In The Quantum Theory of Fields Nobel Laureate Steven Weinberg combines his exceptional physical insight with his gift for clear exposition to provide a selfcontained, comprehensive, and uptodate introduction to quantum field theory. Volume I introduces the foundations of quantum field theory. The development is fresh and logical throughout, with each step carefully motivated by what has gone before, and emphasizing the reasons why such a theory should describe nature. After a brief historical outline, the book begins anew with the principles about which we are most certain, relativity and quantum mechanics, and the properties of particles that follow from these principles. Quantum field theory then emerges from this as a natural consequence. The classic calculations of quantum electrodynamics are presented in a thoroughly modern way, showing the use of path integrals and dimensional regularization. The account of renormalization theory reflects the changes in our view of quantum field theory since the advent of effective field theories. The book's scope extends beyond quantum electrodynamics to ele· mentary particle physics and nuclear physics. It contains much original material, and is peppered with examples and insights drawn from the author's experience as a leader of elementary particle research. Problems are included at the end of each chapter. A second volume will describe the modern applications of quantum field theory in today's standard model of elementary particles, and in some areas of condensed matter physics. This will be an invaluable reference work for all physicists and mathematicians who use quantum field theory, as well as a textbook appropriate to graduate courses.
The Quantum Theory of Fields
Volume I Foundations
" i 'f f )
The Qua
y of Fields
f"~
Volume I Foundations Steven Weinberg
University of Texas at Austin
1~ ABR. 1996
. CAMBRIDGE
UNIVERSITY PRESS
•
Published by the Press Syndicate of the Univenity of Cambridge The Pitt Building, Trurnpington Street, Cambridge CBl IRP 40 West 20th Street, New York, NY 100114211, USA 10 Stamford Road, Oaldeigh, Melbourne 3166, Australia
@ Steven Weinberg 1995
First published 1995 Printed in the United States of America
A catalogue record for this book is available Irom the British Library
Librar)' of Congress cataloguing In publication dara
Weinberg, Steven, 1933The quantum theory of fields I Steven Weinberg. p. em. Includes bibliographical reference and index. Contents: v. L Foundations. ISBN 0521550017 1. Quantum field theory. L Title.
QC174.4SW45 1995 530.1'43dc20 952782
ISBN 0 521 55001 7
CIP
TAG
l
To Louise
:
g
~
r
j
!
)
i
Contents
Sections marked with an asterisk are somewhat out of the book's main line of development and may be omitted in a first reading.
PREFACE NOTATION
xx
xxv
1
1.1
HISTORICAL INTRODUCTION
1
3
Relativistic Wave Mechanics
De Broglie waves D SchrodingerKleinGordon wave equation 0 Fine structure D Spin D Dirac equation D Negative energies D Exclusion principle D Positrons o Dirac equation reconsidered
1.2 The Birth of Quantum Field Theory 15 Born, Heisenberg, Jordan quantized field D Spontaneous emission D Anticommutators D HeisenbergPauli quantum field theory 0 FurryOppenheimer quantization of Dirac field 0 PauliWeisskopf quantization of scalar field DEarly calculations in quantum electrodynamics D Neutrons D Mesons
31 The Problem of Infinities 1.3 Infinite electron energy shifts D Vacuum polarization D Scattering oflight by light D Infrared divergences D Search for alternatives D Renormalization D Shelter Island Conference D Lamb shift D Anomalous electron magnetic moment D Schwinger, Tomonaga, Feynman, Dyson formalisms D Why not earlier?
Bibliography References
39
40
2
2.1
RELATIVISTIC QUANTUM MECHANICS Quantum Mechanics
49 49
Rays D Scalar products D Observables D Probabilities
IX
x
Contents
2.2 Symmetriell 50 Wigner's theorem 0 Antilinear and antiunitary operators 0 ObselVables 0 Group structure 0 Representations up to a phase 0 Superselection rules 0 Lie groups o Structure constants 0 Abelian symmetries
2.3 Quantum Lonmtz Transformations
55
Lorentz transformations 0 Quantum operators 0 Inversions 2.4 The Poincare Algebra 58 JI" and pI' 0 Transformation properties 0 Commutation relations 0 ConselVed and nonconserved generators 0 Finite translations and rotations 0 InoniiWigner contraction 0 Galilean algebra 25 OneParticle States
62
Transformation rules 0 Boosts 0 Little groups 0 Normalization 0 Massive particles 0 Massless particles 0 Helicity and polarization 2.6 Space Inversion and TimeReversal 74 Transfonnation of JI" and pI' 0 P is unitary and T is antiunitary 0 Massive particles 0 Massless particles 0 Kramers degeneracy 0 Electric dipole moments 2.7 Projective Representations' 81
Twococyles 0 Central charges 0 Simply connected groups 0 No central charges in the Lorentz group 0 Double connectivity of the Lorentz group 0 Covering groups 0 Superselection rules reconsidered Appendix A Appendix B Appendix C Problems References 3 3.1 SCATTERING THEORY 'In' and 'Out' States The Symmetry Representation Theorem Group Operators and Homotopy Classes Inversions and Degenerate Multiplets
91
96
100
104
105 107
107
Multiparticle states 0 Wave packets 0 Asymptotic conditions at early and late times 0 Lippmann Schwinger equations 0 Principal value and delta functions
3.2
The Smatrix 113 Definition of the Smatrix. 0 The T·matrix 0 Born approximation 0 Unitarity of the Smatrix
Symmetries of the SMatrix 3.3 116 Lorentz invariance 0 Sufficient conditions 0 Internal symmetries 0 Electric charge, strangeness, isospin, S U(3) 0 Parity conselVation 0 Intrinsic parities 0
Contents
Xl
Pion parity 0 Parity nonconservation 0 Timereversal invariance 0 Watson's theorem 0 PT nonconservation 0 C, CP, CPT 0 Neutral K mesons 0 CP nonconservation
3.4
Rates and CrossSections
134
Rates in a box 0 Decay rates 0 Crosssections 0 Lorentz invariance 0 Phase space 0 Dalitz plots
3.5
Perturbation Theory
141
Oldfashioned perturbation theory 0 Timedependent perturbation theory 0 Timeordered products 0 The Dyson series 0 Lorentzinvariant theories 0 Distorted wave Born approximation
3.6
Implications of Unitarily
147
Optical theorem 0 Diffraction peaks 0 CPT relations 0 Particle and antiparticle decay rates 0 Kinetic theory 0 Boltzmann Htheorem
3.7
PartialWave Expansions·
151
Discrete basis 0 Expansion in spherical harmonics 0 Total elastic and inelastic crosssections 0 Phase shifts 0 Threshold behavior: exothermic. endothermic, and elastic reactions 0 Scattering length 0 Highenergy elastic and inelastic scattering
3.8
Resonances'
159
Reasons for resonances: weak coupling, barriers, complexity 0 Energydependence 0 Unitarity 0 BreitWigner formula 0 Unresolved resonances 0 Phase shifts at resonance 0 RamsauerTownsend effect Problems References
165 166 169
4
4.1
THE CLUSTER DECOMPOSITION PRINCIPLE Bosons and Fermions
170 Permutation phases 0 Bose and Fermi statistics 0 Normalization for identical particles
4.2
Creation and Annihilation Operators
173
Creation operators 0 Calculating the adjoint 0 Derivation of commutation! anticommutation relations 0 Representation of general operators 0 Freeparticle Hamiltonian 0 Lorentz transformation of creation and annihilation operators 0 C, P, T properties of creation and annihilation operators 4.3 Cluster Decomposilion and Connected Amplitudes 177
Decorrelation of distant experiments 0 Connected amplitudes 0 Counting delta functions
XII
Contents
Structure of the Interaction
4.4
182
Condition for cluster decomposition 0 Graphical analysis 0 Twobody scattering implies threebody scattering
Problems References
189 [89
5
5.1
QUANTUM FIELDS AND ANTIPARTICLES
191 191
Free Fields
Creation and annihilation fields 0 Lorentz transformation of the coefficient functions 0 Construction of the coefficient functions 0 Implementing cluster decomposition 0 Lorentz invariance requires causality 0 Causality requires antiparticles o Field equations 0 Normal ordering
5.2
Causal Scalar Fields
201
Creation and annihilation fields 0 Satisfying causality 0 Scalar fields describe bosuns 0 Antiparticles 0 P, C, T transformations 0 11:0
5.3
207 Creation and annihilation fields 0 Spin zero or spin one 0 Vector fields describe bosons 0 Polarization vectors 0 Satisfying causality 0 Antiparticles 0 Mass zero limit 0 P, C, T transformations
5.4 The Dirac Formalism
Causal Vector Fields
213
Clifford representations of the Poincare algebra 0 Transformation of Dirac matrices 0 Dimensionality of Dirac matrices 0 Explicit matrices 0 ('s 0 Pseudounitarity o Complex conjugate and transpose
5.5
219 Creation and annihilation fields 0 Dirac spinors 0 Satisfying causality 0 Dirac fields describe fermions 0 Antiparticles 0 Space inversion 0 Intrinsic parity of particleantiparticle pairs 0 Chargeconjugation 0 Intrinsic Cphase of particleantiparticle pairs 0 Majorana fermions 0 Timereversal 0 Bilinear covariants 0 Beta decay interactions 5.6
General Irreducible Representations of the Homogeneous Lorentz Group'
Causal Dirac Fields
229
Isomorphism with SU(2) 0 SU(2) 0 (A,B) representation of familiar fields 0 RaritaSchwinger field 0 Space inversion
5.7
General Causal Fields'
233
Constructing the coefficient functions 0 Scalar Hamiltonian densities 0 Satisfying causality 0 Antiparticles 0 General spinstatistics connection 0 Equivalence of different field types 0 Space inversion 0 Intrinsic parity of general particleantiparticle pairs 0 Chargeconjugation 0 Intrinsic Cphase of antiparticles 0
Contents
Xlll
Selfchargeconjugate particles and reality relations D Timereversal D Problems for higher spin?
5.8
The CPT Theorem
244
CPT transformation of scalar, vcctor, and Dirac fields D CPT transformation of scalar interaction density D CPT transformation of general irreducible fields D CPT invariance of Hamiltonian
5.9
Massless Particle Fields
246
Constructing the coefficient functions D No vector fields for helicity +1 D Need for gauge invariance D Antisymmetric tensor fields for helicity +1 D Sums over helicity D Constructing causal fields for helicity +1 D Gravitons D Spin > 3 D General irreducible massless fields D Unique helicity for (A, B) fields Problems References
255 256 259
6
THE FEYNMAN RULES
6.1 Derivation of the Rules 259 Pairings D Wick's theorem D Coordinate space rules D Combinatoric factors D Sign factors D Examples 6.2
Calculation of the Propagator
274
Numerator polynomial D Feynman propagator for scalar fields D Dirac fields D General irreducible fields D Covariant propagators D Noncovariant terms in timeordered products
6.3
Momentum Space Rules
280
Conversion to momentum space D Feynman rules D Counting independent momenta D Examples D Loop suppression factors
6.4
Off Ihe Mass Shell
286
Currents D Olfshell amplitudes are eX<lct matrix elements of Heisenbergpicture operators D Proof of the theorem Problems References
7
290 291 292
THE CANONICAL FORMALISM
Canonical Variables 7.1 293 Canonical commutation relations D Examples: real scalars, complex scalars, vector fields, Dirac fields D Freeparticle Hamiltonians D Freefield Lagrangian D Canonical formalism for interacting fields
· XIV
Contents
The Lagrangian Formalism 298
7.2
Lagrangian equations of motion 0 Action 0 Lagrangian density 0 EulerLagrange equations 0 Reality of the action 0 From Lagrangians to Hamiltonians o Scalar fields revisited 0 From Heisenberg to interaction picture 0 Auxiliary fields 0 Integrating by parts in the action 7.3 Global Symmetries 306 Noether's theorem 0 Explicit formula for conserved quantities 0 Explicit formula for conserved currents 0 Quantum symmetry generators 0 Energymomentum tensor 0 Momentum 0 Internal symmetries 0 Current commutation relations
Lorentz Invariance Currents .$lp!'v 0 Generators Smatrix
7.4
314
j>'V
0 Belinfante tensor 0 Lorentz invariance of
7.5 Transition to Interaction Picture: Examples Scalar field with derivative coupling D Vector field D Dirac field
318
7.6 Constraints and Dirac Brackets 325 Primary and secondary constraints D Poisson brackets D First and second class constraints D Dirac brackets D Example: real vector field 7.7 Field Redefinitions and Redundant Couplings· 331
Redundant parameters D Field redefinitions D Example: real scalar field Appendix Problems References Dirac Brackets from Canonical Commutators
332 337 338 339
8
ELECTRODYNAMICS
8.1 Gauge Invariance 339 Need for coupling to conserved current D Charge operator D Local symmetry D Photon action D Field equations D Gauge·invariant derivatives 8.2 Constraints and Gauge Conditions 343
Primary and secondary constraints D Constraints are first class D Gauge fixing D Coulomb gauge D Solution for AO 8.3 Quantization in Coulomb Gauge 346
In
Remaining constraints are second class D Calculation of Dirac brackets Coulomb gauge D Construction of Hamiltonian D Coulomb interaction
8.4 Electrodynamics in the Interaction Picture 350 Free·field and interaction Hamiltonians D Interaction picture operators D Normal mode decomposition
Contents xv 8. linear. derivative coupling 9.5 The PhotQn Propagator 353 Numerator polynomial 0 Separation of noncovariant terms 0 Cancellation of noncovariant terms 8.P .1 Transition amplitudes for infinitesimal intervals 0 Transition amplitudes for finite intervals 0 Interpolating functions 0 Matrix elements of timeordered products o Equations of motion 9.2)form gauge fields 0 Nothing new in four spacetime dimensions Appendix Problems References 9 Traces 372 374 375 376 378 PATHINTEGRAL METHODS The General PathIntegral Formula 9.3 Lagrangian Version of the PatbIntegral Formula Integrating out the 'momenta' 0 Derivatively coupled scalars 0 Nonlinear sigma model 0 Vector field 9. and elliptic polarization 0 Polarization and spin sums 8. vector fields.8 Generalization : pform Gauge Fields' 369 Motivation 0 pforms 0 Exterior derivatives 0 Closed and exact pforms 0 pform gauge fields 0 Dual fields and currents in D spacetime dimensions 0 pform gauge fields equivalent to (D .4 PathMIntegral Derivation of Feynman Rules 395 Separation of freefield action 0 Gaussian integration 0 Propagators: scalar fields.5 Path Integrals for FermiolL'i 399 Anticommuting cnumbers 0 Eigenvectors of canonical operators 0 Summing states by Berezin integration 0 Changes of variables 0 Transition amplitudes for infinitesimal intervals 0 Transition amplitudes for finite intervals 0 Derivation of Feynman rules 0 Fermion propagator 0 Vacuum amplitudes as determinants .6 Feynman Rules for Spinor Electrodynamics 355 Feynman graphs 0 Vertices 0 External lines 0 Internal lines 0 Expansion in cz/4n 0 Circular.7 Compton Scattering 362 Smatrix 0 Differential crosssection 0 Kinematics 0 Spin sums 0 Traces 0 KleinNishina formula 0 Polarization by Thomson scattering 0 Total crosssection 8.2 Transition to tbe SMMafrix 385 389 Wave function of vacuum 0 if' terms 9.
3 Field and Mass Renormalization 436 LSZ reduction formula 0 Rcnormalized fields 0 Propagator poles 0 No radiative corrections in external lines 0 Counterterms in selfenergy parts 10.6 gauges Path integral in Coulomb gauge 0 Reintroduction of aO 0 Transition to covariant 9.7 The KallenLehmann Representation' 457 Spectral functions 0 Causality relations 0 Spectral representation 0 Asymptotic behavior of propagators 0 Poles 0 Bound on field renormalization constant 0 2 = 0 for composite particles 10.1 NONPERTURBATlVE METHODS Symmetries 425 425 Translations 0 Charge conservation 0 Furry's theorem 10.5 Gauge Invariance 448 Transversality of multiphoton amplitudes 0 Schwinger terms 0 Gauge terms in photon propagator 0 Structure of photon propagator 0 Zero photon renormalized mass 0 Calculation of 2 3 0 Radiative corrections to choice of gauge 10.2 Polology 428 Pole formula for general amplitudes 0 Derivation of the pole formula 0 Pion exchange 10.7 Varieties of Statistics' 418 Preparing 'in' and 'out' states 0 Composition rules 0 Only hosans and fermions in > 3 dimensions 0 Anyans in two dimensions Appendix Problems References Gaussian Multiple Integrals 420 423 423 10 10.6 Electromagnetic Form Factors and Magnetic Moment 452 Matrix elements of )0 0 Form factors of )1': spin zero 0 Form factors of ]I'.8 Dispersion Relations' 462 History 0 Analytic properties of massless boson forward scattering amplitude .4 Renormalb:ed Charge and Ward Identities 442 Charge operator 0 Electromagnetic field renormalization 0 Charge renormalization 0 WardTakahashi identity 0 Ward identity 10. spin ~ 0 Magnetic moment of a spin ~ particle 0 Measuring the form factors 10.XVI Contents PathIntegral Formulation of Quantum Electrodynamics 413 9.
and mass renormalization 0 Lagrangian counterterms 11.4 Electron SelfEnergy 493 Oneloop formula for electron selfenergy part 0 Electron mass renormalization o Cancellation of ultraviolet divergences Appendix Problems References 12 Assorted Integrals 497 498 498 499 500 GENERAL RENORMALlZATION THEORY Degrees of Divergence 12.2 Cancellation of Divergences 505 Subtraction by differentiation 0 Renormalization program 0 Renormalizablc theories 0 Example: quantum electrodynamics 0 Overlapping divergences 0 BPHZ renormalization prescription 0 Changing the renormalization point: ¢4 theory 12.3 Is Renormalizability Necessary? 516 . charge.i.Contents XVll o Subtractions 0 Dispersion relation 0 Crossing symmetry 0 Pomeranchuk's theorem 0 Regge asymptotic behavior 0 Photon scattering Problems References 469 470 11 11. 0 Anomalous muon magnetic moment to ordercx 2 1n(ml'/ me) 0 Charge radius of leptons 11.2 Vacuum Polarization 473 Oneloop integral for photon selfenergy part 0 Feynman parameters 0 Wick rotation 0 Dimensional regularization 0 Gauge invariance 0 Calculation of 2 3 0 Cancellation of divergences 0 Vacuum polarization in charged particle scattering o Uehling effect 0 Muonic atoms 11.1 ONELOOP RADIATIVE CORRECTIONS IN QUANTUM ELECTRODYNAMICS Counterterms 472 472 Field.1 Superficial degree of divergence 0 Dimensional analysis 0 Renormalizability 0 Criterion for actual convergence 12.3 Anomalous Magnetic Moments and Charge Radii 485 Oneloop formula for vertex function 0 Calculation of form factors 0 Anomalous lepton magnetic moments to order r.
2 Virtual Soft Photons 539 544 Effect of soft virtual photons 0 Radiative corrections on internal lines 13. Cancellation of Divergences Sum over helicities 0 Integration over energies 0 Sum over photon number 0 Cancellation of infrared cutoff factors 0 Likewise for gravitation 13.4 The Floating Cutoff" 525 Wilson's approach 0 Renormalization group equation 0 Polchinski's theorem 0 Attraction to a stable surface 0 Floating cutoff vs renormalization 12. C.5 Accidental Symmetries' 529 General renormalizable theory of charged leptons 0 Redefinition of the lepton fields 0 Accidental conservation of lepton flavors.1 INFRARED EFFECTS Soft Photon Amplitudes 534 534 Single photon emission 0 Negligible emISSion from internal lines 0 Lorentz invariance implies charge conservation 0 Single graviton emission 0 Lorentz invariance implies equivalence principle 0 Multiphoton emission 0 Factorization 13.4 General Infrared Divergences 548 Massless charged particles 0 Infrared divergences in general 0 Jets 0 LeeNauenberg theorem 13. P.6 The External Field Approximation· 556 Sums over photon vertex permutations 0 Nonrelativistic limit 0 Crossed ladder exchange Problems References 562 562 .5 Soft Photon Scattering· 553 Poles in the amplitude 0 Conservation relations 0 Universality of the lowenergy limit 13. and T Problems References 531 532 13 13.XVlll Contents Renormalizable interactions cataloged 0 No renormalizable theories of gravitation 0 Cancellation of divergences in nonrenormalizable theories 0 Suppression of nonrenormalizable interactions 0 Limits on new mass scales 0 Problems with higher derivatives? 0 Detection ofnonrenormalizable interactions 0 Lowenergy expansions in nonrenormalizable theories 0 Example: scalar with only derivative coupling 0 Saturation or new physics? 0 Effective field theories 12.3 Real Sort Photons.
3 The Lamb Shift in Ligbt Atoms 578 Separating high and low energies 0 Highenergy term 0 Low·energy term 0 Effect of mass renormalization 0 Total energy shift 0 t = 0 0 t+O Numerical results 0 Theory vs experiment for classic Lamb shift 0 Theory vs experiment for Is energy shift Problems References AUTHOR INDEX SUBJECT INDEX 594 596 597 602 OUTLINE OF VOLUME II 15 16 17 18 19 20 21 22 23 NONABELIAN GAUGE THEORIES BACKGROUND FIELD METHODS RENORMALIZATION GROUP METHODS OPERATOR PRODUCT EXPANSIONS SPONTANEOUSLY BROKEN GLOBAL SYMMETRIES SPONTANEOUSLY BROKEN LOCAL SYMMETRIES ANOMALIES TOPOLOGICALLY COMPLICATED FIELDS SUPERCONDUCfIVITY .and angledependence 0 Radial wave equations 0 Energies 0 Fine structure 0 Nonrelativistic approximations 14.2 Radiative Corrections in External Fields 572 Electron propagator in an external field 0 Inhomogeneous Dirac equation 0 Effects of radiative corrections 0 Energy shifts 14.Contents 14 XlX BOUND STATES IN EXTERNAL FIELDS 564 14.1 The Dirac Equation 565 Dirac wave functions as field matrix elements 0 Anticommutators and completeness 0 Energy eigenstates 0 Negative energy wave functions 0 Orthonormalization 0 'Large' and 'small' components 0 Parity 0 Spin.
examples that are chosen from contemporary particle physics or nuclear physics as well as from quantum electrodynamics. I have tried throughout to discuss matters in a context that is as general as possible. I suppose the most distinctive thing about it is its generality. the perspective of this book. Another book will be worth while only if it offers something new in content or perspective.that is. Of course. Some of this traditional approach will be found here in the historical introduction presented in Chapter L This is certainly a way of getting rapidly into the subject. specific examples are frequently used to illustrate general points. and 'quantize' them . apply to various simple field theories the rules of canonical quantization or path integration. Why should we believe in the rules of canonical quantization or path integration? Why should we adopt the simple field equations and Lagrangians that are found in the literature? For that matter. quantum electrodynamics. however. although this book contains a good amount of new material. relying for justification on our experience with electromagnetism. It is.Preface To Volume I Why another book on quantum field theory? Today the student of quantum field theory can choose from among a score of excellent books. rather than its content. but it seems to me that it leaves the reflective reader with too many unanswered questions. 1 aim to present quantum field theory in a manner that will give the reader the clearest possible idea of why this theory takes the form it does. after all. our purpose in theoretical physics is xx . but even more because r think that this generality will help to keep the important points from being submerged in the technicalities of specific theories. and why in this form it does such a good job of describing the real world. This is in part because quantum field theory has found applications far removed from the scene of its old successes. has been to take the existence of fields for granted. several of them quite up·todate. The traditional approach. why have fields at all? It does not seem satisfactory to me to appeal to experience. As to content. since the first papers of Heisenberg and Pauli on general quantum field theory. that provided my chief motivation in writing it.
The most immediate and certain consequences of relativity and quantum mechanics are the properties of particle states. but it is also one that has become newly appropriate. As a spinoff. the Hamiltonian.' that used to bother us when we thought of these theories as truly fundamental. the Hamiltonian. This is a point of view I have held for many years. Also. is to be constructed from creation and annihilation operators. we deduce the CPT theorem and the connection between spin and statistics. The point of view of this book is that quantum field theory is the way it is because (aside from theories like string theory that have an infinite number of particle types) it is the only way to reconcile the principles of quantum mechanics (including the cluster decomposition property) with those of special relativity.' lowenergy approximations to a deeper theory that may not even be a field theory. and the discussions here will reflect these changes. so here particles come first . the reason that quantum field theories describe physics at accessible energies is that any relativistic quantum theory will look at sufficiently low energy like a quantum field theory. Then in Chapter 5 we return to Lorentz invariance. as 'effective field theories. Chapter 3 provides a framework for addressing the fundamental dynamical question: given a state that in the distant past looks like a certain collection of free particles. such as nonrenormalizability and 'triviality. including quantum electrodynamics.they are introduced in Chapter 2 as ingredients in the representation of the inhomogeneous Lorentz group in the Hilbert space of quantum mechanics. We have learned in recent years to think of our successful quantum field theories. It is not until Chapter 7 that we come to Lagrangians and the canonical formalism The rationale here for introducing them is not that they have proved useful elsewhere in physics (never a very satisfying explanation) . On this basis. what will it look like in the future? Knowing the generator of timetranslations. we can answer this question through the perturbative expansion for the array of transition amplitudes known as the Smatrix. and show that it requires these creation and annihilation operators to be grouped together in causal quantum fields. The fonnalism is used in Chapter 6 to derive the Feynman rules for calculating the Smatrix.in terms of a few fundamental principles .Preface XXI not just to describe the world as we find it. In Chapter 4 the principle of cluster decomposition is invoked to describe how the generator of timetranslations.why the world is the way it is. It is therefore important to understand the rationale for quantum field theory in terms of the principles of relativity and quantum mechanics. but something different like a string theory. but to explain . we think differently now about some of the problems of quantum field theories. This is intended to be a book on quantum field theory for the era of effective field theories.
especially in Chapter 3. which historically appeared almost at the very start of relativistic quantum mechanics. but it seems to me that although path integration is by far the best way of rapidly deriving Feynman rules from a given Lagrangian. in part because the results we obtain help us to understand the necessity for field and mass renormalization. in general field theories. Tbe reader may feel that some of the topics treated here. I have met string theorists who have never heard of the relation between time·reversal invariance and final·state phase shifts. . and also as a concrete example of renonnalization in action. Quantum electrodynam· ics finally appears in Chapter 8. which also describes the modern view of nonrenormalizability that is appropriate to effective field theories. Wick rotation. this is the key condition that we need to prove the Lorentz invariance of the S·matrix. Volume I concludes with a series of chapters. Chapter 13 is a digression on the special problems raised by massless particles of low energy or parallel momenta. it rather obscures the quantum mechanical reasons underlying these calculations. is not seen here until Chapter 14. This chapter ends with a treatment of the Lamb shift. the quantum field theory of photons and electrons. The experience gained in Chapter 11 is extended to all orders and general theories in Chapter 12. as we show in Chapter 3.XXll Preface but rather that this formalism makes it easy to choose interaction Hamil· tonians for which the Smatrix satisfies various assumed symmetries. Chapter 11 presents the classic one·loop calculations of quantum chromodynamics. could more properly have been left to textbooks on nuclear or elementary particle physics. dimensional and PauliVillars regularization). and used to justify some of the hand·waving in Chapter 8 regarding the Feynman rules for quantum electrodynamics. Path integration is introduced in Chapter 9. Here too the arrangement is a bit unusual. we start with a chapter on non·perturbative methods. because this equation should not be viewed (as Dirac did) as a relativistic version of the Schrodinger equation. the Lorentz invariance of the Lagrangian density ensures the existence of a set of ten operators that satisfy the algebra of the Poincare group and. The Dirac equation for an electron in an external electromagnetic field. both as an opportunity to explain useful calculational techniques (Feynman parameters. that provide an introduction to the calculation of radiative corrections. So they might. involving loop graphs. without regard to whether the theory contains infinities or not. but in my experience these topics are usually either not covered or covered poorly. In particular. 1014. using specific dynamical models rather than the general principles of symmetry and quantum mechanics. This is a somewhat later introduction of path integrals than is fashionable these days. but rather as an approximation to a true relativistic quantum theory. on bound state problems. bringing the confrontation of theory and experiment up to date.
broken symmetries. electromagnetic. and strong interactions. Rather. I did not always know who was responsible for material presented here. so that they will be ready for whatever new developments in physics may take us beyond our present understandings. I intended that this book should be accessible to students who are familiar with nonrelativistic quantum mechanics and classical electrodynamics. So in the early chapters I have tried to err on the side of inclusion rather than exclusion. anomalies. as for instance in the proof that symmetry operators are either unitary or antiunitary. the use of the cluster decomposition principle. there are parts of this book whose lack of rigor will bring tears to the eyes of the mathematically inclined reader. I have tried to give citations both to the classic papers in the quantum theory of fields and to useful references on topics that are mentioned but not presented in detail in this book. Volume II will deal with the advances that have revived quantum field theory in recent years: nonAbelian gauge theories. the derivation of the reduction formula.Preface XX111 and nuclear theorists who do not understand why resonances are governed by the BreitWigner formula. and the mere absence of a citation should not be taken as a claim that the material presented here is original. I have also supplied problems for each chapter except the first. the discussion of superselection rules. and so on. I hope that I have improved on the original literature or standard textbook treatments in several places. I hope it will suit the physicists and physics students who want to understand why quantum field theory is the way it is. Indeed. Some of these problems aim simply at providing exercise in the use of techniques described in the chapter. ••• Much of the material in this book I learned from my interactions over . I have found that each of the two volumes of this book provides enough material for a oneyear course for graduate students. the analysis of particle degeneracy associated with unconventional representations of inversions. and even the calculation of the Lamb shift. the renormalization group. I assume a basic knowledge of complex analysis and matrix algebra. Nor is this a book for those who seek a higher level of mathematical rigor. but topics in group theory and topology are explained where they are introduced. But some of it is. instantons. This is not a book for the student who wants immediately to begin calculating Feynman graphs in the standard model of weak. others are intended to suggest extensions of the results of the chapter to a wider class of theories. In teaching quantum field theory. the derivation of the external field approximation.
1994 . and to Terry Riley for finding countless books and articles. lowe thanks for help in the preparation of the historical introduction to Gerry Holton. Joe Polchinski. far too many to name here. who prepared the illustrations and typed the It\TB< input files until I learned how to do it. and to my colleagues at the University of Texas: Arno Bohm. and especially to my editar. But I must acknowledge my special intellectual debt to Sidney Coleman. Phil Candelas. Cecile DeWittMarette. Texas October. and Paul Shapiro. Thanks are also due to Alyce Wilson. Vadim Kaplunovsky. Willy Fischler. STEVEN WEINBERG Austin. I am grateful to Maureen Storey and Alison Woollatt of Cambridge University Press for helping to ready this book for publication. Luis Boya. Arthur Miller. for his friendly good advice. Rufus Neal. Joaquim Gomis. Bryce DeWitt.XXIV Preface the years with numerous other physicists. and Sam Schweber. Josh Feinberg. Jacques Distler.
j. ~oo~1. respectively. where defined as the totally antisymmetric Spatial threevectors are indicated by letters in boldface. The 'LeviCivita tensor' quantity with E0 123 = + 1. at the end of an equation indicates the addition of the Hermitian O . v. 0. The Hermitian adjoint of an operator a is denoted at. xxv . with elements 1111 = 1'/22 = 1]33 The d'Alembertian is defined as 0 V2 is the Laplacian i)2ji'Jx i oxi . and so on generally run over the three spatial coordinate labels. 2. and Hermitian adjoint of a matrix or vector A are denoted A AT. with xO the time coordinate. transpose. fl1'lv + }"'}'p = 2~pv. 2. usually taken as 1. Greek indices fl. v~ v/lvl. ls The step function 8(s) has the value +1 for s > 0 and 0 for s < O. or +c. The complex. and f3 = i}'o.c. Dirac matrices }'p are defined so that iYOYI }'Jr'}. unless otherwise indicated. etc. Also.k. +Hc. except where an asterisk is used to emphasize that a vector or matrix of operators is not transposed. conjugate. The spacetime metric 1. EJiVPcr 15 = IJI'I' 82 / Gxl'(jx" = V2  a2 jotl . and At = A oT . 3. Repeated indices are generally summed. A hat over any vector indicates the corresponding unit vector: Thus. generally run over the four spacetime coordinate labels 1. '1/lV is diagonal. A dot over any quantity denotes the timederivative of that quantity. 3.Notation Latin indices i.
= e2 /4n ~ 1/137. 050. Numbers in parenthesis at the end of quoted numerical data give the uncertainty in the last digits of the quoted figure. . Where not otherwise indicated. 1173 (1994). Except in Chapter 1. Throughout e is the rationalized charge of the electron.' Phys.XXVI Notation adjoint or complex conjugate of the foregoing terms. we use units with Ii and the speed of light taken to be unity. Rev. experimental data are taken from 'Review of Particle Properties. so that the fine structure constant is r:r. A bar on a Dirac spinor u is defined by u = ut p.
it is historically correct that quantum field theory grew in part out of a study of relativistic wave equations. not because they are more fundamental. a knowledge of our history is a mixed blessing . If it turned out that some physical system could not be described by a quantum field theory. The underlying theory might not be a theory of fields or particles. including the Maxwell. even though this work was not published until 1939 and did not have a great impact for many years after. For many years after 1950 it was generally assumed that the laws of nature take the form of a quantum theory of fields. Nevertheless. This approach necessarily draws me away from the order in which the subject in fact developed. To take one example. it would be a cataclysm. lately there has been a reaction against looking at quantum field theory as fundamental. it has long seemed to me that a much better starting point is Wigner's definition of particles as representations of the inhomogeneous Lorentz group. I start with particles in this book. and Dirac equations. KleinGordon. more directly derivable from the principles of quantum mechanics and relativity. This is not to say that particles are necessarily more fundamental than fields. In this book we start with particles and get to the wave equations later. if it turned out that the system did not obey the rules of quantum mechanics and relativity. or to profit from their experience. emphasizing the deductive trail that ascends from the physical principles of special relativity and quantum mechanics. but because what we know about particles is more certain. I have tried in this book to present the quantum theory of fields in a logical manner.1 Historical Introduction Our immersion in the present state of physics makes it hard for us to understand the difficulties of physicists even a few years ago.it can stand in the way of the logical reconstruction of physical theory that seems to be continually necessary. 1 . like strings. In fact. it would be a sensation. For this reason it is natural that courses and treatises on quantum field theory introduce these wave equations early. At the same time. but perhaps of something quite different. and give them great weight.
when it finally assumed its modern form. in this chapter I will show explicit factors of Ii and c (and even h). quantum electrodynamics and the other quantum field theories of which we are so proud are mere 'effective field theories.' lowenergy approximations to a more fundamental theory. and is not needed as a basis for the rest of the book. if we want to know why quantum field theories are the way they are. . A word about notation. In order to keep some of the flavor of past times. so that the fine structure constant IX . Thus. readers who have no prior familiarity with quantum field theory may find parts of this chapter too brief to be altogether clear. where I discuss the early attempts to put together quantum mechanics with special relativity. Most of these are listed in the bibliography given at the end of this chapter. I urge such readers not to worry. but that any relativistic quantum theory will look like a field theory when applied to particles at sufficiently low energy. In this case I can only suggest that the reader should skip on to the less familiar parts. I will use the more modern rationalized electrostatic units for charge. we have to start with particles. Some readers may even prefer to start with the next chapter. On the other hand. the reader who is familiar with the history of quantum mechanics may find some material here that he or she already knows. This chapter will therefore present the history of quantum field theory from earliest times to 1949. but in order to facilitate comparison with modem physics literature.. simply setting n=c=1. One problem that I found in writing this chapter is that the history of quantum field theory is from the beginning inextricably entangled with the history of quantum mechanics itself. especially in the first section.. I have based it on books and articles by real historians. This chapter is not intended as a selfMcontained introduction to quantum field theory. On this basis. and come back to the history later. I should add that this chapter is not intended as an original work of historical scholarship. The reader who wants to go more deeply into historical matters is urged to consult these listed works.2 J Historical Introduction From this point of view. In the remainder of the book I will try to keep history from intruding on physics.. But we do not want to pay the price of altogether forgetting our past. However. and in the list of references. 1/137 is e2/4rrlic.. The reason that our field theories work so well is not that they are fundamental truths. In subsequent chapters I will mostly use the 'natural' system of units. for many readers the history of quantum field theory should serve as a good introduction to quantum field theory itself. plus some historical reminiscences and original physics articles that I have read.
and if this phase is to be Lorentz invariant. is an impossibility. the chief clue was Lorentz invariance: if particles are described by a wave whose phase at position x and time t is of the form 271"(K' x . de Broglie was able to derive the old quantization conditions of Niels Bohr and Arnold Sommeneld. v ~ E/h . Pascual Jordan and Wolfgang Pauli in the years 19251926. Max Born. then the vector K and the frequency v must transform like x and t. Also.1. both de Broglie and Walter Elsassec2 suggested that de Broglie's wave theory could be tested by looking for interference effects in the scattering of electrons from crystals. the development of matrix mechanics4 by Werner Heisenberg.1 Relativi8tic Wave Mechanics 3 1. 3 However. relativistic wave mechanics survived as an important element in the formal apparatus of quantum field theory. so it was natural to assume that. as for instance for an electron in a general Coulomb field. The possibility that material particles can like photons be described in terms of waves was first suggested! in 1923 by Louis de Broglie. took a good deal of their inspiration from special relativity. It was only later that it became generally clear that relativistic wave mechanics. so wave packets just keep up with the particle they represent. Louis de Broglie and Erwin Schrodinger. for material particles. Indeed. Nevertheless. Germer.1. despite its many successes.1) just as for photons. For photons. and therefore must be proportional to them.1. = 1/11(1. such effects were established a few years later by Clinton Joseph Davisson and Lester H.vt). The group velocity ov/JI( of the wave then turns out to equal the particle velocity. which though quite mysterious had worked well in accounting for atomic spectra. Thus.1) should be modified for non·free particles. relativistic wave mechanics was ultimately to give way to quantum field theory. with the same constant of proportionality. Wave mechanics was by·passed in the next step in the history of quantum mechanics. By assuming that any closed orbit contains an integral number of particle wavelengths A. Apart from the analogy with radiation.1 Relativistic Wave Mechanics Wave mechanics started out as relativistic wave mechanics. it was still unclear how the de Broglie relations (1. (1. and hence like p and E. and it posed a challenge to field theory. in the sense or a relativistic quantum theory of a fixed number of particles. At least part of the inspiration for matrix mechanics was the . one had the Einstein relation E = hv. as we shall see. the founders of wave mechanics. "~p/h . to reproduce its successes. In order for this to be possible I( and v must have the same velocity dependence as p and E.
~ iliV . Heisenberg's 1925 paper opens with the manifesto: 'The present paper seeks to establish a basis for theoretical quantum mechanics founded exclusively upon relationships between quantities that in principle are observable.t) satisfying the equation obtained by making the same replacements in • This is Lorentz invariant. Schriidinger actually wrote Hand p in terms of partial derivatives of an action function. ¢ would be described by a wave function 1p(x. According to Dirac. Only later. as for instance in the introduction of the Smatrix by John Wheeler and Heisenberg (see Chapter 3) and in the revival of dispersion theory in the 1950s (see Chapter 10).1.e2(p + eA/c)' .vt)}. the de Broglie relations (1. it had already been independently rediscovered by Oskar Klein 7 and Walter Gordon. . Schrodinger guessed that an electron in the external fields A.1. wave mechanics was revived by Erwin Schrodinger. (1.2) For afree particle described by a plane wave exp {2lti(K' X.s and for this reason it is usually called the 'Klein~Gordon equation.1. then became discouraged because it gave the wrong fine structure for hydrogen. such as the energy levels. the Hamiltonian H and momentum p are related by· a ~ (H + e</»2 . iJ E ~ hv ~ Iii iJt ' (1.3) where Ii is the convenient symbol (introduced later by Dirac) for h/2lt. for a 'Lorentz electron' of mass m and charge e in an external vector potential A and Coulomb potential ¢. In his 1926 series of papers.' Schr6dinger's relativistic wave equation was derived by noting first that. and then used to rederive the results of matrix mechanics. though modern quantum field theory is very far from this ideal.6 the history is actually quite different: Schrodinger first derived the relativistic equation. . but this makes no difference to our present discussion. or emission and absorption rates. By an admittedly formal analogy. As everyone knows.4 1 Historical Introduction insistence that the theory should involve only observables. is a relativistic wave equation offered. in the sixth section of the fourth paper.1) can be obtained by the identifications P ~ h. because the quantities A and ¢ have the same Lorentz transformation property as cp and E.m2e' .s the familiar nonrelativistic wave equation is suggested first.' This sort of positivism was to reemerge at various times in the history of quantum field theory. and then some months later realized that the nonrelativistic approximation to his relativistic equation was of value even if the relativistic equation itself was incorrect! By the time that Schr6dinger came to publish his relativistic wave equation. It would take us too far from our subject to describe matrix mechanics in any detail here.
1.1.. This gave a fine structure splitting in agreement with experiment: for instance.' roughly 1/137. (1.. Schrodinger's result (1.6) gives an n = 2 fine structure splitting a4mc2 /12. for the stationary states of hydrogen we have A = 0 and ¢ = el4ltr. according to Dirac.1. and tp has the timedependence exp( iEt/h). On the other hand.7) where m is the electron mass. Balmer. is an integer with 0 < { < n . In contrast. obtained using the rules of the old quantum theory: E ~ me' [1 _ 2_ a'.1. t). The a. split by the observed amount a4 mc2 /32.1 Relativistic Wave Mechanics 5 (1.1.lC.1. to It is instructive here to compare Schrodinger's result with that of Arnold Sommerfeld.] t + '2 4 . considerably larger than observed. 2 term gave good agreement with the gross features of the hydrogen spectrum (the Lyman. the orbital angular momentum in units of h.4) In particular.t + e¢) c 2_ 2 ( ihV + e:) m tp(x. Here k is an integer between 1 and n.] (1.53 x 105 eV. this led George Uhlenbeck and Samuel Goudsmit ll in 1925 to suggest that the electron has an intrinsic angular .4) becomes o ~ [(E + ::.2 n k 4 · a'. (1.1.. _ a4 4 ( 2n 2n n[ _~) + .e2 /4nnc is the 'fine structure constant. Schrodinger correctly recognized that the source of this discrepancy was his neglect of the spin of the electron. n (~_~) + .6 it was this agreement that led Schrodinger eventually to develop his nonrelativistic wave equation. or 4.. The splitting of atomic energy levels by noninversesquare electric fields in alkali atoms and by weak external magnetic fields (the socalled anomalous Zeeman effect) had revealed a multiplicity of states larger than could be accounted for by the BohrSommerfeld theory.2) : 0= [ (m (. which in Sommerfeld's theory is given in terms of the orbital angular momentum th as k = t + 1. etc. for n = 2 Eq.7) gives two levels (k = 1 and k = 2).6) where a .1.1. and t. c 2_ 2 4 ] (1. 11 is a positivedefinite integer. the a4 term gave a fine structure in disagreement with existing accurate measurements of Friedrich Paschen. series) and. so (1.)'  e'h'V'  m'e'] ~(x) (LLS) Solutions satisfying reasonable boundary conditions can be found for the energy values9 E ~ me' [1 _ a'.
1. However. ! . there still was not a thorough relativistic theory which incorporated the electron's spin from the beginning.4.1. p ~  !.9) By accident Sommerfeld's theory had given the correct magnitude of the splitting in hydrogen (j + like k runs over integer values from 1 to n) though it was wrong as to the assignment of orbital angular momentum values t to these various levels. so that Schrodinger's relativistic equation should not be expected to give the correct fine structure splitting..1.1.l 4 Together. At the beginning of his 1928 paper.J . instead. rather than one of the many other types of particles with various spins that are known today.8) 2me It was clear that the electron's spin would be coupled to its orbital angular momentum. in 1928 it was possible to believe that all matter consisted of electrons. _ . instead of being satisfied with the point charge. he approached the problem by posing a question that would today seem very strange. the multiplicity of the fine structure levels in hydrogen was now predicted to be 2 for j = and 2(2j + 1) for j > (corresponding to t values j + in agreement with experiment.6) and experiment. and perhaps something similar with positive charge in the ! ! i). However. There are really two effects here: one is a direct coupling between the magnetic moment (1. by 1927 several authors 13 had been able to show that the spinorbit coupling was able to account for the discrepancy between Schrodinger's result (1. these two effects were found to lift the to the energy (1.7) given level with total angular momentum j = t by Sommerfeld for k = t + 1 = j + while the level with j = t ~! was lowered to the value given by Sommerfeld for k = t = j Thus the energy was found to depend only on nand j. the magnitude of the Zeeman splitting 12 allowed them to estimate further that the electron has a magnetic moment eli . (1. 4 (1. +! +!. this question is like asking why bacteria have only one celL having spin n/2 is just one of the properties that define a particle as an electron. the other is the relativistic 'Thomas precession' caused (even in the absence of a magnetic moment) by the circular motion of the spinning electron. E ~ me' [1 _ . but not separately on t. Indeed. he did not set out simply to make a relativistic theory of the spinning electron.15 he asks 'why Nature should have chosen this particular model for the electron..6 1 Historical Introduction momentum nil. Despite these successes.1. Also. In addition. Such a theory was discovered in 1928 by Paul Dirac. ( 2n 2n j +"2 n 1~) + .2.' To us today.8) and the magnetic field felt by the electron as it moves through the electrostatic field of the atom.
but I was not because of the negative probabilities that it led to.1. worrying about getting a theory which would have only positive probabilities. To quote Dirac's reminiscences l7 about this problem I remember once when I was in Copenhagen.1. which can be formed from solutions of the relativistic Schrodinger equation and which satisfy a conservation law.1 )2 V· (~V"'V.18 Dirac found the answer to this problem on an evening in 1928 while staring into a fireplace at St John's College. The problem therefore . the only probability density p and current J. • • or m so the spaceintegral of IIpI 2 is time~independent. Dirac's question can be restated: 'Why do the fundamental constituents of matter have to have spin n/2?' For Dirac.Vv' )~O (. . (1.1.(1". It was known 16 that the probability density for the nonrelativistic Schrodinger equation is 111'1 2 . Thus.11) (1. According to George Gamow.1. On the other hand. that Bohr asked me what I was working on and I told him I was trying to get a satisfactory relativistic theory of the electron. I just kept on with it. because (with or without an external potential ¢) p does not have definite sign. in the spirit of the times in which it was asked. and that this satisfies a continuity equation of the form . the key to this question was the requirement that probabilities must be positive. and Bohr said 'But Klein and Gordon have already done that!' That answer first rather disturbed me.1 Relativistic Wave Mechanics 7 atomic nucleus. It is not possible to identify p as the probability density. Cambridge. He realized that the reason that the KleinGordon (or relativistic Schrodinger) equation can give negative probabilities is that the p in the conservation equation (1. 2 In .1.10) involves a timederivative of the wave function. This in turn happens because the wave function satisfies a differential equation of second order in the time.12) with N an arbitrary constant.10) are of the form (1.. Bohr seemed quite satisfied by Klein's solution.
15) (1.13) where Jf' is some matrix function of space derivatives.) But this must agree with the freefield form of the relativistic Schrodinger equation (1. In order to have a chance at a Lorentzinvariant theory. where Jij is the Kronecker delta (unity for i = j.1.8 1 Historical Introduction was to replace this wave equation with another one of first order in time derivatives.p n ()'t2 2 = .1.1.1Z3. Suppose the electron wave function is a multicomponent quantity 1pn(x). which just expresses the relativistic relation between momentum and energy.j) and 1 is the unit matrix. we must suppose that because the equation is linear in timederivatives. which satisfies a wave equation of the Conn. ""4  0. alp In Dr ~ Jf'lp. From (1.1 . i and j run over the values 1. it is also linear in spacederivatives. (I4 = 1 0 0 0 0 0 0 1 0 0 0 1 0 0 0 1 . or x.1. c4 Ct:4!P • uXj (The summation convention is in force here.16) (1.1. Therefore. the matrices IX and (14 must satisfy the relations al(Ij +ajeli (Ijel4 = 2J jj l.18) 0 0 1 0 o 1 . so that Jt' takes the form: (1.J(' rp = 2 J2tp n 2C2Clia: j aXjOXj ~ '.13) we can derive the secondorder equation a2.4).14) where 1Zj.1. Inmc 3( Ct:i0C4 a~ 2 + Ct:4::r:i ) ~ + m. and cq are constant matrices. (1. zero for i 1.1. 3.17) + (I4C(i = .1.' . z. . y. 1Z2. (1. like the nonrelativistic Schrodinger equation. Dirac found a set of 4 x 4 matrices which satisfy these relations :XI = o o o o o 0 0 1 1 0 0 1 0 0 1 0 0 0 1 0 0 1 0 0 0 0 el2 = 0 0 0 I 0 0 0 I I 0 0 0 I 0 0 0 . 2. (1.
.23) Dirac used this equation to show that in a central field.. (These matters are discussed more fully. The wave equation (1.1. and correspondingly 14 = 0:4. Dirac multiplied Eq. 3 (L1. a 1p = e (iii :t + e¢) (iltcV + eA)' OC1p + mclCl4VJ..}'v. He concluded from this that AI!. 0.1. in the sense that they are also satisfied by the matrices Alt.1.20) (The Greek indices 11.19) where yO i:X4 .13) on the left with 0:4.2l) . from a rather different point of view. will now run over the values 1. (1. under a Lorentz transformation x ll _ Ali vx v .14) and version of the spin matrix introduced earlier by Pauli 19 the 4 x 4 o o 0 I 0 0 0 1 I 0 0 0 o I 0 0 • (L1. etc. Dirac followed the 'usual procedure' of making the replacements lit ._ lit ~ + e¢ . Dirac used X4 = ict.1.>1'. the conservation of angular momentum takes the form [.24) IS where J't is the matrix differential operator (1. so that it could be put in the form [ ltCyll . v. ~mrxV+h. (1..1..1. where A is any Lorentz transformation.1 Relativistic Wave Mechanics 9 To show that this formalism is Lorentzinvariant.2I) 11=\'=0 ~'fV Dirac noted that these anticommutation relations are Lorentzinvariant.l ~ Sl(A)yPS(A).1.) To study the behavior of electrons in an arbitrary external electromagnetic field.1. 3.13) then takes the form . 8 ()x ll + mel] 1p = 0. (1.1. the wave function undergoes the matrix transformation tp _ S(A)1p.) The matrices 'Ill satisfy the anticommutation relations ~ "~v~1.22) ot lit c as in Eq. (I (1./2] ~O. in Chapter 5. It follows that the wave equation is invariant if. 2.." must be related to yll by a similarity transfonnation: N'.!. with xO = ct.4). (1.iliV _ iltV + A (1. a . (1.2.1.
However.2 {n. Jordan. This theory achieved Dirac's primary aim: a relativistic formalism with positive probabilities.9). with constant total probability J 11p1 2d 3x. B .13) we can derive a continuity equation (L1.1. and Charles G. an 'exact' formula for the hydrogen energy levels in Dirac's theory was derived by Darwin20 and Gordon 21 1/2 1+ . (1. From (1. . the first term dominates.1.23).27) The first three terms of a power series expansion in cr. For a given momentum p. (1.1. J ~ CV. Dirac also iterated Eq. this magnetic moment.8) found by Goudsmit and Uhlenbeck.4 mc 2) with that found by Heisenberg. ll As Dirac recognized. and represents a magnetic moment in agreement with the value (1. x .1.30) Two solutions with E = + Vp 2c2 + m2 c4 correspond to the two spin states of an electron with Jz = +11/2.1.l + [(i+l)2 "r} 1 2 (L1. obtaining a secondorder equation.29) so that the positive quantity 1~)12 can be interpreted as a probability density. the presence of the extra tenn in (1.1.28) with p ~ IV'I'.4) except for the presence on the righthandside of two additional terms [enctJ .EtJ] . there was another difficulty which Dirac was not immediately able to resolve.ienca.24) shows that the electron has intrinsic angular momentum n/2. Ell/' .2 agree with the approximate result (1.13) has four solutions of the plane wave fonn '" oc exp [~(p.""" (1. the wave equation (1. The other two solutions have E = .26) For a slowly moving electron. guaranteed that this theory would give a fine structure splitting in agreement (to order cr.10 1 Historical Introduction Since each component of q has eigenvalues +1.1. Darwin. together with the relativistic nature of the theory. 13 A little later.i .1. (L1. which turned out to have just the same form as the KleinGordon equation (1.1.
The problem of negative energies is much more troublesome in relativistic quantum mechanics. Of course. Since the positive solutions have E > me 2 and the negative ones have E < _me 2. The periodic table of the elements and the systematics of Xray spectroscopy had together by 1924 revealed a pattern in the population of atomic energy levels by electrons: 23 The maximum number N n of electrons in a shell characterized by principal quantum number n is given by twice the number of orbital states with that n .. I I The exclusion principle answered a question that had remained obscure in the old atomic theory of Bohr and Sommerfeld: why do not all the electrons in heavy atoms fall down into the shell of lowest energy? Subsequently Pauli's exclusion principle was formalized by a number of authors 25 as the requirement that the wave function of a multielectron system is antisymmetric in the coordinates. there are two solutions of the form (1.31) Wolfgang Pauli24 in 1925 suggested that this pattern could be understood if N n is the total number of possible states in the nth shell.Vp 2C2 + m2c4. with the energy carried off by two or more photons. Why then is matter stable? In 1930 Dirac offered a remarkable solution. E = + Vp 2c2 + m2 e4 . in classical physics we can simply assume that the only physical particles are those with positive E. . one with positive E and one with negative E. even in classical physics.30). 22 Dirac's proposal was based on the exclusion principle. As Dirac pointed out.1 Relativistic Wave Mechanics 11 .' . and no continuous process can take a particle from positive to negative energy. and no obvious physical interpretation. and as we have seen this was understood a little later as due to the spin of the electron.1. However. 8. This principle was incorporated into statistical mechanics by Enrico Fenni26 and Dirac.27 and for this reason particles obeying the exclusion principle are generally called 'fermions. so a few words about the history of this principle are in order here.1. there is a finite gap between them.. orbital and spin.31) as due to a 'peculiar. and ifin addition there is some mysterious 'exclusion principle' which forbids more than one electron from occupying the same state. the relativistic relation E 2 = p2 c2 + m 2c4 has two solutions. ~ 2L(2t+ 1) ~ 2n' ~ 2. 18. t=o (1.15 the interaction of electrons with radiation can produce transitions in which a positiveenergy electron falls into a negativeenergy state. He explained the puzzling factor 2 in (Lt. classically nondescribable duplexity' of the electron states.1 N. this problem arises also for the relativistic Schrodinger equation: for each p.1. of all the electrons. As Dirac pointed out in his 1928 paper.
For these reasons.' The few vacant states. as well as in chemistry and atomic physics.3o who apparently did not know of this prediction by Dirac. emitting a pair of gamma ray photons. . Anderson. but a discussion of these matters would take us too far afield here. The only particle with positive charge that was known at that time was the proton. or 'holes: in the sea of negative energy electrons behave like particles with opposite quantum numbers: positive energy and positive charge. On August 2.. and decided that the holes would have to appear not as protons but as a new sort of positively charged particle. it was even hoped by some that this would provide an explanation.27a 'the whole climate of opinion at that time was against new particles' so Dirac identified his holes as protons. in fact. then lacking. By itself this would not have created difficulties for the hole theory. 29a The second and third of these problems were eliminated by the discovery of the positron by Carl D. white dwarf and neutron stars... The track was observed to curve in a direction that would be expected for a positively charged particle. of the same mass as the electron. by 1931 Dirac had changed his mind. Dirac hoped that Coulomb interactions between electrons would somehow account for these differences but Hermann Weyl28 showed that the hole theory was in fact entirely symmetric between negative and positive charge. Finally. it was soon pointed out29 by Julius Robert Oppenheimer and Igor Tamm that electronproton annihilation in atoms would take place at much too fast a rate to be consistent with the observed stability of ordinary matter. the title of his 1930 article 22 was 'A Theory of Electrons and Protons: The hole theory faced a number of immediate difficulties. One obvious problem was raised by the infinite charge density of the ubiquitous negativeenergy electrons: where is their electric field? Dirac proposed to reinterpret the charge density appearing in Maxwell's equations as 'the departure from the normal state of electrification of the world: Another problem has to do with the huge dissimilarity between the observed masses and interactions of the electrons and protons. 1932. of the energy source of the stars.' The exclusion principle has played a fundamental role in the theory of metals. and as Dirac later recalled. a peculiar cosmic ray track was observed in a Wilson cloud chamber subjected to a 15 kG magnetic field. Dirac22 predicted the existence of an electronproton annihilation process in which a positiveenergy electron meets a hole in the sea of negativeenergy electrons and falls down into the unoccupied level. .12 1 Historical Introduction just as particles like photons for which the wave function is symmetric and which obey the statistics of Bose and Einstein are called 'bosons. etc. . and yet its range was at least . Dirac's proposal was that the positive energy electrons cannot fall down into negative energy states because 'all the states of negative energy are occupied except perhaps a few of small velocity. However.
for every kind of particle there is an 'antiparticle' with the same mass and opposite charge.for instance.' consisting of a spinless negative 71" or K meson bound to an atomic nucleus. The discovery of the moreorless predicted positron. and therefore do not need to be described by a relativistic wave equation. Yet even in the 1920s particles of zero spin were known .DAD DE :0 i •~ '~ILE . from the stationary solutions of the relativistic KleinGordonSchrodinger equation! Thus. it could be argued that hydrogen atoms and alpha particles are not elementary. UNIV. and so on . such as the 7l"± mesons or W± particles. as would be expected for one of Dirac's holes. it is difficult to agree that there is anything fundamentally wrong with the relativistic equation for zero spin that forced the development of the Dirac equation . (ii) As far as we now know.1.that are no less elementary than the proton and neutron. the hydrogen atom in its ground state.M. and the helium nucleus. Of course.) But how can we interpret the antiparticles of charged bosons. apart from effects of the strong interactions. as holes in a sea of negative energy states? For particles quantized according to the rules of BoseEinstein statistics. gave Dirac's theory a prestige that it has held for over six decades. such as the photon. Anderson quotes press reports of evidence for light positive particles in cosmic ray tracks. together with the earlier successes of the Dirac equation in accounting for the magnetic moment of the electron and the fine structure of hydrogen. are their own antiparticles. not zero.1 Relativistic Wave Mechanics 13 ten times greater than the expected range of a proton! Both the range and the specific ionization of the track were consistent with the hypothesis that this was a new particle which differs from the electron only in the sign of its charge.RSi. Today we know of a large number of spin zero particles . Further. (Some purely neutral particles. there are serious reasons for being dissatisfied with its original rationale: (i) Dirac's analysis of the problem of negative probabilities in Schrod inger's relativistic wave equation would seem to rule out the existence of any particle of zero spin.which seem as elementary as the electron or any other particle. we would today calculate the fine structure of 'mesonic atoms.) Thus it appeared that Dirac was wrong only in his original identification of the hole with the proton. We also know of spin one particles . although there seems little doubt that Dirac's theory will survive in some form in any future physical theory. However.the W± and ZO . but it was not (and still is not) clear how the idea of elementarity is incorporated in the formalism of relativistic quantum mechanics. Blackett.the problem simply is that the electron happens to have spin n12. (This discovery had been made earlier by P.S. obtained by Blackett and Giuseppe Occhialini. K mesons.Jr mesons.:.o. but not immediately published by him.
l]VJF11V (1. Dirac referred to the fact that [or bosons 'we no longer have the picture of a vacuum with negative energy states filled up'. these problems were all eventually to be solved (or at least clarified) through the development of quantum field theory. even though unfortunately it lingers on in many textbooks. As we shall see in this book.) This term would give an additional contribution proportional to K to the magnetic moment of the electron. and then making the substitutions (1. .1.1.) This term could be obtained by first adding a term to the freefield equations proportional to [yll.23). as the magnetic moment (1.yV](a 2/Jxl'ox V)1p. pOl = E" etc. which of course equals zero.14 1 Historical Introduction there is no exclusion principle.30U 'The picture of an infinite sea of negative energy electrons is now best regarded as a historical curiosity. this factor of 2 had remained mysterious until Dirac's theory.' The next section will show how the development of quantum field theory made the interpretation of antiparticles as holes unnecessary. At the point where we brought electric and magnetic fields into the wave equation (1.32) with arbitrary coefficient K. why should we believe it for fermions? I asked Dirac in 1972 how he then felt about this point.3.' In a lecture 271l a few years later. And if the hole theory does not work for bosonic antiparticles. (Here F11V is the usual electromagnetic field strength tensor.1.1. there was no reason to expect any particular value for the magnetic moment of the electron in Dirac's theory. there is really nothing in Dirac's line of argument that leads unequivocally to this particular value for the magnetic moment.' (iii) One of the great successes of the Dirac theory was its correct prediction of the magnetic moment of the electron. and hence nothing to keep positiveenergy particles [rom falling down into the negative·energy states. and so there is no reason why such a term should not be included in the field equations. including Lorentz invariance and gauge invariance.32) is consistent with all accepted invariance principles. and forgotten. he told me that he did not regard bosons like the pion or W± as 'important.22) as before. occupied or not.8) is twice as large as would be expected for the orbital motion of a charged point particle with angular momentum n12.1. However. (See Section 12. with pl2 = B3. A more modern approach would be simply to remark that the term (1. so apart from the possible demand for a purely formal simplicity. To quote Julian Schwinger. This was particularly striking. we could just as well have added a 'Pauli term'3! KCq["I I1 . and remarked that in this case 'the whole theory becomes more complicated.
By analogy with either the case of a string or the full electromagnetic field. t This yields a 'momentum' pdt) ~ L.2.2 The Birth of Quantum Field Theory 15 1. as had been anticipated 20 years earlier by Paul Ehrenfest. the radiation field u(x.q(t)). then lik(t) ~.5) . D (. the 'momentum' pdt) canonically conjugate to qdt) is determined.2. "2 qdt) (1. and Jordan 32 applied their new methods to the free radiation field.2.1. they ignored the polarization of electromagnetic waves and worked in one space dimension.Pk () H(p(t). Born. in one of the central papers on matrix mechanics.3 ) knc/ L.2. as in particle mechanics. Thus it is natural that the formalism of quantum field theory should have been developed in the first instance in connection with radiation and only later applied to other particles and fields.t) = I:qk(t) sin k=l Wk _ (w~x) ( (1. by the condition that if H is expressed as a function of the ps and qs. 32a In particular.2) (1. In 1926.2. the Hamiltonian was taken to have the fonn (1. with coordinate x running from 0 to L.2 The Birth of Quantum Field Theory The photon is the only particle that was known as a field before it was detected as a particle. Heisenberg. (1. so that H ~~ t {libt) + W[q[(t)} .4) k=l Thus the string or field behaves like sum of independent harmonic oscillators with angular frequencies Wk. For simplicity. the field u was expressed as a sum of Fourier components with u = 0 at both x = 0 and x = L: u(x.t) if constrained to vanish at these endpoints thus has the same behavior as the displacement of a string with ends fixed at x = 0 and x = L.1) In order to reduce this expression to a sum of squares.
9) with ak a timeindependent matrix and at its Hennitian adjoint.2. The physical interpretation of a column vector with integer components nt..12) and (1.2. = Vnk + 1 b~. (1.k For a single normal mode.13) do satisfy the commutation relations (1.jf 0 o .qj(t) ~ L J'j..q.!.n. satisfying the commutation relations [ ak .. Heisenberg.!.2.2.(t).m.nt1 II Onjnj ..(t)..3 0 o o o 0 It is straightforward to check that (1. = JYik Oni.6) (1.. (1. The qmatrix is given by qk(t) = In. and Jordan through previous work on the harmonic oscillator. "2. 2 .(t) ~ .2.qj(t)] (1..nk+l II Onjnj . j.L oq.2.(t) ~ w. (1.·..nl.(t) ~ L p..2... (1. is that it represents a state with nk quanta in each normal mode k. oj J = Jkj ...10) and (1.10) [a"aj]~O. [ak exp(iwkt) LWk + areXP(+iwkl)] (1..11)..6)(1.2.2.8) was already known to Born.)2 o o 0 0 0 .12) (1..13) (a!)n'l'n.. the timedependence of qk(t) is governed by the Hamiltonian equation of motion .n2..2.nl.jf00 00.2. j."" one for each normal mode..(t).qj(t)] ~ L p.2.k (1.)20 000. Also. [qk(t).11) The rows and columns of these matrices are labelled with a set of positive integers nl.2..8) The form of the matrices defined by Egs.16 1 Historical Introduction so the canonical commutation relations may be written 2[ ] 2in [iJ.n2... these matrices may be written explicitly as 0. The matrix elements are (ak)n. The matrix ak or acting on such a column vector will respectively at . 2 oH 2 q.2.3 o 0 0 0 0 .7) ~o.
18) A straightforward classical calculation then gives the radiated power. (1.9) and (1. Born.15) We see that the energy of the state is just the sum of energies nWk for each quantum present in the state. . Using (1.4) gives H= 'L.nwk k (alak + ~) .2.1 .n2 . Born and Jordan 33 had assumed in effect that an atom. This interpretation is further borne out by inspection of the Hamiltonian. plus an infinite zeropoint energy Eo = iLk Jiw k • Applied to the radiation field.) However. this formalism justified the Bose method of counting radiation states according to the numbers nk of quanta in each normal mode. energy fluctuations in blackbody radiation. in dropping from a state {1 to a lower state :x.n~.2niv t) + rpil· exp(2niv t) . it is necessary to go back in time a bit.. this approach was soon applied to a more urgent problem._= 'L.s.7). (For this purpose they actually only used the commutation relations (1. it is annihilated by any ak. (1. ct element of the matrix associated with the electron position.2.2. the calculation of the rates for spontaneous emission of radiation. where hv ~ (1.E.1.2.. and dividing by the energy hv per photon gives the rate of photon emission 1611: 3e2v3 2 3hc 3 Irp(l.2.n.2. is the {1. In particular.l exp( . In order to appreciate the difficulties here..2.2. The energy E of such an oscillator is E = ~m (t 2 + (211:v)2r2) = 8n2mv2Irp:t12. (1.6)(1..16) Ep .2. In one of the first papers on matrix mechanics.14) The Hamiltonian is then diagonal in the l1representation (H)n.m..17) and rp(l. (1. they may therefore be interpreted as operators which annihilate or create one quantum in the kth normal mode.2. would emit radiation just like a classical charged oscillator with displacement r( t) = rpf.nwk (nk + 1) II bn. leaving all I1t with t f k unchanged. and Jordan used this formalism to derive an expression for the r.2 The Birth of Quantum Field Theory 17 lower or raise 11k by one unit. the vector with all 11k equal to zero represents the vacuum.2..10) in (1.nj' k j (1. Heisenberg.19) .
and the coefficients were shown to satisfy commutation relations like (1. he was able to derive formulas [or the rates uB(t1.20) in this relation immediately yields the BornJordan result (1. each state of the free radiation field was specified by a set of integers nk.fJ is just r{!~·.) ~ (8nc~.2. (1.2.f) nk x hVk. with matrix coefficients proportional to the matrices ak and at defined in Eqs.10)(1.2.13).20) (Note that the expression on the right is symmetric between states Cl and ft.) of spontaneous emission and the rates uB for induced emission or absorption: A(p ~.2.3) H(P ~ . it was not at all clear why the formulas for emission of radiation by a classical dipole should be taken over in this manner in dealing with spontaneous emission. as in Eq. The crucial result here is the factor )nk + 1 in Eq. one for each nonnal mode. A took the form of a sum over normal modes. t) was expanded in normal modes.2. Finally.2).2. (1.) Einstein34a had already shown in 1917 that the possibility of thennal equilibrium between atoms and blackbody radiation imposes a relation between the rate A(p + 0. 2n 2e2 (1.6). (1.21) Using (1. n k c3u(Vk) . Nevertheless. or nk + 1. The vector potential A(x. 34 By considering the behavior of quantized atomic states in an oscillating classical electromagnetic field with energy density per frequency interval u at frequency (1.)" 3h2 l'pol'.2.) . because T'l.2. ~ P) ~ H(P ~ . and the matrix elements of the electromagnetic interaction er .2. in 1927 Dirac35 was able to give a thoroughly quantum mechanical treatment of spontaneous emission. But in a radiation field with nk photons in a normal mode k. so the rate for emission of radiation in normal mode k is proportional to nk+ 1 =8hv 3 +1. it still seemed unsatisfactory that thermodynamic arguments should be needed to derive formulas for processes involving a single atom.2.13). the probability for a transition in which the number of photons in a nonnal mode k rises from nk to nk + 1 is proportional to the square of this factor.19) for the rate of spontaneous emission. In consequence. the energy density u per frequency interval is U(Vk) = (8.. + {J) and uB(p + Cl) for absorption or induced emlSSlon: H(. (1.17).18 1 Historical Introduction However. A little later a more convincing though even less direct derivation was given by Dirac.
Eq.2. Using his earlier result (1. The first step in this direction seems to have been taken in 1927 by Jordan.2.2 The Birth of Quantum Field Theory 19 The first term is interpreted as the contribution of induced emission.23) O. These matters were put on a finner foundation a little later by Enrico Fermi. 36h Many physicists in the 1930s learned their quantum electrodynamics from Fermi's 1932 review.'. similar methods were used by Dirac to give a quantum mechanical treatment of the scattering of radiation and the lifetime of excited atomic states. without any appeal to thermodynamics. Instead. The relations can be satisfied by matrices labelled by a set of integers .10).20) for B.36 and by Victor Weisskopf and Eugene Wigner to make a detailed study of spectral line shapes.1. The use of canonical commutation relations for q and p or a and at also raised a question as to the Lorentz invariance of the quantized theory. and the second term as the contribution of spontaneous emission. 39 In 1928 an essential element was supplied by Jordan and Wigner. because these relations require nk to take all integer values from o to CX. Gkaj+ajak=Ojk.2. Jordan and Pauli 37 in 1928 were able to show that the commutators of fields at different spacetime points were in fact Lorentzinvariant. (1. (1.22) (1. The electron field therefore cannot he expanded as a superposition of operators satisfying the commutation relations (1.2. Hence. akaj +ajak = (1. they proposed that the electron field should be expanded in a sum of operators ak. in a manner which did not preserve the manifest Lorentz and gauge invariance of classical electrodynamics.11). Dirac could conclude that the ratio of the rates uB for induced emission and A for spontaneous emission is given by the Einstein relation.2. It was not long after the successful quantization of the electromagnetic field that these techniques were applied to other fields. Dirac was thus able to rederive the BornJordan formula 33 (1. at satisfying the anticommutation relations t t . A little later.19) for spontaneous emission rate A. such as the Dirac wave function of the electron. the fields to be quantized were the wave functions used in oneparticle quantum mechanics. At first this was regarded as a 'second quantization'. (These commutators are calculated in Chapter 5.2.21). Bohr and Leon Rosenfeld 38 used a number of ingenious thought experiments to show that these commutation relations express limitations on our ability to measure fields at spacetime points separated by timelike intervals. 36u Dirac in his work was separating the electromagnetic potential into a radiation field A and a static Coulomb potential AD.2. 40 They recognized that the Pauli exclusion principle prevents the occupation number nk of electrons in any normal mode k (counting spin as well as position variables) from taking any values other than 0 or 1.) Somewhat later.
also.. nj 0 otherwise...25) do satisfy the anticommutation relations (1. nj for j 1= k 1= k (1. nk = 1. corresponding to the values unity and zero of n' and n.2. each integer taking just the values zero and one: (ak )n. . rather than to the coefficients of the normal modes appearing in the fields. J Historical Introduction "2. in which case it gives zero. nk = 0. Heisenberg and Pauli took the Lagrangian L as the spaceintegral of a local function of fields and spacetime derivatives of fields. . Ok destroys a quantum in normal mode k if there is one there already.n..20 n1. just as for hosans. nj = 0 otherwise. .. is that it represents a state with nk quanta in each normal mode k. that since each nk takes only the values a and 1. as required by the Pauli exclusion principle.25) ~{ = nj for j For instance.224) and (1..112. The difference is.111 .2. Much later it was shown by Fierz and Pauli40a that the choice between commutation and anticommutation relations is dictated solely by the particle's spin: commutators must be used for particles with integer spin like the photon. Again." (at) n. and anticommutators for particles with halfMinteger spin like the electron.. ~{ 1 n~ = 0. . The interpretation of a column vector characterized by integers "1> "2.23).. and explored the various invariance and . and otherwise gives zero.2.) The theory of general quantum fields was first laid out in 1929. (This will be shown in a different way in Chapter 5.111 . one for each normal mode..22) and (1.n.n2... . They also went on to apply this general formalism to the electromagnetic and Dirac fields.2. there can be at most one quantum in each normal mode.. . 1 "" = 1. the a and at matrices take the form a  1_[0 0'] . a: creates a quantum in normal mode k unless there is one there already. in a pair of comprehensive articles by Heisenberg and Pauli.2.41 The starting point of their work was the application of the canonical formalism to the fields themselves. and the commutation relations were detennined from the assumption that the variational derivative of the Lagrangian with respect to any field's time·derivative behaves like a 'momentum' conjugate to that field (except that commutation relations become anticommutation relations for fermion fields)...24) (1.. for a single normal mode we have just two rows and two columns.. 0 The reader may check that (1. the field equations were then determined from the principle that the action I Ldt should be stationary when the fields are varied. of course.' .
For a free complex scalar field ¢(x) the Lagrangian is taken as (1.26) If we subject ¢(x) to an infinitesimal variation b¢(x). in computing the change in the action f Ldt. Vb</>I </>1 b</> _ _(m.2.2 The Birth of Quantum Field Theory 21 conservation laws. and Lorentz and gauge invariance. The 'momenta' canonically conjugate to the fields ¢ and ¢t are given by the variational derivatives of L with respect to .c2V</> .pt.29) (1. momentum. and write But this must vanish for any b¢ and b¢t.c2V<j'! . The HeisenbergPauli formalism is essentially the same as that described in OUT Chapter 7. the Lagrangian is changed by the amount bL ~ J d1x [<i>1 b<i> + <i>b<i> 1. Thus.2.2.p and .28) and its adjoint.30) These field variables satisfy the usual canonical commutation relations. which we can read off from (1. so ¢ must satisfy the familiar relativistic wave equation (1.1.2.2)' (m. (1. including the conservation of charge.2.2)' </>b</>I] . and so for the present we can limit ourselves to a single example which will turn out to be useful later in this section.227) It is assumed in using the principle of stationary action that the variation in the fields should vanish on the boundaries of the spacetime region of integration. .27) as (1. Vb</> . and energy. we can immediately integrate by parts.
.2.31) (1.35) H~ Jd x[n t n+c2(1'4»t'(I'4»+(m:t)4>t4>] 3 (1.t)] ~ [n t (x. Dirac had proposed that the negativeenergy states of the electron were all filled.n(y.t)] ~ [n(x.I). a great deal of work was put into the development of a formalism whose Lorentz invariance would be explicit. his 'hole theory' was used to calculate a number of processes to the lowest order of perturbation theory. and that of the electron and positron. but it had a number of disadvantages: In particular. After Dirac's idea was seemingly confinned by the discovery of the positron in 1932.4>(I. described in terms of a quantized electromagnetic field.32) (1.2. rather than the negativeenergy electrons themselves. [n(x.26)..4>t(I. Vladimir Fock.').)] ~ [n'(x.2. I).4>'(I. minus the Lagrangian: H~Jd3x[n¢+nt¢t]L or. for instance. production of an electronpositron pair is described as the excitation of a negativeenergy electron to a positiveenergy state. We saw in the last section that in 1930.t)] ~o. Not all physicists felt this to be a disadvantage.I).t).4>t(y. and (1.4>t(y.I).2. The most influential effort was the 'manytime' formalism of Dirac. using (1.t)] ~ [4>t(x.I).n'(y.nt(y.36) After the papers by Heisenberg and Pauli one element was still missing before quantum field theory could reach its final prewar form: a solution to the problem of the negativeenergy states.42 in which the state vector was represented by a wave function depending on the spacetime and spin coordinates of all electrons.'). and the annihilation of an electron and positron is described as the corresponding deexcitation.2. (1.30): (1.I). including electronpositron pair production and scattering. but with only the holes in the negativeenergy sea observable. [4>(X.t). at just about the time of the HeisenbergPauli papers. At the same time.29).22 1 Historical Introduction with a delta function in place of a Kronecker delta [n(x.I)] ~ 0.2. positiveenergy and negativeenergy. the total number of electrons of either positive or negative energy is conserved.33) (1. (1.234) The Hamiltonian here is given (just as in particle mechanics) by the 'sum' of all canonical momenta times the timederivatives of the corresponding fields.4>(y. t)] ~O. [n(x.2.I)] ~ [n t (x. and Boris Podolsky.t)] ~ [4>(x.4>(y. This manytime formalism had the advantage of manifest Lorentz invariance.I)] ~ mo'(xy).2. there was a profound difference between the treatment of the photon. In this formalism.
of course.:w"~) = Lnwkakak. (1. so there were doubts about its physical significance. (1. Since electrons carry a charge. so we might try to write the field as w(x) = • LUk(x)eiWklak.2.38) 2 . .2. .Yi' =inert· V + cqmc ~ (1.2. Furry and Oppenheimer picked up Dirac's idea42 that the positron is the absence of a negativeenergy electron.2. The successful calculation by Fermi 43 of the electron energy distribution in beta decay deserves to be counted as one of the early triumphs of quantum field theory.2.23).25). we would not like to mix annihilation and creation operators.2. suppose we try to construct an electron field in analogy with the electromagnetic field or the BornHeisenbergJordan field (1. (1.37) where uk(x)e.37) H = J d 3x ~)t.1. (1. there was a more practical disadvantage of the manytime formalism: it would have been difficult to use it to describe a process like nuclear beta decay.u.41) The trouble is.24) and (1. (1. the anticommutation relations are . The essential idea that was needed to demonstrate the equivalence of the Dirac hole theory with a quantum field theory of the electron was provided by Fock43a and by Wendell Furry and Oppenheimer44 in 19334. (See Eqs. and sign of the energy): Ji"u.40) and ak are the corresponding annihilation operators. .2). so that he did not expect particles and fields to be described in the same terms.2. To appreciate this idea from a more modern standpoint.half the Wk are negative while the operators a1ak take only the positive eigenvalues t and O.2. spin.2 The Birth of' Quantum Field Theory 23 the electron field unlike the electromagnetic field did not have a classical limit.2. "w.22)(1. Dirac42a conceived of fields as the means by which we observe particles.13) (with k now labelling tbe tbreemomentum.39) I utU{ d3x = i5k{. Though I do not know whether it bothered anyone at the time.iwkl are a complele set of orthonormal planewave solutions of the Dirac equation (1.41 the Hamiltonian is formed by calculating the 'expectation value' of Yf with a 'wave function' replaced by the quantized field (1.2.2. in which an electron and antineutrino are created without an accompanying positron or neutrino. that this is not a positive operator .) In order to cure tbis disease.1. satisfying the JordanWigner anticommutation relations (1. . According to the ideas of 'second quantization' or the canonical quantization procedure of Heisenberg and Pauli. Also.
The Dirac field (1.2. k k (1. bk  at (for Wk < 0) (1.2.2. k k (1.45 in a paper written in part to challenge Dirac's picture of filled negativeenergy states. t) ~ ~L yV k q(k.48) . we can rewrite the energy operator (1. Pauli and Weisskopf expanded the free charged scalar field in plane waves in a cube of spatial volume V _ L 3 : 4>(x.45) In order for this redefinition to be more than a mere formality. so it is not possible to interchange the roles of these operators freely.2. using the anticommutation relations for the hs. by Pauli and Weisskopf. If we measure all energies relative to the vacuum energy Eo. (1.2. and Uk(X) _ uk(x)e.iwkl .2.2.2. respectively. Similarly. ')e'" (1.24 1 Historical Introduction symmetric between creation and annihilation operators.46) (1. it is necessary also to specify that the physical vacuum is a state \flo containing no positiveenergy electrons or positrons: tJk\f'o=O bk\f'o=O (Wk>O). (1. and Eq. The problem of negativeenergy states for a charged spin zero particle was also resolved in 1934. as was the case for fermions.2. then the physical energy operator is H Eo.43) where (+) and () indicate sums over normal modes k with Wk > 0 and Wk < 0.44) gives the energy of the vacuum as just Eo. so they defined the positron creation and annihilation operators as the corresponding annihilation and creation operators for negativeenergy electrons hI Ok . k (1.2. Instead we must return to the HeisenbergPauli canonical formalism 41 to decide which coefficients of the various normal modes are creation or annihilation operators.44) where Eo is the infinite cnumber Eo ~  L ()lilwkl . (Wk<O).44) shows that this is a positive operator. Here the creation and annihilation operators satisfy commutation rather than anticommutation relations.47) Hence (1.41) as H ~ L (+lliwkaiak + L Hlilwklbibk + Eo.42) where the label k on b denotes a positiveenergy positron mode with momenta and spin opposite to those of the electron mode k.2.37) may then be written ~(x) ~ L (+lakuk(x) + L (lbiuk(X).
t ) = JV~p _ 1 " . ~ iM'1 (1.q'(I.p(l.34) yield for the qs and ps: [p(k. _ oH _ 2 t p(k.1 p..t). t) ~ ]v J d]x 4>(x. t (1.2.2. (k ) . We see that. and Jordan.2.31)(1. tJe. By inserting (1.l.tJ] ~ [p(k.2.q (k.50) is just equivalent to the KleinGordonSchrodinger wave equation (1.2 The Birth of Quantum Field Theory 25 with the wave numbers restricted by the periodicity condition.2. a result which in the light ofEq. (1.t) ()q .50) The Fourier inversion formula gives q(k..t).57) (and its adjoint).tJ] ~ [q(k.2.2. we can also write this operator in tenns of ps and qs: H ~ L: [p'(k.2.2. +Ib . fd 3 x el"e.t) k + w[q'(k.2.51) (1252) .2.lb .36) for the Hamiltonian. ( te and therefore the canonical commutation relations (1.tI] ~O (1.49) The minus sign is put into the exponent here so that (1.tjV J d' x n x). Heisenberg.q(l. just as in the case of the 1926 model of Born.29) was expanded as ( nX.56) The timederivatives of the ps are then given by the Hamiltonian equation (1.t).t). (1. (k )1" .49) in the formula (1.2.t)q(k.2.tI] ~ [p(k.t).48) and (1.2.2.53) [p(k.te .tJ] ~ in V .3 should be a set of three positive or negative integers.t).4 the free field behaves like an infinite number of coupled .2.29) now becomes: p(k.t)p(k.t)] . that the quantities kjL/2n for j = 1.55) where w[ _ e2k2 + (". (1.t). .28).tI] ~ [q(k.1.54) together with other relations that may be derived from these by taking their Hermitian adjoints.q(l. ( 1. (1.qt(I. Similarly the canonically conjugate variable (1.2.'2) 2.P'(I.tJ ~ q'(k. (k ) w.2.t).
58) (1.60)(1.65) The existence of two different kinds of operators a and b.2.59) p(k.62) H ~ L nWk[bt(k)b(k) + at(k)a(k)] + Eo. Pauli and Weisskopf could construct p and q operators which satisfy the commutation relations (1. using (1. and (1.48) may he written 1>(X. As emphasized by Pauli and Weisskopf. and if charged have opposite charges.2.53).2.2.2.50) and (1.64) where Eo is the infinite cnumber Eo  L nWk . (1. corresponding to particles and antiparticles: q(k.54).2.2.[a(k)exp(ik'xiwk t ) i~ V k 2Wk bt(k)exp(ik' x + iWktl] and the Hamiltonian (1. Thus. (1.2. which appear in precisely the same way in the Hamiltonian.57).2.2. these two varieties can be identified as particles and the corresponding antiparticles. I) ~ iV 2:k [a(k) exp( iwktl ~ Vn.I)~ "'Ij ruL .61) ~ [at(k). k (1. by introducing annihilation and creation operators a.k [b(k)exP(iwkl) + bt (k) exp(iwktl] at(k) eXP(+iwkl)] (1. .54) and the 'equations of motion' (1. k (1.60) (1.b(l)] (1. shows that this is a theory with two kinds of particles with the same mass.50).2. (1. bt of two different kinds. at.bt(l)] ~ Ski.2.2. [a(k).26 1 Historical Introduction harmonic oscillators.63) H ~ L ! nWk [ht(k)b(k) + b(k)bt(k) + at(k)a(k) + a(k)at(k)] k 2 or.at(l)] ~ [b(k).53)(1.55) takes the form (1.b(I)] ~ [a(k).2.2.bt(I)] ~ [at(k).bt(l)] ~ O.2.2.l) where [a(k).2.a(I)] ~ [h(k).b(I)] ~O. [a(k). The field (1.57).2.62) It is straightforward to check that these operators do satisfy the desired relations (1. b.2.
and which satisfies a conservation law of the form (1. which can be formed from solutions of the KleinGordonSchrodinger free scalar wave equation (1. if it would give zero when applied to the vacuum state.Eo.2. the only probability density p.10).2.63).64) and (1. bosons of spin zero as well as fermions of spin 1/2 can have distinct antiparticles.1..28). we can write (1. as it also is in the FurryOppenheimer fonnalism 44 for spin 1/2. in the 'secondquantized' theory. (1.64) shows that this again is positive. Thus. must be proportional to the quantity p ~ 21m [¢' <7¢] Dt (1.at(k)] '1'0) ~ +1 (1. p is not a positive operator. What about the problem that served Dirac as a starting point. and therefore a(k)'I'o ~ b(k)'I'o ~ O. where ¢ is given by Eq.68) in various forms. We now can tell whether a and h or at and bt are the annihilation operators by taking the expectation values of commutation relations in were an annihilation operator the vacuum state '1'0. so the vacuum expectation value of (1.2. . In this way we can conclude that it is ak and bk that are the annihilation operators.2.2.] . If we measure all energies relative to Eo.66) in conflict with the requirement that the lefthand side must be negativedefinite.1.2. [a(k).2. For instance.269) The spaceintegral of this operator is then easily calculated to be N J aJx ~ L P k (1.2.2. and (1.67) now tell us that Eo is the energy of the vacuum state.t _ o¢' .67) This is consistent with all commutation relations.60) would give a: I Ia(k)'I'o II' ~ ('1'0. which differ by infinite cnumbers. then the physical energy operator is H .58) to be a creation operator.68) and therefore is not necessarily a positive quantity.70) and clearly has eigenvalues of either sign. the canonical formalism forces the coefficient of the e+ iw1 in the field (1.2.2 The Birth of Quantum Field Theory 27 as we stressed above. the problem of negative probabilities? As Dirac had recognized.2. Similarly.2. (1. it proves convenient to write it as P ~~ Ii [D¢ . ilt '+' at '+' (at(k)a(k) . which for bosons cannot be identified as holes in a sea of negative energy particles. Equations (1..bt(klb(k)) (1. Since ¢t(x) does not commute with ¢(x) here.
~ L'+Jat(kja(kl+ LHb(kjbt(kj. and this now has both negative and positive eigenvalues. For charged particles. The solution to this problem provided by quantum field theory is that neither the 1j.(1. in a sense this problem appears in quantum field theory for spin 1/2 as well as spin zero. so the minus signs in (1. (1. k k (1.2. If (fin are a complete orthonormal set of such states. (1. so Furry and Oppenheimer reasoned that the physical number operator is N . which would have to define conserved positive probability densities. In particular. the number of particles minus the number of antiparticles.47). Instead.70) and. are not probability amplitudes at all. The density operator tpt1p of Dirac is indeed a positive operator. then a measurement of particle numbers in an arbitrary state If' will yield a probability for finding the system in state ¢n.2. the physical Hilbert space is sp<mned by states defined as containing definite numbers of particles and/or antiparticles in each mode.2. The wave fields 1>.2.46) and (1.28 1 Historical Introduction However. 1/J. given by P. using the planewave decomposition (1.72) According to Eqs. In particular.71) where No is the infinite constant No ~ L HI. the conservation of charge forces the charge operators to be proportional to these number operators. .2. \fI) is the usual Hilbert space scalar product. no question as to the possibility of negative probabilities will arise for any spin. but in order to construct a physical density we ought to subtract the contribution of the filled electron states. but as number operators: specifically. Hence. ~ 1(<1>" 'l')I' . k k The anticommutation relations for the bs allow us to rewrite this as N No = L(+latak .73) where (¢n.LHb:bk.No.No of Eqs. (1. interactions . just as for spin zero. In this fieldtheoretic formalism. etc.71) are not to be interpreted as total probabilities.71) allow us immediately to conclude that particles and antiparticles have opposite charge. No is the number of particles in the vacuum.2.70) and (1.t".2.43). It would be a good thing if the misleading expression 'second quantization' were permanently retired. k (1.2.2. we may write the total number operator as N [dJx ~.J of Furry and Oppenheimer nor the 1> of Pauli and Weisskopf are probability amplitudes. the operators Nand N . but operators which create or destroy particles in the various normal modes.2.
not protons and electrons. and. This problem (and others) were solved in 1932 with the discovery of the neutron. and various combinations of fieldtheoretic and holetheoretic ideas were used in calculations of physical reaction rates. Nevertheless. of course. in order to have atomic number 7 and atomic weight 14.1. and worked out in detail there for the case of electronphoton scattering. A few years later. positron. Despite its apparent advantages. would have to be composed of 14 protons and 7 electrons. in reasonable agreement with the experimental data. several authors54<1 speculated that nuclear forces might be explained in this theory as due to the exchange of electrons and neutrinos. The conceptual framework described in the above brief remarks will serve as the basis for much of the work in this book.46 e+ + e~ 1' 2}' in 1930 by Dirac.+y+Z and y+Z 1' e+ +e~ +Z (where Z denotes the Coulomb field of a heavy atom) in 1934 by Bethe and Heitler.49 and e+ + e~ 1' e+ + e~ in 1936 by Bhabha.48 e. but it was hard to see how a light particle like the electron could be confined in the nucleus. Throughout the 1920s it was generally believed that heavier nuclei are composed of protons and electrons. N 14 . the nucleus of hydrogen. in conflict with the result of molecular spectroscopy52 that N 14 is a boson. We have already mentioned the electron. Another severe difficulty with this picture was pointed out in 1931 by Ehrenfest and Oppenheimer: 51 the nucleus of ordinary nitrogen. the proton. and the rates of various processes are given by using these interaction operators in a time·dependent perturbation theory. in 1935. photon. After the success of the Fermi theory of beta decay. 5o (Rules for the calculation of such processes are given in Chapter 8. This period saw a number of calculations of cross sections to lowest order in powers of e2 for various processes. . or higher order in field variables. quantum field theory did not imme· diately supplant hole theory. It was clear that a strong nonelectromagnetic force of short range would have to operate between neutrons and protons to hold nuclei together. neutrino.53 and by Heisenberg's subsequent suggestion54 that nuclei are composed of protons and neutrons. One of the reasons for this was the apparent failure of quantum electrodynamics to account for the penetrating power of the charged particles in cosmic ray showers. fourth. such as e~ +}' 1' e~ + "I in 1929 by Klein and Nishina.) These lowest·order calculations gave finite results. a general feeling of dissatisfaction with quantum field theory (whether or not in the form of hole theory) persisted throughout the 1930s.2 The Birth of Quantum Field Theory 29 contribute terms to the Hamiltonian which are of third. the two points of view coexisted for a while. rather.47 e~ + e~ 1' e~ +e~ in 1932 by M~ner.+Z 1' e. and would therefore have to be a fermion. 50a Another cause of dissatisfaction that turned out to be related to the first was the steady discovery of new kinds of particles and interactions. noted in 1936 by Oppenheimer and Franklin Carlson.
At the same time. and when this equation was quantized. the lighter (now called the 11 meson. Ettore Pancini. The observed range of the strong interactions within nuclei led Yukawa to estimate that li./('. Occhialini. In the same year.30 1 Historical Introduction Hideki Yukawa proposed a quite different quantum field theory of the nuclear force. In 1937 such 'mesons' were discovered in cloud chamber experiments 56 by Seth Neddermeyer and Anderson and by Jabez Curry Street and Edward Carl Stevenson. was introduced as a parameter in Yukawa's scalar field equation.. with a dependence on the nucleon separation r given by V(r):x ~ r exp(Ar) (1. and predominates in cosmic rays at sea level.58 and its subsequent experimental confirmation 59 by Cesare Lattes. and it was generally believed that these were the hypothesized particles of Yukawa. The quantity A. but to pursue this story would take us outside the bounds of our present survey. Lothar Nordheim 56a pointed out in 1939 that the same strong interactions by which the mesons are copiously produced at high altitudes (and which are required in Yukawa's theory) should have led to the mesons' absorption in the atmosphere. These discoveries showed clearly that any conceptual framework which was limited to photons. he found that the interaction of a scalar field with nucleons (protons or neutrons) would produce a nucleonnucleon potential. and thus cleared up the problem with these showers that had bothered Oppenheimer and Carlson. it created new difficulties. electrons. being produced by the decay of 11: mesons. 1947. and Cecil Powell . and positrons would be far too narrow to be taken seriously as . In 1947 it was shown in an experiment by Marcello Conversi. This puzzle was cleared up by a theoretical suggestion. entirely new kinds of particles (now known as K mesons and hyperons) were found in cosmic rays by George Rochester and Clifford Butler."/c is of the order of 200 electron masses. 60 From 1947 until the present particles have continued to be discovered in a bewildering variety. The discovery of mesons revealed that the charged particles in cosmic ray showers are not all electrons.2. a result contradicted by their copious appearance at lower altitudes.there are two kinds of mesons with slightly different masses: the heavier (now called the 11: meson or pion) has strong interactions and plays the role in nuclear force envisaged by Yukawa. however. Yukawa found that it described particles of mass 1'11. and therefore could not be identified with Yukawa's particle. or muon) has only weak and electromagnetic interactions. 55 In an essentially classical calculation. and Oreste Piccioni57 that the mesons which predominate in cosmic rays at low altitude actually interact weakly with nucleons.74) instead of the l/r Coulomb potential produced by electric fields.
These calculations treated the electron according to the rules of the original Dirac theory. But an even more important obstacle was presented by a purely theoretical problem .41 Soon after. The problem of infinities in quantum field theory was apparently first noted in the 192930 papers of Heisenberg and Pauli. A few years later Weisskopf repeated the calculation of the electron selfmass in the new hole theory. with an intermediate state consisting of an electron and a photon: for instance. Disappointingly this problem appeared with even greater severity in the early days of quantum field theory. because they arise from intermediate states containing particles of very short wavelength.AIH'ln > 1 mJ 2 _ En ~ Em  Iklc (1. In this case another term appears in secondorder perturbation theory. and therefore blows up for a _ o.3. with all negativeenergy states full. if this infinity is removed by discarding all intermediate states with photon wave numbers greater than l/a. and although greatly ameliorated by subsequent improvements in the theory.61 and of a free electron by Ivar Waller.3 The Problem of Infinities 31 a fundamental theory.. without filled negativeelectron states.3 The Problem of Infinities Quantum field theory deals with fields Ip(X) that destroy and create particles at a spacetime point x. then the selfenergy behaves like l/a 2 as a _ o.. the shift of the energy En of an electron in the nth energy level of hydrogen is given by M. ~ LJd'k 1< m. this mass is e2 /6rrac2 for a surface distribution of charge with radius a.1. the presence of infinities was confirmed in calculations of the electromagnetic selfenergy of a bound electron by Oppenheimer. and H' is the term in the Hamiltonian representing the interaction of radiation and electrons.1. Earlier experience with classical electron theory provided a warning that a point electron will have infinite electromagnetic selfmass. and photon momenta k. 1. which in a nonholetheory language can be described as arising from processes in which the electron in its final state first appears out of the vacuum together with a photon and a positron which then annihilate along with the initial . 62 They used ordinary secondorder perturbation theory. photon helicities . Infinities of this sort are often called ultraviolet divergences. This calculation gave a selfenergy that is formally infinite. further. it remains with us to the present day.1) where the sums and integral are over all intermediate electron states m.the problem of infinities.k.
Nicholas Kemmer and Weisskopfl8 presented arguments that in this case the infinities are spurious. The same calculation was being carried out (at the suggestion of Bohr) at that time by Carlson and Furry. and Heisenberg65 showed in 19356 that these infinities could be eliminated by using a moreOfless arbitrary prescription suggested earlier by Dirac66 and Heisenberg67 . A is logarithmically divergent. me (1. . Infinities also seemed to occur in a related problem.e. Initially Weisskopf found a 1/02 dependence on the photon wavenumber cutoff l/a. 64 He considered the effect of an external static nearly uniform charge density e(x) on the vacuum. The Coulomb interaction between e(x) and the charge density of the negativeenergy electrons produces a 'vacuum polarization. (1.. + B(~)'v2.. was mildly encouraging at the time and turned out to be of great importance later.3.. on the negative~energy electrons in the filled energy levels of hole theory.2) 1t (_n_) . .4) can be derived without any subtraction prescrip~ tion. On the other hand. of order Cllna. Bernard Kockel. After seeing Weisskopfs results. They calculated an effective Lagrangian density for the nonlinear electrody~ namic effects produced by virtual electronpositron pairs: !i'~ 1 2 (E _B ) + 2 2 e41i 360n 2m1e7 [(E _B +7(E. Furry realized that while Weisskopf had included an electrostatic tenn that he and Carlson had neglected.3. Weisskopf found that the 1/a2 terms in the total mass shift cancelled! However.3. An infinity of quite a different kind was encountered in 1933.' with induced charge density 0.3. an infinity remained: with a wavenumber cutoff 1/ a. to In a as compared with the classical l/a or the early quantum l/a 2 . despite this cancellation. in the development of renormalization theory. One bright spot in the struggle with infinities was the successful treat . apparently first by Dirac. Soon after. the scattering of light by light. After hearing from Furry and correcting his own error.B)] +. i. the selfmass was found to be63 3 m. Hans Euler. (1. and that Eq.4) 2 2 2 ) 1 valid for frequencies v <t:: rnec1/h. and of order Cl..3) The constant B is finite. where 1/0 is the wavenumber cutoff.32 1 Historical Introduction electron.. mea The weakening of the cutoff dependence. ~ A. +. Weisskopf had made a new mistake in the calculation of the magnetic selfenergy.m ~ 2• m In (1.
whose elements are the amplitudes for various scattering processes.' Thus in 1938 Heisenberg70 proposed the existence of a fundamental length L. deriving electromagnetic interactions in terms of an interaction at a distance. such as the Smatrix introduced by John Archibald Wheeler71 in 1937 and Heisenberg72 in 1943. This will be discussed in modern terms in Chapter 13. these various infinities were seen not merely as failures of specific calculations. One of the symptoms of this uneasy pessimism was the continued exploration throughout the 1930s and 1940s of alternative formalisms. In 1937 it was shown by Felix Bloch and Arne Nordsieck 6811 that these infinities cancel provided one includes processes in which arbitrary numbers of lowenergy photons are produced. Several specific proposals70cl were made for giving field theory a nonlocal structure. Wheeler and Richard Feynman 74 in 1945 attempted to eliminate the electromagnetic field. and for some theorists a pure Smatrix theory became an ideal. but this was not realized until later. but also between these charges and all the other charges in the universe. those that arise from the lowenergy rather than the highenergy part of the range of integration. they seemed to indicate a gap in the understanding of relativistic quantum field theory on the most fundamental level. Yet another infinity turned up in a calculation by Sidney Michael DancoW9 in 1939 of the radiative corrections to the scattering of electrons by the static Coulomb field of an atom. especially as a possible solution to the problems of the strong interactions. an opinion reinforced by the problems with cosmic rays mentioned in the previous section. 'The preoccupation of the majority of involved physicists was not with analyzing and carefully applying the known relativistic theory of coupled electron and electromagnetic fields but with changing it. or momenta hlL. Perhaps the most radical modification of quantum mechanics suggested during this period was the introduction by Dirac75 of states of negative probability.3 The Problem of Infinities 33 ment of infrared divergences. Field theory was supposed to work only for distances larger than L. As we shall see. Some theorists began to suspect that the formalism of statevectors and quantum fields should be replaced by one based solely on observable quantities. analogous to the fundamental action hand fundamental velocity c.1. The calculation contained a mistake (one of the terms was omitted). As Julian Schwinger69h later recalled. . so that all divergent integrals would effectively be cut off at distances L. They were able to show that a pure retarded (or pure advanced) potential could be obtained by taking into account the interaction not only between source and test charges. Rather. the concept of the Smatrix has now become a vital part of modern quantum field theory.6911 Throughout the 1930s. 73 In yet another direction.
hence the infinities in these quantities can be cancelled by a negative infinity in the 'bare' nonelectromagnetic mass of the electron. it was already known that in any Lorentzinvariant classical theory the electromagnetic selfenergy and selfmomentum of an electron must take the form of corrections to the mass of the electron.3. Perhaps these infinities could all be absorbed into a redefinition. and has therefore always been ignored. (1. the Conference on the Foundations of . the 1928 Dirac theory had predicted complete degeneracy of the 2s 1/ Z2Pl/2 levels of hydrogen to all orders in IX. has also flourished in quantum field theory. as we shall see in Chapter 14. whether all infinities in quantum field theory could be dealt with in this way. to 'TOTAL ~ J*+ d 3 Se) ~ (1 + A)e. any attempt at a quantum electromagnetic calculation of the splitting of these two levels ran into the problem of the infinite selfenergy of a bound electron. (1J. a 'renormalization' of the parameters of the theory.3) shows that the vacuum polarization changes the charge of the electron. this contribution to the splitting is much smaller than 1000 MHz. though not in the form originally suggested. Also. Later Bethe80 recalled that 'This shift comes out infinite in all existing theories.34 1 Historical Introduction as a means of cancelling infinities in sums over states. and verified that known infinities could be eliminated by renormalization of physical parameters in a variety of sample calculations. A more conservative idea for dealing with the infinities was also in the air during the 19305. This idea. In 1936 Weisskopf 76 suggested that this is the case. One notable exception was Edwin Albrecht Uehling. of an 'indefinite metric' in Hilbert space. from e . leaving a finite measurable 'renormalized' mass.J d3X £.' This attitude persisted even in the late 1930s. and of the wrong sign. On June 14. unfortunately. The question was. In particular. it was impossible with the calculational techniques then available to show that infinities could always be eliminated in this way. and Dancoff's calculation69 seemed to show that they could not.78 who realized that the vacuum polarization effect mentioned earlier would produce a 2s 1/z2Pl/Z splitting. Another effect of the appearance of infinities was a tendency to believe that any effect which turned out to be infinite in quantum field theory was actually not there at all. However. 1947. Eq. For instance.5) Vacuum polarization gives finite results in lowest order if observables like scattering crosssections are expressed in terms of eTOTAL rather than e. when spectroscopic experiments 77 began to indicate the presence of a 2sljz2Pl/2 splitting of order 1000 MHz. The gloom surrounding quantum field theory began to lift soon after World War II. therefore the existence of such a splitting was generally not taken seriously.
many of whom had already been working on improved formalisms for calculation in quantum electrodynamics.a few experimental physicists.79a which showed that the difficulties associated with the divergence of the selfenergy in the limit of zero radius do not appear explicitly if the theory is reexpressed so that the mass parameter in the formalism is identified with the experimental electron mass. 77 The impact of this discovery can be summarized in a saying that was current in Copenhagen when I was a graduate student there in 1954: 'Just because something is infinite does not mean it is zero!' The discovery of the Lamb shift aroused intense interest among the theorists at Shelter Island. The atoms in 2p states can decay very rapidly to the Is ground state by onephoton (Lyman cr:) emission. described a decisive measurement 79 of the 2s 1jr2plj2 shift in hydrogen. with a fixed frequency r '"'"' 10 GHz. The beam was passed through a magnetic field. The total (Zeeman plus intrinsic) 2s2p splitting at this value of the magnetic field strength would have to be just hv. so in effect the detector was measuring the number of atoms in the metastable 2s state. One of the experimentalists (or rather theorist turned experimentalist). many in 2s and 2p states.) Almost im . and Weisskopf. in agreement with the earlier spectroscopic measurements. Weisskopf had already assigned this problem to a graduate student. and . A beam of hydrogen atoms from an oven. was aimed at a detector sensitive only to atoms in excited states. A preliminary value of 1000 MHz was announced. At a certain magnetic field strength the detector signal was observed to be quenched. and discussed the matter on the trip to Shelter Island) suggested that since the inclusion of intermediate states involving positrons was known to reduce the divergence in energy level shifts from l/a 2 to Ina. from which the intrinsic splitting could be inferred. Bruce French. The beam was also exposed to a microwavefrequency electromagnetic field. which added a known Zeeman splitting to any 2S 1/ 22plj2 splitting naturally present.of crucial importance . Kramers described his work on mass renormalization in the classical electrodynamics of an extended electron. (In fact. while the 2s states decay only very slowly by twophoton emission. before he learned of Lamb's experiment. perhaps the dftJerences of the shifts in atomic energy levels might turn out to be finite when these intennediate states were taken into account. in 1946. Oppenheimer.1. indicating that the microwave field was producing resonant transitions from the metastable 2s state to the 2p state and thence by a rapid Lyman cr: emission to the ground state. Willis Lamb.3 The Problem of Infinities 35 Quantum Mechanics at Shelter Island. The discussion leaders were Hans Kramers. NY brought theoretical physicists who had been working on the problems of quantum field theory through the 1930s together with a younger generation of theorists who had started scientific work during the war. Schwinger and Weisskopf (who had already heard rumors of Lamb's result.
Fully relativistic calculations using the renormalization idea to eliminate infinities were soon thereafter carried out by a number of other authors. and subsequent measurements of the gyromagnetic ratios in sodium and gallium had given a precise value 83 eli 2 [1. they learned about Lamb's experiment from an article in Newsweek. Hans Bethe80 carried out a nOllRrelativistic calculation. One approach developed by Schwinger85 was based on operator methods and the action principle.87 and described briefly by him at the Pocono Conference. In 1947 he and his group were still out of the loop of scientific communication. He obtained the encouraging approximate value of 1040 MHz. Another exciting experimental result was reported at Shelter Island by Isidor I. at last convinced physicists of the reality of radiative corrections. both Breit and Schwinger described their efforts to calculate this correction.00118 + 0.00003] . W . the successor to the Shelter Island Conference. This. and was presented by him at a conference at Pocono Manor in 1948.81 with excellent agreement with experiment. together with Bethe's calculation of the Lamb shift. Another Lorentzinvariant operator formalism had been developed earlier by Sinltiro Tomonaga 86 and his coworkers in Japan. during a train ride to Schenectady. Measurements in his laboratory of the hyperfine structure of hydrogen and deuterium had suggested 82 that the magnetic moment of the electron is larger than the Dirac value en/2mc by a factor of about 1. An apparently quite different approach was invented by Feynman. Rabi. Instead of introducing quantum field operators.36 1 Historical Introduction mediately after the conference. but using a simple cutoff at virtual photon momenta of order mec2 to eliminate infinities. still without including the effects of intermediate states containing positrons. The mathematical methods used in this period presented a bewildering variety of concepts and formalisms. At Shelter Island. but their work was not at first known in the West. me Learning of these results. Feynman represented the Smatrix as where is the action integral for a a functional integral of exp (iW). Shortly after the conference Schwinger completed a sUlXessful calculation of the anomalous magnetic moment of the electron84 ~ ~ eli ~ ~ 2 [1 me "] + 2 11: eli ~ 2 me [1.0011621 in excellent agreement with observation.0013. Gregory Breit suggested 83a that they arose from an order Cl radiative correction to the electron magnetic moment. Tomonaga had grappled with infinities in ¥ukawa's meson theory in the 1930s.
Oppenheimer61 in 1930 had already noticed that most of the ultraviolet divergence in the selfenergy of a bound electron cancels when one takes the difference between the shifts of two atomic energy levels. and Weisskopfl3 in 1934 had found that most of the divergence in the selfenergy of a free electron cancels when one includes intermediate states containing positrons. an interaction like the Pauli term (1. One result of great practical importance that came out of Feynman's work was a set of graphical rules for calculating S~matrix elements to any desired order of perturbation theory. Freeman Dyson 88 showed that the operator formalisms of Schwinger and Tomonaga would yield the same graphical rules that had been found by Feynman. Unlike the old perturbation theory of the 1920s and 1930s. . I cannot leave the infinities without taking up a puzzling aspect of this story. these Feynman rules automatically lumped together particle creation and antiparticle annihilation processes.1. there was at last a general and systematic formalism that physicists could easily learn to use.3. that the nature of the infinities becomes transparent. in a pair of papers in 1949. this guess would have heen wrong. that varies from one level to another. integrated over all Dirac particle trajectories satisfying certain initial and final conditions for t + +0Cl. and thereby gave results that were Lorentzinvariant at every stage. in the sense that all infinities can be absorbed into a redefinition of a finite number of coupling constants and masses.1. One of the most striking results that could be inferred from Dyson's analysis was a criterion for deciding which quantum field theories are 'renormalizable'. Dyson also carried out an analysis of the infinities in general Feynman diagrams. would spoil the renormalizability of quantum electrodynamics. and outlined a proof that these are always precisely the sort which could be removed by renormalization. but also through a change in the electron kinetic energy. In particular. Finally. including particles and antiparticles on the same footing. As discussed in Section 14.32). that it is only in such calculations. It would have been natural even in 1934 to guess that including positron intermediate states and subtracting the energy shifts of pairs of atomic states would eliminate the ultraviolet divergence in their relative energy shift: There was even experimental evidence 77 for • In fact. With the publication of Dyson's papers. We have already seen in Weisskopfs early calculation63 of the electron selfenergy. which would have changed the predicted magnetic moment of the electron. and that would provide a common language for the subsequent applications of quantum field theory to the problems of physics.3 The Problem of Infinities 37 set of Dirac particles interacting with a classical electromagnetic field. which is the same in all atomic energy levels. radiative corrections to the electron mass affect atomic energy levels not only through a shift in the electron rest energy.
The one missing element was confidence in renormalization as a means of dealing with infinities. But it had become accepted wisdom in the 1930s. Schwinger. Several things happened at Shelter Island to change this expectation. One was news that the problems concerning cosmic rays discussed in the previous section were beginning to be resolved. More important was the fact that now there were reliable experimental values for the Lamb shift and the anomalous magnetic moment that forced physicists to think carefully about radiative corrections. As long as one keeps all tenns up to a given order. In fact. it was made by physicists who though mostly young were playing a conservative role. the first precise calculations81 of the Lamb shift in the USA by French and Weisskopf and Norman Kroll and Lamb were done in just this way. which has only a tiny effect on hydrogen energy levels. and the pions responsible for nuclear forces. and that the solution to its problems could be found only in really adventurous new ideas. there was one such attempt88a in 1939. Probably equally important was the fact that the conference brought together theorists who had in their own individual ways been thinking about renormalization as a solution to the problem of infinities.89 that quantum electrodynamics could not be taken seriously at energies of more than about 100 MeV. after Bethe's work. Robert Marshak presented the hypothesis58 that there were two types of 'meson' with similar masses. . When the revolution came in the late 1940s. The calculation gave a result in rough agreement with the early experiments. though Tomonoga's group 81 in Japan was already using covariant methods to solve this and other problems. 77 This was a mistake. as shown in 1939 by Lamb. using the old nonrelativistic perturbation theory. and a point of view especially urged by Oppenheimer. 88b A fully relativistic calculation of the Lamb shift including positrons in intermediate states could have been attempted during the 19308. but it focused on the wrong part of the problem. the charge radius of the proton. renormalization was widely discussed in the late 1930s. So why did no one before 1947 attempt an numerical estimate of this energy difference? Strictly speaking. the muons that had actually been observed. As we have seen.• 38 1 Historical Introduction a 2s1jr2plj2 energy difference of order 1000 MHz. and Tomonaga. turning away from the search by their predecessors for a radical solution. oldfashioned nonrelativistic perturbation theory gives the same results as the manifestly relativistic formalisms of Feynman.
Cao and S. 58. Beyer. 1949)..2.' Centauros 25. 1.] o L. 'Quantum Theory of Radiation. and Topology II. 87 (1932). o M. Inward Bound (Clarendon Press. Foundations of Nuclear Physics (Dover Publications.1.] o A. 33 (1993). [Sections 1.] o G.] .A. Oxford. 1972). 37. M. Hoddeson. S. [Section 1. Jammer.] o S. 'Development of the Renormalization Theory in Quantum Electrodynamics.' Historia Scientiarum 36. eds. Mod. Amsterdam. New York. [Section 1. 1. Mehra.3. 1994). Cambridge. Brown.' in Aspects of Quantum Theory. 1.] o R. Dirac's Scientific Work from 1924 to 1933. 1.] o J. [Sections 1. Schweber. [Section 1. [Section 1. S. A. 1984). New York. New York. 4. ibid. Dirac. 1. New York. I.] o S.' Rev. Mod. Aramaki.3.2 and 1. [Section 1.1. Wigner. ed.3..2.3.M. Cambridge. T. 1. Salam and E.1.l. S. 1966). UK.. The Birth of Particle Physics (Cambridge University Press. Fermi.2. Miller. Cambridge. 1971).1. [Section 1. [Section 1.A Source Book (Cambridge University Press. 'Yukawa's Prediction of the Meson. 1.] o T. Pais. DeWitt and R. Rev.] (Doubleday and Co. Brown and L.' in Relativity.3. Early Quantum Electrodynamics . ed. Schweber. The Conceptual Development of Quantum Mechanics (McGrawHili Book Co. The Development of Quantum Theory (Gordon and Breach Science Publishers. Groups. 97 (1989).] o L. ed.] o E.3. [Section 1. by B. 'The Golden Age of Theoretical Physics: P. by A. (Cambridge University Press.' Synthese 97. [Sections 1.3. 'The Conceptual Foundations and the Philosophical Aspects of Renormalization Theory.1.3.Bibliography Bibliography 39 o S. Inc.] o A. Gamow.1. [Sections 1. 91 (1989). S. M. 1. 71 (1981). Stora (NorthHolland. P. 1986).2. Phvs. 449 (1986). 1966).2. [Sections 1. 1983). Schweber. 'Some Chapters for a History of Quantum Field Theory: 19381952.. 'Feynman and the Visualization of SpaceTime Processes. Thirty Years that Shook Physics Garden City. [Section 1.] o P.. Phys.2. Y.1.
Ludwig.2 and 1.] o S. New York. Annales de Physique 3.] o V. Brown and L. M. Wentzel.' 1979 Bernard Gregory Lecture at CERN. 1. S.40 1 Historical Introduction o S. New York. 3. Z. L. AI09.. and P. W. 540 (1923). New York. Witten (MIT Press. 1973).] o S. [Sections 1. 2. M. 711 (1925). Schweber.2. 34. Tomonaga.1.3.1. Cambridge. A. cit. 1..3. Z. 'Quantum Theory of Fields (Until 1947)' in Theoretical Physics in the Twen. and E. S. QED and the Men Who Made It: Dyson. 4. C. de Broglie. Phil. in The Physicist's Conception of Nature (Reidel. Fa111977. 336 (1926). Jackiw. 1985). 1973).·I. [Sections 1. Born and P. Nature 112. 1960). 1. 1. Born. 22 (1925) [reprinted in English in Wave Mechan.2 and 1. ed. 642 (1925).] o E.1. Rev. L.2. Paris. 13. Phys. 36.ics. 47. 1924). 879 (1925). A. by M..2 and 1. Naturwiss. 'Growing Up with Field Theory: The Development of Quantum Electrodynamics in Half a Century. 35. MA. f Phys. [Section 1. Comptes Rendus 177. Weisskopf. W.446 (1924).] o J.3. 1. Dirac. 1994). Whittaker.3.1 o s. Hoddeson. H. Mag. These papers are reprinted in Sources of Quantum Mechanics. Pauli.. Phys. 507.. 557 (1925). op. These de doctorat (Masson et Cie. 1. 33. S. 'A Short History of Shelter Island I. [Section 1. M. 705 (1927). 630 (1923). New York.3. Jordan. 30. ed. Z.] o G. published in L. by R. . w. 1958). Schwinger. 1968)].3. Phys. Weisskopf (Interscience Publishers Inc.2. Roy. van der Waerden (Dover Publications. Davisson and L. Heisenberg. W.' Daedalus. ed. by B.1. Soc. Weinberg. 548. Schweber. F. Proe. ed. [Section 1.1. P. by G. f Phys. [Sections 1.] References 1. Jordan. F. Dordrecht. ed.3. 1968).' in Shelter Island II. f. Fierz and V. New York. [Sections 1. 858 (1925). Princeton. and Tomonaga (Princeton University Press. Elsasser. 'The Search for Unity: Notes for a History of Quantum Field Theory. [Sections 1. A History of the Theories oj Aether and Electricity (Humanities Press.tieth Century. Schwinger. Inc. Heisenberg. Germer. (Pergamon Press. Feynman. Weinberg. Selected Papers in Quantum Electrodynamics (Dover Publications Inc.
E. Mag. as in Ref. 145 (1921). 112 (1923). A1l6. Paschen. 3rd edn.953 (1925). Z..j Phys. and his article in Scientific American 208. E. Inst. The fine structure of spectral lines was first discovered interferometrically by A. Deans (Blackie and Son. Sehrodinger. Compton. M. London. e. . P. 425. 227 (1927). 1971). while Dirac quotes only Darwin. Also see Dirac's obituary of Schrodinger. see. Z. L. J. 15. 355 (1961).! Phys. 7. in Wave Mechanics.g. 51. Z.. 9. Quantum Mechanics. At the time. C. 1. ibid. 40. 81. These experiments were actually carried out using He+ because its fine structure splitting is 16 times larger than for hydrogen. A. pp. 242 (1926). 221 (1920). 34. A. For the details of the calculation. lOa. The Development of Quantum Theory (Gordon and Breach. Darwin says that several authors did this work at about the same time. 11. 38. ibid. Z. 895 (1926). 189 (1923). O. 80. 125 (1916).References 41 5.. 338 (1891). 7. M. Naturwiss. Ref. 1968): Section 51. 231 (1921). Jordan. It was only later that the extra angular momentum was recognized. 63. 37. Dirac. Ann. I. Schearer and W. unfortunately in a somewhat abridged form. 263 (1926). trans. Wilson. Goudsmit. 45 (1963). f Phys. Also see Collected Papers on Wave Mechanics. Schiff. Nature 117. 13. The general formula for Zeeman splitting in oneelectron atoms had been discovered empirically by A. 226 (1926).. 901 (1916). SO. 11. Ann. Nature 189. Ann. 264 (1926). G. 117 (1926). New York. 280 (1892).j Phys. H. 5. 13. Inc. the extra nonorbital angular momentum appearing in this formula was thought to be the angular momentum of the atomic 'core. 398 (1921). 70.. Mag. Ann. Lande. Also see V Foek. Uhlenbeck and S. 109 (1926). e.. Soc. Phys. See. Phil. 29. 489. Roy.g. Phys. 6. 1928). 12. F. Darwin. F. Also see W. 192. Miinchner Berichte 1915. to be due to the spin of the electron. These papers are reprinted in English. Frank. Proc. 1. by J. Sommerfeld. 795 (1915). 10. f Phys. 19. The electron spin had been earlier suggested for other reasons by A. ibid. Klein. 79. (McGrawHill. 37. Phil. Z. W. 31. 361. Heisenberg and P. 8. W. Phys. 437.. Gordoo. ibid.429. G. New York. ibid. Phys. Michelson. 32 (1923). ibid.' A Sommerfeld. 39. A.
411 (1926). Soc. Ref. 514 (1926). 125. Fermi. W. 1928). 19. 590 (1926). 654 (1928). 43. 1972). Dirac.42 1 Historical Introduction 14. 37. Phys. ed. Z. G. 16. 34. G. R. Gamow. Paris. (. 38. 719 (1924). 47. 621 (1928).1969. Z. 20. (Doubleday and Co. Z. unpublished. 803 (1926) (reprinted in an abridged English version in Wave Mechanics. Salam and E.. 48. C. M. A. Dirac.f Phys.f Phys. Bohr. 22. P.f Phys. 48. P. Mag. A. P. A. 499 (1926). 1989. ibid. AlIS. Z. Gordon. 11 (1928). (Wiley. Pauli. see M. C. Phil. also see Ref. A120. Gravitation and Cosmology. 43. A126. 17. W. 23. Accad. Rend. Weinberg. 21. f Phys. NY. J. 15. f Phys. 25. N. Phys.. P. 40. Proc. Electrons et Photons . 26. ibid. Soc. Mehra. Thomas. G. Thirty Years that Shook Physics. Stoner. E. M. 39. Z. March 28. Proc. W. 37. A. A112. Slater. Wigner (Cambridge University Press. 1966): p. Soc.Rapports et Discussions du V e Conseil de Physique Solvay (GauthierVillars. 580 (1928). Also see Dirac. Born Z. 863 (1926). 263 (1926). 601 (1927). Also see S. ibid. 765 (1925). Darwin. M. L. 1972): Section 5. M. Crane Lecture at the University of Michigan. Proc. f Phys. f (1926). Z. Rev.f Phys. 610 (1928). 145 27. 351 (1928). 31. Conversation between Dirac and 1. Nature 121. Pauli. Wentzel. Roy. 1293 (1929). . Soc. W. W. by A. Z. Phys. First W. Garden City. Dirac. A117. ibid. 41). Lincei 3. E. Roy. for the application of this theory to the calculation of the Zeeman and Paschen~Back effects and the relative strengths of lines within finestructure multiplets. P.. Heisenberg. A. 661 (1926).483 (1929). New York. For the probabilistic interpretation of nonMrelativistic quantum mechanics. 81 (1927). W Heisenberg. 36. Roy. 25. 902 (1926). Z. Nature 117. Roy. April 17. 27a. 360 (1930).1. A1I8. Proc.. H. M. 38. 18. 41. Dirac.172 (1927). PaUli. C. Dirac. quoted by Mehra in Aspects of Quantum Theory. P. 24. Ref. Natufwissenchaften 17. Cambridge.
35. Weisskopf and E. For a more accessible derivation. 33. Inc. 4.. Ref. Ehrenfes!. J. H. Phys. Proc. Born and Jordan.151 (1928).60 (1931).. Roy. Einstein. Phys. Pauli. by J. Unfortunately the relevant parts of this paper are not included in the reprint collection Sources of Quantum Mechanics. 13. Roy. 431 (1930). F. 121 (1917). P. P. 47. Heisenberg. Quantum Mechanics. Dirac. Phys. New York. 61 (1931). Dirac. ed. M. New York): Chapter IV. The Theory of Groups and Quantum Mechanics. 133. A. 18. Beyer (Dover Publications. Z. Rev. Also see P. 62. Proc. P. Phys.491 (1933). New York. 30a. Lineei Rend. 34. Pauli. 1973): pA15. . 1968): Section 44.. Roy. 36a. 4. 528 (1906). Roy. 29. I. A114. Phys. 19321933). Dirac. Z. 30. 9. 320. Science 76. M. reprinted in English in van der Waerden. Proc. Dirac. P. A. C. 36. The latter paper is reprinted in Foundations of Nuclear Physics. M. 3rd edn. 'A Report on Quantum Electrodynamics. Fermi. Handbuch der Physik (Julius Springer. ed. M. and Jordan. I. Soc. Section 3. Schwinger.References 43 28. 4. T. J. D. P. reprinted in Quantum Electrodynamics. 4. 562 (1930). Z. New York. 4. 43. Z.' in The Physi cist's Conception of Nature (Reidel. Born. 63. A. A. Dordrecht. 7. Berlin. Schwinger (Dover Publications. 1949). 1958). 12. Ref. 710 (1927). 881 (1929). Ref. Dirac. Anderson. E. 34a. A1l2. Mod. Inc. Z. Jordan and W. W. Weyl. Soc. 29a. 37. Tamm. V. Rev. Soc. Robertson (Dover Publications.j Phys. Proc.203 (1941). Phys. Section 12. 545 (1930). by R. translated from the second (1931) German edition by H. Soc. Oppenheimer. 661 (1926): Section 5. Mod. f Phys. A133. 35. Soc. A. Inc. cited in Ref. Rev. Wigner. Schiff. A114. R. M 31. see L. 238 (1932).87 (1932). A. Proc. Roy. M. 54 (1930) 36b. Rev. f Phys. 243 (1927). (McGrawMHill Book Company. P. 32. P.
