) i. Ω2 ). . . and that the states can be normalized. . (6.28) This property follows from the fact that the basis elements are eigenstates of hermitian operators with diﬀerent eigenvalues. (6. |J (1) | + |J (2) |]. | 1 − 2 | + 1. .s.18) applied to Y∗ and to YJ JM (J M | m1 .h.1. 2) = {Y 1 m1 (Ω1 ) Y 2 2 m2 (Ω2 ). = {YJM ( 1 .146 Theory of Angular Momentum and Spin The expansion (6. − + 1.m2 2 M and with (6. . to a new basis B ( 1 . in fact. The orthonormality (Ω1 ) Y 2 m2 (Ω2 )Y 1 m1 Ω1 ) Y 2 m2 (Ω2 ) = δ 1 1 δm1 m1 δ 2 2 δm2 m2 . −J + 1. (6.. However. Our derivation below will also yield real values for the Clebsch-Gordan coeﬃcients.24) The basis B( 1 . the range of values assumed for J is correct. . it holds 1+ 2 ( 2J + 1 ) = (2 J=| 1 − 2 | 1 + 1) (2 2 + 1) . m1 = − 1 . respectively. 2 ). .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. property implies dΩ1 dΩ2 Y 1 m1 corresponding to the l. 2 ) has (2 1 + 1)(2 2 + 1) elements. The basis B ( 1 . . . . . B2 2) should be 2 |Ω1 . (6. J. −J + 1. . J} . Exercise 6. For each quantum number J there should be 2J + 1 elements YJM with M = −J.2: Prove Eq.s. . m2 = − 2 . . 2 ) and B ( 1 . the basis B ( 1 .27) together with (6. 2 ) is also orthonormal2 and must have the same number of elements.26) We will show below in an explicit construction of the Clebsch-Gordan coeﬃcients that. (6. . . − 2} 1 + 1. . In fact.23) corresponding to the r. M = −J. J = | 1 − 2 |. JM 2 |Ω1 . . 2) . 2 |Ω1 . This obviously corresponds quantum mechanically to a range of J–values J = | 1 − 2 |.18) constitutes a change of an orthonormal basis B( 1 .25) We want to state now two summation conditions which follow from the orthonormality of the two basis sets B( 1 .

81) stated above. 2 ) span the same function space.71–5. e.28) allows one to conclude that the coeﬃcients cJ (J M | 1 m1 2 m2 )∗ . 2 ) and B ( 1 .27) yields δJJ δM M = m1 . (6. At this point we like to stress that the construction will be based on the theorems (5.e. 2 |Ω1 .2 Construction of Clebsch-Gordan Coeﬃcients We will now construct the Clebsch-Gordan coeﬃcients. and actually also the properties of Clebsch-Gordan coeﬃcients stated above. (6. 2 |Ω1 .. Ω2 ) and integrating. 2 |Ω1 .6. i.30) The latter property follows from multiplying (6. or baryons multiplets involving three quarks..1 states. Ω2 ) . (6.33) JM The latter summation has not been restricted explicitly to allowed J–values. We can. .e. i.m2 M ( 1. also apply the results. 2 |Ω1 . A similar 2 construction will also be applied to composite systems governed by other symmetry groups. will be based solely on the commutation properties of the operators J and J (k) .34) 6.. Ω2 )Y 1 m1 (Ω1 )Y 2 m2 (Ω2 ) . or J > 1 + 2 (6. One can show readily using the same reasoning as applied in the derivation of (6. Ω2 ). i. The result of this construction will include all the properties previewed above. to composite systems involving spin. the group SU(3) in case of meson multiplets involving two quarks.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 .31) are identical to Comparision with (6. it is possible to expand Y 1 m1 (Ω1 )Y 2 m2 (Ω2 ) in terms of YJM ( 1 . Ω2 ) . 1+ 2 J Y 1 m1 (Ω1 )Y 2 m2 (Ω2 ) = J =| 1 − 2 | M =−J cJ M YJ M ( 1..32) which complements (6. 1 m1 2 m2 ) = 0 if J < | 1 − 2| .g. Hence. rather the convention (J M | has been assumed.18) by Y∗ J orthogonality property (6.18).29) where the expansion coeﬃcients are given by the respective scalar products in function space cJ M = dΩ1 dΩ2 Y∗ J M ( 1. therefore. 2 |Ω1 . 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 (J M | 1 m1 2 m2 ) cJ M .28) from (6. (6.e.2: Construction of Clebsch-Gordan Coeﬃcients 147 The second summation condition starts from the fact that the basis sets B( 1 . (6.

Ω2 ) (Ω1 ) dΩ2 Y 2 2 (Ω2 ) = 1. 1+ 2 In fact. (6. J3 YJJ ( 1 . we can also demonstrate using (??) that G dΩ1 = We. . Ω2 ) = Y 1 1 (Ω1 ) Y 2 2 (Ω2 ) . 2 |Ω1 . Ω2 ) is normalized dΩ2 G dΩ1 Y 1 1 1+ 2.36) = 0 = J YJJ ( 1 . 2 |Ω1 . Ω2 ) Addition of Angular Momenta and Spin (6. Ω2 ) = (J + M + 1)(J − M )YJM ( 1 .42) ( 1. −J. 1+ 2 ( 1. .25) using (6. .39) which we will try for a solution of (6. . 2 |Ω1 . 2 |Ω1 . therefore. Ω2 ) = Y 1 1 (Ω1 ) Y 2 2 (Ω2 ) (6. (6. 2 |Ω1 . Ω2 ) . J} by applying repeatedly J− YJM +1 ( 1 . (6. 2 |Ω1 . (6.37) and replace (1) (2) according to (6. Quantum mechanically this corresponds to a state G 1+ 2. 1+ 2 ( 1. have shown Y 1+ 2.39) into (6. The construction assumes a particular choice of J ∈ {| such J–value seeks an expansion (6. We embark on the suggested construction for the choice J = 1 + 2 .40) (Ω1 ) J+ Y (2) 2 2 J+ Y 1 1 (Ω1 ) Y (Ω2 ) + Y (Ω2 ) = 0 .11) J+ by J+ + J+ . 1+ 2 ( 1. 2 |Ω1 . . Ω2 ). J − 2.43) .38) for M = J − 1.37).68) J+ = (1) + J+ (1) (2) Y 1 1 (Ω1 ) Y 2 2 2 2 (Ω2 ) 1 1 (6. We ﬁrst seek an unnormalized solution GJJ and later normalize. 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.66. 2 |Ω1 . . 2 |Ω1 . M = −J. We obtain using (5. | 1 − 2| + 1. 1 + 2} and for (6.18) which satisﬁes J+ YJJ ( 1 .41) 2) Y (Ω1 ) Y (Ω2 ) . For this purpose we insert (6. −J + 1. . 1+ 2. 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. Similarly. . we can demonstrate condition (6. Ω2 ) 2 |Ω1 . Having determined such YJJ we then construct the whole family of functions XJ = {YJM ( 1 .5.37) The solution needs to be normalized.148 For the construction of YJM we will need the operators J± = J1 + iJ2 . .11) and (5. 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 ) (6.35) 1 − 2 |. .

. To satisfy this equation one needs to choose c = − 1 / 2 . 2 |Ω1 . . and all total angular momentum functions for a given choice of 1 and 2 . Show that the resulting functions are orthonormal. . ). 1 + 2 − 1.38) 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 ).45). ).s. | . . According to the condition (6. 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. . 2 = 1 in Table 1.21) 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. Ω2 ) 2 1+ 2 = (Ω1 )Y 2 2 −1 (6. The result is illustrated for the case 1 = 2. One can obviously continue the construction outlined to determine Y 1 + 2 −2 1 + 2 −2 . We seek for this purpose ﬁrst an unnormalized solution G 1 + 2 −1 1 + 2 −1 of (6. Exercise 6. . . . G 1 + 2 −1 1 + 2 −1 in (6.36.2.45) for some unknown constant c. with the symmetry pattern and with the terms Y 1 m1 Y 2 m2 contributing to each YJ M . 2 |Ω1 . The l.27). One can readily show that (6. This expression is orthogonal to (6. of (6.44) (Ω2 ) .36) we proceed as above and obtain J+ Y = (1) 1 1 −1 (Ω1 ) Y 2 2 2 2 (Ω2 ) + c Y 1 1 1 1 (Ω1 ) J+ Y (2) 2 2 −1 (Ω2 ) (6. We demonstrate the procedure ex(1) (1) plicitly only for M = 1 + 2 − 1. M = −( + − ).37). Ω2 ). The r. 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. all the Clebsch-Gordan coeﬃcients ( 1 + 2 M | 1 m1 2 m2 ). ( + )}. .s. | .h. The reader should familiarize himself with the entries of the Table.6.h. M = −( + ). Normalization yields furthermore Y = 1 1 + 2 −1 1 + 2 −1 We have thereby determined ( 1 2 |Ω1 Ω2 ) (6. . Ω2) for M = 1 + 2 − 2 and J = 1 + 2 . Expression (6. of (6.39) as required by (6. −( 1 + 2 − 1). . To demonstrate (6.2: Construction of Clebsch-Gordan Coeﬃcients 149 We now employ property (6. Having constructed this family we have determined the Clebsch-Gordan coeﬃcients ( 1 + 2 − 1M | 1 m1 2 m2 ). We like to construct now the family of total angular momentum functions B + − = {Y + − M ( .47) (Ω2 ) − 1 1 2 + Y 2 1 1 −1 (Ω1 )Y 2 2 + Y 2 1 1 (Ω1 )Y 2 2 −1 (Ω2 ) .38) yields pression 2( 1 + 2 )Y 1 + 2 1 + 2 −1 ( 1 .1: Construct following the procedure above the three functions YJM ( 1 . 1 + 2 − 3. 1 + 2 − 2.37) is satisﬁed. . etc. thereby. One obtains then Y 1 1+ 2 1 + 2 1 + 2 −1 2 2 ( 1.46) 2 1 +c Y (Ω1 ) Y 2 2 (Ω2 ) = 0 .47) can serve now to obtain recursively the elements of the family B + − for M = 1 + 2 − 2. . 6.38) to construct the family of functions B + = {Y + M ( . in particular. 2 |Ω1 . ( + − )}.

1|Ω1 . Ω2) Y2−2 (2. 1|Ω1 .57735 0.707107 0.774597 0.632456 −0. 1|Ω1 .447214 −0.408248 −0. Ω2) Y31 (2. Ω2) Y10 (2.57735 0.408248 −0. 1|Ω1 .707107 0. Ω2) Y32 (2. Ω2) Y11 (2. 1|Ω1 .57735 0. 1|Ω1 .547723 0. 1|Ω1 .547723 0.707107 0.316228 0.816497 −0. . 1|Ω1 .316228 0. Ω2) Y3−3 (2.258199 −0. 1|Ω1 .707107 0. 1|Ω1 .57735 0.774597 0 2 5 8 15 1 6 3 10 1 3 0.816497 0.258199 0.730297 0.1: Some explicit analytical and numerical values of Clebsch-Gordan coeﬃcients and their relationship to the total angular momentum wave functions and single particle angular momentum wave functions. Ω2) Y3−2 (2. 1|Ω1 .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 6.632456 0. Ω2) − − − − Y22 (Ω1 )Y11 (Ω2 ) 1 Y21 (Ω1 )Y11 (Ω2 ) 2 3 1 3 2 5 1 2 1 10 1 5 1 2 3 10 1 15 1 3 3 5 Y22 (Ω1 )Y10 (Ω2 ) 1 3 2 3 8 15 1 6 0.547723 0.150 Addition of Angular Momenta and Spin Y33 (2.632456 0. 1|Ω1 . Ω2) Y22 (2. Ω2) Y20 (2.57735 −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. 1|Ω1 . Ω2) Y30 (2.447214 0.57735 3 10 3 5 Y2−1 (Ω1 )Y11 (Ω2 ) Y20 (Ω1 )Y10 (Ω2 ) Y21 (Ω1 )Y1−1 (Ω2 ) −0.774597 − − − − 0. 1|Ω1 .730297 −0. Ω2) Y3−1 (2. Ω2) Y21 (2. Ω2) Y1−1 (2.547723 0. Ω2) Y2−1 (2. 1|Ω1 .

b† ζ = 0. Accordingly.ζ ζ. m + 1 . (6.3: Explicit Expressions of CG–Coeﬃcients 151 Exercise 6. aζ bζ . Deﬁnition of Spinor Operators for Two Particles In contrast to Sections 5.53) (6.9.2: Use the construction for Clebsch-Gordan coeﬃcients above to prove the following formulas J+M 1 for = J − 2 2J 1 1 1 = J. a† ζ ζ b† . b† ζ ζ = aζ .ζ a† < ζ |σk | ζ > aζ ζ b† < ζ |σk | ζ > bζ ζ (6. (6.287) to two particles (1) J k k = = (2) J 1 2 1 2 ζ.2. m − 2 . commute with each other ζ ζ aζ . 2 .49) where ζ. bζ = = a† . − 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.10 [cf. 1 (6.3 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 ). J.10 where we studied single particle angular momentum and spin.6. 6.54) 2 + = √ b† ( 2 +m2 )! √− ( 2 −m2 )! . These expressions will be derived now. M | . b† ζ ζ = 0 = 0 . For this purpsose we will employ the spinor operators introduced in Sections 5.10.50) (6. b† refer to diﬀerent particles and. a† ζ .51) The operators aζ .10. 1 . 5. The creation and annihilation operators are again of the boson type with commutation properties aζ . we are dealing now with two particles carrying angular momentum or spin. ζ = ± and the matrix elements < ζ |σk | ζ > are as deﬁned in Section 5. hence. we extent deﬁnition (5. + 2 − J−M +1 for = J + 1 2J+2 2 J−M for = J − 1 2J 2 = .9. M | . bζ = a† . a† and bζ . b † ζ = δζζ = δζζ . 5.52) and | 2 m2 2 According to Section 5. bζ . (5. aζ .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 . It is actually possible to state explicit expressions for any single coeﬃcient (JM | 1 m1 2 m2 ).48) (6.

m2 | 1 . 2) 1. M ( 1 .58) Jk + (2) Jk .59) We seek to determine states |J. M ( 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 . (6. 2) We will also require below the raising and lowering operators associated with the total angular momentum operator (6. J3 . k = 1. (2) J which.65) (6. mj j and. m1 . 2) . 2 ) which are simultaneous eigenstates of the operators 2 2 J2 . 1.300). as usual are denoted by their respective quantum numbers J.61) (6. 1 . M ( 1 .302. M ( 1 . m2 a† + ( 1 2 = a† − ( 1 1 −m1 (6. J3 |J.68) The states |J. in analogy to Eqs. M ( 1 . M ( 1 . M ( 2) 2) = J(J + 1) |J. ˆ kj |J. + 1) |J. (6. M ( 1 . (j) 2 ˆ ˆ J = kj ( kj + 1 ) . M ( 1 . m2 2 . 5. The aim of the present Section is to determine closed expressions for the Clebsch–Gordan coeﬃcents ( 1 . 2 .e. (6.63) according to At this point.67) = j |J.57) b† + ( (2) 2 2 +m2 1 +m1 b† − ( 2 2 −m2 + m1 )! − m1 )! (1) + m2 )! − m2 )! |Ψo . 2 . 2 .303). M ( 1 . k2 = b b+ + b† b− k1 = + − − 2 2 + namely.152 It holds. which describe a two particle system according to (6. m1 1 | 2 . 2 )) . M ( 1 . (5. = = 1( 1 2( 2 2) 2) (6. (1) J 2 |J.. 2) + 1) |J.62) . |J. m1 1 | 2 . 2) 2) can be expressed in terms of Clebsch-Gordan coeﬃcients (6. i. for such states should hold J2 |J. ˆ For the operators kj holds ˆ kj | j . . (1) J 2 . M.18) as follows = | 1 m1 . (6. 2 . m2 |J.60) (6. M ( 1 .55) .59) J± = J1 ± i J2 . = M |J.56) The states | 1 . can be expressed in terms of the number operators 1 † 1 ˆ ˆ a† a+ + a† a− . m1 1 | 2 . M ( (2) J 2 |J. The operator of the total angular momentum/spin of the two particle system is J = with Cartesian components Jk = (1) J + J (6. m1 . m2 |J.54). 2) 2) (6. 2 )). 2. 6. m2 1 2 ( 1 . M ( 1 . 3 . we like to recall for future reference that the operators (5. (1) J .64) (6. hence.66) (6. −J ≤ M ≤ J . (1) J (2) J 3 | 1 m1 1 3 | 2 m2 2 = m1 | 1 m1 = m2 | 2 m2 1 2 (6. are | 1 . j = 1.53. (6.69) − 2| ≤ J ≤ + 2 . (2) J 2 . mj j = j | j . M ( 1 . .

We note that. expressing (k) J 3 through the creation and annihilation operators according to (5. K† J3 .65) and (6. + − − + 2 2 = 1 † 2K . K † + K † 2 2 3 † †ˆ K kj + K .71–6. a† b† − a† a− . 2 4 = 0 = = 0. ˆ A similar calculation yields [k2 . the relationships (6.73) yields J3 . a† b† − a† b† − + − − + 2 + 1 1 = a† a+ .75) The relationships (6.71–6. + − + − 2 2 − + = . + − − + 153 (6.74) + 1 J− J+ + J2 .288). K † 1 † a a+ + a† a− . Employing (6. 2 2 3 ˆ = K † kj + K † . K † ] = (j) J 2 .72) (6.73. 6.50. This operator obeys the following commutation relationships with the other pertinent angular momentum / spin operators ˆ kj . (6. a† b† − a† b† + − + − + − − + 2 1 1 = a† a+ . 6. 4 Using J3 = (1) J 3 + (2) J 3 . due to J2 = K† 1 2 J+ J− 1 † K . a† b† + − − + 2 + 2 − 1 1 − b† a† b+ .73) can be readily proven.6. a† b† + + − − + 2 2 − 1 1 = a† b† − a† b† = K† .64. K † (kj + 1) 2 2 1ˆ † 1 ˆ kj K + K † (kj + 1) 2 2 1 ˆ 1 ˆ K † kj + kj . and applying the relationships (6.3: Explicit Expressions of CG–Coeﬃcients The Operator K † The following operator K † = a† b† − a† b† . K † (j) J 2 .73) (6. For example. (6. K † + kj . K † = 0. K † 2 1ˆ ˆ 1 ˆ ˆ kj kj + 1. j = 1. 6. b† = 0 .71) one can show K† = = = = = 1 ˆ ˆ kj (kj + 1). using (6.70) will play a crucial role in the evaluation of the Clebsch-Gordan-Coeﬃcients.71) (6. b† − b† a† b− . K † 1 a† a+ − a† a− + b† b+ − b† b− . K † J± . j = 1. a† b† + a† a− .74) imply 3 2 J2 .51) one obtains ˆ k1 .

. 6. 2 ) remain eigenstates of J2 and J3 with the same quantum numbers.73.77) where N is another normalization constant.75) ascertain that under the action of K † the states |J.e. 2 (6. One can generalize property (6. j2 ) = |j1 .292) one can derive similarly J+ .76) Here N is an unknown normalization constant. j1 + j2 (j1 + n n . 6. To demonstrate that the resulting states are eigenstates of (1) J 2 and (2) J 2 we exploit (6.63. M ( 1 .62. j1 + j2 (j1 . namely. 2 2 1 + ) . j1 1 |j2 . 2 ) . 2) 2) ˆ K † kj + = K† = j 3 † K + K † j ( j + 1) |J.78) a state which has been used already in the construction of Clebsch-Gordan coeﬃcients in Section 6.67) (j) J 2 K † |J. j + 2 2 2) However. according to (6. M ( 1 + n . 6. K † = Addition of Angular Momentum and Spin a† a− + b† b− .154 Starting from (5. 2) = = (j) 2 J . Action of K † on the states |J. a† b† + b† a† b− . 2 ) produces again total angular momentum eigenstates to the same J and M quantum numbers of J2 and J3 . 2 2 + n ) 2 (6. M ( 1 . j1 + j2 (j1 . 1 2 + 2 . We consider now the case that (K † )n acts on the simplest total angular momentum / spin state. 2) is a state with quantum numbers 1 1 + 1 2 and = N |J. a† b− − a† b† + + + − + = 0.2. on the state |j1 + j2 . j2 2 . 2) = N |J. it holds K † |J. j1 1 + . M ( 1 . M ( 1 . K † ] = 0 is demonstrated in an analoguous way. M ( 1 . = − a† a− . M ( 1 .77). j2 + ) 2 2 n n = N (n. The commutation properties (6. j1 + ) K † 2 2 (6. b† + − + + + − The property [J− . M ( 1 . 4 3 + j ( j + 1) |J. this result implies that K † |J.72) and (6. M ( 1 . 2) We want to demonstrate now that the action of K † on the states |J. Application of (K † )n to this state yields. 2 ) j + 4 1 3 + K † |J. M ( 1 + . K† + K † (j) J 2 |J. (6. M ( 1 . j2 ) .76) and state K† n |J. i. M ( 1 .79) n |j1 + j2 . M ( 1. |j1 + j2 . but for diﬀerent 1 and 2 quantum numbers of the operators (1) J 2 and (2) J 2 .

82.79). M ( 1 .81) = N( 1 + 2 1 2 − J.86) b† + J+ 2 − 1 |Ψo .82) 1 × | (J + 2 1 × | (J + 2 1 (J + 2 1 − 1 ) . − 1. 2) 1 2 = ( 1 + 2 (2J + 1)! − J)! ( 1 + 2 + J + 1)! 1 2 . 1 . 5. The latter states. 2) . M ) (J + × K† 1 − 2 )! (J + 2 − 1 )! J+ 1 − 1 + 2 −J (J− )J−M a† + × 2 (6. (6.105) of the spherical harmonics. J( 1 . 6. For this purpose one needs to choose in (6. M ( 1 . 2 ) for −J ≤ M ≤ J as follows |J. J( 1 . J( 1 . j2 as follows J = j1 + j2 which is equivalent to n j1 j2 Accodingly. j2 + n ). of (6.84) with (6.6. M ) (J− )J−M |J. 1 = j1 + n 2 . = (J + M )! (2J)! (J − M )! 1 2 2) (6. 2 ) can be expressed through the r. 2 = j2 + n 2 (6. M ( 1 . Combining (6. 2 ) ∆(J. j1 .3: Explicit Expressions of CG–Coeﬃcients 155 where we denoted the associated normalization constant by N (n. 2) Our construction of the states |J. holds |J. in analogy to the construction (5.85) ∆(J. Strategy for Generating the States |J.79) n. 2) 2) . M ( 1 . K† 1 2 1 + 2 −J × × .h.104. 2) = ∆(J. j1 + n . J( 1 . 2) = N( 1 + 2 − J. (6.f. (J + 2 − − 2) 1 1) 2 The normalization constant appearing here is actually N( + − J.74)] yields |J. (6. M ( 1 .80) = = = 1 1 2 2 − J 1 n − = (J + 2 2 n 1 − = (J + 2 2 + 1 1 − − 2) 2) . 1. M ) .82) for |J. allow one to obtain the states |J. 2 ) .s. 2 2 It is now important to notice that any state of the type |J. (6.83) The derivation of this expression will be provided further below (see page 158 ﬀ).57) and exploiting the fact that J− and K † commute [c. 2 ) exploits the expression (6.84) (6.

st The desired expansion of |Gr can be obtained from an alternate evaluation of I(λ. x.t (6.87) + − + − i.264).e. y) is a generating function for the states |Gr deﬁned in (6. in analogy to (5.87).69) yields then the Clebsch- Expansion of an Intermediate State We ﬁrst consider the expansion of the following factor appearing in (6. For this purpose we introduce the generating function I(λ. One obtains then using J− = a† a+ + b† b+ (6. y) = λr xs y t r |Gst .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.e. 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. + + Taylor expansion of the two exponential operators yields immediately I(λ.156 Addition of Angular Momentum and Spin Our strategy for the evaluation of the Clebsch-Gordan-coeﬃcients is to expand (6. m2 Gordan-coeﬃcients. (6. b† ] = 0 [cf.92) − − and noting [a+ .88) in terms of monomials (6.s. (6. + − + − ..93) = f (a† + λ a† ) g(b† + λ b† ) |Ψo .91) which. 2 [cf. r!s!t! r. m1 1 | 2 . follows from the commutation properties (6. (6. Comparision with (6.88). in terms of | 1 .52). I(λ. a† ] = [b+ .57)].86) |Gr = Jr a† st − + s b† + t |Ψo (6.50–6. x. x. x.89) i..90) (6.86) in terms of monomials 1 +m1 1 −m1 2 +m2 2 −m2 a† a† b† b† |Ψo . y) = exp (λJ− ) exp x a† exp x b† |Ψo .

t × λ v b† − × r−q v × = t s−u v λu a† − r! q |Ψo λr xs y t r! s! t! r. y) = s. y) = exp a† + λ a† + − exp b† + λ b† + − |Ψo .70) holds K† 1 + 2 −J = s 1 + 2 −J b† − s 1 + 2 −J−s (−1)s 1 + 2 −J−s (6.3: Explicit Expressions of CG–Coeﬃcients We conclude I(λ.86) (J− )J−M a† + = q J+ 1 − 2 b† + J+ 2 − 1 |Ψo 1 (6. 157 (6.90) allows one to infer |Gr s.97) yields.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. x. m2 2 .97).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. using the deﬁnition (6. |J. Expanding the exponentials in st (6.q (−1)s (6. x.98) a† − s a† + b† + s . using the commutation property (6.94) One can infer from this result the desired expressions for |Gr .94) yields I(λ.t = q r! × a† + s q s−q t r−q a† − q × b† + t−r+q b† − r−q |Ψo (6. one can write the right factor in (6. M ( 1 . Operation of this operator on (6.s.t × a† + s−q s q b† + t r−q t−r+q a† − q b† − |Ψo (6.6. to obtain the desired expansion of |J.50).99) . 2 ) in terms of states | 1 .96) and.88). 2) = s.95) Comparision with (6. With K † given by (6. M ( 1 . m1 1 | 2 .

according to (6.158 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. q = 1 − m1 − s and m2 = M − m1 . (6. 2 ) (6. J( 1 . m2 |J.69) if one identiﬁes m1 = 1 −q − s . j2 . For this purpose we introduce j1 = 1 (J + 2 1 2 − J.101) Note that m1 + m2 = M holds. (6.100) ( 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 . M ( 1 .101).q (−1)s × (6.82). J( 1 . n)|ψ(j1 . n) 2 )|J. Using (6. m1 . × 2) = N( 1 + 2 1 − J.53.6.82) and (6. 1 . − 1 .103) = 1. 2 ) we consider the scalar product J.104) . To determine N = N ( 1 + 2 −J. The summation over q corresponds then to the summation over m1 .83) of the normalization constant N ( 1 + deﬁned through (6.54) between creation operator monomials and angular momentum states allow one to write this |J. M ) (J + 2 )! (J + 2 − 1 )! s.69) since.102) We want to determine now the expression (6. m2 = M − 1 + q + s. m2 in (6.103) this can be written 1 = N 2 ψ(j1 . The Clebsch-Gordan coeﬃcents are then ﬁnally ( 1 . M − +q+s One can conclude that this expression reproduces (6. − q − s)!(q + s)! 2 × (M + 1 − 1 + q + s)!( 1 1 + 2 − M − q − s)! 2 − q − s 1 | 2. 2 ) ∆(J. n = 1 + 2 −J. j2 . j2 = 1 (J + 2 2 − 1) . √ 2J + 1 1 2 . M ) 1 = + J)!(− 1 + 2 + J + 1)! 2 2 + 2 − J)!( 1 − ( 1+ + J)! 1 2 × [( × s + m1 )!( − m1 )!( 2 + m2 )!( 2 − m2 )!(J + M )!(J − M )!] 2 (−1)s s!( 1 − m1 − s)!( 2 + m2 − s)! 1 2 1 × The Normalization 1 ( 1 + 2 − J − s)!(J − 1 − m2 + s)!(J − + m1 + s)! (6. 2 ) − 2) . 1 .

112) s .108) = λ=0 (n1 + ν)! n1 ! (6. j2 2 . (6. 1 1−λ n1 +1 1 1−λ n2 +1 = m1 .3: Explicit Expressions of CG–Coeﬃcients where |ψ(j1 .107) s = (n!)2 s 2j1 + n − s 2j1 The latter sum can be evaluated using 1 1−λ a property which follows from ∂ν ∂λν 1 1−λ n1 +1 n1 +1 = m1 n1 + m1 n1 λm1 (6. j2 .110) which can be written 1 1−λ n1 +n2 +2 = r s n1 + r − s n1 n2 + s n2 λr (6.104) 1 = N2 (n!)2 (2j1 )!(2j2 )! (2j1 + n − s)!(2j2 + s)! s!(n − s)! 2j2 + s 2j2 (6. m2 2 .6.m2 n1 + m1 n1 n2 + m2 n2 λm1 +m2 (6.108). n) in terms of states |j1 .109) and Taylor expansion of the left hand side of (6.106) The orthonormality of the states occurring in the last expression allows one to write (6. One obtains then.105) The ﬁrst step of our calculation is the expansion of ψ(j1 .111) Comparision with (6. (6. we obtain |ψ(j1 . n) = K† n 159 |j1 . We employ the expression (6. (6. m1 1 |j2 .70) for the operator K † .57) for these states and the expression (6. applying (6. j1 1 |j2 . j2 . j2 . Accordingly.108) twice.108) yields the identity n1 + r − s n1 n2 + s n2 = n1 + n2 + r + 1 n1 + n2 + 1 . 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 .

g. m2 |J.4 Symmetries of the Clebsch-Gordan Coeﬃcients The Clebsch-Gordan coeﬃcients obey symmetry properties which reﬂect geometrical aspects of the operator relationship (6.120) can be readily derived from the expression (6.116) through formula (6. (6. m1 |J. To derive relationship (6. M ) ( 1 .114) one obtains − J = − J (1) − J (2) The respective Clebsch-Gordan coeﬃcients ( 1 . For the quantum mechanical addition of angular momenta the Clebsch Gordan coeﬃcients ( 1 . 6.102) of the Clebsch-Gordan coeﬃcients. m1 |J. 2 .116) one expresses the Clebsch-Gordan coeﬃcient on the r.114) as long as J. 2 . M ) corresponding to (6. m1 .11) J = J (1) + J (2) . m1 . M | 1 . m1 ) by ( 2 . m1 .160 Applying this to (6.102) by replacing ( 1 .s. m1 . (6. 1 . 2 . (6. −m2 . M ) . and J (2) are vectors in R . M ) . M ) corresponding to (6. −M ) . e. m2 |J. −m1 .120) The symmetry properties (6. m2 |J. m2 .114) For example.116).h.116) If one takes the negatives of the operators in (6. −M ) = (−1) 1 + 2 −J ( 1 . and .h.h. M ) = (−1) 2 +m2 2J + 1 ( 2 . m2 ) by ( 1 . m1 ).118) Finally.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 . namely. 2 .119) The corresponding symmetry property of the Clebsch-Gordan coeﬃcients is ( 1 . (6. M ) = (−1) 1 + 2 −J ( 2 . interchanging the operators J (1) and J (2) results in J = J (2) + J (1) . one can interchange also the operator J on the l. 2 . ( 2 .114) by. m1 . 1 . −m1 2|J.s.s. of (6.115). J (1) . −m1 . m2 ) and. J.115) This relationship is a trivial consequence of (6. m2 |J. m2 .103) one obtains the desired result (6. (6. J (1) on the r. vice versa. −m2 |J. We will demonstrate this now.118). of this equation J (1) = J (2) − J . m1 ) .113) Using the identities (6. (2j1 + 2j2 + 1)! (6.117) are again related in a simple 2 . (6. of (6.83). ( 1 .114) show a simple relationship to the Clebsch Gordan coeﬃcients ( 2 . and (6.. (6. manner to the coeﬃcients ( 1 . m2 |J. 2 . 2 1+1 (6.

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

183) in case θ = 0 becomes in the standard representation with respect to the spin. For this purpose we consider the application of σ · r in the case of θ = 0. ˆ (6. φ) and (5. φ) 2 2 2 .f.185) We have.168 Addition of Angular Momentum and Spin 1 Yjm (j ± 2 . therefore. (6.1 states χ 1 ± 1 2 2 2 [c.180) and (6.177) one notes that for φ = 0 the spin-angular momentum states 2 are entirely real such that σ · r must be real as well. 2 |ˆ) and Yj 1 (j + 1 .148) the particular ˆ 1 1 states Yj 1 (j − 2 .182) σ · r Yj 1 (j − 1 .180). 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 . φ) . 6. We want to determine ˆ its action on the wave functions f (r) Yjm (j ± 1 . ˆ 2 2 2 2 2 2 for θ = 0.188) . 1 |ˆ) + f (r) σ · p Yjm (j ± 1 . hence. identiﬁed the sign of (6. for θ = 0 (6. 6. φ) 2 2 2 = − j+ 1 2 4π (6.182) 0 Since σ · r. ˆ (6. We want to demonstrate ﬁnally that the “−”-sign holds in (6. 1 |θ = 0.181) 0 = j+ 1 2 4π Yj 1 (j + 1 .187) 1 2 .224)] 1 0 σ·r = ˆ .180) where the sign could depend on j.147.174–5.186) Here (( · · · )) denotes again conﬁnement of the diﬀusion operator ∂r to within the double bracket. have proven (6. 1 |θ = 0. φ) = − Yj 1 (j + 1 . 1 |ˆ) r 2 (6. 1 |ˆ) = ((σ · p f (r)))Yjm (j ± 1 .160).180) and. Since f (r) is independent of θ and φ follows ˆ ((σ · p f (r))) = −i (( ∂r f (r) )) σ · r . 1 |ˆ) r r r 2 2 2 2 2 2 (6. given by ˆ σ · r = σ1 sin θ cos φ + σ2 sin θ sin φ + σ3 cos θ . According to (5. (6.181. 1 |ˆ) = −Yjm (j ˆ r 2 2 ˆ The Operator σ · p ˆ The operator σ · p plays an important role in relativistic quantum mechanics. 1 |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 .184) 0 1 space χ 1±1 2 2 one can conclude from (6. 1 |ˆ). 1 |θ = 0. 1 |θ. 1 |ˆ) at θ = 0 are r r 2 2 2 2 Yj 1 (j − 1 . (5.

5: Spin–Orbital Angular Momentum States Using (6.191) (6. h r (6.196) = i j+ ∂r + r 1 2 3 2 1 f (r) Yjm (j − 2 .. r r r Using ∂2 (1/r) = −x2 /r3 . ∂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. r 2 (6.155. Corresponding results are obtained for the other components of p × r and.198) . 6. using ∂r Yjm (j ± 1 .230) allows one to express ˆ ˆ ˆ ˆ ˆ σ · p σ · r = p · r + i σ · (p × r) .190) 3 h − i hr · r 1 − i r· ˆ r h = ∂r h one can conclude ˆ ˆ p · r h = −i 2 + ∂r r h (6. r 2 2 1 J ·S 2 1 ˆ σ · p f (r) Yjm (j ± 2 . To demonstrate this we consider one of its cartision components.193) r r 1 1 1 = −i (∂2 x3 − ∂3 x2 ) h − i h (x3 ∂2 − x2 ∂3 ) . ∂3 (1/r) = −x3 /r3 and ∂2 x3 = x3 ∂2 . x3 x2 ˆ ˆ ( p × r )1 h = −i (∂2 − ∂3 ) h (6. 1 |ˆ) = i [ ∂r + r + ] f (r) Yjm (j 2 r r Using (6.187) one obtains ˆ σ · p f (r) Yjm (j ± 1 .156) this yields ﬁnally 1 ˆ σ · p f (r) Yjm (j + 2 .6. 1 |ˆ) r 2 1 2 (6. hence.g. we conclude the intuitively expected identity 1 ˆ ˆ p×r = − J . 1 |ˆ) r 2 (6.194) ˆ ˆ where J1 is deﬁned in (5. r Altogether we obtain. e.190) can be related to the angular momentum operator.197) 1 ˆ σ · p g(r) Yjm (j − 2 . 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.53).186) for both terms in (6. 1 |ˆ) = 0 and σ = 2S/ . 1 |ˆ) r 2 = i ∂r + −j r 1 g(r) Yjm (j + 2 . 1 |ˆ) r 2 (6.195) . 1 |ˆ) .192) ˆ ˆ The operator p × r in (6.189) The celebrated property of the Pauli matrices (5. 1 |ˆ) = r 2 2 ˆ i ∂r f (r) − f (r) σ · p σ · r Yjm (j ˆ 1 2 169 . 1 |ˆ) r 2 (6.

203) (6. consider ﬁrst the case of the largest m-value m = j. j 2 + 2j + 2 3 4 (6.197. r ∂r2 r 2 (6. 1 |ˆ) r r 2 2 2 2 2 2 2 yields ˆ (σ · p)2 f (r) Yjm (j + 1 . j) 2 2 2 2 (j − . − |j.170 Addition of Angular Momentum and Spin To demonstrate the validity of this key result we note that according to (5.144) and. . r 2 2 2 2 2 2 − r 2 ∂r r + J2 r2 f (r) Yjm (j + 1 . 6. j − 1 .230. 6. adopting the method in Sec. 6. 1 |ˆ) = r 2 2 which agrees with (6. 5. Yjj−1 (j − 1 . j + . in fact. We note ˆ (σ · p)2 f (r) Yjm (j + 1 .45).198). according to (6. In this case holds. 1 |ˆ) r 2 2 r 2 3 i i 1 2 1 r = [ i∂r + ( − j) ] [ i∂r + (j + ) ] f (r) Yjm (j + 2 . according to (6.151). 1 . j) 1 2 1 2 1 2 1 2 = = 1 0 (6.143. 1 |ˆ) = Yj− 1 j− 1 (ˆ) χ 1 1 . using (5.205) which agrees with the expressions (6. 6.144) for m = j. are consistent with this identity.143. 1 |ˆ) r 2 2 r r2 r2 j 2 + 2j + 3 2 2 2 4 = [ −∂r − ∂r + ] f (r) Yjm (j + 1 . For this purpose we use the construction method introduced in Sec.204) (6. 1 |ˆ) r 2 2 The Clebsch-Gordan coeﬃcients are then (j − 1 .199).99) holds ˆ (σ · p)2 = − 2 2 = ∂2 J2 r + 2 . For m = j − 1 one can state. 1 |j.2. 1 |ˆ) r 2 2 i 3 ˆ = σ · p [ i∂r + (j + ) ] f (r) Yjm (j − 1 . 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 .2. 1 |ˆ) r 2 2 r r2 and. 1 |ˆ) = J 2 Yjm (j + 1 .. 1 |ˆ) = r 2 2 2j − 1 Yj− 1 j− 3 χ 1 1 + 2 2 2 2 2j 1 Y 1 1 χ1 1 2j j− 2 j− 2 2 − 2 (6. i.e.201) 1 3 (j + )(j + ) Yjm (j + 1 . 6. We begin with the coeﬃcients (6.146). r (6.200) = (j + 1 )(j + 3 ).206) .143–6.202) Yjj (j − 1 . 2 2 (6. (6. as well as (6.43).101).199) We want to show that eqs. Evaluation of Relevant Clebsch-Gordan Coeﬃcients We want to determine now the Clebsch-Gordan coeﬃcients (6.

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 . 2 2 2 2 2j (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 . The further Clebsch-Gordan coeﬃcients (· · · |jj − 2).f. (6. 2 2 2 2 2j . 6.208) (6.202). is obtained by applying the operator [c. 1 .s. (6. 1 |j.2.f.h. 6.210) to (6. m − 3 . are obtained by iterating the application of (6.211) to the l. Expression (6. etc. j − 1) 2 2 2 2 which again agrees with the expressions (6.143..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 . Let us verify then the expression (6.137)] J− (tot) = J− + S− (6.144) implies for j = m Yjm (j − 1 . j − 3 . as described in Sec.144) for m = j − 1.s. m − 1) 2 2 2 2 = j +m−1 2j (6. 1 |ˆ) = r 2 2 j+m 2j + (j + m − 1)(j − m + 1) Yj− 1 m− 3 χ 1 1 2 2 2 2 j+m Yj− 1 m− 1 χ 1 − 1 2 2 2 2 2j j−m 2j (j + m)(j − m) Yj− 1 m− 1 χ 1 − 1 2 2 2 2 + or Yjm−1 (j − 1 . and J− + S− to the r.143. j − 1 .206). j − 1) 2 2 2 171 = = 2j − 1 2j 1 2j (6.207) (6. 6. 1 . (· · · |jj − 3). 2 |j.210)] yields (j + m)(j − m + 1) Yjm−1 (j − 1 .143.144) for the Clebsch-Gordan coeﬃcients by induction. 1 . 6. (6.5: Spin–Orbital Angular Momentum States The corresponding Clebsch-Gordan coeﬃcients are then 1 (j − 1 . − 1 |j.210).209) (j − 1 . [c.h.

j3 .144) for j = m − 1. j2 m2 ) 2 . the relationship (6.216) 6. m + 1 . using (6. m3 . m1 . m1 . m2 1 (6. − 1 |j. m − 1) 2 2 2 2 = (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 = 0.146) can be obtained from (6. 6. 1 .6 The 3j–Coeﬃcients The Clebsch-Gordan coeﬃcients describe the quantum mechanical equivalent of the addition of two (1) (2) classical angular momentum vectors Jclass and Jclass to obtain the total angular momentum vector (tot) (1) (2) (1) (2) Jclass = Jclass + Jclass .143.120). However. 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 . hence. M. j2 . m3 .214) 2 2 +1 ( 3 . 6. j+m+1 . The Clebsch-Gordan coeﬃcients (6. 2 . In this context Jclass and Jclass play the same role. m) 2 2 2 2 2j + 1 (6. one can obtain expression (6. 1 |j + 1 . − 1 |j.143). all three vectors Jclass . 6. m2 . 2j + 1 (6.215). m1 ) (6. − 1 |j. m + 1 ) = 2 2 2 j+m+1 . 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 . Jclass and Jclass play equivalent roles. 2 .116. m2 | 3 . −m2 | 1 .144).143. The latter relationships applied together read ( 1 .218) where we have replaced the quantum numbers J. m3 ) = (−1) × 2 + 3 − 1 + 2 +m2 × 2 . m − 1 . to reﬂect in the notation the symmetry of these quantities.215) which follows from (6. m.213) which is in agreement with (6. 1 . 1 |j + 1 . proven (6.214) yields (j. (1) (2) (−tot) The coeﬃcients which are the quantum mechanical equivalent to Jclass + Jclass + Jclass = 0 are the 3j–coeﬃcients introduced by Wigner. We have.217) 2j + 2 (j + 1 . 1 . 1 . m + 1 ) = 2 2 2 2 and.143) by applying the symmetry relationships (6. 2j + 2 (6.172 Addition of Angular Momentum and Spin j −m+1 2j (j − 1 . m. m) = 2 2 2 2 Similarly. (1) (2) (−tot) Obviously. m + 1 . 1+1 3 For 1 (j.144) by induction. one obtains (j + 1 .145) from (6. 1 . m1 . by the set j1 .

222) m1 m2 m3 j1 + m1 j2 + m2 j3 + m3 The Regge symbol vanishes.6: The 3j–Coeﬃcients We ﬁrst like to point out that conditions (6. for which no known classical analogue exists. The Regge symbol also vanishes in case that two rows or columns are identical and Σ is an odd integer. m2 and j3 . m2 . e. The two integer entries J − 1 − m2 and J − 2 + m1 in (6. m3 . the Regge-symbol.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 . as follows −j1 + j2 + j3 j1 − j2 + j3 j1 + j2 − j3 j1 j2 j3 j1 − m1 j2 − m2 j3 − m3 .21. −j1 + j2 + j3 . with respect to alltogether 12 symmetry operations. j3 form the sides of a triangle and the condition is symmetric in the three quantum numbers. i. one can exchange columns and one can reﬂect at the diagonal (transposition). According to the deﬁnition of the 3j–coeﬃcients one would expect symmetry properties with respect to exchange of j1 . For this purpose one can use the symmetry transformations to place the smallest element into the upper left corner of the Regge symbol.223) (Σ + 1)!n11 !n22 !n33 !n23 !n32 !)) n=0 n31 n32 n33 .6.218) relating 3jcoeﬃcients and Clebsch-Gordan coeﬃcients. discovered by Regge.102) of the Clebsch-Gordan coeﬃcients. = (6. The Regge symbol reﬂects a remarkable degree of symmetry of the related 3j–coeﬃcients: One can exchange rows. m1 . To represent the full symmetry one expresses the 3j–coeﬃcients through a 3 × 3–matrix. 6. m3 and with respect to sign reversals of all three values m1 . In case of a non-cyclic exchange of rows and columns the 3j–coeﬃcent assumes a prefactor (−1)Σ . there exist even further symmetry properties.. 173 (6. j2 .219) (6. states that j1 .102) are obtained each through the diﬀerence of two entries of the Regge-symbol. Assuming this placement the Regge symbol can be determined as follows (n11 is the smallest element!) n11 n11 n12 n13 n12 !n13 !n21 !n31 ! n21 n22 n23 = (−1)n23 +n32 sn (6. Because of its high degree of symmetry the Regge symbol is very suited for numerical evaluations of the 3j–coeﬃcents. except when all elements are non-negative integers and each row and column has the same integer value Σ = j1 + j2 + j3 . √ safe for the prefactor 2j3 + 1 which is cancelled according to the deﬁnition (6. the so-called triangle condition. The reader may note that the entries of the Regge-symbol.221) j1 j2 j3 −m1 −m2 −m3 where the the results of a cyclic. In this way the values of 12 3j-coeﬃcients are related. are identical to the integer arguments which enter the analytical expression (6. These symmetry operations relate altogether 72 3j–coeﬃcients.g.220) The latter condition |j1 − j2 | ≤ j3 ≤ j1 + j2 . anti-cyclic exchange and of a sign reversal are stated.e. However. 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. j2 .

m) δ(−m1 . m1 + m2 ) (2 − m1 )! (2 + m1 )! √ 6 (2 − m)! (2 + m)! (1 − m2 )! (1 + m2 )! 1 ≥ |m2 | ∧ 2 ≥ |m| ∧ 2 ≥ |m1 | (6.174 where Σ s0 sn = − Addition of Angular Momentum and Spin = n11 + n12 + n13 = j1 + j2 + j3 23 ! 32 ! = (n23n− n11 )! (n32n− n11 )! (n11 +1−n)(n22 +1−n)(n33 +1−n) n(n23 −n11 +n)(n32 −n11 +n) (6. (6. (6.228) √ (−1)2m1 −2m2 7 δ(m.232) .10752786393409395427444450130056540562826159542886 We also illustrate the numerical values of a sequence of 3j-coeﬃcients in Figure 6.225) sn−1 . m2 ) √ 2 1 ≥ |m1 |) 2 (6.231) Here is an explicit value of a Clebsch-Gordan coeﬃcient: 1 1 (70 2 −15 1 |120 −10 60 2 −5 1 ) = 2 2 √ √ √ 4793185293503147294940209340 127 142 35834261990081573635135027068718971996984731222241334046198355 0.229) 1 1 (0m| m1 m2 ) = 2 2 i(−1)1−m2 δ(0.223-6.227) (2m|2m1 1m2 ) = (−1)m+m1 (m + 2m2 ) δ(m.1. m1 + m2 ) (3 − m)! (3 + m)! (3m|2m1 1m2 ) = √ 105 (2 − m1 )! (2 + m1 )! (1 − m2 )! (1 + m2 )! 1 ≥ |m2 |) ∧ 2 ≥ |m1 | ∧ 3 ≥ |m| (6.226) by means of a symbolic manipulation package (Mathematica) (1m|2m1 1m2 ) = (−1)1+m+m1 δ(m. m1 + m2 ) (2 − m1 )! (2 + m1 )! √ 10 (1 − m)! (1 + m)! (1 − m2 )! (1 + m2 )! 1 ≥ |m| ∧ 1 ≥ |m2 | ∧ 2 ≥ |m1 | (6. m1 + m2 ) (1 − m)! (1 + m)! √ 1 1 1 1 6 2 − m1 ! 2 + m1 ! 2 − m2 ! 2 + m2 ! 1 1 ≥ |m1 | ∧ ≥ |m2 | ∧ 1 ≥ |m| 2 2 (6.224) (6.230) 1 1 (1m| m1 m2 ) = 2 2 √ (−1)2m1 −2m2 3 δ(m.226) We like to state ﬁnally a few explicit analytical expressions for Clebsch-Gordan coeﬃcients which were actually obtained using (6.

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

each of the blocks in (6. 1 . 1). Exercise 6. . . . (2( 1 + 2 ) + 1) ×(2( 1 + 2 ) + 1) Partitioning in smaller blocks is not possible. 2 = 1.233) is further block-diagonalized as follows + 2. . . (6.6.6.176 Addition of Angular Momentum and Spin 1 For the basis {YJM ( 1 . 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. . Ω2 ). . 2 |Ω1 . M = −J. .233.2: 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. . .234) Exercise 6. J = | 1 − 2 |.234) .3: Two triplet states |1m1 (1) |1m2 (2) are coupled to an overall singlet state Y00 (1. .. .1: Prove Eqs. 2. Show that the probability of detecting a triplet substate |1m1 (1) for arbitrary polarization (m2 – 1 value) of the other triplet is 3 . Exercise 6.6. .6.

q = −k.42) in case (i) of the position representation of single particle wave functions or the operator (5. 6. . .237) are called tensor operators of rank k. that T belongs to a family of operators {Tkq .g.236) In this equation Dq q (ϑ) denotes the rotation matrix Dq q (ϑ) = kq |R(ϑ)|kq . that in case (i) T is an operator C∞ ( ) → C∞ ( ) acting on single particle wave functions. i.6.222) in case (ii) of single particle spin operators. −k + 1. Such operators T can be characterized through their behaviour under rotational transformations.e. This implies. . ∂x2 3 1 2 2 2 2 case (ii) the operator T could be a spin operator Sk deﬁned in (5.g. Rotations transform |ψ as |ψ = R(ϑ)|ψ and T |ψ as R(ϑ)T |ψ .238) . Let T |ψ denote the state obtained after the operator T has been applied and let R(ϑ) denote a rotation in the representation of SO(3) or SU(2) which describes rotational transformations of the quantum states under consideration. The latter implies that T transforms like an angular momentum or spin state | m . S 2 = S1 + S2 + S3 or any other polynomial of Sk . Note that in the examples mentioned the operator T would be deﬁned within the same representation as R(ϑ).236. . some of the r r examples considered below involve tensor operators T of this type. k} such that k Tkq = q =−k (k) Dq q (ϑ) Tkq . In fact. q = −k. e. for example. the operator (5. . (k) (6. e.223). e. The latter can be written T |ψ where T denotes T in the rotated frame given by T = R(ϑ) T R−1 (ϑ) .235) The property that T imparts onto states |ψ angular momentum or spin corresponds to T behaving as an angular momentum or spin state multiplying |ψ . k} with the transformation property (6. −k + 1.7 Tensor Operators and Wigner-Eckart Theorem In this Section we want to discuss operators which have the property that they impart angular momentum and spin properties onto a quantum state. (6. (k) (6.7: Tensor Operators 177 6. .237) The operators T ∈ {Tkq .g. Examples of Tensor Operators The multiplicative operators C∞ ( ) → C∞ ( ) Y ( ) = def Y ( ) (6. 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). The operator T may also act on multi-particle states like Y 1 m1 (ˆ1 )Y 2 m2 (ˆ2 ).

3. This choice of parametrization allows us to derive conditions which are equivalent to the property (6.237) do not require that the rotation R−1 (ϑ) is expressed in terms of Euler angles according to (5. 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. The fact that these operators form tensor operators of rank 1.1: Show that the following set of spin operators T00 T1±1 T10 T2±2 T2±1 T20 = 1 = 1 √ S± 2 S± 2 = S3 = = = (S3 S ± + S ± S3 ) 2 2 (3S3 − S 2 ) 3 are tensor operators of rank 0.178 are tensor operators of rank k.203). In fact. x3 . Exercise 6.239) These operators can be expressed in terms of the coordinates x1 . 2. .235 . x2 . For the following it is important to note that the rotation matrix elements (6.7.240) 1 where we have deﬁned ϑ± = 2 (ϑ1 iϑ2 ) and L± = L1 ± iL2 . Exercise 6. we will assume presently that the rotation is chosen as follows R(ϑ) = exp ( ϑ+ L+ + ϑ− L− + ϑ3 L3 ) (6.6.237). 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. but are far easier to ascertain for any particular operator.7. 2. 3 follows from the transformation properties of the spherical harmonics derived in Section 1. but rather any rotation and any parametrization can be assumed.