Proc. 46. Camb. 794 (1950). He/v.. Soc. P. Helv. Acta 7. Phys. Y. 44 1 Historical Introduction 38.177 (1940). E. Soc. E. f Phys. M. Henri Poincare 6. 45. Rev. Acta 12. Jordan and O. J. 1. Pauli. 729 (1932). Phys. P. Ref. Fock. Kon. . 58. N. M. Fermi quotes unpublished work of Pauli for the proposition that an unobserved neutral particle is emitted along with the electron in beta decay. L. Nishina.. 468 (1932). pp. Proc. 44. Klein. A. Proc. Zeit. R. V. Phys. Z.f Phys. dansk. Physica 7.8 (1933) (translation in Selected Papers oj Leon Rosenfeld. 40. 453 (1932). Also see W. Rosenfeld.. Klein and Y. M. 453 (1932). Rev.Fys. Zeit. A. 853 (1929). ibid. P. by R. 43a. P.f. Rev. Leningrad 1933. 45. Cambridge. 35. P. Furry and J. Fermi. 473 (1927). M. ibid. Mat. Z . M. Dordrecht. W. 45. lnst. Pauli and F. Proc. Proc. Jordan. 90. Also see L. Also see R. (1939).f Phys. Z. A. Fock. Wigner. Phys. 56.f Phys. 43. XII. Z. 47. Phys. w. Stachel (Reidel.. Z. 41. 709 (1934). P. 4Oa. 729 (1932). Z. 361 (1930). Z. S. R. Phys. 47. A. Pauli and V. 161 (1934). 42a. 631 (1928). London A136. reprinted in English translation in A. Phys. Pauli. Soc. Dirac. also see I. Phil. Z. ed. Cohen and 1. C. 150 (1934). Phys. Early Quantum Electrodynamics (Cambridge University Press. 59. Selsk. 869 (1929). Soc. This article is reprinted in Quantum Electrodynamics. Camh. 26. Belinfante. 209 (1934). Dirac. Heisenberg. H. and B. W.f Phys. 44. Jordan.f. O. Ann. A. der Sowjetunion 2. W. Roy. 751 (1929). 245 (1934). Dirac. Soc. Pierz. Phys. 76. Zeit. 137 (1936). Rosenfeld. der Sowjetunion 3. V. 88. A. 78. 545 (1930). No. 1 (1929). The latter two articles are reprinted in Quantum Electrodynamics. Phys. 700 (1929). f Phys. Roy. M. 42. 1994). Jordan and E. Dirac. 64 (1933). P. 30. M. This paper uses a density matrix formalism developed by P. p. Dirac. W. 30. Nishina. 35. Miller. 267. Rosenfeld. A136. A. Heisenberg and W. f Phys. Z. 146. W. Dirac. 716 (1940).. Peierls. Roy. vid. 168 (1930). Ref. 62. Also see P. Bohr and L. Weisskopf. 76. Z. . Oppenheimer.420 (1934). 1979)). Phil. f Phys. This particle was called the neutrino to distinguish it from the recently discovered neutron. Medd. Pauli. Podolsky. P. 29 and 312. 39. Tamm. 52.
89. 58. C. Conversi. 194. 506 (1947). 54. 51. 54a. 61. 673 (1929). 37. 51. Rev. 837 (1930). M. Oppenheimer. Weisskopf. Rev. 143 (1946). 71. 55. Phys. ibid. F. Rasetti. 855 (1947). Thear. 61. Butler. 59. Marshak and H. Roy.7 (1934). R. Anderson. Webb. The electromagnetic self energy is calculated only to lowest order in r:t in these references. 817 (1934). 333 (1931). f Phys. Ehrenfest and 1. 45. 673 (1930). Chadwick. 494 (1939). Powell. Compt. 598 (1930). J. A136. Oppenheimer. f Phys. 77. Soc. ibid. A146. 884 (1937). 45 Ann. A154. Z. F. 62. Heisenberg. G. R. Proc. Ref. Nordheim and N. 52. Heitler. H. Phys. 62. M~lIer. 30. Z. f Phys. M. C. Rend. Proc. 209L (1947). 453. and Bio. R". Bhabha. Phys. Brown and H. This article is reprinted in The Foundations of Nuclear Physics. 53. E. Phys. 692 (1932). Soc. No. 63. C. R. Bethe. 50a. 560. P.Math. ibid. Pancini. C. Phys. 30.48 (1935). 77. Z. 90. Prog. Roy. Lattes. Carlson and J. For references. 27 (1934). G. M.. G. Racah. 1 (1994). (Japan) (3) 17. 83 (1934). 1003 (1937). Soc. 220 (1937). Ref. the proof that the divergence . Sakata and T. 14. 57. CurieJoliot and F. This article is reprinted in The Foundations of Nuclear Physics. Occhialini. S. 721. 531. W. H. d. Phys. F. also see I. 35. Waller. 168 (1930). 1. 56. S. Proc. 195 (1936). Inoue. Proc. Rochester and C. 49. Nature 160. W. 1 (1932). Street and E.. J. P. J. V. 17. 568 (1932). 52. Naturwiss. Phys. Science. Rechenberg. Rev. L. 61. reprinted in English translation in Early Quantum Electrodynamics. Yukawa. 51. R. E. Oppenheimer. H. and O. 69 (1936). Nuovo Cimento 11. J. 13. Phys. Joliot. 56. Roy. I. also see G. Bethe and W. and C. Neddermeyer and C. Rev.References 48. Stud. 50. C. Hist. D. in Phys. Ref.. 461 (1930). Heitler and G. F. 60. A. 25.. D. J. Rev. Soc. Rev. Rev. S. Nature 160. f Phys. Stevenson.486 (1947). Phys. Phys. Piccioni. ibid. Herzberg. 59. H.273 (1932). see L. Phys. Z.
Weisskopf. 67. Born and L.Fys. Ref. 173 (1948). Also see W. 69a. 66. Z. Henri Poincare 6. Z. ibid. Heisenberg. 3/3. Brown and L. . Dansk. Rev. Lewis. 73. Pauli. A. 141 (1935). (1946). H. Heisenberg. Heisenberg and H. Fierz. Phys. Hoddeson (Cambridge University Press. 35). 55. 20 (1938). Rose. P. 23. Phys. W. 55 (1935). Ref.. Pauli and M. 19. 52. Nordsieck.. Phys. No. Ann. Vid. Wentzel. Z.j. W. W. Also see Furry and Oppenheimer. 714 (1936). 68. 64. Kockel. 48. ReF. Peierls. Schwinger. Rev. Thear. 479. Medd. J. 137 (1936). Epstein. Pauli and M. 44. Phys.462 (1936). 177 (1948). Rev. S. No. [nst. 336. 209 (1934). ~" 46 1 Historical Introduction is only logarithmic in all orders of perturbation theory was given by Weisskopf. 120. Tomonaga. Ref. Sels. Rev. N. A. 65. Phys. in The Birth oIParticle Physics.959 (1939). 1107 (1937). 23. Z. 90. M. reprinted in English translation in Early Quantum Electrodynamics. Sachs. 45. by L.. Wiss. Mat. S. f Phys. 52. I. 69b. Nature 137. f Phys. For subsequent calculations based on less restrictive assumptions. Soc. Comb. Rev. Infeld. 98. see W. Phys. 608 (1946). Rev. Naturfarsch. 203 (1933). 71. Dirac. Also see C. . Cambridge. 32. d. 246 (1935). Euler. A. Serber. 635 (1933). M. Bloch and A.673 (1943). Ann. Phys. 72. Z. Kemmer and V. Akad. XVII Conseil Solvay de Physique. Also see J. Koba and S. 317 (1934). f Phys. 72 (1939). Dancolf. Phys. F. Z. 90. M~ller. Euler and B. Weisskopf. M. R.~. Ref. I (1945). 49. Ref. 23. 49 (1935). Rev. Wheeler. H. f Phys. Prog. Rev. 30. reprinted in Early Quantum Electrodynamics. 56. 73. F. Heisenberg. Proc. G. 45. Phys. Phys. P. Schwinger. Nuovo Cimento 15. 70. 86. Naturwiss. 63. 70a. 44. Kon. Soc. W. Proe. Ref. M. Dirac. reprinted in English translation in Early Quantum Electrodynamics. p. 726 (1034). 69. 290 (1948). 1983): p. 167 (1938). AlSO. f Phys. Z. Heisenberg. 43. Roy. 87. 659 (1936). Z. Phys. Ref. A. Uehling. W. J. 150 (1934). 68a. W. ed. 209 (1934). 45. Rev. 86. W. E. Phil. Phys. (This last article is reprinted in Quantum Electrodynamic. 54 (1937).513. 84.
Rev. A. A. 1961). Brussels. 74. The SMatrix Theory of Strong Interactions (W. Uehling. This suggestion was based on experiments of W. H. Jr and R.j References 47 73. P. ibid. Also see W. A. Rev. C. This article is reprinted in Quantum Electrodynamics. 1950): p. Soc. Phys. 56. T. 75. Bethe. E. 54. Chew. F. 1950). Weinberg Gravitation and Cosmology. SO. Ref. Rev. Houston. l". and W. 79. P. R. Theor. reprinted in English translation in Early Quantum Electrodynamics. Rev. Phys. XIV. For further references and a discussion of the application of actionata~distance theories in cosmology. 74. Mod. Lamb. 21. Phys. 1113 (1938). Proc. Rev. 34 and pp. This article is reprinted in Quantum Electrodynamics. 68a. Phys. M. e. 64. 17. 558 (1938). Medd. Phys. Ref. 388 (1949). This article is reprinted in Quantum Electrodynamics. 76. IS. Rev. New York. Mod. Rev. 749. E. B. Rev. see J. 45. S. 769 (1949). 241. 6 (1936). Inc. Nuovo Cimento IS. W. For a report of contrary data. N. Roy. Ref. 79a. Kramers. 75. 51. Retherford. P. 78. 4. H. AlSO. Phys. Brussels.fys. 77. 241 (1947). Rev. Kon.. 175 (1943). E. G. Ref. 35. 81. and in English translation in Early Quantum Electrodynamics. 367 (1948). Williams.339 (1947). . 35. Mod. 121 (1948). E. 11. Ned. Fierz.1430 (1948). Phys. Wheeler and R. Phys. J. Phys. V. The article by Kroll and Lamb is reprinted in Quantum Electrodynamics. Weisskopf.L1'. E. Peierls in Rapports du sm e Conseil de Physique Solvay 1948 (R. Pauli. 164 (1940). Rapports du S"'e Conseil de Physique Solvay 1948 (R. see S. No. Y. Ref. H. Stoops. Phys. 72. 174.RSIDAD DE SA.V. Kramers. and of yet other attempts to solve the problem of infinities. Rev. 76. 108 (1938). ibid. 1240 (1949). Fukuda.. 1 (1942). H. see W. 79a. 35. See. Ref. J. Vid.446 (1937).. Williams. A. Sel. C. 898 (1949).g. Feynman.. R. Phys. 1972): Section 16. Richardson.u. Weisskopf. (Wiley. 72. 80. Mat. O. Prog. 939. Soc. F. For a review of classical theories of this type. 75.. Tomonaga. Ref. 45. 134 (1944). Pauli and M. Feynman. Dirac. A. 75. Ref. Roy. Phys. Schwinger. Lamb. 440 (1950).3. For a criticism. A. Pasternack. see R.J dO' ~~~~. A. Ref. 54. Phys. J. M. Proc. Rev. French and V. Kroll and W. Mod. Rev. 20. 425 (1949). Stoops.\ UNIVi:. Narwink. 157 (1945). Dan. 47. and S. Drinkwater. Miyamoto. W. 35. especially p. Benjamin.
H. 452 (1951). Z. Julian. Tomonaga. 384 (1939). Phys. 749. Kahn. Prog. Ref. Koba. Ref. Phys. 651 (1949). 914 (1947). Ref. 486. J. P. Phys. Rev. 101 (1948). Quoted by R. 224 (1948).. Schwinger in Ref. Rev. Rev. Heitler. Phys. Z. Phys. G.. 56. F. Rev. 939. 85.790 (1949). Nagel. 82. 1. 101 (1947). 35. Kanesawa and S. Feynman. 664. S.. 74. R. Nafe. The first and fourth of these articles are reprinted in Quantum Electrodynamics. 35. 290 (1948). 1439 (1948). W. Roy. . and S.. Phys. Koba. J. Foley. 269 (1939). 35. 87. Schwinger. ibid. D. 458 (1940). Phys. ibid. I. and I. Rev. Ref. Phys. 35. S. 72. 769 (1949). 37. 72. and J. R. Rev. Rev. Phys. T. Zacharias. Ref. 3. Ito. Tomonaga. Rev. 86. E. Z. Rev. ibid 80. Tati. Pmc. Nat. 440 (1950). 57. and B. Koba and S. 56. 83. 76. Soc. and S. 416 (1948). Acad. E. 1. 367 (1948). 74. Phys. ibid. E. 984 (1947). 84 includes a corrected version of Breit's results. 35. W. 1736 (1949). ibid. 2. in The Birth of Particle Physics. Kusch and H. 1.. Sci. 74. Nelson. 75. 88. M. 88a. 83a. ibid. Serber. 973 (1947). P.. ibid. These articles are reprinted in Quantum Electrodynamics. D. ibid. Mod. p. 1256 (1947). 961 (1939). This article is reprinted in Quantum Electrodynamics. Rev. E. Rev. 713 (1953). 3. Frohlich. 270. 84. Tomonaga. 91. AI7. S. Phys. ibid. 75. B. All but the first two of these articles are reprinted in Quantum Electrodynamics. 73. Breit. 88b. All but the second and third of these articles are reprinted in Quantum Electrodynamics.48 1 Historical Introduction 82. Thear. Mod. Phys. R. Phys.. 1430 (1948). Tomonaga.. 89. 914 (1951). Lamb. Prog. 1. Tomonaga. Rabi. 76. Tomonaga. 71.. Schwinger. Phys. Phys. 69b. 71. Proc. Rev. 20. Rev. Jr. 3. Rev. Dyson. Phys. Ref. S. Theor.27 (1946). 276 (1948).
'P). and others in 192526. (There are also certain technical assumptions that allow us to take limits of vectors within Hilbert space. 2. \{I) also satisfies a positivity condition: (\{I.). \{I) > 0. and vanishes if and only if \{I = O. Born. The reader is assumed to be already familiar with quantum mechanics.(0I>1. (~10I>1 + ~201>" '1') ~ ~.) ~ <1(01). '1'. It has a norm·: for any pair of vectors there is a complex number (11).' (i) Physical states are represented by rays in Hilbert space. 'I'll. for arbitrary complex numbers ~. that is.2) (2. if 11> and 'P are vectors in the space (often called 'statevectors') then so is S"11> + 11\{1. in the generalized version of Dirac.1 Quantum Mechanics First. A Hilbert space is a kind of complex vector space. '1') (01).) A ray is a • We shall often use the Dirac braket notation: instead of ('+'1. 01>( . fl. this section provides only the briefest of summaries of quantum mechanics.2 Relativistic Quantum Mechanics The point of view of this book is that quantum field theory is the way it is because (with certain qualifications) this is the only way to reconcile quantum mechanics with special relativity. 'I'll + <2(01).1. Therefore our first task is to study how symmetries like Lorentz invariance appear in a quantum setting.1. (2. (2. '1'). we may write (112).1. molecular. '1') + ~i(0I>2. nuclear. 49 . Pauli.1 ) + <2'1'. and condensed matter physics. some good news: quantum field theory is based on the same quantum mechanics that was invented by Schrodinger. Heisenberg.3) The norm (\{I. and has been used ever since in atomic. <I '1'1 ~ ('I'. such that (01).
and (iii) If a system is in a state represented by a ray .1. where is an arbitrary complex number with s I~I ~ 1.1.. eigenvectors with different :1 s are orthogonal..5) (<I>.7) where \fI and \fin are any vectors belonging to rays gf and 9't n . if the statevectors If' n form a complete set. \fI) = 1) with qt and \fI' belonging to the same ray if \fI' = ~qt.1. (for instance. (A pair of rays is said to be orthogonal if the statevectors from the two rays have vanishing scalar products.e. ji2. linear in the sense that A(~'I' + ~<I» ~ ~A'I' = + ~A<I> (2.50 2 Relativistic Quantum Mechanic. (\fI.'Jl or :1fl or ~2 .1. where for any linear operator (2.8) 2. .4) and satisfying the reality condition At A the adjoint At is defined by A. by measuring one or more observables) then the probability of finding it in the state represented by . If an observer 0 sees a system in a state represented by a ray .1.) A state represented by a ray ."»n is (2. At'l') . (ii) Observables are represented by Hermitian operators.: A\fI = G(\fI for qt in . These are map~ pings qt + A\fI of Hilbert space into itself.A<I>j" .~ .~ set of normalized vectors (i. (There are also technical assumptions about the continuity of A\fI as a function of \fl.) Another elementary theorem gives a total probability unity: L Pi!!! ~ 91.(A<I>. . '1') ~ ('I'. respectively.:1 IS (2. real. then an equivalent observer 0' who looks at the same system will observe it in a different .) ~ 1 .#l. and an experiment is done to test whether it is in anyone of the different states represented by mutually orthogonal rays Jit. for the observable represented by an operator A if vectors qt belonging to this ray are eigenvectors of A with eigenvalue ..6) An elementary theorem tells us that for A Hermitian. (2.~ has a definite value '1..2 Sy~~etries A symmetry transformation is a change in our point of view that does not change the results of possible experiments.'1. ..
ew' ~ :.4) (2. '1')' . a symmetry transformation that is infinitesimally close to being trivial can be represented by a linear unitary operator that is .2.2) (2. further conditions are discussed in the next chapter. (2. represented by a ray Jt' or Jtj or ~2" observers must find the same probabilities . (2. Instead. of course.).8) There is always a trivial symmetry transformation .f:. (This is only a necessary condition for a ray transformation to be a symmetry.2. the conditions for unitarity or antiunitarity both take the form vt = VI. respectively.ew ~ . they all involve a reversal in the direction of time's flow. A more complete proof is given at the end of this chapter in Appendix A As already mentioned. the adjoint of an antilinear operator A is defined by (<I>. but the two (2. represented by the identity operator U = 1.2.7) With this definition.5) Wigner's proof omits some steps. '1').2.~ then uqt is in the ray £HI.»' of rays we may define an operator U on Hilbert space. while the lefthand side is antilinear in $.2.) In particular. (Symmetries represented by antiunitary antilinear operators are less prominent in physics. the adjoint of a linear operator L is defined by (<I>.6.A''I') . (2.) An important theorem proved by Wigner 2 in the early 1930s tells us that for any such transformation Jt + . See Section 2. This operator is.A"').Lt'l') . Continuity then demands that any symmetry (like a rotation or translation or Lorentz transformation) that can be made trivial by a continuous change of some parameters (like angles or distances or velocities) must be represented by a linear unitary operator V rather than one that is antilinear and anti unitary.(A<I>.2.2.2. U'I') U(~<I> ~ (<I>. (2. such that if qt is in ray .6) This condition cannot be satisfied for an antilinear operator.1) P(.. with V either unitary and linear (U<I>. U'I') U((<I> (<I>.2. '1')' ~ ('I'.6) would be linear in <1>. (2.qf + .ew") ~ p(.2.3) or else antiunitary and antilinear (U<I>. unitary and linear.2 Symmetries 51 state.'.(L<I>. + ry'l') ~ ~U<I> + ryU'I' ~ (2. + ry'l') ~ ~'U<I> + ry' U'I' . because in this case the righthand side of Eq. '1').0f!.
)'P.{T. so it is a candidate for an observable.)'P A + U(T2)U(T.52 2 Relativistic Quantum Mechanics infinitesimally close to the identity: U=l+i€t (2. written T.10) Furthermore.eftn into 9t~.11) on the left with U. which takes 9l~ into 9l n .." U(T2T. Tt} U(T2)U(T. and there is an identity transformation. Indeed. the linearity (or antilinearity) us that these phases are independent of the state qJn. most (and perhaps all) of the observables of physics.2. U(Td must yield a vector U(TdqJn in the ray Jf~.)'P. and T2 is another transformation that takes Jtl~ into Jf~.2. qJ B. unlike the symmetry transformations themselves. Then. ~ e..2.1(TzTd.. arise in this way from symmetry transformations.(jf~. T = 1. Also. of U(T) tells is the proof. such as angular momentum or momentum.2.2. we have then (2. Here Consider any two different vectors qJ A.TdU(T2T. (2.1. then the result of performing both transformations is another symmetry transformation. that takes ~" into . and if T z takes this ray into jfl~.11) Any unitary or antiunitary operator has an inverse (its adjoint) which is also unitary or antiunitary.)('P A + 'P B) ~ U(T2)U(T. But U(T2TdqJn is also in this ray. so these vectors can differ only by a phase ¢n(Tz. then acting on a vector qJn in the ray £if n.9) with r.. For this to be unitary and linear. which are not to each other. If T 1 is a transformation that takes rays 9t n into 9i'~. Since qJ A and qJ B are linearly independent.10) to the state + qJB. rather than on rays.)('P A + 'PH) ~ U(T2)U(T. The set of symmetry transformations has certain properties that define it as a group. but with a complication due to the fact that. applying Eq.2. the operators U(T) act on vectors in the Hilbert space. respectively. If T] takes .12) the upper and lower signs referring to U(TzTd unitary or antiunitary. (2. (2. a real infinitesimal.)'P B = ei¢A U(Tz TdqJ A + e i4>B U(Tz TdqJ B. which we write T2 Ti. Multiplying (2. we have e'. t must be Hermitian and linear. a symmetry transformation T which takes rays Bin into Jf~ has an inverse. proportional qJ AB = qJ A with one significant exception. The unitary or antiunitary operators U(T) corresponding to these symmetry transformations have properties that mirror this group structure. which leaves rays unchanged. this is only . then acting on U(TdqJn it must yield a vector U(Tz)U(Tt}qJn in the ray 9l~.
known as a connected Lie group.2 . and take 4> ~ 0 in Eq. The group multiplication law then takes the form (2.2.2.15) with r(l}. we say that there is a 'superselection rule' between different classes of states. say ea. of special importance in physics. it can tell us whether the group has any intrinsically projective representations at all. For general phases ¢(T2. There is a kind of group.2. For a Lie group. any symmetry group with projective representations can always be enlarged (without otherwise changing its physical implications) in such a way that its representations can all be defined as nonprojective. we have what is called a projective representation. this would say that the U(T) furnish a representation of the group of symmetry transformations. it is widely believed to be impossible to prepare a system in a superposition of two states whose total angular momenta are integers and halfintegers. Ttl act upon.14) is that it may not be possible to prepare the system in a state represented by 'FA + 'F B_ For instance. the transformations of such continuous groups must be represented on the physical Hilbert space by unitary (rather than antiunitary) operators U(T(O)). (2.14) For ¢ = 0.O) a function of the of the identity. The structure of the Lie group cannot by itself tell us whether physical statevectors furnish an ordinary or a projective representation.16) /"(0.7. with ¢ = O.2. the phase in Eq.7. The exception to the argument that led to Eq. with each element of the group connected to the identity by a path within the group. but as we shall see. we must have es and as. These are groups of transformations T(e) that are described by a finite set of real continuous parameters. Taking Oa = ~ 0 as the coordinates (2. We will have more to say about these phases and projective representations in Section 2. As we shall see there.2. (2.3 and the phases ¢(T2. respectively.14). and therefore this can be written as an operator relation (2.0) /"(0.10) is independent of the statevector 'fin.13) So as promised.2. these operators can be . In such cases.0) ~ 0" . (2.2. we will just assume that this has been done. or a representation 'up to a phase'. Ttl.))ln1n1etries 53 possible if (2. Until Section 2. T I ) may depend on which of these classes of states the operators U(T2)U(Td and U(T2. As already mentioned.2.
. + J ObO't". but from the 00 terms we obtain a nontrivial condition (2. i.2. provided we know the firstorder terms. In Section 2.2...2.22) is the single condition needed to ensure that this process can be continued: the complete power series for U(T(O)) may be calculated from an infinite sequence of relations like Eq..21) This shows that if we are given the structure of the group. ·)tb.16).21).0. i.2. (2.. the expansion of rUJ. (2..2.) Then Eq. tbe = teb. the function f(O.23) Such a set of commutation relations is known as a Lie algebra. Suppose that the U(T(O» fonn an ordinary (i.2.2.e. but it . the generators tao This does not necessarily mean that the operators U(T(O)) are uniquely determined for all 0" if we know the t a .2.22) where Cabe are a set of real constants known as structure constants C abe ...7 we will prove in effect that the commutation relation (2. and hence its quadratic coefficient fabe> we can calculate the secondorder terms in U(T(O)) from the generators t a appearing in the firstorder tenns. there is a consistency condition: the operator the must be symmetric in band c (because it is the second derivative of U(T(O)) with respect to b and flC) so Eq. (2. (2.2.0.21) requires that e [tb.e. 0). (2.. (2.) til /1 + 1(0 II + 0 + . tcJ = iCabct(l' (2.in powers of ()a and According to Eq. + ...) (0' + 0' + . (2.. are Hennitian operators independent of the Os.2.(The presence of any terms of order would violate Eq.e.20) iljala The terms of order 1. J 1 + i (flU + Va + rbc(}b(y + .nata + !ObOCtbc + .17) where t a .2. (2.2. etc..20).O) to second order must take the form ea.rhc + F'<'h. (2.. However.2. and 7P automatically match on both sides of Eq.18) Let us see what this condition looks like when expanded.19) with real coefficients rhe. (2. not projective) representation of this group of transformations.] x [I + .. + .18) reads 02 or lP [1 + = + ~ rNh bc + .54 2 Relativistic Quantum Mechanics represented in at least a finite neighborhood of the identity by a power senes U(T(O)) ~ I + iG't.. (2. .16).2.02 .
2. and xO = t a time coordinate.7J) are in this neighborhood. If x ll are the coordinates in one inertial frame (with Xl.2) .2.19) vanish.3 Quantum Lorentz Transformations Einstein's principle of relativity states the equivalence of certain 'inertial' frames of reference. and so do the structure constants coefficients (2.2. The generators then all commute rbc [tb. in such a way that Eq.2.1) or equivalently 'Illv ax .3.e. obeyed by Newtonian mechanics. it is easy to calculate U(T(8)) for all ea.26) 2. we have for any integer N U(T(B))~ [U(T(~))] .\' ox p ox rr = 'Iprr . other inertial frame.24) This is the case for instance for translations in spacetime.] ~ 0. From Eqs.15) is satisfied if o. There is a special case of some importance. t. 0) ~ 0" + ii' (2. the coordinates x 11 must satisfy '111" dx"'d'Px  1/11\' d'Pd' X X (2. we have then N U(T(H)) ~ and hence lim [I NFt) + ~O"t"]N N (2. The extension to all oa is discussed in Section 2. or for rotations about anyone fixed axis (though not for both together). Then the in Eq.18) and (2.x . Suppose that the function f(O.23).25) Such a group is called Abelian.2.2. and keeping only the firstorder term in U(T(8jN)). the speed of light being set equal to unity) then in any . In this case.I' oX . that we will encounter again and again. (2. (2.x~ Cartesian space coordinates. It is distinguished from the Galilean principle of relativity.2. (2. 0) (perhaps just for some subset of the coordinates oa) is simply additive f"(B.7.24).2.2. Letting N . (2.3 Quantum Lorentz Transformations 55 does mean that the U(T(O)) are uniquely determined in at least a finite neighborhood of the coordinates oa = 0 of the identity. and f(8.x2.3. (2. by the transfonnation connecting coordinate systems in different inertial frames.
These transformations have the special property that the speed of light is the same (in our units. then so does AllpAPv.5) with "urN'r and inserting parentheses judiciously.formatlons."onformal trarll. (2.3.dx'~dx" is proportional though generally not equal to /j~.3) and the summation convention is in force: we sum over any index like /l and v in Eg.4). we have 171'" AI' p(A vuA K r'1"') =A = K f! = '1llv 'I VK NIp. with '7 00 = 1. then the effect is the same as the Lorentz transformation xl' _ xl/ Il .· a light wave travelling at unit speed satisfies Idx/dtl = 1. The bar is used here just to distinguish • There is a larger class of coordinate transformations. it is useful to write the Lorentz transformation condition in a different way.4) with all arbitrary constants. written 'II"'.)x V + (iVpa P + a (2. for which /jj!\.3.dt 2 = 0. Multiplying with the inverse of the matrix '7l' v AI' p then gives AI'(TA\l1.2)..3. once upstairs and once downstairs. known as .3. '100 =1 (2. The matrix '1IlV has an inverse. equal to unity) in all inertial frames. If we first perform a Lorentz transformation (2.5). Multiplying Eq. and conditions All v a constant matrix satisfying the (2.17 11 = tl 22 = 11 33 = +1.3. and then a second Lorentz transformation XiI' _ X"I' .2) is linear 3a (2. but the physical relevance of these conformal transformations in four spacetime dimensions is not yet dear.3. (2. (2. with X"I' = AI' px 'P + all = N'p(AP vXv + aP) + all i ). which happens to have the same components: it is diagonal. with elements 'lll = '722 = 1'/33 = +1. (2.dx~dx'.7) (Note that if Al'v and A/v both satisfy Eq. .3.3. or in other words '7l' vdx IJ dx" = dx 2 .6) These transformations fonn a group.56 2 Relativistic Quantum Mechanics Here til'" is the diagonal matrix.5) For some purposes. and which therefore also leave the speed of light invariant.. and hence Idx' jdt'l = 1. with X"I' = (NlpAP. (2. which appears twice in the same term. Any coordinate transformation Xl! _ XiI' that satisfies Eg.3.r '1"K .3. so this is a Lorentz transformation. Conformal invariance in two dimensions bas proved enormously important in string theory and statistical mechanics. from which it follows that also 1J11Vdx'Pdx'V = 0.
of course.5) and (2. and.5) gives (DetA)' ~ 1 (2.O). In accordance with the discussion in the previous section.3. Note that if Nt. and 3. (2.3. The appropriate enlargement is described in Section 2. Further.3.3. But Sq.2. (2.3. from the OOcomponents of Eqs. Those transfonnations with ADO > +1 form a subgroup.a)T(A.A20.A 20 + A ~o ~o ]A 3 0 .a) is properly known as the inhomogeneous Lorentz group.A03) has length .Aa + a). (2.3.13) shows that the threevector (AIO. to avoid the appearance of a phase factor on the righthand side of Eq. we have (A° 0) 2 ~ 1 + NoN. those transformations with all = obviously form a subgroup.9) that either DetA = +1 or DetA = ~ 1. has an inverse. a)'P . necessary to enlarge the Lorentz group.O) ~ T(AA. (2. First.11) (As already mentioned. alUlA.12) known as the homogeneous Lorentz group. those transformations with DetA = +1 obviously form a subgroup of either the homogeneous or the inhomogeneous Lorentz groups. Aa + il) . the transformations T(A.~ 1 + A° iA ° i.3. in general. (2. U(AA.Ao2.  The operators U satisfy a composition rule U(A.10) The inverse of the transformation T(A.3. (2.5) takes the form (A 1)1' v = A/  '7vll'1l'rT N\. (A1)VI' which we see from Eq.a) ~ T(AA. then (AA) 0 0 ~ A oA 0 ° + A IA 0 0 1 0 + A .3.3. We see that either ADO > +1 or AO o < 1.. (2. (2. (2.8) Taking the determinant of Eq. it is.A30) has length J(AOO)2 1. we note from Eq. .2. Also. or Poincare group..a) induce a unitary linear transfonnation on vectors in the physical Hilbert space 'P  ~ UtA.O) T(A.a) is seen from Eq.O).11). and AP v are two such A s.a) induced on physical states therefore satisfy the composition rule T(A.) The transformations T(A.3. a) ~ (2. with ° T(A.9) so N\.3.7. and similarly the threevector (A°I.3.3 Quantum Lorentz Transformations 57 one Lorentz transformation from the other.3. AI a). (2.) The whole group of transformations T(A. It has a number of important subgroups.6).13) o with i summed over the values 1. the identity transformation is T(l.8) to be T(A I. (2.
any Lorentz transfonnation that can be obtained from the identity by a continuous change of parameters must have Dett\. Since it is not possible by a continuous change of parameters to jump from Det A = +1 to Dett\. so the scalar product of these two threevectors is bounded by 0 0 0 IA lA 10+ A 2A 20 + A 3A301 < .14) and so (AA)oo > AOoA D ~ o /(I\.:1° 0 =1.15) and Y is the timereversal matrix.4 The Poincare Algebra As we saw in Section 2. or may be written as the product of an element of the proper orthochronous Lorentz group with one of the discrete transformations . For the inhomogeneous Lorentz group. plus space inversion and timereversal. much of the information about any Lie symmetry group is contained in properties of the group elements near the identity. Until then. (2.3. 2.. = (iI'"..3.(A 0 0)' .11(1\.4. whose nonzero elements are . so we want to study those transfonnations with ~Jl All v=(Jv + (0".1 >t.1) both Will' and ell being taken infinitesimaL The Lorentz condition (2.qp or .6. We will consider space inversion and timereversal separately in Section 2.1. we will deal only with the homogeneous or inhomogeneous proper orthochronous Lorentz group.°0)2 .1. (2.58 2 Relativistic Quantum Mechanics V(AOO)2 .5) .16) Thus the study of the whole Lorentz group reduces to the study of its proper orthochronous subgroup. Any Lorentz transformation is either proper and orthochronous.:133 = I. and AOo of the same sign as [or the identity.0of. I' (2. The subgroup of Lorentz transformations with Det A = + 1 and A°0 > + 1 is known as the proper orthochronous Lorentz group. ::.1 .'7 or f!iJ:Y. the identity is the transformation AI\.2.3. and hence must belong to the proper orthochronous Lorentz group. (AOo)2 .11 =:722=. where & is the space inversion. or [rom AO o > +1 to ADO < 1. = 1. whose nonzero elements are (2. ai' = 0.3.
2) An antisymmetric secondrank tensor in four dimensions has (4 x 3)/2 = 6 independent components. and the dots denote terms of higher order in (jJ and/or f".ign for the f!pPI' term in (2. We are here using the convention.p2. As we shall see.ymmetry groups may depend on the states on which the .1\ an inhomogeneous Lorentz transformation is described by 6 + 4 = to parameters. (2.ymmetries act.4. we see that this condition now reduces to the antisymmetry of W I 1\' (2.and ('independent operators. so including the four components of (. so the .O) acts can be ruled out by the same reasoning that we used in Section 2.€) to be unitary. .4. to be used throughout this book. the possibility that the proportionality factor may depend on the state on which U(I.F) must then equal 1 plus terms linear in w pu and ('p.4 The Poincare Algebra 59 reads here " 'lP(f = 'l!'v (U fI = ) + OJ P)(" <J + OJ 'u p U 2 1Jop + W up + OJ p" + 0(w ). and p 3 are the components of the momentum operators. that indices may be lowered or raised by contraction with '1'11' or 'I Il I' OJ"I)  1J11U(jJ!' P w ll !.4) Since (r)pa is antisymmetric.2. U(l + W.ibility that the phase. the commutation relations do not allow us to di~tinguish between pi' and P~. it must be proportional to the unit operator: and by a choice of phase may be made equal to it. For an infinitesimal Lorentz transformation (2. The consistency of the choice in (2. 1JI"'(jJ(fp. Where superseledion rules apply. it may he neces:.1. J31.3.4. and J12 are the components of the angular momentum vector. in projective repre. We write this as U(1 + w. pl.4.4.>entation.5). of .. Keeping only the terms of first order in OJ in the Lorentz condition (2. Since U(1. In order for U(1 +w.(') = 1 + !iOJpuJP<J  if"pPP +. ppt = pI' .O) carries any ray into itself. J23.1). the operators JI>U and PP must be Hermitian JPat = JPu .2 to rule out the pos.4.3) Here pur and pp are OJ.3) with the wmal definition of the Hamiltonian po is shown in Section 3. or Hamiltonian:· • In the absence of supersclection rules. •• We will see that this identification of the angularmomentum generators is forced on us by the commutation relations of the J!'" _ On the other hand.5) _]"p . we can take its coefficient metric also lpe ~ to be antisym(2. and pO is the energy operator.3) i~ a matter of convention. JPrr (2.ary to redefine V(LO) by phase factors that depend on the sector on which it acts.4..
a)JP"" U. (2. E!. (2.9) For homogeneous Lorentz transformations (with all = 0). (2.7) Equating coefficients of using (2. they tell us that pI' is translationinvariant. where All v and all are here the parameters of a new transformation.8) and (2. For pure translations (with AIl I• = Ollv).AwA la)e pe . Equating coefficients of (J)Il V and find the commutation rules i [jllV .60 2 Relativistic Quantum Mechanics Jpa We now examine the Lorentz transformation properties of We consider the product V(A.11) on both sides of these equations. (2.6) To first order in OJ and E.14) [pe.10) I .wIIV ie' e p pe .8) (2.A.4..3).1(A. (2.4.a)V(I+w. we find UJ pa and ('I' on both sides of this equation (and V(A. i. we . these transformation rules simply say that JIlV is a tensor and pll is a vector. According to Eq.4. + W.t .1(A.e.4. Eqs.1a) U(A.AEAwA1a) .1(A.8).3. the change of the spacespace components of Jpo under a spatial translation is just the usual change of the angular momentum under a change of the origin relative to which the angular momentum is calculated.a) . we have then (AE . so U(A. (2.11). 1/"1' J/lO tlPP pIT "llppO _ "Oll JI'V + "OV JPI' . the product U(Al. a) = A/' A/'(Jllv _ all P" U(A.a) = HAwA1)pvJPv (2.E)V. . Next. All v = oP v + w ll v and all = e!'.a) V(! and PP.E)V.11) that V(A. (2.4. and keeping only terms of first order in Wll v and ell.4. (2. (2.a) = A/Pll. with infinitesimals w P v and eP unrelated to the previous wand e.3.4.9) now become (2. It follows then from (2.a) [!wprrJpa €ppPJ Vl(A. + a" Pll) .12) i [Pll .4. unrelated to wand E. In particular.JpO] = pPI =w ll PP" .4.10)).a).9) to a transformation that is itself infinitesimal.3._Al a) is the Inverse of U(A. This is the Lie algebra of the Poincare group.4. pe) ~ O. [ :.13) (2.4.4.D). U(A.a)pr U1(A.a) ~ V (A(I +w)A1. but Jpo is not.a) equals V(l. let's apply rules (2. ]1)(1] = _1/ll er pI' . Using Eq.4.4.
).j. so by using (2.16) and.2.K j ] = [h Pj ] = [K j .J30} .a) ~ T(l.4. (2.J".15) and the angularmomentum three·vector J ~ {J 23 .24) E.4. The commutation relation (2.18) will be recognized as that of the angular·momentum operator. the Galilean group.4.J20. the energy pO itself. etc.13).4.4.J 12 } (2.4 The Poincare Algebra 61 In quantum mechanics a special role is played by those operators that are conserved.20) (2. (2.17) These are not conserved. In a three·dimensional notation.4.k.iJ+a).18) (2.4.a) form a subgroup of the inhomogeneous Lorentz group with a group multiplication rule given by (2.14) may be written [hJj] = [hKj] = [K.7) as T(I. the commutation relations (2. of course. iH 0. ~iEijkJk. (2. The pure translations T(l.22) (2. run over the values 1. we find that finite translations are represented on the physical Hilbert space by U(I.iJ)T(I.19) (2.4. We .24).3) and repeating the same arguments that led to (2. where i.21) (2.H] iPj . (2..4.a) ~ exp(iP'a.27) It is interesting to compare the Poincare algebra with the Lie algebra of the symmetry group of Newtonian mechanics.jkPk . The remaining generators form what is called the 'boost' three·vector K~ {J1O. and 3. i.25) This is additive in the same sense as (2.4.13) and (2. Pj] = [J"H] ~ [Kj.4.4.H] = iEijkJk. 2. (2. that commute with the energy operator H = pO. 0) ~ exp(iJ .e.4.4.12).2. (2.26).4. we can show that a rotation RO by an angle 101 around the direction of 0 is represented on the physical Hilbert space by U(Re.4.4. (2..26) In exactly the same way.4. ~ 0.23) (2. 9).14) shows that these are the momentum three·vector (2. which is why we do not use the eigenvalues of K to label physical states.) .2. iEijkKk.4.4. [P"H] ~ [H. and Ejjk is the totally antisymmetric quantity with E123 = +1. j (2.3. Inspection of Eqs.
W '"'' mv 2 • Inspection of Egs. In this case physical states provide an ordinary rather than a projective representation of the expanded symmetry group.MJ ~ ~ [K. 2.K j ] =0. W] ~ 0 .M] = [hKj]=if'ijkKk.18){2. but this is not true for the action of these operators on Hilbert space: exp(iK· v) exp(iP' a) = exp(iMa' v/2) exp(i(K· v .5.mv.4. P '"'. [hPj] [J. W] ~ i P. W] [J.24). [K.4. (2. it is easier to obtain the Galilean algebra as a lowvelocity limit of the Poincare algebra.M] ~ [W. [Kj. The difference appears to be a mere matter of notation.W with a total mass M and nonmass energy W (kinetic plus potential) of order M '"'' m.4. (2.. and whose eigenvalues are the masses of the various states. [Ki.. However.P j ] = iMbij.. by what is known as an InoniiWigner contraction4.5 OneParticle States We now consider the classification of oneparticle states according to their transformation under the inhomogeneous Lorentz group. a)) . On the other hand. Note that the product of a translation x _ x + a and a 'boost' x _ x + vt should be the transformation x _ x + vt + a. which commutes with all the other generators. However...4. i€ijk Pk. with K of order l/v. by adding one more generator to its Lie algebra. so it is natural to express physical statevectors in terms of . .p. In this respect. with a superselection rule forbidding the superposition of states of different mass. the mathematics of the Poincare group is simpler than that of the Galilean group. [P. The appearance of the phase factor exp(iMa' v/2) shows that this is a projective representation. since we already have Egs. The components of the energymomentum fourvector all commute with each other. the energy operator is H = M +.62 2 Relativistic Quantum Mechanics could derive this algebra by starting with the transfonnation rules of the Galilean group and then following the same procedure that was used here to derive the Poincare algebra. except that with this reinterpretation of the Galilean group there is no need for a mass superselection rule.. the momentum and the angularmomentum operators are expected to be of order J '"'' 1. ~ [P.24) shows that these commutation relations have a limit for v « J of the form [hJj]=if'jjkh. there is nothing to prevent us from formally enlarging the Galilean group.18)(2. For a system of particles of typical mass m and typical velocity v.M] ~ 0.
4.4. It is natural to identify the states of a specific particle type with the components of a representation of the inhomogeneous Lorentz group which is irreducible. 0' 12.. (2. that have matrices Ca'a(/\..o .. so that the '¥p. such as the lowest state of the hydrogen atom.5. a specific bound state of two or more particles.a)'¥p. we see that the effect of operating on '¥ p.2.0 ~ U(Aj(A. consisting for instance of several unbound particles..1) and (2.u .I"pp)'I'p. (However. is to be considered as a oneparticle state. and will limit ourselves here to that case.1) For general states.a': U(A)'I'p.O) . i. ror instance. but the distinction between composite and elementary particles is of no relevance here.5..5.o ~ Leo'oIA.l.p)'I'A"". in other words.¥p. We take as part of the definition of a oneparticle state.) Eqs.2) Hence U(A)'¥p.5 OneParticle States 63 eigenvectors of the fourmomentum.e. .o with a quantum homogeneous Lorentz transformation U(A.o with pl''¥p.a with (J within any one block by themselves furnish a representation of the inhomogeneous Lorentz group.p) is blockdiagonal.o transform under transla tions: U(1. In some cases it may he convenient to define particle types as irreducible representations of larger groups that contain the mhomogeneous proper orthochronous Lorentz group as a subgroup.p.3) In general.5.9). as we shall sec. Using (2. for massless particles whose interactions rcspoxt the symmetry or space inversion Il is customary to treat all the components of an irreducible representation of the inhomogeneous Lorentz group mcluding space inven>ion as a single particle type. the label (J would have to be allowed to include continuous as well as discrete labels. It is not an elementary particle.o ~ APppl'U(A)'I'p.'P) that are either identical.o must be a linear combination of the statevectors '¥ /\. <liITerent particle species may correspond to representations that are isomorphic. (2.a to choose the (J labels in such a way that the matrix Co'o(A. in the sense that it cannot be further decomposed in this way: Our task now is to work out the structure of the • Of course. or identical up 10 a similarity transrormalion. (2. it may be possible by using suitable linear combinations of the '¥p.o = pl''¥p. that the label (J is purely discrete.u = e1p·(. we thus consider statevectors '¥ p.U(A) is to produce an eigenvector of the fourmomentum with eigenvalue Ap PPU(A)'I'"o ~ UtA) [U'(A)PPUIA)] '1'.. Introducing a label (J to denote all other degrees of freedom. We must now consider how these states transform under homogeneous Lorentz transformations.26) tell us how the states '¥ p.
For this purpose. for each value of p2. (2. for any elements W.o = N(p) U(AL(p»'I'kp ~ N(p) U(L(Ap» U(L 1 (Ap)AL(p)) 'l'k. we can choose a 'standard' fourmomentum.9) .5.' ~ U(W) L Do'o(W)'I'V' ~ L Do'o(W)Do'o"(W)'I'v 17" 17' Il" and so Do'"(W W) ~ L D"'o"(W)D". and express any plJ.5. and then to Ap.6) The point of this last step is that the Lorentz transformation L 1(Ap)AL(p) takes k to L(p)k = p. of this class as (2. and then back to k. we have said nothing about how the a labels are related for different momenta. (2.5) where N(p) is a numerical normalization factor.64 2 Relativistic Quantum Mechanics coefficients Ca'AA.7) This subgroup is called the little group.. we now find U(A)'I'.4) where Lll v is some standard Lorentz transformation that depends on pJi. ". that are left invariant by all proper orthochronous Lorentz transformations Ail v are the invariant square pl := '7!.p ~ U(W)U(W)'I'kp . We can then define the states l.vp1tpV. note that the only functions of plJ. to be chosen later. W we have LDo'"(WW)'I'v ~ U(WW)'I'.• . and also implicitly on our choice of the standard kJ<. also the sign of pO.J1p .5. and for p2 < 0.' (2. so it belongs to the subgroup of the homogeneous Lorentz group consisting of Lorentz transformations WJ\ that leave kJ1 invariant: W>\. Operating on (2.. p) in irreducible representations of the inhomogeneous Lorentz group. Up to this point.o . say k/<.8) 0' The coefficients D(W) furnish a representation of the little group. Hence.5.5) with an arbitrary homogeneous Lorentz transformation U(A).s For any W satisfying Eq. and (for p2 < 0) each sign of pO.5.a of momentum p by '1". that is.o(W). Eq.7).5.k\' = kJl • (2. (25.0 .5) now fills that gap. we have (2.5.5. (2.N(p) U(L(p»'I'k.
) This has the immediate consequence that the representation of the little group in Eqs. (c).2. Of these six classes of four~momenta..5.5.5.5. (2. there would have to be an infinite number of them.5 OneParticle States 65 In particular. 6 Table 2.p)) U (L(Ap)) 'PV "' or. By the usual orthonormalization procedure of quantum mechanics. In what follows we will consider only cases (a) and (c)." ~ (~~. we may apply Eq.5.8) and (2. what about scalar products for arbitrary momenta? Using the unitarity of the operator U(A) in Eqs. (2." ~ N(p) L: D". (2. This is a good place to pause. I) for p2 > 0 and pi' = 0 have no nontrivial finitedimensional unitary representations.11). the problem of detennining the coefficients Cala in the transfonnation rule (2.p) =L1(Ap)AL(p) ' (2. respectively.5.a' are eigenstates of a Hennitian operator with eigenvalues k and k'. (2.5) and (2.3) has been reduced to the problem of finding the representations of the little group. we may choose the states with standard momentum k ll to be orthononnal. and say something about the normalization of these states..1) and SO(3.. it describes the vacuum. we find for the •• The little groups 50(2.8) to the littlegroup transformation W(A. of deriving representations of a group like the inhomogeneous Lorentz group from the representations of a little group.5. so if there were any state~ with a given momentum pi' with p2> 0 or p~ = 0 that transform nontrivially under the little group.11) must be unitary" Dt(W) ~ D1(W). (2.pll = 0.13) Now.5): U(A)'P p .)) ~D"'"(W(A.10) and then Eq.5.P))'PAP..11) Apart from the question of nonnalization. only (a). is called the method of induced representations." (W(A.5. which is simply left invariant by U(A). respectively. This approach.". Not much needs to be said here about case (f) .6) takes the form U(A)'P P. .12) (The delta function appears here because \{Jk. recalling the definition (2.1 gives a convenient choice of the standard momentum k ll and the corresponding little group for the various classes of fourmomenta.5. and (f) have any known interpretations in terms of physical states.5. in the sense that (2. which cover particles of mass M > 0 and mass zero.and \{Jk'.5. (2.
because rotations are the only proper orthochronous Lorentz transformations that leave at rest a particle with zero momentum. Standard momenta and the corresponding little group for various classes of fourmomenta.1) and 80(3. and so the scalar product is (2. Standard kll (a) p2=_M2 <O. consisting of rotations and translations in two dimensions. For p' = p. Here K is an arbitrary positive energy. the delta function b 3(k .O.k) 3 where k' ..po>O (b) p2=M2 <O.and (3 + 1Idimensions.L I(p)p'. The group 180(2) is the group of Euclidean geometry. say I cY.0.p')) :0' 6 (k' .14) It remains to work out the proportionality factor relating b 3(k . Since also k = L l(p)p.pi).0) ~ N(p)N'(p')D (W(L '(p). while 50(2.66 2 Relativistic Quantum Mechanics Table 2.1.5. '1".M) (c) p' ~ O.K) (O.1) SO(3.0" '1'.'.. M) (O.1) are the Lorentz groups in (2 + 1). Note that the Lorentzinvariant integral of an arbitrary scalar function f(p} over four~momenta with _p2 = M 2 > a and pO > 0 (i.k'} is proportional to (53(p . The little groups are mostly pretty obvious: SO(3) is the ordinary rotation group in three dimensions (excluding space inversions). K) (0. K.N. K. the littlegroup transformation here is trivial.e. respectively.O. W(L1(p). 0.pO<O (0. 0) scalar product: ('1'.p) = 1.0.O) (e) p2 = N 2 > (f) pI' = 0 a (0.0) ~ N(p) (V' (L(p))'I'p'P'.rP<O Little Group SO(3) SO(3) 1SO(2) IS0(2) SO(2. ()3(p _ .0.k'} and p').0. 0.O.po > 0 (d) p2=O.1) (0. Its appearance as the little group for l = 0 is explained below.
po) d3 f(p.p) ~ kOP(k' ." (f:) b (p' .16) so we see that the invariant delta function is Since pi and p are related to k ' and k respectively by a Lorentz transformation.15) F(p)~ J ~J F(p'p'(p_p')d 3p' F(p') [Jp'2 + M2b3(p' _ P)] ~"dJ~p'~ Vp'2 + M2 (25.) We see that when integrating on the 'mass shell' p2 + M 2 = 0..p2 . but then we would need to keep track of the pO /ko factor in scalar products.2. J p2 + M2) P 2Jp'+M2 (e(po) is the step function: O(x) = 1 for x > 0.5. I will here adopt the more usual convention that N(p) ~ JkO/pO for which (2. Instead. . L(p). e(x) = 0 for x < 0.5.18) (2.5.k) and therefore ('1'.M 2 )O(po)f(p. and particles of zero mass.5 OneParticle States 67 cases (a) or (c)) may be written J d"pb(p 2 + M 2)I!(po)f(p) ~ ~ J J d 3 pd po b((pO)2 .19) We now consider the two cases of physical interest: particles of mass M > 0."" '1'.17) The normalization factor N(p) is sometimes chosen to be just N(p) = 1.") ~ IN(p)1 2b". we have then p'p(p' . 3 (2.'.5.p). the invariant volume element is d3p/Jp2 + M2 The delta function is defined by (2.
the Wigner rotation W(A../lpl. ~ LO.24) where p.5.O. 1. This is conveniently chosen as L\(p) ~ 0. note that the boost (2..5.(p) ~ p. we need to choose a 'standard boost' L(p) which carries the fourmomentum from kl' = (O.11) now becomes U(A)'!'"o ~ V(AP)O" ° P L.5.1.O" 0' (j) (2.O..!Jl. where R(p) is a rotation (to be defined in a standard way below.24) may be expressed as L(p) ~ R(p)B(lpl)W1(p).Vy2 "1.10): W(A. These can be built up from the standard matrices for infinitesimal rotations R<k = lJ.D". + (. For a particle of mass M > 0 and spin j.5. . (2. To calculate this rotation. (2. Its unitary representations can be broken up into a direct sum of irreducible unitary representations 7 D~). To see this.  p. ~.5.5.p) (the Wigner rotation 5 ) given by Eq.5.j. . in Eq.o(p) LOo(p) ~ Y.68 2 Relativistic Quantum Mechanics Mass PositiveD£ifinite The little group here is the threedimensional rotation group.j .p))'!'AP. '.(W(A.p) ~ L l(Ap)AL(p).21) (2.M) to pJi. " ..(R) of dimensionality 2)+1. . It is very important that when N'v is an arbitrary three~dimensional rotation . L.k + Bib with 8 ik = Ehi infinitesimal: (2. with j = 0.22) with (J running over the values j.5. (2.20) (2. Eq. liM>.p) is the same as at for all p.23) with the littlegroup element W(A.
1(p) and hence as was to be shown. ClebschGordan coefficients. (Wt)Jtki' = tilki' = I .47)) that takes the threeaxis into the direction of p. k ll = (0.0. This is another piece of good news .p) ~ R(81p)B.)"~(".5 OneParticle States 69 (2.1 (lpl)R(8)B(lpI)R. etc.(. Acting on a timelike fourvector ti' = (0.the whole apparatus of spherical hannonics.1).2. and then into the direction :Jtp.0. "' 0 0 1 Then for an arbitrary rotation :Jt But the rotation R1(&p):Jt R(p) takes the threeaxis into the direction p.0. multiparticle states) have the same transfonnation under rotations as in nonrelativistic quantum mechanics. I. and then back to the threeaxis.1(p) ~ R(81p)R(8)R. this now gives °° W(81. with WJt"k" = k ll . Mass Zero First. by extension. such a Lorentz transformation must yield a fourvector Wt whose length and scalar product with Wk = k are the same as those of t: (Wt)I/(Wt)jl = ti'ti' = I. can be carried over wholesale from nonrelativistic to relativistic quantum mechanics. Thus states of a moving massive particle (and. where ki' is the standard fourmomentum for this case. I).5. Consider an arbitrary littlegroup element Wi\. and B(lpl) ~ 1 0 0 0 0 1 0 0 0 0 y VI' I Vy2 . so it must be just a rotation by some angle e around the three·axis cosO sinO 0 0 sinO cosO o 0 1 0 o 0 0 1 Since R(O) commutes with B(lpl).. we have to work out the structure of the little group. Any fourvector that satisfies the second condition may be written (W.p.l+n .
that is.2 + p2)/2.30) R(ii)R(O) ~ R(ii These subgroups are Abelian . laa.25) It follows that the effect of WJi v on transformation s".P)W ~ R(O).5. so Sl(ex. As we have . and for this reason are called semisimple.5. their elements all commute with each other.29){2.5.26) . (2. the subgroup with () = 0 is invariant. The reader will recognize these multiplication rules as those of the group /SO(2).5.5.P + P) + 0). sin 0 + P cos 0) .0. in the sense that its elements are transformed into other elements of the same subgroup by any member of the group R(O)S(" P)R1(O) ~ S(.rnW must be a rotation by some angle () around the threeaxis S'('.p)S(" P) ~ S('  + '.27) where R".. SI'y like Wllv leaves the lightlike fourvector (0.31) From Eqs.1. Groups that do not have invariant Abelian subgroups have certain simple properties.31) we can work out the product of any group elements.p)W is a Lorentz transformation that leaves the timelike fourvector (0. (2.5.1) invariant. .1) invariant. Also.sin () sinO 0 0 o o cosO 0 0 o 1 0 o 0 1 (2.  (2.0. P). o 1 P 1( p P ( 1H (2. but it does mean that Sl(a.P) ~ S('.0.70 2 Relativistic Quantum Mechanics and the first condition then yields the relation ( ~ (.5. cos Ii + P sin 0.5.(O) = cos (1 .28) The most general element of the little group is therefore of the form W(O".29) (2. What group is this? We note that the transformations with () = 0 or with ct = p = 0 form subgroups: S(a. . Furthermore.fJ)) and rotations (by an angle ()) in two dimensions.("P)~ is the same as that of the Lorentz a. (2. P ( This does not mean that W equals Stet. and is therefore a pure rotation. tV (2.P)R(O). consisting of translations (by a vector (ct.5.
2. the little group ISO(2) like the inhomogeneous Lorentz group is not semisimple..b .{J)J\ = (jJI.5.(2. (2. (2.5.3).18)(2.B sin 8. For e. Since A and B are commuting Hermitian operators they (like the momentum generators of the inhomogeneous Lorentz group) can be simultaneously diagonalized by states 'P k•a.33) (2.5.b A'P k•a•b = a'P k •a•b B'P k .VI (R(f1» 'P k .a. U[R(O)]BUI[R(O)] ~Asin8+Bcos8.a.". B] ~ ~ (2. Either from (2.31). = blfk.b .a.29) .cx.fJ infinitesimal.34) and.5.P)) ~ 1 + i. ~r:x. The problem is that if we find one such set of nonzero eigenvalues of A. we see then that the corresponding Hilbert space operator is U( W(8".4.20). a. First. we see that these generators have the commutators [hAJ ~ +iB . .37) iA. [hBJ [A. f/.rx.b . or directly from Eqs.35) (2.5.5 OneParticle States 71 seen. p From (2. and this leads to interesting complications.h . A 'P~. From Eq.36) (2.31). 0. as before.a.32) where A and B are the Hermitian operators A=J 13 +J 1 0=1}+K 1 .5.b .4. where 'Pr. and so. then we find a whole continuum.B.h . (2. for arbitrary fl.5. h = JI2.b = (a sin 0 + h cos O)lff. (2. o 0 win' = (1 0 0 {J p p 0 0 0 0 .l [R(8)J ~ A cosO . let's take a look at the Lie algebra of ISO(2). the general group element is W(O.5.A + iPB + iOf) .4. we have U[R(81]A U. B'PL.a) = (a cos 0 ~ h sin O)IfZ. . + wi\.5. B = _J23 +J 20 = J) +K 2 .
" ~ I(AP)O pO . As we shall see in Section 2.) are eigenvectors of A and B with . just as for massive particles.(W) and therefore Eq.o . (2.p) defined by W(A. The Lorentz transfonnation rule for a massless particle of arbitrary he1icity is now given by Eqs. (2. (2. so that U(W)\{Ik. (2.p)) .. (2. (2.5. or helicity.5..11) and (2. to avoid such a continuum of states." (2..5.5.41) An arbitrary element W of the little group can be put in the form (2.39) Since the momentum k is in the threedirection.5. and p to (2. there are topological considerations that restrict the allowed values of (J to integers and halfintegers. p)) R (B(A. At this point we have not yet encountered any reason that would forbid the helicity a of a massless particle from being an arbitrary real number.. (2." ~ 0.28).5.1T Da. (2. exp ('60(A.32) generalizes for finite ':1..38) These states are then distinguished by the eigenvalue of the remaining generator 1}\{Ik.40) and for finite () to U(R(O)) ~ exp(ihO). 17 gives the component of angular momentum in the direction of motion. = a\{Ik.7.42) with O(A.8) gives = exp(i86)b{rI{T' where () is the angle defined by expressing W as in Eq. p)..18) as U(A)'I'.p)) 'I'A. a = b = 0: A'I'k.2.28).43) We shall see in Section 5. We are now in a position to calculate the Lorentz transformation properties of general massless particle states. p) =L 1(Ap)AL(p) _ S (a(A..<7 = exp(iOl1)\{Ik.72 2 Relativistic Quantum Mechanics Massless particles are not observed to have any continuous degree of freedom like 0.u ~ B'I'.5.9 that electromagnetic gauge invariance arises from the part of the little group parameterized by rt and p. P(A. First note that by use of the general arguments of Section 2.5.5. we must require that physical states (now called \Ilk".5.5. . Eq.5.O" = exp(icr:A + if3B) exp(i81))\IIk.
particles of opposite helicity are related by the symmetry of space inversion. Thus.0. K) to pll.47) where 0 < e < n. the massless particles of he1icity +1 associated with electromagnetic phenomena are both called photons.1) into (sin 8.5.45) and R(p) is a pure rotation that carries the threeaxis into the direction of the unit vector p.5.5.47) is a rotation.) Since (2.5. together with a specification of the range of ¢ and e. suppose we take p to have polar and azimuthal angles 8 and 4>: p= (sin8cos¢.5. (and also to enable us to calculate the effect of space or time inversion on these states in the next section) we need to fix a convention for the standard Lorentz transformation that takes us from kll = (0. and does take the threeaxis into the direction (2. 0. Note that the helicity is Lorentzinvariant.44) where B(u) is a pure boost along the threedirection: B(u) = 1 0 o 1 o o 0 o o 0 0 (u 2 + 1)/2u (u 2 . corresponding to a mere redefinition of the phase of the oneparticle states. cos e).46) Then we can take R(p) as a rotation by angle () around the twoaxis. which takes (0. For instance. cos e). but a different sign for U(R(p)) when acting on halfinteger spin states. (2. sin8sin4>. any other choice of such an R(p) would differ from this one by at most an initial rotation around the threeaxis.5. as we shall see in the next section. we would be justified in thinking of massless particles of each different helicity as different species of particles.46). because shifting o or ¢ by 2n would give the same rotation R(p). K.). and the massless particles of he1icity +2 that are believed to be associated with gravitation are both called gravitons. This may conveniently be chosen to have the form L(p) ~ R(p)B(lpl/K) (2. (2. (We give U(R(p)) rather than R(p).5.0.5 OneParticle States 73 To calculate the littlegroup element (2. 0 < ¢ < 2n. However.1)/2u 0 (u 2 1)/2u (u 2 + 1)/2u (2.43) for a given A and p. because electromagnetic and gravitational forces obey space inversion symmetry. followed by a rotation by angle ¢ around the threeaxis.2. the supposedly massless particles of helicity +1/2 that are emitted in nuclear beta decay have no interactions (apart from gravitation) that . U(R(p)) ~ exp(i¢1) exp(ie]. a massless particle of a given helicity (1 looks the same (aside from its momentum) in all inertial frames. Indeed. On the other hand.
5. 2. a state formed as a linear superposition of oneparticle states with opposite helicities will be changed by a Lorentz transformation into a different superposition. for linear polarizations with a. ::1 I' . and for linear polarization may be adjusted so that a_ = a~..42) shows that under a Lorentz transformation 1\11. Plane polarized gravitons can be defined in a similar way.42).+1 + ·'1.a) ~ U(AA. Circular polarization is the limiting case where either:::r:+ or L vanishes..5.. and linear polarization is the opposite extreme. The overall phase of li+ and 17. with 10:+1 = ILl.Aa + a) . In particular. (2.p). 0 0 I 0 0 0 1 0 0 0 0 0 I 0 0 1 It used to be thought self·evident that the fundamental multiplication rule of the Poincare group U(A. where 1>+1' + ILl' ~ 1 ._'f p. but the relative phase is still important.p). a general onephoton state of four·momenta may be written '¥p.?J or Y or :!i':Y. the phase of 0:+ may be identified as the angle between the plane of polarization and some fixed reference direction perpendicular to p. For instance.= a~. DetA = +1 and ADO > +1) or else equal to a proper orthochronous transformation times either . the state itself is not. and antineutrinos for helicity 1/2. this angle rotates by an amount 8(A.42) has the consequence that a Lorentz transformation A rotates the plane of polarization by an angle 20(A.has no physical significance.3 that any homogeneous Lorentz transformation is either proper and orthochronous (i.e.74 2 Relativistic Quantum Mechanics respect the symmetry of space inversion.6 Space Inversion and TimeReversal We saw in Section 2. with 1:::r:±1 both nonzero and unequal.?J and iT are the space inversion and timereversal transformations 1 0 0 0 1 0 0 0 0 1 0 0 0 1 0 0 .?J1\ = . and here Eq. Even though the helicity of a massless particle is Lorentzinvariant. (2.a) U(A. where . Indeed.o:=o:+'¥p. so these particles are given different names: neutrinos for helicity + 1/2.5.. The generic case is one of elliptic polarization. because of the helicitydependent phase factor exp(i!T e) in Eq. Eq. (2._l.
we obtain the P and T transformation properties of the Poincare generators PiJP<rp. (2. (2. with (1)11" = 0)\. where H _ pO is the energy operator.1 v <rJIIV.4) gives PiHP.9).2).6.. and equating coefficients of O)p<r and E" p in Eqs.1. TU(A. (2.'.'.6. so PHP. i.6.11 and E" J1 both infinitesimal.1) and (2.6.3).6. But then for any state If' of energy E > 0. Timereversal survived for a while.6. but in 1964 there appeared indirect evidence9 that these properties of T are also only approximately satisfied.1 = iH . (2.6) This is much like Eqs.8) and (2.4. TiJf'<l"T. Setting p = 0 in Eq. (2.1 ) (2. it was believed th<lt there are operators corresponding to :JI' and i:J themselves: P such that U (. there would have to be another state pl1f' of . The decision is an easy one.5) TipPT.0) PU (A. aJPI ~ U (:3'A:3'1 . (See Section 3.6.1 = H.6.2) for any proper orthochronous Lorentz transformation All v and translation all.3.tra) (2. (2.4. (2.1.'. These transformation rules incorporate most of what is meant when we say that P or T are 'conserved'.1) and (2.2) actually exist.1 =i.1/pll.1A'. we will make believe that operators P and T satisfying Eqs.1 = i:!l'/:Yv'J JIl" .1) and (2.6. such as those that produce nuclear beta decay.0) T _ U (.>.2.6. except that we have not cancelled factors of i on both sides of these equations. In particular. Let us apply Egs.4.6 Space Inversion and TimeReversal 75 would be valid even if A and/or A involved factors of:!}J or tr or :?fJ::T.1/<'. If P were antiunitary and antilinear then it would anticommute with i.' .6. because at this point we have not yet decided whether P and T are linear and unitary or antilinear and anti unitary.1 = i.a)T. (2.1 = ifjlJ/Pll.6.3) PiP P P.) In what follows.1 = U(.:3'a) .2) in the case of an infinitesimal transformation.e. In 195657 it became understood S that this is true for P only in the approximation in which one ignores the effects of weak interactions.6.g. Using (2.4) (2. hut it should be kept in mind that this is only <In <lpproximation.
11) (2. Note by the way that Eq. There are no states of negative energy (energy less than that of the vacuum).24).4.8) ppp. (2.7)(2. M.6. To avoid this.4.18H2. we can conveniently rewrite Eqs.13). because after timereversal an observer will see all bodies spinning in the opposite direction. From Eqs.7).6.10) (2.6. we see that the same must be true of the state P'I'k.6. setting p = 0 in Eq. TPT.6. and. Now that we have decided that P is linear and T is antilinear.6.1 = H . (2.6.6. because T reverses not only J. The reader can easily check that Eqs.9). (2.6) yields TiHT. because at least the orbital part is a vector product r x p of two vectors.H.9) (2.10) is consistent with the angularmomentum commutation relations J x J = iJ.6.76 2 Relativistic Quantum Mechanics energy E < O. and h with eigenvalues 0. and find THT.1 = THT. and (2.t1 = I}rr'Pk.6. PHP.1 =p .6. (2. If we supposed that T is linear and unitary then we could simply cancel the is.13) It is physically sensible that P should preserve the sign of J.1 = ~J . but also i. as shown before.1 = K.1 ~ +J. and commutes rather than anticommutes with H.rr are defined as eigenvectors of P. Let us now consider what P and T do to oneparticle states: P:M>O The oneparticle states 'Pk.6. we are forced here to conclude that T is antilinear and antiunitary. so we are forced to choose the other alternative: P is linear and unitary.1 ~ p. respectively.6) in terms of the generators (2.6. and cr. .13) are consistent with all the commutation relations (2. (2. and therefore (barring degeneracies) these states can only differ by a phase P'Pk . T reverses J.6. (2. TKr l ~+K.7) PKP. with the again disastrous conclusion that for any state 'P of energy E there is another state Thp of energy E.3)(2.6. (2. (2. On the other hand.4.15)~(2.6. both of which change sign under an inversion of the spatial coordinate system.1 = H.1 =iH.rr with a phase factor (II}I = 1) that mayor may not depend on the spin cr.4.rr.17) in a threedimensional notation PJP. (2.12) TJT. On the other hand. (2.
known as the intrinsic parity." (2. and (2.6. Applying the operator T to (2. we note from (2. O(T'I'k." ~ 1M/pO U(L(p»'I'k. we find '1(1 = '1(1±1 and so '1(1 is actually independent of a.14).(1 is to yield a state with P(T'I'k.o 1]\fI &. Operating on both sides with P. and recalling that T anticommutes with J and i.15).o) H(T'I'k.24): '1'. so using Eqs.(J = (2. we must apply the unitary operator U(L(p)) corresponding to the 'boost' (2.6.20). To get to finite momentum states.6.5.o)..1) and (2.o ~ 1M/pO or in other words PIf' p. Jp2 + M2) U(L(iJ'p))~'I'k. we find (J1 + ihKulf'k. (2. • • where ((1 is a phase factor. and (2. that depends only on the species of particle on which Pacts. (2.6.12).15) with '1 a phase.6. (2.5.10). we see that the squareToot factors cancel.16) T:M>O From Eqs. we see that the effect ofT on the zeromomentum oneparticle state 'P k .o).6.5.21) that (2.8).5.13).o ~ ~'I'k. and so T\fI k (1 = ((1'Pk (1. we have P'I'p.14) where j is the particle's spin. M(T'I'k.6 Space Inversion and TimeReversal 77 To see that 1]u is aindependent.2.14) again on the left.(1 .(1 = V(j + (1)(j + r1 + 1)(".6.6.p.o) 1)(T'I'k.gi>L(p)!r 1 ~ L(iJ'p) iJ'p ~ (_p. (2. We therefore write P'I'.o) ~ ~ ~ 0. and so .6.6. Using Eq.±1 'fk.o' We note that .a:j:J . (2..
k.a.6.78 2 Relativistic Quantum Mechanics We write the solution as (11 = (_)Ja. where R2 is a rotation that also takes k to 2pk.17). just to keep open our options in choosing the phase of the oneparticle states. with' another phase that depends only on the species of particle.jp2+M2) (That is.6. (2. the parity operator P yields a state with fourmomentum (. conveniently chosen as a rotation by 180° around the twoaxis U(R2) ~ e<p(inJ.K. Note th. .tr k. that is defined as an eigenvector of pi' with eigenvalue k ll = (O.11 .O.?f'k)J1 = (O.lf (2.K) and h equal to cr. we have then T'Pp.11 =  dj2'P 'P'k.19) Since U(Ri 1) reverses the sign of J). k.1/2"(_)J"'P. In what follows we will keep the arbitrary phase' in Eq.. This is because we can redefine the oneparticle states by a change of phase 'Pk. we again apply the 'boost' (2.6.*lj2 T 'Pk.) Using Eqs. it is convenient to consider instead the operator U(R Z1 )P.5. changing the sign of each element of U· \' with <In odd number of timeindices is the same as changing the signs of elements with an odd number of spaceindices. ~ ()J"'P' I. To deal with states of finite momentum.20) .rr" = .G" = (()J0'P. As menlioned earlier.5).6. Thus it takes a state of helicity (the component of spin along the direction of motion) (J' into one of helicity cr. the timereversal phase' has no physical significance. (2.6.5.O.rr . (2.24). but it should be kept in mind that this phase is of no real importance. (2.).17) However.t> = I.".2) and (2. in such a way that the phase' is eliminated from the transformation rule I.11 T'P'k._G"' (2.6.K) and an eigenvector of h with eigenvalue (J'.'iPp. this shows that the existence of a spaceinversion symmetry requires that any species of massless particle with nonzero helicity must be accompanied with another of opposite helicity. Because P does not leave the standard momentum invariant.K. Yp~ (_p.G" = '1 lf 'Pk .t :T L(p ):r 1 ~ L(:!J'p) . unlike the 'intrinsic parity' 11. we have U(Ri1)P'Pk.18) P:M=O Acting on a state 'l'k.
This peculiar change of sign in the operation of parity for massless particles of halfinteger spin is due to the convention adopted in Eg. ..22) with the phase nO" or +nO" according to whether the two·component of p is positive or negative.n according to whether a < rjJ < n or n < rjJ < 2n.e)1. helicity J .6.'1'p. k.?J?p) .rJ' commutes with the Lorentz 'boost' (2.YJp = (p.a (2.47) for the rotation used to define massless particle states of arbitrary momentum.5.I (R(pl)U(R(plR2) ~exp( i(ne)1. which has values k Jl = (O. (2. But a rotation of + 1800 around the three·axis reverses the sign of h.) x exp ( . exp (i(n .6 Space Inversion and Time·Reversal 79 with 1/u a phase factor.O.5. some discontinuity of this sort is unavoidable. so by operating on (2. R2 1. but U(R(p)R2) is not quite equal to U(R(p)). so that it remains in the range of 0 to 2n.21) Also. 10 (2. Then U.K) and 0" for PJl and h.K.u. respectively. Because the rotation group is not simply connected.47).5.2.5) with P.45). so U(R(plR z) ~ U(R(pl) exp(+in1)l. Now. and . pO). we find for a general four. R(p)B(lpl/k) is just the standard boost L(. T:M~O Acting on the state \f'k. According to (2. and by itself has nothing to say about whether massless particles of one helicity 0" are accompanied with others of helicity 0"._" Note that R(p)R2 is a rotation that takes the three·axis into the direction of ~p.6.K) and 0" for PJl and l}. Thus T does not change the .O. Because T like P does not leave the standard fourmomentum k invariant.i(rjJ + n)1)) exp(irjJ1)) exp(i81}) exp(inh) =exp( i(n8lh)exp(+inl)lexp( i(n8)h).u = the direction lJu exp( +inO" )'t'. We have then finally p\f' p. the timereversal operator T yields a state which has values (&'k)fI = (O.) with azimuthal angle chosen as rjJ + n or rjJ .K.rJ' commutes with the rotation R(P) which takes the three· direction into the direction of p.5.momentum pit ~~ry"U(R(PJR2BC:I)) U(R( pl) ~ exp (i(¢ + nl1)) '!'.
operating with T on the state (2.26) We get the same result for massless particles. where R2 is the rotation (2. so U(R.25) Again.. and :Y commutes with the rotation R(P). This result has an interesting consequence.5) gIves T'I'". so Eq.'i'p." . ~ C(ia(()j+rrqJp.6. it yields a factor (."'1'. (2. which also takes k into 2Pk. so Eq.45). ~ ~ U [R(P)R2 B C~I)] .o = (" (2.<r ••• 2 = T(()ja\{Jg>p. This commutes with J3."'1".6.6. and recalling that T is antiunitary. If the twoc9mponent of p is positive then the twocomponent of @lp is negative. (2.6.<1' Gexp( +inaK".21). (2. this yields finally exp(+ina)\{I.80 2 Relativistic Quantum Mechanics it is convenient to consider the generator U(R2"l )T. It is interesting that the square T of the timereversal operator has a very simple action on both massive and massless oneparticle states. we get an overall change of sign (2.6. either massive or massless.24) Using Eq. (2. this can be written "1'" T2". (_)1'1'.6.6. Tp. exp( +ina)\{Jp..26).<1 ." ~ . we usually mean the absolute value of the helicity.a or in other words ' .6.6. (2. we see that for massive oneparticle states: T 2 \{Jp. Using Eq. respectively." (2.<1 = = exp(+2ina)\{Jp.')T'I'k.a = T(<f exp( +ina)\{Jf}p.5.6..19).27) is the same as Eq. (2. Since Ri1y commutes with the boost (2.18)._a = T2'1'".23) with (0 another phase. When T2 acts on any state 'P of a system of noninteracting particles.6.a (2627) •• By the 'spin' of a massless particle.25) gives T 2\{Jp. and viceversa.<r' (2. Hence if the state "' contains an odd number of particles of halfinteger spin or helicity (plus any number of particles of integer spin or helicity). the top or bottom sign applies according to whether the twocomponent of p is positive or negative..5.a As long as a is an integer or halfinteger.28) . T'f'p.)2 j or (_)2 11 for each particle. ~ (_)2 1 T p .
as long as these fields are invariant under T. The surprising result is that at least a two·fold degeneracy persists even if rotational invariance is perturbed by external fields. Since T commutes with the Hamiltonian.7 Projective Representations· We now return to the possibility mentioned in Section 2. then T\{I can differ from \{I only by a phase T'P but then ~ ('P . We see that any energy eigenstate \{I satisfying Eg. so such dipole moments are forbidden by timereversal invariance. if any particle had an electric or gravitational dipole moment then the degeneracy among its 2j + 1 spin states would be entirely removed in a static electric or gravitational field.2. these arguments will apply even if our system is subjected to arbitrary static gravitational and electric fields. 2..28). it should be mentioned that P and T can have more complicated effects on multiplets of particles with the same mass. t.7 Projective Representations 81 If we now 'turn on' various interactions. This possibility will be considered in Appendix C of this chapter. even if they do not respect rotational invariance. because the total angularmomentum j of any state of this system would have to be a halfinteger. . (2.28) must be degenerate with another eigenstate of the same energy. U(T). this conclusion is trivial if the system is in a rotationally invariant environment. (For instance. such as electrostatic fields. that a group of symmetries may be represented projectively on physical states. and there would therefore be 2j + 1 = 2. of the symmetry group' may be represented on the physical Hilbert space by unitary operators U(T). which • This section lies somewhat out of the book's main line of development.6.6. etc. Is it the same state? If so. (2.'IO Of course. in contradiction with Eg.2. suppose that \{I is an eigenstate of the Hamiltonian. This is known as a 'Kramers degeneracy.) Now. and may be omitted in a first reading. provided these interactions respect invariance under timereversal.'" degenerate states. In particular. the elements T. etc. For the sake of completeness. this result will be preserved. that is. T\{I will also be an eigenstate of the Hamiltonian. No physically relevant examples are known.4.
and that differ only by functions drjJ(T.1) with rjJ a real phase. We are interested here in whether a symmetry group allows any nontrivial twococyles.2). in the sense that the phase rjJ(T. but a projective representation with such a phase can be replaced with an ordinary representation by replacing U(T) with U(T) U(T)exp(b(T)) for which U(T)U(t) ~ U(Tt) Any set of functions rjJ(T. is called a 'two·cocycle'. which can be eliminated by a redefinition of U(T).7. Using coordinates ()iJ to parameterize group elements (as in Section 2. T(O)) around () = = 0 must start with terms of order HO: e = (2.7. A trivial cocyc1e is one that contains the function ¢ = 0. that is.) The basic requirement that any phase rjJ in Eq.a(T) .7. (2.7. and hence consists of functions of the form (2.7.2) Of course.4) tells us that the expansion of ¢(T(O).1) would have to satisfy is the associativity condition U(TJ)(U(T2)U(TJl) ~ (U(T3)U(T. (A bar is used here just to distinguish one symmetry operator from another.3) will automatically satisfy Eq. In order to answer this question.a(f) (2. the phase ¢ must clearly vanish ¢(T. I) ~ ¢(1.3). T) that satisfy Eq.7. which imposes on rjJ the corresponding condition (2. T) of the form (2. with T(O) 1.7.4) When both T and T are near the identity the phase must be small.7.7. f) ~ O.82 2 Relativistic Quantum Mechanics satisfy the composition rule U(T)U(f) ~ exp (i¢(T. whether it may have a representation on the physical Hilbert space that is intrinsically projective.2).))U(TJl. f) ~ a(Tf) . (2. When either T or T is the identity. f») U(Tf) (2. (2. (2. any phase of the form ¢(T.7. T) cannot be eliminated in this way.3).7. (2.1) on the commutation relations of the generators of infinitesimal transformations.5) .2). Eq. it is useful first to consider the effect of a phase ¢ in Eq. (2.7.
7. d.7. (2. (2.7) The appearance of terms on the righthand side of the commutation relation proportional to the unit element (socalled central charges) is the counterpart for the Lie algebra of the presence of phases in a projective representation of a group.7. and so (2.8) and also (2.tc] = iCabeta . if two conditions are met: (a) The generators of the group in this representation can be redefined (as in Eg.7. and adding the same expressions with b.1).12) A given Lie algebra mayor may not allow solutions of Eq.10). .6) with td.7.6) by a redefinition of the generators (2. (2.7. c.7. For these solutions.1)." 1 + feb· (2.7. C. Taking the commutator of (2. (2. tel = iCabeta Cbe ~ fbe + tCb.2. (2.7 Prqjective Representations 83 where f ub are real numerical constants. The phase of any representation U(T) of a given group can be chosen so that ¢ = 0 in Eg. b. d replaced with c.7. which follows from the Jacobi identity.10) where cPr is an arbitrary set of real constants. (2.22). (2.7.9) always has one obvious class of nonzero solutions for Cab = Cub: CeabcPe. Inserting this expansion in the power series expansion of Eg.11)). the sum of the three doublecommutators on the lefthand side vanishes identically. so as to eliminate all central charges from the Lie algebra. b and d. We can now state the key theorem that governs the occurrence of intrinsically projective representations. we can eliminate the central charges from Eg.7.7. (2.7.7. (2.7.11) The new generators then satisfy the commutation relations without central charges ~b.2.9) other than Eg.6) where Cbe is the antisymmetric coefficient (2. and repeating the steps that led to (2.9) Eg. The constants C bc as well as CUhe are subject to an important constraint. we now have [rb.
4.IlV . (2.1) depends on the particular choice of standard paths.7.1l + C p. (2. Let's now consider each of these possibilities in turn for the special case of the inhomogeneous Lorentz group.. and any two such paths may be continuously transformed into one another. we apply the Jacobi .I' = CJ1.IlI' = _Cl'v.15) (2. jl'O"] = 'I VI' ji'a _ IJllp pa _ "aIJ jPV +1'("jP1l i[PI'.'11'1' P" i[PI\PpJ = Cp. (This corresponds to redefining the phase of the operators U(A. in place of Eqs. It shows that there are just two (not exclusive) ways that intrinsically projective representations may arise: either algebraically.12)(2.14).7.19) .) This theorem is proved in Appendix B of this chapter. leading from the origin to the various group elements.1' = _Cll.4. the commutation relations of the generators of the inhomogeneous Lorentz group would read i[j/'v.pa C P. because the group is not simply connected. and hence a path from 1 to T and then from T to t may not be continuously deformable into some other path from 1 to TT. We will now show that all these constants have additional algebraic properties that allow them to be eliminated by shifting the definitions of jl'\' and pll by constant terms. the phase q. .7. that are used to define the corresponding U operators. _lJllapp + Cpa'll. or topologically. (An equivalent statement is that any loop that starts and ends at the same group element may be shrunk continuously to a point. any two group elements may be connected by a path lying within the group. in Eq. (2. ppJ = '1vp pI' . i. We see that the Cs also satisfy the antisymmetry conditions Cpa.) To derive these properties.7.7.I'V .P .13) (2. because the group is represented projectively even near the identity.7.14) (2..a).pa Cpa. In the latter case.7.e.7. (A) Algebra With central charges.18) (2.84 2 Relativistic Quantum Mechanics (b) The group is simply connected.16) i[jIlV. which also offers comments about the case of groups that are not simply connected.jPO"] = 'IllPpa + Cpa. (2.17) (2.
AJj _1/IlIICP.Il A + I}PACII.7. C PII. (2.20){2.pv _ 1}'lP C Arr .II = 1/IIIlC h1 _ I}AIlCII 17 I} e"v.7. (2.29) (2.22).7.P . Call .28) Also.IIV are not necessarily zero.7. (These expressions automatically satisfy Eqs.7.7 Projective Representations 85 identities [JI".II_ I}IlPC"U.23) with I}. (2."" . [P". (2. CJ1." + 1//:IlCP. but their algebraic structure is simple enough so that they can be eliminated by shifting the definitions of pll and jllV.Il" + 1/11 17 C /:P.{) gives = 0. the constants Ol.13){2. [JP"PPI] + [pc.IlV _ I}f''lCA.27) C J· = .[JJ'.7.I''' .25) with .P 1 o= (2. and hence yields no further information. (2.7.up + I}u}'CJjP _1]11 17 C1.JP'I] + [JI". respectively. 0= '1"PCIlU.) We now see that if the e s are not zero. [PP.II .U (2.25) Contracting Eq.JP'I] ~ 0..24) and (2.pu _ 'I A.A.A.7.'l and CPU. . 1/ vp CIl.AJj _1/IlPC V.17 1} vu CJ1. eM = 1 .J"J] ~o. (2.7.Il" .7.[JI'".P'J] + [p'. (2.32) .7. . I}"p gives CPU.7.IJ" _ 'I flA...7.pu _ 1} vAC/III .21) (2.24) Jj (2. [J".3 1 11 'fp" C p .) Using Eqs.prr + 1} v C IlA. jllU + CIlU . Contracting Eq.II I' +trhCP.P17 + 1/17 A.7. contracting (2.23) (2. (2.31) (2.26) On the other hand.22) (The Jacobi identity involving three Psis automatically satisfied.16) in Eqs.25).30) "P . we obtain algebraic conditions on the Cs + '1IlU CV.U _ '1IlP C''. (2.AJj _ 1/UIlCPI'.7. 0= 1/V P CJ1.7.7.A.7.7.[J". so there is no further information to be gained from the Jacobi identities.).24) with 1]vp gives (2..PPJ] + [pP.7.7.20) (2. [JP'.II CII v.IlII_ I}1I 'CP. [PP. [J'''lJ""JP'J] + [JP'.J"J] ~o.)1I + 1/rr"CPIlJII + 1/1I1l C A" . they can be eliminated by defining new generators pIJ JllU  pll + ell .7.2.2 .JP'J] + [r.
. the full Poincare algebra spanned by }I'" and pI' is not semisimple (the pI' form an invariant Abelian subalgebra). We see that the inhomogeneous Lorentz group satisfies the first of the two conditions needed to rule out intrinsically projective representations. and therefore defines a real fourvector VI'.36) where (J I' are the usual Pauli matrices with ITo 1.86 2 Relativistic Quantum Mechanics and the commutation relations are then what they would be for an ordinary representation i[]IIV.7.35). it is very convenient to represent homogeneous Lorentz transformations by 2 x 2 complex matrices. Indeed.7. but with tildas dropped. Conversely any 2 x 2 Hermitian matrix can be put in this fonn. + 11l1vjPI'. (Semisimple Lie algebras are those that have no 'invariant Abelian' subalgebra. Incidentally.VA = .7. (2.]PV 11VPPI'11VPP". and we needed a special argument to show that its central charges can also be eliminated in this way.34) (2. On the other hand. The property of Hermiticity will be preserved under the transformation v_ A.7. an arbitrary complex 2 x 2 matrix./VP]1'11 _ .33}(2. .4 does allow a central charge. (2. (2.37) with A.]IJ<T] i[]I"'. The commutation relations will always be taken in the form Eqs.7. Furthermore.33) (2.) There is a general theorem ll that any central charges in semisimple Lie algebras may always be removed by a redefinition of generators.7. (2. the covariant . the nonsemisimple Galilean algebra discussed in Section 2.32)./I'p]va _ I}(]!J.i'PJ ~ 0. Any real fourvector VII can be used to construct an Hermitian 2 x 2 matrix (2. consisting of generators that commute with each other and whose commutators with any other generators also belong to the subalgebra.7. How about the second? (B) Topology To explore the topology of the inhomogeneous Lorentz group. the fact that there is no central charge in the algebra of the }I'" could have been immediately inferred from the fact that this algebra is of the type known as 'semisimple'. the mass M.PI'j = = .7.35) iO". as in Eq.
) The group elements depend on 4 . and so correspond to the same Lorentz transformation. is a matrix in S L(2..): (2. C). satisfying Eq. it is easy to see that the matrix produces a Lorentz transformation A(A(8)) which is just a rotation by an angle fI around the threeaxis.7.7.(lVI'(TI). known as SL(2. if. for two such matrices .})t = . = 1.7.t = . C).7..7 Projective Representations 87 square of the fourvector is (2. The Lorentz group is not the same as S L(2. It is therefore convenient to adjust the phase of the AS so that Det.t)}.s that differ only by an overall phase have the same effect on v in Eq.7.. (2. a homogeneous Lorentz transformation A().1. i.1. The 2 x 2 complex matrices with unit determinant form a group. C) is not the same as the Lorentz group.1.1. the same number as the Lorentz group.7.7.7. However. (The S L stands for 'special linear'. while C stands for 'complex'.. (2.40) Furthermore. (2.. (2. but .41) However..)N'v(1.37) provided that IDet 1. then so is I"~ and both A and .39) thus defines a real linear transfonnation of Vil that leaves Eq.42) which is consistent with Eq. = Al'p().) . and hence A = 1 produces a rotation by an angle 2n:.2. two }. 1. SL(2.38) and this determinant is preserved by the transformation (2.38) invariant.37).<)  ~ A('<)A(!.)V "Q"1' t and so AU.7.  (2.37). C).41).39) Each complex 2x2 matrix}. (2. (2.7. Indeed.N\(A)V"(TW1.1 = 3 complex parameters. or 6 real parameters. and }..1 ~ 1.7.l)V1t(TI'(. produce the same Lorentz transformation in Eq. with 'special' denoting a unit determinant. we have ().e.1. (2.
. any complex nonsingular matrix Amay be written in the form _ }" ue . with H an invariant subgroup of G. . the set of pairs of points) of the space of all us and the space of all hs. ifu is unitary then Tr(uvu t ) = TTV. which is the group of complex 2 x 2 matrices with unit determinant.c real but otherwise unconstrained.f.42) requires both exp Trh is real and positive. Any Hermitian traceless 2 x 2 matrix h can be expressed as h~( a + ib c aib) c with a. R3. this decomposition is unique. Detu=l. the threedimensional surface of a spherical ball in flat fourdimensional space. On the other hand.C). so VO = ~Trv is left invariant by A(u). = Since Del u is a phase factor.. when we write GjH. (The factor u simply provides the rotation subgroup of the Lorentz group. identified with }. C) is topologically the same as the direct product R3 x S3.88 2 Relativistic Quantum Mechanics rather*" S L(2. This is simply connected: any curve connecting two points of R3 or S3 can be deformed into any other. what is the topology of the Lorentz group? By the polar decomposition theorem 12 . and Del exph the condition (2. h where u is unitary and h is Hermitian utu = 1. any unitary 2 x 2 matrix with unit determinant can be expressed as U= d + ie ( f+ig f + ig die ) with d.e.b.) Furthermore. so S L(2. C) is topologically just the direct product (i. we mean the group G with elements g and gh identified if g E (J and h f H. and with }.g subject to the single nonlinear constraint d2 + e2 + f2 + g2 = 1. so the space SU(2) of all us is topologically the same as S3. and the same is true of the direct product. so the space of all hs is topologically the same as ordinary threedimensional flat space.7. Now. C)/Z2.e. If h ~ O. (All •• The group 2 2 consists of just elements +1 aml 1. In general. The subgroup 22 is trivially invariant because its elements commute with all elements of SL(2. Thus SL(2.
for instance.) Similarly.44) imposes a superselection rule: we must not mix states of integer and halfinteger spin.  (2.2. or first homotopy group. depending on whether or not they involve an inversion u + u. Identifying A with ).7. where 53 /22 is the threedimensional spherical surface with opposite points of the sphere identified. However. we must have and hence the phase ei. il) ~ +U(AA. they are just the states of integer spin.7. may be continuously contracted to a point. This is not simply connected. (2. and is therefore also doubly connected. In fact. for which these signs are +1 or 1 according to whether the path from 1 to A to AA and then back to 1 is or is not contractible to a point. (2. so the Lorentz group has the topology of R) x 53/22. that goes twice over the same path from any element back to itself. is the same as identifying the unitary factors u and u (because ell is always positive). and the states of half~integer spin. for which the signs in Eqs.7.7.44) These 'representations up to a sign' are familiar. (As discussed in Appendix B. even though these two paths link the same points of 53/22.) However. 53/22 is doubly connected. a path on 53 from u to u' cannot be continuously deformed into a path on 53 from u to u'. this is summarized mathematically in the statement that the fundamental group.p{A. for the inhomogeneous Lorentz group U(A. the inhomogeneous Lorentz group has the same topology as ~ x R3 X S3/Z2.) Thus Eq. Ail + a) . An equivalent statement is that a double loop. (These two cases correspond to the two irreducible representations of the first homotopy group. the limitation to integer or halfinteger spin was pre . and any path in one class can be deformed into another path of that class. of S3/22 is 22.44) are always +1. 22. C). This difference arises because a rotation of 2n around the threeaxis acting on a state with angularmomentum three~component ~ produces a phase e2iltlr . (2. it does have intrinsically projective representations.7 Projective Representations 89 spheres 5n except the circle 51 are simply connected. not 5L(2. the paths between any two points fall into two classes.7.43) or Eq.43) and (2. (2. Because the Lorentz group (homogeneous or inhomogeneous) is not simply connected. we are interested in 5L(2. C)/2 2 . and thus has no effect on a state of integer spin and produces a sign change when acting on a state of half~integer spin. a)U(A. For finite mass. because the double loop that goes twice from I to A to AA and then back to I can be contracted to a point.Aj is just a sign  U(A)U(A) ~ +U(AA) .43) Likewise.7.
and here there is no algebraic reason for a limitation to integer or halfinteger helicity. for zero mass the action of the little group on physical oneparticle states is just a rotation around the momentum. so the only difference is that now the group is simplyconnected. The expanded Lie algebra is then. and it therefore has only ordinary representations. which here are just the angularmomentum matrices . 'Spin(d)'. On the other hand. instead of SL(2. This does not mean that we actually can prepare physical systems in linear combinations of states of integer and halfinteger spin. Ordinary rotation invariance forbids transitions between states of integer and halfinteger total spin. free of central charges. If its Lie algebra involves central charges. There is. where C is a simply connected group known as the 'universal covering group' of G. taking it as SL(2. Instead of working with projective representations and imposing a superselection rule. C) itself. a topological reason: a rotation by an angle 471: around the momentum can be continuously defonned into no rotation at all. not projective representations. Likewise. and H is an invariant subgroupt of C. even though a Lie group G may not be simply connected. and the covering group of the threedimensional rotation group is SU(l).90 2 Relativistic Quantum Mechanics viollsly derived by purely algebraic means from the wellknown repre· sentations of the generators of the little group. it can always be expressed as C/H. of course.4.C). so that we cannot infer a superselection rule.(jl with j integer or halfinteger. This connection with S[~ and SU groups is special to the case of three. the issue of superselection rules is a bit of a red herring. Similar remarks apply to any symmetry group. four. so the factor exp(4nio) must be unity. it mayor it may not be possible to prepare physical systems in arbitrary superpositions 0/ states. then we can always expand the algebra to include generators that commute with anything. we may just as well take the symmetry group as C instead of G. and whose eigenvalues are the central charges. because there is no difference in their consequences. C)/Zl as before. just as we did when we added a mass operator to the Lie algebra of the Galilean group at the end of Section 2. however. We have seen that the covering group of the homogeneous Lorentz group is SL(l. except that G implies a superselection rule. In short. we can just as well expand the Lorentz group. In general. but one cannot settle the question by t The first homotopy group of C/11 is H. while C does not. but only that the observed Lorentz invariance of nature cannot be used to show that such superpositions are impossible. . so the part of the group near the identity has only ordinary representations. and hence (J must be an integer or halfinteger. and does not require any superselection rule. or sh dimensions: for general dimensions d the covering group of SOld) is given a special name.
For our present purposes. For this purpose. [or if there were any nonzero statevector \{I' that was orthogonal to all of the \fI~. say 1't. and therefore ('I'~.4) .A. we single out one of the \{Ik. But (\{I~.) ~ ok!.Appendix A Symmetry Representation Theorem 91 reference to symmetry principles.3) 1'~ It is easy to see that these transformed states also form a complete set. We must now establish a phase convention for the states 1'1. (2. \{I~) is automatically real and positive.'31 t and TfJt2 satisfy (2.2) and let \{I~ be some arbitrary choice of statevectors belonging to the transformed rays T9fk. 'l'IlI' ~ I('l'" 'l'')1 2 ~ Ok!. consider some complete orthonormal set of statevectors 1'k belonging to rays atk. so this requires that it should have the value unity. then the inverse transform of the ray to which \{I' belongs would consist of nonzero statevectors 1''' for which. because whatever one thinks the symmetry group of nature may be.. To start. there is always another group whose consequences are identical except for the absence of superselection rules. with ('l'" 'l'. 'I'll ~ Ok!.2 lfi['l'l + 'I'd (2. From Eq. Appendix A The Symmetry Representation Theorem This appendix presents the proof of the fundamental theorem of Wigner 2 that any symmetry transformation can be represented on the Hilbert space of physical states by an operator that is either linear and unitary or antilinear and antiunitary. and consider the statevectors II .AI). for all k: I('l'" 'l'''I' ~ I('I'~. which is impossible since the \{Ik were assumed to form a complete set. the property of symmetry transformations on which we chiefly rely is that they are ray transformations T that preserve transition probabilities. 'l")I' ~ 0. (2.A.I) We also require that a symmetry transformation should have an inverse that preserves transition probabilities in the same sense.A. (2. in the sense that if \{It and \{I2 are statevectors belonging to rays 9ft and at2 then any statevectors \{lIt and \{IS belonging to the transformed rays T. we have I('l'~.A.
(2.92 2 Relativistic Quantum Mechanics belonging to some ray . From now on. the statevectors T k and 'f'k chosen in this way will be denoted Uik and U\{Ik.9) and (2. 1 U fi['I'.AI) we have and for I of k and I of I: Ckl = O. (2.A.. For any given k./i.. + U'I'.A. k (2. and expand it in the \{Ik: 'I'~~C.W) 1m (Ck/Ctl ~ +Im(C. it still remains to define U\fI fOT general statevectors \fl. may similarly be . Tk = 2::/kl'¥r .A.I'k may be expanded in the statevectors "P. (2.ll) .'I'. ~ 1 fi[U'I'. ICd CII' ~ ICk + C.A. 1: (2. Any statevector lk belonging to the transfonned ray T!. by an appropriate choice of phase of the two statevectors lk and \{Ik we can clearly adjust the phases of the two nonzero coefficients Ckk and Ckl so that both coefficients are just 1/.A8) while the equality of I(Yb '1')1 2 and I(UYh \{I')1 2 tells us that for all k f. (2. 12 The ratio of Eqs.A7) The equality of 1(\{Ib \{I)1 2 and 1 (U\{Ib \{I')1 2 tells us that for all k (including k ~ 1): IC. Now consider an arbitrary statevector If' belonging to an arbitrary ray :J£.I' ~ IC./C.A9) Re(Ck/Ctl ~ Re(Ck/C. From Eq.I'.8) also requires (2.5) However. + '1'.] ~ UY.) (2.. (2.A6) T~ Any state 'P' that belongs to the transformed ray expanded in the complete orthonormal set U\{Ik: '1" ~ ~C'U'I'.) which with Eq.] .Y'k. with k =!= 1.A.A. As we have seen.8) yields the formula (2.
or else Eq. Define a state~vector c:J) _ ~'3 ['P! + 'P k + 'Pd. We see then that for a given symmetry transformation T applied to a given statevector Lk Ck'!:'h we must have either Eq. and in the proof he presented he was considering only symmetries like rotations that do . we can show that the same choice must be made for each k. (This step in the proof was omitted by Wigner.A. But then the equality of the transition probabilities IN).) To see this.)". in other words.12) or else (2. (2.l2) for all k.] . (2. we have CkIC! = Ck/Cl. we must get the same ratios in any statevector $' belonging to the transformed ray: . as well as k =F l. C Cl + 'C! C 2 (Ck Ci) ~ Re (Ck Cl) CICj CIC] (~:) 1m (~:) ~ O. 'fI)1 and 1($'. in contradiction with our assumptions.13) Furthennore. . if 1m Hence either Ck/C! or CrIC! must be real for any pair k.A. a <I> ~ j3[U'P 1 + U'P k + U'P. we have instead CdC! = (CIIC. 'P')I requires that C' C' 1+ k + ' C'1 C'1 and hence 1 + Cl This is only possible if Re 2 Ck C.l3) for all k.A.) We will show that this is impossible. Eq. (2.A.2 + Cj  1 + ". because as he showed any symmetry transformation for which this possibility is realized would have to involve a reversal in the time coordinate. Suppose also that both ratios are complex. Wigner ruled out the second possibility. I. Since all the ratios of the coefficients in this state~vect~r are real. suppose that for some k. so that these are really different cases.B). (This incidentally requires that k =F 1 and 1 =F 1.Appendix A Symmetry Representation Theorem 93 and therefore either (2. or. where IX is a phase factor with 10:1 = 1.A. while for some 1 =F k.
1 both Ai: AI ami B. and choose these four coefficients all to have different phases.A.l.Crl1m(B.A.14) and (2.Aillm(B.em.) fO.13) may apply.A.A. kI r j i (2. kI (2. k or equivalently I:Im(A.A.Bil~O.A. (2.) ~ I:Ck'U'Pk. em . If A:nA" is real.A/ is reR! is handled in just the same way.) Im(A. either Eq.A. Then the invariance of transition probabilities requires that I I: B.A. so we will have to consider that.1l (possibly with either m or n but not both equal to k or I) for which B:'nB" is complex.Il) or Eq.A. and choose these two coefficients to havc different phases.18) • If for some pair k.C.A. then choose all Cs to vanish except for and C". (2.14) and (2.I' ~ I I: Bk Ak l 2 . then thcre must be some othcr pair m. with neither A k nor Bk all of the same phase (so that Eqs. we must make the same choice between Eqs. l. k . Depending on which of these alternatives is realized. or Cl = C. If Ai: AI is complex but B. (2. and en. B[ is real for some pair k. for each symmetry T and statevector I:k C k \fIk. (l. then choose all Cs to vanish except for Ck.A.14) applies for a statevector I:k Ak \fIk while Eq.A. respectively. then choosc all C s to vanish exccpt for Ck and q.15) for arbitrary values of the coefficients Ck. Suppose that Eq. (2. Then either U(I:Ck'Pd ~ I:CkU'P k k k (2. (l. for any pair of such statevectors. q .B[ arc complex.94 2 Relativistic Quantum Mechanics not affect the direction of time.14) or else U(I:Ck'P.Ad kI f 0 (2. (2. Here we are treating symmetries involving timereversal on the same basis as all other symmetries.*. (2. However.l6) We cannot rule out the possibility that Eq.A.17) and also I:Im(C.IS) It remains to be proved that for a given symmetry transformation. .B.A. The case wherc BkBI is complex but A. ! ! i I: 1m (C. we will now define U\fI to be the particular one of the statevectors \}I' belonging to the ray TfJf with phase chosen so that either C 1 = C. we can always find a third statevector Lk Ck '¥k for which* .16) may be satisfied for a pair of statevectors I:k Ak '¥k and Lk Bk\fl k belonging to different rays.A.iS) are not the same). and choo. If also A:nA" is complex.>c these two cocfficients to have diffcrent phase.iS) applies for a statevector I:k Bk '¥k.
20) so U is unitary. . (2.A.14) and (2.A.I5) must be made for Lk B k \f'k and Lk Ck\f'k.'I' + P<l» ~ U I:(aA. (2.17) that the same choice between Eqs. we shall give the details here anyway.A. ~ a I: A.. (2.B.A. but since antilinear operators may be unfamiliar. " . (2. and so: U(. U<I» ~ and hence I: A. (2.A. it follows from Eq. + P' I: B.I5) for all statevectors may be dealt with in much the same way.'I' + P<I» ~ U I:(aAk .B" " . ~ I:(aA. U'I'. It is now easy to show that as we have defined it.( U'I'" U'I'. (2.I4) is satisfied for all statevectors Lk Ck \f'k.Appendix A Symmetry Representation Theorem 95 As we have seen.I4) and (2. (2.I4) or else they all satisfy Eq. U(.A. U(a'l' Using Eq.A.A.A. . We have thus shown that for a given symmetry transformation T either all statevectors satisfy Eq.A. I:A. using Eq. U'I'. (2. (2. .IS) must also be made for the two statevectors Lk A k \f'k and Lk B k \f'k with which we started. + P' B. (2.I5) must be made for LkAk\f'k and Lk Ck\f'k.<1».A.U'I'k .IS) is satisfied for all statevectors Lk Ck\f'k.18) that the same choice between Eqs.. + PBd'l'. The reader can probably supply the arguments without help.14) again. (2. the scalar product of the transformed states is (U'I'. U<I» ~ ('1'. (2.A. + PBd'l'. (2.I4). Any two statevectors \f' and cD may be expanded as before.A. (2.A. so the same choice between Eqs. First. Suppose that Eq.3).) ~ I: A. . and it follows from Eq. + PBdU'I'. Also. (2.14) and (2.)U'I'k ~ . llsing Eqs.A. (U'I'. U'I'k· . the quantum mechanical operator U is either linear and unitary or antilinear and antiunitary.A. this gives + P<I» ~ aU'I' + PU<I>.2) and (2. The case of a symmetry that satisfies Eq.A.A. Any two statevectors \f' and cD may be expanded as and so.19) so U is linear.I5). + P I:B. ~ I:(" A.A. suppose that Eq.
in such a way that the transformations satisfy the composition rule (2. 'I . Group Operators and Homotopy Classes • Appendix B t In this appendix we shall prove the theorem stated in Section 2. running from the origin to each point e.A. (2.A.1) U[Bj U[II] ~ U [flO. (2.22) I .A. We want to construct operators U(T(e)) sponding condition· = UfO] that satisfy the corre(2.A. and define U(J(s) along each such path by the differential • Square brackets are used here to distinguish U operators constructed as functions of the group parameters from those expressed as functions of the group transformations themselves.15) again. .2) and (2. U<I» ~ LA. we introduce a set of real variables OlJ to parameterize these transformations. [ . To do this. rather than a projective representation. As described in Section 2.7. this gives U(a'l' + fi<l» ~ aU'I' + fi'U<I>. • and hence (U'I'. (2.IJ»). Also.2. II)].2. We shall also comment on the projective representations encountered for groups that are not simply connected.8. using Eqs.96 2 Relativistic Quantum Mechanics Using Eq. (2. and (b) the group is simply connected.A.15): T(B)T(IJ) ~ T(J(B. .<1>)'.21) so U is antilinear. provided (a) the generators of the group can be defined so that there are no central charges in the Lie algebra. the scalar product of the transfonned states is (U'I'. we lay down arbitrary 'standard' paths 0(j(s) in group parameter space. with 03(0) = 0 lJ and 83(1) = e .3). let us recall the method by which we construct the operators corresponding to symmetry transformations.Bi(U'I'k' U'I'il ~ LAkB.U<I» ~ ('1'. that the phases of the operators U(T) for finite symmetry transformations T may be chosen so that these operators form a representation of the symmetry group. To prove this theorem. I t so U is Qntiunitary. and their relation to the homotopy classes of the group.
and define a path f!/J that runs from a to 01 and thence to [(82. We therefore conclude that for ~< 1. ! Uo/(s) U~'(s) ~ Ue.(I).(s) along the second segment...6) Matching the coefficients of or3 in the limit 83 _ a yields the result: (2.2) untO) ~ I.2) for U..).5) At the end of the first segment. this does not say that Ue(1) satisfies the desired composition rule (2. !. Od: El:i>(s) .) h'~ ([(02. Along the second segment the differential equation (2.B.B.B. We need to show that Ue(1) is independent of the path from 0 to 8 in order to be able to identify U[8] as Ue(l).B.)) ~ h'b(8.1).4) We are eventually going to identify the operators U[8] with Ue(1). [n order to check the composition rule.(s) is thus the same as the differential equation for Ua z(2s1).(2s1) and in particular U?(l) ~ Un. but U?(s)U~l(1) also satisfies the same differential equation as Uez(2s . They satisfy different initial conditions. 0. (2. To evaluate U&.3) where [h'I'b(O) _ [J[~~. 2 < (2.(s) ~ it.0t}.Uo(s)h"b(Ele(s)) a"s (2. we are at U~(!) = Ual(l).B.{ El[". (2.Oll we have chosen to run directly from 8" = a to W = r(82. we need the derivative of r(@e z(2s1). in general it will not be the same as whatever 'standard' path l @[(O2. consider two points 81 and (h. because although the path @~(s) runs from oa = 0 to 8a = r(82. .B. < ""0. For this purpose.8) However.7) W~~l 0.B. but first we must establish some of the properties of Ue(s).B.l. and in addition the same initial condition: at s= both are unity. we use the fundamental associativity condition: (2. 8d.(!) Ue..(~S)(2sI)" O)? < ss<J!.1).8)t.Appendix B equation Group Operators and Homotopy Classes 97 s with the initial condition d dElb(s) d U.8d. (2.B.
2) gives the differential equation d deb .13) so that Eq. it remains to prove that any projective representation U[Ol of the same group with the same representation generators t a can only differ from U Ie] by a phase: . Taking the variation of Eg.B..c + Caedhebh~. J .9) vanishes hac.12) e In particular.h"h.ll) Eq...22) where habc (without central charges) and rearranging a bit.(E»d + it a Uh bc (8)bE)Is S ' ds + itaUh f)(8) dM~h d s i3h'/.19).B. Antisymmetrizing in band c shows that the last term in Eg.b dEY (h' t a uo :IS ('. Eh (2. so we may now regard U/:I(l) as a pathindependent function of alone: U. (2. (2.9) thus tells us that the quantity U1bU .I).98 2 Relativistic Quantum Mechanics For this purpose. we find for all (}: 0 in the associativity condition (2.tld. as was to be proved.B.iU. (2.. But assumption (b) tells us that any path from 8(0) = 0 to 8(1) = 0 can be continuously deformed into any other. (2.. ~ f'"h(O)dbh(e)'. +1·U (2B9 ) .c + C"ed h'h') ~ b [ + .2. Now that we have constructed a nonprojective representation UfO].B. It follows that Uf/(l) is stationary under any infinitesimal variation of the path that leaves the endpoints 8(0) = 0 and e(l) = fJ (and Uo(O) = 1) fixed.b .1!t)] (2. However.(l) . consider the variation bU of Ve(s) produced by a variation bees) in the path from 0 to O.2.:>.1t a Uh a/Jb8 h is constant along the path fJ(s).r. we have U~(l) ~ u[f(e. by taking the limit 03..B.6).B.B.8) shows that U[fJ] satisfies the group multiplication law (2.hab. this gives ds ~ (UloU) ~ ds ~ (i UI"Uh"h08') l U:o.1> .BIO) r h(e)".B.B. (2.Ule]. ' where f'de is the coefficient defined by (2.deb U d bU = itabU ha. = 0. since the path!/J leads from 0 to f(fJ2./OEY. (2. Using the Lie commutation relations (2.
e) and then back to zero is not continuously deformable to a point..8) Because U[O] and U[O] have the same generators.8 1 "~o' ] Differentiating this result with respect to /J'' and antisymmetrizing in b and c gives immediately o~ 84>. we have thus seen that rjJUh. Setting it equal to its value at a = 0. 4>.(0) . that the loop from zero to e to !(8. consider the operator U[Oj1 U[O'j1 U [O']U [0] ~ U[f(O'.et))U(e2)U(ed can be a phase factor exp(irjJ(lh. or in other words. the derivative of the lefthand side with respect to a'a vanishes at 0' = 0.W. To see this. .1(f(tJ 2 . but ¢ will be the same for all other loops into which this can be continuously deformed. 0)] . 0) cannot be deformed into the standard path we have chosen to go from zero to j(8.tJd) 1. Suppose that the path..Appendix B Group Operators and Homotopy Classes 99 so that the phase ¢ in the multiplication law for [J [61]: U[O']U[O] ~ e'¢W.OIU[f(O'. this requires that rjJh is just a gradient of some function I~: 'Ph ~ (0) ~ ap(O) (je h Thus the quantity U[O]I[r[e]eifJ(tJ J is actually constant in e.'.'fJ from zero to e to [(8.(0) _ 84>.h"b(O) [8~'b 4>(8'. Then U. The set consisting of all loops that start and end at the origin and that can be continuously deformed into a given loop is known as the homotopy class 14 of that loop. O)r l U[f(O'. we see that [r is just proportional to U: UfO] ~ UfO] exp(ip(O) + ip(O)) as claimed above. ed depends only on the homotopy class of . e). The above analysis provides some information about the nature of the phase factors that can appear in the group multiplication law when the Lie algebra is free of central charges but the group is not simply connected..1. and so o~ where a:' {u[or U[O]} i + i<Pb(O)U[Or l UIO].(0) iJe c Of)h A familiar theorem 13 tells us that in a simply connected space.O)] can he removed by a simple change of phase of U[O].
It is not convincing to appeal to the structure of the extended Poincare group. This appendix will explore generalized versions of the inversion operators. But if we are willing to admit this possibility. } ~ . there does not seem to be any good reason to impose Wigner's condition that T2 is proportional to unity. the 'inverse' of the homotopy class of the loop !e is the homotopy class of the loop obtained by going around !e in the opposite direction. but without making some of Wigner's limiting assumptions. The set of homotopy classes forms a group.PI and Sf2 is the homotopy class of the loop formed by going around . Homotopy groups will be discussed in greater detail in Volume II. perhaps with different signs for subspaces separated by superselection rules. I . a possibility that seems to have been first suggested by Wigner 15 in 1964.9'1 and then il2.O) and then back to zero. and going around loop !e gives a phase factor ei¢. in which finite matrices appear in place of the inversion phases. . Because T is antiunitary. Let us start with timereversal.6 we noted in passing that inversions might act in a more complicated way than this on degenerate multiplets of oneparticle states.100 2 Relativistic Quantum Mechanics the loop from zero to to f(O. e Appendix C Inversions and Degenerate Multiplets It is usually assumed that the inversions T and P take oneparticle states into other oneparticle states of the same species. When the sign for T2 on the space of states with even or odd values of 2j is opposite to the sign (_1)2 j found in Section 2.6. In Section 2. Hence we can catalog all the possible types of projective representations of a given group '!J (with no central charges) if we know the onedimensional representations of the first homotopy group of the parameter space of 'lJ. the 'product' of the homotopy class for loops . the physical states involved must furnish representations of the operator T that are more complicated than that assumed so far. then going around both loops gives a phase factor ei1>ei ft. and the 'identity' is the homotopy class of loops that can be deformed into a point at the origin. This group is known as the first homotopy group or fundamental group of the space in question.. perhaps with phase factors that depend on the particle species. Wigner limited the possible action of the inversion operators by assuming that their squares are proportional to the unit operator. it is easy to see that the corresponding proportionality factor for T2 can only be + 1. the only useful definition of any of the inversion operators is one that makes the operator exactly or f i . It is easy to see that the phase factors form a representation of this group: if going around loop !e gives a phase factor eirJ>.
let us assume that on a massive oneparticle state it has the action T'Pp. or 2 x 2 matrices of the form (e~j2 ei~2). (Here is the proof. where the 1> are various real phases. Y must be unitary. the matrix Y is put in the form (2.a.S) where . (2.C2) gives Y'Y'· = lij{lY. we find the same transformation (2.C. and (J are the particle's momentum. as we could if T were unitary. except that because T is antiunitary. Defining new states by the unitary transformation If'~. (The appearance of the factor (_l)i. First. and this may not be the definition that makes T2 proportional to the unit operator.cr.C!). say with phases ei</J" along the main diagonal.n = I:mo"mnlf'p. we have :T ~ D:T T .C.a and the reversal of p and (J are deduced in the same way as in Section 2. and the diagonal elements of f!jJ all vanish.J) This is a unitary transfonnation. and dropping primes.Cl) where p. with the matrix .s1 is symmetric as well as unitary. so it can be chosen to diagonalize the unitary matrix . if ei<l>" = I but ei<l>m i= 1. and spin zcomponent.a.:T•. with the blocks either 1 x 1 phases. note that Eq. By listing first all rows and columns for which eirJ>" = 1. Assuming this to have been done.C2) We cannot in general make :7' diagonal by such a choice of basis of the one·particle states.6.'Tnn vanishes unless ei<l>" = 1. Now let us see how we can simplify this transformation by an appropriate choice of basis for the oneparticle states.:Tmn changed to :7' = IffI yo.) The matrix .7 mn is unknown.m.CA) where D is a unitary diagonal matrix. (2. To explore more general possibilities for timereversal. it can be expressed as the exponential .cr 2:. m (2.n = (l)i.':1 mnlf'p. (2.Appendix C Inversions and Degenerate Multiplets 101 approximately conserved. (2.· . One immediate consequence is that the diagonal component .:T. m are indices labelling members of a degenerate multiplet of particle species.C4) tells us that :Ynm = Y mn = O. Furthermore. Because d is symmetric. and n. then Eq. (2.m. j..:T·!ilI.a. But we can instead make it blockdiagonal. spin.
_"".'1 and hence of ./>i =F 1.3).4) gives :T nm = ei¢"'Y mll and .1l becomes of block diagonal form + I ~. ) .± on which T acts with a matrix (2..CffJ into blockdiagonal form with 2 x 2 blocks of form (2.CS) Then on an arbitrary linear combination of these states. the matrix .i are subjected to the transformations rtf j t Vj1tti wt. . . and then all rows and columns with some other phase ei<p'2 01=.w ~p. so we can clearly choose this transformation to make CCi . L This establishes a correspondence between pairs of individual rows and columns within each block with phases eN).(jJj = ( ( 0 {fa.± T  e±i¢/2( ~ l)j~. it is not possible to choose states to diagonalize the timereversal transformation.+ + c_ m ) _ Tp . Hence :T nm = /7 mn = 0 unless ei<j'"ei<Pm = 1..102 2 Relativistic Quantum Mechanics i of a symmetric antiHermitian matrix. so ..fT.. and with the matrix in the ith block of form ( V.C.'Trim = ei¢>neirPmfTnm and also :Tmn = ei"'"ei¢my /11". so it can be diagonalized by a transformation (2. it is now only necessary to rearrange the rows and columns so that within the ith block we list rows and columns with phase eirh alternating with the corresponding rows and columns with phase eir/>j. Eq. If we list first all rows and columns of :14 with a given phase eir. a 0) W j with Vi and Wi unitary.2) acting only on .. and then all rows and columns with the opposite phase.C6) e~i<jJi~2'1?r ei¢i~2'1/j ) (2. and so on. If we have a pair of states 'f'p.fT. It is therefore only necessary to consider the submatrix [!I that connects the rows and columns for which ei¢>" O. To put the matrix .07 nm . = t6j = 1. (2.'Tml! = ei<i>m.fT.1 not equal to e±f¢]. timereversal gIves ' T (C+ 'P p. the unitarity of . Furthermore.2) with JI1 block~diagonal in the same sense as ].C.C.) It is important to note that where eir/> =F 1.f1.(jJ requires that (6It?.w.jiU( e'.. and e~i(h..=t= I . By applying a transformation (2.. 91~O where . (2. For n =F m. ( ._ 'm " T + e'¢/2 (_ m ~p. . the submatrices '(.+ ) . with the corresponding submatrix of ilIJ real and hence orthogonal.C. then THl p.3).j2 C+ T _p.fT.C7) '1/. and hence t6i is square and unitary.C. (2. and then all rows and columns with opposite phase.
But this choice of basis may not be the one in which timereversal acts simply.II) . But there is no reason not to take a general phase ¢ in Eq. with P'f'p. ± then we can redefine the timereversal operator as T' SIT.C9) If we were to assume with Wigner that T2 is proportional to the unit operator.p/2U1 S UI Tp.± with one another.CS).± .+ .to be transformed under T by a phase necessary that it is But combining these equations gives e±i¢!2 c = IAI 2c±iFi ¢12. (2. As discussed in Chapter 5. here we may always diagonalize this matrix by a choice of basis for the states. It is only in the case where no such internal symmetry exists that we can attribute the doubling of particle states to timereversal itself.u. then we would have to have ei</J = ei. The choice ei¢ = 1 would still require a twofold degeneracy of oneparticle states beyond that associated with their spin. and since the phase is then real it would have to be + 1 or 1.u. which acts on oneparticle states as (Z. Let's come back now to the question of the square of T.C.1. Thus for ei<b 1:.1O) with a unitary but otherwise unconstrained matrix !Y. 103 For c+ 'Pp. so.u. P and T together can impose additional degeneracies that would not be required by PorT alone.m m (2.u.. Thus the fact that observed particles do not show the extra two·fold degeneracy does not rule out the possibility that others might. e=Fi<fi \f'P.<T.C. any quantum field theory is expected to respect a symmetry known as CPT. .Appendix C Inversions and Degenerate Multiplets A.p.n = L. Unlike the case of timereversal.u.<T. We may also consider the possibility of more complicated representations of the parity operator P.rr. one that may vanish for some particles and not for others.= 0 or eN' is unity. in principle.#nm'Pp. and this operator would not mix the states 'Pp.C8) gives _ T2'P p•u. timereversal invariance imposes a twofold degeneracy on these states.'Pp. Of course. and under Wigner's assumptions all particles would show this doubling.+ .+ + c. Repeating the transformation (2.(1) 2. which is impossible unless either c+ = c. (2.± ~ e T p. if there is an additional 'internal' symmetry operator S which subjects these states to the transformation _ ±i. beyond that associated with their spin.
<l".15) On the other hand.13) the CP operator also acts conventionally CP\f'p.. (2. From now on.+ _ e=Fi¢/2m p./1 oc \f' p. In this case.<7.n' . ~ (1) 2' )'1'. (2. Problems 1. A second observer (!)' moves relative to the first with velocity v in the zdirection. CP and T would be what are usually called P and CT.0) with momentum p in the ydirection and spin zcomponent a. where T has the unconventional representation (2. But this is not merely a matter of definition.o.n .C.C. so is the inversion CP _ (CPT)Tl.O" . (2. it is possible that the degeneracy indicated by the label + may be the same as the particleantiparticle degeneracy.) It follows that (CPT) 2 '1'..6.. where nC denotes the antiparticle (or 'chargeconjugate') of particle n.C.. so that the antiparticle (as defined by CPT) of the state \f' ± is \f'f.12) so the possibility suggested by Wigner of a sign _(_1)2j in the action of (CPT)2 does not arise in quantum field theory.n oc qt p.C...14) The operator C = Cppl then just interchanges particles and antiparticles C \f' p._.C.a.IT." • (2. No examples are known of particles that furnish unconventional representations of inversions." (2. Eq. No phases or matrices are allowed in this transformation (though of course we could always introduce such phases or matrices by combining CPT with good internal symmetries.16) In particular.. To the extent that T is a good symmetry of some class of phenomena.C.. on other particles CP and T would still have their usual effect. the inversions will be assumed to have the conventional action assumed in Section 2. How does (!)' describe the W state? .=f" T T (2. For the states that transform under T in the conventional way T\f'p.C.ll) gives Cp m p.104 2 Relativistic Quantum Mechanics I ./1 oc qt p. CP would have the unconventional property of not interchanging particles and antiparticles. Suppose that observer (IJ sees a Wboson (spin one and mass m f. so these possibilities will not be pursued further here.8)..O" .. As far as these particles are concerned..
Derive the commutation relations for the generators of the Galilean group directly from the group multiplication law (without using our results for the Lorentz group).1). A second observer (!/ moves relative to the first with velocity v in the z~ direction. Phys. G. 3a. How does (!/ describe the same photon? 3. . 149 (1939). S. Inc. 5. and E. M. Wightman. For massless particles.a).1. Academic Press. A. see also E.jVPPI:. E. Wick. (Ox~ 2.. 64. Wigner. 1). Rev. Math. New York. Vienna. E. 2513 (English translation. 1958).g. P. 1963): p. P. Nuovo Cimento IX. Suppose that observer (!) sees a photon with momentum p in the y~direction and polarization vector in the z~direction. Wigner. in Theoretical Physics (International Atomic Energy Agency. E. P. 1959). 3. C. consider physics in two space and one time di~ mensions. Inonli and E. Weinberg. P. 1972): Section 2. See. 4. A. Gravitation and Cosmology (Wiley. Oxford. assuming invariance under a 'Lorentz' group 0(2. Wigner. As in Problem 5. The Principles of Quantum Mechanics. Consider physics in two space and one time dimensions. Wigner. New York. and Wp WP commute with all Lorentz transformation operators U(A.PlJ. e. P. P. How would you describe the spin states of a single massive particle? How do they behave under Lorentz transformations? What about the inversions P and T? 6. S. Include the most general set of central charges that cannot be eliminated by redefinition of the group generators. 88. Dirac. Show that the operators PlJ. 4th edn ford University Press. 4. 705 (1952). 1931): pp. 40. 101 (1952). Ann. Wigner. assuming invariance under a 'Lorentz' group SO(2. How would you describe the spin states of a single massless particle? How do they behave under Lorentz transformations? What about the inversions P and T? References 1. 5.References 105 2. where W I1 ~ €/lI'p. Gruppentheorie und ihre Anwendung auf die Quanten~ mechanik der Atomspektren (Braunschweig.
. 14. N. Rev. 1985): Sections 7. see. 99. Tuday. Mackey. Proc.3 and 8. L Friedman and V.265 (1958). R". A. Ann. H. Rose. Lee and C. An Introduction to the Theory of Canonical Matrices (Dover Publications. J. 1957): Chapter 4. Cronin. M. 193 (1953). MA. . Young. Phys.1. 8. 1961): Chapter 4.. 58. Princeton. M.. New York. J. H. L. J. Telegdi. by F. 1977): Section 58. 1964): p. D. 1415 (1957). Hocking and G. 105. S. Bargmann. Wu et ai. 959 (1930). also see F. g. L. Phys. C. 11. 55. P. WuKi Tung. 7. Elementary Theory ~/Angular Momentum (John Wiley & Sons. Phys. Phys. Acta. Induced Representations a/Groups and Quantum Mechanics (Benjamin. e. Group Theory in Physics (World Scientific. G. Lederman. Letters 13. Garwin. E. Kramers. 105. J. Rev. J. Rev. e. to. 105. V. Acad.Non Relativistic Theory. 12. 140 (1962). Flanders. Rev. Edmonds. W. See e. H. 37. and M. Dyson. New York. Sen. 15.1. New York. ed. W. 1413 (1957). Math. 1681 (1957). C. Oxford. L.6. Nash and S. Lifshitz. T. Wigner. 3rd edn. (Pergamon Press. Singapore. 3. 101 (1952). W. G. e. in Group Theoretical Concepts and Methods in Elementary Particle Physics. 1961): p.g. C.g. Sci. Phys. New York. 9.106 2 Relativistic Quantum Mechanics 6. and R. Amsterdam 33. 1963): Section 3.. Math. E. (Princeton University Press. 1968). Angular Momentum in Quantum Mechanics. A. L Fitch. Reading. Math. Yang. R. S. 13. Aitken. 194. H. Differential Forms (Academic Press. D. 138 (1964). Ann. Landau and E. Phys. 104. New York. R. Turnbull and A. See. 1957): Chapter IV.1 (1954): Theorem 7. See. For an introduction to homotopy classes and groups. Weinrich. Quantum Mechanics . 59. London. Christenson. 1983): Chapters 3 and 5. V. Math. Topology (AddisonWesley. Gursey (Gordon and Breach. Topology and Geometry for Physicists (Academic Press.g. J.254 (1956).
The general transformation rule is 107 . 3. the paradigmatic experiment (at least in nuclear or elementary particle physics) is one in which several particles approach each other from a macroscopically large distance.it is only when two or more particles interact with each other that anything interesting can happen. after which the products of the interaction travel out again to a macroscopicallY large distance. or 'crosssections'. for massless particles.1 'In' and 'Out' States A state consisting of several noninteracting particles may be regarded as one that transforms under the inhomogeneous Lorentz group as a direct product of oneparticle states. which includes a specification of its mass. But experiments do not generally follow the detailed course of events in particle interactions. In such an experiment. etc. spin zcomponent (or. Rather. since we now may be dealing with more than one species of particle.3 Scattering Theory The general principles of relativistic quantum mechanics described in the previous chapter have so far been applied here only to states of a single stable particle. all that is measured is the probability distribution. and. To label the oneparticle states we use their fourmomenta pI'. The physical states before and after the collision consist of particles that are so far apart that they are effectively non~interacting. Such oneparticle states by themselves are not very exciting . so they can be described as direct products of the one~particle states discussed in the previous chapter. This chapter will outline the formalism 1 used for calculating these probabilities and crosssections. and interact in a microscopically small region. charge. for transitions between the initial and final states of distant and effectively noninteracting particles. spin. helicity) fI. an additional discrete label n for the particle type.
.10).1.>l[. . Eq. '¥PI .1) requires among other .p».P2))'" lT x x (ApdO(Ap2)o.) ~ S(a' .···.'" 0. the completeness relation for states nonnalized as in Eq.108 3 Scattering Theory U(A. its sign is 1 if this permutation includes an odd pennutation of half·integer spin particles.O.Pt}O".11] .rfj..n2 .~(W(A.p» is replaced with Jul a exp(iaO(A.P2.0._.43).2) is written simply ('1'".P2.O. This will not be important in the work of the present chapter. say Cl. where () is the angle defined by Eq. '...3) reads 'I' ~ J da '1'.'" = exp (  iaj1(p~ + p5 + . we write (3. ) = J3(p't . ( '¥ p.4) In particular.111 .(1) standing for the sum of products of delta functions and Kronecker deltas appearing on the righthand side of Eq.1.1.5.a)'f'PJ.n. . the matrix Dj..Ap2. Setting AP v = bPI' and aP = (O.12) with the term '+ permutations' included to take account of the possibility that it is some permutation of the particle types nl.n2"" that are of the same species of the particle types nl. Eq.r).. '1'. (2..5.n'l .1.O. in summing over states.) ll"j<1j 202 (3.o2on.1.1) is only possible for particles that for one reason or another are not interacting.2).02.<11.1. letting one Greek letter. (3. (3.p) is the Wigner rotation (2. and D~~(W) are the conventional (2j + 1)dimensional unitary matrices representing the threedimensional rotation grou (This is for massive particles.) The states are normalized as in Eq. and otherwise +1..('1'" '1').· '" pOpO.>l2.5) The transformation rule (3. CT2. + permutations (3. 1 2 'f't\PI.u. ..pd)D'1l (W(A. for which U(A. ») (W(A. . (As discussed more fully in Chapter 4.a) = exp(iHr).1..) We often use an abbreviated notation."2.U2.. nl .P2)ba.>lI.1) where W(A. .1. (3.0. n2..P2. .a. for any massless particle.a) (3.nI03(p.3) with 15(0:' .5.1. (3.19) 61 . . stand for the whole collection pI. In this notation..'" ~ d~. (2.n2. (3... 0"1.oIOn.p . Also.
1. statevectors do not change with time . (!)' will see the system in a state U(1. and end with particles at t _ +00 so far apart that they have ceased interacting. Of course. As explained at the beginning of this chapter. (This is known as the Heisenberg picture.1. but they seem to have become traditional.1 'In' and 'Out' States 109 things that \IIrx be an energy eigenstate H\II rx = E rx \IIrx (3. the two observers' time coordinates are related by t' = ( . However.3.To Then if (!) sees the system to be in a state \II.t if observations are made at t _ 00 or t _ +00. In particular. implicit in the definition of the states is a choice of the inertial frame from which the observer views the system.. that is. (3.will be found to contain the particles described by the label r. the transformation rule (3.T)\II = exp(~iHT)'f'.16). in the formalism we are using here. We therefore have not one but two Sets of states that transform as in Eq. Thus the appearance of the state long before or long after the collision (in whatever basis is used by trl ) is found by applying a timetranslation operator exp(iHr) with! _ 00 or r _ +00. different observers see equivalent statevectors. we must consider wavepackets.1.1): the 'in' and 'out' states· 'f'rx + and \IIrx. but not the same statevector. then it cannot be localized in time . Therefore. with an amplitude g(rt) that is nonzero and smoothly varying over some finite • The labels '+' and '' for 'in' and 'out' states may seem backward. . while some other observer <!JI at rest with respect to the first uses a clock set so that (' = 0 is at a time ( = !. (3. respectively.) Thus we do not say that \IIrx± are the limits at t _ +00 of a timedependent statevector \II(t).a statevector \II describes the whole spacetime history of a system of particles. respectively. To maintain manifest Lorentz invariance. terms that would involve more than one particle at a time. On the other hand.1.the operator exp(iHr) yields an inconsequential phase factor exp(iErxr). Note how this definition is framed.7) I and with no interaction terms. where the operators are constant and the states change with time. (3. in distinction with the Schrodinger picture. suppose that a standard observer <!J sets his or her clock so that t = 0 is at some time during the collision process. They arise from the signs in Eq. in the typical scattering experiment we start with particles at time t _ 00 so far apart that they are not yet interacting. superpositions J drt g(rt)'Prx of states.6) with an energy equal to the sum of the oneparticle energies Eo ~ pl + pg +.1. if the state is really an energy eigenstate.1) does apply in scattering processes at times ( _ +cn.
respectively. Hlf'rz± = Erz'Prz± which satisfy the condition (3. not Ho.110 3 Scattering Theory range £:::.1. in nonrelativistic problems we can include the binding potential in H o. To make this concrete. not Ho.12) f d.1.rg(a)'P x± has the appearance of a corresponding superposition of freeparticle states for T «: l/. Eq. of the complete Hamiltonian (3." (3. Also.£ of energies. This is sometimes rewritten as a formula [or the 'in' and 'out' states: (3.}> +1/6£.12) can be rewritten as the requirement that: exp(iHr) r ! t • J dcx g(a)'Pa:± + exp(iHor) J da g(a)<b rz .18) in such a way that Ho has eigenstates cD".13) •• Alternatively.1. H ~Ho+ V (3.1. which are not necessarily the same as the 'bare' mass terms appearing in H. .6. The 'in' and 'out' states are defined so that the superposition exp(iH!) J det. the difference if there is any must be included in the interaction V. respectively.E or '[:. g(a)'Po:± = J da eiE. In the application of this method to 'rearrangement collisions. suppose we can divide the timetranslation generator H into two terms. .' where some bound states appear in the mitial state but not the [mal state.tl) J da eiE"g(a)'Pa:± + J ! . respectively. a freeparticle Hamiltonian Ho and an interaction V. one must use a different split of H into Ho and V in the initial and final states.10) Note that Ho is assumed here to have the same spectrum as the full Hamiltonian H. any relevant bound states in the spectrum of H should be introduced into Ho as if they were elementary particles:" The 'in' and 'out' states can now be defined as eigenstates of H. or viceversa. This requires that the masses appearing in H o be the physical masses that are actually measured.1.9) (3. that have the same appearance as the eigenstates \fit and 't'. (3.x eiE"g(a)<ba: for r + 00 or r + +x.1. for r + 00 or r + +00. \. .
One immediate consequence of the definition (3.1.±) ~ (<I>p.14) However.1..12). we tentatively write the formal solutions as /1)" plus a term proportional to V: \(1. (3.17) at the end of the next section to give a slightly less unrigorous proof of the orthonorm<llity of the 'in' and 'out' states.)g(. (J.. . (3.)exp(iHor).1.1.<I>.Ho is not invertible.± = V'¥.I d.)· Since this is supposed to be true for a1l smooth functions g(Cl).12) is obtained by letting the unitary operator exp(IH1") act on a timewindependent state.<I>.Ho)'¥.fJ_ IE (3. it annihilates not only the freeparticle state /1)".1. (3. (3. 'I' ± " ~ <I> + J d" Tp.11) as (E" . .) ~ b(P . (3.1.± = l1let + (E"  Ho + 110)1 V \(1 r/ .Ho.1 'In' and 'Out' States 111 where !l(.± . (11.1. These are known as the LippmannSchwinger equations.. the scalar products must be equal ('I'p±.12): J d.18) with 10 a positive infinitesim<ll quantity.. . the norm of the righthand side of Eq.1.1.)g·(P)('I'p±.3.13) gives meaningful results only when acting on a smooth superposition of energy eigenstates.e. dP exp(i(E. its nann is independent of time.11) satisfying the conditions (3. dP exp(i(E. but also the continuum of other freeparticle states ¢p of the same energy. note that since the lefthand side of Eq.).1.16) or.15) It is useful for some purposes to have an explicit though formal solution of the energy eigenvalue equation (3.12) is that the 'in' and 'out' states are normalized just like the freeparticle states.E p ).Ep)r)g(·)g·(P)(<I>p. la We shall use Eq. . expanding in a complete set of freeparticle states.1. it should be kept in mind that 0(+00) in Eq. inserted to give meaning to the reciprocal of EC/.1. The operator E et .± <I>p "EE+" C/. 'I'. and therefore equals the norm of its limit for 1" _ x.17) (3. For this purpose.) _ exp(+iH. i. To see this.. (3. Since the 'in' and 'out' states become just l1let for V _ 0.±) ~ . 'I'. write Eq.
In the same way.Ep + i€ For t _ 00. in agreement with the defining condition (3. their contribution is exponentially damped for t _ 00.1. The functions g(lX) and Tf3ll± may. (3.u" . . > O. (3. we can close the contour of integration for the energy variable Ell in the upper halfplane with a large semicircle.1.1.1. with the contribution from this semicircle killed by the factor exp(iEllt).. (3.12) for an 'in' or an 'out' state.20) 00 We want to show that 'Pg +(t) and 'Pg (t) approach I1l g (t) for t _ t _ +00. We conclude then that f {/+ vanishes for t _ rx:. (3.(1) + Jda JdP eiE"g(. which is exponentially small for t _ 00 and 1m E:l'. In general. Ell Ef3 + l€ da (3. '(E)  E + l1r. (3. respectively. For future use. consider the superpositions 'Pg +(t)  <1>.F p. in general. t must be much greater than both the timeuncertainty in the wavepacket g(o:) and the duration of the collision. for t _ +00 we must close the contour of integration in the lower halfplane.±(t) and ~ <1>.Fp. . (Specifically. with a +i€ or i€ in the denominator.vanishes in this limit.19) eiE"g(a)<I>. respectively. and so .) This leaves the singularity in (Ell . which respectively govern the location of the singularities of g(lX) and Tf3rJ. and consider first the integrals .17) in Eq.17). For this purpose.22) .1. we can write (E ~ ..:i.1. We conclude that 'Pg±(t) approaches I1l g (t) for t _ +rx:.± in the complex Ell plane.(1)  J Jda da eiE. we note a convenient representation of the factor (Ell Eft + i€)l in Eq. .F p+.Pf3± J eiEotg(rt)T + (h Ex .Ef/ + i€)l. which is in the upper halfplane for .21) Let us recklessly interchange the order of integration. + I€ )1  _ fY'.1. be expected to have some singularities at values of Ert with finite positive imaginary parts.±<I>p .)Tp. satisfies the condition (3. (3.12).17).112 3 Scattering Theory It remains to be shown that Eq.19) gives 'I'. but just as for the large semicircle. The integral is then given by a sum over the singularities of the integral in the upper halfplane. (3.1.lg(a)'P. Using Eq.but not .1.1.
1.5) to the 'out' states.(E) . for t 4 co. and then Spo: would be just 0(1Y. '1'. .1. and vanishes for E 4 0.24) The function (3. +) . The rate for a reaction IY. (3. With this understanding. and then measures what this state looks like at r 4 +00. (3.25) The S~matrix An experimentalist generally prepares a state to have a definite particle content at r 4 eJ::. ~ 13).4 what Spo: has to do with measured rates and cross*sections. and gives unity when integrated over all E. and if it is found to have the particle content fJ at t 4 +co.23) (3. + fJ is thus the scalar product Sp.2. 'I' P)('1'P.1). If the state is prepared to have a particle content IY.E2 + e2 ' b. with expansion coefficients given by the Smatrix (3. + fJ is thus proportional to ISpo: .Sp.1. They differ only in how they are labelled: by their appearance either at t + co or t + +00. we can drop the label e in Eq. Since SfJO: is the matrix connecting two sets of orthonormal states.2. by excluding an infinitesimal interval around E = O.24) is of order e for lEI ::P e. it must be unitary.1.22). then it is the 'out' state \f'p. '1'.3. Any 'in' state can be expanded as a sum of 'out' states. The function (3. and write simply (E + ie)I 3.1. . ~ J dP ('I'. The probability amplitude for the transition IY. +. so for e + 0 it behaves just like the 'principal value function' f!JJIE. 2 If there were no interactions then 'in' and 'out' states would be the same. +. and write J dP Sp.1) This array of complex amplitudes is known as the Smatrix.23) is just liE for lEI ::P e. We shall see in detail in Section 3.0(11 ~ 13)1 2.~('I'I''I'.2 f? ~ E + ino(E). apply the completeness relation (3. e (3.1.+). Perhaps it should be stressed that 'in' and 'out' states do not inhabit two different Hilbert spaces. so in the limit e + 0 it behaves just like the familiar delta function o(E). (3. +) ~ ('I'.2 The Smatrix 113 where f!JJ" _ E E . then it is the 'in' state \f' 0: +. To see this in greater detail. which allows us to give meaning to integrals of l/E times any smooth function of E.n(E2 + e2) .1.
1. By the method of residues.114 Using (3.2. we now do pick up a contribution from the singular factor (EC( .1. for t + +00 the integral over a in (3. sst = 1. but this time take t + +x.2.5 to derive a formula for the Smatrix in timedependent perturbation theory.:x) make no contribution for t + +00. completeness for the 'in' states (3.3) or.21) for the 'in' state qJ+. so it circles the singularity in a clockwise direction. . b(EIJ.6) This will be used in the next section to examine the Lorentz invariance of the Smatrix.21) has the asymptotic behavior . TO) (J. .2. times a factor ~2in.TO)) exp( iHoTo). Note that for infinite ·matrices.15).2. and although as before the singularities in TfJx + and g(.4) The explicit though highly formal expression (3. In the same way.Efl + iF)l.1. in other words.5) Q(T) t Q(TO) ~ exp(iHoT) exp( iH( T.2. (3. this gives 3 Scattering Theory (J. The contour runs from Ex = 00 to Ex = +00. The methods of the previous section can be used to derive a useful alternative formula for the Smatrix.2) or • • gIves In brief.1. (3.iF.iEp1 ' / d':J.oo) .13) for the 'in' and 'out' states yields a formula for the Soperator: s ~ Q(oo/Q(ool ~ U(+oo. It is often convenient instead of dealing with the Smatrix to work with an operator S. Let's return to Eq.· the unitarity conditions StS=1 and sst = I are not equivalent.• ~ r/ + _2ine. 3. We must now close the contour of integration for Ex in the Im'v'er halfEC(plane. where U(T.EfJ)g(a)TfJC( + • All altcmative proof is given at thc cnd of this section. this contribution to the integral over EIJ. That is. sts  1. and then back to EC( = 00 on a large semicircle in the lower halfplane. defined to have matrix elements betweenjreepartide states equal to the corresponding elements of the Smatrix: (3. is given by the value of the integrand at Ex = Efl. in the limit E + 0+. and in Sec.
and hence. without having to deal with limits at t + +co.) J dP 'P p. this may be rewritten \fIg+(t) = J dfJ 'Ppe.1.Ert ).Ep)g(')Tfi' + or in other words Sflrt = (5(fJ IX) . *** We can use the LippmannSchwinger equations (3.lSp.. in which case Eq./ dP <l>p e.2. 6(E. g(. V\fIt) and equating the results. this has the asymptotic behavior for t + +co 'Pg+(t1 ~ .J. But expanding (3.1.2. ~ gUI) lin Comparing this with our previous result. and. 3 Higherorder terms are discussed in Section 3.1.I V\fI rt ±) ~ (<I>p.19) for \fig + in a complete set of 'out' states gives 'Pg +(t) ~ J d.7) This suggests a simple approximation for the S~matrix: for a weak interaction V. / d.16) on either the left. (3.2iTl:J(E rt  Ep)TfJrt + . we find J d. $.1. we find that (\fI fJ ±.or righthand side of the matrix element (\fIff.2. +(t) ~ J dP eiE"<I>p [g(P1  2in J d. this gives the .18). V'P.Ho + iF)1 V'P.1. using the defining property (3.Efi)g('lTp.if" . .12) for 'out' states.7).. by using (3.2 The Smatrix 115 +co. . . for t + 3.±).2. Since SfJrt contains a factor b(Ep . V <brt) + (\fI fl±.5.lSp.iE /11 J dlXg(a)Sp. g(.Sp.7) gives (3. 6(E.Ho + id.8) This is known as the Born approximation. J d. V (Ert . First. as well as Eq.±l + ('Pfi±. eiE"g(./ Summing over a complete set of intermediate states. t 'P. +] . V(Ep . we can neglect the difference between 'in' and freeparticle states in (3. (3.16) for the 'in' and 'out' states to give a proof 4 of the orthonormality of these states and the unitarity of the Smatrix. (3.
1.a) acts as in (3. covariance) property of the Smatrix: for arbitrary Lorentz transformations Ai'v and translations aIJ .a) is unitary. Lorentz lnvariance For any proper orthochronous Lorentz transformation x _ Ax + a.1) on either the 'in' or the 'out' states.3 Symmetries of the SMatrix In this section we will consider both what is meant by the invariance of the Smatrix under various symmetries.Ea +)' + + lie Tp ± (1.J instead of (E a .Ell if" + + if" .9) with r5(E p .T{irt. we mean that the same operator U(A. (3. 3. divide Eg.E" . this is just the statement that the 'P:>: + fonn two orthonormal sets of statevectors. We see then that b(P .E rl + iE) is unitary.116 3 Scattering Theory equation: T~{i ±* .a)'I'I. Ep . we may define a unitary operator U(A. we obtain the Lorentz invariance (actually. (3.U(A.2. and what are the conditions on the Hamiltonian that will ensure such invariance properties.9) by Eo: . The 210 s in the denominators on the lefthand side can be replaced with since the only important thing is that these are positive infinitesimals. This gives ( ES.1) on both 'in' and 'out' states.17). Since the operator U(A.Ep d =  +)' J( + TrP'Y + 2iE Ty:y.a)'I'. J" x dy T}'(J ±* TI"~ ± (lE. T ~I' . we may write 5p.2. (3.E".  Ey + iEr l  lEp .1.9) To prove the orthonormality of the 'in' and 'out' states.E p EfJ .12) + Tpy'±j(Ep . . + iErl).2. When we say that a theory is Lorentzinvariant. ~ ('1'1''1'.+) ~ (U(A. a) by specifying that it acts as in Eq.+) so using (3. ± .1). The unitarity of the Smatrix can be proved in a similar fashion by multiplying (3. With (3.1. E~ .1.Ep + 2ie.E.E p + 2ie)I.
because it is only for certain special choices of Hamiltonian that there exists a unitary operator that acts as in (3.1) on both 'in' and 'out' states. (3. it will be convenient to work with the operator S defined by Eg. We need to formulate conditions on the Hamiltonian that would ensure the Lorentz invariance of the Smatrix..1.4): 5/.)) x x X p?pg .UI.3.(W(A. so must be the righthand side.S<I>.) Eg..nl. the amplitude Mpl'J."2."I.pJ)DY:l.3. bars are used to distinguish summation variables.1) (Primes are used to distinguish final from initial particles. (3. 'J) (ApIlO(Ap2)O . As we have defined the freeparticle states 4lC( in Chapter 2.o(A .3.. since the lefthand side is independent of all. L Dg'l. (3. X <l>Ar j. ~ (<I>p.(p't' ..' ' = exp ( .3 Symmetries afthe SMatrix 117 = exp (iall(PI Jt + P2 1l + .1.. as we will see in the next chapter. " n .1) will thus hold if this unitary operator commutes with the l .itself contains terms that involve further delta function factors.P2. p?pg. so we can always define a unitary operator Uo(A. a)wPI.. We can therefore write the part of the Smatrix that represents actual interactions among the particles in the form: (3. For this purpose... they furnish a representation of the inhomogeneous Lorentz group.) In particular.U!.pi " .3. ~I ~ ~I Eg.x + p".. and so the Smatrix vanishes unless the fourmomentum is conserved. t1 2..  pt .ia.P2)) °IOr' (ApIl O (Ap2)o . p?pJ ..2) (However.. + .1) on these states: V. rather than a theorem..n2.2.1) should be regarded as a definition of what we mean by the Lorentz invariance of the Smatrix...). a) that induces the transformation (3.(W(A.3.UI. (APlJO(APSJ O'" L (3.'Ar 2.
Kj] = I.f " i j k . Clod a boost generator K o.3.3.j.7) (3.3.118 3 Scattering Theory Soperator: Uo(A.3.3. Just as in Section 2. pj] ~ iHo 'J ' [ii.4.S] ~ [Po. that together with Ho generate the infinitesimal version of inhomogeneous Lorentz transformations when acting on freeparticle states.12) (3.3. (3.3. [K6.' operators P.3.4) (3.1) on. i 0 [J0.3. where i. and 3. .1) is equivalent to the statement that the Smatrix is unaffected by such transfonnations.13) (3. say. f"ijk K' 0.8) (3. a) ~ 5 .jk Jk pk [J i.3.9) (3. etc. H] ~ [pi.3) Because the operators Ho. Pi] ~ 0 .3.H] = ipi. an angular momentum Jo.4.18)(2. In the same way. J' (3. Ho] ~ [Po.6) (3.3. run over the values 1. 5] ~ [Ko. they automatically satisfy the commutation relations (2. i [K 0' Kj] . pj] ~ 0. [K i. (3. H] ~ [pi.' If"ijk J' . or in other words. Eg. pj] = I. and f"ijk is the totally antisymmetric quantity with f"m = + 1.11) (3.Hol =iPJ. . Kj] = If". (As already mentioned.. Kj] = t f " iK' . a momentum Po.4. [J i . [Jo. Eijk J' 0 0 i [J0' pj] 0 = I f"ijk (3. This condition can also be expressed in terms of infinitesimal Lorentz transformations. 2. Jj) .3. what is not obvious is that the same operators generate the same transformations on the 'out' states.3.1. (3.S] ~ [J o.16) (3.5) . and Ko generate infinitesimal inhomogeneous Lorentz transformations on the cD". that the Soperator commutes with these generators: [Ho..S] ~ o.3. J o. Po. P61 = iHobij .14) (3.) The group structure tells us that these exact generators satisfy the same commutation relations: [J i. a)I S Uo(A. K that together with the full Hamiltonian H generate the transformations (3.17) [Ki. Ho] ~ [Po./. J.10) . (3. the 'in' states. we may define a set of 'exact generators.15) [K i. there will exist a set of Hermitian operators.3. .k. pk 0 ' [K6. jk .24): i [J0' Jj] 0 =tf"ijk 0' .
17).16) are satisfied provided that the interaction commutes with the freeparticle momentum and angularmomentum operators [V.21) is not enough.16) or equivalently from (3. we already know that the Soperator commutes with Ho.2. it is not possible to set the boost generator K equal to its freeparticle counterpart Ko.3.(3.3.3.15) and (3. Also we easily see that Po and J o commute with the operator U(t. Further. (3.7).19) equation (3. (3. Eq.H].6). (0). where the angular momentum of states depends on the interactions.3. Recall that the crucial point in the Lorentz invariance of a theory is not that there should exist a set of exact generators satisfying Eqs.E tl ).3.3. V]'P~)/(Ep . This leaves just the boost generator Ko which we need to show commutes with the Soperator. (3J. which is certainly not true in the presence of interactions. (3. the condition (3. (3.20) It is easy to see from the Lippmann~Schwinger Of the remaining commutation relations.3. while leaving the momentum and angular momentum unchanged: H ~ Ho+ V .1.13) that the operators that generate translations and rotations when acting on the 'in' (and 'ouf) states are indeed simply Po and J o. because there are energyconservation delta functions in both tenns in (3. and in particular should not have singularities of the form (Ef:J . because for any V we could always define W by giving its matrix elements between H eigenstates 'P ~ and 'P p as ('P il . [Ko. p = Po .18) (The only known exceptions are theories with topologically twisted fields.3.tol defined by Eq. (3.3.3. but rather that these operators should act the same wayan 'in' and 'out' states.3. (3. and hence with the Soperator U(co.11)~(3. (3. because then Eqs. let us concentrate on Eq. We shall now show that .3.1. the effect of interactions is to add an interaction term V to the Hamiltonian.3.14). we must also add a correction W to the boost generator: (3. merely finding an operator K that satisfies Eq.3 Symmetries oj the SMatrix 119 In virtually all known field theories.21) is empty.3. which may now be put in the form [Ko.18) implies that the commutation relations (3. J = Jo .2.Eo:)~l. Thus when we add an interaction V to Ho. (3.8) would give H = Ho.17). Pol ~ [V.3.Jol ~ O. (3.21) does become significant if we add the requirement that matrix elements of W should be smooth functions of the energies. such as those with magnetic monopoles.3.2I) By itself. V] ~ [W.11).) Eq. and (3. On the other hand.
as required in order to justify the very idea of an Smatrix. Also.22) where W(t) .exp(iHot)W exp( iHot). (3. we have Hrl(+oo) ~ Q(+oo)Ho. then matrix elements of W(t) between smooth superpositions of energy eigenstates vanish for 1 _ +00. since all ¢a. (3.exp(iHot)] ~ tPoexp(iHot) ~ while Eq.21).3.13) the operator that converts a freeparticle state <1>~ into the corresponding 'in' or 'out' state '¥~+. (3. (3.19) that the same is true for the momentum and angular momentum: PQ( +00) Jrl(+oo) ~ ~ rl(+00)Po . it follows trivially from Eqs. does imply the remaining Lorentz invariance condition [Ko.S].26) (3.(3. Using Eq.3.3.3. so Eq.to) for finite t and to. S] = O. together with an appropriate smoothness condition on W.3.oo)] ~ [Ko. To prove this. (3. let us consider the commutator of Ko with the operator U(t. because it is much like the condition on matrix elements of V that is needed to make V(t) vanish for t + +00. (3. This is the essential result: Eq. U(oo. (3.17» yields [K. rl(+oo)Jo.3.10) and the fact that Po commutes with Ho yields: [Ko.22) gives in effect: o ~ [Ko. exp(iHt)] ~ tPexp(iHt) tPo exp(iHt) .3.3. (3.18) and (3.3. We can also use Eq.3.3.3.3.27) Finally.23) If the matrix elements of W between Hoeigenstates are sufficiently smooth functions of energy. The momentum operators then cancel in the commutator of Ko with U. (3.21) together with the smoothness condition on matrix elements of W that ensures that W(t) vanishes for t + +00 provides a sufficient condition for the Lorentz invariance of the Smatrix.28) .120 3 Scattering Theory Eq. and 'P ~± are eigenstates of Ho and H respectively with the same eigenvalue Ea.25) where 0(+00) is according to (3. (3.24) as was to be shown. This smoothness condition is a natural one. and we find: (3.3.22) with 1: = 0 and "'0 = +00 to show that Krl(+00) ~ rl( +00)Ko .3. (3. (3.1.3.21) (which is equivalent to Eq.
3 Symmetries of the SMatrix 121 Eqs.3.25)(3.P2 U2 fi 2. (3.32) Finally.(T)"'.". (3. taking the scalar product of the states obtained by acting with U(T) on two different 'in' states or two different 'out' states.31) Also. Internal Symmetries There are various symmetries. x 'PPlll"lill.P2ll"21l2.2).'" " ""'N°. Po.29) with U(T)..28) show that with Our assumptions.3. J. (3. Plll"IN1. .N. we see that @(T) must be unitary ".(T)"· "'II.'" = Sp'Ill".p. taking the scalar product of the states obtained by acting with U(T) on one 'out' state and one 'in' state shows that 9 commutes with the Smatrix.P2U2N2.Pl<l"lnl.'" (3.n.12).. or the 'chargeconjugation' symmetry between particles and antiparticles.3.lTj'" • '" x Sp'Ill".ll".3.3. P. and (3. then some other transfonnation T. Acting on Eq.. (3.J L.30) where TT is the transformation obtained by first performing the transformation T.'" .". and H satisfy the same commutation relations as Ko.29) In accordance with the general discussion in Chapter 2. L. that have nothing directly to do with Lorentz invariance. like the symmetry in nuclear physics under interchange of neutrons and protons.J 1111 "2"2 1 1 (T) i'tIl"..N. that induces linear transformations on the indices labelling particle species U(T)\{IPUTlnl.. and using the normalization condition (3. in the sense that NINe N.3.P2ll"2n2.3. the U(T) must satisfy the group multiplication rule U(T) U(T) ~ U(TTj. we see that the matrices::» satisfy the same rule "'(T)"'(T) ~ "'(TT).3. Also. and H o. Jo. (3.'" .3. Such a symmetry transformation T acts on the Hilbert space of physical states as a unitary operator U(T). and further appear the same in all inertial frames.N.1.t(T) ~ Sr'(T).3. This is why it turned out to be unnecessary in proving the Lorentz invariance of the Smatrix to use the other commutation relations (3.'" = L nlnr' @filnl(T)9 fi2n2 (T)'" .. 'in' and 'out' states do transform under inhomogeneous Lorentz transformations just like the freeparticle states. since these are similarity transformations.Il.p~ll".15) that involve K.13).33) .3.. we now see that the exact generators K. (3.3... (3.
3..13).33) we still need to show that the . (3.. ll)PI t11'~'1 . N2n. (3.1. we see that the operator Uo(T) will induce the transformations (3.3. minus the number of their antiparticles) but as we shall see in Volume II.+q'+'~q q" j " 2 "1"2 +q + . P2(J]N2 . in this case the corresponding Hilbertspace operators must take the form U(T(O)) ~ exp(iQO) (3.34) and that commutes with both the freeparticle and interaction parts of the Hamiltonian Uo1(T)HOUo(T) ~ Ho. = L NINo"  !!2 NI "1 (T) q.3. (3.3. all known processes conserve baryon number (the number of baryons. and neutrinos. where T is a function of a single parameter 0. This will be the case if there is an 'unperturbed' transformation operator Uo(T) that induces these transformations on freeparticle states. (3.3. I A special case of great physical importance is that of a oneparameter Lie group. such as protons. such .40) The classic example of such a conservation law is that of conservation of electric charge. .39) where q" are a set of real speciesdependent numbers. because to derive Eq.36) From either the LippmannSchwinger equation (3.1.3. (3.122 3 Scattering Theory Again.37) As shown in Section 2. these conservation laws are believed to be only very good approximations.>( T) . and hyperons. neutrons.3.3. so that we can derive Eg... Here Eq.17) or from (3. minus the number of their antiparticles) and lepton number (the number ofleptons.29) on 'in' and 'out' states as well as freeparticle states. muons. ! particles.35) (3. with T(O)T(B) ~ T(O  + 0). U01(T)VUo(T) ~ V. Likewise the matrices fi(T) take the form (3. Also.29) on both 'in' and 'out' states. (3..~ame unitary operator U(T) will induce the transformation (3.3.33) simply tells us that the q s are conserved: Sflx vanishes unless . There <:Ire other conservation laws of this type that are definitely only approximate." (3. Uo( T)ll)PIO"I nl : P2"'2"Z.38) with Q a Hermitian operator..3..3. this is a definition of what we mean by a theory being invariant under the internal symmetry T.29) taking U(T) as Uo(T).3.2. such as electrons.
These and t) ~.4.2.with T = 1 and t3 = +1. and the AD hyperon with T = and t3 = 0. as in reactions like n+ + n + K+ + AO.~. This relation was originally inferred from observed selection rules. while the more familiar protons. :E+. just like the degenerate spin multiplets required by rotational invariance.0. the baryon number B. neutrons. based on the nonAbelian group SU(3). like the covering group of the group of threedimensional rotations. The conservation of strangeness in strong interactions explains why strange particles are always produced in association with one another. . and n mesons (or pions) are taken to have strangeness zero. These examples illustrate the relation between electric charge Q. quantum chromodynamics.:E.s and Imit baryon rmmber. it requires particles to form degenerate multiplets lahelled with an integer or halfinteger T and with 2T + 1 components distinguished by their t3 values. but it was interpreted 8 by GellMann and Ne'eman in 1960 to be a conse• quence of the embedding of both the isospin T and the 'hypercharge' y = B + S in the Lie algebra of a larger but more badly broken nonAbelian internal symmetry.18): [ti. As we will see in Volume II.1 . while the relatively slow decays of strange particles into nonstrange ones such as AD + p + n. • Superscripts <Jenote e1ectnc ch'ITgcs in units of the absolute value of the electronic charge. and n. its generators are denoted ti. Mathematically. :E0. The classic example of a 'nonAbelian' symmetry whose generators do not commute with one another is isotopic spin symmetry.tj] = if'ijktk.K+ and K Oare assigned strangeness +1 and the hyperons AD. the mesons now called. the pions n+. which was suggested 6 in 1937 on the basis of an experimcne that showed the existence of a strong protonproton force similar to that between protons and neutrons. which was introduced to explain the relatively long life of a class of particles discovered by Rochester and Butler5 in cosmic rays in 1947. include the nucleons p and n with T nO.3.3.are assigned strangeness 1. today both isospin and SU(3) symmetry are understood as incidental consequences of the small masses of the two or three lightest quarks in the modern theory of strong interactions. the third component of isotopic spin t). with j = 1. and satisfy a commutation relation like (2.and K+ + n+ + nO show that the interactions that do not conserve strangeness are very weak.3 Symmetries ql the SMatrix 123 as the conservation of the quantity known as strangeness. and the strangeness S: =! = ° Q ~ . For instance. To the extent that isotopic spin symmetry is respected. the group is SU(2).] + (B + 5)/2. A 'hyperon" is any particle carrying nonzero strangene.
(3.3. I . an operator P satisfying Eq.124 3 Scattering Theory The implications of isotopic spin symmetry for reactions among strongly interacting particles can be worked out by the same familiar methods that were invented for deriving the implications of rotational invariance. (This is for massive particles.33) requires that the Smatrix may be put in the form (suppressing all but isospin labels): + StnlDJ/AJIBJ = L CTcTD(Tt).3. This is because we are always free to redefine P by combining it with any conserved internal symmetry operator. respectively. I'. there must exist a unitary operator P under which both 'in' and 'out' states transform as a direct product of singleparticle states: (J. and . T. for a twobody reaction A + B __ C D. but neither can provide a unique determination of the '7 s. to. may be inferred either from dynamical models or from experiment. tV) .3.42) Just as for internal symmetries. Eq. Parity To the extent that the symmetry under the transformation x ~ x is really valid. because this symmetry is not respected by electromagnetic (and other) interactions. and ST is a 'reduced' Smatrix depending on T and on all the suppressed momentum and spin variables. (3. is only approximate. 4. if P is conserved. tA3 t B3)ST . totD3)C T TB( Tt 3.41) will actually exist if the operator Po which is defined to act this wayan freeparticle states commutes with V as well as Ho. The phases 1/.3. (J.t] where C jjj2 (jcr.) The parity conservation condition for the Smatrix is then. this.:JjJ reverses the space components of pp. . tBJ. Of course. For instance. In particular. like all of the consequences of isotopic spin invariance.. . the modification for massless particles is obvious. cr] (j2) is the usual ClebschGordan coefficient9 for fanning a spin j with three·component (T from spins )1 and h with threecomponents at and (J2. as shown for instance by the fact that different members of the same isospin multiplet like p and n have different electric charges and slightly different masses. but not on the isospin threecomponents tAJ. .4l) where '7/1 is the intrinsic parity of particles of species n.
3. consisting of the operators exp(ict(B + L)) with arbitrary real a. so as far as we know. once we have done this the intrinsic parities of other particles like the charged pion (which can be emitted in a transition n _ P+l'C) are no longer arbitrary. the operator p2 behaves just like an internal symmetry transformation: If this internal symmetry is part of a continuous symmetry group of phase transformations. and y. p. If this is true. in which case the intrinsic parities can only take the values + 1. In any case.. P. This is conserved to the same extent as P. if p2 = exp(ictB + . All known particles of halfinteger spin have odd values of the sum B + L of baryon number and lepton number. L. hence either P or P' could be called the parity operator. say lp. and electron have different combinations of values for B. The neutron. and electric charge. whether or not p 2 = 1. where B. However.. if p 2 = (It then P can be . (For instance. The foregoing remarks help to clarify the question of whether intrinsic parities must always have the values + 1. and Q are respectively baryon number. The only sort of theory in which it is not necessarily possible to define parity so that all intrinsic parities have the values + I is one in which there is some discrete internal symmetry which is not a member of any continuous symmetry group of phase transformations. proton. andy are arbitrary real phases. it is a consequence of angularmomentum conservation that the total number F of all particles of halfinteger spin can only change by even numbers. and has an inverse square root exp(in(B + L)j2). then take I p = exp(!iaB + . and ct. (It = (_l)B+L. however. such as the group of multiplication by the phases exp(iaB + ifJL + iyQ) with arbitrary values of ct. so by judicious choice of the phases ct. the intrinsic parity of any particle like the neutral pion nO which carries no conserved quantum numbers is always meaningful. p. then its inverse square root must also be a member of this group. Also. but rather may differ from it by a phase transformation of some sort.). the parity operator that is conserved may not be this one. so there is no reason why we should not call this the parity operator. with I~p2 = 1 and [Ip.. and Q. ). and y we can define the intrinsic parities of all three particles to be + 1. then (It is part of a continuous symmetry group.PIp with p I2 = 1.) We can then define a new parity operator pi . lepton number. so the internal symmetry operator (_I)F is conserved.P] = O. In this case. L. It is easy to say that space inversion P has the group multiplication law p 2 = 1.3 Symmetries of the S·Matrix 125 then so is pi _ Pexp(ictB + ipL + iyQ). lO For instance.
as would be expected from the symmetry (isospin invariance) among these three particles. (The triple scalar product PI' (P2 X p)) formed from the three pion momenta vanishes because PI + P2 + P3 = 0. In this case.3.) For the same reason. and as we have seen we can take the neutron and proton to have the same intrinsic parity. of course. so we can conclude that the intrinsic parities of the particles in this reaction must be related by: t ! '1d11n. S = 0 . It follows from Eq. it was observed ll in 1951 that a pion can be absorbed by a deuteron from the t = 0 ground state of the nd atom. = 1. S = 1 . we would have p4 = 1. with j = and B + L = 0).) The initial state has total angularmomentum j = 1 (the pion and deuteron having spins zero and one. the matrix element is odd under reversal of the direction of all threemomenta. which are allowed by angularmomentum conservation. respectively).= I. The n+ and nO have also been found to have negative parity. and we can conclude that '1.+ d 10 n + n. so the final state must have orbital angularmomentum t = 1 and total neutron spin s = L (The other possibilities t = 1. A spin zero particle that decays into three pions must have intrinsic parity '1. then the Smatrix must be respectively even or odd overall in the threemomenta. then it would be possible to have p2 = (I without our being able to redefine the parity operator itself to have eigenvalues + 1.126 3 Scattering Theory redefined so that all intrinsic parities are +L However. because in the Lorentz frame in which the decaying particle is at rest. In particular. rotational invariance only allows the matrix element to depend on scalar products of the pion momenta with each other.. that is. S = 1. in the reaction Jr. a spin zero particle that decays into two pions must have intrinsic parity 11. among the strange particles discovered in the late 19408 there seemed to be two different particles of . t = 0.) Because the final state has t = 1. The negative parity of the pion has some striking consequences. <:lnd t = 2. or equal to minus this product. so '1d = '1~. = + 1.= '1'1' 2 The deuteron is known to be a bound state of a proton and neutron with even orbital angularmomentum (chiefly t = 0). the orbital angularmomentum quantum number (can be used in relativistic physics in the same way as in nonrelativistic wave mechanics. (As discussed in Section 3. are forbidden by the requirement that the final state be antisymmetric in the two neutrons.7. (3.. so all particles would have intrinsic parities either +1 or (like the Majorana neutrino) +i. if there were to be discovered a particle of half~integer spin and an even value of B + L (such as a socalled Majorana neutrino. all of which are even under reversal of all momenta.42) that if the product of intrinsic parities in the final state is equal to the product of intrinsic parities in the initial state. For instance. the negative pion is a pseudoscalar particle.
(3.6 that the timereversal operator T acting on a oneparticle state \f'p.. while the other. As we have seen. + fJ is invariant under the reversal of direction of all threemomenta. Nevertheless. with proportionality factors that are invariant under reversal of all threemomenta.) The electrons were found to be preferentially emitted in a direction opposite to that of the spin of the decaying nucleus. for reasons discussed in Section 12. was identified by its decay into three pions. so Eq.. For example.42) have no effect on IS{Jof.12 As we shall see in detail in the next section of this chapter. Wu together with a group at the National Bureau of Standards measured the angular distribution of the electron in the final state of the beta decay C060 + Ni6() + e. parity is conserved in the strong and electromagnetic interactions. TimeReversal We saw in Section 2.n gives a state \f'&p. (now known as the K+) and that parity is simply not conserved in the weak interactions that lead to its decay. it became clear that parity is indeed not conserved in the weak interactions responsible for these decays. + fJ (with a. neutrino. following theoretical suggestions by Lee and Yang. of course.42) would imply that the rate for a.states '1.3. (3. the T. As long as the.3. and hence was assigned a parity 1. but it is a nontrivial restriction on rates in more complicated processes. be impossible if the decay rate were invariant under a reversal of all threemomenta. this is a trivial consequence of rotational invariance for the decays of a K meson into two or three pions. and antineutrino. A multiparticle state transforms as usual as a direct product of oneparticle states. which would. times a phase (n(I)ja. the 0.. A similar result was found in the decay of a positive muon (polarized in its production in the process n+ + Jl+ + v) into a positron. =1= fJ) is proportional to ISpx1 2. After many suggested solutions of this puzzle. was identified by its decay into two pions and was assigned a parity +1. and proposed that the T and 0 are the same particle.. and therefore continues to play an important part in theoretical physics. The trouble with all this was that as the T and f:I were studied in greater detail.+ v with a polarized cobalt source.5.. Lee and Yang in 1956 finally cut the Gordian knot..3 Symmetries oj the SMatrix 127 zero spin (inferred from the angular distribution of their decay products): one. they seemed increasingly to have identical masses and lifetimes. 13 (No attempt was made in this experiment to measure the momentum of the antineutrino or nickel nucleus. the phase factors in Eq. except that since this .n with reversed spin and momentum. the rate for a physical process·'1. 14 In this way.3. and fJ contain definite numbers of particles of each type.
: ..Ho + iErlT'P~ .T:y!I.50).44) where fT indicates a reversal of sign of threemomenta and spins as well as multiplication by the phase factors shown in Eq. with obvious modifications being required for massless particles.43).3. qi! .P2 Cf 21l 2.'" • (3.13) or (3.'3"p2... operating on the LippmannSchwinger equation (3... (3.3.) It will be convenient to abbreviate this assumption as i TlU+ _ luf T.~(l + J dy [E:.47) i .'" .2( _l)J2 "2 . The S~matrix will satisfy this transformation rule if the operator To that induces time·reversal transformations on freeparticle states (3.3..3.3. . Because T is antiunitary.. (3..3.. (l)h "i 01 2 ••.¥Pl<Tjt1j. X SJ'PI 011 n].16) with T and using Eqs.3. T'I'lil. ' .43) (Again.46) ! I . or in more detail Sp'~IMI'pl~IM" 1"1"1' 2"2"2"'" .349) (3. (3.350) t In this case we can take T = To.Pl<l"2n2"  <"n]  l)jja j Y ( i.1.'[ n'l ..' (3. (3.1. we have ('I'li.1. For instance.16) to show that time·reversal transformations do act as stated in Eq.l" P2 . Pl"1 t1 I.'" '/ I  '"'1 (1)11 "1 (n. (3.n2  1)h(J2 . .. • • Note that in addition to the reversal of momenta and spins. J'P2 012 n2.a 2n 2. this is for massive particles.J'P'I . the role of initial and final states is interchanged.48}(3.44).45) I t I so the timereversal invariance condition for the S·matrix is (3. 128 3 Scattering Theory is a time·reversal transformation.3. GJ _l)JI a] (.. and use either (3. n.3. as would be expected for a symmetry involving the reversal of time. we have PP~ = tD..3.48) commutes not only with the freeparticle Hamiltonian (which is automatic) but also with the interaction: (3. 'I'~) ~ (T'I'~. we expect 'in' and 'out' states to be interchanged: pp± _!" ( Pl(JJ/lj.
3. (3.51) where S(l) is small. then this can be put in the form (1) 5fh = ~ JJ dY 10 ' dY' 5(0) 5(1). they are real because .46) does nat in general tell us that the rate for the process a _ f1 is the same as for the process Yo: ..YfJ. In some cases S(O) is simply the unit operator. while S{Ol happens to have matrix element zero for some particular process of interest.3.3. _ P and ff a .Y fJ are thus the same if summed over sets .3. and similarly for . ~ Q(oo)oI>".46).51) in cases of this type.+ v. fl'/ y}.3. Section 3. this gives a reality condition (3. so that TQ(oo)oI>. However. the timereversal invariance condition (3.53) Since S(Ol is unitary. In contrast with the case of parity conservation. (For instance. N _ N' + e.17 consisting of just one state each.44).I.3.) In the simplest case we have 'complete' sets . again leading to Eq. and either a or af are in oF. the rates for the processes r:J. something like this is true in cases where the Smatrix takes the form SPrt = S~~ + Sft~ .3.3. then both states are in. the unitarity condition for the Soperator reads 1= st s = S{O)tS(Ol + S{O)tS(ll + S(1ltS(O).3.':9'.44).F and /F of final and initial states that are complete with respect to S(O). with S{Ol the Smatrix produced by the strong nuclear and electromagnetic interactions alone. This is just the LippmannSchwinger equation for 'P~!t' thus justifying Eq. (3.3 Symmetries a/the S~MQtrix 129 with the sign of +iE reversed because T is antiunitary. (The 6rt and 6{J are called 'phase shifts'. though it generally has much larger matrix elements than Sli). both the initial and the final states are eigenvectors of S{Ol with eigenvalues e 2io" and e 2iOp .F .r·1 5.52) If S(1) as well as SlO) satisfies the timereversal condition (3. because T is antiunitary it changes the sign of the j in the exponent of Q(t).) To first order in SOl. that is.0:' (3. and S(ll the correction to the Smatrix produced by the weak interactions. respectively. (3. the process might be nuclear beta decay.5 shows how the use of the 'distortedwave Born approximation' leads to an Smatrix of the form (3.F and . (By being 'complete' here is meant that if S~?~ is nonzero. = Using the zerothorder relation S(oJt S(Ol for S(l): 1. Similarly.
54): S(1) Pel = _ei(b. for instance.55) which just says that iS~~.rfI .54) reads (3. timereversal invariance is consistent with the observation that electrons from the decay C0 60 _ Ni 60 + e. has the phase bel + op mod n.3. This is known as Watson's theorem. Eq. they did show immediately that the product PT is not conserved.3. If conserved this operator would have to be antiunitary for the same reasons as for T. (3. ' . As described below.l 4 that discovered the nonconservation of parity.) In this case.+b~)S(1)·_ . ls The phases in Eqs.p)S{l)o :7'1' :rp (3354) .'1p ~ p. the angular distribution of the pion relative to the A spin involves the interference between these states. for which Eq. For instance. so in processes like nuclear beta decay its consequences would take the form of relations like Eq.3. In some cases we can use a basis of states for which /7 li = li and . the final state can only have orbital angularmomentum t = 0 or t = 1. This is the case for instance in nuclear beta decay (in the approximation in which we ignore the relatively weak Coulomb interaction between the electron and nucleus in the final state).+<. because both the initial and final states are eigenstates of the strong interaction Smatrix (with 0:>: = op = 0).54) or (3. PT Although the 1957 experiments on parity violation did not rule out timereversal invariance.55) may be measured in processes where there is interference between different final states. (3.130 s(O) 3 Scattering Theory is unitary.3.>. and it is clear that the absolute value of the Smatrix for the process a + # is the same as for the process "Fa + :rp. indirect evidence against timereversal invariance did emerge in 1964..'7 CI. (3.3. and hence according to Watson's theorem depends on the difference Os op of their phase shifts. but it remains a useful approximate symmetry in weak as well as strong and electromagnetic interactions. This prediction was not contradicted in the 1956 experiments 13. (3..+ II are preferentially emitted in a direction opposite to that of the C060 spin. the differential rate for a beta decay process should be unchanged if we reverse both the momenta and the spin zcomponents (J of all particles.53) becomes simply: (l) S{h   _e 2i(". in the decay of the spin 1/2 hyperon A into a nucleon and a pion. Thus if timereversal invariance is respected.3.o/'.
In reactions involving only completely neutral particles. so the observation of the neutral pion decay nO _ 2'{' requires that 1111° = + 1.3. such as exp(bB + if3L + irQ). the only particles whose chargeconjugation parities are individually measurable are those completely neutral particles like the photon or the neutral pion that carry no conserved quantum numbers and are their own antiparticles. (3.T reverses the signs of all spin zcomponents but not any momenta. in this case.57) As with other internal symmetries.3.57) tells us that the product of the chargeconjugation parities in the initial and final states must be equal. For these two particles. C.3. Neglecting the finalstate Coulomb interaction. as is in fact known to be the case. as we shall see the photon is required by quantum electrodynamics to have chargeconjugation parity 'I}' = 1. it then follows that the process nO _ 3r should he forbidden. the Smatrix will satisfy this condition if the operator Co that is defined to act as stated in Eq. either +1 or 1. this will always be the case if all internal phase transformation symmetries are e" . known as chargeconjugation. which interchanges particles and antiparticles. whose effect on multiparticle states is: (J. (3. If this is true for both 'in' and 'out' states then the S·matrix satisfies the invariance conditions (J.56) where nC. it would then follow that there could be no preference for the electron in the decay Co6() _ Ni fio + e.J. The phases e" are called chargeconjugation parities.+ v to be emitted in the same or the opposite direction to the Co60 spin. the are in general not uniquely defined. for instance. CPo and CPT As already mentioned. Just as for ordinary parity.56) on freeparticle states commutes with the interaction V as well as Ho. (3. is the antiparticle of particle type n. we can find another such operator with different ~n by multiplying C with any internal symmetry phase transformation. Just as for the ordinary parities 1]". Eq.3.56).3. and ~" is yet another phase. because for any operator C that is defined to satisfy Eq. there is an internal symmetry transformation. in contradiction with what was observed. the chargeconjugation parities are real. this entails the existence of a unitary operator C. we take C = Co.3 Symmetries of the SMatrix 131 where f!/. Formally.
the argument may be equally well be based on CP conservation. If we arbitrarily define the phases in the CP operator and in the KO and KO states so that CP"PKO = "P K{) and CP"P KO = "P KO then we can define selfchargeconjugate oneparticle states 1 "PKo. with the result that the new C satisfies C2 = 1.3. (As we shall see below.) It is understood today that C as well as P is not conserved in the weak interactions responsible for processes like beta decay and the decay of the pion and muon. For general reactions. Although the early experiments on parity nonconservation indicated that neither C nor P are conserved in the weak interactions. because then we can redefine C by multiplying by the inverse square root of the internal symmetry equal to C2. In 1954 GellMann and Pais 16 had pointed out that because the KO meson is not its own antiparticle (the KO carries a nonzero value for the approximately conserved quantity known as strangeness) the particles with definite decay rates would be not KO or KO. This had particularly important consequences for the properties of the neutral K mesons. but with C not conserved in the weak interactions. they left open the possibility that their product CP is universally conserved. Eq. For some years it was expected (though not with complete confidence) that CP would be found to be generally conserved. 'i 2 and . not just in the particular theory considered by Lee and Yang. but the linear combinations KO + KO.57) requires that the rate for a process equals the rate for the same process with particles replaced with their corresponding antiparticles. the observed violation of TP conservation would imply a violation ofC conservation in any field theory of weak interactions. This was not directly contradicted by the 1957 experiments on parity nonconservation (it will be a long time before anyone is able to study the beta decay of anticobalt).n['PKo+"Pi O] . This was originally explained in terms of C conservation. (3. though both C and P are conserved in the strong and electromagnetic interactions. but these experiments showed that C is not conserved in the theory of weak interactions as modified by Lee and Yang 12 to take account of parity non~conservation.132 3 Scattering Theory members of continuous groups of phase transformations.
5 that (without any approximations) the total decay rate of any particle is equal to that of its antiparticle. These theories were field theories. although it seems more nearly conserved than C or P individually. at least in any quantum field theory. Because CPT is antiunitary. there are good reasons to believe that although neither C nor CP is strictly conserved. !lnd hence P = +1.0 may differ from the partial rates for the decay of the antiparticle mto the corresponding final states "Cf?fJ:Yflt and "C&':!Ifh. and in particular it is the fact that CPT commutes with the Hamiltonian that tells us that stable particles and antiparticles have exactly the same mass. and also for an t = n+ n. Nevertheless. Further. and therefore automatically conserved CPT. It is CPT that provides a precise correspondence between particles and antiparticles. As we shall see in Chapter 5. C = +1 for two nOs becanse the nO has C = +1. in cases where the Smatrix can be divided into a weak term S(l) that produces a given reaction and a strong term S(O) that acts in the initial and final states. In particular. ° . and neutrino. although the partial rates for decay of the particle into a pair of final states fh. but CP conservation would allow this only" for the K 1.3. fh with S~~12 I. it relates the Smatrix for an arbitrary process to the Smatrix for the inverse process with all spin threecomponents reversed and particles replaced with antiparticles. we can use the same arguments that were used above in studying the implications ofT conservation to show that the rate for any process is equal to the rate [or the same process with particles replaced with antiparticles and spin threecomponents reversed. respectively. The K~ would thus be expected to decay only by slower modes. Since the experiments showed that PT conservation but not T conservation is badly violated in nuclear beta decay. However.state because C interchanges the two pions. any theory that was consistent with these experiments and in which CPT is conserved would have to also incorporate C but not CP nonconservation. •• The neutral Xmesons have spin zero. 17 The conclusion was that CP is not exactly conserved in the weak interactions. muon or electron. so the twopion final state has t = 0. not the K2.3 Symmetries of the SMatrix 133 which have CP eigenvalues +1 and 1. We can now understand why the 1957 experiments on parity violation could be interpreted in the context of the existing theory o[ weak interactions as evidence that C conservation as well as P conservation are badly violated but CP is not. the product CPT is exactly conserved in all interactions. we shall see in Section 3. into three pions or into a pion. it was found by Fitch and Cronin in 1964 that the longlived neutral K meson does have a small probability for decaying into two pions. provided that we sum over sets of initial and final states that are complete with respect to S(O). The fastest available decay mode of these particles is into twopion states.
(2n)3 p'. using wave packets to represent particles localized far from each other before a collision. actually more a mnemonic than a derivation. we can take this box as a cube. with the excuse that (as far as r know) no interesting open problems in physics hinge on getting the fine points right regarding these matters. which ensures the conservation of the total energy and momentum. This has since been verified 18 by more detailed studies of the KO~Ko system. and then following the timehistory of these superpositions of multiparticle states.n2. but what does this have to do with the transition rates and crosssections measured by experimentalists? In particular.p.2) thus implies that the states we have been using have scalar products in a box which are not just sums of products of Kronecker deltas.P"J. Then all threedimensional de1tafunctions become 03 ( ' ) I d3 i(pp')x _ V J v p . and L 3 = V.. 2n (3. 3. In what follows we will instead give a quick and easy derivation of the main results.'. equal to one if the subscripts are equal and zero otherwise. r (34 2) .4.2) shows that Spa has a factor 04(pp .134 3 Scattering Theory Similarly.3.p ~ (2n)3. (3.1. but with points on opposite sides identified. but also contain a factor [V j(2n)3]"'. but it has so far been impossible to find other direct evidence of the failure of invariance under timereversal. We consider our whole system of physical particles to be enclosed in a large box wi~h a macroscopic volume V. For instance. so what are we to make of the factor [J4(pp _ p.n3).4 Rates and CrossSections The Smatrix SPit is the probability amplitude for the transition rx ) fi. where N is the number of particles in .1v x e .:)] 2 in the transition probability IS/h:1 2 ? The proper way to approach these problems is by studying the way that experiments are actually done.1) where the ni are integers. so that the singleMvaluedness of the spatial wave function requires the momenta to be quantized p ~ L(n1. the observation in 1964 of small violations of CP conservation in the weak interactions together with the assumed invariance of all interactions under CPT allowed the immediate inference that the weak interactions also do not exactly conserve T. where op"p is an ordinary Kronecker delta symbol.. The nonnalization condition (3.
(3. is a product of Kronecker deltas. To calculate transition probabilities we should use states of unit norm.4) where 0PC/.5) where S:. so let us introduce states normalized approximately for our box (3. one for each threemomentum. One immediate consequence is that the energy~conservation delta function is replaced with 3T (E. We shall define the finalstate interval dfJ as a product of d3 p for each final particle.Ep)t ) dt. plus terms with particles permuted. .4.n3 for which the momentum (3.1) lies in the momentumspace volume d3 p around p. To calculate a meaningful transition probability we also have to put our system in a 'time box'. the Smatrix may be written S{lC/. = [V j(2n)3rN~+N~)/2s:.3. is found in a state {l after the interaction is turned off.4. because this is the number of triplets of integers nj.8) Hence the total probability for the system to wind up in a range dP of final states is (3.x is calculated using the states (3.¥BOX .6) The probability that a multiparticle system.4.3). Correspondingly.1 2n T12 1 T 2 / exp ( irE. so the total number of states in this range is dX jl ~ [v /(2n)3(" dP .4. which is in a state rJ.4 Rates and CrossSections 135 the state. then every possible transition will occur again and again.¥BOX) = (5 {I' 'x ( Ih (3.3) with norm . (3.4. if we just leave our particles in the box forever. The number of oneparticle box states in a momentumspace volume d3p is Vd 3pj(2n)3. We suppose that the interaction is turned on for only a time T. . spin. (3. is (3.4.4.4. and species label.9) .E p) ~ .7) This is the probability for a transition into one specific box state fl.4. before the interaction is turned on. Of course.n2.x.
) For such states.)dp . this means that we are considering only the connected part of the Smatrix.)]' ~ oT(Ep 1 E.4. which gives the transition rate for a singleparticle state Ci. We will come back to the interpretation of the factor J4(plX . (T /2rr)IMp. ~ ".11) This is the master formula which is used to interpret calculations of Smatrix elements in tefms of predictions for actual experiments.)Mp.2 V 1.)V /(2rr)3 .12) P) dP(.p.p. but that also satisfy the more stringent condition. In this limit. (3. (In the language to be introduced in the next chapter.)".10) Our introduction of the box allows us to interpret the squares of delta functions in IS/I. the transition probability is simply proportional to the time T during which the interaction is acting.. we may define a delta functionfree matrix element M{J(J.4.E..(E p E. (3.1 2 x o~(pp .: 5p.p.N 'IMp.4.1 2 0'(pp . (3.E.p(l).p.=I: Here the volume V cancels in Eq.13) . the delta function product here may be interpreted as an ordinary fourdimensional delta function . If we let V and T be very large.)oT(Ep . that no subset of the particles in the state fJ (other than the whole state itself) has precisely the same fourmomentum as some corresponding subset of the particles in the state a.)T /2rr.p.4. (3.E. There are two cases of special importance: Na. 2irro~(pp . with a coefficient that may be interpreted as a differential transition rate: dl'(' ~ where now (3. (0) [oT(E p .136 3 Scattering Theory We will restrict our attention throughout this section to final states f3 that are not only different (however slightly) from the initial state 11:.a as [o~(pp  Pol]' ~ o~(pp .9) gives a differential transition probability dP(" ~ P) ~ (2rr)2 [(2rr)3/Vr.4. so Eq.54(pfJ .)oT(Ep .)dP .4.)o~(OI ~ o~(pp. ~ PI/T ~ (2rr)3N.PfJ)df3 later in this section.11).1 2 for fJ i=. to decay into a general multiparticle state p as (3.
3. We next take up the question of the Lorentz transformation properties of rates and crosssections. (3.3.EtJ). also known as the crosssection. (3. and sum over all spins.1) for the Smatrix is complicated by the momentumdependent matrices associated with each particle's spin.4. this makes sense only if the time T during which the interaction acts is much less than the mean lifetime tel of the particle a. so we cannot pass to the limit T + [jJ in i5T(E~ . and some of them are very important in chemistry.) . consider the absolutevalue squared of (3. which will help us to give a more general definition of the relative velocity U a in Eq. two protons and an electron turn into a deuteron and a neutrino.1264(PI_ p.4. the sum is Lorentz~invariant. (By flGl E and TIp E is meant the product of all the singleparticle energies po = J p2 + m2 for the particles in the states!'J.11) is proportional to I/V.13) is only useful if the total decay rate 1/rrx is much less than any of the characteristic energies of the process. (For instance. (3. The flux of either particle at the position of the other particle is defined as the product of the density IIV and the relative velocity UC(: ~rx = u':\O/V . so Eq.16) is a scalar function of the fourmomenta of the particles in states If.4.3.I' II E II E = Rp. the quantity L SpillS IMp.4. N rx =2: Here the rate (3.6 presents an application of the master transition rate formula (3. in one of the main reactions that release energy in the sun.3.1) (after factoring out the Lorentzinvariant delta function in Eq. (3. There is an unremovable width 6E =::::: liT ~ 1/t rx in this delta function.4. and {3. Experimentalists generally report not the transition rate per density.) Section 3. transition rates for N rx > 3 are all measurable in principle.11) for general numbers N a of initial particles. etc.14) (A general definition of U rx is given below. The unitarity of the matrices D~l(W) (or their analogs for zero mass) then shows that. The Lorentz transformation rule (3. and {3. That is.1). ~ (2n)4u~1IMI. fJ ':\0 (3.12)).4.4 Rates and CrossSections 137 Of course. but the rate per flux.15). or in other words.)dp. To avoid this complication.4. to the density of either particle at the position of the other one.4. (3.) Thus the differential crosssection is da(a ~ p) =dqa ~ P)/il>. astrophysics.15) Even though the cases Nrx = 1 and N(l = 2 are the most important. [or the moment we will content ourselves with specifying that if either particle is at rest then U rx is defined as the velocity of the other. apart from the energy factors in (3.
Uo = vEi m~/E2 = 1P2iIE2 .138 3 Scattering Theory We can now write the spinsummed singleparticle decay rate (3. It is conventional to define the crosssection to be (when summed over spins) a Lorentzinvariant function of fourmomenta. the slower it decays. and so Eq.. . (j4(pfJ .13) as I: dr(u ~ P) ~ 2nE. where the total three~momentum vanishes.p:. P2)2  mlm~/ £1£2 (3. . ..5. our result (3.>. . This uniquely determines U(l to have the value in general Lorentz frames· U(1 = J(PI .17) makes it obviou8 lhal EIE. when particle 1 is at rest. Also.4.) and here Eg. of course. (3. we have PI ~ (p. i f' f r I: II fJ .17) give. . . just the usual specialrelativistic time dilation . where Ert.Ed.)dPIII E. fJ The factor dP/ ITp E may be recognized as the product of the Lorentzinvariant momentumspace volume elements (2. so this means that we must define the relative velocity u" in arbitrary inertial frames so that u"E]E 2 is a scalar. and dfJ/TI/lE are already Lorentzinvariant.b4(pp spms p. so it is Lorentzinvariant. we have PI = O. which is Just the velocity 01' particle 2.17) where PI. so PI . Uo: is the velocity of the other particle.I R1 . > ! 1 .4.2.1 Ell E. is the energy of the single initial particle.the faster the particle. So also are Rfi~ and J4(pp . P' ~ (p. OUf conclusion then is that the decay rate has the same Lorentz transformation property as l/E I1 This is.15). we note that in the 'centerof~mass' frame.15) for the spin·summed crosssection may be written as du(a ~ PI ~ (2n)4 u.) E1E2 PI P2 £1 E2 (H18) • Eq.Pa)..17) gives u" = Ipl(EI + E.ppldP I E . P2 = ml F..4.b 4(p.u o is a scalar. We also mentioned earlier that in the Lorentz frame in which one particle (say.4. particle 1) is at rest.. (3. £1 = ml. P2 and m].E. m2 are the fourmomenta and mass of the two particles in the initial state a. (3.4.J.4.I Rp. Spill' where £1 and £2 are the energies of the two particles in the initial state (1. leaving just the noninvariant factor 1/ EC1 . . As a bonus. Similarly. The factors RfJct.
.p.4 Rates and CrossSections 139 as might have been expected for a relative velocity. and dO .xo)/ If'(xo)1 ' e where f(x) is an arbitrary real function with a single simple zero at x = xo.4.4. (3..p. (3.' P2=PI E) Ipi 'I'dipi 'Idn. + E of the delta function in Eq. .15) for decay rates and crosssections.19) (For N:x = 1.' + Vipi '1 2 + m.E)d3p~ d3p~ . This can be simplified by using the standard fonnula b(f(x)) ~ b(x .21) gives (j4(pl1 . . it can take values as large as 2.4. . in particular.4. .p(t)dfJ.4.4..  E)d 3 p2' In more detail.22) PI = P 2 . ..23) £. Here (3. We can similarly use the remaining delta function to eliminate anyone of the remaining integrals. this is J4(pp .. P~ E)d3p~'" (3.11) for transition b rates. where the total threemomentum of the initial state vanishes p.3. + . (3.p. Eq.sin2 dO d¢ is the solidangle differential for PI '.Jd{/  b(Ej + E..21) with the understanding that wherever replaced with appears (as in ED it must be (3...4.) If the final state consists of particles with momenta P~' P 2" . ')D(Ej + E. is the total energy of the initial state.13) and (3. (3. say over P'l' can be done trivially by just dropping the momentum delta function J4(PI! _ p.  . in this frame UCl is not really a physical velocity. ~ 0.)dp ~ b3 (p ~ +p ~ + . .. We here specialize to the case of the 'centerofmass' Lorentz frame. . However. (3.4. the argument E. there are just two particles in the final state. (3. .18) shows that for extremely relativistic particles. We now turn to the interpretation of the socalled phasespace factor 4 (pf:! .)dp ~ . which appears in the general formula (3.P3 .20) where E _ E. . In the simplest case.)dp ~ b ( Vipi '1 2 + m. Anyone of the P~ integrals.23) where .. this is just the case of a particle decaying at rest. and also in Eqs.4. then b4(pp. + .4. In our case.4...)(Ei + E.
27).21) gives 4 3 3 b (Pfl . Eq.15) as drr(11 + fJ) dQ (2n:)4k'E.4. but there is one nice result for NfJ = 3 that is also worth recording.4.27) _ k' E'I + k' E' 2 _ k'E'I E'2 E We can thus drop the delta function and the differential Eg.E. 12 _ (2n:)4k'EiE2EIE2 E'k I Mp.25) (3.13) for decay of a oneparticle state of zero momentum and energy E into two particles IS ! i I .E 2 I Eu.4. pl d3 p3 = Ipll 2 dlplllp31 2 dlp31 d0 3 dcP23 dcos 823 .l=k' (3. For N p = 3.26). and E2 are given everywhere by Eqs.p.24) k' = J(E2 m. (3. ~~~~~ (3.4.29) 4 and the differential crosssection for the twobody scattering process 12 1'2' is given by Eq. . . '4( o pp  dlpl'l in (3. The above case Nfl = 2 is particularly simple. 12 .23). ~ Mp.4.4.. (3. I dr(ct 4 fJ) _ 2nk'EiE21M 2 dO E {Jet 1 (3. (3.JdfJ 4 d Pl d p3 x J (J(Pl +P3)2 +m? + JP1 2 +m? + VP32+m32 We write the momentumspace volume as d 3 E) • .30) where k =IpI] Ip21. where (3.140 has a unique zero at 3 Scattering Theory Ip'll = k'. i ! j .4.2 _m?)2 4m?m?/2E.4.4. the differential rate (3.4.28) PIl )dfi 4 k' Ei E2dO E with the understanding that k'. by dividing by (3.4. In particular.24}(3.4.26) with derivative d [dl PI 'I (VIPI '1 2+ m? + VlpI '1 2 + m2'  E)] ip. (3. (3.4.4.
19 because of its use by Dalitz in 1953 to analyze the decay K+ .3. If we approximate this scalar as a constant.. The orientation of the plane spanned by P2 and P3 is specified by </123 and the direction of p}. 3.. plotted in the E2 E} plane is uniform.h is the differential element of solid angle for P3' and 023 and rP23 are the polar and azimuthal angles of P 2 relative to the p} direction. This is known as a Dalitz plot. this is finally J4(pfJ . Eqs.18) give the Smatrix as SpCl = J(fJ !X) . is a scalar function of fourmomenta. including possible centrifugal barriers or resonant intermediate states.ECl)TpCl + Tprt+ = (<Dp. the distribution of events .7) and (3. (3. an expansion in powers of the interaction term V in the Hamiltonian H = Ho + V.5 Perturbation Theory 141 where df.4. obtained by summing IMpCll 2 over spins and multiplying with the product of energies.31) But recall that the quantity (3. (3.5 Perturbation Theory The technique that has historically been most useful in calculating the Smatrix is perturbation theory.2inb(E p . .4.16).. n+ + n+ + n.4.2.. The derivative of the argument of the delta function with respect to cos 023 ~o'C"OSC'e"23 eEi Ip 211p}1  E 1 so we can do the integral over cos 823 by just dropping the delta function and dividing by this derivative Replacing momenta with energies. V'f'rt+).. Any departure from a uniform distribution of events in this plot thus provides a useful clue to the dynamics of the decay process.31) tells us that for a fixed initial state. (3..PCl)dP 1 E 1E2E} dE 2 dE} d03 drP23 . then Eq.1. with the remaining angle 823 fixed by the energyconservation condition V2 + 21p 211p 31cos P 2 IS 823 + Ip}1 2 + mj2 + )IP21 2 + m22 + )lp3 2 1+m? = E.
.3).E y + /E)(£').5). CD).5.4) where V(t) exp(Hot)Vexp(iHot). (E(1 .5.. (3.x:}+IE ' . The easiest way to derive the timeordered perturbation expansion is to use Eq.5..1. For the most part in this book.exp(iHOT) exp( ~iH(T .2.Vlo + J. Vp. TO) .(t +111" 'E ...( + = <I>x + .. Differentiating this formula for U(t. V"7 I'. Its obvious drawback is that the energy denominators obscure the underlying Lorentz invariance of the Smatrix. (3. Operating on this equation with V and taking the scalar product with $/l yields an integral equation for T+ Tlo + . !l i' IE . + is obtained by iteration from Eg.. which has the virtue of making Lorentz invariance much more transparent. VI" C! .2) The perturbation series for TpC1. (3. (3. (3.5. + J d yd y V'i}' V/}'. (3. known as timedependent perturbation theory.TO)) exp( iHoto). Ey' . (3./ dy E T.E" + IE . s~ where U(w.Tr . which dominated calculations of the Smatrix in the 1930s.5.1) where (3. .1) TfJ'l + = Vflx + J I dr' Vp..3). we will rely on a rewritten version of Eq.3) The method of calculation based on Eg. is today known as oldfashioned perturbation Theory.. which gives the Soperator as ! I f I f i .. in clarifying the way that singularities of the Smatrix arise from various intermediate states.142 3 Scattering Theory where '¥" + satisfies the LippmannSchwinger equation (3.17): '¥(. U(r.+ dy E _ E +.5) . to) with respect to t gives the differential equation (3.5..5. however. It still has some uses.Ey' + IE) + . while somewhat obscuring the contribution of individual intermediate states.5.
by using the Fourier representation of the energy factors in Eq. with f" + 0+..)V(t. TO) = 1 .f: dt] .5 Perturbation Theory 143 (Operators with this sort of timedependence are said to be defined in the interaction picture.l~l dtz V(tt) V(t2) + (i)1l' dtl l~l dt2l:l dt3 o V(tt) V(t2) V(t3) + .5. each o[ which gives the same integral over all .5.) + 8(t.5. There is a way of rewriting Eq. (J.5.)V(t. (3.  t.. T(V(t)} ~ V(t) .[~ dtz V(tj)V(tz) (3.4) as well as the initial condition U(ro. For instance.3): (Eo: .8) dt] V(t. equal to +1 for r > 0 and to zero [or r < O. (3.5. The timeordered product of n Vs is a sum over all n! permutations of the Vs. This can also be derived directly from the oldfashioned perturbation expansion (3. and so on. T{V(t. + (_i)Z +(i)] L: l~.i l: dt V(t)U(t.5.)V(t. .) Eq.8) that proves very useful in carrying out manifestly Lorentzinvariant calculations.t.5.YJ dTexp(i(E Cl  E}')r) (3..3). (3..)V(t. the nextlatest next to the leftmost.n in the integrand. to distinguish their timedependence from the timedependence 0H(t) = exp(iHt)OH exp( iHt) required in the Heisenberg picture of quantum mechanics. and so on. '0).9) with the understanding that such integrals are to be evaluated by inserting a convergence factor e. TO) in powers of V U(r. (3.) V(t]) + .5.3.)V(t. TO) = 1 is obviously satisfied by the solution of the integral equation U(" '0) ~ 1. we obtain an expansion for U(T..i I:. dt.)} ~ O(tl .E" + if")l = l.i l: (itl V(tt) + (_i)2 lo' dtj .). dtl V(td dt.6) By iteration of this integral equation.) V(t. Define the timeordered product of any timedependent operators as the product with [actors arranged so that the one with the latest timeargument is placed leftmost. I: i . where O(r) is the step function.7) Setting T = 00 and TO Soperator: S = = ct:' then gives the perturbation expansion for the I.
in general (3. this is not usually the case.Jt'(Ax + a).X2 is spacelike. what we want is that the Soperator should commute with the operator Uo(A.)}.10) does not even converge. and is at best an asymptotic expansion in whatever coupling constant factors appear in V.5).T{V(td···V(t. let's try the hypothesis that V(t) is an integral over threespace V(t) ~/ d3x .X2)2 > 0. etc.5.8) may be written ~ (i)' 1~ S~l+L 11=1 n.5.13) introduces no special Lorentz frame if .5. and Ko. However Eq.. 20 This series can be summed if the V(t) at different times all commute.K(x) a scalar in the sense that Uo(A. Po Jo.13) Everything is now manifestly Lorentz invariant. (3. Of course.5. Equivalently.Jt'(xd·. I dttdt2 .t) (3.a): Ho.a) ~ .5. the sum is then S ~ exp (i 1: dt Viti) . This is sometimes known as the Dyson series. (3.. Since the elements of the Smatrix are the matrix elements of the Soperator between freeparticle states $a. 3 Scattering Theory so Eq.10) ex. the Soperator must commute with the generators or Uo(A.11) with . To satisfy this requirement. i. except for the timeordering of the operator product.5.)}. $.10) is sometimes written in the general case as S ~ T exp (i I: dt V(t)) with T indicating here that the expression is to be evaluated by timeordering each term in the series expansion for the exponential.12) (By equating the coefficients of a O for infinitesimal transformations it can be checked that Jf'(x) has a timedependence consistent with Eq. T{. Now.. unless (Xl .Jt'(x. so the timeordering in Eq. (3.Jt'(x)U"I(A. a) that produces Lorentz transformations on these freeparticle states.a). (3. dt. (3. (3.5. t l1 .e.144 t1 .B.. (3. the timeordering of two spacetime points Xl.5..X2 is Lorentzinvariant unless Xl .) Then S may be written as a sum of fourdimensional integrals S ~ ~ 1+ ~ ~! ( i)'/ a'XI'a'X.. We can now readily find one large class of theories for which the Smatrix is manifestly Lorentzinvariant.Jt'(x.5.
.18) This condition would follow from the 'causality' condition (3. (3. t)1 ~ tVJ1'(x. 3 where W jd3XXJf(·(X. VI ~ j d 3x j d y x [. 3 (3. It is this condition that makes the combination oj Lorentz invariance and quantum mechanics so restrictive. W]. The other condition for Lorentz invariance. but the most general Lorentz invariant theories are not very different. Lvrenlz invariance can be dislUrbed by troublesome singularities at x = x'. • We write the condition on x and x' here as (x ~ x')2 2': 0 instead of (x .5. the commutation relation (3. J1'(y.5. then the same is true of V.3 to give a formal nonperturbative proof that an interaction (3. there is always a commutation condition something like (3.5. J1'(x.K(x.3. as is necessary for the validity of scattering theory. is also valid if and only if o ~ [W.O) between eigen· states of Ho are smooth functions of the energy eigenvalues. = so integrating over x and setting t [Ko. (3. for which timeordering is always Galileaninvariant.5.14). t) . V] 0. (3. 0)] .3. t) + x :('""(X. j d x J1'(x. Theories of this class are not the only ones that are Lorentz invariant. For an infinitesimal boost. .5 Perturbation Theory 145 (though not only if) the Jf(x) all commute at spacelike or lightlike* separations: We can use the results of Section 3. because as we shall sec in Chapter 6.14) that needs to be satisfied.5. and also true of W. Eq.x ' )2 > 0.. but provides a somewhat less restrictive sufficient condition for Lorentz invariance of the Smatrix. 0)] ~ [Ho.17) If (as is usually the case) the matrix elements of JF(x.16) ~ [Ko.5..12) and (3.12) gives i [K o. (3. This condition has no counterpart for nonrelativistic systems.5. which is necessary in the proof of Lorentz invariance. 0).21). In particular.11) satisfying Eqs.5.0).5.14) does lead to an Smatrix with the correct Lorentz transfonnation properties.
5.2l) T pll + = (\{Jsp .16) as TfI.l + = (<DfI. (3. [V.I/l' VW \{l'1+)' (3.23).20) We can then write (3.21) is most useful when this second term vanishes: that is.).5.'+) = (\{IsfJ' Vs<1\. Vw'l'o+) + ('1'.146 3 Scattering Theory The methods described so far in this section are useful when the interaction operator V is sufficiently small. that is useful when the interaction conwins two terms v ~ V.i.x± as what the 'in' and 'out' states would be if Vs were the whole interaction + (EC(  Ho + iF)l V.iE)l V. this way of rewriting the T ~matrix becomes worthwhile when Vw is so weak that we may neglect its effect on the state \{Io: + in Eq. (3. The second term on the righthand side is just what TJlrx + would be in the presence of strong interactions <:llone T s {ic/+ . (For example. (3. just drop Vw everywhere in the derivation of Eq. • This is valid to first order in Vw.5. (3. This approximation is ubiquitous in physics.23) So far. + Vw \fIs:/ = ¢"l (3.. the matrix element (3. + Vw)'I'.5. In this approximation. this is all exact.w/)..23) becomes Tpo:+~(\{IsfJ' Vw'P. known as the distortedwave Born approximation. Vw~ 0: +) + ('1'. when the process rx . (3. +) ~ ('I'..)..24) ..) For such processes.. and hence replace 'Po: + with the state \{Iso: +.21). and so V. V.8 . (3.5.(E p .22) vanishes.5. There is also a modified version of this approximation. even though we cannot ignore the presence of the strong nuclear force acting in the initial and final nuclear states. the Smatrix element for nuclear .5. in nuclear beta decay we need a weak nuclear force to turn neutrons into protons.5.Ho .5. We can define ~s. which takes account only of the strong interaction V.Ho + iE)l(V.cD".'I'. V'P x +) ~ (['I'.22).5. for instance.5. (3. (3..22) (For a proof of Eq.p.p ] .(Ep .21) reads Tpy+ = ('P. + Vw)] '1'0 +) (3..I..5.(V.p .) Eq. so Eq. but to all orders in V.19) with Vw weak but Vs strong..f3 cannot be produced by the strong interactions alone.1.\{J. However.5. (3. VS\{l. Eq.(cD fJ .
4. The unitarity condition then gives 6(1' rL) = I d[J Sp"S{Jex = 6(1' ~ '1.6.2): Sfh = NP  '1.p.)IMp. this can be expressed as a formula for the total rate for all reactions produced by <In initial state ct.6. where the state II mayor may not be the same as the state .2nir5 4(pp . where it reads (3.. + 4n 2 J dP 6 4(p!l .. and tIC( is the total . ImM:l"" In particul<lr.C()M~/. (3. for fornrard scattering in an arbitrary multiparticle state rL to the total rate for all reactions in that state. 1 rr J dfi 3 4(pp_ p.'!J.'1.I' (3. we find that for P)' = p.y). 3.4) where U x 15 the relative velocity (3.4.2) ~ rr J dfi 34 (PI .1 2 Using Eq.3. (3.2 V 'N. and with 't's/l..and 'PsC( + the final and initial nuclear states. (3.6..)MprMfJ!J.11). . this can be written (3.6. + 2n dfJ 6 4(pfJ rL = J p!J.) ~ 2ni()4(Pr  pex)M. the Smatrix may be written as in (3.5.6 Implications of Unitarity The unitarity of the Smatrix imposes an interesting and useful condition relating the amplitude MC(!J.3) = __ (2n)3Nc2VlN.24) with V the strong s nuclear interaction and Vw respectively either the weak nuclear interaction or the electromagnetic interaction.C(.Pi' )J4(p!l .C( 0= iM.)M.6 Implications of Unitarity 147 beta or gamma decay is calculated using Eq.)IMp.p.JC( +2nib 4(Pr  p. + iM~.) . .yM!l&. in <I volume V r _ .p. J d" dr(' ~ fi) "dfi ~ (2rr)3N.'1.) <lnd a factor 2n6 4 (p" . Cancelling the term b(/' ..1 ) This is most useful in the special case 1m M" I'.. where rL is a twoparticle state.3.17) in state ct.pC()Mp.. Recall that in the general case.
3) (3.6. M p.9) Using (3. • The phase of J is conventional. (3. Eq. for elastic twobody scattering. U(X.30) shows that the differential crosssection for twobody scattering in the centerofmass frame is da(Ci.1.15) as a. We therefore define the scattering amplitude as· f(a ~ Pl =E 4n' (3.4. so there must be some solid angle .6) where k' and k are the magnitudes of the momenta in the initial and final states.).6.3) is known as the optical theorem. usually f is defined so that a ratio of final and initial velocities appears in the formula for the differential crosssection.E. within a factor 2) as in the forward direction. . The total crosssection is then bounded by a.6.p. (3.4.4n. In dil (2n)4k'E 1 2 E ElE2 2 kE2 IMp.8) In particular.I'6 4 (pl . (3..p). The scattering amplitude f may be expected to be a smooth function of angle.148 3 Scattering Theory crosssection in this state.1 .10) a) ~ k 4n a.6. J dP da(a ~ P)/dP ~ (2n)'u.6. This form of the unitarity condition (3. > J Ifl 2dil > ~ If (a ~ a)I'Ail > ~ IImf(a ~ a)I'Ail . given by (3. we have f(a ~ P) = . The normalization of f used here is slightly unconventional for inelastic scattering.5) This is usually expressed in a slightly different way.6. (3.18) for the relative velocity now reads Imf(a ~ the unitarity prediction (3.6.~.' J dP IMp.0 within which Ifl 2 has nearly the same value (say.7) so that the differential crosssection is simply dar:.6.. and is motivated by the wave mechanical interpretation 21 of f as the coefficient of the outgoing wave in the solution of the timeindependent Schrodinger equation. P) ~ If(a ~ Pll2 (3. 22 There is a pretty consequence of the optical theorem that tells much about the pattern of scattering at high energy. in terms of a scattering amplitude f(a .4.
12) where C(/fY:Y indicates that we must reverse all spin zcomponents.6. so Eq.6.. we can use Eq. (3. .15) In particular.3. (3.pp) in Spi7.'" ~ Mp'_u'nc·p'_lpn c . its conservation does not in general imply any simple relation between a process a + P and the process with particles replaced with their antiparticles.6.6. Instead. Since CPT invariance also requires particles to have the same masses as their corresponding antiparticles.PI (Tj rll .13) In particular. " p'(T'rl"p'(T'rl" ' . . 'I'" "2' 'I' . when the initial and final states are the same the phase factors all cancel. (3. we see that the decay rate of any particle equals the decay rate of the antiparticle with reversed ..13) says that M P1 (TI rll . . (3..11) As we shall see in the next section.6.. it provides a relation between a process and the inverse process involving antiparticles: we can use the same arguments that allowed us to deduce (3. The generalized optical theorem (3. total crosssections are usually expected to approach constants or grow slowly at high energy.11) shows that the solid angle around the forward direction within which the differential crosssection is roughly constant shrinks at least as fast as 1/k2 for k + roo This increasingly narrow peak in the forward direction at high energies is known as a diffraction peak.6.6 Implications of Unitarity 149 Using Eq.6.2) then tells us that the total reaction rate from an initial state consisting of some set of particles is the same as for an initial state consisting oj the corresponding antiparticles with spins reversed: (3. and Eq.10) then yields an upper bound on 8Q Ll!l < 32. and multiply the matrix element by various phase factors for the particles in the initial state and by their complex conjugates for the particles in the final state.6.46) from timereversal invariance to show that CPT invariance requires the Smatrix to satisfy the condition Sp.P2(T2rl2. (3.2) together with CPT invariance to say something about the relations for total interaction rates of particles and antiparticles. = S'f/!lYi7. the same relation holds for the coefficient of 04(pi7. = M'f/!lY'X.'" .: M p . . (3.6.6. change all particles into their corresponding antiparticles. 'f/YyP .' /k'a.i7. (3. applying this to oneparticle states..i7. Because CPT is antiunitary.14) where a superscript c on n indicates the antiparticle of n.3. (3. Returning now to the general case of reactions involving arbitrary numbers of particles.<£&'Yp .. P2(T2rl2.
19) ~ .:.f P dP .15) is that unstable particles and their corresponding antiparticles have precisely the same lifetimes.6..f dfi .2) also allows us to use the other unitarity relation sst = 1 to derive the result (3. + {J)/d{J..16) .f dfi J4(PI . The same argument that led from the unitarity condition sts = I to our result (3.) .6.f d>/ dli PI In (~) df(~. ~ . (3. Rotational invariance does not allow particle decay rates to depend on the spin zcomponent of the decaying particle.:. due to transitions from all other states is f d{J Pp df(fJ + rx)/det.17) This result can be used in deriving some of the most important results of kinetic theory.2) then yields the reciprocity relation .. (3. P". 1nP.P dfJ Interchanging the labelling of the integration variables in the second term. In P.4(PI .18) ~ .6. is I I . J dfJ df(et.. while the rate of increase of P". Now.6. this may be written .) On the other hand.fdP dr(a ~ P) dt p det.)IM.p. the rate of change of P".6. .pl' (3.p.19). do (3. + 1) Ii)] df(P ~..)IMI. for any positive probabilities Prt and PfI' the function Pfl In(PtJ/ P". the rate of change of the entropy J drxPx InP". 0) . d{J' It follows immediately that (3.f d> P. ~ .f dfi (In P. ~ . ~ rt x [PfJ det. Putting this together with Eq.23 If Pexdct is the probability of finding the system in a volume drx of the space of multiparticle states cD"" then the rate of decrease in Px due to transitions to all other states is P". .6. Prtdrx is timeindependent.I' or in other words d dr(a ~ fi) ~ fdfi dr(fi ~ a} .6.fdPP dnp ~ 0) _ .150 3 Scattering Theory spin.6. (Just interchange the labelling of the integration variables in the integral in the second term in Eq. is then dP.f d> .f do P. so a special case of the general result (3.) dr(.
approaches the positive quantity {I'.6. In this case the conservation laws· require dr(P + a)jd'1 to vanish unless PC( = Pp. so we can replace PfJ with PC( in the first term of Eq.6. neither the Born approximation nor timereversal invariance are exact. Here again.6. so we may conclude that the entropy always Increases: .6.7 PartialWave Expansion8 151 satisfies the inequality" Pflln (~) > PfJ  PC(' The rate of change of the entropy is thus bounded by t. for which IM1 2 is symmetric in et and P so that dr(p + et)/det = hl dr(a + P)/dP.and righthand side. In P.18) again then shows that in this case. < Fp. and may be omitted in a first reading. a derivative with respect to P. (3.:.7 PartialWaye Expansions· It is often convenient to work with the Smatrix in a basis of freeparticle states in which all variables are discrete.6. j do P. > O. except for the total momentum •• The difference between the left.19).20) This is the 'Boltzmann H theorem. so it is a good thing that the unitarity result (3. which would tell us that IMfl:>1 2 is unchanged if we interchange a and fJ and also reverse all momenta and spins. 3.' This theorem is orten derived in statistical mechanics textbooks either by using the Born approximation. + 1'110 and ha.18) is all we need in order to derive the H theorem. we need only the unitarity relation (3. Using Eq. not the Born approximation or timereversal invariance. that is positive. Pill' /2P~ for P. • This section lies somewhat out or the hook's main line of development.or negativedefinite for all P. (3. da or interchanging variables of integration in the second term dd (do P. InP.] drtP ~ 0) ~jdoP InP >jd (dPP [drtP~O)_drto~P)] dt C( a a. > P~ or P.6. . > jdo (dP [Pp _ P. .18). PC( is timeindependent. respectively. The increase of the entropy stops when the probability Pa becomes a function only of conserved quantities such as the total energy and charge. fJ da dfJ But the unitarity relation (3.3. Of course.18) tells us that the integral over a on the righthand side vanishes. or by assuming timereversal invariance. (3.
' . ti2 with nonzero masses Ml. the spin zcomponents ai.3) and the relation (3. This is possible because the components of the momenta PI. These states may conveniently be chosen to be normalized so that their scalar products are: (<I>£. We may thus define a basis for these nparticle states that apart from the continuous variables p and E is discrete: we label the freeparticle states in such a basis as I1l EpN .7. we can form a convenient discrete basis by using ClebschGordan coefficients9 to combine the two spins to give a total spin 5 with zcomponent Jl.. T<I>EpN) ~ J'(p' p)MN'. for instance. defined by their scalar products with the states of definite individual . (3.N is a unitary matrix. say.P)ON'. consider a state consisting of two nonidentical particles of species nl. such as the spherical harmonics that are commonly used in representing functions on the twosphere.N' (3.p' N. ti2. (3.18). S<I>EPN) ~ olE' _E)03(P' .P)SN'. for the present we will concentrate on reactions in which the initial state involves just two particles. in an nparticle state of definite total momentum p and total energy E form a (3n .p) (3. . Alternatively. This gives a basis of states I1l Epjo (s/i (with n a 'channel index' labelling the two particle species' nl. Similarly.p).4)dimensional compact space. for n = 2 particles in the center~ofmass frame with p = 0. Pl. the energy E. p. ti2). (3.2.1. the species labels til.1) The Soperator then has matrix elements in this basis of the form (<I>£'p'N'.<I>£pN) ~ OlE' .152 3 Scattering Theory and energy. the Toperator. and then using ClebschGordan coefficients again to combine this with the orbital angular momentum t with threecomponent m to form a total angular momentum j with threecomponent a.. a2. may be expressed in our new basis (in accordance with Eq.E)J'(P' . For example.4) We shall use this general formalism in the following section.2) where SN'.7. the states may be labelled by their total momentum p = Pi + Pl. and a pair of integers t. 52.8' Tl1lo:) are defined to be the quantities TfJ'l + defined by Eq.7) is now an ordinary matrix equation: (3.N(E. this space is a twodimensional spherical surface. whose freeparticle matrix elements (111..7.4. M2 and arbitrary spins 51. with the index N incorporating all spin and species labels as well as whatever indices are use to label the generalized partial waves.N(E.12) as (<I>Ep'N'.m (with Iml < t) that specify the dependence of the state on the directions of.7. Any function on such a compact space may be expanded in a series of generalized 'partial waves'. In this case.
m. 24 The factor (lp11EIEz/ E)1/2 is inserted so that in the centerofmass frame these states will be properly normalized: (IDE' p' j' a' t' s' 11" IDEO jot S11) = 03(p')03(E' . (37.7) (The fact that this is diagonal in j and IT follows from the commutation of 0 with J2 and h.7.n') kla~k'a.7 Partial.7.lI'.E)o j'.(k)Mt'5'nl.8) The differential scattering crosssection is If1 2.7.7) in the centerofmass system takes the fonn f(klTj.l. it follows that the scattering amplitude (3. x Cs'l s.jOa' . O".aOt'.Wave Expansions 153 momenta and spin threecomponents: (IDp1 UI P2U2rr .0"2)CtAj.lTl.J1) " ( ' . 25 ) Applying this to the operator M whose matrix elements are the quantities MfJrt.6) yt For identical particles to avoid double counting we must integrate only over half of the twoparticle momentum space.sbll'.k0"2.2 + k' O"i. In the centerofmass frame the matrix elements of any momentumconserving and rotationallyinvariant operator 0 must take the fonn: J2 (3.6).0"t>0"2 ('l}.PI .5) x L m.E E 2k 4. ')C (.M + Mr  vpj + Mi) 6"" (3.J1j .~ " j x Yt'f/(k')Yt . This is a special case of a general result known as the WignerEckart theorem. We will take the direction of the initial momentum k to be along the threedirection.p CS1S2(S..Il.0"2) C(Aj.3.J.0". in which case ym(k) t ~ 6mOV~' fU+l (3.m.n' (3.7.' T L ju('m' ~J pi fms/' C51S2 (s.".9) . IDEpjafsn)  (lpIIEIEz/ E)1/203(p ..k(Tlka21l " k' E1 1E2 M E. .j1. and the further fact that the coefficient of O(TUI is independent of IT follows from the commutation of 0 with JI + ill.k' O".6.m.P2) x 6( E.7.a.n __4.tbs'.S.J1)Yt (pd· where are the usual spherical harmonics.t51l(E). so an extra factor should appear in the scalar product (3.0"j.
.10) Summing (3.E)=) k~(2s1 X L (2)+1) + 1)(2S2 + 1) jist's' 1r 2 Ib!'t()"""bn'n . m. and thus [0. " ' ) [ l (E)]{sIlIm. 2j+L L.9). (3..Sj(E))t(l.m(E)1 . and the ClebschGordan sum rules·· give the spinaveraged forward scattering amplitude as f(n.12) and (3.sr.E) = k 2(2 l)(?' L..'2{S.13) I I ..'.' and finally Ct..o'J = and the same witb prime:. if. we u~e standard slim rulcs 9 for the ClehsebGordan eoenleients: first :2. (3. (3.7.8)..E) ~ 2k(2s + :)(25)+ II L(2j + 1)[1.sfw.. i I L . (l.7.7.t.12) O"total(n. if channels involving three or more particles are open.7.. (3.. a."I . 0) = 0jJo~" ".7..4).11) •• In Jerivlllg tbis result.:= c".6.!.7.11) are equaL On the other hand.7.. 0.(). Jl: "I . m.10) over all twobody channels gives the total crosssection for all elastic or inelastic twobody reactions: L"(n n' ~ 11'. (3. 0..7.11) For comparison. (3.7.) = 2( + I <'itt '" . (3.Sj(E»)t(1  sj(E))LsIlJ.'OI'j.10) then gives the total crosssection as 2. then the difference of(3.(2J+I Re IS SI+ _s2+1)jls If only twobody channels can be reached from the channel n at energy E. a )Cr. then the matrix Sj(E) (or at least some submatrix that includes channel n) is unitary. Cd)..on'.)Crs!J.12) and (3.Sl(E))] ("n. i so (3. . (~. i . /.11l = 2 Re [l ~ Sj(E)]rsll/sn. It. tben o.J 1 Its The optical theorem (3. O"(n+n'. ."2)C" I.154 3 Scattering Theory Integrating Ifl 2 over the direction of the final momentum k' and respectively summing and averaging over final and initial spin threecomponents gives the total crosssection·· for transitions from channel n to n'.{j. E) ~ k 2(2s X I + ~)(2S + 1) L(2j + 1) 2 Jls [(1./.7.7.{8>1 ..7. Eqs.
only on unitarity and invariance principles. the elastic and total cross·sections are then given by Eq. for instance in pionnucleon scattering we can have j = t + ~ or j = t . (3. The classic example of such an internal symmetry is isotopic spin symmetry.). n.10) or Eq. so they cannot be connected by nonzero Smatrix elements.O) all vanish unless N' is the twobody state j. t.7. and no other channels are open at this energy. This is the case.E) = k 2(2s] + ~)(2S2 + 1):. (3.l.7. (3. then unitarity requires that S. and also to emphasize that it depends on no particular dynamical assumptions. E) ~ k'(2s] +~~(2S' + 1) ~(2j + 1) sin' OJ!.Wl(E.~.n . (2j + 1) II x [1. For real phase shifts. The derivation given here is offered both to show that the partial wave ex pansion applies for elastic scattering even at relativistic velocities.15) where ()jbn(E) is a real phase. s.14) positive. This formula is also often used where the twobody part of the Smatrix is diagonal but channels containing three or more particles are also open.14) and this must be positive. (3.Wave Expansions 155 gives the total crosssection for producing extra particles: O"productioll(n. but for a given j these two states have opposite parity.3. The partial wave expansion is particularly useful when applied to processes where the relevant part of the Smatrix is diagonal. It is also possible for the Smatrix to be diagonal in certain processes involving particles with spin.'. to keep (3.jl. for instance.Sj(E)'Sj(E)] (.7.7 Partial.=."().tsn(E) = ex p [2ib jtsn( E) pnb"""'()n'n .7. For a pair of spinless particles we have j = t. 7. if the initial channel n contains just two spinless particles.nO scattering at energies below the threshold for producing extra pions (provided one ignores weak and electromagnetic interactions). (3. in such cases the phase shift must have a positive imaginary part. if for some nand E the Smatrix elements SN'.n. for which the channel index n includes .16) This familiar result is usually derived in nonrelativistic quantum me· chanics by studying the coordinate space wave function for a particle in a potential. and angular momentum conservation keeps the Smatrix diagonal. commonly known as the phase shift.12) as: rr(n 4 n.l n'.n+ or JT+ . It is also often llseful to introduce phase shifts in dealing with problems where several channels are open.7. E) = rT1ota[(n. In any case. as in JT+ . forming a few irreducible representations of some internal symmetry group.
18) jisT! For instance. we note that eYiIl(k) is a simple polynomial function of the threevector k.{sn must go as ki+ik'!+! when k and/or k ' go to zero. Ii.ll)· Suppose that for the channels and energy of interest the Smatrix is diagonal in t and s as well as j. The chief exception is for scattering involving longdistance forces.t : tl. while for pionnucleon scattering we have phase shifts 0). Unitarity and isospin symmetry then allow us to write the Smatrix as SJ.2. For instance. in the Born approximation (3. the coefficients Ml's'n'.17) with bjtsT(E) a real phase shift. .YT't'..tsTt = exp[2iDjtsT(E)]DI'"tbts()T'T6t't. (3.t2.). in pionpion scattering we have phase shifts 0rtoT(E) with T = 0 or T = 2 for each even t and T = 1 for each odd t. and the total crosssection is given by Eq. T2 of the two particles together with their threecomponents tl. with coefficients given by the familiar ClebschGordan coefficients CT]T2(T. k (TI k(fl n to be an analytic function of the threemomenta k and k' near k = 0 or k' = 0. We can gain some useful insight about the threshold behavior of the scattering amplitudes and phase shifts from considerations of analyticity that are nearly independent or any dynamical assumptions. so in order for Mk'". T. it is only the lowest partial waves in the initial and/or final state that contribute appreciably to the scattering amplitude. such as the Coulomb for~"e.12) as O"total(tl.7.E) = k2(2sl + 1)(2s2 + 1) 4. x L(2j+l)CTi. Hence for small k and/or k'.7. (3. [2. tindependent according to the WignerEckart theorem.10).7. the states in channel n may be expressed as linear combinations of the tth components of irreducible representations T. and t. k(fl kal n to be an analytic function t of the threemomenta k and k' near k = 0 or k' = 0 or (for elastic scattering) k = k' = O.~.T1(T. The partial crosssections can again be calculated from Eq.t2)2 sin 2 ojtST(E). We have three possible cases: t " ! .\on'n sj'snl. we would expect the matrix element Mk'!T~ k' a.g).+ l I T with T = or T = ~. and hence is analytic at zero momenta as long as these matrix elements fall olf sufficiently rapidly at large separations.7. n'.156 3 Scattering Theory a specification of the isospins T I . Unless there are special circumstances that would produce singularities in momentum space. M is proportional to the Fourier transform of the coordinate space matrix elements of the interaction.t.7. Turning to the partial wave expansion (3.fsn or equivalently o('(o. (3.8) for M.tI.'. (3.'1 k' ".
11"1. 11"2J as(n + n') .~~ (s.J'"'n S. aside from the higherorder effects of Coulomb forces.11"2.'s'nl. for instance.11"1'k' . known as the scattering length. or the production of electronpositron pairs in the scattering of photons. for instance.) The reaction rate is the crosssection times the flux. so in the limit k + 0 the scattering amplitude (3. In the most usual case ( = 0. However. (This is the case.7.n') + L '" CSP2 (S' 11". defined by the limit (3.S.7.20) .) In elastic scattering the partial waves with t = (' = 0 are always present.'s'nl.11) in this case goes as k 2t .7. 11"]. .k . and hence like JE Elhreshold. so the rate for an exothermic reaction behaves like a constant for k + O. or where I'l' consists of particles in the same isotopic spin multiplets as are those in n. or for the annihilation of electronpositron pairs into photons at low energy.19) where a is a constant. k' .3.) Elastic Reactions Here k = k'.11"2.tb'~'sbn'n .7. 11".Wave Expansions Exothermic reactions 157 Here k ' approaches a finite value as k + 0. so the reaction crosssection goes as 11k. 11"2)Cs.7. and in this limit bt. in the absorption of slow neutrons by complex nuclei.11) in this case goes as (k ' )2t'+I.hn goes as (k')f'+L The crosssection (3. it is the crosssection rather than the reaction rate that determines the probability of absorption when a beam crosses a given thickness of target material.n + . so k and k' go to zero together. (3.{b~J. Endothermic Reactions Here the reaction is forbidden until k reaches a finite threshold value. and the factor 11k makes this probability very high for slow neutrons in an absorbing material like boron. in the associated production of strange particles. (This is the case where n' = n. (This is the case. In the most usual case t' = 0. 11"~.8) becomes a constant: i(k.1.tsn goes as kt+L The crosssection (3. where ( is here the lowest orbital angular momentum that can lead to the reaction. where (' is here the lowest orbital angular momentum that can be produced at threshold. 7 Partial. Just above this threshold b{. where k' = o. which goes like k. so the reaction crosssection rises above threshold like k ' .
.. 14~2 s (3.) On the other hand. Summing 4nlfl 2 over final spins and averaging over initial spins gives the total crosssection for the transition n + n' at k = k' = 0: a(n ~ n'.7.7.24) gives an inelastic crosssection nR.(n ~ n') .7. with the spin singlet length ao considerably larger than the spin triplet length ai_ The partial wave expansion can also be used to make a crude guess about the behavior of crosssections at high energy. we may expect scattering to be described more or less classically: a particle of momentum k and orbital angular momentum t would have an impact parameter t /k.7. the same analysis gives a bound on the real part of the forward scattering amplitude IRef(n.7.Si.E)  2. so the sum over j and s in Eq.24) The difference between Eqs.158 3 Scattering Theory for k = k' + o. if we assume along with Eq. where there are two scattering lengths.E)I· (3.7. ] 1 The total crosssection IS then given for k» l/Rn by Eq.E)  nR~. which is what we would expect for collisions with an opaque disk of radius Rn.10) gives the elastic scattering crosssection O"(n _ n.ln~ ~i {O 1 t« kR.7. there are 25 + 1 values of j.r. (3. (3.22) that S}sn. may be attributed to diffraction by the disk.k ~ 0) ~ (2 51 +) 52 + 1) 2.)2s + I)G.12) merely gives a factor of order 2)2j + 1) ~ (2t + I) 2)2s + I) ~ (2t + 1)(2s[ + 1)(2"2 + I).25) . all close enough to t to approximate 2) + 1 ..kR" In exactly the same way. Eq.~n. (3. (3.22) where Rn is some sort of interaction radius for channel n..7. This can be interpreted as a statement about Smatrix: elements: 'tsn/.E)1 < 2kR. For a given t A> s.23) and (3. and will therefore strike a disk of radius R if t < kR. t>kR n ' (3. (3.. (The somewhat surprising elastic scattering crosssection nR. With decreasing wavelengths. then using the inequality 11m (1 . .12) as O"\Olal(n.tsn)1 < 2.« IImf(n.7. " k L 2 (2/+ 1)  2 2nR11 • (37.ln is complex only for impact parameters t /k within a small range of width till «:: Rn around t /k = Rn .23) t5."'..21) The classic instance of the use of this formula is in neutronproton scattering. .7. (3. 2t + I..
In 1989 the ZO particle was seen as a resonance" in electronpositron collisions at CERN and Stanford.8 Resonances 159 The smallness of the real part of the forward scattering amplitude at high energy is confinned by experiment. it has been rigorously shown26 on the basis of very general assumptions that the total crosssection can grow no faster than (log E)2 for E . including the initial and final states ''Y.. so this rough picture of highenergy scattering does seem to have some correspondence with reality. known as a resonance. but with a total decay rate that is much less than the ZO mass.t frame. 3. which allows R to decay into various states. Mz of o~cil1alion of the ZO wavc function in its re.25 seconds. For instance there is a neutral particle. and in fact the observed protonproton total crosssection rises something like (log Ef at high energies. What is important is that the decay rate is 36 times smaller than the ratc n. and may be omitted in a first reading. at the energy of the intermediate state R. the ZO lifetime is 2. •• Incidentally. . this example . plus a weak perturbation Vw . and the crosssections go as (log £)2. we may take R n as the distance at which the factor exp(w) in the Yukawa potential (1.. which is not long enough for a ZO travelling near the speed of light to cross an atomIC nndeus. These interactions allow the ZO to decay into electronpositron pairs. As a very crude guess. a 'strong' Hamiltonian Ho + V" which has the particle R as an eigenstate. with j = 1 and mass 91 GeV. etc. which is a good thing since there are a number of very different mechanisms that can produce a nearly stable state: (a) The simplest possibility is that the Hamiltonian can be decomposed into two terms. muonantimuon pairs. As it happens. the Zo. We shall see that the behavior of the crosssections near a resonance is pretty much prescribed by the unitarity condition alone.2. co. in which case Rn goes as log E for E _ co. that would be stable in the absence of the electroweak interactions. that eventually decays into the particles observed as the final state. p of our collision process.74) takes a value proportional to some unknown power of E.3. • This section Iie~ somewhat out of the hook's main line of development. If the total decay rate of R is small the crosssection exhibits a rapid variation (usually a peak). So far we have not said anything about whether the interaction radius R n itself may depend on energy.8 Resonances· It often happens that the particles participating in a multiparticle collision can form an intermediate state consisting of a single unstable particle R.hows that a resonant state only need~ to deeay relatively slowly.6 x 10.
a large part of its energy is concentrated on a single neutron. 3 Scattering Theory + e t ZO _ e+ + e. and gammaunstable nuclei of high spin.1.rt/2). the lowestenergy state of the Be8 nucleus is unstable against decay into two alpha particles. element for a reaction near a resonance. The decay then can proceed only by quantum mechanical barrier penetration. beta. so it corresponds to a state whose probability decays like exp(rt).) (c) It is possible for complicated systems to be nearly unstable for statistical reasons. The classic example is nuclear alpha decay: it may be energetically possible for a nucleus to emit an alpha particle (a He 4 nucleus) but the strong electrostatic repulsion between the alpha particle and the nucleus creates a barrier region around the daughter nucleus. an excited state of a heavy nucleus may be able to decay only if. This state will then show up as a resonance in the scattering of a neutron on the daughter nucleus. t . (In addition to Coulomb barriers.1.' As mentioned in Section 3.lt) of 'in' states has a time·dependence given by (3. without the presence of any potential barriers or weak interactions.if 12 yields a term in the amplitude that behaves like exp(iERt . For instance. and is exponentially slow.l + exp(iE(. there are also centrifugal barriers that help lengthen the life of alpha. through a statistical fluctuation. E (. These mechanisms for producing longlived states are so different that it is truly fortunate that most of the properties of resonances follow from unitarity alone. Specifically. a pole at E Cl = ER . without regard to the dynamical mechanism that produces the resonance. We conclude then that a longlived state of energy ER with a slow decay rate i I . (b) In some cases a particle is long~1ived because there is a potential barrier that nearly prevents its constituents from escaping. Such an unstable state shows up as a resonance in the scattering of the alpha particle on the daughter nucleus. + in the lowerhalf complex E Cl plane would make a contribution to the second term that decays exponentially as t _ co. a pole in the function TfJrt.l E fJ + u· .160 in the reactions e+ etc.iEot = J +J J det g(et)l1>Cle iEot eiEot g(et) TfJ + dP <I>p d. which the alpha particle is classically forbidden to enter. let's consider the energy dependence of the matrix. e+ + e _ ZO  /1+ + Jr. For instance.19) I f I ! i l I I l I J det g(et)\{Ite. and is seen as a resonance in He 4_He4 scattering. First. A wave packet J det g(et)\{I(.
4) Applied to Eq.3. the S·matrix may be written Sp'E'N'.1) To go further. this tells us that the nonresonant background S·matrix is unitary y'ot..ERI:S r.8. A .E)SN'N(P.9'ot.8 Resonances 161 r produces a term in the scattering amplitude that varies as Tprt + . E).~r.3). (3. (3. (3.8.'+C.8.. Such matrices can always be expressed as a sum of dyads of orthonormal vectors u{r). and N is an index that takes only discrete (though infinitely many) values.9'(E) ~ 1 ..9'o.U N'N = LJ uN' UN " (r) (r)o . (3. E) ~ . we expect the centerofmass frame amplitude S(O.$"o = 1.8) . In this basis.:l u:.9'N'N(E) ~L N" ["N'N"  j E _ Er .11) . pEN ~ "](p' .01 .8.8.'I'o ~ 0 .8.8.'l'o+91'9!~O. (3. 4 2 ~ Any such Hermitian idempotent matrix is called a projection matrix.10) The discrete part of the Smatrix is then .7) .8. the unitarity of the Smatrix is an ordinary matrix equation . (3. UN = urs' (s) ~ (3..8. (3. These conditions can be put in a more transparent form by setting 9! _ W4.6) (3. it will be convenient to adopt as a basis the orthonormal discrete multiparticle states lb pEN discussed in the previous section. (E rt . p and E are the total momentum and energy. E) 9"(E) to have the form .51'0 and (jf are approximately constant at least over the relatively small range of energies IE .'J!+ ~r9!t.8.'I' N'N(E) = SN'N(O.W""I""2 :JfN'N ' (3.8. and also that the residue matrix :Ji satisfies the two conditions .8. .9!I .'I'ON'N + TEcE"'R"'.9'0 I 9! .'I'(E)'. 12 L R+lr r u:.5) (3.:l:] . .2) Near a resonance.3) where .'I'OWN· (3.9) The unitarity conditions on the matrix d are then simply . In this basis.p)"(E' . ' " LJUN N (r).wi ~ 4 .ER + ir/2)1 + constant.
Eq.7.8. 27 We can also use these results to calculate the crosssection for resonant scattering from an initial twobody channel n to a final twobody channel . (3. Eq. (3.8. then r just labels the value of (TR. (lR takes 2jR + 1 values. setting .. .162 3 Scattering Theory Each term in the sum over r can be thought of as arising from a different resonant state.14) Also.8. (3.16) rI: t. I Utsnl 2 .13) where U{sn are a set of complex amplitudes that (because of the WignerEckart theorem) are independent of (T. the quantities IUtsnl2 have the interpretation of branching ratios for the decay of the resonant state into the various accessible twobody states. If there is no other degeneracy. As we shall see.12) In all cases the label r will include an index (lR giving the zcomponent of the total angular momentum of the resonant state.\'1 + 1)(2s2 + l) (E E where In (3.8. for a resonant state of total angular momentum jR.8. all these states having the same values for E R and 1.7) as r E R + il/2 Ut's'n' Utsn 1 . (3. Now Eq.7. Eq.8. we will come back to the more general case a little later. = (3. ) = k2(2.8.8.8.12) now gives the total crosssection for all reactions in channel n: n(2jR + 1) O"tolal(n. Then for the twobody discrete centerormass states described in the previous section. (3. (3. What has this to do with rates and crosssections? For simplicity let's now ignore the nonresonant background scattering.17) This is a version of the celebrated BreitWigner singlelevel formula. and (3.12) gives the amplitude sj defined by Eq.10) now reads I: tsn I Utsnl 2 + . (3.'/ON1N equal to (jN'N..15) with the dots representing the positive contribution of any states containing three or more particles.11) reads: (3.
16) and (3. ( The individual f n are often called partial widths. This rule.8. to the extent that isospin symmetry is respected. (3.14). (3. and hence can only couple . this is J (Jtotal(n. taking values TR. This formalism can also be applied when the resonant states with a given spin zcomponent form a multiplet related by some symmetry group. For the total crosssection (3.TR+ 1. (3.7. because each twobody channel n has definite values tl.E) ~ n(2jR + 1) r nfn' k2(2.8. In this case. it often happens that a resonance is detected.10) gives a(n ~ . (2n/k)2.15).8. the branching ratio for oscillations in any laboratory mass to lose their energy through gravitational radiation is tiny. (3. We see in Eqs. or gravitational waves with gravitational wave antennae. n .19) Such experiments can revei:ll only the partial width for decay of the resonant state into the initial particles.8 Re. as for instance in the resonant interaction of sound waves with bubbles in the sea. so we can conclude that f n is just the rate for the decay of the resonant state into channel n. (3. (In the latter case.wmances 163 n'. In this case there is no change in the above results for the total and partial cross~sections. For instance.18) This shows that the probabilities that the resonant state will decay into anyone of the final two·body channels n' are proportional to the fnt.'" T R. the sum of the f n (including contributions from fmal states containing three or more particles) is just equal to the total decay rate r. q( 2s l + 1)( 252 + I)' 2n2(2jR (3.) Since f n < r. t2 of the isospin zcomponents of the two particles. is universally i:lpplicable even in classical physics (where energy conservation plays the role played here by unitarity).8. for a resonance of total isospin TR the index r labelling resonant states includes a specification not only of the angularmomentum zcomponent (JR. not the total width or branching ratios. so the crosssection even at a resonance peak is vastly less then a square wavelength.8." + 1)(2s2 + I)(E E R )2 + r'/4 .3. that crosssections at a single resonance are roughly bounded by a square wavelength. the total crosssection at the peak of the resonance is roughly bounded by one square wavelength. but energy measurements are insufficiently precise to resolve its width. Using Eq. but also of an isospin threecomponent fR.8. According to Eq. with a width (the full width at half maximum) equal to the decay rate f. 28 ) Incidentally.14) in Eq.8.t8) the characteristic resonant peak at energy E R . what is measured experimentally is the integral of the crosssection over the resonant peak.E)(ZE = + l)f.
(3. The unitarity requirement (3. it is necessary that the same is true of Sl'ON'N. the the 'eigenphase' (j(rl(E) jumps from a value V1t (with v a positive or negative integer) below the resonance to (v + 1)n above it. in order to use this result to say something about reaction rates. spin.20) 'anb(')(E) ~ _ r/2 exp(2iboN )b N 'N and Eq. However.164 3 Scattering Theory to the resonant state with the single value II +t2 for fR. in general. Eq. we see from Eq.O.8. (3.10) requires that so that there can be only one term r in Eq.IR.11) for which u~) this case.tt.11) (but still taking Yo = 1).8. The partial widths r ll here depend on tl and t2 only through factors C T1 . In order for Y'N'N to vanish for some particular N and all N' 1. in other words. Returning to the general formula (3.11) gives YN'N(E) ~ IN'N 1.O. it is not difficult to include the effects of a nonresonant background scattering matrix Y'o in the general result (3. These results are much more useful in those special cases when the particles in a particular channel N are forbidden (usually by conservation laws) from making a transition to any other channeL With this assumption.8. and also true of urJ for any r for which ut.5) then requires that for this N Y'ONIN = r R + ir /2 (J.T2(TR.8.bWN exp(2ibN (E)) 1. and species. (3.} 1. The presence of a resonance shows itself in a characteristic behavior of phase shifts near the resonance.8.8.21) We see that over an energy range of order r entered around the resonant energy ER.11). the phase shift bN(E) jumps from a value bON below the . ~3. there is an eigenvector u% of Y'N'N(E) with eigenvalue exp(2ib(')(E)) ~ 1 . E ER We see that over an energy range of order r centered around the resonant energy. we need to know the eigenvectors u~).8.10) that for each individual resonant state r. have components with arbitrary numbers of particles with various momenta.8.8.N. which.t2)2. In [ ir] + ir/2 exp(2iboN ) with total phase shift bN(E) ~ JON  arctan ( E r/2 ) E R (3.i E _ E or.E _ E R .
Consider a theory with a separable interaction. Suppose that a resonance of spin one is discovered in e+ e.7. that is.34 cm2 . in pionnucleon scattering there is a resonance at 1232 MeV called 6. in which case the crosssection will exhibit a sharp dip as the phase shift rises through n near ER . and T. and summed over final spins) for elastic e+e. Problems 1. For instance. solu~ 2.scattering at the peak of the resonance equal to 10. Such dips were first observed by Ramsauer and Townsend 29 in 1922. There are famous resonances in these channels: in pionpion scattering there is a resonance at 770 MeV called p with j = t = 1. Inspection of Eq. as we saw in the previous section.18) shows that the total crosssection reaches a peak when the resonant phase shift passes through nl2 (or oddinteger multiples of nI2. Vlb!l:) = g u{J u: .Problems 165 resonance to bON + n above it.12) or Eq. it sometimes happens that the nonresonant background phase shift bON is near n12.scattering at a total energy of 150 GeV and with a crosssection (in the centerofmass frame. T = ~ and r = 110 to 120 MeV.+ e+? What is the total . The WignerEckart theorem allows the phase shifts to depend only on j. t = 1. However. t. at an energy close to ER. in the scattering of electrons by noble gas atoms. and quantities with U!I: IS a set of complex Use the LippmannSchwinger equation (3. What is the branching ratio for the decay of the resonant state R by the mode R + e. with N incorporating the total and orbital angularmomenta j. (3. averaged over initial spins. (Flota] will exhibit a sharp peak when the phase shift b{ goes through n12. (3. t (with j = t for pionpion scattering) and the total angularmomentum zcomponent (F.1. (lb p. where g is a real coupling constant. T = 1.16) to find explicit tions for the 'in' and 'out' states and the Smatrix.7.) The nonresonant phase shifts are typically rather smaiL so as we saw earlier. these assumptions are satisfied in various two~body reactions such as pionpion and pionnucleon scattering at energies below the threshold for producing extra pions. not on t or (j. as well as the total isospin T and its threecomponent t. and r = 150 MeV. due to destructive interference hetween the resonance and the nonresonant background amplitude. with j = ~.