6.237) yields ﬁrst k R(ϑ) Tkq R −1 (ϑ) = q =−k kq |R(ϑ)|kq Tkq .242) from which follows by means of kq |1 1|kq equation ϑ+ [L+ .80) follows kq |L+ |kq = −i (k + q + 1)(k − q)δq and. 1 (6.248) a tensor operator? Exercise 6.237) for transformations characterized by inﬁnitesimal values of ϑ+ .6. (6. Show that S (1) · S (2) is a tensor operator of rank 0 in the space of the products of the corresponding spin states | 1 m1 (1) | 1 m2 (2) .247). Tkq ] = = = Tkq+1 Tkq−1 q 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) . .4: Is the 1-particle Hamiltonian 2 H = − 2 2m + V (|r|) (6.235 . 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) .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 . J− . ϑ3 )T . The property (6. ϑ− . J3 yields [J+ .7: Tensor Operators 179 The conditions can be derived if we consider the property (6. Exercise 6. 0)T in case of small ϑ+ . 0)T . 0.6. hence.246) (6. (6. We ﬁrst consider a rotation with ϑ = (ϑ+ .247) These properties often can be readily demonstrated for operators and the transformation properties (6.246.7.237) be assumed then.7. Tkq ] = = δqq and by subtracting Tkq on both sides of the k ϑ+ kq |L+ |kq Tkq .5: Consider a system of two spin.7.235 . q =−k q+1 (6.243) From (5. 6.6. Exercise 6.244) [L+ . ϑ3 . ϑ− .245) (6. Tkq ] [J− . Expressing the results in terms of the angular momentum operators J+ .6: Form tensor operators of rank 1 in terms of the three components of acting on the space of 1-particle wave functions. Tkq ] [J3 .235 . 2 Exercise 6.3: Derive Eqs. (0. Tkq ] = −iTk q+1 (k + q + 1)(k − q) . (6.7. 0. (k + q + 1)(k − q) (k + q)(k − q + 1) (6. Similar equations can be derived for inﬁnitesimal rotations of the form ϑ = (0.

e. 2 |γ2 ) is an eigenstate of J 2 with eigenvalue + 1). 2 |γ2 ) which correspond to total angular momentum states. J3 | 2 m2 γ2 = m2 γ2 | 2 m2 and the property (6. 2 |γ2 ) can be shown readily as follows using (6.m2 ( 1 m1 |kq 2 m2 ) (J3 Tkq − Tkq J3 + Tkq J3 ) | 2 m2 γ2 = qTkq = q. 1 m1 |Φ 1 m1 (k.180 Theory of Angular Momentum and Spin 6.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 .8 Wigner-Eckart Theorem A second important property of tensor operators Tkq beside (6. J 2 | 2 m2 γ2 = 2 2( 2 + 1) | 2 m2 γ2 . (6. J3 are the generators of the rotation R(ϑ).235 . hence. Before we proceed we like to point out that the states | 2 m2 γ2 are also eigenstates of J 2 . i. J2 .18) are deﬁned through Φ 1 m1 (k.247). construct states Φ 1 m1 (k.1: Show that Φ 1 m1 (k. Starting point of the derivation is the fact that the states Tkq | 2 m2 γ2 behave like angular momentum states of a composite system of two particles each carrying angular momentum or spin.m2 ( 1 m1 |kq m1 q. | 1 m1 γ1 denotes an angular momentum (spin) state. 2 |γ2 ) = Cδ 1 1 δm1 m1 2 1( 1 (6.e. i. one can show that Φ 1 m1 (k. 2 |γ2 ) = q. The hermitian property of J3 and J 2 implies that states Φ 1 m1 (k.e. m1 . the stated property. i.8.237) is that their matrix elements 1 m1 γ1 |Tkq | 2 m2 γ2 obey simple relationships expressed in terms of Clebsch-Gordan coeﬃcients. 2 |γ2 ) are orthogonal to | 1 m1 in case of diﬀerent quantum numbers 1 . One can. . The relationships among the matrix elements 1 m1 γ1 |Tkq | 2 m2 γ2 are stated by the Wigner–Eckart theorem which we will derive now. 2 |γ2 ) is an eigenstate of J 2 with eigenvalue 2 1 ( 1 + 1).250) q. which is characterized also by a set of other quantum numbers γ1 which are not aﬀected by the rotational transformation R(ϑ). To prove this we consider the transformation of Tkq | 2 m2 γ2 R(ϑ) Tkq | 2 m2 γ2 = R(ϑ) Tkq R−1 (ϑ) R(ϑ) | 2 m2 γ2 = q m2 Dq q Tkq Dm2 m2 | 2 m2 γ2 2 (k) ( ) (6. J3 .m2 2 m2 ) (q + m2 ) Tkq | 2 m2 γ2 | 2 m2 γ2 (6. possibly the total angular momentum– spin state of a compositie system. (6. 2 |γ2 ) = ( 1 m1 |kq 2 m2 ) Tkq | 2 m2 γ2 .249) which demonstrates.253) Exercise 6.252) ∼δm1 q+m2 = ( 1 m1 |kq 2 m2 ) Tkq Similarly. in fact.21) of Clebsch-Gordan coeﬃcients J3 Φ 1 m1 (k.251) The corresponding property for Φ 1 m1 (k. behave like |kq | 1 m1 .6. J3 | 2 m2 γ2 = m2 | 2 m2 γ2 . These states according to (6.

m2 .257) which establishes the m1 –independence of 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) . γ2 is determined by applying (6.260) Exercise 6. γ2 .h. γ2 = 2 1 +1 1 m1 .s. γ2 (−1)k− 2 + 1 ( 1 m1 |kq 2 m2 ) (6.259) to a combination of magnetic quantum numbers 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 ) (6.261) .8: Wigner-Eckart Theorem 181 In order to evaluate the matrix elements (6.8.32) Tkq | 2 m2 γ2 = and orthogonality property (6. m1 = q = m2 = 0.256) and noting that the operator adjoint to J+ is J− . (6.g. γ2 ( 1 m1 |kq (6. One can then evaluate also the corresponding Clebsch-Gordan coeﬃcient ( 1 m1 |kq 2 m2 ) and determine 1 . To prove this property we consider 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) for a diﬀerent m1 value.255) At this point the important property can be proven that 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) is independent of m1 . γ1 ||Tk || 2 . can be evaluated as easily as possible. its m1 –dependence being expressed solely through a Clebsch-Gordan coeﬃcient.259) The socalled reduced matrix element 1 . 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. e. γ1 ||Tk || 2 . say m1 + 1. for which the l. Using | 1 m1 + 1γ1 = 1 ( 1 + m1 + 1)( 1 − m1 ) J+ | 1 m1 γ1 (6.258) We can then ﬁnally express the matrix elements of the tensor operators Tkq as follows 1 m1 .6. (6. q . γ1 ||Tk || 2 . γ1 |Tkq | 2 m2 .e. the matrix elements 1 m1 γ1 |Tkq | 2 m2 γ2 can be reduced to an m1 –independent factor. 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 .254) = 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) ( 1 m1 |kq 2 m2 ) . γ2 k− 2 + 1 √ 1 2 m2 ) (−1) 2 1 +1 = 1 . i. γ1 ||Tk || 2 . γ1 |Tkq | 2 m2 .2: Determine the matrix elements of the gradient operator d3 rF (r) G(r) of the type (6.

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

for example. in case of the so-called bag model of hadronic matter. i.. The corresponding wave functions serve basis functions for multi-electron systems in atoms.1. The ﬁrst potential V (r) = 0 0 ≤ r ≤ R ∞ r > R (7. in potentials which are solely a function of r and are independent of the angles θ.1) conﬁnes a freely moving particle to a spherical box of radius R. and crystals. Switzerland.4) governs the motion of electrons in hydrogen-type atoms. The potential (7. 1993) 1 183 . φ. See.e. Talmi (Harwood Academic Publishers. Potential (7.Chapter 7 Motion in Spherically Symmetric Potentials We describe in this section the stationary bound states of quantum mechanical particles in spherically symmetric potentials V (r). (7. 7. molecules.3) describes an isotropic harmonic oscillator . 7000 Chur. for example. The potentials (7.3) describes the motion of a charge in a uniformely charged sphere and can be employed to describe the motion of protons and neutrons in atomic nuclei1 . The fourth potential V (r) = − Z e2 r (7.2) serve as schematic descriptions of quantum particles.4) is by far the most relevant of the four choices. Four examples will be studied.2) V (r) = 0 r > R The third potential V (r) = 1 mω 2 r2 2 (7. Simple Models of Complex Nuclei / The Shell Model and the Interacting Boson Model by I. Poststrasse 22. It leads to the stationary electronic states of the hydrogen atom (Z = 1). The second potential is of the square well type −Vo 0 ≤ r ≤ R .

(7. Jk } where H is the Hamiltonian. J ψE. 3 . 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 .184 Spherically Symmetric Potentials 7.m (r) m ψE.m (r) = = = E ψE.6) ˙ Since Jk is also equal to the poisson bracket {H.m (r) . ˆ dt (7. k = 1. k = 1.14) (7. 2. Jk } = 0 . Accordingly. (7. ( + 1) ψE.m (r) .5) and r = v. i. ˆ A classical particle moving in a potential V (r) is governed by the Newtonian equation of motion (7. one can state ˙ J = m r2 θ ..11) (7.e.99)] 2 1 2 J2 2 2 = − ∂r r + . For this purpose one concludes ﬁrst that the conservation of angular momentum J implies a motion of the particle conﬁned to a plane.8) This property can also be proven readily employing the expression of the kinetic energy operator [c.85– 5. 3.13) .m (r) (7. 2 .7) ˆ The correspondence principle dictates then for the quantum mechanical Hamiltonian operator H and angular momentum operators Jk ˆ [H. Jk ] = 0 . θ for the distance from the origin and for the angular position. Employing in this plane the coordinates r. (5. one can conclude ˙ Jk = {H.m (r) can be chosen as simultaneous eigenstates of the Hamiltonian ˆ operator H as well as of J 2 and J3 . using (7.87) for Jk as well as the commutation property (5.m (r) .1 Radial Schr¨dinger Equation o ˙ mv = − er ∂r V (r) .61) of J 2 and Jk .10) (7. . stationary states ψE. J3 ψE.9) − 2m 2m r 2mr2 the expressions (5. 2 . 2. In fact. (7.15) (7.12) .f. ˆ H ψE. d ˙ J = v × mv + r × mv = 0 + r × er (−∂r V (r)) = 0 . . 2 2mr2 (7. the time variation of J can be written. 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.5) In the case of an angular independent potential angular momentum J = m r × v is a constant of ˙ motion.

(r) = 0 . but can also trap particles in the latter space giving rise to strong scattering resonances (see Section ??).20) Since this equation is independent of the quantum number m we drop the index m on the radial wave function vE.10) − 1 2 ∂ r + 2m r r 2 2 ( + 1) + V (r) − E 2mr2 . (7. . i.11. φ) (7. Employing the kinetic energy operator in the form (7.22) which is the original potential 2 ( +1) V (r) with an added rotational barrier potential 2mr2 .m (r) = 0.m (r) and E .24) . φ) are the angular momentum eigenstates deﬁned in Section 5.23) This demonstrates that the function r vE. r(t) is determined the angular motion follows from (7.m (r) Y m (θ.20) in the form of the one-dimensional Schr¨dinger equation o 2 − where 2m 2 ∂r + Veﬀ (r) − E φE (r) = 0 .1: Radial Schr¨dinger Equation o 185 This is a diﬀerential equation which governs solely the radial coordinate.17) In analogy to the classical description one can derivem. Multiplying (7.11). (7.m (r) 1 2 J2 ∂r r + + V (r) − E 2m r 2mr2 the functional form ψE. This barrier. .m (r) = 0. . by integration of ˙ θ = J . using (7. in the present case..4.16) m 2mr2 Once.21) Veﬀ (r) φE (r) = = V (r) + r vE. (7. . equations (7.m ψE.e. ( + 1) 2mr2 .m vE. ∞[ governed by the eﬀective potential (7.m (r) describes the radial motion as a one-dimensional motion in the interval [0.9) one can write the stationary Schr¨dinger o equation (7.20) by −2mr/ 2 yields the so-called radial Schr¨dinger equation o 2 ∂r − ( + 1) − U (r) − κ2 r2 r vκ. . .18) = vE. for the wave function of a quantum mechanical particle a diﬀerential equation which governs solely the r-dependence.10). 7.12) are obeyed and one obtains for (7.19) where Y m (θ. (7. .m (r) 2 . One can write (7.7. mr2 (t) (7. (7. 2 (7.m (r) .13). can exclude particles from the space with small r values. − Adopting for ψE. together with the original potential.m . It can be solved by integration of 1 2 2 J2 r = ± ˙ E − V (r) − .22) (7.

for > 0 is not integrable.25) (7. (r) ∼ A r + B r− −1 +1 + B r− (7. Galindo and P. the term Br−( +1) would contribute |B|2 0 dr r−2 +1) (7. r → 0 (7. .m (r) ∼ √ B . New York. For = 0 the contribution of Br−( function is.182) for Y00 . 2nd Ed. (r) ∼ A r or vκ.29) Only the ﬁrst term is admissable. This follows for > 0 from consideration of the integral which measures the total particle density. II by A. accordingly. 4π|r| (7.33) A detailed discussion of the proper boundary conditions. according to a well-known result in Classical Electromagnetism3 .30) and. In this case the solution is governed my ( + 1) 2 r vκ.20) one needs to specify proper boundary conditions2 . using the expression (5.26) 2m 2 In case E < 0. at r = 0 is found in the excellent monographs Quantum Mechanics I. Pascual (Springer. 2 2 2m 2 ψE. For r → 0 one may assume that the term ( + 1)/r2 becomes larger than the potential U (r). ”Classical Electrodynamics. assumes the general form r vκ. Berlin.28) (7. 1975). .20) the index E by the equivalent index κ.31) to the complete wave which. κ assumes real values. (r) (7. see. Jackson (John Wiley. The radial part of this integral is ∞ 0 2 dr r2 vκ.” by J.32) The total kinetic energy resulting from this contribution is. (7. (r) = 0 . 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).27) ∂r − r2 and.m (r) ∼ √ 4π B δ(x1 )δ(x2 )δ(x3 ) .D. in particular. We replaced in (7. for example.186 where we deﬁned U (r) κ2 = = − − 2m 2 Spherically Symmetric Potentials V (r) E (7. Boundary Conditions In order to solve (7. ψE. . hence.

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.50).98) by ∞ eik3 x3 = =0 b j (kr) P (cos θ) . Since the spherical harmonics Y 0 (θ.s.178) are given in terms of Legendre polynomials P (cos θ) one can replace the expansion in (7.e.100) Deﬁning x = cos θ.106).102) −1 d −1 +1 −1 dx −1 (x2 − 1) .m a mj (kr) Y m (θ. One can expand the former solution in terms of solutions (7.101) 2 ! ∂x −1 −1 Integration by parts yields +1 dx eizx P (x) −1 = 1 2 ! − 1 2 ! eizx d −1 2 (x − 1) dx −1 dx d izx e dx +1 (7. namely.150) one obtains +1 +1 1 ∂ dx eizx P (x) = dx eizx (x2 − 1) . This holds for −1 and one can conclude +1 consecutive integrations by part d izx e dx (7.s. In this case the expansion on the r.98) The l. 2 +1 (7. the surface term ∼ [· · ·]+1 vanishes. −1 .179) yield from (7. φ) for non-vanishing m has a non-trivial φ-dependence as described by (5. z = kr. according to (5.45) and one given by (7.h. in case of a free particle moving along the x3 -axis one expands eik3 x3 = .99) We want to determine the expansion coeﬃcients b .104) dx eizx P (x) −1 = (−1) 2 ! (iz) = 2 ! +1 dx (x2 − 1) −1 +1 dx (1 − x2 ) eizx .43) has two solutions. φ) .194 Generating Function of Spherical Bessel Functions Spherically Symmetric Potentials The stationary Schr¨dinger equation of free particles (7. of (7. One can show d −1 2 (x − 1) ∼ (x2 − 1) × polynomial in x dx −1 (7.98) does not involve any non-vanishing m-values since Y m (θ.44. φ). one given o by (7. (7. hence. i. and using the Rodrigues formula for Legendre polynomials (5. 7.. can be written exp(ikr cos θ). (7. (7. the wave function does not depend on φ.99) +1 d cos θ eikr cos θ P (cos θ) = b j (kr) −1 2 . For example.50). The orthogonality properties (5.103) and.

of (7.114) .2: Free Particle Comparision with (7.111) Employing (7.109) results in the integral representation of j (z) j (z) = (z) 2 +1 ! +1 dx (1 − x2 ) eizx . as the generating function of the spherical Bessel functions. 2 +1 −1 dx (1 − x2 )ν− 2 eizx . −1 (7.112) We want to consider the expression Gν (z) = aν z ν fν (z) where we deﬁne fν (z) = C (7.110) One refers to the l.113) dt (1 − t2 )ν− 2 eizt .107) 1 2 (2 )! 1 · 3 · 5 · · · (2 + 1) 2 + 1 1 · 2 · 3 · 4 · · · (2 − 1) · 2 (7.105) and (7.s.106) yields ﬁnally b = i (2 + 1) or. z i or i (2 + 1) z (2 )! 2 2 2 +1 2 ! [1 · 3 · 5 · · · (2 − 1)] 1 (7. 1 (7. in particular.109) eikr cos θ = =0 i (2 + 1) j (kr) P (cos θ) . −1 (7.f. for the leading power x [c.7. (7.69)] b z 2 (iz) = 1 · 3 · 5 · · · (2 + 1) 2 + 1 2 ! +1 dx (1 − x2 ) .h.106) Employing (5.117) one can write the r.h.s.105) must hold for all powers of z.h. after insertion into (7. Integral Representation of Bessel Functions Combining (7. (7.108) where the last factor is equal to unity.105) This expression allows one to determine the expansion coeﬃcients b . The identity (7. ∞ (7.s. using ν = Jν (z) = √ 1 πΓ(ν + 1 ) 2 z 2 ν + 1. Comparision with the l.96) one can express this.99). 1 (7. −1 (7.100) gives b j (kr) 2 (iz) = 2 +1 2 ! +1 195 dx (1 − x2 ) eizx .

93).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 . 7.93) for arbitrary ν. (7. t2 .116) fν (z) = − or fν (z) = − From this we can conclude fν (z) + 2ν + 1 fν (z) + fν (z) = 0 . 2. 3. it holds for the derivatives of Gν (z) Gν (z) G (z) = = ν aν z ν fν (z) + aν z ν fν (z) z 2ν ν(ν − 1) aν z ν fν (z) + aν z ν fν (z) + aν z ν fν (z) 2 z z (7. for properly chosen endpoints t1 .123) Insertion of these identities into Bessel’s equation leads to a diﬀerential equation for fν (z) which is identical to (7.121) (7.119) We note that equations (7.124) . z (7. In fact. We can now demonstrate that Gν (z) deﬁned in (7.121).113) satisﬁes (7.4: Prove (7. (7.114.93) as long as the integration path in (7.2. 1 j = 1. t2 of the integration paths C.120) such that we can conclude that Gν (z) for proper integration paths is a solution of (7.120) z 2ν + 1 dt (1 − t2 )ν− 2 eizt + C C 1 1 t2 t1 = 0 (7.116) In case that the endpoints of the integration path C satisfy (1 − t2 )ν+ 2 eizt one can write (7.117). To prove this we note fν (z) = i C dt (1 − t2 )ν− 2 t eizt .118) z 2ν + 1 fν (z) + fν (z) . 1 (7. C 1 (7.196 Spherically Symmetric Potentials Here C is an integration path in the complex plane with endpoints t1 .113) obeys the Bessel equation (7. 4. We consider now the functions u(j) (z) = √ 1 πΓ(ν + 1 ) 2 z 2 ν Cj dx (1 − x2 )ν− 2 eizx . Gν (z).122) (7. obeys Bessel’s equation (7.117) dt (1 − t2 )ν− 2 t2 eizt 1 (7. Exercise 7.120) imply also the property (2ν + 1) fν (z) + z fν+1 (z) = 0 .

134) and (7.124) is analytical in the part of the complex plane surrounded by the path C we conclude 4 u(j) (z) ≡ 0 j=1 (7. for Rez > 0 and ν ∈ R obey (7. 1 (7.130) holds Jν (z) = (1) 1 2 1 (1) (2) Hν (z) + Hν (z) .129) The integrand in (7. Accordingly.124) vanishes along the whole path C3 and.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 . 1 (7.134) Similarly.112) shows u(1) (z) = Jν (z).126) −1 ≤ s ≤ 1 (7.131) According to (7. Comparision with (7.132) For Hν (z) one derives.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. (7. u(3) (z) ≡ 0. the integral expression (1) Hν (z) 1 1 1 1 (7. therefore.127) 0 ≤ s < ∞ (7.117) and.130) We note that the endpoints of the integration paths C2 and C4 .7. (7. using t = 1 + is. 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 .128) C = C1 ∪ C2 ∪ C3 ∪ C4 is a closed path. hence. . one can state Jν (z) = − u(2) (z) + u(4) (z) . dt = i ds. we introduce the so-called Hankel functions (1) Hν (z) = −2 u(2) (z) . u(2) (z) and u(2) (z) are both solutions of Bessel’s equation (7.135) are known as the Poisson integrals of the Bessel functions. one can derive (2) Hν (z) = e −i(z−πν/2−π/4) ∞ 0 √ ds [s(1 − is/2)]ν− 2 e−zs .93).125) 0 ≤ s < ∞ (7.135) (7. Since the integrand in (7. (7. Following convention. (2) Hν (z) = −2 u(4) (z) .

One obtains W (z) = − 4i . Hν (z) as well as H−ν (z) are solutions of this equation.140) Bessel Functions with Negative Index Since ν enters the Bessel equation (7.135) which read for ν = + 1 2 H where f The binomial formula yields f (1) 2 (1) 2 1 +2 (1. hence. We employ for this purpose the Poisson integrals (7..136) (1) 2 (z) = 0 ds [s(s ± is/2)] e−zs .137) (z) = r=0 r ± i 2 r 0 ∞ ds s +r e−zs . the Wronskian W (g.140). πz (7. For the Wronskian connected with the Bessel equation (7. h(z) to be linearly independent. i. The general solution of (7.198 Asymptotic Behaviour of Bessel Functions We want to obtain now expansions of the Hankel functions H Spherically Symmetric Potentials the expansions converge fast asymptotically.93) holds the identity 1 W = − W z (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. converge fast for |z| → ∞.139) (z) = 2 ±i[z − ( e πz +1) π ] 2 r=0 ( + r)! r!( − r)! i ± 2z r (7.e.93) only as ν 2 .141) (1) (1) the derivation of which follows the derivation on page 192 for the Wronskian of the radial Schr¨dinger equation.143) . For such solutions g(z). h) must be a non-vanishing function.134) and (7. z (7. we obtain H (1) 2 1 +2 (1) 2 (z) = r=0 ( + r)! ! r!( − r)! ± i 2 r 1 z +r+1 (7.141) is o W (z) = − In case of g(z) = H (1) (z) +1 2 c . 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. (7. (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.

140). According to (7. We conclude 2 H−( Simarly.149) We want to establish also the relationship between h (z) and the irregular spherical Bessel function n (z) deﬁned in (7. (7. 2z 2 (1. 7. (7.149.97) one deﬁnes the spherical Hankel functions h (1.132.2) H + 1 (z) . H (1. in fact. 7. (7.2) (7. 7.53). 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 (7.153) .147) (z) = π (1. 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| → ∞ .150) According to (7.151) Equations (7. 7.147) this can be written n (z) = 1 2i h (1) (z) − h (2) (z) .96.7.144) The expansion coeﬃcients A.152) (7. 7. B can be obtained from the asymptotic expansion (7.90.148) holds for the o regular spherical Bessel function j (z) = 1 (1) (2) [ h (z) + h (z) ] . 2 (1.2) (7. (7. From (7.146. linearly independent.2) (z) +1 2 199 are.e. one can show H−( Spherical Hankel Functions In analogy to equations (7.151) are equivalent to h h (1) (2) (z) (z) = = j (z) + i n (z) j (z) − i n (z) .2) (2) +1) 2 (1) +1) 2 (z) = i (−1) H (1) (z) +1 2 .97. (7.132) follows n (z) = 1 2 π (−1) 2z +1 H−( (1) 1 +2) (z) + H−( (2) +1) 2 (z) . (7.145) This equation can hold only for B = 0 and A = exp[i( + 1 )π]. H−ν (z) = A H (1) (1) (z) +1 2 One can then expand + BH (2) (z) +1 2 ..97). (7.2: Free Particle i.146) (z) = −i (−1) H (2) 1 (z) +2 .

157) i q (z) 2z = i Im r=0 i 2z i 2z r = [ −1/2] ( +2r+1)! r=0 (2r+1)!( −2r−1)! −1 r 4z 2 0 ≥ 1 = 0 (7.162) The leading terms in these expansions. are j (z) n (z) = = (7.156) ( + r)! r!( − r)! ( + r)! r!( − r)! i 2z r [ /2] = r=0 ( + 2r)! (2r)!( − 2r)! −1 4z 2 r (7. z (7. is h (1) 2 (z) = ( i) z +1 e±iz .159) (7.164) . Hence. From (7.200 Asymptotic Behaviour of Spherical Bessel Functions Spherically Symmetric Potentials We want to derive now the asymptotic behaviour of the spherical Bessel functions h and n (z). ne ll(z) = Im[h (1) (z)] . 2 2z z (7.140.161) (7. it follows z ∈ R : Using p (z) = Re r=0 j (z) = Re[h (1) (z)] .148) one obtains readily h (1) 2 (1. at large |z|. are complex conjugates. as given by (7.154) The leading term in this expansion.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) . at large |z|.155) (z) and To determine j (z) and n (z) we note that for z ∈ R the spherical Hankel functions h (2) h (z). 7. (7.160) Employing cos[z − ( + 1) π ] . = sin(z − π/2) and sin[z − ( + 1) π ] . j (z) (z) = ( i) z +1 e±iz r=0 ( + r)! r!( − r)! ± i 2z r (7.2) (z). = −cos(z − π/2) results in 2 2 the alternative expressions j (z) n (z) = = sin(z − π/2) cos(z − π/2) p (z) + q (z) z 2z 2 cos(z − π/2) sin(z − π/2) q (z) − p (z) .158) one can derive then from (7. (1) (7.148).140) and (7.163) (7. 2z 2 z sin(z − π/2) z cos(z − π/2) − .

7.169) (7.2: Free Particle Expressions for the Spherical bessel Functions j (z) and n (z) 201 The identities (7. j (z) and n (z). One can combine (7. For this purpose we need to demonstrate only that the (1.170) Recursion Formulas of Spherical Bessel Functions The spherical Bessel functions obey the recursion relationships g g +1 (z) +1 (z) = = (1.148) and express g (z) = h (1. 1 (7.174) We want to prove these relationships.114) we can write g (z) = a z f 2 ! + 1 (z) 2 .2) (z) = −2 π 1 √ 2z πΓ(ν + 1 ) 2 z 2 ν C2. (7.173) A (z) = z ( +1) z2 +2 z (7.151) follows then that the relatiosnhips hold also for g (z) = j (z) and g (z) = n (z). From the linearity of the relationships (7.165) (7.131.171–7.7. 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 (7.2) To demonstrate that (7.166) (7. 7.175) Using Γ( + 1) = !. 1.172) where g (z) is either of the functions h obtain the recursion relationship g g (z). deﬁning a = −1 and employing fν (z) as deﬁned in (7.171.176) .167) (7.2) relationships hold for g (z) = h (z).162) allow one to provide explicit expressions for j (z) and n (z). One obtains for = 0.168) (7.172) to +1 (z) +1 (z) = A (z) 1− g (z) g (z) −1 (7.174) and from (7.124.161.4 dx (1 − x2 )ν− 2 eizx . 7. 7. (1.149. 7.2) g (z) − g (z) z 2 +1 g (z) − g z (7.171) −1 (z) (7.171) holds for g (z) = h (z) we employ (7.

173) is equivalent to (7. Exercise 7..177) (z) = − z f 2 +2 + 3 (z) 2 . n0 (z) using (7.182).173.58) it holds o 2 g (z) = − g (z) + z ( + 1) g (z) − g (z) . n2 (z) from n0 (z).171). z2 (7. In order to prove (7. in fact. To prove (7.172) we diﬀerentiate (7.180) yields g +1 (z) = g (z) − ( + 2) g (z) + z2 +2 g (z) . j0 (z) using (7. is equivalent to (7.176) g (z) = z g (z) − g +1 (z) +1 2 Spherically Symmetric Potentials z g (z) + a z f 2 ! +1 2 (z) .180) Since g (z) is a solution of the radial Schr¨dinger equation (7. (7.173.174) we start from (7. and (b) n1 (z).181) Using this identity to replace the second derivative in (7. 7. together with (7.166). . Exercise 7.174) to determine (a) j1 (z).172).171).2. 7.173).179) g (z) = − z2 g (z) + g (z) − g +1 (z) z .179) from which follows (7. The ﬁrst component of (7.171).178) (7.6: Employ the recursion relationship (7.165).202 The derivative of this expression is g (z) = Employing (7.172). f yields. (7. z (7. (7. The second component of (7. i.e.121).2.179) leads to (7.5: Provide a detailed derivation of (7. j2 (z) from j0 (z).182) Replacing all ﬁrst derivatives employing (7.

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

(8. t) and A(r. but leave the ﬁelds unaltered: A(r. t) and B(r. As is well known. Scalar and Vector Potential Setting B(r. The following substitutions. Equation (8.5) where p is the momentum of the particle and ro (t) it’s position at time t. t) (8. t) = 0 = 0. The reader is referred to the well-known textbook ”Classical Electrodynamics”. t) in (8. the relationship between ﬁelds and potentials is not unique. t) + ∂t A(r. t) + χ(r. t) Interaction of Radiation with Matter = 4 π ρ(r. t) × B(r. t) + ∂t A(r. t) −→ 1 A(r. 2nd Edition. t) (8. t) · B(r. called gauge transformations.3) reads then × E(r. t) where V (r.1) the term qδ(r − ro (t)) and to the current density ˙ J(r. t) . t) − ∂t E(r.6) for some vector-valued function A(r. solves implicitly (8. t) is a scalar function.204 Maxwell equations1 · E(r. t] (8. t) .25). (8.1) (8. t) − ∂t χ(r. We assume so-called Gaussian units. called the scalar potential.4) Lorentz Force A classical particle with charge q moving in the electromagnetic ﬁeld experiences the so-called Lorentz force q[E(r.2) the term q r˙o δ(r − ro (t)). t). t) which is solved by E(r. From this follows E(r. t) in (8. . Jackson (John Wiley & Sons. t] + v × B[ro (t). t) − ∂t A(r.8) = 0 (8. contributes to the charge density ρ(r. New York.3) (8. t) −→ V (r. t) = × A(r. t)] and. t) = − V (r. D.4). in turn. called the vector potential. The particle. accordingly. obeys the equation of motion d p = q dt E[ro (t). the two homogeneous Maxwell equations hold × E(r.11) V (r. t) = 4 π J(r. t) through the scalar and vector potentials V (r.2) In addition. t) + ∂t B(r. by J. In the non-relativistic limit holds p ≈ mr and (8. t) = − V (r. alter the potentials. t) + v × B(r. t). t) .7) Gauge Transformations We have expressed now the electric and magnetic ﬁelds E(r. (8.10) (8.5) above agrees with the equation of motion as given in (1.9) (8. 1975) for a discussion of these and other conventional units.

16) and the vector potential are uncoupled. t) Using · ∂t A(r. t) = Ω∞ for r ∈ ∂Ω∞ (8. (8. t) (8.16) This is the potential commonly employed in quantum mechanical calculations for the description of Coulomb interactions between charged particles. such that the latter describes the electromagnetic radiation. (8. t) in (8. t) + ∂t V (r. The vector potential A(r. In fact. t) can be obtained employing (8.20) .6) and (8. t) − ∂t A(r. (8. t) . (8. t) and the current due to moving net charges. the second inhomogeneous Maxwell equation. equation (8.19) couples the vector potential A(r. t) − 2 × A(r. t) = − 4 π ρ(r. In fact. One would prefer a description in which the Coulomb potential (8.18) together with condition (8. Using the expressions (8. be achieved. the gauge freedom allows us to impose on the vector potential A(r.2).17) A(r. t) the condition · A(r. t).13) · A(r. this potential results from inserting (8.12) leads us to 2 2 A(r. 4π (8. and the former the Coulomb interactions in the unperturbed bound particle system. t) = 0 the solution is given by the Coulomb integral V (r. t) = 1 ∂t V (r.1) · − V (r.14) In case of the boundary condition V (r. t) together with (8.12) The corresponding gauge is referred to as the Coulomb gauge.19) Unfortunately. a name which is due to the form of the resulting scalar potential V (r.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. For this purpose we examine the oﬀending term ∂t V (r. t) = · A(r. t) = ∂t = 4 π ρ(r. in fact. t) .15) d3 r ρ(r . t) − ∂t V (r. and a component due to the remaining currents. t) = 0 . t) + ∂t A(r. t) . Such description can.9) into (8. t) |r − r | (8. t = 4 π J(r. (8. t).19) and deﬁne J (r. t) = − 4 π J(r.9) for the ﬁelds results in × The identity × × A(r. t) .12) yields then the Poisson equation 2 V (r. t) . t) and V (r. t) − ∂t A(r.8.

t) .22) 1 ∂t 4π · J (r. t) = 0 (8. The deﬁnitions of J and Jt applied to (8. t) be the total current of the system under investigation and let Jt = J − J . t) = 0 .23) This continuity equation identiﬁes J (r. no divergence-free currents. V (r. t) = − 4 π Jt (r. t) + ∂t ρ(r.21) ∂t and the Poisson equation (8. t) contributes an electrical ﬁeld component as well as the total magnetic ﬁeld Et (r. for example.30) . t) − ∂t A(r. t) holds.25) Because of properties (8. hence. exist in the space considered. (8. (8. t) = − V (r.12) and the Coulomb potential (8. For J also holds the continuity equation · J(r. t) These ﬁelds are obviously divergence -free (e.24) (8.21) and (8.g. t) . t) = 0).26) This equation does not couple anymore scalar and vector potentials. V (r. A(r. t) = 0 . t) = 0 and from this follows · Jt (r.19) yield 2 2 A(r. t) = 0 . t) = 0). such current densities exist. 8. The vector potential determined through (8. the name longitudinal electric ﬁeld. t). t) + ∂t ρ(r. In this case (8.26) 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) = × A(r. i. t) (8. t) contributes solely an electric ﬁeld component E (r.e. yet a current. (8. respectively. hence.29) · Et (r. t) (8.206 For the curl of J holds × J (r. = − ∂t A(r. t) Bt (r.14). (8.28) (8.27) which is obviously curl-free ( × E (r.2 Planar Electromagnetic Waves The current density Jt describes ring-type currents in the space under consideration. t) = or = 2 Interaction of Radiation with Matter (8.. Let J(r. we want to assume that no ring-type currents. t) = − ∂t ρ(r. using ∂t · J (r. the name transverse ﬁelds. Presently.26) and (8. For the divergence of J (r. t) − ∂t A(r.16) yield ﬁnally the electric and magnetic ﬁelds. t) as the current due to the time-dependence of the charge distribution ρ(r. in a ring-shaped antenna which exhibits no net charge.

39) 8π The deﬁnition incoming waves and outgoing waves is rationalized below in the discussion following Eq. 2 . hence. t) = ± ω2 ˆ |Ao |2 k sin2 ( k · r − ωt ) 4π (8. the constant k is referred to as the wave vector. is orthogonal to k.31). (8. A complete set of solutions is given by the so-called plane waves A(r.12) yields u·k = 0. t) = ±i ω A(r. Here the “-” sign corresponds to so-called incoming waves and the “+” sign to outgoing waves2 . t) = Ao u exp i(k · r ˆ where the dispersion relationship |k| = ω (8.34) (8.35). Note that in the units chosen the velocity of light is c = 1.38. Time average over one period 2π/ω yields S(r. are Et (r. according to (8. t) Bt (r.29).34.158).33) ωt) (8. The latter vector describes the direction of propagation of the energy ﬂux connected with the plane wave electromagnetic radiation. t) = i k × A(r.35) are complex-valued quantities.34. t) in (8. there exist two linearly ˆ u independent orientations for u corresponding to two independent planes of polarization. t) and Bt (r.31) and the resulting ﬁelds (8. accordingly.38) ˆ In this expression for the energy ﬂux one can interprete k as the propagation velocity (note c = 1) and. The Coulomb gauge condition (8. (8. obviously. 8. This ﬂux is given by 1 S(r. 8. ω2 = |Ao |2 (8. 8π (8. ˆ We want to characterize now the radiation ﬁeld connected with the plane wave solutions (8. The corresponding electric and magnetic ﬁelds.35) The vector potential in (8. see also the comment below Eqs. at each point r and moment t.31. Et (r. In applying the potential and ﬁelds to physical observables and processes we will only employ the real parts. t) = Re Et (r.35) one obtains S(r. t) × Re B(r. t) = ± ω2 ˆ |Ao |2 k .31) u is a unit vector (|ˆ| = 1) which. 8. (8. 8.2: Planar Electromagnetic Waves 207 which describes electromagnetic ﬁelds in vacuum.36) 4π Using the identity a × (b × c) = b (a · c) − c (a · b) and (8. 8. 8. ˆ (8. Obviously.39). are orthogonal to each other and are both orthogonal to the wave vector k. t) . t) . (8.28.32.8.32) holds.37) ˆ ˆ where k is the unit vector k = k/|k|.34. 8.

38) implies that for incoming waves. t) 2m 2 + qV (r) . t) . These photons each carry the energy ω. t) = H Ψ(r. t).43) Here the ﬁelds are deﬁned through Eqs. 8. A ‘poor man’s’ quantization of the ﬁeld is possible at this point by expressing the energy density (8. we will neglect the respective terms in the Hamiltonian (8. the energy of the plane wave is transported in the direction of −k. We are left with a classical Hamiltonian function which has an obvious quantum mechanical analogue ˆ H = ˆ p − q A(r. deﬁned below Eqs. (8.39) and (8. a ˆ speciﬁc k and a speciﬁc u. (8. The integrals express the integration over the energy density of the ﬁelds.40) = V It should be pointed out that Nω represents the number of photons for a speciﬁc frequency ω.31). ˆ (8. (8.27.41) ωV Inserting this into (8. 8.31. A correct description of the electromagnetic ﬁeld requires that the ﬁeld be quantized. |k| = ω . We assume that the energy content of the ﬁelds is not altered signiﬁcantly in the processes described and.208 Interaction of Radiation with Matter as the energy density.45) .39) through the density of photons connected with the planar waves (8. t) is real 1 and that Et (r.32). t). i (8.8. u · k = 0 .43).29) together with the potentials (8. whereas in the case of outgoing waves the energy is transported in the direction of k. (8. t) 2m + 1 8π Ω∞ 2 H = + qV (r) 1 16π d3 r |Et |2 + |Bt |2 Ω∞ d3 r E 2 + .3 Hamilton Operator The classical Hamiltonian for a particle of charge q in a scalar and vector potential V (r) and A(r. hence. Comparision of (8. If we consider a volume V with a number of photons Nω the energy density is obviously Nω ω .42) 8. t) ˆ replacing the classical momentum p by the diﬀerential operator p = of the particle is then described by the Schr¨dinger equation o ˆ i ∂t Ψ(r.44) . The sign in (8. Bt (r. Note that E (r.16. The wave function Ψ(r. is p − q A(r.28. (8. t) are complex leading to the diﬀerence of a factor 2 in the energy densities of the lontitudinal and transverse components of the ﬁelds.31). respectively. 8.31) allows one ﬁnally to state for the planar wave vector potential A(r.40) allows one to express then the ﬁeld ˆ amplitudes 8πNω Ao = . t) = 8πNω u exp i(k · r − ωt) ˆ ωV . (8.

t)/ Ψ(r. (8. not to phenomenological interactions like the molecular van der Waals interaction.t)/ should not aﬀect a system and that.t)/ constitutes a unitary transformation of a scalar quantity. hence.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. In the classical case such question is mute since the gauge transformations do not alter the ﬁelds and.49) (8. Obviously. in the Schr¨dinger equation o 2 ˆ p − qA i ∂t Ψ(r.10..10. t) = 2m where ((· · ·)) denotes derivatives in (( χ)) and ((∂t χ)) which are conﬁned to the function χ(r. however. t) = + qV eiqχ(r.46). does not change expectation values which contain the probability densities3 .10.47. (8. t) = eiqχ(r. (8. t) enter whereas in the classical equations of motion ˙ m¨ = q E(r. t) . The idea just stated can be generalized by noting that multiplication by a phase factor eiqχ(r. 8. hence.t)/ Ψ(r. 8. An important conceptual step of modern physics has been to turn the derivation given around and to state that introduction of a local phase factor eiqχ(r. t) r (8.48) implies that the gauge transformation (8.t)/ [ i ∂t − q((∂t χ)) ] Ψ(r. an element of the group U(1).e.11) to (8. t) inside the double brackets.8.g. 8. t) (8. t) are necessary to compensate terms which arise through the phase factor.48) 2m For this purpose one notes i ∂t eiqχ(r.t)/ .45) leads to the Schr¨dinger equation o 2 ˆ p − q A − q(( χ)) + qV − q((∂t χ)) Ψ(r.47) i ∂t Ψ(r. This leads to the question in how far the gauge transformations (8. t) = + qV Ψ(r. t) ˆ = eiqχ(r. t) and A(r.50) The equivalence of (8. Applying the gauge transformations (8. accordingly.. Elementary constituents of matter which are governed by other symmetry groups. 8. have no eﬀect on the motion of the particle described by (8. 8.46) the ﬁelds enter. t) by a local and time-dependent phase factor eiqχ(r.t)/ p + q(( χ)) Ψ(r. t) .t)/ Ψ(r.44.11) aﬀect the quantum mechanical description. such phase factor does not change the probability density |Ψ(r. t)|2 and. by the group 3 The eﬀect on other expectation values is not discussed here. It should be noted. t) + q r × B(r.47) is equivalent to 2 ˆ p − qA i ∂t eiqχ(r.51) 2m the potentials A(r. that this principle applies only to fundamental interactions. t) and V (r. One can show that (8. t) .11) of the potentials is equivalent to multiplying the wave function Ψ(r. . t) ˆ p eiqχ(r. e. i.t)/ Ψ(r.

54) ˆ2 p q ˆ q2 2 − p·A + A + qV .52) This expression vanishes in the present case since since ˆ ˆ p · Af = A · pf and.59) .57) satisﬁes · A = 0 and.56). Using Cartesian coordinates this yields 2 H = − 2me 2 2 2 ∂1 + ∂2 + ∂3 + 2 eBo e2 Bo 2 x1 ∂2 + x i me 2me 1 (8. The Hamiltonian (8. k2 .55). likewise can demand the existence of ﬁelds which compensate local transformations described by eiσ·χ(r. 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.58).55) 8. consequently.56) due to the gauge freedom. · A = 0 [cf. Accordingly. t). x2 .t) where σ is the vector of Pauli matrices. t) = Bo . (8. For this purpose we use the wave function in the form Ψ(E. aﬀects the complexity of the mathematical derivation of the wave functions. 2.12)]. (8. 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.g.45) with Hamiltonian (8. holds (8. (8. the generators of SU(2). The stationary homogeneous magnetic ﬁeld B(r.210 Interaction of Radiation with Matter SU(2). one can employ the Hamiltonian (8.. The vector potential (8.55). the so-called Landau gauge ˆ associates the vector potential AL (r) = Bo x1 e2 ˆ (8. The choice of a vector potential aﬀects the form of the wave functions describing the eigenstates and. (8. thereby.53) ˆ ˆ p·A + A·p + q2 2 A + qV 2m (8. In case of a homogeneous potential pointing in the x3 -direction. H = 2m m 2m (8. can be described by various vector potentials. t) ≡ 0.57) with a homogeneous magnetic ﬁeld Bo . The resulting ﬁelds are called Yang-Mills ﬁelds. x1 . for Bo = Bo e3 in (8. We want to describe the stationary states corresponding to the Hamiltonian (8. In this o case holds V (r. therefore.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. k3 . x3 ) = exp(ik2 x2 + ik3 x3 ) φE (x1 ) .58) where ∂j = (∂/∂xj ). j = 1. e.

therefore. according to (8. We want to employ . Without aﬀecting the energy one can form wave packets in terms of the solutions (8.56) through the vector potential A(r) = 1 Bo × r . the states are degenerate in the quantum number k2 describing displacement along the x2 coordinate. (8.62) is that of a displaced (by x1o ) harmonic oscillator with o 2 k 2 /2m ) energies. 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. From this observation one can immediately conclude that the wave shifted (by e 3 function of the system. 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 wave function (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 ) .66) Obviously.59).60) Completing the square 2 2 e2 Bo 2 eBo k2 e2 Bo x1 + x1 = 2me me 2me x + k2 eBo 2 − 2 k2 2 2me (8.57). Unfortunately.67) . Solution for Symmetric Gauge The solution obtained above has the advantage that the derivation is comparatively simple. k2 . the familiar harmonic oscillator quantum number.64) is the classical Larmor frequency (c = 1). the so-called symmetric gauge which expresses the homogeneous potential (8.63) this induces a spread of the wave function in the x1 direction. k3 . (8.65) where we replaced the parameter E by the integer n.65) which localize the electrons. like the corresponding gauge (8. is not symmetric in the x1 .63) (8. according to (8. The stationary Schr¨dinger equation (8. x1 . x2 .59). However.65). The energies corresponding to these states are E(n.and x2 -coordinates. k2 . k3 ) = ω(n + 2 k2 1 3 ) + .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 ) . 2 (8. is Ψ(n. 2 2me (8.62) x1o = and where ω = k2 eBo eBo me (8.8.

ˆ (8. using × (uf ) = −u × f +f ˆ ˆ Bo · p × r f = Bo · r × p f .72) allows one to interpret µ = − e L 2me (8. the magnetic moment (8. Vmag = −µclass · Bo (8. The theory of classical electromagnetism predicts an analogue energy contribution .76) e 2me class class · Bo .70) ˆ ˆ p · A f = − Bo · p × r f . (8. namely.71).52) becomes ˆ2 e e2 p + Bo · L + H = 2me 2me 8me Of particular interest is the contribution Vmag = e L · Bo 2me Bo × r 2 .74) is in the present case 1 µclass = − e r v e3 .77) Comparision with (8. In this case the current density measures −e v oriented tangentially to the ring.73.69) × u and × r = 0.71) (8. Accordingly. y-plane with constant velocity v on a ring of radius r.212 Interaction of Radiation with Matter One can readily verify that this vector potential satisﬁes the condition (8.78) . the Hamiltonian (8.12) for the Coulomb gauge. (8.68) × u yields.67) one can write ˆ p·A = Using · (u × v) = − u · function f (r) ×v + v· 2i · Bo × r . for any (8. namely. an electron moving in the x. in the present case of constant Bo .72) to Hamiltonian (8.75) 2 The latter can be related to the angular momentum µclass = − and. The latter can be rewritten. (8. 8. Vmag = e 2me class = r me v e3 of the electron ˆ (8. ˆ Identifying r × p with the angular momentum operator L. (8. For the vector potential (8. accordingly.74) We consider a simple case to relate (8.72) and (8.74).73) where µclass is the magnetic moment connected with a current density j µclass = 1 2 r × j(r) dr (8.

described by the operator S. x2 ) obey then ˆ Ho ψ(x1 . the so-called gyromagnetic ratio of the electron. x2 . x3 ) = E ΨE (x1 . To obtain these eigenstates we introduce the customary dimensionless variables of the harmonic oscillator Xj = (8. x3 ) = exp(ik3 x3 ) ψ(x1 . E = E − (8. The functions ψ(x1 . 10 that the spin of the electron.80) (8. the Hamiltonian (8.71) is ˆ H = 2 ˆ2 p e2 Bo + 2me 8me x2 + x2 + 1 2 eBo L3 . x2 ) where ˆ Ho = − 2 2 2 ∂1 + ∂2 (8.86) 1 2 2me 1 2 ˆ as well as of the angular momentum operator L3 . x2 ) = E ψ(x1 . A derivation of his property and the value of g. In this case. we separate the variable x1 .81) (8.72) with an associated magnetic moment − g 2me S where g ≈ 2. x2 . requires a Lorentz-invariant quantum mechanical description as provided in Sect. 10. e gives rise to an energy contribution (8. for Bo = Bo e3 .84) .8. 2 . j = 1.56) pointing in the x3 -direction the symmetric gauge (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 . i.82) 1 ω (x1 ∂2 − x2 ∂1 ) i 2me + 1 me ω 2 x2 + x2 1 2 2 2 k2 3 + (8. Accordingly. x3 ) .79) To obtain the stationary states. likewise. (8.88) me ω xj . 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. the solutions of ˆ H ΨE (x1 .e. x2 from x3 setting ΨE (x1 . We will demonstrate in Sect.85) The Hamiltonian (8.67) yields a more symmetric solution which decays to zero along both the ±x1 . ˆ i.87) .and the ±x2 -direction. x2 ) ..e. For a magnetic ﬁeld (8.4: Electron in Homogenous Magnetic Field 213 as the quantum mechanical magnetic moment operator for the electron (charge −e). (8. 2me (8. x2 .83) describes two identical oscillators along the x1 -and x2 -directions which ˆ are coupled through the angular momentum operator L3 .83) (8. 2me We have used here the expression for the angular momentum operator ˆ L3 = ( /i)(x1 ∂2 − x2 ∂1 ) .

x2 ) = ω 2j + 2m + 1 Ψ (j. .as well as for the x2 -oscillator. x1 . .. i. x2 ) (8. Such eigenstates can be expressed. −j + 1. (8. 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 .95) (j) ˆ We want to choose the coeﬃcients αmm such that (8. 1. 2 (8.90) a† a2 − a† a1 1 2 .89) . in fact. .e. (8. it holds a† a1 + a† a2 + 1 1 1 2 Ψ (j. obeyed. +j .214 Employing the creation and annihilation operators 1 a† = √ j 2 and the identity ˆ ω L3 = which can readily be proven. m. m . m = −j.e. x2 ) = m=−j αmm Ψ(j. j = 1. x2 ) . m . x1 . the total vibrational energy being ω(2j + 1). 1 2 since one phonon is annihilated and one created. aj = √ 2 Interaction of Radiation with Matter Xj + ∂ ∂Xj .94) Since this state is a linear combination of states which all have vibrational energy (2j + 1) ω. x1 ..92) are not eigenstates of L3 . through a combination of states j Ψ (j. x2 ) . such that 1 i a† a2 − a† a1 1 2 Ψ (j. (8. x1 . m . x1 . m as follows: j = 0. x1 . 2 2 . x2 ) = 2m Ψ (j. x1 . If this property is. (8. . x2 ) = ( 2j + 1 ) Ψ (j. . m . however. (8. x1 .91) We note that the operator a† a2 − a† a1 leaves the total number of vibrational quanta invariant. (8. x2 ) = a† 1 j−m (j + m)! (j − m)! Ψ(0.96) ˆ holds.97) . x1 . 1 3 . (j) (8. x2 ) (8. 0.94) is also an eigenstate of L3 . therefore.94) is an eigenstate of Ho ˆ Ho Ψ (j. 0. We. m. . x2 ) is the wave function for the state with zero vibrational quanta for the x1 . .94) is an eigenstate of the vibrational Hamiltonian. m .93) ˆ The states (8. (8.92) where Ψ(0. . attempt to express eigenstates in terms of vibrational wave functions a† 1 j+m Ψ(j. (8. m . m .92) represents a state with j + m quanta in the x1 -oscillator and j − m quanta in the x2 -oscillator. i. x1 . x2 ) . In order to cover all posible vibrational quantum numbers one needs to choose j. x1 .

x2 ). 5. The external radiation ﬁeld is accounted for by the vector potential A(r.10. x2 ) = Dmm ( .11.98) then the states Ψ(j.5. m) in Sect. the second rotation a factor dmm ( π ). 5. the 2 2 latter representing the Wigner rotation matrix of Sect. x1 .97). (8. m . is in the notation of the present section 1 ˆ a† a2 − a† a1 . t) = 1/4πr conﬁning an electron with energy E < 0 to move in the well known orbitals of the hydrogen atom. This implies that ˆ we can obtain eigenstates of L3 by rotation of the states Ψ(j. which corresponds there to the angular momentum in the x2 –direction.8. If we identify 2 a† .55) conﬁnes the particle to stationary bound states. a1 .309) yields ﬁnally Ψ (j.5. cf..5. According to Sect.100) J2 = 1 2 2i 1 ˆ i.9.5. b† . an example is the Coulomb potential V (r. x1 . Eq. 8. b− + − notation in Sects. t) introduced above. x1 .288)] 1 ˆ J3 = 2 a† a1 − a† a2 1 2 (8.11 (8.11.9.9. i.10.5: Time-Dependent Perturbation Theory (j) 215 In order to obtain coeﬃcients αmm we can proﬁtably employ the construction of angular momentum states in terms of spin– 1 states as presented in Sects. 5. 0) Ψ(j. a transformation moving the x3 –axis into the x2 – 2 axis. 5. m .10. the eigenstates of (8. The required rotation must transform the x3 –axis into the x2 –axis. x1 . thus. We assume that the scalar potential V in (8. The connection with the present problem arises due to the fact that the operator J2 in Sect. x2 ) (8.102) We have identiﬁed. x1 . The ﬁrst rotation contributes a factor exp(−im π ). x2 ) . is identical to the operator L3 introduced in (8.99) with eigenvalue m.e. except for a factor 2 . b+ .55). the states are eigenstates of the operator [we use for the operator the notation of Sect. According to the derivation given there. m.10. a2 1 2 present notation ←→ b† .101) 2 2 where Dmm ( π .e. 5.11 such transformation is provided through π π (j) Ψ (j. 5. m. x2 ) deﬁned in (8.(5. 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 .31). In the simplest case the radiation ﬁeld consists of a single planar electromagnetic wave described through the potential (8.92) correspond to the eigenstates |Ψ(j. (8. . m. Using the explicit form of the Wigner rotation matrix as given in (5. 5.84) above.. 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.83) and conﬁrmed the eigenvalues stated in (8. Other radiation ﬁelds can . a† .5 Time-Dependent Perturbation Theory We want to consider now a quantum system involving a charged particle in a bound state perturbed by an external radiation ﬁeld described through the Hamiltonian (8. m. x1 . π .

can be considered a weak perturbation.. the term ∼ p · A(r.e. Estimate of the Magnitude of VS We want to demonstrate now that the interaction VS (t). The states |n are thought to form a complete. t) + A (r.103) (8. However. one can estimate that the perturbation.105) as follows4 . (8. is much 2 (r. . i. t) is given by (8.42). . orthonormal basis. This result will allow us to neglect the larger than the second term. For an electron charge q = −e and an electron mass m = me one can provide the estimate for the ﬁrst term of (8.e. using (8.. the term ∼ A second term in (8. The Hamiltonian of the particle in the radiation ﬁeld is then described through the Hamiltonian H Ho VS = = = Ho + VS ˆ2 p + qV 2m q ˆ q2 2 p · A(r. t). In fact. We ﬁrst note. Here the so-called unperturbed system is governed by the Hamiltonian Ho with stationary states deﬁned through the eigenvalue problem Ho |n = n |n . (8.104) (8. as given in (8.107) 1 = 1 n=0 |n n| ..105) in further calculations and to expand the wave function in terms of powers of VS (t) in a perturbation calculation. 1. t). in this case.105). n = 0. i. 2 ..108) We assume for the sake of simplicity that the eigenstates of Ho can be labeled through integers. i. 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. is much smaller than the eigenvalue diﬀerences near ˆ typical atomic bound states.106) where we adopted the Dirac notation for the states of the quantum system.105) for the case of radiationinduced transitions in atomic systems. ωV (8.109) The reader should note that the estimates are very crude since we are establishing an order of magnitude estimate only.e. and that the ﬁrst term in (8. t) − m 2m (8. i.41) e ˆ e p2 p·A ∼ 2me me me 2me 4 1 2 8πNω .216 Interaction of Radiation with Matter be expanded through Fourier analysis in terms of such plane waves. .105) where A(r.e. this assumption can be waved as our results below will not depend on it. we discount the possibility of a continuum of eigenstates. we assume n|m = δnm and for the identity ∞ (8.