Goldberger and K. 2. 38. For more details. New York. Express the differential crosssection for twobody scattering in the laboratory frame. 6. J. parity.( . Wheeler.7. Phys.( =111(. 863 (1926). 79. C.J(l1ltJ. 469 (1950). 23. Watson.6).5.E r. Show how to express the Smatrix in terms of the Kmatrix. G. 3 M. Z. in which one of the two particles is initially at rest.scattering at the peak of the resonance? (In answering both questions. Schwinger. 1982). and isospin. 803 (1926). Derive the perturbation expansion (3. A. Phys. Rev. W. Medd. 37. 2nd edn (SpringerVerlag. Z. V'¥xo) is Hermitian. Phys. (3. No. M. 4. New York. 52. without llsing the results derived in this chapter for the differential crosssection in the centerofmass frame. Danske Videnskab. M~ller. L. ignore the nonresonant background scattering. B. No. see M.3) of oldfashioned perturbation theory.ln defined by Eq. 1107 (1937).(+ tp:x 0 by a modified version of ° E't. Express the differential crosssection for elastic n+ proton and Jr proton scattering in terms of the phase shifts for states of definite total angular momentum.1 (1945). Kgl.) 3. Show that the states I1lEpja. R. 5. Fys..5. Scattering Theory of Waves and Particles. We can define 'standing wave' states the LippmannSchwinger equation \fI(. ~H ° ° V\fI x • Show that the matrix K(i(. Phys. 22. Rev. I . . 19 (1946). Born.5) are correctly normalized to have the scalar products (3. 120. 1964). 7. 513. 673 (1943). la. r .8) directly from the expansion (3. (Derive the result directly.nr5(EtJ . Mat.) 4. Heisenberg.166 3 Scattering Theory crosssection for e+e.. in terms of kinematic variables and the matrix element M(iC/. Newton. Collision Theory (John Wiley & Sons. References 1. Lippmann and J.7.
Cronin.. Butler. Lee and C. L. Report of an International Conference on Fundamental Particles and Low Temperatures. J. 12. Rev. Phys. Elementary Theory of Angular Momentum (John Wiley & Sons. Phys. Rev. Lett. 1946 (The Physical Society. 6. G. 16. A. 17. Yang. Rev. Condon. Cassen and E. Tuve. Lederman. Letters 13.~. see G. Phys. J. Weinberg. W. 105. Y. 1947). 105. S. 1387 (1955). Rev. and M. T. Ret? 50. Condon.References 167 5. and finds that the part of the CPviolating amplitude that conserves . Phys. also see B. 850 (1936). Phys. Rev. 1561 (1954). Phys. Friedman and V. 846 (1936). G. 88. Phys. 50. M. Rev. 13. Present. K. Feinberg and S. C. G. E. Phys. ed. Christenson. Phys. and R. 14. New York. Feenberg. L. 1413 (1957). D. and L. G. 95. L. Rochester. Synchotron Laboratory Report CTSL20 (1961). Ne'eman Nucl. Tech. Heydenberg. Cambridge. H.g. 1957): Chapter 3 (where CM2 (Jm. and R. Fitch. Schubert et al. M. I.mlm2) is denoted (Jlh. 7. 125.imIJjmlhm2)). Brown. 1487 (1955).Particle Physics in the 1950s. E. See e. R. Rochester and C. Weinrich. 50. D.. R. Dresden. Rev. London. 571 (1959). A. M. 825 (1036). 11. U. Watson. ml m2 m)). N. B.mlm2) is denoted C(jj h J. Phys. Breit and E. 26. 10. N. For a historical review. Princeton. Rev. 97. 15. 31B. Cal. Phys. Te1egdi. Pais and O. M. Hoddeson (Cambridge University Press. W. Rev. GellMann and A. Garwin. Ferretti. Breit. Rose. R. K. D. Nature 160. 138 (1964). M. Phys. 1415 (1957). Edmonds. 8. GellMann. 855 (1947). 1957): Chapter III (where CM2 (Jm. UK. Rev. Phy. Hafstad. Rev. 222 (1961). in Pions to Quarks . U. This reference analyzes neutral kaon data without the assumption of CPT invariance. Steinberger. by L. 105. Camhridge. 18. 100. 806 (1936). C. J. Angular Momentum in Quantum Mechanics (Princeton University Press. 662 (1970). 254 (1956). V. Phys. Turlay. Rev. WU et aI. 1681 (1957). 1067 (1962). 104. 1989). also see A. and L. 9. Piccioni. Nuovo Cimento Serie X. S. Ret? 50. Phys. R. M. 14. Chinowsky and J. M. PhY8. Pais. 1163 (1952).
c. This was first proved in classical electrodynamics. Dalitz. 20. g. 1053 (1961). Also see A Aharony. 305 (1930). while the part that conserves T and violates CPT is within a standard deviation of zero. Kollath. G. P. 1957): Appendix III. E. Rev. P.. Feenberg. M.g.479 (1954). 985 (1931). Lifshitz. Angular Momentum in Quantum Mechanics. Breit and E. Dyson. Quantum Mechanics. I. Gravitation and Cosmology (Wiley. H. L. 49. Edmonds. 1973): pp. 22. A. Rev. 519 (1936). E. 24. 2. Phys. Fisici. P. Weinberg. ( . 3rd edn (Pergamon Press. Mod. See. J. Braunschweig. A R. 1949): Section 19. 19. For the proof in quantum mechanics. and G. 1st edn (McGrawHill. 21. 486. Rose. in Modern Developments in Thermodynamics (Wiley. D. 1931). Phys. e. R..g. N. Phil. 1927. Amsterdam.. New York. Collected Scientific Papers (NorthHolland.Non Relativistic Theory. Rev. Quantum Mechanics . A. and references therein. Mag. M. 1977): Section 28. Froissart. Wigner. H. e. also see E. see E.1736 (1949). Rev. 23. R. Atti Congr. S. 26. See. Phys. Kramers. Schiff. 1957): Chapter 2. Nature 144. Phys. (Princeton University Press. R. 75. Landau and E. Zeit.40 (1932). Yang. 28. Kramers. Peierls. Eckart. Elementary Theory of Angular Momentum (John Wiley & Sons. Oxford.g. Princeton. Placzek. New York. E.200 (1939). 29. e. N. .7. Nuovo Cimento 11. See. 25. 1972): Section 10.168 3 Scattering Theory CPT and violates T has both real and imaginary parts five standard deviations from zero. 95114. Phys. 31. 44. Rev. L. See. 40. Phys. New York. Yang and C. New York. e. A general version of this argument was given in the late 1960s in unpublished work of C. 1956). reprinted in H. 27. 123. F. M. Intern. 1068 (1953). Bohr. Gruppentheorie (Friedrich Vieweg und Sohn. Fabri. Como.. Wigner.
where the number of particles is typically fixed. Equivalently. They provided a natural formalism for theories in which massive particles as well as photons can be produced and destroyed. The great advantage of this formalism is that if we express the Hamiltonian as a sum of products of creation and annihilation operators. it is for this reason that the formalism of creation and annihilation operators is widely used in nonrelativistic quantum statistical mechanics. which goes beyond the need to quantize any preexisting field theory like electrodynamics.4 The Cluster Decomposition Principle Up to this point we have not had much to say about the detailed structure of the Hamiltonian operator H. beginning in the early 1930s with Fermi's theory of beta decay. However.2 but such efforts have always run into trouble in sectors with more than two particles: either the threeparticle Smatrix is not Lorentzinvariant. There have been many attempts to formulate a relativistically invariant theory that would not be a local field theory. In this chapter we will first discuss the basis of states containing ar169 . and has nothing to do with whether particles can actually be produced or destroyed. We saw in Chapter 1 that such creation and annihilation operators were first encountered in the canonical quantization of the electromagnetic field and other fields in the early days of quantum mechanics. there is a deeper reason for constructing the Hamiltonian out of creation and annihilation operators. the cluster decomposition principle. and it is indeed possible to construct theories that are not field theories and yet yield a Lorentzinvariant Smatrix for twoparticle scattering. with suitable nonsingular coefficients. as we shall show here. or else it violates the cluster decomposition principle. This operator can be defined by giving all its matrix elements between states with arbitrary numbers of particles. the cluster decomposition principle plays a crucial part in making field theory inevitable. Indeed. 1 which says in effect that distant experiments yield uncorrelated results. then the Smatrix will automatically satisfy a crucial physical requirement. In relativistic quantum theories. any such operator may be expressed as a function of certain operators that create and destroy single particles.
(4."n . 2. and dots representing other particles that may be present in the state.11' and pi. then the statevectors l1l.·.. 4. As far as we know. all particles are either bosons orfermions... for definiteness we shall deal here with the freeparticle states l1l P1 ""1 n].. In this section we shall offer a nonrigorous argument that all particles must be either bosons or fermions.' the coefficients of these multiparticle basis vectors in physically allowable state·vectors.' n'" p... and finally show how their use facilitates the construction of Hamiltonians that yield Smatrices satisfying the cluster decomposition condition.. the difference being that a state is unchanged by the interchange of two identical bosons.. or 'out' states. That is <D"'p. respectively. Since the two statevectors are physically indistinguishable. this could be stated as a condition on the 'wave functions. for massless particles) and n labels particle species. These can be freeparticle states. As usual. but we shall not need this information in the present chapter.· p . respectively.n"'p''''n''' = + <D"'p''''n'''p. if this were not the case then the particles would be distinguished by their order in the labelling of the statevector. and changes sign under the interchange of two identical fermions.) These two cases are often referred to as Bose or Fermi 'statistics'... a' belong to identical species n. 1. .n'" represent the same physical state.170 4 The Cluster Decomposition Principle bitrary numbers of bosons and fermions... pa'I"·Va'n .P2""2nc. and then set up normalization conditions for multiboson or multifermion states. We must now go into a matter that has been passed over in Chapter 3.. p'.n''' (4.2) where CL n is a complex number of unit absolute value..1) with an upper or lower sign if n is a boson or a fermion. they must belong to the same ray.. or 'in' states. We may regard this as part of the definition of what we mean by identical particles.1 Bosons and Fermions The Hilbert space of physical states is spanned by states containing 0.. 11' labels spin zcomponents (or helicities. = Gl n l1l. . We will see in the next chapter that Bose and Fermi statistics are only possible for particles that have integer or halfinteger spins. and so $ . First note that if two particles with spins and momenta p. and the first listed would not be identical with the second." but all our results will apply equally to 'in' or 'ouf states.1. p'a'n"·p.1. then define the creation and annihilation operators. free particles. (Equivalently.. the symmetry properties of these states.n'" p'(f'n'" and 11l.
What about interchanges of particles belonging to different species? If we like. it is convenient to adopt a convention that generalizes Eq..1.'''·pa II .1 Bosons and Fermions 171 The crux of the matter is to decide on what the phase factor an may depend. pa n . we find . but this would lead to the uncomfortable result that the symmetry of statevectors under interchange of particles here on earth may depend on the presence of particles elsewhere in the universe. then we are nearly done. (4. The logical gap in the above argument is that (although our notation hides the fact) the states IPplajll.1. but not for three or more spatial dimensions. p' a' n  so that a~ = 1. because then these spindependent phase factors would have to furnish a representation of the rotation group.y. .1. The phase :'til might conceivably depend on the momenta of the two particles that are interchanged.1) and (4.  _ an 2.p' a' 11 . We will see in Section 9. and therefore such dependence would not change the argument leading to the conclusion that a~ = 1. by phase factors. we can allow the particle labels to appear in any order. '¥". If it depends only on the species index n. This is the sort of thing that is ruled out by the cluster decomposition principle.2)). yielding Eq. but Lorentz invariance would require an to depend only on the scalar rJ. (4. (4. etc.y.1): the statevector will be taken to be symmetric .. Interchanging the two particles in Eq. this is symmetric under interch<lllge of particles I and 2. and define the statevectors with particle labels in an arbitrary order as equal to the statevector with particle labels in some standard order times phase factors. and there are no nontrivial representations of the threedimensional rotation group that are onedimensional .4. whose dependence on the interchange of particles of different species can be anything we like. to be discussed later in this chapter.. In order to deal with symmetries like isospin invariance that relate particles of different species. we can avoid this question by simply agreeing [rom the beginning to label the statevector by listing all photon momenta and helicities first.1. The phase:'t ll cannot have any nontrivial dependence on the spins of the two particles that are interchanged. so that a~ 1.1. (4.P2(J"211..P2/l. 'J.7 that this is a real possibility in twodimensional space.that is. In this case the interchange of two particles twice might change the state by a phase factor. then all electron momenta and spin zcomponents. and so on through the table of elementary particle types. Alternatively.2) again.1.may carry a phase factor that depends on the path through momentum space by which the momenta of the particles are brought to the values pl.1) as the only two possibilities. P2. On what else could an depend? It might depend on the numbers and species of the other particles in the state (indicated by dots in Eqs.
in all cases... and also under interchange of the qj. More generally.. We adopt the C01went!()U that the statevector is symmetric or antisymmetric nnder interchange of identical bosons or fermions of different helicities or spin zcomponents in order to facilitate the use of rotational invariance. in the first term in Eq. p. l1lq1q2 ) = b(ql .q2) + b(q~  ql) b(q.." II b(qj " i q&J (4.. It is easy to see that Eg.172 4 The Cluster Decomposition Principle under interchange of any bosans with each other.1.5) On the other hand. for N = 2 the states l1l q and l1l q are physically the .7) has the desired symmetry or antisymmetry properties under interchange of the qi. then the momenta of all photons of helicity I. we will use a label q to denote all the quantum numbers of a single particle: its momentum. . f!jJ is the identity.1. This obviously is consistent with the above stated symmetry properties of the states. q~f' l1lqJ Q2 . . . * The normalization of these states must be defined in consistency with these symmetry conditions. (4. helicity) a.q. 7) The sum here is over all permutations f!jJ of the integers 1.q) _ 03(p' .if both particles are fermions and + otherwise. or any bosans with any fennions.q). (l1lq.) Also. (4. is a sign factor equal to 1 if &' involves an odd permutation of fermions (an odd number of fermion interchanges) and +1 otherwise. .6) the sign + being . o(q' . The Nparticle states are thus labelled tllq. so here we must take (l1l q .o"".. spin zcomponent (or.". same. for massless particles.QN) = bNM L b.3) and (<1>". (4. whether the particles are of the same species or not.1.) For N = 0 and N = 1 the question of symmetry does not arise: here we have (<1>0.4) where b(q'_q) is a product of all the delta functions and Kronecker deltas for the particle's quantum numbers. • In fact. (4. &2 = 2. .q2) (4.. the symmetry or antisymmetry of the statevector under interchange of particles of the same species but different helicities or spin ~components is purely conventional.q.1. +!. <l>q) ~ o(q' . .J!J2 = 1. &'1 = 1. To save writing. then the momenta of all electrons of spin ~component and so on. because we could have agreed from the beginning to list first the momenta of photons of helicity +1. <1>0) ~ 1 (4. (For instance. .q. and antisymmetric with respect to interchange of any two fermions with each other. O&. N.p)o.ql) b(q~ .cP1 = 2. qN (with N = 0 for the vacuum state tllo. while in the second term . q.6).1. 2. by the same reasoning. and species n.1.'.1. .
) r=1 j> i=l N M = bN.1. As we shall now show.. (4. (4.9J of the remaining integers .. .qN) .2"".) .. In particular...N be written as a sum over the integer r that is permuted into the place.00" .1) In particular. a(q) removes a particle from any state on which it acts.M+I L(+)'1 (cIl q.r 1. The sum over permutations rY of 1. = ().7).. (4.2 Creation and Annihilation Operators Creation and annihilation operators may be defined in terms of their effect on the normalized multiparticle states discussed in the previous section.N into I.N ~ 1.7) twice. We can first 1..2.4.O". and over mappings .q". (4. qN) = bN.2." ·.1..3 ) with a +1 or 1 sign for bosons or fennions. i. qN) = (l1lqq. the Nparticle state can be obtained by acting on the vacuum with N creation operators at(q.'·' .g.'" a(q)cIl qj ··· qN ) . Hence.q.2 Creation and Annihilation Operators 173 4.q~. ·at(qN)oI>o ~ 01>".q:I1"' a(Q)l1l qj" . when the particles q ql .. The creation operator at(q) (or in more detail.7). respectively... which is then called a(q). now use Eq.1).. . We want to calculate the scalar product of a(q)l1lqlq2"'qN with an arbitrary state Wq'j"'q'.) II J(q. and is therefore known as an annihilation operator. Furthermore.2.1.. at(p.n») is defined as the operator that simply adds a particle with quantum numbers q at the front of the list of particles in the state (4. qN are either all bosons or all fermions.. + with upper and lower signs for bosons and fermions. we have N a(q)lDqlqrqN = :2)+)'+lb(q . (Here is the proof."qr_jqr+j ... this is (l1lq'I···q~.(a t (q)l1l q. respectively.2.e.·qN ) . the sign factor is Uf! . may be calculated from Eq. fJ'r = 1. its adjoint.. (l1l q.w' Using Eq. r + 1.Ii>. (4. using Eq.'''q~' r=1 N I1lqj .M+I x L2)+y1Jc>J(q .qr)wq!"'qr_jqr+I"qN ' r=1 (4.q:I1"' I1l qj" .)at(q.2) It is conventional for this operator to be called at(q). I1lqj .
6) and so also a(q')a(q) (4.7) As always. and anticommute if both are fennions.3) gives a(q')at(q)lDqj···qN = J(q' . we note that for both bosans and fermions. N + ~)+y+20(q' r=1 qr)lDqql"''I'_''Irc' "'iN • (The sign in the second term is (+1'+2 because qr is in the (r + l)th place in <I>qql"'q.3) thus have the same matrix element with any state lDql . (4. we could have started with the commutation or anticommutation relations Egs. Eq.5) In addition. (4.q.2.~.2. the creation and/or annihilation operators for particles of two different species commute if either particle is a boson. applying the operator at(q) to Eq.2.2.3). (4. and are therefore equal.) On the other hand."'q.. Applying the operator a(q') to Eg. a(q) annihilates the vacuum a(q)<I>o ~ O.5}(4. (4.)lD qqj 'qrlqr+I'"'q~' r=[ .2. q' .. The above discussion could have been presented in reverse order (and in most textbooks usually is). (4.2... Multiparticle .174 4 The Cluster Decomposition Principle Both sides of Eq.1) and using Eg.2) gives immediately at(q')at(q) + at(q)at(q'l ~ 0 + a(q)a(q') ~ o.) As a special case ofEq'~(4.3) gIVes N at(q)a(q')lDq[ .7). derived from the canonical quantization of some given field theory. (4.2.q)<I>ql''q. (4. as was to be shown. Subtracting or adding. According to the conventions discussed in the previous section. (4. = b(q' ~ q)lDql'"'qN . respectively.2. we have then [a(q')at(q) + at(q)a(q')] (J)qj'"'q".q. the top and bottom signs apply for bosons and fermions.~. the creation and annihilation operators satisfy an important commutation or anticommutation relation. But this holds for all states /1)ql. That is.2.~' (and may easily be seen to hold also for states containing both bosons and fermions) and therefore implies the operator relation a(q')a'(q) + at(q)a(q') ~ o(q'  q).4) As defined here.2.2.2. (4.v = ~)+y+lb(q' .
irrespective of the values of CNM with N > 0 and/or M > O. if the formula for some operator has the creation and annihilation operators in some other order. In fact.... Now suppose that the same is true for all matrix elements of (!> between N.7) derived from the commutation or anticommutation relations. We need only use Eq. we can always bring the creation operators to the left of the annihilation operators by repeated use of the commutation or . We will now prove the fundamental theorem quoted at the beginning of this chapter: any operator (1:: may be expressed as a sum of products of creation and annihilation operators o ~ I: X X.2. and their scalar products Eq.2..4.and Kparticle states. (4.2). Of course.ql . CI. M < K or N Whatever values have already been given to CNAt with N < L. an operator need not be expressed in the form (4.2.4) to see that Eq.8) to evaluate (ll>q. we can give (cDo. That is. that is. (4. and rather wish to understand why field theories are the way they are.1.M < K or N < L.8) x CNM(qj'" q~ql'" qM). (4. use Eq.8) has the vacuum expectation value (<1>0. &<1>0) ~ em .) However. it is trivial that by choosing Coo properly.)"'at(q~)a(qM)"'a(qJl (4. (4. We do this by mathematical induction.8).2. M < K . (This is often called the 'normal' order of the operators. . dq~dql dqM at(q. M < K or N < L.M < K.qK) = L!K ! C LK (q. such a treatment would be much closer to the way that this formalism developed historically. First. as discussed in Chapter 1. + terms involving CNM with N < L.". "'q~' IY'cDqj.and Mparticle states.2.. We have followed an un historical approach here because we want to free ourselves from any dependence on preexisting field theories. with N < L.. . we want to show that the C NM coefficients can be chosen to give the matrix elements of this expression any desired values. N=OM=O I: Jdq.2 Creation and Annihilation Operators 175 states would have then been defined by Eq.2. that these matrix elements have been given some desired values by an appropriate choice of the corresponding coefficients C NM • To see that the same is then also true of matrix elements of (!) between any L. qK) < L. (~'cDo) any desired value. q~. there is clearly some choice of CLK which gives this matrix element any desired value. (4. with all creation operators to tbe left of all annihilation operators. M < K.
.2. For instance.P2<J2n2.~ 01 111 . This is all for m f..2.j°(Ap2)O .pJl)Dg2~2(W(A.p) ."" 'Vpl':rlnJ.p) is the particular rotation W(A. (4.5). consider any sort of additive operator F (like momentum. (4." = at (PI 0"1 nda t (P20"2n2) .11) where E(q) is the singleparticle energy E(p. where l1lo is the Lorentzinvariant vacuum state . where L(p) is the standard 'boost' that takes a particle of mass m from rest to fourmomentum pJ'.n)=Jp2+m~.P2)) (rIOT" I1lpl.C 1 (Ap)AL(p).P2A 0'2"2.• !flo.. we will return to the massless particle case in the following chapter. Recall that the N~particle states have the Lorentz transformation property ")~ Uo(A .P20"2112.2. x _~ D~I~JW(A.'" ~ c:: ei(Ap!l·!t ei(Ap::l"x (Ap. but using only the term withN=M=l: F ~ J dq at(q)a(q)j(q) . (Of course. picking up new terms from the delta function in Eq. First. (4.2..9) Such an operator can be written as in Eq.10) In particular. charge.) for which (4.) Now.5.2. and W(A. etc. m and j depend on the species label n.2. (4. p?p~ .2) I1lPW1 "I . let's consider inhomogeneous proper orthochronous Lorentz transformations.. Here PA is the threevector part of Ap.0.8).CT." .176 4 The Cluster Decomposition Principle anticommutation relations. We will need the transformation properties of the creation and annihilation operators for various symmetries. these states can be expressed as in Eq. the freeparticle Hamiltonian is always Ho ~ J dqat(q)a(q)E(q) (4. DY1(R) is the same unitary spinj representation of the threedimensional rotation group as used in Section 2.
the cluster decomposition principle states that if multiparticle processes at .13) (4. and/or T are conserved. In Smatrix theory.P2. in which case we would introduce operators ain and aout defined in the same way by their action on these states. . J(Ap)O Ip o x LD¥)(W(A.3 Cluster Decomposition and Connected Amplitudes 177 In order that the state (4.Pl. although we have been dealing with operators that create and annihilate particles in freeparticle states. fL)a t (pan)Uol(A. . fL).12). that induce charge·conjugation. (4. for which it is necessary to distinguish betweeu the operators U(A.a} and Uo{A.fLA" . The probabilities for various collisions measured at Fermilab should not depend on what sort of experiments are being done at CERN at the same time. P.2.14) (4.2. (4. it is necessal)' and sufficient that the creation operator have the transformation rule Uo(A.p))at(PA. space inversion.2. fL) = ei{Ap)·'J. and T.2) should transform properly.2. then the Smatrix element for the overall process • We omit the subscript '0' on these operalon>. of all science) that experiments that are sufficiently separated in space have unrelated results.' ./( are studied in % very distant laboratories. the operators C.15) As mentioned in the previous section. then we could never make any predictions about any experiment without knowing evel)'thing about the universe. the operators that induce these transformations on 'in' and 'out' states are the same as those defined by their action on freeparticle stales.2. and timereversal transformations on free particle states· transform the creation operators as: (4. fL) instead of the free· particle operator Uo(A. This is not the case for continuous Lorentz transformations.3 Cluster Decomposition and Connected Amplitudes It is one of the fundamental principles of physics (indeed.4. but with the true Lorentz transformation operator U(A.P. a2 . the whole formalism can be applied to 'in' and 'out' states.n).a).2. These operators satisfy a Lorentz transformation rule just like Eq. P. 4. because in virtually all cases where C.. If this principle were not valid. In the same way.12) .
(4. into clusters Clj."2. partition functions..cd in the theory of noise 6 to decompose tbe correlation function of several ranllom variables into its 'cumulant~'. and in a box the llelta function become~ a Kronecker delta times thc volume.2) Here the sum is over all different ways of partitioning the particles in the state Ci. and Ii. and likewise a sum over all ways of partitioning the particles in the state Pinto clusters Pl. The sign is + or ~ according to whether the rearrangements rx .pcl:ie.fJdh··· involve altogether an even or an odd number of fermion interchanges.' + (II' •• This decomposition has been used IJl classll:ul stahstil:ul mechanics by Ur. and in quantum statistical mechanics by Lee and Yang and others. That is: (4. with a total number of particles in each of the states ':i. The term 'connected' is used because of the interpretation of S~~ in terms of diagrams representing different contributions in perturbatIOn theory.C(2. the sum on the righthand side of Eq.3. by the formula·· Sia."'. amI . For each r:t. (4. because as we shall see. XI + . Sf/x = L PART (+) Sf:. spin."'.). The cluster decomposition is also the same formal device as that us. Greek letters (l or fJ stand for a collection of particles.3. . and others.P2. to be discussed in the next section. Mayer. + IX" is the .li Sh·12··· (4.j... including for each particle a specification of its momentum. +. There is a combinatoric trick that allows us to rewrite Eq. respectively.rxlrx2'" and fJ ..el1. .) It has also been used to calculate manybody ground state energies by Goldstone 4 and Hugenholtz. Suppose we define the connected part of the Smatrix.3. ano X.2) consists of a term Six' plus a sum 1:' over products of two or more SCmatrix elements.1) in a more transparent way. etc. if tbe random variable receives contributions from a large number N of independent fluctuations. 5 In all of these applil:ations the porpose of isolating the l:onnected parts of Green's functions."'. corresponding to uncorrelated experimental results. then eacb cumulant is proportional to N. all of the particles in states (Xi and #i are at a great spatial distance from ali of the particles in states elj and Pj.tate formed by combining all the particles in the states "1. is to deal with objects with a simplc volumc dependence. This is a recursive definition. ano likewise for {II + f32 +. This is essentially our purpose too. This factorization of Sif for all i matrix elements will ensure a factorization of the corresponding transition probabilities.. not counting as different those that merely arrange particles within a given cluster or permute whole clusters.j and f3j that is less than the number of particles in • We arc here returning to the notation used in Chapter 3.3.j.1) 1. resolvents.178 4 The Cluster Decomposition Principle factorizes. Also. the crucial property of the connel:tell parts or the Smatnx is that they are proportional to a single momentumconservation delta function.
so that there are no transitions between singleparticle states and any others. and otherwise +. Eq. so here Sf~ is just (S l)pll' But the general case is more complicated. (4. containing together fewer than. (4. (4. Suppose that the SCmatrix elements in this sum have already been chosen in such a way that Eq.q) follows from conservation laws.q2)6(q.2) is just itself.2) contains no information in itself.'1~.3.2) reads Sq't'1. For transitions between threeparticle or fourparticle states. (4.q3. so for oneparticle states Sfi: Sq'q ~ Sq'q ~ o(q . (4. (4.2) is also satisfied for states a.1.3.6). Eq. it is merely a definition of SC.j and fii both contain no particles il[ ilB. the slates'.ql)()(q. We must thereltlre de/ine [he connected vacuumvacuum element So:o to be zero.qlqZ + J(q') q!lJ(q. (4..'.3.. which in the absence of timevarying external fields is simply uefined to be unity.3. We recognize that the two delta function terms just add up to the norm (4. 50. (4.QI'1Zq. . N particles.3. The absence of any proportionality factor in Eq.>C Eq.jSft.Qlq2q) = I + J(q'J sign + is  Q2)O(Q2 . (433) (Apart from possible degeneracies.([lIJ2 = S~. . we can always choose the remaining term Sffll so that Eq. such as the vacuum.q3) + permutations (43.5) t A technicality should be mentioned here.2). then the only term on the righthandside of Eq.o. For transitions between twoparticle states.4. say.q). (4. t Thus Eq. If the states 'J.) We are here assuming that singleparticle states are stable.3) is based on a suitable choice of the relative phase of 'in' and 'out' states.2) is satisfied for states p.3.qd .3. We do not u. Then no matter what values are found in this way for the Smatrix elements appearing in the sum 1:'. the fact that Sqlq is proportional to 6(q' .4) sS . .3. + L ' (+) S}i. and P Sfl'.3. 'J. This argument works only if we neglect the possibility that for one or more of the connecteu Smatrix elements in Eg.2) reads S'1I"Q2q J. c _ .3).2) for the vacuum vacuum Smatrix So.3.q2) (We are here using Eq.p containing a total of N particles.3 Cluster Decomposition and Connected Amplitudes 179 the states 'J. (4. say with quantum numbers q and q' respectively.) The if both particles are fermions.0 = 1. and P each consist of just a single particle.3. (4. = S~.. q1Q2 +J(q'j . We will have more to say about the vacuum vacuum amplitude in the presence of external fields in Volume II. .1l2'" PART . .qdSqCiq'3'q2q) + permutations Z +6(qj .
(4.3.< for threeparticle states.'" and ajI. 3'.7) where :Eli) is a sum over all different ways of partitioning the clusters Pj and Ct. there are a total of 1 + 9 + 6 = 16 terms in Eq.Pz. (4. (4.3. and that all particles in the set ai + Pi are far from all particles in the set Ct.6). (4. Then if Sio: vanishes when any particles in P and/or rt are far from the others.) into subclusters (3)I."'. we let particles 1. and 2' be very far from 3.3. Then if Sf.3.2. there would be even more terms.3. we will have to Fourier transform SC. it vanishes if any particles in these states are in different clusters..3.3.5) when we define Sf.j + (3) for any j f. :):2.4) to define SiC! for twoparticle states.180 4 The Cluster Decomposition Principle and (Taking account of all permutations.3. (4. and so on.1).6) that survive (in an even more abbreviated notation) are :I: In order 1.a)z. the definition of Sfi~ is recursive: we use Eq.5) and 1 + 18 + 16 + 72 + 24 = 131 terms in Eq. (4.'" and ':):1.4.i. For instance.0 give a meaning to 'far'.PjZ. . (4.t To see this. this is just the desired factorization property (4.) As explained previously.3.' are far from the others. The point of this definition of the connected part of the Smatrix is that the cluster decomposition principle is equivalent to the requirement that must vanish when anyone or more of the particles in the states f3 sg and/or ex are far away in space from the others. suppose that the particles in the states p and ex are grouped into clusters [h. so the definition (4.6) to obtain the definition of Sfo: for fourparticle states. so that each threemomentum label p is replaced with a spatial coordinate threevector x. then use both of these definitions in Eq. and 4' . 1'. If we had not assumed that one·particle states are stable.···. suppose that in the fourparticle reaction 1234 + 1'2'3'4'.3. But referring back to Eq. (4.czl vanishes if any particles in pi or Ct. the only terms in Eq. then use this definition in Eq.2) yields .2).
3. like the Smatrix itself.34 03'404'3)SI'2'. 12S3'4'34 We have formulated the cluster decomposition principle in coordinate space.) If IS~lP. Translational invariance teUs us that the connected part of the Smatrix. we can write S~IP. (4.'" (438) (We are here temporarily dropping spin and species labels. with their differences held constant. The coordinate space matrix elements are defined as a Fourier transfonn SC ~Jd3 PI 3P2 .3."'P1P2""1 not Lebesgue integrable)..4. if C itself in Eq.3. as well as the energyconservation delta function required by scattering theory. as the condition that S(x vanishes if any particles in the states p or a. ifit were Lebesgue integrable) then according to the RiemannLebesgue theorem? the integral (4. then this principle would not be satisfied.""XIX2'" = SC P.34 + (01'102'2 + + (03'304'4 + + (0\'10212 + 01'202'l)Sf4'.ll34 ) 181 Sfll.12S~41.. It is convenient for us to reexpress this in momentum space. x~X.1) S1'21 3141 •1234 + SI'l' • • .3 Cluster Decomposition and Connected Amplitudes SI12' 3'4' .. However. become large.3. and therefore does not change at all if aU of the Xi and xj vary together. be proportional to a threedimensional delta function that ensures momentum conservation (and makes IS~P.9) This is no problem: the cluster decomposition principle only requires that Eq. For instance.4) shows that this is just the required factorization condition (4."'.12 01'202'1)(03'304'4 C + °3' 4b4' 3).8) vanish when the differences among some of the Xi and/or x.P.''''P1P2'' = b 3(p'l + P2 + . ) x o(Ei + E~ +  EI .P1 .8) would vanish when any combination of spatial coordinates goes to infinity. e'p.x. (4.3. this is certainly too strong a requirement. This requires that the elements of SC in a momentum basis must. like those of S. Comparison with Eq. are far from any others.P2 _ . suppose that there were a delta function in C that required that the sum of the and Pj for some subset of the particles vanished. Then Eq.8) would not vary if all of the x.3. can only depend on differences of coordinate vectors. d 3 PI d 3P2'" 'd ."'P1P2'" P 'I L. 'p"X' ip"x' XeJ1e22"'e jp".. (4.3. (4. which just go along with the momentum or coordinate labels.9) contained additional delta functions of linear combinations of the threemomenta. That is.E2 " ')Cp . Now. and p: .PIP2'" (4."PJP2"_1 were sufficiently well behaved (to be specific.
SC can have poles at certain values of the P and p'. in fact.x. qj"'q". (4. and. the requirement of analyticity is not met in all theories.. an exponential falloff of SC is not an essential part of the cluster decomposition principle. After Fourier transforming.1) with coefficient functions h NM that contain just a single threedimensional momentumconservation delta function (returning here briefly to a more . But how smooth? It would be most straightforward if we could simply require that Cp 'IP."dq~dql'dqM N=O M=O Q' a:) x a"(q~)"'at(q:'I')a(qM)"'a(qd x hNM(ql"'q~. However. as we will see in Chapter 10. .P3)2.. if a massless particle can be emitted in the transition 1 _ 3 and absorbed in the transition 2 _ 4. The answer is contained in the theorem that the Smatrix satisfies the cluster decomposition principle if (and as far as I know.."'. For instance.3.. = PI = P2 = .. = O.12 will have a term proportional to 1/(P1 . but not singularities as severe as delta functions.. In order to put this a bit more precisely.9) is a smooth function of its momentum labels.PIP~'" in Eg. Most notably. unlike the Smatrix itself: contains just a single momentumconservation delta junction. Loosely speaking then. such poles yield terms in x' x.and Xi. 1 There is no need to formulate the cluster decomposition principle so stringently that such behavior is ruled out.tl (4.182 Xi 4 The Cluster Decomposition Principle for the particles in that subset moved together (with constant differences) away from all the other xi.4 Structure of the Interaction We now ask.8): H~L LJdq.4. This requirement would indeed guarantee that S~x. we can say that the coefficient function Cp'p~"'. only if) the Hamiltonian can be expressed as in Eg. Thus the 'smoothness' condition on SC should be understood to allow various poles and branchcuts at certain values of the p and p'. 4. the cluster decomposition principle simply says that the connected part of the Smatrix.that fall sS I 2' L off only as negative powers of coordinate differences. then SS.. in theories with massless particles.2.xlxr vanishes exponentially fast when any of the x and x' is very distant from any of the other x and x'.PIP2'" be analytic in all of the momenta at PI = P2 = .. in contradiction to the cluster decomposition principle. what sort of Hamiltonian will yield an Smatrix that satisfies the cluster decomposition principle? It is here that the formalism of creation and annihilation operators comes into its own. (4.
.)}iD.1) by itself has no content . '"()' 00 [oodtj'dt.x + i€jl is not.5) we may move each annihilation operator in turn to the right past all the creation operators.'l':) in S~". we make use of perturbation theory in its timedependent form. (4. the proof has some instructive features.4.4. For each annihilation operator moved to the right past a creation operator we have two terms.T{V(t!l'V(t.2.4. However. so each term in the sum (4. and V(t) is itself a sum of products of creation and annihilation operators. • We are now adopting the convention that for n '" 0.4.(iDl.1) combined with the requirement that h NM has only the single delta function shown in Eq.5. (4. The trusting reader may prefer to skip the rest of the present chapter.2. as shown by writing Eq.2) that guarantees that the Smatrix satisfies the cluster decomposition principle.4.we saw in Section 4. and move on to consider the implications of this theorem in Chapter 5. It is only Eq.) The Smatrix is given by Eq.10) as· Sp.3) is taken as the unit operator.2) where hNM contains no delta function factors.4. and V(t) =exp(iHot)Vexp(iHot). To prove this theorem. By using the commutation or anticommutation relations (4. Note that Eq.4 Structure oj'the Interaction 183 explicit notation) (4. (4. and will help to clarify in what sense the field theory of the next chapter is inevitable. The validity of this theorem in perturbation theory will become obvious when we develop the Feynman diagram formalism in Chapter 6.2.2) as products of creation operators acting on the vacuum lDo.~~ ~'.3) where the Hamiltonian is split into a freeparticle part Ho and an interac· tion V.4) Now. (4.4. (4.4. ! (4. so the n '" 0 term ltl the sum just yields the term J(1l . if E is a sum of oneparticle energies then e.4.iEt is a product of functions of the individual energies. (One of the advantages of timedependent perturbation theory is that it makes the combinatorics underlying the cluster decomposition principle much more transparent.5) in the form a(q')at(q) ~ + at(q)a(q') + D(q'  q) . (4.3) may be written as a sum of vacuum expectation values of products of creation and annihilation operators. (3. the states ID x and l1lfl may be expressed as in Eq. while [E .E. (4. the time·ordered proJuct in Eq.).2 that any operator can be put in this form.
so in the end all we have left is the delta functions. we are not including any terms in which an annihilation operator in one vertex destroys a particle that is produced by a creation operator in the other vertex .. (This is not yet the full Feynman diagram formalism. because for this diagram. right through the diagrams. each term equal to a product of delta functions and + signs from the commutators or anticommutators. spins. For each delta function produced when an annihilation operator in the adjoint of the final state lPfl moves past a creation operator in one of the V(t). Each of the delta functions associated with one of these lines enforces the equality of the momentum arguments of the pair of creation and annihilation operators represented by the line. In this way. Finally. (4. Each of the terms generated in this way may be symbolized by a diagram. and species in the arguments of the delta functions. But Eq.4) shows that any annihilation operator that moves all the way to the right and acts on $0 gives zero.4. It follows that each term in Eq. draw a line coming into the diagram from below that ends at the corresponding vertex. draw a line from the corresponding vertex upwards out of the diagram. it breaks up into a number of connected pieces. which enforces the conservation of the total threemomentum at the vertex. integrated over all the times and integrated and summed over the momenta. For each delta function produced when an annihilation operator in one V(t) moves past a creation operator in another V(t) draw a line between the two corresponding vertices. called vertices.184 4 The Cluster Decomposition Principle Moving other creation operators past the annihilation operator in the first term generates yet more terms. draw a line from bottom to top. The V(t) operator associated with a vertex in one connected component effectively commutes with the V(t) associated with any vertex in any other connected component. the two vertices would be in the same connected . There is also at least one delta function contributed by each of the vertices. we are using the diagrams here only as a way of keeping track of three~momentum delta functions. For each delta function produced when an annihilation operator in one of these V(t) operators moves past a creation operator in the initial state lP:t.if we did. (4. each term equal to a product of delta functions and ± signs from the commutators or anticommutators and whatever factors are contributed by V(t).) Draw n points. because we are not yet going to associate numerical quantities with the ingredients in the diagrams.2.3) may be expressed as a sum of terms. Such a diagram may be connected (every point connected to every other by a set of lines) and if not connected. one for each V(t) operator. for each delta function produced when an annihilation operator in the adjoint of the final state moves past a creation operator in the initial state. the vacuum expectation value of a product of creation and annihilation operators is given by a sum of different terms.
Some of the clusters in Eq.4. .) } <1>. Now let us use Eq. t n are sorted out into each cluster.. nl" PART I~! ..4. ell' and Pl .{3))).4. T {V(tIl'" V(t. Ct. one from each connected component: (<1>1.4. because the only connected diagrams without vertices consist of a single line running through the diagram from bottom to top.2. nj = 0.3) can be expressed as a sum over products of contributions.5) may contain no vertices at all.3).. The factor n! here cancels the lin! in Eq.. that is. (4. (4.L(+) L fl\+' nt . (i)Il" so instead of summing over nand . and likewise for the final state... for these factors. n2.) nl ..4. Thus the matrix element in Eq. . ) L clusterings (+) j=l II (<I>lj' T {V(tj.l ' .(<I>p.4. Every time variable is integrated from 00 to +00. (4. " V(tj'J) }<I>'j) C The first sum here is over all ways of partitioning the particles in the initial and final states into clusterSCi.4 Structure of the Interaction 185 component.. i.) ...j are both oneparticle states (in which case it is just a delta function (j(elj .5) can be written as a product (_i)n 1 . each containing nl.5) means that we exclude any contributions corresponding to disconnected diagrams. we must take the matrix element factor in Eq. el v . T{V(tIl'" V(t..5) in the sum (4.4. equal to the number of ways of sorting out n vertices into v clusters. V(tj.5) Here the sum is over all ways of splitting up the incoming and outgoing particles and V(t) operators into v clusters (including a sum over v from 1 to n) with the nj operators V(tj j ) ' " V(tjn) and the subsets of initial particles rtj and final particles Pj all in the jth cluster. Most important.=n x (<1>/1' T {V(tj.)}<I>.4.. (4. ' '+n.. (4.e.3). Of course. + nv and also the set rt is the union of all the particles in the subsets C11. this means that n = nl + . and the factor (on in the perturbation series for (4.. P" (including a sum over the number v of clusters)..) }<I>'j)C (4.."dtJ'.' .) . any contributions in which any V(t) operator or any initial or final particle is not connected to every other by a sequence of particle creations and annihilations. fl. vertices: L:  dttdt. (4. nl ..5) to vanish unless Pj and Ct.!. The sum over clusterings therefore yields a factor n !/nl !n2! .4.n2· . so it makes no difference which of the tl.4. 1=1co i:Il~dtJ. the subscript C in Eq.
What we want to prove is that Sfix contains no other delta functions... momentum is conseI>'ed at each vertex and along every line. dtj'i PART j=1 nj=O x (1I>Pi' T{V(tj.) simply keep the momentum of any particle that is not created or annihilated at the corresponding vertex unchanged. so the connected parts of the Smatrix individually conserve momentum: Sf~ contains a factor (p(PfJ .)}1I>. + n v = we can simply sum independently over each nl.186 4 The Cluster Decomposition Principle n. (The other delta functions in matrix elements VI'. (Any line which if cut leaves the diagram disconnected carries a momentum that is fixed by momentum conservation as some linear combination of the momenta of the lines coming into or going out of the diagram. so it is also then true separately for the interaction V.Pal. (Wj 1 nJ' jOC . this means that each vertex contributes one threedimensional delta function.) We see that Sfi~ is calculated by a very simple prescription: SiC( is the sum of all contrihutions to the Smatrix that are connected.1) of the Hamiltonian in terms of creation and annihilation operators are proportional to a single threedimensional delta function.4.. (1I>p..X.4. . This gives finally Sl' ~ I: (+) II I: IX. This is automatically true for the freeparticle Hamiltonian H o.2) of the connected matrix elements SFr).f in the expansion (4. as these are now mere integration and summation variables.)c· (4.6) (The subscript j is dropped on all the t sand n s. then summing separately over nl. We now make the assumption that the coefficient fractions hNll. in the sense that we drop all terms in which any initial or final particle or any operator V(t) is not connected to all the others by a sequence of particle creations and annihilations. dt j. that ensures momenta conservation. As we have seen. then we say it has L independent loops.2.T{V(tt!'V(t. most of these delta functions simply go to fix the momentum of intermediate particles. Returning to the graphical interpretation of the matrix elements that we have been using. The only momenta that are left unfixed by such delta functions are those that circulate in loops of internal Jines. If the diagram has L lines that can all be cut at the same time without the diagram becoming disconnected. This justifies the adjective 'connected' for Se.' .'" nv.)"'V(tj'i}1I>'i)C' Comparing this with the definition (4.) Now. and there are L .' we see that these matrix elements are just given by the factors in the prod uct here sfr>ro ~! OC ( i)' jOC _ocdtl'dt. n v constrained by nl + .
Then Eq.) With V vertices. It was not important in the above argument that the time variables were integrated from 00 to +00.1) should have only a single threedimensional momentum conservation delta function factor is very far from trivial. the numbers of vertices.M in the Hamiltonian contain just single delta functions. I internal lines.P3) + permutations.momenta that are not fixed by momentum conservation. so I = V + L 1 and C = I. leaving V . each containing a single momentumconservation delta function factor.PI P2P1 = V3.ls of C such connected parts. (4.) But then the matrix element of the interaction between threeparticle states is VpIIP.4. to) can also be decomposed into connected parts. and C = 1. But a wellknown topological identity*· tells that for any graph consisting of C connected pieces. the connected part of the Smatrix also contains an energyconservation delta function.1) must contain a term with N = M = 2. there are V delta functions. i5(Ep .2(P'IP'2. Thus exactly the same arguments can be used to show that if the coefficients hN . and loops are related by (4.3(P'IP2P~. On the other hand. of which I . [f a disconnected graph eon.9) + V2.Pl P2P3) . L = 0. and L loops.. contains only a single energyconservation delta function. L = 0. and 1. and has farreaching implications. (4.I + L delta functions relating the momenta of incoming or outgoing particles.4.4. It should be emphasized that the requirement that hNM in Eq.i. and when we come to Feynman diagrams in Chapter 6 we shall see that S1".E.4. and C = 1.x ). in each connecled part will than satisfy LI~LV+LLC . Any additional intemallines attached (without new verti"es) to the same connected graph produce an equal number of loops. the sums of 1. assume that V has nonvanishing matrix elements between twoparticle states. internal lines. to) contains no energyconservation delta functions at all. V.8) (We are here temporarily dropping spin and species labels. For instance. (4. we find just a single threedimensional delta function b 3(pp Px). P1P2) i53(p~ •• A graph consisting of a single vertex has V = 1. while U(t.4. as was to be proved. then U(t. If we add VI vertices with just enough internal lines to keep the graph connecled. and coefficient (4.p.7) Hence for a connected matrix element like Sffx' which arises from graphs with C = 1.L go to DX internal momenta. we have 1 = VI.
2(p ~ p. In consequence.2 so that the twobody Smatrix is Lorentzinvariant.3 to cancel the other terms in Eq. PI P2 P3) = V2.4. provided that the Hamiltonian satisfies our condition that the coefficient functions hN. this would mean that each term in function factors (recall that V2. but they have not turned out to be useful in relativistic theories and so will not be described in detail here. PI P2) b 3(p 3. it is necessary to replace the LippmannSchwinger equation with one that has a connected righthand side. (4.' PIP2) has a factor b 3(p'] + P 2  contains two delta PI ~ P2) ) and this would violate the cluster decomposition principle.4. Our arguments in this section have so far relied on perturbation theory. However. To solve problems involving three or more particles. Thus in a theory satisfying the cluster decomposition principle. • •• When we set out to solve threebody problems in quantum theories that satisfy the cluster decomposition principle. even one of very large rank.9) V3.188 4 The Cluster Decomposition Principle As mentioned at the beginning of this chapter. as required by the cluster decomposition principle.3 IP.3(p 'I P 2P. the existence of scattering processes involving two particles makes processes involving three or more particles inevitable. but the extra delta function in the other terms makes the LippmannSchwinger equation difficult to solve directly.10) However. Such equa..8.9 and in nonrelativistic scattering problems they can be solved recursively. (4. it cannot be approximated by a finite matrix. we might try to make a relativistic quantum theory that is not a field theory by choosing V2. but it has been shown9 that these reformulated nonperturbative dynamical equations are consistent with the requirement that UC(t.EfJ + iEjl VP:t of this equation not squareintegrable.9) gives no particular trouble. We would then have to take V3. The problem is that these delta functions make the kernel [E(t .P3) + permutations.3 in Eq.2(P V3. I do not know of any nonperturbative proof of the main theorem of this section.4. (4. .M each contain only a single momentumconservation delta function. recasting the LippmannSchwinger equation in this manner is useful in another way. even after we factor out an overall momentum conservation delta function.tions have been developed for the scattering of three or more particles. and adjusting the rest of the Hamiltonian so that there is no scattering in states containing three or more particles.. to) (and hence Sc) should also contain only a single momentumconservation delta function. the term V3.
Soc. Yang.Problems 189 Problems 1...ql . The cluster decomposition principle seems to have been first stated explicitly in quantum field theory by E.M. Bakamijian and L. 267 (1957). Rev.qMdqi ···dq~.. H. Lee and C. Proc. Crichton. See. Wichmann and 1. For references. e. 3. H. References 1. 1300 (1953). 1. (You may consider the purely bosonic case. ./ v'(qi)'''v'(q. Rev. 2. Define generating functionals for the Smatrix and its connected part: F[v] 1+ ~ltl N!~!. q~. N. Consider an interaction 3 3 3 3 3 V = g d pl d p2 d p3 d p4 6 (Pl J + P2  P3 .qM dqi ···dq~dql·· 'dqM Fe [v] 2: 2: N.'q~"q]" . B.")v(q!l"'V(qM) N=lM=1 I 00 1 x s~ . Goldstone. Phys. see T. D.. Phys. 113. where g is a real constant and a(p) is the annihilation operator of a spinless boson of mass M > O. London A239.l is defined to be an eigenstate of the annihilation operators a(q) with eigenvalues . H.) 2. Phys..1.'')v(q!l'''V(qM) 00 x Sq. 2788 (1963). Rev.g. 132.dql"·dqM' Derive a formula relating P[v] and pe[u].. 1165 (1959). A coherent state /1). Use perturbation theory to calculate the Smatrix element for scattering of these particles in the centerofmass frame to order g2. Construct such a state as a superposition of the multiparticle states /1)qlqr'QN.(q). Thomas. .'/v'(qi)"'V'(q. What is the corresponding differential crosssection? 3. Roy.P4) x a t (Pl)a t (P2)a(P3)a(P4). 4. 92.
C. Phys. Nouk. Rec.. Physica 23. R.8.190 4 The Cluster Decomposition Principle 5. Akod. 39.JETP 12. 9. 565 (1961) and 145. 481 (1957). 133. L. 7. e. 384 (1961) and 7. 1937): Section 1. 6. Kubo. 30 (1962) (translations Soviet Physics Doklady 6.g. Phys. Fiz. Faddeev. 8. M. 174 (1963). Hugenholtz. B232 (1964) . Oxford. Math. 1459 (1961) (translation: Soviet Phys . E. Dokl. 4. Weinberg. i Teor. Zh. J. 1014 (1961)). See. 600 (1963». Ekxper. SSSR 138. Introduction to the Theory of Fourier Integrals (Oxford University Press. D. Titchmarsh. N. S.
or can also include space inversions. (5.both 191 . and various relationships between particles and antiparticles.1. the existence of antiparticles.>!"(Ax + a).2. (5.5 Quantum Fields and Antiparticles We now have all the pieces needed to motivate the introduction of quantum fields. (5. under Lorentz transformations each such operator is multiplied by a matrix that depends on the momentum carried by thal operator.a). 5.! In the course of this construction. including the celebrated CPT theorem.JI"(x')] ~ 0 for (x . t Free Fields We have seen in Chapter 3 that the Smatrix will be Lorentz+invariant if the interaction can be written as V(t) ~ J d'x £(x.) In order to facilitate also satisfying the cluster decomposition principle we are going to construct Jf'(x) out of creation and annihilation operators. (4.1. How can we couple such operators together to make a scalar'? The solution is to build Jf(x) out of fields . there are more general possibilities.a) ~ . but none of them are very different from this.2) and satisfies the additional condition: [£(x). but here we face a problem: as shown by Eq.1) where <)I{' is a scalar. in the sense that Uo(A.1. we shall encounter some of the most remarkable and universal consequences of the union of relativity with quantum mechanics: the connection between spin and statistics. (For the present we are leaving it as an open question whether A here is restricted to a proper orthochronous Lorentz transformation. t).x')2 > O.3) As we shall see.JI"(x)Uo'(A.12).
a) ~ L L l ..O". but as we shall see.p. a) ~ D. (5. we do not require at this point that D(A) be irreducible. D.(x) J ~L J d' "" d3p u.1) ~ D«AA)I). That is.n)a(p. we see that the Dmatrices furnish a representation of the homogeneous Lorentz group: (5. have different transformation matrices D± for the annihilation and creation fields. These particular representations are irreducible.a.) By applying a second Lorentz transformation X. (J .(A 1)'I'i'(Ax + a) . However. n)at (p. (5.(A I )'I'!(Ax + a) . in the sense that it is not possible to by a choice of basis to reduce all D(A) to the same blockdiagonal fonn.1. and a host of tensor and spinor representations.. (5.(x.1.p.(x)U"I(A. as well as another that runs over the components of the individual irreducible representations. the vector D(A)I\ = N\. a.7) (We might.1.(x. the index t here includes a label that runs over the types or particle described and the irreducible representations in the different blocks.1.4) p v. so taking Al = (A)l and A2 = (A)I.. including the scalar D(A) = 1.5) with coefficients· u/(x. in principle. in general it is a blockdiagonal matrix with an arbitrary array of irreducible representations in the blocks. with two or more blocks. we find that D(A1)D(A.1.6) Uo(A.8) There are many such representations. (5. Later we will separate these fields into irreducible fields that each describe only a single particle species (and its antiparticle) and transform irreducibly under the Lorentz group. a)'I'i(x)U"I(A. it is always possible to choose the fields so that these matrices are the same.n) and vAx. p.a. • A reminder: the labels II and respectively_ run over all dilferent particle species and spin zcomponenls.192 5 Quantum Fields and Antiparticles annihilation fields lpt(x) and creation fields 1p{(x): 'l'i(x) ~L "" 'I'.n) .a.a)'I'.n) chosen so that under Lorentz transformations each field is multiplied with a positionindependent matrix: Uo(A. a.p. n).
(We have used the unitarity of the rotation matrices D~G) to put both Eqs. (4.1. b) ~ exp ( .)(W'(A. (5. (5.1.""(x) ~"" L.UI) LD~. what shall we take as the coefficient functions Ut(x.a.1. a.~.N (x)·'I'/(X) 1 1 ''1'/ (x) M (5. b) J(Ap)O Ipo (S.J NMt'·. as we saw in Section 2.p»)at(PA. because we regard the derivatives of components of these fields as just additional sorts of field components. n)Uo'(A.1.7).. in the sense that for all A: (5.11) and (5. Later we will be able to combine creation and annihilation fields so that this density also commutes with itself at spacelike separations.12) and its adjoint give the transformation rules" for the annihilation and creation operators Uo(A.w that satisfy Eq.2) if the constant coefficients gl'lt~" II '·(14 are chosen to be Lorentz covariant.1.10) (Note that we do not include derivatives here. where in is the spin of particles of species n. .5 the volume element d3 pj pO is Lorentzinvariant.n).p.·t't 1 ·1 M 1 N x ~.12) in the form shown here.5.10) is no different in principle (and no\ much more difficult in practice) than that of using ClebschGordan coefficients to couple together various representations of the threedimensional rotation group to form rotational scalars.1. Now...a.n)? Eq.1.i(Ap) . n)Uo 1(A.6) and (5. we will be able to construct the interaction density as .~.(X). t'~" {I(.J L. so we •• This is for massive pariides.a.a.9..2.) Also.J " g t 'I{ 'N'· I I>f ("t L. The case of zero mass will be taken up III Section 5.9) and this will be a scalar in the sense ofEq.p.) The task of finding coefficients g(' .I'(Wl(A.n) .1 Free Fields 193 Once we have learned how to construct fields satisfying the Lorentz transformation rules (5. b) J(Ap)OIpo (5. Uo(A.b) X ~ exp (i(Ap) . b)at(p.Ul) x L D~.p»)a(PA.n) and w(x. .b)a(p. (5. and PA is the threev~ctor part of Ap.1.
h) ~~ ""V J d3(Ap) ur(x.(Ax +h.)D).p..n) We see that in order for the fields to satisfy the Lorentz transformation rules (5. o)D). it is necessary and sufficient that ~ Dtl(A I )u.1. p)) / po j(Ap)O a(PA.194 5 Quantum Fields and Antiparticles can replace d3 p in Eqs.5) with d3(Ap)pOj(Ap)o.(Ax + h.n). a. ".b)'I'i(x)UOI(A.1. we find Uo(A..6) and (5.a.p))~ VPO/(Ap)O " and x ~Dl. (5.n) t (5. .p.14) These are the fundamental requirements that will allow us to calculate the Ut and Ut coefficient functions in terms of a finite number of free parameters. n) and Uo(A.4) and (5.)' ( W1(A. n) ~ t V / (Ap)O po x ~D~..i(Ap)' b)Vt(x.a.p.p.a.(A)exp (i(Ap). J ""V d3(Ap) Vt(x.i(Ap) ..a.n) .1. n) ~ VPO / (Ap)O t x ~DW'(WI(A. or somewhat more conveniently ~u(Ax+h. b)DW ( WI(A.)' (W(A. PA. t (5. h)D~.p))exp( + i(Ap)'b)ut(x.p)) exp( i(Ap)' b)vt(x.13) ~ v.)(WI(A.a.a.n) x exp (i(Ap) .7).)(W(A.t(x)U01(A. and ~ Dt7(A I )v?(Ax + h. ".p.1.b)V. p)) ~ VPO/ (Ap)O " x ~Di. a. PA. PA. p)) / po j(Ap)O at (PA. n) .1. b)ut(x.1.b) ~~ x exp( .n) . Putting this all together.(A)exp ( .p.PA.
1.)0 and :>.n)eiP'Xa(p.14) are satisfied if and only if ~ uI(PA.a.a.) Using Eqs.5.1.e. (5.16).n) . n) (:.1.1.1.L l(Ap)AL(p) ~ L. (5.n).l'(W(A.1.1.16) so the fields are Fourier transforms: 'Pt(x) = i)2n)3/2! d3p Ut(p.14) in three steps.P)) (5. n) t (5.15) (5.O".p.n)DY.1.n) and vr(x.1. (5.) (W(A.191 ~ V(:.1.1.1. (5.17) and lPr(X) = i)2n)3/2 J d3p vr(p.n)eipxat(p.".".O".15) and (5.19) and (5. We see immediately that Ut(x. e. (T. ". Boosts Next take p = 0 in Eqs. nIDY.1(q)L(q) ~ 1 .1.20). Then L(p) = 1.a.1.)O ~Dir(A)V(p." (The factors (2n)3/2 could be absorbed into the definition of ur and Vr. vr(x.1 Free Fields 195 We will use Eqs.a. and let A be the standard boost L(q) that takes a particle of mass m from rest to some fourmomentum qll. (5. Hence in this special case.a.1.1.20) for arbitrary homogeneous Lorentz transformations A. (5.a. but it is conventional to show them explicitly in these Fourier integrals.19) and (5. (5.13) and (5.n).1.p.p.p) . we see that Eqs.n) must take the form Ut(x. considering in turn three different types of proper orthochronous Lorentz transformation: Translations First we consider Eqs.18) ".n) = (2n)3/2 eiP'X ut (p.n). and W(A.20) give Ut( q.1.13) and (5.". n) ~ (m/ qO)l!2 I: Drf(L(q)) Ut(O.O".n • (5.n) = (2n)3/2eiPXvt(p.13) and (5. (5.p.1.:"I(PA.n). (5. p)) ~ V ~ DrAA)ur(p. Eqs.a.14) with A = 1 and h arbitrary.21) .
. with the coefficients Ut(O.6. 0".) = LJltUt(O. n) .(O.1..n) (5. respectively. (5.(L(q)) V.)(R) ~ L ~ D/. then for a given representation D(A) of the homogeneous Lorentz group. a. n) .n). as a rotation R...n)DY.5 that each irreducible representation of the proper orthochronous Lorentz group contains any given irreducible representation of the rotation group at most once.26) tell us that if the field lpi(x) is to describe particles of some particular spin j. We shall see in Section 5.24) or equivalently LUl(O.1. (Explicit formulas for the matrices Dtt(L(q)) will be given for arbitrary representations of the homogeneous Lorentz group in Section 5. Here obviously W(A.n) and vr(p.22) In other words. then this representation D(R) must contain among its irreducible components the spinj representation DUJ(R). n.23) ( and L vi(O.nlJ¥.20) read Lu/(O. 6.6.n) and Vt(O. and so Eqs. More generally. the number of free .26) L"7(0. .25) and (5.a.. nl ~ (m/ qO)I/2 L D.1. take p = 0.1.6.6. t where JUl and J are the angularmomentum matrices in the representations DU)(R) and D(R)..19) and (5. be a Lorentz transformation with PA = 0. 6.n)JV.a.(R)ut(0. n) for zero momentum. (5.. but this time let i\.n) (5. if we know the quantities ut{O.1.a.Lhrvt(0.7. (5.) Rotations Next. that is.1. n) and vitO. (5.n) simply describing how the spinj representation of the rotation group is embedded in D(R). take i\.1.)' (R) ~ L " t D7t(R)vr(0. then they are unique up to overall scale. n) D¥.1.p) = R.'· ~ . Any representation D(A) of the homogeneous Lorentz group obviously yields a representation of the rotation group when A is restricted to rotations R.1. so that if the fields lpt(x) and tp7(x) transform irreducibly.25) " and r (5. we know the functions u(p. Eqs. .O".n) for all p.196 and 5 Quantum Fields and Antiparticles v/(q.
: t. (5. n) and vt{o. and the cluster decomposition principle requires that the coefficient 1' NM may be written as in Eq.: L. t For general functions the indices (' and t may have to run over an infinite range.1.17) and (5.1.19) and (5.O". L. ( 'PI 0"1 "l .27) . . d3p'l . PMO"MI1M) = (2n)33N/23M/2 x L. . di:.0". Inserting Eqs.1. (5. (5.tThe cluster decomposition principle together with Lorentz invariance thus makes it natural that the interaction density should be constructed out of the annihilation al1d creation fields.. In fact. ) = b 3(p'l X '" 1'NM(PIO"lnl"'. L.1. (5..28) as the product of a single momentumMconservation delta function times a smooth coefficient function.O".1.1.27).1..n) and v('(p..1.5.29).1. dJp. (5. t~.1.~l(PMO"MnM)' This interaction is manifestly of the form that will guarantee that the SMmatrix satisfies the cluster decomposition principle: 1'NM has a single delta function factor. . (5.. This is left as an exercise for the reader. t.~1 gt'l·..(PN O"N n~) (5.22)...... with Ut(O.1.28) where ~NM(p~ 0"1 nl'" P~ O"~ n~ ."'a~al"'aMn... PIO"Inl . The reasons for re:.~l Vt'l (P'I 0"1 nl) .. . we could turn this argument around.9) and integrating over x. pMO"MnM ) 11 " "N Y NM with coefficient functions given by 1' N M(Pt 0"1 nl . ..  + .y NM that (at least for a finite number of field types) has at most branch point singularities at zero particle momenta.tricling (' and t ' to a finite range have to do With the principle of renormalizability..  P I .p]O"ll1l''').18) in Eq. with a coefficient ..'n~nl"'nM X at (pi] 0"1 nlJ .' d3pl" d]PM L.(~ t J" t. Vt~.... n) satisfying Eqs.1 Free Fields 197 parameters in the annihilation or creation fields (including their overall scales) is equal to the number of irreducible representations in the field. L.. It is straightforward to show that coefficient functions u('(p. Let us now return to the cluster decomposition principle.n) given by Eqs.0". ..23) and (5.. a(PI SI nd (5. will automatically satisfy the more general requirements (5..21) and (5. (5.29) x U('l(PIO"tn])··· ut.1. P10"Ini ..1.24). PN O"N. p] 0"] nl .20). any operator can be written as in Eq.1. Any sufficiently smooth function (but not one containing additional delta functions) can be expressed as in Eq. the interaction Hamiltonian is V ~ L. ) (5.cussed in Chapter 12.h . at (p~ 0":'" n:"') a(PMO"MnM) . NM X J a..1.
for then the interaction could not be Hermitian.1.1. (5.'f(x).).y is spacelike then no signal can reach y from x. The only way out of this difficulty is to combine annihilation and creation fields in linear combinations: (5.yj (5.) and in general this does not vanish even for x .31) [~t(x).32) is often described as a causality condition. for the Lorentz iovariance of the Smatrix it is necessary also that the interaction density satisfy the commutation condition (5.y spacelike [1pr(x). or tp~.1. This condition is not satisfied [or arbitrary functions of the creation and annihilation fields because (with the sign + indicating a commutator or anticommutator .1.'.1.(y)].31) satisfying .130) with the constants K and Aand any other arbitrary constants in the fields adjusted so that for x . at point y. It is obviously not possible to avoid this problem by making the interaction density out of creation or annihilation fields alone.<p{'{y)]+ ~ [. with coupling coefficients g{'l"<~" {I"t. we will be dealing here with fields like the Dirac field of the electron that do not seem in any sense measurable.32) is needed for the Lorentz invariance of the S matrix.10) (and a suitable reality condition). (5. respectively. 2 However.198 5 Quantum Fields and Antiparticles If all we needed were to construct a scalar interaction density that satisfied the cluster decomposition principle.1.. ~ 0. ~ (2rr)3 I: "" J dJp U{(p. so that a measurement of tpt at point x should not be able to interfere with a measurement of lpt. Such considerations of causality are plausible for the electromagnetic field. (5. a.f the particles destroyed and created by the components lPt and lJ'i are bosons or fermions..1.. because if x .y spacelike. anyone of whose components may be measured at a given spacetime point.. However. n)e'P.132) We will see in subsequent sections of this chapter how to do this for various irreducibly transforming fields. (By including explicit constants K and Ain Eq.3) if it is constructed out of such fields and their adjoints.9). . The point of view taken here is that Eq. then we could combine annihilation and creation operators in arbitrary polynomials (5. There is an obstacle to the construction of fields (5.f subject only to the invariance condition (5.3). without any ancillary assumptions about measurability or causality.(p. The condition (5. v'i(Y)].1. as shown in a classic paper of Bohr and Rosenfeld. n)v.) The Hamiltonian density Jf'(x) will satisfy the commutation condition (5. with an even number of any field components that destroy and create fennions.1.31) we are leaving ourselves free to choose the overall scale of the annihilation and creation fields in any way that seems convenient.
Now.6) and Vt(p.. (5. = O. we shall have to consider many different such fields.n)] ~ q(n)a(p. [Q. such that fields that belong to different blocks do not transform into each other under Lorentz transformations. In order that ff(x) should commute with the charge operator Q (or some other symmetry generator) it is necessary that it be formed out of fields that have simple commutation relations with Q: (5. Also. we shall from now on restrict our attention to fields that destroy only a single type of particle (dropping the label n) and create only the corresponding antiparticle. then the same component of the creation field must create particles of a species ti. If the representation D(A) is not irreducible. if particles of species n carry a value q(n) for the electric charge Q.n). there must be a doubling of particle species carrying non~ zero values of such quantum numbers: if a particular component of the annihilation field destroys a particle of species n. then we can adopt a basis for the fields in which D(A) breaks up into blocks along the main diagonal.a.. such that qfl + qt2 + ..a(p. In the following sections we are going to finish the determination of the coefficient functions Ur(p. 0").. known as the antiparticles of the particles of species n.1.5.. and that transform irreducibly under the Lorentz group (which as mentioned above mayor may not be supposed to include space inversion).1 Free Fields 199 (5. some perhaps formed as the derivatives of other fields. This is the reason lor antiparticles. We see that in order for such a theory to conserve quantum numbers like electric charge.33) is satisfied for one particular component lpt(X) of the annihilation field if and only if all particle species n that are destroyed by the field carry the same charge q(n) = qr. which have opposite values of all conserved quantum numbers. in general.n).1.33) for then we can make Jf'(x) commute with Q by constructing it as a sum of products of fields 'Prj 'Pt 2 ••• and adjoints 'PlllP~2 ..a.l. with the understanding that. and it is satisfied for one particular component vJ7(x) of the creation field if and only if all particle species n that are created by the field carry the charge q(n) = qr.  qml  qm2  .a. For instance.n)] ~ +q(n)at(p. It may be that the particles that are destroyed and created by these fields carry non~zero values of one or more conserved quantum numbers like the electric charge.at(p. . Eq. induding many irreducible components and many particle species.32).a. Lorentz transformations have no effect on the particle species. instead of considering one big field. Therefore. then [Q.
After that we will repeat the analysis for a completely general irreducible representation. and Dirac spinae representations. .18) shows that all the components of a field of definite mass rn satisfy the KleinGordon equation: (0 . (5. Rewriting : §l : in this way makes it obvious that despite the normal ordering. • •• A technicality must be mentioned here. we start with the particles. (5. with the field equations arising almost incidentally as a byproduct of this construction.1.1. and derive the fields according to the dictates of Lorentz invarianee.m')'J'((x) ~ o.4. (5. but including minus signs for permutations of fermionic operators) so that all creation operators stand to the left of all annihilation operators. the condition that guarantees that a theory will satisfy the cluster decomposition principle is that the interaction can be expressed as a sum of products of creation and annihilation operators. A word about field equations. and with coefficients that contain only a single momentumconservation delta function. and (5.1. vector. with all creation operators to the left of all annihilation operators. and deduce the relations between the properties of particles and antiparticles first for fields that belong to the simplest irreducible representations of the Lorentz group. or with the Lagrangian from which they are derived.34) Some fields satisfy other field equations as well.17). According to the theorem proved in Section 4. (5. In the approach followed here.be written as a sum of ordinary products of the fields with cnumber coefficients.31).y is spacelike. : ff(<p(x).1. with even numbers of any fermionic field components. For this reason. and then uses them to derive the expansion of the fields in terms of oneparticle annihilation and creation operators. By using the commutation or anticommutation relations of the fields.32). the scalar.35) the colons indicating that the enclosed expression is to be rewritten (ignoring nonvanishing commutators or anticommutators. any such normalordered function of the fields can just as well.<pl(y» : when x . we should write the interaction in the 'normal ordered' form (5. if it is constructed out of fields that satisfy Eq.<pt(x») : will commute with: ff(<p(y).200 5 Quantum Fields and Antiparticles fix the relative values of the constants K and A.1.1. depending on whether or not there are more field components than independent particle states. Traditionally in quantum field theory one begins with such field equations. Inspection of Eqs.
4) A Hamiltonian density £(x) that is formed as a polynomial in c/>+(x) and c/>(x) will automatically satisfy the requirement (5. Restricted to rotations.25) and (5. (5. so ¢+(x) either commutes or anticommutes with c/>+(y) for all x and y. It remains to satisfy the other condition for the Lorentz invariance of the Smatrix. with D(A) = 1.26) have no solutions except for j = 0.17) and (5. and ¢+(x) does not commute or anticommute with ¢(y) for general spacelike separations.1. the scalar. . </>+(x) ~ J dlp (2n)l/2(2 p o)1/2 a(p)e iP > (5.1. (5.*'(y) for all x and y. with no distinct antiparticle (and dropping the species label n as well as the spin label cr and the field label f). in the scalar case.1. for which J = 0. Using the commutation (for bosons) or anticommutation (for fennions) relations . It is conventional to adjust the overall scales of the annihilation and creation fields so that these constants both have the values (2m)lj2. (5. according to whether the particle is a boson or fermion.22) then give simply u(p) ~ (2 pO)1/2 (5. of course.1) and v(p) ~ (2p')1/2 .1. any such even polynomial) will commute with . Eqs. (5.3) and q.1. All annihilation operators commute or anticommute. cr take only the value zero. in order to be Hermitian.2.9). Thus a scalar field can only describe particles of zero spin. so Eqs.2 Causal Scalar Fields 201 5.1. If .*'(x) must involve c/>+t(x) = c/>(x) as well as c/>+(x). this is just the scalar representation of the rotation group. there would be no problem.5) Hence any £(x) formed as a polynomial in ¢+(x) (or.21) and (5. is that.2) The fields (5.2.2. the quantities ut(Ocrn) and vt(Ocrn) are here just the numbers u(O) and v(O).2 Causal Scalar Fields We first consider onecomponent annihilation and creation fields c/>+(x) and c/>(x) that transform as the simplest of all representations of the Lorentz group.*'(x) were a polynomial in ¢+(x) alone. in which case cr.2.2.5.(x) ~ J dlp (2n)l/2(2po)1/2at(p)eiP> ~ </>+I(x).*'(x) commute with Jf(y) at spacelike separations x .1. The problem. that it transform as a scalar. respectively: (5. Assuming also for the moment that the field describes only a single species of particle.y.18) are then. for fermions. that .
. (5.2. we have then for x . (5.p ') ....Y ["'+( x.7) then gives d+(X) = 1 (2rr)3 J Jo G 3 2j p2 + m' p 2ap dp eip ' x 4rr .. .'p.2 . Eq. suppose we try to construct *(x) out of a linear combination Using Eq. Ixl ~ j.8) Changing the variable of integration to 1I+(x) ~ plm.. </>+(y)].7) depend only on the invariant square x 2 > O.5).2 . </>(y)l+ ~ 1)·1' W(x). this is m fooo udu 0 Ju 2 + 1 or. </>(y)]+ + ~ (IKI' + 1. </>1 (y)h ~ IKI'W(x).. G sm(m"x 2 u) (5.~ 'I' J P P ip·x iP'Y63( (2rr)3(2 pO. for x 2 > 0 it is even in x/l.) KA(1 + 1)1I+(x  y).2) (5.6). so what are we to do with it? Note that even though d+(X) is not zero. 3 . a3 a3 I which collapses to the single integral [</>+(x)"P(Y)!cc ~ 1I+(x . ( )1 Ll+ X = (2n)3 J 2po d p .2.(W(X).2. </>+(Y)h + W(x).2. in terms of a standard Hankel function. (5.y spacelike [</>(x). (5.2.6) where d+ is a standard function: .2) 2jp2 + U m' pj.2.(2rr)3 rx sin(pj.9) This isn't zero.y). we have 5 Quantum Fields and Antiparticle8 ) '1'.202 (4. '" ()] .<1 2)1I+(x . We can thus evaluate d+(X) for spacelike x by choosing the coordinate system so that xO This is manifestly Lorentzinvariant. 4n 2 "x 2 1I+(x) ~ 4n2 x 2 P K1 (mj.y) [</>(x). Instead of using only ¢+(x). and therefore for spacelike x it can ~ 0.2.. 21"0)1/2 e e p . </>(y)]+ ~ K).
Taking 0. we have then (5. But this is impossible i[ £'(x) is formed as a polynomial in ¢(x) = ¢+(x) + ¢+t(x). This ¢. Even though the choice of the relative phase of the two terms in Eq. respectively. an arbitrary phase. like q. at(p) _ eiClat(p).2..r 5.~ q4>+(x) .2.. it is the top sign that applies) and K and }.Jt'(x) shonld commute with the charge operator Q (or some other symmetry generator) it is necessary that it be formed out of fields that have simple commutation relations with Q...8).+(x) + eirtq. but charges +q and q.y if formed as a normalordered polynomial in the selfadjoint scalar field 4>(x). but also another scalar field for the same particle ¢(x) = eilXq. but ¢(x) would not commute with q.. [Q.y is spacelike. suppose that the interaction density involved not only the field (5. it is a convention that once adopted must be used wherever a scalar field for this particle appears in the interaction Hamiltonian density. = !Arg(J. (5.(x) carry some conserved quantum number like electric charge.+t(x) with fJ. Redefining 4>(x) to absorb the overall factor K = A... 4>+t(x)]_ ~ +q4>+t(x). by redefining the phases of the states so that a(p) ) eitl:a(p) .e.2 Causal Scalar Fields 203 Both of these will vanish if and only if the particle is a boson (i. and hence K _ irx iCl Ke .(y) at spacelike separations. and therefore we cannot have both of these fields appearing in the same theory. we must suppose that there are two spinless bosons. are equal in magnitude We can change the relative phase of K and ). 4>+(x)].+(x) and its adjoint." would be causal in the sense that ¢(x) commutes with ¢(y) when x . For instance. for which [Q. we can in this way make K and A. and hence equal. This is true for q. If the particles that are destroyed and created by q. then Jf(x) will conserve the quantum number if and only if each term in Jf(x) contains equal numbers of operators a(p) and a(p)t.10) is a matter of convention. In order to deal with this problem. A _ }. but not for the selfadjoint field (5. in order that . To put this another way.2.2./K).e.10) The interaction density Jf'(x) will commute with £(y) at spacelike separations x . equal in phase. with the same mass m. Let ¢+(x) and ¢+C(x) denote the annihilation fields [or these .10).
In deriving this result. = 0. By redefining the relative phase of states of these two particles. we have tacitly assumed that the particle and antiparticle have the same mass.pt(y)]'F ~ IKI'[.y). and for particles with distinct antiparticles (for which aC(p) =F a(p)). because it is not possible that ¢(x) should anticommute with ¢t(y) at spacelike separations unless K = . it is necessary and sufficient that IKI 2 = l.p+(x)]_ ~ q. So a spinless particle must be a boson. [Q.. so that .11) This is the essentially unique causal scalar field. with u"(p) = u(p).pix) ~ .pHt(X).p(x)]_ ~ q. so that the commutators or anticommutators involve the same function d+(x . This common factor can again be eliminated by a redefinition of the field ¢...p+<1(x) or in more detail . in which case K = A.p(x).204 5 Quantum Fields and Antiparticles two particles. so that[Q.p(x) ~ J (2n)3/~.I'[. .. . as well as for the particle and antiparticle to have the same mass. This formula can be used both for purely neutral spinless particles that are their own antiparticles (in which case we take aC(p) = a(p))..p+(x).p(x).p+(x) + . we can again give K and ).2.p(x) ~ K.p+t(y)] + 1!.p+(x). in which case the fields simply vanish..p+'(x). the same phase.~pO)II' [a(p)e"" + act(p)e"'] .. (5.p+'(x)]_ ~ +q. in order that a complex ljI(x) shou!d commute with ljIt(y) at spacelike separations. For Bose statistics.p+'(y)]'F ~ (IKI 2 f IAI 2)1I+(x . at spacelike separation [.. Fermi statistics is again ruled out here. Define ¢(x) as the linear combination .p+'t(x). • The label 'c' denotes 'charge conjugate'. while ¢(x) and ¢(y) automatically commute or anticommute with each other for all x and y because ¢+ and ¢+ct destroy and create different particles.y). The commutator or anticommutator of ¢(x) with its adjoint is then. It should be kept in mind that a particle that carries no conserved quantum numbers mayor may not be its own antiparticle.lf.p+(X) + !. which manifestly has the same commutator with Q as ¢+(x) alone [Q.
pt(y)J ~ A(x .2. The only way to preserve Lorentz invariance as well as parity conservation and the hermiticity of the interaction is to require that rjJp be proportional to rjJ. the intrinsic parity '1'1 c of a state containing a spinless particle and its antiparticle is even. and T. We have now simply (5. and changing the variable of integration from p to p.d+(y .19) •• We are. because io virtually all cases where these inversions are good symmetries.2.17) where as before &'x = (x. the same operators induce inversion transformations on 'in' and 'out' states and on freeparticle slates.2.I ~ ~'a(p). First. but if rjJ and rjJ~ appear in the same interaction then we are in trouble.15) where tI and TIc are the intrinsic parities of the particle and antiparticle. (5.2.2.y) =d+(X  y) .y). 2po(2n)3 (5.2.14) (5. We see that in general applying the space inversion to the scalar field rjJ(x) = rjJ+(x) + rjJ+ct(x) would give a different field rjJp = r(rjJ+. we see that (5.2.4). Both fields are separately causal.18) That is.16) (5. we note here that the commutator of the complex scalar field with its adjoint is [.+ '1 crjJ+c t .x) = J 3 d p [eiP(x..13) Let's now consider the effect of the various inversion symmetries on this field. because in general they do not commute at spacelike separations.2. C. Applying these results to the annihilation field (5.3) and the charge·conjugate of the creation field (5. . from the results of Section 4. omitting the subscript 0 on inversion operators P.xo). and hence that (5.2.2. respectively..5. we can readily see that the effect of the spaceinversion operator on the annihilation and creation operators is :•• Pa(p)p.2.p(x).2 Causal Scalar Fields 205 For future use. (5.y) _ eiP·(xy)j.12) where d(X .
2. (5.p(9x).2. From the results of Section 4.25) Again. where ~c = ¢.31) .23) In order that C¢(x)C. we have Ca(p)c' ~ (' a'(p) .pHl(x)c' ~ '°.24) Just as for ordinary parity.26) (5.28) (5.p+ol(_9x). (5. the intrinsic chargeconjugation parity ~~c of a state consisting of a spinless particle and its antiparticle is even. (5.p+(x)r' ~ ('. we must have (' and then ~ (' (5. It follows then that e C. (5.2. for which I{ = Il.2.2.20) (5.p+(9x) T.pH(X) .p+'(x).1 should be proportional to the field t. we find that T.p+(x)C' ~ (' .2.p(x)r' ~ ('. We now have simply (5.206 5 Quantum Fields and Antiparticles These results also apply when the spinless particle is its own antiparticle.27) Recalling that T is antiunitary.2.p+ol(x)r' ~ (0.tJt(x) with which it commutes at spacelike separations.2. and again changing the variable of integration from p to p.2 we have Ta(p)r' ~ ('a(p). Finally we come to timereversal.21) where and (c are the phases associated with the operation of chargeconjugation on oneparticle states.2.2. (5.30) T.2.29) In order for T¢(x)Tl to be simply related to the field ¢ at the timereversed point #x. it is evidently necessary that (5.22) C. In this case the charge~conjugation parity like the ordinary parity must be real. these results apply also in the case where the particle is its own antiparticle.2. and imply that the intrinsic parity of such a particle is real: 7/ = + 1.2. From Section 4. (5. Chargeconjugation can be handled in much the same way. ~ = + 1.
(5.o) ~ (m/p')I/2L(p)'.5. For the moment we will suppose that only one species of particle is described by this field (dropping the species label n).$'.f''"VV(O.0) and (5.3.. 0) ~ (m/pO)I/2L(p)". etc.) Also. 0). then we shall consider the possibility that the field describes both a particle and a distinct antiparticle.v'(0.:)21t)3/2 / d3p uJJ(p.3) v"(p.1.1) The annihilation and creation parts of the vector field are written: ¢+It(X) = :~. v. the simplest nontrivial representation of the homogeneous Lorentz group. the coefficient functions at zero momentum aTe subject to the conditions (5.2) "''(x) The coefficient functions uJ1 (p.3."'(0.25) and (5.1. (5. rr)JV.tpx .3. lT)a(p. .1.3. with (5. etc. In the four~vector representation of the Lorentz group. so this example is not merely of pedagogical interest.1."'(O. IT) for arbitrary momentum are given in terms of those for zero momentum by Eqs.3 Causal Vector Fields 207 5.5) ~ 2)2")3/2 " J dlp v'(p.a) and vl'(p.26): I: "'(0.o)e. which here read: "'(p. (5.3.o).4) (5. (Also. (5. the W± and Zo. that at low energies are described by such fields and that play an increasing role in modern elementary particle physics.22).7) . ~ . (We are using the usual summation convention for spacetime indices 11.o)a'(p.IT). although we are here considering only massive particles.6) " . (5. which transforms as a fourvector. the rows and columns of the representation matrices D(A) are labelled with fourcomponent indices 11.3 Causal Vector Fields We now take up the next simplest kind of field. one approach to quantum electrodynamics is to describe the photon in terms of a massive vector field in the limit of very small mass).L vJl(O.21) and (5. v.3. There are massive particles.ij)JV~· = .3.a)eiPX ff .
)oo ~ (f')o.3. i1)(J(j)·)~lT = 0.jk .16) . we can take the only nonvanishing component of uJl(O) and vi'(O) to have the conventional values: . Let us look in a little more detail at each of these two possibilities. 2. (1) (5. (5.9) with i. Clki j ~ i'.3. (5.5) yield for general momenta .11) )i j = 2t5 i j ' From Eqs.3.4) and (5. and 3.15) we see that there are just two possibilities for the spin of the particle described by the vector field: either j = 0.3. and 2ui (O. (The label (T here takes only the single value zero.3.$2 takes the form C?.10) (5. or else j = 1 (so that j(j + 1) = 2).14) (5. ~ (f'/ ~ 0.3.13) 2:>°(0.3.3. . (/ 2 ° (5.?Il ~ o in the four~vector representation are given 0 i CIk) °~ (h) .7) it follows then that LUO(O. i1)(J(j)~lT = . .3. We note in particular that .) Then Eqs. we recall the familiar result that (J(j)~<T = j(j + 1)t5 iTo • From Eqs.3. (5. for which at p = 0 only the spacecomponents u i and vi are nonzero.3.12) L ui(O. Spin Zero By an appropriate choice of normalization of the fields.3. (5.3.208 5 Quantum Fields and Antiparticles The rotation generators by Eq.6)(J(j})~<1 = 0.3.j.12){5. (5.15) Also. and is therefore dropped."(p) ~ ip"(2P')1/2 (5. (5.3.1) as .3.6) and (5. for which at p = 0 only uO and vO are nonzero. ~ CIk) °~ 0.k here running over the values 1.0(0) ~ i(m/2)1/2 vO(O) ~ i(m/2)1/' .8) (5. (5.
O) and vi(O.20) the other components.+I) ~ }z(2m)1/2 Applying Eqs.3.3 Causal Vector Fields and 209 (5..6).9) to calculate the effect of the raising and lowering operators JpJ + lJ on u and v. 0) ~ (2m)1/2 (5.I) ~ v'(O.) (5.3. 2.. Spin One From Eqs.5.7) we see immediately that the vectors ui(O.3.6) and (5.3.(p. we can take these vectors to have the values ° ° 1 ° with fourvector components listed always in the order 1..3.3. 3.19) Hence we need not explore this case any further here. (5.3. By a suitable normalization of the fields.22) u'(p.3.+1) ~ v'(O.3.3. To find u'(O.23) where e"(p. we use Eqs.24) . (5.) ~ v"'(p.21) u'(O.(p)e'(O.3. and (5.) ~ (2pO)1/2 e. This gives: iJ1 1 u'(O.3.O) for (J = 0 are in the 3direction.18) It is obvious that the causal vector field for a spinless particle is also just the derivative of the causal scalar field: (5..3.. (5.3.3.) L".I) ~  }z(2m)1/2 +i ° ° 1 I (5. O.7). (5.17) The vector annihilation and creation fields here are nothing but the derivatives of the scalar annihilation and creation fields ¢± for a spinless particle that were defined in the previous section: (5.) (5.O) ~ v'(O.4) and (5.5) now yields ° ° (5.
3. + (2n)J2 pO (5.27) may then be written in terms of the d+ function defined in the previous section. we [arm a linear combination of annihilation and creation fields for which.3./2 +i o o .3.b+lt(X) rpV(y)].210 5 Quantum Fields and Antiparticles with e"(O..= / J3 p eip(xy) Illt\'(p) .24) then shows that IlW(p) is the projection matrix on the space orthogonal to the [ourvector pI': Ill"'(p) = 'Ill\' + pJlpl' 1m2 .are (5.29) The commutator (or anticommutator) (5.25) shows that IlIl"(O) is the projection matrix on the space orthogonal to the timedirection.l1)e"·(p.y spacelike.330) For our present purposes.y spacelike it does not vanish and is even in x . the important thing about this expression is that for x .3. (5. for x .28) A straightforward calculation using Eq. but rjJ+I'(X) and ¢V(y) do not.25) The annihilation and creation fields (5.27) where IT/H'(p) Lel'(p. e"(O. u (5.3. Their commutator (for bosons) or anticommutator (for fermions) is [c. 1 o o (5.3.y.3.3.26) The fields ¢+I'(X) and ¢+V(y) of course commute (or anticommute) for all x and y. .l)~ )2 1 .3. We can therefore repeat the reasoning of the previous section in seeking to construct a causal field. as (5.3. and Eq.u).+I) ~~ 1 .2) and (5. (5. (5. 0) ~ o o o 1 1 e"(O.3) here.
2. spin one particle that is its own antiparticle. This again is a causal field. Instead.y). First.3. and then drop the common factor K by redefining the overall normalization of the field. we must suppose that there is another boson of the same mass and spin which carries an opposite value of Q. by simply setting aC(p) = a(p).13).y..y) is the function (5.31) We note that this is real: (5. the field satisfies the KleinGordon equation: (5.5.33) or in more detail vP(x) ~ (2n)3/2 L x [e1t(p. (5. (5.P(x) ~ <jJ+P(x) + <jJ+pt(x).34) where the superscript c indicates operators that create the antiparticle that is chargeconjugate to the particle annihilated by ¢+It(X). since p/l in the exponential in Eq. The real and complex fields we have constructed for a massive. (T )act (p.3.26) satisfies p2 = _m 2 .3. (5. if the particles it describes carry a nonzero value of some conserved quantum number Q. but no longer real. the commutator of a vector field with its adjoint is [vP(x).35) where il(x . it is necessary and sufficient that the spin one particles be bosons and that IKI = 111.3.3 Causal Vector Fields and 211 In order for both to vanish for spacelike x.t(y)] ~ [ryp.3.v. We can also use this formula for the case of a purely neutral.  a:n A(x . then we cannot construct an interaction that conserves Q out of such a field. (5. spin one particle satisfy interesting field equations.3.32) However.36) . (T )e " JM X iP p  d3 + eJl* (p.3. After all this. In either case. we find that the causal vector field [or a massive particle of spin one is . (T )a(p. and construct the causal field as (5. By a suitable choice of phase of the oneparticle states we can give K and A the same phase. (T)e iPX] . so that K = }.
In addition. cause the emission rate to blow up when m + O.3.3. Using Vv( p) = !!fJ1' pLP r(p)!!fJr v and Eg. The inversions can be dealt with in much the same way as for the scalar field discussed in the previous section.'(p.3.38) In the limit ofsmall mass. is an arbitrary [our·vector current. at x = 0) between the initial and final states of all other particles. the need for conservation of the current can be seen by simply counting states.2.(1). and 1.37) we now have another equation 8. (5. The term pl'pV 1m2 in Ill'V(p) will. (5. Using el' . Ill'\'(p).39) Also.24) shows that e'(p.v'(x) ~ O. The trouble can be seen by considering the rate of production of a spin one particle by an interaction density J't = Jp. in the sense that 0l'JI' = O. (1). where J p. (5. (5. which in coordinate space is just the statement that the current JI' must be conserved. 0")"1 2 =< JI' >< J >. since Eq.36) and (5. However.3. Egs. Indeed. L Il v where p is the momentum of the emitted spin one particle.u) ~ iJ"/(p.3. we have e"(p. we find (_1jl+"e"'(_p.3. we cannot obtain electrodynamics from just any theory of massive spin one particles by letting the mass go to zero. A massive spin one particle has three spin states. 0. in general.u). . spin one particle like the photon can only have helicities + 1 and 1: the current conservation condition just ensures that the helicity zero states of the spin one particle are not emitted in the limit of zero mass. (5.212 5 Quantum Fields and Antiparticles just as for the scalar field. it is straightforward to .3. while any massless. Squaring the matrix element and summing over the spin zcomponents of the spin one particle gives a rate proportional to I < J > eP(p. (5. The only way to avert this catastrophe is to suppose that < J I' > pI' vanishes._u) ~ iJ". (5.VIl.u).40) Using these results and the transformation properties of the annihilation and creation operators given in Section 4. and < JI' > is the matrix element of the current (say. which can be taken as the states with helicity + 1.(p).38) are just the equations for the potential fourvector of electrodynamics in what is called Lorentz gauge. we need a formula for el'( p. to evaluate the effect of timereversal we need a formula for (_l)I+<1 el'*(p. To evaluate the effect of space inversion.3.24).u) P" ~ 0.((1) = el'((1) and the above formula for £I\.
From the point of view we are following here. (5. of their covering groups see Section 2. the minus sign in Eq. As we saw in Section 1.7) in any number of dimensions.4 The Dirac Formalism 213 work out the inversion transformation properties of the annihilation and creation fields. with no extra phases or signs accompanying the matrix {!}'Ji v . and timereversal phases for spin one particles and their antiparticles be related by 'Ie = 11· . Once again we find that.3. Cv"(x)C.3 but as so often happens it was already known to mathematicians. this representation was introduced into the theory of the electron by Dirac.v'(fY'x). it is necessary that the intrinsic space inversion.3.41) (5.46) Tv"(x)r ~ ('fY'". (5.4 The Dirac Formalism Among all the representations of the homogeneous Lorentz group. so it will be natural for us to describe the Dirac formalism as it first appeared in mathematics. we mean a set of matrices D(A) satisfying the group multiplication law D(A)D(A)  ~ D(AA) .3.45) (5.4 because it provides the basis of one of the two broad classes of representations of the rotation or Lorentz groups (actually. (5. there is one that plays a special role in physics.34) has the inversion transformation properties Pv"(X)pl ~ ~'fY'".3.3.3.44) (5. our causal vector field (5.43) (In particular all phases must be real if the spin one particle is its own antiparticle.3.) With these phase conditions satisfied. 5.42) (' ~ (' . in order for causal fields to be transformed into other fields with which they commute at spacelike separations. rather then as it was introduced by Dirac.3.5. charge~ conjugation. (5. ' In particular.44) means that a vector field that transforms as a polar vector.1.  .v'i&'x). the structure and properties of any quantum field are dictated by the representation of the homogeneous Lorentz group under which it transforms. describes a spin one particle with intrinsic parity 'I = ~1. By a representation of the homogeneous Lorentz group.1 ~ ('v"l(x) .
d'" (5.4. (5.v'1/lP (5. (5. suppose we first construct matrices that satisfy the anticommutation relations and tentatively define (5.7) can be summarized by saying that yP is a vector. we can study the properties of these matrices by considering the infinitesimal case.rl"/l /1'" + 'IavfPII. or a direct product of a spinor and a tensor.pur  '11'1' /vo .4.4. are irreducible. (5.9) .6).4. with an irreducible set of {liS.7) and from this we easily see that Eq.5) P' ~ ~[Y" [ / J1 v.6) does indeed satisfy the desired commutation relation Eq. (5. to show that yP] = i yl'I}"P + i}.6) It is elementary.12): i[J11I'. (5. that there is no proper subspace that is left invariant by all these matrices.4.4.4).214 5 Quantum Fields and Antiparticles Just as for the unitary operators U(A).5).4. (5.4. We shall further assume that the matrices rp. (5.6) that the most general irreducible representation of the Lorentz group is either a tensor. more accurately. (5.3) and (5. Any set of matrices satisfying a relation like Eq.4.1) (5.1/lV = . which would transform as in Eqs.4.4) I'll To find such a set of matrices.4. Otherwise we could choose some smaller set of field components.6).4.4. (5.2) for which D(A) ~ 1 + ~ w. the unit matrix is trivially a scalar D(A) 1 D1(A) ~ 1 (5.3) with . y').4. The commutation relation (5. that is.4. (5.4. with 1'f1l1' replaced with a Kronecker delta) is called a Clifford algebra. its covering group) arises from the fact (shown in Section 5.fp(TJ = 'I vp .5) (or its Euclidean analog./VII a set of matrices satisfying the commutation relations (2. using Eq.3) satisfies D(A)'yPD1(A) ~ AoP y" . (5. in the sense that Eq.4.4. The importance in mathematics of this particular representation of the homogeneous Lorentz group (or.8) In the same sense.4.3) and (5.4. or a spinor transfonning as in Eqs.
PIPCf . that no totaJly antisymmetric tensor can have more than four indices.4) shows that . respectively. terminates with the tensor (5.4. i'l'.4.4.4. so the totally antisymmetric tensors form a complete basis for the set of all matrices that can be constructed from the Dirac matrices.4.. (5.5. By repeated use of Eq.1. This formalism automatically contains a parity transfonnation.15) (5.416) Everything so far in this section applies in any number of spacetime dimensions and for any 'metric' l1J11" In four spacetime dimensions. so the sequence of tensors 1.4. each of these tensors transforms differently under Lorentz and/or parity transformations so .11) (5. (5. however.4.13) Applied to the Dirac matrices. then yields just a plus or minus sign. Furthermore. "PCf. according to whether the product contains an even or an odd number of ys with spacelike indices. For instance. pyOp1 ~ + yO. (5. (5.jPCf is an antisymmetric tensor D(A)/PCfD1(A) = A/AvCf"PI'.4.14) (We here label indices so that J1 runs over values 0. respectively. (5.) The same similarity transformation. In particular. conventionally taken as (5.4.. =yP'/y' _ ')iPY'yCf _ yCfyP y' + y'yPyCf + yCfy'yP _ Y'''(yf' .2. applied to any product of y·matrices. d()(fT. .11) is shorthand for .··· . this gives py.pl ~ i.4.5) we can write any product of 'Is as a sum of antisymmetrized products of ')is times a product of metric tensors. indicating that we are to sum over all permutations of the indices within the brackets. (5.10) The matrices yP can be used to construct other totally antisymmetric tensors (5.4 The Dirac Formalism 215 and Eq. with a plus or minus sign for even or odd permutations.12).12) The brackets here are a standard notation. Eq. there is a special feature.
lilIi2·'·J.216 5 Quantum Fields and Antiparticles they are all linearly independent.. In spaces or spacetimes with odd dimensionality. aud hence equal to ±L .d y.. for r = 0. the totally antisymmetric tensors of rank nand d .4. (More generally. because each component of each of these tensors is proportional to a product of different ymatrices.l.4.4. d ~ 1. We shall therefore take the yIJ. 1.n) t) There are at most v2 independent v x v matrices.JI6 = 4 rows and columns. yli'] oc (. This constraint does not interfere with the inclusion of space inversion in the Dirac representation of the Lorenlz group in odddimensional spacetime. six for . to be 4 x 4 matrices.2.?PrJ.?jJJ.l.rJ. and one for .) Returning now to four spacetime dimensions. Dirac matrices of the minimum dimensionality are necessarily irreducible. with the scalar product of two matrices defined by the trace of their product.. (The general rule is that a totally antisymmetric tensor with n indices in d dimensions has a number of independent components equal to the binomial coefficient d!/n !(d . we shall choose an explicit set of 4 x 4 ymatrices. one can form antisymmetric tensors with 0. Note that none of these matrices can vanish.n can be linearly related by the conditions" "y[P'yIi2 . so they must have at least . Under these conditions there are only z1~1 independent tensors. fOUf for . because here the tensor £~I~l~d is even under inversion of space coordinates.. " d indices.19) and (5. fOUf for yP. with ('JlIJ.. in any even number d of spacetime dimensions. . and the lefthand side taken as the unit matrix for r = O. If we don't care about space inversion.d totally antisymmetric. if reducible. requiring 'Imatrices of dimensionality at least 2(d~lJ!2. .· The number of linearly independent components of these tensors is one for 1.. which contain altogether a number of independent components equal to d I: 11=0 d! = 2d n!(dn)! ' so theymatrices must have at least 2d/ 2 rows and columns. . An example is provided by the submatrices in Eqs..17) where 1 is the unit 2 x 2 matrix..Vp".l. One very convenient choice is 'I 0"[01] 1 0 = 1 ' "I = I "[ ° . these matrices can be shown to be linearly independent by noting that they form an orthogonal set.20) below. we can also construct 2(dlJ/2_dimensional irreducible representations of the proper orthochronous Lorentz group in eveu spacetime dimensions by imposing the above condition relating antisymmetrized products of rand dr Dirac matrices.!. (5.1..+2 """ YlidI . and such a product has a square equal to plus or minus the product of the corresponding squares. or 16 independent components in aU. (5. the subspace left invariant by these matrices would furnish a representation of lower dimensionality."y lJ'r+l 1i. and the components of a are the usual • Alternatively.s/pur.
1.6): if .26) .22) (5.2. we find (5. we see that (5.2. It is convenient to write the totally antisymmetric tensors (5. with em ~ +1.4 The Dirac Formalism 217 Pauli matrices (Tl = (~ ~) .4.4. and the pseudoscalar 1'5.3 or 0. It is easy to see that the matrix 1'5 has unit square (5.u_! E'l]k ["' 2 .4.3.0 2 (Here E'ijk (Ii 0] . 7: equal in turn to 0. (T2 = (~ ~i) . Setting p.] ~ 0.3 or 1. we can easily calculate the Lorentz group generators (5. py. y. and therefore proportional to the pseudotensor E'par rr .4.4.4. and by setting p.25) The 16 independent 4 x 4 matrices can therefore be taken as the components of the scalar 1.12) in a somewhat simpler way.pI ~ Y5 . From Eq.21) where Y5 . (5.4. the antisymmetric tensor /pa. The matrix (5.~iyOyly2y3 . Similarly. 1.4.23) (5.) We note that these are blockdiagonal.12) is totally antisymmetric. U..4.10 ~ +.19) if"' ".4. (5. the direct sum of two irreducible representation with fij = +ieijkfkO .2 or 0.20) is the totally antisymmetric tensor in three dimensions. the 'axial' vector '/51'". defined as a totally antisymmetric quantity with E'0m = + 1.3.24) The matrix '15 is a pseudoscalar in the sense that [j'P".1.2..4.1.4.) It can be shown 5 that any other irreducible set of 'Imatrices are related to these by a similarity transformation.5. (5.18) (The (Ii are just the 2 x 2 ymatrices in three dimensions. so the Dirac ma~ trices provide a reducible representation of the proper orthochronous Lorentz group. 0  Uk 0] (5. respectively. the vector yP.4. '1 equal to 0. sfpar must be proportional to E'parf/ contracted with some matrix sf".11) and (5.17).4. (T. (13 = (~ ~1)' (5.
1 that was originally introduced by Dirac. (5. Such reality conditions can conveniently be written in a manifestly Lorentzinvariant fashion by introducing the matrix p _ iyo of Eq.y3. (5. In particular.4.17) shows that P.26) and (5. but j'iO is antiHermitian.4.17) we have fij Hermitian.29) Inspection of Eq. Y5 is Hermitian and anticommutes with PI5tp~~Y5 P" so (5. because Dirac was mostly interested in electrons in atoms where v « c./I'tP = and it follows then that _yJl (5. together with Eq.) The representation of the homogeneous Lorentz group we have constructed here is not unitary. (5. (5.27).4. As we shall see. v l' c.4.4.4.4.y2.4.218 5 Quantum Fields and Antiparticles and anticommutes with all iJ. and in this case it is more convenient to adopt a representation [or which yO rather than Y5 is diagonal.33) and it follows that (5.27) The notation 1s is particularly appropriate.17) takes the fonn (5.32) Also.yl. the matrices D(A) satisfy the pseudounitarity relation (5.4.28) This representation is convenient because it reduces .31) Hence. (It is not. For the particular 4 x 4 representation (5. the matrix }'S is (5.PI> and "/5 to blockdiagonal form. this makes it particularly useful for dealing with particles in the ultrarelativistic limit.30) (5. which in the representation (5.4.4. because the anticommutation relations (5. though not unitary.34) . in the representation (5.4. however. the representation described in Section 1.4.17) of the {'matrices. because the generators f per are not all represented by Hermitian matrices.4.4.5) show that yO.13).Y5 provide a Clifford algebra in five spacetime dimensions.4.
1 .1 T.5 Causal Dirac Fields 219 The Dirac and related matrices also have important symmetry properties.40) (5.38) (YsY.'6.1.5.5. 2 and antisymmetric for 1.17) and (5.4.4. (5.4.4.35) where T denotes a transpose./jl'l') .18): 1J'i(x) and ~ (2rr)3/2 L " Jd'p ut(p.5.4.u).41) (5.Y2P It follows immediately that =i[ ~ ~2] .4. we can combine our results for adjoints and transposes to obtain the complex conjugates of the Dirac and allied matrices: (5.4.2) . = 1.4. Of course. In general these take the form given in Eqs.4.43) 5.4. (5.39) These signs will prove significant when we consider the charge·conjugation properties of various currents in the next section.18) shows that 'YII is symmetric for t. " (5.l = 0.I . (5.42) (5.1. (5.37) (5. discussed in the previous section. Inspection of Eqs.17) and (5.u)eiP·:<ad(p.1) 1jJ{C(X) = (21t)3/2 L Jd 3p Vt(p.)T ~ +'6YsY.36) /~v = 'C//1vr:e. (5.a) (5. so Y /1 T _ rL'"" ([. and rc.4.a)e'P')a(p. (5.5 Causal Dirac Fields We now want to construct particle annihilation and antiparticle creation fields that transform under the Lorentz group according to the Dirac representation of this group.1.3.
we have in matrix notation (5.1. and replacing the fourcomponent index t with a pair of indices.25) and (5. labelling the rows and columns of the supennatrix in Eqs.5. must be the same as !O' up to a similarity transformation.1.4. we must first use Eqs. (5. if we regard um +(O.4. (2.{p.5. one 2valued index'll labelling the rows and columns of the submatrices in Eqs. A general theorem of group theory known as Schur's lemma6 tells us that when a matrix like U ± or V± connects two such representations as in Eqs.a)J q i7 <7 Ul* = L m !tJmmvm±(O.lT) appearing in these formulas.22) to calculate them for arbitrary momenta.5.25) and (5.22) of the rotation generators.5) • We are here dropping the species label n.4.19) and (5.4. (5.26) read* and .1.a) as the m.3) and (5.4).3) and (5.20).1. we have J(I/2) = 10' and _J(I/2}· = 40"20'(12. and a second index taking values ±. and then apply Eqs.19). (5. and hence must be proportional to the unit matrix: 1 (5.5.4.l1) and vm±(O.5. It follows then that U± and V±a2 must commute with 0'.20).Lvm ±(O. the (2j + I)dimensional matrices JU) and JU)· and the 2 x 2 matrices 10' all provide irreducible representations of the Lie algebra of the rotation group.a).21). (so that Hence the Dirac field can only describe particles of spin j = 2j + 1 = 2) and the matrices J(I/2) and _J(I/2). with DrAA) in both cases taken as the 4 x 4 Dirac representation of the homogeneous Lorentz group discussed in the previous section.1.220 5 Quantum Fields and Antiparticles with the particle species label omitted here. in the standard representation (2.19) and (5. Using Eq.5.4) Now. (5.26) to find Ut and Vr for zero momentum. In other words.cr) and vr. (5.CT elements of matrices U± and Vi.1.5. . the zeromomentum conditions (5. the matrix must either vanish (a possibility of no interest here) or else be square and non·singular. In order to calculate the coefficient functions Ut(p.21) and (5. (5. In fact.
we are no longer making a distinction between f3 and pl.) ~ Jm/pOD(L(p))u(O. In general.5.1 = 17((211:)3/2 J LJ " d' p ur( p.5.11) Also.) Pa't(p. 0). these are quite arbitrary .. The only physical principle that could tell us anything abollt the relative values of the c± or the d± is the conservation of parity. 1: ') .6) (5. 0) ~ J m/pO PD(L(p))pu(O.5 Causal Dirac Fields 221 In other words u(O.o)p~l ~ ~·a(p..((x)P. so that the Dirac field would have only two nonvanishing components. . (5. and (5.10) d3p Vt(p.5.12) (5.. Eqs..5. (5. 1: ') c+ 0 c~ u(O. (Since p2 = 1. v(O. (5. D= 0 c+ 0 c~ .~ D=  d+ 0 d~ 0 and the spinors at finite momentum are u(p.7) It now only remains to say something about the constants c± and d±.).o) o(p. (5. 0) (5. <T)eip·YXact(p.5.o) (5. (1). (5..21). ~ Jm/pOD( L(p))v(O.13) ~ Jm/po PD(L(p))pv(O. 0) .1.) In order that the parity operator should transform the annihilation and creation fields at the point x into something proportional to these fields .4.8) (5. the particle annihilation and antiparticle creation operators undergo the transformations: Pa(p.5.5. 0 0 d+ 0 d~ 0(0.we could even choose c_ and d_ or c+ and d+ to be zero if we liked..) o(p.1. We recall that under a space inversion.9) and so P'I't(x)p~1 ~ ~'(2n)~3!2 L " Plp.22) give u( p.16).)e.).5.p9'o(p.)p~l ~ ~'o't(p.5.
In this case.t 2(2.a).14) where hu and bv are sign factors. 1 1 0 v(O. A straightforward calculation gives [tpt(x).l/JJ(y)h: = J d3p lIKI 2 N tl (p)e ip '(xy) + 111 2M t ?(p)e. (5.5.t 2(2.(p.17) h.I>t(p. 0 (5. it is necessary that PU(O. . ) ~ . u(O.5.(p.16) By adjusting the overall scales of the fields.) (5.a)v'(p.21) " Mtl(p) .5.I>.18). we can choose the coefficient functions at zero momentum to have the form: u(O. we find at zero momentum: N(O) " ~ 1 + b. (5.5.J2 1 0 h. Now let's try to put together the annihilation and creation fields in a linear combination (5. ') _ 1 .17) and (5.C1) and I1v(O.cr) and v(O.18) h.l) ~ J2 h.20) where Nt 7(P) .a)u. respectively: (5.a).ip ·(xy)] . J2 1 .(1) be proportional to u(O.5. the fields have the simple spaceinversion properties: (5.5.) ' M(O) ~ 1 + b. v(O. (5. 0 0 1 0 .5.5.5.l)~ ~ 0 1 0 .23) .15) (5.5. (5.19) that commutes or anticommutes with itself and its adjoint at spacelike separations.5. h~ = h~ = 1. .22) By using either the eigenvalue conditions (5.5.a).14) or the explicit formulas (5.222 5 Quantum Fields and Antiparticles at fJ'x.5.
29) '"+()~J x  d'p 'p' 2pO(2np e . We saw in Section 5. N(p) ~ 2 o [Ip'y. we find'· I . so that m appears in place of pO in the denommalors of the spin sums (5.'0. (5.25) 2. (5. or course.y) is an even function of x .5.2: X))ti.y space·like d+(X .30) •• Sometimes an cxtra factor pO 1m is included in the Dirac spillors.8) we have D ( L(p) )pD.7) we have then N(p) M(p) ~ ~ 2.5.(y)]" ~ (IKI 2 [l'O" + b"m]p". (5. (5.4.y) +1. M(p) ~ 2 o [lp'y" p + b"m]p .6) and (5. (5.28) Using this in Eg.27) (5.20) yields finally [w(x).(1 2 [_.28).OD(L(P)) [I + b"PJDt(L(p)) .26) Putting this together.5.y.1 (L(p)) ~ iL"o(p)y" ~ ip. p I .y and. + b.24) (5.4. (5. + b"mlp .5. Hence.27) and (5. The normalization convention nsed here has the advantage that it goes smoothly over to the case m = 0. ~.5. in order that both the derivative and the nonderivative terms in the commutator or anticommutator should vanish at spacelike separations. this implies that its first derivatives are odd functions of x .5.5.5 Causal Dirac Fields 223 From Egs.5.5.mJP"'+(y where d+ is the function introduced in Section 5.y" 1m.0 D (L(p)) [I +b"P1Dt(L(p)). The pseudounitarity condition (5.5.5.2 that for x .+(x .5.5. it is necessary and sufficient that (5. so by using the Lorentz transformation rule (5. J .32) yields and We also recall that fJ = iyo.
(5.. and .5. For this to be possible the phases in Eqs. (1)] " (5. a )eiP'Xa(p. N(p) ~ 2 o [W"l" + mJP .5. Just as for scalars we have the freedom to redefine the relative phase of the creation and annihilation operators to make the ratio KI)" real.30) is only possible if we choose the bottom sign. ~ b.5. the particles described by a Dirac field must be fermions.5. so we can always take b.39) Now let's return to the requirement that under a space inversion the field 1p(x) must transform into something proportional to 1p(&x). (1) + Vt(p.') 1 V2 0 1 0 0 1 0 I u(O. (5.16) must be equal.38) so the anticommutator is given by Eq..34) 1 while the coefficient functions at zero momentum are _ u(O.37) 1 .5.mJP. It is also then necessary that 11(1 2 = 1.5. (5. IT)eip'x act (p.5. (5. v(O.35) v(O.(x).=1.1. j')  V2 1 .31) Clearly Eq.224 5 Quantum Fields and Antiparticles and (5./. which changes the sign of both bll and hr .lJ~ .1 2 and bu = hr. and by adjusting the overall scale and phase of the field rp we may then take 1(=.. that is. + = +. (5.5. (5.36) The spin sums are 1 .j) ~ }z V2 1 0 1 0 1 I 0 1 0 . ~ +1. in which case PC = ).5. ~!(y)J+ ~ {[y"8" + mJP}tl ~(x  y) . (5.5. p (5.33) For future use. p (5.15) and (5.5.5.1. . (5.5.20) as [. we record here that the Dirac field is now W(x) = (2n )3/2 J L ! d P [Ut(p.32) Finally if we like we can replace rp with Y5rp. M(p) ~ 2 O[W"y" .
We also note that rc~lfJu(O. (5. + m)v(p. From Eq. a) ~ P'8v(p. (5. In order for the field to transfonn under chargeconjugation into another field with which it commutes at spacelike separations.14). = u(O. mverSlOn. (ip'y.5.41) ! Before going on to the other inversions. it is necessary . (5. + m)u(p. Eqs. a).a) and v(p.(p.a) are eigenvectors of ipJlYJl/m with eigenvalues +1 and 1.)f = IN?exp( F/JiVWJl~Y6'~lfJ and so in particular D(L(p))" ~ P'8D(L(p))'C. a) ~ 0 .41) we see that for general real wJlv: [exp( ti/JlVwJi\. but to obtain the coefficient functions at finite momentum we have to multiply with the complex matrix D(L(p)).4.cr) = v(O.5. so (5. (5.15) and (5.5.5. we will need expressions for the complexconjugates of the u and v coefficient functions.5.5. (5.40) That is.43) This is the celebrated Dirac equation for a free particle of spin From the point of view adopted here.a).5. (5. respectively: (il"'y.33) satisfies the differential equation (y'o.Bv(O.a) and "C~l.a) u.33).5. P'I'(i'h).5. a) .5.5.45) v.16) now give the transfonnation of the causal Dirac field under space inversion as p'I'(x)pl ~~. this is a good place to mention that Eqs.26) show that u(p. the intrinsic parity '"1' of a state consisting of a spin particle and its antiparticle is odd. and (5. the freeparticle Dirac equation is nothing but a Lorentzinvariant record of the convention that we have used in putting together the two irreducible representations of the proper orthochronous Lorentz group to form a field that transforms simply also under space .a) ~ P'8u(p.5 Causal Dirac Fields 225 so the intrinsic parities of particles and their antiparticles must be related by (5.5. It is for this reason that negative parity mesons like the pO and J /tp can be interpreted as swave bound states of quarkantiquark pairs. !.44) (5. (5.5. a) ~ 0 . In order to work out the chargeconjugation and timereversal properties of the Dirac field. .1P .(p. + m)'I'(x) ~ O. These functions are real for zero momentum.42) If follows then that the field (5.
the Dirac field of such a particle must satisfy the reality condition ~(x) ~ P'ii'~'(xj ..·')a'(p.p'. we can rewrite this as C<I> ~ L Jdlp Jd'p' X(p'.5. a. not a row.jad(p'...47). . (5.46) In this case. Following the same reasoning that led to Eq.47) (We are calling the Hermitian adjoint of the field on the right·hand side If..'.'j<I>o. respectively. There is an important difference between fermions and bosons in the intrinsic chargeconjugation phase of states consisting of a particle and its antiparticle. the field transforms as C~(x)Cl ~ tP'ii'~·(x). in the sense that if the wave function X of the state is even or odd under interchange of the momenta and spins of the particle and antiparticle. (5. this state transformed into C<I> IS ~ C(" L Jdlp Jalp' X(p. The classic example here is positronium.'j<I>o.·V·t(p. ~ = + l.5. The two lowest states are a pair of nearly degenerate swave states with total spin s = 0 and s = 1.5.. a".'j<I>o.p.a' where <1lo is the vacuum state. the intrinsic chargeconjugation parity of a state consisting of a particle described by a Dirac field and its antiparticle is odd. while the chargeconjugation parity must be real. then the chargeconjugation operator applied to such a state gives a sign lor + 1. '7 = +i.46).. (5. the bound state of an electron and a positron.48) For Majorana fermions the intrinsic space·inversion parity must be imaginary.' That is. Such a state may be written <I>  L Jdlp Jd'p' x(p. (5.a' Interchanging the variables of integration and summation and using the anticommutation of the creation operators and Eq. we have not ruled out the possibility that the two are actually identical.·)a'(p.p'.. Such spin ~ particles are called Majorana fermions.226 5 Quantum Fields and Antiparticles again that the charge conjugation parities of the particle and antiparticle be related by (5.) Although we have been distinguishing particles from their antiparticles.... Under chargeconjugation.)a'(p'..5.)a"(p'.instead of 1ft to emphasize that this is still a column vector.5. a.
5. (4. (5. we shall redefine the variables of integration and summation as p and a.5._a).a). Also. Eg5.a) in terms of u.. a) + Cv.5. 50 (1)! +a u· (p.&D(L(p))'r'ys.a). so para. respectively.and orthopositronium have C = + 1 and C = ~ 1. a )e. so they must have C = I. a )eiP Xact (p. respectively.a)T~l = t(l)~~<1a(p.50) 'C(_1)~<1d:t(_p.a).(p. which is consistent with their interpretation as quarkantiquark bound states with orbital angular momentum zero and total quark spin one.5.a) ~ (I)'Uu(O.52) ..49) (5.51) (5. a)] " In order to put this back in the form given for tp. we can use the fact that cliO anticommutes with f3 and commutes with rS together with our former result for D(L(p))* to write D'(L(p)) ~ YsPD'(L(p}Pls ~ Ys.a).(p.a). Recall the transformation properties of the particle annihilation and antiparticle creation operators given by Eg.(p. (1)' "v (p.a) and v. so we need formulas for ur(p.36) and (5. a).5.35)(5.34) thus gives T1pr(x)r' ~ (2n)3/'L x J d'p (_l)!u [C u.. These values are dramatically confirmed in the decay modes of positronium: parapositronium decays rapidly into a pair of photons (each of which has C = 1). Now we come to timereversal. The wave function for these two states is even under interchange of momenta and odd or even respectively under interchange of spin zcomponents. In the same way. Timereversal of the field (5. For this purpose. . while orthopositronium can only decay much more slowly into three or more photons.5.a) and v.5.5 Causal Dirac Fields 227 known respectively as para~ and orthopositronium. Ys'&'v(O.(p. single pO and roO mesons are produced as resonances in highenergy electron~positron annihilation through a onephoton intermediate state. a) = Ys'tl' u(p.2.a) (5.5.15): Ta(p..a)T1 = (5.iP ' x a( p.36) give 7s.&l u(0. Tact(p.+ ' ~ Ys'&v(p.a) ~ (I)'"v(O.4.(p.
5. (The terms 'axial' and 'pseudo' indicate that these have space· inversion properties opposite to those of ordinary vectors and scalars: a pseudoscalar has negative parity. (As discussed in Chapter 2. As already mentioned the Dirac representation is not unitary.54) Now let us consider how to construct scalar interaction densiti~s out of the Dirac fields and their adjoints. (5.pv and also ijJpM'Pn lpeMyS'Pv. For instance. while the space and time components of an axial vector have positive and negative parity. '11'. the original Fermi theory of beta decay involved an interaction density proportional to 1ppYP1. so 'P t 'P is not a scalar. If the neutrino is massless then its parity may also be defined as +1. or 'Is. they expanded the list of possible nonderivative interactions to include ten terms proportional to lppM'Pn ijJeM1. with YP.replaced with anyone of the five covariant types of 4 x 4 matrices 1. .jJJv. . and electron all have parity +1.) These results apply also when the two fermion fields in the bilinear refer to different particle species. To deal with this complication it is convenient to define a new sort of adjoint: ip ='I' t p.4. "ISyP.5. axial vector.pn 1peYp'Pv. respectively. (5.5. we are defining the spaceinversion operator so that the proton. except that in this case a space inversion also yields a ratio of the intrinsic parities.5. .228 5 Quantum Fields and Antiparticles We see then that in order for timereversal to take the Dirac field into something proportional to itself at the timereversed point (with which it would anticommute at spacelike separations) it is necessary that the intrinsic timereversal phases be related by (5. fpv. neutron. or 'Is yields a bilinear ijJM'P that transforms as a scalar. YSyP.53) and in this case (5.57) Taking the matrix M as 1.jpv. YSyP.32). and pseudoscalar. under a space inversion (5. with M running over the matrices 1. 'II'.) When Lee and Yang? called parity conservation into question in 1956. if necessary by replacing the neutrino field with '/s'Pv. respectively. we see that the fermion bilinears constructed with ijJ have the Lorentz transformation property Uo(A)[ip(x)M'I'(x)]UOI(A) ~ ip(Ax)D1(A) M D(A)'I'(Ax). vector.56) Also.5. tensor. yP.55) Using the pseudounitarity condition (5. or 'IS. Later it was realized that the most general Lorentzinvariant and parityconserving nonderivative beta decay interaction takes the form of a linear combination of products like this.
IllY are as usual raised or lowered by contraction with 'lilY or 'lilY') To see how to construct such matrices.Iptr] = i (.IllY' (The minus sign in the first line arises from Fermi statistics.120.IllY = .for 'Ill and . we have C(.pM'P (5. A general representation of the proper orthochronous homogeneous Lorentz group (or.2) and a boost Jf"1=.6..pM'P)C. (5.1) (Of course. We ignore a cnumber anticommutator.6 General Irreducible Representations 229 It is also of some interest to study the chargeconjugation properties of these bilinears.47) and (5. lies somewhat out of the book's main line of development.6 General Irreducible Representations of the Homogeneous Lorentz Group· We shall now generalize from the special cases of vector and Dirac fields to the case of a field that transfonns according to a general irreducible representation of the homogeneous Lorentz group. and C = 1 for vectors or antisymmetric tensors. pseudoscalars.IllY satisfying the same commutation relations (5. more properly. Using Eqs.4.6.4.5. .IJltT'lyp) .IllY into two threevectors: an angular momentum matrix (5. its infinitesimal part) is provided by a set of matrices .1 ~ (P'lI'P)'PM(p'tI'P') ~ (P'lI'P')'MT'lI>p ~ .58) the sign in the last expression being + for the matrices I.IYIl' and indices on . while the photon hasC=l. and may be omitted in II first reading.6. or axial vectors.1) then reads (5.4.Illp 'lvo  .3) Eq.39).p'llIMT'lI'P ~ +. $'2=. All fields may be constructed as direct sums of these irreducible fields. This is one way of seeing that the nO (which couples to pseudoscalar or axialvector nucleon currents) has C = +1. . (5.6. first divide the six independent components of .6.Il0.IOy'lPIl . Y5YIl' and Y5.35){5.5. (5.4) • This section. and .4) as the generators of the group LfIlV' .5.IPY'l01l + . 5. (5.) A boson field that interacts with the current f{JMtp must therefore have C = +1 for scalars.
17) . (5. k run over the values 1. b ~ B .6.6) are equivalent to [.13) UJI)db'.6.14) .10) (5.~jJ = if"ijk .. (5. and Eq.6.ab = e5 b'b (5. (5. we label the rows and columns of these matrices with a pair of integers and/or halfintegers a.8) It is easy to see that the commutation relations (5.6.f!1 j] = i Eijk .5) just represents the fact that . .a±l V(A (5.Jf' with two decoupled spin threevectors. 1/2) indices..as a direct sum.e" as a symmetric 5U(2) tenwr with 2j twovalued indi~es.. and plays a crucial role in what follows. running over the values a ~ A .W k [~j.16) + o)(A + a + 1).6. the latter written with dots to distinguish them from the former. (5.6.cBk [. .s based on the fact that the spin j representation of the rotation group can be written as the symmetrized direct product of 2j spin 1/2 representations .0) indices and 2B twovalued (0..ab = J~1J ' bal a Jb~~ (5. There is an alternative formalism.7) f!i9 _ ~(.6. .. +A ". Eq.6. The minus sign in the righthand side ofEq.6.9}(5.6. B + 1" .6.  iX). (5..9) (5. A + 1 . writing •<$ ~ _ 1 2 ( .6. 3.6.6.$k.4)(5. + .6.6) where i. ~ ao".11) in the same way that we find matrices representing the spins of a pair of uncoupled particles . That is.4) just says that the matrices.11) . (5.6.i.6.12) (5.if"alj] ~ 0. (5. and f" ijk is the totally antisymmetric quantity with Em _ +1.6.b. We find matrices satisfying Eqs... (5. It is very convenient to replace the matrices f and .X"j] = if"ijk. j. (5. +B and take·> (d)u'b'.'.6.6) arises from the fact that '100 = 1.Jf' is a threevector.15) where J{A) and J(B) are the standard spin matrices for spins A or B: (JiA+ iJiA1) a'a = ) (J\A)). We can therefore write fields belonging to the (A.230 5 Quantum Fields and Antiparticles (5. bd. B) representation with 2A twovalued (1/2. (5. .6.2.$ generate a representation of the rotation subgroup of the Lorentz group.Xl.d'i.5) [X"j.
and do not need to have a Lorentzinvariant positive norm. but instead is the noncompact group known as SO(3. and corresponds to an antisymmetric tensor FI'\" that . with j~A+B.A) field. which is required by the minus sign in (5.6 General Irreducible Representations 231 and likewise for J{Bl. A+BI.e. we can see that a field that transforms according to the (A.5. the rotation group is represented unitarily.7) and (5.. 1). and therefore f is Hermitian but jf' is antiHermitian.2A .. and corresponds to a traceless symmetric tensor of rank 2A. by the identity).0) field is obviously scalar.0. The finite~dimensional representations of the homogeneous Lorentz group are not unitary. with its generators represented by the Hermitian matrices (5. }'5 = +1) and bottom (Y5 = 1) two components of the Dirac spinoL A (1. (5.!) field can only have j = +!.0) or (0. More generally.1) field can only have j = 1.6. because the objects we are now concerned with are fields.6. This is because of the j in Eqs.8).rtf and fJ!J are Hermitian. .6. It is only compact groups that can have finitedimensional unitary representations (aside from representations in which the noncompact part is represented trivially. not wave functions. '. a compact group. a (0. these are the top (i.!) field has components with j = 1 and j = 0. (Note that the number of independent components of a symmetric tensor of rank 2A in four dimensions is 4'5"'(4+2Al) (3+2A)1 (2A)! 6(2A)! and the tracelessness condition reduces this to (3+2A)! (1+2A)! 2 6(2A)1 ~ 6(2A~2)! ~(2A+l) ° as expected for an (A. A (!. We see that the (A. because .) One more example: a (1. In contrast. The representation is labelled by the values of the positive integers and/or halfintegers A and B.6. B) representation of the homogeneous Lorentz group has components that rotate like objects of spin j. with only a single j = component.B) representation has dimensionality (2A + 1)(2B + 1). A) field contains terms with only integer spins 2A.6).1."'. IA~BI. For instance. and hence stems from the fact that the homogeneous Lorentz group is not the same as the fourdimensional rotation group SO(4). corresponding to the spatial part v and timecomponent vO of a fourvector vI'. an (A. B) representations with the perhaps more familiar tensors and spinors.O) or (O. There is no problem in working with nonunitary representations.18) By the usual rules of vector addition. This is enough to identify the (A.
vlp F.. these (A. ( I. tbe field describes a single particle of spin j = 3/2. The spin representations.6. 'P for (1.8).. As we have seen..0) EEl (O. representation t by requiring that YjJtpll = O.0) and (0.2 ~p.!) representation gives a spinorvector 1pIJ. '. .6. B) for which A + B is an integer.I ~ +J.!) representation and the Dirac (1. The quantity Y/ltplJ would transform as an ordinary (1. j) El> (0. taking the direct product of the vector (~.1) EEl (1. that transforms according to the reducible representation (!.6. N /2 I. A general tensor of rank N transfonns as the direct product of N fourvector representations.6.0) El> (0.232 5 Quantum Fields and Antiparticles satisfies the further irreducibility 'duality' conditions FI'V = +~ . for which A + B is half an odd integer. In terms of the matrices (5.19) pdpI ~ 1iIJ. the vector. For instance. . In this way. and B ~ N /2. the irreducible representations of the Lorentz t According to Eq. respectively.'i. PriBp.7) and (5. For A f B. and the symmetric traceless tensors. It can therefore be decomposed (by suitable symmetrizations and antisymmetrizations and extracting traces) into irreducible terms (A. B) representation of the proper orthochronous homogeneous Lorentz group does not provide a representation of the Lorentz group including space inversion unless A = B.1) fields. and in particular 1) PJP.O) El> (1. DJ ~ ( j. (5. The doubling is eliminated by imposing the Dirac equation [yV 0. This is the RaritaSchwinger jield. + mlll'~ = 0.0) EEl (O.18). l). A) representations are the scalar. such a field transforms under ordinary rotations as a direct sum of two j = 3/2 and two j = components. 0) EEl (0. With these conditioll. can similarly be constructed from the direct product of these tensor representations and the Dirac representation (!. 1) El> ( j. !) 1). there must be a matrix It which reverses the signs of tensors with odd numbers of space indices. this is (5. it is only in four dimensions that an antisymmetric twoindex tensor F/l v can be divided into such 'selfdual' and 'antiselfdual' parts.!) Dirac field. and the remaining j = component is eliminated by requiring that a~1pp = O. N /21. 9 So far in this section we have only considered the representations of the proper orthochronous Lorentz group. we can construct any irreducible representation (A. ! ! . so we can isolate the (1.1 ~ d . l) ® [(j.20) Thus an irreducible (A. B) with A ~ N /2..6. In any representation of the Lorentz group including space inversion. Of course. (5.
abUab(O..1.O) $ (O. a) (5.4.o .Jaavab(O.O").)3/2 I: " J d'p [K a(p.14)(5.Vab(O..1. The 4 x 4 matrix (5.O")Jr. One of these is the (!.A).5.. The fundamental conditions (5..a). 5.JbbVab(O. We are here leaving open the possibility that this particle is its own antiparticle. running over the ranges (5.7.2) is the defining condition for the ClebschGordan coefficients CAB(jO".4..!) Dirac representation discussed in Section 5.O"). .6.12).L. of dimensionality 2(2A + 1)(2B + 1).UJo _ " (A) _ .7. Our first task is to find the zeromomentum coefficient functions Uab(O.B~matrix for this representation. b (8) + "L.29) provides the .6.7. The index t is replaced here with a pair of indices a.U) .B) representations described in the previous section.7 General Causal Fields· We now proceed to construct causal fields that transform according to the general irreducible (A. (5. if)JV~ = L J~~)Uab(O. (5. 0")] with K and A arbitrary constants. including both self~dual and antiselfdual parts. + L J~~)Uab(O.it)JV~ = Lc?ab.25)(5.26) on u(O. (5.2) a _ . tT )eiP''''Vab(P.a) read here LUab(O.6. and may be omitted in a first reading. tT) . (5.O)EB(O.O") and v(O. u) r.13).a) = a(p. which as we have seen isjust the antisymmetric tensor of second rank. ij a." L. 0") and vab(O.B) $ (B. b.1) +A act (p. in which case aC(p. so the fields are now written as 'M(x) ~ (2.. 1) representation.15) L uiib(O. Another familiar example is the (l.7 General Causal Fields 233 group including space inversion are the direct sums (A.7. a)eip <U"b(P.6.ab)! These coefficients are defined by the requirement that • This section lies somewhat out of the book's main line of development.U) ij (5.b or using Eqs.3) a b But Eq.
(5.7) and write LPv(e) in place of U'v(p). we have [L(O)]i\' where wio = WOi = Pie and Wij = woo = O.7. the unique solution for v(O.234 5 Quantum Fields and Antiparticles if qJ ab are states that under an infinitesimal rotation transform as (jqJah = iL:0' J~~}qJab + a iL: (J.6. under the same rotation. (5.I The advantage of this parameterization is that L(Il)L(B) ~ L(1l For infinitesimal e.O") is v.4) This result is unique because each irreducible (A. (5.sinhB. and therefore. (5.cr). ab). where Lik(O) Jik + (cosh (} .7.6) We must now perform a boost to calculate the coefficient functions for finite momentum. = sinhB ~ Ipl/m (5.I )pdJk. u) is just CAB(jer. Uab(O.2).24) as a function of a parameter defined by e coshB ~ yip 2 + m2 /m.9) .7. This constant is conventionally chosen so that ". 0"). LOo((}) = cosh (. a) ~ (2m)1/2CAB(ja.ab)\{Iab transforms as " (j'¥J IJ = i ~ (} .ab). iihpJ_f . L'o(B) ~ LO. + 0). (5. inspection of Egs.0' • (5. + Jl'v + wl'v.a). we can write the boost (2.7.8) . it takes the same form as Eq.5) Therefore if we write Eq. the state qJj(f  L:CAB(jO".16)(5.7.7. (5. up to a possible proportionality factor.5.7.. ~ 1111 ( u Inspection of Eq.17) shows that the complex conjugates of the angular momentum matrices are .") ~ (1V+ u"'b(O.7.6.JUl· = (1 )aa' JUl a a' 0.3) in terms of (_l)ja Vab (p. For a fixed direction p . Jt~hPab h then.7. Similarly.p/lpl. Following the same reasoning that led from Eq.(0. B) representation of the homogeneous Lorentz group contains a given spin j representation of the rotation group at most once. (5. (5.(O.2) shows that this requirement is satisfied by the coefficients Uab(O..(p) ~ p. With a suitable adjustment of a constant factor.
6.10) This is for any representation of the homogeneous Lorentz group.13) Eqs.15) D(L(p)) a'b'. (5.'a".1.(Ax).7. .r4 and f1l are commuting matrices (5.17) alar·'a" b\I12'·'h" by simply taking &alu2'.7..6. lt is very easy in this formalism to construct Lorentz scalar interaction densities.5.7) and (5.. A2.."'0) exp(+p .2.(p.Ke).b j hr b" as the product of a coefficient for coupling spins A}. a'b' (5..cWO) .).. Thus we can construct scalars of the form '" L.7...) ~ ~ L (exp ( .15) show that the matrix generators of the (A.7. Eqs.p' jlAie)) a'a (5.7 General Causal Fields (2.(AI)"".2.O) and (0.11) D(L(p)) ~ exp(p . to make a scalar and a coefficient for coupling spins Bt.. (5.B) representations. B2. for the irreducible (A. The (A. B) representations.7. Eqs. it follows then that 235 D(L(p)) ~exp(jp.K ~ sf .I gaja2'··a". (2. 8. (Even though we do not explicitly consider interactions involving derivatives.8) give i. respectively.6) then give the coefficient functions at finite momentum as u.6..6.) ~ (l)j+'U'h(p. (5. B) representation of the homogeneous Lorentz group is just the direct product of the (A.. B) representations are just the spin matrices for spin A and B. so the general Lorentz transformation rules (5.ab ~ (exp (.14) and (5. tpa"b n (x ) (5.7.6). using Eqs.· .4) and (5.. we will in this way obtain the most general interaction involving n fields.6. . '" L.O) and (0.14) and (5. to make a scalar. (exp (+ p' J(Bi8l) bib (5.··· A..7.7..iJII and since .(p.7. because the derivative of a field of . (5.(x)UOI(A) ~ LD:~(AI)Dg~.. so the field (5.a'b') u.p' jlAiel) aa (exp ( + p' J(Blel) btr 2 O p db ' X CABU.14) and (5.12) In more detail.15) These results give the field explicitly for a given transformation type (A.7.. (5.26).7) read here Uo(A)""..16) Furthermore. (5..1.31) of this type is unique up to the choice of the constant factors K and J.6.24) to Eq.1.h1h2 "b" ~)ajbl ( X ) tpa2b2 () ••. (5. (II (2l X (. B).
17) denote the Wigner "threej" symbols: lO h ( m\ m2 12 13) =2:: Cj1hU3m.22) .m3) m3 m' .(Y)].. it is also necessary that Jf(x) should commute with Jf'(y) at spacelike separations x .b(p.7.ll What concerns us here is the fact that it turns out to be the massshell value of a polynomial function P of p and pO: (5. ! ' I~ ~ (2n)3/2 J Jlp (2 po)l n'b.) For the Smatrix to be Lorentzinvariant it is not enough that the interaction density Jf'(x) be a scalar like (5.4.b. To sce how to satisfy this condition. .B2 ).)V. Ed. B). mlm2)CiJiJ (OO."(P) ~ 2: a' hi 2: 2: CAB(J.'i':...7.)il. B) can always be decomposed into fields of other types without derivatives.) In more detail n.b(P."(P)  (5. B).(XY) +AX"eiP'(XYI] .) For instance.7.18) with a single free parameter g. the top and bottom signs are for bosons and fermions.) (5. where n(p) is the spin sum (2pO)ln'b. and the adjoint . We find ['M(X). ..7.b(p. (The brackets in (5.21) The function 1t'(p) has been calculated explicitly.18).a'b') x(exp( _p'J(A1e)L (exp(p. al A 2 Al) (Bl a2 Q3 1I1(J2113 bI B2 b2 Bl) b3 (I) (2) (l) 1falbl1fa2h21faJb3 (5.) ~ 2: V.7.236 5 Quantum Fields and Antiparticles type (A." if hi r1 (5. m. .a'b') CitB (J..J(B1e))w x(exp( _pJ(Ale))~ (exp(pJ(B)e)). which describe the coupling of three spins to make a rotational scalar. (A 2.. the most general Lorentz scalar formed from a product of three fields of transformation types (AI.19) 2: U.(p.y..7.7. respectively. a field 1p of type (A. This is the most general three~field interaction.20) and as usual. (We allow here for different coefficients K and 1 in the rp field.p t of a field ip of type (. '" '" (AI b J b2b3 L. consider the commutator or anticommutator of two fields for the same particle species.Bl) is g L."(P) x [KKoeip. and (Al.
.. In order that this should vanish when x .23) We shall check this here for just one particular direction of p.(P) = Pab .26) Q(_p) ~ _(_)2A+28Q(p). we can adopt a Lorentz frame in which x O = yO.7. Taking p in the threedirection._pO) ~ (_)2A+28 P(p. (b > a) (a > b) [(pO _ p3)/m] where pO 2.2' We see that 1t(p) can indeed be written as the mass~shell value of a polynomial P(p.'b(P) ~ L CAB(jU. 2b . Any polynomial in p and Vp 2 + m2 can be written in a form linear in Vp 2 + m 2 (by expressing even powers of VP 2 + m2 in terms of p) so 1t(p) can be written 1taMi.y spacelike.po). • f. a + b. (5.27) . (5.( iV)~+(x .po). ab)CAB(Jer.. Also. ab) CA8 (ju. ab) x [(po + p3)/m] 2b2.a + b = 2a + er + 2b .4+28 )'x'jQ.24) (5.25) (5.1p~(Y)h ~ [KK' + (_)2A+28 AX']P"..19) as [1p'b(x).a1i(P) + 2Jp 2 + m2Qab .!( iV)b 3(x . ilb) exp ([u + b u il + b] (J e) = The ClebschGordan coefficients vanish unless a = a + band so we can replace a + b.1)2A+28'" A. so the polynomial satisfies the reflection condition (5.y) .2a.iib(P) = L CA8(ja . +( K K = _ .y. For x .5.7.21) gives "b.2a equals 28 + 2A minus an even integer.7 General Causal Fields 237 and that P is even or odd according to whether 2A + 2B is an even or odd integer P(_p. so here 1t ab.7.7. We can write exp(+O) as (po + p3)/rn.a = 2b .y.7. (5.7. 0) +[KK' + (_)2. and write Eq.ab(P) ' where P and Q are now polynomials in p alone.A.23).7. we must have (5. with P(_p) ~ (_)2A+2B p (p) = VP 2 + m 2• (5.7.
7..30) where c is the same factor for all fields of a given particle. we have ~ = (_1)2B+2B ~ . Starting with a (0. the possible fields for j = 0 are those of type (A. Eq. so in particular A = A and B = B. vector.l12. We can therefore eliminate e for all fields by a redefinition of the relative phase of the operators a(p. Using Eq.. so that e = 1.b(p. it also involves 1pt. because the hermiticity of the Hamiltonian requires that if £'(x) involves lp. for any field. where the fields 1p and ijJ may be different.~u)eiPX]. (5. and dividing both sides by IRf = IXI 2.(j)act(p.)a(p. This is the general relation between spin and statistics. The different fields for a given particle do not really represent possibil~ ities that are physically distinct. We emerge from all this with a formula for the (A. This is possible if and only if + (_1)2A+2B and = +1 (5.u). (5.28) (5.7.27) gives IKI' ~ +(_)2A+ 2B I.Jr(y)].) In this case.12 of which we have already seen special examples for particles described by scalar. 2A + 2B differs from 2) by an even integer.u) and acf(p.29) shows that c is just a phase. = (_)2B K. Eqs.238 5 Quantum Fields and Antiparticles Now let us consider the special case where 1p and ijJ are the same. so Eq.1.7. Also.31) +(_)2BVab(P. A) (because the triangle inequality IABI < j < A+B here requires A = B).7.7.7. the factor K for each field type may be eliminated by a redefinition of the overall scale of the field.7. B) field of a given particle. or Dirac fields. (5.7. (5.29) Of course. we can easily .. Furthermore.)e iP ' (5. Now let's return to the general case.28) says that our particle is a boson or fermion according to whether 2) is even or odd.*'(x). lei = 1. (5.27). For instance.0) scalar field 4>. and hence. that is unique up to overall scale 'M(X) ~ (2n)3/2 L p J d'p [U. K A It follows that. (It is unavoidable that such commutators or anticommutators will appear in [.
(5.7. for elll:h ii.m") rp~(x) = 0. a IS just the Kronecker symbol 1J~~. since D(A) has an inverse D{A.33) This transforms as the direct product of the representations (B. beginning with space inversion. .ii) = O.7. and hence by the usual rules of vector addition.ible unless all the M~(ip) vanish. Now let us consider the behavior of these fields under inversions. {u PI "'OPLJ4> . where {} here denotes the traceless part. respectively. . .7. B) for a given particle of spin j.. consider the field ° {ap> . More generally. (5..BI < A < i + B. the~.32) {a"a.) But Eq. B) with Ii . (Recall that a traceless symmetric tensor of rank N transforms according to the (N /2.7. Using the results of Section 4.31) for j = can be nothing but linear combinations of the 2Ath derivatives (5. B) for a given particle of spin j can be expressed as a differential operator of rank 2B acting on the field 13 iPa(x) of type (j. .1 ~ ~'a(p.} ~" .7.34) (5. To see this.0).7. (5. The only possible flaw in this argument would be if some of the (A.2.7 General Causal Fields 239 construct such (A.O) field 'P~ would satisfy a field equation L~ M~(cl03.O) fields rp~(x) thus satisfy no field equation other than the KleinGordon equation (0 .7. ap. Since Eq. N /2) representation... The (j.7J3) can vanish. B) and (j.7. B) field obtained from the derivatives (5. But in this case.. For the (j. the spaceinversion properties of the particle annihilation and antiparticle creation operators are: Pa(p. The general causal (A.7.7..x)'P~{x) = 0 and hence..a)p. A) fields (5. for instance (5..a).A) fields from the 2Ath derivative ! . which is impos. 5. ~M~(ip)u~(p..B) field (5.33).31) represents the unique causal (A.a) repre· sentation the ClebschGordan coefficient C/lJ(jii.}_ 02 8x Pox v j~..j».31) represents the unique field of type (A. any field (A. (5. or equivalently IA . it can be decomposed into fields transforming according to all the irreducible representations (A. it can be nothing" but the (A.I ).o. so the (A.31) thus transforms under .0) (or a differential operator of rank 24 acting on the field of type (O. and therefore none of Ihe (A. B) field for a given particle of spin j.35) where I} and I}c are the intrinsic parities for the particle and antiparticle..(L(p))=O. B) fields obtained from (5.BI < i < A + B. so this would require L~ M~(ip)DM. .32) of a scalar field.B) fields obtained in this way actually vanished.
.)J/2 L . and 1 intrinsic parity for fermions.7.1 ~ (2. (5.7.a)eiPX v1:(p. (5. (5.7.31) is the unique causal field of any type.)3/2 L! .(x. a) J.3. Ptp1bB(X)p~1 = .37) U::(p.u).31) because. (5. and for this purpose we will need to evaluate Uab(p.1 ~ (2. so this gives ~'~ ~'(_)2j . a )eip''?xvfaA(p. This gives (5.7. Hence the ratio of the coefficients of the two terms in Eq. the of a particleantiparticle pair is +1 for bosons. d3 p (_I)A+Bj X [.7.42) in Eq.ab) ~ (t+BjCBA(ju. This is the causal field tpfaA evaluated at . Using Eq.*(_)A+Bjtp:aA(_x.7.ba). (5.xu::(p.36) +ttC(_)2Bact(_p.31) (but with B replaced with A because this is supposed to be a (B. (5.7.14) and (5.15).39) v::(p.A) field): ~'(_)2B /~.7.B differs from the spin j by only an integer. (5.31). where j = 0.a)]..7.7. our final result for space .2.7."(x)P. (5.u)eiP< U::(p.5. We want to change the integration variable from p to p. xo).<?. j = 1. We now see that the result is general. (5. and j = respectively. ! dJp [~'a(p.240 5 Quantum Fields and Antiparticles the parity operator Pinto P". (5. the sign (_I)A+B~j is just . (T) + 'I C( )2B act(p. ~ (_)2A.7.a) = (_)A+Bj vfaA(p.u) (5.40) must be the same as in Eq..I1). But these coefficients must be the same up to an overall constant factor as in Eq.( a(p.41) But A . InverSIOn IS "e" 1.42) We saw special cases of this result in Sections 5. aside from scale.7.\. and 5.u) .38) (5. (5. . To do this.7. a )eip. and use the symmetry property of the ClebschGordan coefficient 14 CAB(ju. 0) and bottom (0. a) = so (_)A+Bj u::(p. P".40) where. 5..7. as before. (5.~x.43) Let's see how this applies to the Dirac field. H components of the Dirac field.a) and V b(p.40).7. a we need only glance back at (5. [JJx .::(x)P. Eq. except that the coefficients of the annihilation and creation terms may not be the same as called for in Eq. FaT the top ( ~.7.xO).
we find C'I'1t(x)C. so ug.. J1B1e)) bIi L. Now let us consider chargeconjugation. (5.31).46) It is useful to compare this formula for the chargeconjugate of an (A. 1p~:~a(x) = (2n:)3/22: J a) d3p (_)a+brru::(p._a(p.)3/2 e e L " J dJp ~t(p.7.O'.47) is then (replacing a _ a. CfJ au oc (ly.(p.7.fl.J 2p a'b' ' ' .5.1 = ~cat(p.47) x [(_1)2A(_l)jU aC(p. respectively.I ~ C a'(p. (5. Its effect on the particle annihilation and antiparticle creation operators is Ca(p.7.A) field for the same particle: 'I'~.).7. The reversal of the top and bottom components of the Dirac field is accomplished by the matrix fJ in (5.7.41). J1A1e)) a a (exp (ji .7. Applying this transformation to the field (5. so the parity operator simply takes x into x.or = (_)a+brr u:~(p. we use our previous result J(jl· = _'6'JUlr. (5.1 ~ (2.At(x) ~ (2n)3/2 L J d3putt (p. Cact(p.) x [Ca'(p.O')eiP 'X] " To calculate the u·..) C.7.a') = CAB(jO'.7..14) is real..f ~ ~ "( exp( ji . The ClebschGordan coefficient in Eq. We use the reflection property of the ClebschGordan coefficients 14 CBAU.O')eiP·X]. b _ b. (5._u._O')e iP 'x +at(p..O')eiP'X +at(p.5.48) and the fact that those coefficients vanish unless a' + b' = a.a'b' ) u~t.O') x [(_)2A.O'). " .7 General Causal Fields 241 +1.) (5._.b'd).O").. B) field with the adjoint of the (B. X (j""(_)b'bCBAU.7..49) 0"  The field adjoint (5. and multiplies the field with 11·.O') C.)e'P·x +~'(_)2Bat(p.(_y+rr aC(p.)(_)j"ejpx].. to write (5.44) (5.b'.u ba. reverses the top and bottom components.45) where and c are the chargeconjugation parities of the particle and antiparticle.
~(J)ejp·xJ. 1 = ~.7. _(1)e. it is necessary that it be proportional to 1p~t~a(x). (5. . and 5.a. (5.~a(x) = (2n)3/2 L x [a\.52) is satisfied without any chargeconjugation operator on the left~hand side or phase C on the right: ! 1f!ab AB( x ) . we use Eq. (5. this gives 1 (5.7.ip xJ.a. (5.7.46). Eq.7.7. .(J)(1)j11 x [Ca(p. We have already encountered the relation (5.14) and the standard formula 14 CAB(j.7.50) with Eq.7. for a particle that is its own antiparticle. (J")eip~ J d 3p u::(p.5. Comparing Eq. Finally we come to time·reversal.53) We have already seen an example of this sort of reality condition for Majorana spin particles in Section 5. and in Sections 5.7.1 should commute or anticommute with all ordinary fields at spacelike separations. BAt ( ) (5752) .56) To calculate the complex conjugate of the coefficient function.7. and noted some of its implications for electronpositron and quarkantiquark states in Section 5.31) thus has the transformation property T1P:bB(X)T.57) .51) for spins 0. we see that this is only possible if the chargeconjugation parities are related by (5. (5. Applied to particle annihilation and antiparticle creation operators. In particular.7.a) " + (_lJu+2B at(p.1 = (2n)3/2 L " J d3pu:t(p.51) in which case C 1jJ un xC.5. AB() .7.55) The irreducible field (5.b) (5.A). this is ~)~t. 1.(_)2Aahj 1jJ ba X . transformation type (B. (5. 5. because this is the adjoint of the unique causal field of .:(p.2.a.50) In order that C JP:bB(x) C.( _ )2Aabj ~)ba BAt (x ) .3.242 5 Quantum Fields and Antiparticles (_)~2Aj = Using the sign relation (_)2Aabj (_)28+).a.54) (5.7.~(J)eiPX + 'C(_1)2Bact(_p.7.b) ~ (_)A+BJCAB(j.5. (5.
these are encountered in the study of the propagation of a higher spin field in the presence of cnumber external field.7. there is no reason to doubt that this is always possible.)1l+h+(J"+A+Bj~~.5. it is necessary that ~ eO = (' . These are free fields. unphysical mass states. Thus any difficulties with higher spin can only arise when we try to go beyond perturbation theory. even for weak external fields.58) Changing the variables of integration and summation in Eq. so no question of inconsistency or unphysical mass states can arise. . I will not go into details about these problems here. the difficulties encountered include noncausality. because it seems to me that they are not relevant to the calculational scheme described in this chapter. (5.6. there can be no difficulty with unitarity.7. But the results obtained in this way depend on all the details of the interaction . Lorentz invariance is guaranteed in perturbation theory as long as we add appropriate local but noncovariant terms in the Hamiltonian density. (5. As long as the interaction Hamiltonian is Hermitian.7.7. though a rigorous proof is lacking.7 General Causal Fields 243 and find: U:f( p. the solution of field equations in the presence of a cnumber background field (the context where all the problems with higher spin have been found) does go beyond perturbation theory. B) field to be transformed by timereversal into something proportional to another (A. B) field. we can use perturbation theory to calculate Smatrix elements that automatically satisfy the cluster decomposition principle. if the fields are sufficiently slowly varying. This partial summation is justified._b(X. inconsistency._b(P. rr) = (. for the following reasons: (I) The fields lpab(X) have been constructed here directly from the creation and annihilation operators for physical particles. (2) As discussed in Section 13. but by incorporating them into an interaction Hamiltonian density in the interaction picture.XO). (5. the smallness of energy denominators making up for the weakness of the fields. in that the results correspond to summing an infinite subset of terms in the perturbation series. rr) (5.60) ••• It should be mentioned that from time to time various difficulties have been reported 15 in the field theory of particles with spin j > 3/2. Generally. Depending on the details of the theory. and violation of unitarity. we find that in order for an (A.59) in which case Tlp:~(x)Tl = (_)Il+h+A+B2j C 1iJ~~.56) to ~p and cr.
those whose interactions with external fields are particularly simple. particles of higher spin are expected to have interactions of all possible types allowed by symmetry principles. that is. B) that contains the j = 2 representation of the rotation group. and as we shall see in Chapter 12. that can be summarized in the statement that . So the requirement of simplicity does not seem to have any objective content. a point generally neglected in earlier work. No one has shown that the problems persist for arbitrary interactions. (4) Also. It should be kept in mind that the requirement of simplicity depends on the choice of which field we choose to represent the higherspin particle. but interactions that are simple when expressed in tenns of a field of one type may look complicated when expressed in tenns of a field of another type. For one thing. If there is any problem with higher spin. there is no doubt about the existence of higherspin particles. The problems reported l5 with higher spin have been encountered only for higherspin particles that have been arbitrarily assumed to have only very simple interactions with external fields. the spin two particle is represented by a (1. including various stable nuclei and hadronic resonances. it can only be for 'point' particles. the interactions may be reexpressed in the interaction picture in terms of any field type (A. Remember that any free field types for a given particle can be expressed as a derivative operator acting on any other field type. there are good reasons to believe that the problems with higher spin disappear if the interaction with external fields is sufficiently complicated. Not only is it necessary that every particle have an antiparticle (which may for a purely neutral particle be itself). 16 (It was found that the consistency of the theory depends on the assumption of realistic external fields that satisfy the field equations.1) free field. 5. so in the interaction picture any interaction with external fields may be written in tenns of any field types we like.244 5 Quantum Fields and Antiparticles of the highspin particle with the external fields: not only the multipole moments of the particle but also possible terms in the interaction that are nonlinear in the external fields. there is a precise relation between the properties of particles and antiparticles. both higherdimensional 'Kaluza~Klein' theories and string theories provide examples of consistent theories of a charged massive particles of spin two interacting with a electromagnetic background field. but as mentioned above. (3) In fact.8 The CPT Theorem We have seen that the demands of relativity combined with quantum mechanics require the existence of antiparticles.) Reformulating this work in the interaction picture.
and the bilinear therefore satisfies Eq.5) (Any complex numerical coefficient appearing in these tensors is transformed into its complex conjugate because CPT is antiunitary.5 give (5.}i" formed from any set of scalar and vector fields and their derivatives transforms into (5. then using complex UJrentz transformations to prove a reflection property of vacuum expectation values of products of fields.pl(x)y.1 7 It has been proyed rigorously in axiomatic field theory. (5. then M is a product of n modulo 2 Dirac matrices.MY''P2(X)]t .8.8.8..6) (A minus sign from the anticommutation of P and Y5 is cancelled by the minus sign from the anticommutation of ferrnionic operators.4) Then any tensor q. so Y5MY5 = (1) n M.IS by using commutativity assumptions to extend the Lorentz invariance of the theory to the complex Lorentz group. This is the celebrated CPT theorem: As a first step in the proof.'P.8.}il .8 The CPT Theorem 245 for an appropriate choice of inversion phases.) We can easily see that the same transformation rule applies to tensors formed from bilinear combinations of Dirac fields. vector.3) (Of course.1 ~ 'P[ (x)y.2. and then using this reflection property to infer the existence of an antiunitary operator that induces CPT transformations on the fields.5). For a scalar. and 5. .PM'y.3.(x) ~ [. the phases " ~. or Dirac field the results of Sections 5.8. Applying Eq. A Hermitian scalar interaction density Jf(x) must be formed from • The original proofs of this theorem were by Liiders and Pauli.) If the bilinear is a tensor of rank n. . 5.1) (5. the product CPT of all the inversions is conserved.8.) We are going to choose the phases so that for all particles (5. (5. and 11 depend on the species of particle described by each field. let us work out the effect of the product CPT on free fields of various types.2) (5.5.8. (5.8.3) to such a bilinear gives CPT[iPJ(x)M'P2(X)][CPTJ.
>I'(x) [CPT]~I ~ . B) fields that can be constructed for finite mass. Thus the operator CPT.8.7) More generally (and somewhat more easily) we can see that the same is true for Hermitian scalars formed from the fields tp1tJ(x) belonging to one or more of the general irreducible representations of the homogeneous Lorentz group. such as the scalar and Dirac fields discussed in Sections 5. (5. and therefore CPT . (5..8..9) Also. The physical consequences of this symmetry principle have already been discussed in Sections 3.3.8.3)..>I'(x).7). and Bl + B2 +. in any theory CPT commutes with the freeparticle Hamiltonian H o. it is necessary that both AI + A2 +. there is no special problem in passing to the limit of zero mass. (5.9 Massless Particle Fields Up to this point we have dealt only with the fields of massive particles. let us attempt to construct a general free field for a massless particle as a linear combination of the annihilation operators a(p.) In order to couple together a product tp:/b/(X) tp:22b~2(X)'" to form a scalar £'(x).6.2 and 5. so (_1)28 1+28 2+'" = I. This peculiar limitation on field types will lead us naturally to the introduction of gauge invariance.8.3 and 3. bl (5. (5. (5.O): CPT V [CPT]~I ~ V .8. From Eq. For some of these fields. we saw in Section 5. acts on 'in' and 'out' states in the way described in Section 3. .O") for particles of momentum p and he1icity 0". and so a Hermitian scalar X(x) will automatically satisfy Eq.. In fact. be integers.7) it follows immediately that CPT commutes with the interaction V _ fd3xJt'(i.8. 5. Just as we did for massive particles.8) (For the Dirac field the factor (_1)28 is supplied the matrix "5 in Eq. Putting together our results in the previous section for the effects of inversions on such fields.5. On the other hand.3 that there is a difficulty in taking the zeromass limit of the vector field for a particle of spin one: at least one of the polarization vectors blows up in this limit.246 5 Quantum Fields and Antiparticles tensors with an even total number of spacetime indices. we shall see in this section that the creation and annihilation operators for physical massless particles of spin j > 1 cannot be used to construct all of the irreducible (A. which has been defined here by its operation on freeparticle operators. we find CPT tp::(x) [CPT)~I ~ (_I)'8tp::'(x).
1) + Aact(p..) eip ..20). (5.11) eiPx] where now pO ~ Ipl. PA ..9. " a.. (5.9. (2. I1)Vt(p.Ap. 11) for the antiparticles:· w(x) ~ (2rr)3/21 alp L " [Ka(p..1. The creation operators transform just like the oneparticle states in Eq.8(p.9.5) then we must take the coefficient functions u and v to satisfy the relations (5.5.(A)vr(p.(A I )'I'l(Ax) .I(A) ~ V(AP)O exp ( .A) ) a(PA.5..9.O(p.43).2) (5.i..42) (5..: are constant coefficients to be determined by the requirement of causality wIth some convenient choice of normalization of the coefficient functions Ut and V(..9.) As in the massive particle case.) exp (.9.. and drop the species label n.).6) Vl(PA.A)) ~ Yfi\PiO { (PO)O LD. we can satisfy these requirements by setting (in place of • We deal herc with only a single species of particle.7) in place of Eqs. Hence if we want the field to transform according to some representation D(A) of the homogeneous Lorentz group U(A)'I'r(xjUI(A) ~ L I = D.jU .) (5.1.5. Also. (2.9 Massless Particle Field!! 247 and the corresponding creation operators aC+(p.19) and (5. .4) where PA Ap.3) and hence also U(A)a(p.9. (5. and 0 is the angle defined by Eqs.)Ut(p. (5. (Again. po .nd .
22» "lip.9.O.(k.(k.9. W») ~ LDtt(W)"t(k.ioe(k.5.8) Vl(P.e.{J) in the x .27). S'.10) v.(a.9.(k.9.23) and (5. and !t(p) is a standard Lorentz transformation that takes a massless particle from momentum k to momentum p.9. (5.11) 0 0 1 ".(e)~ case sine 0 0 sine case 0 0 0 0 1 0 o we find from Eqs.o)e'U' ~ LDlt(R(e»)"t(k. (5. (5. W») ~ LD. Also. Ikl). 0) t (5.26).9.k).12) v.o) t (5.1. 1 0 a a 0 1 p P a p 1y I a P I 1+1 . in place of Eqs.o).(k.21) and (5.9.9.1.9. (5. i. R'.y plane.28). given by Eq.p) 1 ~ =(a' + P')/2.o)e'U' ~ LD?t(R(e»)v.o) exp (ioe(k. 0) ~ fikI ypo L t Dtt Ut'(p)) ". say (O.1. the coefficient functions at the standard momentum must satisfy ".5. given by (2. 0) . (2.9) where k is a standard momentum.9. (5.o) ~ J'kJ L.10) and (5. For a rotation R(lJ) by an angle IJ around the zaxis. .(k. We can extract the content of Eqs.24).5. (2. P Dl t C5i'(p))vt(k.10) and (5.1.o) t (5.o) exp ( ..(k.11) by considering separately the two kinds of littIe·group elements in Eq.o) t (5.13) For combined rotations and boosts S(IX.. (5.(W)v.11) where WI\ is an arbitrary element of the 'little group' for fourmomentum k = (k.248 5 Quantum Fields and Antiparticles Eqs.(k. an arbitrary Lorentz transformation that leaves this four momentum invariant.
let's try to construct the fourvector [(!. It is conventional to write the coefficient function u ll here in terms of a 'polarization vector' ell: U.u) _ (2 pO)1/2 e.(p. (5.8) and (5. (5.u) ~ I:Dit(S('.5.9.20) Eq.19) (5. we have simply D.18) Also.8) gives e'(p.16) The problem is that we cannot find a Ut that satisfies Eq.O. The equations for v are just the complex conjugates of the equations for u. (5.14) for general representations of the homogeneous Lorentz group.9. (5. we may normalize the coefficient functions so that Vt(p.1.9.'l'(p)'.15) Eqs.P))ut(k. which is impossible for general real (X. Eqs.9 Massless Particle Fields Eqs.(p.9. (5. In the fourvector representation.12) and (5.9. 0') eiu(l = R(lJ)/l vev (k. a) . Eqs. (5.17) so that Eq.9) then give them at arbitrary momenta.j2 .9.9. To see what goes wrong here.u).21) But then Eq.u) t (5.9. (5.9. (5.{3. (5.9.uv(A) = N'v .P))vt(k.9.e'(k.9.9. u) ~ Ut(p. u)' .+i.). e"(k.u) t (5. We therefore cannot satisfy the fundamental .12)(5.9. so with a suitable adjustment of the constants K and .O)/.u) ~ .20) would require also that (X + i{3 = 0.15) are the conditions that determine the coefficient functions u and v at the standard momentum k.10) and (5.9.u)..9. even for those representations for which it is possible to construct fields for particles of a given helicity in the case m =1= O. (5. (5.14) vi(k. (5.14) read here eJ'(k.9.+l) ~ (1.11) give 249 ui(k.9. !)] field for a massless particle of helicity +1.19) requires that (up to a constant which can be absorbed into the coefficients K and ). (5.9.u) ~ I:Du(S('.9.
25) (5.9.I~) k" }.+l) = 0 and k 'e(k.+l) is a purely spatial vector with only x and y components. (T ( e. Note that the Lorentz transfonnation 2(p) that takes a massless particle momentum from k to p may be written as a 'boost' iH(lpl) along the zaxis which takes the particle from energy Ikl to energy Ipl.+l) = 0 so eO(p.).9. Let's temporarily close our eyes to this difficulty. (W(B. P)R\(B)e'(k. (5. + 1) ~ R(fi)".+l)~O. + 1) ~ S")(.9.23) of course satisfies Da"(x) ~ O.9. Since e"(k. e'(k. + 1) ~ exp (+iBl { e"(k.22) We have thus come to the conclusion that no fourvector field can be constructed [rom the annihilation and creation operators for a particle of mass zero and helicity + 1.9. instead. (We shall need these properties of the polarization vector later when we come to quantum electrodynamics. (5.10).24) Other properties of the field follow from those of the polarization vector.250 5 Quantum Fields and Antiparticles requirement (5. +1) . (5.9.27) It follows that aO(x) ~ 0 (5. (1 )eiP ' Xa(p.26) and p·e(p. PI)e'(k. we have here D". it is unaffected by the boost along the zaxis. and go ahead anyway.21) to define a polarization vector for arbitrary momentum.23) We will come back later to consider how such a field can be used as an ingredient in a physical theory.9. and so e"(p..28) and v· .9. (5. The field (5.9. and take the field as a.14) or (5. using Egs. ~a)] . eO(k.9. (5.. In particular.9.9. (T) + e/. followed by a standardized rotation R(p) that takes the zaxis into the direction of p.18) and (5.9.(x) ~ J x d'p(2n)3/2(2P')1/2 L u=±l [e pep.+l)~O (5..29) .(p. (5.(x) ~ o. + 1) + (:'.ip ' xact (p.
9. (5.9. but are also invariant under the 'gauge' transformations al' + all + 01'0. because the extra tenn in Eq.30) so that under a general Lorentz transformation U(A)u.U(x. This is accomplished by taking the couplings of aJl to be of the form allP'. (5.31) where O(x.32) ~eHO(kPe'(k.34).+I)) This shows that the coefficient function that satisfies Eq. The fact that aO vanishes in all Lorentz frames shows vividly that aJl cannot be a fourvector.9.9. (5. fi») D' "( W(8. +1) . Note also that Eqs.22) and the invariance of k Jl under the little group we see immediately that D' p (W(8. (5.9.22) shows that for a general momentum p and a general Lorentz transformation /\. there is no problem in constructing an antisymmetric tensor field for such particles. invariant under formal Lorentz transfonnations under which al' + AJl v aV ). A). (5. +1) ~ i(2rr)3/2(2p o)3/2 [PPe'(p.9.9 Massless Particle Fields 251 As we shall see in Chapter 9. Instead.34) Note that this is a tensor even though all is not a fourvector.33) where el'(p. (5. we will be able to use a field like aJl(x) as an ingredient in Lorentzinvariant physical theories if the couplings of all(x) are not only formally Lorentzinvariant (that is. As we will see in more detail in Chapter 8. and (5.A) is a linear combination of annihilation and creation operators.9. (5.23) gives the general antisymmetric tensor field for massless particles of helicity +1 in the form (5. in place of Eq.9. >.28).9.6) we have (5.5.6) for the antisymmetric tensor representation of the homogeneous Lorentz group is (with an appropriate choice of normalization) ". Although there is no ordinary fourvector field for massless particles of helicity +1.9. (5.9. From Eq.9. whose precise form will not concern us here.1(A) ~ A' . these are the conditions satisfied by the vacuum vector potential of electrodynamics in what is called Coulomb or radiation gauge. Using this together with Eq. (5.(x)U.(p.34). (5.24).. (5.9.9. + I)) (5. (5.+I)k'e'(k.9.p' e"(p. >. + I) .25).a.9. (5.9. Eq. k" eP(k.31) drops out in Eq. +1) is given by Eq.29) show that f" satisfies the . Pl) (kP e"(k. where P' is a fourvector current with 8J1r = o.(Ax) + iJ. +1) ].
faster than the usual inversesquare law..) ~Jij112' '. fp.39) in which case since illv is a tensor the commutator also vanishes for all spacelike separations.9.9.9.. '" e(p.25). Eq.9. but gaugeinvariant theories that use vector fields for massless spin one particles represent a more . This is perfectly possible.)e1(p.38) dl p(2po)1 [1<I'e"< lll'e"<] yO if and only if This clearly vanishes for xO = 1<1 2 ~ Ill' (5.(y)t] = (2n)3 [~1JJ.39) also implies that the commutator of the all vanishes at equal times.. (5. and as we shall see in Chapter 8 this is enough to yield a Lorentzinvariant Smatrix. the fields are then Hermitian if the particles are their own chargeconjugates.)" ~J'ilkI2 c:r=±1 and so. using Eq.9. The relative phase of the creation and annihilation operators can be adjusted so that K = .. p'pi ~ G=±I (5. as is the case for the photon..9.34) means that an interaction density constructed solely from !/lV and its derivatives will have matrix elements that vanish more rapidly for small massless particle energy and momentum than one that uses the vector field aw Interactions in such a theory will have a correspondingly rapid falloff at large distances. (5.9.)e1(k... k'ki ~ e'(k.. Why should we want to use fields like aP(x) in constructing theories of massless particles of spin one..252 5 Quantum Fields and Antiparticles vacuum Maxwell equations: 0/l[IJV = 0.9.!pava.37) P A straightforward calculation gives then [f/lv(x).+ 'fl1'pO/J0(f x I + 'flIJI1OvOp ~ 17voOJiOp] (5.9.21) gives ". (5.36) To calculate the commutation relations for the tensor fields we need sums over helicities of the bilinears eJ'ev*.35) (5. rather than being content with fields like fIlV(x) with simple Lorentz transformation properties? The presence of the derivatives in Eq. (5.1.. The explicit formula (5.
aside from possible total derivative terms that do not affect the longrange behavior of the force produced:· The fields of massless particles of spin j > 3 would have to couple to conserved tensors with three or more spacetime indices.8i"·"i'N = 0. and the energymomentum four~vector. satisfying 0f/OIlV = O.(J/B))b'b Dala] .ab = i where J(j) are the angularmomentum matrices for spin j.ions. then d J x eO!'2"J'N is a conserved quantity that transforms like a tensor of rank N .ab = €ijk [(J~A))al<l r5b'b + (Jk(B))blb bua ] [(J~A))<l'<l bh'b . Parallel remarks apply to gravitons.. B)."1'1'/ is a tensor current satisfying Ji'. However.6. The problems we have encountered in constructing fourvector fields for helicities +1 or symmetric tensor fields for helicity +2 are just special cases of a more general limitation. in order to incorporate the usual inversesquare gravitational interactions we need to introduce a field h)lv that transforms as a symmetric tensor. would forbid all but forward colIis. any representation D(A) of the homogeneous Lorentz group can be decomposed into (2A + 1)(2B + 1)dimensional representations (A. this is achieved by coupling the field to a conserved 'current' Of/V.1. From the annihilation and creation operators for such particles we can construct a tensor Rllvp • with the algebraic properties of the RiemannChristoffel curvature tensor: antisymmetric within the pairs Il. . so highspin massless particles cannot produce longrange forces. up to gauge transformations of the sort associated in general relativity with general coordinate transformations. As we saw in Section 5.9 Massless Particle Fields 253 general class of theories. The only such conserved tensor is the energymomentum tensor. including those that are actually realized in nature. .V and p.rr. As in the case of electromagnetic gauge invariance.5. To see this. but aside from total derivatives there are none. C/IU)alb'. massless particles of helicity +2. now with two spacetime indices. Thus in order to construct a theory of massless particles of helicity ±2 that incorporates longrange interactions. For 8 infinites •• If 81'1 . The conservation of any other fourvector. or any tensor of higher rank. The only such conserved tensors are the scalar 'charges' associated with various continuous symmetries. f . let's consider how to construct fields for massless particles belonging to arbitrary representations of the homogeneous Lorentz group. it is necessary for it to have a symmetry something like general covariance. and symmetric between the pairs.. for which the generators of the homogeneous Lorentz group are represented by Clij)alb'.
0') B = (AA) + iJiA))adUa1b(k. and create antiparticles of helicity + 1 and . t1vab(k. so Eqs. 0= ($32 + .b./238.1. We can now see why it was impossible to construct a vector field for massless particles of helicity +1.(k. B)' can be formed only from the annihilation operators for a massless particle of helicity a and the creation operators for the antiparticle of helicity . the (1. + b )vab(k.1 and + 1 respectively. Also.0).J(B. (J) must vanish unless 0' = a + band (J' = a ." and fJ become infinitesimal in Eq.9.1) satisfy Eq. we see that a field of type (A.1 and antineutrinos helicity + 1 in this theory. By the same methods as in Section 5.iAB))wUab.9. (J) .U) vanishes unless a ~ A. 0") = (a + b)Uah(k. there is only a ( 1. (B' 2 (J 1 +1.9. (1) = (a and so uab(k.$02)ab.iJ1B))wuab.)) fields for massless particles of spin) (i.) h/l uab' (k) = 0 .0) and (0. respectively. It is easy to see that the fields for a massless particle of spin) of type (A.. b ~ +B (5.9.254 imal.(.a). letting. A vector field transforms according to .14) gives Cf 31 + .a) A = (Ji ) + iJiA))adUdb(k. D(R(8)) ~ 5 Quantum Fields and Antiparticles 1 + i.12) and (5.39).. respectively. (5.( J . The relative phase of the annihilation and creation operators may then be adjusted so that these coefficients are equal. Putting this together.9. (J) + (JiB) .B) are just the lAth or 2Bth derivatives of fields of type (0.i AA)) .(k..0") + (A ) .40) and the same is obviously also true of vab(k.a) = 0. 0= or more simply (AAl .a). a) and vab(k. (5.7. (5.41) For instance. respectively. In the 'twocomponent' theory of the neutrino. (1').101 )ab.I'1I Uu'b. rr) .e.U These require that Uab(k. helicity +)) commute with each other and their adjoints at spacelike separations if the coefficients of the annihilation and creation terms in Eq.. (5.(k.A + j) or (B + ). (5.9.13) give t1u ab(k.9.)) or (j." udb(k. so neutrinos have helicity .0) field and its adjoint.(k.dh' ua'h. it can be shown that the (j. where "~BA.0) and (0. so these more general fields do not need to be considered separately here. ~) parts of the Dirac field for a massless particle can only destroy particles of helicity ...
which multiply the annihilation operators a(p.(p). C.9. in such a way that the field transforms under Lorentz transformations like a Dirac field llV with an extra fourvector index 11.19) can only describe helicity zero. that is.Problems 255 the ( L 0 representation. 2): a fourth rank tensor which like the RiemannChristoffel curvature tensor is antisymmetric within each pair of indices and symmetric between the two pairs. Show that if the zeromomentum coefficient functions satisfy the conditions (5. possible to construct a vector field for helicity zero .19) and (5.K}.0) Ef:l (0.0) Ef:l (0.) The simplest covariant massless field for helicity + 1 has the Lorentz transformation type (1.23) and (5. Problems 1.21) and (5.'. (1). and hence according to Eq. Similarly. then the coefficient functions (5. defined by " + L u"(p.d(p). defined (for p2 = _m 2) by Lui(p.22) for arbitrary momentum satisfy the defining conditions Eqs.(T) in this field.1. in such a way that the field transforms under Lorentz transformations like a tensor. the simplest covariant massless field for helicity +2 has the Lorentz transformation type (2. What field equations and algebraic and reality conditions does this field satisfy? Evaluate the matrix pJlV(p).20). T given in the previous section can be carried over to the case of zero mass with only obvious modifications.1. (1) in this field. Show how to calculate the coefficient functions ui(p.1. (5. What field equations does this field satisfy? Evaluate the function PJl".(T). (5. Consider a free field hJlV(x) satisfying hJlV(x) = h"J'(x) and 1flJ'(x) = 0. 2.1). which multiply the annihilation operators a(p. it is an antisymmetric tensor fJlv. What are the commutation relations of this field? How does the field transfonn under the inversions P. which annihilates and creates a particle of spin two and mass m O. " . u) =(2 pO)1 p.(T) =(2pO)lp~~(p). The discussion of the inversions P.(T)u~·(p. T? 3.24).just take the derivative a/A) of a massless scalar field ¢.1. Consider a free field tpi(x) which annihilates and creates a selfchargeconjugate particle of spin ~ and mass m f.1. of course. Show how to calculate the coefficient functions UJ'I'(p.O. u)ud'(p.1. (It is. C.
1979)).0). 181.O) + (0. j) or (j. . B882 (1964). Cartan. 12 (1933) (translation in Selected Papers of Leon Rosenfeld.0) + (0. T? 4. Work out the transformation properties of fields of transformation type (j. Kgl. Show that the fields for a massless particle of spin j of type (A. 138.m and definite helicity? (Assume that the Fourier transform of the current has values for different a. France 41. Danske Vidensk. Consider a generalized Dirac field 1p that transforms according to the (j. List the tensors that can be formed from products of the components of'P and 'Pt. M."(x)]. Suppose it has an interaction Hamiltonian of the form V ~ J d]x [>P. Phys.256 5 Quantum Fields and Antiparticles What are the commutation relations ofthis field? How does the field transform under the inversions P. Dordrecht. E. Selskab Mat. 78. Rosenfeld.. by R. where Jab is an external cnumber current. A similar approach has been followed in unpublished lectures by E. N. Rev. Rev. respectively. 5. 610 (1928). 4. 794 (1950).b(X)}'b(x) + rbl(x)>p. Dirac. Proc. Bohr and L. B1318 (1964). Check your result against what we found for j = t. Soc. 6.A + j) or (B + j. b that are of the same order of magnitude. Soc. A. 134. C. ed. j) for massless particles of helicity +j under the inversions P. B988 (1965).)) representation of the homogeneous Lorentz group. Weinberg. The point of view adopted in this chapter was presented in a series of papers: S. and that do not depend strongly on E. Math. Phys. Roy. B) representation of the homogeneous Lorentz group. No. Wichmann. What is the asymptotic behavior of the matrix element for emitting these particles for energy E >. 2. Consider a general field 'Pab describing particles of spin j and mass m =1= 0. C. Medd. 3. P. 1893 (1969). T. (London) A1l7.) References 1. 7. 53 (1913). Cohen and J.Fys. B) are the 2Ath or 2Bth derivatives of fields of type (0. Stachel (Reidel. S. that transforms according to the (A. 133. Bull.
. Kong. Liiders. Rarita and J. 61 (1941). 807 (1958).. S. e. 6. Phys. Argyres and C. Phys. Die gruppentheoretische Methode in der Quantenmechanik (Springer Verlag. Nuavo Cimenta 6. Rev. Pauli. 89 (1989). Nonperturbative proofs in axiomatic field theory were given by G. and A. G. Jauch and F. 58. 188. R. Mat. Weinberg. Rev. MA. Phys. A. H Georgi. VelD and D. A. W.j) representation were introduced by H Joos. Sitz. Dedijer (Pergamon Press. The original reference is L Schur. Phys. Medd. D. See. New York. 1960). D 40. Nappi. 16. New York. and J. Zumino. E. A. 406 (1905). Lett. 1095 (1989). 12. 15. 8. 28. R". B224. by T. 10. e. See. Rev. S. e. Rev. Spin & Statistics. 60. peT. Schwinger. 1955): Appendix A2. in Proceedings of the Fifth Coral Gables Conference on Symmetry Principles at High Energy. Dansk. Section V. Nuovo Cimento 8. Acta 12. Zwanziger. When . Rev. Burgoyne.. Elementary Theory ofAngular Momentum (John Wiley & Sons. 104. Gudehus. 186. MA. and All That (Benjamin. Phys. Lyubarski. New York. 9. Rohrlich. van der Waerden. 1968). Lee and C. 1 (1957). 1932). 716 (1940). p. The Applications of Group Theory in Physics. 1450 (1958) and N. Fierz. 7. G. M. 110. Selskab. 1969). Wightman. Phys. 1982): pp. translated by S. R. 13. 181. Phys. 0) + (O. L. F. Phys. C. 2927 (1970).g. 133. e. W. G. R. Witten. 15. Perlmutter (Gordon and Breach. 14. 1893 (1969). Phys. Also see R. W. Vid. Princeton. See. N. Weinberg. Lie Algebras in Particle Physics (Benjamin/Cummings. Rev. 65 (1962). The Theory of Photons and Electrons (AddisonWesley. Seiler.Fys. Wightman.. R". 1957): Chapter III . Schroer. Rose. Ref. 1337 (1969). 11. 17. Phys. (Princeton University Press. ed. Phys.. New York. 254 (1956). 10. Ann. Yang. Streater and A. Reading. Phys.References 257 5. S. Angular Momentum in Quantum Mechanics. T. M. 2218 (1969). 198.g. Edmonds. G. Cambridge. 1957): Chapter 3. J. 5 (1954). D 2. A. 3 (1939). R. Edmonds. Preuss. 10. Rev. P. B. B. 204 (1957). Phys. and other references quoted therein. Fortschr. See.g.g. Pauli. Swieca. Liiders and B. C. Fields in the (j. BB18 (1964). Akad. or M. 2. S. Nappi and L. Kaiser. R. Ya. Berlin. Helv.
409 (1957). so his theorem stated that C conservation is equivalent to T invariance. 12. Phys. 110. F. Acta 30. Rev. . Phys. it was still taken for granted that P is conserved. Ref. R. 579 (1958).258 5 Quantum Fields and Antiparticles Luders first considered how the inversions are related. J. Jost. Also see Streater and Wightman. Dyson. Helv. 18.
we shall use the approach described by Dyson! in 1949. Schwinger. This chapter will outline the diagrammatic calculational technique first described by Feynman at the POCODOS Conference in 1948.1 Derivation of the Rules Our starting point is a formula for the Smatrix. 6.2) for the freeparticle states: 259 .2. Feynman was led to these diagrammatic rules in part through his development of a pathintegral approach. which until the 1970s was the basis of almost all analyses of perturbation theory in quantum field theory.6 The Feynman Rules In previous chapters the use of covariant free fields in the construction of the Hamiltonian density has been motivated by the requirement that the Smatrix satisfy Lorentz invariance and cluster decomposition conditions. described at the beginning of Section 3. Nevertheless.10) with expression (4. and still provides a particularly transparent introduction to the Feynman rules. obtained by putting together the Dyson series (3. in which Lorentz invariance and cluster decomposition properties are transparent throughout. With the Hamiltonian density constructed in this way. there are obvious practical advantages in using a version of perturbation theory in which the Lorentz invariance and cluster decomposition properties of the Smatrix are kept manifest at every stage in the calculation. and Tomonaga in the late 1940s was to develop perturbative techniques [or calculating the Smatrix. it makes no difference which form of perturbation theory we use to calculate the S·matrix.5.5. the results will automatically satisfy these invariance and clustering conditions in each order in the interaction density. which will be the subject of Chapter 9. This was not true for the perturbation theory used in the 1930s. The great achievement of Feynman. now known as 'oldfashioned perturbation theory'. In this chapter.
which destroy antiparticles and create particles. freeparticle vacuum state. spin. taken as a polynomial in the fields and their adjoints J!"(x) ~ Lg.3).n') e. cr.1.2) each tenn *"1 being a product of definite numbers of fields and field adjoints of each type. and n label particle momenta.260 6 The Feynman Rules As a reminder: p.o.. T indicates a timeordering. a and at are annihilation and creation operators. .5. (For instance. The field of a particle of species n that transforms under a particular representation of the homogeneous Lorentz group (with or without space inversions) is given by w(x) ~ L(2n)l/2 q J alp [Ut(p.(x) in an order in which the arguments xO decrease from left to right. as well as including a running index labelling the components in this representation..(x). in the scalar field the UI for a particle of energy E is simply (2E)1/2. and exp(+ip· x) is calculated with pO set equal to P2 + m~.n)e.CT. Of course.1. (6. and species.1.o. the derivative of a field (6. The coefficient functions UI and VI depend on the Lorentz transformation properties of the field and the spin of the particle it describes. and those we call 'antiparticles'. such as positrons and antiprotons. We will here make a distinction between some particle species that we arbitrarily call 'particles'.1. V .3) is just another field described by (6. primes denote labels for particles in the final state.J!". their adjoints. which puts the *. and *.n) a(p. (6.n )aI(p.P<] .cr. while in a Dirac field UI and VI are the normalized Dirac spinors introduced in Section 5. with different UI and VI.3) Here nC denotes the antiparticle of the species n.. from our point of view. lbo is the. etc. are called 'field adjoints'. protons. for instance electrons. they were calculated in Chapter 5. There is no need to deal separately with interactions that involve derivatives of fields. The field operators that destroy particles and create antiparticles are called simply 'fields'.(x) is the interaction Hamiltonian density." + VI (p.) The index t on the field should here be understood to indicate the particle type and the representation of the Lorentz group by which the field transfonns.
6.1 Derivation of the Rules
261
some particle species like the photon and nO are their own antiparticles; for these the field adjoints are proportional to the fields. We now proceed to move all annihilation operators to the right in Eq. (6.1.1), repeatedly using for this purpose the commutation or anticommutation relations:
a(p rr n)at(p'rr'n' ) = +at(p'rr'n')a(p rr n)
H'(p' p)o.,.o"
a(p a n)a(p I rr' n') = +a(p'rr' n')a(p rr n) at(p rr n)at(p'rr'n') = +at(p'rr'n')at(p rr n)
(6.1.4) (6.1.5)
(6.1.6)
(and likewise for antiparticles), the + sign on the right being  if both particles n, n' are fermions, and + if either or both are bosons. Whenever an annihilation operator appears on the extreme right (or a creation operator on the extreme l~ft), the corresponding contribution to Eq. (6.1.1) vanishes, because these operators annihilate the vacuum' state:
a(pan)<I>o~O,
(6.1.7)
<l>lat(p a n) ~ O.
(6.1.8)
The remaining contributions to Eq. (6.1.1) are those arising from the delta function tenus on the righthand side of Eq. (6.1.4), with every creation and annihilation operator in the initial or final states or in the interaction Hamiltonian density paired in this way with some other annihilation or creation operator. In this way, the contribution to Eq. (6.1.1) of a given order in each of the terms Jt'i in the polynomial Jt"(tp(x),tpt(x)) is given by a sum, over all ways of pairing creation and annihilation operators,2 of the integrals of products of factors, as follows: (a) Pairing of a final particle having quantum numbers p',a',n ' with a field adjoint tpj(x) in Jt'i(X) yields a factor
[a(pl a'n'), tp;(x)L = (2n)3/2eiP·Xut(plrr'n').
(6.1.9)
(b) Pairing of a final antiparticle having quantum numbers pi, ai, n'c with a field tp{(x) in Jt"j(x) yields a factor
[a(p'rr' n' C), tpr(x)] 1= = (2n )3/2 eip'XVt(p Irr ' n') .
(6.1.10)
(c) Pairing of an initial particle having quantum numbers p,rr,n with a field tpAx) in Jt"i(X) yields a factor [w(x),at(pan)lf ~ (2n)'f2 e'P·'ur(pan).
(6.1.11)
262
6 The Feynman Rules
(d) Pairing of an initial antiparticle having quantum numbers p,6,ne with a field adjoint tp;(x) in Jf'i(X) yields a factor ['I')(x),at(p" n')] + ~ (2n)3/2 e'p·,v;(P" n).
(6.1.12)
(e) Pairing of a final particle (or antiparticle) having numbers p',u',n' with an initial particle (or antiparticle) having quantum numbers p,O",n yields a factor
[a(p '(1' n'), at (p
(1
n)] =F = 03(p I

p)b".'u(),.l n .
(6.1.13)
(f) Pairing of afield 1p{(x) in Jfj(x) with a field adjoint tpl(y) in ,}f'j(Y) yields a factor*
6(x  y) ['I'i(x), 'I'~t (y)] + + O(y  x) ['I';t(y), 'l't(x)] +
 j ~rm(x,y),
(6.1.14)
where tp+ and tp are the terms in tp that destroy particles and create antiparticles, respectively:
'Pt(x)
=
(211:)3/2
'Pt"(x) =
f (2n)~3!2 f
d3p
L Ut(p
"
(1
n) eip'xa(p
(J
n),
(6.1.15) (6.1.16)
d3p LUt(p
,
(J
n) eiP·Xat(pO"nC).
Recall that (I(x  y) is a step function, equal to + 1 for xO > yO and zero for xO < yO. These step functions appear in Eq. (6.1.14) because of the timeordering in Eq. (6.1.1); we can encounter a pairing of an annihilation field tp+(x) in Jt'(x) with a creation field tp+t(y) in Jt'(y) only if .Jf'(x) was initially to the left of £'(y) in Eq. (6.1.1), i.e., if xO > yO; similarly, we encounter a pairing of an annihilation field tpt(y) in Jr(y) with a creation field tp(x) in £'(x) only if .Jf'(y) was initially to the left of .Jf'(x) in Eq. (6.1.1), i.e., if yO> xO. (The + sign in the second tenn in (6.1.14) will be explained a little later.) The quantity (6.1.14) is known as a propagator; it is calculated in the following section. The Smatrix is obtained by multiplying these factors together, along with additional numerical factors to be discussed below, then integrating over Xl ... XN, then summing over all pairings, and then over the numbers of interaction of each type. Before filling in all the details, it will be convenient first to describe a diagrammatic formalism for keeping track of all these pairings.
• If the interaction J!"(x) is written in the normalordered form. as in Eq. (5.1.33), then there is no pairing of fields and field adjoints in the same interaction. Otherwise some sort of regulari7.ation is needed to give meaning to A(m(O).
6.1 Derivation of the Rules
~,
263
po n
, ,
p 0 n
~,
, 'c
e ip'.,
ul(pcr'n')
e,pxv,Jpon)
(21t)31l
'.
~"
,
e'p·, uf(pon)
(21t) 312
.
~
(21t)31l
£.x
Lx
(bj
(ej
f,x
p
~,
0
,
n
,
po n
~,
, 'c
e ip . x v}(Pon)
(21t)31l
~
p
on
(ej
(tij
)0
•
m,y
.(
iAfm (x ,y)
Figure 6.1. Graphical representation of pamngs of 0rerators arising in the coordinatespace evaluation of the Smatrix. The expressions on the right are the factors that must be incluued in the coordinatespace integrand of the Smatrix for each line of the Feynman diagram.
The rules for calculating the Smatrix are conveniently summarized in terms of Feynman diagrams. (See Figure 6.1.) The diagrams consist of points called vertices, each representing one of the Jri(x), and lines, each representing the pairing of a creation with an annihilation operator. More specifically: (a) The pairing of a final particle with a field adjoint in one of the Jr(x)
264
6 The Feynman Rules
is represented by a line running from the vertex representing that Jf'(x) upwards out of the diagram, carrying an arrow pointed upwards. (b) The pairing of a final antiparticle with a field in one of the Jf'(x) is also represented by a line running from the vertex representing that .Jf'(x) upwards out of the diagram, but carrying an arrow pointed downwards. (Arrows are omitted throughout for particles like y, nO, etc. that are their own antiparticles.) (e) The pairing of an initial particle with a field in one of the Jf'(x) is represented by a line running into the diagram from below, ending in the vertex representing that £(x), carrying an arrow pointed upwards. (d) The pairing of an initial antiparticle with a field in one of the .Jf'(x) is also represented by a line running into the diagram from below, ending in the vertex representing that Jf'(x), but carrying an arrow pointed downwards. (e) The pairing of a final particle or antiparticle with an initial particle or antiparticle is represented by a line running clear through the diagram from bottom to top, not touching any vertex, with an arrow pointed upwards or downwards for particles or antiparticles, respectively. (f) The pairing of a field in Jf'(x) with a field adjoint in £(y) is represented by a line joining the vertices representing .Jr(x) and £(y), carrying an arrow pointing from y to x. Note that arrows always point in the direction a particle is moving, and opposite to the direction an antiparticle is moving. (As mentioned above, arrows should be omitted for particles like photons that are their own antiparticles.) The arrow direction indicated in rule (f) is consistent with this convention because a field adjoint in .Jfiy) can either create a particle destroyed by a field in Jf'r(x), or destroy an antiparticle created by a field in £j(x). Note also that since every field or field adjoint in Jf'j(x) must be paired with something, the total number of lines at a vertex of type i, corresponding to a tenn ,)f'i(X) in Eq. (6.1.2), is just equal to the total number of field or field adjoint factors in .Jf'j(x). Of these lines, the number with arrows pointed into the vertex or out of it equals the number of fields or field adjoints respectively in the corresponding interaction term. To calculate the contribution to the Smatrix for a given process, of a given order N; in each of the interaction terms £';(x) in Eq. (6.1.2), we must carry out the following steps: Draw all Feynman diagrams containing N j vertices of each type i, and containing a line coming into the diagrams from below for each particle or antiparticle in the initial state, and a line going upwards out of the diagram for every particle or antiparticle in the final state, together
0)
6.1 Derivation of the Rules
265
with any number of internal lines running from one vertex to another, as required to give each vertex the proper number of attached lines. The lines carry arrows as described above, each of which may point upwards or downwards. Each vertex is label1ed with an interaction type i and spacetime coordinate xIJ • Each internal or external line is labelled at the end where it runs into a vertex with a field type t (corresponding to the field 1p((x) or 'P;(x) that creates or destroys the particle or antiparticle at that vertex), and each external line where it enters or leaves the diagram is labelled with the quantum numbers p, a, n or pi, ai, n' of the initial or final particle (or antiparticle). (ii) For each vertex of type i, include a factor i (from the (~i)N in Eq. (6.1.1)) and a factor gi (the coupling constant multiplying the product of fields in Jf'j(x)). For each line running upwards out of the diagram, include a factor (6.1.9) or (6.1.10), depending on whether the arrow is pointing up or down. For each line running from below into the diagram, include a factor (6.1.11) or (6.1.12), again depending on the arrow direction. For each line running straight through the diagram include a factor (6.1.13). For each internal line connecting two vertices include a factor (6.1.14). (iii) Integrate the product of all these factors over the coordinates Xl> X2, .•• of each vertex. (iv) Add up the results obtained in this way from each Feynman diagram. The complete perturbation series for the Smatrix is obtained by adding up the contributions of each order in each interaction type, up to whatever order our strength permits. Note that we have not included the factor liN! from Eq. (6.1.1) in these rules, because the time~ordered product in Eq. (6.1.1) is a sum over the N! permutations of X1X2· ··XN, each permutation giving the same contribution to the final result. To put this another way, a Feynman diagram with N vertices is one of N! identical diagrams, which differ only in permutations of the labels on the vertices, and this yields a factor of N! which cancels the liN! in Eq. (6.1.1). (There are exceptions to this rule, discussed below.) For this reason, henceforth we do not include more than one of a set of Feynman diagrams that differ only by relabelling the • vertices. ~ In some cases, there are additional combinatoric factors or signs that must be included in the contribution of individual Feynman diagrams: (v) Suppose that an interaction £j(x) contains (among other fields and field adjoints) M factors of the same field. Suppose that each of these fields is paired with a field adjoint in a different interaction (different for each one), or in the initial or final state. The first of these field adjoints
266
6 The Feynman Rules
Figure 6.2. EXample of a graph requiring extra combinatoric factors in the Smatrix. For an interaction involving, say, three factors of some field (as well as other fields) we usualIy include a factor 1/31 in the interaction Hamiltonian, to cancel factors arising from sums over ways of pairing these fields with their adjoints in other interactions. But in this diagram there are two such factors of 1/3! and only 3! different pairings, so we are left with an extra factor of 1/3!.
can be paired with anyone of the M identical fields in Xj(x); the second with anyone of the remaining M  1 identical fields; and so on, yielding an extra factor of M!. To compensate for this, it is conventional to define the coupling constants gi so that an explicit factor 1/ M! appears in any ,*,j(x) containing M identical fields (or field adjoints.) For instance, the interaction of Mth order in a scalar field ¢(x) would be written g¢M / M!. (More generally, one often also displays an explicit factor of 11M! when the interaction involves a sum of M factors of fields from the same symmetry multiplet, or when for this or any other reason the coupling coefficient is totally symmetric or antisymmetric under permutations of M boson or fermion fields.) However, this cancellation of M! factors is not always complete. For instance, consider a Feynman diagram in which the M identical fields in one interaction £j(x) are paired with M corresponding field adjoints in a single other interaction £j(Y)' (See Figure 6.2.) Then by following the above analysis, we find only M! different pairings (since it makes no difference which of the field adjoints we call the first, second, ... ), cancelling only one of the two factors of l/M! in the two different interactions. In this case, we would have to insert an extra factor of 11M! 'by hand' into the contribution of such a Feynman diagram. Other combinatoric factors arise when some of the permutations of vertices have no effect on the Feynman diagram. We noted earlier that the factor liN! in the series (6.1.1) is usually cancelled by the sum over the N! diagrams that differ only in the labelling of the N vertices. However, this cancellation is incomplete when relabelling the vertices does not yield a new diagram. This happens most commonly in the calculation of vacuumtovacuum Smatrix elements in a theory with a quadratic interaction £ = lpJMItI1.ptl, where M may depend on external fields. (The physical significance of such vacuum fluctuation diagrams is discussed in detail in Volume II.) The Feynman diagram of Nth order in ft is a
6.1 Derivation of tl1e Rules
267
ring with N corners. (See Figure 6.3.) There are only (N 1)1 different diagrams here because a permutation of labels that moves each label to the next vertex around the ring yields the same diagram. Hence such a graph is accompanied with a factor
(NI)!
N! N (vi) In theories involving fermion fields, the use of Eqs. (6.1.4)(6.1.6) to move annihilation and creation operators to the right and left introduces minus signs into the contribution of various pairings. To be specific, we get a minus sign wherever the permutation of the operators in Eq. (6.1.1) that is required to put all paired operators adjacent to one another (with annihilation operators just to the left of the paired creation operators) involves an odd number of interchanges of fermion operators. (This is because to compute the contribution of a certain pairing, we can first permute all operators in Eq. (6.1.1) so that each annihilation operator is just to the left of the creation operator with which it is paired, ignoring all commutators and anticommutators of unpaired operators, and then replace each product of paired operators with their commutators or anticommutators.) One immediate consequence is to produce the minus sign in the relative sign of the two terms in Eq. (6.1.14) for the fermion propagator. Whatever permutation puts the annihilation part lP+(X) of a field in X(x) just to the left of the creation part lp+tCr) of a field adjoint in Jf{v), the permutation that puts the annihilation part lJ)t(y) of the field adjoint just to the left of the creation part tp(x) of the field involves one extra interchange of fermion operators, yielding the minus sign in the second term of Eq. (6.1.14) for fermions. In addition, minus signs can rise in the contribution of whole Feynman diagrams. As an example, let us take up a theory in which the sole interaction of fermions takes the form
(6.1.18) .""(x) ~ 2..:gtmk~,)(x)"'m(X).pk(X), tmk where gtmk are general constants, w(x) are a set of complex fermion fields, and <Pm(x) are a set of real bosonic (but not necessarily scalar) fields. (Not only quantum electrodynamics, but the whole 'standard model' of weak, electromagnetic, and strong interactions, has fermionic interactions that can all be put in his form.) Let us first take up the process of fermionfermion scattering, 12 _ 1'2', to second order in ,1{'. The fermion operators in the secondorder term in Eq. (6.1.1) appear in the order (with obvious abbreviations)

1
(6.1.17)
a(2')a( I ')~ t (x )",(x IV' t (y )v,(y)at (l)a t (2) .
(6.1.19)
There are two connected diagrams to this order, corresponding to the
268
6 The Feynman Rules
Figure 6.3. An eighthorder graph for the vacuumtavacuum amplitude with particles interacting only with an external field. In this diagram the external field is represented by wiggly lines. There are 7! such diagrams, differing only by relabelling the vertices, and not counting as different those labellings that simply rotate the ring. The factor 1/8! from the Dyson formula (6.1.1) is therefore not entirely cancelled here, leaving us with an extra factor 1/8.
pamngs
(6.1.20)
and
(6.1.21)
(See Figure 6.4.) To go from (6.1.19) to (6.1.20) requires an even permutation of fermionic operators. (For instance, move tp(x) past three operators to the right, and then move a(l') past one operator to the right.) Thus there is no extra minus sign in the contribution of the pairing (6.1.20). This in itself is not so important; the overall sign of the Smatrix does not matter in transition rates, and in any case depends on sign conventions for the initial and final states. What is important is that the contributions of pairings (6.1.20) and (6.1.21) have opposite sign, as can be seen most easily by noting that the only difference between these two pairings is the interchange of two fermionic operators, a(l') and a(2'). In fact, this relative minus sign is just what is required by Fenni statistics: it makes the scattering amplitude antisymmetric under the interchange of particles I' and 2' (or 1 and 2). However, it must not be thought that all sign factors can be related in such a simple way to the antisymmetry of the final or initial states, even in the lowest order of perturbation theory. To illustrate this point, let's now consider fermionantifermion scattering, 12c + 1'2\ to second order in the same interaction (6.1.18). The fermionic operators in the secondorder
6.1 Derivation of the Rules
2' 2'
269
2
2
Figure 6.4. The connected secondorder diagrams for fermionfermion scattering in a theory with interaction (6.1.18). Here straight lines represent fermions; dotted lines are neutral bosons. There is a minus sign difference in the contributions of these two diagrams, arising from an extra interchange of fermion operators in the pairings represented by the second diagram. term in Eq. (6.1.1) appear in the order:
a(1")a( 1')'1' t (x )'!'(xlv' t (y ),!,(y)at (I)a t (2')
(6.1.22)
Here again there are two Feynman diagrams to this order, corresponding to the pairings
[a(1")'I'(x)] [a(I')'I"(x)] [,!,(y)at(I)] [,!,'(y)a t (2')] (6,1.23)
and
(6,1.24)
(See Figure 6.5.) To go from (6.1.22) to (6.1.23) requires an even permutation of fermionic operators (for instance, move tp(x) past two operators to the left and move tpt(y) past two operators to the right) so there is no extra minus sign in the contribution of the pairing (6.1.23). On the other hand, to go from (6.1.22) to (6.1.24) requires an odd permutation of fermionic operators (the same as for (6.1.23), plus the interchange of tpt(x) and tpt(y)) so the contribution of this pairing does come with an extra minus sign." Additional signs are encountered when we consider contributions of higher order. In theories of the type considered here, in which the interactions of fennions all take the form (6.1.18), the fermion lines in
•• Actually, this sign is not wholly unrelated to the requirements of Fermi statistics. The same field can destroy a particle and create an antiparticle, so there is a relation, known as 'crossing symmetry', between proces~s in which initial particles or antiparticle, are exchanged with final antiparticles or particles. In particular the amplitudes for the process 12c > 1'2'c are related to those for the 'crossed' process 12' > 1'2; the two pairings (6.1.23) amI (6.1.24) just correspond to the two diagrams for this process, which differ by an interchange of 1 and 2' (or l' and 2), so the antisymmetry of the scattering amplitude under interchange of initial (or final) particles naturally requires a minus sign in the relative contribution of these two pairings. However, crossing symmetry is not an ordinary symmetry (it involves an analytic continuation in kinematic variables) and it is difficult to use it with any preciSIOn for general processes.
270
6 The Feynman Rules
l'
2"
"/
, ,
2'
Figure 6.5. The connected secondorder diagrams for fermionantifermion scattering in a theory with interaction (6.1.18). Here straight lines represent fermions or antifermions, depending on the arrow direction; dotted lines are neutral hosans. There is again a minus sign difference in the contributions of these two diagrams, arising from an extra interchange of fermion operators in the pairings represented by the second diagram.
general Feynman diagrams form either chains of lines that pass through the diagram with arbitrary numbers of interactions with the boson fields, as in Figure 6.6, or else fermionic loops, like that shown in Figure 6.7. Consider the effect of adding a fermionic loop with M corners to the Feynman diagram for any process. This corresponds to the pairing of fermionic operators
(6.1.25)
On the other hand, these operators appear in Eq. (6.1.1) in the order
(6.1.26)
To go from (6.1.26) to (6.1.25) requires an odd permutation of fermionic operators (move J{l(Xt} to the right past 2M  1 operators) so the contribution of each such fermionic loop is accompanied with a minus sign. These rules yield the full Smatrix, including contributions from processes in which various clusters of particles interact in widely separated regions of spacetime. As discussed in Chapter 4, to calculate the part of the Smatrix that excludes such contributions, we should include only connected Feynman diagrams. In particular, this excludes lines passing clean through the diagram without interacting, which would yield the factors (6.1.13). To make the Feynman rules perfectly clear, we will calculate the loworder contributions to the Smatrix for particle scattering in two different theories.
6.1 Derivation of the Rules
271
, 2'
,
,
,
,
,
''r'>~ ,
,
,
,,
',2
,,
,,
,,
'2
Figure 6.6. Thc connected secondorder diagrams for bosonfermion scattering in <l theory with interaction (6.1.18). Straight lines are fermions; dashed lines are neutral bosons.
,, ,,
,
, r+( ,
,
, ,,
,
,,
,
Figure 6.7. The lowestorder connected diagmm for boson boson scattering in a theory with interaction (6.1.18). Such fermion loop graphs yield an extra minus sign, arising from permutations of the paircd fermion fields.
Theory I
Consider the theory of fermions and selfchargeconjugate bosons with interaction (6.1.18). The lowestorder connected diagrams for fermionboson scattering are shown in Figure 6.6. Following the rules outlined in Figure 6.1, the corresponding Smatrix element is
(2n)6
L
k'I'm' kim
(i)2 gl 'm'k'
gmlk
Ui'(Pl
'oi nil) U/(Pl OlntJ
X./ tf"x
X
J
tf"y (  i6. m 'm(Y  x))eiP'I'YeiPI'X
[eiP;'Y Uk'(P2 'O2,n;)e iP2X Uk(P2<"T2n2) eiP;XUk(P2'lT2nS)eiP2'Yudp2lT2n2)]
(6,1,27)
+
272
6 The Feynman Rules
(The labels 1 and 2 are used here for fermions and bosans, respectively.) For fermionfermion scattering there are also two secondorder diagrams, shown in Figure 6.4. They yield the Smatrix element
SPI '".;/1'1 Pl'lT;"; , PI rTl 111 P2 (T2 'J2 =
(2n)6
L
k'I'm'klm
(_i)2 g l/JlmJel g/'Ik
XU~'(P2
x
J J
d4 x
I"; nz) ui'(PI' 0"1 nil) Um(P2 0"2 nz)u/(pJ ITI nl)
cry
eiP~'XeiP'I'YiP2'XejP1"Y(i)D.klk(X 
y)
 [1' ""
2'J
(6.1.28)
with the last term indicating subtraction of the preceding term with interchange of particles I' and 2' (or equivalently 1 and 2). There are no secondorder graphs for bosonboson scattering in this theory; the lowestorder graphs are of fourth order, such as that shown in Figure 6.7. More specific examples of formulas like Eqs. (6.1.27) and (6.1.28) will be given in Section 6.3, after we have had a chance to evaluate the propagators and go over to momentum space. .m de /D/.:YQCU'OH (aLkf'yd//dep~Dge.JYH.?pd: {de (dfl;t';" .kAff Off all different. It is instructive afso to look at an example with a trilinear interaction in which all three fields are the same, or at least enter into the interaction in a symmetric way.
Theory II
Now take the interaction density to be a sum of terms that are trilinear in a set of real bosonic fields ¢t(x):
1 J!"(x) ~ ]I
L gtm,q>{(x)q>m(x)q>,(x)
(6.1.29)
. tmn
with gtmn a real totally symmetric coupling coefficient. Suppose we want to consider a scattering process 1 2 _ 1'2' to second order in this interaction. Each of the two vertices must have two of the four external lines attached to it. (The only other possibility is that one of the external lines is attached to one vertex and three to the other vertex, but the vertex with three external lines attached to it would have no remaining lines to connect it to the other vertex so this would be a disconnected contribution.) The additional line required at each vertex must then just serve to connect the two vertices to each other. There are three graphs of this type, differing in whether the other external line that is attached to the same vertex as line I is line 2 or I' or 2'. (See Figure 6.8.) Following the rules given above, the contribution to the Smatrix from those three
. if the bosons in this theory are spinless particles of a single species. + P2)' x) exp(i(pl + P2)' Y) p'd' x) exp(i(pl .(P10"1nl) e iP 2'Y .P2. . .1. .. . The connectedsecond order diagrams for bosonboson scattering in a theory with interaction (6.6. .1..><.1 Derivation of the Rules j' 273 . ')< .y) x [exp(i(P.. P. )e iP1 "x x U~'(P'. . .(p 21'T2n2) eip. n. L gtt't" gmm'm" / d'x Jd'y(  ill.8. ..1. 'x .(x. 1 . . " ../ 1 "  . . [ui(p. . . .29) in the form (6. x Ut(P11'T1 n. . (p 20"2 n2) eip../ " ". . " .. ..) e'P(X Ut(PtO"l nj)e'PP x U~I(p20"2n2) eip~'y U..(x . then we write the interaction (6. 0".l' Yl]. .... Figure 6.. . . E2E1E1 J Jd'y d'x II. .. .y)) X Um(PI 0"1 nI )e iPl Y Um'(P20"2n2) e iP2 'Y + ui.) eip. 0".P1P2 ig' ~ (2n)6 J16E. . " " . (6. . n. .. . .... . . n~) eip'l 'x u. ' + ui. .P2)' Y) + exp(i(pl  +exp(i(p.p~l' xl exp(i(P2  p'. 'y Um(P10"2n2) eiP2'Y].1.31) and the Smatrix element (6.Plalnl P2 a 2n2 I" ~ (i)'(2n)6 X tt'{"mm'm" .30) To be even more specific." . ".30) for scalarscalar scattering is s' . " " . . diagrams is Sill Plal"1 P2a2n2 .29).1. . 0".(p.
) . the top and bottom signs refer to bosonic and fermionic fields.6) . It is absent in the pairing ofW(x) with ij)(y) . f' Prm ( p. If W(x) and tpm(y) are Dirac fields for a particle of spin Ptm(P) then (625) ~ [(iy"p" +m)PL. an essential ingredient in the Feynman rules that arises in the pairing of a field W(x) with a field adjoint 'P~(y). we showed that LUt(pan)u~(pun)~(2J p2+ m .) IfW(x) and lpm(Y) are vector fields VIl (:\:) <lnd Vv(y) for a particle of spin one. we have immediately idtm(x.1) " In the course of calculating commutators and anticommutators in Chapter 5. Inserting Eqs.1.2. There are no terms of third order in JF(x).2. (6.2.2. 2 " (6.274 6 The Feynman Rules where 8F(X ~ y) is the scalar field propagator.2.) For instance.4) ~.Y') . 6.) " LVt(pa .tpt(y)[J.1. respectively.14).f' p tm (p. (6. calculated in the next section.1. (Here as in Eq. then we have simply P(p)~l. (6. (6. y) = e(x .16) in Eq.w) is a polynomial in P and w. or of any odd order in Jf(x).P. (6. then (6. if 1/V(x) and tpmCV) are scalar fields ¢(x) and ¢(y) for a particle of spin zero.3) where Ptm(P.14).2.15) and (6.1. (6.2 Calculation of the Propagator We now turn to a calculation of the propagator (6.1). and using the commutation or anticommutation relations for annihilation and creation operators. _Jp + m. where t and m are here fourvalued Dirac indices.J p2+ m.y )(271")3/2 + J d3p L Ut(p " (J" n)u~(prr n)eip(X y) elY  x)(2n)3/2 J d3p I>~(pa n)Vt(pa n)e. (The matrix [J appears here because we are considering the pairing of 1p{(x) with lj"~(y).2) n)v~(pa n) ~ + (2Jp2 + m.
(j.2.b run by unit steps from A to +A.2.7) where sinh 8 = Ipl/m.4))] [exp(+8jJ .!ab(Y) for a particle of spin j. if 1p(x) and tpm(Y) are components of fields 1J. A to +A. Eqs.9) in which po is taken as +/r + m1.2) and (6. b'.!ab(X) and 1J. and is otherwise constant) to move .B) and (A. a'b')C. JIBI)]" bb' ' (6. i.e.a.ll are polynomials depending only on q. (6.10) p(L)(q) = p(O)(q) + qOp(1)(q) (for general qll). respectively.li'l(jf5.2.2.Y) ~ 8(x  Y)Pt. respectively. we must say a bit about how to extend the definition of the polynomial P(p).2.8) is the function introduced in Chapter 5 L1+(x) =(2n)3 Jd3p (2 pO)le'P> (6. (6. because any power (pO)2v or (pO)2v+l can be written as (p2 + m1)V or pO(pl + m2)".y) ox + O(y where ~+(x) x)Pt. where p(O. while a. Thus we can define a polynomial p(Ll(q) by the conditions that pILI(p) ~ P(p) (6. then Pab.ilb(P) = L L L CAB(jf5.3) only define P(p) for four~ momenta 'on the mass shell'. and likewise for the running indices a'.il..2. (6. We can now use the relations (6. (6.B) repre~ sentations of the homogeneous Lorentz group.1) yields iLlt.11) (recall that 8(x) has a unit step at xu.h.:) L1+(x . To go further. (j'i)') a'b' a'b' 11 x [exp(8jJ. (i :J ) L1+(y  x). B to +B.3) in Eq.2) and (6.2. and It Inserting Eqs. JIB)] b/l! [exp(8jJ . Any polynomial function of such four~momentum can always be taken as linear in po.6.2. JIA))] X aa' aa' [exp(+8jJ.(x.2.2 Calculation of the Propagator 275 More generally.2. with pO = +/p2 + m2. and B to +B. in the irreducible (A. JI.
F(X .15) . We introduce new integration variables.9) for D. yielding lLJ.q2 _ (qOf (In the denominator we have replaced 2€Jq2 + m2 with e.) (i(. (6.iqOxO) CL' (2n)3J2 yq 2 + m2 ° x [(qOJQ2+mLiEt + (_qOJQ2+mLiEt] Combining denominators and adopting a fourdimensional notation. since a change x + ~X in Eq. giving an integral equal to zero.9) can be compensated by a change p + p in the integration variable.6. ()_ I! 1 x . because the only important thing about this quantity is that it • To prove this. (X exp(ist) ds.14) It will be most useful to use the expression of the Feynman propagator as a Fourier integraL The step functions in Eq. (6.13) However. (6.2.2.F O(t) ~ 1. qO = pO + s in the first term of Eq.2.y) 1I+(y .l ( i :x) ~F(X  y) [1I+(x .y0)p)~ (iV) where D.2.16) where q2 .12).y) = p).+(x) is even in x. .2.2. We can therefore drop the second term in Eq. 2m d3 q 00 dq exp(iq .13) have the Fourier representation* 2m }00 S + IE This can be combined with the Fourier integral (6.tm(X.. q _ p.y).2. (6. (6.. for xO = 0 the function .2. (6. note that if t > then the contour of integration can be closed with a largc clockwise semicircle in the lower halfplane.. (6.276 6 The Feynman Rules the derivative operators to the left of the 0 functions in Eq.2.(6.2.8) ~(m(x. Ie (6. and write simply D.2.13).:) D. x .12) + 6(x o . If 1< 0 then the contour can be closed with a large counterclockwise !lcmicircle in thc uppcr halfplanc.F is the 'Feynman propagator' (6. whcre thc intcgrand is analytic.+(x).y) = Pg. we have simply dF(x)=(2n)4!d4q q +m ~xp(iq2·x). so the integral picks up a contribution of 2/ri from the pole at s = if.xl] . ° .
6.2 Calculation of the Propagator
277
is a positive infinitesimal.) This shows incidentally that dF is a Green's function for the KleinGordon differential operator, in the sense that
(0 m')"'F(x) ~ 04(x) (6.2.17)
with boundary conditions specified by the iE in the denominator: as shown by Eq. (6.2.13), dF(X) for xO _ +oc: or xO _ 00 involves only positive or negative frequency terms, exp( ixo Vp2 + m2) or exp( +ixo";p2 + m 2 ), respectively. Inserting Eq. (6.2.16) in Eq. (6.2.14) now gives the propagator as
"'fm(X,y)
~ (2.)4
J
d4q
p(L)() 'q'('y) ,~q e . q +m2 IE
(6.2.18)
There is one obvious problem with this expression. The polynomial P(p) is Lorentzcovariant when p is on the mass shell, p2 = _m 2, but in Eq. (6.2.18) we integrate over all qll, not restricted to the mass shell. The polynomial p{LJ(q) is defined for general q/t to be linear in qO, a condition that clearly does not respect Lorentz covariance unless the polynomial is also linear in each spatial component qi as well. We can instead always define our extension of the polynomial P(p) to general fourmomenta qll, which we shall call simply P(q), in such a way that P(q) is Lorentzcovariant for general qll, in the sense that
Ptm(Aq) ~ D,dA)D;",,(A)P"m,(q),
where All v is a general Lorentz transfonnation, and D(A) is the appropriate representation of the Lorentz group. For instance, for scalar, Dirac, and fourvector fields, these covariant extensions are obviously provided by just replacing pll with a general fourmomentum q/l in Eqs. (6.2.4), (6.2.5), and (6.2.6). For the scalar and Dirac fields, these are already linear in qO, so here there is no difference between p{LJ(q) and P(q):
p;~J(q) = Prm(q)
(scalar, Dirac fields).
(6.2.19)
On the other hand, for the vector field of a spin one particle, the 00 components of the covariant polynomial PlJv(q) 1J/tl. + m 2 q/lqv are quadratic in qO, so here there is a difference:
P~~l(q)
= 1J/l" =
+ m2 [q/lqv  bZ6~(q6  q2 P!IV(q) + m2(q2 + m2)()Z6~ .
m )]
2
(6.2.20)
(The extra term here is fixed by the two conditions that it must cancel the (QO)2 term in Poo(q), and must vanish when q!l is on the mass shell.) Inserting this in Eq. (6.2.18) gives the propagator of a vector field as
"'w(q)
~ (2.)
4
J
d q,
4
Pllv(q)eiq'(xyj 2 . q +m IE
° + m2 04 (x  y)o"o,° .
(6.2.21)
278
6 The Feynman Rules
The first term is manifestly covariant, and the second lenn, though not covariant, is local, so it can be cancelled by adding a local noncovariant term to the Hamiltonian density. Specifically, if VjI(x) interacts with other fields through a term Vp(x)JI'(x) in JF(x), then the effect of the second term in Eq. (6.2.21) is to produce an effective interaction
 Uf'eJJ(x) =
![
jJJl(x)] [ 
ir(x)] [ im2o~o~] .
(The factors i are the usual ones which always accompany vertices and propagators. The factor ! is needed hecause there are two ways to pair other fields with Jr"eff(x), differing in the interchange of JIl and P.) Thus the effect of the noncovariant second term in Eq. (6.2.21) can be cancelled by adding to JF(x) the noncovariant term
1 J ,It'Ndx) ~ .>fC'ff(x) ~ 2m 2 [0 (xl ] 2 '
(H22)
It is the singularity of the equaltime commutators of vector fields at zero separation that requires us to employ a wider class of interactions than those with a scalar density. A detailed nonperturbative proof of the Lorentz invariance of the S~matrix in this theory will be given in the next chapter. It should not be thought that this is solely a phenomenon associated with spins j > 1. For instance, consider the vector field associated with a particle of spin j = 0, equal (as discussed in Chapter 5) to the derivative cA¢(x) of a scalar field. For the pairing of this field with a scalar ¢t(y), the polynomial P(p) on the mass shell is
P,(p)
~
iPi, ,
(6.2.23)
while the pairing of ih¢(x) with cl/¢\V) yields a polynomial
P).,Ij(p) = PAPIj .
(6.2.24)
The covariant polynomials for general offshell fourmomenta q/l are again obtained by just substituting ql' for pi in Eqs. (6.2.23) and (6.2.24). Eg. (6.2.23) shows that PA(q) is already linear in qQ, so here there is no difference between P).(q) and pjLl(q). However, for Eq. (6.2.24) there is a difference:
pl.~l(q)
=
q).qlj  (q6  q2  m2)t5~t5~
= p;',I/(q)
+ (q2 + m2)b?o~,
(6.2.25)
so here the propagator is
~).,Ij(x,y) = (2rr)
4
J
d q)
4
q/~ql/eiqx 2 . q~ + m  IE
(6.2.26)
6.2 Calculation of the Propagator
279
Just as before, the noncovariant effects of the second term may be removed by adding to the interaction a non~covariant term
Jl'Nc(X)
~ j [JO(x)r,
,
(6.2.27)
where JI'(x) is here the current which multiplies op¢(x) in the covariant
part of .IF(x). It should be clear that (at least for massive particles) the effects of
of the propagator can always be c<:lncelled in this way by adding noncovariant loc<:ll terms to the Hamiltonian density. This is because the numerator pJ~\q) in the propagator must equal the covariant polrnomial Ptm(q) when qP is on the mass shell, so the difference between pJ~ (q) and p{m(q) must contain a factor q2 + rn 2. This factor cancels the denominator (q2 + m2  if') in the contribution of this difference to Eq. (6.2.18), so Eq. (6.2.18) always equals a covariant term plus a term proportional to the delta function 04(x  y) or its derivatives. The effect of the latter term may be cancelled by adding to the interaction a term quadratic in the currents to which the paired fields couple, or in their derivatives. In what follows, it will be assumed tacitly that such a term has been included in the interaction, and in consequence we shall use the covariant polynomial p{m(q) in the propagator (6.2.18), and will thus henceforth drop the label 'L'. It may seem that this is a rather ad hoc procedure. Fortunately, in the canonical formalism discussed in the following chapter, the non~covariant term in the Hamiltonian density needed to cancel non~covariant terms in the propagator arises automatically. This, in fact, fonns part of the motivation for introducing the canonical formalism.
non~covariant parts
...
Before closing this section, it may be useful to mention some other definitions of the propagator, equivalent to Eq. (6.2.1), that appear commonly in the literature. First, taking the vacuum expectation value of Eq. (6.1.14) gIves
i11I"(x,y)
~ O(x 
([v,t(X), v,~t(y)] ,) ° + ely  x) ( [v',;;' (Y), v'r(X)],) °
Y)
(6.2.28)
(Here (AB" ')0 denotes the vacuum expectation value ($o,AB··· $0).) Both "P;(x) and ~);t(y) annihilate the vacuum, so only one term in each commutator or anticommutator in Eq. (6.2.28) actual1y contributes to the propagator:
280
6 The Feynman Rules
Also, lpt and tp+ would annihilate a vacuum state on the right, and lp and tp+t would annihilate the vacuum state on the left, so 1p+ and lp may be replaced everywhere in Eq. (6.2.29) with the complete field
1/) = 11'+
+ tp:
 iA'm(x,y) ~ 8(x  y)(w(x)'I'l(y»o + O(y  x)('I'l(Y)w(x»o. (6.2.30)
This is often written
(6.2.31)
where T is a timeordered product, whose definition is now extended" to all fields, with a minus sign for any odd permutation of fermionic operators.
6.3
Momentum Space Rules
The Feynman rules outlined in Section 6.1 specify how to calculate the contribution to the S·matrix of a given Nth order diagram, as the integral over N spacetime coordinates of a product of spacetimedependent factors. For a final particle (or antiparticle) line with momentum p'Ji leaving a vertex with spacetime coordinate xl', we get a factor proportional to exp(ip'· x), and for an initial particle line with momentum pit entering a vertex with spacetime coordinate xl', we get a factor proportional to exp(+ip' x). In Section 6.2 we saw that the factor associated with an internal line running from y to x can be expressed as a Fourier integral, over offshell fourmomenta ql', of an integrand proportional to exp(iq' (x  y)). We can think of qJi as the fourmomentum flowing along the internal line in the direction of the arrow from y to x. Hence the integral over each vertex's spacetime position merely yields a factor
(2.)4 0 '
(I: p + I: q  I: p'  I: q')
z=
,
(6.3.1)
where L p' and L p denote the total fourmomentum of all the final or initial particles leaving or entering the vertex, and q' and q denote the total fourmomentum of all the internal lines with arrows leaving or entering the vertex, respectively. Of course, in place of these integrals over x Ji s, we now have to do integrals over the Fourier variables ql', one for each internal line. These considerations can be encapsulated in a new set of Feynman rules
z=
•• This is not inconsistent with our previous definition of the timeordered product of Hamiltonian densities in Chapter 3, because the Hamiltonian density can only contain even numbers of fermionic Held factors.
6.3 Momentum Space Rules
l,
281
p
cr "
,
'
p
4,
cr"
,
'e
(
£
£
I"
(b)
pan
(0)
4,
£
p
cr "
,
'
p'a'n'c
(21t
r 3/2 V;( Pan)
"'
Id)
Ie)
,
q
I
P£",(q)
q2+
(21t)4
m1ie
In
Figure 6.9. Graphical representation of pairings of operators arismg in the momentumspace evaluation of the Smatrix. The expressions on the right are the factors that must be included in the momentum space integrand of the Smatrix for each line of the Feynman diagram.
(see Figure 6.9) for calculating contribution to the Smatrix as integrals over momentum variables: (i) Draw all Feynman diagrams of the desired order, just as described in Section 6.1. However, instead of labelling each vertex with a spacetime coordinate, each internal line is now labelled with an offmassshell fourmomentum, considered conventionally to flow in the
282
6 The Feynman Rules
direction of the arrow (or in either direction for neutral particle lines without arrows.)
(ii) For each vertex of type i, include a factor
 i(2n)'g; 0'
(2..: p + 2..: q  2..: p'  2..: q')
(6.3.2)
with the momentum sums having the same meaning as in (6.3.1). This delta function ensures that the fourmomentum is conserved at every point in the diagram. For each external line running upwards out of the diagram, include a factor (2Jr)3/2ui(p'a!n') or (21t)3/2 vt (p ' a ' n'), for arrows pointing upwards or downwards, respectively. For each external line running from below into the diagram, include a f<lctor (2JT)3/2 ut (pan) or (2n)3/2 v;(pan), for arrows pointing upwards or downwards, respectively. For each internal line with ends labelled t and m, the arrow pointing from m to t, and carrying a momentum label q/l, include as a factor the integrand of the integral for i.6.{m(q):
i(21r)4Ptm(q)/(q2+m~ic).
(6.3.3)
A reminder: for scalars or antiscalars of fourmomentum q, the us and vs are simply (2 qO)1/2, while the polynomial P(q) is unity. For Dirac spinors of fourmomentum p and mass M, the us and vs are the normalized Dirac spinors described in Section 5.5, and the polynomi<ll P(p) is the matrix (h'IIPI' + M)f. (iii) Integrate the product of all these factors over the fourmomenta carried by internal lines, and sum over all field indices I,m, etc. (iv) Add up the results obtained in this way from each Feynman diagram. Additional combinatoric factors and fermionic signs may need to be included, as described in parts (v) and (vi) of Section 6.1. Examples will be given at the end of this section. We have a fourmomentum integration variable for every internal1ine, but many of these are eliminated by the delta functions associated with vertices. Since energy and momentum are separately conserved for each connected part of a Feynman diagram, there will be C delta functions left over in a graph with C connected parts. Hence in a diagram with I internal lines and V vertices, the number of independent four·momenta that are not fixed by the delta functions is I  [V  C]. This is clearly also the number L of independent loops:
L
~
I  V
+C ,
(6.3.4)
which is defined as the maximum number of internal lines that can be cut without disconnecting tbe diagram, because any such and only
6.3 Momentum Space Rules
283
such internal lines can be assigned an independent fourmomentum. We can think of the independent momentum variables as characterizing the momenta that circulate in each loop. In particular a tree graph is one without loops; after taking the delta functions into account there are no momentumspace integrals left for such graphs. For instance, in a theory with interaction (6.t.18), the Smatrix (6.1.27) for fermionboson scattering is given by the momentumspace Feynman rules as
L
k'l'm'klm
(i)2(2n)8 gl 'm'k' gmlk Ur(P1'IT;n~)ul(pIO"lnJ)
2 Pm'm;q) . )
X Jd 4q (i(2n)4
q +mlllIE
x (2n)6[Uk'(P2 '0"~n;)Uk(p20"2nl)84(p2 
4 P2'  q)b (P1
4
+q+ q)]
pd ,
+
ui:(P2 '0";n~)Uk,(p20"2n2)b4(p2  PI'  q)b (Pl  P2'
with labels 1 and 2 here denoting fermions and bosons, respectively. The momentumspace integral here is trivial, and gives
SPI'rI~Il'1 P2',,;n~
x
,Plll"lnl P2<l"21l2
=
4 i(2n)2b (PI
+ P2 
P; 
p~)
2:
k' I' m'klm
gl'm'k' gmlk uj,(p1 'IT; n'l) UI(PI 0"1 nj)
x [(
Pm'm(P1' ~p,) '('" ) )2 1 . Uk' P2 1T2nl)udp20"2n2 PI'  PI + mm IE '( " ' ) ( )] Uk P2 (J2n2 Uk' P2 1T2n2
+ (P2' Pm'm(P2'~pd IE )2 +) .  PI mr;j
(6.3.5)
In the same way, the Smatrix element (6.1.28) for fermionfermion scattering in the same theory is
Spl''';Il\
P2'a;n~
. PI "1111 P2 <l"2 112 = i(2n)2b
4
(Pl
+ P2 
P;  P2)
x
L gm'mk' gl'lk k'l'm'klm (PI' 
'"
P"k(pk' 2
ptl
2
pd + mk 
IE
.
X U;,,(P2' 0"; n~) uj'(PI ' 0"1 n~) U:(P2 0"2 n2)uj(PI 0"1 nj)
 [1' ""
2'1 .
(6.3.6)
These results illustrate the need for a more compact notation. We may define a fermionboson coupling matrix
(6.3.7)
The matrix elements (6.3.5) and (6.3.6) for fermionboson and fermion
284
6 The Feynman Rules
fennion scattering can then be rewritten in matrix notation as
Spl/u;n;
P2'(T~n;
'" O"t n
, PIUIU] P2'>2"2 =
i(21t)~264(pl
+ P2 
p~  pS) ~
k'k
t ( U
(PI
P(Pl'  pJl d 1 k'( Pl'Pt+M 2 ' lk u (PJCJwd ) )? It:
x uk'(P2' 0"2n;.)Uk(P20"2n2)
+
t '" P(p2'PJl ( U (PI 0"lndlk'(P21 pl)2+M2
x U;'(P2 '0";nS)Ukl(P20"2nZ) ]
and
(6.3.8)
(Ut(P20"2nZ)fkU(PI 0"1
nd)
(6.3.9)
where M 2 and m2 are the diagonal mass matrices of the fermions and bosons in Eqs. (6.3.8) and (6.3.9), respectively. The general rule is that in using matrix notation, one writes coefficient functions, coupling matrices, and propagators in an order dictated by following lines backwards from the order indicated by the arrows. In the same notation, the Smatrix for bosonboson scattering in the same theory would be given by a sum of oneloop diagrams, shown in Figure 6.7:
SpJ'a;n;P2'a;Il;,Pl<l"1I1JP2a2IlZ
= b (Pl+P2p 'IP2)
4
x
L
klk2 k'lk;
uZ; (p~, ai, n; )uZ; (P2, a;, n; )Uk l (PI, aI, nl )Uk, (Pl, a2, n2)
P(q) P(q
z q 2+ M2 iE
x !a'qTr{rk , x
r k,
r
+ pj 
+ pj)2 + M2  iE pil r P(q  p~)
l(q
2
P(q+pjl
}
(6.3.10)
k' (Q+Plp])2+M2_iE k (q_P2)2+M2_iE
+... ,
where the ellipsis in the last line indicates terms obtained by permuting bosons 1',2', and 2. The minus sign at the beginning of the righthand side is the extra minus sign associated with fennionic loops. Note that after elimination of delta functions there is just one momentumspace
6.3 Momentum Space Rules
285
integral here, as appropriate for a diagram with one loop. We shall see how to do this sort of momentumspace integral in Chapter 11. To make this more specific, consider a theory with a Dirac spinor field 1p(x) of mass M and a pseudoscalar field ¢(x) of mass m, interacting through the interaction igipY51p. (The factor i is inserted to make this interaction Hennitian for real coupling constants g.) Recall that the polynomial P(q) for the scalar is just unity, while for the spinor it is [iYpqll+M]{l. Also, the u for a scalar of energy E is (2E)1/2, while for the spinor u is the conventionally normalized Dirac spinor discussed in Section 5.5. Eqs. (6.3.8), (6.3.9), and (6.3.10) give the lowestorder connected Smatrix elements for fermionboson scattering, fermionfermion scattering, and bosonboson scattering;
Sp'IO";
P~
, PIO"I P2
=
i(2:n:)2g2(4El E2)1/2J4(PI
i"/ (PI" PI'  PI
+ P2 
P;  P2)
X [ ( U(PI
I
(Tt h'5 (
"
)2
pll" + M ) + m2  IE YsU(PI (TI) .
+
SPI'O";
( U(PI'(Tt)'/5(
PI 0"1 P20"2
iy (P2  P,)" ] ")2+ 2 .Y5 U(Pl(T]) P2'  PI m  IE
i(2:n:)2 g 2J4(PI
p~IO";,
=
+ P2  P; 
p~)
x (U(P2 (Tll Ys U(PI CT~)) (U(P2 (T2hS U(Pl (TI))
X~_~;l_~~ (PI'  pd 2 + m2  iE
[1'~2'],
Sp'l
V2 '
PI P2
=
_(2:n:)6 g2 (16E1E2E;E;)1/2 b4(PI
+ P2 
PI  P2)
xfa' q Tr{"f5q2+M2_iEY5(q+p;)2+M2_iE iYJiql'+M iYJi(q+p;)I'+M
iYJi(q + Pl)Ji + M iYJi(q  P2)Ji + M } X Y5 I 2 2 Ys I 2 2 (q + PI  p,) + M  ic (q P2) + M  iE
p; 
+ ...
,
where in the last formula the ellipsis indicates a sum over permutations of particles 2, I', 2'. The factors p in the fermion propagator numerators have been used to replace ut with u.
'*.
Another useful topological result expresses a sort of conservation law of lines. For the moment we can think of all internal and external lines as being created at vertices and destroyed in pairs at the centers of internal lines or when external lines leave the diagram. (This has nothing to do with the directions of the arrows carried by these lines.) Equating the
286
6 The Feynman Rules
numbers of lines that are created and destroyed then gives
2l+E=Ln j Vi ,
(6.3.11 )
where [ and E are the numbers of internal and external lines, Vi are the numbers of vertices of various types labelled i, and nj are the number of lines attached to each vertex. (This also holds separately for fields of each type.) In particular, if all interactions involve the same number ni = n of fields, then this reads
(6.312)
where V is the total number of all vertices. In this case, we can eliminate I from Eqs. (6.3.4) and (6.3.11). and find that for a connected (i.e., C = 1) graph the number of vertices is given by
V~2L+E2
n2 For instance for a trilinear interaction the diagrams for a scattering process (E = 4) with L = 0, 1,2' .. has V = 2,4,6' .. vertices. In general, the expansion in powers of the coupling constants is an expansion in increasing numbers of loops.
(6.3.13)
6.4
Off the Mass Shell
In the Feynman diagrams for any Smatrix element all external lines are 'on the mass shell'; that is, the fourmomentum associated with an external line for a particle of mass m is constrained to satisfy p/tpll = _m 2 . It is often important also to consider Feynman diagrams 'off the mass shell', for which the external line energies like the energies associated with internal lines are free variables, unrelated to any threemomenta. For one thing, these arise as parts of larger Feynman diagrams; for instance, a loop appearing as an insertion in some internal line of a diagram could be regarded as a Feynman diagram with two external lines, both off the mass shelL Of course, once we calculate the contribution of a given Feynman diagram off the mass shell, it is easy to calculate the associated Smatrix elements by going to the mass shell, taking the fourmomentum pll flowing along the line into the diagram to have pO = Vp2 + m2 for particles in the initial state and po = _J p2 + m2 for particles in the final state, and including the appropriate external line factors (2n)3/2 ut or (2n)3/2v; for initial particles or antiparticles and (2n)3/2 u;' or (2n)3/2 v( for final particles or antiparticles. Indeed, when we come to the path integral approach in Chapter 8 we shall find it easiest first to derive the Feynman
6.4 OJ] the Mass Shell
287
rules for diagrams with all external lines off the mass shell, and then obtain Smatrix elements by letting the momenta associated with external lines approach their appropriate mass shells. Feynman graphs with lines off the mass shell are just a special case of a wider generalization of the Feynman rules that takes into account the effects of various possible external fields. Suppose we add a sum of terms involving external fields f'a(x) to the Hamiltonian, so that the interaction V(t) that is used in the Dyson series (3.5.10) for the Smatrix is replaced with
Ve(t)
~
V(t)
+~
J ,
3 d x ,,(x,t)o,(x,t).
(6.4.1)
The 'currents' oa(t) have the usual timedependence of the interaction picture
Oa(t)
=
exp(iHot)oa(O)exp(iHot),
(6.4.2)
but are otherwise quite arbitrary operators. The Smatrix for any given transition .:1  4 fJ then becomes a functional Sflrt[f'] of the cnumber functions f'a(t). The Feynman rules for computing this functional are given by an obvious extension of the usual Feynman rules. In addition to the usual vertices obtained from V(t), we must include additional vertices: if ou(x) is a product of na field factors, then any Oa vertex with position label x must have nu lines of corresponding types attached, and makes a contribution to the positionspace Feynman rules equal to if'a(x) times whatever numerical factors appear in oa(x). It follows then that the rth variational derivative of Sflrt[i'] with respect to f'a(x), f'b(Y)'" at f' = 0 is given by position space diagrams with r additional vertices, to which are attached respectively na, nh ... internal lines, and no external lines. These vertices carry position labels X,y'" over which we do not integrate; each such vertex makes a contribution equal to i times whatever numerical factors appear in the associated current 0a' In particular, in the case where these currents are all single field factors,
I.e.,
Ve(t)
~
Vir)
+~
f
.
f
d 3x ,/(x, t)V'f(X, t),
the rth variational derivative of Sp,,[€] with respect to €t{x), f'm(Y)'" at € = 0 is given by position space diagrams with r additional vertices carrying spacetime labels x, y"', to each of which is attached a single internal particle line of type t, m' . '. These can be thought of as offshell external lines, with the difference that their contribution to the matrix element is not a coefficient function like (2n)3/2ur(p,rr)eiPX or (2n)3/2 u;(p,rr)eiPX but a propagator, as well as a factor i from the vertex at the end of
288
6 The Feynman Rules
the line. We obtain a momentum space Feynman diagram with particles in states r.r. and fJ on the mass shell plus r external lines of type t, m' .. carrying momenta p, p' . .. from the variational derivative
0'5,,[eJ ] [Oedx)Oe.(y)"·
u;,
,~O
by stripping away the propagators on each of the offshell lines and then taking the appropriate Fourier transforms and multiplying with etc. and a factor (i)'. appropriate coefficient functions Ut, It is very useful for a number of purposes to recognize that there is a simple relation between the sum of contributions from all perturbation theory diagrams for any offshell amplitude and a matrix element, between eigenstates of the full Hamiltonian, of a timeordered product of corresponding operators in the Heisenberg picture. This relation is provided by a theorem,3 which states that to all orders of perturbation theory"
[oe,(~;5;;:~~)L ~ ('Pp, T{ 
iO,(x), iO,(y)
}'P, +) ,
(6.4.3 )
,
where Ou(x), etc. are the counterparts of ou(x) in the Heisenberg picture O,(x, ,) ~ exp(iR') o,(x, 0) exp( iR') ~ 0(,) o,(x, ,) 0 1 (,)
(6.4.4) (6.4.5)
and 'l'p+ and 'l'p are 'in' and 'out' eigenstates of the full Hamiltonian H, respectively. Here is the proof. From Eq. (3.5.10), we see immediately that the lefthand side of Eq. (6.4.3) is
For definiteness, suppose that x? > xg > ~ x~. Then we can denote by Til TINt all TS between and by rOI ... rONo all rs greater than x~, and so on; finally denoting by Tri ... erN, all rs that are less than x~ .
x?;
x?
• For a single 0 operator, this is a version of the Schwinger action principle. 4
t) satisfies the differential equation d d U(t'. we can just sum independently over No. NJ.4.(x..4..4.(xr) ] <.4.5).4.4 0JI the Mass Shell 289 Eq.9) with the obvious initial condition U(t.'" N r of these t s. (6. The factor N!/No!N\!"'Nr ! is the number of ways of sorting N 1'S into r + 1 subsets.6) then becomes: brS[E] [ bE(. Joo x (lD{J' T {V(1'od'" V(1'ONo )} 0UI (xdT {V(1'll)'" V(1'JNI)} x"'o."'. t) ~ 1. setting N where it appears in (_i)N equal to No+NJ + ..... subject to the condition N o+ N 1 +.4. Nl.11) with n given by Eq.10) This has the solution U(t'. + N r .8) The operator U(t'.t) t' (6.. Instead of summing over No.4.7) where (6.=0 = f N=O (_i)N+r L NoN]"'N. (6. + Nr = N..lj(xd···bEa.~? J~ ri.4. N].t) ~ iV(t')U(t'.No+Nl+·+N. d1'lNl . (6..t)) exp( iHot) ~ rr'(t')(J(t) (6. . r"'? d1' r \ . each containing No. NJb N.). and then summing over N.. N r. .11) in Eq.? d1'll . t) ~ exp(iHot') exp( iH(t' .)}<I>.Nr.. Inserting Eq.6.) Oa2(X2)'" T{V(T'Il"V(T'N..4..7) and using . N! No!NJ!"'Nr! x rre· d1'01 ··· d1'oNo .. (6. (6. This gives (6. d1'rN.
Problems 1.12)) x? .(xll . responding Smatrix elements for the processes F C + B . x? 'I'/± ~ n(+oo)<I>~ . 3. Calculate the connected Smatrix element for scalarscalar scattering to second order in g.)'" O. we saw in Section 3.13) is the desired result (6. Consider a theory involving a neutral scalar field ¢(x) for a boson B and a complex Dirac field l.4).(xr) . (3. (6. 2.pc + B. with interaction (in the interaction picture) V = ig..(x.(X r ) e=O ~(i)'(n(oo)<I>~.3). and pc + F _ B + B (where F C is the antiparticle of F). so this relation holds whatever the order of the times Also. we have 6 The Feynman Rules [ Of'al(xt}"'Of"a.. Calculate the Smatrix for scalarscalar scattering to order g.. Consider the theory of a real scalar field ¢. > x?. O..4. expressing the result as an integral over a single fourmomentum.=0 6S['1 ] ~ (i)' ( n(oo)<I>~. with interaction V = g.13) But now both sides are entirely symmetric (or for fermions antisymmetric) in the as and xs.1.(x.4. ·o.p(X) for a fermion F.4. Draw all the connected orderg 2 Feynman diagrams and calculate the cor.4. and use the result to calculate the differential scattering crosssection.r d3x ¢(x)4/4!.. Do all integrals. . T{o..4. doing all integrals.)}n(OO)<I>.290 Eg.). Consider the theory of a real scalar field ¢(x). F + F C _ F + F C .12) In deriving this result we supposed that x? > xg > .) . with interaction (in the interaction picture) V = g J dJx ¢(x)3/31. (6.4.14) Hence Eq.r d3 x1jJ(X)Y51p(X)¢(x). so we could just as well replace the product of operators on the righthandside with the timeordered product: 6S[..)n(oo)<I>.(x. Use the results to calculate the differential crosssection for scalarscalar scattering in the centerofmass system. (6. Calculate the correction terms in the Smatrix for scalarscalar scattering to order g2. but do all xintegrals.. (6.1 that (in the sense of Eq.] ] [ Of'aj(x!l'" DEa. (6.
F. 4. 3. 1. Phys. T{<l>(x). to orders g and g2. 75. The formal statement of this result is known as Wick's theorem. References \. 1. cI>(x)'¥o) and (1'0. Wick. 2. Rev. What is the contribution in Feynman diagrams from the contraction of the derivative oJ'VJt(x) of a Dirac field with the adjoint tpl(y) of the field? S. GellMann and F. Rev. Schwinger. 268 (1950). Dyson. Phys.cI>(y)}'¥o) in the theory of Problem 1. Use the theorem of Section 6. Phys. respectively. Low. 1736 (1949). including M. 914 (1951). E. see G.References 291 4. 82.486. Rev. It was known in the early 1950s to several theorists. SO. . I do not know who first proved this theorem.4 to give expressions for the vacuum expectation values of Heisenberg picture operators (1'0. C.