112) yields e2 A2 ∼ 10 eV · 10−8 = 10−7 eV . altogether. e ˆ p · A ∼ 10 eV · 10−4 = 10−3 eV . me (8. hence. with ao ≈ 0.5: 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.e. m (8.8. and as long as the photon densities Nω /V are small.. 2me (8. λ = 2πc/λ. (8. We can.118) . replace the perturbation (8. one obtains for (8. u) p · u exp i(k · r − ωt) ˆ ˆ ˆ kV . Nω = 1.109) using V = λ 4π 2 c2 /ω 2 e2 e ˆ p·A ∼ me 4πao 2 ao ω π λ me c2 1 2 (8. Using again (8.114) Employing for the second factor the estimate as stated in (8.115) (8. (8.41) one can state e2 e2 1 8πNω ω A2 ∼ 2me 2me ω 2 V For the same assumptions as above one obtains e2 e2 A2 ∼ · 2me 8πao ao 4 ω λ me c2 . Assuming a single photon. and me c2 ≈ A A 500 keV 2 ao ω ≈ 10−8 (8.ˆ u 4πNk α(k. one can neglect this term as long as the ﬁrst term gives non-vanishing contributions.105) due to a radiation ﬁeld by VS = − q ˆ p · A(r.110) where ao is the Bohr radius.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. Consequently. i.42) it holds VS ≈ e m k.111) For ω = 3 eV and a corresponding λ = 4000 ˚ one obtains.e.117) In case that such perturbation acts on an electron and is due to superpositions of planar waves described through the vector potential (8.105) for radiation ﬁelds can be considered a small perturbation.5 ˚.112) π λ me c2 and with e2 /ao ≈ 27 eV. Hence. t) . i. the ﬁrst term in (8..105). a volume V = λ3 where λ is the wave length corresponding to a plane wave with frequency ω. We want to estimate now the second term in (8.116) This term is obviously much smaller than the ﬁrst term.

108) as the eigenstates of the unperturbed Hamiltonian Ho .121) (8.119) In order to determine the wave function ΨS (t) we choose the so-called Dirac representation deﬁned through i |ΨS (t) = exp − Ho (t − to ) |ΨD (t) (8.218 Interaction of Radiation with Matter where we have replaced ω in (8. .127) i ∂t |ΨD (t) = VD (t) |ΨD (t) .125) (8. This state obeys the Schr¨dinger equation o i ∂t |ΨS (t) = [ Ho + VS (t) ] |ΨS (t) subject to the initial condition |ΨS (to ) = |0 .119) 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) . describes the eﬀect of a radiation ﬁeld on an electron system and which will be assumed below to describe radiative transitions. under ordinary circumstances.118) is the form of the perturbation which. |Ψ(to ) = |0 (8. Using i i ∂t exp − Ho (t − to ) and (8. n = 0 at some later time t ≥ to .120) (8. Multiplying the latter equation by the operator exp i (8. |n are deﬁned in (8. A factor α(k.124) (8.106–8. (8. u) has been added to describe eliptically or circularly polarized ˆ waves. For this purpose one needs to determine the state |ΨS (t) of the particle. . n = 1. 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 . The states |0 . Equation (8. the sum over u involves the two possible polarizations of planar ˆ electromagnetic waves. 2. The probability to ﬁnd the particle in the state |n at time t is then p0→n (t) = | n|ΨS (t) |2 .123) i = Ho exp − Ho (t − to ) (8. The sum runs over all possible k vectors and might actually be an integral.42) through k = |k| = ω. One seeks to predict the probability to observe the particle in one of the states |n .126) Ho (t − to ) yields ﬁnally . ..122) where |ΨD (to ) = |0 .

137) |ΨD (t) = (1) (0) (8. In order to determine |ΨD (t) described through (8.127.128).5: Time-Dependent Perturbation Theory where VD (t) = exp 219 i i Ho (t − to ) VS (t) exp − Ho (t − to ) . We will consider here only the two leading contributions to |ΨD (t) . . (8.8. = . .127) we assume the expansion ∞ |ΨD (t) = n=0 (n) |ΨD (t) (n) (8. Equations (8.130) = exp − i n (t − to ) − to )]| = 1.137) follows |ΨD (t) = |0 .132) where |ΨD (t) accounts for the contribution due to n-fold products of VD (t) to |ΨD (t) . Accord(n) ingly.136) together with the initial conditions |ψD (to ) = |0 0 for n = 0 for n = 1. we deﬁne |ΨD (t) through the evolution equations i ∂t |ΨD (t) i ∂t |ΨD (t) i (2) ∂t |ΨD (t) (1) (0) = = = . 8. (8.135) i ∂t |ΨD (t) (n) VD (t) |ΨD (n−1) (t) (8.137) can be solved recursively. . . n| (8. Due to the Hermitean property of the Hamiltonian Ho holds n|Ho = i n|exp − Ho (t − to ) from which we conclude. (8. 0 VD (t) |ΨD (t) (1) VD (t) |ΨD (t) (0) (8.132–8. .139) .128) We note that the transition probability (8.133) (8.137) One can readily verify that (8. 8.133. to (8.134) (8. . 8.134. 2.131) p0→n (t) = | n |ΨD (t) |2 . From (8. using |exp[− i n (t n (8.137) are consistent with (8. Employing this result one obtains for (8.138) 1 i t dt VD (t ) |0 .133–8.121) expressed in terms of ΨD (t) is i p0→n (t) = | n|exp − Ho (t − to ) |ΨD (t) |2 . . consequently.129) n| and.

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. . (8. however. yields for (8. The factor III in (8. according to (8.e. to → −∞ (8. II III (8. the direction of ˆ the propagation being determined by k = k/|k|.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. 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.137) |ΨD (t) = (2) Interaction of Radiation with Matter 1 i 2 t t dt to to dt VD (t ) VD (t ) |0 . 8.220 This result. λ → 0+ . 8..144) describes the propagation of the planar wave. A2 are deﬁned through (8.41). non-singular.142) combining an incoming and an incoming or outgoing wave. i.141) + . The resulting perturbation on an electron system.. in turn. In (8. such that the combined radiation ﬁeld is decribed by the vector potential A(r. For the sake of including the eﬀect of superpositions of plane waves we will assume.135. We want to apply now the perturbation expansion derived to such perturbations. This factor will serve mainly the purpose of keeping in the following derivation all mathematical quantities properly behaved. A factor exp(λt). + m=0 dt to to dt n|VD (t )|m m|VD (t ) |0 8. The coeﬃcients A1 .142)..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). (8. note that uj .118). 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. deﬁnes the direction of the E-ﬁeld of the ˆ radiation. λ → 0+ has been introduced which describes that at time to → −∞ the perturbation is turned on gradually.34.144) r is the position of the electron and p = ( /i) is the momentum operator of the electron. is VS = ˆ ˆ V1 exp(−iω1 t) + V2 exp( iω2 t) eλt . that two planar waves simulataneously interact with an electronic system.143) ˆ ˆ where V1 and V2 are time-independent operators deﬁned as e ˆ Vj = m 8πNj ωj V I ˆ ˆ p · uj eik·r .6 Perturbations due to Electromagnetic Radiation We had identiﬁed in Eq.

140. employing the perturbation (8.149) Using the deﬁnition of VD stated in (8. (8.6: Perturbations due to Electromagnetic Radiation 1st Order Processes 221 We employ now the perturbation (8.130) and (for m = 0) i i exp − Ho (t − to ) |m = exp − for (8. For the 1st order contribution to the transition amplitude n|ΨD (t) = (1) 1 i t dt n|VD (t ) |0 to (8. (8. using (8. 8.143) to the expansion (8. (8. (8.128) one obtains k|VD (t) | = k|VS (t)| exp i ( k − ) (8.145) we obtain then.145) n|ΨD (t) (1) m (t − to ) |m .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 .8.146) = 1 λ→0+ t→−∞ i lim lim × t dt exp to i ( n − o − i λ) t iω2 t × . (8.141). 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).150) and. According to (8.151) ˆ ˆ n|V1 |m m|V1 |0 exp ˆ ˆ + n|V2 |m m|V2 |0 exp i i n − − m − ω1 − i λ)t exp ω2 − i λ)t exp m o n m m o .128).148) We consider now the 2nd order contribution to the transition amplitude.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 .

155) λ→0+ ( o We are actually interested in the transition rate.153) (8. We assume ﬁrst that the ﬁrst order transi(1) tion amplitude n|ΨD (t) is non-zero. We also assume for the ﬁnal state n = 0 such that n|0 = 0 and p0→n (t) = | n|ΨD (t) |2 holds.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. i. provide now the transition probability p0→n (t) according to Eq. in which case one can expect that it is larger than the 2nd (2) order contribution n|ΨD (t) which we will neglect. For this rate holds . the time derivative of p0→n (t).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 (8.131).154) yields p0→n (t) = lim e2λt + + 2 Re ˆ n|V1 |0 |2 ( o + ω1 − n )2 + (λ )2 ˆ n|V2 |0 |2 ( o ± ω1 − n )2 + (λ )2 ˆ ˆ n|V1 |0 0|V2 |n exp i (± ω2 − ω1 ) t + ω1 − n + iλ ) ( o ± ω2 − n − iλ ) (8.148) and ∗ |z1 + z2 |2 = |z1 |2 + |z2 |2 + 2Re(z1 z2 ) (1) (8.e. Using (8.152). respectively.. (8.148) and (8.

the matrix elements can vary strongly for diﬀerent states |n of the system.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. ˆ (8. one should average the rate over many periods of the radiation. to cancel the third term in (8. describing either an incoming or an outgoing plane wave due to the vector potential A2 u2 exp i (k1 · r ˆ ω2 t) . Hence. is much shorter than could be resolved in any observation. For the resulting expression the limit limλ→0+ can be taken. The time average will be denoted by · · · t .142) ˆ ˆ The matrix elements n|V1 |0 and n|V2 |0 in (8.159) leads to ﬁnal states |n with energy n = o + ω1 . The result supports our deﬁnition of incoming and outgoing waves in (8. however.160) leads to ﬁnal states |n with energy n = o ± ω2 .156) such that the 1st order contributions of the two planar waves of the perturbation simply add. (8. of this expression describe the individual eﬀects of the two planar wave contributions of the perturbation (8.142–8. First. a property.8. due to the vector potential A1 u1 exp i (k1 · r − ω1 t) . any attempt to do so. 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. i. 8.157) one can conclude for the average transition rate k = d p0→n (t) dt t = 2π ˆ | n|V1 |0 |2 δ( n − n o − ω1 ) o (8.e.158).143. . In case the matrix elements are non-zero.158) play an essential role for the transition rates of radiative transitions. which is observed through the so-called spectral intensities of transitions |0 → |n .144).156) ( o The period of electromagnetic radiation absorbed by electronic systems in atoms is of the order 10−17 s..144).31) and (8.158) ˆ + | n|V2 |0 |2 δ( − ω2 ) Obviously. the two terms apearing on the rhs. Using lim x2 + 2 δ→0+ = π δ(x) (8. The δ-functions appearing in this expression reﬂect energy conservation: the incoming plane wave contribution of (8. The second contribution to (8. in fact. The result of such average is. due to the uncertainty relationship would introduce a considerable perturbation to the system.

e..k=1 j=k 2 2 2 2 ∗ zj zk (8.168) absorption of 2 photons ω1 . Time average · · · t of expression (8. (8.166) over many periods of the radiation yields exp[ i (±ω2 − ω1 )] t = 0 and.. We deﬁne ∞ z1 = m=0 ˆ ˆ n|V1 |m m|V1 |0 m − o − ω1 − i λ exp i × (8. i.163) × (8. similarly. hence. in analogy to (8.158). k = = d p0→n (t) dt 2π ∞ m=0 t 2 ˆ ˆ n|V1 |m m|V1 |0 m − o − ω1 − i λ δ( m − o − 2 ω1 ) (8.166) 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.g. i.e.164) (8. in Eqs.157) yields. |z1 + z2 + z3 |2 t = |z1 |2 + |z2 |2 + |z3 |2 (8.152). it holds p0→n (t) = | n|ΨD (t) |2 . e.. exp[ i (±ω2 − ω1 )]. as we did in the the case of 1st order transitions.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. (2) (8.158).167) Taking now the limit limλ→0+ and using (8.153) vanishes such that the leading contribution to the transition probability p0→n (t) arises from the 2nd order amplitude (8. ∞ − 2 ω1 − 2i λ)t n − o − 2 ω1 − 2i λ ( n o − z2 = m=0 ˆ ˆ n|V2 |m m|V2 |0 ω2 − i λ m− o exp i × (8.162) × and.224 2nd Order Radiative Transitions Interaction of Radiation with Matter We now consider situations where the ﬁrst order transition amplitude in (8.153–8.161) To determine the transition rate we proceed again.

describes processes which lead to ﬁnal states |n with energy n = o + 2 ω1 and.168) describe similar histories of the excitation process. The factors | · · · |2 in (8.e..169) pert. . accordingly. which is stationary and energy is conserved. describe the absorption/emission of two photons..e. i.e. each of energy ω1 .168) describes two intermediate histories. i. For this purpose we will employ the transition rate as given in Eq. Similarly. according to its δ-function factor. a second perturbation. 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.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. i. The remaining two contributions in (8. m = o + ω1 ) and the evolution of state |m is described by a factor 1/( m − o − ω1 −i λ). it must hold n = o + 2 ω1 .168) the interpretation is ∞ ˆ n|V1 |m m=0 1 m − o − ω1 − i λ virtually occupied state |m ˆ m|V1 |0 pert.e.168) describe the time sequence of the two photon absorption/ emission processes. promotes the system then to the state |n . the system is perturbed through absorption of a photon with energy ω1 from the initial state |0 into a state |m . Similarly. this state is only virtually excited. The expression sums over all possible virtually occupied states |m and takes the absolute value of this sum. there is no energy conservation necessary (in general. through absorption of a photon. interference between the contributions from all intermediate states |m can arise.158) which accounts for such transitions. the third term in (8. ﬁrst absorption of photon ω1 and then absorption/ emission of photon ω2 .. The ﬁrst term.. Most remarkably. (8.8. namely absorption/ emission ﬁrst of photon ω2 and then absorption of photon ω1 and. |m ← |0 (8. describe the absorption of two photons. accordingly. 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 second term describes the processes leading to ﬁnal states |n with energy n = o ± 2 ω2 and. i. |n ← |m . vice versa. In case of the ﬁrst term in (8. each of energy ω2 .

t) = 8πN ωV u exp ˆ ı (k · r − ωt) .m=0 (r. the vector potential actually assumed is A(r. according to (8. φ) a3 ao o 2p (8. in this range one can expand eik·r ≈ 1 + ik · r + . . φ) = 6 r −r/2ao e Y1m (θ. i. m denote the relevant quantum numbers) ψn=1. φ) a3 o 1s (8. and the integral is =1. φ) = 2 − 1 2 1 −r/ao e Y00 (θ. (8.178) These wave functions make signiﬁcant contributions to this integral only for r-values in the range r < 10 ao . ˆ (8. (8. (8. θ.144) VS = where e 2πN ˆ ˆ p · u e±ik·r .226 Absorption of a Plane Polarized Wave Interaction of Radiation with Matter We consider ﬁrst the case of absorption of a monochromatic..158). in case of a radiative transition from the 1s state of hydrogen to one of its three 2p states.143. λ → 0+ .2 = m ωV Only the ﬁrst term of (8.173) V1.174) We seek to evaluate the matrix element ˆ M = n| p eik·r |0 . θ.175) The matrix element involves a spatial integral over the electronic wave functions associated with states |n and |0 . For example.172) δ( n − o − ω) (8. the second term can be discounted in case of absorption.143) will contribute to the absorption process.179) . φ) (8. 8.170) We will employ only the real part of this potential.m (r. However.177) √ M = 6 0 ∞ 1 2π ia4 o r2 dr −1 dcosθ 0 ∗ dφ r e−r/2ao Y1m (θ. The absorption rate. 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 . .e. (8. . plane polarized wave described through the complex vector potential A(r. t) = 2πN ωV u exp ˆ ı (k · r − ωt) + 2πN ωV u exp ˆ ı (−k · r + ωt) .176) ψn=2.171) The perturbation on an atomic electron system is then according to (8. =0. φ) × × eik·r e−r/ao Y00 (θ. the wave functions are (n. to → −∞ (8.

We are now in a position to obtain an alternative expression for the matrix element (8.182) one ﬁnds [ r. V (r) = − .179) and other terms are never larger than 20π ao /λ.187) . pj ] = i δkj one obtains ˆ [ r.8.7: One-Photon Emission and Absorption 227 One can estimate that the absolute magnitude of the second term in (8.529 ˚ one concludes that in the signiﬁcant integration range in (8. in case of the hydrogen atom.k=1 i ˆ p m (8.182) one obtains M ≈ Insertion into (8.186) δ( n − o − ω) (8.184) and the commutation property [xk . is Ho = ˆ (p)2 e2 + V (r) . This simpliﬁcation results from the identity m ˆ p = [ r. Transition Dipole Moment A further simpliﬁcation of the matrix element (8.175) can then be ˆ achieved and the diﬀerential operator p = i replaced by by the simpler multiplicative operator r.k=1 i m 3 pk ek = ˆ j.104) and.181) and (8. (8. pk ] + ˆ ˆ k=1 1 2me 3 [ r.174) yields kabs = 4π 2 e2 N ω V u · n| ˆ |0 ˆ r 2 m( o − m n| [r. V (r) ] 2me =0 = Using r = 3 ˆ j=1 xj ej 1 2me 3 pk [ r.182).181) For the commutator in (8. Ho ] = i m 3 pk ej δjk = ˆ j.182) (8.185) from which follows (8. Using (8. ] + [ r. Ho ] i where Ho is the Hamiltonian given by (8.183) (8.178) holds eik·r ≈ A 1 1 + O( 50 ) such that one can approximate eik·r ≈ 1 . Ho ] = ˆ2 p [ r. 2me r (8.175). Using |k| = 2π/λ. the value of the wave length for the 1s → 2p transition 2π c λ = = 1216 ˚ A (8. pk ] pk ˆ k=1 (8. One refers to this approximation as the dipole approximation. Ho ] |0 = i i n) n|r |0 .180) ∆E2p−1s and ao = 0.

u2 . there are ony two linearly independent directions of u possible.. the direction of u1 is described by the ˆ ˆ ˆ ˆ angles ϑ1 = π/2. ρ) and ˆ u β = ∠(ˆ2 . one can express | u · n| r |0 |2 = |ρ|2 ( cos2 α + cos2 β ) = sin2 θ .228 Interaction of Radiation with Matter where we used the fact that due to the δ-function factor in (8.189) where the factor 1/ arose from the integral over the δ-function.31–8. 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. three-dimensional vector. u2 and k ˆ ˆ coordinate system. x2 . i. 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. in practical situations. Integrating and summing accordingly over all contributions as given by (8. ˆ u ˆ (8. and all polarizations of the radiation. x3 -axes of a right-handed cartesian vectors u1 .174) one can replace n − o by ω. and (2) the fact that strictly monochromatic radiation cannot be prepared such that any radiation source provides radiation with a frequency distribution.188) ˆ Here dk is the integral over all orientations of k. ϕ2 = π/2. For the two angles α = ∠(ˆ1 .191) (8. We have demonstrated in our derivation of the rate of one-photon processes (8. 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. say u1 and u2 .35). will actually be replaced by a distribution function which reﬂects (1) the ﬁnite life time of the states |n . a very hot oven. ρ) holds then u cosα = cosϑ1 cosϑ + sinϑ1 sinϑ cos(ϕ1 − ϕ) = sinϑcosϕ and cosβ = cosϑ2 cosϑ + sinϑ2 sinϑ cos(ϕ2 − ϕ) = sinϑsinϕ .190) (8. The unit ˆ ˆ ˆ ˆ can be chosen to point along the x1 . hence. The δ-function appearing in this expression. according to (8. ϕ1 = 0 and of u2 by ϑ2 = π/2. k frame is described by the angles ϑ. planar waves as deﬁned in (8. obtain the transition rate in the present case by adding the individual transition rates of all planar waves present in the oven. ˆ In order to carry out the integral dk we note that u describes the possible polarizations of the ˆ ˆ and u. k ˆ As a result. We can. ϕ. The direction of this vector in the u1 .33) are orthogonal to each other.158) above that in ﬁrst order the contributions of all components of the radiation ﬁeld add. |0 . directions.e.193) . Absorption of Thermal Radiation We want to assume now that the hydrogen atom is placed in an evironment which is suﬃciently hot. Accordingly.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.

172. For practical evaluations we provide an expression which eliminates the physical constants and allows one to determine numerical values readily..37 × 1019 1 ao | n|r|0 |2 × s λ λ2 .7: One-Photon Emission and Absorption 229 This geometrical average. i. in agreement with observations..173) contributes. The radiation ﬁeld is described. ..e. there is also a contribution to the emission rate which is proportional to Nω . (8. The transition dipole moment can vanish for many transitions between stationary states of a quantum system. Transition Dipole Moment The expression (8. i.194) combined are typically somewhat smaller than (1 ˚/1000 ˚)3 = A A 10−9 . The most intensely absorbing molecules are long. The quantum nature of radiation manifests itself in that the number of photons Nω msut be an integer.197) o (tot) (8.8. The value of | n| r |0 | determines the strength of an optical transition. in particular. Nω photons before absorption. . for atoms or symmetric molecules.e. The corresponding process is termed spontaneous emission.194) for the absorption rate shows that the essential property of a molecule which determines the absorption rate is the so-called transition dipole moment | n| r |0 |. ˆ In case of emission only the second term V2 exp(+iωt) in (8. by planar waves with vector potential (8. the calculation of the emission rate proceeds as in the case of absorption. However. the resulting total rate of emission bears a diﬀerent dependence on the number of photons present in the environment. 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. 2 . In case of a classical radiation ﬁeld one would expect that emission cannot occur. (8. 1.195) = Nω 1. however. However. as for the absorption process. Accordingly. the absorption rate is of the order of 109 s−1 or 1/nanosecond. . ﬁnally. linear molecules.171) and perturbation (8. Otherwise.37 × 1019 × 3 3 c3 3 ao λ3 s λ and kabs where λ = 2πc n − (8.196) The last two factors in (8. for Nω = 0. This poses. Nω = 0. that emission occurs even if no photon is present in the environment. However.189) to yield the total absorption rate kabs (tot) = Nω 4 e2 ω 3 | n|r|0 |2 3 c3 . This diﬀerence between emission and absorption is due to the quantum nature of the radiation ﬁeld. a problem in case of emission by quantum systems in complete darkness. can be inserted into (8. Emission of Radiation We now consider the rate of emission of a photon.173). For this purpose we use ω/c = 2π/λ and obtain 4 e2 ω 3 32π 3 e2 ao 1 ao = = 1. This dependence predicts.194) For absorption processes involving the electronic degrees of freedom of atoms and molecules this radiation rate is typicaly of the order of 109 s−1 . 8.

202) It follows.230 Interaction of Radiation with Matter which is termed induced emission since it can be induced through radiation provided. Nω + 1 (8. The number of atoms undergoing absorption in unit time are Nω ksp No and undergoing emission are (Nω +1)ksp Nn . the well-known Planck radiation formula. respectively.201) ksp = 3 c3 the rates of absorption and emission are Nω ksp and (Nω + 1)ksp .8 Two-Photon Processes In many important processes induced by interactions between radiation and matter two or more photons participate. For these numbers holds Nn / No = exp[−( n − o )/kB T ] (8.200).204) Nω . exp[− ω/kB T ] = This equation yields Nω = 1 .199) Planck’s Radiation Law The postulate of the Nω + 1 dependence of the rate of emission as given in (8. To demonstrate this property we apply the transition rates (8.198) (8.. In the following we discuss several examples.198) is consistent with Planck’s radiation law which reﬂects the (boson) quantum nature of the radiation ﬁeld. accordingly. . 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. in lasers.200) where kB is the Boltzmann constant. 8. = (8. using (8. Under stationary conditions the number of hydrogen atoms undergoing an absorption process |0 → |n must be the same as the number of atoms undergoing an emission process |n → |0 .e.198) to determine the stationary distribution of photons ω in an oven of temperature T .203) i. exp[ ω/kB T ] − 1 (8. e. Examples are radiative transitions in which two photons are absorbed or emitted or scattering of radiation by matter in which a photon is aborbed and another re-emitted. Deﬁning the rate of spontaneous emission 4 e2 ω 3 | n|r|0 |2 (8. We assume n − o = ω.g. it must hold Nω ksp N0 = (Nω + 1) ksp Nn (8. The total rate of emission.195) and (8.. Hence. respectively. Let No and Nn denote the number of atoms in state |0 and |n .

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.207) for the rate of 2-photon transitions shows that the transition |0 → |n becomes possible through intermediate states |m which become virtually excited through absorption of a single photon. The 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. (8. even if n| r |0 vanishes. however. ﬁnally. We assume that a planar wave with wave vector k1 and polarization u1 . If the sum in (8.208) has been prepared.207) does not converge rapidly. 8.104) with stationary states |n deﬁned through (8.182) yields. in case ˆ of absorption only V1 contributes. Employing the dipole approximation (8.187). (8. k = 2π e2 2πNω m2 ωV e × δ( m 2 ∞ m=0 n|ˆ · p eik·r |m m|ˆ · p eik·r |0 u ˆ u ˆ m − o − ω1 − i λ 2 × (8. a transition between the states |0 and |n may be possible.207) does not provide a suitable avenue of computing the rates of 2-photon transitions. One obtains.181) and using (8.207) Expression (8. (8. as described through the vector potential ˆ A(r. In this case one needs to choose the energy of the photons to obey n = o + 2 ω. then expression (8. but then requires the absorption of two photons.106). 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. The electron absorbs the radiation and emits immediately a second photon. induce single photon absorption processes as described by the transition rate Eq. In applying (8.207) one is.206) − o − 2 ω) . under ordinary circumstances. dropping the index 1 characterizing the radiation. The resulting rate of the transition depends on 2 Nω . however. The transition requires that the transition dipole moment n| r |0 does not vanish for two states |0 and |n .168) where the ﬁrst term describes the relevant contribution. . which is not necessarily the case. The intense radiation ﬁelds of lasers allow one to increase transition rates to levels which can readily be observed in the laboratory. t) = Ao1 u1 cos(k1 · r − ω1 t) . However.8.205) The respective radiative transition is of 2nd order as described by the transition rate (8.173).8: Two-Photon Processes Two-Photon Absorption 231 The interaction of electrons with radiation. faced with the dilemma of having to sum over all intermediate states |m of the system. 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) . ˆ (8.172.

208) and the emission of the radiation in (8. In the present case we consider only scattering processes which absorb radiation corresponding to the vector potential (8.210). hence. 8. The second term in (8.e. 8.208). This expansion yields. i.211) is o2 A− = o2 8π (N2 + 1) ω1 V (8. The relevant terms of the perturbation (8.211) are given by VS (t) = e ˆ p · A+ u1 exp[i(k1 · r − ω1 t)] + A− u2 exp[−i(k2 · r − ω2 t)] o1 ˆ o2 ˆ 2me contributes in 2nd order + e2 4me A+ A− u1 · u2 exp{i[(k1 − k2 ) · r − (ω1 − ω2 ) t]} ˆ o1 o2 ˆ contributes in 1st order The eﬀect of the perturbation on the state of the electronic system is as stated in the perturbation expansion (8.211) accounts for the emitted wave and. We know already from our description in Section 8.214) .212) where N1 /V is the density of photons for the wave described by (8. Let us assume for the present that the emitted photon has a wave vector ˆ k2 with cartesian components sinθ2 cosφ2 k2 = k2 sinθ2 sinφ2 (8.210) is accounted for by the following contributions of (8.210) The vector potential which describes both incoming wave and outgoing wave is a superposition of the potentials in (8. according to the ˆ description of emission processes on page 229. the amplititude A− deﬁned in (8.105). ˆ The perturbation which arises due to the vector potential (8. the wave characterized through k1 . however.209) cosθ2 where the value of k2 has been ﬁxed.208.232 Interaction of Radiation with Matter We assume that the experimental set-up also includes a polarizer which selects only radiation with a certain polarization u2 .210) A(r.213) where N2 /V is the density of photons characterized through k2 .105) using the vector potantial (8.208.208) and emit radiation corresponding to the vector potential (8. for the components of the wave (8. t) = Ao2 u2 cos(k2 · r − ω2 t) . in the present case. u2 .141). o1 ˆ o2 ˆ (8. u1 .211) The ﬁrst term describes the absorption of a photon and.. later we will allow the quantum system to select appropriate values. The vector potential describing the emitted plane wave is then A(r.211) is stated in Eq. the amplitude A+ is given by o1 A+ = o1 8πN1 ω1 V (8.210). t) = A+ u1 exp[ i (k1 · r − ω1 t) ] + A− u2 exp[ i (k2 · r − ω2 t) ] .6 above that the absorption of the radiation in (8. ˆ (8. (8.

212.215) 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 (8.215) 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.e. evaluating the time integrals in (8. Accordingly.217) 2 Here ro denotes the classical electron radius ro = e2 = 2. Inserting also the values (8. one needs to integrate the rate as given by (8. only these two terms contribute to the scattering amplitude.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 . i.215) correspond to scattering processes in which the radiation ﬁeld ‘looses’ a photon ω1 and ‘gains’ a photon ω2 . Following closely the procedures adopted in evaluating the rates of 1st order and 2nd order radiative transitions on page 222–225. 8.8 · 10−15 m . i.8. one needs to carry out the integration 2 V(2π)−3 k2 dk2 · · ·.216) over all available modes of the ﬁeld..181) in stating this result.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.. Only the second (1st order) and the third (2nd order) terms in (8.213) for the amplitudes A+ and A− results o1 o2 in the Kramers-Heisenberg formula for the scattering rate k = N1 c 2 ω2 r (N2 + 1) dΩ2 u1 · u2 n|0 ˆ ˆ V o ω1 1 n|ˆ2 · p |m m|ˆ1 · p |0 u ˆ u ˆ n|ˆ1 · p |m m|ˆ2 · p |0 u ˆ u ˆ − + me m m − o + ω2 m − o − ω1 (8. me c2 (8.e. Hence.218) .

. in fact. .224) = m 1 − o ( m ω ( ω)2 + + . 2 . .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. one assumes the even stronger condition ω1 << | o − m| .e.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 ) (8.234 Interaction of Radiation with Matter The factor N1 c/V can be interpreted as the ﬂux of incoming photons. for all states |m of the electronic system (8. ω1 = ω2 .217) to the scattering cross section deﬁned through dσ = It holds then dσ = 2 ro rate of photons arriving in the the solid angle element dΩ2 ﬂux of incoming photons ω2 (N2 + 1) dΩ2 u1 · u2 n|0 ˆ ˆ ω1 1 n|ˆ1 · p |m m|ˆ2 · p |0 u ˆ u ˆ me m m (8. − o )2 ( m − o )3 (8.219) (8. Rayleigh scattering is deﬁned as the limit in which the wave length of the scattered radiation is so long that none of the quantum states of the electronic system can be excited.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. consequently.228) S (0) = m S (0) = 2 m .223) m Condition (8. Rayleigh Scattering We turn ﬁrst to an example of so-called elastic scattering.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. (8.. a process in which the electronic state remains unaltered after the scattering. Accordingly.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 + .221) suggests to expand S( ω) S( ω) = S(0) + S (0) ω + Using 1 − o ± ω one can readily determine m (8. i.221) Using |n = |0 and. it follows 2 dσ = ro (N2 + 1) dΩ2 |ˆ1 · u2 − S( ω) |2 u ˆ (8.. one can relate (8.

For this purpose we apply (8.k=1 (ˆ1 )j (ˆ2 )k δjk = u1 · u2 u u ˆ ˆ (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 ) .226). 8.236) 2 × m 0|ˆ∗ ·r |m m|ˆ2 ·r |0 u1 u 0|ˆ∗ ·r |m m|ˆ1 ·r |0 u2 u + m − o m − o We have applied here a modiﬁcation which arises in case of complex polarization vectors u which ˆ describe circular and elliptical polarizaed light.226) into S(0) = 1 i ( 0|ˆ1 ·r |m m|ˆ2 · p |0 − 0|ˆ2 · p |m m|ˆ1 ·r |0 ) u u ˆ u ˆ u m . (8. It explains.227) vanishes.235) We can now combine eqs.235) and obtain the leading contribution to the expression (8.229) (8. for example.230) According to (8.108) for the identity operator. Expression (8.224.234) where we used for the second identity the fact that u1 ·r and u2 ·r commute.232. honey bees.231) (ˆ1 )j (ˆ2 )k 0|[xj .108) this is 1 0|ˆ1 ·r u2 · p − u2 · p u1 ·r |0 .222). this term cancels the u1 · u2 term in (8.228) we obtain ˆ ˆ ˆ ˆ S (0) = 2me 2 m 0|ˆ2 ·r |m m|ˆ1 ·r |0 u u 0|ˆ1 ·r |m m|ˆ2 ·r |0 u u + m − o m − o (8. pk ] = i δjk yields ﬁnally ˆ S(0) = 1 S(0) = i 3 3 (8.229) both for the u1 · p and the to u2 · p terms in (8.228) provides then the ﬁrst non-vanishing contribution to the scattering cross section (8. 8. u1 ·r ] |0 = 0 u ˆ (8. i.e.108) yields S (0) = me 2 0| [ˆ2 ·r.232) Obviously.233) m Employing again (8.8. as a compass.182) 0|ˆ1 · p |m u ˆ 1 = 0|ˆ1 ·r |m u me ( m − o ) i This transforms (8.k=1 j.236) is of great practical importance. 8.234. ˆ ˆ We want to prove now that expression (8. pk ]|0 = u u ˆ j.222) of the cross section for Rayleigh scattering dσ = 2 ro m2 ω 4 (N2 + 1) dΩ2 × e (8. m|ˆ1 · p |0 u ˆ 1 = − m|ˆ1 ·r |0 u me ( m − o ) i (8. ˆ We want to simplify ﬁrst (8. u ˆ ˆ ˆ ˆˆ i The commutator property [xj . the blue color of the sky and the polarization pattern in the sky which serves many animals.. u u u u (8. Using again (8. For this purpose we replace p using (8.222).182) for p and the expression (8.8: Two-Photon Processes 235 ˆ These three expressions can be simpliﬁed using the expression (8. . ˆ ˆ S (0) given in (8.

239) We will show below that this expression decribes the non-relativistic limit of Compton scattering. though. u2 = u2 ˆ ˆ ˆ ˆ (1) (2) (2) (1) (1) (2) (1) (2) (2) (1) needs to be (8. we choose for the polarization of the emitted radiation two perpendicular directions u2 ˆ (2) and u2 which are also orthogonal to the direction of k2 .. however. (8. the emitted u ˆ radiation is decribed by the wave vector sinθ2 cosφ2 k2 = k1 sinθ2 sinφ2 (8.e.. u ˆ 1 ao . for all states |m of the electronic system .240).and the x2 -axes 1 0 (1) (2) u1 = 0 . The ﬁrst choice is ˆ sinφ2 k2 × k1 (1) u2 = ˆ = −cosφ2 (8.240) cosθ2 We choose for the polarization of the incoming radiation the directions along the x1 . u2 = u2 ˆ ˆ ˆ ˆ for u1 = u1 . 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 .220) 2 dσ = ro (N2 + 1) dΩ2 |ˆ1 · u2 |2 . u2 = u2 ˆ ˆ ˆ ˆ cos2 φ2 cos2 θ2 sin2 φ2 for u1 = u1 .244) . To evaluate |ˆ1 · u2 |2 we assume that k1 is oriented along the x3 -axis and.237) The resulting scattering is called Thomson scattering. We want to assume.236 Thomson Scattering Interaction of Radiation with Matter We consider again elastic scattering.241) 0 0 Similarly. i. hence. ao Bohr radius .238) (8.242) |k2 × k1 | 0 where the second identity follows readily from k1 = e3 and from (8. u2 = u2 ˆ ˆ ˆ ˆ (8. (8. that the dipole approximation is still valid which restricts the applicability of the following derivation to k1 >> One obtains immediately from (8.243) cos2 θ2 cos2 φ2 for u1 = u1 . u1 = 1 ˆ ˆ (8. |n = |0 and ω1 = ω2 = ω in (8.220). assume now that the scattered radiation has very short wave length such that ω >> | o − m| .

)n . the scattering is termed Brillouin scattering.247) ( 1 − cosθ2 ) (8.. Since the scattering is inelastic. In the case of such scattering an electron system absorbs and re-emits radiation without ending up in the initial state. in case of a diatomic molecule |n = |φ(elect. but most often involves other degrees of freedom. In the case that the scattering excites other than electronic degrees of freedom. 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.. φ(vibr.e. Raman Scattering and Brillouin Scattering We now consider ineleastic scattering described by the Kramers-Heisenberg formula. (8. deﬁned in (8. ˆ In case that one measures the scattered radiation irrespective of its polarization.e. The Compton scattering cross section which is derived from a model which treats photons and electrons as colliding relativistic particles is dσ tot = dσ tot where (rel) = 2 ro (N2 + 1) 2 ω2 ω1 2 ω1 ω2 + − sin2 θ2 ω2 ω1 dΩ2 (8. the resulting scattering cross section is 2 ro (N2 + 1) ( 1 + cos2 θ2 ) dΩ2 . molecular vibrations or crystal vibrational modes. if energy is released.246).8. In case that the nuclear degrees of freedom excited absorb very little energy. The energy deﬁcit is used to excite the system. or in case of translational motion of molecules in liquids. for c → ∞ the Compton scattering cross section (8. one speaks of Stokes scattering.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 .)n . one speaks of anti-Stokes scattering. The radiation scattered at right angles is even completely polarized (1) along the u2 -direction. the states |n etc. i. the scattered radiation is polarized to some degree. For electronic systems in molecules or crystals the degrees of freedom excited are nuclear motions.220) represent actually electronic as well as nuclear motions.249) dσ = ro (N2 + 1) ω1 −1 −1 ω2 − ω1 = . the ﬁrst term in (8.247. as in the case of excitations of accustical modes of crystals. Such scattering is called Raman scattering. If energy is absorbed by the system.220) vanishes and one obtains in case of Raman scattering ω2 2 dΩ2 | u2 · R · u1 |2 ˆ ˆ (8.245) 1 cos2 θ for u = u(2) ˆ ˆ 2 2 2 2 The result implies that even though the incident radiation is unpolarized. e. The excitation can be electronic.248) me c2 One can readily show that in the non-relativistic limit..246) 2 This expression is the non-relativistic limit of the cross section of Compton scattering.247) becomes identical with the Thomson scattering cross section (8.g. i. 8.

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

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

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

. xk ) is symmetric in its two arguments. In the latter case all even permutations are represented by ‘1’ and all odd permutations by ‘-1’. usually a very large number. k = 1... the Hamiltonian for permuted indices P HP −1 is identical to H. ) are the same for all particles and for all pairs of particles. all particles being governed by the same interactions.8) |Ψ (9. it is a postulate of quantum mechanics that for any description of many–particle systems the permutation operators must be diagonal. In fact.5) implies that a permutation eﬀectively changes the indices of the particles) and. or the corresponding group of two elements 1. 1 together with multiplication. ∀P ∈ SN . particles with half–integer spin values. i. 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 . (9. i.9. S where S denotes the number of available single particle states.. P ] = 0 . In fact. a diagonal representation can only berealized in a simpler group. From the latter results [H.11) . The terms of the Hamiltonian will be discussed further below. (9. {1} (trivial) and {1. i. i.6) which describes one–particle interactions and two–particle interactions of the system.e. Since permutations do not necessarily commute. Fock-Space Our derivation will start by placing each particle in a set of S orthonormal single particle states x|k = φk (x). In fact.10) We will construct now wave functions which exhibit the appropriate properties. The single particle states are assumed to be orthonormal. the representation is either isomorphic to the group composed of only the ‘1’. for the scalar product holds j|k = δj k .e. −1. G(xj .. provide a representation of the group structure represented by the multiplication table of even and odd permutations given above. The ﬁrst case applies to bosons. .e. in turn. particles with integer spin values. (9. xk ) j. 2.1: 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. From this property follows that H is independent of any speciﬁc assignment of indices to the particles (note that (9. hence.-1}. Presently. i.e.9) = 1 for P even −1 for P odd (9. the second group to fermions. it is essential that the functions F ( ) and G( .7) This property.k=1 (9.e. Obviously. both groups. the Hamiltonian must be of the type N H = j=1 1 F (xj ) + 2 N G(xj . . implies that the permutation operators can be chosen in a diagonal representation.

. λN ) = j=1 δλj λj . |λ2 . 2. The numbers nj . .12). . N x1 . . nS ).e. ... the so-called Fock-states. . 9. λ2 . S} of the type (9.12) where the second factor represents the scalar product between spin states. . λN ) ⊂ SN is the set of all permutations involving only particles j with diﬀerent λj . each particle j into a speciﬁc state |λj . . appear severalfold in (9. The reader is well advised to pause and grasp the deﬁnition of the nj . which do not yet obey the symmetries (9.15) with λk = j. . . It holds N ΨFock (λ1 .9). λN ) do not exhibit the symmetries dictated by (9.. We ﬁrst consider orthonormal many–particle wave functions. hence. . |λN . therefore. .13) These states form an orthonormal basis of N -particle states. . λ2 . the states |ΨFock (λ1 . . . . .14) Obviously. .. k) ∈ Λ(λ1 . λj ∈ {1. T (j. . λN ) only if λj = λk .e. . . . . . .. λN ) = j=1 φλj (xj ) . . k.15) is that the sum over permutations does not involve permutations among indices j. . . λ) in which two particles occupy the same single– particle state |λ does not allow transposition of particles.15) completely and.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. λN )|ΨFock (λ1 . the nj specify how often a single particle state |j is occupied. . with λj = λk = . . . . . the reason is that such transposition duplicates the state which would. . The Fock states represent N particles which are placed into S states |λ1 . . n2 . x2 . . xN |ΨFock (λ1 . nS ) = S j=1 (9. . x2 . Λ(λ1 . are essential. . .λN ) on the rhs. . 9. xN |Ψ Fock (λ1 . . x2 .9). . (9. This detail is connected with the fact that a two–particle state of the type ΨFock (λ. . (9.. λ2 .15) (nj !) N! 1 2 P ∈Λ(λ1 . A wave function which obeys the boson symmetry (9. . n2 . e. 9. . .15). xN |ΨB (n1 . . . .g. i. characterize the wave function (9. is a particular choice of placing N particles consistent with the occupation numbers (n1 .λN ) P 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.λN ) Here the states (λ1 . . . . . . In a second step we form linear combinations of Fock states with the desired symmetries. i.8. The integers nj are equal to the number of λk in (9.8) is x1 . λ2 . . . An important detail of the deﬁnition (9. .11. . referred to as the occupation numbers. λ2 . . . λ2 .8. . .

nS ) = φλ1 (x1 ) φλ1 (x2 ) . . . .15) describe.1: (a) Show that the states (9. . .15) holds (9. . nS ) and N = (n1 . . for two sets of occupation numbers N = (n1 . . nS ) holds S Ψ (N )|Ψ (N ) = j=1 B B δnj nj .16) Exercise 9.λN ) on the rhs. (9. φλN (xN ) Here.15) obey the boson symmetry (9. n2 . .λN ) (9.. correspond to a particular choice of placing N particles consistent with the occupation numbers (n1 . n2 . .e. . . . .16).e. . x2 . xN |ΨFock (λ1 . . . . . Wave Functions with Fermion Symmetry in the Occupation Number Representation One can construct similarly a wave function which obeys the fermion symmetry property (9. . . nS ) holds S ΨF (N )|ΨF (N ) = j=1 δnj nj . (b) Show that for the states (9. .8). .. . for two sets of occupation numbers N = (n1 .19) φλ1 (xN ) φλ2 (xN ) . xN |ΨF (n1 . Using the identity governing the determinant of N × N –matrices A N det(A) = P ∈SN P j=1 Aj P (j) (9. . . .. a molecule or a crystal. . . The important property holds that nj can only assume the two values nj = 0 or nj = 1.17) Here the states (λ1 . for example. (9. φλN (x1 ) φλN (x2 ) . xN |ΨF (n1 . . .17) describe. i.. . . . the integers nj denote the occupancy of the single–particle state |j . . . . The fermion states (9. n2 .20) . φλ2 (x1 ) φλ2 (x2 ) .. the photons of the electromagnetic ﬁeld or the phonons in a crystal. . . . nS ). . . electrons in an atom. . . . nS ) = √1 N! P ∈SN P P x1 . . λ2 . . n2 . nS ) and N = (n1 .9. . These states form an orthonormal basis.9). . √1 N! (9. n2 . . . for example. . . . . . For any value nj > 1 two or more of the columns of the Slater matrix are identical and the wave function vanishes. ..18) this wave function can also be expressed through the so-called Slater determinant x1 .1: Bosons and Fermions 243 The wave functions (9. i.1. n2 . . Such wave function is x1 . x2 . These states form an orthonormal basis. λ2 . n2 . x2 .

.23) (9. . . .17) obey the fermion symmetry (9. . n ) are normalΨ 1 j 1 j S S ized according to (9. These operators allow one to construct and manipulate many–particle wave functions while preserving permutation symmetry. To prove that (9.16) one obtains (Aj )N = ΨB (N )| aj |ΨB (N ) = √ S N nj δnj nj −1 k=1 k=j δnk nk . . . . . .24) [aj . . n . .22) The operators aj and a† obey the commutation properties j [aj . since both the N -particle wave function B (n .1. . . nj − 1. . . n − 1. . . [a† . . . . nS ) = j nj + 1 ΨB (n1 . . are the so-called operators of 2nd quantization. .2: (a) Show that the states (9. . .2 Operators of 2nd Quantization An important tool to describe mathematically systems of many bosons or fermions. . Employing the superindices N and N introduced above and using the orthonormality property (9. .9). (b) Show that the states (9. nj + 1. 9. ak ] = 0 . n ) and the N − 1-particle wave function ΨB (n . guaranteeing the correct permutation properties of the many–particle states.17) obey the orthonormality property (9. nS ). . nS ) nj ≥ 1 0 nj = 0 √ . .21) we consider the matrix Aj corresponding to aj in the basis of many–particle states (9. Creation and Annihilation Operators for Bosons We consider the operator aj deﬁned through aj ΨB (n1 . . Exercise 9. . . a† ] = 0 j k (9. . nS ) = nj ΨB (n1 . . . . . nj .244 Many–Particle Systems The representation of wave functions (9. . . nj .15). For the adjoint operator a† holds j a† ΨB (n1 .21) √ The factor nj appears here on the rhs. . n2 . . nS ) . . . . . (9.17) is commonly referred to as the occupation number representation since the wave functions are uniquely speciﬁed by the set of occupation numbers N = (n1 . . (9. .15) and (9. .25) . a† ] = δj k . . . 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 . .15).22) follows from (9.20). . (9.

e. a† ] = 1 For j = k follows similarly 1. . .. . . . . . One refers to Nj as the occupation number operator.33) is called the particle number operator since.32) ˆ ˆ i. nS ) = j=1 nj |ΨB (n1 . 9. (9. The creation operators a† allow one to construct boson states ΨB (N ) from the vacuum state j |0 = |ΨB (n1 = 0. nk + 1 .26) From this result one can immediately conclude that (9. .34) = .9. nS ) . nS ) = |ΨB (N ) . . (9. i. . We will prove the commutation properties (9. . .27. The related operator S ˆ N = j=1 ˆ Nj (9. .28) (9. . nS = 0) as follows S (9.21. nS ) = j=1 a† j nj nj ! |0 . j (9. (9.24). .29) yield the commutation relationship (9. n2 = 0. the eigenvalues of Nj are the occupation numbers nj . n2 . . nS ) N |ΨB (n1 . the properties (9. . One can readily show using (9. . .21) and vice versa. n2 . n2 . . n2 .. 9. j j Since this holds for any |Ψ(N ) it follows [aj . nj − 1 .22) follows from (9. One obtains for its matrix elements j A† j N N 245 = ΨB (N )| a† |ΨB (N ) = (Aj )N N j S k=1 k=j nj δnj nj −1 δnk nk = nj + 1 δnj nj +1 S k=1 k=j δnk nk . . . It holds for j = k (aj a† − a† aj )|Ψ(N ) = (nj + 1 − nj )|Ψ(N ) = |Ψ(N ) .e. . obviously. .27) Equations (9.24). . j (aj a† − a† aj )|Ψ(N ) = k k ( nj (nk + 1) − (nk + 1)nj )|Ψ(n1 . . . for basis states |ΨB (n1 . S ˆ N |ΨB (n1 .31) This operator is diagonal in the occupation number representation.29) |ΨB (n1 . (9. .23) follow in an analogous way. .30) Of particular interest is the operator ˆ Nj = a† aj .2: Operators of 2nd Quantization Let A† be the matrix adjoint to Aj . n2 . . nS ) = 0 . (9.22) ˆ Nj |ΨB (N ) = nj |ΨB (N ) . .