We saw in the previous chapter that in theories with derivative couplings or spins j > 1. Heisenberg. all of the most familiar quantum field theories furnish canonical systems. this does not in itself explain why we should prefer to use the Lagrangian formalism as a starting point in constructing various quantum field theories. However. Dirac.7 The Canonical Formalism Ever since the birth of quantum field theory in the papers of Born. when we come to nonAbelian gauge theories in Volume II. Yet historical precedent is not a very convincing reason for using this fonnalism. That is. The canonical fonnalism with a scalar Lagrangian density will automatically provide these extra tenns. it would be 292 . would it bother us if it could not be derived by the canonical quantization of some Lagrangian? To some extent this question is moot because. Jordan. its development has been historically linked to the canonical formalism. The point of the Lagrangian formalism is that it makes it easy to satisfy Lorentz invariance and other symmetries: a classical theory with a Lorentzinvariant Lagrangian density will when canonically quantized lead to a Lorentzinvariant quantum theory. This is the approach used in most books on quantum field theory. and Pauli in the late 1920s. there is no proof that every conceivable quantum field theory can be formulated in this way. it is not enough to take the interaction Hamiltonian as the integral over space of a scalar interaction density.1. and these can easily be put in a Lagrangian form. and therefore leads to a Lorentzinvariant Smatrix. so much so that it seems natural to begin any treatment of the subject today by postulating a Lagrangian and applying to it the rules of canonical quantization. And even if it can. Fermi. If we discovered a quantum field theory that led to a physically satisfactory Smatrix. This is not so trivial. this extra convenience will become a necessity. we shall see here that such a theory allows the construction of suitable quantum mechanical operators that satisfy the commutation relations of the Poincare algebra. as we shall see in Section 7. Later. we also need to add nonscalar terms to the interaction density to compensate for noncovariant terms in the propagators.
t) that satisfy the familiar canonical commutation or anticommutation relations: [q'(x. and anticommutators if both particles are fermions.1 Canonical Variables In this section we shall show that various quantum field theories that we have constructed so far satisfy the commutation rules and equations of motion of the Hamiltonian version of the canonical formalism. J2k~~1r. t)..1. t).(y.1 Canonical Variables 293 just about hopeless to try to guess at the form of the Hamiltonian in such theories without starting with a Lorentzinvariant and gaugeinvariant Lagrangian density. t)J+ ~ iO'(x .O) ~ _i0 3(x).(x. where the subscripts + indicate that these are commutators if either of the particles created and destroyed by the two operators are bosons. the real scalar field ¢(x) for a selfchargeconjugate particle of zero spin was found in Section 5. For instance. It is the Hamiltonian formalism that is needed to calculate the Smatrix (whether by operator or pathintegral methods) but it is not always easy to choose Hamiltonians that yield a Lorentzinvariant Smatrix. (7. p.¢(y))_ where Ll is the function Ll(x) ~ lI(x .y)] with kO .)3 [eiH~Y)  eik. q'(y. The purpose of the present section is to identify the canonical fields and their conjugates in various field theories. t) and canonical conjugates Pn(x.p. We first show that the free fields constructed in Chapter 5 automatically provide a system of quantum operators q"(x. t)l+ ~ 0. [q'(x.jk"''+'::m''2.y)o~ .1) (7. .y).3) [P. and use it to derive physically satisfactory Hamiltonians. t)).(X.1. t). We note that li(x.2 to obey the commutation relation [¢(x).1. 7. and incidentally to reassure us that the canonical formalism is indeed applicable to physically realistic theories.7.(y. In the balance of this chapter we shall take the Lagrangian version of the canonical formalism as our starting point.2) (7. ~ 0 . to tell us how to separate the freefield terms in the Lagrangian.
t)]_ ~ 0 .t). p(x.12) '( ) avo(x. t).(x.1)(7.t Equivalently.t) ~ (7.294 7 The Canonical Formalism (A dot denotes the derivative with respect to the time xo.3).1.<I>(x.1)(7. t) .10) (7.) It is easy then to see that the field and its timederivative (p obey the equaltime commutCltion relations: [<I>(x. (7.1.t) ~ v'(x. <p(y.) Here the freeparticle canonical variables may be taken as q'(X.1. <p(y.t) . [<I>(x).I).1.13) ox' with i = 1. For the complex scalar field of a particle of spin zero with a distinct antiparticle. We may therefore define the freeparticle canonical variables as the complex operators q(x.3.l) Pi(X.1.8) (7.t).3).2. t)]_ ~ (7.<I>(x.t) ~ <I>.1)(7. (7.1.7) which satisfy the canonical commutation relations (7. Therefore we may define canonical variables q(x.5) (7.y).11) and these satisfy the commutation relations (7.4) (7.1. writing rjJ (¢1 we have canonical variables + i(hl/ji with ¢k Hermitian fOf k 1.l3) satisfy the commutation relations (7.1.l) .t) ~ (7.t+ ) ~' . l(x. (7. The field equations (5.2.3 as [v"(x).y). t).3). Pk(X. For the real vector field of a particle of spin one.12) and (7.9) = <I> (X.36) and .t) =<p(x. p(x. (We are using Vii rather than VI' for the vector field because we want to reserve upper case letters here for the fields in the Heisenberg picture. the commutation relations are given by Section 5.1. t).1. [<I>(x. [<p(x. <l>t(y)]_ ~ L1(x .1.1. <I>(y)] O. The reader may check that (7. .3. t).1.1.1. v'(y)]_ ~ [ry"'  a:n L1(x  y) .6) 0. t)]_ ~ i0 3 (x .(x.1.t) .t) ~ <P.1. <I>(y. the commutation relations are [<I>(x).t=V1x.1.
}') for all x at Iixed y.1.18) are respectively just the left.q"( x.p(t)] is written with all qs to the left of <111 ps..and rightderivatives with respect to qn and pn.1.3.15) and (7.r) is a function of two classes of variables collectively called _~ and .p(I)] bpn(x.1 )(7.[ I p"(X.1.1.p(t)] 3q"(x.6 shows that the <lnticommutator is and Here it would be inconsistent to take 'Pn and lflt to be independent canonical variables. _ (7. It is conventional instead to define q"(x) ~.18) This definition is motivated by the fact that if F[q(t).t) ~ I'[F[ q (I) . we define· 3F[q(I).16) Pn(X) _ i.1 Canonical Variables (5. F[q(I). then F[f(}')] indicate.7..~(x). By a bosonic functional we mean one in which each term contains only even numbers of fermionic fields. I)] .1.p(I)] . if f(x. .t) at a fixed time t. That is.1.14) so vO is not regarded as one of the q s.3).3).1.1. The extension of these results to complex vector fields m<lY be h<lndled just as for complex scalar fields.(x). ] (7. we may define a quantum mechanical functional derivative: [or an arbitrary bosonic functional F[q(t).17) 3F[q(t).t) and Pn(x. (7.1.1. It is easy then to see that (7.1)(7. (7.16) satisfy the canonical anticommut<l tion rel<ltions (7.38) together with Eq. a functional that depends on the values of f(x.l) .1. Section 5.17) and (7. For any system of operators that satisfy commutation or anticommutation relations like (7. then (7.p(t)] of qn(x. for an • We are here using a notation that will be adopted henceforth.15) (7. For the Dirac field of a non·Majorana spin! particle.1.1..I).1'. because their anticommutator does not vanish at equal times.p (I)] .13) allow us to express other variables as vO 295 in terms of the (7.1.
t (7. (7.p of the qs and ps.1. (7.1. (7.(x. liq" and lip" are nnderstood to commute or anticommute with all fermionic operators.20) plJ(X.a.1.1.7) and the Fourier representation of the scalar field ¢. we have 6F[q(t).23) is equal up to a constant term to the functional ! d3 x [ 1p2 + !(Vq)2 + 1m2 q2] .24) becomes: Ho~ ~ 1 . respectively.17) and (7. p.1. (7. The freeparticle Hamiltonian is given as always by Ho ~L 1 '.24) To be more precise. (7.t) For more general functionals we need the definitions (7.n) jk 2 + m. In particular.t) = exp(iHot)Pn(x.p(t)] 6P.22) We recognize these as the familiar dynamical equations in the Hamiltonian formalism. a.O) exp( iHot) . Ho is the generator of timeMtranslations on freeparticle states in the sense that: qll(x. t) = exp(iHot)q"(x. using (7. 6p.18) to pin down various signs and equal~time commutators that may appear. so the freeparticle operators have the timedependence il'(x.296 arbitrary c~nllmber" 7 The Canonical Formalism variation Jq and (.Ho = d 3k a'(k.t»).19) (7. . we find that Eq.6 '( ) (7. (7.t) ~ .~.1.(X. Eq.t) ~ t[P.1. For instance. .Ho] ~ . n)a(k.1. .t)6Fi:~.~\t)] + 6F[q(t).1. it is easy to see that for a real scalar field.23) This H o may be rewritten in terms of the qs and ps at time t.1.1.a'(k)].p(t)] ~ 1 d'x ~ (6q'(X.(x.21) .t).[Ho.q'(x.t)] ~ 6:'7x~t)' 6& q x. and to commute with all bosonic operators.25) d3 kkO (a'(k)a(k)+ j6 3(kk)) •• Where q" and p" are bosonic or fermj'Jnic.1 d'kkO[u(k).(x.O)exp(~iHot).1.
5.27) d3 x [ ~a.(x.29) changes in such terms due to changes in the boundary conditions for the fields.23). We have seen that various freefield theories can be formulated in canonical terms. are physically significant. Whatever we suppose the complete Lagrangian of the scalar field may be. (7.1.25) for spinless particles. but from now on we shall content ourselves with guessing the form of the freefield Lagrangian and then confirming that it gives the correct freeparticle Hamiltonian. (7.25) up to a constant term. (7. we should ask what freefield Lagrangian gives Eq. I .1. the freefield Lagrangian is given by Lo[q(t). for Eq. Such terms only affect the zero of energy. tlq'(x. and have no physical significance in the absence of gravity. For instance.t) .1. from the Hamiltonian (7.!(Vq)2 _ !m 2q2] (7. t However.t.23). as we shall see.24) and (7. if some assumed freeparticle Lagrangian did not give Eq. Rather.1. A similar exercise may be carried out for the other canonical systems described in this section. except for the infinite constant term. (7.O)exp(iHt).1.exp(iHt)p. gives the freeparticle Hamiltonian (7. (7.q(t)] 3 ~ ~ a x p.28) P. This question may be answered by the wellknown Legendre transformation from the Hamiltonian to the Lagrangian. defined by Q'(x.25) as a consequence of Eq. which in turn is derived from a Lagrangian density. or more generally.tm2 ¢2]. as for instance qnantizing in the space between parallel plates rather than in infinite space.1 Canonical Variables 297 This is the same as Eq.1.(x. perhaps some auxiliary fields as well).7. H is usual in textbooks on quantum field theory to derive Eq. We can introduce canonical variables in what is called the 'Heisenberg picture'. (7.25) must hold. t) exp(iHt)q'(x. It is then a short step to show that the same is true of the interacting fields.24). 0) exp( iH t). we would conclude that it was the wrong Lagrangian.1.1. and have even been measured.A)al'¢. This seems to me backward. (7.(x.1.26) it being understood that pn is replaced everywhere by its expression in terms of qn and qn (and.1. J  Ho.1.7) we can derive the freefield Lagrangian for a scalar field: La = = J J d3x [pq  !p2 . Explicit forms for Ha as a functional of the q and p variables for other fields will be given in Section 7. t) .1. (7. this is the term that must be separated out and treated as a term of zeroth order in perturbation theory.1. (7.
t). t)] ~ 6P~~. the Heisenberg picture operators again satisfy the canonical commutation or anticommutation relations: [Q'(x. t) . P. Q'(y.y)6.(y. as we shall see.34) For instance.p] . t)]" ~ (7. t).6Q'(x. There is not much loss of generality in this.30) (7. because Eqs. the relation between the canonical conjugates Pn(x) and the field variables and their timederivatives is in general not the same as for the free~particle operators.35) In this case the canonical conjugate to Q is given by the same formula as for free fields: (7.29) define a similarity transformation.1. so that in terms of Heisenbergpicture variables H ~ J d3 x [ jp 2 + HIlQ)2 + jm2 Q2 + ff(Q)] . (7. Because this is a similarity transformation that commutes with H.1. [P. .1.24) plus the integral of a scalar interaction density Jf.1.(x.33) and (7. (7. t)h ~ i6 3 (x .2 The Lagrangian Formalism Having seen that various realistic theories may be cast in the canonical for~ malism.p]e.1. [Q'(x. t)]f ~ 0.. the easiest way to enforce Lorentz in~ variance and other symmetries is to choose a suitable Lagrangian and use it to derive the Hamiltonian. t).34).(x. P] ~ eiH'H[q.t) ~ i[P. t) ~ i[H.1. we must now face the question of how to choose the Hamiltonian. However. 7. (7. we might take the Hamiltonian for a real scalar field as the freeparticle term (7.t). the total Hamiltonian is the same functional of the Heisenberg picture operators as it was of the qs and ps: H[Q.iH ' ~ H[q.P.(x. Q'(x. but must be inferred from Eqs.1.H] ~ .1.298 7 The Canonical Formalism where H is the full Hamiltonian.t) (7.(y. 6H P.LJl) (7.32) O.1. (7.1. As we will see in the next section.28)(7. Also. they now have the timedependence Q'(x.33) .36) However. .1.
~ I d"x [J~~X) :t b~~X)] b'P{(x). f d 3 x [3 'P x) 3~. t) and their timederivatives 'P{ (x. 'i'(t)] .lj. As far a~ 1 know. t = .7. in general. We use upper case 'I'~ and lis to indicate that these are interacting rather than free fields. a functional like L in whieh we display the variable 1 is understood to depend on thc fields 'I'{(x. we may integrate by parts. t) . known as the action I['¥] I: dt L['P(t).2. 3'P{(x) + b'P (xl ".2). a functional· L['t'(t). no important i~. J'i'{(X)] Assuming that 6tpt(x) vanishes for t and write b/['P] + +00. (The derivation of Eq. (7. rather than Hamiltonians.1) iI ( t X. IS (7. • Recall that in the notation we use for functional~. it is easier to explore physically satisfactory theories by listing possible Lagrangians. 'i'(t)] btpt(x. The Lagrangian is. Instead.2. ( The equations of motion are (7. .2) These field equations can be usefully reformulated as a variational prin~ ciple.2. the change in I [tp] 3f['P] ~ (~ dt J~. by reversing the process that we are going to describe here of deriving Hamiltonians from Lagrangians.) But although we can go from Hamiltonians to Lagrangians or Lagrangians to Hamiltonians. •• Boxausc the 't' ~ and of's do not in general satisfy simple commutation or anticommutation relations. t ~ ) _ 3L['P(t). The conjugate fields flAx.3) Under an arbitrary variation of tp(x).oh/( t) U T X. (7.2.2. we will simply specify that the variational derivatives are what they would be for cnumber variable~ with minus signs and ellualtime commutators or anticommutators supplied as needed to make the formulas correct qnantummechanically.t) and 'V(x.ues hinge on the details here. we cannot give a ~imple definition of the functional derivatives occuring here a~ we did for functional derivatives with respect to the Qs and Ps in the previous section. We define a functional of tpt(x) Over all spacetime. (7.26) gives one example of this reconstruction.t) are defined as the variational derivatives" ) _ 3L['P(t). t).1. with the undisplayed variables { and x running over all their values at a flxed value of the displayed variable I. 'i'(t)] III x.2 The Lagrangian Formalism 299 given a realistic Hamiltonian.4) We see that the action is stationary with respect to all variations (jtpt that vanish at t + +co if and only if the fields satisfy the field equations (7. we can generally reconstruct a Lagrangian from which it could be derived. 'P(t)] of a set of generic fields 'PI (x.
we find a variation in L: oL ~ = f f d x dx 3 3 [Off [( 8'1't 0'1' i)\f'{  t + 8(V'I't) Vo'l' off (ost>. we guess that L should itself be a spaceintegral of an ordinary scalar function of'P(x) and o\{l(x)/oxi'.2. Varying qJ{(x) by an amount J\{It(x). (7.2.I)) (7.2. In addition to being Lorentzinvariant. then the real and imaginary parts of the conditions that I be stationary (the EulerLagrange equations) would yield 2N field equations for N fields.V' 8(V'I't) . (7. We will see in the next section that the reality of the action also ensures .2. say N of them. By breaking up any complex fields into their real and imaginary parts. I). since I [\II] is a time·integral of L[\f'(t). + o'i't 0'1' t] 02 V' D(V\f'/) 82) J\f't + a1't {PI't] 02 . If I were complex. 'i'(X.0'l'(x)/ox").8) The field equations (7. In particular. 'P(t)]. I). with independent real and imaginary parts. This is because we want just as many field equations as there are fields. the action I is required to be real.9) These are known as the EulerLagrange equations. if . and integrating by parts. it is natural in trying to construct a Lorentzinvariant theory to make I ['¥] a scalar functional. V'I'(x. oL 02 b\jJt = a\jJt . As expected. (7. we can always think of I as being a functional only of a number of real fields.300 7 The Canonical Formalism Because the field equations are determined by the functional I [\{I). 'i'(I)] ~ f d3 x 2 ('I'(x.2.:£ is a scalar then these equations are Lorentzinvariant. too many to be satisfied except in special cases.1) then read a az axil D(a\f'tj8xll) (7.6) All field theories used in current theories of elementary particles have Lagrangians of this form.7) so (with obvious arguments suppressed) oL 02 82 o'l't ~ 0'1" .2.5) so that the action is /['1'] ~ f d"x2('I'(x). known as the Lagrangian density :e: L['I'(I).
t)'i'I(X..2. Using the defining equation (7.12) The equations of motion (7.34).1. + J .t) y.L .'¥'(x.t) . .H . (7. In general. . t' .t) .(x.( I Y I' ).'¥ (y.2. the Hamiltonian is given by the Legendre transformation H ~~ J d3x 1I..¥I(x. t) d Y 'iclIdy.13) are the same as the Hamiltonian equations of motion (7. (7.1.'¥'(x.33) and (7. t) .t) n ' .t) '¥ Jd3".t) n ~ II.t .II.H ~ Jd3 I: Jd3 I:.2. (7. .t) n and  ~. Y L u'¥(I (y.t).12) and (7.x.2. (7.1.1I'(x. ..H ) ~ '¥ (x.'¥ (y.t) .'¥ (y.1).t'f 'II ( .L['¥(t). t) . I' " ~ . so that Eqs.t '¥ 0 IIt()' " where subscripts denote the quantities held fixed in these variational derivatives. .2.2. so in general it is a functional only of \{It and lIt.t) .L Y " . (7.2) are then equivalent to • '¥'(x.30)(7.13) It is tempting now to identify the generic field variables \fit and their conjugates ITt with the canonical variables Qn and Pn of the previous section.'¥'(x.pt to be expressed uniquely in terms of "P f and lIt.t) n . it is easy to see that Eq. Although the Lagrangian formalism makes it easy to construct theories that will satisfy Lorentz invariance and other symmetries.2.2. (7.(x.1.t) I' '¥ .t) .1) for ITt.1) does not in general allow.2.'i"'(y.2 The Lagrangian Formalism 301 that the generators of various symmetry transfonnations are Hermitian operators.H '¥ ~ '¥ (x.¥I(x.(x.11) U{x. I (7. Its variational derivatives with respect to these variables are . 'i'(t)] .¥I(x.'¥ (y.2.t) 3 " '¥ .t) n ..32). to calculate the Smatrix we need a formula for the interaction Hamiltonian. (7.7. and impose on them the same canonical commutation relations (7.10) has vanishing variational derivative with respect to \jIf for any \jIf satisfying Eq.t) ~ .H .10) Although Eq.t) . this simplifies to n .2. This is indeed the ..L 1I.
The constant m is known as the bare mass.' (7.I d]x (rrti> . Matters are not always so simple.2. We shall denote the field variahles 'Pc whose timederivatives do not appear in the Lagrangian as C'. (7. that jm 2$2 + £('11) must be bounded below as a function 01" $. From the point of view of the Lagrangian formalism. .302 7 The Canonical Formalism case fOT the simple example of the real scalar field III with nonderivative coupling.1. a negative constant here would lead to a Hamiltonian that is not bounlled below.(I1l)] .36) if we identify lP and [J with the canonical variables Q and P. becau. The Hamiltonian is now given by Eq.2') d 3x [ ~rr2 + ~(VI1l)2 + ~m211l2 + <*..2.2.<lJi}I''1I.ation of '11. with X'('1I) a quartic polynomial in '11. H ~ = J .17) which we recognize as the Hamiltonian (7. The most general Lagrangian that satisfies the principle of renormalizability (discussed in Chapter 12) is of this form..2. The remaining condition i. the special character of field variables like the time component of a vector field or the Hermitian conjugate of a Dirac field arises from the fact that although they appear in the Lagrangian. As we shall see. The Euler~ Lagrange equations here are (7. we calculate a canonical conjugate to 111: 0<1> ' which is the same as Eq.35). (7. their timederivatives do not. We have already seen in the previous section that there are field variables. but rather as a validation of the Lagrangian (7. yet Lorentz invariance dictates that these must appear in the Lagrangians fOT the vector and Dirac fields.2.14) as a possible theory of scalar fields. The positivity of the first two terms .14). (7.15) From this Lagrangian density. t In order for H to be interpreted as an energy.hows that we guessed correctly as to the sign in the lirst term in Eq. that are not canonical field variables Qn and do not have canonical conjugates.<1>0"<1>  2<1>2 . it should be bounlled below. . such as the time component of a vector field or the Hermitian conjugate of a Dirac field.14) which can be obtained by adding a real function Jf'(lP) of lP to the freefield Lagrangian density found in the previous section.Jt"(<I». ~<I> 02' .1. (7. the remaining independent field variables are the t We llo not inclulle a free constant factor in the term 0} <!!.10) as~ rr~.2.e any such constant if positive can be absorbed into the normali. Consider the Lagrangian densityt 2' ~ 1 2 <1. This little exercise should not be regarded as another derivation of this Hamiltonian.
to use perturbation theory we must make a transition to the interaction picture. C(t)] . The Hamiltonian is timeindependent.2.2. Section 7. Finally. We shall present a number of examples of this procedure in Section 7.C(t)] (x. C(t)] .30){7. the timedependence equations (7. .1)(7.7. the scalar field with Hamiltonian (7. In gauge theories like electrodynamics other methods must be used: either choosing a particular gauge.6 shows how in such cases one can avoid the task of actually solving for the cr and Qf. We split H into a freeparticle term and an interaction H~Ho+V Ho = V J ~J (7. for the moment we will give only one example of the simplest type. Because bLjtJCr = 0.32). as in Chapter 8.2 The Lagrangian Formalism 303 canonical variables Q".23) d'x ""(<1».17). JQ'(x.22) and the commutation or anticommutation relations (7.2.Q(t).21) d3 x [ ~n2 + t(V¢)2 + ~m2 $2] (7.20) o~ JL[Q(t). The Hamiltonian derived in this way may then be expressed in terms of the qs and ps of the interaction picture. so it can be written in terms of the PI! and QI! at t = 0.1. Q(t).22) (7.21) and (7.t) ' (7.2.t ) P.1. t) In the simple cases to be discussed in this chapter. or the more modern covariant methods to be discussed in Volume II. and split into two parts. which are equal to the corresponding operators PI1 and ql! in the interaction picture at t = O.2.2) are used to express the qs and ps in V(t) as linear combinations of annihilation and creation operators.18) can be solved to give the cr and Qf in terms of the Q s and P s.5. Q(t).1.1. but there are no canonical conjugates for the cr.2. The Q" have canonical conjugates _ JL[Q(t). (7.2. bC'(x.2. Once we have derived a Hamiltonian as a functional of the Heisenberg picture Qsand P s. a suitable freeparticle term Ho and an interaction V.2. The equations of motion of the their first timederivatives cr and (t in tenns of the cr involve only fields and (7.18) and satisfy the commutation relations (7.1.1. these equations together with Eq. Q'  L[Q(t). the Hamiltonian (7.19) but this is not yet useful until we express the Qs and P s. (7.10) is in general H ~L m J d3 x P.
and a(p) some asyetunknown operator function of p.26) then gives the canonical conjugate as .2.P>al(p)] (7.t) ~ .5. but as we shall see this is not to be expected in general.! (This happens to be the same relation as in Eq.tlL ~O. (7. (7. since they are defined by Eqs.1. (7.2.1.304 7 The Canonical Formalism Here ~ and II are taken at the same time t.2. bHo 2 2 n(x.29) to be equal at that time. t») .(x) ~ _i(2n)3/2 J d3p (po/2)l/2 [eiP>a(p) .25) The relation between nand (p is dictated by Eq.2./) .t).30) In order to get the desired commutation relations. (7. [</J(x.31) (7.eiP>al(p)] . (7.2.t) ~ +V </J(x. II with the interaction picture variables 4>.2.33) .t). To calculate the interaction V(t) in the interaction picture.28) (7.23).21) 6 ( ) ~ .(x.t) . tlL ~ i 63 (x . (7.t)+ ~ 1(V</J(X. </J(y.26) n x. we can simply replace 4>.2.2.24) The same transformation applied to Ho leaves it constant: ~ J . We now pass to the interaction representation.1. (7.2.y) .16). (7. expUHot) Ho exp( iHo(t)) d)x [ln 2(x. which together with Eq. I). (7.5) V(t) ~ expUHo/) V exp(iHot) ~ Ho ~ J d3x Jf(</J(x.2.tlL ~O.) Also. (7. the equation of motion for </J is dictated by Eg. we apply the similarity transfonnation (3.2.29) with pO = Vp2 + m2 understood. (7. [</J(x.6</J(x. though Ho and V usually are not.26) yields the field equations (0 .22) and (7. Eq. (7.t).2. (7.2.m2 )</J(x) ~ O.t).tl)2j .27) JHo The general real solution may be expressed as </J(x) ~ (2n)3/2 J d3p(2P')l/2 [e"'a(p) + e.m </J(x.2.32) (7.1.28) and (7. [. n(y. Taking t = 0 in Eqs. and H is independent of t.n(y.22): </J(x. 'Tt.(x.2.
• IQ(t)] Q"(x.~ What is less obvious and worth noting here is that a timederivative 80 :7'0 in the Lagrangian density also does not affect the quantum structure of the theory. that the fields vanish at infinity.2.2. . We can now proceed to use perturbation theory to calculate the Smatrix.(x. we have already shown in the previous section that using these • expansions in Eq.2.35) Also. ~D". (7. (7. and (7.2. up to an inconsequential additive constant.34) (7.14) as the correct freeparticle Lagrangian for a real scalar field. (7. To see this. This is under the usual assumption. with the field ¢(x) given by Eq.29).2. JQ"(x. These results do not necessary apply when we allow fields of different topology..2 The Lagrangian Formalism 305 we must take the as to satisfy the familiar commutation relations [alp).7. (7.t) ~.2. at(p')] ~ J3(p . It is obvious that such total derivative terms do not contribute to the action and hence do not affect the field equations. as discussed in Volume ll.2. (7.2. alp')] ~o.t).35) (which were obtained in Chapter 5 on quite other grounds) but rather as a validation of the first two terms ofEq. these results should not be regarded so much as an alternative derivation of Eqs.37) It follows that there is no change in the Hamiltonian as expressed as a . treating as equivalent any Lagrangian densities that differ only by total derivatives 0IIYII. The procedures illustrated here will be carried out for examples that are more complicated and more interesting in Section 7. As remarked before.t) .2.and xdependent functional of the values of Q at a given time.2.2.24) as V(t).2.25) gives the usual fonnula (4. ••• In considering the various possible Lagrangian densities for physical theories it is common to apply integration by parts.36) where D is an arbitrary n. (7. let's first consider the effect of adding a term to the Lagrangian of the more general form LlL(t) ~ J d 3 xD. taking (7. (7.11) for the freeparticle Hamiltonian.5. [alp). (7.34). This changes the formula for the conjugate variables p(t) as functionals of Q(t) and Q(t) by the amount J LlL(t) LlP.29). It is also obvious that a spacederivative term V' §' in the Lagrangian density does not contribute to the Lagrangian and hence does not affect the quantum theory defined by the Lagrangian.p').IQ(t)].
2. I) .36) has no effect on the quantum structure of the theory.2. and since. The commutators of the Q" with each other and of the Q" with the Pm are given by the usual canonical relations. I). I) + liPm(Y. t)] [IIP.41) In this case the commutator (7. Pm(Y.(x. then D in Eq.IIL(I) ~ O. 7. We have seen that a change of the form (7. I)Q'(X.y[Q(t)] I  bQm(y. Pm(y.39) In general this doesn't vanish. the addition to the Lagrangian of the term (7. I). I) + LIPm(y. LIPm(y. I).(x. t). that it provides a natural framework for the quantum mechanical implementation ..2. (7240) . I) JI + IIP.(x. I)] + [lIP.IIPm(y. I).(x.(x.38) Hence also there is no change in the Hamiltonian as expressed as a functional of the old canonical variables Q" and Pn . the Hamil· tonian is not the same functional of the new canonical variables Q" and PI1 + tlPn as it was of the Q" and Pn .bDm.(x. Different Lagrangian densities obtained from each other by partial integration may therefore be regarded as equivalent in quantum as well as classical field theory. However. and in a theory described by the new Lagrangian !f' + 1'1. (7. Pm(Y.39) vanishes.( x.bD". but the commutators of the Pn with each other are now [P.(x. (7.2. I) . as we have now shown.36) is of the special form bG[Q(I)] D".36) in the Lagrangian does not change the form of the Hamiltonian as a functional of the Qn and Pn.2. the commutation relations of these variables are also unchanged. (7.2./) bG[Q(I)]Q"( x.£ it is the new canonical variables Q" and PI1 + tiPn rather than the Q" and Pn that would satisfy the canonical commutation relations. (7.[Q] ~ bQ'(x.2.t)] __ . I)] ~ [P.t)] [P.[Q(/)] + .l) J bQ'(x. but if the added term in the Lagrangian is a total timederivative IIL~ dG dl ~j'd3 x bQ"(x.306 7 The Canonical Formalism functional of the Q( t) and Q( t): J d3x IIP.3 Global Symmetries We now come to the real point of the Lagrangian formalism./) . I) + IIP. so the variables Qn and Pn satisfy the usual commutation relations.t) .
This is the case. the principle of stationary action.3. ax J r 'I' (x) (7. in general. such symmetries are known as global symmetries.1) that leaves the action (7.3 Global Symmetries 307 of symmetry principles.3) invariant: o~ M ~ <E . including variations of the form (7. we see that P(x) must satisfy a conservation law: o ~ o]l'(x) .3) then.2) is automatically satisfied for all infinitesimal variations of the fields if the fields satisfy the dynamical equations.3.3.2. Consider any infinitesimal transformation of the fields If'{(x) + If'/(x) + if'~t(x) (73. the variation of the action will not vanish. Many symmetry transformations leave the Lagrangian and not just the action invariant. so in this case (7.5) M ~ x' It follows immediately that o ~ dF dt ' (7. often referred to as Noether's theorem: symmetries imply conservation laws.3. This rep· resents a general feature of the canonical fonnalism. (7.5F{ (x).3. (7. Ifwe now consider the same transformation with E' an arbitrary function of position in spacetime: '1" (x) ~ 'I't (x) + i«x) S" (x).4) should vanish.2) (With E' a constant.) Of course Eq. This is because the dynamical equations in the Lagrangian formalism take the form of a variational principle. by an infinitesimal symmetry transformation we mean one that leaves the action invariant even when the dynamical equations are not satisfied. In general.6) where F I d]xJo .3.4) ox 1j in order that it should vanish when E'(x) is constant. but it will have to be of the form Ia'x J'(x)o.(x) (7. If we now take the fields in I [If'] to satisfy the field equations then I is stationary with respect to arbitrary field variations that vanish at large spacetime distances. for translations and . . (7.3.7) There is one such conserved current ]P and one constant of the motion F for each independent infinitesimal symmetry transfonnation.7. for instance.I" M['I'] . Integrating by parts.7/ depends on the fields and their derivatives at x.3.3.3). 0 (7.
and write an explicit formula for the conserved quantities F. In such cases we can go even further.'P(t)]3"'( t)+ 6L['P(t).3. . and write an explicit formula for the current P'(x). Other symmetry transformations such as isospin rotations leave not only the action and the Lagrangian invariant but also the Lagrangian density. t) ~ ( ) ~t( 6'1' )l] .'P(t)] d 3"'(x t)] (739) 6'1"(x. + 6L['I'(t).2. (7. Writing the action as in Eq. t) .6) as the integral of the Lagrangian density.1 308 ! 7 The Canonical Formalism rotations in space and also isospin transformations and other internal symmetry transformations.'#' x.t " (x. so for general fields (whether or not the field equations are satisfied) the variation in the action is M ~ iJdt !d]x 6L['I'(t).t).9). In this case the variation in the action is M ~iJ dt J d 3x [JL['I'(t)..8) The requirement that the Lagrangian be invariant under this transformation when E is a constant yields 0~Jd3X [JL['I'(t). 'P(t)] 3"(x t). the reader can easily check that this F is indeed timeindependent for any fields that satisfy the dynamical equations (7. (7. Consider a field variation (7.10) Comparing this with Eq. 'P(t)J E(t)3'" (x t) J'I't(x.3. t) x. t) . When the Lagrangian is invariant we can go further. 6'1"(x. though not for general Lorentz transformations.3. 6'1"(x.3. its variation under the transformation (7.4) gives F ~ iJ d]x 6L['P(t).3) in which E(X) depends on t but not x. 'P(t)] '(t)3"(x. 'P(t)] dt ( £of.2. (7.2).12) The invariance of the Lagrangian density when E is a constant requires . t) dt ' . (7.3.3.3. 6'1"(x.3. . t) (7.11) Using the symmetry condition (7.3) with a general infinitesimal parameter E(X) is (7.
(7.3.t) ~ i Ja Ja 3 x 3 yp.'I"(X)) u.'I"(x)) ".x].t)Y'[Q(t). (7. For such transformations.3.13) so for arbitrary fields the variation of the action is "T 'I[m] ~·f.t) at an arbitrary time t. The quantum properties of the conserved quantities F are most easily seen for symmetries of the Lagrangian (not necessarily the Lagrangian density) that transform the canonical fields Qn(x. o2'('I'(x). To see this. and then use the equaltime .tW'(x. (7.3.t) ~ .3. (7.13). €( ) '.7.11) in the form F ~ i Ja 3 xp. ' (7.3.3. hence we may rewrite Eq. those of the 'PI whose time derivatives appear in the Lagrangian) into xdependent functionals of themselves at the same time. 0.14) Comparison with Eq.3. the functional derivative i5Llti'itl is equal to the canonical conjugate Pn.3.""[Q(t).3.15) Using the symmetry condition (7. For all such symmetries the operator F is not only conserved.'F n a linear functional of the Qm. we can invoke Eq. but we will not need to assume here that the symmetry is linear.17) To calculate the commutator (not anticommutator) of F with a canonical field Qm(x.3.'I'(x)). ax o(o. (7.(x.16) As we shall see.'I'(x)) Y'(x) o o'l"(x) 02'('I'(x). (7. note first that when 'PI is a canonical field Qn. (7. it also acts in quantum mechanics as a generator of this symmetry. 0" 'I'(x)) .6) to evaluate F as a functional of the Qs and P s at the time t.3 Global Symmetries 309 that ~ 02'('I'(x).o. Y'(x) + O(O.2. it is easy to see directly that DpP vanishes when the fields satisfy the EulerLagrange equations (7.3. Note also that the integral of the time component of the current (7.3.15) has the previously derived value (7.9). with . So far everything we have said would apply to classical as well as quan~ tum mechanical field theories.11). this functional derivative vanishes. while when 'PI is an auxiliary field cr. infinitesimal spatial translations and rotations as well as all infinitesimal internal symmetry transformations are of the form (7.(x.x]. t) (that is.16).""(X. x.3. we have .4) shows that off =1 JP' '0(0'1" /ox') !#' (7.1).
~. conventionally grouped together in the energymomentum tensor Ti' \' : (7.~"T.3.17) and the canonical commutation rules give also [F.1.t)]_~ J 3F m (Q(t). Eq.t)VQ"(x.t) 3Q"(x. Thc only exceptions are certain symmetries known as supersymmetries.(x.3.3.3.310 7 The Canonical Formalism canonical commutation relations (7. (7. 1 (7. J dJxP.25) Using the equaltime commutation relations (7. As a first example. with four independent parameters corresponding transfonnation functions ${ m{ . so in accordance with the above general results we can conclude that the spatial components of Pv take the form p.1).1.30}(7.. (7. .3.y) d'yPm(y. This is of the fonn (7.3./ d3 x TO. consider the symmetry transformation of spacetime translation: 'I'{ (x) ~ 'I'{ (x + <) ~ 'I'{ (x)+ <'8" 'I'{ (x) . (7. so that F is bosonic.'I'.19) tells us that PI1 transforms contragredicntly to Q".3.3.18) is an anticommutator if Q" is also fermionic.3. Eq. (7. (7. we also find the commutator of this operator with the canonical fields and • We are here assuming that for QIJ bosonic or fermionic thc variation iF" is also respectively bosonic or fermionic.1.P"(x.t) . Q"(x. . for which f' is fermionic and (7.32).3.30){7.24) The Lagrangian is invariant under spatial translations.32) to obtain· [F.19) Where F m is linear.t).3. (7.22) from which we can derive time·independent quantItIes as the spatial integrals of the time components of the translation 'currents' (not to be confused with the canonical conjugate field variables PI1(x.t). '8 In consequence we have four independent conserved currents. t)): P.16). dt Pv = O. (7.1.318) It is in this sense that F is the generator of the transformation with Eq.t)]_ ~ 3'"(x.20) tel' and four (7321) .23) d (7.
26) (7. it is the Hamiltonian pO H.3.(x.30) 1S U 'I["'J T ~ Jd x (OY "' ".3.32) Integrating by parts. so JJ M['P] = / trx (~~EIl + O(~~{)OIIIfl18YEIl) (jJ = .t)]_ explicitly on x. Instead. (7.t). so we cannot use Eq. we see that this takes the form of Eq. If we further assume that the Lagrangian is the integral of a Lagrangian density.3. However.t = 0 . then we may also obtain an explicit formula for the energymomentum tensor Til v .3. (7. However. which as we know satisfies the commutation relation = [H.3.34) As a check. (7. iVP.t5) here. sJ(x. (7.3. (7.{ + DlfltE 4 tilT oY.3..7./ (7.t)J~ ~ iVQ'(x. note that the change in the action under a spacetimedependent translation (7.31) The EulerLagrange equations (7.4) d4x TV"o"e JJ (7.tll ~ ici'(x.t).33) with 'currents' Ty Il  uIJ no!f _ (/ :1(' iJ.5i' :1 m! ()UII Ov IfI T ( 7.(x. Q'(x.3 Global Symmetries 311 conjugates: [P.3. In contrast.3. we have It follows that for any runction 'lJ of Qs and P s that does not also depend [p.29) for any function 'lJ of Heisenberg picture operators. while for }.2.27) [P.3.3. (7.3. timetranslations do not leave the Lagrangian L(t) invariant.25) for P. we may note that the spatial components of Eq. E 8(8"IfI{)C.23) are the same as our previous formula (7.t) (7. sJ(x)] ~ iVsJ(x).9) show that the terms proportional to e add up to EIl O 5i'.3. the Lagrangian density 2(x) is not invariant under spacetime translations. we already know the generator of timetranslations. [" OJJT "'fJ) . P.28) These results show that the operator P is indeed the generator of space translations.
under a set of linear coordinateindependent transformations of the canonical fields (7. (7. (For instance..3.40) The equal·time commutation relations here give [To. (7. (7.3. in electrodynamics there is such a symmetry.41) (7.. for which the one matrix tTl m is diagonal.3.34) is not in general symmetric.(x)] ~ +(t. and we sum over repeated group indices a.3.3. b.Q'.42) (Where ta is diagonal.'i']. we can infer the existence of another set of conserved currents Jg: (7.39) When the Lagrangian as well as the action is invariant under the trans~ formation (7.3.23) gives the usual formula for the Hamiltonian: HPo~ jd]x [~P.11) provides an explicit formula for the T a : T a = i J d 3 xP. and therefore cannot be used as the righthand side of the field equations of general relativity. with the charges carried by each field on the main diagonal. we can calculate the commutator of Ta with .(x.) In many theories there are also one or more symmetry principles that state the invariance of the action. Eq. (7. etc. [T.36).Pm(x) . The correct energymomentum tensor to use as the source of the gravitational field is the symmetric tensor E)/lV introduced in the next section. Q'(X)] ~ (t.)m.312 7 The Canonical Formalism Eq. (7. this tells us that Q" and P" respectively lower and raise the value of Ta by an amount equal to the nth diagonal element of t a .36) together with a set of suitable transformations on any auxiliary fields C': (7.3.3.3.3. P. (7.35) (A warning: the tensor T/l V obtained by raising the second index in Eq.)' mQm(x).38) whose time components are the densities of a set of operators time~independent Ta=JiXJ~.) From any such symmetry.3.3.) Using these results.37) Here t a and Ta are sets of Hermitian matrices furnishing some representa~ tions of the Lie algebra of the symmetry group.t). (7.t)(t a tmQm(x.
45) This confinns that the quantities (7.7.t)] ~ J)(xy)(t. since the Lagrangian density does not involve timederivatives of the auxiliary fields.3.)' mP.49) (7.Tb]_=ij d 3x [~Pm(ta)mn(tb)\Qk+ fYn(tb)\(ta)kmQm].y) (t. (7.3. t). t). ta iJ(8C iJx") ase / ()'s .15) to provide an explicit formula for the currents associated with these global symmetries: jJ _ .!m2 (J)I . TbJ= ifab(' T c · (7. (7.47) This is invariant under a linear transfonnation like (7.3. In particular. with Lagrangian density :f =  2 ! aJl~lall(J)1 . [Ta . (7.50) [J~(x.3 Global Symmetries 313 another generator Tb.= ifabCt('.3. J a = 1 8(iJQ' iJx P ) ase / ()' mQm .t).(x. tbJ. If the auxiliary fields are constructed as local functions of the P sand Qs in such a way that they transfonn according to a representation of the .3.I. 't"a C' (7. we have J~ = iPn(tat mQ m .3. Where the Lagrangian is the integral of a Lagrangian density which is invariant under (7.46): Jil = <D2aJ'~1  ~1 c)ll~2 . (7.3. but also with the densities J~: [J~(x.Jf'(~I + ~~). and use Eq.36): so there is a conserved current (7.4m (J)1. Q'(y.3.40) are correctly normalized to qualify as generators of the symmetry group.46) for the current can be used to derive other useful commutation relations. [Ta.3.44) then so do the quantum operators T a . Pm(y.3. (7.3. (7. The explicit formula (7.4all~2ajJ(J)2 .)'mQm(x.t)] ~ J)(x . if the matrices fa fonn a Lie algebra with structure constants tab C .43) Thus.3. suppose we have two real scalar fields of equal mass. 46) As an illustration. [ta.37) we can go further.3.3.t).36) and (7.3.48) We can then derive the equaltime commutators of general fields not only with the symmetry generators Ta.
49) and (7.3) (7. one current for each independent component of Will" The integrals of the timecomponents of these 'currents' then provide us with a set of timeindependent tensors: JltV / d3 x . then also (7.3. Consider an infinitesimal Lorentz transformation (7. so we cannot immediately use the results of the previous section.3. translation invariance allows us to formulate Lorentz invariance as a symmetry of the Lagrangian density .f)] ~ _03(x .J'.tl' . However.A(f'VIl .49){7. (7. but Lorentz transformations act on the coordinates and hence cannot leave the Lagrangian density invariant. f).y)(T.3. JI" ~0. the invariance of the action under such transformations tells us immediately that there are a set of conserved 'currents' uIt PIlV : iJ p . 7.4.4.4 Lorentz Invariance We are now going to show that the Lorentz invariance of the Lagrangian density implies the Lorentz invariance of the Smatrix.51) will be used in Chapter 10 to derive relations called Ward identities for matrix elements involving the current jJl. We would like to have an explicit formula for the tensor Jt(lJ1\'.4.3. C'(y.4t'" jt PIlV = ~ 0.1) (7.4) _.A/0J.3.5) :.52) Commutation relations like (7.314 7 The Canonical Formalism symmetry algebra with generators 'a.4.51 ) [J2(x.3. (7.4.6) The JltV will turn out to be the generators of the homogeneous Lorentz group. (7. t).4. We often summarize Eqs.2) According to the analysis of the previous section.C'(x.51) in the single commutation relation (7. (7.
.7) transforms like another such field. B) we have Jq. we have an irreducible field of type (A. ~A = /1}P/i.4 Lorentz Invariance 315 under a set of transformations on the fields and fieldderivatives alone.j jJV are a set of matrices satisfying the algebra of the homogeneous Lorentz group (7.?11I' )t. so fll" = 0.:'Pt) (. w"'(.4.9).<~! v/:'P .2" D'P{ (f/lv m i iJ!I! + )1 ~ In 2" O(OK'P1) J'II" mVK'P i rl ( o!l! ( + 2" OU\: 'Pt) ('1I.34) for the energyMmomentum tensor Till" we may write this as o= OK 2 o(o. )' ". respectively. '(~ ) . We specially note that for a covariant vector field. d" iJ(iJK'P ') 2 )( .d. for a scalar field q. while for . where .4.9) The Lagrangian density is assumed to be invariant under the combined transformations Eqs.8) For example. The derivative of a field that transforms as in Eq. but with an extra vector index (7. (7. 7. so here ( "PtJ ) Ie ) . II" mUK "...m ./l)'P .1iBc).3.m T + (."0 i(. . so o_ o'P1 if!£'  2" W iJ!I! i IIV( fill' )! 'Pm m + "'"~cc!i'= !..rrI and JI are spin matrices for spin A and B.4.. (7. Setting the coefficient of (J)llV equal to zero gives o_ )t 'Pm .7) and (7.4./lC" 1 o!l! ~ . + IlJlTKUp ~A . OK'P (WI.9).4 7) • where .T"/l) ..10) This immediately suggests the definition of a new energymomentum . Using the EulerLagrange equations (7.2 Till' .. m'P m] . = ~ 0.'1Kl'(. and our formula (7. The fields undergo the matrix transformation UT nu! = 2 ~ w" (. (7.2.f!tr .4. d"m T (7 . (j VIC = OJ/( A VA.[ i c!i' 1( .4.
4.lV satisfies the same conservation law as TjJv.3. unlike TJlV.  xVeo~). (7. However. (7.4. so the derivative term here drops out when we integrate over all space J0o'd]x~ JTO'd3x~p'.4. D(D. just as well as TJlv. we have O J3(tO j 0 Ok .'Pr)" m o(o. (7.l = 0 in Eq.18) Also.\' that acts as a source of the gravitational field. (7. we may construct one more consenred tensor density: 8J..J tk 1= "2 €jtk e OI'.. in the sense that O. the Belinfante tensor ell" is not only conserved but also symmetric: (7. Ie.16) Thus Lorentz invariance allows us to define one more timeindependent tensor J~" = J J( 0IlI'd 3x = Jd3x(x~eOV ~o.xkox j S Ot) d x x ox = i€ijk Jd3 e Ok x .J] (7.4.l and so 0J. known as the Belin/ante tensor: 2 0" ~ T" _ 2 ~0 [ K o(o"lf'() / o!£ (. (7.4. applying Eq. so it commutes with the Hamiltonian [H.H.4.'P t ) " m (7411) •• The quantity in square brackets is manifestly antisymmetric in f.J(AIl" = Sill' .el'~ = O..hl = 1 j 2 €itk[Pj.4.15) This is conserved.lV (7.14) It is rather than pl.. but also has no explicit time·dependence.11) the index K runs over space components only. Thus ejJv can be regarded as the energymomentum tensor. (7. when we set f. 3 In consequence of the symmetry of e llV .17) The rotation generator Jk = €ijkJijj2 is not only timeindependent. Eq.4.''')' 'Pm m _ o!£ (.28) to the function [Pj. (7.13) where pO .)r 'Pm _ o!£ (d")' 'Pm].12) For the same reason.316 7 The Canonical Formalism tensor.0" ~0 .4. 0.10) tells us that.
O)e~iHot must vanish for t _ +ro in order to allow the introduction of 'in' and 'out' states and the Smatrix. and conclude that the Smatrix is Lorentzinvariant.J kO . •• • The same arguments were also used in Section 3.3.4..7.4. K]. Also. or more explicitly (7.19) On the other hand. (7. though timeindependent.iHot vanish· for t ~ +O'J.Kk] = i and therefore J d xx 3 k o~ 0 00 = iliik J 3 d xe oo [Pj. (Note that the interaction tenns in eiHot I d3x eoo(x.23) Since the Lagrangian density does not depend on the timederivatives of the auxiliary fields.20) will be 'smooth' in the sense used in Section 3. the 'boost' generator Kk .4.4. which here take the form ij = J d 3x ::r (xiv/P{ .4. (7.3 to verify that the remaining commutation relations of the Lorentz group.K] ~ ~jP. and the rotation generators do not mix canonical and • When we say that some interactionpicture operator vanishes for t 4 ±cc.3.21) in hand. the interaction terms in eiHot I d3x x eoo(x. KkI ~ j oJ' H . .20) Since this is a constant. (7.4 Lorentz Invariance 317 and therefore [Pj.3.O)e.22) For any reasonable Lagrangian density. take the proper fonn. we have ° = It = p + i[H. we mean that its matrix elements between Slates lhat are smooth superpositions of energy eigenstates vanish in this limit.28) again gives [Pj. does explicitly involve the time coordinate Kk = J d 3 x(x k e OO  XOe Ok ) .) With this smoothness assumption and the commutation relation (7. those of the jllV with each other. applying Eq. we can repeat the arguments of Section 3.e.4. This can also be shown directly for the commutators of the rotation generators.XiOiqJt . i.21) [H. and therefore (7. the operator (7.i CtiJt mqJm) . (7.4.Jd = ~jf'ijkPk.
1) where JJl is either a c~number external current (unrelated to currents lJl introduced earlier).. .5.P. Cl'J)' "Q' (x). or a functional of various fields other than iD (in which case terms involving these other fields need to be added to (7. .5.)Q'(x) [pi.4. this can also be written as a sum over canonical fields alone: } 11 = that .3. but now with derivative coupling. (7.)P'(x) · .i (. Thc commutator of the auxiliary fields must thu> bc consistent with the rotational invariance of L. so the direct proof of the commutation relations of the JiO with each other has to be given on a case by case basis.5 Transition to Interaction Picture: Examples At the end of Section 7. I (7. Q'(X)]_ ~ i(x.Jijr /l1'Pn') .4.2 we showed how to use the Lagrangian of a simple scalar field theory to derive the structure of the interaction and the free fields it contains in the interaction picture. this is not needed for the proof of the Lorentz invariance of the S~matrix given in Section 3. 7.4. Jij with . it commutes with L. sinl:e Jij commutes with Hand PnQ".(x)]_ ~ i(x.!m2 <1>2  lJloJliD . We take the Lagrangian as f£ =  ! oJl<1>iWiD .1)). We will now turn to somewhat more complicated and revealing examples.(x). (7.xjO.26) + (J'i1)" These results can be used to derive the usual commutation relations of the Jij with each other and other generators"· If there are no auxiliary fields then the same arguments may be applied to the 'boost' generators to complete the demonstration that the P!t and JII\' satisfy the commutation relations of the inhomogeneous Lorentz group.24) It follows immediately then from the canonical commutation relations [J'I. . Scalar Field. P.xjO. Fortunately. Derivative Coupling First let's consider a neutral scalar field.I 3 d XP/l (Xia jQ/l  Xj OiQ/l . However the 'boost' matrices fiO will in general mix canonical and auxiliary fields (such as the components Vi and V O of a vector field).318 7 The Canonical Formalism auxiliary fields.oi . The •• Also.25) (7.oi .Jt'($).
with the remainder called the interaction.2.2.m'</>'(x.'<1>+ Hlo).4) (7.35). we can write this as Ho V = J + V. Indeed. and leads as in Section 7. The last step is to replace 11: in the interaction Hamiltonian with its value (p in the interaction picture (not its value (p_J o in the Heisenberg picture) : .5./ d 3x[010 +J. l)l o(x.. (7.l) + j ('. '.5 Transition to Interaction Picture 319 canonical conjugate to <1> is now 0~82' ~<i>10 8<1> (7. +*'(<1»] .7) The freeparticle Hamiltonian is just the same as Eq.35).2.6) V(I) ~ J 3 d x [rr(x.7.5. though we will not bother to indicate this explicitly): Ho~ J X[ 3 d jrr'(x.5) ~. (7.5. As explained in Section 7.I)]' + "'''(</>(X.[10(x.5.'</>(x.'<1> + 1°(0 + 1°) + *'(<1»] . (7. because as we have seen it is this form of the freeparticle Hamiltonian that leads to the correct expansion (7. I) +J(x.26)(7.tJ]. + jm' <1>' + H Ho Collecting terms. whatever the total Hamiltonian may be.2.3) n2 d3 x [ 1 + WV'<1»2 m211l2 +1 ] .))] . (7.2.5. I»)' + .25). (7.I) + .I)' '. we must take (7.1(0+1°)' J. we can pass to the interaction picture by simply replacing nand I1l with 11: and <b (and likewise for any fields in the current JII.34). '.2 to Eqs.29) of the scalar field in terms of creation and annihilation operators that satisfy the commutation relations (7. (7.5.2) and the Hamiltonian is H J ~J ~ ~ d3 x[0<i>  2'1 d 3x[0(0+10) + 1('.2.'</>(x.'<1»'. (7.6) as the part we split off and call the freeparticle part.5.2.
5.. V' + m2 a.5.2 is needed to cancel a noninvariant tenn in the propagator of or/>._ (iJp Vi') + m 2 VI' ~ = Jil • Taking the divergence gives (. and m 1 are so far arbitrary constants.5. Vector Field. can be absorbed in the definition of Vp . and III is either a cnumber external current.320 7 The Canonical Formalism The extra noninvariant tefm in Eq.J. VO .5.!m 2 Vi'Vii . and can therefore be used as a constraint that .III VV . We want to describe a theory containing only particles of spin one.V.~~ a Vv 811 VV . in which case additional terms involving these fields must be added to :e. and VO is thereFore an auxiliary field. so yO does not appear in the Lagrangian. This causes no serious difficulty: the fact that J:£ I ayO vanishes means that the field equation for V O involves no second timederivatives. The EulerLagrange field equations for VII read cr:O VV + f3 0.. The constant a. so the Vi are canonical fields. U} _ _  iF I V' 4 pv FP".5.8) isjust what we saw in Section 6.. + fJ).13) This is nonvanishing for 11 a spatial index i. (7. P. we take IX = fJ." I(a.2m 2V/1 V'J I t . V' all! .5.10) This is the equation for an ordinary scalar field with mass m 2 /(a.11) where (7. Spin One Similar results are obtained in the canonical quantization of the vector field Vp for a particle of spin one. (7. (7. Let's here keep an open mind.9) p :e where ct." can be expressed in terms of an external current or other fields. and therefore oZ. or an operator depending on fields other than Vii. oY 0 (7. as J. + p)oa.5." 1m 2. with conjugates II' ~ F'o ~ V' + a. so to avoid the appearance of 8j yl as an independently propagating scalar field. (7. + fJ) and source JAJ. (7.. in which case J.14) On the other hand F on = 0.!/3 op Vv OV Vi' . and write the Lagrangian density in a fairly general form = . not spin zero. so we can take IX = fJ = 1.12) The derivative of the Lagrangian with respect to the timederivative of the vector field is ayp=FI'.
5 Transition to Interaction Picture 321 eliminates a field variable.5. (7. Eq..m2V(V·. though not shown explicitly. + 2~2 (10)2] .2]. (7. it is not related by a similarity transformation to any interaction·picture object va. m H ~ J  d3x [II2 +m2(V· II)(V' II .14) allows us to write V in terms ofn and J O : .5. n  +J . n  jO) ] Again. (7.V  m. the EulerLagrange equation for v = a is (7. (7.vf+ jO)2 ~m2V2 ~ m.') ~.2') for this theory. (7.)2 + ~(V X.2 jO(V .5. The relation between and v is then .)2 + ~2 . Instead.) J.5. 1 0 m (7.5. likewise for whatever fields and conjugates are present in jll): Ho V ~ Jd3 x [~.m11: 2 JOV·.5.16) Now let us calculate the Hamiltonian H = I d3x (n· V .V(V· IIJ ° ). V ~ VV°+II so ~ 1 II.5.15) or using Eq.20) J.~ bHo('.7.5.1I:) =+V2vV(V'v)m2v.14) V0 ~. Specifically. we can invent a quantity (7.22) .21) Since VO is not an independent field variable.J O) ~n2+ Hv.(v·nJ).5.19) ~ Jd'x [J" .18) (7.17) and pass to the interaction picture by replacing the Heisenberg·picture quantities V and n with their interactionpicture counterparts v and 11: (and. and the 'field equation' is 1i:=_OHo(v. (7.2 + 2~2 (V' .5. we split this up into a free·particle term H o and interaction V: H =Ho+ V.2(V .
a)e'P"}. a) and (7.25) and (7.5. V. v) '1 . (J) for vectors satisfying = + 1.5.26) can be expressed as a Fourier transform v'(x) ~ (2.5.20) then allows us to write as (7. and (7. (7. (7.5. (T where pO = J P 2 + m 2 .27) + e"'(p.29) to calculate that v and 1t satisfy the correct commutation relations ["(X.28) and normalized so that L eJ1 (p. vj(x.23) x=v+Vvo.5.322 7 The Canonical Formalism 1t Eq.a)at(p. I). [. (7. a(p'.5.')] ~ O..I)] ~ 0.I).26) A real vector field satisfying Eqs.).5. (7.30) at (p.j(y.a) = 'II!" + pl'l/m2 .)] ~i6'jJ'(xy).I)] ~ ["(X.5.m2 vO = 0.)3/2 I: " J d3p (2 pO)1/2{ e'(p.5. a) satisfy the commutation relations (7. provided that a(p. tT) are operator coefficients.29) and the a(p. (7. [a(p. (7.5.22) and (7.vl' = a (7.0. It is straightforward using Eqs.5...27).32) . (7.'(X. nj(x.m 2 )" ~ 0 .5.5.a)e V O(p. V2y _ V(V . 1 are three independent (7.a).31) [a(p. ale'"~ (7.5..24) Taking the divergence gi yes c/. a)a(p.21) gives our field equations in the form v 2VO + V .a')] ~ 6 3 (p' p)6 c'c.5.5.Vb? ~ m2y = 0 .a).23). (7. (7. These can be combined in the covariant form D v Jl j)IIOVVV  m 2 v ll = o. Inserting this in Eqs.5. at(p'.5.25) and hence (0 . the el'(p.
J!"('I'. This is not real. but rather as proportional to the canonical conjugate of \fl. It is easy to check also that Eq.5.27). '1') . + m)'I' . The Hamiltonian is H ~ .I d3 x[[]. we tentatively take the Lagrangian as !£ ~ 'I'(y"a. Spin One Half For the Dirac field of a particle of spin 1/2.19) yields the interaction in the interaction picture V(t) ~ . (7. Dirac field.18) gives the correct freeparticle Hamiltonian for a massive particle of spin one.p !£] ~ .5.22) in Eq.u). 0'1' ~ 'I'y  0 (7.u)a(p.I d'x [J.37) (7. 'l + m]'I' . '1') (7. (7.5.I d x nyo!y .5.35) so we should not regard \f' as a field like \fI. as L" f d3p pO at(p.>1'] .V" + 2~2 (J O)'] .(qt}'llop\fl)t = 'Py P8p\fl + (op'P)'l\fl = cp(qtyP\fl). J  . (7. (7. Finally. but the action is. The canonical conjugate to 'P is n ~ a!£ .33) is just what we found in Chapter 6 is needed to cancel a noninvariant term in the propagator of the vector field.7.38) d x .. 3 (7.5. as necessary if we are to avoid having too many field equations. using Eq.5.5 Transition to Interaction Picture 323 We already know that the vector field for a spin one particle must take the form (7. Hence the field equations obtained by requiring the action to be stationary with respect to If are the adjoints of those obtained by requiring the action to be stationary with respect to \fI.36) where Ho ~ ~ V .5.5.34) with £' a real function of qt and \fl. (7.5. because qtyf1iJp\fl .I d3 x[ ny o[y.I . (7.5.18) may be written (up to a constant term) in the standard form of a freeparticle energy.5.>1'('1'.33) The extra noninvariant term in Eq. We write this as H ~ Ho + V. so our derivation of these results serves to verify that Eq. 'l +m]'I' +. (7.5.
5. In order to obtain the desired anticommutators [<p. This gives the equation of motion 1P ~ oHo on ~ YO(Y .u(p.5.(y.39) Likewise. + D are the two independent solutions of (7.46) '" p L. (7.44) '" _ (iy'p. eigenvalucs ±m. + mlj2m amI {i..P'b' (p.5.5.5. a(p.5.p yO . +m)v(p.45) (7.a)~ _ (iy'p.1. a)e.47) (7. The overllll sign is determined by positivity: Tr :Euufi = Lu t u and Tr :EVttJ = :Ev t v must be positive. yields just the adjoint of this one. so :Euil and :EVt must be proportional to the projection matrices (i)'l'p!. 20 + m) . (7.37) and (7.5.29) yields immediately n ~ .(x.pp(y. tlt(y0).5.p (7.P'a(p.y). a) + v(p.v(p.a) normalized so that· ~ 0 (7. + m)'I' ~ O. .5. . respectively. and u(p. p where pO yip 2 + m2.42) (iy'p.5.. +m)u(p. V + m)v.5. + mjj2m.5. ft = bHo/ btp. L.p"](x .. a) are operator coefficients.48) ['I'. a) e. The proportionality factor may be adjusted up to a sign by absorbing it in the dcfinition of 11 and v. the similarity transformation (7. t). fJ) and bt(p. Since Eq. (7. tlt ~ 0. Ho and V(t) can be calculated by replacing 'P and II with tp and 1t in Eqs. 'l'p(y. t). hal. (7.(x.) Any field satisfying Eq.5. (7.(x.a) ~ 0 and likewise (7.41) (The other equation of motion. (7. a)} .5.35) does not involve the time. (7.43) (iy'p. tlt ~ ['I'.a) ~ 2 p °+ m) ' p p (7.a)v(p.a)u(p.28). n.38). t). • The matrix il'p/.40) or more neatly (y'o.1.41) can be written as a Fourier transform 'I'(x) ~ (2n)]/2 J d] P L {u(p.. ~ i(yo).324 7 The Canonical Formalism We now pass to the interaction picture.l'p!.
0 )b(p. The standard analysis of this problem is that of Dirac. 0) + bt (p. returning to Dirac's approach in the next chapter.11) of a massive vector field VJl interacting •• Note the negative sign of the cnumber term. The conje. otherwise here. b(p'. bt(p'.p)]. a(p'.7.52) is only important if we worry about gravitational phenomena.ol.5.5.501 (7.:ts the numbers of boson and fermion fields. Primary constraints are either imposed on the system (as when in the next chapter we choose a gauge for the electromagnetic field) or arise from the structure of the Lagrangian itself. olbt(p.o'it ~63(p'p)6u'"' [a(p. thus verifying that (7.o'it ~ [a(p.o'it ~ [b(p. (7.at(p'. In tenns of the as and bs. "I) .o')t ~ 0. For an example of the latter type. plus another infinite cnumberu Ho ~ L Jd3p pO [at(p. consider the Lagrangian (7.o).6 Constraints and Dirac Brackets The chief obstacle to deriving the Hamiltonian from the Lagrangian is the occurrence of constraints.5. o)a(p. H o is a positive operator.5. Dirac's analysis is not really needed for the simple theories discussed in this chapter. in such a way that the cnumbers in Ho all cancel. We shall use the theory of a real massive vector field for illustration here. These agree with the results obtained in Chapter 5. With this understanding.5 whose terminology we will follow here.o'it ~ [b(p. b(p'.o).").5.o). . (7.5.b'(p'.37) is the correct freeparticle Hamiltonian for spin !. (7. 7.5.o')]+ ~ [a(p. 0) u 6 3(p . we can throw it away.511 We can rewrite this as a more conventional freeparticle Hamiltonian.6 Constraints and Dirac Brackets 325 we must adopt the anticommutation relations [u(p. (7. where it will be actually useful. as for the scalar field.49) and their adjoints. just as for bosons. this Hamiltonian is Ho ~ L Jd3p pO( ut(p.:tnred symmetry known as supersymmerry4 conne.o).521 The cnumber term in Eq. where it is easy to identify the unconstrained canonical variables. since it only affects the zero of energy with respect to which all energies are measured.o) u b(p. o)a(p.
6.4) More generally. We should then define the conjugates II. Such Lagrangians are called irregular.1) where (7.3) We immediately find the primary constraint: (7. Then there are secondary constraints. etc.16) for Va: (7. iJL c'P<l (7. .2) Suppose we try to treat all four components of V" on the same basis.6) .F '(" _ oSf' _ 0". The constraints we have found for the massive vector field are of a type known as second class.oa(t) and their timederivatives \jJa(t). To explain the distinction between first and second class constraints. ~ o vo A J1 ) . and the prescription used to deal with second class constraints.6. (The Lagrangians of quantum field theory are a special case..5) Here we are finished.6. which arise from the requirement that the primary constraints be consistent with the equations of motion. this is just the EulerLagrange equation (7. with the index a running over all pairs of t and x.) We can define canonical conjugates for all of these variables by II" ~ . \jJ) that depends on a set of variables q. b2 LjlJ(oo\f't)b(oo\f'm) has vanishing determinant.5.6. For the massive vector field. This will be the case if and only if the matrix.326 7 The Canonical Formalism with a current J /1: (7. and so on. it is useful first to recall the definition of the Poisson brackets of classical mechanics. secondary. we will treat them all together here. The distinction between primary. . but in other theories we might encounter further constraints by requiring consistency of the secondary constraints with the field equations. There is another distinction between certain types of constraint that is more important.6. we encounter primary constraints whenever the equations ITt = lJLjooo"P( cannot be solved to give all the eJo'Pt (at least locally) in terms of ITt and 'P t . constraints is not important. for which there is a universal prescription for the commutation relations. Consider any Lagrangian L('P. (7.6..
but may instead be related by various constraint equations. and time arguments are everywhere dropped.12) con~ (In field theory these are local transformations.9) [A. BC]p ~ ~ [B.A]p. the set of first class constraints XN = 0 is always associated with a group of symmetries. (7. The Poisson bracket is then defined by [A . But the constraints do not always allow this. (7. Il b] = 0. (Here and below all fields are taken at the same time. the set of all the constraints is necessarily consistent with the equations of motion A = [A.H]p. C]p]p + [B.6. We call a constraint first class if its Poisson bracket with all the other constraints vanishes when (after calculating the Poisson brackets) we im~ pose the constraints.6.B]p. In particular. with the constraints ignored in calculating the derivatives with respect to \fIa and n a . and therefore (7.7.A]r. The constraints may in general be expressed in the form XN = 0. we always have ['¥a. B]p]p O. because the index N .) These brackets have the same algebraic properties as commutators: [A. C]p. We shall see a simple eX<lmple of such a constraint in the quantization of the electromagnetic field in the next chapter. [A. both primary and secondary. Bec<luse we are including secondary constraints along with the primary constraints. under which an arbitrary quantity A undergoes the infinitesimal transformation bNA . [\fI a. ~ including the Jacobi identity [A. [C.A]p]p + [C. Ilh] = ibg. B] r = DA aB _ oB oA D\fIa ana o\fl<J alla 67) (7.B] = i[A. \fIh] = [Il a .B]p [A. where the first class constraint arises from a symmetry of the action. [B.8) (7. then the commutator of any two functions of the 't's and Ils would be just [A.10) If we could adopt the usual commutation relations [\fI a .Ilb]p = bg. B]pC + B[A. In fact.6. N (7.6.6 Constraints and Dirac Brackets 327 The Ils and \fIs will in general not be independent variables.6. where the XN are a set of functions of the 't's and fls.2:f'N [XN.11) when the constraint equations X"l = 0 are satisfied. electromagnetic gauge invariance..
6. Note that there must always be an even number of second class constraints. As we have seen.6.E]o is a generalization of the Poisson bracket known as the Dirac bracket: (7.15) where XIx = Ilo(x).Blo.15) are second class. Such first class constraints can be eliminated by a choice of gauge.lx = [XIx. (7. (7. in the case of the massive real vector field the constraints are Xix = X2x = 0. because an antisymmetric matrix of odd dimensionality necessarily has vanishing determinant.6. It follows that the matrix of the Poisson brackets of the remaining constraints is nonsingular: DetCfO.328 7 The Canonical Formalism tains a spacetime coordinate.XM]P of the Poisson brackets of these constraints with each other vanishes.19) where {A.11) shows that this transfonnation leaves the Hamiltonian invariant.20) (Here Nand M are compound indices including the position in space. After all of the first class constraints have been eliminated. so the constraints (7.) Eq. (7.B] ~ i[A.6.6.6. or treated by gaugeinvariant methods described in Volume II.ly = C2x.14) Constraints of this sort are called second class.6.6.17) and. the remaining constraint equations XN = 0 are such that no linear combination LNUNU:N.6. Cl x. (7.X2y]P = m26 3(x ~ y) (7. . Dirac suggested that when all constraints are second class.16) The Poisson bracket of these constraints is Cl x.13) where (7.2y = C2y. and for first class constraints it also respects all other constraints. of course. the commutation relations will be given by [A.18) This 'matrix' is obviously nonsingular.6. (7.2y = o. (7.
6.!!2r and their respective conjugates Pn.B]o]D 0.lnical transformation.i'J bll' = i>. This Cmatrix has inverse C.B]DC [A. but only by a timederivative.B]o ~ [B. It follows that the Poisson brackets for any functions A. But all these agreeable prop~ erties do not prove that the commutators are actually given by Eq. the Dirac brackets are unchanged if we replace the XN with any functions iN for which the equations X~ = 0 and XN = 0 define the same submanifold of phase space. [A.6.6.2s = [. [A. .6.C]D]D (7.BClD ~ [A.19) in terms of the Dirac brackets. we mean a transformation from a set of phase space coordinates '1'''.X2r]P vA = ::lJI ' 02. . B [A. .6. Also.23) and also the relations (7.22) + BIA.qpr ' = DA [A.x and 2. such that the constraints read !!2r = rlPr = O. n". the Poisson brackets hcing calculated in terms of the '1'" and I1". and for any functions A. rr".: and ['fa. = [!!2r.b 0. and redefining the constraint functions as Xlr = Jr.B are the same whether calculated in terms of 'fa and n" or in terms of of" and It also follows that if '1''' and n a satisfy the Hamiltonian equations of motion.21) (7. such that [>¥". X2r = rlPr . II a governed by second class constraints.20) are • Recall that by a can'. nr]P =  D.rIP~]p = Clr. it is always possible by a canonical transformation* to construct two sets of variables Qn. we have Cl r . The Lagrangian is changed by a canonical transformation.) He noted that the Dirac bracket like the Poisson bracket satisfies the same algebraic relations as the commutators [A. so the Dirac brackets (7. ~ + [B.i'Jb]p = 0. [B. 'fb]p = [i'J".!!2r. Using these coordinates to calculate Poisson brackets.19) consistent with the constraints XN = O.C]o.A]D.24) which make the commutation relations (7.J}Jr.6 Constraints and Dirac Brackets 329 taking values like 1. which does not affect the action.6.A]D]D + [C.~s]p = 0. then so do 'fa and Ii". This issue is illuminated if not settled by a powerful theorem proved by Maskawa and Nakajima. . 6 They showed that for any set of canonical variables \fIa. with the same Hamiltonian. Il a to some other phase space coordinates 'f".6.7. (7. [C. (7.x in the vector field example.I C.
]P(n"Bjp O.(y)]p ~ J3(x . IT.'. (7. ~ (C l )'. canonical commutation relations of the Q. P" \:onstructod by the MaskawaNakajima eanotllcal tran.17) and (7.26) (7.'P .6.27) Therefore the Dirac prescription (7.19) in terms of the Dirac brackets:· We now return to the massive vector field. we have simply ITo = 0. B are given by Eq. In the AppenJix to this chapter we di. B]p .6.'.25) aB DA .. PI]If we assume that these unconstrained variables satisfy the canonical commutation relations. (7.B]D ~ [A. B]p + im2 J d3z ([A. This is a case where it is easy to express the constrained variables VO and no in tefms of the unconstrained ones t Vi and IIi.6. (7. and Ps. the Hamiltonian defined in terms of the unconstrained Ifs and fls may be writlen just as well in tenns of the constrallled variables.6.18).y)J. discussed In Part A of the Appendix.]dXi"B]p = [A. uQn (I J] uQn OPf! 8A aB aB 8A In other words. (7.'lll + v.xl. the test of such canonical commutation relation. [V"(X).. V'(y)]p ~ [IT.z eT r [A.B]p  ~. and VO is given by Eq.z L:r r ~)Ir :')0» (7.tion whether we should adopt canonical commutation relations for the unconstrained variables Q".. VJ(y)] ~ [V°(x).play two largo c1as:.X2.tielJ commutation relations derived in Chapter 5. to see how it can be quantized using Dirac brackets. then the commutators of general operators A. VO(y)] ~ 0.. (7.(z) .jlr 8A aB :j.m' VO(z) .!' ~ m'J 3(x .. is their consistency with the frce. such that the Dirac commutation relation:..19) follow from the ordinar.29) Hence [V'(x).JO(z). IT.y).6.6. but to apply this test we need to know what the fZ' and P" are. .6. t ThIS is a spocial\:ase of the theorie:.IT. It is still an open que:.28) By definition.(x).A ~ B) ..20) yields the equaltime commutators [A. (7.6. From Eqs. We shall also show that in these cases. the Dirac hracket is equal to the Poisson bracket calculated in terms of the reduced set of unconstrained canonical variahles QJl.es of theories in which we can identify a set of unconstrained Qs and Ps. Ultimately. we see that CNM here has the inverse (C l )!>. we have [V'(x).6.B]p + [A.formation. ITo(z]p [a... (7.6.(y)]p ~ O. .6..6.330 7 The Canonical Formalism here [A.5). (7. B] ~ irA.19).
nO(Y)] ~ 0.'I'(x). • This .·· though.7.2) may be compensated by a field redefinition 'l'i(x) ~ 'l'i(x) + EPI('I'(X). that have nonvanishing matrix elements between the va<:uum amI the oneparticle states of the particles participating in the rea<:tion. it would change the values of matrix elements of the fields themselves.e. ' ax b ( '¥ (x) (7. y). [n'(x). for which the change in the action is of the form " " Jg. [V'(X). '). 8'1'(x).ection hes .30) [V°(X). . ITj(y)] = 0. [Vi (x). ITj(y)] = ioj o3(x ..Oj(Y)] ~ [V'(X). Smatrix elements can be obtained from the vacuum expectation value of a timeordered product of 'my operator.". of course. How can we tell whether some variation in the parameters of a the· ory can be cancelled by a field redefinition? A continuous local field redefinition will produce a change in the action of the form OJ ['1'] ~ E~ J f d'x J'I'i(x) pi('I'(X). O. .7 Field Redefinitions and Redundant Couplings· Observables like masses and Smatrix elements are independent of some of the coupling parameters in any action.7 Field Redefinitions and Redundant Couplings 331 [Vi(X). the theorem of Section 10.3(. and may be omitted in a first reading. This is because changes in these parameters can be undone by simply redefining the field variables.2i\J 3(x . and using the constraints to evaluate the commutators involving ITo and Va. VO(y)J = _iln. (7.y).71 ) So <lny change ogi in the coupling parameters gi.."'). that as long as we multiply by the correct field renormalization constants.tam. 7. V j(y)] = [ITi(x).6.2 show.o'l'(x)."').O'(y] ~ These are indeed just the commutation relations that we would find by assuming that the unconstrained variables satisfy the usual canonical commutation relations [Vi(x). clearly cannot affect any ohservahle of the lt theory. known as the redundant parameters. such as an infinitesimal local transformation \{II (x) + \{It (x) + E" pI(\{I( x). A continuous redefinition of the fields.omewhat Ollt of the book's main line of development.···) 61['1'] (7. OJ(Y)] ~ iOj. •• For in.7. L oJ i cgi ~ E L "J ~ 61['1'] P i ('I'(x). v \{I( x).y) .
ummed and integrated.u _ 24' ~gz2<D4 The constant Z is an redundant coupling.2. with the common lime argument dropped everywhere. In other words.!gZlD3 ) 2 6 ' and this vanishes when we use the field equation 0111 m2$ = igZlD3. . and an infinite subset of the parameters of the theory would be redundant. neither the bare mass m nor the coupling g are redundant. and r include a space coordinate x as well as discrete indices. The quantities !1' are the same as the C' introduced in Section 7. that the formula giving commutators as Dirac brackets times j follows [rom the usual canonical commutation relations for a reduced set of variables.. the field redefinition needed to compensate for a change of Z is a simple rescaling. (For this reason Z is called a field renormalization constant. and no function of m and g is redundant.) This is the most general field transformation that leaves the general fonn of this action invariant. in which labels like a. All quantum variables are uuderstood to be evaluated at the same time. 11. Repeated labels are . But for the more general actions considered in Sections 12. in which F is proportional to 111. Appendix Dirac Brackets from Canonical Commutators In this Appendix we shall show. suppose we write the Lagrangian density of a scalar field theory in the form Ie = . In this example. we would have to consider nonlinear as well as linear field redefinitions.332 7 The Canonical Formalism and therefore can have no effect on any observables.!z (aJilD a lD + m1$1) 2. A Suppose (as in the case of a massive vector field VIJ) that the quantum variables 'Fa and Il'l appearing in the Lagrangian L may be divided into two classes:· one set Qn of independent canonical variables (like Vi(x)) with independent canonical conjugates Pn = iJL/8QI1. For example.3 and 12. because 81 az = Ijd"X<1>(DlDm2<D. in theories of two types. with arbitrary numbers of fields and derivatives. a coupling parameter is redundant if the change in the action when we vary this parameter vanishes when we use the field equations JI I (5lf't = O. and another set • We are again using a compact notation.4. On the other hand.
I = [xlr.7) Now let us compare these commutators with the Dirac brackets. all commutators involving vanish: (7.ll) . The Poisson brackets of the constraint functions are Cl r .6) and.) We assume that the independent Qs and Ps satisfy the usual canonical commutation rules: (7.A.A.) It is easy to see that the Cmatrix has the inverse (c.A.P nJ =laQn' . the nil and Qs. fY'r (7.5) where ps is the Poisson bracket rrs  [!'.~nx) (like VOl whose timederivatives do not appear in the Lagrangian.Q'J ~ _jot'n ' ap " [fl.4) (7. of course.i.8) (7. but the discussion here will apply also for nonvanishing rrs.l) are the variables conjugate to the flr.Xls]P = 0. ~!l' .A.A. (7. where (7. The secondary constraints arise from the equations of motion 0 = aLjiJJJ/ for the fJ/.A.A.1)2r.1O) (In the example of the massive vector field ps vanishes. we suppose that these constraints can 'solved' .6. with X2r in the fonn Xl.A.f'(Q. The primary constraints are the conditions Xlr = 0. (7.A. which gives VO in terms of the independent P s (here.3) The constraint X2r = 0 yields the commutators involving !!2: [!l' .2s = 0. P) .2) (An example is provided by Eq.1. (7.r]p. they may be written in the form X2r = 0.5).Appendix Dirac Brackets 333 .9) (7.ar (7.A.A. (7.e..
A. .]p ~ . In particular [2'. + v. which can be solved by expressing the n a in terms of a smaller set of unconstrained PII' (We will see an example in the next chapter.7) shows that in all cases.16) with Eqs.ern oB [.12) th<:lt contributes. and an equal number of separ<:lte conditions X2r(IT) = 0 on the ITa.A.A.Qm]D ~ [P".z C. where A is ffJ r and B is anything. .6 all Dirac brackets involving the constraint functions vanish. (7. we have only the first and third terms. {i of' [2'.A. so the Dirac brdcket is equal to the Poisson bracket.334 7 The Canonical f"'ormalism Also. we have only the fourth term (7. (7. B] P _ j. B Next consider the case where the constraints take the form of conditions XlrC¥) = 0 on the 'P a.P.A.f Jp Now.14) Where both A and Bare i?s. aiP .z .7r. cfL (7.f'(Q.f p aiLt' (7. (7. (7. In particular. B] 0 .0 .]D ~ Il + 0af' 'Q" . because as remarked in Section 7.Q']o ~ .. the commutators are equal to the Dirac brackets times i.[0' BPl + .P)]p.16) Comparison of Eqs.ilfa0.12) oR DA + [)jJr r 8A rs oB 8A [ .'P .3)~(7.A. [Q'. it is only the fifth term on the righthand side of Eq.~" + [ '] DB A.A.13) Where A is !:i' and B is a function of Qs and P 5. so the Dirac brackets involving it and/or {jIJs are given by using the constraint equations to express fLr and/or!!J's in tenns of the independent Qs and Ps. oA Hence the Dirac bracket is oA oR [A.r .82/ B. then cAI8fL' = oB/ ail' = 0. Xlr]P= ad" [A".13)(7.7r :j"l. (7. the Poisson brackets of any function A with the constraint functions are iJA [A.PmI D ~ O.0&.A. B] D _ [A .[A. which can be solved by expressing them in terms of a smaller set of unconstrained variables QII.PmI D ~ 6.7. where the constraints on the 'Fa are .il (.A. if A <:Iod B are both functions only of the independent canonical variables Qn and Pn." [Q'.A. which cancel: a" .15) Finally. This is only to be expected.
Now. since Xis depends only on the qJ and X2r depends only on the these are the only nonvanishing Poisson brackets of constraints.O. It follows that 0= [qJa. . To determine these coefficients.r /j c~ as the Poisson bracket :olD h = [Xb.18) implies that U.2r. n.21) We recognize the factor multiplying DXlAqJ) OX2r(n) .) We assume that the unconstrained variables satisfy the usual canonical commutation relations [Qn.lulb (. and the constraints on the n" are secondary constraints arising from the consistency of the first class constraints with the field equations. Hence Eg.A. we have then . the constraint XlrC¥) = 0 is satisfied for qJa qJ<l(Q) for all Q. [Qn.17) or. that X2r(n) = O.A.('!') oX2. c.r i7L DL DqJh = ()\jJh DQn = nh oQn (lqJh . D] = IUb '~a b + le a (lqJb . in other words.A. (7. Also.19) 0 .i6') o'!" ~ O.) = i[qJ'\nb]()~~Dr(n) (/ b .Qm] = {Pn.(D) .20) with some unknown coefficients c~. we make use of the other constraint.A. .X2' . 'oQ" = (7 A 18) .2. (['I'" D " 1. so (7. .A.Pml = O.20).Appendix Dirac Brackets JJ5 gauge fixing conditions that are used to eliminate first class constraints. then there would be additional constraints on the qJa.o".1 = O. ana ~ . so Clr. "XL.Pml = ib~.X2r(n)] Using Eq.X2r]P ~ Cl s. (7.ls = C2r. r eXlr (7. cqJb oQn  eXlr DqJb Furthermore. DqJ" c nb (7." [ T. the vectors (Vr )/1 0Xlr /oqJh form a complete set perpenc'l'h/cQ".(".D". because if there were some dicular to all the vectors (Un)h other vector Vb with Vb(Un)b = 0 for all n. (7. The constrained and unconstrained momenta are related by Pn = It follows that ai.A.
A. the other Dirac brackets are ['1". (7.] ~ . all ' a OXlr [Xlr. In theories of both types considered in this Appendix.Or.A.jJaL. for which "1m = Q" and II" = P" are tbe independent canonical variables. we also need an explicit formula for the Hamiltonian. For second class constraints.N (7.]o (7. II.23) II a ['I" .27) ['P a. . IIh]p = 0 . (7.b II ] ~ I.1 has no 11 or 22 components.] h Ih IU" ] L~ . this is trivial. . so trivially (7.30) For theories of type A.20) shows that = eXlr _a (c. II. o\f'h aX2' ax. 'Ph] = i['¥". with the constraint functions are an.24) is the Dirac bracket ['1".[6' _ ana (C' )" . we can easily see that because C. 'Ph] 0 .Xlr] P =0. (7. Also.25) [.A.26) as was to be shown. a = c s Cls.21) may be written eXN 'II v .A. (7.A. the sum over a runs over values n.22) with N running over all constraint functions. (7.29) the sum running over independent canonical variables..A. p = (7..336 7 The Canonical Formalism Thus Eq. nb1p = 8.A.A.['1". this Hamiltonian may be written in terms of the constrained variables as H=nal.A.ls on" Using this in Eq..28) *** In addition to commutation rules. (7.ls.A.¥h ' (7. so the quantity in brackets on the righthand side of Eq. The usual canonical formalism tells us to take (7.A. this has the unique solution c~ = ~ ~XN (C1)N.24) The Poisson brackets of qw and [\f' .A.1)2r.0.. 'l'bl o ~ [II" IIb]o ~ 0.
Derive an explicit expression for the symmetric energymomentum tensor el'l".ig Vp and Fl'v 01' VI' .Problems 337 together with values r. with Lagrangian density:£ = L::mnoplDnoPl1lm!nm(I1I). YI 'mb '/y which again yields Eg.op . Fl'v 41"· where D p.6. VI" and .A. 5. where fYlm(lD) is an arbitrary real matrix function of the field. we note that Eg. Problems I.J¥' is an arbitrary function. (This is called the nonlinear (1model.¥J. 6. = iEL::jr'j. OV!' = O. 3. Derive an explicit expression for the conserved current associated with this symmetry. 7. Consider a theory of real scalar fields I1IYI and Dirac fields 'Pi. In the theory of Problem 4. = !m 2 VI' VII . o'¥. Prove that the Dirac bracket satisfies the Jacobi identity (7. suppose that the Lagrangian density is invariant under a global infinitesimal symmetry ol1l Yl = iEL::mtllm<D"'.17) gives P. Consider the theory of a complex scalar field 111 and a real vector field VI'. Consider the theory of a set of real scalar fields I1IYI .Q ~ lib 'Q Q ~ lib 'I' o " 'n O r ..) Carry out the canonical quantization of this theory. Derive the interaction V[¢(t). with Lagrangian density :£ =:£0 + :£1.A. Carry out the canonical quantization of this theory. 4.#(oI>tol» 2 ' . Derive the interaction in the interaction picture.30).23). where :£0 is the usual freefield Lagrangian density. For theories of type B.a. but not their derivatives.¢(t)] in the interaction picture. with Lagrangian density !l' = (D I' oI»t DI'I1I. derive expressions for the symmetric energymomentum tensor Sill" and for the conserved current associated with the symmetry under 0111 = iEI1I.!F . Prove that the Dirac bracket is independent of the choice of constraint functions XN used to describe a given submanifold of pbase space. ! 2. and Sf1 is an interaction term involving I1IYI and 'Pi. (7. for which fly = 9 y = O. (7. In the theory described in Problem 2.
Nature 180. 1983). New York.. G. Math. Supersymmetry and Supergravity (Princeton University Press. Akad.g. Maskawa and H. I am grateful to J. Rosenfeld. and original references quoted therein. M. G. Proc. 2. Roy. Lectures on Quantum Mechanics (Yeshiva University.338 7 The Canonical Formalism References 1. Soc. Dirac. Bagger. J. S. 3. Ned. A. Wet. 1295 (1976). e. H. e. Phys. ser. 326 (1958). Helv. Wess and J. Prog.g. 246. 4. A. Can. 51. Phys. London. F. 56. 887 (1939). Gravitation and Cosmology (Wiley. P. Theor. Belinfante. K. Proc. 79 (1956). P. J. M. M. T. 6. See. Memoires de l'Academie Roy. Dirac. 2. Spaarnay. IV. Also see P. Weinberg. . 1972): Chapter 12. Princeton. 5. A.. Nakajima. Bergmann. See. 334 (1957). B. 30 (1930). 1964). 635 (1948). J. also see L. Physica 6. 129 (1950). New York. Belgique 6. Casimir. Feinberg for bringing this reference to my attention. Acta Suppl.
and then deduce the main features of electrodynamics from the gauge invariance principle. 8. and then from that deduces the gauge invariance of the effective field theory that describes such particles.9. After that we shall follow a more conventional modern approach. It will probably not surprise the reader that this book will follow a different path.9 that there is no difficulty in constructing an antisymmetric tensor free field f/1V(X) for such particles. We saw in Section 5. given by Eq. Eg. This field can be expressed in terms of the fourpotential al'(x).1 Gauge Invariance Let's start by recalling the problems encountered in constructing covariant free fields for a massless particle of helicity + 1. through the familiar relation (8. Most theorists have tended to take gauge invariance as a starting point. one first notices a state of mass zero and unit spin among the normal modes of a string.1) However. as we shall see.23).1. We shall first infer the need for a principle of gauge invariance from the peculiar difficulties that arise in formulating a quantum theory of massless particles with spin. in which one takes gauge in variance as the starting point and uses it to deduce the existence of a vector potential describing massless particles of unit spin.9. still the paradigmatic example of a successful quantum field theory. (5. but in modern string theories) the argument runs the other way.8 Electrodynamics The original approach to quantum electrodynamics was to take for granted Maxwell's classical theory of electromagnetism. using either approach one is led to the quantized version of Maxwell's theory. (5. and quantize it. It is too soon to tell which of these two alternatives corresponds to the logical order of nature. At any rate.23) shows that the al'(x) transforms as a fourvector 339 .
not AIl(x).(Ax)+ 8. in fact.A).5) This is trivially true if 1M involves only Fllv(x) and its derivatives.4) Hence the Lorentz invariance of 1M requires that bl M a/I JAIl(x) = O.(x) + ". m2 1'11'2 + QIIQv/ . " 2 q +m Ie which prevent us from dealing with massless particles of helicity +1 by simply passing to the limit m _ 0 of the theory of a massive particle of spm one..340 8 Electrodynamics only up to a gauge transfonnation Uo(A)a.1.y ) ( ~ (21t )'!d4 xeiq·(xy) .dx).avAJl{x) and its derivatives. no way to construct a true fourvector as a linear combination of the creation and annihilation operators for helicity + 1. (8.2) There is.(x) Uol(A) ~ A/ a.1. bAIl(x) bF11V(x) But if 1M involves AIl(x) itself then Eq.«x) (8. The change in the matter action under the transformations (8.3) (at least when the matter fields satisfy the field equations) so that the extra term in Eq. This is one way of understanding the presence of singularities at m = 0 in the propagator of a massive vector field '"!.1. .5) is a nontrivial constraint on the theory. Instead of banishing AIl(x) from the action.1. but this is not the most general possibility.1.1. and not the one realized in nature.vx.!l(x. (8. (8. we shall require instead that the part of the action 1M for matter and its interaction with radiation be invariant under the general gauge transformation A. (8.2) should have no effect. We could avoid these problems by demanding that all interactions involve only· FIl"(x) _ 0IlAI'(x) .3 that infinitesimal internal • We now use A p and rj" for the electromagnetic potential vector and the Jield strength tensor because these are interoiCting fields. Now. along with matter fields.(x) ~ A. (8. what sort of theory will provide conserved currents to which we can couple the field AIl(X)? We saw in Section 7. x) blM (8.3) may be written bl M ~ " J ax M. In this case OIM = 20 1' OI M .1.
11) The conservation of electric charge only allows us to fix the values of all charges in terms of the value of anyone of them. tHere '¥t is understood to run over all independent fields other than AJ'.use the summation convention for sums over ficld indices. so we may simply set these quantities equal: M.. ~ 8(8 '1") q.J (x .8.3 that if 1M is the integral of a function !f'M of 'lit and a.1. then for general infinitesimal functions f"(x) the change in the matter action must take the form OIM ~  J d4 x J'(x)O'_(x).9) Q= J i'xJo . and hence o/JI' = 0. " 82M .1 Gauge Invariance 341 symmetries of the action imply the existence of conserved currents. where Q is the timeindependent charge operator (8. '1" (xl. (8. so there is no sum over {in Eq.7) When the matter fields satisfy their field equations. " and this generates the transformations (8.1.6).(x) is taken to be proportional to J!'(x). . ) (8.1. if the transformation" leaves the matter action invariant for a constant f". We use a capital psi to indicate that these are Heisenbergpicture fields. we saw in Section 7.6) in the sense that [Q.'¥t. the matter action is stationary with respect to any variation of the \}Jr.1. 'l'i (x)1 ~ q. (8. In particular. in the sense that OIM/bA/. Any constant of proportionality may be absorbed into the definition of the overall scale of the charges qt. so in this case (8.1.1. whose timedependence includes thc effects of interactions_ Of course.10) We can therefore construct a Lorentzinvariant theory by coupling the vector field A" to the conserved current JI'.(x) . I . then the conserved current is given by t J ~ . . this '1'( is not to be confused with a statevector or wave function.1. 'I' .1. (8. OIM _ .7) must vanish. (8.8) In particular. conventionally taken to •• I:lecause thc field transformation matrix is taken now to be diagonal it is not convenient here to .
4. but with m = 0: (8.1. (x) iE(X)q. without higher derivatives. A symmetry of this type with an arbitrary function E(X) is called a local 8ymmetry. denoted e.11) that gives a definite meaning+ to the value of e.13). Maxwell equations (8.(!vA/.14) This is the same as the action used in classical electrodynamics.> In the above discussion. we can take this to be the same as for massive vector fields.15) We recognize these as the usual inhomogeneous Maxwell equations. L11) fixes the defimtion of e only after we have dellned how we are normalizing Ajl(x). or a gauge invariance of the second kind.1.o/r(x).1. the derivation runs in the opposite direction.12).12) (8.1.14). Eq_ = (8. (8. (See Section 12. (8. It is Eq. we have started with the existence of massless spin one particles. (8.1.13) ~ J 0/. as we will see in the next section.1.11) and (8.1. Also. or a gauge invariance of the first kind.1. A symmetry under a transformation with f' constant is called a global symmetry. If there are any terms in the action of with higher derivatives and/or of higher order in FI'\' they can be lumped into what we have called the matter action. The requirement (8. it leads to a consistent quantum theory. homogeneous. Using Eqs. the field equation for electromagnetism now reads O~. Several exact local symmetries are now known.1. one starts t Of course. The question of electromagnetic field normalization is taken up in Section 10.11) may be restated as a principle ofinvariance: 1a the matter action is invariant under the joint transformations J A/l(x) = C1jE(X) .5.b [Ir+IM]~apF'r+J'. oA v (8.) We have not yet said anything about the action for photons themselves. with current J". There are also other. As a guess. but the only purely global symmetries appear to be accidents enforced by other principles.16) which follow directly from the definition PI'V DI'A\' .1.342 8 Electrodynamics be the electron charge.1. but its real justification is that it is (up to a constant) the unique gaugeinvariant functional that is quadratic in PIIV . . (8. and have been led to infer the invariance of the matter action under a local gauge transformation (8. That is. As usually presented.
19) In order to cancel the second tenn here.21) 6 D.D\{I) that is formed only out of \{If and DIl\{lI will be invariant under the transformations (8.". (8. f 0IJ.Dp'l"(x). But all realistic Lagrangians do involve field derivatives..L. (We could also include F llv and its derivatives in :EM') From this point of view..AIl in the Lagrangian density would violate gauge invariance.1. we 'invent' a vector field AII(x) with transformation rule .1.1. DIl 'l' t = 011 '1" .'I"(x).8. (i Il q.'I"(x) ~ ie(x)q.11)..(x)q..18) If the Lagrange density !£ depended only on fields \{II (x) and not on their derivatives then it would make no difference whether E is constant or not.22) A matter Lagr<lllgian density "~)MCP.1.1..1.0' 'I'.)Ap(x) ~ upe(x) OIJ\{If (8.17) and asks what must be done to promote this to a local symmetry 3'1"(x) ~ i. OffM  'L. (8.1D'I" I ( 8ff M . we have MM _ ' " v 'A II . and here we have the problem that derivatives of fields tr<lllsform differently from fields themselves: (j 0/1\{It (x) = i E(X )qtc/pf (x) + i q/pl (x)OJtf'{x) .q.'I' . (8. im2 8. f which is the same as Eq..2 Constraints and Gauge Condition8 with a global internal symmetry (PP{(x) = iEl'P{(x) 343 (8.2 Constraints and Gauge Conditions There are aspects of electrodynamics that stand in the way of quantizing the theory as we did for various theories of massive particles in the . the masslessness of the particles described by All is a consequence of gauge invariance rather than an assumption: a term AIJ. . if it is invariant with E a constant. (8. invariance with E constant would imply invariance with E a function of spacetime position..18).1.20).1. (8. which transforms just like \{If and All only in (8.1. with E(x) an arbitrary function. With the Lagrangian of this form..20) and require that the Lagrangian density depend on the combination .lq{ A II '1" . \{It) __ .
and I'. (8.1. we have e.3.I)] ~ io. while the canonical electromagnctic fields and canonical conjugates are A. = O. For matter Lagrangians!l'M that mvolve only >pf and D~>pt.".4) Hence Eq. Even though the matter Lagrangian may generally depend on AD. t).2) This is called a primary constraint. i. more directly.n'(y.2. and therefore will vanish for I'.. The symbols QR and p" are now reserved for the canonical matter fields and their canonical conjugates. (8. ' (8. As usual.(x. 'll ~ o. (8.03(x . and therefore nO(x)~o.2.j. the prescription (8. There is also a secondary constraint here. which vanishes for 11 = 0 because FW is antisymmetric. t Due to exhaustion of alphabetic resources.= . which follows from the field equation for the quantity fixed by the primary • constramt: •• i 8!£ D!:e 0 D. •• As usual. t).A~) will be again antisymmetric in I'."\/iJ{iJQA~} = _Fa~. But this is not possible here. . and fl.~ } .. Even if the mattcr Lagrangian depends also on F~" D!I' .n ~ 8. t).y) and [Q'(x. ~ PnqnQ .2. W(y. n'(y.. run over the values 1. We encountered a similar problem in the theory of the massive vector field. (8../4. W(y. .".y). etc. The first constraint arises [rom the fact that the Lagrangian density is independentO of the timederivative of Ao.3) is a functional relation among canonical variables.2. (8.(x.2. .. t). because All and II" are subject to several constraints. 8!!' L. because it follows directly from the structure of the Lagrangian.344 8 Electrodynamics previous chapter. havc had to adopt a notation here that is different from that of the previous chapter.. ~ .2.3) are inconsistent with the usual as· sumptions that [A. Both Eq. In that case we found two equivalent ways of dealing with it: either by the method of Dirac brackets or. the charge density depends only on the canonical rna tter fields t Qn and their canonical conjugates Pn : J o = I .3) 8FiQ (JA o the timederivative term dropping out because Foo = O.1) Quantization by the usual rules would give [A.(x.2) and Eq.2.O'(x . respectively.21) tells us that Y'M does not depend on any derivatives of any A'. we may define the canonical conjugates to the electromagnetic vector potential by n' _ 8!!' 8(8oA.) (8.dV(il.2. by treating only • For :1\ _ FpvF~'. 1)1 ~ [P.r o(oo'P/ )qt'P ( = I .t)] ~ io.
.JI{(x) +..t) at later times. if Eq. Eq. to A). There are various gauges that have been found useful in various applications::!: t Here oIl is any complex scalar field with q 1= 0. because this theory has a local gauge symmetry that makes it. aiDO . Given any solution AJl(x.3) are first class.t).8. It is clear that here we cannot use Dirac brackets. Because of this partial arbitrariness of A11(x. the constraint functions X here are nO and ain i ..5) oFo j (J + OiOj of!' . is to exploit the gauge invariance of the theory.(x))l.exp (iqri. t) with the same value and timederivative at t = 0 (by choosing ~ so that its first and second derivatives vanish there) but which differs from A/j(x.A1tCx) + u/l}. ::.2 Constraints and Gauge Conditions Ai and 345 as canonical variables. to be discussed in Volume II. we have of!' 00 [Oi . t) of the field equations.JO (as compared with Oirri . in principle. (8. .2. two are particularly useful. as for finite mass. t) + DJlt3{x. (8.2.JI{(x) to impose a condition on AJl(x) that will allow us to apply the methods of canonical quantization. to 'choose a gauge'.3) is satisfied at one time.JO) and these obviously have vanishing Poisson brackets. then it is satisfied for all times.ooJ ° ° (82.. we make a finite gauge transformation AI.(x) +. (8. Of the various approaches to this difficulty.(x.2) and (8. it is not possible to apply the canonical quantization procedure directly to All (or. the constraints (8.2.ooJ ~F" l' and the current conservation condition then gives 00 la. which will be roHowed here.(x). That is.2.3) to calculate AO in tenns of these variables.2.3) is a mere initial condition. In Dirac's terminology. The other. impossible to infer the values of the fields at arbitrary times from their values and rates of change at anyone time.m2Ao . because (using the field equations for the other fields Ai). we can always find another solution Aj.aJi . solving the analog of Eq.J aFiQ ni 0] = = of!'. Instead of giving AO for all time. this gauge condition is used when the gauge symmetry is spontaneously broken by a nonvanishing vacuum expectation value of <1'.: _10] ~O It should not be surprising that we still have four components of AJl with only three field equations. l. One is the modem method of gaugeinvariant quantization. Nor can we eliminate AO as a dynamical variable by solving for it in terms of the other variables.
i = V. so we will adopt Coulomb gauge here. (8.3. then the gaugetransfonned field AP + all A. subject to the gauge condition V .t l =! d3y JO(y. but of course Coulomb gauge keeps manifest rotation invariance in a way that axial gauge does not. As mentioned earlier. so Eg. the charge density depends only on the canonical matter fields Qn and their canonical conjugates Pn.8) which can be solved to give A()(X.2. so that (8.9) to eliminate AO (and ITo) from the list of canonical variables. A.6) It will be convenient henceforth to limit ourselves to theories in which the matter Lagrangian 2M may depend on matter fields and their timederivatives and also on AJi but not on derivatives of Ai'.2. Even after we use Eq.2.6).2. 2 To check that this is possible. this yields _ V2 AO = J O .2.2. with i = 1. will.FI'".9) represents an explicit solution for the auxiliary field AO. we cannot apply the usual .2.346 8 Electrodynamics Lorentz (or Landau) gauge: Coulomb gauge: Temporal gauge: Axial gauge: Unitarity gauge: GIJAP = 0 V·A~O AO =0 A3 = 0 <D real The canonical quantization procedure works most easily in the axial or Coulomb gauge.2. From now on. we assume that this transformation has been made.t) 4nlx YI (8. (8. (8.2.9) The remaining degrees of freedom are Ai.7) Together with the Coulomb gauge condition (8. provided we choose A so that V2 . (The standard theories of the electrodynamics of scalar and Dirac fields have Lagrangians of this type.3 Quantization in Coulomb Gauge There is still an impediment to the canonical quantization of electrodynamics in the Coulomb gauge. note that if All does not satisfy the Coulomb gauge condition.3) reads _ OjpjO = JO . A = O. (8. 8.) Then the only term in the Lagrangian that depends on Fw is the kinematic term ~Fi'\. and the constraint equation (8.
(x) C NM is nonsingular. then the commutators of the constrained variables and their c<:lnonical conjugates are given (aside from a factor i) by the corresponding Dirac brackets (7.3) where here. Note that the constraint functions have the Poisson bmckets Clx.Xly]P = [X2x. x 2. the canonical conjugates IIi to Ai may likewise be expressed in terms of the canonical conjugates P xl and P2x to Qh and Q2x. for any functionals U and V. independent QIx = A1(x). the field vari<:lbles Ai may be expressed in terms of canonical vari<:lbles. In such C<:lses. which identifies these as second class Also. s) + a2A 2(x l .(x) _ bV bA'(x) JV JU ] bA'(x) bIT.1): A \x) =  IX) ds [a A (x j I I. . These constraints <:lre of a type known as 8econd class.3 Quantization in Coulomb Gauge 347 canonical commutation relations to Ai and Ih because there are two remaining constraints on these variables" One of them is the Coulomb gauge condition (8.3.3. X2y]P = _V (52(x .2y 2 [Xix. etc. .1) The other is the secondary constraint Eq. which may. (8. with A 3 given by the solution of Eq.IT'(x) + 1 0 (x) ~ O. Ph.G.3). 0.3.arne time. and omitting the time argl. X2y']P = 0. • In this section i.20).6. We continue the practice of taking all operators at the .3. V] p ~ The 'matrix' constraints.p:5 3 (x .1ment. discussed in Section 7.2). J d] x [b U bO.2. (8. for which there is a univers<:ll prescription for the commutation relations. j.2y = C2y. (8. This prescription has the great advantage that we do not have to do use explicit expressions for the dependent variables in terms of the independent ones.3. be taken as Q2x = A 2(x). 8)] .6. for instance. Part B of the Appendix to the previous chapter tells us that if the independent variables Qh. (8. (8. [u.y).3.y). ITj(y)] = iO. and P2x satisfy the usual canonical commutation relations. which requires that X2.lx Clx. Using Eq..ly C2x. Q2x. [Xlx.8.2) Neither constraint is consistent with the usual commutation relations [Adx). because operating on the righthand side with either Oji'Jx i or iJ/8yj does not give zero. run over the values 1.x 2.2.
F 1tI' in the Lagrangian depends on A. Thus all we can conclude from inspection of the Lagrangian is that n equals A(x)+VAo(x) plus the gradient of some scalar. we must face the complication that n does not commute with matter fields and their canonical conjugates. _. () Xl a 3 Hence according to Eqs. but its Dirac . A = 0. +.6) But with A constrained by the condition V . which requires that V· n = _Jo = V2 AO.Xi. ( 4nxyI)' <1: 11 [Ai(x) .A° (OXl (8.2Y = O.5) are reasonably simple. (8.0/'· JA. If F is any functional of these matter degrees of freedom. (8. then since V' <sA = 0."'03(x . IIi(y)] ~ O.6.y) ox' . the nonvanishing Poisson brackets of the Ai and constraint functions are n j with the [Ai(x).3. what is in the previous section where only the kinematic term ~~ I d3 xF/I\. we also have bL = f d3 x [&" + V. .348 8 Electrodynamics To calculate the Dirac brackets. then its Dirac bracket with A vanishes.]P ~ +OCO (x . varying the Lagrangian with respect to A without worrying about the constraint V . as is guaranteed by the general properties of the Dirac bracket. bA for any scalar function .3.3.]p ~ _'. and [II i (x). Because V· A = 0.'F(x). Although the commutation relations (8. the equaltime commutators are [Ai(x).2). AI(y)] ~ [IIi (x)._ OAl(x) n in electrodynamics? For the class of theories discussed JL '.'F] .3.19) and (7.3.X2. (7.6. .3.1) and (8. variational derivatives with respect to A are not really well defined. Also. we note that the matrix C has the lllverse (C1hl:.. we conclude that Eq.ly = (C.20). If the variation of L under a change JA in A is bL = f d 3 x.y).6) does indeed give the correct formula for IIi.03(xY)+iaxJoxl.1hx. Now. This ambiguity is removed by condition (8.2). A = 0 gives II.A j (x) ~ (x).35) Note that these are consistent with the Coulomb gauge conditions (8.IIh)] ~iO.
3.Jo(x)]p 4'lx _ YI VJ'(y z) 3 3 3 3 1 1 d3 y [F. .) Replacing A everywhere with "t.8) Now we need to construct a Hamiltonian.Ao(y)]pVo 3 (y .9) because V· A ~ 0. this gives (8.3. we shall write it in terms of A and llt.9) where.3.3.3.l(y) = VAo(y) and that ojAo(x) commutes with ojAo(y).3.10) where iJ1 is a current that does not involve All. A + iOAO] + HM. and also the simple constraint (8.TI(Z)]p d xd y [F.3 Quantization in Coulomb Gauge 349 bracket with II is [F. as mentioned earlier.8. instead of expressing the Hamiltonian in terms of A and ll. In electrodynamics. without first having explicitly to write the Hamiltonian in tenns of the unconstrained Qs and Ps.lx _ yl [xly.) To be specific.n. In order to facilitate the transition to the interaction picture. but the electrodynamics of spinless particles is more compli~ cated. VAo(z)]o. consider a theory with a Lagrangian density of the form (8. VAo(z)]p ~ [F. By using the facts that "t(x) commutes with ll(y) . According to the general results of the Appendix to Chapter 7.5) as n(x). (We can use llt in place of TI in Eq. where llt is the solenoidal part of ll: TI"_TIVA °~A. this gives a Hamiltonian (8. nt(z)] vanishes. it is easy to see that "t(x) satisfies the same commutation relations (8. (8.Xl..]p 4. (8. (The electrodynamics of spin! particles has a Lagrangian of this [onn.TI(z)]o ~  J ~ J ~ J d xd y [F.10).z) ~ [F. (8. .3. and 2"maHer is the Lagrangian for whatever other fields do appear in iI'.9) of the form H = J d3x [lli + HV x A)2  Hllt + VAof  J. Q" and PI! are to be understood as the matter canonical fields and their canonical conjugates. we can apply the usual relation between the Hamiltonian and Lagrangian using the constrained variables A and llt.3.7) for which [F. which are given explicitly by the term iJ1AI' in Eq. aside from their electromagnetic interactions.
(8. in particular at t = O.11) into a freeparticle term Ho and an interaction V H = Ho + V. this is H = J !J d 3x [ ~ni + HV x A)l . H].3.11) oAO may look peculiar. the transition to the interaction picture is made by applying the similarity transformation V(t) ~ ~ exp(iHot) V[A.O.J 1 3[12 dx 2"l+2(VxA) . P].2) and (8.4) where P here denotes the canonical conjugates to the matter fields Q.] 3 d xJA+Vcoul+ Vmatter.5).1 d3 x (P. 0) .~o exp(iHot) V [art). and any operator o(x. (8. n(t).3. t) in the interaction picture is related to its value O(X. Using the solution (8.4.Ix YI (8.12).3) can be evaluated at any time we like (as long as both are evaluated at the same time). 8.3.4.9) for AU.4.. q(t).{t .4. and II Caul is the Coulomb interaction (8.2. so Eqs. where Hmatter. p(t)] . The total Hamiltonian (8. n".4.1) (8.3) Ho = J v=.4 Electrodynamics in the Interaction Picture We now break up the Hamiltonian (8.3. (8. . As in Chapter 7.2) (8.4. 4. +Hmatler. that the rate of change of any operator function F of A and n is given by iF = [F.4.O and Vmailer are the freeparticle and interaction terms in Hmatter. but this is nothing but the familiar The term Coulomb energy Veou] = 1 J d 3x fl AO ~ j Jd 3xJd3/O(X)JO(YI . as it should be. using the commutation relations (8. A + ±f1AO] + HM .1) is timeindependent. Q. excluding their magnetic interactions HM =. (8.12) The reader can verify.3.350 8 Electrodynamics electro~ where H M is the Hamiltonian for matter fields.2"malter).J .
(8. O.4.8) still apply V' a V' n ~ ~ 0.t).9) [n'(x.t).Y)+. (8.4.n'(~I)]~i3"3IXY)+'a' 4nlx [ ox' xl [a'lx. (8. (8.Ho] .l) ~ = 351 0 by (8.6) to evaluate Ii: iilj(x.8) (8.4 Electrodynamics in the Interaction Picture in the Heisenberg picture at t o(x.4.4. (8.4.10) and (8. 1 (ix1{}Xl a 2 .6) and (8.4.O) exp(iHot).t)] . integrate by parts.1J) x (V x V x a(y.4.4.11).aJIy.5) exp(iHot) O(x.nJ(y.4. 4nlx _ YI 1 ] We can replace OJ ox J in the second term with tJ / ayJ.t) = [nj(x.4. Hol ~ iJd]Y [3.3](X. we must use Eq.6) (We are dropping the subscript ~ on 1t(x).2. (8. For the same reason.) Since Eq. t).3.t).4.5) is a similarity transformation.12)) just yields the usual wave equation o a = O.I))) .4. and use Eq. (8. 3 0 '2 I] yl .8.I).I) ~ [a(x. and likewise for the matter fields and their conjugates. yielding a=1t (8.7) ~ 0. which (using Eqs.10) (8. .12) just as in the Heisenberg picture. t) = [aj(x.Hol = I 'Jd' Y [OW)'I 2 X  Y) + l. the equaltime commutation relations are the same as is the Heisenberg picture: [ a'(~t).4. the constraints (8.4.11) To establish the relation between 1t and Ii. (8. The field equation is likewise determined by Urj(x.t)] ~ 0. so that i O(X.'x1uXl ::l a '::l  4n Ix  I] y! (8.
4.P ') 0uu'. and a(p.4. (8.15) in Eq. a) are any two independent 'polarization vectors' satisfying p' .2) gives the correct Hamiltonian for free massless particles of helicity + 1.(p.14) The most general real solution of Eqs. 0") . but rather a functional (8. 0") ] . (8. we can easily see that the commutation relations (8. (8. (8. (8. ~ O.7)(8.4.352 8 Electrodynamics Since AD is not an independent Heisenbergpicture field variable.O") a(p.4.P(p.a')] (8.o(p. we can normalize the ell(p.16) (8.21) As remarked before for massive particles.15) where pO .2) to calculate the freephoton .4. (8.4. (8. this result should be regarded not so much as an alternative derivation of Eqs.15) satisfy [a(p.20) [alp.4.12) and (8.4.a).2.9: ~ .4.19) where R(p) is a standard rotation that carries the threeaxis into the direction of p. (8.4.+lt R(p) 1/y2 +i/y2 o o .a) so that the completeness relation reads (8.17) . (8. eJl(p.4.21). In the same spirit one can also use Eqs. (8.12).4.4.9) are satisfied if (and in fact only if) the operator coefficients in Eq. 0") at(p. at.14) may be written ai'(x) = (2n)3/2 J dJp. J2i!i ~ [eIP'Xe!"(p.4. we do not introduce any corresponding operator aD in the interaction picture. hut rather take aO ~ O. 0") to be the same polarization vectors that we encountered in Section 5.4. By adjusting the normalization of a(p.4.4.10).4. Using Eqs.13).4.4. and (8. (8. (8.20) and (8. (8. + elPxe!".(p.4.18) For instance. we could take the e(p.u).9) of the matter fields and their canonical conjugates that vanishes in the limit of zero charges.a')] ~ 03(p . a'(p '. hut rather as a verification that Eq.4. with cr a twovalued index.a)~O.a) ~ 0. alp '.O") are a pair of operator coefficients.18) and (8.Ipl.
4. a in Eq.i".at d3 (at(p.t) 4nlx YI (8. (8. Inserting our formula (8.4. (851) where T as usual denotes a timeordered product.(<I>YAC.. Finally.26) We have written jpall instead of j .15) for the electromagnetic potential then yields . T {a..a)a(p.(y)} <l>vAC) ..t) .expUHot) J. a).t)l(y.22) c which (aside from an inconsequential infinite cnumber term) is just what we should expect.O)exp(iHot). given by the propagator: .4) in the interaction picture IS V(t) =  J d3 x jp(x.(x.y) + e.5 The Photon Propagator 353 Hamiltonian Ho J Ip ~ J p Lpo ~ d]PL c Q [a(p.4. t) aP(x.p(y'JB(y  x)] .4.23).(x .i".4..8.24) while Vcou](t) is the Coulomb term VCon](t) = exp(iHot) Veonl exp( iHot) ~ Jd3x d]y l(x.y) ~ J(2~~IPI P. we record that the interaction (8.4. t) + VCoul(t) + Vrnatter(t).y) . 8.4. (8.25) and Vrnatter(t) is the non·electromagnetic part of the matter field interaction in the interaction picture: Vmatter(t) = exp(iHot) Vmatterexp(iHot).a) + lb](pp») (8.(x.5 The Photon Propagator The general Feynman rules described in Chapter 6 dictate that an internal photon line in a Feynman diagram contributes a factor to the corresponding term in the Smatrix.(x .(p) [e'P'l>YJB(x .2) . at(p.23) where in terms of the current J in the Heisenberg picture J.5. (8. but these are equal because aP has been defined to have aO = O. (8. (8. a.(x) .4.
5. We recall from pipl ~ Ipl' ' Poo(p) ~ P.5. The result has been justified by a detailed analySIS of Feynman diagrams.3) and p!J.5.=±1 ett(p. q2 as usual is q2 . yielding a term that is the same as would be produced by a tenn in the action: 1 i ~ .s.5.0.2) may be rewritten ~JI"(x  y) = (2n)4 J cfq Pt.354 8 Electrodynamics where p/H.7) where nil (0. (8. We shall choose qO in Eg.5. because the factors ql! or q.) but the easiest way to treat this problem is by pathintegral method.CTj' (8. the contribution of an internal photon line carrying fourmomentum q that runs between vertices where the photon is created and destroyed by fields all and aV is i PJI"(q) (2JT)4 q2 .5) Thus in using the Feynman rules in momentum space.7) to be given by fourmomentum conservation: it is the difference of the matter pas flowing in and out of the vertex where the photon line is created.(p) L .2) may be expressed as integrals over an independent timecomponent qO of an offshell fourmomentum ql!.q«y) r 2. (8. are coupled to currents jI' and that satisfy the conservation condition c!'J!1 = 0" The term proportional to I1 ll l1 v contains a factor q2 that cancels the q2 in the denominator of the propagator.o(p) O.6) It will he very useful (though apparently perverse) to rewrite Eq..5. (8. but qO is here entirely arbitrary.5. in the exponentials is taken with pO = Eqs. 4 ix Hl(x)]Hl(Y)] 'i e.5. (8.(qOf. (2rr)4 JIql2 • This argument as given here is little hetler than handwavmg. (8.18) and (8.4. .4) as (8.6. (8. and the photon fields all and a.(p) ~ Ipl.0.iE . so that Eq.a)ev(p.. (8.j Po.(q? q IE eiq{xy) . /d /d"y . (8.17) that Pij(p) ~ iJ.4. The terms proportional to q!' and/or ql' then do not contribute to the S·matrix. act like derivatives all and 0"..5. the theta functions in Eq. as discussed In Section 9.1) is a fixed timelike vector.4) As we saw in Chapter 6.
6. of the fann ~Jd3XJd3y l(x. The simplest gauge.FjJl'  \f Cr.2) " • In Chapter 12 we will Jiscuss rea.9) and the Coulomb interaction is dropped. that as noted in Section 5.+ i eAJi] + m)  \f' . why more complicated terms are excluded fwm the Lagrangian density. .6 Feynman Rules for Spinor Electrodynamics 355 The integral over qO here yields a delta function in time.5. the contribution of an internal photon line is simply given by I l1jJl' (2n)4 q2 . but the same formalism applies to muons and other such particles..4. For definiteness.6 Feynman Rules for Spinor Electrodynamics We are now in a position to state the Feynman rules for calculating the Smatrix in quantum electrodynamics.if' (8..~(x .6. (8. We see that the apparent violation of Lorentz invariance in the instantaneous Coulomb interaction is cancelled by another apparent violation of Lorentz invariance. 8. Our result is that the photon propagator can be taken effectively as the covariant quantity . we will consider the electrodynamics of a single species of spin particles of charge q = e and mass m.1.t j 2 4nlxYI .9 the fields all(x) are not fourvectors.I'[01'. From a practical point of view.8.1 ) The electric current fourvector is then simply JI' = 3Y cA = i e \f'yJi\f' .y) '"' = (2n)4 Jtt q 2'71'\". We will call these fermions electrons. so this is equivalent to a correction to the interaction Hamiltonian V(t).5. eiq{xy) q (8.25).. (8. This is just right to cancel the Coulomb interaction (8.t)l(y. the important point is that in the momentum space Feynman rules.and Lorentzinvariant Lagrangian for this theory is· 1 !£' =  ~ FI'.8) If' with the Coulomb interaction dropped from now on.om. and therefore have a noncovariant propagator.
with the lines joined at vertices. electrons are represented by external lines carrying arrows pointing upwards into or out of the diagram. while positrons are represented by lines carrying arrows pointing downwards into or out of the diagram. we can state the momentum space Feynman rules for the connected part of the Smatrix in this theory as follows: (i) Draw all Feynman diagrams with up to some given number of vertices. (ii) Associate factors witil tile components of the diagram as follows: Vertices Label each vertex with a fourcomponent Dirac index IZ at the electron line with its arrow coming into the vertex. .o'(k . There are also as many internal lines as are needed to give each vertex the required number of attached lines.k' + q).) As we have seen. and a spacetime index Jl at the photon line. Following the general results of Section 6. (8. For each such vertex. There is one external line coming into the diagram from below or going upwards out of the diagram for each particle in the initial or final states. (8. the Coulomb term Vcou[(t) just serves to cancel a part of the photon propagator that is noncovariant and local in time. t)"II'tp(X. at each of which there is one incoming and one outgoing electron line and one photon line.6.4.) Each external1ine is labelled with the momentum and spin zcomponent or he1icity of the electron or photon in the initial and final states. t)l Qp(x. t) + VCout(t) .3. include a factor (2n)4 e (l")p. and q is the photon fourmomentum entering the vertex (or minus the photon momentum leaving the vertex).356 8 Electrodynamics The interaction (8.I d 3x (1J'(x. The diagrams consist of electron lines carrying arrows and photon lines without arrows. a Dirac index fi at the electron line with its arrow going out of the vertex.3) (There is no Vmatter here. respectively.6.4) where k and k' are the electron fourmomenta entering and leaving the vertex. Each internal line is labelled with an offmassshell fourmomentum flowing in a definite direction along the line (taken conventionally to flow along the direction of the arrow for electron lines.23) in the interaction picture is here v(t) = +ie .
(p. include a factor vc.6.6. (8. include a factor· up(p.10) The ell are the photon polarization fourvectors described in the previous section. a) (8.6. on this line./p.t.8) The us and vs are the fourcomponent spinors discussed in Section 5.6.l on this line.9) For each line [or a photon in the initial state connected to a vertex carrying a spacetime label J1 on this line.t(p.(p.6.6 Feynman Rules for Spinor Electrodynamics 357 External lines: Label each external line with the threemomentum p and spin zcomponent or helicity (J of the particle in the initial or final state. include a factor ". (J) (2n)3/2M' (8. on this line. .4).8. a) (2n)'/2 .o) (2nV/ 2 . include a factor el. so that Ii amI f appear instead of ut and ~.7) For each line for a positron in the initial state running out of a vertex carrying a Dirac label P on this line.o) (2n)3/2 ' (8. a) (2n)3/2 ' (8.6) For each line for an electron in the initial state running into a vertex carrying a Dirac label r:J. (8.6. For each line for an electron in the final state running out of a vertex carrying a Dirac label P on thIS line.5. Internal lines: For each internal electron line carrying a four~momentum k and running from a vertex carrying a Dirac label f3 to another vertex carrying a Dirac • A matrix {j has been extracted from the interaction in (8. include a factor ux(p. For each line for a photon in the final state connected to a vertex carrying a spacetime label f. include a factor e.6.5) For each line for a positron in the final state running into a vertex carrying a Dirac label (I.
and sum over all Dirac and spacetime indices. propagators. and momentum space integrals. These quantities are not independent. E external lines. a factor (2n)4 from each internal line. (iv) Add up the results obtained in this way from each Feynman diagram. so it is important to have some idea of what numerical factors tend to suppress the contributions of the more complicated diagrams.V".if' I (8. The volume element in fourdimensional Euclidean space in terms of a radius parameter K is n 2 /(2dK 2 . include a factor [i ~ + m]~p (2JT)4 k 2 + m 2 .) For each internal photon line carrying a fourmomentum q that runs between two vertices carrying spacetime labels J. and a fourdimensional momentum space integral for each loop.3: There is a factor e(2n)4 from each vertex. Thus the diagram will contain a factor (2nJ'V e V (2n)~41 n 2L ~ (2n)4eE~2 (1::') L . and L loops. I internal lines. We shall estimate these numerical factors including not only the factors of the electronic charge e associated with vertices. roc any fourvector Vi" p denotes }'IJ.358 8 Electrodynamics label ex.81 4n X 10.6. The number E of external lines is fixed for a given process. The difficulty of evaluating Feynman diagrams increases rapidly with the number of internal lines and vertices.l and v include a raetar I 'l/(V (2JT)4 q2  jf' . but are subject to relations already used in Section 6.l. so we see that the expansion parameter that governs the suppression of Feynman graphs for each additional loop is 2 e 16n 2 = ~ = 5. as described in parts (v) and (vi) of Section 6.4 . Additional combinatoric factors and fermionic signs may need to be included. (8. Consider a connected Feynman diagram with V vertices.12) (iii) Integrate the product of all these factors over the fourmomenta carried by the internal lines.11) (We are here using the very convenient 'Dirac slash' notation.6. but also the factors of 2 and n from vertices. so each loop contributes a factor n 2 .
6. in general. we simply replace eJt(p. the polarization vectors are e(p.+I) ~ R(p) 1/J2 +i/J2 o o where R(iJ) is the standard rotation that takes the zaxis to the p direction.15) The polarization vectors for definite helicity satisfy the normalization condition (8. For linear polarization. by an adjustment of the overall phase of the state (8. for which L = 0 or rt+ = 0.18) . where not every particle in the initial and final states has a definite known helicity or spin zcomponent. which in practice are often characterized by a state of transverse or elliptical polarization rather than helicity.6. and linear polarization.6 Feynman Rules for Spinor Electrodynamics 359 Fortunately this is small enough that good accuracy can usually be obtained from Feynman diagrams with at most a few loops.6.6.+1 + IX.17) The two extreme cases are circular polarization.14) To calculate the Smatrix element for absorbing or emitting such a photon. These are not the only possible photon states. ••• We must say a little more about the spin states of photons and electrons in realistic experiments.±1 0:+ 'f'p.8. (8.13).6. (8.6. for which 10:+1 = IlLI = 1/J2. so that they can be expressed as "± ~ exp(+i<!>l/J2. we can make Cl+ and 0:_ complex conjugates. for photons of helicity +1.6. a photon state can be a linear combination of helicity states 'f'p. This consideration is especially important for photons.16) and therefore in general (8.13) which is properly normalized if 1'+1' + ILl' ~ 1 . As we saw in the previous section.'Pp. + 1) in the Feynman rules with (8.l (8.