. .e. . nS ) = . Exercise 9.s. 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 . showing that the property holds for nj = 0 and. (9. e. xN |ΨF (n1 .24) and the property aj |0 = 0. To prove this property we notice that the l. i. 9.9). 9.2.19) without aﬀecting the fermion symmetry (9. . x2 .37) (9.24) and aj |0 = 0 imply that the the properties (9.. x2 .23) hold for the state deﬁned through (9.29. j Exercise 9. .. f (a† )] j [a† .30). .g. nS ) .. xN |ΨF (n1 . xN |ΨF (n 1 . .23. . . 9.. . . φj+1 (x1 ) φj+1 (x2 ) . nS ) = − x1 . .2: Show that the boson operators a† and aj satisfy j [aj . . .246 Many–Particle Systems In using the boson creation and annihilation operators. . . i. of (9. x1 . . . it also holds for nj + 1. nj (9.35) and where the derivative operation is deﬁned through ∂f (A) = ∂A ∞ n=1 1 f (n) (0) An−1 . .38) corresponds to x1 . 9. nj+1 .. . . . . . . x2 . . . . in general. .. (n − 1)! (9. .36) Creation and Annihilation Operators for Fermions We want to consider now creation and annihilation operators for fermions. √1 N! . in case it holds for nj . .. 9... .17. . from the vacuum state |0 or states |ΨB (N ) . one does not need to worry ever about proper symmetries of wave functions.39) .2. nj . nj+1 .22. .1: The commutation relationships (9. one needs to apply only the commutation propertes (9. . . Obviously. 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. . the fermion wave function changes sign when one exchanges the order of the occupancy. . . operators which can alter the occupancy of the wave function (9. . φj (xN ) φj+1 (xN ) . . in case that the single particle states |j and |j + 1 are both occupied. nj .h. As long as one starts from a wave function with the proper boson symmetry. .23. Prove this by induction.21.38) .. since they are induced through the algebra of a† and aj . nj+1 = 1.e. φj (x1 ) φj (x2 ) .

. ... n2 . nj = 0. . .19). nj+1 . nj → 0.. nS = 0) φ2 (x1 ) φ5 (x1 ) 1 1 = √2 = √2 [φ2 (x1 )φ5 (x2 ) − φ5 (x1 )φ2 (x2 )] . . .e. . . at which column of the Slater determinant (9. ..40) . . For this purpose one must adhere to a strict convention: the labelling of single–particle states by indices j = 1. i. . A proper example is the two particle fermion wave function x1 . . .39) and (9. xN |ΨF (n1 . .42) i. . . .2: Operators of 2nd Quantization The r. in order to be consistent with this convention. . the expressions (9.e.19). . n2 = 1. . . . n2 . . . . nS ) qj states occupied In case that the single particle state |j is not occupied. x2 |ΨF (n1 = 0. (9. . . at the ﬁrst column of the Slater determinant (9. nS ). nj+1 . We are now ready to deﬁne the annihilation operator cj for fermions in the single–particle state |j as follows cj |ΨF (n1 . . i. . .. One adopts the convention that particles are created and annihilated by these operators always at the ﬁrst position of the wave function. particles are being created or annihilated. (9.h. We will be adopting below the notational convention that qj corresponds to occupancies N = (n1 . φj (x1 ) φj (x2 ) . nS ) as follows: j−1 qj = k=1 nk .s. Because of the property of the determinant to change sign when two columns are interchanged. nS ) = nj (−1)qj |ΨF ( n1 . φj+1 (x1 ) φj+1 (x2 ) . 247 − √1 ! N .. n4 = 0. This requires one. . . it is necessary to specify for a fermion wave function the order in which the single– particle states |λj are occupied. 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.. . .e..e. n6 = n7 = . 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). of (9. the rhs. . . however.. nj−1 . nS ) = . This change of position is connected with a possible sign change (−1)qj where qj is deﬁned for a given N = (n1 . Obviously.. . n5 = 1.38) reads − x1 .9. . x2 .40) are identical.43) . n3 . . n3 = 0. φ2 (x2 ) φ5 (x2 ) (9. n2 . . 2. . . . nj . (9.41) 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 . φj+1 (xN ) φj (xN ) . that occupancies are always ordered according to a monotonous increase of the single–particle state index.. .. n2 . . i. vanishes. qj is the number of states |k with k < j which are occupied in a fermion wave function. .

nS ) = j j (−1)qj (1 − nj + 1)(−1)qj nj |ΨF (N ) = nj ΨF (N ) (9. nj → 1. the last factor on the rhs. We consider (9.44). . .48) vanishes. . . .248 The operator c† adjoint to cj exhibits the following property j c† |ΨF (n1 . .45) (9. otherwise. .44) follows from the deﬁnition (9. . c† ]+ = δj k k where we have used the deﬁnition of the so-called anti-commutators [A. nj−1 . 9. nj .47) Let C† be the matrix adjoint to Cj .19). .50) . .46). of (9. nS ) = j Many–Particle Systems (−1)qj (1 − nj ) |ΨF ( n1 . . We will prove now the anti-commutation property (9. It holds c† cj |ΨF (N ) = c† (−1)qj nj |ΨF (n1 .49) The derivation involves realization of the fact that a non-zero result is obtained only in case nj = 1. nS ) = j (−1)qj (nj + 1))(−1)qj (1 − nj ) |ΨF (N ) = (1 − nj ) |ΨF (N ) (9.48) From this follows (9.46) [cj . Let Cj be the matrix corresponding to the operator cj in the basis of fermion states (9. n3 .43). nS ) qj states occupied The operators thus deﬁned obey the commutation properties [cj . nj → 0 . . . [c† .44) (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 . (9. (9. nj+1 . The anti-commutation properties (9.45) follow in an analogous way and will not be derived here. ck ]+ = 0 . We have used here the deﬁnition qj = k<j nk along with qj = qj . 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 . .46) ﬁrst for the case j = k. Similarly one obtains cj c† |ΨF (N ) = cj (−1)qj (1 − nj ) |ΨF (n1 . B]+ = AB + BA.17. nj → 1 . c† ]+ = 0 j k (9. n2 . . We ﬁrst show that (9. n2 .

. . in the basis |ΨF (N ) . i. of this equation we meant to indicate only those N occupation numbers nλj which are non-vanishing. . .. j → j. Correspondingly. . nλ2 .2: Operators of 2nd Quantization (9. .) = c† 1 c† 2 · · · c† N |0 . i. Equation (9.46). . hence. x|j → φj (r) | 1 σ .e.s. . λ λ λ (9.e.52) One can obtain similarly the same relationship in the case j > k. S (9. with eigenvalues equal to the occupation numbers nj ˆ Nj |ΨF (N ) = nj |ΨF (N ) ˆ Nj . hence.52) hold for any |ΨF (N ) one can conclude (9.55) is called the particle number operator. σ 1 (σ = ± 2 ) . .9. i. is referred to as the occupation number operator. On the r.e. .49) shows that the operator ˆ Nj = c† cj j (9. 9. The stated property ˆ follows from the idempotence of Nj which can be derived as follows † ˆ2 ˆ Nj = c† cj c† cj = c† (1 − c† cj )cj = c† cj − c† c† cj cj = cj cj = Nj . Since (9. . 2 (9. 1 In the following we will apply fermion operators c† and cj only to electrons which carry spin 2 . .. We j will often separate the states |j .s. coordinates x and index j into a space part and a spin part.h. nλN . . nk = 1 .50) yield ( cj c† + c† cj ) |ΨF (N ) = |ΨF (N ) .29)) |ΨF (. the creation operators must operate in the proper order.53) is diagonal in the occupation number representation.51. . nk = 1 . nj = 0 . j j j j j j j (9. nj = 0 . ( cj c† + c† cj ) |ΨF (N ) = 0 .51) (9. 1}.54) ˆ N = j=1 ˆ Nj (9. . nλ1 . an operator cj must be on the left of an operator ck for j < k. . j j For j < k one obtains cj c† |ΨF (N ) = (−1)qj +qk nj (1 − nk )|ΨF (n1 . 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. for which purpose one needs to envoke only the algebraic (anti-commutation) properties (9. k k 249 (9.46). nS ) k c† cj |ΨF (N ) = (−1)qj +qk −1 nj (1 − nk )|ΨF (n1 . ˆ It is an interesting exercise to demonstrate that Nj only has eigenvalues 0 or 1.58) . nS ) k and.56) Here we have made use of cj cj = 0 and c† c† = 0 which follows from nj ∈ {0..h.57) On the l.49) and (9. . . . .

15. N are not all identical. j = 1.63) ˆ ˆ This expression has been speciﬁed for the purpose of these notes to distinguish F and f and is not common ˆ terminology.59.e.F (N )|G|ΨB. xk ) ˆ j. ˆ . The operators g (x. (9. .17) and integrating over all particle space-spin coordinates x1 .F (N ) . 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.F (N )|F |ΨB. x2 . x ) which constitute G are ˆ called the associated generic two–particle operators. 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. .15. .3 One– and Two–Particle Operators Deﬁnition Operators acting on N particle systems of the type N ˆ F = j=1 ˆ f (xj ) (9.60) is that the sum over particles in (9. respectively – which are all identical. xN . 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 ) . . .k=1 j=k (9. (9. .59. .60) ˆ correspondingly are called two–particle operators.59) ˆ ˆ are called single–particle operators. 9. i. (9. An operator N ˆ R = j=1 rj (xj ) ˆ (9. 2. The essential aspect of deﬁnition (9. 9. The action of the operators boson and many–fermion states |Ψ is obviously described through the many–particle state matrix elements ˆ ΨB. Operators of the type 1 ˆ G = 2 N g (xj . These operators had been introduced already in Eq. namely. The operators f (x) which constitute F are called the associated generic single–particle operators1 .6) above.59) and (9.59) and over pairs of particles in (9. xk .60) involves generic operators – acting on xj and on xj . . 9. . 9.62) These matrix elements can be obtained by chosing the operators and many-particle states in the position representation. . ˆ ΨB. respectively. We adopt a similar expression to distinguish two–particle operators G and g . Since the many–particle states are built-up from a basis |r .F (N ) . σs ) (9. 2.17). ˆ r|f |s = 1 ∗ ˆ d3 r ψr (r. S of single–particle spinorbital states (see page 241) and the operators are speciﬁed through the associated generic operators ˆ f and g .61) is not a one particle operator as long as the N operators rj (x)..250 Many–Particle Systems 9. σr ) f (x) ψs (r.

69). s|ˆ|t. σt )ψu (r . ∂r 2 (9. ˜ (9. g g (9.k=1 j=k q2 |rj − rk | (9.69) ˆ In this case the generic operator is f (x) = δ(r − r). i. ˆ 251 (9. The matrix elements of this operator in a single–particle basis are ˆ rσ|t|sσ = δσσ d3 r φ∗ (r) r 2 − 2m ∂2 φs (r) .65) This symmetry will be exploited repeatedly below. 9. u which follows directly g from the deﬁnition (9. u = g d3 r ∗ ∗ d3 r ψr (r.66) and (9.e. ∂rj2 (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 .. is t(x) = (− 2 /2m)(∂ 2 /∂r 2 ). that the symmetry implies that one can switch simultaneously the pairs of indices r.64) r.59). (9. s|ˆ|t. s and t.71) . We will show that the boson and fermion creation and annihilation operators serve this purpose.66) Here we adopt the notation for the Laplacian ∂2 ∂2 ∂2 ∂2 = + + . x ) ψt (r. u. are spin-independent as is evident from the factor δσσ . s|ˆ|t. ∂r 2 ∂x2 ∂x2 ∂x2 1 2 3 (9.63. σs ) g (x. one cannot switch the indices of one pair individually.9. σu ) . Notice.3: One– and Two–Particle Operators r.64) We like to note an important symmetry of the matrix element r. r|ˆ|u.70) Both operators.67) ˆ In this case the generic single–particle operator. u = s. An example for a two–particle operator is the Coulomb repulsion operator 1 ˆ V = 2 N j.64). σr )ψs (r . according to (9. The aim of the present section is to express the matrix elements of single– and two–particle operators in a basis of many–particle states (9.62) in terms of the matrix elements of single–particle states (9.68) Another single–particle operator is the one–particle density operator N ρ(r ) = ˆ˜ j=1 δ(r − rj ) . 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. t .

The generic operator is given through the Pauli matrices. σ = 2 q φ (r )φ (r ) δ δ d3 r2 φ∗ (r1 )φ∗ (r2 ) r s |r1 − r2 | t 1 u 2 σσ σ σ d 3 r1 (9. s. σ.73) is a one-particle operator.252 Many–Particle Systems which is also spin-independent. u. one can obviously extract the term j Sj to be left with a two–particle operator in the strict sense of our deﬁnition. not in the strict sense of our deﬁnition above since the restriction 2 j = k does not apply in the summation. Sj is the spin operator for particle j with components (in the spherical representation) Sj. (9. s = 1 σ.e.76) the single–particle state matrix elements of which are r. α|ˆ+ |s.− S2. σ. α|ˆ3 |s.3 2 2 (9. σ . α s r.− . .74) = N j=1 Sj = 2 N Sj · Sk j.+ S2. β s The operator S 2 = δrs = δrs = = + − 1 2 1 2 δrs δrs (9. The generic operator is s12 = S1 · S2 = ˆ 1 1 S1. i. β|ˆ− |s. u. The ˆ matrix elements of the generic operator in terms of single–particle states are ˆ r. σ | t |t. s. However. α s r.77) δσ σ ( δσσ − δ−σσ ) The operators (9. x ) = q 2 /|r1 − r2 |. β s r.74) and (9.+ . β|ˆ3 |s. σ | s12 |t. σ ˆ + 1 δ− 1 σ δ 1 σ δ 1 σ δ− 1 σ 2 2 2 2 2 = δrt δsu + 1 4 δσσ 1 2 δ 1 σ δ− 1 σ 2 2 δ− 1 σ δ 1 σ 2 2 + . however.3 S2. Here.77) are orbital-independent as evidenced by the factors δrs and δrt δsu in the respective formulas. The operator N S = j=1 Sj (9. σ .+ + S1. Sj. The non-vanishing matrix elements of the spherical components of s are (α = + 1 . Sj. In this case the generic operator is v (x.3 .72) As a ﬁnal example of one– and two–particle operators we consider operators of total spin. β = ˆ 2 2 1 −2) r.− + S1.75) is a two–particle operator..k=1 (9.

23.9. s|ˆ|t. namely cu ct .46).64).60) as follows N ˆ f (xj ) → j=1 N 1 2 1 2 S r. The matrix elements are then actually those of the number operator c† cr which is diagonal in the occupation number representation. 9.83) t=1 t=r. i.62) are correct one needs to compare the result derived with conventional derivations of the matrix elements2 . g and not ct cu .78) and (9. respectively.82) ΨF (N )|c† cs |ΨF (N ) = r cr 2 (9. Expressions (9.81) For the latter matrix elements simple rules can be derived from the algebraic properties of the boson and fermion operators (9.63.176 where the matrix elements of fermion states can be found.62) through single–particle state matrix ˆ ˆ elements (9.24) and (9.79) It is crucial to notice in the expression for the fermion two–particle operator that the order of the annihilation operators in (9. 9. To show that the resulting values of the matrix elements (9. .u=1 (9.79) in the basis of many–particle states |ΨB. and then in g a second step the matrix elements ΨB (N )|a† as |ΨB (N ) . s|ˆ|t. u . 9.59) and G in (9. These rules will be provided below only for the case of fermions.s.63) and (9. r ΨF (N )|c† cs |ΨF (N ) . s|ˆ|t. (9.t. u a† a† at au bosons g r s † † r. xk ) → ˆ j. r ΨB (N )|a† a† at au |ΨB (N ) r s ΨF (N )|c† c† ct cu |ΨF (N ) r s bosons fermions (9.171 and pp.64) one replaces the operators F in (9.s=1 ˆ r|f |s a† as bosons r ˆ r|f |s c† cs fermions r r.s.79) is opposite to that in the matrix element r.u=1 S r.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’.. pp.79) have the following implication: The operators (9.78) 1 2 g (xj .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. respectively. s|ˆ|t.s=1 S r.45.F (N ) have the same values as the respective matrix elements (9. The Matrix Elements ΨF (N )|c† cs |ΨF (N ) r We consider ﬁrst the case r = s. r S Ψ (N In case of r = s one obtains F )|c† cr |ΨF (N ) r = nr r=1 δnr nr .62).80) (9.e.t.78.k=1 j=k S r. In order to determine these matrix elements one needs to evaluate ﬁrst the matrix elements ˆ of the generic operators r|f |s r. u cr cs cu ct g fermions (9. 9. u as deﬁned in (9.

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

r s|ˆ|r t − r s|ˆ|t r g g ˆ c† ct Nr .t>u r s|ˆ|t u − r s|ˆ|u t g g c† c† cu ct .s r. r s|ˆ|r t c† c† ct cr g r s r s|ˆ|t s c† c† cs ct g r s r s|ˆ|s t c† c† ct cs g r s r s|ˆ|t r c† c† cr ct . One obtains 1 ˆ ΨF (N )|G|ΨF (N ) = 2 S nr ns r.t=1 a.3: One– and Two–Particle Operators 255 Obviously.90) ˆ Since Nr is diagonal.t>u S r. i.s.92) ˆ This contribution to G has non-vanishing matrix elements only when N diﬀers from N in four ˆ occupation numbers nr .t.s. s (9.d.i.t=1 a.u r>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. employing the symmetry property (9. Commutation of ct cu according to (9.s. r g g .d.88).s. ˆ The second contribution G1 is ˆ G1 = + + + 1 2 1 2 1 2 1 2 S r. ˆ Case 0 This case covers diagonal matrix elements. ns .45). s − r. r s (9.i. nu .86).t=1 a. this contribution obviously behaves similarly to the oﬀ-diagonal part (9.i. As we will ﬁnd. S r.t>u S r.t.t=1 a.i. N = N or nr = nr for all r.u r<s.93) . this contribution to the two–particle operator is diagonal in the occupation number ˆ representation. g r s (9.9.s.u r>s.i.. (9. Only G0 .s. nt .t.u r<s.t=1 a.89) S r.d.d.s.84) ˆ of the one–particle operator F .t.t<u S r.s. s|ˆ|r.s. contributes in this case. (9.s. given in (9.91) ˆ G2 = r. For this purpose we consider three ˆ We can state now the matrix elements Ψ cases which actually correspond to the three cases considered below Eq.t. A similar contribution does not arise for F .e.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 (9.65). F (N )|G|ΨF (N ) .d. this part accounts for all diagonal contributions to G.u r>s. S r. ˆ A similar series of transformations allows one to express G2 which can be written ˆ G2 = + + + as follows S 1 2 1 2 1 2 1 2 r. s|ˆ|s.

u g g (9.95) In the latter formula the ‘+’-sign applies for s < r. We like to express these operators through ˆ fermion creation and annihilation operators.92) from (9.62) for bosons. nt = nt + 1. given in (9.e. nv = nv − 1. respectively. one–particle and two–particle operators. u < v. One obtains for the three components of S ˆ S3 ˆ S+ ˆ S− = 1 2 S ˆ ˆ Nrα − Nrβ = = r=1 S c† crβ rα r=1 S c† crα rβ r=1 (9.90).76). The matrix elements of the corresponding generic operators are (9. r=1 r=1 Exercise Exercise Exercise Exercise Exercise 9.1: 9. for ˆ non-vanishing matrix elements ΨF (N )|G|ΨF (N ) at most four occupation numbers nr and nr can diﬀer.. t for which holds ˆ ns = ns + 1 and nt = nt − 1. Furthermore.256 Many–Particle Systems Case 1 This case covers oﬀ-diagonal elements with nr = nr except for r = s. i. v for which holds ˆ ns = ns + 1.4: 9.3.90) from (9. Derive (9. ˆ ˆ Show that the matrix elements of F and of G conserve particle number. u > v or s > r. u < v or s > r.3. Derive (9. the ‘−’-sign applies for s < r.74) and (9.5: Derive (9. u > v.91). Only G1 . s|ˆ|t. In particular. u. All matrix elements which are not covered by the three cases above vanish. Only G2 .92).75) which are. s|ˆ|u.3.73) and S 2 given in (9. nu = nu − 1. s|ˆ|r.3. v − r.92).94) Case 2 This case covers oﬀ-diagonal elements with nr = nr except for r = s. One obtains ˆ ΨF (N )|G|ΨF (N ) = ± (−1)qs +qr +qu +qv (1 − ns )(1 − nr )nu nv r. Derive the non-vanishing matrix elements (9. contributes in this case. t − r.3: 9. contributes in this case.89).t nr r. it holds S nr = S nr . s|ˆ|v. t.3. given in (9.2: 9. particle number is conserved.96) .95) from (9. One obtains ˆ ΨF (N )|G|ΨF (N ) = (−1)qs +qt (1 − ns )nt r r=s. Spin Operators ˆ ˆ In this section we will brieﬂy consider the spin operators S given in (9. r g g (9.

e. rα sβ (9.4 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. the two–particle operator makes the description of many–particle systems a very hard problem.3.9. v c† c† cu ct . in crystals. the concerted behaviour of interacting systems can deviate qualitatively from that of systems made up of independent components. 9. Often.98) In many cases the Hamiltonian is spin-independent and the equivalent notation S H = r. superconductivity. The outcome of these studies is that even when interactions between components (particles) are simple.t. In the latter case the indices r.s=1 σ 1 ˆ rσ r|t|s c† csσ + 2 S r. If it were not for the two–particle contribution to the Hamiltonian.g..97) is over the orbital states. which represents the interactions between particles. Unfortunately. s|ˆ|u.t.s=1 c† c† crβ csα . 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. e. s.σ (9. u c† c† cuσ ctσ v rσ sσ r. 9. s|ˆ|t. molecules. Examples is the laser action. 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 systems investigated involve a macroscopic number of components such that statistical mechanical approaches are envoked.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.99) has preoccupied an important part of all intellectual eﬀorts in Theoretical Physics during the past ﬁfty years. u refer only to the orbital part of the single–particle basis. that the two–particle operator representing interactions between particles induces a variety of interesting phenomena. The great generality of the concepts developed in the context of many–particle systems can be judged from the fact . Actually.u S (9.s. 2 2 Exercise 9. but also ordinary phenomena like freezing and evaporation. atoms and nuclei. In fact. would be a simple exercise in linear algebra.97) The summation in (9. The fortunate side of this is.96.96) and (9.98.6: Derive (9. the description of many–particle systems.g.99) will be used. the spin part of the single–particle states is represented by α = 1 and β = − 1 .97).u=1 σ. evaluation of their stationary states and excitation energies. t. v r s 2 r. 9.s. the study of problems governed by Hamiltonians of the type (9.s=1 r.

4. a Hamiltonian without a two–particle operator contribution accounting for interactions among the particles. namely of neurons3 . r=1 3 Urm U ∗sm = δrs m=1 (9. .103) An account of some of these eﬀorts can be found in Modelling Brain Function – The World of Attractor Neural Networks by D. . New York. another prototypical systems of many strongly interacting components.100) in a form in which the factors c† csσ become rσ occupation number operators. (9. We will restrict our description in the following to systems composed of an even number (2N ) of particles and to spin-independent interactions. The electronic properties of atoms with diﬀerent numbers of electrons would diﬀer little. ultimately depend in a crucial way on the fermion character of its constituent building blocks. Amit (Cambridge University Press.100) ˆ ˆ We will denote the matrix elements of t as r|t|s = trs . . Before continuing one may ﬁnally point out that the material world as we know it and as it establishes the varieties of natural substances. 1989) . In such a representation Ho is diagonal and the stationary states can be stated readily.. would not exist. ranging from nuclei to the molecules of living systems . can be described in a rather straightforward way. i. S}. electrons and nucleons..1: Imagine that in a closed. Transformation to a New Set of Creation and Annihilation Operators Our aim is to represent the Hamiltonian (9.J.e. according to which nuclei and atoms change their qualitative properties when they grow larger. 2. If it were’nt for the fermion nature.101) The unitary property of Urn U† U = U U† = 1 1 or S S (9.102) U ∗rm Urn = δmn . all material systems would condensate into states which would depend little on particle number.s=1 σ ˆ rσ r|t|s c† csσ (9. m = 1. Chemistry essentially would know only one element and Life would not exist. water-ﬁlled jar all electrons of water turn their fermion nature into a boson nature. {|m). What might happen? Independent-Particle Hamiltonian A many–fermion system governed by an independent-particle Hamiltonian. The Hamiltonian of such system is S Ho = r. Exercise 9. To alter the representation of Ho we apply a unitary transformation U to the single–particle states |r to obtain a new basis of states |m).258 Many–Particle Systems that today these concepts are providing insight into the functioning of biological brains. i. where S |m) = r=1 Urm |r .e. The Aufbau principle.

101) of states |r . d† ]+ = δmn δσσ nσ (9. d† ]+ = 0 (9. (9. for nσ rσ example.108) and (9. (9. hence.108) obey the same anti-commutation properties (9. In a basis of states N d† j σj |0 n j=1 (9. (9.s=1 ˆ U † tU mn (9. nσ We demonstrate here the anti-commutation property (9. behave exactly like the operators c† and crσ behave for states |ΨF (N ) . csσ = n=1 Usn dnσ .104)] S S ∗ Urm d† mσ m=1 c† rσ = . the anti-commutation properties (9. . dnσ = r=1 ∗ Urn crσ (9. accordingly.9. The unitary property (9.s.103) allows one to express [c.and two-particle operators hold in an analogous way for d† and dnσ .111) and.111).45. (9.107) Urn c† rσ . [dmσ .101) are occupied. the operators d† and dnσ . 9.h.4: Independent-Particle Models allows one to express conversely |r in terms of |m) S 259 |r = m=1 ∗ Urm |m) . they are Fermion creation and annihilation operators.109) The operators (9.106) which together with (9. using (9. [d† . i. ∗ ∗ Urm Urn . c† ]+ = δσσ [dmσ . d† ]+ = sσ nσ r..46) as c† and crσ .110) can be demonstrated similarly. the expressions drived above for the matrix elements of one.s r The unitarity property r ∗ Urm Urn = δmn yields immediately identity (??).110) mσ nσ [dmσ .113) Urm Usn [crσ .100) yields ˜ Ho = tmn d† dnσ mσ where d† nσ = r=1 S S (9.f. rσ namely.46).104) One can. dnσ ]+ = 0 . of (9.112) . expand S trs = mn ∗ ˜ tmn Urm Usn (9. of N -Fermion states in which single particle states |nj ) as deﬁned in (9.111) can be written.e.108) These operators describe the creation and annihilation operators of fermions in states |n) which are linear combinations (9. The l.105) where ˜ ˆ tmn = (m|t|n) = S ∗ trs Urm Usn = r.

We apply for this purpose (9. a simple matter to nσ state many–particle states in the new representation.106) and the unitary property of Urn .101) gives us the freedom to choose the matrix (9.e. 2. transformation matrices Urn and energy values n . S (9. i. j j (9. . n = 1. USn ). 1 as well as a fermion in a spin state | 1 . i. 1 and |β = | 1 . the so-called ground state. i. .114) Ho = n=1 † n dnσ dnσ (9.116) which follows from (9.106) is equivalent to S trs Usn = s=1 n Urn ∀r. ) = j=1 nj . n2N . σ2 .115) can be stated immediately since any many–particle wave function in the occupation number representation is suitable.114) together with (9. . Eigenstates of (9. σj ) = (nk . σ1 . 2 2 2 2 i. . . .119) In this state the N lowest orbital eigenstates of trs are occupied each by an electron with spin |α = | 1 . is obtained. the associated ( properly normalized) eigenvectors Vn are the columns of (Urn ). .260 Diagonal Representation Many–Particle Systems The transformation (9.57) to the case of 2N particles. Before proceeding we like to address the issue how the new representation..e.. . 2.. is N |Φo = j=1 d† α d† β |0 .118) Ground State In case of an ordering m < n for m < n the state of lowest energy. . hence. r = 1. a fermion in a spin state | 1 . ˜ tmn = In this case Ho has the desired form S n δmn .106) diagonal.115) ˜ and involves solely occupation number operators N nσ = d† dnσ .117) are eigenstates of (9. U2n . (9. The condition (9. σ2N . Vn = (U1n . (9. It is. . S are T real. The 2N fermion state 2N d† j σj |0 n j=1 where (nj . . − 1 .e. σk ) for j = k (9.115) with eigenvalues 2N E(n1 . n2 .. A non-degenerate orbital state which is occupied by two fermions. The corresponding eigenvalues n . This equation poses the eigenvalue problem for the hermitian matrix (trs ).e. . − 1 is called a closed 2 2 2 2 .

n=1 d† d† dmβ dnα mα nβ (9. both with eigenvalues 2 ˆ S 2 |α. 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. 1 S 4 m.120). β |β.4: Independent-Particle Models 261 shell. namely.120) where the occupation number operators refer to the single-particle states |m).n=1 † † if either |m) or |n) are closed shells. One can conclude immediately that the ˜ ˜ ˜ ˜ (N mα − N mβ )(N nα − N nβ ) does not give any contribution second term in (9. the third term − S m. Excited States We want to construct now excited states for the system governed by the independent-particle Hamiltonian (9.123) 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. β ˆ are eigenstates of S 2 given in (9. the states with energy closest to the ground state are those in which a particle is promoted from the highest occupied state |N ) to the lowest unoccupied state |N + 1).. We will demonstrate now that this property endows the ground state with total spin zero. of ∆E = and |β. Obviously. is a singlet state.97). 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. 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.. 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.n=N (9.n=1 m. The two states |α. α |α.e. α N +1 − N . The action of the operator (9. |α.124) .121) All four states have the same excitation energy. However.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. ˆ S 2 ||β.n=1 m. ˆ one can conclude that |Φo is an eigenstate of S 2 with eigenvalue zero.120) is particularly simple for closed shells.9. energy above the ground state. In case of m = n it gives a contribution ‘1’ for each closed shell which cancels the contribution of the ﬁrst term. β .100). α . β = 2 |α.e.n=1 cmα cnβ cmβ cnα does not give a contribution if m = n and either |m) or |n) are closed shells. The ground state has only closed shells.n=1 d† d† dmβ dnα mα nβ (9. (9. namely. α |β. i. i. the states diﬀer in their spin character. α = 2 |β.97). There are four such states. Alltogether. Similarly.

. β both with eigenvalues 1. mα nβ (9. also eigenstates of Σ2 .127) ˆ are triplet and singlet states.2: Construct four operators O ˆ O ˆ such that O ˆ S3 . etc. conclude that (9. |N + 3 . of course. the linear combinations 1/2(|α. r = 1. m |Φo m m m γσ. α as triplet states. for example. ˆ The remaining states |α. β =− m. β qualify as eigenstates.128) The cyclic property of the system is expressed through c2N +1 σ = c1 σ .σ = (9. α . (9. |β. i. 2N ) drσ = √ 1 2N 2N e−irnπ/N cnσ n=1 (9. apropriate eigenstates of the operators S 2 and Example: 2N Independent Electrons on a Ring In order to illustrate the procedure to obtain eigenstates of one-particle Hamiltonians (9. The same result holds for |β. We identify the states |2N + 1 = |1 to avoid cumbersome summation limits. ˆ 3 on the two states produces . .4.n=N d† d† dmβ dnα d† +1α dN β |Φo = 0 mα nβ N (9.100) outlined above we consider a system of 2N electrons which move in a set of atomic orbitals |r .n=N Many–Particle Systems N +1 d† d† dmβ dnα . β are not eigenstates of S 2 . α and |β.e. . 2N which are located on a ring. .σ d† +1. . however. for |α. α are. 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 . .122) is correct. . ˆ An eigenvalue ‘2’ of the spin operator S 2 identiﬁes the states |α. do not need to be considered since they ˆ give vanishing contributions. One can. c† +1 σ = c† σ 1 2N (9. hence. The action of Σ N +1 ˆ Σ3 |α. The system is assumed to posses a 2N –fold symmetry axis and interactions connect only orbital states |r → |r ± 1 .129) The symmetry of the Hamiltonian suggests to choose a new representation deﬁned through the operators (r = 1.262 and ˆ Σ3 = − m.130) . . one with eigenvalue zero (singlet) and one with eigenvalue ‘2’ (triplet). The two states |α..σ dN σ N σ. vanish. 2. β . of course. . α ± |β. 2.125) ˆ Contributions to S 2 acting on states |N + 2 .126) which follows from the occurence of squares of fermion operators which. β and |β. ˆ Exercise 9.

1−a (9. thereby.137) where 1 ˜ trs = e−irπ/N 2N Using (9.s.f. indeed.132) one can conclude 2N ei(s−r)nπ/N + eisπ/N n=1 1 2N 2N ei(r−s)nπ/N . One can then express[. one obtains the new representation of Ho 2N ˆ Ho = −t r.132) gives for r = 0. eisπ/N 1 − e2N isπ/N 1 − eisπ/N (9.136) ˆ and. 9.132) an identity which can be derived employing the well-known expression for a ﬁnite geometric series m as = a s=1 1 − am . (9. .131) constitute a unitary transformation U deﬁned through Urn = unitarity property follows from 2N eirsπ/N = 0 r=1 for 0 < s < 2N .9.s=1 σ ˜ rσ trs d† dsσ (9. . n=1 (9. vanishes for integer s.131) The (9.130.4: Independent-Particle Models d† = √ rσ 1 2N 2N 263 eirnπ/N c† nσ n=1 √ 1 eirnπ/N . . 4N.140) The Hamiltonian is now diagonal and the construction outlined above can be applied.133) Application to the l.134) which. rσ (9. (9. .135) c† nσ 1 = √ 2N e−irnπ/N d† rσ r=1 (9.139) −2t cos r=1 σ rπ N d† drσ . . 2N (9.109)] ¸ cnσ = √ 1 2N 2N eirnπ/N drσ r=1 2N (9. 2N. of (9.138) ˜ trs = from which follows ˆ Ho = eirπ/N + e−irπ/N 2N δrs (9.h.

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

. . U = ( Urm ) as deﬁned in (9.108). . . r = 1. |r = m=1 ∗ Urm |m . . u c† c† cuσ ctσ v (9. j j (9. t. u refer to the initial basis of single-particle states |r . S is a unitary matrix.145) which will play a pivotal role: it is this the state in which the fermions are assumed to be moving and relative to which the mean interactions acting on individual fermions is determined. S) ˜ S S |m = ˜ r=1 Urm |r . .142) is based.142) over this reference state to turn the contribution into an eﬀective one-particle operator. m = 1. as in in (9.9. . . N .σ c† c† rσ sσ cuσ ctσ + c† c† rσ sσ c† ctσ cuσ rσ c† cuσ rσ ctσ cuσ ctσ + c† c† cuσ rσ sσ ctσ + c† sσ − c† c† ctσ cuσ − c† rσ sσ sσ (9. ˜ (9.s. s. . In our formulas below we will adopt strictly the convention that indices r. n.t. 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. . 2. s|ˆ|t. the corresponding U = ( Urm ) forms an S × N matrix and.5: Self-Consistent Field Theory 265 In this section we will consider systems of 2N fermions described by the spin-independent Hamiltonian S S 1 ˆ rσ H = r. S on which (9.s=1 σ r. If one restricts m = 1.146) .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. 2. . . q.143) with r. . naturally. 2. u v r.144) deﬁned as in (9. This is done as follows: ˆ Vmf (|Φo (U) ) = 1 2 S r. S} the elements of which will be labelled by indices ˜ m. .u=1 σ. m = 1. .145). p.t. The states |m serve to deﬁne a reference state of the system ˜ N |Φo (U) = j=1 d† α d† β |0 . 2.142) r|t|s c† csσ + rσ sσ 2 r.u=1 σ. The states |m are connected with the states |r through (m = 1. We will adopt a second set of single-particle states {|m . is not unitary. s|ˆ|t. .σ and derive a single-particle operator which approximates this Hamiltonian. 2.

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

˜ v ˜ sσ m≤N s. s ) ˆ ˆ m≤N ∗ Utm Uum (9. n c† ctσ δσ σ ˜ v ˜ rσ − r.u ( 2 r. as introduced here.m. Exploiting the symmetry r. The mean ﬁeld approximation. t| v |u..S.. 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.5: Self-Consistent Field Theory 1 2 1 2 1 2 S 267 m.u=1 σ S − m. The one-particle operator (9. u − r.. We will address now a proper choice of the reference state.151) Carrying out the sum σ δσσ = 2 leads to a factor 2 for the ﬁrst two terms. s| v |t.156) which deﬁnes the reference state |Φo (U) . m|ˆ|˜ .9.u=1 σ.t=1 σ (9.u=1 σ.t=1 σ.σ S r.155) which is a function of the S × N –matrix U = ( Urm )r=1. t| v |s. leaves open the question how the reference state should be chosen.2.m.s σ where r| vmf (|Φo (U) ) |s = ˆ t..152) can then be written in the form ˆ Vmf (|Φo (U) ) = r| vmf (|Φo (U) ) |s c† csσ ˆ (9..2. At this point this state is completely arbitrary. s|ˆ|t.153) rσ r.σ r. u c† cuσ ˜ vn rσ (9. . n c† ctσ .σ S + − r. m|ˆ|˜ ..142) by Hmf (U) deﬁned through S ˆ Hmf (U) = r. u = s. u c† cuσ ˜ vn rσ m≤N r. u c† cuσ δσσ ˜ vn sσ m≤N s.N (9.154) ˆ The mean ﬁeld approximation replaces then the Hamiltonian (9.149) for the matrix elements in the mixed representation one can state the right hand side explicitly in terms of Urm . m|ˆ|t. r| v |u.152) By means of expression (9.s=1 σ ˆ r|t|s + r| vmf (|Φo (U) ) |s ˆ c† csσ rσ (9. s|ˆ|˜ . m=1.

i.119. i. We will argue below that the self-consistent ﬁeld ground state. . often also referred to as the Hartree-Fock approximation. . S. however.145). .s=1 σ ˆ r|hmf |s c† csσ rσ (M ) (9. (9. Step 1: Choosing an Initial Reference State One deﬁnes the Hamiltonian ˆ (1) Hmf = r. m = 1. 2.159) One deﬁnes the reference state |Φo (U(1) ) using the deﬁnition (9.268 Many–Particle Systems 9. S . 2. The state thus determined is referred to as the self-consistent independent-particle ground state.155) following the method outlined in Section 9.. 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 (9.145). S . . 2. (9. m = 1.s=1 σ S ˆ rσ r|t|s c† csσ (9. is the lowest energy independent-particle ground state one can construct. . Step 2M. 2. is exact only for an independent-particle Hamiltonian. r = 1.160) . steps (2) and (3) are being repeated M times until the procedure converges. 2. 9.e. SCF-Algorithm. SCF-Algorithm.4 and deﬁning this as the new reference state. the reference state resulting from step 2M + 1 (within numerical error) is equal to the reference state resulting from step 2M − 1. .155). . .142). M = 1. Let us state now the construction of the self-consistent ﬁeld ground state in more detail. the independent-particle nature stemming from the fact that the functional form of the ground state. S . This procedure. U(1) is the S ×N –matrix (1) of the ﬁrst N column vectors of Urm . . determining in a step (2) the corresponding mean ﬁeld Hamiltonian (9. r = 1.s=1 σ S ˆ r|t|s + r| vmf |Φo (U(M ) ) ˆ |s c† csσ rσ = r.158) where the labels m are ordered to obey the condition (1) m < (1) n for m < n . under conditions which often are realized. can achieve this goal only iteratively. i.6 Self-Consistent Field Algorithm The Self-Consistent Field (SCF) approximation. assuming in an intial step (1) a properly chosen reference state. (9. . and obtaining in a step (3) the ground state of this one-particle Hamiltonian according to the construction in Section 9.4. . .e.: Determine Mean Field Hamiltonian In this step the mean ﬁeld Hamiltonian S ˆ (M ) Hmf = r.157) and determines the associated diagonal representation deﬁned through S (1) ˆ r|t|s Usm = s=1 (1) (1) m Urm . . . The procedure determines the reference state as the ground state of the independent-particle Hamiltonian (9.e. ..

. 2. ˆ Exercise 9. . S . .1: Let Prs be the one-particle operator with the generic operator prs = |r s|. (9.. . 2 .e. .162) is solved adopting an ordering of the labels m which obeys the condition (M +1) m < (M +1) n for m < n . . . S. 2. . t| v |s.166) ˆ (d) Deﬁne a similar two-particle operator Prstu . 2. u − r. the S × S–matrix S (M ) r|hmf |s N 269 ˆ = r|t|s + t. ˆ (b) How can the expectation values of the operator Prr be interpreted. M = 1. σ p = δtr δsu δσσ . (c) Show that any spin-independent one-particle operator S ˆ F = r. t| v |u. i. .e. U(M +1) is the (M +1) S × N –matrix of the ﬁrst N column vectors of Urm . S (9. S .6. . 2..6: Self-Consistent Field Algorithm is evaluated. . SCF-Algorithm. (9. S do not exceed a preset threshold.167) .s=1 σ ˆ r|f |s c† csσ rσ (9. i. etc. 2. 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. r = 1.145). . m = 1.164) ˆ (a) Determine the expectation value of Prs for the state as deﬁned in (9.165) can be written ˆ F = S ˆ ˆ r|f |s Prs r. indicating that the state converged.161) is calculated. ˆ t. r = 1.163) The result yields the reference state |Φo (U(M +1) ) using the deﬁnition (9.u=1 ( 2 r. i. . SCF-Algorithm: Continuation and Convergence Condition When step 2M + 1 is completed. (e) Consider the creation operators N † gqσ = m=1 Vmq d† mσ (9. σ|ˆrs |u. . m = 1.: 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 . s = 1. .9.145).e. . Step 2M + 1. s ) ˆ ˆ m=1 Utm ∗(M ) (M Uum ) (9. step 2M + 2 is carried out.s=1 (9. .

171) where the convention (9.σ + 2v ( c† c1α c† c1β + c† c2α c† c2β ) + v 1α 2α 1β 2β + c† c2σ c† 2σ 1σ c1σ ) .7 Properties of the SCF Ground State We want to investigate now the properties of the SCF ground state. For this representation holds N m| t |˜ + ˜ ˆn m 2 mm | v |˜ m ˜ ˜ ˆ n˜ − mm | v |m n ˜˜ ˆ ˜ ˜ = (SCF ) δmn m (9. 2.170) (SCF ) Here |r . r = 1.. . Equation (9.270 Many–Particle Systems which are connected with d† through a unitary N × N –matrix (Vqm ).171) implies that for the mean ﬁeld Hamiltonian holds S ˆ Hmf (U(SCF ) ) = m=1 σ (SCF ) † dmσ dmσ m . (9. Note that this is not an mσ S × S–matrix! Show that the reference state deﬁned through N † † gqα gqβ |0 q=1 (9. (f) Can one distinguish the states (9.169) 9. (9.163).172) is obeyed. (9.9.173) .145) and (9.168) through a physical observation? Exercise 9. We denote the representation which results from the SCF algorithm after its convergence as follows: S |m = ˜ r=1 (SCF Urm ) |r . S denotes the initial single-particle states and Urm is the unitary S × S– transformation matrix obtained in (9.168) ˆ ˆ has the same expectation values for Prs and Prstu as the reference state (9. We begin by summarizing the result of the SCF algorithm. (SCF ) m < (SCF ) n for m < n (9. .145).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.6.e. i.162. .163).

Employing expressions (9. mα mβ (9.177) 2 mm | v |mm ˜˜ ˆ ˜˜ − mm | v |m m ˜˜ ˆ ˜ ˜ .. Exercise 9.1: Reformulate (9.142). 2 m=1 m|t|m + ˜ ˆ˜ m. .177). r.174) The ﬁrst property we consider is the total spin character of SCF .e.7.162) is non-linear in Urm .93) for the expectation values (diagonal elements) of the one-particle and two-particle part of the Hamiltonian (9. We like to determine next the energy expectation value of SCF .146) counts a particle twice.142).171) yields ˆ SCF H SCF N = N (9. and once as a probe particle being subject to the mean pair interaction. Within the mean ﬁeld approximation as described by (9. This over-counting leads to the correction term in (9.142). Since the system under consideration has only 2N particles altogether. the mean-ﬁeld Hamiltonian is diagonal in the SCF representation. once as a member of the ground state.171) in terms of matrix elements r|t|s . 2 m=1 (SCF ) m − m. s|ˆ|t.173) holds N ˆ SCF Hmf (U(SCF ) ) SCF = 2 r=1 (SCF ) m .176) 2 mm | v |mm ˜˜ ˆ ˜˜ − mm | v |m m ˜˜ ˆ ˜ ˜ . the mean ﬁeld potential ˆ Vmf (|Φo (U) ) deﬁned in (9.2: Derive (9. exploiting the spinindependence of (9. one can also determine the expectation value of SCF for the Hamiltonian (9. u and Urm .m =1 Comparision with (9. Finally.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.9.7: Properties of the SCF Ground State 271 i. Since this state is composed only of closed shells one can conclude following Section 9. one obtains ˆ SCF H SCF N = N (9.175) However. (SCF ) ˆ Exercise 9.177).7. i.4 that SCF is a total singlet state. (9. ˆ SCF H SCF . v (SCF ) Show that the resulting eigenvalue problem of the type (9. the SCF ground state is N SCF = m=1 d† d† |0 .e.84) and (9..

t is assumed to be a positive. t describes the coupling of state |r to states |r ± 1 at the two neighboring lattice sites. The ﬁrst term on the r. r ∈ Z.s. The one-dimensional Hubbard model is described by the Hamiltonian ˆ H = −t rσ † c† r+1. is characterized through the parameters t. i.σ rσ The potential energy term. s|ˆ|u. . of (9. except in case of one-dimensional systems.272 Many–Particle Systems 9.σ crσ + crσ cr+1.180) V = rσ sσ 2 r. the second term on ther. .4–9. rσ rσ r σσ (9. u c† c† cuσ ctσ . . as stated through Hamiltonian (9.σ + v 2 c† c† crσ crσ .178). The case n = 1 is termed the half ﬁlled band case.128). N0 which are populated by 2N electrons (we choose an even number of electrons for convenience).e. of (9. s|ˆ|t.178) v contributes only in case that two electrons occupy the same lattice site.h. We assume that there are altogether S = 2N0 lattice sites. real number. v (9. is identical to that of the independent–particle Hamiltonian H0 (9.s. referring to the fact that in this case the band of single particle energies (9.178) and through the so-called ﬁlling factor n n= 2N N = . In the present Section we want to adapt the algorithm to systems containing a macroscopically large number of fermions.e.178). reads 1 ˆ r. According to (9.u σσ where r. (9.178) Here the index r. 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.8 Mean Field Theory for Macroscopic Systems The SCF algorithm has been developed in Sections 9.. respectively) in state |r at lattice site r. † ˆ Ho = −t c† . i. v represents the ‘on-site’ Coulomb repulsion.179) r+1.7 for ﬁnite systems.181) The Hubbard model.σ crσ + crσ cr+1. describes the sites of a linear lattice. in the usual two–particle operator form. v of the Hubbard Hamiltonian (9.e.s.. t = v δru δst δrs . in the two-dimensional copper oxide lattices of high temperature superconductors. S N0 (9. Due to the lack of an exact solution in dimensions higher than one. for example.192) derived below is ﬁlled half. . v (9.. labeled r = −N0 + 1.178). Both N0 and N are macroscopically large numbers.t. i. We consider as an example the Hubbard model of an inﬁnite linear lattice and apply the model to describe magnetic instabilities in metals. In spite of the simplicity of its Hamiltonian the Hubbard model. cannot be solved exactly. β for ‘up’ and ‘down’ rσ spins. approximation schemes like the self-consistent ﬁeld approximation play an important role.182) n can assume values 0 ≤ n ≤ 2. The operator c† (crσ ) creates (destroys) an electron with spin σ (σ = α.h. .

9. For this purpose we apply the SCF theory presented in Section 9. The self-consistent ﬁeld Hamiltonian Hmf which corresponds to the Hubbard Hamiltonian (9. Accordingly.187) The energy levels are labeled such that 0 < ±1 < . N0 (9. a freedom which we employ to adopt so-called periodic boundary conditions like those for the example “2N independent electrons on a ring” on page 262.181. < ±(N0 −1) < N0 holds as can be readily veriﬁed. The unitary transformation which diagonalizes H0 .183) where the operators d† are related to the original creation operators c† through the unitary mσ rσ ˆ transformation U deﬁned in (9.140). .184) ˆ We are now ready to apply step 1 of the SCF algorithm and diagonalize term H0 . rσ rσ (9. in fact.8: Macroscopic Systems 273 Since we are dealing presently with a macroscopic system.155.178) within the framework of the self-consistent ﬁeld theory.179).185) ˆ According to (9.186) one obtains for the mean ﬁeld Hamiltonian N0 ˆ (2) Hmf = m=−N0 +1 σ m d† dmσ + v mσ N 2N0 c† crσ . by N −1 ||SCF = m=0 d† d† |0 .184) and using (9. 2N0 (9. Using (9. we identify |r + N0 = |r and later let N0 go to inﬁnity. We can now embark on step 2 of the SCF algorithm.185) into (9. 9.6.154).. boundary eﬀects are assumed to be negligible. mα mβ (9.9. we assume that the SCF ground state is given by (9.186) where m = −2t cos mπ . one obtains for the mean ﬁeld Hamiltonian N −1 ˆ ˆ Hmf (U) = H0 + v r. Inserting (9.178) can be determined by applying (9. .174).144). Accordingly.e. rσ (9.188) . This task has ˆ been solved already on page 262. i. is Urm = √ 1 exp (irmπ/N0 ) .σ m=0 ∗ Urm Urm c† crσ . the operator H0 can be re-written in the diagonal form N0 ˆ H0 = m=−N0 +1 σ m d† dmσ mσ (9. SCF Ground State for the Hubbard Model We seek to determine the ground state of the Hubbard model stated by (9.

π and.191) where m is given in (9. Indeed. The energy m can be expressed as a function of a continuous variable k ∈] − π. π]. as given in (9. . To demonstrate this we consider an alternative formulation of the mean ﬁeld approximation for Hamiltonian (9.187). the state with the lowest possible energy.. even for one and the same Hamiltonian. One can see that −2t ≤ (k) ≤ 2t holds. the results of diﬀerent formulations agree qualitatively. the electrons in the present description behave like independent particles with energies ¯m = m + vn . π]. can be obtained. is already diagonal.185) one can prove rσ mσ N0 c† crσ = rσ rσ m=−N0 +1 σ d† dmσ . i. i.f. (9. Indeed. The bottom of the energy band corresponds to (0) = −2t and the width of the energy band is 4t. . the self–consistency condition is exactly met and the SCF algorithm converged after two steps. in the limit N0 → ∞. mσ (9. Apparently. Alternative Description of the Mean Field Approximation The state (9. dnσ according to (9.e. denoted by F (see Fig.192) This dispersion law is presented in Fig.1). called the Fermi energy. When N0 becomes macroscopically large the single particle discrete energy levels (9. there exist several ways of formulating mean ﬁeld approximations. 9. through the so-called dispersion law (k) = −2t cos k . . In other words. hence. The energy Emf of this state can be calculated readily by using (9. 9. The ground state of the system.190). the ground state of the Hubbard model in the self-consistent ﬁeld approximation can be determined exactly. N0 ˆ (2) ˆ Hmf = Hmf = m=−N0 +1 σ m + vn † dmσ dmσ 2 (9.182)].182) of the ﬁlling factor n [c. by ﬁlling up this energy band with electrons from the bottom of the band (which corresponds to k = 0) to a maximum energy.183) is not the only candidate for the mean ﬁeld ground state of the Hubbard model.109) and using (9. namely.177) and mm |ˆ|mm = mm |ˆ|m m = ˜˜ v ˜˜ ˜˜ v ˜ ˜ which holds in the present case. for km = mπ/N0 holds km ∈ −π + N0 . 4 So far we have not exploited the fact that N0 and N are macroscopically large quantities. but may diﬀer quantitatively. . 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. according to the Pauli principle. Usually.178)..274 Many–Particle Systems Expressing c† .189) and. holds k ≡ km ∈] − π. crσ through d† . ˆ The Hamiltonian Hmf .190) We have used in the latter expression the deﬁnition (9. −π + N0 .e. −π ≤k ≤π . (9.187) form a quasi1 2 continuous energy band.1.

say the one given in (9.8: Macroscopic Systems 275 Figure 9. First. Inserting (9.. The mean ﬁeld approximation neglects the . F = (±kF ) denotes the Fermi energy.193) where.178) and performing the summation over the spin indices results in ˆ ˆ H = H0 + v r ˆ ˆ Nrα Nrβ . The anti-commutation properties (9. we have employed (9.194) ˆ For a given reference state.45. the deviation of Nrσ from its mean value Nrσ for the corresponding reference state..183).9.e. i. 9. the operator Nrσ can be written ˆ Nrσ = Nrσ + δ (Nrσ ) . (9. (9. where Nrσ is the mean occupation number of the one particle state |rσ . Nrσ = SCF ||Nrσ ||SCF .56).1: The dispersion law (k) for independent electrons on an inﬁnite one dimensional lattice. let us express the interaction term in (9.178) in terms of the occupation number operaˆ tors Nrσ = c† crσ .193) into (9. i.e.195) ˆ ˆ ˆ and where δ (Nrσ ) = Nrσ − Nrσ describes the ﬂuctuations of Nrσ .46) of the fermionic creation and rσ annihilation operators yield c† c† crσ crσ = −c† c† crσ crσ = rσ rσ rσ rσ ˆ ˆ ˆ2 c† crσ c† crσ − c† crσ δσσ = Nrσ Nrσ − Nrσ δσσ . in the last term on the right hand side.196) (9. rσ rσ rσ (9.

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

192).206). This last equation tells us that the electrons with spin α (β) are accommodated by an energy sub–band α (k) ( β (k)) which is obtained from the dispersion law of the non-interactive electrons (k) by an overall shift of vnβ (vnα ). one can expect the system to lower its energy by assuming nα = nβ and. N0 ∼ 1023 .206) needs to be evaluated. if nα > nβ then β (k) > α (k). i.208) . a spin–polarized ground state. nβ and µ.201) ( m ˆ + vnβ ) Nmα + ( m ˆ + vnα ) Nmβ − 2N0 vnα nβ . the so-called Fermi energy F = µ (see below). (9..205) is met. the smaller is σ (k) for a given k.206) ascertains that condition (9. β) are determined by minimizing the energy density of the ground state Hmf 1 = [ mσ Nmσ ] − vnα nβ (9.e. (9. The ground state (corresponding to the lowest possible energy) of the many electron system is obtained by ﬁlling these two energy sub–bands with electrons up to the same maximum energy value.. E with respect to nα .9.206) ˜mf is minimized where µ is the Lagrangian multiplier considered an independent variable. (9. Since β (k) − α (k) = v(nα − nβ ).202) This Hamiltonian describes a system of non-interacting electrons with a spin-dependent dispersion law (9. At this point we exploit the fact that our system is macroscopically large. For this purpose the sum in (9.203) is provided by the continuous function σ (k) = (k) + vn−σ . 2π (9. The additional term in (9.g.207) where (k) is given by (9. m (9. 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.205) Such minimization is realized through the method of Lagrangian multipliers. hence.2a.e.204) Emf ≡ 2N0 2N0 m. In the limit N0 → ∞ this sum can be expressed as an integral. Therefore. (9. as is shown schematically in Figure 9. subject to the constraint n = nα + nβ .8: Macroscopic Systems from which we obtain nσ = The mean ﬁeld Hamiltonian is then ˆ Hmf = m 277 1 2N0 Nmσ .. According to this method one minimizes ˜ Emf = Emf + µ (n − nα − nβ ) . 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) .σ with respect to nα and nβ . In this limit the discrete energy spectrum (9. such that the larger nσ .203) mσ = m + vn−σ . e. The values of nα and nβ are determined by minimizing the energy density (9. The actual values of nσ (σ = α. i. In fact.

The shaded areas represent the ﬁlled portions of these bands by electrons.2 = ±(π − arccos( /2t)). δ(k + π − arccos( /2t)) + δ(k − π + arccos( /2t)) (2t)2 − 2 . Since the equation f (k) ≡ 2t cos k + = 0 has two simple roots.278 Many–Particle Systems (a) (b) Figure 9. namely k1.210) Here δ(x) is the Dirac–delta function and ρ( ). unit energy and a given spin orientation of the particle. and noting |f (k)| = |2t sin k| = 2t one obtains δ( − (k)) = 1 − cos2 k = (2t)2 − 2 . For a given dispersion law (k) the density of states ρ( ) can be calculated by employing equation (9. gives the the number of available one particle states per site.210). If the function f depends explicitly only on (k) one can state π −π dk f ( (k)) = 2π π ∞ ρ( )d f ( ) −∞ (9. therefore.209) where ρ( ) = −π dk δ ( − (k)) .2: (a) Relative position of the two energy sub–bands α (k) and β (k) for ∆ = nα − nβ > 0.211) and using π dkδ(k ± (π − arccos( /2t))) = θ(2t − | |) .212) where xi are the simple roots of the function f (x). 2π (9. 2π (9. (b) Graphical deﬁnition of the limiting energies α and β . |f (xi )| (9.192) and. holds π ρ( ) = −π dk δ( + 2t cos k) . Inserting this last result into (9. −π .211) In order to calculate the integral we recall the following property of the Dirac–delta function δ [f (x)] = i δ(x − xi ) . In our case (k) is given by (9. which is usually called the density of states.

if we assume that is restricted to the interval ] − 2t. β . namely.221) and (9.206)] depends only on continuous quantities. π α.217) and µ.219.e.9.214) are given by the sum (9. θ(x) = 1 if x > 0 and θ(x) = 0 if x < 0. (9. in principle. F = α + vnβ = β + vnα .. namely.f. (9.2]. From (9.216) allow one. on conditions for a ground state of minimum energy are ˜ ∂ Emf =0. of t.201).221) ∂µ Equations (9. (9. ∂ β ˜ ∂ Emf =0.215) nα and nβ in (9. (9. ∂ α ˜ ∂ Emf =0. ∂µ (9.213) and carrying out the resulting integral yields nσ = 1 arcsin( σ /2t) . + vnα − µ = 0 .216) Using (9. to determine the unknown quantities α . (9. Replacing again the sum by an integral over energy one can write σ nσ = −2t ρ( )d (9. 2t[ then the θ function in (9. 9. i.e.. i.219) (9.. nα and nβ as a function of the parameters of the Hubbard model. v and n.222) .e.213) can simply be replaced by 1. (9. Fig. β (9. the necessary ˜ Since Emf [c.8: Macroscopic Systems 279 where θ(x) is the step function.218) The derivatives can be readily determined and the conditions (9.220) µ can be readily obtained µ= 1 ( 2 α + β) + vn .218) read ˜ ∂ Emf = ∂ α ˜ ∂ Emf = ∂ β and α + vnβ − µ = 0 .219–9. 2 (9.220) β ˜ ∂ Emf = n − nα − nβ = 0 . i. 9.213) The presence of the θ function in the above formula guaranties that the density of states vanishes for | | > 2t. One obtains ˜ Emf = α β ρ( )d + −2t −2t ρ( )d + vnα nβ + µ(n − nα − nβ ) . We can employ the results obtained to express the ground state energy density (9. µ.204) through an integral expression.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. one arrives at the following expression of the density of states ρ( ) = θ(2t − | |) π (2t)2 − 2 .f. outside the energy band (k).

the slope of f (∆) at the origin is larger than 1 [c. and only if. non trivial. Figure 9.e. First. as reﬂected by the expression d[f (∆)] = vρ( d∆ β + ∆) = π (2t)2 v − (v∆ + 2 β) >0. Indeed. from (9.f. For v above a critical value vc . (9. (9.226) one can obtain ∆ by solving (cf. (9. One still needs to determine α and β .215) reveals that µ is indeed equal to the Fermi energy F (see also Figure 9.227) will have a second.217) and (9.e.227) Since f (0) = 0 one can infer that ∆ = 0 is always a solution (the so called paramagnetic solution) of condition (9.224) v∆ ∆= β ρ( )d = 0 ρ( β + )d .280 Many–Particle Systems Comparision of (9.2b). For this purpose we consider the magnetization of the mean ﬁeld ground state ∆ ≡ nα − nβ . a magnetic ground state.223) follows α α − v β .225) Deﬁning the function v∆ f (∆) = 0 ρ( β + )d .220) and (9.227) can be solved graphically.223) ∆ = 0 corresponds to the ground state (9. the Hubbard model has a ground state with ∆ = 0. If the slope of f (∆) at the origin is less than 1. . so called ferromagnetic solution.183) without magnetization.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.223 one obtains ∆= and second. One can determine vc by equating the slope of f (∆) at the origin to 1.3]. from (9. condition (9. through the condition [c. Accordingly. (9.. i.228)] vc ρ ( β ) = 1 . The slope of f (∆) at the origin depends on the Hubbard model parameter v.228) Since the total number of states per site is ﬁnite. i.222) and (9. The Stoner criterion (9.219–9.226) ∆ = f (∆) . For this purpose one plots f (∆) versus ∆ as is shown schematically in Figure 9. (9.3. (9. We assume in the following discussion that the implicit ∆ dependence of β in f (∆) can be neglected.229) has a wider range of validity than one can infer from the present analysis of the Hubbard model. if. condition (9. the slope of f (∆) is positive for all ∆ ≥ 0.227). (9.. f (∆) converges to a maximum value as ∆ → ∆max = n. (9.226) follows that f (∆) is a monotonically increasing function of ∆. (9.f. One can express ∆ as a function of α and β in two diﬀerent ways.227) has only the trivial solution ∆ = 0. From the deﬁnition (9. Equation (9.

3: Graphical solution of the mean ﬁeld equation ∆ = f (∆).233) Employing the approximation sin(y + δ) − sin y ≈ δ cos y for small δ. n). 2t As already mentioned.220.8: Macroscopic Systems 281 Figure 9.219.231) (9. hence.205) the following set of equations determining x. t. the plot of vc vs.231) follows y= δ π π (n − 1) − ≈ (n − 1) . i. 2 2 2 (9..234) 2t 2 The resulting phase diagram of the ground state. 9. β x+y x−y = = π(n − 1) 2t π (sin x − sin y) = π∆ . Since ∆ vanishes near vc we set x = y + δ where δ is a small quantity. (9.233) yield the desired expression vc π = π cos (n − 1) .4. the magnetization ∆ of the ground state vanishes and remains zero for values v < vc .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. By parameterizing α and β α = 2t sin x . 9. is presented in Figure 9. Accordingly. We like to determine ﬁnally the dependence of vc on the ﬁlling factor n.9. (9.230) one obtains from (9. β = 2t sin y . From (9. the ground state is characterized by the quantities (v. The corresponding 2t analytical expression provides the phase diagram of the ground state: when v approaches vc from above. (9.e. y and. α. v (9. Equation (9.232) states that x − y is proportional to ∆. any .232) together with (9. n.

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

the strong eﬀect of quantum ﬂuctuations of the strongly correlated electron system described by the Hubbard Hamiltonian makes all the presently existing solutions questionable. a reliable microscopic theory is still lacking. In fact. and on the other hand. in the case of the copper oxide high temperature superconductors. . the lack of exact solutions. which involve strongly correlated quasi two-dimensional electron systems.9.8: Macroscopic Systems 283 In case of two spatial dimensions things are more complicated. On the one hand. The available mean ﬁeld theories cannot account for all the striking. unusual physical properties of these materials.

284 Many–Particle Systems .

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. 2 As mentioned. 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. spin– 1 particles. t) ∂ i ψ(r.2) will result. x2 .g. x3 = x3 (10.. t = t− 1 − v1 x c2 1 v1 2 c . t) in one frame with space time coordinates (x1 . The representation 285 . t) are vectors of dimension larger one. the components representing the spin attribute of particles and also representing together with a particle its anti-particle. t) (10.f. 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.e. Furthermore. t) c 2 i + qV (r. x3 . t ) in another frame. x2 . etc. v = v1 x1 .e. e.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. t) ψ(r. these ˆ transformations are x1 = x1 − v1 t 1 − v1 2 c . The transformations beween such frames according to the Theory of Special Relativity are described by Lorentz transformations. i. Obviously.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..1) which connect space time coordinates (x1 . It is not possible to uncouple the equations to describe only a single type particle without aﬀecting negatively the Lorentz invariance of the equations. Here c denotes the velocity of light. x3 .2) The replacement of (10. spin–0 particles. t) and a vector potential A(r. 8.45)] described by an electrical potential V (r. t) = ∂t 1 2m q − A(r. any of these equations being speciﬁc for certain classes of particles. some of the equations describe a particle together with its anti-particle. x2 = x2 . i. We will ﬁnd that actually several Lorentz–invariant equations which replace (10. (refeq:ham2. 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.. In case that v is oriented along the x1 –axis.

Rather than following this avenue we will introduce Lorentz transformations within a setting which does not require real and imaginary coordinates. i. t) as components. i. 10. but in a setting of representation theory methods as applied in Secti..7 a heuristic derivation of the two most widely used and best known Lorentz–invariant ﬁeld equations. 10. x 1 . the group of Lorentz transformations (10.e. x3 ) (10. This space is called the Minkowski space. x2 . We will ﬁnd that this deﬁnition will lead us back to the transformation law (10.286 Relativistic Quantum Mechanics in Sect.1 will be extended in Sect. This will provide us with a general set of Lorentz–invariant equations which for various particles take the place of the Schr¨dinger equation.4 to cover ﬁelds.1). t) and vectors with functions ψ(r.5) √ One can interprete the quantity −s2 as a distance in a 4–dimensional Euclidean space if one chooses the time component purely imaginary.4) invariant. Note that the components of xµ all have the same dimension.e. i. assume new units for time such that the velocity of light c becomes c = 1. henceforth. x 3 ) holds (x0 )2 − (x1 )2 − (x2 )2 − (x3 )2 = (x )2 − (x )2 − (x )2 − (x )2 . wave functions ψ(r. Minkowski Space Historically. The reason for this choice is that the transformations (10. namely the Klein–Gordon (Sect.5) and the Dirac (Sect. namely that of length. 10. x1 . The Group of Lorentz Transformations L = O(3.3) where x0 = ct describes the time coordinate and (x1 . 10. The elements L ∈ L act on 4–dimensional vectors of position– and time–coordinates. for the transformed space-time–cordinates x µ = (x 0 . 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. This choice implies dim(time) = dim(length). 0 1 2 3 (10.5.7) equation. x 2 . x2 . We will. Before introduco ing these general Lorentz–invariant ﬁeld equations we will provide in Sects. we will provide a more basic deﬁnition of the transformations. 10.. 5 to the groups SO(3) and SU(2) of rotation transformations of space coordinates and of spin.1). In such a space Lorentz transformations correspond to 4-dimensional rotations. 10.e.1 Natural Representation of the Lorentz Group In this Section we consider the natural representation of the Lorentz group L.1) The Lorentz transformations L describe the relationship between space-time coordinates xµ of two reference frames which move relative to each other with uniform ﬁxed velocity v and which might be reoriented relative to each other by a rotation around a common origin. We will denote these vectors as follows def xµ = (x0 . x3 ) = r describes the space coordinates. however.1) leave the quantity s2 = (x0 )2 − (x1 )2 − (x2 )2 − (x3 )2 (10. Rather than starting from (10. 10.1). Denoting by xµ the .

summation over that index is assumed without explicitly noting it.8) new old old The summation convention allows us to write (10.] The following possibilities exist for the positioning of the indices µ.7) The reason for the notation is two-fold. Aµ ν B ν ρ = ν=0 Aµ ν B ν ρ . (10. This will be explained further below.10.e. 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. Aµ ν xν = µ=0 new old Aµ ν xν .6) yields Lµ ρ gµν Lν σ xρ xσ = gρσ xρ xσ . ν = 0.e. The second reason is that upper and lower positions allow us to accomodate the expression (10.. the Lorentz transformations constitute a linear transformation which we denote by 3 x µ = ν=0 Lµ ν xν .11) This condition speciﬁes the key property of Lorentz transformations. (10. The subset of GL(4..6) Here Lµ ν are the elements of a 4 × 4–matrix representing the Lorentz transformation. We will exploit this property below to determine the general form of the Lorentz transformations. The Lorentz transformations form the subgroup of all matrices which leave the expression (10. Aµν . 4 × 4 tensor: Aµ ν .6) x µ = Lµ ν xν . 2. the latter set constituting a group.5) invariant. xµ . The Lorentz transformations are non-singular 4 × 4–matrices with real coeﬃcients. however. 3: 4-vector: xµ .6) allows us to introduce the so-called summation conventon: any time the same index appears in an upper and a lower position.12) (10. the latter notation will be used for diﬀerent quantities further below. 3 3 3 yµ xµ = new yµ xµ . i. R). the . [ A more conventional notation would be Lµν . the notation in (10. L ∈ GL(4. i.9) and (10. 0 0 0 −1 (10. First. ν=0 (10. This condition can be written xµ gµν xν = x gµν x where µ ν (10. Aµν .5) into scalar products.10) Combining condition (10. Since this holds for any xµ it must be true Lµ ρ gµν Lν σ = gρσ . (10.1: Natural Representation of the Lorentz Group 287 coordinates in one reference frame and by x µ the coordinates in the other reference frame. 1. Aµ ν .9) 1 0 0 0 −1 0 0 ( gµν ) = 0 0 −1 0 = g . R).

LT gL = g } .s. L2 ∈ O(3.288 Relativistic Quantum Mechanics elements of which satisfy this condition.1) ⊂ GL(4. i.1). From this one can obtain using g2 = 1 1 (gLT g)L = 1 and.20) results in the desired property (L−1 )T g L−1 = g .1). 1) = { L.12) LT gL = g ..14) L0 0 −L1 0 −L2 0 −L3 0 −L0 1 L1 1 L2 1 L3 1 . in fact. employing expressions (10. To simplify the following proof of the key group properties we like to adopt the conventional matrix notation for Lµ ν 0 L 0 L0 1 L0 2 L0 3 L1 L1 1 L1 2 L1 3 (10. (10. L−1 ∈ O(3.15) (10.17) one obtains (L−1 )T g L−1 = gLgggLT g = gLgLT g . (10.14) holds for the inverse of L.e..e.. the associated inverse obeys (10.e.22) (10.. .14).16. hence. 3 2 1 2 1 2 (10. L3 0 L3 1 L3 2 L3 3 Using the deﬁnition (10. is called O(3. R) is a group.19) i. (10.20) Multiplying (10.1).e.15) one can derive LT gLgLT = LT and multiplying this from the left by by g(LT )−1 yields L g LT = g Using this result to simplify the r.16) The corresponding expression for (LT )−1 is obviously (LT )−1 = (L−1 )T = g L g . i.1) since it satisﬁes (10. of (10. R).h.14).1).13) L = ( Lµ ν ) = 2 0 L 0 L2 1 L2 2 L3 3 . a group. This set is identical with the set of all Lorentz transformations L. We want to show now L = O(3. To demonstrate the group property of O(3. This stipulates that O(3. = g LT g = 0 1 2 −L 2 L 2 L 2 L3 2 −L0 3 L1 3 L2 3 L3 3 (10. For L3 = L1 L2 holds LT g L3 = LT LT g L1 L2 = LT (LT gL1 ) L2 = LT g L2 = g .10) of g one can rewrite the invariance property (10. the inverse of L 1 L−1 (10.14) from the right by gLT and using (10.1).1).1) is. We consider 1 then L1 .21) i. In fact. One can also show that if L ∈ O(3.18) (10. property (10.17) we note ﬁrst that the identity matrix 1 is an element of O(3. of O(3. L3 ∈ O(3. 10. L ∈ GL(4.

det L = −1. For this purpose we consider ﬁrst the value of det L. (10. (10.29) (10. 1). A statement on this value can be made on account of property (10.27) (10.14).14). Schwartz’s inequality yields j=1 2 3 3 3 j=1 A jB 0 ≤ 0 j A 0 2 j Bj 0 2 . 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 . (10.34) j=1 j=1 . L0 0 ≥ 1} . M1 = a M2 }. The above shows that the set of proper Lorentz transformations L↑ allows one to generate all Lorentz transformations.1: Natural Representation of the Lorentz Group Classiﬁcation of Lorentz Transformations 289 We like to classify now the elements of L = O(3. Properties (10. = { L. 1).33) where we used the deﬁnition aM = {M1 ..30) ≤ −1} . Lorentz transformations in L↑ . L0 0 ≥ 1} .24) or since (L1 0 )2 + (L2 0 )2 + (L3 0 )2 ≥ 0 it holds (L0 0 )2 ≥ 1.14) which we employ in the formulation (10. One can also 1 verify that one can write L↑ − L↓ + L↓ − = gL↑ + = = −L↑ . M2 ∈ M.e. L0 0 (10. A second class property follows from (10. L0 0 ≤ −1} . Considering in (10. Using det AB = det A det B and det AT = det A yields (det L)2 = 1 or det L = ±1 . 1). L↑ contains 1 1. L ∈ O(3. L ∈ O(3. ∃M2 . B ∈ L+ holds C = AB ∈ L+ . hence. there exist two other distinct classes. (10. = { L.10. entirely suitable to investigate ﬁrst only 1. + except for the trivial factors g and −1 It is. It holds g ∈ L and −1 ∈ L as one can readily verify testing for property (10. L↓ are disjunct sets deﬁned as follows + + − − L↑ + L↓ + L↑ − L↓ − = { L.31) (10.32) (10. L↑ . 1).12) the case ρ = 0. + − gL↑ + = L↑ g . L ∈ O(3. det L = −1. det L = 1.28) (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↑ .23) and (10. = { L. + + ↑ ↑ We can also demonstrate that for A. σ = 0 yields L0 0 2 − L1 0 2 − L2 0 2 − L3 0 2 = 1.1).12). + = − L↑ g + (10.23) One can classify Lorentz transformations according to the value of the determinant into two distinct classes. L↓ . L ∈ O(3. Obviously.26) (10. + We start our investigation by demonstrating that L↑ forms a group.25) i. det L = 1. From this we can conclude L0 0 ≥ 1 or L0 0 ≤ −1 .

obviously j=1 A0 0 B 0 0 ≥ 1.1) ascertains CT gC = g it must hold C 0 0 ≥ 1. .13).38) For these transformations. 1 0 0 = ≥ 1 and. 1 We will then employ the Lie group properties to generate all transformations in L↑ . holds L0 0 > 0 and the value of the determinant is close to unity. + i..1). We also note that the identity operator 1 has elements 1 + 1 µν = δµν 1 where we deﬁned1 δµν = 1 for µ = ν 0 for µ = ν (10.290 Relativistic Quantum Mechanics From (10. orthochronous Lorentz transformations.e.1). 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. µ ν small . To order O( 2 ) holds g + g = 0. accordingly.14) actually L0 0 ≥ 1 and det L = 1 must hold.14) implies 1 + T g (1 + ) = g 1 1 (10. + Accordingly.34) provides then the estimate conclude from (10. The next group property of L↑ to be demonstrated is the existence of the inverse.36) It holds.. from which one can conclude L−1 ∈ L↑ .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 .e. Using the above expression for C 0 0 one can state C 0 0 > 0. 1 ∈ L↑ . therefore. obviously. (10. Since the associative property holds for matrix multipli1 1 + ↑ cation we have veriﬁed that L+ is indeed a subgroup of SO(3. In fact.40) and. we consider transformations Lµ ν = δ µ ν + µ ν . + transformations in a small neighborhood of 1 which we parametrize by inﬁnitesimal parameters. since the group property of O(3. Property (10. For the inverse of + ↑ any L ∈ L+ holds (10. Similarly. (10.39) where we have employed the matrix form T 1 deﬁned as in (10. Since A0 0 ≥ 1 and B 0 0 ≥ 1. one can j=1 j=1 3 0 )2 = (A0 )2 − 1. In the following we + will consider solely this subgroup of SO(3.12) follows (B 0 0 )2 − 3 (B j 0 )2 = 1 or 3 (B j 0 )2 = (B 0 0 )2 − 1. | 3 A0 j B j 0 | < A0 0 B 0 0 . This relationship shows (L−1 )0 0 = L0 0 . (10. L↑ is called the subgroup of proper. i. hence. Inﬁnitesimal Lorentz transformations The transformations in L↑ have the property that they are continously connected to the identity 1 1.37) (10.35) One can conclude. ρ ν (10. if we enforce (10.16). δ00 = 1 and It should be noted that according to our present deﬁnition holds δµν = gµρ δ δ11 = δ22 = δ33 = −1.

j. 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 . 2.47) 0 −1 0 0 0 0 −1 0 0 −1 0 0 0 0 0 0 0 0 K1 = .43) k = − Inspection shows that the matrix has 6 independent elements and can be written 0 −w1 −w2 −w3 −w1 0 −ϑ3 ϑ2 . ϑ3 . 2.10.50) . = − k m Km (10. (10.42) This relationship implies µ 0 j µ j = 0 = j 0 . 3) Jk = (ϑk = 1. k j j = 1. w1 . k = 1. ϑ2 . K ] [ Jk . The generators are explicitly 0 0 0 0 0 0 0 0 J1 = 0 0 0 −1 0 0 1 0 (10.49) . 2. J ] [ Kk . J = 0 3 0 1 0 0 0 0 0 0 0 0 0 0 0 (10.45) (10. 3 (10. K ] The operators also obey J · K = J1 J1 + J2 J2 + J3 J3 = 0 = = k m Jm k m Jm 0 −1 0 0 0 0 0 0 (10. (ϑ1 . K2 = .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 .48) (10.15) one can conclude T 291 = −g g (10. 3 .1: Natural Representation of the Lorentz Group Using (10. other ﬁve parameters zero) .44) This result allows us now to deﬁne six generators for the Lorentz transformations(k = 1.46) 0 0 0 0 . J2 = 0 0 0 −1 0 0 0 0 0 0 0 0 0 0 0 −1 0 1 . w3 ) = −w2 ϑ3 0 −ϑ1 −w3 −ϑ2 ϑ1 0 (10. w2 .

52) where the 3 × 3–matrices Lk are the generators of SO(3) deﬁned in Chapter 5. k = 1. 2π[ .50). 3 (10.53) Here R(ϑ) are the 3 × 3–rotation matrices constructed in Chapter 5. that the transformations (10. ϑ..35) it issuﬃcient to consider in the present case the 2 × 2–matrix L = exp since w1 0 −1 −1 0 ∞ = n=0 n w1 n! 0 −1 −1 0 0 0 . (10. 3 is closed. and using the relationship Jk = 0 0 0 Lk (10. + The commutation relationships imply that the algebra of the generators Jk . w = 0) = 0 0 0 R(ϑ) . 3 where we have deﬁned ϑ · J = k=1 ϑk Jk and w · K = following the algebra in Chapter 5. 2. L(ϑ. Relativistic Quantum Mechanics The commutation relationships (10. A boost in the x1 –direction is L = exp(w1 K1 ). however. For the parameters ϑk of the Lorentz transformations holds obviously ϑk ∈ [0. k = 1.51) for w = 0 correspond to rotations of the spatial coordinates. To determine the explicit form of this transformation we evaluate the exponential operator by Taylor expansion. 10. w ∈ R3 3 k=1 wk Kk . In analogy to equation (5. Exercise 7. 0 1 n (10. 2.57) . constitutes an overcomplete parametrization of the rotations (see Chapter 5). (10. w) = exp ϑ · J + w · K .49.51) One can readily show. Kk . i. We consider now the Lorentz transformations for ϑ = 0 which are referred to as ‘boosts’.292 as can be readily veriﬁed.56) = 1 0 0 1 = 1 1 (10.1: Demonstrate the commutation relationships (10.e. Following the treatment of the rotation group SO(3) one can express the elements of L↑ through + the exponential operators L(ϑ.49) deﬁne the Lie algebra associated with the Lie group L↑ .55) L 0 0 0 0 exp (w1 K1 ) = exp Using the idempotence property 0 −1 −1 0 2 0 0 1 0 (10.54) which.

for wk wk ∈ ] − ∞. We expect that vk can only assume values less than the velocity of light c which in the present units is c = 1. (10.58) = cosh w1 1 + sinh w1 1 0 −1 −1 0 The conventional form (10. From (10.1: Natural Representation of the Lorentz Group one can carry out the Taylor expansion above: ∞ 293 L = n=0 2n w1 1 + 1 (2n)! ∞ n=0 2n+1 w1 (2n + 1)! 0 −1 −1 0 = cosh w1 −sinh w1 −sinh w1 cosh w1 . (10. cosh2 w1 − 1 and coshw1 = cosh2 w1 − 1 = cosh w1 sinh w1 sinh2 w1 + 1 . This property of the wk -values contrasts with the property of the parameters ϑk specifying rotational angles which assume only values in a compact range. Accordingly. (10. consistent with (10.1) of the Lorentz transformations is obtained through the parameter change sinh w1 v1 = = tanh w1 (10.59) cosh w1 Using cosh2 w1 − sinh2 w1 = 1 one can identify sinhw1 = sinh2 w1 + 1. we can state vk ∈ ] − 1. t = t − v1 x1 2 1 − v1 . x2 = x2 . (10.59) follows. 1[. The range of the parameters wk can now be speciﬁed.63) which agrees with (10.1). ∞[ . Correspondingly. 10.59) v1 = These two equations yield cosh w1 = 1/ and (10.51) the explicit transformation for space–time–coordinates is then x1 = x1 − v1 t 1 − 2 v1 .59) for the case k = 1 corresponds to the relative velocity of two frames of reference. however. x3 = x3 (10.62) According to (10.61) √ √−v1 √ exp (w1 K1 ) = 2 1 − v1 1 2 1 − v1 0 0 0 0 0 0 0 0 1 0 0 1 (10. in fact. . This property is.60) sinh w1 = v1 / 2 1 − v1 . 10. one obtains from (10.59) can be written 1 2 1 − v1 √−v1 2 1 − v1 2 1 − v1 .3.6.64) We note that the range of wk -values is not a compact set even though the range of vk -values is compact.10. vk deﬁned in (10. 10.56.59).

scalars like r = x2 + x2 + x2 . We will see below how true scalars under Lorentz transformations can be constructed. not in the so-called natural representation associated with the 3–dimensional Euclidean space E3 . tensor operators Tkm . we have not yet deﬁned representations of Lorentz transformations beyond the ‘natural’ representation acting in the 4–dimensional space of position– and time–coordinates. a2 .2 Scalars. t) and J(r. Presently. namely.66) Examples of 4-vectors beside xµ are the momentum 4-vector pµ = (E. Nevertheless. (10. ﬁnally.3) are the so-called 4–vectors aµ . Scalars The quantities with the simplest transformation behaviour are so-called scalars f ∈ R which are invariant under transformations. J) (10.e. Counterexamples are the energy. 4-Vectors The quantities with the transformation behaviour like that of the position–time vector xµ deﬁned in (10.4). t) and A(r. not any physical property f ∈ R is a scalar.69) where V (r. . x2 . However. Some of these r ˆ quantities were actually deﬁned with respect to representations of the rotation group in function spaces..65) An example is s2 deﬁned in (10.68) where ρ(r. 2 |ˆ1 . we will encounter an impressive example of physical properties. t) are the electrical and the vector potential of an electromagnetic ﬁeld. the square of the electric ﬁeld |E(r. (10. r2 ) and. The 4-vector character of J µ and of Aµ will be demonstrated further below. t)|2 or the scalar product r1 · r2 of two particle positions. vectors like r = (x1 . the charge density.67) the transformation behaviour of which we will demonstrate further below.294 Relativistic Quantum Mechanics 10. f = f. of a system of charges. Hence. t) are the charge density and the current density. total angular momentum r 1 2 3 states of composite systems like Y m ( 1 . Such quantities appear in the description of physical systems and statements about transformation properties are often extremely helpful and usually provide important physical insight. A) (10. Another example is the potential 4-vector Aµ = (V. 4–Vectors and Tensors In this Section we deﬁne quantities according to their behaviour under Lorentz transformations. i. a3 )T and transform according to a µ = Lµ ν aν . E = √ m mv . A third 4-vector is the so-called current vector J µ = (ρ. a1 . p) . the z–component x3 of a particle. These quantites always come as four components (a0 . spherical harmonics Y m (ˆ). x3 )T . another example is the rest mass m of a particle. our deﬁnition of quantities with special properties under Lorentz transformations presently is conﬁned to the natural representation. respectively. We have encountered examples in connection with rotational transformations. p = √ 2 1−v 1 − v2 (10.

17) can be written (L−1 )ν µ = Lµ ν .73) Note that according to (10. If aµ and bµ are 4-vectors aµ gµν bν (10.71) Contravariant and Covariant 4-Vectors It is convenient to deﬁne a second class of 4-vectors.17) Lµ σ is the transformation inverse to Lσ µ . lower. In fact.. aµ (10. one can employ the tensors g µν to raise and lower indices of L ν establish here the consistency of the ensuing notation.72) where aν is a vector with transformation behaviour as stated in (10.12). In any case one can express (10. and . (10.74) g µν aν .78) To prove the 4-vector property of ∂µ we will show that g µν ∂ν transforms like a contravariant 4vector. −a2 . xν = g νσ Lρ σ gρµ x µ . (10.75) (10. accordingly.16). We do not fact.79) This is the inverse Lorentz transformation consistent with (10. using (10. We have duplicated the expression for the inverse of Lµ ν to obtain the correct notation in terms of covariant. This property follows from (10. Covariant 4-vectors transform like a µ = gµν Lν ρ g ρσ aσ where we deﬁned g µν = gµν . 4–Vectors and Tensors Scalar Product then 295 4-vectors allow one to construct scalar quantities.e.10.72) of the covariant 4-vector follows = In µν and g µ as well. i. ∂x µ (10.2: Scalars. ∂t (10.66). We like to point out that from deﬁnition (10.77) The transformed diﬀerential operator will be denoted by ∂µ = def ∂ . One calls 4-vectors aµ covariant and 4-vectors aµ contravariant. The respective vectors aµ are associated with the 4-vectors aµ . The 4-Vector ∂µ (10.12) a gµν b µ ν = Lµ ρ gµν Lν σ aρ bσ = aρ gρσ bσ (10. Multiplication (and summation) of x µ = Lµ ν xν by Lρ σ gρµ yields. the relationship being aµ = gµν aν = (a0 .73) aµ = Lµ σ aσ .76) An important example of a covariant 4-vector is the diﬀerential operator ∂µ = ∂ = ∂xµ ∂ .70) is a scalar. g µν ∂ν = Lµ ρ g ρσ ∂σ . i. gσν xν = Lρ σ gρµ x µ and g µσ gσν = δ µ ν . −a1 .66) together with (10. one can express [(L−1 )T ]µ ν = (L−1 )ν µ and.e.. (10. We start from x µ = Lµ ν xν . −a3 ) (10.

67) transforms like (10. upper.g. d’Alembert Operator We want to construct now a scalar diﬀerential operator. For this purpose we consider the scalar diﬀerential (dτ )2 = dxµ dxµ = (dt)2 − (dr)2 It holds dτ dt from which follows 2 (10.86) One can write (10. i. e. 2 2 dt 1 − v 1 − v (10. this operator is equal to the d’Alembert operator which is known to be Lorentz-invariant. indices. dτ 1 − v 2 dt p0 = E = √ m m dt = √ ..85) 1 d d = √ .83) is a scalar under Lorentz transformations.82) (10.− ∂t .81) ∂ ..66). 2 dt dτ 1 − v (10. i.e.89) .88) One can express then the momentum vector pµ = √ d m dxµ = m xµ . For this purpose we deﬁne ﬁrst the contravariant diﬀerential operator ∂ µ = g µν ∂ν = Then the operator 2 ∂µ ∂ µ = ∂t − 2 (10. (10. In fact. 1 − v2 1 − v 2 dt (10.e.. Proof that pµ is a 4-vector We will demonstrate now that the momentum 4-vector pµ deﬁned in (10.79) allows us to determine the connection between ∂µ and ∂µ . (10. ∂µ does indeed transform like a covariant vector. 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. p1 = √ m v1 m dx1 = √ .87) The remaining components of pµ can be written.80) Multiplication by g λµ (and summation over µ) together with g λµ gρµ = δ λ ρ yields g λµ ∂µ = Lλ σ g σν ∂ν .84) = 1 − (v)2 (10.296 Relativistic Quantum Mechanics contravariant.

We will demonstrate now the 4-vector property of Aµ .s. t) = 0 .90) (10. m2 m2 v 2 − = m2 1 − v2 1 − v2 (10.92) (10. namely pµ pµ = pµ gµν pν = E 2 − p 2 Evaluation of the r. · A = 0.94) 1 m In the non-relativistic limit the rest energy m is the dominant contribution to E.. the r. of (10.89) must be a 4-vector. in fact. (10.93) (10. pµ on the l.s. of (10. a contravariant 4-vector then the l.77) which allow one to express (10.s. kinetic energy ± 2m and relativistic corrections.e. This gauge is not Lorentz-invariant and we will adopt here another gauge. namely. 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. t) + · A(r. Expansion in should then be rapidly convergent. . Since xµ transforms like a contravariant 4-vector. i. (10.h.3: Relativistic Electrodynamics 297 d The operator m dτ transforms like a scalar. respectively.96) in the form ∂µ Aµ = 0 .97) and. and.h.69) for the electrodynamic potential and the 4-vector derivative (10.96) The Lorentz-invariance of this gauge. We like to rewrite the last result E 2 = p 2 + m2 or E = ± p 2 + m2 . t) and A(r. 10.97) We have proven already that ∂µ is a contravariant 4-vector.12). Lorentz Gauge In our previous description of the electrodynamic ﬁeld we had introduced the scalar and vector potential V (r.s. The momentum 4-vector allows us to construct a scalar quantity. in (10. ∂t V (r. Lorentz-invariant. is a scalar.h. in (10. 8. equivalently.69) is. hence. hence. can be demonstrated readily using the 4-vector notation (10.h. for these potentials. (10. yields according to (10. t).95) 2 p This obviously describes the energy of a free particle with rest energy ±m.67) E2 − p 2 = or pµ pµ = m2 which. in fact. If we can show that Aµ deﬁned in (10.89) alltogether transforms like a contravariant 4-vector.91) (10.10. and had chosen the so-called Coulomb gauge (8.96) is a scalar and. the so-called Lorentz gauge.

105) .101). t) = ∂t · A(r. m. ∂µ J µ (xµ ) = 0 . · A(r. t) = −∂t V (r.103) B .98) which reﬂects the principle of charge conservation. One can readily verify.83) and (10. t) = −∂t V (r. n = 1.13).82) as follows F µν = ∂ µ Aν − ∂ ν Aµ . Using · ∂t A(r. transforms like a 4-vector.96). according to (10.17) using the identity (8. i.h. t) .6) and (8.101) Combining equations (10. i.77) and (10. (8.96). 10. t) − 2 V (r.102) In this equation the r. t) and current density J(r. This principle should hold in any frame of reference. one obtains for A(r. k. t) = 0 (10.69).e. 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 . t) = 4πρ(r.298 Transformation Properties of J µ and Aµ Relativistic Quantum Mechanics The charge density ρ(r.h. thereby. (10. (10. (10.s. The l. using (10.32).99) must be a scalar. Since ∂µ transforms like a covariant 4-vector.100) Similarly. Consequently. t) − 2 A(r.h. t) from (8.. must transform likewise. 3 (10.s. Since ∂µ ∂ µ transforms like a scalar one can conclude that Aν (xσ ) must transform like a 4-vector. also the l.100. t) together with (10. We want to derive now the diﬀerential equations which determine the 4-potential Aµ in the Lorentz gauge (10. t) 2 ∂t A(r. t) = 4 π J(r.104) where mn = mn is the totally anti-symmetric three-dimensional tensor deﬁned in (5. has to transform like a contravariant 4-vector. must be a scalar.98) can be written. 2. F µν = x Ey B z 0 −Bx Ez −By Bx 0 Alternatively. that F µν can be expressed through the potential Aµ in (10.s. t) are known to obey the continuity property ∂t ρ(r.69) and ∂ µ in (10. t).97) and. Equation (10. · A(r. (10. this can be stated F k0 = −F 0k = E k . using (8. in fact.e.99) Since this equation must be true in any frame of reference the right hand side must vanish in all frames. using (10. t) + · J(r. prove that Aµ is. The respective equation for A0 = V can be obtained from Eq.18) and. it follows that J µ . in fact. (10.68). yields ∂µ ∂ µ Aν (xσ ) = 4 π J ν (xσ ) .9). F mn = − mn (10. . a 4-vector.. one obtains 2 ∂t V (r. of (10. t) .

102) ∂µ F µν = 4π J ν . In a new frame of reference holds F µν = Lµ α Lν β F αβ (10.107) 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 (10.106) In case that the Lorentz transformation Lµ ν is given by (10. E⊥ = B⊥ E⊥ + v × B √ 1 − v2 B⊥ − v × E = √ 1 − v2 (10. one can conclude from (10. (10. Noting ∂µ F µν = ∂µ ∂ µ Aν − ∂µ ∂ ν Aµ = ∂µ ∂ µ Aν − ∂ ν ∂µ Aµ = ∂µ ∂ µ Aν .4).97).112) (10.108) yields then the expressions for the transformed ﬁelds E and B . The results can be put into the more general form E B = = E . B and E⊥ .103.63).113) which can be shown to be equivalent to the two homogeneous Maxwell equations (8. 8. Maxwell Equations in Lorentz-Invariant Form One can express the Maxwell equations in terms of the tensor F µν in Lorentz-invariant form.111) One can readily prove that this equation is equivalent to the two inhomogeneous Maxwell equations (8. 8.1.3. B . where we used (10.10.105) and (10. t).109) (10. 10. B⊥ are. respectively. .105) of the tensor F µν one can conclude the property ∂ σ F µν + ∂ µ F νσ + ∂ ν F σµ = 0 (10.3: Relativistic Electrodynamics 299 The relationships (10. These equations show that under Lorentz transformations electric and magnetic ﬁelds convert into one another. From the deﬁnition (10. t) and B(r.2).110) where E . by (10. one obtains Ey −v B Ez +v B 0 −Ex − √ 1 2z − √ 1 2y 1−v1 1−v1 Bz −v1 Ey By +v1 Ez √ 2 Ex 0 − √ 2 1−v1 1−v1 F µν = E −v B (10.104) establishe the transformation behaviour of E(r. equivalently. the components of the ﬁelds parallel and perpendicular to the velocity v which determines the Lorentz transformation.62) or.

e. .114) reads then.104). using (10.116) (10. Equation (10. We have.120) (10.103) dpx q = ( EEx + py Bz − pz By ) .117) (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.114) is. dτ m √ Employing again (10. dt E From this one can conclude.86).5).93).119) which is the x-component of the equation of motion (8.g.114) reads using (10. for µ = 1. dE q = p·E dτ m or with (10.67) and charge q dpµ q = pν F µν dτ m (10.86) dE q = p·E .67).87)] and retain the deﬁnition E for the electric ﬁeld. 1 dE 2 1 dp 2 = = qp·E 2 dt 2 dt (10.86) and (10. (10. It holds for a particle with 4-momentum pµ as given by (10. provides an alternative description of the action of the Lorentz force.120) is referred to as the Lorentz force. yields dpx = q dt Ex + (v × B)x (10. equivalent to (8. For the spatial components.115) This equation follows.114). We want to express this force through the tensor F µν ..h.s. i. hence. hence. E = m/ 1 − v 2 . however. employing (10.5).e.118) p· dt √ where we exploited the fact that according to p = mv/ 1 − v 2 holds p v. in fact. We want to √ demonstrate now that this equation is equivalent to the equation of motion (8. To avoid √ confusion we will employ in the following for the energy of the particle the notation E = m/ 1 − v 2 [see (10. of (10. The term on the r. demonstrated that (10. also from the equation of motion (8.5) taking the scalar product with p dp = qp · E (10. The µ = 0 component of (10.5) where p = mv/ 1 − v 2 ..114) where d/dτ is given by (10.

10. However. in the present case we exclude the factor ‘−i’ [cf. We like to express now ψ (x µ ) in terms of the old coordinates xµ .51) for the natural representation of the Lorentz group. 10..125) def (10. i.g.48) of SO(3).e. and which can be constructed following the procedure adopted for the function space representation of SO(3).124) In this expression J = (J1 . in as far as SO(3.125)]. i. w) = exp ϑ · J + w · K .7.47). def (10. J3 ) and K = (K1 . J2 .1) is a generalization (rotation + boosts) of the group SO(3).e.121) by (L−1 )µ ν xν and obtains ψ (xµ ) = ψ((L−1 )µ ν xν ) . 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. K2 .3 or the bi-spinor wave function of electron-positron pairs in Sect.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 ) .42) of SO(3). 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. ψ ∈ C∞ (4). w) which correspond to the generators Jk and Kk in (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. w) = ρ(Lµ ν (ϑ. For such functions holds in the transformed frame ψ (Lµ ν xν ) = ψ(xµ ) (10. the charge-current density in case of Sect. We will denote the intended representation by L(ϑ.48) and (10. The resulting expression should be a generalization of the function space representation (5. We need to emphasize that (10. taken at the pairs of points (x µ = Lµ ν xν .. we seek an expression which corresponds to (10. For this purpose one replaces xµ in (10. e. (10. (5. w.4: Function Space Representation of Lorentz Group 301 10.127) .10. (10. K3 ) are the generators of L(ϑ. Accordingly.123) This deﬁnition corresponds closely to the function space representation (5. Functions which represent 4-vectorsor other non-scalar entities. xµ ).121) covers solely the transformation behaviour of scalar functions. obey a diﬀerent transformation law. w)) = ρ eϑ·J + w·K which we present in the form L(ϑ.. (10. In this section we will investigate the transformation properties of scalar functions ψ(xµ ).126) ρ ewk Kk − 1 1 .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 .

K obey the same Lie algebra (10.128) eϑ1 J1 = e−ϑ1 J1 1 0 = 0 0 0 0 0 1 0 0 0 cosϑ1 sinϑ1 0 −sinϑ1 cosϑ1 (10. One can determine similarly K1 starting from coshw1 sinhw1 sinhw1 coshw1 = 0 0 0 0 0 0 1 0 0 0 . which we like to demonstrate for J1 and K1 . the expression for K1 in (10. reproduces the expression for J1 in (10. The generators J . K ] = = k m Jm k m Jm = − k m Km .132) and. x3 ) 2 = ψ(xµ ) + w1 (x1 ∂0 + x0 ∂1 ) ψ(xµ ) + O(w1 ) (10. [ Jk . In order to evaluate (10. i.129) which yields for small ϑ1 ρ eϑ1 J1 ψ(xµ ) = ψ(x0 .133) We demonstrate this for three cases.130) ew1 K1 −1 = e−w1 K1 (10. and [J1 . cosϑ1 x2 + sinϑ1 x3 .e.134) . x3 ∂1 ] = −x1 ∂2 + x2 ∂1 = J3 . K2 ] = K3 : [ J1 . J3 = x2 ∂1 − x1 ∂2 . namely [J1 . −sinϑ1 x2 + cosϑ1 x3 ) = ψ(xµ ) + ϑ1 (x3 ∂2 − x2 ∂3 ) ψ(xµ ) + O(ϑ2 ) . [K1 . x2 . x1 ∂3 − x3 ∂1 ] = [x3 ∂2 .128). J2 ] = J3 .126). J2 = x1 ∂3 − x3 ∂1 . K ] [ Jk . 1 This result. (10.49) as the generators of the natural representation.302 One obtains J1 = x3 ∂2 − x2 ∂3 . (10. J2 ] = [x3 ∂2 − x2 ∂3 . sinhw1 x0 + coshw1 x1 .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. J ] [ Kk . x1 .131) This yields for small w1 ρ ew1 K1 ψ(xµ ) = ψ(coshw1 x0 + sinhw1 x1 . obviously. 0 1 (10. obviously. K2 ] = −J3 . x1 ∂3 ] − [x2 ∂3 .

The relationships (10. In both cases the radial coordinate is the quantity conserved under the transformations. Eqs. a relationship which can also be stated R = and + − (x0 )2 − (x3 )2 (x0 )2 − (x3 )2 if x0 ≥ 0 if x0 < 0 (10. 0 ∂cothΩ ∂x x = ∂R x0 = − ∂x3 R (10. x0 ∂2 ] = −x2 ∂1 + x1 ∂2 = − J3 . x0 ∂2 + x2 ∂0 ] = [x3 ∂2 .10.141) (10. Ω which are connected with x0 .87)]. [ J1 . One-Dimensional Function Space Representation The exponential operator (10. cothΩ = 3 .138) x3 x0 ..e.137) where K3 is given in (10.139) x3 = R sinhΩ (10. x0 ∂2 + x2 ∂0 ] = [x0 ∂1 . 303 (10.125) in the case of a one-dimensional transformation of the type L(w3 ) = exp w3 K3 . i. x2 ∂0 ] − [x1 ∂0 .139. (5.4: Function Space Representation of Lorentz Group [ K1 . can be simpliﬁed considerably.135) (10. x3 as follows x0 = R coshΩ . We note tanhΩ = ∂R x0 = .85-5. K2 ] = [x3 ∂2 − x2 ∂3 .143) . For this purpose one expresses K3 in terms of hyperbolic coordinates R. ∂Ω .137).128). 3 ∂R 3 ∂Ω ∂x ∂x R ∂R x ∂Ω = ∂Ω ∂tanhΩ 1 = cosh2 Ω 0 3 ∂tanhΩ ∂x x ∂Ω ∂cothΩ 1 = = − sinh2 Ω 3 . (10. In the following we consider solely the case x0 ≥ 0. x0 ∂2 ] = x3 ∂0 + x0 ∂3 = K3 .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. x2 ∂0 ] − [x2 ∂3 . ∂3 in terms of ∂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 .140) allow one to ∂ ∂ express the derivatives ∂0 . K2 ] = [x0 ∂1 + x1 ∂0 .136) (10.142) (10. x2 + x2 + x2 1 2 3 in the case of SO(3) and (x0 )2 − (x3 )2 in case of transformation (10.

145) ∂Ω 10.151) reads ∂µ ∂ µ + m2 ψ(xµ ) = 0 (10. is not very insightful. − ) = i ∂ µ . in 4-vector notation reads pµ =⇒ pµ = i (∂t . The historic route to these two equations. ) = i ∂µ . therefore. We will provide a ‘derivation’ of these two equations which stem from the historical development of relativistic quantum mechanics. extremely useful. (length)−1 . In these units the quantities energy. In natural units the Klein–Gordon equation (10. ∂Ω Relativistic Quantum Mechanics (10.92) one obtains the wave equation − or 2 2 µ ∂ ∂µ ψ(xν ) = m2 ψ(xν ) ψ(xν ) = 0 .122). mass. however.150) . and (time)−1 all have the same dimension. The latter property implies that upon change of reference frame ψ(xµ ) transforms according to (10. Further below we will provide a more systematic.121. ˆ pµ =⇒ pµ = i(∂t . (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 ) .128) yields K3 = x0 ∂3 + x3 ∂0 = ∂ . Free Particle Case A quantum mechanical description of a relativistic free particle results from applying the correspondence principle. but certainly is short and.147) (10. complex-valued function. (10. representation theoretic treatment.144) The action of the exponential operator (10.304 Inserting these results into the deﬁnition of K3 in (10. 10.148) ∂ µ ∂µ + m2 (10. In the position representation the correspondence principle states ˆ E =⇒ E = − ∂t i ˆ p =⇒ p = i which. In the following we will employ so-called natural units = c = 1. momentum.151) is called the Klein-Gordon equation.146) Applying the correspondence principle to (10. which allows one to replace classical observables by quantum mechanical operators acting on wave functions. The partial diﬀerential equation (10. namely the Klein–Gordon and the Dirac equations.149) where ψ(xµ ) is a scalar. ˆ (10.5 Klein–Gordon Equation In the following Sections we will provide a heuristic derivation of the two most widely used quantum mechanical descriptions in the relativistic regime.

t) jKG (r. t) ψ(r. In order to obtain the conservation law connected with the Klein–Gordon equation (10. t) ) 2m 1 = ( ψ ∗ (r. This follows from the fact that ∂µ ∂ µ and m2 are scalars.150) becomes ∂µ ∂ µ + m2 ψ (x µ ) = 0 (10. t) ψ(r. and that (as postulated) ψ(xµ ) is a scalar. t) ] 2mi (10.158) which yields 2 2 ψ ∗ ∂t ψ − ψ∂t ψ ∗ − ψ ∗ 2ψ +ψ ψ∗ − 2ψ∗ = ψ) = 0 (10.157) 2m which is equivalent to (10. (10. t) ψ ∗ (r. t) = · jKG (r. Under Lorentz transformations the free particle Klein–Gordon equation (10. t)∂t ψ ∗ (r. t) 1 [ ψ ∗ (r. t)∂t ψ(r. t) + where ρKG (r.155) describes the positive deﬁnite probability to detect a particle at position r at time t and where jS (r.159) (10.160) ∂t (ψ ∗ ∂t ψ which corresponds to − ψ∂t ψ∗) + · (ψ ψ∗ ∂t ρKG (r.161) . t) (10. 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. 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 (10.154) describes the current density connected with motion of the particle probability distribution. t) = − 1 2m 2 ψ(r.153) is associated with a conservation law for particle probability ∂t ρS (r. t) = 0 (10. t) = (10. t) = ψ ∗ (r.152) which has the same form as the Klein–Gordon equation in the original frame.5: Klein–Gordon Equation or 2 ∂t − 2 305 + m2 ψ(xµ ) = 0 .150) is invariant under Lorentz transformations. t) − ψ(r.154). t) ψ ∗ (r.150) one considers Im ψ ∗ ∂µ ∂ µ + m2 ψ = 0 (10.151) One can notice immeadiately that (10. 2mi (10.156) · jS (r. t) + where ρS (r. t) ) .10. t) − ψ(r. t) ψ(r. t) − ψ(r. t) = 0 i ( ψ ∗ (r.

t) = 0 (10.93] dispersion relationship connecting E0 . corresponding to λ = −.151) yields 2 2 Eo − p0 − m2 ψ(r. the Klein-Gordon equation is considered as a suitable equation to describe spin–0 particles. with p = 0.161) and (10.154). (10. λ|xµ )ψo (p.e.167) m which is positive for λ = + and negative for λ = −.150) are actually determined by po and by the choice of sign ±.163) which results in the expected [see (10. and of negative energies E ≤ −m. λ = ± (10. λ = ± . The proper interpretation of ρKG (r. not of particle probability. λ|xµ ) according to (10.164) The corresponding energy is Eo (po . t).166) is Eo (p) ∗ ρKG (p. in that the expression for ρKG is not positive deﬁnite. We denote this by summarizing the solutions as follows ∂µ ∂ µ + m2 ψo (p. Generating a Solution Through Lorentz Transformation A particle at rest. for example pions. λ|xµ ) = 0 2 m2 + po . λ) actually describes charge density rather than probability. λ|xµ ) = Nλ.162) Inserting this into the Klein–Gordon equation (10. according to (??) is decribed by the r–independent wave function ψo (p = 0. it had been realized later.. (10. corresponding to λ = +. ±) = ± 2 m2 + po (10.p ei(p·r − λEo (p)t) Eo (p) = The spectrum of the Klein–Gordon equation (10.306 Relativistic Quantum Mechanics This conservation law diﬀers in one important aspect from that of the Schr¨dinger equation (10. The proper interpretation of the two cases is that the Klein–Gordon equation describes particles as well as anti-particles.165) This result together with (10. is actually that of a charge density. (10. Today.162) shows that the solutions of the free particle Klein-Gordon equation (10. p0 .150) is a continuum of positive energies E ≥ m. λ) = λ ψo (p.168) . m 2 2 E0 = m2 + po . 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. o namely. The density ρKG (p. the anti-particles carry a charge opposite to that of the associated particles. Solution of the Free Particle Klein–Gordon Equation Solutions of the free particle Klein–Gordon equation are ψ(xµ ) = N e−ipµ x µ = N ei(p0 ·r − Eo t) . λ|xµ ) (10. λ|xµ ) = N e−iλmt . λ) associated with the corresponding wave functions ψo (p. i. and the density ρKG (p.166) ψo (p.