(p) ~ R(f» o o (8.s) are the two orthonormal eigenvectors of P with eo(p.s).2 Ase.s)M'I'. each with probability Pro The rate for absorbing such a photon in a given process will then be of the form .22) where ep(p. which is only possible for linear polarization. s) with probabilities A.s) and ~ e. s=1. In between the extremes of circular and linear polarization are the states of elliptic polarization. an initial photon may be prepared in a statistical mixture of spin states. where P is the density 11U1trix Pvp (8.1>0. More generally..(p.6.. if we know nothing whatever about the initial photon polarization. with . Note that the photon polarization vector here is real.6.6. ¢ is the azimuthal angle of the photon polarization in the plane perpendicular to p.r)(p)e~)*(p). (8. then the two probabilities As for the polarization vectors .21 ) Since P is obviously a Hermitian positive matrix of unit trace (because Lr Pr = 1) with PvO = POjJ = and Pvppv = Pv/tp!J = 0..s)p' ~ 0 (8.~1.(p.6.6.2 We may then write the rate for the photon absorption process as [' ~ I: l.s)e~(p.le.19) That is. for which 10:+1 and la~1 are nonzero and unequal.6.20) =L Pr e. it may be written as ° PI'P = L s=1.I • In particular.(p. s=1. (8.23) ":s are the corresponding eigenvalues.'.360 8 Electrodynamics Then in the Feynman rules we should use a polarization vector cos rP sin cP e.2 (8. an initial photon may have any number of possible polarization vectors etl(p).24) Thus any statistical mixture of initial photon states is always equivalent to having just two orthonormal polarizations ev(p. I: l. In the most general case.
. "') A.(p'. such as those with spin zcomponent a = +!.s) over which we average. (8.8. Hence if neither electron nor positron spins are observed. a') J"t (lp up(p.5.27) ap where pO = V 2 + m2.5. if we make no attempt to measure a final electron's or positron's spin state. such as those with spin zcomponent a = +~.2 ei(p.1=1.PiPj) .25) Fortunately.2 = 361 !' and the density matrix (and hence = the absorption rate) is an average over initial polarizations Pij = ! L .38): '" _ (i 20 p+m) LU'(P. if we make no attempt to measure the polarization of a photon in the final state.6.")up(p.s) are equal. if the initial state contains an p electron with momentum p and spin zcomponent {1. then the Smatrix element for the process will be of the form (V~(p'.")~ ° a p (8. so that ).")vp(p.s) ~ (e5 ij .37) and (5. The same remarks apply to electrons and positrons. a». then we must sum the rate over any two orthonormal initial spin states. then the rate may be calculated by summing over any pair of orthonormal final photon polarization vectors. Similarly.1.(1 I(v.")~ a p ~P ' (8. then the rate is to be calculated by averaging over any two orthononnal initial spin states.6.. if (as is usually the case) we make no attempt to prepare an electron or positron so that some spin states are more likely than others. for unpolarized photons we can average the absorption rate over any pair of orthonormal polarization vectors.6 Feynman Rules for Spinor Electrodynamics ev(p.1 =. "))1 2 ~ !Tr{PAt # 4 (i 2jJ'p m) A (i 2p+m)} p O' O Techniques for the calculation of such traces are described in the Appendix to this chapter. Such sums may be perfonned using the relations (5. For instance. and a positron with momentum p' and spin zcomponent ai.26) '" _ (i 2pm) ' LV'(P.p Up(p.s)ej(p. this result does not depend on the particular pair of polarization vectors ei(p.6. the rate will be proportional to 1L a'.
O"')W (2rrj3/2 e~' U(P'O")CI elJ (2rrl3/2J2ko' (2rr)3/2 (2rr)3/2J2ko x +m j d" q [i] [q2i q2 if: ] CI'/i (211")4 +m X { [e(2rrl 4y.kl] + [e(2rrl 4Yi". e'll .' a .e ~ p.e') = u(p'. tJ' and p'lJ. where kO = Ikl and kO' = Ik'i. 0 ~ p.k'l] [e(2rrj4y. 8.".6 4(q + k . Straight lines arc electrons.e _ p'.1l . the corresponding Smatrix element is S(p.I. ai.1. 0 Figure 8. Also. to lowest order in e.O c>' .p'l] [e(2rr)4 y. q a ~" ". p a ~ a k. The two lowestorder Feynman diagrams for Compton scattering.7 Compton Scattering As an example of the methods described in this chapter. wavy lines are photons..7. Using the rules outlined in the previous section.1.p . the initial and final electron momenta and spin zcomponents are labelled pit. ell and k'Jl. We label the initial and final photon momenta and pOlarization vectors by kll .a' + k'. 8 Electrodynamics .0 k . .362 ~..6 4(q . p.6 4(q + k' . ~.pj] }.O" + k.p' .6 4(q . with m the electron mass. where pO = J p 2 + m2 and p'o = Vp'2 + m2 . p . The lowest order Feynman diagrams for this process are shown in Figure 8. we shall consider here the scattering of a photon by an electron (or other particle of spin! and charge e). (8.
[i(1W)+m] (/Pk'}U(p.7.7 Compton Scattering 363 Performing the (trivial) qintegral.7. the laboratory frame for highenergy (X ray or gamma ray) photonelectron scattering experiments is usually (though not always) one in which the initial electron can be taken to be at rest. (876) The differential crosssection is given in terms of M by Eq. because the denominators here do not vanish.k')2 + m2 ~ 2p' .10) . (8.3.3) (p .) Because p2 = _m 2 and k 2 = k'2 = 0. collecting factors of i and 2n.7. which here reads da = (2n )4 u.17) for the initi<ll velocity gives (8.7. (~. we drop the ~it:. which (because some scattering is assumed to take place) here reads r s ~ 2ni6 4 (p' + k' so p ..5) M~ 4(2n):~kOkO'U(P'"'){ ( [ i(1+ ~)+m] '/p.~~l2:2m) "]u(p. pO~m.a) (Here means e. the 'Feynman amplitude' M is defined in general by Eq. 2 (872) + .1IMI 2(i4(p' + k' ~ P ~ k )d3pi d3k' . and rewriting the result in matrix notation.k)M .4. Eq.7.8) To go further. we have more simply S ~ _ie b (p' + k' .J2k"'2ko (p. k.' (2n)2. Since electrons in atoms move nonrelativistically.7.7) Since one of the particles here is massless.k .7. .7. (3. (3. (8. not (f)*. (8. so that p~O.4) Also.8) is then simply u = 1.2).p _k)u ' a') [ . (8. Also.8. it will be convenient to adopt <l specific coordinate fr<lme. (3.. the denominators can be simplified (p+k)' +m 2 ~ 2p'k.15).4.yJi. We will adopt this frame here.9) The velocity (8. 2 4 (i( JI + (p+k)2+ m ~) + m) . (8.a). (8.
7.k'. (8.w.15) p'OW' I J(w .)w 2 2wo/ cos e + 0)'2 + m2 + ro']/8w'l J(w' .7.12) The threemomentum delta function in Eq.(O)).7.7. where 8 is the angle between k and k'. Squaring both sides and cancelling w'2 terms gives· w ~w m+w (I m COSo ") .'~i+""11 (cc" (8.' w  IcosO m The verification of this formula in thc scattering of X rays by electrons by A. is the solid angle into which the final photon is scattered.. (8.7.16).7.7.w. (8. (8. 2ww'cos8 + w'2 +m  w' .15) just serves to eliminate the differential dw' in Eq.(O)) "I w 'wcCC'os"OcC)l'ip.7.7.7) just serves to eliminate the differential d3p'. mw given by Eq.11) (8.17) • Equivalently. This leaves the remaining energy delta function J(p'o +k'o _p' _k o) ~ J (j(kk')' +m' +w' mw) + m2 = OJ (8..14) The energy delta function (8. and III £0' dO. leaving us with a differential crosssection du ~ with p'o = (2n)'IMI 2 '0 '3 m + £0 .14).13) This fixes w' to satisfy Vw 2 . .13) can be written J(p'o + k'o _ pO _ kO) ~  J(w' .7. we denote the photon energies by w~ko~lkl~p'k/m. which soon after Compton's experiments came to be known as the photon.(O)) [a[ . The final delta function in Eq.H. (8.w'. mw Also.w.7. (8.364 8 Electrodynamics To save writing.7. w'~k"~lk'I~p·k'/m. p w (8. Compton in 19223 played a key role in confirming Einstein's 1905 proposal of a quantum of light. there is an increase wavelength 1 1 ". the differential d 3k' can be written ~ d3 k' ~ w"dw'do. (8.16) where dO.(O) .w. setting pi = k .
not (epyll)".pi.6) gives now "[MI' ~ .u. we must sum over (1' and average over (1.u)~ 2 " p 0 (8.[i Ii + m] ~ [i ~ .e _ p' + k'..u)up(p. a))(u(p.21) can therefore be written in .P" + m2 ) ~ 0 .' [i(li  p'k ' ~') + m] .e 4 0 '0 (8.) such as for instance Coulomb gauge in the laboratory frame. and likewise for work in a 'gauge' in which r t.7. or in other words take half the sum over (1 and a': du(p + k.7.[i(li.a a'.e . This implies that [i Ii + m] .a' Recalling that f3Y~P = 'II" Eq.. a )1 ~ 2 L(u(P'. (8. (8.8. a)pAtpu(p'. u')Au(p. (8.. and r.' [i(li + p'k ~)+m].18) To calculate this. It follows that for an arbitrary 4x4 matrix A L a.V')+m] t}(ili+m) p'k ' p'k ~)+ m] . where eO = e'o = 0 and p = O. we use the standard formula " _ (ili+m).(p.a' a. We (Recall again that means eZyP.7.p L. and likewise for ¢*. a' + k'.7. Eq. 64(2n) ww'p p ..} (i Ii' + m)] .7.21) x Tr [{ t[i(li+ x {.( Ii' + m2 ) ~ .(p.[i Ii + m] Ii + m] .a' lu(p'..e' ). ¢'. a')Au(p.19) and likewise for the sum over a ' . In such cases.7 Compton Scattering 365 Usually we do not measure the spin zcomponent of the initial or final electron.' _ ..e')  ~ Lda(p. a')) a.a + k._ .
7..7. 12 ~ Tr { I' ~ / N r} . where ell and e'll are real. (p·k)(p·k') + T) (p'k)(p'k') m2t4 ) l ro f2 rolf} +(p'k')2 (p'k)2 .7.) To simplify the calculation further. b. .(p'k')2 (8.7.24) where TI r u' ti'} . d. . (8.: 4 e (TI 0p'0 (p'k)2 64(2n)6 ww 'p T4 rolf! + T. c.7. respectively. T) ~ Tr { U' I' u u' ti'} . we also have k·k = k'·k' = O.7.. in addition to Eg.32). T2 ~Tr{ I' U tII'~' r ti'} . .(p'k)(p'k') .7.31) (8.32) I' u r u} .) ~ Tr { / ~' I' r U. T4 ~ Tr{ /~' I' iJ /' ¥' r ti'} . Dropping the asterisks in Egs..} . '4 ~ Tr { / ~' I' 0' r} .22).(p'k)(p'k') ..26) (8. ~ Tr { I' ~ II! (8.7. traces of products of 6 or 8 gamma matrices like the tk or Tk would be given by a sum of 15 or 105 terms.7.7.366 8 Electrodynamics the greatly simplified form The trace of any product of an odd number of gamma matrices vanishes. we have then . (Furthermore. In general.} as a sum of products of scalar products of the fourvectors a. let us specialize to the case of linear polarization.7.27) (8. but fortunately here most scalar products vanish.29) (8.7..28) (8.25)(8.25) (8. (8. so this breaks up into terms of zeroth and second order in m: "IMI' ~ :!. e'e* = e"e" = 1. II ~ Tr { The Appendix to this chapter shows how to calculate any trace Tr{ ¢ ~ ¢ I.30) (8.
A. p')(p' k'). Combining all these terms in Eq.35) (8.7.7.34) (8. In the lahoratory frame.4(k . [p + k . ' 1) w w ~ ~ P .33) A similar (though more lengthy) calculation gives T 2 = T3 = 8(e' k')2(p. so and hence T r ~ 2p' k Tr { " Using Eq. ~ 4 e .k'] = e' . k [2e' . e')2] (8. p) + 32(e.8. k' = mw' .Hp~k')2 ~ ~m2 = p' k' so T j = 16 p' k(e'· k)2 + 8 p' k p' k' . e) + 8(e' . kJlk l1 = 0. we have the special results k· k = ww (cosO 1) = mww p .(I . p'] . k k· p' = ~ Hp' ~k)2  ~m2 = .4(k . k') . TI = ~8p . e + 8(p' k')(e' e')2 .7. t2 = t3 = 8 e' e' k .k . this is U' l} . k) + 16 (e' e')2 p ' k'p' k + 8(e' e')2k' k' m2 (8.36) (8. (8. we have so Also. (8. e' k' .6). k')(p· p') . It is convenient to make the substitutions e' . k)'p' k' 4(k' p)' + 4(k .7. k = mw " .7.7.:. 64(2n)6 ww 'pOpo [ 8(k' k')' (k' p)(k'. p' . . p' = e' .7.7 Compton Scattering 367 Since eJipJl = 0 and eJie l1 = 1.38) All this applies in any Lorentz frame.24) gives '" IMI2 . e')(k' . T4 ~ 16 p' V(e' k')2 + 8(p' k)(p' V).37) 8(e' e')m'(k . k e' . tl = t4 = 0 . (8.
In the nonrelativistic case.O" k. . As discussed in Section 8.7.O"' + (8.O" 0. . using " This gives l L e.40) over e'. if the incoming photon is (as usual) not prepared in a state with any particular polarization. Klein and Y.7. Nishina4 in 1929.7.e e.e') (8.7. perpendicular to the plane in which the scattering takes place.38) with Eq.jk.7.e' .e') = e w II [W W 222 ~+~2(k" 'e) 64n: m OJ OJ' OJ 4 '2d ' 2] . (8.e + p'.0.o.0' + k.e 0/ + p'. !(b.368 8 Electrodynamics Combining Eq.7.17).kj ) and the differential crossMsection is then ~ " Lda(p. This average gives ~Leiej= x [ ~+~ OJ " . (8. cooler. ' .0' dO"(p.. (8. This polarization is normally undetectable because it cancels when the astronomer adds up light from all parts of the star's disk.O" + k.o 1 + "+ p.6. the laboratory frame crosssection is ! LdO"(p. we must sum Eq.40) We see that the scattered photon is preferentially polarized in a direction perpendicular to the incident as well as the final photon direction.O"' + k'. (8. responsible among other things for the polarization of light from eclipsing binary stars"· To calculate the crosssection for experiments in which the final photon polarization is not measured.a+ k. •• The light from one of the stars is polarized when it is scattered by free electrons in the outer atmosphere of the other. « m. star when both are along the same line of sight. The polarization has been observed in eclipsing binary stars at times when the cooler star blocks the light from just one side of the hotter star.39) w' w This is the celebrated formula derived (using oldfashioned perturbation theory) by O.e. then we must average over two orthonormal values of e.41) (J) e4 w'2dQ w w' ~+~~1+cos2tJ ] 32n: 2m2(l)2 [w' W where eis the angle between k and k'. . This is a wellMknown result. i.
43) is called the Thomson crosssection.7.42) and (8.8. and may be omitted in a first reading.8. (8.43) This is often written O"T = 8nrfi/3. the discoverer of the electron. u=3' giving a total cross·section for < m: 6 "T ~ .8 Generalization: p. J. + (l)Po!Jp+l tl'lJl2""'J!P (8.8.. where ro = e2 /4nm = 2.8 Generalization : pform Gauge Fields Eg.m2' e' 2 (8. 8. while a pform that is itself an exterior derivative is called exact. Expression (8..7.818 x 10. Because derivatives commute.7. by calculating the reradiation of light by a nonrelativistic point charge in a plane wave electromagnetic field. Thomson.43) were originally derived using classical mechanics and electrodynamics. in part because the exterior derivative of a tensor transforms like a tensor even though it is calculated using ordinary rather than covariant derivatives.41) gives 369 (8.7. (8. From a pform tJ!I. From Eq. Eqs.42) The solid angle integral is J+ [1 cos' O]dfl ~ 10'" d</> fa" [1 + cos' OJ Q) Slnu .form Gauge Fields· The antisymmetric field strength tensor F1lV of electromagnetism is a special case of a general class of tensors of special importance in physics and mathematics.1) with square brackets indicating antisymmetrization with respect to the indices within the brackets.J!2""J!p one may construct a (p + I)form called the exterior derivative" dt by taking the derivative and then antisymmetrizing with respect to all indices: (dt)J!lM'"'J!P+1  0!Jlj tM1W'J!p+il 01'2 t J!1!J3'J!P+l 0J!I tJ!21'3"'J!P+l  + . (8.7. "d" 16. after J. •• Exterior derivalives and pforms playa special role in general relalivity. 1 .8. (8.7.2) A pform whose exterior derivative vanishes is called closed.13 cm is known as the classical electron radius. repeated exterior derivatives vanish d(dt) ~ o.2) • This section lies somewhal out of the book's main line of development. A pform is an antisymmetric covariant tensor of rank p.
The formalism of pforms and exterior derivatives makes it natural to consider the possibility of massless particles described by pfonn field gauge fields+ AI""'.2)..8.up~1 + J. From such a pform gauge field.uA v ~ o. We are speaking loosely in calling A~I'~" a pform. we might expect the Lagrangian density for A to take the form . Poincare's theorem then shows that it is also exact. we have already seen that in four spacetime dimensions.. "'n .1. (8. + .!' which will be discl. (8. we can construct a gaugeinvariant field strength tensor given by (884) or in detail (8. There is a deep relation between the de Rham cohomology groups of a space and its topology.ul"'.lsscd further in Volume II.5) (Alternatively. In fact.1.16) tell us that the electromagnetic field strength twoform FIl " is closed. we see that the twoform FJiv is invariant if All is changed by an exterior derivative." rp+ rr I' .8.7) t III multiply connected spaces dosecJ forms are not necessarily exact.F!'l "'.) By analogy with electrodynamics... makes up what is called the pth de Rham cohomology group of the space.u J.e. i. we can start with a (p + I)form F. although it is possible to write a dosed pform as an exterior derivative locally.ur = ~ dO (8. and from an assumed condition dF = 0 infer the existence of a pform A with F = dA.6) where J is an antisymmetric tensor current (either a cnumber.8. the homogeneous Maxwell equations (8. i. or a function of fields other than A) that in order to make the action gaugeinvariant must satisfy the conservation condition (8.uI . so that it can be written as an exterior derivative. because for F = dA to be a ten. it is not possible to construct a fourvector field from the creation and annihilation operators of physical massles~ particles of helicity ±I.or it IS only necessary for A to be a tensor up to a gauge transformation. so we have to deal with an A~(r) that a\:cording to Eq. .Sf =  2(p 1 + 1) F". any closed pform is exact t For instance. this cannot in general be done smoothly throughout the space.ul"'!'r An.2) transforms as a fourvector only up to a gauge transformation.8. by a gauge transfonnation bAit = 0llfl.!tr_1 is an arbitrary (p .. p where n.A w Again using Eq. a famous theorem 6 of Poincare states that in a simply connected region. The set of closed pforms.up.e" as FJlv = 8.3) a[Pln/12 "'. modulo exact p·forms. (8.1)form.8.370 8 Electrodynamics it follows that any exact pform is closed. with an invariance under gauge transformations OA or in more detail ()A.
the field strength F may be expressed in terms of a dual (D .12) tell us that the difference of ff and c'/ is closed. (8. (8.8. it may be written (D . p·forms offer no new possibilities.8.8. and therefore according to Poincare's theorem.2)form. df ~ 0 .1. There is an exception for the case p = D .12) Eqs."'/lJ41' = 0 /11. (8. To see this. the homogeneous 'Maxwell' equations dF = 0 read :1 S!i.e.8. For instance. note first that in D spacetime dimensions there are no antisymmetric tensors with more than D indices. For p < D . there is a normal mode of the string represented at low energies by a twoform gauge field A/Iv.9) 10 Likewise.8 Generalization : pjiJrm Gauge Fields 371 The EulerLagrange equations are then i'J " FIIIII "'!lp  _Jill "!ip (8.13) yields the field equation for ¢: 0111 (d¢ )JlI'"'IIIJp = _(dY)Il::'''!iJ)P ./ as terms of a (8.8.I)form .2. as FJiI "'/lr+l  _ E II] '"'/ID of ~ /lp+2' 'JiD .7) then read simply d.8) and conservation condition (8.8. III (8.15) This is invariant under a new set of gauge transformations ¢ ~ ¢ + dw.8. scalars. and the condition dff = d/f simply tells us that Y and Y diff. as terms of a dual (8.8.2 = 0 the gauge transformation that leaves F . (8.8.I)form . (8. the pform current J may be expressed (D .7 0. except that where D .8) These pform gauge fields play .p .10) The field equation (8. We may therefore concern ourselves only with the cases p < D . In this case the gauge field describes no degrees Jf freedom at all.p .p)form current f.'!I'. (8. i. in the simplest string theories in 26 spacetime dimensions.11) and (8.'F=/.14) which with Eq.13) with ¢ a (D .I is closed..8. may be written as /ff' ~. Like any other (p + I)form with p + 1 < D. But in four spacetime dimensions.p . so in general we must take p + 1 <D.:In important role in theories with more than four spacetime dimensions.8.11) Because the dual current .8.'I' + d¢ .p . where ff and Yare zeroforms.2. r only by a constant.8.8.
or p = 0..2. for which Eq. For products of even numbers of gamma matrices. according to the general result quoted above.··· p. with current dY'. P3 < P4. Finally.8.5) reads F!J.372 8 Electrodynamics invariant is ¢ + ¢ + c. P2. in which we pair {lPI with {lP2.lS within a pair yields the same pairing. As we have seen. . This is just the theory of a massless scalar field with only derivative interactions.'i op II 1Jpaired paIrs liS' (8. We can now understand why pform gauge fields offer no new possibilities in four spacetime dimensions. A zeroform gauge field is a scalar S. P . just as in electrodynamics. 2N into some order PI. 2.A..2)form gaugejield ¢.2) We can avoid summing over equivalent pairings by requiring that PI < P2.I) Here the sum is over all different ways of pairing the indices {II. We see that in D spacetime dimensions. . and the field equations (8.4) With this convention. (8. the trace is given by !. l1p3 with l1p4.3) and PI < P3 < P5 < . (8. The product in Eq. It will therefore be useful to give formulas for these traces that can be used in all such calculations.P. where c is an arbitrary constant. the factor b p is + 1 or 1 according to whether the pairing involves an even or odd permutation of indices. A oneform gauge field is a fourvector AIl(x) coupled to a conserved fourvector current. = ails. (2N . (2N) (8A..A.. and so on. The gauge invariance here is invariance under a shift S _ S + c.I) < P . (8A. a twofonn gauge field in four spacetime dimensions is equivalent to a zeroform gauge field."'. A pairing can be regarded as a permutation of the integers 1.8) read simply oS = J. or 2. (2N). the theory of a pform gauge field A with current J is equivalent to the theory of a (D .8.'·· {l2N. Tr{YIlIYlll· ··YIl2!J = 4 L pairing. 1. which as we have seen is equivalent to a massless derivatively coupled scalar field. so the number of different pairings is (2N)!/N!2 N ~(2N~I)(2N3)'''1 (2NI)!'. with c a constant. Permuting pairs or permuting J.. we need only consider the cases p < D . Appendix Traces In calculating Smatrix elements and transition rates for processes involving particles of spin we often encounter traces of products of Dirac gamma matrices.
I) is true for N < M 1.8) The proof of Eq. the nth pair contributing a factor ~IlP... in agreement with Eq.} ~ .I) also correctly gives the trace of any product of 2N Dirac matrices.Tr {I' Jl2 Yill 1'113 •.. + 211P11l2M Tr {YP2 . "').v. I} ~ + B~w ' so Tr{7I'Yv} = 4'1/1V.. All commutators have zero trace.A.} = 4 ['1/1V ~P" Tr {I'llI' .1 Tr{}'/12'1/L1'''YJl2M} + 11111/L1 Tr{Y/12YI'3YII5 . '/11 = YSYp(YS)I.Appendix Traces 373 (8.A. }'Jl2M YIII} . /12. yI' I' u I' '1IlP11vo" ~1(1f I( 11I'P~"" + ~w. (8.Tr {Mp} + 2Tr {~p. Traces are unaffected by such similarity • There are now computer programs 7 available for the calculation of traces of products of large numbers of Dirac matrices.(2"I)IlP'(2")' For instance (writing Il. Next...A. (8.  (B.. (B. We then have Tr{ I' III Y1'2 ••• y1'2M} = 2 '11'1Jl2 Tr{ I' III ..6) Y'1} = 4 [11 I'V 11 pO" 111('1 (flf  ~1'" 11pK ~ 0"1f + 11I'v 111''11101( + 11 I'P ~ "1(11 + 11JUT11 vp111('1 ~ '1110" 11\' K~ 1''1 + '11''111vp'IO"/C (8..• YI'2MI} . The easiest way to see that the trace of an odd number of Dirac matrices vanishes is to note that I'I' is related to }'P by a similarity transformation. (8.5) Tr {I' Ilh '}' p y.I) correctly gives the trace of any product of 2N .~vP] ~ I'P11v'1 11 (fl( .A. YP2M} = 2111'l!12 Tr{ 71'3 ••.A.2111'1/13 Tr {'Y1'21'/4 .~1l/C11vp'Iulf + ~1'/C11vo11p'1 . so the last term subtracted here is the same as the lefthand side. "YIl2M} . suppose that Eq. /13.p. Tr{Y/ll"l1'2'" YJl2N+I} = For an odd number of gamma matrices.A. + ~1l11'2M Tr {}'Jl2 .2 Dirac matrices.l '''YIl2M} = 2 111l11'2 Tr{ 7113 .A. (8.. and 3 we have· Tr{YIlY.·} = 4"llv. First note that Tr {ypy. for N = 1. }'Jl2M} ~ 2 11/11113 Tr {"I1121'1'4 ••• Y112M} + 211PIIl4 Tr{7112Y/1 3 YP5 ···7112M} ~ .. YI'2MI} • (8.A.Tr {'11l2 •. (8.4.:1 .'" in place of Ill. the result is much simpler o.9) shows that Eq.1) is over all N pairs.9) If we assume that Eq.A. .2.Tr{M...I). (8.) (8. and so Tr{YI'I'YP2 "'i'1'2M} = '11'1Jl2 Tr{}'1l3" 'YJl2M} '1//11'.A.•• }'1l2N} + Tr{YIl2YIl3'Y1l 1Y/l. 114... yJl2M} . 'Y1'2. .a..AJ) is by mathematical induction.?) +'1I'O"'1v'1~p/C .111'1('1V1/11p(f ~'1I''111vu11PK+'1I''1'1V/('1pu]. then Eq..
Use Coulomb gauge.374 8 Electrodynamics transformations. It also vanishes for n = 0 and n = 2: Tr {y.) ~ 0 (RAIO) (RAIl) Tr{m"Y. V. 1.}~O.m 2<1>t<1> . Use the simplest Lagrangian for the electrodynamics of electrons and muons. (8. v. V.. or more Dirac matrices may be calculated by the same methods used above to verify Eq. p. Carry out the canonical quantization of the theory of a charged scalar field <I> and its interaction with electromagnetism. Furthermore this trace must be odd under permutations of 11. Thus the trace Tr {i'sYtt}'vYPY"'} must be proportional to the totally antisymmetric tensor e/lI'p". One occasionally encounters another class of traces. Y/IJ . Calculate the differential and total crosssections for the process e+e. with Lagrangian density. (l are some permutation of 0. p.'s iYO}'lr2r'3. but only if /1. . Express the Hamiltonian in terms of the fields A.p.3. of the form Tr {}'5}'111 (112 •.. 2. 1. 2.!Fl'vFI'" . Evaluate the interaction .iq A /l<1> .(J since gamma matrices with different indices anticommute. F/ll' .'5i'/1i'1' }'pi'"._ 11+JC to lowest order in e. eight..' The constant of proportionality may be worked out by letting 11.2. <1>.} = 4ie/ll'p'" . and note that there is no way of pairing the indices in Tr {"Ioi'rf2Y3} or in Tr {YoY) r2r3'j'jl'fl'} so that the spacetime indices in each pair are equal. (8. and <1>t and their canonical conjugates.) For n = 4 it is possible to pair the indices in Tr {rO}'Ii'2Y3}'/l'Yvrpru} so that the spacetime indices in each pair are equal.AI2) The trace of products of}'s with six. and recalling that eom I.(J/lA 1.i(<I>t<1»2 . (J take the values 0. so the trace of an odd number of Dirac matrices is equal to minus itself.!£ = (D/l<I»t(D/liD) . This vanishes for odd n for the same reason as given above for traces without a rS. Assume that electron and muon spins are not observed. Problems 1. In this way we find Tr {. and hence vanishes. 3.AI). where DIJ<1> a/l<1> . (To see this just recall that .  i\A/l .
Also see F. Rev.f Phys..· References 375 V(t) in the interactionpicture in terms of the interaction picture fields and their derivatives. . 3. 78. See. Comput. 5. Klein and Y. 375 (1927). N. H. 56. Weyl. 52. 330 (1929). 77. Phys. Flanders. Assume that final and initial spins are not measured. Z. 2. Materie.769 (1949): Section 8. 1972): Section 4. 101.. P. 1987): Section 2. O. 6. West. Weyl. 62. References 1. 39. 853 (1929). Tamm. Z. ibid. B. R. Phys. For a readable general introduction to the geometry and topology of pforms. Commun. 7. Y. M. Phys. Schwarz. Use the results of Problem 2 to calculate the differential and total crosssections for photon scattering by a massive charged scalar particle to lowest order in e. The use of Coulomb gauge in electrodynamics was strongly advocated by Schwinger on pretty much the same grounds as here: that we ought not to introduce photons with helicities other than + 1. Weinberg. 5. Calculate the differential crosssection for electronelectron scattering to lowest order in e. Z. S. 226 (1927). V Fock. The term 'gauge invariance' derives from an analogy with earlier speculations about scale invariance by H. also see I.2. 1920). 3. Schwinger. and E. Nisbina. 10. 869 (1929). Differential Forms (Academic Press. f Phys. New York. Witten. see H. talk at City College (unpublished). Feynman. London. 1963). (SpringerVerlag. New York. J.g. f Phys.. See. 4. Nishina. Berlin. This history has been reviewed by C. 324 (1962). H. 3rd ed. Write a gaugeinvariant Lagrangian for a charged maSSIve vector field interacting with the electromagnetic field.g. e. e. 127. See J. Green. Yang. 545 (1930). Gravitation and Cosmology (Wiley. 1439 (1948). Rev. Superstring Theory (Cambridge University Press. 4. Z. Phys. 286 (1993).11. Nuaea Cimenta 30. Zeit. Z. T. 278 (1963). Cambridge. 42.I Phys. in Raum.
such as the scalar field with derivative coupling or the vector field with zero or nonzero mass. The interaction Hamiltonian turned out to contain a covariant term. when Faddeev 376 . though it is local in time. which were shown by Dyson in 1949 to lead to the same diagrammatic rules that had been obtained by Feynman by his own methods. 2 The pathintegral approach played a part (along with inspired guesswork) in Feynman's later derivation of his diagrammatic rules. D. 3 However. although Feynman diagrams became widely used in the 1950s. In this respect. equal to the negative of the interaction term in the Lagrangian. This was first presented in the context of nonrelativistic quantum mechanics in Feynman's Princeton Ph. Yet the final results are quite simple: the Feynman rules are just those we should obtain with covariant propagators. most physicists (including myself) tended to derive them using the operator methods of Schwinger and Tomonaga. One would very much prefer a method of calculation that goes directly from the Lagrangian to the Feynman rules in their final. such a method does exist. like the nonAbelian gauge theories to be discussed in Volume II.9 PathIntegral Methods In Chapters 7 and 8 we applied the canonical quantization operator formalism to derive the Feynman rules for a variety of theories. and also general relativity. In the case of electrodynamics this noncovariant term (the Coulomb energy) turned out to be not even spatially local. which was bad enough for the theories considered in Chapters 7 and 8.l as a means of working directly with a Lagrangian rather than a Hamiltonian. plus a noncovariant term. The awkwardness in obtaining these simple results. it was inspired by earlier work ofDirac. the procedure though straightforward was rather awkward. It is provided by the pathintegral approach to Quantum mechanics. thesis. and using the negative of the interaction term in the Lagrangian to calculate vertex contributions. Lorentzcovariant form. becomes unbearable for more complicated theories. which served to cancel noncovariant terms in the propagator. The pathintegral approach was revived in the late 1960s. Fortunately. In many cases.
and that also depends on the way in which we define the scalar field. Soon after. using the naive Feynman rules derived directly from the Lagrangian density would yield an Smatrix that is not only wrong but even nonunitary. the only way to show that the pathintegral formalism yields a unitary Smatrix is to use it to reconstruct the canonical formalism. it was discovered that the pathintegral method allows us to take account of contributions to the Smatrix that have an essential singularity at zero coupling constant and therefore cannot be discovered in any finite order of perturbation theory. For most theorists. In such theories. Since the path~integral approach is here derived from the canonical approach. There is a kind of conservation of trouble here. in a gauge that made the high energy behavior of these theories transparent. or the pathintegral approach.the two approaches yield the same Smatrix. Feynman seems at first to have thought of his pathintegral approach as a substitute for the ordinary canonical formulation of quantum mechanics. . Since then. when 't Hooft6 used pathintegral methods to derive the Feynman rules for spontaneously broken gauge theories (discussed in Volume II). the turning point came in 1971. As far as I know. in which unitarity is obvious. is simply wrong.Pathlntegral Methods 377 and Popov4 and De Witt 5 showed how to apply it to nonAbelian gauge theories and general relativity. we know that . we can use the canonical approach. with Lagrangian density Sf = 1gk/(¢)oprjla Jl c/l. The second reason for introducing the canonical formalism first is more practical: there are important theories in which the simplest version of the Feynman pathintegral method. as also discussed in Volume II. At this point the reader may be wondering why if the pathintegral method is so convenient we bothered in Chapter 7 to introduce the canonical formalism. in which unitarity is obvious and Lorentz invariance obscure. One example is the nonlinear 6model. it does not make clear why the Smatrix calculated in this way is unitary. Indeed. 7 In this chapter we shall derive the pathintegral formalism from the canonical formalism. including in particular the theory of weak and electromagnetic interactions. so that the Smatrix must indeed be both Lorentzinvariant and unitary. which is manifestly Lorentzinvariant but far from manifestly unitary. the pathintegral methods described here have become an indispensable part of the equipment of all physicists who make use of quantum field theory. and in this way we will see what additional sorts of vertices are needed to supplement the simplest version of the Feynman pathintegral method. The first is a point of principle: although the pathintegral formalism provides us with manifestly Lorentzinvariant diagrammatic rules. There are two reasons for starting with the canonical formalism. in which propagators and interaction vertices are taken directly from the Lagrangian.
t = 0).1. Our results will be generalized to include fermionic operators in Section 9.1) and (9.1.1.1.1.n .lq) ~ q. and any remaining ~econd cla.m .m :J. These are 'Schrodingerpicture' oper· ators.1.378 9 PathIntegral Methods 9. Since the Qa all commute.1. ~ q.m .6) (We are using lower case qs and ps here to denote eigenvalues rather than to denote operators in the interaction picture as in Chapter 7. for the present it will be convenient to use the more compact notation of Eqs. Also. we can find a simultaneous eigenstate Iq}. the index a consists of a position x and a discrete Lorentz and species index m. (9. but since we will not be using the interaction picture in this chapter no confusion should arise.q). (q'lq) ~ II 6(q~ .b 3(x . with eigenvalues qa: Q. described by Faddeev.hrect apphcation or g pathintegral methods to con~trained ~yslem~ i.1. .) The eigenvectors can be taken to be orthonormal. the Kronecker delta in Eq. (9.1.1.5) However.) 6(q' .1) is interpreted in a field theory as bx. (9.) In a field theory.J. Iq) (ql . (9. The timedependent operators in the Heisenberg picture will be considered a little later.2). (9.6. and we conventionally write Qx.s constraints are eliminated by choosing a gauge.4) Px.Qm(x). as in Section 7.Pm(x). which satisfy commutation rather than anticommutation relations. (9.7) so that the completeness relation reads 1 JII " dq.5. with Hermitian operator 'coordinates' Qa and conjugate 'momenta' Pb.8) • We are tacitly assuming here that any Jirst cla. (9. taken at a fixed time (say.olved' by writing the constrained degrees of freedom in terms of the unconstrained Qu and P".2) (We shall restrict ourselves in this and the next three sections to bosonic operators. (9. The . satisfying the canonical commutation relations·: [Qa.traints arc '.s con.1 The General PathIntegral Formula We start with a general quantum mechanical system. Pb] = i (jab.3) (9.y)b mn . Iq) . (9.1) ~ [Q" Qb] ~ [P" Pbl o.
p) . This is why its timedependence is given by a factor exp(iHt) rather than exp(iHt).LlO).. (9.t. (9.(J/!Jqh on ..1. (p'.t)~b(q'q).tlq. the ~ame lines as in the quantum mechanics of point partide:. Iq. P.lm Eq.1. (9120) (9.11) JII .19) If] dp.t) P.1.1. ('i..1.1.9) (P'lp) ~ 1~ II J(p~  p. exp(iHt). p. •• The prouf rulluw:.).t) (p.(i) .22) dq. These have eigenstates Iq. we ~ee that Ph act!.1. t) ~ b(p' .1.lp) . Ip. we can find a complete orthonormal set of eigenstates of the P" : P.1 The General PathIntegral Formula 379 Similarly. tJ Q.Ul) i~ then ~een to be the wave function in this basi:. (9.1. TI .1.14) where H is the total Hamiltonian. The righthand ~ide of lOq (9. 1. As usual. t) given by q.t) ~ "p(iHI)lp)· (9. the Q and P operators are given a timedependence Q.. and not the result of letting the state Iq) evolve for a time t. dp.(t)lp.. Ip. is the eigenstate of Qa(t) with eigenvalue q".1.exp(iHt)Q.1.18) (Note that Iq.iIii: is f1~ed by the normalization requiremcnt.1.p.16) Iq.17) (9.lq. I).) These states obviously satisfy the completeness and orthonormality conditions (q'.lp. 'i n 1 (9.lp) ~ p.13) (9. as ·_.(t) = exp(iHt)Pa exp(iHt).t) ~ exp(iHt)lq).ave function~ in a qbasls. . it follows from Eq. (9. Fn. tJ and Ip.15) (9.(t)lq. tip. Ip) (pI.t). I) (q. t) .10) (9.12) In the Heisenberg picture. Eq.1) that these two complete sets of eigenstates have the scalar product·· (q Ip) ~ II IF exp(iq.21) (9. tl ~ 1..9. ~ ~ (9. I. (9. of an eigenstate of P Thc factor 1/. (9.tl ~ JII ..) =b(p'  p).1.
x(p.t) . (9.23) to expand Iq.···"t'N.i babP". (9. Using Eq.380 9 PathIntegral Methods and also (q. the Qa(t)s in the Hamiltonian in Eq. it can equally well be written as the same function of Q(t) and p(t) H =H(Q.tlp.13) and (9.14) are similarity transformations.)P']' . t).tl.q.P).1. For instance. say t ' = r + dr and t = r. t'lq. r in' exp [iH(q"P)dt+i~)q. by using the commutation relations (9.P) ~ eiH'H(Q.tlexp( iH(Q(t).27) t This is only possible because with dr infinitesimal.t' ..iH' ~ H(Q(t).tlexp(iHdt)lq.1.t+dt I q. Now let's return to the more general case of a finite timeinterval.1. With this convention. with t'+l .1.1. we find that OUf system is in a definite state Iq.t) ~ j r. and H commutes with itself. This is easy when t f and t are infinitesimally close.1. with different constant coefficients.tk ~ dt ~ (t' I)/(N + 1) (9. tl.P)e. It will be convenient to adopt a standard form.1. (9.23) If..).t'lq. 2n ~exp(iq. by measurements at time I. OUf central dynamical problem is to calculate this scalar product. (9. t). To deal with the P(t). and find (q'."t'l in Peigenstates Ip. we use Eq."t'2.24) may be replaced t with their eigenvalues q~.tl ~ rr a .1) and (9.1. (9.1.P(t)jdt)lp.1.1.t). exp(iH dr) is linear in H. we have (q'.17).p.1. (q'. .tl ~ (q'. P(t»). in which all Qs appear to the left of all Ps. (9. but since (9.tlq. (9. we break up the timeinterval from t to t' into steps t.2) to move the Qs and P s past each other.t +dtlq.24) The Hamiltonian H is given as a function H(Q. given a term in the Hamiltonian of form PaQbPc.26) with each Pa integrated from OCJ to +00. we would rewrite it as PaQbPc = QbPaPc . t') is the scalar product (q'.25) This function can be written in various different forms."t'I.tl ~ jrrdp. with t < t' . then the probability amplitude for measurements at time r to give a state Iq'.. To calculate (q'.
q(r) and p(r). the argument of the exponential in Eq." .a x exp [it dt{ 2.H (q(t).p(T)) } d T.1.t'lqN. as well as over all p(r).1.a qo ~ . (9.t).9.e." k=! qk1.t) ~ f dql···dqN(q'..31) In the limit dr + 0 (i. rk) at each time rk: 381 (q'.1.~T) r. 271' dr. Eq.a)Pkl.1..b 2n (9.1. (9.32) l {2. Further.")Pkl. (9.a .TNl) ."] l lu If dp .qkl.1 The General PathIntegral Formula and sum over a complete set of states Iq.1.H(q"p..a .hm ..p(r) by fn r..O fn r.29) becomes just an integral over r ~ {L(q.Pk. t'lq.1..Pk_ddr}] q. because we integrate over all paths that take q(r) from q at r = t to q' at r = t'.34) This is called a path integral.qk.. t) ~ f ~ If dq.=1i"(T)P"(T)H(q(T). ' k.29) x exp where [i ~l {~(qk. .26). The .1./2n] (9.H(qk.t) ~ f qoftl = qa qat/'j = q~ ndq"(T)ndP.. Define smooth interpolating functions.33) Eq.a n dPkb .= n r. N + 00).. (ql.TNlqNl.1. this becomes (q'.a k.28) Inserting Eq. (9. (9.29) then becomes a constrained path integral (q'.a . (9.t'lq. can be put in a much more elegant form.t'lq.b 1i"(T)p"(T) . (9. Pa(rk) . we may define integrals over the functions q(r)..1._tld t} a ~ ~ {~?"(TklP'(Tk) ..H(q(T').P(t))}] (9.a dqa(r) dPh(T) .29).1. such that qa(rk) .30) Our result..b dqk.P(T')) } d ~ t + O(dT 2) (9..TN)(qN.tllq.
> . rl (p. we now find the general pathintegral formula . t.. if the time tA falls between"tk and 't"k+l.) II dp.382 9 PathIntegral Methods great advantage of writing matrix elements in this way is that. This result is only valid if the times are ordered.35).(. but with the operators arranged in order (from left to right) of decreasing time.TkJ.1. (9..Q(tA)) between (qk+l. r) ~ If) ?: exp [iH(q'..36) refers to the order of timearguments.3. with t < t').1. For instance.Q(I))lq.. x (q'. rl0 (P(r).. (9. but also the matrix elements between states (q'. 1'10A(P(IA). (9. t) II dq. (9. with t > tA > tB > .I (q'. for . It will be convenient to define these operators with (unlike H) all P s moved to the left and all Qs to the right.35) In order to calculate the matrix element of a product (GA(P(tA). 0.1.)P.Q(tB))'" of operators with tA > tn >"'. " 'h qu{t) = qu '.] !0(p.. Note that in Eq..1.1. we can insert the (I>operators between the appropriate states in Eq.1.1. we have (q'.36).Q(tA)) (!.1. (9.36). t) of timeordered products of general operators (D(P(t).q(tA))0 B(p(tB).. Q(t)] in Eg.. with tA.Q(t)).1. Q(r)) Iq. (9.q).tfh'" in arbitrary order (all between t and t' . t'l and Iq. The pathintegral formalism allows us to calculate not only transition probability amplitudes like (q'.28).28) each successive sum over states is at a later time. That is.'n(P(fB). Following the same steps as before. Iq. nothing on the righthand side of Eq. Q(tB)) . Hence if we are presented with a path integral like the righthand side of Eq. then insert (G'A(P(tA).26).. (9. tl exp ( iH (Q( r). the path integrals are easy to calculate when expanded in powers of the coupling constants in H.t) ~ III d dp. so this is only possible because of our assumption that fA > tn > . t'lq.37) However. (9.~t) 0A (P(tA). as shown in Section 9. (9.q(tB)) .Tk+ll and Iqk. Q(IA)) ~)B (P(tB). then this path integral will equal a matrix element like the lefthand side of Eq. Then by inserting any such operator ~)[P(t). q.a qa(l') = q. and use Eq.P)dt + i:S=(q. Pit) )dt) Ip..t +dt!0(P(t).
pl) ~ i 0 ~ ) exp (iI[q.:~.(r) p. cqa tA qa tA the argument of the exponential in Eg. We note that (for t < tA < t') (.38) vanishes if (9A(p. ) .p(r))}.(r) .40).1.p(r)) in Eg. Pa(r) in Eq.1.1.t) J II q"{t') = q~ dq.'" in arbitrary order. appearing in any of the other functions (T'H. (9. 0 q..p] . This simple rule applies only if the integration variables qa(tA). oqa r (91.'" }Iq. in Eg. q(tA)) e.1.1.38).40).p(r)) }] .1.38) are mere variables of integration.39) or Eq. the integral of such variational derivatives must vanish.1.39) or (9. (9. ( ( PAtA) where j] 1S exp (iI[q.h1r 0 A(P(tA).p]  t dr{~q.1 The General PathIntegral Formula tA. (9.1. we have 383 (q'.(tA) exp . (9. 0 .38) where T denotes the usual timeordered product. (p(tB). (9.t'IT{eA(p(tA).(r) II dP r) .1. Pu(tB). fS'c. (9.p(r)) ~.f(r)) ~ 0.Q(tB)). q" . t ) =q" "a x exp [i l dr {~q.( T. r DH~.q) is taken to be the lefthand side of either of the equations of motion (9.38). tB. (9. Suppose that one of the functions in Eq.(tA)Gp. GH(q(tA). and hence . etc.39) (9.(tA) exp .pJ). there is a limited sense in which path integrals do respect these equations of motion.Q(tA)).) Nevertheless. q(tA)).1. () (For this reason.(r)H(q(r).(r)p. ) .. happens to be the lefthand side of either Eg.(tAl.(r)p.9.1.H (q(r).38) is not constant in r.I [q.P(tA))) (. etc.38): GH(. and so with reasonable assumptions about the convergence of these integrals.1. As long as tA does not approach t or t'. 0B(P(tB). Pa(tA) are independent of any of the variables qa(tB). ( )+ DH(q(r). Hence the path integral (9. It should perhaps be stressed that the cnumber functions qa(r). and in particular are not constrained to obey the equations of motion of classical Hamiltonian dynamics q. say (I)A(p(tA).~(tA))) I[q. the integrations over qa(tA) and Pa(tA) are unconstrained.p] ~ 'Op. (9. q(tB)) .I [q. the Hamiltonian H(q(r).40) .
all Ps on the right.1. which of many possible quantum mechanical Hamiltonians H(Q. For these purposes it is better to use the pathintegral method to calculate amplitudes in Euclidean space. and we generally formulate theories with these constants left as arbitrary parameters anyway. The question is not an urgent one.38). is appropriate only if the measure is interpreted according to Eqs. where t is replaced with an imaginary quantity iX4.38) is a negative real quantity.p). Ifwe were to define a theory by the path integrals. There are a great many ways of interpreting the measure n dqaCr) IIdpb(') appearing in path integrals like (9.33).1. sa Under certain plausible assumptions.tB) or its derivatives. Our derivation has provided an answer: the quantum Hamiltonian is to be taken with all Qs on the left.38) for numerical calculations or as a source of rigorous theorems. In this way. These delta functions are the same as would be found in the operator formalism from time·derivatives of the step functions implicit in the definition of the timeordered product. It is difficult to use the general path integral in Eq.1.384 9 PathIntegral Methods only if we prohibit fA from approaching IB. the cnumber function H(q. . But it would be a mistake to give this prescription too much significance. etc. we only need to know the classical Hamiltonian. of putting all Qs to the left of all Ps.38). (9. Our prescription. the question would naturally arise.1.1.31)~ (9. quantum field theory may be fonnulated from the beginning in terms of Feynman amplitudes in Euclidean spacetime. tB. instead of jagged paths prOducing rapid oscillations of the integrand from one path to another. it is possible to reconstruct the Feynman amplitudes in Minkowskian spacetime from their Euclidean counterparts.34) and (9. all jagged paths are exponentially suppressed. because different prescriptions for ordering operators in the Hamiltonian just correspond to different choices of the constants that appear as coefficients of the various tenns in the Hamiltonian. 8b But we may as well stick to the Minkowski space formulation of the path integral if we are only going to use it to calculate Feynman amplitudes in perturbation theory. and the argument of the exponential in Eq. Other measures would lead to other prescriptions for operator ordering. Though we shall not go into it here.34) or (9. (9. When fA approaches. say.1. In evaluating the path integrals (9. as well as t or {'. the path integral will be found to involve a nonzero term proportional to O(tA .1.1. (9. P) (differing is the order of Qs and Ps) governed the quantum theory that corresponds to these path integrals. te.
9.m r. tltx. spin zcomponent (or helicity).t') = q~{x) X (!JA [p(tA).1) However.x. Eq.I qm(x. Experimentalists do not measure probability amplitudes for transitions between eigenstates (q'.". J (!JB [P(tB).1) by the 'wave functions' (ft. we can just as well do an unconstrained integral over qm(x.1 to a notation appropriate to quantum field theory..2) and then integrating over q~(x) and qm(X). To calculate a matrix element of a timeordered product (perhaps empty) between such states.2. (9.t)Pm(X. These are called 'in' and 'out' states.x. to remind us that fl[q(r). we can easily convert the general quantum mechanical results of Section 9. the probability amplitudes for transitions between states that at t + CIJ or t + +00 contain definite numbers of particles of various types. respectively." ·JI q. t) II dPmi:' t) r. but rather Smatrix elements. t'l and Iq. taken for convenience here to be 00 and +00. in) at any fixed times t and t ' . (X ) X.". tJ of the quantum field Q.p(t)] and (l. we need to multiply Eq.1. and replacing Qa(t) and PaUl with Qm(x.2): • We are now writing H and the (}'s with square brackets. But instead of constraining the path integral over qm(x. T)).2. where tx and f3 denote sets of particles characterized by the various particles' momenta.1) is not exactly what we want. outlq'. .q(tA)] (!JB [p(tB).2 Transition to the SMatrix 385 9.. and species.2.2 Transition to the SMatrix As already mentioned.2.t} and Pm{x.'[p(t). respectively. by letting the index a run over points x in space and over a spinandspecies index m.q(t)] arefunclionals of q". and set the arguments of the wave functions equal to the values given by Eq. ( ) = 'I. (9. in) and 1.t) II dqm(x. r) by the conditions (9. (9. Itx.8. t).Q(tAl] .m 'I. (9.q(tBl] . x exp [i t d t { j d3xqm(x. t') and (q.(x. and then perform an integral over the 'arguments' qm(X) and q~(x) of these wave functions.t' IT {(!JA [P(tAl.Q(tB)]. t) and Pm(x. out).38) then reads· (q'.t) at a fixed time l.t) H[q(tl. in field theory Eq. (9.P(tlJ}].2. T) (and also over Pm(x.
X.2..x.lo +00.tj.out) = 0.q(tA)] "'B[P(tB). The left·hand ~ide of Eq. in).3) is the derivative of this expression with re. (p)e'P' + H. rJ /2rr. matrix elements may be calculated as if there were no interactions. Incidentally.H [q(rt p(r)] x ({J. } la. for a Hamiltonian H[P(I}.2.I.v) ('1((£).pect to "'a.. (9. ..c.2. (6. Let's first consider the simplest and most important case. (6. outlq(+oc.n)IYAC.T)Pm(X. rJPm(x. x exp .q.4.3). <'I B [P(tB).]. (9.3). T) IT (dpm(x.). (9. etc. (9.3) a.4 that Smatrix elements may be easily calculated from the vacuum expectation values of timeordered products. outlq(+co).3).m dqm(x.386 9 PathIntegral Methods (p. T)/2rr) . and u~ing Eq. we have in effect <I>(x. ({J.. the vacuum.T)H[q(T).Q(t)] + LA.. in) ~ / IT dqm(x. "'n.m X <'IA[p(tA). (We saw in Section 6.lo OC.4) where ain and aout are the operators appearing in the coefficients of exp(ip' x . the Smatrix is given by Eq./ J'.5) "Il is only necessary to note that.f d]X<lm(X.oc+:. this result leads immediately" to Eq. outlC(.4. in).:.P(T)J}] (9.O".jEt) in the planewave expansion of the operator Qm(x. respectively.) We assume as usual that for t _ +:x.~oo (2rr)]/2'/ d]p (2E)1/2 [a . and t . . q".no.x.4. Q(tB)]. It is necessary now to consider how to calculate the wave functions appearing as the final pair of factors in Eq.} L>r 3 d x qm(x.I II [i {.3).. (6.n)IVAC.+co) (q(co).2.I 2. n}] = r.2.1II r.3) again then immediately gives the righthand side or F.in) =0. in).O". t) . r) . For instance.I)&A(X.t) at t . out IT { <'I A [P(tA). (9. r.m r.2. r} II(dPm(X. Q(tA)].+. (9. a theorem that we use repeatedly to relate sums of offshell Feynman graphs to matrix elements of Heisenbergpicture operators between exact energy eigenstates. for the real scalar field of a neutral spinless particle. which yields the righthand side of Eq.cola.2.q(tB)]'" x exp [iL: dT{. at to = 0.3) x (P. The 'in' and 'out' vacua may thus be defined by the conditions ain(p. aout(p.r d3x fA(X.".
.2.:. but we may note in passing that for x f. in other words.2.2. (9.4) read (9.8) The analogous ordinary differential equation has a wellknown Gaussian solution. (9. and we here use the conventional $ and n where pO rather than Q and P for a scalar field.2. .] (9. 6. +ooIVAC. if J d3 x eip·xC(x. we have _ a in ( p) oul l' /_+:>:.2.1 ~X exp (  l J d'x d' y $(x. we see that the functional differential equation for the vacuum wave functional is satisfied if for all ¢ o~ J d)xe.ip .t) • t~OC.K_ (mr) ) .y) ~ (2n)3 J d)peiP('Y(E(p).E(P)¢(X)] (9. the 'momentum' n(x. n..y) = In d 2n: 2 r dr (1.J p 2 + m2 ).2. t) 1.10) or. (9. 1"!!1 (2n:)· ei". [j 3 d y 6'(x. +:. and drop the unnecessary labels m.2.9.. 3/' ~ j' d3 xe ip·x X [v1oI>(X.2.12) (Recall that E(p) .t). Inverting the Fourier transforms and taking a linear combination of the resulting expressions. t) acts on wavefunctions in a ¢basis as the variational derivative ib/6¢(x.2.8). 1 (9.y .t)+ij2~n(X. $ may also be written in terms of a Hankel function of negative order $(x. so ill this basis the conditions (9..6) E = J p2 + m2.0 to the SMatrix 387 ~ <l>(x.t)] (9.11) The solution is easily found by inverting the Fourier transform 6'(x...9) with kernel g and constant A/' to be detennined. so let's try a Gaussian ansatz here: (¢(').13) .2 Transition nix. This is actually the most useful representation for the kernel C. y)¢(y) .y.7) As mentioned in Section 9.1. Substituting this in Eq. y)¢(X)¢(y)) . (9. y) = E(p)eip . _i(2n)3/2 J d3p (E/2)1/2[a ~~t (p)e ip ·x  RC.2.
2. f(+ro) + [(ro) ~ cu. [PB(tB). r) II(dn(x.16) We shall see in Section 9.c'.y) [<!>(x. ro)]) ~ lXI' exp ( je J 3 d' xd y i: (9.16) is to provide the . roIVAC. (9.388 9 PathIntegral Methods where r _ Ix . dqm(X.3) IS (VAC.T)] [.2.€ in the denominator of the scalar field propagator in momentum space.9) may be formally obtained from the normalization condition for the vacuum state. (9.14) C "'). in) ~ 1.. we have used the fact that for any reasonably smooth function f(r).15) Inserting Eq. . . We will not go into the corresponding details for fields of general spin.n(r)] + lie J i: (l)A [n(tA)' <!>(tA)] 3 x ¢(x.y)e""<!>(x.14) in Eq. in) "x X m B[n(tB).X (VAC.p(T)] i: P d m2(:. ro)<!>(y. (I). (9. <]l(tB)]. (9. <]l(tA)]. in calculating vacuum expectation values in the theory of a scalar field.2..} IVAC. T)Pm(X.r)<!>(y. but will simply state that in general (VAC.n.YI. q(tB)] . exp H[q(T). r) dT {J d d]xd]y 8(x.17) + ieterms}]. outl T { (I) A[PA(tA). (9. in) ~ lXI' J [.. r)/2n) >. r) n(x. (9.2. } IVAC.y)<!>(x.. Q(tA)]. dr8(X.2.4 that the whole effect of the last term in the argument of the exponential in Eq. the product of the last two factors in Eq.2. [p2 + m2 _ iE]l.r)e where E' is a positive infinitesimal. (l)B [IT(tB).' .r)<!>(y.+oc)<!>(y.q(t A)] dT { J 3 d x ~ im(X... (9.n. ¢(tB)J . r) (9.T)}] (92."1' exp ( I J d' xd'y8(x. out T { (I) A[IT(tA). r)] (l)A [p(tA). [i x (l)B [P(tB). but we will not need this result. According to Eq..2.2.9). The constant .2. Q(tB)]. outl<!>(ro).AI in Eq.. exp [i H[<!>(r). To obtain the final expression.3) now gives I ~ lXI' JII d<!>(x. +ro) (<!>( ro).+ro) +</>(x.'!?+ e i: dr[(r) e.
3. d)xd)yA". We do not need to work all this out here. This is a good place to mention that fieldindependent factors in Eq. (9.1. there is a large and important class of theories in which the integral over the 'momenta' can be done by just replacing them with the values dictated by the canonical fonnalism. + I: J .17) are all we need in order to be able to calculate Smatrix elements.7) to the vacuum state.3 Lagrangian Version of the PathIntegral Formula The integrand in the exponential in Eqs. However. are not important. We could go on and calculate matrix elements between mUltiparticle states. like the constant IA'~12. because as shown in Section 6.17) looks like the Lagrangian L associated with the Hamiltonian H. outiVAC.38) or (9. These can be calculated by applying the adjoints of annihilation operators such as (9.2.2.(y) d3 x B..P] ~ l I: J . This appearance is somewhat misleading because here the 'momenta' Pu(t) or Pll(x. outiVAC. t) are independent variables.17).2. and nonsingular. and any constant factors in the vacuum expectation values cancel in this ratio.. positive.9. symmetric.3 Lagrangian Version of the PathIntegral Formula 389 where the 'jf' terms' just have the effect of putting the correct if' in the denominators of all propagators. 9. these wave functionals turn out to be Hennite polynomials in the field times the vacuum Gaussian. just as for the harmonic oscillator. the vacuum expectation values (9. (9.in the language of field theory H[Q.(x) + C[Q] (9. (9.[Q] P. This is because such factors contribute also to the matrix element (VAC. The . In calculating the connected part of vacuum expectation values of timeordered products (or the Smatrix) we eliminate the contribution of disconnected vacuum fluctuation subgraphs by dividing by (VAC.y. in). not yet related to qu(t) or qll(x.4. t) or their derivatives.3).2. in).[Q] P. These are theories with a Hamiltonian that is quadratic in the 'momenta' . by inserting the appropriate 'wave functionals' in Eq. in which case the integrand in the exponential in the path integrals really is the Lagrangian.1) with a 'matrix' A that is real.2.(x) p.
Mx. T') 'e[q].7) (For a proof of this [annnla. see the Appendix to this chapter.6) S where ~ is the stationary point . (9.(.r'ym[q] Pn(x.5) J d.2)  ~/ d] x / dt PlIm [q] p.\ COO (IT dr.¢.(. . (9.p('l] } !L I . t) dictated by the canonical .2. s .390 9 PathIntegral Methods argument of the exponential in Eq.2) will be proportional to the exponential evaluated at the stationary point of its argument. For a finite number of real variables (I" this formula reads .17) by setting these variables at the stationary point of the quadr<ltic expression in the argument of the exponential. (9.i I: PlI. (9.3.i'&}' (9.3.) Thus the stationary 'point' Pn(x.Axn. =  U~ d] x p. in Eq.3.4) (9.. as long as the f!/'A.) Hence. .3. Sll'rxn[q] '&[q] Bxn[q(r)]  q"(x. r).)'m[q(r)]b(r ..t) where this vanishes is just the value of Pn(x. 'l  H [q('l.r'ym[q] . But the variational derivative of this quadratic is (The ie terms depend only on the qs.w"U. d3 x d\' dr dr' sf'rxn.3.2."r.3) (9. t)  where . Now. etc. for such a Hamiltonian we can evaluate the path integral over the ps in Eq. .3. C[q('l] . in general the integral of the exponential of a quadratic expression like (9.(x.(x.17) are independent of the ps.i2:>1I.2.ti'txn. '). r) Pm(Y. (9.) exp {liI: sr d. (9.3.17) is then quadratic in the ps: / d. eB .T').100 .If i'&} ~ (Det lid /2rr]) 1/2 exp {i l I: .
P(t)]) (9.3. out ~ 1%1' J IT {(!)A [Q(tAlj. This is not as restrictive as it may seem. (9.10) is the determinant of sl[q].(DB. For instance. In deriving Eq. with dr _ O. in which we divide by a vacuumvacuum amplitude proportional to the same constant factor.3. (9.Pit)]] '() up" x." x (9A[q(tA)] (DB [q(tB)] .3. <i(t)] + iderms} ] (9.. (9.10). .in) II dq"(X. one of which is d>(t).<ii t )] = J d]x (~<i"iX'T)Mx.3.9) and we can write Eq. by taking the difference of matrix elements in which this operator is replaced with <D(t + dr) and $(r).9.$n] " V(<D). The one serious remaining complication in Eq.10).3.. in a scalar field theory for which the canonical conjugate to $ is n = 4.3. we have already noted that overall constants make no contribution to the connected parts of vacuum expectation values. (9. x exp I: dT { L [q( T).3.17) as (VAC. eB [QUB)]"') I VAC. (9.t)  . we can simply differentiate Eq.t)R[qit).<D"aA<D" + J/'D. (9.17) is the ordinary Lagrangian L[qit).8) With p..3.3 Lagrangian Version of the Pathintegral Formula fonnula 391 q"(x.10) (We have combined the 1/2n factors in the integrals over the p" with the detenninant coming from Eq. Equivalently..X.2. This is the case for instance for the theory of a set of scalar fields $" with nonderivative coupling to each other and/or derivative coupling to external currents J".2. it was necessary to assume that the operators (!JA. the argument of the exponential in Eq. If d[q] is fieldindependent.6).r p={i (9. as long as t is not equal to any of the other timearguments of the operator in Eq.T) (Det[2insl[qJlr'/2 [i '.'" were independent of the canonical 'momenta'. _ [OR [qit).) This is the desired Lagrangian form of the pathintegral formula.10) with respect to t. (9. t) set equal to this value. it is possible to calculate the matrix element of a timeordered product of operators. then this is no problem. The Lagrangian density here is 2? =  L [ ~ai.(x. and then dividing by dr.
'<1>.my = jV<I>.' '.x') 61111.y. The determi .'<I>m (1+ y) . + . let us consider the socalled nonlinear amodel. using the relation Det. (1 + U(<I»):: Om + s1 nx . we may interpret the delta function in '#nx. (9.r d4 x··· .O)'] + J • d)x V(<I».) In general there is a nontrivial term that is linear in the nn.)' +J. but the coefficient of the quadratic term is a constant.'<1>. As a second example. Here sl is the fielddependent quantity [1 + U(lD(X))[~ b4(x  In cases of this sort. we note that Lx'" = 0.0. so that with the logarithm of a matrix defined now by its power series expansion In(1 + U) ~ U 2" +:3 '" U2 UJ To evaluate the trace. +J.1 .olbx.oO. (The 11/1 are taken to be real scalars. but complex scalars can be accommodated by separating them into real and imaginary parts.X'n' = J4(x . with Lagrangian density 5// =  tL ""' OAlDnoilDm [b nm + Unm(lD)]  V(lD) . However. and therefore is without effect.'d = exp Tr In sf .10) is here a field·independent constant. A straightfonvard calculation gives the Hamiltonian as H ~ Jd)x [ jO.y) = . . matters are not always so simple.3. U(<I»L + V(<I» J.' '.my as r5 4 (x . the determinant may be reexpressed as a contribution to the effective Lagrangian. + HJ.5 from one to several derivatively coupled scalars shows that this Lagrangian implies the Hamiltonian H ~ J d)x L [ 10.('. The factor (Det [4i7LW"[q]]) 1/2 in Eg. just the unit 'matrix': d XIl.392 9 PathIntegral Methods An obvious extension of the results of Section 7. By replacing the continuum of spacetime positions with a discrete lattice of points surrounded by separate regions of very small spacetime volume .
. we see that the Hamiltonian is again with the understanding that other terms must be added involving any fields that appear in ln ll . The constant of proportionality (which arises from the In 0 term) is fieldindependent and therefore of no present interest. Even where the factor (Det. let's consider the theory of a set of real vector fields. with Lagrangian density :£ ~  L [ l (a. 1 .l.  a.1)1/2 in the pathintegral formula (9. and would also be inconsistent with any symmetries of the Lagrangian under transformations of the scalar fields. where 'tr' is to be understood as the trace in an ordinary matrix sense.!iOItr In[l + U(¢(x))]. Here the coefficient of the quadratic term is .1 may be written as an ultraviolet divergent integral Wi ~ 64(x  xl ~ (2nl4/ rJ4 p .10) is fieldindependent.. As an example. We can regard this determinant as providing a correction to the effective Lagrangian density t!!E = . We shall not show this here. the Lagrangian in this formula may not be the same as the one with which we started.9. By a simple extension of the results of Section 7.5. U A + l:m n An.·An + J li·lA ] where the currents lull are either externally produced cnumber quantities or depend on other fields (in which case terms describing these other fields are to be added to the Lagrangian)..3 Lagrangian Version of the PathIntegral Formula 393 nant factor here is then DeL'" x exp [WI/ rJ'x If 1n[1 + U(<I>(X))]] .A". " ' 2 . but the extra terms in the Feynman diagrams for this theory contributed by t!!E could also have been derived in the canonical formalism by taking account of the equaltimecommutator terms in the propagator of timederivatives of the scalar field.3.A". 7 Ignoring this correction would lead to a spurious dependence of the Smatrix on the way that the scalar field is defined./') . The factor 0.a'A.) (a' A"' ..
In order to take account of the possible need to introduce auxiliary fields like A~. Ln I d3 x I1 n ...r) IVA[~(tA)] IVB[~'(t')] . the Lagrangian (9.: d' { L['1'( 'l. On the other hand.H .im~(A~f +J An J~A~ +A~V" nil] Il " This is still quadratic in nn with a fieldindependent (and somewhat simpler) coefficient of the quadratic term. suppose that we now integrate over the nn before integrating over the A~. ['¥B(tBI]"'}1 VAC. we may reintroduce the auxiliary field. y)  ~ VS jb 4(x 2m.17) is here replaced with H+ilH= Jd3xL[~n~+ i(VxAn)2+ im~An2 " . This can only introduce a fieldindependent overall factor. However.IV. An .n X exp [i 1.2.w)1/2 has no effect. (9.10) is far from obvious. Suppose we add to the Hamiltonian a term LlH =  ~ Lm~ " J 3x [A~ d . it is expressed entirely in terms of A and its spacetime derivatives.2J~r and integrate over the A~ as well as the A" and Un.394 9 PathIntegral Methods somewhat more complicated than in our first example: sfnix mjy = (jnm [bij(i4(X  .ilH: Il n = All +VAn . 0 With nn eliminated in this way.X. II A T. from now on we shall write the pathintegral formula after elimination of the canonical conjugates in terms of fields Ipr that include both canonical fields qn and auxiliary fields Cr : (VAC. with no dependence on any timecomponent AO. in ) oc J d~. All . "'('I] + i€terms}] (9311) . the Lorentz invariance of Eq. The Hamiltonian in the path integral (9.3. so that the factor (Det . To remedy this. ."(x.m.2V' nil + m.H . For this reason. y)] but it is fieldindependent.9) is here not the one with which we started. since /1H is a quadratic in AD ( with a fieldindependent coefficient in the term of second order in AO) whose stationary value vanishes.3. out IT {0 ['¥A(tAl] ..ilH is just the Lorentzinvariant Lagrangian with which we started.. so the integral over the nns can be done by just replacing nn with its value at the stationary point of the functional Ln I d3 x n .
T)) (9.T).4.in) ) AXB (VAC.4 PathIntegral Derivation of Feynman Rules We are now ready to use the pathintegral formalism to derive the Feynman rules in a wide class of theories.10).4.r:). (9.\"O(~(X..3.4.(xB)"'lIVAC.'I'(X. and summing over the indices on these coefficient functions.4 PathIntegral Derivation of Feynman Rules 395 it now being understood that L includes any terms arising from a possible fielddependent factor (DeLw)1/2 .4) by stripping off the final propagators associated with each field.9.outIVAC. plus a Lagrangian interaction density 21: L['I'(T). For the simpler theories whose Hamiltonian is quadratic in the TIs. Let us now suppose that the Lagrangian is the integral of a Lagrangian density. Eq. We will concentrate here on the vacuum expectation values of timeordered products of field operators (and their adjoints). replacing them with the coefficient functions that multiply creation or annihilation operators in the corresponding free fields. consisting of a quadratic term 20 which would be present in the absence of interactions.11) gives (9.p(X.in) (9.2) where I [~)] is the action I[~'] ~ l: dT{L[~'(T)'V'(T)] ~ +ic terms} (9.4. That is.4.. M{A(B  ( ~ (VAC.V'(T)] J dx 3 [. 9. X .4) +~l(ip(X. (9. the action (9. OJl1.r:))].outIT{'Pr.3.3) is . O.(xA).'Pr.3 ) with L now including any terms that may arise from a field·dependent determinant in Eq..1) from which Smatrix elements may be obtained (as shown in Section 6.
9) and the if' terms in 10 are given by Eq. With [o[~)] of the form (9. originally present in the numerator of Eg.rxYI(". exp(il[~.O/l1/J(X)) + Ill".(X)). We can therefore .10) so here fZ xx' = 8 • .]) ~ exp(i1o[~']) L '. f d4 x d4 x' L (. x!J a iJ f'!x'It  (5 (xx)+m 4 2 4( I .8. JJ dt I d3 xd 3 x' 8(x.f' :1JtX. The general integrals that we encounter in the numerator and the denominator of Eq. I 0 xx)if'&'(x. x. we will expand the exponential in powers of II.396 9 PathIntegral Methods I[~.4.5) 4 if' terms.1 J + ~ J.4.j(XA) etc. (9.2) are of the form where the field factors "Pt l (xd.4.] ~ 10["'] IO[~J] = Ill".4.11) (We are now dropping the factor eelrl in the if' term.. "P1 2(X2).x I )o(tt).) To deal with interactions.6) (9. (9.2). (9.t) (9. (1 1 ["'1) N=O Q"N II{ (9.12) N. as ji. (9. the unperturbed Lagrangian (9.(x).13) is of the same form as the integral evaluated in the Appendix to this chapter. (9.4.2.4. (9.16). etc.{I_~1 ~)t{x) "Pdx').4. for a real scalar field of mass m.8). the integral (9.8) For instance.4.7) Since Yo and the 'if' terms' are quadratic in the fields. with the discrete index s replaced with the pair of labels t.4. since it produces a correction of higher order in E.4.]. and then expand II in powers of the fields. (9.~.4. arise from Id"P] and/or from the field factors ~'(. we may always write 10 in the generalized quadratic form [o[~)] IS = 1 . d x 2'0 ( 1p(X).4.t)¢(x'.x') ¢(x.
!1>:2 " C~ = (211: )4 J d4P eir(xjx.4. so the propagator is d(x.y) = (211:)4 d4 p eip(xy) (p2 iEE(P))l .y = (211:)4 J J d 4p eip·(xYI (p2 +m2+m2_ iE E(p)) .4. where (9..I] paired fields .18) where 9. which give here .1 is the ordinary inverse of the matrix !Z'. (9.16) is then '"t l 1 2 (x 2. (9.1S).x3)bt l() . but in any case this factor cancels in the ratio (9..12) and (9.4.2) in their covariant form: we expand in the interaction 11. the iE terms have the effect of making the inverse welldefined for all real values of p..(. x) 3 ~ (2rr)4 J d4 p eip{~IX2) !tlI (p) (It2 ' (9.14) actually represents the contribution of graphs with unlimited numbers of single loops unattached to any other lines. (9.(XIX2) 397 ~ [Det (~~) r 2 I: pairing. (9.'.A. translation invariance will make f2 necessarily a function only of Xl .4.9.4. (9.15).j . for which the kernel !!lJ takes the form (9.4. (9. and then sum over the ways of pairing the fields in the II s with each other and with the fields WA(XA).4 ParhIntegral Derivation of Feynman Rules use Eqs. We can write this as a Fourier integral 9 x .4.:Q) = 4 b (XI .11). which can be written as a Fourier integral =IIXI.4. (9.4. First consider a massive scalar field.vi! P (9..~ of pairs II [.15) as an integral equation L J"X2 2?tjXjl2X2d(2/)(X2.A. etc.4.17) The solution of Eq. with the contribution of each pairing being given by the spacetime integral of the product of the coefficients of the fields in 11 [tp] and the product of the 'propagators' id.LJtlt2 () . We have thus reduced the problem of calculating the propagator to that of taking the inverse of a finite matrix.4.2).16) In the absence of external fields.) It remains to calculate the propagators (9. .X2.g. We interpret Eg.14) Helds This just amounts to the coordinatespace Fcynman rules for calculating the numerator of Eq.4. As we will see.15) (The factor [Det(j9j211:)]1/2 in Eg.
<TY = = _ ['1 PU oXI'C'Y/l a:  a ~~ .p] (9. The unperturbed Lagrangian is Yo =  ~(CtlAv ~ i\Att)(C!IA" . They are important in the denominator in defining how the integrand is to be treated near the mass shell.4. E( ) pmIE p 2 d'p e"('Y) [ '1PIT + p p ] P . the pairing of (}ji¢ with i\.iE). since both are just positive infinitesimals.PpPa • " .8). These noncovariant terms were previously needed to cancel noncovariant terms in the interaction Hamiltonian.m +. The factor arising from the pairing of a field derivative O/l1JJt(X) with any other field lPm(Y) (perhaps itself a field derivative) is [n d1JlAxl] (}/I1j)((X) tpm(Y) eil[~)l (O.¢ gives a momentum space propagator kjik v /(k 2 + m2 .398 9 PathIntegral Methods We recognize this as the same scalar propagator previously obtained by operator methods.. Also.(x.t [ ] I n d1p{(x) x. for a real scalar field. I ei1[.4. Theories with derivative coupling are equally simple.IAII . .' [ ap eip·(H) [ 2 . except that the noncovariant terms proportional to b(xo .t o ~ ox' ('Pi (x) 'Pm(Y)) .IE t erms] We will not bother to show it here. For instance.x oyP + m2 '1 Pu] b 4 (x . consider a real massive vector field. (The difference between e: and E E(p) is immaterial. but the '+ if' terms' here take the simple form ie: E(P)l1puo The vector field propagator is then given by simply inverting the 4x4 matrix in the integrand Jip.y) + ie: terms (2 11: ) . and no such cancellation is needed. vertices in the theory of a scalar field with derivative couplings to other fields may be read ofr from the Lagrangian. We can again write Io[tp] in the form (9.) For a second example. but the vertex contributions in the Feynman rules are now obtained directly by inspection of the covariant Lagrangian..19) Such propagators have no noncovariant pieces. as we saw in the previous section.IT m• (Terms proportional to E are dropped in the numerator.y) = (2n)4 J + 2 .aVA/I)  1m2 A. p2 = _m 2 . YfpaP + m2 11. with kernel ::2 px .yO) are now absent. and are separately covariant.'Pi(x) V'm(Y)) ~ ..) This is the same as the propagator derived by operator methods.
we can take where If) and (gl are any kets and bras for which these expressions do not vanish. (9.4) and a 'bra' state (01 annihilated (from the right) hy all Pa : (01 P.1) (9.5. (9. (9. we have Q~ = P.5 Path Integrals for Fermions We now turn to the problem of extending the pathintegral formalism to cover theories containing fermions as well as bosons. Ph} = i bah.5) are unique up to constant factors. 9 As before.5. They are not in general unique. and will assume that the states satisfying Eqs. but for simplicity we will limit ourselves here to the case where the only degrees of freedom are those described by the fermionic operators Qa and Pa. Note that for any given a. as we did for bosons. with 'coordinates' Qa and canonical conjugate 'momenta' Pa.5. (9. we will here derive the pathintegral formalism for fermions directly from the principles of quantum mechanics. (9.5. Instead.5) For instance.5.3).9. which we assume not to be the case.5. because there may be other bosonic degrees of freedom that distinguish the various possible 10/ and (01.3) It follows that there will always be a 'kef state 10) annihilated by all Qa: (9. which we choose so that (010)~1. with the justification that this gives the 'right' Feynman rules.) These states satisfy Eqs.5. = O.5 Path Integrals for Fermions 399 9. we will start with a general quantum mechanical system.) Later we will replace the discrete index a with a spatial position x and a field index m.6) .5.5. by analogy with the bosonic case.5) by virtue of Eq.2) {Q"Qb} ~ {P"Pb} ~O. We wish first to construct a complete basis for the states on which the Qs and Ps act. ~ 0 (9. or in other words Heisenbergpicture operators at time t = 0.4) and (9.4) and (9. but now satisfying anticommutation rather than commutation relations: { Qa.5. It would be easy to proceed in a purely formal way. (9. (They cannot vanish for all If) and (g I unless the operators fIa Qa and fIa Pa vanish. (These are Schrodingerpicture operators.5.
5. there are fermionic operators (such as the 'ghost' fields to be introduced in Volume II) for which Pa is unrelated to the adjoint of Qu.(01 . the result of acting on these states with any operator function of the P sand Qs can be written as a linear combination of the same set of states. in the Dirac theory Qa is not Hermitian.400 9 PathIntegral Methods (Note that this nonnalization convention could not be imposed if we had defined (01 as the lefteigenstate of the Qa with eigenvalue zero.12) Using Eqs.Ph .b. 10) (9. In what follows we will not need to assume anything about the adjoints of Qu or Pa. I a.5. 1.) As we saw in Section 7. / (9.. because in this case (OI{QI. we would like to be able to rewrite sums over intermediate states like (9.b. which with Eq.c. h.) P.5. In particular. In deriving the Feynman rules.5.4){9.1).5. (9." ../ .d.5."') =0. " we can always rewrite the state (possibly changing its sign) so that a is the first of these indices.13) where {. (9. . (iQh)( iQ. d = b. if a is equal to one of the indices in the sequence. b.10) (9.7) with any number of different Ps acting on 10). "}f{a. .) . consisting of (01 and the states (also antisymmetric in the indices) (a.h. in which case (01 can be regarded as simply the adjoint of 10).5./.5.. d. However.5. That is. b. then Qa be"') ~ " 1 °." '/. ) 1 a. .8) (9.l. C.. it is not possible for these operators to have . (9.c. but instead has an adjoint ~iPa. Ph" '10) (9. } here denotes the set of indices within the brackets. irrespective of order. we may define a complete dual basis.5. h. (9.6) and the anticommutation relation (9.11) Pa!a. if an index a is unequal to any of the indices appearing in I h.) ~ (01" (iQd)( iQ. etc. However." ) P. in which case we have Ql abc"'\~ilbc"'\ ' a". A complete basis for the states of this system is provided by 10) and the states (antisymmetric in indices a. .7) as integrals over eigenstates of the Qa or the Pa. Similarly.5.5.1) would imply that (010) ~ 0.' .5. On the other hand..Pb}IO) would vanish. or about any relation between 10) and (01.  _{ ° if {c. we see that the scalar products of these states take the values (e..9) Plbc"')~labc"'\ a" ."'} 1 if c = a.
(9. .5.5.q. (9. satisfying Eq.] exp (i L. (To verify Eq.) exp ( i.14): with the exponential defined as usual by its power series expansion. By the same argument as for Eq.5.Pb9b) 10) boF' q. we see that .] [1 . Suppose we try to find a state Iq) that satisfies (for all a) (9. Pbqb) 10) boF' ~ [i{Q"P.) ~ (IJ Q.2) we see that (9. We can now construct eigenstates Iq. We will require further that where q and q' denote any two 'values' of these variables.Pbqb) 10) boF' ~O as required by Eq.5. (9.15). as (ql . (9.  9..14).p..19) where rIa is the product in whatever order we take as standard.q.q. but which still satisfy the anticommutation relations (9. which act like cnumbers as far as the physical Hilbert space is concerned.) .5.= q. (9.5 Path Integrals for Fermions 401 eigenvalues (other than zero) in the usual sense.14).q. use the fact that all Paqa commute with one another and have zero square.5.] 19) ~ [Q.18) (9.14) From Eq.9. ~ [Q. (9.= p.] exp (iL.] exp(iP.14).) (01 exp ( +i.5. so that [Q.) We can also define lefteigenstates (ql (not the adjoints of Iq}).iP.5.}q.(01 (IJ Q. However.) exp (iL.q.5.15) which is impossible for ordinary numbers.5. nothing can stop us from introducing an algebra of 'variables' (known as Grassmann variables) qa.
.P.5.24) ~ (01 exp ( i ~ P. which we move to the right Ollt of the scalar product.5.) (I] ~ (I] ~ and so eXP(iq"P.lp) .p. (9. so We shall see that Eq.  qb)) 10) (11(1 + iPh(q. with scalar product (now derived by moving the Ps to the left) (pIp) ~ IlIp~ (9. ~ ~ p. .21) (9. and IP)~exP(i~Q.20) Moving each Qa to the right (starting with the rightmost) yields factors p(q~ . we can construct right.20) plays the role of a delta function in integrals over the qs.5.) (I] p") 10) p.) .5.)) (ql eXP(iq. In the same way.5.402 9 PathIntegral Methods These eigenstates have the scalar product (q'lq) ~ (01 ~ (01 (I] (I] Q.5.5.qb))) 10) . (9.p.qa).) (i~?h(q.q.) .25) The scalar products of these two sorts of eigenstate with each other are (qlp) ~ (ql exp ( i~ Q.) exp Q.and lefteigenstates of the Pa : P.Q.p.) 10) 10) (qlp) ~ XN exv( .5.i~ q.23) (9. (9. (pI (9. " p. (9.22) where the Pa arc like qa anticommuting cnumbers (taken for convenience to anticommute with the qu and all fermionic operators as well as each other).)) (01 (I] p") (I] (I] Q.).) ~ XN eXv(i~ .lp) (pI P. (pip.26) .p.) (11 Ph) 10) .
10 that is designed to pick out the coefficients of such products of anticommuting cnumbers. For any set of such variables ~Il (either ps or qs or both together)..5 Path Integrals for Fermions 403 where XN is a phase that depends only on the number N of Qa operators: XN (01 (!JQ. the integral is likewise antisymmetric under the interchange of any two d¢s. .) (!JP.5. it will be very convenient to introduce a sort of integration over fermionic variables. (9. . (9. Since this product is antisymmetric under the interchange of any two ~ s. Therefore we can write any state If) in the form If) ~ folq)o + Lf.29) JWd~") f(O=c with the tilde in Eq.··· in an expansion of Iq) in a sum of products of qs.b.. + Lf"lq)'b + . in which case it is important to standardize the definition of c by moving all ¢s to the left of c before integrating over them.) From the definitions (9.5. the coefficient c may itself depend on other unintegrated cnumber variables that anticommute with the ¢ s over which we integrate. as we have done in Eq.···) in the general basis is (up to a phase) just the coefficient of the product qaq.5.29) indicating that we use the convenient convention that the differentials are written in an order opposite to that of the product of integration variables in Eq.30) Also. known as Berezin integration.28).5.28) and the integral over the ~s is defined simply by (9.. we see that the state la. we also find It is easy to see that the states Iq) are in a sense a complete set (and so also are the Ip).28).··· on Iq) denotes the coefficient of qa%··· in Iq).27) Somewhat more simply.17).5.5.9.5. so these 'differentials' effectively anticommute (9. In summing over states.)IO)~iN(l)N(NII/2 (9. and a subscript a. (9. b.lq). the most general function !(O (either a cnumber or a statevector like Iq)) can be put in the form f(0 ~ (!J ~") c + terms with fewer ~ factors (9.5. a w# where the fs are numerical coefficients.
28) and (9. (9. This definition of integration shares some other properties with multiple integrals (from 00 to +co) over ordinary real variables. we could have first defined the integral over a single anticommuting cnumber Sl. (It was in order to obtain this result without extra sign factors that we took the product of differentials in Eq.~2) ~ ~2ct2 + 'I.28).) Indeed.404 9 Path~Integral Methods For instance.32) where a(¢/) is any function (including a constant) of any anticommuting c·numbers s~ over which we are not integrating.~2) ~ ~1~2CI2 + ~I 'I + ~2'2 + d because the squares and all higher powers of Sl and S2 vanish. J d~2 f(~\'~2) ~ ~I '\2 + '2.f d~lf(~t. but there are significant differences.5. (9.~2.f d6 d~1 f(~I."') ~ '(~2. Note that the multiple integral is the same as a repeated integral: J d~2 d~1 [(~1.~2) ~ J d¢. Obviously. (9. in the sense that J (1] d~") [Jm+g(~)] ~ / (1] d~") [(0+ (1] d~") g(~) J (9531) and also J(1] d~") [f(~)a(n] ~ [J (1] d~") f(()] a(~'). (9. This function has the integrals J d~1 f(~I. Berezin integration is linear. and its integral over Sl is defined as . ~2.29) to be in the opposite order to the product of variables in Eq.''') ~ b(~2''') = + ~1'(~2"') 0). ~2) ~ ct2 . However. Repeating this process leads to the same multiple integral as defined by Eqs.5. The most general function of anticommuting cnumbers is linear in anyone of them (because a f(~I. linearity with . and then defined multiple integrals in the usual way by iteration.5.[J d~1 f((1'~2)]' a result that can easily be extended to integrals over any number of fermionic variables. . the most general function of a pair of anticommuting cnumbers Sl and S2 takes the form f(~I."') .29).5.5.
It follows that the coefficient of TIn (~ in any function .e.m. On the other hand. The product of the new variables is IH. (9. m (9.)(~ + (~(d(.) ~ 0 (9. as long as we take the ~~ in the same order. consider a change of variables (n + f with fewer than (~ = L Y'nm~m .5.) m) ~ aWl J(IJ de.) n I'TIlmr· n But TIn (171" here is just the same as the product (in the original order) TIn (n.) f(0· (9535) Another similarity with ordinary integration is that.37) where Y' is an arbitrary non~singular matrix of ordinary numbers. ~ L (II Y.. since shifting ~ by a constant only affects the terms in the total number of ~·variables.5 Path Integrals for Fermions 405 respect to left~multiplication is not so obvious. we have and so (9.33) reads more simply Ja(tl (IJ de. for an arbitrary anticommuting c~number ~' independent of~. then since ~~ is assumed to anticommute with all ~n.33) It is therefore very convenient (though not strictly necessary) to take the differentials d~n to anticommute with all anticommuting variables (including the ~n): (d(. If we are integrating over v variables. except for a sign dmJ which is +1 or I according to whether the permutation n 10 mn is an even or odd permutation of the original order: This applies whatever order we take for the ~n.534) in which case Eq.9.5.5.
q')) and expand in powers of q .39) with some bra (q'l (with q' any fixed Qeigenvalue).) To find the function f(q) for a given statevector Ifl. in the state Iq) by integrating the product of Iq) with an qa with a not equal to b. All terms beyond the lowest order vanish when multiplied with the product .17) used to define Iq) involves only even numbers of fermionic quantities. We can now use this definition of integration to write the completeness condition as a formula for an integral over eigenvalues.1 times the coefficient of TIn en. where N is now the total number of qu variables. As already mentioned. qaqb. a statement we write as (9. We shall use Eq.38) and the linearity properties (9. and (9..q'.5.d. in the Qeigenstate Iq).5.35) and (9.c. so We can rewrite f(q) as f(q' + (q .5.32). except that (DeL</') appears to the power 1 instead of + 1. h).5.31). According to the definition of integration here. we can pick out the coefficient of any product q/lqcqd . (9.5. and these states are (up to a phase) the coefficients of the products 1. take the scalar product of Eq.39) (We can move Iq) to the left of the differentials without any sign changes because the exponential in Eq. by choosing a function f(q) as a suitable sum of such products of qs. etc..5. etc.5.35) later to evaluate the integrals encountered in deriving the Feynman rules for theories with fermions. any state If) can be expanded in a series of the states 10).38) This is the usual rule for changing variables of integration.5.20).···. (9.5. qu. this is Moving every factor (qa q~) to the right past every differential dql> yields a sign factor (. Thus.)N' = (_)N. According to Eqs. (9. jal. (9.406 9 PathIntegral Methods f(~) is just (Det YT.5. la. (9. we can write any state If) as an integral: (9.
we can also show th'Jt (9.32).5. we now have The term in the integrand proportional to TI qa has coefficient f(q'). Using Eq. Inserting this back in Eq.5 Path Integrals for Fermions 407 (!J(q. (9. Q) with all P s to the left of all Qs. we define timedependent operators Q.tl =exp(iHt)lq) . exp(iHt) (9. .5. = (qlexp(iHt) .40) which partly justifies our earlier remark that Eq. (9.41) or as an operator equation In exactly the same way. As before.t) (q. (9.q)dT).42) We are now in a position to calculate transition matrix elements.'f(q') . Ip. (p.5.5.5.) ~ (q'lexp(iHd')lq)· Now insert Eq. (9.43) (9. (9.5.(t) ..5.lq.42) to the left of the operator exp(iHdr). (9.544) PaUl = exp(iHt)Pa exp(iHt) and their right. so that (for dr infinitesimal) (plexp( iH(P.t!.5.exp(iHt)lp) . so according to our definition of integration (q'lf) = ()·II.45) (9.)) f(q') .9.)) f(q) ~ (!J(q.  q.q.exp(iHt)Q.!) .5.T+d.39) gives our completeness relation If) ~ (_)N jlq) (~dqb) (qlf) .(plexp(iHt).5.Q)d')lq) ~ (plq)exp( iH(P.20) plays the role of a delta function for integrals over the qs. (9.46) The scalar product between qeigenstates defined at infinitesimally close times is then (q'. . It is convenient here to define the Hamiltonian operator H(P.and lefteigenstatcs Iq.
We then find (q'. Up to this point we have kept track of the overall phase factor (i)N XN .(q. t) /" ([a(l)=([a.B (P(tB).5. However.48) The symbol T here denotes the ordinary product if the times are in the order originally assumed.q(r))}] (9.r+drlq.(T1dP'<T)) X (1. To calculate the matrix element (q'. . (9. > t). t) of a product of operators (with (' > fA > tB > .5. the righthand side is totally symmetric in the mA.47) The rest of the derivation follows the same lines as in Section 9.27).47) to evaluate the resulting matrix elements (with (!}A... q) to either side of the matrix element without any sign changes because each term in the Hamiltonian is assumed to contain an even number of fermionic operators. .m B . and then introduce functions qa(t) and PaCt) that interpolate between the values of qa and pa at each step. we find (q'.q)dT).) This gives (q'. and noting that the products Paqa and Paq~ commute with all anticommuting cnumbers. with an overall minus sign if timeordering the operators involves an odd number of permutations of fermionic operators. etc. j Iq.r + drlq. Q(tA I).5. (_i)N XN (1.26) and (9.  q.41).{!}B.A (P(tA). divide the timeinterval from t to t f into a large number of very close time steps.q(tB))'" x eXP[i ldr{~p'(T)q"(r)H(p(rJ. inserted where appropriate). t'IQA(P(tA)) QB(P(tB). Q(tB))" ·Iq."'" (except for minus signs where anticommuting cnumbers are interchanged) so this formula holds for general times (between t and t'). tA > tB > """. move all differentials to the left (this introduces no sign changes. provided T is interpreted as the timeordered product.) exp [i ~p.408 9 Path~Integral Methods (We could move the cnumber H(p. q(tA)) mB(p(tB).5. because at each step we have an equal number of dps and dqs).) (plexp(iHdrJlq) ~ j(q'IP) (~dP') (plq)exp( iH(p. Using Eqs. use Eq.1.r) ~ j(q'IP) (l)dP. Q(tB J).5.5. at each time step insert the completeness relation (9.) . (9.qJdrj (9..(ja(l )=q" (U dq.iH(p.  t'IT {'" A(P(tA I.r) ~ j (~i dP.
(9.m exp 3 x Pm(X.49) as it stands is the Lagrangian L. so that the quantity I d3 x Lm Pm qrrr .17): (VAC. QUAI]. r)qm(x. The transition to quantum field theory follows along the same lines as described for bosonic fields in Section 9.(YI Ho [q(T). (9.p(T)] + if' terms} ~L m" Jd"x d"y (9. Each term in the Hamiltonian for a fermionic field that carries a non~vanishing quantum number (like the electron field in quantum electrodynamics) generally contains an equal number of ps (proportional to qt) and qs. f!)B. since it only affects the constant phase in the path integral. in the standard model of electroweak interactions (and in other theories.. . [i I: d'{ J ~ d 'I .5.H in Eq..491 H[q('l.5. The interaction Hamiltonian V H Ho is a sum of products of equal numbers of fennionic qs and ps (with coefficients that may depend on bosonic fields) so when we expand Eq. Pm(xlq.p('l] + i"erms} ] where the proportionality constant is the same for all operators (C'A. the freeparticle term Ho in the Hamiltonian is bilinear in p and q. qUBI] . T. such as the older Fermi theory of beta decay) the canonical conjugates Pm are auxiliary fields unrelated to the qm. In particular. and hence will not be of importance to us...2. Q(tBI]. } IVAC. we have replaced each discrete index like a with a space position x and a field index m. The vacuum expectation value of a time~ordered product of operators is given by a fonnula just like Eq.X.. and the 'if' terms' again arise from the wave function of the vacuum.501 with 9 some numerical 'matrix'. Indeed.5.m T. A major difference between the fermionic and bosonic cases is that here we will not want to integrate out the ps before the qs. in) oc X f [IT dqm(x"I] [IT dPm(x..)] eA[p(tAI. etc. so that l: J dT{ 3 dx ~ Pm(x. As before.49) in powers of the V we encounter a sum of fermionic integrals = .5 Path Integrals for Fermions 409 But in fact these phases contribute only to the vacuumvacuum transition amplitude. (9.. out IT { e A[PUAI.2. <im(X. e B[P(tB).X.9. r) '0=".q(tAI] eB[PUB I.) (9. We are also dropping the tilde on the product of differentials. and the Lagrangian is linear in the ifm.5..
g) ~ exp (iI: .. To calculate this sort of integral.53) The integral is a constant (i. first consider a generating function for all these integrals: J(f. (9. Y2 .nyPm(x)qn(Y) .(y) J a'x (9. 4 4 (9. independent of the functions f and g) which can be shown using Eg. We shift to new variables of integration p.r)] x.:! d xd y::2 mx. xN.5..(y)) .i I:/ . .r.e.. m ..(y)q..".(y)).r)] x.(y)) .(x) J a'y g. Of more importance to us is the first factor. one such term for each possible set of vertices in the Feynman diagram.iI: Ja'y g.m x exp ( iI: Jd4xd4y2l=".)".(y) + I: m i2. yI.36).g) • J[II dqm(x.m2""/lNmN(xj.Pm(X)q.m x exp ( .5.x.(y)(9 1 )".52) where fm(x) and gn(Y) are arbitrary anticommuting cnumber functions.51) r. g. q..(y)fm(x)) x J[II dq~(x.m x exp( iI: Ja'x Pm(x)fm(x) . we then find f(j'.r)dPm(x.38) to be proportional to Det @. r)] (9. with coefficients contributed by each vertex given by i times the coefficient of the product of fields in the corresponding term in the interaction. YN) = J[II dqm(x.r.m. f mix) . Expanding this factor in powers .T)]qmJxdPn1(ydqm2(X2)Pn2(Y2) "'qmN(XN)PnN(YN) r..410 9 PathInte~ral Methods of the form .~ J d xd y 4 4 UT 1)".5. (9..m.')dp~(x.5.m x [II dpm(X. ~ Pm(x) + I: ." . 4 4 p~(x) q. d xd y 9=".5. Xl. ~.f nlfflln.. ~ q. 1 Using the translation invariance condition (9.
not counting as different pairings that only differ in the order of the pairs.(Y)}.57) l1 m iel) eik.5.IlY = J d4k (2JT)4 ([iYP k l1 +m iejl [·. let us calculate the field dependence of the .k(Xy) .) The propagator is then (.5.+ m. (9. Comparing this with Eq. TJ<im(X. we find here £?mx.58) just as we found in the operator formalism.52). and the factors (.[ip(YbOj. we sum over the N! permutations either of the ps or the qs.j.01 t i l e.IIY = [rO (i'lli.5. T)  HO[q(T).ji'. The sign factor Jpairing is +1 if this permutation is even. (2JT) (Xli mil ()4(x  y) (9. (9. ~ Pm(X. not T{~'m(x).5 Path Integrals for Fermions 411 of gf and comparing with the direct expansion of Eq. This sign factor and sum over pairings are just the same as we encoun~ tered in our earlier derivation of the Feynman rules. the ie term here arises in much the same way as for the scalar field in Section 9.54) with a proportionality constant that is independent of the x. or n.QiI )mX.ny playing the role of the propagator for the pairing of qm(x) with Pn(Y). (9. with the sum over pairings corresponding to the sum over ways of connecting the lines asso· ciated with vertices in the Feynman diagrams.50) .9.Qil)mx. In other words.2..5.(xy). and also independent of the number of these variables. As one example of a problem that is easier to solve by path·integral than by operator methods. I if it is odd. we see that (9.5.55) J d4 x lfl(X) [yll U + m] 1p(x) .56) with m a fourvalued Dirac index.5. 11 where in the usual notation the canonical variables here are (9.5. The sum is over aU different ways of pairing ps with qs.y.ie)] ~ I d'\ (yO[i}'1' k + .P(T)]} (9. In the Dirac formalism for spin ~' the freeparticle action is I: =  dT{ J 3 d .m. The extra factor r'O arises because this propagator is the vacuum expectation value of T{'Pm(x). mil (Though we shall not work it out in detail.
x. in). the vacuum persistence amplitude in the presence of this external field is (VAC.." ~ (9. (9.5. (9. (9. .5.T)] r .5. (9.if'] q} 1 (9.x. 4 4 r. 64 (x  and expand in powers of ~[l]. we change the variables of integration qn(x) to L Jd4y Jf[r]=. According to Eq.y q. in).5.5.49).W[rn ) ( n 1 n . Eq.T)] [IT dPm(X.x.60) with a proportionality constant that is independent of r(x).T)] r. x X J[IT ! dqm(X.i L Jd x d y Pm(x) q. in). We write this as (VAC. (9.5.67) .5.5.66) (yO qx)lm. outiVAC.y } .5.o q~(x) (9.. 1)'+1 Tr (01.65) (9. x X J[IT dqm(x. To recover the results of perturbation theory.5.(y) .. outiVAC.412 9 PathIntegral Methods VaCUUffiloVaCUUm amplitude for a Dirac field that interacts only with an external field.(y) %[r]m.ny = (}.m exp {  if trx pT .liD/ + m+ r .. x Det %[r] .64) = '0 + O§[r] .ob.59) where r(x) is an xdependent matrix representing the interaction of the fermion with the external field.t)] [IT dPm(X.5. y). outiVAC. (9. let us write %[r] O§[r]=..63) The remaining integral is now r independent.x.5.m exp { . outiVAC. so the whole dependence of the vacuum persistence amplitude is contained in the determinant arising according to Eq. Take the Lagrangian as :£ = l/)[yJl V!( + m + r]V'.m (9.62) To evaluate this.64) gives then (VAC. x Det ('0[1 +'0IO§[r]]) = [Det01] exp ~( ?.m .61) where "" %[r]".x.38) from the change of variables: (VAC. (9. in).
"'] = HM+ J d3x[!ni. 9.2 + HVXA)2_A. This will be pursued further in Volume II.5. lin is the usual combinatoric factor associated with such loops (see Section 6. More to the point.9. In Chapter 8 we found that in Coulomb gauge. satisfying the . without ever having to think about the details of Feynman diagrams.~'[r]. the trace of the product of n factors of _gl'§[l] thus corresponds to a loop with n vertices connected by n internal lines.ni. But as we shall now see.6.6 PathIntegral Formulation of Quantum Electrodynamics The pathintegral approach to quantum field theory really comes into its own when applied to gauge theories of massless spin one particles. the sign factor is (_l)n+l rather than (_l)n because an extra minus sign is associated with fermion loops. (9.1 and ~iW[r].6 PathIntegral Formulation C!f Quantum Electrodynamics 413 This is just what we should expect from the Feynman rules: the contributions from internal lines and vertices in this theory are i9J. and the sum over n appears as the argument of an exponential because the vacuum persistence amplitude receives contributions from graphs with any number of disconnected loops. and that the purely timelike terms would just cancel the Coulomb term in the Hamiltonian. To give a real justification of this result using the methods of Chapter 8 would involve us in a complicated analysis of Feynman diagrams. The derivation of the Feynman rules for quantum electrodynamics in the previous chapter involved a fair amount of handwaving.J]+VcOU ] ' (9. it represents the contribution of any number of fermion loops that carry no vertices. the Hamiltonian for the interaction of photons with charged particles takes the form H[A.64) allows us to derive nonperturbative results by using topological theorems to derive information about the eigenvalues of kernels like . so that the effective photon propagator could be taken as 111'" I q 2.is the solenoidal part of its canonical conjugate. the pathintegral approach yields the desired form of the photon propagator. The rindependent factor Det § is less easy to derive from the Feynman rules.1). such as quantum electrodynamics.6. in arguing that the terms in the photon propagator L~r(q) proportional to qll or q" could be dropped.2) while ni. a formula like Eq. subject to the Coulomb gauge condition (9.1) Here A is the vector potential.
6.2) and (9. The argument of the exponential in Eq."2 and "1. (9.J+YM] if dtVCOU1 } (965) x[\lJ(l'a(x))] [1]6(1' "(X))] . hut which unlike n commutes with all canonical matter variables. where ~)t(x) are generic matter fields. the integral over 1t can be done (up to a constant factor) by setting 1t equal to the stationary point of the argument of the exponential.2) and (9. (9.5) to enforce the constraints (9.5) only by a factor Detc~. say.3)... with fieldindependent coefficients in the term of second order in It.n .A. Thus. we are assuming that HM is local and either linear in the matter Jr5 (as in spinor electrodynamics) or quadratic with fieldindependent coefficients (as in scalar electrodynamics).5) is evidently quadratic in the independent components of 1t (say. then we should insert the delta functions "L. is not strictly accurate. this differs from the product of della functions in Fq. (9. which although infimte is fieldindependent and hen~. 7[1 and 11"2).9). according to Eq.6.3). II J (al(X) + C:Jl (Cl adx) + ('2a2(X)) ) J ( "3{X) + 6J 1(CI "dx) + C2"2(X)) ) . (9. (9. •• Thi.4.6. We have inserted delta functions" in Eq.3) Also.1l2.6. (9. "1. However.6. In wntmg Eq. HM is the matter Hamiltonian and VCoul is the Coulomb energy Just as for any other Hamiltonian system. (9.414 9 PathIntegral Methods same constraint V'TIl= o. cancels in ratios like (9.6. • Note that nix) is the interpolating cnumocr field for the quantum operator whose commutation relations with each other and with A are the same as those of n.5) in terms of a matter Lagrangian density. we can calculate vacuum expectation values of timeordered products as path integrals· (T{~A0B"'})vAC~ J [gda'(X)gdn. If we take the canonical variables to be.6.6. with aJ and It) regarded as functionals of these variables given by Eqs.(X) gdV'!(X)] 0'A0)B'" x eXP{if d4 x[1t"a ~1t2_ 1(Vxa)2+ a .6.
7) just gives the Coulomb action ~ J dt VCou!.and gaugeinvariant.l) ~ Jdly i"(y.V'ao(x) ~ 0 or in other words.8) Using this in Eq.6.t) . i.6) To bring out the essential covariance of this result. and replace the Coulomb term .6 PathIntegral Formulation of Quantum Electrodynamics 1t 415 = a : (T{(I)A@B})VAC~ J[lJdai(X)gd~'t(X)] x (l)A(y'B··· exp {i Jd x [~a2 4 ~(V x a)2 +a ·j+2'M] iJ J dIVC.9) x 0A@B'" exp(illa.J dt VCoul in the action with d4x [aO(x)}"(X) + J(vaO(x))'] .6.a(x))].6.(X)] [g dW(X)] (9.10) Now everything is manifestly Lorcntz.6.6.6. ~)] = J J4x [ t fl".e.6. except for the fmal product of delta functions which enforce the Coulomb gauge . the integral over aD can be done (up to a constant factor) by setting aO(x) equal to the stationary point of (9. (9. we use a trick.6.6. 4nlxyl (9. the path integral (9.ill" + alIi' + .7).. Introduce a new variable of integration aO(x).6.'PI) II o(V'a(x)).6) is now (T{@A@)B"))VAC oc J[IT da. That is. (9.7) is quadratic in aD."l + iderms} [!f0(v. (9. (9. (9.HV x a)2 +a·j +2"'M aol + HVaO)2 lll ... to the solution of 1"(x) .t fi/vi ' + ald li + 2 M + total derivatives with Illv = alia" ~ ava l " and integrate over aO as well as over a and matter fields.6) as = 192 .::t'M] + ie terms. Hence we can rewrite the argument of the exponential in Eq.9. to aO(x. where I is the original action I [a.7) Since (9.
6. respectively. (9. we shall use a simple version of a trick 4. 'P]. and (!.6..] xexp(iI[a.6. 'P]. <!J B [A.!.~]) II b(V·a(x)+V2A(x)). We shall exploit this fact to put the path integral in a much more convenient form.6. &' B [A.functions with the original fields a!'(x) and W(x).15) •• Note that now aO(x) is not equal to the value (9. . the function A(x) was chosen at random.'P]. like changing an integral I::'DJf(x)dx to read J~CLf(y)dy. replace the field variables of integration QI'(x) and rp(x) everywhere in Eq.6. 'P].14) (where IX is an arbitrary constant).6. Multiply Eq. which would lead back to Eq.13) by the functional B[A. For simplicity.6.11) 'PtA (x)  exp (iqtA(x))w(x) (9. Eq.9) then becomes (T { <!J A [A. Next.!. (9. and integrate over A(x). we see that the effect is simply to multiply Eq.. (9. We will not integrate over Jl(x) first.13) cannot depend on this function. .6.al ~ exp (ji. .' X })VAC J[g da. This step is a mathematical triviality.6.13) with the fieldindependent constant (9. as well as the action I [a. (9. so despite appearances the righthand side of Eq.6. but instead will treat it in tandem with a(x). measure.6. (9.• 416 9 PathIntegral Methods condition:* To make further progress.V2 2) A) (9.]<!J B [a. (9.6).6. and noting the actual Aindependence of (9.13) Now. and does not require use of the postulated gauge invariance of the theory. tp] and measure [TI daHTI dtp 1are gaugeinvariant.12) with arbitrary finite A(x).9) with the new variables (9.13).. By shifting the integration variable A(x). we shall deal here with the case where the operators (r) A [A.5 that in Volume II will be used to treat the more difficult case of nonAbelian gauge theories. use gauge invariance to replace a/IA(x) and lJ'tA(X) in the action. . (9. First. Ja'x (aoaO .~). but is an independent variable of integration.(X)] [W'I't(X)] <!J A [a.6.
6. (9.9) now becomes (T{<!JA0 B' "})vAC oc J[IT da.] li 4 (x .19) tJ tf"x t y ajJ(x)aV(y) DjH.6. (9.aVail) ~ tex:( ojJa /l)2 =  + iE terms] (9.6.6. which yields the propagator in Feynman gauge: (q A JLx .y) + iE terms d4q[.'y=(2n)4 .9. in which case the factor Il"b(V'a(x)+V2A) in Eq.6.Ci)qltqv .22) . and thus has no physical effect.tiex: J tf"x (Oltall)2).20) The photon propagator is then found immediately by inverting the 4 x 4 matrix in the integrand of Eg.'f]).16) where ":x" again means proportional with a fieldindependent factor. yl.6.i\. Dropping constant factors.20) iE) We are free to choose ex: as seems most convenient.6 PathIntegral Formulation of Quantum Electrodynamics 417 This factor cancels out in the connected part of the vacuum expectation value.v~~an = (2n)4 J d4 q [q2": iE] eiq'(xy) (9. lp] . We consider the new term in (9.(X)] [IT d'!'f(X)] iTiA<!J B "'exp(iI.t(opa v .(XY).1I" q2 . But (9.et) " = (2n)4 J a~ oxf/ .6.6.6. Two common choices are a·= 1.6. Eq.iE'1lJv]eiq. This is now manifestly Lorentzinvariant.lp] = I[a. (9.13) is only AMindependent after we integrate over af/(x) and Ip(X).6.ff[a.18) as a contribution to the turbed part of the action. whose photonic part now reads unper~ Io[a] = J tf"x [.all)(iJllav . (9.6. (9.13) is replaced with J[I] dA(X)] (jia Jd'x (aoaO I] "(va(x) +V'A) V'Al') exp oc exp (.(1 .18) where Ielda.vy = ["pv J~ ()XPcyp :l (1 .~ex: J(Ollall)2d4x.I'Y' where f2 JLX . We can just as well integrate over A(x) before we integrate over all(x) and W(x).17) (9. (9. J d4q [q2 iE + (1 a I" (1 ) 2 qq ' ] " 2 eiq(xy). (9.21) A~~..
. The pathintegral formula (9. this is not a problem. and so on. but has no particular symmetry under separate permutations of the ps or ks.. k2.lI except that they apply thc pathmtegral method to the whole scattering process. In a relatiVIstic theory the possibility of particle creation and annihilation makes it neeessal')' to apply the path"integral method to fields rather than particle orbits.x.34) gives an amplitude (Pl. whether or not the particles are bosons or fermions or something else. they should be understood to accompany momentum labels. this amplitude is symmetric under pennutations of the ps and simultaneous permutations of the ks.418 9 PathIntegral Methods or Cf. In particular. For us. 9..d.) To calculate the amplitude for this process we can use the pathintegral method:· taking the qs and ps of Section 9.. But • This . These always satisfy canonical commutation rather than anticommutation relations. and may be omitted in a first reading. q2 _ if: (q2 .6.q = (2n)4 d4q'7 _qq.' .23) Practical calculations are made far more convenient by working with such manifestly Lorentzinvariant interactions and propagators. because we limit such calculations to sul1kiently early or late times.: = 00. etc. and we obtain the propagator in Landau gauge (often also called Lorentz gauge): dLan. . by some sort of slowly varying external fields.~omewhat out of the book's mam line of development. P2. we will consider the preparation of the initial or final states in a scattering process. rather than just the preparation of initial or nnal states. •• Here I am following the discussion of Laidlaw and C. keeping the particles far enough apart in the process to justify the use of nonrelativistic quantum mechanics. so at this point we are not committing ourselves to any particular statistics. another identical particle is brought continuously from momentum k2 to momentum P2. k2.1.1 as particle positions and momenta.6. The subscript 'D' indicates that this is the amplitude we would calculate for distinguishable particles. etc.uV I [ "" ''"] eill(xy).14) acts as a delta function. (9.7 Varieties of Statistics· We can now take up a question raised in Chapter 4: what are the possibilities for the change of statevectors when we interchange identical particles? For this purpose. in which case the factor (9.'leetion lie~ . ')0 as an integral over paths in which one particle is brought continuously from momentum k 1 to momentum PI. rather than fields and their canonical conjugates. when the particles participating in a scatlering process are all far apart. Suppose that a set of indistinguishable particles in either of these states is brought to a particular configuration with momenta PI. OeWitl. (Spin indices are not shown explicitly here. . P2. from a standard configuration with momenta kl. Ikl.au !.
d]q." 'IQ"Q2. )D (q."'lo ~ .71 ) the sum running over all N! pennutations of the N indistinguishable particles in the state. »'2. 3 X (P. C••' J J d q. "'(p"P2." (p"P2.Y'j.>"]. . Ik]. . . . Iq. .' j. (9.' X (P9""Y' I> py" . q2 .. 1 so the composition rule for the physical amplitudes may be written LC..···lkj. q9""2..!LC"C. . ".."'lq"q2.d'q... Ik" k2. /0 satisfy the composition rule for distinguishable particles (p"p"" 'lk"k2.1" JiP' X (p'P". . . . For space dimensionality d > 3. excludmg dveclors that coincide with (or are within an arbitrary limiting distance 00 each other. (Pi'" Pen." 'lk"k2. '}O (q.p".3) (q"q" "Ik"k..2) .0/'2. .(p"..7. c"..k2..P . t This is expressed formally in the statement that the first homotopy group of configuration space in d 2': 3 is the permutation group." y d]q." 'l..1'''2..1)..' • ." ')0 ~ X J d]q.''') x (Q"q2.1>1' b q.'C. But the amplitudes (PI.~.. Applying a permutation &" to both the initial and final states in the first amplitude on the right.9. Ik I..) ~ L .. k2. and Cg> a set of complex constants.) ~ N! J] ] d q.1'''2..· ")0 ~ . . '10. /D .. and identifying configurations that diller only by a permutation of the vectors.···.."'lD~ N' . (9. meant thc spacc of N dvectors."lk"k...···lk"k..P2..1' L C. the only such paths t are those that take k l .n" ·Ik"k.1"'.k 2. Iql... Hence the true amplitude should be written (p" P'. These amplitudes must satisfy a composition rule appropriate for indistinguishable particles: 1 (p"p2.0>'2.. k 2.: C? (p".pace' lor N distinguishable particles i..'?."lk"k2.. this is the requirement (9. . .7.···)D...·.P. " . this is LC? (P?hPen.d]Q2'" .7. .9' I> P. Ik!.'" into some nontrivial permutation:Y of PI. d]q.12 By 'configuration . P. q.' . k2. Using Eq." 'lo. Ik" k"" 'lo .' . but that yield the same final configuration. P2. then there are other paths that are topologically distinct. ." ')0 (9.. C.7 Varieties of Statistics 419 if the particles are really indistinguishable.
with K. In this case there is a much richer variety of topologically distinct paths. L" and M are all real. In consequence. with ego = + 1 or Cil' = 1 according to whether P is an even or odd permutation.l5 particles with more general permutation properties than just Fermi or Bose statistics. It has bccn shown 11 that parastatistics theories in d ~ 3 space dimensions are equIvalent to theories in which all particles are ordinary fennions or bosons. respectively.y'" .. For this purpose. but a larger group known as the hraid group. except that the matrix K is required to be symmetric and nonsingular.. L" and M are arbitrary constants. thcre is a matrix [/ with transpose yT = 9"1 such that (. Therefore.420 9 PathIntegral Methods which will be satisfied if and only if C:!I"(iI''' = C)'I C.TKY)" ~ b"K. This is expressed in the statement that the first homotopy group of conJiguration space in two space dimensions is not the permutation group. in two space dimensions it is possible to have anyons.:l: The nice feature of this argument is that it makes it clear why the case of two space dimensions is an exception.~ For instance. we begin by considering the case where K . and the other is the alternating representation.A. 14 . so that wave functions could have nnusual properties under permutations of momenta and spins.A.7. over a finite number of real variables ~" of the exponential of a general quadratic function of ~: (9.4) That is. the coefficients Cf!} must furnish a onedimensional representation of the permutation group.l) (9.s also positive. (9.o/ = +1 for all permutations.3) :j: There has been much discussion in the literature of possibilities other than Bose or Fermi statistics. But the permutation group has only two such representations: one is the identity.A. . often under the label parastatist'cs.s. but carrying an extra quantum number. (9. Sa Appendix Gaussian Multiple Integrals We wish first to calculate the multiple integral. These two possibilities correspond to Bose or Fermi statistics. a path in which one particle circles another a definite number of times cannot be deformed into a path where it does not. with C. Any real symmetric matrix can be diagonalized by an orthogonal matrix.2) where K rs ./. The result in the general case can then be obtained by analytic continuation.
(9.A. K~l = But the determinant and the reciprocal of Eq.1) for K rs .5) may be written } f~ (Det(~)r'exp{ q?.A8) as f ~ ( K Det ( 2n ))1/2 exp{Q(()} .\ is analytic everywhere in L r and M.5) ~ e.L. L r . Since (9. (9. The sign of the square root is fixed by this analytic continuation. so Eq.A.A. with K rs also positive. . (5'. with a cut required by the square root.A. (9. (9. (9A.3) give l:= YIl/Y { mtKt1 . (9. L r .Appendix Gaussian Multiple Integrals 421 Because K is assumed positive and nonsingular. In field theory K rs is actually imaginary.1 is real and positive.A7) (9. Eg.r. ~ ('  (9. so the multiple integral (9. and M all real.6) in terms of the stationary point of the function (9.6) equals (9.A9) This is the result to remember: Gaussian integrals can be evaluated up to a determinant factor by setting the integration variable equal to the point where the argument of the exponential is stationary.A4) The Jacobian IDetYI of this transformation is unity.A.} DetK = .(} .Al) is now given by a product of ordinary integrals: f ~ e.K. except for a small real part due to the 'iE term'. It is useful to express Eq. the eigenvalues K r are positivedefinite. (9.I. (9.. We can use the matrix Y' to perform a change of variables: ~r = l:= ''/'rs(~ .A. II .\. Kr d( exp { (SPTL). .M M Ill: .r L ).M (9.l) defines a function of K r.A6) provides an analytic continuation of Eq. and M that is analytic in K rs in a finite region around the surface where K r.r j~~ exp {2~.A. where the integral converges. and for such K r.A6) Eq.Al) to the whole complex plane.2): (9.
} [Det (~) K r'/2 exp x. leaving us with eN = K [ Det ( 2n )]'i' .12) is equal to the number (2N)! of permutations of the indices.422 9 Path· Integral Method8 We next wish to use this result to calculate the integrals I"". To calculate the constant factor eN.15) .A. we have the sum rule ~ / (I! d(.l). Comparing the coefficients of L" L r2 ••• Lr'N on both sides.(.A.(.K. '2N.A./) (9.) exp { . } (9A 10) (Integrals of this sort with an odd number of S'factors in the integrand obviously vanish..A.1.A.A. . (9.. which symmetry requires to take the form Jrlrrr2!\i = eN (9.12) of 'I "'rZN Here the sum is over all ways of pairing the indices q .ll) shows that the factors (2N)! and N!2 N are cancelled by ~'N. N L.} (9All) [ Del ( 2:n: )]'/2 ~ N!2''\/ ( ~LrLsK.~.ur.l )N I L II (K1)paired indices· pamngs p.~ L.'. (9. ~/ (I! d(.(..A.(.A.12) gives L" Lr'2 '1'2···'2N ••• L'2N1'lrr'2N = V/>/CN (L ".L"K... (9..Lr Lr~r) in Eq. exp { . divided by the number of ways N! of pennuting index pairs and by the number 2 N of permutations within index pairs VN = N !2 N (2N)! .~ ~ K. {~ ~L'L.13) Therefore.14) Comparing this with Eq.(.) (" (" <. we note that the number VN of terms in the sum over pairings in Eq. (9. (9. (9. or by the order of indices within a pair. we see that I"frr2N must be proportional to a sum of products of elements of K.j ~ K. Eq..) From the powerseries expression of exp (. with two pairings being considered the same if they differ only by the order of the pairs.
I )rlrl(K. The Lagrangian density of the free spin 3/2 RaritClSchwinger field If'1' is :£ = ijJl'( l (:Iv + m )'PI'  ! ijJl'(YltOV + }'v 0. 3.)"r4 (9. Find the wave function in field space of the vacuum in the theory of a neutral vector field of mass m 1. Find the wave function in field space of a state consisting of a single spinless particle of mass m 1. Ann Arbor). For a general reference.} (9A18) ~ [Det(~)r2. moving along the xaxis in a potential V(x) = mw 2 x 2 /2.A. 1942. 2. References 1.j ~ K"". Problems 1. 4. R P. Use pathintegral methods to find the propagator of this field. then it is between x and x + dx at time t.17) and so on. New York. Jaffe. Glimm and A. .O. 2948. Quantum PhysicsA Functional Integral Point of View. Quantum Mechanics and Path Integrals (McGrawHill. Use pathintegral methods to find the probability that if the particle is at Xl at time t). University Microfilms Publication No.A. R.References 423 For instance.)1// + t ijJl'y Jt( yU au  m)'/ 'Pv. Feynman.O./ (I. 1965). 2nd edn (SpringerVerlag. 1987). Also see R. Use the result to derive the form of the iE terms in the propagator of this field.I d'. Use the result to derive the Feynman rules for emission or absorption of such a particle. P.1)r2'4 + (Kl)rlr4(KI)r2rJ]' (9. The Principle of Least Action in Quantum Mechanics (Princeton University..16) + (K. I r1r2 = Io(K1)rlr2 ' 1 1 Irlr2rJr4 = In [(K.. where 10 is the integral with no indices /0 . Feynman and A. New York. see J. Consider a nonrelativistic particle of mass m.)wl(K.) exp { . Hibbs.