172) as the energy [c. λ|xµ ) = = exp w3 ∂ ∂Ω N e−iλmRcoshΩ 3) N e−iλmRcosh(Ω+w . λ = −|xµ ) = N ei(px 3 + Eo (p)x0 . (10. 2 2 1−v 1−v (10.177) .e. for the wave function (10. in fact.10.174) which agrees with the expression given in (10. for negative energy solutions.168). λ = +|xµ ) = N ei(px 3 − Eo (p)x0 (10.f. Aµ (xµ ) = (V (r.e. Ω. using (10.61) yield for this expression −iλ √ m mv x0 − iλ √ x3 .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. one has to interprete p = mv/ 1 − v 2 (10.145) choosing m = sinhw3 .169) The addition theorem of hyperbolic functions cosh(Ω + w3 ) = coshΩ coshw3 + sinhΩ sinhw3 allows us to rewrite the exponent on the r..h. (10. t).170) The coordinate transformation (10. i.166). For this purpose we consider the wave function for a particle moving with momentum velocity v in the direction of the x3 –axis.168). t)) (10.. −A(r.175) as the momentum of the particle and one obtains L(w3 )ψo (p = 0.171) One can interpret then for λ = +. i.166)] of the particle. L(w3 )ψo (p = 0. (10. for positive energy solutions. 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. (10. In case of λ = + one obtains ﬁnally L(w3 )ψo (p = 0.169) −iλ ( m coshw3 ) ( R coshΩ ) − iλ ( m sinhw3 ) ( R sinhΩ ) . A(r. In case of λ = −. t). t)) . of (10.s.138) and the relationships (10.6: Klein–Gordon Equation with Electromagnetic Field 307 We want to demonstrate now that the wave functions for p = 0 in (??) can be obtained through appropriate Lorentz transformation of (10.173) (10.176) 10.138) to replace t = x0 by hyperbolic coordinates R. Such wave function should be generated by applying the Lorentz transformation in the function space representation p (10. This yields.

t) 2 (10. i∂t and −i in (10.1. albeit in a form. A) free quantum particle quantum particle in ﬁeld (V. m2 ) ψ(xµ ) = 0 . momenta for classical particles and the respective operators for quantum mechanical particles in the manner shown.147)..e.180) + m2 ψ(r. is much larger than all other energy terms. in . A) or Aµ = (V. t))2 ψ(r.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 − .93).178) According to the replacements in Table 10. we deﬁne ψ(r.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. According to the so-called minimum coupling principle the presence of the ﬁeld is accounted for by altering energy.. To obtain the appropriate wave equation we follow the derivation of the free particle Klein–Gordon equation above and apply again the correspondence principle to (10. (10.150). i. (10. (10. (10.e. t) = e−imt Ψ(r. −A). t).179) (10. i.e. For this purpose one writes the Klein-Gordon equation (10. For this purpose we split-oﬀ a factor exp(−imt) which describes the oscillations of the wave function due to the rest energy. t) .308 free classical particle classical particle in ﬁeld (V.1: Coupling of a particle of charge q to an electromagnetic ﬁeld described by the 4-vector potential Aµ = (V. in keeping withnthe non-relativistic limit. which couples a particle of charge q to an electromagnetic ﬁeld described through the potential Aµ (xν ).. it’s rest energy. and focus on the remaining part of the wave function. t) = −i − q A(r.182) and seek an equation for Ψ(r. We will also assume. that the mass m of the particle.69)] one replaces the quantum mechanical operators.150) −g µν (i∂µ )(i∂ν ) + m2 ψ(xµ ) = 0 . t) . i. In terms of space-time derivatives this reads (i∂t − qV (r. See also Eq. according to the rules shown in Table 10. According to the principle of minimal coupling [see (10. A) Relativistic Quantum Mechanics energy momentum 4-vector E p pµ E − qV p − qA pµ − qAµ i∂t ˆ p = −i i∂µ i∂t − qV ˆ ˆ p − qA = π i∂µ − qAµ = πµ Table 10.

t) = − Ze and A(r. (10.181) reads then i ∂t Ψ(r. ﬁltered out of the beam and ﬁnally fall as slow pions onto elements for which a pionic variant is to be studied. t) 2m Ψ(r. | i∂t Ψ | << m .6: Klein–Gordon Equation with Electromagnetic Field particular. Ψ |q V | << m . spin-0 mesons. To ‘manufacture’ pionic atoms. t)]2 + qV (r. The Klein-Gordon equation (10. i.e.g.s.183). This application demonstrates that the Klein–Gordon equation describes spin zero particles.184) where we neglected all terms which did not contain factors m. (10.. t) ≡ 0 r + Ze2 r 2 + 2 − m2 π ϕ(r ) = 0 . larger than |i∂t Ψ/Ψ| and alrger than qV .183) The term on the l. 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− . The process of π − meson capture involves the so-called Auger eﬀect. i. (10.. The approximation is justiﬁed on the ground of the inequalities (10. of (10. identical to the Schr¨dinger equation (10.186) then are slowed down.e.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..181) to a physical system we consider pionic atoms.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 (10.181) yields (we assume now that the Klein–Gordon equation describes a 2 particle with mass mπ and charge −e) for qV (r.188) Inserting this into (10. 309 (10.189) . t) (10.185) This is.h. however. π − mesons are generated through inelastic proton–proton scattering p + p −→ p + p + π − + π + .2) of a non-relativistic spin-0 particle o moving in an electromagnetic ﬁeld. (10. A stationary state of the Klein–Gordon equation is described by a wave function ψ(xµ ) = ϕ(r ) e−i t . t) = ˆ [p − q A(r. e. atoms in which one or more electrons are replaced by π − mesons.10.

191) m (θ. . . 1.193) (10. r r r r sin θ r sin θ Ze2 2 r ) (10.190) With this expression and after expanding ( + one obtains 2 ˆ d2 L2 − Z 2 e4 2 Ze2 − + + dr2 r2 r − m2 π r φ(r) = 0 . the eigenfunctions of the operator L2 ˆ L2 Y m (θ.. φ) .192) ˆ are spherical harmonics. (10. The similarity of (10.196) In this expression the number n deﬁned through n = n− −1 (10.310 Relativistic Quantum Mechanics Because of the radial symmetry of the Coulomb potential we express this equation in terms of spherical coordinates r. n − 1 .198) . (10. θ. φ) R (r) Y r m (θ. . . i.195) The latter problem leads to the well-known spectrum En = − mπ (Ze2 )2 .194) Bound state solutions can be obtained readily noticing that this equation is essentially identical to 2 that posed by the Coulomb problem (potential − Ze ) for the Schr¨dinger equation o r d2 − dr2 ( + 1) 2mπ Ze2 + + 2mπ E r2 r R (r) = 0 (10. (10. 2.197) counts the number of nodes of the wave function.194) and (10. (10.e. φ) (10. 2n2 = 0. φ) = ( + 1) Y m (θ. . i. 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 .191) ˆ The operator L2 in this equation suggests to choose a solution of the type ϕ(r ) = where the functions Y in (10. φ.192) leads then to the ordinary diﬀerential equation d2 − dr2 ( + 1) − Z 2 e4 2 Ze2 + + r2 r 2 − m2 π R (r) = 0 .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..e. this quantity deﬁnitely must be an integer. n = 1.199) ( + 1) − Z 2 e4 . . (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 ) . Introducing α = Z 2 e4 and 1 β = + 2 (10. 1. 1.203) spectrum . 2mπ 2 −→ λ( ) . + (10.. EKG = Using (10.196) if one makes the replacement 2 values which can be obtained E −→ One obtains − m2 π . . . . m) = mπ 1 + Z 2 e4 (n− − 1 2 + ( + 1 )2 − Z 2 e4 )2 2 n = 1.10. (10.200) The bound state solutions of this equation should correspond to from (10. .204) reads 1 (10. results in the (10.199) and deﬁning EKG (n. .204) In order to compare this result with the spectrum of the non-relativistic hydrogen-like atom we expand in terms of the ﬁne structure constant e2 to order O( 8 ).e. . (10. .6: Klein–Gordon Equation with Electromagnetic Field and one can write (10. . . . . − + 1. . 2mπ 2 (n + λ( ) + 1)2 (10. . 2.202) Solving this for state) yields (choosing the root which renders = 1 + mπ Z 2 e4 ( n + λ( )+1 )2 ≤ mπ . . .194) d2 λ( ) ( λ( ) + 1 ) 2 Ze2 − + + dr2 r2 r 2 311 − m2 π R (r) = 0 . . 1.201) mπ Z 2 e4 m2 − m2 π π = − . which corresponds to a bound = 0. = 0.197. . . 2 e2 −→ e2 mπ . 10. . n = 0. . i. n − 1 m = − .

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

this was not so. The discovery by Dirac.211) Obviously. 2. one route to important discoveries in physics is the creation of new mathematical descriptions of nature. . yet to be discovered. It turns out that no 4-vector of real or complex coeﬃcients can satisfy these conditions. spin. j = 1.. Initially. However. the realization that the linearization can be carried out only if one assumes a 4-dimensional representation of the coeﬃcients γ µ . γ 3 can only be realized by d×d–matrices requiring that the wave function Ψ(xµ ) is actually a d–dimensional vector of functions ψ1 (xµ ). (10. therefore. Accordingly. For µ = ν condition (10. Starting point is to boldly factorize. the 4-dimensionsional representation discovered by Dirac involved new physical properties. In fact. these descriptions ultimately merging with the true theory of matter. according to (10. 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 .7: Dirac Equation 313 to the problem involved an ingenious step. ψ2 (xµ ).213) reads (γ µ )2 = 1 µ = 0 −1 µ = 1. the operator of the Klein–Gordon equation ∂µ ∂ µ + m2 = −( P + m ) ( P − m ) (10. γ 2 . the quantities γ 0 . are associated with a positive-deﬁnite particle density. exploited ∂µ ∂ν = ∂ν ∂µ . namely. but did not commute the. j = 1. quantities with the transformation law (10.66). 3 . γ 1 . it was assumed that the 4-dimensional space introduced by Dirac could be linked to 4vectors. 3 has imaginary eigenvalues ±i. unspeciﬁed algebraic objects γ µ and γ ν .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 (10.208). Properties of the Dirac Matrices Let us now trace Dirac’s steps in achieving the linearization of the ‘square root operator’ in (10. 2.213) We want to determine now the simplest algebraic realization of γ µ .210) where P = iγ µ ∂µ . .209).e. Inserting (10. We seek to identify the coeﬃcients γ µ . . one can impose the condition γ0 is hermitian . (10. 3 are anti-hermitian . ψd (xµ ). (10. γ j . beautiful mathematical theory and that. achieved through a beautiful mathematical theory.215) .212) where we have changed ‘dummy’ summation indices.214) From this follows that γ 0 has real eigenvalues ±1 and γ j . this would lead to the two wave equations (P − m)Ψ = 0 and (P + m)Ψ = 0 which have a ﬁrst order time derivative and. strengthens the believe of many theoretical physicists today that the properties of physical matter ultimately derive from a. therefore.211) into (10.10. γ ν ]+ = 2 g µν (10. i. 2.1 2 and anti-particles. Instead. so far.

ψ2 . For d = 2 there exist only three anti-commuting matrices. the γ µ are anti-commuting. From this equation one can conclude that also the following should hold ( iγ µ ∂µ − m ) Ψ(xµ ) = 0 which is the celebrated Dirac equation. σ j σ k = − σ k σ j for j = k . (10. 1 (10.314 For µ = ν (10. ψ3 . Relativistic Quantum Mechanics (10. rather than a scalar wave function.217) Obviously. in fact. σ 3 for which. 2. except for similarity trasnformations.218) The Pauli matrices allow one. (10. (10.213) is satisﬁed.222) and multiplication from the right by γ 0 yields the adjoint Dirac equation Ψ† (xµ ) γ 0 iγ µ ∂µ + m ← =0.224) . is unique.221) =0 (10.213) reads γ µ γ ν = −γ ν γ µ . σ 2 . One can readily show using the hermitian property of the Pauli matrices (γ 0 )† = γ 0 and (γ j )† = −γ j for j = 1. The Dirac Equation Altogether we have shown that the Klein–Gordon equation can be factorized formally ( iγ µ ∂µ + m ) ( iγ µ ∂µ − m ) Ψ(xµ ) = 0 (10.223) Inserting this into (10. γj = 0 −σ j σj 0 . as long as det(γ µ ) = 0 the dimension d of the square matrices γ µ must be even so that (−1)d = 1.216) i..215). namely the Pauli matrices σ 1 . A proper choice is γ0 = 1 1 0 0 −1 1 .e.219) Using property (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 ) . ψ4 ) and where ∂µ denotes the diﬀerential operator ∂ µ operating to the left side. to construct four matrices γ µ for the next possible dimension d = 4. in fact. 3 which. (10. holds 2 σj = 1 . however.218) of the Pauli matrices one can readily prove that condition (10.222) ∗ ∗ ∗ ∗ where we have deﬁned Ψ† = (ψ1 . is implied by (10. The Adjoint Dirac Equation The adjoint equation is Ψ† (xµ ) i(γ µ )† ∂µ + m ← ← (10. This property can also be written (γ µ )† = γ 0 γ µ γ 0 .220) where Ψ(xµ ) represents a 4-dimensional wave function. We will argue further below that the choice f γ µ . rather than to the right side.

227) leaves the algebra of the Dirac matrices unaﬀected and commutation property (10. (10.e. 2. . t) = 0 (10.1: Derive (refeq:Dirac-intro20a) from (10.231) ˆ whereα has the three components αj .229) The similarity transformation (10.219) is the socalled chiral representation deﬁned through ˜ Ψ(xµ ) = S Ψ(xµ ) .228) (10.227). (10. and γ0 = ˜ 0 1 1 1 0 1 .225) . γ ˜ Exercise 10. γj = ˜ 0 σj 1 S = √ 2 −σ j 0 1 1 1 1 1 −1 1 1 (10. [˜ µ .10.7: Dirac Equation Similarity Transformations of the Dirac Equation . (10..213) still holds.227) A representation often adopted beside the one given by (10.230) Schr¨dinger Form of the Dirac Equation o Another form in which the Dirac equation is used often results from multiplying (10. 10. 3 .233) The eigenstates and eigenvalues of H correspond to the stationary states and energies of the particles described by the Dirac equation.7. j = 1.226) (10.222. 3 .Chiral Representation 315 The Dirac equation can be subject to any similarity transformation deﬁned through a non-singular ˜ 4 × 4–matrix S. (10. 3 and ˆ β = 1 1 0 0 −1 1 . γ ν ]+ = 2 g µν .221) from the left by γ 0 ˆ ˆ i ∂t + i α · − β m Ψ(r. 2. αj = ˆ 0 σj σj 0 . i. j = 1. 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 = 1.213). ˆ ˆ ˆ Ho = α · p + β m . (10.232) 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 ) . 2.

i.316 Cliﬀord Algebra and Dirac Matrices The matrices deﬁned through dj = iγ j . j = 1.238) These elements form a group since obviously any product of two Γr ’s can be expressed in terms of a third Γr .6 and R. Γ±11 = ±d3 d4 Γ±12 = ±d1 d2 d3 . Γ±3 = ±d2 .236) where ‘⊗’ denotes the Kronecker product between matrices.213). The representations of this group are given by a set of 32 4 × 4–matrices which are equivalent with respect to similarity transformations.234) (10. 3 . One can conclude then that also any set of 4 × 4–matrices obeying (10.Mod.235) can diﬀer only with respect to unitary similarity transformations. . 2. To complete the proof in this section the reader may consult Miller ‘Symmetry Groups and their Application’ Chapter 9.237) by means of the property (10. d3 . d4 is called a Cliﬀord algebra C4 . This property extends then to 4 × 4–matrices which obey (10. hence. Γ±15 = ±d2 d3 d4 Γ±16 = ±d1 d2 d3 d4 (10. i. j1 < j2 < · · · < js s ≤ 4 (10. m = Aj Bkm . The three Pauli matrices also obey the property (10. (1955). d2 .235) as can be readily veriﬁed from (10. to Dirac matrices. Γ±5 = ±d4 Γ±6 = ±d1 d2 . any product 1 of the dj ’s can be brought to the form (10. . There are (including the diﬀerent signs) 32 elements in G4 which we deﬁne as follows Γ±1 = ±1 1 Γ±2 = ±d1 . For example. Since the Γj are hermitian the similarity transformations are actually given in terms of unitary transformations. 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 . satisfy the anti-commutation property dj dk + dk dj = 2 for j = k 0 for j = k Relativistic Quantum Mechanics d4 = γ 0 (10. The Cliﬀord algebra C4 entails a subgroup G4 of elements ± dj1 dj2 · · · djs .213). d2 = .235) and. Γ±8 = ±d1 d4 . page 187.. d4 . d4 = 1 0 0 −1 1 0 0 −1 0 1 1 0 0 i −i 0 (10. Rev. in case of C4 .Good.237) which are the ordered products of the operators ±1 and d1 . The representations of Cliﬀord algebras Cm are well established.. Γ±14 = ±d1 d3 d4 . The reader may .e. Γ±13 = ±d1 d2 d4 .H. 27. the matrix elements of C = A ⊗ B are Cjk.Phys. Γ±9 = ±d2 d3 Γ±10 = ±d2 d4 .e. Γ±4 = ±d3 .235). Obviously. Γ±7 = ±d1 d3 . form a Cliﬀord algebra C3 . The associative algebra generated by d1 .

. (10.80) by Lµ ρ and using (10.213.80). Exercise 7.e. We can..218) of the Pauli matrices σ j .2: Demonstrate the anti-commutation relationships (10.. the form of the Dirac equation is identical in equivalent frames of reference.76) yields ∂ρ = Lν ρ ∂ ν .4: Show that from (10. Inﬁnitesimal Bispinor State Transformation We want to show now that a suitable transformation of Ψ(xµ ) does.240) . in a moving frame holds iγ µ ∂µ − m Ψ (x ) = 0 . a result one could have also obtained by applying the chain rule to (10. i. i. j = 1.219). 10. x = Lµ ν xν . µ µ (10. we do not know yet how to transform the 4-dimensional wave function Ψ and the Dirac matrices γ µ .241) .3: Demonstrate the anti-commutation relationships (10. in fact.236) with the Dirac representation (10. namely. Except for a similarity transformation.10.218) of the Dirac matrices γ µ . and derivatives according to (10.. i. 10. i.e.. i.e. in a moving frame should hold iγ ∂µ − m Ψ (x ) = 0 . these properties determine the matrices γ µ uniquely. The transformation is assumed to be linear and of the form Ψ (x ) = S(Lµ ν ) Ψ(xµ ) µ µ (10.215). The latter transformations imply that coordinates transform according to (10. Exercise 7. and γ j .239) Form invariance implies that the matrices γ µ should have the same properties as γ µ . exist. therefore. x µ = Lµ ν xν (10. it must hold γ µ = γ µ .6).8: Lorentz Invariance of the Dirac Equation 317 also want to establish the unitary transformation which connects the Dirac matrices in the form (10. Multiplication and summation of (10.8 Lorentz Invariance of the Dirac Equation We want to show now that the Dirac equation is invariant under Lorentz transformations. 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. transform coordinates and derivatives of the Dirac equation. Hence.e.6). in frames connected by Lorentz transformations. Lorentz Transformation of the Bispinor State Actually. Exercise 7. However.214) follows that γ 0 has real eigenvalues ±1 and can be represented by a hermitian matrix. 2. 3 has imaginary eigenvalues ±i and can be represented by an anti-hermitian matrix.e.

One can ﬁnally conclude multiplying both sides by g ρµ S(Lη ξ )γ µ S −1 (Lη ξ ) = Lν µ γν . however.241) implies a similarity transformation Sγ µ S −1 .318 Relativistic Quantum Mechanics where S(Lµ ν ) is a non-singular 4 × 4–matrix.s. It holds then S( µν i ) = 1 − σµν 1 4 µν µν = S(− νµ i ) = 1 + σµν 1 4 µν .12) in the form Lν µ gνσ Lσ ρ = gµρ = gρµ . g is an anti-symmetric matrix. it must hold (10. in the expression of the inﬁnitesimal transformation ρ Lµν = g µν + µν (10. (10. Ob1 viously.248) Here σµν denote 4 × 4–matrices operating in the 4-dimensional space of the wave functions Ψ.246) (10.249) from which we can conclude σµν = −σµν νµ = −σνµ i.250) σµν = −σνµ . hence.247) the inﬁnitesimal matrix µν is anti-symmetric. One can.e. The elements of g are. S( µν ) should not change its value if one replaces in its argument µν by − νµ . We had introduced in (10.38) the inﬁnitesimal Lorentz transformations in the form Lµ ν = δ µ ν + µ ν where the inﬁnitesimal operator µ ν obeyed T = − g g. i. .221) upon Lorentz transformation yields iS(Lη ξ )γ µ S −1 (Lη ξ ) Lν µ ∂ν − m Ψ (x ) = 0 .243) We want to determine now the 4 × 4–matrix S(Lη ξ ) which satisﬁes this condition.243) in a form in which the Lorentz transformation in the form Lµν is on the r. follows S(Lη ξ )γρ S −1 (Lη ξ ) = Lσ ρ γσ .243) from the left by Lσ ρ gσν yields S(Lη ξ )γ µ S −1 (Lη ξ ) gρµ = Lσ ρ gσν γσ .h.244) from which. (10. the transformation (10. hence. using gρµ γ µ = γρ . (10. Multiplication of this property by g from the right yields ( g)T = − g.247) can be expanded S( µν i ) = 1 − σµν 1 4 µν (10.245) The construction of S(Lη ξ ) will proceed using the avenue of inﬁnitesimal transformations. µ g ρν = µν and. µ (10. νµ (10. Multiplication of (10. The inﬁnitesimal transformation S(Lρ σ ) which corresponds to (10.242) The form invariance of the Dirac equation under this transformation implies then the condition S(Lη ξ )γ µ S −1 (Lη ξ ) Lν µ = γ ν . The proper starting point for a constructiuon of S(Lη ξ ) is actually (10.e. state that the Dirac equation (10.. the coeﬃcients of which depend on the matrix Lµ ν deﬁning the Lorentz transformation in such a way that S(Lµ ν ) = 1 for Lµ ν = δ µ ν holds.. For this purpose we exploit (10. of the equation.

(10. of (10. like the expression on the l. 2. 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. 10. (10.e..s. For this purpose we inspect the properties of the generators in a particular representation. (10.252) also as a sum over both indices of αβ .h.253) 1 2 αβ ( δ µ β γα − δ µ α γβ ) . γ µ ]− = 2i ( δ µ β γα − δ µ α γβ ) .254) The proper σαβ must be anti-symmetric in the indices α. a solution of condition (10.213. 10. σjk = [˜j .. γj = ˜ 0 −σ j σj 0 . 3 . β and operate in the same space as the Dirac matrices. Algebra of Generators of Bispinor Transformation We want to construct the bispinor Lorentz transformation by exponentiating the generators σµν . ˜ ˜ σ only six generators need to be determined) σ0j = ˜ i [˜0 .257) .228.246) results then in a condition for the generators σµν − i ( σαβ γ µ − γ µ σαβ ) 4 αβ = νµ γν . For this purpose we need to verify that the algebra of the generators involving addition and multiplication is closed.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.h.256) One can readily verify that the non-vanishing generators σµν are given by (note σµν = −˜νµ . (10. In fact. the chiral representation introduced above in Eqs. this condition can actually be expressed through six independent conditions.255) which can be demonstrated using the properties (10.5: Show that the σαβ deﬁned through (10. γj ] = γ ˜ 2 −iσ j 0 0 iσ j .252) Since six of the coeﬃcients αβ can be chosen independently. j = 1.s. must be symmetric with respect to interchange of the indices α and β.251) into (10. i.s. of (10.h. (10. Furthermore. Exercise 7.216) of the Dirac matrices.h. 10.10. For this purpose one needs to express formally the r. γk ] = ˜ γ ˜ jk σ 0 0 σ .254) is σαβ = i [ γα . In this representation the Dirac matrices γµ = (˜ 0 .255) satisfy condition (10.254).229).252) results in the condition for each α. namely.251) S −1 ( µν ) = 1 + σµν µν 1 4 Inserting (10.248.s. Comparing this with the l. −γ ) are ˜ γ ˜ γ0 = ˜ 0 1 1 1 0 1 . γβ ]− 2 (10. the expression on the r. β [ σαβ .

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

Obviously. The conservation law (10.263) does hold.224) from the right by Ψ(xµ ). and addition yields Ψ† (xµ )γ 0 The last result can be written ∂µ Ψ† (xν )γ 0 γ µ Ψ(xν ) = 0 . the property applies then in any representation of S.263) allows one to conclude that j µ must transform like a contravariant 4-vector as the notation implies.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 .267) has the desired property of being positive deﬁnite. The reason is that the r.268) holds for the bispinor Lorentz transformation.224). (10. This transformation behaviour can also be deduced from the transformation properties of the bispinor wave function Ψ(xµ ). (10. of (10. j) = Ψ† (xµ )γ 0 γ µ Ψ(xµ ) . in fact. ϑ) = S(−w.263) Starting point are the Dirac equation in the form (10. for j µ (xµ ) = (ρ. (10. 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 . Multiplying (10.265) (10.221) and the adjoint Dirac equation (10. We will now determine the relationship between the ﬂux j µ = Ψ (x )γ 0 γ µ Ψ (x ) † µ (10.h. The time-like component ρ of j µ 4 iγ µ ∂µ + iγ µ ∂µ ← Ψ(xµ ) = 0 . (10.221) from the left by Ψ† (xµ )γ 0 . j µ must transform like a contravariant 4-vector. Since ∂µ transforms like a covariant 4-vector..e. (10.10.271) .268) We will prove this property in the chiral representation.264) (10. the conservation law (10.270) We conclude that (10.s.263) obviously is a scalar under Lorentz transformations and that the left hand side must then also transform like a scalar. −ϑ) = e− 2 (w − iϑ)·σ 0 1 (w + iϑ)·σ 0 e2 1 (10.269) One can readily show that the same operator is obtained evaluating γ S (w.266) ρ(xµ ) = Ψ† (xµ )Ψ(xµ ) = s=1 |ψs (xµ )|2 (10. i. For our proof we note ﬁrst ˜ ˜ S −1 (w. For this purpose we prove ﬁrst the relationship S −1 = γ 0 S † γ 0 .

282. These operaˆ tors. One obtains using (10. (10. bispinor wave functions and we need to determine corresponding components of the wave function..273) where we have employed (10. The operators commute also with Ho in (10. In relativistic quantum mechanics a Dirac particle can also be characterized through a momentum.9 Solutions of the Free Particle Dirac Equation We want to determine now the wave functions of free particles described by the Dirac equation. The solutions provide also a complete.233) o i∂t Ψ(xµ ) = Ho Ψ(xµ ) . commute with each other. We will start from the Dirac equation in the Schr¨dinger form (10. a property which leads to the energy– momentum (dispersion) relationship of the Dirac particle. 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. orthonormal basis and allows one to quantize the Dirac ﬁeld Ψ(xµ ) much like the classical electromagnetic ﬁeld is quantized through creation and annihilation operators representing free electromagnetic waves of ﬁxed momentum and frequency. t) and is determined through the momentum p ∈ R3 . As pointed out. i. but also as providing a basis in which the wave functions of interacting particle systems can be expanded and characterized. 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. Like in non-relativistic quantum mechanics the free particle wave function plays a central role. Combining this with (10. the Dirac particles are described by 4-dimensional.268) µ j = Ψ† (xµ )S † γ 0 γ µ SΨ(xµ ) = Ψ† (xµ )γ 0 S −1 γ µ SΨ(xµ ) .233). (10.231.f.272) results in the expected transformation behaviour µ j = Lµ ν j ν .274) 10. however. In case of non-relativistic quantum mechanics the free particle wave function has a single component ψ(r. as is required for the mentioned property. speciﬁc for the Dirac free particle. the identiﬁcation of observables and their quantum mechanical operators.e. The additional degrees of freedom described by the four components of the bispinor wave function require.283)].232. as just mentioned.272) With S −1 (Lη ξ ) = S((L−1 )η ξ ) one can restate (10.246) S −1 (Lη ξ )γ µ S ( Lη ξ ) = (L−1 )νµ γν = (L−1 )ν γ ν = Lµ ν γν . Note that we have assumed in (10. (10. In the following we want to provide this characterization.275) The free particle wave function is an eigenfunction of Ho . the wave function has four components which invite further characterization of the free particle state. Like for the free particle wave functions of the non-relativistic Schr¨dinger and the Klein–Gordon o equations one expects that the space–time dependence is governed by a factor exp[i(p · r − t)].76). of which the wave functions are eigenfunctions as well. not only as the most simple demonstration of the theory. additional characterizations. As it turns out. 10. For this purpose we consider . (10.271) that the Dirac matrices are independent of the frame of reference. µ (10.

280) σ·p (σ · p)2 φ0 = χ0 = +m ( + m) ( + m) p2 χ = χ0 .234) and (10.278) provides us with information about the components of the bispinor wave function (10. reproduce the classical relativistic energy–momentum relationships (10. 10. (10.282) (10. the Dirac equation.277) To solve this problem we write (10.234) of Pauli matrices holds (σ · p)2 = p 2 1 One can.280) which has a positive and a negative solution = ± E(p) . conclude 1.278) 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 .276) into (10.277) explicitly ( − m) 1 φo − σ · p χo 1 − σ · p φo + ( + m) 1 χo 1 = 0 = 0. . E(p) = m2 + p 2 . two-dimensional spinor state.94) Equation (10. χo each represent a constant.276).283) imply that the bispinor part of the wave function allows only two degrees of freedom to be chosen independently. (10. 10.276) where p and together represent four real constants.282.279) the well-known relativistic dispersion relationship 2 = m2 + p 2 (10. +m (10. one ﬁnds using (5. In fact. namely φo and χo are related as follows φ0 χo = = σ·p χ0 −m σ·p φ0 . Inserting (10.281). the socalled helicity of the particle.231. We want to show now that these degrees of freedom correspond to a spin-like property.281) Obviously. and φo . 10. from (10. hence.284) The relationships (10. (10.232) leads to the 4-dimensional eigenvalue problem m σ·p σ · p −m φo χo = φo χo . later to be identiﬁed with momentum and energy.93. 1 (10.10.9: Solutions of the Free Particle Dirac Equation the following form of the free Dirac particle wave function Ψ(xµ ) = φ(xµ ) χ(xµ ) = φo χo ei(p·r − t) 323 (10.283) where is deﬁned in (10. like the Klein–Gordon equation.279) According to the property (5. These two relationships are consistent with each other. 2 − m2 0 (10.

+|xµ ) = N+ (p) u σ·p E(p) + m u ei(p·r − t) .286) for p = 0 yields. ±)γ 0 Ψ(p.292) . (10. −|xµ ) = N− (p) −σ·p E(p) + m u u ei(p·r − t) .268). (10. For this purpose we consider ﬁrst the positive energy solution. ±).281)] separately. S −1 = γ 0 S † γ 0 and.. the l. ±) = S Ψ(0. u† u = |u1 |2 + |u2 |2 = 1 .241). ±) γ 0 Ψ(0. i.286) Here N+ (p) is a constant which will be chosen to satisfy the normalization condition Ψ† (p. +) γ 0 Ψ(0.h.h. +) γ 0 Ψ† (p.e. and let Ψ(0. (10.287. using γ 0 as given in (10. The connection between the solutions. Employing (10. i. ±) γ 0 S −1 γ 0 γ 0 S Ψ(0. +) = −1 . (10. ±) denote the solution of a free particle moving with momentum p in the laboratory frame. = +E(p) . S † = γ 0 S −1 γ 0 . The form of this condition and of (10. 10.s. ±) Ψ† (0.h. +) = |N+ (0)|2 (u† .290). (10. we present the negative energy solution. hence.288) corresponding to the wave function Ψ(p. ±).287) will be justiﬁed now. we demonstrate that the product Ψ† (p. the solution for = −E(p). ±) γ 0 Ψ(p. 0) γ 0 u 0 = |N+ (0)|2 . of (10. according to (10.f. = −E(p) .290) should diﬀer. ±) . Ψ† (p.259). (10.219) and u† u = 1 [c. Similarly. i. is Ψ(p. We want to demonstrate now that the signs on the r. the solution for = +E(p).290) which diﬀers from the normalization condition (10. First. For the positive energy solution. we present φo through the normalized vector φo = u1 u2 = u ∈ C2 . +) γ 0 Ψ† (p. ±) denote the corresponding solution of a particle in its rest frame. Ψ† (0. of (10.e.287.s. 10.287) the form of which will be justiﬁed further below. One can see this as follows: Let Ψ(p. (10.. Hence.324 Relativistic Quantum Mechanics For our further characterization we will deal with the positive and negative energy solutions [cf.s.285)].287) in the minus sign on the r. (10. +) = 1 .. according to (10. ±) S † γ 0 S Ψ(0. ±) Ψ† (0.285) The corresponding free Dirac particle is then described through the wave function Ψ(p. (10.291) Note that we have used that. where S is given by (10. u† u = |u1 |2 + |u2 |2 = 1 . ±) = = = Ψ† (0. through χo given by χo = u1 u2 = u ∈ C2 .289) Here N− (p) is a constant which will be chosen to satisfy the normalization condition Ψ† (p. is invariant under Lorentz transformations.e. (10.

h. 2m (10.299) ( m + E(p) )2 . using γ 0 as given in (10.286). (10.302) . ( m + E(p) )2 − p 2 (10.9: Solutions of the Free Particle Dirac Equation The same calculation for the negative energy wave function as given in (10.294) We want to determine now N± (p) for arbitrary p. −) γ 0 Ψ(0. of (10.298) becomes N+ (p) = m + E(p) .285) results in 2 N+ (p) 1 − p2 (E(p) + m)2 = 1 (10.296) Replacing (σ · p)2 by p 2 [c.287) written explicitly using (10.6: Show that the normalization condition 2 N+ (p) (u∗ )T .f. 2 E(p) (10.s.s. −) = |N− (0)|2 (0.10. σ·p u∗ E(p) + m T γo u σ·p E(p) + m u = 1 (10.297) from which follows N+ (p) = Noting ( m + E(p) )2 − p 2 = m2 − p 2 + 2mE(p) + E 2 (p) = 2(m + E(p)) m the normalization coeﬃcient (10. We consider ﬁrst the positive energy solution.234)] and using the normalization of u in (10. (10.219) yields 2 N+ (p) u† u − u† (σ · p)2 u (E(p) + m)2 = 1. σ·p u∗ E(p) + m T u σ·p E(p) + m u = 1 (10.293) Obviously.h. 325 (10.290) to assign a positive value to |N− (0)|2 .300) (10. u† ) γ 0 0 u = − |N− (0)|2 .286) is 2 N+ (p) (u∗ )T .289) yields Ψ† (0. (5. Condition (10. Exercise 7.295) Evaluating the l.301) yields the normalization coeﬃcient N+ (p) = m + E(p) .298) This result completes the expression for the wave function (10. this requires the choice of a negative side on the r. We can also conclude from our derivation N± (0) = 1 .

(10.306) i. Helicity The free Dirac particle wave functions (10. Condition (10. however.275) yields Ho Ψ(p. deﬁned as the component of the particle spin along the direction of motion.305) and. Ho ] = 0 follows from (10. The commutation property [σ · p. have completed the determination for the wave function (10. λ|xµ ) (10.308) 2 |p| ˆ Note that p represents here an operator. The wave functions are also eigenstates of the momentum operator i∂µ . λ|xµ ) = pµ Ψ(p. thereby. λ|xµ ) in 4-vector notation.289. (10. 10. −p).h.s. identical to the condition (10. Inserting (10. We can.e.e.290) written explicitly using (10. j = 1. The corresponding operator which measures this observable is 1 p ˆ Λ = σ· . hence.289) is T −σ·p −σ · p 2 E(p) + m u N− (p) u∗ . yields 2 N− (p) u† (σ · p)2 u − u† u (E(p) + m)2 = −1 .296) for the normalization constant N+ (p) of the positive energy solution. 10.303) E(p) + m u Evaluating the l. i. Rather than considering the ˆ observable (10. 2.e. This degree of freedom 1 describes a spin– 2 attribute..275).. (u∗ )T γ o = −1 (10. i∂µ Ψ(p.309) . The property [σ · p.289). the two components of u indicate another degree of freedom which needs to be deﬁned.286.304) This condition is. exp[i(p · r − t)] = exp(ipµ xµ ).286.289) are not completely speciﬁed. λ|xµ ) = λ E(p) Ψ(p. 3 is fairly obvious.300) have been constructed to satisfy the free particle Dirac equation (10.286) into (10. The wave functions (10. pj ] = 0 . λ|xµ ) . 10. the wave functions constructed represent eigenstates of Ho . This attribute is the so-called helicity.326 Relativistic Quantum Mechanics We consider now the negative energy solution.307) where pµ = ( . This can be veriﬁed expressing the space–time factor of Ψ(p. We ˆ want to show that this operator commutes with Ho and p to ascertain that the free particle wave ˆ function can be simultaneously an eigenvector of all three operators..233) and ˆ ˆ ˆ from the two identities 1 1 0 0 −1 1 σ 0 0 σ ·p − ˆ σ 0 0 σ ·p ˆ 1 1 0 0 −1 1 = 0 (10.307) we investigate ﬁrst the observable due to the simpler operator σ · p. conclude N− (p) = m + E(p) 2m (10. not a constant vector. (10. i.

= +E(p).. Ho and Λ deﬁned in (10.310) We have shown altogether that the operators p. t) = Np e 2 1 p m + Ep 0 0 i(px3 − E t) 1 p e Ψ(pˆ3 . In this case Λ = 2 σ 3 . +. i.313) 2 0 1 where Ep and Np are deﬁned in (10. + 1 |r.286) are eigenfunctions of Λ as well will specify now the vectors u.311) 2 −p 0 m + Ep 1 where e3 denotes the unit vector in the x3 -direction and where ˆ Ep = m2 + p2 . i. −. 1 p = (0.e. 0. 327 0 (σ · p)2 ˆ 2 (σ · p) ˆ 0 0 (σ · p)2 ˆ 2 (σ · p) ˆ 0 (10. t) = Np e (10. +. 2 the wave functions which are eigenstates of the helicity operator. + 2 |r. 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 .e..312) We assume now particles with negative energy. are 1 i(px3 − E t) 0 p e Ψ(pˆ3 .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. Ho and σ·p commute with each other and. in principle.224) 1 of σ3 the two u vectors (1. Since helicity is deﬁned relative to the direction of motion of a particle the characterization of u as an eigenvector of the helicity operator. According to the deﬁnition (5. − 1 |r.312). We consider ﬁrst the simplest case that particles move along the x3 –direction.308) above. Np = m + Ep .. p3 ). 0)T and (0. −.10. ˆ The condition that the wave functions (10. 2m (10. t) = Np e 1 0 0 p m + Ep i(px3 + E t) 1 p e Ψ(pˆ3 . is independent of the direction of motion of the particle.e. − 1 |r. 1)T are eigenstates of 2 σ 3 with eigenvalues ± 1 . = −E(p). hence. t) = Np e (10. i. We assume ﬁrst particles with positive energy. Therefore. States which are simultaneously eigenvectors of these three operators are also simulteneously eigenvectors of the three operators p. . can ˆ ˆ be simultaneously diagonal.

223) as follows 1 u(p.123). a wave function which represents free particles at rest. t) 2 = Np u(p. Generating Solutions Through Lorentz Transformation The solutions (10. + 1 |r. − ) 2 where ϑ(p) = = = exp exp i − ϑ(p) · σ 2 i − ϑ(p) · σ 2 1 0 0 1 .] The corresponding free particle wave functions are then Ψ(p. − 2 |r.312) can be obtained also by means of the Lorentz transformation (10.312). 10.315) e3 × p ˆ ∠(ˆ3 .313) except that the states (1. (10. − 2 ) u(p. 1)T have to be replaced by eigenstates u± (p) of the spin operator along the direction of p. + 2 ) u(p.317) Ψ(p. i. For this purpose one starts from the solutions of the Dirac equation in the chiral representation (10.328 Relativistic Quantum Mechanics To obtain free particle wave functions for arbitrary directions of p one can employ the wave functions (10.. − 1 |r.318) 1 Ψ(p. t) = N√ ei(p·r + Ep t) (10. + 2 |r. + 2 ) p 1 m + Ep u(p. p) e |p| (10.220). for an r–independent wave function.229). The corresponding wave functions are determined through ˜ i˜ 0 ∂t − m Ψ(t) = 0 . 10. −. 2 1 0 (10.321) . β. + 2 ) ei(p·r − Ep t) (10. t) 2 = Np 1 u(p.262) for the bispinor wave function and the transformation (10. +. These eigenstates are obtained through a rotational transformation (5. − 1 ) 2 −p 1 m + Ep u(p.222. 0)T and (0.314) (10. 10.316) describes a rotation which aligns the x3 –axis with the direction of p. γ and are 1 0 1 ˜ Ψ(p = 0. + 1 ) 2 p 1 m + Ep u(p. t) = Np ei(p·r + Ep t) (10.e. +.311. denoted by ˜. 1 |t) = √2 e−imt .320) where Ep and N are again given by (10.319) 1 Ψ(p. −.226. − 1 ) 2 ei(p·r − Ep t) (10.311. in which case the transformation is given by (5. γ . + ) 2 1 u(p. [One can also express the rotation through Euler angles α. +. 5. − 2 ) −p 1 m + Ep u(p.

10.176) imt → i ( p3 x3 and for the bispinor part according to (10.228) yields the corresponding solutions (10. χo are eigenstates of σ 3 with eigenvalues ±1. −1 e 0 0 1 +imt 1 ˜ Ψ(p = 0. + 1 |r.9: Solutions of the Free Particle Dirac Equation 0 1 1 ˜ Ψ(p = 0. 2 0 1 1 0 +imt 1 1 ˜ Ψ(p = 0. 2 0 e −1 329 (10.325) to (10. Applying (10. 2 The solutions (10.10. ± 1 of the wave functions stated in (10. For the resulting wave functions in the chiral representation one can use then a notation corresponding to that adopted in (10.174. 10. χo ∈ 1 0 ..323) should yield the solutions for non-vanishing momentum p in the x3 –direction.324. 2 |t) = √2 . −.325) One should note that φo . +. +.311) e 1 w 2 3 ˜ Ψ(p(w3 )ˆ3 .e. t) e 2 e− 2 w3 1 = √ 1 2 e 2 w3 i(px3 − E t) p e .311.262) φo χo → e 2 w3 σ 0 1 3 Et ) (10.322) The reader can readily verify that transformation of these solutions to the Dirac representationsas deﬁned in (10. w3 ).323) Transformation (10. −. − 1 |t) = √2 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 .322). This correspondence justiﬁes the characterization ±.322) can be written in spinor form 1 √ 2 φo χo e imt . − 1 |t) = √2 . φo .262) for a boost in the x3 –direction. 0 1 (10. − 1 |r. i. yields for the exponential space–time dependence according to (10. for w = (0.324) 0 e − 1 w3 σ 3 2 φo χo = e 2 w3 σ φo 1 3 e− 2 w3 σ χo 1 3 . +. 0.313) in the p → 0 limit. 10. (10.

2 sinh x = 2 coshx − 1 .61) p(w3 ) = m sinhw3 .228) yields w3 cosh 2 = sinh w3 2 1 Ψ(p(w3 )ˆ3 . −. −. −. t) e = i(px3 + E t) p e (10.329) . − 2 |r. +.328) the relationship (10. +. − 1 |r.311) for Ep one obtains 1 = √ 2 1 1 √ 2 + 1 = 1 − v3 2 m2 + = 2 m2 v3 2 1 − v3 cosh w3 2 1+ 2 v3 2 + 1 1 − v3 +m = 2m Ep + m 2m (10.327) Employing the hyperbolic function properties cosh x = 2 coshx + 1 . − 2 |r. and the expression (10.326) where according to (10. + 1 |r. t) e 1 0 1 0 0 1 0 1 1 0 1 0 0 1 0 1 i(px3 − E t) p e 1 Ψ(p(w3 )ˆ3 .330 e 1 w 2 3 Relativistic Quantum Mechanics 1 0 1 0 0 1 0 1 1 ˜ Ψ(p(w3 )ˆ3 . Transformation to the Dirac representation by means of (10. t) e 2 e 1 = √ 1 2 − e 2 w3 i(px3 + E t) p e (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 . + 2 |r. t) e 2 = i(px3 + E t) p e 1 Ψ(p(w3 )ˆ3 . t) e 1 = √ 2 −e− 2 w3 − 1 w3 2 1 i(px3 + E t) p e ˜ Ψ(p(w3 )ˆ3 . −. 2 (10. + 2 |r.61) between the parameter w3 and boost velocity v3 .

123) above and characterized there.10. In the new derivation we consider the particle described by the wave function transformed. = γν . that if Ψ(xµ ) is a solution of the Dirac equation that Ψ (xµ ) as given in (10.313) for −p.243). Actually.331) i S(Lη ξ )γ µ S −1 (Lη ξ ) ρ(Lη ξ )∂µ ρ−1 (Lη ξ ) − m Ψ (xµ ) = 0 (10. reproduces the positive energy wave functions (10. 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 (10.9: Invariance of Dirac Equation Revisited and similarly sinh w3 = 2 Ep − m = 2m p 2m (Ep + m) 331 (10.243) considered solely the limit of inﬁnitesimal transformations anyway. the essential property stating the Lorentz–invariance of the Dirac equation. Such description will be essential for the formal description of Lorentz invariant wave equations for arbitray spin further below. which however.327).176). when we generated the solutions Ψ(p.331) is a solution as well. ρ(Lη ξ ) has been deﬁned in (10. This transformation. Λ|t). Invariance of Dirac Equation Revisited At this point we like to provide a variation of the derivation of (10. λ.329. we will derive this equation only for inﬁnitesimal transformations.333) . Making this expectation a postulate allows one to derive the condition (10. To show this we rewrite the Dirac equation (10. expresses the system in the old coordinates.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. but not the observer. (10.243) and. in general.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 ).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 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 .221) using (10.334) (10. The transformation is Ψ (xµ ) = S(Lη ξ ) ρ(Lη ξ ) Ψ(xµ ) (10. λ.330) Inserting expressions (10.168–10.330) into (10. We expect. The reason why we provide another derivation of (10. of course. and since (2) the calculations following (10. Such transformation had been applied by us.311) as well as the negative energy solutions (10. Λ|xµ ) from the solutions describing particles at rest Ψ(p = 0. the proper transformation S(Lη ξ ). 10. thereby. refered to as the active transformation. is suﬃcient since (1) it must hold then for any ﬁnite transformation. indeed.

336). ∂µ ] = (K )ν µ ∂ν . ∂µ ] = [x2 ∂1 − x1 ∂2 . 0). w = (0. ∂µ ] = (10. ϑ = (0. ∂µ ] = [x0 ∂3 + x3 ∂0 . ∂µ ] = (10.340) [K2 . The second condition is identical to (10.338) [J3 .e. 1) .341) [K3 . .337) [J2 . .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. of course. 0). the Dirac equation is invariant under active Lorentz transformations. 0.262). . ϑ = (0. ∂µ ] = (10. As mentioned already we will show condition (10. ∂µ ] = [x0 ∂2 + x2 ∂0 . ∂µ ] = [x1 ∂3 − x3 ∂1 . ∂µ ] = (10. that any solution Ψ(xµ ) transformed according to (10. 0).335) = ∂µ + M ν µ ∂ν + O( 2 ) The result will show that the matrix M ν µ is identical to the generators of Lν µ for the six choices ϑ = (1. . therefore. 0.342) One can readily convince oneself that these results are consistent with (10.332 Relativistic Quantum Mechanics We will demonstrate now that the ﬁrst condition is satisﬁed by ρ(Lη ξ ). Inspection of (10.243) and. ∂µ ] = (10. w = (0. 0). w = (0.. We will proceed by employing the generators (10. 0. i. ∂µ ] = (J )ν µ ∂ν . 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 (10.335) shows that we need to demonstrate [J . We have demonstrated. [K . ∂µ ] = [x0 ∂1 + x1 ∂0 . 0. it is met by S(Lη ξ ) as given in the chiral representation by (10.336) We will proceed with this task considering all six cases: [J1 .331) is again a solution of the Dirac equation. ∂µ ] = [x3 ∂2 − x2 ∂3 . .334) for inﬁnitesimal Lorentz transformations Lη ξ . ∂µ ] = (10.339) [K1 . 0). 0.

351) X (x ) µ . Accordingly. σ3 )T one obtains then i∂t φ i∂t χ = = ˆ σ · π χ + qV φ + mφ ˆ σ · π φ + qV χ − mχ .6 above). (10.10: Dirac Particles in Electromagnetic Field 333 10.346) ˆ ˆ ˆ α · π + qV + β m Ψ(xµ ) (10.345) (10.233) by replacing i∂t by (see Table 10. we replace the operator ∂µ by ∂µ + iqAµ . The Dirac equation (10.345) holds |qV | << m . accordingly.221) reads then [ iγ µ (∂µ + iqAµ ) − m ] Ψ(xν ) = 0 (10.g.1) i∂t − qV and p by ˆ ˆ π = p − qA .344) Using the notation σ = (σ1 .349) We want to focus on the stationary positive energy solution. The Dirac equation in the Schr¨dinger form reads then o i∂t Ψ(xµ ) = ˆ ˆ where α and β are deﬁned in (10.350) (10.232).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. For this purpose we choose the decomposition Ψ(xµ ) = φ(xµ ) χ(xµ ) . (10. e.1 in Sect 10. for the scalar ﬁeld V in (10. σ2 .. 10.345) in the so-called non-relativistic limit in which all energies are much smaller than m. Equivalently.348) (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. This solution exhibits a timedependence exp[−i(m + )t] where for holds in the non-relativistic limit | | << m. we deﬁne φ(xµ ) = e−imt Φ(xµ ) χ(x ) = e µ −imt (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 10.6 we assume that the ﬁeld is described through the 4-vector potential Aµ and. Following the respective procedure developed for the Klein-Gordon equation in Sect.10.