1604 (1968). 51. 749. 742 (1964). 76. New York. 15. Phys. Acta Phys. 13. by S. Myrheim. Wilczek. Artin. G. 1430 (1948). l. Gerstein. Schrader. Mande1stam. and S. Rev. Euclidean covariance. and J. 204 (1970). Reading. Lett. Leinaas and J. E. See The Collected Papers of E. L. 44. 49. Phys. De Witt. 8b. Kamefuchi. Phys. M. translation in Thear. Nuovo Cimento 37 B. Sci. P. P. Dirac.. 29. 12. 939. F. 9. G. 956 (1958). 1966). R. Leinaas and J. Math. F. 1580. 7. D. Artin. 62 (1933). 2486 (1971). Phys. Phys. Haag. B25. 8a. Also see R. This section was largely inspired by discussions with 1. Lett. Math. Phys. 1 (1970). 281 (1975). 1423 (1972). ed. Phys. Rev. and cluster decomposition. "Hoof!. 697 (1963). Phys. Phys. 1279 (1968). 167 (1971). Cambridge preprint DAMTP9490 (1994). K. A. Rev. D. 74. Ref. Fredenhagen. Gaberdiel. Mod. 957 (1982). 14. L. Proc. Acad. and S. S. Polchinski. Phys. M. D 3. Zeits. R. The OsterwalderSchrader axioms require smoothness. 20. Rev. Roberts. Lett. Phys. Lang and J. 12. The Method oj Second Quantization (Academic Press. Popov. Rev. Tate (AddisonWesley. 769 (1949). A. B. 1. Lett. M. M. y. 11.337 (1969). 1. Feynman. Jackiw. 29 (1967). Math. 83 (1973). 1 (1977). MA. Phys. K. Rev. Ann. 4. 440 (1950). 5. Phys. The braid group was introduced by E.1375 (1970). 367 (1948). Also see J. Myrheim. I. M. 80. 3 (1969). Rev. Faddeev. P.424 9 Path·lntegral Methods 2. Osterwalder and R. J. Laidlaw and C. D3. Phys. 1965). Nat. B35. Nucl. Feynman. 12. G. E. Schwinger. Math. 'reflection positivity. Ohnuki and S. S. Faddeev and V. R. N. 24. Phys. K. Berezin. 170. Commun. Rev. Fizika.' permutation symmetry. Ruger. 6. De Witt. 8. Sowjetunion 3. Lee. Phys. Phys. 31. Mat. B. DrUhl. Teor. 42. 10. R. Commun. Rev. Pol. Weinberg. 3. M. W. 175. M. Commun. . 18.
etc. etc. where the operators Oa(x). to parts of Feynman diagrams.4: the sum of all diagrams for a process CI: + f3 with extra vertices inserted corresponding to operators oa(x). consider the symmetry of spacetime translational invariance. this matrix element equals the sum of all Feynman diagrams with incoming lines on the mass shell corresponding to the state ct. As a special case.}'P. iO. Ob(X). is given by the matrix element of the timeordered product of the corresponding Heisenbergpicture operators ('Pp. are elementary particle fields.T{ iO.10 NonPerturbative Methods We are now going to begin our study of higher~order contributions to physical processes. Ob(Y). This symmetry has as a consequence the existence of a Hermitian 425 . corresponding to Feynman diagrams involving one or more loops. It will be very useful in this work to have available a method of deriving results valid to all orders in perturbation theory (and in some cases beyond perturbation theory). with some or all external lines off the mass shelL For instance.(x). The essential bridge between the Heisenberg picture and the Feynman diagrams of perturbation theory is provided by the theorem proved in Section 6. After exploring some of the nonperturbative results that can be obtained in this way we will be in a good position to take up the perturbative calculation of radiative corrections. etc. Ob(X).(y).1 Symmetries One obvious but important use of the theorem quoted above is to extend the application of symmetry principles from S~matrix elements. where all external lines have fourmomenta on the mass shell. and lines off the mass shell (including propagators) corresponding to the operators Oa(x).+). 10. outgoing lines on the mass shell corresponding to the state p. In this chapter we will exploit the field equations and commutation relations of the interacting fields in the Heisenberg picture for this purpose.
426
10 NonPerturbative Methods
fourvector operator pll, with the property that, for any local function O(x) of field operators and their canonical conjugates,
[p,(x),O(x)]
~
,
ia:,O(X).
~
(10.1.1)
(See Eqs. (7.3.28) and (7.3.29).) Also, the states to be eigenstates of the fourmomentum:
pll,¥
:l':
and
 PfJ
p are usually chosen
fJ
~
+
= .Jl,¥ +
f''Y.:l':'
pll,¥  _
fJ
i',¥
.
(10.1.2 )
It follows that for any set of local functions Oa(x), Ob(X), etc. of fields
and/or field derivatives
(pp,  p,,,) ('Pp~, T {O,(xIl, O,(X2)'" }'P,+) ~ ('P p, [p" T{O,(xIl, O,(x2)}] 'P,+)
~i(,8,+ vX + ...) ('P pT{0,(xIl,O,(x2)''''}'P,+) ,0, eX
l
2
(10.1.3)
This has the solution
('Pp,
=
T{O,(xIl,O,(X2),'"
}'P,+)
X2,"'),
exp (i(pp  Px) . x) Fab",(Xl 
(10.1.4)
where x is any sort of average spacetime coordinate
Xli = clxt
+ C2X~ + ... ,
Cl
+ C2 + ... =
1
(10.1.5)
and F depends only on differences among the xs. (In particular, a vacuum expectation value can depend only on the coordinate differences.) We can Fourier transform Eq, (10,1.4) by integrating separately over xl' and the coordinate differences, with the result that
I
d4xl ",X2'"
('P p T {O,(xIl,O,(x2),'" }'P,+) ,
x exp(~ikl . Xl  ik2' X2 _ ... ) oc /j4(pfJ  Pel  k 1 k2 ~" .). (10.1.6)
We saw in Section 6.4 that the matrix element of the timeordered product is given by applying the usual coordinatespace Feynman rules to the sum of all graphs with incoming particles corresponding to particles in :x, outgoing particles in {3, and external lines that simply terminate in vertices at Xj,X2,···, The Fourier transform (10.1.6) is correspondingly given by applying the momentumspace Feynman rules to the same sum of Feynman diagrams, with offshell external lines carrying fourmomenta kJ,k 2," into the diagrams. Eq. (10.1.6) is then just the statement that this sum of Feynman graphs conserves fourmomentum. The result is obvious in perturbation theory, because fourmomentum is conserved at
10.1 Symmetries
427
every vertex, so it is not surprising to see the same result emerging without having to rely on perturbation theory. With somewhat more effort, one can use the Lorentz transformation properties of the Heisenberg·picture fields and the 'in' and 'out' states to show that the sum of all graphs with a given set of on and offshell lines satisfies the same Lorentz transformation conditions as the lowestorder terms. Similar arguments apply to the conservation of internal quantum numbers, like electric charge. As shown in Section 7.3, a field or other operator Oa(x) that destroys a charge qa (or creates a charge qa) will satisfy
[Q,O,(x)]
~
q,O,(x)
in the Heisenberg and interaction pictures alike. Also, if the freeparticle states 0; and f3 have charges qrt and q(l, then so do the corresponding 'in' and 'out' states. We then have
(qlq,)
Thus the charge is conserved
}'I',+) ~ ('1'1' [Q, T{O,(X),Oh(Y),}] '1',+) ~ (q, + qb +"'j ('I'p, T {o,(xj,Oh(Y),'" }'I',+) amplitude (If' fJ , T { Oa(x), Oh(Y),' .. } 't' ~ +) vanishes
('1'1'
T{O,(X),Oh(Y),'
unless
(10.1.7)
A somewhat less trivial example is provided by the symmetry of chargeconjugation invariance. As we saw in Chapter 5, there is an operator C that interchanges electron and positron operators Ca(p,IT,e)C 1 = (a(p,IT,e+) , C a(p, IT, e+)C 1 = ( a(p, IT, e) , with
~
a phase factor. For the free electron field lp(x), this gives
C~(x)C1 ~ CPCV'(x)',
where PC is a 4 x 4 matrix, which (for the Dirac matrix representation we have been using, with Y5 diagonal) takes the fonn
o o o
0 o 0 1 1 o 1 0 o
1
0 0
0
Applied to the freeparticle electric current in spinor electrodynamics, this
428
gives
10 NonPerturbative Methods
C(ipY"'ljJ)C 1 = _ijJCyltTCtp = _ipyJJrp.
If C is to be conserved in electrodynamics, it must then also he defined to anticommute with the free photon field
C(aJl)C1 = _all .
In theories like electrodynamics for which C commutes with the interaction as well as Ho, it also commutes with the similarity transfonnation Q(t)
between the Heisenberg and interaction pictures, and so it anticommutes with the electric current of the interacting fields
C('!'y"I')C 1 ~ '!'I"1' (10.1.8)
and the electromagnetic field in the Heisenbergpicture
C(A")C I ~ A" .
(10.1.9)
It follows then that the vacuum expectation value of the timeordered
product of any odd number of electromagnetic currents and/or fields vanishes. Therefore the sum of all Feynman graphs with an odd number of external photon lines (off or on the photon mass shell) and no other external lines vanishes. This result is known as Furry's theorem.! It can be proved perturbatively by noting that a graph consisting of electron loops t, to each of which are attached nt photon lines, must have numbers I and E of internal and external photon lines related by an analog of Eq. (6.3.11):
2l+E~Lnt.
{
Hence if E is odd at least one of the loops must have attached an odd number of photon lines. For any such loop there is a cancellation between the two diagrams in which the electron arrows circulate around the loop in opposite directions. Hence Furry's theorem is a somewhat less trivial consequence of a symmetry principle than translation or Lorentz invariance; it is not true of individual diagrams, but rather of certain sums of diagrams. Figure 10.1 illustrates the application of Furry's theorem that was historically most important, its use to show that the scattering of a photon by an external electromagnetic field receives no contributions of first order (or any odd order) in the external field.
10.2
Polology
One of the most important uses of the nonperturbative methods described in this chapter is to clarify the pole structure of Feynman amplitudes as
10.2 Polology
429
Figure 10.1. The lowestorder diagrams for the scattering of a photon by an electromagnetic field. Here straight lines represent virtual electrons; wavy lines represent real and virtual photons; and the double line represents a heavy particle like an atomic nucleus that serves as a source of an electromagnetic field. The contributions of these two diagrams cancel, as required by chargeconjugation . . mvanance.
functions of the momenta carried by external lines. Often the Smatrix for a physical process can be well approximated by the contribution of a single pole. Also, an understanding of this pole structure will help us later in dealing with radiative corrections to particle propagators. Consider the momentumspace amplitude
J
d4 xl'" tf'x n eiql'xl
... eiqn'xn
(T{ Al(Xl)'"
An(xn)}) 0
=G(qj··· q,).
(10.2.1)
430
10 NonPerturbative Methods
The As are Heisenbergpicture operators of arbitrary Lorentz type, and (. '}O denotes the expectation value in the true vacuum '¥o+ = \fIo 'fa. As discussed in Section 6.4, if AJ,'" An are ordinary fields appearing in the Lagrangian, then (10.2.1) is a sum of the terms calculated using the ordinary Feynman rules, for all graphs with external lines corresponding to the fields Ai,'" An, carrying offshell fourmomenta ql ". qn into the graph. However, we will not be limited to this case; the Ai may be arbitrary local functions of fields and field derivatives.
=
We are interested in poles of G at certain values of the invariant squares of the total fourmomenta carried by various subsets of the external lines. To be definite, let's consider G as a function of q2, where
q  ql
+'" + qr =
qr+l '"  qn
(10.2.2)
with 1 <r < nl. We will show that G has a pole at q2 = _m 2, where m is the mass of anyoneparticle state that has non~vanishing matrix elements with the states pole is given by
Ar·,· At 'flo and Ar+1 ." An\flo, and that the residue at this
2i«+ m2 2 2 .
q
G~
+m
o
IE
(2,) 6 (q,
7 4
+ ... + q")
(10.2.3)
XL
M o;O},,,.(q2 ... qr )Mq,aIO(Qr+2 ... qn)
where the M s are defined by"
J J
LfXI,,·tx r eiql'xi
···eiq,.'x,
('fa, T{Adxl)···Ar(xr)}'fp,a)
(10.2.4)
~ (2,)4 64(q,
+ ... + q, 
p)M01p•o (q2··· q,),
•• ,
LfXr+l'" d4x n e iq ,.+l·Xr+l
eiq,,·x"
x ('fp,.,., T{Ar+I(Xr+l)"'An(xn)}'f'O)
=
(2n)4 o4(qr+1
+ ... + qn + p)Mp,aIO(Qr+2'"
qn)
(10.2.5)
(with pO _ VP 2 + m2 ), and the sum is over all spin (or other) states of the particle of mass m. Before proceeding to the proof, it will help to clarify the significance of
• Recall that in the absence 01" limevarying external fielJs, there is no Jistinetion between 'in' and 'out' oneparticle states, so that 'i'p," ~ = 'i'P,"  = 'PM'
10.2 Polology
(10.2.3) if we write it in the somewhat longwinded form
431
G(q1 ... q,)
~
J
1
J'k
3'2 x [ (2n) 4 J 4 (q1+"'+q,k)(2n)' ( / 2 +m2 ) 1/2 M 01 •.Aq2···q,)] 2vk
x [ i
(2n)4 k 2 + m2 
]
jf'
x [(2n)4 J4(k + q,+1 + ... + q,)(2n)3/2
x (2Vk 2 +m2t2 M','IO(q'+2' .. q,)]
(10.2.6)
This is just what we should expect from a Feynman diagram with a single internal line for a particle of mass m connecting the first r and the last n  r external lines:' However, it is not necessary that the particle of mass m correspond to a field that appears in the Lagrangian of the theory. Eqs. (10.2.3) and (10.2.6) apply even if this particle is a bound state of the socalled elementary particles whose fields do appear in the Lagrangian. In this case, the pole arises not from single Feynman diagrams, like Figure to.2, but rather from infinite sums of diagrams, such as the one shown in Figure 10.3. This is the first place where the methods of this chapter take us beyond results that could be derived as properties of each order of perturbation theory. Now to the proof. Among the n 1possible orderings of the times xV ... x~ in Eq. (10.2.1), there are n!/r!(n  r)! for which the first r of the x? are all larger than the last n  r. Isolating the contribution of this part of the volume of integration in Eq. (10.2.1), we have
G(ql"'qn)=
J
tfXj···tfxneiql·XI···eiIJ.·x"
xfJ (min [xV'" x~]  max [X~+l ... x~l)
x('Po,
T {A1 (xJl ... A,(x,)} T {A'+1(X,+Jl" . A,(x,)}
'Po)
(10.2.7)
+ QT,
where 'OT' denotes the other tenns arising from different timeorderings. We can evaluate the matrix element here by inserting a complete set of intermediate states between timeordered products. Among these may
., See Figure 10.2. The factors (br)J/2[2(k 2 + m2)]I/2 just serve to remove kinematic factor, associated with the mass m external line in M{)lk,~ and Mk,,, o. Also, the sum over IT of the product of coefficientfunction factors from these two matrix clcmcnts yields the numerator of the propagator associated with the intemalline in Figure 10.2.
432
10 NonPerturbative Methods
,
,+ 1
k
Figure 10.2. A Feynman diagram with the pole structure (10.2.6). Here the line carrying a momentum k represents an elementary particle, one whose field appears in the Lagrangian.
,
. ..
"
Figure to.3. Figure 10.3. A Feynman diagram of the class whose sum has the pole structure (10.2.6). Here the pole is due to a composite particle, a bound state of two elementary particles. The elementary particles are represented by straight lines, and interact by the exchange of particles represented by wavy lines.
be the singleparticle state \f'p,rJ of a definite species of mass m. Further isolating the contribution of these oneparticle intermediate states, we have
G(Ql"'Qn) =
J
tfXt" ··tfxneiqlXJ. ··eiQnxn
e (min [x?"· x~] + QT,
max
('Po, T{Al(xIl···A,(x,)}'P P•• ) ('PP,.' T{A,+l(x,+Il· .. A"(x")}'Po)
(10.2.8) where 'QT' now denotes other terms, here arising not only from other timeorderings, but also from other intermediate states. It will be convenient
[X~+l··· x~]) L •
J
d3p
10.2 Polology
to shift variables of integration, so that
Xi=XI+Yi,
Xi = X r+!
433
+ Yi
,
i=23"'r , " i=r+2,'" n,
and use the results of the previous section to write
('Po, T{A I (xIl"·A,(x,l}'I'".) ~ eip " ('1'0, T {AI (0)A'(Y2) " . A,(y,) } '1'".)
(10.2.9)
('I'
',ff>
T { A,+ (x,+ Il " . A,(x,) }
eip'X,+1 r
=
I '1'0) (\f'p,,,., T {A +l(O) ... An(Yn) }\f'o)
x~]
(10.2.10)
Also, the argument of the theta function becomes min [x?'" x~l
=

max (x~+!
x?  x~+!
+ min [0 yg
1
y~]  max [0 Y~+l ... Y~l .
We also insert the Fourier representation (6.2.15) of the step function
O(r)
~X r+l
1°C
oc
dw
w
e
1wr
2ni
+ iE
The integrals over
Xl
and
now just yield delta functions:
G(ql '''q,)
~
J
d"Y2 ,,·d"y,d'Y,+2 . "d"y,
x
eiQ2'Y2 ••• e iq,'y, e iQ,+2'y,+2 .•• eiqn'Yn
I x  2.
1rl
1°C
ex; W
dw. exp (iw[min[Oyg ... y~] max[O Y~+!'" y~]])
X
L Jd r('I'o, T {AI(O). ,
3
+ If'
A,(y,) }'I'".)
x ('I'"ff> T{A'+I(Ol"·A,(Y,l}'I'o)
X (2n)4b 3(p_qj '" qr) b(J p 2 + m2 + w  q? '" 3 X (2n)4 b (qr+1
q~)
+ ... + qn + p) i5 (q~+l + ... + q~ + Jp 2 + m2 +
w)
+ OT
.
(10.2.11)
We are interested here only in the pole that arises from the vanishing of the denominator w + iE, so for our present purposes we can set the factor exp(iw[minmax]) equal to unity. The integrals over both p and ware
434
10 NonPerturbative Methods
now trivial, and yield the pole
G(q! ... qn)
X
+
i(21ll J4(qJ
+ ... + qn) [qO _ Jq 2 + m2 + ito]1 +,,'
(10.2,12)
I: ,
MOI_q,,(qz'" q,) M q,'IO(q'+2'" q,)
where now
M01q,a( q2 ... q,)
JIf J
Y2 ... tf y, e iqn,
... eiqr'}"
x
('1'0,
T{Al(O)Az(yz)",A,(y,)}'I'q,,) ,
d4Yr+2'" d4Yn eiq'f2"Yrt2
(10,2.13)
M q,l110(Qr+2'" qn)
••.
eiqn"l'n
x
('1'"" T{A'+1(OlA'+2(Y,+zl"'A,(y,l}'I'0)
,
(10.2,14)
and the final '...' in Eq. (10.2.12) denotes terms that do not exhibit this particular pole. (The 'other terms' arising from other singleparticle states produce poles in q at different positions, while those arising from multiparticle states produce branch points in q, and those arising from other timeorderings produce poles and branch cuts in other variables.) Using Eqs. (10.2.9) and (to.2.1O), it is easy to see that these Ms are the same as defined by Eqs. (10.2.4) and (10.2.5). Also, near the pole we can write 1
qO J q 2+ m2 +if' _qO _
vQ2+ni! + i~
~
_(qO)2+(J q 2+ m2 if')2
2Jq 2 + m2 q2+ m2_if'
(We again redefine f' by a positive factor 2Jq 2 + m2 , which is permissible since ~ stands for any positive infinitesimal.) Eq. (10.2.12) is thus the same as the desired result (to.2.3). This result has a classic application to the theory of nuclear forces. Let <1>a(x) be any real field or combination of fields (for instance, proportional to a quarkantiquark bilinear IlYs1: a q) that has a nonvanishing matrix element between a onepion state ofisospin a and the vacuum, normalized so that
(10,2,15)
The matrix element of <J)a between onenucleon states with fourmomenta p,p' then has a pole at (p  p')2 + which isospin and Lorentz invariance (including space inversion invariance) dictate must take the
m;
10.2 Polology
435
form"
(10.2.16)
where u and u' are the initial and final nucleon spinor coefficient functions, including the nucleon wave functions in isospin space, and 1"a with a = 1,2,3 are the 2 x 2 Pauli isospin matrices. The constant Gll is known as the pionnucleon coupling constant. This pole is not actually in the physical region for the matrix element (10.2.16), for which (p  p')2 > 0, but it can be reached by analytic extension of this matrix element, for instance by considering the offshell matrix element
where Nand N' are appropriate components of a field operator or product of field operators with nonvanishing matrix elements between onenucleon states and the vacuum. The theorem proved above in this section shows then that exchange of a pion in the scattering of two nucleons with initial fourmomenta PI,P2, and final fourmomenta Pl, yields a pole at (Pl  P'I)2 = (P2  p;)2 + m~:
P;
SN' N'
1 2"
NjN,
•
+ i(2:n:)4b (p'l
4
+ P2 
PI  P2),ccc'"ccco (PI Pl)2+m~
G'
x (2n)3(UI'}'STaUl) x (2:n:)3(J2Y5TaU2).
(10.2.17)
(The easiest way to get the phases and numerical factors right in such formulas is to use Feynman diagrams; our theorem just says that the pole structure is the same as would be found in a field theory in which the Lagrangian involved an elementary pion field.) Again, this pion pole is not actually in the physical region for scattering of nucleons on the mass shell, for which (PI  P'I)2 > 0, but it can be reached by analytic extension of the Smatrix element, for instance by considering the offshell matrix
•• Lorentz and isospin invariance requires this matrix to take the form (ii' r to u), where r is a 4 x 4 matrix for which the bilinear (ij,' r 'I') transforms as a pseul.1oscalar_ Like any 4 x 4 matrix, r can be expanded as a sum of terms proportional to the Dirac matrices 1, /'~, ['/1" "~], }"s/~' and Yo. The coefficients must be respectively pseudoscalar, pseudovector, pseudotensor, vector, and scalar. Out of the two momenta p and p' it is possible to construct no pseudos.calars or pseudovectors; just one pseudotcnsor, proportional to eJl"Po Pp p~; two independent vectors, proportional to p~ or P~; and a scalar proportional to unity, in each case with a proportionality factor depending on the only independent scalar variable, (p  p')<. By using the momentum·space Dirac equations for u and II, it is ea.o;y to see that the tensor and pseudovector matrices in r give contributions proportional to Y5.
436
element
10 NonPerturbative Methods
Jtf
Xl
cr Xl d4 x~ d4 xS e~iPl
Xl
eip2' XI elp; Xl eip~ Xl
x ( T (N, (xd, N2(X2), N: (x',), N,(x,)} hAC.
Although this pole is not in the physical region for nucleonnucleon scattering, the pion mass is small enough so that the pole is quite near the physical region, and under some circumstances may dominate the scattering amplitude, as for instance for large t in the partial wave expanSIOn. Interpreted in coordinate space, a pole like this at (PI  P'lf = (pzP'2)2 _ implies a force of range Ijm lt _ For instance, in Yukawa's original theory2 of nuclear force the exchange of mesons (then assumed scalar rather than pseudoscalar) produced a local potential of the form exp(m1!r)/4nr, which in the first Born approximation yields an S~matrix for non~relativistic nucleon scattering proportional to the Fourier trans~ form:
m;
J
d3XI d3x2 d3x; d3Xl e ixlp] eixZ'Pzeixl 1'PI 1eiX2 1,P'2 1
exp( ~m1!lxl~x21) x 4 I I 03(XI  xI')b 3(xl  XI') 1"1: Xl  X2
=(21"1:)3b3(PI+P2~PI'~P2')(
1 )2+ 2' PIP1 ' m
The factor l/[(PI PI ')2 + m2 ] is just the nonrelativistic limit of the propagator l/[(PI  p;)2 + m;l in (10.2.17). (In (10.2.17) the energy transfer 2 p~  p~ for Ipil < mN and Ipil < mN equals [P1 2 ~ P1 l/2mN, which is negligible compared with the magnitude Ipi I of the momentum trans· feLl When Yukawa's theory was first proposed, it was generally supposed that this sort of momentumdependence arises from the appearance of a meson field in the theory. It was not until the 1950s that it became generally understood that the existence of a pole at (P1  p~)2 _ follows from the existence of a pion particle and has nothing to do with whether this is an elementary particle with its own field in the Lagrangian.
PI
m;
10.3
Field and Mass Renormalization
We will now use a special case of the result of the previous section to clarify the treatment of radiative corrections in the internal and external line of general processes. The special case that concerns us here is the one in which the four
10.3 Field and Mass Renormalization
437
momentum of a single external line approaches the mass shell. (In the notation of the previous section, this corresponds to taking r = 1.) We will consider a function
Gt(qlq2"')
=
J
tfxltfx2'"
eiQI'Xl
e~iQ2·X2 ...
x ('1'0' T{Cir(xd,
A2(X2),'"
}'I'o) ,
(10.3.1)
where I!Jt(x) is a Heisenbergpicture operator, with the Lorentz transformation properties of some sort of free field 1jJt belonging to an irreducible representation of the homogeneous Lorentz group (or the Lorentz group including space inversion for theories that conserve parity), as labelled by the subscript t, and A2, A3, etc. are arbitrary Heisenbergpicture operators. Suppose there is a oneMparticle state \{Iqj,a that has nonvanishing matrix elements with the states \{Io and with A2A3'" \{Io. Then according to the theorem proved in the previous section, Gt has a pole at = _m 2, with
e;J
q?
Gr(qlq2"')~
X
Jd"
2iJqj 2+ m2 3 2 2 . (2n) ql + m  It:
...
L a
(
'I'o,Ci r (O)'I',,,"
)
X2 "·e iq" ,
('1'".• T{A2(X')'" }'I'o) ,
(10.3.2)
We use Lorentz invariance to write
('I'o,Ci r (O)'I',,,") ~ (2n)3/2 N Ut(qJ,u), (10.3.3)
where ut{q,a) is (aside from the factor (271")3/2) the coefficient function" appearing in the free field !Pt with the same Lorentz transformation properties as (!)t, and N is a constant. (It was in order to obtain Eq. (10.3.3) with a single free constant N that we had to assume that (!)e transforms irreducibly.) We also define a 'truncated' matrix element Me by
J
J.4x2
"'eiQ2'X2 ... (\{Iqj,a
r{A2(X2)'" }\{Io)
 N 1(2n)3/2
L U;(ql, u) Mr(q,···)·
r
(10.3.4)
Eq. (10.3.2) then reads, for q2
Gr
t
t
ml
~2iJqt 2 + m 2
2
qj
+m
2
.
 It:
L
Ut(q\,a)u;,(qt,a)M{'.
(10.3.5)
a,l'
According to Eqs. (6.2.2) and (6.2.18), the quantity multiplying M t ' in (10.3.5) is the momentum space matrix propagator idu,(q!l for the free
• For instance, for a conventionally normalized free scalar field, U{(qj,u) = [2Jqf +m 2jI/2.
ur(qj. According to Eg. t ql. Eg. One important aspect of this result is that it applies to any sort of operator.I f ' IT X .3. qr ('Po.2) reads J~XI..'" in Eg. and the particle it creates may be a bound state composed of those particles whose fields do occur in the Lagrangian. (1)/ need not be some field that actually appears in the Lagrangian. (10.3. • This is just the usual behavior of a propagator (the sum of all graphs with two external lines) near its pole.A 2. which we have proved here by a somewhat different method that has allowed us easily to generalize this result to the case of arbitrary spin.0".6) is called a renormalized. Symanzik.3. The field renormalization constant N shows up in another place. so (10.3.t 0.(Xl) } 'Po) iqj 2iyqj' + m (2n)3 ' " ('P 0 (O)'P ) 2+ l ' L. (10.o) ~ (2n)3/2 U{(q./ qi+m' ~ d4 Xl eiqlXj e iqP2 2 (If'o. (10. The only discrepancy with the usual Feynman rules is the factor N.3. Suppose that there is just one of the operators Al... Then Eg. 'Pr(O)'P. Q2'" corresponding to the operators rs:{.5) allows us to identify Mt as the sum of all graphs with external lines carrying momenta qt. T { (!J'tCq )ll).AJ. and contract with the usual external line factor (2n)3/2 u.438 10 NonPerturbative Methods field with the Lorentz transfonnation properties of fI:r (or at least its limiting behavior for + _m 2 ).6) A field normalized as in Eg.(O)lf'o ) 4 4 2iINI\/ql l ' ml + 2 2 ql+m Jf' """' L.u).4) is then just the usual prescription for how to calculate the matrix element for emission of a particle from the sum of Feynman diagrams: strip away the particle propagator. and Zimmerman.6). (I)~.field.3.. It provides an important lesson even where (S'/ is some field lf1t in the Lagrangian: if we are to use the usual Feynman rules to calculate Smatrix elements.3 known as the reduction formula.O" ql m .3."'.u)ut . (10. Thus a renormalized field is one whose propagator has the same .I d4 Xl eiq2'X2 e" 'Xl (If' ql . this factor is absent in the propagator of the renormalized field If'r.(qj. The above theorem is a famous result due to Lehmann.1). except for the factor INl l . then we should first redefine the normalization of the fields by a factor 1/ N. but with the final propagator for the (!h line stripped away.u)(2n) 0 (qj +q2). (10. and take it to be the adjoint of a member of the same field multiplet as (!Jr. so that (with apologies for the multiple use of the symbol 'P). (10.
Then the corrections to the complete propagator are given by a sum of chains of one. In calculating the corrections to the complete momentum space propagator of the renormalized scalar field. An example is shown in Figure lOA.y.11) (10. (10. The Lagrangian density is taken as usual as (10. the subscript B being added here to remind us that so far this is a 'bare' (i.6).2 2m 'I' .3 Field and Mass Renormalization 439 behavior near its pole aslor afreefield.3.7) may then be rewritten (10.10. It is conventional to write the sum of all such graphs. conventionally called d'(q). with the asterisk to remind us that these are oneparticleirreducible graphs.3.. two.9) with Z to be chosen so that <D does satisfy Eq.. ~jJ.3.3. or more of these oneparticleirreducible subgraphs connected with the usual uncorrected .y. and the renormalized mass is defined by the position of the pole. and 15m2 chosen so that the pole of the propagator is at q2 = _m 2 . (10.e. To see how this works in practice.y. as i(2rr)4fl'(q2).8) (10.7) In general there would be no reason to expect that the field <D n would satisfy condition (10. (The use of the symbol Z in this context has become conventional. it is convenient to consider separately the oneparticleirreducible graphs: those connected graphs (excluding a graph consisting of a single scalar line) that cannot be disconnected by cutting through anyone internal scalar line.iE)1 omitted.6).3.3.JZ<I».12) where V(<I» =V B( . jUp'l'() 'I'  I 2. consider the theory of a real selfinteracting scalar field <DB..10) W ~ 0  _  I '" .3. there is a different Z for each field in the Lagrangian.3.) The Lagrangian density (10. with the two external line propagator factors _i(2rr)4(q2 + m2 . unrenormalized) field. nor that the pole in q2 would be at m~.3.. so let us introduce a renormalized field and mass (10.
.. Figure 10. like the theory of a scalar field ¢ with interaction proportional to ¢4.17) Also.4 q2 + ~2 iE] +.4(a): 0·(q2) ~ (Z .3.(q2)] [ (2n:)4 q2 ! + m 2 ~ IE 1. propagator factors: j d'(q) = I (2n)4 (271")4 q2 + m2  1 it: + [(2n:)4 q2 + m2 IE ] i 1.3. we encounter a tree graph arising from a single insertion of vertices corresponding to the terms in Eq. (10. the condition that the pole of the propagator at q2 = _m 2 should . this gives (10.14) + m2 _ Summing the geometric series.3. ] i 1 1 + [(2n)4 q 2+ m2_iE ]['(2)4 0 ..l)[q2 + m2] + Z 3m2 + OLOOp(q2) . (10.3.. oneparticle irreducible. so that (10. or (b) are not.15) In calculating II·. These diagrams are drawn for a theory with some sort of quadrilinear interaction. Diagrams that (a) are.440 10 NonPerturbative Methods () (b) ()«).3. x [i(2n)40·(q2)] or more simply d'(q) = [q2 [i(2n)40.12) proportional to 0il<Ili1il <Il and <Ill.3. (10. _ (10..16) The condition that m2 is the true mass of the particle is that the pole of the propagator should be at q2 = ~m2..(2)][i q 2+ m2ie ] I 11: q (2n)4 [(2~.13) ier J + [q2 + m2 _ ier 1rr*(q2)[q2 + m2 _ iEr1 +[q2 + m2 _ iEr1rr*(q2)[q2 + m2 _ iEr1rr*(q2)[q2 + m2 _ jEri + . Figure lOA. plus a term IIioop arising From loop graphs like that in Figure lO.4..
) The Lagrangian density is then rewritten st' = st'o + st'l . . justifying the treatment of the first two terms in Eq. [n·(q2)[q2 + m2 _ if"]l + +. this subtraction procedure incidentally cancels the infinities that arise from the momentum space integrals in n LOOp .3. (10.19) (10.3.3. As we shall see.3.20) d. An important consequence of the conditions (10.3.12) as part of the interaction st'l. (10.3.18) These conditions allow us to evaluate 2 and om2 : 20m2 = IlLOOP(O).3. However.10. as this discussion should make clear. For instance.3. (10. 2 = (10.3.18) is that it is not necessary to include radiative corrections in external lines on the mass shell. 2 ] 1 + [ dq2Ilwop(q) q2=_m 2 This incidentally shows that 20m 2 and 2 1 are given by a series of terms containing one or more coupling constant factors.3.3 Field and Mass Renormalization 441 have a unit residue (like the uncorrected propagator) is that d.18). ]q2+_m2 ~O.3.3.3.21) Similar remarks apply to particles of arbitrary spin.25) (10. That is.22) We introduce renormalized fields and masses . [ .24) (The subscript 2 on 22 is conventionally used to distinguish the renormalization constant of a fermion field. (10. the renormalization of masses and fields has nothing directly to do with the presence of infinities.17) and (10. :£ ~ n·(q2)[q2 + m 2 _ if"jI n"(q2)[q2 + m2 _ if"r l (10. m=mB+om.. for the 'bare' Dirac field the Lagrangian is 'P B[? + mBI'l'B  VB('I'B).23) (10.26) 2'0 ~ 'I'[? + m]'I'.3. (10. (10..1/2 'I' =2 2 \f'B. In actual calculations it is simplest just to say that from the loop terms n wop (q2) we must subtract a firstorder polynomial in q2 with coefficients chosen so that the difference satisfies Eqs. and would be necessary even in a theory in which all momentum space integrals were convergent.17) and (10.q! I (q) d 2] q2=_m 2 ~O.
~ [i ¥ + m . (Lorentz invariance is being used to justify writing r as an ordinary function of the Lorentz scalar matrix ~ kll'lJJ.im tells us that radiative corrections may be ignored in external fermion lines.442 10 NonPerturbative Methods (10.3.27) Let i(21!)4I:*(~) be the sum of all connected graphs. (10.27) proportional to \{J WP and lptp as well as loop contributions: L'W ~ (22 I)[i ¥ + m] + 226m + L~oopW . and hence im ~O. and with external line propagator factors i(2n: )4 and [i ~ + m .33) !f=im Just as for scalars..iEjIr(¥)[i ¥ + m . Corresponding results for the photon propagator will be derived in Section 10.) in the limit Jf. Recall that the invariance of the Lagrangian density . + m]tI:" (J/.3.i<jI + [i ¥ + m .if'jI omitted. 10.iErlrW[i ¥ + m .5. the vanishing of [i J/. with one fermion line coming in with fourmomentum k and one going out with the same fourmomentum.29) The condition that the complete propagator has a pole at k 2 = _m 2 with the same residue as the uncorrected propagator is then that L'(im) ~ 0.3.28) In calculating I:*(.iErl + .3. (10.3. . (10. that cannot be disconnected by cutting through any single internal fermion line.3. (10..30) (10.31) JJL'WL iJ Jf.i<jIL'(¥Hi ¥ + m .4 Renormalized Charge and Ward Identities The use of the commutation and conservation relations of Heisenbergpicture operators allows us to make a connection between the charges (or other similar quantities) in the Lagrangian density and the properties of physical states.L'(¥) . (10.) Then the complete fermion propagator is S'(k) ~ [i ¥ + m.iErl + [i ¥ + m .¥") we take ioto account the tree graphs from the terms in Eq.32) 2 ~ 1 _ i JJL~oopW 2 JJ ¥ (10.3.iErl .3.
4.4 Renormalized Charge and Ward Identities 443 '.1.4.) Also.a. and hence (assuming no vacuum degeneracy) must be proportional to 'Po itself. But the proportionality constant must vanish. momentum. Q] ~O. (10.6) It follows that Q acting on the true vacuum If'o must be another Lorentz invariant state of zero energy and momentum.llil (10. Q acting on anyoneparticle state 'Pp. Hence Q'I'o~O.4.a. and Lorentz transformation properties.n . = O.4. (10. This eigenvalue is what is known as the electric charge (or whatever other quantum number .n must be another state with the same energy.10. that the integral (10.3) where Q J d3x JO • (10.8) The Lorentz invariance of Q ensures that the eigenvalue q(n) is independent of p and (J. (10.4) may not exist if there are longrange forces due to massless scalars in the system. since it is a spaceintegral.5) and since JII is a fourvector.1) satisfying the conservation condition {}J. and thus (assuming no degeneracy of oneparticle states) must be proportional to the same one·particle state Q If'p. ( ." a:e alaII 'I' t ) q. Q is manifestly translation· invariant (10.4. Q is invariant with respect to homogeneous Lorentz transformations [1'".4) (There is a very important possible exception here. depending only on the species of the particle.n = q(n) If'p.4.4. (10. HI ~0. (10. We will return to this point when we consider broken symmetries in Volume II.4. with respect to global gauge transformations If'{ _ exp(iq{CI:)1f'{ (with an arbitrary constant phase) implies the existence of a current J" _ . 'I'! .I L.4. Q ~ [Q.a.7) Also.2) This implies that the spaceintegral of the time component of JIl is timeindependent: i :. because Lorentz invariance requires that (If'o. JJilf'o) vanish.
whether or not their fields appear in the Lagrangian. we have ('1'0.10) The same is true of any local function F(y) of the fields and field derivatives and their adjoints.444 10 NonPerturbative Methods of which ]Jl may be the current) of the one~particle state.4.8).4. containing definite numbers of each: (10.t). we note that the canonical commutation relations give [JO(x.t)p(x . (10.exp(iqto:)\{It. . the physical electric charge is just equal to a parameter qr appearing in the Lagrangian (or to a sum of such parameters.rI' For a oneparticle state corresponding to one of the fields in the Lagrangian we could take F = 'Pt. (1004.) The qualification that has to be added here is that the requirement. The physical electric charges are those that determine the response of matter fields to a given renormalized electromagnetic field AP.14) is the condition that assures that momentum space Greens functions involving F have poles corresponding to the oneparticle state \{Ip.7) and (10.) (qF Hence we must have qt') ~ O.4.13) as long as (10. Taking the matrix element of this equation between a oneparticle state and the vacuum. that the Lagrangian be invariant under the transfonnations 'P t .4. 'I't(y.11) where qF is the sum of the qt for all fields and field derivatives in F(y). the scale of the qt is fixed by requiring that the renormalized electromagnetic field appears in the matter Lagrangian 'zM in the linear combinations [011 .4. minus the sum of the qt for all field adjoints and their derivatives. in which case qF = qt. tells us that despite all the possible highorder graphs that affect the emission and absorption of photons by charged particles. like qF .U. does nothing to fix the over·all scale of the quantities qt.t)] ~ qt'l't(y. That is. but our results here apply to general oneparticle states. and using Eqs.iqtAp]\{It.9) or integrating over x: (10. This almost.4.12) (10. To relate this to the qr parameters in the Lagrangian. (10. Eq.4.. F(y)'I'••. (10.y).4.14) As we saw in the previous section. but not quite.
4. consider the Greens function for an electric current J1l(X) together with a Heisenbergpicture Dirac field qtn(Y) of charge q and its covariant adjoint 'Pm(z). is (10. We can see a little more deeply into the nature of these cancellations by making use of the celebrated relations between these charged particle propagators and vertices known as the Ward identities. It is only necessary to assume that for some reason the charges qat in the Lagrangian are equal and opposite for the electron and for those particles (two u quarks and one d quark) that make up the proton. We define the electromagnetic vertex function r ll . can have the same charge as the positron. we should define the renormalized charges by (10.16) The renonnalized electromagnetic field (defined to have a complete propagator whose pole at p2 = 0 has unit residue) is conventionally written in tenns of A~ as All = 2 31/ 2 A~ . the effect of higherorder corrections then appears solely in the common 1/2 f actor 2 3 . that is surrounded by a cloud of virtual mesons and other strongly interacting particles.4. whose interactions are all much weaker.lOA Renormalized Charge and Ward Identities 445 so that the current JJI. and qr are not the same as the 'bare electromagnetic field' A~ and 'bare charges' qHt that appear in the Lagrangian when we write it in its simplest form !:f =  HOIlABv .ovABIl)(oJl. with 1 2 a proportionality constant 2 3 / that is the same for all particles. [011. (10. there must be cancellations among the great variety of other radiative corrections to the propagators and electro~ magnetic vertices of the charged particles.15) But AJI.oVA~) +!:fM ('lit. In order for charge renormalization to arise only from radiative cor· rections to the photon propagator. For instance. This helps us to understand how a particle like the proton.17) so in order for the charge qt to characterize the response of the charged particles to a given renormalized electromagnetic field.18) We see that the physical electric charge q of any particle is just proportional to a parameter qB related to those appearing in the Lagrangian.lqtt) (10.4.4.AB .iqBtABJl.
4. these functions take the values The one·loop diagrams that provide corrections to these limiting values are shown in Figure 10.(y)'P m(z)}'I'o) eikYe+'h. (104. the complete Dirac propagator.iqS.4.19) ..t)S~'m(t)J'(p is.(k..19) gives the sum of all such graphs with an extra photon line attached. T{J"(x)'I'. so P' is the sum of "vertex" graphs with one incoming Dirac line.446 10 Non·Perturnative Methods Figure 10. Eq. and one photon line. (10. but with the complete Dirac external line propagators and the bare photon external line propagator stripped away_ To make the normalization of S' and rJ< perfectly clear.(k)r:'m. Also. i. of the charged particle by J where  d'xd'yd'ze'P'eik'e W ' ('1'0.'''(k)J'(k t)  J d'yd'z ('1'0' T{'I'. we mention that in the limit of no interactions. Here straight lines are electrons.e. Eq.4.5.(y)'P m(z)}'I'o) +k t). one outgoing Dirac line.5.20) According to the theorem of Section 6. (10. (10.20) gives the sum of all Feynman graphs with one incoming and one outgoing fermion line.4. Diagrams for the first corrections to the electron propagator and vertex function in quantum electrodynamics. wavy lines are photons.. .
The conservation condition (10.(y)'f'.25) by letting t approach k.27) " .. 1)6 3(x .4. (10. (10.(y)'f'.4.(y) [JO(x). (10. derived earlier by Wards [rom a study of perturbation theory. 'I' .4.9).4 Renormalized Charge and Ward Identities 447 We can derive a relation between p' and S' by use of the identity .21) then reads O~" T {J"(x)'I'.22) and its adjoint [JO(x.4 The original Ward identity..(y.28) " S'I(k) ~ i ~ +m~'(M.. + q b4(x  z) T {'I'.4.4. (10.4.t) S'(t) ~ i S'(t)  is'(k) or in other words (10.(y.25) This is known a~ the generalized Ward identity. . (10.26) may be written P(k. I)] ~ q'l'.4.(y)'f'.y) (10.23) Eq. (10.I).3.(z)} . 'f'.25) gives P(k.y) .yO) T { [f'(x).(y)] 'f'm(Z)} + b(xo  zo) T {'I' .k) ~ i o~ S'I(k).k)~y"+i o~ ~·(M· (10.(z)] } .. can be obtained from Eq. 'f'.4.(y.k)" S'(k) P(k. (10.4.2) tells us that the first term vanishes.(y.24) Inserting this in the Fourier transform (10..4.4.(X . I)..(y)'f' m(Z)} ~ q 6 4(x .k) by Eq. In this limit. which here give [JO(X..y) T {'I' .19) gives (t .(z)} oxl ' + b(xo .4.4.(z)} ~ T{a"J"(x)'I'.10. 'I'. first derived (by these methods) by Takahashi. while the second and third terms can be calculated using the commutation relations (10. (10.4..0 T{J"(x)'I'. so Eq. I)". (10.(z) } .(y)'f'. Eq.21) where the delta functions arise from timederivatives of step functions.26) The fermion propagator is related to the selfenergy insertion r:(. I)] ~ q'f' .
etc. it will be sufficient to show the vanishing of the first of these quantities.2) Since M is defined symmetrically with respect to the photon lines.27) tell us that on the mass shell (10.3) The electric current JI'(x) is conserved.5 Gauge Invariance The conservation of electric charge may be used to prove a useful result for the quantities M~~"··(q. q'. (10. 10.5. (10.. (10.q'.·. (10. For this purpose.3) vanishes. JI" (x') . ( \f'b.. leaving the electric charge again unchanged.q'. but this does not immediately imply that Eq. (and/or q.)=i Jd Jd 4x 4 x' ..28) where [jYttkll + m]uk = [iY/lk Ji + m]uk = O. (105. x eiq·xeiq'·x' .4. } \f' a+).or offshell photons having fourmomenta q.). T{J"(x). note that by an integration by parts qI'M~~"·'(q..1) vanishes when contracted with anyone of the photon fourmomenta: (105.. as is the case when we set out to measure the fermion's electric charge. because we still have to take account .1) In theories like spinor electrodynamics in which the electromagnetic interaction is linear in the field AP..31) and (10. in an arbitrary transition (J. a~1' T { JII(X). with external1ine photon coefficient functions or propagators omitted.448 10 NonPerturbative Methods For a renonnalized Dirac field. t p. this is the matrix element for emission (and/or absorption) of on.4.J"(x')'" }'I'.···) J J trx crx'···eipeiq'·x .3. Thus the renormalization of the fermion field ensures that the radiative corrections to the vertex function r J1 cancel when a fermion on the mass shell interacts with an electromagnetic field with zero momentum transfer. etc. If we had not used a renormalized fermion field then the corrections to the vertex function would have just cancelled the corrections due to radiative corrections to the external fermion lines.5.5..+). Eqs. x ('I'b... Our result is that Eq. q'.
It follows that [JO(x. F(j.4) vanishes. t). For instance. directly from terms in the Lagrangian.5) (10.l) b3(x  y). 6 Where the current includes terms arising from a charged scalar field lD. However. In many cases it turns out that there are nonvanishing contributions to the commutator of JOCi. and will use a regularization procedure (dimensional regularization) that does not lead to Schwinger terms.5. either arising from the regulator procedure (if gaugeinvariant) or. all these Schwinger terms are cancelled in multiphoton amplitudes by the contribution of additional interactions that are quadratic in the electromagnetic field. We will be dealing mostly with charged spinor fields. t) with the regulated current JiG.b(yO .5. minus the sum of the qtS for the field adjoints and their derivatives. we get an equaltime commutator like this (inside the timeordered product) for each current aside from JIl(X) itself.xo) l' (y) JO(x) ~ b(xo . In deriving Eq. (l0. I). P(J.t)] ~ 0 and therefore Eq.5) we should take into account the fact that a pl'Oduct of fields at the same spacetime point y like the current operator r(y) can only be properly defined through some regularization procedure that deals with the infinities in such products. so that Eq. For the electric current.6) as was to be proved. There is an important qualification here.5. there are additional regulatorindependent Schwinger terms involving lDtlD..5. (10. JV(y) is itself an electrically neutral operator. for just two currents T {JI'(x)J"(y)) ~ e(xo . so .5. I)] ~ qF F(x. where qF is the sum of the qtS for the fields and field derivatives in F.4) With more than two currents. we recall that (as shown in the previous section) for any product F of field operators and their adjoints and/or derivatives [JO(X. (10.yO) [JO(x). known as SchWinger terms.5.10. taking account of the conservation of P(x): ~ Il oX T (J"(x) 1'(y)} ~ b(xo  yO)J°(x) J"(y) . as for charged scalars.yO)J"(x)1'(y) + e(yO  xO)1'(y)JP(x) so. To evaluate this commutator. Ply)] .3) gives (10.5 Gauge Invariance 449 of the xOdependence contained in the theta functions that appear in the definition of the timeordered product. qJ is zero.I). (10.
2"') (10.2). and LL W #I" and c are entirely arbitrary (not necessarily constants.. and not necessarily the same for all propagators or polarization vectors.1 (kl. kept in the states \{I fJ . I 1'1!12"' xMha v I1'2"'P1P2 ""1"2"'( k k ql.1) calculated in the absence of electromagnetic interactions: The invariance of Eq. The same argument yields a result like Eg. but wilh photons deleted.5)."'.. (10..5.5.. The complete photon propagator. i. such as those we encountered in the derivation of the Ward identity in the previolls section.9) under the 'gauge transformations' (10.ql. (10.5..7) and (10.5. Nole that the arguments of M are all taken to be j/l('(Jm."'.) .5. (10.5.2) receives contributions from nonvanishing equaltime commutators.8) follows immediately from the conservation conditions (10.450 10 NonPerturbative Methods in what follows we will ignore this issue and continue to use the naive commutation relation (to. but only to sums of diagrams in which the current vertices are inserted in all possible places in the diagrams.2) is that Smatrix elements are unaffected if we change any photon propagator 8 1IV (q) by 8Jtv(q) + L'illv(q) + ·':J.5.9l· .) This result is not as elementary as it looks. (In Section 9.5. provided that all charged particles are taken on the mass shell.5. (10.q2.{~ k.21). as it applies not to individual diagrams.e. which IS why we have to insert various signs for some of the arguments of Min Eq. t: al (k I A ka2 (k2.6.6 we used the pathintegral formalism to prove a special case of this theorem.and \{I C( +.2 (k~A.5.q2.9) where MPtf··· is the matrix element (10.7) or if we change any photon polarization vector by (10.2) even if other particles besides photons are off the mass shell.ng [ourmomenta.5.! trql d4q2'" ~1'1"2(qdAlt"V2(q2)'" X E.) This is (somewhat loosely) called the gauge invariance of the Smatrix. that vacuum expectation values of timeordered products of gaugeinvariant operators are independent of the constant:x in the propagator (9.8) where kO _ Ikl./lqv + QlJ3 v (10.. To prove this result it is only necessary to display the explicit dependence of the Smatrix on photon polarization vectors and propagators Spa ex::. (IO. • The Slates a and h are the same as ~ and fJ.5.5. Otherwise.k'1'k' 2"'. (10..1. There is a particularly important application of Eq.h) . One consequence of Eq.5. the lefthand side orEg.2) to the calculation of the photon propagator.
(q )II' (q )t!.5.(q) ~ [t!.' ( ) _ Jlyqq "t!. (q) ~ t!. takes the fonn t!.2) we have here qIlMJlv(q) = 0.~.~.5 Gauge Invariance 451 conventionally called t!.5. (10.5.. (There is an important exception in the case of broken gauge symmetry.iE][1 .16) also has no such pole.5.n(q')] . and so the pole in the complete propagator (10.5.5.5." (q )MP"(q )t!.' ( ) ~ ry" .5.qPq")n(q').5.(q)..' (q) .1) with two currents and a and It both the vacuum state.10. just as we did for scalar and spinor fields in Section 10.13) yields a complete propagator of the form t!.(q) + t!.12).(q)fl" (q )t!. Jl " q q2 ' Ie (10. (10. (10..11) From Eq.(q)I .17) where (10.!q p' q [q2 .12) On the other hand. because Il~y(q) receives contributions only from onephotonirreducible graphs. we must have qpII'P"(q) O. it is expected not to have any pole at q2 = O. (10.((q )q"q. so that q 't!. written here in a general Lorentzinvariant gauge as • ( ) ~ ry" _ ~(q2)q. (10.) In particular.(q).15) This together with Lorentz invariance tells us that Il*(q) must take the form n'pP(q) ~ (q'ryPP .' (q) ~ t!.5. (10.5.5..q.~\.5. we may express the complete photon propagator in terms of a sum fl*(q) of graphs with two external photon lines that (unlike M) are onephotonirreducible: t!..5.3. ~ (10.10) where Mpa is proportional to the matrix element (10.(I/H·q q 2' ~(q'» IE (10.IlI' is the bare photon propagator. to be discussed in Volume II.(q) ~ t!. ( ) _ q.5.l.16) Then Eq. the absence of poles at q2 = a in the qJlqv term in fl~l'(q) tells us that the function n(q2) in Eq.~..!q2 L.13) or in other words t!.14) Then in order to satisfy Eq. (10. "2 2 (10. (q) + t!..(q )II' (q )t!. and t!.(q) + .II'(ql]1 + t!.18) Now.".17) is . (10.(q)II'pe(q)t!.(q) + t!.e.
1) . this sum is given by the one~ particle matrix element of the electromagnetic current JI'(x).19) This condition leads to a detennination of the electromagnetic field renor· malization constant Z).5. indicating that radiative corrections do not give the photon a mass.17). Eq.6. According to the theorem of Section 6. It follows that Z3 ~ 1 + "wop(O) . Incidentally. so n(O) ~ 0.20) The function rr(q2) in the onephotanirreducible amplitude is then + nwop( q2) . the one~particle matrix element of the electromagnetic current takes the form ('PP'. but to all orders in all other interactions (including electromagnetic) of our particle. Recall that when expressed in terms of the renormalized field (10. The current conservation condition 01. both on the mass shell.6.17).0 the gauge term in the photon propagator is altered by radiative corrections.6 Electromagnetic Form Factors and Magnetic Moment Suppose that we want to calculate the scattering of a particle by an external electromagnetic field (or by the electromagnetic field of another particle). (l0." J'(x)'P". for which ~ = ~ = 1 for all q2. the electrodynamic Lagrangian takes the form 5e = . plus a photon line.Z3 +2 M (\{It.iZ3(CjJAv . (10. (10. (10.5. ('PP'.18) shows that for ql of=.ovA/l)(8JlAV . Let us see what governs the general form of this matrix element.2) (10. radiative corrections should also not alter the gaugeinvariant part of the residue of the photon pole in Eq.5. 10. The one exception is the case of Landau gauge.. x)('PP'. where lrLOOP is the contribution of loop diagrams.•" JI'(O)'P.5.iZWtA/l]\{Ir) .4. According to spacetime translation invariance. (10. we just calculate the loop contributions and subtract a constant in order to make rr(O) vanish.21) In practice.452 10 NonPerturbative Methods still at q2 = 0.p'). to first order in this electromagnetic field. For this purpose. we need to know the sum of the contributions of all Feynman diagrams with one incoming and one outgoing particle line. [0/1.•) ..p)..4.•. (10. which may be on or off the mass shell. For a renormalized electromagnetic field.• ) ~ O.J'(O)'P.aVAil) n(q') ~ 1.}ll = 0 then requires (p' .) ~ exp(i(p .5.
10.4.6. or equivalently of k' ..6.) Obviously. (10.3) where q is the particle charge. (We have extracted a factor of the charge q of the particle from .I for future convenience.8) The function F(k 2 ) is called the electromagnetic form factor of the particle.p) ~ (p' + p)'F(k 2 ) + i(p'  p)'H(k 2 ). (10.6. while (pi _ p)2 = k 2 is not generally zero. the most general such fourvector function takes the form of a linear combination of p'll and pll. or equivalently of pili + pll and pill . Spin Zero For spin zero.p) is a fourvector function of the two fourvectors pll and pI'.11J. and comparing with Eq. 1. F(O) ~ 1.p)" = JIl(p.2p' p' .6. (pi + p) vanishes. so that both F(k 2 ) and H(k 2 ) are real.(P'. (10. we find that = 0 in Eq.6 Electromagnetic Form Factors 453 Also.(p . (10. We also have at our disposal the constraints on the current matrix elements imposed by Lorentz invariance. Now (pi _ p). we will limit ourselves to the simplest cases: spin zero and spin The analysis presented here provides an example of techniques that are useful for other currents.6. and . p). so the scalar variables that can be formed from pi' and p'll can be taken as functions only of p' pi.4) where pO and p'O are the massshell energies (pO = J p .p').pI'. such as those of the semiIeptonic weak interactions. To explore these.5) Thus the function fJl(p'. setting pi = P and J.8).4).) ~ (2n)3 qoo'o.7) Also. setting /1 = 0 and integrating over all x gives ('1'". with scalar coefficients. (10.p')2 ~ _2m 2 . so the condition of current conservation is simply H(k2)~0. (10.6." Q'I'." JO(O)'I'"o)' Using Eq. But the scalars p2 and p'2 are fixed at the values p2 = p'2 = _m 2. (10.. .6.!'(p'. this gives ('1"'0" JO(O)'I'".6.3).11i(p'.l (10.) ~ q(2n )J(2p'O)1/2(2 po)1/2 J"(p'.p) must take the form . (10..2 + m 2).) ~ (2n)3 0 3(p_p')('1'".6) The fact that JJl is Hennitian implies that . Lorentz invariance requires the oneparticle matrix element of the current to take the general form ('1"" J'(O)'I'.
fer]. which may be replaced respectively with impl'. The [P./ /' vi' ) .ptl. (i p+m)u(p. P. and 'Iv.6. imp~. which may be replaced with 2m 2 + 2p' p' = _k 2. We conclude that..6. a) (10. we can drop· all but the first three entries: pP.+2p'p'. which are the same as tenns already on our list. Lorentz invariance requires the oneparticle matrix element of the current to take the general form ('I' p'.5). p'l'.a)~O. a linear combination of tenns already on our list.454 10 NonPerturbative Methods Spin! For spin ~. J"(O)'P p. the quantity (to. and yI'. a')P(p'. lJ. which may be replacell with 2im'f~ same applies to [)'~. p'~.1=2'/ IiV"l 2p~. pill P.a). [p. pP [Yp. lJP'" "51' P EP/lv(Jp vp' r (J NONE with the coefficient of each term a function of the only scalar variable in the problem.)~}" + i'"/ l  y" y~l  "l}'~ y" . rp. we may expand pi in the 16 covariant matrices 1.l can therefore be written as a linear combination of 1 "If! pll . Just as for any 4 x 4 matrix. Also. and j.. amI p'~ p'. The most general rourvector rJ.o'. to lleal with the last term we may use the relation 'f5l'p c'l"q IIp'jl give = ~j (). .a') (i l +m) ~ 0. and u is the usual Dirac coefficient function. and p'll. We have extracted a factor iq to make the normalization of P' the same as in the previous section. '{'srI" and Ys.a')[y"F(k') _i (p 2m + p')"G(k2) + (p  p')' H(k')]u(p.9) where pt is a fourvector 4 x 4 matrix function of pl'. on the fennion mass shell.p)u(p. lJ. [Yp.1pJ 15Yp }'s [i'. and p~ and then moving all . . Contracting this with p.10) • This is obvious for the terms p~" p'~ p. which we choose to write as u(p'. pill Ii' pJ. liP". imp'~. Finally. pl'l.a')P(p'. pi' 1'. pill yJi. Also.o) ~ iq (2rr) 3 u(p'. and y~. p' factors to the . =2}'~ '2p~. r tl may be expressed as a linear combination ofytl.a) ~ u(p'. [1"."p'~. [p.Il. 2m (10."'~ + y~. factors to the right and left again gives a linear combination of pI'.p'I'. we can write b'~. Iknce the terms [" I'Jp~ and [" nothing new.6. iJ =21' . In consequence.+(" il=2p' . This can be greatly simplified by using the Dirac equations satisfied by u and Q: u(p'. p)u(p.
10).11) so that F(k 2 ). The conservation condition (10.') [y'F(O) . (10." [F(O) + G(O)] . On the other hand. letting pi + P in Eqs. /J (p' .2) is automatically satisfied by the first two terms in Eq.6. (/)u(p. we find ('I'"o'J'(O)'I'"o) ~ j q(2rr)3 u(p.p).10..") ~ q(2rr)3(p'lpQ)"".14) It may be useful to note that the electromagnetic vertex matrix rl' is commonly written in terms of two other matrices. =  u(p. (10.rr ' ) ti [ylt.rr) = u(p'.6. we may rewrite the matrix appearing in the second tenn in terms of those used in defining F(k 2 ) and G(k 2 ): u(p'. (10.16) .6.rr ' ) [yt<F 1{k 2 ) + ji [y". and H(k 2 ) must all be real functions of k 2 .p)u(p.9) and (10.6.) ..).6. (10. rr) . we also have ..UP+ m)] and (p' _p)' (p' +p) ~ p" _p2.)..") [i(p" pi] u(p.6.F. m .p). Recall also that and therefore ('1"'"" J"(O)'I'. i rJ + m} = 2m')'I' + 2ipJi.3) yields the normalization condition + G(O) ~ 1 .12) Also.p)2 does not in general vanish.~ P'G(O)] u(p. rr) ~ u(p. .p') . because (p' .13) Comparing this with Eq. G(k 2 ). so current conservation requires the third term to vanish H(k 2 ) ~ O.) + 1"')+ 2my"] u(p."') [ih'+ ji{y'. Using the identity {I'I'.6.6.10). (10.6. ~ u(p'. as u(pl. (pi . (10. (10.u) ~u(p'. (J')Yl'u(p.p)P ~ P(p.15) As already mentioned. p')iy' p+ ji{y"..yI' ~ i [U II' + m) . ''/] (p'  p)v u(p.6.6 Electromagnetic Form Factors 455 The hermiticity ofJI'(O) implies that ppt(p'. F(O) (10.rr')P(p'. (10.6.(k')] u(p.
.o) ia i.F(k 2)] u(p.6.20) as u(O. (f IT a x e'(pp)'(J(j))o'o . H''I'p. we find F(k') ~ FI(e) + 2mF2(k2) G(k 2) ~ 2m F2(k 2). B bJ(p J p) . . let us consider the spatial part of the vertex function in the case of small momenta.6.19) and (5.17) (10.6. (10. the matrix element of the interaction Hamiltonian is I qF(O) el 3 I ('I'p. (10.p)u(p.22) The magnetic moment Jl for an arbitrary particle of general spin j is defined by the statement that the matrix element of the interaction of the particle with a weak static slowly varying magnetic field is ('I'P. it is useful to use Eq.H'I'•• )~.21) where B = VxA is the magnetic field. I I "(p.4.15) with Eq.10) (with H = 0) in a third form: "(p'. Hence to 1 I el [(pp) x J' Jo'oF(O)..m (J'l. tT') [). a) = 0.6. (10.a ' u(O. a' )[yi.p)u(p.6.a ) p. (10.I d3 x J(x)' A(x) between oneparticle states of small momentum is therefore ~ ('I'. (10. ~ iqF(O) Jd 3 x m(2n")3 ~ ~~~i3 .20) m In a very weak timeindependent external vector potential A(x) the matrix element of the interaction Hamiltonian H' = .6. is the angular momentum matrix [or spin where J(!l = first order in the small momenta.ilpp)"A(x)' [(pp') x J(jl] .6. 0') [(p  + p')'{F(k')+ G(k 2)) (10.'o·Bb (pp). In order to evaluate the magnetic moment of our particle in terms of its form factors.6.6.6.19) ~[yP'Y'](p' _p). 0) ~ 2~ "(p'.6.o I )r(p.18) The normalization condition (10. yO]u(O. Ipl.J3 .23) . yj] u(O.o). For zero momenta the matrix elements of the commutators of Dirac matrices are given by (5. (J) = 4iejjk (Jk i J) a'.456 10 NonPerturbative Methods Comparing Eq. (10. B(x). For this purpose. Ip'l ~ m.14) now reads FI(O)~I. Hence in the limit of a slowly varying weak magnetic field.4.a.i. (10. (10. H''I' P•• ) ~  P (JIj))o'•.16) to rewrite Eq.10).6. o')r"(p'. (10.•'.".
7 The KallenLehmann Representation 457 Hence Eq.24) This contains as a special case the celebrated Dirac result? J.2.6. We mention without proof that the form factors F(k 2) and G(k 2) of the proton may be measured for k2 > a by comparison of experimental data for electronproton scattering with the Rosenbluth formula 8 for the laboratory frame differential crosssection: 4 2 do= e cos (012) [ l+sm '(01 2)]1 2Eo. mass m. (10. ois the scattering angle. But in the special case of a vacuum expectation value involving just two operators. dQ 4(4n)2E6 sin'(012) m x { (F(k 2 ) + G(k 2)) 2 + 4~' 2 (2F (k') tan'(O 12) + G'(k 2)) } where Eo is the energy of the incident electron (taken here with Eo» me). and spin ! as: ~ ~ qF(O) 2m (10.6. . When combined with the positivity requirements of quantum mechanics.2 that the presence of oneparticle intennediate states leads to poles in Fourier transfonns of matrix elements of timeordered products. we have a convenient representation that explicitly displays the analytic structure of the Fourier transfonn. where the two operators are the fields of elementary particles.l = q/2m for a spin! particle without radiative corrections. which mayor may not be an elementary particle field. 1 + (2Eo/m) sin 2(0 12) 10.10. this representation may be used for propagators.7 The KallenLehmann Representation· We saw in Section 10. Consider a complex scalar Heisenbergpicture operator l1>(x). this representation yields interesting bounds on the asymptotic behavior of propagators and the magnitude of renonnalization constants. The vacuum expectation • This section lies somewhat out of the book's main line of development. and k'~ 4E6sin2(OI2) . and may be omitted in a first reading.22) gives the magnetic moment of a particle of charge q. Multiparticle intermediate states lead to more complicated singularities.1). which are difficult to describe in general. In particular. like (10.
458
10 NonPerturbative Methods
value of a product ¢(x)¢t(y) may be expressed as
(<I>(x)<I>'(y»)o ~ L(OI<l>(x)ln) (nl<l>t(Y)IO) , (10.7.1)
where the sum runs over any complete set of states. (Here the sum over n includes integrals over continuous labels as well as sums over discrete labels.) Choosing these states as eigenstates of the momentum fourvector pI', translational invariance tells us that
(OI<l>(x)ln) ~ exp(ip" . x)(OI<l>(O)ln) , (nl<l>t(Y)IO) ~ exp(ip"' y)(nl<l>t(O)IO)
(10.7.2)
and so
(<I>(x)<I>t(y»o ~ Lexp(ip"' (x  y» I(OI<l>(O)ln)I' .
(10.7.3)
It is convenient to rewrite this in terms of a spectral junction. Note that 2 the sum Ln b 4 (p  p,,) I (OI11l(O)!n,1 is a scalar function of the fourvector pI', and therefore may depend only on p2 and (for p2 < 0) on the step function e(po). In fact, the intermediate states in Eq. (10.7.3) all have p2 < 0 and pO > 0, so this sum takes the Conn
(10.7.4)
with p(_p2) = 0 for p2 > O. (The factor (21T)3 is extracted from p for future convenience.) The spectral function p(~p2) = 0 is clearly real and positive. With this definition, we can rewrite Eq. (10.7.3) as
(<I>(x)<I>' (y) 1o
J ~ (2rr)~3 J 1
~ (2rr)~ 3
d'p
d'p expUp' (x  y)] e(po) p( p')
00
dp' exp[ip' (x  y)] e(pO) (10.7.5)
x p(p') o(p' + p') .
Interchanging the order of integration over p/1 and p2, this may be expressed as
(<I>(x)<I>'(y»)o
~
1
00
dp' p(p')LI+(x  y;p'),
(10.7.6)
where A+ is the familiar function
LI+(x  y; p')
= (2rr)~J
.I
d' p exp[ip . (x  y)] e(po) O(p'
+ p').
(10.7.7)
In just the same way, we can show that
(<I>t(y)<I>(x»o
~
1X
dp' p(p2) LI+(y _ X:I")
(10.7.8)
10.7 The KallenLehmann Representation with a second spectral function p(J./) defined by
459
L J'(p ,
p,) I (nl<l>' (0)10) 1
2
~ (2n)] O(pO) P( _p2) .
(10.7.9)
We now make use of the causality requirement, that the commutator [lIl(x),lIl t (y)] must vanish for spacelike separations x  y. The vacuum expectation value of the commutator is
([<I>( x), <I> t (y)])o
~
fox d~2
(p(~2) "'+(x _
y; ~2)
_ p(~2) "'+(y _ x; ~2») .
(10.7.10) As noted in Section 5.2, for x  y spacelike the function ~+(x  y) does
not vanish, but it does become even. In order for (10.7.10) to vanish for arbitrary spacelike separations, it is thus necessary that
(10.7.11)
This is a special case of the CPT theorem, proved here without the use of perturbation theory; for whatever states with p2 = _/12 have the quantum numbers of the operator 1Il, there must be corresponding states with p2 = Ii that have the quantum numbers of the operator lilt. Using Eq. (10.7.11), the vacuum expectation of the timeordered product
IS
(10.7.12)
where ~F(X  y;/12) is the Feynman propagator for a spinless particle of mass J.1: _ i"'dx  Y; ~2)  e(xo _ yO)"'+(x _ y; ~2) _ e(yO _ xO)"'+(y _ x; ~2) .
(10.7.13)
Borrowing the notation introduced in Section 10.3 for complete propagators, we introduce the momentum space runction
 i""(p)
=
J
d'x exp[ip' (x  y)] (T {<I>(x)<I>'(y) })o.
(10.7.14)
Recall that
./
d4x
exp[ip'(XY)]~F(XY;J})= P + j . l  I f. ' 2 \ '
(10.7.15)
This yields our spectral representation;9
""(p)
~
f
p(~2) P' +d;:
Ie
(10.7.16)
One immediate consequence or this result and the positivity of p(J.I2) is
460
10 NonPerturbative Methods
that d'(p) cannot vanish for Ip2 1+ Cfj faster" than the bare propagator 1/([,1 + m2  iE). From time to time the suggestion is made to include higher derivative terms in the unperturbed Lagrangian, which would make the propagator vanish faster than l/pl for Ip 2 1 + ro, but the spectral representation shows that this would necessarily entail a departure from the positivity postulates of quantum mechanics. We can use the spectral representation together with equaltime commutation relations to derive an interesting sum rule for the spectral function. If (J)(x) is a conventionally normalized (not renonnalized) canonical field operator, then
O<l>(X, t) [
ot
,<I> (y,t)
t
]
~,.
. 3
(xy).
(10.7.17)
We note that O x
a ,"A+(x 
y)
xO=yO
~ ,.
. 3(x 
y)
so the spectral representation (10.7.10) and the commutation relations (10.7.17) together tell us that
hoc p(~2)di ~ 1 .
, 1
(10.7.18)
This implies that for Ip2 1~ ro, the momentum space propagator (10.7.16) of the unrenormalized fields has the freefield asymptotic behavior
A(P)~2' p
This result is only meaningful within a suitable scheme for regulating ultraviolet divergences; in perturbation theory the unrenormalized fields have infinite matrix elements, and their propagator is illdefined. Now consider the possibility that there is a oneparticle state Ik) of mass m with a nonvanishing matrix element with the state (0111>(0). Lorentz
.. In fact, it is not even certain that !I.'(p) vanishes for III > en at all, even though this would seem to follow from the spectral representation. The problem arises from the interchange of the integrals over pi' and 1'2. What is certain is that !I.'{p) is an analytic function of _p2 with a discontinuity across the positive real axis I = JlJ. given by IIp(Jll), as can be shown by the methods of the next section. From this, it follows that Sip) is given by a dispersion relation with spectral function p(J12) and possible subtractions:
!I.'(p) = p{p2)+(_p2
+I'5}"
where n is a positive integer, Jl5 is an arbitrary positive constant, and PIp!} is a ~dependent polynomial in p2 of order n  1 that is absent for n = O.
'" (:w) (' 1
dJ12 2 )" 2+ 2 01'+1'0 P I '
i€
10.7 The KiillenLehmann Representation
invariance requires this matrix element to take the form
461
(Olol>(O)lk) ~ (2,)3/2 (2y'k2 +m2fl/2 N,
(10.7.19)
where N is a constant. According to the general results of Section 10.3, the propagator ~I(p) of the unrenormalized fields should have a pole at p2 + _m 2 with residue Z INI 2 > O. That is,
=
p(p2) ~ Zb(l m2)
+ .(p2) ,
(10.7.20)
where (T(J12) > 0 is the contribution of multiparticle states. Together with Eq. (10.7.18), this has the consequence that
I
~ Z + fooo .(p2)dp2
Z < 1
(10.7.21)
and so
(10.7.22)
with the equality reached only for a free particle, for which (OII1l(x) has no matrix elements with multiparticle states. Because Z is positive, Eq. (10.7.21) can also be regarded as providing an upper bound on the coupling of the field III to multiparticle states:
fooo .(p2) dl < 1
(10.7.23)
with the equality reached for Z = O. The limit Z = 0 has an interesting interpretation as a condition for a particle to be composite rather than elementary.lO In this context, a 'composite' particle may be understood to be one whose field does not appear in the Lagrangian. Consider such a particle, say a neutral particle of spin zero, and suppose that its quantum numbers allow it to be destroyed by an operator F('¥) constructed out of other fields. We can freely introduce a field III for this particle by adding a term to the Lagrangian density of the form t ~!£' = (11l F('¥)f, because the path integral over III can be done by setting it equal to the stationary point III = F('P), at which ~ff = O. But suppose instead we write ~!£' = .6.20 + A2}, where ~!£'o ~aJ.ll1lalll1l im211l2 is the usual freefield Lagrangian, and treat .6.21 = .6.ff  ~!£'o as an interaction. A term !o).l11l8J.l11l in the interaction is nothing new. We encountered such a term in Eq. (10.3.12), multiplied by a factor (1  Z); the only new thing is that here Z = O. Instead of adjusting Z to satisfy the field renormalization condition rl*'(O) = 0, here we must regard this as a condition on the
=
t This is known in condensed matter physics as a 'HubbardStratonovich transformation'.lI It win
be used to introdnce fields for pairs of electrons in our discussion of superconductivity in Volnme
[1
462
10 NonPerturbative Methods
coupling constants of the composite particle. Unfortunately, it has not been possible to implement this procedure in quantum field theories, because as we have seen Z = 0 means that the particle couples as strongly as possible to its constituents, and this rules out the use of perturbation theory. The condition Z = 0 does prove useful in nonrelativistic quantum mechanics; for instance, it fixes the coupling of the deuteron to the neutron and proton. 12 Although the spectral representation has been derived here only for a spinless field, it is easy to generalize these results to other fields. Indeed, in the next chapter we shall show that to order e2, the Zfactof for the electromagnetic field (conventionally called 23) is given by 23
=
1
1~:2In (~;)
(where A
(10.7.22).
::»
me
IS
an ultraviolet cutoff), in agreement with the bound
10.8
Dispersion Relations·
The failure of early attempts to apply perturbative quantum field theory to the strong and weak nuclear forces had led theorists by the late 1950s to attempt the use of the analyticity and unitarity of scattering amplitudes as a way of deriving general nonperturbative results that would not depend on any pClrticular field theory. This started with a revival of interest in dispersion relations. In its original form,13 a dispersion relation was a formula giving the real part of the index of refraction in terms of an integral over its imaginary part. It was derived from an analyticity property of the index of refraction as a function of frequency, which followed from the condition that electromagnetic signals in a medium cannot travel faster than light in a vacuum. By expressing the index of refraction in terms of the forward photon scattering amplitude, the dispersion relation could be rewritten as a formula for the real part of the forward scattering amplitude as an integral of its imaginary part, and hence via unitarity in terms of the total crosssection. One of the exciting things about this relation was that it provided an alternative to conventional perturbation theory; given the scattering amplitude to order e2 , one could calculate the crosssection and the imaginary part of the scattering amplitude to order e4 , and then use the dispersion relation to
• This section lies somewhat out of the book's main line of develupment, and may be omitted in a first reading.
10.8 Dispersion Relations
463
calculate the real part of the forward scattering amplitude to this order, without having ever to calculate a loop graph. The modern approach to dispersion relations began in 1954 with the work of GellMann, Goldberger, and Thirring. 14 Instead of considering the propagation of light in a medium, they derived the analyticity of the scattering amplitude directly from the condition of microscopic causality, which states that commutators of field operators vanish when the points at which the operators are evaluated are separated by a spacelike interval. This approach allowed Goldbergerl5 soon thereafter to derive a very useful dispersion relation for the forward pionnucleon scattering amplitude. To see how to use the principle of microscopic causality, consider the forward scattering in the laboratory frame of a massless boson of any spin on an arbitrary target :x of mass mel > 0 and Prt = O. (This has important applications to the scattering not only of photons but also pions in the limit m" = 0, to be discussed in Volume II.) By a repeated use of Eq. (10.3.4) or the LehmannSymanzikZimmerman theorem,3 the Smatrix element here is
S=
1 , I' hm" 0 1m,,' 0 (2n)3J4ww'INI' ~ ~
X
J J
a'x
d4 y e"" etk<(iDy)(iD,) (oIT{At(y), A(x)}I') , (10.8.1)
Here k and k' are the initial and final boson fourmomenta, with OJ = kG, Wi = k'O; A(x) is any Heisenbergpicture operator with a nonvanishing matrix element (VAClA(x)lk) = (2n)3j2(2w)lj 2Ne ih between the oneboson state Ik) and the vacuum; and N is the constant in this matrix element. In photon scattering A(x) would be one of the transverse components of the electromagnetic field, while for massless pion scattering it would be a pseudoscalar function of hadron fields. The differential operators iD x and iDj' are inserted to supply factors of ik '2 and ik 2 that are needed to cancel the external line boson propagators. Letting ' these operators act on At(y) and A(x), we have
S
=
1 I' 1m,' 0 I' 1m", 0 (2rr)3J4ww'INI' ~ ~
X
J J
d4 x
d4y e tk ", e"«'IT{Jt(y), J(x)}lo)
+ ETC,
(10,8,2)
DxA(x), and 'ETC' denotes the Fourier transform of equal where J(x) time commutator terms arising [rom the derivative acting on the step functions in the timeordered product. The commutators of operators like A(x) and At(y) (or their derivatives) vanish for x O = yO unless x = y, so the 'ETC' term is the Fourier transform of a differential operator acting on a4 (x  y), and is hence a polynomial function of the boson
=
464
10 NonPerturbative Methods
fourmomenta. We are concerned here with the analytic properties of the Smatrix element, so the details of this polynomial will be irrelevant. Using translation invariance, Eq. (10.8.2) gives the Smatrix element as S = 2trib 4 (k' k)M(w), where
M(w) ~ F(w) 
2wlNI'
,
F(w),
(10.8.3)
J
d"x e'wt>(aIT{Jt(O), J(xllla)
+
ETC,
(10.8.4)
it now being understood that k Jl = w{!', where t is a fixed fourvector with (Pt Ii = a and tJ = 1. The timeordered product can be rewritten in terms of commutators in two different ways:
T{Jt(O),J(x)} ~ O(xo)[Jt(O),J(x)] +J(x)J'(O)
~ O(xo)[J'(O),J(x)]
+ Jt(O)J(x).
(10.8.5)
Correspondingly, we can write
(10.8.6)
where
FA(w)
h(w)  
J J
d'x O(xo) (al[Jt(O), J(x)]I,) e'wf'
+
+
ETC,
(10.8.7)
d'x O(xo) (al[Jt(O), J(x)]la) e'wh
ETC,
(10.8.8)
F+(w)
=
L(w) 
J J
d'x (a1J(x)1t(0)I') e'wf',
(10.8.9)
d'x (aIJt(O)J(x)la) e'wi, .
(10.8.10)
Microscopic causality tells us that the integrands in (10.8.7) and (to.8.8) vanish unless xII is within the light cone, and the step functions then require that xli is in the backward light cone in (10.8.7), so that x' t > 0, and in the forward light cone in Eq. (10.8.8), so that x·t < O. We conclude that FA(w) is analytic for Imw > 0 and FR(W) is analytic for Imw < 0, because in both cases the factor eiwh provides a cutoff for the integral over Xli. (Recall that the 'ETC' term is a polynomial, and hence analytic at all finite points.) We may then define a function
Imw>O Imw <0
(10.8.11)
10.8 Dispersion Relations
465
which is analytic in the whole complex w plane, except for a cut on the real axis. We can now derive the dispersion relation. According to Eq. (10.8.6), the discontinuity of 9"(w) across the cut at any real E is
If §"(w)/w ll vanishes as Iwl _ co in the upper or lower half·plane, then by dividing by any polynomial P(w) of order n we obtain a function that vanishes for Iwl _ co and is analytic except for the cut on the real axis and poles at the zeroes W v of P(w). (Where .~(w) itself vanishes as Iwl  co, we can take P(w) = 1.) According to the method of residues, we then have
9'(w)
P(w)
L
9'(w,)
1 ,[
+
v (w v  w)PI(w,,) = 2rr:i
Jc
.!F(z) dz (z  w) P(z) ,
(10.8.13)
where w is any point off the real axis, and C is a contour consisting of two segments: one running just above the real axis from co + iE to +00 + iE and then around a large semicircle back to co + iE, and the other just below the real axis from +co  iE to x;  iE and then around a large semicircle back to +co  iE. Because the function Y(z)/P(z) vanishes for Izi  co, we can neglect the contribution from the large semicircles.
Using Eq. (10.8.12), Eq. (10.8.13) becomes
9'(w)
~ Q(w) + P(w) 1+~ F
2ni
_~
(E)  F~(~) dE , (E w) P E
(10.8.14)
where Q(w) is the (n l)thorder polynomial
Q(w)P(W)L(
"
Y(w v)
OJ" (f)
)P'()' w"
A dispersion relation of this form, with P(OJ) and Q(w) of order nand n 1 respectively, is said to have n subtractions. If we can take P = 1 then Q = 0, and the dispersion relation is said to be unsubtracted. If we now let w approach the real axis from above, Eq. (10.8.14) gives
FA(W)~Q(w)+ P(~)l+~
2m
_~
L(E)F+(E) dE. (E  w  Ie) PtE)
(10.8.15)
Recalling Eqs. (10.8.6) and (3.1.25), this is
1 F(w) ~ Q(w)+ :/_(w)
oc 1 P(w) F_(E)  F+(E) F+(w) + 2ni _~ (E _ w) PtE) dE +2 (10.8.16) with 1/(E OJ) now interpreted as the principal value function 9/(E OJ).
l+
466
10 NonPerturbative Methods
This result is useful because the functions F±(E) may be expressed in terms of measurable crosssections. Summing over a complete set of
in Eqs. (10.8.9) and (10.8.10) (including integrations over the momenta of the particles in p) and using translation multiparticle intermediate states
f3
invariance again, we have
F+(E)
~ (2n)' 2: I(PIJ(OJII')I' b'(p, + Et + PI),
p
(10.8.17)
F_(E) ~ (2n)'2: fi
I(PIJ(O)I,)I' J'(p, + Et 
PI)'
(10.8.18)
But the matrix elements for the absorption of the massless scalar boson B in B + Cl + fJ or its antiparticle W in Be + Cl  f3 are
_
2inMB'+'~" ~ ,,,
(2n)4 (PI1'(O)I') (2n)3/2.j2EB'N '
(10.8.19)
 2inM B+'_1 ~ (2n)3/2 .J2EBN (PIJ(O)I.) .
(2n)'
(10.8.20)
Comparing with Eq. (3.4.15), we see that F±(E) may be expressed in terms of total crosssections" at energies +E:
F+(E) ~ O(E) (2n)3
21EIINI'
rr,+B,(IEI)"
(10.8.21)
2EINI 2 L(E) ~ OlE) J .,+B(E). (2n)
(10.8.22)
The scattering amplitude (10.8.3) is now, for real w > 0,
M(w) ~
iQ(w)
_
2wlNI2 P(w)4 w(2n)
i 2(2n)3"'+B(W)
Jo
r [
(E
"'+B(E) w)P(E)
+
",+B,(E) ] E dE. (10.8.23) (E +w)P( E)
It is more usual to express this dispersion relation in terms of the amplitude f(w) for forward scattering in the laboratory frame, defined
so that the laboratory frame differential crosssection in the forward direction is 1/(w)1 2 . This amplitude is given in terms of M(w) by
•• In some cases where selection rules allow the transition ::t: + ::t: + Band ::t: + ::t: + 8', the functions F+(E) also contain terms proportional to ,1(E) arising from the contribution of the oneparlide state C1. in the ,urn over intermeliiate states p, This docs not occur for transversely polarized photons, or for pseudoscalar pions in the limit mIT + O.
10.8 Dispersion Relations
f(w) ~ 4rr 2w M(w) ~ 2rr'iF(w)/INI 2, so Eq. (10.8.23) now reads
467
f(w) = R(w)
+ 4n (T:x+B(W)
IW
+
P(w)
4.'
roo [a'+B(E)
Jo (E w)P(E)
+
a,+B.(E)] EdE (E + w)P(E) ,
where R(w) ~2in2 Q(w)/INI 2 . The optical theorem (3.6.4) tells us that the second term on the righthand side equals Hm/(w), so this can just as well be written in the more conventional form
Ref(w)
~
R(w) + P(w)
4rr 2
r [(Ew)P(E) + (E+w)P(E) ] E dE, (10.8.24) a,+B(E) "HB'(E) Jo
In particular, we see that R( w) is real if we choose P (w) real. The forward scattering amplitude also satisfies an important symmetry condition. By changing the integration variable in Eqs. (10.8.7) <lnd (10.8.8) from x to x and then using the translationinvariance property
(,llJt(O), J(x)]lal ~ ('llJt(x), J(O)]I')
we see that for Imw < 0, FA(W) is the same as FR(W), except for an interchange of J with Jt. That is,
FA(w) = FR{w)
for
Imw <0,
where a superscript c indicates that the amplitude refers to the scattering of the antiparticle Be on 0:. (We leave it to the reader to show that this relation is not upset by the equaltime commutator terms in Eqs. (10.8.7) and (10.8.8).) In the same way, we find
FR(W)
=
F:4(w)
for
Imw>O,
and for real
OJ
F±(w) ~ Ff(w).
Using these relations in (10.8.6), and recalling th<lt /(w) is proportional to F(w), we find the crossing symmetry relation, that for real w
/(w)
~
f"(w).
(10.8.25)
We are free to take P(w) as any polynomial of sufficiently high order, but R(w) then depends not only on P(w) but also.on the values of y(w) at the zeroes of P(w). For pew) real and of nth order, the only free parameters in Eq. (10.8.16) are the n real coefficients in the real (n 1)th order polynomial R(w). Hence Eq. (10.8.16) contains just n unknown real independent constants, the coefficients in the polynomial R(w) for a given P(w). We therefore wish to take the order n of the otherwise arbitrary polyomial P(w) to be as small as possible.
468
10 NonPerturbative Methods
We might try taking P(w) = 1, but this doesn't work. The analysis of Section 3.7 suggests that the forward scattering amplitude should grow like OJ or perhaps as fast as w 1n 2 w. In this case for f(w)/ P(w) to vanish as ill 10 0, it is sufficient to take P(w) as a second order polynomial, so that R(w) is linear in w. Choosing P(E) = £2 for convenience, Eq. (10.8.24) then becomes
Ref(w)~a+bw
+
w
2
4rr 2 /0
r;. [(T:t+B(E) + (T:x+w(E)] If' dE (E (E +
w) w)
(10.8.26)
with a and b unknown real constants. The crossing symmetry condition (10.8.25) tells us that the corresponding constants in the dispersion relation for the antiparticle scattering amplitude r(w) are
d' = a,
be
=
b.
(to.8.27)
If we assume for instance that the crosssections O"a+B(E) and (T:l+w(E) behave for E + CD as different constants times (In E)', then (10.8.26) would give
Re f(w)  [aO+B(w)  "O+B' (w)] In w  w(ln W)'+1
(10.8.28)
so the real part of the scattering amplitude would grow faster than the imaginary part by a factor In w. This is implausible; we saw in Section 3.7 that the real part of the forward scattering amplitude is expected to become much smaller than the imaginary part for (JJ + CD, as confirmed by experiment. We conclude that if O"Cl+B(E) and O"Cl+Bc(E) do behave for E + CD as constants times (In E)' then the constants must be the same. Because we are concerned here with the highenergy limit, this result does not depend on the assumption that B is a massless boson, so in the same sense, the ratio of the crosssections of any particle and its antiparticle on a fixed target should approach unity at high energy. This result is a somewhat generalized version of what is known as Pomeranchuk's theorem. 16 (Pomeranchuk considered only the case r = 0, while Section 3.7 and the observed behavior of crosssections both suggest that r = 2 is more likely.) Although Pomeranchuk took his estimates of the asymptotic behavior of scattering amplitudes from arguments like those of Section 3.7, today high energy behavior is usually inferred from Regge pole theoryP It would take us too far from our subject to go into details about this; suffice it to say that for hadronic processes the asymptotic behavior of f(w) as w goes to infinity is a sum over terms proportional to WCl"{oJ, where Ct:n(t) are a set of 'Regge trajectories', each representing the exchange of an infinite family of different onehadron states in the collision process. The leading trajectory (actually, a complex of many trajectories) in hadron
Problems
469
hadron scattering is the 'Pomeron,' for which a(O) is close to unity. It is this trajectory that gives crosssections that are approximately constant for E _ ro. According to Pomeranchuk's theorem, the Pomeron couples equally to any hadron and its antiparticle. We can estimate an(O) for the lower Regge trajectories from the spectrum of hadronic states. A necessary though not sufficient condition l8 for a mesonic resonance of spin j to occur at a mass m is that m2 equals the value of t where one of trajectories an(t) equals j. Apart from the Pomeron, the leading trajectory in pionnucleon scattering is that on which we find the j = 1 P meson at m = 770 MeV, the j = 3 g meson at m = 1690 MeV, and a j = 5 meson at m = 2350 MeV. Extrapolating these values of a(t) down to t = 0, we can estimate that this trajectory has a(O) ~ 0.5. This trajectory couples with opposite sign to 11:+ and 11:, so for pionnucleon scattering we expect JIW)  J'(w) to behave roughly like )05. For photon scattering there is no distinction between B and Be, so here Eq. (10.8.27) gives b ~ 0, and Eq. (10.8.26) reads
J(w)~a+,
w21.~
()
211:
E 
.(E) , ,dE.
OJ
(10.8.29)
This is essentially the original KramersKronig 13 relation. As we shall see in Section 13.5, for a target of charge e and mass m the constant a has the known value Re j(O) = e2 jm.
Problems
1. Consider a neutral vector field v",(x). What conditions must be imposed on the sum rr;,,(k) of oneparticleirreducible graphs with two external vector field lines in order that the field should be properly renormalized and describe a particle of renormalized mass m? How do we split the freefield and interacting terms in the Lagrangian to achieve this?
2. Derive the generalized Ward identity that governs the electromagnetic vertex function of a charged scalar field.
3. What is the most general form of the matrix element(p20"2Ij!'(x)lpuTd
of the electromagnetic current j!'(x) between two spin! oneparticle states of different masses ml and m2 and equal parity? What if the parities were opposite? (Assume parity conservation throughout.) 4. Derive the spectral (KallenLehmann) representation for the vacuum expectation value (T{j!'(x) JV(y)t})o. where j!'(x) is a complex conserved current.
125 (1937). 10. Use dispersion theory and the results of Section 8. Phys. Y. 1. Rev. Without using any assumptions about the asymptotic behavior of the scattering amplitude or crosssections. H. Phys. Takahashi. Dakl. 4. Furry. M. 5. K. Howard and B. 342 (1954). Zimmerman. NUODO Omenta 1. Sf}/). 8. P. L. A. Aaron. J. Derive the spectral (KallenLehmann) representation for a complex scalar field by using the methods of dispersion theory. S. M. 683. 205 (1955). Proc. 1972). Proc. HelD.466 (1960). Lehmann. Stratonovich.6. Rev. N. 2. Lett.7 to calculate the amplitude for forward photonelectron scattering in the electron rest frame to order e4 . R. 1962): p. Rev. Rev. Amado. J. 51. W. 7. 416 (1957). 182 (1950).48 (1935). Rev.Math. Ser. 2. where rp(x) is a Dirac field. NuoDa Omenta. (Landon) A117. 6. C. Hubbard. D. Derive the spectral (Kal1en~Lehmann) representation for the vacuum expectation value (T{tpn(x) V\1!(Y)})o. 296 (1959). 78. Nuovo Omenta 11. 6. Jouvet. 610 (1928). Rev. 3. Acta 25. 1254. 77 (1959). 615 (1950). Lett. Vaughan. C. H. 417 (1952). Kallen. Yukawa. G. 1258 (1961). and R. H. 137. Quantum Electrodynamics (SpringerVerlag. M. Roy. Soc. Symanzik. Ward. Rosenbluth. Japan 17. B672 (1965). Phys. Phys. 79. 9. in Proceedings of the 1962 HighEnergy Conference at CERN (CERN. 7. Weinberg. Schwinger. 12. 8. 3. Weinberg. Phys.470 10 NonPerturbative Methods 5. Phys. R. 10. Geneva. H. J. J. References 1. show that it is impossible for forward photon scattering amplitudes to satisfy unsubtracted dispersion relations. Berlin. Rev. Phys. Lehmann. Soc. Phys. 370 (1957). 11. Phys. 3. Phys. and W. Dirac. Nuova Omenta 18. . S.
979 (1955). Intern. B. Regge. by G. The nonperturbative nature of this result was shown by M. Lett. 18. 1961). and W. Cambridge. 124. Kunde 9. by C.. Nuovo Omenta 14. 17. P. Expt. GellMann. Toll. M. Thesis 1952). Fisici. 15. Amsterdam. Phys. 95.g. 1612 (1954). ed. Goldberger. Goldberger in Dispersion Relations and Elementary Particles. Ned. Kramers. The graph of spin versus squared mass is known as a ChewFrautschi plot. Thirring. Rev. Tyd. 947 (1960). reprinted in H. De Witt and R. 499 (1958). de Kronig. 16. Jackson in Dispersion Relations. 1956). English version: Soviet Physics . 41 (1962). Nat.) 34. 97. R. L. For historical reviews. (USSR. Phys. An Introduction to Regge Theory and High Energy Physics (Cambridge University Press. A. Physica 12. 14. L. For a generalization.References 471 13. The Dispersion Relation for Light and its Application to Problems Involving Electron Pairs (Princeton University Ph. Atti Congr. Phys. M. 1927). Rev. Goldberger. Omnes (Hermann. Collected Scientific Papers (NorthHolland. The original references are T. e. 1960). Weinberg. see G. M. C. J. F. see S. L. R. Phys. Bologna. Paris.2049 (1961). Goldberger. Theor. Frautschi. 725 (1958). D. Kramers. 402 (1942). Rev. D. 99.951 (1959): 18. See. J. M. 1977). Chew and S. H. D. 543 (1946). 508 (1955). ed. . L lao Pomeranchuk. Como (Nicolo ZanichelHi. see J. A. 8. Phys. Collins. S. Edinburgh. Phys. Rev. L. Rev. Screaton (Oliver and Boyd.JETP 34(7).
we will move on to the calculation of the vacuum polarization in Section 11. and both the dimensional regularization of 't Hooft and Veltman and the older regularization method of Pauli and Villars.1 Counterterms The Lagrangian density for electrons and photons is taken in the fonn· S!' =  ~F~v FB !IV VJB [YII [ail + ieBA~] + mB] tpB (11. After some generalities in Section 11.11 OneLoop Radiative Corrections in Quantum Electrodynamics In this chapter we shall proceed to carry out some of the classic oneloop calculations in the theory of charged leptons . m which the an upper case A and a lower case Ij! are used to denote the photon and charged particle fields.and interactionpicture operators. we will introduce a number of the mathematical techniques that prove useful in such calculations. Wick rotation.3.' though most of OUf calculations will apply equally to muons and tauons. unrenormalized) fields of the photon and electron.avA~. including the use of Feynman parameters.1. 1 11. For definiteness we shall refer to the charged particles in our calculations here as 'electrons. and the electron selfenergy in Section 11. and ~eB and mB are the bare charge and • In this chapter we will not be making transformations between Heisenberg. Along the way.1.4.a!l AiJ . Although we shall encounter infinities. In the next chapter we shall extend what we have learned here about renormalization to general theories in arbitrary orders of perturbation theory. There are three known species or 'flavors' of leptons: the electron and muon..1) where F~v . so we shall return to a conventional notation. A~ and lpB are the bare (i. respectively. the anomalous magnetic moment of the electron in Section 11.e. and the heavier. more recently discovered tauon. it will be seen that the final results are finite if expressed in terms of the renormalized charge and mass. 472 .massive spin particles that interact only with the electromagnetic field.2.
7) (11.2) (11.1. 11.1.6miiJ ~ + m]tp (11. It will turn out that all of the terms in 2 2 are of second order and higher order in e.5) ~ A ' .1.1.ie(Z2 1)A.1. and that these terms just suffice to cancel the ultraviolet divergences that arise from loop graphs. (11.3) (11.1.11. the socalled vacuum polarization effect. Z3. and Om adjusted so that the propagators of the renonnalized fields have poles in the same position and with the same residues as the propagators of the free fields in the absence of interactions.1. The Lagrangian may then be written in terms of renonnalized quantities.ZI/2A .4) (11.8) and 22 is a sum of 'counterterms' 22 = .1.2 Vacuum Polarization We now begin our first calculation of a radiative correction involving loop graphs. Vacuum polarization produces measurable shifts in the energy levels of hydrogen. 3 B' e = Z3 _ +1/2 eB.(Z2 l)ip [Ytt (lll +Z. the calculation of the vacuum polarization provides a key .9) . and makes an important correction to the energies of muons bound in atomic orbits around heavy nuclei. as we shall see in Volume II. as (11. with the constants Z2. we introduce renormalized fields and charge and mass: _ 1/2 ~ ~ Z. As described in the previous chapter.HZ3 .iiJ y'~ .6) where (11. consisting of the corrections to the propagator associated with an internal photon line. Also.2 Vacuum Polarization 473 mass of the electron. V'B.l)pv PI/V .
IlV(q) is the photon propagator without radiative corrections. Here wavy lines represent photons.2. not including photon propagators for the two external lines. 1 We use the elementary fonnula 1 il dx AB ~ Jo [(1x)A+xB]' (11. In lowest order there is a oneloop contribution to n°.5. element in the calculation of the high energy behavior of electrodynamics and other gauge theories.1: i(2n)4n~~~op(q) =  J tpTr {[(2~.2.i4 it d)2+ ] [(2rr)4 ey "]} (n) (pq) +m .5.1) where L~.dl+ m] yO) lloopq (2n)4 p (p2+m2id((p~q)2+m2_iE) . More simply.474 11 OneLoop Radiative Corrections p p ~ q Figure 11. with the first minus sign on the right required by the presence of a fermion loop.2) m x [(2rr) 4ey l'] [2. Our task here is to calculate the leading contributions to n Op<1(q).4) 0'1'" ( ) ~ _ie 2 Ja' . we define i(2n:)4nO Il<1(q) as the sum of all connected graphs with two external photon lines with polarization indices }1 and v and carrying fourmomentum q into and out of the diagram.1. As in Section 10. (1123) The first step in doing this integral is to use a trick introduced by Feynman.1.13): !j. this is Tr {[i 1 + m]y' [i(1 .I f ' 1._miE ] (1l.'~LI[IO'Llrl. lines carrying arrows represent electrons. corresponding to the diagram in Figure 11.4 p2~:. (1l. (l 0. The complete photon propagator il'lll' (q) is given by Eq. The oneloop diagram for the vacuum polarization in quantum electrodynamics. and with the asterisk indicating that we exclude diagrams that can be disconnected by cutting through some internal photon line.
this step is only valid in convergent integrals.5) Using the results of the Appendix to Chapter 8. the trace here can easily be calculated as Tr {[i(j/ + qx) + mlr P [i U  q(1 .x»qPO (11. i.xl)" + (p + qx)' (p  (p+qx)"(pq(lx))P+m'qP"].3) as (p2 r' + m2 . (See Figure 11.iE + q2 x (1  x)r 2 x Tr {[i(j/ + qx) + miY P [i U  q(1 . Our next step is called a Wick rotation.) We can rotate the contour of integrations of po counterclockwise withollt crossing either of these poles. (11.q(l.3) becomes rr?~op(q) = (.iE) . In principle. with p4 again integrated over real values from 00 to • Strictly speaking.2 Vacuum Polarization 475 to write the product of scalar propagators in Eq.2.2.11.f .:~: . so that Eq. with p4 integrated over real values from a:.e..I dx trp[p2 +m 2 .2. (This is a special case of a class of integrals given in the Appendix to this chapter.10 2 + ((p _ q)2 + m2 _ iE)Xr dx = [(p2+ m'i<)(Ix) 10 10 1 [p2+m2ie_2p'qx+q2xr2dx 1 = [(pqx)2+m2iE+q2x(1x)r2dx. such as the dimensional regularization scheme discussed below. (11. to +00.(p +qx)P (p . we should introduce some regulator scheme to make all integrals converge.5) are at po = + Vp 2 + m2 + q2 x (1 x) iE. then we would have been setting pO = _ip4.2. in order to justify the ~hift of variable~.2.x») + mlr"J.) We can now shift the variable of integration in momentum space· p~p+qx.q)2 + m2 . so that instead of integrating pO on the real axis from 00 to +::1).2.iE)((p1. (If an if" instead of it: had appeared in the denominator of the propagator. just above the negative real axis and just below the positive real axis. (11. so the poles in the integrand of Eq. the quantity m2 + q2 x (l_ x) is positive for all x between a and 1.6) ~ 4[ . 2 As long as _q2 < 4m 2.x») + miY"} q(1 . . (11. we integrate it on the imaginary axis from joo to +i::c. That is. we can write pO = ip4.
q(l.x» ~P" (11. Small x's mark the poles in the pO complex plane. and the Ward identity (10. from the real to the imaginary pOaxes.25) shows that in order to maintain .~~op(q) = (~:~4 10 X [  1 dx J 4 (d p)E [p2 + m2 + q2 x (1.2.) Eg.~~p(q).5) now becomes rr. The integral (11.4. but to see this it is necessary at intermediate stages of the calculation to use some sort of regularization technique that makes the integrals finite. It would not do simply to cut off the integrals at some maximum momentum A.x))" + (p + qx)' (p  q(1 . +00.x)]2 (p + qx)P(p .2. because this would amount to introducing a step function O(A 2 .476 11 One~Loop Radiative Corrections x x Figure 11. 4. the arrow indicates the direction of rotation of the contour of integration. 2. with the indices running over 1. or as the usual Minkowski tensor. f/ PI1 can be taken as either the Kronecker delta.2. 3. integrating only over pi' with p2 < A2. Also. Eventually all infinities will cancel. Wick rotation of the pO contour of integration.2. The effect would be a change of sign of rr. (11. O.7) (p+qx)"(pq(1xlJP+ m2~P"]. with the indices running over 1. where and all scalar products are evaluated using the Euclidean norm a' b = alb! + a 2b2 + a 3b3 + a4b4 with the understanding that q4 = _iqo .p2) into the electron propagator.7) is badly divergent. 2. 3.
jpi. v.x)]' +q )tl pa + m2'1pa] . This amounts to carrying out angular averages in integrals like (11.7). the infinities then reappearing as factors (d . while the factors may be found by requiring that botb sides give the same result when contracted with ljs.(2n)4 Jo dx Jo K dK K [2 + m2' x(1. based on a continuation from four to an arbitrary number d of spacetime dimensions.11) fax Kd+'[K2+v212dK~ l(v2)Hr(l+dI2)r(ldI2) (11.2.2. .2. (11. (11. radiative corrections would induce a photon mass. dimensional regularization gives 2 . etc.4)1. If fact.2..2. with an ordinary cutoff A.7) by dropping all terms that are odd in p.2. For the integral (11. any modification of the electron propagator must be accompanied with a modification of the electronphoton vertex. aside from the even integers).2. a clear violation of the requirements of gauge invariance. p.7) now converges for complex spacetime dimensionality d.2 Vacuum Polarization 477 gauge invariance. We use the wellknown formulas (given in greater generality in the Appendix to this chapter): fow I<"I[K' + V'r'dK ~ l(v')~2 r (dI2) r (2 dI2) . x [ ~K2 '7pa +2q Pqa x(1x)+ (K 2 _ q2 x (l_x) The integrals over K can be carried out for any complex d (or for any real d.2.12) •• These expressions may most easily be derived hy noting that their form is dictated by Lorentz invariance and the symmetry among the indices J1.9) Also. We can continue the integral through complex dvalues to d = 4.10) The integral (11. where K . and Qd is the area of a unit sphere in d dimensions (11. (11.8) Vpv pP pU + (p2)2 [f//'v pa tl + '1JJP'I va + 'Ilia f/ VP] / d(d + 2) . and replacing the terms that have even numbers of pfactors with" vpv + p2t1llv / d.pa _ 4e Qd [I fer) dl n11oop(q) .11. the volume element d4pE is to be replaced with QdKd1dK. after writing the integrand in this way as a function only of p2. Experience has shown that the most convenient method for regulating divergent integrals without impairing gauge invariance is the dimensional regularization technique introduced by 't Hooft and Veltman 3 in 1972.
2.d12) in EQ.2.1)F1 Pll' in the .2.21d) r (1+ d12) r (1.13) blows up for d + 4.oJo dx x(1 .16) with n(q ) ~  2 4e20d (2n)4 2 r r( 0 r (2. It was precisely to achieve this result that we adopted the dimensional regularization scheme.13) We note the very important result that this contribution to n*pu satisfies the relation 'P" qpnlloop(q) = a (11.P" 2e2 (J:d n11oop(q) = (211:)4 xl + dX[ (12Id) ~P" (". arising from the term hZ3 .d12) We find ~ r(dI2) r(2 . (11. The gamma function r(2 .d12) ]. (11. (11. (11.2. the complete II" has the form rropO"(q) = (q2'1PO" _ qPqU)n(q2) .JCf x (lx)+m 2/1P(1) (m2+q2x(1_X))~2 x r(dI2) r(2 .1). (11. using (I .17) .2.v interaction Lagrangian.13) Il:0~(q) ~ (Z3 _1)(q2~P" . as we saw in Section 11. (11.2. The reason that dimensional regularization gives this result is that the conservation of current does not depend on the dimensionality of spacetime. Fortunately.5 on the basis of the conservation and neutrality of the electric current.15) so to order e2. The two terms in the integrand can be combined.2 +q2X(Ix))~Ir(1 +dI2) r(1dI2) (2 q Pq"x(1x)q2/1.1. This term has a structure like Eq.dI2).2. there is another term that must be added to nopO"(q).478 11 One·Loop Radiative Corrections and find .qP q") .x)(m + q2x (1 l " x)p2 (Z3.14) that was derived in Section 10.
(q')~.' d _ 4 (11.19) Now we can remove the regularization.(2.(2n)4 r( 4e Q d ')r(2. (11. and replacing d everywhere else by 4: 4e 2 . Therefore. As mentioned before. because for d = 4 both (m 2 +q2 x (l_x))1. there is an infinity in the oneloop contribution. The infinite part of 2 3 . that arise from the product of the pole in r(2d/2) with the linear terms in the expansion of Qd1(dI2) around d = 4. 2n 2 1 (23 . Moreover.d12) for 1(2 . though it does make a finite contribution to 231. arising from the limiting behavior of the Gamma function 1 . because as shown by inspection of Eg. For the same reason. where}' is the Euler constant. where Ii is some quantity with the units of mass. .20) We shall see in Volume II that this result may be used to derive the leading term in the renormalization group equation for the electric charge.dI2). The poles at d = 4 obviously cancel in n(q2).') 2 2 !o1 dxx(lx) 0 . to order .18) so that. e2 cannot be supposed to be dindependent. but these also cancel in the totaln(q2).1. the term ')1 in 1(2dI2) cancels in the totaln(q2). then . aside from gaugedependent terms). (11. }' = 0. it has the ddependent dimensionality [mass]4d.2.2.0 ~ (2d(2) . but not of n(q2).11.l)w ~ .6(2n)4 2 _ d(2 e2 1 6. Indeed.2.): r( 4e 2Q 0 r (2 2 (11. unity. " 23 ~ 1 .d . allowing d to approach its physical value d = 4.1 is given by using 1/(2 . and these too would have contributed to the finite part of 22 . There are other finite contributions to 23 1. in carrying out our dimensional regularization.2.5772157. If we take e2 IX 1i4. to order e2. we might have replaced (2n)4 with (2n)d.2 Vacuum Polarization 479 As we saw in Section 10.13).2 and (m2)1~2 have the same limit. the definition of the renormalized electromagnetic field requires that n(O) = 0 (in order that the residue of the pole in the complete photon propagator at q2 = 0 should be the same as for the bare propagator. x [(m2+q2X(I_x))~2 _(m2)~2]." . r(2. and the factor Tr 1 = 4 might have been replaced with the dimensionality 2dj2 of gamma matrices in arbitrary even spacetime dimensionalities d.5.
(11.2') = (2n )12/204(PI' X [ + P2'  PI .3 make contributions to the scattering Smatrix element of the form Sa(I.P2) [e l (2n )4 UI .x») dx. Eq.2. and likewise for particle 2.2) = lei e2 4 2 2 [1 n q + 11" (Q2)] u<'(Pl' + P2'  Pl  P2) (11.yi u1 =::: 0.2. iJ<7.3.PI .2 in powers of d .P2)Ja'<7Ar'<72 . Using the conservation property QIlUl'yllul = 0 the two diagrams together yield an Smatrix element: Sa+b(1.2  " 1. The only terms that do contribute to :n:(q2) in the limit d _ 4 are those arising from the product of the pole in r(2 ..2 and (m 2 )5.2') = (2n)12/2 04(PI' !.. these cancel between Z3 .<71 while ul.x)ln (1 + q2X~2.22) The physical significance of the vacuum polarization can be explored by considering its effect on the scattering of two charged particles of spin The Feynman diagrams of Figure 11.2 _ 1'.P2)[el(2n)4ul.2 ~ (i 2) In (1 + q2X~2. and qll is the momentum transfer q _ PIPI' = P2'~P2. where et and e2 are the charges of the two particles being scattered.d) In tl in the expansion of /14d in powers of 4 ..x))~2 _(m 2 )i. Also.dll) with the linear terms in the expansion of (m 2 + q2 x (1 ~ x»5. 2 _ 1' .22) the magnitude of the charge of the particle circulating in the loop in Figure 11.1 and the oneloop contributions to n:(q2). In the nonrelativistic limit.2.23) x [U1" yIlUl] [U2IYIlU2].i(2n)4 .yllut] x [ . 1t q 1 2 (11.2. but again. arising from the product of the pole in r(2 .23) in this limit becomes Sa+b(1.21) This gives at last n:(q2) = 2:2 1 1 x(l.2 _1'.2') = ~1~le22 [1 + n(q2)] J4(pl' + P2' . + P2'  (11. (11.yIlUI] i(2n)\12t[i(2n)4(q2'7 Il Y+qllqy)n(q2)] [e2(211")4u2IYYU2].. in this limit qO is negligible compared with Iql.4: (m 2+q'x(l. PI .2.dll) with the tenn (4 .2] [e2(2n)4u2'I'IlU2] .X)) .480 II OneLoop Radiative Corrections there are additional finite terms in Z2 1.d..24) . (11. 1I"(q2) is calculated using for e in Eq. Ul'yOUt . Sb( 1.2.
(11. are the same constants el and e2 that govern the interactions of the renormalized electromagnetic field.28) Eq.2 X 1'. as detennined from the longrange part of the Coulomb potential.11.2').1/. V(r)e. (1L2. (11."lb".2.t) = b".2. T Born (1. Comparing this with Eq.2 Vacuum Polarization 481 This may be compared with the Smatrix in the Born approximation due to a local spinindependent central potential V(r): SBorn(1.q" .29) where (11.P2)b.2.. For 1rl 1= a the integral (11.2.E2)TBorn(1..31) so the total charges of particles 1 and 2.2 . xy+rl (11.3 yield the same Smatrix element as a potential V(r) such that J d 3r V(r) eiq"r = ele2 1 +q~(q2) or.30) Note that / d).02 (11.30) can be carried out by a straightforward .2.2. inverting the Fourier transfonn.27) d3 .2') = 271"ib(E1' + E2 1  E1 . (11."X2eiPJ "XI e iP2 "X2 Setting Xl = X2 + r.1' .2.28) is to first order in the radiative correction the same potential energy that would be produced by the electrostatic interaction of two extended charge distributions et/1(x) and e2'1(Y) at a distance r: V(lrl) ~ ee /d'x l 2 Idly 4n~(X)~(y) .X 21) (11.25) d3x 1 3 d x2 .26) (271" )12/2eiP11"XI eiP2.2.2 . this gives ! SHorn = 4n2b (PI' X / 4 + P2'  Pl .23) shows that in the nonrelativistic limit the diagrams of Figure 11. ~(r) ~ 1 + In(O) ~ 1.2. (11."2 J J V(IXI.o1lb".
2. t1(r) must contain a term (1 +L)b 3(r) that is singular at r = 0. yielding a renormalized charge smaller by a factor I/O + L).) (bl Figure 11.2 = 12.32) The complete expression for the charge distribution function is then ~(r) ~ (I + L)b 3 (r)  811: r "3 3 10 0 x(l. we have seen that the integral of l1(r) over all r equals +1. As a check.2. Hence if we identify the momentum space cutoff A with the divergent .31): L= ~ 8.2. (11. 3 f 10r 3 d r3 x(1 _ x) dx [1 + .jx(i mr 1 x) ] exp ( .2.jx(7 xl] exp (~) (11. so that the bare charge is partially shielded. Diagram (b) represents the lowestorder vacuum polarization correction to the tree approximation graph (a). we find that the part that is divergent for a + 0 is LX! . we may note that if we cut off the divergent integral (11. wavy lines are photons. (11.jx(i mr x) ). Here lines carrying arrows are charged particles.2.32) by taking the integral to extend only over r > a. repelling their antiparticles to infinity. Two diagrams for the scattering of charged particles.3. Therefore. 2 Ina. However. r X)]exp(~).482 2 11 OneLoop Radiative Corrections (.x)dx x [I + r .33) The physical interpretation of this result is that a bare point charge attracts particles of charge of opposite sign out of the vacuum.34) aI. This expression is negative everywhere. (11. contour integration: t1(f)=8::r 3!o1 X(lX)dX[l+ Jx(7 1 . with L chosen to satisfy Eq.1 .
The effect of vacuum polarization is therefore very much larger for t = 0 than for higher orbital angular momenta. for hydrogenic orbits of electrons around a nucleus of charge Z e we have a = 137/Z m (where here m = me). the wave function of electrons in ordinary atoms will generally be confined within a much larger radius a:. (11.'o 'V( ) ~ ~ (2np ele2 Ja3 qe i" [rr(q2)] ' q2 (11.2. the effect of Feynman graph (b) in Figure ll.37) becomes LIE ~ 1'1'(0)1 2 Ja 3 . (11. (l + L) I eBt.3 is to shift the energy of an atomic state with wave function v'(') by LIE ~ Ja 3 2 r IIV(. (11. (11.) 1'1'(')1 ..11.2. For t = 0 the wave function is approximately equal to the constant W(O) for r less than or of the order of m. On the other hand.4.y.38) This perturbation falls off exponentially for r:. As we shall see in Chapter 14.. (11.39) Using Eqs.1. For orbital angular momentum t.2. .2.2.37) where 1\ V (r) is the perturbation in the potential (11.). the wave function behaves like rt for r « a.22). the integral of the shift in the potential (for eje2 = _Ze 2) is J d 3r AV(r) = Ze2JT'(O) = _ 4ZCl:~ 15m .y. IIV(.2 Vacuum Polarization 483 part of L is related to the divergent part of Z3 .38) and (11. m.36) Eq.40) Also.2.2. for instance. so Eq.35) because to order e2 the renormalized charge (10. so Eq.2.2.41) .1 by (Z3 .35) is confirmed below.2.37) gives 1\£ proportional to a factor (ma)2l+3.2. (11. (11.( 1 + HZ) 1) ) eBf . The energy shift will then depend only on the behavior of the wave function for r « a. Vacuum polarization has a measurable effect on muonic atomic energy levels. (11.18) is given by et = Z31/2 eBl .1)00 = 2La.1. (11.28): l. in states of hydrogenic atoms with t = 0 and principal quantum number n the wave function <It the origin is '1'(0) ~ y'4ii 2 (Zam)3/2 ' n (11.2.1 . m.2.
2.20). then in place of Eq. ••• For the purposes of comparison with later calculations. and obtain (23 1)~ ~ .2.13 MHz. 4 As discussed in Chapter 1. but the agreement between theory and experiment is good enough to verify the presence of the 37. Although vacuum polarization contributes only a small part of the radiative corrections in ordinary atoms.2. in the 25 state of hydrogen this energy shift is .2.2. it dominates the radiative corrections in muonic atoms. such tiny energy shifts became measurable because in the absence of various radiative corrections the pure Dirac theory would predict exact degeneracy of the 25 and 2p states of hydrogen.37. giving an energy shift proportional to m~m.484 11 OneLoop Radiative Corrections so the energy shift (11.L 122 X 107 eV. so the approximate result (11.13 MHz shift due to vacuum polarization. As we shall see in Chapter 14. note that if we had cut off the integral (11.20) we would have encountered an integral of the form (Z3~I)cv=6n2Jp K e1 fA d5 e2 dK= 6n2 /Jd4 ~ Ad4 d4 ' where J1 is an infrared effective cutoff of the order of the mass of the charged particle circulating in the loop of Figure 11. corresponding to a frequency shift 11£/2n11 of . (The easiest way to find the constant factor here is to require that the limit of this expression for d < 3 and A _ co matches Eq.) With such an ultraviolet cutoff in place. most of the +1058 MHz 'Lamb shift' between the 25 and the 2p states comes from other radiative corrections. (11. This is because most radiative corrections give energy shifts in muonic atoms that on dimensional grounds are proportional to mil' while the integrated vacuum polarization energy J d3 r ~ V due to an electron loop is still proportional to m. in this case the muonic atomic radius is not much larger than the electron Compton wavelength.39) is (11.2 = (210)2 mp . (11. e2 (11.2. This is sometimes called the Uehling effect.2 as in Eq.43) .1.2.6n2In(A/~).42) For instance. (11. we can pass to the limit d _ 4.39) only gives the order of magnitude of the energy shift due to vacuum polarization.7) at K = A.40).2. in which a muon takes the place of the orbiting electron. However.
3 Anomalous Magnetic Moments and Charge Radii 485 (0) (h) (0) (dl Figure 11. those involving insertions in incoming or outgoing lepton lines vanish because the lepton is on the mass shell.4.4) that needs to be calculated here: rllOOp(P. Of these graphs. here we do . and (d) is the term calculated in Section 11. " 'fd'['p 4][1 I(p'¥)+m]l'"] key(2n) (2n)4(p' k)2+ m2_ie y x . The graph involving an insertion in the external photon line is the vacuum polarization effect. respectively. Oneloop diagrams for the photonlepton vertex function P'. (The contribution of the vertex connecting the external photon line and the internal lepton line is taken as yl'. The oneloop graphs and renormalization corrections for the photonlepton vertex are shown in Figure 11. (2n)' k Ie eYp(2n) 21 je]' (11. Diagrams (a) and (b) are cancelled by lepton field renormalization terms. Unlike the case of the vacuum polarization. we shall calculate the shift in the magnetic moment and the charge radius of an electron or muon due to lowestorder radiative corrections. as discussed in Section 10.4.1) where pi and p are the final and initial lepton fourmomenta. 11.3.11.2.(p¥)+m] [ + m2 '] [. [ (2. This leaves one oneloop graph (the last in Figure 11.3.)' (p _ k)2 .p). other lines represent electrons or muons. discussed in the previous section. because a factor e(2n)4 was extracted in defining rl'. diagram (c) arises from the vacuum polarization calculated in Section 11.3 Anomalous Magnetic Moments and Charge Radii For our next example.) This integral has an obvious ultraviolet divergence. roughly like J d4k/(k 2 )2.3. Here wavy lines represent photons.
¥. In any case.p'y) + m]yp. .3.3) where q .p) = 2ie2 [I (2n)4 In r dx Jo dy ~ y) dk [k 2 + m2x 2 + q2 y (x .Y) +(k'i€)(lxf' = 2 [(k . In what follows we shall leave the integrals for the vertex function in their infinite [ann.pi is the momentum transferred to the photon. using a repeated version of the Feynman trick described in the Appendix to this chapter 1 ABC ~2 Jo [I dx Jo r dy [AY+B(xY)+C(Ixl]J (11. as we shall see the anomalous magnetic moment and charge radii can be calculated without encountering any ultraviolet divergences at all.p . (11. this gives 1 1 1 (p'_k)2+ m2_iE (pk)2+ m 2_iE k 2 IE = 2 fol dx fo1 dx foX dy fox dy [((pi _k)2+ m iE)Y 2 + ((pkf +m2  iE)(X .1). because the photon is a neutral particle and so the integral may be rendered finite by suitable modifications of the photon propagator (for instance by including a factor M 2 /(k 2 + M 2 ) with a large cutoff mass M).y)) + m]Yp i( p(l.p'y _ p(x _ y)) 2+ m2x 2 + q2 y (x  y) _ iEf3 .4) xyP [  x [ ¥.1) becomes ql IOOp(p'.x + y). without having to introduce modifications elsewhere to maintain gauge invariance.486 11 OneLoop Radiative Corrections not need a fancy regularization procedure like dimensional regularization to maintain the structure required by gauge invariance. i( 1'(1  Our next step is a Wick rotation.y) the integral (11. Shifting the variable of integration k ~ k +p'y +p(x. We start by combining denominators.3. (11.y)  4 iEr (11. with it being understood that if necessary any divergent integrals can be expressed in terms of a cutoff mass M.p(x .3.2) Applied to the denominators in Eq.3.3. As explained in the previous section.
y that multiply pill and pP are interchanged.Ie dK o o 2 2 .i( 1'( I . After a straightforward but tedious calculation.3 Anomalous Magnetic Moments and Charge Radii 487 the it: in the denominator dictates that when we rotate the kG contour of integration to the imaginary axis we must rotate counterclockwise./ . (11. (11.y. We can therefore simplify this expression by using the anticommutation relations of the Dirac matrices to move all factors P' to the right and all factors p to the left.'e2 (21t)4 {' r r ] xy . we drop terms in the denominator of odd order in k.~x(lx).11.2 + Zm (x 4x + 2) + Zq'(y(x .5) then becomes Ul~neloop(P'. so both may be replaced with their average: !Cvx+xy)+ Hx 2 xy  y) = .p)u u' {y'[  Jo dx J dy .h) +m]yp} x [K2+m2x2+q2y(x_Y)r3 (11. Eq.6) We next exploit the symmetry of the final factor under the reflection y + x . and replace the volume element d4k = idk l dk 2dk 3dk 4 with 2i1t 2K3dK. (11.y)}u +4im pip (y  x + xy) + 4im pi' (.y)) + m] y' x [i( p(1x+y).xy . Under this reflection.p(x . Putting this all together.4) now becomes rtloop(p'. the functions y .4).lu dx lX dy 10C! K3dK {K 2yP''/yll}'a f P/4 +yP [ . We also exploit the rotational symmetry of the denominator in Eq. Eq.p) = ~~:~~ .3. or equivalently over real values of k 4 _ ik o from 0Cl to +0Cl.3. .3. where K is the Euclidean length of the fourvector k.:2  x [K2+m2x2+q2y(x_y)r3. (11.y).y) + 1. so that the integral over k O from 0Cl to +0Cl is replaced with an integral over imaginary values from iOCl to +iOCl.3. replace ek a with '1Mk2/4.x + xy and x 2 .5) l We are interested here only in the matrix element ij'['Pu of the vertex function between Dirac spinors that satisfy the relations u'[i l' +m] ~O.x)] = 4.3. [i p+mJu~O. replacing them when they arrive on the right or left with im.
.1 in the correction tenn (11.2q'(y(x .) + qly(x  ..9) The form of each of these tenns is in agreement with the general result (10.' . the effect of insertions of corrections to the external photon propagator is a term.10) To order e2.x) 2 /(3 dK 3 ' (11. There are other diagrams that need to be taken into account.p) Yv .x)] .12) (2n) 0 0 0 [K1+m2x2+q2y(x_y)] where n(q2) is the vacuum polarization function (11.3. ~ (Z. (11. 1 p)2 (11..2. r~ac pol(P'.. + m2 x 2 + q2 y (x  y)] (11. y) (11. there is the zerothorder term yJl in rJi. or course.3...488 This gives finally 11 OneLoop Radiative Corrections u'r~nelOop(P'..4x + 2) .10) (with H(q') ~ 0) u'P(p'. 1)yI' .y) + 1.3. The inte