Ak ] + q 2 [Aj . due to the m−1 factor.357) Equation (10. ∂k + qAk ] i i 1 1 1 1 [ ∂j . (10.334 and assume that for the time-derivative of Φ and X holds ∂t Φ << m . For this purpose we employ the identity (5.356) i∂t Φ ≈ Φ + qV Φ . does not need to be considered anymore.357) for Φ can be reformulated by expansion of (σ · π)2 . ∂k ] + [∂j . (10. i i (10. Ak ] i i i i =0 =0 = q q [Aj . 10. 5.353) (10.354) ˆ 0 ≈ σ · π Φ − 2m X .348.361) .230). Ak ] ) f = ( ∂j Ak − Ak ∂j + Aj ∂k − ∂k Aj ) f . ∂k ] + [∂j .350. derived in Sect.360) For an arbitrary function f (r) holds ( [Aj . Φ Using (10.349) yields i∂t Φ i∂t X = = ∂t X X Relativistic Quantum Mechanics << m .352) ˆ σ · π X + qV Φ ˆ σ · π Φ + qV X − 2mX . henceforth.358) ( πj πk − πk πj ) = jk [πj .7. π ] . ∂k ] + q [ ∂j .352) allow one to approximate (10.355) ˆ σ·π Φ 2m (10. 10.353) by X ≈ to obtain a closed equation for Φ ˆ σ·π 2m 2 (10. (10.359) We want to evaluate the latter commutator. 10. one can replace X in (10. ˆ Equation (10.351) in (10. and. πk ] = = 1 1 [ ∂j + qAj . One obtains [πj . (10. which in the present case states ˆ ˆ2 ˆ ˆ (σ · π)2 = π + i σ · (π × π) . ∂k ] + q [Aj .346. identiﬁes X as the small component of the bi-spinor wave function which. (10. accordingly. ˆ ˆ For the components of π × π holds ˆ ˆ π×π = jk (10.354) The properties (10. Ak ] .356).

362) or. −A).223) shows that the propagator in (10. Equations (10. [πj .1 with the magnetic ﬁeld. t) + q V (r. ∂k ] + [∂j .363) allow us to write (10. the spin. 0.1 which emerged in the Lorentz-invariant theory as an algebraic 2 necessity. πk ] = q q (( ∂j Ak − ∂k Aj )) = − i i ×A jk q = − B i jk (10.364) governing a two-dimensional wave function Φ ∈ C2 with the corresponding non-relativistic Schr¨dinger equation (10.358. The respective bispinor wave function Ψ(xµ ) ∈ C4 is described as the stationary solution of the Dirac equation (10.2) governing a one-dimensional wave function ψ ∈ C.363) where we employed B(r. Comparision of (10.344. 10.2) it introduces the o extra term qσ · B Φ which describes the well-known interaction of a spin.Spectrum We want to describe now the spectrum of a relativistic electron (q = −e) in the Coulomb ﬁeld of a nucleus with charge Ze.364) which is referred to as the Pauli equation.357) in the ﬁnal form i∂t Φ(r.359.360) and Aµ = (V. 10.2 around the magnetic ﬁeld.222. t) [see (8. Ak ] ) f = [ ((∂j Ak )) − ((∂k Aj )) ] f (10. o reveals a stunning feature: the Pauli equation does justice to its two-dimensional character. r (10. The formal solution of this equation is Φ(t) = e−iqtB·σ Φ(0) . t)]2 q − σ · B(r. Let us consider brieﬂy the consequences of the interaction of a spin.e. the interaction qσ · B induces a 1 precession of the spin. t) = × A(r.365) Comparision of this expression with (5. 5. 0. (10. Dirac Particle in Coulomb Field . one obtains ( [Aj . t) 2m 2m Φ(r. while agreeing in all other respects with the non-relativistic Schr¨dinger equation (10.366) can be interpreted as a rotation around the ﬁeld B by an angle qtB.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 ((· · ·)). 0) .1 particle with a magnetic 2 ﬁeld B. does not leave the theory again when one takes the non-relativistic limit. i.6)].367) . t) ≈ ˆ [p − q A(r. but rather remains as a steady “guest” of non-relativistic physics with the proper interaction term. using (10. 10. t) (10.10.. In other words.366) (10. For 2 this purpose we disregard the spatial degrees of freedom and assume the Schr¨dinger equation o i∂t Φ(t) = q σ · B Φ(t) .343) for the vector potential Aµ = (− Ze2 .

1 one can write (10. respectively.369) For our solution we will adopt presently a strategy which follows closely that for the spectrum of pionic atoms in Sect.228) and in (10.368) ˜ ( γ µ πµ − m ) Ψ(xµ ) = 0 . 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 . γ ˆ2 ˆ2 (10.370) resembles closely the Klein-Gordon equation (10. (5.e. 10.373) (10. This follows from [ i˜ µ (∂µ + iqAµ ) + m ] [ i˜ µ (∂µ + iqAµ ) − m ] γ γ = [ i˜ µ (∂µ + iqAµ ) − m ] [ i˜ µ (∂µ + iqAµ ) + m ] γ γ according to which follows from (10. but diﬀers from it in an essential way.180). noting from (10. (˜ 0 )2 = 1 and (˜ j )2 = γ 1 γ −1 j = 1.375) Following the algebra that connected Eqs.370). 2.ν=1 µ =ν γ µ γ ν πµ πν . indeed.336 Relativistic Quantum Mechanics For the purpose of the solution we assume the chiral representation.369) from the left by γ ν πν + m. but the converse is not necessarily true. multiplying (10. Equation (10.374).370) is obtained then ˜ [ i˜ µ (∂µ + iqAµ ) + m ] Ψ(xµ ) γ is a solution of (10.371).230).370) Any solution of (10. ˜ (10. 1.230) γ µ γ ν = − γ ν γ µ .369).369). Employing πµ as deﬁned ˜ in Table 10.368) ˜ where Ψ(xµ ) and γ µ are deﬁned in (10. The diﬀerence arises from the term γ µ πµ γ ν πν in (10.231).229).h. 3. can be rewritten using.s. once a solution Ψ(xµ ) of (10. µ = ν and altering ‘dummy’ summation ˜ ˜ ˜ ˜ . 5. is a 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. γ ˆ ˜ ˆ (10. according to (10.370) ˜ [ i˜ µ (∂µ + iqAµ ) − m ] [ i˜ µ (∂µ + iqAµ ) + m ] Ψ(xµ ) = 0 γ γ (10.371) such that we can conclude that (10. we solve ˜ [ i˜ µ (∂µ + iqAµ ) − m ] Ψ(xµ ) = 0 γ (10.368) is also a solution of (10.6.372) (10. (5. ˜ ˜ ˆ ˆ (10. i.374) The ﬁrst term on the r. For this purpose we ‘square’ the Dirac equation. 3 µ=0 ˆ2 (˜ µ )2 πµ = π0 − π .7 one can write the second term in (10. ˜ However.232) in Sect.

γ ] [ˆ0 .380) where f = f (r.382) .378) According to the deﬁnition (10. πν ] = 0 π ˆ for µ.381) According to (10. Altogether. it holds 1 4 [˜ µ .380) γ µ γ ν πµ πν = i ˜ ˜ ˆ ˆ µ. πj ] .10: Dirac Particles in Electromagnetic Field indices. πν ] γ ˜ π ˆ µ. πj ] in (10.ν=1 µ =ν 337 1 2 ( γ µ γ ν πµ πν + γ ν γ µ πν πµ ) ˜ ˜ ˆ ˆ ˜ ˜ ˆ ˆ µ.379) The commutators [ˆ0 .10. πν ] γ ˜ π ˆ µ.367) of Aµ .376) This expression can be simpliﬁed due to the special form (10.. r2 (10.376–10.377) which follows readily from the deﬁnition (10.ν=1 µ =ν 3 j=1 3 3 = 1 4 1 2 1 [˜ . one can summarize (10.ν=1 µ =ν σ 0 0 −σ · (( qA0 )) (10. Since [ˆµ . γ 0 ] [ˆj .ν=1 µ =ν (10.344). −i∂j ] = − [ (∂t + iqA0 ) ∂j − ∂j (∂t + iqA0 ) ] f i ((∂j qA0 )) f (10. ν = 1. πj ] + γ ˜ π ˆ 4 0 j [˜ j . 2. γ ν ] [ˆµ .ν=1 µ =ν = [˜ µ . i. πj ] π ˆ = = (−i∂t + qA0 . 3 (10. the commutators [˜ 0 . γ j ] [ˆ0 .378) can be evaluated using (10.229).e.367) holds qA0 = r ˆ Ze2 . π0 ] γ ˜ π ˆ j=1 = [˜ 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.367) and the deﬁnition (10. γ µ γ ν πµ πν = ˜ ˜ ˆ ˆ µ. γ j ] are γ ˜ [˜ 0 . γ j ] γ ˜ = = 2 0 1 1 1 0 1 σj 0 0 −σ j 0 σj −σ j 0 − 0 σj −σ j 0 0 1 1 1 0 1 (10. γ ν ] [ˆµ .344) π ˆ [ˆ0 . due to A = 0. γ ˜ π ˆ j=1 (10. t) is a suitable test function and where ((· · ·)) denotes the range to which the derivative is limited.

(10. (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. only the two states {Yjm (j − 2 .384) can be interpreted as the energy of the stationary state and.383) We seek stationary solutions of this equation. (10. 10. 6. 1 |ˆ). Insertion of (10. couples the orbital angular momentum of the electron to its spin. −j + 1. (10.6.1 .e. hence.. 1 |ˆ). .5 which describe the coupling of orbital angular momentum and spin 1 {( Yjm (j − 2 .148).384) into (10. 6. Yjm (j + 1 . Such solutions are of the form ˜ ˜ Ψ(xµ ) = Φ(r) e−i t .388) in terms of states introduced in Sect.381). .387) The expression (10. Yjm (j + r 2 1 1 .385) 1 We split the wave function into two spin. The term iσ · r. m values are coupled. . however. i.147. 2 |ˆ)} as given in (6.374). a solution of the form ∼ Y m (ˆ) can be obtained.386) + 2 ± iσ · r ˆ Ze2 − m2 r2 φ± (r) = 0 .388) Except for the term iσ · r this equation is identical to that posed by the one-dimensional Kleinˆ Gordon equation for pionic atoms (10.189) for the Laplacian and expansion of the term (· · ·)2 result in the twodimensional equation 2 ∂r − ˆ L2 − Z 2 e4 iσ · r Ze2 ˆ 2Ze2 + r2 r + 2 − m2 r φ± (r) = 0 . in r ˆ fact. Combining this result with (10.389) According to the results in Sect. In the latter case. (10. (10. . We note that these states are also eigenstates of the angular r 2 .5 the operator iσ · r is block-diagonal in this basis such that only ˆ 1 the states for identical j.2 components ˜ Ψ(r) = and obtain for the separate components φ± (r) Ze2 + r 2 ˜ φ+ (r) ˜ φ− (r) (10. r r 2 2 2 j = 1. m = −j.338 Relativistic Quantum Mechanics where r = r/|r| is a unit vector. + j } (10. 2 3 2 . . 6. we express 2 the solution of (10.191) solved in Sect. Accordingly.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 . it is this quantity that we want to determine.375) the ˆ ‘squared’ Dirac equation (10. 1 |ˆ) ) . is genuinely two-dimensional and. .

392) into diagonal form. − 1 r2 (j − 1 )(j + 1 ) − Z 2 e4 2 2 ±i Ze2 (j + )(j + ) − Z 2 e4 We seek to bring (10. such transformation can be chosen as to diagonalize the second term.2 (j) + 1) 2 Ze2 + + r2 r 2 − m2 f1. We can. therefore. λ1 (j) and λ2 (j). (6. we want to determine solely the spectrum. . 1 |ˆ) = − Yjm (j ˆ r 2 2 1 2 h± (r) g± (r) Yjm (j − 1 .10.395) (10.151]. Any similarity transformation leaves the ﬁrst term in (10.199).396) and (10. property (6. except for the slight diﬀerence in the expression of λ1. (10. 10.392) h± (r) g± (r) = 0.390) (10.396) and λ2 (j) [λ2 (j) + 1] (10. 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 . 1. 1 |ˆ) are r 2 2 ˆ eigenfunctions of L2 .2 (j) being 2 j = 1 . m and expand φ± (r) = Using σ · r Yjm (j ± 1 . rather than = 0. in the present treatment. . as in the case of pionic atoms.2 (j) as given by (10.394) Equation (10.2 (j). hence.2 (r) = 0 (10. . 1 |ˆ) r 2 2 2 2 r r . conclude that the . a speciﬁc pair of total spin-orbital angular momentum quantum numbers j. namely.397) This equation is identical to the Klein-Gordon equation for pionic atoms written in the form (10. . involving the 2 × 2 unit matrix. However.392) reads then in the diagonal representation 2 ∂r − λ1. 3 . .f. . We select. 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. and except for the fact that 2 2 we have two sets of values for λ1.393) but do not explicitly consider further the wavefunctions.10: Dirac Particles in Electromagnetic Field 339 ˆ momentum operator L2 [cf. 6. Since. namely. 1 |ˆ) + r Yjm (j + 1 .395. the eigenvalues are independent of m. together with the orthonormality of these two states leads to the coupled diﬀerential equation 2 ∂r + 2Ze2 + r 2 −m 1 0 0 1 ±iZe2 1 2 3 2 (10. the missing additive term − 1 .5 [c.186)].391) derived in Sect.200). 1 |ˆ) r 2 (10. Obviously. which states that the states Yjm (j ± 1 . unaltered. (6. not the wave functions.392).151).. the values of the argument of λ1.2 (j)[λ1.

401) if one attributes to the respective 1 states the angular momentum quantum numbers = j + 2 and = j − 1 . . −j + 1. . Using (10. . Expanding the energies in terms of this parameter allows one to identify is determined by Z the relationship between the energies 1 and 2 and the non-relativistic spectrum.403) √ Z 2 e4 2 ED (n. One can equate (10.395) and 2 corresponds to λ2 (j) as given in (10. . .401) corresponds to a non-relativistic state 2 with quantum numbers n. (10. . The magnitude of the relativistic eﬀect 2 2 e4 . .400) (10. . . 2 2 1 ≈ m − (10. albeit with some modiﬁcations.398) 2 = 1 + m . (10. 10. j . . 2n2 = 0.397) is again given by eq. 2. and spin-orbital angular momentum j = − 1 . m = − . . −j + 1. . One may also state 2 this in the following way: (10. . . 2 2 m = −j. . . . . 1. n − 1 . . j = − 2 . j = 1 . .398. the second term on the r.402) with (10. j = + 1 . accordingly. j where 1 corresponds to λ1 (j) as given in (10. . counts the nodes of the wave function. n − 1 . j = 1.399) 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. For a given value of n the energies 1 and 2 for identical j-values correspond to mixtures of states 1 with orbital angular momentum = j − 2 and = j + 1 .400) corresponds to a non-relativistic state with quantum numbers n.2 .396). . of these equations describe the binding energy. . . . 2.404) = 0. 1 ms = ± 2 (10. and spin-orbital angular momentum j = + 1 . . .h. m = −j. . 1. . (10. Obviously. . .s. 10. . including spin.396) we obtain. (10.203). . . .395. . 3. 2. . 1. . . m) 2 n = 1. 3 . the stationary states have binding energies E = − mZ 2 e4 n = 1. . These considerations 2 are summarized in the following equations 1 ED (n. . . . j .399) n = 0. It is given by n = n + + 1 where is the orbital angular momentum quantum number and n = 0.401) These expressions can be equated with the non-relativistic spectrum. One obtains in case of (10. . 1. .340 Relativistic Quantum Mechanics spectrum of (10. 1 = 1 + m Z 2 e4 (n + 1 + (j+ 1 )2 − Z 2 e4 )2 2 Z 2 e4 (n + (j+ 1 )2 − Z 2 e4 )2 2 . . In case of non-relativistic hydrogen-type 1 atoms. .402) In this expression n is the so-called main quantum number. − + 1. . = 1 + m Z 2 e4 √ ( n − − 1 + 2 − Z 2 e4 )2 . (10. −j + 1. . ( +1) − Z 2 e4 )2 (10. . 2. .400) and (10. . . m) = 1 + m (n− + . m = −j. 1.

in fact.3.405).405)] spectrum of the hydrogen atom.402)] and the relativistic [cf. 2. relativistic. 10. . ± 3 . r 2 2 2 . (10. j (10. One can combine the expressions (10. .551176 spectr.40147 -1. . .0041 keV and e2 = 1/137. in case of heavier nuclei the kinetic energy of bound electrons in the ground state scales with the nuclear charge Z like Z 2 such that in case Z = 100 the kinetic energy is of the order of the rest mass and relativistic eﬀects become important. −j + 1. The reason is that the mean kinetic energy of the electron in the hydrogen atom. 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. .1.e.60601 -3.51176 ⇑ ⇑ ⇑ ⇑ 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 10.551177 ⇑ .403. binding energy / eV Eq. m) = m 1 + Z 2 e4 (n−j −1 + 2 (j+ 1 )2 − Z 2 e4 )2 2 n = 1. 0 0 1 1 0 1 1 2 2 spinorbital ang. Degeneracies are denoted by ⇑. The table entries demonstrate that the energies as given by the expression (10. the non-relativistic and relativistic energies are hardly discernible. . i. j 1 2 1 2 1 2 3 2 1 2 1 2 3 2 3 2 5 2 341 rel. case these six states are split into energetically diﬀerent 2p 1 and 2p 3 states. . mom.402.10: Dirac Particles in Electromagnetic Field main quantum number n 1 2 2 2 3 3 3 3 3 orbital angular mom. the 2p 3 states with r 2 2 2 2 1 involve four degenerate states corresponding to Y 3 m (1. for example..404) ﬁnally into the single formula ED (n. (10. notation 1s 1 2 non-rel. the nonrelativistic degeneracy for the six 2p states of the hydrogen atom: in the present.60583 -3. much less than the rest mass of the electron (511 keV). 1.2: Binding energies for the hydrogen (Z = 1) atom.3. (10. .405) In order to demonstrate relativistic eﬀects in the spectrum of the hydrogen atom we compare in Table 10. n − 1 j = 1 2 1 2 ± m = −j. j. . is in the range of 10 eV. 10. i. . The energies were evaluated with m = 511. binding energy / eV Eq. . relate closely to the corresponding nonrelativistic states. 2 |ˆ) for m = ± 1 .551178 ⇑ .405) in terms of the non-relativistic quantum numbers n. This is clearly demonstrated by the comparision of non-relativistic and relativistic spectra of a hydrogen-type atom with Z = 100 in Table 10.40146 ⇑ ⇑ -1.402) -13..036 by means of Eqs.2 the non-relativistic [cf. However.405 ) -13. Of particular interest is the eﬀect of spin-orbit coupling which removes.1.40151 ⇑ . (10. 1 |ˆ) for m = ± 1 . (10. .10. = 0.e.

In order to investigate further the deviation between relativistic and non-relativistic spectra of hydrogen-type atoms we expand the expression (10. 0 0 1 1 0 1 1 2 2 spinorbital ang. mom. notation 1s 1 2 2s 1 2 2p 1 2 2p 3 2 3s 1 2 3p 1 2 3p 3 2 3d 3 2 3d 5 2 Table 10.1 ⇑ . (10.405 ) -161. . binding energy / keV Eq.405) to order O(Z 4 e8 ).405) reads 1 (10. j 1 2 1 2 1 2 3 2 1 2 1 2 3 2 3 2 5 2 Relativistic Quantum Mechanics non-rel.407) 1 2 This expression allows one. Introducing α = Z 2 e4 1 and β = j + 2 (10. 10.1 ⇑ ⇑ ⇑ ⇑ rel.3 spectr.2.342 main quantum number n 1 2 2 2 3 3 3 3 3 orbital angular mom. for example. Tables 10.406) α √ 1+ (n − β + β 2 −α)2 The expansion (10.9 ⇑ . (10.402) -136.402. 32 (10. (10. to estimate the diﬀerence between the energies of the states 2p 3 and 2p 1 (cf.409) where V (r) is spherically symmetric. j.3). 0. 0) (10. An example for such potential is the Coulomb potential V (r) = −Ze2 /r considered further below.35.2 -17.3: Binding energies for the hydrogen-type (Z = 100) atom.206) provides in the present case ED (n.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).15.1 -34.405).036 by means of Eqs. We assume for the wave function the stationary state . 1 2 2 2 2 E (2p 3 ) − E (2p 1 ) ≈ − 2 2 mZ 4 e8 2 · 23 1 −1 2 = mZ 4 e8 .15. Degeneracies are denoted by ⇑. It holds for n = 2 and j = 3 . m) ≈ m − mZ 2 e4 mZ 4 e8 − 2n2 2n3 1 j+ − 3 4n + O(Z 6 e12 ) . The energies were evaluated with m = 511. (10.8 ⇑ .6 -42.0 ⇑ ⇑ -15.0041 keV and e2 = 1/137. 0.10. binding energy / keV Eq.

according to (10. 10. 1 |ˆ) r 2 2 = i ∂r + j+ r 1 2 3 2 f (r) Yjm (j − 1 .413.413) ˆ σ · p g(r) Yjm (j − 1 . The arguments above allow one to eliminate the angular dependence by expanding Φ(r) and X (r) 1 in terms of Yjm (j + 2 .197. m-values. but are timeindependent..417) c(r) d(r) X (r) 1 1 1 1 Yjm (j + 2 . 1 |ˆ) r 2 2 = i ∂r + −j r 1 g(r) Yjm (j + 2 . The Dirac equation reads then. 2 |ˆ) r r r . 1 |ˆ).f.2 degrees of freedom.416) The diﬀerential equations (10. i. X (r) ∈ C2 describe the spatial and spin. i.5. 1 |ˆ) r 2 1 1 = ∂r r r r (10. 6. 2 |ˆ) and does r 2 ˆ has odd parity and it holds [c.411. 10. ˆ σ · p X + m Φ + V (r) Φ ˆ σ · p Φ − mX + V (r) X = = Φ X (10.10. 1 |ˆ) introduced in r 2 2 1 Sect. 6. 2 |ˆ) + r Yjm (j − 2 . (6.e.412) ˆ In this equation a coupling between the wave functions Φ(r) and X (r) arises due to the term σ · p. 1 |ˆ) and Yjm (j − 1 .e. σ · p ˆ σ · p f (r) Yjm (j + 1 . 1 |ˆ) .411) (10. This term has been discussed in detail in Sect. r 2 2 (10.198)] not couple states with diﬀerent j.10: Dirac Particles in Electromagnetic Field form Ψ(xµ ) = e−i t 343 Φ(r) X (r) . 6. in particular. 168]: the term is a scalar (rank zero tensor) in the space of the spin-angular momentum states Yjm (j ± 1 . r 2 (10. 1 |ˆ) r 2 2 (10. 2 |ˆ) + r Yjm (j − 2 . pp.345). 1 |ˆ) r 2 = i ∂r − 1 2 r g(r) Yjm (j + 1 . the term is block-diagonal in the space spanned by the states Yjm (j ± 1 . 1 |ˆ) r 2 2 = i j+ ∂r + r j+ r 1 2 1 r f (r) Yjm (j − 2 .414) ˆ σ · p rf (r) Yjm (j + 1 .415) 1 ˆ σ · p rg(r) Yjm (j − 2 . = r (10. (10. r r 2 2 2 a(r) b(r) 1 1 1 1 Yjm (j + 2 .410) 1 where Φ(r).232. 2 |ˆ) r Φ(r) r . 1 |ˆ) .5 [see.414) These equations can be brought into a more symmetric form using ∂r + which allows one to write (10.. 10.412) are four-dimensional with r-dependent wave functions.

Inserting (10. 2 |ˆ) r r where the factors i and −1 have been introduced for convenience.421) g2 (r) X (r) 1 1 − Yjm (j + 2 . such expansion must include states with all possible j. 1 |ˆ) r Φ(r) 2 2 r = (10. only a(r).417) of the form f1 (r) i Yjm (j + 1 .157).415.411. Accordingly.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 (10.419) Obviously. g2 (r) ∂r + 1 2 f1 (r) − [ + m − V (r) ] g1 (r) = 0 (10.417) into (10.418) holds for f1 (r).416). 2 |ˆ) = (10. using (10. m values. g1 (r) ∂r − j+ r j+ r 1 2 g1 (r) + [ − m − V (r) ] f1 (r) = 0 ∂r + and for f2 (r).423) . m pair contribute. 10.412). 2 |ˆ) r r f2 (r) 1 1 r Φ(r) i r Yjm (j − 2 . and multiplying by r results in the following two independent pairs of coupled diﬀerential equations i ∂r − j+ r j+ r 1 2 d(r) + [ m + V (r) − ] a(r) = 0 i and i ∂r + 1 2 a(r) + [ −m + V (r) − ] d(r) = 0 (10. we consider the case that only states for one speciﬁc j. 10. (10. Presently.420) g1 (r) X (r) 1 1 − Yjm (j − 2 .418) ∂r + j+ r 1 2 c(r) + [ m + V (r) − ] b(r) = 0 i j+ ∂r − r 1 2 b(r) + [ −m + V (r) − ] c(r) = 0. According to (10. c(r) are coupled. there exist two independent solutions (10. d(r) are coupled and b(r).344 Relativistic Quantum Mechanics In general. the orthonormality property (6.

except for the opposite sign of the term (j + 1 ).367) which is spherically symmetric such that equations (10. We note that (10. one demonstrates ﬁrst that the wave function at r → 0 behaves as rγ for some suitable γ. Dirac Particle in Coulomb Field .405) of the relativistic hydrogen atom the corresponding states have quantum numbers n = 1. 2s 1 .422)..423) apply for V (r) = −Ze2 /r. one demonstrates then that the wave functions for r → ∞ behaves as exp(−µr) for some suitable µ. 3d 3 . and obtains ﬁnally a polynomial function p(r) such that rγ exp(−µr)p(r) solves (10. Similarly. to states with angular momentum = j + 2 .409). . in the non-relativistic limit wave functions f2 (r) 1 1 i Yjm (j − 2 .10: Dirac Particles in Electromagnetic Field 345 Equations (10. . Hence. The solution of (10.422) describes the states 2p 1 . enforcing the polynomial to be of ﬁnite order leads to discrete eigenvalues . We assume the 4-vector potential of pure Coulomb type (10. Behaviour at r → 0 We consider ﬁrst the behaviour of the solutions f1 (r) and g1 (r) of (10.422. 10. 2 |ˆ) r Ψ(xµ ) ≈ . According to the discussion of the spectrum (10.422) follows in this case from the same procedure as that adopted for the radial wave function of the non-relativistic hydrogen-type atom.422) and (10.420) is the large component and X is the small component.422). together with the appropriate boundary conditions at r = 0 and r → ∞. 1 |ˆ) r i 2 2 . 3p 3 . the ones given in (10. 1. 2p 3 .e. 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. In the non-relativistic limit..422). According to this procedure. .Wave Functions We want to determine now the wave functions of the stationary states of hydrogen-type atoms which correspond to the energy levels (10.426) . (10.422) corresponds to states f1 (r) Yjm (j + 1 . (10. 3s 1 . (10.421). the radial wave functions for Dirac particles in the potential (10. 3d 5 . Φ in (10.10. Equation (10.423).405). correspodingly the states We consider ﬁrst the solution of (10. = 0. . .e.420). for small r. Hence.405). 2 2 2 2 2 2 = j − 1 2 and. 2 2 2 (10. . . 3p 1 . . i. namely.425) r 0 covers states with angular momentum 1s 1 . (10.422) determines solutions of the form (10. n − 1.423) are identical. etc. etc. the 2 equations determine. determining wave functions of the type (10.424) Ψ(xµ ) ≈ r 0 1 i. 2. (10.422) near r = 0.

hence.266.427) makes only the positive solution 2 possible. in case j + 1 )2 < Ze2 . ) f1 (r) ( + m ) g1 (r) (10. Only the exponentially decaying solution is admissable and. 10. determined that the solutions f1 (r) and g1 (r). for small r. for the particle density holds then ρ(r) ∼ |rγ−1 |2 = Behaviour at r → ∞ For very large r values (10. (10.429) This equation poses an eigenvalue problem (eigenvalue −γ) for proper γ values. g1 (r) →∞ ∼ e−µr .432) = = − ( − m.435) . Such r-dependence 2 would make the expectation value of the potential r2 dr ρ(r) 1 r (10. (10. becomes imaginary. according to (10.428) = 0.427) = = 0 0. we conclude f1 (r) →∞ ∼ e−µr .431) inﬁnite since.267. hence. assume the r-dependence in (10. 10.434) √ The solutions of these equations are f1 . r2 (10.427).430) 2 Note that the exponent in (10.346 Setting f1 (r) yields γ b rγ−1 − (j + 1 ) b rγ−1 + Ze2 a rγ−1 2 γ a rγ−1 + (j + 1 ) a rγ−1 − Ze2 b rγ−1 2 or γ + (j + 1 ) −Ze2 2 2 Ze γ − (j + 1 ) 2 a b →0 Relativistic Quantum Mechanics ∼ a rγ .433) = = ( m2 − 2 ) g1 (r) ( m2 − 2 ) f1 (r) (10.420). g1 (r) →0 ∼ b rγ (10. g1 ∼ exp(± m2 − 2 r). (10. The assumed r-dependence in (10.427) with 1 γ = (j + )2 − Z 2 e4 .422) becomes ∂r g1 (r) ∂r f1 (r) Iterating this equation once yields 2 ∂r g1 (r) 2 ∂r f1 (r) 1 . We have. µ = m2 − 2 (10. One obtains γ = ± (j + 1 )2 − Z 2 e4 .

437) where they ˜ ˜ cancelled in case f1 = g1 = a.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 ˜ (10.441) The last two terms on the l.10: Dirac Particles in Electromagnetic Field For bound states holds 347 < m and. without loss of generality we can choose ˜ f1 (r) = φ1 (r) + φ2 (r) . However. hence.440) (10.436). √ g1 (r) = − m − a e−µr .10.h.438) (10. we set f1 (r) g1 (r) = = √ √ ˜ m + e−µr f1 (r) −µr = = 0 0 (10. µ is real.441) correspond to (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.s.439) − m− e g1 (r) ˜ where µ is given in (10.445) . Accordingly. We also introduce the new variable ρ = 2µ r . Equation (10.436) f1 (r) = Insertion into (10. g1 (r) = φ1 (r) − φ2 (r) ˜ (10.435 ).434) √ m + a e−µ r . (10.437) (10.444) (10.442) which leads to a partial cancellation of the asymptotically dominant terms. (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. Let us consider then for the solution of (10.434) results in √ √ (m − ) m + a − (m − ) m + a √ √ (m + ) m − a − (m + ) m − a which is obviously correct. of both (10. Solution of the Radial Dirac Equation for a Coulomb Potential To solve (10. ρ m+ ρ [∂ρ − (10.443) From this results after a little algebra j+1 m + Ze2 2 ] (φ1 − φ2 ) − (φ1 + φ2 ) + φ2 = 0 ρ m− ρ j+1 m − Ze2 2 [∂ρ + ] (φ1 + φ2 ) + (φ1 − φ2 ) − φ2 = 0 .440) and (10.

448) φ2 (ρ) = ρ (10.447) of the form n φ1 (ρ) = ργ s=0 n γ s=0 αs ρ2 βs ρ2 (10.449) into (10.f.430) which conform to the proper r → 0 behaviour determined above [c. 10.448.453) (s + γ)2 + Z 2 e4 − (j + 1 )2 2 2 √ Ze m2 − 2 βs .451) follows βs−1 = = Ze2 s+γ + √ m2 − s+γ − βs + m2 Z 2 e4 m2 − 2 − (j + 1 )2 2 2 √ Ze m2 − 2 2 s+γ − βs (10.446) (10.451) 1 √mZe − (j + 2 ) 2 2 αs .446.430) one can write this βs = s+γ − 2 √ Ze m2 − 2 s (s + 2γ) βs−1 . 10. Inserting (10. = m− Ze2 βs s + γ − √m2 − 2 2 (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 (10.449) for γ given in (10.450) follows Ze2 m2 − 2 βs − βs−1 =0 (10. (10. Using (10. 10.446 .430)].454) .447) We seek solutions of (10.450) −√ mZe2 m2 − 2 βs ρs+γ−1 = 0 mZe2 m2 − αs (s + γ) β2 + (j + 1 ) αs + √ 2 s 2 +√ From (10.426–10.447) leads to (s + γ) αs + (j + 1 ) βs − √ 2 s Ze2 m2 − 2 αs (10. (10.452) From (10.

for the states 2p 1 . According to the deﬁnitions (10.448. 3p 1 .10. .457) One can relate the polynomials φ1 (ρ) and φ2 (ρ) deﬁned through (10. Comparision with (10. c c(c + 1) 2! (10.456) − so 2 √mZe m2 − 2 (−so )(1 − so )(2 − so ) .10: Dirac Particles in Electromagnetic Field Deﬁning so = √ one obtains βs = = .452) follows j+ αs = 1 2 349 Ze2 m2 − 2 −γ (10. hence. The expressions (10. c. 10. 2γ + 1.462) 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. m (n ) = . with the associated Laguerre polynomials L(α) = F (−n. ρ) .456) and (10. (s − so ) β0 s! (2γ + 1)(2γ + 2) · · · (2γ + s) (10.. is an integer. i. .405).430. For example. . equivalently. 10. . namely. . x) . conclude 2 2 2 that the polynomials in (10. n = 0. n It holds φ1 (ρ) φ2 (ρ) = = ργ F (−so .461) In order that the wave functions remain normalizable the power series (10.405) and the ensuing so values ( 10.. α + 1. 1.457) with the conﬂuent hypergeometric functions F (a. (10.459) − 2 √mZe m2 − 2 (10. 2γ + 1. β0 j+ 1 2 (10.455) s − so βs−1 s(s + 2γ) (s − 1 − so )(s − so ) βs−2 (s − 1)s (s − 1 + 2γ)(s + 2γ) (1 − so )(2 − so ) . We can. 3d 3 holds n = 1. 2.449) must be of ﬁnite order. (s − so ) β0 s! (2γ + 1)(2γ + 2) · · · (2γ + s) (10. . so = n . 10.449) and (10.457) for βs and αs imply that so must then be an integer.461 for values given by (10. .456. 2. 1.448.455) are ﬁnite.e. x) = 1 + a a(a + 1) x2 x + + . = From (10..455) this conﬁnement of so implies discrete values for . 10. in fact. . ρ) so β0 ργ F (1 − so .458) or.460) (10. .405) allows one to identify n = n − j + 2 which. . This requires that all coeﬃcients αs and βs must vanish for s ≥ n for some n ∈ N.

461) are to be chosen to satisfy a normalization condition and to assign an overall phase.465) (10. given by expression (10.460.464) 0 The evaluation of the integrals. are two-dimensional vectors determined through the explicit form of the spin-orbital angular momentum states Yjm (j ± 1 .460. The complete wave function is given by the following set of formulas 1 Ψ(n. The normalization integral is then ∞ 0 π 2π r2 dr 0 sin θdθ 0 dφ (|Φ(r)|2 + |X (r)|2 ) = 1 (10. 2µr) (2µ) 2 Γ(2γ + 1) 3 N = and m 4m )Γ(2γ + n + 1) (n +γ)m (n +γ)m (10.147. = j + 2 . Berlin. m. = j + 2 . (n + γ)m N (2µr)γ−1 e−µr − κ F (−n .267). 6. Sect.421). Due to the form (10.439). 10. m|xµ ) = e−i t iF1 (r) Yj.148). 10. j.421) with radial wave functions f1 (r). where2 F± (κ|r) = G1 (r) = F+ (κ|r) .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 . The coeﬃcients β0 in (10.m (j − 1 . (10. 1990). 1 |ˆ) in (6. 10.461). and (10. 1 |ˆ) 2 2 r κ = j+ 1 2 (10. 10. which involve the conﬂuent hypergeometric functions in (10. j.461). (Springer.463) where Φ and X .158) of the latter states absorb the angular integral in (10.6. 2µr) (10. 2 .438.460.421) correspond to non-relativistic states with orbital angular momentum 1 1 = j + 2 . 1 |ˆ) 2 2 r G1 (r) Yj. 2γ + 1.467) ± n F (1 − n . as in (10.157.421)] ∞ dr (|f1 (r)|2 + |g1 (r)|2 ) = 1 (10. 9. 6. can follow the procedure adopted for the wave functions of the non-relativistic hydrogen atom and will not be carried out here. Greiner. g1 (r) determined by (10. The orthonormality 2 2 r properties (6. They are described through quantum numbers n. 2γ + 1. is time-independent. (10.463) and yield [note the 1/r factor in (10.350 Relativistic Quantum Mechanics Altogether we have determined the stationary states of the type (10.410) of the stationary state wave function the density ρ(xµ ) of the states under consideration. The wave functions (10.442).466) F1 (r) = F− (κ|r) .469) This formula has been adapted from ”Relativistic Quantum Mechanics” by W.

One can verify. .421) are governed by the radial Dirac equation (10.469) that the following wave functions result Ψ(n. in the non-relativistic limit.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.m (j + 1 . 1 |ˆ) 2 2 r G2 (r) Yj. become f2 (r) 1 1 i Yjm (j − 2 . 2s 1 .469). 3d 5 . 2 accordingly. describes the complementary set of states 1s 1 .10. 2 |ˆ) r Ψ(xµ ) ≈ e−1 t . = j − µ 1 2 .472) F2 (r) = F− (κ|r) .469).465–10.10: Dirac Particles in Electromagnetic Field 351 We want to consider now the stationary states of the type (10. this wavefunction has an orbital angular momentum quantum number = j − 1 and. obtained closed expressions for the wave functions of all the stationary bound states of relativistic hydrogen-type atoms. etc. (10. not not 2 2 2 2 2 2 covered by the wave functions given by (10. G2 (r) = F+ (κ|r) . hence.467–10. 2p 3 .422) to (10. tracing all steps which lead from (10.471) (10. where F± (κ|r) are as given in (10. j. 3s 1 . m|x ) = e −i t iF2 (r) Yj.421) which. We have. 1 |ˆ) 2 2 r κ = −j − 1 2 (10. The radial wave functions f2 (r) and g2 (r) in (10.470) r 0 Obviously.m (j − 1 . 3p 3 .

352 Relativistic Quantum Mechanics .

i..229). 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. diﬀerent L(w. indeed. must be two-dimensional irreducible representations of the Lorentz group. a(z) and b(z).2) (11. may be exploited to express the Lorentz-invariant equations of relativistic quantum mechanics. This is. ϑ)µ ν correspond to diﬀerent a(z) and b(z). isomorphic to the natural representation.225–10. z ∈ C ˜ Because of its block-diagonal form each of the diagonal components of S(z). a complex variable z. ϑ) for the ˜ in the chiral representation as given by (10. We will proceed by building as much as possible on the results obtained sofar in the chiral representation of the Dirac equation. ˜ We have altered here slightly our notation of S(w. i.1) (11. expressing its dependence on w.. 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. in a sense.4) = exp = w − iϑ .262). what will be achieved in the following. The lower dimensionality of a(z) and b(z) implies. ϑ through 3.e.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. ϑ)µ ν . We will characterize the space on which the transformations a(z) We will see below that the representations a(z) and b(z) are. in 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 (11.3) (11. ϑ). in fact. 1 353 .e.Chapter 11 Spinor Formulation of Relativistic Quantum Mechanics 11. Starting point is the Lorentz transformation S(w.

± 1 represents the familiar spin– 1 states. Ψ2 (xµ ) and Ψ3 (xµ ). in 2 fact. 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 .6) as given by (5. This characterization allows one to draw conclusions regarding the state space in which a(z) and b(z) operate. ϑ ∈ R3 (11. 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. formulate the Dirac equation. − 1 2 2 z = iϑ.7) where “∼” stands for “transforms like”. +1 2 2 |1.8) We like to stress that there exists. A First Characterization of the Bispinor States We note that in case w = 0 the Dirac transformations are pure rotations. are elements of SU(2) and correspond. to the rotational transformations of spin. namely. +1 2 |1.354 Spinor Formulation and b(z) act. ϑ ∈ R3 . (11. a distinct diﬀerence in the transformation behaviour of ˜ ˜ ˜ ˜ Ψ1 (xµ ). In this case a(z) and b(z) are identical and read a(i ϑ) = b(i ϑ) = exp 1 − ϑ·σ 2 . + 1 2 2 state 2 ∼ | 1 .e. For ϑ = (0. Here | 1 . the operator ∂µ and the Pauli matrices σ in the new representation and.. a space of vectors state1 for which holds state2 state 1 ∼ | 1 . Aν . neutrino equation. the so-called spinor space. (11. Ψ4 (xµ ) transform according to ˜ ˜ ˜ and Ψ b(z). to 1 ∗ ∗ z ·σ (11.e. Ψ4 (xµ ) in case z = w + iϑ for w = 0. and the Klein–Gordon equation in the spinor representation. Relationship Between a(z) and b(z) The transformation b(z) can be related to the conjugate complex of the transformation a(z). usually expressed 2 as product of rotations and of functions of Euler angles α. 2 2 2 ˜ Since a(z) acts on the ﬁrst two components of the solution Ψ of the Dirac equation.. ϑ ∈ R3 . i. express 4-vectors Aµ . (11. β.9) a∗ (z) = exp 2 . γ (see Chapter 5). −1 2 2 z = iϑ.243). i. however. θ ∈ R3 . β. −1 2 2 |1. for z = iϑ. b(z). Ψ2 (xµ ) transform according to a(z) whereas Ψ3 (xµ ). will establish the map between L(w.5) The transformations in this case. ﬁnally. In this case holds a(z) = b(z) ˜ 1 (xµ ). ϑ)µ ν and a(z).1 states as described by Dm m (ϑ).

17) We conclude. one can express a∗ (z) −1 = exp 1 − z ∗· σ 2 = b(z) .10) From this one can derive b(z) = where = 0 1 −1 0 .18) with a(z) given by (11. therefore. Ψ (xµ ) as well as Ψ (xµ ). 11. that they alter w into −w. . σ2 .f. using f (a) −1 = f ( a −1 ). The eﬀect of inversion on Lorentz transformations is.11) one ﬁrst demonstrates that for and −1 as given in (11.11.14) . hold −1 = 1 One notices then.1) can be written in the form ˜ S(z) = a(z) 0 0 a∗ (z) −1 (11. however. in fact.1: Lorentz Transformation of Dirac Bispinor ∗ ∗ ∗ where σ ∗ = (σ1 . but leave rotation angles ϑ unaltered.224.12) To prove (11. µ ). −1 a∗ (z) −1 (11.12). 1.2. Hence. Ψ (xµ ) even though these pairs of ˜ ˜3 ˜4 featuring the components Ψ1 (x ˜ 2 components transform independently of each other. σ3 ). one can show −1 −1 −1 = −σ1 = σ2 = −σ3 ∗ = −σ1 ∗ = −σ2 σ2 −1 (11.4) and . 355 (11.15) (11. a∗ (z) −1 = exp 1 ∗ z σ∗ 2 −1 . (11. Spatial Inversion ˜ One may question from the form of S(z) why the Dirac equation needs to be four-dimensional. 11.16) σ = −σ ∗ . that the transformation (11. 11. (11. One can readily verify [c. (11. −1 = ∗ −σ3 = −σ ∗ . ∗ σ 3 = σ3 . (11.13) Explicit matrix multiplication using (5.224)] ∗ σ1 = σ1 .12) does.11) 0 −1 1 0 = . a result to be used further below. 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.10. This demonstrates that a(z) is the ˜ transformation which characterizes both components of S(z).12) yields σ1 σ3 or in short σ Similarly. ∗ σ 2 = − σ2 . −1 given by (11. (5.

ϑ ∈ R3 . (11. 2.e. In fact. (11.f. One can verify this by evaluating the determinant of a(z) det( a(z) ) = det e 2 z·σ 1 (11.2. For the general case the proof.21) = etr( 2 z·σ) = 1 1 (11.22) which follows from the fact that for any complex.G.24) = etr(M ) (11.21) together with matrix multiplication as the binary operation forms a group.20) ˜ Obviously. ˜ ˜ P S(w + iϑ) P = S(−w + iϑ) = b(z) 0 0 a(z) . 2 = 1 i. does not suﬃce for particles like the electron which obey inversion symmetry.3.224)] tr( σj ) = 0 .A. hence. Part (Elsevier. 11. the Lorentz invariant equation for neutrinos is only 2–dimensional.10. 1990). The transformation S(z) in the transformed space is then ˜ P 1. C) = { M. B¨uerle and E. M is a complex 2 × 2–matrix. C) deﬁned in (11. (5. P −1 = P. non-singular matrix M holds2 det eM and from [c. However. de Kerf Lie a Algebras. j = 1.1: Show that SL(2.1) We have pointed out that a(iϑ).356 Spinor Formulation ˜ Let P denote the representation of spatial inversion in the space of the wave functions Ψ. This implies ˜ Ψ1 ˜ Ψ2 P ˜ Ψ3 ˜ Ψ4 ˜ Ψ3 ˜ Ψ4 = Ψ1 ˜ ˜ Ψ2 . a(w + iϑ) for w = 0 is an element of SL(2.C) and SO(3.19) i. Amsterdam. 3 . However. the space spanned by only two of the components of Ψ is not invariant under spatial inversion and.2 Relationship Between the Lie Groups SL(2. which describes pure rotations..23) Exercise 11. (11. for particles like the neutrinos which do not obey inversion symmetry two components of the wave function are suﬃcient. Exercise 1. 2 .e. can be found in G. det(M ) = 1 } . Obviously. is an element of SU(2). based on the Jordan–Chevalley theorem.A. The proof of this important property is straightforward in case of hermitian M (see Chapter 5). the transformations a(z) and b(z) become interchanged..

Argue why M (Aµ ) = σµ Aµ provides a bijective map. σ2 . Noting that for covariant vectors according to (10. −→ µ Consistency of (11.32) . in fact. The function M (Aµ ) is bijective. one wishes that the deﬁnition (11.2: Show that σ0 . the matrix M (Aµ ) is hermitian as can be seen from inspection of (11.28) σµ = 1 0 0 1 . We will prove below [cf.2. 1 0 0 −1 . (11. or+ thochronous Lorentz transformations.C) and SO(3. (11.26) or from the fact that the matrices σ0 . (11. one can provide a simple expression for the inverse of M (Aµ ) M = M (Aµ ) ↔ Aµ = 1 tr M σµ 2 .30).27) −→ Straightforward transformation into another frame of reference would replace σµ by σµ . (11.29) where we used (10. 0 −i i 0 .30) Since the components of Aµ are real.1) Mapping Aµ onto matrices M (Aµ ) 357 We want to establish now the relationship between SL(2.25) of the matrix M (Aµ ) is independent of the frame of reference.2: Lie Groups SL(2.135)] this transformation behaviour.25) σ0 σ1 σ2 σ3 The quantity σµ thus deﬁned does not transform like a covariant 4–vector. (11. Starting point is a bijective map between R4 and the set of two-dimensional hermitian matrices deﬁned through M (Aµ ) = σµ Aµ where (11. i.75) holds aν = Lν µ aµ one realizes that σµ transforms inversely to covariant 4–vectors.76).25) can also be written M (Aµ ) = A0 + A3 A1 − iA2 A1 + iA2 A0 − A3 . σ2 .28) and (11. in a transformed frame should hold Lρ ν σ µ Aµ σµ A µ . σ3 provide a linear–independent basis for the space of hermitian 2 × 2–matrices. The following important property holds for M (Aµ ) det ( M (Aµ ) ) = Aµ Aµ which follows directly from (11. Demonstrate that (11. C) and the group L↑ of proper. σ1 .e.31) holds. (11. M (Aµ ) according to (11. σ3 are hermitian.27) requires then σ µ Aµ Lν µ σµ = σν (11.. 0 1 1 0 . σ1 . In fact.11. 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 .26) (11.31) Exercise 11.

33) conserves the determinant of M . C) and for any Aµ there exists an A µ such that M (A µ ) = a M (Aµ ) a† .33) to M (Aµ ) describing the action of the transformation in terms of transformations of Aµ .35) holds for this transformation A µ Aµ = Aµ Aµ (11.36) The suitable A µ can readily be constructed using (11.39) Exercise 11. Due to det(a) = 1 the transformation (11.31)] 1 a Aµ −→ A µ = tr a M (Aµ ) a† σµ 2 Because of (11. (11.37) which implies that (11.38) . a ∈ SL(2.358 Transforming the matrices M (Aµ ) Spinor Formulation We deﬁne now a transformation of the matrix M (Aµ ) in the space of hermitian 2 × 2–matrices a M −→ M = a M a† . (11.37) using (11. C) . for any a ∈ SL(2. The linear character of the transformation becomes apparent expressing A µ as given in (11. 11.40) 1 tr a σν a† σµ 2 Aν (11. In fact.f.33) det(M ) = = det( a M a† ) = det(a) det(a† ) det(M ) [det(a)]2 det(M ) = det(M ) . C) deﬁnes the transformation [c.37) deﬁnes actually a Lorentz transformation.31). In fact. (11. any a ∈ SL(2. Accordingly.40) is an element of L↑ .35) We now apply the transformation (11.32. it holds for the matrix M deﬁned through (11. (11. (11.33) This transformation conserves the hermitian property of M since (M )† = ( a M a† )† = (a† )† M † a† = a M a† = M (11.2.34) where we used the properties M † = M and (a† )† = a.3: Show that L(a)µ ν deﬁned in (11. + .25) Aµ = which allows us to express ﬁnally A µ = L(a)µ ν Aν . L(a)µ ν = 1 tr a σν a† σµ 2 . (11.

C) and of SO(3.1) L(a)µ ν provides a homomorphism 359 We want to demonstrate now that the map between SL(2.C) and SO(3.11.β γ. (11. (11. 1 ¯ and using the deﬁnition of Lµ ρ in (11.41) results in 1 ¯ Lµ ρ = 4 where Aαβγδ = ν Γ = a2 σρ a† 2 (11.42) = ν Deﬁning Γ = a† σµ a1 . (11. i.α.e. β δ = δ = = δ = δ 1 2 = 2 = 1 Aαβγδ (11.47) This completes the proof of the homomorphic property of L(a)µ ν .1).41) 1 tr a1 σν a† σµ 1 2 1 tr σν a† σµ a1 1 2 1 tr a2 σρ a† σν 2 2 1 tr a2 σρ a† σν 2 2 . β 2.1) deﬁned through L(a)µ ν [cf.40)] respects the group property of SL(2. C) and SO(3.43) (σν )αβ Γβα Γγδ (σν )δγ = ν. 2 (11. ..δ (11. ¯ Lµ ρ = L(a1 )µ ν L(a2 )µ ρ = L( product in SO(3.46) 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 .45) One can demonstrate through direct evaluation 2 α = 2 α = 2 α = = 2 α = 0 else β β γ γ = = = = γ = γ = 1.β γ.2: Lie Groups SL(2.C) )µ ρ (11.1) For this purpose one writes using tr(AB) = tr(BA) L(a1 )µ ν L(a2 )µ ρ = ν a1 a2 product in SL(2.44) (σν )αβ (σν )δγ .δ 1 4 Aαβγδ Γβα Γγδ α.

(11.52) where we employed (11. J3 of the Lorentz transformations Lµ ν in the natural representations. 0. 0.h. of (11.59) = = = = κ1 σ 0 + σ 0 κ† = κ 1 + κ† 1 1 κ1 σ 1 + κ1 σ 2 + κ1 σ 3 + σ 1 κ† 1 σ 2 κ† 1 σ 3 κ† 1 (11. 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 . 1 (11.36) M (Lµ ν Aν ) = a M (Aµ ) a† (11. 1.25) in the last step.s. 0. 0). 0. C) which correspond to K. 0. J Spinor Formulation The transformations a ∈ SL(2.53) Equating (11. J2 .2) which correspond to the generators K1 . Because of the condition det(a) = 1 only six independent real numbers actually suﬃce for the deﬁnition of a. −A0 . J1 . 1) − σ1 σ0 0 0 One can verify readily that 1 κ1 = − σ 1 2 (11. i. (11. noting the linearity of M (Aµ ). 0) ) σ0 A 1 − σ1 A 0 (11. To obtain the generator of a(z) corresponding to the generator K1 .48).49) we obtain using (11. For the r. correspond to the generators given by (10. K2 .48) yields for the l. 1 (11.58) .52) and (11. eight real numbers. correspondingly. 10.56) (11.48) We want to determine now the generators of the transformation a(z) as deﬁned in (11.49) assuming (note that g µ ν is just the familiar Kronecker δµν ) Lµ ν a = = gµν + 1 + 1 (K1 )µ ν κ1 . 1.54) This reads for the four cases Aµ = (1. C) as complex 2 × 2–matrices are deﬁned through four complex or. denoted below as κ1 . Aµ = (0. K3 .53) results in the condition σ 0 A1 − σ 1 A0 = ( κ 1 σ µ + σ µ κ † ) Aµ . 0). Aµ = (0. 0). To this end we consider inﬁnitesimal transformations and keep only terms of zero order and ﬁrst order in the small variables. we write (11. 0.e.47.57) .49).h. of (11.51) Insertion of (K1 )µ ν as given in (10.360 Generators for SL(2.51) ( 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 ) . Aµ = (0.50) (11.s.55) (11.. (11. M (Aµ + (K1 )µ ν Aν ) = = M (Aµ ) + M (Aµ ) − M ( (−A1 . 0.

one can show that the generators κ2 .1) acts on 4–vectors Aµ . λ3 corresponding to J1 .4: Show that the generators (11.60) We can. λ2 . − 1 . a1 1 a1 2 a2 1 a2 2 (11. κ3 of a(z) corresponding to K2 . 2 (11. 2 λ = i σ. hence. Exercise 11.63) where we extended the summation convention of 4-vectors to spinors. w describing boosts and ϑ describing rotations.3: Spinors 361 obeys these conditions. Similarly.3 Spinors Deﬁnition of contravariant spinors We will now further characterize the states on which the transformation a(z) and its conjugate complex a∗ (z) act. a(w − iϑ) = e− 2 (w − iϑ)·σ ∈ SL(2. J3 are given by 1 κ = − σ. (11.61) This identiﬁes the transformations a(z = w − iϑ) as representations of Lorentz transformations. 2 2 φ1 φ2 ∈ C2 . K3 and λ1 . ϑ) = ew·K + ϑ·J ∈ SO(3.64) .11. J2 . We consider ﬁrst the transformation a(z) which acts on a 2-dimensional space of states denoted by φα = According to our earlier discussion holds φ1 φ2 1 transforms under rotations (z = iϑ) like a spin– 2 state | 1 . C) correspond to the generators K and J of Lµ ν .2. C) acts on spinors φα ∈ C2 (characterized below) 1 (11. the so-called contravariant spinors. 11. state that the following two transformations are equivalent L(w.62) We denote the general (z = w + iϑ) transformation by φ α = aα β φβ = aα 1 φ1 + aα 2 φ2 .60) of a ∈ SL(2. Here a(z) = ( aα β ) = describes the matrix a(z). def α = 1. 2 (11. + 1 2 2 1 transforms under rotations (z = iϑ) like a spin– 2 state | 1 .

0. 1 .65) where | · · · 1 describes the spin state of “particle 1” and | · · · 2 describes the spin state of “particle 2” and (0. In this case spinors φα transform like spin– 1 states and an invariant.e. ± 1 . 0 = 2 2 m=± 1 2 1 1 (0.e. 0| 1 . which 2 can be constructed from products φ1 χ2 . . . invariant and as such has the necessary properties of a scalar3 product. . 1 ) stands for the Clebsch–Gordon coeﬃcient.. . . ± 1 . exchange of φα and χβ alters the sign of the expression. those of “particle 2” with the spinor χβ one can state that the quantity 1 Σ = √ 2 φ1 χ2 − φ2 χ1 (11. . 1 .. χβ . −m 2 2 2 2 1 1 1 1 2 (11. Using (0.69) (11. To arrive at a suitable scalar product we consider ﬁrst only rotational transformations a(iϑ). is the singlet state.68) (11. 0| .e. i. χβ .66) is a bilinear form.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.. invariant under transformations a(z).66) should constitute a singlet spin state. .. as contravariant spinors and to φα . etc. as we will demonstrate below such states are invariant under general Lorentz transformations a(z). In fact. this scalar product is anti-symmetric. The existence of a scalar product gives rise to the deﬁnition of a dual representation of the states φα denoted by φα . 1 ) = 2 2 2 2 2 2 2 2 √ ±1/ 2 and equating the spin states of “particle 1” with the spinor φα . The relationship between the two can be expressed φ1 φ2 = = 3 φ1 φ2 0 1 −1 0 (11.67) We will refer to φα . m 1 | . . Deﬁnition of covariant spinors Expression (11. i. (11. indeed. 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. Such a scalar product does. . should remain invariant under transformations a(iϑ). In the notation developed in Chapter 5 holds for the singlet state | . as covariant spinors. i. However.70) ‘Scalar’ implies invariance under rotations and is conventionally generalized to invariance under other symmetry trasnformations. The corresponding states are deﬁned through φ1 χ2 − φ2 χ1 = φ1 χ1 + φ2 χ2 It obviously holds χα = χ1 χ2 = χ2 −χ1 . 0| 1 . m. −m) | .

77) is Lorentz invariant.79) = a −1 T a κβ .70) and (11.70) is φ1 φ2 −1 363 = = −1 φ1 φ2 .72) 0 −1 1 0 Exercise 11.74) = = 0 1 −1 0 0 −1 1 0 (11.74)]. In the transformed frame holds φ α χα = aα β αγ aγ δ δκ φβ χκ . (11.75) is given by φα = as can be readily veriﬁed. Proof that φα χα is Lorentz invariant We want to verify now that the scalar product (11.72) in a notation analogous to that chosen for contravariant and covariant 4–vectors [cf.63. g µν of the Minkowski space [cf.72)] φα φ α = = = = αβ αβ φβ φβ 11 21 11 21 12 22 12 22 (11.77) (11. (10.69. (11.3: Spinors as can be veriﬁed from (11. 11. 11. Accordingly. 11.73) (11. For this scalar product holds φα χα = −χα φα .81) .10. (10. we will express (11.69.3.80) One can write in matrix notation aα β γ δκ αγ a δ αβ (11.11.73.76) αβ αβ The scalar product (11. (11. 10.78) aβ γ γδ φδ (11.71. 11. The transformation behaviour of φα according to (11. The inverse of (11.75) (11. 11.67) will be expressed as φα χα = φ1 χ1 + φ2 χ2 = φ1 χ2 − φ2 χ1 .1: Show that for any non-singular complex 2 × 2–matrix M holds M −1 = det(M ) M −1 T The matrices .68). −1 connecting contravariant and covariant spinors play the role of the metric tensors gµν .71) (11.

it holds φk = (φk )∗ .63). from (11..364 Using (11. As discussed above.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 (11.90) extending the summation convention to ‘dotted’ indices. the ˙ conjugate complex spinors φα need to be considered separately from the spinors φα . Obviously.e.2) and (11. k = 1. (11. ˙ (11.84) (11.88) which one veriﬁes taking the conjugate complex of (11.87) which we will employ from now on.83) into (11.86) A concise notation of the conjugate complex spinors is provided by ˙ (φα )∗ = φα = φ1 ˙ φ2 ˙ ˙ (11. 2. Hence. We denote ˙ (aα β )∗ = aα β ˙ (11.85) φ1 φ2 ∗ ∗ .83) Here we have employed the hermitian property of σ. i. 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 (11.89) such that (11.88) reads ˙ ˙ φ α = aα β φβ ˙ ˙ (11. a∗ (z) provides a representation of the Lorentz group which is distinct from that provided by a(z). hence. Insertion of (11.14) one can write a −1 Spinor Formulation = e 2 z·σ 1 −1 = e 2 z· 1 σ −1 = e− 2 z·σ 1 ∗ (11. The transformation ˙ behaviour of φα is ˙ φ α = (aα β )∗ φβ ˙ (11. ˙ We also deﬁne covariant versions of φα φα = (φα )∗ .81) yields aα β and.80) φ α χα = φβ χβ . σ † = (σ ∗ )T = σ.91) .

A tensor tα1 α2 ···αk β1 β2 ···β (11. one can state ˜ and (11.4 Spinor Tensors ˙ ˙ ˙ We generalize now our deﬁnition of spinors φα to tensors.94) (11.. a∗ (z) −1 accounting for the ˜ transformation behaviour of the third and fourth spinor component of Ψ. (11.75). − 1 .98) ˜ ∗ (z) −1 Ψ3 a ˜ Ψ 4 obviously implies that the solution of the Dirac equation in the chiral representation can be written in spinor form ˜ 1 µ Ψ1 (xµ ) φ (x ) ˜ Ψ2 (xµ ) φ2 (xµ ) φα (xµ ) = = . is governed by the operator a∗ (z) −1 which arises in ˜ the Lorentz transformation (11. a property which is rather evident. For the spinors φα and χα thus ˙ deﬁned holds that the scalar product ˙ φα χα = φ1 χ1 + φ2 χ2 ˙ ˙ ˙ ˙ ˙ (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 = (11.73.75.99) Ψ3 (xµ ) χ ˙ (xµ ) ˜ χβ (xµ ) ˙ 1 ˜ χ2 (xµ ) Ψ4 (xµ ) ˙ 11.18) of the bispinor wave function Ψ. Ψ4 .92) (11.93) (11. i.11.18) ˜ 3 . in matrix notation. A comparision of (11.95) ˙ where αβ and αβ are the real matrices deﬁned in (11. (11.97) implies then that φα transforms like Ψ ˙ φ1 ˙ φ2 ˙ transforms under rotations (z = iϑ) like a spin– 1 state | 1 .100) .e. 11.76) φα ˙ ˙ φα αβ ˙ ˙ αβ ˙ ˙ = = = = αβ ˙ ˙ αβ ˙ ˙ αβ αβ φβ φβ ˙ ˙ (11.97) The transformation. The transformation behaviour of φα is ˙ φα = ˙ ˙ β γδ ˙˙ φδ ˙ ˙ αβ a γ ˙ ˙ . 11.4: Spinor Tensors 365 The relationship between contravariant and covariant conjugate complex spinors can be expressed in analogy to (11. + 1 2 2 2 1 transforms under rotations (z = iϑ) like a spin– 2 state | 1 .96) is Lorentz invariant.

103) Similarly.m aj a∗ t km m . ˙ ˙ ˙ ˙ (11. i. i.102..e.m aj akm t m (11.107) which reads in spinor notation [cf.25] M (Aµ ) = σµ Aµ . we want to express Aµ through a spinor tensor. . i. 11..104) Indices on tensors can also be lowered employing the formula tα β = and generalizations thereof. αβ = αβ (11. An example of a tensor is αβ and tions. According to (11.105) This tensor is actually invariant under Lorentz transforma= αβ . tjk = a t a† jk ˙ ˙ ˙ ˙ (11.106).102) = . it holds αβ . the transformation bevaviour of a tensor tαβ reads in spinor and matrix notation t αβ = aαγ aβδ tγδ .109) . ˙ ˙ ˙ ˙ (11. using conventional matrix indices j.4.108) Obviously. (11.103) A αβ = aα γ aβ δ Aγ δ . The 4–vector Aµ in spinor form We want to provide now the spinor form of the 4-vector Aµ . An obvious candidate is [cf.e.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 . m.101) An example is the tensor tαβ which will play an important role in the spinor presentation of the Dirac equation.1: Prove equation (11.25) the tensor is explicitly Aαβ = ˙ A1 1 A1 2 ˙ ˙ A2 1 A2 2 ˙ ˙ = A0 + A3 A1 − iA2 A1 + iA2 A0 − A3 . (11. αβ ˙ αγ t ˙ γβ (11.e. (11. We had demonstrated that M (Aµ ) transforms according to M = M (Lµ ν Aν ) = a M (Aµ ) a† (11. k. This tensor transforms according to t αβ = aα γ aβ δ tγ δ ˙ This reads in matrix notation. with ˙ contravariant indices αβ. (11. this transformation behaviour is in harmony with the tensor notation adopted. This task implies that we seek a tensor.. the elements of which are linear functions of Aµ . tjk = a t aT jk = .106) Exercise 11.

114) We ﬁnally note that the 4–vector scalar product Aµ Bµ reads in spinor notation Aµ B µ = 1 αβ ˙ A B αβ . Aµ and tensors tαβ can be applied to the partial diﬀerential operator ∂µ .110) one obtains Aα β = ˙ A0 − A3 −A1 − iA2 −A1 + iA2 A0 + A3 A0 + A3 A1 + iA2 A1 − iA2 A0 − A3 . (11.2: Prove that (11.4: Spinor Tensors One can express Aαβ also through Aµ Aα β = ˙ ˙ 367 A0 − A3 −A1 + iA2 −A1 − iA2 A0 + A3 .4. (11. (11.109. 11.115) is correct.112) A22 −A21 ˙ ˙ −A12 A1 1 .113) Aα β = ˙ (11. Aµ can also be associated with tensors Aα β = ˙ This tensor reads in matrix notation A11 A12 ˙ ˙ A21 A22 ˙ ˙ = = 0 1 −1 0 ˙ ˙˙ αγ β δ A ˙ γδ . ˙ 2 (11.11.111) A11 A12 ˙ ˙ A21 A22 ˙ ˙ ˙ 0 −1 1 0 (11.115) Exercise 11.114) yields ∂αβ = ˙ ∂0 + ∂3 ∂1 + i∂2 ∂1 − i∂2 ∂0 − ∂3 . (11. (11.116) . ∂µ in spinor notation The relationship between 4–vectors Aµ . employing (11.110) one can state ∂ αβ = Similarly. (11. Using (11. Hence.110) The 4-vectors Aµ .117) ˙ ˙ ∂0 − ∂3 −∂1 + i∂2 −∂1 − i∂2 ∂0 + ∂3 .

i. (11. . hence. γµ = ˜ 0 µ )∗ (σ −1 0 σµ σµ 0 (σ µ )∗ −1 . σµ .15) one can write σµ = and.f. Note that (10.s. of this equation is 0 ∗ (σ µ )∗ a and. (11. (11. .124) constitutes the essential transformation property of σ µ and will be considered further. holds also in the chiral representation.122) aσ µ −1 a−1 −1 (a∗ )−1 0 Lν µ (11.e.124). hence.124) .. −1 0 0 1 (11. Hence. (10.118) 0 −1 −1 0 0 i −i 0 For this purpose we consider ﬁrst the transformation behaviour of σ µ and σµ . Equation (10.243)]..125) a (σ ) Equation (11.18) and γ µ by (11.h. We will obtain the transformation behaviour of σ µ .125) is equivalent to µ ∗ −1 −1 a∗ (σ µ )∗ −1 −1 a Lν µ = (σ ν )∗ (11. (11. σµ building on the known transformation behaviour of γ µ . This is ˜ possible since γ µ can be expressed through σ µ .243) in the chiral representation expressing S(Lη ξ ) by (11.124). 0 −i i 0 .121) One expects then that the transformation properties of σ µ should follow from the transformation properties established already for γ µ [c.123) (a∗ )−1 Lν µ a L ν µ = = σν (σ ) ν ∗ −1 (11.118) yields ˜ γµ = ˜ Using (11. (11. .243) holds independently of the representation chosen.125) is equivalent to (11.26) and its contravariant analogue σ µ σµ = σµ = 1 0 0 1 1 0 0 1 .e.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.119) (11. i.368 σµ in Tensor Notation Spinor Formulation We want to develop now the tensor notation for σµ (11. To obtain the transformation properties of σ µ we employ then (10. one can conclude aσ µ ∗ −1 Lν µ σν 0 = . 1 0 0 −1 .121).126) which is the complex conjugate of (11. Comparision of (10. (11. 0 1 1 0 .229) and (11.120) σµ 0 .

In fact. (11. e.11.g. This transformation behaviour.2) −1 (a∗ )−1 = e2z 1 ∗ ·(σ ∗ )T = e2z 1 ∗ ·σ ∗ T = [a∗ ]T . . = .124) a σ µ [a∗ ]T Lν µ = σ ν .124) using (11.4).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 . 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. according to (11.107) to (11. 3) σj −→ a(iϑ) σj a(iϑ) † = a(iϑ) σj a∗ (iϑ) T . equation (11.2) −1 369 (a∗ )−1 = −1 − 1 z ∗ ·σ ∗ 2 e = e− 2 z 1 ∗ · −1 σ ∗ = e2z 1 ∗ ·σ . state that σ µ in a new reference frame is σ µ = a σ µ a† (11.103) ˙ identiﬁes σ µ as a tensor of type tαβ .102.4: Spinor Tensors One can rewrite (11.133) Combining (11..2. (11.134) .16.127) Exploiting the hermitian property of σ. One can. 11. 2.129. −1 0 0 1 µ=3 . 0 1 1 0 µ=1 . 0 −1 −1 0 µ=1 . (11. (11. 11.128) One can express.132) 1 0 0 1 µ=0 . (11..118) (σ µ )11 (σ µ )12 ˙ ˙ (σ µ )21 (σ µ )22 and (σµ )11 (σµ )12 (σ µ )21 (σ µ )22 ˙ ˙ 1 0 0 1 µ=0 ˙ ˙ ˙ ˙ = (11. 0 −i i 0 µ=2 . 1 0 0 −1 µ=3 .2. 11. hence. It holds according to (11. therefore.131) where a is given by (11.4) with w = 0. i. (11. transformations (11. under rotations the Pauli matrices transform like (j = 1. 11. 11.e. (σ ∗ )T = σ yields using (11.108) that the same transformation behaviour applies then for general Lorentz transformations.

i. Spinor Formulation (11. state ˙ ˙ ˙ ˙ (σ11 )11 (σ11 )12 (σ12 )11 (σ12 )12 ˙ ˙ ˙ ˙ ˙ ˙ ˙ ˙ ˙ γδ (σ11 )21 (σ11 )22 (σ12 )21 (σ12 )22 ˙ ˙ ˙ ˙ . ˙˙ (11.133) is characterized through a 4-vector ˙ index µ.224) ˙ Using (11. however.12) for of σµ yields together with (5.139) g −f g and. in fact. account for the 4–vector index µ.e. 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 . We want to express now σ µ and σµ also with respect to the 4–vector index µ in spinor form employing (11. We will now consider the representation of σ µ . β.29)].s.370 Inverting contravariant and covariant indices one can also state Lν µ σµ = σν .132) and (11.136) σαβ = = 2 ˙ σ1 − iσ2 σ0 − σ3 0 0 0 0 1 0 0 1 where on the rhs. the submatrices correspond to tαβ spinors.e.114). hence. µ = 0. i. the transformation is inverse to that of ordinary 4–vectors. i..135) This is the property surmised already above [cf. 1. ˙ ˙ 0 −1 1 0 ˙ (11.141) σ αβ = σ0 − σ3 σ1 + iσ2 −σ1 − iσ2 σ0 + σ3 = 2 0 0 0 1 0 −1 0 0 . We can.136) results in the succinct expression 1 2 ˙ γδ ˙ (11.. whereas ˙ account for the elements of the individual Pauli the ‘outer’ contravariant spinor indices.e. The ˙ desired change of representation(σµ )αβ −→ (σµ )αβ corresponds to a transformation of the basis of ˙ spin states f g f −→ = (11.138) Note that all elements of σαβ are real and that there are only four non-vanishing elements.137) σαβ ˙ = δαγ δβ δ . δ. ˙ ˙ In (11. γ. (11.138) the ‘inner’ covariant spinor indices. Each of the 4 × 4 = 16 matrix elements in (11. matrices. of (11. This yields 1 0 0 1 0 0 0 0 σ0 + σ3 σ1 + iσ2 (11.h.. = σαβ ˙ ˙ ˙ ˙ ˙ (σ21 )11 (σ21 )12 (σ22 )11 (σ22 )12 ˙ ˙ ˙ ˙ ˙ ˙ ˙ ˙ 21 (σ )22 21 (σ )22 (σ21 ) (σ22 ) ˙ ˙ ˙ ˙ 21 22 Equating this with the r. 3 as well as through two spinor indices αβ. We can summarize that σ µ and σµ transform like a 4–vector. σµ in which the contravariant indices are moved ‘inside’.110) to express σ αβ in terms 0 −1 1 0 0 0 0 1 (11. account for the 4-vector µ. and the covariant indices are moved outside.140) where we employed the expressions (11. α. 2.

This expression implies that the element of γ µ given by σ µ −1 is in the expression (11.144) The Dirac Matrices γ µ in spinor notation We want to express now the Dirac matrices γ µ in spinor form.145) Let Aµ be a covariant 4–vector. employing then transformation (11. ˙ ˙ ˙ ˙ (11.147) . ˙˙ (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 ˙ ˙ .115) γ µ Aµ ˜ = 0 ((σ µ )αβ )∗ Aµ ˙ 0 ˙ ∗ 1 γδ ˙ ˙ 2 ((σ )αβ ) Aγ δ (σ µ )αβ Aµ 0 ˙ ˙ αβ γ δ 1 ˙ 2 (σγ δ ) A ˙ = 0 .4: Spinor Tensors 371 and. One can write then the scalar product using (11.143) Combined with (11. (11. ˙˙ (11.11. Accordingly. (11. For this purpose we start from the ˜ µ . we write ˜ ˙ γµ = ˜ 0 (σ µ )αβ µ ) )∗ ((σ αβ 0 ˙ ˙ .121) of γ ˜ ˜ ˙ basis |αβ whereas the element of γ µ given by σ µ is in the basis |αβ .138) one can conclude that the following property holds 1 2 ˙ γδ σ αβ ˙ = 1 2 σ αβ ˙ ˙ γδ = δαγ δβ δ .146) Exploiting the property (11.144) results in the simple relationship γ µ Aµ = ˜ 0 Aα β ˙ Aα β 0 ˙ . (11.140) to transform each of the four submatrices. which in ˙ (11.142) = This can be expressed 1 2 σ αβ ˙ ˙ γδ = δαγ δβ δ .

˙ (11.99) yields the Dirac equation in spinor form i ∂ αβ χβ ˙ i ∂ αβ φ ˙ The simplicity of this equation is striking.147) allows us to rewrite the Dirac equation in the chiral representation (10.372 Spinor Formulation 11.148) ˜ Employing Ψ(xµ ) in the form (11.226) ˜ 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 .149) (11. (11.

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

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 .λ1 . this implies a degeneracy in the spectrum. 3p and 3d all have the same energy. Classically. In chapter 7 the dynamics of a particle moving in three dimensions under the inﬂuence of a spherically symmetric potential. namely motion in a spherically symmetric potential described by the group SO(3) (or its double covering SU (2) as discussed in section 5. Sk . The energy levels are mk 2 En = − 2 2 .λ1 . on an energy eigenstate. For example. Jk . (7. Sk .. Jk ] = 0 . This degeneracy follows from the fact that any rotation. 2m r (12. 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. k = 1. as represented by an element of SO(3). . l.. (12. (7.5) with m = −l.. . As an example of this we will consider the Coulomb problem with the Hamiltonian H = p2 k − . l and the corresponding energy levels are independent of m. (7. n − 1. which yields a set of eigenstates of the form ψE. . leaves the energy of the state invariant H exp(iαk Sk ) ψE.. · · · . (12.λn .. as studied in chapter 7.m (r) Y m (θ... can be identiﬁed with the angular momentum operators. . φ).18). . the states 3s..8) . The energy levels are totally independent of l. (12..39). each energy level is (2l + 1)-fold degenerate.3) If the newly obtained state is linearly independent of the original one. V (r).8) ˆ [H.. generates a state which has the same energy as the original one. Therefore. In order to illustrate this. Presence of additional symmetries may further increase the degeneracy of the system. we ﬁrst consider a particular example of the implications of symmetry.λ1 .12). is allowed to take values in 0..λn = E exp(iαk Sk ) ψE. ψE. 3 . We will investigate this shortly in more detail in the case of systems with spherical symmetry.24).374 Spinor Formulation The action of any symmetry generator. Lie groups play an essential role in the discussion of mass degeneracies in particle physics. can then be reduced to a one-dimensional radial o equation.4) The stationary Schr¨dinger equation. has been discussed. The spherical symmetry implies the commutation of the Hamiltonian with angular momentum operators (7. . (12. .λn . m r (12.7) 2 n where the orbital angular momentum.6) From elementary quantum mechanics we know the spectrum of the hydrogen atom.m (r) = vE. where the symmetry generators. 2..

19) (12. 2m r 1 x2 (J1 p3 − J3 p1 + i p2 ) − k . 2 1 B = (J − K). In order to understand the aforementioned extra degeneracy.20) (12.10) (12. we will compute the hydrogen spectrum using the additional symmetry. (12. Jj ] = i [Ki . ijk Jk . (12. Aj ] = i ijk Ak .18) which complement the familiar angular momentum algebra of section 5. 2m r 1 x1 (J3 p2 − J2 p3 + i p1 ) − k . (12.8) is not a hermitian operator.21) .11.16) Through some algebra [4] the following commutation relations can be veriﬁed [Ki . in the subspace of the Hilbert space corresponding to a certain energy we can replace H by E.3.5: 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. Now we scale the eccentricity vector as follows K= − m . We introduce the following new operators 1 A = (J + K). The corresponding quantum mechanical hermitian operator can be deﬁned by 1 1 2 3 = = = = x1 1 (p2 J3 − p3 J2 + J3 p2 − J2 p3 ) − k . which is valid even when a and b do not commute.13) It can be veriﬁed explicitly that its components commute with the Hamiltonian.12) (12.15) which can be proved after very considerable algebra.17) (12. 2m r (12. Therefore. The vector in (12. which have a negative energy E. We will also need the following identity [4] 2 = 2H J2 + m 2 + k2 .14) This follows from a · (a × b) = (a × b) · a = 0. In the following we consider the bound states.11) (12. 2m r 1 x3 (J2 p1 − J1 p2 + i p3 ) − k .9) (12. (12. Kj ] = i ijk Kk . For this purpose we ﬁrst note that J · = 0. 2E (12. 2 which can be shown to satisfy [Ai .

A.33) (12.27) (12. . Using this equation the energy eigenvalues can be written in terms of the eigenvalues of A2 and B 2 operators. . ijk Bk . 1. (12. . . In order to arrive at the spectrum a ﬁnal bit algebra is needed A2 + B 2 = J 2 + K 2 m 2 = J2 − 2E mk 2 1 = − − 4E 2 (12. l. 2 (2a + 1)2 2 2 (12. Spinor Formulation (12. . a + b = n − 1}. H B.15). .32) A comparison with (12.28). The most extreme case of this is the quantum analogue of a classically chaotic system. .21) and (12. a = 0. the energy eigenvalues are found to be E=− mk 2 1 . . . a = 0.28) . . Bj ] = i [Ai .25) So far we have shown that the symmetry generators form an algebra. .14) we note that A2 − B 2 = J · = 0. which is identical to the the direct sum of the Lie algebra of of two SO(3) (or SU (2)) algebras. 1. 2 (12.31) where we have used (12. .376 [Bi .23) (12. namely that J J > < A − B =0 A + B =2 A (12.29) (12. = 0.22): A2 = a(a + 1) B 2 = b(b + 1) Following (12.7) tells us that (2a + 1) = n. b = 0. can be seen to follow from the triangle inequality as applied to J = A + B. H = 0.30) 2 2 2 1 . a lack of symmetry implies a lack of degeneracy in the energy levels of a quantum mechanical system. 1. Chaos is described classically as exponential . Noticing that A2 and B 2 have the same eigenvalues because of (12. 1. . In contrast to the discussion above about extra symmetries. . . 2 1 .26) (12. . . By comparing to the rotation algebra introduced in chapter 5.22) (12.34) It follows that l has to have values in {0 = |a − b|.24) (12. . This implies that a = b. . Furthermore the bound on the orbital angular momentum. we can read oﬀ the eigenvalues of A2 and B 2 from (12. Bj ] = 0. This illustrates the eﬀect of additional symmetries to the degeneracy structure of a quantum mechanical system.

12. (Strong interactions bind the atomic nucleus together. another manifestation of chaos is the lack of independent operators commuting with the Hamiltonian. we will be able to use the familiar Clebsch-Gordan coeﬃcients to combine the isospin of two particles the same way we added spin in chapter 6. 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. (12. n= 0 1 . For example. who. for the . which is a particle in box problem in two dimensions with a boundary which can be chosen arbitrarily. it is natural to represent them in terms of a two component vector. for example. This (approximate) degeneracy led into the idea of the existence of an (approximate) symmetry obeyed by the underlying nuclear interactions. the isospin symmetry is also governed by an SU (2) group ‘rotating’ components in (12. Originally it was believed that isospin was an exact symmetry of strong interactions and that it was violated by electromagnetic and weak interactions. in the sense that nearby trajectories in the phase space diverge from each other over time. after the discovery of the neutron in 1932. suggested that the proton and the neutron can be regarded as two states of a single particle. This enables us to utilize what we already know about the SU (2) symmetry group from the study of angular momentum.57M eV /c2 . mp = 939. that the proton and the neutron behave identically under the so-called strong interactions and that their diﬀerence is solely in their charge content. (12.2 Isospin and the SU (2) ﬂavor symmetry The concept of isospin goes back to Heisenberg.11.) If the proton and the neutron are to be viewed as two linearly independent states of the same particle.5: Lorentz Invariant Field Equations 377 sensitivity to initial conditions. A typical example of this so called quantum chaos is the quantum billiard problem. Following the mass-energy equivalence of special relativity E = mc2 . (Weak interactions are responsible. which was discovered by observing degeneracies in the particle spectrum.36) In analogy to the concept of spin regarding the rotations in 3-space as discussed in chapter 5.2 system p= 1 0 . This is known as level repulsion. This was motivated by the observation that their masses are approximately equal: mp = 938. the classical trajectories will diverge from each other after successive bounces from the boundary. However.36) into each other in abstract isospin space. For a more detailed discussion of quantum chaos in billiard systems we refer the reader to [7] and the references therein. In the next section we will proceed with the discussion of a symmetry. It is important to place the isospin concept in its proper historical context. namely.35) this mass equivalence can be viewed as an energy degeneracy of the underlying interactions.28M eV /c2 . analogous to the 1 spin-up and spin-down states of a spin. If the chosen boundary is ‘irregular’ in a suitably deﬁned sense.

2 2 1 2 2 2 (12. 2 2 1 2 2 2 1 1 1 1 1 √ . d= 1 1 . . If it were otherwise the proton would be unstable by decaying into the neutron. . the three pions in (12.42) (12. albeit it is a good approximate one. In the framework of the quark of model. They can be constructed with the same rules that have been used for angular momentum addition ¯ u in chapter 6. 2. 2 2 1 1 1 . −1 . −1 = = = 1 1 1 1 . We can describe isospin multiplets the same way we have described the angular momentum and spin multiplets. sec.36) as a multiplet with I = 2 1 1 p = I = .11.− 2 2 . π 0 = |1. I3 = 2 2 1 1 .− . Further than that. I3 = − 2 2 .41) (12.− 2 2 1 2 2 2 1 1 1 1 . 0 π − = |1. (mπ± = 139. the proton had to be heavier. Isospin symmetry is not an exact symmetry of strong interactions. the proton is the lighter of the two.378 Spinor Formulation beta decay). (12. as we shall see below. The mass diﬀerence between the neutron and the proton could then be attributed to the charge content of the latter. Therefore it remains a useful concept. I.39) All other isomultiplets. Shortly we 1 will see how to describe both the pion and nucleon states as composites of more fundamental I = 2 states. π − = |1. 1 we can re-write (12. For example. . However. the fundamental representation of the isospin symmetry corresponds to the doublet that contains the so-called up and down quarks u= 1 1 . respectively. (12. as good quantum numbers. If the mass diﬀerence (or the energy diﬀerence) were to be to be purely electrostatic in nature. 0 . spelling disaster for the stability of matter. and its third component. it can be seen as part of a larger (and more approximate) symmetry which is of great utility to classify observed particle families. Denoting the total isospin. 1 .0M eV /c2 ).− 2 2 . (12. .6M eV /c2 . are made up of these two quarks. we refer the reader to [1]. For a precise description of the two nucleons as composite states.43) Similarly. u¯ and d¯ states. I3 .− .38) are ud.40) + 2 1 1 . 2 2 . 1 π 0 = |1. They u form an isotriplet: π + = |1.37) As an example of a multiplet with I = 1 we have the three pions or π-mesons π + = |1.38) which have all nearly identical masses. the proton and the neutron can be written as totally symmetric uud and udd states. n = I = . mπ0 = 135. including the proton and the neutron. 2 (12. including the spin and color quantum numbers of their constituent quarks.

strongly interacting particles) which is about a GeV /c2 . −1 = |p > |p >. 1 |1. If taken literally. The names. The up and down quarks have strangeness zero. This is why isospin is such a good symmetry and why isomultiplets have nearly identical masses. 0 = √ (|p > |n > −|n > |p >) 2 and that of an isotriplet |1. It shall be noted here that the quark model was not invented until 1960’s. baryons generally are heavy. Therefore the deuteron has to be an isosinglet state (|0. the pions being examples thereof. such as proton and the neutron. where leptons (electron.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. originally refer to the relative weight of particles. baryon and meson. Therefore it has to have isospin. a hydrogen isotope. By convention ¯ baryons have baryon number 1. by accidental convention. . Following (12. All antiparticles have their 3 quantum numbers reversed. (12.45) Is the deuteron an isosinglet state or an isotriplet? If it were an isotriplet (|1. 1 = √ (|p > |n > +|n > |p >). the up and down quarks are not the only quark ‘species’ . The value of the strangeness quantum number is taken. The reader may know that the deuteron. In the late 1940’s and early 1950’s. 0 |1. to be −1 for the strange quark.e. mesons have intermediate mass ranges. but at the time the empirical concepts like isospin and strangeness quantum numbers were in use. 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. muon. whereas mesons are q q states. 0 ).or ﬂavors as they are commonly called. a number strange particles have been found which presumably contained a third quark species: the strange quark.43). All other composite states have their strangeness given by the sum of the strangeness content of their constituents. 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).11. 2 = |n > |n > . this remains only an inaccurate naming convention today.47) (12. the neutrinos etc.44) 2 where B is the baryon number and S is the strangeness. we shall setup some terminology: baryons are qqq states.48) (12. Now let us consider another example of combining the isospins of two particles. Before proceeding further.) are light.46) (12. Naturally.37) and in analogy to (12. (12. I3 = 0. this will be mathematically identical to adding two spins. As it later turned out. The possibilities are that of an isosinglet 1 |0. mesons have baryon number 0. consists of a proton and a neutron. and quarks have baryon number 1 . as some mesons discovered later are heavier than some baryons and so on. but such states do not exist. 0 ) we should have seen nn and pp bound states of comparable energy in nature (because of isospin symmetry). We will now try to describe its wave function in terms of its constituent nucleons.

such as the spatial dependence of the wave function and spin. The initial and ﬁnal states can be denoted in more detail as | initial | ﬁnal = = I (i) . γ (i) is a reduced matrix element deﬁned in the same sense as in section 6. I3 = 1. σ. 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. For example. γ (f ) T00 I (i) = 1. For completeness let us start by restating the Wigner-Eckart theorem (6. 1 and (1/ 2)(|1. I3 .49) The only assumption that we put in will be that the interaction is of the form V = α I(1) · I(2) . Exercise. We will eventually be able to compute ratios of scattering cross-sections between diﬀerent processes. which is an isoscalar.38) and the fact that the deuteron is an isosinglet we know that the isospins of the ﬁnal states in (I) and (II) are |1.7 to the case of isospin. We will now employ the (isospin analogue) of the Wigner-Eckart theorem (6.7 and 6. The dot product here refers to the abstract isospin space.7. as discussed in the exercise above. γ (i) . (12. Show that the expectation of I(1) · I(2) is state. is proportional to |M|2 . γ (f ) ||T0 ||I (i) . The cross-section.52) where γ (i) and γ (f ) denote degrees of freedom other than isospin. Exercise. γ (f ) ||T ||I (i) .55) . Show that I(1) ·I(2) is an ‘isotensor’ of rank zero.53) 0 √ 1 2I (i) +1 I (f ) . 1 and |1.8 to compute the ratio of the scattering amplitudes MI and MII . 0 .259) in the present context: I (f ) I3 . (f ) (i) (I (f ) I3 |00I (i) I3 ) is a Clebsch-Gordon coeﬃcient and I (f ) . 0 ). I3 (f ) = 1.51) (12. γ (f ) (f ) (i) (12.45) and (12.5 for the spin-analogue of the same problem. Consider the generalization of tensor operators discussed in section 6. Refer to exercise 6. γ (i) = (f ) (i) (I (f ) I3 |00I (i) I3 ) (f ) (i) (−1)I −I (f ) (i) (12.380 Spinor Formulation As an exercise on the implications of isospin symmetry we will consider nucleon-nucleon scattering. . γ (i) (i) (12. 0 + |0. γ (i) I (f ) . consider (I) p + p → d + π + (II) p + n → d + π 0 (12. 1 4 in an isotriplet state and − 3 in an isosinglet 4 Using (12. γ (f ) |T00 |I (i) I3 .50) . where M is the scattering amplitude given by M = ﬁnal | α I(1) · I(2) | initial . (12.8. I3 . √ According to (12.48) the initial states in (I) and (II) have isospin values |1.54) Here T00 ≡ V = α I(1) · I(2) . respectively.259) discussed in detail in sections 6.

5: Lorentz Invariant Field Equations 1 (11|0011) √ I (f ) = 1.58) (12. (c) π − + p → π 0 + n . particles with strangeness.59) (12. but these are not dominant at relatively low energies.63) where common dynamical factors (which would not be as easy to compute) have dropped out thanks to the Wigner-Eckart theorem. for example. We want to compute the ratio of total cross-sections assuming a similar interaction as in the previous example σ( π + + p → anything ) . involving. 1 π − + p : |1. γ (f ) ||T00 || I (i) = 1. 0 1 1 3 3 2. (12. σ( π − + p → anything ) The possibilities are (a) π + + p → π + + p .61) MII Note that the second term in MII vanishes due to the isospin conservation. . γ (f ) ||T00 || I (i) = 1. γ (f ) T00 I (i) = 0.66) = 2 3 1 3 2. It follows σI = 2. Once again we need the isospins for the initial and ﬁnal states.62) We can now write the ratio of the scattering amplitudes: MI 1 √ .57) (12. I3 = 0. we will consider pion-nucleon scattering. σII which is in approximate agreement with the observed ratio.64) There are more exotic possibilities. I3 = 0. 2 . 2 = 3 2. γ (i) 2 1 = (10|0010) √ I (f ) = 1. I3 = 0. (b) π − + p → π − + p . The relevant Clebsch-Gordon coeﬃcients are easily evaluated: (11|0011) = (10|0010) = 1. −2 1 1 √ . 1 1 1 1 3 √ 2. which are obtained by a standard Clebsch-Gordan expansion π + + p : |1.11. γ (i) = √ 2 1 (f ) (i) +√ I (f ) = 1. which is also manifested by a vanishing Clebsch-Gordon prefactor.65) (12. 2 = 2. −2 . I3 = 0. −1 2 3 2 − + . γ (i) 3 +0.60) (12. [2] As a further example. (12.56) (12. γ (i) 3 1 (f ) (i) I (f ) = 1. −2 2 1 1 3 2. −2 1 1 2. = MII (1/ 2) (12. −1 π0 + n : |1. = 381 (12. (12. γ (f ) T00 I (i) = 1.

which can be produced by π − + p → K + + Σ− .68) − 2 3 M1 2 Guided by empirical data we will further assume that M 3 >> M 1 .3 The Eightfold Way and the ﬂavor SU (3) symmetry The discovery of the concept of strangeness. however.65) Ma = M 3 2 Mb = 1 M 3 + 2 M 1 3 3 √ 2 Mc = 2 3 M3 2 √ 2 (12. m 1 2. m 1 2. while weak interactions did not. etc. S(Σ− ) = −1 and S(π) = S(N ) = 0. For instance. m .67) which are independent of m. Hence the strangeness changing strong decays of K + (or Σ− ) were forbidden. S. 0 0 1 . for example. It was assumed that strong interactions conserved S (at least approximately). which leads to the following 2 2 ratios for the cross-sections σa : σb : σc = 9 : 1 : 2. are up to 40% heavier. . 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. mentioned in the previous section. m V V 3 2. it appears an obvious generalization to add another component for an extra quark to the isospin vector space 1 0 0 (12. [2] (12.382 Spinor Formulation As in the example of nucleon-meson scattering we deﬁne the relevant matrix elements M3 = 2 M1 = 2 3 2. K + . such that S(K + ) = 1. called strangeness. (12.70) (12.69) 12. A computation similar to the previous example of nucleon-nucleon scattering yields (apart from common prefactors) the following amplitudes for the reactions in (12. In the light of the quark model. was motivated by the existence of particles that are produced strongly but decay only weakly. Strange partners of the familiar nucleons.s = 0 . The classiﬁcation of the newly found particles as members of some higher multiplet structure was less obvious then the case of isospin. giving it a higher than usual lifetime.71) u = 0 . Hence Gell-Mann and independently Nishijima postulated the existence of a separate quantum number. making an identiﬁcation of the underlying symmetry and the multiplet structure less straightforward.d = 1 . has a lifetime which is comparable to that of π + albeit being more than three times heavier.

74) for any two unitary matrices U and V .71) into each other. giving rise to the quark model. of three dimensions. the same way nucleons and pions were identiﬁed as representations of the isospin SU (2) symmetry.73) To verify that all such matrices form a group. The reader is also invited to revisit section 5. Flavor SU (Nf ) symmetry on the other hand. it is the group . bottom (beauty) and top (truth) are signiﬁcantly heavier than the hadronic energy scale. called quantum chromodynamics. The unitarity relation imposes 9 constraints on the total of 18 real degrees of freedom of a 3 × 3 complex matrix. becomes increasingly inaccurate for Nf > 3. This group of 3 × 3 unitary matrices is denoted by U (3). It is seen that such a matrix U has to satisfy the following k U † = U −1 . have to be elements of the group SU (3) (which we will investigate closely very soon). Given a complex vector. eiφ . while preserving the norm. Then came the question of what the fundamental representation. However SU (3) symmetry appears in another and much more fundamental context in strong interaction physics. which set the hadronic energy scale. but there are a number of subtle diﬀerences. namely charm. However. where Nf is the number quark ﬂavors. Lie algebras and related concepts. This is believed to be an exact symmetry of strong interactions. (The ‘bare’ mass of the charm quark is already heavier than the two nucleons. (The reader is referred to section 8. Because of this additional constraint SU (3) has 8 dimensions. Since arbitrary phase factors are of no physical interest. we want to ﬁnd those transformations ak → Ukl al that preserve the norm. In many respects it will resemble the more familiar group SU (2) discussed in some detail in chapter 5. but we will stick to N = 3 in the following. (12.71). Hence the group U (3) has 9 dimensions. In this perspective the ﬂavor SU (3) symmetry may appear to be mainly of historical interest. of a. ak . called color.72) a∗ ak . SU (N ). The reader shall note that most of what is being said trivially generalizes to other unitary groups. in reference to Lie groups.1 whenever necessary. as in (12.5: Lorentz Invariant Field Equations 383 In this case the transformations that ‘rotate’ the components of (12. in fact modern theory of strong interactions is a ‘gauge theory’ of this color group.3 for a brief discussion of gauge transformations). The bottom and top are even heavier. should correspond to. As quarks were never directly observed. we observe that (U V )† = V † U † = V −1 U −1 = (U V )−1 .11. Multiplying U by a phase. (12. First particle multiplets were identiﬁed as representations of the SU (3) group. history followed the reverse of this path. The reason is that the other known quarks. still leaves the norm invariant. 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. for a period they were considered as useful bookkeeping devices without physical content. which again form representations of an SU (3) group. we will establish some mathematical preliminaries about the group SU (3). unitary relation k (12. The quarks posses another quantum number. [2] ) Before discussing the signiﬁcance and the physical implications of the quark model.

) The unit determinant condition requires that all λk are traceless.79) (12. which can be veriﬁed explicitly as matrix identities. the meaning being apparent from the context. . 0 0 0 . The Lie algebra structure is given by the commutators of λk [λj .80) This is the fundamental or deﬁning representation of SU (3).75) αk λk (12. (12. The reader is invited to compare the structure of SU (3) to that of SU (2) discussed in section 5. λk . (12.32). λ3 = 0 −1 0 . As in the case of SU (2) higher dimensional representations obeying the same structure can be found.384 Spinor Formulation SU (3) and not U (3) that is of main interest. Therefore we will express elements of SU (3) as U = ei k (12. δjkl . λ7 = 0 0 . We can also introduce the constants. . can be written in the form U = eiH where H is a hermitian matrix.77) λ1 . regardless of the dimension of the representation. λ2 = . 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 .79). U . λk ]+ = δjk + 2δjkl λl . 3 (12.78) where fjkl are the antisymmetric structure constants similar to the jkl of SU (2) given in (5.7. λ5 = . λk ] = 2ifjkl λl . since det(eA ) = etrA . The fundamental relation to be preserved is (12. −2 λ4 λ6 λ8 The generators.1. (We shall at times refer to the Lie algebra with the name of the group. As discussed in section 5. any unitary matrix.76) where λk are 8 linearly independent matrices forming the basis of the Lie algebra of SU (3). via the anticommutator relation 4 [λj . obey the following relation tr(λj λk ) = 2δjk .

we have [Y. (12.90) (12.87) the commutation relations between the generators can be expressed in a succinct manner.44). T0 } and [Y.94) . T± ] = ±T± .86) (12. J− } proved useful. V± = F4 ± iF5 .5: Lorentz Invariant Field Equations 385 As it turns out the set of generators in (12. U± ] = ±U± .77) is not the most useful basis in the study of SU (3). 2 Y = √ F8 . 1 [T0 . 3 (12. T 0 = F3 .11. V± ] = ± V± . The generators of SU (3) can be arranged into a very similar form to that of SU (2). (Recall that the strange quark has S = −1 by convention). [Y. (12. In the basis (12. which satisﬁed an ‘eigenvalue equation’ [J0 . J± ] = ± J± . 2 1 [T0 . these relations have been used to construct the angular momentum spectrum as well as the function space representation of the rotation algebra. namely spherical harmonics. I3 . In the case of SU (2) the identiﬁcation of ‘ladder’ operators {J+ . (12.82) With another change of basis we arrive at the ‘standard’ form of the generators of the Lie algebra of SU (3) T± = F1 ± iF2 . Derive the Gell-Mann–Nishijima relation (12. We ﬁrst introduce the F-spin operators 1 Fi = λ i . [Y. Using the convention in (12.81) In chapter 5. U± ] = U± .71) show that T0 is the isospin operator. V± ] = ±V± . T± ] = 0.87) Exercise. Y is called the hypercharge.83) (12. starting with the observation that Y = B + S.89) (12.91) [T0 .93) (12.92) (12.85) (12. Exercise.84) (12. T0 ] = 0. 2 (12. U± = F6 ± iF7 . First. 2 (12.88) which deﬁnes (not uniquely) a maximal set of mutually commuting operators {Y.

similar to the way it was done in section 5. U− ] = 0.105) C2 = jkl djkl λj λk λl In general the number of independent Casimir operators of a Lie group is equal to its rank. [U+ .102) (12.101) (12. (12. Thus. A formal deﬁnition and a detailed discussion of the Cartan-Weyl form is beyond the scope of this chapter. jkl (12.88). We can construct two such independent Casimir operators for the group SU (3). The interested reader is instead referred to a very readable account given in chapter 12 of [4]. U− ] = Y − T0 = 2U0 . V+ ] = 0. [U+ . J 2 ) which commutes with all of the generators. Another important property of SU (2) is the existence of an operator (namely the total angular momentum. [T+ . The maximum number of mutually commuting generators of a Lie algebra is called its rank. T0 }. V± . since there are two mutually commuting generators in SU (3) as given in (12. Casimir operators can be used to label irreducible representations of the Lie algebra. we will use these commutation relations to construct representations of SU (3). However. In the case of SU (2) the representations lay on a line on the J0 axis. [V+ . 3 [U+ . T− ] = 2T0 . U± } to this maximal set by ‘eigenvalue equations’ and [T+ . SU (2) has rank 1. Note that.99) (12. [T+ .5. Finally. V+ ] = 0. V− ] = 2 (12. 2 3 Y + T0 = 2V0 . the representations will now lie in a T0 − Y -plane.97) which relate commutators of generators with opposite eigenvalues to the maximal set {Y. it is said to be in Cartan-Weyl form.386 Spinor Formulation which relate the remaining generators {T± . . [T+ . The same way the angular momentum ladder operators have been used to construct the representations of SU (2). V− ] = T− . U+ ] = V+ . U0 and V0 are linear combinations of T0 and Y . As in the case of J 2 and SU (2). V− ] = −U− .98) (12. This form is essential for easy labeling of the representations of the group. C1 = k λ2 = − k 2i 3 fjkl λj λk λl .100) (12. as the relation between the states in a given representation can be conveniently expressed in terms of ladder operators.96) (12.104) (12.95) (12. An operator which commutes with all generators of a Lie group is called a Casimir operator. while SU (3) has rank 2. When the basis of a Lie algebra is expressed in such a way to satisfy the form of the eigenvalue relations as given above. we have [T+ .103) Any remaining commutators follow from hermiticity.

108) (12. All other states of the representation are then constructed by successive application of ladder operators T± . For example. 1) correspond to the triplets of quarks and antiquarks. The representations for hexagons with sides of length p and q in the T0 − Y -plane. y . we have T± |t3 . The interested reader is referred to [4]. especially chapters 7 and 8.112) The eﬀect of these operators to the states in the y − t3 plane have been outlined in Fig. The details of this procedure is beyond the scope of this chapter. 2 1 = γ t3 .2) shows the representation D(2.88) we can label states by the eigenvalues of T0 and Y operators.y ± 1 . y : T0 |t3 .11. Because of (12.107) From the commutation relations we have presented above (the Cartan-Weyl form) we can write down the eﬀect of various generators on the state |t3 . Figure (12. |t3 . y Y |t3 . (12. 2 (12. 2 (12.5 how to label the irreducible representations of the angular momentum algebra SU (2) in terms of the value of the total angular momentum. Such a representation is labeled as 1 D(p. This property can be used to label representations in terms of the values of the Casimir operators. y . y V± |t3 . y .111) (12. U± . y . For example. the pion family forms part of an octet corresponding to the D(1. V± . As another example for the representations of SU (3).109) The same way that J± |m is proportional to |m ± 1 in the case of the angular momentum algebra. y 3 = ( y− 4 3 = ( y+ 4 1 t3 ) |t3 . y = t3 |t3 . y ± 1 . y . we have U0 |t3 . 2 1 = β t3 ± . For example. 1) representation. (12.106) (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.1).3). q) and it has a dimensionality of 2 (p + 1)(q + 1)(p + q + 2). it was shown in section 5. (12. . y 1 = α t3 ± . y . respectively. y V0 |t3 .antiquark pair. For example the pion octet (or any other meson octet) is therefore realized as states of a quark . 2 1 t3 ) |t3 .) All other representations can be constructed by combining these two conjugate representations. 0) and D(0. = y |t3 . A notation suggestive of the dimensionality of the representation can be used to identify representations. The representations D(1.110) (12.5: 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 U± |t3 . 1) as an example. Now we will construct explicit representations of SU (3). 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. (See Fig.

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

This way the quark-antiquark states can be represented as follows [3] ⊗ [¯ = [8] ⊕ [1] 3] (12. This expansion can be compared.118) ¡ ¢ ¨ ©§ ¨¥¦¤ " ! .114) Figure 12. 3]. to the case of adding two spin triplet states.117) The Gell-Mann–Nishijima relation can then be succinctly expressed as 1 Q = y + t3 . . where we would write [3] ⊗ [3] = [1] ⊕ [3] ⊕ [5]. £ (12. 2 (12. 3 (12.113) The additional singlet state corresponds to the η meson.5: Lorentz Invariant Field Equations 389 D(1. in the case of SU (2). 0) or the fundamental representation of ﬂavor SU (3) symmetry. The octet D(1.11.3: D(1. 2 3 2 0. − . 0) and D(0.116) (12. The three quarks in the fundamental representation can now be written as u = d = s = 1 1 . . 1) is written as [8] etc. 1) are denoted by [3] and [¯ respectively. for example. 2 3 1 1 − .115) (12.

1987. [2] Introduction to elementary particles. [3] Microscopic theory of the nucleus. L. North holland. W. 3 £ Spinor Formulation (12. D. 1984. Eisenberg. Springer-Verlag u [5] Principles of symmetry. Greiner.-F. W. G.4: The baryon octet as a D(1. and spectroscopy. Griﬃths. D. L. 3 1 Qd = − . 133. 1972. I. 1989. D. [6] Gauge theory of elementary particle physics.120) (12. Kosztin. M¨ller. J. 1993. ¡ ¢ S - S 0 ¤ ¥ S + . Wiley. Wiley.-P. Li. Schulten. A.121) L 0 X - X 0 Figure 12. 67 (2). B. W. M. [7] Expansion method for stationary states of quantum billiards. Kaufman. K. J. References [1] Quarks and leptons: an introductory course in particle physics. Oxford.119) (12. p. F. Greiner. (1999). 1984.390 from which the quark charges follow 2 Qu = . T. Am. [4] Quantum mechanics: symmetries. Halzen. Harter. dynamics. Cheng. Wiley. 3 1 Qs = − . 1) representation of SU (3). Phys. Martin.

Sign up to vote on this title

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

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